The Fly River, Papua New Guinea, Volume 9: Environmental Studies in an Impacted Tropical River System (Developments in Earth and Environmental Sciences)

Developments in Earth & Environmental Sciences, 9 THE FLY RIVER, PAPUA NEW GUINEA: ENVIRONMENTAL STUDIES IN AN IMPACTED...

Author: Barrie R. Bolton

15 downloads 459 Views 21MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Developments in Earth & Environmental Sciences, 9

THE FLY RIVER, PAPUA NEW GUINEA: ENVIRONMENTAL STUDIES IN AN IMPACTED TROPICAL RIVER SYSTEM Edited by

BARRIE BOLTON Environment Department, Ok Tedi Mining Limited, Tabubil, W.P., Papua New Guinea

Amsterdam – Boston – Heidelberg – London – New York – Oxford Paris – San Diego – San Francisco – Singapore – Sydney – Tokyo

Elsevier 30 Corporate Drive, Suite 400, Burlington, MA 01803, USA Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands Linacre House, Jordan Hill, Oxford OX28DP, UK First edition 2009 Copyright r 2009 Elsevier B.V. All rights reserved No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (þ44) (0) 1865 843830; fax (þ44) (0) 1865 853333; email: [email protected]. Alternatively you can submit your request online by visiting the Elsevier web site at http://www.elsevier.com/locate/permissions, and selecting Obtaining permission to use Elsevier material Notice No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent veriﬁcation of diagnoses and drug dosages should be made British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress ISBN: 978-0-444-52964-0 ISSN: 1571-9197 For information on all Elsevier publications visit our website at books.elsevier.com

Printed and bound in Hungary 09 10 11 12 13

10 9 8 7 6 5 4 3 2 1

List of Contributors Simon C. Apte

Centre for Environmental Contaminants Research, CSIRO Land and Water, PMB 7, Bangor Sydney, New South Wales 2234, Australia

Stephen J. M. Blaber

CSIRO Marine & Atmospheric Research, PO Box 120, Cleveland, Queensland 4163, Australia

Barrie R. Bolton

Environment Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, Western Province, Papua New Guinea; Barrie Bolton Consulting Pty Ltd., PO Box 3053, Ripponlea, Victoria 3185, Australia, and School of Geosciences and Water Studies Centre, Monash University, Clayton, Victoria 3800, Australia

Gregg J. Brunskill

84 Alligator Creek Road, Alligator Creek, Queensland 4816, Australia

John S. Crockett

School of Oceanography, University of Washington, Seattle, WA 98195, USA

Yantao Cui

Consulting Geomorphologist, Stillwater Sciences, 2855 Telegraph Avenue, Berkeley, CA 94705, USA

Geoff Day

Newcrest Mining Ltd., Level 8, 600 St Kilda Road, Melbourne, Victoria, 3004, Australia

William E. Dietrich

Department of Earth & Planetary Science, University of California, Berkeley, CA 94720, USA

Boga Figa

Livelihood Programs Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, Western Province, Papua New Guinea

Miguel A. Goni

College of Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, OR 97331, USA

Henry Kundapen

Environment Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, Western Province, Papua New Guinea

Jessica Lynas

School of Animal Biology (M092), The University of Western Australia, Crawley, Western Australia 6009, Australia

xiv

List of Contributors

Andrew R. Marshall

Andrew Marshall & Associates Pty Ltd., 43 Warrangarree Drive, Woronora Heights, New South Wales 2233, Australia

David A. Milton

CSIRO Marine & Atmospheric Research, PO Box 120, Cleveland, Queensland 4163, Australia

Max S. Moulds

Entomology Section, Australian Museum, 6 College Street, Sydney, New South Wales 2010, Australia

Charles A. Nittrouer

School of Oceanography, University of Washington, Seattle, WA 98195, USA

Andrea S. Ogston

School of Oceanography, University of Washington, Seattle, WA 98195, USA

John Pﬁtzner

Australian Institute of Marine Science PMB 3, Townsville MC, Queensland 4810, Australia

Geoff Pickup

Consulting Geomorphologist, 1538 Sutton Road, Sutton, New South Wales 2620, Australia

Jesse L. Pile

Environment Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, Western Province, Papua New Guinea

Monica T. Rau

Community Relations Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, Western Province, Papua New Guinea

Douglas P. Reagan

Doug Reagan & Associates, 350 Gordon Drive, Castle Rock, CO 80104, USA

Peter V. Ridd

Marine Geophysics Laboratory, School of Mathematical; and Physical Sciences, James Cook University, Townsville, Queensland 4811, Australia

Nicola J. Rogers

Centre for Environmental Contaminants Research, CSIRO Land and Water, PMB 7, Bangor Sydney, New South Wales 2234, Australia

Joel C. Rowland

Department of Earth & Planetary Science, University of California, Berkeley, CA 94720, USA

John P. Salini

CSIRO Marine & Atmospheric Research, PO Box 120, Cleveland, Queensland 4163, Australia

Jenny L. Stauber

Centre for Environmental Contaminants Research, CSIRO Land and Water, PMB 7, Bangor Sydney, New South Wales 2234, Australia

List of Contributors

xv

Andrew W. Storey

School of Animal Biology (M092), The University of Western Australia, Crawley, Western Australia 6009, Australia

Charles Tenakanai

Livelihood Programs Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, Western Province, Papua New Guinea

John P. Walsh

Department of Geological Sciences and the Institute for Coastal Science and Policy and Coastal Resources Management Program, 301B Graham Building, East Carolina University, Greenville, NC 27858, USA

Markson Yarrao

Environment Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, Western Province, Papua New Guinea

Irena Zagorskis

Australian Institute of Marine Science PMB 3, Townsville MC, Queensland 4810, Australia

Preface Ever since the discovery of rich copper and gold mineralization under Mount Fubilan in the rugged central highlands of western Papua New Guinea there have been concerns about the potential for environmental impacts on the nearby Fly River and in particular its tributary, the Ok Tedi. These concerns prompted a wide-ranging and comprehensive series of studies of the river system designed, in the ﬁrst instance, to develop an understanding of the premine environmental setting, and then following the initiation of mining in 1984, the nature and extent of mine-related impacts. Through the combined efforts of various universities, government departments, individuals and in particular, the management of Ok Tedi Mining Limited, the owner and operator of the Ok Tedi mine, this on-going attempt to monitor environmental impact has resulted in the assembly of an enormous body of work on this important river system. While much of the data acquired from these studies has been presented in workshops, conferences, and a wide range of specialist publications, there have been few attempts at bringing together the results of this work into a single volume. The 17 chapters comprising this volume are intended to summarize aspects of the natural history of the system as well as provide a multidisciplinary appraisal of the nature and extent of mine-related impacts. They provide in most instances both a synthesis of existing knowledge and a review of the status of current research and development in the respective ﬁelds addressed. The chapters have been grouped under ﬁve main themes. The ﬁrst group includes six chapters that deal with aspects of the physical environment. The ﬁrst chapter describes the geomorphology, hydrology, and climate of the river system. It also provides an account of the changes in the geomorphology that have occurred as a result of the release of overburden and tailings from the Ok Tedi mine. This is followed by a chapter that examines the changes that have occurred to the texture, geochemistry, and mineralogy of the sediments carried down the river system before and after the start of mining at Mount Fubilan. Another explores the impact of mine tailings and waste rock on the river ﬂoodplain and particularly the speed of these changes as evidenced by the concentrations of copper in ﬂoodplain sediments. The remaining chapters in this group examine the processes,

xviii

Preface

sediments, and stratigraphy of the Fly River delta, the fate of mine-derived copper in the sediments of the shallow Gulf of Papua adjacent to the mouth of the Fly River and ﬁnally, the variable styles of sedimentation, including material from the Fly River, in the clinoform developing at the mouth of the river. The two chapters of the second group deal with predicting the impact of mining on the physical environment. The ﬁrst presents the results of efforts to model the changes in the rate of river bed aggradation brought about mainly by the discharge of waste rock and tailings from the Ok Tedi mine, while the second focuses on the resultant increase in the incidence of ﬂoodplain inundation and how this is likely to change in the future. The two papers of the third group discuss aspects of the mine’s inﬂuence on the chemistry and biotoxicity of river waters. The ﬁrst provides a review of the processes controlling the fate of copper, the main metal contaminate originating from the mine, in the river system while the second discusses the considerable research into the speciation, bioavailability, and toxicity of copper. The fourth group of three chapters deals with the ﬁsh population of the river and how it has been affected by mining. The ﬁrst of these chapters’ focuses on the biology of barramundi in the river, one of the main food sources for the communities living along the river’s edge. A second chapter examines the effects of mine-derived waste materials on ﬁsh habitat while the third and last explores the spatial and temporal changes in ﬁsh assemblage to assess the impact of the mine. The last group of four chapters deals with the fauna of the region, the riparian vegetation, and the development of a food web for the system. It is hoped that publication of these contributions will stimulate further discussion and research on the impact of mining on riverine and their adjacent near-coastal environments. Finally, the job of an editor of a volume such as this would be impossible without the help and support of a lot of people. First I wish to thank Keith Faulkner, a former Managing Director of Ok Tedi Mining Limited (OTML) and Jim Veness, current Manager of the OTML Environment Department, for their unfailing support and encouragement throughout this project. OTML Board members, Allan Roberts, Ross Garnaut, Jochen Tilk, Dick Zandee, and Simon Tosali are also thanked for their support and encouragement in the production of this volume. The ﬁnancial support provided by Ok Tedi Mining Limited for the publication of this book is also gratefully acknowledged. I would also like to acknowledge the efforts of the following people who provided critical reviews and many editorial improvements to the

Preface

xix

contributions of this volume: Geoff Pickup, Len Murray, Barry Hart, Mead Allison, Simon Apte, Ted Edwards, David Heggie, Kent Hottle, Paul Humphries, Denis Mackey, Alan Orpin, Max Moulds, John Chapman, Wade Hadwen, Robert Hughes, Peter Teasdale, Franc- ois Edwards, Scott Markich, Susan Adams, John Rinne, Katie Farnsworth, Robert Dalrymple, Scott Miller, Jim Veness, Jacky Croke, Ian Cowx, Alan Whitﬁeld, and H. Gill. And ﬁnally, I wish to thank the support of the very patient staff at Elsevier for this project and speciﬁcally, Linda Versteeg-Buschman, Suja Narayana and Femke Wallien.

Barrie Bolton Environment Department, Ok Tedi Mining Limited

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00401-1

Chapter 1

Geomorphology, Hydrology, and Climate of the Fly River System Geoff Pickup1, and Andrew R. Marshall2 1

Consulting Geomorphologist, 1538 Sutton Road, Sutton, New South Wales 2620, Australia 2 Andrew Marshall & Associates Pty Ltd., 43 Warrangarree Drive, Woronora Heights, New South Wales 2233, Australia

1.1. Introduction The Fly River system drains an area of about 75,000 km2 in western Papua New Guinea (Fig. 1.1). There are three major rivers in the system: the Ok Tedi, which drains the Hindenburg Ranges; the Upper Fly, which drains the southern part of the Victor Emanuel Range; and the Strickland, which drains the Victor Emanuel and Central Ranges. The Upper Fly and the Ok Tedi meet at D’Albertis Junction to form the Middle Fly, which meanders down a 400-km-long ﬂoodplain with extensive scroll bar complexes, cutoffs, and blocked valley lakes. The Middle Fly and the Strickland meet at Everill Junction before entering the Fly Delta. The delta covers about 10,000 km2 and extends downstream for another 400 km before entering the ocean in the Gulf of Papua. The Ok Tedi receives waste rock and tailings from Ok Tedi while the Strickland receives sediment load from mining operations at Porgera. This chapter covers the climate, geomorphology, and hydrology of the Fly. It examines rainfall distribution and basin hydrology. It describes the river system from the mountains to the lower end of the ﬂoodplain, including linkages between the river channel and the many off-river water bodies (ORWBs) in tributary valleys and on the ﬂoodplain. It concludes with a summary of mine-related impacts on the river system. Corresponding author. Tel.: þ61 26238 3427;

E-mail: [email protected] (G. Pickup).

4

G. Pickup and A. R. Marshall

Figure 1.1: Location map. Mining operations at Tabubil and Porgera are shown by arrows.

Geomorphology, Hydrology, and Climate of the Fly River System

5

1.2. Climate The Fly basin has a humid tropical climate. Average rainfall varies with elevation (Moi et al., 2001). Falls in excess of 10,000 mm/year occur at the Ok Tedi mine site (1,500–2,000 m), declining to about 8,000 mm/year along the upper and middle Ok Tedi. Further decreases occur downstream with values of 5,250 mm/year being recorded at Kuambit just downstream of D’Albertis Junction, 3,869 mm/year at Manda about two-thirds of the way down the Middle Fly ﬂoodplain, and 1,847 mm/year at Obo near Everill Junction. Heavy rainfall occurs throughout the year in the mountains, but there is a slightly less wet period from September to November. Seasonal variations are also stronger over the southern part of the ﬂoodplain, where rainfalls are lower than in the mountains. Based on other stations in lowland Papua (McAlpine et al., 1975), annual evaporation is probably within the range 1,500–2,000 mm, which is less than annual rainfall in many areas. Rainfall not only varies between seasons. The region experiences both El Nin˜o and La Nin˜a events, including severe drought episodes in 1972, 1983, and 1997. There may also be longer-term variations in rainfall. For example, Moi et al. (2001) report that rainfalls for 1999–2000 exceeded long-term averages across the region, with the largest percentage increases occurring in lowland areas. The rainfall record for Tabubil, just below the mine site, shows some connection with the Southern Oscillation Index (SOI) (Fig. 1.2) but mainly during severe droughts. There is no obvious relationship at other times, making it difﬁcult to use the SOI as a predictive tool.

1.3. River System Geomorphology In its upper reaches, the Ok Tedi and its tributaries drain a heavily dissected ridge and ravine landscape. The ridges rise to over 2,000 m in the north, but most of the basin lies between 200 and 800 m. The eastern part of the basin is karst country, including the massive Hindenburg Wall escarpment, and contains large areas of landslide debris and old debris ﬂow deposits. In the west, igneous rocks are exposed on high mountains, but much of the area consists of shales, limestones, and sandstones. Slopes are unstable in spite of a dense rain forest cover, and landslides and debris ﬂows are common. Mine waste is disposed of into two tributaries of the Ok Tedi: Sulphide Creek (which joins the Ok Gilor, and then the Ok Mabiong in its lower section)

6

G. Pickup and A. R. Marshall

1200

30

1000

10 SOI Value

800 0 600 -10 400 -20 200

-30 -40 1970

Tabubil Monthly Rainfall (mm)

20

0 1980

1990

2000

Figure 1.2: Monthly rainfalls for Tabubil compared with the Southern Oscillation Index. Both datasets have been smoothed using a ﬁve-point moving average. and the Ok Mani (Fig. 1.3). These tributaries are referred as the Mine Area Creeks. These systems occupy steep, narrow valleys and, prior to mining, were cut down to bedrock for most of their length. Sediment deposits were restricted to local boulder chokes and short reaches with thin layers of armored cobbles. Sulphide Creek contains a waterfall, as does Harvey Creek, a small tributary of the Ok Mani that now receives waste rock from dumps on the southern side of the mine. These creeks are typical of the supply zone (Table 1.1) and were transporting virtually all delivered load before mining operations started. The Upper Ok Tedi is a fast-ﬂowing river capable of transporting large quantities of coarse sediment, including boulders. For much of its length, it occupies a bedrock gorge less than 200 m wide at its base, although there are wider sections with a few islands developing in the lower sections. Debris ﬂow deposits ﬂank the upper reaches of the gorge, and bed levels have ﬂuctuated over time in response to massive landslides from the Hindenburg Wall and other areas, including one of 7 km3 about 8,800 BP (e.g., Pickup et al., 1979; Blong, 1991). Prior to mining, this reach of the river was classiﬁed as a supply zone (Table 1.1). However, unlike the Mine Area Creeks, the bed had a veneer of coarse sediment in transit, much of which probably came from the 1977 Hindenburg Wall landslide. Just upstream from Ningerum, the Lower Ok Tedi emerges from its gorge into a gradually widening valley ﬁlled with what are probably Pleistocene

Geomorphology, Hydrology, and Climate of the Fly River System

7

Figure 1.3: A SPOT image of the Ok Tedi mine and the Mine Area Creeks. alluvial deposits. These include weathered and indurated sand and gravel deposits topped by distinctive red soils. Initially, these ‘‘red beds’’ form valley edges and conﬁne the channel but, as the valley widens, form terraces or terrace remnants, and the Ok Tedi develops a currently active ﬂoodplain within them. A braided channel develops, and forested islands have formed on some of the more stable braid bars. Before mining increased sediment input to the river, this reach was classiﬁed as an armored zone merging into a gravel–sand transition zone (Table 1.1). Finer material, including a large volume of sand, is now being deposited in this area. Prior to mining, the armored zone ended upstream of the junction with the Ok Mart where braiding ceases, and there was a distinctive gravel front. Beyond this point, there was a gravel–sand transition zone (Table 1.1). Flow in the gravel–sand transition zone was (and still largely is) concentrated in a single channel, and meandering begins to develop. However, although there are several meander cutoffs and some channel traces, the ﬂoodplain is conﬁned to a narrow belt, 1–4 km wide within the alluvial piedmont, and shows some traces of minor incision. The outer sections of some meanders also cut into Pleistocene terrace remnants, and lateral movement is restricted by indurated and highly weathered soils developed on former ﬂuvial sediments.

8

Zone

Supply

Armored

Gravel–sand transition

Sand

Backwater

Location

Mountain tributaries and upper gorges. Mine Area Creeks, Ok Tedi to Ningerum

Sediments

250 mm–W1.0 m 100–200 mm at but mainly coarse surface, mixed sand and gravel below

Bimodal Well-sorted sands distribution with mainly in 0.2- to gravels in 15- to 0.4-mm range 60-mm range and well-sorted sands

Channel form

Bedrock gorges Wide, shallow and boulder-lined channels, often torrents in braided but with conﬁned reaches. many stable bars Wide, shallow and islands channels, in wider sections, often braided but connected by steeper reaches. Few deposits

Varies from wide, Meandering with Meandering with shallow, and levees, welllow levees, scroll straight to deep developed bars, and and sinuous. ﬂoodplains and frequently Conﬁned by off-channel water inundated valley walls, bodies backswamps. terraces and (backswamps and Sand and silt clay terrace remnants, cutoffs). Terrace bed deposits although there is remnants restrict depending on some ﬂoodplain meandering local variations in development locally ﬂow velocity

Middle and lower Upper ﬂoodplains. Floodplains. gorges to gravel Bige to Atkamba Atkamba to front upstream of Manda Bige

Fly upstream of the Strickland between Manda and Everill Junction. Minor ﬂoodplain tributaries Well-sorted very ﬁne-medium sand alternating with silts and clays

G. Pickup and A. R. Marshall

Table 1.1: Classiﬁcation scheme for river reaches on the Fly River System prior to mining.

Sediment regime

Controlling factors Underlying Behavior is Possibly an Supply controlled, Backwater effects bedrock controls sediment supply armored reach at mainly by sand. force deposition slope, valley walls controlled but by lower sea levels. Local controls on but local scour conﬁne channel material in armor Channel forms lateral movement may occur when layer typical of a mixed imposed by backwater effects load sediment terrace remnants are temporarily regime removed Source: Adapted from Pickup (1984) and Higgins et al. (1987).

Geomorphology, Hydrology, and Climate of the Fly River System

Transport capacity Transport capacity Transport capacity Mobile sandy Unable to usually greater greater than just greater than bedload and ﬁne transport all load than supply, supply, except for supply for sand. suspended load. when backed up except after major sizes in the armor Gravels rarely Close to shortlandslides. Bed layer that are only move and may be term equilibrium covered by a transported in locally derived with respect to veneer of material large ﬂows from Pleistocene bed material load. in transit or swept terraces as well as Often below clean. Material upstream reaches capacity for wash often moves in load slugs or waves

9

10

G. Pickup and A. R. Marshall

The Middle Fly extends for about 420 km from D’Albertis Junction to Everill Junction, where it meets the Strickland. There are two distinct reaches: a sand zone and a backwater zone (Table 1.1). Prior to mining, the sand zone had a riverbed consisting of well-sorted ﬁne-to-medium sand and extended as far downstream as Manda. This reach has a steeper water surface slope than the backwater zone that extends from Manda downstream to Everill Junction. The pre-mining backwater zone had a bed composed of mainly ﬁne sand but with an increasing proportion of silt clay particles further downstream until less than 20% sand was present. A localized body of medium sand occurred downstream from Bosset and is probably associated with the slightly steeper river long proﬁle there. It may have originated from the catchment of the blocked valley lake of Bosset Lagoon. The evolution of the Middle Fly may be as follows. During the Pleistocene, the river repeatedly entrenched and dissected surrounding hills during periods of low sea level (Lofﬂer, 1977). As sea level rose, reaching its current height around 5,000 BP, the river aggraded at about 1 mm/year (Pickup et al., 1979), declining to about 0.1 mm/year (Dietrich et al., 1999). The aggradation process created a number of distinctive landforms (Paijmans et al., 1971) (Figs. 1.4 and 1.5). First, there is a central meander belt made up of the river channel, a wide and sometimes discontinuous levee, scroll bar complexes, Pleistocene terrace remnants, and partly ﬁlled meander cutoffs, often connected to the main river by tie channels. This system forms an inner ﬂoodplain, accumulates most of the sediment from upstream that is deposited, and is perched above the surrounding outer ﬂoodplain. Second, the aggradation and levee buildup of the inner ﬂoodplain was sufﬁcient to block off or back up most of the lowland tributary systems. This created a system of blocked valley lakes, backswamps, ﬂood basins, and drowned tributary channel systems on each side of the central meander belt. These systems may be up to 7 m lower than the levees that border the main river channel. The elevation difference decreases downstream and is smallest in the backwater zone upstream of Everill Junction (Dietrich, 2000). However, it remains a signiﬁcant feature in determining both ﬂoodplain behavior and landforms down as far as the Fly Delta. While the Middle Fly contains spectacular evidence of meandering in the form of cutoffs and scroll complexes, it has not experienced rapid lateral migration. Pickup et al. (1979) found that only three loops were cut off between 1899 and 1979. More extensive analyses by Barr Engineering (1995) and Dietrich et al. (1999) found that bank erosion and meander migration rates were highest close to D’Albertis Junction and decreased by a factor of three, downstream to the backwater-affected reach.

Geomorphology, Hydrology, and Climate of the Fly River System

11

Figure 1.4: A LANDSAT TM image showing the ﬂoodplain systems of the upper Middle Fly during low-ﬂow conditions. Flooded off-river water bodies are shown in black or blue. Dried-out areas that are normally ﬂooded are pink or purple. Green areas are mainly rain forest. Yellow squares are gauging stations or sampling sites. The vegetation of the ﬂoodplain corresponds fairly closely with the steepness of the long proﬁle. Rain forest extends from D’Albertis Junction to a point beyond Mabaduam, entering a transitional reach close to the junction with the Agu. The lower, backwater-affected reach, beyond Manda,

12

G. Pickup and A. R. Marshall

Figure 1.5: A LANDSAT TM image showing the ﬂoodplain systems of the lower Middle Fly and lower Strickland during low ﬂow conditions. Flooded off-river water bodies are shown in black or blue. Dried-out areas that are normally ﬂooded are pink or purple. Green areas are mainly rain forest. Meander cutoffs, blocked valley lakes, and swamps behind the main channel levees are all clearly visible. Yellow squares are gauging stations or sampling sites. largely consists swamp grass or riparian vegetation (Day et al., 1993). The low gradients down the Middle Fly and the small transverse differences in ﬂoodplain elevation make ﬂoodplain vegetation systems sensitive to changes in water level on the Fly if they are of sufﬁcient duration.

Geomorphology, Hydrology, and Climate of the Fly River System

13

It is likely that the Fly has evolved only slowly since the sea level reached its present level. Before sediment load increased due to mining, the levee system was probably growing upward and extending downstream through deposition of sediment. However, given the limited sediment load reaching the Middle Fly, the rate of progress was very slow. For example, Pickup (1984) estimates that it would take somewhere between 1,400 and 27,000 years for the sand zone in the Middle Fly to extend as far as Everill Junction, provided that the bed level of the Strickland remained constant. The backswamps receive limited amounts of material from the main channel, apart from close in to the levees, and may grow upward more by accumulation of organic material than by deposition of sediment. Studies using copper as a marker show that most of the sediment leaving the main river channel is deposited on the levees, in off-channel water bodies with tie channels to the main river, and by backﬂow up tributaries (Day et al., 1993; Dietrich, 2000). The Strickland has a catchment area of 36,740 km2 at Everill Junction and is a much larger river than the Fly (catchment area of 18,400 km2). It is also closer to the main sources of natural sediment load in the mountains. Like the pre-mining Ok Tedi and the Middle Fly, the Strickland has an armored zone, a gravel–sand transition zone, and a sand zone, but these extend much further down the system. The Strickland has a steeper gradient than the Fly, and its ﬂoodplain probably had more active channel shifting. The Strickland backs up some very large off-channel water bodies, including Lake Murray. It is separated from the Middle Fly ﬂoodplain by a 10- to 30-km-wide preHolocene former ﬂoodplain surface that forms an extensive set of terraces. Once the Strickland and the Middle Fly join, river characteristics change. Prior to mining, the Strickland transported perhaps 8–10 times the natural load of the Fly. Even now, it probably contributes more than 60% of the total load (Moi et al., 2001). Also, the bed material of the Strickland and the Lower Fly contain much more sand than the Middle Fly. The Fly Delta is a funnel-shaped system dominated by tidal action. The tidal range is about 3.5 m at the mouth and 5 m at the apex (Wolanski and Eagle, 1991). Channel bed material is mainly sand, and the majority of ﬁne material is advected offshore (Harris et al., 1993). Terrace remnants in the delta are substantially higher than those surviving in the Middle Fly. The channel also seems to become more conﬁned downstream and does not have the extensive ﬂoodplains and backswamps of the Middle Fly. This may suggest minor uplift due to tectonic activity or isostatic adjustment. The conﬁned channel, increased river discharges, and tidal ﬂushing all limit the deposition of mine-related sediment reaching the delta.

14

G. Pickup and A. R. Marshall

1.4. Hydrology 1.4.1. Runoff and Flooding Most of the runoff in the Fly River originates in the mountains. Runoff at Kuambit just below the junction of the Ok Tedi and the Upper Fly is about 6,600 mm/year or about 70–80% of the total rainfall. However, once runoff enters the ﬂoodplain reach of the Middle Fly, ﬂow patterns change (Fig. 1.6). At the upstream end of the ﬂoodplain, water levels vary rapidly as ﬂoods from the Ok Tedi and the Upper Fly enter the system. By the time ﬂow reaches Manda, about 230 km downstream, much of the short-term variability has disappeared, and most of the changes in water level occur at a scale of weeks to months. The main cause of this ﬂood wave attenuation is exchange of water between the main river and many off-river ﬂoodplain storages through tie channels, levee breaches, and, if water levels are high enough, directly across the ﬂoodplain. Dietrich (2000) describes the ﬂoodplain storages as follows: There are 38 oxbow lakes within the scroll complex of the study area, 25 of which have well maintained tie channels which transport water and sediment over a large range of main channel stage. A further 8 of the oxbows have smaller tie channels which transport water and sediment only over the upper ranges of main channel ﬂow with the remainder (5 oxbows) having no apparent tie channel. There are at least 35 blocked valley lake systems within the backswamp areas of the ﬂoodplain ranging in size from o0.1 km2 to W 50 km2. The off-river water bodies (ORWB) (oxbow and blocked valley lakes) comprise 10% of the total area of the ﬂoodplain or approximately 350 km2.

20

Elevation (m)

15

10

5

0 1/1/95

1/1/96

1/1/97

1/1/98

1/1/99

1/1/00

1/1/01

Figure 1.6: Water levels on the Middle Fly 1995–2000.

Geomorphology, Hydrology, and Climate of the Fly River System

15

The ORWBs can hold 2–3% of the annual ﬂow of the Fly before general ﬂooding occupies most of the ﬂoodplain. However, because there is continuous inﬂow and outﬂow between the ﬂoodplain storages and the main channel during ﬂoods, a much larger proportion of ﬂow is cycled through them. Dietrich and Day (2004) has suggested that at least 20%, and possibly up to 40%, of the Fly River water passes through the ﬂoodplain. Model calculations (see Pickup, 2009) suggest that the value is closer to 10% but could increase to between 14 and 16% by 2050, depending on the level of bedload extraction by dredging from the Lower Ok Tedi. Most of the transfer between the main river channel and the ﬂoodplain is inﬂow and outﬂow to and from storages rather than ﬂow down backplains between levee breaches. However, there is evidence on satellite imagery of backplain ﬂow on the west side of the Fly near Bosset. Dietrich and Day (2004) have suggested that water may leave the middle reaches of the Middle Fly via tie channels and ﬂow into the Agu River. This water then returns to the lower Middle Fly via the Agu junction below Manda. It is not clear how important this effect is. Three small tie channels connect ORWBs to both the Fly and the Agu, and it is possible for water to ﬂow into the Agu across the ﬂoodplain at high water levels. At the same time, ﬂood levels on the Fly propagate up the lower Agu very quickly, so the ﬂow path from the Fly via the upper Agu would soon be restricted by rising water levels. 1.4.2. Links between the River and its Floodplain – The Fly River Tie Channels The term ‘‘tie channel’’ was coined by Blake and Ollier (1971) to describe the narrow channels that link the main river channel and the oxbow lakes that occupy meander cutoffs on the Fly ﬂoodplain. These channels experience bidirectional ﬂow, depending on differences in water level between the oxbow lake and the main river. Typically, water ﬂows into the oxbow on the rising stage of ﬂoods in the main river and out of it as ﬂoodwater levels recede (Fig. 1.7). Hydraulic gradients therefore vary with time but may be steep, and ﬂow velocities of 1–2 m/sec are quite common at times of rapidly changing water levels. This is enough to keep many tie channels clear of sediment and some have functioned for hundreds and perhaps thousands of years (Rowland et al., 2005). It is likely that tie channels play a crucial role in the ecology of ORWBs. They provide a route for the transfer of water and sediment, transport contaminants, and allow the exchange of carbon. They may also serve as refugia and breeding sites for ﬁsh species, especially during drought.

16

G. Pickup and A. R. Marshall

Figure 1.7: Some features of tie channels. Upper left shows rapid and turbulent ﬂow through a narrow vegetation-lined tie channel. This is sufﬁcient to move sandy bed material from the Middle Fly into the receiving water body. Upper right shows a tie channel when water levels on the main river and the connected water body are similar. Virtually no ﬂow is occurring. Lower left shows a tie channel draining clean water from the ORWB into the Fly. Lower right shows what occurs when water levels in the river and the ORWB have equalized during a large ﬂood. The levee breach tie channel is completely submerged and contains still, clear water.

Water circulation through tie channels may have a role in maintaining water quality in many ORWBs, especially if they are small. Most ORWBs will be replenished by rainfall, and virtually all become connected with the main river during major ﬂoods. However, during periods of low-medium ﬂow on the main river, ﬂow through tie channels will be the main process of water circulation. This may be a factor limiting eutrophication in the smaller ORWBs, where tie channel ﬂow could replace 5–10% of the total water volume in a day. In larger water bodies, input from rainfall, inﬂows from the surrounding catchment, and outﬂow to the main river would play a greater role in water circulation. However, inﬂow and outﬂow of main channel water is still likely to be the dominant process.

Geomorphology, Hydrology, and Climate of the Fly River System

17

If we deﬁne a tie channel as one that links the main river channel and an ORWB and experiences bidirectional ﬂow, then there are at least seven types on the Lower Ok Tedi and the Middle Fly (see Fig. 1.8 for examples). Each type may have evolved in a different way, and there may be variations in the potential for silting up. The main types are: Cutoff tie channels that connect oxbow lakes of meander cutoffs to the main river. The oxbows are the deepest ORWBs on the ﬂoodplain and usually hold water during drought, even if water levels fall below the elevation of the tie channel bed. They are therefore likely to be important refugia for a whole range of aquatic species. Several subtypes occur. The most common form drains through a sediment plug at one or both ends of the former meander channel. Several others exit closer to the middle of the cutoff through the swales between adjacent former point bars. A third variation occurs where the tie channel links with a tributary, and there are even cases where the tributary crosses the cutoff and functions as the cutoff tie channel as well as draining its own catchment. Levee breach channels. These may vary in size and usually connect the river to areas of standing water in the backswamps behind the levee. These backswamp lakes vary in depth and area over time and may dry out completely when river water levels are low during drought. Levee breach channels are usually short and may develop when localized levee collapses occur and open a path for water to ﬂow. There are shallower and probably more short lived than the cutoff tie channels. Many levee breach channels may have silted up, as levee deposition has accelerated in response to deposition of mine waste. Tributary tie channels. Over the last 5,000–7,000 years, the central meander belt of the Fly has grown in height faster than the outer edges of the ﬂoodplain and the ﬂoors of tributary valleys. This has created a set of blocked valleys often containing permanent or ephemeral lakes that provide the largest areas of standing water on the ﬂoodplain. The blocked valley lakes cover a large area, and some of the larger ones may have water depths similar to the meander cutoff oxbows close to the main river. The larger lakes hold water during drought, and many are distant from the main river channel. The more distant lakes are largely unaffected by deposition of sediment from the main river. Tributary tie channels can be highly sinuous and often have well-developed levees. While no dating has been done, it is likely that most of these channels are hundreds to thousands of years old, especially in areas distant from the main Fly River channel. Drowned tributary tie channels. The backwater zone of the Middle Fly extends from Manda to Everill Junction. Backwater from the Strickland

18

G. Pickup and A. R. Marshall

A

B

C

D

E

F

Figure 1.8: Examples of tie channel types. (A) A blocked valley tie channel with typically sinuous form entering the Lower Ok Tedi through a levee breach. (B) A cutoff tie channel. (C) A levee breach into a ﬂooded backswamp. (D) A levee breech into a ﬂooded area in an old scroll bar/swale complex. (E) Submerged former channel levees in a blocked valley lake. (F) A complex tie channel system at D’Albertis Junction. The large tie channel in the foreground connects the upper Fly and the Ok Tedi along a meander trace. The smaller tie channel above it connects a meander cutoff.

Geomorphology, Hydrology, and Climate of the Fly River System

19

keeps large areas of the ﬂoodplain inundated for long periods. The tie channels that originally linked ORWBs to the river are now ﬂooded for most of the time, and only their levees may be visible. These channels may offer a faster path for water ﬂowing from the river to the inundated area when water levels rise after a drought, but for most of the time, they behave as part of the lake. Again, they are likely to be hundreds to thousands of years old. Meander neck channels. A few meanders on the Fly and the Lower Ok Tedi have narrow channels that cut across the meander neck. Typically, they occupy a swale between adjacent scroll bars. Some provide a shortcut for canoe and small-boat trafﬁc. They may experience bidirectional ﬂow if they provide access to large areas of intermittently ﬂooded backswamps behind levees. Fossil tie channels. The ﬂoodplain contains traces of tie channels of all types that are no longer functional or only operate intermittently. Many of the blocked valley lakes show sinuous channel traces that are only visible at low water. Many disappear well before they reach the main river. Others appear to have been blocked by long-term levee development on the main channel. There are also some abandoned tie channels connecting the main river with meander cutoff oxbows. In at least one case, the cutoff tie channel has been abandoned when meander migration on the main river has intersected the oxbow and opened up a new path for water ﬂow (Rowland et al., 2005). Composite forms consisting of more than one tie channel type are fairly common. Both cutoff and tributary tie channels often have levee breaches that drain adjacent backswamps. Cutoff and tributary channels also sometimes capture each other and intersect. Very little work has been done on tie channel behavior. The most detailed information in the literature comes from Rowland et al. (2005) and Rowland and Dietrich (2006). These papers describe the forms, sedimentology, and evolution of meander cutoff tie channels, together with deposition rates in associated oxbow lakes. Rowland describes tie channel evolution as follows. Sediment-laden water from the main river enters the cutoff as a jet that extends a small delta outward into the cutoff water body and advects sediment laterally to its margins. This allows a set of submerged levees to build up. Over time, the levees extend upward and outward, gradually ﬁlling up the river end of the cutoff with sediment. Flow is usually sufﬁcient to keep the bed of the tie channels scoured, and ﬁner sediment deposited on the banks when ﬂow velocities are low tends to collapse into the channel during more active periods. Flow from the cutoff water body back to the river, as water levels

20

G. Pickup and A. R. Marshall

recede after a ﬂood, also plays a role in keeping the tie channel open since it scours sediment that may be deposited when water levels are similar in both the main river and the cutoff, and tie channel ﬂow velocities are low. An example of tie channel evolution from the Fly ﬂoodplain is shown in a series of satellite images in Fig. 1.9. Several factors govern the rate of tie channel extension into the cutoff oxbow lake and whether a tie channel is likely to be maintained. First, there is the amount of water ﬂowing into and out of the oxbow. This is a function of lake size and water level variations on the main river channel. More water will ﬂow into larger oxbows. Also, inﬂow and outﬂow will occur more frequently when river levels are more variable. Second, the sediment concentration in river water will affect delta growth rates since most of it will settle out as the jet enters the still water of the lake. Sediment concentration will also inﬂuence tie channel levee growth and backﬁlling of areas behind these levees. Tie channels not only grow outward into meander cutoffs but may also extend toward the river if meander migration moves the channel away from the original tie channel mouth (Rowland et al., 2005). Many of the long, straight tie channel sections on the Middle Fly ﬂoodplain have developed in this way. Rowland et al. (2005) have dated two active tie channels connecting the Fly with meander cutoffs. One was 347744 years old while the other proved to be at least 9287128 years old. This indicates how persistent tie channel features can be. It also suggests caution in interpreting ‘‘blocked’’ tie channels. They may, in fact, have been blocked for hundreds of years and long before mine sediments were added to the river. Rowland’s model can be applied to levee breach tie channels as well as those connecting meander cutoffs to the river. However, the tie channel levee complex and the delta will not develop to any great extent. This occurs because the ﬂoor of the levee breach remains close to the elevation of the backswamp behind the main channel levee. This limits the water surface slope that can develop between the ORWB and the river and keeps ﬂow velocities lower than in cutoff tie channels. It also prevents much of the sandy main riverbed material from getting into tie channel ﬂow. Most of the sediment reaching the ORWB will therefore be ﬁne main river wash load that settles out slowly and disperses more widely in the ORWB. Blocked valley tributary tie channels may also evolve differently from cutoff tie channels. Many of these systems have long reaches with welldeveloped levees that have built up over thousands of years in response to bidirectional ﬂow. However, because these channels are much longer than the cutoff tie channels, they do not tend to develop the steep water surface

Jan 1998

Feb 2000

Oct 2002

Feb 2004

Jul 2005

Figure 1.9: Evolution of a cutoff on the Middle Fly. Inﬁlling and tie channel development begins very quickly once the cutoff has occurred. Initially, both ends of the cutoff remain open to water from the river, but one arm of the former meander develops a tie channel. Eventually, a tie channel begins to develop in the other meander arm. Imagery includes LANDSAT TM, PACRIM 2000 SAR, and ENVISAT SAR.

22

G. Pickup and A. R. Marshall

slopes that occur in cutoff tie channels during rising and falling ﬂood stages. Deposition therefore tends to be dispersed over a longer distance, and the jetdeposited fans of cutoff oxbow lakes are less likely to occur. Many of the blocked valley tributaries also drain and ﬁll very large ORWBs and have sufﬁcient ﬂow to remain open. 1.4.3. Backwater Effects from the Strickland Backwater effects from the Strickland partly determine water levels in the lower reaches of the Middle Fly. The backwater effect extends upstream for about 200 km, and the gauging record at Manda (FLY16) closely follows that of Obo (FLY15) (Fig. 1.6). Lack of data makes it difﬁcult to determine the precise inﬂuence of the Strickland. Gauging stations on the Lower Fly at Ogwa (FLY22) and the lower Strickland (STR01 and more recently SG4) have only operated for short periods, and there are many missing records. Obo (FLY15) has a longer record, but gaps in the data frequently coincide with periods when the other gauges were operating. There is a common dataset for Obo (FLY15) and the lower Strickland (STR01) with few missing records for 1989–1992 (Fig. 1.10). STR01 is 73 km upstream of Everill Junction and below Massey Baker Junction, where the 16 14

Elevation (m)

12 10 8

STR01 (LOWER STRICKLAND) FLY15 (OBO)

6 4 2 0 1/1/88

1/1/89

1/1/90

1/1/91

1/1/92

1/1/93

1/1/94

Figure 1.10: Water levels on the lower Middle Fly and the lower Strickland. Strickland height datum is arbitrary.

Geomorphology, Hydrology, and Climate of the Fly River System

23

ﬂows from Lake Murray join the Strickland. There are no major inﬂow points between Station STR01 and Everill Junction, so the STR01 record should give a reasonable indication of ﬂows in the Strickland at Everill Junction. The datasets show both differences and similarities among stations. Some of the low water levels at Obo (FLY15) and on the lower Strickland (STR01) coincide, but this is not always the case. There is also some matching of ﬂood peaks but, again, not in every case. A very crude multiple-regression analysis for a short period shows that, after adjusting for lag, Stations FLY10 (Kuambit) and STR01 have a similar degree of inﬂuence on water levels at Obo (FLY15). However, these results do not cover a wide enough range of water levels to be deﬁnitive. They show that water surface elevations on the lower Middle Fly are a product of ﬂows coming down both the Fly and the Strickland. 1.4.4. Tidal Effects Very few data are available to show the inﬂuence of tidal cycles on Middle Fly water levels. A tide gauge operated at Daru, southwest of the Fly Delta for short periods in mid-1983, but no water level data are available in the OTML database for the lower Middle Fly. The JTides model (http:// www.arachnoid.com/JTides/) gives a good reproduction of the limited Daru Roads dataset. This model may be used to calculate tidal variation for other periods, including those when lower river system gauges were operating. The longest gauging record for the lower river system is for Station FLY15 at Obo. However, this station is affected by backwater from the Strickland as well as weak tidal effects. A Lower Fly station (FLY22) operated at Ogwa, just downstream of Everill Junction (Fig. 1.1) in late 1996 and early 1997. This station measures water levels for the combined ﬂows from the Middle Fly and the Strickland and gives a clearer picture of tidal inﬂuences. Ogwa (FLY22) water levels are compared with the modeled tidal variation in Fig. 1.11. Daily tidal variation can be seen on the lower graph. By the time tidal inﬂuence reaches Ogwa, there is only about 0.1-m variation at low water and less during medium ﬂows. At high ﬂow, daily tidal variation is almost indistinguishable. Longer-term tidal variation (upper graph) is also difﬁcult to identify, but during periods of steady ﬂow, it may possibly reach 0.3–0.5 m at Ogwa. However, some of this variation could be due to small ﬂoods coming downriver rather than tidal effects.

24

G. Pickup and A. R. Marshall

3

6

2

5

1

4

0

3

-1

2

-2 1/09/1996

1/10/1996

1/11/1996

1/12/1996

1/01/1997

Elevation at FLY22 (m)

Tide Elevation (m)

DARU TIDES (MODEL) FLY22 (OGWA )

1 1/02/1997

3

6

2

5

1

4

0

3

-1

2

-2 2/12/1996

9/12/1996

16/12/1996

23/12/1996

Elevation at FLY22 (m)

Tide Elevation (m)

DARU TIDES (MODEL) FLY22 (OGWA )

1 30/12/1996

Figure 1.11: Modeled tidal variation at Daru compared with water levels at Station FLY22 (Ogwa) on the Lower Fly just downstream of Everill Junction. The upper graph shows medium-term tidal variation. The lower graph compares FLY22 water levels with individual tidal cycles over a period of low-medium river ﬂow. The tide gauge datum is arbitrary.

Geomorphology, Hydrology, and Climate of the Fly River System

25

1.5. Geomorphic Changes Associated with Mining The Ok Tedi mine is located close to the headwaters of the Ok Tedi on what was formerly Mount Fubilan, although much of the mountain has now been removed (Fig. 1.12). The operation is open cut, producing an average of 83,000 tonnes/day of ore and a peak of 152,000 tonnes/day of overburden. About 80 million tonnes of waste rock and tailings are discharged into the Fly River system each year. The original mine design incorporated a tailings dam, and construction of a tailings dam began. However, the foundations collapsed in a landslide in 1984. An Interim Tailings Scheme contained tailings at start-up. Since this storage area is ﬁlled, tailings have been discharged directly to the river system (Ok Tedi Mining Ltd., 1999). The discharge of waste rock and tailings has greatly increased the sediment load of the river. While much of the waste rock consists of coarse material, about 60% of it breaks down during transport into ﬁner material in the sand–silt size range. While the gravels are not transported beyond the middle reaches of the Ok Tedi, the ﬁner material can pass into the Fly. Prior to mining, the natural load of the Ok Tedi at D’Albertis Junction was 3–5 Mt/ year with a similar amount coming from the Upper Fly (Higgins et al., 1987). Over the period 1985–2000, the load of the Ok Tedi has increased to an

Figure 1.12: The Ok Tedi mine and the Ok Mani in early 2004 (photo: Andrew Marshall).

26

G. Pickup and A. R. Marshall

average about 45 Mt/year. Some reaches of the river have experienced extensive deposition, raising the riverbed, adding material to levees and ﬂoodplains, and increasing water levels.

1.5.1. Mine Area Creeks Waste rock is disposed of in failing dumps on the north and south sides of the mine. These dumps feed a range of sediment sizes, from boulders to clay, into Sulphide Creek (which becomes the Ok Gilor, and then the Ok Mabiong in its lower section) and the Ok Mani. Mill tailings with a median size of about 75 mm are discharged directly into the Ok Mani downstream of the dump inputs. A summary of inputs is presented in Table 1.2. The Sulphide Creek system, on the north side of the mine, receives material from the northern dumps. This material accumulates in a steep fan at the dump toe, ﬁlling the whole valley of Sulphide Creek. The creek then enters a steep and narrow slot with waterfalls. All sediment delivered from upstream passes through this reach. Beyond the slot, a second valley-wide fan develops, extending down to the junction with the Ok Tedi. This system experienced a major landslide in 1989, adding about 125 Mt of hillslope material to the system. This is equivalent to about 80% of the waste rock dumped into Sulphide Creek between 1985 and 2000.

1.5.1.1. Ok Mani and the southern dump system The southern dumps feed Harvey Creek, a small, very steep northern tributary of the Ok Mani. The middle section of Harvey Creek is too steep to hold dumped material, so the dump toe ends abruptly in cliffs and there are several waterfalls. There is also substantial erosion and landsliding in this area, with cliff retreat of 25–50 m between November 1998 and November Table 1.2: Major sediment inputs to mine area creeks during mining operations. Location

Ok Mani Sulphide Creek

Quantity (Mt) 1985–2005 Waste rock

Tailings

Slides & erosion

292 163

490 –

190 125

Geomorphology, Hydrology, and Climate of the Fly River System

27

2000 (Marshall, 2001). Beyond the waterfalls, much of the dumped material accumulates in a large alluvial fan that extends out from Harvey Creek (Fig. 1.13). This fan blocks the valley of the Ok Mani, creating a small upstream lake, and pushes the channel of the Ok Mani against its southern valley wall, causing undercutting and erosion. Below the Harvey Creek fan, the valley of the Ok Mani is ﬁlled with deposited waste rock as far as the Ok Tedi junction (Fig. 1.14). Mill tailings enter the Ok Mani downstream of the Harvey Creek fan. These mix with the waste rock, adding to both stream discharge and valley ﬁll. Harvey Creek, which currently receives most of the waste rock, behaves as a typical alluvial fan in its lower reaches (Fig. 1.13). As material accumulates, the fan builds up, gradually increasing both its elevation and slope until the rate of transport equals the rate of removal from the fan base. The rate of removal is largely governed by the sediment transport capacity of the Ok Mani. At present, not all delivered material is being removed, so the fan base is building up, blocking the valley of the Ok Mani and diverting its channel toward its southern valley wall. This wall is presently being undercut and eroded intermittently. The Harvey Creek channel is also being pushed against its western valley wall, causing erosion.

Figure 1.13: The southern dumps and the Harvey Creek fan (photo: Barrie Bolton).

28

G. Pickup and A. R. Marshall

Figure 1.14: The Ok Mani and the middle Ok Tedi showing changes between 1979 and 2004. The behavior of the Ok Mani is typical of a river where the rate of sediment delivery exceeds transport capacity. The original concave long proﬁle is gradually building up and becoming less concave. The bulk of deposition is occurring in the middle reaches. There is less deposition at the lower end of the system because the Ok Tedi removes most, if not all, of the delivered load from the Ok Mani and sets a local base level. 1.5.1.2. Patterns of erosion and deposition on the Ok Mani Over the period 1996–2000, the Ok Mani experienced large variations in rates of sediment delivery from the southern dump area in response to variations in rainfall, valley wall erosion, and dumping rates (Marshall, 2001). In total, 172 Mt of waste rock, 50.7 Mt of valley wall material, and 112.2 Mt of tailings entered the system, and of this, 201.7 Mt reached the Ok Mani. Figure 1.15 shows maps of the spatial pattern of erosion and deposition from digital elevation models for March 1996, March 1997, November 1998,

Geomorphology, Hydrology, and Climate of the Fly River System

29

Figure 1.15: Patterns of erosion and deposition through time on the Ok Mani and the lower Harvey Creek fan. and November 2000. The pattern of erosion and deposition suggests that the Ok Mani may be divided into four reaches, each of which is separated by a reduction in valley width. These reaches are shown in Fig. 1.15 as: A–B: the toe of the Harvey Creek fan to the ﬁrst valley constriction. B–C: a relatively steep and narrow valley section of fairly constant width. C–D: a section in which the valley doubles in width and then narrows again. D–E: a reach of variable width ending at the junction with the Ok Tedi. Reach A–B shows patterns largely related to sediment delivery rates from the Harvey Creek fan. When delivery is lower, the toe of the fan experiences

30

G. Pickup and A. R. Marshall

scour, but there is rapid ﬁll at high delivery rates. There also seems to be erosion and deposition related to shifts in position of the main channels on the fan itself. Reach B–C has experienced consistent buildup over time, both during the dry period in 1997 and in the wetter period since then. This reach probably restricts supply to the reaches downstream but is likely to be increasing its slope and sediment transport capacity. Reach C–D has also experienced consistent buildup in elevation but at a lesser rate than upstream. The narrower section at the lower end of the reach probably also restricts outﬂow. Reach D–E is affected by the local base level imposed by the Ok Tedi. Deposition rates are lower than upstream reaches, and there was net erosion during 1997–1998. This shows that the Ok Tedi was able to remove delivered material and generate a wave of incision that progressed up the system as far as the ﬁrst main valley constriction. These patterns of behavior suggest that the rate of transport through each of the reaches is a function not only of slope but also of valley width, with more rapid buildup of material in the narrow sections. It also seems that rates and locations of incision during periods of lower sediment delivery from Harvey Creek are affected by the valley width. Projecting 1996–2000 deposition rates forward for 10 years suggests a further 40–50 m of deposition in the upper Ok Mani (Sections A–B and B–C in Fig. 1.15), decreasing to about 20 m in the middle section (C–D), and to about 10 m in the D–E section. Deposition rates will be lower close to the Ok Tedi junction, which sets the local base level. Channel forms will remain similar, and there will be signiﬁcant sidewall erosion, particularly in the upper reaches and in valley constrictions. The Harvey Creek fan will continue to grow upward and outward, and further signiﬁcant sidewall erosion can be expected, particularly during phases when fan growth pushes watercourses against valley walls. While the general trend in both Harvey Creek and the Ok Mani is for material to accumulate, progressively over time, several other types of behavior may be expected in future: The rate of accumulation of material will decrease over time as gradients become steeper, and sediment transport capacity increases. Marshall (2001) reports a decline in sediment retention from about 40% in 1992 to 30–35% currently. The retention rate will vary with the amount and characteristics of dumped material. Higher dumping rates will result in more sediment being retained. However, this may be offset by the removal of material during periods when

Geomorphology, Hydrology, and Climate of the Fly River System

31

the dumping rate is lower. As a rough guide, it may take 3–12 months for a large slug of dumped material to pass down the system to the Ok Tedi junction. Waves of erosion that may develop during times of limited dumping may take longer to propagate through the system. Rates of retention and sediment movement will vary with rainfall and stream discharge. During periods of low rainfall, dumped material may be held higher up the system, and lower reaches may scour in response to reduced supply of sediment from upstream. Nick points may also develop and travel upstream from the Ok Tedi junction. Seismic activity and landslides/debris ﬂows will affect sediment retention rates, especially in upper parts of the system. There will be considerable local variation in erosion and deposition within the system as channels migrate laterally and are diverted by localized accumulation, particularly close to the lower reaches of the Harvey Creek fan. Individual landslides/debris ﬂows may cause quite dramatic shifts in channel locations. The Harvey Creek fan, the Ok Mani, and the Sulphide Creek system show extreme forms of channel braiding, with up to 10 active channels operating, as waves of material pass through the system. Channels transporting high sediment loads are wide, shallow, and may experience rapid lateral shifting. Where sediment supply is reduced, ﬂows may be conﬁned to a single channel that may incise and develop some degree of bed armoring. The lake that is developing as the valley of the Ok Mani is blocked by the toe of the Harvey Creek fan will increase in depth and extend further upstream. There may also be periodic dam-break ﬂoods down the Ok Mani at times of high ﬂow.

1.5.1.3. Sulphide Creek/Ok Gilor/Ok Mabiong system and the northern dumps The northern dump system received much less material than the southern dumps between the 1996 and 2000 surveys with the bulk of dumping after 1998. Marshall (2001) cites 44 Mt of waste rock and 9.7 Mt of valley wall erosion. Of this, only 0.18 Mt of material remained in Sulphide Creek and 2.55 Mt in the Ok Mabiong and the Ok Gilor. The Sulphide Creek system is much steeper than the Ok Mani. In the reach upstream of the waterfall, at the toe of the dumps (F–G in Fig. 1.16), there is a mass of material, but it is basically in transit. Most would be removed if dumping ceased. Indeed, the F–G reach shows extensive sidewall erosion as well as little potential to store waste rock. The F–G reach ends in a narrow slot with a waterfall, and all delivered material goes

32

G. Pickup and A. R. Marshall

Figure 1.16: Patterns of erosion and deposition on the Sulphide Creek system. through. Beyond that, the system has a gradient similar to the upper Ok Mani that is maintained right down to the Ok Tedi junction. This reach has experienced buildup of sediment in the section marked G–H on Fig. 1.16, possibly related to several valley constrictions. Beyond that, the reach H–J shows erosion of previously deposited material down as far as the Ok Tedi junction. Unlike the Ok Mani, which has a concave long proﬁle, the Sulphide Creek system long proﬁle is virtually straight downstream of the waterfall. This means that there is little capacity to store waste rock over the long term. Delivered material will accumulate over time, but it is essentially in transit, and given time, or a reduction in the rate of dumping, it will be scoured out fairly rapidly. Base level reduction due to scour on the Ok Tedi will also be transmitted up the lower reaches of the Sulphide Creek system fairly rapidly, inducing erosion. These patterns of behavior are evident in the H–J reach. The future behavior of the Sulphide Creek system will very much depend on dumping rates. The area appears to be less stable than the valley of the Ok Mani (excluding Harvey Creek) and is relatively narrow. The upper catchment appears highly unstable where forest cover has been lost. High dumping rates would induce rapid deposition, especially in constricted reaches, even though material would ﬂush out fairly rapidly once dumping ceased. This would generate signiﬁcant sidewall erosion and landsliding. The retention rate would be less than that of the Ok Mani. The lower dumping rates that prevailed between 1996 and 2000 have allowed the system to erode over much of its length and remove previously deposited material. This trend will continue at low dumping rates.

Geomorphology, Hydrology, and Climate of the Fly River System

33

The waterfall at G (Fig. 1.16) is experiencing erosion and may become unstable. If so, the storage area upstream, beyond the dump toe, could be undermined.

1.5.2. Upper Ok Tedi 1.5.2.1. Deposition and river behavior The Upper Ok Tedi has a steep, concave long proﬁle with some local irregularities, often associated with gorge constrictions and openings. Steeper sections of the river that are conﬁned by valley walls have a single channel scoured to bedrock or lined with boulders while wider sections have braided channels with central and lateral bars. Flow is often critical or supercritical, and the river has a high sediment transport capacity. It is also capable of transporting virtually all sediment sizes delivered both from landslides and the mine. For most of its length, the river behaves as a source zone system, transporting a large quantity of delivered sediment. Some of this sediment breaks down into ﬁner particles during transit. Most of the deposition in the Upper Ok Tedi is actually bed material in transit and can be expected to pass downstream fairly quickly, especially in the steep upper reaches. For example, the zone between the Ok Mabiong junction and the Ok Mani junction experienced net erosion in the 1996–2000 period (Marshall, 2001) in response to reduced input from the northern dump system. The lower section of the long proﬁle is ﬂatter, and deposits here are likely to persist for longer, with some being long term. Patterns of erosion and deposition in the Upper Ok Tedi are highly variable (see Fig. 1.5 of Pickup and Cui, 2009). Close to the Ok Mani junction, bed levels have increased by 6–8 m since 1984, but there are periods of occasional scour of 2–3 m. There is also extensive lateral erosion and deposition as the channel shifts (Moi et al., 2001). Further down the system, especially in the steeper reaches, changes in bed level are minimal. Close to the end of the gorge, the gradient becomes ﬂatter and sediment begins to accumulate with about 4 m of deposition at its downstream end. Much of this deposition is in wider, ﬂatter reaches, and there are still sections where deposition is limited. The Upper Ok Tedi is unlikely to experience major changes in its current behavior in the period up to 2010. Projecting current trends forward suggests another 2–4 m of deposition in some areas, but periods of scour and ﬁll are likely, and the channel will continue to shift within the conﬁnes of its valley. Bed levels may rise locally, especially in lower reaches, or if there is a major

34

G. Pickup and A. R. Marshall

landslide in the mine area or elsewhere. There may also be some terrace and island formation in wider reaches.

1.5.3. Middle and Lower Ok Tedi 1.5.3.1. Deposition and river behavior Extensive deposition of mine-derived material has occurred in the Middle and Lower Ok Tedi. Minimum bed levels in this reach (Fig. 1.17) show little change over time, close to where the Lower Ok Tedi begins. By Ningerum, there is a bed-level rise of about 2 m, increasing to about 8 m at Konkonda, and remaining high (at about 5 m) down to D’Albertis Junction. The impact of dredging that began in March 1998 can also be seen with falls in bed level of 2–3 m at the TED25 and TED36 cross sections, which lie just upstream and downstream of the Ok Mart/Ok Tedi junction. The impact of dredging is examined in more detail below.

98 96 94 92 1980

18

TED20 NINGERUM

TED58 LOWER OK TEDI GORGE

68

Elevation (m)

Elevation (m)

100

1985

1990

1995

62

TED25 KONKONDA

1985

1990

1995

2000

TED36 NEAR D'ALBERTIS JUNCTION 16 14 Elevation (m)

Elevation (m)

64

60 1980

2000

16 14 12 10 8 6 1980

66

12 10 8 6

1985

1990

1995

2000

4 1980

1985

1990

1995

2000

Figure 1.17: Changes in minimum bed levels over time in measured cross sections from the lower end of the Ok Tedi gorge to D’Albertis Junction.

Geomorphology, Hydrology, and Climate of the Fly River System

35

There are three distinct patterns of river behavior within the Lower Ok Tedi: What was originally the armored zone (from the end of the Ok Tedi gorge to the former gravel front near Bige) is now a broad sand bed river with extensive central and lateral bars. Many of the former gravel islands are now partially or fully buried. In the upper section of this reach, the river is conﬁned by Pleistocene terrace material, so its capacity to widen is limited. Further downstream, where the valley opens out, the channel has become wider, and there is extensive overbank deposition. Much of the ﬂoodplain below the conﬁned reach is affected by rain forest dieback, although there has been some recovery since dredging commenced. The former gravel–sand transition zone (from the gravel front near Bige through the Konkonda meander to Atkamba) has remained as a single channel, but there has been major deposition on the channel bed. Water levels have risen and ﬂow has gone through the neck of the Konkonda meander, leaving overbank deposits. Lateral shifting of the channel has been limited by indurated Pleistocene terrace material, but water levels have been high enough to stress rain forest on all but the highest terrace remnants and to cause dieback in backswamps in lower areas between the terrace remnants. The ﬂoodplain below Atkamba contains few terrace remnants, and the river is able to migrate laterally. It is also affected by backwater from the Middle Fly. There has been major deposition on the channel bed, overbank deposition, lateral shifting of channel cross sections, and one meander cutoff. Some meander cutoffs that predate the mining operation are now being used as ﬂoodwater channels and are ﬁlling with sediment. A large bar has formed just upstream of D’Albertis Junction that is currently diverting ﬂow to the east across the ﬂoodplain into the Upper Fly. This reach shows extensive forest dieback.

1.5.3.2. Impact of dredging Future behavior of the Lower Ok Tedi largely depends on whether current dredging operation continues. Without dredging, the pre-1998 trends would continue, suggesting perhaps 2 m of additional deposition in the upper section near Ningerum, rising to 4–6 m in the riverbed at Konkonda and 2–4 m in the reach between Atkamba and D’Albertis Junction by 2010. This would mean: Channel widening throughout the reach upstream of the Konkonda meander. Extensive ﬂoodplain deposition in that reach but especially around Bige.

36

G. Pickup and A. R. Marshall

Possible cutoff of the Konkonda meander. Continued high and probably rising water levels (subject to no change in present rainfall conditions). Continuing forest dieback throughout the reach. Active meandering, ﬂoodplain deposition, and possible changes in river course between Atkamba and D’Albertis Junction. Evidence from cross section surveys suggests that dredging has slowed, halted, or begun to reverse some of these effects but not everywhere in the Lower Ok Tedi. In the dredge slot itself and upstream, bed levels have been reduced by about 1 m between the 1997 and 2000 cross section surveys. Immediately downstream of the dredge, there is a short zone of deposition. This is common in dredging operations and develops as material thrown into suspension by the dredging operation settles on the bed once more. In the former gravel–sand transition zone between Bige and Atkamba, average bed levels have been reduced by about 1 m between 1997 and 2000. Between D’Albertis Junction and Atkamba, bed levels have ﬂuctuated, and there is both erosion and deposition, with the deposition concentrated close to the junction with the Fly.

1.5.4. Middle Fly River 1.5.4.1. Major trends in river behavior Several physical changes have occurred in this reach of the river since waste rock and tailings have been discharged into the Upper Ok Tedi. These include: A gradual increase in bed elevation or partial constriction of the channel by enlarged point bars or other lateral deposits in the upper reaches of the Middle Fly. A substantial increase in levee elevation that either began, or has accelerated, with the increase in rainfall after the 1997 El Nin˜o drought. An increase in the rate of meander migration in some reaches of the river. Blocking of tie channels to some off-channel water bodies. Higher water levels in the period after the 1997 El Nin˜o drought. Backup of the lower reaches of the Upper Fly, resulting in higher water levels. Possible increase in the size of riverbed sediments. Very rapid weathering of material deposited in levees.

Geomorphology, Hydrology, and Climate of the Fly River System

37

The increase in bed elevation and partial channel constriction can be seen in Fig. 1.18. It is most severe in the upper reaches of the river with losses of more than 1,000 m2 of channel capacity at Kuambit and Niukamba, declining to about 750 m2 at Wygerin. Further downstream, the effect is minor, with no obvious changes at Mabaduan and only minor rates of decline at Bosset and Manda. This may be because backwater effects from the Strickland and tidal effects in the delta limit downstream transport and, therefore, prevent the coarser fractions of both mine-derived and natural sediment on the bed from reaching this far downstream. The location of, and downstream decline in the amount of channel ﬁlling/constriction suggests a mine-related impact. However, there is also substantial short-term variability about the central trend that may be due to variations in ﬂow, such as scour during ﬂoods and ﬁll during low ﬂows, and localized scour and ﬁll as large bars move through the system. There may also be some measurement error, given the difﬁculties of depth measurement using sounding weights in deep, fast-ﬂowing rivers.

1.5.4.2. Bed-level changes since 1997 On ﬁrst inspection, the long-term trends in channel area (Fig. 1.18) point to a mass of deposition moving gradually downstream. However, there is considerable local variation over time, and the deposition is not a clear-cut feature. The long-term cross section survey network is also very sparse, considering the complexity of the Middle Fly. A more densely spaced network of cross sections has been surveyed since 1997. Patterns are beginning to emerge (Fig. 1.19), and these may be used to make some rough estimates of future behavior on the Middle Fly. The upper graph of Fig. 1.19 shows the trend from December 1997 to March 2002. This period spans a time of river readjustment after the 1997 drought, a period of high ﬂows, and a period with closer-to-average ﬂows. Clear trends may be partially masked by this variability in hydrologic conditions. The pattern of bed-level change shows channel scour in the dredge slot gradually decreasing downstream until deposition begins in the backwater-affected reach of the Lower Ok Tedi. On the Middle Fly, there is deposition down to the start of the backwater zone close to the 180 km point and then substantial ﬂuctuations in bed level. If the survey data are divided into El Nin˜o/wet period conditions and closer-to-average ﬂow conditions, two different patterns emerge. Between 1997 and 1999, sediment supply to the Middle Fly was reduced by low ﬂows in the Ok Tedi and interruptions to mine operations. A wet period

Area (sq m)

FLY10 - KUAMBIT

FLY11 - NIUKAMBA

3500

3500

3000

3000

2500

2500

2000

2000

1500 1982

1986

1990

1994

1998

2002

1500 1982

1986

Year

Area (sq m)

FLY12 - WYGERIN

3000

3000

2500

2500

2000

2000

1990

1994

1998

2002

1500 1982

1986

Year

Area (sq m)

FLY16 - MANDA

1500

1500

1000

1000

1990

1990

1994

1998

2002

1998

2002

FLY14 - BOSSET 2000

1986

2002

Year

2000

500 1982

1998

FLY13 - MABADUAN 3500

1986

1994

Year

3500

1500 1982

1990

1994

Year

1998

2002

500 1982

1986

1990

1994

Year

Figure 1.18: Change in channel area with time for the Middle Fly River. Areas were calculated for a ﬁxed reference level set as the lowest observed bankfull stage in each cross section.

1997-2002

2 Middle Fly Backwater Zone

Lower Fly

Middle Fly Sand Zone

Ok Tedi

Aggradation(m)

1 Dredge Slot

0

-1

-2 0

100

200 300 400 1997-1999 El Nino/Wet Period

0

100

500

2

Aggradation(m)

1

0

-1

-2 200

300

400

500

400

500

1999-2000 Average Period 2

Aggradation (m)

1

0

-1

-2 0

100

200 300 Distance Upstream (km)

Figure 1.19: Bed-level changes in the Lower Ok Tedi and Middle Fly, 1997–2002.

40

G. Pickup and A. R. Marshall

immediately followed the El Nin˜o. In this phase, the middle graph of Fig. 1.19 shows that scour associated with dredging occurred at the upstream end of the Lower Ok Tedi, but there was deposition closer to D’Albertis Junction. Scour occurred on the upper Middle Fly as the sediment supply from upstream was reduced. The scoured material then moved down to the upper end of the backwater zone where it seems to have lodged. With a return to more average conditions (lower graph of Fig. 1.19), the main wave of deposition stays in the upper Middle Fly and ends close to the 300 km point. A second, smaller wave of deposition occurs in the lowest 100 km of the river. This deposition gradually increases downstream until it is truncated upstream of Everill Junction, presumably by periodic drawdown and increased velocities associated with low tide and low water on the Strickland.

1.5.4.3. Changes to banks and levees Channel cross section surveys carried out between 1997 and 2002 (Marshall, 2002) show that levee heights have increased over much of the Middle Fly, and many of the low points have been ﬁlled in. The process may have been occurring earlier, but there are not enough measured cross sections in the database to demonstrate this. While there is much variation between cross sections (Fig. 1.20), the regression lines suggest a rise of about 1 m over the whole of the Middle Fly from Everill Junction to D’Albertis Junction. This means that many levee breaches connecting the ﬂoodplain surface to the channel are now blocked. It is also likely that tie channels to off-channel water bodies are being blocked more frequently and for longer periods. Field inspection shows that much of the sediment is very fresh, and it often occurs unconformably on top of much older bank material. This implies an unusually rapid period of levee buildup. The increase in levee heights may be attributed to: Rising bed levels (and therefore ﬂoodwater levels) in the upper part of the Middle Fly, associated with the deposition of mine-related sediments. Higher water levels associated with the increase in rainfall since 1997. Increased suspended sediment concentrations, making more material available for deposition once ﬂow velocities decrease at the ﬂoodplain– channel boundary. There does not seem to be any downstream decline in the rate of levee construction. Indeed, it may be slightly greater in the backwater zone of the Middle Fly.

Geomorphology, Hydrology, and Climate of the Fly River System

41

20 Oct 1997 Dec 2000 Mar 2002

Elevation (m)

15

10

5

0 0

100

200

300

400

Distance from Everill Junction (km)

Figure 1.20: Levee top elevations between D’Albertis Junction and Ogwa, 1997–2002. 1.5.4.4. Changes in water levels The effect of deposition on water levels in the Middle Fly is shown in Fig. 1.21. At Kuambit, mean daily values have increased by about 3 m, with most of the changes occurring since 1992. Minimum values have also increased, as would be expected with rising bed levels. Some of the increase in water level is related to the higher rainfalls since 1997, but the upward trend was apparent before that. Fewer data are available for Manda at the upstream end of the backwater zone, but a weak upward trend in water level seems to be present. Water levels alone make it difﬁcult to say when this trend began, and there is a case for relating it to the higher rainfalls after 1997. Water levels on the Upper Fly River have been high since the 1997 drought and associated with dieback of rain forest along the ﬂoodplain and backed up tributary drainage basins. The dieback is extensive along the reach between D’Albertis Junction and Kiunga, where the river is subject to backwater from the Middle Fly. There has been some debate over whether high water levels in this part of the river are the result of downstream deposition of mining waste, higher rainfalls since the 1997 drought, or both. Cross section surveys show that there is little deposition in the Upper Fly except very close to D’Albertis

42

G. Pickup and A. R. Marshall

20

Elevation (m)

15

10

5

0 1/1/90

1/1/92

1/1/94

1/1/96

FLY10 (KUAMBIT)

1/1/98

1/1/00

1/1/02

FLY16 (MANDA) FLY15 (OBO)

Figure 1.21: Daily water levels for Middle Fly gauging stations. Junction, so the dieback is not the result of local changes in the channel in the reach upstream toward Kiunga. The record of water levels for the Upper Fly at Kiunga (Fig. 1.22) provides useful evidence suggesting that: The consistently high water levels since 1997 have no parallel in the record of the last 30 years at Kiunga. Not only has the average level remained high, but also there have been fewer periods of low water than in the past. The 1997 drought and the period of higher minimum water levels that followed it bear some similarities to the drought of 1972. However, after 1972, although water levels were high, the river fell to low levels much more frequently. The density of points at high water levels seems to increase from about 1994 or 1995. This precedes the change in rainfall, strongly suggesting that the underlying cause of the high levels is not a climatic shift. 1.5.4.5. Deposition and tie channel blocking Tie channels are long-term features in the landscape and largely selfsustaining. Indeed, virtually all recent meander cutoffs maintain a tie channel

Geomorphology, Hydrology, and Climate of the Fly River System

43

26 24

Elevation (m)

22 20 18 16 14 12 10 1970

1975

1980

1985

1990

1995

2000

Figure 1.22: Water levels on the Upper Fly at Kiunga since 1985. linking them with the main river. Abandoned cutoff tie channels are not widespread and usually occur where meander migration creates a new cutoff containing the river end of the original tie channel. Even then, the original tie channel may continue to operate and may link the original cutoff with the new one. The lack of abandoned tie channels shows just how effective these channels are at keeping clear of deposited sediment. While tie channels occasionally become blocked, there is some debate over whether tie channels are becoming blocked more frequently. The jet model of Rowland et al. (2005) provides some clues on how tie channels become blocked. Figure 1.23 shows an abandoned tie channel on the lower Middle Fly ﬂoodplain. All the morphological characteristics of the channel survive, including levees that decrease in height and a channel cross section that becomes smaller away from the main river. The channel even remains connected to the oxbow lake, although it now exits through a ﬂoating grass mat. What seems to have happened in this case is that a tributary, originally connected to the west bank of the channel prior to cutoff, has now opened a route across the cutoff and down the swale behind the former main channel east bank levee. This has also become the tie channel draining the oxbow lake in the meander cutoff as well as the tributary outﬂow channel. The southern tie channel has not been kept open due to lack of ﬂow and is now choked with vegetation and sediment.

44

G. Pickup and A. R. Marshall

Figure 1.23: An abandoned tie channel on the Middle Fly. The tie channel has ceased to operate because ﬂow from a blocked valley tributary has entered the cutoff and overﬂowed down the swale behind the old levee on the right-hand side of the cutoff. This has opened a path to the river and created a new tie channel. A delta has formed where the tributary crosses the cutoff.

Geomorphology, Hydrology, and Climate of the Fly River System

45

While there is plenty of evidence that deposition rates in cutoffs and other ORWBs have increased, reports of tie channel blocking are less clear. Some tie channels have become blocked to small boats but have subsequently cleared. This may be a cyclic process whereby sediment is deposited in tie channels during periods of extensive ﬂooding (e.g., the 2000 La Nin˜a event) when both river and ORWB water levels remain high and water surface slopes between the two are low. The sediment then clears when river heights become more variable and tie channel water surface slopes and ﬂow velocities become greater. Perhaps the best known case of a blocked tie channel is the one at Pangua on the lower Middle Fly, which seems to have become blocked during the extended ﬂooding of 1999–2000. Figure 1.24 shows how this may have occurred. In the 1998 image, the tie channel can be seen connecting the southern end of the cutoff closest to the channel. During 1999 or 2000, much of this area remained inundated, and there was 0.5–1 m of deposition on the main channel levees. It is likely that this deposition also blocked the tie channel entrance, which was not regularly scoured out because of continuously high water. High water levels and backwater from the Strickland forced some of the ﬂoodwater out of the Middle Fly, and a major ﬂood path developed through backswamps on the western edge of the ﬂoodplain over a distance of about 30 km. Flow down this ﬂood path was far greater than the amount the Pangua tie channel could handle even if it had remained open. Instead, a whole series of breaches were cut through the levee, and some of these have remained open since then, replacing the original tie channel. More evidence that the principal cause of natural cutoff tie channel blockage is lack of ﬂow comes from the maps of tie channels and ORWBs in Pickup (2006). Most upper Middle Fly meander cutoffs have tie channels linking them to the main river. However, in the backwater zone of the Middle Fly, water level variation has greatly decreased as Strickland water levels have risen over several thousand years. The area now remains ﬂooded for much of the time. Many of the backwater zone cutoffs now have no obvious active tie channels. The major difference between the upper and lower Middle Fly is water level variability (Fig. 1.22). Tie channel inﬂows and outﬂows have therefore decreased to the point that active channels can no longer be maintained in many cases. The foregoing discussion shows that tie channels are self-forming and, once developed, are a largely self-sustaining feature of the landscape. While they may be subject to occasional blocking by sediment during high ﬂows, they can also reopen when ﬂuctuating river levels create large differences between the main river and the connected ORWB. Rapid, sediment-scouring

46

G. Pickup and A. R. Marshall

Figure 1.24: The Pangua tie channel. The upper PACRIM 2000 Synthetic Aperture Radar image shows the major ﬂoodway that developed through backswamps on the western margin of the Middle Fly ﬂoodplain. The lower two images show how a meander cutoff tie channel had been replaced by broader multiple-levee breaches by February 2000.

Geomorphology, Hydrology, and Climate of the Fly River System

47

ﬂows then occur in both directions along the tie channel. Tie channels only seem to become permanently blocked under natural conditions when their water supply is removed or reduced. Many of the tie channels that are connected to large ORWBs are unlikely to become blocked because enough water ﬂows through them to maintain an opening. Connections with smaller ORWBs are more likely to be subject to intermittent blockages because of increased sediment concentration in the Fly. Levee breach channels may be particularly affected because of recently increased levee deposition. However, they may be replaced by outﬂows through other low points in the levee. The main threat to all upper Middle Fly tie channels comes from rising water levels in response to deposition on the bed of the main river channel. Until the early 1990s, water levels on the upper Middle Fly ﬂuctuated rapidly and covered a range of about 12 m. From about 1999 onward, the range has decreased and, for most of the time, is 6 m or less. This means that there is less scouring of tie channels because ﬂow velocities are reduced. Also, local base levels, which are set by main river ﬂow heights and govern tie channel depth, have risen. Taken together, these two processes will mean smaller and shallower tie channels. It is also likely to occur along the whole length of cutoff tie channels. The larger tributary tie channels will be less affected but could show some shallowing close to the river. Predictions of deposition and water level rise (see Pickup and Cui, 2009) show that a wave of deposition and water level increase will move down the Middle Fly over the next 40–50 years. The reduction in water level variation at Kuambit will affect much of the upper Middle Fly, although it will decrease in intensity downstream. The backwater zone of the Middle Fly will also experience higher water levels, but ﬂow variability will still be largely determined by the Strickland. Overall, this means less scouring potential in tie channels but more overbank connections between the Fly and its ﬂoodplain.

REFERENCES Barr Engineering Company (1995). Investigation of the effect of mine tailings on meander rate of the Fly River. Report to Ok Tedi Mining Ltd. Blake, D. H., & Ollier, C. D. (1971). Alluvial plains of the Fly River, Papua. Zeitschrift fur Geomorphologie, Supplementband, 12, 1–17. Blong, R. J. (1991). The magnitude and frequency of large landslides in the Ok Tedi catchment. Report to Ok Tedi Mining Ltd. Day, G. M., Dietrich, W. E., Apte, S. C., Batley, G. E., & Markham, A. J. (1993). The fate of mine-derived sediments deposited on the middle Fly River

48

G. Pickup and A. R. Marshall

ﬂood-plain of Papua New Guinea. In: R. J. Allen, & J. O. Nriaugu (Eds). International Conference on Heavy Metals in the Environment, Vol. 1. CEP Consultants Ltd, Edinburgh, pp. 423–426. Dietrich, W. E. (2000). Quarterly and fourth monthly report – Middle Fly Sedimentation Project. Ok Tedi Mining Ltd. and University of California, Berkeley. Dietrich, W. E., & Day, G. (2004). Rates and patterns of ﬂoodplain sedimentation, middle Fly ﬂoodplain (1985–1994). Presentation at the Sediment Transport Model Workshop, Cairns, Australia, January 2004. Environment Department, OK Tedi Mining Ltd. Dietrich, W. E., Day, G., & Parker, G. (1999). The Fly River, Papua New Guinea: Inferences about river dynamics, ﬂoodplain sedimentation and fate of sediment. In: A. Miller, & A. Gupta (Eds). Varieties of Fluvial Form. Wiley, New York, pp. 345–376. Harris, P. T., Baker, E. K., Cole, A. R., & Short, A. S. (1993). A preliminary study of sedimentation in the tidally dominated Fly River Delta, Gulf of Papua. Continental Shelf Research, 13, 441–472. Higgins, R. J., Pickup, G., & Cloke, P. S. (1987). Estimating the transport and deposition of mining waste at Ok Tedi. In: C. R. Thorne, J. C. Bathurst, & R. D. Hey (Eds). Sediment Transport in Gravel Bed Rivers. Wiley, Chichester, pp. 949–976. Lofﬂer, E. (1977). Geomorphology of Papua New Guinea. ANU Press, Canberra. Marshall, A. R. (2001). Upper Ok Tedi sediment storage analysis – November 2000. Environment Department, Ok Tedi Mining Ltd. Marshall, A. R. (2002). Ok Tedi–Middle Fly bed aggradation survey. Epoch 6 – March 2002. Environment Department, Ok Tedi Mining Ltd. (and associated spreadsheets). McAlpine, J. R., Keig, G., & Short, K. (1975). Climatic tables for Papua New Guinea. CSIRO Division of Land Use Research Technical Paper 37. Moi, A. S., Dremdap, T., & Simbina, P. (2001). APL compliance and additional monitoring program. 1999/2000 Annual Report. Appendix 1 – Hydrology. Report ENV01-02, Environment Department, Ok Tedi Mining Ltd. Ok Tedi Mining Ltd. (1999). Community and environment programs. Paijmans, K., Blake, D. H., Bleeker, P., & McAlpine, J. R. (1971). Land resources of the Morehead-Kiunga area, territory of Papua New Guinea. CSIRO Land Research Series 29. Pickup, G. (1984). Geomorphology of tropical rivers. 1. Landforms, hydrology and sedimentation on the Fly and lower Purari, Papua New Guinea. Catena Supplement, 5, 19–41. Pickup, G. (2006). Comments on a possible program for maintaining tie channel openings by hydraulic cleaning. Report to Environment Section, Ok Tedi Mining Ltd. Pickup, G. (2009). Floodplain inundation modeling and forecasting for the Middle Fly. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 291–318.

Geomorphology, Hydrology, and Climate of the Fly River System

49

Pickup, G., & Cui, Y. (2009). Modeling the impact of tailings and waste rock disposal on the Fly River system. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 257–289. Pickup, G., Higgins, R. J., & Warner, R. F. (1979). Impact of Waste Rock Disposal from the Proposed Ok Tedi Mine on the Fly River and its Tributaries. Department of Minerals and Energy and Ofﬁce of Environment and Conservation, Papua New Guinea. Rowland, J. C., & Dietrich, W. E. (2006). The evolution of a tie channel. In: G. Parker & M. Garcia (Eds). River Coastal and Estuarine Morphodynamics: RCEM 2005, Vol. 2. Taylor & Francis/Balkema, London, pp. 725–736. In: Proceedings of the 4th IAHR Symposium on River, Coastal and Estuarine Morphodynamics. Rowland, J. C., Lepper, K., Dietrich, W. E., Wilson, C., & Sheldon, R. (2005). Tie channel sedimentation rates, oxbow formation age, and channel migration rate from optically stimulated luminescence (OSL) analysis of ﬂoodplain deposits. Earth Surface Processes and Landforms, 30(9), 1161–1179. Wolanski, E., & Eagle, A. M. (1991). Oceanography and sediment transport, Fly River estuary and Gulf of Papua. 10th Australasian Conference on Coastal and Ocean Engineering, Auckland, New Zealand.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00402-3

Chapter 2

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System Barrie R. Bolton, Jesse L. Pile and Henry Kundapen Environment Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, W.P., Papua New Guinea

2.1. Introduction In May 1984 mining began at the Mount Fubilan copper and gold mine located in the remote Star Mountains of western Papua New Guinea (PNG) (Fig. 2.1). The mine, which is owned and operated by Ok Tedi Mining Limited (OTML), commenced operations with a 24,000 tpd gold plant that has since been expanded to a 90,000 tpd copper plant. Waste materials from mining operations, at present comprising approximately 55 Mt of overburden and 30 Mt of tailings, are discharged annually into the headwaters of mine area creeks where erosion and sediment transport processes carry signiﬁcant proportions of the total waste produced into the Ok Tedi and Fly River system. The potential for adverse impacts to the ﬂuvial environments downstream of the mine resulting from these waste materials has been of concern since before the start of mining. These impacts include the release of copper and other metals from these sediments and their toxicity to aquatic organisms (Apte, 2009) and riverbed aggradation and its potential to cause an increase in the frequency and duration of ﬂoodplain inundation (Pickup, and Pickup and Cui, 2009). In response to these concerns OTML has commissioned numerous studies and ongoing monitoring of a range

Corresponding author. Tel.: +61-3-9769-6099; Fax: +61-3-9769-6099;

E-mail: [email protected] (B.R. Bolton).

Figure 2.1: Maps showing the location of the Fly River System in western PNG (A) and drainage around the Mount Fubilan (Ok Tedi) mine (B).

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

53

of environmental parameters throughout the river system to assess the extent and nature of these impacts. This chapter examines the temporal and spatial impact of these mine waste materials on the texture, geochemistry, and mineralogy of the sediments deposited in the channel and on the banks of the downstream riverine environment, from the mine to the Fly River delta. We also present information on the distribution of heavy minerals in the river system and discuss the implications of these data for the potential for the development of environmental issues such as acid rock drainage (ARD).

2.2. Texture of River Sediments This section describes how bed and bank sediment textural characteristics have changed in the Ok Tedi and Fly River system as a result of mining at Mount Fubilan, in terms of sediment particle size, sediment type, and the degree of sorting that has occurred.

2.2.1. Before Mining An extensive investigation of river behavior and sediment characteristics was undertaken prior to mining and is reported by Pickup et al. (1979), Pickup (1984), and Higgins et al. (1987). The following description draws heavily on these works as well as the detailed description of the geomorphology, hydrology, and climate of the Fly River system provided elsewhere in this volume (Pickup and Marshall, 2009). Before mining began in 1984, the steep, fast-ﬂowing streams draining Mount Fubilan at the headwaters of the Ok Tedi (Fig. 2.1) displayed many of the features typical of source-area streams including narrow bed-rock and coarse-grained channels lying within steep valley walls with the river largely occupying the entire valley ﬂoor at peak discharges (Fig. 2.2A) (Pickup and Marshall, 2009). These streams, known locally as the mine area creeks, comprise Harvey Creek and Ok Mani, to the south of the mine and Sulﬁde Creek, Ok Gilor, and Ok Mabiong to the north (Fig. 2.1). Together they form part of a ‘‘supply’’ zone in which mine-derived materials, along with natural sediments, are delivered to the river system through a range of processes including massive landslides and debris ﬂows. One such landslide that occurred about 8,800 BP from the Hindenburg Wall, a limestone escarpment

54 B. R. Bolton et al.

A

C

E

B

D

F

Figure 2.2: (A) Upper Ok Tedi at Bukrumdaing. Note the steep valley walls and coarse-bedded river channel. (B) Aerial view looking west up Ok Mani toward the mine (upper right). (C) Poorly sorted, massive debris ﬂow deposits typical of upper reaches of mine area creeks. Section seen here is just over 1 m thick. (D) Gravel-bedded chute typical of armored zone of Ok Tedi about 70 km downstream of the mine. (E) Alternating massive, poorly sorted gravels and cross-bedded sands typical of the gravel–sand transition zone of upper Ok Tedi after the start of mining. Note the presence of dark-colored sulﬁde-rich laminae partially oxidized in places to form reddish colored secondary reaction products. (F) Lateral gravel-sand bars in upper Ok Tedi located approximately 45 km downstream of the mine. Note the gravel deposits adjacent to the active channel and more distal sand deposits characterized in part by elongate, darker colored heavy mineral segregations.

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

55

300–1,000 m in height, resulted in the addition of an estimated 7 km3 of material into the headwaters of the Ok Tedi system (Blong, 1991). The transport capacities of the streams in this part of the system typically exceeded supply and therefore stream beds were generally either covered by a thin bed of sediment ranging in size from boulders to silts and clays, or clean bed rock was exposed. The grain size, sediment type, and sorting characteristics of material from this zone are well illustrated in Figs. 2.3, 2.4, and 2.5. The supply zone is represented on Fig. 2.3A by samples collected within 15 km of the mine and has a median (d50) particle size of less than 1 mm but a d90 particle size of almost 100 mm. Sediment types present in this zone include medium sand (up to 50%) and lesser amounts of gravel (up to 35%) (Fig. 2.3A). Figure 2.5(A–E) illustrates the bimodal distribution of these sediments with ﬁner material typically ﬁlling the interstices of accompanying gravels. Furthermore, sediments in this zone, which mostly occurred as a series of channel bar deposits, typically ﬂuctuated widely in grain size distribution with time, as shown in Fig. 2.5A and E, as material was transported downstream. The supply reach passed downstream into a zone referred to by Pickup and Marshall (2009) as the ‘‘armored’’ zone. The moderate-to-well-sorted coarse gravels characteristic of this zone formed a distinctive protective pavement or ‘‘armor’’ over ﬁner grained sands (see Bar 2, Fig. 2.5A, B, and C). In the Ok Tedi/Fly system before mining began, the armored zone was located about 50–110 km downstream of the mine. Further downstream the river entered a reach characterized by a mixed sand/gravel bed referred to by Pickup (2009) as the ‘‘gravel–sand transition zone.’’ Located 110–160 km downstream of the mine, grain size in this zone ﬂuctuated widely (Fig. 2.3A), as did the degree of sorting (e.g., Fig. 2.5D). At some sampling localities in this zone sediments displayed bimodal size distribution patterns (e.g., Fig. 2.5A, Bars 4, 5, and 6), while elsewhere sediments were well sorted and composed of either sand (e.g., Fig. 2.5D, site 64) or coarse gravels, such as at site 6 (Fig. 2.5A), in the lower part of the Ok Tedi. Figure 2.5A shows grain size distribution curves for samples collected in June 1981 for the supply, armored, and gravel–sand transition zones. This shows that generally there is very little sediment in the size range 1–10 mm at sites within these reaches, as observed previously by others (Maunsell and Partners, 1982; Higgins et al., 1987). Sediments in this size range are infrequent in most ﬂuvial deposits and probably reﬂect a tendency for gravels, which may contain a range of minerals, to break down into sizes more closely related to the size of the individual constituent minerals (Pickup, pers. comm.). Furthermore, Fig. 2.5E shows the distribution of

56

B. R. Bolton et al.

1000.00

D'albertis Jct.

Everill Jct.

Particle size, mm

100.00 d90

10.00

1.00

0.10 d50 0.01 0

100

200

300

400

500

600

700

800

900

800

900

Distance downstream from mine, km

A

1000.00

D'albertis Jct.

Everill Jct.

Particle size, mm

100.00

10.00 d90

1.00

0.10 d50 0.01 0 B

100

200

300

400

500

600

700

Distance downstream from mine, km

Figure 2.3: Downstream variation in size of riverbed material below the Mount Fubilan mine. (A) Pre-mine (Pickup et al., 1979; Maunsell and Partners, 1982; OTML, 1986) and (B) for period 2001–2003, after the start of mining operations (OTML, 2004). Symbols d50 and d90 describe median grain size and the size for which 50 and 90% of the bed is ﬁner, respectively.

Figure 2.4: Downstream variation in grain size distribution. (A) Bed samples collected before the start of mining operations (Pickup et al., 1979; Maunsell and Partners, 1982; OTML, 1986); (B) bed samples collected in 2001–2003 during mining operations (OTML, 2004); (C) bank samples collected before mining (Pickup et al., 1979); (D) bank samples collected in 2003 (OTML, 2004). Vertical arrows at top of plots mark conﬂuence with major tributaries. Note change in horizontal scale.

58

B. R. Bolton et al.

100 June 1981 ( Pre-mine)

90 80 70 60 50 40 30 20 10

70 60 50 40 30 20

0 0.10

1.00 10.00 Particle Size (mm)

A Bar 1

Bar 2

Bar 3

Bar 4

100.00

0.01

1000.00

0.10

1.00 10.00 Particle Size (mm)

E Bar 5

Bar 6

Bar 1

100

100

90

90

80

80

70

70 Percent finer

Percent Passing Sieve Size (%)

80

10

0 0.01

60 50 40 30

Bar 2

Bar 3

100.00

Bar 4

Bar 5

1000.00 Bar 6

Upper Middle Fly (Upstream of D'albertis Junction: 0 - 38 km)

60 50 40 30

20

20

10

Sand Only (< 15 mm) - June 1981

0 0.01

0.10

1.00

10.00

100.00

10

Bar 1

Bar 2

Bar 3

0 0.01

1000.00

0.10

F

Particle Size (mm)

B

Bar 4

Bar 5

S1 ( 0 km) S5 (24 km)

Bar 6

1.00 10.00 Particle Size (mm) S2 (6 km) S3 (12 km) S6 (33 km) S7 (38 km)

100.00 S4 (17 km)

100

100 90

Coarse Only (>15 mm) - June 1981 80

80 70 Percent finer

Percent Passing Sieve Size (%)

January 1984 (Pre-mine)

90 Percent Passing Sieve Size (%)

Percent Passing Sieve Size (%)

100

60 50 40

60 Upper Middle Fly (44 - 225 km)

40

30 20

20 10 0 0.01

0.10

1.00 10.00 Particle Size (mm)

C Bar 2

Bar 3

0 0.01

1000.00

0.10

G

Bar 4

Bar 5

S9 (51 km) S17 (152 km)

Bar 6

100

100

90

90

80

80

70

70 Percent finer

Percent Passing Sieve Size (%)

Bar 1

100.00

60 50 40

1.00 10.00 Particle Size (mm) S11 (70 km) T10 (107 km) S18 (162 km) S22 (225 km)

60 50 Lower Middle Fly 'Backwater Zone' (308 - 456 km)

40 30

30

100.00 S15 (127 km)

20

20 Lower Ok Tedi - 1979

10 0 0.01

0.10

D 41 64

42.5 64.5

1.00 Particle Size (mm) 44

46

10.00

10 0 0.01

100.00 H

54.5

62.5

S28 (308km) S33 (386km)

0.10

1.00 Particle Size (mm) S29 (323 km) S34 (400 km)

10.00

S30 (335km) S37 (456km)

100.00 S32 (370 km)

Figure 2.5: Bed material size distributions for Ok Tedi and Fly River before the start of mining (Pickup et al., 1979; Maunsell and Partners, 1982; OTML, 1986).

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

59

grain size for the same six bars in January 1984, again before the start of mining. This plot shows a signiﬁcant decrease in the amount of ﬁner grained material (up to 10%) and near-uniform median sizes through the entire length of the upper Ok Tedi. The gravel–sand transition zone passed in turn downstream into a long reach dominated by a sand bed – the ‘‘sand zone’’ of Pickup (1984). Premine, the sand zone extended from the midway down the Ok Tedi to about midway down the Fly River. Sediments deposited on the bed of the river along this reach are dominantly composed of medium–ﬁne sands with lesser amounts of silts and clays (Figs. 2.3A, 2.4A, and 2.5D). Figure 2.6 shows in detail the variation in grain size of both bed and bank material from the Fly River between Kiunga and Everill Junction together with regression analyses of grain size against distance for pre-mine data collected by Pickup et al. (1979). In general, grain size decreases in the bed with increasing distance downstream (as evidenced by the slope of the regression line), while the correlation coefﬁcient of 0.60 for the d50 highlights the signiﬁcant amount of variability in the data. It also shows local areas of coarser-grained sediment deposition, such as at 335 km, where the upper 10% of bed material approaches 0.6 mm or coarse sand when very little material of this size was found downstream of D’Albertis Junction at 41 km (Pickup et al., 1979). Figure 2.6A also shows a signiﬁcant decrease in grain size, as reﬂected in both the d50 and d90 curves, between about 270 and 415 km, with the exception of the coarse-grained sampled at 335 km, noted above. The sorting characteristics of sediments deposited in the sand zone are illustrated in Fig. 2.5G. This ﬁgure indicates that sediments are in general well sorted (as evidenced by the steepness of the distribution curve), differ little in the degree of sorting downstream, and vary somewhat in median grain size from 0.09 to 0.3 mm, as shown by the spread of distribution curves along the 50% ﬁner horizontal line. Still further downstream the median grain size abruptly decreases to the lowest values seen anywhere in the system, particularly between about 430 and 600 km downstream of the mine (280–452 km downstream of Kiunga) (Figs. 2.3A and 2.6A). Sediments in this reach are composed of a mixture of well-sorted ﬁne sands and ﬁner-grained silts and clays (Fig. 2.4A). Grain size distribution plots for this reach show two distinct sets of curves representing the two size classes and suggest the presence of alternating sand and silt/clay beds in channel deposits which formed in response to changing ﬂow conditions (Fig. 2.5H). Pickup et al. (1979), Pickup (1984), and Higgins et al. (1987) referred to this zone as the ‘‘backwater zone.’’ Textural data in the pre-mine period for the Fly River downstream of Everill Junction are rare. Spencer (1974), in a study mainly devoted to

0.70

0.35 Bank Material - 1979 0.30

y = -0.0005x + 0.293 R2 = 0.6076 d90

0.40 0.30 0.20 0.10

d90

0.25 0.20

d90

y = -0.0002x + 0.1601 R2 = 0.1801

d50

y = -0.0001x + 0.0682 R2 = 0.3613

0.15 0.10 0.05

d50

d50

0.00

0.00 0

100

A

200

300

400

500

0

100

B

Distance downstream from Kiunga, km

0.70

200

300

d90 d90

0.50

d50

Bank Material - 2003

y = -0.0007x + 0.4221 R2 = 0.5151 y = -0.0004x + 0.2337 R2 = 0.6325

0.40 0.30 0.20 0.10

0.30 Particle size, mm

0.60 Particle size, mm

500

0.35 Bed Material - 2003

d90

0.25 0.20

y = -0.0002x + 0.153 R2 = 0.2191

d50

y = -0.0002x + 0.0952 R2 = 0.2945

0.10 0.05

0.00

d90

0.15

d50

d50

0.00 0

C

400

Distance downstream from Kiunga, km

100

200

300

400

Distance downstream from Kiunga, km

500

0 D

100

200

300

400

500

Distance downsream from Kiunga, km

Figure 2.6: Downstream variation in particle size from Kiunga on the Upper Fly River to Everill Junction. (A) Pre-mine bed material from Fly River (Pickup et al., 1979); (B) pre-mine bank material from Fly River (Pickup et al., 1979); (C) bed material from Fly River collected in May 2003 (OTML, 2003, 2004); (D) bank material from Fly River collected in May 2003 (OTML, 2003b, 2004). Symbols d50 and d90 designate median grain size and the size for which 50% and 90% of the bed is ﬁner, respectively.

B. R. Bolton et al.

Particle size, mm

d50

0.50

Bed Material - 1979

60

0.60

y = -0.0007x + 0.4831 R2 = 0.5895

Particle size, mm

d90

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

61

understanding the sedimentation history of the Fly River delta, reported on the results of 16 riverbed samples taken from the Fly and Strickland Rivers. Spencer noted that the single sample from the bed of the Strickland was slightly coarser (ﬁne to very ﬁne sand) than those samples collected from the bed of the lower Middle Fly (upstream of Everill Junction) and that grain size initially increased in the Fly downstream of Everill Junction in response to the larger sediment load in the Strickland before ﬁning again toward the delta (Spencer, 1974) (Table 2.4). Similar ﬁndings were also reported by (Everett and Associates, 1980) based on four samples collected from the riverbed downstream of Everill Junction. We know of no grain size data for the banks of the Ok Tedi; however, Pickup et al. (1979) provide a comprehensive data set for sediments deposited on the levees of the Fly River between Kiunga and Everill Junction. These workers noted that the grain size of material collected from the top of the river bank is considerably smaller than from the adjacent channel bed and that grain size generally decreases downstream (Fig. 2.6B). They also noted that bank top material shows marked variability in size (d50 of 0.01–0.1 mm; d90 of 0.06–0.15 mm) through the ﬁrst 250 km then progressively ﬁnes but with much less variability (d50 of 0.01–0.02 mm; d90 of 0.06–0.15) over the remaining 200 km down to Everill Junction. Figure 2.4C shows the sediment types for the bank material from Kiunga to Everill Junction before mining began (Pickup et al., 1979). This plot shows that in the upper 270 km the bank is mostly composed of clays and silts (26–91%) alternating with ﬁne sand (9–74%). Coarser material occurs only spasmodically and mainly comprises medium sand which, as seen on this plot, was recorded over the ﬁrst 17 km and then at locations 51 km (18%), 103 km (4%), 127 km (9%), 203 km (6% medium sand and 4% coarse sand), and 225 km (27%) downstream of Kiunga. From 270 km to Everill Junction (452 km) clays and silts dominate with ﬁne sand relatively steady at around 15%. 2.2.2. After the Start of Mining The grain size and grain size distribution of sediments in both the channel bed and on the levees of the Ok Tedi and Fly River is a function of the following: (1) source material characteristics; (2) distance from source; and (3) river energy. In general terms, faster ﬂowing waters transport all sizes but the ﬁner material passes through as wash load and is rarely present in bed material in large quantity. In contrast, ﬁner grained material tends to be deposited in quieter water. The degree of sorting, another important measure

62

B. R. Bolton et al.

frequently used in describing sediment characteristics, reﬂects what is supplied from upstream and the competence of the ﬂow to transport sediment. Assuming that all size fractions are supplied, the upper-size limit is set as the maximum size the ﬂow can transport. The lower limit is determined by the washload/suspended bed material load cutoff. This is a function of the river’s energy and sediment particle fall velocity. Fall velocity is largely determined by sediment size, shape, and density. The sediment presently found in the channel beds and levees of the Ok Tedi and Fly Rivers is derived from a number of sources: (1) reworked bed and bank material; (2) material eroded from late Pleistocene and Holocene terraces cut by the active river channel; (3) material supplied from presentday erosion of the upland catchment areas particularly by major tributaries such as the Ok Tedi and Strickland River; and (4) mine wastes discharged from the Mount Fubilan copper-gold mine located in the headwaters of the Ok Tedi. Since the start of mining OTML has monitored sediment grain size, in both the Ok Tedi and Fly Rivers, to assess the impact of waste materials released from the mine on the downstream ﬂuvial environment (e.g., OTML, 1986, 1997, 2000, 2001, 2002, 2003a, 2003b, 2004). Figures 2.3, 2.4, and 2.6 summarize the results of selected parts of this work as a basis for comparison with pre-mine conditions. With the riverine deposition of mine wastes, the mine area creeks changed rapidly from those typical of source-area streams to streams in which sediment supply clearly exceeds the transport capacity. Rather than being mostly swept clean of sediment, as was characteristic during pre-mine times, sediment, now unable to be transported downstream, was deposited resulting in bed aggradation. It has been estimated that between 30 and 35% of all material entering the southern mine area creeks is retained and that bed levels have increased up to 70 m or more in their headwaters close to the mine (Marshall, 2000, 2003). Marshall (2000) estimated that up to November 2000 in excess of 200 Mt of mine waste materials had accumulated in the Harvey Creek–Ok Mani network and that new material was being added at an annualized rate of around 25 Mt and that approximately 66 Mt is delivered to the upper reaches of the Ok Tedi, each year. The material now being deposited in the mine area creeks is coarse-grained with a median grain size generally between 40 and 60 mm while the d90 is consistently close to 100 mm (Fig. 2.3B). These sediments are mainly composed of coarse sands (typically 50%) with lesser amounts of gravel and ﬁner fractions (Fig. 2.4B). They are also poorly sorted and in most cases show bimodal size distributions with ﬁner sands inﬁlling intergranular spaces in gravels (Fig. 2.6A and B). Exposed deposits in this area reveal a sediment

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

63

proﬁle that is typically massive and largely devoid of internal structure (Fig. 2.2C). These features are typical of ﬂuvial environments where deposition is rapid and subject to mass-ﬂow processes. The sediments delivered to, and deposited in, the upper reaches of the Ok Tedi continue to be mainly coarse grained with median sizes in excess of 40 mm, for at least the ﬁrst 100 km downstream of the mine (Fig. 2.3B). While sediments deposited in this reach are comparable to pre-mine sediments in terms of grain size, a well-deﬁned armored zone is present but only to a minor extent (Fig. 2.2D). Near absent also is the gravel–sand transition zone of pre-mine times. As shown in Fig. 2.4B, at around 100 km downstream from the mine the composition of the bed changes abruptly from coarse sand and gravels to being almost entirely composed of medium to ﬁne sands and silts. This change in the nature of these sediments is also reﬂected in the sorting characteristics, as shown in Fig. 2.6(C–E). This ﬁgure shows a transition from bimodal sediments typical of the upper reaches of the Ok Tedi downstream of the mine area creeks, through to a relatively short reach in which sediments deposit in segregated gravel and sand bars, to well-sorted sands in the lower reaches of the river. This change is in contrast to the much longer zone of gravel rifﬂes and discontinuous sand bars typical of pre-mine conditions, as described above (Pickup, 1984) (Figs. 2.2A, 2.4A). The Ok Tedi joins the Fly River at D’Albertis Junction, about 190 km downstream of the mine (Fig. 2.1). While there have been regular campaigns to characterize the texture of sediments deposited on the bed and banks of the Fly River, very few of these campaigns have employed the same sampling methodologies, making direct comparisons difﬁcult. Earlier studies have typically analyzed particle size in single samples collected from the thalweg, mid-channel, or from speciﬁed locations across the channel (usually, 0.25, 0.5, and 0.75 channel width). We are aware of only two studies that have adopted a similar approach to sampling: the comprehensive survey undertaken by Pickup et al. (1979) before the start of mining and an intensive survey of sediment properties carried out by OTML in May 2003 (OTML, 2004). Both studies collected samples from the levees at either end of river transects. Figure 2.6 is a plot of the d50 (median grain size) and d90 for the 1979 and 2003 surveys. This ﬁgure shows a high degree of similarity between the two surveys. Both data sets show that d50 and d90 decrease downstream with increasing distance from Kiunga. Furthermore, these plots show that very little difference exists in the rate of change in grain size in both bed and bank sediments, based on the slope of the lines of best ﬁt. Pre-mine and post-mine plots are also similar in that while grain size does appear to decrease

64

B. R. Bolton et al.

downstream the trend is not systematic; rather, the data show change mostly occurs over short distances separated by longer intervals with relatively little change. Both data sets also show a pronounced ﬁning in grain size in the bed sediment in the lower Middle Fly upstream of Everill Junction (Fig. 2.6A, C). These similarities are also reﬂected in the downstream variation in size distributions in the increase in proportion of ﬁner size fractions with distance from Kiunga, in both bed and bank sediments (Fig. 2.4). The most noticeable difference between the 1979 and 2003 is in bed material grain size; the sediments of the more recent survey are signiﬁcantly ﬁner. This is reﬂected not only in the d50 and d90 (Fig. 2.6A and C) plots but also the increase in ﬁner size fractions as shown in the size distribution plots (Fig. 2.4A and B). Signiﬁcantly, the opposite trend is shown in sediments of the bank where overall there appears to be a slight increase in grain size and the proportion of coarser size fractions, compared to the pre-mine survey (Figs. 2.4 and 2.6). This may be because deposition of ﬁner mine sediments is greater in upstream reaches (Geoff Pickup, pers. comm.). Another signiﬁcant difference to emerge from Figs. 2.4 and 2.6 is the rapid change in grain size and the proportion of medium and ﬁne sand immediately downstream of D’Albertis Junction. While not shown for the 1979 survey, there is a similar dramatic change in both grain size and grain size distribution downstream of Everill Junction reﬂecting the input of coarser material carried by the Strickland. We interpret the sudden change in grain size characteristics of the Fly channel bed sediment, compared to the 1979 survey, to largely reﬂect the addition of Mount Fubilan mine waste. OTML studies have shown that there has been at least a 5-fold increase in the sediment load entering the Middle Fly from the Ok Tedi as a result of mining (Markham and Repp, 1992). Pile et al. (2003) has also shown that mine wastes supplied to the lower reaches of the Ok Tedi as bed load are dominantly composed of ﬁner sandand silt-sized materials that now also characterize the Middle Fly downstream of D’Albertis Junction. We suggest therefore, that the dominant source of sediment to the Middle Fly and in turn the main control on sediment texture, are waste materials released from the mine. Input from previously deposited sediments while important, is considered to be of secondary importance and are estimated to comprise only a small fraction of sediment released into the active channel (OTML, 2001). Similarly, while perhaps locally important (Pickup et al., 1979), contributions to the sediment load from erosion of bedrock are considered minor. Tributaries, other than the Ok Tedi and Strickland River, appear to play no signiﬁcant role in inﬂuencing the grain size characteristics of either the bed or bank sediments in the Fly River.

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

65

2.2.2.1. Cause of downstream ﬁning in grain size Downstream ﬁning of grain size is a feature typical of many gravel- and sand-bedded rivers around the world, including rivers elsewhere in PNG (Leopold et al., 1964; Pickup, 1984; Parker, 1991a, 1991b; Paola et al., 1992; Gomez et al., 2001). Grain size investigations on the Fly River system mostly conform to this pattern, but with some variation. As discussed above, median grain size in the Fly decreases rapidly over the ﬁrst 45 km downstream from Kiunga then remains relatively constant for the next 250 km before again decreasing rapidly over the next 50 km. Grain size remains relatively constant until Everill Junction where it increases over just a few kilometers before decreasing as the river enters the delta (Figs. 2.3B and 2.4B). It is generally agreed that three factors, often in combination, control the textural characteristics of riverbed materials: (1) the nature and location of material delivered to the catchment from the surrounding hinterland; (2) the ability of the river to sort and transport delivered sediment; and (3) the susceptibility of delivered material to abrasion and breakdown (i.e., chemical and physical weathering) (e.g., Gomez et al., 2001; Knighton, 1998; Paola et al., 1992; Pickup, 1984). As previously described, the main source of sediment to the Fly River system (along with most river systems in PNG) is the headwater streams located in the central cordillera. This natural supply of sediment has been greatly supplemented in recent times by the discharge of waste materials from the Mount Fubilan mine at the headwaters of the Ok Tedi and, to a far lesser extent, the Porgera mine which discharges waste materials into the Lagaip River, an upland tributary of the Strickland River. Downstream grain size variation diagrams (Figs. 2.3 and 2.6) indicate sediment supplied by the Ok Tedi and the Strickland River greatly inﬂuences the textural characteristics of the Fly River bed. As for the other factors that might inﬂuence downstream grain size variation, there appears to be little evidence that selective sorting and deposition of coarser material plays a major role in downstream ﬁning. As shown in Fig. 2.7, the degree of sorting in bed material (as evidenced by the slope of the distribution curve) does not increase systematically downstream as might be expected in a river in which coarser material is progressively removed by deposition. Indeed, the data presented here suggest the degree of sorting deteriorates between D’Albertis and Everill Junctions, before increasing again in the upper Lower Fly under the inﬂuence of the greater ﬂow competence of the combined Strickland/Fly Rivers. The deterioration in the degree of sorting as well as the pronounced decrease in grain size and

100

100

80 Percent finer

Percent Finer

Harvey Creek December 2000 (0.5 - 2.5 km downstream of mine) 80

60

40

60

40

0 0.01

0 1

10

A

100

1000

10.00

100.00

80 Percent finer

Percent Finer

1.00 Particle Size (mm)

100

Ok Mani October 2003 2.5 - 9 km downstream of mine 80 60 40

60 40

20

20

0 1

10

B

100

0 0.01

1000

Particle Size (mm)

1.00

10.00

100.00

Particle Size (mm)

80

60

40

60

40

20

20

0 0.01

Lower Middle Fly (296 - 441 km) 0.10

C

1.00 Particle Size (mm)

10.00

0 0.01

100.00

0.10

1.00

10.00

100.00

Particle Size (mm)

H 100

100

Upper Ok Tedi (Braided Reach: 120 km downstream of mine)

60

80

Percent finer

80

40

60

40

20

0 0.01

0.10

100 Upper Ok Tedi (Gorge Reach: 21 km downstream of mine)

Percent finer

80

Upper Middle Fly (58 -107 km)

G

100

Percent finer

0.10

F

Particle Size (mm)

100

Percent finer

Upper Fly (Upstream of D'albertis Junction: 0 - 38 km)

20

20

20

0.10

1.00

10.00

0 0.01

100.00 I

Particle Size (mm)

D

Lower Fly (482 - 700 km) 0.10

1.00 Particle Size (mm)

10.00

100.00

100

Percent finer

80

60

40 Lower Ok Tedi (Meander Reach: 130 kmdownstream of mine)

20

0 0.01 E

0.10

1.00

10.00

100.00

Particle Size(mm)

Figure 2.7: Bed material size distribution for Ok Tedi and Fly River after the start of mining (OTML, 2001, 2004).

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

67

increase in proportion of silts and clays observed in the lower Middle Fly (296–450 km) may in part be due to the backwater effect from the Strickland River (Pickup et al., 1979; Pickup, 1984; Dietrich et al., 1999). Under this scenario episodic backup of the Fly River waters above Everill Junction due to higher ﬂows in the Strickland cause changes in ﬂow velocity which in turn cause a decrease in the transport capacity of the river and its competence. Alternating deposition of ﬁne and coarse materials in this reach of the river, in response to this mechanism (Pickup, pers. comm.), would explain the increased spread of grain size distribution curves, deterioration in the degree of sorting, increase in the proportion of silts and clays, and the overall decrease in grain size, compared to river reaches both upstream and downstream of this part of the river. Downstream ﬁning of bed material in the sand-bedded Fly River as a result of selective deposition of coarser material was recently proposed by Dietrich et al. (1999). This team suggested that rather than a large loss of coarser grained materials and a signiﬁcant decrease in the river’s ability to transport sediment, ﬁning can occur with only modest loss of coarser grained material (15%), either through deposition on a slowly aggrading riverbed, or, on the adjacent levees and ﬂoodplain. The presence of medium sand (B60%) in bank crest sediment of the upper Middle Fly would appear to support this idea. While there are no detailed quantitative studies of particle shape for the Fly River system as a whole, abrasion is known to occur in bed sediments, as evidenced by the increase in the degree of roundness of sediment particles, along the upper reaches of the Ok Tedi (Pile et al., 2003). In contrast, Day and Associates (1996), in a study of bed sediments from the Fly River found no evidence of a measurable change in grain shape from samples taken in the upper reaches of the Fly above D’Albertis Junction. A similar observation was noted by OTML (2003b). In this study visual classiﬁcation of grain shape and roundness indicated little systematic change downstream. Quartz and feldspar particles, the dominant mineral components of the bed material, were consistently well rounded and of high sphericity throughout the entire 700 km length of the river sampled in this survey. Furthermore, between D’Albertis and Everill Junctions, there was little change in the proportion of mineral and rock fragments that might indicate abrasion and breakdown of less-resistant components (see discussion on sediment mineralogy below). Weathering, as suggested by Knighton (1998), for rivers in general, probably only plays a minor role in determining grain size characteristics in the Fly River. To be effective as a means of bringing about particle disintegration this mechanism requires prolonged exposure to weathering agents. While this might be true for some of the sediments stored in terraces,

68

B. R. Bolton et al.

channel bars and the ﬂoodplain environments of the lower reaches of the system, evidence from sediment transport studies show that supply of material from these sources are relatively minor. It is estimated that erosion of ‘‘natural’’ material along the river course supplies about 10% of the total transported load entering the Fly downstream of D’Albertis Junction (OTML, 2001). If grain size reduction was due to chemical weathering of sediment temporarily stored in the channel bar and ﬂoodplain deposits then it might also be expected that the clay mineralogy of the bed material might also reﬂect this contribution. As discussed below, no observable change in the proportion of clay mineral species was observed in the Middle Fly reach. In summary, while downstream ﬁning of bed material is apparent in the data, this trend is complicated by input from tributaries (mainly the Ok Tedi, with its substantial load of Mount Fubilan mine wastes, and the Strickland), and the backwater effect in the lower Middle Fly during period of high ﬂow in the Strickland. In light of these factors it may be more useful to examine changes in particle size on the basis of individual samples, major geomorphic elements, and the source of supply, as ﬁrst suggested by Pickup (1984). According to this approach the Fly River system can be divided into at least ﬁve separate zones each with its own characteristic grain size distribution. While three of these zones are restricted to the gravel-bedded upper reaches of the Ok Tedi, it is signiﬁcant that the remaining two zones – named by Pickup the ‘‘sand’’ and ‘‘backwater’’ zones, respectively – continue to be well represented despite the large increase in overall sediment load in the system resulting from mining at Mount Fubilan. It would appear that the only major difference between the 1979 and 2003 surveys in terms of the grain size distribution is an overall decrease in grain size in bed materials in both zones as a result of mine inputs. Furthermore, it is apparent from the data that the boundary between the sand and ﬁner grained sediments of the lower Middle Fly is migrating upstream. A comparison between the 1979 and 2003 surveys (Fig. 2.4) shows that this boundary has moved from about 330 km downstream of Kiunga to about 260 km – a shift of approximately 70 km upstream. This shift might be best explained in terms of the continuing inﬂuence of the Strickland River and the backwater effect and the ﬁner nature of mine inputs.

2.3. Metal Geochemistry of River-Deposited Sediments In this section we look at what is known about the abundance of selected metals in sediments deposited in the channel bars and levees of the Ok Tedi

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

69

and Fly River from before the start of mining at Mount Fubilan in 1984 through to 2004. For reviews of metal geochemistry in sediments carried in suspension, and deposited on the Fly River ﬂoodplain and in the Fly Estuary and the Gulf of Papua, the reader is referred to companion chapters in this volume by Apte et al., Brunskill et al., and Nittrouer et al. 2.3.1. Before Mining Information on the geochemistry of pre-mine river-deposited sediments in the Fly River system is scarce. Table 2.1 summarizes these data but it should be pointed out that the paucity of data together with variation in sampling methodology, the size fractions analyzed, and the analysis method used, makes comparisons between surveys and estimates of baseline or background metal concentration difﬁcult. One of the earliest studies dealing with the geochemistry of river sediments was conducted by the Cambridge Expedition of 1974 (Boyden et al., 1974). This study presented the results of trace metal determinations on 39 samples (o 200 mm fraction) collected from the Ok Tedi, six of its tributaries (Ok Gilor, Ok Mani, Ok Ningi, Ok Mabiong, Ok Menga, and Ok Ma), and the Fly River near the township of Kiunga (Fig. 2.1) (Table 2.1). While the samples of this data set are too sparse to provide an accurate picture of baseline sediment geochemistry in this region, the results do identify the presence of anomalously high copper, lead, zinc, manganese, and possibly cadmium, in the creeks surrounding the mine and upper reaches of the Ok Tedi, compared with downstream tributaries, Ok Menga (except in Cd), and the Ok Ma, as well as the lower Ok Tedi (Ningerum), and the Fly River at Kiunga. Pre-mine copper values, for example, averaged about 173 ppm in deposited sediments of the Ok Tedi tributaries compared to an average of 62 ppm in the Ok Tedi itself. Figure 2.8, which shows the variation in elemental abundance from the mine to D’Albertis Junction, indicates that trace metal concentrations generally decreased downstream from the Mount Fubilan. Interestingly, the highest average values for Co, Fe, Mn, as well as Ni and Zn, were all recorded at D’Albertis Junction. It was the anomalous geochemistry of these river sediment samples, together with more intensive studies carried out by the Kennecott Australian Explorations Pty Ltd beginning in 1968, which in large part led to the discovery of the Mount Fubilan copper-gold ore body (Davies et al., 1978; Rush and Seegers, 1990). Everett and Associates (1980), as part of a series of government-sponsored studies of the river system in 1979, prior to the start of mining, provide

70

Table 2.1: Geochemistry of pre-mine river bed and bank sediments.

Mine area creeks Ok Mani Ok Mani

Al (%) Fe (%) S (%)

As (ppm)

Cd (ppm)

Co (ppm)

Cu (ppm)

Mn (ppm)

Ni (ppm)

Pb (ppm)

Zn (ppm)

Type

Fraction

– –

3.6 3.3

– –

11 –

– 2.2

9 19

36 30

620 380

135 32

25 34

88 82

Bed Bed

Clay o200 mm

Ok Kam Ok Gilor

– –

– 9.7

– –

– –

1.0 5.5

– 29

15 852

308 1680

– 38

40 185

30 267

Bed Bed

? o200 mm

Ok Ningi

–

4.9

–

–

3.1

17

233

677

24

60

274

Bed

o200 mm

Ok Mabiong

–

4.6

–

–

2.4

24

198

768

36

69

252

Bed

o200 mm

Mean Median

– –

4.7 4.1

– –

– –

2.9 2.4

20 19

173 33

661 586

47 36

58 37

145 88

–

2.8

–

–

4.1

17

9

305

29

24

79

Bed

o200 mm

Ok Ma

–

4.1

–

–

2.2

26

9

552

38

26

87

Bed

o200 mm

Mean Median

– –

3.5 3.5

– –

– –

3.2 3.2

22 22

9 9

429 429

34 34

25 25

83 83

–

4.0

–

–

2.4

20

90

553

31

49

149

Bed

o200 mm

–

4.7

–

–

2.3

28

34

528

28

27

98

Bed

o200 mm

– –

– 2.9

– –

– 8

1.0 –

– 7

19 23

485 310

– 170

43 23

31 61

Bed Bed

NG Clay

Ok Tedi tributaries Ok Menga

Ok Tedi Upper Ok Tedi (Tabubil) Upper Ok Tedi (Ningerum) Upper Ok Tedi Upper Ok Tedi (Tabubil)

Source

Everett (1980) Boyden et al. (1974) Maunsell (1982) Boyden et al. (1974) Boyden et al. (1974) Boyden et al. (1974)

Boyden et al. (1974) Boyden et al. (1974)

Boyden et al. (1974) Boyden et al. (1974) Maunsell (1982) Everett (1980)

B. R. Bolton et al.

Location

–

4.6

–

–

–

38

120

697

53

72

–

Bank

Clay

Everett (1980)

–

6.8

–

–

–

44

120

1162

52

73

–

Bank

Clay

Everett (1980)

–

4.6

–

–

–

19

17

1239

142

25

–

Bed

Clay

Everett (1980)

–

3.9

–

–

–

35

92

697

53

70

–

Bed

Clay

Everett (1980)

–

29.6

–

–

4.2

88

44

2352

46

24

453

Bed

o200 mm

Boyden et al. (1974)

Mean Median

– –

7.6 4.6

– –

– –

2.5 2.4

35 32

62 44

891 697

72 53

45 43

158 98

Fly River U Fly

–

6.0

–

–

2.2

22

17

787

35

29.8

132

Bed?

o200 mm

U Fly U. Middle Fly L. Middle Fly

– – 8.6

4.5 5.3 3.7

– – 0.14

– – 4.6

– – 0.3

39 38 13

100 95 45

775 775 289

65 56 33

80 65 18

– – 142

Bank Bank Bed

Clay Clay o100 mm

L. Middle Fly L. Middle Fly

–

– 4.9

–

– –

2.6 –

– 42

28 99

– 697

– 58

25 73

88 –

Bed Bank

Clay Clay

Boyden et al. (1974) Everett (1980) Everett (1980) Hettler et al. (1997) Maunsell (1982) Everett (1980)

Bed Bank Bank

Silt/clay Clay Clay

Maunsell (1982) Everett (1980) Everett (1980)

Bank

Clay

Everett (1980) Bowen, 1979

Mean Median

4.9 4.9

– –

– –

1.7 2.2

31 38

64 70

665 775

49 56

48 47

121 132

– 8.2 7.9

– 4.8 4.9

– 0.06 0.05

– – –

2.5 – –

– 39 39

20 110 98

– 775 813

– 88 78

33 73 76

57 – –

Mean Median

8.1 8.1

4.9 4.9

0.06 0.06

– –

– –

39 39

76 98

794 794

83 83

61 73

– –

L. Fly Av. crustal abundance

– 8.2

4.8 4.1

– 0.03

– 1.5

– 0.1

41

100 50

852 852

63 80

71 14

– 75

Strickland River Strickland R. Strickland R. Strickland R.

71

– –

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

Lower Ok Tedi (Konkonda) Lower Ok Tedi (Aran) Lower Ok Tedi (D’Albertis Jct) Lower Ok Tedi (D’Albertis Jct) Lower Ok Tedi

72

B. R. Bolton et al.

Metal concentration (ppm)

10000

1000

100

10

ga un

D

'a

lb

Ki

Jc

t.

m er

in

tis

ge

ru

bi

l

a

bu

M k O

Ta

ga

i

en

an

M

N

O

O

k

O

k

ab M

M

io

in N k

O

k

gi

r ilo G k O

ng

1

Sample location Cd

Co

Cu

Fe%

Mn

Ni

Pb

Zn

Figure 2.8: Variation in metal concentration in sediment samples collected from the Ok Tedi and selected tributaries (data from Boyden et al., 1974). additional baseline geochemical data on the clay-sized fraction of a small number of river-deposited sediments. Table 2.1 which includes the results of this work, shows that Cu concentrations in the clay-sized fraction were generally higher in the samples collected from the river banks compared to those collected from the bed; whether in the higher energy environments of the Ok Tedi or the more sedate reaches of the Fly and Strickland Rivers. A similar trend appears to be typical for Pb, Co, and possibly Cr while Ni showed an opposite trend with highest concentrations occurring in the sandrich channel-bar deposits of the upper reaches of the Ok Tedi and the Ok Mani. The results presented by Everett and Associates (1980) for the Ok Mani, one of the creeks located immediately downstream of the mine, are broadly similar to those reported by Boyden et al. (1974). In detail however, the concentrations of Co, as shown in Table 2.1, are considerably higher in the 1974 survey. Nickel by contrast, is signiﬁcantly higher in the 1980 survey; a similar result can also be seen when comparing riverbed samples from the Ok Tedi near Tabubil. Furthermore, the samples collected from the banks of the Ok Tedi just upstream of D’Albertis Junction during the 1980 survey generally had lower metal concentrations than those collected in the 1974

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

73

survey except for Cu. The ﬁner grained sample from the 1980 survey gave a copper concentration of 92 ppm compared to 44 ppm recorded in the earlier study. Additional pre-mine sediment geochemistry is presented by Maunsell and Partners (1982) as part of an OTML-sponsored environmental study to examine potential impacts of the mine on the Fly River system. In this study, results of analysis of four grab samples collected from the bed of the active channels of the lower Middle Fly and Strickland Rivers were presented, along with data on four samples from the upper reaches of the Ok Tedi, upstream of the mine and one sample from the Ok Kam, a headwater tributary of the Ok Tedi. Table 2.1 shows metal concentrations were broadly in line with those previously determined in the 1974 and 1980 studies. Hettler (1995), as part of a UNDEP study of the environmental impacts of mining at Mount Fubilan, reported the results of a comprehensive study of cores collected from the bed of the Fly River and adjacent ﬂoodplain environments. Using probability plots of copper distribution, this worker concluded that average metal abundance of riverbed sediments present at Obo, on the lower Middle Fly, at about 45 ppm, were broadly similar to those presented by Everett and Associates (1980) and Maunsell and Partners (1982). Table 2.1 also shows that pre-mine metal concentrations in sediments, when compared to average crustal abundance, were enriched in Fe, As, Cd, Cu, Pb, and Zn, as well as Al, S, Ag, Mo, V, and Sc, based on the average background metal concentrations for the Middle Fly determined by Hettler (1995). The table also shows that pre-mine sediments of the Strickland were enriched relative to average crustal abundance, in Fe, S, As, Cd, Cu, Pb, and Zn. Figure 2.9 shows box plots of elemental abundance in the various river reaches as well as average crustal abundance of each element. This plot summarizes pre-mine metal geochemistry for the river system. It shows that most elements maintain relatively constant concentration down the river system except for Cd, which appears to increase downstream, and Mn, which shows the reverse trend. Zn, Cd, Cu, Fe, and Pb generally occur in concentration above average crustal abundance while Mn and Ni are generally below. 2.3.2. After the Start of Mining Since the start of mining at Mount Fubilan in 1984 various sampling campaigns have been conducted mainly to monitor the impact of mining on downstream environments. The results of these sampling campaigns are

B. R. Bolton et al.

Cd (ppm)

3 2 1 0 Ok Tedi Tributaries Ok Tedi Upper Fly Middle Fly Lower Fly Strickland Av. Crustal Abund.

Ok Tedi Tributaries Ok Tedi Upper Fly Middle Fly Lower Fly Strickland Av. Crustal Abund.

-1

Ni (ppm)

Ok Tedi Tributaries Ok Tedi Upper Fly Middle Fly Lower Fly Strickland Av. Crustal Abund.

Total Fe (%)

Pb (ppm) 200 180 160 140 120 100 80 60 40 20 0

Ok Tedi Tributaries Ok Tedi Upper Fly Middle Fly Lower Fly Strickland Av. CrustalAbund.

4

Mn (ppm) 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 400 200

Ok Tedi Tributaries Ok Tedi Upper Fly Middle Fly Lower Fly Strickland Av. Crustgal Abund.

5

180 160 140 120 100 80 60 40 20 0

Cu (ppm) 900 800 700 600 500 400 300 200 100 0 -100

6

35 30 25 20 15 10 5 0 Ok Teditributaries Ok Tedi Upper Fly Middle Fly Lower Fly Strickland Av Crustal Abund.

Zn (ppm) 500 450 400 350 300 250 200 150 100 50 0

Ok Tedi Tributaries Ok Tedi Upper Fly Middle Fly Lower Fly Strickland Av. Crustal Abund.

74

Figure 2.9: Box plots showing metal variation in river deposited sediments for selected reaches of the Ok Tedi/Fly River system downstream of Mount Fubilan mine. Also shown for comparative purposes is average crustal abundance (from Bowen, 1979). summarized in Table 2.2 along with the geochemistry of the Mount Fubilan orebody and its waste products. The earliest report on the geochemistry of river-deposited sediment after the start of mining is by Kyle (1988). The single riverbed sample analyzed in this work was collected from the lower Middle Fly and perhaps not surprisingly, given that it was collected only about six months after the start of mining, gave metal concentrations similar to those recorded in the premine (Maunsell and Partners, 1982) study for the same region. The ﬁrst major attempt at characterizing the geochemistry of riverdeposited sediment throughout the river system is presented in OTML (1994). This study, which only deals with copper abundance, was completed about 10 years after the start of mining and includes samples collected along

Table 2.2: Geochemistry of deposited river sediments after the start of mining at Mount Fubilan. Location

Fe (%)

S (%)

As (ppm)

Cd (ppm)

Cu (ppm)

Mn (ppm)

Ni (ppm)

products 8.6 3.2 6.6 13.8 – 16.4 6.1 14.6 8.4 3.1

0.94 3.54 1.76 2.36 0.86

3 – 135 5 5

5.60 – 5.00 0.51 92.40

2,050 1,461 7,100 8,083 2,030

520 – 555 451 629

12 – 14 52 15.5

Pb (ppm)

Zn (ppm)

Source

26 114 27 57 140

42 117 94 415 66

EGi (1996) OTML OTML EGi (1991) EGi (1996)

Mine area creeks Harvey creek Harvey creek Ok Mani

– 0.8 1.1

– 2.2 3.9

– 1.08 1.95

– 13 12

– 0.94 0.67

5,510 819 1,143

– 442 410

– 11 12

– 71 58

– 330 290

OTML (1993) OTML (2004) OTML (2004)

Ok Tedi Upper Ok Tedi Upper Ok Tedi

– 6.1

– 4.8

– 1.52

– 30

– 1.60

1917 1523

– 700

– 17

– 463

– 541

Upper Ok Tedi Lower Ok Tedi

0.8 –

4.7 –

2.05 –

20 –

0.59 –

893 864

377 –

7 –

36 –

163 –

OTML (1993) Hettler et al. (1997) OTML (2004) OTML (1993)

0.8

2.0

0.003

3

0.04

8

269

9

4

29

OTML (2004)

–

–

–

–

–

415

–

–

–

–

OTML (1993)

1.4

4.9

0.64

10

0.69

479

668

10

43

219

OTML (2004)

9.3

4.3

–

12

–

o83

621

o74

–

129

Delft (1987) 75

Fly River Upper Fly River Upper Middle Fly Upper Middle Fly Lower Middle Fly

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

Orebody and waste Ex-mill tailings Ex-mill tailings Orebodya Orebody Waste rock

Al (%)

76

Location

Al (%)

Fe (%)

S (%)

As (ppm)

Cd (ppm)

Cu (ppm)

Middle

–

5.8

–

–

1.70

Middle

–

–

–

–

Middle

7.4

4.6

0.22

Middle

8.1

4.0

Middle

1.6

Lower Fly Lower Fly Strickland River Strickland Strickland

Lower Fly Lower Fly Lower Fly Lower Fly Lower Fly

Mn (ppm)

Ni (ppm)

Pb (ppm)

Zn (ppm)

58

515

–

15

120

–

314

–

–

–

–

15

1.80

760

520

15

50

185

EGi (1996)

0.25

8

0.50

530

528

21

79

211

3.8

0.64

11

0.79

740

543

11

68

298

Hettler et al. (1997) OTML (2004)

– 1.7

– 4.1

– 0.04

– 11

– 0.30

60 103

– 540

– 50

– 23

– 124

OTML (1993) OTML (2004)

8.6 1.7

4.4 3.8

– 0.02

12 13

– 0.35

o190 28

735 548

o35 45

– 24

107 122

Delft (1987) OTML (2003a, 2003b)

Note: Also given are typical metal concentrations for tailings, Mount Fubilan orebody, and waste rock. ¼ Not determined. a Median values given.

Source Kyle (1988) OTML (1993)

B. R. Bolton et al.

Table 2.2: (Continued ).

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

77

the entire length of the river system – a total river distance of about 1,000 km (Table 2.2). Figure 2.10, from OTML (1994), plots copper abundance, along with particle size distribution and copper distribution against distance downstream from the mine for riverbed (deposited) and suspended sediment load. The plots also show how each of these parameters change according to the size fraction analyzed. It is clear from these plots that copper concentrations in both deposited and suspended sediments decreases downstream from the mine (Fig. 2.10A and D). It is also apparent that this change is not progressive but characterized by sudden decrease at both D’Albertis and Everill Junctions (190 and 598 km downstream of the mine, respectively). Furthermore, the plots show these changes across all particle size fractions and bulk samples. The pronounced change at Everill Junction is thought to be the result of dilution by sediment input from the Strickland River. Figure 2.10 also shows that the highest copper concentrations, irrespective of location or sample type, typically occur in the ﬁner grain size fractions (see also Pile et al., 2005). The only signiﬁcant variation from this pattern appears in deposited sediment of the Ok Tedi and in the Middle Fly immediately downstream of D’Albertis Junction. Here the coarser fractions (W 63 mm) typically contain about 80% of the total copper content of sediments (Fig. 2.10F). Furthermore, this ﬁgure shows that the o5 mm fraction consistently contributes about 50% of total copper compared to the contributions by the W63 mm fraction which consistently contributed less than 2% of the total copper. A similar trend was described for benthic sediment; however, in this case, it was found that closer to the mine the largest size fraction (W63 mm) contributed the highest percentage of total copper while in the Middle Fly, and in particular downstream of about 400 km from the mine, this pattern was reversed with the ﬁnest fraction (o5 mm) contributing consistently over 50% of the total copper (except for a brief interval just downstream of Everill Junction). The increase in copper concentration, in the ﬁrst instance between about 10 and 100 km from the mine then again from about 400 to 600 km from the mine, described above, may be related to the varying role of a particular grain size fraction in controlling copper chemistry in sediments. In this interpretation the peak in copper concentration observed in the upper reaches of the Ok Tedi may relate to the preferential deposition of copper contained in the larger size fractions while the second peak observed in the lower Middle Fly may relate to the higher copper content of the ﬁner fractions dominant in this reach (see section of grain size variation above). Signiﬁcant also is the observation that, at least for the Ok Tedi copper peak as deﬁned in the 2003 survey, heavy mineral studies have shown that the

5000 4000 3000 2000 1000 0 200

90 80 70 60 50 40 30 20 10 0

400 600 800 Distance from the mine (km)

1000

D. Copper concentration 10000 1000 100 10 1

1200

0

B. Particle size distribution

0

200

400

600

200

800

1000

1200

200

400

600

800

0

200

63 um -20 um

20 um -10 um

Percent of total copper

Percent of total copper

1000

1200

400

600

800

1000

1200

1000

1200

1000

1200

F. Percent of total copper

100 80 60 40 20 0 0

200

Distance from mine (km) >63 um

800

Distance from the mine (km)

C. Percent of total copper contributed

0

600

E. Particle size distribution

100 90 80 70 60 50 40 30 20 10 0

Distance from the mine (km) 80 70 60 50 40 30 20 10 0

400

Distance from the mine (km)

Percent

Percent of total sediments

0

100000

400

600

800

Distance from the mine (km) 10 um -5 um

<5 um

>63 um

63 um - 20 um

20 um -10 um

10 um - 5 um

Figure 2.10: Downstream variation in copper concentration, particle size distribution, and copper distribution for suspended sediment (A, B, C) and deposited bed load sediment (D, E, F) from the Fly River (data from OTML, 1994).

B. R. Bolton et al.

Copper concentration (ug/g)

A. Copper concentration

6000

78

Copper concentration (ug/g)

7000

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

79

largest percentage of heavy minerals as a percentage of the total sediment, occurs coincident with the copper peak (i.e., just upstream of Bige). A similar increase in the lower Middle Fly in the relative abundance of sulﬁde minerals (mainly pyrite) may also explain the sudden increase in copper concentration, at least for the interval between 392 and 600 km (Everill Junction). A second system-wide survey of deposited sediment geochemistry was undertaken in 2004 by OTML to better understand the chemical impact of mine operations on the river (OTML, 2004). Summary statistics for metal analysis of composite samples from a total of 85 river channel transects completed in this survey are given in Table 2.2. Table 2.2 indicates that signiﬁcant differences occur in the metal geochemistry of riverbed sediments according to river reach. Downstream of D’Albertis Junction, the Fly Riverbed sediments are relatively enriched in Al, Fe, As, Cd, Cu, Mn, Pb, and Zn, when compared to sediments of the Upper Fly. Sulfur, also low in the Upper Fly, increases dramatically downstream of D’Albertis Junction and then decreases again to relatively low concentrations below Everill Junction. Nickel is relatively low throughout the Upper and Middle Fly reaches but increases signiﬁcantly below Everill Junction. These trends become even more clearly deﬁned when these data are examined in terms of enrichment. Figure 2.11 plots metal enrichment for the Ok Tedi/Fly River system downstream of the mine relative to average Upper Fly concentrations and average crustal abundance, respectively. Figure 2.11A shows that most of the elements covered in this survey are enriched relative to concentrations in the deposited sediments of the Fly River upstream of D’Albertis Junction, with greatest enrichment occurring in copper and sulfur. Sulfur concentrations, in all but the sediments of the Lower Fly, downstream of Everill Junction, show the greatest enrichment compared to the Upper Fly sediment with enrichment factors typically in excess of 100. This, as is discussed below, has important environmental implications related to the increased risk of acid generation wherever these sediments might become exposed to the atmosphere. Figure 2.11B shows that since the start of mining, sediments deposited downstream of the mine are enriched in Se, S, Cu, As, Cd, Zn, Pb, and Fe, relative to average crustal abundance. The ﬁgure also shows these sediments are depleted in Mn, Ni, and Al, when compared to average crustal abundance. Figure 2.11A also shows that there is very little dilution of metal abundance from the mine (Harvey Creek) to Everill Junction. Relative metal abundance is almost constant down the river system until the junction with the Strickland at which point dilution by the sediment carried by the later causes a sudden decrease in relative abundance of most metals. Figure 2.12 shows the variation in element abundance in composite riverbed samples downstream from Kiunga based on the closely spaced

80

B. R. Bolton et al.

Enrichment Factor (compared to av. U. Fly River sediment)

1000.0

Al (%) Fe (%)

100.0

S (%) As (ppm) Cd (ppm) Cu (ppm)

10.0

Mn (ppm) Ni (ppm) Pb (ppm) Se (ppm)

1.0

Zn (ppm)

0.1 Harvey Creek

Ok Mani

Enrichment Factor (compared to av. crustal abundance)

A

B

Upper Ok Tedi Reach

Middle Fly

Lower Fly

100.0

Al (%) 10.0

Fe (%) S (%) As (ppm) Cd (ppm)

1.0

Cu (ppm) Mn (ppm) Ni (ppm) Pb (ppm)

0.1

Se (ppm) Zn (ppm)

0.0 Harvey Creek

Ok Mani

Upper Ok Tedi Reach

Middle Fly

Lower Fly

Figure 2.11: Downstream variation in elemental abundance relative to (A) average concentration in sediments of the Upper Fly River (i.e., above D’Albertis Junction) and (B) average crustal abundance (data from Bowen, 1979). Sediment input from the Strickland River, which marks the boundary between the Middle and Lower Fly, is the major cause of the sudden change in enrichment factor seen in both plots.

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

D'albertis Jct.

Everill Jct.

D'albertis Jct.

Total Cu (mg/kg)

1200 800 400

Concentration

Concentration

1600

0 0

200

400

600

800

Total S (%S)

1000

0

200

400

D'albertis Jct.

Everill Jct. Concentration

Concentration

0.8 0.6 0.4 0.2

Everill Jct. As (ppm)

25 20 15 10

200

400

600

800

5

1000

0

200

400

Distance from Mine (km) D'albertis Jct.

600

800

1000

Distance from Mine (km) D'albertis Jct.

Everill Jct.

Everill Jct.

0.08

2.0 Cd (ppm)

1.5 1.0 0.5

Hg (ppm)*

Concentration

Concentration

1000

0 0

0.0

0.06 0.04 0.02 0.00

0

200

400

600

800

0

1000

200

400

Distance from Mine (km) D'albertis Jct.

600

800

1000

Distance from Mine (km) D'albertis Jct.

Everill Jct.

14 12 10 8 6 4 2 0

Everill Jct.

1000 Fe (%)

Concentration

Concentration

800

30 Ag (ppm)

1.0

0.0

Mn (ppm)

800 600 400 200 0

0

200

400

600

800

1000

0

200

400

Distance from Mine (km) D'albertis Jct.

140 120 100 80 60 40 20 0

600

800

1000

Distance from Mine (km)

Everill Jct.

D'albertis Jct.

600 Pb (ppm)

Concentration

Concentration

600

Distance from Mine (km)

1.2

Everill Jct.

500

Zn (ppm)

400 300 200 100 0

0

200

400

600

800

0

1000

200

400

D'albertis Jct.

D'albertis Jct.

Everill Jct.

2.0 1.5 1.0 0.5

Al (%)

Concentration

2.5

0.0 0

200

400

600

800

1000

800

1000

Everill Jct.

70 60 50 40 30 20 10 0

Ni (ppm)

0

200

Distance from Mine (km) D'albertis Jct.

600

Distance from Mine (km)

Distance from Mine (km)

Concentration

Everill Jct.

3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0

Distance from Mine (km) D'albertis Jct.

81

400

600

800

1000

Distance from Mine (km)

Everill Jct.

D'albertis Jct.

Everill Jct.

Se (ppm)

5 4 3 2 1 0

Concentration

Concentration

6 d50 (mm)

10.00 1.00 0.10 0.01

0

200

400

600

Distance from Mine (km)

800

1000

0

200

400

600

800

1000

Distance from Mine (km)

Figure 2.12: Variation in concentration of selected metals and the d50 in river deposited sediments downstream of the Mount Fubilan mine (OTML, 2004).

82

B. R. Bolton et al.

transects sampled by OTML in 2003 (OTML, 2004). This ﬁgure shows Cu increases rapidly in concentration downstream of D’Albertis Junction from a value consistently less than 15 mg/kg to concentrations in excess of 600 mg/kg. In detail, while the data are noisy, the Cu concentration after rising to about 800 mg/kg generally decreases to about 100 mg/kg. At 237 km downstream of Kiunga it then increases to levels marking the highest recorded during this survey at around 1200 mg/kg (at 315 km, just downstream of Bosset Lagoon). Copper levels remain relatively high (350–1,000 mg/kg) until Everill Junction where they decrease rapidly over a short distance to concentrations generally less than 200 mg/kg. Zinc, lead, selenium, and to a slightly lesser extent, arsenic, silver, cadmium, and sulfur show a similar pattern of variation to that of copper. Manganese differs from copper in that after increasing rapidly at D’Albertis Junction it then, with some scatter, progressively decreases downstream to the delta without any apparent increase in the lower Middle Fly. Iron also increases at D’Albertis Junction and while showing considerable scatter between transects, relatively high concentrations (4–7%) are maintained through to about 161 km downstream of Kiunga. At this location, concentrations then decrease rapidly to between 3.8 and 4.1% with a slight rise discernible below Everill Junction before decreasing again toward the delta. Aluminum, unique among the elements discussed here, shows a progressive but noisy increase downstream from Kiunga to the delta. Mercury, for the ﬁrst 250 km or so downstream of Kiunga, hovers close to the detection limit. From just downstream of Manda (at approximately 282 km) concentrations of this element then increase substantially and remain at an elevated level (0.02–0.06 mg/kg), until the river enters the delta. Nickel remains at relatively constant concentrations throughout the middle and upper reaches of the Fly and then, downstream of Everill Junction, increases dramatically before decreasing rapidly again as the river enters the delta.

2.3.2.1. Source of metals Sediments deposited on the bed of the Fly River downstream of D’Albertis Junction are enriched in As, Cd, Cu, Fe, S, Se, and Zn when compared to average crustal abundance as well as Al, Mn, Ni, and Pb, when compared to the average composition of the riverbed in the Upper Fly (i.e., upstream of D’Albertis Junction) (Fig. 2.11A). Earlier work by EGi (1996) based on multi-element analysis of a single sample from the Fly River at Obo indicated enrichment in Ag, As, Bi, Cd, Cu, Mo, S, and Se with respect to average crustal abundance (see also Hettler et al., 1997).

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

83

It has long been argued that the source of metals that causes the metal enrichment now observed in the Ok Tedi and Middle Fly is most likely the waste materials from the Mount Fubilan porphyry copper-gold mine (e.g., Hettler et al., 1997). This interpretation has been largely based on the similarity between the chemical composition of river sediments deposited downstream of the mine along the Ok Tedi and Fly Rivers and the trace metal composition of the Mount Fubilan orebody. A comparison between the abundance of trace metals found in these sediments, the orebody, and the mine’s waste materials highlights this similarity (Table 2.2). The trends described for trace metal distribution and abundance in the river system described above (Figs. 2.11 and 2.12) (Table 2.2) are interpreted to reﬂect the following: (1) the trace metal enrichment in Mount Fubilan orebody (primary control); (2) the distribution pattern of heavy minerals (particularly pyrite and associated Cu-bearing minerals such as chalcopyrite); and (3) the proportion of ﬁner fractions in deposited sediment. The concentration maxima for Cu, Ag, Cd, Pb, Se, Zn, Mn, and S, within river-deposited sediments located in the upper reaches of the Ok Tedi is generally thought to reﬂect preferential deposition of mine-derived heavy minerals in this reach (Fig. 2.12). The presence of river channel bar deposits with up to 13wt% heavy mineral (mainly, magnetite, pyrite, and a range of Cu-bearing sulﬁde minerals) provides support for this interpretation (Pile et al., 2005). A similar concentration maxima for most of these elements in the lower Middle Fly (between 392 and 460 km, Fig. 12), while probably due in part to deposition of metal-bearing sulﬁde minerals (Fig. 2.18), may also in part reﬂect preferential deposition of metal-enriched ﬁner grained silts and muds in the backwater zone upstream of Everill Junction (see discussion on sediment texture above and Fig. 2.4). The large increase in mercury in the lower Middle Fly and Lower Fly is thought to be the result of runoff from Hg-bearing geologic units in the ridge dividing the Fly River and basin from Lake Murray. Anomalous mercury has been identiﬁed in the Lake Murray region for some time and while we are aware of no previous geochemical evidence for enrichment in the Middle Fly region, high levels of mercury have been described from hair samples collected from people living in villages along the Middle Fly (Kyle and Ghani, 1982). The rapid increase in nickel downstream of Everill Junction is interpreted to be the result of inﬂux of Ni-bearing sands and silts from the Strickland River. The source of these Ni-bearing sediments is not known with certainty; however, the Strickland traverses in its middle reaches the western ﬂank of Mount Bosavi, an extinct volcano comprised in part of alkaline and transitional olivine basalts known to have anomalously high Ni (Mackenzie and Johnson, 1984).

84

B. R. Bolton et al.

The abrupt decrease in most metal concentrations in bed material downstream of Everill Junction is in large part due to dilution by the much larger sediment load of the Strickland, which is also considerably coarser in grain size. The important role of the Strickland (and even more so the Ok Tedi) in inﬂuencing the geochemistry of Middle Fly deposited sediments is in contrast to the other tributaries of the Middle Fly, which appear to have little or no inﬂuence in this regard. The progressive increase in aluminum downstream from the mine including the Lower Fly reach is not understood. It may reﬂect increasing Al-bearing clay input to bed load as a result of weathering. Certainly the dominant clay mineral of the numerous off-river water bodies (and entering via the main tributaries of the Middle Fly) is kaolinite. This may in turn say something about exposure times to chemical weathering processes across the Fly alluvial plain.

2.3.2.2. Temporal and spatial changes in geochemistry In this section we examine how metal concentrations in river-deposited sediments vary with time and space. Comparison of Tables 2.1 and 2.2 indicates a dramatic increase in the concentrations of most elements after the commencement of mining in 1984 along much of the Ok Tedi/Fly River system. For example, copper concentrations in Ok Mani, just to the south of the mine, have increased in bulk sediments from around 30 ppm in 1974 to in excess of 1,000 ppm in 1993. By 2004, metal concentrations in Ok Mani and Harvey Creek, even closer to the mine, were consistently close to those seen in mine waste or even the orebody. Further downstream in the lower Middle Fly metal concentrations show a similar rapid increase with time. Near Obo, for example, just upstream of Everill Junction, copper concentrations in riverbed sediments has increased approximately 12-fold from background concentrations of around 45 ppm to over 500 ppm. This increase in copper concentration, both with increasing distance from the mine and with time, can also be seen in Fig. 2.13 which compares the variation in copper concentration in river-deposited sediments collected from various sampling campaigns before the start of mining and detailed surveys conducted by OTML in 1993 and 2003. In detail, this ﬁgure shows copper concentrations, though highly variable, decrease from the mine to the conﬂuence of the Ok Tedi and Fly, in both the 1993 and 2003 data sets. It is interesting to note however, that there is a pronounced increase in concentration of copper peaking at 65 km in the 1993 survey (upstream of Ningerum in the gorge reach of Ok Tedi) and 117 km in

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

Copper concentration (mg/kg)

10000

D'albertis Jct.

Everill Jct.

85

1993 2003

1000

Pre-mine

100

10

1 0

200

400 600 800 Distance from the mine (km)

1000

1200

Figure 2.13: Variation in copper concentration (note log-scale) downstream from Mount Fubilan mine for samples collected in 1979 (Data from Everett and Associates, 1980; Hettler et al., 1997), 1993 (OTML, 1994), and 2003 (OTML, 2003b).

the later survey. Downstream of D’Albertis Junction (at 187 km from the mine), copper concentrations again increase rapidly in both of the more recent data sets and then, with some scatter, decrease to relatively low concentrations at around 448 km from the mine for the 1993 survey and 392 km for the 2003 survey. Downstream of these locations copper concentrations then increase rapidly to form zones of relatively elevated levels until Everill Junction. Comparison of the 1993 and 2003 data sets suggest that this zone of elevated copper has shifted upstream with time by some 50 km or so and, increased in concentration from an average of 314–695 mg/kg, an increase of 121% over the 10-year period. Downstream of Everill Junction, copper concentrations decrease rapidly to the lowest levels recorded in either survey and continue, with some minor ﬂuctuations, to remain either constant or decrease further through to the estuary. While data are again scarce and often highly variable, it is also illuminating to compare the downstream variation in copper abundance in sediments from the Fly River levees with time. Figure 2.14, which shows the results of two surveys conducted on the levees of the Middle Fly between D’Albertis and Everill Junctions, indicates a substantial increase in the copper content of river-deposited bank sediments with time particularly in the upper part of this 411 km long reach. The plot also indicates that while copper concentrations are highly variable during the mid-90s they typically oscillate around an average of about 890 mg/kg with possibly a trend toward a slight increase in concentration downstream toward the junction with the

86

B. R. Bolton et al.

2500 1994 2003 Linear (1994) 5 per. Mov. Avg. (2003) Linear (2003)

Cu concentration (mg/kg)

D'albertis Jct.

2000

Everill Jct.

1500

1000

500

0 0

50

100

150

200

250

300

350

400

450

500

Distance downstream from Kiunga, km

Figure 2.14: Variation in copper concentration in levee sediments along the Middle Fly River between D’Albertis and Everill Junctions for surveys conducted in 1994 and 2003 (data from OTML 1994, 2003b). Strickland. By contrast, the 2003 data set is again highly variable but with a pronounced decreasing trend in concentration downstream. The average concentration of copper in the levees of the Middle Fly from the 2003 survey is 1083 mg/kg – an increase of over 20% compared to the 1993 survey results. The trend lines for both data sets have been plotted on this ﬁgure to support these observations. Also shown on this ﬁgure is the rolling average for the 2003 data set. This curve, while again giving support to the overall downstream decrease in copper concentration in levee sediments, appears to indicate a bimodal distribution pattern with concentration peaks about 160 and 400 km downstream of Kiunga. This distribution pattern is similar to that described above in relation to the copper concentration in the riverbed sediments and may again reﬂect the dual role of preferential deposition of Cu-bearing sulﬁde minerals in the upper part of the reach and the tendency for increased deposition of silt and ﬁner grained sediments on the levees in the lower Middle Fly in response to the Strickland backwater effect described earlier to explain the prominent textural changes in sediments in this region. We are aware of only two geochemical assays for sediments collected from the levees of the Middle Fly before the start of mining; both reported by Everett and Associates (1980) (Table 2.1). These samples, collected from the upper Middle and lower Middle Fly gave copper concentrations of 95 and 99 mg/kg, respectively. Compared to the 2003 results this represents an increase in the copper content in the order of almost 1000%.

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

87

2.3.2.3. Implications for the risk of acid rock drainage Since mining began sulfur concentrations have progressively increased in sediments deposited from the Ok Tedi and Fly Rivers (Tables 2.1 and 2.2) (Fig. 2.12). This increase in sulfur content has long been predicted as a consequence of the variable but generally high sulfur grade of the Mount Fubilan orebody and the riverine disposal of the mine’s waste (e.g., Pickup et al., 1979). Of particular concern has been the tailings generated by the copper recovery plant located near the mine. Sulfur grades in tailings between late 1999 and mid-2006 averaged close to 3%S, varying on the type of ore being fed to the mill (Table 2.2). Most of the sulfur present in these sediments is derived from pyrite (FeS2), along with generally lesser amounts of other sulfur-bearing minerals such as chalcopyrite and chalcocite, discharged from the mine in both waste rocks and tailings (see following section on sediment mineralogy). Whenever these minerals are deposited from the river in areas that may become exposed to atmospheric oxygen and water, such as river point bars and the crests of levees, there is a risk that they will be oxidized in a process that generates a solution equivalent to dilute sulfuric acid and other potentially harmful metal-rich liquids that are collectively known as ARD. Environmental damage resulting from ARD is well known from many parts of the world and may occur not only with mining activities but anywhere where sulfur-bearing minerals are exposed to the air (Parker and Robertson, 1999). Since 2000 OTML has conducted intensive monitoring of the riverdeposited mine wastes throughout the river system (including the mine wastes recovered from the bed of the Ok Tedi through its dredging operation at Bige) to assess the potential for the development of ARD. This monitoring includes determination of the sulfur content of sediments together with that of a parameter known as the acid neutralizing capacity (ANC) – essentially a measure, in this instance, of the amount of carbonate in the sediment that is available to neutralize any acid formed. With this information it is possible to calculate a third parameter known as the net acid producing potential (NAPP) (usually expressed in units of kg H2SO4/t), which is simply derived according to the following equation: NAPP ¼ MPA ANC where the maximum potential acidity (MPA) is equal to the sulfur content in %S multiplied by a conversion factor of 30.6, commonly used in acid–base accounting (AMIRA, 2002). When the NAPP is a negative number ANC exceeds the MPA and the resultant material is thought to be unlikely to

88

B. R. Bolton et al.

generate acid. In contrast if the NAPP is a positive number then acid generation is considered likely (AMIRA, 2002). Figure 2.15A shows how sulfur concentrations in river-deposited sediment on the levees of the Fly River downstream of D’Albertis Junction have changed with time. In order to improve interpretation of this data the trend lines shown in this plot represent running averages of sulfur concentration in samples collected every 5 nautical miles (9.3 km), in the Middle Fly, or either 5 (2005), 10 (B18.5 km; 2001) or 20 nautical miles (B37 km; 2003) in the Lower Fly. This ﬁgure indicates that sulfur concentrations increased from 2001 to 2003 with the later samples showing concentrations consistently greater than 1%S between D’Albertis and Everill Junctions. By contrast, the trend for samples collected in 2005 indicates sulfur concentrations continued at high levels (in excess of 1%S) in the upper Middle Fly (between about 350 and 436 nautical miles upstream from the estuary mouth) while decreasing, sometimes below 2001 concentrations, in the lower Middle Fly (between about 350 and 214 nautical miles). Downstream of Everill Junction sulfur concentrations in the three surveys depicted in this plot decrease progressively toward the estuary. Figure 2.15B shows the running average for ANC for the three surveys discussed here. While all three surveys show a general trend of decreasing ANC with distance downstream, the ANC of levee sediments has decreased markedly in the Middle Fly, between about 370 and 200 nautical miles, to levels consistently below those of earlier years. Interestingly, peaks in ANC are also apparent at about 430 and 170 nautical mile marks with implications for ARD risk. This change in the acid–base balance with time in the levee sediments is in turn reﬂected in the increase in the number and distribution of NAPP positive samples, along much of the Middle Fly, as shown in Fig. 2.15C. Figure 2.16 is an acid–base plot for samples collected from the levee (right bank only) of the Fly River downstream from D’Albertis Junction from 2001 to 2005 and shows the total sulfur content compared to the ANC. The NAPP positive and NAPP negative zones are identiﬁed on this plot. The plot shows that with time, more samples are plotting in the NAPP positive ﬁeld indicating they are increasingly likely to generate acid. Analyses of these data indicate that since 2003 in excess of 20% of the samples collected from the Fly River levees gave NAPP positive results, peaking at almost 40% in 2004. The results of recent water quality testing of drainage and pore waters from several locations in the sand bars of the upper reaches of the Ok Tedi as well as the levee environments of the lower Middle Fly conﬁrms that sulﬁde oxidation is occurring. Oxidation and neutralizing

A

Everill Jct.

D'albertis Jct.

2.0 1.5 1.0 0.5

percent sulphur

2.5

0.0 50

150

250

350

450

n. miles from mouth of Fly River estuary Mouth

B

D'albertis Jct.

Everill Jct.

120 100 80 60 40

ANC (kg H2SO4/t)

140

20 0 50

150

250

350

450

n. miles from mouth of Fly River estuary

Everill Jct.

D'albertis Jct.

20 10 0 -10 -20 -30 -40 -50 -60 -70 -80 -90

NAPP positive/ potentially acid generating

NAPP negative/ unlikely to generate acid

50

150

250

350

NAPP (kg H2SO4 /t)

C

450

n. miles from mouth of Fly River estuary 4 per. Mov. Avg. (Right Bank (2001)) 4 per. Mov. Avg. (Right Bank (2003)) 4 per. Mov. Avg. (Right Bank (2005))

Figure 2.15: (A) Moving average sulfur concentration. (B) Moving average ANC. (C) Moving average NAPP for levee sediments collected from the right bank of Fly River for surveys conducted in 2001, 2003, and 2005 (data from OTML, 2001, 2003a, 2003b, 2005). Also shown in (C) are the NAPPpositive and NAPP-negative ﬁelds.

90

B. R. Bolton et al.

3.0 NAPP positive

Sulphur (%)

2.5

NAPP=0

2.0 1.5 1.0 NAPP negative 0.5 0.0 0

50 100 ANC (kg H2SO4 /t) 2001 (RB)

2003 (RB)

2004 (RB)

150

2005 (RB)

Figure 2.16: Acid-base plot for sediment collected from the levee of the Fly River right bank (RB) only between D’Albertis Junction and the delta (data from OTML 2001, 2003b, 2004, 2005). Also shown is a line representing NAPP ¼ 0 and the NAPP-positive and NAPP-negative ﬁelds.

products including sulfate and manganese are elevated along with metals such as copper, aluminum, and zinc. pH values as low as 2.6 have been measured in several of these areas together with high levels of acidity, and again indicates acid generation is locally and intermittently produced. Overall, the results to date of monitoring suggest that there is a high risk of widespread acid drainage and elevated copper (and other metals) from the deposited sediments in the longer term. Although high ANC in places is likely to buffer acid generation from most parts of the river system and particularly the levees, for a number of years, observed pyrite and carbonate segregation suggests that localized hot spots could develop in the short term, as has been observed with river-deposited sediments in the upper reaches of the Ok Tedi (Pile et al., 2005) and the lower Middle Fly (OTML, 2005), as described above. In recognition of the risks posed by ARD to the Fly River system ecology, including the communities who live along its length, OTML has recently initiated plans to recover sulﬁde minerals from mine tailings together with other measures, to signiﬁcantly mitigate this risk.

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

91

2.4. Sediment Mineralogy In this section we examine the impact of mining on the mineralogical composition of sediments deposited in the bed and levees of the Ok Tedi and Fly rivers.

2.4.1. Before Mining Prior to the onset of mining, very little work was undertaken on the mineralogy of the Fly River sediments. Everett and Associates (1980) conducted mineralogical analysis on a small number of samples collected in 1979. This study presented information on the distribution of clastic sediment components, quartz, volcanic rock fragments, carbonate rock fragments, mudstone and soil zone clasts from the Kiunga, D’Albertis Junction, and Everill Junction regions of the river system. These data showed that components were variably distributed in the three regions sampled. The Upper Fly contained the highest percentage of quartz in the bed load, while breakup of mudstone rock fragments and soil zone clasts during transport was suggested as the main source of matrix present in both sediments of the bed and levees of the Middle and Lower Fly. These workers also documented the presence of volcanic rock fragments in sand-sized material of the Upper Fly and the Strickland Rivers. They suggested that volcanic ashes from andesitic central PNG volcanoes may also be a local source in the Fly River system due to the presence of orthopyroxene and clinopyroxene, strongly zoned plagioclase phenocrysts, and a general absence of olivine.

2.4.2. After the Start of Mining Mineralogical studies of river sediments since mining began in 1984 are given by Hettler (1995) as part of a study of the environmental impact of mining on the river system (see also Hettler and Lehmann, 1995; Hettler et al., 1997). This work mainly targeted the heavy mineral and clay fractions of ﬂoodplain sediments from the Middle Fly region. These workers note the similarity of the heavy mineral component from the Middle Fly and in particular the sulﬁde component, to that of the sulﬁde mineral assemblage in the Mount Fubilan orebody as well as in riverdeposited tailings in the upper reaches of the Ok Tedi and mine area creeks (Hettler et al., 1997).

92

B. R. Bolton et al.

The clay fraction was also the focus of mineralogical studies conducted on river-deposited sediments by Delft on behalf of OTML in 1987 (Delft, 1987; Salomon and Eagle, 1990). The results of these studies are presented below. However, it is work conducted by OTML as part of its monitoring activities along the river system, which is the basis for the most detailed mineralogical analysis of the system’s sediments and most of this is drawn from the results of a research cruise of the M.V. Western Venturer in May 2003 (OTML, 2003b).

2.4.2.1. Mineralogy of riverbed sediments Quartz (48%) and feldspar (37%) are the dominant minerals of the contemporary riverbed. They occur with generally minor amounts (o10%) of clay, carbonate, mica, hornblende, and opaque minerals (Table 2.3). Also present in trace amounts in some samples, mainly from the Upper and Middle Fly, are clinopyroxene, andalusite, staurolite, and tourmaline. Signiﬁcantly, the sediments of the bed (OTML, 2003b) are typically monomineralic in nature. Particles composed of two or more different minerals or rock fragments were generally rare or absent. Figure 2.17 is a plot of the variation in the estimated proportions of these minerals downstream from Kiunga in both the sediments of the bed and levee. Figure 2.17A shows that between Kiunga and the conﬂuence with the Ok Tedi (D’Albertis Junction), quartz is the dominant mineral of the riverbed (over 80%) with lesser amounts of feldspar (plagioclase mainly but with moderate to trace amounts of K-feldspar in some samples), clays (mostly chlorite), and opaques. Downstream of D’Albertis Junction there is a rapid increase in the relative abundance of feldspar (mainly plagioclase) at the expense of quartz such that within just a few kilometers of the junction Table 2.3: Estimated mineral percentages for o3.32 Sp. Gr. component of Fly River bed sediments during research cruise WV03-11.

n Mean Median SD Min Max

Quartz

Feldspar

43 48 45 13 30 90

43 38 40 11 5 55

Source: OTML (2003b).

Carbonate Hornblende Mica/chlorite 43 2 2 2 0 5

43 o1 o1 o1 0 1

43 3 3 1 1 6

Clay Opaques 43 5 5 3 1 10

43 3 3 2 2 10

60

40

40 30

Strickland R.

10

Agu R.

Ok Tedi

20

Binge R.

30

50

20 10

0

0

A Feldspar

Carbonate

Hornblende

Mica/Chlorite

404

482

589

2

700

Clay

Distance downstream from Kiunga (km)

Opaques

Pyrite

100

100

90

90

80

80

70

70

60

60

50 40

Chalcopyrite

Goethite

Magnetite

Hematite

Ti minerals

Non-Opaques

50 40 30

Strickland R.

10

Agu R.

Ok Tedi

Binge R.

30 20

39 54 73 93 107 123 141 163 176 200 219 237 282 315 348 387 424 482 552 633 700

B

% Mineral

% Mineral

Quartz

123 154 176 209 237 296 348 Distance downstream from Kiunga (km)

Strickland R.

99

Agu R.

73

Binge R.

45

Ok Tedi

2

20 10

0

0 2

44

73

98

124

154

180

209

237

296

348

404

552

Distance downstream from Kiunga (km)

C Quartz

Feldspar

Carbonate

Hornblende

Mica/Chlorite

2

659

44

73

98

Clay

Opaques

124

154

180

209

237

296

348

404

552

659

Distance downstream from Kiunga (km)

D Pyrite

Chalcopyrite

Goethite

Magnetite

Hematite

Ti minerals

Non-Opaques

93

Figure 2.17: Variation in mineral composition in Fly River sediments collected during WV03-11 (OTML, 2003b). (A) Variation in bed material; (B) variation in heavy mineral component of bed material; (C) variation in mineralogy of bank material; (D) variation in heavy mineral component of bank material.

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

50

Strickland R.

70

60

Agu R.

70

Binge R.

80

Ok Tedi

90

80

% Mineral

100

90

% Mineral

100

94

B. R. Bolton et al.

these two minerals occur in almost equal quantities. Chlorite, biotite, amphibole, and traces of calcite and muscovite are also present in this reach. With some minor ﬂuctuation, as evidenced by the scatter on the distribution diagram, mineral abundances thereafter remain more-or-less constant through to the Fly River delta. There is a suggestion in the data of a slight increase in the relative abundance of quartz with respect to feldspar in the reaches between the Binge and Agu Rivers and in the Lower Fly. Heavy minerals are an important component of the river-deposited sediments. Heavy liquid separation test work on bed sediments indicates that heavy minerals vary considerably from reach to reach (Table 2.4) but consistently exceed 2% of the total sediment. Table 2.4 shows that average heavy mineral content of the total sediment is just over 2% in the Upper Fly, almost 7% in the Middle Fly, and 2.7% in the Lower Fly. Signiﬁcantly, one sample from the Middle Fly has a heavy mineral content of close to 30%. Figure 2.18 illustrates the downstream variation in the relative abundance of heavy minerals in the bed material of the Fly River. This plot shows that Table 2.4: Summary statistics for heavy mineral component of Fly River deposited sediment collected during research cruise WV03-11. Reach

n

Mean

Median

SD

Min

Max

Upper Fly Middle Fly Lower Fly

3 33 7

2.2 6.9 2.8

2.0 4.7 2.6

0.7 5.4 0.8

1.6 2.3 1.6

3.0 27.3 4.1

Wt% >3.32 Sp. Gr.Product

Source: OTML (2003b).

D'albertis Junction

30

Everill Junction

25 20 15 10 5 0 0

50 100 150 200 250 300 350 400 450 500 550 600 650 700 Distance downstream of Kiunga, km

Figure 2.18: Variation in the percent total heavy mineral content of bed sediments from Kiunga to the delta.

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

95

heavy mineral content is relatively low in the Upper Fly at between 1.6 and 3%. Just below D’Albertis Junction the heavy mineral content increases over the next few kilometers to a maximum value of 27% (near the village of Nukumba). Heavy mineral abundance then remains high (generally above 5%) for the next 100 km at which point it abruptly decreases to about 5 wt% and then progressively decreases to the river delta. Downstream of Everill Junction heavy mineral content is relatively constant averaging 2.7 wt% – a concentration similar to that seen in a single composite sample collected from the Strickland River (2.6 wt%). Table 2.5 presents a statistical summary of the estimated relative abundance of the heavy minerals identiﬁed in bed materials collected during the May 2003 research cruise (OTML, 2003b). This table indicates that the main heavy minerals present are magnetite (av. 27%) followed by hematite (13%), and the iron sulﬁde mineral, pyrite (9%). The ‘‘non-opaques,’’ which form the dominant component in the greater than 3.32 Speciﬁc Gravity (Sp. Gr.) product, is mainly composed of epidote, augite, and hypersthene. Samples collected from the Middle Fly are usually dominated by epidote while augite is the dominant non-opaque mineral in the Strickland River sample. In addition to the minerals noted above trace amounts of marcasite were observed in almost all samples collected and in the Middle Fly, trace amounts of copper minerals, bornite, chalcopyrite, idaite, covellite, and chalcocite were also identiﬁed. Molybdenite, galena, and pyrrhotite also occur in trace amounts in samples from the Middle Fly. Figure 2.17 above also demonstrates the variation in the proportions of heavy minerals in bed sediments. Figure 2.17B shows that pyrite, and to a lesser extent magnetite and hematite, increase abruptly in abundance over a short distance downstream of D’Albertis Junction reﬂecting input from the Ok Tedi. The relative proportions of these minerals are then maintained, Table 2.5: Estimated heavy mineral percentages for Fly River bed sediments collected during research cruise WV03-11. Pyrite Chalcopyrite Goethite Magnetite Hematite Ti-minerals Non-opaques n Mean Median SD Min Max

43 9 10 6 0 25

Source: OTML (2003b).

43 0 0 0 0 1

43 3 2 3 1 15

43 27 25 5 18 40

43 13 15 4 5 20

42 0 0 0 0 2

43 47 45 10 25 73

96

B. R. Bolton et al.

with some minor ﬂuctuations, over the next 315 km downstream from Kiunga. At this point there is a sudden increase in the relative proportion of pyrite (up to 25%), and to a lesser extent, goethite, mainly at the expense of non-opaque minerals. These high levels are maintained until Everill Junction where there is a rapid decrease, particularly in regard to pyrite, to levels not seen since the Upper Fly (at around 1–3%). The size of pyrite grains, the most important mineral in terms of ARD risk, varies downstream and is clearly inﬂuenced by input of material from the Ok Tedi mine. Figure 2.19 indicates that the maximum size of pyrite after decreasing along the Upper Fly suddenly increases at D’Albertis Junction to a maximum of about 200 mm at 58 km from Kiunga. The size of particles then progressively decreases over the next 200 km and thereafter remains more-or-less constant through to the delta. In addition to the visual estimation of pyrite particle size reported above, the o3.32 and W3.32 Sp. Gr. products of heavy liquid separation test work of some of the samples collected in the WV03-11 cruise (OTML, 2003b) were dry-screened at 500, 355, 250, 212, 150, 106, 75, 53, and 38 mm to determine their size distributions. The results of this work are shown in Figs. 2.20 and 2.21. The size distributions of the W3.32 Sp. Gr. and o3.32 Sp. Gr. fractions were found to be different. The o3.32 Sp. Gr. fraction was coarser than the ‘‘heavies’’ and covered a greater size range. This bimodal distribution 250 Maximum size of Pyrite (microns)

D'albertis Junction

Everill Junction

200

150

100

50

0 0

100

200

300

400

500

600

700

Distance downstream of Kiunga (km)

Figure 2.19: Variation in the size of the largest pyrite grains downstream from Kiunga.

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

97

0119/CB - T13 (124 km)

100.0 Cumulative % Passing

90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 0.01

0.1 Particle Size (microns) Floats

1

Sinks

Figure 2.20: Size distribution plot showing characteristic curves for ‘‘sinks’’ (W3.32 Sp.Gr.) and ‘‘ﬂoats’’ (o3.32 Sp. Gr.) for a sample from the upper Middle Fly, 124 km downstream of Kiunga. between density fractions is not unexpected given that the sample is river sediment and, as such, has undergone natural hydraulic classiﬁcation resulting in co-deposition of particles possessing similar settling velocities. The higher density material is balanced by a ﬁner particle size compared to the lower density particles having coarser particle size. The disparity in size distributions and speciﬁc gravity of the two fractions are typical of mineral mixtures that have similar hydraulic settling velocities during co-deposition. Figure 2.20 is a typical size distribution plot illustrating the characteristic curve for light and heavy fractions found in the Middle Fly. Figure 2.21A shows that for the W3.32 Sp. Gr. products the d50 decreases in the Upper Fly until D’Albertis Junction at which point it increases over a short distance, as seen in Fig. 2.19 above, for maximum pyrite size. Median particle size continues to remain at relatively high levels for the next 200 km or so and then decreases rapidly to about 0.1 mm or less and remains at this level until Everill Junction. Downstream of Everill the d50 increases again under the inﬂuence of sediment from the Strickland River. A similar distribution pattern is also reﬂected in the curves for the d10, d80, and d95, indicating variations are consistent across both ﬁne and coarse fractions. Figure 2.21B also shows the variation in grain size for the o3.32 Sp. Gr. separation products. An almost identical pattern of variation to that seen for the ‘‘heavy’’ products is clearly apparent.

98

B. R. Bolton et al.

0.6 D'albertis Jct.

Grain Size (mm)

0.5

Everill Jct.

0.4

d10 d50

0.3

d80

0.2

d95

0.1 0 0

A

100

200

300

400

600

D'albertis Jct.

0.3

Everill Jct.

0.25 Grain Size (mm)

500

Distance downstream of Kiunga (km)

0.2

d10 d50

0.15

d80 d95

0.1 0.05 0 0

B

100

200

300

400

500

600

Distance downstream of Kiunga (km)

Figure 2.21: (A) Downstream variation in grain size for heavy minerals (W3.32 Sp. Gr.) in bed sediments from Fly River. (B) Downstream variation in the grain size of ‘‘ﬂoats’’ (o3.32 Sp. Gr.).

2.4.2.2. Mineralogy of the levees The mineralogy of samples collected from the crest of the levees during the 2003 research cruise (OTML, 2003b) is presented in Tables 2.6 and 2.7 and the variation in relative abundance of mineral species is shown in Fig. 2.17. Table 2.6 shows that the banks, consistent with the channel material, are dominantly composed of feldspar (49%) and quartz (33%) with minor amounts of clay, carbonate, and mica. Figure 2.17 shows the variation in mineral composition of levee sediments from Kiunga to the delta. As described for the bed sediments the most obvious feature of this plot is the

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

99

Table 2.6: Estimated mineral percentages for o3.32 Sp. Gr. component of levee crest sediments from WV03-11. Quartz Feldspar Carbonate Hornblende Mica/Chlorite Clay Opaques n Mean Median SD Min Max

28 33 30 11 20 65

28 49 55 14 24 65

28 4 3 3 0 10

28 0 0 0 0 1

28 3 2 2 1 10

28 7 5 6 2 30

28 5 4 3 2 15

Source: OTML (2003b).

Table 2.7: Estimated heavy mineral percentages for levee sediments from WV03-11. Pyrite Chalcopyrite Goethite Magnetite Hematite Ti-minerals n Mean Median SD Min Max Strickland

43 27 30 10 0.5 40 25

43 2 1 1 0 3 Tr

43 10 10 4 5 15 10

43 26 25 5 20 40 25

43 5 4.5 4 0 15 10

42 1 0 1 0 5 0

Nonopaques 43 31 30 5 20 40 30

Note: Strickland data from a single sample (OTML, 2003b). Tr ¼ trace.

large increase in relative proportion of feldspar with respect to quartz downstream of D’Albertis Junction. This trend persists as far as Everill Junction at which point the proportion of feldspar rapidly decreases coincident with an increase in the proportion of quartz and clay. The other major feature of the mineral distribution plot is the remarkable uniformity of the composition of the sediments overall, particularly between D’Albertis and Everill Junctions. The relative abundance and distribution of heavy minerals in the levee crests was also examined in this study. Analysis of the overall heavy mineral content of these sediments shows that heavy minerals comprise about 4 wt% of the total bulk sediments. Figure 2.22 shows the variation in total heavy mineral content of the left bank levee sediments collected in the 2003 research cruise (OTML, 2003b), from Kiunga to the delta. As shown for heavy minerals in the bed sediments,

100

B. R. Bolton et al.

D'albertis Junction

Wt% >3.32 Sp. Gr. Product

9

Everill junction

8 7 6 5 4 3 2 1 0 0

50

100 150

200

250

300

350 400

450

500

550 600

650

700

Distance downstream from Kiunga, km

Figure 2.22: Variation in the percent total heavy mineral content of left bank levee sediments from Kiunga to the delta. there is a sudden increase in their content downstream of D’Albertis Junction. This high proportion of heavy minerals persists (over 8% at 237 km downstream of Kiunga), except for a sustained decrease between about 150 and just over 200 km downstream of Kiunga, until about 300 km (just downstream of the conﬂuence between the Agu and Fly Rivers). Thereafter the heavy mineral content progressively decreases, through Everill Junction and onward to a location close to the headwaters of the delta. Table 2.7 presents a summary of the relative abundance of the main heavy minerals identiﬁed in Fly River levee sediments in 2003 (OTML, 2003b). These data show that pyrite and magnetite are the major heavy minerals present in levee crest sediments. Non-opaque minerals such as epidote and augite, together account for the bulk of the heavy minerals in these deposits. This table also indicates that the relative abundance of heavy minerals in the Fly is similar to those seen in the banks of the Strickland River upstream of Everill Junction. The downstream variation in the relative abundance of heavy mineral species in sediments deposited on the levee of the Fly River in 2003 is shown in Fig. 2.18D. This ﬁgure shows that pyrite and hematite increase rapidly downstream of D’Albertis Junction and are maintained at relatively high levels through to a location near the Agu River conﬂuence. Pyrite abundance then slowly decreases at the expense of non-opaque minerals. There appears to be no measurable change to this pattern downstream of Everill Junction under the inﬂuence of the Strickland. As noted above for bed sediments, the heavy mineral fraction of the levee crests also contain trace amounts of a distinctive suite of copper minerals

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

101

including chalcocite, covellite, bornite, and idaite. Marcasite occurs in trace amounts throughout the system, including the single sample from the Strickland River bank.

2.4.2.3. Clay mineralogy

Relative abundance

Clay minerals constitute an important component of both the riverbed and levee deposits. Figure 2.23 shows the variation in clay mineral distribution in bed sediments from Kiunga to the Fly River delta, as determined from samples collected during OTML research cruise WV03-11 (OTML, 2003b). This ﬁgure indicates that clay/phyllosilicate mineralogy in bed sediments, while highly variable over relatively short distances, appears to characterize several river reaches. The Upper Fly is characterized by approximately equal amounts of illite-smectite, chlorite, and mica. Downstream of D’Albertis Junction, while the proportion of each mineral type varies widely over short distances, overall illite-smectite and mica dominate, with lesser and approximately equal amounts of the remaining clay mineral groups. Downstream of Everill Junction the dominant mineral is mica (up to about 60%; mainly biotite) and chlorite (20–30%). This shift toward increasing amounts of mica and chlorite reﬂects the input from the Strickland River. Samples analyzed from the single transect investigated during this cruise averaged 54% mica and 27% chlorite; the composite sample averaged 45% mica and 30% chlorite. One sample contained 80% mica and 20% chlorite. 100 90 80 70 60 50 40 30 20 10 0

D’albertis Junction

0

Everill Junction

50 100 150 200 250 300 350 400 450 500 550 600 650 700 Distance downstream from Kiunga, km

Mica

Chlorite

Smectite

Kaolinite

Illite-smectite

Talc

Figure 2.23: Distribution of clay minerals in bed sediments collected during WV03-11 (OTML, 2003b).

102

B. R. Bolton et al.

Clay analyses were also performed on the samples collected from the levee crest. The results of this work are shown in Fig. 2.24. This ﬁgure shows that upstream of D’Albertis Junction the illite-smectite group of minerals dominates the clay fraction of the river bank sediments. Downstream of D’Albertis Junction, however, the proportion of mica minerals, together with kaolinite increase at the expense of illite-smectite minerals and this trend continues to about 200 km downstream of Kiunga. At this point micas account for approximately 50% of the clay fraction and chlorite replaces kaolinite as the second most abundant clay mineral. These proportions are then maintained, with some variation, through to the delta. Below Everill Junction there is an increase in the proportion of illite-smectite at the expense of kaolinite and smectite as the river approaches the delta. 2.4.2.4. Controls on mineralogical composition

Relative abundance

The composition of the Fly Riverbed sediments is controlled mainly by the composition of sediments delivered by its main tributaries, the Ok Tedi and the Strickland River. Upstream of D’Albertis Junction (conﬂuence with the Ok Tedi), the Fly Riverbed is almost entirely composed of quartz with minor–trace amounts of feldspar, clay, and opaque minerals such as epidote, augite and hypersthene, heavy minerals, mainly magnetite and hematite (Fig. 2.17A). This composition is consistent with an upstream source area dominated by the sandstone, siltstone, mudstone, and limestone units of Mesozoic to Tertiary age found in the Blucher and Victor Emanuel Ranges. 100 90 80 70 60 50 40 30 20 10 0

D’albertis Junction

0

50

Everill Junction

100 150 200 250 300 350 400 450 500 550 600 650 Distance downstream from Kiunga, km

Mica

Chlorite

Smectite

Kaolinite

Illite-smectite

Talc

Figure 2.24: Distribution of clay minerals in bank (left) sediments collected during WV03-11 (OTML, 2003b).

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

103

Downstream of D’Albertis Junction the composition abruptly changes. Plagioclase feldspar rapidly becomes the dominant mineral species in bed sediments together with slightly less abundant quartz. This composition is maintained, with some minor variation, through to the junction with the Strickland River, a further 410 km downstream (Fig. 2.17A). Bed sediments of the Middle Fly are further characterized by the presence of accessory chlorite, biotite, amphibole, and traces of calcite and muscovite. A suite of heavy minerals including magnetite, pyrite, and a variety of copper sulﬁde minerals are also characteristic of this reach (Fig. 2.17B). Signiﬁcantly, this heavy mineral suite, after rapidly increasing at D’Albertis Junction, is maintained at elevated levels relative to both the Upper Fly and Lower Fly reaches. Also signiﬁcant is a sudden increase in the relative proportions of pyrite and goethite, mainly at the expense of non-opaque minerals such as epidote and augite, in the reach between 282 and 456 km downstream from Kiunga. This change in the composition of the heavy mineral fraction coincides with a pronounced change in the grain size characteristics of the bed sediments in this reach with a sudden and persistent decrease in grain size (Fig. 2.6) and the grain size distribution (Fig. 2.7). We interpret this change in the proportion of heavy minerals to reﬂect a decrease in the ﬂow conditions in this reach such that a greater proportion of pyrite along with magnetite are deposited in the riverbed. The mineral assemblage observed in the Middle Fly is very similar to that found in the Mount Fubilan orebody. The orebody is largely composed of a mesothermal stockwork and disseminated copper-gold mineralization hosted in an altered quartz monzonite porphyry stock of Plio-Pleistocene age (Rush and Seegers, 1990). High-grade mineralization also occurs in calc-silicate sulﬁde and magnetite skarns that have formed at the contact between intrusives and host sediments. Overburden which is released into the river system is dominantly composed of limestone and hornfelsed sandstones, siltstones, mudstones. The porphyries are characterized by large and approximately equal amounts of potassium (orthoclase) and sodic (andesine, oligoclase) feldspar, quartz, hydrothermal biotite, hornblende, and accessory apatite, sphene, rutile after sphene, and magnetite. Mineralized areas include a range of opaques, including pyrite, marcasite and pyrrhotite, a variety of copper sulﬁdes (chalcopyrite, bornite, covellite, chalcocite), and hydrocarbonates (malachite, azurite), with the sulﬁdes in various stages of oxidation (digenite, chalcocite, native copper, covellite, goethite, and cuprite). The limestones are predominantly calcitic in composition. Therefore, consistent with earlier workers (e.g., Hettler and Lehmann, 1995), we interpret the composition of bed sediments in the Middle Fly to largely reﬂect the input of Mount Fubilan mine wastes via the Ok Tedi.

104

B. R. Bolton et al.

The main control on the distribution of these minerals along the Fly, downstream of the junction with the Ok Tedi, is the progressive decrease in carrying capacity of the river in the upper part of the Middle Fly resulting from a decrease in the slope of the riverbed brought about by sediment aggradation, and, for the lower Middle Fly, by reduced carrying capacity brought about by the backwater effect during times of high ﬂow in the Strickland. The decrease in carrying capacity in the upper Middle Fly is further evidenced by the coincident decrease in heavy mineral abundance and the maximum size of pyrite grains (Fig. 2.19), as well as the overall d50 of the heavy mineral fraction (Fig. 2.21). Downstream of Everill Junction the composition of the bed sediment varies once more. As shown in Fig. 2.17A, there is a small but persistent increase in the amount of quartz relative to feldspar and more characteristically, a sudden decrease in the proportion of pyrite (and to a lesser extent goethite) at the expense of non-opaque minerals (Fig. 2.17B). The inﬂuence of the main tributaries, the Ok Tedi and Strickland, is also apparent in the composition of the levee sediments. Figure 2.17C shows the banks of the Upper Fly mainly composed of quartz and to a slightly lesser extent, plagioclase feldspar as well as minor amounts of carbonate (mainly calcite) and trace amounts of phyllosilicate (mica and chlorite) and nonopaques such as epidote. Again this is consistent with an upland source area dominated by terrigenous clastic sediments and limestone. Downstream of D’Albertis Junction the composition of the levee crest again changes abruptly with a sudden increase in the proportion of feldspar, mainly at the expense of quartz. Feldspar continues to dominate these sediments (B55–60%) through to Everill Junction where there is a sudden increase in quartz and most noticeably, clay content through to about half way to the estuary. From this location downstream, quartz continues to increase in abundance but mainly at the expense of clay minerals (Fig. 2.17C). The inﬂuence of the Ok Tedi and therefore the mine on the sediment composition is also reﬂected in the heavy mineral composition of the levee crests sediments. Fig. 2.17D shows a rapid increase in the proportion of pyrite as well as goethite downstream of D’Albertis Junction, largely at the expense of magnetite, hematite, and Ti-minerals. These mineral proportions are then maintained more or less, through to about 180 km downstream of Kiunga in the middle part of the Middle Fly. Thereafter the proportion of pyrite progressively decreases at the expense of goethite and non-opaque minerals. In contrast to bed sediments, there does not appear to be a sudden change in levee crest sediment composition due to inﬂux of material from the Strickland River. Close examination of mineralogy variation plots from both the riverbed and levee crests (Fig. 2.17) indicates that apart from the Ok Tedi and

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

105

Strickland River, there appears to be little evidence for signiﬁcant input from the other tributaries joining the Fly, particularly in the Middle Fly reach. The clay mineralogy of deposited sediments collected is largely controlled by input from tributaries, mainly the Ok Tedi and Strickland. As noted above the dominant clay minerals of the Upper Fly are chlorite and mixed species, illite-smectite. Downstream of D’Albertis Junction, following input from the Ok Tedi, there is a rapid increase in the relative amount of smectite at the expense of the illite-smectite group of minerals. Further downstream the clay mineral shows considerable variation over short distances mainly reﬂected in the relative abundance of illite-smectite minerals as a component of the clay fraction (Fig. 2.23). The cause of this high degree of variability is uncertain; however, input from both smaller tributaries entering the Middle Fly together with erosion of deeply weathered bank material including older sediments, may be a factor. The inﬂuence of input of clay minerals from the Strickland River is evident downstream of Everill Junction with the sudden increase in the relative amount of chlorite in bed sediments together with mica at the expense of kaolinite, and illite-smectite group of minerals (Fig. 2.23). Kaolinite is largely restricted to the Upper and Middle Fly in its distribution in bed sediments. This pattern of clay mineral distribution is broadly mirrored in the clay mineral assemblages observed for the Fly River levees (Fig. 2.24). The Upper Fly is again dominantly composed of illite-smectite and chlorite. Downstream of D’Albertis Junction there is a rapid increase in the relative amount of smectite, mainly at the expense of illite-smectite, as seen with the bed sediments, at least in the upper Middle Fly (to about 150 km downstream of Kiunga). Further downstream the clay mineralogy of the levees is relatively constant and dominated by chlorite, smectite, and to a lesser extent kaolinite. Micas are the dominant phyllosilicate from D’Albertis Junction to the delta, averaging about 30–40% of the total. Downstream of Everill Junction, chlorite and to a lesser extent, illite-smectite minerals gradually increase at the expense of kaolinite and smectite. Again, most kaolinite is deposited in the Upper and Middle Fly.

2.4.2.5. Comparison with earlier studies Figure 2.25 shows a comparison between the estimated mineral abundances in data from 1979 and 2003 (Everett and Associates, 1980; OTML, 2003b). This ﬁgure shows that the composition of sediments in the bed of the Upper Fly has changed little with time. These sediments were, and continue to be, dominantly composed of quartz sand. The upper Middle Fly, in contrast,

117 (1979)

O pa

C

qu

la y

ite hl or C M ic a/

H

C

or

ar

nb

bo

le n

na

de

te

r ld sp a Fe

Q ua

Relative abundance (%)

es

Wv03-11-0009/CA (2003)

rtz

Relative abundance (%)

Upper Fly

100 90 80 70 60 50 40 30 20 10 0

Upper Middle Fly

70

108 (1979)

60

Wv03-11-0055/CA (2003)

50 40 30 20 10

es

y

qu

la

a/

O

C

pa

hl

C

or ite

e nd ic M

H

C

or

ar

nb

bo

le

na

sp Fe

ld

ua Q

te

ar

rtz

0

77 (1979)

50

Wv03-11-0325/CA (2003)

40 30 20 10

es

y

qu

M

H

ic

or

a/

O

C

pa

hl

C

or

la

ite

e nd nb

bo ar

Fe

C

Relative abundance (%)

le

na

ar sp ld

ua Q

te

0 rtz

Relative abundance (%)

Lower Middle Fly 60

Lower Fly

60

73 (1979)

50

Wv03-11-0343/CA (2003)

40 30 20 10

es

y

qu

ic a

O

/C

pa

C

hl or

la

ite

e M

H

C

or

ar

nb

bo

le

na

nd

te

r pa ld s Fe

Q

ua

rtz

0

Relative abundance (%)

Strickland River

70 60

75 (1979)

50

Wv03-11-0334/CA (2003)

40 30 20 10 es

y

qu

la

pa O

C M

ic

a/

C

ite hl

en H

or

nb l

bo n ar C

or

de

e at

ar sp Fe ld

Q

ua

rtz

0

Figure 2.25: Plots of relative mineral abundance in bed sediments from the Fly and Strickland Rivers collected during the 1979 and 2003 (Everett and Associates, 1980; OTML, 2003b).

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

107

shows considerable variation with time as evidenced by the change in the relative abundance of quartz and feldspar. In the 1979 survey this reach of the riverbed was dominated by quartz and to a far lesser extent, feldspar and opaques. In 2003 the bed was dominated by feldspar grains with moderate but lesser amounts of quartz particles. The amount of opaque mineral has also increased slightly in the most recent survey. In the lower Middle Fly and Lower Fly, while the proportion of feldspar has increased signiﬁcantly relative to quartz, it is the large decrease in the abundance of clay minerals that characterize these two reaches. This change is interpreted to reﬂect the changing grain size distribution of the lower Middle Fly discussed above. The trend toward increasing amounts of silt-sized quartzo-felspathic material deposited, particularly under the inﬂuence of the Strickland River ‘‘backwater effect,’’ has diluted the dominant clay mineral deposition of pre-mine conditions. As might be expected, there is little evidence of any signiﬁcant change in the composition of the bed materials in the Strickland River. Figure 2.25 shows that while feldspar has increased slightly, along with mica/chlorite and clay, in 2003 compared to 1979, the relative amounts of quartz and carbonate have decreased. What this comparison has shown is that the mineralogy of bed sediments has become dominantly feldspathic in the Middle Fly. Interestingly, there also appears to be an increase in the feldspathic component in the Strickland over the same time interval as well. In terms of the heavy mineral component the 2003 data show a small but signiﬁcant increase in content in the upper Middle Fly as well as the lower Middle Fly. The Lower Fly and Strickland Rivers show a small decrease in the relative abundance of the heavy mineral component. The increase in the relative proportion of heavy mineral in the 2003 survey is consistent with the increasing impact of mine-derived materials on the composition of the bed sediments. The distribution of clay minerals as determined from the 2003 survey compared with that in Delft (1987) is shown in Fig. 2.26. This ﬁgure shows that clay mineralogy has changed considerably since the 1987 survey. The dominant clay of the Upper Fly (and lower Middle Fly) in 1987 was smectite. In 2003 the dominant clay was illite with slightly lesser amounts of chlorite (for the lower Middle Fly also). In 1987, under the inﬂuence of sediment input from the Strickland, the Lower Fly was dominated by illite. This is in contrast to 2003 when the dominant clay was found to be chlorite. Both surveys show illite and to a lesser extent, chlorite to be the most abundant clay minerals of the Strickland. The cause of the change in the Upper and Middle Fly from a clay mineral assemblage dominated by smectite to illite is

108

B. R. Bolton et al.

80 70 60 50 40 30 20 10 0 Upper Fly (1987)

Upper Fly L. Middle L. Middle (2003) Fly (1987) Fly (2003) Smectite

Illite

L. Fly (1987) Kaolinite

L. Fly (2003)

Strickland Strickland (1987) (2003)

Chlorite

Figure 2.26: Comparative plot showing relative clay mineral content in o2 micron fraction of river sediments collected in 1987 (Delft, 1987; Salomon and Eagle, 1990) and 2003 (OTML, 2003b). unclear as is the change from illite to chlorite in the Lower Fly. Further work on the nature and distribution of these clay minerals is required to better understand the controlling processes.

2.5. Conclusions This review has shown that the texture, geochemistry, and mineralogy of the sediments deposited in the channel and on the levee banks of the Fly River system have been distinctly affected by the release of waste rock and tailings from the Mount Fubilan mine. Texturally, the sediments deposited in the channel of the Ok Tedi and Middle Fly have generally become ﬁner since the start of mining in 1984 while levee sediments have coarsened slightly, particularly in the lower reaches of the Middle Fly. Metal concentrations and in particular that of copper, have increased markedly in deposited sediments, in some places by up to an order of magnitude above background levels. Similar increases have also been observed in sulfur abundance in levee sediments down through the system. This has largely been due to an increase in the proportion of mine-derived pyrite and other sulﬁde-bearing heavy minerals deposited during ﬂooding events. This increase in sulﬁde minerals has led in turn to an increase in the risk of acid generation and metal leaching in periodically exposed levee sediments with a subsequent risk of adverse ecological impacts in downstream environments. These impacts, while varying in intensity, can be traced from the creeks surrounding the mine, down through the Ok Tedi and Middle Fly, to the Fly

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

109

River delta – a total river distance of over 1,000 km. The lack of periodic monitoring of sediment characteristics prohibits a clear understanding of the rate of these changes but available data indicate substantial change, particularly in regard to sediment geochemistry, occurring within 10 years of the start of mining and near-pervasive change in all three parameters in the Ok Tedi and Middle Fly by 2003.

ACKNOWLEDGMENTS The authors wish to thank the management of Ok Tedi Mining Limited for their support in the preparation and publication of this chapter. We also wish to thank past and present staff members of the OTML Environment Department who are too many to name individually but without whose tireless efforts in often very trying conditions, the telling of this story would not be possible. Geoff Pickup and Jim Veness are thanked for providing constructive comments on an earlier version of the manuscript.

REFERENCES AMIRA (2002). ARD Test Handbook. AMIRA International, Melbourne. Apte, S. C. (2009). Biogeochemistry of copper in the Fly River. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 321–373. Blong, R. J. (1991). The magnitude and frequency of large landslides in the Ok Tedi catchment, Report to Ok Tedi Mining Ltd, Tabubil, Papua New Guinea, 62 pp. Bowen, H. J. M. (1979). Environmental Chemistry of the Elements. Academic Press, London, 333 pp. Boyden, C. R., Brown, B. E., Drucker, R. F., & Tuft, S. J. (1974). Report of the 1974 Cambridge Expedition to the Western District, Papua New Guinea. 42 pp. Davies, H., Howell, W. J. S., Fardon, R. S., Carter, R. J., & Bumstead, E. D. (1978). History of the Ok Tedi porphyry copper prospect. Economic Geology, 73(5), 796–809. Delft. (1987). The contribution of the Ok Tedi, Fly and Strickland river to sediments deposited in the delta. A feasibility study. Unpublished Report to Ok Tedi Mining Limited, dated July 1987, 11 pp. Dietrich, W. E., Day, G., & Parker, G. (1999). The Fly River, Papua New Guinea: Inferences about river dynamics, ﬂoodplain sedimentation and fate of sediment. In: A. J. Miller, & A. Gupta (Eds). Varieties of Fluvial Forms. John Wiley and Sons, New York, US, pp. 346–376. EGi (1991). Mine Rock Geochemistry and Acid Mine Drainage, Letter Report to Ok Tedi Mining Limited, Tabubil, Papua New Guinea.

110

B. R. Bolton et al.

EGi (1996). Tailings Geochemistry, Unpublished Memorandum to Ok Tedi Mining Limited, dated 14 May, 1996, 6 pp. Everett & Associates (1980). Estimate of geochemical processes and chemical concentrations in the Fly River drainage that may result from development of the Ok Tedi project, Western Province, Papua New Guinea (2 volumes), Unpublished Report to Ok Tedi Mining Limited, Tabubil, Papua New Guinea. Geoff Day & Associates (1996). Particle size and shape of characteristics of deposited sediments in the Middle Fly River, and sediment deposition rates in the tie-channels and off-river water bodies: Volume one – main report, Unpublished Report to Ok Tedi Mining Limited, Tabubil, Papua New Guinea, 20 pp. Gomez, B., Rosser, B. J., Peacock, D. H., Hicks, D. M., & Palmer, J. A. (2001). Downstream ﬁning in a rapidly aggrading gravel bed river. Water Resources Research, 37, 1813–1823. Hettler, J. (1995). Bergbau und umwelt in Papua-Neuguinea: Die Pk Tedi-mine und das Fly River-Flubokosystem. Berliner Geowissenschaftliche Abhandlungen, Reihe A Reihe A, B, 176, 110. Hettler, J., & Lehmann, B. (1995). Environmental impact of large-scale mining in Papua New Guinea: Sedimentology and potential mobilisation of trace metals from mine-derived material deposited in the Fly River ﬂoodplain. South Paciﬁc Regional Environment Programme, SPREP Reports and Studies Series 90: 1–71, Apia, Western Samoa. Hettler, J., Irion, G., & Lehmann, B. (1997). Environmental impact of mining waste disposal on a tropical lowland river system: A case study on the Ok Tedi mine, Papua New Guinea. Mineralium Deposita, 32, 280–291. Higgins, R. J., Pickup, G., & Cloke, P. S. (1987). Estimating the transport and deposition of mining waste at Ok Tedi. In: C. R. Thorne, J. C. Bathurst, & R. D. Hey (Eds). Sediment Transport in Gravel Bed Rivers. Wiley, London, pp. 949–976. Knighton, D. (1998). Fluvial Forms & Processes: A New Perspective. Arnold Publishing, London, 383 pp. Kyle, J. (1988). Water quality and sediment chemistry of Lakes Bosset, Pangua and Daviumbu. In: J. C. Pernetta (Ed.). Potential Impacts of Mining on the Fly River. United Nations Environment Programme Regional Seas Reports and Studies 99 and SPREP Topic Review 33, Nairobi, pp. 9–18. Kyle, J. H., & Ghani, H. (1982). Mercury concentrations in ten species of ﬁsh from Lake Murray. Science in New Guinea, 9, 48–58. Leopold, L. B., Wolman, M. G., & Miller, J. P. (1964). Fluvial Processes in Geomorphology. Freeman, San Francisco. Mackenzie, D. E., & Johnson, R. W. (1984). Pleistocene volcanoes of western Papua New Guinea highlands: Morphology, geology, petrography, and modal and chemical analyses. BMR Report 246, Microform 200. Markham, A. J., & Repp, K. (1992). Erosion and sediment transport in Papua New Guinea. Network design and monitoring. Case study: Ok Tedi copper mine: Erosion and sediment transport monitoring programmes in river basins, Proceedings of the Oslo Symposium, August 1992, pp. 517–526.

Texture, Geochemistry, and Mineralogy of Sediments of the Fly River System

111

Marshall, A. R. (2000). Ok Tedi-Fly River cross-section survey & sediment storage analysis. Epoch 3-June 2000, Unpublished Report to Ok Tedi Mining Limited, Tabubil, Papua New Guinea, 17 pp. Marshall, A. R. (2003). Ok Mani storage estimate, Unpublished Memorandum Report to Ok Tedi Mining Limited, Tabubil, Papua New Guinea, 3 pp. Maunsell & Partners (1982). Ok Tedi Environmental Study: Supporting Studies, Volume 4, Sedimentation analysis and results; chemical effects, Unpublished Report to Ok Tedi Mining Limited, Tabubil, Papua New Guinea, 245 pp. OTML (1986). Ok Tedi mining limited sixth supplemental agreement environmental study: Review of hydrologic data for the Ok Tedi and Fly River catchments, Ok Tedi Mining Limited, Environment Department Report ENV/86-07, Tabubil, Papua New Guinea, 41 pp. OTML (1993). Fly River ﬂoodplain copper monitor (Addendum to the APL Compliance and Additional Environmental Monitoring Report 1993 Annual Report). OTML (1994). Fly River ﬂoodplain copper monitor (Addendum to the APL Compliance and Additional Environmental Monitoring Program: 1993/1994 Annual Report). OTML Report No. ENV 94-14 dated 1994. OTML (1997). APL compliance and additional monitoring program annual report1995/96, OTML Report ENV 97-01. Ok Tedi Mining Limited, Tabubil, Papua New Guinea, 119 pp. OTML (2000). Hydrology Annual APL Report 1998–1999, Ok Tedi Mining Limited, Environment Department Report ENV2000-02, Tabubil, Papua New Guinea, 57 pp. OTML (2001). Hydrology Annual APL Report 1999–2000, Ok Tedi Mining Limited, Environment Department Report ENV2001-02, Tabubil, Papua New Guinea. OTML (2002). Annual Environmental Report FY02, OTML Report ENV020923. Ok Tedi Mining Limited, Tabubil, Papua New Guinea, 155 pp. OTML (2003a). Annual Environmental Report FY03, OTML Report ENV030927. Ok Tedi Mining Limited, Tabubil, Papua New Guinea, 207 pp. OTML (2003b). WV03-11 Cruise Report: Sampling bed and suspended sediment load and levee bank deposits of Fly between Kiunga and ARM 80, OK Tedi Mining Limited, Environment Department Report ENV030530, Tabubil, Papua New Guinea, 13 pp. OTML (2004). WV03-11 Cruise Report: A report on the results of sediment sampling conducted during R.V. Western Venturer research cruise WV03-11 on the Fly River, May 2003, Ok Tedi Mining Limited, Environment Department Report ENV040919, dated 10th September 2004, 80 pp. OTML (2005). TC05-04 Report of Investigation: A Report on the Results of Sediment and Water Sampling and Ecological Soil, Plant, and Fish Sampling Conducted during R.V. Tahua Chief Research cruise TC05-04 on the Fly River, March 2005, 100 pp. Paola, C., Parker, G., Seal, R., Sinha, S. K., Southard, J. B., & Wilcock, P. R. (1992). Downstream ﬁning by selective deposition in a laboratory ﬂume. Science, 258, 1757–1760.

112

B. R. Bolton et al.

Parker, G. (1991a). Selective sorting and abrasion of river gravel: Theory. Journal of Hydraulic Engineering, 117(2), 131–149. Parker, G. (1991b). Selective sorting and abrasion of river gravel: Applications. Journal of Hydraulic Engineering, 117(2), 150–171. Parker, G. K., & Robertson, A. (1999). Acid Drainage. Australian Minerals & Energy Environment Foundation, Melbourne, 227 pp. Pickup, G. (1984). Geomorphology of tropical rivers: 1. Landforms, hydrology and sedimentation in the Fly and lower Purari, Papua New Guinea. In: A. P. Schick (Ed.). Channel Processes – Water, Sediments, Catchment Controls. Catena Verlag, pp. 1–17. Catena Supplement 5: Cremlingen-Destedt. Pickup, G. (2009). Floodplain inundation modeling and forecasting for the Middle Fly. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 291–318. Pickup, G., & Cui, Y. (2009). Modeling the impact of tailings and waste rock disposal on the Fly River system. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 257–289. Pickup, G., & Marshall, A. R. (2009). Geomorphology, hydrology and climate of the Fly River System. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 3–49. Pickup, G., Higgins, R. J., & Warner, R. F. (1979). Impact of Waste Rock Disposal from the Proposed Ok Tedi Mine on the Fly River and Its Tributaries. Papua New Guinea Department of Minerals and Energy and Ofﬁce of Environment and Conservation, Port Moresby, 139 pp. Pile, J. L., Bolton, B. R., Kundapen, H., & Davies, H. (2003). Geochemical characteristics of river deposited mine wastes and associated contact waters downstream of the Ok Tedi mine, Papua New Guinea – Implications for ARD management. In: Proceedings of 6th International Conference on Acid Rock Drainage, The Australian Institute of Mining and Metallurgy, Melbourne, pp. 163–172. Pile, J. L., Bolton, B. R., & Kundapen, H. (2005). Acid forming characteristics, hydrogeochemistry and sedimentology of gravel and sand bars in the upper Ok Tedi, Papua New Guinea. In: 7th Geology, Exploration and Mining Conference, Lae, Morobe Province, Papua New Guinea, May 2005. Rush, P. M., & Seegers, H. J. (1990). Ok Tedi copper-gold deposits. In: F. E. Hughes (Ed.). Geology of the Mineral Deposits of Australia and Papua New Guinea. The Australian Institute of Mining and Metallurgy, Melbourne, Australia, pp. 1747–1754. Salomon, W., & Eagle, A. M. (1990). Hydrology, sedimentology and the fate and distribution of copper in mine-related discharges in the Fly River system, Papua New Guinea. The Science of the Total Environment, 97/98, 315–334. Spencer, L. K. (1974). Sedimentation and geological history of the Fly River region, Papua, MSc. Preliminary Thesis, University of Sydney, Sydney, Australia, 72 pp.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00403-5

Chapter 3

The Rapid Spread of Mine-Derived Sediment across the Middle Fly River Floodplain Geoff Day1, William E. Dietrich2,�, Joel C. Rowland2 and Andrew R. Marshall3 1

Newcrest Mining Ltd., Level 8, 600 St Kilda Road, Melbourne, Victoria, 3004, Australia 2 Department of Earth & Planetary Science, University of California, Berkeley, CA 94720, USA 3 Andrew Marshall & Associates Pty Ltd., 43 Warrangarree Drive, Woronora Heights, New South Wales 2233, Australia

3.1. Introduction The introduction of mining-related waste into the Fly River system by Ok Tedi Mining Limited (starting in 1985) suddenly elevated both the total amount of sediment and the concentration of copper in that sediment. Most of the tailings and the rock waste that was ground to ﬁne sediment during transport traveled rapidly down the Ok Tedi, and entered the Middle Fly River (Fig. 3.1). Annual sediment load on the Fly increased approximately ﬁvefold (Day et al., 2008). Concerns about the consequences of this sediment loading on the system led to several requirements in the 1989 Applied Particulate Level monitoring agreement with the Papua New Guinea government (OTML, 1990). Beginning in 1990 and continuing annually until 1994, a ﬂoodplain sampling program was conducted to quantify the deposition rate and copper concentration of sediments dispersed across the Middle Fly River ﬂoodplain. Given the very low background copper levels in the sediment, the detection of elevated copper not only demonstrated arrival of mine-derived sediment �Corresponding author. Tel.: 510-642-2633; Fax: 510-643-9980;

E-mail: [email protected] (W.E. Dietrich).

114

G. Day et al.

Figure 3.1: Generalized depositional environments of the Fly River and

lower Strickland area (From Day et al., 2008; Journal of Geophysical Research,

American Geophysical Union).

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

115

but also gave us a time marker (1985, i.e., start of dumping) for the sediment accumulation. The initial sampling program called for 200 samples to be taken: 100 from the ﬂoodplain and 100 from off-river water bodies (ORWB) (OTML, 1993a, b). Previous estimates of historical sedimentation rates suggested that the deposition rate on average was about 1 mm yr�1 for the past 10,000 years (Pickup et al., 1979; Dietrich et al., 1999). This was suggested by the presence of pre-Holocene bright red sediment exposed in deep sections of the Fly River bed, the widespread preservation of scroll bars across the ﬂoodplain (Fig. 3.2), and the extensive regions of low elevation (backswamps) bordering the lower Middle Fly. Furthermore, the common observation of brown, sediment-rich ﬂow along the mainstem Fly channel constrained by sediment-free black waters on the ﬂoodplain, even when the river was well above bankfull, suggested that advection of sediment-laden ﬂows across the ﬂoodplain was prevented by the chronically ﬂooded state of the surrounding ﬂoodplain (Dietrich et al., 1999; Day et al., 2008). These expectations of low sedimentation rates led the initial sampling program to collect and analyze only the ﬁrst 1 cm of ﬂoodplain sediments.

Figure 3.2: Aerial view of the forested upper Middle Fly in 1990. Flow is from top to bottom, average channel width is about 320 m, and horizontal distance from top to bottom of the scene is about 4 km (north is at top of image). River entering from east is the Binge (see Fig. 3.1). Note the tie channels connecting the mainstem to oxbows and blocked valley lakes, and the extensive areas of scroll bars.

116

G. Day et al.

Early ﬁndings, however, revealed that the rate and extent of ﬂoodplain deposition of mine-derived sediment was much greater than expected. Hence, we had to modify and amplify our sampling program. We added a systematic survey along 17 transects and took cores up to 1 m deep which allowed us to document the spatial pattern of deposition and the rate more reliably. Our observations also prompted a hydrologic study to document outﬂows and water levels across the ﬂoodplain. Here we report the ﬁndings of the sedimentation study which demonstrate that suspended sediment on the Fly is dispersed over a surprisingly large distance via tie channels and tributaries. This creates what Day et al. (2008) have called a ‘‘depositional web.’’ Many of the detailed observations reported here are summarized in the recent Day et al. (2008) publication. Sedimentation is inﬂuenced by sea-level rise, and Lauer et al. (2008) use numerical modeling to infer that the Fly aggraded only about 10 m from low sea stand during the glacial maximum to present levels. Aalto et al. (2008) report ﬂoodplain deposition rates for the Strickland River based on 210Pb analysis of core samples which compare favorably with rates documented over a shorter time-scale from mine tracer studies (Swanson et al. 2008). These rates are locally much faster than that found on the Fly. This chapter on the Fly River provides more detailed reporting of the sampling strategy and documented sedimentation patterns and then considers the environmental consequences of these observations.

3.2. Methods 3.2.1. APL 200 Sampling Program The annual coring program between 1990 and 1994 was established according to the protocol detailed in the ‘‘Floodplain Copper Monitor’’ section of the ‘‘Acceptable Particulate Level (APL) And Additional Environmental Monitoring Program’’ (OTML, 1990) which describes Ok Tedi Mining Limited’s (OTML) monitoring program on the physical, biological, and geochemical effects of the mine’s ﬂuvial waste disposal. The APL protocol involved establishing a grid cell network dividing the ﬂoodplain into 10 equal-area reaches, or river-zones, by adjusting the north– south dimensions of each zone. This procedure produced river-zones with short north–south extent where the ﬂoodplain is widest. Within each riverzone, ﬁve equal area zone-bands were delineated across the ﬂoodplain by dividing each 1 km northing into ﬁve equidistant units across the ﬂoodplain

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

117

and joining them downriver by interpolation. This created ﬁve zone-bands per river-zone, giving a total of 50 ﬂoodplain zone-bands, or grid cells, between D’Albertis and Everill junctions. From within each of the 50 zonebands, two sediment cores were collected from ﬂoodplain sites and two from ORWB sites (total of 200 samples) and the particulate copper (pCu) in the top 5–10 mm analyzed. Figure 3.3 shows the layout of the zone-band sampling grid along with the site codes for the 200 zone-band core locations; the coordinates of each sample location and the measured pCu are reported in Appendix 1. Samples were only collected within those areas of the zonebands that were east of the Papua New Guinea border with Indonesia. Core samples were collected from a ﬂoat-equipped helicopter due to the difﬁcult access throughout the Middle Fly resulting from the thick vegetation and the large ﬂoodplain area to be sampled. An early model single-frequency GPS receiver (JRL model) with position accuracy of approximately 100 m radius, and visual observation were used to relocate the 200 zone-band sample locations each year. The selection of the two ﬂoodplain sites within each zone-band was governed primarily by helicopter access and where this was not a constraint, cores were generally collected 0.25 and 0.75 of the distance along the length of each zone-band, as close as possible to the middle of the zone-band. Cores from oxbows were generally collected about halfway along the length and in the middle of each oxbow; blocked valley lakes (BVL) were usually sampled in the center of the lake. Depending on water depth at each location, cores were collected using a specially made hand corer or a commercially available KBTM gravity corer. Both corers were constructed of 50 mm diameter stainless-steel tube with interchangeable polycarbonate core liners to aid in sampling and to avoid contamination between cores. The length of cores collected ranged from 0.05 to 1 m. In 1993, the polycarbonate core liners were modiﬁed with a series of 6 mm slits every 20 mm along the length of the liner to achieve greater control over the thickness of collected subsamples. 3.2.2. Transect Sampling Starting in 1992, sampling along prescribed transects was initiated. One transect was located in each of the river-zones deﬁned in the APL survey. Where helicopter access was possible, the spacing of cores along each transect was generally equidistant. Review of the 1992 pCu results showed that the equidistant core spacing did not adequately capture the higher deposition occurring closer to the channel, so the transect program was further modiﬁed in 1993 and repeated in 1994.

118

G. Day et al.

Oxbow lake sample Floodplain sample

PNG

Irian Jaya

Zone 10

Zone 9

Zone 8

Zone 7

Irian Jaya

Zone 6

PNG

Figure 3.3: Floodplain sample locations for the initial APL program. This

shows the APL system of zones and sample strips (delineated by dashed lines).

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

119

Oxbow lake sample Floodplain sample

Zone 5

Zone 4

Zone 3

Zone 2

Papua New Guinea Irian Jaya

Zone 1

Figure 3.3: (Continued).

The 1993 and 1994 revised transect program involved closer core spacing nearer to channels and increasing the spacing between cores further from the channels. Seventeen transects were established, one approximately every 10 km (down valley) along the Middle Fly (Fig. 3.4). Where helicopter access permitted, cores were collected from: the top of the main channel levee, 50, 100, 250, 500, and then every 500 m to the ﬂoodplain boundary along each transect. Depending on ﬂoodplain width and helicopter access, between 5 and 30 cores were collected along each of the 17 transects in addition to the

120

G. Day et al.

Figure 3.4: Map of entire Middle Fly River showing APL points and transect sample locations. Labels (T1, T2, etc.) indicate transect identiﬁca tion numbers. Sample sites not along transects record the location of the original 200 APL sites and the additional oxbows analyzed (reproduced from Day et al., 2008 with permission from the American Geophysical Union). 200 cores from the APL zone-band program. Extra cores were also collected from within the ORWB. Within each oxbow, three cores (in the middle and 10 m from each bank) were collected from three transects across each oxbow (both ends and the center of the oxbow), resulting in nine cores per oxbow. BVL vary signiﬁcantly in size and shape, and additional cores collected from these lakes were based arbitrarily on increasing the number of cores according to lake size. In 1993 and 1994, the transect cores as well as each of the 200 APL zoneband cores were sampled from several horizons to determine the thickness of copper-rich sediment. While the core sample locations remained the same over the 2 years, the horizons from which subsamples were collected varied. In 1993, each core was sampled from, on average, three horizons including: 0–6, 95–101 mm, and the base of the core. If the core was less than 100 mm long,

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

121

then the top and bottom 6 mm was sampled. The pCu results from the 1993 cores revealed that spacing between these horizons was too coarse, and in 1994 the horizons sampled generally included 0–6, 20–26, 60–66, 100–106 mm, and the base of the core. In some instances, particularly at the bottom of the deeper oxbow lakes, the retrieved core was poorly consolidated and comprised a slurry of water and sediment. In such cases, a subsample of the entire slurry was analyzed, representing an average pCu over the length of the collected core. 3.2.3. Sample Variance Test The 840 core locations represent only about 1.5 m2 of the W3 � 109 m2 of ﬂoodplain surface. To determine how representative the core samples were from this low sampling density, in 1991 multiple cores were collected from 10 of the 200 APL zone-band sites sampled in 1990. Based on the 1990 results, ﬁve ﬂoodplain and ﬁve ORWB sites were chosen covering a range of pCu, and at each site 10 cores were collected randomly within a 3 m radius. 3.2.4. Cu Analyses With the various modiﬁcations made to the sampling program between 1990 and 1994, cores were collected from a total of 423 ﬂoodplain and 420 ORWB sites with approximately 4,000 samples being analyzed for pCu. Samples were thawed and wet-sieved through a 100 mm mesh nylon sieve as per the OTML APL protocol (OTML, 1990). The purpose of the sieving was to ensure that the analysis speciﬁcally targeted copper-rich sediment from the mine, given the maximum grain size of the copper-rich mine-derived sediment was o100 mm (OTML, 1989). The sediment passing the sieve was dried at 1041C, digested with hot acid and the pCu analyzed by ﬂame atomic absorption spectroscopy using standard analytical procedures and quality control protocols. All pCu results were reported on a microgram copper per gram of sediment dry weight basis. Note that the 1994 pCu results from the 200 APL zone-band sites are not reported due to an unresolved analytical problem at the OTML Laboratory. 3.2.5. Grain-Size Analysis In 1995, transect cores were collected and used for grain-size analysis. Grainsize analysis of deposited ﬂoodplain sediment was performed using a Coulter

122

G. Day et al.

LS130 Laser Diffraction Analyzer (Beckman–Coulter Corporation). The LS130 calculates grain-size distribution based on the laser diffraction pattern of a sample for all particles in the effective diameter range of 0.4–940 mm. The method is not a direct measurement of sediment grain size, rather it is a statistical approximation based on the diffraction pattern of the whole sample. The method is based on several important operational assumptions in relation to sample homogeneity, shape, and refractive index. It also does not take into account particle density, other than in the pump speed used to keep all particles fully suspended during analysis. Results are reported as volume percent of sediment within 74 size classes, rather than the traditional weight percent within a size class. Operational efﬁciency was a consideration in the choice of using laser diffraction for the grain-size analysis. Other methods of grain-size analysis, such as pipette sedimentation or Sedigraph analysis which may provide more physically relevant information in relation to deposition, were found to be too time consuming for the number of samples analyzed. One of the most critical factors in using the LS130 is sample preparation. Sample pretreatment and analysis required extensive test-work to ensure: (1) particle segregation during sample preparation and analysis, and (2) the subsample analyzed was representative of what was collected in the ﬁeld. During the initial test-work, it was determined that chemical dispersants, which are sometimes used for grain-size analysis to disaggregate cohesive ﬁne-grained sediments, were not required. Extensive testing by the manufacturer and ongoing quality control analysis as part of the ﬂoodplain sample analyses (including certiﬁed reference materials) provided reasonable conﬁdence that the method produced a good approximation to the grain-size distribution of the sediments analyzed. Microscopy of core sample thin sections was also performed as a check on the LS130 results and provided qualitative conﬁrmation on the grain sizes reported. Initial results of ﬂoodplain samples often showed distinctive tails at the upper grain sizes of the distribution. In such cases, the reported grain sizes were so large that it was extremely unlikely that sediment of the size indicated were transported under known hydrodynamic conditions across the ﬂood plain. These same samples also had observable quantities of coarse particulate organic matter (CPOM) consisting of leaf litter, twigs and grass fragments, only some of which could be removed by sieving the sample. These samples were subsequently combusted at 5501C for 8–12 h to determine the quantity of CPOM present and assess if the distribution tails were removed by this process. A large number of ﬂoodplain samples required combustion based on either visual presence of CPOM or where tails in the size distribution plots

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

123

were evident. Combustion was not required for ORWB and river sediment as they generally had insigniﬁcant amounts of CPOM. Suspended sediment and core samples were prepared differently for grainsize analysis. Core samples were air dried, or combusted as required, and transferred to plastic bags where large aggregates were broken apart by hand or by gentle crushing with a roller and then thoroughly mixed in the plastic bag. An aliquot of water was added and a portion of the sediment suspension was transferred to a 50 ml beaker and made up to approximately 40 ml with water. The beakers were then placed in an ultrasonic bath for between 5 and 10 min to disaggregate further the sediment particles immediately prior to analysis. The sediment concentration in the river and tie channel suspended sediment samples was generally too low for the LS130 to measure grain size directly and required preconcentration prior to analysis. The samples, collected in 500 and 1,000 ml plastic bottles, were transferred to 1,000 ml plastic separating funnels and allowed to settle. About every 6–8 h, the settled sediment at the base of the funnel was collected in 50 ml plastic vials. The funnels were then swirled to resuspend particles adhered to the funnel surface, which were then allowed to settle again. This process was repeated 3–5 times. 3.2.6. Calculation of Deposition Rate As described by Swanson et al. (2008) and Day et al. (2008), there are two ways to use vertical proﬁles of tracer metal concentration. The simplest is to assume that the deepest occurrence of elevated metal records the arrival of mine-derived sediment. Dividing the thickness of the overlying core by the time since start of mine-waste discharge gives an annual deposition rate. This was the preferred approach by Swanson et al. (2008) for samples collected on the Strickland ﬂoodplain upstream of the junction with the Fly River. The alternative is to allow for the possibility that sediment, once deposited, may be subjected to bioturbation and physical mixing (associated with mud cracking), leading to anomalously deep penetration of mine-derived sediment and to metal concentrations well below that arriving in the suspended sediment. Vertical variation in metal concentration could also arise from postdeposition dissolution or by temporal variation in arriving suspended sediment concentration. As reported by Day et al. (2008), these latter two effects appear to be unimportant on the Fly. Suspended samples collected near the Ok Tedi Junction indicate pCu values of 800 mg g�1 by the late 1980s, and from 1990 to 1994 more intense sampling of suspended load documented a systematic increase from 900 to 1300 mg g�1. These values, however, are all greatly in excess of the elevated values typically found at

124

G. Day et al.

depth in the core samples. Also, for this period of observation, ﬁeld measurements of pore waters did not indicate copper remobilization post deposition (Day et al., 2008). Hence, we consider proﬁles of pCu that decline signiﬁcantly with depth to be a result of bioturbation. Here we summarize the method described by Day et al. (2008) for calculating deposition rates in which bioturbation is assumed to occur. Day et al. (2008) derive the following expression for calculating the total thickness of copper-rich sediment (LCu) from a vertical proﬁle of pCu measurements (Fig. 3.5): LCu ¼

n X i¼1

zi � rb ð�si � �b Þ rc ð�ci � �si Þ � rb ð�b � �si Þ

(3.1)

εs1 εs2

εs3

z1 z2

Lcu

ρc

z3

Lcu = ∑ ni=1 εs4

zi · ρb (εsi − εb) (ρc (εci − εsi ) − ρb (ρc (εb − εsi )

z4 ρb, εb

Figure 3.5: Schematic diagram of a sediment core with mixing between incoming Cu-rich (green) and pre-mine, Cu-poor (brown) ﬂoodplain sediments. The diagram on the left shows graphically the effect of mixing on measured pCu (color gradation). The diagram on the right provides deﬁnitions of the components of the mass balance calculations used to determine ﬂoodplain deposition rates based on the Cu inventory of the core as described in the text.

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

125

In essence, for each of the subsample length zi, the ratio of the other terms range from zero (when measured sample concentration, �si , equals back ground concentration, eb) to 1.0 (when measured concentration equals the concentration of the incoming contaminated sediment, �ci ). In Eq. (3.1), the background bulk density is rb and the bulk density of copper-enriched sediment is rc. The average bulk density of ﬁve core subsamples from different locations (each 5 cm in length) of both entirely copper-rich sediment deposits and background ﬂoodplain sediments was measured. A mean value of 1.2 g cm�3 was obtained for the copper-rich sediment deposits. Back ground bulk density, varied depending on whether the deposits were mineralsediment dominated or organic-sediment dominated (such as occurs extensively throughout much of the backswamp areas). The average bulk densities of mineral and organic sediments were 1.1 and 0.5 g cm�3, respectively. Based on deep and distant samples, eb for mineral-rich samples is approximately 31 mg g�1 and for organic-rich samples it is 15 mg g�1. To account for variations in suspended sediment concentrations with distance downstream and in time, we varied the value of �ci based on location and depth. The samples nearest to the surface were assigned �ci values corresponding to the year of sample collection and location along the river (Table 3.1). Deeper samples were assigned earlier values, with suspended load concentrations measured in 1991 assigned to all deep samples (Table 3.1). If �sn 4�b (i.e., the core had not sampled all the way through the copper-rich sediment), a value of zi for this particular horizon subsample was rounded up to the next 50 or 100 mm based on a review of the [pCu] through the core proﬁle. In 20 cases, the measured core concentration exceeded the assigned incoming concentration. For these cases, the incoming concentration was used for the core concentration in the sediment. 3.2.7. Topographic Surveys Additional ﬂoodplain cross-sections and spot height surveys were performed in 1997 using dual frequency ZXII Ashtech Global Positioning System Table 3.1: pCu mg g�1 in suspended load samples during the period 1990–1994 at two OTML monitoring stations just downstream from the Ok Tedi River (Nukumba) and just upstream from the junction with the Strickland (Obo). Stations

1990

1991

1992

1993

1994

Nukumba Obo

900 700

900 700

1000 800

1100 1000

1300

1200

126

G. Day et al.

(GPS) receivers in kinematic and static modes. High-resolution static GPS measurements were made for ground control of the aerial photography and at 69 water level recorder sites throughout the ﬂoodplain used to examine ﬂoodplain hydrodynamics (Day et al., 2008). Kinematic GPS measurements were made along each of the 17 ﬂoodplain coring transects. The detailed GPS and aerial survey work for this project was performed by Andrew Marshall & Associates, Sydney, Australia. The satellite imagery used for the GIS base map was a Landsat TM image (path 100; rows 64 and 65; and bands 3, 4, and 5) captured on January 10, 1998. Two sets of aerial photography (1984 and 1992) were conducted by BHP Engineering, Australia on behalf of OTML. The accuracy of each type of survey varied. Vertical accuracy ranged from: 72.0 m (aerial photography); 70.3 to 70.5 m (laser surveys); 70.1 to 70.2 m (kinematic GPS); 70.01 m (locally) to 70.05 m (regionally) for static DGPS; and 70.001 m locally for the level and staff surveys. Horizontal accuracy for each was 50–75% of the vertical accuracy. The GIS was used to examine: (i) the distribution of copper-rich sediment across the ﬂoodplain; and (ii) to model the results of copper-rich sediment thickness from which the total ﬂux of sediment across the ﬂoodplain was calculated. For this GIS work ArcView GIS 3.2 (Environmental Systems Research Institute, Inc.) was used. The ﬁrst step in establishing the GIS was to deﬁne: (i) the ﬂoodplain boundary; and (ii) delineate high-elevation areas within the ﬂoodplain boundary above the level of maximum inundation. A mask (a theme or layer within the GIS) was draped over high-elevation areas above the level of maximum inundation where deposition does not occur. The ﬂoodplain boundary was assigned based on a combination of satellite imagery, aerial photography, and ﬁeld observations of maximum sediment-laden Fly River water inundation. The mask areas were delineated based on ﬁeld observations and elevation data from the 1992 aerial photography. Masked areas included the remnant Pleistocene terraces (e.g., Kuambit and Manda) and the high-elevation areas to the east of the Agu River in the vicinity of Kuem Village, for example. Excluding the boundary and masked regions, the ﬂoodplain area was calculated to be 3,300 km2, which is larger than the 2,600 km2 reported by Pickup et al. (1979).

3.3. Patterns and Rates of Floodplain Deposition of Mine-Derived Sediment The two core sampling protocols described above provide two data sets. Data from the 1990–1993 surface samples from the APL zone-band program

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

127

were used to construct histograms of the number of samples with elevated pCu over time. These data were also used to document the spatial distribution (not total sediment deposition) of copper-rich sediment using the GIS. The 1993 zone-band and the 1993 and 1994 transect depth proﬁle samples provided the data to determine, both numerically and by GIS, the rate of sediment deposition and the percentage of the river’s suspended sediment load deposited across the ﬂoodplain. These analyses are based on pCu samples which had sampling uncertainty associated with each collection. Table 3.2 shows the results of 10 replicate samples at various ORWB and ﬂoodplain samples taken at 10 different locations at the time of the original APL coring program. The coefﬁcient of variation ranged from 2 to 108% with an average of 42%. This is a signiﬁcant uncertainty, but as subsequent analysis shows, this variance does not obscure strong spatial and temporal trends; hence, we think our conclusions are quantitatively sound.

3.3.1. Spatial and Temporal Pattern of the Spread of Mine-Elevated pCu Figure 3.6 shows the frequency of occurrence of pCu for samples in the ORWB sites and across the ﬂoodplain for the 200 APL samples collected annually during the period 1990–1993. In 1990, about 70% of the ﬂoodplain samples were at, or close to, background pCu (o30 mg g�1) values, reﬂecting the early stages of copper-rich sediment deposition across the ﬂoodplain even though the mine had been operating since 1984 (Fig. 3.6a). Between 1990 and 1993, the left-skewed distribution progressively shifted to a strongly rightskewed distribution as copper-contaminated sediment spread throughout the ﬂoodplain. In contrast, the initial 1990 ORWB data show a clear bimodal distribution of pCu reﬂecting background (pre-mine) sediment and copperrich sediment (Fig. 3.6b). Even by 1990, approximately 75% of ORWB sites had pCu elevated above pre-mine conditions and this increased to more than 95% by 1992. This records the rapid dispersion via tie and tributary channels. The temporal change in the spatial distribution of copper-rich sediment is mapped in Fig. 3.7. The data shown are for pCu at the sediment surface (approximately ﬁrst 6 mm), not the deposition thickness or rate. The data show that in 1990, elevated pCu sediments were generally conﬁned to core locations in the forest reach and sites close to the river (Fig. 3.7a). In just three years (by 1993), this pattern of deposition changed dramatically (Fig. 3.7b). Mine-derived sediment by this time had traveled great distances, in some cases tens of kilometers up tributary channels and out tie channels, dispersing widely across the ﬂoodplain. Note that for the upper half of the

128 G. Day et al.

Table 3.2: Results from the analyses of 10 replicate samples from 10 different locations from the Original APL

Floodplain Coring Program throughout the Middle Fly River ﬂoodplain.

Site

Raw data (mg/g) dry weight

Mean

s.d.

c.v. (%)

W10 W87 W42 W24 W25 16 98 95 31 42

547, 1003, 1011, 318, 953, 1237, 1441, 792, 1321, 722 306, 972, 744, 884, 832, 1018, 1032, 719, 260, 258 8, 55, 86, 10, 34, 26, 46, 77, 26, 40 52, 427, 117, 112, 41, 36, 111, 161, 566, 55 777, 844, 783, 476, 319, 838, 882, 856, 242, 746 818, 774, 792, 781, 777, 768, 806, 747, 792, 780 272, 322, 380, 55, 669, 235, 603, 326, 84, 1162 1256, 1287, 1311, 1318, 1277, 1294, 1291, 1269, 1280, 1149, 1229 52, 52, 84, 67, 75, 80, 53, 69, 60, 73 10, 8, 17, 14, 14, 22, 10, 14, 18, 18

934 702 41 168 676 784 411 1269 66 14

350 313 26 180 238 20 328 49 12 4

38 45 64 108 35 2 80 4 18 30

Note: The water body (ORWB) site numbers are preceded by W.

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

129

40 a. 35

Floodplain samples

Number of samples

30

90 91 92 93

25 20 15 10 5 0 10

100

1000

pCu (ug/g) 40

b.

35

Number of samples

30

90 91 92 93

ORWB samples

25 20 15 10 5 0 10

100

1000

pCu (ug/g)

Figure 3.6: Histograms of pCu in (a) shallow ﬂoodplain cores and (b) offriver water bodies (ORWB) between 1990 and 1993. maps shown in Fig. 3.7, the western side of the Fly, lying in Indonesian territory, was inaccessible and no samples were collected. Note, too, that the 1994 pCu results from the 200 APL zone-band sites are not reported due to an unresolved analytical problem at the OTML Laboratory.

130

G. Day et al.

Figure 3.7: Distribution and magnitude of pCu in ﬂoodplain surface samples along the Middle Fly River in 1990 and 1993. Comparison of the two maps highlights both the downstream progression of pCu distribution along the Middle Fly and the increase in pCu across the entire ﬂoodplain system. The junction of the Middle Fly and Strickland rivers is located in the bottom right of each image.

3.3.2. Rates of Deposition Across the Floodplain The average pCu of core surface samples decreased exponentially as a function of distance over the ﬁrst 1 km from the nearest channel bank (mainstem, tie channel, tributary, or ORWB) for the transect samples collected in 1993 and 1994 (Fig. 3.8). At the channel bank, the concentra tions were similar to the incoming values of the suspended load. By 1 km from the bank, values approached background concentrations, but remained above estimated background values of 15–31 mg g�1 well across the

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

131

10000 1993 Transect data 1994 Transect data

pCu (ug/g)

1000

100

10

1 0.0

5.0

10.0

15.0

20.0

25.0

30.0

Shortest distance to channel bank, km

Figure 3.8: Surface pCu in samples collected along the 17 transects in 1993 and 1994. Data from distal tie channels are not included. ﬂoodplain. The data plotted in Fig. 3.8 do not include low values found near tie channels, which are at great distance from the mainstem Fly. For the 1993 and 1994 transect surveys, deposition rate of sediment based on the application of Eq. (3.1) also decreases exponentially for the ﬁrst 1 km (Fig. 3.9). The exponential ﬁt was chosen in part due to its use in other studies of overbank deposition rates (Bridge, 2003). The data shown represent cumulative deposition from the start of mining to 1993 and 1994, respectively, and not an annual rate. For the distance beyond 1 km from the nearest bank, the cumulative deposition shown in Fig. 3.9 is between 0 and 1 mm since mining commenced with an average cumulative deposition of 0.5 and 0.7 mm by 1993 and 1994, respectively. We grouped deposition thickness versus distance data into two classiﬁcations (reach types and nearest channel types) and tested for signiﬁcance of regression statistics for the ﬁrst 1 km of deposition. The reach type consisted of forest, transition, or swamp (Fig. 3.1). We hypothesized that the forest reach may experience the greatest deposition rate because of visibly stronger overbank ﬂows across the ﬂoodplain, although one could also argue that the swamp reach, with longer periods of ﬂooding might trap more sediment. Our second grouping was by the channel type closest to the sample site: the mainstem Fly River, a tie or tributary channel or any channel type (no distinction made). We expected the overbank deposition rate to be less on tie and tributary channels because they carry sediment away from the mainstem source. Separate exponential functions for the ﬁrst 1 km of rapid decline were derived for each of

G. Day et al.

Equivalent depth of mine-derived sediment (mm)

132

1000

1993

100

10

1

0.1 0.01 0.001 0

1

2

3

4

5

6

7

8

9

10

Equivalent depth of mine-derived sediment (mm)

Shortest distance from a source channel (km) (a)

1000 1994 100 10

1

0.1 0.01 0.001 0

1

2

3

4

5

6

7

8

9

10

Shortest distance from a source channel (km) (b)

Figure 3.9: Cumulative deposition of sediment by (a) 1993 and (b) 1994, based on the 17 transect samples. the sorted datasets for both 1993 and 1994 data. Linear regression analysis of each of the above test variables (reach and channel type) was performed by comparing the slope and intercept of the exponential functions using the one-tailed t test for slope and two-tailed t test for the intercept (Zar, 1984, Chapter 18). Signiﬁcance was determined at the Po0.05 level. Table 3.3 summarizes the regression statistics of exponential ﬁts for the ﬁrst 1 km distance from the channel. The exponential functions describe the

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

133

Table 3.3: Regression results of ﬁts of exponential function, d ¼ ae�bx , for all samples collected in 1993 and 1994 distinguished by ﬂoodplain type (forest, transition, swamp) and closest channel (mainstem Fly or ﬂoodplain channel, i.e., tie channels, tributaries, TC, and off-river water bodies, ORWB). a (mm)

R2

b (1/km)

Forest Trans. Swamp Forest Trans. Swamp Forest Trans. Swamp 1993 Data All channels Fly R. only TC/ORWB only

243 243 245

165 169 159

139 128 179

3.79 3.78 3.82

3.69 3.66 3.60

3.23 2.92 3.77

0.96 0.95 0.94

0.97 0.96 0.98

0.85 0.81 0.89

1994 Data All channels Fly R. only TC/ORWB only

288 307 204

169 187 119

122 113 187

3.64 3.62 3.36

3.18 3.20 2.70

2.45 2.04 3.59

0.92 0.91 0.94

0.90 0.87 0.93

0.75 0.70 0.93

Note: The distance from the channel, x, is in kilometers, and a is deposition in millimeters since start of mine loading.

rate of decline in deposition with distance from a channel. The expression takes the form, d ¼ ae�bx , in which d is the total sediment deposition (in millimeters) since the start of elevated mine-derived sediments delivered to the system, a is the deposition at the bank (x ¼ 0), b has units of km�1, and x is the distance from the channel bank (in kilometers). Deposition decreases more rapidly with larger values of b (due to the negative sign). Linear regression analysis comparing the slopes and intercepts of the exponential functions were used to determine if differences existed between the test variables (reach and channel type). The regression tests for 1993 showed signiﬁcant differences in deposition between ﬂoodplain reach type, i.e., deposition in each reach is described by a separate exponential function, and the rate of deposition decreases from forested reach to swamp grass. Tests for channel type showed no signiﬁcant difference between any of the channel groups in the forest or transition reach. In the swamp reach, however, the exponential function representing the tie channels differed from that for mainstem Fly River. If accurate, this difference indicates that the amount of ﬂoodplain deposition from the tie channels is greater than that from the mainstem Fly as indicated by the higher values of the exponential constants for the tie channel compared to the mainstem Fly (Table 3.3).

134

G. Day et al.

In comparison, the regression tests for 1994 showed no signiﬁcant differences between any of the test groups (reach or channel type). The reasons for the apparent difference between 1993 and 1994 statistical results may be due to the decreased sample size in 1994 (n ¼ 146) compared to 1993 (n ¼ 276), and the resultant decreased statistical power of the analysis (note ‘‘n’’ represents the number of ﬂoodplain samples within 1 km of a channel and does not include oxbow samples or samples W1 km from a channel). The decreased sample number in 1994 is a consequence of an unresolved analytical problem in the 1994 APL zone-band samples. To calculate the total deposition on the ﬂoodplain based on the best-ﬁt exponential relationships, we use a GIS model in which the appropriate deposition-distance function is applied to each reach of the ﬂoodplain. Based on the regression analysis of the 1993 data, the following exponential functions were used as input to the GIS model for calculating the deposition thickness within 1 km of any channel: d ¼ 243e�3.79x d ¼ 165e�3.69x d ¼ 128e�2.92x d ¼ 179e�3.77x tributary.

for the forest reach, for the transition reach, for the swamp reach within 1 km of the Fly River, and for the swamp reach within 1 km of a tie channel or

While no signiﬁcant difference existed in deposition between reach or channel type in the 1994 data, the same groups used for the 1993 calculations were used for 1994 to ensure consistency between years. The GIS model was run using the following exponentials for 1994: d ¼ 288e�3.64x d ¼ 169e�3.18x d ¼ 113e�2.04x d ¼ 187e�3.59x tributary.

for the forest reach, for the transition reach, for the swamp reach within 1 km of the Fly River, and for the swamp reach within 1 km of a tie channel or

Beyond 1 km from a channel, uniform deposition thicknesses of 0.5 and 0.7 mm were used for 1993 and 1994, respectively. Figure 3.10 shows the calculated pattern of sediment deposition and sample locations that were used to derive the local regressions. A graded color scheme is applied to illuminate the exponential function, which ends at 1 km. The ﬁgure highlights how channel paths across the ﬂoodplain dictate the pattern of sediment deposition. The cumulative volume of copper-rich sediment deposited on the ﬂood plain between the commencement of mining and 1993 calculated from the GIS model was 1.11 � 108 m3 (133 Mt based on the bulk density of 1.2 t m�3

Figure 3.10: Predicted pattern of deposition from the GIS model of the exponential functions (1993) for the Middle Fly ﬂoodplain. The magnitude of deposition decreases from dark orange to light orange with distance from the channel; white areas are regions of the ﬂoodplain outside the inﬂuence of the depositional web or the channel system. Sample locations are shown as crosses. The margins of ORWB are outlined in gray. The channel network is shown at low water level to reveal the extensive network that extends into the blocked valley lakes (as shown in Fig. 3.1).

136

G. Day et al.

Figure 3.10: (Continued)

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

137

for copper-rich sediment deposits). For the 1994 model, the calculated volume was 1.37 � 108 m3 (164 Mt). A sensitivity analysis was performed to determine how these loads varied in response to minor changes in the exponential functions. Using the forest reach exponential as an example and adjusting the deposition value at the levee, either higher (to 275 mm) or lower (to 200 mm), such that the R2 value of the exponential function changed by no more than 0.05, the GIS model recalculated the deposition volumes. For 1993 the sensitivity range was 0.97 � 108 to 1.30 � 108 m3, and for 1994 the range was 1.21 � 108 to 1.57 � 108 m3. The sensitivity analysis shows that these small changes in the exponential functions alter the GIS-calculated copper-rich sediment load by about 10–15%. We propose that this error is equal to or greater than that caused by local variance in individual samples, though the coefﬁcient of variation is larger (Table 3.2). The values of 1.11 � 108 m3 and 1.37 � 108 m3 deposited up to 1993 and 1994 equate to an average deposition across the ﬂoodplain within 1 km of a channel of 42 and 52 mm, respectively, or approximately 5–6 mm yr�1 from 1986 which is the year when the mine-waste input exceeded the natural sediment load of the Middle Fly (Day et al., 2008). As a check on the deposition values derived from the GIS model, estimates of the total volume of deposition were calculated manually. Using the appropriate constants from Table 3.3 for each reach and channel type, and integrating the exponential function from 0 to 1 km, we can estimate an average deposition thickness ðDCu Þ for each reach. Multiplying DCu by twice the measured channel lengths in each reach (to account for deposition from each bank) provides the total thickness of copper-rich sediment deposited across the ﬂoodplain. The results of the above calculations are presented in Table 3.4 and show the calculated volumes deposited up to 1993 and 1994 as 1.56 � 108 and 1.81 � 108 m3, respectively, i.e., approximately 25% higher than the value derived from the GIS model. The average deposition ranged from 9.6 mm yr�1 in the forested reach to 6.0 mm yr�1 in the swamp grass reach. The higher rate arises for at least two reasons: (1) this calculation does not take into account those areas of the ﬂoodplain above the level of maximum inundation (i.e., the masked areas in the GIS) where deposition does not occur, and (2) in the GIS model, where curvature causes projected deposition to overlap, only one value is chosen, whereas this calculation essentially assumes the channels are straight. Given these differences, the similarity of values supports the GIS results. Sediment deposited directly into the ORWB (and blocked valley lakes) via tie and tributary channels is not accounted for in this GIS analysis. Measured copper-rich thickness from all cores in each oxbow were averaged and applied

138

G. Day et al.

Table 3.4: Calculation of total ﬂoodplain deposition based on the product of the 1 km average deposition rate (from integration of exponential expressions whose coefﬁcients a and b are reported). Reach

Channel Exp. constants DCu (mm) from Table 3.4 (a)

Channel length (km)

Deposition (m6 � 106)

(b)

1993 Data Forest Transition Swamp Swamp Total

Closest Closest Fly River TC & OB

243 165 128 179

3.79 3.69 2.92 3.77

63 44 42 46

1,120 1,150 320 410

70 51 13 19 153

1994 Data Forest Transition Swamp Swamp Total

Closest Closest Fly River TC & OB

2.88 169 113 187

3.64 3.18 2.04 3.59

77 51 48 51

1120 1150 320 410

86 59 15 21 181

over the oxbow area. This volume multiplied by a bulk density of 1.1 t m�3 results in an estimate of 6 Mt of sediment deposited across those oxbows that were cored up to 1994. The cored oxbows cover an area of approximately 24.9 km2 of the approximately 34.3 km2 of total area occupied by oxbows across the Middle Fly. Therefore, including the additional area of oxbows not cored, the estimate for the total mass of copper-rich sediment in all oxbows is about 8.5 Mt by 1994. As a check on this estimate, we alternatively calculate oxbow deposition by assuming that Fly River water with an average sediment concentration of 500 mg l�1 enters oxbows at an average ﬂow of 25 m3 s�1 10– 20% of the time (Day et al., 2008), to obtain an estimate of 11–19 Mt by 1994. The oxbows cores under sample the extent of deposition because the samples rarely penetrated beyond the mine-contaminated level while the ﬂux calculation estimate is crude. As such, the agreement between the two approaches is encouraging. The estimated range of oxbow deposition (11– 19 Mt) represents an average deposition thickness over the 34.3 km2 of oxbows of between 32 and 55 mm yr�1, which is approximately an order of magnitude higher than that reported for the ﬂoodplain core sites. Adding the copper-rich sediment thickness from the ﬂoodplain, BVL and oxbow core results (164 Mt from the ﬂoodplain and BVL cores and between 8.5 and 19 Mt into oxbows) equates to a total mass of copper-rich sediment

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

139

Table 3.5: Sediment budget of the Fly River. Tailings, waste rock, and wall erosion (erosion due to waste dumping) deﬁne the input to the Ok Tedi River. Year

Tailing (t � 106)

Waste rock (t � 106)

Wall erosion (t � 106)

Sediment loada (t � 106)

Sediment loadb (t � 106)

1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 Total

2.38 3.10 9.72 14.93 23.60 27.42 27.01 26.74 28.62 29.71 193

3.80 9.50 11.80 23.40 29.70 27.80 27.80 21.96 14.03 27.47 197

0 0 0 0 0 1 19 24 16 22 82

17 22 32 51 53 62 49 51 56 62 455

7 12 22 41 43 52 39 41 46 52 355

Source: Data supplied by Ok Tedi Mining, Ltd.

a The total suspended sediment load entering the Middle Fly at D’Albertis Junction from the Ok Tedi and

Upper Fly.

b The total mine-derived sediment input to the Middle Fly (assuming an annual background load to the

Middle Fly of 5 � 106 t each from the Ok Tedi and Upper Fly).

deposited across the entire ﬂoodplain system up to 1994 of between 172 and 183 Mt. Table 3.5 shows the amount of sediment discharged to the ﬂuvial system from the mine between 1985 and 1994, and the calculated suspended sediment load entering the Middle Fly River at D’Albertis Junction (Day et al., 2008). If we simply divide the range of values by the total load since 1985 (455 Mt), the fraction of the suspended load that is deposited on the Middle Fly ﬂoodplain is about 38–40%. Delivery of signiﬁcant mine-derived sediment did not begin until 1987, so dividing by this shorter duration, instead of the ten-year total, gives a 41–44% deposition in the ﬂoodplain. The difference between the total sediment deposited on the ﬂoodplain between 1993 and 1994 of 31 Mt is the estimated deposition rate for the year 1994. According to Table 3.5, the suspended load entering the Middle Fly River in 1994 was 62 Mt; hence, our measurements suggest that as much as 50% of the suspended load was deposited on the ﬂoodplain. About half of the ﬂoodplain deposition occurs via overbank ﬂows from the ﬂoodplain channels (and injection into ORWB) and the other half from overbank ﬂows from the mainstem channel. Rate of ﬂoodplain deposition may have been increasing during this ﬁrst half of the 1990s. Table 3.5 shows that from 1988, the delivery of mine waste

140

G. Day et al.

to the system has been 4–5 times that of the natural load, whereas prior to 1988 it was less than or equal to twice the natural load. Channel bed aggradation along the Fly River was occurring at this time and advancing downstream from Kuambit. By 1994, the bed of the Fly River had aggraded from approximately 2–3 m at Kuambit to 1–2 m at Wygerin. Beyond Wygerin, by 1994, the amount of aggradation was apparently minimal (o1 m). While the timing of the onset of aggradation in the Middle Fly is not known with any certainty, it appears reasonable to assume that it was low in the early years of mining, perhaps until at least 1988 to 1990, coinciding with the increased delivery of mine waste to the system. The effect of aggradation of the Fly River bed would be to increase the frequency of overbank ﬂow events and the frequency at which Fly River water is directed out along tie channels. Aggradation between Kuambit and Wygerin (both within the forest reach) leading to increased overbank transfer of water and sediment is also consistent with the results of the exponential functions. Between 1993 and 1994, the greatest change in deposition was observed in the forest reach as evident from the relatively higher increase in the exponential constant a (the intercept), which increased by approximately 14% compared to the transition and swamp reach, which only increased by 3–5% between 1993 and 1994. 3.3.3. The Grain Size of Floodplain Sediment Figure 3.11 summarizes the size distribution of sediment in the suspended load and in surface sediment from the channel bank top to well across the ﬂoodplain. Several striking features are apparent. As expected, the levee is sand-rich (about 22%) relative to either the suspended load or to sediments farther across the ﬂoodplain. The sand content drops immediately away from the levee, but unexpectedly, does not drop to zero, but instead remains around 5%. We made thin sections of distal samples and conﬁrmed the presence of this sand. The sand content on the channel banks of the mainstem Fly is greater than the banks of its tie and ORWB. Also, surprisingly, the ﬁne silt and clay content (o8 mm) is greatest within the ﬁrst 10–50 m of the channel bank and then progressively declines across the ﬂoodplain. This decline is compensated by a progressive increase in coarse silt (16–32 mm). The high o8 mm concentration near the channel bank and decline across the ﬂoodplain may inﬂuence the calculated sediment deposi tion rate, as the pCu is greatest on the ﬁner particles (Fig. 3.12). This may cause the calculated sediment deposition rate near the channel bank to be somewhat overestimated.

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

141

Figure 3.11: Plots of grain-size distributions of surface ﬂoodplain deposits (ﬁrst B10 mm) with distance from channels. Measured grain sizes are plotted in ﬁve bin classes from o8 mm to sand size. The ﬁrst panel presents data for all channel types and presents a representative size distribution of the suspended sediment load of the Middle Fly River. The other two panels present data separated by whether samples bordered the Fly River or tie channels and ORWBs. The greatest percentage of sand are found closest to the channel (0–0.01 km), though sand is present at all distances from the channel. There is surprisingly little change in the size distribution with distance from the channel, and the data suggest that a slight coarsening of the distribution may occur away from the channel. Samples were collected in 1995, and depending on the proximity to channels they are a mixture of background and mine-derived sediments.

142

G. Day et al.

Figure 3.12: Variation in pCu as a function of grain size in samples collected at various locations along the Middle Fly.

3.4. Discussion As Day et al. (2008) propose, and as illustrated in Fig. 3.10, suspended sediment is spread great distances across the Fly River ﬂoodplain via tie and tributary channels, forming a depositional web. This network of ﬂoodplain channels is over 900 km and creates an efﬁcient trap for sediment. Rates of dispersion and deposition far exceed expectations when the original 1989 APL sampling program was designed. While similar processes likely occur on other rivers, as Day et al. (2008) review, this appears to be the ﬁrst systematic documentation of it. The trapping efﬁciency of 40% or more of the total suspended load by the middle Fly ﬂoodplain is equivalent to about 0.09% sediment deposition per kilometer of mainstem channel length, but drops to 0.03% if the length of the full channel dispersal system is used. As Day et al. (2008) report, this number is less than on the Strickland (0.05%), Brahmaputra–Jamuna (0.07%), Amazon (0.065–0.1%) but is comparable to that on the Mississippi (0.02%). So the rate loss per unit length is modest, but the extensive ﬂoodplain channel system leads to large net loss. Importantly, Day et al. (2008) report a multiyear hydrologic study of the middle Fly ﬂoodplain system which concluded that as much as 40% of

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

143

the total annual discharge of the Fly may enter the ﬂoodplain system (approximately one-half via ﬂoodplain channel outﬂow). This ﬁnding adds considerable support to the sediment deposition rate analysis. Based on suspended sediment measurements on the middle Fly before mining and during this monitoring period, the ten-year average load increase was about 4.6 times relative to pre-mining rates. Given the modest amount of channel aggradation by 1994, we infer that the deposition rate on the ﬂoodplain was probably increased by a comparable amount. Hence, if the mean aggradation per year across the ﬁrst 1 km of ﬂoodplain was 5–6 mm in 1994, then the background was about 1 mm yr�1, supporting previous interpretations by Pickup et al. (1979) and Dietrich et al. (1999). The exponential decline in deposition rate occurs over 1 km, but the levee banks are much narrower, typically 56–93 m on the mainstem and 43–58 m on the tie and tributary channels (with heights of less than 2 m). Furthermore, the grain size across the ﬂoodplain does not systematically decline from the channel bank, but instead has a peak silt and clay content close to the bank. As Day et al. (2008) summarize, these observations contradict theory for levee formation and grain-size distribution across the ﬂoodplain (James, 1985). There are several implications of our ﬁndings for predicting the long-term fate of the Fly River ﬂoodplain in response to sustained mine-waste discharge to the system. First, mine-derived sediment will be widely dispersed across the system, with greatest deposition bordering the mainstem and ﬂoodplain channels (as well as in the ORWB). Signiﬁcant amounts of metals will be deposited on the ﬂoodplain. For example, by 1994, perhaps as much as 131,000 tonnes of copper had been delivered to the ﬂoodplain (164 Mt times a mean pCu of 800 mg g�1). The greatest deposition rate will be on the levees. Subsequent topographic monitoring of levee heights by OTML between 1997 and 2002 (OTML, 2004) showed that the height increased by 1 m in 6 years, which means that the rate of levee growth increased signiﬁcantly after our monitoring program ended. The fastest average levee growth we documented occurred between 1993 and 1994 and was only 45 mm in 1 year or 270 mm in 6 years, a much slower rate. The acceleration recorded in the repeat topographic surveys may be due to a combination of increased mine-waste loading, La Nina ﬂooding, and channel aggradation (causing more frequent and higher ﬂooding levels). It is not clear how levee growth may affect ﬂoodplain channel dynamics, or whether, it may lead to new processes, such as frequent crevasse splay formation. Rowland et al. (2005) used OSL dating to document that in two oxbows the rate of tie channel advancement into the oxbows increased by 6–17 times since the early 1980s to mid 1990s. This increase in rate corresponds to

144

G. Day et al.

increased mine loading. By 2005, at least one major tie channel (just upstream from Obo) had completely plugged with sediment. Given the crucial role the tie channels play in connecting the mainstem to ﬂoodplain ecosystems, the fate of these channels and the rapid accumulation of minederived sediment in the oxbows should be a concern. Modeling of bed material transport on the Fly suggests that there will be a slow migration of a mine-derived sediment aggradational front downstream. Frequency of overbank deposition and ﬂushing of sediment out of tie and tributary channels should increase with aggradation. As Dietrich et al. (1999) pointed out, the lower Middle Fly channels, despite having meandering form, showed no evidence of lateral bank migration for decades. If the aggradational front persists through this reach, and forces the development of bars, lateral migration may be reactivated. Bank migration will cut into overbank sediment rich in mine-derived sediment. Hence, in this lower reach, as well as along the entire upstream Middle Fly River, lateral migration will cause mine-derived sediment to remain a signiﬁcant load in the river well after mine closure. Given the active distributary-like ﬂoodplain channel network, it is also a concern that progressive bed aggradation could lead to the potential for channel avulsion. Figure 3.13 shows the pattern of elevation along the east side of 4 of the 17 transects (also see Fig. 3.4). The ﬂoodplain generally decreases in height away from the channel, especially once it leaves the elevated scroll-bar zone (Fig. 3.1). Note that there are some very low elevations in the peripheral 14 transect 4

Elevation (m AHD)

12

transect 9 transect 12

10

transect 15 8

6

4

2

0

5,000

10,000

15,000

20,000

25,000

30,000

Distance from channel bank (m)

Figure 3.13: Elevation proﬁles along four of the transects shown in Fig. 3.4.

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

145

ﬂoodplain zone. For example, along the distal portions of transect 9, elevations reach as low just 4 m above sea level. Such signiﬁcant crossﬂoodplain topographic gradients set up favorable conditions for avulsion (Slingerland and Smith, 2004) and the potential for this happening on the Fly, given continued bed aggradation, deserves consideration.

3.5. Conclusions During the ﬁrst 10 years of mine operation, mine-derived sediment, enriched in pCu traveled quickly down the Ok Tedi and spread rapidly over great distances across the Fly River ﬂoodplain. This rapid and extensive dispersal of sediment was caused by the over 900 km of tie and tributary channels which serve as avenues for suspended sediment pumped out from the mainstem Fly. The sediment traveled along the channels and then spilled out across their banks. By 1994, mine-derived sediment was found across the entire ﬂoodplain in areas adjacent to channel sources. This overbank deposition decreased exponentially from levee top out to 1 km, beyond which very little mine-sediment arrived, and whose deposition showed no spatial pattern. This depositional web of sediment had dispersed about 180 Mt of sediment to the ﬂoodplain by 1994, depositing over 130,000 tonnes of copper, with about half deposition occurring via outﬂow along the ﬂood plain channels. All ORWB connected to the mainstem via tie or tributary channels were immediate recipients of mine-derived sediment. The consequence of this broad and efﬁcient dispersal system is that about 40% (or more) of the total load delivered to the middle Fly River was deposited on its ﬂoodplain before reaching the junction with the Strickland. The environmental legacy of this deposition is manyfold. Nearly all ORWB have large amounts of mine-derived sediment. Levee heights have grown signiﬁcantly with potential morphodynamic consequences. Minederived sediment has traveled tens of kilometers up tributaries, dispersing the sediment widely. An exponentially thinning package of mine-derived sediment borders the channels and will remain a source of mine-related sediment due to lateral migration long after mine closure. Tie channel advance has quickened and plugging has occurred. Lateral migration is accelerated by bar growth due to increased sediment supply (Dietrich et al., 1999). The combination of a propensity for outﬂow on tie channels and tributaries, large areas of low elevation, accelerated channel shifting, and channel bed aggradation driven by mine-derived sediments may create the potential for channel avulsion.

146

G. Day et al.

ACKNOWLEDGMENTS

This work was supported by Ok Tedi Mining Limited (OTML). We especially thank OTML environment managers, Murray Eagle and Eric Woods. Cliff Riebe assisted in ﬁeldwork. Analysis of the data was also supported in part by the National Center for Earth-Surface Dynamics and an NSF Margins Source to Sink grant. Deanna Sereno and Dino Bellugi performed GIS analysis of ﬂoodplain deposition and Mark Stacey provided valuable guidance. The paper beneﬁted from a review by Geoff Pickup and careful editing by Barrie Bolton.

REFERENCES Aalto, R., Lauer, J. W., & Dietrich, W. E. (2008). Spatial and temporal dynamics of sediment accumulation and exchange along Strickland River ﬂoodplains (PNG), over decadal-to-centennial time scales. Journal of Geophysical Research, 113, F01S04. doi:10.1029/2006JF000627. Bridge, J. S. (2003). Rivers and ﬂoodplains: Forms, processes, and sedimentary record. Blackwell Publishers, UK, 491 pp. Day, G., Dietrich, W. E., Rowland, J. C., & Marshall, A. (2008). The depositional web on the ﬂoodplain of the Fly River, Papua New Guinea. Journal of Geophysical Research, 113, F01S02. doi:10.1029/2006JF000622. Dietrich, W. E., Day, G., & Parker, G. (1999). The Fly River, Papua New Guinea: Inferences about river dynamics, ﬂoodplain sedimentation and fate of sediment. In: A. J. Miller, et al., (Eds). Varieties in Fluvial Form. Wiley, New York, pp. 345–376. James, C. S. (1985). Sediment transfer to overbank sections. Journal of Hydraulic Research, 23(5), 435–452. Lauer, J. W., Parker, G., & Dietrich, W. E. (2008). Response of the Strickland and Fly River conﬂuence to post-glacial sea-level rise. Journal of Geophysical Research, 113, doi:10.1029/2006JF000626. Ok Tedi Mining Ltd (OTML) (1989). Supplemental Environmental Investigations. OTML, Tabubil, Papua New Guinea. Ok Tedi Mining Ltd (OTML) (1990). APL Compliance and Additional Environ mental Monitoring Program. OTML, Tabubil, Papua New Guinea. Ok Tedi Mining Ltd (OTML) (1993a). APL compliance and additional environmental monitoring program: 1992 annual report. OTML Report ENV 93-04. Ok Tedi Mining Ltd (OTML) (1993b). Fly River ﬂood-plain copper monitor. June, ENV 93-06.

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

147

Ok Tedi Mining Ltd (OTML) (2004). Summary report: Effects of mine life extensions and dredging on the Ok Tedi and Fly River–A sediment transport model study. Environment Section, September 10, 68 pp. Pickup, G., Higgins, R. J., & Warner, R. F. (1979). Impact of waste rock disposal from the proposed Ok Tedi mine on sedimentation processes in the Fly River and its tributaries. Report to Department of Minerals and Energy and Ofﬁce of Environment and Conservation, Papua New Guinea, 138 pp. Rowland, J. C., Lepper, K., Dietrich, W. E., Wilson, C. J., & Sheldon, R. (2005). Tie channel sedimentation rates, oxbow formation age and channel migration rate from optically stimulated luminescence (OSL) analysis of ﬂoodplain deposits. Earth Surface Processes and Landforms, 30(9), 1161–1179. Slingerland, R., & Smith, N. D. (2004). River avulsions and their deposits. Annual Review in Earth and Planetary Sciences, 32, 257–285. Swanson, K. M., Watson, E., Dietrich, W. E., Apte, S., Lauer, J. W., Aalto, R., Bera, M., Marshall, A., & Taylor, M. (2008). Decadal sedimentation rates on the ﬂoodplain of the Strickland River, Papua New Guinea, Journal of Geophysical Research, 113, F01S03, doi:10.1029/2006JF000623. Zar, J. H. (1984) Biostatistical Analysis. 2nd Edn., Prentice Hall, New Jersey.

Appendix: Results of Applied Particulate Level (APL) Annual Survey. Analyses are from Shallow Cores on the Floodplain and in Off-River Water Bodies (Sample Code Starts with W) Sample code 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Sampling location

Benthic Cu (mg Cu/g of sediment)

Latitude 1S

Longitude 1E

1990

1991

1992

1993

1994

6-10.43 6-22.96 6-20.89 6-21.15 6-13.02 6-16.37 6-16.88 6-16.30 6-16.72 6-17.14 6-22.45 6-26.77 6-26.01 6-25.72 6-30.68 6-26.20 6-36.82

141-10.16 141-06.93 141-04.28 141-04.28 141-04.74 141-02.88 141-00.78 141-03.76 141-04.65 141-05.80 140-58.61 140-56.69 140-55.38 140-55.64 140-56.68 140-57.24 140-55.17

26 31 19 18 592 26 30 193 48 30 112 50 64 56 46 800 61

19 61 30 45 610 41 21 548 13 82 200 578 167 47 34 736 88

17 350 38 42 n/s 125 47 564 280 194 611 341 434 53 171 649 236

23 826 75 24 n/s 475 40 950 500 716 456 680 456 523 137 692 670

32 691 30 36 1455 1019 48 987 772 526 983 693 591 370 102 1312 165

148

G. Day et al.

Appendix 1: (Continued ) Sample code 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59

Sampling location

Benthic Cu (mg Cu/g of sediment)

Latitude 1S

Longitude 1E

1990

1991

1992

1993

1994

6-36.61 6-36.63 6-36.62 6-25.70 6-31.08 6-34.09 6-38.13 6-41.61 6-51.79 6-52.62 6-52.37 6-51.60 6-55.49 6-56.55 6-58.11 7-33.06 7-35.31 7-34.43 7-33.12 7-32.41 7-30.06 7-30.95 7-32.18 7-30.82 7-29.51 7-26.88 7-27.06 7-24.13 7-21.90 7-21.61 7-21.78 7-24.04 7-22.53 7-23.03 7-19.76 7-19.02 7-17.48 7-18.22 7-17.21 7-17.08 7-17.01 7-16.73

140-52.66 140-50.79 140-52.52 141-00.26 140-59.97 140-58.43 140-53.04 140-51.28 140-54.79 140-58.55 140-59.67 141-01.03 141-02.22 141-02.26 141-02.12 141-21.37 141-18.77 141-16.73 141-17.06 141-17.48 141-16.53 141-18.26 141-14.73 141-11.80 141-08.57 141-10.57 141-11.86 141-16.36 141-16.75 141-14.56 141-13.86 141-15.20 141-10.16 141-09.43 141-08.86 141-13.11 141-15.28 141-16.57 141-17.15 141-13.92 141-12.44 141-10.26

71 679 49 39 61 54 80 65 67 63 71 113 40 112 30 52 10 9 10 42 63 83 10 19 10 10 10 30 32 58 62 21 10 10 19 157 21 19 20 20 22 21

42 n/s 37 33 43 60 331 124 233 182 93 163 658 213 72 43 23 36 62 174 248 319 37 23 35 20 13 105 63 119 103 68 25 52 75 124 34 59 19 30 264 475

221 724 165 41 96 61 545 151 478 213 539 265 536 321 625 84 43 95 64 227 570 383 34 61 40 42 42 66 94 227 151 177 100 38 50 149 71 41 49 107 187 149

444 837 356 164 41 50 236 427 737 225 688 1116 1109 233 1066 351 321 110 135 192 30 688 193 116 1109 15 18 27 69 340 95 49 137 33 91 93 247 63 166 37 619 708

41 1009 126 n/s 95 41 269 150 1011 266 274 64 1199 725 58 38 39 32 35 184 859 1550 39 12 16 18 17 26 54 97 134 75 105 17 61 140 20 18 29 26 614 1265

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

149

Appendix 1: (Continued ) Sample code 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101

Sampling location

Benthic Cu (mg Cu/g of sediment)

Latitude 1S

Longitude 1E

1990

1991

1992

1993

1994

7-14.65 7-15.59 7-10.65 7-12.63 7-12.45 7-12.95 7-12.78 7-07.72 7-06.59 7-10.14 7-07.68 7-08.96 7-05.43 7-05.36 7-04.91 7-03.63 7-02.58 6-59.26 7-00.16 7-02.28 7-00.15 7-00.64 6-56.53 6-55.52 6-56.65 6-51.15 6-51.04 6-51.13 6-50.77 6-50.84 6-46.72 6-45.92 6-47.77 6-47.78 7-24.22 6-44.18 6-40.35 6-31.30 6-32.90 6-28.68 6-17.03 6-14.43

141-07.26 141-05.29 141-04.34 141-06.28 141-08.70 141-12.13 141-13.59 141-12.79 141-11.05 141-13.26 141-07.59 141-07.69 141-03.67 141-02.84 141-05.69 141-08.59 141-10.79 141-09.34 141-08.42 141-06.09 141-01.99 141-04.19 141-06.57 141-07.20 141-07.62 141-04.12 141-00.40 140-58.86 140-56.68 140-54.07 140-55.92 140-54.83 140-52.89 140-53.54 141-12.39 140-53.67 140-55.11 140-56.00 140-55.71 140-57.42 141-00.86 141-04.36

10 20 10 20 14 14 27 12 15 14 28 25 11 13 12 73 13 13 9 50 20 31 20 21 20 32 63 20 32 135 21 21 36 264 29 443 108 42 233 64 301 598

26 43 23 23 45 27 48 75 33 51 69 56 22 28 75 68 63 51 114 120 37 340 70 61 48 63 91 114 27 210 42 91 144 77 37 61 141 160 350 180 75 392

27 48 41 36 45 36 52 45 24 50 54 107 77 41 752 39 60 87 292 132 104 301 575 108 175 191 85 176 188 63 300 188 671 591 62 81 155 963 941 562 234 310

28 343 49 120 673 444 109 168 58 695 s/s 86 139 220 684 215 179 212 213 417 218 1911 1119 706 190 116 139 286 334 58 635 635 920 845 96 1122 169 147 836 569 564 628

29 10 18 21 39 31 20 36 10 23 58 119 15 6 38 90 47 85 n/s 414 33 185 114 40 29 231 142 93 205 438 652 110 88 47 61 91 234 78 1193 1330 70 908

150

G. Day et al.

Appendix 1: (Continued ) Sample code W-1 W-2 W-3 W-4 W-5 W-6 W-7 W-8 W-9 W-10 W-11 W-12 W-13 W-14 W-15 W-16 W-17 W-18 W-19 W-20 W-21 W-22 W-23 W-24 W-25 W-26 W-27 W-28 W-29 W-30 W-31 W-32 W-33 W-34 W-35 W-36 W-37 W-38 W-39 W-40 W-41 W-42

Sampling location

Benthic Cu (mg Cu/g of sediment)

Latitude 1S

Longitude 1E

1990

1991

1992

1993

1994

6-11.55 6-12.58 6-14.86 6-18.07 6-21.25 6-20.81 6-20.38 6-20.92 6-17.55 6-11.12 6-16.52 6-17.33 6-20.41 6-21.28 6-24.43 6-24.65 6-25.00 6-30.05 6-28.02 6-35.52 6-36.41 6-36.63 6-35.57 7-32.80 7-32.39 7-32.50 7-31.77 7-31.41 7-32.53 7-30.52 7-30.41 7-29.42 7-29.01 7-25.89 7-27.91 7-27.58 7-27.44 7-22.31 7-21.85 7-22.01 7-22.53 7-22.77

141-08.53 141-07.47 141-07.69 141-04.51 141-02.79 140-59.59 140-59.40 140-59.29 141-00.38 141-01.68 141-02.01 141-02.28 140-58.59 140-58.97 140-58.37 140-58.35 140-58.48 141-00.98 141-03.15 140-54.62 140-54.27 140-54.78 140-55.35 141-18.37 141-17.04 141-15.92 141-13.15 141-11.52 141-07.50 141-09.73 141-11.96 141-15.67 141-16.19 141-15.87 141-12.55 141-11.96 141-11.77 141-11.55 141-11.35 141-10.52 141-10.57 141-10.05

302 854 35 994 57 205 617 70 933 521 716 717 931 115 642 640 641 31 30 767 535 417 1134 146 912 43 43 24 33 19 33 54 553 540 52 44 10 105 189 65 65 10

1066 671 50 923 48 36 888 954 1401 51 709 751 938 637 632 659 701 39 43 67 38 1021 941 528 215 334 43 32 42 29 188 215 484 634 390 31 29 698 199 342 605 48

1268 828 62 748 1009 838 1564 1264 519 1399 995 843 898 460 975 896 831 46 36 1009 649 1084 1061 998 1002 466 107 154 72 70 359 49 750 734 78 55 51 776 895 143 687 18

1240 683 38 1334 206 543 963 1321 912 1443 963 1095 1152 1857 1087 885 11041 28 130 230 620 784 1031 827 468 299 352 183 37 160 424 525 876 876 238 154 306 1027 326 716 478 40

1622 1276 127 1015 64 520 299 1734 75 1345 1556 1487 1239 795 1267 781 866 52 44 105 505 96 1286 1107 1218 185 40 43 32 35 131 361 1368 47 173 37 40 1131 640 642 919 144

Rapid Spread of Mine-Derived Sediment across Middle Fly River Floodplain

151

Appendix 1: (Continued ) Sample code W-43 W-44 W-45 W-46 W-47 W-48 W-49 W-50 W-51 W-52 W-53 W-54 W-55 W-56 W-57 W-58 W-59 W-60 W-61 W-62 W-63 W-64 W-65 W-66 W-67 W-68 W-69 W-70 W-71 W-72 W-73 W-74 W-75 W-76 W-77 W-78 W-79 W-80 W-81 W-82 W-83 W-84

Sampling location

Benthic Cu (mg Cu/g of sediment)

Latitude 1S

Longitude 1E

1990

1991

1992

1993

1994

7-21.96 7-14.51 7-15.16 7-14.29 7-14.54 7-19.40 7-18.18 7-18.42 7-15.16 7-14.78 7-14.13 7-12.55 7-12.31 7-10.49 7-11.74 7-10.16 7-09.05 7-09.62 7-07.15 7-07.64 7-07.99 7-03.83 7-02.50 7-02.79 7-01.96 7-03.15 7-00.13 7-01.28 7-03.21 7-05.54 7-06.22 6-58.18 6-57.66 6-56.79 6-54.51 6-53.75 6-52.69 6-53.26 6-54.56 6-57.45 6-58.64 6-43.02

141-09.35 141-02.85 141-05.11 141-04.95 141-05.81 141-15.54 141-15.20 141-16.96 141-14.07 141-09.89 141-07.22 141-05.00 141-09.95 141-12.34 141-14.78 141-14.51 141-15.87 141-10.01 141-10.04 141-07.66 141-05.59 141-11.60 141-09.94 141-09.84 141-07.83 141-06.63 141-06.20 141-01.60 141-03.31 141-02.04 141-05.00 141-05.14 141-05.56 141-04.81 141-03.40 141-02.55 141-01.38 141-05.22 141-08.42 141-06.66 141-08.42 140-51.68

17 51 29 39 125 28 33 21 31 810 40 40 220 109 74 64 36 39 524 304 42 53 119 81 289 52 712 222 88 230 127 72 66 994 n/s 47 60 209 556 492 908 671

33 44 70 224 503 152 23 23 550 958 625 127 68 42 757 533 155 29 294 708 161 127 95 618 747 689 611 63 145 172 682 35 77 158 69 42 100 259 548 325 206 150

150 50 150 100 563 299 52 28 605 916 849 97 492 105 542 591 88 109 636 45 150 869 389 654 831 811 803 476 693 666 864 50 151 1285 202 81 150 211 479 374 783 586

363 60 179 340 670 1050 333 62 1362 1889 323 93 1492 115 476 354 61 97 427 1306 763 557 488 766 846 947 820 1695 n/s 727 285 146 104 1867 294 240 146 503 460 365 485 483

145 68 130 127 127 358 36 65 841 965 640 201 204 76 330 511 66 101 602 1432 169 88 659 875 640 480 1135 758 320 825 868 70 136 1311 65 65 83 201 852 339 519 105

152

G. Day et al.

Appendix 1: (Continued ) Sample code W-85 W-86 W-87 W-88 W-89 W-90 W-91 W-92 W-93 W-94 W-95 W-96 W-97 W-98 W-99 W-100

Sampling location

Benthic Cu (mg Cu/g of sediment)

Latitude 1S

Longitude 1E

1990

1991

1992

1993

1994

6-43.48 6-45.16 6-46.60 6-48.09 6-51.19 6-51.03 6-51.80 6-50.98 6-44.24 6-24.75 6-31.90 6-32.50 6-36.63 6-37.01 6-42.09 6-42.25

140-54.08 140-58.26 140-53.80 140-53.56 140-56.68 141-00.68 141-02.99 141-04.86 140-53.42 140-57.69 140-58.97 140-56.19 140-53.15 140-52.88 140-52.01 140-52.50

192 192 1446 692 41 244 64 117 815 600 30 158 998 61 82 126

251 78 988 635 856 99 174 389 622 681 45 41 988 491 183 59

990 161 635 683 923 199 116 244 938 1009 29 504 871 952 321 494

813 191 s/s 978 1210 167 75 282 493 1233 153 72 1042 82 170 732

1450 412 1606 1288 1037 100 168 533 1055 1012 38 1024 1188 998 256 719

43 22

69 34

101 50

126 64

92 46

Number of samples W200 mg/g % Samples W200 mg/g Notes: n/s, not sampled; s/s, spilled sample.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00404-7

Chapter 4

Processes, Sediments, and Stratigraphy of the Fly River Delta John P. Walsh1,� and Peter V. Ridd2 1

Department of Geological Sciences and the Institute for Coastal Science and Policy and Coastal Resources Management Program, 301B Graham Building, East Carolina University, Greenville, NC 27858, USA 2 Marine Geophysics Laboratory, School of Mathematical; and Physical Sciences, James Cook University, Townsville, Queensland 4811, Australia

4.1. Introduction Voluminous rainfall and rugged, tectonically active mountains are two primary ingredients that make Indo-Paciﬁc islands incredible suppliers of sediment to the ocean. It is estimated that six islands alone (Papua New Guinea (PNG), Sulawesi, Borneo, Sumatra, Java, Timor) supply 20–25% of the total annual sediment load transported to the ocean globally (Milliman et al., 1999). Many rivers draining these islands discharge onto broad, low-gradient continental shelves characterized by large tidal ranges, and these are locations for tide-dominated deltas. The Fly River of PNG is one example. The Fly River delta is a large funnel-shaped estuary with three major distributary channels; its morphology is a classic example of a tidedominated delta (Fig. 4.1; Coleman and Wright, 1975; Galloway, 1975; Wright, 1985). The delta is approximately 100-km wide at the coast and its three main distributary channels merge at Sumogi Island around 100 km inland. Although the Fly River and its delta are not as large as the Amazon or Ganges–Brahmaputra rivers, it is signiﬁcant by world standards. The system has a high sediment yield (1,118 t km�2 year�1, Harris et al., 1993), �Corresponding author. Tel.: (252) 328-5431; Fax: (252) 328-4391;

E-mail: [email protected] (J.P. Walsh).

154

J. P. Walsh and P. V. Ridd

Figure 4.1: Map of the Fly River drainage basin. The Fly River drains the Papuan Fold Belt with mountains exceeding 4,000 m in elevation, and it ﬂows across the Papuan Foreland Basin in southwestern Papua New Guinea. Note, the location of Fig. 4.2 is shown.

and the delta receives a large natural annual sediment load (85 � 106 tonnes; 17th in the world). In fact, the Fly River carries more sediment to the sea than all the rivers draining Australia combined (Harris et al., 1993). Interestingly, the average daily tidal water ﬂux within the estuary is about 18 times greater than the river discharge (Spencer, 1978). Sediment transport associated with tidal, ﬂuvial, and marine processes is responsible for the complex erosion and deposition patterns throughout the delta, making portions of this environment highly dynamic. Because of the large size and energetic tidal nature of the delta, the signal of sediment preserved in the stratigraphic record and leaving this part of the dispersal system is altered considerably in space and time. This paper is a digestion of the research conducted in the Fly River delta, much of which was supported by Ok Tedi Mining Limited. In this review of the physical processes and sediments of the Fly River delta, we will discuss the geologic setting, ﬂuvial characteristics, hydrodynamic processes, sediment transport, and the resulting stratigraphy. Also, this overview identiﬁes and discusses topics for future research.

The Fly River Delta

155

4.2. Geologic Setting and Fluvial Characteristics The island of New Guinea is located immediately north and only a short distance from the Cape York Peninsula of northeastern Australia (Fig. 4.1). The Fly River drainage basin lies on the southern side of the island. Between 1842 and 1846, Captain F.P. Blackwood aboard the H.M.S. Fly ﬁrst surveyed the mouth of the river (Jukes, 1847), and D’Albertis (for whom the junction of the Ok Tedi and Fly rivers is named) was the ﬁrst explorer to travel the full navigable length of the Fly River in 1876 (Harris et al., 1993). Geologically, the Fly River drains a foreland basin that lies immediately southwest of the Papua Fold Belt. The basin has developed since the Oligocene in response to loading of material onto the Australian plate associated with subduction of the Paciﬁc plate; evidence for variable subsidence in the basin is provided by thickness differences in sediments across the Papuan Basin (Veevers, 1984; Pigram et al., 1989). Approximately 30% of the drainage basin is in the mountains of the Papuan Fold Belt (Dietrich et al., 1999). This portion of the basin is extensively faulted and has rocks of varied lithology including, partly metamorphosed Jurassic volcanics and clastics, metamorphosed and unmetamorphosed Cretaceous sediments, Miocene limestones, and Quaternary volcanics (D’Addario et al., 1976). After leaving the mountains, the Fly River ﬂows across a long, lowgradient ﬂood plain, B850 km to the delta mouth (Dietrich et al., 1999). En route, the Fly River meets the larger Strickland River at Everill Junction. The mean annual discharge of the Strickland River is about 1.4 times that of the Fly River at Obo (2,244 m3 s�1), and the Strickland River contributes as much as 80% to the total Fly River sediment load (Harris et al., 1993). Nonetheless, the Fly River keeps its name below the junction. As river ﬂow increases downstream, the hydrograph is more stable and less sensitive to individual rain events. Annually, the hydrograph remains relatively steady; greatest discharge variability is associated with interannual climate change associated with the El Nin˜o Southern Oscillation (Walsh et al., 2004). For example, in 1996 the discharge ﬂuctuated roughly between 2,000 and 3,000 m3 s�1 while in 1997 the river was below 2,000 m3 s�1 for about three-quarters of the year and below gauge level (o1,000 m s�1) for about 50% of the year (Walsh et al., 2004). Tidal inﬂuence on the river can be observed upstream of Everill Junction (Dietrich et al., 1999). The gravel–sand transition in ﬂuvial sedimentation is located near the conﬂuence of the Ok Tedi and Fly Rivers. The majority of the system is bedded with medium sand. Bed grain size generally becomes ﬁner along its course as the river slope decreases (Pickup, 1984; Dietrich et al., 1999;

156

J. P. Walsh and P. V. Ridd

Bolton et al., 2009). However, ﬁner sediments are observed on the bed in a backwater region just above the conﬂuence with the Strickland River. Sediment is carried to the delta mainly (B90%) in suspension (ﬁner than 100 mm) with the remaining 10% carried as bed load (Ok Tedi Mining Limited, 1988). The natural annual sediment load of the Fly River is thought to be 85 million t y�1, and sediment discharge is estimated to have reached 125 million t y�1 in the early 1990s after the height of mining (Ok Tedi Mining Limited, 1988). Modern sediment discharge lies between these values.

4.3. Delta Geomorphology The term ‘‘delta’’ includes the subaerial and subaqueous river-sediment accumulations near a source stream, including all that is reworked by oceanographic processes (Wright, 1985). The Fly River delta can be subdivided into four regions: (1) tidal-ﬂuvial channel, (2) distributary channels and islands, (3) distributary mouth bars, and (4) delta front and prodelta (Dalrymple et al., 2003). Of course, the four regions are inherently connected by processes and lateral stratigraphy. This paper focuses primarily on the second deltaic region (i.e., the portion with distributary channels and islands). For the remainder of this paper, the term ‘‘delta’’ will refer to this region unless speciﬁed otherwise. This region also behaves as an estuary (following Pritchard, 1967), and may be (and has been) referred to as such. The Fly River delta has the classic morphology of a tide-dominated delta with seaward-radiating, straight, and funnel-shaped distributary channels creating a triangular shape (Coleman and Wright, 1975; Galloway, 1975; Wright, 1985). It is worth noting that because of its sheer size and dynamic nature, a comprehensive, detailed bathymetric dataset is lacking for the delta. More recently, as part of the National Science Foundation, MARGINS Source-to-Sink initiative, portions of the delta front and prodelta have been mapped at high resolution with multibeam (Crockett et al., 2005). But in the delta, only single-beam methods have been used over discrete areas at different times, and seabed dynamics are poorly known. The apex of the delta lies near Sumogi Island (Fig. 4.2B), and below this point three major distributaries exist: the Far Northern Entrance, the Northern Entrance, and the Southern Entrance. The tidal-ﬂuvial channel region above the delta apex has one primary channel and a few islands, with morphological characteristics of a meandering stream, e.g., point bars, cut banks, and a thalweg on the outside of meander bends. Channel width and

The Fly River Delta

157

Figure 4.2: Maps of Fly River delta. Unprocessed SRTM elevation data obtained from the Jet Propulsion Lab are shown in (A). Although meters of error are characteristic of these data, the data nicely illustrate that the head of the Fly River delta is regionally incised. The distribution of mangroves in the Fly River delta as determined by Robertson et al. (1991) are presented in (B). The absence of mangroves from Kiwai Island and the heads of several of the other islands, especially in the southern part of the delta, might indicate that these islands are covered by pre-Holocene sediments.

158

J. P. Walsh and P. V. Ridd

meander wavelength increase seaward towards the delta apex (Dalrymple et al., 2003). Here the channel is most dynamic, experiencing regular morphological change (Hughes and Baker, 1996). The major distributaries between the apex and the mouth are generally broad, shallow, and relatively straight channels, usually o10-m and rarely W12-m deep, and range from a few kilometers wide at the upstream end to tens of kilometers wide near the mouth (Spencer, 1974). Long subtidal bars with relief of 5–8 m, widths of 0.5–2 km, and up to several kilometers long subdivide the main channels (Dalrymple et al., 2003). These are maintained by energetic ﬂood- and ebb-tidal ﬂows. Many ﬂood and ebb channels pinch out in the landward or seaward direction, and these are referred to as ﬂood and ebb barbs, respectively. A signiﬁcant portion of the shorelines are steep, eroding banks, but elsewhere prograding shorelines are evident, at least temporarily (Fig. 4.3). Colonization of the long subtidal bars and prograding shorelines by vegetation is a means of forming elongate islands and

Figure 4.3: Photographs of sedimentary environments in the Fly River delta. The following are shown: (A) an eroding bank, (B) the accretionary bank of an emergent island, (C) a prograding bank, and (D) an island interior. Note, data in Fig. 4.4 are from cores collected in an island interior (D).

The Fly River Delta

159

increasing land area. In some cases, it appears that islands are rooted to preHolocene strata (Spencer, 1978). Seaward of the delta islands on the delta front is a region characterized by linear bars (referred to as ‘‘distributary mouth bars’’) which in some cases look to be a continuation of distributary islands (Dalrymple et al., 2003). Bars are 1–6 km wide and are evident up to 15–20 km offshore. Dalrymple et al. (2003) explain that the interﬁngering of ﬂood and ebb channels in this region reﬂects a zone of bed load convergence. Bathymetric contours also reveal that this zone is dissected by at least one submarine channel that extends into the Far Northern Channel (Harris et al., 1993). Interestingly, this channel, referred to as Umuda Channel, is aligned with an underlying paleochannel, suggesting the geologic framework may exert some control (Walsh et al., 2003). The delta front is a gently sloping, sandy region on the inner shelf of the Gulf of Papua (Harris et al., 1993). This region is part of a large depositional feature which has been constructed on the continental shelf, known as a subaqueous delta clinoform (Walsh et al., 2004). The delta front is also the location for the topset beds of the subaqueous delta clinoform. Seaward of the delta front is the prodelta and distal delta which includes the more steeply dipping foreset beds and the nearly ﬂat-lying bottomset beds of the subaqueous delta clinoform, respectively. Details and dynamics of the delta front, prodelta, and distal delta portions of the Fly River delta system are discussed in Harris et al. (1993, 2004), Walsh et al. (2004), Palinkas et al. (2006), and Crockett et al. (2009).

4.4. Hydrodynamics 4.4.1. Tides The tides in the delta are semidiurnal with a large diurnal inequality. Spring tide variations up to 4 m are reduced to around 1 m during the neap tides. The effect of bottom friction in generating higher order tidal harmonics (Aubrey and Friedrichs, 1988) is evident with the M4, S4, and M8 constituents accounting for up to 0.65 m of the tidal range at Lewada (Fig. 4.2). During spring tides, the tidal variation deviates grossly from a sinusoidal pattern. Aubrey and Friedrichs (1988) have indicated that a tidal bore is generated upstream of Lewada (near the delta apex), but personal experiences of the lead author indicate a bore can form closer to the river mouth (i.e., in the central portion of the delta).

160

J. P. Walsh and P. V. Ridd

4.4.2. Low-Frequency Water-Level Fluctuations Water-level oscillations up to 0.1 m occur at the mouth of the delta with periods from days to weeks, and these are forced by sea-level oscillations in the Coral Sea and the Gulf of Papua. In addition to this external forcing, considerable variations in the longitudinal water-level slope occur within the delta. The water-level difference between the mouth and apex of the delta varies by as much as 0.2 m in conditions when river discharge remains constant. This effect has been attributed by Wolanski et al. (1997) to two causes: (1) an increase in the bottom friction coefﬁcient with spring tides (Provis and Lennon, 1983; Wolanski, 1994), requiring a larger water slope to maintain a constant freshwater discharge, and (2) trade winds from the southeast, which ﬂuctuate at periods of 1–2 weeks and have speeds averaging around 12 m s�1, causing a set up at the coast and an increase in water level within the delta.

4.4.3. Currents Currents within the delta are dominated by the tides (Wolanski et al., 1995). The net ﬂow velocity due to the freshwater discharge is relatively small, i.e., on the order of 1–2 cm s�1 at the mouth and 10 cm s�1 at Sumogi Island. This compares with tidal currents commonly reaching 1 m s�1 and up to 2 m s�1 in the apex of the delta and 1.5 m s�1 at the mouth. The net discharge of freshwater is highly asymmetrical with up to 80% of the discharge being diverted to the south of Sumogi Island. In addition, high-salinity seawater intrudes further upstream in the Northern Entrance than the other two channels. It is estimated that the Southern Entrance channel carries between 55% and 80% of the freshwater and river-supplied sediment ﬂux. The tidal currents show little vertical shear except near the bed due to bottom friction (Wolanski et al., 1995; Harris et al., 2004). Vertical current shear due to buoyancy is small and difﬁcult to measure, and the water column is often well mixed by the strong tidal currents. It should be noted that while buoyancy currents are small compared to the tidal currents, they are doubtless present and are responsible in part for the formation of the turbidity maximum zone in the central delta. Hydrodynamic modeling of the delta by Wolanski et al. (1995) duplicates well the observed currents, but requires a smaller than expected Manning coefﬁcient (0.15). It is postulated that this is due to the presence of a nearbottom ﬂuid–mud layer which decreases bottom frictional stresses (Wolanski et al., 1995).

The Fly River Delta

161

4.4.4. Waves Because the delta faces the east–southeast it is directly exposed to waves generated by the southeast trade winds between April and November. The fetch for these waves is on the order of a thousand kilometers, and thus the genera tion of long period waves that impinge on the sea bed is expected. The gently shoaling continental shelf offshore from the delta is about 150-km wide and is expected to modify the wave spectrum as waves propagate towards the shore. Unlike other oceanographic parameters such as current and tides, no published data examining the systematics of waves in the delta are known to the authors, though Harris et al. (2004) has modeled the waves at the mouth of the estuary. Wave-rider buoy measurements at Kerema (B200 km to the northeast of the Fly River delta in the Gulf of Papua) indicate signiﬁcant wave heights exceed 0.25 m for 90%, 0.7 m for 50%, 1.5 m for 10%, and 2.0 m for 1% on an annual basis (Thom and Wright, 1983). The wave climate offshore of the delta is likely to be more energetic than that at Kerema due to the aspect of the coastline with respect to the wave direction. The affect of waves generated in the Coral Sea on sediment transport at the mouth of the delta and across the delta front is thought to be signiﬁcant (Harris et al., 2004; Walsh et al., 2004). Furthermore, there is considerable seasonality in the winds and waves; Thom and Wright (1983) determine the median signiﬁcant wave height to be 1.3 m during the SE trade-wind period (December–March) and 0.3 m during the NW monsoon season (March–November). This seasonal cycle in waves has been hypothesized to be important to across-shelf transport of ﬂuvial sediments (Harris et al., 1993, 2004; Walsh et al., 2004). In conditions of strong winds that generally blow along the axis of the three main distributary channels, there is a considerable generation of short-period waves within the delta. Although the magnitude of these waves has not been directly studied, they likely inﬂuence suspended-sediment concentration (SSC)s. This is hypothesized as these waves, though short in wavelength and typically of less than 0.5 m in height, probably can resuspend bottom sediment on the shores of the numerous unconsolidated mudbanks and islands within the delta. Observations of Wolanski et al. (1995) support this notion.

4.5. Sediment Dynamics 4.5.1. Suspended-Sediment Concentrations Tidal currents are the primary control on SSC within the delta (Wolanski et al., 1995). During spring tides, near bed maximum concentrations may

162

J. P. Walsh and P. V. Ridd

brieﬂy exceed 30 g l�1 and are persistently above 2 g l�1. In contrast, during neap tides, the SSC is generally less than 0.5 g l�1 (Wolanski et al., 1995). There are considerable variations of SSC within a tidal cycle with resuspension occurring during accelerating strong currents and deposition occurring when currents are decelerating, particularly at slack tides. Some SSC data was collected by Wolanski et al. (1995) during a rough weather event (winds exceeding 30 knots from the southeast). These data from the mouth of the Southern Entrance during spring tides show a short-lived (a few hours) factor-of-two increase in SSC. A very pronounced turbidity maximum is present in the delta and is located approximately 20 km from the mouth of the delta in the southern channel at salinities of 5–10 (Wolanski et al., 1995). Water entering the head of the delta typically has an SSC of ca. 0.5–1.0 g l�1 (Salomons and Eagle, 1990) but SSCs at the turbidity maximum may reach an average value of 3 g l�1 during spring tides. It is interesting to note that during the neap tides, the SSCs in the delta are apparently comparable to, or lower than the average SSCs in the river above the delta. The physics behind turbidity maximums is relatively well described as they are a common feature of many estuaries. In the case of the Fly River delta, Wolanski et al. (1995) suggest that tidal pumping and baroclinic circulation are both responsible for the turbidity maximum. The baroclinic circulation is particularly effective at transporting sediment upstream during spring tides due to the very high near bed SSC and strong vertical SSC gradients. 4.5.2. Particle Size and Flocculation The particle-size distribution of the suspended sediment changes markedly from the head of the delta to the ocean. The freshwater entering the delta has a typical d50 of 5 mm, i.e., ﬁne silt, with corresponding d10 and d90 of 1 and 18 mm respectively (Wolanski and Gibbs, 1995). Clay-sized particles accounted for only 20% of the suspended sediment. Within the delta, the particle size is smaller, for example in the Southern Entrance, approximately 20 km from the ocean where the salinity averages about 5, the particle size d50 is about 2 mm at the surface (i.e. clay-size) and 4 mm at the bottom (Wolanski and Gibbs, 1995). The process of ﬂocculation is a fundamental process affecting ﬁne-grained particles and sedimentation in most estuaries, and the Fly River delta is no exception. Sediment enters the delta in an unﬂocculated form but ﬂocs between 30 and 60 mm are commonly found in the delta (Wolanski and Gibbs, 1995). Because of the signiﬁcant fraction of ﬁne silt (as opposed to clay), many of the ﬂocs are relatively weak and easily break-up with strong tidal

The Fly River Delta

163

currents. Typically ﬂocs at slack tide during the neap cycles may reach 60 mm but are smaller, only 30 mm in size, during spring tides. Floc size also varies signiﬁcantly within a tidal cycle; they are partially destroyed during peak currents, and reform during slack water periods. Flocculation is very important in the formation of the turbidity maximum zone as large ﬂocs have larger fall velocities, allowing an appreciable vertical SSC gradient to form. Flocculation settling rates in saline waters are non-linear (Wolanski et al., 1995). This relationship affects the water-column suspended-sediment proﬁle, and thus is important to modeling sediment ﬂuxes and the turbidity maximum in the delta (Wolanski et al., 1995). Wolanski and Gibbs (1995) suggested that the reduction in mean particle size from ﬁne silt in the freshwater zone to clay in the delta is at least partially caused by the weakness of the ﬂocs composed of ﬁne silt compared with ﬂocs composed of clay. Flocs composed of clays are more likely to hold together in periods of high ﬂow and thus are more likely to settle to the bed where they may be advected upstream by the density-driven current. Silt particles may thus preferentially bypass the delta. 4.5.3. Biological Processes Affecting Sediment Dynamics Biological factors also have a signiﬁcant inﬂuence on the dispersal and ultimate fate of sediments. Once over the continental shelf, the Fly River surface plume loses sediment by a process which is heavily inﬂuenced by biological process. Ayukai and Wolanski (1997) studied the sediments, nutrients, and organic material in the plume on an occasion when the plume was partially impinging upon the coral reefs to the south west of the river mouth. It was found that high concentrations remained in the plume until the salinity was about 23 after which there was an abrupt loss of sediment from the plume. It was found that in areas experiencing the sudden loss of sediment, the settling particles were dominated by colonies of phytoplank ton, fecal pellets, and macroscopic aggregates. Blooms of phytoplankton and other biological components are determined by the availability of light and nutrients. In the delta and immediately seaward of the delta mouth, although nutrient levels are high, biological production is limited by the availability of light caused by very high turbidity. Indeed, in the delta, the photic zone is limited to only the top few centimeters of the water column much of the time. As a result of low food availability and dynamic seabed conditions, infaunal abundances within the delta are low (Alongi, 1991; Alongi et al., 1992), and for these reasons, bioturbation is limited in most areas of the delta (Alongi, 1991; Alongi et al., 1992; Dalrymple et al., 2003; Walsh and Nittrouer, 2004).

164

J. P. Walsh and P. V. Ridd

The prograding character of numerous coastlines along the delta gives the impression that mangroves play a key role in trapping a considerable amount of muddy Fly River sediments. However, remote-sensing work suggests most islands are relatively persistent features, and there is little net accretion in the delta (Harris et al., 1993). This topic is discussed further below.

4.6. Sediments and Stratigraphy 4.6.1. Sediment Lithology and Facies Most of the sediment (90%) reaching the Fly River delta is ﬁne-grained (i.e., o63 mm), so not surprisingly, a considerable amount of mud comprises the sediments and strata of the delta. Compared to others, the Fly River delta can be viewed as a muddy system (Dalrymple et al., 2003). However, the large tidal range and strong ﬂuvial discharges produce rapid ﬂows in the distri butary channels, particularly during spring ﬂood and ebb tides. Consequently, surﬁcial sediments of distributary channels (at least the major ones) are 80–90% sand; locally, mud percentages are W80% (Harris et al., 1993). Close examination of delta cores reveals considerable variation in lithological characteristics which are related to controlling processes. The modern facies indicate that heterolithic (i.e., composed of alternating sand and mud layers) sediments are widely present, but with varying sand/mud content and structural conﬁguration. The sands are black in color because of their source and immature nature; the most common petrology of sand grains in decreasing order are lithic fragments, quartz, plagioclase feldspar, pyroxene, and hornblende (Spencer, 1974; Baker, 1998). Earlier clay mineralogical work indicated that the smectite-rich Fly River sediment load was overwhelmed by illite-rich sediments of the Strickland River in the delta (Salomons and Eagle, 1990), but more recent research reveals chlorite-group minerals now appear to dominate (Bolton et al., 2009). Using core lithology, geophysical, and sediment transport data from four discrete areas in the delta, in combination with insights from previous work, Dalrymple et al. (2003) describe 10 facies that characterize the entire deltaic system (including the delta front and prodelta; Table 4.1). It should be noted that the facies of Dalrymple et al. (2003) are not numbered in the order they would be anticipated to occur stratigraphically, and considerable variation in the vertical succession and thickness of the facies (i.e., in a deep core) is expected. The spatial arrangement of facies in the modern system is summarized below; these insights can be used to

The Fly River Delta

165

Table 4.1: Discrete facies that characterize the Fly River deltaic system (after Dalrymple et al., 2003). Facies 1: Channel-ﬂoor sands

Facies 2: Mud–pebble conglomerates

Facies 3: Channel-ﬂoor muds Facies 4: Thin-bedded heterolithic stratiﬁcation deposited on tidal bars and point bars Facies 5: Thinly laminated heterolithic tidal-ﬂat deposits Facies 6: Rooted tidal-ﬂat deposits

Facies 7: Distributary–Mouth–Bar Sands Facies 8: Bioturbated, heterolithic stratiﬁcation Facies 9: Interbedded heterolithic stratiﬁcation Facies 10: Prodelta sandy muds

Massive (appearing) sand beds with occasional cross stratiﬁcation and thin mud layers Mud pebbles (generally 0.5–1 cm in diameter), shell fragments, and wood mixed with sand Composed of thick, homogenous mud layers commonly W0.5 cm thick Mud layers typically thinner than in Facies 3, interbedded with sand layers Laminae generally 1–2 mm where pinstripe lamination and lenticular bedding can be observed Usually muddy, mottled sediments containing mangrove roots, oxidized zones, and dessication cracks Clean, well-sorted sand, averaging 5% clay, and containing cross-bedding and laminations Similar in appearance to Facies 4 and 5 but with increased evidence of bioturbation Thick beds (5–15 cm) of sand and mud generally form the stratiﬁcation Bioturbated mud-rich sediments, carbonate content increases seaward

Note: These facies are not numbered in the order they anticipated stratigraphically. See text for additional information.

construct an idealized view of stratigraphic development by the application of Walther’s law (see Dalrymple et al., 2003). A facies model of coastal strata in the Fly, Kikori, and Purari area was created by Walsh and Nittrouer (2004) from vibracores, and it agrees well with the upper facies (facies 4–6) of Dalrymple et al. (2003; Table 4.1). Work across the region suggests a consistent stratigraphy is preserved in the intertidal and shallow subtidal zone throughout the western Gulf of Papua, and the characteristics of the deposits are analogous to those of other intertidal areas (with and without mangroves) experiencing a large tidal range (W2 m) and sediment supply (e.g., Ganges–Brahmaputra and Yangtze

166

J. P. Walsh and P. V. Ridd

river deltas; Walsh and Nittrouer, 2004). Missing from both of these facies models is representation of pre-Holocene strata. Spencer (1974) identiﬁed pre-Holocene strata on the northern end of Kiwai Island and near the southern mainland shoreline of the delta (Fig. 4.2). Based upon these observations, it was hypothesized that some distributary islands may be nucleated on pre-Holocene topographic highs. Walsh and Nittrouer (2004) collected (after much effort) a 65-cm-long vibracore (core ‘‘D’’) on a mainland bank near the apex of the delta. It was composed of indurated, oxidized sediments, which are likely pre-Holocene in age. 4.6.2. Distribution of Pre-Holocene and Modern Facies It is difﬁcult to be certain of the depth to pre-Holocene strata in the delta. Pre-Holocene exposures include the northern end of Kiwai Island (Spencer, 1974) and the southern bank near the delta apex (Walsh and Nittrouer, 2004), and the sequence boundary (i.e., the unconformity between the preHolocene and Holocene sediments) has been identiﬁed in seismics in several locations (Dalrymple et al., 2003). Mapping efforts by Lo¨fﬂer (1977) and Robertson et al. (1991) indicate that a number of the islands in the central Fly River delta (including Kiwai) are vegetated by freshwater plants (non-mangrove species), suggesting these areas may be composed of preHolocene strata (note the distribution of mangroves in Fig. 4.2 following Robertson et al., 1991). The relative sea-level curve for northern Australia (and presumably southern PNG) indicates a sea-level peak (W1 m above modern sea level) in the mid-Holocene (Larcombe et al., 1995), providing another possible explanation for their non-marine nature. Nevertheless, welldeﬁned channels are evident in seismic proﬁles farther offshore (Harris et al., 1993, 1996; Harris, 1994; Crockett et al., 2005), and it is reasonable to expect that pre-Holocene ﬂuvially derived relief (i.e., a paleovalley) extends throughout the delta. Although an isopach map cannot be created with the existing data, it is noted that the Holocene sediment thickness likely ranges from zero to tens of meters where pre-Holocene lows existed. 4.6.3. Modern Facies Distributions The deepest facies (i.e., Facies 7–10; Table 4.1) are accumulating in reverse order on the prodelta, delta front, and distributary mouth bar (Dalrymple et al., 2003). Detail on the nature of the strata accumulating offshore is described in Harris et al. (1993, 1996), Harris (1994), Walsh et al.

The Fly River Delta

167

(2004), and Palinkas et al. (2006). Within the delta, distributary channels typically contain Facies 1–4. The prevalence of mud beds in Facies 3 and 4, mud layers in Facies 1, and mud pebbles in Facies 2 can be explained by the accumulation and reworking of ﬂuid muds in the delta (Dalrymple et al., 2003). During periods of energetic ﬂows (i.e., spring tides) considerable resuspension occurs, resulting in the development of very high SSCs and near bed ﬂuid muds (Wolanski et al., 1995). During periods of slack water, mud layers are deposited in the distributary channels, particularly in deeper areas where the highest sediment concentrations are found (Dalrymple et al., 2003). As a result, mud beds (Facies 3) are expected to deposit at deeper sites and mud pebbles (Facies 2) in erosional areas. The sandy beds of Facies 1 and the thinly bedded mud and sand of Facies 4 are anticipated in shallower subtidal depths and in low- to mid-tidal areas of banks where energetic currents are found. But, in fact, Facies 1 is common in all channel areas, including the deepest portions (Dalrymple et al., 2003). The facies succession moving vertically from an erosional surface in a distributary channel is likely to proceed in the following manner: Facies 2 may mark the erosional boundary. Facies 1 strata may overlie or underlie these sediments for 0.5–1 m, and these will transition gradually into Facies 3 (for 1–3 m) and 4 (Dalrymple et al., 2003). Note, Facies 4 contains beds of mud and sand, generating a coarsening-upward sequence from Facies 3–4. Finally, moving from the upper subtidal to supratidal strata, a ﬁning-upward trend is produced (including a decreasing mud-layer thickness; Facies 4–6). This transition (common to many intertidal areas) is likely a result of decreasing inundation time and SSCs. Facies 6, composed of muddy, mottled mangrove sediments, marks the limit of regular inundation by tidal and ﬂuvial processes. However, at the most seaward islands of the delta, the ﬁning-upward trend into mangrove strata may be less obvious or absent because of winnowing by wave processes (Walsh and Nittrouer, 2004). 4.6.4. Sedimentation Rates A few studies have attempted to quantify sedimentation rates in the Fly River delta. Unfortunately, 210Pb methods do not work well within the delta. 210 Pb activities are low and often variable with depth (i.e., non-steady state), likely because of water, sediment, and scavenging dynamics (Baker et al., 1995; Walsh and Nittrouer, 2004). Also, 137Cs activities are low and variable in the Indo-Paciﬁc region (Kineke and Sternberg, 2000); this coupled with the concern of mobility of 137Cs precludes its use as a radiochemical marker and reliable measure of sedimentation in the delta.

168

J. P. Walsh and P. V. Ridd

In a few cases, 210Pb proﬁles from cores have produced trends which may be interpreted as steady state and from which sediment accumulation rates may be estimated (Fig. 4.4), but as with any proﬁles without independent veriﬁcation (e.g., from 137Cs), they must be viewed with caution. A rate of 8–10 cm y�1 was apparent from one core collected from a subtidal area, near an island in the Southern Entrance (Baker et al., 1995). Although very rapid, such rates are conceivable in this deltaic setting and are supported by lithologic evidence. Cycles (10–30 cm thick) of sand–mud lamination couplets are preserved in many delta cores, and are likely deposited by

Figure 4.4: Two 210Pb proﬁles from cores collected among mangroves in the interior of islands. The supported activity (226Ra) was estimated to be 1 dpm g�1 for both cores. Data are normalized by clay content. Methods are described in Walsh and Nittrouer (2004). Apparent accumulation rates are indicated, but these rates must be viewed with caution, considering the low activities and non-steady-state nature of the sediment supply. See Fig. 4.3d for a photograph near VCR; VCN was collected in a similar mangrove environment.

The Fly River Delta

169

complex interaction of ﬂuvial and tidal processes, with neap–spring changes in current velocities playing an important role, such as in the Amazon system (Baker et al., 1995; Jaeger and Nittrouer, 1995; Dalrymple et al., 2003; Walsh and Nittrouer, 2004). In most cases, the number of couplets present in these cycles is less than an ideal semidiurnal tidal sequence (28 couplets over a neap–spring cycle). Nevertheless, the strata provide strong evidence for the impact of tides, and furthermore, suggest that at least portions of the sedimentary strata accumulate at rates on the order of 1–W4 mm d�1, lending credence to high measured accumulation rates (Chappell, 1992; Baker et al., 1995; Dalrymple et al., 2003). Operation of Ok Tedi Mine in the head waters of the Fly River began in 1984. More information about the nature, history, and impacts of the Ok Tedi Mine can be found in other chapters of this book. Copper concentrations above pre-mine levels in the Fly River system provide a useful new tool for quantifying sedimentation, yet limited research has used this approach in the delta. Chappell (1992) and Baker (1998) report that a pronounced post-mine signal is preserved in discrete areas near the delta apex; a peak concentration (144 mg g�1) was found at 80-cm depth in a core collected on a mudbank west of Kiwai Island (Chappell, 1992; Baker, 1998). Farther seaward, the copper signal was less prominent in 1990, presumably because of dilution. It is hypothesized that given more time a post-mine copper signal will be prevalent in sediments over a greater extent of the delta (Chappell, 1992). Accumulation rates can be calculated using the proﬁles of copper, assuming an approximate time for the local copper concentration increase. Also, sedimentation rates can be compared with lamination cycles, remotely sensed data, and mangrove forest age estimates. Rates from vibracores in shallow, subtidal areas on mud banks in the apex region are estimated to be 1.75–3.5 m y�1 from lamination thickness, and copper proﬁles reveal sedimentation at these sites likely was deposited within 3 years (prior to sampling). Due to the shallow nature of the sample sites, these rates cannot be sustained for long. Ephemeral deposition over multiannual and shorter timescales is undoubtedly important in the delta, and consequently, storage over this large system is difﬁcult to estimate. Two cores collected in mangrove areas in the interior of inlands have accumulation rates of B1 cm y�1, estimated from 210Pb (Fig. 4.4). If these rates are valid and are assumed to represent sedimentation across the mangrove portions of the islands, the delta may be storing more sediment than previously estimated based on shoreline progradation rates alone (480,000 m3; Harris et al., 1993). For example, if sediment accumulation rates in mangrove areas (covering 874 km2; Robertson et al., 1991) occurs sympathetically with the average the rate of eustatic sea level rise

170

J. P. Walsh and P. V. Ridd

(B0.2 cm y�1), commensurate with creation of new accommodation space, storage of an additional 1.8 � 106 t y�1 may be possible. In fact, this estimate can be increased if a similar assumption is made for channel areas. Nevertheless, when compared to the large load of the Fly River (B1.0 � 108 t y�1), modern storage is likely o10% of the total sediment discharge. Although numerous back-of-the-envelope calculations like this can be made, a true budget of sediment sequestered in the delta is elusive (and perhaps is impossible to determine) due to the sheer size of the system and its dynamic nature.

4.7. Shoreline Change Rates Aerial photograph and remotely sensed data have been used to examine shoreline changes in the delta over periods of decades (Chappell, 1992; Hughes and Baker, 1996; Baker, 1998; Walsh and Nittrouer, 2004). Conclusions from this research include: (1) little net change in island area (Baker, 1998); (2) a complex pattern of minor erosion and accretion is evident along many discrete shoreline segments throughout the delta; and (3) a number of small islands have appeared and disappeared over time, particularly near the delta apex. Rates of progradation and erosion in isolated locations are reported up to hundreds of meters per year. More work is needed to quantify this change and understand controlling factors. It is likely that channel migration has an important impact on distributary channel switching and thus on the modern dispersal of sediment seaward. The reader is directed to Dalrymple et al. (2003) for more insights on this topic.

4.8. Topics for Further Study 4.8.1. Fluid–Mud Dynamics Fluid muds have been observed within the Fly River delta (Wolanski and Eagle, 1991; Wolanski et al., 1995) and are hypothesized to be important in moving sediment across the prodelta (topset and foreset beds of the subaqueous shelf clinoform) (Harris et al., 1993, 2004; Dalrymple et al., 2003; Walsh et al., 2004). The low 210Pb and high clay content of deposits on the foreset region of the clinoform imply a ﬂuid–mud transport mechanism on the shelf (Walsh et al., 2004), which may be an important process enhancing terrestrial carbon storage. Formation of ﬂuid muds in the delta by tidal

The Fly River Delta

171

processes is reasonably well documented (Wolanski and Eagle, 1991; Wolanski et al., 1995), but it is not clear if and how these ﬂuid muds traverse the shelf (Harris et al., 2004; Walsh et al., 2004). Tidal currents alone do not appear to be sufﬁcient to transport ﬂuid muds from the delta to the foreset region, suggesting seasonal wave dynamics also must be important (Harris et al., 1993, 2004; Walsh et al., 2004). More research is needed to examine ﬂuid-mud formation, transport, and accumulation throughout the Fly River dispersal system. Sediment and carbon dynamics on the shelf actively are being examined by Chuck Nittrouer, Andrea Ogston, Miguel Goni, John Crockett, and others as part of the Source to Sink (S2S) Initiative of the US National Science Foundation Margins Program (Crockett et al., 2009). 4.8.2. A Field, Remote Sensing and Modeling Approach to Understand Delta Sedimentation The elevated copper signal produced by mining upstream of the delta provides a unique tool to quantify sedimentary processes in the delta. A core collection effort guided by a detailed remote-sensing analysis and geophysical survey work (chirp and multibeam) could provide an improved understanding of modern sediment dynamics in the delta and insight on how the geologic history (e.g., delta switching or allocyclic factors like sediment supply) may have inﬂuenced it. However, the sheer size and dynamic nature of this system precludes a ﬁeld study from thoroughly characterizing the system. For this reason, any ﬁeld-based study should run concurrently with (or at least be guided by) a modeling effort. Sergio Fargherazzi (Florida State University), Irina Overeem, and Scott Peckham (University of Colorado at Boulder) are currently using a modeling approach to evaluate the morphological and stratigraphic evolution of this delta system, also as part S2S. 4.8.3. Tectonic Control on Delta Evolution The Fly River ﬂows across a tectonically active foreland basin, experiencing spatially varying rates of subsidence (Veevers, 1984; Pigram et al., 1989). The ‘‘Oriomo Uplift’’ 36,000 y.b.p. is hypothesized to have shifted the Fly River from ﬂowing southwestward to its present course (Blake and Ollier, 1971; Torgersen et al., 1985). It is unknown how this tectonic history has impacted the evolution and stratigraphy of the delta. A comprehensive seismic, coring, and drilling expedition is needed to determine the evolution of the delta and evaluate the degree to which the past is the key to the present.

172

J. P. Walsh and P. V. Ridd

4.9. Conclusions Based on insights from studies cited above, we understand the modern Fly River delta is a region of great and complex sediment transport and ephemeral deposition, yet it experiences relatively little modern sediment accumulation and storage. Through lithologic and transport data, it appears the delta behaves primarily as a discharge signal ﬁlter for sediments accumu lating in the delta seabed and those on route to the Gulf of Papua shelf. This concept is captured in a cartoon which illustrates the major forcing factors for sedimentation in the Fly River dispersal system (Fig. 4.5; Walsh et al., 2004). The large, interannually varying discharge of water and sediment to the Fly River delta is met by strong, semidiurnal tides propagating landward from the ocean. Here, energetic tidal currents produce the morphological characteristics of a tide-dominated delta and a muddy stratigraphy dominated by associated signatures (e.g., cyclic, interlaminated mud and sand deposits). Ultimately, these tides with help from river discharge and seasonal wave processes are thought to control the transfer of sediment farther along the dispersal, across the continental shelf. More work is needed to better understand sediment dispersal processes today and those recorded in delta strata.

Figure 4.5: Forcing functions driving sedimentation processes in the Fly River dispersal system (edited from Walsh et al., 2004, with permission from Elsevier). Note, the river-discharge, tidal and wave signals of sediment transport converge and are convolved at the Fly River delta.

The Fly River Delta

173

ACKNOWLEDGMENTS

This paper is dedicated to Bruce Parker who passed away from cancer in 2003. Bruce, an employee of the Australian Institution of Marine Science assisted J.P. Walsh and Pete Swarzenski (USGS) with the collection of vibracores in the Fly River delta during September 1997. Bruce, who was head of the coral drilling program at the Australian Institute of Marine Science, was bafﬂed by our desire to work in the delta, an extremely muddy habitat for crocodiles. But, he provided invaluable knowledge, unwavering support, and great companionship to help make the research successful. Many others are acknowledged for enabling our research in this area, especially, Gregg Brunskill, Chuck Nittrouer, Dick Sternberg, and Irena Zagorskis. Funding from a MARGINS S2S National Science Foundation grant (OCE-0452166) helped support the writing of this paper.

REFERENCES Alongi, D. M. (1991). The role of intertidal mudbanks in the diagenesis and export of dissolved particulate materials from the Fly Delta, Papua New Guinea. Journal of Experimental Marine Biology and Ecology, 149, 81–107. Alongi, D. M., Christofferson, P., Tirendi, F., & Robertson, A. I. (1992). The inﬂuence of freshwater and material export on sedimentary facies and benthic processes within the Fly Delta and adjacent Gulf of Papua (PNG). Continental Shelf Research, 12, 287–326. Aubrey, D. G., & Friedrichs, C. T. (1988). Seasonal climatology of tidal non linearities in a shallow estuary. In: D. G. Aubrey, & L. Weishar (Eds). Hydrodynamics and Sediment Dynamics of Tidal Inlets. Springer, Berlin, pp. 103–124. Ayukai, T., & Wolanski, E. (1997). Importance of biologically mediated removal of ﬁne sediments from the Fly River plume, Papua New Guinea. Estuarine, Coastal and Shelf Science, 44, 629–639. Baker, E.K. (1998). Sedimentation in a tidally dominated delta and the impact of upstream mining: Fly River, Papua New Guinea. Unpublished Ph.D. Thesis, University of Sydney. Baker, E. K., Harris, P. T., Short, S. A., & Keene, J. B. (1995). Patterns of sedimentation in the Fly River Delta. In: B. W. Fleming & A. Bartholoma (Eds). Tidal Signatures in Modern and Ancient Sediments: International Association of Sedimentologists Special Publication, Wiley Blackwell Publishing, Oxford, Vol. 24, pp. 122–193. Blake, D. H., & Ollier, C. D. (1971). Alluvial plains of the Fly River delta. Z. Geomorph. Zeitschrift fu¨r Geomorphologie, Suppl., 12, 28–32.

174

J. P. Walsh and P. V. Ridd

Bolton, B. R., Pile, J. P., & Kundapen, H. (2009). Texture, geochemistry, and mineralogy of sediments of the Fly River system. In: B. Bolton (Ed.), The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 51–112. Chappell, J. (1992). Effects of increased sediment ﬂux in the Fly River delta: Sediment core results. Australian National University, School of Biogeography & Geomorphology. Report to Ok Tedi Mining Limited dated 1992. 64 pp. Coleman, J. M., & Wright, L. D. (1975). Modern river deltas: Variability of processes and sand bodies. In: M. L. Broussard (Ed.). Deltas, Models for Exploration. Houston Geological Society, Houston, pp. 99–149. Crockett, J., Nittrouer, C., Ogston, A., Sternberg, R., Driscoll, N., Babcock, J., Milliman, J., Slingerland, R., Naar, D., Donahue, B., Walsh, J., Dietrich, W., Parker, G., Bera, M., Davies, H., Harris, P., Goni, M., Aller, R., & Aller, J. (2005). Where rivers and oceans collide: Geological complexities of this interface revealed by the NSF MARGINS program. EOS, Transactions, 83(6), 1–32. Crockett, J., Nittrouer, C. A., Ogston, A. S., & Goni, M. A. (2009). Variable styles of sediment accumulation impacting strata formation on a clinoform: Gulf of Papua, Papua New Guinea. In: B. Bolton (Ed.), The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 177–204. Dalrymple, R. W., Baker, E. K., Harris, P. T., & Hughes, M. G. (2003). Sedimentology and stratigraphy of a tide-dominated, foreland basin delta (Fly River, Papua New Guinea). In: H. Sidi, D. Nummedal, P. Imbert, H. Darman, & H. W. Posamentier (Eds). Tropical Deltas of Southeast Asia-Sedimentology, Stratigraphy and Petroleum Geology. SEPM Special Publication No. 76, pp. 147–173. D’Addario, G. W., Dow, D. B., & Swoboda, R. (1976). Geology of Papua New Guinea 1:2,500,000. Bureau of Mineral Resources, Canberra, Australia. Dietrich, W. E., Day, G., & Parker, G. (1999). The Fly River, Papua New Guinea: Inferences about river dynamics, ﬂoodplain sedimentation and fate of sediment. In: A. J. Miller, & A. Gupta (Eds). Varieties of Fluvial Form. Wiley, New York, pp. 345–376. Galloway, W. E. (1975). Process framework for describing the morphologic and stratigraphic evolution of deltaic depositional systems. In: M. L. Broussard (Ed.). Deltas, Models for Exploration. Houston Geological Society, Houston, pp. 87–98. Harris, P. T. (1994). Incised valleys and backstepping deltaic deposits in a forelandbasin setting, Torres Strait and Gulf of Papua, Australia. In: R. W. Dalrymple, R. Boyd, & B. A. Zaitlin (Eds). Incised-Valley Systems: Origin and Sedimentary Sequences. SEPM Special Publication No. 51, pp. 98–108. Harris, P. T., Baker, E. K., Cole, A. R., & Short, S. A. (1993). A preliminary study of sedimentation in the tidally dominated Fly River Delta, Gulf of Papua. Continental Shelf Research, 13, 441–472. Harris, P. T., Pattiaratchi, C. B., Keene, J. B., Dalrymple, R. W., Gardner, J. V., Baker, E. K., Cole, A. R., Mitchell, D., Gibbs, P., & Schroeder, W. W. (1996). Late Quaternary deltaic and carbonate sedimentation in the Gulf of Papua

The Fly River Delta

175

foreland basin: Response to sea level change. Journal of Sedimentary Research, 66, 801–819. Harris, P. T., Hughes, M. G., Baker, E. K., Dalrymple, R. W., & Keene, J. B. (2004). Sediment transport in distributory channels and its export to the prodeltaic environment in a tidally dominated delta: Fly River, Papua New Guinea. Continental Shelf Research, 24(10), 2431–2454. Hughes, M. G., & Baker, E. K. (1996). Using satellite remote sensing data to identify coastal changes in the Fly River Delta, PNG – pilot study. Ocean Science Institute Report No. 68. Jaeger, J. M., & Nittrouer, C. A. (1995). Tidal controls on the formation of ﬁne-scale sedimentary strata near the Amazon River mouth. Marine Geology, 125, 259–281. Jukes, J. B. (1847). Narrative of the Surveying Voyage of H.M.S. Fly, Commanded by Captain F.P. Blackwood, R.N., in Torres Strait, New Guinea and Other Islands of the Eastern Archipeligo During the Years 1842–1846, Vols I and II. T&W. Boone, London. Kineke, G. C., & Sternberg, R. W. (2000). Sediment dispersal from the Sepik River, Papua New Guinea, via surface and subsurface plumes. EOS, Transactions, 80, 49, OS218. Larcombe, P., Carter, R. M., Dye, J., Gagan, M. K., & Johnson, D. P. (1995). New evidence for episodic post-glacial sea-level rise, central Great Barrier Reef, Australia. Marine Geology, 127, 1–44. Lo¨fﬂer, E. (1977). Geomorphology of Papua New Guinea. Australian National University Press, Canberra, 258 pp. Milliman, J. D., Farnsworth, K. L., & Albertin, C. S. (1999). Flux and fate of ﬂuvial sediments leaving large islands in the East Indies. Netherlands Journal of Sea Research, 41, 97–107. Ok Tedi Mining Limited (1988). Sixth supplement agreement environmental study 1986–1988. Final Draft Report OTML, PNG. Palinkas, C. M., Nittrouer, C. A., & Walsh, J. P. (2006). Inner-shelf sedimentation in the Gulf of Papua, New Guinea: A mud-rich shallow shelf setting. Journal of Coastal Research, 22, 760–772. Pickup, G. (1984). Geomorphology of tropical rivers: I. Landforms, hydrology and sedimentation in the Fly and lower Purari, Papua New Guinea. In: A. P. Schick (Ed.). Channel Processes – Water, Sediment, Catchment Controls. Braunschweig, Catena Supplement 5, Catena-Verlag, Reiskirchen, Germany, pp. 1–17. Pigram, C. J., Davies, P. J., Feary, D. A., & Symonds, P. A. (1989). Tectonic controls on carbonate platform evolution in southern PNG: Passive margin to foreland basin. Geology, 17, 199–202. Pritchard, D. W. (1967). What is an estuary, physical viewpoint. In: G. H. Lauf (Ed.). Estuaries. American Association for the Advancement of Science, Washington D.C., publ. no. 83. Provis, D. G., & Lennon, G. W. (1983). Eddy viscosity and tidal cycles in a shallow sea. Estuarine and Coastal Shelf Science, 16, 351–361.

176

J. P. Walsh and P. V. Ridd

Robertson, A. I., Daniel, P. A., & Dixon, P. (1991). Mangrove forest structure and productivity in the Fly River estuary, Papua New Guinea. Marine Biology, 111, 147–155. Salomons, W., & Eagle, A. M. (1990). Hydrology, sedimentology and the fate and distribution of copper in mine-derived discharges in the Fly River system, Papua New Guinea. The Science of the Total Environment, 97/98, 315–334. Spencer, L. K. (1974). Sedimentation and geological history of the Fly River region, Papua. M.Sc. Preliminary Report. University of Sydney, 58 pp. Spencer, L. K. (1978). The Fly estuarine delta, Gulf of Papua, Papua New Guinea. Unpublished M.Sc. Thesis, University of Sydney, 278 pp. Thom, B. G., & Wright, L. D. (1983). Geomorphology of the Purari Delta. In: T. Petr (Ed.). The Purari: Tropical Environment of a High Rainfall River Basin. Dr. W. Junk Publishers, The Hague, 624 pp. Torgersen, T., Jones, M. R., Stephes, A. W., Searle, D. E., & Ullman, W. J. (1985). Late Quaternary hydrologic changes in the Gulf of Carpentaria. Nature, 13, 785–787. Veevers, J. J. (1984). Phanerozoic Earth History of Australia. Clarendon Press, Oxford, 418 pp. Walsh, J. P., & Nittrouer, C. A. (2004). Mangrove-bank sedimentation in a mesotidal environment with large sediment supply, Gulf of Papua. Marine Geology, 208, 225–248. Walsh, J. P., Driscoll, N. W., & Nittrouer, C. A. (2003). Controls on subaqueousdelta clinoform development. EOS, Transactions, 84(46), OS12A-0188. Walsh, J. P., Nittrouer, C. A., Palinkas, C., Ogston, A. S., Sternberg, R. W., & Brunskill, G. J. (2004). Clinoform mechanics in the Gulf of Papua, New Guinea. Continental Shelf Research, 24, 2487–2510. Wolanski, E. (1994). Physical Oceanographic Processes of the Great Barrier Reef. CRC Press, Boca Raton, USA, 185 pp. Wolanski, E., & Eagle, M. (1991). Oceanography and sediment transport in the Fly River estuary and Gulf of Papua. In: Proceedings 10th Australasian Conference on Coastal and Ocean Engineering, Auckland, pp. 453–457. Wolanski, E., & Gibbs, R. J. (1995). Flocculation of suspended sediment in the Fly River estuary, Papua New Guinea. Journal of Coastal Research, 11(3), 754–762. Wolanski, E., King, B., & Galloway, D. (1995). Dynamics of the turbidity maximum in the Fly River estuary, Papua New Guinea. Estuarine, Coastal and Shelf Science, 40, 321–337. Wolanski, E., King, B., & Galloway, D. (1997). Salinity intrusion in the Fly River estuary, Papua New Guinea. Journal of Coastal Research, 13, 983–994. Wright, L. D. (1985). River deltas. In: R. A. Davis (Ed.). Coastal Sedimentary Environments. 2nd Edn., Springer-Verlag, New York, pp. 1–75.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00405-9

Chapter 5

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform: Gulf of Papua, Papua New Guinea John S. Crockett1, Charles A. Nittrouer1,, Andrea S. Ogston1 and Miguel A. Goni2 1

School of Oceanography, University of Washington, Seattle, WA 98195, USA College of Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, OR 97331, USA 2

5.1. Introduction Clinoforms are sigmoidal-shaped deposits on the continental shelf, and have been recognized as the basic building blocks of continental margin morphology (Mitchum et al., 1977). Clinoforms are well represented in the ancient geological record (e.g., Asquith, 1970; Leithold, 1993) and are common on modern shelves adjacent to large rivers (e.g., Ganges–Brahmaputra, Kuehl et al., 1989; Huang He, Alexander et al., 1991; Liu et al., 2004; and Amazon, Nittrouer et al., 1986a). While a general understanding of the factors that affect the evolution of clinoform strata exists, we are still improving our ability to quantify processes responsible for the sedimentary deposits that are the building blocks of these stratigraphic environments. The across-shelf anatomy of a clinoform includes a landward topset, where sediment accumulation is slow due to enhanced shear stresses created by waves and currents; a foreset in deeper water, where shear stresses decrease and accumulation is rapid; and a seaward bottomset, where accumulation is Corresponding author. Tel.: 206-543-5099; Fax: 206-543-6073;

E-mail: [email protected] (C.A. Nittrouer).

178

J. S. Crockett et al.

wave- and current-induced shear stress inhibits sediment deposition in shallow water

rollover point

topset

deeper water creates a “stress refuge” and allows accumulation on the foreset

reduced sediment supply causes slow accumulation on the bottomset

foreset

bottomset

Figure 5.1: Schematic illustration of two-dimensional clinoform growth. Signiﬁcant processes for each region are indicated. The distribution of accumulation rates allows the clinoform to prograde into deeper water while maintaining a self-similar shape. slow due to reduced sediment supply (Fig. 5.1). The dimensions, shape, and sedimentary fabric of clinoforms contain information about the depositional conditions. Interpretation of this record in the context of sea-level change, basin subsidence, and sediment supply is a fundamental element in the stratigraphic approach to investigating sedimentary sequences (Vail et al., 1977; Vail, 1987; Posamentier et al., 1988; Van Wagoner et al., 1990). Previous studies quantifying the factors that inﬂuence clinoform shape have focused predominantly on the rate and mechanisms of clinoform progradation into deeper water (e.g., Wright et al., 1988; Kuehl et al., 1989; Alexander et al., 1991; Nittrouer et al., 1996). Clinoforms have been modeled primarily as two-dimensional (2-D) features with variation only in the across-shelf dimension (as a function of distance from shore and water depth). The 2-D approach has been reasonable and has led to important discoveries concerning the processes that govern clinoform growth. However, most clinoforms extend for hundreds of kilometers along shelf, which suggests that different processes likely operate to cause diversity of sedimentation over this distance. It is the aggregate of processes that creates the gross morphology. However, the three-dimensional (3-D) complexity of clinoform development is poorly known. Depending on the characteristics of the river discharge (e.g., amount, grain size, and variability of discharge) and ocean conditions (e.g., waves and currents), sediment can accumulate in either a steady-state fashion, an episodic fashion, or both. On the continental shelf adjacent to the Fly River in Papua New Guinea, clinoform accumulation was examined by earlier

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

179

studies (e.g., Harris et al., 1993; Walsh et al., 2004). However, these studies of sedimentation included limited sediment cores near the mouth of the Fly River. Through the present study, data addressing the along-shelf heterogeneity of clinoform progradation have been obtained and can be used to appraise the 3-D character of clinoform processes. The objectives of this paper are to (1) identify the styles of sediment accumulation and their distribution on the Fly River shelf, (2) relate the styles of accumulation to oceanographic processes, and (3) present a more complete picture of the complex processes that are responsible for creating the clinoform deposit found on the Fly River shelf.

5.2. Background 5.2.1. Physical Setting The Gulf of Papua (GOP) is located off the southeastern coast of New Guinea (Fig. 5.2). The Fly River discharges into the western side of the GOP and supplies 1.15 108 t/y of sediment (Milliman, 1995), which makes it one of the world’s top 20 rivers in terms of sediment supply to the ocean (Milliman and Meade, 1983). New Guinea has a rugged mountain range running down the center of the island, with peaks W4,000 m in elevation that receive W10 m/y of precipitation (Harris et al., 1993). Rainfall decreases to B2 m/y at the coast. The steep slopes and high rainfall produce large denudation rates of 3–4 mm/y (Pickup, 1984). The great relief of the highlands gives way to an extensive alluvial plain across which the Fly River system meanders for the ﬁnal 800 km, with only a 20-m change in elevation. Under typical Paciﬁc Ocean conditions, warm water resides as a pool north of New Guinea, causing high rates of precipitation and creating wet tropical conditions for New Guinea. The extensive ﬂoodplains and relatively constant rainfall combine to produce a hydrograph for the Fly River that typically varies by less than a factor of 2–3 over the seasonal cycle (Nittrouer et al., 1995). However, the GOP can be subject to some large-scale climatic effects. During El Nin˜o conditions, the warm pool is displaced to the eastern Paciﬁc, precipitation is reduced in the highlands of New Guinea, and rivers dramatically decrease their discharge into the GOP. The Fly River mouth has the classic morphology of a tide-dominated delta (Fisher et al., 1969; Galloway, 1975; Wright, 1977, 1985; Miall, 1984). It is generally mesotidal (2–4 m range, reaching a maximum of B5 m at the delta apex during spring tides; Wolanski and Alongi, 1995) and produces tidal

180

J. S. Crockett et al.

0 °N

New Guinea 8°0′S

10°S

5m

15m

8°15′S

T13-20 T13-30 T9-18 T9-30

8°30′S T8-16

T13-48 T13-65 T12-35

T10-17 T10-40 T10-60

T8-18 T8-20 T8-30

8°45′S

150°E

140°E

10m

T8-50 T7-30 T6-30

T8-60 Depth in Meters

T5-40 T4-30 T4-50 T2-17

9°0′S T1-17 T1-23

20 40

T2-30

60

9°15′S

80

0 143°30′E

10

100

20 Kilometers 143°45′E

144°0′E

144°15′E

144°30′E

Figure 5.2: The Fly River margin was divided into 13 shore-perpendicular transects, which were resampled on a seasonal basis. Transect T1 is located at the southwest end of the study area, and transect T13 is in the northeast. Each transect is separated from the next by B10 km. T8 corresponds to Umuda shelf valley, and T4 is Kiwai Valley. Station locations are superimposed on a multibeam bathymetric survey (colored isobaths) undertaken for portions of the Gulf of Papua (see Crockett et al., 2008). currents W100 cm/s in the distributary channels of the delta (Ogston et al., 2008). The climate of the GOP is dominated by seasonal ﬂuctuations between monsoon and trade-wind seasons. During the monsoon season (December–March), winds blow from the northwest and waves are

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

181

fetch-limited, creating wave heights that are typically o0.3 m. Waves generated by the trade winds (May–October) enter the GOP from the southeast with a very long fetch and typically are W1.3 m in height (Thom and Wright, 1983). Large frontal storms rarely travel within 101 of the equator (Pielke, 1990) and have little direct impact on the GOP. 5.2.2. Sedimentary Setting The large sediment supply and a broad (W150 km) continental shelf in the GOP together provide an environment conducive to the creation of a clinoform (Walsh et al., 2004). The modern clinoform of the Fly River is building over the continental shelf and extends to B65-m water depth (Slingerland et al., 2008). The shelf gently slopes from the base of the modern clinoform to the shelf break at B150-m water depth. By using mean wave heights (Thom and Wright, 1983) and tidal currents from one location, Walsh et al. (2004) demonstrated that the depth to the ‘‘rollover point’’ separating topset from foreset (Fig. 5.1; B25–40 m, deepening toward the northeast) is likely inﬂuenced by the bed shear stress resulting from combined waves and currents. Due to seasonal variability, shear stresses are sufﬁcient to transport sediment from the topset to the foreset during the trade-wind season and insufﬁcient to transport sediment during the monsoon season. Several subaqueous valleys incise the Fly River shelf (Crockett et al., 2008). Kiwai Valley is narrow, steep-sided and extends seaward from the Fly River delta (Fig. 5.2). Umuda Valley is a broad valley and extends seaward from the northernmost distributary of the Fly River delta (Fig. 5.2). These shelf valleys presently accumulate a small fraction of the overall sediment supplied from the Fly River (o1%, Crockett et al., 2008) because of their small areal extent. Sediment from the prograding clinoform accumulates in the valleys, and they also serve as conduits for moving sediment across shelf via gravity-driven ﬂows. 5.2.3. 210

210

Pb Geochronology

Pb is a member of the 238U natural radioactive series and is found in seawater as a result of runoff, atmospheric precipitation, and the decay of its 226 Ra precursor in the water column (Bruland, 1974). 210Pb has a half-life of 22.3 y and therefore is appropriate to use in the examination of sediment accumulation over the past century (Nittrouer et al., 1979). In seawater, 210 Pb is permanently adsorbed onto the surfaces of sediment particles and

182

J. S. Crockett et al.

the relatively rapid removal of 210Pb from seawater causes it to occur in sediment at activities above those supported by the decay of 226Ra there. This process creates activities of two types in the seabed: ‘‘excess’’ activity, which indicates the presence of 210Pb scavenged from seawater to the surfaces of particles; and ‘‘supported’’ activity, which is generated by the decay of 226Ra contained within particles. Because of analytical limitations associated with relatively low initial activities supplied to the surface of the seabed in coastal environments, excess activity can be detected only until four to ﬁve half-lives of the radioisotope have elapsed. If excess 210Pb activity is present, the sediment was deposited in the past century. 210 Pb geochronology has been utilized in most aqueous environments. In the marine environment, it has been used effectively to measure steady-state sediment accumulation rates on the continental shelf. When sediment is deposited on the seabed, it is effectively removed from the presence of excess 210Pb in the water column, and the acquired activity decays toward the supported level. At a location where sediment is steadily accumulating, a proﬁle of 210Pb activity shows a logarithmic decrease. The slope of a line ﬁt to the activities can be used to calculate the rate of sediment accumulation at that site. In some cases, physical and biological reworking of the seabed can create a surface mixed layer (SML) of uniform 210Pb activity, below which sediment accumulates in a steady-state fashion. 210 Pb proﬁles do not always follow this simple pattern. Initial 210Pb activity reaching the seabed can be inﬂuenced by particle size (210Pb preferentially adsorbs to clay particles, which have relatively large surface areas). Other factors affecting excess 210Pb activities are the amount of dissolved 210Pb in the water column, suspended-sediment concentration during deposition, and resuspension that exposes particles to dissolved 210Pb. All of these factors either separately or together can conspire to produce activity proﬁles that vary with depth in core and do not decrease logarithmically. This is an indication that sediment accumulation is a non-steady-state process and is impacted by variability in the factors described above. Non-steady-state sedimentation often results from episodic events that have an easily identiﬁed signature. On the Eel River shelf, California, ﬂoodrelated deposition creates beds with low, uniform 210Pb activity because of competitive scavenging by particles during high-concentration deposition events (Sommerﬁeld and Nittrouer, 1999). Similar proﬁles have been observed on the Fly River shelf and termed ‘‘event horizons’’ (Walsh et al., 2004). The adsorption of 210Pb onto sediment particles is limited for high sediment concentrations due to (1) the presence of a lutocline that inhibits mixing of sediment and water in the bottom boundary layer (BBL) (Ross and Mehta, 1989), and (2) competitive scavenging by which many sediment particles share the available dissolved 210Pb. Both of these processes cause low

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

183

activity for individual particles when transport and deposition rates are high (Sommerﬁeld and Nittrouer, 1999).

5.3. Methods To provide comprehensive coverage of the study area, the shelf at the mouth of the Fly River was divided into 13 shore-perpendicular transects. Transect T1 is located at the southwestern boundary of the delta and transect T13 is located beyond the northeastern boundary of the Fly River mouth, in the central GOP (Fig. 5.2). Each transect is separated from the next by B10 km. Stations are denoted by their transect number followed by the nominal water depth (e.g., site T13-20 is on transect T13 in B20 m of water). Two to seven stations were reoccupied seasonally on each transect. At each site occupied, a box core (B50 cm in length) and a kasten core (B300 cm in length) were collected. Box cores were used to capture the sediment–water interface for evaluation of seasonal changes in the seabed and collected at least once each season. Kasten cores were used to evaluate the longer-term sedimentary history of the site. 5.3.1. Sedimentary Analysis From each box and kasten core, 2.5-cm-thick slabs (covering the entire core length) were removed and X-rayed to reveal internal sedimentary structures. A 15-cm PVC tube was inserted into box cores and was subsampled every 1 cm down to 10-cm depth, then every 2 cm to the bottom of the core. To measure grain size, the samples were wet sieved at 63 mm to separate the sand fraction from the mud (i.e., silt and clay). The sand was subsequently dried and weighed. The grain size of the mud fraction was analyzed by a Sedigraph 5100. Clay and silt mass percentages were corrected for the total mass of the sample. Grain-size variations were used to corroborate the changes observed in Xradiographs. Kasten cores were subsampled directly from the core barrels. One sample was collected every 2 cm down to 50-cm depth, then every other 2 cm to the bottom of the core. Grain size was also measured on kasten-core samples. 5.3.2.

210

Pb Analysis

To document decadal-scale sediment accumulation rates, 210Pb was used. 210 Pb activities were determined using a method modiﬁed from Nittrouer et al. (1979). Sediment samples were dried and ground to expose surface areas.

184

J. S. Crockett et al.

Approximately 5 g of sample was spiked with a known amount of 209Po to quantify the efﬁciency of the laboratory procedures. 210Po, the effective daughter of 210Pb, was chemically released from the sediment by leaching in 16N HNO3 and 6N HCl. Both 209Po and 210Po were removed from the leachate onto silver planchets by spontaneous electrodeposition. Alpha decay detected from the planchets was measured over B24 h. From the measured activity of 210Po, the activity of 210Pb was calculated. Activities were normalized to the salt-corrected dry mass and expressed as decay per minute per gram (dpm/g). All activities were decay-corrected to the time of collection.

5.3.3. Bottom-Boundary-Layer Tripods To measure sediment-transport conditions near the seabed, a series of BBL tripods were deployed, and more details of the data can be found in Ogston et al. (2008) and Martin et al. (2008). A tripod at T13-20 is used in this paper to represent conditions on the outer topset. The tripod was conﬁgured to measure nearbed currents (acoustic Doppler velocimeter, electromagnetic current meter), tides and waves (pressure sensor), suspended-sediment concentration (optical backscatterance sensor, OBS), and water-column currents (upward-looking acoustic Doppler current proﬁler). The tripod was deployed during trade-wind conditions for 4 weeks in August and September 2003. Subsequent deployments were during monsoon conditions for 4 weeks in January and February 2004, and during transition conditions for 4 weeks in April and May 2004.

5.3.4. Bottom-Boundary-Layer Data Processing 5.3.4.1. Currents Current observations were averaged over an 8-min burst collected each hour. The wave-orbital velocity was calculated using the root-mean-square of the velocity ﬂuctuations around the mean velocity in each hourly burst. The wave height and period were evaluated from the spectrum of pressure within the individual bursts. The dominant period was identiﬁed in the spectrum, and the signiﬁcant wave height was determined from the spectral energy. Current and wave velocities were used to estimate bottom shear stress (tb) according to the Grant–Madsen model (Grant and Madsen, 1979). A bed roughness of 0.2 cm was used, based on ﬁeld observations of the seabed recovered in box cores. Bottom stresses are reported as shear

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

velocities (u): rﬃﬃﬃﬃﬃ tb un ¼ r

185

(1)

where r is the density of sea water (1.027 g/cm3).

5.3.4.2. Suspended-sediment concentration Suspended-sediment concentration was determined from OBS measurements, which were calibrated with surﬁcial seabed sediment before and after each deployment. Calibrations were performed by mixing speciﬁc concentrations of suspended sediment and documenting the relationship between the concentration and the sensor response. No evidence was seen at the T13-20 site of the very-high-concentration turnaround in OBS response that has been documented by other studies (e.g., Kineke et al., 1996; Ogston et al., 2000).

5.4. Results 5.4.1. Sediment Accumulation Seven distinct styles of sediment accumulation have been identiﬁed on the Fly River shelf (Fig. 5.3). An eighth sedimentary condition is typiﬁed by uniform, supported 210Pb activities, indicating that negligible modern accumulation occurs – a condition common in the southwestern portion of the study area (transects T1, T2, and T3). Steady-state accumulation proﬁles that exhibit logarithmic decay with no SML are conﬁned to Umuda Valley (T8) and southwest of that feature. Steady-state proﬁles are further subdivided by accumulation rates with those W1.0 cm/y designated ‘‘rapid’’ or Type I, and those o1.0 cm/y designated ‘‘slow’’ or Type II (Fig. 5.3). Neither style of steady-state accumulation (rapid and slow) is present northeast of Umuda Valley and they occur primarily on transects T4, T6, T7, and T8. Site T8-60 has a proﬁle that indicates a clear change in accumulation rate because of the slope change in the decay proﬁle, and this is designated Type III. Event layers (Walsh et al., 2004) are present near the surface of cores T5-40 and T10-40 and this style is Type IV. A thick, uniform SML can unconformably overlie sediment with supported activity (T8-16, T8-18, and T10-17), which is Type V. Type VI is deﬁned as a uniform SML overlying a

186

J. S. Crockett et al.

0

0

T6-30

T8-50 20

10

40 20 60 30

80

SAR = 1.2 cm/y

100 0.1

1

SAR = 0.2 cm/y 10

40 0.1

Type I: Steady-state accumulation rate with rapid accumulation 0

0

0

SAR = 0.2 cm/y

1 Type II: Steady-state accumulation rate with slow accumulation 20

40

60

80

10

100 T5-40

event layer

20

50 T8-60

40

SAR = 0.5 cm/y

100

60 80 0.1

0 20 40 60 80 100 120 140 160 180 0.1

150

1 Type III: Steady-state accumulation at different rates

10

T10-17

200 0.1

SAR = 2.0 cm/y

1 Type IV: Event layer in surface sediment with steady-state accumulation below

10

0 T9-18 50 100 SAR = 3.0 cm/y

150

1 Type V: Uniform surface activity with background activity below

10

200 0.1

1 Type VI: Uniform surface layer with steady-state accumulation below

0 T12-35 50 Type VII: Quasi-steady-state accumulation with varying activity

100 150 SAR = 5.5 cm/y 200 0.1

1 210

Pb Activity (dpm/g)

10

10

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

187

region of logarithmic decrease (T7-30, T8-20, T9-18, T9-30, and T10-60). A ﬁnal sedimentary style is characterized by variable activity with a general trend of decreasing activity with depth in core (T11-36, T12-35, T13-30, T13-48, and T13-65) and is labeled Type VII. Non-steady-state accumulation is common northeast of T5, and is particularly prevalent in shallow depths of Umuda Valley (T8) and on transects farther northeastward. X-radiography of non-steady-state cores reveals sedimentary structure that is dominated by physical stratiﬁcation.

5.4.2. Bottom-Boundary-Layer Sediment Transport 5.4.2.1. Trade-wind conditions Waves during trade-wind conditions had a signiﬁcant height of 1.2 m with a period of B8 s. They generated a signiﬁcant wave-orbital velocity of 13.8 cm/s in 20 m of water on the open clinoform (site T13-20). The signiﬁcant tidal-current velocity was 41.6 cm/s at 1 mab (meter above bed) for the same site. Shear velocities and suspended-sediment concentrations were analyzed to compare seasonal variability in sediment resuspension. During trade-wind conditions, maximum current shear velocities (u) reached 3.4 cm/s on the open clinoform. Currents are predominantly generated by tides in this region, and periods of slack water occur with Figure 5.3: Seven styles of sediment accumulation have been identiﬁed within the seabed (vertical axis is depth in core, cm; closed circles are total 210 Pb activity; open circles are excess 210Pb activity). Type I: Steady-state sedimentation with high sediment accumulation rate (SAR) indicates relatively rapid and constant input of sediment to the seabed. Type II: Slow accumulation results from less sediment supply to a site. Type III: Changing slopes in the activity proﬁles indicate that the rate of sediment accumulation has changed at these sites. Type IV: Event layers are identiﬁed by layers with a decrease in 210Pb activity correlated with an increase in clay content (% clay shown by triangles and scale across top). Type V: Excess activity lies unconformably on sediment with only background activity, which is found predominantly in shallow (o20 m) water. Type VI: A thick uniform surface layer is a common feature in the northeastern portion of the study area. Type VII: Variable activity with a decreasing trend down core results from variations in the initial activity of accumulating sediment on timescales that are short compared to the timescale of accumulation.

188

J. S. Crockett et al.

negligible velocity. Combined wave–current shear velocities reach a maximum of 4.6 cm/s and a minimum of near zero for some slack tidal periods (Fig. 5.4). Suspended-sediment concentration varies with shear velocity (r2 ¼ 0.44), and reaches a maximum of 0.58 g/l and a minimum of 0.02 g/l at 13 cmab (centimeter above bed) (Fig. 5.5). Along-shelf proﬁling indicates that high suspended-sediment concentrations are present both to the southwest (in Umuda Valley maximum, W10 g/l) and farther northeast of transect T13 (Martin et al., 2008). T13-20 is a representative site for this region of the shelf. Comparing the shear velocity and suspended-sediment concentration shows that a critical value of wave–current shear velocity for resuspension is in the range 1.5–2.0 cm/s (Fig. 5.5). During the 28-day record of the trade-wind season, wave-current shear velocities exceeded 1.5 cm/s during 81% of the study period and 2.0 cm/s during 59% of the period.

Trade-wind conditions, T13-20 5.0 Combined wave/current shear velocity Current shear velocity (cm/s)

4.5

Shear velocity, u (cm/s) *

4.0 3.5 3.0 2.5 2.0

~Threshold of motion

1.5 1.0 0.5 0 235

240

245

250

255

260

265

Day of Year, 2003

Figure 5.4: Shear velocity during trade-wind conditions at tripod site T13-20. Shear velocities are driven predominantly by tidal ﬂow, and the fortnightly cyclicity is clearly visible. Wave–current combined shear velocities were calculated according to Grant and Madsen (1979). Maximum combined shear velocity (u) exceeds 4.0 cm/s.

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

189

0.6 ~Threshold of Motion

Suspended-sediment concentration (g/l)

0.7

0.5

0.4

0.3

0.2

0.1

0

0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Shear velocity, u wc (cm/s) *

Figure 5.5: Relationships between suspended-sediment concentrations and combined wave–current shear velocity. Suspended-sediment concentrations increase with increasing shear velocity and indicate that sediment motion begins at shear velocities between 1.5 and 2.0 cm/s.

5.4.2.2. Monsoon conditions Monsoon conditions had a signiﬁcant wave height of 0.45 m with a period similar to that of trade-wind conditions, and signiﬁcant wave-current orbital velocity of 5.5 cm/s at 20-m water depth. Signiﬁcant tidal currents were similar to those during the trade-wind season (B40 cm/s). This resulted in combined shear velocities that generally varied from 0.5 to 3.5 cm/s. Suspended-sediment concentrations were also lower during this period than in the trade-wind season, reaching a maximum of 0.3 g/l at 13 cmab, but having an average concentration an order of magnitude less. During monsoon conditions, suspended-sediment concentration also increased with shear velocity (r2 ¼ 0.52), and initiation of motion occurs at shear stresses similar to those of the trade-wind period. Combined wave–current shear velocities over the 28-day period exceeded 1.5 cm/s during 51% of the time and exceeded 2.0 cm/s during 31% of the time (28 days).

190

J. S. Crockett et al.

5.5. Discussion Sediment accumulation on the Fly River shelf should be viewed in the larger context of clinoform development, which has occurred in several episodes. Recent seismic and multibeam studies of the GOP bathymetry document the modern clinoform foreset, which extends from B20 to B65 m and downlaps onto an erosional surface. Variability of sediment accumulation is expected because the history of clinoform evolution in the GOP has revealed an irregular basal surface (i.e., stream-eroded; Slingerland et al., 2008) over which the modern clinoform is prograding. Many mechanisms for clinoform progradation have been proposed and documented around the world. These include wave-supported turbidity currents (Kostic et al., 2002), ﬂuid-mud ﬂows (Kineke et al., 1996; Kuehl et al., 1996), and advection/diffusion of suspended sediment (Pirmez et al, 1998). Recently, gravity-driven processes have gained popularity as a necessary mechanism of clinoform progradation. Turbidity currents, ﬂuid muds, and hyperpycnal plumes are able to deliver large masses of sediment necessary to account for the high accumulation rates found on the foreset of clinoforms, and they have all been implicated in the creation of the largest modern clinoforms (Amazon, Kineke et al., 1996; Kuehl et al., 1996; Ganges–Brahmaputra, Kuehl et al., 1997; Michels et al., 1998). These gravity-driven processes leave non-steady-state signatures in the seabed, similar to those found in the study area (Fig. 5.3, Type IV). The Fly River shelf exhibits eight recognizable styles of sedimentation that can be grouped into three general categories: negligible accumulation, steady-state accumulation, and non-steady-state accumulation. The spatial variability in distribution of accumulation signatures (Fig. 5.6) indicates that multiple mechanisms are in operation on this shelf.

5.5.1. Negligible Accumulation At several stations, there is no excess 210Pb and this indicates that negligible accumulation is occurring at these sites. These sites occur on the outer topset (15–25 m), where waves and tidal currents regularly rework the seabed and prevent accumulation of new sediment. These areas were observed to be capable of resuspension at least 59% of the time during trade-wind conditions and 31% of the time during monsoon conditions. Consequently, physical stratiﬁcation is observed in X-radiographs (Fig. 5.7). Transects T1 and T3 have no excess 210Pb in cores collected from W25-m water depth, which likely reﬂects limited sediment supply in this region of the study area.

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

8°15′

191

NA NA NSS VH,V

NA 8°30′S

NSS VH,V NA NSS H.V NSS VH,V NSS VH,V

UL BKD UL BKD UL SS SSL

8°45′S

SSL UL SS

NA SSL

9°S

NA NA

NA SSL

UL BKD NA NA UL GDSSh UL SSH UL SSH UL SSH SSH UL SS,CR

NA UL GDSSH NA SS L NA Title (surface layer profile) SSH NA NA - no apparent accumulation

9°15′S

143°45′E

SS - Steady-state accumulation NSS - Non-steady-state accumulation UL - Surface uniform layer Subscript (sub-layer profile) L - Steady-state accumulation (< 1.0 cm/y) H - Steady-state accumulation (> 1.0 cm/y) GD - Gravity-driven layer SS - Steady-state accumulation BKD - Background activity only CR - Changing accumulation rates VH - Very high accumulation rate (> 2.0 cm/y) V - Variable activity, steady-state accumulation

144°E

144°15′E

Figure 5.6: The distribution of sedimentary styles on the Fly River shelf. The ﬁrst designation indicates the character of the surface sediment. The subscripts describe the character of the subsurface sediment. Sites with negligible accumulation are dominant in the southwestern portion of the study area. The northeastern portion is dominated by sites with very high accumulation rates and variable 210Pb activities. The central region has transitional sedimentary styles.

192

J. S. Crockett et al.

T3-20

T4-17

T5-20

T6-20

0 cm

5 cm

30 cm

Figure 5.7: X-radiographs (negatives) of cores from the topset are generally dominated by physical stratiﬁcation generated by the strong waves and tidal currents, even in the southwestern area of relatively slow sediment accumulation (transects T3 to T6). A variation on negligible accumulation at B20-m water depth occurs on transects T8 and T10. Sites T8-16, T8-18, and T10-17 all have a deposit with uniform excess activity at the surface and supported activities beneath (Type V). These stations are in or near Umuda Valley, which maintains some of the highest suspended-sediment concentrations on the entire shelf (Martin et al., 2008). Today, Umuda Valley is a major focus of sediment supply from the Fly River. When large amounts of sediment are delivered to local areas, stresses due to tidal currents are insufﬁcient to keep it all in suspension and temporary deposition results, but not long-term accumulation.

5.5.2. Steady-State Accumulation Steady-state accumulation occurs at many sites throughout the study area, and many of these proﬁles also have an SML. Two sites without a mixed layer are located on the foreset of the clinoform entering Kiwai Valley (T4-30 and T4-50), two other sites are between Kiwai Valley and Umuda Valley (T6-30 and T7-20), and one site is within Umuda Valley (T8-50). Northeast

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

193

of Umuda Valley, there are no examples of steady-state accumulation without uniform surface layers. The absence of an SML implies that physical and biological processes are not active in mixing the seabed. At the sites that additionally exhibit steady-state accumulation, the generation of the observed proﬁle is driven by accumulation, the initial activity of the accumulating sediment, and the decay constant of 210Pb. Two of the sites with slow steady-state sediment accumulation are located on the foreset of the clinoform west of Kiwai Valley (e.g., T4-30 and T4-50). Similar proﬁles are observed for the Holocene clinoform of the Huang He (Yellow River) (Alexander et al., 1991). Because of river avulsion and sea-level rise, today the river mouth is displaced by several hundred kilometers from the clinoform built B11 kya (thousand years ago). Modern accumulation rates on the clinoform are steady state (topset 0.6–1.2 cm/y, foreset 0.3 cm/y, bottomset negligible; Liu et al., 2004), but maximum rates occur on the topset, which is incompatible with models of clinoform progradation. Most of the Huang He sediment is deposited via hyperpycnal plumes at the river mouth (Bornhold et al., 1986; Wright et al., 1986, 1988), indicating that the clinoform is not a likely sink for much sediment today. Slow rates of steady-state accumulation also are observed in the Adriatic Sea (Palinkas and Nittrouer, 2006), where a clinoform is generated by sediment from the Apennine mountain range. The Apennine rivers were collectively dammed at the end of World War II (Capelli et al., 1997; Coltorti, 1997; Farroni et al., 2002; Syvitski and Kettner, 2007), which dramatically decreased their sediment supply to the Adriatic Sea. Prior to that change, the collective line source of rivers was actively building a clinoform on the western side of the Adriatic Sea. Modern accumulation rates are slow, and do not always show the highest accumulation rates on the foreset, again a condition that is not conducive to the active progradation of a self-similar deposit. Considering the documented conditions in the Yellow Sea and the Adriatic Sea, the low rates of sediment accumulation in the southwestern portion of the Fly River clinoform likely reﬂect a stage of negligible progradation.

5.5.3. Non-Steady-State Accumulation 5.5.3.1. Deposits of uniform

210

Pb activity

Surface layers with uniform 210Pb activities are created by biological mixing (e.g., Washington shelf; Nittrouer et al., 1979) or physical mixing

194

J. S. Crockett et al.

(e.g., Amazon shelf; Kuehl et al., 1996) that homogenize the 210Pb activity to the depth of mixing. Under steady-state accumulation with an SML, sediment is deposited on the seabed, an equal amount is displaced below the region of active mixing, and the SML migrates upward with the sediment surface. The 210Pb in the displaced sediment continues to decay with its characteristic half-life and an accumulation proﬁle develops below the SML. At most locations with uniform 210Pb proﬁles on the Fly River shelf, the SML is B1 m thick. Some examples of bioturbation are found along the shelf, although burrows and trace evidence of biological mixing are generally not common. Organisms found in the GOP mix only the upper few centimeters of the seabed (Aller and Aller, 2004). Typically, biological mixing on continental shelves does not extend much below 10 cm into the sediment (Rhoads et al., 1985), so benthic organisms are unlikely to account for the thickness of many mixed layers on this shelf. Physical mixing can resuspend sediment deep into the seabed. On the Amazon shelf, resuspension has been observed in the form of a seasonal surface layer that extends more than a meter into the seabed (Kuehl et al., 1996). The magnitude of resuspension is driven by the energetic oceanographic conditions present on the Amazon shelf. Tides, which are a dominant physical forcing mechanism there, have a maximum range of B6 m and generate currents as great as 200 cm/s (Gibbs, 1982; Nittrouer et al., 1986b; Kineke et al., 1996). Tidal currents play a signiﬁcant role in reworking the seabed up to 10-cm depth over a diurnal tidal cycle (Jaeger and Nittrouer, 1995). In addition, signiﬁcant wave heights average 1–2 m (Meserve, 1974) and maximum signiﬁcant wave heights W3.0 m (13 s period) (Sternberg et al., 1996) contribute to seabed mixing. In contrast, maximum tidal currents on the Fly River shelf are o100 cm/s (Martin et al., 2008). The maximum signiﬁcant wave height measured during the study period was 3.0 m. The mean signiﬁcant wave height during trade-wind conditions was 1.2 m, but decreased to 0.5 m during the monsoon. Resuspension depths due to wave-orbital velocities have been calculated for a variety of wave conditions on the Eel River shelf (Wiberg, 2000), which is a silt-dominated environment (Borgeld, 1985; Borgeld et al., 1996; Sommerﬁeld and Nittrouer, 1999; Crockett and Nittrouer, 2004) similar to the Fly River shelf. Under wave-orbital velocities that vary from 50 to 200 cm/s, resuspension on the Eel River shelf was always o10 cm into the seabed (Wiberg, 2000). On the Fly River shelf, signiﬁcant wave-orbital velocities at 20-m water depth during the trade-wind period averaged only 13.8 cm/s, and wave resuspension is not expected to extend into the seabed more than a few centimeters.

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

195

5.5.3.2. Mechanisms for deposition of uniform layer Tidal currents generate a signiﬁcant portion of the shear stress on the Fly River shelf, with waves imprinting a variable increase in calculated shear velocities (Fig. 5.4). The combined stresses are sufﬁcient to transport sediment (at least in shallow water) a majority of the time, and this transport likely plays a role in the development of the thick mixed layers. In shallow water where thick deposits with uniform 210Pb are observed, physical sedimentary structures indicate that sediment resuspension and deposition is important in creating the stratigraphy. The unconformable contact between the surface deposit with excess 210Pb and the underlying deposit with supported activities (Fig. 5.8) indicates an erosive boundary between the two strata. The regular occurrence of tidal currents combined with variations in sediment supply may control sedimentation in this region. When sediment is in short supply, the currents and waves progressively erode the seabed. The depth of erosion is not the result of any one period of exceptional shear stress; rather it is the result of regular incidence of shear stresses sufﬁcient to transport sediment. This process continues until new sediment is supplied. In cores that have excess 210Pb activity lying Water Content (%) 0

10

20

30

40

50

60

0 T8-18 50

Depth (cm)

100

150

Total 210Pb Activity

200

Water Content 250 0.1

1

10

Total 210Pb Activity (dpm/g)

Figure 5.8: Water content correlates with 210Pb activity. More-consolidated subsurface sediment is resistant to erosion during transport events.

196

J. S. Crockett et al.

unconformably on supported sediment, a discontinuity in water content (and porosity) also exists (Fig. 5.8). More-consolidated subsurface sediment is resistant to erosion, creating conditions for the surface layer to be periodically removed and redeposited with little impact on the subsurface layer. When large amounts of sediment are available, the transport capacity of tidal currents may be overwhelmed and sediment deposited. Sediment deposition occurs on a timescale that is short compared to the resolution of 210 Pb. With initial (surface) excess activities of o2 dpm/g and analytical errors of about 70.3 dpm/g, sedimentation can be resolved for a timescale of B6 y (Kuehl et al., 1995). This allows rapid deposition on a tidal scale (fortnightly or monthly) to create thick surface layers with uniform 210Pb activities. In deeper water, wave and tidal shear stresses are weaker and not able to facilitate this process, conﬁning thick uniform layers to shallow water where wave and tidal shear stresses are strongest.

5.5.3.3. Event deposition Low-activity sediment associated with elevated clay content has been observed previously by Walsh et al. (2004) on the foreset northeast of transect T13. In the present study, sites T5-40 and T10-40 exhibit 210Pb and grain-size proﬁles that are consistent with an interpretation of event sedimentation (Fig. 5.9). The event recorded near the surface of these cores overlies a region of logarithmic decay, indicating that under usual conditions, the site was undergoing steady-state accumulation of sediment. Most known processes would not be able to transport ﬁne-grained sediment to the observed accumulation depth (B40 m) while preserving the low-activity signature. BBL transport in the absence of a lutocline would generate mixing with ambient water, thereby increasing the surface activity of the suspended sediment. Settling through the water column would similarly expose the accumulating sediment to higher concentrations of dissolved 210Pb. Furthermore, the sedimentary characteristics of the deposits (e.g., physical structure and thickness; Fig. 5.10) imply that they were emplaced during a single depositional event (possibly with multiple pulses), and involved a large amount of sediment in transport. Water-column studies have identiﬁed nearbottom suspended-sediment concentrations greatly exceeding the 10 g/l threshold needed for gravity ﬂows (Ogston et al., 2008; Martin et al., 2008). This demonstrates that ﬂuid-mud conditions exist at discrete locations. Shortperiod waves (3–8 s), like those found in the GOP, have been demonstrated to create thin (o2 cm) layers of ﬂuid mud in experiments (Lamb and Parsons, 2005), and these thin layers are difﬁcult to observe under ﬁeld conditions.

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

197

T5-40 % Clay Composition 0

20

40

60

80

100

0

event layer

Depth (cm)

50

Total 210Pb activity Excess 210Pb activity

100

150 SAR = 2.0 cm/y 200 0.1

1 210Pb

10

Activity (dpm/g)

Figure 5.9: Details of core T5-40 showing an event layer with decreased 210 Pb activity correlated with increased clay content. The event layers are physically stratiﬁed in X-radiographs (see Fig. 5.10). The suite of evidence strongly suggests that high-concentration gravity-driven ﬂows are responsible for the event deposits in the GOP.

5.5.3.4. Fast accumulation rates with variable

210

Pb activity

Variable initial activity of accumulating sediment precludes development of a steady-state 210Pb proﬁle. However, in other systems (e.g., Amazon, Ganges– Brahmaputra) that exhibit variable 210Pb proﬁles with a general decay trend, using the steady-state accumulation model with best-ﬁt lines has generated accumulation rates in good agreement with independent geochronometers (e.g., 228Ra/226Ra; Dukat and Kuehl, 1995; Michels et al., 1998). The best-ﬁt approach seems to be a reliable estimate where a general decay trend is observed in the 210Pb activity proﬁle. Such a trend indicates that accumulation and radioactive decay dominate the activity proﬁle even in the presence of variations in initial activity. Typically, steady-state proﬁles exhibit minimal scatter around a regression line, but some variability is unavoidable

198

J. S. Crockett et al.

Core T5-40 0 cm

10 cm

Event Layer

40 cm

Figure 5.10: X-radiograph (negative) of core T5-40. The event layer contains interlaminated sand (light colored) and mud (dark colored). Overlying sediment shows evidence of biological activity through the presence of biogenic sedimentary structures. because of factors mentioned previously. In the case of proﬁles from the northeast part of the study area, the scatter is large and the proﬁles are interpreted to be non-steady-state because the variations in activity are great compared to the mean activity in the sediment. In these conditions, it is valuable to examine the proﬁles to gain insight regarding sediment dynamics in the context of longer-term accumulation. Low-activity sediment is found on the topset because of limited interaction with the oceanic source of dissolved 210Pb. In addition, high concentrations of suspended sediment that compete for the small amount of excess 210Pb in the water column (see Section 5.2) are common in shallow water and can be transported to accumulate in deeper water. Low concentrations of sediment that have ample interaction with ocean water can conversely be expected to exhibit higher activity. An example of this effect is shown by a study performed near the mouth of the Yangtze River. Where the suspended-sediment

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

199

concentration exceeds 100 mg/l, McKee et al. (1984) found that 95% of the 234 Th inventory (another particle-reactive radioisotope produced from 238U in seawater) is associated with particles. In regions where the suspended-sediment concentration is lower by more than an order of magnitude (10 mg/l), particle association of 234Th is decreased to only B60%. This indicates that under lower-concentration conditions, individual particles are able to scavenge a much higher proportion of the available particle-reactive isotopes in the water column. Therefore, low-concentration conditions can produce a higher speciﬁc activity in the seabed, and temporally variable suspended-sediment concentrations may contribute to the generation of non-steady-state proﬁles. Non-steady-state proﬁles also have been observed on the Amazon continental shelf, where they result from thick seasonally mixed surface layers on the clinoform topset that are mobilized as ﬂuid muds and ﬂow to the foreset region (Kuehl et al., 1995). 210Pb proﬁles in foreset cores demonstrate variable activity reﬂecting pulses of sediment input from the topset region (Kuehl et al., 1986). These are subsequently buried by sediment with high concentrations of dissolved 210Pb, which is adsorbed by particles settling through isotope-rich open-ocean water (Moore et al., 1996). The periodic input of low-activity sediment from near-bottom ﬂuid-mud ﬂows interspersed with high-activity sediment deposited under conditions of lower suspended-sediment concentration results in the observed quasi-cyclic proﬁles (Kuehl et al., 1996). Foreset cores from the GOP have 210Pb proﬁles with similarities to cores from the Amazon shelf, and we interpret low-activity strata to form through deposition of high-concentration nearbottom suspensions that originate on the topset.

5.6. Conclusions A high level of heterogeneity characterizes the sedimentary record along and across the Fly River shelf. Seven distinct styles of accumulation are recognized, resulting in sediment accumulation rates that range from 0.2 to 5.0 cm/y, with an eighth pattern described by the lack of any detectable modern (100 y) accumulation. Preservation of sedimentary strata is also variable on this shelf and is largely controlled by the intensity of physical processes and the local rates of sediment accumulation. Cores collected from the continental shelf directly off the Fly River mouth reveal accumulation rates that are slow (o1 cm/y) or negligible. These observations indicate that progradation of the clinoform is presently limited in the southwestern portion of the study area.

200

J. S. Crockett et al.

Cores collected northeast of the Fly River mouth contain sediment layers with low 210Pb activity and high clay content. Low-activity sediment is generally deposited under conditions of large suspended-sediment concentrations. To emplace the thick deposits observed at the water depths where they are found, sediment gravity ﬂows are a likely mechanism. Strong tidal currents occur at regular intervals on this shelf and shear stresses are sufﬁcient to transport sediment on the outer topset a majority of the time during all seasons (up to 80% during trade-wind conditions and 50% during monsoon conditions); thick, uniform 210Pb layers are observed at many sites in the central and northeastern portion of the study area. Some uniform layers are likely mixed by biology, but those layers rarely extend deeper than 10 cm into the seabed and biological mixing cannot account for the thickness of all the observed uniform layers. Shear stresses are also insufﬁcient to resuspend sediment in a single event and homogenize the thicknesses observed. Generation of the uniform layers is likely related to recurring tidal currents. When supply is limited, tidal currents and waves erode the surface layer. When supply is ample, rapid deposition (in relation to the timescales measured by 210 Pb) occurs and creates a deposit that exhibits uniform activities.

ACKNOWLEDGMENTS We thank all US, PNG, and Australian scientists who helped with ﬁeldwork, including the crew of the R/V Melville. This paper is based on work supported by the National Science Foundation under Grant Nos. OCE0203351 and OCE0504616 (to Nittrouer and Ogston) and OCE0220600 (to Goni).

REFERENCES Alexander, C. R., DeMaster, D. J., & Nittrouer, C. A. (1991). Sediment accumulation in a modern epicontinental-shelf setting: The Yellow Sea. Marine Geology, 98, 51–72. Aller, J. Y., & Aller, R. C. (2004). Physical disturbance creates bacterial dominance of benthic biological communities in tropical deltaic environments of the Gulf of Papua. Continental Shelf Research, 24, 2395–2416. Asquith, D. O. (1970). Depositional topography and major marine environments, Late Cretaceous, Wyoming. American Association of Petroleum Geologists Bulletin, 54, 1184–1224. Borgeld, J. C. (1985). Holocene stratigraphy and sedimentation on the northern California continental shelf. Ph.D. Dissertation, University of Washington, Seattle, 177 pp.

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

201

Borgeld, J. C., Clarke, J. E. H., Goff, J. A., Mayer, L. A., & Curtis, J. A. (1996). Acoustic backscatter of the 1995 ﬂood deposit on the Eel Shelf. Marine Geology, 154, 197–210. Bornhold, B. D., Yang, Z.-S., Keller, G. H., Prior, D. B., Wiseman, W. J., Jr., Wang, Q., Wright, L. D., Xu, W. D., & Zhuang, Z. Y. (1986). Sedimentary framework of the modern Huanghe (Yellow River) delta. Geo-Marine Letters, 6, 77–83. Bruland, K. W. (1974). Geochronology in the coastal marine environment. Ph.D. Dissertation, University of California, San Diego, 106 pp. Capelli, G., Miccadei, E., & Rafﬁ, R. (1997). Fluvial dynamics in the Castel di Sangro plain: Morphological changes and human impact from 1875 to 1992. Catena, 30, 295–309. Coltorti, M. (1997). Human impact in the Holocene ﬂuvial and coastal evolution of the Marche region, Central Italy. Catena, 30, 311–335. Crockett, J. S., & Nittrouer, C. A. (2004). The sandy inner shelf as a repository for muddy sediment: An example from northern California. Continental Shelf Research, 24, 55–73. Crockett, J. S., Nittrouer, C. A., Ogston, A. S., Naar, D. F., & Donohue, B. T. (2008). The origin, evolution and ﬁlling of exposed submarine valleys on the Fly River shelf. Journal of Geophysical Research, 113, F01S12, doi: 10.1029/2006JF000674. Dukat, D. A., & Kuehl, S. A. (1995). Non-steady-state Pb-210 ﬂux and the use of Ra-228/Ra-226 as a geochonometer on the Amazon continental shelf. Marine Geology, 125, 329–350. Farroni, A. M. V., Magaldi, D., & Tallini, M. (2002). Total sediment transport by the rivers of Abruzzi (central Italy): Prediction with the RAIZAL model. Bulletin of Engineering Geology and the Environment, 61, 121–127. Fisher, W. L., Brown, A. J. S., & McGowan, J. H. (1969). Delta Systems in the Exploration for Oil and Gas. Texas Bureau of Economic Geology, Austin, 97 pp. Galloway, W. E. (1975). Process framework for describing the morphologic and stratigraphic evolution of deltaic depositional systems. In: M. L. Brousard (Ed.). Deltas Models for Exploration. Houston Geological Society, Houston. Gibbs, R. J. (1982). Currents of the shelf of north-eastern South America. Estuarine Coastal and Shelf Science, 14, 238–299. Grant, W. D., & Madsen, O. S. (1979). Combined wave and current interaction with a rough bottom. Journal of Geophysical Research, 84, 1797–1808. Harris, P. T., Baker, E. K., Cole, A. R., & Short, S. A. (1993). A preliminary study of sedimentation in the tidally dominated Fly River delta, Gulf of Papua. Continental Shelf Research, 13, 441–472. Jaeger, J. M., & Nittrouer, C. A. (1995). Tidal controls on the formation of ﬁnescale sedimentary strata near the Amazon river mouth. Marine Geology, 125, 259–281. Kineke, G. C., Sternberg, R. W., Trowbridge, J. H., & Geyer, W. R. (1996). Fluidmud processes on the Amazon continental shelf. Continental Shelf Research, 16, 667–696.

202

J. S. Crockett et al.

Kostic, S., Parker, G., & Marr, J. G. (2002). Role of turbidity currents in setting the foreset slope of clinoforms prograding into standing fresh water. Journal of Sedimentary Research, 72, 353–362. Kuehl, S. A., DeMaster, D. J., & Nittrouer, C. A. (1986). Nature of sediment accumulation on the Amazon continental shelf. Continental Shelf Research, 6, 209–225. Kuehl, S. A., Hariu, T. M., & Moore, W. S. (1989). Shelf sedimentation off the Ganges-Brahmaputra river system: Evidence for sediment bypassing to the Bengal fan. Geology, 17, 1132–1135. Kuehl, S. A., Pacioni, T. D., & Rine, J. M. (1995). Seabed dynamics of the inner Amazon continental shelf: Temporal and spatial variability of surﬁcial strata. Marine Geology, 125, 283–302. Kuehl, S. A., Nitttouer, C. A., Allison, M. A., Faria, L. E. C., Dukat, D. A., Jaeger, J. M., Pacioni, T. D., Figueiredo, A. G., & Underkofﬂer, E. C. (1996). Sediment deposition, accumulation, and seabed dynamics in an energetic ﬁne-grained coastal environment. Continental Shelf Research, 16, 787–816. Kuehl, S. A., Levy, B. M., Moore, W. S., & Allison, M. A. (1997). Subaqueous delta of the Ganges-Brahmaputra river system. Marine Geology, 144, 81–96. Lamb, M. P., & Parsons, J. D. (2005). High-density suspensions formed under waves. Journal of Sedimentary Research, 75, 386–397. Leithold, E. L. (1993). Preservation of laminated shale in ancient clinoforms: Comparison to modern subaqueous deltas. Geology, 21, 359–362. Liu, J. P., Milliman, J. D., Gao, S., & Cheng, P. (2004). Holocene development of the Yellow River’s subaqueous delta, North Yellow Sea. Marine Geology, 209, 45–67. Martin, D. P., Nittrouer, C. A., Ogston, A. S., & Crockett, J. S. (2008). Tidal and seasonal dynamics of a muddy inner-shelf environment, Gulf of Papua. Journal of Geophysical Research, 113, F01S07, doi: 10.1029/2006JF000681. McKee, B. A., DeMaster, D. J., & Nittrouer, C. A. (1984). The use of Th-234/U-238 disequilibrium to examine the fate of particle-reactive species on the Yangtze continental shelf. Earth Planetary Science Letters, 68, 431–442. Meserve, J. M. (1974). US Navy marine climatic atlas of the world, Vol. 1, North Atlantic Ocean. United States Government Printing Ofﬁce, Washington, DC, 371 pp. Miall, A. D. (1984). Deltas. In: R. G. Walker (Ed.). Facies Models. Geological Association of Canada, St. John’s, NL, pp. 106–118. Michels, K. H., Kudrass, H. R., Huebscher, C., Suckow, A., & Wiedicke, M. (1998). The submarine delta of the Ganges-Brahmaputra; cyclone-dominated sedimentation patterns. Marine Geology, 149, 133–154. Milliman, J. D. (1995). Sediment discharge to the ocean from small mountainous rivers: The New Guinea example. Geo-Marine Letters, 15, 127–133. Milliman, J. D., & Meade, R. H. (1983). World-wide delivery of river sediments to the oceans. Journal of Geology, 91, 1–21. Mitchum, R. M., Jr., Vail, P. R., & Sangree, J. B. (1977). Seismic stratigraphy and global changes of sea level, part 6: Stratigraphic interpretation of seismic

Variable Styles of Sediment Accumulation Impacting Strata Formation on a Clinoform

203

reﬂection patterns in depositional sequences. In: C. E. Payton (Ed.). Seismic Stratigraphy – Applications to Hydrocarbon Exploration. American Association of Petroleum Geologists, Tulsa, pp. 117–133. Moore, W. S., DeMaster, D. J., Smoak, J. M., McKee, B. A., & Swarzenski, P. W. (1996). Radionuclide tracers of sediment-water interactions on the Amazon shelf. Continental Shelf Research, 16, 645–665. Nittrouer, C. A., Sternberg, R. W., Carpenter, R., & Bennett, J. T. (1979). The use of Pb-210 geochronology as a sedimentological tool: Application to the Washington continental shelf. Marine Geology, 31, 297–316. Nittrouer, C. A., DeMaster, D. J., Kuehl, S. A., & Kowsmann, R. O. (1986a). The deltaic nature of Amazon shelf sedimentation. Geological Society of America Bulletin, 97, 444–458. Nittrouer, C. A., Curtin, T. B., & DeMaster, D. J. (1986b). Concentration and ﬂux of suspended sediment on the Amazon continental shelf. Continental Shelf Research, 6, 151–174. Nittrouer, C. A., Brunskill, G. J., & Figueiredo, A. G. (1995). The importance of tropical coastal environments. Geo-Marine Letters, 15, 121–126. Nittrouer, C. A., Kuehl, S. A., Figueiredo, A. G., Allison, M. A., Sommerﬁeld, C. K., Rine, J. M., Faria, E. C., & Silveira, O. M. (1996). The geological record preserved by Amazon shelf sedimentation. Continental Shelf Research, 16, 817–841. Ogston, A. S., Cacchione, D. A., Sternberg, R. W., & Kineke, G. C. (2000). Observations of storm and river ﬂood-driven sediment transport on the northern California continental shelf. Continental Shelf Research, 20, 2141–2162. Ogston, A. S., Sternberg, R. W., Nittrouer, C. A., Martin, D. P., Goni, M. A., & Crockett, J. S. (2008). Sediment delivery from the Fly River tidally dominated delta to the nearshore marine environment and the impact of El Nino. Journal of Geophysical Research, 113, F01S11, doi: 10.1029/2006JF000669. Palinkas, C. M., & Nittrouer, C. A. (2006). Clinoform sedimentation along the Apennine shelf, Adriatic Sea. Marine Geology, 234, 245–260. Pickup, G. (1984). Geomorphology of tropical rivers: I. Landforms, hydrology and sedimentation in the Fly and lower Purari, Papua New Guinea. In: A. P. Schick (Ed.). Channel Processes – Water, Sediment, Catchment Controls. Catena Supplement, 5, 1–17. Pielke, R. A. (1990). The Hurricane. Routledge, New York. Pirmez, C., Pratson, L. F., & Steckler, M. S. (1998). Clinoform development by advection-diffusion of suspended sediment: Modelling and comparison to natural systems. Journal of Geophysical Research, 103(B10), 24141–24157. Posamentier, H. W., Jervey, M. T., & Vail, P. R. (1988). Eustatic controls on clastic deposition I. Conceptual framework. In: C. K. Wilgus, C. G. Hastings, S. C. Kendall, H. W. Posamentier, C. A. Ross, & J. C. Van Wagoner (Eds). Sea-Level Changes: An Integrated Approach. Society for Sedimentary Geology, Tulsa, pp. 125–154. Rhoads, D. C., Boesch, D. F., Zhican, T., Fengshan, X., Liqiang, H., & Nilsen, K. L. (1985). Macrobenthos and sedimentary facies on the Changjiang delta platform and adjacent continental shelf, East China Sea. Continental Shelf Research, 4, 189–213.

204

J. S. Crockett et al.

Ross, M. A., & Mehta, A. (1989). On the mechanics of lutoclines and ﬂuid mud. Journal of Coastal Research, 5, 51–61. Slingerland, R., Driscoll, N. J., Milliman, J. D., Miller, S. R., & Johnstone, E. A. (2008). Anatomy and growth of a Holocene clinothem in the Gulf of Papua. Journal of Geophysical Research, 113, F01S13, doi: 10.1029/2006JF000628. Sommerﬁeld, C. K., & Nittrouer, C. A. (1999). Modern accumulation rates and a sediment budget for the Eel shelf: A ﬂood-dominated depositional environment. Marine Geology, 154, 227–241. Sternberg, R. W., Cacchione, D. A., Paulson, B., Kineke, G. C., & Drake, D. E. (1996). Observations of sediment transport on the Amazon subaqueous delta. Continental Shelf Research, 16, 697–715. Syvitski, J. P. M., & Kettner, A. J. (2007). On the ﬂux of water and sediment into the northern Adriatic. Continental Shelf Research, 27, 296–308. Thom, B. G., & Wright, L. D. (1983). Geomorphology of the Purari Delta. In: T. Petr (Ed.). The Purari: Tropical Environment of a High Rainfall River Basin. Dr. W. Junk Publishers, The Hague, 624 pp. Vail, P. R. (1987). Seismic stratigraphy interpretation using sequence stratigraphy. Part 1. Seismic stratigraphy interpretation procedure. AAPG Studies in Geology, 27, 1–10. Vail, P. R., Mitchum, J., Todd, R. G., Widmier, J. M., Thompson, S., III, Sangree, J. B., Bubb, J. N., & Hatlelid, W. G. (1977). Seismic stratigraphy and global changes of sea level. In: C. E. Clayton (Ed.). Seismic Stratigraphy: Applications to Hydrocarbon Exploration. American Association of Petroleum Geologists, Tulsa, pp. 49–212. Van Wagoner, J. C., Mitchum, R. M., Campion, K. M., & Rahmanian, V. D. (1990). Siliciclastic sequence stratigraphy in well logs, cores and outcrops: Concepts for high resolution correlation of time and facies. American Association of Petroleum Geologists, Tulsa. Walsh, J. P., Nittrouer, C. A., Palinkas, C. M., Ogston, A. S., Sternberg, R. W., & Brunskill, G. (2004). Clinoform mechanics in the Gulf of Papua, New Guinea. Continental Shelf Research, 24, 2487–2510. Wiberg, P. (2000). A perfect storm: Formation and potential for preservation of storm beds on the continental shelf. Oceanography, 13, 93–99. Wolanski, E., & Alongi, D. M. (1995). A hypothesis for the formation of a mud bank in the Gulf of Papua. Geo-Marine Letters, 15, 166–171. Wright, L. D. (1977). Sediment transport and deposition at river mouths: A synthesis. Geological Society of America Bulletin, 88, 857–868. Wright, L. D. (1985). River deltas. In: R. A. Davis (Ed.). Coastal Sedimentary Environments. Springer-Verlag, New York, pp. 1–76. Wright, L. D., Yang, Z.-S., Bornhold, B. D., Keller, G. H., Prior, D. B., & Wiseman, W. J., Jr. (1986). Hyperpycnal plumes and plume fronts over the Huanghe (Yellow River) delta front. Geo-Marine Letters, 6, 97–105. Wright, L. D., Wiseman, W. J., Bornhold, B. D., Prior, D. B., Suhayda, J. N., Keller, G. H., Yang, Z.-S., & Fan, Y. B. (1988). Marine dispersal and deposition of Yellow River silts by gravity-driven underﬂows. Nature, 332, 629–632.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00406-0

Chapter 6

A Mass Balance for Sediment and Copper in the Rivers, Estuaries, Shelf and Slope of the Gulf of Papua, Papua New Guinea Gregg J. Brunskill1,, Irena Zagorskis2 and John Pﬁtzner2 1

84 Alligator Creek Road, Alligator Creek, Queensland 4816, Australia Australian Institute of Marine Science PMB 3, Townsville MC, Queensland 4810, Australia 2

6.1. Introduction A steady-state ocean is a useful paradigm in chemical oceanography. All that is added to the ocean must be removed on decadal to millennial scales, so that major and minor element concentrations and ratios in seawater remain similar over geologically long periods of time (Rubey, 1951; Mackenzie and Garrels, 1966; Garrels and Mackenzie, 1971; Lowenstein et al., 2005). Continental movements from polar zones to the tropics, tectonic uplift of mountain ranges, rifting ocean basins and submarine vents, climatic change, glacial/ interglacial sea level changes, biological evolution, and anthropogenic alterations of the land and rivers have changed the rate of addition of different elements (and their isotopic ratios) to the sea (Edmond, 1992; Richter and Turekian, 1993; Edmond and Huh, 1997; Schlesinger, 1997; Hooke, 2000; Peizhen et al., 2001; Vorosmarty et al., 2003; Meybeck and Vorosmarty, 2005; Syvitski et al., 2005). For elements that do not have a gas phase (that can escape to the atmosphere), changes in the rate of supply of the elements to the ocean can often be seen in the sedimentation history book. For river inputs, most dissolved and particulate elements are removed by sedimentation in estuaries and on the continental shelf. Of the trace elements, 50–95% of river inputs are probably trapped in organic and inorganic phases close to the estuaries of the major rivers of the world (Martin and Windom, 1991). Corresponding author. Tel.: 07 4778 8142; Fax: 07 4772 5852;

E-mail: [email protected] (G.J. Brunskill).

206

G. J. Brunskill et al.

Dissolved and particulate copper (pCu) is supplied to the world ocean from rivers, atmospheric dust, and submarine vents. Global river supply of Cu to the ocean is estimated to be 30 109 mol yr1 (76% in the particulate form, according to Martin and Meybeck, 1979; Carey et al., 2002), and input from atmospheric precipitation and dust might be 3–6 109 mol yr1 (Mackenzie and Wollast, 1977). The global inﬂux of Cu from submarine vents to the world ocean is poorly known (James et al., 1995), but Lackschewitz et al. (2004) indicate that black smoker vents in the Manus Basin (NE Papua New Guinea) produce mineral deposits greatly enriched in Cu and other trace elements. Various authors (e.g., Gibbs, 1973) have shown that most (75%) Cu supply from rivers is in the crystalline mineral particulate form, and that riverine dissolved Cu (dCu) (or Cu–ligand, Cu–colloid) is usually removed to suspended particles or sediments along the salinity gradient in estuaries on the inner shelf (Sholkovitz and Copland, 1981; Boyle et al., 1982; Windom et al., 1988; Cobelo-Garca and Prego, 2003; Chadwick et al., 2004; Cobelo-Garcia et al., 2004; Monbet, 2004; Jiann et al., 2005). In estuarine water and seawater, most Cu is in the particulate form as organic matter or aluminosilicate phases. Dissolved copper is thought to be almost completely associated with dissolved organic matter (proteins and polysaccharides), the colloidal phases, or in ionic ligand form with carbonate, hydroxide, and chloride (Massey, 1973; Gordon, 1992; Gordon et al., 1996; Moffett and Brand, 1996; Li, 2000; Shafer et al., 2004; Heidmann et al., 2005). Variations in the distribution coefﬁcient of Cu between particulate and dissolved phases (Kd ¼ [mmol Cu g1 particulate matter]/[mmol Cu g1 ﬁltered water]) are thought to be 105 to 106 in seawater, if the colloid phase is considered (Shafer et al., 2004). The oceanic residence time (t ¼ annual removal rate/mass in ocean water, or annual addition rate/ mass in ocean water) for copper is about 2–5 104 years, but the residence time for the surface layer and euphotic zone of the open ocean is probably shorter, perhaps on the order of 30 years (Goldberg, 1963; Brewer and Spenser, 1975; Broecker and Peng, 1982; Li, 2000). In ocean margin sediments, most river-supplied Cu is probably buried in the aluminosilicate phases (Gibbs, 1973; Mantoura et al., 1991). According to water column speciation and sediment trap studies, most delivery of Cu to marine sediments from the pelagic water column may be with organic matter (Kumar et al., 1996). As most organic matter is decomposed in deep water and the surface sediments, the liberated pore water copper may be sequestered by sulﬁde in anaerobic sediments (Di Toro et al., 1990), or with clay minerals and the Fe and Mn oxide phases in oxidized surﬁcial sediments (Strawn et al., 2004; Heidmann et al., 2005). The mobility of Cu during sedimentary diagenesis and transport from sediments to the water column is

Sediment and Copper in the Gulf of Papua, Papua New Guinea

207

greatly facilitated by natural organic ligands, which are present in huge excess over Cu concentrations (Seritti et al., 1986; Moffett and Brand, 1996; Moffett et al., 1997; Skrabal et al., 1997; Kozelka and Bruland, 1998). In some coastal seas, the annual diffusive ﬂux of ligand–Cu from margin sediments is greater than the annual supply from rivers and the atmosphere (Williams et al., 1998; Audry et al., 2004). Copper is an essential element for biological growth, and is an important component of enzyme systems for nitrogen ﬁxation, denitriﬁcation, metabolic oxygen reduction, methane oxidation, and blood oxygen transport in many organisms (Herskovits and Hamilton, 1991; Kim et al., 2004; Prigge et al., 2004; Tocheva et al., 2004; Lieberman and Rosenzweig, 2005). Copper, and other essential trace elements (Fe, Mo, Zn, Mn, etc.) probably limit plankton growth in some parts of the ocean, away from sources of river and dust inputs (Springer-Young et al., 1992). As in seawater, the concentrations of free ionic Cu in living cells are probably o1018 M, and all normal Cu metabolism is facilitated by metallothionein proteins, carbohydrate ligands, and enzymes (Lippard, 1999; Rae et al., 1999). In the ocean, Cu concentrations in offshore surface water are probably regulated by phytoplankton uptake and release, with an expected Cu/OC ¼ 3–5 mmol Cu/mol OC (Sunda and Huntsman, 1995; Luoma et al., 1998). When copper supply to estuaries is enhanced from technological activities, biological toxicity is sometimes reduced by physiological reactions in cells that sequester Cu in excretory organs or vesicles (Sarasquete et al., 1992; Canli et al., 1997; Meyer et al., 1999), or by trapping of Cu in microbial mucilaginous exudations (Seritti et al., 1986; Moffett et al., 1997; Dryden et al., 2004; Rivera-Duarte et al., 2005). In spite of these cellular protection mechanisms, CuSO4 (bluestone) is an excellent biocide for microbes, crustacea, and mollusks when applied in high concentrations to small bodies of water (van Hullebusch et al., 2003; Garcia-Villada et al., 2004). High concentrations of copper in sewage (0.2–0.5 mg L1, Islam and Tanaka, 2004), mine wastes, and metal industries have been shown to be deleterious to many estuarine and marine, benthic, and planktonic invertebrates and ﬁshes (Jonas, 1989; Sharp and Stearns, 1997; Negri and Heyward, 2000; Agell et al., 2004; Chelazzi et al., 2004; Geracitano et al., 2004; Kim and Kang, 2004; Lehman et al., 2004; Rivera-Duarte et al., 2005; Virkutyte et al., 2005). Over the last century in North America, approximately 1.6 1014 g of Cu has been extracted from the lithosphere for human technological use, and over half of this mined Cu was disposed to waste in landﬁlls, tailings, and slag reservoirs (Spatari et al., 2005). These authors show that since 1940, there has been an order of magnitude increase in the proportion of industrial copper that is lost to postconsumer waste.

208

G. J. Brunskill et al.

The geography of New Guinea (Fig. 6.1) and its coastal oceanography have been reviewed by Brunskill (2004). The catchment area of rivers draining into the Gulf of Papua is 142 109 m2, and these combined rivers deliver about 330–400 109 m3 of freshwater to the Gulf, containing 150–200 1012 g yr1 of suspended sediments. The catchments of these rivers are largely undisturbed, except for two large mines, Ok Tedi Mining Ltd (OTML) in the headwaters of the Fly River, and Porgera (Placer Dome Asia Paciﬁc) in the headwaters of the Strickland River. Davies (2004), Harris et al. (2004), and Hemer et al. (2004) describe the distribution of nutrient elements, sediment, and sediment deposition climate for the Fly River delta. Aller and Aller (2004) and Aller et al. (2004) describe some of the biogeochemical aspects of the inner-shelf mud deposits, and Walsh et al. (2004) describe the large clinoform delta front that forms from the combined inputs of the rivers of the Gulf of Papua. Baker et al. (1990), Baker and Harris (1991), and Alongi et al. (1991) give data on the Cu concentration in biota and surface sediments of the Fly River delta. Apte and Day (1998) show the distribution of dCu (and other trace elements) in the water of the Gulf of Papua and Torres Strait. Wolanski and Gibbs (1995), Wolanski et al. (1995a, b, 1997, 1998) describe estuarine dynamics, sediment ﬂuxes, and the circulation of the inner-shelf water, and Pickard et al. (1977) and Burrage (1993) describe the larger-scale Hiri Current Gyre in the northern Coral Sea (Fig. 6.1). Our objectives in this paper are to present the annual rate of supply of Cu to the Gulf of Papua from rivers, and the removal rate of Cu by burial in sediments. The excess of river Cu input over observed Cu burial rates will be delivered to the Hiri Gyre of the Coral Sea. We speculate that this system is not in steady state, as there has been an increase in dissolved and particulate Cu supply rate from the Fly River, because of mining operations in the headwaters.

6.2. Methods Field and laboratory methods for this work are given in detail in Brunskill et al. (2003), and are brieﬂy summarized here with emphasis on methods of measurement of copper. 6.2.1. Field Sampling End member fresh river water and estuarine salinity gradient samples were collected from the ship in acid-cleaned plastic buckets or Niskin bottles. To collect suspended sediment, immediately after sampling, the water sample

130°E

140°E

N

D

O

N

E

S e p ik

Ja y a

S

I

A

Tabubil Kiunga

S

kl tric R.

a

ko

ri

0m

20

Sea

Torres

10°S

ur

N E W

ari

Gulf of Papua

Solomon

A

U

S

T

R

A

L

I

Sea

Port Moresby

Strait

Coral

Darwin

G U I N E A

Lae

Kerema

Daru

Arafura

Madang

Porgera

Ki

ly

Merauke

Rabaul

Sea

P A P U A

F

Di

Bismarck

R.

P

g ul

I

o

A

200

m

Sea

Figure 6.1: A map of the island of New Guinea, showing the main rivers (the Fly, Kikori, and Purari) that ﬂow to the Gulf of Papua.

Sediment and Copper in the Gulf of Papua, Papua New Guinea

Irian

Jayapura

ra m

nd

am M

be

150°E

R.

0°

209

210

G. J. Brunskill et al.

was pulled by vacuum through preweighed ﬁlters in the ship laboratory. Suspended sediment samples for ICP-AES analyses were collected on cellulose acetate 0.45 mm pore diameter ﬁlters stored in glass scintillation vials, and samples for organic carbon (OC) were collected on precombusted glass ﬁber GF/F ﬁlters and stored in aluminum foil envelopes. All suspended sediment samples on ﬁlters were kept frozen until they were oven dried and weighed for analyses. Water for dissolved element analyses was also subsampled from the above water samples. Water samples for dissolved element analyses were ﬁltered through 0.45 mm Minisart disposable ﬁlters, and for dissolved organic carbon (DOC) analyses were ﬁltered through polycarbonate membrane ﬁlters. These samples were acidiﬁed with concentrated HCl after sampling. Larger river suspended sediment samples were extracted from 60 to 1,000 L plastic barrels ﬁlled with zero salinity river water. The barrels were allowed to settle for several days to weeks, and the supernatant water was siphoned to allow recovery of the settled sediment slurry. Sediment trap cylinders were constructed of stainless steel 2 m high and 25 cm in diameter (height/diameter ¼ 8) with no top aperture bafﬂes. There were six replicate cylinders in each aluminum trap frame. The trap frame was ﬁtted with a 2 3 m tailﬁn of plywood, and the frame was suspended on a gimballed galvanized steel shaft that was attached to the plastic coated mooring wire above and below the trap frame. Current meters (Aanderaa RCM5) were positioned 10 m below each sediment trap frame on the mooring wire. Sediment traps were positioned at water depths of 300 m below the surface and 100 m above the bottom at locations identiﬁed by the ship’s Global Positioning System (GPS) (Fig. 6.2). The trap mooring was anchored to a 2 tonne concrete weight by means of chain and an acoustic release device. The teﬂon bottles attached to the bottom of each cylinder cone were ﬁlled with salinity 60 water with added HgCl2 for preservative. The bottles were immediately removed from the cylinders upon recovery, and the water was vacuum ﬁltered through GF/A ﬁlters, 1.6 mm nominal pore size, to obtain the particulate matter. Grab samples and sediment cores were collected from sites located by GPS estimations (Figs. 6.2, 6.5, and 6.6). We used a modiﬁed van Veen grab in shallow locations, or a weighted Smith–MacIntyre grab at deeper locations, both having a top opening port for sampling surface (1–3 cm) sediment, on transects from the intertidal zone to the continental slope. Sediment cores from intertidal mangrove mud banks were taken with a custom built 150 mm i.d. plastic barrel push corer attached to a block and tackle tripod hook. A small gravity corer (steel 150 mm i.d. tubes, 0.5–2 m long) with 300 kg

Sediment and Copper in the Gulf of Papua, Papua New Guinea

211

Figure 6.2: A bathymetric map of Gulf of Papua, showing the locations of sediment cores and sediment traps. The cores displayed in Figs. 6.8–6.11 are numbered on this map. Data for all other cores are given in Table 6.4. driving weight, a top closing lid upon retrieval, and an orange-peel type core catcher was used from the R/V Harry Messel. Sediment slices from both the mangrove corer and the small gravity corer were obtained by piston extrusion from the core bottom upward. Open water and deep water Kasten cores (Kuehl et al., 1985) were taken from the R/V Lady Basten and the R/V Franklin, using over 500 kg driving weight. The core tubes were 150 mm diameter square steel (2–4 m long, opening on one side for subsampling) with core top closure upon retrieval and a bottom core catcher. Sediment was subsampled for chemical analyses from grab samples and core slices, using cleaned plastic or stainless steel tools, and stored in plastic vials frozen on board the ship. Porites sp. coral bommies near Bramble Cay were cored in triplicate in 1994–1996, using a hand-held drill corer powered by a hydraulic motor. Cores

212

G. J. Brunskill et al.

were split sawed at AIMS, X-rayed, and studied under ultraviolet light to see the annual growth bands. Subsamples of coral aragonite from individual annual growth bands were obtained using a dremel tool with a steel bit. 6.2.2. Chemical and Radiochemical Analyses Measurements of concentrations of Cu, Al, and S were determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES, Varian Liberty 220). For sediments they were analyzed after nitric and perchloric acid digestion of the dried and ground sediment sample (Loring and Rantala, 1992; Thompson and Walsh, 1993). Digestion was carried out in Pyrex tubes in a programmable Block Digester at 1201C for 3 h followed by 1801C for 3 h. Cu was measured from the emission at 324.754 nm. These ICP measurements represent strong acid extracts of the bulk sediment, and do not include elements contained in acid-resistant minerals such as quartz and the heavy minerals. Filters with suspended sediment from river water, estuarine salinity gradient water samples, and sediment trap samples were weighed dry, digested, and analyzed for metals in the same manner. Filtered water samples were analyzed directly by ICP-AES to measure dissolved elements. Aragonite powder from coral growth bands was dissolved in nitric acid and analyzed by ICP-AES for Ca, Fe, Sr, Ba, Pb, and Cu. Cu analytical precision was 5% or better for all element measurements on water and sediment. Digestion and recovery of Cu in certiﬁed reference materials 1646 Estuarine Sediment (NIST), Standard Reference Sediment MESS 2 from the National Research Council of Canada, and BCJS-1 Marine Sediment (NRCC) were within certiﬁed limits or within 95% conﬁdence limits of several years of past data for that digest. No samples were below our detection limit for Cu in particulate material, and determinations on our reference samples suggest that we are measuring total Cu by this procedure. Measurements of organic carbon (OC) in sediments analyzed prior to 1995 were performed using a Beckman Carbon Analyzer with an infrared carbon dioxide detector, after acidiﬁcation of the sediment slurry to digest inorganic carbon. Later OC analyses used a Shimadzu TOC Analyzer with solid sampler, after HCl extraction. Particulate OC, collected from water samples, was also analyzed using the Shimadzu TOC Analyzer after acidiﬁcation of the suspended sediment on the glass ﬁber (GF/F) ﬁlters. DOC was determined by high-temperature catalytic oxidation on a Shimadzu TOC5000 Analyzer (Hedges et al., 1993). Gamma spectrometric measurements of 210Pb, 226Ra, 137Cs, and other isotopes were made on 50–150 g of dried and ground bulk sediment core

Sediment and Copper in the Gulf of Papua, Papua New Guinea

213

slices packed (with a 10 tonne hydraulic press) into a custom designed gas-tight Perspex container. The energy spectra of the gamma spectrometers were calibrated with Amersham and CANMET standards, of known low-activity spikes of suitable nuclides, in cleaned silica and carbonate sand of geometry and mass similar to the sediment samples. IAEA marine sediment reference material IAEA-315 was used to check the calibrations. Counting errors of the sample measurements were less than 10%, except for some very low-activity 137Cs samples which had errors of 30%. Interpretations of radiochemical tracer sedimentation history were done with several submodels described by Robbins (1978, 1986). These utilize a sediment mixed layer thickness, a decadal-century scale average input of excess 210Pb, Brisbane measurements of thermonuclear bomb fallout 90Sr (for 137Cs) over 1950–1990 (McCallan et al., 1980; Torgersen et al., 1983; B400 Bq 137Cs m2), and diffusion coefﬁcients for 210Pb and 137Cs in marine sediments (Li and Gregory, 1974). Estimates of the annual atmospheric ﬂux of 210Pb and the 137Cs inventory have been obtained from soil proﬁles in north Queensland (approximately 50 Bq 210Pb m2 yr1, 300 Bq 137 Cs m2; Pﬁtzner et al., 2004). Mass accumulation rate (MAR) for each core proﬁle is derived from the dry mass (g cm2) accumulated over several half-lives (22.26 years) of 210Pb in excess of its grandparent 226Ra, when there is a good log-linear relationship between accumulated sediment mass and excess 210Pb, in units of kg m2 yr1. In some intertidal and o20 m water depth cores, excess 210Pb and bomb fallout 137Cs are uniformly or irregularly distributed down the core, and sometimes we do not capture the entire inventory of these radiotracers of sedimentation. This may be the result of sediment mixing in situ, resuspension and mixing in the nepheloid layer, and subsequent redeposition or massive short-term (weekly, monthly, yearly) transient accumulation of uniformly labeled recent sediment (RS). For these later cores, we estimate ‘‘transient mass accumulation’’ (TMA) in units of the mass accumulated in four half lives of excess 210Pb (g cm2). This measure is intended to describe sediments of varying grain size and composition that are being resuspended, deposited, and moved as ﬂuid mud toward sites of more permanent (century to millennial scale) sediment accumulation. Areas of the Gulf of Papua river catchment and the bathymetric contours for the adjacent continental shelf were determined from digitized map reading software. Annual and mean river discharge of water were provided by Ok Tedi Mining Ltd. (personal communication), and the Bureau of Water Resources of the PNG Department of Environment and Conservation (Mr. Narua T. Lovai, Boroko, NCD, PNG).

214

G. J. Brunskill et al.

6.3. Results We will present our results in the order useful for our mass balance of Cu: annual river supply rate, removal rate to sediments, removal rate to sediment traps, and the remainder being available to the Hiri Current of the Coral Sea. 6.3.1. River Supply Rate Freshwater data from 1993 to 2003 samples of the Fly, Kikori, and Purari Rivers (Fig. 6.2) for salinity, suspended sediment, dissolved and pCu, dissolved and particulate organic matter, and pAl and sulfur are presented in Table 6.1. The dCu measurements are from Simon Apte’s work (ref, this book), and dCu concentrations in the upper Fly River (above the conﬂuence of the Strickland and Fly Rivers) are in the range 50–150 nM. The Strickland River contributes a large amount of water to the lower Fly River (catchment area ¼ 36.7 109 m2), with dCu in the range 7–16 nM (Apte, personal communication 2004). pCu in these river water samples is one to two orders of magnitude higher (1–4 mmol g1 of suspended sediment, or 0.6– 1 mmol L1) in concentration, compared to dCu. DOC varies little in the Fly River samples (0.15–0.28 mmol L1), but DOC in the Kikori and Purari Rivers appears to be somewhat lower in concentration (0.06–0.20 mmol L1). Particulate OC varies in concentration from 0.5 to 1.5 mmol g1 (or 0.08 to 0.25 mmol L1) amongst the three river freshwaters. Large particulate organic matter, such as living and dead terrestrial and mangrove trees, and rafts of ﬂoodplain and mangrove vegetation are commonly seen in the river, but are difﬁcult to sample quantitatively (see Robertson et al., 1993). pAl varies little (2.3–4.4 mmol g1) in the suspended sediment of these rivers, whereas particulate sulfur (pS) varies by a factor of 7 (10–71 mmol g1) in suspended sediment samples. The wet tropical catchment areas of the Gulf of Papua rivers have average erosion rates of 1.0–1.2 kg m2 yr1, a factor of 5–10 higher than the other rivers listed in Table 6.2. High erosion rates are expected in catchments that have high relief (4.5 km) and high rainfall (4–10 m yr1). The rate of release of dCu from the Strickland, Kikori, and Purari River catchment soils and bedrock exposures is 21–36 mmol dCu m2 yr1, whereas the upper Fly River gives up much more dCu to the sea (243 mmol m2 yr1). Particulate Cu ﬂuxes from the catchments of the Kikori and Purari Rivers were 920–1,940 mmol m2 yr1, whereas the upper Fly River annual transport of pCu was 2,460 mmol m2 yr1. Higher river catchment yields of dCu and pCu for the upper Fly River are due to mine tailings disposal into the headwater

Table 6.1: River zero salinity chemistry for the Gulf of Papua. River Fly

SS Sample Type Barrel SS Filter GFC

Barrel SS Filter c.a. Filter c.a. Barrel SS Filter c.a. Filter c.a. Barrel SS Filter c.a. Mean

Sep ‘97 Sep ‘97 Jan ‘99

0.1 0.1

dCu (nM)

271 62a 48a

0.1 0.7 0 0 0.8

217 243 250 141 164a

0.13 0.13 0.9 0 0 0.1 0 0 0.1 0.1 0.1

353 170 296 165 286 239 44 46 45

91

pCu (mmol g1)

DOC (mM)

1.07 1.52

0.276

2.8a 2.96 1.07 1.78 2.47 1.77 4.0 1.87 2.19 1.56 2.30 2.64 2.33 2.41 3.10 2.2 0.93 2.65 1.62 1.7

POC (mmol g1)

pAl pS (mmol g1) (mmol g1)

0.60

2.45

15

0.52

2.93 2.83 2.33 3.17 3.48

25 10 17 20 13

0.67

2.81

0.94

2.38 2.81 3.00 2.30 2.85 2.8

18 18 14 31 32 19 49 37 23

2.64 2.80 4.41 3.3

39 44 27 37

0.281 0.243 0.257 0.256

0.155 0.152 0.220b 0.215 0.216 0.23

0.69 0.87 0.72 1.76

0.059b 0.100 0.08

1.8

215

Kikori Barrel SS Filter c.a. Filter c.a. Mean

Jan ‘93 Jan ‘93 Feb ‘93 Jul ‘93 Feb ‘94 Feb ‘94 Feb ‘94 Feb ‘94 Feb ‘94 Mar ‘94 Mar ‘95 Mar ‘95 Mar ‘95 Aug ‘97 Aug ‘97 Jan ‘99 Feb ‘03 Feb ‘03

SS (mg L1)

Sediment and Copper in the Gulf of Papua, Papua New Guinea

Filter c.a. Filter c.a. Barrel SS Filter c.a. Filter c.a.

Date Salinity

216

Table 6.1: (Continued ). SS Sample Type

Date Salinity

Purari Filter c.a.

Feb ‘95

0.1

Wame Barrel SS Filter c.a. Filter c.a. Barrel SS Filter c.a. Mean

Sep ‘97 Sep ‘97 Jan ‘99 Feb ‘03 Feb ‘03

0.4 0.4 0.1 0.1 0.1

SS (mg L1) 191 30 40 187 112

dCu (nM)

pCu (mmol g1)

DOC (mM)

0.80

0.103

0.80 0.71 0.83 0.94 1.04 0.85

POC (mmol g1)

44 1.47

0.095b 0.111 0.200 0.13

pAl pS (mmol g1) (mmol g1)

0.75 0.94 1.1

2.40 2.75 3.59 2.84 3.71 3.1

71 64 30 28 15 42

Note: The Fly River samples were collected upstream of Sumogi Island, well below the conﬂuence of the upper Fly River and the Strickland River. SS, surface water suspended sediment; dCu, copper that passes through a ﬁlter; pCu, particulate copper retained on a ﬁlter; DOC, dissolved organic carbon; POC, particulate organic carbon; pAl, particulate aluminum; and pS, particulate sulfur; GFC, Whatman glass ﬁber coarse ﬁlter; c.a., cellulose acetate 0.45 mm pore diameter ﬁlter. a Data from Simon Apte, Centre for Advanced Analytical Chemistry, Sydney, NSW, Australia ([email protected]) Chapter 9 (319–371) of this book. b Peter Davies data, See Davies (2004).

G. J. Brunskill et al.

River

Sediment and Copper in the Gulf of Papua, Papua New Guinea

217

Ok Tedi River. The mountainous and tropical Amazon River transports a similar amount of dCu from its catchment (per unit area, and compared to the Strickland, Kikori, and Purari Rivers), but temperate zone (and technologically developed) European river catchments yield much lower dCu and pCu transport to the sea (Table 6.2). We estimate the average annual river input of suspended sediment, dCu and pCu to the Gulf of Papua in Table 6.3. Approximately 275 106 mol of dissolved and pCu are added to the Gulf of Papua annually, and about 70% of this is from the Fly River. This computation is based on the freshwater concentrations of Cu multiplied by estimates of average annual river ﬂow. The estimate for dCu (10 106 mol yr1) is probably too high, as half of this dCu appears to be removed from solution and adsorbed to particulate phases in the lower salinity gradient. We have no dCu data for the Kikori and Purari Rivers, so we have used Apte’s data for the Strickand River (B12 nM) as an estimate of this input.

6.3.2. The Salinity Gradient Simon Apte’s data for dCu along the salinity gradient show that over half of the dCu is removed from the solution phase to the particulate phase at salinities of 0–3 (Apte, 2009). After this initial removal of dCu, there appears to be conservative mixing of high concentrations of dCu (50–150 nM) in the Fly River with low concentrations of dCu from the Strickland River (12 nM) and in the Gulf of Papua (1 nM). DOC concentrations decline in an irregular manner along the salinity gradient (Fig. 6.3), from 0.2 to 0.3 mmol L1 in low-salinity water, to o0.1 mmol L1 in outer shelf waters. Our DOC salinity gradient proﬁle for August 1997, and Robertson et al. (1993) proﬁle for 1991, show an increase in DOC at salinity 15–20. We have no dissolved Al data, but sulfate mixes conservatively from low concentrations in the rivers (B1 mM) to high concentrations (28 mM) at salinity 35 in Coral Sea water. Solution pH varies from 7 to 8.2 along the salinity gradient. Our pCu data for the salinity gradient (Fig. 6.3) contain data for suspended sediment from the Fly, Kikori, and Purari rivers, their deltas, and offshore waters. River water varied from 1 to 3 mmol pCu g1, and most samples from the salinity gradient were close to 1 mmol pCu g1. A few samples of suspended sediment from February 1994 (near Bramble Cay), February 2003, and November 2003 (from offshore of the Kikori River Delta) had higher pCu concentrations of 3–6 mmol g1 at salinity 20–35. We have no reason to believe that these samples were contaminated by us, as the

Upper Fly Strickland Fly and Strickland Purari Kikori Amazon Zaire (Congo)

Chanjiang (Yangtze) Huanghe (Yellow) Zhujiang (Pearl) Mekong GangesBrahmaputra Netravati and Gurpur Mississippi Loire Morlaix Scheldt Danube

SS (g m2 yr1)

pAl (mol m2 yr1)

1,118 1,200

dCu pCu (mmol m2 yr1) (mmol m2 yr1) 243 28

1,100 1,143 192

3.1 3.4 21 0.84

11

0.03

36 21 30

2,460 920 1,940 806

1.6

18

256

205

1,460 249 200 941

614 192 340 442

0.83 2.7

331 105 15 23 102

250 0.40

0.5

80 17 4 8 8.4

Note: The average annual load of river sediment, Al, and Cu was divided by the river catchment area.

12 78 184

Source This study, Apte (2009). Apte (2009) This study This study This study Boyle et al. (1982); Martin and Meybeck (1979) Eisma and van Bennekom (1978); Martin et al. (1978); Meybeck (1979); Sholkovitz et al. (1979) Zhang et al. (1992); Zhang (1999) As above As above Martin and Meybeck (1979) As above Shankar and Manjunatha (1994) Martin and Meybeck (1979) Waeles et al. (2004a, b) Monbet (2004) Zwolsman and van Eck (1999) Guieu et al. (1998)

G. J. Brunskill et al.

River

218

Table 6.2: The annual rate of river transport of suspended sediment, particulate Al, dissolved and particulate Cu from river catchment areas.

Table 6.3: The annual river supply rate of water, surface suspended sediment, and copper to the Gulf of Papua. River

Qa SSa (109 m3 yr1) (1012 g yr1)

dCu (nM)

39.3

81723

34

91(1)

36.7 76.0

112721 193

51 85

12(1)

33.7

97

37

17.5 15 142

30 12 332

20 14 156

pCu S dCu S pCu S Cu (mmol g1) (106 mol yr1) (106 mol yr1) (106 mol yr1) 7.4

2.2

1.3 8.7

187

196

[12]

0.85

1.2

31

33

[12] [12]

1.7 [0.85]

0.36 0.14 10.4

34 12 264

34 12 275

Note: Dissolved Cu input was calculated as average dCu concentration times mean annual river ﬂow rate (Qa). Particulate copper was calculated as measured average pCu times mean annual suspended sediment discharge rate (SSa). Fly and Strickland River water ﬂow data are from OTML. Dissolved Cu data(1) is from Simon Apte. Numbers in square brackets are estimates based on Apte’s Strickland River data. SdCu=Qa dCu for an average year. SpCu ¼ SSa pCu for an average year. SCu=SdCuþSpCu for an average year. Ad is the land drainage area for each river.

Sediment and Copper in the Gulf of Papua, Papua New Guinea

Fly River at Obo, upstream of Strickland River Strickland River Total Fly River input to delta Purari (Wame Channel) River Kikori Other Total for Gulf of Papua

Ad (109 m2)

219

220

G. J. Brunskill et al.

Figure 6.3: The variation of dissolved and particulate organic carbon (OC), particulate Cu and Al, and suspended sediment concentration along the salinity gradient from fresh river water to offshore Gulf of Papua seawater, over 1993 to 2003. same sampling and analytical methods were used on samples from lower salinity waters. Figure 6.3 also gives a view of the variation of suspended sediment, pAl, and POC along the salinity gradient. A maximum in suspended sediment concentration is usually seen between salinity 5 and 15 in the Fly River delta. The composition of the suspended sediment does not change much along the salinity gradient, with most values between 2.5 and 3.5 mmol Al g1, 1–2 mmol POC g1, and 10–30 mmol S g1. A few samples from the mid shelf

Sediment and Copper in the Gulf of Papua, Papua New Guinea

221

Figure 6.4: The variation of the ratios of particulate Cu/Al, Cu/OC, Cu/S along the salinity gradient from fresh river water to offshore Gulf of Papua seawater. south of the Kikori Delta, and near Bramble Cay, are greatly depleted in pAl and enriched in pS and POC, and these samples usually had high pCu concentrations (Fig. 6.4). Per unit volume, pAl, and POC were found in high concentrations, and declined in an irregular manner to low concentrations offshore at salinity 30–35. The ratios pCu/pAl, pCu/POC, and pCu/pS along the salinity gradient are shown in Fig. 6.4. The ratio of Cu/Al in suspended sediment along the salinity gradient changes little (3–8 104) from salinity 0 to 25, but Cu is enhanced relative to aluminum in the low suspended sediment concentrations of the outer Gulf of Papua at salinities greater than 30. Particulate Cu declines in proportion to POC and pS along the salinity gradient (Fig. 6.4).

6.3.3. The Distribution of Cu in Surface Sediments Copper concentrations in surface sediment grab samples (0–3 cm) are higher (600–1,000 nmol g1) in the aluminosilicate clay-rich deltas and inner

222

G. J. Brunskill et al.

shelf, and much lower in the hard bottom carbonate sediments of the outer shelf (Fig. 6.5). Some of the deep water pelagic carbonate sediments of the continental slope and northern Coral Sea have higher Cu concentrations (150–700 nmol g1), especially in the eastern Gulf offshore from Port Moresby. This distribution of Cu appears to be related to the relative abundance of the aluminosilicate clay and coarse carbonate fractions, with ﬁne-grained sediments that have high Al concentrations usually having high Cu concentrations (Figs. 6.6 and 6.7). The sediment molar ratio Cu/Al varies 1–5 104, with enhanced Cu concentrations being found on the Fly River delta front and seaward near the 100 m depth contour, and along the central and eastern inner shelf on the top and foreset beds of the clinoform (Fig. 6.6). Sediments with low Cu/Al were usually black sand-rich (SE of the Bamu and Turama Inlets) or hard-bottom carbonate-rich sediments of the outer shelf from 80 to 150 m water depth (Fig. 6.6). The relationship between Cu and Al in Fig. 6.7 indicates a family of outliers which probably represent ‘‘hotspots’’ of undiluted recent Fly River sediment in zones of transient sediment deposition, which have much higher concentrations of Cu (1,000–4,000 nmol g1, Table 6.1). Large variation in concentrations of sedimentary organic matter, sulfur (Fig. 6.7), manganese, and iron (not shown) do not give useful predictions of Cu concentrations.

Figure 6.5: The variation of Cu concentration in surface sediment grab samples from the Gulf of Papua. The cross (þ) symbols indicate locations of surface sediment grab samples that were analyzed for Cu in this study.

Figure 6.6: The variation of the ratio Cu/Al in surface sediment grab samples from the Gulf of Papua. The cross (þ) symbols indicate locations of surface sediment grab samples that were analyzed for Cu in this study.

Figure 6.7: The relationships between copper, aluminum, OC, and sulfur in surface sediment grab samples from the Gulf of Papua. The locations of these grab samples are shown by the (þ) symbols in Figs. 6.5 and 6.6, and we have sorted the data points in these graphs by the depth zones 0–20 m, 20–100 m, and W100 m.

224

G. J. Brunskill et al.

6.3.4. Sediment Core Chemistry and Accumulation Rates of Copper in Sediments Table 6.4 and Figs. 6.8–6.11 show the variation in Cu and likely host elements with sediment core depth, and core proﬁles of radiochemical tracers that we use to estimate sediment accumulation rate. We present these data according to the depth zones indicated in Table 6.4: the intertidal cores, subtidal cores to 10 m depth, cores from 10–100 m depth of the major sedimentary clinoform, and cores from the continental slope and northern Coral Sea at water depths of 200–2,000 m. The depth zone from 100 to 200 m is largely hard-ground carbonate sediment, from which we have not recovered any cores, and which we consider to have zero or very low sediment and copper accumulation rates on our recent century time scale.

6.3.5. Intertidal Zone, 0–3 m Tidal range in the Gulf of Papua is 3–4 m, and tidal energy moves sediment seaward, landward, and along-shore. Most medium to large tidal channels are nondepositional zones with hard sandy bottoms, and there are many sandbars and islands that appear to be composed of sand and gravel. Fine-grained sediments accumulate on the low-current mangrove aprons opposite the channel cut bank, and sometimes at the junctions of smaller channels with larger channels. We have observed the transient nature of some of the mangrove mud banks, some of them accumulating and then vanishing over several years of observation. With very few exceptions in the intertidal zone, excess 210Pb and bomb fallout 137Cs proﬁles in cores of mud and sandy mud suggest that sequential packages of sediment are deposited episodically and are sometimes mixed and homogenized during frequent resuspension and redeposition events. As an example for the intertidal zone, cores 205 (in the Fly River delta, see Fig. 6.2) and 3618 (in the Kikori River delta, Ai Ai River) do not show the expected exponential decline in excess 210 Pb or the known history of 137Cs deposition (Fig. 6.8), and thus cannot easily be used to estimate sediment accumulation rate. They do indicate that a large thickness (1–1.5 m) of sediment is labeled with our short-lived radiotracers, and this is roughly quantiﬁed by our RS parameter in Table 6.4. We consider RS (in g cm2) to be equivalent to an inventory of sediment that is transient on a time scale less than four half-lives of 210Pb (89 years). In both of these intertidal cores, the RS inventory is incomplete,

Table 6.4: Gulf of Papua sedimentation rates, sediment Cu concentrations, and Cu burial rate. Depth interval Area Core m (109 m2) Number 0–10

20–30

2.93

2.18

MAR Ex210PbI 2 1 (kg m yr ) (Bq m2) 5,208 704 4,325 5,972 400 1,595 W26,580 W21,410 8,504 12,382 4,684 12,879

137

RS (g cm2)

CsI Cu (nmol (Bq m2) g1)

75 33 W66 99 5.1 24 W85 W100 50 50 40 105

500 134 244 406 53 195 W1,032 W960 384 536 177 446

787 787 820 800 700 658 800 700 720 630 1,100 1,420

Cu Cu (mmol m2 (proﬁle) yr1)

IT IT IT IT 8 8 IT 10 6 9 5 8

EoM EoM EoM EoM EoM EoM EoM EoM 6.6 EoM EoM EoM 6.9 Transient

223 224 1,206 2,067 3,163 3,615 3,639 Mean

15 15 16 13 20 15 14

3.6 M, 1.4 M, 3 M, 0.6 M, 2.5 M, 0.7 EoM 1.7

5,322 2,288 5,706 3,566 8,789 2,667 5,247

35 16 50 14 48 10 24

384 250 667 159 441 60 125

580 500 628 680 700 650 600

O O O

2,048 3,161 3,154 3,616 Mean

24 27 27 24

28.5 M, 0.3 26.7 6.5 18.5

15,618 1,039 10,710 5,048

W120 7 110 60

943 44 828 133

600 300 400 410

O O O

I O I S O O O I O

O O

6.63 11.7 W6.1 8.9 0.40 1.8 W7.6 W7.9 4.8 3.5 4.9 16 (6.8) Transient 2.0 0.70 1.9 0.41 1.8 0.46 0.81 1.1 17.1 0.090 10.7 2.7 7.6

225

203 204 205 206 217 218 3,618 3,634 3,638 3,636 3,666 3,668 Mean

Sediment and Copper in the Gulf of Papua, Papua New Guinea

10–20

12.57

Z (m)

226

Table 6.4: (Continued ).

30–50

50–100

3.97

14.3

Z (m)

MAR Ex210PbI 2 1 (kg m yr ) (Bq m2)

RS (g cm2)

137

CsI Cu (nmol (Bq m2) g1)

Cu Cu (mmol m2 (proﬁle) yr1)

208 210 215 216 230 231 2,046 3,123 3,128 3,139 3,150 3,617 3,619 3,623 Mean

48 48 42 42 45 45 40 42 46 42 48 52 37 49

8.0 9.6 10 13.5 3.5 4.8 2.6 2.3 3.5 EoM 22 9.8 9.5 25.6 9.6

14,093 15,762 12,500 13,713 11,803 16,677 3,717 13,160 9,336 45,526 41,025 20,214 16,280 50,767

70 65 82 86 38 46 22 40 32 W220 W150 65 85 W180

338 408 844 904 19 99 30 190 329 1,494 1,777 573 620 1,764

400 425 600 600 400 400 176 520 610 600 600 580 470 600

S S O S S I S S I O I S I

220 221 234 235 3,134 3,144 3,148 3,158 3,620 3,622 Mean

55 55 66 66 68 58 76 83 76 70

4.3 5.5 2.0 1.4 14.7 2.4 4.1 2.0 3.8 1.9 4.2

9,106 11,648 15,392 11,776 49,340 9,140 20,792 18,384 20,312 15,473

30 35 W20 18 180 55 35 30 40 28

198 233 61 38 1,501 171 137 69 134 23

381 390 420 480 600 400 400 425 390 200

O O S O O O, I O S I

3.2 4.1 6.0 8.1 1.4 1.9 0.46 1.2 2.1 0.59 13.2 5.7 4.5 15.0 4.8 1.6 2.1 0.84 0.67 8.8 0.92 1.6 0.85 1.5 0.38 1.9

G. J. Brunskill et al.

Depth interval Area Core m (109 m2) Number

100–200 200–1,000

4.7 5.7

1,000–2,000

9.75

923 923 473 473

232 233 240 241 1,611 1,616 1,632 1,630 Mean

1,618 1,618 1,117 1,117 1,509 1,245 1,472 2,045

0.25 0.25 0.40 0.50 0.35

13,676 19,398 3,132 6,464

W8 6 7 13

ND ND ND ND

550 450 150 180

O S S S

0.14 0.11 0.060 0.090 0.10

0.85 0.40 o0.1 0.5 o0.2 o0.3 o0.01 0.55 0.36

11,604 19,357 4,677 5,110 2,496 658 750 5,300

20 W22 8 W5 4 4 1.3 10

ND ND ND ND ND ND ND ND

700 625 440 420 600 400 500 700

S S I O I S D S

0.60 0.25 o0.044 0.21 o0.12 o0.12 o0.005 0.39 0.22

Note: Locations of sediment core sites are given in Fig. 6.2. Z, water depth at core site; MAR, sediment mass accumulation rate; Ex 210PbI, inventory of excess 210Pb down core in Bq m2; RS, Recent sediment labeled with excess 210Pb in g cm2; 137CsI, Inventory of 137Cs down the core in Bq m2; IT, intertidal zone; EoM, Episodic sedimentation or mixed core; M, thick 210Pb mixed layer; ND, not detected; Cu core proﬁle code: I, Cu increasing at core top; D, Cu decreasing toward top; S, no variation in Cu; O, Oscillating Cu concentrations down core. RS was converted to Transient Mass Accumulation (TMA) ¼ g m2 yr1 accumulated over 4 210Pb half lives (89 years). This TMA was multiplied by the Cu concentration of the sediment to obtain a transient Cu accumulation rate in sediments.

Sediment and Copper in the Gulf of Papua, Papua New Guinea

236 237 238 239 Mean

227

228

G. J. Brunskill et al.

Figure 6.8: Intertidal and deltaic (0–3 m water depth) sediment core proﬁles of Cu, Cu/Al, Cu/OC, Cu/S, excess 210Pb, and 137Cs for representative cores 205 and 3618 from the Gulf of Papua. See Table 6.4 for an interpretation of the depositional history of these cores. There is no exponential decline in excess 210Pb in most of the cores from this depth zone, and we interpret these locations to have massive episodic sediment deposition or deep mixing of sediment supplied in the last century. as our maximum depth of core samples had easily measurable excess 210Pb and 137Cs. We designate this type of sedimentation as ‘‘episodic or mixed’’ (EoM) in Table 6.4. For purposes of the Cu mass balance, we have divided the sediment mass inventory RS in Table 6.4 by 89 years, to obtain a semiquantitative transient mass accumulation (TMA) rate, but we do not give this TMA equal status to estimates of MAR obtained from cores that display exponential declines in excess 210Pb and a reasonable proﬁle of 137 Cs. In the intertidal cores in Table 6.4, RS varies from 33 to 99 g cm2, and the calculated TMA ranges from 3 to W11 kg m2 yr1. The inventories

Sediment and Copper in the Gulf of Papua, Papua New Guinea

229

Figure 6.9: Subtidal (3–10 m water depth) sediment core proﬁles of Cu, Cu/ Al, Cu/OC, Cu/S, excess 210Pb, and 137Cs for representative cores 3638 and 3668. See Table 6.4 for an interpretation of the depositional history of these cores. There is no exponential decline in excess 210Pb in most of the cores from this depth zone, and we interpret these locations to have massive episodic sediment deposition or deep mixing of sediment supplied in the last century. of excess 210Pb and 137Cs in the cores in Table 6.4 are related to RS and the thickness of the transient sediment mass. Concentrations of Cu in core 205 increase smoothly from 490 nmol g1 at the core bottom to 800 nmol g1 at the sediment surface, which suggests that the sediment is not wholly homogenized by mixing (Fig. 6.8). The ratios Cu/Al and Cu/OC also increase smoothly toward the sediment surface, indicating a 50% enrichment in Cu with respect to the aluminosilicate phase of the sediment in recent time. Core 205 was taken in mangrove sediments in

230

G. J. Brunskill et al.

Figure 6.10: Inner shelf (10–100 m) sediment core proﬁles of Cu, Cu/Al, Cu/ OC, Cu/S, excess 210Pb, and 137Cs for representative cores 3134, 3154, and 3639. See Table 6.4 for an interpretation of the depositional history of these cores. The arrow indicates the approximate core horizon of deposition of sediment in 1900.

Sediment and Copper in the Gulf of Papua, Papua New Guinea

231

Figure 6.11: Continental slope and northern Coral Sea sediment core proﬁles of Cu, Cu/Al, Cu/OC, Cu/S, excess 210Pb, and 137Cs for representative cores 241 and 1630. The arrow indicates the approximate core horizon of deposition of sediment in 1900. See Table 6.4 for an interpretation of the depositional history of these cores. the channel between Purutu and Aibinio Island, in the middle of the estuarine salinity gradient of the Fly River delta. Concentrations of Cu in core 3618 show no trend with time, and vary between 700 and 830 nmol g1 (Fig. 6.8). These variations in Cu concentration appear to be due to variations in the abundance of the aluminosilicate phase, as Cu/Al ratios are fairly constant down core at 2.5–3 104. This core was taken in a mangrove apron mud bank at the juncture of a tidal creek and a major tidal channel in the Kikori River delta. The other intertidal cores listed in Table 6.4 show similar proﬁles of Cu, Cu/Al, Cu/OC, and Cu/S.

232

G. J. Brunskill et al.

6.3.6. Subtidal Zone, 3–10 m Table 6.4 gives sedimentation rate and Cu data for seven cores from the subtidal zone of the Gulf of Papua. This region is generally difﬁcult to core with our equipment, as much of the area is black sand or wave-packed stiff muddy sand. Only one core (3638 in Table 6.4 and Fig. 6.9) from a small patch of mud south of Turama Inlet (see Fig. 6.2) gave us an accumulation rate from our radiotracers. Excess 210Pb MAR was 6.6 kg m2 yr1, and 137 Cs was distributed down to 50 cm in this core. The other cores for this zone listed in Table 6.4 are similar to core 3668 (Fig. 6.9) from south of Bamu Inlet (Fig. 6.2), where excess 210Pb and 137Cs are distributed down the core in an irregular way to 106 cm core depth. Core 3668 appears to consist of two packages of sediment labeled with excess 210Pb homogeneous sediment, the top 10 cm package of higher activity, and then a homogeneous package from 10 to 106 cm. Other cores from this zone often show 2–5 separate packages of radiotracer labeled sediment, and some have occasional laminations of coarse black sand and ﬁner muddy sand. As deﬁned in the previous section, we estimate RS for these cores to be 5 to W100 g cm2, with a TMA of 0.5 to W11 kg m2 yr1 (Table 6.4, Fig. 6.9). We think that our estimate of ﬁne-grained sediment and sedimentary Cu accumulation in this 3–10 m depth zone is likely to be over-estimated, because: (1) most of our coring attempts were successful in small patches of mud within a larger area that is hard packed sand and copper-poor, (2) we do not have coring equipment that works well in sand, (3) we have few cores from this largest area of the Gulf, the topset of the clinoform, and (4) the observed nonchronological deposition is difﬁcult to quantify. Our methods clearly do not quantify the rate of accumulation of sandy sediments. Concentrations of Cu, and the ratios Cu/Al and Cu/OC, show a small increase from the core bottom to the surface 40 cm in core 3638 (Fig. 6.9), with a dramatic increase in the Cu/S ratio. Core 3668 displays a somewhat oscillatory down core variation in Cu, Cu/Al, Cu/OC, and Cu/S, with a general enhancement in Cu concentration in the surface 40 cm, relative to its likely host phases Al, organic matter, and sulfur (Fig. 6.9). We estimate an average TMA for the 3–10 m water depth zone of 6.7 kg m2 yr1, where RS varies from 5 to W100 g cm2. Average pCu accumulation rate in this zone is a transient 6.7 mmol m2 yr1 (Table 6.4). Obviously, the uncertainty in extrapolation of these parameters in this depth zone is large.

Sediment and Copper in the Gulf of Papua, Papua New Guinea

233

6.3.7. Inner Shelf, 10–100 m In general, the quality of our sediment cores, and the application of our radiotracer methods to estimate MAR, were much better offshore, where mud accumulation occurs in a more chronological fashion well below wave base and the effects of tidal scouring. The surface of the cores usually showed a 1–3 cm rusty red oxidized iron layer, sometimes with surface sediment polychaete worm tubes intact. MAR varied from 0.3 to 28.5 kg m2 yr1 (Table 6.4, Fig. 6.10), and most of the cores had 210Pb surﬁcial mixed layers of 10–50 cm depth. We present three cores as examples of this depositional zone in Fig. 6.10. Core 3639 from 14 m water depth (Fig. 6.2) has a 25 cm thick mixed layer of excess 210Pb and 137Cs labeled sediment (RS ¼ 12 g cm2), below which is an older sediment with undetectable activities of our radiotracers. This may represent a ﬂuid mud layer resting temporarily on much older sediments. Concentrations of Cu, and ratios of Cu/Al and Cu/OC, vary little from the core bottom to the surface, but there is a large increase in Cu/S within the surface 25 cm layer. Core 3154 from 30 m water depth (Fig. 6.2) has a good exponential decline in excess 210Pb with core depth (or accumulated mass per unit area), which suggests a MAR of 26.7 kg m2 yr1 with little indication of surface sediment mixing in the core. 137Cs was detected to 130 cm core depth, and this core probably represents o30 years of deposition history. RS ¼ 110 g cm2, which also suggests a large thickness of RS that contains our radiotracer labels. Cu concentrations, and the ratios Cu/Al, Cu/OC, and Cu/S are nearly constant from the bottom of the core (at 260 cm) to 50 cm, but all these parameters increase and then decline slightly in the top 50 cm of the core (Fig. 6.10). Core 3134 from 67 m water depth at the bottomset toe of the clinoform, south of the Kikori Delta (Fig. 6.2), has a thin 210Pb mixed layer of 11 cm, and displays an exponential decline in excess 210Pb activities over four halflives and 150 cm of core depth (Fig. 6.10 and Table 6.4). This chronology and MAR of 14.7 kg m2 yr1 is conﬁrmed by the 137Cs proﬁle, which was found to be undetectable below 50 cm core depth. RS ¼ 180 g cm2, which indicates a very large inventory of RS of deposition age o100 years. The inventory of excess 210Pb and 137Cs is also very large, approximately 50 and 5 times greater than atmospheric supply rates respectively. These data suggest that this site receives ﬁne-grained sediment with our attached radiotracers from a much larger area, probably the 0–20 m water depth zone. Cu concentrations, Cu/Al and Cu/S vary little from the core bottom to about 50 cm, where these

234

G. J. Brunskill et al.

parameters increase slightly. Cu/OC varies little in this core, but declines in the top 50 cm. All of the other 33 cores listed in Table 6.4 show similar trends to our example cores. MAR varied from 0.3 to 28 kg m2 yr1, RS varied from 7 to W220 g cm2, and sedimentary Cu accumulation rate varied from 0.09 to 17 mmol m2 yr1. 6.3.8. Continental Slope and Northern Coral Sea, 200–2,000 m The average accumulation rate of sediment and Cu is one to two orders of magnitude smaller on the seaﬂoor of the slope and northern Coral Sea (Table 6.4). No 137Cs was detected in these deeper water cores, and we rely on the proﬁle of excess 210Pb alone to estimate MAR. In most cores, excess 210Pb was detected only in the top 6–10 cm, but an exponential decline was observed, suggesting MAR in the range o0.1 to 0.85 kg m2 yr1 (Table 6.4, Fig. 6.11). These sediments are a golden tan pelagic foraminifera and pteropod ooze with a variable and small aluminosilicate fraction. We give proﬁle data for two example cores in Fig. 6.11. Core 241 was taken at the base of the slope offshore directly south of the mouth of the Purari River at 1,117 m water depth (Fig. 6.2). The excess 210 Pb proﬁle is based upon the top 6 cm, where excess 210Pb activities are very high (100–300 Bq kg1), and the proﬁle suggests a MAR ¼ 0.5 kg m2 yr1. If this chronology is correct, then the changes in Cu concentration in this core shown in Fig. 6.11 occurred well before 1900. Cu concentration is relatively constant at 400 nmol g1 from the core bottom at 160 cm up to 20 cm core depth, where there is a rapid decline in Cu concentration to 300 nmol g1, followed by an increase back to 400 nmol g1. This near surface variation in Cu concentration was not caused by variations in the supply of aluminosilicate minerals, as the Cu/Al ratio changes over this near surface interval. The abundance of Cu relative to organic matter and sulfur increases from about 100 cm core depth to the surface (Fig. 6.11). Core 1630 was taken from 2,045 m water depth in the northern Coral Sea, SE of Eastern Fields (Fig. 6.2). A good exponential decline in excess 210Pb was observed, again with very high activities in the top core slices, suggesting a MAR of 0.55 kg m2 yr1 (Fig. 6.11). Cu concentrations vary from 220 to 800 nmol g1 over the 160 cm core length, with the top 10 cm having Cu concentrations of about 700 nmol g1. The ratios Cu/Al, Cu/OC, and Cu/S all show an oscillatory behavior, with Cu/S declining greatly in the top 6 cm slices of the core.

Sediment and Copper in the Gulf of Papua, Papua New Guinea

235

The other 10 cores listed in Table 6.4 have similar characteristics, with an average MAR of 0.3 kg m2 yr1 and a sedimentary Cu accumulation rate of 0.06 to 0.6 mmol m2 yr1. The inventory of sediment mass labeled with our radiotracer (RS) ranged from 1 to 22 g cm2, and the inventory of excess 210 Pb was usually greater than expected from atmospheric fallout for this region (B50 Bq m2 yr1).

6.3.9. Removal Rate of Copper to Sediment Traps The rate of removal of Cu from the pelagic water column at the base of the continental slope and northern Coral Sea was determined during TROPICS cruises in May–July 1997 (Fig. 6.2 and Table 6.5). This time period is during the SE Trade Winds season, when maximum wave energy is directed into the crescent of the Gulf of Papua (Fig. 6.1). We tried to estimate the Cu removal rate from the warm surface layer of the northern Coral Sea with traps set at 300 m, below the thermocline at 80–130 m. Deeper traps set on the mooring wire at 900–1,360 m (100 m above the seaﬂoor) probably collected the sediment rain from the euphotic zone as well as horizontal ﬂuxes of sediment from the continental shelf and slope. Total sediment ﬂux to the traps was 0.1 to 0.2 g m2 d1 (Table 6.5), but much higher trap ﬂuxes were observed from the deep Kerema Canyon trap site (0.6 g m2 d1) where shelf sediment is probably discharged into the Coral Sea. The sediment trap material is greatly enhanced in Cu concentration, relative to sources in the Gulf of Papua rivers and continental shelf (Table 6.5) and the sediment composition beneath the sediment trap mooring sites (cores 236, 237, 240, 241, 1630) in Table 6.4. In proportion to Al, OC, and S, Cu concentration in the sediment trap material is enhanced 5 to over 10 fold, compared to river, shelf, and slope sediments. If we extrapolate these sediment trap results to a full year, and multiply the composition by the annual mass ﬂux to the traps, we obtain the annual Cu removal rates shown in Table 6.6. These removal rates of Cu (0.1–0.5 mmol m2 yr1) for trap sites south of the Fly River and in the northern Coral Sea are about one tenth of Cu accumulation rates on the inner shelf, but are very similar to sedimentary Cu burial rates on the slope and northern Coral Sea (Table 6.4). In contrast, the sediment trap mooring in the Kerema Canyon collected material that was enriched in Cu (higher Cu concentration than found in rivers), had higher sediment accumulation rates, and relatively high sedimentary Cu accumulation rates (0.7–4.3 mmol Cu m2 yr1).

236

Trap site

Water depth (m)

Trap depth (m)

Mooring duration days

SSE Fly River Slope Kerema Canyon Eastern ﬁelds Northern Coral Sea

Cu Al OC S Mass ﬂux (g m2 d1) (mmol g1) (mmol g1) (mmol g1) (mmol g1)

dates

1,040

300 940

50 50

Jun–Jul ‘97 Jun–Jul ‘97

0.115 0.232

970

300 870

44 44

Jun–Jul ‘97 Jun–Jul ‘97

0.181 0.574

1,460

300

29

May ‘97

0.118

1,360

29

May ‘97

0.106

Note: Locations of sediment trap moorings are given in Fig. 6.2.

3.55 5.80 11.6 20.4 2.71 10.0

2.34 7.23

12 3.23

0.73 0.45

2.96 31.5

1.10 2.6

0.23 0.83

0.25 0.98

15.9 5.33

0.16 0.14

G. J. Brunskill et al.

Table 6.5: Mooring details, bulk sediment mass ﬂux, and chemical composition of Gulf of Papua sediment trap material (copper, aluminum, organic carbon, and sulfur).

Trap site

Trap depth (m)

Mass (g m2 yr1)

Cu (mmol m2 yr1)

S (mmol m2 yr1)

Al (mmol m2 yr1)

OC (mmol m2 yr1)

SSE Fly River slope

300 940

42 85

149 491

31 38

98 612

504 274

Kerema Canyon

300 870

66 210

766 4,270

15 174

196 6,600

72 545

Eastern Fields

300

43

117

6.9

11

685

1,360

38

387

5.4

38

206

Northern Coral Sea

Note: Annual ﬂuxes were calculated by multiplying the average trap sediment composition by the sediment mass ﬂux per unit area per day, and then by 365 days yr1. These data were derived from Table 6.5.

Sediment and Copper in the Gulf of Papua, Papua New Guinea

Table 6.6: Sediment trap annual ﬂuxes for total sediment dry weight, copper, sulfur, aluminum, and organic carbon in the Gulf of Papua in 1997.

237

238

G. J. Brunskill et al.

6.3.10. Copper in Annual Growth Bands of Corals One of three replicate Porites sp. coral cores from near Bramble Cay in Torres Strait was scanned by gamma densitometry to determine the chronology of the growth bands. Corals in this region have extension rates of 12–18 mm yr1 and calciﬁcation rates of 1.7–2 g cm2 yr1, and some cores show annual records back to 1776 (Lough et al., 1999). Samples were taken of growth bands from the years 1906 and 1911 to determine the average Fe and Cu concentration of river-supplied elements before mine tailings disposal began in the Fly River catchment. Samples from growth bands representing 1984–1990 and 1990–1992 were taken to determine the average Fe and Cu concentration in corals after mine tailings had been added to the upper Fly River for a decade. The results are presented in Table 6.7 as the ratio Fe/Ca and Cu/Ca, which indicates that recent coral growth bands have Cu/Ca ratios 40–60% greater than this ratio in the early 1900s, whereas Fe/Ca ratios have changed little over this time period. Fe and Cu were approximately equal in concentration in the growth band aragonite (0.2–0.3 mmol g1) in 1906–1911, but Cu concentrations doubled in coral growth bands of 1984–1992 (0.40–0.46 mmol g1).

6.4. Discussion 6.4.1. Rate of Addition of Copper to the Gulf of Papua The mass balance for Cu in the Gulf of Papua is given in Table 6.8. The dominant input is pCu from rivers (264 106 mol yr1), whereas dCu input was estimated from Simon Apte’s data as 10 106 mol yr1. Over 70% of this dCu and pCu came from the Fly River (Table 6.3). The rate of transport of Cu from these wet tropical river catchment areas is relatively high (1–2 mmol Cu m2 yr1) compared to temperate zone river catchments in industrialized countries (Table 6.2). This is partly due to the high relief of the headwater catchments, high rainfall (5–10 m yr1), relatively young bedrock, and the tectonic uplift rate (3–6 cm yr1) of these mountain ranges. The weathering rate of Cu in the headwater mountains of the Fly and Purari Rivers is probably higher than that shown in Table 6.2, as the high relief fraction of the river catchment areas will provide most of the riverine sediment and copper. The Ok Tedi and Porgera mine tailings disposal will provide an additional sediment and trace element load to the river, and the Ok Tedi mine has probably doubled the pCu supply rate of

Sediment and Copper in the Gulf of Papua, Papua New Guinea

239

Table 6.7: The ratio Fe/Ca and Cu/Ca from annual growth bands of a massive Porites sp. coral bommie near Bramble Cay, directly south of the mouth of the Fly River. Sample (no.)

Growth band (year)

Luminescence

110F 49F 845F 890F

1984–1990 1990–1992 1911 1906

Yes No Yes No

Fe/Ca ( 106)

Cu/Ca ( 106)

23 23 20 50

39 47 23 29

Note: This coral core was collected in 1995, and the annual growth bands were tracked using gamma densitometry and measurement of luminescence according to Isdale et al. (1998), Lough et al., (1999). Coral growth bands that are strongly luminescent are thought to be more affected by river runoff, while growth bands with little or no luminescence have had little river inﬂuence.

Table 6.8: The copper mass balance for the Gulf of Papua. Parameter River input Loss to sediment burial Loss to sediment traps, 300 m Loss to sediment traps, 1,000 m Export to Coral Sea

dCu (106 mol yr1) 10.4

0.000119

pCu (106 mol yr1)

S Cu (106 mol yr1)

264

275 156 2.0 6.8

119

119

Note: The annual river input includes the sum of the Fly, Strickland, Purari, and Kikori Rivers, as well as an estimate for ungauged catchment areas (using the unit area annual catchment yield for pCu from the Purari and Kikori Rivers). Loss to sediment burial was calculated from the average Cu burial rate for each depth zone multiplied by the area of that depth zone in Table 6.4. The unit area ﬂuxes to the sediment traps were multiplied by the seaﬂoor area of the slope and northern Coral Sea (200–2,000 m) listed in Table 6.4, to obtain the average annual removal rate from the water column. The dCu exported to the Coral Sea is estimated based upon Kd ¼ 106.

the upper Fly River, and increased dCu by a factor of B10 (Table 6.2, and Simpson et al., 2005). Simpson et al. (2005) show that chemolithotrophic microbes oxidize the fresh sulﬁdic mine tailings (containing chalcopyrite, digenite, and chalcocite) within a few days of discharge to the river, and that downstream pCu is largely nonsulﬁdic. The extensive ﬂoodplains on the catchment lowlands of these rivers are likely to be a sink for sediments and Cu. If this annual river supply rate of 275 106 mol of Cu to the Gulf of Papua was deposited in the sediments of the 0–100 m depth zone

240

G. J. Brunskill et al.

of area 36 109 m2, this would provide an average Cu burial rate of 7.6 mmol m2 yr1. Input of Cu to the Gulf of Papua from dust to this region is likely to be very low (Duce et al. 1991; Hesse and McTainsh, 2003; Jickells et al. 2005), contained in 0.2–0.5 g dust m2 yr1. If the dust contained 5 mmol Al g1, and if Cu/AlB104 in this dust, then the input might be o0.25 mmol Cu m2 yr1, or o15 103 mol yr1 over the Gulf of Papua. Inputs of Cu from upwelling of deep Coral Sea water are not known, but likely to be large, because of the large volume of Coral Sea water advected onto the shelf, compared to river inputs (Coral Sea water/river water ratio B300). 6.4.2. Rate of Removal of Copper From the Gulf of Papua The measured removal of Cu from the water column of the Gulf of Papua to sediment burial at the seaﬂoor from 0–2,000 m water depth was 156 106 mol yr1, or about 56% of the river input rate (Table 6.8). Most of this Cu removal was in the depth zone from 0 to 100 m on the inner shelf, where sedimentation rates were high and Cu burial rates were 0.4 to 15 mmol m2 yr1 (see Table 6.4). The largest area of ﬁne-grained and Al–Cu rich sediment accumulation was in the region between Turama Inlet, the Kikori and Purari River deltas, whereas the delta front clinoform directly southeast of the Fly River contains relatively small amounts of ﬁne sediment and lower accumulation rates (Fig. 6.6). Removal rates of Cu from the slope and pelagic water column of the Northern Coral Sea were much smaller. About 2 106 mol Cu yr1 were scavenged from the clear blue euphotic zone south of the slope (to the 300 m sediment traps), and additional 6.8 106 mol Cu yr1 was delivered to the deep sediments (the 900–1,300 m sediment traps) of the northern Coral Sea by lateral injection (turbidity currents, nepheloid layers) from the shelf (Tables 6.6 and 6.8). We propose that much of this off-shelf transport occurs south of Kerema and eastward toward Port Moresby, where the width and area of the shelf is greatly reduced. Evidence to support this suggestion comes from the 870 m deep Kerema sediment trap (Tables 6.5 and 6.6), which collected ﬁve times more sediment with much higher Cu concentration (20.4 mmol Cu g1), compared to the other two trap sites (Fig. 6.2). Deep water oceanic removal rates of Cu from the water column to sediments are in the range 7–14 mmol Cu m2 yr1, and much of this Cu appears to be associated with organic matter (Fischer et al., 1986; Heggie et al., 1987; Noriki et al., 1998). Surface sediment diagenesis of organic

Sediment and Copper in the Gulf of Papua, Papua New Guinea

241

matter probably releases some of this Cu back to the deep water (Kumar et al., 1996). These sediment accumulation rates of Cu are several orders of magnitude smaller than those found in the inner shelf of the Gulf of Papua, but are within a factor of 10 of some of the data from the 1–2 km water depth zones of the northern Coral Sea (Tables 6.4 and 6.6).

6.4.3. Annual Export from the Shelf and Slope to the Coral Sea This removal of Cu by sedimentation leaves 43% of the river annual input to escape the Gulf into the northern Coral Sea, the Hiri Gyre, and the New Guinea Coastal Current. If a Kd of 106 is appropriate (Shafer et al., 2004) for the salinity 34–35 water of the Hiri Gyre, then most of this Cu must be in the particulate or colloid phase. This annual input of 119 106 mol of Cu per year must be the largest riverine input of Cu to the Coral Sea, as water and sediment inputs from north Queensland are small. Due to its very large volume (B20 Sv, Burrage, 1993), the Hiri Gyre (and the South Equatorial Current) would carry a very large input of Cu. Some perspectives upon how this loss of Cu from the shelf happens can be seen in Table 6.9, where we present the mole ratios Cu/Al, Cu/OC, and Cu/S for the river suspended sediment input, the fraction being buried on the shelf, the Kerema Canyon trap particles, and the largely pelagic particles caught by Table 6.9: Changes in the abundance of copper relative to likely host mineral phases (aluminosilicate clays, organic matter, sulﬁde minerals) for components of the Gulf of Papua Cu mass balance. Mass balance parameter River freshwater suspended sediment Shelf and slope sediments Pelagic and shelf particles in Kerema Canyon traps Pelagic particles in Eastern Fields traps

Cu/ Al 104

Cu/ OC 104

Cu/ S 103

Comments

3–8

8–30

20–96

1–4

3–15

2–20

6–39

78–105

25–50

Loss of Cu, compared to river input Gain of Cu

102–108

2–19

17–71

Gain of Cu

Note: These ratios are taken from Figs. 6.5, 6.7, 6.9 and 6.10 and Tables 6.3, 6.5, and 6.6.

242

G. J. Brunskill et al.

sediment traps SE of Eastern Fields Reef in the northern Coral Sea. We use these ratios to imply the dominant host particle types (aluminosilicate clay minerals, organic matter, and iron/manganese oxide/sulﬁde minerals) that carry pCu from the river inputs to the Coral Sea (Gibbs, 1973) as shown in Figs. 6.4 and 6.7. River suspended sediments have Cu/Al ¼ 3–8 104, whereas most sediments carrying pCu to burial sites on the shelf and slope have reduced Cu concentrations and Cu/Al ¼ 1–4 104 (Figs. 6.8–6.11). This reduction in Cu/Al suggests a release of Cu from the aluminosilicate phase along the reaction pathway. This ratio increases greatly in particles caught by the offshore sediment traps (Tables 6.5, 6.6 and 6.9) where Cu/Al ¼ 40–100 104, and where the aluminosilicate mineral fraction in hemipelagic carbonate sediment is relatively small. However, seaﬂoor sediments below the sediment traps at the base of the slope and northern Coral Sea do not have such high Cu/Al ratios (Figs. 6.8–6.11), indicating that these Cu-rich particles must be carried further seaward, or that diagenesis releases sedimentary Cu into deep Coral Sea water. This same sequence is observed for Cu/OC and Cu/S (high in river suspended sediment, decrease in shelf sediments, much higher Cu enrichment in the Kerema Canyon sediment trap particles), except that the northern Coral Sea (Eastern Fields) trap sediments had similar or lower Cu/OC (2–19 104) compared to the river input sediments (Table 6.9). These variations in Cu/Al and Cu/OC suggest that some transformation of the riverine sediment happens along the salinity gradient and on the innershelf sediment accumulation sites. This transformation of particle surfaces must release Cu from the riverine sediments, sorb Cu to a more mobile (colloidal?) aluminosilicate and/or organic matter phase that can be transported off the shelf to the northern Coral Sea. Perhaps much of the repeated mixing, resuspension, oxidation, redeposition, and diagenetic reduction of transient sediments in the intertidal and subtidal zone (Table 6.4) is the location of this transformation process (Simpson et al., 1998; Aller, 2004; Aller et al., 2004; Aller and Blair, 2004; Robert et al., 2004). Fresh river water Kd ¼ [moles Cu/g SS]/[moles Cu/g ﬁltered water] B2–6 104, whereas at the other end of the salinity gradient in seawater we ﬁnd Kd ¼ 5–10 105 (derived from Simon Apte’s dCu and pCu data, and our data in Fig. 6.4). This suggests that riverine pCu is more labile and reactive, and that marine particles have more sorption sites that attract dCu. Whatever the process might be, Cu accumulation rates in our sediment cores have been relatively constant for several millennia over most of the shelf, slope, and northern Coral Sea (Figs. 6.9–6.11). If Cu uptake into living Porites sp. aragonite comes from the dCu pool at Bramble Cay, then our coral cores suggest that there has been at least a

Sediment and Copper in the Gulf of Papua, Papua New Guinea

243

doubling of the availability of dCu to these corals in the last century (Table 6.7). Inoue et al. (2005) have shown that Cu/Ca varies in coral growth bands according to increase and decrease of anthropogenic Cu supply to the coral habitat. The isopleths of Cu concentration over the Gulf of Papua (Fig. 6.5) give the impression that Cu is equally abundant in the river deltas and inner shelf, and in the northern Coral Sea deep sediments. Table 6.4 shows that the sediment accumulation rate of Cu to deltaic and inner-shelf sediments is 10–100 times greater than that of the deep water Coral Sea sediments, and that the factor controlling this difference is bulk sediment accumulation rate. Pelagic and shelf-derived particles with relatively high Cu concentrations (Tables 6.5 and 6.6) rain to the seaﬂoor at 1–2 km water depths, but most of this Al or OC-bound Cu must be diagenetically released back to the water column and transported elsewhere. The variation in Cu concentration in suspended and bottom sediments is best predicted by Al (Figs. 6.3–6.7), whereas organic matter, sulfur, iron, and manganese do not appear to vary in proportion to Cu concentration in sediment. The equation for this relationship (Fig. 6.7) is Cu ¼ 0:2187ðAlÞ þ 5:7421; R2 ¼ 0:48 In the outer shelf relict carbonate sediments, where both Al and Cu are found in very low concentrations, the sedimentary ratio of Cu/Al may give some spurious results (Fig. 6.6). The Cu/Al relationship (Figs. 6.6 and 6.7) also reveals a few Cu enriched sediment locations to the SE of the North Channel of the Fly River Delta, which are probably undiluted recent Fly River suspended sediments inﬂuenced by Ok Tedi Mine tailings additions to a Fly River headwater tributary (Ok Tedi) over the last 25 years. Notice that our sediment cores have been taken over the period 1993–2003, and that many of the cores with Cu enrichment at the surface are more recent cores. We might expect increasing delivery of Cu-rich sediment to the Gulf of Papua in the next several decades, when the upper delta turbidity maximum becomes saturated with Cu-rich sediments from the Fly River. The addition of 29 1012 g yr1 of Cu-rich mine tailings over the past 25 years of the OTML operations (Simpson et al., 2005) can be seen in a few core sites in the Fly Delta mangrove channels and the intertidal zone eastward in the Gulf of Papua (Figs. 6.8 and 6.9), but natural variations in Cu concentrations over the last millennium are of similar magnitude. Most of these longer term pre-1900 natural variations in Cu concentration (Figs. 6.9–6.11) are the result of variations in supply rates of aluminosilicate (Cu-rich) and carbonate (Cu-poor) minerals to the seaﬂoor.

244

G. J. Brunskill et al.

Mine tailings rich in copper are added to the headwaters of the Ajkwa River in Irian Jaya (West Papua), and the downstream Cu signal is very strong (enhancement by factors of 20–40 above pre 1950 sediment Cu accumulation rates) in the intertidal zone mangrove mud (Brunskill et al., 2004). There must be some large difference in hydrological and sediment transport mechanisms for the Ajkwa River, compared to the Fly River, as the Fly River somehow sequesters much of the mine tailings in the ﬂoodplains and upper delta. It may be that the volume of ﬂuid mud (Wolanski et al., 1995b, Wolanski and Gibbs, 1995) moving in and out of the Fly River delta is so large, that 20 years of mine tailings input is greatly diluted and slowly dispersed over the shelf. If this is the case, then we would expect the titration of the Fly River delta ﬂuid mud with Cu-rich mine tailings to proceed over the next few decades, increasing the pCu concentration in the surface mixed layer of inner-shelf sediments of the entire Gulf of Papua.

6.5. Conclusions About 57% of the river-supplied particulate and dCu can be found in the sediments of the continental shelf and slope of the Gulf of Papua. The remainder (119 106 mol of Cu per year) must be delivered in ligand, colloid, or particulate form to the Hiri Gyre of the Coral Sea. This system is not in steady state, as there has been a large increase in the river supply rate of Cu from mining operations in the headwaters of the Fly River. There are a few sites in the inner shelf where this excess Cu can be observed, but most of the shelf and slope sediment cores do not show large increases in Cu in recent sediments. This is probably due to estuarine dilution by the natural ﬂuid mud reservoir, and century scale bottom sediment mixing on the inner shelf.

ACKNOWLEDGMENTS This work is the result of over a decade of research in the Gulf of Papua as part of Project TROPICS, involving many ships and ship crews, scientiﬁc collaborators, and logistic and diplomatic support from Australia, Papua New Guinea, and USA. We particularly wish to thank the Masters and crew of the R/V Lady Basten and R/V The Harry Messel from the Australian Institute of Marine Science, the R/V Franklin from the Australian National Facility and CSIRO Marine ship managers and technicians, Field Operations Managers John Hardman and Tony McKenna of AIMS, and

Sediment and Copper in the Gulf of Papua, Papua New Guinea

245

David Vaudrey and Ron Plaschke of CSIRO Marine. Ok Tedi Mining LTD (Murray Eagle, Ian Wood, Jim Veness, Barrie Bolton) provided their measurements of river ﬂow for the Fly and Strickland Rivers, and access to the R/V Western Venturer for research in the Fly Delta. We thank Simon Apte for access to his measurements of dCu and pCu in the Fly River salinity gradient. We thank Ruth Blunden and Greg French of the Australian Department of Foreign Affairs and Trade in Canberra, and Sarah Story of the Australian Embassy in Port Moresby, for their assistance in arranging research ship clearances in foreign waters. Don Niles, Michael Laki, and Jim Robbins (PNG National Research Institute), Patricia Pepina (Marine Research Subcommittee, Government of Papua New Guinea), and Dr. Hugh Davies (Earth Sciences, University of Papua New Guinea) assisted with our research visa applications and helped in other ways. Coral cores were collected by Glen Shen, Roberta Hamme, and Irena Zagorskis, and Bruce Parker (deceased) provided expertise in coral growth band history and sampling for chemical analyses. In the AIMS analytical laboratory, Steve Boyle and Cassie Payn did the ICP-AES and DOC analyses presented here. The AIMS Engineering Workshop built all of our coring equipment, and the AIMS library kept us up to date with the biogeochemical literature. We thank Sheila M. Brunskill for help with mangrove coring, careful proof reading of this manuscript, and a lot of other things.

REFERENCES Agell, G., Turon, X., De Caralt, S., Lopez-Legentil, S., & Uriz, M. J. (2004). Molecular and organism biomarkers of copper pollution in the ascidian Pseudodistoma crucigaster. Marine Pollution Bulletin, 48, 759–767. Aller, R. C. (2004). Conceptual models of early diagenetic processes: The muddy seaﬂoor as an unsteady, batch reactor. Journal of Marine Research, 62(6), 815–835. Aller, J. Y., & Aller, R. C. (2004). Physical disturbance creates bacterial dominance of benthic biological communities in tropical deltaic environments of the Gulf of Papua. Continental Shelf Research, 24(19), 2395–2416. Aller, R. C., & Blair, N. E. (2004). Early diagenetic remineralization of sedimentary organic C in the Gulf of Papua deltaic complex (Papua New Guinea): Net loss of terrestrial C and diagenetic fractionation of C isotopes. Geochimica et Cosmochimica Acta, 68(8), 1815–1825. Aller, R. C., Hannides, A., Heilbrun, C., & Panzeca, C. (2004). Coupling of early diagenetic processes and sedimentary dynamics in tropical shelf environments: The Gulf of Papua deltaic complex. Continental Shelf Research, 24(19), 2455–2486. Alongi, D. M., Tirendi, F., & Robertson, A. I. (1991). Vertical proﬁles of copper in sediments from the Fly Delta and Gulf of Papua, Papua New Guinea. Marine Pollution Bulletin, 22, 253–255.

246

G. J. Brunskill et al.

Apte, S. C., & Day, G. M. (1998). Dissolved metal concentrations in the Torres Strait and Gulf of Papua. Marine Pollution Bulletin, 36, 298–304. Apte, S. C. (2009). Biogeochemistry of copper in the Fly River. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 321–373. Audry, S., Schafer, J., Blanc, G., Bossy, C., & Lavaux, G. (2004). Anthropogenic components of heavy metal (Cd, Zn, Cu, Pb) budgets in the Lot-Garonne ﬂuvial system (France). Applied Geochemistry, 19(5), 769–786. Baker, E. K., Harris, P. T., & Beck, R. W. (1990). Cu and Cd associated with suspended particulate matter in Torres Strait. Marine Pollution Bulletin, 21, 484–486. Baker, E. K., & Harris, P. T. (1991). Copper, lead, zinc distribution in the sediments of the Fly River Delta and Torres Strait. Marine Pollution Bulletin, 21, 484–486. Boyle, E. A., Huested, S. S., & Grant, B. (1982). The chemical mass balance of the Amazon Plume. II. Copper, nickel, and cadmium. Deep Sea Research, 29, 1355–1364. Brewer, P. G., & Spenser, D. W. (1975). Minor element models in coastal waters. In: T. M. Church (Ed.). Marine Chemistry of Coastal Environments. American Chemical Society, Washington, D.C., pp. 80–96. Broecker, W. S., & Peng, T.-H. (1982). Tracers in the Sea. Published by LamontDoherty Geological Observatory, Columbia University, Palisades, New York, 690 pp. Brunskill, G. J., Zagorskis, I., & Pﬁtzner, J. (2003). Geochemical mass balance for lithium, boron, and strontium in the Gulf of Papua, Papua New Guinea (Project TROPICS). Geochimica et Cosmochimica Acta, 67(18), 3365–3383. Brunskill, G. J. (2004). New Guinea and its coastal seas, a testable model of wet tropical processes: An introduction to Project TROPICS. Continental Shelf Research, 24, 2273–2295. Brunskill, G. J., Zagorskis, I., Pﬁtzner, J., & Ellison, J. (2004). Sediment accumulation rates and geochemical history in sediment cores from the Ajkwa River estuary mangroves in Irian Jaya (West Papua, Indonesia). Continental Shelf Research, 24, 2535–2551. Burrage, D. M. (1993). Coral sea currents. Corella, 17, 135–145. Canli, M., Stagg, R. M., & Rodger, G. (1997). The induction of metallothionein in tissues of the Norway lobster Nephrops vorvegicus following exposure to cadmium, copper and zinc: The relationships between metallothionein and the metals. Environmental Pollution, 96, 343–350. Carey, A. E., Nezat, C. A., Lyons, W. B., Kao, S.-J., Hicks, D. M., & Owen, J. S. (2002). Trace metal ﬂuxes to the ocean: The importance of high-standing oceanic islands. Geophysical Research Letters, 29(23), 2099, doi:10.1029/2002GL015690. Chadwick, D. B., Zirino, A., Rivera-Duarte, I., Katz, C. N., & Blake, A. C. (2004). Modeling the mass balance and fate of copper in San Diego Bay. Limnology and Oceanography, 49(2), 355–366.

Sediment and Copper in the Gulf of Papua, Papua New Guinea

247

Chelazzi, G. M., De Pirro, M., & Williams, G. A. (2004). Different cardiac response to copper in limpets from metal polluted and clean shores of Hong Kong. Marine Environmental Research, 58, 83–93. Cobelo-Garca, A., & Prego, R. (2003). Land inputs, behaviour and contamination levels of copper in a ria estuary (NW Spain). Marine Environmental Research, 56(3), 403–422. Cobelo-Garcia, A., Prego, R., & Labandeira, A. (2004). Land inputs of trace metals, major elements, particulate organic carbon and suspended solids to an industrial coastal bay of the NE Atlantic. Water Research, 38(7), 1753–1764. Davies, P. (2004). Nutrient processes and chlorophyll in the estuaries and plume of the Gulf of Papua. Continental Shelf Research, 24(19), 2317–2341. Di Toro, D. M., Mahony, J. D., Hansen, D. J., Scott, K. J., Hinks, M. B., Mayr, S. M., & Redmond, S. M. (1990). Toxicity of cadmium in sediments: The role of acid volatile sulﬁde. Environmental Science and Technology, 9, 1487–1502. Dryden, C. L., Gordon, A. S., & Donat, J. R. (2004). Interactive regulation of dissolved copper toxicity by an estuarine microbial community. Limnology and Oceanography, 49(4), 1115–1122. Duce, R. A., Liss, P. S., Merill, J., Atlas, E. L., Buat-Menard, P., Hicks, B. B., Miller, J. M., Prospero, J. M., Arimoto, R., Church, T. M., Ellis, W., Galloway, J. N., Hansen, L., Jickells, T. D., Knap, A. H., Reinhardt, A. H., Schneider, B., Soudine, A., Tokos, J. J., Tsunogai, J. J., Wollast, R., & Zhou, M. (1991). The atmospheric input of trace species to the world ocean. Global Biogeochemical Cycle, 5, 193–259. Edmond, J. M. (1992). Himalayan tectonics, weathering processes, and the strontium isotope record in marine limestones. Science, 258, 1594–1597. Edmond, J. M., & Huh, Y. (1997). Chemical weathering yields from basement and orogenic terrains in hot and cold climates. In: W. F. Ruddiman (Ed.). Tectonic Uplift and Climate Change. Plenium, NY, pp. 329–351. Eisma, D., & van Bennekom, A. J. (1978). The Zaire River and estuary and the Zaire outﬂow in the Atlantic Ocean. Netherlands Journal of Sea Research, 12(3/4), 255–272. Fischer, K., Dymond, J., Lyle, M., Soutar, A., & Rau, S. (1986). The benthic cycle of copper: Evidence from sediment trap experiments in the eastern tropical North Paciﬁc Ocean. Geochimica et Cosmochimica Acta, 50, 1535–1543. Garcia-Villada, L., Rico, M., Altamirano, M., Sanchez-Martin, L., Lopez-Rodas, V., & Costas, E. (2004). Occurrence of copper resistant mutants in the toxic cyanobacteria Microcystis aeruginosa: Characterisation and future implications in the use of copper sulphate as algaecide. Water Research, 38(8), 2207–2213. Garrels, R. M., & Mackenzie, F. T. (1971). Evolution of Sedimentary Rocks. Norton Publishers, New York. Geracitano, L. A., Bocchetti, R., Monserrat, J. M., Regoli, F., & Bianchini, A. (2004). Oxidative stress responses in two populations of Laeonereis acuta (Polychaeta, Nereididae) after acute and chronic exposure to copper. Marine Environmental Research, 58, 1–17.

248

G. J. Brunskill et al.

Gibbs, R. J. (1973). Mechanisms of trace metal transport in rivers. Science, 180, 71–73. Goldberg, E. D. (1963). The oceans as a chemical system. In: The Sea, Volume 2. Wiley, New York, pp. 3–25. Gordon, A. S. (1992). Isolation of compounds with afﬁnity for copper from seawater using immobilized copper ion afﬁnity chromatography. Marine Chemistry, 38, 1–12. Gordon, A. S., Dyer, B. J., Kango, R. A., & Donat, J. R. (1996). Copper ligands isolated from estuarine water by immobilized metal afﬁnity chromatography: Temporal variability and partial characterization. Marine Chemistry, 53, 163–172. Guieu, C., Martin, J.-M., Tankere, S. P. C., Mousty, F., Trincherini, P., Bazor, M., & Dai, M. H. (1998). On trace metal geochemistry in the Danube River and Western Black Sea. Estuarine, Coastal and Shelf Science, 47, 471–485. Harris, P. T., Hughes, M. G., Baker, E. K., Dalrymple, R. W., & Keene, J. B. (2004). Sediment transport in distributary channels and its export to the prodeltaic environment in a tidally dominated delta: Fly River, Papua New Guinea. Continental Shelf Research, 24(19), 2431–2454. Hedges, J. I., Bergamaschi, B. A., & Benner, R. (1993). Comparative analyses of DOC and DON in natural waters. Marine Chemistry, 41, 121–134. Herskovits, T. T., & Hamilton, M. G. (1991). Higher order assemblies of molluscan hemocyanins. Comparative Biochemistry and Physiology, 99B, 19–34. Heggie, D., Klinkhammer, G., & Cullen, D. (1987). Manganese and copper ﬂuxes from continental margin sediments. Geochimica et Cosmochimica Acta, 51(5), 1059–1070. Heidmann, I., Christl, I., & Kretzschmar, R. (2005). Sorption of Cu and Pb to kaolinite–fulvic acid colloids: Assessment of sorbent interactions. Geochimica et Cosmochimica Acta, 69(7), 1675–1686. Hemer, M. A., Harris, P. T., Coleman, R., & Hunter, J. (2004). Sediment mobility due to currents and waves in the Torres Strait-Gulf of Papua region. Continental Shelf Research, 24(19), 2297–2316. Hesse, P. P., & McTainsh, G. H. (2003). Australian dust deposits: Modern processes and the Quaternary record. Quaternary Science Review, 22(18/19), 2007–2035. Hooke, R. LeB. (2000). On the history of humans as geomorphic agents. Geology, 28, 843–846. Inoue, M., Suzuki, A., Nohara, M., Kan, H., Edward, A., & Kawahata, H. (2005). Coral skeletal tin and copper concentrations at Pohnpei, Micronesia: Possible index for marine pollution by toxic anti-biofouling paints. Environmental Pollution, 129(3), 399–407. Isdale, P. J., Stewart, B. J., Tickle, K. S., & Lough, J. M. (1998). Palaeohydrological variation in a tropical river catchment: A reconstruction using ﬂuorescent bands in corals of the Great Barrier Reef, Australia. The Holocene, 8, 1–8.

Sediment and Copper in the Gulf of Papua, Papua New Guinea

249

Islam, M. S., & Tanaka, M. (2004). Impacts of pollution on coastal and marine ecosystems including coastal and marine ﬁsheries and approach for management: A review and synthesis. Marine Pollution Bulletin, 48, 624–649. James, R. H., Elderﬁeld, H., & Palmer, M. R. (1995). The chemistry of hydrothermal ﬂuids from the Broken Spur site, 291N mid-Atlantic Ridge. Geochimica et Cosmochimica Acta, 59(4), 651–659. Jiann, K.-T., Wen, L.-S., & Santschi, P. H. (2005). Trace metal (Cd, Cu, Ni and Pb) partitioning, afﬁnities and removal in the Danshuei River estuary, a macro-tidal, temporally anoxic estuary in Taiwan. Marine Chemistry, 96, 293–313. Jickells, T. D., An, Z. S., Anderson, K. K., Baker, A. R., Bergametti, G., Brooks, N., Cao, J. J., Boyd, P. W., Duce, R. A., Hunter, K. A., Kawahata, H., Kubilay, N., La Roche, J., Liss, P. S., Mahowald, N., Prospero, J. M., Ridgwell, A. J., Tegen, I., & Torres, R. (2005). Global iron connections between desert dust, ocean biogeochemistry and climate. Science, 308, 67–71. Jonas, R. B. (1989). Acute copper and cupric ion toxicity in an estuarine microbial community. Applied and Environmental Microbiology, 55, 43–49. Kim, S.-G., & Kang, J.-C. (2004). Effect of dietary copper exposure on accumulation, growth and haematological parameters of the juvenile rockﬁsh, Sebastes schlegeli. Marine Environmental Research, 58, 65–82. Kim, H. J., Graham, D. W., DiSpirito, A. A., Alterman, M. A., Galeva, N., Larive, C. K., Asunskis, D., & Sherwood, P. M. A. (2004). Methanobactin, a copperacquisition compound from methane-oxidizing bacteria. Science, 305(5690), 1612–1615. Kozelka, P. B., & Bruland, K. W. (1998). Chemical speciation of dissolved Cu, Zn, Cd, Pb in Narragansett bay, Rhode Island. Marine Chemistry, 60, 267–282. Kuehl, S. A., Nittrouer, C. A., DeMaster, D. J., & Curtin, T. B. (1985). A long square-barrel gravity corer for sedimentological and geochemical investigation of ﬁne-grained sediments. Marine Geology, 62, 365–370. Kumar, N., Anderson, R. F., & Biscaye, P. E. (1996). Remineralization of particulate authigenic trace metals in the Middle Atlantic Bight: Implications for proxies of export productivity. Geochimica et Cosmochimica Acta, 60, 3383–3397. Lackschewitz, K. S., Devey, C. W., Stoffers, P., Botz, R., Eisenhauer, A., Kummetz, M., Schmidt, M., & Singer, A. (2004). Mineralogical, geochemical and isotopic characteristics of hydrothermal alteration processes in the active, submarine, felsic-hosted PACMANUS ﬁeld, Manus Basin, Papua New Guinea. Geochimica et Cosmochimica Acta, 68(21), 4405–4427. Lehman, J. T., Bazzi, A., Nosher, T., & Nriagu, J. O. (2004). Copper inhibition of phytoplankton in Saginaw Bay, Lake Huron. Canadian Journal of Fisheries and Aquatic Sciences, 61(10), 1871–1880. Li, Y.-H., & Gregory, S. (1974). Diffusion of ions in seawater and in deep-sea sediments. Geochimica et Cosmochimica Acta, 38, 703–714. Li, Y.-H. (2000). A compendium of geochemistry, from solar nebula to the human brain. Princeton University Press, Princeton, NJ, 475 pp.

250

G. J. Brunskill et al.

Lieberman, R. L., & Rosenzweig, A. C. (2005). Crystal structure of a membranebound metalloenzyme that catalyses the biological oxidation of methane. Nature, 434(7030), 177–182. Lippard, S. J. (1999). Free copper ions in the cell?. Science, 284, 748–749. Loring, D. H., & Rantala, R. T. T. (1992). Manual for the geochemical analyses of marine sediments and suspended matter. Earth-Science Reviews, 32, 235–283. Lough, J. M., Barnes, D. J., Devereux, B. J., & Tobin, S. (1999). Variability in Growth Characteristics of Massive Porites on the Great Barrier Reef. CRC Reef Research Centre Technical Report No. 28, Townsville, Queensland, 95 pp. Lowenstein, T. K., Timofeeff, M. N., Kovalevych, V. M., & Horita, J. (2005). The major ion composition of Permian seawater. Geochimica et Cosmochimica Acta, 69, 1701–1719. Luoma, S. N., van Geen, A., Lee, B.-G., & Cloern, J. E. (1998). Metal uptake by phytoplankton during a bloom in south San Francisco Bay: Implications for metal cycling in estuaries. Limnology and Oceanography, 43(5), 1007–1016. Mackenzie, F. T., & Garrels, R. M. (1966). Chemical mass balance between rivers and oceans. American Journal of Science, 264, 507–525. Mackenzie, F. T. & Wollast, R. (1977). Sedimentary cycling of global processes. In: E. D. Goldberg, I. N. McCave, J. J. O’Brien & J. H. Steele (Eds). The Sea-Ideas and Observations on Progress in the Study of the Seas, Vol. 6: Marine Modeling. Wiley, New York, pp. 739–788. Mantoura, R. F. C., Martin, J.-M., & Wollast, R. (Eds). (1991). Ocean Margin Processes in Global Change. Wiley, New York, p. 469. Martin, J.-M., Thomas, A. J., & van Grieken, R. (1978). Trace element composition of Zaire suspended sediments. Netherlands Journal of Sea Research, 12(3/4), 414–420. Martin, J.-M., & Meybeck, M. (1979). Elemental mass balance of material carried by major world rivers. Marine Chemistry, 7, 173–206. Martin, J.-M., & Windom, H. L. (1991). Present and future roles of ocean margins in regulating marine biogeochemical cycles of trace elements. In: R. F. C. Mantoura, J.-M. Martin, & R. Wollast (Eds). Ocean Margin Processes in Global Change. Wiley, New York, pp. 45–67. Massey, A. G. (1973). The chemistry of copper. In: The Comprehensive Inorganic Chemistry Series, Chapter 27. Pergamon Press, Paris, pp. 1–78. McCallan, M. E., O’Leary, B. M., & Rose, C. W. (1980). Redistribution of Caesium137 by erosion and deposition on an Australian soil. Australian Journal of Soil Research, 18, 119–128. Meybeck, M. (1979). Note on dissolved elemental contents of the Zaire River. Netherlands Journal of Sea Research, 12(3/4), 293–295. Meybeck, M., & Vorosmarty, C. (2005). Fluvial ﬁltering of land-to-ocean ﬂuxes: From natural holocene variations to anthropocene. Comptes Rendus Geoscience, 337, 107–123. Meyer, J. S., Santore, R. C., Bobbitt, J. P., Debrey, LD., Boese, C. J., Paquin, P. R., Allen, H. E., Bergman, H. L., & DiToro, D. M. (1999). Binding of nickel and

Sediment and Copper in the Gulf of Papua, Papua New Guinea

251

copper to ﬁsh gills predicts toxicity but free ion activity does not. Environmental Science and Technology, 33(6), 913–916. Moffett, J. W., & Brand, L. E. (1996). Production of strong, extracellular Cu chelators by marine cyanobacteria in response to Cu stress. Limnology and Oceanography, 41, 388–395. Moffett, J. W., Brand, L. E., Croot, P. L., & Barbeau, K. A. (1997). Cu speciation and cyanobacterial distribution in harbours subject to anthropogenic Cu inputs. Limnology and Oceanography, 42, 789–799. Monbet, P. (2004). Dissolved and particulate ﬂuxes of copper through the Morlaix River estuary (Brittany, France): Mass balance in a small estuary with strong agricultural catchment. Marine Pollution Bulletin, 48, 78–86. Negri, A. P., & Heyward, A. J. (2000). Inhibition of coral fertilisation and larval metamorphosis by tributyltin and copper. Marine Environmental Research, 51, 17–27. Noriki, S., Arasahitani, Y., Minakawa, M., Harada, K., & Tsunogai, S. (1998). Vertical cycling of Cu and Ni in the western North and Equatorial Paciﬁc. Marine Chemistry, 59, 211–218. Peizhen, Z., Molnar, P., & Downs, W. R. (2001). Increased sedimentation rates and grain sizes 2–4 Myr ago due to the inﬂuence of climate change on erosion rates. Nature, 410, 891–897. Pﬁtzner, J., Brunskill, G. J., & Zagorskis, I. (2004). 137Cs and excess 210Pb deposition patterns in estuarine and marine sediment in the central region of the Great Barrier Reef Lagoon, north-eastern Australia. Journal of Environmental Radioactivity, 76(1-2), 81–102. Pickard, G. L., Donguy, J. R., Henin, C., & Rougerie, F. (1977). A Review of the Physical Oceanography of the Great Barrier Reef and Western Coral Sea. Australian Institute of Marine Science Monograph Series, volume 2, 135 pp. Prigge, S. T., Eipper, B. A., Mains, R. E., & Amzel, L. M. (2004). Dioxygen binds end-on to mononuclear copper in a precatalytic enzyme complex. Science, 304(5672), 864–867. Rae, T. D., Schmidt, P. J., Pufahl, R. A., Culotta, V. C., & O’Halloran, T. V. (1999). Undetectable intracellular free copper: The requirement of a copper chaperone for superoxide dismutase. Science, 284, 805–808. Richter, F. M., & Turekian, K. (1993). Simple models for the geochemical response of the ocean to climatic and tectonic forcing. Earth and Planetary Science Letters, 119, 121–131. Rivera-Duarte, I., Rosen, G., Lapota, D., Chadwick, D. B., Kear-Padilla, L., & Zirino, A. (2005). Copper toxicity to larval stages of three marine invertebrates and copper complexation capacity in San Diego bay, California. Environmental Science and Technology, 39(6), 1542–1546. Robbins, J. A. (1978). Geochemical and geophysical applications of radioactive lead. In: J. O. Nriagu (Ed.). Biogeochemistry of Lead in the Environment, Vol. 1A. Elsevier Scientiﬁc Publishers, pp. 285–393.

252

G. J. Brunskill et al.

Robbins, J. A. (1986). A model for particle-selective transport of tracers in sediments with conveyor belt deposit feeders. Journal of Geophysical Research, 96, 17081–17104. Robert, S., Blanc, G., Schafer, J., Lavaux, G., & Abril, G. (2004). Metal mobilization in the Gironde Estuary (France): The role of the soft mud layer in the maximum turbidity zone. Marine Chemistry, 87(1–2), 1–13. Robertson, A. I., Daniel, P. A., Dixon, P., & Alongi, D. M. (1993). Pelagic biological processes along a salinity gradient in the Fly Delta and adjacent river plume (Papua New Guinea). Continental Shelf Research, 13(2–3), 205–224. Rubey, W. W. (1951). Geologic history of sea water. Bulletin of the Geological Society of America, 62, 1111–1147. Sarasquete, M. C., Gonzales de Canales, M. L., & Gimeno, S. (1992). Comparative histopathological alterations in the digestive gland of marine bivalves exposed to Cu and Cd. European Journal of Histochemistry, 36, 223–232. Schlesinger, W. H. (1997). Biogeochemistry, An Analysis of Global Change. Academic Press, Sydney, Australia, 588 pp. Seritti, A., Pellegrini, D., Morelli, E., Barghigiani, C., & Ferrara, R. (1986). Copper complexing capacity of phytoplankton exudates. Marine Chemistry, 18, 351–358. Shafer, M. M., Hoffmann, S. R., Overdier, J. T., & Armstrong, D. E. (2004). Physical and kinetic speciation of copper and zinc in three geochemically contrasting marine estuaries. Environmental Science and Technology, 38(14), 3810–3819. Shankar, R., & Manjunatha, B. R. (1994). Elemental composition and particulate mental ﬂuxes from Netravati and Gurpur Rivers to the coastal Arabian Sea. Journal of the Geological Society of India, 43, 255–265. Sharp, A. A., & Stearns, D. E. (1997). Sublethal effects of cupric ion activity on the grazing behaviour of three calanoid copepods. Marine Pollution Bulletin, 34, 1041–1048. Sholkovitz, E. R., van Grieken, R., & Eisma, E. (1979). The major element composition of the suspended matter in the Zaire river and estuary. Netherlands Journal of Sea Research, 12(3/4), 407–413. Sholkovitz, E. R., & Copland, D. (1981). The coagulation, solubility and adsorption properties of Fe, Mn, Cu, Ni, Cd, Co, and humic acids in river water. Geochimica et Cosmochimica Acta, 45, 181–189. Simpson, S. L., Apte, S. C., & Batley, G. E. (1998). Effect of short-term resuspension events on trace metal speciation in polluted anoxic sediments. Environmental Science and Technology, 32, 620–625. Simpson, S. L., Apte, S. C., & Davies, C. M. (2005). Bacterially assisted oxidation of copper sulﬁde minerals in tropical river waters. Environmental Chemistry, 2(1), 49–55. Skrabal, S. A., Donat, J. R., & Burdige, D. J. (1997). Fluxes of copper-complexing ligands from estuarine sediments. Limnology and Oceanography, 42, 992–996. Spatari, S., Bertram, M., Gordon, R. B., Henderson, K., & Graedel, T. E. (2005). Twentieth century copper stocks and ﬂows in North America: A dynamic analysis. Ecological Economics, 54, 37–51.

Sediment and Copper in the Gulf of Papua, Papua New Guinea

253

Springer-Young, M., Piotrowicz, S. R., & Harvey, G. R. (1992). Impact of an episodic event on copper and zinc concentrations and speciation in surface waters of the Gulf of Mexico. Deep Sea Research, 39, 953–964. Strawn, D. G., Palmer, N. E., Furnare, L. J., Goodell, C., Amonette, J. E., & Kukkadapu, R. K. (2004). Copper sorption mechanisms on smectites. Clays and Clay Minerals, 52(3), 321–333. Sunda, W. G., & Huntsman, S. A. (1995). Regulation of Cu concentration in the oceanic nutricline by phytoplankton uptake and regeneration rates. Limnology and Oceanography, 40, 132–137. Syvitski, J. P. M., Vorosmarty, C. J., Kettner, A. J., & Green, P. (2005). Impact of humans on the ﬂux of terrestrial sediment to the global coastal ocean. Science, 308(5720), 376–380. Thompson, M., & Walsh, J. N. (1993). Handbook of Inductively Coupled Plasma Spectrometry (2nd ed.). Chapman and Hall Publishers, pp. 155–156. Tocheva, E. I., Rosell, R. I., Mauk, A. G., & Murphy, M. E. P. (2004). Side-on copper-nitrosyl coordination by nitrite reductase. Science, 304(5672), 867–870. Torgersen, T., Chivas, A. R., & Chapman, A. (1983). Chemical and isotopic characterisation and sedimentation rates in Princess Charlotte Bay, Queensland. Bureau of Mineral Resources Journal of Australian Geology and Geophysics, 8, 191–200. van Hullebusch, E., Chatenet, P., Deluchat, V., Chazal, P. M., Froissard, D., Botineau, M., Ghestem, A., & Baudu, M. (2003). Copper accumulation in a reservoir ecosystem following copper sulfate treatment (St. Germain Les Belles, France). Water, Air and Soil Pollution, 150, 3–22. Virkutyte, J., Velizarova, E., Ribeiro, A. B., & Sillanpaa, M. (2005). Copper and chromium electrodialytic migration in CCA-treated timber waste. Water, Air, Soil Pollution, 160(1–4), 27–40. Vorosmarty, C. J., Meybeck, M., Fekete, B., Sharma, K., Green, P., & Syvitski, J. P. M. (2003). Anthropogenic sediment retention: major global impact from registered river impoundments. Global and Planetary Change, 39, 169–190. Waeles, M., Riso, R. D., Maguer, J.-F., & Le Corre, P. (2004a). Distribution and chemical speciation of dissolved cadmium and copper in the Loire estuary and North Biscay continental shelf, France. Estuarine, Coastal and Shelf Science, 59(1), 49–57. Waeles, M., Riso, R. D., & Le Corre, P. (2004b). Seasonal variations of dissolved and particulate copper species in estuarine waters. Estuarine Coastal and Shelf Science, 62(1–2), 313–323. Walsh, J. P., Nittrouer, C. A., Palinkas, C. M., Ogston, A. S., Sternberg, R. W., & Brunskill, G. J. (2004). Clinoform mechanics in the Gulf of Papua, New Guinea. Continental Shelf Research, 24(19), 2487–2510. Williams, M. R., Millward, G. E., Nimmo, M., & Fones, G. (1998). Fluxes of Cu, Pb, and Mn to the NE Irish Sea: The importance of sedimental and atmospheric inputs. Marine Pollution Bulletin, 36, 366–375. Windom, H., Smith, R., Rawlinson, C., Hungspreugs, M., Dharmvanij, S., & Wattayakorn, G. (1988). Trace metal transport in a tropical estuary. Marine Chemistry, 24, 293–305.

254

G. J. Brunskill et al.

Wolanski, E., & Gibbs, R. J. (1995). Flocculation of suspended sediment in the Fly River estuary, Papua New Guinea. Journal of Coastal Research, 11, 754–762. Wolanski, E., Norro, A., & King, B. (1995a). Water circulation in the Gulf of Papua. Continental Shelf Research, 15, 185–212. Wolanski, E., King, B., & Galloway, D. (1995b). Dynamics of the turbidity maximum in the Fly River Estuary, PNG. Estuarine, Coastal and Shelf Science, 40, 321–337. Wolanski, E., King, B., & Galloway, D. (1997). Salinity intrusion in the Fly River estuary, PNG. Journal of Coastal Research, 13, 983–994. Wolanski, E., Gibbs, R. J., Spagnol, S., King, B., & Brunskill, G. J. (1998). Inorganic sediment budget in the mangrove-fringed Fly River Delta, Papua New Guinea. Mangroves and Salt Marshes, 2, 85–98. Zhang, J., Huang, W. W., Liu, S. M., Liu, M. G., Yu, Q., & Wang, J. H. (1992). Transport of particulate heavy metals towards the China Sea: A preliminary study and comparison. Marine Chemistry, 40, 161–178. Zhang, J. (1999). Heavy metal compositions of suspended sediments in the Changjiang (Yangtze) River) estuary: Signiﬁcance of riverine transport to the ocean. Continental Shelf Research, 19, 1521–1543. Zwolsman, J. J. G., & van Eck, G. T. M. (1999). Geochemistry of major elements and trace metals in suspended matter of the Scheldt estuary, southwest Netherlands. Marine Chemistry, 66, 91–111.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00407-2

Chapter 7

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System Geoff Pickup1, and Yantao Cui2 1

Consulting Geomorphologist, 1538 Sutton Road, Sutton, New South Wales 2620, Australia 2 Consulting Geomorphologist, Stillwater Sciences, 2855 Telegraph Avenue, Berkeley, CA 94705, USA

7.1. Historical Background Sediment transport modeling has a long history at the Ok Tedi. It began in 1979, after reconnaissance investigations predicted signiﬁcant deposition in both the Ok Tedi and the Fly (e.g., Pickup, 1978). Disposal of tailings and loss of material from the waste rock dumps into the Kawerong and Jaba Rivers from the Bougainville copper mine during the early 1970s had already caused signiﬁcant environmental and social problems. The Ok Tedi mine was expected to have similar effects, but the likely extent and severity were unknown. More information was therefore sought by both the Papua New Guinea government and the Mining Consortium prior to negotiation of the mining agreement. Sediment transport models had previously been developed for the Bougainville mine (Higgins, 1979; Pickup and Higgins, 1979) and for the Purari River in the Gulf of Papua (Pickup, 1980). It therefore seemed feasible to apply them to the Ok Tedi. Before mine construction began, it was expected that tailings would be held behind a dam on the Ok Ma and it would be possible to build conventional, relatively stable waste rock dumps at the mine site. Only soft and highly weathered overburden was to be disposed of via the river system, although Corresponding author. Tel.: þ61 26238 3427;

E-mail: [email protected] (G. Pickup).

258

G. Pickup and Y. Cui

Pickup et al. (1979) anticipated up to 30% loss from the conventional waste rock dumps. The ﬁrst modeling studies therefore concentrated on the Mine Area Creeks (Ok Mani and Sulphide Creek) and the Ok Tedi. These are steep and highly active systems with braided channels and, in the case of the Ok Tedi, a number of gorge reaches. The early models focused on gravel transport and deposition, as only limited amounts of sand and ﬁner sediment were expected and would pass through the Ok Tedi to reach the Middle Fly. Construction difﬁculties caused the tailings dam proposal to be abandoned in 1984, and this radically changed the impact of the mine. Tailings were now disposed of into the river system. Large volumes of ﬁne sediment began to accumulate in the ﬂoodplain reaches of the Lower Ok Tedi and the upper Middle Fly. Rising water levels and rain forest dieback in these areas had become apparent in the mid-1990s, and the focus of modeling shifted to the sand-bed ﬂoodplain channels. Work also began on ﬂoodplain and levee deposition and increased potential for ﬂooding. The most recent sediment modeling studies have combined upper and lower river system deposition models with studies of ﬂoodplain hydraulics (Cui and Parker, 1999; Pickup and Cui, 2003; Pickup, 2003, 2005) and predict deposition, water level rise, and the lateral extent of ﬂooding. However, attempting to model the full implications of mine waste deposition throughout the river system remains a daunting task, and there are still many gaps in knowledge. Since the ﬁrst study by Pickup et al. (1979), many different types of sediment transport model have been used at the Ok Tedi. The techniques have varied from empirical estimates based on riverbed sediment particle sizes (Vanoni and Henderson, 1979) to sophisticated scale models (Foster et al., 1984) and two-dimensional hydrodynamic approaches (Hranisavljevic et al., 1996). Some modeling exercises have been successful, while others are less so, often because data on sediment inputs or system characteristics are inaccurate or lacking (Higgins et al., 1987). Indeed, matching model sophistication to data accuracy and availability has always been a problem, and every modeling study has faced signiﬁcant unknowns. Overall, however, both data and model accuracy have improved over time. In this chapter, rather than looking at each modeling study, we review the range of approaches taken and some of the problems faced. We also present results from a recent sediment modeling study and preview some future activities. The review is not exhaustive and provides a somewhat personalized view of a 25-year-long modeling effort at the Ok Tedi.

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

259

7.2. Challenges in Modeling the Impact of the Ok Tedi Project Predicting the impact of the Ok Tedi mine is one of the world’s most difﬁcult sediment modeling challenges, with almost every kind of ﬂuvial environment and transport process represented in the system. No single model has proved capable of representing all these environments, and many of the processes involved go beyond the limitations of the sediment transport theory. We examine these issues in detail below, but some common problems include: Lack of knowledge about natural sediment loads in the system. While some baseline information was collected for the Ok Tedi before mining started, initial estimates of load varied widely (Higgins et al., 1987). For example, between 1978 and 1985, ﬁgures of 1.5–21.4 Mt/year were used for the natural load of the Upper Ok Tedi and 3–30 Mt/year for the Lower Ok Tedi. More recently, natural sediment loading has been treated as a model calibration parameter and varied so that a model receiving only natural load shows limited or zero deposition. Lack of understanding of the long-term, large-scale tectonic and geomorphic processes, such as possible basin subsidence and continued basin response to the Holocene sea level rise. It is possible that the river is still recovering from the Holocene sea level rise (e.g., Dietrich et al., 1999) and continues to aggrade as a natural background condition to mine-related deposition. Basin subsidence, if it exists, provides an additional sediment sink. The response of the river to sea level rise and basin subsidence could be a signiﬁcant element of the overall sediment budget, given the 500-km-long river reach that has to be modeled. Problems with the sediment transport theory. Existing transport equations do not perform well in the river system. In the mine area and the Upper Ok Tedi, gravel transport equations are pushed to their limit by the very high sediment concentrations. In the backwater zone of the Middle Fly, there are extensive silt deposits on the channel bed, and silt may even travel as bedload. No sediment transport equation handles this situation. Modeling studies that have attempted to validate their results by comparison with observed behavior have usually had to resort to a model calibration to get the models to work. Lack of information on channel changes. Most modeling has been one dimensional and used very simple models of change in channel cross section. Deposition in the upper river system has involved transitions from a single channel to extreme forms of braiding. Erosion in the post-mining period will create terraces and a reversion to a single channel in many reaches. We lack

260

G. Pickup and Y. Cui

even the basic knowledge to replicate these processes in sediment transport models. Flow varies downstream in the Middle Fly as water goes overbank and is added to off-channel water bodies, or as these bodies drain back into the main channel. We know little about these processes. Also, the ﬂoodplain usually has too much cloud cover for monitoring using visible/near-infrared monitoring from remote-sensing satellites. We lack a ﬂoodplain deposition model apart from an early attempt at modeling reported by Dietrich (2000). Existing theory about ﬂoodplain processes is limited, and construction of an empirical model is a daunting logistical task, given the size and variability of the ﬂoodplain. Many of these challenges have remained unchanged since the beginning of the Ok Tedi project. Other areas have advanced, including recent developments in transport of ﬁne sediment (e.g., Wright and Parker, 2005a, b). We describe progress and results in the following sections.

7.2.1. Approaches to Upper River System Geomorphology 7.2.1.1. System behavior Coarse sediment is delivered from the waste rock dumps to Harvey Creek and headwaters of Sulphide Creek by mechanical failures, landslides, and debris ﬂows (see Fig. 1.1 of Pickup and Marshall, 2009). In the northern dumps, material is fed into Sulphide Creek down steep slopes and over a waterfall. Below the southern dumps, the upper part of Harvey Creek also has debris ﬂows and landslides, but further downstream, a large alluvial fan that extends into the valley of the Ok Mani has developed. The fan behaves as a debris ﬂow runout zone, but its toe is continuously eroded by the Ok Mani. Although the bed of Harvey Creek has built up by more than 50 m in places since dumping started, it has also experienced major sidewall erosion as water ﬂow has moved to fan edges and undercut adjacent hillsides. This process added an estimated 180 Mt of sediment to the river system between 1990 and 2003. Sidewall undercutting may also have played a role in the Vancouver landslide in the upper reaches of Sulphide Creek, which released about 120 Mt between 1989 and 1992. Sidewall erosion occurs throughout the Mine Area Creeks as laterally migrating braided channels undermine adjacent hillsides. The Ok Mani and lower Sulphide Creek are both highly active braided systems, and it is quite common to see 10 or more individual channels in reaches experiencing rapid deposition. Suspended sediment concentrations

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

261

below the tailings discharge point in the Ok Mani can exceed 100,000 mg/l, and the water–sediment mixture in many channels may ﬂow as a supercritical slurry. The Ok Tedi provides a range of geomorphic environments (Fig. 7.1). Narrow gorge sections alternate with wider reaches, with several channels in upstream reaches. However, there has been only limited long-term deposition because of steep slopes and high discharges. Further downstream, the river has a braided gravel channel, a mixed-load braided and single channel reach, and ﬁnally a single sand-bed channel. Deposition has raised the bed of each of the channel systems to the extent that both ﬂow and deposition now regularly encroach onto the ﬂoodplain.

7.2.1.2. Representing sediment transport in the Mine Area Creeks Early attempts to represent dump behavior involved application of a simple dump loss rate that did not vary through time. Initially, there was much controversy about what these rates might be, and initial designs involved stable dumps on the northern side of the mine and failing dumps feeding into the Ok Mani. Pickup et al. (1979) used measured dump loss rates from the Bougainville mine but ran a range of optimistic and pessimistic scenarios. Others used more conservative loss rates (e.g., Vanoni and Henderson, 1979). Later studies have used observed dump loss rates and tried to represent the interaction between the debris ﬂow zones below the dumps and the braided systems downstream. In the OkGrav model, progressively developed by Gary Parker since the late 1980s (e.g., Parker and Dietrich, 1993), the river system has debris ﬂow runout zones below the northern and southern dumps. These feed into braided channel reaches covering lower Sulphide Creek, the Ok Mani, and the Upper Ok Tedi above the Ok Mani junction. These feed into the Middle Ok Tedi, which is treated as a gravel reach capable of transporting sand and ﬁner material as wash load. Rock waste is fed into the model at the northern and southern dumps at rates based on observed or projected annual dumping rates. The rock waste is initially partitioned into gravel, sand, and silt fractions based on expected breakdown from the dump site to the base of the debris ﬂow runout zones. Much of the gravel, and some of the sand and silt, is captured in the debris ﬂow runout zones as they aggrade with constant slope (Klohn Leonoff, 1993). The rest of the gravel is delivered to the ﬂuvial sub-reaches immediately downstream. Sediment is also produced indirectly in the form of landslide material as a consequence of the dumping of rock waste. This material includes the

262

G. Pickup and Y. Cui

Figure 7.1: Geomorphic zones of the Mine Area Creeks and Middle Ok Tedi. Upper left: alluvial fan and debris ﬂow runout zone, Harvey Creek, and Ok Mani. Upper right: debris ﬂow runout zone, upper Sulphide Creek. Centre left: a hyperbraided reach on the lower Ok Mani. Centre right: gravel bars in a conﬁned reach of the Middle Ok Tedi. Lower left: a formerly armored and stable braided reach on the Middle Ok Tedi now experiencing major deposition. Lower right: the dredging operation at the downstream end of the Middle Ok Tedi in what was formerly the gravel–sand transition zone. Photos: Barrie Bolton, Stuart Miller, and Geoff Pickup.

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

263

Vancouver slide on the upper reach of Sulphide Creek, which added about 120 Mt of material between 1989 and 1992. It also includes sidewall erosion in upper Harvey Creek arising from undercutting by streams at the edge of the Harvey Creek fan. Observed and predicted schedules for this delivery are occasionally updated based on new information from consultants working for the mine. The initial breakdown into gravel, sand, and silt fractions is assumed to be identical to that for rock waste. The debris ﬂow runout zones are assumed to aggrade at a constant slope of 8%, based on ﬁeld measurements. The fractions of gravel, sand, and silt captured in this zone are also based on ﬁeld measurements. The area of deposition expands as the bed aggrades and the river valleys widen. The downstream ends of the debris ﬂow runout zones are constrained by waterfalls or a shift to ﬂuvial sediment transport as slope decreases downvalley and water discharge increases. In the OkGrav model, it was assumed that the debris zones aggrade and degrade at the same vertical rate as the uppermost node of the ﬂuvial zone. After the mine ceases operation, degradation processes will be driven from downstream by the ﬂuvial system. Degradation of the debris zones is then adjusted to keep pace with incision in the ﬂuvial zones downstream.

7.2.1.3. Modeling braided gravel channel systems in the Ok Mani, Sulphide Creek, and the Ok Tedi The ﬁrst attempts to model sediment transport in the Mine Area Creeks and the Ok Tedi used models that had been developed and calibrated for transport of mine waste on the Kawerong and Jaba Rivers in Bougainville (Higgins, 1979; Pickup and Higgins, 1979). The Bougainville experience showed that it was not feasible to use standard sediment transport equations, and some form of direct or indirect model calibration was necessary. Problems were also experienced in obtaining sufﬁcient topographic data to adequately represent system geometry, and eventually, two approaches evolved (Pickup et al., 1979; Maunsell and Partners, 1982). Where detailed topography was available, notably the Mine Area Creeks and the Upper Ok Tedi, the system was divided into short reaches, and sediment routing was applied. This involved calculation of a reach-by-reach sediment mass balance that is progressively updated after each time step. Deposited sediment is allowed to build up the bed in each reach and change river slope. This process feeds back to and affects the transport rate, continuously changing system geomorphology. Where topographic data were sparse, a ‘‘representative reach’’ approach was taken in which short reaches in different sections

264

G. Pickup and Y. Cui

of the river were surveyed in detail. A sediment transport equation is then used to deﬁne sediment outﬂow for these reaches and any excess deposited upstream. Slope and channel geometry remain unchanged in the representative reaches throughout this process. Since the late 1980s, most work on the upper river system has used the OkGrav model, which applies a sediment routing approach, although some reaches are very long for this procedure. OkGrav has gone through many iterations, and the current version operates as follows. The river is assumed to be steep enough to allow quasi-normal ﬂow, so backwater equations are not used to predict ﬂow velocities and slope. Also, rather than using daily discharges as input, the ﬂows are computed for a number of points on a ﬂow duration curve. This curve is speciﬁed for each node along the system but assumed to be invariant in time. The model thus cannot distinguish between wet and dry years. The model focuses on gravel transport. It assumes three sources of gravel: natural gravel, gravel derived from rock waste from the mine, and gravel derived from landslide material. The model also tracks mine-derived sand and silt. Sand and silt derive from rock waste, slide material, and tailings. An additional source of sand and silt is abrasion (see below). Gravel transport is calculated using the Parker (1990) surface-based transport relation. It treats sediment mixtures ranging from 2 to 512 mm on a fractional basis. Three layers are assumed in the model: a bedload layer, a bed surface (active) layer, and a substrate layer. The average transport rate and transport grain size distribution for a year is based on calculations from the ﬂow duration curve. The gravel-sized rock waste is relatively easily abraded. The products of abrasion are computed to be silt and sand but mostly the former. The coefﬁcient of abrasion is assumed to be relatively large near the mine and relatively smaller farther downstream to reﬂect the gradual dominance of resistant gravel as the friable material is ground out. Although the model allows for different abrasion rates for several lithologies, only two lithologies have been used: harder natural gravel and softer mine-derived gravel. The basis for the abrasion calculations is given in Parker (1991a, b). Within the Mine Area Creek ﬂuvial zones (Ok Gilor–Ok Mabiong–Ok Tedi and Ok Mani), the bed is allowed to aggrade or degrade between any pair of nodes, according to whether or not the gravel supply exceeds the sediment delivery. The width of deposition is divided into a channel width that is used to compute sediment transport and a wider valley width over which aggraded sediment is allowed to deposit uniformly. This conﬁguration approximates the reworking of the valley ﬂat by a braided or wandering gravel-bed stream. The channel widths were determined from aerial

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

265

photography and adjusted in the ‘‘zeroing’’ procedure discussed below. Valley geometry was assumed to be V-shaped, so that valley width increases with bed aggradation. This was particularly important for the Ok Mani, which is now an order of magnitude wider than the pre-mine state. Valley shape was obtained from topographic maps. The Mine Area Creek deposition is assumed to consist gravel particles, but sand and some silt ﬁll the pore spaces as the bed aggrades. This material is released as the bed later incises into its own deposits. Exactly the same algorithm is used for the Middle Ok Tedi between the Ok Mani junction and Konkonda, with the exception that valley width is not allowed to change according to bed height. Because the river later degrades into its own deposits upon mine closure, it is necessary to store in memory both the thickness of the deposit and the grain size distribution. The stored grain size distributions consist averages of the deposit over a speciﬁed thickness. Sand and silt deriving directly from rock waste, slide material, and tailings are allowed to deposit in or be eroded from the Mine Area Creeks and the Upper Ok Tedi. They are neither deposited in nor eroded from the Middle Ok Tedi. The sand and silt load at the Ok Tedi–Ok Mani junction is delivered directly to Konkonda (gravel–sand transition) with no intervening storage. The sand and silt produced by abrasion in the Middle Ok Tedi is also delivered directly to Konkonda with no storage. This assumption was reasonable in the early days of mining, but there is now about 15–30 Mt of stored sand between Ningerum and Konkonda. It therefore proved necessary to extend the sediment transport model for the sand-bed reach of the Middle Fly upstream to account for this material. Models of bed evolution are prone to large errors if used without prudent adjustment. The model was adjusted in the following way. Assuming that no mine was ever built, gravel infeed and channel width were adjusted so as to predict a river system that changes a negligible amount over the planned lifetime of the mine. The natural gravel load was determined in this way. This load is very small compared with sediment delivery from the mine. 7.2.1.4. OkGrav model calibration/validation Most of the ﬁeld data for model validation came from Marshall (2001). They include:

Sediment deposition history in the Ok Mani and Harvey Creek. Erosion in Sulphide Creek between 1996 and 2000. Erosion in the Upper Ok Tedi between 1996 and 2000. The depositional history at selected cross sections.

266

G. Pickup and Y. Cui

The sediment deposition history in the Ok Mani and Harvey Creek is presented in Fig. 7.2. The model closely predicts the observed sediment deposition. The amounts of erosion in Sulphide Creek and on the Upper Ok Tedi between 1996 and 2000 are also closely predicted (Figs. 7.3 and 7.4). The depositional histories in cross sections TED44, TED50, TED54, and TED56 are presented in Fig. 7.5. Comparison between the model results and ﬁeld data indicates that the model correctly predicts the general trend and the magnitude of sediment deposition in three of the four cross sections, that is, TED44, TED50, and TED56. The model did not reproduce the general trend of bed degradation between 1999 and 2001 for cross section TED54. This discrepancy may be caused by the fact that ﬁeld data are for thalweg elevation while the model results represent the average bed elevation. A closer inspection of the survey data for TED54 revealed that only a few surveys covered the entire cross section, so the attempt to validate the model against average bed elevation failed. Also, the cross section is close to a valley constriction that may have resulted in localized scour at times. Field data at TED50 and TED56, which are located approximately 9.8 km upstream and 4.8 km downstream of TED54, respectively, all show

Cumulative Gravel, Sand and Silt Deposit (Mt)

250 Simulated total deposit in Ok Mani and Harvey Creek 200

150

Estimated total deposit in Ok Mani and Harvey Creek (read from Figure 6.10 in Marshall 2001) Estimated total deposit in Ok Mani and Harvey Creek (derived from tables in Section 5.1 in Marshall 2001)

100

50

0 1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

Year

Figure 7.2: Predicted and estimated sediment deposition in the Ok Mani and Harvey Creek.

Amount of Erosion* in Sulphide Creek - Ok Mabiong - Ok Gilor between 1996 and 2000 (Mt)

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

267

8

6

4

2

0 Estimate from DEM**

Simulation

Figure 7.3: Predicted and estimated sediment erosion in Sulphide Creek between 1996 and 2000. Data were estimated from digital elevation models and differ slightly from the values in Marshall (2001).

Amount of Erosion in Upper Ok Tedi between 1996 and 2000 (Mt)

8

6

4

2

0 Marshall's Estimate

Simulation*

Figure 7.4: Predicted and estimated sediment erosion in the Upper Ok Tedi (Ok Mabiong junction to Ok Mani junction) between 1996 and 2000. The model does not directly estimate ﬁne sediment deposition and erosion in this reach. The result presented here assumes 30% of ﬁne sediment in the deposit.

9

3 TED44

268

TED50

Incremental Elevation (m)

7

5

3

1 Field data

G. Pickup and Y. Cui

Incremental Elevation (m)

2.5 2 1.5 1 0.5 0

Field data Simulated results

Simulated results -1 1985

1987

1989

1991

1993

1995

1997

1999

2001

-0.5 1985

2003

1987

1989

1991

1995

1997

1999

2001

2003

6

5

TED56

TED54 5 Incremental Elevation (m)

4 Incremental Elevation (m)

1993

Year

Year

3

2

1

0 Field data Simulated results -1 1985

1987

1989

1991

1993

1995

Year

1997

1999

2001

2003

4 3 2 1 0 -1 1985

Field data Simulated results 1987

1989

1991

1993

1995

1997

1999

2001

2003

Year

Figure 7.5: Predicted and measured change in bed elevation of the Ok Tedi at cross sections TED44, TED50, TED54, and TED56. The predicted values are for average bed elevation; the ﬁeld data are for thalweg elevation.

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

269

continued aggradation between 1999 and 2001. With this in mind, we suggest that the discrepancy for cross section TED54 may be discounted.

7.2.2. Model Approaches to the Lower River System 7.2.2.1. System behavior The Lower Ok Tedi and much of the Middle Fly have a sand-bed channel that is partly conﬁned by terrace remnants in upper reaches but is free to meander over much of its length. The lower Middle Fly is affected by backwater from the Strickland and by tides, especially at low ﬂow. The backwater reach has silt on the bed for much of the time but still transports sand under some conditions. There is frequent ﬂooding when ﬂow goes overbank, but the ﬂoodplain does not always act as a conveyance system. Instead, standing water on the ﬂoodplain conﬁnes sediment-transporting ﬂow to the channel for much of the time. Sediment load in the Lower Ok Tedi and beyond has been reduced by dredging since 1998. The dredging operation extracts more than 15 Mt/year on average, most of which is sand.

7.2.2.2. Characteristic approaches: Lower Ok Tedi and Middle Fly In the early stages of the Ok Tedi project, little was known about the geomorphology of the Fly River system beyond the results of the CSIRO land systems survey (Blake and Ollier, 1971; Paijmans et al., 1971). Given this lack of information, two basic approaches were used to model the lower river system: in the Lower Ok Tedi and the Upper Middle Fly, a representative reach approach was taken, with reach nodes surveyed at Ningerum, Konkonda, and Kuambit. Transport rates were calculated using an early version of HEC-6 (Hydrologic Engineering Centre, 1992). Further downstream, it was assumed that mine waste would behave in the same way as natural load. Existing bed material sizes were then used to predict deposition, assuming that mine-derived sediment ﬁner than the material on the bed would pass downstream (Pickup et al., 1979; Vanoni and Henderson, 1979). Mine-derived sediment that was of the same size as the bed material would either be transported or deposited depending on whether the river had sufﬁcient capacity to transport the additional load. This approach is likely to underestimate deposition, especially where large increases in sediment load are involved. However, at this stage of the project, tailings were to be stored behind a dam, so only a limited increase in load was expected.

270

G. Pickup and Y. Cui

Later attempts to model the river were based on more extensive data. While many attempts have been made, two are of particular note since they illustrate widely differing approaches to lower river system modeling and inﬂuenced later approaches. Both, however, were still limited by the lack of data on river system geomorphology. Klohn-Crippen (1996) applied an early version of the MIKE11 model (Danish Hydraulic Institute, 1996) to the Lower Ok Tedi and the Middle Fly. MIKE11 is a one-dimensional hydraulic and sediment model capable of calculating water surface proﬁles as well as sediment transport and deposition. Channel slope and geometry were constructed on the basis of eight measured cross sections, and a number of scenarios were run, including various dredging rates and several major landslide scenarios. Sediment inputs from the upper river system were calculated using OkGrav. Little information was available to validate the model. The basic approach used by Klohn-Crippen (1996) has been followed by most of the modeling studies carried out since then. MIKE11 predicted extensive bed level rise throughout the system but with signiﬁcant recovery within the Middle Fly sand zone within 20–40 years of mine closure. Further downstream, the bed level increase was greater than that predicted in more recent modeling studies but with a shorter recovery time. However, backwater effects on the lower Middle Fly from the Strickland were not allowed for. Floodplain inundation, based on river water surface proﬁles and estimated ﬂoodplain height, was also calculated and found to increase downstream over time. However, once again, there was signiﬁcant recovery in the upper part of the system within several decades after mine closure. A second study in 1996 applied a hydrodynamic approach to river system modeling using the two-dimensional RMA hydraulic and sediment model (King, 1996). Hranisavljevic et al. (1996) constructed a two-dimensional ﬁnite element grid and attempted to model ﬂow and sediment transport both in the channel and on the ﬂoodplain. The grid consisted of a ﬁne mesh covering the channel and the meander belt and a coarser mesh covering the outer ﬂoodplain area. A trapezoidal channel was assumed, and topographic data for the ﬂoodplain came from 35 east–west laser proﬁles ﬂown between Kuambit and a point just downstream of Everill Junction. While reasonable estimates of in-channel deposition were obtained for the Ok Tedi, results were less satisfactory for the Middle Fly. Estimates of ﬂoodplain deposition were also made, and while the broad spatial pattern may look reasonable, it is clear that the lack of topographic data limited what could be achieved. While the two-dimensional model was an interesting experiment, its value was limited by the lack of data for calibration and validation at the time.

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

271

Also, the model did not adequately reproduce ﬂow patterns on the ﬂoodplain. Much of the ﬂoodplain is occupied by large bodies of standing water during ﬂood times and by inﬂow and outﬂow to and from off-channel water storages. The two-dimensional model shows ﬂoodplain ﬂows parallel to the channel or across meander loops. These patterns occur on the Fly but are not the dominant form. Floodplain ﬂow patterns are also made complex by the presence of pre-Holocene terrace remnants, which may divert water ﬂow and isolate some areas from the adjacent channel. The topographic data used to describe the ﬂoodplain did not capture many of these features. Data availability has improved greatly since the 1996 study, but even now, there are insufﬁcient data to justify going beyond a one-dimensional model of the river channel. There is also still a shortage of information on ﬂoodplain topography, ﬂow patterns, and hydraulic roughness, which makes it difﬁcult to justify the two-dimensional modeling needed to represent ﬂoodplain behavior. Subsequent modeling efforts have used a onedimensional approach. Parker et al. (1996) developed the customized OkPlain model to simulate the sand-bed reach of the Ok Tedi between Konkonda and D’Albertis Junction. Again, the upstream sediment supply was produced with the OkGrav model, and OkPlain simulates only the transport of ﬁne sediment downstream of the gravel-bed reaches. OkPlain employed a standard backwater formulation for ﬂow calculation and the Brownlie (1981) bed material load equation for sediment transport capacity calculation. One of the distinctive features of OkPlain is that it has an ad hoc ﬂoodplain deposition module to simulate the amount of overbank sediment deposition parameters that can be calibrated based on ﬁeld observations. In addition, the model tracks different types of sediment by size fraction. Experience with the OkPlain model indicated that setting the downstream boundary condition at D’Albertis Junction is inadequate because in-channel sediment deposition continues to progress downstream. With that, the model became OkFly and was expanded further downstream, and the lower boundary was set at Obo just upstream of Everill Junction. The model also included the Upper Fly River between D’Albertis Junction and Kiunga to account for water and sediment supply (Cui and Parker, 1999). OkFly results show river-wide aggradation, extending into the Upper Fly. The main depositional wedge forms in the Ok Tedi and progrades downstream during mine life and for many years after mine closure. The general conclusion from this modeling is that aggradation and ﬂoodplain inundation will continue to increase in the Fly River even after the mine closure as the sediment wedge progresses downstream, and that recovery of the system will be slow.

272

G. Pickup and Y. Cui

OkFly was used to evaluate various management scenarios before the completion of the most recent modeling efforts. Examination of dredging with OkFly indicates that the operation will result in some recovery in the sand-bed reach of the Lower Ok Tedi and that the effect will gradually shift downstream into the Middle Fly.

7.2.2.3. Lower Ok Tedi and Middle Fly modeling with HEC-6: The 2003 study HEC-6 (Hydrologic Engineering Centre, 1992) is a one-dimensional openchannel ﬂow numerical model designed to predict changes in river proﬁles resulting from scour and deposition over periods of years. The model was developed by the US Army Corps of Engineers and has been in use for more than 20 years. HEC-6 divides the river into a series of reaches whose geometries are described by measured cross sections. A continuous ﬂow record is partitioned into a series of steady ﬂows of variable discharge and duration. A water surface proﬁle is calculated for each ﬂow providing information on water surface elevation, ﬂow velocity, energy slope, width, and depth at each cross section. Potential sediment transport rates are calculated at each cross section from these ﬂow parameters and a sediment mass balance obtained for each reach, based on sediment inﬂow and outﬂow. Each cross section is then changed depending on the amount of scour or deposition. The ﬂow computations proceed to the next ﬂow in the sequence and are repeated with the updated channel geometry. Sediment calculations are performed for each particle size fraction in the load and allow for both sorting and bed armoring. HEC-6 represents a river system by a main stem plus a set of tributaries. Water and sediment ﬂow in at the upper end of the model with outﬂow at the lower end. Any difference between sediment inﬂow and outﬂow is accounted for by erosion or deposition. Inﬂow and outﬂow may also occur at tributary junctions. The layout used in this study is shown in Fig. 7.6. Inﬂow of mine waste at the upper end of the system was calculated using sediment-rating curves derived in OkGrav. However, OkGrav does not allow for the sand deposition that has occurred in the Konkonda–Ningerum reach, which may include 15–30 Mt of material. This was accounted for by extending HEC-6 upstream to overlap part of the lowest reach modeled in OkGrav. HEC-6 then produces about 25 Mt of sand deposition upstream of Konkonda. Inﬂows of natural load must also be provided for the main stem and the various tributaries. While various estimates have been made of the natural

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

Iogi

MIDDLE OK TEDI

Overlap with OKGRAV Konkonda

273

Natural load & mine waste inflow from OKGRAV

OK MART Natural load inflow

UPPER FLY Natural Load inflow

MIDDLE FLY

Manda

AGU No net inflow of sediment Inflow & outflow of water

EVERILL JUNCTION Sediment outflow to Lower Fly

Figure 7.6: HEC-6 model layout showing main inﬂows and outﬂows. rate of sediment delivery to the Lower Ok Tedi and from the Upper Fly (Higgins et al., 1987), amounts vary substantially. In the absence of anything better, sediment inputs from the Ok Mart and the Upper Fly were taken from GHD (1999). However, a 4-km reach of the Upper Fly was used as a buffer zone between Kuambit and the Upper Fly sediment input point in HEC-6 to allow for backwater. The natural load of the Ok Tedi was estimated during model ‘‘zeroing’’ runs (see below) in which natural sediment input was gradually increased to the point where there was little or no net scour in the 1984 reconstructed proﬁle over a 13-year period. Sediment load particle size data came from riverbed samples collected before mine construction began (Pickup et al., 1979). Channel cross sections for the Lower Ok Tedi and the Middle Fly came from survey data provided by A. R. Marshall. The December 1997/March 1998 survey was used as a baseline while the March 2002 survey provided an endpoint for the model calibration/validation runs. The 1997 survey was also used to reconstruct the pre-mining river geometry. Roughness coefﬁcients for ﬂow proﬁle calculations were derived by comparing model results with observed water surface proﬁles (Pickup, 2001).

274

G. Pickup and Y. Cui

Flow data for modeling may be in the form of a ﬂow duration curve or a sequence of discharges. All model runs described in this chapter used a sequence of observed daily discharges for 1985–2002. Discharges for future periods were generated by repeating this record.

7.2.2.4. HEC-6 model calibration/validation HEC-6 offers a choice of sediment transport equations for clay, silt, sand, and gravel. There are also different methods for dealing with erosion and deposition of material in the silt and clay fractions. Most other parameters can be directly measured or calculated from stream, gauging data, and bed material samples. Once transport equations and other parameters have been selected, the model should be validated from observed data where possible. On the Ok Tedi and the Middle Fly, there are enough data to allow the following tests of model accuracy: Model ‘‘zeroing’’ using a crude reconstruction of riverbed levels in late 1984 from a small number of long-term cross sections, including Konkonda, Kuambit, Wygerin, and Obo. The aim of model zeroing is to show that the river does not experience signiﬁcant erosion or deposition over time with estimated natural sediment loads. It may also be used to estimate natural sediment loads where they are uncertain. A test of whether the model can reproduce 1985–1997 deposition on the Lower Ok Tedi and the Middle Fly, using the reconstruction of 1984 riverbed levels. A test of whether HEC-6 can reproduce the complex pattern of erosion and deposition that developed after the introduction of dredging at Bige on the Ok Tedi in 1998. Since then, more than 120 cross sections have been surveyed to a very high standard, up to twice a year, along the Lower Ok Tedi and the Middle Fly. 1997–2002 included a drought and a very wet period, both of which affected river behavior and created a challenging test of the model. The ﬁrst and second tests of model accuracy are not particularly rigorous because only eight cross sections can be used to estimate changes over more than 400 km of river. The third test provides a good indication of what level of accuracy the model can achieve but covers only a short time period. Model calibration runs showed that none of the sand transport equations in HEC-6 produced transport rates great enough to move sand down into the

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

275

Middle Fly in sufﬁcient quantity to reproduce observed deposition rates. Instead, deposited material built up in the Lower Ok Tedi. The greatest concentration of deposition was at the upper end of the reach between Iogi and Konkonda and produced increases in riverbed and water levels far higher than those observed in the reach. Underestimation of transport rates is common when using sediment transport equations in heavily laden rivers, and models frequently require calibration. It is possible to calibrate HEC-6 by adding user-derived sediment discharge-rating curves for each particle size range used in the model. However, insufﬁcient data are available to develop suitable empirical curves for the Fly. An alternative approach to calibrating HEC-6 is to rescale the time interval used in the model, which, in effect, increases the transport rates and moves sediment down the system more quickly. This approach proved effective, and the model was run using the Yang transport equation speeded up by a factor of 8. Model zeroing and calibration runs produce close agreement between modeled and observed river behavior. Model calculations in the zeroing run (Fig. 7.7) show slight erosion immediately below Iogi and slight deposition in the Lower Ok Tedi. The upper Middle Fly above Wygerin is close to being stable, but downstream, deposition gradually increases with both distance and time, with about 60 Mt accumulating in the Manda–Obo reach. This is the Strickland backwater zone and, in both pre-mining and current riverbed sediment surveys (see Bolton et al., 2009), showed a marked reduction in the amount of medium sand present. This is consistent with a river experiencing deposition. The deposition in the ﬁrst 50 km at the downstream end of the lower Middle Fly is affected by the model boundary condition and may be inaccurate. The overall pattern of deposition in the Middle Fly resembles a deposition sequence suggested by Pickup (1984). In this scheme, the downstream edge of the Middle Fly sand zone, which extends from D’Albertis Junction to Manda, slowly moves downstream as a front, into the backwater zone. Behind this front, the bed slowly rises to maintain slope, but this effect dissipates upstream. The vertical striping shows the effect of local variability in the river channel characteristics. This variability takes the form of channel openings, where sediment accumulates, and constrictions, which hold back material in upstream reaches but have limited local deposition. The vertical line marked ‘‘A’’ shows an extreme example of this behavior. The openings and constrictions are often associated with outcrops of resistant ‘‘red beds,’’ where the pre-Holocene ﬂoodplain survives in elevated terrace remnants. The vertical striping may be broken up into repeating cells of higher and lower deposition (e.g., the middle part of the vertical line marked ‘‘B’’).

276

G. Pickup and Y. Cui

Erosion/Deposition Mt/km 2060

-1.5 -1.0 -0.5 0.0 0.5 1.0

2050

2040 B Note: Erosion is indicated by negative values

2030

2020

A

2010

2000

1990 1985 0

100

200

300

400

IOGI

Distance Upstream from Everill Junction km MIDDLE FLY OBO

BOSSET MANDA

OK TEDI MABADUAM WYGERIN

D'ALBERTIS JCT KUAMBIT NIUKAMBA

Figure 7.7: HEC-6 model zeroing results. Erosion and deposition on the Lower Ok Tedi and the Middle Fly underestimated natural sediment load. These result from periods of high and low discharge such as La Nina and El Nino sequences, several of which were built into the input data. The case for rising bed levels under natural sediment loading is summarized by Dietrich et al. (1999). It includes: The medium sand fraction present in the riverbed material of the upper Middle Fly did not seem to reach the lower Middle Fly prior to mining (Pickup et al., 1979).

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

277

An apparent rise in the level of the Strickland, backing up the lower Middle Fly and virtually halting meander migration in the lower Middle Fly because of lower bank shear stress values associated with reduced slope. Possible subsidence associated with a fault that crosses the Middle Fly ﬂoodplain. This may explain why terrace remnant elevations decrease down the Middle Fly but rise again in the Lower Fly (Pickup, 2001). The Middle Fly channel levee system may be up to 7 m higher than the surrounding ﬂoodplain.

Estimates of the aggradation rate vary, but Dietrich et al. (1999) suggest a rate of 0.1–1 mm/year for the zone away from the channel, 1–10 mm/year for ﬁlling of meander cutoff oxbows, and perhaps 5 mm/year for the channel system. However, the channel system aggradation rate may be lower because some sand is transported out of the channel by overbank ﬂow and through tie channels. These rates may be conservative. For example, a detailed stratigraphic study of Magela Creek in the uranium province of Australia’s Northern Territory shows that the whole ﬂoodplain has risen at an average rate of about 4 mm/year since the post-glacial rise in sea level ended 5,000–7,000 years ago (Clarke et al., 1992). Magela Creek is much smaller than the Fly but would have a similar history of post-glacial sea level rise. It is also backed up by the much larger East Alligator River, so the situation is analogous to the Middle Fly and the Strickland. However, the Magela Creek catchment is much ﬂatter and more geologically stable than that of the Fly, so delivered sediment load would be less. It therefore seems reasonable to expect a higher natural rate in the Middle Fly. In fact, the 70-year modeling run suggests a natural accumulation of about 10 mm/year if all deposited sediment stays in the channel. If 30% is lost to the ﬂoodplain, the rate comes down to 7 mm/year, which is close to the Dietrich et al. (1999) estimate of 5 mm/year. This suggests that, though our estimates of deposition under natural sediment load are high, they are not unreasonable. The reconstructed 1985 river channel therefore produces a reasonable pattern of modeled river behavior. After model zeroing, a number of calibration runs were carried out using model runs with natural load, tailings, and waste rock input for 1985–1997. There are two major sources of uncertainty in these model runs. First, the 1984 bed reconstruction may contain substantial errors. Second, output of sand and silt from OkGrav may not always be accurate since there are not enough data to calibrate sand storage and transport at the downstream end of the model. In spite of these uncertainties, HEC-6 should still reproduce

278

G. Pickup and Y. Cui

the general pattern of deposition in the river. The water surface proﬁle should also change to reﬂect rising bed levels. The results are good by river modeling standards. Rising bed levels upstream of the 450-km point are closely reproduced. The model is slightly less successful on the Lower Ok Tedi, and the 1997 bed levels close to D’Albertis Junction (429 km) are underestimated by about 1 m. There is also a tendency to underestimate 1997 bed levels on the upper part of the Middle Fly between 300 and 410 km, but the average is less than 0.5 m. Given uncertainties in the input data and less-than-optimal cross section spacing, this is an acceptable degree of discrepancy. Bed level change in the 0- to 300-km reach (lower Middle Fly) largely consists proﬁle smoothing as holes are ﬁlled in and some constrictions are reduced by local scour. There is also a buildup of material close to the downstream end of the model where the backwater effect from the Strickland is greatest. While these results seem reasonable, available data are not good enough for a convincing test of model performance. Little is known about tributary inﬂows and outﬂows in this reach, and data from only three cross sections were available for the 1984 river reconstruction. Overall, however, results seem reasonable. The pattern of observed and modeled changes in the water surface proﬁle is broadly similar (Fig. 7.8). However, there are some differences in particular reaches. The modeled water surface proﬁle above 450 km is very close to the proﬁle calculated from 1997 cross sections (labeled as 1997 Observed in Fig. 7.8). In the 300- to 450-km reach, it is about 1 m too low. A difference of this magnitude over 150 km is not great, given that bed levels have risen by about 4 m. Below 300 km, the proﬁles are very similar. Comparison of observed and modeled deposition in the 1985–1997 and 1985–2002 calibration periods shows reasonably good results, given that the reconstructed 1985 proﬁle could contain errors. However, changes to the upper 300 km of the river system upstream of Mabaduam are modeled more closely than the lower river. Also, there seems to be a tendency to underestimate deposition in the Fly based on the 2002 proﬁle. A more detailed test is to compare modeled changes between 1997/1998 and 2002 with observed changes from cross section surveys (Fig. 7.9). This is quite a severe test of model performance because the cross section surveys cover a period with both wet and dry phases as well as more average years. Also, sediment inputs from OkGrav would not represent the wet and dry periods very well because it uses discharges derived from long-term average ﬂow duration curves. The 1997/1998–2002 comparisons show that the impact of dredging upstream of the 400-km point is reproduced very closely. However, the

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

10

25 20 15

279

1984 ESTIMATE 1997 OBSERVED 1997 MODEL

1984 ESTIMATE 1997 OBSERVED 1997 MODEL

0

10 -10 5 0

-20

-5 -10 300

(a)

-30 350

400

450

500

25

20

0

(c)

Distance from Everill Junction (km)

150

200

250

300

1984 ESTIMATE 1997 OBSERVED 1997 MODEL

15

10

10

5

(b)

100

20 1984 ESTIMATE 1997 OBSERVED 1997 MODEL

15

5 300

50

Distance from Everill Junction (km)

0 350

400

450

Distance from Everill Junction (km)

500

0

(d)

50

100

150

200

250

300

Distance from Everill Junction (km)

Figure 7.8: 1984–1997 bed levels and water surface proﬁles (10% ﬂow duration) with natural load, waste rock, and tailings input. modeling results show that the main wave of deposition extends as far as the 300-km point (upper right graph) while the survey data suggest that it has moved 100 km further downstream (upper left graph). If the survey data are divided into El Nino/wet period conditions and closer-to-average ﬂow conditions, two different patterns emerge (see Pickup and Marshall, 2009). Between 1997 and 1999, sediment supply was reduced by low ﬂows in the Ok Tedi and interruption to mine operations. A wet period immediately followed the El Nino. In this phase, deposited material in the upper Middle Fly moved down to the upper end of the backwater zone where it seems to have lodged. With a return to more average conditions, the main wave of deposition ends close to 300 km, as the model predicts. A secondary wave of deposition at the downstream end of the river is also shown in both observed and modeled results. These results suggest that the model performs reasonably well under average ﬂow conditions but may not deal with extreme variability in the short term. The 1997 drought, incidentally, had the longest period of low

280

G. Pickup and Y. Cui

20

1.0

OBSERVED 1997/8-2002

0.5

15

0.0

10

-0.5

5

1984 ESTIMATE 1997 OBSERVED 2002 OBSERVED 2002 MODEL

0

-1.0 0 (a)

100

200

300

400

500

0 (c)

Distance from Everill Junction (km)

100

150

250

200

300

25

1.0 MODEL 1998-2002 0.5

20

0.0

15

-0.5

10

-1.0 0 (b)

50

Distance from Everill Junction (km)

100

200

300

400

Distance from Everill Junction (km)

500

5 300 (d)

1984 ESTIMATE 1997 OBSERVED 2002 OBSERVED 2002 MODEL

350

400

450

500

Distance from Everill Junction (km)

Figure 7.9: Observed and modeled results for 1997/1998–2002. The upper graphs show changes in bed level. The lower graphs show water surface proﬁles. ﬂows since 1984 and possibly 1972, according to Fly River gauge height graphs contained in SMEC (1978). Water surface elevation results are closer to observed river behavior than the bed level simulations. The pattern of change between 1998 and 2002 is almost identical, and by 2002, the proﬁles are almost identical over the whole of the river. This is a very good result, given that we are attempting to predict changes in elevation of only a few meters over almost 500 km of river.

7.3. Predictions of Future Deposition, Water Level Change, and Inundation Assuming mine closure in 2010 and dredging at the current rate, patterns of deposition and water level change are shown in Figs. 7.10 and 7.11.

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

281

Erosion/Deposition Mt/km 2060

-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0

2050

2040

2030

2020

Note: Erosion is indicated by negative values

C

MINE CLOSURE

2010

2000

DREDGING BEGINS

1990 1985 0

100

200

300

400

Distance Upstream from Everill Junction km MIDDLE FLY OBO

BOSSET MANDA

MABADUAM WYGERIN

IOGI

OK TEDI D'ALBERTIS JCT KUAMBIT

NIUKAMBA

Figure 7.10: Predicted deposition of natural and mine-related sediment in the Lower Ok Tedi and the Middle Fly between 1985 and 2064, assuming mine closure in 2010 and dredging at the current rate. At its simplest, the pattern of deposition may be seen as a ﬂow of material with a wave front that gradually moves down the system over time. This behavior shows as a prominent diagonal break between the green (little or no deposition) feature in the lower left of the image and the red and yellow (greater deposition) zone. The ﬂow attenuates with distance and with time once mining ceases. It also breaks up into separate cells in the lower part of the Middle Fly but leaves behind long-term deposits in many reaches. Initially, heavy sediment accumulation occurs in the upper part of the Ok Tedi between Iogi and D’Albertis Junction. This deposition is reduced by

282

G. Pickup and Y. Cui

Water Level Change (m) 2060

0.0-0.5 0.5-1.0 1.0-1.5 1.5-2.0 2.0-2.5 2.5-3.0 3.0-3.5 3.5-4.0 4.0-4.5 4.5-5.0 5.0-5.5 >5.5

2050

2040

2030

2020 MINE CLOSURE

2010

2000

DREDGING BEGINS

1990 1985 0

100

200

300

IOGI

400

Distance Upstream from Everill Junction km MIDDLE FLY OBO

BOSSET MANDA

OK TEDI D'ALBERTIS JCT

MABADUAM WYGERIN

KUAMBIT NIUKAMBA

Figure 7.11: Predicted water level change between 1985 and 2064 in the Lower Ok Tedi and the Middle Fly, assuming mine closure in 2010 and dredging at the current rate. dredging and natural erosion after mine closure. By about 2060, much of this deposition has cleared, and some of the irregularities in the bed that were ﬁlled during the lead-in period to stabilize the model begin to reappear. The Lower Ok Tedi holds material longer, and deposition persists in some reaches to at least 2060. The bulk of deposition initially occurs on the Middle Fly in the 240- to 420-km reach. Some, but not all, of this material shifts downstream after mine closure. There is some clearing of deposited material from the upper

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

283

70–100 km after about 2040, but deposition increases further downstream and shows little sign of clearing by 2060. There are local buildups of sediment in the lower part of the Middle Fly after about 2020. These show as short, vertical red stripes in the top left-hand part of the ﬁgure. There is also a prominent scour zone marked ‘‘C,’’ where a scour hole buried during the pre-mining lead-in period reappears after 2020. While modeled behavior in the lower reaches of the Middle Fly appears logical, little is known about water inﬂows and outﬂows in this reach, so questions of accuracy remain. The downstream end of the system is also affected by the assumption of a stable-rating curve at Obo as a boundary condition. There is no evidence of rating shift between 1987 and 2001 (Pickup, 2001), but data quality is poor. Water level changes tend to be more regular than raw deposition data because rises in water level affect upstream reaches and smooth out local increases in bed level. Water level increase is greatest at the upstream end of the system but decreases rapidly after dredging and mine closure. Further downstream, water levels continue to rise until about 2060 when the system stabilizes. Water level changes decrease with distance downstream. However, by 2060, they are still signiﬁcantly higher than the pre-mining situation over most of the Middle Fly and all of the Lower Ok Tedi. 7.3.1. Deposition on Channel Levees and the Floodplain Predicting deposition rates on the ﬂoodplain remains a problem for sediment modeling. Deposition occurs on the ﬂoodplain surface, in the tie channels that connect the main river to ﬂooded meander cutoffs, in the cutoffs themselves, and on main channel and tie channel levees. Using early data on copper distribution in sediments, Dietrich (2000); Dietrich and Day (2004) has shown that the bulk of deposition occurs on main channel and tie channel levees and decreases exponentially with distance from the channel across the ﬂoodplain surface. There is also the potential to increase deposition in the off-river water bodies. However, as Day and Associates (1996) notes, most are not under threat of inﬁlling. Indeed, Day estimated that the larger off-river water bodies might lose 1–2% of capacity while smaller ones might lose up to 10% over the life of the mine. These calculations apply to oxbow lakes in meander cutoffs rather than the much larger blocked valley lakes. Rowland et al. (2005) looked at deposition rates in off-river water bodies through extension of tie channels and their levees into cutoffs. They conclude that tie channel extension is the dominant process of off-river water body

284

G. Pickup and Y. Cui

inﬁlling and that rates have increased 6–17 times since 1980. They also concur with Day and Associates (1996) that most off-river water bodies are not in danger of inﬁlling. The pattern of ﬂoodplain deposition can be inferred from airborne gamma ray data acquired during drought [see Pickup and Marks (2000, 2001) for a description of how gamma ray emissions can be linked to deposition]. Figure 7.12 shows a typical pattern from a meander on the lower Middle Fly with the gamma ray data in the left-hand image and panchromatic LANDSAT coverage of the same area on the right. The concentration of

Figure 7.12: Total gamma emissions from an airborne gamma ray survey of meanders on the lower Middle Fly during drought conditions when the ﬂoodplain was virtually dry. Emissions are highest in the red areas and decrease through yellow, green, light blue, dark blue, and purple. The LANDAT ETM image on the right shows ground features in more detail. High emissions are thought to indicate freshly deposited mine sediment. The low-emission areas behind the levees indicate sediments with a high organic content, but some gamma ray emission may be blocked by the presence of soil–water in these areas.

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

285

Increase in Levee Height (m)

4

3

2

1

0 2000

2020

2040

2060

2080

Figure 7.13: Modeled increases in levee height for selected reach locations on the Middle Fly after 1997. material on the levees and levee backslopes can be clearly seen. There is also a zone of heavier deposition along a tie channel. The most recent effort to predict ﬂoodplain deposition was conﬁned to the main channel levees where recent surveys have shown about 1 m of deposition on average (see Pickup and Marshall, 2009). Pickup and Cui (2003) constructed a simple model relating deposition to mean sediment concentration the main river and the amount time ﬂow was overbank. The model was calibrated from survey data and shows that the levees will continue to increase in height while water levels on the Fly continue to rise, even after disposal of mine waste has ceased (Fig. 7.13). The amount of increase decreases downriver as the depth of inundation over the levees decreases. This results from the gradual damping of ﬂood waves downstream (see Pickup and Marshall, 2009).

7.4. Future Paths The modeling techniques currently in use are close to their limits of accuracy, given available input data, sediment transport technology, and capacity for model calibration. Future advances are most likely to come from new datasets that remove bottlenecks and open up new opportunities for

286

G. Pickup and Y. Cui

prediction. There will also be changes in emphasis as the need for prediction of system behavior after mine closure increases. In the upper river system, further progress is limited by the lack of detailed topographic data showing deposition rates over time for the Upper and Middle Ok Tedi. OkGrav is reasonably well calibrated for deposition but largely untested for river behavior after mine closure when the system will begin to scour. This process may leave sections of the current riverbed as terraces, which will be exposed to more rapid weathering than at present. Future modeling activities in the upper river system are therefore increasingly likely to concentrate on post-mining river behavior. In the lower river, any improvement to sediment transport calculations is likely to be incremental. Information about water losses to the ﬂoodplain and backﬂow to the channel was a signiﬁcant data gap in the 2003 modeling exercise. What lower boundary condition to use in modeling is also a problem, given the complex interaction between the Fly and the Strickland and tidal effects. Some of these issues have been addressed in a recent study of ﬂoodplain hydraulic behavior (see Pickup, 2009). However, the improvements have yet to be incorporated in a sediment transport model. On the ﬂoodplain, there is still no satisfactory approach to predicting the amount and location of deposition. While we know what the spatial pattern is and the general form it takes, a predictive model will need extensive calibration from observed data. Airborne gamma ray surveys provide the easiest method of acquiring these data but can only be usefully undertaken on rare occasions when the ﬂoodplain is dry.

REFERENCES Blake, D. H., & Ollier, C. D. (1971). Alluvial plains of the Fly River, Papua. Zeitschrift fur Geomorphologie Supplementband, 12, 1–17. Bolton, B. R., Pile, J. L., & Kundapen, H. (2009). Texture, geochemistry and mineralogy of sediments of the Fly River System. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 51–112. Brownlie, W. R. (1981). Prediction of ﬂow depth and sediment discharge in open channels. Report No. KH-R-43A, W. M. Keck Laboratory of Hydraulics and Water Resources, California Institute of Technology, Pasadena, California, USA, 232 pp. Clarke, R. L., Guppy, J., Mahon, D., McBride, P., & Wasson, R. J. (1992). Late quaternary evolution of the Magela plain. In: Wasson, R. J. (Ed.). Modern Sedimentation and Late Quaternary Evolution of the Magela Creek Plain.

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

287

Australian Government Publishing Service, Canberra, pp. 81–157. Supervising Scientist for the Alligator Rivers Region Research Report 4. Cui, Y., & Parker, G. (1999). Sediment transport and deposition in the Ok Tedi-Fly River system, Papua New Guinea: The modeling of 1998–1999. St. Anthony Falls Laboratory, University of Minnesota, 27 pp. Danish Hydraulic Institute (1996). MIKE 11. A microcomputer based modeling system for rivers and channels. Danish Hydraulic Institute. Day, G., & Associates (1996). Particle size and shape characteristics of deposited sediments in the Middle Fly River and sediment deposition rates in the tie channels and off-river water bodies. Volume 1. Ok Tedi Mining Ltd., Environment Department. Dietrich, W. E. (2000). Quarterly and fourth monthly report – Middle Fly sedimentation project. Department of Earth and Planetary Science, University of California, Berkeley, 51 pp. Dietrich, W. E., & Day, G. (2004). Rates and patterns of ﬂoodplain sedimentation, middle ﬂy ﬂoodplain (1985–1994). Presentation at the Sediment Transport Model Workshop, Cairns, Australia. January, 2004. Environment Department, OK Tedi Mining Ltd. Dietrich, W. E., Day, G., & Parker, G. (1999). The Fly River, Papua New Guinea: Inferences about river dynamics, ﬂoodplain sedimentation and fate of sediment. In: A. Miller, & A. Gupta (Eds). Varieties of Fluvial Form. Wiley, New York, pp. 345–376. Foster, D. N., Cox, R. J., Miller, C. A., & Wallace, B. C. (1984). Ok Mani diversion scheme – weir, intake channel and tunnel. University of New South Wales Water Research Laboratory Technical Report 84/07, 42 pp. GHD (1999). Hec6 model for Ok Tedi and Fly River. Ok Tedi Mining Ltd. Hranisavljevic, D., King, I. P., Chadwick, M. J., Nittim, R., & Cox, R. J. (1996). Ok Tedi and Fly Rivers. Hydrodynamic and sediment transport modeling. University of New South Wales Water Research Laboratory Technical Report 96/15, 59 pp. Higgins, R. J. (1979). Sediment transport in a river with high induced load. Civil Engineering Transactions, Institution of Engineers Australia, CE21, pp. 111–117. Higgins, R. J., Pickup, G., & Cloke, P. S. (1987). Estimating the transport and deposition of mining waste at Ok Tedi. In: C. R. Thorne, J. C. Bathurst, & R. D. Hey (Eds). Sediment Transport in Gravel Bed Rivers. Wiley, Chichester, UK, pp. 949–976. Hydrologic Engineering Centre (1992). HEC-6 scour and deposition in rivers and reservoirs. Users manual, US Army Corps of Engineers. King, I. (1996). Program Documentation RMA-2: Finite Element Model for Flow Estimation in Rivers and Streams. Resource Management Associates, Suisin, CA. Klohn-Crippen (1996). Waste management study. Ok Tedi and Fly River model report. Ok Tedi Mining Ltd., 2 vols. Klohn Leonoff (1993). Ok Mani diversion and waste storage. Ok Tedi Mining Ltd., Appendix IV. Marshall, A. R. (2001). Upper Ok Tedi sediment storage analysis – November 2000. Environment Department, Ok Tedi Mining Ltd., 25 pp.

288

G. Pickup and Y. Cui

Maunsell and Partners Pty. Ltd. (1982). Ok Tedi environmental study. Report to Ok Tedi Mining Ltd., Tabubil, 7 vols. Paijmans, K., Blake, D. H., Bleeker, P., & McAlpine, J. R. (1971). Land Resources of the Morehead-Kiunga Area, Territory of Papua and New Guinea. CSIRO, Melbourne, Australia, Land Research Series 29. 124 pp. Parker, G. (1990). Surface-based bedload transport relation for gravel rivers. Journal of Hydraulic Research, 28(4), 417–436. Parker, G. (1991a). Selective sorting and abrasion of river gravel: Theory. Journal of Hydraulic Engineering, 117(2), 131–149. Parker, G. (1991b). Selective sorting and abrasion of river gravel: Applications. Journal of Hydraulic Engineering, 117(2), 150–171. Parker, G., & Dietrich, W. A. (1993). Numerical modeling of the Klohn Leonoff waste control scheme for the Ok Mani, Papua New Guinea. In: Klohn Leonoff. Ok Mani Diversion and Waste Storage. Ok Tedi Mining Ltd., Appendix IV. Parker, G., Cui, Y., & Imran, J. (1996). Flooding in the lower Ok Tedi, Papua New Guinea due to the disposal of mine tailings and its amelioration. International Seminar on Recent Trends of Floods and their Preventive Measures, Hokkaido River Disaster Prevention Research Center, June 20–21. Pickup, G. (1978). Report on environmental aspects of Ok Tedi project waste disposal. Australian National University North Australia Research Unit, Darwin, 15 pp. Pickup, G. (1980). Hydrologic and sediment modeling studies in the environmental impact assessment of a major tropical dam project. Earth Surface Processes, 5, 61–75. Pickup, G. (1984). Geomorphology of tropical rivers. I. Landforms, hydrology and sedimentation in the Fly and lower Purari, Papua New Guinea. Catena Supplement, 5, 1–18. Pickup, G. (2001). Geomorphic processes in the Ok Tedi and Fly River system: Current status, future trends and rehabilitation. Report to Ok Tedi Mining Ltd., 62 pp. Pickup, G. (2003). Effects of mine life extensions and rates of dredging on the Ok Tedi and Fly River. Uncertainty analysis for the null scenario. Report to Environment Section, Ok Tedi Mining Ltd., 38 pp. Pickup, G. (2005). Floodplain inundation modeling and forecasting for the Middle Fly. Report to Environment Section, Ok Tedi Mining Ltd., 88 pp. Pickup, G. (2009). Floodplain inundation modeling and forecasting for the Middle Fly. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 291–318. Pickup, G., & Higgins, R. J. (1979). Estimating sediment transport in a braided gravel channel – the Kawerong River, Bougainville, Papua New Guinea. Journal of Hydrology, 40, 283–297. Pickup, G., & Marks, A. (2000). Identifying large-scale erosion and deposition processes from airborne gamma ray remote sensing and digital elevation

Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System

289

models in a weathered landscape. Earth Surface Processes and Landforms, 25, 535–557. Pickup, G., & Marks, A. (2001). Regional-scale sedimentation process models from airborne gamma ray remote sensing and digital elevation data. Earth Surface Processes and Landforms, 26, 273–279. Pickup, G., & Cui, Y. (2003). Effects of mine life extensions and rates of dredging on the Ok Tedi and Fly River – A sediment transport model study using Okgrav6 and HEC-6. Report to Environment Department, Ok Tedi Mining Ltd., 113 pp. Pickup, G., & Marshall, A. R. (2009). Geomorphology, hydrology and climate of the Fly River System. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River system, Vol. 9. Elsevier, Amsterdam, pp. 3–49. Pickup, G., Higgins, R. J., & Warner, R. F. (1979). Impact of waste rock disposal from the proposed Ok Tedi mine on the Fly River and its tributaries. Department of Minerals and Energy and Ofﬁce of Environment and Conservation, Papua New Guinea, 139 pp. Rowland, J. C., Lepper, K., Dietrich, W. E., & Wilson, C. (2005). Tie channel sedimentation rates, oxbow formation age, and channel migration rate from Optically Stimulated Luminescence (OSL) analysis of ﬂoodplain deposits. Earth Surface Processes and Landforms, 30(9), 1161–1179. SMEC (1978). Hydrologic investigations for the Ok Tedi project. Report prepared for Dampier Mining Co. Ltd., Snowy Mountains Engineering Corporation, 115 pp. Vanoni, V. A., & Henderson, F. M. (1979). The effect of mine sediments on the Ok Tedi-Fly River system. Ok Tedi feasibility study. Volume 2 – Environment. The Ok Tedi Consortium, pp. 74–121. Wright, S., & Parker, G. (2005a). Modeling downstream ﬁning in sand-bed rivers, I: Formulation. Journal of Hydraulic Research, 43(6), 612–619. Wright, S., & Parker, G. (2005b). Modeling downstream ﬁning in sand-bed rivers, II: Application. Journal of Hydraulic Research, 43(6), 620–630.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00408-4

Chapter 8

Floodplain Inundation Modeling and Forecasting for the Middle Fly Geoff Pickup Consulting Geomorphologist, 1538 Sutton Road, Sutton, New South Wales 2620, Australia

8.1. Introduction Modeling of deposition in the Ok Tedi and Fly River systems (Cui and Parker, 1999; Pickup and Cui, 2009) has shown that dredging will slow or halt bed level rise in the lower Ok Tedi and upper Middle Fly. However, there is still a large body of sediment that was downstream of the dredge site at Bige by the time dredging operations began in 1998. Channel cross-section surveys show that this material continues to move downstream (Marshall, 2002) and modeling indicates that it will produce a signiﬁcant rise in water levels in future years. Given the low gradient of the Middle Fly and the large area of ﬂoodplain, higher water levels will increase the extent and frequency of ﬂoodplain inundation. The Pickup and Cui sediment model results presented elsewhere in this volume have limitations because they do not allow for processes such as downstream ﬂood-wave attenuation, ﬂows in and out of off-river water bodies, local rainfall, tidal variations, and backwater effects from the Strickland and major tributaries such as the Binge and the Agu (Pickup, 2003). Some of these processes may reduce the amount by which water levels will rise. Capacity to predict changes to the amount and frequency of ﬂoodplain inundation has been limited by lack of topographic data. Apart from widely separated laser proﬁler transects ﬂown in 1995, little was known about ﬂoodplain topography at the resolution needed for inundation mapping and Corresponding author. Tel.: +61 26238 3427

E-mail: [email protected] (G. Pickup).

292

G. Pickup

modeling. It was therefore not feasible to calculate the volume of ﬂoodplain storages such as the many large off-river water bodies or to identify ﬂoodplain ﬂow paths that link these storages to the main river. It was also not possible to model the lateral extent of ﬂoodplain inundation. In late 2003, NASA released a digital elevation model (DEM) from the Shuttle Radar Topography Mission (SRTM) covering Papua New Guinea. This DEM puts modeling of ﬂoodplain inundation a step closer. However, with a stated accuracy of only 5–7 m in elevation and limited rainforest canopy penetration by the radar, the SRTM DEM is not good enough for inundation modeling in its raw form. Several other developments have made it possible to correct the SRTM DEM data to a point where ﬂood inundation modeling becomes feasible. First, a body of remotely sensed data from the LANDSAT Thematic Mapper has accumulated since 1997. These data may be used to map the extent of ﬂoodplain inundation over a range of river heights on the few occasions where cloud cover is limited. Second, the European Space Agency ENVISAT satellite carries a high resolution, synthetic aperture radar that allows imaging of ﬂooded areas through cloud. These datasets have been used together with water surface proﬁles derived from river gauge heights to adjust SRTM DEM elevations so that they reproduce observed ﬂooding behavior more closely. This chapter describes the development of a ﬂoodplain inundation model based on corrected SRTM data. The model incorporates unsteady ﬂow down the main river channel, inﬂows and outﬂows to off-river water bodies, and backwater effects from the Strickland. Tests of model accuracy show that it is capable of producing water surface proﬁles that vary with time as ﬂoods pass down the Middle Fly and off-river water bodies ﬁll and empty. They also show that the model can reproduce observed extents of ﬂoodplain inundation. The model has been used with predicted channel deposition rates from Pickup and Cui (2009) and shows how ﬂoodplain inundation increases with time as mining waste passes downriver.

8.2. Digital Elevation Data for the Floodplain – the Shuttle Radar Topography Mission (SRTM) 8.2.1. Background SRTM was the primary payload on the STS-99 Mission of the Space Shuttle Endeavour which launched on February 11, 2000 and ﬂew for 11 days

Floodplain Inundation Modeling and Forecasting for the Middle Fly

293

covering most of the Earth’s land area (Farr and Kobrick, 2000). SRTM data spacing is either 1 arc-second or 3 arc-seconds, referred to as SRTM-1 and SRTM-3, respectively. Since 1 arc-second at the equator corresponds to roughly 30 m in horizontal extent, the sets are sometimes referred to as ‘‘30 m’’ or ‘‘90 m’’ data. Only the 90 m data are currently available for areas outside the United States. There are ﬁve potential sources of error in the SRTM DEM when used for inundation modeling on the Fly ﬂoodplain: These are: Differences between the SRTM reference geoid and AHD as used in OTML elevation data. These may be up to 1.8 m (Marshall, personal communication). There is also a long-wavelength roll-induced error of about 2 m in SRTM data (Farr, personal communication). Errors associated with radar speckle and the use of interferometry to derive elevations. Data voids due to radar shadowing by locally higher terrain and over water bodies. Inability to capture small features (Teng, 2002) such as tie channels when using a 3 arc-second grid. Some tie channels may only be a few meters wide yet they may be the main path for ﬂooding of some off-river water bodies from the main channel of the Fly. The SRTM DEM is a DEM rather than a ground surface elevation dataset. As such, elevation data describe what the sensor detects rather than true ground levels. This may well be the vegetation canopy, especially in forests. There has been considerable discussion of SRTM DEM accuracy on Internet message boards. Error levels seem to vary with terrain and surface characteristics, with claims of accuracy to within 1–2 m in ﬂat areas and 5–7 m on steeper slopes (see also Smith and Sandwell, 2003; Sun et al., 2003). Water bodies present speciﬁc problems. If the water is too smooth, no radar signal returns and there will be a void in the SRTM data. Rugged terrain may also have data voids because some areas may be in the radar shadow of adjacent hills. Vegetation effects on SRTM elevations are greatest over closed canopy rainforest. Comparisons between forested and adjacent cleared areas show that forest adds 4–25 m to SRTM elevations. Elevation errors in grassland and open forest are less than in rainforest but can also be signiﬁcant. Some are associated with isolated stands of trees and shrubs. Others seem to be the result of radar speckle. Proﬁles taken across the SRTM DEM in ﬂat areas with grass cover or bare sediment suggest that most errors are 72 m or less.

294

G. Pickup

8.2.2. Correction Procedures for SRTM Data SRTM elevation data were corrected by involved identifying inundated areas on airborne and remotely sensed satellite imagery and linking them with observed water surface proﬁles in the Middle Fly. The DEM was then adjusted iteratively so that observed ﬂoodplain inundation patterns are generated by the GRADFLOW model (Pickup et al., 2002) for each known water surface proﬁle. The water surface proﬁles along the main river are calculated by linear interpolation between the gauging stations at Obo (FLY16), Manda (FLY15), and Kuambit (FLY10)1. The DEM correction procedure assumes that ﬂoodplain water surface elevations adjust to main channel water surface elevations quickly. If so, inundated areas on the ﬂoodplain detected on satellite imagery will have similar water surface elevations to the main channel where the two are connected. Channel water surface elevations can then be used to set a maximum elevation for inundated areas of the ﬂoodplain with only limited loss of accuracy. This assumption was tested using water-level data for tie channels and offriver water bodies collected in the mid-1990s. Overall, it seems that ﬂoodplain water levels adjust rapidly and closely approach those of the Fly in times of ﬂood. Also, because the Fly rises and falls relatively slowly, ﬂoodplain water levels can be close to water levels in the river. Some errors will occur when water levels are mismatched but they are likely to be less than 1 m in most cases. This is small compared with other sources of error in the SRTM DEM. There were many steps in correcting the DEM. They included testing and use of synthetic aperture radar data from the ENVISAT satellite (Fig. 8.1). The ENVISAT instrument can be set to acquire many combinations of radar incidence angle and polarization. These had to be tested to determine the best way of identifying inundated areas. LANDSAT data were also used but many scenes were obscured by cloud. Tests of ENVISAT data over the Fly ﬂoodplain show that the largest incidence angles (ENVISAT swath IS7, 42.5–45.2 degrees) give the best water body discrimination. However, swath width is too narrow to cover the ﬂoodplain in a single satellite pass. Inundation monitoring would, therefore, require multiple passes on different days. This would make it difﬁcult to establish a pattern of inundation across the whole ﬂoodplain that was matched to a single set of ﬂow conditions. Low incidence angle data (ENVISAT swath IS1, 15.0–22.9 degrees) cover the whole ﬂoodplain but detect what appear to be wind-generated waves on the water surface, which are indistinguishable from forest backscatter (Fig. 8.1). ENVISAT swath IS4

Figure 8.1: ENVISAT IS1 (upper) and IS4 (lower) HH polarization images of the Middle Fly ﬂoodplain. There is more water in the channel and on the ﬂoodplain in the lower image. Water bodies are clearly shown in black on the lower image. In the upper image some channel reaches (marked by red arrows) cannot be distinguished from the surrounding ﬂoodplain. This may be the effect of wind-driven surface waves on radar returns.

296

G. Pickup

data (31.0–36.3 degrees) gives the best compromise between water body detection and areal coverage and was used here. Reducing the effect of vegetation on SRTM elevations requires data on vegetation canopy heights. However, canopy heights vary from one vegetation type to another, and also within vegetation types This is a common problem in the open forest, mixed forest/grassland systems, and areas of rainforest dieback that are common on the Fly ﬂoodplain. There is limited information in the literature on correcting SRTM elevations for vegetation effects. Kellndorfer et al. (2004) were able to calculate vegetation canopy height by comparing SRTM elevations with a national reference elevation dataset in the United States but this approach is not feasible on the Fly. De Ruyver (2004) attempted to correct SRTM elevations in the Pantanal Wetlands of Brazil. This area has some similarities to the Middle Fly. His approach was ﬁrst to classify vegetation using LANDSAT TM data, creating a basic stratiﬁcation of cover types. He then calculated normalized difference vegetation index2 (NDVI) values and assumed a correlation with vegetation height but with a different range in each vegetation type. The range was based on observation but the approach was not ﬁeld tested or veriﬁed. This approach does not work in the Fly because NDVI values seem unrelated to canopy height in most vegetation types. An alternative method was therefore sought. As a ﬁrst step in reducing the effect of vegetation on elevation, the ﬂoodplain has been classiﬁed into land cover types using LANDSAT TM data. A simpliﬁed version of the classes is listed in Table 8.1. Once the DEM has been corrected, the change in SRTM elevations can be mapped and overlaid on satellite imagery classiﬁed into different land cover types. Adding all the changes together for each cover type and averaging gives a mean change in elevation for each type. This may give a ﬁrst-order elevation correction where vegetation has a uniform height. However, it fails because vegetation height varies within and across cover types, so another approach is needed. Many attempts were made to detect a pattern in the SRTM elevation changes generated by the correction procedure. While there were clear differences associated with each land cover type locally, other factors were at work and a general relationship could not be found across the ﬂoodplain. The best result was obtained by comparing SRTM elevation changes with height above local river base level for each cover type. This required three steps: Identiﬁcation of ﬂow paths, calculation of local base level and calculation of the height of each grid cell above that base level. This was done using the

Floodplain Inundation Modeling and Forecasting for the Middle Fly

297

Table 8.1: Aggregated land cover types on ﬂoodplains and adjacent areas classiﬁed from transformed LANDSAT data. Name Forested areas Terrace forest 1 Terrace forest 2

Open forest

Wet forest

Class description Closed canopy forest on terraces bordering the ﬂoodplain, forest on lower terrace surfaces Forest on terraces bordering the ﬂoodplain in drier areas on the west side of the lower Middle Fly. May include areas with openings in the canopy. Some areas may have been logged or burnt by bushﬁre. Open forest with bare soil or sediment exposed in some areas. Occupies lower terrace slopes and tributary valley ﬂoors. May include occasionally inundated rainforest areas on the main ﬂoodplain Wet and sometimes sparsely forested areas on low-lying ground on and bordering the ﬂoodplain. Bare sediment may be visible through the canopy. Many areas are affected by forest dieback.

Grassed ﬂoodplain and frequently ﬂooded areas Open Grass or bare surfaces with sparse tree or shrub cover. ﬂoodplain 1 Frequently wet and may include forest dieback areas with scattered trees Open Grassed ﬂoodplain surfaces, some localized tree and shrub ﬂoodplain 2 cover Wet grassed As open ﬂoodplain 2 but lower and more frequently ﬂoodplain inundated Swamp 1 Frequently inundated areas with exposed mud at low water levels. May also have sparse grass, shrub, or tree cover Swamp 2 Low lying and occasionally waterlogged grassed ﬂoodplain Flooded areas at low-medium water surface elevations Flooded River water and ﬂooded areas Unclassiﬁed Unclassiﬁed

Small unclassiﬁed areas that are often a mix of existing classes

DEM processing routines in the SAGA GIS (http://www.saga-gis. uni-goettingen.de/html/index.php). Use of Gradient Analysis3 to extract a simple relationship between the SRTM elevation corrections and elevation above local base level. Fitting a predictive relationship to the Gradient Analysis results to allow correction of SRTM elevations. This could then be applied to locations that

298

G. Pickup

were too high to be reached by the ﬂows observed on the satellite image database and so could not be corrected by the procedures described at the beginning of this section. Examples of the correction factors for three different land cover types are shown in Fig. 8.2. The correction factors were only applied to SRTM DEM elevations less than 15 m above local base level. Beyond that, there are insufﬁcient points in the Gradient Analysis to ﬁt a predictive relationship. An approximate test of the accuracy of the corrected SRTM DEM can be made by comparing the elevation of channel bank grid cells with banktop elevations from channel cross-section surveys. However, some differences between these two datasets can be expected for the following reasons: Surveyed cross sections do not tend to be in locations where one or both channel banks end in low lying, ﬂooded zones, Corrected SRTM data are 30 m grid cell averages whereas the crosssection survey data are for point locations, and There is only limited satellite coverage at high water levels so the SRTM DEM elevations tend to cluster around a few water surface proﬁle elevations. 5 Open Forest Terrace Forest 1

SRTM Elevation Correction (m)

0

Open Floodplain 2 -5

-10

-15

-20

-25

-30 0

5

10

15

20

25

Elevation above Local Base Level (m)

Figure 8.2: Examples of correction factors for SRTM DEM elevations for different land cover types.

Floodplain Inundation Modeling and Forecasting for the Middle Fly

299

Where there are two sets of parallel proﬁles, this indicates that some data have been corrected using satellite data on ﬂood extent while others have only had vegetation effects stripped off. These locations lie above the highest water level on the satellite data. More ﬂood information would produce more variability. In spite of these issues, the two datasets compare reasonably well (Fig. 8.3). The main clusters of SRTM data, represented by lines of points on Fig. 8.3, are within 1–1.5 m of the surveyed cross-section banktops. The scatter of points above these lines comes from locations where the channel bank intersects the Pleistocene terrace that borders the ﬂoodplain. Cross-section surveys that extend onto the terrace show similar elevations above the general banktop level. There is also a group of SRTM DEM points 3–4 m below the cross-section survey banktop levels. These low points compare reasonably well with the lower elevations in the 1995 laser proﬁler data.

25 Bank Elevations from Corrected SRTM DEM Left Bank Elevation Dec, 2000 Right Bank Elevation Dec, 2000

ELEVATION m

20

15

10

5

0 0

100

200

300

400

DISTANCE FROM EVERILL JUNCTION Km

Figure 8.3: Banktop elevations from corrected SRTM data compared with December 2000 cross-section survey data.

300

G. Pickup

8.3. Floodplain Hydraulic Model – Development and Testing 8.3.1. The Unsteady Flow Model The unsteady ﬂow calculations are based on the HEC-RAS model developed by the US Army Corps of Engineers (2003). The ﬂoodplain system is represented in HEC-RAS as: A channel described by a series of surveyed cross sections. A set of off-channel storages with known storage volume–height relationships. A set of connections between the channel and the off-channel storages. These are included in the model as a series of lateral weirs. Flows pass down the channel and are attenuated during transmission and by transfer of water to and from ﬂoodplain storages at rates, which vary with differences in water surface elevation between the main channel and the storages. This has the effect of reducing ﬂood height on the rising stage as water ﬂows from the channel into off-channel storages. Flow is then augmented on the falling stage of the ﬂood as the off-channel storages empty back into the main channel of the Fly. Model boundary conditions are set by a time series of discharges at the upstream end of the model and a time series of water levels at the downstream end. The water-level record for the gauging station at Obo (FLY15) is used as the lower boundary condition since it incorporates both backwater effects from the Strickland and tidal effects. The storage-volume curves were derived from the SRTM DEM and runs with the GRADFLOW model. A base water surface proﬁle was interpolated from gauging data for January 11, 1998 when the river was very low and LANDSAT imagery showed that the ﬂoodplain storages were largely dry. The water surface proﬁle was progressively raised above the base proﬁle, 1 m at a time, and the volume of the inundated area was calculated. These calculations were initially carried out at a spacing of one LANDSAT TM pixel (B30 m) but were then aggregated into a single storage on each side of the channel between each surveyed cross section. Various attempts were made to deal with inputs from ﬂoodplain tributaries. These contribute about 9% of total ﬂow to reach upstream of Manda. However, no information is available on the amount and timing of inﬂows so it was not feasible to build them into the model directly. An alternative approach is to assume that tributary inﬂows partly ﬁll ﬂoodplain storages and reduce capacity for inﬂows from the main river.

Floodplain Inundation Modeling and Forecasting for the Middle Fly

301

The resultant changes to available ﬂoodplain storage volumes are fairly low when water levels are high. Numerical trials with changes of up to 5% in ﬂoodplain storage volumes have virtually no effect on the main river water surface proﬁle so it seems fairly safe to ignore tributary inﬂows.

8.3.2. Unsteady Flow Model Testing Results of model testing are shown in Fig. 8.4. The modeled water levels at Manda (FLY16) show acceptably close correspondence with observed values. The water levels at Kuambit (FLY10) are reasonably accurate and many of the features of the backwater dominated water-level record at FLY16 are reproduced. The main errors are overestimation of the water level of some high ﬂows and underestimation of low-ﬂow elevations. A comparison between model predictions for Manda (FLY16) both with (brown line) and without storage effects (dark green line) is also shown in Fig. 8.4. It is clear that ﬂoodplain storage plays a major role in damping ﬂood Observed 20 FLY10

18 16

Elevation (m)

14 Model - No Storage Model

12 10

FLY16

8

Observed

6 FLY15 4 1/11/98

1/3/99

1/7/99

1/11/99

1/3/00

1/7/00

1/11/00

Figure 8.4: Observed and modeled water levels for Manda (Station FLY16) compared with water levels for Kuambit (FLY10) and Obo (FLY15). Note the effects of ﬂoodplain storage on water-level variability at FLY16.

302

G. Pickup

waves and reducing water levels downriver. It also seems that a large volume of river water passes through the ﬂoodplain storages at various stages.

8.3.3. Floodplain Inundation Model The GRADFLOW model uses the water surface proﬁles produced by the hydraulic model to inundate the ﬂoodplain. The channel and the ﬂoodplain DEM are divided into 30 30 m grid cells. The algorithm begins by ﬂooding the main channel and any tie channels directly connected to it. After that, ﬂoods are extended laterally from each channel grid cell one step at a time by searching all immediate neighbors. As each neighboring grid cell is encountered, it is tested to see whether (a) it is already ﬂooded and, (b) it lies above the water level in the main river channel. If either condition is met, the cell is not ﬂooded on this pass. If the cell lies below the main river water level and is not already ﬂooded, it is inundated. In the next iteration, all neighbors of the newly ﬂooded cell are tested and the search gradually proceeds outwards from the ﬂooded area until stopped by higher ground or blocked by cells that are already ﬂooded. This procedure works well in areas where water moves onto the ﬂoodplain during ﬂooding and returns to a similar location on the main channel as the ﬂood recedes. It is less successful where water leaves the channel through a levee breach, ﬂows down the backplains and reenters the main channel further downstream. The GRADFLOW algorithm contains a range of procedures to allow for this effect (Pickup and Marks, 2000, 2001). Tests of the inundation model are presented in Figs. 8.5–8.7. These compare areas ﬂooded by the model with observed areas of water on LANDSAT TM images. Water surface proﬁles used in this test come from HEC-RAS model runs with channel cross section data generated by the HEC-6 sediment model (Pickup and Cui, 2009) rather than observed cross sections. We are, therefore, testing both the inundation model and the channel geometry used to generate the water surface proﬁles into the future. The LANDSAT TM image shown in Fig. 8.5 was used in DEM correction so the modeled pattern of ﬂooding should match with what is observed. The images shown in Figs. 8.6 and 8.7 were not used in DEM correction and provide an independent test. While there are local anomalies, the results are generally good. The observed and modeled ﬂoods in Fig. 8.5 show a close match. Where there are differences, they often occur in areas with, or adjacent to cloud shadow. These were not corrected in the DEM because ﬂooding was obscured.

Figure 8.5: Comparison of modeled inundation area (blue area on upper image) with observed ﬂooding on a LANDSAT TM image on February 27, 2001 on the upper Middle Fly near D’Albertis Junction. Flooded areas on the TM image are blue and black. Flow duration is about 20%.

Figure 8.6: Comparison of modeled inundation area (blue area on upper image) with observed ﬂooding on a LANDSAT TM image on October 22, 2000 on the lower Middle Fly at Aiambak just north of Everill Junction. Flooded areas on the TM image are blue, black, and the darker purple tones. Some bright green areas close to the river may be inundated vegetation or ﬂoating grass mats. Flow duration is about 30%.

Figure 8.7: Comparison of modeled inundation area (blue area on upper image) at 10% ﬂow duration with observed ﬂooding on a LANDSAT TM image (February 11, 2001) at Everill Junction. Flooded areas on the TM image are blue, black, and most of the purple tones. The whole of the Middle Fly ﬂoodplain is inundated. Some bright green areas close to the river may be inundated vegetation or ﬂoating grass mats. Inundation on the Strickland ﬂoodplain (lower right) was not modeled. Flow duration is about 20%.

306

G. Pickup

Other results from missing tie channels that leave small off-river water bodies unconnected to the main river channel. The area shown in Fig. 8.6 also indicates a close match although the extent of ﬂooding appears to be slightly exaggerated in some areas. This may occur because waterlogged grass and ﬂoating grass mats on the TM image do not show as ﬂooded areas. This area is also very ﬂat and changes in the water surface proﬁle as small as 0.2 m can change the ﬂooded area quite substantially. The ﬂood in Fig. 8.7 provides a severe test of model performance. There is minimal relief so small differences in the water surface proﬁle mean large changes in ﬂooded area. Large areas of ﬂoating grass mat and ﬂooded vegetation obscure the ﬂooded area on the TM image while cloud and shadow also make it difﬁcult to recognize surface water. Overall however, the match between observed and modeled inundation is fairly close although model results may slightly overestimate the ﬂooded area. The model should not only be able to predict water-level elevations and inundation extents, but it should also preserve the ﬂow frequency characteristics of the water-level record. A test of this capacity is shown in Fig. 8.8. 12

Model Observed

Water Surface Elevation (m)

10

8

6

4

2 20

40 60 % Time Flow Equalled or Exceeded

80

Figure 8.8: Observed and modeled water-level duration curves for Manda (FLY16). Model data were generated from river proﬁle and cross-section data for 2000 from the HEC-6 sediment model.

Floodplain Inundation Modeling and Forecasting for the Middle Fly

307

The two curves in this ﬁgure show how well the model predicts ﬂow durations at Manda (FLY16). The model water-level data were generated using cross sections from the 2003 sediment model so we are testing both the HEC-RAS model and the HEC-6 simulations. The results are, again, extremely close, indicating that model results reproduce the natural ﬂow frequencies very well. The two curves deviate at low water surface elevations but this may be a result of inaccuracies in the input data rather than the performance of the model. This is not really an issue in model performance since we are mainly interested in forecasting inundation at higher ﬂows.

8.4. Forecasting the Extent and Frequency of Floodplain Inundation 8.4.1. Methods Once calibration and testing were completed, the hydraulic model was run on the scenario described in Chapter 7 (Pickup and Cui, 2009). This involves dredging at 15.1 Mt/year and mine closure in 2010. A synthetic ﬂow record was used in each model run based on observed data for Kuambit and Obo. This record has similar ﬂow frequency characteristics to the long synthetic ﬂow record used in the sediment model. Water surface proﬁles from the hydraulic model were used to derive a set of water surface elevation frequencies down the length of the Middle Fly. These were used to inundate the ﬂoodplain using GRADFLOW. The resultant inundation maps and water surface proﬁle changes with time are presented in Figs. 8.9–8.16. The hydraulic model and the GRADFLOW routines can also be used to estimate ﬂows between the river and the ﬂoodplain. These values were used to estimate sediment delivery to the ﬂoodplain surface. This value does not include levee deposition and further development of the levee model described in Chapter 7 is planned. The HEC-RAS model cannot generate the continuously changing bed levels produced by the HEC-6 sediment transport model used in the 2003 sedimentation study. However, it is possible to represent the effects of deposition on ﬂoodplain inundation by taking the river channel characteristics produced in HEC-6 model runs at 10 year intervals and using them in HEC-RAS. Changes in the pattern of ﬂoodplain inundation over time can then be generated.

308

G. Pickup

Change in Water Surface Elevation (m)

3.0 2000 2010 2020 2040 2050

2.5

2.0

1.5

1.0

0.5

0.0 0

100

OBO

200

300

400 D'ALBERTIS JUNCTION

Distance Upstream (km)

Figure 8.9: Change in water surface elevation with time for mine closure in 2010, dredging at 15.1 Mt/year. 20 18

% OF TOTAL DISCHARGE

16 14 12 10 8 6 4 2 0 1980

1990 1985

2000

2010

2020

2030

2040

2050

YEAR

Figure 8.10: Model estimates of inﬂow from the main channel of the Middle Fly to the ﬂoodplain as a percentage of total river ﬂow.

Floodplain Inundation Modeling and Forecasting for the Middle Fly

309

Figure 8.11: Pre-mining situation 1985, modeled ﬂoodplain inundation patterns for the upper Middle Fly.

8.5. Results Figure 8.9 shows the modeled changes in water surface elevation through time. Behavior is typical of a river experiencing an increase in sediment load. Deposition gradually progresses downstream raising water levels until the

310

G. Pickup

Figure 8.12: Pre-mining situation 1985, modeled ﬂoodplain inundation patterns for the lower Middle Fly.

river establishes a new long proﬁle with a steeper gradient. This allows it to convey delivered load further downstream. As water levels rise, more water will circulate through off-channel storages but this effect is not strong enough to interfere with the basic pattern of water-level rise that is progressing downstream.

Floodplain Inundation Modeling and Forecasting for the Middle Fly

311

Figure 8.13: Modeled ﬂoodplain inundation patterns for the upper Middle Fly in 2010. When there is a reduction in sediment load, the proﬁle will begin to ﬂatten from the upper reaches downstream. However, the effect of increased load may continue to raise water levels in downstream reaches even once proﬁle ﬂattening begins at the upper end of the system. The effects of increased load take many years to pass down the river and water levels continue to rise long after mine closure.

312

G. Pickup

Figure 8.14: Modeled ﬂoodplain inundation patterns for the lower Middle Fly in 2010. There are some irregularities in water surface proﬁle changes. This results from not having enough surveyed cross sections for detailed hydraulic modeling and the use of a one-dimensional sediment model. In reality, the proﬁles would be smoother.

Floodplain Inundation Modeling and Forecasting for the Middle Fly

313

Figure 8.15: Modeled ﬂoodplain inundation patterns for the upper Middle Fly in 2050. The proportion of total main channel discharge entering the ﬂoodplain varies from about 9% at the start of mining to about 14% in 2050 (Fig. 8.10). This is less than Dietrich and Day’s (2004) estimate of 20–40%. Most of the increase is due to the body of sediment already

314

G. Pickup

Figure 8.16: Modeled ﬂoodplain inundation patterns for the lower Middle Fly in 2050. on the bed of the lower Ok Tedi and Middle Fly prior to the start of dredging. The ﬂoodplain inﬂow data may be used together with sediment transport data to obtain a crude estimate of sediment delivery to the ﬂoodplain.

Floodplain Inundation Modeling and Forecasting for the Middle Fly

315

Suspended sediment data for Station FLY11 at Niukamba obtained by depth integrated sampling show a mean sediment concentration of 570 mg/l. However, these data were not obtained using a large enough sampler and have very similar values to surface gulp samples. A correction factor of 1.2 gives a mean concentration of 684 mg/l and a total inﬂow to the ﬂoodplain of about 5 Mt/year. This does not include sediment deposited on the levees, which could involve 1–2 times as much material. The SRTM DEM and the GRADFLOW model have been used to calculate the duration and extent of ﬂoodplain inundation. Maps are presented here for the pre-mining channel based on the 1985 river reconstruction used in the 2003 sediment modeling study, for 2010 and for 2050 when the river is close to adjusting to the increased sediment load. The ﬂoodplain maps show that there is a major increase in ﬂooding in the upper part of the Middle Fly during mine life. The lower Middle Fly experiences major changes after mine closure, no matter what the scenario. Recovery in the upper river is very limited as inundation remains more frequent than in the pre-mine situation. Once the initial rise in water occurs, the changes pass downstream and ﬂooding occurs more frequently. Lateral extension is limited because the ﬂoodplain is conﬁned by terraces. Flooding does extend up tributaries as water levels rise but the main effect is in the downstream direction. The precise impact varies with location but some areas go from experiencing ﬂooding for 10% or less of the time to 70% or more. This is likely to produce substantial changes in the whole ﬂoodplain ecosystem. Many of these changes will occur long after mine closure. While continued dredging does improve the situation, the bulk of the impact comes from the sediment that had already passed Bige before dredging started. There are few options available to deal with this.

8.6. Conclusions The ﬂoodplain inundation model and maps show how the passage of mine waste down the Middle Fly may affect the ﬂoodplain. Over time, inundation depth and frequency will increase in many areas. This will affect a whole range of geochemical, biological, and social processes. It will also change the amount, type, and accessibility of both terrestrial and aquatic resources used by ﬂoodplain communities. The inundation maps may be used in association with other knowledge to derive estimates of the timing and extent of some of these changes.

316

G. Pickup

Frequency and extent of inundation is one of the variables affecting the rate of weathering through the process of wetting and drying. Floodplain inundation also delivers sediment containing sulﬁdes to the ﬂoodplain with the amount of deposition declining exponentially with distance from the channel (Dietrich, 2000). There will also be greater deposition in low-lying areas close to the channel, all other things being equal. Inundation depth and frequency maps have some potential for predicting ﬂoodplain areas where ARD problems may arise. The distribution of many ﬂoodplain plant communities reﬂects frequency and duration of ﬂooding. While there may be some tolerance to variations in ﬂooding allowing plant communities to persist, other areas are likely to experience changes in plant cover. These may occur rapidly and allow invasive species to colonize large areas. Aquatic environments will also change. Many areas will be ﬂooded for longer periods and terrestrial environments may become increasingly aquatic. Major changes to the whole ﬂoodplain ecosystem can be expected. The ﬂoodplain model is close to the limits of what can be done with available data. It is not a full hydraulic model so there will be some imprecision in the predictions. Also, the SRTM DEM can be improved as more satellite imagery becomes available. These will produce differences of degree rather than major changes in the inundation predictions. Ultimately, the accuracy of the ﬂoodplain inundation rests on the accuracy of predictions from the sediment transport model. While there are questions about this model in some areas (Pickup and Cui, 2009; Booth, 2004), the overall pattern of predicted deposition is thought to be reasonable.

Notes 1. Gauging station locations are shown in Figs. 4 and 5 of the ‘‘Geomorphology’’ chapter. 2. NDVI is a widely used vegetation greenness index based on the difference between red and near-infrared values in satellite imagery. 3. Gradient Analysis (Pickup and Marks, 2000) is a technique for identifying complex relationships between variables in satellite images and Geographic Information System layers with very large amounts of data and signiﬁcant noise. When the values in every grid cell are plotted, relationships between variables may be obscured by the noise, especially where thousands-to-hundreds of thousands of observations are involved. In the Gradient Analysis carried out here, elevation above base level is split into classes and the average value of the SRTM elevation correction is then calculated for each elevation above base level class on a grid cell by grid cell basis. The result may then be plotted and a normal or weighted regression equation ﬁtted to give a predictive relationship.

Floodplain Inundation Modeling and Forecasting for the Middle Fly

317

REFERENCES Booth, S. (2004). Sediment Transport Model Workshop. Cairns, Australia, 21st–23rd January 2004. Final Report. Environment Department, Ok Tedi Mining Ltd. and Total Catchment Management Services. Cui, Y., & Parker, G. (1999). Sediment transport and deposition in the Ok Tedi-Fly River System, Papua New Guinea: the modeling of 1998–1999. St. Anthony Falls Laboratory, University of Minnesota, Minnesota. De Ruyver, R. (2004). DEM Optimization for Hydrological Modelling using SRTM for the ‘Pantanal Region’, Brazil. M.Sc. Thesis. International Institute for Geo-Information Science and Earth Observation, Enschede, the Netherlands. Dietrich, W. E. (2000). Quarterly and Fourth Monthly Report, Middle Fly Sedimentation Project. Department of Earth and Planetary Science, University of California, Berkeley. Dietrich, W. E., & Day, G. (2004). Rates and Patterns of Floodplain Sedimentation, Middle Fly Floodplain (1985–1994). Presentation at the Sediment Transport Model Workshop, Cairns, Australia. January, 2004. Environment Department, Ok Tedi Mining Ltd. Farr, T. G., & Kobrick, M. (2000). Shuttle Radar Topography Mission produces a wealth of data. American Geophysical Union Eos, 81, 583–585. Kellndorfer, J., Walker, W., Pierce, L., Dobson, C., Fites, J. A., Hunsaker, C., Vona, D., & Clutter, M. (2004). Vegetation height estimation from Shuttle Radar Topography Mission and National Elevation Datasets. Remote Sensing of Environment, 93, 339–358. Marshall, A. R. (2002). Ok Tedi – Middle Fly Bed Aggradation Survey. Epoch 6 – March 2002 (and associated spreadsheets). Environment Department, Ok Tedi Mining Ltd. Pickup, G. (2003). Effects of Mine Life Extensions and Rates of Dredging on the Ok Tedi and Fly River. Uncertainty Analysis for the Null Scenario. Report to Environment Section Ok Tedi Mining Ltd. Pickup, G., & Marks, A. (2000). Identifying large-scale erosion and deposition processes from airborne gamma ray remote sensing and digital elevation models in a weathered landscape. Earth Surface Processes and Landforms, 25, 535–557. Pickup, G., & Marks, A. (2001). Regional-scale sedimentation process models from airborne gamma ray remote sensing and digital elevation data. Earth Surface Processes and Landforms, 26, 273–279. Pickup, G., & Cui, Y. (2009). Modeling the impact of tailings and waste rock disposal on the Fly River system. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, pp. 257–289. Pickup, G., Marks, A., & Bourke, M. (2002). Palaeoﬂood reconstruction on ﬂoodplains using geophysical survey data and hydraulic modeling. In: P. Kyle

318

G. Pickup

House, R. H. Webb, V. R. Baker, & D. R. Levish (Eds). Ancient Floods, Modern Hazards: Principles and Applications of Palaeoﬂood Hydrology. Water Science and Application. American Geophysical Union, Washington D.C., Vol. 5, pp. 47–60. Smith, B., & Sandwell, D. (2003). Accuracy and resolution of Shuttle Radar Topography Mission data. Geophysical Research Letters, 30, 1467–1470. Sun, G., Ranson, K. J., Kharuk, V. I., & Kovacs, K. (2003). Validation of surface height from Shuttle Radar Topography Mission using shuttle laser altimeter. Remote Sensing of Environment, 88, 401–411. Teng, L. W. (2002). Drainage analysis using DEMs from different sources. http:// www.crwr.utexas.edu/gis/gishydro03/Classroom/trmproj/Teng/Summary.htm US Army Corps of Engineers, Hydrologic Engineering Centre (2003). HEC-RAS Hydraulic Reference Manual.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00409-6

Chapter 9

Biogeochemistry of Copper in the Fly River Simon C. Apte Centre for Environmental Contaminants Research, CSIRO Land and Water, PMB 7, Bangor Sydney, New South Wales 2234, Australia

9.1. Introduction Since mining of the Mount Fubilan porphyry copper–gold ore body by Ok Tedi Mining Limited (OTML) began in 1984, the Ok Tedi/Fly River system (Fig. 9.1) has received high loads of mine-derived sediments enriched in particulate copper. The potential release of copper from these sediments and its toxicity to aquatic organisms in the river system was an obvious concern to mine management. In order to understand this issue, it was necessary to understand the fate, mobility, and transport of mine-derived copper. The factors governing the dissolution of particulate copper from mine-derived sediments and the chemical speciation of dissolved copper were of particular concern. This chapter gives an overview of the geochemical studies undertaken over the ﬁrst 15 years of mine life, before acid rock drainage (ARD) was detected in the system. A number of very detailed studies were carried out to gain an understanding of copper geochemistry in river, offriver water body (ORWB), estuarine, and coastal shelf environments. The studies were made challenging because of the remoteness of the study environments and the logistical demands imposed. As a result of these efforts, a quite extraordinary depth of knowledge on the copper geochemistry of the Fly River system was obtained, much of this information being of relevance to other aquatic systems. The factors inﬂuencing dissolved copper concentrations in the aquatic environments of the Fly River system are covered in this chapter. Corresponding author. Tel.: þ61-2-9710-6838; Fax: þ61-2-9710-6837;

E-mail: [email protected] (S.C. Apte).

322

S. C. Apte

Figure 9.1: Location of sampling sites on the Ok Tedi and Fly rivers.

Biogeochemistry of Copper in the Fly River

323

9.2. Copper and Mine-Derived Suspended Sediment Distributions in the River System Details of the basic geochemistry and hydrology of the Fly River system may be found in the review by Salomons and Eagle (1990). The Ok Tedi, Fly, and Strickland rivers have quite similar bulk chemical compositions (Salomons and Eagle, 1990) and are dominated by calcium and bicarbonate ions which reﬂect the limestone geology of the upland areas. The pH of water in the Ok Tedi and Fly River are generally in the range 7.0–8.5; however, the waters draining from the Middle Fly ﬂoodplain that ﬁll the ORWB of this region are characterized by low hardness, low pH, and high dissolved organic carbon (DOC). These waters are also introduced into the main river system following rainfall events. Since 1989, when the copper extraction circuits at the OTML mill were fully commissioned, the mine, with approval from the State, has discharged typically 70,000 tonnes/day of ﬁne-grained tailings and 140,000 tonnes/day of waste rock overburden directly into the river system. An additional 20,000 tonnes/day of natural sediment enters the river system as a result of valley wall erosion directly related to the mine’s activities. Once introduced into the river system, the majority of the silt and clay load travels down the river in suspension. In-river grinding (particle abrasion) of coarse waste rock, particularly in the upper reaches of the river system, also generates ﬁne particles. Aside from the Strickland River, the inputs of naturally suspended sediments are relatively small and mine-derived sediments dominate the sediment load. Over the last 20 years, OTML has extensively monitored a wide range of parameters at sites in the river, estuary, and ORWB of the Fly River system. Thousands of samples have been collected and analyzed at the mine environment laboratory in Tabubil and at off-site laboratories. A detailed analysis of this data and the inherent trends are outside the scope of this chapter; however, the general concentration versus distance trends of some important geochemical parameters are worthy of note (Fig. 9.2). Total suspended solids (TSS) are highest closest to the mine and then decrease sharply with distance down the Fly River. TSS concentrations then increase slightly below the junction with the Strickland River owing to the sediment input by this river. Particulate copper concentrations in suspended sediments (Fig. 9.2) are highest closest to the mine and gradually decrease through the Middle Fly owing to mixing with natural sediments low in copper. There is a step-change decrease in particulate copper concentration at the junction with the Strickland River owing to dilution with natural sediments from the

S. C. Apte

5000

8000 Junction of Fly & Strickland Rivers

(a)

TSS (mg/L)

6000 3000

1993 1994 1996 Discharge 1993-1996

Junction of Ok Tedi & Fly River

2000

4000

2000 1000

0 0

200

400

600

800

0 1000

Distance from the mine (Km) 1600 (b) 1200 1993 1994 1996

800

400

0 0

200

400

600

800

1000

Distance from the mine (Km) 20 Dissolved Cu (µg/L)

(c) 15 1993 1995 1996

10

5

0 0

200

400

600

Distance from the mine (Km)

800

1000

Discharge (Cumecs)

4000

Particulate Cu (µg/g)

324

Biogeochemistry of Copper in the Fly River

325

Strickland which are relatively low in copper. Depending on the type of ores being processed, suspended particulate copper concentrations in the Ok Tedi, just below the mine site, range from typically 1,000 to 1,800 mg/g. This represents an increase in particulate copper concentrations of over 20 times the pre-mining concentrations of suspended sediments (typically 50 mg/g) in many parts of the system. Dissolved copper concentration proﬁles with distance along the Ok Tedi and Fly River show a trend that is somewhat counter-intuitive (Fig. 9.2). Copper concentrations in the water column actually increase with distance from the mine and are highest in the Middle Fly region of the river between Bosset and Obo, some 400–600 km from the mine site. At the junction of the Fly and Strickland Rivers, there is a sharp decrease in dissolved copper concentration. Laboratory modeling experiments showed that this phenomenon is not just a consequence of dilution, but partly caused by the rapid adsorption of copper onto the large load of natural sediments introduced by the Strickland River which are low in particulate copper (Apte et al., 1995a). This process adsorption results in typically a 30% reduction in dissolved copper concentration (Fig. 9.3). It should be noted that in the work discussed in this chapter, the term ‘dissolved copper’ is operationally deﬁned and includes any copper species that passes through a 0.45 mm diameter pore size ﬁlter. This may include colloidal species.

9.3. Factors Affecting Copper Concentrations in the Fly River Most mining-related studies concerning trace metal mobility have either been directed to the effects of ARD, where metals are released in soluble forms and following pH neutralization can then precipitate (Chapman et al., 1983; Johnson, 1986), or the leaching of metals from deposited sediments (Nimick and Moore, 1991; Pedersen et al., 1993). The Fly River system is unusual as, at the point of discharge, most copper is in particulate form and dissolved copper comprises only a minor proportion (o1%) of the metal’s total load. Understanding the factors affecting copper release from mine-derived Figure 9.2: Summary of routine OTML monitoring data. (a) mean discharge (1993–1996), total suspended solids (TSS), (b) particulate copper, and (c) dissolved copper concentrations downstream of the mine. The data are annual average values. The years 1993, 1994, and 1996 were classiﬁed by OTML as being climatically dry, normal, and wet years, respectively.

326

S. C. Apte

8

Dissolved Cu (µg/L)

Unfiltered Filtered control 6

4

2

0 0

5

10

15 Time (hour)

20

25

30

Figure 9.3: Adsorption of dissolved copper as a function of equilibration time. Mixing of Fly River (Obo) and Strickland River waters at 1:1 dilution. Initial copper concentration 7.4 mg/L, TSS 535 mg/L. particles was therefore critical in unraveling the geochemical copper cycle and understanding copper toxicity to aquatic organisms. During the 1990s, the ores being processed at the Ok Tedi mine were dominated by copper sulﬁde minerals such as chalcopyrite with lesser quantities of chalcocite, digenite, covellite, and bornite. In addition, there was some oxide mineralization containing copper oxide minerals such as malachite, cuprite, and azurite. Thermodynamic equilibrium calculations clearly show that the solubility of copper sulﬁde mineral phases such as amorphous copper sulﬁde and chalcopyrite at circumneutral pH are in the submicrogram per liter range and cannot account for the observed riverinedissolved copper concentrations in the system (typically range of 0.5–20 mg/ L). This indicates the importance of copper oxide/carbonate minerals and copper sulﬁde oxidation in determining dissolved concentrations in the river system. In the absence of oxygen, the aqueous solubilities of the principal copper sulﬁde mineral phases, chalcocite and chalcopyrite, were negligible (o0.5 mg/ L) (Apte et al., 1999). However, suspension of these minerals in Fly River water in equilibrium with the atmosphere results in rapid surface oxidation reactions and large increases in solubility. Chalcopyrite was signiﬁcantly less susceptible to oxidation than chalcocite. This may be attributed to the role of iron in controlling the solubility of CuFeS2.

Biogeochemistry of Copper in the Fly River

327

Copper solubility tests using ﬁltered waters from the Fly River were conducted in order to better understand copper dissolution from minederived solids (Apte et al., 1999). The measured and calculated solubilities of Cu(OH)2, malachite, azurite, and tenorite over the pH range 6.8–8.0 in Fly River water are shown in Fig. 9.4. Solubility calculations were carried out using the program MinteqA2 (Allison et al., 1991). The predicted solubility of copper varies widely depending on which solid is assumed to control dissolved copper concentrations. Experiments indicated that despite the differences in minerals added to the waters, dissolved copper concentrations after 16 h equilibration of the copper mineral with the river water at 251C were remarkably similar and within an order of magnitude of each other. The experimental results suggest that the minerals undergo transformation during resuspension in oxygenated water to form the same solid phase which controls copper solubility, regardless of the composition of the parent mineral. Based on experimental measurements described above, the maximum attainable dissolved copper concentration (o0.45 mm) in Fly River water at pH 7.8 was found to be ca. 250 mg/L. The experimentally observed solubility is at least eight times higher than dissolved copper concentrations typically 1000

Dissolved Cu (µg/L)

800 Cu(OH)2 600 Azurite 400

Tenorite

200

Malachite 0 6.5

7.0

7.5 pH

8.0

8.5

Figure 9.4: Measured solubility of Cu(OH)2 (K o0.45 mm, o0.1 mm), malachite (’), azurite (7), and tenorite (~) in Fly River water, pH 6.8–8.0. The solid lines are the solubilities predicted by MinteqA2 for the four mineral phases.

328

S. C. Apte

observed in the river system. The fact that dissolved copper concentrations of this magnitude have never been observed in the ﬁeld, indicates the importance of adsorption reactions in attenuating dissolved copper concentrations. In the ORWB of the Fly River, where water pH can be much lower than the river system, the maximum attainable dissolved copper concentrations are much higher. For instance, at pH 7.0 the maximum solubility is 550 mg/L, and at pH 6.6 the maximum value is 850 mg/L. Tailings leaching studies were conducted in order to understand the factors affecting the initial release of copper from mine-derived sediments as would occur when the mine sediments are introduced into the river system. The experiments were conducted by immersing known concentrations of tailings solids into freshwaters and measuring the subsequent release of copper over a time period of typically 24 h. Three freshwaters were used in the leaching studies: Ok Mani (an upper tributary of the Ok Tedi), the Fly at Kiunga (above the junction with the Ok Tedi), and Bosset Lagoon (a slightly acidic ORWB in the Middle Fly). The water quality parameters for these samples are summarized in Table 9.1. In order to characterize temporal variations in copper release as would occur as ore composition changes, a series of 24 h composite ﬂotation plant tailings (November 1994 to January 1995) were used in the study. The particulate copper concentrations of the 25 tailings samples varied over the range of 730–2,130 mg Cu/g. Typically, 4075% of the tailings particles were o63 mm with coarse particles (W180 mm) comprising 1377% of total mass. When immersed in river water, the initial release of copper from tailings material was characterized by a rapid initial release of copper in the ﬁrst hour followed by a slower release over the ensuing 17 h (Fig. 9.5). The copper released in the ﬁrst hour comprised 58–79% of the dissolved copper concentrations observed at the end of the experiment (48 h). Copper release was essentially independent of particle concentration over the range 0.5–4 g/ L (Fig. 9.6). As expected, pH played a signiﬁcant role in determining copper solubility. The amount of copper released into solution increased with decreasing pH (Fig. 9.7). In waters having a pH of less than 7.0, there was a Table 9.1: Chemical properties of the waters used in the leach tests. Water

pH Dissolved Dissolved Complexing DOC (mg/L) Cu (mg/L) Ca (mg/L) Capacity (mg/L)

Ok Mani 7.8 Fly River at Kiunga 7.8 Bosset Lagoon 6.6

0.8 1.0 4.1

33 24 6

16.9 15.9 40.0

2.1 2.6 6.4

Biogeochemistry of Copper in the Fly River

329

Dissolved Cu (µg/L)

80

Ok Mani Kiunga Bosset

60

40

20

0 0

10

20

30

40

50

Time (hour)

Figure 9.5: Effect of time on the release of copper from mine tailings (tailings concentration 1 g/L, three different water types).

50 Ok Mani Kiunga Bosset

Dissolved Cu (µg/L)

40

30

20

10

0 0

2

4

6

8

10

12

Tailings Concentration (g/L)

Figure 9.6: Effect of tailings concentration on the release of copper from mine tailings. Sample T13 leached with three different water types (leaching time 18 h).

330

S. C. Apte

Dissolved Cu (µg/L)

80

60

40

20

0 5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

9.5

pH

Figure 9.7: Effect of pH on dissolved copper release from mine-derived sediments. Unﬁltered river water sample from Konkonda (total suspended solids 1,150 mg/L) titrated with acid and base. signiﬁcant increase in the amount of copper leached from tailings particles. This has a bearing on processes occurring in ORWB where pH values below 7.0 can occur. Mine-derived particles transported into ORWB would therefore be expected to release more copper into solution. Water type had a signiﬁcant effect on copper release with leachable copper concentrations increased in the order: Bosset Lagoon W Fly River W Ok Mani (Fig. 9.8). The highest copper concentrations were attained in the samples containing the highest DOC concentrations and copper complexation capacities (CuCCs). It was therefore postulated that dissolved organic matter (DOM) played a role in the release of copper from mine-tailings particles. This important mechanism was proven experimentally by employing UV irradiation to destroy dissolved organic matter of the test waters prior to leach testing. A dramatic reduction in the amount of copper released into solution was observed in the leach tests using UV-irradiated waters (Fig. 9.9). Further experiments conducted on other tailings samples yielded similar results. The mechanism of copper release may involve the initial complexation of surface-bound copper by DOM followed by release of the Cu-DOM complex into solution. The composition of DOM and speciﬁcally, the nature of functional groups which are able to complex metals are potentially important factors in determining the extent of copper mobilization. It was hypothesized that stronger complex formation is observed with DOM derived from lowland ﬂoodplain sources (i.e., Fly River and Bosset Lagoon). Increasing the calcium concentration of the leach test waters

Biogeochemistry of Copper in the Fly River

331

100 Ok Mani Kiunga Bosset

Dissolved Cu (µg/L)

80

60

40

20

0 0

5

10

15

20

25

Tailings Sample

Figure 9.8: Variability of dissolved copper release from different tailings samples. Results of batch leach tests performed on samples T1–T25 with three different water types (tailings concentration 1 g/L, leaching time 18 h). 80 Untreated Water UV-irradiated Water

Dissolved (Cu µg/L)

60

40

20

0 Ok Mani

Kiunga

Bosset Lagoon

Figure 9.9: Effect of UV irradiation of leach test waters on copper release from mine tailings. Comparison of the amounts of dissolved copper leached from tailings sample T13 with three different water types (tailings concentration 1 g/L, leaching time 18 h). Subsamples of water were irradiated with UV light (650 W Hg vapor lamp for 3 h) to destroy dissolved organic matter prior to use in the experiment.

332

S. C. Apte

Dissolved Cu (µg/L)

20

15

10

5

0 0

50

100

150

200

250

Dissolved Ca (mg/L)

Figure 9.10: Effect of dissolved calcium concentration on copper release from sample tailings sample T12. (Fly River water, tailings concentration 1 g/L, leaching time 18 h). (Fig. 9.10) did not enhance the release of copper from mine tailings. Ionexchange processes involving the displacement of copper ions from surface binding sites by other divalent cations are therefore unlikely to contribute signiﬁcantly to copper release. Signiﬁcant between-sample variability was observed in the amount of copper leached from the tailings (Fig. 9.8) showing that tailings composition played an important role in determining the amount of copper released into solution. The amount of copper released into solution was typically between 1 and 8% of the total copper present in the tailings particles. Statistical analysis of the data sets showed there were no signiﬁcant correlations between the ﬁnal concentration of copper measured in the leach tests and major element concentrations (including Cu, Mn and Fe, S) and particle size. Highly signiﬁcant correlations (Po0.05) were however found between the fraction of copper that dissolved in cold, dilute nitric acid (acid-soluble copper, ASM-Cu) and the three sets of leach test data (Fly River: r ¼ 0.850, Ok Mani: r ¼ 0.755, Bosset Lagoon: r ¼ 0.800). Under the conditions adopted for the analysis of ASM-Cu (pH 2, 10 min extraction time), sulﬁde copper phases are insoluble (Svehla, 1987) and only oxidized forms of copper are measured (e.g., copper hydroxides and carbonates). The relationship between ASM-Cu and the Fly River leach test data is shown in Fig. 9.11. ASM-Cu concentrations ranged from 208

Biogeochemistry of Copper in the Fly River

333

Dissolved Cu (µg/L)

30

20

10

0 0

200

400

600

800

1000

ASM-Cu (µg/g)

Figure 9.11: Relationship between cold acid-soluble copper concentrations of the tailings samples (ASM-Cu) and the amount of copper released in the batch leach tests conducted with Fly River water. Dissolved Cu ¼ 0.020 [ASM-Cu]þ6.65. (r ¼ 0.850, po0.05). to 974 mg/g and were typically 35% of total particulate copper concentrations. Oxidation of copper sulﬁde minerals may occur either in the mine pit (on exposure to air) or during the processes of crushing and grinding of the ores. Over the pH range typically encountered in the river system, complexation by DOM was the most important single process determining the release of copper from mine-derived particulates observed in this study. The amount of copper released into solution depended on the concentration and composition of DOM and the amount of copper soluble in dilute acid in the tailings particles. Copper release from waste rock also occurs via the same mechanisms identiﬁed for mine tailings. While some waste rock types contain particulate copper concentrations comparable to mine tailings, the proportion of copper present in an oxidized form is generally much lower than in mine tailings (e.g., typically 10% ASM-Cu in waste rock samples compared to 30% ASMCu in tailings samples). This is because ore particles are subject to additional oxidation during mineral processing (e.g., during crushing and grinding) which increases ASM-Cu concentrations of the tailings particles. Nevertheless, waste rock particles may undergo oxidation during riverine transport, which will increase copper mobility.

334

S. C. Apte

9.4. Riverine Processes This section considers the in-river geochemical transformations that affect copper geochemistry. Dissolved copper concentrations in the river are determined by a combination of geochemical processes during the 10–15 day transit time of suspended particulates through the river system. These processes include: copper release from mine-derived sediments, copper complexation by natural organic matter, sorption reactions involving natural and mine-derived sediments and copper sulﬁde oxidation. A conceptual model of copper transport and reactivity in the Fly River is presented in Fig. 9.12.

9.4.1. Copper Complexation by Dissolved Organic Matter To a large extent, dissolved copper concentrations in the river system reﬂect changes in the concentration and composition of DOM. CuCC comprises the concentration of organic ligands that are present in the sample and available to strongly complex copper. It is determined by titration of the sample with copper and, after an appropriate equilibration period, measurement of the Key dCu Dissolved Copper pCu Particulate Copper L Organic Ligand

L inputs from rivers and floodplain

dCu

dCu

dCu Cu

Cu Cu

Cu Cu

dCu,pCu

Cu Cu

Cu

Cu O2

desorption Cu + L

Cu

L

Cu Cu

organic complexation

oxidation

adsorption

dCu,pCu

CuL

Estuary

solution complexation

from upper river

dCu

Cu

Cu

M

dissolution

dCuL Cu

dCu Sedimentation and resuspension River Bed

diffusion pCu

dissolution

dCu in porewater

Figure 9.12: Conceptual model of copper cycling in the river system.

Biogeochemistry of Copper in the Fly River

335

uncomplexed copper (Apte et al., 2001). CuCC is an operationally deﬁned parameter which is inﬂuenced by the measurement technique used to estimate uncomplexed copper. In the work described in this chapter, anodic stripping voltammetry (ASV) was used to measure CuCC. CuCC is lowest in the Ok Tedi and increases progressively downstream to a maximum in the Middle and Lower Fly (Fig. 9.13). The highest complexing capacities (30–50 mg/L) are found in the ORWB of the Middle Fly. These gradients in complexing capacity concentrations largely reﬂect changes in vegetation type, i.e., the transition from upland forest to lowland swamp, and the associated differences in dissolved organic matter production rates and inputs. OTML has routinely monitored CuCC at several sites in the river on a quarterly basis since 2002. Analysis of the data set to date (Rogers et al., 2005) indicates temporal variability, but no identiﬁable changes with time. The data analysis indicated little evidence of a relationship between ﬂow and CuCC (Apte et al., 1999). The size distribution of copper complexing ligands present in ﬁltered (o0.45 mm) water samples collected from the river system were characterized using a combination of equilibrium dialysis and differential pulse ASV (Apte et al., 2001). The proportion of CuCC associated with small molecular weight (o1,000 Da) ligands increased progressively from 40% in the upper catchment to 79% at lowland ﬂoodplain/wetland locations (Fig. 9.14). This

Complexing Capacity (µg/L)

40

30

20

10

n io ct un iJ

re Bu

ric kl St

Le w ad a

O gw a

er an

d

R

iv

O bo

Bo ss Bo et ss et La go on

ba Fl

y

uk N

at

um

ga un Ki

da on nk Ko

ge in N

O

k

M

ru

an

m

i

0

Figure 9.13: Typical copper complexation capacities of waters from the Fly River system. Locations are shown in Fig. 9.1.

336

S. C. Apte

Copper Complexation Capacity (µg/L)

20 <1,000 MWCO 1,000 - 14,000 MWCO >14,000 MWCO 15

10

5

0 Ok Mani

Kiunga

Oxbow 6

Lake Pangua

Sample

Figure 9.14: Variation of copper complexing capacity with molecular size fraction as determined by dialysis. The molecular weight cutoffs (MWCO) for the dialysis membranes used in the study are given in Daltons.

increase illustrated the role of the ﬂoodplain as a source of low molecular weight ligands to the river system. A signiﬁcant proportion of copper complexation was also associated with ligands in the W14,000 Da fraction. As the majority of ﬁlterable iron was also found in this fraction, it is likely that this complexing pool contains organic ligands adsorbed to the surface of inorganic colloids as well as large molecular weight ligands. Destruction of dissolved organic matter by sample pretreatment with UV radiation resulted in a complete loss in CuCC detectable by ASV conﬁrming that the observed complexation of copper was associated with organic ligands and not inorganic components of the samples.

9.4.2. In-River Oxidation of Copper-Containing Minerals On entry to the river system, typically 30% of the copper in the mine-derived sediments is present in an oxidized form as a result of oxidation processes that have occurred at the mine (e.g., during ore crushing and grinding). As noted earlier, oxidation of mine-derived copper sulﬁdes to mineral forms that are more amenable to solubilization (e.g., malachite and amorphous copper

Biogeochemistry of Copper in the Fly River

337

Table 9.2: Copper concentrations measured on river samples (January 1995). Site Ningerum Konkonda Nukumba Fly at Bosset Obo Ogwa Burei Junction

TSS (mg/L)

Dissolved Cu (mg/L)

2380 1150 170 118 7 124 58

12.7 8.5 12.4 13.8 8.6 2.3 3.4

Particulate Acid leachable Cu (mg/g) Cu (mg/g) 1190 1240 1400 1230 1580 159 158

328 534 809 970 1110 79 105

Acid leachable % of total Cu 28 43 58 79 71 50 67

hydroxide) is an important process which determines metal release into solution. Typical copper concentrations in dissolved and particulate phases in the Fly River are shown in Table 9.2. As can be seen, the proportion of ASM-Cu increases with distance from the mine. When river water samples from close to the mine were incubated under controlled conditions (temperature 301C, pH: 8.0–8.3), signiﬁcant dissolved copper release was observed which was accompanied by an increase in ASM-Cu, indicating further oxidation of copper sulﬁdes (Figs. 9.15 and 9.16). This experiment clearly indicated the role of in-river oxidation in the mobilization of copper from mine-derived particulates. Dissolved copper release and copper sulﬁde oxidation were markedly lower in sterile control samples demonstrating biological mediation (Figs. 9.15 and 9.16). Attempts to isolate the bacteria responsible for copper mineral oxidation were unsuccessful; however, laboratory investigations provided strong evidence that the copper sulﬁde oxidizing bacteria were chemolithotrophs (Simpson et al., 2005). The results suggest that such bacteria are naturally present in mineralized areas and are actively involved in the cycling of particulate copper between sulﬁde and hydroxy-carbonate forms, thus inﬂuencing the solubility and bioavailability of copper. It is well known that certain chemolithotrophic bacteria (e.g., Thiobacillus and Leptospirillum species) obtain energy from the oxidation of copper sulﬁde minerals at low pH (Nordstrom and Southam, 1997). However, these bacteria do not oxidize metal sulﬁdes at pH values of greater than 6 (Kuenen et al., 1992; Blowes et al., 1995, 1998). The results point to a hitherto unidentiﬁed process affecting copper sulﬁde mineral stability in tropical river systems. The experiments demonstrated extensive biologically mediated copper

338

S. C. Apte

35 Tabubil

30

Dissolved Cu (µg/L)

Dissolved Cu (µg/L)

35 25 20 15 10 5

Ningerum

30 25 20 15 10 5 0

0 0

4

8

12

16

20

0

24

4

8

15

16

20

24

15 Nukumba

13

Dissolved Cu (µg/L)

Dissolved Cu (µg/L)

12 Time (days)

Time (days)

10 8 5 3 0 0

4

8

12 16 Time (days)

20

24

13 10 8 5 ARM 380

3 0

0

4

8

12 16 Time (days)

20

24

Figure 9.15: River water incubation experiments showing the variation of dissolved copper concentrations with time. The symbols represent the mean of duplicate treatments for the unmodiﬁed waters (K), and waters amended with formalin (3) and a single ﬁltered water treatment ( ). The error bars encompass the range of duplicate measurements.

oxidation at pH 8.0–8.3 and provided a clear link between ASM-Cu concentrations and increasing dissolved copper concentrations. It is likely that these bacteria are particularly active in the Fly River system owing to the practice of in-river tailings disposal. The discharge of ﬁne particles high in copper sulﬁde provides an ideal environment for these bacteria to ﬂourish. The oxidation of sulﬁde minerals increases the reactivity of the solid copper phase, thereby making it easier for other processes (e.g., complexation by dissolved organic matter) to solubilize copper. The oxidation rates followed zero-order kinetics for the ﬁrst 150 h, after which the oxidation processes appeared to be strongly inhibited presumably following the consumption of easily oxidizable solid copper sulﬁde phases or armoring of particles with insoluble oxidation products such as Fe- and Cu- hydroxy/ carbonate phases, sulfur or sulfates, which are slow to dissolve at circumneutral pH. Particle-size fractionation studies indicate that the bulk of the oxidation occurs in the o20 mm fraction. Copper release rates calculated from experimental data obtained during an extensive study in 1995 (Apte et al., 1995b) are summarized in Table 9.3.

Biogeochemistry of Copper in the Fly River

339

700

700 Tabubil ASM-Cu (µg/g)

ASM-Cu (µg/g)

600 500 400 300

600 500 400 300

Ningerum

200

200 0

4

8

12

16

20

24

0

4

8

12

16

20

24

Time (days)

Time (days)

700

700 600

ASM-Cu (µg/g)

ASM-Cu (µg/g)

Nukumba

500 400 300 200

0

4

8

12 16 Time (days)

20

24

600 500 400 300 200

ARM 380 0

4

8

12 16 Time (days)

20

24

Figure 9.16: River water incubation experiments showing the variation of acid-soluble copper concentrations with time. The symbols represent the means of duplicate treatments for the unmodiﬁed waters (K), and waters amended with formalin to eliminate bacteria (3). The error bars encompass the range of duplicate measurements.

Table 9.3: Dissolved copper release rates measured in sample incubation studies (5-day period). Site Ningerum Konkonda Nukumba Fly at Bosset

Release rate, mg Cu/g sediment/day 0.6270.06 0.2770.07 3.1371.01 Not detectable

9.4.3. Adsorption and Desorption Reactions Involving Copper As noted earlier, copper release from mine-derived particles occurs on initial contact of mine sediments with river water, slow release as a response to mineral oxidation, and changes in CuCC. An important step change in dissolved copper concentrations also occurs at the junction of the Fly River and Ok Tedi. Laboratory mixing tests (Apte et al., 1995b) demonstrated

340

S. C. Apte

signiﬁcant release of dissolved copper from mine-derived particulates when the Ok Tedi and Fly rivers join at D’Albertis Junction. The release of copper is rapid (less than 10 min to attain equilibrium). Instead of a dilution effect (40% Ok Tedi to 60% Fly River), dissolved copper concentrations increased from the expected value of 4 to 16 mg/L. This is the result of copper desorption from mine-derived particulates which is mediated by copper complexation by dissolved organic matter introduced by the upper Fly River. Copper adsorption from solution onto suspended sediments is also important in the river system. The results of adsorption experiments conducted on unﬁltered water samples from the Fly River are shown in graphical form in Fig. 9.17 as a plot of the ﬁnal dissolved copper concentration (after 18 h equilibration) versus the initial dissolved copper concentration (concentration measured in the ﬁltered controls). The straight lines in the plots indicate the anticipated dissolved copper concentrations if no adsorption had taken place. In all experiments, extensive adsorption of inorganic copper was observed. A striking feature of the plots was the similar ﬁnal copper concentration, irrespective of the initial amount of copper added. This illustrates the role of the sediments to ‘buffer’ dissolved copper concentrations and is a consequence of the large excess of particulate copper present in an exchangeable form. 9.4.4. Short-Term Variations in Dissolved Copper Concentrations OTML’s routine monitoring data for sites in the Ok Tedi and Fly River downstream of mine inputs are characterized by large ﬂuctuations in dissolved copper concentrations with time (spiky data). For a long time, there was much speculation as to whether these trends were real, or simply reﬂected contamination during sampling and analysis. In order to resolve this problem, a 12-month intensive monitoring study was conducted in 1999 at Nukumba on the Fly River, followed by a further period of 4 months of intensive monitoring at Bige (located in the Ok Tedi). Samples were collected on an hourly basis for most of the study. The resulting dissolved copper data set is one of the most intensive ever produced by OTML. The study conﬁrmed that the large variability inherent in OTML’s monthly monitoring data is indeed real (Fig. 9.18, Table 9.4). On some occasions, dissolved copper concentrations doubled within a period of several hours (Fig. 9.18). The study greatly altered the appreciation of the timescales of the processes affecting dissolved copper concentrations in the river system. The inherent variability of the dissolved copper concentration data indicates the pitfalls

Biogeochemistry of Copper in the Fly River

341

90

Final dissolved Cu (µg/L)

Konkonda TSS: 1150 mg/L

60

30

0 0

30

60

90

Initial dissolved Cu (µg/L) 90

Final dissolved Cu (µg/L)

Fly, Bosset TSS: 118 mg/L

60

30

0 0

30

60

90

Initial dissolved Cu (µg/L)

Figure 9.17: Adsorption of dissolved copper added to unﬁltered water samples from the river system. Samples were spiked with inorganic copper and equilibrated for 18 h prior to the measurement of dissolved copper. The straight lines indicate the expected dissolved copper concentrations if no adsorption had taken place.

342

S. C. Apte

(a) 25

dCu (µg/L)

20

15

10

5 02/11/1999 02/01/2000 02/03/2000 02/05/2000 02/07/2000 02/09/2000 Date

30

(b)

dCu (µg/L)

25

20

15

10

5 21/11/2000

19/12/2000

16/01/2001

13/02/2001

Date

Figure 9.18: (a). Time series plot of hourly dissolved copper concentrations at Nukumba (352 days of data with a range of 6–22 readings taken per day). (b). Time series plot of hourly dissolved copper concentrations at Bige. Ninety-ﬁve days of data with approximately 18 readings taken per day are depicted.

Biogeochemistry of Copper in the Fly River

343

Table 9.4: Intensive monitoring study: summary statistics of dissolved copper concentrations (mg/L) of each of the months of sampling at Nukumba. Month January February March April May June July August September October November December All data

Minimum

Median

Maximum

Mean

Variance

5.0 7.9 6.4 8.7 9.0 9.3 8.5 6.4 6.9 11.7 6.1 6.8 5.0

13.0 13.0 14.0 13.2 14.2 13.5 15.4 15.9 12.0 11.5 13.2 17.3 13.6

20.0 21.1 25.8 22.9 26.2 22.1 19.6 23.4 22.5 27.4 25.4 21.0 27.4

13.6 13.4 14.4 13.9 15.1 16.0 12.2 11.8 13.5 17.8 13.5 12.9 13.9

8.3 4.6 15.4 8.7 8.1 5.0 3.6 4.7 11.4 12.8 7.7 5.4 9.9

associated with the interpretation of trends in dissolved copper concentrations based on low-frequency sampling (e.g., monthly monitoring data). Failure to take into account the variability that occurs at timescales of hours and days may result in erroneous or biased interpretation. The application of time series analysis techniques (Shao et al., 2002) failed to detect any noteworthy cyclic or noncyclic trends in the data at hourly, daily, weekly, and monthly timescales. No strong correlations between dissolved copper concentrations and river level or mine production data such as waste rock output and waste rock type could be found. However, some correlations between dissolved copper concentrations and rainfall were observed at Bige when time lags were incorporated into predictive regression models. This could involve the rainfall-related release of dissolved copper from stored dredge material as well as from deposited sediments located on the river banks and adjacent ﬂoodplain. The climatic conditions encountered during the studies described earlier were characteristic of a ‘wet’ phase. As a result, the data set does not provide information on dissolved copper concentrations that may develop in the system under ‘dry’ climatic conditions. OTML’s monitoring data following the El Nin˜o period of 1996–1997 provides a very good indication of climatic effects, and shows elevated dissolved copper concentration in the Ok Tedi and Middle Fly following the cessation of the dry period. For instance, concentrations as high as 48 mg/L were recorded at Obo on the Middle Fly

344

S. C. Apte

during early 1998. These elevated copper concentrations may result from aerial oxidation of exposed riverine sediments and release of copper on rewetting. Indeed, recent ﬁeldwork by OTML has conﬁrmed this with dissolved copper concentrations up to 44,000 mg/L in porewaters collected from an area subject to acid generation (B. Bolton, personal communication, May 2006). This illustrates the importance of catchment processes and the role of stored sediments as a potential source of dissolved copper. The impacts of dry weather events (including El Nin˜o) on dissolved copper concentrations still remains a concern for OTML. The possible causes of copper spikes may also include reactive minerals discharged infrequently from the mine (e.g., skarn ores) and ARD-related groundwater inputs. The speciation of copper during spike events has not been determined, but copper spikes are likely to be highly bioavailable especially when CuCC concentrations are exceeded.

9.5. Copper Cycling in Floodplain and Off-river Water Body Environments The 2,400 km2 ﬂoodplain of the Middle Fly River is relatively narrow, widening from ca. 4 km near Kiunga to over 14 km near Everill Junction. The inner ﬂoodplain contains numerous oxbow lakes whereas the outer ﬂoodplain is composed of backswamps and blocked valley lakes. Before extensive forest die-back occurred, the ﬂoodplain was covered with dense vegetation with little exposed sediment. The high annual rainfall in the headwaters of the Fly River and Ok Tedi result in periodic ﬂooding of the Middle Fly ﬂoodplain, as well as inundation of over some 118 ORWB. Typically, ﬂoodplain areas can be covered to a depth of 2–3 m for periods of up to 8 months before the waters recede. In the early 1990s, an investigation conducted by OTML conﬁrmed signiﬁcant deposition of mine-derived sediments as a network across the ﬂoodplain during ﬂooding events (Day et al., 1993). ORWB are also recipients of mine-derived sediments during these periods of ﬂoodplain inundation, particularly when there is high river ﬂow and low standing water on the ﬂoodplain. Subsequent studies were conducted to assess the stability of mine-derived copper deposited on the ﬂoodplain and in ORWB. A key concern was the stability of the deposited particulate copper and the potential for copper release into solution. This problem was addressed by use of sediment porewater dialysis peepers (Teasdale et al., 1995) which were deployed to measure in situ depth proﬁles of dissolved copper across the

Biogeochemistry of Copper in the Fly River

345

sediment/water interface (Apte et al., 1993, 1995b). This approach allowed a ﬁeld-based assessment of copper mobilization. Field deployments of porewater peepers were carried out during June 1994 and January 1995 at established ORWB and ﬂooded ﬂoodplain sites. Details of porewater peeper sampling procedures may be found in the review by Teasdale et al. (1995). Brieﬂy, each peeper comprised a narrow rectangular acrylic block with 96 water-ﬁlled slots located 1 cm apart down the length of the peeper. Both sides of the peeper were covered by a porous polysulfone membrane (0.45 mm pore size). At each site, a number of peepers were positioned in the sediments with about 75% of the peeper chambers below the sediment–water interface. The peepers were then allowed to equilibrate for a period of at least 14 days, during which time dissolved species diffused across the semipermeable membranes, after which they were recovered and sampled immediately. Sediment cores were also collected at each site. The particulate copper proﬁles of sediment cores at each site (Fig. 9.19) conﬁrmed elevated particulate copper concentrations in the surﬁcial sediments at all sites. Mine-derived copper signatures were detectable to depths of over 30 cm at many sites (Fig. 9.19). Dissolved metal concentrations of the overlying waters are shown in Table 9.5. During the time of the study, the ﬂoodplain was covered with river water (as indicated by the dissolved calcium concentrations) with copper concentrations typical of the Middle Fly. Peeper proﬁles of dissolved copper, iron, and manganese are shown in Figs. 9.20 and 9.21. Porewater copper concentrations were generally lower than dissolved copper concentrations found in the overlying water column. The peeper proﬁles indicated a net downward ﬂux of copper from the water column to the sediments. These proﬁles clearly indicate that there was little detectable mobilization of particulate copper in porewaters, rather, the key process was removal of dissolved copper from the water column. Dissolved iron and manganese concentrations were consistently higher in the deeper sediment porewaters and decreased toward the sediment–water interface. This is consistent with the bacterially mediated reductive dissolution of iron and manganese oxyhydroxides in the anoxic sediment layers (Davison, 1993; Hamilton-Taylor et al., 1996). At some sites, it is evident that the maximum production of these dissolved species occurs close to the sediment–water interface. This probably reﬂects the natural abundance of bacteria capable of reducing particulate iron and manganese to their soluble lower oxidation sites. Dissolved iron and manganese concentrations decreased dramatically above the sediment–water interface but were detectable in measurable quantities indicating signiﬁcant diffusion of these species from the porewaters to the overlying waters (Table 9.5).

346

S. C. Apte

Floodplain 15

Depth (cm)

0-5 5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-45 45-50

0

200

400 600 800 1000 Particulate Copper (µg/g)

1200

Lake Pangua: tie-channel

Depth (cm)

0-5 5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-45 45-50

0

200 400 600 800 Particulate Copper (µg/g)

1000

Lake Pangua: floodplain 0-5

Depth (cm)

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-45

0

100

200

300

400

500

600

Particulate Copper (µg/g)

Figure 9.19: Particulate copper proﬁles in ﬂoodplain and ORWB sediment cores (January 1995). Refer to Fig. 9.1 for site locations. At some sites, very localized high-concentration spikes of dissolved copper were observed at depth. These cannot be attributed to contamination as they were observed in replicate peeper measurements. The precise reason for this localized copper mobilization at depth is currently unknown. These events are very localized and do not give rise to signiﬁcant ﬂuxes of dissolved copper to the overlying waters.

Biogeochemistry of Copper in the Fly River

347

Table 9.5: Dissolved metals and suspended solids in overlying waters during the 1995 peeper deployment. Site

Sample location

Ca (mg/L)

Cu (mg/L)

Fe (mg/L)

Mn (mg/L)

Floodplain 15 (Water depth 1.8 m)

Surface Middle Bottom

26.4 26.5 26.5

19.8 18.7 17.9

79 69 106

o2 o2 3

L. Pangua: Tie-channel (Water depth 1.6 m)

Surface Middle Bottom

24.3 23.9 24.6

15.6 15.7 16.5

58 63 59

14 o2 o2

L. Pangua: Floodplain (Water depth 1.6 m)

Surface Middle Bottom

23.0 22.5 22.4

12.2 11.2 10.6

63 60 65

o2 o2 o2

The ﬂooded ﬂoodplain and ORWB which have received mine-derived inputs appear to act as a sink for copper rather than a source to the overlying water column. The absence of any net ﬂuxes of copper from the sediments to the water column, despite very high particulate copper concentrations, suggests that sediment-bound copper is stabilized in an unreactive form (possibly as sulﬁdes) in the deeper anoxic sediments. A proposed model of copper cycling on the ﬂoodplain and ORWB is shown in Fig. 9.22. It is hypothesized that the natural iron cycle (and possibly the manganese cycle) plays a major part in controlling dissolved copper concentrations under wet conditions. The removal of dissolved copper from the water column is considered to take place by adsorption onto iron(III) oxyhydroxides generated close to the sediment–water interface. Based on the peeper proﬁles, quantitative estimates of the times for a major reduction in copper concentration in the ORWB were calculated (Apte et al., 1995b). It was estimated that it would take between 7 and 28 years for dissolved copper concentrations to return to pre-mining levels. It should be noted however, that the studies described in this section were conducted under ﬂooded ﬂoodplain conditions. The stability of copper in ﬂoodplain sediments during prolonged periods of drought and release of copper during postdrought ﬂooding has not been studied. Hot, dry climatic conditions may lead to the aerial oxidation of reduced particulate copper phases (e.g., copper sulﬁdes), which in turn increases the potential for subsequent release into solution on contact with ﬂood waters. Such release may be episodic and may well be greatest during the ﬁrst ﬂood event after a

Copper -25

-25

25 50

Peeper 3

Depth (cm)

Depth (cm)

Peeper 2 0 25 50

75 5

10

15

20

25

25 50

75 0

0

S. C. Apte

Depth (cm)

Peeper 1 0

348

-25

75 0

5

10

Cu (ug/L)

15

20

25

0

5

10

Cu (ug/L)

15

20

25

Cu (ug/L)

Iron -25

-25

-25

Peeper 2

25 50 75

Peeper 3

0

Depth (cm)

0

Depth (cm)

Depth (cm)

Peeper 1

25 50

20

40

60

80

100 120 140

25 50 75

75 0

0

0

20

40

Fe (mg/L)

60

80

100 120 140

0

20

Fe (mg/L)

40

60

80

100 120 140

Fe (mg/L)

Manganese -25

-25

-25

Peeper 3

0 25 50

Depth (cm)

Peeper 2

Depth (cm)

Depth (cm)

Peeper 1 0 25

75

75 0

1

2

3

4

Mn (mg/L)

5

6

7

25 50

50

75

0

0

1

2

3

4

Mn (mg/L)

5

6

7

0

1

2

3

4

5

6

7

Mn (mg/L)

Figure 9.20: Porewater depth proﬁles obtained using dialysis peepers of dissolved copper, iron, and manganese at the Floodplain 15 site (January 1995).

Copper -25

-25

-25

Peeper 2

25 50 75

Peeper 3

0

Depth (cm)

0

Depth (cm)

Depth (cm)

Peeper 1

25 50 75

0

10

20

30 40 Cu (µg/L)

50

60

70

0 25 50 75

0

10

20

30 40 Cu (µg/L)

50

60

70

0

10

20

30 40 Cu (µg/L)

50

60

70

Iron -25

0 25 50

-25

Peeper 2

Depth (cm)

Depth (cm)

Peeper 1

0 25

75

0

10

20

30

40

50

25 50

50

75

Peeper 3

0

75

0

10

Fe (mg/L)

20

30

40

50

0

10

Fe (mg/L)

20

30

40

50

Fe (mg/L)

Manganese -25

-25

-25

25 50

Depth (cm)

Depth (cm)

Depth (cm)

0

0 25

75

75 2

4

6

Mn (mg/L)

8

10

0 25 50

50

0

Peeper 3

Peeper 2

Peeper 1

75 0

2

4

6

Mn (mg/L)

8

10

0

2

4

6

8

10

Mn (mg/L)

349

Figure 9.21: Porewater depth proﬁles obtained using dialysis peepers of dissolved copper, iron, and manganese at the Lake Pangua tie-channel site (January 1995).

Biogeochemistry of Copper in the Fly River

Depth (cm)

-25

350

S. C. Apte

Figure 9.22: Conceptual model of copper cycling in ORWBs and areas of the ﬂoodplain subject to frequent inundation. prolonged dry period. Clearly, extensive copper release during the ﬁrst ﬂoods that follow a prolonged El Nin˜o event is a potential cause for concern. Riverine-dissolved copper concentrations in the middle Fly are likely to be at their highest under this climatic scenario.

9.6. Copper Cycling in the Fly River Estuary 9.6.1. Background to the Estuary The Fly River Estuary and delta region (Fig. 9.23) cover an area of approximately 7,100 km2 (Salomons and Eagle, 1990). The estuary is split into southern, northern, and far northern channels and is partially stratiﬁed with an average water depth of less than 8 m. Between 60 and 80% of the freshwater ﬂows through the southern channel (Wolanski et al., 1995), and saline water usually intrudes no further than the junction of the three channels. The estuary is subject to large tides and is very turbid (Wolanski and Eagle, 1991). The sediment inﬂow to the estuary before mining commenced was estimated to be 85 106 tonnes/yr and has increased with mining to 120 106 tonnes/yr, about 90% of which moves as suspended load (Wolanski et al., 1992). As a result of mine-derived sediment inputs, riverine end-member concentrations of dissolved and particulate copper in the Fly River Estuary have increased since pre-mining times from approximately 0.5 to 5 mg/L and

Biogeochemistry of Copper in the Fly River

351

Bamu Estuary Site TM

Feb 93 July 93 March 94

0

40 km

Figure 9.23: Map of the Fly River estuary showing sampling locations.

40 to 300 mg/g, respectively (Apte et al., 1995b). From the start of mining, the estuary was noted as a potential sink for mine-derived sediments and an important transition zone in terms of the physical and chemical processes that may affect copper mobilization. The processes affecting the behavior and speciation of copper in the Fly River Estuary were subsequently investigated through a combination of ﬁeld studies and laboratory modeling experiments carried out during the ﬁrst half of the 1990s. Four extensive ﬁeld surveys were conducted over the period February 1993 to March 1994. The ﬁrst two surveys focused on determining surface-waterdissolved copper concentrations throughout the estuary, and the chosen sampling sites represented a compromise between a good spatial coverage of the estuary (e.g., all three channels) and the sampling waters over the full salinity range (Fig. 9.23). The March 1994 survey focused on determining

352

S. C. Apte

depth proﬁles of dissolved and particulate copper at key locations. In March 1995, additional ﬁeld studies were conducted to elucidate the mechanism of copper removal in the low-salinity zone of the estuary. In order to avoid sample contamination and erroneously high results, ultratrace sampling and analysis protocols were used throughout the study. To avoid metal contamination, surface water samples were collected using a small dinghy positioned approximately 100 m upstream of the main ship. Surface water samples were collected from a depth of approximately 0.5 m below the surface. Suspended sediment proﬁles obtained during the surveys (Fig. 9.24) illustrate the turbid nature of the Fly River Estuary with concentrations ranging from 2 to 2,100 mg/L. The pH of estuarine waters varied in a consistent manner between surveys with a marked increase of 0.1–0.4 pH units in the 0–1 salinity range followed by a more gradual increase over the remaining salinity range (Fig. 9.25). Suspended particulate copper concentrations dropped sharply from greater than 200 to 100 mg/g in the 0–1 salinity region and then dropped to a concentration of 57714 mg/g at salinities W12 (Fig. 9.26). This sharp decrease in particulate copper concentration in the low-salinity zone of the estuary was attributed to a rapid and efﬁcient mixing of mine-derived particulates with natural suspended particulate material low in copper. In July 1993, the decrease in particulate copper concentration was coincident with the presence of a turbidity maximum in the low-salinity region of the estuary; however, in the March 1994 survey, a turbidity maximum was not visible and yet the same sharp decrease in particulate copper concentrations was observed. This indicated that there were sufﬁcient estuarine particulates in solution to dilute mine-derived particulates at all tidal states. The suspended particulate copper concentrations in midestuary are comparable with particulate copper concentrations of 12–71 mg/g in benthic sediments from the Fly River Estuary (Alongi et al., 1991; Baker and Harris 1991). It should be noted that particulate metal concentrations vary with changes in particle size. Generally, metal concentrations are inversely related to particle size. Dissolved copper displayed a remarkably consistent relationship with salinity throughout the surveys (Fig. 9.27). A rapid decrease in dissolved copper concentration was observed in the very low-salinity region of the estuary (0–3), which accounted for the removal from solution of typically 50% of the river-borne dissolved copper. Over the rest of the estuarine salinity gradient, dissolved copper could be explained by dilution alone (conservative behavior), decreasing in a linear fashion with increasing salinity. Despite variations in riverine copper concentrations, the proportion of dissolved copper removal observed over the three main surveys was quite

Biogeochemistry of Copper in the Fly River

353

2500

(a)

TSS (mg/L)

2000 1500 1000 500 0 0

5

10

15 Salinity

20

25

30

1000

(b)

TSS (mg/L)

800 600 400 200 0 0

5

10

15

20

25

30

Salinity 250

(c)

TSS (mg/L)

200 150 100 50 0 0

5

10

15

20

25

30

Salinity

Figure 9.24: Total suspended solids versus salinity: (a) February 1993, (b) July 1993, (c) March 1994 (all depths).

354

S. C. Apte

8.2 Feb-93

Jul-93

Mar-94

8

pH

7.8 7.6 7.4 7.2 7 0

5

10

15 Salinity

20

25

30

Figure 9.25: Surface water pH versus salinity: (’) February 1993, (7) July 1993, (K) March 1994.

300

Particulate Cu (µg/g)

Jul-93

Mar-94

200

100

0 0

5

10

15

20

25

30

Salinity

Figure 9.26: Particulate copper versus salinity. similar. No marked difference in dissolved copper behavior was observed between the three main channels and little variation of dissolved copper concentration with depth was noted that could not be accounted for by differences in salinity. Depth proﬁle samples (March 1994) collected at nine estuarine sites (Fig. 9.23) (data not shown) also failed to indicate any systematic increase or decrease in dissolved copper with depth that could be attributed to remobilization from sediments or scavenging onto suspended sediments (Apte et al., 1995b). In spite of the size of the estuary and the

Biogeochemistry of Copper in the Fly River

355

5 (a) Dissolved Cu (µg/L)

4

3

2

1

0 0

5

10

15

20

25

30

35

Salinity 5 Dissolved Cu (µg/L)

(b) 4

3

2

1

0 0

5

10

15

20

25

30

35

Salinity 12 Dissolved Cu (µg/L)

(c) 10 8 6 4 2 0 0

5

10

15

20

25

30

35

Salinity

Figure 9.27: Dissolved copper versus salinity: (a) February 1993, (b) July 1993, (c) March 1994 (all depths).

356

S. C. Apte

complexity of its physical structure, dissolved copper concentrations were largely determined by salinity and were not appreciably inﬂuenced by changes in suspended sediment concentration. Substantial removal of CuCC from solution was observed at low salinity (probably through precipitation of humic materials) (Fig. 9.28). The DOC versus salinity proﬁles (Fig. 9.29) also showed nonconservative behavior with removal occurring over the 0–15 salinity range. The differences in behavior between DOC and CuCC are consistent with the organic ligands comprising 20

CuCC (µg/L)

15

10

5

0 0

5

10

15

20

25

30

35

Salinity

Figure 9.28: Copper complexing capacity versus salinity (July 1993 survey). 4 Feb-93

Jul-93

DOC (mg/L)

3

2

1

0 0

5

10

15

20

25

30

35

Salinity

Figure 9.29: Dissolved organic carbon versus salinity: (’) February 1993, (7) July 1993.

Biogeochemistry of Copper in the Fly River

357

the CuCC being only a small subset of DOC. Mantoura and Woodward (1983) noticed similar relationships in the Severn Estuary, UK. Humic materials precipitate in the low-salinity zone, but DOC remained conservative (Mantoura and Woodward, 1983). Despite the elevated copper inputs, complexing capacity was still in excess of dissolved copper and copper organic complexes are the major solution species. The change in CuCC concentrations with salinity were similar to the changes observed in dissolved copper concentrations. This may involve precipitation of complexing ligands and removal from solution. 9.6.2. Mechanisms of Dissolved Copper Removal Laboratory simulations of estuarine processes were carried out in order to understand the relative importance of ﬂocculation and adsorption as controls of dissolved copper concentrations. The ﬂocculation of dissolved organic matter and iron which often occurs in the low-salinity regions of estuaries (Sholkovitz, 1976, 1978) may reduce dissolved copper concentrations through coprecipitation of copper (and possibly the coagulation of copper–organic complexes). Dissolved copper may also be adsorbed onto estuarine particulates, particularly in the turbidity maximum region. Metal adsorption processes may also be enhanced by the slight but consistent increase in pH in this region. Conversely, the increase in the concentration of calcium and magnesium with salinity may increase the competition for binding sites and lead to the desorption of copper from particulate material. In order to reduce sampling artefacts, all experiments were conducted onboard ship within hours of sample collection. Flocculation experiments were performed with both ﬁltered and unﬁltered river water. The experiments involved mixing known quantities of river water and seawater together to give a ﬁnal salinity of 6, followed by measurement of dissolved metal concentrations. The chemical characteristics of the samples used are summarized in Table 9.6. The tests indicated extensive precipitation of iron and appreciable coremoval of copper at low salinities (Table 9.7). The amount of iron removed from solution varied from 90% in the unﬁltered experimental treatment to 76% in the ﬁltered treatment. This suggests that the presence of natural particulate material enhances the coagulation of iron colloids. The precipitation of copper was less affected by the presence of particulate material with 23% removal occurring in the ﬁltered water experiment compared to 30% removal with unﬁltered solutions. The 30% removal in laboratory experiments compares to typically 50% removal as observed in the three ﬁeld surveys.

358

S. C. Apte

Table 9.6: Typical composition of low-salinity region samples (collected during March 1995). Parameter

Site Lewada

TM

176 o0.01 7.50 6.8 26.3 133 100 75

164 0.27 7.72 4.2 25.9 121 86.0 71

TSS (mg/L) Salinity PH Dissolved Cu (mg/L) Copper complexing capacity (mg/L) Particulate Cu (mg/g) Acid-soluble copper (mg/g) % ASM/pCu

Table 9.7: Results of the estuarine mixing experiments (24 h equilibration). Sample

Treatment

Dissolved Fe (mg/L)

Dissolved Cu (mg/L)

Filtered water

Control Final (n ¼ 5)

29.872.0 7.170.4

4.070.2 3.170.3

Unﬁltered water

Control Final (n ¼ 5)

56.671.4 5.570.3

4.770.2 3.370.1

Note: 100 mL aliquots of river water (Lewada) were mixed with 20 mL of seawater to give solutions with a ﬁnal salinity of approximately 6. The solutions were mixed on a bottle roller for 24 h, ﬁltered through 0.45 mm membrane ﬁlters, and subsequently analyzed for dissolved metals.

The adsorption of copper onto estuarine suspended material was also evaluated in laboratory experiments. These tests used samples taken from a brackish water location (Site TM, Fig. 9.23, 0.27 salinity) and at Lewada. The mean particle-size diameters, measured by laser particle sizing were 9.2 and 7.2 mm for Lewada and TM, respectively. The adsorption experiment (Fig. 9.30) showed that estuarine sediments have a strong afﬁnity for dissolved copper. A release of copper relative to the ﬁltered controls was observed in the zero copper addition experimental treatment which may be explained by disaggregation of particles during agitation and generation of colloidal particles small enough to pass through the 0.45 mm membrane ﬁlters. The proportion of dissolved copper removed from solution actually increased with the concentration of dissolved copper added, from 20% at the lowest added copper concentration to 52% at the highest spiked concentration. The observed increase in adsorptive afﬁnity with added copper concentration

Biogeochemistry of Copper in the Fly River

359

Final Dissolved Cu (µg/L)

25

20

15

10

5

0 0

5

10 15 Initial Dissolved Cu (µg/L)

20

25

Figure 9.30: Adsorption of copper onto estuarine particulates in the lowsalinity zone of the estuary. Note that the straight line indicates the expected dissolved copper concentration if no adsorption had taken place. Adsorption of copper onto estuarine suspended material was tested onboard the ship. 20 mL aliquots of ﬁltered river water (Lewada) were spiked with copper and equilibrated for 2 h. A 180 mL portion of sample TM (salinity 0.27, suspended sediment concentration 164 mg/L) was added to each container to give a ﬁnal concentration of 200 mL. The samples bottles were placed on a bottle roller for 24 h, after which the solutions were ﬁltered and analyzed for dissolved copper. All samples and ﬁltered controls were prepared in triplicate. could be explained by changes in copper speciation in solution, in particular the increase in concentration of inorganic copper species. Partition coefﬁcients, Kd ([particulate copper (mg/g)]/[dissolved copper (mg/L)] were calculated for the samples collected during the March 1994 survey. Above salinity W5, there was a signiﬁcant increase in Kd with salinity (Fig. 9.31) which implies that the adsorptive afﬁnity of particulate material increased with increasing salinity. This cannot be explained by any of the processes identiﬁed during the ﬁeld and laboratory modeling studies. Despite being used widely in estuarine studies, the partition coefﬁcient approach has several criticisms. The speciation of copper is ignored, and this form of analysis assumes that the particulate copper is fully exchangeable with the dissolved fraction which is not necessarily the case. For example, the ASM-Cu concentrations measured on the estuarine particulates (Table 9.6) indicate that about 70% of the particulate copper is present in an oxidized (exchangeable) form. Not all of this fraction will be exchangeable with the solution phase.

360

S. C. Apte

100000 Jul-93

Mar-94

5

10

Kd (L/kg)

80000

60000

40000

20000

0 0

15

20

25

30

Salinity (‰)

Figure 9.31: Kd versus salinity (July 1993 and March 1994 data). Overall, the modeling experiments indicated that the removal of copper in the low-salinity zone of the estuary can be attributed to a combination of ﬂocculation and adsorption onto estuarine particulates. A combination of efﬁcient particle mixing and the substantial removal of dissolved copper at low salinity signiﬁcantly attenuates the impacts of mine-derived sediments on the Fly River Estuary and surrounding coastal waters. The ability of the estuary to attenuate mine-derived impacts will depend on the size of the pool of natural suspended sediments available to dilute the mine-derived sediments. Sediment transport studies (Wolanski et al., 1996) suggest that there is signiﬁcant input of silt-sized particulates from the Gulf of Papua via the Northern Channel which replenishes the estuarine particulate pool. The magnitude of this ﬂux is much greater than the input of sediments from the Fly River (about sevenfold greater). These observations suggest that the capacity of the estuarine sediments to dilute the mine-derived sediments will not change signiﬁcantly over mine life.

9.7. Modeling of Copper Geochemistry Geochemical modeling involves the use of mathematical models based on fundamental thermodynamic and in some cases kinetic concepts to calculate

Biogeochemistry of Copper in the Fly River

361

the concentration and chemical forms of an element under speciﬁed conditions. In the context of aquatic systems, this includes the distribution between dissolved and particulate phases and the chemical forms (speciation) of the element that exist in solution. The prediction of the behavior of metals in river systems requires detailed knowledge of both the chemical processes, the hydrology of the system, and the movement of sediments. Attempts to model the chemistry of the Fly River date back to the late 1980s. The Fly River model (Kersten and Kerdijk, 1991) was an interactive computer model speciﬁcally developed as a management tool to assess the effects of mine tailings and overburden from the Ok Tedi mine. The model was based on the general purpose model CHARON (Chemistry Applied to the Research of Natural systems) developed by Delft Hydraulics (de Rooij and Kroot, 1991) which combined chemical equilibrium, slow reaction kinetics, and water–sediment transport submodels. The critical geochemical parameters determining the distribution of copper between dissolved and particulate phases were assumed to be: the adsorbing capacity of the suspended matter, the amount of exchangeable copper on the particles, and the chemical composition of the water (e.g., dissolved organic matter and pH). Calculations were performed in 1988 (OTML, 1988) and predicted the potential effects of tailings and waste rock disposal on the river system from 1987 to 2008. Dissolved copper predictions were satisfactory during the cyanide phase of mine operation because all available copper was complexed by cyanide and remained in solution. Dissolved copper concentrations in this period were therefore driven by river ﬂows. In 1991, following the transition from the ‘cyanide’ to ‘copper’ phase of mine life, deﬁciencies in the predictive capabilities of the Fly model were recognized. In particular, the model signiﬁcantly underpredicted dissolved copper concentrations at key monitoring sites. A major assumption of predictions was the concept of ﬁxed reactivity of mine-derived particulate copper. A ‘site speciﬁc’ version of the Fly model was used in 1992 (OTML, 1992) to predict annual average dissolved copper concentrations at Kuambit, Obo, and Ogwa during 1992–1996. The model was calibrated with mean monitoring data from the 1991 calendar year. It was recognized however that modeling could not be successful if the system was treated as a ‘black box.’ From August 1998 to May 1999, a 10-month intensive study was carried out to develop a predictive mechanistic model for the assessment of various mine-waste retention schemes. The investigations described earlier in this chapter provided a platform for a more sophisticated, mechanistic approach to modeling copper geochemistry. The resulting OkCSIRO geochemical model incorporated the key chemical processes of complexation,

362

S. C. Apte

partitioning, and sulﬁde mineral oxidation. The geochemical model was used to predict annual average concentrations of dissolved copper, dissolved inorganic copper (bioavailable copper), and particulate copper at sites in the river system for the period 1990–2052. The model considered eight sites along the Ok Tedi and Fly Rivers: Ningerum, Konkonda, Nukumba, Bosset, Obo, Ogwa, Burei Junction, and Lewada. The model utilized a single pH-dependent, site-speciﬁc partition coefﬁcient which described the partitioning of inorganic copper between solid and solution phases coupled with a single ligand complexation model describing copper binding by natural organic matter in solution. The model input data were: (i) Predicted suspended sediment concentrations at each site (split into tailings, waste rock, valley wall erosion, and natural riverine sediments at each site): These were obtained from the Parker sediment transport model (developed for OTML by Professor Gary Parker, University of Minnesota, St. Anthony Falls Laboratory). (ii) Information from the OTML mine plan on the future copper concentrations of mine tailings and waste rock: This information was derived from the mine production schedule data supplied by OTML which provided the tonnages and annual average copper content of waste rock and tailings. The o63 mm suspended sediment fraction predicted by the sediment transport model was assigned as the TSS concentration at each site. As noted in previous studies (Apte et al., 1999), particulate copper concentrations increase with decreasing particle size. The mine plan information on copper concentrations in bulk mine tailings and waste rock was therefore likely to underestimate the copper concentration of suspended sediments in the river system because of the presence of coarse particulates. In order to overcome this problem, a conversion factor was developed to predict the pCu o63 mm concentration in both the waste rock and tailings solids fractions. This was derived from the observed relationship between total copper and particulate copper in the o63 mm fraction of waste rock and tailings samples. A very consistent linear relationship was observed between total particulate copper and the copper present in the o63 mm fraction (Fig. 9.32). Both tailings and waste rock obeyed the same relationship. (iii) River ﬂow data (real and synthetic) for the period 1992–2052: The same ﬂow data was used in both the Parker sediment model and OkCSIRO. The synthetic ﬂow data was constructed from OTML hydrological monitoring data and comprised a predetermined sequence of annual hydrological data for 1996 (wet), 1992 (dry), 1994 (normal), and 1997 (El Nin˜o).

Biogeochemistry of Copper in the Fly River

363

(iv) Annual average values of water pH at each site: These were derived from OTML monitoring data (Table 9.8). (v) ASM-Cu: This is the concentration of particulate copper that is oxidized and most likely exchangeable with the solution phase. It is calculated from the total copper concentration using a site-speciﬁc reactivity parameter (i.e., the %ASM-Cu), which increases with distance from the mine. A ﬁxed value of %ASM-Cu was assigned for each site. The values were derived from previous CSIRO monitoring data where ASM-Cu and pCu 6000

pCu <63 µm (µg/g)

5000 4000 3000 2000 1000 0 0

500

1000

1500

2000

2500

3000

3500

4000

Total pCu (µg/g)

Figure 9.32: Relationship between total particulate copper and particulate copper in the o63 mm fraction (7 tailings samples, ’ waste rock samples). pCu (o63 mm) ¼ 1.36pCu[total]þ132 (r2 ¼ 0.927).

Table 9.8: Copper complexing capacity concentrations and pH values used in the model. Site Ningerum Konkonda Nukumba Fly Bosset Obo Ogwa Burei Junction Lewada

River distance from the mine (km)

CuCC (mg/L)

pH

68 127 191 470 587 609 754 889

17 18 24 25 26 27 29 25

8.3 8.1 8.0 7.8 7.8 7.8 7.9 7.9

364

S. C. Apte

were measured in the river system. The plot of ASM-Cu versus distance for the mine is given in Fig. 9.33. The maximum value of %ASM-Cu was set at 70%. This is the point at which armoring of particles through the buildup of other insoluble oxidation products prevents any further oxidation of copper sulﬁdes. (vi) Kd: This partition coefﬁcient describes the partitioning of inorganic copper between solution and solid phases. It was assumed that inorganic copper only was exchangeable with the particulate phase, and Cu-DOM complexes did not signiﬁcantly adsorb onto particles. The values of Kd were derived from laboratory adsorption studies (Apte et al., 1995b). For inclusion in a model to predict annual average concentrations, it was necessary to select values of Kd for each site that reﬂect mean annual values. It was found that only three Kd equations were needed to describe the observed partitioning phenomena. These were split into three sections: Ok Tedi, Middle Fly River, and Fly River downstream of the Strickland River. The Kd equations were not changed with time or climatic regime because the value of Kd is related to the physico-chemical properties of the mine-derived sediments and natural sediments in the system. The mine plan data indicated that on an annual average basis, the mixture of ore types being processed at the mill did not vary appreciably. Similarly, it can be reasoned that the composition of natural sediment will not vary appreciably on an annual average basis. Undoubtedly, short-term variations in dissolved copper are, in part, associated with ﬂuctuations in Kd values caused by changes in ore types. 80

% ASM

60

40

20

0 0

200

400

600

800

1000

Distance from mine (km)

El Nino

Dry

Medium

Wet

Figure 9.33: Percentage ASM-Cu values selected for use in the model.

Biogeochemistry of Copper in the Fly River

365

(vii) CuCC: This is the mean annual average CuCC at each site. The CuCC values used in the model are given in Table 9.8. (viii) Log Ku: It is the conditional stability constant for the formation of copper–organic matter complexes in solution: Ku ¼ [CuOrg] ¼ [CuInorg]/ [Org], where [CuInorg] is the dissolved inorganic copper concentration, [CuOrg] is the dissolved organically complexed copper concentration, and [Org] is the dissolved organic ligand concentration (ligand not bound to copper). Based on previous CSIRO studies (Apte et al., 1995b), a ﬁxed value of log Ku ¼ 9 was used for all sites. The model output was particulate copper, dissolved copper, and also dissolved inorganic copper for each site. The inorganic copper concentration was taken to represent the potentially bioavailable copper fraction. Free copper ion concentrations (Cu2þ) at each site were computed from the inorganic copper concentrations and the mean site values for pH and alkalinity. The model was validated using OTML annual average monitoring data for which the necessary mine plan data was available (1992–1997). The graph of observed versus predicted pCu concentrations (Fig. 9.34) indicated, with the exception of ﬁve outliers, an excellent ﬁt. This demonstrated the effective coupling of mine plan and sediment transport model data. The ﬁve major 2500

pCu Predicted (µg/g)

2000

1500

1000

500

0 0

500

1000

1500

2000

2500

pCu Observed (µg/g)

Figure 9.34: Final model output: predicted versus observed annual average particulate copper concentrations for the period 1992–1997 (r2 ¼ 0.864). The straight line is the 1:1 equivalence line. If the model were perfect, all data points would fall on this line.

366

S. C. Apte

model outliers were for all overpredictions of pCu and were all from the year 1996. The large positive bias was due to the high pCu concentrations of the waste rock data for this year. Model predictions for dissolved copper are shown in Fig. 9.35. Considering the complexity of the system, the model ﬁt was excellent. Sensitivity analyses were conducted using OTML annual average data for 1995 to determine the effect of variations in key input parameters on dissolved copper concentrations. As observed in laboratory studies, dissolved copper concentrations increased markedly with decreasing pH. A pH change from 8.0 to 7.4 resulted in a ca. 30% increase in dissolved copper concentration. It should be noted that the model was designed to predict the effects of pH over the range 7.4–8.2 (the pH range over which the partitioning reactions were studied). Currently, the model is not conﬁgured to predict dissolved copper concentrations in ARD-affected regions of the river system; this would require alteration of the copper partitioning and complexation submodels. Increasing the %ASM-Cu content of the particulate copper fraction resulted in increased dissolved copper concentrations (Fig. 9.36a). Variations in TSS concentration over the range anticipated in the river system had little effect on dissolved copper (Fig. 9.36b). The 25

dCu Predicted (µg/L)

20

15

10

5

0 0

5

10

15

20

25

dCu Observed (µg/L)

Figure 9.35: Final model output: predicted versus observed annual average dissolved copper concentrations for the period 1992–1997. The straight line is the 1:1 equivalence line. If the model were perfect, all data points would fall on this line.

Biogeochemistry of Copper in the Fly River

367

partition coefﬁcient Kd has a crucial role in determining dissolved copper concentrations (Fig. 9.36c). Kd values below 10 result in extremely high dissolved copper concentrations. Such Kd values would correspond to highly reactive solid phases (e.g., high proportion of amorphous copper hydroxide). Dissolved copper concentrations increased markedly with CuCC reﬂecting the role of copper complexation with natural organic matter in mobilizing copper from mine-derived particulates. The model predictions showed that implementation of the mine-waste retention options proposed in 1999 would not appreciably affect dissolved copper concentrations in the system (Fig. 9.37). It was predicted that mean annual dissolved copper concentrations in the system would plateau between 2000 and 2010. Dissolved copper concentrations in the Ok Tedi and Middle Fly would take over 40 years to return to pre-mining levels. Full model validation was achieved in 2005 when monitoring data for 2000–2005 was compared to the predictions made in 2000 (Fig. 9.38). As can be seen, the predictions made in 2000 were remarkably accurate. On an annual average basis, inorganic copper concentrations (an indication of copper bioavailability) were predicted to be below the levels of ecotoxicological concern. However, it was noted that frequent spikes of bioavailable copper would be observed in the systems throughout mine life and beyond. The model was not able to predict the occurrence of the spikes which occur on a timescale of hours and possibly minutes. The spike generation mechanism has still not been fully elucidated, but is most likely to involve inputs of reactive copper from the catchment (generated by ARD, in situ oxidation and wetting) and possibly changes in ore type processed by the OTML mill.

9.8. Overview Since the studies described in this chapter were conducted, two major changes to the system have occurred. Firstly, the occurrence of ARD in the Ok Tedi owing to the discharge of pyritic material from the mine and extensive forest die-back throughout the river system which has been caused by widespread riverbed aggradation. In terms of copper biogeochemistry, ARD will result in greater inputs of reactive copper from the river channel, ﬂoodplain, and groundwaters. These changes are likely to manifest themselves as pulse events in the system, e.g., wetting following a prolonged dry period, and are likely to result in increased concentrations of dissolved copper and greater bioavailability. The effects of forest die-back on copper

S. C. Apte

30

(a) 25

dCu (µg/L)

20

15

10

5

0 Ningerum

Konkonda

Nukumba

20%

Fly Bosset

40%

Obo

60%

Ogwa

80%

Burei Jn.

Lewada

100%

25

(b)

dCu (µg/L)

20

15

10

5

0 Ningerum

Konkonda

Nukumba

Fly Bosset

0.1

0.5

Obo

1

Ogwa

2

Burei Jn.

Lewada

4

70

(c) 60 50

dCu (µg/L)

368

40 30 20 10 0 Ningerum

Konkonda

Nukumba

10

Fly Bosset

100

1000

Obo

Ogwa

5000

Burei Jn.

10,000

Lewada

Biogeochemistry of Copper in the Fly River

369

Obo

25

There never was amine No sediment control is implemented

Dissolved Copper (µg/L)

Close mine end FY 2000

20

Dredge to 2001 & store tailings to 2010 Cease dredging operations in 1999 Dredge only to 2010, 15 Mt/a Dredge only to 2010, 19 Mt/a

15

10

5

0 1990

2000

2010

2020

2030

2040

2050

Calendar year

Figure 9.37: Model predictions for Obo. geochemistry are harder to predict. A reduction in vegetative cover will change the supply and composition of DOM to the river system. As noted in this chapter, complexation of copper by DOM plays an important role in mobilizing copper from mine-derived particulates and also in determining copper bioavailability. Recent speciation and bioavailability monitoring data points to changes taking place (see chapter by Stauber et al., this volume), which could be related to changes in DOM cycling brought on by forest die-back. Overall, the Ok Tedi project has provided some salutary lessons for environmental management. At the start of mining, environmental science did not possess the predictive capability required to conﬁdently assess mine impacts, as a result of which many of the assessments of environmental impact were overoptimistic. Knowledge gaps were identiﬁed and progress was made in understanding this complex system. However, because of the nature of mine operations, intense periods of study were often followed by long periods where further research was not a priority to OTML. Undoubtedly, lack of progress was compounded by logistical problems of

Figure 9.36: Model sensitivity analysis: (a) effect of varying the proportion of ASM-Cu on dissolved copper predictions; (b) effect of varying TSS concentration on dissolved copper predictions; (c) effect of varying Kd on dissolved copper predictions.

370

S. C. Apte

30

dCu (µg/L)

25

Error bars: ± Std. Dev.

OTML Mean dCu: 2000-05 OkCSIRO predicted mean dCu: 2000-2005

20 15 10 5 0 Ningerum

Atkamba

Nukumba Site

Obo

Ogwa

Lewada

Figure 9.38: Model validation for dissolved copper. Predictions made in 2000 compared to actual ﬁeld data collected between 2000 and 2005. working in a 1,000 km river system with no roads and where there is access by helicopter and boat only. Clearly, there was a need for a systemwide appreciation of mining impacts which extended over a much wider geographical area than initially considered. The Ok Tedi project highlights the need for a multidisciplinary approach to environmental assessment and a need for the integration of many disciplines including chemistry, hydrology, sediment transport, and biology. Through the research work on the Fly River funded by OTML, our understanding of copper geochemistry in tropical environments has increased appreciably; nevertheless, the environmental costs have been high. The Ok Tedi project is hopefully a never-to-be repeated chapter in the history of the global mining industry and, indeed, environmental management.

ACKNOWLEDGMENTS The majority of the work described in this chapter was funded by Ok Tedi Mining Ltd. The author wishes to thank the following people: the various staff members of the OTML Environment Department over the last 15 years, notably Herman Arugani, Dominic Bainok, Barrie Bolton, Don Carroll, Bobby Dara, Murray Eagle, Roger Higgins, Henry Kundapen, Monica Rau, Michael Ridd, Marshall Lee, Eric Schwamberger, Charles Tenakanai, Jat Tinkerame, Sel Towal, Jim Veness, Ken Voigt, Ian Wood, and Ben Yaru; people from CSIRO: Graeme Batley, Cheryl Davies, Phillip Ford, Kin Hong Friolo, Leigh Hales, Jacqui Lassau, Stuart Simpson, Sim Te, Robert

Biogeochemistry of Copper in the Fly River

371

Rowland, Peter Teasdale, Mathew Tiltman, and Ian Webster for their various technical contributions. The various crews of the MV Western Venturer for their expert assistance with ﬁeldwork. Kevin Brix, Yantao Cui, Bill Dietrich, Andrew Marshall, Gary Parker, Ross Smith, Wim Salomons, and Andrew Storey for their support and technical guidance. A special thanks goes to Geoff Day (formerly OTML) who was the driving force behind the copper geochemistry investigations described in this chapter.

REFERENCES Allison, J. D., Brown, D. S., & Novo-Gradac, K. J. (1991). MINTEQA2/ PRODEFA2, a Geochemical Assessment Model for Environmental Systems, Version 3.0 User’s Manual: US Environmental Protection Agency. US EPA/600/ 3-91/021, Athens, GA. Alongi, D. M., Tirendi, F., & Robertson, A. I. (1991). Vertical proﬁles of copper in sediments from the Fly Delta and Gulf of Papua (Papua New Guinea). Marine Pollution Bulletin, 22, 253–255. Apte, S. C., Batley, G. E., Day, G. M., & Wood, I. B. (1993). Factors inﬂuencing the partitioning and fate of copper in the Ok Tedi and Fly River systems: Year 1 report. OK Tedi Mining Ltd./CSIRO Joint Report, 109 pp. Apte, S. C., Benko, W. I., & Day, G. M. (1995a). Partitioning and complexation of copper in the Fly River, Papua New Guinea. Journal of Geochemical Exploration, 52, 67–74. Apte, S. C., Batley, G. E., Day, G. M., Schwamberger, E. C., & Wood, I. B. (1995b). Factors inﬂuencing the partitioning and fate of copper in the Ok Tedi and Fly River systems: Final report. OK Tedi Mining Ltd./CSIRO Joint Report, 139 pp. Apte, S. C., Simpson, S. L., & Webster, I. T. (1999). Geochemical modelling and prediction of copper speciation in the Fly River system. CSIRO Investigation Report CET/LH/IR 219R prepared for Ok Tedi Mining Ltd., 111 pp. plus appendices. Apte, S. C., Rowland, R., & Patney, H. (2001). Size distribution of copper complexing ligands in tropical freshwaters. Chemical Speciation and Bioavailability, 12, 79–88. Baker, E. K., & Harris, P. T. (1991). Copper, lead and zinc distribution in the sediments of the Fly River Delta and Torres Strait. Marine Pollution Bulletin, 22, 614–618. Blowes, D. W., Al, T. A., Lortie, L., Gould, W. D., & Jambor, J. L. (1995). Chemical, mineralogical and microbiological characterisation of the Kidd Creek mine-tailings impoundment, Timmins area, Ontario. Geomicrobiology, 13, 13–31. Blowes, D. W., Jambor, J. L., Hanton-Fong, C. J., Lortie, L., & Gould, W. D. (1998). Geochemical, mineralogical, and microbiological characterisation of a sulphide-bearing carbonate-rich gold mine tailings impoundment, Joutel, Quebec. Applied Geochemistry, 13, 687–705.

372

S. C. Apte

Chapman, B. M., Jones, D. R., & Jung, R. F. (1983). Processes controlling metal ion attenuation in acid mine drainage systems. Geochimica Cosmochimica Acta, 47, 1957–1973. Davison, W. (1993). Iron and manganese in lakes. Earth Sciences Review, 34, 119–163. Day, G. M., Dietrich, W. E., Apte, S. C., Batley, G. E., & Markham, A. J. (1993). The fate of mine derived sediments deposited on the middle Fly River ﬂoodplain of Papua New Guinea. Heavy Metals in the Environment. In: Proceedings of the 9th International Conference, 12–17 September, Toronto, Canada, pp. 423–426. de Rooij, N. M., & Kroot, M. P. J. M. (1991). Charon: A mathematical model to simulate biochemical processes: User’s manual. Delft Hydraulics report, 458 pp. Hamilton-Taylor, J., Davison, W., & Morfett, K. (1996). The biogeochemical cycling of Zn, Cu, Fe, Mn and dissolved organic C in a seasonally anoxic lake. Limnology and Oceanography, 41, 408–418. Johnson, C. A. (1986). The regulation of trace element concentrations in river and estuarine waters contaminated with acid mine drainage: the adsorption of Cu and Zn on amorphous Fe oxyhydroxides. Geochimica Cosmochimica Acta, 50, 2433–2438. Kersten, R. H. B., & Kerdijk, H. N. (1991). Fly Model. A water quality model for the Tedi-Fly river system. Manual and report. Delft Hydraulics report, 104 pp. Kuenen, J. G., Robertson, L. A., & Tuovinen, O. H. (1992). The genera Thiobacillus, Thiomicrospira and Thiosphaera. In: A. Balows, H. G. Truper, M. Dworkin, W. Harder, & K. H. Schleifer (Eds). The Prokaryotes – A Handbook on the Biology of Bacteria: Ecophysiology, Isolation, Identiﬁcation, Applications. Springer-Verlag, New York, pp. 2638–2657. Mantoura, R. F. C., & Woodward, E. M. S. (1983). Conservative behaviour of riverine dissolved organic carbon in the Severn estuary: Chemical and geochemical implications. Geochimica Cosmochimica Acta, 47, 1293–1309. Nimick, D. A., & Moore, J. N. (1991). Prediction of water-soluble metal concentrations in ﬂuvially deposited tailings sediments, Upper Clark Fork Valley, Montana, USA. Applied Geochemistry, 6, 635–646. Nordstrom, D. K., & Southam, G. (1997). Geomicrobiology of sulﬁde mineral oxidation. In: J. E. Banﬁeld, & K. H. Nealson (Eds). Geomicrobiology: Interactions between Microbes and Minerals. Reviews in Mineralogy and Geochemistry 35. Mineralogical Society of America, Washington, DC, pp. 361–390. OTML (1988). Sixth Supplemental Agreement Environmental Study, 1986–1988. OTML Environment Department Report, Vol. 1, 75 pp. OTML (1992). Preliminary predictions of suspended sediments, particulate and dissolved copper in the Fly River 1992–1996. OTML Environment Department Report, ENV/92-08, 9 pp. Pedersen, T. F., Mueller, B., McNee, J. J., & Pelletier, C. A. (1993). The early diagenesis of submerged sulphide-rich mine tailings in Anderson Lake, Manitoba. Canadian Journal of Earth Sciences, 30, 1099–1109.

Biogeochemistry of Copper in the Fly River

373

Rogers, N. J., Apte, S. C., Stauber, J. L., & Storey, A. W. (2005). Copper speciation and toxicity in the Fly River: A review. CSIRO Energy Technology Report No: ET/IR745R, 61 pp. Salomons, W., & Eagle, A. M. (1990). Hydrology, sedimentology and the fate and distribution of copper in mine-related discharges in the Fly River system, Papua New Guinea. Science of the Total Environment, 97/98, 315–334. Shao, Q., Harch, B., Apte, S. C., & Simpson, S. L. (2002). Temporal variability of dissolved copper concentrations in the Ok Tedi and Fly River. CSIRO Investigation Report ET/IR 501R prepared for Ok Tedi Mining Limited, 48 pp. Sholkovitz, E. R. (1976). Flocculation of dissolved organic and inorganic matter during the mixing of river water and seawater. Geochimica et Cosmochimica Acta, 40, 831–845. Sholkovitz, E. R. (1978). Flocculation of dissolved Fe, Mn, Al, Cu, Ni, Co and Cd during estuarine mixing. Earth and Planetary Science Letters, 41, 77–86. Simpson, S. L., Apte, S. C., & Davies, C. M. (2005). Rapid bacterial oxidation of copper sulphide minerals in tropical rivers. Environmental Chemistry, 2, 49–55. Svehla, G. (1987). Vogel’s Qualitative Inorganic Analysis. Wiley, New York, 310 pp. Teasdale, P., Batley, G. E., Apte, S. C., & Webster, I. T. (1995). Pore water sampling with sediment peepers. Trends in Analytical Chemistry, 14, 250–256. Wolanski, E., & Eagle, M. (1991). Oceanography and ﬁne sediment transport, Fly River Estuary and Gulf of Papua. In: Proceedings of. Coastal EngineeringClimate for Change. 10th Australian Conference on Coastal and Ocean Engineering, Auckland, December. Wolanski, E., Ridd, P., King, B., & Trenorden, M. (1992). Fine sediment transport, Fly River Estuary, Papua New Guinea. Australian Insitute of Marine Sciences report, 31 pp. Wolanski, E., King, B., & Galloway, D. (1995). Dynamics of the turbidity maximum in the Fly River Estuary, Papua New Guinea. Estuarine, Coastal and Shelf Science, 40, 321–337. Wolanski, E., Galloway, D., & Spagnol, S. (1996). Field and model studies of the fate of mine-derived contaminants in the Fly River Estuary. Australian Insitute of Marine Sciences report, 44 pp.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00410-2

Chapter 10

Speciation, Bioavailability and Toxicity of Copper in the Fly River System Jenny L. Stauber1,, Simon C. Apte1 and Nicola J. Rogers1 1

Centre for Environmental Contaminants Research, CSIRO Land and Water, PMB 7, Bangor Sydney, New South Wales 2234, Australia

10.1. Introduction The Ok Tedi Mining Ltd. (OTML) porphyry gold and copper mine at Mt. Fubilan is an open-pit operation located within the Star Mountains in the Western Province of Papua New Guinea and is situated near the headwaters of the Ok Tedi-Fly River system (Fig. 10.1). Discharge of mine tailings into the Fly River since 1984 has resulted in elevated concentrations of particulate and dissolved copper, with documented detrimental impacts on ﬁsh populations and other aquatic biota. Potential impacts of dissolved and particulate copper, released from tailings particles under riverine conditions, are just one of a number of impacts of the OTML mine, including increased turbidity, acidiﬁcation, channel aggradation, forest dieback, and associated habitat loss. While in situ monitoring, e.g., ﬁsh abundance and species diversity, in the Fly River system gives an indication of the impact on populations of these multiple stressors, it does not identify the impact of dissolved copper alone. The most useful data for assessing the effects of dissolved copper come from laboratory-based toxicity tests conducted on aquatic species representative of those found in the Fly River. OTML is currently required to monitor copper bioavailability and aquatic toxicity at four speciﬁed locations on the Ok Tedi-Fly River system. The monitoring program is based on the chemical measurement of copper

Corresponding author. Tel.: +61-2-97106808; Fax: +61-2-97106837;

E-mail: [email protected] (J.L. Stauber).

376

J. L. Stauber et al.

Bukrumdaing Tabubil Profile 3

Ok Tedi Ningerum Kiunga

Konkonda Nukumba

DÁlbertis Junction

Strickland River N

Oxbow 3 Fly at Bosset Lagoon Bosset Strickland River Obo

Everill Junction

0

Indonesia Papua New Guinea

Ogwa

50

100

Kilometres Burei Junction

Fly River Lewada

Figure 10.1: Map of the river system showing sampling locations.

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

377

speciation and bioassays using copper-sensitive algae and bacteria, originally intended to provide an early warning of copper toxicity to aquatic organisms resident in the Fly River system. Toxicity tests, together with chemical analyses and biological monitoring, provide several complementary lines of evidence of potential impact of copper in the Fly River system. The accompanying chapter ‘‘Biogeochemistry of copper in the Fly River’’ by Apte (2009) provides a detailed review of historical monitoring data of dissolved copper and copper complexation in the Fly River system. In this chapter, we review speciation data over the last 10 years where labile copper and toxicity in Fly River waters were determined concurrently on the same samples. We also assess recent trends in copper bioavailability in the Fly River system and highlight data gaps in our current understanding of copper effects on biota in the Fly River.

10.2. Overview of Copper Speciation and Bioavailability The potential impacts of copper in aquatic systems depend on the total concentrations of copper, the chemical speciation of copper, interactions of copper at the surface of the organism (e.g., ﬁsh gill, algal cell wall), and copper uptake into the organism, with either subsequent adverse effects or intracellular detoxiﬁcation. Ultimately, adverse effects on cells or individuals may lead to effects at the population or community level. The term ‘bioavailable’ is deﬁned as the fraction of total copper that an organism accumulates, i.e., the fraction of copper that binds to and traverses the cell membrane. The bioavailability of dissolved copper and ultimately its toxicity to aquatic organisms is primarily inﬂuenced by the speciation (chemical form) of copper in solution and its binding to receptor sites (e.g., ﬁsh gills or algal cell membranes), both of which depend on a range of water quality parameters such as pH, hardness, and dissolved organic matter (DOM) (Campbell, 1995; Markich et al., 2001). For copper to be taken up intracellularly where it may exert an adverse effect, it must ﬁrst bind with ligands present on the organism/cell surface. Lipid bilayer membranes are generally impermeable to charged and polar species, so copper must ﬁrst be bound to membrane transport proteins that carry copper (and the ligand) into the cell. For copper, the rate of formation of the copper-transport complex is often, but not always, relatively fast compared to copper transport through the membrane, so that a pseudoequilibrium exists between copper in the external medium and that bound to transport ligands. Thus copper uptake is largely dependent on the free copper

378

J. L. Stauber et al.

ion concentration, which is related to its free-ion activity. In some cases, the uptake of metals may be related to the concentration of kinetically labile inorganic species (free ion plus inorganic complexes) rather than the free metal ion (Hudson, 1998). Such kinetic control of metal uptake may be important when copper transport becomes diffusion limited. Generally, the free copper ion or weak or labile complexes that are able to dissociate at the cell membrane, are more bioavailable than copper in strong or inert complexes or adsorbed to colloidal and particulate matter. There are some exceptions, notably lipid-soluble copper complexes of organic ligands such as xanthates, used as mineral ﬂotation agents (Florence and Stauber, 1986; Stauber and Florence, 1987; Phinney and Bruland, 1994). These complexes are highly toxic because they can diffuse directly through cell membranes, allowing both copper and the ligand to enter the cell. Although lipid-soluble copper complexes generally represent less than 1% of total copper in natural waters, they may be of greater environmental signiﬁcance in mine-impacted waters. Over the pH range 7–8, typical of the Fly River system, copper speciation is dominated by complexation by natural organic matter, with copper– organic matter complexes comprising W90% of total dissolved copper concentrations (Apte and Day, 1993). In systems receiving elevated inputs of dissolved copper, the complexing capacity of natural organic matter may be exceeded, leading to an increase in the amount of inorganic copper in solution. In most natural waters, copper toxicity is lower than predicted by the dissolved metal concentration owing to the complexation effects of DOM (Apte and Day, 1993). However, DOM may also exert a direct effect (depending on pH) at the biological surface such as ﬁsh gills and algal cells, thereby modifying metal–organism interactions (Batley et al., 2004). Solution pH will also affect copper speciation, with low pH giving rise to larger proportions of the free copper ion. Carbonate concentrations also inﬂuence inorganic copper concentrations with copper–carbonate complexes dominating the inorganic copper pool under alkaline solution conditions. The binding of free copper ions to receptor sites on aquatic organisms is inﬂuenced by pH, and the concentration of competing cations such as calcium and magnesium (hardness) and sometimes other cations. Unfortunately, it is not possible to generalize on the role of pH and hardness on metal bioavailability. Different organisms have different responses. Although increasing pH can decrease the proportion of free metal ion in solution, there is also a decrease in competing Hþ, which can actually result in greater copper-cell binding and hence greater copper uptake and toxicity (Franklin et al., 2000; Wilde et al., 2006). Traditionally, high hardness has been accepted as having an ameliorating effect on the toxicity of copper and other metals (Erickson et al., 1996); however, recent work in our laboratory

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

379

has shown that hardness has little protective effect on copper-sensitive freshwater biota (Markich et al., 2005). A range of analytical chemical techniques (Tessier and Turner, 1995) and geochemical modeling approaches (Allison et al., 1991) have been used to measure and predict copper speciation in natural waters. Direct chemical measurement techniques include anodic stripping voltammetry (ASV), ionselective electrodes, ligand competition methods, ion exchange resins, e.g., Chelex, diffusive gradients in thin ﬁlms, and size-based separations (Tessier and Turner, 1995; Zhang et al., 1996; Batley et al., 2004). ASV has been successfully used to detect a labile (inorganic and weakly bound organic) copper fraction and to determine copper complexation capacity. While the labile copper measured at natural pH is operationally deﬁned by the measurement technique, it is generally thought to be related to bioavailable copper. There are exceptions, particularly when waters contain high DOM (Stauber et al., 2000). ASV-labile copper in the Fly River samples was measured using a Metrohom 746VA trace Analyser and a hanging mercury drop electrode, according to the method outlined in Apte et al. (2005). PTFE (Teﬂon) voltammetric cells were pre-equilibrated with the river water for at least 30 min, before being discarded. Buffered test solutions (20 mL) were weighed into the PTFE cells and sodium nitrate added (40 mL of 5 M). ASV-labile copper was determined after purging the samples for 5 min with nitrogen. Geochemical speciation models have been used for many years to calculate copper speciation at a range of water quality characteristics (Turner et al., 1981; Soli and Byrne, 1989). Most assume chemical thermodynamic equilibrium and until recently, their use has been limited by the lack of known binding constants for natural DOM and inability to account for metal adsorption. Franklin et al. (2000) showed that although copper speciation changes in a synthetic soft water, as calculated using geochemical speciation modeling, were negligible over the pH range 5.7–6.5, the toxicity of copper to a freshwater green alga varied by more than a factor of 20, indicating that speciation modeling may have limited ability in predicting copper bioavailability to aquatic organisms. Moreover, speciation models based on chemical thermodynamic equilibrium cannot predict metal uptake that is under kinetic control.

10.3. Bioassays While chemical measurement techniques and geochemical speciation modeling may detect or predict the different forms of copper in aquatic

380

J. L. Stauber et al.

systems and hence give some information about likely toxicity, they do not provide direct data on adverse biological effects. Bioassays or toxicity tests are generic tests that use living organisms as indicators of contaminant bioavailability in aquatic systems. Acute bioassays (short-term tests) typically measure organism survival over 96 h or a sublethal effect such as bioluminescence or enzyme inhibition. Chronic tests, such as inhibition of growth of microalgae, determine toxicity over several generations of cells. Such tests may be of long duration (weeks) or short-term in the case of single-celled algae (that divide approximately once per day). Copper may cause a decrease in ﬁnal cell numbers, a decrease in exponential growth rate, or an increase in the time to commencement of growth (lag phase). Because different organisms have different sensitivities to copper, batteries of toxicity tests, using sensitive species from different trophic levels, are usually used. Several bioassays using copper-sensitive bacteria and microalgae isolated from reference sites in Papua New Guinea, have been used over the past 10 years to monitor copper toxicity in the Fly River system, in conjunction with chemical analyses of dissolved copper, labile copper, and copper-complexing capacity. These bioassays respond to concentrations of biologically available copper in the low micrograms per liter range. These bioassays, developed speciﬁcally for application in the Fly River, are described in more detail below. 10.3.1. Bacterial Bioassays Bacteria have several attributes that make them useful as test organisms for the screening of metals in natural waters. They have relatively short life cycles, respond quickly to environmental change, and have enzymatic processes common to those of higher organisms. Bacterial processes are also of vital importance in the aquatic environment, mediating the degradation of organic compounds, biogeochemical cycling of nutrients, and transformations of trace metals (Lee et al., 1990). The response of bacteria to contaminants such as copper ranges from relative tolerance to extreme sensitivity, with some species exhibiting sensitivity at near-ambient metal concentrations. Sensitive bacterial bioassays were developed with an isolate from the Fly River, upstream from its conﬂuence with the Ok Tedi (Davies et al., 1998), and this was used in early monitoring of copper toxicity in the Fly River system in the mid-1990s. However, problems with maintaining consistent sensitivity of the bacterium in culture to copper necessitated the isolation of a temperate species (Erwinia persicinus) from Sydney for subsequent toxicity

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

381

monitoring. These bioassays, developed in our laboratory, are described below.

10.3.1.1. Growth inhibition bioassay The bacterial growth inhibition test utilizes a copper-sensitive freshwater bacterium, E. persicina, isolated from the Woronora River, New South Wales (Rogers et al., 2005). The bioassay is highly reproducible and exhibits a sensitivity to copper at o2 mg/L. The inhibition of cell division (growth) is measured over 48 h in various dilutions of freshwater samples supplemented with nutrient media, and compared to a pristine control water and a solution of known copper concentration. Growth is determined by optical density measurements (420 nm) 48 h after inoculation. The concentration of the test sample causing a 50% inhibition in bacterial growth and that at which no effect is observed are determined using standard statistical methods. Criteria for test acceptability include a ﬁnal optical density at 420 nm of 0.2070.05 for the control sample, a statistically signiﬁcant reduction in growth in the presence of 2 mg/L copper, and less than 20% coefﬁcient of variation in the controls. This type of bacterial growth test (with another species) has been used to examine the toxicity of copper in a number of freshwaters (Davies et al., 1998). The sensitivity of the bacterial isolate to other metals in unknown.

10.3.1.2. Radiochemical bioassay The uptake and assimilation of radiolabeled metabolites such as glucose by bacteria can also be used to quantify microbial responses to metal contamination (Gillespie and Vaccaro, 1978.). A radiochemical freshwater bioassay has been developed using the same metal-sensitive bacterial isolate used in the growth inhibition test (Rogers et al., 2005). A 50% reduction in the assimilation of radiolabeled glucose is observed at o1 mg/L copper, which corresponds well with the growth bioassay response. Due to the sensitivity of radiochemical techniques, it is possible to use environmentally relevant bacterial concentrations and short incubation periods thus minimizing the potential for changes in metal speciation during the bioassay. Furthermore, metal-impacted waters may be tested without the need to add growth-stimulating nutrients. A bacterial culture is grown in a deﬁned medium for 24 h. The cells are harvested by aseptic centrifugation, washed twice, and resuspended in sterile

382

J. L. Stauber et al.

synthetic water. Test solutions are inoculated with the washed starter culture to achieve a ﬁnal cell density of 105 cells/mL. The inoculated test solutions are equilibrated for 2 h followed by addition of an aliquot of D-glucoseUL-14C solution. A further 2 h is allowed for cellular glucose assimilation and then formalin is added to terminate respiratory activity. The cells are harvested by ﬁltration and the b-activity associated with each sample is measured using a liquid scintillation counter. The assimilation of 14C-glucose by copper-exposed cells is expressed as percentage assimilation compared to the control. Criteria for test acceptability include an initial optical density (420 nm) between 0.1 and 0.2 (corresponding to a cell density of 1.470.2 107 cells/mL), o20% coefﬁcient of variation in controls and a statistically signiﬁcant reduction in glucose assimilation in the presence of 1 mg/L copper compared to the control. 10.3.2. Algal Bioassays Microalgae are particularly important in tropical aquatic ecosystems, being responsible for most of the primary production at the base of the aquatic food chain. Recent stable-isotope studies have clearly shown the importance of microalgae in the food webs of the Fly River system Storey (2005). A bioassay developed in our laboratory measures the decrease in cell division rate (growth rate) of the unicellular alga Chlorella sp. in a 3-day exposure to toxicant. This species was originally isolated from Lake Aesake, Papua New Guinea. It is cultured axenically on a 12:12 h light/dark cycle (cool white ﬂuorescent light, 100 mmol photons m2 s1) at 271C. Cells in log-phase growth are used in the algal bioassays after washing and centrifuging three times to remove culture medium. The bioassay follows the OECD Guideline 201 (OECD, 1984) and the protocol of Stauber et al. (1994). Controls in synthetic soft water, together with Fly River water samples, each in quadruplicate, are prepared. Fifty milliliters of each is dispensed into 200 mL silanized (Coatasil, BDH) Erlenmeyer ﬂasks. To each ﬂask, 0.5 mL of 26 mM sodium nitrate and 0.05 mL of 1.3 mM potassium dihydrogen phosphate is added as nutrients. Each ﬂask is inoculated with 103 or 104 cells/mL of a prewashed algal suspension and incubated at 271C on a 12:12 h light/dark cycle at 100 mmol photons m2 s1 for 72 h. The pH in each ﬂask is measured on Day 0 and Day 3 of the bioassay. Cell densities in each ﬂask are determined daily for 3 days by counting cells in either a Coulter Multisizer II particle analyser with 70 mm aperture or a ﬂow cytometer. A regression line is ﬁtted to a plot of log10 cell density versus

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

383

time (h) for each ﬂask and the cell division rate (growth rate) per hour (m) determined from the slope. Cell division rates per day (3.32 m 24) are calculated for each water sample and expressed as a percentage of the control growth rate. The test is considered acceptable if the algal cell division rate in the soft water controls is 1.470.4 doublings/day and the variability in the controls is o20%. The reference toxicant copper (tested at at least three concentrations) is included in the bioassay to ensure that the algae are responding to a known toxicant in a reproducible way. The 72-h IC50 (i.e., the inhibitory concentration to cause a 50% decrease in algal cell division rate compared to controls) is calculated and the bioassay is acceptable if the copper IC50 is within quality control chart limits. 10.3.3. Limitations of the Algal and Bacterial Bioassays While bioassays have a number of advantages over other monitoring techniques such as the ability to detect bioavailability of mixtures of contaminants, high sensitivity, ecological relevance, and reproducibility, they also suffer from some limitations (Stauber and Davies, 2000). In particular, bioassays can underestimate copper toxicity if copper in the test solutions is depleted over the duration of the test, whether by adsorption to cells or production of cell exudates, which bind copper and reduce its bioavailability. Franklin et al. (2002) showed that copper toxicity to Chlorella sp. 12 decreased with increasing initial cell density from 102 to 105 cells/mL. With the recent application of ﬂow cytometry to algal (Franklin et al., 2000) and bacterial (Boswell et al., 1998) toxicity testing, there is now the capability to conduct bioassays at much lower cell densities (down to 102 cells/mL for algae) that are more typical of environmental concentrations. Such low cell densities have been used for toxicity testing of Fly River waters since 2004 to improve the environmental relevance of the bioassays. Different organisms respond differently to contaminants such as copper, hence the need for a range of toxicity test species. Recent work in our laboratory has shown that the effects of copper on Chlorella sp. were ameliorated in natural water samples, i.e., less toxic than predicted based on the free copper ion concentration, whereas the bioavailable copper fraction for bacteria was much greater than the free metal alone (Apte et al., 2005). A possible explanation for the observed amelioration of algal toxicity in the presence of natural DOM is the adsorption of DOM to the surface of the alga and the consequent blocking of metal receptor sites. In contrast, the bacteria could have responded to some forms of organically complexed

384

J. L. Stauber et al.

copper, which were bioavailable, hence copper toxicity may be greater than that predicted by the free metal ion. Thus the two different bioassays appear to be measuring different fractions of bioavailable copper.

10.4. Historical Toxicity Monitoring Data in the Fly River System Few data have been collected on the direct toxicity of copper on aquatic organisms indigenous to the Fly River system. The two most studied groups of aquatic organisms in the Fly/Strickland system are algae and large ﬁsh, with only limited data available on invertebrate species, despite the fact that they make up a signiﬁcant proportion of the aquatic fauna. There are no data available, at this time, on the toxicity of copper or Fly River water to planktivorous ﬁsh. 10.4.1. Invertebrate and Fish Bioassays Laboratory-based ecotoxicological tests of OTML mining wastes have been reviewed by Smith (1997). Toxicity testing using ﬁsh, prawns, cladocerans and mayﬂies showed that these species had sensitivities to dissolved copper comparable to published sensitivities for Australian species. Smith et al. (1990) reported the toxic effects of particulate copper to the freshwater prawns Macrobrachium rosenbergii and Macrobrachium handschini, and the catﬁsh Neosilurus ater. The test medium was synthetic Fly River water reconstituted from Woronora River (NSW) water with the addition of sodium bicarbonate to control pH and alkalinity. If the dissolved copper concentrations were kept below those likely to occur in the Fly River, little impact from particulate copper was observed at concentrations ranging from 14,300 to 15,100 mg Cu/g, much higher concentrations than those expected to occur at sites downstream from the mine. Recent studies of acute copper toxicity to juveniles (mean length 306721 mm) of a native barramundi species in the Fly River, reported 96-h LC50 values between 410 mg/L (initial measured concentration) and 270 mg/L (ﬁnal measured concentration) dissolved copper at pH 7.0–7.6 and 20.5 mg/L CaCO3 (Australian Water Technologies, 2002). The dilution water used for the test medium was carbon-ﬁltered Melbourne mains water, which would not have been representative of the natural water in the Fly River system.

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

385

10.4.2. Microalgal Bioassays In terms of copper ecotoxicology, microalgae represent the most studied group of aquatic organisms in the Fly River. It is clear that indigenous algal species are sensitive to relatively low concentrations of bioavailable copper. Early bioassays to investigate the potential toxicity of copper in the waters of the Fly River used a temperate green algal species Chlorella protothecoides obtained from the CSIRO Division of Fisheries Culture Collection, Hobart (Stauber and Critelli, 1993). These tests compared the growth of C. protothecoides in pristine laboratory water controls, matched to the same hardness, pH, nitrate and phosphate concentrations as Fly River water, to the growth of this alga in Fly River water at pH 7.9. Eight river water samples from Ok Tedi, Ok Mani, and the Fly River, together with two offriver water bodies (Bosset Lagoon and Lake Daviumbu) were tested (Fig. 10.1). No toxicity was observed for any of the 10 water samples from sites on the Fly River, despite the fact that they contained between 0.5 and 13 mg/L of dissolved copper, well above the lowest observed effect concentration (LOEC) of 2.5 mg Cu/L for this alga. Copper speciation measurements were not made on these early samples. In 1995, Stauber carried out a brief taxonomic survey of algal species present in the Fly/Strickland River system, with the aim of isolating suitable copper-sensitive species for use in the development of an algal growth inhibition bioassay with local tropical species. Tropical algae from eight sites in the Fly and Strickland River systems were collected in January 1995. Two sites, Lake Daviumbu and Lake Pangua in the Fly River ﬂoodplain, were potentially impacted by copper, while six reference sites in the Fly River oxbows near Kiunga or near the Strickland River (Lake Aesake and Oxbow Levamme) were not copper impacted. Good species diversity was found in all lakes at the time of this survey. Over 60 algae were identiﬁed to genus level, with the dominant taxa at all sites being green algae (Chlorophyceae), diatoms, and blue-green algae (Cyanophyceae) (Stauber and Apte, 1996). There was little difference in the genus composition or diversity between the impacted lakes in the Fly River system and the reference lakes in the Strickland River system. Thirty-ﬁve genera of green algae were identiﬁed, of which nine were only present in nonimpacted waters and two were present only in copper-impacted waters. The ﬁlamentous blue-green algae were distributed mainly in Lake Daviumbu, Lake Pangua, and Lake Aesake and were typical of the ﬂora of shallow tropical water bodies. Of the 14 genera of diatoms identiﬁed, only 2 (Cocconeis and Cyclotella) were present exclusively

386

J. L. Stauber et al.

in the Strickland River reference sites. The distribution of chrysophytes between samples from the Fly River and Strickland River systems was very different, although few genera were found. Four of the ﬁve genera identiﬁed were only found in nonimpacted sites and one genus (Bicosoeca) was found only in impacted lakes. In general however, the ﬂora of both the Fly River and Strickland River waters were similar, and broadly similar to algae of the Sepik River ﬂoodplain in Papua New Guinea (except for the absence of desmids) (W. Vyverman, personal communication). The diversity of euglenoids found (ﬁve genera) was typical of mixed water lakes that receive sediment-loaded water during the wet season and ﬂood plain-derived water during the dry season. Dominant algal genera in phytoplankton net samples during this 1995 survey are shown in Table 10.1. Algal abundance was also determined by cell counts in bottle-collected samples. All samples from both impacted and nonimpacted sites contained large numbers and diversity of blue-green algae, diatoms, dinoﬂagellates, green algae, and a cryptomonad (1–640 cells/mL). Stauber and Apte (1996) isolated 21 strains of microalgae, comprising 9 species, from reference sites in the Fly/Strickland River system. Most of these were green algae from Lake Aesake, together with two diatoms Table 10.1: Dominant algae in 20 and 63 mm net samples from the Fly and Strickland Rivers. Site

Impacted Lake Pangua

Nonimpacted Lake Aesake

Oxbow Levamme Downstream Oxbow Kiunga Upstream Oxbow Kiunga Oxbow 4 Kiunga Source: From Stauber and Apte (1996).

Algae in highest abundance Blue-green algae

Diatoms

–

Melosira Navicula Synedra

Scenedesmus

Dactylococcopsis Lyngbya

Navicula Synedra Diatoma –

Scenedesmus Ankistrodesmus

– Navicula Synedra –

Phacus –

Lyngbya Oscillatoria Dactylococcopsis Dactylococcopsis –

Green algae

–

Chlorella

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

387

(Nitzschia palea and Achnanthes sp). Unialgal (and some bacteria-free) cultures were established, and taxonomically identiﬁed. All isolated species were ubiquitous, widely distributed and identiﬁed in the original ﬁeld samples. Nine isolates of three common species (Chlorella, Chlamydomonas, and Monoraphidium arcuatum) were also tested for copper sensitivity in screening bioassays. All species were sensitive to copper, with 72-h IC50 values ranging from 7 to 17 mg Cu/L, LOEC values of o10 mg Cu/L and NOEC values of o5 mg Cu/L. A copper-sensitive green alga, Chlorella sp. 12, isolated from Lake Aesake on the Strickland River, was selected and subsequently used in the development of the site-speciﬁc growth rate inhibition bioassay, described earlier. Chlorella sp. 12 is particularly sensitive to copper, with copper toxicity depending on the initial cell density used in the bioassay (Franklin et al., 2002) and water quality parameters. For the standard test that used 2–4 104 cells/mL as the initial inoculum in Fly River monitoring up until 2004, the 72-h IC50 (mean72SD) was 8.574.6 mg Cu/L, with a no observed effect concentration (NOEC) at 5 mg Cu/L. More recent tests using lower cell densities (2–4 103 cells/mL) to improve environmental realism and sensitivity, have given slightly lower IC50 values of 5.574.5 mg Cu/L, with a LOEC at 4 mg Cu/L and NOEC at 3 mg Cu/L. This is of comparable sensitivity to copper as the temperate C. protothecoides used in the early bioassays of Fly River waters, suggesting that any changes in copper toxicity over time were unlikely to be due to changes in bioassay procedures. Recent work has shown that copper toxicity to Chlorella sp. 12 is highly dependent on water quality characteristics (Wilde et al., 2006). Copper toxicity in synthetic soft water decreased about 20-fold as the pH decreased from 8.0 to 5.5, and decreased about 80-fold as the DOC increased from 0 to 20 mg/L. Interestingly, hardness had no effect on copper toxicity to Chlorella sp. 12 in the synthetic water bioassays. A comparative study (Stauber and Apte, 1996) used C. protothecoides, Chlorella sp. 12, a bacterial bioassay (Apte et al., 1995; Davies et al., 1998), and ASV to measure the complexation capacity of water samples from seven sites on the Fly River, which had been collected over the preceding 12-month period. To measure complexing capacity from bioassays, water samples were spiked with additional copper, and assessed for toxicity in the usual way. The IC15 (i.e., the concentration of copper to cause a 15% reduction in growth rate) was taken as a measure of the copper-complexing capacity of the sample. All of the samples had copper complexation capacities (4–33 mg/L) in excess of their respective dissolved copper concentrations (1.8–17 mg/L), and again no toxicity was observed at any of the sites (Table 10.2). In fact,

388

J. L. Stauber et al.

Table 10.2: Complexation capacities of Fly River waters. Date

Jan-1995 Jan-1995 Jan-1995 Jan-1995 Jan-1995 Jan-1995 Jan-1995

Sample

pH

Konkonda Nukumba Lewada Obo Ogwa Burei Junction Kiunga

7.77 7.77 7.99 8.15 8.20 7.28 8.10

Dissolved Complexation capacity (mg/L) copper Electro- Bacterial C. protothecoides Chlorella (mg/L) chemical bioassay bioassay sp. 12 bioassay 5.3 9.3 1.8 17.9 8.0 11.3 2.3

9.1 23 27 20 19 33 18

– – – 22 27 – 11

8 14 4 30 17 25 10

– – – 23 17 36 9

Source: From Stauber and Apte (1996).

signiﬁcant growth enhancement of Chlorella sp. 12 was observed in the four samples tested with this species compared to controls.

10.4.3. Monitoring of Copper Speciation and Toxicity using Anodic Stripping Voltammetry and Bacterial and Algal Bioassays Total dissolved copper concentrations and copper speciation data, in conjunction with ecotoxicological bioassay data, have been periodically collected by OTML and CSIRO since 1993. The aim was to gain temporal as well as spatial information on copper speciation and bioavailability, and most importantly, to link the chemical form of dissolved copper with biological measurements of toxicity. Early surveys in 1993 and 1995, using C. protothecoides and Chlorella sp. 12, together with a bacterium isolated from the Upper Fly River (Isolate 37), found no toxicity at any sites in the Fly River, despite the fact that dissolved copper concentrations (0.5–17 mg/L) exceeded the test species LOEC values for copper. The absence of toxicity was attributed to DOM complexation lowering copper bioavailability. There were considerable logistical difﬁculties in the collection and transport of Fly River waters for testing in our Sydney laboratories. To avoid contamination, water samples had to be ﬁltered on arrival in Sydney, rather than on-site. Maintaining the temperature of the samples at 41C in transport was also difﬁcult, yet holding temperature was found to be a critical factor in determining the amount of labile copper in the samples. Labile copper decreased markedly within a few weeks of storage of the water samples at room temperature, so it is possible that, on occasions, particularly

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

389

in the early survey work, labile copper and hence toxicity was underestimated.

10.4.3.1. Speciation/toxicity survey 1996–1997 A more detailed study of copper speciation and copper toxicity was carried out over the period September 1996–June 1997 using both electrochemical speciation techniques and copper-sensitive algal and bacterial growth bioassays (Apte et al., 1997). A total of 48 river water samples in 7 surveys were analyzed for total dissolved copper, ASV-labile copper, copper complexation capacity, and algal and bacterial growth inhibition. Total dissolved copper concentrations ranged from 5.4 to 42.6 mg/L at sites downstream from the mine (41 samples) and from 1.0 to 3.8 mg/L at the riverine reference site Kiunga. The highest dissolved copper concentration was measured at Ningerum, the site closest to the mine. ASV-labile copper (0.5–8.3 mg/L) was detected in 26 of the 41 samples, but was not detected at the reference site. In eight of the samples, labile copper was detected even though the copper complexation capacity had not been exceeded. It was suggested that this was due to the presence of weakly bound copper–organic complexes which dissociated during the ASV analysis. Algal and/or bacterial growth inhibition was observed at four out of the seven sites sampled (at Ningerum six times and, D’Albertis Junction, Nukumba, and the Fly at Bosset each on one occasion). Samples from Ningerum were consistently toxic on six out of eight occasions. For both the algal and bacterial bioassays, dissolved copper in the river water samples was less toxic than for an inorganic copper calibration bioassay carried out in laboratory water, illustrating the role of the natural organic matter present in the natural waters in ameliorating copper toxicity. There was no general relationship between algal growth inhibition and labile copper, but inhibition was only observed when labile copper was present. These data could be split into two populations (Fig. 10.2): 1. No toxicity observed in the presence of labile copper. For these samples, labile copper concentrations were generally below the LOEC for inorganic copper in synthetic water (about 5 mg Cu/L) for Chlorella sp. 12, i.e., there was insufﬁcient labile copper to cause any growth inhibition. 2. More toxic than predicted by the labile copper concentration. In these samples, the concentrations of copper to cause growth inhibition were typically 5 mg Cu/L lower than the concentrations in the inorganic copper calibration bioassays. The most likely explanation for this was the presence

390

J. L. Stauber et al.

Algal Response (% Control)

120 100 80 60 40 20 0 0

5

10

15

20

25

30

35

40

45

Dissolved copper µg L-1

a)

Algal Response (% Control)

120 100 80 60 40 20 0 0 b)

2

4

8 10 6 ASV-labile copper µg L-1

12

14

Figure 10.2: Relationship between algal growth inhibition and (a) dissolved copper, (b) ASV-labile copper; (7) no inhibition, () inhibition, in 1996– 1997 toxicity monitoring. The solid line represents the inorganic copper calibration curve in synthetic water (Apte et al., 1997). of easily dissociable copper–organic complexes, or the presence of lipidsoluble copper complexes which have a greater toxicity than ionic copper (Florence and Stauber, 1986). Alternatively, other toxic metals may be present in the sample, i.e., the growth inhibition observed was not solely due to copper. In the ﬁnal survey conducted in June 1997, consistent trends in copper speciation and algal and bacterial growth inhibition were observed. Labile copper was detected at three sites closest to the mine input (Ningerum,

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

391

D’Albertis Junction, and Nukumba) and both bioassays showed growth inhibition at these sites. The linear relationship between algal growth inhibition and labile copper concentration (m ¼ 12.872.4, r2 ¼ 0.967) was similar to that for ionic copper in the calibration bioassay (m ¼ 16.370.2, r2 ¼ 0.999). Less labile copper was needed to elicit a similar degree of growth inhibition, the labile copper curve being displaced from the ionic copper calibration curve by about 5 mg/L. Again, this suggested that a fraction of the copper–organic complexes were bioavailable to the algae. Finally, copper uptake experiments were performed using three of the samples. Cellular copper concentrations increased with increasing amounts of labile copper in the samples, and both the algae and bacteria showed the greatest amount of copper uptake from the Ningerum sample.

10.4.3.2. Speciation/toxicity surveys 2004-onwards Since early 2004, monitoring surveys have analyzed a total of 53 water samples for labile copper and toxicity to bacteria (E. persicina) and microalgae (Chlorella sp. 12). To avoid the confounding factor of algal growth or bacterial respiration stimulation due to nutrients in the water samples, matrix-matched controls were prepared by the addition of the chelating agent EDTA to one subsample of each of the test waters (Apte et al., 1997). EDTA complexes metals such as copper, rendering them nontoxic. Microalgal growth or bacterial respiration in the EDTA-amended samples was then compared to growth/respiration in the non-amended samples. In this way, each water sample served as its own control. Of the 53 water samples tested, toxicity has been detected by either or both the algal or bacterial bioassays in 38 samples (72%). The sites and frequency at which toxicity was observed are shown in Table 10.3. Nine different sites were tested and toxicity was observed at seven sites (shown in Table 10.3). In general, toxicity was observed consistently at sites on the Upper Fly River and only intermittently at sites on the lower river or in off-river water bodies. No toxicity was observed at the control site Kiunga (two samples) or at Oxbow four (two samples) (not shown). This suggests that the number of potentially toxic samples has increased compared to the 1996–1997 surveys. Good agreement was obtained between the algal and bacterial bioassays during this period, with differences occurring only for 7 of the 53 samples. Although there was again no relationship between dissolved or labile copper and toxicity, labile copper was always present when toxicity was observed. For the microalga, no clear relationship was discernable between dissolved copper and algal growth inhibition (Fig. 10.3a). This plot also includes the

392

J. L. Stauber et al.

Table 10.3: Sites impacted by copper toxicity during the 2002–2004 routine speciation surveys. Site

Total number of Number of times analyzes toxic

Ningerum Nukumba Obo Ogwa Burei Junction Lewada Oxbow 5

Number of times labile copper present when toxic

10 10 10 4 3

10 10 10 4 2

10 10 10 4 2

9 3

2 0

2 0

Algal Response (% EDTA amended control)

120 100 80 60 40 20 0 0

5

a)

10

15

20

25

Dissolved copper µg

L-1

4

8

30

35

Algal Response (% EDTA amended control)

120 100 80 60 40 20 0 0 b)

2

6

10

12

ASV-labile copper µg L-1

Figure 10.3: Relationship between algal growth inhibition and (a) dissolved copper, (b) ASV-labile copper; (7) no inhibition, () inhibition. The solid line represents the inorganic copper calibration curve in synthetic water (Apte et al., 1997).

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

393

algal bioassay response to inorganic copper in laboratory synthetic freshwater, represented by the inorganic copper calibration curve. As in the previous surveys, the concentration of copper in the test waters required to elicit an inhibitory response was much higher than for inorganic copper (i.e., to the right of the inorganic copper calibration curve). Eight test waters exhibited toxicity in the region of the calibration curve, suggesting that there was little ameliorative effect of DOM in these eight waters. The relationship between algal growth inhibition and labile copper measured by ASV is shown in Fig. 10.3b. Four regions are discernable. First, concentrations of labile copper in the samples that were not inhibitory to algal growth were generally at or below 6 mg/L, close to the threshold lowest observable effect concentration for ionic copper for this bioassay. Therefore insufﬁcient labile copper was present to cause an inhibitory effect. Second, some of the waters tested exhibited toxicity in the region of the inorganic copper calibration curve suggesting that the measured labile copper was responsible for the observed toxicity. The rest, and majority, of the data fell either to the right of the calibration curve, indicating that the measured labile copper was less toxic than predicted, or to the left of the curve indicating greater than predicted toxicity. The latter may be due to the presence of additional metals or other toxicants in the water samples. Addition of the metal chelating agent EDTA to water samples consistently removed the observed toxicity, indicating that the toxic effects are most likely related to the presence of metal contaminants. These may be independently toxic, or have an additive, antagonistic or synergistic effect on copper toxicity to Chlorella sp. 12. Monitoring of other metals in the water samples, including aluminum, cadmium, chromium, iron, manganese, nickel, lead, and zinc, has shown that lead exceeded its guideline trigger value (ANZECC/ARMCANZ, 2000) of 3.4 mg/L at all sites and aluminum exceeded its guideline trigger value of 55 mg/L at Ningerum and Nukumba (on two occasions). There was no relationship between the concentrations of lead (11–12 mg/L at all sites on one occasion it was measured) and the observed toxicity, suggesting that lead was not responsible for the additional toxicity. In contrast, samples containing slightly elevated aluminum (58 and 56 mg/L) were more toxic than predicted from their labile copper concentrations, suggesting that aluminum may contribute to toxicity. However, the toxicity of aluminum alone to Chlorella sp. has not been determined, so it is difﬁcult to assess whether this may have contributed to the algal growth inhibition on several occasions. For the bacteria, increasing dissolved copper concentrations generally increased inhibition in the bacterial bioassay (Fig. 10.4a). This plot also

394

J. L. Stauber et al.

Bacterial Response (% EDTA amended control)

120 100 80 60 40 20 0 0

5

10

15

20

25

30

35

Dissolved copper µg L-1

a)

Bacterial Response (% EDTA amended control)

120 100 80 60 40 20 0 0 b)

2

6 4 8 ASV-labile copper µg L-1

10

12

Figure 10.4: Relationship between inhibition of bacterial respiration and (a) dissolved copper, (b) ASV-labile copper; (7) no inhibition, () inhibition. The solid line represents the inorganic copper calibration curve in synthetic water (Rogers et al., 2005). includes the bacterial bioassay response to inorganic copper in laboratory synthetic freshwater, represented by the inorganic copper calibration curve (Rogers et al., 2005). In many of the test waters, no inhibitory response was observed, despite the presence of dissolved copper concentrations up to 10 mg/L, illustrating the role of dissolved natural organic matter in ameliorating bacterial copper toxicity. However, the concentration of copper in the test waters required to elicit an inhibitory response was much higher than for ionic copper (i.e., to the right of the inorganic calibration curve). The relationship between bacterial inhibition and labile copper measured by ASV is shown in Fig. 10.4b. Increasing concentrations of labile copper were correlated with inhibition of the bacterial bioassay (R=0.703).

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

395

However, despite the presence of ASV-labile copper concentrations between 1.3 and 3.5 mg/L in six of the test waters, no bacterial toxicity was observed.

10.5. Spatial and Temporal Trends in Copper Toxicity in the Ok Tedi/Fly River System It is difﬁcult to determine temporal trends in labile copper and toxicity, as previous studies in which dissolved copper concentrations were intensively monitored at Nukumba and Bige (Shao et al., 2002), indicated large inherent variability in dissolved copper concentrations. With such large variability in dissolved copper at hourly, daily, weekly, and monthly timescales, it is not possible to draw ﬁrm conclusions about long-term trends in copper bioavailability. It is likely that the greatest copper toxicity is observed after events such as drying and wetting of the ﬂoodplain (e.g., the ﬁrst rainfall after a prolonged dry period). Event-based sampling is therefore required to better interpret trends. In addition, acidiﬁcation of the Fly River system, ﬁrst detected in 2000, confounds interpretation of toxicity changes over time, as pH and copper can interact to have additive or antagonistic effects on biota. Nevertheless, general comments on labile copper and toxicity data at four main sites in the Fly River system monitored since 1996 are given below.

10.5.1. Ok Tedi Dissolved copper concentrations at Ningerum ranged from 6 to 23 mg/L but appear to have decreased in the later survey period (Fig. 10.5a). Concentrations from April 2002 to July 2004 (6–7 mg/L) were signiﬁcantly (p ¼ 0.02) lower than from September 1996 to February 2002 (11–23 mg/L). Labile copper concentrations at Ningerum were generally low (o4 mg/L) during the 1996–1997 survey period, although algal toxicity was observed on three occasions (Fig. 10.5b). A small increase in labile copper was observed in the period February 2002 to April 2003, with measured concentrations being consistently above 4 mg/L and rising to a maximum of 9.2 mg/L. This was accompanied by consistent algal growth inhibition. Labile copper concentrations have subsequently decreased to between 3 and 5 mg/L but this has not been accompanied by a concurrent decrease in algal growth inhibition at this site. There is a weak correlation (R ¼ 0.657) between the labile copper concentration and the extent of algal growth inhibition at Ningerum.

396

J. L. Stauber et al.

40 35

Copper (µg L-1)

30 25 20 15 10 5 0 Sep- Oct- Nov- Feb- Apr- May- Jun- Feb- Apr- Jul- Oct- Feb- Apr- Aug- Dec- Feb- Jul96 96 96 97 97 97 97 02 02 02 02 03 03 03 03 04 04

Sampling Date

a) 100

30

90

Algal Inhibition (%)

70

20

60 15

50 40

10

30 20

Labile Copper µg L-1

25

80

5

10 0

0 Sep- Oct- Nov- Feb- Apr- May- Jun- Feb- Apr96 96 96 97 97 97 97 02 02

b)

Jul02

Oct- Feb- Apr- Aug- Dec- Feb02 03 03 03 03 04

Jul04

Sampling Date

Figure 10.5: Copper speciation and algal inhibition at Ningerum, 1996– 2004. (a) Copper speciation; dissolved copper (); copper complexation capacity (~); labile copper (7); (b) algal inhibition (red bars) and labile copper concentrations (~). 10.5.2. Middle Fly River At Nukumba, dissolved copper concentrations were similar to those observed at Ningerum and ranged from 9 to 23 mg/L (Fig. 10.6a). At this site, labile copper concentrations were o4 mg/L during the early survey

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

397

40 35

Copper (µg L-1)

30 25 20 15 10 5 0 Nov- Feb- Apr- May- Jun- Feb- Apr96 97 97 97 97 02 02

Oct- Jul-02 Feb- Apr- Aug- Dec- Feb- Jul-04 02 03 03 03 03 04

Sampling Date

a)

30

100 90

Algal Inhibition (%)

70

20

60 15

50 40

10

30 20

Labile Copper µg L-1

25

80

5

10 0

b)

Nov- Feb- Apr- May- Jun- Feb- Apr96 97 97 97 97 02 02

Jul02

Oct- Feb- Apr- Aug- Dec- Feb- Jul02 03 03 03 03 04 04

0

Sampling Date

Figure 10.6: Copper speciation and algal inhibition at Nukumba, 1996– 2004. (a) Copper speciation; dissolved copper (); copper complexation capacity (~); labile copper (7); (b) algal inhibition (red bars) and labile copper concentrations (~). period, November 1996–June 1997. Algal inhibition was observed on only one occasion but was accompanied by the lowest measured labile copper concentration (0.6 mg/L) (Fig. 10.6b). Labile copper concentrations increased during the period February 2002–February 2003 being consistently above 5 mg/L and rising to a maximum of 9.3 mg/L, but algal growth inhibition was

398

J. L. Stauber et al.

not always observed. Labile copper concentrations appeared to decrease at Nukumba from April to December 2003, but algal growth inhibition was still consistently observed. Overall there was a very poor correlation (R ¼ 0.394) between algal growth inhibition and labile copper concentrations at Nukumba. At Obo, dissolved copper concentrations (11–28 mg/L) and copper complexation capacities (9–38 mg/L) remained high and were comparable to those observed at sites on the upper river. Labile copper concentrations were low (o0.5–8 mg/L) (Fig. 10.7a). However, there has been a general, and statistically signiﬁcant (po0.001), increase in labile copper concentrations between the 1996/1997 surveys and the 2002/2004 surveys. No algal inhibition was observed at Obo during the 1996/1997 surveys when labile copper concentrations ranged from o0.5 to 2.5 mg/L. After February 2002, labile copper concentrations were consistently W3 mg/L, reaching a maximum value of 8.2 mg/L. Algal inhibition was also observed at Obo from February 2002 to July 2004 (Fig. 10.6b) but there was only a poor correlation (R=0.317) between labile copper concentrations and the magnitude of algal inhibition during this latter survey period. Lower dissolved copper concentrations (5–14 mg/L) were observed at Ogwa compared to sites further up the Fly River (Fig. 10.8a). Labile copper concentrations however, showed a similar trend to those observed at Obo. During the early survey period, September 1996 to June 1997, labile copper concentrations were o1 mg/L and no algal inhibition was observed. Increased labile copper concentrations ranging from 2 to 4 mg/L were observed from August 2003 and consistent algal inhibition was observed (Fig. 10.8b). Further work is required to determine if the recent increase in algal inhibition at Ogwa is signiﬁcant, and if there is any correlation between increasing labile copper concentrations and the extent of algal growth inhibition observed at this site. The poor correlations between labile copper concentration and algal growth inhibition, and the lack of an obvious relationship between dissolved copper concentrations, copper complexation capacities, and labile copper concentrations at these sites, strongly suggests that factors such as the concentrations of other dissolved metals, e.g., aluminum, or suspended particulate matter in the river may be inﬂuencing the observed algal toxicity.

10.6. Risks to Aquatic Biota in the Fly River System The speciation monitoring data indicate that both dissolved copper concentrations and copper-complexing capacity in the Upper and Middle

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

399

40 35

Copper (µg L-1)

30 25 20 15 10 5 0 Sep- Oct- Nov- Feb- Apr- May- Jun- Feb- Apr- Jul- Oct- Feb- Apr- Aug- Dec- Feb- Jul96 96 96 97 97 97 97 02 02 02 02 03 03 03 03 04 04

Sampling Date

a)

30

100 90

Algal Inhibition (%)

70

20

60 50

15

40

10

30 20

Labile copper µg L-1

25

80

5

10

0

0 Sep- Oct- Nov- Feb- Apr- May- Jun- Feb- Apr- Jul- Oct- Feb- Apr- Aug- Dec- Feb- Jul96 96 96 97 97 97 97 02 02 02 02 03 03 03 03 04 04

b)

Sampling Date

Figure 10.7: Copper speciation and algal inhibition at Obo, 1996–2004. (a) Copper speciation; dissolved copper (); copper complexation capacity (~); labile copper (7); (b) algal inhibition (red bars) and labile copper concentrations (~).

Fly River have remained relatively constant since 1996. Although there were some logistical difﬁculties in the early speciation surveys, which may have resulted in an underestimation of labile copper and copper toxicity, there is a general trend of increasing labile copper concentrations in the Middle Fly.

400

J. L. Stauber et al.

40 35

Copper (µg L-1)

30 25 20 15 10 5 0 Sep-96 Oct-96 Nov-96 Feb-97 Apr-97 May-97 Jun-97 Aug-03 Dec-03 Feb-04 Jul-04 Sampling Date

a)

30

100 90

Algal Inhibition (%)

70

20

60 15

50 40

10

30 20

Labile Copper µg L-1

25

80

5

10 0

0

b)

Sep-96 Oct-96 Nov-96 Feb-97 Apr-97 May-97 Jun-97 Aug-03 Dec-03 Feb-04 Jul-04 Sampling Date

Figure 10.8: Copper speciation and algal inhibition at Ogwa, 1996–2004. (a) Copper speciation; dissolved copper (); copper complexation capacity (~); labile copper (7); (b) algal inhibition (red bars) and labile copper concentrations (~).

The reason for this increase is unknown, but may be due either to inputs of more reactive copper from a change in mining throughput and/or the dredge stockpiles at Bige, or to vegetation dieback on the ﬂoodplain and consequent reduced organic matter input or changed organic matter composition, which in turn inﬂuences copper speciation.

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

401

The frequency of algal growth inhibition has also increased since 1996. At sites on the Middle Fly River, the increase in the frequency of algal growth inhibition corresponds to an increase in labile copper concentrations. The magnitude of algal inhibition however, has not signiﬁcantly increased and there is no signiﬁcant correlation between the amount of algal inhibition and labile copper concentrations in the river. Addition of EDTA to water samples consistently removed the observed toxicity in both the algal and bacterial bioassays. This indicates that the toxic effects are likely to be related to the presence of metal contaminants. Of the dissolved metals measured in river waters since 2002, only aluminum and lead exceeded guideline trigger values (ANZECC/ARMCANZ, 2000) on one or more occasions. In an attempt to understand the impacts of copper toxicity on the Fly River aquatic ecosystem, a summary of the mean and maximum labile copper concentrations at four sites in the Ok Tedi and Fly Rivers is given in Table 10.4, together with the cumulative frequency distribution plot of labile copper at all monitoring sites in the Ok Tedi and Fly Rivers over the recent monitoring campaign (Fig. 10.9a). A comparison of Fig. 10.9a monitoring data for labile copper with known sensitivities of freshwater biota to dissolved copper in the literature [Fig. 10.9b, summarized in Rogers et al. (2005) and Parametrix (1999)] shows that freshwater biota are at risk of chronic effects from copper in the Fly River system. At the labile copper concentrations measured during the monitoring program, acute copper toxicity is not an issue. At worst, acute toxicity may be observed in fewer than 10% of invertebrate species, with no acute effects on ﬁsh. However, at current labile copper concentrations, chronic effects of copper resulting from long-term exposure to elevated bioavailable copper concentrations may be expected in 50–80% of freshwater species. Any further increases in labile copper in the river may result in chronic toxicity to the majority of freshwater species (Rogers et al., 2005).

Table 10.4: Maximum and mean measured labile copper concentrations at sites on the Fly River (1996–2004 monitoring). Site

Maximum labile Cu [mg/L] (date)

Mean labile Cu (SD) [mg/L]

Ningerum Nukumba Obo Ogwa

9.2 9.2 8.2 4.2

4.0 4.1 3.6 1.4

(April 2002) (February 2002) (February 2002) (December 2003)

(2.3) (2.4) (2.4) (1.2)

n ¼ 17 n ¼ 15 n ¼ 17 n ¼ 11

402

J. L. Stauber et al.

Food web stable-isotope studies (Storey, 2005; Storey and Yarrao, 2009) clearly indicate the importance of microalgae, such as periphyton, in the food webs of the Fly River system. Studies by Bunn et al. (1999) indicate that about 40% of the riverine and 70% of the ﬂoodplain ﬁsh biomass is supported by carbon derived from algae. Recent work by Storey (2005), suggests that periphyton abundance is reduced in the river sections downstream of D’Albertis Junction, and the contribution of algal carbon to the aquatic food web in these sections is also reduced compared with areas upstream of the mine. The factors affecting periphyton abundance in the Fly River remain to be established. At this stage, it is not possible to deconvolute the effects of copper, turbidity (with respect to light production or scouring) and loss of habitat. Nevertheless, the concentrations of labile copper observed in the river system are in the range that may cause growth reduction in algae. Loss of algal carbon food sources and consequent effects on higher-level consumers, irrespective of the cause, is an important potential impact.

10.7. Looking Ahead While the labile copper and toxicity monitoring data over the past 10 years have provided valuable information on spatial and temporal trends in copper in the Fly River system, further event-based sampling will enable the assessment of variability in the system and better analysis of long-term trends. Changing mine practices and increased forest dieback currently confound our understanding of the causes of the observed toxicity and we have a very limited understanding of the effects of colloidal copper (included in the estimates of ‘‘dissolved’’ copper) or other metals on biota in the Fly River. The potential interactive effects of acidiﬁcation from acid rock drainage (ARD) and metal toxicity have not been assessed, but have the potential for serious consequences if pH reductions are signiﬁcant. While ecotoxicological studies under controlled conditions with algae isolated from the Fly River have shown effects on algal growth in Fly River water, extrapolating this response to the response of complex aquatic communities such as those found in the Fly River is difﬁcult. In particular, effects such as ecosystem recovery, depletion of food sources, prey switching, and algal succession are not considered in laboratory studies. In a recent review by Chariton and Apte (2005), they concluded that, while copper concentrations below 2 mg/L have no effect on composition, diversity, and abundance of biota, at slightly higher concentrations (4 mg Cu/L) changes in

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

403

a) 100 90 Cumulative frequency (%)

80 70 60 50 40 30 20 10 0 0

2

6 4 Labile Cu (µg L-1)

8

10

100

Cumulative Frequency (%)

90 80 70 60 50 40 30 20 10 0 1 b)

10 Dissolved Copper (µg L-1)

100

Figure 10.9: Cumulative frequency distribution of (a) labile copper concentrations observed in the Ok Tedi and Fly River systems during the monitoring campaign; (b) chronic copper sensitivities for freshwater species at water hardness 30 mg/L CaCO3 (data from ANZECC/ARMCANZ, 2000).

algal assemblages, including the dominance of taxa, have been demonstrated. Concentrations of copper above 10 mg/L can lead to a reduction in algal biomass and a shift to copper-tolerant species. Reduced pH (o6.2) can independently lead to an imbalance in the ratio of algal production to

404

J. L. Stauber et al.

respiration, resulting in the dominance of acid-tolerant diatoms. Combined low pH (o5) and elevated aluminum concentrations also have the potential to impact the Fly River system in future. The presence of localized ARD on levee banks of the Middle Fly, resulting in elevated dissolved copper and other metals in ﬂoodplain waters adjacent to the ARD patches, is an increasing cause for concern. The levee ARD occurs primarily during periods of low river levels when levees are exposed and oxidizing, after which localized rainfall leaches metals from the levees resulting in elevated metals in adjacent surface waters (A. Storey, personal communication). OTML 2005 monitoring data show that metals released include copper, aluminum, cadmium, manganese, magnesium, nickel, zinc, and lead. In situ toxicity tests are proposed to determine the ecotoxicity of levee ARD. These will involve sampling of resident ﬂora/fauna to assess community effects and ﬁeld exposure trials to determine responses in selected taxa deployed in 96-h exposure experiments. Given that food web studies showed that periphyton provided an important carbon source supporting aquatic food chains in the Fly River system, further work is underway to determine the effect of both increased bioavailable copper concentrations and increased turbidity on periphyton communities in the Fly River. Periphyton samples collected from both the Middle Fly and Kiunga are currently being examined for species diversity and abundance, and the sensitivity/tolerance to copper and turbidity of isolated species is being compared. Further research is aimed at developing an algal model to predict algal biomass responses to contaminants in ﬁeld populations. There is also a lack of basic ecotoxicological data on the effects of copper on primary consumers (algal grazers) such as small invertebrates and zooplankton, planktivorous ﬁsh, and other key aquatic organisms in the Fly River system. This is another important knowledge gap currently being addressed through the use of in situ bioassays with ﬁeld-collected invertebrates. Alternative, more robust methods to ASV to measure labile copper have recently been developed in our laboratory and applied to Fly River water samples (Apte et al., 2005; Bowles et al., 2006). The water sample is passed rapidly through a small Chelex resin column, which captures only ionic metal plus metal complexes that are able to dissociate within the short contact time with the resin. Similar to ASV, colloidally bound metals and metals bound in strong complexes do not interact with the resin. The robustness and relative simplicity of this technique makes it attractive for future routine monitoring of labile copper in the Fly River and should allow collection of more data on labile copper and on-site analysis.

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

405

REFERENCES Allison, J. D., Brown, D. S., & Novo-Gradic, K. J. (1991). MINTEQA2/ PRODEFA2. A Geochemical Assessment Model for Environmental Systems. Version 3.0. Users Manual. EPA/600/3-91/021. United States Environmental Protection Agency, Washington, DC. ANZECC/ARMCANZ (2000). Guidelines for Fresh and Marine Water Quality. Australia and New Zealand Environment and Conservation Council, Agriculture and Resource Ministers Council of Australia and New Zealand, Canberra, Australia. Apte, S. C. (2009). Biogeochemistry of copper in the Fly River. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, pp. 321–373. Apte, S. C., & Day, G. M. (1993). Organic complexation and partitioning of copper in river waters. A review. Ok Tedi Mining Ltd./CSIRO Joint Report, 49 pp. Apte, S. C., Batley, G. E., Davies, C. M., & Stauber, J. L. (1995). Assessing the fate, transport and toxicity of mine derived contaminants. In: Proceedings of the 20th Annual Environment Workshop Managing Environmental Impacts—Policy and Practice, Minerals Council of Australia, Darwin 2–6 October 1995, pp. 377–384. Apte, S. C., Stauber, J. L., Davies, C. M., & Hales, L. T. (1997). Bioavailability of dissolved copper in waters downstream of the Ok Tedi mine. CSIRO Energy Technology Investigation Report CET/IR 27R prepared for Ok Tedi Mining Ltd., 58 pp. Apte, S. C., Batley, G. E., Bowles, K. C., Brown, P. L., Creighton, N., Hales, L. T., Hyne, R. V., Julli, M., Markich, S. J., Pablo, F., Rogers, N. J., Stauber, J. L., & Wilde, W. (2005). A comparison of copper speciation measurements with the toxic responses of three sensitive freshwater organisms. Environmental Chemistry, 2, 320–330. Australian Water Technologies. (2002). Testing of the toxicity of copper to barramundi. AWT Report Number, 1999/0059. Batley, G. E., Apte, S. A., & Stauber, J. L. (2004). Speciation of trace metals in waters: Progress since 1982. Australian Journal of Chemistry, 57, 903–919. Boswell, C. D., Hewitt, C. J., & Macaskie, L. E. (1998). An application of bacterial ﬂow cytometry: Evaluation of the toxic effects of four heavy metals on Acinetobacter sp. with potential for bioremediation of contaminated wastewaters. Biotechnology Letters, 20, 857–863. Bowles, K. C., Apte, S. C., Batley, G. E., & Rogers, N. J. (2006). A rapid Chelex column method for the determination of metal speciation in natural waters. Analytical Chimica Acta, 558, 237–245. Bunn, S., Tenakanai, C., & Storey, A. W. (1999). Energy sources supporting Fly River ﬁsh communities. Ok Tedi Mining Limited Report, 44 pp. Campbell, P. G. C. (1995). Interactions between trace metals and aquatic organisms: A critique of the free-ion activity model. In: A. Tessier, & D. R. Turner (Eds). Metal Speciation and Bioavailability in Aquatic Systems. Wiley, Toronto, pp. 45–102.

406

J. L. Stauber et al.

Chariton, A., & Apte, S.C. (2005). Field and experimental responses of biota to copper-enriched and acidic water: Literature review. CSIRO Energy Technology Report ET/IR8143R, 17 pp. Davies, C. M., Apte, S. C., & Johnstone, A. L. (1998). A bacterial bioassay for the assessment of copper bioavailability in freshwaters. Environmental Toxicology and Water Quality, 13, 263–271. Erickson, R. J., Benoit, D. A., Mattson, V. R., Nelson, H. P., & Leonard, E. N. (1996). The effects of water chemistry on the toxicity of copper to fathead minnows. Environmental Toxicology and Chemistry, 15, 181–193. Florence, T. M., & Stauber, J. L. (1986). Toxicity of copper complexes to the marine diatom N. closterium. Aquatic Toxicology, 8, 11–26. Franklin, N. M., Stauber, J. L., Markich, S. J., & Lim, R. P. (2000). pH-dependent toxicity of copper and uranium to a tropical freshwater alga (Chlorella sp.). Aquatic Toxicology, 48, 275–289. Franklin, N. M., Stauber, J. L., Apte, S. C., & Lim, R. P. (2002). Effect of initial cell density on the bioavailability and toxicity of copper in microalgal bioassays. Environmental Toxicology and Chemistry, 21, 742–751. Gillespie, P. A., & Vaccaro, R. F. (1978). A bacterial bioassay for measuring the copper-chelation capacity of seawater. Limnology and Oceanography, 23, 543–548. Hudson, R. J. M. (1998). Which species control the rates of trace metal uptake by aquatic biota? Observations and predictions of non-equilibrium effects. Science of the Total Environment, 219, 95–115. Lee, K., Tay, K. L., Ewing, C. N., & Levy, E. M. (1990). Toxicity and environmental impact assessment tests based on the activity of indigenous bacteria. Ocean Dumping Report 4, Environment Canada, Atlantic Region, Darmouth, Nova Scotia, 138 p. Markich, S. J., Brown, P. L., Batley, G. E., Apte, S. C., & Stauber, J. L. (2001). Incorporating metal speciation and bioavailability into water quality guidelines for protecting aquatic ecosystems. Australasian Journal of Ecotoxicology, 7, 109–122. Markich, S. J., Batley, G. E., Stauber, J. L., Rogers, N. J., Apte, S. C., Hyne, R. V., Bowles, K. C., Wilde, K. L., & Creighton, N. (2005). Hardness corrections for copper are inappropriate for protecting sensitive freshwater biota. Chemosphere, 60, 1–8. OECD (1984). Alga, growth inhibition test. In: OECD Guidelines for Testing of Chemicals, Guideline 201, Vol. 1. Organisation for Economic Cooperation and Development, Paris, France, pp. 1–14. Parametrix Inc., & URS Greiner Woodward Clyde (1999). Assessment of human health and ecological risks for proposed mine waste mitigation options at the Ok Tedi mine, Papua New Guinea. Detailed Level Risk Assessment, Final Report. Prepared for Ok Tedi Mining Ltd., 214 pp. Phinney, J. T., & Bruland, K. W. (1994). Uptake of lipophilic organic Cu, Cd and Pb complexes in the coastal diatom Thalassiosira weissﬂogii. Environmental Science and Technology, 28, 1781–1790.

Speciation, Bioavailability and Toxicity of Copper in the Fly River System

407

Rogers, N. J., Apte, S. C., Knapik, A., Davies, C. M., Bowles, K. C., & Kable, S. H. (2005). A rapid, radiochemical bacterial bioassay to evaluate copper toxicity in freshwaters. Archives of Environmental Contamination and Toxicology, 49, 471–479. Shao, Q., Harch, B., Apte, S., & Simpson, S. (2002). Temporal variability of dissolved copper concentrations in the Ok Tedi and Fly River, CSIRO Energy Technology Investigation Report CET/LH/IR 501R prepared for Ok Tedi Mining Ltd., 56 pp. Smith, R. E. W. (1997). Review of laboratory based ecotoxicological testing of Ok Tedi Mining Limited Wastes. R&D Environmental Pty Ltd., prepared for Ok Tedi Mining Limited, 18 pp. Smith, R. E. W., Ahsanullah, M., & Batley, G. E. (1990). Investigations of the impact of efﬂuent from the Ok Tedi copper mine on the ﬁsheries resource of the Fly River, Papua New Guinea. Environmental Monitoring and Assessment, 14, 315–331. Soli, A. L., & Byrne, R. H. (1989). Temperature dependence of Cu(II) carbonate complexation in natural seawater. Limnology and Oceanography, 34, 239–244. Stauber, J. L., & Florence, T. M. (1987). The mechanism of toxicity of ionic copper and copper complexes to algae. Marine Biology, 94, 511–519. Stauber, J. L., & Critelli, C. (1993). Toxicity of waters from the Fly River system to freshwater and estuarine algae, CSIRO Energy Technology Investigation Report CET/IR 187 prepared for Ok Tedi Mining Ltd., 20 pp. Stauber, J. L., & Apte, S. C. (1996). Bioavailability of copper to algae in the Fly River system, CSIRO Energy Technology Investigation Report CET/LH/IR 464R prepared for Ok Tedi Mining Ltd., 78 pp. Stauber, J. L., & Davies, C. M. (2000). Use and limitation of microbial bioassays for assessing copper bioavailability in the aquatic environment. Environmental Reviews, 8, 255–301. Stauber, J. L., Tsai, J., Vaughan, G., Peterson, S. M., & Brockbank, C. I. (1994). Algae as indicators of toxicity of BEKM efﬂuents, National Pulp Mills Research Program Technical Report Series No. 3, Canberra, CSIRO, pp. 1–82. Stauber, J. L., Benning, R. J., Hales, L. T., Eriksen, R. S., & Nowak, B. (2000). Copper bioavailability and amelioration of toxicity in Macquarie Harbour, Tasmania, Australia. Marine and Freshwater Research, 51, 1–10. Storey, A. W. (2005). Comparison of carbon sources supporting aquatic food webs above and below D’Albertis Junction, Unpublished Report by Wetland Research & Management to Ok Tedi Mining Ltd., pp. 1–42. Storey, A. W., & Yarrao, M. (2009). Development of aquatic food web models for the Fly River, Papua New Guinea, and their application in assessing impacts of the Ok Tedi Mine. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, pp. 575–615. Tessier, A., & Turner, D. R. (1995). Metal Speciation and Bioavailability in Aquatic Systems. Wiley, Chichester, UK.

408

J. L. Stauber et al.

Turner, D. R., Whitﬁeld, M., & Dickson, A. G. (1981). The equilibrium speciation of dissolved components in freshwater and seawater at 251C at 1 atm pressure. Geochimica et Cosmochima Acta, 45, 855–882. Wilde, K. L., Stauber, J. L., Markich, S. J., Franklin, N. M., & Brown, P. L (2006). The effects of pH on the uptake and toxicity of copper and zinc in a tropical freshwater alga (Chlorella sp.). Archives of Environmental Contamination and Toxicology, 51, 174–185. Zhang, H., Davison, W., & Statharn, P. (1996). In situ measurements of trace metals in seawater using diffusive gradients in thin ﬁlms (DGT). In: S. Bottrell (Ed.). International Symposium of Geochemistry and Earth’s Surface. University of Leeds, Leeds, pp. 138–142.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00411-4

Chapter 11

The Biology of Barramundi (Lates calcarifer) in the Fly River System Stephen J. M. Blaber, David A. Milton and John P. Salini CSIRO Marine & Atmospheric Research, PO Box 120, Cleveland, Queensland 4163, Australia

11.1. Introduction Lates calcarifer is a large species of Centropomidae that occurs in coastal waters, estuaries, and freshwaters from western India, around Sri Lanka to the Bay of Bengal, and through the whole of Southeast Asia to Papua New Guinea and northern Australia. In Australia and Papua New Guinea, its common name is Barramundi, but it has a large number of other common names, such as kakap putih in Indonesia and siakap in Malaysia. Throughout its range it is a very well-known ﬁsh and supports extensive commercial, aquacultural, and recreational ﬁsheries, and in Papua New Guinea it is the most economically important species in the commercial component of the artisanal ﬁsheries of Western Province (Fig. 11.1). In Papua New Guinea, it occurs on the coast and inland in most rivers in the south of the country, with the largest populations in the Fly River. Barramundi is the most valuable species in the Fly River, and in the past, it has represented up to 80% of the catch in coastal waters. An artisanal ﬁshery in the Fly River is expanding as part of community development initiatives by Ok Tedi Mining Limited. The coastal and lower Fly River ﬁshery appeared to collapse in the early 1990s, following overﬁshing in the 1980s; and while the middle Fly River ﬁshery has, in the past few years, developed annual catches up to 170 t, the coastal ﬁshery is still relatively minor. There is

Corresponding author. Tel.: þ61-7-38267104; Fax: þ61-7-38267222;

E-mail: [email protected] (S.J.M. Blaber).

412

S. J. M. Blaber et al.

Figure 11.1: Adult Barramundi caught for research purposes on the middle Fly River. The measuring board is 1 m long. now considerable interest in the cage culture of Barramundi in Papua New Guinea, following the methods developed for the very valuable industry in west Malaysia. Monitoring of Fly River ﬁsh populations, which began in 1983, has shown that although ariid catﬁsh, plotosid catﬁsh, and Nematalosa herrings are the most numerous species, L. calcarifer comprise the greatest biomass, forming over 30% by weight of research catches (Swales et al., 2000). This research on Barramundi over the past 20 years in the Fly River and Western Province indicates that their ecology and life history in Papua New Guinea differs considerably from that of their relatives in northern Australia, about which there is much detailed knowledge (Copland and Grey, 1987; Blaber, 2000).

Biology of Barramundi in the Fly River System

413

Prior to work in Papua New Guinea, almost all our information about Barramundi was derived from studies in Australia and India, where habitats and climatic regimes are rather different. The ﬁrst of these differences concerns the genetic makeup of the Barramundi populations of Papua New Guinea.

11.2. Stock Structure and Genetics The population structure of Barramundi in northern Australia had been investigated using both allozyme electrophoresis and mitochondrial DNA methods. Initial studies by Shaklee and Salini (1985), Salini and Shaklee (1988), and Shaklee et al. (1993) identiﬁed 14 genetically distinct populations in Australian waters and a separate population from the Gulf of Papua. Keenan (1994) extended this work with more intensive sampling along the Northern Territory coast and Queensland east coast and detected another four genetically distinct populations. The historical basis of this allozyme population structure was extended with mitochondrial DNA (mtDNA) markers by Chenoweth et al. (1998a, b) and Doupe´ et al. (1999). The mtDNA haplotype distribution suggested some genetic exchange between adjacent populations, but reduced exchange with more distant populations (isolation by distance). Chenoweth et al. (1998a) concluded that their results suggest historical separation between the western Paciﬁc and Indian Oceans at the Torres Straits followed by recent recontact. Life history studies by Russell and Garrett (1983, 1988) and tagging experiments have conﬁrmed very limited coastal migration by Barramundi in Australia. This lack of coastal movement is the most likely mechanism for the observed genetically distinct populations in adjacent rivers in Australia. However, tagging studies in Papua New Guinea suggested that Barramundi in the Fly River system and other coastal waterways of southern Papua New Guinea do move signiﬁcant distances. Moore (1982) and Moore and Reynolds (1982) showed that 20% of adult ﬁsh moved more than 100 km and about 5% more than 500 km – much further than their Australian counterparts. Fish tagged in the Kikori and Purari rivers, east of the Fly River, were also recaptured on the coast around Daru. Only 72 ﬁsh were tagged in the Bensbach and Morehead rivers and the single recapture was from Irian Jaya. These results suggest that there is some mixing of ﬁsh from different river systems and that there is probably exchange of Barramundi between nearby rivers in southern Papua New Guinea. In order to elucidate stock structure in Papua New Guinea and to identify the source of juvenile Barramundi marketed in a southwest coastal ﬁshery,

414

S. J. M. Blaber et al.

both mtDNA and microsatellite DNA methods have recently been used (Salini et al., 2004). This genetic research has shown that, unlike the situation in northern Australia where there are multiple small genetic stocks of Barramundi, there is just one large stock of Barramundi living in the Fly River and associated coastal waters. This stock extends into Irian Jaya (West Papua), thereby constituting a cross-border resource, but differs from a small isolated stock of Barramundi found in the far east of Papua New Guinea. These ﬁndings are extremely important for ﬁsheries management, and mean that the Fly River and Western Province Barramundi can be managed as a single unit – unlike the situation in northern Australia where it has been necessary to manage on an almost river-by-river scale.

11.3. Life Cycle 11.3.1. Reproduction L. calcarifer has a complex life history that varies somewhat across its geographic range (Blaber, 2000); however, all individuals start life as males and then change to females, that is, they are protandrous hermaphrodites (Moore, 1979). Garrett (1987) reviewed the reproductive biology, from which the following account is largely taken. The gonads are dimorphic and a complete reorganization of gonad structure and function takes place after sex inversion, probably under the inﬂuence of hormones. L. calcarifer spawn as males for several years before sex inversion. The sex reversal is initiated as the testes ripen for the last time, and the change to ovary takes place rapidly within a month of spawning. The change to female usually takes place at about 7 years of age and a length of about 800 mm, but is apparently more related to age than to length. The length at which sex change occurs varies somewhat across its extensive geographic range, probably due to habitat, food, and genetic differences. Moore (1980) postulated that protandrous sex reversal in L. calcarifer allows the larger and more successful ﬁsh (the females) to make the greatest contribution to the gene pool of a particular population. Movement to spawning sites and gonad maturation takes place at the end of the dry season in estuarine waters (28–36m), usually near the mouth. The eggs and larval stages have a narrower range of salinity and temperature tolerance than the adults (Russell and Garrett, 1983). There is a tidal-based monthly cycle of spawning, with postlarvae entering coastal swamps on spring tides as the wet season progresses. Spawning at this time ensures that the juveniles can enter and remain in the shelter of ﬂooded

Biology of Barramundi in the Fly River System

415

backswamps and ﬂoodplains until they dry up during the early part of the dry season. There is thus a strong link between the reproductive cycle of L. calcarifer and the monsoon pattern in Australia and Asia. The fecundity of L. calcarifer is among the highest of any teleost ﬁsh, with estimates of from 0.6 to 2.3 106 eggs kg1 of body weight (Garrett, 1987). It is reported that some individuals spawn all at once whereas others may be multiple spawners. The high fecundity may be a response to the late onset of female reproductive function in the population and the relatively low numbers of females in the overall population (Moore, 1980; Davis, 1984). This implies that most recruitment is derived from small numbers of large female ﬁsh (Garrett, 1987) and makes the species very vulnerable to overﬁshing. Sex reversal in L. calcarifer may, on the one hand, allow the most successful individuals (those that have survived to large size) to contribute most to the gene pool, but it results in a need for very high fecundities in the low number of large surviving ﬁsh. On the other hand, as an alternative strategy, there may be the potential in some circumstances for selection to favor change of sex at a smaller size and hence allow breeding more often. In the Fly River region, large female Barramundi move to the coastal spawning grounds near Sigabaduru (west of Daru) (Fig. 11.2) during the late dry season (September–October) when gonad maturation takes place in estuarine waters (salinities of 28–36m). Spawning takes place in the sea because the eggs and larval stages have a narrower range of salinity and temperature tolerance than the adults. Between October and February, as the wet season progresses, there is a tidal-based monthly cycle of spawning, with larvae moving into shallow coastal areas and backswamps shortly after hatching at around 5.0-mm long, usually on the spring high tides. Spawning at this time in the sea ensures that the juveniles can enter and remain in sheltered coastal areas, ﬂooded backswamps, or ﬂoodplains until they dry up during the early part of the dry season. By entering shallow, temporary brackish or freshwater pools, and tidal swamps, the young Barramundi can exploit the high prey abundance and low predator numbers in these habitats. They rapidly become the dominant predators of other ﬁsh larvae and crustaceans in this habitat. Environmental conditions in these pools can be quite extreme and juvenile Barramundi must adapt to a wide range of temperatures and salinities. The growth rate of Barramundi is rapid, with most ﬁsh reaching a length of at least 300 mm within their ﬁrst year and a meter by about 10 years (Reynolds and Moore, 1982). However, as the life history is complex, involving both environmental and sex changes, growth varies considerably during the different phases and between habitats. For example, Reynolds (1978) showed that the growth rate in freshwater is higher than that in

416

S. J. M. Blaber et al.

Figure 11.2: Western Province and Fly River region showing the movements and spawning grounds of Barramundi. saltwater. Nevertheless, it is the high initial growth rate that has made this species an attractive and successful candidate for culture. They can be raised to ‘‘plate size’’ within about 6 months.

11.3.2. Movement Patterns Earlier studies (Moore and Reynolds, 1982; Mobiha, 1993) found that juveniles left the coastal nursery swamps from March onward at about 6 months of age and a length of about 200 mm. Tagged juveniles usually stayed in the vicinity of the tagging site until July–August when the proportion recaptured at the same site declined dramatically. Moore and Reynolds (1982) suggested that this corresponded with the period when they migrate along the coast. Hence, while some juveniles stay in coastal waters, the majority migrate into the lower Fly River and continue to move upstream, reaching the middle Fly River, the Strickland River, and Lake Murray where they grow to become adults. Older (1 to 3þ year-old) ﬁsh tended to migrate eastward and were mostly recaptured in the lower and

Biology of Barramundi in the Fly River System

417

middle Fly River. Tagging studies by Moore and Reynolds (1982) showed that most recaptures of ﬁsh that had moved relatively longer distances (W15 km) were of ﬁsh that had moved to the east. Two 0 þ ﬁsh were recaptured in the Fly River estuary as 1þ ﬁsh and another as a 1þ in Lake Murray in the middle Fly River, suggesting that spawning ﬁsh returned to the area from which they had originally migrated. They recaptured adult ﬁsh at the same site in Lake Murray after up to 7 years and in the Gulf of Papua after 3 years. There were no recaptures of ﬁsh from the Fly River in the Gulf of Papua or vice versa. Movements to and from both of these areas were of ﬁsh caught on the coast west of Daru. We now know that the adults do not, however, follow a regular annual migration to the sea as they do in Australia. Recent research has shown that the movement patterns of Barramundi can be traced by measuring the strontium (Sr) concentration across their otoliths (Milton et al., 2000). The concentration of Sr in water changes in relation to salinity whereas that of Ca does not. The age of a ﬁsh is also reﬂected by the daily and annual rings in the otolith, and hence measurements of the Sr/Ca ratios and concentrations across the otolith can be linked with age data to show when the ﬁsh moved between marine and freshwater (Fig. 11.3). The chemical composition across the otolith, from its core (when the ﬁsh was born) to its edge, is carried out using laser ablation of a series of sites across the otolith, with the material from the ablations being analyzed using inductively coupled plasma mass spectrometry (Fig. 11.3). This research conﬁrms that all Fly River Barramundi were born in the sea, but unlike the situation in northern Australia where they undertake very regular annual breeding migrations between river and coast, many Papua New Guinea ﬁsh only visit the sea once during their lives, with some individuals remaining upriver for extended periods of many years. Examples of the very different movements are shown in Figs. 11.3 and 11.4, the former showing a female caught at Baimuru in the eastern Fly River delta and the latter females captured at Obo in the middle Fly River and on the spawning grounds at Sigabaduru. Summarizing the results from transects across the otoliths from 122 mature ﬁsh (5- to 14-year old) shows that about half the spawners stay on the coast. Most ﬁsh (80%) have migrated into freshwater at least once during their life, but less than half the adults in the middle Fly River (46%) have migrated back to the sea. Males that have migrated back to the sea have only done so once, but most females that have migrated between freshwater and the sea have done so at least twice. On the spawning grounds less than half the ﬁsh spawn each year and of these spawners none had come from the Fly River and most (6 of 8) had not migrated into freshwater at all although a few (2) appear to have migrated into local rivers along the coast. Many ﬁsh

418

S. J. M. Blaber et al.

(a)

(b)

Born

(c)

2400

Sr concentration ppm

2000

1600

1200

800

400 0

2

4 Age (years)

6

8

Figure 11.3: (a) Section across a Barramundi otolith, (b) scanning electron micrograph of the transect across the otolith showing laser ablation pits for determining strontium (Sr) concentration, and (c) plot of the wherabouts since birth of a female Barramundi captured at Baimuru, based upon the relationship between Sr and salinity.

Biology of Barramundi in the Fly River System

419

4000 Male

Female

Sr concentration ppm

Sea 3000

2000

1000

Freshwater

Female spawner at Sigabaduru

0 0

2

4 6 Age (increments)

(a)

8

10

4000 Middle Fly River at Obo

Sr concentration ppm

3000

Sea 2000

Freshwater

1000

0 0 (b)

2

6 4 Age (increments)

8

10

Figure 11.4: The life histories of (a) a 9-year-old female Barramundi captured on the spawning grounds at Sigabaduru, and (b) another 9-year-old female caught in the middle Fly River.

420

S. J. M. Blaber et al.

stay on the coast for several years before migrating, and of those sampled none had migrated to spawn. Results from the middle Fly River show that ﬁsh move regularly between off-river bodies and the main river. A key factor then is that coastal and estuarine populations of Barramundi appear to form the majority of spawners, with little input from middle and upper Fly River ﬁsh, which constitute only a small percentage of the total spawning population (o5%). This has important ﬁsheries management consequences because it indicates a necessity to minimize catches of large ﬁsh (spawning females) on the coast, rather than to restrict catches of large female (W90 cm) ﬁsh in the middle Fly River that may have ﬁnished spawning. Differences in the movement patterns between Australian and Papua New Guinea ﬁsh may possibly be linked to both environmental and climatic differences. The Fly River, with its massive freshwater discharge, great length, and huge catchment, is very different from the rivers of northern Australia, which have very marked short seasonal ﬂows and are relatively small with small catchments. There is also a complete absence in northern Australia of the coastal swamps found in Western Province, the nearest equivalent being the upriver freshwater billabongs favored by juvenile Barramundi. 11.3.3. Diet No detailed data have been published on the food and feeding ecology of Barramundi in Papua New Guinea, but in a comprehensive study of its diet in northern Australia, Davis (1985) showed that from a length of 5–1,200 mm it is an opportunistic predator with an ontogenetic progression in its diet from microcrustacea to macrocrustacea to ﬁsh. Small L. calcarifer (o80 mm) feed mainly on copepods and amphipods. Such microcrustacea are absent in ﬁsh longer than 80 mm where they are replaced by macrocrustacea such as penaeids and palaeomonids. Fish and macrocrustacea are equally important up to about 300 mm, but the proportion of ﬁsh increases thereafter until it represents more than 80% of the diet in larger ﬁsh. A wide variety of ﬁsh species are consumed, with Mugilidae, Engraulidae, and Ariidae among the most frequent. There is a strong correlation between the length of L. calcarifer and that of its prey, with most prey ﬁsh being about 30% of the length of the predator. Research on the diets of Barramundi in the Embley and Norman estuaries in the Gulf of Carpentaria (Salini et al., 1990, 1998) conﬁrmed the ﬁndings of Davis (1985), and moreover showed that they are signiﬁcant predators of commercially important penaeid

Biology of Barramundi in the Fly River System

421

prawns. Davis (1985) reports that the transition from eating prawns to eating ﬁsh entails spending more time feeding on pelagic species in mid-water or on the surface rather than on benthic forms. This could result in greater exposure to predators, but this may be a factor of decreasing importance as the ﬁsh attain a large size [with the possible exception of crocodiles – see Blaber (2000)]. The prey chosen by the smaller size classes of L. calcarifer may reﬂect the types of food available in areas where predation risk is reduced, such as in vegetated backswamps and shallow creeks in the upper reaches. Cannibalism is common in L. calcarifer (up to 11% of the diet in some areas), but is probably reduced by the different size groups living in different habitats. For example, the small juveniles occupy shallow habitats that are not accessible to larger conspeciﬁcs. From the information available, it appears that the diet of Barramundi in the Fly River follows the same general pattern as in Australia. However, many of the prey species are different. In a study of the energy sources supporting Fly River ﬁsh communities, Bunn et al. (1999) indicated that L. calcarifer is at the apex of the food web. During ﬂood periods many Barramundi move into off-river water bodies, particularly oxbow lakes, where there is an abundance of crustaceans and small ﬁsh, such as Fly River Herring (Nematalosa spp.). Swales et al. (1999) stated that the dominance of the ﬁsh communities of oxbow lakes by Nematalosa spp. is a very notable feature, and furthermore that these pelagic, shoaling ﬁsh provide an important food source for Barramundi. Other ﬁsh species recorded in the diet of Barramundi include Melanotaenia splendida, Thryssa scratchleyi, Porochilus meraukensis, Variichthys lacustris, Craterocephalus randi, Ambassis agrammus, Amniataba percoides, and various juvenile ariid catﬁsh. Freshwater prawns (Macrobrachium spp.) are also a common item in the diet. The dominance of the Nematalosa spp. in the diet of Barramundi, particularly in the various off-river water bodies and lakes, is interesting because these ﬁsh are mainly phytoplankton feeders and their consumption emphasizes the importance of the plankton pathway in terms of energy transfer in the Fly River system (Bunn et al., 1999).

11.4. The Fisheries for Barramundi The various ﬁsheries for Barramundi in the Western Province and the Fly River have a long and checkered history. The commercial ﬁshery for Barramundi in Western Province began after the establishment of processing and distribution centers in the province in the early 1960s. By 1969

422

S. J. M. Blaber et al.

commercial gill-net operations were established on the coast near the mouth of the Fly River at Daru and in the middle Fly River around Lake Murray (Opnai and Tenakanai, 1987). The ﬁshery comprised commercial ﬁshing vessels operating gillnets (15.2 and 17.5 cm mesh size) and also artisanal ﬁshers in both regions. The total catch from both areas reached 330 t year1 in the early 1970s (Opnai and Tenakanai, 1987). An artisanal ﬁshery for large adult Barramundi developed alongside the commercial ﬁshery during the 1970s. These ﬁshers sold their catch to village cooperative freezer plants or commercial buyers. Three commercial buyers in Daru purchase ﬁsh, including Barramundi, but usually only ﬁsh of over 3 kg. In the late 1980s, Barramundi ranked fourth among commercial ﬁsheries in Papua New Guinea in terms of total ﬁsh production and foreign exchange earnings (Opnai and Tenakanai, 1987). The Barramundi catch represented B80% of the total weight of commercial seafood sold at Daru in 1995 (Anonymous, 1997). This export commercial ﬁshery caught 200–300 t of Barramundi a year, with the majority of this being caught in the Daru area. This ﬁshery was important to coastal communities in Western Province because of the large number of artisanal ﬁshers involved and the cash income generated in areas with few alternative sources of income. For the economy of the region, it is obviously vital to maintain the long-term sustainability of Barramundi populations. However, in the early 1990s the catches from the Daru-based ﬁshery were as low as 4 t, suggesting that the ﬁshery for adult Barramundi had all but ceased in coastal Western Province, and the commercial ﬁshery was forced to close. Until about 1997, only the artisanal Barramundi ﬁsheries in the coast and the middle Fly River continued to operate. During 1998, the Daru-based commercial ﬁshery resumed buying, and production by the coastal artisanal ﬁshery has now reached more than 100 t annually. This is indicative of a recovering Barramundi stock in the Western Province and the Gulf of Papua. In addition, a freezer-based local community owned ﬁshery was set up in the middle Fly River at Obo in 1998. This ﬁshery employs local people in the Obo area and continues to increase production with about 100 t of Barramundi processed in 2003. In addition to the ﬁshery for adult Barramundi there are subsistence and artisanal ﬁsheries catching immature ﬁsh, with artisanal ﬁshers selling their catch locally in Daru. These ﬁsheries use smaller mesh gillnets (8.9 and 10.2 cm) to catch mainly juveniles in coastal and estuarine waters during the peak period (March–September) of migration of ﬁsh from the coastal nurseries to the riverine adult habitats (Moore and Reynolds, 1982; Mobiha and Murri, 1993). By 1970, the estimated 1.5 t of juvenile Barramundi from the artisanal ﬁshery was the second-most important component of

Biology of Barramundi in the Fly River System

423

the seafood catch sold at the Daru market (Moore and Reynolds, 1973). By 1995, the sale of juveniles in Daru had increased to almost 6 t (W15,000 ﬁsh), and a similar amount was eaten at home (Anonymous, 1997). This is almost double the weight of the previous year’s entire commercial catch in the province and represents several times as many ﬁsh. These large catches of immature ﬁsh may have contributed as much to the observed decline in the catch of adult ﬁsh as the overexploitation of the adults themselves. Although several management measures, including a minimum size limit for Barramundi (38 cm), were introduced after 1983 to reduce the catch of juvenile and immature ﬁsh in Western Province, none of these measures succeeded, probably because of a lack of will to enforce them (Kare, 1995). A new Barramundi Management Plan was gazetted into law in 2003. This plan took into account the research results from projects carried out by Ok Tedi Mining Limited, National Fisheries Authority and CSIRO, Australia, together with commercial and artisanal ﬁshery consultations, to produce a new set of regulations to try and ensure the long-term sustainability of the subsistence, artisanal and commercial components of the ﬁshery. The main points of the new law are that it (1) stipulates a total allowable catch (TAC) of 260 t per annum (whole weight) for the Western Province (subject to periodic review); (2) prohibits the sale or export of Barramundi with a total length of less than 36 cm (14 in.); (3) prohibits the use of gill and beach seine nets with mesh sizes greater than 15 cm (6 in.) – in order to protect the larger spawning females; (4) prohibits the use of gill and beach seine nets with mesh sizes between 6.35 cm (2.5 in.) and 12.7 cm (5 in.) during the peak periods of juvenile recruitment, March 1 to April 30, in the coastal waters from Sui village in the east to the Papua New Guinea/Irian Jaya border in the west; and (5) prohibits the use of gillnets with mesh sizes greater than 12.7 cm (5 in.) during the peak spawning migration period, September 1 to October 31, in the coastal waters from Sui village in the east to Buzi village in the west. In addition, the main spawning and breeding grounds between Sigabaduru village and Papua New Guinea/Irian Jaya border are closed to commercial ﬁshing during peak spawning periods, October 1 to November 30, each season.

REFERENCES Anonymous (1997). Final report of the Gulf and Western Provinces Coastal Zone Management Feasibility Study. Volume I. (AusAID: Canberra).

424

S. J. M. Blaber et al.

Blaber, S. J. M. (2000). Tropical Estuarine Fishes: Ecology, Exploitation and Conservation. Blackwell, Oxford, 372 pp. Bunn, S., Tenakanai, C., & Storey, A. (1999). Energy sources supporting Fly River ﬁsh communities. Report to Ok Tedi Mining Limited, 38 pp. Chenoweth, S. F., Hughes, J. M., Keenan, C. P., & Lavery, S. (1998a). When oceans meet: a teleost shows secondary intergradation at an Indian-Paciﬁc interface. Proceedings of the Royal Society of London Series B, 265, 1–6. Chenoweth, S. F., Hughes, J. M., Keenan, C. P., & Lavery, S. (1998b). Concordance between dispersal and mitochondrial gene ﬂow: Isolation by distance in a tropical teleost, Lates calcarifer (Australian Barramundi). Heredity, 80, 187–197. Copland, J. W., & Grey, D. L. (Eds.) (1987). Management of wild and cultured Sea Bass/Barramundi (Lates calcarifer). ACIAR Proceedings, 20, 1–210. Davis, T. L. O. (1984). Estimation of fecundity in Barramundi, Lates calcarifer (Bloch) using an automatic particle counter. Australian Journal of Marine and Freshwater Research, 35, 111–118. Davis, T. L. O. (1985). Seasonal changes in gonad maturity, and abundance of larvae and early juveniles of Barramundi, Lates calcarifer, in Van Diemen Gulf and the Gulf of Carpentaria. Australian Journal of Marine and Freshwater Research, 36, 177–190. Doupe´, R. G., Horwitz, P., & Lymbery, A. J. (1999). Mitochondrial genealogy of Western Australian Barramundi: Applications of inbreeding coefﬁcients and coalescent analysis for separating temporal population processes. Journal of Fish Biology, 54, 1197–1209. Garrett, R. N. (1987). Reproduction in Queensland Barramundi (Lates calcarifer). ACIAR Proceedings, 20, 38–43. Kare, B. (1995). A review of research on Barramundi, reef ﬁsh, dugong, turtles and Spanish mackerel and their ﬁsheries in the Torres Strait adjacent to Papua New Guinea. Science in New Guinea, 21, 43–56. Keenan, C. P. (1994). Recent evolution of population structure in Australian Barramundi, Lates calcarifer (Bloch): An example of isolation by distance in one dimension. Australian Journal of Marine and Freshwater Research, 45, 1123–1148. Milton, D. A., Tenakanai, C., & Chenery, S. R. (2000). Can the movements of Barramundi in the Fly River region, Papua New Guinea be traced from their otoliths. Estuarine, Coastal & Shelf Science, 50, 855–868. Mobiha, A. M. (1993). Abundance estimates of juvenile Barramundi (Lates calcarifer) along the coast of Western province, Papua New Guinea. PNG Fisheries Research and Survey Branch Technical Paper No. 93-04. 11 pp. Mobiha, A. M., & Murri, P. (1993). Preliminary estimates of the effort involved in the commercial Barramundi (Lates calcarifer) ﬁshery in the Western Province, Papua New Guinea. PNG Fisheries Research and Survey Branch Technical Paper No. 93-03. 6 pp.

Biology of Barramundi in the Fly River System

425

Moore, R. (1979). Natural sex inversion in the giant perch (Lates calcarifer). Australian Journal of Marine and Freshwater Research, 30, 803–813. Moore, R. (1980). Reproduction and migration in Lates calcarifer (Bloch). Ph.D. Thesis, University of London, 213 pp. Moore, R. (1982). Spawning and early life history of Barramundi Lates calcarifer in Papua New Guinea. Australian Journal of Marine and Freshwater Research, 33, 647–661. Moore, R., & Reynolds, L. F. (1973). Fish sales at Daru market. Science in New Guinea, 1, 11–14. Moore, R., & Reynolds, L. F. (1982). Migration patterns of Barramundi, Laters calcarifer (Bloch), in Papua New Guinea. Australian Journal of Marine and Freshwater Research, 33, 671–682. Opnai, L. J., & Tenakanai, C. D. (1987). Review of the Barramundi ﬁshery in Papua New Guinea. In: Copland, J. W., Grey, D. L. (Eds). Management of wild and cultured Sea Bass/Barramundi (Lates calcarifer). ACIAR Proceedings, 20, 50–54. Reynolds, L. F. (1978). Population dynamics of Barramundi, Lates calcarifer (Pisces: Centropomidae) in Papua New Guinea. M.Sc. Thesis, University of Papua New Guinea, 120 pp. Reynolds, L. F., & Moore, R. (1982). Growth rates of Barramundi Lates calcarifer in Papua New Guinea. Australian Journal of Marine and Freshwater Research, 33, 663–670. Russell, D. J., & Garrett, R. N. (1983). Use by juvenile Barramundi, Lates calcarifer (Bloch), and other ﬁshes of temporary supralittoral habitats in a tropical estuary in northern Australia. Australian Journal of Marine and Freshwater Research, 34, 805–811. Russell, D. J., & Garrett, R. N. (1988). Movements of juvenile Barramundi, Lates calcarifer (Bloch), in North-Eastern Queensland. Australian Journal of Marine and Freshwater Research, 39, 203–218. Salini, J. P., & Shaklee, J. B. (1988). Genetic structure of Barramundi (Lates calcarifer) stocks from northern Australia. Australian Journal of Marine and Freshwater Research, 39, 317–329. Salini, J. P., Blaber, S. J. M., & Brewer, D. T. (1990). Diets of piscivorous ﬁshes in a tropical Australian estuary with particular reference to predation on penaeid prawns. Marine Biology, 105, 363–374. Salini, J. P., Brewer, D. T., & Blaber, S. J. M. (1998). Dietary studies on the predatory ﬁshes of the Norman River Estuary, with particular reference to penaeid prawns. Estuarine, Coastal & Shelf Science, 46, 837–847. Salini, J. P., Hughes, J., Milton, D. A., & Hurwood, D. (2004). Genetic structure and diversity of Barramundi in Papua New Guinea and implications for managing the ﬁshery. Journal of Fish Biology (in press). Shaklee, J. B., & Salini, J. P. (1985). Genetic variation and population subdivision in Australian Barramundi, Lates calcarifer (Bloch). Australian Journal of Marine and Freshwater Research, 36, 203–218.

426

S. J. M. Blaber et al.

Shaklee, J. B., Salini, J. P., & Garrett, R. N. (1993). Electrophoretic characterization of multiple genetic stocks of Barramundi (Lates calcarifer) in Queensland, Australia. Transactions of the American Fisheries, 22, 685–701. Swales, S., Storey, A., Roderick, I., & Figa, B. (1999). Fishes of ﬂoodplain habitats of the Fly River system, Papua New Guinea, and changes associated with El Nino droughts and algal blooms. Environmental Biology of Fishes, 54, 389–404. Swales, S., Storey, A., & Bakowa, K. (2000). Temporal and spatial variations in ﬁsh catches in the Fly River system in Papua New Guinea and the possible effects of the Ok Tedi copper mine. Environmental Biology of Fishes, 57, 75–95.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00412-6

Chapter 12

Use of Changes in Fish Assemblages in the Fly River System, Papua New Guinea, to Assess Effects of the Ok Tedi Copper Mine Andrew W. Storey1,, Markson Yarrao2, Charles Tenakanai3, Boga Figa3 and Jessica Lynas1 1

School of Animal Biology (M092), The University of Western Australia, Crawley, WA, 6009, Australia 2 Environment Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, W.P. Papua New Guinea 3 Livelihood Programs Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, W.P. Papua New Guinea

12.1. Introduction The Fly River in Papua New Guinea is one of the largest rivers in Australasia (mean annual discharge B6,000 m3 s1). It has a catchment area of 76,000 km2 and ﬂows a distance of over 1,200 km from its source in the central highlands of New Guinea to the Gulf of Papua. Much of the catchment, particularly in the upper reaches, consists of dense primary tropical rainforest, while in the middle and lower reaches open savannah forest, swamp forest, and seasonally inundated grasslands predominate. Although the upper catchment extends to altitudes of more than 3,500 m, the majority of the drainage basin is low-lying and ﬂat, to the extent that the port of Kiunga, which is 800 river km from the coast, is only 20 m above sea level. The combination of low topography and high rainfall has resulted in a broad ﬂoodplain with extensive shallow lake systems, occupying an area of 4.5 million ha, making it the largest wetland system in the country. The wetlands Corresponding author. Tel.: (618) 6488 1482; Fax: (618) 6488 1029;

E-mail: [email protected] (A.W. Storey).

428

A. W. Storey et al.

of the Fly River are highly productive and play a vital role in the ecology of the river system. The area is sparsely populated, with an average human density of one–two persons per square kilometer. Prior to 1960, the highlands area had little contact with the outside world, and the river downstream was relatively pristine, with no mining or logging, and only low-scale commercial ﬁshing for barramundi. Following the discovery of the Mt Fubilan ore deposit, scientists began surveying the aquatic fauna of this previously unstudied, remote river system (Boyden et al., 1978; Roberts, 1978; DPI, 1979, 1980; Robertson and Baidam, 1983). These studies revealed that the river system has the most diverse freshwater ﬁsh fauna in the Australasian region, and is known to support over 115 freshwater and marine vagrant species (Roberts, 1978; Maunsell and Partners, 1982; Allen, 1991; Coates, 1993; Swales et al., 1999). Of these, 17 are endemic to the Fly Basin, and over 30 are known only from the Fly and one or more of the large rivers in central-southern New Guinea (Roberts, 1978). The ﬁshes of the Fly River basin are characterized by the large individual sizes of some species, an abundance of endemic ﬁshes, and the presence of species that are poorly represented in other parts of the world, particularly the ariid and plotosid catﬁshes. In most other ways, the composition of the freshwater ﬁsh fauna is largely determined by its position in the Australasian zoogeographical zone (Roberts, 1978; Coates, 1993). Given the uniqueness of the ﬁsh fauna and concerns over the potential impact of a mine on its headwaters, environmental monitoring of the Fly River for the proposed mine commenced in 1981 (Maunsell and Partners, 1982). Monitoring began with an impact assessment which involved an expeditionary survey of water quality, ﬁsh communities in different habitats and metal levels in biota (Maunsell and Partners, 1982). Soon after, Ok Tedi Mining Ltd (OTML) implemented an extensive biological monitoring regime that commenced during the mine construction phase in 1983 and included all river reaches, from the headwaters to the delta and into the Gulf of Papua (Wood et al., 1995; Swales et al., 1999). An important aspect of this program was monitoring ﬁsh populations, partly in recognition of their value as a tool for assessing anthropogenic impacts, and as an indicator of ecosystem health and productivity (Fausch et al., 1990; Harris, 1995). It is also because the ﬁsh fauna forms an important subsistence food source for village communities along the river (Hortle, 1986). Due to this reliance on ﬁsh, the main concern of the State of Papua New Guinea was that this resource had to be protected (OTML, 1988, 1990). As a result, monitoring tended to concentrate on numbers and weight of ﬁsh available at locations throughout the river system. This emphasis has continued through mine life, and has inﬂuenced sampling methods, data collected, and how analyzed.

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

429

There have been three operating stages in mine life to date. Stage One (the treatment of gold ores by cyanide extraction) commenced in 1984. Stage Two commenced in mid-1987 and consisted of the gold extraction circuit running in parallel with a ﬂotation circuit to produce copper concentrate. Stage Three commenced in mid-1988 when the gold circuit was decommissioned (and all use of cyanide ceased) and the mine became a copper concentrate producer. Additional infrastructure was completed in 1989, enabling the mine to treat 80,000 t of ore per day at peak production. The next major change was in 1998 when a dredge was deployed on the lower Ok Tedi to dredge the channel with the intention of alleviating bed aggradation in the lower Ok Tedi and Middle Fly. Currently, approximately 50 Mt per annum (pa) of waste rock and 30 Mt pa of tailings are discharged directly into the Ok Tedi and its tributaries, and approximately 15 Mt pa of sediment are dredged from the lower Ok Tedi and deposited in ‘‘cells’’ on the adjacent ﬂoodplain. The original mine plan was to operate with riverine disposal of waste rock, but storage of tailings in a dam in the upper catchment. However, the tailings dam collapsed during construction, and to allow the mine to proceed, the State permitted riverine disposal of detoxiﬁed tailings. The Ok Tedi mine subsequently received notoriety in the mid-1980s when the cyanide detoxiﬁcation system failed on several occasions, releasing tailings with elevated cyanide levels into the Ok Tedi, resulting in the death of ﬁsh, turtles, and crocodiles for approximately 100 km downstream. Since the closure of the gold extraction circuit (1988), the mine has operated with a copper ﬂotation circuit, which recycles the ﬂotation chemicals. As such, the tailings have no elevated chemicals, but are elevated in a suite of particulate and to a lesser extent dissolved metals including Cu, Pb, Zn, Cd, As, and Fe, compared to average crustal abundance (Bolton et al., 2009). Monitoring the effects of the Ok Tedi mine on the river system has continued to present, principally to document and understand impacts resulting from the mode of operation of the mine (Pickup and Cui, 2009). Relationships between changes in ﬁsh stocks in the main river channel and mine waste discharges have been reported (Smith et al., 1990; Smith and Hortle, 1991; Smith and Morris, 1992), as have impacts to the ﬁsh fauna of riverine and ﬂoodplain habitats (Swales et al., 1998, 1999, 2000). Since these publications, additional data have been collected and further analyses conducted. This chapter describes temporal variations in ﬁsh catch and assemblage composition, and updates previous assessments of species diversity and biomass from sites downstream of the mine (OTML, 1994, 1995, 1996; Swales et al., 1998, 1999, 2000).

430

A. W. Storey et al.

12.2. Sampling and Analyses Although monitoring only commenced during mine construction, providing limited baseline data, the sampling method adopted at that time has been maintained throughout the project, providing a standard sampling approach, now covering almost 25 years. This provides an immensely valuable time series, unlikely to be matched from any other river in the world, let alone a large, tropical river system. Such intensive sampling is known to yield a high number of species and more accurate and precise estimates of richness (Cao et al., 2001; Hughes et al., 2002; Reynolds et al., 2003; Kennard et al., 2006). Methods were based on those used by Maunsell and Partners (1982) using a standard set of 13 gill-nets, ranging in stretched mesh size from 25 to 175 mm (Table 12.1). The selection of gill nets as the main method was a compromise over what was readily available and could be used and maintained in a remote region, and what was effective in the conditions and across habitats. It is also a sound, simple, and technically robust method to obtain catch per unit effort (CPUE) data, ideally suited to a remote location where failure of more technical equipment (i.e., electroﬁshers or hydroacoustic gear) can result in many months delay. Seine netting, rotenone, trapping, and electroﬁshing were also used selectively in certain habitats to supplement catches; however, these data are not included in the current

Table 12.1: Gill nets used in the Standard Gill Net Set used by OTML since 1983, giving stretched-mesh size of each net, line type, number set, dimensions, and area. Stretched-mesh size and type 1v (25 mm) Monoﬁlament 1½v (38 mm) Monoﬁlament 2v (50 mm) Monoﬁlament 2½v (63 mm) Monoﬁlament 3v (75 mm) Monoﬁlament 3½v (88 mm) Monoﬁlament 4v (100 mm) Monoﬁlament 5v (125 mm) Monoﬁlament 6v (150 mm) Monoﬁlament (6M) 6v (150 mm) Multiﬁlament (6C) 7v (175 mm) Multiﬁlament (7C)

Number in net set

Length (m)

Depth (m)

Area (m2)

1 1 1 1 1 1 1 1 1 2 2

40 40 45 40 45 45 45 45 50 25 25

2.3 1.7 2.1 2.8 3.2 3.5 4.2 4.9 6 2.8 3.1

92 68 95 112 144 158 189 221 300 70 78

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

431

analyses as they can confound CPUE data across sites and times. Monitoring was infrequent during the early years (1983–1988), but increased to quarterly during the late 1980s and 1990s, with monthly sampling at some locations. However, this frequency has since been reduced to biannually, and at fewer sites (2000 onward). Sites were initially selected to provide a good geographical spread and ease of access to satisfy monitoring programs designed to assess the state of the indigenous subsistence ﬁshery, rather than to provide ecological data. Often sites were adjacent to villages with airstrips, which provided access to the river. Details of the sites sampled and their locations are presented in Table 12.2 and Fig. 12.1, respectively. Sites were stratiﬁed between riverine and ﬂoodplain habitats, respectively. Riverine sites ranged from fast-ﬂowing, forested, cobble/gravel bed channel approx 100 m across and 2–3 m deep in the upper catchment, to slower ﬂowing, meandering forested channel, 150 m wide and 5–10 m deep with silt/ sand substrate in the middle reaches, to slow-ﬂowing sections meandering through seasonally inundated grassed ﬂoodplain, approximately 200–300 m wide, 10–20 m deep with ﬁne silt/mud substrate. Floodplain sites were separated between forested ﬂoodplain in the upper half of the catchment and grassed ﬂoodplain in the lower half. Within these broad regions, there were sites in oxbow lakes, being cut-off meander bends of the river, as wide

Table 12.2: Key long-term monitoring sites and their period of sampling. Site name

Site code

Start of sampling

End of sampling

Riverine sites Ningerum Atkamba New Atkamba Kuambit Bosset Obo Ogwa Strickland River

TED20 TED30 TED35 FLY10 FLY14 FLY15 FLY20 STR01

April 1983 December 1983 January 1993 June 1983 June 1983 April 1987 April 1987 April 1987

March 1996 October 1992 March 2002 October 2003 February 2004 February 2002 February 2004 February 2002

Floodplain sites Bosset Lagoon Daviumbu Sembe Oxbow Lake Pangua Oxbow at ARM345 Strickland

BOS10 DAV01 OXB03 OXB05 OXB06 OXB08

June 1983 April 1987 October 1990 June 1989 June 1993 March 1992

May 1996 February 2002 October 1998 February 2002 March 2002 February 2002

432

A. W. Storey et al.

Figure 12.1: Location of riverine and ﬂoodplain sampling sites on the Fly River system. (150–300 m) and deep (10–20 m) as the adjacent river and lined with either primary rainforest or ﬂooded grasses (Saccharum and Phragmites). The other main ﬂoodplain sites were blocked valley lakes, being broad, shallow lagoons 3–4 m deep and heavily vegetated around the margins with ﬂoating grasses, and submerged and emergent macrophytes. The blocked valley lakes recede

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

433

each dry season and dry totally in El Nin˜o droughts, whereas the deeper oxbow lakes never dry out, providing drought refuge. At riverine sites, nets were set at approximately 301 to the bank in decreasing order of mesh size with distance upstream, and in ﬂoodplain sites they were set perpendicular to the bank, with large mesh nets at either end of the site, and small mesh nets set progressively toward the center. Nets were set at least 50 m apart. Nets of the same mesh size tended to be set at the same location at each site on each occasion (i.e., the same mesh size would be deployed at a speciﬁc backwater at a site on each sampling occasion) to minimize temporal variability. At each site, on each sampling occasion, the nets were deployed for 24 h, checked at dusk and dawn, and all ﬁsh caught identiﬁed to species. Fish length and weight were also recorded. This sampling method provided abundance, biomass, and assemblage composition data for analysis. Monitoring data were then analyzed to determine temporal changes in species richness, biomass, and assemblage composition. To assist detection of mine effects, and acknowledging the different phases of mining, data for each site were subdivided into time periods corresponding to different ‘‘mine operating periods’’ (Pickup and Cui, 2009) (Table 12.3). Temporal changes in species richness and biomass were examined using linear regression for each site. Biomass data were transformed (ln(xþ1)) prior to analysis. Where a signiﬁcant relationship (po0.05) was found, percentage change was determined using the observed data and predictions from regression analysis (i.e., change from beginning of sampling period versus end of period). Changes in the frequency of occurrence of each species in each time period were tabulated and between-time period differences in observed over expected frequency of occurrence tested by Chi-square contingency table analysis (Zar, 1974). Multivariate analyses were also performed to detect changes in ﬁsh assemblages at sites over time. Ordinations were performed using the semistrong hybrid multidimensional scaling (SSH MDS) procedure in the PATN pattern analysis software (Belbin, 1995), using abundance of each species at a site on each sampling occasion. Dissimilarity between samples at a site over times was determined using the Bray-Curtis dissimilarity coefﬁcient, and species occurring in less than 10% of samples in any data set were omitted to avoid ‘‘rare’’ species having a disproportionate effect on the analyses (Gauch, 1982). Sites were analyzed separately, and samples were grouped according to time periods and mining method (Table 12.3) and these groups illustrated on the ordinations. Finally, principal axis correlation (PCC) (Belbin, 1995) was used to test for signiﬁcant gradients in community descriptors (viz. species richness and biomass) through ordinations of each

Start of period End of period Site Riverine (code) Ningerum (TED20) Atkamba (TED30) New Atkamba (TED35) Kuambit (FLY10) Bosset (FLY14) Obo (FLY15) Ogwa (FLY20) Strickland R. (STR01) Floodplain (code) Bosset Lagoon (BOS10) Daviumbu (DAV01) Sembe Oxbow (OXB03) Lake Pangua (OXB05) Oxbow at ARM345 (OXB06) Strickland (OXB08)

Baseline & gold only (Au)

Gold/copper & coppero80,000 tpd (Au & Cu)

CopperW80,000 tpd (1–5 years) (CuW80k 5 year)

CopperW80,000 tpd (6–10 years) (CuW80k 10 year)

Dredge in operation (Dredge)

December 1983 July 1987

August 1987 July 1989

August 1989 August 1993

September 1993 February 1998

March 1998 March 2004

15 13

6 9

4

14 14

9 7 8 9 9

4 23 8 43 14 15 15 14

30 49 14 16 16 16

35 33 16 14 16 14

14

7 7

15 12 12 14 7

10 12 12 14 8

9 10 4 14 13

5

9

4

Note: Au ¼ gold cyanide leachate phase; Au & Cu ¼ gold cyanide leachate and copper ﬂotation circuit; CuW80k 5 year ¼ copper ﬂotation circuit in isolation, running atW80,000 tpd for the ﬁrst ﬁve years of this phase (1989–1993); CuW80k 10 year ¼ the second ﬁve years of this phase (1993–1998); and, Dredge ¼ copper ﬂotation circuit with the dredge operating in the lower Ok Tedi.

A. W. Storey et al.

Time periods

434

Table 12.3: Mine operating periods used to assess temporal changes in ﬁsh assemblage structure at key monitoring sites, indicating the number of samples in each period from each site.

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

435

site. The technique identiﬁes changes in species richness and biomass across ordination space, with the gradient indicating the direction of increasing values of the parameter (i.e., increasing in the direction of samples with high richness/biomass). Gradients of time in days from start of sampling (i.e., as an indication of cumulative exposure to mine effects) were also tested. Although a comprehensive geochemical monitoring program is conducted by OTML, unfortunately the sites and sampling times do not match the ﬁsh sampling sites/times, and as a result there are no quantitative physicochemical data that exactly match individual ﬁsh catch data at each site on each occasion to assist interpretation. However, changes in ﬁsh assemblages correspond with gross patterns of changes in water quality reported elsewhere in this volume (Apte, 2009; Bolton et al., 2009; Pickup and Cui, 2009).

12.3. Changes in Biodiversity A number of methods have been developed worldwide to evaluate anthropogenic disturbance in river systems, including assessments of macroinvertebrates, chemical variables, and ﬁsh diversity (Karr, 1991; Hugueny et al., 1996). While historically, assessments of chemical variables (i.e., nutrients, heavy metals, and salinity) were popular, they do not integrate environmental quality over time. More recently, ﬁsh diversity has been used as an indicator of ecosystem health worldwide (Karr, 1981; Oberdorff and Hughes, 1992; Hugueny et al., 1996; An and Choi, 2003). Because ﬁsh continually inhabit the receiving water, they integrate the chemical, physical, and biological histories of the river. Most ﬁsh species have a long life span and therefore reﬂect both long-term and current water quality. Sampling ﬁsh assemblages can be used to assess a range of environmental disturbances, such as changes in habitat, water quality, and land use (Hugueny et al., 1996). Fish are also one of the most conspicuous biota in the Fly River system, they are relatively easy to sample and identify, but they are also a signiﬁcant indigenous resource.

12.3.1. Riverine Sites Since monitoring began in 1983, 86 species representing 32 families have been recorded from sites on the Fly, Ok Tedi, and Strickland rivers (Swales et al., 2000). Of these, catﬁsh of the families Ariidae (16 spp) and Plotosidae

436

A. W. Storey et al.

(9 spp) were highly dominant in terms of diversity, but with Nematolosa herrings being the most abundant species (forming over 37% of the catch). A number of species known from the Fly River are considered important owing to their large size, including anchovy (Thryssa scratchleyi), catﬁsh (Arius augustus), Papuan bass, (Lutjanus goldiei), and barramundi (Lates calcarifer). Barramundi provides valuable commercial and artisanal ﬁsheries. Of importance, from both an ecological and social perspective, is the fact that declines in species diversity have been recorded from the majority of sites, with the exception of New Atkamba (TED35) and Strickland River (STR01, a control site) (Table 12.4 and Fig. 12.2). Since the commencement of monitoring, reductions in species diversity ranging from 21% (FLY20) to 80% (TED20) have been recorded (Table 12.4). The greatest reductions in the number of species were reported from the more heavily impacted sites, closest to the mine, i.e., Ningerum TED20 (80%) and TED30 (69%). Nonetheless, considerable declines in the number of ﬁsh species collected were also reported from the middle Fly at Bosset FLY14 (46%) and Obo FLY15 (35%). Sampling at TED35 commenced in January 1993, as a replacement for Atkamba (TED30), which was effectively lost when the meander loop was cut off as an oxbow lake. It is likely that the short sampling period for TED35 limited the ability to detect mine effects, but particularly since the period of record covers only the copper ﬂotation period, and the lower Ok Tedi had already been heavily impacted prior to this time. Future monitoring of this site is important since it is the closest riverine site downstream of the dredge. The dredge was implemented to remove sediment from the lower Ok Tedi, prior to discharge into the Fly River and thereby reduce bed aggradation and alleviate overbank ﬂooding and associated environmental impacts downstream. Chi-square contingency table analysis of changes in the frequency of occurrence of individual species from each site over time, showed that the greatest loss and declines in occurrence of species occurred at sites closest to the mine (i.e., TED30, FLY10, FLY14, and FLY15) (Table 12.5). Species particularly noted for their declines from riverine sites include many of the forktailed catﬁshes; Arius latirostris, Arius carinatus, Arius augustus, Cinetodus crassilabris, Arius macrorhynchus, Cinetodus froggatti, and Nedystoma dayi. The mullet Liza alata (diadema) has also declined at the most impacted riverine sites. Most of these species are resident in the river system, except the mullet, which likely has an estuarine/marine afﬁnity. Little is known of the biology of the individual species, which limits the ability to interpret reasons for their declines; however, issues such as increased suspended sediment loads, mobile bed loads, smothering of habitats, loss of

Site

Reduction in species richness

Reduction in biomass

p

r-square

% reduction

p

r-square

% reduction

Riverine

Ningerum (TED 20) Atkamba (TED30) New Atkamba (TED35) Kuambit (FLY10) Bosset (FLY14) Obo (FLY15) Ogwa (FLY20) Strickland R (STR01)

po0.0001 po0.0001 pW0.05 po0.0110 po0.0001 po0.0001 p ¼ 0.0048 pW0.05

0.620 0.485 – 0.043 0.463 0.262 0.133 -

90 75 None 22 46 35 21 None

po0.005 po0.0001 pW0.05 po0.0001 po0.0001 p ¼ 0.0008 pW0.05 p ¼ 0.0075

0.257 0.515 – 0.272 0.278 0.200 – 0.132

86 87 None 79 75 57 None 49

Floodplain

Bosset Lagoon (BOS10) Lake Daviumbu (DAV01) Sembe Oxbow (OXB03) Lake Pangua (OXB05) Oxbow ARM345 (OXB06) Strickland Oxbow (OXB08)

p ¼ 0.0112 p ¼ 0.038 pW0.05 p ¼ 0.0157 p ¼ 0.0082 pW0.05

0.115 0.106 – 0.137 0.240 –

32 26 None 25 25 None

p ¼ 0.0021 p ¼ 0.044 pW0.05 p ¼ 0.0102 p ¼ 0.0103 pW0.05

0.165 0.100 – 0.154 0.228 –

72 64 None 66 52 None

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

Table 12.4: Regression statistics and percent reduction in species richness and biomass from riverine and ﬂoodplain sites.

437

438

A. W. Storey et al.

Figure 12.2: Regressions of species richness against time at riverine sites, showing signiﬁcance level, r-square, percent reduction from start of sampling, and 95% prediction intervals around the linear regression line.

food resources, and possible issues such as chronic toxicity to eggs/larvae could all play a role in their decline. Concomitant with declines in ﬁsh biodiversity has been the deposition of mine-derived waste sediment, resulting in elevated sediment loads, increased particulate, and dissolved copper levels, die-back of ﬂoodplain forest and

Family Ambassidae

Anabantidae Apagonidae Ariidae

Ambassis agrammus Ambassis macleayi Ambassis spp Parambassis gulliveri Anabas testudineus Glossamia aprion Glossamia trifasciata Arius augustus Arius agreutes Arius berneyi Arius carinatus Arius latirostris Arius leptaspis Arius macrorhynchus Arius taylori Arius sp. A Cinetodus crassilabris Cinetodus froggatti Cochlefelis spatula Cochlefelis danielsi Nedystoma dayi Tetranesodon conorhynchus Craterocephalus randi Strongylura kreffti Carcharhinus leucas Lates calcarifer

TED30 TED35 FLY10 ns – – k – ns – – ns ns k ns k ns – ns ns ns – – ns – k

ns – – ns ns ns – – ns m ns m k – ns ns ns – – – ns ns – ns

m – ns k m ns – ns k k ns k ns – k k k ns k – – k

FLY14

FLY15

FLY20

STR01

ns ns – ns m ns – k – ns k ns k – – k k ns – – ns – ns

ns – – ns ns ns – – ns ns ns ns k – – ns ns ns – – – ns

– – ns ns ns – k ns ns k ns – – k ns ns ns ns ns – ns ns ns

ns – – ns ns ns – ns ns ns k ns ns ns ns k ns ns ns ns ns – ns – ns

439

Atherinidae Belonidae Carcharhinidae Centropomidae

Species

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

Table 12.5: Summary of changes in species occurrences at each riverine site.

440

Table 12.5: (Continued ).

Clariidae Clupeidae Datnioididae Eleotridae

Engraulidae

Gobiidae

Hemiramphidae Kurtidae Lutjanidae Megalopidae Melanotaeniidae Mugilidae

Species Clarias batrachus Clupeoides papuensis Nematalosa spp Datnioides quadrifasciatus Mogurnda cingulata Mogurnda mogurnda Ophieleotris aporos Oxyeleotris ﬁmbriata Oxyeleotris herwerdini Oxyeleotris nullipora Oxyeleotris lineolatus Oxyeleotris spp Thryssa rastrosa Thryssa scratchleyi Thryssa spp Glossogobius concavifrons Glossogobius giurus Glossogobius sp. Zenarchopterus novaeguinae Kurtus gulliveri Lutjanus argentimaculatus Lutjanus goldiei Megalops cyprinoides Melanotaenia splendida Crenimugil labiosus Liza alata (diadema)

TED30 TED35 FLY10 – ns k ns – – – ns – – ns – – – – – – ns ns – ns ns ns –

ns ns m ns – – – – ns – ns – ns ns – ns ns – ns – – ns ns ns – ns

m ns ns ns ns – ns m – ns – – ns ns ns k k ns ns k

FLY14

FLY15

FLY20

STR01

m ns ns ns – – – m – ns ns ns – – – – – ns ns ns k ns ns k

ns ns ns ns – ns – ns ns ns ns – k ns – – – – – – ns k – k

ns – ns ns – – ns ns ns ns ns – ns ns – – – ns ns ns ns ns ns – ns

ns ns ns ns – – – ns ns – ns – ns ns – – – – – ns ns ns – – ns

A. W. Storey et al.

Family

Pristidae Scatophagidae Sciaenidae Sparidae Terapontidae

Toxotidae No.of species lost ( ) No. of species in decline (k) No. of species increasing (m) No. of species with sign variability (-)

– – – ns ns ns ns – – – – – – ns – – ns – k – 7 6 0 0

– – ns ns – – k ns – – – – ns – ns – ns ns ns – 2 2 3 2

– ns k ns ns ns ns m k – – – ns k ns ns m – 3 15 6 10

– ns ns ns – – ns ns ns ns ns k ns ns ns ns ns ns ns 8 8 3 0

– – ns ns – – – – ns – – ns ns ns ns – ns – ns ns ns 5 4 0 2

ns – ns ns – – – ns – – ns ns ns ns ns – – ns ns ns ns 3 3 0 2

– – ns ns – – – – ns ns – ns – ns ns ns – – – ns ns ns 1 2 0 1

441

Notes: ¼ species previously common, but not recorded in last time period; k ¼ species still present but showing a signiﬁcant decline (po0.05) according to Chi-square test; - ¼ a signiﬁcant change between time periods, but with lowest occurrences in intermediate years; ns ¼ no signiﬁcant change in occurrence according to Chi-Square test; ‘‘–’’ ¼ species never recorded from site. NB analysis was not conducted for TED20 because this site was seldom sampled after 1987 because ﬁsh catch had declined so severely.

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

Osteoglossidae Plotosidae

Liza macrolepis Liza subviridis Scleropages jardini Neosilurus ater Neosilurus equinus Neosilurus sp.C Oloplotosus luteus Plotosus papuensis Porochilus obbesi Porochilus meraukensis Porochilus spp Pristis microdon Scatophagus argus Nibea semifasciata Acanthopagrus berda Amniataba percoides Hephaestes fuliginosus Hephaestes roemeri Pingalla lorentzi Terapon lacustris Toxotes chatareus Toxotes lorentzi

442

A. W. Storey et al.

aggradation of the riverbed. Previous studies have ruled out acute copper toxicity to ﬁsh, as levels of dissolved copper were not sufﬁciently elevated to exceed riverine complexing capacity; labile copper levels were generally low (Smith et al., 1990; Smith and Hortle, 1991; Stauber, 1995). Habitat loss due to riverbed aggradation is likely one major reason for reductions in ﬁsh biodiversity from riverine sites of the Ok Tedi and Fly River (Storey et al., 2009). As sediment is transported and deposited downstream, diversity of potential ﬁsh habitat is reduced through the inﬁlling of backwaters, smothering of woody debris, reduction in bank proﬁle, and loss of emergent vegetation and channel heterogeneity. In addition, ﬁsh may be affected by the increased water velocity, and indirectly by changes to food web structure, including declines in invertebrate diversity, reduced terrestrial inputs (insects and fruits) from the forest, and likely loss of in-stream algal production (Storey and Yarrao, 2009). The changes in occurrence of some species may not be mine related. For example, Allen (1991) recorded the freshwater saw ﬁsh, Pristis microdon, as being common from the middle Fly River, but over the last 15 years this species has not been collected upstream of Everill Junction. It is, however, still commonly caught in gill nets downstream of this point and within the Strickland River. Owing to its elongate snout and rows of laterally projecting teeth, P. microdon is easily caught in gill nets. It is because of its morphology that it is likely to have been ‘‘ﬁshed-out’’ of the middle Fly by the increased incidence of gill netting by villagers. Because of its susceptibility to ﬁshing pressure and habitat loss across its range, this species has been listed as ‘‘endangered’’ on the IUCN Redlist (2004). However, it is also equally plausible that this species is susceptible to mine impacts and avoids the Fly River upstream of the major point of dilution at Everill Junction. It is generally accepted that there has been a loss of species from the Ok Tedi (Wood et al., 1995). However, it was previously suggested that all species were either still present in tributaries or within the main river and therefore, the system retained its capacity to return to pre-mine composition (Storey, 1997). Current data and analyses indicate however that some species are becoming even more restricted in their distribution and may be missing from the Ok Tedi and the Middle Fly. This would seriously restrict the capacity of the system to recover post-mine closure. In response to this concern, an extensive ﬁsh diversity survey was conducted in mid-2005 (WRM and Hydrobiology, 2007). The study showed that numbers of migratory species were much reduced in the middle and upper Fly, including Barramundi (Lates calcarifer), Oxeye herring (Megalops cyprinoides), mullet species (Liza diadema/alata, Crenimugil heterocheilus), Croakers (Nibea squamosa), Pikey bream (Acanthopagrus berda), gobies (Glossogobius spp.),

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

443

snappers (Lutjanus goldei, L. argentimaculatus), Soles (Asseraggodes klunzingerii), and Tongue soles (Cynoglossus heterolepis). Many of the Ariidae (forktailed) catﬁsh species expected to be taken from the middle Fly were absent or rarely caught, such as A. latirostris, Co. spatula, Co. danielsi, A. taylori (robertsi), A. augustus, Ci. carinatus, Ci. froggatti, Ci. crassilabris, and N. dayi. This probably reﬂects mine-related impacts. Although these catﬁsh were absent/rare in the middle Fly, most of these species were found elsewhere in the river system, either in the lower Fly/Strickland or the upper Fly tributaries (Elevala/Palmer). However, ﬁshing pressure, especially in the upper Fly may undermine the value of this area as a refuge area. Tributaries of the upper Fly (Palmer/Elevala) and the larger Ok Tedi tributaries (Ok Mart, Ok Menga) continued to support a relatively diverse range of species, and continue to provide a refuge area from which species could colonize the middle Fly/Ok Tedi post-mine closure. However, abundances in these tributaries were generally lower than occurred in the Middle Fly, increasing the risk of loss of genetic diversity. The Fly River below Everill Junction also supported species absent/rarely taken from other parts of the system, but this area is less secure, especially if mine-related impacts progress into the lower Fly (WRM and Hydrobiology, 2007).

12.3.2. Floodplain Sites The Fly River ﬂoodplain has been found to support diverse and abundant assemblages of freshwater ﬁsh, with 66 species from 33 families being recorded (Swales et al., 1999). In a similar way to the main river channel, ﬁsh diversity on the ﬂoodplain was found to be dominated by catﬁsh of the families Ariidae (11 species) and Plotosidae (7 species). Species richness ranged from 32 (OXB06) to 46 (BOS10). Generally, ﬁsh diversity was more consistent at oxbow sites in comparison to blocked valley lakes and grassed ﬂoodplain sites. Marked inter-annual variations were evident from the latter sites, with a notable decline in diversity recorded during 1993/94. This coincided with El Nin˜o drought conditions which resulted in very low river levels, with most habitats on the ﬂoodplain drying. The exception was the oxbow lakes, which, being deeper (generally 10–20 m deep), provided a drought refuge (Smith and Bakowa, 1994). Observed differences in ﬁsh communities from the different ﬂoodplain habitats likely reﬂect differences in physical habitat, and the tendency for the grassed ﬂoodplain and shallower blocked valley lakes (3–4 m deep) to dry out in times of drought.

444

A. W. Storey et al.

Since monitoring began in 1983, reductions in ﬁsh biodiversity have also been detected from ﬂoodplain sites. In fact, at four of the seven sampling sites, a signiﬁcant decline in species diversity has been recorded since the commencement of biological monitoring. However, reductions were generally less severe than was noted for riverine sites, and ranged from 25% (OXB05 and OXB06) to 32% (BOS10) (Table 12.4 and Fig. 12.3). The site closest to the Ok Tedi (OXB06) experienced a 25% reduction in species diversity. Chi-square contingency table analysis showed that within the ﬂoodplain sites, the highest rates of loss and decline in species were from Lake Pangua (OXB05), Bosset Lagoon (BOS10), Sembe Oxbow (OXB03), and Oxbow at ARM345 (OXB06) (Table 12.6). Species which recorded consistent declines at ﬂoodplain sites included Oxyeleotris ﬁmbriata and Oxyeleotris lineolatus.

Figure 12.3: Regression of species richness against time for ﬂoodplain sites, showing signiﬁcance level, r-square, percent reduction from start of sampling, and 95% prediction intervals around the linear regression line.

Table 12.6: Summary of changes in species occurrences at each ﬂoodplain site.

Anabantidae Ambassidae

Apagonidae Ariidae

Anabas testudineus Ambassis agrammus Ambassis macleayi Ambassis spp Denariusa bandata Parambassis gulliveri Glossamia aprion Glossamia trifasciata Arius agreutes Arius augustus Arius berneyi Arius carinatus Arius latirostris Arius leptaspis Arius macrorhynchus Arius taylori Arius sp.A Cochlefelis spatula Cochlefelis danielsi Cinetodus crassilabris Cinetodus froggatti Nedystoma dayi Tetranesodon conorhynchus Craterocephalus randi Strongylura kreffti Carcharhinus leucas Lates calcarifer Clarias batrachus

BOS10

DAV01

OXB03

OXB05

OXB06

OXB08

m ns ns – ns ns – – ns ns – – ns – – – ns – – – – – – k ns

ns ns – ns ns ns ns ns – – ns – ns – – – – – – – – – ns – ns ns

ns ns – – – ns ns – – – ns – – ns – – – – – – – ns – ns ns – k ns

ns ns ns – – ns ns – – ns ns – – ns – – – ns ns – ns – – ns ns – ns

ns ns ns – ns ns – – – ns – – ns – – – – – – – – – – ns – ns –

ns ns ns – – ns ns – – ns ns ns – ns – – – – ns ns ns m – – ns – ns –

445

Atherinidae Belonidae Carcharhinidae Centropomidae Clariidae

Species

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

Family

Table 12.6: (Continued ).

Datnioididae Eleotridae

Engraulidae

Gobiidae

Hemiramphidae Kurtidae Lutjanidae Melanotaeniidae Megalopidae Mugilidae

Clupeoides papuensis Nematalosa spp Datnioides quadrifasciatus Mogurnda mogurnda Mogurnda cingulata Ophieleotris aporos Oxyeleotris ﬁmbriata Oxyeleotris nullipora Oxyeleotris lineolatus Oxyeleotris herwerdini Oxyeleotris spp Thryssa scratchleyi Thryssa rastrosa Thryssa spp Glossogobius giurus Glossogobius sp. Glossogobius concavifrons Stenogobius lachneri Zenarchopterus novaeguinae Kurtus gulliveri Lutjanus argentimaculatus Lutjanus goldiei Melanotaenia maccullochi Melanotaenia splendida Megalops cyprinoides Crenimugil labiosus Liza alata (diadema) Liza macrolepis Liza subviridis

BOS10

DAV01

OXB03

OXB05

OXB06

OXB08

– ns – – – – m – k ns – – – – – ns ns ns – ns ns – ns – –

– ns ns – – ns – m – – – – – – – – ns ns – ns ns – – –

– ns – – – – ns m – ns – – – – – ns – ns ns ns ns – – –

– ns ns – – – – m – ns – – – – ns ns – – k – ns k – k – –

– ns k – – – ns – ns ns – ns ns – – – – – ns ns – ns ns – ns – –

– ns ns – – – ns – ns ns – ns ns – – – – – ns ns ns – ns ns – ns – –

A. W. Storey et al.

Clupeidae

Species

446

Family

Osteoglossidae Plotosidae

Toxotidae No. species lost ( ) No. species in decline (k) No. species increasing (m) No. species with sign variability (-)

k ns – – ns ns – ns m – – – – – – ns ns – ns 6 3 3 4

ns ns – ns ns ns – ns ns ns – – – ns – – ns ns ns ns 3 0 1 6

ns ns – – ns – – ns ns ns – – – – – – – ns m – k ns 4 2 2 1

ns ns – – ns – – ns – – ns ns k – – ns ns – ns 5 4 1 2

ns – – – ns – k ns – – ns – – – ns ns – ns – 3 2 0 3

ns ns – – – ns – ns – ns – ns ns – – – ns ns – ns – 2 0 1 0

447

Notes: ¼ species previously common, but not recorded in last time period; k ¼ species still present but showing a signiﬁcant decline (po0.05) according to Chi-square test; - ¼ a signiﬁcant change between time periods, but with lowest occurrences in intermediate years; ns ¼ no signiﬁcant change in occurrence according to Chi-Square test; ‘‘–’’ ¼ species never recorded from site.

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

Pristidae Scatophagidae Sciaenidae Sparidae Terapontidae

Scleropages jardini Neosilurus ater Neosilurus equinus Neosilurus sp.C Neosilurus brevidorsalis Plotosus papuensis Oloplotosus luteus Porochilus obbesi Porochilus meraukensis Porochilus spp Pristis microdon Scatophagus argus Nibea semifasciata Acanthopagrus berda Hephaestes roemeri Hephaestes fuliginosus Pingalla lorentzi Amniataba percoides Terapon lacustris Terapon jarbua Toxotes chatareus Toxotes lorentzi

448

A. W. Storey et al.

However, since Oxyeleotris herwerdini increased in occurrence at most sites from which the former species declined, and these species are very similar, it is likely the reported declines are actually due to misidentiﬁcation, although species replacement cannot be discounted. The mullet Liza alata (diadema) has also declined at a number of ﬂoodplain sites (i.e., OXB05, OXB03, and BOS10), probably reﬂecting the declines in riverine sites. Natural environmental change likely explains the signiﬁcant declines in species diversity reported from Daviumbu (DAV01), Pangua (OXB05), and Bosset (BOS10). El Nin˜o droughts in recent years (1983, 1986, 1992, 1993, and 1997), caused large areas of the ﬂoodplain to dry out with adverse effects on ﬁsh communities through habitat loss and associated changes in water quality. These effects can be seen in the monitoring data. Chi-square analysis showed that for a number of species, frequency of occurrence reduced in intermediate years, but subsequently recovered (i.e., Arius berneyi, Porochilus obbesi, Porochilus meraukensis, and Amniataba percoides). This undoubtedly reﬂects declines in populations of these species at blocked-valley lake sites in El Nin˜o years, with subsequent lags in recovery. Lakes Daviumbu and Bosset, as with other blocked valley lakes dry during El Nin˜o droughts. As they re-ﬂood they become covered in a continuous, dense (1 m thick) mat of ﬂoating grass (wild rice – Oryza sp.) which seems to prevent recolonization by ﬁsh. This is likely due to low light and low dissolved oxygen levels under the grasses. Assuming the lagoon does not dry in the intervening time, it appears to take 12–18 months for the ﬂoating grasses to break up, disperse, and die off. It is not until the water body is predominantly open water that ﬁsh numbers and diversity start to recover. These changes do not appear at oxbow lake sites, because they are much deeper and never dry. However, changes observed at Lake Pangua appear to be due to a different process. This oxbow is very deep (þ22 m) and becomes strongly stratiﬁed in oxygen and temperature (AWS, unpublished data). The observed algal blooms and declines in ﬁsh catch are likely the result of the occasional breakdown of stratiﬁcation and mixing of the water column leading to algal blooms and intermittent anoxia. This process appears to recur in Lake Pangua, but not in other oxbow lakes. The exact reason is not known, but could be because it is deeper than most oxbow lakes (W22 m), which may result in stronger stratiﬁcation and anoxia of hypolimnetic waters; there may also be nutrient enrichment at this oxbow from a nearby village, which helps drive algal blooms. While in earlier years all changes in ﬁsh structure of ﬂoodplain sites were attributed to natural events and climatic phenomena (Swales et al., 1999), mine-related declines in ﬁsh diversity are now becoming apparent (viz. OXB06). A number of explanations associated with mine operations have

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

449

been postulated. As a result of bed aggradation in the main channel, there is increased frequency and duration of inﬂow to the oxbow lakes. This results in increased suspended sediment (TSS) levels in the water column, increased dissolved metal levels with the potential for chronic or possibly acute toxicity to fauna, and loss of riparian vegetation around the water bodies associated with forest die-back. A number of investigations are currently underway to try to separate the possible effects of these stressors. Riverbed aggradation likely impacts ﬂoodplain habitats through overbank ﬂooding, inputs of mine-derived sediments, and die-back of ﬂood-intolerant vegetation (Pickup and Cui, 2009). Recent analysis has detected a change in the algal contribution to the food web of OXB06, which may be related to die-back and sediment inputs (Storey and Yarrao, 2009). An issue likely to affect ﬂoodplain (and riverine) habitats is the development of localized acid rock drainage (ARD) on levee banks and tie-channel banks of the middle Fly (Bolton et al., 2009; Apte, 2009). Oxidization (exposure to air) of acidgenerating sediments on levees and subsequent wetting and leaching lead to the release of metals, such as copper, into adjacent surface waters. This process seems particularly prevalent at the end of a dry period, with the onset of localized rainfall which ‘‘leaches’’ the levees. This has resulted in highly elevated levels of dissolved metals in surface waters immediately adjacent to the levees. Tie-channels connecting oxbow lakes to the main channel are particularly susceptible to deposition of mine-derived sediments and the establishment of ARD. It is also known that ﬁsh are sensitive to increases in heavy metals (including copper) and show an active preference for waters with low metal levels (Sorrenson, 1991). Therefore, ﬁsh may avoid areas leaching metals, including tie-channels, dissuading migratory species such as mullet, barramundi, ox-eye herring, and bass from entering these water bodies. Reductions in species diversity within ORWBs is of particular concern since it suggests ﬁsh lost from riverine sites are not using ﬂoodplain habitats as refuges against adverse conditions in the Ok Tedi and Fly rivers. This perhaps implies the loss of species from parts of the Fly River system, with implications for the maintenance of a subsistence ﬁshery reliant on these species, and also for ecosystem recovery following mine closure.

12.4. Changes in Fish Catch (Biomass) At the onset of mine development, the importance of the ﬁshery resource of both the river channel and ﬂoodplain to local people was recognized. In fact,

450

A. W. Storey et al.

ﬁsh catch was set as a compliance indicator, to be reported to the State of PNG on a regular basis to ensure protection of the subsistence ﬁshery (OTML, 1988, 1990). Emphasis has therefore been placed on analyzing monitoring data to assess changes in ﬁsh catch over time, using total biomass of ﬁsh taken in the standard 24 h gill net set. 12.4.1. Riverine Sites Over the entire monitoring period, considerable reductions in biomass have been detected from the majority of riverine sites. Signiﬁcant reductions in biomass ranged from 49% at the Strickland River site (STR01) to 92% in the Ok Tedi (TED20) (Table 12.4 and Fig. 12.4). The site on the Ok Tedi in closest proximity to the mine (TED20) recorded the most extreme reduction in biomass (92%), closely followed by the next downstream site (TED30, 88%). Within the Fly River, the extent of declines in biomass decreased with increasing distance from the mine, i.e., Kuambit FLY10 (79% reduction), Bosset FLY14 (75%), and Obo FLY15 (57%) (Table 12.4 and Fig. 12.4). Barramundi, which formed a high proportion of catch biomass at many sites, particularly in the middle Fly, declined in number at most sites following peak numbers in the early 1990s following a period of good recruitment of this catadromous species from coastal areas. While Ogwa (FLY20) showed a reduction in species diversity over time, it did not show declines in biomass. This can perhaps be explained by the remaining ﬁsh at this site being comprised of large species. However, the linear regression plot suggests a trend in decreasing ﬁsh catch, with the last three sampling events recording lower than average biomass. If this trend continues, declines may become signiﬁcant in the future. Interestingly, ﬁsh biomass also decreased at the site on the Strickland River over the monitoring period (STR01); however, this site exhibited the lowest level of decline. STR01 was originally selected as a control site, but the subsequent development upstream of the Porgera gold mine has meant that it can no longer be regarded in this way. The possible effects of the Porgera mine on the Strickland River are unknown. 12.4.2. Floodplain Sites Since 1983, ﬁsh catch has generally been higher from oxbow lake sites (mean biomass range 214–280 kg) than at either blocked valley lakes (54–144 kg) or ﬂoodplain sites (47–104 kg). Such differences can perhaps be explained by the

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

451

Figure 12.4: Regressions of ﬁsh catch biomass against time at riverine sites, showing signiﬁcance level, r-square, percent reduction from start of sampling, and 95% prediction intervals around the linear regression line. habitat at these sites, with the shallower, well-vegetated blocked valley lakes supporting more of the smaller ﬁsh species (i.e., Iriatherina werneri, Ambassis agrammus, and Melanotaenia sp), and the deeper oxbow lakes supporting a greater number of the larger predatory species, such as Lates calcarifer and Scleropages jardinii. In 1995–1996, 11 species recorded from the oxbow lakes

452

A. W. Storey et al.

(Cinetodus froggati, Nedystoma dayi, Lates calcarifer, Parambassis gulliveri, Clarias batrachus, Ophieleotris aporos, Thryssa rastrosa, Thryssa scratchleyi, Kurtus gulliveri, Lutjanus goldiei, and Pristis microdon) were not recorded from blocked valley lakes. Only one species (Iriatherina werneri) recorded from blocked valley lakes was not found in oxbow lakes. Some of these differences may be inﬂuenced by El Nin˜o droughts, drying of wetlands, and rates of recolonization. A large number of ﬂoodplain sites also showed trends of decreasing biomass (Table 12.4 and Fig. 12.5). In fact, a signiﬁcant decline in ﬁsh biomass was detected from most sites, with reductions as high as 72% (BOS10). Earlier suggestions of natural environmental change no longer account for all reported changes in ﬁsh catch, with sites in close proximity to the mine increasingly showing effects (i.e., OXB06). Possible natural and

Figure 12.5: Regressions of ﬁsh catch biomass against time at ﬂoodplain sites, showing signiﬁcance level, r-square, percent reduction from start of sampling, and 95% prediction intervals around the linear regression line.

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

453

mine-related causes for declines in biomass are the same as those reported above for reductions in biodiversity in ﬂoodplain habitats.

12.5. Changes in Assemblage Composition Much of the previous work on ﬁsh fauna of the Fly River and its ORWBs has utilized univariate approaches for analyses, with the exception of Smith and Morris (1992) and OTML (1993). Although univariate analyses provide an indication of how an individual species may change temporally or spatially, they do not indicate how ﬁsh assemblage structure as a unit may have changed. The total loss and subsequent replacement of a species (i.e., by an introduced species) would not be detected using univariate methods, for example, since the total number of species would remain unchanged. However, multivariate analysis is able to identify a change in overall assemblage structure, statistically representing the degree of change (percentage similarity) between assemblages (Belbin, 1995). Multivariate approaches are widely used and accepted in community ecology (Gauch, 1982).

12.5.1. Riverine Sites All sites showed signiﬁcant separation among the different mine-operating time periods (Analysis of Similarity; ANOSIM, po0.05). Generally, samples from the most recent mine-periods (i.e., CuW80k & Dredge) displayed the greatest separation, with earlier time periods showing some overlap (Fig. 12.6). This suggests ﬁsh assemblage change has been more pronounced in recent times. Multidimensional Scaling (MDS) supported results from linear regression, with Ok Tedi (TED20 and TED30) and Fly River sites closer to the mine (FLY10, FLY14, and FLY15) showing reductions in species diversity and biomass over the monitoring period. This was evident from the signiﬁcant gradients in time from these sites, indicating a progressive change in assemblage structure over time (Fig. 12.6). Furthermore, gradients in biomass and species diversity were directly opposite to the time gradient, indicating a decline in these parameters over time (Fig. 12.6). Sites geographically further from the mine recorded less change in their ﬁsh assemblages than those from sites closer to the mine. Gradients in biomass and diversity from ordinations of the Fly River at Ogwa (FLY20) and the Strickland River (STR01), for example, were not directly opposite to the time gradient, indicating less severe changes to assemblage structure than reported from impacted sites (Fig. 12.6).

454

A. W. Storey et al.

Figure 12.6: MDS ordination of species assemblage data from riverine sites, indicating signiﬁcant gradients in assemblage descriptors and time. New Atkamba (TED35), although on the Ok Tedi and close to the mine site, showed no signiﬁcant change in either species richness or biomass using linear regression, as discussed above. Multivariate analyses did however detect a signiﬁcant difference in ﬁsh assemblages between dredge and predredge operating periods (ANOSIM, po0.05). Signiﬁcant gradients in

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

455

species richness, biomass, and time were in the direction of samples collected post-dredge construction (Fig. 12.6). This infers an increase in species richness and biomass at this site following implementation of the dredge, suggesting possible improvements in ﬁsh assemblages.

12.5.2. Floodplain Sites Floodplain sites generally demonstrated less change in ﬁsh assemblage structure than riverine sites. The majority of sites along the ﬂoodplain did not record any signiﬁcant separation of mine-operating periods. Furthermore, gradients in time were not signiﬁcant for most sites, indicating temporal changes in assemblage composition were not great (Fig. 12.7).

Figure 12.7: MDS ordination of species assemblage data from ﬂoodplain sites, indicating signiﬁcant gradients in assemblage descriptors and time.

456

A. W. Storey et al.

The ﬂoodplain site closest to the mine (OXB06) appears to be showing signs of a developing change. Although separation of mine-operating periods was not signiﬁcant (ANOSIM, pW0.05), samples collected between April 1999 and March 2002 (within the dredge period) showed considerable separation from the majority of samples in earlier time periods and from other samples in the dredge period (Fig. 12.7). The signiﬁcant gradient in time was in the direction of this group of samples, indicating a recent, but consistent change in ﬁsh assemblages in this oxbow. Gradients in species richness and biomass were also signiﬁcant, but were perpendicular to the time gradient, indicating a weak reduction in these parameters over time. Lake Pangua (OXB05) not only recorded a signiﬁcant gradient in time, but also signiﬁcant gradients of species richness and biomass which were in the opposite direction to time through the ordination (Fig. 12.7). This suggests that ﬁsh assemblages from Lake Pangua have declined in species richness and biomass over the monitoring period. Although changes to ﬁsh assemblages in this system are not entirely understood, and may be the result of environmental (El Nin˜o, algal blooms, and anoxia) and social conditions (artisanal ﬁshing and input of nutrients from the nearby village), the possibility of a mine-related impact should not be discounted. The oxbow at Sembe (OXB03) did have a signiﬁcant gradient in time, indicating a systematic change over time. Gradients in biomass were also signiﬁcant, and the earliest samples (CuW80K, 5 years) tended to separate from samples collected during later time periods (Fig. 12.7). As with Lake Pangua, this may also be mine related, but more data are required.

12.6. Conclusions Analyses presented here, combined with results from a recent ﬁsh diversity study of the Fly River system (WRM and Hydrobiology, 2007) demonstrated that the majority of ﬁshes of the Fly River system continue to maintain populations within the catchment, but there has been a marked reduction in the diversity and biomass of ﬁshes in most reaches downstream of the Ok Tedi mine, as far as Everill Junction. Mine closure is currently planned for 2013; however, because of the volume of waste material stored in the upper catchment, and the rate of erosion and sediment transport through the system, the full impacts in the middle and lower Fly River will not be apparent for up to 50 years after mine closure (Pickup and Cui, 2009). It is therefore likely that impacts in the middle Fly may become worse, and ﬁsh populations downstream of Everill Junction may be affected in the future.

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

457

More than half the ﬁsh species formerly resident in the habitats of the Ok Tedi and Middle Fly sub-catchments are effectively no longer resident there. Populations of most species still occur in other parts of the catchment, and this provides the potential for recovery of the ﬁsh assemblages post-mining. However, four species (Clupeoides venulosus, Glossogobius celebius, Oxyurichthys papuensis, and Hephaestus raymondi) have not been found in the Fly River system since 2000 (WRM and Hydrobiology, 2007). If lost, this obviously limits the potential of the system to recover post-mine closure. The impacts of the mine on the ﬁshery are indisputable; however, an additional factor that will further limit the ability of the system to recover post-mine is the presence of three exotic ﬁshes – Climbing perch (Anabas testudineus), Striped snakehead (Channa striata), and Walking catﬁsh (Clarias batrachus). These species have all invaded the Fly River system since 1988 (Storey et al., 2002). They were most likely brought from Indonesia by transmigrants settling in West Papua close to the border with Papua New Guinea. From here they likely escaped into the Trans-Fly ﬂoodplain system. All three are now well established throughout the Fly River, from the Lower Fly to the upland streams along the Kiunga-Tabubil Highway. This is of concern with respect to the potential recovery of native ﬁsh populations after mine closure. Exotic ﬁshes commonly have competitive advantages over native ﬁsh species in river systems that have been highly modiﬁed (see discussions in Storey et al., 2002 and Dudgeon and Smith, 2006). The Fly River is now a highly modiﬁed system, with substantial habitat alteration resulting from the operations of the Ok Tedi mine, but also from the increased human population attracted by the mine. There are many new settlements downstream of the mine, with associated pressures on water courses as a result of clearing for gardens, everyday activities (e.g., washing, disposal of efﬂuent, etc.), and increased ﬁshing pressure. It can be expected that these exotic species will exert adverse pressure on the recovery of native ﬁsh populations. A recent study on the Fly River (August 2007) noted these exotic species in creeks and backwaters, and where exotics were prevalent native species were almost absent (A.W. Storey, personal observation). Unfortunately, it will not be possible to eradicate these exotic species, and the long-term impacts of these (and potentially other) introduced species on the native ﬁsh fauna of the Fly River will likely be far greater than the relatively short-term impacts of the Ok Tedi mine (G.R. Allen, Western Australian Museum, personal communication). Another signiﬁcant recent observation is the proliferation of small mesh (r2v), lightweight Indonesian-made gill nets in the middle and upper Fly. These nets intentionally target small-bodied species, and it is a well-accepted progression that subsistence (and artisanal) ﬁsheries resort to smaller gear

458

A. W. Storey et al.

size as a ﬁshery becomes depleted (see discussion in Welcomme, 1985). CPUE may not decline, but the gear targets smaller species, reﬂecting the loss of slower growing, low-fecundity large-bodied species, and the increase in faster growing, more fecund smaller-bodied species. The appearance of these small mesh nets could be as a result of a cultural transfer of methods from West Papuan refugees coming into the area. However, theory on the adoption of small mesh nets in subsistence/artisanal ﬁsheries is supported by the results of OTML ﬁsh catch monitoring data, which show signiﬁcant declines in ﬁsh catch at riverine sites, and now at some oxbow sites. It seems likely that villagers are speciﬁcally targeting smaller species out of necessity. Biological monitoring of the Fly River has intentionally targeted ﬁsh catch (abundance and biomass) at the request of the State of Papua New Guinea, to ensure a subsistence ﬁshery is maintained for communities downstream of the mine. Analysis of the resultant data (viz. changes in abundance, biomass, species richness, and ﬁsh assemblage composition) clearly show the impacts due to the mine. That these impacts are detectable partly reﬂects the effectiveness of the monitoring approach, but also that the impacts are very signiﬁcant, and so relatively easy to detect with simple, but robust sampling methods and data analyses. The monitoring program demonstrates the suitability of ﬁsh as indicators of ecosystem health, particularly in that they are relatively easy to sample and their mobility and longevity means they can ‘‘integrate’’ temporal and spatial effects at both the catchment scale and at the local scale (Harris, 1995; Harris and Silveira, 1999; Simon, 1999). In recent years monitoring programs for ﬁsh (and macroinvertebrates) are moving toward an Index of Biological Integrity (IBI) approach. Originally developed by Karr (1981), it is a multi-metric approach which uses a number of quantiﬁable ﬁsh metrics (parameters) to monitor changes in ﬁsh community structure and function. Metrics include species richness, abundance, number of exotic species, ﬁsh health, and trophic and habitat guild composition. The information gathered can be broadened by community participation to include data on recreational species (e.g., redﬁn perch, trout), ﬁsh response to short-term ﬂuctuations (spikes) in water quality parameters (e.g., ﬁsh kills), and anecdotal information on the historic condition of the rivers and their ﬁsh populations. Although detailed knowledge of the life history, habitat requirements, trophic position, and sensitivity to pollutants of all Fly River ﬁsh species has not been documented, future adoption of an IBI approach to monitoring ﬁsh assemblages in the Fly River may prove productive. In conclusion, over its 25 years of operation, the Ok Tedi Mine has signiﬁcantly changed the Fly River ﬁshery. Other pressures, such as commercial barramundi ﬁsheries, a growing population, and increased

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

459

artisanal ﬁshing with greater access to more efﬁcient nets and boats are also inﬂuencing the system. The pressures from exotic species currently in the system, and additional species that may yet be introduced (i.e., carp, tilapia) will further impact the ﬁsh fauna. Perhaps the greatest threat in the short term however, is the appearance of ARD along river bank levees, with associated mobilization of metals from mine-derived sediments with a high sulﬁde content (Bolton et al., 2009; Apte, 2009). ARD is of growing concern for riverine and ﬂoodplain ﬁsh resources, and will need to be managed carefully from now until after mine closure.

ACKNOWLEDGMENTS The authors thank all past and present staff of the Environment Department, Ok Tedi Mining Limited, who assisted in the collection of ﬁsh-catch data. Past coordinators of the biological monitoring programs at Ok Tedi are acknowledged for their role in directing the collection of data and inﬂuencing the project: David Balloch (1982–1985), Kent Hortle (1983–1987), Ross Smith (1988–1992), Andrew Storey (1993–1995), Stephen Swales (1996–1998), Charles Tenakanai (1998–2001), and Markson Yarrao (2001–present).

REFERENCES Allen, G. R. (1991). Field Guide to the Freshwater Fishes of New Guinea. Christensen Research Institute, Madang, PNG, Publication No. 9, 268 pp. An, K., & Choi, S. (2003). An assessment of aquatic ecosystem health in a temperate watershed using the index of biological integrity. Journal of Environmental Science & Health, Part A: Toxic/Hazardous Substances & Environmental Engineering, 38, 1115–1131. Apte, S. C. (2009). Biogeochemistry of copper in the Fly River. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 321–373. Belbin, L. (1995). PATN Pattern Analysis Package. Division of Wildlife & Ecology, CSIRO, Canberra, Australia. Bolton, B. R., Pile, J. L., & Kundapen, H. (2009). Texture, geochemistry and mineralogy of sediments of the Fly River System. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 51–112. Boyden, C. R., Brown, B. E., Lamb, K. P., Frucker, R. F., & Tuft, S. J. (1978). Trace elements in the Upper Fly River, Papua New Guinea. Freshwater Biology, 8, 189–205.

460

A. W. Storey et al.

Cao, Y., Larsen, D. P., & Hughes, R. M. (2001). Evaluating sampling sufﬁciency in ﬁsh assemblage surveys: a similarity-based approach. Canadian Journal of Fisheries and Aquatic Sciences, 58, 1782–1793. Coates, D. (1993). Fish ecology and management of the Sepik-Ramu, New Guinea, a large contemporary river basin. Environmental Biology of Fishes, 38, 345–368. DPI (1979). Fisheries Survey of the Ok Tedi Mining Region. In: Fisheries Research Annual Report. Fisheries Division Department of Primary Industry. DPI (1980). Ok Tedi Fisheries. In: Fisheries Research Annual Report. Fisheries Division Department of Primary Industry. Dudgeon, D., & Smith, R. E. W. (2006). Exotic species, ﬁsheries and conservation of freshwater biodiversity in tropical Asia: the case of the Sepik River, Papua New Guinea. Aquatic Conservation: Marine and Freshwater Ecosystems, 16, 203–215. Fausch, K. D., Lyons, J., Karr, J. R., & Angermeier, P. L. (1990). Fish communities as indicators of environmental degradation. American Fisheries Society Symposium, 8, 123–144. Gauch, H. G. (1982). Multivariate Analysis in Community Ecology. Cambridge University Press, New York, NY. Harris, J. H. (1995). The use of ﬁsh in ecological assessments. Australian Journal of Ecology, 20, 65–80. Harris, J. H., & Silveira, R. (1999). Large-scale assessments of river health using an index of Biotic Integrity with low-diversity ﬁsh communities. Freshwater Biology, 41, 235–252. Hortle, K. G. (1986). A Review of Biological Sampling of the Ok Tedi and Fly River systems, April 1983 to June 1986. OTML Report ENV86-9. Hughes, R. M., Kaufmann, P. R., Herlihy, A. A., Intelmann, S. S., Corbett, S. C., Arbogast, M. C., & Hjort, R. C. (2002). Electroﬁshing distance needed to estimate ﬁsh species richness in raftable Oregon rivers. North American Journal of Fisheries Management, 22, 1229–1240. Hugueny, B., Camara, S., Samoura, B., & Magassouba, M. (1996). Applying an index of biotic integrity based on communities in a West African river. Hydrobiologia, 331, 71–78. IUCN (2004). IUCN red list of threatened species. www.redlist.org. Karr, J. R. (1981). Assessment of biotic integrity using ﬁsh communities. Fisheries, 6, 21–27. Karr, J. R. (1991). Biological integrity: a long-neglected aspect of water resource management. Ecological Applications, 1, 66–84. Kennard, M. J., Pusey, B. J., Harch, B. D., Dopre, E., & Arthington, A. H. (2006). Estimating local stream ﬁsh assemblage attributes: sampling effort and efﬁciency at two spatial scales. Marine and Freshwater Research, 57, 635–653. Maunsell & Partners. (1982). Ok Tedi Environmental Study, Maunsell and Partners, Pty. Ltd. Oberdorff, T., & Hughes, R. M. (1992). Modiﬁcation of an index of biotic integrity based on ﬁsh assemblages to characterize rivers of the Seine Basin, France. Hydrobiologia, 228, 117–130.

Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine

461

OTML (1988). Sixth Supplemental Agreement Environmental Study, 1986–1988. Final Draft Report, Volume 1. Report prepared by Ok Tedi Mining Limited, November 1988. OTML (1990). APL Compliance and Additional Environmental Monitoring Program. Unpublished report to the State of Papua New Guinea by Ok Tedi Mining Limited. OTML (1993). Biology Annual Report, data collected to September 1992. OTML Report ENV 93-01. OTML (1994). Biology Annual Report, data collected to 31 September 1993. OTML Report ENV 94-10. OTML (1995). Biology Annual Report, data collected to September 1994. OTML Report ENV 95-03. OTML (1996). Biology Section Annual Report 1994–1995. OTML Report ENV 96-05. Pickup, G., & Cui, Y. (2009). Modeling the impact of tailings and waste rock disposal on the Fly River system. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 257–289. Reynolds, L., Herlihy, A. T., Kauffman, P. R., Gregory, S. V., & Hughes, R. M. (2003). Electroﬁshing effort requirements for assessing species richness and biotic integrity in western Oregon streams. North American Journal of Fisheries Management, 23, 450–461. Roberts, T. R. (1978). An ichthyological survey of the Fly River in Papua New Guinea with descriptions of new species. Smithsonian Contributions to Zoology, 281, 1–72. Robertson, C. H., & Baidam, G. (1983). Fishes of the Ok Tedi area with notes on ﬁve common species. Science in New Guinea, 10, 16–26. Simon, T. P. (1999). Assessing the Sustainability and Biological Integrity of Water Resources using Fish Communities. CRC Press, New York, NY. Smith, R. E. W., & Hortle, K. G. (1991). Assessment and predictions of the impacts of the Ok Tedi copper mine on ﬁsh catches in the Fly River system, Papua New Guinea. Environmental Monitoring and Assessment, 18, 41–68. Smith, R. E. W., & Morris, T. F. (1992). The impacts of changing geochemistry on the ﬁsh assemblages of the Lower Ok Tedi and Middle Fly River, Papua New Guinea. The Science of the Total Environment, 125, 321–344. Smith, R. E. W., & Bakowa, K. A. (1994). Utilisation of ﬂoodplain waterbodies by the ﬁshes of the Fly River, Papua New Guinea. Mittenbach International Vereinen Limnology, 24, 187–196. Smith, R. E. W., Ahsanullah, M., & Batley, G. E. (1990). Investigations of the impact of efﬂuent from the Ok Tedi copper mine on the ﬁsheries resource in the Fly River, Papua New Guinea. Environmental Monitoring and Assessment, 14, 315–331. Sorrenson, E. M. (1991). Metal Poisoning in Fish. CRC Press Inc, Boca Raton, 374 pp.

462

A. W. Storey et al.

Stauber, J. L. (1995). Toxicity testing using marine and freshwater unicellular algae. Australian Journal of Ecotoxicology, 1, 15–24. Storey, A. W. (1997). Multivariate analysis of temporal and spatial changes in the structure of ﬁsh communities in the Fly River. Unpublished report by Wetland Research and Management to Ok Tedi Mining Ltd. December 1997. Storey, A. W., Roderick, I. D., Smith, R. E. W., & Maie, A. Y. (2002). Spread of the introduced climbing perch (Anabas testudineus) in the Fly River System, Papua New Guinea, with comments on possible ecological effects. International Journal of Ecology and Environmental Sciences, 28, 103–114. Storey, A. W., & Yarrao, M. (2009). Development of aquatic food web models for the Fly River, Papua New Guinea, and their application in assessing impacts of the Ok Tedi mine. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 575–615. Storey, A. W., Marshall, A. R., & Yarrao, M. (2009). Effects of mine-derived river bed aggradation on ﬁsh habitat of the Fly River, Papua New Guinea. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 416–487. Swales, S., Storey, A. W., Roderick, I. D., Figa, B. S., Bakowa, K. A., & Tenakanai, C. D. (1998). Biological monitoring of the impacts of the Ok Tedi copper mine on ﬁsh populations in the Fly River system, Papua New Guinea. The Science of the Total Environment, 214, 99–111. Swales, S., Storey, A. W., Roderick, I. D., & Boga, S. F. (1999). Fishes of ﬂoodplain habitats of the Fly River system, Papua New Guinea, and changes associated with El Nin˜o droughts and algal blooms. Environmental Biology of Fishes, 54, 389–404. Swales, S., Storey, A. W., & Bakowa, K. A. (2000). Temporal and spatial variations in ﬁsh catches in the Fly River system in Papua New Guinea and the possible effects of the Ok Tedi copper mine. Environmental Fish Biology, 57, 75–95. Welcomme, R. L. (1985). River Fisheries. FAO Fisheries, Technical Paper No. 262, 330 pp. Wood, I. B., Day, G. M., Storey, A. W., & Markham, A. J. (1995). Environmental monitoring and research programs at the Ok Tedi copper mine. Proceedings of the 1994 PACOM conference, Townsville, Qld, Australia. WRM and Hydrobiology (2007). Fly River Freshwater Fish Diversity Survey – July 2005. Unpublished report prepared by Wetland Research & Management and Hydrobiology Pty Ltd for Ok Tedi Mining Limited. February 2007, 33 pages plus appendices. Zar, J. H. (1974). Biostatistical analysis. Englewood Cliffs, NJ: Prentice-Hall, 620 pp.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00413-8

Chapter 13

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River, Papua New Guinea Andrew W. Storey1,, Andrew R. Marshall2 and Markson Yarrao3 1

School of Animal Biology (M092), The University of Western Australia, Crawley, WA, 6009, Australia 2 Andrew Marshall & Associates Pty Ltd., 43 Warrangarree Drive, Woronora Heights. New South Wales 2233, Australia 3 Environment Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, W.P. Papua New Guinea

13.1. Introduction The Ok Tedi mine, located on Mount Fubilan in the remote Star Mountains of the Western Province of Papua New Guinea, is one of the largest producers of copper concentrate in the world (Smith and Hortle, 1991; Swales et al., 1999, 2000). The mine initially commenced operations in 1984 using a cyanide leachate circuit to recover gold. In 1987 the operation had a dual gold leachate/copper ﬂotation recovery process, and by 1988 the mine had changed to copper concentrate production, with gold and silver also recovered in the concentrate (Smith and Hortle, 1991). Waste disposal at the Ok Tedi Mine has been controversial. The initial intention during mine development was to store tailings in a dam close to the mine, however, high rainfall (B10 m per annum at the mine), geological instability (limestone rock), and regular seismic activity meant that any tailings dam had a high risk of failure at some point in the future. This was evident in 1984 when the foundations of a tailing dam under construction were buried by a landslide Corresponding author. Tel.: +(618) 6488 1482; Fax: +(618) 6488 1029;

E-mail: [email protected] (A.W. Storey).

464

A. W. Storey et al.

(Swales et al., 2000). Consequently, on-site waste storage was considered unfeasible. Rather than close the mine, the State of Papua New Guinea authorized disposal of waste rock and detoxiﬁed tailings directly to the Ok Tedi (‘ok’ in the local Min language means ‘river’), a major headwater tributary of the Fly River. Consequently, the mine has always operated using riverine disposal, and currently, approximately 120,000 metric tonnes per day (tpd) of waste rock and 80,000 tpd of tailings are discharged directly into the Ok Tedi and its tributaries to be transported downstream into the Fly River (Wood et al., 1995; Swales et al., 2000). Regular and extensive environmental monitoring of the Fly River system has been conducted by Ok Tedi Mining Limited to examine the effects of this waste disposal on the river system and surrounding environment (Wood et al., 1995; Apte, 2009; Bolton et al., 2009; Pickup and Cui, 2009). Changes in ﬁsh diversity, abundance, and biomass in the Ok Tedi and Fly River downstream of the mine have been documented since 1983 (Swales et al., 1998, 2000). Decreasing ﬁsh catch biomass in the Fly River had been noted since the late 1980s (Smith and Morris, 1992). These changes became statistically signiﬁcant declines in the early 1990s (Swales et al., 1998, 2000), with losses of between 65% and 96% relative to pre-mine levels in the Ok Tedi and middle Fly River channel, with the greatest reductions recorded from sites closest to the mine (Swales et al., 2000; Storey et al., 2009). There have also been losses of species diversity in the Ok Tedi and middle Fly River (Swales et al., 1998, 2000), and an extensive survey in mid-2005 determined that most species still occurred in tributaries or other parts of the system, although several species were not recoded and may have been lost (Storey and Smith, unpublished data; Storey et al., 2009). The reductions in ﬁsh biomass (and diversity) indicate signiﬁcant biological effects on the river system. There were initial concerns that observed effects were due to metal toxicity to biota, given the elevated concentrations of total copper in the Fly River downstream of the mine. However, the majority of metals are particulate-associated, and therefore not readily bioavailable (Smith et al., 1990; Smith and Hortle, 1991), and concentrations of dissolved copper are generally low (o15 mg/L) (Apte, 2009). Furthermore, the majority of the dissolved copper is non-labile due to the complexing capacity of the receiving waters, which are alkaline and high in dissolved organic carbon (Stauber, 1995; Stauber et al., 2009). This has been supported by a range of laboratory toxicity studies which have failed to identify acute toxic effects on Fly River species or equivalent species from northern Australia exposed to equivalent concentrations of copper currently found in the river channel (Smith et al., 1990; Smith and Hortle, 1991; Stauber, 1995; Swales et al., 1998, 2000).

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

465

The other major impacts in the river channel which could cause the observed declines in ﬁsh biomass were increased total suspended sediments (TSS) and river-bed aggradation resulting from the disposal of mine waste and tailings to the river (Smith and Hortle, 1991; Swales et al., 1998, 2000). Signiﬁcant bed aggradation was ﬁrst recorded in the middle Ok Tedi in the early 1990s, and it soon spread to the lower Ok Tedi and into the Fly River immediately below the conﬂuence with the Ok Tedi at D’Albertis Junction (Fig. 13.1). Aggradation has subsequently progressed further into the middle Fly River (Pickup and Cui, 2009). Currently there is approximately 6 m of aggradation in the middle Ok Tedi, decreasing to 4.5 m in the lower Ok Tedi and 4 m in the Middle Fly immediately downstream of D’Albertis Junction. Aggradation gradually reduces with distance from the conﬂuence, to the point where it is not detectable below Everill Junction (B220 nautical miles downstream) (Pickup and Cui, 2009) (Fig. 13.1). In the apparent absence of any ecotoxicity effect, bed aggradation and associated changes provide a mechanism for reductions in ﬁsh catch by impacting on ﬁsh habitat in the main channel. Although TSS is elevated, and is often reported as impacting on aquatic fauna, the ﬁsh fauna of the Fly River system is considered relatively tolerant of high TSS (Smith et al., 1990), as are ﬁsh assemblages of many turbid, lowland tropical rivers. The Strickland River, a major tributary of the Fly River (Fig. 13.1), naturally has a higher TSS concentration than the Fly, yet supports a healthy and diverse, although slightly different ﬁsh fauna (Smith et al., 1990). Therefore, aggradation and associated changes are thought likely of greater effect than the increased TSS. It was therefore hypothesized that declines in diversity and biomass of ﬁsh in the Fly River channel most likely related to loss of habitat associated with bed aggradation. The relationship between habitat diversity and ﬁsh diversity is well established in the literature, with many studies reporting signiﬁcant positive correlations (Gorman and Karr, 1978; Bishop and Forbes, 1991; Lobb and Orth, 1991; Cowx and Welcomme, 1998), with attributes such as water depth, substrate type, aquatic vegetation, and bank cover shown to be important. Similarly, reductions in ﬁsh diversity as a result of loss of habitat have also been reported (Hortle and Lake, 1983; Mann, 1988; Scarnecchia, 1988; Harvey and Nakamoto, 1997; Cowx and Welcomme, 1998), with straightening of channels, removal of in-stream structures such as pool-rifﬂe sequences and large woody debris, and reduction in depth shown to result in loss of ﬁsh diversity. It is reasonable therefore to suggest that the generalities of such relationships will also apply to ﬁsh in the Fly River channel, although speciﬁc habitat requirements of ﬁsh assemblages have yet to be determined. To make the link between declines in

466

A. W. Storey et al.

Figure 13.1: Fly River catchment, with inset of D’Albertis Junction and immediate area, showing sampling reaches.

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

467

ﬁsh catch and loss of habitat, it was ﬁrst necessary to establish if the mine was impacting on ﬁsh habitat. The aim of the current study was to assess the effects of bed aggradation on ﬁsh habitat, by measuring changes in parameters likely to constitute or reﬂect aspects of ﬁsh habitat.

13.2. Study Area The study was conducted in the main channel of the Fly River in the vicinity of D’Albertis Junction (Fig. 13.1). The Fly River in this area is a large, slowﬂowing, meandering lowland ﬂoodplain river. The channel is 150–200 m wide, 5–10 m deep (depending on stage height, but up to 20 m deep in places), with a raised levee along each bank. The channel is almost trapezoidal in cross-sectional shape, with relatively steep banks and a broad central bed of irregular depth. The channel form has long meander loops (2–3 km between bends), with eroding banks on the outside of bends and depositional point bars on the inside. Pre-mine, the banks in the lower Ok Tedi and upper middle Fly were lined by tall, overhanging primary rainforest, which was regularly ﬂooded for short periods (days to weeks). The vegetation holds the banks together, and provides trailing root mats, undercuts, and inputs of large woody debris to form snags in the channel. The forest also provides inputs of allochthonous carbon such as leaves, fruits, seeds, and insects to support riverine food webs (see Storey and Yarrao, 2009). Apart from localized areas of harder clays, sediments are predominantly silt/sand. The banks of the channel tend to be relatively uniform except for features locally referred to as ‘backwaters.’ These are small embayments or ‘scalloping,’ ranging from 10 to 50 m long and 5 to 10 m wide, characterized by relatively deepwater and back-eddies with a general upstream ﬂow direction. These ‘backwaters’ are distinct from the more classic shallow backwaters (shallow, off-channel areas with zero ﬂow, usually colonized by emergent and submerged macrophytes), which do not occur along the Fly River main channel, but are present on the ﬂoodplain. Access from the main channel onto the ﬂoodplain is either over the levee when river levels are high and the forest is ﬂooded, or into oxbow lakes (cut-off meander loops) or ﬂooded lagoons via ‘tie-channels’ (narrow channels connecting these features to the main channel). As noted above, aggradation in this area progressively decreases with distance downstream from the mine, from 4.5 to 5.0 m in the lower Ok Tedi to approximately 2 m at Nukumba (Fig. 13.1). Physical changes in the channel due to aggradation are obvious and include reduced depth across the

468

A. W. Storey et al.

channel, increased water velocity as a result of reduced depth, increased bank erosion as a result of increased velocity, inﬁlling of the ‘backwaters,’ smothering of snags (woody debris), reduction in bank angle, reduced ‘‘heterogeneity’’ of the channel bed and increased TSS. Inﬁlling of backwaters is particularly signiﬁcant, as these areas provide deepwater habitat out of the main ﬂow where larger ﬁsh tend to reside. These areas are targeted for gill netting as part of OTML monitoring, providing high diversity and biomass in ﬁsh catch (see Swales et al, 2000; Storey et al., 2009). Loss of these habitats results in loss of the associated catch which partially explains the observed reductions in biomass in monitoring data (see Storey et al., 2009). In-stream physical changes also cause changes on the ﬂoodplain. Bed aggradation results in increased over-bank ﬂooding (Pickup and Cui, 2009). During over-bank ﬂooding events sediment is deposited on levees and across the ﬂoodplain, whereby greater ﬂooding occurs under greater aggradation, giving a greater depth of sediment deposition. The combination of sedimentation and water logging on the ﬂoodplain then results in forest dieback, whereby vegetation intolerant of smothering/prolonged ﬂooding dies, and is gradually replaced by ﬂood-tolerant species (i.e., Phragmites karka) (Pickup and Cui, 2009). This results in the loss of forest along the banks of the channel.

13.3. Methods The study aimed to quantify aspects of ﬁsh habitat along impacted reaches of the lower Ok Tedi and upper Middle Fly and nonimpacted reaches upstream. Twenty reaches were selected to cover a range in aggradation from zero aggradation (six reaches), through low aggradation (1–2 m; four reaches), medium aggradation (2–3 m; six reaches) to high aggradation (4–5 m; four reaches) (Fig. 13.1). Subsequently, one reach (reach 15 at Adopted River Mile 431, immediately upstream of Nukumba) was dropped from the data set because it was too unstable, suffering signiﬁcant over-bank ﬂooding, a high rate of channel migration and high turbulence at the time of survey. Such conditions prevented accurate hydrographic mapping. Aggradation estimates for each remaining site were taken from the bed aggradation surveys performed by Ok Tedi Mining Limited in September 2002. As far as possible, surveys of each reach were standardized to allow comparisons across reaches. Each reach consisted of 1,000 m of channel as measured by GPS, with the reach commencing at the downstream end on the outside of a meander bend and continuing downstream. At each reach,

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

469

‘reach-level’ and ‘point-level’ parameters considered indicative of ﬁsh habitat were recorded (Appendix 1). Reach-level attributes were surveyed by taking total counts of selected parameters for the whole reach (i.e., counting the number of backwaters, number of snags, etc.), and point-level parameters were measured at 200 m intervals along each reach (i.e., 0, 200, 400, 600, 800, and 1,000 m), and then averaged for the reach. Point-level variables consisted either of the measurement of physical parameters (i.e., water depth, water velocity, or bank angle), or percentage cover of attributes estimated for a 10 m length of bank at the point (i.e., percentage cover of features such as riparian vegetation, trailing root mats, and undercuts). No previous studies assessing effects of river-bed aggradation on the ﬁsh fauna of a large lowland tropical river system have been reported. Therefore a methodology was developed by selecting habitat parameters previously shown to inﬂuence ﬁsh diversity such as water depth, water velocity, substrate composition, snags, aquatic vegetation, and cover such as tree roots and bank undercuts (Sheldon, 1968; Gorman and Karr, 1978; Bishop and Harland, 1982; Schlosser, 1982; Moyle and Vondracek, 1985; Grossman et al., 1987a, b; Bain et al., 1988; Bishop and Forbes, 1991; Rabeni and Jacobson, 1993; Sheldon and Meffe, 1995; Cowx and Welcomme, 1998). The amount of algal growth on submerged surfaces was assessed for each reach, with growth ranked as 0 for no observable growth, 1 for present, 2 for common, and 3 for abundant algal cover. The amount of cover available for ﬁsh is an important habitat attribute, providing areas for resting, spawning, and predator avoidance. As an indication of cover available along the edge of the channel, the percentage of different types of cover along the bank was recorded as: riparian canopy, trailing vegetation (vines/creepers/branches), trailing vegetation (P. karka; a colonizing plant characteristic of disturbance/ forest dieback), trailing root mats, moss/algae on sediments, undercuts, and woody debris. Cover was recorded at each measurement point and then averaged to give a percent cover for the whole reach. Backwaters were categorized by size for each reach: small (o5 m long), medium (5–10 m long), large (10–30 m long), and very large (W30 m long). The number in each size category of backwater was then counted for each reach. An area of shoreline was classiﬁed as a backwater if there was visible reversal in ﬂow direction relative to the adjacent open channel. Total number of backwaters was calculated by summing the number of backwaters in each size class. The percent of the reach occupied by backwaters was calculated by multiplying the number of each size class of backwater by the approximate median length of each size class (5, 7.5, 15, and 40 m, respectively), and expressing the total number of meters of backwater as a percentage of the length of the reach.

470

A. W. Storey et al.

Extent of mine-derived sediment deposition along the linear length of the shoreline and the depth of sediment deposition on the ﬂoodplain adjacent to the channel was also recorded for each reach. The number of minor projections and irregularities along the shoreline (i.e., features smaller than backwaters), referred to as nodes, was also measured. A large number of projections provides high heterogeneity, and as a result would likely indicate greater diversity in habitat and near-shore ﬂows, as opposed to few projections indicative of a uniform bank with low habitat diversity. A likely effect of bed aggradation is modiﬁcation of the channel proﬁle, from a trapezoidal form with deepwater close to the shore, to one with more gradual banks and a shallow-angle providing greater area of shallow water closer to the shore. Such changes would likely affect different species in different ways. The angle (901 for vertical and 01 for horizontal) and linear height of the bank above the water line was recorded at each point measurement. To assess changes in the water column, the relationship between depth and velocity was also characterized at each point measurement. Water depth and velocity were recorded at 1, 4, and 10 m distance from the bank, with velocity measured at approximately 0.6 total depth at each position. This spacing represented velocity and depth conditions in areas close to the bank where ﬁsh predominantly reside (Storey, unpublished data). For each distance from the bank (1, 4, and 10 m), the variance for depth and velocity were also calculated for the reach to reﬂect level of variability in these parameters. Depth and velocity were then integrated for each zone (0–1, 1–4, 4–10, and 0–10 m from the bank). Using water depth at each distance from the bank, the wetted cross-sectional area was calculated, and this value was then divided by water velocity so that deepwater with low velocities resulted in a high score, and as water depth decreased or velocity increased, then the score decreased. Not all relevant habitat parameters could be surveyed from the water surface, therefore an EchoTrac MkII dual frequency echo sounder with a Side Scan Sonar transducer was used to gather additional information. It was hypothesized that smothering by bed aggradation would likely result in a decline in the number and size of snags over a reach, therefore snag abundance and size was determined using a standardized, repeatable qualitative approach using the Side Scan Sonar. A longitudinal proﬁle of each reach was derived by surveying a transect parallel to the bank, with the nadir point of the pass approximating the thalweg (e.g., deepest point) of the channel. Snags and obstructions in the water column between the scanner and the bank were recorded on a paper plot and snags categorized as ‘‘simple,’’ ‘‘medium,’’ or ‘‘complex’’ based on the size and strength of

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

471

the image. Small snags (o0.1 m diameter) and ﬂoating debris (residing only in the upper 0.5 m of the water column) were not recorded, and large rafts of ﬂoating logs supported by only a single log anchored into the bed of the river were recorded as a single, simple snag. The number of snags in each category along each reach was calculated, as was the total number of snags by summing all simple and medium snags and adding to the count of complex snags. A weighting of 7.5 was applied to the count of complex snags to allow for the greater areal extent of complex snags (and the likely greater role they play in providing habitat). This weighting was determined by measuring the spatial two-dimensional coverage of complex snags and comparing it to a single large snag, giving a conversion factor of between 5 through to 15, but with a mean of approximately 7.5. All data were converted to either measurements or counts for each reach, and then analyzed initially using Spearman Rank Correlation of each variable against aggradation. Linear regression was then applied, using reaches as replicates, aggradation as the independent variable and with individual habitat parameters as the dependent variables to determine if a cause-effect relationship could be established between aggradation and habitat parameters. Principal Components Analysis (PCA) in the PRIMER package (Clarke and Gorley, 2001; Clarke and Warwick, 2001) was then used to reduce the 55 measured and derived habitat variables to summary variables as principal components. The principal components were then regressed against aggradation to determine whether they gave a better relationship than individual ﬁsh habitat parameters.

13.4. Results Channel surveys were conducted over a 10-day period in November 2002. In all, 55 variables were measured or derived (Appendix 1). Spearman Rank Correlation indicated that in many instances there were very strong relationships between habitat variables and aggradation, with the majority showing an inverse relationship to aggradation (Appendix 1). Linear regression indicated that the most statistically signiﬁcant relationships explained between 75% and 93% of variation in the individual habitat parameters (Fig.13.2). For some parameters, stronger relationships were achieved using log10(xþ1) transformed data for aggradation, indicating that a lower threshold had been reached in the habitat parameter, and the rate of decline in these parameters decreased with increasing levels of aggradation.

472

A. W. Storey et al.

Figure 13.2: Plots of linear regressions of river-bed aggradation (log10(xþ1)) against habitat parameters demonstrating a high rank correlation coefﬁcient (Appendix 1), giving regression equations and R2 values. Analyses showed many highly signiﬁcant interdependencies in the data set, indicating that many of the habitat parameters were closely related, including habitat parameters measured by independent methods (i.e., visual assessment versus hydrographic survey). This was to be expected given the nature of the data. For example, ﬁsh cover and cover of root mats, trailing

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

473

10 Y = 6.7261 - 3.3194 x Aggradation PCA Factor 1

5

Adj R2 =0.9336

0

-5

-10 0

1

2

3

4

5

River Bed Aggradation (m)

Figure 13.3: Linear regression of PCA Factor 1 against river-bed aggradation, giving regression equation, R2 value, and 95% CI about the regression line. vegetation, undercuts, riparian canopy, and woody debris all reﬂect the state of the riparian vegetation, loss of which occurs due to aggradation and overbank ﬂooding. PCA reduced the data set to ﬁve principal components, with 58% of the variation in the data explained by PC1 and PC2. Habitat parameters contributing to PC1 included percent ﬁsh cover per reach, percent of bank comprised of root mats, percent of bank covered by forest canopy, percent of bank comprised of undercuts, percent of bank covered by trailing forest vegetation, total number of snags, number of snags weighted by water velocity (large and total), snags weighted by velocity/depth (simple, large, and total), percent of reach occupied by backwaters, and the number of backwaters (large and total). Habitat parameters contributing to PC2 included the angle of the upper bank; water depth at 4 and 10 m distance from the waters edge; extent of erosion; and, hydrographic measures of bed heterogeneity such as the total number of nodes per meter of reach. Only PC1 gave a signiﬁcant relationship when regressed against aggradation (po0.05), however, this analysis indicated that aggradation explained 93% of the variation in PC1 (Fig. 13.3).

13.5. Discussion The original hypothesis of this study was that river-bed aggradation resulted in the loss of ﬁsh habitat which in turn caused the observed declines in ﬁsh

474

A. W. Storey et al.

biomass and diversity in the Fly River channel. Field measurements showed that many of the parameters chosen to be indicative of ﬁsh habitat availability declined with increasing bed aggradation. These included numbers of medium, large, and very large backwaters, numbers of medium and complex snags, total number of snags, bank heterogeneity, algal cover on ﬂoating debris, cover of woody debris, riparian canopy, roots, undercuts and trailing vegetation, overall ﬁsh cover, and depth weighted by velocity at 0–1 and 1–4 m zones from the bank. Changes in habitat heterogeneity associated with river-bed aggradation were supported by observations during the ﬁeld study, whereby sites affected by aggradation appeared more uniform, with fewer trailing root mats, undercuts, trailing riparian vegetation, canopy cover, woody debris, and more uniform banks with fewer backwaters (Figs. 13.4 and 13.5). Inﬁlling of backwaters due to siltation of the banks was common along the lower Ok Tedi and middle Fly downstream of Kuambit (Fig. 13.6). This represents a loss of deep, low-velocity habitat utilized by many of the larger, deep-bodied species. Routine ﬁsh catch monitoring at Kuambit during the 1980s and 1990s targeted individual backwaters with speciﬁc large mesh gill nets because these backwaters repeatedly produced good catches. These backwaters are now in-ﬁlled and no longer contain ﬁsh, partly explaining the declines in ﬁsh catch at this

Figure 13.4: Section of bank on the Fly River between Kiunga and D’Albertis Junction with no aggradation, showing high diversity of ﬁsh habitat: abundant riparian canopy cover, trailing riparian vegetation, large woody debris, in-stream snag habitat, trailing root mats, and undercuts.

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

475

Figure 13.5: Section of bank on the Fly River downstream of D’Albertis Junction with B4 m of aggradation, showing low diversity of ﬁsh habitat: reduced riparian canopy cover, minimal trailing riparian vegetation, no large woody debris or snag habitat, and no trailing root mats or undercuts. Note open canopy and colonizing grasses on upper slopes.

Figure 13.6: Sections of river bank on the lower Ok Tedi and middle Fly River downstream of D’Albertis Junction, showing examples of backwaters in-ﬁlled with deposited mine waste.

476

A. W. Storey et al.

location. The importance of backwaters to ﬁshes cannot be overstated, with studies by Humphries et al. (1999), Rosenberger and Angermeier (2003), King (2004), and Feyrer et al. (2005) all highlighting the importance of backwater habitat to ﬁsh. Independent hydrographic surveys during this study indicated that snags were also adversely affected by aggradation. In slower ﬂowing reaches with high aggradation, there were few snags observed in locations where snags normally would be expected. It is likely that snags in these locations were totally inundated by sediment. Some reaches with high aggradation appeared to have a greater loading of woody debris (i.e., lower reaches of the Ok Tedi), likely as a result of forest die off and subsequent erosion of the dead trees into the river. However, surveys indicated that this material either had little structure (i.e., consisted of single branches/trunks), provided little cover (i.e., was lying ﬂat on the river bed), or was located in mid-channel in high to very high water velocity and as such would likely provided little cover for ﬁsh. Snags and woody debris provide an important function in rivers, they improve the quality of food and habitat resources available to ﬁsh (Angermeier and Karr, 1984; Reeves et al., 1993; Lehtinen et al., 1997), and also provide foraging sites (Van Den Avyle and Petering, 1988), spawning substrates (Van Den Avyle and Petering, 1988), protection from ﬂow (Todd and Rabeni, 1989), and shelter from predators (Angermeier and Karr 1984; Johnson et al., 1988). A number of studies on small river systems have established that snags inﬂuence hydraulic processes, affecting water depth, water velocity, and substrate composition (Angermeier and Karr, 1984; Reeves et al., 1993). Removal of snags has also been shown to be detrimental, as reported by Angermeier and Karr (1984), where the removal of woody debris from a small Illinois stream led to a reduction in water depth and incidence of benthic organic matter, and a concomitant increase in water velocity and proportion of sand bottom. Lehtinen et al. (1997) studied the use of snags by ﬁshes within the upper Mississippi River and found species assemblages were signiﬁcantly different between sites with snag and control sites, with average ﬁsh biomass greater at sites with snags. Similarly, Cowx and Welcomme (1998) noted that ﬁsh density declined eight-fold when large woody debris was removed from a lowland river in the United Kingdom. As noted previously, bed aggradation causes the channel to become shallower, and as a consequence water velocities increase as the system attempts to convey the same volume of water in a smaller channel. This study found greater velocities in the near-bank habitats compared with areas with no aggradation. Increased velocity is unlikely to be conducive for ﬁsh species adapted to deep, slow-ﬂowing water as more typical of the lowland reaches of the Fly River. Velocity previously has been shown to inﬂuence biomass,

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

477

richness, mean ﬁsh size, and density of ﬁshes (Moyle and Vondracek, 1985; Sheldon and Meffe, 1995). Aggradation also results in over-bank ﬂooding under high ﬂows, when the channel is no longer able to convey the same ﬂows. Riparian forest vegetation is lost as it becomes inundated by water and sediment. Therefore, habitats and conditions dependent on healthy riparian vegetation (i.e., extensive cover of root mats, trailing vegetation, undercuts, woody debris, shade from canopy, etc.) are lost. Vegetation also tends to stabilize banks, and as a result of dieback, there is less root structure and so greater bank instability. These effects are particularly evident on the outsides of bends in reaches with high aggradation, where the banks are actively eroding, with sections continually collapsing (Fig. 13.7). Although the banks may provide areas that physically resemble backwaters, these features tend to be transient, lasting hours to days before collapsing, especially under high channel migration rates. These backwaters also tend to have high water velocities, and combined with the unstable, eroding nature of the banks, they are likely unsuitable for ﬁsh. In contrast, reaches with no aggradation have healthy riparian vegetation, which promotes stable meander bends which support backwaters that are permanent features lasting months to years. A major effect of bed aggradation is a reduction in water depth (with a concomitant increase in velocity, as discussed above). Studies have repeatedly shown depth to be an important parameter in deﬁning ﬁsh communities. Often ﬁsh have evolved to exploit speciﬁc habitats deﬁned by depth (Gorman and Karr, 1978). Examples of such evolutionary biology include highly adapted body form and mouth structure, as well as vertical segregation of sympatric species in pools based on morphological and behavioral adaptations (Gorman and Karr, 1978). Similar specialization was

Figure 13.7: Examples of severe bank instability in the middle Fly River below D’Albertis Junction, showing effects of active erosion, regular overbank ﬂooding, extensive forest loss, and high rates of channel migration.

478

A. W. Storey et al.

noted by Welcomme (1985), whereby tropical stream ﬁshes in African and Asian streams separated vertically, based on morphology and feeding habit. A reduction in depth will force species to alter their optimum habitat utilization, and cause the co-occurrence of species which are normally segregated, and ultimately lead to the loss/exclusion of species (Welcomme, 1985). This was observed by Harvey and Nakamoto (1997) who found that decreased water depth resulting from aggradation led to a reduction in largesized ﬁsh. Unfortunately little is known of the habitat preferences or utilization by ﬁsh of the Fly River. Some insights are provided from dietary data (see Storey and Yarrao, 2009), but there is insufﬁcient knowledge at this time to infer how changes in speciﬁc habitat parameters would adversely affect speciﬁc species of ﬁsh. As well as the direct physical effects of aggradation on ﬁsh habitat, there are also less obvious, indirect effects on ‘habitat suitability’ that are more difﬁcult to quantify. They include aspects such as loss of spawning habitat, reduction in food resources, and disruption to food webs (i.e., reduced aquatic invertebrate diversity, loss of terrestrial insects and fruits through forest dieback, and loss of in-stream algal production due to the reduced photic zone and smothering). For example, species of ﬁsh which are obligate frugivores, dependent on fruits falling into the river from terrestrial vegetation are no longer recorded from river reaches affected by forest dieback (i.e., the arid catﬁsh Arius latirostris), and this likely reﬂects loss of this food resource. Similarly, species dependent on inputs of terrestrial insects are much reduced, as are visual feeders that are affected by elevated TSS. Recently it has been shown that the contribution of carbon (energy) derived from in-stream algal sources is reduced downstream of the mine (Storey and Yarrao, 2009). Whether the effect is due to aggradation, habitat loss, smothering of surfaces, reduced photic zone, or toxicity from metals is currently being resolved, but this disruption to the food web is likely to have inﬂuenced higher consumers directly or indirectly dependent on these energy sources. Therefore, mine-related inﬂuences on ﬁsh assemblages in the Fly River are likely numerous, but also likely change with proximity to the mine. However, physical changes related to aggradation appear to be considerable in the Ok Tedi and upper Middle Fly River. Assessment of the quality and quantity of a speciﬁc habitat in an instance in time, although an important means of considering faunal habitat, may be inadequate and misleading if other spatial or temporal habitat requirements are ignored. Maintenance of habitat required by all life stages is critical to the survival of a species, and an inadvertent loss of a key habitat(s) could result in a loss of species diversity. Cowx and Welcomme (1998) commented that a ﬁsh seldom spends its entire life in the same habitat, and usually

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

479

requires a different habitat with suitable microhabitat conditions for each speciﬁc life stage. They further noted that ﬁsh in rivers depend on undamaged interactive pathways along four dimensions: longitudinal, lateral, vertical, and temporal. The longitudinal pathway encompasses upstream and/or downstream migration routes for spawning/recruitment. These migrations may be over tens or hundreds of meters, or many hundreds or even thousands of kilometers, encompassing movement between headwaters, estuaries, and open ocean. The lateral pathway comprises the lateral movement to backwaters, vegetated margins, and ﬂooded oxbow lakes, lagoons, and ﬂoodplain in search of spawning, feeding, rearing/nursery, and predator avoidance areas. Lateral movement also occurs to avoid high ﬂows, particularly for young-of-the-year ﬁsh with poor swimming abilities. The vertical pathway includes riverine/groundwater interactions for species spawning in benthic substrates whereby oxygenated water must penetrate to a sufﬁcient depth to ensure survival of buried eggs. This involves interactions with sediment particle size and water velocity. The temporal pathway refers to seasonal changes in the availability of different microhabitats and the availability of the appropriate microhabitat at the appropriate time of year/life history stage. Fish in the Fly River will also depend on habitat in different dimensions (sensu Cowx and Welcomme, 1998), with species migrating from headwaters and mid-reaches to the ocean, migrations occurring at different times of the year, and lateral migrations onto and off the extensive ﬂoodplain. The current study however, did not consider the broader context of habitat usage, or attempt to look at habitat requirements for different life stages, but concentrated on describing differences in habitat in one area at one instant in time using as broad a range of habitat descriptors as possible. Even so, the parameters used and the design adopted showed signiﬁcant impacts on ﬁsh habitat that would affect utilization by ﬁsh. However, it is possible the mine affects other aspects of ﬁsh habitat in the Fly River which may also contribute to the observed declines in ﬁsh catch. As noted above, the authors were unable to source literature describing studies that assessed ﬁsh habitat of large, lowland tropical rivers, and therefore had to develop a speciﬁc approach for the Fly River. The selection of parameters used in this study was based on habitat parameters previously shown in the literature to inﬂuence ﬁsh diversity. Cowx and Welcomme (1998) noted that depth, substrate, current, and presence of coarse woody debris in lowland rivers were important in providing ﬁsh cover. Sheldon and Meffe (1995) found depth and velocity were the primary correlates of biomass, richness, mean ﬁsh size, and density in temperate rivers in North America. Bishop and Forbes (1991) and Bishop and Harland (1982)

480

A. W. Storey et al.

correlated species diversity to habitat diversity measured in four dimensions; water depth, substrate type, aquatic vegetation, and bank cover. Gorman and Karr (1978) and Schlosser (1982) assessed habitat diversity using water depth, water velocity, substrate type and cover, such as tree roots and bank undercuts. Bain et al. (1988) used measures of water depth, current velocity, substrate coarseness, and substrate heterogeneity (measures of diversity in particle size) to assess habitat diversity. Sheldon (1968) and Moyle and Vondracek (1985) also determined depth and velocity as important habitat variables affecting ﬁshes. Grossman et al. (1987a, b) noted that ﬁshes responded to depth, velocity, substrate composition, and cover, and Rabeni and Jacobson (1993) found that ﬁsh were affected by depth, current velocity, substrate particle size, and cover characteristics. Although not all parameters were considered relevant to the lowland reaches of the Fly River, the selected suite of parameters showed a strong response to aggradation, and likely reﬂects a change in ﬁsh habitat. As noted, the current study demonstrated a strong relationship between aggradation and habitat; however, the effect of reduced habitat availability on ﬁsh diversity for the Fly River channel has not been demonstrated. Fish catches at monitoring sites in the Fly River are inherently very variable, being inﬂuenced by many factors, known and unknown. For example, river stage and whether rising, falling, or steady is known to affect the catching efﬁciency of gill nets (Storey, unpublished data), the standard gear used in ﬁsh catch monitoring (Storey et al., 2009). Similarly, variability in the interannual recruitment of barramundi (Lates calcarifer Bloch) from coastal nursery areas is also known to inﬂuence the biomass taken by gill nets, and seasonal effects (wet versus dry season), as well as less predictable climatic effects (El Nin˜o/La Nin˜a events) also affect riverine catches (Swales et al., 1999, 2000). Although monitoring data show signiﬁcant declines in ﬁsh catch compared with pre-mine baseline data, it has taken many years (i.e., 1983 to B1992) before these changes became statistically signiﬁcant declines using regression analysis. This not only in part reﬂects sampling frequency, but also variability in catches. This is shown for the Ok Tedi at Atkamba (TED30) and the Fly River at Kuambit (FLY10) (Fig. 13.8), where declines in ﬁsh biomass are approximately 88 and 79%, respectively. Given the temporal variability in these data, and the logistics of taking replicate samples from each reach, it was not considered feasible to statistically demonstrate a relationship between aggradation/habitat availability and ﬁsh diversity in the current study. The relationship between habitat diversity and ﬁsh diversity in rivers, however is well established in the literature, with many studies reporting signiﬁcant positive correlations (Gorman and Karr, 1978; Bishop and

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

481

Figure 13.8: Changes in ﬁsh catch biomass (in kilograms) over time at (top) Atkamba (TED30) on the lower Ok Tedi and (bottom) Kuambit (FLY10) on the middle Fly, giving regression equation, R2 value, and 95% CI about the regression line. Forbes, 1991; Lobb and Orth, 1991; Cowx and Welcomme, 1998). For example, Lobb and Orth (1991) noted that structurally complex habitats inﬂuence the assemblage structure of ﬁshes from a large stream in West Virginia, with total ﬁsh densities highest in complex habitats such as snags, backwaters, edge pool, and edge rifﬂes. Bishop and Forbes (1991) documented a strong positive relationship between habitat and ﬁsh diversity in tropical rivers in northern Australia. Cowx and Welcomme (1998) similarly reported that in lowland sections of rivers, the potential capacity of a reach to support a rich ﬁsh assemblage was dependent upon habitat complexity. Reductions in ﬁsh diversity as a result of loss of habitat have also been reported (Hortle and Lake, 1983; Mann, 1988; Scarnecchia, 1988; Cowx and Welcomme, 1998). Channelization (the straightening of streams and removal of in-stream structures such as pool-rifﬂe sequences and large

482

A. W. Storey et al.

woody debris), in particular, has been shown to be detrimental to ﬁsh communities (Hortle and Lake, 1983; Scarnecchia, 1988; Cowx and Welcomme, 1998). Cowx and Welcomme (1998) reported a 60% reduction in ﬁsh diversity in channelized river sections when compared to natural conditions. Similarly, Hortle and Lake (1983) recorded decreased diversity from channelized reaches of a river in Victorian, Australia and concluded that loss of habitat was responsible. Given the consistency and strength of the relationships between habitat and ﬁsh diversity, there is no reason to believe that neither the relationship, nor the variables playing a role, as widely reported in the literature are substantially different from those in the Fly River. Therefore, the authors argue it may be validly concluded that the reduction and/or loss of features considered important to ﬁshes and indicative of habitat complexity, such as backwaters, woody debris, snags, undercuts, and root mats are likely responsible for the observed decline in ﬁsh diversity and biomass in the Fly River system. These results support the contention that aggradation as a result of the discharge of waste rock into the river is impacting ﬁsh habitat, and this is likely a major cause of declines in ﬁsh catch in the Ok Tedi and upper Middle Fly. Visually dramatic changes to the channel support this argument, as does the absence of acute toxicity effects from dissolved metals in laboratorybased toxicity tests (Smith et al., 1990; Smith and Hortle, 1991; Smith and Morris, 1992; Swales et al., 1998). Because these impacts to ﬁsh habitat are the result of physical processes, it may be possible to predict future effects on ﬁsh habitat, and possibly therefore ﬁsh diversity by modeling sediment delivery, sediment transport, and associated aggradation/degradation of the river bed. A considerable effort has been expended over the years by Ok Tedi Mining Limited to model sediment transport in the system (Pickup and Cui, 2009), with the aim of not only predicting longitudinal and lateral dispersal of waste material through the river, but also to predict future aggradation and erosion in the system. However, sediment transport modeling tends to work at the reach scale and was not designed to predict effects at the near-bank zone, which is where most ﬁsh tend to reside in the channel. Therefore, it is currently not possible to model whether ﬁsh habitat will recover post mine-closure. Even so, in general terms, once a reach passes the depositional phase and the channel stabilizes, and in some areas begins to scour, it is likely that sedimentary structures and snags will begin to redevelop (Pickup, personal communication). Channel scouring and incision has been observed in the upper Ok Tedi above Tabubil following the Vancouver Ridge landslide. This natural event in the upper catchment in 1989 delivered equivalent to 3 years of mine waste production into the river in one instant. Within a matter of months the channel in the affected

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

483

tributary (Sulphide Creek) and the upper Ok Tedi began (and continues) to down-cut through the deposited material, leaving benches that are still visible today. This reﬂects the higher energy of the river in the upper catchment, with greater capacity to recover channel form as soon as the delivery of waste to the system ceases. Current sediment transport modeling suggests however, that waste material, once deposited in the lower Ok Tedi and middle Fly River channel will reside there for many years (Pickup, personal communication), and it is quite possible that backwaters and snags will not scour of deposited sediment. This reﬂects the lower energy of this lowland, meandering river. In reality, it may not be possible to predict future changes in ﬁsh habitat in the lower Ok Tedi and middle Fly with any certainty. It is proposed that monitoring of the system post mine-closure will continue for several decades, and only then will we see how ﬁsh habitat and ﬁsh diversity recover in the channel of this heavily impacted tropical river.

ACKNOWLEDGMENTS The authors thank all past and present staff of the Environment Department, Ok Tedi Mining Limited who assisted in the collection of survey data, in particular Philip Atio and Tenong Dremdap in the ﬁeld; and Lawrence Pimi, Andy Ikuavia, and Nelson Opa for logistical support. Constructive comments from two anonymous referees are acknowledged, and Geoff Pickup is thanked for discussion on sediment transport modeling.

REFERENCES Angermeier, P. L., & Karr, J. R. (1984). Relationships between woody debris and ﬁsh habitat in a small warmwater stream. Transactions of the American Fisheries Society, 113, 716–726. Apte S. C. (2009). Biogeochemistry of Copper in the Fly River. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 321–373. Bain, M. B., Finn, J. T., & Booke, H. E. (1988). Streamﬂow regulation and ﬁsh community structure. Ecology, 69, 382–392. Bishop, K. A. & Harland, W. G. (1982). Further ecological studies on the freshwater ﬁshes of the Alligator Rivers region (ﬁnal report). Open File No. OFR34, Supervising Scientist for the Alligator Rivers Region. Bishop, K. A., & Forbes, M. A. (1991). The freshwater ﬁshes of northern Australia. In: C. D. Haynes, M. G. Ridpath, & M. A. J. Williams (Eds). Monsoonal Australia, Landscape, Ecology and Man in the Northern Lowlands. A. A. Balkema, Rotterdam & Brookﬁeld, pp. 79–107.

484

A. W. Storey et al.

Bolton, B. R., Pile, J. L., & Kundapen, H. (2009). Texture, geochemistry, and mineralogy of sediments of the Fly River System. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 51–112. Clarke, K. R., & Gorley, R. N. (2001). Primer v5: User Manual/Tutorial. Primer E: Plymouth, Plymouth Marine Laboratory, Plymouth, UK. Clarke, K. R., & Warwick, R. M. (2001). Changes in Marine Communities: An Approach to Statistical Analysis and Interpretation (2nd ed.). Primer E: Plymouth, Plymouth Marine Laboratory, Plymouth, UK. Cowx, I. G., & Welcomme, R. L. (1998). Rehabilitation of Rivers for Fish. FAO; Fishing News Books, Oxford, UK; Malden, MA. Feyrer, F., Sommer, T. R., & Baxter, R. D. (2005). Spatial-temporal distribution and habitat associations of age-0 splittail in the lower San Francisco estuary watershed. Copeia, 1, 159–168. Gorman, O. T., & Karr, J. R. (1978). Habitat structure and stream ﬁsh communities. Ecology, 59, 507–515. Grossman, G. D., de Sosta, A., Freeman, M. C., & Lobon-Cervia, J. (1987a). Microhabitat use in a Mediterranean riverine ﬁsh assemblage: Fishes of the upper Matarran˜a. Oecologia, 73, 501–512. Grossman, G. D., de Sosta, A., Freeman, M. C., & Lobon-Cervia, J. (1987b). Microhabitat use in a Mediterranean riverine ﬁsh assemblage: Fishes of the lower Matarran˜a. Oecologia, 73, 490–500. Harvey, B. C., & Nakamoto, R. J. (1997). Habitat-dependent interactions between two size classes of juvenile steelhead in a small stream. Canadian Journal of Fisheries and Aquatic Sciences, 54, 27–31. Hortle, K. G., & Lake, P. S. (1983). Fish of channelized and unchannelized sections of the Bunyip River, Victoria. Australian Journal of Marine and Freshwater Research, 34, 441–450. Humphries, P., King, A. J., & Koehn, J. D. (1999). Fish, ﬂows and ﬂoodplains: links between freshwater ﬁshes and their environment in the Murray-Darling River system, Australia. Environmental Biology of Fishes, 56, 129–151. Johnson, D. L., Beaumier, R. A., & Lynch, W. E., Jr. (1988). Selection of habitat structure interstice size by bluegills and largemouth bass in ponds. Transactions of the American Fisheries Society, 117, 171–179. King, A. J. (2004). Ontogenetic patterns of habitat use by ﬁshes within the main channel of an Australian ﬂoodplain river. Journal of Fish Biology, 654, 1582–1603. Lehtinen, R. M., Mundahl, N. D., & Madejczyk, J. C. (1997). Autumn use of woody snags by ﬁshes in backwater and channel border habitats of a large river. Environmental Biology of Fishes, 49, 7–19. Lobb, M. D., & Orth, D. J. (1991). Habitat use by an assemblage of ﬁsh in a large warmwater stream. Transactions of the American Fisheries Society, 120, 65–78. Mann, R. H. K. (1988). Fish and ﬁsheries of regulated rivers in the UK. Regulated Rivers: Research and Management, 2, 411–424.

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

485

Moyle, P. B., & Vondracek, B. (1985). Persistence and structure of the ﬁsh assemblage in a small Californian stream. Ecology, 66, 1–13. Pickup, G., & Cui, Y. (2009). Modeling the Impact of Tailings and Waste Rock Disposal on the Fly River System. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 257–289. Rabeni, C. F., & Jacobson, R. B. (1993). The importance of ﬂuvial hydraulics to ﬁsh-habitat restoration in low-gradient alluvial streams. Freshwater Biology, 29, 211–220. Reeves, G. H., Everest, F. H., & Sedell, J. R. (1993). Diversity of juvenile anadromous salmonid assemblages in coastal Oregon basins with different levels of timber harvest. Transactions of the American Fisheries Society, 122, 309–317. Rosenberger, A., & Angermeier, P. L. (2003). Ontogenetic shifts in habitat use by the endangered Roanoke logperch (Percina rex). Freshwater Biology, 48, 1563–1577. Scarnecchia, D. L. (1988). The importance of streamlining in inﬂuencing ﬁsh community structure in channelized and unchannelized reaches of a prairie stream. Regulated Rivers: Research & Management, 2, 155–166. Schlosser, I. J. (1982). Fish community structure and function along two habitat gradients in a headwater stream. Ecological Monographs, 52, 395–414. Sheldon, A. L. (1968). Species diversity and longitudinal succession in stream ﬁshes. Ecology, 49, 193–198. Sheldon, A. L., & Meffe, G. K. (1995). Path analysis of collective properties and habitat relationships of ﬁsh assemblages in coastal plain streams. Canadian Journal of Fisheries and Aquatic Sciences, 52, 22–33. Smith, R. E. W., & Hortle, K. G. (1991). Assessment and prediction of the impacts of the Ok Tedi copper mine on ﬁsh catches in the Fly River system, Papua New Guinea. Environmental Monitoring and Assessment, 18, 41–68. Smith, R. E. W., & Morris, T. F. (1992). The impacts of changing geochemistry on the ﬁsh assemblages of the Lower Ok Tedi and Middle Fly River, Papua New Guinea. The Science of the Total Environment, 125, 321–344. Smith, R. E. W., Ahsanullah, M., & Batley, G. E. (1990). Investigations of the impact of efﬂuent from the Ok Tedi copper mine on the ﬁsheries resource in the Fly River, Papua New Guinea. Environmental Monitoring and Assessment, 14, 315–331. Stauber, J. L. (1995). Toxicity testing using marine and freshwater unicellular algae. Australian Journal of Ecotoxicology, 1, 15–24. Stauber, J. L., Apte, S. C., & Rogers, N. J. (2009). Species, Bioavailability, and Toxicity of Copper in the Fly River System. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 375–408. Storey, A. W., & Yarrao, M. (2009). Development of Aquatic Food Web Models for the Fly River, Papua New Guinea, and their Application in Assessing Impacts of the Ok Tedi Mine. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 575–615.

486

A. W. Storey et al.

Storey A. W., Yarrao, M., Tenakani, C., Figa, B., & Lynas, J. (2009). Use of Changes in Fish Assemblages in the Fly River System, Papua New Guinea, to Assess Effects of the Ok Tedi Copper Mine. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System. Elsevier, Amsterdam, Vol. 9, 427–462. Swales, S., Storey, A. W., Roderick, I. D., Figa, B. S., Bakowa, K. A., & Tenakanai, C. D. (1998). Biological monitoring of the impacts of the Ok Tedi copper mine on ﬁsh populations in the Fly River system, Papua New Guinea. The Science of the Total Environment, 214, 99–111. Swales, S., Storey, A. W., Roderick, I. D., & Boga, S. F. (1999). Fishes of ﬂoodplain habitats of the Fly River system, Papua New Guinea, and changes associated with El Nin˜o droughts and algal blooms. Environmental Biology of Fishes, 54, 389–404. Swales, S., Storey, A. W., & Bakowa, K. A. (2000). Temporal and spatial variations in ﬁsh catches in the Fly River system in Papua New Guinea and the possible effects of the Ok Tedi copper mine. Environmental Fish Biology, 57, 75–95. Todd, B. L., & Rabeni, C. F. (1989). Movement and habitat use by stream-dwelling smallmouth bass. Transactions of the American Fisheries Society, 118, 229–242. Van Den Avyle, M. J., & Petering, R. W. (1988). Inundated timber as nursery habitat for larval gizzard and threadﬁn shad in a new pumped storage reservoir. Transactions of the American Fisheries Society, 117, 84–89. Welcomme, R. L. (1985). Distribution of species in river systems. River Fisheries. FAO Fisheries Technical Paper, 262, 102–113. Wood, I. B., Day, G. M., Storey, A. W., & Markham, A. J. (1995). Environmental monitoring and research programs at the Ok Tedi copper mine. In: Proceedings of the 1994 PACOM Conference. Townsville, Queensland, Australia.

Appendix 1: Measured and Derived Habitat Attributes Taken at Each Site to Provide an Indication of Fish Habitat Condition, Listing Parameter, Abbreviation, Units and Type (Reach or Point), with Correlation Coefﬁcient and Signiﬁcance Level for Spearman Rank Correlation against Aggradation Habitat parameter

Secchi depth No. of small backwaters (o5 m long) No. of medium backwaters (5–10 m long)

Abbreviation

Reach/ point

Units

Spearman coefﬁcient

p-value

Secchi Smallback

Reach Reach

m Count

0.699 0.255

0.0009 ns

Medback

Reach

Count

0.848

o0.0001

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

487

Appendix 1: (Continued ) Habitat parameter

No. of large backwaters (10–30 m long) No. of very large backwaters (W30 m long) Total no. of backwaters Percent of reach occupied by backwaters No. of simple snags No. of medium snags No. of complex snags Total no. of snags Ease of sediment penetration Bank heterogeneity (no. of ‘‘nodes’’) Algae cover on debris % of reach with sediment deposition along edge Angle of upper bank Linear height of upper bank Angle of lower bank Linear height of lower bank Percent of bank covered by forest canopy Percent of bank covered by trailing forest vegetation Percent of bank covered by trailing Phragmites Percent of bank covered by all trailing vegetation

Abbreviation

Reach/ point

Units

Spearman coefﬁcient

p-value

Largeback

Reach

Count

0.853

o0.0001

VLargeback

Reach

Count

0.705

0.0008

Totalback

Reach

Count

0.891

o0.0001

Percentbacks

Reach

Count

0.824

o0.0001

Simpsnag Medsnag Compsnag Totsnag Sedpen

Reach Reach Reach Reach Point

Count Count Count Count cm

0.500 0.780 0.906 0.917 0.063

0.0292 o0.0001 o0.0001 o0.0001 ns

No_knobs

Reach

Count

0.798

o0.0001

Algae Sed_deposit

Reach Reach

Rank Percent

0.449 0.836

ns o0.0001

Upper_ang Upper_lgth

Point Point

Degrees m

0.613 0.276

0.0053 ns

Lower_ang Lower_lght

Point Point

Degrees m

0.463 0.526

0.0459 0.0207

Canopy

Point

%

0.832

o0.0001

Trailveg

Point

%

0.903

o0.0001

Trailphrag

Point

%

0.763

Trailall

Point

%

0.388

0.0001

ns

488

A. W. Storey et al.

Appendix 1: (Continued ) Habitat parameter

Percent of bank comprised of root mats Percent of bank covered by moss and algal growths Depth of mine sediment deposited on upper banks Percent of bank covered by woody debris/snags Percent of bank comprised of undercuts Water velocity at 1 m distance from waters edge Water depth at 1 m distance from waters edge Water velocity at 4 m distance from waters edge Water depth at 4 m distance from waters edge Water velocity at 10 m distance from waters edge Water depth at 10 m distance from waters edge Mean water velocity for reach Percent ﬁsh cover per reach Depth velocity for zone 0–1 m from bank Depth velocity for zone 1–4 m from bank

Abbreviation

Reach/ point

Units

Spearman coefﬁcient

p-value

Roots

Point

%

0.917

o0.0001

Moss

Point

%

0.167

ns

Sedontop

Point

m

0.864

o0.0001

Snag on bank

Point

%

0.796

o0.0001

Undercuts

Point

%

0.883

o0.0001

Vep1m

Point

cm/sec

0.834

o0.0001

Dep1m

Point

cm

0.397

ns

Vel4m

Point

cm/sec

0.784

o0.0001

Dep4m

Point

cm

0.079

ns

Vel10m

Point

cm/sec

0.580

0.0092

Dep10m

Point

cm

Meanvel

Reach

cm/sec

Fishcover

Reach

%

0.913

o0.0001

DxV0-1

Point

Count

0.811

o0.0001

DxV1-4

Point

Count

0.816

o0.0001

0.143

0.699

ns

0.0009

Effects of Mine-Derived River Bed Aggradation on Fish Habitat of the Fly River

489

Appendix 1: (Continued ) Habitat parameter

Depth velocity for zone 4–10 m from bank Total weighted usable area Score for simple snags weighted by DXV Score for medium snags weighted by DXV Score for large snags weighted by DXV Score for total number of snags weighted by DXV No. of simple snags weighted by water velocity No. of medium snags weighted by water velocity No. of large snags weighted by water velocity Total no. of snags weighted by water velocity Variance of water velocity measurements at 1 m Variance of water velocity measurements at 4 m Variance of water velocity measurements at 10 m Mean variance in water velocity for reach Variance of water depth measurements at 1 m

Abbreviation

Reach/ point

Units

Spearman coefﬁcient

p-value

DxV4-10

Point

Count

0.655

0.0024

Total DxV

Point

Count

0.816

o0.0001

DxV simpsnag

Reach

Count

0.814

o0.0001

DxV medsnag

Derived

Count

0.858

o0.0001

DxV larsnag

Derived

Count

0.919

o0.0001

DxV totsnag

Derived

Count

0.942

o0.0001

Mvel simpsnag

Derived

Count

0.786

o0.0001

Mvel medsnag

Derived

Count

0.845

o0.0001

Mvel larsnag

Derived

Count

0.913

o0.0001

Mvel totsnag

Derived

Count

0.920

o0.0001

Var-vel1m

Point

cm/sec

0.411

ns

Var-vel4m

Point

cm/sec

0.544

0.0161

Var-vel10m

Point

cm/sec

0.554

0.0138

Var-meanvel

Point

cm/sec

0.801

o0.0001

Var-dpth1m

Point

cm

0.417

ns

490

A. W. Storey et al.

Appendix 1: (Continued ) Habitat parameter

Variance of water depth measurements at 4 m Variance of water depth measurements at 10 m

Abbreviation

Reach/ point

Units

Spearman coefﬁcient

p-value

Var-dpth4m

Point

cm

0.089

ns

Var-dpth10m

Point

cm

0.199

ns

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00414-X

Chapter 14

Insects of the Fly River System Max S. Moulds Entomology Section, Australian Museum, Sydney, New South Wales 2010, Australia

14.1. Introduction The Fly River basin encompasses much of the southwest corner of Papua New Guinea. The catchment is high (above 3,000 m) in the central mountain chain. The lower Fly River meanders across a giant ﬂood plain for more than 800 km (500 miles) over which distance it drops just 70 m. The ﬂood plain incorporates Lake Murray, Papua New Guinea’s largest lake. The region remains substantially isolated; there is still no road access, and until the establishment of the Ok Tedi Mine in the early 1970s, access was limited to a few bush airstrips or river navigation. The mountains of the catchment include some of the wettest places on Earth. The Ok Tedi Mine (2,000 m) averages over 7,800 mm annually. At higher elevations rainfall exceeds 10 m. Moss forest covers much of the higher mountain slopes, giving way to rainforest on the lower slopes and across much of the lowlands. Elsewhere, the lowlands are swampy grasslands or open woodland. The insect fauna of the Fly River and its tributaries is very rich; a conservative estimate is that it is likely to exceed 20,000 species. This diversity is a direct result of the considerable diversity of habitat that includes lowland swamps, open forest, rainforest, moss forest, and mountain grassland. The richest areas appear to be the rainforests of the foothills and lower mountains to around 1,600 m. These are areas rich in plant diversity, an essential criterion for insect diversity.

Corresponding author. Tel.: +61 07-40938359;

E-mail: [email protected] (M.S. Moulds).

494

M. S. Moulds

Collecting of insects from the Fly River and its tributaries has a relatively long history, although a somewhat irregular one. However, few in-depth studies of these insects have been made and even today comparatively little is known about them compared to those of just about any other region of New Guinea. This, of course, reﬂects the inaccessibility of the region, with access having been restricted to boating up the river until air services became available in relatively recent years. The poor documentation of the fauna, with much of the literature very old and scattered, has precluded an in-depth account of the insect fauna here. This chapter attempts to provide an overview of what is known of the Fly River insects, highlighting notable species and providing key references. The ﬁrst insects to be collected from the Fly River were taken during Luigi D’Albertis’ voyages of exploration up the Fly River, ﬁrst with the Congregational missionary and amateur naturalist Samuel Macfarlane in 1875 and then in two famous trips far up the river in the steam launch Neva during 1876–1877 (D’Albertis, 1877, 1880). D’Albertis collected some 80,000 insects from New Guinea, for the Civico Museo di Genova, many of which were taken during his exploits along the Fly River. In 1885, the Geographical Society of Australasia sent the steam launch Bonito to the Fly and Strickland Rivers, on which Walter Froggatt was assistant zoologist and entomologist. Froggatt was then 27 but already an experienced and enthusiastic entomologist, and he assembled an extensive collection of Fly River insects that were sent to the Australian Museum, Sydney. During 1889–1890, Charles Kowald, a patrol ofﬁcer with the then British New Guinea Protectorate (later to become Papua), collected butterﬂies on Kiwai Island at the mouth of the Fly River (Tryon, 1890). Thereafter interest in the Fly River waned, although collecting of plants and animals, including insects, continued in other more accessible parts of New Guinea. It was not until the Second Archbold Expedition to New Guinea in 1936–1937 that the natural history of the Fly River again attracted attention. Like the other six Archbold Expeditions, this was a well-organized ambitious undertaking by the American Museum of Natural History and funded and led by Richard Archbold, an oil millionaire and associate of the museum. The Second Expedition surveyed the Fly River and parts of the Palmer and Strickland tributaries on which G. H. H. Tate collected mammals but also some 2,000 insects, later studied by F. E. Lutz (Rand and Brass, 1940). The intense entomological interest in Papua New Guinea during World War II and post–World War II years largely passed by the Fly River basin. W. W. Brandt, however, spent 4 months in the Kiunga area, including the Eliptamin Valley, during 1957 collecting insects for the Bishop Museum in

Insects of the Fly River System

495

Honolulu and the Australian Department of External Territories (Upton, 1997). Brandt had previously worked for Sir Edward Hallstrom collecting butterﬂies, until Hallstrom donated his collection to the Australian National Insect Collection, Canberra, in 1955 (Parsons, 1999). E. N. Marks, a research scientist with the Queensland Government’s Mosquito Control Committee, collected mosquitoes at Kiunga in February 1958 supported by funding from the Bishop Museum. The Museum Leiden Netherlands New Guinea Expedition collected near Sibil (1,260 m) in the Star Mountains in July 1959. L. W. Quate and S. Quate collected in the Star Mountains in 1961 and H. W. Clissold collected around the lower Fly River basin in 1964, both as part of the ambitious Bishop Museum entomological survey. D. K. McAlpine from the Australian Museum collected Diptera and some other insects around Lake Murray in 1963. Sir Alan Mann, chief justice of the then Territory of Papua New Guinea, had a broad interest in insects and collected brieﬂy at Kiunga in April 1963. J. Sedlacek and M. Sedlacek also collected around Kiunga for the Bishop Museum in August 1969. K. H. L. Key and J. Balderson collected insects (mainly Orthoptera) for the Australian National Insect Collection on the lower Fly River in August 1970. The establishment of the Ok Tedi Mine in the 1970s provided the ﬁrst opportunities for accessing many of the rich mountain habitats on the upper Fly River and its tributaries, especially around the Ok Tedi itself, and the Mine’s town of Tabubil. P. Imlay and Owen McCaw, a geologist, collected insects for the Australian Museum during the early years of mine construction in the 1970s; McCaw died in a helicopter accident soon after he began collecting in 1970. Roger Grund collected butterﬂies during a brief visit to Lake Murray in 1973 working with an oil drilling operation. D. A. Polhemus and J. T. Polhemus, entomologists from the USA, collected aquatic Heteroptera between Kiunga and Tabubil in September 1983. Since the 1980s, the Environment staff of Ok Tedi Mining Limited also has collected aquatic insects as part of ongoing monitoring of the Fly River for assessing any impact of mining operations. In particular, B. R. Bolton collected a large variety of insects for the Australian Museum. Y. Roisin and M. Leponce, both from the Universite´ Libre de Bruxelles, Belgium, visited the lower Fly River to collect termites and termite inquilines in March 1989, returning again in May/June 1990 to collect around Tabubil, Lake Murray, and the Nomad River. R. B. Lachlan, a schoolteacher, amateur entomologist, and associate of the Australian Museum, worked at the Tabubil International School from 1991 to 1993. He made extensive collections of butterﬂies, hawk moths, beetles, and cicadas, and also other insects during these years and again during return visits in 1994 and 2000. He did extensive light collecting at Tabubil and on some mountain peaks accessible by

496

M. S. Moulds

helicopter from the mine. Lachlan’s collections remain the most comprehensive so far made from the upper Fly River region. M. S. Moulds, S. A. Cowan, and M. A. Humphrey, all associated with the Australian Museum, also visited the Tabubil area during 1993 and 1996, collecting mainly cicadas and hawk moths but also some other insects and spiders. Collections of mosquitoes were made by G. J. T. Schuurkamp during the 1980s and 1990s while conducting malaria studies, both around the Tabubil and along the Fly River itself. In recent years, entomologists from the Australian Quarantine and Inspection Service have collected insects from the lower Fly River as part of an ongoing monitoring program for insect pests.

14.2. Beetles (Coleoptera) The only extensive studies of the Fly River beetles remain those resulting from the early expeditions. Macleay (1886) described many of the beetles collected by Froggatt during the Geographical Society’s expedition of 1885. The collection numbered 295 species and 914 specimens. Macleay notes the sparsity of beetles compared to Australia, especially of ground beetles (Carabidae) and blossom-feeding Buprestidae, Cetoniinae, and Scarabaeidae generally. The best-represented families were the Curculionidae (50 species), Brentidae (9 species), Anthribidae (6 species), Cerambycidae (48 species), and Chrysomelidae (64 species). Among the 40 new taxa that Macleay described were the aptly named genus Stricklandia (Carabidae) and the water beetle Enhydris froggatti. Pascoe (1886) described 98 species of Curculionidae beetles from D’Albertis’ voyages. Many other D’Albertis beetles from other families have been described by other authors including Jacoby (1883, 1885, 1886), Cande`ze (1880), Fauvel (1877, 1878a, b, 1879), and Reitter (1880). The jewel beetle Cyphogastra albertisi [Fig. 14.1(12)] and the leaf beetle Solenia albertisi are two of several beetle species named in honor of D’Albertis. Revisions of the water beetle families Dytiscidae and Hydrophilidae have been published by Balke (1995), Gentili (2002), Hebauer (2000), and Shaverdo et al. (2005), and include records of species from the Fly River basin. Bourguignon and Roisin (2006) have recently described three staphylinid beetles from the tribe Pseudoperinthini that are inquilines in termite nests; they have another paper in preparation describing further staphylinid termitophiles from the tribe Trichopseniini. In addition to the works of Fauvel, H. R. Last has also described the Fly River staphylinids (e.g., Last, 1980, 1987). Gressitt (1959) reviewed the New Guinea Cerambycidae, recording 49 species from the Fly River, 15 of which were

Figure 14.1: Coleoptera (localities of specimens in square brackets): (1) Rhinoscapha richteri (Faust.), Curculionidae [Tabubil, 650 m]; (2) Eupholus schoenherri (Gue´rinMe´neville), Curculionidae [Tabubil]; (3) Rosenbergia vetusta (Ritsema), Cerambycidae [Tabubil]; (4) Xylotrupus ulysses (Gue´rin-Me´neville), Scarabaeidae [Tabubil]; (5) Lomaptera fasciata (Moser), female, Cetoniinae, Scarabaeidae [Tabubil]; (6) Eupholus schoenherri (Gue´rin-Me´neville), Curculionidae [Tabubil]; (7) Micterophallus sp., Certoniinae, Scarabaeidae [Tabubil]; (8) Sphingnotus mirabilis (Boisduval), Cerambycidae [Tabubil]; (9) Xixuthrus microcerus (White), Cerambycidae [Tabubil]; (10) Eupatorus beccarii (Gestro), Scarabaeidae [Tabubil]; (11) Ischiopsopha hyla (Heller), Cetoniinae, Scarabaeidae [Tabubil]; (12) Cyphogastra albertisi, Buprestidae [Tabubil]; (13) Megaphonia adolphinae (Lansberge), Cetoniinae, Scarabaeidae [Tabubil]; (14) Cyphogastra sp., Buprestidae [Tabubil]; (15) Batocera wallacei (Pascoe), Cerambycidae [Tabubil]; (16) Prosopocoilus bison (Olivier), Lucanidae [Tabubil].

498

M. S. Moulds

described as new. Other papers that mention the Fly River beetles include those of Baehr (1996) (Carabidae: Odacanthinae), Chuˆjoˆ and Chuˆjoˆ (1987) (Erotytidae), Gressitt (1971) (Chrysomelidae), Howden (1989) (Scarabaeidae: Geotrupinae), Riedel (2002) (Attelabidae), Setliff (2007) (Curculionoidea), Stebnicka (1998) (Scarabaeoidea), Thompson (1996) (Curculionidae: Zygopinae), and Uhmann (1995) (Anthicidae). There are many more papers mentioning Fly River beetles but they are too numerous to list here. Beetles collected by R. B. Lachlan around Tabubil (600–2,000 m) during his residency and subsequent return trips between 1991 and 2000 (Fig. 14.1) numbered more than 50 species. These included several spectacular species such as the very large cerambycids Batocera wallacei [Fig. 14.1(15)] and Rosenbergia vetusta [Fig. 14.1(3)], the tri-horned dynastids Scapanes australis and Eupatorus beccari [Fig. 14.1(10)], the brilliant blue weevil Eupholus schoenherri [Fig. 14.1(2, 6)], and seven lucanid species of which the largest was Cyclommatus ﬁnschi. None of these species are endemic to the Fly River basin, but are widespread through New Guinea rainforests. Lachlan found beetles to be generally uncommon compared to the rainforests of tropical Australia. The same groups that Macleay noted as poorly represented in the lowlands were also found to be uncommon in the mountains. Ground and litter beetles were scarce, probably because of the exceptionally high rainfall that saturates the ground and washes away litter. Blossom-feeding beetles were also uncommon, although it is difﬁcult to draw sound conclusions because the rainforest blossom is largely inaccessible for collecting. In the lower foothills and ﬂoodplains of the Fly River where sago palm is a staple diet, the larvae of the sago-palm weevil, Rhynchophorus bilineatus (Curculionidae), are regularly eaten, as they are throughout much of Papua New Guinea. The larvae are collected when the sago is harvested and are a by-product of the harvest.

14.3. Butterﬂies and Moths (Lepidoptera) The Lepidoptera are perhaps the best-collected order of insects from the Fly River. Oberthu¨r (1880) published on D’Albertis’ Somerset and New Guinea butterﬂies but Fly River records are not clearly distinguished. Meyrick (1886) documented the 25 moth species collected by Froggatt during the Geographical Society Expedition. The butterﬂies from this expedition were studied by Tryon (1890), together with the butterﬂies collected from Kiwai Island at the mouth of the Fly River by Charles Kowald; in total 33 species were recorded, including 2 described as new although these are now considered

Insects of the Fly River System

499

junior synonyms. The extensive collection of butterﬂies from Kiunga (60 m) assembled by W. W. Brandt in 1957 was studied by Michael Parsons when compiling his comprehensive text of Papua New Guinea butterﬂies (Parsons, 1999). However, the detail in Parsons’ book precludes extraction of most Fly River records, so an analysis of the species Brandt collected would have to be based on an examination of specimens in collections. Roger Grund (2007) collected 27 species of butterﬂies during a brief visit to lowland rainforest near Lake Murray. All were common widespread species. R. B. Lachlan had a keen interest in butterﬂies, and during his 3 years of residency at Tabubil (600 m) he assembled a large collection of over 2,030 specimens representing more than 220 species (Lachlan, personal communication). However, these included few lowland species as he made only two short visits to lower altitudes at Kiunga and Matkomrae (50 km NNW of Kiunga). He is currently preparing a manuscript documenting these species in detail as well as Brandt’s records from Kiunga. The number of species Lachlan collected compares more than favorably with the 250 species recorded from a comprehensive survey of the nearby, and somewhat similar, Mimika subdistrict over the border in West Papua where butterﬂies were extensively collected from both lowlands and mountains (Gotts and Pangemanan, 2001). Lachlan was particularly keen on Delias [Fig. 14.2(9)] and collected 36 species, mostly on remote mountain creeks above Tabubil, which he accessed by helicopter courtesy of the Ok Tedi Mine (Lachlan, 1999a, 2000b). Unusually large numbers of species (in some cases as many as 15–20) were encountered on single creeks (Lachlan, personal communication), and 4 new species and 6 new subspecies were discovered, all of which are endemic to the area. Other notable butterﬂy species from the Fly River basin include the birdwings Troides oblongomaculatus, Ornithoptera priamus, O. goliath [Fig. 14.2(2)], O. chimaera, and the rare O. meridionalis. All these spectacular species are widespread through much of New Guinea, although O. meridionalis tends to be localized. The larvae feed on Aristolochia and Pararistolochia vines that for the most part grow in rainforest. The female of O. goliath is one of the two largest butterﬂies in the world, surpassed only by the female of O. alexandrae. Lachlan also assembled a comprehensive collection of over 2,250 hawk moths (Sphingidae) documented by Moulds and Lachlan (1998), Lachlan (1999b, 2000a), and Lachlan and Kitching (2001). Sixty-six species were recorded from Tabubil and the mountains beyond, including the ﬁrst New Guinea records of the death’s head hawk moth Acherontia lachesis [Fig. 14.3(8)] and seven other species, plus the ﬁrst Papua New Guinea records for seven others. The 66 species from this region compare more than favorably with the 80 known from the whole of New Guinea and the 68

Figure 14.2: Lepidoptera (localities of specimens in square brackets): (1) Hypochrysops plotinus (Grose-Smith), Lycaenidae [Tabubil, 650 m]; (2) Ornithoptera goliath goliath (Oberthu¨r), Papilionidae [Tabubil]; (3) Graphium weiskei (Ribbe), Papilionidae [Tabubil]; (4) Artipe grandis grandis (Rothschild & Jordan), Lycaenidae [Mt. Akrik ¼ Mt. Ian, 1,625 m]; (5) Libythea geoffroy maenia (Fruhstorfer), Libytheidae [Tabubil]; (6) Arhopala leo (Druce), Lycaenidae [Tabubil]; (7) Arhopala axina (Evans), Lycaenidae [Tabubil]; (8) Dichorragia ninus distinctus (Ro¨ber), Nymphalidae [Tabubil]; (9) Delias mysis lara (Boisduval), Pieridae [Tabubil]; (10) Polyura jupiter jupiter (Butler), Nymphalidae [Tabubil]; (11) Taenaris bioculatus (Gue´rin-Me´neville), Nymphalidae [Tabubil]; (12) Cethosia cydippe damasippe (Felder), Nymphalidae [Tabubil]; (13) Argyreus hyperbius niugini (Samson), Nymphalidae [Tifalmin, 1,300 m].

Figure 14.3: Lepidoptera (localities of specimens in square brackets): (1) Aenetus sp., Hepialidae [Tabubil, 650 m]; (2) Haemabasis calodesma (Rothschild & Jordan), Catocalinae, Noctuidae [Tabubil]; (3) Milionia sp., Ennominae, Geometridae [Tabubil]; (4) Ovimilionia ovata (Rothschild & Jordan), Ennominae, Geometridae [Tabubil]; (5) Callhistia dohertyi (Rothschild), Ennominae, Geometridae [Tabubil]; (6) Milionia sp., Ennominae, Geometridae [Tabubil]; (7) Milionia isodoxa (Prout), Ennominae, Geometridae [Tabubil]; (8) Acherontia lachesis (Fabricius), Sphingidae [Tabubil]; (9) Phyllodes imperialis (Druce), Noctuidae [Tabubil]; (10) Coscinocera hercules (Miskin), Saturniidae [Tabubil]; (11) Pterocyclophora huntei (Warren), Noctuidae [Tabubil]; (12) Lyssa toxopeusi (Altena), Uraniidae [Tabubil]; (13) Eupanacra pulchella (Rothschild & Jordan), Sphingidae [Mt. Akrik, 1,625 m]; (14) Othreis fullonia (Clerck), Noctuidae [Tabubil]; (15) Neodiphthera venusta (Rothschild & Jordan), Saturniidae [Tabubil]; (16) Alcides agathyrsus (Kirsch), Uraniidae [Tabubil]; (17) Macroglossum stevensi (Clark), Sphingidae [Mt. Akrik ¼ Mt. Ian, 1,625 m]; (18) Syntherata sp., Saturniidae [Tabubil].

502

M. S. Moulds

known from Australia. Like most other parts of New Guinea, the Fly River is particularly rich in moths but other families have not been studied. Among the other moths, one notable species that is known to occur in the lower mountains is Coscinocera hercules (Saturniidae) [Fig. 14.3(10)], the largest New Guinea moth, but two other species of Coscinocera also occur in the area (Lachlan, personal communication). Other large and spectacular moths common in the mountains are the tailed Lyssa spp. (Uraniidae). Some beautifully colored moths found around Tabubil by R. B. Lachlan include Haemabasis calodesma (Noctuidae) [Fig. 14.3(2)], Ovimilionia ovata (Geometridae) [Fig. 14.3(4)], and several species of Milionia (Geometridae) [Fig. 14.3 (3, 6, 7)]. The only endemic Lepidoptera known from the Fly River basin are the four species and eight subspecies of Delias butterﬂies and the hawk moths, Macroglossum mouldsi and Altijuba oktediensis, both known only from Tabubil. The eight hawk moth species recorded only from Tabubil by Moulds and Lachlan have since been found elsewhere in Papua New Guinea.

14.4. Flies and Mosquitoes (Diptera) Little is known of the true ﬂies (non-Nematocera). D. K. McAlpine collected near Lake Murray, concentrating on acalyptrates. Diptera were also collected by D’Albertis and Froggatt during their early voyages up the Fly River. W. W. Brandt also collected Diptera during his time at Kiunga. McAlpine published a revision of the genus Achias (McAlpine, 1994), a group of platystomatids or signal ﬂies of which many of the males have stalked eyes. Notable among them are the Fly River endemics, A. polyonychus, A. wallacei, and the aptly named A. ﬂyensis, all known from just a few specimens taken at Kiunga by Brandt. A. lachlani is known from a single specimen collected by R. B. Lachlan near Mt. Akrik in 1992. Other Fly River Diptera have been described as part of broader revisions, for example, Clements (2000) who described the large robber ﬂy Brachyrhopala grandis also collected by Brandt at Kiunga; James (1971) who reviews the Calliphorinae; and Mackerras (1964) who revised the Tabanidae. An interesting mosquito, Toxorhynchites nepenthicola, was described by Steffan and Evenhuis (1982) and is known only from Olsobip, near Kiunga, from larvae reared by M. Sedlacek from the pitchers of Nepenthes. Mosquito-borne diseases are rife in the Fly River basin, particularly malaria and elephantiasis (ﬁlariasis). The ﬁrst studies of these diseases in the region were undertaken by R. Knight, G. J. T. Schuurkamp, and others during the

Insects of the Fly River System

503

early years of development of the Ok Tedi Mine (e.g., Knight et al., 1979; Schuurkamp, 1987, 1992; Schuurkamp et al., 1987). An overview of the mosquitoes of Western Province has been provided by Cooper et al. (1997).

14.5. Cicadas and Other Hemiptera Comprehensive collections of cicadas from around Tabubil and Matkomrae (near Kiunga) have been made by M. S. Moulds and R. B. Lachlan. Thirtythree species were collected but several remain undescribed. Some of this material, and some other collected by W. W. Brandt and J. Sedlacek and M. Sedlacek, has been studied by Arnold de Boer (e.g., de Boer, 1992, 1993, 1994a, b, 1995a, b, 1999). Baeturia bicolorata [Fig. 14.4(3)] is the common dusk-singing cicada in Tabubil. The large and beautiful green-mottled Cosmopsaltria mimica and similar, but more brown colored, C. aurata [Fig. 14.4(9)] are common in the moss forests above Tabubil as is C. gigantea [Fig. 14.4(6)], the largest New Guinea cicada species. These three species are widespread through the mountains of New Guinea (Duffels, 1983) and their songs dominate at dusk. The pale green bladder cicada Thaumastopsaltria globosa (=T. glauca) is restricted to the lower Fly River and tropical northeastern Australia (Moulds, 1990; de Boer, 1992). This species sings only at dusk with a call somewhat resembling radio static. The very small leaf green Guineapsaltria ﬂava is widespread from the Fly River mouth to the foothills and also occurs in tropical northeastern Australia (Moulds, 1990; de Boer, 1993). Diceropyga subapicalis [Fig. 14.4(8)] is a common day-singing cicada around Tabubil with records also from the lower Fly River. It is a widespread species across the southern half of New Guinea and the northern tip of Australia (Duffels, 1977; Moulds, 1990). There is not a single cicada species known to be endemic to the Fly River basin. Cicadas, like most New Guinea insects, are not conﬁned to river basins but are instead distributed according to altitude along the mountains of the central cordillera. de Boer and Duffels (1996) summarize much of the biogeography of the New Guinea cicadas. Some aquatic Hemiptera (Aphelocheiridae, Gerridae, Hydrometridae, Mesoveliidae, Naucoridae, and Veliidae) have been studied by D. A. Polhemus and J. T. Polhemus and I. Lansbury (Polhemus and Polhemus, 1986, 1989, 1993, 1994, 1997, 2000, 2001a, b; Polhemus and Lansbury, 1997). They recorded 17 species, 9 of which were described as new including the aptly named marsh treader (Hydrometridae) Hydrometra kiunga and the water striders (Gerridae) Ptilomera kiunga and Cilometra kiunga, and the very large Limnometra grallator. While most of the species have extensive distributions

Figure 14.4: Orthoptera, Phasmatodea, and Hemiptera: Cicadidae (localities of specimens in square brackets). (1) Extatosoma popa (Sta˚l), female, Phasmatodea, Phasmatidae [Tabubil, 650 m]; (2) Baeturia pigrami (Boer), Cicadidae [Mt. Akrik ¼ Mt. Ian, 1,625 m]; (3) Baeturia bicolorata (Distant), Cicadidae [Tabubil]; (4) Siliquofera grandis (Blanchard), Orthoptera, Tettigoniidae [Tabubil]; (5) Cosmopsaltria mimica (Distant), Hemiptera, Cicadidae [Tabubil]; (6) Cosmopsaltria gigantea (Distant), Hemiptera, Cicadidae [Tabubil]; (7) Phyllium (Pulchriphyllium) ?schultzei (Giglio-Tos), Phasmatodea, Phylliidae [Tabubil]; (8) Diceropyga subapicalis (Walker), Hemiptera, Cicadidae [Tabubil]; (9) Cosmopsaltria aurata (Duffels), Hemiptera, Cicadidae [Tabubil]; (10) E. popa (Sta˚l), male, Phasmatodea, Phasmatidae [Tabubil].

Insects of the Fly River System

505

including Mesovelia horvathi that occurs in Australia, throughout most of Southeast Asia and possibly beyond, eight species were recorded only from the Fly River basin and may be endemic: Calyptobates simplex, C. kiunga, Iobates somare, and P. kiunga (Gerridae); H. kiunga (Hydrometridae); M. stysi (Mesoveliidae); Aphelocheirus pallens (Aphelocheiridae); Idiocarus isolatus (Naucoridae); and Neusterensifer pseudocyclops (Veliidae). Other papers that mention Fly River Hemiptera include those of van Doesburg (2004) (Velocipedidae), Evans (1971, 1981) (Cicadellidae), Fennah (1977, 1980) (Fulgoridae and Cixiidae), Grimshaw and Donaldson (2007) (Coccidae), Jansson (1982) (Corixidae), and Slater (1961) (Lygaeidae); but this list is not exhaustive.

14.6. Other Insects Other orders of insects from the Fly River basin are at best poorly known and poorly represented in collections. Some termites (Isoptera) have been described from the Fly River basin (Roisin, 1990; Roisin and Pasteels, 1996, 2000) including Microcerotermes ﬂyensis, a possible endemic with characteristic very long mandibles. A revision of the termites of the Termes–Capitermes group (Termitinae) that incorporates Fly River species is nearing completion (Roisin, personal communication). The aulacid wasp Pristaulacus kiunga (Hymenoptera) has recently been described by Jennings and Austin (2006). Some other papers that mention Fly River Hymenoptera include those of Brothers (1971) (Mutillidae), Gibson (2005) (Eupelmidae), and Smith (1980) (Pergidae). Among the Orthoptera (grasshoppers, locusts, and katydids) Kirby (1899) described the large katydid Phyllophora eburneiguttata (Tettigonidae), which is known only from the Fly River. R. B. Lachlan encountered the similar but giant katydid, Siliquofera grandis [Fig. 14.4(4)], at Tabubil and Matkomrae, an impressive insect with a wingspan of over 20 cm. I have not been able to ﬁnd other references to Fly River Orthoptera but they may well exist. Certainly, Orthoptera are plentiful, in both the lower mountains and the lowlands. Lachlan also collected two remarkable phasmids (Phasmatodea) near Tabubil: the giant spiny leaf insect, Extatosoma popa [Fig. 14.4(1)], and the leaf insect, Phyllium sp. [Fig. 14.4(7)]. The praying mantid, Kongobatha papua (Mantodea: Mantidae), has been described from Kiunga (Beier, 1965). This is one of the snake mantids, so named because the very agile nymphs can wind their way rapidly through grass. Beier also describes Machairima papua, one of the tree-running

506

M. S. Moulds

mantids that is known from just two specimens, also collected at Kiunga, and records ﬁve other species from the same locality. Storey and Maie (1993) summarize collections of aquatic macroinvertebrates from monitoring programs undertaken by Ok Tedi Mining Limited. While 86 species are listed, most are identiﬁed only to family or genus with only two mayﬂies (Ephemeroptera), Plethogenesia pallida and P. papuana, carrying species identiﬁcation. Dragonﬂies and damselﬂies (Odonata) are conspicuously rare in the fast-ﬂowing mountain creeks and rivers, but are more plentiful along the lower Fly River. Although M. A. Lieftinck and others wrote many papers on the dragonﬂies of New Guinea there appears to be no publication concerning the dragonﬂies of the Fly River. I know of no publications on the insects of other orders including Blattodea (cockroaches), Psocoptera (psocids), and Neuroptera (lacewings), but it is difﬁcult to conﬁrm this and there may be publications that mention these groups.

14.7. Concluding Remarks Few endemic insects are known from the Fly River basin. This is probably because the distribution of most New Guinea insects is not so much dependent upon river basins but rather on altitude. Species tend to be distributed along altitudinal bands so that distributions go laterally along the central cordillera. However, to some extent, the central cordillera does divide the insect fauna between its northern and southern sides. An exception is likely to be those aquatic insects dependent upon the aquatic environment of the river itself, but very little is known about these. With time, it is possible that the Fly River basin will prove to be one of the richest regions of insect diversity in Papua New Guinea, especially the mountainous regions. To date, only butterﬂies, hawk moths, beetles, ﬂies, and cicadas have been collected in any serious way. Much more collecting is needed before we can say that the insect diversity is reasonably documented. Most notable is the lack of knowledge and collection of the major orders Hymenoptera, Hemiptera (excluding cicadas), and Orthoptera, all of which are common insects. There is a similar lack of collections and knowledge for aquatic insects: Odonata (dragonﬂies and damselﬂies), Trichoptera (caddisﬂies), Megaloptera (alderﬂies), and the like. The lowlands, in particular, have been largely neglected, certainly since the early expeditions of the 1800s. The remoteness of the Fly River basin has been undoubtedly one of the contributing factors to the comparative lack of knowledge of the insects.

Insects of the Fly River System

507

Specialist collectors of speciﬁc insect groups have tended not to venture to the Fly River but conﬁne their research to more accessible regions. The Fly River is the last frontier for entomologists in Papua New Guinea, a frontier that is not yet challenged.

ACKNOWLEDGMENTS I am grateful to Ok Tedi Mining Limited for ﬁeld support, especially to staff of the Environment Section and senior geologists for logistical support. United Paciﬁc Drilling kindly provided accommodation at drill sites. R. B. Lachlan generously allowed access to his extensive collection of Fly River insects and allowed photography of specimens for the ﬁgures. For comments on the manuscript, I am especially grateful to B. R. Bolton, E. D. Edwards, R. B. Lachlan, S. E. Miller, and D. C. F. Rentz; B. R. Bolton also collected many insect specimens for study, many of which remain undescribed. C. Pardoe-Matthews kindly assisted with photography for the ﬁgures. S. A. Cowan assembled the ﬁgures ready for publication, and S. F. McEvey provided technical expertise relating to ﬁgure preparation. G. B. Monteith kindly assisted with information on collectors. For assistance with literature searches, I thank E. D. Edwards, D. K. McAlpine, S. F. McEvey, S. E. Miller, G. B. Monteith, D. A. Polhemus, Y. Roisin, and R. I. Storey. For identiﬁcation of specimens, I am grateful to A. de Boer (Cicadidae), P. Brock (Phasmatodea), J. P. Duffels (Cicadidae), E. D. Edwards (Lepidoptera), R. de Keyzer (Coleoptera: Cerambycidae), M. Delpont (Coleoptera: Cetoniinae), G. Hangay (Coleoptera: Scarabaeidae), R. B. Lachlan (Lepidoptera), U. Nylander (Coleoptera: Buprestidae), G. P. Setliff (Curculionoidea), A. Sundholm (Coleoptera), and T. A. Weir (Coleoptera).

REFERENCES Baehr, M. (1996). The ground beetle genus Casnoidea Castelnau: Taxonomy, phylogeny and zoogeography (Insecta: Coleoptera: Carabidae: Odacanthinae). Invertebrate Taxonomy, 10, 1041–1084. Balke, M. (1995). The Hydroporini (Coleoptera: Dytiscidae: Hydroporinae) of New Guinea: Systematics, distribution and origin of the fauna. Invertebrate Taxonomy, 9, 1009–1019. Beier, M. (1965). Die mantodeen Neu-Guineas. Paciﬁc Insects, 7, 473–502. de Boer, A. J. (1992). The taxonomy and biogeography of the genus Thaumastopsaltria Kirkaldy, 1900 (Homoptera, Tibicinidae). Beaufortia, 43, 17–44.

508

M. S. Moulds

de Boer, A. J. (1993). Guineapsaltria, a new genus from the Australian-New Guinea region (Homoptera, Tibicinidae), with notes on its taxonomy and biogeography. Bijdragen tot de Dierkunde, 63, 15–41. de Boer, A. J. (1994a). The taxonomy and biogeography of the exhausta group of the genus Baeturia Sta˚l, 1866 (Homoptera, Tibicinidae). Beaufortia, 44, 127–158. de Boer, A. J. (1994b). Four species added to the Baeturia nasuta group, with notes on taxonomy and biogeography (Homoptera, Tibicinidae). Tijdschrift voor Entomologie, 137, 161–172. de Boer, A. J. (1995a). The taxonomy, phylogeny and biogeography of the cicada genus Gymnotympana Sta˚l, 1866 (Homoptera, Tibicinidae). Invertebrate Taxonomy, 9, 1–81. de Boer, A. J. (1995b). The taxonomy and biogeography of the cicada genus Papuapsaltria gen. n. (Homoptera, Tibicinidae). Tijdschrift voor Entomologie, 138, 1–44. de Boer, A. J. (1999). Taxonomy and biogeography of the New Guinea Cicadettini (Hemiptera, Tibicinidae). Mitteilungen aus dem Museum fu¨r Naturkunde in – Berlin, Deutsche Entomologishe Zeitschrift, 46, 115–147. de Boer, A. J., & Duffels, J. P. (1996). Biogeography of Indo-Paciﬁc cicadas east of Wallace’s Line. In: A. Keast, & S. E. Miller (Eds). The Origin and Evolution of Paciﬁc Island Biotas, New Guinea to Eastern Polynesia: Patterns and Process. SPB Academic Publishing, Amsterdam, pp. 297–330. Bourguignon, T., & Roisin, Y. (2006). Revision of the termitophilous tribe Pseudoperinthini (Coleoptera: Staphylinidae) in New Guinea. Insect Systematics and Evolution, 37, 443–456. Brothers, D. J. (1971). Ascetotilla, a new genus of Mutillidae from New Guinea (Hymenoptera). Paciﬁc Insects, 13, 471–485. Cande`ze, E. C. A. (1880). Addition au releve´ des Elate´rides Malais. Annali del Museo Civico di Storia Naturale di Genova, 15, 188–198. Chuˆjoˆ, M., & Chuˆjoˆ, M. (1987). Erotylidae from New Guinea and her adjacent islands. I (Coleoptera). Esakia, 25, 5–36. Clements, A. N. (2000). A revision of Brachyrhopala Macquart, an Australian region genus (Diptera: Asilidae). Invertebrate Taxonomy, 14, 77–114. Cooper, R. D., Waterson, D. G. E., Kupo, M., Foley, D. H., Beebe, N. W., & Sweeney, A. W. (1997). Anopheline mosquitoes of the Western Province of Papua New Guinea. Journal of the American Mosquito Control Association, 13, 5–12. D’Albertis, L. M. (1877). Journal of the Expedition for the Exploration of the Fly River. Frederick White, Sydney, 43 pp. D’Albertis, L. M. (1880). New Guinea: What I Did and What I Saw. Sampson Low, Marston, Searle and Rivington, London. vol. 1, pp. 1–424; vol. 2, pp. 1–406. van Doesburg, P. H. (2004). A taxonomic revision of the family Velocipedidae Bergroth, 1891 (Insecta: Heteroptera). Zoologische Verhandelingen, Leiden, 347, 5–110. Duffels, J. P. (1977). A revision of the genus Diceropyga Sta˚l 1870 (Homoptera, Cicadidae). Monograﬂiee¨n van de Nederlandsche Entomologische Vereeniging, 8, 1–227.

Insects of the Fly River System

509

Duffels, J. P. (1983). Taxonomy, phylogeny and biogeography of the genus Cosmopsaltria, with remarks on the historic biogeography of the subtribe Cosmopsaltriaria (Homoptera: Cicadidae). Paciﬁc Insects Monograph, 39, 1–127. Evans, F. (1981). The Tartessinae of Australia, New Guinea and some adjacent islands. Paciﬁc Insects, 23, 112–188. Evans, J. W. (1971). Leafhoppers from New Guinea and Australia belonging to the subfamilies Macropsinae and Agalliinae with notes on the positions of Nionia Ball and Magnentius Pruthi (Homoptera: Cicadelloidea). Paciﬁc Insects, 13, 343–360. Fauvel, A. (1877). Les Staphylinides de l’Australie et de la Polyne´sie. Annali del Museo Civico di Storia Naturale di Genova, 10, 168–298. Fauvel, A. (1878a). Les Staphylinides de l’Australie et de la Polyne´sie. Annali del Museo Civico di Storia Naturale di Genova, 13, 465–598. Fauvel, A. (1878b). Les staphylinides des Moluques et de la Nouvelle Guine´e. (2e. Me´moire). Annali del Museo Civico di Storia Naturale di Genova, 12, 171–315. Fauvel, A. (1879). Les staphylinides des Moluques et de la Nouvelle Guine´e. (2e. Me´moire). Annali del Museo Civico di Storia Naturale di Genova, 15, 63–121. Fennah, R. G. (1977). New species and new records of Fulgoridae (Homoptera: Fulgoroidea) from New Guinea. Paciﬁc Insects, 17, 373–403. Fennah, R. G. (1980). The genus Bajauana and two allied genera in New Guinea (Fulgoroidea: Cixiidae). Paciﬁc Insects, 22, 237–328. Gentili, E. (2002). Descrizione di nuove specie del genere Paranacaena Blackburn, 1889 (Coleoptera, Hydrophilidae). Giornale italiano di Entomologia, 10, 77–97. Gibson, G. A. P. (2005). The world species of Balcha Walker (Hymenoptera: Chalcidoidea: Eupelmidae), parasitoids of wood-borning beetles. Zootaxa, 1033, 1–62. Gotts, R., & Pangemanan, N. (2001). Mimika Butterﬂies. A Guide to the Butterﬂies of the Mimika Subdistrict of Papua. PT Freeport Indonesia Biodiversity Centre, Timika, 287 pp. Gressitt, J. L. (1959). Longicorn beetles from New Guinea, I (Cerambycidae). Paciﬁc Insects, 1, 59–171. Gressitt, J. L. (1971). Chrysomelid beetles from the Papuan subregion, 7 (Donaciinae). Paciﬁc Insects, 13, 607–609. Grimshaw, J. F., & Donaldson, J. F. (2007). New records of mango shield scale Milviscutulus mangiferae (Green) (Hemiptera: Coccidae) and Brevennia rehi (Lindinger) (Hemiptera: Pseudococcidae) in north Queensland. Australian Journal of Entomology, 46, 96–98. Grund, R. (2007). Some old butterﬂy records from Lake Murray, Papua New Guinea. Victorian Entomologist, 37, 38–40. Hebauer, F. (2000). The New Guinean species of the genus Platycyon Hansen, 1999 (Coleoptera: Hydrophilidae). Acta Coleopterologica, 16(1), 3–16. Howden, H. F. (1989). The Geotrupinae of New Guinea (Coleoptera: Scarabaeidae). Invertebrate Taxonomy, 3, 261–289.

510

M. S. Moulds

Jacoby, M. (1883). Descriptions of new genera and species of phytophagous Coleoptera from the Indo-Malayan and Austro-Malayan subregions, contained in the Genoa Civic Museum. Annali del Museo Civico di Storia Naturale di Genova, 20, 188–192. pp. 193–233. Jacoby, M. (1885). Descriptions of new genera and species of phytophagous Coleoptera from the Indo-Malayan and Austro-Malayan subregions, contained in the Genoa Civic Museum. Annali del Museo Civico di Storia Naturale di Genova, 22, 20–76. Jacoby, M. (1886). Descriptions of new genera and species of phytophagous Coleoptera from the Indo-Malayan and Austro-Malayan subregions, contained in the Genoa Civic Museum. Annali del Museo Civico di Storia Naturale di Genova, 24, 41–121. James, M. T. (1971). New species and records of Australasian Calliphorinae, with special reference to the fauna of New Guinea (Diptera: Calliphoridae). Paciﬁc Insects, 13, 1–12. Jansson, A. (1982). Notes on some Corixidae (Heteroptera) from New Guinea and New Caledonia. Paciﬁc Insects, 24, 95–103. Jennings, J. T., & Austin, A. D. (2006). Aulacid wasps (Hymenoptera: Aulacidae) of New Guinea, with descriptions of ﬁve new species. Zootaxa, 1365, 19–35. Kirby, W. F. (1899). Notes on the orthopterous genus Phyllophora. Annals and Magazine of Natural History, 7(4), 302–311. Plate 6. Knight, R., McAdam, K. P. W. J., Matola, Y. G., & Kirkham, V. (1979). Bancroftian ﬁlariasis and other parasitic infections in the middle Fly River region of Western Papua New Guinea. 1. Clinical, parasitological and serological studies. Annals of Tropical Medicine and Parasitology, 73, 563–576. Lachlan, R. B. (1999a). A new species of Delias Hu¨bner (Lepidoptera: Pieridae) from the Star Mountains, Papua New Guinea. Australian Entomologist, 26, 33–35. Lachlan, R. B. (1999b). A new genus and species of hawk moth (Lepidoptera: Sphingidae) from Papua New Guinea. Australian Entomologist, 26, 91–95. Lachlan, R. B. (2000a). Corrections to an annotated list of the hawk moths (Lepidoptera: Sphingidae) of Western Province, Papua New Guinea. Australian Entomologist, 27, 31. Lachlan, R. B. (2000b). New species, subspecies and records of Delias Hu¨bner (Lepidoptera: Pieridae) from the upper Ok Tedi and Tari regions, Papua New Guinea. Australian Entomologist, 27, 71–96. Lachlan, R. B., & Kitching, I. J. (2001). A new hawk moth (Lepidoptera: Sphingidae) from Papua New Guinea. Australian Entomologist, 28, 117–122. Last, H. R. (1980). Records of New Guinea Staphylinidae (Coleoptera) in the Hungarian Natural History Museum. Annales Historico-Naturales Musei Nationalis Hungarici, 72, 139–161. Last, H. R. (1987). Staphylinidae from Papua New Guinea in the collection of Bernice P. Bishop Museum Honolulu, Hawaii (Insecta, Coleoptera). Entomologische Abhandlungen, 51(3), 25–56.

Insects of the Fly River System

511

Mackerras, I. M. (1964). The Tabanidae (Diptera) of New Guinea. Paciﬁc Insects, 6, 69–210. Macleay, W. J. (1886). The insects of the Fly River, New Guinea, ‘‘Coleoptera’’. Proceedings of the Linnean Society of New South Wales, 2(1), 136–157. pp. 183–204. McAlpine, D. K. (1994). Review of the species of Achias (Diptera: Platystomatidae). Invertebrate Taxonomy, 8, 117–281. Meyrick, E. (1886). On some Lepidoptera from the Fly River. Proceedings of the Linnean Society of New South Wales, 2(1), 241–258. Moulds, M. S. (1990). Australian Cicadas. New South Wales University Press, Kensington, 217 pp., 24 plates. Moulds, M. S., & Lachlan, R. B. (1998). An annotated list of the hawk moths (Lepidoptera: Sphingidae) of Western Province, Papua New Guinea. Australian Entomologist, 25(2), 45–60. Oberthu¨r, C. (1880). E´tude sur les collections de Le´pidopte`ra Oce´aniens appartenant au Muse´e Civique de Geˆnes. Annali del Museo Civico di Storia Naturale di Genova, 15, 461–530, 3 plates. Parsons, M. (1999). The Butterﬂies of Papua New Guinea: Their Systematics and Biology. Academic Press, London, xvi, 736 pp., 26þ136 plates. Pascoe, F. P. (1886). List of the Curculionidae of the Malay Archipelago collected by Dr. Odoardo Beccari, L. M. D’Albertis, and others. Annali del Museo Civico di Storia Naturale di Genova, 22, 201–332, plates 1–3. Polhemus, D. A., & Polhemus, J. T. (1986). Naucoridae of New Guinea, II. A review of the genus Idiocarus Montandon (Hemiptera: Naucoridae) with descriptions of three new species. Journal of the New York Entomological Society, 94, 39–50. Polhemus, D. A., & Polhemus, J. T. (1989). The Aphelocheirinae of tropical Asia (Heteroptera: Naucoridae). Rafﬂes Bulletin of Zoology, 36, 167–310. Polhemus, D. A., & Polhemus, J. T. (1997). A review of the genus Limnometra Mayr in New Guinea, with the description of a very large new species (Heteroptera: Gerridae). Journal of the New York Entomological Society, 105, 24–39. Polhemus, D. A., & Polhemus, J. T. (2000). Additional new genera and species of Microveliinae (Heteroptera: Veliidae) from New Guinea and adjacent regions. Tijdschrift voor Entomologie, 143, 91–123. Polhemus, D. A., & Polhemus, J. T. (2001a). A revision of the genus Ptilomera (Heteroptera: Gerridae) on New Guinea and nearby islands. Journal of the New York Entomological Society, 109, 81–166. Polhemus, J. T., & Polhemus, D. A. (1993). The Trepobatinae (Heteroptera: Gerridae) of New Guinea and surrounding regions, with a review of the world fauna. Part 1. Tribe Metrobatini. Entomologica Scandinavica, 24, 241–284. Polhemus, J. T., & Polhemus, D. A. (1994). The Trepobatinae (Heteroptera: Gerridae) of New Guinea and surrounding regions, with a review of the world fauna. Part 2. Tribe Naboandelini. Entomologica Scandinavica, 25(3), 333–359. Polhemus, J. T., & Lansbury, I. (1997). Revision of the genus Hydrometra Latrielle in Australia, Melanesia, and the Southwest Paciﬁc (Heteroptera: Hydrometridae). Bishop Museum Occasional Papers, 47, 1–67.

512

M. S. Moulds

Polhemus, J. T., & Polhemus, D. A. (2001b). The genus Mesovelia Mulsant & Rey in New Guinea (Mesoveliidae: Heteroptera). Journal of the New York Entomological Society, 108, 205–230. Rand, A. L., & Brass, L. J. (1940). Results of the Archbold Expeditions. No. 29 Summary of the 1936–1937 New Guinea Expedition. Bulletin of the American Museum of Natural History, 77, 341–380. Plates 21–42, 2 maps. Reitter, E. (1880). Neue Nitidulidae des Museo Civico di Storia Naturale in Genova. Annali del Museo Civico di Storia Naturale di Genova, 15, 124–128, pp. 454, 460. Riedel, A. (2002). Revision of the subgenus Neosynaptops Voss of Euops Schoenherr (Coleoptera, Curculionoidea, Attelabidae) from the Papuan region Zoologica Scripta, 31, 135–165. Roisin, Y. (1990). The termite genus Amitermes Silvestri in Papua New Guinea. Indo-Malayan Zoology, 6, 185–194. Roisin, Y., & Pasteels, J. M. (1996). The nasute termites (Isoptera: Nasutitermitinae) of Papua New Guinea. Invertebrate Taxonomy, 10, 507–616. Roisin, Y., & Pasteels, J. M. (2000). The genus Microcerotermes (Isoptera: Termitidae) in New Guinea and the Solomon Islands. Invertebrate Taxonomy, 14, 137–174. Schuurkamp, G. J. T. (1987). Epidemic malaria in the hyperendemic North Fly Region of Western Province, Papua New Guinea. Papua New Guinea Medical Journal, 30, 39–44. Schuurkamp, G. J. T. (1992). The epidemiology of malaria and ﬁlariasis in the Ok Tedi region of Western Province, Papua New Guinea. Ok Tedi Mining Limited, Medical Department (Public Health), Tabubil. 341 pp., þplates. Schuurkamp, G. J. T., Matango, M., Napil, J., & Vergara, R. (1987). Malaria control and the development of the Ok Tedi Mining Project in the Star Mountains of Papua New Guinea: The initial construction phase. Papua New Guinea Medical Journal, 30, 27–38. Setliff, G. P. (2007). Annotated checklist of weevils from Papuan region (Coleoptera, Curculionoidea). Zootaxa, 1536, 1–296. Shaverdo, H. V., Sagata, K., & Balke, M. (2005). Five new species of the genus Papuadytes Balke, 1998 from New Guinea (Coleoptera: Dytiscidae). Aquatic Insects, 27, 269–280. Slater, J. A. (1961). Dentisblissus: A new genus of Blissinae from New Guinea (Hemiptera: Lygaeidae). Paciﬁc Insects, 3, 481–484. Smith, D. R. (1980). Pergidae (Hymenoptera) from New Guinea and Australia in the Bishop Museum. Paciﬁc Insects, 22, 329–346. Stebnicka, Z. T. (1998). The Aphodiinae of New Guinea, the Bismarck Archipelago and the Solomon Islands (Coleoptera: Scarabaeoidea). Invertebrate Taxonomy, 12, 833–895. Steffan, W. A., & Evenhuis, N. L. (1982). A new species of Toxorhynchites from Papua New Guinea (Diptera: Culicidae). Mosquito Systematics, 14, 1–13. Storey, A. W., & Maie, A. Y. (1993). Biological monitoring of aquatic macroinvertebrate assemblages in the Fly River system. In: R. Ho¨ft (Ed.).

Insects of the Fly River System

513

Proceedings of the Biological Society of New Guinea Annual Meeting 1993. Wau Ecology Institute, Wau, pp. 119–128. Thompson, R. T. (1996). The species of Phaenomerus Scho¨nherr (Coleoptera: Curculionidae: Zygopinae) of the Australian Region. Invertebrate Taxonomy, 10, 937–993. Tryon, H. (1890). Rhopalocera collected in British New Guinea during year 1889–90, under the auspices of his Honour the Administrator. Report of the Administrator British New Guinea, 2, 112–115, Appendix 5. Uhmann, G. (1995). Anthicidae (Insecta: Coleoptera) from New Guinea in the Hornabrook Collection. Journal of the Royal Society of New Zealand, 25, 517–526. Upton, M. S. (1997). A Rich and diverse Fauna. The History of the Australian National Insect Collection 1926–1991. CSIRO Publishing, Collingwood, 386 pp.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00415-1

Chapter 15

Vegetation of the Ok Tedi–Fly River System Monica T. Rau1, and Douglas P. Reagan2 1

Community Relations Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, W.P. Papua New Guinea 2 Doug Reagan & Associates, 350 Gordon Drive, Castle Rock, CO 80104, USA

15.1. Introduction The Fly River is the biggest river in Papua New Guinea (PNG) with a catchment area of some 76,000 km2, a length of more than 1,000 km, and width of about 93 km at the mouth (Fig. 15.1). It drains the Hindenburg and Victor Emanuel Ranges to the north and the Muller Range to the northeast. From its headwaters to the sea, the riparian vegetation of the Ok Tedi–Fly River system reﬂects signiﬁcant climatic, geographical, and geomorphological differences. Vegetation of the Basin ranges from mid- to lower-montane forests at 2,000–3,000 m above sea level (asl), foothill forests at around 600– 1,000 m, to lowland tropical and ﬂoodplain forests, swamp forests, open grassed wetland ﬂoodplains, Eucalyptus savannah grassland at 20–150 m asl, and ﬁnally to the mangrove forests at the Fly River delta (Paijmans, 1976). Detailed taxonomic and species distribution information on many plant species in the Ok Tedi–Fly River system can be found in several contributions in the Handbook of the Flora of Papua New Guinea, Volume 1 edited by Womersley (1978) and Volume 3 edited by Conn (1995). On its way to the sea, the river system passes through three distinct climate zones, deﬁned fundamentally by altitude, rainfall, temperature, and the prevailing southeasterly and northwesterly winds that bring about the dry season and the rains, respectively. Mean annual rainfall ranges from over Corresponding author. Tel.: +675-5483144; Fax: +675-5489603;

E-mail: [email protected] (M.T. Rau).

516

M. T. Rau and D. P. Reagan

Figure 15.1: Map of the Ok Tedi–Fly River system showing the extent of the drainage basin (dashed line) and location of selected villages.

Vegetation of the Ok Tedi–Fly River System

517

12,000 mm in the headwaters of the Ok Tedi to between 1,500 and 2,000 mm in the lowland, subhumid region at the Fly River estuary (McAlpine et al., 1983). In between these extremes, around Ningerum, Kiunga, and the upper Middle Fly, the rainfall is 4,000–5,000 mm. Below Everill Junction, downstream to Suki, Sturt Island, Tapila, and the South Fly region including Daru, the climate changes to a distinct wet and dry similar to the rest of the Papuan coastline. The dry season is brought on by the prevailing southeasterly winds from May to September and the monsoon rains brought on by the northwesterly winds from December to March. The daily temperature ranges from 131C to 281C around the Ok Tedi (Mount Fubilan) mine pit and the Tabubil area. Humidity averages around 90% most of the year reﬂecting the very high rainfall in the region. As the altitude decreases around Ningerum, Kiunga, and the Middle Fly region, the temperature is warmer ranging from 261C to 331C reﬂecting normal tropical conditions. Humidity is lower at around 80–90% reﬂecting a decrease in precipitation. Toward the coast, with decreasing rainfall, humidity is around 70–80%, being affected by prevailing winds and drier conditions. This chapter describes the major vegetation types found along the Ok Tedi and the Fly River system from the headwaters to the estuary. Descriptions include key ecological processes, including plant colonization and reproduction.

15.2. Regional Vegetation The high mountains of the upper Ok Tedi catchment, the foothills bounding the middle reaches of the Ok Tedi, and the extensive swampy lowlands of the Middle Fly and South Fly areas form distinct vegetation regions (Fig. 15.2). These vegetation communities have been variously affected by the operation of the Ok Tedi Mine, located near the headwaters of the Ok Tedi. A brief survey of major habitats in the Basin is provided in the following sections.

15.2.1. Upper Ok Tedi Catchment Mixed-montane association forests occur at 1,500–2,000 m asl and have dense 20- to 30-m-high evergreen trees. Common tree species include, Castanopsis acuminatissima, Lithocarpus sp., and genera Elmerrillia, Myristica, Weinmannia, and Araucaria. The thick shrub undergrowth layer and lianas are not common in this association. Tree ferns, genera Dicksonia and Cyathea, sometimes form pure stands or are mixed with the forest as

518

M. T. Rau and D. P. Reagan

Figure 15.2: Map of the Ok Tedi–Fly River system showing distribution of vegetation types.

Vegetation of the Ok Tedi–Fly River System

519

undergrowth. The lower-montane forest below 1,500 m asl merges into the foothill forests around Finalbin, Bultem, and Korkit villages; Tabubil township; and the Lukwi, Ok Ma, Migalsimbip, and Olsobip areas (Fig. 15.3). Here the lianas, with a thick understory layer, form an impenetrable structure permanently wet, rich in lichens, mosses, algae, and ferns (Rau, 1996). Prominent species include Albizia falcataria, Metrocideros spp., free-living and epiphytic Schefﬂera spp., Elaeocarpus spp., and Finschia chloroxantha. Several species of rattan (Korthalsia spp. and Calamus spp.) constitute the large prominent climbers above the canopy (Fig. 15.4). The epiphytes are most abundant and diverse at this altitude. Predominant species include numerous ferns, orchids, some herbaceous species, and thick layers of lichens and mosses. The high rainfall results in very high rates of nutrient leaching through the soil. In most places, the topsoil is either very thin or nonexistent. Landslides are a major feature of this region and contribute to the suspended sediment load of streams and rivers in the upper catchment. The major nutrient source is above ground in the form of humus, dead logs, and leaf litter. Plant growth rates are generally high in the foothill forest region, despite extreme poor soil conditions. Some secondary tree species can grow more than 2 m a year. Examples of these species include Casuarina spp., Spondias dulcis, and A. falcataria. The lush green cover from the air is deceiving. Ground survey

Figure 15.3: Foothills forest typical of the high-rainfall area between Ningerum and Tabubil in the upper Ok Tedi (river can be seen in lower right). View looks toward the Ok Tedi Mine (area of exposed rock on the left just below skyline) and the rugged Star Mountains that form the backdrop to photograph.

520

M. T. Rau and D. P. Reagan

Figure 15.4: Albizia falcataria as an emergent species in a climax forest, near Boungkim village, upper Ok Tedi catchment. studies in this region showed that timber volume of trees with a diameter of greater than 60 cm is around 10 m3/ha, compared to about 15–20 m3 in West New Britain or Vanimo in the West Sepik region.

15.2.2. Lower Ok Tedi Further downstream at 20–100 m asl, the vegetation consists of a rich maximum tropical rainforest with a 35- to 40-m-high closed canopy and sparse undergrowth. Creeping Piper spp., Araceae members, ferns, Donax cannaeformis, Zingiberaceae members, and Leea spp. make up the sparse undergrowth. The epiphytes are less common than in the foothill mixed lower-montane forests. Before the development of forest dieback on the ﬂoodplain of the lower Ok Tedi and its tributaries (Fig. 15.5), canopy cover in undisturbed sections of the forest ranged from 80% to 95% (Rau, 1994), the primary cause of the sparse undergrowth. Prior to the start of mining at Mount Fubilan in 1984, the riparian vegetation around the lower Ok Tedi and the upper Middle Fly below the D’Albertis Junction (where the Ok Tedi joins the Fly River) was composed of some of the most luxuriant diverse lowland ﬂoodplain rainforests in the tropics. Immediately along the riverbanks, secondary forest species, including Artocarpus communis, Ficus pungen, Ficus adernosperma, Barringtonia sp., Macaranga spp., Glochidion spp., Nauclea orientale, Timonius timon, and Bischoﬁa javanica, were the most important small tree species

Vegetation of the Ok Tedi–Fly River System

521

Figure 15.5: Forest dieback on the lower Ok Tedi.

Figure 15.6: Wild breadfruit Artocarpus communis is a common secondary tree species around Ningerum and lower Ok Tedi. Photo taken at Wogam village inland from Ningerum. (Fig. 15.6). The grass Saccharum robustum sometimes mixed with the common reed, Phragmites karka, formed a narrow fringe on the depositing banks from the lower Ok Tedi and downstream into the Fly River. A. communis and F. pungen fruits provided an important food source to people, birds, ﬂying foxes, and a large variety of insects and small mammals such as bats and bandicoots.

522

M. T. Rau and D. P. Reagan

Immediately behind the riverbanks, the ﬂoodplain supported up to 45-m tall, ﬂoristically rich rainforest mixed with sago palms. Common canopy tree species included Cryptocaria spp., Vatica papuana, Myristica sp., Pometia pinnata, Terminalia sp., Maniltoa psilogyne, Syzygium sp., Octomeles sumatrana, Ervatamia sp., Ficus benjamina, Celastraceae sp., Alstonia brassii, Dysoxylum sp., Semecarpus spp., and Pangium edule. Palms in the ﬂoodplain forest were uncommon but Gulubia costata occurred on the higher margins of the ﬂoodplain while Metroxylon sagu (sago palm) occurred in concentrated stands in low-lying areas (Piskaut et al., 2000) often mixed with other forest tree species. Sago was and still is a staple diet of the river communities. Now the communities walk long distances inland to make sago or buy it from the neighboring villages where sago palms are not affected by ﬂoods and dieback. Another palm species that occurred on the ﬂoodplain and is important to the communities is Caryota rumphiana (Fig. 15.7). The stem of this palm is used extensively for traditional house ﬂooring and walls. Some of the richest stands of C. rumphiana in PNG are found in the lower Ok Tedi and the upper Middle Fly regions. The fertile ﬂoodplain humus-rich soils were the most important gardening sites for the communities who live along the Ok Tedi. With the changes brought about by the inﬂux of mine-derived sediment and increased overbank ﬂooding, the soil composition has become less fertile and is now largely alkaline ﬁne silt. This soil is not suitable for gardening and the communities cannot plant crops in this mine-derived sediment. Gardening is now restricted to higher ground inland that is less susceptible to ﬂoods and sedimentation.

Figure 15.7: Caryota rumphiana palm, Boungkim village, looking toward Olsobip District in the background.

Vegetation of the Ok Tedi–Fly River System

523

Dredging since 1997 at Bige in the lower Ok Tedi has mitigated some of the impacts of mine-related sedimentation and ﬂooding. Most of the area is under good recovery. Secondary forest is well established in places along the riverbanks and in adjacent areas of the ﬂoodplain. With continued dredging the succession should develop into a well-established secondary forest. However, some degree of uncertainty remains about what happens after mine closure in 2013, when the dredging ceases. Clearly some strategy needs to be included into the current mine closure plan where ﬂoodplain recovery can be monitored post mine closure. 15.2.3. Middle Fly Vegetation Published data on the vegetation of the Middle Fly is scarce. Leach and Osborne (1985) provide a list of aquatic plants and their distributions from the Middle Fly wetlands. Specimens from the National Herbarium and the University of Papua New Guinea Herbarium indicate that a number of plant surveys were conducted in some of the localities within the Middle Fly, especially Lake Daviumbu and Bosset Lagoon. A number of reports on the vegetation of the Western Province are available at the Papua New Guinea Forest Research Institute but are sketchy and could not be veriﬁed or quoted. Information on some weed species distribution throughout PNG including the Ok Tedi and Fly River ﬂoodplain habitats can be found in Department of Forest, Botany Division’s botanical bulletin publications (Henty and Pritchard, 1982). The Fly River ﬂoodplain between D’Albertis and the Everill Junctions covers an area of about 2,500 km2 (Smith and Bakowa, 1994). In this chapter, this area will be referred to as the Middle Fly. Approximately 90% of this area used to be covered by vegetation including lowland and ﬂoodplain rainforest, wetland grass species, and sedges of the family Cyperaceae, weedy herbaceous species, shrubs, and ﬂood-tolerant trees (Fig. 15.8). The aquatic ﬂora can be grouped into strictly aquatic species, that is, plants that remain and grow in water as their normal habitat, or semiaquatic that can grow both under ﬂood conditions as well as on dry land. The remaining 10% of the area comprised of open areas, largely occupied by the various types of lakes (off-river water bodies) found in the ﬂoodplain. The plant cover in the lakes is subjected to changes in water-level ﬂuctuations caused by the rainfall and drought patterns. Prior to forest dieback, the vegetation of the Middle Fly could be broadly categorized into two major vegetation types: the ‘‘open forest on the plains’’ and the ‘‘swamp grassland in river ﬂoodplain’’ (Paijmans, 1976). The open

524

M. T. Rau and D. P. Reagan

Figure 15.8: Low-altitude aerial view of swampy grasslands typical of lower reaches of the Middle Fly ﬂoodplain. forest vegetation covered the upper Middle Fly, which included the areas from the D’Albertis Junction to around Kwem [adopted river mile (ARM) 346], just above Manda, while the swamp grassland covers the lower Middle Fly, which includes the river reach from Manda to the Everill Junction. The canopy structure of the ‘‘open forest on the plain’’ tended to be very irregular; large gaps existed between clumps of trees. These areas were often dominated by climbing rattan palms, especially in the gaps and along the riverbanks. The dominant canopy tree genera were Pometia, Terminalia, Alstonia, O. sumatrana, Planchonia, Bischoﬁa, Cananga, and Nauclea. A back plain swamp forest that comprised common tree species such as Campnosperma, Terminalia, Syzygium, Nauclea, and Myristica was associated with the open forest. This divide between the inner plain with open canopy forest and an outer plain with swamp forest reﬂects a change in the local topography. The inner plain was characterized by topographic features such as scroll bars and crevasse splays. The outer plain was of lower elevation and was inhabited by unique swamp forests probably endemic to the Fly River (e.g., the swamp forests of the upper Binge River behind Membok village). At the time of writing, much of this area was under permanent standing water; nearly all of this swamp forest had undergone severe dieback and was mostly gone. Much of the open swamp grassland and associated communities between Manda and the Everill Junction reﬂect an environment permanently under water. This vegetation type is dominated by aquatic grass species: Leersia hexandra, Oryza ruﬁpogon, Hymenachne acutigluma, P. karka, S. robustum,

Vegetation of the Ok Tedi–Fly River System

525

Echinochloa praestans, Eleocharis sphacelata, and Coix lachryma-jobi (Fig. 15.9). Associated with the swamp grassland areas are the swamp woodlands, which occupy permanent shallow swamp habitats. There are two types of woodlands characterized by the predominant tree species. One is the Melaleuca woodland that is composed almost entirely of Melaleuca cajuputi and the other is the N. orientale woodland. Both woodland types were the distinguishing features of the Middle Fly from the Kwem/Mipan/Manda area and Bosset. Associated species in the woodland included water lilies Nelumbo nucifera, Nymphaea macrosperma; fully submerged species including Hydrilla sp.; and the free-ﬂoating fern, Azolla pinnata. In the last 10 years, these woodlands either died or are in the process of dying. P. karka covers thousands of square kilometers of the Middle Fly as well as the area below the Everill Junction. S. robustum that prefers coarse sediment and well-regulated waters is restricted to the riverbanks and around some of the lakes and oxbow tie channels particularly around Manda, Bosset, Aiambak, and Kaviananga and up the Strickland River to Levame and upstream. Small aquatic, erect, prostrate, or creeping herbaceous plants are also an important component of the open grassed wetland areas of the lower Middle Fly. They grow on the lake margins, riverbanks, or among the grasses on ﬂoating islands of vegetation. The predominant species belong to two genera: Polygonum and Ludwigia. Some of them display specially adapted structures for ﬂoating, for example, Ludwigia adscendens, which can be rooted in mud or ﬂoating in water, with conspicuous, thick white buoyant pneumatophores

Figure 15.9: Echinochloa praestans and Hymenachne acutigluma in Middle Fly ﬂoodplain.

526

M. T. Rau and D. P. Reagan

that arise in clusters at the nodes (Leach and Osborne, 1985). An important semiaquatic plant is kangkong or Ipomoea aquatica that the river communities consume as a green vegetable. Sedges (Cyperus spp.) are also common plant group in the ﬂoodplain, and around lakes. Communities make baskets and mats from the Cyperus stalks. An important feature of the area around Mipan, Manda, and Bosset is the open grassed ﬂoodplain, which appears to be dominated by strictly aquatic plant species H. acutigluma, L. hexandra, Scirpus grossus, and E. sphacelata. These species are indicative of an area that is under water all or most of the year. The predominance of these species indicates a deeper depression in the topography of the Middle Fly in this area. Downriver from here, there is a slight change in vegetation from around and before Lake Pangua down to Obo and below. P. karka predominates in this area, reﬂecting comparatively less ﬂooding and/or shallower ﬂood levels for relatively shorter periods. A checklist of aquatic plant species from the Middle Fly is presented in Table 15.1. Observations in February and March 2002 showed that the forest along the banks of the Fly River at ARM 346 was undergoing severe dieback. Some of the large trees on both sides of the banks had died due to severe ﬂoods during the 1999–2001 La Nina period. The riverbanks at ARM 346 are actually higher than the level of the river, but the open forest behind this high ground was lower and thus was inundated. Further downriver from ARM 346, the effect of dieback was observed to be severe although there was evidence of slow recovery. The open areas created by the dieback had created suitable habitats for the invasive rattan (Calamus spp.), which vigorously suppresses regeneration. The species recorded from around ARM 346 include Terminalia sp., Macaranga aleuritoides, Althoﬁa sp., Timonius sp., Leea indica, Hibiscus tiliaceus, Gluta sp., Anthocephalus chinensis, Glochidion sp., Ficus nodosa, Syzygium sp., Alstonia scholaris, and Rhus sp. Other shrubs, vines, and palms included Deris elagens, Calamus sp., Pandanus sp., Bambusa sp., C. rumphiana (palm), P. karka, Cocos nucifera (coconut), Desmodium sp., D. cannaeformis, Calamus hollrungii, Mucuna sp., Hornstedtia sp., Platycerium sp., and Phrynium sp. Studies carried out by Kiapranis et al. (2001) grouped the Middle Fly wetlands into three main plant communities: (a) Obo Floodplain, (b) Bosset Floodplain, and (c) Oxbow 6 Swamp Forest Lake. These three communities were differentiated by the species composition and the levels of ﬂooding observed. The Obo Floodplain is S1 and the Bosset Floodplain is S2, while the S3 plant community included the Oxbow 6 Swamp Forest Lake and adjacent main Fly River channel.

Vegetation of the Ok Tedi–Fly River System

527

Table 15.1: A checklist of all aquatic species recorded for the Middle Fly Botanical Survey and their presence and absence within each of three vegetation communities. Family Araceae Azollaceae Ceratophyllaceae Convolvulaceae Cyperaceae

Eriocaulaceae Flagellariaceae Hydrocharitaceae

Lentibulariaceae Menyanthaceae Najadaceae Nymphaeaceae

Onagraceae Parkeriaceae Poaceae

Name

S1

S2

S3

Pistia stratiotes L. Azolla pinnata R. Brown Ceratophyllum demersum L. Ipomoea aquatica Forsk. Actinoscirpus grossus (L. f.) Goetgh. & Simpson Cyperus bifax C. B. Clarke C. brevifolius (Rottb.) Hassk. C. diffusus Vahl C. odoratus L. Fimbristylis dichotoma (L.) Vahl F. globosa (Retz.) Kunth Fuirena ciliaris (L.) Roxb. Schoenus sp. Scleria ciliaris Nees Eriocaulon australe R. Brown Hanguana malayana (Jack) Merr. Hydrilla sp. Hydrilla verticillata (L.f.) Royle Hydrocharis dubia (Bl.) Backer Utricularia sp. Nymphoides indica (L.) Kuntze Najas indica (Willd.) Cham. Nelumbo nucifera Gaertn. Nymphaea nouchali Burm. f. N. pubescens Willd. Ludwigia adscendens (L.) Hara L. hyssopifolia (G. Don) Exell Ceratopteris thalictroides (L.) Brongn. Echinochloa stagnina (Retz.) P. Beauv. E. praestans Hymenachne amplexicaulis (Rudge) Nees H. acutigluma Isachne globosa (Thunb.) Kuntze Ischaemum polystachyum L. Leersia hexandra Sw. Ophiuros tongcalingii (Elmer) Henrard

P P P P P

a P P P a

a a a a a

P P P P a a P P P a P a P P P P a P P P P P P P P P

a P P a P P P P P P P a a P a P a P P a P P a P P P

P a a a a a p a a a a p a a a a P a a a a P a P P

P P P P a

P P P P P

P a P P a

528

M. T. Rau and D. P. Reagan

Table 15.1: (Continued ). Family

Polygonaceae Potamogetonaceae Psilotaceae Scrophulariaceae

Name

S1

S2

S3

Oryza ruﬁpogon Phragmites karka (Retz.) Trin. Saccharum robustum E. W. Brandes & Jeswiet ex Grassl Setaria pallide-fusca (Schumach.) Stapf & C. E. Hubbard Polygonum attenuatum R. Brown P. barbatum L. Potamogeton sp. Psilotum sp. Limnophila aromatica (Lamk) Merr. L. indica (L.) Druce

P P P

P P P

a P P

a

P

a

P P P P P P

P a P a a P

a a a a a a

Notes: S1 ¼ Obo Floodplain, S2 ¼ Bosset Floodplain, S3 ¼ Oxbow 6, P ¼ present, a ¼ absent.

Table 15.2: Number of plant families and species collected from each study site and the number of aquatic species. Site Obo–Daviambu–Pangua (S1) Bosset Lagoon/ Floodplain (S2) Oxbow 6 & ARM 346 (S3)

Number of families

Number of Number of strictly aquatic species species

33

58

32

24

48

31

29

44

14

The total number of species recorded for each vegetation community is given in Table 15.2. These species numbers did not reﬂect the full species diversity and abundance at each site because the collection was limited by accessibility to the sites. In this study, the highest number of species sampled and recorded was from the Obo Floodplain followed by Bosset Floodplain and Oxbow 6 with 58, 48, and 44, respectively. The numbers for Obo could be biased as more areas and time was spent sampling around the Obo Floodplain than in the other two vegetation communities.

Vegetation of the Ok Tedi–Fly River System

529

Forty-six species were collected from all three communities: 12 species (26%) were collected only from the Obo Floodplain, 8 species (17%) from the Bosset Floodplain, and only 2 species (4%) recorded from Oxbow 6 Swamp Forest. In terms of shared species, 8 (19.6%) were present in all three sites and 16 (35%) were present only within the Obo Floodplain and the Bosset Floodplain. There was no shared species exclusively between Oxbow 6 Swamp Forest and Bosset Floodplain or between Oxbow 6 Swamp Forest and Obo Floodplain. Obo and Bosset therefore share similar vegetation and both are open grassed ﬂoodplains. Oxbow 6 has less aquatic grass species reﬂecting the drier and higher elevation, which favored drier forest vegetation. The species recorded from the Obo Floodplain and the Bosset Floodplain represented about 52% of the species recorded and also the most common species. The most common and abundant species were P. karka, S. robustum, N. nucifera, L. hexandra, Hymenachne amplexicaulis, Polygonum attenuatum, L. adscendens, Eleocharis stagnina, and Ischaemum polystachyum. This is a mixture of strict aquatic and semiaquatic species, reﬂecting a permanently inundated environment. 15.2.4. South Fly The South Fly region of the Fly River Basin extends from below Everill Junction, down to Sturt Island, further south to Tapila and into the Fly River delta. The vegetation in this area consists of four main types: gallery forest, lowland rainforest, open Eucalyptus and Acacia grassland savannahs, and open wetlands. The open wetland areas, which are an extension of the Middle Fly ﬂoodplains, stretch downriver past Everill Junction to Ogwa. The plant species distribution here is similar to that described for the wetland of the Middle Fly above. Large areas of sago on the outer margin of the ﬂoodplain between Ogwa and Suki had died out after the El Nino bushﬁres in 1997 and the La Nina ﬂoods of 1999–2001. Downstream of Ogwa the climatic conditions and a gradual change in topography resulted in a drier and higher elevation. The vegetation here is a mixture of gallery forests, lowland rain forest, and open Eucalyptus and Acacia grassland savannahs typical of those found around the Morehead District to the west. Other common savannah species in the area include Banksia dentata, Grevillea spp. (Foreman, 1995), A. scholaris, and Garuga ﬂoribunda. Further down toward Sturt Island and Tapila are found some of the most luxuriant virgin lowland tropical rainforest stands with closed canopy attaining more than 40 m. Very few studies have been conducted in this area of the Trans-Fly River Basin.

530

M. T. Rau and D. P. Reagan

Many changes in local relief occur halfway between Everill Junction and Suki. The landscape is dotted with small forest patches, high dry areas where kunai grass (Imperata cylindrica) predominates, with a mixture of P. karka, sedges, and other grass species. The presence of large grazing rusa deer herds (Rusa timorensis) indicates a less wet area. Grazing by the rusa deer could alter the plant species composition on the open ﬂoodplains and the undergrowth by altering the water table level. The small shallow rooting species could be excluded by a drop in water table due to increased rate of evaporation as a result of grazing and replaced by larger deeper rooting tree species. Further south toward Sapuka and Sturt Island, some canopy tree species of the rain forests include genera Diospyros, Horsﬁeldia, Calophyllum, Maniltoa, Citronella, Garcinia, and the palm C. rumphiana. The ﬂoodplain on the eastern bank of the Fly River below the conﬂuence with the Strickland River toward Sturt Island illustrates the recent signiﬁcant changes in ﬂood levels and surface ﬂood ﬂow routes. Forest trees are dying out and sago palms appear to be invading large areas. Some areas are bare or have very little plant cover. These contain stagnant water. Patches of Melaleuca spp. are common. Large areas are also covered by P. karka. Hunters from Suki and Balimo hunt for deer, wild pigs, wallabies, cassowaries, and other game in this area, which is otherwise uninhabited. Further downstream, large lakes and sago swamps are also common away from the riverbanks. These extensive herbaceous swamp vegetation communities have been described by Paijmans (1976). Key species that characterize these vegetation communities include the large lilly Hanguana malayana, sedges Thoracostachyum sumatranum and Scleria sp., as well as the fern Cyclosorus sp. Recent studies by Bito (2004) in the Morehead District to the west of Middle Fly and Suki recorded 240 species of plants, indicating the rich and varied plant communities found in Western Province. The vegetation in this region is characteristic of dry and wet climate similar to the savannahs and gallery forests that occur to the west of Suki and further south at Wipim and the Oriomo Plateau. Enormous herds of rusa deer have been present since the 1960s and probably longer; their impact on the plant biology of the area is an interesting aspect for future studies. 15.2.5. Fly River Delta and Estuary The Fly River delta is characterized by constantly changing sandbars, mudﬂats, and mangrove-covered islands. Downstream from Sturt Island, stands of Pandanus and the mangrove species Sonneratia lanceolata line the

Vegetation of the Ok Tedi–Fly River System

531

Figure 15.10: Pure stand of Sonneratia lanceolata on an accreting bank on the Fly River, between Sturt Island and Sapuka village. ever-widening accreting banks (Fig. 15.10). The presence of S. lanceolata indicates the tidal inﬂuence, but the Pandanus and other herbaceous species (e.g., Cyperacea) indicate freshwater dominance and very low salinities (around 0–11 parts per thousand) most of the year (Rau measured salinity). The mangrove species numbers increase with increasing salinity toward the sea, where mangroves dominate low-lying land in the vast 7,000 km2 of the Fly River estuary. Mangroves cover in excess of half a million hectares (Tomlinson, 1986) in PNG. The exact total area is difﬁcult to determine due to lack of accurate maps and the difﬁculty in distinguishing where the Nypa swamps (tidal inﬂuence) end and sago/freshwater swamps begin in areas such as the enormous, complex wetland systems of the Gulf of Papua. Most of the mangrove forests in PNG are intact; their use by village communities is on a very small local scale. To date, little research has been done on the ecology, distribution, and the different types of the mangrove communities. A total of 28 mangrove species found in the Fly River estuary and nearby coastal areas are listed in Table 15.3. A number of speciﬁc studies were carried out in the Gulf of Papua; for example, the Purari Hydroelectricity Scheme Study publication series in the late 1970s, and several studies carried out by a number of people from the Australian Institute of Marine Sciences and UNESCO Project on Management and Mangrove Research in the Asia Paciﬁc Region (1986). The taxonomy of the common key species is relatively well studied and several books were published (Frodin et al., 1975; Percival and Womersley, 1975; Frodin and Leach, 1982).

532

M. T. Rau and D. P. Reagan

Table 15.3: Common and important mangrove species of the Fly River. Family Acanthaceae Pteridaceae Plumbaginaceae Myrsinaceae Avicenniaceae Bombacaceae Caesalpiniaceae Rhizophoraceae

Euphorbiaceae Sterculiaceae Combretaceae Palmae Rutaceae Myrtaceae Rubiaceae Sonneratiaceae

Meliaceae

Species Acanthus ilicifolius Acrostichum aureum Aegialitis annulata Aegiceras corniculatum Avicennia ofﬁcinalis A. marina Camptostemon schultzii Cynometra ramiﬂora Bruguiera gymnorrhiza B. parviﬂora B. sexangula Ceriops decandra C. tagal Rhizophora apicualata R. mucronata R. stylosa Excoecaria agallocha Heritiera littoralis Lumnitzera racemosa Nypa fruticans Merope angulata Osbornia octodonta Scyphiphora hydrophyllacea Sonneratia alba S. lanceolata S. ovata Xylocarpus granatum X. mollucensis

Productivity studies of Rhizophora apiculata and Rhizophora stylosa through litterfall analysis were carried out by Leach and Burgin (1985) in Central Province. Johnstone published a number of his studies on the zonation patterns and herbivory on the Central Province mangroves (Johnstone, 1981, 1983). Several good publications and data on mangrove ﬁsheries exist in the Department of Fisheries and Marine Resources, but most of it is unpublished. There are, however, a few existing published accounts of mangrove communities of the Fly River delta. These include general descriptions and taxonomic references by Percival and Womersley (1975), Paijmans (1976),

Vegetation of the Ok Tedi–Fly River System

533

and more recently Robertson et al. (1993). Environmental monitoring by Ok Tedi Mining Limited (OTML) has provided good information on the aquatic fauna, chemistry, geochemistry, and hydrology of the estuary, which can be found in the OTML Environment Department annual reports from 1984 to present. Signiﬁcant studies were undertaken by Wolanski et al. (1996) of the sediment input and output of the Fly River estuary in the 1990s for the OTML Environment Department and the reports are archived in the OTML Environment Department library. The importance of the Gulf and Fly River mangrove ecosystems to the adjacent coastal waters is discussed in several papers by a number of authors in recent years. Alongi (1990), Alongi et al. (1992, 1993), Aller and Aller (2004), Robertson et al. (1993), and Kristensen et al. (2008) discuss the fate of organic carbon, detritus, nutrient regeneration and transport processes, and sediment proﬁles of the Gulf of Papua, the Fly River delta (see also Harris et al., 1992), Coral Sea, and the Great Barrier Reef.

15.2.5.1. Mangrove species distribution This section describes the occurrence of mangroves from its farthest upstream limit at Suki down to the seaward margin at Umuda Island (Fig. 15.11). An attempt is also made to describe the factors that inﬂuence the mangrove species distribution in the Fly River estuary. Information has been obtained from the studies mentioned above and from ﬁrsthand observations during OTML Community Relations patrols up and down the river since 2002 and from satellite imagery.

1

2

3

UPSTREAM SUKI, SIALOWA

4 MIDDLE OF ESTUARY WABIURA, KIWAI

5

6

7

8

9

OUTER MARGINS OF ESTUARY PURUTU PASSAGE, TOROPASS, PARAMA, UMUDA

SPECIES 1. Sonneratia lanceolata

4. Bruguiera parviflora

7. Rhizophora stylosa

2. Nypa fruticans

5. Bruguiera gymnorrhiza

8. Avicennia marina

3. Rhizophora apiculata

6. Sonneratia alba

9. Aegialites annulata

Figure 15.11: Typical mangrove distribution proﬁles in the Fly River estuary from its farthest point upstream at Suki to its outermost seaward margin, at Umuda Island. Only the key predominant species are shown.

534

M. T. Rau and D. P. Reagan

The Fly River and other large rivers in PNG transport huge quantities of ﬁne sediment and freshwater into the Gulf of Papua. The Fly River estuary supports an estimated 80,000 ha of mangroves (Robertson et al., 1993). Mangroves occur on the deltaic islands forming the main vegetation system of the islands and in low-lying areas along the main riverbanks. Island sizes vary from a few hectares to the biggest, the Kiwai, which is about 45-km long and 7-km wide. The riverbanks on both sides do not have large mangrove areas because these are largely high above the extreme high watermark and are occupied by lowland rainforest. Overall, mangrove vegetation covers a smaller area than expected for a large dynamic estuary. Upstream of the estuary small stands of S. lanceolata can be found in narrow strips up to around 200-m wide on ﬁne sand or mudﬂats on accreting banks. Examples of these mangrove fringes can be seen near O’uwo village, on the northwestern quadrant of Kiwai Island, where sedge species (family Cyperaceae) form an outer fringe in front of the S. lanceolata, while on the older and relatively higher elevated ground Nypa fruticans forms a band behind the S. lanceolata (Fig. 15.11). As indicated previously, a total of 28 species are listed for the estuary, and this is far from complete (Table 15.3). This list is a result of recordings made during 2003 and 2004 OTML Community Relations village patrols in the area by the ﬁrst author. Additional information was from work done by Percival and Womersley (1975) and Robertson et al. (1993). Of interest is the collection and recording of the mangrove species Merope angulata from the estuary, which is a ﬁrst record for PNG. This species has only been recorded in Asia as far as Malaysia and Singapore, but not further east. The voucher specimens are in the Herbarium, Biology Department, University of Papua New Guinea. Toward the seaward side, Avicennia marina may be found on well-drained mud and sandﬂats on accreting banks (Fig. 15.11). A. marina is less common and is found only in the outer high salinity, more marine inﬂuenced, exposed parts of the islands. For example, it can be seen in small stands right at the seaward edge of Wabada Island near Gesoa village, exposed areas of Wapi Island, and the seaward tip of the western bank near Samari village and Parama Island. It forms pure well-spaced stands on the southwestern corner of Kiwai Island between Wapaura and Agobaro villages on sand. These Avicennia communities are stunted and are up to about 5–7 m in height. It is also common on Daru Island and westward toward the border with Papua Province of Indonesia along the mainland coast. On the exposed seaward side of Daru Island, Aegialitis annulata may be found. This is a species that occurs in high wind and wave action, sandy or rocky habitats similar to Daru Island, and the mainland and parts of Central Province coastline.

Vegetation of the Ok Tedi–Fly River System

535

The six main species occupying signiﬁcant parts of the islands are N. fruticans, R. apiculata, Bruguiera parviﬂora, Bruguiera gymnorrhiza, S. lanceolata, and Camptostemon schultzii (Fig. 15.12). The palm N. fruticans is a predominant species, covering the entire interior portions of the islands. Pure stands of Nypa allow little or no space both in and above ground for other species to colonize. Wabiura Island, which is one of the biggest islands, is almost 70% covered by mature stands of Nypa. A few species of creeping Ficus spp. and Finlaysonia sp. climb on the Nypa fronds, which can be as tall as 30 m. In the mature mangrove communities, the undergrowth is predominated by Derris sp. and Acanthus ilicifolius, which abound in slightly higher areas with the fern Acrostichum aureum. Behind the mangroves at the extreme high watermark, A. ilicifolius and A. aureum abound on crustacean (Thalassina anomala) mounds. There is a signiﬁcant number of small herbaceous terrestrial plant species that form part of the undergrowth, mainly in the extreme high water-mark areas. In the inner part of the estuary, where S. lanceolata is the dominant primary colonizer, a thick mat of Cyperacea spp. can be found growing among S. lanceolata. From a distance, it gives the impression of lawn grass.

9

8

15

6

7

SEAWARD SIDE Mixed high salinity, sandy, strong wave action tolerant species.

3

13

5

2

10

14

MIDDLE OF ISLAND

1

11

12

4

PROTECTED SIDE

Nypa dominated, high to extreme high tide, low salinity Species, common in protected middle part of the island.

Protected, low salinity species on low wave action sides of the island

SPECIES 1 Sonneratia lanceolata

6 Sonneratia alba

11 Bruguiera sexangula

2 Nypa frutican

7 Rhizophora stylosa

12 Camptostemon schultzii

3 Rhizophora apiculata

8 Avicennia marina

13 Ceriops spp.

4 Bruguiera parviflora

9 Aegialites annulata

14 Xylocarpus granatum

5 Bruguiera gymnorrhiza

10 Heritiera littoralis

15 Avicenniao officinalis

Figure 15.12: The ﬁgure shows the combination of mangrove species distribution patterns on the different islands in the delta. The distribution pattern at any one location is affected by the tidal regime, salinity range, sediment type, and the degree of exposure to the sea, currents, wind, and wave action. Only the large predominant key tree species are used to illustrate the pattern. This distribution pattern may be observed around the Wapi Island.

536

M. T. Rau and D. P. Reagan

The presence of this non-marine plant reﬂects the strong freshwater inﬂuence in this part of the delta. Relatively diverse mangrove communities are found on large wide sand and mudﬂats in the protected passage waterways, for example, at Purutu and the Wabada passages. These passages are deeper than 15 m and are very well regulated by the strong and very fast-ﬂowing incoming and outgoing tidal ﬂows. In such habitats, a zonation pattern can be seen from the edge toward the middle of the islands. At the water’s edge in newly accreting banks is a 100- to 200-m-wide, thick mixture of young Avicennia ofﬁcinalis and S. lanceolata, then Nypa with Heritiera littoralis, R. apiculata at the back. Behind this on higher ground is pure Nypa stand or a mixture of Nypa and other high-ground species like H. littoralis, Xylocarpus moluccensis, and tall Excoecaria agallocha. In older well-established steep eroding banks, R. apiculata is predominant; while on the accreting banks, A. marina, R. stylosa, R. apiculata, and Sonneratia alba may be found. In the Wabada passage, well-developed, 35-m-tall stands of C. schultzii mixed with B. parviﬂora, and the odd Xylocarpus granatum line the banks. Even though C. schultzii has been recorded from the Gulf of Papua in the Pie River estuary near Baimuru and Kikori (Womersley (1978); Rau, personal observation, 1986), concentrated stands, such as seen in the estuary of the Fly River, have not been described elsewhere and are presumed to be uncommon. The distribution of C. schultzii is restricted to only the southwest Papuan coast. Extensive mud and sandﬂats characterize the estuary. Their presence indicates the enormous amount of sediment transported by the river into the estuary. They are constantly changing in shape and size as they are subjected to the currents of the tidal and freshwater forces. The most important mangrove species growing on this sand and mudﬂats is S. lanceolata. It is the primary colonizer of these newly deposited sand and mudﬂats. Examples of these extensive S. lanceolata may be observed between Wabiura and Wabada Islands (Fig. 15.1). At low tide one of the largest sand banks, about 10-km long, is visible between Tapila and Lewada. Common epiphytes observed include the ant plant (genus Hydnophytum), several fern species (including the staghorn fern), and orchids.

15.2.5.2. Factors inﬂuencing species distribution Mangroves are found in all coastal and island provinces in PNG. Signiﬁcant areas occur mostly on sheltered shorelines, on deltas and estuaries of large rivers, although small stands may be found on most high islands (volcanic in

Vegetation of the Ok Tedi–Fly River System

537

origin). Most mangrove community types described by Thom (1982) occur in PNG reﬂecting the varied geomorphic, geophysical, and biological conditions under which they occur. Thom (1982) classiﬁed mangrove communities into tide dominated, river dominated, wave dominated, river and wave dominated, and others. The largest mangrove areas in PNG are found in the Gulf of Papua (about 300,000 ha), and the Murik Lake in the Sepik River estuary on the northwest coast. Signiﬁcant areas are also found in Central Province (Galley Reach approximately 14–20,000 ha) and Cloudy Bay in the Robinson River estuary (30,000 ha), Milne Bay, and Northern Province (Mambare Wetlands). Similar mangrove areas exist in West New Britain, New Ireland, and Manus Provinces. There is very little or no recorded data for most of these mangrove communities. East New Britain coast has very limited mangrove stands in small coves and bays presumably as a result of very high geological and uplifting activities. In the Fly River delta, the mangrove formations and the species composition reﬂect the interactions of the powerful currents created by tides, the freshwater ﬂowing seaward, the prevailing southeasterly winds that blow from May to September, and the northeasterly monsoon that brings the rains from December to April each year. They would fall into the ‘‘river- and wave-dominated’’ mangrove systems (Thom, 1982). The coastline experiences extreme offshore currents and wave action. Around the islands, there are few locations that can be considered truly sheltered from the winds, wave, and current action. Erosion and accretion rates are very high. The powerful inﬂuence of the tides is obvious with the presence of the lowsalinity-tolerant species S. lanceolata as far upriver as Suki Creek that is more than 200 km inland. In fact, the local inhabitants conﬁrmed that the tidal bores travel upriver as far as Suki Creek. The Fly River introduces about 6,000 m3/s of freshwater into the sea with freshwater-induced currents of up to about 0.1 m/s. In the lower reaches of the estuary under the inﬂuence of tides, currents are even stronger with peak tidal velocities of about 1.5 m/s (Wolanski et al., 1995). Tidal bores, brought about by contrasting tidal currents during spring tides, are a characteristic of the estuary particularly in its upper reaches (Wolanski et al., 1995). These powerful water movements also have a signiﬁcant impact on the distribution and the size of the mangrove forests along the riverbanks. Examples of this impact are portrayed around and upstream from Tapila, where the S. lanceolata appear stunted, with a uniform height of about 2 m with ﬂat canopies and small short branches. Substrate type is also a major determinant of the species distribution. Species that prefer sand mixed with silt are found in the outer seaward

538

M. T. Rau and D. P. Reagan

margin. A. marina, S. alba, and R. stylosa occur in this type of sediment, and this can be seen on the main Kiwai Island near Sagapari and Agobaro villages. The A. marina trees are stunted to about 5–7 m and their pneumatophores and stems are covered in oysters and barnacles. A species A. annulata, which is found in southern PNG, North Queensland, and Northern Territory coastlines, can only be found in high salinities of around 28–30 parts per thousand, in sandy and rocky substrates. This mangrove is one of the few species that inhabits high wave action coastlines. In the Fly River delta, it is found on the seaward side of the Daru Island. Species such as B. parviﬂora, R. apiculata, N. fruticans, S. lanceolata, and X. granatum prefer very ﬁne clay and silty sediment.

15.3. Plant Ecology Plants play a critical role in supporting the fauna and human communities and in the rehabilitation of damaged ecosystems of the Basin. The processes of ﬂowering and fruiting are fundamental aspects of plant reproduction essential for the recolonization of damaged areas and for providing important food resources to inhabitants of the Basin.

15.3.1. Dieback and Colonization Forest dieback from mining operations has altered the vegetation of large portions of the lower Ok Tedi and the Fly River ﬂoodplains (Fig. 15.5). Dredging has reduced downstream ﬂooding effects and resulted in colonization of once-ﬂooded areas of the lower Ok Tedi.

15.3.1.1. Forest dieback Vegetation dieback is a natural phenomenon that occurs in vegetation communities worldwide. It can be affected by extreme climatic conditions, pathogenic outbreaks, or human disturbance. The development and spread of mine-related dieback in the lower Ok Tedi and the Middle Fly has been reported and documented by various authors since its onset in 1991 (Rau, 1994; Duff, 1992; OTML Environment Department annual reports since 1992; Prendergast et al., 1996; Marshall and Rau, 1999; Piskaut et al., 2000).

Vegetation of the Ok Tedi–Fly River System

539

Dieback has affected the forested ﬂoodplains in the lower Ok Tedi and downriver to the upper Middle Fly to around Kwem village. In the affected areas, the dieback process has and continues to alter the forest structure and species composition. Although hydrological modeling has been used to predict the extent and recovery of the dieback (Carroll et al., 1999), there is little detailed information on the impact on vegetation communities, speciﬁc plant communities, species diversity, and rates of recovery.

15.3.1.2. Regeneration and colonization Regrowth of the dieback areas involves two basic strategies, depending on the sources of regenerating species. In the absence of further disturbances, affected areas recover to approximate their former condition. The ﬁrst type of regeneration occurs where plant stress has resulted in defoliation and death of some of the large forest trees and the species of plants least tolerant to low levels of oxygen in the soil. In this instance, most of the regeneration is regrowth from the initial plant population and, therefore, growth and plant cover is rapid. Many seeds germinate under large parent trees producing large numbers of seedlings, most of which usually die out due to their low tolerance of higher light intensity, and competition for space and nutrients from existing vegetation. The second type of regrowth occurs in severely stressed dieback areas where 95% or more of the initial vegetation is dead, or has been defoliated and the forest ﬂoor is totally covered by newly deposited sediment. The regeneration here then almost entirely comprises new plants transported as seeds, spores, or vegetative portions, from elsewhere. The initial growth rate and plant cover is slower in comparison to the ﬁrst type of regeneration. The rate and success of natural colonization by plants of the newly deposited mine-derived sediment on the banks of lower Ok Tedi depends on several key environmental factors. These include soil type and degree of saturation, nutrient content, soil surface temperature, rainfall, the type of dispersal agents (e.g., water, wind, animal), availability of parent plants, and the local topography of the exposed surface. The frequency and extent of overbank ﬂooding and associated sediment deposit also have a profound impact on the rate and success of regeneration. The monitoring of the dieback regeneration in seven permanent marked out transects since January 1994 showed a normal colonization development as observed elsewhere in PNG for similar vegetation types at the same altitude (Paijmans, 1976). Normal successional development can only occur if there is no interruption by ﬂoods and sedimentation. At any stage of development,

540

M. T. Rau and D. P. Reagan

seedlings can be damaged or uprooted by strong ﬂood ﬂow or a rise in the water table. A rise in the water table can kill all the larger tree saplings that have deeper rooting systems in comparison to the grasses and weeds that have shallow roots. Due to the apparently rapid response in regeneration when the ﬂood frequency drops and hence water table levels, reducing the aggraded riverbed level through dredging should result in more regeneration and a faster recovery to a climax forest. However, if the present conditions are left as they are, the long-term regeneration may develop into an open grassed ﬂoodplain similar to the vegetation type found in the Middle Fly ﬂoodplain around Manda, Bosset, Aiambak, and Obo. The OTML-operated dredge at Bige has been in operation since March 1997, and it has signiﬁcantly halted the aggradation and overbank ﬂooding in the lower Ok Tedi. The Ok Tedi banks have dried up now long enough to let the natural regeneration to proceed and this could be seen to be very successful immediately near to the dredge. Growth of recolonizing trees is rapid. In places, the Parasponia sp. reached about 3 m inside 15–18 months of self-establishing, and other trees have grown to about 20 m. Predominant secondary species include Glochidion spp., T. timon, B. javanica, Ficus adenosperma, Ficus wassa, A. falcataria, etc. New sago seedlings are also healthy and growing well. Homalanthus novoguineensis is almost absent from the regeneration areas. This exclusion seems to reﬂect this species inability to establish in alkaline soils, which is predominant in the mine-derived sediments. H. nova-guineensis is one of the most common secondary invading trees in lowland vegetation communities. It occurs throughout the region from Tabubil to Kiunga and the Middle Fly. Further upstream from the dredge, above Dome and Iogi, and downstream, around the Konkonda bend and Ieran villages, large areas have now 100% ground cover, with trees around 20- to 25-m tall. There is evidence of good faunal invasion as the secondary forest regeneration advances, habitats become more complex, and food sources increase in abundance. Initial groups include insects (small butterﬂies, ants, and beetles), reptiles (snakes), and birds. As discussed above, this fast growth rate also enables a very fast ground cover by creepers, and some ferns. The rate of the cover increases as more species invade and obviously the leaf litter and mychorrizal activity improves the soil with the increasing number of species with time. On ﬂat surfaces, the plant cover can be between 90% and 100% in 2 years. The cover rate increases with time due to a number of conditions: (1) self-seeding, (2) soil changes due to more plant growth, and (3) increase in viability as the

Vegetation of the Ok Tedi–Fly River System

541

mychorrizal effect increases with the increase in species diversity. Decomposition of dead trees and other dead plant materials in dieback areas increases organic input into the soil. In the region primary colonizers differ with altitude, although some secondary tree species appear to be common from about 2,000 m to the lowland rainforest at 20 m asl. In newly exposed surfaces such as the mine pit, along newly graded roadsides, and landslide surfaces to around lower Ok Tedi, green algae invade the soil surface before any higher plants establish. After about a year, mosses and lichens form the ﬁrst cover before ferns or ﬂowering plants establish. The mosses provide initial topsoil stability by binding the substrate over which they grow. This is only effective on the surface layers and is thus ineffective as a preventive measure against erosion or landslides. Higher ﬂowering plants provide a more permanent hold onto the soil due to deeper rooting system and reduce erosion. Common secondary species include Duabanga moluccanum, Neonauclea neonauclea, and species of Bischoﬁa, Homalanthus, Ficus, Macaranga, Piper, Glochidion, and Timonius. Herbaceous secondary colonizers Urticaceae, tree Piper spp., and Impatiens are common species along the road to the mine from Tabubil. The free-living orchid genus, Spathaglottis spp., adorns the steep dump slopes of the mountains and the highway right down to the lower Ok Tedi area. The most common are the lilac/deep purple and, to a lesser degree, the white variety. Flowers of the little herbaceous Impatiens sp. is an attractive colorful sight on the roadsides under the shades of larger trees hanging over the road in the higher altitudes. In the lower warmer areas, grass species are the predominant primary colonizers, while in the higher altitudes including Sisimakam, Haidawogam, Tabubil, Lukwi, and up to the mine, the ferns predominate as the ﬁrst colonizers. At 1,000 m above Finalbin and the mine area, the ferns Nephrolepis sp., Pteridium aquilinum, Sphaerostephanos unitus, Sphaerostephanos telefominicus, and Pteris orientalis are the major primary colonizers (Fig. 15.13). Mixed in with the ferns sometimes two grass species Imperata conferta and Eulalia leptostachys may be found. An ancient creeping plant genus, Gunnera spp., is also a common colonizer of new habitats above 1,000 m. This genus is represented by three species in the area. It is an important species as it ﬁxes nitrogen and grows well in these poor, highly leached alkaline, mainly rocky soils. The low-pH-tolerant grass species Isachne confusa is found in pools of acidic water runoff, wet acidic soil conditions. It was commonly observed in the western wall of the Ok Tedi Mine pit in waters with pH values as low as 2.71 (OTML, 1997).

542

M. T. Rau and D. P. Reagan

Figure 15.13: Fern invading an exposed rock surface along the TabubilKiunga highway near Migalsim village. There are also a number of shrubs, trees, and grasses that occur throughout the Basin, from the Ok Tedi Mine to the D’Albertis Junction and further downriver including Macaranga spp., A. falcataria, Glochidion spp., Parasponia sp., F. adernosperma, F. pungen, D. molucanum, H. novaguineensis, T. timon, Timonius densiﬂora, B. javanica. The grass species Paspalum conjugatum, Axonopus compressus, I. polystachyum are also widespread throughout the entire area as primary colonizers with the latter two more common in the lower altitude and warmer habitats. Landslides are a common feature in areas above Ningerum, as the altitude increases and the terrain becomes increasingly rough with narrow ridges and near vertical slopes (Fig. 15.14). The abrupt increase in rainfall with altitude is a major contributing factor to the high frequency of landslides. The sizes of the landslides range from a few tons to hundreds of thousands of tons and occasionally millions of tons. Hindenburg Wall standing close to 1-km high and 7-km long was a result of a massive landslide that occurred 8,800 years ago. In August 1989, 160 million tons of material slid into the Ok Tedi from the Northern dump next to the mine pit. The colonization rate of these slip surfaces depends on how stable they are. Some landslide surfaces continue to shed material as debris ﬂow. In general, it can be concluded that the steeper the slopes the slower the colonization rate. Establishment of seedlings is prevented by the continuous runoff and because the slopes cannot support the weight of the larger heavier trees. Most often only small herbaceous species, lichens, mosses, and ferns colonize and provide the main cover over many years.

Vegetation of the Ok Tedi–Fly River System

543

Figure 15.14: Rugged terrain near the mine lease area at Mount Fubilan. Note the area affected by landslides in the middle center of the photograph. 15.3.2. Flowering and Fruiting In PNG, two ﬂowering and fruiting peaks are observed per year. Some species will ﬂower and fruit during the two peaks. Ferns, fern allies, and other ﬂowering species produce fruits, spores, and ﬂowers throughout the year. First of the two is early in the year from February to April and the second late in the year from August to October. During the peak period, there is a notable increase in hay fever and other respiratory related illnesses due to high pollen load in the air. The fruits of the ﬁrst peak mature around April and May for grasses, and for trees around June and July. For the second ﬂowering peak, the fruits set in late September and reach maturity in December and January. Many big tree species shed leaves around September, October, and November. Examples of these include Terminalia spp., Pterocarpus indicus (rosewood), Ficus spp., and Acacia sp. Early ﬂowering period involves all grass species, ferns, some trees, orchids, and herbaceous species. Many ferns also produce sori and the spores mature around March and April. Grass species from high altitude to lowland areas produce ﬂowers from February to April. These include the common widespread grass species P. conjugatum, A. compressus, Digitaria spp., I. polystachyum, I. cylindrica, and all the aquatic wetland species in the Middle Fly ﬂoodplains listed above. The extensive P. karka ﬂower heads provide a gentle khaki or light brown appearance for many hundreds of square kilometers of the Middle Fly and upper South Fly ﬂoodplain in April,

544

M. T. Rau and D. P. Reagan

May, and June. In contrast, the large ﬂower heads of S. robustum are white and form narrow white strip along the Fly River banks where it is mainly distributed. The late-year ﬂowering period involves most of the large forest trees, orchids, and herbaceous species. At this time of the year, some trees lose their foliage too including members of Ficus spp., P. indicus, etc. At Tabubil and the surrounding areas, in species A. falcataria, Homalanthus spp., Macaranga spp., Glochidion spp., and Elaeocarpus spp., ﬂowering commences in August and September and their fruits mature around December and January. Many of the orchids also ﬂower around the same period and set fruit by November and December. The begonias also ﬂower around August and September, and by October and through to December the fruits mature and the seeds are dispersed by the wind and rain.

15.4. Summary and Conclusions Plants play several critical roles in all ecosystems of the Fly River Basin. They are the basic producers of energy and organic matter. Their roots hold the soil in place. Their fruits, leaves, and branches provide food for animals and human communities throughout the Basin. Large trees, lianas, shrubs, and tall grasses provide structural habitat for epiphytes and animals. Plants of all taxa are integral to ecosystem function and essential for the reestablishment of habitats affected by mining. The mangrove areas of the estuary and riverbanks of the lower Fly River perform all the functions of plants in other riparian habitats; in addition, they provide critical habitat for marine and semiaquatic species. Fish, crabs, and a variety of other invertebrates use mangrove root systems as nurseries, thus supporting marine ﬁsheries of the Gulf of Papua. Many bird species also feed and reproduce in mangrove habitat. Mangrove communities in the estuary are also important in stabilizing the shifting islands, sand and mudﬂats of the delta. They play a major role in slowing down the erosion and act as an effective barrier to the winds and wave action around the islands and the mainland coastline. The shifts in organic matter and nutrients from areas of dieback to the Gulf of Papua have not been investigated. Lack of essential nutrients due to these losses from riparian habitat are signiﬁcant, but colonization by a variety of plant species following the reduction of overbank ﬂooding is encouraging. However, the success of these mitigating actions on the vegetation of the riverine ecosystems will require monitoring and management for decades.

Vegetation of the Ok Tedi–Fly River System

545

REFERENCES Aller, J. Y., & Aller, R. C. (2004). Physical disturbance creates bacterial dominance of benthic biological communities in tropical deltaic environments of the Gulf of Papua. Continental Shelf Research, 24(19), 2395–2416. Alongi, D. M. (1990). Effect of mangrove detrital outwelling on nutrient regeneration and oxygen ﬂuxes in coastal sediments of central Great Barrier Reef lagoon. Estuarine, Coastal and Shelf Science, 31(5), 581–598. Alongi, D. M., Christoffersen, P., Tirendi, F., & Robertson, A. I. (1992). The inﬂuence of freshwater and material export on sedimentary facies and benthos processes within the Fly Delta and Adjacent Gulf of Papua. Continental Shelf Research, 12(2–3), 287–326. Alongi, D. M., Tirendi, F., & Christoffersen, P. (1993). Sedimentary proﬁles and sediment water solute exchange of ion and manganese in reef- and riverdominated shelf regions of the Coral Sea. Continental Shelf Research, 13(2–3), 287–305. Bito, B. (2004). Plant species diversity and forest resources of trees W10 cm dbh in Monsoon and Beach Ridge Forests in seventy 0.5 ha plots in Morehead District of Transﬂy, Western Province, Papua New Guinea. Scientiﬁc Report No. 1/9S0739.08. Report for WWF Transﬂy Project. Carroll, D. G., Marshall, A. R., & Moi, A. S. (1999). The results of dieback modelling on the Ok Tedi and Fly River Floodplains. Unpublished Internal Report by Ok Tedi Mining Limited, Tabubil, Papua New Guinea. Conn, B. J. (Ed.) (1995). Handbook of the Flora of Papua New Guinea. Melbourne University Press, Melbourne, Australia, Vol. III. Duff, G. A. (1992). Effects of sediment deposition on rain forest trees on the Ok Tedi: Final report to Ok Tedi Mining Limited. Environment Department, Ok Tedi Mining Limited, Tabubil, Western Province. Foreman, D. B. (1995). Proteaceae. In: B. J. Conn (Ed.). Handbook of the Flora of Papua New Guinea. Melbourne University Press, Melbourne, Australia, Vol. III. Frodin, D. G., & Leach, G. J. (1982). Mangroves of the Port Moresby Region. Biology Department Occasional Paper No. 7, University of Papaua New Guinea, Port Moresby, PNG. Frodin, D. G., Huxley, C. R., & Kirina, K. W. (1975). Mangroves of the Port Moresby Region. Department of Biology Occasional Paper No. 3, University of Papua New Guinea, Port Moresby, PNG. Harris, P. T., Baker, E. K., Cole, A. R., & Short, S. A. (1992). A preliminary study of sedimentation in the tidally dominated Fly River Delta, Gulf of Papua. Continental Shelf Research, 13(4), 441–472. Henty, E. F., & Pritchard, G. H. (1982). Weeds of New Guinea and their Control. Botany Bulletin No. 7, Published by Division of Botany, Department of Forests, Lae, Papua New Guinea. Johnstone, I. M. (1981). Consumption of leaves by herbivores in mixed mangrove stands. Biotropica, 13, 252–259.

546

M. T. Rau and D. P. Reagan

Johnstone, I. M. (1983). Succession in zoned mangrove communities: Where is the climax? In: H. J. Teas (Ed.). Biology and Ecology of Mangroves. Dr W. Junk Publishers, The Hague, pp. 131–140. Kiapranis, R., Rau, M., & Bau, B. (2001). A botanical survey of ﬂoodplain, lakes, lagoon and open swamp forests along the Middle Fly, Western Province, Papua New Guinea. Unpublished Report for Environment Department, Ok Tedi Mining Limited, Tabubil, Western Province. Kristensen, E., Bouillon, S., Dittmar, T., & Marchand, C. (2008). Organic carbon dynamics in mangrove ecosystems: A review. Aquatic Botany, 89, 155–185. Leach, G. J., & Burgin, S. (1985). Litter production and seasonality of mangroves in Papua New Guinea. Aquatic Botany, 23, 215–224. Leach, G. J., & Osborne, P. L. (1985). Freshwater plants of Papua New Guinea. The University of Papua New Guinea Press, Port Moresby, Papua New Guinea, 254 pp. Marshall, A. R., & Rau, M. T. (1999). Lower Ok Tedi and Middle Fly – Estimate of current vegetation dieback and classiﬁcation of ﬂoodplain vegetation. Unpublished Internal Report by Ok Tedi Mining Limited, Tabubil, Papua New Guinea. McAlpine, J. R., Keig, G., & Falls, R. (1983). Climate of Papua New Guinea. United Publishers Services, Ltd., Tokyo, Japan. OTML (1997). Annual environmental report. Unpublished Report by Environment Department, Ok Tedi Mining Limited to State of Papua New Guinea. Paijmans, K. (1976). New Guinea Vegetation. ANU Press, Canberra, Australia. Percival, M., & Womersley, J. S. (1975). Floristics and ecology of the mangrove vegetation of Papua New Guinea. Botany Bulletin No. 8. Department of Forests, Division of Botany, Lae, Papua New Guinea. Piskaut, P., Kiapranis, R., & Rau, M. T. (2000). The effect of dredging on vegetation recovery in dieback forests along the lower Ok Tedi River. Unpublished Internal Report for Environment Department, Ok Tedi Mining Limited, Tabubil, Western Province, Papua New Guinea. Prendergast, J. B., Rau, M. T., Moi, A. S., & Marshall, A. (1996). An evaluation of the effectiveness of mitigation works in reducing dieback on the Ok Tedi and Fly river ﬂoodplains. Ok Tedi Mining Limited, Western Province, Papua New Guinea. Rau, M. T. (1994). Overbank ﬂooding effects on vegetation in the Lower Ok Tedi. Report to the OTML Environment Department, Tabubil, Western Province, Papua New Guinea. Rau, M. T. (1996). Preliminary ﬂora survey of the proposed mitigating sites. In: Environmental, Financial and Risk Analysis of Various Dredging and Tailings Storage Schemes to Mitigate Mining Impacts in the Ok Tedi/Fly River System, 4: Report to the State of Papua New Guinea by Ok Tedi Mining Limited. ENV96-08. Robertson, A. I., Daniel, P. A., Dixon, P., & Alongi, D. M. (1993). Pelagic biological processes along a salinity gradient in the Fly Delta and adjacent river plume (Papua New Guinea). Continental Shelf Research, 13(2–3), 205–224.

Vegetation of the Ok Tedi–Fly River System

547

Smith, R. E. W., & Bakowa, K. A. (1994). Utilisation of ﬂoodplain water bodies by the ﬁshes of the Fly River, Papua New Guinea. Mitteilungen Societas Internationalis Limnologiae, 24, 187–196. Thom, B. G. (1982). Mangrove ecology: A geomorphological perspective. In: B. F. Clough (Ed.). Mangrove Ecosystems in Australia, Structure, Function, and Management. Australian National University Press, Canberra, Australia, pp. 3–17. Tomlinson, P. B. (1986). The botany of mangroves. Cambridge University Press, New York, USA; Melbourne, Australia. Wolanski, E., King, B., & Galloway, D. (1995). Dynamics of the turbidity maximum in the Fly River Estuary, Papua New Guinea. Estuarine, Coastal and Shelf Science, 40, 321–337. Wolanski, E., Galloway, D., & Spagnol, S. (1996). Field and model studies of the fate of mine-derived contaminants in the Fly River estuary. Unpublished Report by Australian Institute of Marine Science, Townsville to Ok Tedi Mining Limited, Tabubil, Papua New Guinea. Womersley, J. S. (Ed.) (1978). Handbook of the Flora of Papua New Guinea. Melbourne University Press, Melbourne, Australia, Vol. 1, 278 pp.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00416-3

Chapter 16

Fauna and Food Webs of the Fly River Basin Douglas P. Reagan1, and Monica T. Rau2 1

Doug Reagan & Associates, 350 Gordon Drive, Castle Rock, CO 80104, USA Community Relations Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, W.P. Papua New Guinea 2

16.1. Introduction After visiting New Guinea in 1858, Sir Alfred Russel Wallace described New Guinea as ‘‘ya country which contained more strange and new and beautiful objects than any other part of the globe’’ (Wallace, 1890). The statement is still true a century and a half later, though recent impacts are rapidly affecting the environment. South of the central cordillera, the Fly River Basin of New Guinea comprises one of the most diverse and distinctive assemblages of animals found anywhere on earth (Bourliere, 1983a, b; Sekhran and Miller, 1995; Fig. 16.1). Principal factors responsible for the present faunal composition of the lush ecosystems of the Fly River Basin are the prolonged isolation of Australia and New Guinea from other tropical areas of the world, the recent isolation of New Guinea from Australia, and the diversity of habitats in the region. High topographic diversity (sea level to over 4,000 m) inﬂuences climate, which in turn determines ﬂoral and faunal composition. Wallace’s Line forms the boundary separating the zoogeographical regions of Asia from Australia, New Guinea, and adjacent islands. The line runs through the Malay Archipelago, between Borneo and Sulawesi (Celebes); and through the Lombok Strait between Bali (in the west) and Lombok (in the east). Species common to Asia are found west of the line; to the east, Corresponding author. Tel.: 303-688-0754; Fax: 303-688-1716;

E-mail: [email protected] (D.P. Reagan).

550

D. P. Reagan and M. T. Rau

Figure 16.1: Map showing location of Fly River System in Western Province of Papua New Guinea. The region outlined with dots is the approximate boundary of the drainage basin.

Fauna and Food Webs of the Fly River Basin

551

mostly organisms related to Australian species. The line is named after Alfred Russel Wallace, who noticed the apparent dividing line during his travels through the East Indies in the 19th century. The underlying basis for this separation into different zoogeographic regions is the presence of a deep ocean trench that separated the land masses to the west from those to the east, even during periods of lower sea level, when many of the islands of the archipelago comprised much larger land masses. New Guinea was joined to Australia, and the islands of Java, Borneo, and Sumatra were joined to Asia, but the land masses east and west of the line remained separated for more than 50 million years. Though the distance between the islands of Bali and Lombok is only about 35 km, their faunas are strikingly dissimilar. Many birds, such as parrots, refuse to cross even the smallest stretches of open water. Many bat species have distributions that cross Wallace’s Line, but nonvolant species are usually limited to one side or the other. The farming practices and communities of humans, who have inhabited the Fly River Basin for millennia, are integrated into the web of the Basin’s ecosystems. During the past century, however, incursions by outsiders for purposes of mining, harvesting timber, and other exploitative activities have become the primary inﬂuences on the environment (Frodin and Grissitt, 1982; UNEP, 1995; CSIRO, 1996; Parametrix and URS Greiner Woodward Clyde, 1999). These human-induced disturbances have combined with droughts and ﬂooding precipitated by climatic ﬂuctuations such as El Nino Southern Oscillation (ENSO) to rapidly alter the structure and function of the terrestrial ecosystems of the region. Because of the Basin’s remote location and difﬁcult access, its fauna are poorly known. New species continue to be discovered, and records of even the commonest species are spotty (Beehler et al., 1986; Nightingale, 1992; Flannery, 1995). Knowledge of even the basic biology of many species can only be inferred from related species present elsewhere. This chapter provides a general overview of the animals present, their diverse roles in the food webs of ecosystems of the Fly River Basin, and a discussion of current conditions. It is intended to provide a basis for understanding current impacts and an appreciation for the ecological values to be preserved for future generations.

16.2. Habitats Much of the Fly River Basin is covered with forests, ranging from closed canopy montane forests at higher elevations in the Star Mountains of the

552

D. P. Reagan and M. T. Rau

Hindenburg Range through lower montane forest at middle elevations and foothill forest before descending to the swampy lowlands. Until recently, the lower two thirds of the Fly River Basin was covered with a mosaic of ﬂoodplain forest, eucalyptus and grassland savannahs (around Suki, Morehead, and Wipim areas; see Fig. 16.1), and gallery forest and herbaceous swampland interspersed with shallow oxbow lakes (10–20 m deep) and lagoons (typically up to 4 m deep) and constantly shifting channels. Extensive mangrove habitat occurs at the mouth of the river. In the past two decades, much of the forest has disappeared from the middle portion of the Basin. Mountains near the headwaters receive some of the highest annual rainfall on the planet, often in excess of 10 m (400 inches) a year. Consequently, the stream channels in the mountains and foothills have carved narrow, steepsided gorges. In the vast lowland areas, water spreads seasonally over broad areas of low topographic relief that extend to the coast and the Gulf of Papua. These are interlaced with a complex network of channels, rather than a single deﬁned estuary for each major river. This network and the areas of low relief create one of the world’s largest wetlands in which all but the highest land areas are ﬂooded for a portion of each year. Seasonal ﬂoods dictate that animals living in ﬂoodplain forest or swamp habitat must be able to swim, climb trees, or ﬂy to avoid drowning. This limits the distribution of land dwelling species which require closed canopy forest or dry land, including pademelons (Thylogale browni), rodents, and many rodent-like marsupials (Flannery, 1995). Such areas provide suitable habitat for highly mobile and semiaquatic species, such as pythons, lizards, and crocodiles. Extensive ﬂoodplains provide a broad interface between the terrestrial and aquatic ecosystems of the Ok Tedi/Fly River system, making it difﬁcult to delineate precise boundaries between these ecosystems. Numerous species are semiaquatic and/or feed across these ecosystem boundaries. Crocodiles, turtles, and monitor lizards are semiaquatic; they occur throughout the ﬂoodplain and at the edges of water bodies. Cassowaries can be seen foraging towards the landward edge of the mangroves, where they abut patches of lowland rainforest. The boundaries between aquatic and terrestrial ecosystems in the lower Fly River Basin are indistinct, and many species inhabit or use resources from both systems. The archer ﬁsh feeds on terrestrial insects, and other ﬁsh species forage on seeds and fruit in ﬂooded forest areas. Crocodiles prey on turtles, ﬁsh, wild pigs, and occasionally humans. Stilts and ibis feed along exposed ﬂats and shallow areas. Ducks dabble in the shallows. Brahminy kites and white-bellied sea-eagles feed on ﬁsh and small terrestrial fauna

Fauna and Food Webs of the Fly River Basin

553

(e.g., snakes, tortoises). Cormorants, the Australian pelican, darter, and herons feed on ﬁsh and other aquatic fauna. Wallabies and wild pigs (Sus scrofa) migrate to the dry open grassed ﬂood plains during the dryer months of the year, August to mid December and during ENSO in search of food and water. During the 1997–1999 ENSO event, many animal species, including wallabies, pigs, snakes, turtles, crocodiles, and cassowaries ﬂocked to the main river channel banks and lakes for water and to avoid the massive bush ﬁres, caused by lightening and humans, which ravaged the thousands of square kilometres of ﬂoodplains. Mangrove communities are important because of their contribution to biological diversity, their fundamental role in primary productivity, and because they provide food and habitat for a broad variety of birds. The roots and submerged areas of these habitats provide critical nursery areas for aquatic and coastal marine organisms. Wild pigs are also in abundance in the mangrove areas (Rau and Reagan, 2009). Cuscuses (Phalanger and Spilocuscus spp.), and the large fruit bats, commonly called ﬂying foxes, also feed on the mangrove species; Sonerratia lanceolata and Sonerratia alba ﬂowers and fruits. These bats are commonly found in mangrove areas during the ﬂowering and fruiting times.

16.3. Animal Groups Because of New Guinea’s long period of separation from Asia, the fauna of New Guinea, including those of the upper and middle Fly River Basin, differs from rain forests in Africa, Asia, and the Americas by lacking large mammalian predators (e.g., leopards, tigers, or jaguars). Also absent are monkeys. The largest species in the ecosystem are snakes (python) and birds (cassowaries); the latter can attain body weights of 60 kg and a stature of 1.5 m (5 feet) (Beehler et al., 1986). The diversity of vertebrate species of New Guinea is considered impoverished compared to those of larger continental areas (Bourliere, 1983a, b), but ranks high in numbers of unique endemic species. Insect diversity is high (Moulds, 2009), and like tropical rain forests worldwide, many, if not most of the species in New Guinea rain forests occur in the forest canopy (Sutton, 1989). The open grassed ﬂoodplains of the Middle Fly and downstream of the conﬂuence of the Strickland and Fly Rivers are teaming with abundant and highly diverse insect fauna, though few studies and collections of insects have been conducted in the area (Moulds, 2009).

554

D. P. Reagan and M. T. Rau

In regional terms the Fly River also supports a very diverse ﬁsh fauna (115 species), which is covered by Storey et al. (2009). 16.3.1. Mammals The mammalian fauna of New Guinea exhibits close afﬁnities to Australia and immediately adjacent islands. Intermittent contact with Australia over the last several million years ended over 10,000 years ago. These intermittent contacts permitted exchange of species in a series of invasions, but preserved the high endemism found in marsupials, murid rodents (family: Muridae, old world rats and mice), and bats. Over 200 species of mammals comprise a diverse assemblage of herbivores, omnivores, and insectivores. Flannery (1995) reports that that the wild dog (Canis familiaris) was introduced to New Guinea about 2,000 years ago, but humans ﬁll the role of large mammalian predators. Several mammal species are important in the diet of humans throughout the Basin. Humans hunt or trap ground-dwelling marsupials and hunt cuscus with bows and arrows. Wild pigs and rusa deer (Cervus timorensis) are particularly important in the diet of some communities, providing evidence for the close connection between ecosystem components of the Basin and the indigenous inhabitants. Wallabies, both grass and bush species, are important delicacies for many communities along the Fly River. Most or all of the mammals inhabiting the rain forests of Africa, South and Central America, India, and Southeast Asia are placental mammals. In New Guinea and its close neighbor Australia, most of the mammals, except bats, are marsupials (pouched mammals). This divergence came about as the result of the break-up of the supercontinent, Gondwanaland, isolating Australia and New Guinea from the land masses of Africa, Madagascar, New Zealand, Antarctica, and South America. The marsupials that occupied Australia before the separation continued to pursue their independent course of evolution. Many have developed into species that have undergone adaptive radiation to occupy many functional niches that are occupied by placental mammals in other tropical ecosystems. The largest ground mammals in the Fly river Basin are wild pigs and rusa deer, both introduced by humans, though the dates of introduction are the subject of much controversy (Flannery, 1995). Common house cats, which were brought by people into villages and urban areas, have a signiﬁcant impact on rodents, frogs and bird populations around inhabited areas. Dugongs are primarily herbivores and occur in the estuaries but concentrate more to the outer clearer coastal waters of the Western Province

Fauna and Food Webs of the Fly River Basin

555

in the coral reef and sea grass meadows. These large mammals are an important traditional food source for the coastal tribal communities. Studies undertaken by Helgen and Oliver (2004) for the World Wildlife Fund’s South Paciﬁc Program noted 72 species of terrestrial mammals, which include 34 bats, 17 rodents, 20 marsupials, and one monotreme in the Trans-Fly region. This study provides a thorough review of mammal work carried out in PNG and the Western Province, focusing on the Trans-Fly region. Monotremes (Order Monotremata) are primitive but highly specialized egg-laying mammals. The members of this group in New Guinea are spiny anteaters, or more speciﬁcally, echidnas. In the Fly River Basin they inhabit the mountainous forest region and lowlands, but are rare in most areas. Tachyglossus aculeatus in captivity have been observed consuming ants, but Zaglossus bruijnii, potentially occurring in the upper Fly Basin, is excellently equipped to eat worms, which presumably comprise most of its diet. This species is the largest living monotreme and attains a body weight of up to 16.5 kg (about 36 pounds). The distribution and natural history of both species in New Guinea are poorly known (Flannery, 1995). Marsupials occupy most of the terrestrial niches in the Basin’s terrestrial ecosystems. Bandicoots (Microperoryctes spp., Peroryctes spp.) are small to medium-sized marsupials occurring primarily in forested areas. They are generally insectivorous to omnivorous, but their habits are poorly known. Cuscuses (family: Phalangeridae) are another important group of arboreal animals. These medium to large sized possums are more agile than the tree kangaroos. They posses long prehensile tails, which are completely naked on the terminal half, except for the ringtail possums, Pseudocheirus species. They use their tails for climbing and can move rapidly through the branches and foliage in search of leaves, bark of trees (Menzies, 1991), ﬂowers and fruits or to escape predators. The southern common cuscus (Phalanger intercastellanus) is common in the lowlands of southern New Guinea, including the lower Fly River Basin. The species, as other members of the genus, is herbivorous, eating ﬂowers, fruits, and leaves. Menzies (1991) also reported that they feed on lianas, ferns, orchids, and other epiphytic plants. Cuscuses range from the upper catchment of the Basin to the delta and the lowland coastal habitats. Species of cuscus and bandicoots are important food items in the diet of the inhabitants of many communities in the Basin. Perhaps the strangest and most curiously adapted of all mammals in New Guinea are the tree kangaroos. Though uncommon and susceptible to hunting pressure, tree kangaroos occur in the forest of the Fly River Basin (Flannery, 1995). All tree kangaroos belong to the genus Dendrolagus,

556

D. P. Reagan and M. T. Rau

represented by two species in Australia and ten in New Guinea (Martin, 2005). Their hind limbs are short and broad with granulated soles and sharp curved claws. The forelimbs are proportionately larger and more powerful than any other macropods. The tail is long and nearly cylindrical and apparently functions primarily for balance, as it is not prehensile. These adaptations allow them to climb into the forest canopy and thus gain access to the leaves and fruits of rain forest trees. As a result of their ‘‘redirected’’ evolution, their means of locomotion is curious combination of climbing, walking, and hopping. In trees they climb and walk on branches and are the only macropods (kangaroos and their close relatives) able to walk backwards. On the ground, however, they bound along on their hind legs. The leaves of many rain forest trees are difﬁcult to digest, so tree kangaroos have large distended stomachs and a pot-bellied appearance similar to that of many leaf-eating monkeys. Food remains in the digestive system for long periods, allowing bacteria to break down the cellulose and obtain more nutrients (Nightingale, 1992). Many of the placental mammals of the Fly River Basin are small to medium-sized rodents. The murid fauna of New Guinea is abundant and diverse and one of the most species-rich for a land mass of its size of any place in the world (Flannery, 1995). It includes many insectivorous species, leaf eaters, as well as tree mice and typical old world rats. The common water rat (Hydromys chrysogaster) inhabits a variety of habitats, always near water and including mangrove areas of the Fly River estuary. It is widespread and common from sea level to montane forest. Other species occurring in the basin include the white-bellied melomys (Melomys leucogaster) in forest and forest patches and the cane ﬁeld rat (Rattus sordidus) found in savannah habitat and rain forest margins. Bats are diverse and include species that feed on insects, fruit, and nectar (Flannery, 1995; Seri and Rau, 1996). The large fruit bats or ﬂying foxes, as they are sometimes called, belong to the order Chiroptera, suborder Megachiroptera. In New Guinea, the family Pteropidae contains just 20 species, of which there are few records in the Fly River Basin. One such species, the greater ﬂying fox (Pteropus neohebernicus), has one of the largest wingspans of any bat – nearly 1.5 m (almost 5 feet). In densely populated areas like the lower Ok Tedi region, larger mammals and marsupials are over-hunted, and the most common mammals are bats and wild pigs. With parts of the ﬂoodplain forest converted to early successional grasslands due to dieback (Rau and Reagan, 2009), the fruit-eating and nectar-feeding bats have apparently migrated elsewhere. A few mammal species are entirely restricted to foothill forest, but other species also occur in the adjacent montane zone that extends into the lowland

Fauna and Food Webs of the Fly River Basin

557

alluvial plain forest (Flannery, 1995). Lowland alluvial ﬂoodplain forests support a limited endemic mammal fauna. Only the gray dorcopsis (Dorcopsis luctuosa), a few murid rats, and some bats are found in this habitat type (Flannery, 1995). Savannah habitat along the lower Fly River supports a diverse and abundant assemblage of mammals including wallaby, wild pig, and rusa deer. The delicate mouse (Pseudomys delicatulus) and brush-tailed rabbit rat (Conilurus penicillatus) are restricted to this habitat. Both are reported from the Trans-Fly plains (Flannery, 1995). 16.3.2. Birds New Guinea hosts about 650 bird species of which 568 breed there (Rand and Gilliard, 1968). Distinctive groups include bowerbirds, cassowaries, and more than three dozen species of birds of paradise. Both the dwarf cassowary (Casuarius bennetti) (Fig. 16.2) and southern cassowary (Casuarius casuarius) occur in the Basin. The southern cassowary is among the largest terrestrial animals in New Guinea. Standing up to 1.5 m (5 feet) tall, these birds inhabit lowlands and swamp forests, and willingly swim (Beehler et al., 1986). Kingﬁshers (Alcedinidae), storks (Ciconiidae), herons and egrets (Ardeidae), Australian pelican (Pelecanidae), and cormorants (Phalacrocoracidae) feed primarily on ﬁsh and invertebrates in the aquatic ecosystem. Whimbrel (Numenius phaeopus), sandpipers (Calidris spp.), and other shorebirds obtain their food from shallow sediments in marshes, mudﬂats, and along beaches. The dry lowlands of the trans-Fly area, including a small section bordering the north bank of the lower Fly River, supports the most distinctive regional bird fauna in New Guinea (Beehler et al., 1986). Large expanses of savannah and grassland, and patches of lowland rain forest provide an important winter haven for Australian migrants and a few true endemics (e.g., the Fly River grassbird, Megalurus albolimbatus). The magpie goose (Anseranas semipalmata) migrates to New Guinea by the millions from Australia and occurs only in the extensive ﬂood plains of the Trans-Fly. Rain forest patches also contain isolated remnants of the forest fauna found farther inland. Several species of birds of paradise occur through much of the Basin. The superb bird of paradise (Lophorina superba) is found at higher elevations around Ok Tedi. The magniﬁcent bird of paradise (Cicinnurus magniﬁcus) is reportedly common in hill forest areas closer to the Fly River (Gregory, 1995) near Ningerum on the upper Ok Tedi and also up the Fly River near and above Kiunga town. Gregory also reported that the greater bird of

558

D. P. Reagan and M. T. Rau

Figure 16.2: Dwarf Cassowary (Casuarius bennetti) from the South Fly River Basin (Photo: C. Tenakanai). paradise (Paradisaea apoda) is the commonest bird of paradise species found up to 750 m elevation in the vicinity of Ok Tedi, Ok Ma, and Ok Menga and in the upper Fly River Basin. Raptors (e.g., eagles, hawks, and owls) range throughout the basin. Brahminy kites (Haliastur indus) and white-breasted sea-eagles (Haleaeetus leucogaster) are common along the river and major tributaries. The crested hawk (Aviceda cristata) is one of the commonest hawks at low and middle elevations. The gray (variable) goshawk (Accipiter novaehollandia) is common through much of the area, particularly in the vicinity of human habitation. Shorebirds and wading birds are abundant throughout the ﬂoodplain of the middle and lower Fly River. Plovers, whimbrels, snipe, stints, herons, egrets, and spoonbills occur along waterways over much of the basin (Beehler et al., 1986; Gregory, 1995; Halse et al., 1996). Berrypeckers, cuckoos, drongos, ducks, frogmouths, honeyeaters, kingﬁshers, manucodes, meliphagas, mynas, parrots, pittas, robins, starlings, swallows, swifts, and many other species of birds occur in the Basin. The magniﬁcent riﬂebird (Ptiloris magniﬁcus), Blyth’s hornbill (Rhyticeros plicatus), and eclectus parrot (Eclectus roratus) are strikingly beautiful species that exemplify the striking diversity and brilliant colors of birds in the forested areas of middle and low elevations (Gregory, 1995). The vulturine parrot (Psittrichas fulgidus) and black sicklebill (Epimachus fastuosus) occur in middle forest levels in tributary drainages of the Fly River. Both are found in the vicinity of the Ok Tedi Mine (Gregory, 1995).

Fauna and Food Webs of the Fly River Basin

559

The Fly River grassbird (Megalurus albolimbatus), occurs only in isolated portions of the middle and lower Fly River Basin. This species is listed as vulnerable in the IUCN Redlist of Threatened species (IUCN, 2007), but expansion of its preferred habitat may be beneﬁting the species. As the impacts of the Ok Tedi mine (sedimentation and ﬂood levels) in the Middle Fly are predicted to continue increasing and persist for at least a century after mine closure, survival of many bird species in portions of the lower Basin are in serious question. The forests of New Guinea are also rich in species of pigeons that feed on fruits and seeds of forest trees, vines, and other forest plants. The ornate fruit dove (Ptilinopus ornatus), superb fruit dove (Ptilinopus superbus), Zoe imperial pigeon (Ducula zoeae), and Papuan mountain pigeon (Gymnophaps albertisii) are just a few of the species found in forested areas at middle elevations in the upper Basin. These and other pigeon species play an important role as seed dispersers in tropical forests (Janzen, 1983). Crowned pigeons (Goura sp.) (Fig. 16.3) are very common in the southern Fly River region below the Strickland Fly junction. Many of these species appear to be moving onto higher grounds of the ﬂoodplains as ﬂoodwaters remain over much of the lowlands for longer periods than in the past. Most of the middle Fly River ﬂoodplain is now under permanent standing water. Many species of ﬁgs (Ficus spp.) which inhabited the river banks of the upper and lower Ok Tedi, as well as the upper Middle Fly below the D’albertis Junction were an important source of food for fruit-eating birds, mice, ﬁsh, ﬂying foxes, and people. Ficus pungens which produces proliﬁcally,

Figure 16.3: Victoria crowned pigeon (Goura victoria) (Photo: C. Tenakanai).

560

D. P. Reagan and M. T. Rau

thousands of small round fruits about 7 mm in diameter on stocks attached to the roots and up the main base of the tree trunk. When ripe the fruits are pinkish in color and attract large numbers of animals and insects. Several bird species are restricted to particular habitats. Bird species, such as the mangrove fantail (Rhipidura phasiana), broad-billed ﬂycatcher (Myiagra ruﬁcollis), and red-headed honeyeater (Myzomela erythrocephala) occur in mangrove habitat in the lower portion of the Fly River Basin. 16.3.3. Reptiles Reptiles are abundant throughout the Fly River Basin. Lizards and snakes occur in all terrestrial habitats. Pythons, boas, and poisonous snakes, such as the death adder, occur in both forest and savannah habitats (Mehrtens, 1987). Agamid lizards (family Agamidae) and goannas , a type of monitor lizard, (family Varanidae) are widely distributed in habitats along the lower Fly River. Crocodiles, turtles, and monitor lizards are semiaquatic; they occur throughout the ﬂoodplain and at the edges of water bodies. Allison (2005) provides a comprehensive review of the reptiles of the trans-Fly region. He lists 97 species including 9 turtles, 2 crocodiles, 48 lizards, and 38 snakes. Detailed discussion on the distribution of these species in New Guinea and Indo West Paciﬁc area and Australia are also provided. Two species of crocodiles also inhabit the Fly River Basin. While crocodiles are primarily aquatic species, they must come on shore to lay eggs, and feed on terrestrial species. Among the largest predators on the island, the species are considered components of both terrestrial and aquatic ecosystems. The New Guinea crocodile (Crocodylus novaeguineae) is a small to medium-sized species, attaining a maximum length of 3.5 m (about 12þfeet) and found extensively in swamps throughout the Island of New Guinea. It feeds on ﬁsh, water birds (rails and grebes), amphibians, and reptiles. Like other crocodilians, females construct a mound nest in which the eggs hatch. Both males and females have been observed opening the nest and moving the hatchlings in their mouths to nearby water. Large areas of wetland habitat and low human population have beneﬁted the species. Species management and monitoring programs have reversed the unsustainable harvesting trends of the 1950s and 1960s, and the species is currently not considered threatened (IUCN, 2007). The species is eaten by humans in some communities (Montague, 1984). The saltwater crocodile (Crocodylus porosus) or saltie, as it is called in Australia, is by weight the world’s largest living reptile species. A male 6 m

Fauna and Food Webs of the Fly River Basin

561

long can weigh over 1,000 kg. Adult males can measure up to 7 m (about 23 feet) in length. It and the Nile crocodile are the two species known to actively stalk and prey upon humans. Young salties consume a variety of small prey, and adults consume crabs, goannas (monitor lizards), snakes, shorebirds, wading birds, and occasionally wild pigs (Taylor, 1979). As its name implies, it inhabits saltwater and brackish water in the Fly River Estuary, but extends long distances upstream into freshwater reaches of the middle Fly and Strickland Rivers. Females construct nests on land in freshwater areas during the wet season, thus avoiding the hazard of ﬂooding. It is estimated that less than 1% of the hatchlings survive to maturity because of predation by other species. As in Australia, many of the snakes in New Guinea belong to the family Elapidae, which contains some of the most poisonous snake species in the world. These include members of the genera Demansia, Furina, Notechis, Pseudonaja, and Toxicocalamus (Mehrtens, 1987). The distribution of these species is poorly known. Many are presumed to occur in the Fly River Basin, and most are considered dangerous to humans. Many members of human communities along the river are killed every year by the three most venomous snakes found in Western Province: Death Adder (Acanthophis sp.), Papuan Black (Pseudechis sp.), and the Papuan Taipan (Oxyuranus scutellatus canni). The two latter species are more common in the open hot, grassland and savannah habitats while the death adder is relatively common in the cool leaf litter covered forest ﬂoors. Papuan Black and the Papuan Taipan are more commonly found in the open grassed ﬂoodplains of the Lower Middle Fly, down towards Suki and the open coastal habitats around Tapila, Balimo, and the Wipim area spreading westwards towards the Morehead District. Studies carried out by WWF in 2004 provide a more detailed species list of the Morehead area (Allison, 2005). Yet to be conﬁrmed, there may also an equivalent of the Australian brown. Pythons (Fig. 16.6) also inhabit the Fly River Basin. The green tree python (Chondropython viridis) is an arboreal species inhabiting mountain and ﬂoodplain rain forests of the Basin. The D’Albertis water python (Liasis albertisii) is a medium-sized python, known from the lowlands. Boelen’s python (Morelia boeleni) is a strikingly colored species, attaining a length of 2þm (6–8 feet). The upper part of the snake is shiny black with white diagonal marking on its belly and white bars on the lips. The species is endemic to New Guinea and occurs in midmontane rain forests of Papua New Guinea, favoring habitats with high humidity and low light conditions. It is the most protected reptile species in Papua New Guinea. Pythons are a delicacy routinely consumed by humans within the Basin, especially the Awins, Yongoms, and the Ningerum tribes of the Ok Tedi catchment.

562

D. P. Reagan and M. T. Rau

The New Guinea snapping turtle (Elseya novaeguineae) is a scavenger feeding on live and dead ﬁsh in lake and river habitat. The herbivorous pignosed turtle (Carettochelys insculpta) forages on aquatic plants in the Fly River and in similar habitat in Australia (Ernst, 1989; Cogger, 1994). It is a particular delicacy for the communities living in the Middle Fly, Lake Murray and to the east of the Fly River, the Gogodala tribe of the Aramia River system. All of the tortoise species are considered delicacies by all riverine communities. The sand banks of the Fly River and the banks of the major lakes used to provide nesting habitats for C. insculpta. With the river bank levies undergoing profound changes by the deposition of very ﬁne mine-derived sediments and subject to an increased frequency of ﬂooding, C. insculpta may be absent from the Middle and South Fly main channel habitats for a long time. Its eggs were also an important source of protein for the people. 16.3.4. Amphibians Amphibians are diverse in New Guinea, but only a few families are represented. There are no salamanders or caecilians on New Guinea, and tree frogs (Hylidae) dominate the fauna (Darlington, 1957). Tree frogs (family: Hylidae) occur worldwide, including Australia and New Guinea. True frogs (Ranidae) are present, as are old world tropical tree frogs (Microhylidae) (Zweifel, 1972). Narrow-mouthed toads (family: Microhylidae) likewise have a worldwide distribution that includes New Guinea. Two species in the mountains of New Guinea (Liopphryne shlaginhaufeni and Sphenophryne cornuta), like other microhylids in New Guinea, skip the tadpole stage and undergo direct development from egg to adult form. However, these species exhibit unusual parental behavior; the male frog guards the eggs then transports the young on his back. They drop individually as he moves across the forest ﬂoor at night. This behavior is thought to be an adaptation to disperse the young and increase their chances of survival by reducing competition and exposure to predation (Bickford, 2002). The distribution of these species in the Fly River Basin is unknown, but would likely be limited to the mountainous areas. A number of biologists have undertaken ecological and taxonomic studies of the herpetofauna and the amphibians of New Guinea (Alison et al., 1998; Gunther and Richards, 2000; Richards, 2001; Richards and Iskandar, 2000, 2001; Gunther et al., 2001) and most notably the work of Richards and Johnstone on the frogs of the Star Mountains and the Middle Fly in Western Province (Richards and Johnstone, 1993; Johnstone and Richards, 1994).

Fauna and Food Webs of the Fly River Basin

563

Figure 16.4: New frog species (Litoria sp.) from Tabubil (Photo: F. Tekei). A beautiful light green and brown species of frog, family Hylidae was found by Francis Tekei at Tabubil near the workers’ living quarters in 2007 (Fig. 16.4). Yet to be identiﬁed, it resembles a new species; Litoria spartacus described by Richards and Oliver (2006) which was found several hundred kilometres east of Tabubil at a similar habitat near the Moro Camp, at the base of Iagifu Ridge in the Southern Highlands Province. The family Bufonidae (true toads) has a worldwide distribution, except for Australia and New Guinea. However, the cane toad (Bufo marinus) (Fig. 16.5) has been introduced into the region and is considered a pest species. Cane toads were sighted around Tabubil in 2006, even though the species has been present in Daru and Kiunga for many years. The movement of containers and vehicles between Kiunga and Tabubil is the most probable means of the spread of this toad. Although the Morehead and Bensbach areas are far west of the Fly River Basin, located near the border of PNG and the Indonesian province of West Papua, it is important to note that recent studies and survey by Richards (2005) recorded 14 frogs and 15 reptile species. The species were typical of lowland savannah habitats showing assemblages of fauna with close afﬁnities to northern Australia with many species common to the Papua Province of Indonesia and the Torres Strait. The Bensbach and Morehead areas experience typical distinct dry and wet climatic conditions of the Papuan coastal areas, such as the coast of Central and Western Provinces. The vegetation resembles those of the Arnhem Land in the Far North Queensland and Kakadu in the Northern Territory, Australia.

564

D. P. Reagan and M. T. Rau

Figure 16.5: Cane toad (Bufo marinus) from Tabubil (Photo: Francis Tekei). 16.3.5. Invertebrates As in other tropical areas that receive high amounts of rainfall, vegetation is lush, and there is a high diversity of animals, especially insects. They occupy all strata in the forest and play signiﬁcant roles in the food web. Ants are common throughout the area, as are mosquitoes, which carry a variety of diseases that are harmful to humans including malaria, dengue fever, and Japanese encephalitis. Insects of the Basin are discussed in Moulds (2009). The highest diversity of arthropods, including insects, in tropical areas is generally considered to occur in forest canopies (Erwin, 1982; Rinker and Lowman, 2004). It can be logically assumed, therefore, that overall faunal diversity in the Fly River Basin has been adversely affected by loss of forest habitat. This reduction in species reduces the complexity which underpins the stability of affected ecosystems (Paine, 1966; Trombulack et al., 2004), though by how much, is almost impossible to determine. These changes are taking place over very short time frames. Invertebrates play a central role in the structure and function of terrestrial ecosystems throughout the Fly River Basin. The group includes predators (e.g., spiders), leaf eaters (e.g., Lepidoptera larvae), grazers (e.g., snails), pollinators (e.g., butterﬂies, bees), and detritus feeders (e.g., mites). Because much of the Fly River Basin is covered with forest, arboreal invertebrates are particularly important in the healthy functioning of the ecosystem. Large aggregations of spiders are common in disturbed areas along rightsof-way in the vicinity of Tabubil, along the Ok Tedi tributary of the

Fauna and Food Webs of the Fly River Basin

565

Fly River. This species constructs gigantic communal webs, which can exceed 60 m in length in the forest canopy and along power lines. These massive webs are home to thousands of individuals of all stages from eggs to adults. The invertebrate fauna of habitats of the Fly River basin are poorly known. Few scientists have had the opportunity to collect in the area, but what sampling has been done conﬁrms that the diversity is great. Many, if not most, of the species inhabiting tropical rain forests remain undescribed (Erwin, 1982; Hubbell, 2001). In the upper portions of the Basin, which do not ﬂood, soil organisms play an important role in decomposing dead organic matter and converting it to simpler forms that can be recycled into living matter. In middle and lower sections of the Basin, much of the area is ﬂooded for a portion of the year, and interactions between terrestrial and aquatic ecosystems are tied closely together. At higher altitudes, where the soil is saturated constantly, fungus and bacteria rapidly decompose dead plant and animal matter. Denser plant material (e.g., twigs, bark, branches, and logs) are broken down by millipedes, beetle larvae, and decomposers, including lichen and fungus. Constantly high humidity (80–90%) at this altitude provides ideal conditions for high microbial activity. Mangrove areas support a rich fauna including molluscs and crustaceans. Geloina coaxans, Batisa violacia, Anadara sp., and a number of oyster species are the predominant bivalves in the estuaries and mudﬂats. The predominant gastropods include Telescopium telescopium, Nerita spp., littorinids and several Onchidium spp. The most important commercial crustacean species are the mud crab Scylla serata, and the prawn species which form the commercial ﬁsheries for Western Province. The colorful little crabs of the genus Uca are widespread in the mud ﬂats and must certainly contribute to the detritus break down and the oxygenation of the mudﬂats with their burrows.

16.4. Food Webs and Functional Organization Food webs are the sum total of feeding relationships among all living things in an ecosystem. As such, they provide essential structural organization in ecosystems (Gallopin, 1972). Understanding basic structure is an important means of understanding current conditions and potential impacts of current changes in faunal assemblages and ecosystem function in the Fly River Basin. Though the Basin’s fauna is poorly known, one can still identify basic functional components of the Basin’s terrestrial ecosystem. Information on

566

D. P. Reagan and M. T. Rau

the species present in the region and on their habitats, food habits, and foraging strategies were obtained from Mackay (1976), Beehler et al. (1986), and Flannery (1995), conversations with regional experts, and direct observations. As with the terrestrial fauna having an indistinct boundary with aquatic fauna, the terrestrial food webs have an indistinct boundary with aquatic food webs. The latter are discussed in detail by Storey et al. (2009) and are not considered further here. All complete and functional food webs contain three trophic categories: Producers – organisms that manufacture food from inorganic compounds by photosynthesis or chemosynthesis (e.g., green plants); Consumers – organisms that ingest other organisms (e.g., animals that consume plants or other animals); and Decomposers – organisms that derive their nourishment from dead organic matter (e.g., scavengers, fungi, and bacteria). These categories are further subdivided into functional components on the basis of structure (e.g. trees and grasses) and for animals, where they obtain their food (e.g., on the ground or in trees). A generalized food web that depicts the interrelationships among functional groups is shown in Fig. 16.6. Species inhabiting the Fly River Basin that belong to each of the compartments presented in Fig. 16.6 are presented in Table 16.1. The arrows in this diagram indicate the direction of energy and nutrient ﬂows through the Arboreal Predators

Omnivores

Fruit & Seed Eaters

Herbaceous Plants Detritivores & Scavengers

Terrestrial Predators

Intermediate Predators

Leaf Eaters & Browsers

Shrubs & Small Trees Chemical Decomposers

Small Terrestrial Predators

Nectar Feeders

Canopy Trees Nutrients Detritus

Figure 16.6: Functional food web, showing the interrelationships among major compartments common to tropical rain forests. Representative species for the tropical forests of the Fly River Basin are listed in Table 16.1.

Fauna and Food Webs of the Fly River Basin

567

Table 16.1: Representative taxa for functional food web groups shown in Fig. 16.6. Functional group Producers Herbaceous plants & grasses (& creepers/climbers)?

Shrubs and small trees

Canopy trees

Representative taxa Cogon grass, Phragmites karka, wild sugar cane; Saccharum robustum, Imperata cylindrical, water lilies, Cyperaceae, Hymenachne acutigluma, Azolla pinnata, Ludwigia spp., creeping piper, Ageratum conizoides, creeping Convolvulaceae spp., Vigna sp., Compositae members, Laea indica Erect Piper spp., Homalanthus spp., Ficus adenosperma, F. pungen, F. wassa, Parasponia spp., Bischoﬁa spp., Timonius densifolius, Macaranga spp., Glochidion spp., Melaleuca spp., Nauclea orientale, Mallotus sp. Barringtonia spp., Abelmoschus sp. Artocarpus cummunis, Gnetum gnemon Octomeles sumatrana, Albizia falcataria, Dysoxylum sp., Myristica sp., Cryptocaria sp., Syzygium sp., Semecarpus spp., Ficus benjamina, Pometia pinnata, Calophyllum sp., Ervatamia sp., Pangium edule, Harpullia sp., Diospyros sp., Canarium moluccanum, Caryota rumphiana (palm), Anthocephalus chinensis, Terminalia sp.

Consumers Fruit and seed eaters

Cassowaries, birds of paradise, rats, mice, possums, cuscuses, parrots, ﬂying foxes, insects Leaf eaters and browsers Rusa deer, tree kangaroos, ringtail, insects Nectar feeders Blossom bats, honeyeaters, ﬂowerpeckers, butterﬂies, moths Omnivores Wild pigs, berrypeckers Predators (small, intermediate, Echidnas, long-ﬁngered bats, lizards, snakes, arboreal, and terrestrial) frogs, falcons, hawks, owls, harpy eagle, kites

Decomposers Scavengers and detritus feeders Chemical decomposers Detritus

Black kite, termites, millipedes Bacteria, fungi All nonliving organic matter (dead trees, animal carcasses, etc.)

568

D. P. Reagan and M. T. Rau

food web. Dashed lines indicate the recycling and ﬂux of energy and nutrients as the result of decomposition processes. Organisms of different sizes are grouped in a single functional component based on similarity in food and foraging mode. It is understood that all species of one functional group do not interact with all the species in related groups. Each of these groups contains species with additional attributes that contribute to the health and sustainability of the ecosystem. For example, bats and butterﬂies pollinate different plants, and some relationships are species-speciﬁc. Such an approach permits a general analysis without the necessity to know the detailed interactions among all organisms in this complex environment. A general listing of functional groups and their ecological attributes is provided in Table 16.2. Additional human use values, such as food, construction materials, and medicines are values of fundamental importance to communities along the Fly River. Loss of species and disruption of ecosystems can adversely affect the integrity of ecosystems and the human communities which depend on them (Reiger, 1993). No detailed studies are available by which to measure the changes now occurring within the Basin, but the extensive loss of forest cover Table 16.2: Functional groups of terrestrial ecosystems of the Fly River Basin. Functional groups Herbaceous plants Shrubs and small trees Canopy trees Fruit and seed eaters Leaf eaters and browsers Nectar feeders Omnivores Intermediate and small predators Top predators Scavengers and detritus eaters Chemical decomposers

Food Habitat

Primary Pollination Seed Decomposition Control production dispersal

X

X

X

X

X

X

X X

X

X X

X X X X

X X

X X X

Fauna and Food Webs of the Fly River Basin

569

and prolonged ﬂooding of large areas within the Basin can be assumed to have signiﬁcant effects. Though the functional food web includes all species in the ecosystem, it does not address other important interactions such as reciprocal predation (food loops) and the division of the food web into day and night compartments such as occur in other tropical forest food webs (Reagan et al., 1996). The same basic components of terrestrial food webs present in rain forests of other tropical regions occur within the Basin. Losses of forest habitat, due to overbank ﬂooding, have reduced the distribution, and presumably the populations of forest species, while expanding populations of others. Based on observations of the junior author, generalist species, such as pigs, appear to be favored by recent changes at the expense of forest species. Even at the simplistic scale of understanding, as presented in Fig. 16.6, it is apparent that recent changes in over bank ﬂooding are producing large-scale changes in food webs along the middle and lower Fly River (Rau, 1994, 1995). Dead vegetation reduces production of energy by photosynthesis by plants and creates vast amounts of dead organic matter. This pulse of organic matter in the form of decomposing leaves and dead wood ﬁnds its way into the aquatic ecosystem and downriver to the Gulf of Papua, affecting ecosystems far from the point of initial impact. Loss of trees translates into loss or redistribution of animals that inhabit forests. The complexity of interactions and incomplete knowledge of the Basin’s ﬂora and fauna and of the current human impacts make it difﬁcult to predict what the Basin will look like in the future. It is safe to say, however, because of the ongoing changes, it is likely to look far different than it does today.

16.5. Summary and Conclusions Many of the species present in the Fly River Basin are found only in New Guinea or small rain forest area remaining in northern Australia. The structure and composition of the riparian ecosystems of the Basin have been dictated by the long history of New Guinea’s close association with Australia and long separation from other terrestrial regions of the world. Consequently, monkeys, large predatory mammals, and large native mammalian herbivores are absent from New Guinea and the Basin. Marsupials dominate the mammalian fauna, and species such as tree kangaroos and spiny anteaters inhabit forests of the upper Basin. Current conditions are but a snapshot of an ever-changing tapestry of life in this Basin. Archeological evidence indicates the presence of humans on

570

D. P. Reagan and M. T. Rau

New Guinea for more than 50,000 years. Their farming practices and communities are integrated into the web of all life in the Basin. During the past century, however, resource exploitation activities, such as mining, timber harvesting, and other exploitative activities have become the primary inﬂuences on the environment. These human-induced disturbances have combined with climatic ﬂuctuations such as ENSO to rapidly alter the structure and function of the terrestrial ecosystems of the region. The changes in the sediment types along the levees of the river banks and on the ﬂoodplains through mine-derived sediment deposition is causing the simpliﬁcation of plant and animal communities. As established in annual mine reports, deposition of mine-derived sediment on the river’s banks has increased grain size and locally reduced soil pH through the introduction of sulfde minerals such as pyrite. The distribution patterns of many of the large animals described above will permanently change due to the ﬂoods and decreased availability of food. Movement of animals away from the ﬂoodplains is observed during high ﬂoods, and conﬁrmed by reports from the communities during regular community consultations. Where there is no forest as a result of dieback, there are no animals. Terrestrial food webs in the Basin are comprised of the same basic components as are rain forests in other tropical areas. Recent losses of forest habitat in the Basin have reduced the distribution, and presumably the populations of forest species, while expanding populations of others. As in large-scale disturbances in other forests, generalist species are favored at the expense of those conﬁned to a particular habitat, such as forest canopy, which are adversely affected. The increased and prolonged ﬂooding caused by the sedimentation from the Ok Tedi Mine has drastically reduced the available drier areas of ﬂoodplain for grazing. Though no recent studies have been performed, minerelated ﬂooding almost certainly reduces wild pig, wallaby, bandicoot and cassowary populations. Rusa deer are adept swimmers and seem to be the most abundant mammal on the ﬂoodplains under current conditions. The overall faunal diversity in the Fly River Basin may be adversely affected by losses of forest habitat. This reduction in species reduces the complexity which can reduce the stability of affected ecosystems, but data to address this concern is currently unavailable. These changes are taking place over very short time frames, decreasing the possibilities that species and populations can adapt effectively to such changes. Mine-related overbank ﬂooding has shrunk plant and animal populations in the region and created barriers to animal movement. The impacts of the mine operations are predicted to remain for at least the next 40–50 years, especially in the Lower Ok Tedi, the Middle Fly, and below the Everill

Fauna and Food Webs of the Fly River Basin

571

Junction on the Fly River (see Pickup and Marshall, 2009). Long-term effects on ﬂora and fauna of the Basin and the human communities that depend on them continue to be major concerns.

REFERENCES Allison, A. (2005). The Reptiles of the Trans Fly Region, New Guinea. Draft report to the WWF Paciﬁc Program. Allison, A., Bickford, D., Richards, S. J. & Torr, G. (1998). Herpetofauna in A Biological Assessment of the Lakekamu Basin, Papua New Guinea. Rapid Assessment Program Working Paper Number 9. Washington, Conservation International. Beehler, B. M., Pratt, T. K., & Zimmerman, D. A. (1986). Birds of New Guinea. Princeton University Press, Princeton. Bickford, D. (2002). Male parenting of New Guinea froglets. Nature, 418, 601–602. Bourliere, F. (1983a). Animal species diversity in tropical forests. Chapter 5. In: F. B. Golley (Ed.). Ecosystems of the World 14A: Tropical Rain Forest Ecosystems. Elsevier Scientiﬁc Publishing Co., Amsterdam. Bourliere, F. (1983b). Savannas of Australia and Southwest Paciﬁc. In: F. Bourliere (Ed.). Ecosystems of the World 13: Tropical Savannas. Elsevier Scientiﬁc Publishing Company, Amsterdam. Cogger, H. G. (1994). Reptiles and Amphibians of Australia (5th Ed.). Reed Books, Chatswood, NSW. CSIRO (1996). Review of Riverine Impacts: Porgera Joint Venture. CSIRO Environmental Projects Ofﬁce, Australia. Darlington, P. J., Jr. (1957). Zoogeography: The Geographic Distribution of Animals. Wiley. Ernst, C. H. (1989). Turtles of the World. Smithsonian Institution Press, Washington, D.C. Erwin, T. L. (1982). Tropical forests: Their richness in Coleoptera and other arthropod species. Coleopterist’s Bulletin, 36, 74–75. Flannery, T. F. (1995). Mammals of New Guinea. Cornell University Press, Ithaca, NY. Frodin, D. G., & Gressitt, J. L. (1982). Biological exploration of New Guinea. In: J. L. Gressit (Ed.). Biogeography and Ecology of New Guinea Volume 1 Part 2: Man and his Impact on the Environment. Dr. W. Junk Publishers, The Hague. Gallopin, G. C. (1972). Structural properties of food webs. In: B. C. Patten (Ed.). Systems Analysis and Simulation in Ecology. Academic Press, New York. Gregory, P. (1995). The Birds of the Ok Tedi Area. Independent Publishing, Port Moresby. Gunther, R., & Richards, S. J. (2000). A new species of the Litoria gracilenta group from Irian Jaya (Anura: Hylidae). Herpetozoa, 13, 27–43.

572

D. P. Reagan and M. T. Rau

Gunther, R., Richards, S. J., & Iskander, D. T. (2001). Two new species of Oreophryne from Irian Jaya, Indonesia (Amphibia, Anura, Microhylidae). Spixiana, 24, 257–274. Halse, S. A., Pearson, G. B., Jaensch, R., Gregory, P., Kulmoi, P., & Storey, A. W. (1996). Water bird surveys of the Middle Fly River ﬂoodplain, Papua New Guinea. Wildlife Research, 23, 557–569. Helgen, K. M., & Oliver, P. M. (2004). A Review of the Mammal Fauna of the TransFly Ecoregion, Report to WWF South Paciﬁc Program, WWF Project: TransFly Ecoregion Program, Project No: 9S0739.02 and PG0022.04. Hubbell, S. P. (2001). The Uniﬁed Neutral Theory of Biodiversity and Biogeography. Princeton University Press, Princeton. IUCN (2007). 2007 IUCN Red List of Threatened Species. IUCN, Gland, Switzerland. Janzen, D. H. (1983). Food webs: Who eats what, why, how, and with what effect in a tropical forest? Chapter 11. In: F. B. Golley (Ed.). Ecosystems of the World 14A: Tropical Rain Forest Ecosystems. Elsevier Scientiﬁc Publishing Co., Amsterdam. Johnstone, G. R., & Richards, S. J. (1994). A new species of Litoria (Anura: Hylidae) from New Guinea and redeﬁnition of Litoria leucova (Tyler, 1968). Memoirs of the Queensland Museum, 37, 273–280. Mackay, R. D. (1976). New Guinea. Time-Life Books, Amsterdam. Martin, R. (2005). Tree Kangaroos of Australia and New Guinea. CSIRO Publishing, Collingwood. Mehrtens, J. M. (1987). Living snakes of the world. Sterling Publishing Co., New York. Menzies, J. (1991). A handbook of New Guinea Marsupials and Monotremes. Publ. Kristen Pres Inc., Madang, Papua New Guinea. Montague, J. J. (1984). Morphometric analysis of Crocodylus novaeguineae from the Fly River drainage, Papua New Guinea. Australian Wildlife Research, 11, 395–414. Moulds, M. (2009). Insects of the ﬂy river system. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River system. Elsevier, Amsterdam, Vol. 9, pp. 493–513. Nightingale, N. (1992). New Guinea: And Island Apart. BBC Press, London. Paine, R. T. (1966). Food web complexity and species diversity. American Naturalist, 100, 65–75. Parametrix, Inc. & URS Greiner Woodward Clyde (1999). Assessment of Human Health and Ecological Risks for Proposed Mine Waste Mitigation Options at the Ok Tedi Mine, Papua New Guinea. Ok Tedi Mining Limited. Volumes 1, 2, 3. Pickup, G., & Marshall, A. R. (2009). Geomorphology, hydrology and climate of the Fly River System. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River system. Elsevier, Amsterdam, Vol. 9, pp. 3–49. Rand, A. L., & Gilliard, E. T. (1968). Handbook of New Guinea Birds. Natural History Press, New York.

Fauna and Food Webs of the Fly River Basin

573

Rau, M. T. (1994). Overbank Flooding Effects on Vegetation in the lower Ok Tedi. Report by Ok Tedi Mining Limited, Tabubil, Papua New Guinea. Rau, M. T. (1995). Overbank Flooding Effects on Vegetation in the Lower Ok Tedi. Report by Ok Tedi Mining Limited, Tabubil, Papua New Guinea. Rau, M. T., & Reagan, D. P. (2009). Vegetation of the Ok Tedi–Fly River system. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River system, Vol. 9. Elsevier, Amsterdam, pp. 515–547. Reagan, D. P., Camilo, G. R., & Waide, R. B. (1996). The community food web: Major properties and patterns of organization. In: D. P. Reagan, & R. B. Waide (Eds). The Food Web of a Tropical Rain Forest. Chicago University Press, Chicago, pp. 461–510. Chapter 14. Reiger, H. A. (1993). The notion of natural and cultural integrity. In: S. Woolly, J. Kay, & G. Francis (Eds). Ecological Integrity and the Management of Ecosystems. St. Lucia Press, Ottawa. Richards, S. J. (2001). A new species of torrent-dwelling frog (Anura: Hylidae, Litoria) from the mountains of Indonesian New Guinea (West Papua). Memoirs of the Queensland Museum, 46, 733–739. Richards, S. (2005). Herpetofauna of the Trans-Fly Region, Papua New Guinea. Report for World-Wide Fund for Nature (Trans-Fly Project). Richards, S. J., & Iskandar, D. T. (2000). A new minute Oreophryne (Anura: Mycrohylidae) from the mountains of Irian Jaya, Indonesia. In: A. Mack, & L. Alonso (Eds). A Biological Assessment of the Wapoga River Area of Northwestern Irian Jaya, Indonesia. RAP Bulletin of Biological Assessment 14. Conservation International, Washington D.C. Richards, S. J., & Iskandar, D. T. (2001). A new tree frog (Anura: Hylidae, Litoria) from the mountains of Irian Jaya, Indonesia. Rafﬂes Bulletin of Zoology, 48, 257–262. Richards, S. J., & Johnstone, G. R. (1993). Amphibian Fauna of Off-River Water Bodies Along the Fly River, Western Province, Papua New Guinea. July 12–16, 1993. Unpublished report to Environment Department, Ok Tedi Mining Limited, September 1993. Richards, S. J. & Oliver, P. M. (2006). A New Species of Torrent-Dwelling Litoria (Anura: Hylidae) from the Kikori Integrated Conservation and Development Project area, Papua New Guinea. Salamandra: 42–4, 231–238. Rinker, H. B., & Lowman, M. D. (2004). Insect herbivory in tropical rain forests. Chapter 18. In: M. D. Lowman, & H. B. Rinker (Eds). Forest Canopies. 2nd Ed., Academic Press, New York, pp. 359–386. Sekhran, N., & Miller, S. (Eds) (1995). Papua New Guinea Country Study on Biological Diversity. Africa Centre for Resources and Environment. Seri, L. G., & Rau, M. (1996). Preliminary mammal fauna survey of the proposed waste retention scheme sites. Appendix 24, Volume 4 of Environmental, Financial and Risk Analysis of Various Dredging and Tailings Storage Schemes To Mitigate Mining Impacts In The Ok Tedi/Fly River System. Ok Tedi Mining Ltd. ENV96-08.

574

D. P. Reagan and M. T. Rau

Storey, A., Yarrao, M., Tenakanai, C., Figa, B. & Lynas, J. (2009). Use of spatial and temporal changes in ﬁsh assemblages in the Fly River system, Papua New Guinea, to assess effects of the Ok Tedi copper mine. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River system. Elsevier, Amsterdam, Vol. 9, pp. 427–462. Sutton, S. L. (1989). The spatial distribution of ﬂying insects. Chapter 24. In: H. Lieth, & M. J. A. Werger (Eds). Ecosystems of the World 14b: Tropical Rain Forest Ecosystems. Elsevier Scientiﬁc Publications, Amsterdam. Taylor, J. A. (1979). The foods and feeding habits of subadult Crocodylus porosus Schneider in Northern Australia. Australian Wildlife Research, 6, 347–359. Trombulack, S. C., Omland, K. S., Robinson, J. A., Lusk, J. J., Fleichner, T. L., Brown, G., & Damroese, M. (2004). Principles of conservation biology: Recommended guidelines for conservation literacy for the Education Committee of the Society for Conservation Biology. Conservation Biology, 18(5), 1180–1190. UNEP (1995). Environmental impact of large-scale mining in Papua New Guinea: Mining residue disposal by the Ok Tedi Copper–Gold Mine. Selbstverlag Fachbereich Geowissenschaften, Berlin, Germany. Wallace, A. R. (1890). The Malay Archipelago (tenth edition). Paperback edition (2000). Periplus Press, Singapore. Zweifel, R. G. (1972). A Revision of the Frogs of the Subfamily Asterophryinae, Family Microhylidae. Bulletin of the America Museum of Natural History, Volume 148, Article 3.

Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00417-5

Chapter 17

Development of Aquatic Food Web Models for the Fly River, Papua New Guinea, and their Application in Assessing Impacts of the Ok Tedi Mine Andrew W. Storey1,� and Markson Yarrao2 1

School of Animal Biology (M092), The University of Western Australia, Crawley,

WA 6009, Australia

2 Environment Department, Ok Tedi Mining Limited, PO Box 1, Tabubil,

W.P. Papua New Guinea

17.1. Food Webs and Their Use in Environmental Assessment Environmental management programs often concentrate on how individual stressors affect the physiology or ecology of certain focal or keystone species (i.e. ANZECC/ARMCANZ, 2000). Although a necessary component of rigorous investigation, understanding ecological linkages of species to their food sources and predators is also vital to understand interactions amongst species and how impacts to one component can negatively (or positively) affect other components (Power et al., 1994). Such an understanding may be achieved through the construction of food chains and ultimately food webs, and allows investigators to determine how species of concern might respond to natural or anthropogenic change in their environment. Power (1990, 1992) highlighted the importance of understanding strong food chains (i.e. chains representing dominant or numerically abundant species) in complex food webs, to identify bottom–up and top–down interactions and the potential for ‘‘trophic cascades’’. Trophic cascades occur when the removal or reduction

�Corresponding author. Tel.: (618) 6488 1482; Fax: (618) 6488 1029;

E-mail: [email protected] (A.W. Storey).

576

A. W. Storey and M. Yarrao

of predators releases consumer populations, which in turn suppresses populations of their own resources, producing alternating release and suppression of trophic levels that often reach down to primary producers (Carpenter et al., 1985, 1987, 1992; Power et al., 1995; Schnetzer and Caron, 2005; Sinistro et al., 2007). Understanding such reactions is important when interpreting changes observed in a system, particularly those brought about through anthropogenic actions (i.e. changes in water quality, loss of species at different trophic levels and from different faunal groups). Although there is debate over the importance of trophic interactions in modifying populations of animals, particularly over the relative contribu tions of ‘‘bottom–up’’ versus ‘‘top–down’’ control (Carpenter et al., 1985, 1987, 1992; Schnetzer and Caron, 2005; Sinistro et al., 2007), food webs are acknowledged as a valuable tool to understanding the dynamics of communities and species (Power, 1992). Using food webs to understand the effects of the Ok Tedi mine on the ecology of the Fly River, Papua New Guinea is no exception. The Ok Tedi mine is one of the world’s largest producers of copper concentrate, and is located on the headwaters of the Ok Tedi, a major tributary of the Fly River (‘ok’ in the local Yongom language means river). The mine commenced operations in 1984 and is due to close in 2013. Initially the mine used cyanide extraction to recover gold, but in 1987–1988 the gold circuit was phased-out and replaced by a copper ﬂotation circuit. Currently the mine produces approximately 95 million tonnes per annum (mta) of waste (40 mta waste rock, 30 mta tailings, and 25 mta valley wall erosion from the waste rock dumps). The area where the mine is situated receives 10 m annual rainfall, is seismically active, and is geologically unstable. Because of these conditions, the construction of conventional waste rock and tailings storage facilities was not possible and waste material is discharged directly into the headwaters of the Ok Tedi. As a result, levels of total suspended sediment (TSS), dissolved and particulate copper (dCu and pCu) are all elevated above premine concentrations throughout the Ok Tedi and the Middle Fly River (Swales et al., 1998). An extensive biological monitoring program conducted by Ok Tedi Mining Limited (OTML) has detected a loss of some ﬁsh species from the main Ok Tedi channel, and signiﬁcant declines in ﬁsh catch biomass in the Ok Tedi and middle reaches of the Fly River (Smith et al., 1990; Smith and Hortle, 1991; Swales et al., 1998, 1999; Storey et al., 2009) indicative of impacts from the mine. The importance of describing and understanding trophic interactions amongst the ﬁshes of the Fly River has been acknowledged from the earliest investigations, initially through the collection of dietary data (Roberts, 1978; Maunsell and Partners, 1982), and subsequently through the construction of

Aquatic Food Web Models for the Fly River, Papua New Guinea

577

food webs (Maunsell and Partners, 1982; OTML, 1988; Storey and Smith, 1998; WRM, 1998). It was argued that without such information it would not be possible to assess likely indirect effects of the Ok Tedi Copper Mine on different components of the ﬁsh fauna (e.g. Power, 1990, 1992; Wootton and Power, 1993). Understanding interactions between predators and prey would assist with interpretation of monitoring data. For example, loss of invertebrate (aquatic insects) or terrestrial (fruits) food sources could result in loss of ﬁsh species dependent on these sources, alternatively, loss of predatory species could result in the increase in prey items through ‘‘predatory release’’. This paper describes the progressive development of aquatic food webs for the Fly River, illustrating how preliminary models were developed on minimal data, and how these models have evolved in their accuracy and complexity as they were progressively revised to include additional species and incorporate additional understanding of the various components of the system. The paper illustrates the general accuracy of early attempts of ecologists using little data, but basing their interpretation on experience and sensible assumptions, when compared with the more informed models using extensive datasets. Finally, the paper details the recent developments using stable isotopes of carbon and nitrogen to elucidate energy ﬂow through the aquatic ecosystem, and how this approach has advanced the understanding of current and possible future impacts of the mine on the aquatic system.

17.2. Premine/Early Mine Description of Food Webs Early investigators working on the ﬁsh fauna and aquatic resources of the Fly River were conscious of the need to document interactions amongst the diverse ﬁsh fauna, both from a purely ecological perspective (Roberts, 1978) and from a future mine management perspective (Maunsell and Partners, 1982). Roberts (1978) conducted the ﬁrst ichthyological study of the Fly River in 1975. Although no attempt was made to construct a food web per se, Roberts (1978) reported the species dietary dependencies and interactions for the diverse ﬁsh fauna described. Dietary intake of 12 species was detailed, with less extensive comments made on the food habits of an additional 29 of the 115 species recorded. Observations by Roberts (1978) have been summarized by the current authors (Table 17.1) using taxonomy applied at the time and the ten dietary categories developed by Maunsell and Partners (1982) and subsequently adopted by OTML for their database (see below).

578

A. W. Storey and M. Yarrao

Table 17.1: Dietary items for (a) detailed analysis of selected species, and (b) additional observations as reported by Roberts (1978), and subsequently classiﬁed to 10 feeding categories (sensu OTML) by the authors.

Aquatic Food Web Models for the Fly River, Papua New Guinea

579

Table 17.1: (Continued ).

Note: Nematalosa have commonly been cited as ingesting mud, and so classiﬁed as detritivores.

Examination of their gut contents under magniﬁcation, shows it to be phyto- and zooplankton, and

supports the observation that this species is a mid-water planktivore. Therefore, Nematalosa is classiﬁed as

a planktivore.

Feeding guilds are given where noted by Roberts (op. cit.). Dominant categories, as deﬁned by Roberts

(op. cit.) are highlighted (Dietary Categories: C1, Aquatic insects; C2, Terrestrial insects; C3, Aquatic

plants; C4, Terrestrial plants; C5, Crustaceans; C6, Fish; C7, Other vertebrates; C8, Molluscs; C9,

Worms; C10, Detritus/sediment).

�Indicates the presence of this dietary category in the diet. Roberts (1978) notes size range of specimens of each species taken at each site, but does not provide sample sizes for each species.

Feeding guilds for each species as noted by Roberts (1978) are also presented. Dietary intake tended to be inferred from small sample sizes (1–20 individuals, majorityo12), and from a limited number of locations within the river system, however, detailed identiﬁcations are given for some prey items (e.g. aquatic insects identiﬁed to Order, ingested ﬁsh identiﬁed to species level, Crustacea identiﬁed to genus level). These data provide an indication of the specialized feeding mode of some species, compared with the generalist approach of other species (Table 17.1). Although a very comprehensive survey, Roberts (1978) did not report data for barramundi Lates calcarifer (Bloch, 1790), which is an important predatory species in the system, and consistently comprises a high proportion of ﬁsh catch biomass. Moore (1982), Moore and Reynolds (1982), and Reynolds and Moore (1982) report data for 101 specimens of L. calcarifer taken from the middle Fly River, with dietary items summarized as percentages based on number of items encountered. These data demon strated that L. calcarifer is a predator, principally a piscivore, but ingests a

580

A. W. Storey and M. Yarrao

wide range of prey species, including in descending order of percent; unidentiﬁed ﬁsh, herring (Nematalosa spp.), freshwater prawns (Macrobrachium spp.), fork-tailed catﬁsh (Ariidae), eel-tailed catﬁsh (Plotosidae), other Crustacea, mouth-almighty, or cardinalﬁshes (Glossamia sp.), freshwater longtom (Strongylura kreffti) (Gunther, 1866), and glassﬁshes (Ambassidae). Although Nematalosa spp. were the dominant single identiﬁable species, the vast majority of stomach contents (60%) were unidentiﬁed ﬁsh, presumably because they were semidigested. Macrobrachium prawns were also a signiﬁcant component of the diet. The next major study on the Fly River was in 1981 by Maunsell and Partners (1982), in which 21 locations were sampled (mostly riverine as opposed to ﬂoodplain) along the river, from the headwaters to the estuary. The study was part of the baseline environmental study for the then proposed Ok Tedi Mine, and reported dietary data for ﬁsh collected (Table 17.2). Maunsell and Partners (1982) categorized species into carnivores, detritivores, and omnivores, and food items were classiﬁed as either allochthonous (originating from riparian zones) or autochthonous (derived from in-stream production). Trophic relationships amongst species for which there were detailed dietary data were also derived, and using this information the ﬁrst food chain for the Fly River was constructed, with an emphasis on the major biological links within the system (Fig. 17.1). Maunsell and Partners (1982) commented that sampling was conducted over a short time period, when river levels were high and ‘‘when many insects had been washed out of their terrestrial habitats and drowned.’’ This could account for the predominance of terrestrial insects in the diets of many species. Maunsell and Partners (1982) also noted that most of the common species appeared to feed opportunistically upon the array of food items available, with few specialized feeders. The main observations by Maunsell and Partners (1982) were: � The importance of allochthonous material, especially drowned insects opportunistically incorporated in the food web, � Opportunism in selection of food items by many species of ﬁsh, and therefore trophic levels could only be broadly identiﬁed, � Three species of mullet (Crenimugil labiosus, Liza diadema, Rhinomugil nasutus) and the herring (Nematalosa spp.) were all browsers on algae and detritus, consuming sediment, � Lates calcarifer, Lutjanus argentimaculatus (Forsska˚l, 1775), L. goldiei (Macleay, 1884), Arius augustus (Roberts, 1978), A. leptaspis (Bleeker, 1862) and Scleropages jardini (Saville-Kent, 1892) were all top carnivores, feeding on ﬁsh, macroinvertebrates/crustacea, and vertebrates (frogs, lizards, etc.),

Aquatic Food Web Models for the Fly River, Papua New Guinea

581

Table 17.2: Summary of food items recorded in species of ﬁsh by Maunsell and Partners (1982), giving percentage of each dietary category, and feeding guilds as deﬁned by Maunsell et al. (op. cit.). Species Liza diadema Rhinomugil nasutus Crenimugil labiosus Nematalosa erebi Megalops cyprinoides Thryssa scratchleyi Thryssa rastrosa Arius acrocephalus Arius leptaspis Arius carinatus Arius augustus Arius berneyi Cochlefelis danielsi Neosilurus ater Melanotaenia oktediensis Melanotaenia rubrostriata Craterocephalus randi Ambassis agrammus Lates calcarifer Hephaestus trimaculatus Glossamia aprion

Guild Detritivore Detritivore Detritivore Detritivore Carnivore Carnivore Insectivore Omnivore Omnivore Insectivore Carnivore Omnivore Carnivore Omnivore Insectivore

C1

C2

C3

C4

C5

C6

C7

C8

C9

C10

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 40 o5 0 0 28 24 6 o5 13 3 0 o5 43 40 0 0 87 5 0 0 0 0 0 0 23 31 0 29 12 o5 0 0 19 10 o5 28 13 19 7 0 78 13 0 o5 o5 0 0 0 11 0 0 0 22 66 0 0 15 26 0 13 5 12 o5 9 o5 30 0 o5 65 0 o5 0 56 0 7 0 0 0 0 16 31 67 0 0 0 0 0 0

0 0 0 0 0 0 0 0 7 0 0 17 0 0 0

100 100 100 100 0 0 9 o5 o5 o5 o5 o5 o5 20 o5

0 o5

o5

Insectivore

16

68

4

8

0

0

0

Insectivore Insectivore Carnivore Carnivore

88 74 0 24

0 5 12 3

0 0 0 0

0 7 0 0

0 11 51 13

6 0 37 19

0 0 0 0

0 0 0 11

0 0 0 22

6 o5 0 8

Car/ o5 Omnivore

54

0

0

0

38

0

0

0

5

Note: Maunsell and Partners (1982) refer to Category 10 as ‘‘miscellaneous, mud/detritus.’’

Dietary categories: C1, Aquatic insects; C2, Terrestrial insects; C3, Aquatic plants; C4, Terrestrial plants;

C5 Crustaceans; C6, Fish; C7, Other vertebrates; C8, Molluscs; C9, Worms; C10, Detritus/sediment. Data

for species for which there were at least six full or partially full stomachs are presented.

� The majority of species fell between carnivores and detritivores, consuming a wide variety of food items (i.e. omnivorous), � Aquatic primary productivity in all riverine habitats was low, probably as a result of (naturally) high turbidity and low light penetration in the lowland reaches, and high velocity and turbulence in mountainous reaches, � River-borne detritus and allochthonous plant and animal material formed much of the basis of the food chain, � Bosset Lagoon supported some plankton (NB Maunsell and Partners (1982) did not deﬁne whether this was zooplankton, phytoplankton, or

582

A. W. Storey and M. Yarrao

Figure 17.1: Inferred food chain of the Fly River (reproduced from

Maunsell and Partners, 1982).

both), most other ﬂoodplain habitats examined were poor in plankton, as was the main river channel. (NB a recent survey (2006) of six oxbow lakes and two lagoons has revealed an extremely diverse phytoplankton ﬂora (154 taxa, average of 57 per site) and zooplankton fauna (total of 217 taxa, average of 88 per site) (WRM, 2007)).

Aquatic Food Web Models for the Fly River, Papua New Guinea

583

The food chain constructed by Maunsell and Partners (1982) was based on relatively few data, incorporating only 21 of the B 115 species of ﬁsh found in the river system. However, it beneﬁted from the experience gained by the authors from other tropical systems. The food chain identiﬁed the main links between primary, secondary and higher consumers, noted the role of freshwater prawns in the food web, the apparent minor role of in-stream production (algal growth), and noted the important role of food items derived from the riparian zone. This food chain underpinned environmental management at the Ok Tedi mine during the initial years of mine development, until it was revised during preparation of the Sixth Supplemental Agreement Environmental Study Draft Final Report (OTML, 1988); the agreement between the State and OTML, under which the mine operated from 1986 until implementation of the APL Compliance and Additional Environmental Monitoring Program in 1990 (OTML, 1990; Wood et al., 1994). Revision of the food web utilized data collected over early mine life (1984–1986). OTML summarized feeding habits of Fly River ﬁshes and produced a food web for the middle Fly River channel incorporating a range of primary producers and consumers, including a range of ﬁsh at species, genus, and family levels (Fig. 17.2). Functionally, this food web was similar to that produced by Maunsell and Partners (1982), but it beneﬁted from a better understanding of how the system functioned, and from additional data collected as part of routine monitoring from 1983–1986. As a result, the food web was more detailed in the understanding of various linkages and in the number of species of ﬁsh included.

17.3. Use of Dietary Data in Food Webs Following preparation of the food web for the Sixth Supplemental Agreement (OTML, 1988), the next major development in food webs for the Fly River was not until the screening-level and detailed Health and Ecological Risk Assessments (Parametrix et al., 1999a, b). A review of available data highlighted the need for a food web of Fly River ﬁshes that would conﬁdently identify and detail interactions between the ﬁsh species and different food sources within the system. It was argued that without such a food web it would not be possible to assess likely indirect effects of the mine on different components of the ﬁsh fauna either via bottom-up (i.e. changes ﬂowing up through the food web due to effects on primary producers) or top-down responses (i.e. increased or reduced pressure on

584

A. W. Storey and M. Yarrao

END CONSUMERS Homo sapiens, avifauna, Crocodilia

Terrestrial vertebrate eating fish H. taylori, Eleotridae

Piscivorous fish Lates, Thryssa, Lutjanus, A. augustus, Megalops

Omnivorous fish A. leptaspis, A. berneyi, C. spatula, A. carinatus

Terrestrial vertebrates Squamata (lizards & snakes) Amphibia (frogs)

Crustacean eating fish Nibea, Lutjanus, A. augustus, A. leptaspis Crustaceans Macrobrachium other Malacostraca Terrestrial arthropod eating fish T. chatareus, Zenarchopterus, H. macrorhynchus, Clupeoides Molluscivorous fish

Terrestrial plants, fruit, seed eating fish H. macrorhynchus, A. leptaspis, H. taylori, A. acrocephalus

Cinetodus froggati

Aquatic invertebrate eating fish H. macrorhynchus, H. taylori, A. acrocephalus, Ambassis, M. splendida Terrestrial arthropods Odonata adults, Orthoptera, Hymenoptera Mollusca

Detritus, mud eating fish L. diadema, N. dayi, Nematalosa, C. labiosus, M. splendida

Aquatic invertebrates Coleoptera, Trichoptera, Diptera, Ephemeroptera Detritus, mud

Terrestrial plants, fruit and seeds

SURFICIAL SEDIMENTS

Figure 17.2: Food Web for the ﬁsh fauna of the Middle Fly River channel (from OTML, 1988). lower trophic levels due to a change in higher order consumers). Therefore, WRM (1998) conducted a review and summary analysis of available data, with the aim of identifying data gaps and data needs. The main data source identiﬁed by WRM (1998) was the extensive database developed by OTML, detailing the stomach contents of 10,961 ﬁsh from the Fly River, covering 64 species, collected from 1983 to 1994 (principally collected over 1984–1986),

Aquatic Food Web Models for the Fly River, Papua New Guinea

585

from 12 riverine sites (six Ok Tedi and upper catchment and six Fly River sites), and eight ﬂoodplain sites. These data were taken from ﬁsh collected as part of routine monitoring programs using a combination of gill netting, seine netting, rotenone poisoning and trapping. Each specimen was dissected and the fullness of each stomach was ranked as 0 (empty), 1 (quarter full), 2 (half full), 3 (full), or 4 (distended), and the contents were classiﬁed into the same 10 general categories developed by Maunsell and Partners (1982) (ref Tables 17.1 and 17.2), and the approximate volume of the stomach contents occupied by each category recorded to the nearest 10%. These values were then scored from 1 (10%) to 10 (100%). WRM (1998) classiﬁed sites as either riverine or ﬂoodplain habitat and used two-way ANOVAs on those individuals with food items in the stomach (n ¼ 10,961) to test for between category and between habitat differences in dietary composition for each species. Thirty species were sampled from both habitats, whilst 34 species were recorded from one habitat only (29 species from riverine and 5 species from ﬂoodplain habitat). The majority of species recorded from riverine habitat reﬂected the greater historical sampling intensity at riverine as opposed to ﬂoodplain sites. Ten species demonstrated a signiﬁcant difference in gut fullness between habitats, and in all but one instance, individuals from ﬂoodplain habitats had fuller guts compared to those from riverine habitats. This result was interpreted as indicating better feeding efﬁciency in ﬂoodplain as opposed to riverine habitats for those species exhibiting signiﬁcant differences. This may reﬂect reduced feeding efﬁciency for ﬂoodplain-adapted species when in turbid, ﬂowing water (i.e. species normally associated with ﬂoodplain habitat enter the main river channel during severe droughts to avoid desiccation and then have difﬁculties feeding in the turbid water), increased feeding efﬁciency of riverine species when on the ﬂoodplain (even though these species tend to have tactile and chemo-adaptations to assist feeding in turbid riverine conditions), or simply greater food availability at ﬂoodplain over riverine sites. Further analysis determined that the majority of species showed a preference for one or more dietary categories, and that the dominant dietary components differed between riverine and ﬂoodplain habitats (WRM, 1998). For example, Thryssa scratchleyi (Ramsay and Ogilby, 1886) had a greater proportion of aquatic insects in its diet at ﬂoodplain as opposed to riverine sites, but a greater proportion of crustaceans at riverine as opposed to ﬂoodplain sites. Arius leptaspis had a greater proportion of aquatic insects in its diet at ﬂoodplain compared to riverine sites and the opposite trend for terrestrial insects. The diet of Cinetodus froggatti (Ramsay and Ogilby, 1886) from ﬂoodplain sites consisted entirely of gastropods, whilst most categories (in addition to gastropods) were represented in its diet from riverine sites.

586

A. W. Storey and M. Yarrao

In ﬂoodplain habitats, the diet of Lates calcarifer consisted predominantly of ﬁsh, whilst in riverine sites, ﬁsh and Crustacea were the dominant categories. These results in part reﬂected differing availability of certain food items between habitats, but also likely reﬂected differences in feeding ability/ efﬁciency between habitats (WRM, 1998). Whilst this was a large and comprehensive database covering many sites and species, the summary analyses by WRM (1998) did not attempt to look at seasonal differences or changes associated with different life history stages within species. In-depth analysis of the database may reveal such patterns. The targeted review by WRM (1998) also determined reasonable con cordance in dietary intake reported by Roberts (1978), Maunsell and Partners (1982) and the OTML database, although there were some differences, which probably related to habitat differences (riverine versus ﬂoodplain populations, or upper versus lower catchment sites), or opportunistic feeding on items available seasonally/sporadically. WRM (1998) considered it unlikely that a species-speciﬁc food web model could be constructed for all Fly River ﬁsh species without extensive additional work. There were, however, sufﬁcient, good quality data for the classiﬁcation of the majority of common species to functional feeding groups/guilds (e.g. piscivores, terrestrial insectivores, molluscivores etc), from which food webs could be constructed at the ‘‘functional’’ level for the ﬁsh assemblages if not for the whole community. Because of the reported differences between ﬂoodplain and riverine ﬁsh assemblages, it was recommended that habitat-speciﬁc functional food web models, for ﬂoodplain and riverine ﬁsh populations be constructed. Storey and Smith (1998) subsequently used the OTML database to construct functional food webs to show the relative importance of each food source to each species, and to the ecosystem. The former objective was achieved by calculating the mean percentage composition of each dietary category for each species. This quantiﬁed the importance of each food source to the diet of each species, but did not reﬂect the importance of that food source at the ecosystem level (e.g. how much does that food source contribute to carbon ﬂow through the whole ecosystem). This latter objective was determined by multiplying the mean percentage composition of each dietary category for a species by the relative biomass of that species in the system, determined as the proportion that the species comprised of total ﬁsh catch biomass, taken from OTML biomonitoring data (see Storey et al., 2009). This produced scores for the proportion contributed by each dietary category in each species to the total ﬁsh community. The resulting scores when summed for each dietary category gave the relative proportion of all stomach volume taken-up by each dietary category. Dietary categories equate to the food sources in the food webs, therefore indicating the relative

Aquatic Food Web Models for the Fly River, Papua New Guinea

587

importance of each food source to the ecosystem (Storey and Smith, 1998). The use of total catch biomass of each species as opposed to the numerical abundance of each species, was justiﬁed to prevent placing too much importance on small, but very numerous species (e.g. Rainbow ﬁshes), which, although numerically abundant, would utilize a relatively small volume of a food source compared with a few individuals of a much larger species. In the resultant functional food webs for Riverine and Floodplain ﬁsh assemblages (Figs. 17.3 and 17.4, respectively), species are listed under each food source utilized by that species, and the Dietary Score in parenthesis beside each species name reﬂects the relative contribution to stomach content of that dietary source to each species. The relative proportion of stomach volume taken-up by each dietary category was then used to weigh the importance of each food source in each food web, with heavier arrows indicating greater utilization of a food source (Figs. 17.3 and 17.4). The 10 highest ranked species in each food source were then indicated in each food web (species names which are underlined). Storey and Smith (1998) noted that the mud/detritus category is probably over-emphasized in these models because all unidentiﬁable organic matter tended to be allocated to this category, resulting in most species of ﬁsh appearing to consume detritus/mud, even pelagic species such as Megalops cyprinoides (Broussonet, 1782) and Strongylura kreffti. In particular, this practice classiﬁed herrings, Nematalosa spp. as detritivores. Microscopic examination of the stomach contents of herrings has shown that this material is actually phyto- and zooplankton (Ian Roderick, OTML, unpublished data), as would be expected for a midwater planktivore. Therefore, this very abundant species was classiﬁed as planktivorous by Storey and Smith (1998) to more appropriately shows it functional role in the food web. The food webs developed by Storey and Smith (1998) support the differences in feeding preferences between riverine and ﬂoodplain habitats identiﬁed by WRM (1998), with plankton, aquatic invertebrates and aquatic plants more important food sources in the ﬂoodplain than the river, and the opposite true for crustaceans, terrestrial invertebrates, and ﬁsh. The food webs highlight piscivory as a dominant feeding mode in both the riverine and the ﬂoodplain habitats, with planktivory co-dominant on the ﬂoodplain, principally because of the numerically abundant Nematalosa herrings. Storey and Smith (1998) noted various limitations and assumptions in the data, and in the interpretation of the resultant food webs: � Dietary data were collected over a long time period, but using standardized methods (Hortle, 1986). It was assumed that all data were comparable.

588 A. W. Storey and M. Yarrao

Figure 17.3: Food web for ﬁsh populations in the Fly River channel (from Storey and Smith, 1998). Arrows indicate the proportional use of each food source by the ﬁsh fauna, broken lines indicate assumed linkages. Species names underlined are the top 10 species utilizing each food source in terms of stomach volume and abundance of the species in the river channel. Numbers in parentheses are dietary categories indicating percentage of stomach volume occupied by each food source (1 ¼ r5%; 2 ¼ W5%, and r20%; 3 ¼ W20% and r50%; 4 ¼ W50%).

Aquatic Food Web Models for the Fly River, Papua New Guinea 589

Figure 17.4: Food web for ﬁsh populations on the Fly River ﬂoodplain (from Storey and Smith, 1998). Arrows indicate the proportional use of each food source by the ﬁsh fauna, broken lines indicate assumed linkages. Species names underlined are the top 10 species utilizing each food source in terms of stomach volume and abundance of the species on the ﬂoodplain. Numbers in parentheses are Dietary Categories indicating percentage of stomach volume occupied by each food source (1 ¼ r5%; 2 ¼ W5% and r20%; 3 ¼ W20% and r50%; 4 ¼ W50%).

590

A. W. Storey and M. Yarrao

� The data cover a long time period and may be affected by changes in mine operating periods (e.g. samples span four mine operating periods: construc tion, gold circuit only, gold and copper circuits both operating and copper circuit in isolation). They also cover a broad range of sites (gravel bed versus sand bed rivers, and oxbow versus blocked-valley lake versus ﬂoodplain sites) sampled in different seasons (wet and dry season). The food webs therefore integrate all these factors into a broad representation of dietary intake by ﬁsh species in each main habitat type. � The broad categories used have limited value, since species may feed on speciﬁc items within a major category. In particular, ﬁsh ingested by piscivores were not identiﬁed to species, therefore it was not possible to separate and quantify the contribution of each primary ﬁsh trophic group to piscivorous ﬁshes. � The database likely over-emphasizes the importance of slowly digested items retained in the stomach relative to rapidly digested food items (i.e. plankton and many aquatic invertebrates), and provides no information on different rates of digestion across species. � The food webs were based on dietary intake of ﬁsh, and assume the linkages between food sources and primary producers/primary consumers (e.g. it was assumed aquatic invertebrates consume periphyton and detritus in equal proportions). Given the extensive data base on dietary data developed by the mine, it was only sensible that these data should be analyzed to their full extent, acknowledging the various shortcomings and assumptions. However, WRM (1998) also recommended the use of stable isotopes as a valuable addition to dietary intake studies and the construction of food webs. It was argued that determination of both d13C carbon and d15N nitrogen signatures for ﬁsh species and primary food sources would assist in conﬁrming dietary sources, help in differentiating between ingested and assimilated food items, identify trophic levels of consumers, and determine the relative contributions of nonﬁsh compartments to the food webs (Bunn et al., 1999; Peterson, 1999; Bowles et al., 2001; Douglas et al., 2005; West et al., 2006). Most importantly, this approach should provide a reliable indication of the ultimate source(s) of organic carbon that underpin the food web. This recommendation, and resulting studies (see below) led to a change in environmental awareness at OTML from 1998 to 2006, an expanded under standing of how the mine was affecting the ecology of the river system and gave rise to a series of targeted studies looking speciﬁcally at food web interactions, carbon sources and energy ﬂow.

Aquatic Food Web Models for the Fly River, Papua New Guinea

591

17.4. Use of Stable Isotopes to Describe Food Webs As detailed above, the traditional method for constructing food webs has been to determine which food items animals are ingesting by looking at their feeding mode and gut contents. However, this approach has inherent deﬁciencies, including identiﬁcation of partly digested items, regurgitation of items under capture stress, and underestimation of the importance of those foods which are digested at more rapid rates (Peterson and Fry, 1987; Lancaster et al., 2005). In addition, gut content only shows what was ingested, not whether ingested items are assimilated (i.e. components of the diet that contribute to the C and N signatures of consumer muscles, bones or organs). Determination of the components of an animal’s diet that are actually assimilated has been simpliﬁed with stable isotope tracing techniques (Peterson and Fry, 1987; Lajtha and Michener, 1994). The measurement of stable isotope ratios of carbon and nitrogen in tissue samples has furthered our understanding of predator–prey relationships and the energy sources supporting aquatic food webs (Peterson and Fry, 1987; Lajtha and Michener, 1994; Peterson, 1999; West et al., 2006). The basis of this technique is the accurate measurement of nitrogen isotopes 14N and 15N and carbon isotopes 12C and 13C in source materials and consumer tissues (Peterson and Fry, 1987; Peterson, 1999; West et al., 2006). Nitrogen stable isotopes are particularly useful in determining trophic position, as signatures undergo fractionation during trophic transfer from food items to consumers (Minagawa and Wada, 1984). This fractionation typically leads to enrichment of the 14N:15N ratio due to digestion of nitrogen-containing biomolecules (mainly proteins) by organisms and the preferential excretion of 15N-depleted urea and ammonia (Michener and Schell, 1999). As a consequence, d15N increases by approximately 3m with each trophic level (Minagawa and Wada, 1984; Michener and Schell, 1999; Peterson, 1999). However, there is variability in the degree of trophic fractionation of nitrogen and this distinction between trophic levels using d15N does not appear to be the case in all aquatic systems where, because omnivory is prevalent, animals exist along a trophic continuum rather than at discrete trophic levels (France et al., 1998). In contrast to the situation for nitrogen stable isotopes, the fractionation of carbon stable isotopes occurs mostly at the primary producer level (De Niro and Epstein, 1978; Rounick and Winterbourn, 1986; Peterson, 1999). To this end, subsequent trophic transfer of carbon does not appreciably affect 13C: 12C ratios. Different photosynthetic routes such as the Calvin cycle (C3), Hatch–Slack cycle (C4) and Crassulacean Acid

592

A. W. Storey and M. Yarrao

Metabolism (CAM) result in different d13C values. Terrestrial C3 plants (d13CB�28m) take their carbon from the atmosphere (13CB�7m) and a net fractionation of about 21m occurs during photosynthesis (Peterson and Fry, 1987). In contrast, the photosynthetic pathway of C4 plants (a group that includes tropical and salt grasses) results in a smaller fractionation of about 6m (Peterson and Fry, 1987). The carbon cycle of aquatic plants is different again, as photosynthesis utilizes dissolved CO2 which may be supplied by carbonate rock weathering, mineral springs, from the atmosphere or from respired organic matter (Peterson and Fry, 1987). This causes widely varying d13C values that are subject to site speciﬁc and seasonal variations (Peterson and Fry, 1987; Jones et al., 1999; Peterson, 1999; Kankaala et al., 2006). Stable isotope analysis of carbon has proved particularly effective in the study of aquatic food webs where the variation in carbon isotope ratios for different plant groups allows consumer organisms to be traced back to speciﬁc primary producers (Rounick and Winterbourn, 1986; Peterson and Fry, 1987; Rosenfeld and Roff, 1992; Boon and Bunn, 1994; Lajtha and Michener, 1994). The utility of the d13C method for determining carbon sources relies upon 13C not being fractionated during food assimilation. In practice, a slight enrichment of d13C occurs with each trophic transfer, however, the fractionation is small (0.5–1m) in relation to the overall span of d13C values (typically �10m to �40m for freshwater plants) (De Niro and Epstein, 1978; Rounick and Winterbourn, 1986). The approach also relies on there being a signiﬁcant separation in the d13C of suspect primary carbon sources, otherwise it is very difﬁcult to discriminate between source contributions to consumers if primary sources overlap. 17.4.1. Establishing Applicability of SIA and Baseline Conditions On the recommendation of WRM (1998) and Storey and Smith (1998), a study was conducted by Bunn et al. (1999) to test whether stable isotope analysis could identify and separate the major primary sources of organic carbon driving aquatic food webs in the Fly River system, and establish baseline conditions for carbon sources driving food webs at ﬂoodplain and riverine habitats. A river reach upstream of D’Albertis Junction, above the inﬂuence of the mine (Fly River at Kawok), and an oxbow lake downstream of the mine, but not showing any effects from the mine at that time (OXB06 at ARM 346) were selected for investigation using stable isotopes of carbon and nitrogen. Samples of ﬁsh and macroinvertebrates, together with potential primary food sources were collected from the two locations between the 16th and 21st of October 1998.

Aquatic Food Web Models for the Fly River, Papua New Guinea

593

Primary sources were collected by hand at each site. Epiphytic algae were gently scraped from trailing vegetation (e.g. Ischaemum or Saccharum). Larger strands of ﬁlamentous algae were picked from submerged vegetation and logs. Phytoplankton at Oxbow 6 was collected by ﬁltering through GF/C paper, though there was insufﬁcient for analysis. Samples of seston (phytoplankton and ﬁne detritus) were obtained using a ﬁne plankton net at Kawok. Aquatic macrophytes were collected where present from each sampling location, including Utricularia exoleta, U. aurea, Azolla sp., Ludwigia sp., and Cyperus platistylis. Samples of conspicuous fringing vegetation were collected by hand from each site, particularly Saccharum robustum, Echinochloa praestans, Phragmites karka, and Ischaemum poly stachium. Leaves and (where present) fruits from riparian trees were also collected by hand, including Ficus polyanther, F. cf. wassa, Laportea decumana, Eleocarpus sp., Anthocephalus chinensis, and Leer indicus. Five samples of coarse particulate organic matter (whole leaves) and ﬁne particulate organic matter (0.25–1.00 mm) were obtained from Ekman grab samples at each location (one per site). Aquatic macroinvertebrates, mostly atyid shrimps (Atyidae), baetid and caenid mayﬂies, and odonates (Zygoptera and Anisoptera), were collected with a long-handled sweep-net (250 mm) from fringing vegetation or submerged logs at each site. Aquatic mites were also collected at Oxbow 6. Chironomid larvae were the only species found in the benthos, with only one or two individuals obtained from each grab sample, as a result insufﬁcient larvae were collected for analysis. Zooplankton were sampled after dark near the fringing vegetation at Oxbow 6, using a 60 mm plankton net. Freshwater sponges were also prevalent on submerged riparian vegetation at Oxbow 6 and samples were taken to provide an indication of the phytoplankton isotope signatures. At Kawok, vertical clay banks on the outside meander bends were riddled with mayﬂy burrows (Plethanogesia). Samples of clay were removed by spade and carefully washed through 4 and 1 mm sieves to retain larvae. At one site, larvae of a free-living caddis (Hydrobiosidae?) were also found in the burrows. Four large baited traps were set at each location to catch Macrobrachium prawns (Palaemonidae). No animals were collected from Oxbow 6, however, at least ﬁve species were sampled from Kawok, including M. rosenbergii, M. lorentzi (possibly a complex including an undescribed species), M. handschini, and M. weberi. Grasshoppers were sampled from fringing vegetation as representative of terrestrial primary consumers. Moths (Lepidoptera) and adult caddis (Trichoptera) were also collected at night around the lights on board the sampling vessel; the Western Venturer.

594

A. W. Storey and M. Yarrao

Some adult dragonﬂies (Anisoptera) were also sampled along the fringing vegetation at Oxbow 6. Most of the larger ﬁsh were caught in standard OTML gill net sets at each location (Storey et al., 2009). Smaller individuals were sampled from trailing vegetation using a long-handled sweep-net. Some specimens were collected from ﬂooded forest at Oxbow 6 using rotenone, and juveniles and small species were collected by spotlighting in the river channel. All of the larger ﬁsh were individually measured in the ﬁeld and samples of dorsolateral muscle removed for analysis and frozen. Carapace lengths of large Macrobrachium rosenbergii were also recorded and samples of tail muscle removed for analysis. All samples of primary sources and invertebrates were kept on ice in the ﬁeld and subsequently frozen until prepared for stable isotope analysis. In the laboratory, all samples were cleaned, rinsed in distilled water and oven-dried at 601C for 36–48 h. Dried material was then ground using a mortar and pestle (small samples) or ring grinder (large samples). Before drying, atyid shrimp, zooplankton and sponges were acid washed in 20% HCl for 2 min and rinsed, prior to drying, to avoid possible contamination from nondietary carbonates (see Bunn et al., 1995). Acid-washed individuals were used to obtain d13C values, and nonacid washed individuals were used to obtain d15N values. Exoskeletons of Macrobrachium were also removed to avoid possible contamination. Individual carapace lengths of Macrobra chium were ﬁrst recorded, and the digestive tracts also removed. Dried, ground samples were oxidized at high temperature and the resultant CO2 and N2 were analyzed with a continuous ﬂow-isotope ratio mass spectrometer (Europa Tracermass and Roboprep, Crewe, UK). Ratios of 13 C/12C and 15N/14N were expressed as the relative per mil (m) difference between the sample and conventional standards (PDB carbonate and N2 in air) where: dX ¼

Rsample � 1 � 1000ð%Þ Rstandard

where X ¼ 13C or 15N and R ¼ 13C/12C or 15N/14N. The possible contributions of algae and terrestrial vegetation to the assimilated carbon in primary consumers were calculated using the following simple two end-point mixing model: Palgae ¼

d13 C consumer � f � d13 Criparian d13 C algae � d13 C riparian

!

Aquatic Food Web Models for the Fly River, Papua New Guinea

595

where Palgae ¼ proportion of algal carbon; and f ¼ isotopic fractiona tion (m). In all cases, an isotopic fractionation (f ) of 0.2 was used (see France, 1995). The same approach was used to assess the likely contributions of primary consumers to the assimilated carbon of secondary consumers. Stable nitrogen isotope values and data on diet (Storey and Smith, 1998) were used to determine which species were secondary (or tertiary) consumers. In total 512 samples were analyzed, consisting of 38 and 49 samples of primary sources from 13 and 15 different species at Kawok and Oxbow 6 respectively, 72 and 56 samples from 17 and 11 species of invertebrates, and 206 and 91 samples from 24 and 14 species of ﬁsh from Kawok and Oxbow 6 respectively. Aquatic algae from both locations were 13C-depleted compared with riparian forest vegetation (C3 plants) and markedly so compared with C4 plants such as Saccharum and Ischaemum. Riparian vegetation was also more 15N-depleted compared with aquatic plants, especially at Kawok. Analysis showed that macroinvertebrates collected from submerged or trailing vegetation obtained their biomass carbon primarily from algae. Only a few chironomid larvae were collected from benthic samples at both locations and these too had isotopic signatures similar to algae. Interactions amongst the various sources and consumers were summarized in stylized food web plots (Fig. 17.5), which generally supported many of the previously conceived trophic interactions (Maunsell and Partners, 1982; OTML, 1988; Storey and Smith, 1998; WRM, 1998), but also showed a greater contribution of algal-derived carbon to food webs than previously determined. This highlighted the utility of stable isotope analyses when compared to gut contents analyses, and showed that consumed algae is often very difﬁcult to identify and is often lumped into a miscellaneous or detritus category. The stable isotope approach also showed the trophic relationships between periphytic algae, grazing macroinvertebrates, small ﬁshes, and larger predatory species. It conﬁrmed phytoplankton and zooplankton as the main source of carbon for Nematalosa herrings, identiﬁed species of ﬁsh dependent on direct riparian inputs, and showed that secondary and higher consumer ﬁsh tended to exist along a trophic continuum rather than in discrete trophic positions, which is indicative of a high level of omnivory (France et al., 1998). Based on the food webs, Bunn et al. (1999) identiﬁed four major pathways for transfer of carbon: 1. There was a very important plankton pathway that provided carbon to planktivorous Nematalosa, a numerically dominant prey item for piscivores.

litter

terrestrial (C3) inputs Oxbow 6

fruits insects Megalops cyprinoides Arius leptaspis Strongylura kreffti, Scleropages jardini

Lates calcarifer

Toxotes chatareus

Glossomia aprion

Porochilus meraukensis Variichthys lacustris Nematalosa sp

Arius berneyi Melanotaenia splendida Oxyleotris herwerdenii

Neosilurus ater

Craterocephalus randi Nematalosa sp (<90mm) odonates

sponges

mayflies atyids

mites zooplankton

plant detritus

epiphytes

phytoplankton

animal detritus

aquatic sources litter

terrestrial (C3) inputs Kawok

fruits insects

Lates calcarifer Strongylura kreffti Lutjanus goldiei Scleropages jardini Thryssa scratchleyi Glossomia aprion

Porochilus meraukensis Variichthys lacustris Melanotaenia splendida Craterocephalus randi Nematalosa spp., Ambassis agrammus Amniataba percoides

zooplankton phytoplankton

Arius latirostris Arius leptaspis Toxotes chatareus

Neosilurus ater Hephaestus roemeri Z. novauguineae Plotosus papuensis

Glossogobius Macrobrachium

odonates mayflies atyids epiphytes

plant detritus

animal detritus

aquatic sources

Figure 17.5: Simpliﬁed food web for Oxbow 6 and Kawok based on stable isotope and dietary data reproduced from Bunn et al. (1999). Thickness of lines is indicative of the strength of the interactions and broken lines indicate assumed linkages.

Aquatic Food Web Models for the Fly River, Papua New Guinea

597

2. There was also an obvious periphyton-invertebrate grazer pathway representing an important carbon pathway to small ﬁsh species and juveniles of larger species that utilize fringing vegetation and logs as habitat, and feed on associated aquatic insects and crustaceans. 3. Most species of Macrobrachium prawns, an important food resource for predatory ﬁsh in the river channel, derived their biomass carbon from terrestrial sources, but this was not through direct consumption of riparian vegetation because the prawns occupied too high a trophic position, and, 4. Selective feeders, such as Toxotes chatareus (Hamilton, 1822) feeding on terrestrial insects and Arius latirostris (Macleay, 1884) feeding upon fruits, clearly derived most of their biomass carbon and nitrogen from terrestrial sources at both locations. An independent study of the food web of Lake Murray (Bowles et al., 2001) showed strong consistency in d13C signatures for primary producers, seston and for species of ﬁsh that were recorded in both studies, providing conﬁdence in the approach, whereby species from Lake Murray appear to be assimilating carbon from comparable energy sources. Bunn et al. (1999) considered that microalgae were clearly important to the food web, and of great signiﬁcance to the ﬁshery. Using estimates of the contribution of algal-derived carbon to the biomass of each ﬁsh species, and the proportion of each species in ﬁsh catch monitoring data, it was estimated that algal carbon supported approx. 40% of ﬁsh standing stock biomass at the riverine site, and 70% at the ﬂoodplain site. The relatively high contribution of algal carbon to riverine food webs was a surprise, especially for a turbid tropical river that was considered not to support much in-stream productivity. This was a major change in thinking (cf. Maunsell and Partners, 1982), and brought the realization that factors such as increased turbidity (WRM, 2005, 2006) and labile dissolved copper (Stauber, 1995; Apte et al., 2009), all known to affect the productivity and species composition of aquatic algae, could be disrupting the supply of algal carbon to higher trophic levels, and could explain observed declines in ﬁsh catch and loss of species below D’Albertis junction (Swales et al., 1998, 1999, 2000; Storey et al., 2009). In addition, the study highlighted that forest dieback as a result of river bed aggradation and resultant increased frequency and duration of ﬂooding, as reported for the lower Ok Tedi and extending into the middle Fly (Pickup, 2009, Pickup and Cui, 2009) had the potential to affect the supply of riparian carbon to the food web. Therefore, this study by Bunn et al. (1999) brought the realization that combined impacts of the mine on the river channel had the potential to impact the delivery of fundamental sources of organic

598

A. W. Storey and M. Yarrao

carbon driving the aquatic food webs. This initiated a series of targeted studies using stable isotopes to test for some of the possible effects of mining operations. 17.4.2. Comparison of Contribution of Algal Carbon to Food Webs Above and Below D’albertis Junction The importance of algal carbon to the riverine food web (Bunn et al., 1999), and the known toxicity to algae of observed levels of labile copper in the middle Fly River (Stauber, 1995; Apte, 2009) resulted in a targeted stable isotope investigation in October 2003 (WRM, 2005). Sampling was conducted in the Fly River immediately above and below D’Albertis Junction (ARM450 and Kuambit, respectively) to compare the contribution of algal carbon to the aquatic food webs at each location. The study was designed to test the hypothesis that mine impacts had reduced the contribution of algal carbon, with the result that riverine food webs downstream of the mine were now more dependent on riparian (rainforest) inputs. Sampling involved the collection of as many primary sources (algae and terrestrial plants) and primary and higher consumers (prawns, aquatic insects, and ﬁsh) as possible at each site. The same collecting and processing methods as used by Bunn et al. (1999) were again employed, with stable isotopes of nitrogen and carbon analyzed with a continuous ﬂow-isotope ratio mass spectrometer, and the same two-point mixing model used to compare sources of carbon (energy) contributing to aquatic food webs at each site. A total of 172 and 191 samples were collected, covering 21 and 27 species of primary sources, 7 and 12 species of invertebrates (aquatic and terrestrial), and 17 and 21 species of ﬁsh at Kuambit and ARM450 respectively. Analyses indicated that the aquatic food web in the river channel downstream of the mine (Kuambit) had a reduced contribution of algal carbon compared with the food web above the inﬂuence of the mine (ARM450). Comparisons at the species level were constrained by there being few species occurring at both sites, and few replicates of each species, particularly at Kuambit. This likely reﬂects the gross impacts of the mine on resident species below D’Albertis Junction (i.e. increased TSS, bed aggradation, and elevated particulate and dissolved metal concentrations) rather than changes in resource availability, with large mine-related declines in ﬁsh catch reported from Kuambit reach since before 1990 (Swales et al., 1998, 2000; Storey et al., 2009). Even so, those species which did occur at

Aquatic Food Web Models for the Fly River, Papua New Guinea

599

80 (a) 60 40 1 2

2

2

6 A. berneyi

P. meraukensis

P. gulliveri

D. quadrifasciatus

L. calcarifer

2

A. macrorhynchus Z. novaeguinea (a) Z. novaeguinea (sa)

1

4

2

4 S.kreffti

1

Z. novaeguinea

(j)

Z. novaeguinea

(vj)

1

0

80 (b) 60 40 4

6

12

3

3 1 N. ater

11 2

M. splendida

20 1

1

1

6 M. sp. nov.

M. rosenbergii

O. herwerdinii

0 A. leptaspis

% carbon from algal sources

1

C. danielsi

10

T. chatareus

20

C. papuensis

% carbon from algal sources

both sites consistently had a more enriched signature (i.e. greater algal carbon contribution) at ARM450 than at Kuambit (Fig. 17.6). Similarly, for species that occurred only at one site, those at ARM450 consistently had a greater percent algal carbon contribution than species taken from Kuambit (Fig. 17.6). As is standard practice when comparing carbon isotope data across sites, particularly when carbon signatures of primary sources differ between sites, carbon isotope signatures of consumers were standardized using mixing models (Bunn et al., 1997, 1999; Phillips, 2001; Phillips et al., 2005), to

Figure 17.6: Mean (71 S.E.) percent contribution of algal carbon to species of ﬁsh and prawns that (a) occurred at ARM450 (&) or Kuambit (’), and (b) occurred at only ARM450 (&) and Kuambit (’), illustrating betweensite differences. Numbers ¼ number of specimens of each species sampled, ‘‘a’’ ¼ adult, ‘‘sa’’ ¼ sub-adult, ‘‘j’’ ¼ juvenile, ‘‘vj’’ ¼ very juvenile.

600

A. W. Storey and M. Yarrao

present data as percentage of carbon derived from algal sources. In this study, d13C signature of the algal sources at Kuambit were more enriched (�20.5) than at ARM450 (�23.8), whilst riparian sources had the same signatures across sites (�29.8 & �29.9 respectively). Photosynthesis in aquatic plants (algae, macrophytes etc.) utilizes dissolved CO2 which is obtained from a range of sources, including weathering of carbonate rock, mineral springs, the atmosphere, or from respired organic matter (Peterson and Fry, 1987). Variation in sources of CO2 is known to cause widely varying d13C values in algae, which, as a result may be subject to site speciﬁc and seasonal variations (Peterson and Fry, 1987; Peterson, 1999; Kankaala et al., 2006). Differences in algal signatures between Kuambit and ARM450 may reﬂect a difference in the dominant source of dissolved carbon available for assimilation during photosynthesis. As part of mine environmental management, limestone is continually fed into the waste rock stream to mitigate ARD risk in the mine area creeks, and this also inﬂuences buffering capacity in the Ok Tedi and middle Fly. The input of limestone may result in different carbonate content in the river downstream of the mine (i.e. Kuambit) compared with that coming from the upper Fly, which would likely exhibit a greater inﬂuence from respired organic matter (i.e. ARM450). Whether this could be sufﬁcient to modify the carbon signature of algae downstream of the mine is unknown. Alter natively, samples of algal at Kuambit were collected from higher on the banks than at ARM450 due to there being no visible algae on the lower banks/below the water line. Algae at Kuambit would therefore experience less frequent inundation, and as a result, may depend more on atmospheric CO2, compared with populations at ARM450 which were inundated, and therefore exposed to dissolved sources of carbon. However, as with riparian vegetation which depend on atmospheric CO2, it would be expected that algae at Kuambit would be more depleted than at ARM450, whilst the reverse was true. The difference in algal signatures between sites was not readily explicable, but may indicate a functional difference in the predominant source of CO2 being assimilated by algae during photosynthesis (i.e. atmospheric versus dissolved organic versus that derived from carbonate rock) and requires further investigation. The difference in algal carbon contribution between the two sites was supported by observations on the apparent algal cover whilst sampling. At ARM450 algal growth was visible along the banks and into the water (Fig. 17.7), on exposed unconsolidated mud banks, and on objects ﬂoating close to the water surface, however, at Kuambit there was no observable algae growth on these habitats except for small patches of algae higher up the banks. An obvious ‘‘tide’’ mark was apparent at Kuambit, with no apparent

Aquatic Food Web Models for the Fly River, Papua New Guinea

601

Figure 17.7: Algal growth extending into the water column, at ARM450

reach, with a photic zone extending B30 cm below water surface.

Figure 17.8: Exposed depositional river bank at Kuambit with no observable algal growth.

algal growth below the upper water level, reﬂecting recent past river stage height (Figs. 17.8–17.10). Based on observations in this study, WRM (2005) concluded that in-stream production is a minor, but signiﬁcant source of the algal carbon present in the food web of upper, forested riverine reaches of the Fly River. These ﬁndings supported those from the previous study by Bunn et al. (1999) and are in agreement with a recent review of how Australian tropical rivers function (Douglas et al., 2005). Furthermore, these data are consistent with the Riverine Productivity Model (RPM) of Thorp and Delong (1994), which

602

A. W. Storey and M. Yarrao

Figure 17.9: Exposed river bank at Kuambit showing algal growth above a

distinct ‘tide’ line.

Figure 17.10: River bank at Kuambit showing algal growth above a distinct ‘‘tide’’ line, but with no algae below and no observable algal growth in the water column. emphasizes the importance of local carbon inputs (i.e. in-stream primary production from phytoplankton, benthic algae, and other aquatic plants, and the local inputs of carbon in the form of leaf litter, fruits and terrestrial insects from the adjacent riparian zone) in providing energy (carbon) to river reaches. The study identiﬁed a reduction in the contribution of algal carbon downstream of the mine. Likely mine derived stressors affecting algal growth included chronic toxicity of algae (i.e. growth retardation) from labile dissolved copper, abrasion of algae due to increased TSS, limited primary

Aquatic Food Web Models for the Fly River, Papua New Guinea

603

production due to a reduced photic zone, and physical smothering of algal bioﬁlm and microhabitats by silt (WRM, 2005). 17.4.3. Comparison of Food Webs of Floodplain Habitats The most recent study to investigate possible mine effects on Fly River food webs using stable isotopes was conducted in February 2005 (WRM, 2006). Previously, mine-related impacts had been restricted to riverine habitats, with no detectable mine-related effects on the ﬂoodplain (Swales et al., 1998, 1999, 2000; Storey et al., 2009), however, a series of events in recent years have raised concerns about impacts spreading to the ﬂoodplain. First, OTML monitoring data have shown that ﬁsh catch in some oxbow lakes downstream of the mine have declined, particularly at Oxbow 6 in the Middle Fly at Kwem (OTML, unpub. dat.). In addition, aggradation of the river bed has progressed throughout the middle Fly resulting in overbank ﬂooding and extensive forest dieback throughout this area (Pickup, 2009, Pickup and Ciu, 2009), and ﬁnally, Acid Rock Drainage (ARD) has started to appear on levees in the middle Fly, with highly elevated levels of copper and other metals recorded in ﬂoodplain waters immediately adjacent to areas on ARD on the levees (Bolton et al., 2009). These events raised concern that declining ﬁsh catch on the ﬂoodplain may be a consequence of increased concentrations of bioavailable copper impacting either directly on ﬁsh species (i.e. acute toxicity of early life stages, or chronic toxicity causing reduced overall ﬁtness of populations) or indirectly via disruption to the food web. Such disruption could be to the base of the food web, in the form of reduced primary productivity (i.e. reduced algal growth) or other potentially sensitive food chain linkages, such as the zooplankton – Nematalosa herring link, or the periphyton – macroinvertebrate-ﬁsh pathway. In response to these concerns and hypotheses, a comparative study was implemented in February 2005 to compare aquatic food web structure and test for possible mine effects at three control forested oxbow lakes upstream of the inﬂuence of the mine (Drimdenasuk, Kiunga, and Moian oxbows), three exposed forested oxbow lakes downstream of the mine (Kuambit, Erekta, and Kwem (OXB06) oxbows), and two shallow lagoon systems (Bosset Lagoon and Lake Daviumbu). Using the same methods as Bunn et al. (1999) and WRM (2005), sampling involved the collection of as many primary sources (algae and terrestrial plants) and primary and higher consumers (prawns, aquatic insects and ﬁsh) as possible at each site. A total of 1164 individual samples were analyzed,

604

A. W. Storey and M. Yarrao

with an average of 145 samples per site (maximum of 166 at Oxbow 6 and minimum of 92 at Drimdenasuk oxbow), with up to ﬁve replicates per species, providing comparable sampling intensity per site as previous studies. Stable isotopes of nitrogen and carbon were then used to determine sources of carbon (energy) supporting the aquatic food webs at each site. Analysis of variance, with samples nested within sites, was used to compare carbon and nitrogen signatures of sources and consumers between exposed and control oxbow lakes, with sites as replicates. Mixing models, as per previous studies on the Fly River (Bunn et al., 1999; WRM, 2005), were then used to standardize data across sites to allow among-site comparisons in the contribution of algal and riparian carbon sources to the food webs at each site. Although sources were variable within sites, results indicated signiﬁcant changes in both the d13C and d15N signatures between exposed and control oxbows suggesting a shift in the food webs of the exposed oxbow lakes. Fauna of exposed lakes appeared to derive more algal carbon from periphyton sources (unicellular algae growing on submerged surfaces) over phytoplankton, with control oxbow lakes displaying the reverse. It appears that the die-off of rainforest along the shores of exposed oxbow lakes due to increased ﬂooding has reduce near-shore shading, facilitating the establish ment of submerged macrophytes (Figs. 17.11 and 17.12), and ﬂoating grasses (Fig. 17.13), and this has provided complex habitat and substrate for colonization by periphytic algae and aquatic macroinvertebrates. Signiﬁ cantly, ﬂoating grasses and macrophytes will rise and fall with changing

Figure 17.11: Collecting macroinvertebrates from submerged macrophytes

in Kuambit oxbow (Note degraded condition of rainforest in background).

Aquatic Food Web Models for the Fly River, Papua New Guinea

605

Figure 17.12: Beds of submerged macrophytes in Kuambit oxbow.

Figure 17.13: Encroaching margin of ﬂoating grass in Oxbow 6, where previously there was no such habitat due to shading by rainforest (note degraded condition of rainforest in background). water levels, providing a constantly inundated habitat, allowing establish ment of algal and invertebrate communities. The food chain link between periphyton and macroinvertebrates provides a carbon source to higher consumers (predatory invertebrates, small and large ﬁsh) and therefore, the periphytic carbon signature will be transferred through all levels of the food web (Douglas et al., 2005). In comparison, algae and primary consumers colonizing inundated rainforest, apart from contending with light limitation along the margins due to shading (see Fig. 17.14), will have to

606

A. W. Storey and M. Yarrao

Figure 17.14: Checking gill nets at Drimdenasuk oxbow. Note dense wall of rainforest along waters edge in background, and absence of any ﬂoating grasses. relocate as water levels change and new vegetation is inundated/exposed. As a result, phytoplankton-derived carbon appeared more dominant in control oxbow lakes, where near-shore shading by intact rainforest has limited the growth of ﬂoating grasses and macrophytes (Fig. 17.14), thereby limiting habitat for periphyton and aquatic invertebrates. A recent study of the macroinvertebrate fauna of the same lakes (A.W. Storey, unpublished data), showed signiﬁcantly greater abundance and species richness in the fauna at exposed lakes compared with control lakes, again reﬂecting the inferred increase in habitat diversity provided by the submerged macrophyte and ﬂoating grasses. Analyses also detected a signiﬁcant depletion of the d13C signature of periphyton and phytoplankton at exposed sites compared with control oxbow lakes (WRM, 2006), hypothesized to be related to the loads of organic material entering these lakes. Exposed oxbow lakes on the Fly River were characterized by extensive forest dieback (Pickup, 2009, Pickup and Ciu, 2009), resulting in heavy loads of decaying riparian vegetation in the ﬂooded margins. This high organic load would likely result in depleted d13C signatures of primary sources because CO2 respired during the breakdown of decaying vegetation is known to be depleted in 13C and as such algae assimilating this CO2 would also be depleted (A. Revill, CSIRO, pers. comm.). A similar ﬁnding was reported from the Keep River in the Kimberley, Western Australia, where zooplankton had depleted d13C values of up to �40 (Storey, unpublished data). It is likely that this was also due to the inﬂuence of respired CO2 since the Keep River site exhibited similar

Aquatic Food Web Models for the Fly River, Papua New Guinea

607

characteristics to exposed oxbow lakes; relatively deep, stratiﬁed oxygen levels, anoxic sediments, dense zooplankton blooms, and high organic loads. In other food web studies worldwide, depleted d13C have often been reported in lake zooplankton (i.e. Jones et al., 1999; Grey et al., 2000; Kankaala et al., 2006), and have been considered due to selective feeding on isotopically light photosynthetic algae (which have utilized dissolved inorganic carbon from microbial respiration) or feeding on isotopically light methanotrophic bacteria; the anaerobic decomposition of organic matter within the sediment can lead to high concentrations of isotopically light methane (CH4) in the hypolimnion (Kortelainen et al., 2000; Grey et al., 2004). The other main effect detected in the ﬂoodplain food web study was enrichment of the nitrogen signature of primary sources, in particular riparian vegetation and periphyton at Oxbow 6 (Fig. 17.15). Signatures of these sources were enriched (B10 d15N) in comparison with the other sites sampled in this study (and in comparison with this site sampled in 1998 by Bunn et al. (1999). Enriched nitrogen signatures of algae and macrophytes have been reported in other studies (i.e. Peterson, 1999; Udy and Bunn, 2001), and usually relates to enrichment of the site with nitrogen, often from sewage efﬂuent (Rau et al., 1981; EHMP, 2004). This is unlikely to be the reason for elevated nitrogen signatures in Oxbow 6, since there is no source of sewage efﬂuent at this site. However, forest dieback, with subsequent breakdown and release of nutrients from the organic loading may be sufﬁcient to result in this enrichment. Elevated nitrogen signatures were also observed in Bosset Lagoon and Lake Daviumbu, where there is a naturally large load of decaying grasses and macrophytes. Interestingly, nitrogen signatures of sources collected from Kuambit Oxbow were not enriched, yet this site has been more severely affected by forest dieback than Oxbow 6. However, this site was ﬁrst affected by dieback 5–7 years ago, and it is possible that the ‘‘ﬂush’’ of nutrients has passed through the system. The study identiﬁed a further disparity in the food web structure at Oxbow 6 compared with other forested oxbows, in that the enriched nitrogen signature of the primary sources (periphyton and riparian) placed them at a higher trophic position than the primary and secondary consumers supposedly dependent upon them for their carbon. Given that nitrogen goes through a predictable enrichment with each trophic transfer (Minagawa and Wada, 1984; Michener and Schell, 1999; Peterson, 1999), it does not seem possible that ﬁsh were deriving carbon from these sources. Phyto and zooplankton were not as enriched and were in an appropriate trophic position below primary consumers, however, it seems unlikely that plankton support the whole food web at this site. In addition, the carbon signature of plankton was relatively depleted compared with that of consumers, even

608

A. W. Storey and M. Yarrao

14

Oxbow Lake 6 (major trophic groups)

12

T. scrat A. lept

Δ15 N (‰)

10

T. chat

N. ater

8

POM

A. bern

S. kreff Fish (l)

A. agra (j) M. splen

S. jard

6 4 Nematalosa sp.

Phytoplankton (Theoretical)

2 0 -2 -41

14

-38

-35

-32 -29 Δ13 C (‰)

Fish

Invertebrates (Predatory)

Macrophytes

Riparian

Invertebrates (Primary)

Periphyton

Grass

Zooplankton

Algae

-26

-23

-20

Phytoplankton (Theoretical)

Drimdenasuk Oxbow (major trophic groups)

12 A. bern A. perc

10

Δ15 N (‰)

A. agra

8

C. rand

A. lep M. splen (j) Nematalosa sp.

6

N. ater

G. apri Zoop

4

Snails

2 Phytoplankton (theoretical)

POM

0 -2 -41

-36

-31

-26

-21

Δ C (‰) 13

Fish

Invertebrates (Predatory)

Macrophytes

Riparian

Invertebrates (Primary)

Periphyton

Grass

Zooplankton

Algae

Phytoplankton (Theoretical)

Figure 17.15: Representative bi-plots of d13C by d15N (means 795% CI) for an exposed (Oxbow 6) and a control (Drimdenasuk) oxbow, showing depleted carbon signature of plankton and enriched nitrogen signature of primary sources at Oxbow 6.

Aquatic Food Web Models for the Fly River, Papua New Guinea

609

allowing for a high level of fractionation with trophic transfer, making plankton unlikely to be the sole carbon source. As indicated, WRM (2006) have identiﬁed that the food web structure at Oxbow 6 appears to be altered relative to other oxbow sites sampled, but anomalies exist that this study was not able to readily explain. Although WRM (2006) identiﬁed signiﬁcant differences in food web structure between control and exposed oxbow lakes, the differences should not necessarily be interpreted as negative impacts, possibly more of a change in state. Exposed oxbow lakes may be becoming more like the shallow grassed lagoon systems (Bosset Lagoon and Lake Daviumbu) in terms of prevalence of grasses, macrophytes and associated periphyton and inverte brate fauna. The increased reliance of the food web on periphyton and the greater abundance and diversity of macroinvertebrates in exposed oxbow lakes may indicate these systems becoming more productive compared with unimpacted forested oxbows. Over time, recovery and regrowth of forest with ﬂood tolerant species may well shade-out the grasses and macrophytes currently encroaching from the shores returning the systems to their previous state. Similarly, the organic loading and depleted carbon signatures of plankton and periphyton may be a transitory state whilst the dead forest is processed.

17.5. Conclusions Knowledge of the biology and trophic interactions of the Fly River aquatic ecosystem has expanded greatly since the ﬁrst surveys conducted in the 1970s (Boyden et al., 1978; Roberts, 1978; Robertson and Baidam, 1983; DPI, 1979, 1980). In particular, the premine impact assessment by Maunsell and Partners (1982), and subsequent intensive collections of various data by OTML, from 1983 to present, have seen food web models evolve to include a greater number of species and interactions, and a greater overall under standing of trophic dependencies in riverine and ﬂoodplain habitats. Credit should be given, however to early investigators who only had limited data, but applied sound assumptions and experience from other systems, to generate relatively accurate outputs. More recent investigations initiated by OTML (Bunn et al., 1999; WRM, 2005, 2006) have adopted a process-related approach to understanding how the mine may be affecting the aquatic ecosystem, rather than attempting to ﬁnd patterns in ‘monitoring’ data (see Storey et al., 2009). Stable isotope analysis of carbon has proved particularly effective in the study of aquatic

610

A. W. Storey and M. Yarrao

food webs elsewhere (e.g. Rounick and Winterbourn, 1986; Peterson and Fry, 1987; Rosenfeld and Roff, 1992; Boon and Bunn, 1994; Douglas et al., 2005), and the Fly River is no exception. Stable isotopes have helped to clarify trophic relationships in riverine and ﬂoodplain food webs, identiﬁed the role that algae play as a carbon source, and have assisted in the detec tion of impacts to the bottom of the food web, which have ﬂow-on effects to higher trophic levels. As with many scientiﬁc investigations, however, as one question is answered, others appear. Issues such as quantifying the contribution of in-stream algal production (sensu Thorp and Delong, 1994; RPM) versus supplementation of riverine food webs with ﬂoodplain produc tion (sensu Junk et al., 1989; Flood-pulse concept) have not been resolved. In addition, studies to date (WRM, 2005, 2006) have been snap-shots in time, which provide no information on the degree of temporal variation in isotopic signatures. It is not known how consistent the observed patterns are within and between sites over time, which limits the ability to differentiate mine-related impacts from natural variability. The hypothesis of a mine-related effect on riverine organic carbon and its inﬂuence on algal carbon signatures above and below the mine could be investigated further, as could the conundrums in Oxbow 6 related to possible nutrient-mediated enrichment of nitrogen signatures of riparian and algal sources and depleted CO2-driven signatures of algal sources. No doubt, as investigations continue, some of these questions will be answered – but others will arise.

ACKNOWLEDGMENTS The authors thank all present staff of the Environment Department, OTML who assisted in the collection of dietary and ﬁsh catch data. In particular, the authors thank Kent Hortle who processed the majority of stomach samples prior to 1987 to produce the very extensive database on ﬁsh stomach contents, and Barre Kare and Ian Roderick who were responsible for the majority of the post-1987 stomach content analyses. Stuart Bunn and Ross Smith are thanked for contributions to food web analyses through their respective studies summarized in this chapter. Andy Revill, CSIRO, Tasmania is thanked for his insights, discussions and advice on incon sistencies in the stable isotope data. Jess Lynas is thanked for formatting and revising the draft. Finally, past coordinators of the biological monitoring programs at Ok Tedi are acknowledged for their role in directing the collection of data and inﬂuencing the project; David Balloch (1982–1985), Kent Hortle (1983–1987), Ross Smith (1988–1992), Andrew Storey (1993–1995), Stephen Swales (1996–1998), Charles Tenakanai (1998–2001),

Aquatic Food Web Models for the Fly River, Papua New Guinea

611

and Markson Yarrao (2001–present). This chapter is in memory of Barre Kare and Kayeman Bakowa, dedicated biologists and Papua New Guineans who contributed much to this story, and had much more to offer, but sadly are no longer with us.

REFERENCES ANZECC/ARMCANZ (2000). Australian and New Zealand guidelines for fresh and marine water quality. www.ea.gov.au/water/nwqms/index.html Apte, S. C. (2009). Biogeochemistry of copper in the Fly River. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 321–373. Bolton, B., Pile, J., & Kundapen, H. (2009). Texture, geochemistry, and mineralogy of sediments of the Fly River System. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 51–112. Boon, P. I., & Bunn, S. E. (1994). Variations in the stable-isotope composition of aquatic plants and their implication for food-web analysis. Aquatic Botany, 48, 99–108. Bowles, K. C., Apte, S. C., Maher, W. A., Kawei, M., & Smith, R. E. W. (2001). Bioaccumulation and biomagniﬁcation of mercury in Lake Murray, Papua New Guinea. Canadian Journal of Fisheries and Aquatic Science, 58, 888–897. Boyden, C. R., Brown, B. E., Lamb, K. P., Druckers, R. F., & Tufts, S. J. (1978). Trace elements in the Upper Fly River, Papua New Guinea. Freshwater Biology, 8, 189–205. Bunn, S. E., Loneragan, N. R., & Kempster, M. A. (1995). Effects of acid washing on stable isotope ratios of C and N in seagrass and penaeid shrimp: Implications for food web studies using multiple stable isotopes. Limnology and Oceanography, 40, 622–625. Bunn, S. E., Davies, P. M., & Kellaway, D. M. (1997). Contributions of sugar cane and invasive pasture grass to the aquatic food web of a tropical lowland stream. Marine and Freshwater Research, 48, 173–179. Bunn, S. E., Tenakenai, C. D., & Storey, A. W. (1999). Energy Sources Supporting Fly River Fish Communities. Unpublished report to Ok Tedi Mining Ltd., 38 pp. Carpenter, S. R., Kitchell, J. F., & Hodgson, J. R. (1985). Cascading trophic interactions and lake productivity. Bioscience, 35, 634–639. Carpenter, S. R., Kitchell, J. F., Hodgson, J. R., Cochran, P. A., Elser, J. J., Elser, M. M., Lodge, D. M., Kretchmer, D., He, X., & von Ende, C. N. (1987). Regulation of lake primary productivity by food web structure. Ecology, 68, 1863–1876. Carpenter, S. R., Kraft, C. E., Wright, R., He, X., Soranno, P. A., & Hodgson, J. R. (1992). Resilience and resistance of a lake phosphorus cycle before and after food web manipulation. The American Naturalist, 140, 781–798.

612

A. W. Storey and M. Yarrao

De Niro, M., & Epstein, S. (1978). Inﬂuence of diet on the distribution of carbon isotopes in animals. Geochimica et Cosmochimica Acta, 42, 495–506. Douglas, M. M., Bunn, S. E., & Davies, P. M. (2005). River and wetland food webs in Australia’s wet-dry tropics: General principles and implications for manage ment. Marine and Freshwater Research, 56, 329–342. DPI (1979). Fisheries Survey of the Ok Tedi Mining Region. In: Fisheries Research Annual Report. Fisheries Division Department of Primary Industry. DPI (1980). Ok Tedi Fisheries. In: Fisheries Research Annual Report. Fisheries Division Department of Primary Industry. EHMP (2004). Ecosystem Health Monitoring Program 2002–2003. Annual technical report. Moreton Bay Waterways and catchments Partnership, Brisbane City Council, Brisbane, Qld., Australia. France, R., Chandler, M., & Peters, R. (1998). Mapping trophic continua of benthic food webs: Body size-del15N relationships. Marine Ecology Progress Series, 174, 301–306. France, R. L. (1995). Critical examination of stable isotope analysis as a means for tracing carbon pathways in stream ecosystems. Canadian Journal of Fisheries and Aquatic Sciences, 52, 651–656. Grey, J., Jones, R., & Sleep, D. (2000). Stable isotope analysis of the origins of zooplankton carbon in lakes of differing trophic state. Oecologia, 123, 232–240. Grey, J., Kelly, A., & Jones, R. I. (2004). High intraspeciﬁc variability in carbon and nitrogen stable isotope ratios of lake chironomid larvae. Limnology and Oceanography, 49, 239–244. Hortle, K. G. (1986). A Review of Biological Sampling of the Ok Tedi and Fly River systems, April 1983 to June 1986. OTML Report ENV86-9. Jones, R. I., Grey, J., Sleep, D., & Arvola, L. (1999). Stable isotope analysis of zooplankton carbon nutrition in humic lakes. Oikos, 86, 97–104. Junk, W. J., Bayley, P. B., & Sparks, R. E. (1989). The ﬂood-pulse concept in riverﬂoodplain systems. In: Dodge, D. P. (Ed.), Proceedings of the International Large River Symposium LARS. Canadian Special Publication of Fisheries and Aquatic Sciences, 106, 110–127. Kankaala, P., Taipale, S., Grey, J., Sonninen, E., Arvola, L., & Jones, R. I. (2006). Experimental d13C evidence for a contribution of methane to pelagic food webs in lakes. Limnology and Oceanography, 51, 2821–2827. Kortelainen, P., Huttunen, J. T., Va¨isa¨nen, T., Mattsson, T., Karjalainen, P., & Martikainen, P. (2000). CH4, CO2 and N2O supersaturation in 12 Finnish lakes before and after ice melt. Verhandlungen Internationale Vereinigung fu¨r Theoretische und Angewandte Limnologie, 27, 1410–1414. Lajtha, K., & Michener, R. H. (1994). Stable Isotopes in Ecology and Environmental Science. Blackwell Scientiﬁc Publications, Oxford, UK. Lancaster, J., Bradley, D. C., Hogan, A., & Waldron, S. (2005). Intraguild omnivory in predatory stream insects. Journal of Animal Ecology, 74, 619–629.

Aquatic Food Web Models for the Fly River, Papua New Guinea

613

Maunsell & Partners (1982). Ok Tedi Environmental Study: Supporting Studies, Volume 6: Ok Tedi-Fly River Aquatic Survey-Heavy Metal Review. Unpublished report to Ok Tedi Mining Limited, 344 pp. Michener, R. H., & Schell, D. M. (1999). Stable Isotopes in Ecology and Environmental Science. Blackwell Scientiﬁc Publications, Oxford, UK, pp. 138–156. Minagawa, M., & Wada, E. (1984). Stepwise enrichment of 15N along food chains: Further evidence and the relation between d15N and animal age. Geochimica et Cosmochimica Acta, 48, 1135–1140. Moore, R. (1982). Spawning and early life history of barramundi Lates calcarifer in Papua New Guinea. Australian Journal of Marine and Freshwater Research, 33, 647–662. Moore, R., & Reynolds, L. F. (1982). Migrations patterns of barramundi Lates calcarifer in Papua New Guinea. Australian Journal of Marine and Freshwater Research, 33, 671–682. OTML (1988). Sixth Supplemental Agreement Environmental Study, 1986–1988. Final Draft Report, Volume 1. Report prepared by Ok Tedi Mining Limited, November 1988. OTML (1990). APL Compliance and Additional Environmental Monitoring Program. Unpublished report to the State of Papua New Guinea by Ok Tedi Mining Limited. Parametrix, I., & URS Greiner Woodward Clyde (1999a). Assessment of Human Health and Ecological Risks for Proposed Mine Waste Mitigation Options at the Ok Tedi Mine, Papua New Guinea. Detailed Level Risk Assessment. Parametrix, Inc., URS Greiner Woodward Clyde for Ok Tedi Mining Ltd., Washington. Parametrix, I., & URS Greiner Woodward Clyde (1999b). Assessment of Human Health and Ecological Risks for Proposed Mine Waste Mitigation Options at the Ok Tedi Mine, Papua New Guinea. Screening Level Risk Assessment. Parametrix, Inc., URS Greiner Woodward Clyde for Ok Tedi Mining Ltd., Washington. Peterson, B. J. (1999). Stable isotopes as tracers of organic matter input and transfer in benthic food webs: a review. Acta Oecologica, 4, 479–487. Peterson, B. J., & Fry, B. (1987). Stable isotopes in ecosystem studies. Annual Review of Ecology and Systematics, 18, 293–320. Phillips, D. L. (2001). Mixing models in analyses of diet using multiple stable isotopes: A critique. Oecologia, 127, 166–170. Phillips, D. L., Newsome, S. D., & Gregg, J. W. (2005). Combining sources in stable isotope mixing models: Alternative methods. Oecologia, 144, 520–527. Pickup, G. (2009). Floodplain inundation modeling and forecasting for the Middle Fly. In: B. Bolton (Ed.). The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 291–318. Pickup, G., & Cui, Y. (2009). Modeling the impact of tailings and waste rock disposal on the Fly River System. In: B. Bolton (Ed.). The Fly River, Papua New

614

A. W. Storey and M. Yarrao

Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 257–289. Power, M. E. (1990). Effects of ﬁsh in river food webs. Science, 250, 811–814. Power, M. E. (1992). Top–down and bottom–up forces in food webs: Do plants have primacy. Ecology, 73, 733–746. Power, M. E., Rau, M., & Tenakanai, C. D. (1994). Preliminary Description of Food Web Structure of the Fly River, Papua New Guinea, Based on Inferences from Stable Isotopes. Unpublished report to Ok Tedi Mining Limited, 17 pp. Power, M. E., Parker, G., Dietrich, W. E., Sun, A., & Wootton, J. T. (1995). Hydraulic food chain models: An approach to the study of food web dynamics in large rivers. Bioscience, 45(3), 159–167. Rau, G. H., Sweeney, R. E., Kaplan, I. R., Mearns, A. J., & Young, D. R. (1981). Differences in animal 13C, 15N and D abundance between a polluted and unpolluted coastal site: Likely indicators of sewage uptake by marine food webs. Estuarine Coastal and Shelf Science, 13, 701–707. Reynolds, L. F., & Moore, R. (1982). Growth rates of barramundi Lates calcarifer in Papua New Guinea. Australian Journal of Marine and Freshwater Research, 33, 663–670. Roberts, T. R. (1978). An Ichthyological Survey of the Fly River in Papua New Guinea, with Descriptions of New Species. Smithsonian Contribution to Zoology, Vol. 281. Smithsonian Institution Press, Washington. Robertson, C. H., & Baidam, G. (1983). Fishes of the Ok Tedi area with notes on ﬁve common species. Science in New Guinea, 10, 16–26. Rosenfeld, J. S., & Roff, J. C. (1992). Examination of the carbon base in southern Ontario streams using stable isotopes. Journal of the North American Benthological Society, 11, 1–10. Rounick, J. S., & Winterbourn, M. J. (1986). Stable carbon isotopes and carbon ﬂow in ecosystems. BioScience, 36, 171–177. Schnetzer, A., & Caron, D. A. (2005). Copepod grazing impact on the trophic structure of the microbial assemblage of the San Pedro Channel, California. Journal of Plankton Research, 27, 959–971. Sinistro, R., Sa´nchez, M. L., Marinone, M. C., & Izaguirre, I. (2007). Experimental study of the zooplankton impact on the trophic structure of phytoplankton and the microbial assemblages in a temperate wetland (Argentina). Limnologica, 37, 88–99. Smith, R. E. W., Ahsanullah, M., & Batley, G. E. (1990). Investigations of the impact of efﬂuent from the Ok Tedi copper mine on the ﬁsheries resource in the Fly River, Papua New Guinea. Environmental Monitoring and Assessment, 14, 315–331. Smith, R. E. W., & Hortle, K. G. (1991). Assessment and predictions of the impacts of the Ok Tedi copper mine on ﬁsh catches in the Fly River system, Papua New Guinea. Environmental Monitoring and Assessment, 18, 41–68. Stauber, J. L. (1995). Toxicity testing using marine and freshwater unicellular algae. Australian Journal of Ecotoxicology, 1, 15–24.

Aquatic Food Web Models for the Fly River, Papua New Guinea

615

Storey, A. W., & Smith, R. E. W. (1998). Food Webs for the Fly River Fish Fauna. Unpublished report for Ok Tedi Mining Ltd by Wetland Research and Management and R and D Environmental, 27 pp. Storey, A. W., Yarrao, M., Tenakanai, C., Figa, B., & Lynas, J. (2009). Use of changes in ﬁsh assemblages in the Fly River system, Papua New Guinea, to assess effects of the Ok Tedi copper mine. In: Bolton, B. (Ed.), The Fly River, Papua New Guinea: Environmental Studies in an Impacted Tropical River System, Vol. 9. Elsevier, Amsterdam, pp. 427–462. Swales, S., Storey, A. W., Roderick, I. D., Figa, B. S., Bakowa, K. A., & Tenakanai, C. D. (1998). Biological monitoring of the impacts of the Ok Tedi copper mine on ﬁsh populations in the Fly River system, Papua New Guinea. The Science of the Total Environment, 214, 99–111. Swales, S., Storey, A. W., Roderick, I. D., & Boga, S. F. (1999). Fishes of ﬂoodplain habitats of the Fly River system, Papua New Guinea, and changes associated with El Nin˜o droughts and algal blooms. Environmental Biology of Fishes, 54, 389–404. Swales, S., Storey, A. W., & Bakowa, K. A. (2000). Temporal and spatial variations in ﬁsh catches in the Fly River system in Papua New Guinea and the possible effects of the Ok Tedi copper mine. Environmental Fish Biology, 57, 75–95. Thorp, J. H., & Delong, M. D. (1994). The riverine productivity model: An heuristic view of carbon resources and organic matter processing in large river systems. Oikos, 70, 305–308. Udy, J. W., & Bunn, S. E. (2001). Elevated D15N values in aquatic plants from cleared catchments: Why? Marine and Freshwater Research, 52, 347–351. West, J. B., Bowen, G. J., Cerling, T. E., & Ehleringer, J. R. (2006). Stable isotopes as one of nature’s ecological records. Trends in Ecology and Evolution, 21, 408–414. Wood, I. B., Day, G. M., Storey, A. W., & Markham, A. J. (1994). Environmental Monitoring and Research Programs at the Ok Tedi Copper Mine. Proceedings of the Mining and Environmental Management Conference, Nevada, USA, pp. 85–98. Wootton, J. T., & Power, M. E. (1993). Productivity, consumers, and the structure of a river food chain. Proceedings of the National Academy of Science, 90, 1384–1387. WRM (1998). Review of Dietary Data for Fish from the Fly River System: A Precursor to Constructing a Food Web. Unpublished report by Wetlands Research and Management to Ok Tedi Mining Limited, 28 pp. WRM (2005). Comparison of Carbon Sources Supporting Aquatic Food Webs Above and Below D;Albertis Junction. Unpublished report to Ok Tedi Mining Ltd by Wetland Research and Management, 42 pp. WRM (2006). Stable Isotope Analysis of Food Web Structure of Fly River Floodplain Waterbodies: February 2005. Unpublished report to Ok Tedi Mining Limited by Wetland Research and Management. WRM (2007). Floodplain Impact Study: Assessment of Ecological Health of Control and Exposed Forested Oxbow Lakes Using Plankton and Aquatic Invertebrate Assemblages. Unpublished report to Ok Tedi Mining Limited by Wetland Research & Management. February 2007.

Subject Index Acid rock drainage (ARD), 87–90, 321, 325, 344, 366–367, 600, 603 acid neutralizing capacity (ANC), 87–90 maximum potential acidity (MPA), 87 net acid producing potential (NAPP), 87–88 oxidation, 88, 103 risk, 79, 82, 87–88, 108 sulfur, 79, 82, 87–88, 108 sulfuric acid, 87 water quality testing, 88 Acid-soluble copper (ASM-Cu), 332–333, 337–338, 359, 363–364, 366 Adsorption, 325, 328, 339–340, 347, 357, 359–360, 364 Agu River, 11, 15, 94, 100, 126, 291 Algae, 380, 382–384, 385–386, 391, 402 epiphytic, 593 ﬁlamentous, 593 Aluminum, 82, 84, 90 Amphibians, 562–564 ANC (see Acid neutralizing capacity (ANC)) Animals, 552, 557, 560, 564, 566, 569–570 Archbold Expedition, 494 ARD (see Acid rock drainage) Bacteria, 380–383, 391, 393 Barramundi, 411–415, 417, 420–423, 434, 578 Beetles, 496–498 Bioassay, 294, 379–389, 391, 393, 401, 404 Bioavailability, 377–380, 383, 388, 395 Biomass, 448 (see also Fish catch) Birds, 557–560 Bornite, 95, 101, 103, 326 Bosset Lagoon, 10, 15, 37, 82, 325, 328, 330, 332, 362, 385, 389, 436, 444, 446, 448, 450, 523, 525–526, 581, 603, 607, 609 Butterﬂies, 498–502 Carbon cycle, 592 Catﬁsh eel-tailed, 580 fork-tailed, 580 Chalcocite, 87, 95, 101, 103, 326

Chalcopyrite, 83, 87, 95, 103, 326 Chemolithotrophic bacteria, 337–338, 345 Cicadas, 503 Climate, 5, 155, 180, 515, 517, 529–530, 549 Clinoform, 177–179, 181, 187, 192–193, 199 Coleoptera, 496–498 Colonization, 538–543 Competence, of ﬂow, 62, 65, 67 Copper, 13, 51, 62, 69, 73–74, 77, 79, 82, 84–87, 90, 95, 103, 108, 113, 116–117, 120–121, 124–127, 134, 137–138, 143, 145, 169, 171, 206, 213, 221–224, 232, 235–244, 321, 323, 325, 375, 377–378, 429, 436, 438, 449, 463, 464, 576–577, 590, 597–598, 602–603 bioavailability, 337, 367, 369, 375, 377 complexation capacity (CuCC), 330, 333–336, 339–340, 344, 356–357, 362, 365, 367, 377 dissolved, 325, 327–328, 334, 337–347, 350–362, 366–367 mobilisation, 330, 337, 345–346, 351 solubility, 327–328, 337 sulﬁde, 326, 333–334, 336–338, 364 toxicity, 326, 377–378, 380, 383–384, 387, 389, 392–399, 401, 442 Covellite, 95, 101, 103, 326 Crocodiles, 552–553, 560 D’Albertis, Luigi, 494, 496, 498, 502 D’Albertis Junction, 63, 340 Daviumbu, Lake, 385, 446, 448, 523, 603, 607, 609, Delta, Fly, 154–176, 179, 181, 183, 208, 217, 220, 222, 243–244, 350, 417, 515, 530–538 (see also Estuary) Deposition, 154, 162, 169, 257–261, 265–266, 270–272, 274–285 Depositional web, 116, 135, 142, 144 Desorption, 339–340, 357 Dieback effects, 556 Diet, of ﬁsh, 420–421 Digenite, 326 Diptera, 502–503

618

Subject Index

Dissolved organic carbon (DOC), 323, 330, 356–357 Dissolved organic matter (DOM), 330, 333–334, 369 Ecological effects, 569–571 Ecosystems, 533, 538, 544 El Nin˜o, 5, 343–344, 350, 362, 431 Environmental monitoring, 462 Ephemeroptera, 506 Estuary, of Fly, 69, 323, 350–354, 360, 417, 517, 530–537 Fauna, 549–569 macroinvertebrate, 606 Fish assemblages, 427–459 biomass, 429, 433, 435, 449 change in assemblage composition, 453–455 ﬂoodplain sites, 451–455 riverine sites, 453–455 change in biodiversity, 433–447 change in ﬁsh catch, 449–453 ﬂoodplain sites, 450–453 riverine sites, 450 chi-square contingency table analysis, 433, 436, 444, 448 exotic ﬁshes, 457–459 monitoring ﬁsh populations, 435–436, 444, 448, 450, 453 multivariate analyses, 433 sampling method, 428 Fisheries, 421–423 Fish fauna dietary intake, 577 Fish habitat, 465 backwaters, 465 bed aggradation, 461, 463 Ok Tedi, 465 riparian vegetation, 475 snags, 467–471, 473–474, 476, 481–483 Flies, 502–503 Flooding, 292–293, 302, 306, 315, 522–523, 526, 538–540 Floodplain, 292–309, 313–315, 515, 520, 522–524, 526–530, 539–540 Floodplain deposition, 116, 124, 126, 133–134, 138–139, 141 Flowering plants, 543–544 Fly River, 3, 14–15, 17, 25, 36, 38, 41–42, 51, 69, 72–73, 154–156, 163–164, 169–171, 178–181, 183, 192, 291, 294, 296, 300, 377, 380, 384–388, 402, 404, 462, 576 alluvial plain, 84

backwater effect, 67–68 backwater zone, 59, 68 catchment area, 425 geochemistry of pre-mine river-deposited sediments, 69 grain size, 61 Food chain, 583 Food webs, 549–569, 575–610 algal carbon, 598–603 ﬂoodplain habitats, 603–609 use in environmental assessment, 575–577 use of dietary data, 581–588 use of stable isotopes, 591–609 Forest dieback, 344, 367, 369, 520–521, 538–539 Frogs, 562–563 Genetics, 413–414 Geochemical modelling, 360–367 Geographical Society, 494, 496, 498 Geomorphology, 5–13, 25–47 Grain size, 59, 121–123, 140–143 cause of downstream ﬁning, 65–68 distribution, 59, 67 selective deposition, 67 Grassland, 523–525, 529 Gulf of Papua, 159–161, 165, 179–181, 183, 189–190, 194, 196–197, 199, 208–209, 211, 213, 215, 217, 219, 221, 224–225, 232, 235, 238–241, 243, 360 Habitats, 551–553 ﬂoodplain, 582, 585 riverine, 585 Harvey Creek, 6, 26–27, 29–32, 53, 62, 79, 84 Heavy mineral, 79, 83 chalcopyrite, 83, 87, 95, 103, 326 copper minerals, 95 magnetite, 83, 95 pyrite, 79, 83, 87, 90, 95–97, 99–100, 103–104, 108 Hemiptera, 503–505 Herring, 580, 587 Hindenburg Wall, 53 Hydraulic modeling, 300–307, 312 Hydrology, 14–24 Hymenoptera, 505 Impacts, 51, 73, 576 acid rock drainage (ARD), 447 bed aggradation, 62, 447, 463, 468, 597 dissolved copper, 462

Subject Index

ﬂooding, 292–293, 302, 306, 315, 522–523, 526, 538–540 forest dieback, 447, 468, 478, 597 loss of habitat, 463 monitoring, 53 Insects, 493–506 Invertebrates, 564–565 Isoptera, 505 Kiunga, 64, 69, 494–495, 499, 502–503, 505 Kuambit, 5, 14, 23, 37, 41, 47, 126, 140, 269–270, 273–274, 294, 301, 307, 361, 431, 450, 474, 480, 598–600, 602–603, 607 Lagaip River, 65 Lake Murray, 83, 493, 495, 499, 502 d13C signatures, 597 food web, 597 mercury, 83 Landslides, 519, 541–543 Lates calcarifer (see Barramundi) Lepidoptera, 498–502 Macroinvertebrates, 593 Mammals, 554–557 Manda, 5, 10–11, 14, 22, 37, 41, 82 Mangroves, 530–538 Mantodea, 505–506 Mass Balance, 214, 228, 238–239, 241 Maximum potential acidity (MPA), 87 Metal geochemistry, 68–90 after start of mining, 73–90 baseline, 72 before mining, 69–73 channel bars, 68, 72 compared to average crustal abundance, 73 source of metals, 82–84 temporal and spatial changes in, 84–86 Middle Fly ﬂoodplain, 323, 344–345 Migrations, 413, 417, 422–423 Mine area creeks, 51, 53 landslides, 53 sediment types, 55 Mine-derived sediments/particulates, 323–325, 328, 330, 333–334, 336–337, 340, 344, 350–352, 360–361, 364, 367, 369, 522, 539–540 Mineralogy, 91–108 Al-bearing clay, 84 clay, 101–102 kaolinite, 84

619

Mine waste, 5, 17, 51, 53, 62, 64, 84, 87, 103, 113, 116, 123, 137, 139–140, 143, 207, 258, 263, 269, 272, 285, 292, 315, 361, 367, 429, 465, 475, 482 Mining impact, 3, 34–35, 37 Mixing models, 604 Monitoring biological, 576 Montane vegetation, 515, 517–520 Mosquitoes, 502–503 Moths, 498–502 Mount Bosavi, 83 Mount Fubilan, 51, 69, 73, 461 Ok Tedi Mining Limited (OTML), 51 orebody, 69, 83, 103 overburden, 103 tailings, 87 MPA (see Maximum potential acidity (MPA)) Mullet, 580 NAPP (see Net acid producing potential (NAPP)) Natural organic matter, 334, 362, 367 Net acid producing potential (NAPP), 87–88 Northern Coral Sea, 208, 222, 224, 231, 234–235, 240–244 Obo, 5, 22–23, 73, 82, 84, 125, 144, 271, 274–275, 283, 294, 300, 307, 325, 343, 361–362, 398, 417, 422, 436, 450, 526, 528–529 Odonata, 506 Ok Gilor, 5, 26, 31, 53, 69 Ok Kam, 73 Ok Ma, 69 Ok Mabiong, 53 Ok Mani, 53, 72 Ok Menga, 69 Ok Tedi, 51, 68, 72, 257–283, 286, 576 gravel-bedded, 68 gravel-sand transition zone, 59 metal enrichment, 79 sand zone, 59 trace metal, 69 waste, 62 Ok Tedi mine, 5, 7, 25, 169, 238, 243, 257, 259, 326, 361, 429, 456, 458, 463, 493, 495, 499, 503, 541–542, 558–559, 570, 576–577, 580, 583 Ok Tedi Mining Limited (OTML), 57 Orthoptera, 504–505 Oxbow 6, 603, 607 Oxidation, 326, 333–334, 336–339, 344–345, 347, 362, 364, 367

620

Subject Index

Pangua, Lake, 446 Particulate copper, 117, 121, 123–125, 127–130, 325, 328, 333, 337, 340, 344–345, 347, 350, 352, 359, 361–363, 365–366 Partition coefﬁcient (Kd), 359, 364, 367 Phasmatodea, 504–505 Porewater copper concentrations, 344–345 Porewater peepers, 344–345 Porgera mine, 65, 448 Porphyry, 83 Prawns, 580 macrobrachium, 594 Pyrite, 79, 83, 87, 90, 95–97, 99–100, 103–104, 108

Spawning, 414–415, 417, 420, 423 Stable isotopes, 577 d13C, 590–594 d15N, 590–594 Star Mountains, 51, 495 Stock structure, 413–414 Stratigraphy, 153–154, 156, 164–172 Strickland River, 61, 65, 72–73, 77, 84, 323, 325–326, 364, 463, 494 Sulﬁde Creek, 53 Sulfur, 79 enrichment, 79

Rainforest, 520–523, 529, 534, 541 Reproduction, 414–416 Reptiles, 560–562 Riparian vegetation, 595

Tabubil, 495–504 Tailings, 3, 25–28, 36, 51, 87, 90–91, 108, 113, 139, 207, 214, 238–239, 243–244, 257–258, 261, 264–265, 269, 277, 279, 323, 328, 330, 332–333, 338, 361–362, 375, 429, 463–465, 576 Tailings leaching studies, 328 Termites, 495–496, 505 Terrestrial ecosystem, 551–552, 555, 564–565, 568 Texture, 53–68 after start of mining, 61–68 before mining, 53–61 grain size distribution curves, 55, 58 Tide-dominated delta, 153, 156–159, 160 Trophic cascades, 575–576 Trophic interactions, 576

Savannah, 515, 529–530 Sediment, and sedimentation, 155, 162, 167–171, 205, 208, 213, 210–214, 217, 221–230, 232–238, 240–244, 522–523, 539 accumulation, 177–179, 181–183, 185–187, 190, 193, 196–197 deposition rates, 115–116, 123–142 Sediment mineralogy, 91–108 after start of mining, 91–108 before mining, 91 chlorite, 105 clay mineralogy, 101–102 comparison with earlier studies, 105–108 controls on mineralogical composition, 102–105 heavy minerals, 94 illite, 107 kaolinite, 102 micas, 105 mineral distribution, 99 mineralogy of levees, 98–101 mineralogy of riverbed sediments, 92–98 tributaries, 105 Sediment modeling, 258–259, 283 Snakes, 553, 560–561 Sorting, 59

Vegetation, 515–544 Waste, 5, 17, 51, 53, 62, 64, 84, 87, 103, 113, 116, 123, 137, 139, 140, 143, 207, 258, 263, 269, 272, 285, 292, 315, 361, 367, 429, 465, 475, 482 (see also Mine waste) Waste disposal, 461 Waste rock, 257–258, 260–261, 263–265, 277, 279 Weathering, 65, 67, 84 chemical, 65, 84 physical, 65 role in determining grain size, 67

Developments in Earth & Environmental Sciences, 7

Climate, Environment, and Society in the Pacific during the Last Millennium, Volume 6 (Developments in Earth and Environmental Sciences) (Developments in Earth and Environmental Sciences)

The Fly River, Papua New Guinea, Volume 9: Environmental Studies in an Impacted Tropical River System (Developments in Earth and Environmental Sciences)

Developments in Earth & Environmental Sciences, 7

Climate, Environment, and Society in the Pacific during the Last Millennium, Volume 6 (Developments in Earth and Environmental Sciences) (Developments in Earth and Environmental Sciences)

Environmental Oriented Electrochemistry. Studies in Environmental Sciences, Volume 59

Physical Processes in Earth and Environmental Sciences

Papua New Guinea

Oil Pollution and its Environmental Impact in the Arabian Gulf Region, Volume 3 (Developments in Earth and Environmental Sciences)

Economic Valuation of River Systems (New Horizons in Environmental Economics)

Gender Analysis in Papua New Guinea

Peatlands, Volume 9: Evolution and Records of Environmental and Climate Changes (Developments in Earth Surface Processes)

Geosciences, Environment and Man (Developments in Earth and Environmental Sciences, Volume 1)

Fish versus Power: An Environmental History of the Fraser River (Studies in Environment and History)

Antarctic Climate Evolution, Volume 8 (Developments in Earth and Environmental Sciences)

Managing Animals in New Guinea: Preying the Game in the Highlands (Studies in Environmental Anthropology, 7)

Environmental Foresight and Models Developments in Environmental Modelling

Environmental Contamination (Studies in Environmental Science)

Environmental Radioanalysis (Studies in Environmental Science)

Grand Challenges in Environmental Sciences

Environmental Sciences)

Environmental Sciences)

Global Tropical Cyclogenesis, Second Edition (Praxis Books in Environmental Sciences)

Grand Challenges in Environmental Sciences

Advances in Research in Karst Media (Environmental Earth Sciences)

Environmental Sciences)

Neutron applications in earth, energy and environmental sciences

Commercial and Business Organizations Law in Papua New Guinea

Neutron applications in earth, energy and environmental sciences

And Then the Engines Stopped: Flying in Papua New Guinea

In the River

In the River

River in the Sea

The Fly River, Papua New Guinea, Volume 9: Environmental Studies in an Impacted Tropical River System (Developments in Earth and Environmental Sciences)

Developments in Earth & Environmental Sciences, 7

Climate, Environment, and Society in the Pacific during the Last Millennium, Volume 6 (Developments in Earth and Environmental Sciences) (Developments in Earth and Environmental Sciences)

Environmental Oriented Electrochemistry. Studies in Environmental Sciences, Volume 59

Physical Processes in Earth and Environmental Sciences

Papua New Guinea

Oil Pollution and its Environmental Impact in the Arabian Gulf Region, Volume 3 (Developments in Earth and Environmental Sciences)

Economic Valuation of River Systems (New Horizons in Environmental Economics)

Gender Analysis in Papua New Guinea

Peatlands, Volume 9: Evolution and Records of Environmental and Climate Changes (Developments in Earth Surface Processes)

Geosciences, Environment and Man (Developments in Earth and Environmental Sciences, Volume 1)

Fish versus Power: An Environmental History of the Fraser River (Studies in Environment and History)

Antarctic Climate Evolution, Volume 8 (Developments in Earth and Environmental Sciences)

Managing Animals in New Guinea: Preying the Game in the Highlands (Studies in Environmental Anthropology, 7)

Environmental Foresight and Models Developments in Environmental Modelling

Environmental Contamination (Studies in Environmental Science)

Environmental Radioanalysis (Studies in Environmental Science)

Grand Challenges in Environmental Sciences

Environmental Sciences)

Environmental Sciences)

Global Tropical Cyclogenesis, Second Edition (Praxis Books in Environmental Sciences)

Grand Challenges in Environmental Sciences

Advances in Research in Karst Media (Environmental Earth Sciences)

Environmental Sciences)

Neutron applications in earth, energy and environmental sciences

Commercial and Business Organizations Law in Papua New Guinea

Neutron applications in earth, energy and environmental sciences

And Then the Engines Stopped: Flying in Papua New Guinea

In the River

In the River

River in the Sea

Recommend Documents