Forests and Dragonflies
FORESTS
AND
DRAGONFLIES
Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005
1
2 Adolfo Cordero Rivera
This page intentionally left blank
Forests and Dragonflies
FORESTS
AND
DRAGONFLIES
Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005
Edited by
Adolfo Cordero Rivera
Sofia–Moscow 2006
3
4 Adolfo Cordero Rivera
FORESTS
AND
DRAGONFLIES
Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005 Edited by Adolfo Cordero Rivera
First Published 2006 ISBN-10: 954-642-278-9 ISBN-13: 978-954-642-278-1 Pensoft Series Faunistica No 61 ISSN 1312-0174
Cover: A male of Macromia splendens perched amongst tree roots in a road that goes through a Quercus forest near the river Avia, in NW Spain. The small picture shows three trophic levels in a forest stream in Central Ialy: leaves of Acer campestre (producer), a female of a Heptageniid mayfly (herbivore) and the shade of a male of Calopteryx haemorrhoidalis (predator). Photos A. Cordero.
© PENSOFT Publishers All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the copyright owner.
Pensoft Publishers Geo Milev Str. 13a, Sofia 1111, Bulgaria
[email protected] www.pensoft.net
Printed in Bulgaria, October 2006
Forests and Dragonflies
5
CONTENTS Introduction: Dragonflies as forest-dependent animals ADOLFO CORDERO RIVERA Forests as habitats for dragonflies (Odonata) CORBET, P.S.
7
13
Allochthonous organic matter as a food resource for aquatic invertebrates in forested streams GRAÇA, M.
37
THE IMPORTANCE OF FORESTS FOR DRAGONFLIES IN DIFFERENT CONTINENTS Odonata in Bornean tropical rain forest formations: diversity, endemicity and implications for conservation management ORR, A.G. The importance of forests to neotropical dragonflies PAULSON, D.
79
Use of forest and tree species, and dispersal by giant damselflies (Pseudostigmatidae): future prospects in fragmented forests FINCKE, O.M. Thoughts from Africa: how can forest influence species composition, diversity and speciation in tropical Odonata? DIJKSTRA, K.-D. & CLAUSNITZER, V.
127
Specialists vs. generalists among dragonflies - the importance of forest environments in the formation of diverse species pools SAHLÉN, G. Dragonfly habitat maps based on landcover and habitat relation models TSUBAKI, Y. & TSUJI, N.
181
51
153
103
6 Adolfo Cordero Rivera
CONSERVATION
AND
BEHAVIORAL ISSUES
Threat levels to odonate assemblages from invasive alien tree canopies SAMWAYS, M. Movement behaviours of a forest odonate in two heterogeneous landscapes 225 TAYLOR, PH. The structure of the Coenagrion mercuriale populations in the New Forest, southern England 239 THOMPSON, D.J. & WATTS, PH.C. Mate location and competition for mates in relation to sunflecks of forest floors 259 WATANABE, M. Differences in immune ability in forest habitats of varying quality: dragonflies as study models 269 CÓRDOBA-AGUILAR, A. & CONTRERAS-GARDUÑO, J. The present role and future promise of conservation genetics for forest Odonates 279 HADRYS, H.
209
Forests and Dragonflies Adolfo Cordero Rivera (ed) 2006 Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 7-12.
7
© Pensoft Publishers
Sofia–Moscow
Introduction: Dragonflies as forest-dependent animals Adolfo Cordero Rivera In a famous paragraph of the Origin of species Darwin (1859) describes the effect of cats on the abundance of red clover (Trifolium pratense) with these words: “Humble-bees alone visit red clover, as other bees cannot reach the nectar. [...] Hence we may infer as highly probable that, if the whole genus of humble-bees became extinct or very rare in England, the heartsease and red clover would become very rare, or wholly disappear. The number of humble-bees in any district depends in a great measure upon the number of field-mice, which destroy their combs and nests; and Col. Newman, who has long attended to the habits of humble-bees, believes that “more than two-thirds of them are thus destroyed all over England.” Now the number of mice is largely dependent, as every one knows, on the number of cats; and Col. Newman says, “Near villages and small towns I have found the nests of humble-bees more numerous than elsewhere, which I attribute to the number of cats that destroy the mice.” Hence it is quite credible that the presence of a feline animal in large numbers in a district might determine, through the intervention first of mice and then of bees, the frequency of certain flowers in that district!”
These indirect effects of one trophic level on organisms that apparently have no connection may well be the first example of a trophic cascade in an ecosystem (Polis et al. 2000) that has explicitly been proposed. Since Darwin, many studies have detected trophic cascades in terrestrial and aquatic ecosystems, but my favourite is one example involving dragonflies that affect the abundance of flowers near ponds, by preying on their insect pollinators (Knight et al. 2005) (Fig. 1). Near fishless ponds, dragonflies are more abundant, and flowers scarce, but if fish are present, they reduce dragonfly larvae, and hence adult populations, allowing insect pollinators to be more common and increasing plant reproductive success. Dragonflies are top predators, first in the
8 Adolfo Cordero Rivera
Fig. 1. Trophic cascade among ecosystems. In fishless ponds, dragonflies are voracious predators of flying insects, having a significant effect on pollinators. Dragonflies reduce pollinator visits to flowers, and hence determine a diminution of fructification and plant abundance. The introduction of fish in the pond drastically reduces dragonfly larval density. This has cascading (positive) effects over flying insects and plants near the pond. Direct effects in white and indirect effects in gray. Source: based on Knight et al. (2005).
aquatic habitat as larvae, and then as adults (Corbet 1999). They are good candidates for cascading effects, as this example shows. Over the world, forests are disappearing at an accelerating rate (FAO 1997), sometimes substituted by monocultures of fast-growing tree species, converted into agricultural land, industrial soil, and urban areas. Forests are ecosystems that provide diverse habitats for a range of organisms, including dragonflies and other animals, that at a first sight seem not to depend on forests. For instance, Macromia splendens, one of Europe’s most endangered dragonflies, uses forest roads as hunting places, and larvae are sometimes found amongst tree roots (Cordero Rivera et al. 1999; Cordero Rivera 2000). Therefore, in this example, native forests are essential for the maintenance of this species. Forests are feeding places for many dragonflies, and movements between ponds and forests are therefore continuous. If roads are constructed between these two landscape elements, mortality of dragonflies by collision with vehicles can be surprisingly high (Riffell 1999). As the authors of this book show, dragonflies are highly dependent on forest cover
Forests and Dragonflies
9
and composition, and this is true from the boreal forests to the tropics. Dragonflies can even be good indicators of forest structure and composition of tropical forests (Clausnitzer 2003), and vascular plants in boreal landscapes (Sahlén & Ekestubbe 2001). Tree plantations near streams have been shown drastically to affect odonata assemblages, because they shade most of the stream (Kinvig & Samways 2000), and forestry practices, notably logging, affect odonate communities of nearby lakes in Scandinavia (Sahlén 1999). The aim of this book is therefore to explore the ways in which forests affect dragonfly life, and to show that forests are much more than places where timber is produced. The multifunctional approach to forest ecology is now badly needed all over the world, and forest managers need to acknowledge that timber production no longer holds primacy. Other forest functions, such as the conservation of biodiversity and the regulation of the water cycle, are of equal importance (Lindenmayer & Franklin 1997). In other words: when managing a forest plot, rather than concentrating on what can be removed from the plot, the focus should be on what will be left after logging (Hammond 1997). In many regions of the world, calopterygids are the most clearly forestdependent damselflies (Córdoba-Aguilar & Cordero Rivera 2005), and for this reason Calopteryx haemorrhoidalis has been selected for the cover of this book. The importance of riparian forests for Calopteryx is so crucial that these species can be absent from rivers that lack forests, or can even have a different wing morphology in deforested areas (Taylor & Merriam 1995). Nevertheless, if forests are too dense and sunny places are absent for most of the day, adults may avoid such river sections. The presence or absence of riparian forests can even change reproductive behaviour, as previous work on C. haemorrhoidalis in Central Italy has shown: areas with dense forest cover were avoided and most males assembled in a small sunny area on the stream, resulting in such high densities that males could not court females due to the interference of rivals, and forced matings with ovipositing females (Cordero 1999). Nevertheless, when a section of the forest was felled, males dispersed over a large area, reducing density and increasing the proportion of matings that were preceded by courtship (Cordero Rivera & Andrés 2002). Dragonflies are sun-loving insects, and most of them use forests for only a part of their adult life. In the first chapter of this book, Philip Corbet describes the importance of forests as habitats for dragonflies. In temperate forests, dragonflies perch in open areas, but many use forests for aestivation, unlike in tropical regions, where forests are suitable as a permanent adult habitat. In Mediterranean regions, riparian forests may be the only habitat patches with temperatures suitable for adults, and they clearly have a direct effect on river invertebrates, as the review by Manuel Graça describes in the second chapter of the book.
10 Adolfo Cordero Rivera
The first part of the book describes the importance of forests for dragonflies in different continents. Chapters include an analysis of Bornean forest Odonata by Albert Orr, and a description of odonate communities in neotropical forests by Dennis Paulson. Orr estimated that at least 70 % of the Bornean odonate fauna is presently confined to forest habitats and probably depends on forest for its survival. Paulson arrives at the same figure when estimating the proportion of Costa Rican odonates that depend on forests. Ola Fincke presents evidence for the effect of forest fragmentation on the maintenance of populations of giant damselfies (Pseudostigmatidae) in Panama. She concludes that, as predators of phytotelm mosquito larvae, some of which are disease vectors, the demise of pseudostigmatids may affect not only forest food chains, but also human health. The African forest Odonata are studied by KlaasDouwe Dijkstra & Viola Clausnitzer. Their analysis suggests that low insolation in forest habitats and interspecific competition are key factors segregating forest and non-forest species. The Nordic perspective is presented by Göran Sahlén. His elegant study shows for instance that in constructed wetlands under 10 years of age those close to forest habitats (even small clumps of trees) had, on average, more than twice as many breeding Odonata species than those in more open areas. This is another unexpected example of the importance of forests for dragonflies. The last chapter in this part addresses a methodological issue. If we visit a pond and record the species present, it is likely that we shall miss some breeding species. Therefore we need to repeat sampling until we have a “complete” list. Yoshitaki Tsubaki and Nobuyuki Tsuji develop a method designed to obtain reliable lists of presence and absence from Odonata inventories, and using this method discovered that, in Japan, the area of broad-leaved forest within a grid-square (10x10 km) had positive effects on the occurrence of 57 species, indicating that at least 50% of dragonflies depend on forest. The second part of the book explores the implications of this forestodonate relationship for conservation biology and behavioural problems. Michael Samways reviews the efforts made by South-African government to restore native riparian vegetation by an extensive removal of invasive exotic tree species. The recovery of some endangered odonate species as a direct result of alien tree removal has been remarkable. The chapter by Philip Taylor summarizes his work on landscape ecology of riparian forests and Calopteryx maculata. His results suggest that C. maculata move more extensively in open habitats when compared to the more closed, forested landscape, a finding that has implications for landscape-scale population structure. David Thompson and Philip Watts analyse the structure of Coenagrion mercuriale populations in the New Forest, England, using a mark-recapture and molecular approach. This species lives in small streams in heathlands, and is a nonforest odonate (whose main English populations are inside the New Forest
Forests and Dragonflies
11
National Park!). But, even for sun-loving species like C. mercuriale, forests are important landscape elements, because they can act as barriers to dispersal and probably provide shelter areas for tenerals. Mature males of some odonates tend to remain in forest. To locate females in forests, males mainly perch in sunflecks (sunlit sites on the forest floor) and adopt a sit-and-wait tactic. Mamoru Watanabe describes the territorial behaviour of males of the damselfly Platycnemis echigoana, at sunflecks in climax deciduous forests of Japan. For species like this, forest cover is crucial because it not only provides rendezvous (sunflecks) but also oviposition substrates. In their chapter Alex Córdoba and Jorge Contreras review the potential use of dragonflies for testing current ideas of differences in immune ability related to habitat quality. They suggest that wing pigmentation of forest species like Calopteryx is a good indicator of habitat quality that promises to be of relevance for future studies. In the last chapter of the book, Heike Hadrys, Viola Clausnitzer and Linn Groeneveld present the first genetic analysis of afrotropical forest odonates. Their study of three species of Pseudagrion revealed strong inter- and intraspecific differences in population genetic patterns, and indicated that the natural isolation of the mountain forests has resulted in radiation of P. bicoerulans in at least three significant units of conservation. Therefore, forests not only provide habitats for odonates; they also contribute to speciation for those odonates that are true forest species. This book is the result of a special session of invited talks on the relationship between dragonflies and forests, held during the 4th Worldwide Dragonfly Association Symposium of Odonatology, at the Forestry School of the University of Vigo (Pontevedra, Spain), in July 2005. The editor whises to thank all contributors and participants in the Symposium for their enthusiasm for the project, their sharing of lively ideas and the warm atmosphere during the meeting (even when it was raining…). Many thanks also to Philip Corbet who greatly improved the draft of this Introduction. This book could not have been possible without financial support from the University of Vigo, the Spanish Ministry of Science and Education (grant CGL2004-21004-E), and the government of Galicia, through the Departments of Environment, Innovation and Industry and Universities and Education. Our sincere thanks are due to all the people and agencies involved in the project.
REFERENCES CLAUSNITZER, V. 2003. Dragonfly communities in coastal habitats of Kenya: indication of biotope quality and the need of conservation measures. Biodiversity and Conservation 12: 333-356.
12 Adolfo Cordero Rivera
CORBET, P. S. 1999. Dragonflies. Behaviour and ecology of Odonata. Harley Books, Colchester, UK. CORDERO, A. 1999. Forced copulations and female contact guarding at a high male density in a Calopterygid damselfly. Journal of Insect Behavior 12: 27-37. CORDERO RIVERA, A. 2000. Distribution, habitat requirements and conservation of Macromia splendens Pictet (Odonata: Corduliidae) in Galicia (NW Spain). International Journal of Odonatology 3: 73-83. CORDERO RIVERA, A. & ANDRÉS, J. A. 2002. Male coercion and convenience polyandry in a Calopterygid damselfly (Odonata). Journal of Insect Science 2: 14 - Available online: insectscience.org/2.14. CORDERO RIVERA, A., UTZERI, C. & SANTOLAMAZZA CARBONE, S. 1999. Emergence and adult behaviour of Macromia splendens (Pictet) in Galicia, northwestern Spain (Anisoptera: Corduliidae). Odonatologica 28: 333-342. CÓRDOBA-AGUILAR, A. & CORDERO RIVERA, A. 2005. Evolution and ecology of Calopterygidae (Zygoptera: Odonata): Status of knowledge and future research perspectives. Neotropical Entomology 34: 861-879. DARWIN, C. 1859. The origin of species. John Murray, London. FAO 1997. State of the world’s forests. Roma: Food and Agriculture Organization of the United Nations (FAO). HAMMOND, H. 1997. What is ecoforestry? Global Biodiversity 7: 3-7. KINVIG, R. G. & SAMWAYS, M. J. 2000. Conserving dragonflies (Odonata) along streams running trough commercial forestry. Odonatologica 29: 195-208. KNIGHT, T. M., MCCOY, M. W., CHASE, J. M., MCCOY, K. A. & HOLT, R. D. 2005. Trophic cascades across ecosystems. Nature 437: 880-883. LINDENMAYER, D. B. & FRANKLIN, J. F. 1997. Re-inventing the discipline of forestry a forest ecology perspective. Australian Forestry 60: 53-55. POLIS, G. A., SEARS, A. L. W., HUXEL, G. R., STRONG, D. R. & MARON, J. 2000. When is a trophic cascade a trophic cascade? Trends in Ecology & Evolution 15: 473-475. RIFFELL, S. K. 1999. Road mortality of dragonflies (Odonata) in a Great Lakes coastal wetland. Great Lakes Entomologist 32: 63-73. SAHLÉN, G. 1999. The impact of forestry on dragonfly diversity in central Sweden. International Journal of Odonatology 2: 177-186. SAHLÉN, G. & EKESTUBBE, K. 2001. Identification of dragonflies (Odonata) as indicators of general species richness in boreal forest lakes. Biodiversity and Conservation 10: 673-690. TAYLOR, P. D. & MERRIAM, G. 1995. Wing morphology of a forest damselfly is related to landscape structure. Oikos 73: 43-48.
CyanMagentaYellowBlack Odonata page 13
Forests habitats for dragonflies (Odonata) Adolfo Cordero Rivera (ed)as2006 Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 13-36.
13
© Pensoft Publishers
Sofia–Moscow
Forests as habitats for dragonflies (Odonata) Philip S. Corbet I.C.A.P., University of Edinburgh, Scotland, U.K. Present address: Crean Mill, St. Buryan, Cornwall TR196HA, U.K.
[email protected]
ABSTRACT The ways in which forests can be inferred, or shown, to meet the habitat requirements of dragonflies are reviewed globally. The relationship between dragonflies and forests is examined along a latitude spectrum in the Northern Hemisphere, from the Arctic Circle to the equator, a transect along which species diversity progressively increases, and the microclimate within forest becomes steadily more permissive for occupancy by the several stages in the dragonfly life history. In mid-temperate latitudes dragonflies use forests mainly for aestivation as prereproductive adults, a strategy functionally similar to the siccatation exhibited by tropical dragonflies in seasonal-rainfall regions. Tropical rainforest is the planet’s most diverse terrestrial ecosystem, with regard to species and habitats. It provides habitats for many species of dragonflies, for some or all of their life- history stages. Many such species, and their behaviour and ecology, remain undescribed. For biologists, including odonatologists, the foremost challenge of our time is that this irreplaceable storehouse of biological information faces imminent threat of destruction before its contents can be placed on record. Key words: Odonata, dragonfly, forests, habitat requirements, latitude, tropical rainforest, species diversity.
INTRODUCTION I warmly commend the organisers on their inspired choice of an opening theme for the 4th WDA Symposium of Odonatology. This has obliged all
CyanMagentaYellowBlack Odonata page 14
14 Philip S. Corbet
participants to view dragonflies from a global perspective and to reflect on their relationship with one of the planet’s most conspicuous, threatened and biologically diverse terrestrial ecosystems. Dragonflies are top predators among insects. In order to maintain their life style, they must consume large numbers of animals, and they expend a lot of energy while doing so. As we shall see, dragonflies use forests in many different ways. Or, to put this another way: without forests, a high proportion of the world’s dragonflies would not exist. This fact has searching implications for the future of dragonfly species diversity and its progressive erosion by anthropological impact. My topic is less specialised than those of other contributors; so I shall try to make a strength of its general approach by offering a template against which we can view the special adaptations, in morphology and behaviour, which dragonflies exhibit in their exploitation of the sylvan environment worldwide. Respecting the advice I used to give to students, I begin by defining some key terms. A forest is a climax ecosystem dominated and characterised by a dense growth of trees, namely woody, upright plants growing close together and generally more then 5 m high. Such an ecosystem has its own characteristic microclimate, different from the environment outside it. A habitat is a place, part of an ecosystem, in which a given species, in one or more of its developmental stages, lives. With my prospective readership in mind, I shall not presume to define the word dragonfly, except to stress that I use it to mean a member of the order Odonata and not merely the suborder Anisoptera. My approach today will be to revisit the known habitat requirements of dragonflies and then to view forests as habitats for dragonflies across a spectrum, from latitudinal tree line to the equator, placing special emphasis on the microclimate offered by forests; then, for forests of high-temperate latitudes, to consider (1) correlations between the distribution of dragonflies and forests; (2) the microclimate within forest and (3) the ways in which this microclimate might be expected to meet the habitat needs of dragonflies; and then (4) for temperate latitudes, to consider examples of dragonfly behaviour that support or illustrate such inferences. Next, with this reasoning as a basis, I examine the use of forests by dragonflies in mid-temperate latitudes. Finally, and separately, I review what is known about the use made by dragonflies of tropical rainforest.
HABITAT REQUIREMENTS OF DRAGONFLIES Attributes that any habitat, including a forest, must possess in order to meet the requirements of dragonflies are as follows: 1. A microclimate that permits effective thermoregulation by adults.
CyanMagentaYellowBlack Odonata page 15
Forests as habitats for dragonflies (Odonata)
15
2. A milieu for successful foraging by adults. 3. Provision for nocturnal roosting or daytime shelter for adults from inclement weather and from predators. 4. Provision of, or proximity to, a water body suitable as a reproductive site and as a locus for larval development and survival. To obtain food and to reproduce, the adult dragonfly needs to fly, for which it is obliged to maintain a certain minimum thoracic temperature, namely the temperature that permits spontaneous flight.
THE LATITUDE SPECTRUM Forests exist in different types, each encircling the planet in a broad swathe bounded by limits of latitude. The full range of forest types is evident only in the Northern Hemisphere where land exists as far as, and beyond, latitudinal tree line. High temperate latitudes 1. Clues from the distribution of dragonflies. The forest type that extends to tree line is the boreal, coniferous forest, also termed taiga. Dragonflies do not extend much beyond latitudinal or altitudinal tree line; this fact alone suggests that dragonflies need forests as habitats. Indeed, most dragonflies of Europe were probably originally “forest species”, inasmuch as the natural postglacial ground cover was forest. In the climax state such forest was probably never completely closed and dark, but interspersed with clearings, ponds, rivers etc. (Wildermuth 2005a). A caveat to bear in mind is that (as might be expected) some species may use forests in different ways at different latitudes or depending on the weather. For example, even Boyeria irene may be regarded as a forest species in the Mediterranean region (Wildermuth 2005b) and adults of Somatochlora flavomaculata patrol completely in the shade when the ambient temperature exceeds 32oC (Wildermuth 2005a). In western Europe Somatochlora metallica seems to be confined to wooded areas (Schorr 1990). In mountainous regions of Central Europe (except at higher altitudes in the Alps) in the Black Forest Aeshna caerulea is “regionally stenotopic”, i.e. in contrast to its occurrence in the taiga and tundra it is restricted to bogs surrounded by coniferous forests; and during days featuring high insolation and high temperature, adults move to the cool environment of forest (Sternberg & Sternberg 2000). Elesewhere, dragonfly species diversity tends to be correlated with the presence of forest. So that the use of forests by dragonflies can be inferred by analysing the effect on odonate diversity of forest destruction.
CyanMagentaYellowBlack Odonata page 16
16 Philip S. Corbet
In Central Sweden (ca 60o30’N) about 30 species of Odonata (9 Zygoptera and 21 Anisoptera) occupy lakes surrounded by boreal forest; odonate species diversity there correlates with aquatic plant richness (Sahlén 1999; Sahlén & Ekestubbe 2001). If riparian forest is logged, odonate diversity is reduced, especially among partivoltine species. These findings establish a correlation between the presence of forest and odonate diversity but do not reveal the behavioural or ecological links that bring this about. Further south, in Minnesota (ca 47o28’N) Rith-Najarian (1998) found that odonate species diversity (among 39 species, all but one being Anisoptera) in a midcontinental mixed-hardwood-coniferous forest was greatest in oldgrowth forest and least in recently cleared areas. Mature second-growth biomes possessed intermediate odonate species diversity. Recovery of odonate diversity after logging was slow, taking several decades after the disturbance. The detailed causes of diversity loss (i.e. the direct effects on dragonflies) were not identified, but it was noted that logging affected aquatic habitats by damming streams and making ponds stagnant. It was further noted that stenotypic species might have been differentially affected by loss of trees and shade. As in the example from Central Sweden, a clear correlation existed between odonate species diversity and the presence of undisturbed forest, and again the causal links were not revealed.
Fig. 1. Readings of incident light intensity from above the canopy to ground level in raised wet forest in southern Uganda, at about 00o07’N. A platform at 120 ft on the tower from which the readings were taken causes the first significant drop in illumination and the canopy causes the second. (Modified from Corbet 1964.)
CyanMagentaYellowBlack Odonata page 17
Forests as habitats for dragonflies (Odonata)
17
2. The microclimate inside forest. An important feature of a forest, at any latitude, is its microclimate. Inside a forest (compared with outside) incident solar radiation (and with it light intensity) are reduced, especially towards ground level (Fig. 1), air movement is less, and saturation deficit (the inverse of humidity) is usually less (Fig. 2). Because the canopy intercepts radiation, it prevents the forest from becoming very cold on clear nights whereas during daylight it reduces the ambient temperature, especially near the ground (Fig. 3), obliging dragonfly adults of some species to bask in sunflecks and forage in places where they can be exposed to sunlight, namely clearings and forest edges. The ambient temperature at ground level in forest will be latitude-dependent and is unlikely to permit spontaneous activity by adult dragonflies except at low latitudes.
Fig. 2. The diel march of saturation deficit measured in different places within and outside raised wet forest in Uganda, at about 00o07’N. Means are derived from ten rainless days. (Modified from Corbet 1963.)
CyanMagentaYellowBlack Odonata page 18
18 Philip S. Corbet
Fig. 3. The diel march of ambient temperature measured in different places within and outside raised wet forest in Uganda, at about 00o07’N. Means are derived from ten rainless days. (Modified from Corbet 1963.)
3. How might dragonflies be using forests? Despite the dearth of relevant observations, it is reasonable to assume that adult dragonflies near tree line use coniferous forest mainly, perhaps exclusively, for shelter from wind and predators and for nocturnal roosting, and perhaps for mid-air foraging within forest clearings and in lee sites at forest margins. Because of the temperature differential, it is unlikely that adults frequent the forest itself during daytime, even for foraging. I have been unable to find descriptions of the physical environment within and adjacent to taiga but we can be sure that air temperature within taiga will be much reduced by the shading effect of trees, which will also mitigate the most severe effects of high wind. Thus the value of forests to dragonflies near tree line may be confined to physical shelter (for avoiding wind-chill,
CyanMagentaYellowBlack Odonata page 19
Forests as habitats for dragonflies (Odonata)
19
for roosting and for foraging in lee sites and glades) and perhaps as a source of small flying insects. (Sometimes it appears that the density of the small flying insects that form potential prey for dragonflies is greater inside forest, although this remains to be confirmed and may depend on local conditions.) It is reasonable to infer that even taiga provides habitat components essential for dragonflies. We may note, however, that, taiga reduces incident solar radiation. So we may expect to find adult dragonflies compensating for this by basking on insolated surfaces. Indeed, the habit of resting on exposed ground, even when the sun’s elevation is low, exploits the reduced frequency of temperature inversions which is a feature of high latitudes during the summer (Corbet 1969). 4. Illustrative examples of forest use. Here I explore the observed use of forests by dragonflies, in the light of the inferences above, by reviewing behavioural observations made at different temperate latitudes. These are only examples and, although illustrative, may not all be representative or typical. All, however, throw light on the ways in which dragonflies use forests. A species whose distribution broadly coincides with latitudinal or altitudinal tree line is Somatochlora sahlbergi, studied in Fennoscandia (Valle 1931; Sahlén 1994), in the Yukon (Cannings et al. 1991; Cannings & Cannings 1985), and in Southern Siberia (Belyshev 1973; Kosterin 1992), where this species occurs chiefly in mountains. In the Yukon S. sahlbergi is invariably found within 100 km or so of latitudinal tree line and usually within 300 m of altitudinal tree line. Larvae develop in clear, cold, deep, mossy ponds (Fig. 4), usually surrounded by spruce woods where adults seek shelter (e.g. during copulation and nocturnal roosting, Sahlén 1994), also doing so sometimes within stands of dwarf birch (Kosterin 1992) which likewise harbour other insects. On the tundra of northern Sweden a tandem that had formed beside water flew towards a birch wood; and at night adults seemed to rest in nearby birch forest (Sahlén 1994). Adults of Aeshna caerulea at about 56oN were reported to fly 300-1600m from their reproductive sites on exposed moorland to nocturnal roosting sites in neighbouring forest (Smith 1995). At about 51o37’N, in southern England, habitats favoured by populations of Cordulia aenea are within extensive woodland, have scattered bankside trees, feature sections of the bank with open, sunny margins, a moderate accumulation of (broad-leaved) leaf litter and areas of open water with floating-leaved aquatic plants (Brooks et al. 1995). In ponds among mixed beech woodland, larvae were invariably found amongst large fragments of leaf litter, a microhabitat that would exist only where broad-leaved deciduous trees grew along the margins. Once again, however, we encounter regional variation in habitat occupancy: in continental Europe and Japan larvae of this species often occur beneath Sphagnum mats (Ubukata 1984; Schorr 1990). The maiden flight in southern England was directed up to tree-top height and then
CyanMagentaYellowBlack Odonata page 20
20 Philip S. Corbet
Fig. 4. Larval habitat of Somatochlora sahlbergi: an oxbow pond of the Blackstone River, just below altitudinal tree line in TheYukon Territory, Canada at 65o4’N and ca 840 m a.s.l., June 1982. The pond, where many adults were copulating and ovipositing, contained numerous larvae. The Arctic tundra is visible beyond the spruce taiga flanking the pond. (From Corbet (1999); photograph by Lynn and Rich Moore.)
away into the surrounding woodland. Adults also flew to alight on tree tops to complete copulation. The study by Brooks et al. (1995) in southern England demonstrates an intimate functional relationship between certain populations of C. aenea and woodland, but obviously cannot permit one to draw general conclusions about the species, which in nearby continental Europe (ca 48o30’N) occupies a wide variety of moorland habitats and is by no means confined to woodland (Sternberg & Schmidt 2000). In a population studied in Switzerland the maiden flight of C. aenea was directed towards nearby forest and not to adjacent treeless areas (Ha et al. 2002). In the Austrian Alps copulating pairs of Somatochlora arctica perch mostly in sunlit spots at the edges of pine trees at heights between 0.8 and 12 m (Wildermuth 2003). Mid-temperate latitudes Towards lower latitudes forests become more diverse biologically and more permissive microclimatically, and they come to include a progressively higher proportion of broad-leaved, deciduous angiosperms, until at the lowest latitudes we encounter the richest ecosystem on the planet – tropical rainfor-
CyanMagentaYellowBlack Odonata page 21
Forests as habitats for dragonflies (Odonata)
21
est (TRF). Almost all metazoan groups (including Odonata) that have been studied exhibit progressively greater species diversity along this latitude spectrum , from tree line to the equator, although not necessarily to the same extent (Lawton et al. 1998). Surveys of forests in Western Europe have shown them to harbour diverse odonate faunas, typical examples being the Forest of Rambouillet (at 46o40’N) with 46 species (Arnaboldi 1997) and the Forest of Notre-Dame (48o45’N) with 31 species (14 Zygoptera, 17 Anisoptera) (Le Calvez 1998). These observations support the view that forests provide habitats useful to Odonata but, again, do not tell us how. In contrast, the use made by dragonflies of the forest ecosystem in southern Japan at 35o05’N and in northeastern Algeria at 36o51’N in climates featuring a hot, dry summer can be securely inferred. In these places species belonging to the Lestes-Sympetrum-Aeshna mixta complex (sensu Landmann 1985) clearly use forests as refuges in which to aestivate. One example is Lestes temporalis, studied by Uéda & Iwasaki (1982) in southern Honshu, Japan. Here the benefit of aestivating in forest is compellingly illustrated by the profile of the adult daily survival rate during the three months of the prereproductive period (Fig. 5). In the population studied this approximated to the extraordinarily high value of 0.996. This is even higher than the daily survival rate of a population of Hetaerina cruentata, recorded as 0.978 by Córdoba-Aguilar (1994) in virgin riparian forest in Veracruz, Mexico (at ca
Fig. 5. Survivorship during adult life of Lestes temporalis in southern Honshu, Japan at about 35o N. Survivorship, especially of males, remains extremely high during the prereproductive period throughout which adults aestivate in woodland. (Modified after Uéda and Iwasaki (1982).)
CyanMagentaYellowBlack Odonata page 22
22 Philip S. Corbet
19o30’N). When adults of Lestes temporalis became reproductively active (and left the forest) their daily survival rate fell abruptly. Another example is provided by three species of Anisoptera (also members of the same complex) in northeastern Algeria (Samraoui et al. 1998): Aeshna mixta, Sympetrum meridionale and S. striolatum. The species involved are typically univoltine in the warmer parts of their range, emerging from lowland habitats in late spring or early summer when their reproductive sites are already becoming dry. In Algeria adults repair to cork-oak woodland (featuring Quercus suber L.) at an altitude of 500-1000 m where they spend the next three or four months, sheltering and foraging, while attaining sexual maturity. Then, in late summer or early autumn, they leave forest, descend en masse to the lowlands, where they reproduce (Samraoui et al. 1998). The use of the forest as a refuge and protection against high temperature and saturation deficit is obvious in these examples from Japan and Algeria. In our progression along the latitude spectrum these examples foreshadow the life-cycle patterns exhibited by certain tropical Odonata in seasonalrainfall areas (see below) in which the long-lived adults repair to forest to spend the dry season, a process termed ‘siccatation’, not aestivation, with which, strictly speaking, it is not phenologically equivalent (Corbet 1999). Tropical latitudes Tropical rainforest (TRF) encircles the planet between the Tropics of Cancer and Capricorn as a belt about 5,000 kilometres wide. Although this ecosystem features by far the greatest species diversity of Odonata, it has been proportionately little studied. In reviewing the ways in which dragonflies use tropical forest, we have relatively few studies to call upon. Most derive from brief glimpses that do not embrace the changing seasons. Two accounts provide exceptions: Jochen Lempert (1988) spent six consecutive months studying dragonflies in TRF in Liberia at sites occupied by 98 species of Anisoptera and 58 species of Zygoptera, and Steffen Oppel (2005) spent eight consecutive months studying an odonate fauna comprising 16 species of Anisoptera and 45 species of Zygoptera in lower montane rainforest in Papua New Guinea. The species diversity of TRF far exceeds that encountered in temperate latitudes: one of the most species-rich sites so far investigated is the Tambopata-Candamo Reserved Zone in southeastern Peru, studied by Paulson (1985) and Butt (1995). This remarkable site, on the eastern side of the Andean foothills, comprises large and small oxbow lakes and forest swamps and streams. Although only about 5.3 km2 in area, this forest supports at least 151 known species of Odonata, a total that with closer study may reach 200 (Paulson 1985). In the accounts that follow, all sites mentioned are within 10o of latitude of the equator unless otherwise specified.
CyanMagentaYellowBlack Odonata page 23
Forests as habitats for dragonflies (Odonata)
23
TRF clearly functions as a refuge for Odonata, both during the daytime and during the dry season. At both times the forest offers a microclimate that is benign compared with that outside. Thus, as mentioned above, inside forest during the daytime, light intensity, ambient temperature and saturation deficit are significantly lower than outside (Corbet 1963, 1964), the contrast becoming much reduced at sunset when many insects, dragonflies included, leave the forest at its margins – at ground level and above the canopy – to engage (in the case of dragonflies) in a frantic episode of foraging while the fading light still permits. Well known examples are species of Gynacanthini, adults of which can be encountered inside forest during the heat of the day, roosting or patrolling along shaded tracks. I retain a vivid memory of meeting a mature male Gynacantha bullata face-to-face as it patrolled along a forest track in Uganda, its green eyes appearing luminous against the deep shade of the surrounding forest. It was Robert Gambles (1960) who revealed that certain species (e.g. Lestes virgatus and Acanthagyna vesiculata) in Nigeria use the rainforest as a microclimatic refuge during the dry season and as a means of maintaining a univoltine life cycle. With the onset of the seasonal rains, adults, by then reproductively mature, leave the forest and oviposit in the newly formed pools outside in which their larvae develop very rapidly. In this regard they form the tropical counterparts of members of the Lestes-Sympetrum-Aeshna mixta complex of mid-temperate latitudes. This kind of life cycle, featuring siccatation by a long-lived adult (sometimes required to survive for as long as nine months) is probably commonplace in the Tropics. Other known examples include Bradinopyga geminata (Kumar 1973) Megaloprepus coerulatus and Mecistogaster spp. (Fincke 1992), Erythrodiplax funerea (Dunkle 1976; Morton 1977) and Uracis imbuta (Campanella 1975). This type of life cycle (A.2.1.2 of Corbet 1999: 220) may be prevalent among dragonflies of TRF and it provides convincing support for the inference that such Odonata are using the forest as a shelter from the harsh conditions of the dry season. We may assume further that they are also gaining access to the prey that would likewise be seeking refuge in forest and also that, by analogy with Lestes temporalis, they are probably enhancing daily survival of the stage that has to bridge the long, dry season. Species we have been considering so far use TRF as a dry-season refuge but reproduce in open situations outside forest. Many other species, however, remain within rainforest during the whole of their life cycle. In doing so, such species are exploiting the fact that rainforest is rich, not only in α (species) diversity but also in β (habitat) diversity (see Magurran 1988). Both Lempert (1988) in Liberia and Orr (2003) in Borneo have described the diversity of habitats within TRF and some of the ways in which different species of Odonata use them. A habitat-species cluster analysis performed in Papua New Guinea by Oppel (2005) likewise revealed a high α diversity associated with a species-rich odonate community.
CyanMagentaYellowBlack Odonata page 24
24 Philip S. Corbet
Legrand & Couturier (1985) proposed an ecological zonation for running water Odonata in forests of the Ivory Coast. Peter Miller (1993) classified the Odonata of the Budongo Forest, Western Uganda, identifying four types of Odonata with respect to habitat and reproductive behaviour (Table 1). Such segregation is probably typical of Odonata in TRF and beautifully illustrates both the diversity of habitat opportunities within that ecosystem, and the ways in which dragonflies are exploiting them. One may suppose that such a wealth of opportunities must have been a potent force for speciation among tropical Odonata. Reflecting on Table 1, on the heavily shaded environment at ground level inside forest, and on the need for Odonata to thermoregulate, we may marvel at the size of Miller’s categories 1 and 4. Inside tropical forest there are usually sun-flecks at ground level caused by the high elevation of the sun and gaps in the canopy, and there are many examples of sylvan Odonata following these around during the day (e.g. Shelly 1982) and, indeed, sometimes using them as a sexual rendezvous, as in Argia vivida in Alberta (Conrad & PritTable 1. Habitat choice and reproductive behaviour of Odonata in the Budongo Forest, Uganda. (Source: Miller 1993). Category Species 1
2
3
4
Key:
Aeshna scotias xx Macromia aureozona x M. funicularia xx Micromacromia camerunica xx Notogomphus butoloensis xx Onychogomphus styx xx Pseudagrion melanicterum xx P. spernatum spernatum Trithemis sp. (near congolica) x Chlorocypha straeleni xx Pseudagrion hageni xx P. kersteni Umma saphirina x Ceriagrion glabrum x Orthetrum julia julia xx Palpolpleura lucia x Trithemis nuptialis T. stictica Aeshna ellioti xx Hadrothemis coacta x Orthetrum julia julia Micromacromia camerunica
Behaviour Active in densely shaded parts of streams, usually perching or flying in the shade.
Active in shady parts of streams but usually choosing to perch in sun-flecks.
Active only in parts of streams or pools well exposed to sunlight.
Active at or near shaded forest pools
x = reproductive behaviour witnessed. Xx = evidence of breeding, i.e. emergence, exuviae or tenerals observed.
CyanMagentaYellowBlack Odonata page 25
Forests as habitats for dragonflies (Odonata)
25
chard 1988) and Libellago hyalina in Borneo (Orr 2004) but a species like Aeshna scotias, which is placed by Miller in category 1, is reported by Miller as always perching or flying in shade to the extent that a male patrolling in dense shade, on reaching a sunny patch, would turn back and re-enter shade or fly rapidly upwards. Such behaviour resembles that reported by Kotarac (1993) for Somatochlora meridionalis in Eastern Slovenia (at ca 46oN), males of which, while patrolling in shade, would try to avoid sunny spots. Perhaps a flier, like A. scotias, can compensate for low ambient temperature by the endothermic warming effected by persistent flight. The same might be true for the two species of Macromia in category 1. However the other species in Miller’s category 1 are all perchers (sensu Corbet 1962: 126), and this raises the question of how they maintain flight readiness in the shaded environment of the forest floor. A taxon that might be expected to throw light on this apparent paradox is the subfamily Tetrathemistinae, most species of which are dedicated forest dwellers and ombrophiles. A quantitative study of Notiothemis robertsi in the Kakamega Forest, Western Kenya by Viola Clausnitzer (1998) helps to resolve the paradox. N. robertsi, a typical percher, spends only 5% of its time in flight during the daytime. Analysis of 2754 flights by males near the forest floor (Fig. 6) showed that by far the most frequent flights were what Clausnitzer termed ‘sun-flights’ into the forest canopy, presumably for
Fig. 6. The habitat of adult Notiothemis robertsi, males of which perch at ground level in rainforest. Pictured is the Kakamega Forest, western Kenya, at about 00o16’ N., the study site used by Clausnitzer (1998). (Photograph by V. Clausnitzer.)
CyanMagentaYellowBlack Odonata page 26
26 Philip S. Corbet
thermoregulatory purposes. By interspersing frequent, intermittent sun-flights among reproductive and foraging flights, males were apparently able to maintain territorial perches by shaded and ephemeral rainforest pools, where (one supposes) interspecific competition would be far less than in insolated pools in forest clearings and outside forest. In Barro Colorado Island, Panama, the coenagrionid Heteragrion erythrogaster, a percher, frequents deep shade where its thoracic temperature can be within 1oC of ambient. Shelly (1982) compared H. erythrogaster with the sympatric Argia difficilis that perched in the most brightly illuminated areas. During sunny weather A. difficilis foraged five times more often than H. erythrogaster, made longer flights and maintained a thoracic temperature 4–8oC above ambient; so it needed more food. On overcast days the two species resembled each other closely in their thoracic temperature and level of activity. Shelly does not record whether H. erythrogaster made compensatory sun-flights in the manner of N. robertsi. If it did not, we encounter a contradiction when faced with these two apparently conflicting strategies employed by perchers. If the cost of remaining cool and foraging less is relatively insignificant, why, one may ask, does A. difficilis invite a greater predation risk by maintaining a higher level of activity? This is just one of the many evolutionary questions raised by our very superficial and fragmentary knowledge of the way that Odonata use the resources of TRF. TRF, with its equable conditions and its profusion of species, offers an unrivalled natural laboratory for studying the behaviour and ecology of dragonflies. In pristine, closed rainforest conditions may favour Zygoptera over Anisoptera. In such an ecosystem in Papua New Guinea Oppel (2005) hypothesized that the preponderance of Zygoptera he encountered (>70%) reflected a habitat that failed to provide for the elevated thermal needs of Anisoptera. The ways in which Odonata have become adapted to ecological niches offered by TRF are varied and wonderful to behold. Many species, mainly Zygoptera, use phytotelmata as larval habitats (Corbet 1983, 1999: 144). Phytotelmata (Fig. 7) provide normal, occasional or sole habitats for larvae of at least 24 genera and 47 species of Odonata, virtually all of which are confined to the Tropics. Most are Zygoptera, conspicuous among which are species of Pseudostigmatidae, such as Coryphagrion grandis (Clausnitzer & Lindeboom 2002), Mecistogaster species and Megaloprepus coerulatus (Fincke 1992) and Leptagrion species (Coenagrionidae) (Santos 1966). Among Anisoptera, for example species of Lyriothemis and Indaeschna in Borneo, the use of phytotelmata is facultative, tree holes being favoured when there is a paucity of standing water, e.g. on sloping ground (Orr 1994). Phytotelmatous Anisoptera are, unsurprisingly, confined to larger tree holes (Orr 1994; Copeland et al. 1996), probably because females ovipositing exophytically have difficulty escaping from a confined space (e. g. Hadrothemis camarensis, Corbet 1961).
CyanMagentaYellowBlack Odonata page 27
Forests as habitats for dragonflies (Odonata)
27
The occupancy of phytotelmata as larval habitats by sylvan Odonata, partly it appears in response to a dearth of ground pools, has its logical extension in the evolution of terrestrialism in a few species of sylvan Odonata. Terrestrialism was first detected in Hawaiian species of Megalagrion (Coenagrionidae), some of which are completely aquatic whereas others (e.g. M. amaurodytum and M. oahense) (Fig. 8) are virtually terrestrial, occupying
Fig. 7. Epiphytic bromeliads growing on an Erythrina branch in Costa Rican rainforest. Larvae of species of Leptagrion and Mecistogaster typically occupy such phytotelmata. (From Corbet (1962) after Calvert (1911).)
Fig. 8. The terrestrial larva of Megalagrion oahuense, probably in the penultimate stadium. It occupies damp leaf litter beneath stands of Gleichenia fern in upland forests of Hawaii, and is about 13 mm long when fully grown. (From Corbet (1962) after Williams (1936).)
CyanMagentaYellowBlack Odonata page 28
28 Philip S. Corbet
moist leaf litter on the forest floor and becoming restless if placed in free water (Williams 1936). The caudal lamellae of these two species are saccoid and triquetral and (unusually for Zygoptera) the body is covered with dense setae (Fig. 8). The few other species that have terrestrial larvae all occupy the same kind of habitat, though not all are on islands: Calydopteryx uniseries in New Caledonia (Lieftinck 1976; Winstanley 1983), Psudocordulia species in high-altitude mist forest on the Atherton Tablelands, northern Queensland (Watson 1982) (Fig. 9) and Idomacromia proavita in the Ivory Coast (Legrand 1983). I have already mentioned the Tetrathemistinae as a taxon most of whose members are intimately associated with TRF. A further example will serve to emphasise this. In Gabon, West Africa two species of Malgassophlebia develop in small, shaded forest streams. M. aequatoris reproduces briefly at the onset of each of the two annual rainy seasons. It oviposits epiphytically, the female hovering close to a leaf overhanging a stream and placing its eggs within a coating of jelly on the underside of the tip of the leaf whence, in due course, the prolarvae will drop into the stream. Its larvae develop rapidly in the stream, having a life cycle that appears to be bivoltine (Legrand 1979).
Fig. 9. The terrestrial larva, probably in the penultimate stadium, of a gomphomacromiine corduliid (probably a species of Pseudocordulia) that inhabits ground litter in upland rainforest in Queensland, Australia The larva shown is about 17 mm long. (From Corbet (1999); photograph by the Commonwealth Scientific and Industrial Research Organization, Australia.)
CyanMagentaYellowBlack Odonata page 29
Forests as habitats for dragonflies (Odonata)
29
M. bispina likewise hovers to oviposit on the undersides of leaves overhanging streams in high forest in Liberia (Lempert 1988). Lempert suggests that this habit may protect the eggs against predation by ants although, in the case of M. aequatoris, it seems not to protect them against predation by drosophilid flies (Legrand 1979). TRF also features glades and clearings, often originating from tree-fall locations. The absence of a canopy makes such places benign hotspots offering advantages over the savannah outside forest on account of the shelter they provide from wind. Such clearings are often frequented by savannah species that reproduce there if aquatic habitats are available (Lempert 1988). Sometimes forest species may use such clearings as sources of warmth and food of which the provision may exceed that available within the closed forest. An example is Megaloprepus coerulatus, a tree-hole breeder, adults of which are unique among Odonata in being specialized (as distinct from generalized) predators. They prey exclusively on spiders which they pluck from a web. By preying on web-building spiders in sunlit clearings in forest, these pseudostigmatids can tap a source of food energy that has already been harvested by the spider and sequestered within its body, giving the predator a large reward per foraging episode. Indeed, the pseudostigmatid may even be ‘farming’ its prey, because spiders tend promptly to occupy vacant webs (Wilson 1992; Rowe 1993). This account of ways that Odonata use TRF is necessarily brief, not least because we still know little about the behaviour and ecology of species in this environment. Especially needed now are longitudinal studies, spanning at least a year, underpinned by authoritative taxonomy. Also, there is a pressing need for comprehensive inventories of the fauna, which certainly contains many undescribed taxa. The science of odonatology as a whole, and especially the fields of phylogeny and systematics, stand to benefit immeasurably if comprehensive fauna lists for the planet’s TRF can be made available soon. Indeed, these branches of science will remain woefully incomplete and provisional until this can be accomplished. Carl Cook (1992), in a stirring review, has reminded us that TRF provides the sole habitat for more than half the species of Metazoa on the planet. If, as is likely, this statistic applies to Odonata, then hundreds of dragonfly species await description; yet, as a result of accelerating anthropological impact, only 6% or less of the original acreage of the world’s TRF remains, and even this is disappearing at an alarming rate, powered by human greed, demand for short-term profit and indifference by so-called legislators. It is frustrating beyond description to stand as impotent witnesses to this comprehensive destruction and to know that an irreplaceable storehouse of biological information is being eliminated before our eyes. A litany we encounter repeatedly as a taxonomist describes another new species from TRF is that he or
CyanMagentaYellowBlack Odonata page 30
30 Philip S. Corbet
she hopes that the odonate fauna of the habitat can be adequately described before it disappears for ever. Any species occupying forest subject to commercial logging faces such a threat, and in some cases this has an immediacy that is impossible to ignore. We may consider the example of Risiocnemis seidenschwarzi, a beautiful stream-dwelling platycnemidid, apparently endemic to Cebu in the Philippines. Cebu is already the most denuded of the primary forests in the Philippines: only a few patches (totalling about 45 ha) remain. R. seidenschwartzi, which is apparently confined to a stretch of about 30 m in a creek 2.5 – 3.5 m wide, may be one of the most vulnerable species of dragonfly in existence (Hämäläinen 2000). From his study in Papua New Guinea Oppel (2005) concluded that, because the habitat requirements of most closed-forest Zygoptera are unlikely to be met in degenerated forest or in a largely deforested landscape, species of this sub-order could provide useful indicators of undisturbed rainforest. Consequently most such species can probably be regarded as prone to local extinction following habitat modification and it can be assumed that deforestation will severely endanger their survival (Clausnitzer 2003). The recent publication documenting the conservation status of Odonata across the world (Clausnitzer & Jödicke 2004) has performed a great service by drawing attention to the scale and urgency of this problem. The inescapable conclusion to be drawn from this review is that forests (especially TRF) provide an essential habitat for the world’s Odonata. In particular the planet’s TRF forms an irreplaceable storehouse of the Odonata that have survived to the present day. Unless the destruction of TRF is halted very soon we face the certain prospect of seeing a great part of the planet’s endowment of Odonata disappearing without trace.
CONCLUSIONS Dragonflies seek habitats that can: permit effective thermoregulation by adults; provide a milieu for foraging; provide sites for nocturnal roosting or daytime shelter for adults; and provide, or offer access to, a body of water suitable as a reproductive site and locus for larval development and survival. The journey we have taken along the latitude spectrum, from the Arctic Circle to the equator, has followed a gradient of increasing ambient temperature, decreasing saturation deficit (inside forest) and increasing biodiversity. At the highest latitudes we suppose that dragonflies use forests only as a shelter, as distinct from a source of habitat opportunities for different activities and stages of development. At the Arctic Circle, taiga presumably offers essential protection from the unremitting harshness of the adjacent tundra. So one might say that dragonflies use the boreal forest, or taiga, not for its habitat richness but as a ‘port in a storm’.
CyanMagentaYellowBlack Odonata page 31
Forests as habitats for dragonflies (Odonata)
31
As we proceed towards lower latitudes we find that dragonfly species diversity is progressively correlated with the presence of forests, although we are ignorant of the causal links that bring this about. At mid-temperate latitudes, where summers can be long, hot and dry, the use of forests by dragonflies becomes more structured, as prereproductive adults routinely aestivate there during the dry season, thus enhancing their survival and postponing reproduction until late summer, thereby maintaining a univoltine life cycle. This protocol finds its most conspicuous manifestation in the Tropics, especially where aquatic insects are subject to the tyranny of a monsoon climate. There several, perhaps many, species repair to forest after completing larval development in ephemeral pools in adjacent savannah, siccatating inside forest until advent of the next rains triggers their reproduction. Lestids and aeshnids feature prominently among species that adopt this life cycle, in both tropical and temperate latitudes. Our survey ends in equatorial latitudes, in the unrivalled richness of the tropical rainforest. This biological treasure-house, on account of its uniquely high α and β diversity, offers incomparably rich habitat opportunities for all stages of the dragonfly life history. It contains a high proportion of the planet’s dragonfly fauna, many species of which still remain to be described, and most of which pursue a way of life almost unknown to us. In contrast to the taiga, where we can only speculate as to the use of forest made by dragonflies, in the tropical rain forest we can see clearly the many ways in which dragonflies use the resources on offer. Indeed, we may suppose that the α diversity of the tropical rain forest has been, and continues to be, a potent force for dragonfly speciation.
CONSERVATION IMPLICATIONS Arrival at the terminus of our journey is attended by a bitter irony. The forests on which dragonflies depend most for habitats and therefore survival are in imminent danger of terminal destruction. If present trends of deforestation continue, then within a human generation most of the planet’s remaining tropical rain forests will have disappeared and with them countless organisms, including dragonflies, of great antiquity and biological interest. The causes for this destruction lie in the pressure exerted by human numbers and human greed, forces which governments seem unwilling to acknowledge or check. To summarise: dragonflies use forests as habitats wherever forests exist. Indeed dragonflies depend on forests for survival. We are witnessing an inexorable process whereby dragonflies are being rapidly deprived of that prerequisite for their survival. Anything that odonatologists can do to check this process or mitigate its effects will constitute a major contribution to biological science.
CyanMagentaYellowBlack Odonata page 32
32 Philip S. Corbet
ACKNOWLEDGEMENTS It is a pleasure to thank Sally Corbet and Sarah Jewell for helpful comments on the text, and Hansruedi Wildermuth for generously sharing with me his extensive knoweldge of odonate biology. I thank the originators of the photographs (acknowledged in the text) and the first publishers of four figures for permission to reproduce them, namely: Blackwells Publishing, Oxford (Fig. 1); Manney Publishing, London (Figs. 2, 3); and B. Kiauta, for Societas Internationalis Odonatologica (Fig. 5).
REFERENCES ARNABOLDI, F. 1997. Note sur les odonates de la Forêt de Rambouillet. Martinia 13: 86. BELYSHEV, B.F. 1973. The dragonflies of Siberia. 2 volumes. Nauka, Siberian Branch, Novosibirsk. [In Russian.] BROOKS, S.J., A. HINE, S.A. CHAM & A. MCGEENEY. 1995. A study of the ecology of the Downy Emerald Dragonfly (Cordulia aenea (L.)) (Odonata: Corduliidae) in southeast England. Unpublished MS. BUTT, M. 1995. Odonata collected from the Tambopata-Candamo Reserved Zone, southeastern Perú, August 1992-January 1993. Notulae Odonatologicae 4: 93-97. CALVERT, P.P. 1911. Studies on Costa Rican Odonata. II. The habits of the plantdwelling larva of Mecistogaster modestus. Entomological News 22: 402-411. CAMPANELLA, P.J. 1975. Letter, 18 December. CANNINGS, S.G. & R.A. CANNINGS. 1985. The larva of Somatochlora sahlbergi Trybom, with notes on the species in the Yukon Territory, Canada (Anisoptera: Corduliidae). Odonatologica 14: 319-330. CANNINGS, S.G., R.A. CANNINGS & R.J. CANNINGS. 1991. Distribution of dragonflies (Insects: Odonata) of the Yukon Territory, Canada with notes on ecology and behaviour. Contributions to Natural Science, Royal British Columbia Museum 13: 1-27. CLAUSNITZER, V. 1998. Territorial behaviour of a rainforest dragonfly Notiothemis robertsi (Odonata: Libellulidae): proposed functions of specific behavioural patterns. Journal of the Zoological Society of London 245: 121-127. CLAUSNITZER, V. 2003. Dragonfly communities in coastal habitats of Kenya: indication of biotope quality and the need of conservation measures. Biodiversity and Conservation 12: 333-356. CLAUSNITZER, V. & R. Jödicke (eds). 2004. Guardians of the watershed. Global status of dragonflies: critical species, threat and conservation. International Journal of Odonatology 7: 111-430.
CyanMagentaYellowBlack Odonata page 33
Forests as habitats for dragonflies (Odonata)
33
CLAUSNITZER, V. & M. LINDEBOOM. 2002. Natural history and description of the dendrolimnetic larva of Coryphagrion grandis (Odonata). International Journal of Odonatology 5: 29-44. CONRAD, K.F. & G. PRITCHARD. 1988. The reproductive behaviour of Argia vivida Hagen (Odonata: Coenagrionidae). Canadian Journal of Zoology 67: 298-304. COOK, C. 1992. Questions and answers about tropical rainforests. Argia 4: 2-7. COPELAND, R.S., W. OKEKA & P.S.CORBET. 1996. Treeholes as larval habitat of the dragonfly Hadrothemis camarensis (Odonata: Libellulidae) in Kakamega Forest, Kenya. Aquatic Insects 18: 129-147. CORBET, P.S. 1961. Entomological studies from a high tower in Mpanga Forest, Uganda. XII. Observations on Ephemeroptera, Odonata and some other orders. Transactions of the Royal Entomological Society of London 113: 356-361. CORBET, P.S. 1962. A biology of dragonflies. Witherby, London. CORBET, P.S. 1963. The oviposition-cycles of certain sylvan culicine mosquitoes (Diptera, Culicidae) in Uganda. Annals of Tropical Medicine and Parasitology 57: 371-381. CORBET, P.S. 1964. Observations on mosquitoes ovipositing in small containers in Zika Forest, Uganda. Journal of Animal Ecology 33: 141-164. CORBET, P.S. 1969. Terrestrial microclimate: amelioration at high latitudes. Science 166: 865-866. CORBET, P.S. 1983. Odonata in phytotelmata. In: Frank, J.H. & L.P. Lounibos (eds), Phytotelmata: terrestrial plants as hosts of aquatic insect communities, pp. 29-54, Plexus, Marlton, New Jersey. CORBET, P.S. 1999. Dragonflies. Behaviour and ecology of Odonata. Harley Books, Colchester, U.K. CÓRDOBA-AGUILAR, A. 1994. Adult survival and movement in males of the damselfly Hetaerina cruentata (Odonata: Calopterygidae). Florida Entomologist 77: 256-264. DUNKLE, S.W. 1976. Notes on the Anisoptera fauna near Mazatlán, Mexico, including dry to wet season changes. Odonatologica 5: 207-212. FINCKE, O.M. 1992. Interspecific competition for tree holes: consequences of mating systems and coexistence in Neotropical damselflies. American Naturalist 139: 80-101. GAMBLES, R.M. 1960. Seasonal distribution and longevity in Nigerian dragonflies. Journal of the West African Science Association 6: 18-26. HA, L.Y., H. WILDERMUTH & S. DOM. 2002. Emergenz von Cordulia aenea (Odonata: Corduliidae). Libellula 21: 1-14. HÄMÄLÄINEN, M. 2000. Risiocnemis seidenschwarzi spec. nov., an endangered damselfly from Tabunan forest in Cebu, the Philippines (Odonata: Platycnemididae). Entomologische Berichten, Amsterdam 60: 46-49. KOSTERIN, O. 1992. New distribution records of Somatochlora sahlbergi Trybom (Odonata, Corduliidae). Acta Hydroentomologica Latvia 2: 22-26. KOTARAC, M. 1993. Dragonfly observations in the Raka area, Lower Carniola, Eastern Slovenia, with a note on the behaviour of Somatochlora meridionalis Nielsen (Anisoptera: Corduliidae). Notulae Odonatologicae 4: 1-4.
CyanMagentaYellowBlack Odonata page 34
34 Philip S. Corbet
KUMAR, A. 1973. The life history of Bradinopyga geminata (Rambur) (Odonata: Libellulidae). Gurukul Kangri Vishwa Vidayalya. Journal of Scientific Research 5: 50-57. LANDMANN, A. 1985. Stukturierung, Ökologie und saisonale Dynamik der Libellenfauna eines temporären Gewässers. Libellula 4: 49-80. LAWTON, J.H., D.E. BIGNELL, B. BOLTON, G.F. BLOEMERS, P. EGGLETON, P.M. HAMMOND, M. HODDA, R.D. HOLT, T.B. LARSEN, N.A. MAWDSLEY, N.E. STORK, D.S. SRIVASTAVA & A.D. WATT. 1998. Biodiversity inventories, indicator taxa and effects of habitat modification in tropical forest. Nature, London 391: 72-76. LE CALVEZ, V. 1998. Les odonates de la forêt domaniale de Notre Dame (Départements du Val-de-Marne et de Seine-et-Marne). Martinia 14: 137-145. LEGRAND, J. 1979. Morphologie, biologie et écologie de Malgassophlebia aequatoris, n. sp., nouveau Tetratheminae du Gabon (Odonata: Libellulidae). Revue Française d’Entomologie (N.S.) 1: 3-12. LEGRAND, J. 1983. Personal communication, 18 August. LEGRAND, J. & G. COUTURIER. 1985. Les odonates de la forêt de Tai (Côted’Ivoire). Premières approaches faunistique, répartition écologiques et association d’espèces. Revue d’Hydrobiologie Tropicale 18: 133-158. LEMPERT, J. 1988. Untersuchungen zur Fauna, Ökologie und zum Fortplanzensverhalten von Libellen (Odonata) an Gewässern des tropischen Regenwaldes in Liberia, Westafrika. Diplomarbeit, Rheinischen FriedrichWilhelms-Univ., Bonn. LIEFTINCK, M.A. 1976. The dragonflies (Odonata) of New Caledonia and the Loyalty Islands. Part 2. Immature stages. Cahiers Office de la Recherche Scientifique et Technique Outre-Mer 10: 165-200. MAGURRAN, A.E. 1988. Ecological diversity and its measurement. Croome Helm, London. MILLER, P.L. 1993. Some dragonflies of the Budongo forest, western Uganda (Odonata). Opuscula Zoologica Fluminensia 102: 1-12. MORTON, E.S. 1977. Ecology and behavior of some Panamanian Odonata. Proceedings of the Entomological Society of Washington 79: 273. OPPEL, S. 2005. Habitat associations of an Odonata community in a lower montane rainforest in Papua New Guinea. International Journal of Odonatology 8: 243-257. ORR, A.G. 1994. Life histories and ecology of Odonata breeding in phytotelmata in Bornean rainforest. Odonatologica 23: 365-377. ORR, A.G. 2003. A guide to the dragonflies of Borneo. Their identification and biology. Natural History Publications (Borneo), Kota Kinabalu, Sabah. ORR, A.G. 2004. Territorial behaviour associated with feeding in both sexes of the tropical zygopteran, Libellago hyalina (Odonata: Chlorocyphidae). International Journal of Odonatology 7: 493-504. PAULSON, D.R. 1985. Odonata of the Tambopata Reserved Zone, Madre de Dios, Perú. Revista Peruana de Entomologia 27: 9-14. RITH-NAJARIAN, J.C. 1998. The influence of forest vegetation variables on the distribution and diversity of dragonflies in a northern Minnesota forest landscape: a preliminary study (Anisoptera). Odonatologica 217: 335-351.
CyanMagentaYellowBlack Odonata page 35
Forests as habitats for dragonflies (Odonata)
35
ROWE, R.J. 1993. Letter, 21 June. SAHLÉN, G. 1994. Tundratrollslandan Somatochlora sahlbergi funnen I nordligaste Sverige. Entomologiske Tijdskrifte 115: 137-142. SAHLÉN, G. 1999. The impact of forestry on dragonfly diversity in Central Sweden. International Journal of Odonatology 2: 177-186. SAHLÉN, G. & K. EKESTUBBE. 2001. Identification of dragonflies (Odonata) as indicators of general species richness in boreal forest lakes. Biodiversity and Conservation 10: 673-690. SAMRAOUI, B., S. BOUZID, R. BOULAHBAL & P.S. CORBET. 1998. Postponed reproductive maturation in upland refuges maintains life-cycle continuity during the hot dry season (Anisoptera). International Journal of Odonatology 1: 119-135. SANTOS, N.D. 1966. Contribuiçao ao conhecimento da fauna do Estada da Guanabara. 56. Notas sobre Coenagriidae (Odonata) que se criam em bromélias. Atas da Sociedade de Biologia do Rio de Janeiro, Nova.Serie, Zoology 256: 1-5. SCHORR, M. 1990. Grundlagen zu einem Artenhilfsprogramm Libellen der Bundesrepublik Deutschland. Ursus Scientific Publishers, Bilthoven. SHELLY, T.E. 1982. Comparative foraging behavior of light- versus shade-seeking adult damselflies in a lowland Neotropical forest (Odonata: Zygoptera). Physiological Zoology 55: 335-343. SMITH, R.W.J. 1995. Personal communication. STERNBERG, K., & R. BUCHWALD (eds). 2000. Die Libellen Baden-Württembergs. Volume 2. Ulmer, Stuttgart. STERNBERG, K. & B. SCHMIDT. 2000. Cordulia aenea. In: Sternberg, K., & R. Buchwald (2000), pp. 209-218. STERNBERG, K. & S. STERNBERG. 2000. Aeshna caerulea. In: Sternberg, K., & R. Buchwald (2000), pp. 23-38. UBUKATA, H. 1984. Oviposition site selection and avoidance of additional mating by females of the dragonfly, Cordulia aenea amurensis Selys (Corduliidae). Research on Population Ecology 26: 285-301. UÉDA, T. & M. IWASAKI. 1982. Changes in the surivorship, distribution and movement pattern during the adult life of a damselfly, Lestes temporalis (Zygoptera: Odonata). Advances in Odonatology 1: 281-291. VALLE, K.J. 1931. Materialen zur Odonatenfauna Finlands. II. Somatochlora sahlbergi Trybom. Notulae Entomologicae 11: 41-51. WATSON, J.A.L. 1982. A truly terrestrial dragonfly from Australia (Odonata: Corduliidae). Journal of the Australian Entomological Society 21: 309-311. WILDERMUTH, H. 2003. Fortpflanzenverhalten von Somatochlora arctica (Zetterstedt) (Anisoptera: Corduliidae). Odonatologica 32: 61-77. WILDERMUTH, H. 2005a. Personal communication to P.S. Corbet, 23 May. WILDERMUTH, H. 2005b. Beitrag zur Larvalbiologie von Boyeria irene (Odonata: Aeshnidae). Libellula 24: 1-30. WILLIAMS, F.X. 1936. Biological studies in Hawaiian water-loving insects. Part II. Odonata or dragonflies. Proceedings of the Hawaiian Entomological Society 9: 273-349.
CyanMagentaYellowBlack Odonata page 36
36 Philip S. Corbet
WILSON, E.O. 1992. The diversity of life. Harvard University Press, Cambridge, Massachusetts. WINSTANLEY, W.J. 1983. Terrestrial larvae of Odonata from New Caledonia (Zygoptera: Megapodagrionidae; Anisoptera: Synthemistidae). Odonatologica 12: 389-395.
CyanMagentaYellowBlack Odonata page 37
Allochthonous organic matter Adolfo as a food resource for aquatic invertebrates in forested streams Cordero Rivera (ed) 2006 Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 37-47. © Pensoft Publishers
Sofia–Moscow
Allochthonous organic matter as a food resource for aquatic invertebrates in forested streams Manuel A. S. Graça IMAR – Departamento de Zoologia, Universidade de Coimbra, 3004-517 Coimbra, Portugal.
[email protected]
ABSTRACT This paper summarises the role of organic matter in the ecology of forested low order streams. Forests are among the most productive systems on Earth. More than 90% of forest primary production will end in detrital pathways, in soil and water. The amount of energy in the form of plant litter entering forested low order streams is several times higher than the energy synthesized by aquatic producers; therefore leaves produced in the riparian zones are a main energy source and decomposition is an important ecological process in those systems. Decomposition is mainly a biological process initiated by aquatic fungi and shredding invertebrates. Those organisms promote the transformation of leaves into fine particles used by bacteria, collectors and filter-feeding invertebrates. Therefore, much of the energy allocated into secondary production in streams has an allochthonous origin. Nutrients liberated as a result of decomposition are used further downstream, in lakes or estuaries by primary producers. The rate at which leaf litter is decomposed is controlled by intrinsic leaf properties (nutrient content, plant chemical and physical defences) as well as environmental factors (e.g. nutrients in water). Disturbances of riparian zones and eutrophication can affect decomposition and, for this reason, changes in decomposition rates could be used as a functional parameter to assess stream health. Given that the standing stock of leaf litter has a positive effect on leaf consumers, allowing high biomass and diversity, it is likely to also affect top invertebrate predators including odonates; however, the literature on this subject is still scarce.
37
CyanMagentaYellowBlack Odonata page 38
38 Manuel A. S. Graça
LEAVES ARE AN IMPORTANT SOURCE OF ENERGY IN STREAMS Forests are among the most productive terrestrial systems, with net primary production ranging from 1800 g dry mass m-2 year-1 in tropical rain forest to 850 g dry mass m-2 year in boreal forests. Those values are above the well fertilized cultivated land, with primary production around 650 g dry mass m-2 year-1 (Ricklefs 2000). Data from several ecosystem studies have shown that only approximately 10% of the energy produced at a given trophic level will be incorporated into secondary production in the next one (range from 0.5 to 20% in invertebrates feeding on trees; Townsend et al. 2000). The remaining 90% will enter the detritus food web (up to 95% in forests). The material entering the detritus food web in forests is made of bark, flowers, seeds, fruits, twigs and the most important in terms of mass: leaves (Bray & Gorham, 1964; Abelho, 2001). Forests produce large amounts of leaves. That is particularly evident in deciduous forests (300 to 800 g dry mass m-2 year-1), but the highest values have been measured in tropical forests (> 1000 g dry mass m-2 year-1; see Abelho 2001 for a review). With such an amount of leaves, it is virtually impossible that some of this material does not reach streams. The amount of organic material reaching streams is variable, depending on the type of trees in the riparian zone, canopy cover, slope of margins, format of the valley and winds. Values up to 2800 g dry mass m-2 year-1 of litter input have been measured (reviewed by Abelho 2001). Although rivers have a unidirectional flow, it is wrong to assume that leaves and other plant detritus entering streams would be washed away. In fact, small streams (1st to 4th order) are very retentive, with typically 90% of leaves entering streams being retained in the first 60 meters (reviewed by Abelho, 2001). However this value is strongly dependent on depth, current, substrate, hydrological regime, etc. The key aspect is that leaves tend not to travel long distances after entering streams and accumulate in riffles, margins and among substrates projected above the water level (e.g. twigs, wood) where it decomposes. Documented values for standing stock of benthic organic matter range from 12 to 3000 g AFDM (ash free dry mass) m-2 in 1st - 3rd order streams (Abelho 2001). Those values are several times higher than the algal biomass covering stream substrates. Indeed, some authors calculated that in low order streams running through forests, up to 99% of the respired energy entered the rivers as leaves and other plant remains (Abelho 2001). In streams, the energy respired by decomposers and detritivores is approximately 150% of the primary production. This value in streams is only possible with subsidizing energy in the form of organic matter from riparian zones (Abelho 2001).
CyanMagentaYellowBlack Odonata page 39
Allochthonous organic matter as a food resource for aquatic invertebrates in forested streams
39
Riparian vegetation is therefore very important for the ecology of small streams. It provides organic matter and limits primary production by reducing the amount of light reaching the stream bed. This is one of the core ideas of the River Continuum Concept (Vannote et al. 1980): headwaters are heterotrophic systems and therefore, invertebrates feeding on coarse particulate organic matter are an important functional feeding group in those areas. As we move downstream, the distance between margins increases and even if the riparian vegetation is dense, more light enters the river and primary production can replace organic matter as the main energy source. In headwaters wood and twigs can also enter streams and be decomposed in situ. However, because they are difficult to use as food they are more important as trappers of leaves and sediments thus creating heterogeneity in the stream bed. This heterogeneity allows more microhabitats favourable for fish and other invertebrates.
DECOMPOSITION OF ORGANIC MATTER IS AN IMPORTANT BIOLOGICAL PROCESS IN THE ECOSYSTEM Leaves are highly energetic. However, this energy is not easily accessible to consumers because leaves are constituted mainly of cellulose and lignin; in general, animals do not have the enzymatic ability to break down these compounds. That is why many large size herbivorous mammals (e.g. cows and horses) have large guts working as fermentation chambers, in which bacteria, yeasts and protozoa initiate the digestion of such compounds. It is generally accepted that herbivores are not limited quantitatively by energy, but by nutrients. Senescent leaves are even worse as a food resource since before senescence, nutrients such as nitrogen are reabsorbed by plants (Slansky & Scriber 1985). Unlike animals, fungi and bacteria are equipped with the necessary enzymes to breakdown plant structural compounds. The decomposition of leaves in streams occurs in four overlapping steps (Gessner et al. 1999). The first step consists in the leaching of soluble compounds such as phenolics, amino acids and simple sugars. Those compounds will be further decomposed by bacteria in the water column. Leaves are then attacked by fungi, followed by invertebrates. As a result, nutrients are incorporated into secondary production. Fungal and invertebrate attack also results in the production of fine particles of organic matter which are transported downstream and used by bacteria and invertebrates (filterers and gatherers). Leaf mass loss often occurs in a negative exponential motion, with rapid initial decomposition (including leaching) and a slower rate in later stages when the more recalcitrant compounds remain. The rate of decomposition can then be expressed as the slope (k) of a regression of mass over time (for details see Boulton & Boon 1991). The decomposition coefficient k has typi-
CyanMagentaYellowBlack Odonata page 40
40 Manuel A. S. Graça
cally values ranging from 0.01 d-1 to 0.001 d-1 (but values above 0.1 d-1 and below 0.0001 d-1 have been reported; Abelho 2001). Factors affecting decomposition rates include nutrient content of senescent leaves, nutrients in the water, leaf hardness, presence of long lived plant chemical defences and temperature. All these factors are important for the activity of microorganisms. Moreover, during decomposition, nitrogen and ATP content of leaves generally increases, which is considered as an indicator of microbial colonization (Cortes et al. 1995, Abelho 2005). Ergosterol content (a component occurring in the cell membrane of some fungi) also increases, as well as respiration in leaves (Gessner 2005, Graça & Abelho 2005).
IN STREAMS LEAVES ARE DECOMPOSED BY FUNGI AND CONSUMED BY INVERTEBRATES Fungi involved in decomposition of allochthonous organic matter in streams are known as “aquatic hyphomycetes” or “ingoldian fungi” (Fig. 1). They are characterized by asexual reproductive structures (conidia) with sigmoid or tetra radiate shapes. Those shapes are ideal for fixing to substrates in flowing waters. Aquatic hyphomycetes generally account for > 90% of the microbial biomass in decomposing leaves and up to 8 – 16% of the leaf mass (Suberkropp 1995, Baldy and Gessner 1997).
Fig. 1. Hyphae of aquatic fungi on a leaf surface and penetrating the leaf through the stomas (Photo by C. Canhoto).
CyanMagentaYellowBlack Odonata page 41
Allochthonous organic matter as a food resource for aquatic invertebrates in forested streams
41
Invertebrates feeding on leaves (shredders) are represented in rivers by several species of trichoptera (Fig. 2), plecoptera, amphipoda, diptera, and other minor groups. Leaves colonized by aquatic hyphomycetes are more palatable to shredders than fresh fallen leaves. This can be a consequence of leaf digestion by fungal enzymes (Graça & Ferreira 1995, Rodrigues & Graça 1996), increase of nitrogen and decrease of leaf toughness. Consumption on
Fig. 2. Cases of shredder trichoptera (a) Sericostomatidae, (b) Lepidostomatidae (Photos by C. Canhoto).
CyanMagentaYellowBlack Odonata page 42
42 Manuel A. S. Graça
microbial colonized leaves results in high growth, survivorship and invertebrate reproductive output (Graça 1993, 2001). In streams, shredders are frequently more abundant in “fast” (high quality) than in “slow” (low quality) decomposing leaves (Malmqvist & Oberle 1995; Basaguren & Pozo 1994), which is an indication that shredder invertebrates actively search for high quality leaf resources. However, shredders exhibit a high feeding plasticity and are able to feed and grow well in other food sources such as algae, macrophytes and FPOM (fine particulate organic matter) (Friberg & Jacobsen 1994, Mihuc & Mihuc 1995). The functional role of shredders can be very important in leaf decomposition, being responsible for leaf mass losses up to 60% (Chergui & Pattee 1991, Hieber & Gessner 2002). Experimental treatments of streams with insecticides resulted in decreased decomposition rates, accumulation of leaves and reduction of fine particulate organic matter (Cuffney et al. 1990). The importance of shredders to decomposition is generally assessed in decomposition experiments using litterbags of fine mesh size (e.g. 0.5 mm; accessible to fungi but not to macroinvertebrates) and large mesh size (e.g. 5 mm; accessible to both invertebrates and fungi). Given the large numbers of shredders in the upper reaches of streams, detritivores can have a primordial role on decomposition there. However, in the lower sections (stream order >3), the number of shredders decreases, the nutrient content in the water and temperature increases and therefore microbial decomposition seems to overcome invertebrate feeding (Graça et al. 2001).
DETRITUS BASED SYSTEMS ARE AFFECTED BY ANTHROPOGENIC ACTIVITIES Given the importance of riparian vegetation to streams, it is plausible that changes in the riparian zone can affect aquatic communities. This has been demonstrated by several studies. Streams running through mature forests were reported to have a larger standing stock of leaves, higher proportion of slow decomposition leaves and lower production of shredders than streams running through managed forests (Stout et al. 1993). In the New Mexico Rocky Mountains, aspen forests give way to pine forests after wildfires. Streams running through pine forests had a larger standing stock of organic matter and a higher number of shredder caddisflies than streams running through aspen forests (Molles 1982). The replacement of a mixed forest with monocultures of trees is likely to affect headwaters. Eucalyptus is one of the trees most extensively planted in the planet. There are nearly 600 species of Eucalyptus native from Australia and New Zealand, but large plantations of this genus can be found in Morocco, South
CyanMagentaYellowBlack Odonata page 43
Allochthonous organic matter as a food resource for aquatic invertebrates in forested streams
43
Africa, Brazil, Angola, India and the Iberian Peninsula. In Central Portugal monocultures of Eucalyptus globulus (Labill.) occupy > 20% of forest area. Plantations with eucalypts change the timing of litter fall in the Iberian Peninsula (reviewed by Graça et al. 2002). Instead of a single and large input of leaves during autumn, in streams running through eucalypt plantations litter input occurs mainly in summer or is distributed along the year (Fig. 3). The standing stock of leaves in streams increase in eucalypt plantations. This may be related to two factors. Firstly, Eucalyptus globulus shed the bark which accumulates in streams, increasing retention. Secondly, litterfall in eucalypt plantations tend to be more intense during summer, which is the dry period. Given the low amount of N and P in eucalypt leaves, nutrient input to streams is reduced in eucalypt streams (e.g. Molinero & Pozo 2004). Several studies in Portugal have shown that the diversity of aquatic hyphomycetes in eucalypt forest streams was lower than in reference deciduous forests. The same was true for the species richness and density of invertebrates (Graça et al. 2002). Finally, if decomposition is an important ecosystem functional parameter, can we use decomposition rates as a tool to assess environmental quality of a river? This was the objective of the RivFunction (http://www.ladybio.upstlse.fr/rivfunction). Preliminary results have shown that nutrient enrichment (nitrogen and phosphorous) resulted in accelerated decomposition of leaves. This subject is still under investigation. Autumn
Summer 70
AFDW (g.m-2)
60 50 40 30 20 10 0 F
M
A
M
J
J
A
S
O
N
D
J
Time (months) Fig. 3. Patterns of litterfall in deciduous forests (solid line) and eucalyptus plantations (dotted line) in Central Portugal (from Abelho and Graça, 1996). Rains occur in Autumn and they are generally absent in Summer.
CyanMagentaYellowBlack Odonata page 44
44 Manuel A. S. Graça
DRAGONFLIES AND LEAF LITTER Several studies have addressed top-down effects on detritus based systems, looking mainly to fish as top predators (e.g. Konishi et al 2001; Ruetz et al 2002; MacIntosh et al 2005). The results of such studies are inconclusive since in some cases the presence of predators acting on shredders decreased the density of shredders or decomposition rates of leaves, but in other cases there were no significant effects. Besides rivers and streams, organic matter is an important energy source for other aquatic environments such as lakes and tree holes (e.g. Gessner et al., 1996; Paradise 2004). If odonates are top predators in some of those systems, how do they affect litter decomposition? The number of studies addressing this question is low, probably because odonates never reach densities comparable to fish or because invertebrates feeding on leaves are not a numerically dominant prey of odonates. However, in tree hole systems in forests, leaves can be a main resource for several mosquito larvae, facilitating their decomposition (Yanoviak, 1999). Fincke et al (1997) reported that in Barro Colorado Island in Panama, odonates can be a key stone species in tree holes. The presence of odonates increases the mortality of mosquito larvae but their role on the decomposition process is unclear (Yanoviak 1999, 2001). Odonates may not reach numbers capable of affecting decomposition rates in many systems. However, given the large proportion of energy derived from litter inputs to streams and the positive effect of litter accumulation on numbers of stream invertebrates (e.g. Pretty & Dobson, 2004) it is plausible that litter accumulation increase resources for odonates in streams, rivers and lakes, affecting growth and survival. Indeed, in tree holes, increases in the amount of leaves increased growth of odonates (Yanoviak 2001). Hall et al (2000) reported the presence of organic matter in the gut of odonate larvae sampled in a forested stream, which was attributed to the ingestion of shredder invertebrates. Moreover, the same authors referred that approximately 92% of the diet of predators (including odonates) was derived from organic matter. If these values can be generalized for forested systems, litter has a strong bottom-up effect upon odonate assemblages.
REFERENCES ABELHO, M. 2001. From litterfall to breakdown in streams: a review. The Scientific World 1: 656-680. ABELHO, M. 2005. Extraction and Quantification of ATP as a Measure of Microbial Biomass. In: M.A.S. Graça, F. Bärlocher & M.O. Gessner (eds), Methods to Study Litter Decomposition: A Practical Guide, Springer, Dordrecht.
CyanMagentaYellowBlack Odonata page 45
Allochthonous organic matter as a food resource for aquatic invertebrates in forested streams
45
ABELHO, M. & M.A.S. GRAÇA. 1996. Effects of eucalyptus afforestation on leaf litter dynamics and macroinvertebrate community structure of streams in Central Portugal. Hydrobiologia 324: 195-204. BALDY, V. & M.O. GESSNER. 1997. Towards a budget of leaf litter decomposition in a first-order woodland stream. C. R. Acad. Sci. Paris, sciences de la vie / Life sciences 320: 747-758. BASAGUREN, A. & J. POZO. 1994. Leaf litter processing of alder and eucalypt in the Agüera stream system (Northern Spain). II. Macroinvertebrates associated. Archiv fur Hydrobiologie 132: 57-68. BOULTON, A.J. & P.I. BOON. 1991. A review of methodology used to measure leaf litter decomposition in lotic environments: Time to turn over an old leaf? Australian Journal of Marine and Freshwater Research 42: 1-43. BRAY, J.R. & E. GORHAM. 1964. Litter production in forest of the world. Advances in Ecological Research 2: 101-157. CHERGUI, H. & E. PATTEE. 1991. An experimental study of the breakdown of submerged leaves by hyphomycetes and invertebrates in Morocco. Freshwater Biology 26: 97-110. CORTES, R.M., M.A.S. GRAÇA, J.N. VINGADA & S. VARANDAS DE OLIVEIRA. 1995. Stream typology and dynamics of leaf processing. Annales de Limnologie 31: 119-131. CUFFENEY, T.F., J.B. WALLACE & G.J. LUGTHART. 1990. Experimental evidence quantifying the role of benthic invertebrates in organic matter dynamics of headwater streams. Freshwater Biology 23: 281-300. FINCKE, O.M., S.P. YANOVIAK & R.D. HANSCHU. 1997. Predation by odonates depresses mosquito abundance in water-filled tree holes in Panama. Oecologia 112: 244-253. FRIBERG, N. & D. JACOBSEN. 1994. Feeding plasticity of two detritivore-shredders. Freshwater Biology 32: 133-142. GESSNER, M.O., B. SCHIEFERSTEIN U. MULLER, BARKMANN & U.A. LENFERS. 1996. Aquatic Botany 55: 93-105. GESSNER, M. 2005. Ergosterol as a Measure of Fungal Biomass. In: M.A.S. Graça, F. Bärlocher & M.O. Gessner (eds), Methods to Study Litter Decomposition: A Practical Guide, Springer, Dordrecht. GESSNER, M.O., E. CHAUVET & M. DOBSON. 1999. A perspective on leaf litter breakdown in streams. Oikos 85: 377-384 GRAÇA, M.A.S. 1993. Patterns and processes in detritus-based stream systems. Limnologica 23: 107-114. GRAÇA, M.A.S. 2001. The role of invertebrates on leaf litter decomposition in streams – A review. International Review of Hydrobiology 86: 383-393. GRAÇA, M.A.S. & ABELHO M. 2005. Respirometry. In: M.A.S. Graça, F. Bärlocher & M.O. Gessner (eds), Methods to Study Litter Decomposition: A Practical Guide, Springer, Dordrecht. GRAÇA, M.A.S. & R.C.F. FERREIRA. 1995. The ability of selected aquatic hyphomycetes and terrestrial fungi to decompose leaves in freshwater. Sydowia 47: 167-179.
CyanMagentaYellowBlack Odonata page 46
46 Manuel A. S. Graça
GRAÇA, M.A.S., R.C.F. FERREIRA & C.N. COIMBRA. 2001. Litter processing along a stream gradient: the role of invertebrates and decomposers. Journal of the North American Benthological Society 20: 408-420. GRAÇA, M.A.S., J. POZO, C. CANHOTO & A. ELOSEGI. 2002. Effects of Eucalyptus plantations on detritus, decomposers and detritivores. The Scientific World Journal 2: 1173-1185. HIEBER, M. & M.O. GESSNER. 2002. Contribution of stream detrivores, fungi, and bacteria to leaf breakdown based on biomass estimates. Ecology 84: 10261038. KONISHI, M., S. NAKANO & T. IWATA. 2001. trophic cascading effects of predatory fish on leaf litter processing in a Japanese stream. Ecological Research 16: 415-422. MACINTOSH, A.R., H.S. GREIG, S.A. MCMURTRIER, P. NYSTRÖM & M.J. WINTERBOURN. 2005. Top-down and bottom-up influences on populations of a stream detritivore. Freshwater Biology 50: 1206-1218. MALMQVIST, B. & D. OBERLE. 1995. Macroinvertebrate effects on leaf pack decomposition in a lake outlet stream in Northern Sweden. Nordic Journal of Freshwater Research 70: 12-20 MIHUC, T.B. & J.R. MIHUC. 1995. Trophic ecology of five shredders in a Rocky Mountain Stream. Journal of Freshwater Ecology 10: 209-216. MOLINERO, J. & J. POZO. 2004. Impact of a eucalyptus (Eucalyptus globulus Labill.) plantation on the nutrient content and dynamics of coarse particulate organic matter (CPOM) in a small stream. Hydrobiologia 528: 143-165. MOLLES, M.C. 1982. Trichopteran communities of streams associated with aspen and conifer forests: long-term structural change. Ecology 63: 1-6. PARADISE, C.J. 2004. Relationship of water and leaf litter variability to insects inhabiting treeholes. Journal of North American Benthological Society 23: 793-805. PRETTY, J.L. & M. DOBSON. 2004. The response of macroinvertebrates to articially enhanced detritus levels in plantation streams. 2004. Hydrology and Earth System Sciences 8: 550-559. RICKEFS, R.E. 2000. The Economy of nature. Freeman. RODRIGUES, A.P.L. & M.A.S. GRAÇA. 1997. Enzymatic analysis of leaf decomposition in freshwater by selected aquatic hyphomycetes and terrestrial fungi. Sydowia 49: 160-173. RUETZ, C., R.M. NEWMAN & B. VONDRACEK 2002. Top-down control in a detritus based food web: fish, shredders, and leaf breakdown. Oecologia 132: 307-315. SLANSKY, F.JR. & J.M. SCRIBER. 1985. Food consumption and utilization. In: Kerkut G.A. & L.I. Gilbert (eds), Comprehensive insect physiology. Vol. 4. Chapt. 3, Pergamon Press. Oxford. STOUT, B.M., E.F. BENFIELD & J.R. WEBSTER. 1993. Effects of a forest disturbance on shredder production in a sourthern Appalachian headwater streams. Freshwater Biology 29: 59-69 SUBERKROPP, K. 1995. The influence of nutrients on fungal growth, productivity, and sporulation during leaf breakdown in streams. Canadian Journal of Botany 73 (Supp.1): 1361-1369.
CyanMagentaYellowBlack Odonata page 47
Allochthonous organic matter as a food resource for aquatic invertebrates in forested streams
47
TOWNSEND, C.R., HARPER J.L. & BEGON M. 2000. Essentials of Ecology. Blackwell. VANNOTE, R.L., G.W. MINSHALL, K.W. CUMMINS, J.R. SEDELL & C.E. CUSHING. 1980. The river continuum concept. Canadian Journal of Fisheries and Aquatic Sciences 37: 130-137. YANOVIAK, S.P. 1999. Effects of Mecistogaster spp. (Odonata: Pseudostigmatidae) and Culex mollis (Diptera : Culicidae) on litter decomposition in neotropical treehole microcosms. Florida Entomologist 82: 462-468. YANOVIAK, S.P. 2001. Predation, resource availability, and community structure in Neotropical water-filled tree holes. Oecologia 126: 125-133.
CyanMagentaYellowBlack Odonata page 48
This page intentionally left blank
CyanMagentaYellowBlack Odonata page 49
Odonata in Bornean tropical rain forest formations
THE IMPORTANCE OF FORESTS FOR DRAGONFLIES IN DIFFERENT CONTINENTS
49
CyanMagentaYellowBlack Odonata page 50
50 A. G. Orr
This page intentionally left blank
CyanMagentaYellowBlack Odonata page 51
Odonata Bornean Adolfo Cordero Riverain(ed) 2006 tropical rain forest formations Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 51-78.
51
© Pensoft Publishers
Sofia–Moscow
Odonata in Bornean tropical rain forest formations: diversity, endemicity and implications for conservation management A. G. Orr School of Australian Environmental Studies, Griffith University, Nathan, Q 4111, Australia
ABSTRACT The island of Borneo was originally almost completely covered by closed canopy tropical rainforest. Owing to an aseasonal, hot, perhumid climate and high rainfall, forests were well supplied with streams and standing water. Consequently the rich, largely endemic odonate fauna must have evolved in association with these forests, and non-forest species, common today in disturbed land, must formerly have been rare opportunists in forest gaps or localised lacustrine species. It is estimated that at least 70 % of the fauna is presently confined to forest habitats and probably depends on forest for its survival. This study relates odonate distribution to a mosaic of complex tropical rain forest formations in Brunei. The tiny sultanate of Brunei still enjoys about 80% forest cover, representative of all the seven major formations found on the island and a great many of the 30+ sub-formations, and results from a nation-wide survey of odonates from most habitats are considered to be broadly applicable to the entire island of Borneo and many other parts of equatorial south-east Asia. Greatest odonate diversity, both a and b, and greatest endemicity, is found in the primary lowland mixed dipterocarp forests, especially those growing in highly dissected landscapes such as occur at the KBFSC, at the edges of the central uplands. High diversity and endemicity is also found in swamp forest, especially freshwater swamp, with certain endangered peat swamp formations also important. The highly vulnerable kerangas forest harbours fewer species, none uniquely, and the mangrove fauna is still more depauperate, with only a single wide-ranging specialist restricted to this habitat.
CyanMagentaYellowBlack Odonata page 52
52 A. G. Orr
Secondary dipterocarp forest is certainly less rich in odonates than primary forest, but lack of sites for parallel comparisons makes it difficult at present to state how serious this effect is. These results emphasise the importance of conserving a wide range of primary forest formations to achieve satisfactory odonate conservation, a strategy congruent with the conservation of charismatic land-based vertebrates and forest peoples.
INTRODUCTION The ecology of tropical forest Odonata is one of the most significant gaps in our knowledge of the biology of the order. It is well known that, globally, many more odonates occur in tropical ecosystems than in all others combined, and that a probable majority of species are forest dwellers dependent on forests for their survival. At present however, we can neither quantify nor explain this phenomenon. Even in Corbet’s (1999) encyclopaedic review of odonate biology, there are few direct references to the topic, indicative of a dearth of published work. Furtado (1969), in a definitive study of Malaysian dragonfly biotopes and habitat requirements, virtually ignores the surrounding forest in his habitat characterisations, suggestive of an intellectual demarcation line between freshwater biologists and forest botanists. It is most probable that traditional boundaries of research and scholarship have tended to discourage the synthesis of information and ideas from different disciplines necessary to progress in this field. As dragonflies are non-phytophagous aquatic insects, the forest environment in which they occur may seem peripheral to their needs. However when one considers the role forests play in mediating macro- and microclimates, the concomitant effects they have on riparian vegetation, landscape, hydrology and water quality, and the many potential prey items which they support, and the added foraging space created by the third dimension of the tropical rainforest and its canopy, it is obvious that tropical rain forests are very important indeed to the survival of many, perhaps most, odonate species. No clearer evidence of this dependence is provided than by the paucity of stream odonates in equatorial south-east Asian lowlands once surrounding forest has been removed. In a state of nature virtually the entire land surface of Borneo, the third largest island in the world, was covered with forest. Only small alpine areas and inland lakes interrupted the continuous, closed canopy tree cover. These forests contained an extraordinary diversity of plant life, with 10,000-15000 species of seed plants alone (Merrill 1921, 1950). Owing to the perhumid, aseasonal, high rainfall climate all forest habitats were richly supplied with permanent streams, and where topography permitted, permanent or semipermanent standing water. With the encroachment of man, especially over the last 100 years, a good deal of forest has been felled and has given way to
CyanMagentaYellowBlack Odonata page 53
Odonata in Bornean tropical rain forest formations
53
non-forest, variously agricultural or urban and wasteland, or is in varying stages of regeneration. Based on data now 15 years old, Collins et al (1991) estimated forest cover on Borneo at 72.2% of total land area, comprising 53% lowland forest (mixed dipterocarp and secondary formations), 6.2 % montane forests, 10.7 % inland swamp (mostly peatswamp) and 2.3 % mangrove. These figures surely must now be revised down considerably after a disastrous decade of continuing legal and illegal logging and cycles of drought followed by accidental or deliberately set fires. The assessment of Collins et al (1991) was rather inexact in its designation of forest types. Generally six major natural formations are recognised (Cranbrook and Edwards 1994, Ashton 1964, Anderson and Marsden 1984, Whitmore 1984) i.e.: Littoral Forests, Mangrove, Peatswamp Forests, Riparian (mainly freshwater swamp) Forests, Mixed Dipterocarp (including montane) Forests and Heath Forests, each of which has several subdivisions, so that the total number of recognisable formations is well over 30. A seventh major formation, secondary forests, is now a highly significant and diverse component of the island’s vegetation. The various forest formations differ greatly, not only in the physiognomic and floristic differences by which they are defined, but also in edaphic, topographic and hydrological characteristics, all of which have profound consequences for their suitability as dragonfly habitats. The great predominance of perhumid forest habitats in Borneo is reflected in the habitat preferences of its odonate fauna (Table 1). These figures are approximate, with more than 11% classified as indeterminate due either to a gross deficiency of data or to the difficulty of classifying those species occurring in multiple habitats. Habitat definition here is restricted to presumed breeding habitat as many species, especially crepuscular feeders, regularly forage in open areas but return to forest pools to breed, while conversely, species which breed in open ponds and lakes may forage in the forest canopy. In total nearly 70% of all species are confined to forests or require forest habitats in which to breed. For the Zygoptera this figure is over 80%. The 19.3% predominantly found in non-forest are now mostly very abundant, eurytopic species. Presumably these were once localized on lakes, an uncomTable 1. Proportions of Bornean Odonata confined or nearly confined to forest and non-forest breeding habitats (Based on 280 known species and compiled from Lieftinck 1954, Hämäläinen 1994, Orr 2001, Orr 2003, Dow 2005, Kalkman 2005 and Orr, unpublished data).
Zygoptera Anisoptera TOTAL
Forest
Non-Forest
Indeterminate
82.7% 56.7% 69.3%
10.7% 27.3% 19.3%
6.6% 16.0% 11.4%
CyanMagentaYellowBlack Odonata page 54
54 A. G. Orr
mon habitat in most of Borneo, and temporary pools in forest gaps, and in many cases would have been quite rare. However, despite changing conditions, very large areas of Borneo remain forested and provide habitat for the majority of the rich odonate fauna, especially its endemic elements. From a conservation viewpoint, it is important to know how this forest fauna is distributed within the diverse forest formations, and how well protected these formations are.
FAUNISTIC STUDIES OF DRAGONFLIES IN BORNEO To date, regional or local faunistic surveys of dragonflies in Borneo have been very limited, especially those that relate occurrence to habitat (in a sense understood by modern forest ecologists). Exceptions to this are Brunei, (Thompson and van Tol 1993, Orr 2001, Kalkman 2005), Gunong Kinabalu (Laidlaw 1934, Hämäläinen 1994), in which old or uncorrelated records can easily be related to modern phytogeographical knowledge, and the Danum Valley Field Centre in eastern Sabah, where similar lists can be related to a well known landscape. Approximately 64 percent of the Bornean odonate fauna has been recorded from within the small sultanate of Brunei (area 5765 km2), covering less than one percent of the land area of Borneo, hence this may serve as an exemplar for the whole island.
LANDFORMS AND FOREST HABITATS OF BRUNEI Brunei provides a microcosm in which almost every major vegetational formation found on the island of Borneo is well represented. (Lacking are the mixed dipterocarp forests on ultramafic soils of eastern Sabah (Proctor et al. 1988), pure limestone formations and upper montane forest; the montane forest present is very difficult to access and too limited in extent to support extensive stream systems). Moreover, with over 1900 tree species and 3500 species of all seed plants (Wong, 1999), Bruneian forests are both exceptionally rich, and exceptionally well studied floristically (Ashton 1964, Anderson and Marsden 1984), hence it is possible to state with some precision the floristic differences between forests of very similar physiognomy. Because of sound conservation policy and practice (more than 20 % of the land area is protected), microhabitats can be revisited over many years without suffering significant anthropogenic disturbance. Despite some destruction by fire in 1997, total forest cover is probably still nearly 80% of land area (Table 2). Because Brunei is relatively low lying, the area of peatswamp forest is substantially greater than the average for the island of Borneo.
CyanMagentaYellowBlack Odonata page 55
Odonata in Bornean tropical rain forest formations
55
Table 2. Extent and composition of Brunei forests (based on Anderson and Marsden, 1984). Forest formation Primary lowlandmixed dipterocarp Montane oak/laurel Peatswamp Mangrove Kerangas (heath) Alluvial and other freshwater swamp Littoral Secondary TOTAL
Area km2
% of land area
2158 72 909 184 35 127 Figures unavailable 1279 4764
36.5 1.2 15.4 3.1 0.6 2.1 Small 21.6 80.5
Fig. 1. Distribution of major forest formations of Brunei (secondary forests not separated from primary mixed dipterocarp)
CyanMagentaYellowBlack Odonata page 56
56 A. G. Orr
Figure 1 shows the distribution of main forest formations in Brunei. Remaining littoral forest, kerangas, and most types of freshwater swamp forest, including alluvial forest, occupy areas too small to be indicated on this map, and either exist as small patches interdigitating with or else reticulating throughout other forest types, especially peat swamp and lowland mixed dipterocarp forest; nevertheless some minor formations provide very important odonate habitats. Throughout Borneo, forest formation is influenced primarily by topography and substratum. Mangroves, grow on tidally inundated silty deltas and are highly salt tolerant. Littoral forests grow in slightly more elevated situations near the sea, mostly growing on undeveloped sand or mixed clay and sand. Generally the forest is fairly open and water, running and standing, is scarce. On the low plains inland behind the beaches are broad expanses of peatswamp forest. Thickness of peat varies considerably, to a maximum of about 20 m. Deposits tend to be laid in patches several km in diameter with the peat, lens shaped in section, deep and raised in the middle and thin at the edges (Anderson 1964). In normal (non-drought) conditions the peat is completely waterlogged, although exposed standing water tends to be more available around the edges. The pH of water in pools deep in the forest may reach 4.0 or even lower. Peatswamp occurs in several sub-formations, which form concentric bands around the peat dome. Anderson (1961) recognised a catenary sequence of 6 formations, ranging from a mixed or Campanospermun dominated outer band, through several Shorea albida dominated formations, (the most impressive being virtual monocultures, recognised from the air by the grey-green canopy of nearly uniform height, up to 80 m), through to stunted pole forest in the centre. The complete sequence is present only in the oldest and deepest peat deposits, up to 11,000 years old. Often the central area is occupied by inner tall Shorea albida formations. Interspersed among the peat deposits are low islands of pure white sand, and sometimes higher outcrops of various shale/sandstone/clay formations. On the sand grows kerangas forest, often dominated by the conifer Agathis borneensis. On the shale/sandstone/clay formations grows lowland mixed dipterocarp forest. At the ecotones between kerangas and peatswamp, and between mixed dipterocarp forest and peatswamp, are often found tannin stained freshwater swamp associations, distinguished from true peatswamp by the lack of peat in the substratum and the higher pH of the water, usually above 5.5. Another form of freshwater swamp occurs in alluvial forest, growing in low lying, flat or gently undulating country, beside streams and rivers with a developed floodplain. The most extensive kerangas formations were formerly found just inland in a long belt between coastal littoral formations and inland peat; almost all are now destroyed, principally by repeated fire over the last three decades.
CyanMagentaYellowBlack Odonata page 57
Odonata in Bornean tropical rain forest formations
57
As the terrain rises to more than a few metres a.s.l., the forest becomes almost entirely lowland mixed dipterocarp, a formation which may vary greatly floristically according to topography and soil type, while retaining an almost uniform physiognomy. Bornean mixed dipterocarp forests are among the tallest statured rainforests in the world (Ashton 1964, Whitmore 1984), with highly complex vertical zonation, and, with over 231 tree species per hectare (Paulsen et al 1996), also the most floristically diverse. The principle odonate habitats are streams, both small and large, although springs, seepages, and phytotelmata are also significant. The most diverse microhabitats are provided by highly dissected landscapes in steep terrain. Lowland mixed dipterocarp formations undergo a transition to the similar hill dipterocarp, from about 500-1000 m, above which they are succeeded by montane oak/laurel associations, then elsewhere in Borneo by ericacious upper montane forest. The latter is present above 2000 m and is not represented in Brunei.
GENERAL HABITAT ASSOCIATIONS OF DRAGONFLIES IN BRUNEI Orr (2001), lists 174 species from 35 sites in Brunei, 25 of which were forest habitats of known formation. Later records (Kalkman 2005, Orr unpublished) bring this total to 179 species. A full list is provided in Appendix 1. Of these a total of 12 species (14.3%) of Zygoptera and 36 species (37.9%) of Anisoptera were generally found unassociated with forest (although a few occasionally entered forest near its margins or were routinely present on exposed hilltops and forest canopies). Information concerning almost all species in this category was supported by large samples from multiple sites, and for each species, at least 95% of records were made outside forest, including areas well removed from forest. Table 3 summarises the family composition of nonforest and forest odonate faunae, from which it may be seen that the former is highly unbalanced and unrepresentative of the fauna as a whole, being composed mainly of libellulids and coenagrionids. It is concluded that these, mostly wide ranging, common and eurytopic species did not require forest as essential habitat, and, at most, used it facultatively for foraging or occasionally for breeding. It is believed that if the forests of Brunei were felled, this assemblage, 26.8 % of the known fauna, would continue to flourish and many species would probably increase in numbers and distribution. These need concern us no further. Of the remaining 73.2 % of species, it is probable that most are completely dependent on forests (or at least forest margins) for their existence as none was ever found associated with water at any distance from forest. Several crepuscular aeshnids were observed foraging in open country or were at-
CyanMagentaYellowBlack Odonata page 58
58 A. G. Orr
Table 3. Numbers of species in major families and family groups associated with forest and non-forest habitats in Brunei (Data from Orr, 2001; Kalkman, 2005) Family Calopterygoidea Coenagrionidae Other Zygoptera Libellulidae Other Anisoptera TOTAL
Non-forest
Forest
Total
0 11 1 31 5 48
26 13 34 26 32 131
26 24 35 57 37 179
tracted to light, as were a few stray coenagrionids which entered my house, about 100 m from the nearest closed forest, during the six years I lived in Bandar Seri Begawan, but it is quite certain that the breeding habitats of these species were well inside closed canopy forest.
HABITAT ASSOCIATIONS OF DRAGONFLIES WITHIN BRUNEI FORESTS Of the forest dependent species the questions remain: to what extent are species associated uniquely with particular forest formations? And, to what extent are the species assemblages supported by different forest formations characteristic to those formations? The analysis presented by Orr (2001), suggests answers to both these questions: similar habitat types clustered together in a similarity analysis, and the family profiles of mixed dipterocarp forest sites were rather similar, and quite different from those of swampy forested areas (Figure 2). The present analysis is more directly species oriented, with a definite aim of defining habitat preferences in relation to various forest formations. To the 25 forest sites listed by Orr (2001), it was possible to add another 7 forest sites (Orr, unpublished), giving a total of 32 sites. This approximates the total number of forest formations recognised for Brunei (Cranbrook and Edwards 1994), and although several of these (Mangrove and Littoral formations) are largely irrelevant to this exercise, since they support poor faunas and were poorly sampled, and montane forest was omitted for logistic reasons, it is obvious that duplication of similar formations was in many cases impossible. The five lowland mixed dipterocarp forest formations sampled represent three different soil types with up to 80 % difference in the tree species present between sites (Ashton 1964, and pers com). Added to this is the problem that it was impossible to duplicate the stream habitats within each sampling area, and it is difficult to say whether observed differences
CyanMagentaYellowBlack Odonata page 59
Odonata in Bornean tropical rain forest formations
59
Fig. 2. Average family profiles of odonate species assemblages from sites in mixed dipterocarp forest, swamp forests & kerangas, and non forest (modified from Orr 2001).
relate in some fundamental way to forest formation or its topographic or geological/edaphic correlates, or rather are just a consequence of local topography. Therefore, in the first instance, I consider the associations of dragonflies relative to a coarse-grained classification of forest formations, namely, lowland mixed dipterocarp, secondary (formerly mixed dipterocarp), peatswamp, freshwater swamp and alluvial forest, kerangas (tropical heath) and mangrove. In order to avoid the spurious inference of habitat association which may result from single records, I have adopted the following protocol in my analysis of associations – species are included only if they are recorded from at least three sites or at least five individuals, these latter represented in at least two samples separated by not less than one month. As Table 4 shows, the greatest number of species occurs in mixed dipterocarp forest with nearly half of all forest species confined to mixed dipterocarp. Eleven species were found in both mixed dipterocarp and freshwater swamp,
CyanMagentaYellowBlack Odonata page 60
60 A. G. Orr
Table 4. Total species in six main forest formations and degree of overlap between formations (principal diagonal gives number of species confined to a particular formation, cross references give number of shared spp.). Based on 100 forest species for which minimum data requirements are met, data mainly from Orr (2001) and the parent data set. Figures represent numbers of species, but may also be read as percentages (because of fortuitous sample size). md, mixed dipterocarp; fw, freshwater swamp; ps, peatswamp; kg, kerangas; mg, mangrove; sd, secondary dipterocarp.
md fw ps kg mg sd Total spp. in habitat
md
fw
ps
kg
mg
sd
47
11 8
2 21 3
2 6 4 0
0 1 1 0 1
62
44
23
7
3
10 13 7 2 1 0 18
generally in areas where the two habitats lay close together. Other overlaps between mixed dipterocarp and other forest types were few, except for secondary dipterocarp, with which 10 species were shared. The next richest habitat was freshwater swamp forest, with 44 species, 8 of which were found nowhere else. True peatswamp forest was about half as rich with 23 species, 21 of which were shared with freshwater swamp. Only three species were confined to peatswamp. Only seven forest species were present in kerangas (although some non-forest species were present as vagrants or foraged there sporadically), none of which was unique to the habitat. Similarly mangrove was inhabited by only three forest species, one specialist, with two others shared. Secondary dipterocarp forest was rather poor compared with its parent formation (but see below), with only 18 species, many shared with other formations. Non-forest species were more likely to encroach in secondary formations than in primary mixed dipterocarp forest.
ENDEMICITY AND FOREST FORMATIONS Of forest species found in Brunei, 52.7% are Bornean endemics, a further 29.0 % are regional endemics, confined to Sundaland and Palawan and only 18.3% are wide ranging. Conversely, only one non-forest species, Pseudagrion lalakense, is endemic and only 10.4% of non-forest species are regional endemics; 87.5% are wide-ranging Table 5. Of species found in mixed dipterocarp forest, 56.5% are Bornean endemics and 29.0% are regional endemics. A relatively small proportion are wider
CyanMagentaYellowBlack Odonata page 61
61
Odonata in Bornean tropical rain forest formations
Table 5. Levels of endemicity and regional endemicity amongst forest and non forest species as indicated by numbers and percentage of total forest and non-forest species respectively. Borneo endemic Sundaland + Palawan
Confined to
Wider ranging
Total
69 (52.7%) 1 (2.1%) 70
38 (29.0%) 5 (10.4%) 43
24 (18.3%) 42 (87.5 %) 66
131 48 179
Forest Non-Forest TOTAL
Table 6. Levels of endemicity and regional endemicity amongst mixed dipterocarp dwelling species versus swamp forest species, expressed as numbers of species and percentages of totals found in each forest type (data deficient spp excluded) Borneo endemic Mixed dipterocarp Swamp Forest (all formations) TOTAL
Confined to Wider ranging Total Sundaland + Palawan
35 (56.5%) 17 (35.4 %)
18 (29.0%) 22 (45.8%)
9 (14.5%) 9 (18.8%)
62 48
52
40
18
100
ranging. Within swamp forest, a lower percentage (35.4 %) are Bornean endemics but 45.8% are regional endemics. Less than 20 % are wide ranging. Clearly, both mixed dipterocarp forests and swamp forests provide key habitat for endemic and regional endemic species. This is essential information for conservation planning.
ACCOUNTS FOR EACH FOREST TYPE Lowland Mixed Dipterocarp Formation Of the forest assemblages sampled by Orr (2001), the greatest numbers of species (a diversity) and the greatest variability between sites (b diversity) was among the mixed dipterocarp formations, especially stream habitats. In some cases distinct assemblages occupied different microhabitats within the same forest formation and were separated by less than 50 m (notably in the steep, highly dissected terrain of the Kuala Belalong Field Studies Centre KBFSC), as shown by Figures 3-6, where 54 species have been recorded from within an area about 100 m square. As an obvious corollary, such sites were also were home to the highest proportion of endemic species. Of the sites sampled extensively, several associations were recognised, based on a similar-
CyanMagentaYellowBlack Odonata page 62
62 A. G. Orr
ity analysis of species assemblages (Orr, 2001). The following summary demonstrates the considerable differences which occur between selected sites. Kuala Belalong formation (soil: Setap shale formation, grey clay and shale) Myrmeconauclea/ cobbled open canopy stream association (Figure 4) – In Brunei this was represented only by the Sungei Belalong (KBFSC) and nearby Sungei Temburong, but similar broad streams in the Crocker ranges and Danum Valley, Sabah have similar faunas (Hamalainen in lit, Orr, unpublished data). At the point illustrated in figure 4 the stream runs north south, and receives direct sunlight for approximately half the day. Humidity remains high all day (>80% RH) owing to the closely bordering vegetation. The rheophyte Myrmeconauclea probably plays little role in the lives of odo-
20 sp forest 3 sp
20 sp
tributary
1 sp 2 sp
main stream 15 sp
N 100 m Fig. 3. Contour map of Kuala Belalong Field studies Centre (KBFSC), Brunei, showing proximity of sites with distinct species assemblages; numbers within areas enclosed by dashed lines indicate number of species at each site; circled numbers indicate number shared between sites. (contour intervals 5 m). Data from Orr, 2001.
CyanMagentaYellowBlack Odonata page 63
Odonata in Bornean tropical rain forest formations
63
period when direct sunlight reaches stream 1200h 0900h
1500h
0600h
1800h
height (m)
30
20
10 0
10
20
30 m
0
Fig. 4. Profile of stream and riparian forest on the main stream, shown in figure 3, at KBFSC. (Compilation of forest mensuration data from several sources).
nates, but serves to identify the association. This was the only habitat in Brunei which supported many calopterygoid species, such as Neurobasis longipes, Vestalis amoena, Heliocypha biseriata, Libellago stictica, Rhinocypha aurofulgens, and, Euphaea tricolor. Libellago semiopaca, Rhinocypha cucullata and Dysphaea dimidiata, shared with other mixed dipterocarp streams were also present. The chlorocyphids all oviposited in large, semi-submerged logs and developed in leaf trash in eddies behind such obstructions, whereas N. longipes, V. amoena and E. tricolor larvae lived in riffles, often in leaf packs. The only other zygopteran was Prodasineura verticalis, which oviposited in root masses where the stream banks were undercut. Several gomphids (Megalogomphus, Sieboldius, Microgomphus ) were present as were corduliids such as, Macromidia fulva, Macromia westwoodii and other Macromia species. In most cases gomphids and corduliids were more easily sampled in the larval stage. The libellulids Onychothemis coccinea and Orthetrum pruinosum, both present in other mixed dipterocarp were common, while Zygonyx iris was rare. Rocky closed canopy stream association (Figure 5)– a small tributary ascending abruptly from the main stream at KBFSC, the Sungei Mata Ikan, with mainly rocky bed and steep banks on which few rheophytes grew. At the site shown in profile in Figure 5 a short reach runs nearly north south. Differ-
CyanMagentaYellowBlack Odonata page 64
64 A. G. Orr
20
10
height (m)
30
0
0
10
20
30 m
Fig. 5. Profile of stream and forest on the tributary, shown in figure 3, at KBFSC . (Compilation of forest mensuration data from several sources).
ent sections of the stream are sequentially illuminated by broad patches of sun throughout the day, especially from 0900-1500h. Humidity remains close to that of the forest understorey (RH > 85%). Only two species were shared with the main stream. Typical species were Devadatta podolestoides, Vestalis amaryllis, V. atropha, Rhinocypha humeralis, R. stygia, Euphaea impar (rare), E. subcostalis, Rhinagrion borneense (rare), Coeliccia cyaneothorax and Drepanosticta rufostigma, among Zygoptera. Indaeschna grubaueri and Cratilla metallica, widespread forest understorey species, here occurred beside a permanent leafy rock pool. Heliogomphus blandulus is believed to prefer this habitat. Macromia westwoodi was more common here than on the main stream. Other gomphid and corduliid species may be present. Seepage/ forest understorey association (Figure 6) – This formation is exemplified by the area around a marshy spring at the head of the Sungei Mata Ikan and in the surrounding forest. The forest stature is greater than beside the stream banks and only small sun flecks reach the forest floor.
CyanMagentaYellowBlack Odonata page 65
Odonata in Bornean tropical rain forest formations
65
20
10
height (m)
30
0
marshy area
0
10
20
30 m
Fig. 6. Profile of forest, including marshy area, within the ‘forest’ site shown in figure 3, at KBFSC. (Compilation of forest mensuration data from several sources).
Humidity at ground level is always high (RH > 90%). At this site the following species are found: Devadatta podolestoides, Vestalis beryllae, Bornargiolestes sp, Stenagrion dubium, Coeliccia borneensis, C. nigrohamata, Drepanosticta attala, D. forficula, D. versicolor, Drepanosticta sp. A, Drepanosticta sp. B, Protosticta sp. A, Protosticta sp. B, Leptogomphus pasia, L. pendleburyi. Some species are known to perch in sun flecks higher in the subcanopy 10-20 m above the ground. A similar, but less well sampled fauna, is known from seepages in the dipterocarp forest at Sungei Ingei. Pinanga/Dipteris/gravel and sand open stream formation (Soil: Lambir formation, sandstone and ahale with thin limestone and marl) A common association in several localities in Brunei, especially the Labi Hills is open or semi open streams with rheophytic vegetation dominated by the fern Dipteris lobata, in the rockier less exposed places, and the low palm Pinanga tenella. In general the terrain is more gentle than KBFSC, and it is difficult to separate small and large stream associations, hence slight overlap with both stream associations at KBFSC will be noted. Species present include Devadatta podolestoides, Vestalis amabilis, Libellago aurantiaca, Sundacypha petiolata, Dysphaea dimidiata, D. lugens, Euphaea ameeka, E. impar, Rhinagrion born-
CyanMagentaYellowBlack Odonata page 66
66 A. G. Orr
eense, Coeliccia sp. A aff macrostigma, Drepanosticta rufostigma, Elattoneura analis, Prodasineura sp. aff hosei, Leptogomphus mariae, Macrogomphus parallelogrammus, Megalogomphus sumatranus, Oligoaeschna platyura, Macromia cincta, M. corycia, Orthetrum pruinosum, Onchothemis coccinea (rare), O. culminicola, Zygonyx iris. Pandanus/ fine gravel-muddy open stream (merges into freshwater swamp) (soil: Belait formation, soft sandstone, clay and lignite). This association occurs in a number of locations in Brunei, notably Sg Ingei, where Thompson sampled in 1991 (Thompson and van Tol 1993). Species recorded were rather similar to the last formation and included Vestalis amabilis , Libellago aurantiaca, Rhinocypha cucullata, Sundacypha striata, Dysphaea lugens, Euphaea ameeka, E. impar, Rhinagrion borneense, Elattoneura analis, Prodasineura sp. aff hosei, Burmagomphus insularis, Macrogomphus parrallelogrammus, Megalogomphus sumatranus, Idionyx yolanda, Macromia corycia, Orthetrum pruinosum, Onychothemis coccinea and O culminicola. Phytotelmata association – The mixed dipterocarp phytotelmata association appears to be fairly uniform across all lowland mixed dipterocarp formations. Even in forests where no phytotelmata could be located, placing artificial humus and water filled containers near the base of tree trunks invariably attracted members of the association (Artificial containers did not attract odonates in swamp or kerangas). It is probable however that large buttress pans, which host the richest communities (Orr, 1994, Kitching and Orr, 1996), tend to be most common where the topography is steep, as the buttresses on the upper side of the slope are complexly folded, creating watertight depressions. Moreover, most such trees sampled at KBFSC were ‘Belian’ (Eusideroxylon Lauraceae) which appears particularly given to such basal growth forms. This tree species is rare or absent from many mixed dipterocarp formations. Most phytotelmata do not dessicate, even in drought conditions (Orr, 1994), unlike those of seasonal forests in Panama (Finke 1992, in litt). Pericnemis triangularis is present in most phytotelmata including small rot holes, but also large buttress pans. Lyriothemis cleis is almost always present in medium to large phytotelmata. The larvae may suffer extreme dessication without harm (Orr, 1994). Indaeschna grubaueri is present in most buttress pans, but also breeds in ground pools where the terrain permits their presence. Cratilla metallica prefers ground pools but will also develop in large ‘log holes’ and disturbed buttress pans. The very rare Camacinia harterti has been reported ovipositing in buttress pans (Lieftinck, 1954) and may belong to this assemblage, (although listed below as a non-forest species). The phytotelmata assemblage, although small, is interesting in that it is clearly absolutely dependent on the presence of intact forest for survival. Each phytotelma represents an isolated community, with an allochthonous energy supply in the form of leaf litter. Bruneian communities are notable for their
CyanMagentaYellowBlack Odonata page 67
Odonata in Bornean tropical rain forest formations
67
Predators Indaeschna Lyriothemis Pericnemis Detritovores Tipulidae Scirtidae Other 0
50 100 relative biomas (% total)
Fig. 7. Pyramids of biomass and proportional contribution by major species for five large treeholes in Brunei. After Orr, 1997.
relatively simple structures, with few species of detritovores and lower level carnivores, and a high biomass of top predators, invariably odonates, which often exceed in biomass, lower trophic levels Orr (1997), indicated by Figure 7. It is possible that small phytotelmata are found in the sub canopy, and might be utilized by certain Zygoptera which have been collected in canopy fogging operations. However the existence of such microhabitats remains hypothetical, and if present, they may not be sustainable as long term sources of standing water given low diurnal relative humidity in the canopy and resulting high evaporation. Freshwater Swamp Formation The freshwater formations best sampled were alluvial forest and stunted forest growing in depressed waterlogged areas in dipterocarp or at the kerangas/peatswamp boundary. These are believed to support the richest odonate assemblages, with at least 8 species not found in other formations . Tidal brackish swamp forest - Poorly sampled but known to support an odonate assemblage, believed to be species poor and comprised of mainly eurytopoc species. Tidal freshwater swamp forest - poorly sampled but known to support an odonate assemblage, grading into the next category. Alluvial forest with silty streams – Low growing forest, many of the small trees with prop roots and pneumatophores, present beside many lowland streams and larger rivers. Limited variation between similar formations but any given site may be very rich. Thompson and van Tol (1993) record 35 forest species from this formation in an area adjoining three other forest types. Most species are found in pools and runnels above the stream level and include: Vestalis amabilis, Libellago aurantiaca, L. hyalina, Sun-
CyanMagentaYellowBlack Odonata page 68
68 A. G. Orr
dacypha striata, Orolestes wallacei, Podolestes orientalis, Amphicnemis remiger, A wallacei, Archibasis incisura, A. tenella, Teinobasis rajah, Copera vittata, Prodasineura sp. aff hosei, Gynacantha basiguttata, G. dohrni, Leptogomphus coomansi, Macrogomphus quadratus, Rhisiophlebia dohrni, Brachygonia oculata, Nannophyopsis chalcosoma, Tyriobapta torrida, Agrionoptera insignis, Cratilla lineata, Nesoxenia lineata, Orchithemis pulcherrima, Pornothemis serrata. Tannin stained, non-acidic freshwater swamps at borders of peatswamp and kerangas. Low formation of small trees, dense undergrowth, (possibly should be classified as marginal peatswamp forest). A rich assemblage of swamp forest species, including several possibly confined to this formation, or at least most common in it: Libellago hyalina, Amphicnemis erminea, Archibasis melanocyana, A. viola, Coeliccia macrostigma, Prodasineura collaris, Prodasineura dorsalis, Gynacantha basiguttata, Heliaeschna crassa, Oligoaeschna sp, Ictinogomphus acutus, Hemicordulia tenera, Metaphya micans (DD), Macromia cincta, Brachygonia oculata, Tyriobapta laidlawi, Cratilla lineata, Orchithemis pulcherrima, O. pruinans, O. xanthosoma, Pornothemis serrata, Pseudagrionoptera diotima (DD) Shallow depressions in MDF (tannin stained, non acidic water). Depending on the size of the basin, this is a distinct forest formation with low trees with prop roots and a dense undergrowth of rattans, or may be a small patch in Dipterocarp forest, with few small trees growing in swampy ground. Typical swamp forest assemblage with few unique: Vestalis amabilis, Podolestes orientalis, Onychargia atrocyana, Amphicnemis erminea, A. wallacei, Archibasis viola, Coeliccia sp. A aff macrostigma, Prodasineura collaris, P. haematosoma, P, hyperythra , Prodasineura sp. aff dorsalis, Prodasineura sp. aff hosei, Gynacantha basiguttata, Tyriobapta kuekenthali, T. laidlawi, T. torrida, Orchithemis pulcherrima. Peatswamp forest – Of six recognised formations (Anderson 1964, Davies and Kamariah, 1999), four were sampled. In general formations of mixed species at the outer margins of the peat deposit where the peat layer is thinner were richer in species than Shorea albida monocultures on deeper peat deposits. In areas where the peat is raised ground water may be very limited and few odonates present. Most peatswamp species are shared with freshwater formations. Three species which appear confined to outer to middle formations in low wet forest are Podolestes chrysopus, P. harrissoni and Amphicnemis martini. Kerangas – formations close to swamp forest often harbour swamp species which may establish feeding territories at high abundance (Orr 2004b), especially Libellago hyalina and several libellulids. Rare ground water attracts widespread swamp and dipterocarp forest species such as Cratilla metallica. Small, clear streams in elevated kerangas supported Prodasineura sp.
CyanMagentaYellowBlack Odonata page 69
Odonata in Bornean tropical rain forest formations
69
aff hosei, and rarely, P. dorsalis. Kerangas growing on hilltops in dipterocarp forest was generally dry and devoid of odonates, except in hilltop clearings where non-forest species foraged. Mangroves - A total of six formations are recognised (Cranbrook and Edwards 1994) Typically 3-5 zones seaward to landward – in the Rhizophora formation the specialist Raphismia bispina occurs, apparently breeding in oligohaline water. Other species occurring in mangrove do so facultatively, including several non-forest generalists, and seldom venture beyond the landward Xylocarpus or Bruguiera zones. Nipah occurs upstream in brackish water, and barely qualifies as forest. No specialized odonata are known from this formation. Secondary dipterocarp -A heterogeneous formation, often with swampy elements. Secondary formations sampled were not as diverse as the total primary dipterocarp forest, in terms of soils, topography and stream types. The numbers of species recorded are boosted by invasions from nearby swampy areas. Consequently, although it is clear that secondary dipterocarp is less rich than the parent primary forest, it is very difficult to quantify this effect with present data. Pericnemis stictica, a key element of the phytotelmata assemblage, appears to be absent. Ecotones: megadiverse sites – in many situations on the plains of Brunei several different forest types may interdigitate forming a mosaic over an area of as little as one km2. Thompson and van Tol (1993) report 43 forest species from Sungei Ingei, over a mosaic of mixed dipterocarp forest, alluvial forest, peatswamp forest and kerangas. Such sites are of undoubted conservation value, although as sites for ecological investigations they may be less useful, as the proximity of different habitats may obscure the species associations usual for each forest formation.
DISCUSSION Orr (2001) concluded that the richest odonate assemblages in Brunei were present in primary mixed dipterocarp formations, especially in sites with a highly dissected landscape, allowing high species turnover between microhabitats, and although family profiles tend to remain similar between dipterocarp forests on different soils, species assemblages vary considerably. Thus both a and b diversity of the total fauna is maintained principally by assemblages occurring within this formation. Orr (2001) also concluded that the contribution to a diversity of freshwater swamp formations was very important. The information presented above, based on almost the same data set, affords similar conclusions.
CyanMagentaYellowBlack Odonata page 70
70 A. G. Orr
The data set of the Orr (2001) study included about 1250 specimens collected within forests, supplemented by well over 10,000 observational records, and yet in some respects it was limited as a survey of habitat associations. The uniqueness of many sites within Brunei, and impracticality of working beyond its boundaries meant sites were not adequately duplicated. Some sites were sampled for larvae and others were not. Locating equivalent stream habitats within different dipterocarp formations was at times impossible, partly because of the inaccessibility of many places in Brunei. Clearly a study over a more extensive area, including sites in western Sabah and northern Sarawak would yield a broader-based data set from which generalisations could be made with greater confidence. In tropical rainforest surveys it is desirable to sample both adults and larvae to achieve a proper balance in the data. There is also a constant need for more baseline data. A significant number of species known from north Borneo have not yet been named. More probably await discovery. Life histories are known in detail for only a handful of species, and in many cases early stages are completely unknown. We understand little about the way odonates utilize the physical structure of the forest. For example many gomphids and females of other families appear on stream beds only briefly during the day, and may be seen disappearing into the subcanopy, where presumably they spend most of their time. The canopy and subcanopy as dragonfly habitat remains virtually unknown, so that even a few days dedicated observation from one of the many canopy towers and walkways now available throughout the tropics might contribute significantly to our knowledge. The best understood system within the forest is the simplest; the phytotelmata association. The reasons various species are restricted to forest habitats are poorly understood. It was not possible in the Brunei study to document the effect of forest clearance, except in the case of one freshwater swamp association adjoining peat swamp and extensive areas of peat swamp, in which the forest fauna was replaced by a depauperate assemblage of non-forest species following removal or death of trees. Parallel surveys in forested areas and equivalent areas from which forests have been cleared, especially stream habitats, are desirable firstly to determine which species are genuinely dependent on the presence of intact forest, and secondly to document as many environmental parameters as possible which might be responsible for observed associations. (A little work on this theme has been done at Danum Valley centre in eastern Sabah, where a known history of logging and regrowth makes it ideally suited to such studies, but most have been carried out by undergraduates and are of very limited scope.) For deep forest species, forest clearance obviously radically alters the physical environment, raising the temperature of the air and water in seepages, lowering the relative humidity, possibly exceeding the innate tolerances of
CyanMagentaYellowBlack Odonata page 71
Odonata in Bornean tropical rain forest formations
71
both adults and larvae. For larger streams, clearing forests may radically alter the hydrology of the system, causing erosion of banks and increasing turbidity of water. Lack of forest to retard run-off can result in more violent spates following every rain storm, leading to scouring of sand and silt deposits (Ross & Dykes, 1996). The amount of vegetable detritus in leaf packs in riffles and in deep macerated deposits may be greatly reduced, depriving larvae of living space and detritivorous prey items. Rheophytes essential for oviposition may not grow in exposed situations or may be out-competed by weedy grasses. The relative humidity of the air above the stream may drop significantly with forest clearance. Many rainforest insects are poorly adapted to withstand low humidity, and it is known that in cleared urban areas in Brunei RH drops to 60% or lower in the middle of the day, an effect that also occurs in the forest canopy. An observation I have often made, is that in the Australian wet tropics in open forest and in Thailand, especially at higher altitudes and in what was formerly monsoon forest, rivers and streams support quite rich faunas, including some species which occur also in deep closed canopy forest. In Brunei, at 5° latitude deforested streams support poor faunas. It is not clear if this difference is due to a lesser reduction in RH in cleared areas in tropical areas 10-15° latitude from the equator, or whether the forest odonates from more seasonal and drier regimes are better adapted to dry conditions (in some cases the same species are involved), or if other variables are involved. Simple measurements of environmental parameters would help resolve these questions. In brief, we understand to some extent the pattern of habitat associations in Borneo, and these results are expected to apply at least to most of Sundaland. Further data would refine and broaden the picture, but probably not radically alter our present view. On the other hand we have a very poor understanding of the underlying process(es) which have produced this pattern. It is difficult to say exactly what most needs investigation, as any study of ecology or behaviour is likely to further our understanding in some important way. An understanding of the ecological requirements of odonates is critical to their conservation. Our present understanding of patterns of distribution indicate that both primary mixed dipterocarp and swamp forest, especially freshwater swamp, are key habitats, necessary to the continuing existence of the greater part of the Bornean odonate fauna. As conservation is most effective when concentrated on unusual and endemic forms (Moore 1997), the same conclusions are reinforced by analysis of endemicity among forest species. Unfortunately these same forests are under intensive pressure from logging concerns, especially mixed dipterocarp forest and Shorea albida peatswamp forest, the latter designated as an endangered ecosystem (Davies and Kamariah 1999). Kerangas forest, which may include odonate rich swamps
CyanMagentaYellowBlack Odonata page 72
72 A. G. Orr
around its margins, exists now only in tiny pockets and is highly vulnerable to fire. Draining of peatswamps (Davies and Kamariah 1999) has catastrophic consequences for the forests, the hydrology and dragonflies. It does not seem likely that odonate conservation can compete with legalised logging, except in Brunei which is supported by its petroleum industry, and where, with 10% of the land area and all critical habitats protected, the odonate fauna can be regarded as secure, despite unfortunate habitat destruction due to fire in peatswamp and kerangas during el Niño droughts (Orr, 2001). For the rest of Borneo, including Sarawak, Sabah and Kalimantan, about 5% of the total land area lies in gazetted reserves and adequate areas of a majority of essential habitats, as presently understood, are included in this category (Orr 2003). There is however considerable doubt as to whether nominal protection is adequately enforced, and the disastrous cycle of forest fires seems likely to continue, despite the fact that the associated smoke pall makes human habitation in cities in the region extremely unhealthy and uncomfortable. Thus, in Borneo, and much of south-east Asia, odonate conservation equates with forest conservation. This has the advantage that the same conservation policies are good not just for endemic dragonflies, but more charismatic species, orang-utan, rhinoceros, gibbons, birds, butterflies and indigenous forest-dwelling humans. Nevertheless, this does not negate the importance of special efforts directed at Odonata, and it is an unfortunate fact that the island of Borneo is so poorly studied that not a single species can be with certainty given a definite conservation ranking (Orr 2004a), although certain genera are believed to be at least vulnerable. Conservation is as much a political as a scientific exercise. Having the information to flag an iconic species as endangered appeals directly to political sensibilities. Vague statements of data deficiency do not carry same weight. There is a pressing need to involve local people in the study of odonatology and dragonfly conservation. To this end, there is a very great need for local field guides. At present the expertise and the will to produce such guides exists, but funding does not. Field guides must compete in a commercial publishing environment, and seldom qualify for scientific grants. An exception of which I am aware is a proposal to fund guides in local languages. This is a gimmick, and serves only to encourage obscure and poorly produced booklets. In south-east Asia the educated minority who buy books and have disproportional political influence are also mostly highly competent in English and proud of it. Good field guides rely heavily on their illustrations, and subtleties of literary expression are not an issue. For them English has become the new Latin, a universal language of scholarship, and the greatest difficulty they experience in understanding English scientific books is the old Latin, represented abundantly in terminology and Linnean names.
CyanMagentaYellowBlack Odonata page 73
Odonata in Bornean tropical rain forest formations
73
ACKNOWLEDGEMENTS I would like to express my sincere thanks firstly to Adolfo Cordero, who invited me to write this chapter, and arranged funding which allowed me to attend the 4th WDA conference in Pontevedra, Spain, and participate in the ‘forests and dragonflies’ programme. Thanks are also due to my botanical mentors in Brunei, Peter Ashton, Timothy Whitmore, David Edwards, Webber Booth and Kamariah abu Salim. Jan van Tol assisted throughout with dragonfly identification and Rory Dow and Vincent Kalkman have recently provided interesting new data. I would also like to thank Rory Dow for his detailed reading of the text and thoughtful comments made in the light of his recent and extensive field work in Sarawak and Sabah.
REFERENCES ANDERSON, J.A.R. 1961. The ecology and forest types of the peat swamp forests of Sarawak and Brunei in relation to their silviculture. Ph.D. thesis, Edinburgh University. ANDERSON, J.A.R. 1964. The structure and development of the peat swamps of Sarawak and Brunei. Journal of Tropical Geography 18: 7-16. ANDERSON, J.A.R. & D. MARSDEN. 1984. Brunei Forest Resources and Strategic Planning Study. Unpublished report to the Government of His Majesty the Sultan and Yang Di-Pertuan of Negara Brunei Darussalam. ASHTON, P.S. 1964. Ecological Studies in the Mixed Dipterocarp Trees of Brunei State. Oxford Forestry Memoirs No 25. Clarendon Press, Oxford. COLLINS, N.M., J.A. SAYER, & T.C. WHITMORE (eds). 1991. The Conservation Atlas of Tropical Forests – Asia and the Pacific. IUCN/ Simon & Schuster, London. CORBET, P.S. 1999. Dragonflies. Behaviour and Ecology of Odonata. Harley Books, Colchester. CRANBROOK, LORD & D.S. EDWARDS. 1994. Belalong, A tropical rainforest. Suntree Publishers, Singapore. DAVIES, J. & A.S. KAMARIAH. 1999. The rain forests of Brunei. In: Wong, K.M. & A.S. Kamariah (eds), Forests and trees of Brunei Darussalam, pp. 15-34, Universiti Brunei Darussalam, Brunei. DOW, R.A. 2005. Odonata, burglary and ballistic cicadas in south-east Asia. Agrion 9: 10-12. EDWARDS, D.S., W.E. BOOTH & S.C. CHOY, 1996. Tropical rainforest research – current issues. Kluwer, Dordrecht. FINKE, O. M. 1992. Interspecific competition for treeholes: consequences for mating systems and coexistence in Neotropical damselflies. American Naturalist 139: 80-101.
CyanMagentaYellowBlack Odonata page 74
74 A. G. Orr
HÄMÄLÄINEN, M. 1994. Dragonflies of Mount Kinabalu (the highest mountain in Borneo). Malangpo 11: 77-81. LAIDLAW, F.F. 1934. A note on the dragonfly fauna (Odonata) of Mount Kinabalu and of some other mountain areas of Malaysia: with a description of some new or little known species. Journal of the Federated Malay States Museum 17: 549-561. LIEFTINCK, M.A. 1954. Handlist of Malaysian Odonata. A catalogue of the dragonflies of the Malay Peninsula, Sumatra, Java and Borneo, including the adjacent small islands. Treubia 22 (Suppl.). KALKMAN, V. 2005. some notes on dragonflies observed at the KBFSC, Brunei. Agrion 9: 13-14. KITCHING, R.L. & A.G. ORR. 1996.The food-webs from water-filled treeholes in Kuala Belalong, Brunei. Raffles Bulletin of Zoology 44: 405-413. MERRILL, E.D. 1921. A Bibliographic Enumeration of Bornean Plants. Journal of the Straits Branch of the Royal Asiatic Society, Special Number. 637 pp. MERRILL, E.D. 1921. A brief survey of the present status of Bornean botany. Webbia, 7: 309-324. MOORE, N.W. 1997. Dragonflies – status survey and conservation action plan. IUCN/SSC Odonata Specialist Group. IUCN, Gland and Cambridge. ORR, A.G. 1994. Life histories and ecology of Odonata breeding in phytotelmata in Bornean rain forest. Odonatologica 23: 365-377. ORR, A.G. 1997. Odonate predation in Bornean treehole communities: some observations on predator density and prey diversity. In: Ulrich, H. (ed.), Tropical Biodiversity and Systematics, Proceedings of the International Symposium on Biodiversity and Systematics in Tropical Ecosystems, Bonn, 27 May 1994, pp. 223-228, ZFMK, Bonn. ORR, A.G. 2001. An annotated checklist of the Odonata of Brunei with ecological notes and descriptions of hitherto unknown males and larvae. International Journal of Odonatology 4: 167-220. ORR, A.G. 2003. A Guide to the Dragonflies of Borneo – their identification and biology. Natural History Publications Borneo, Kota Kinabalu. ORR, A.G. 2004a. Critical species of Odonata in Malaysia, Indonesia, Singapore and Brunei. International Journal of Odonatology 7: 371-384. ORR, A.G. 2004b. Territorial behaviour associated with feeding in both sexes of the tropical zygopteran, Libellago hyalina (Odonata: Chlorocyphidae). International Journal of Odonatology 7: 493-504. PAULSEN, AD., I.C. NIELSEN, S. TAN & H. BASLEV. 1996. A quantitative inventory of trees in one hectare of mixed dipterocarp forest in Temburong, Brunei Darussalam. In: D.S. Edwards et al. (eds), Tropical Rainforest Research – Current Issues, pp. 139-150, Kluwer, Dordrecht. PROCTOR, J., Y.F. LEE, A.M. LANGLEY, CM. MUNRO & T. NELSON. 1988. Ecological studies on Gunung Silam, a small ultrabasic mountain in Sabah, Malaysia. I. Environment, forest structure and floristics. Journal of Ecology 76: 320-340. ROSS, S.M. AND A. DYKES, 1996. Soil conditions, erosion and nutrient loss on steep slopes under mixed dipterocarp forest in Brunei Drussalam. In: D.S. Edwards
CyanMagentaYellowBlack Odonata page 75
Odonata in Bornean tropical rain forest formations
et al. (eds), Tropical Rainforest Research – Current Issues, pp. 259-270, Kluwer, Dordrecht. THOMPSON, D.J. & J. VAN TOL. 1993. Damselfies and dragonflies from four forest types in Brunei. Brunei Museum Journal 8: 57-72. WHITMORE, T.C. 1984. Tropical Rain Forests of the Far East, (2nd edition). Clarendon Press, Oxford. WONG, K.M. 1999. The nature of the Brunei flora. In: Wong, K.M. & A.S. Kamariah (eds), Forests and trees of Brunei Darussalam, pp. 53-73, University Brunei Darussalam, Brunei. WONG, K.M. & A.S. KAMARIAH (eds). 1999. Forests and trees of Brunei Darussalam, Universiti Brunei Darussalam, Brunei.
75
CyanMagentaYellowBlack Odonata page 76
76 A. G. Orr
APPENDIX 1 List of forest and non-forest species recorded from Brunei with known habitats listed for each forest species: md, lowland mixed dipterocarp; fw, freshwater swamp forest; ps, peatswamp forest; kg, kerangas (heath forest); mg, mangrove; sd, secondary mixed dipterocarp; DD, data deficient. Data from Orr (2001) Records representing less than 5% of records in dominant habitat are in parentheses and are disregarded in analysis above.
Non Forest Lestes praemorsus Agriocnemis femina A. pygmaea Argiocnemis rubescens Mortonagrion falcatum Aciagrion borneense Ischnura senegalensis Xiphiagrion cyanomelas Ceriagrion cerinorubellum Pseudagrion lalakense P. microcephalum P. perfuscatum Anax guttatus A. panybeus Ictinogomphus decoratus Epophthalmia vittigera Macromia cincta (enters forest) Tetrathemis irregularis Brachydiplax chalybea Chalybeothemis fluviatilis Nannophya pygmaea Orthetrum chrysis (md, ps) O. glaucum O. sabina
O. testaceum P. starrei Diplacodes. trivialis Neurothemis fluctuans N. ramburii N. terminata Rhodothemis rufa Pseudothemis jorina Trithemis aurora Camacinia harterti Hydrobasileus croceus Pantala flavescens Rhyothemis aterrima (ps) R. obsolescens (ps) R. phyllis R. pygmaea (md, fw) R. triangularis Tholymis tillarga T. phaeoneura Tramea transmarina Zyxomma obtusum (kg) Z. petiolatum (fw, ps) Aethriamanta gracilis Urothemis signata
CyanMagentaYellowBlack Odonata page 77
Odonata in Bornean tropical rain forest formations
Devadatta podolestoides md (sd) Neurobasis longipes md Vestalis amabilis md, fw, ps, kg V. amaryllis md V. amnicola md V. amoena md V. atropha md V. beryllae md Heliocypha biseriata md Libellago aurantiaca md, fw, sd L. hyalina ps, fw, kg L. lineata ps, fw L. semiopaca md L. stictica md Rhinocypha aurofulgens md R. cucullata md R. humeralis md R. stygia md Sundacypha petiolata md S. striata fw Dysphaea dimidiata md D. lugens md Euphaea ameeka md, (fw), sd E. impar md E. subcostalis md E. tricolor md Orolestes wallacei fw Bornargiolestes nigra md ,DD Podolestes chrysopus ps P. harrissoni ps P. orientalis (md), fw, ps Rhinagrion borneense md Mortonagrion alcyone md DD Onychargia atrocyana md, fw Amphicmenis erminea fw, ps, (kg) A. martini ps A. remiger fw, ps, kg A wallacii fw, ps Archibasis incisura fw DD A. melanocyana fw, ps A. tenella fw A. viola fw Pericnemis triangularis md Stenagrion dubium md Teinobasis rajah md, sd
Forest Coeliccia borneensis md C. cyaneothorax md C. macrostigma fw, ps C. nigrohamata md Coeliccia sp. A fw, DD Coeliccia sp. B fw, DD Copera vittata fw, (md) Copera sp. fw, ps, (md) Drepanosticta attala md, DD D. forficula md D. rufostigma md D. versicolor md Drepanosticta sp. A md Drepanosticta sp. B md, DD Drepanosticta sp. C md, DD Protosticta feronia md DD Protosticta sp. A md Protosticta sp. B md, DD Elattoneura analis md, fw Prodasineura collaris md, fw P. dorsalis fw, ps P. haematosoma md, sd, fw P. hosei md, DD P. hyperythra md, sd, fw P. verticalis md, sd Prodasineura sp. A ps, DD Prodasineura sp. B md, fw, ps, sd Gynacantha basiguttata fw, mg, sd G. bayadera fw, sd G. dohrni fw, sd G. maclachlani kg, DD Heliaeschna crassa fw, ps, sd H. idae sd, DD H. simplicia sd, DD Indaeschna grubaueri md Tetracanthagyna degorsi md T. plagiata md, sd Oligoaeschna buehri fw, sd O. foliacea md, fw, sd O. platyura sd, DD Oligoaeschna sp. fw, DD Burmagomphus insularis md Heliogomphus blandulus md DD Leptogomphus coomansi md, fw L. mariae md, DD
77
CyanMagentaYellowBlack Odonata page 78
78 A. G. Orr
L. pendleburyi md, DD L pasia md, DD Macrogomphus parallelogrammus md M. quadratus fw, sd Microgomphus chelifer md Megalogomphus sumatranus md Sieboldius japponicus md Ictinogomphus acutus fw, ps, DD Chlorogomphus sp. md, DD Hemicordulia tenera fw, ps Metaphya micans ps, DD Idionyx yolanda md, fw DD Macromidia fulva md Macromia corycia, md M. westwoodii, md Hylaeothemis clementia fw, DD Risiophlebia dohrni fw, ps B. farinosa fw, DD Brachygonia oculata fw, ps, kg, sd B. ophelia ps, DD Nannophyopsis chalcosoma fw, DD
Raphismia bispina mg Tyriobapta kuekenthali fw T. laidlawi fw, ps T. torrida fw, (md), kg Agrionoptera insignis mg, sd A. sexlineata fw, kg Cratilla lineata fw, ps C. metallica md, (fw), kg Lyriothemis biappendiculata md, sd L. cleis md Nesoxenia lineata fw Orchithemis pruinans fw O. pulcherrima fw, ps (md) O. xanthosoma fw Orthetrum pruinosum md Pornothemis serrata fw, ps Pseudagrionoptera diotima fw, DD Onychothemis coccinea md O. culminicola md Zygonyx iris md
CyanMagentaYellowBlack Odonata page 79
The Importance Forests to Neotropical Dragonflies Adolfo Cordero Rivera (ed)of2006 Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 79-101.
79
© Pensoft Publishers
Sofia–Moscow
The Importance of Forests to Neotropical Dragonflies Dennis Paulson University of Puget Sound, Tacoma, Washington, USA
ABSTRACT Dragonflies are quintessential forest animals, and forests are essential to them. The majority of odonate species are associated with forests, especially in the neotropical region. Forests are important in furnishing a variety of larval habitats and favorable conditions for adults. Adult odonates can use both sunshine and shade available in forests, but forests also offer constraints to odonate activity. Forest odonates are poorer dispersers than those of open country, this factor contributing to the very high biodiversity of the tropics.
This paper is presented as an overview of the subject matter. Although an outcome of the author’s long interest in neotropical dragonflies, it is not based on a single research track. Many of the data presented are unpublished and are indicated as such. “Dragonflies” and “odonates” are used interchangeably throughout.
DRAGONFLIES ARE FOREST ANIMALS Forested landscapes covered about half of prehistoric Earth’s subarctic land surface (Global Forest Watch 2005). Subarctic lands that were not covered by forests included deserts, savannahs, grasslands, and low-latitude alpine and subalpine zones, many of which were too dry and/or cold to support populations of Odonata. Forest cover is a consequence of adequate rainfall, which also produces abundant wetlands. Thus it is likely that the existing Odonata fauna would be highly adapted to forested landscapes.
CyanMagentaYellowBlack Odonata page 80
80 Dennis Paulson
Forest Odonata are defined here as species that are normally found in forested habitats, with the assumption that forest, or at least a band of riparian vegetation, is essential to the survival of adults or larvae or both. The literature is surprisingly poor in categorizing terrestrial habitats used by adult Odonata, so I have relied largely on my own field experience in classifying genera and species as forest-based or not. All extant families of Odonata contain forest species, with the exception of Hemiphlebiidae, represented by a single Australian species of open marshes. Within each family, basal groups (e.g., Gomphaeschna and Boyeria in the Aeshnidae, Tetrathemistinae in the Libellulidae, and Argia in the Coenagrionidae) all appear to be forest-dwellers, typically associated with streams, while derived genera (e.g., Anax and Aeshna in the Aeshnidae, Pantala and Tramea in the Libellulidae, and Ischnura in the Coenagrionidae) occur in open habitats, typically associated with ponds (phylogenies from Fraser 1957, von Ellenrieder 2002, Rehn 2003). It is apparent that life in forests is typical of the Odonata, life in open habitats a more recent adaptation. Of 217 odonate genera in the New World, 175 (80.7%) include forest species. Of 164 neotropical genera, 144 (87.8%) include forest species. Of 71 nearctic genera, 39 (54.9%) include forest species. I cannot assign all New World species to forest or nonforest habitats, but I am confident that the majority of the species in genera that include forest species are in fact forest-based species.
Alaska
31
REGION
Washington
80
Iowa
108
Florida
166
Veracruz
208
Costa Rica
271 0
20 40 60 80 PERCENT SPECIES RESTRICTED TO FOREST
100
Fig. 1. Proportion of Odonata species restricted to forest habitats in six New World regions (total number of species inside bars). Note that the values increase with decreasing latitude.
CyanMagentaYellowBlack Odonata page 81
The Importance of Forests to Neotropical Dragonflies
81
MANY SPECIES ARE RESTRICTED TO FOREST, ESPECIALLY NEOTROPICAL FORESTS Figure 1 indicates regions at different latitudes and the proportion of odonate species in them that are restricted to forested habitats. The regions were chosen to cover a wide range of latitudes and because their odonate fauna was sufficiently familiar to me that I was able to categorize species. There is a distinct latitudinal gradient, as forests become ever more important at lower latitudes. It is apparent that the proportion of forest species also increases with an increase in species diversity. Complete lists of odonate species are available for relatively few neotropical sites (Table 1), but they show that forest species dominate or are at least well-represented at all sites, whether they are entirely or partially forested. With five weeks of field work and the notes on many specimens collected by others, I attempted to categorize the odonate species of the Tambopata Reserve, Madre de Dios, Peru, by habitats (Figure 2). I made such designations for 146 of the 188 species (78%). Although such efforts are fraught with difficulty for the poorly known tropical fauna, the data, as best as they can be construed, show the 80
NUMBER OF SPECIES
70 60 50 unique shared
40 30 20 10 0 cochas
swamps
stream
forest
MAJOR NABITATS
Fig. 2. Total number of Odonata species from Tambopata Reserve, Madre de Dios, Peru, found in each habitat divided into those unique to it and those shared with one or more other habitats. Many species are restricted to single major habitats. Cochas are oxbow lakes, with open beds of grass at shore but forested just back from that; swamps are wooded lentic wetlands; streams are forested; and forest consists of forested areas away from water.
Tuxtlas, Mexico Taboga, Costa Rica La Selva, Costa Rica Barro Colorado Island, Panama Rincón de Osa, Costa Rica Limoncocha, Ecuador Tambopata, Peru
LOCALITY 18.5° N 10.4° N 10.4° N 9.2° N 8.7° N 0.4° S 12.8° S
134 99 93 91 93 148 177
LATITUDE SPECIES 52 43 82 64 72 82 84
% FORESTBASED
much open habitat much open habitat all forested surrounded by open lake scattered open habitat all forested all forested
HABITATS
González Soriano 1997 D. Paulson unpubl. D. Paulson unpubl. May 1979 D. Paulson unpubl. D. Paulson unpubl. D. Paulson unpubl.
REFERENCE
Table 1. Neotropical localities adequately surveyed and with a well-enough known fauna to categorize the Odonata species. From about half to about four-fifths of the species at a neotropical locality are forest-based.
CyanMagentaYellowBlack Odonata page 82
82 Dennis Paulson
CyanMagentaYellowBlack Odonata page 83
The Importance of Forests to Neotropical Dragonflies
83
importance of forest at a single site. Some forest species are known to breed either in cochas, swamps, or streams, the source of the substantial number of shared habitats, while the exact breeding habitat of others is not known.
WHAT MAKES TROPICAL FORESTS SPECIAL? Tropical forests are well-known for their biological complexity, with a tremendous diversity of species and a bewildering variety of biological interactions (Kricher 1997). The great number of species of animals includes many that may represent prey of, predators on, or competitors with Odonata. The physical complexity of these forests, however, may be more significant in their importance to dragonflies. Tropical forests feature among the tallest trees in the world, and their size promotes vertical stratification, with many physical attributes varying sharply between the top of the canopy and the forest floor. The most important physical attributes are light, temperature, humidity, and wind, and these in turn affect biological attributes such as species diversity, species presence, and abundance of individuals. Light falling on the canopy of a typical tropical wet forest at Barro Colorado Island, Panama, was reduced to around 25% just below the canopy and 1% just above the forest floor (Richards 1966). Daily temperature and humidity ranges fluctuate much less just above the forest floor than in the canopy, and wind speed may be reduced to almost nothing, becoming undetectable (Richards 1966).
IMPORTANCE OF TROPICAL FORESTS TO DRAGONFLIES The importance of forests to Odonata is based on a great variety of factors. Forests may be important as breeding sites, thus to the larval stage of the life history, or they may be important only to adults. Probably for most forestbased species, the forest environment is important to both larvae and adults.
IMPORTANCE TO LARVAE For breeding sites, neotropical forests provide forested wetlands, temporary wetlands, phytotelmata, and ground litter. The first two habitats are shared with the temperate zone, but the second two are used by odonates only in the tropics. Forested wetlands The great majority of odonate species are tied to fresh water as the breeding habitat in which the larvae develop, and freshwater wetlands are
CyanMagentaYellowBlack Odonata page 84
84 Dennis Paulson
very varied. Common types of wetlands may be running or still, shallow or deep, small or large, acid or alkaline, and vegetated or open, but any type of wetland can be entirely within or at the edge of forest. Thus many wetlands will be forested wetlands. Because forests are extensive, it is reasonable to assume that many odonate species would inhabit forested wetlands, and that is in fact the case. The majority of the species considered forest-based breed in these forested wetlands, both streams and ponds, and may spend their lives associated with them. As adult Odonata are ectotherms, their body temperature regulated by ambient temperature (May 1976), they are dependent on relatively high air temperatures for their activity. Air temperatures in the shade are often too low for odonate activity, and this becomes increasingly the case at higher latitudes, so most species are active in the sun (Corbet 1999). Thus, forested wetlands should not be favored breeding places at high latitudes. As an example, of the 80 species in Washington state, USA, only three breed in forested streams, and none breeds in swamps (unpubl. data). At lower latitudes, where air temperature is relatively high in the shade as well as in the sun, a greater proportion of species is able to be active in the shade, including males maintaining territories and searching for females, both sexes mating, and females ovipositing. Some anisopterans are functional endotherms, able to elevate body temperature by wing-whirring and flight (May 1976), but zygopterans are not thought to perform this behavior and are thus more dependent on air temperature. It is of interest, therefore, that this suborder is the one that shows a very strong latitudinal diversity gradient (Table 2). Not only are there many more species of wide-ranging zygopteran families in tropical latitudes, but seven families are restricted to the tropics. Note that, conversely, five anisopteran families are restricted to temperate latitudes, so the two suborders exhibit somewhat different patterns. However, overall diversity (total species) is higher in the tropics, and Zygoptera diversity is even more tied to the tropics. Eighty-six per cent of New World Zygoptera species are tropical, while 70% of Anisoptera are so distributed. Temporary wetlands Some wetlands are temporary, present relatively briefly during and after rains or lasting a substantial part of the year during an annual rainy season. Temporary wetlands can form anywhere it rains, but they are more likely to be a feature of tropical forests for two reasons. First, precipitation tends to be heavier in tropical latitudes (Walter 1973), and second, the relatively stable conditions of the forest floor (less extreme high temperatures, lower saturation deficit, much less wind) must reduce evaporation, leading to longer persistence of temporary waters. This in turn makes them of much greater potential as odonate breeding habitats. Although the details are poorly known,
CyanMagentaYellowBlack Odonata page 85
The Importance of Forests to Neotropical Dragonflies
85
Table 2. Number of Odonata species recorded from tropical and temperate latitudes in the New World. Some species occur in both zones. Most New World species are distributed in the tropics, but Zygoptera are even more characteristic of the tropics than are Anisoptera. Only in the Zygoptera are entire families restricted to low latitudes. Note, however, that in the Anisoptera, there are more temperate families. TROPICAL SPECIES
TEMPERATE SPECIES
Polythoridae Calopterygidae Dicteriadidae Amphipterygidae Lestidae Synlestidae Perilestidae Megapodagrionidae Platystictidae Coenagrionidae Pseudostigmatidae Protoneuridae TOTAL ZYGOPTERA
58 60 2 3 39 1 19 127 42 321 18 92 782
0 7 0 0 14 0 0 0 1 101 0 1 124
Petaluridae Austropetaliidae Aeshnidae Gomphidae Neopetaliidae Cordulegastridae Macromiidae Corduliidae Libellulidae TOTAL ANISOPTERA
0 0 100 258 0 0 0 32 327 717
3 5 49 98 1 10 9 55 79 309
TOTAL SPECIES
1499
533
species of at least 26 genera of four families are known or strongly suspected of breeding in temporary wetlands in the New World tropics (Table 3). Phytotelmata Phytotelmata, or small “containers” of water associated with plants (leaf axils, tree holes, usually above the forest floor), are characteristic of tropical wet forests and tropical dry forests with sufficiently long rainy seasons. The reduced evaporation within forests promotes retention of water in these small volumes, just as it does on the forest floor. Some of these phytotelmata will contain no water in the dry season, but I am not including them in the “temporary wetlands” classification.
CyanMagentaYellowBlack Odonata page 86
86 Dennis Paulson
Table 3. Species of at least 26 genera of four families are known or strongly suspected to breed in temporary wetlands in the Neotropics (unpubl. data). FAMILY
GENUS
LESTIDAE COENAGRIONIDAE
Lestes Calvertagrion Leptobasis Metaleptobasis Nehalennia Neoerythromma Telebasis Anax Coryphaeschna Gynacantha Triacanthagyna Anatya Brachymesia Cannaphila Erythemis Erythrodiplax Miathyria Micrathyria Orthemis Pantala Perithemis Rhodopygia Tauriphila Tholymis Tramea Uracis
AESHNIDAE
LIBELLULIDAE
Twenty-three genera of Odonata are known or strongly suspected to breed in phytotelmata, many of them restricted to this habitat (Corbet 1999), and 10 of these genera are neotropical. Not all species of some of the genera are phytotelmata breeders, but all of them are associated with forests. Phytotelmata furnish two advantages as breeding habitats for Odonata: (1) absence of predation by fish and many other freshwater predators; (2) occurrence throughout forest, not dependent on topography that holds water. Ground litter For the most part, the same factors that promote breeding in phytotelmata promote terrestrial breeding in Odonata. Nine genera of seven families of Odonata are known or strongly suspected to have terrestrial larvae (Corbet 1999), so it is quite interesting that no neotropical genus is suspected of ground
CyanMagentaYellowBlack Odonata page 87
The Importance of Forests to Neotropical Dragonflies
87
breeding. I propose that this is because of the abundance of ants in the New World Tropics (Forsythe and Miyata 1984). Regions that support terrestrial larvae include islands lacking in ants (Hawaii, Wilson and Taylor 1967, although not any more, Roderick and Gillespie 1998) and those in which ants perhaps are not as abundant as they are in mainland tropics (New Caledonia). Australian rain forests support several species of Odonata with terrestrial larvae, and ants are less diverse in its rain forests than in the more open, drier environments that make up most of Australia (Shattuck and Barnett 2001).
IMPORTANCE TO ADULTS Not only do forests provide aquatic breeding habitats for odonates, but they are especially suitable for the persistence of the adults. They provide shade, high humidity, shelter against both physical and biotic factors, daily and seasonal retreat, and mating and oviposition sites. Shade and humidity The problems of desiccation and overheating, both potential stresses on adult odonates, are reduced by the relatively high humidity and shade and reduced evaporation potential of the tropical forest understory. There is some evidence that tropical Odonata thermoregulate less effectively than temperate ones (May 1976). Many forest species are small and delicate, in particular a large number of slender-bodied zygopterans in the families Platystictidae, Coenagrionidae, and Protoneuridae. Temperate-zone Coenagrionidae include species as small in total length and weight, but none as slender as those at tropical latitudes. It is possible that their restriction to the tropical-forest understory is because their slender abdomens, and thus high surface-to-volume ratio, make them more subject to either overheating or desiccation. Many Odonata are thought to be associated with the sun or the shade but not necessarily both, so this is an important dichotomy to odonates (Osborn & Samways 1996). Shelter The dense foliage of a tropical forest provides shelter against both physical and biological threats. Physical threats include wind and rain, which are usually much more extreme outside the forest and above its canopy. The fact that storms regularly uproot large forest trees (Whitmore 1998) makes it clear that odonates in the open would be under some threat from these storms. The argument made above that small and delicate species may survive best in the tropical-forest understory is significant here as well. Biological threats include dragonfly-eating birds, which are diverse and abundant in open and semiopen areas in the neotropical region. Certain small
CyanMagentaYellowBlack Odonata page 88
88 Dennis Paulson
falcons (Falconidae) and large flycatchers (Tyrannidae), if not specialists on Odonata, at least take them in good numbers (unpubl. data). A dragonfly would surely have a greater chance of escaping such aerial insectivores in the dense vegetation of the forest understory than in the open air over a pond or clearing. The delicate zygopterans that are a prominent feature of these forests might be especially vulnerable to birds as well as climatic factors. Bird predation is, of course, not absent within the forest, which is full of puffbirds (Bucconidae), trogons (Trogonidae), motmots (Momotidae), and others that take large aerial insects. In fact, jacamars (Galbulidae) are a forest group that specializes on large flying insects such as butterflies and dragonflies, but in my experience, they are not as common within the forest as flycatchers are at its edges. Retreat Corbet (1999) mentions the importance of forests as both daily and seasonal retreats for odonates (see also Corbet, this volume). Dusk-flying aeshnids (Gynacantha, Neuraeschna, Triacanthagyna) and libellulids (Tholymis) retire to the forest during the day, but they also breed within the forest. On the other hand, a large number of neotropical genera that appear to be forest-based have their sexual rendezvous in the open, at lentic or lotic habitats, much like Sympetrum infuscatum in paddy fields in Japan (Watanabe et al. 2004). Table 4 lists neotropical genera that include species that occur in open areas for mating but retire to the forest to roost and feed. Forests may also be important as seasonal retreats. Some species of Lestes, Sympetrum, and Aeshna are known to retreat soon after emergence to forests, usually at higher elevation, during the warm, dry summer in a few temperatezone localities (Japan, Miyakawa 1994; Algeria, Samraoui et al 1998). They then emerge from the forests in autumn to return to breeding habitats. Less is known about the movements of tropical species, but there is some evidence that forests play a significant role in the annual cycle of many species. In Venezuela, some savannah species moved into gallery forest during the extended dry season (De Marmels 1998, 1990). Similarly, certain open-country species moved into rain forest during the dry season in Panama (Morton 1977). I observed a spectacular movement of anisopterans from dry forests to recently flooded marshes at the beginning of the rains in June 1967 in Guanacaste Province, Costa Rica (unpubl. data). Mating and oviposition sites Some odonates mate away from the wetlands where they deposit their eggs (Corbet 1999), and forested environments may be important as rendezvous sites, perhaps because both sexes feed there. For example, males of both temperate and tropical Macromia are seen patrolling over water, yet mating often appears to take place away from it (Cordero Rivera et al 1999). In the
CyanMagentaYellowBlack Odonata page 89
The Importance of Forests to Neotropical Dragonflies
89
Table 4. Species of at least 35 neotropical genera of six families occur in open areas for mating but retire to the forest to roost and feed (unpubl. data). FAMILY
GENUS
CALOPTERYGIDAE LESTIDAE PROTONEURIDAE
Hetaerina Lestes Epipleoneura Neoneura Protoneura Acanthagrion Aeolagrion Anisagrion Argia Leptobasis Telebasis Coryphaeschna Gynacantha Remartinia Rhionaeschna Staurophlebia Aphylla Phyllocycla Phyllogomphoides Progomphus Brachymesia Brechmorhoga Cannaphila Diastatops Dythemis Erythemis Erythrodiplax Macrothemis Nephepeltia Oligoclada Orthemis Planiplax Rhodopygia Tholymis Zenithoptera
COENAGRIONIDAE
AESHNIDAE
GOMPHIDAE
LIBELLULIDAE
North American aeshnids Coryphaeschna ingens, Epiaeschna heros, and Gynacantha nervosa, males encounter females in woodland rather than at the water, and females later go to water to oviposit (Dunkle 1989). Probably most neotropical Gynacantha species meet to mate in forest away from water (unpubl. data).
CyanMagentaYellowBlack Odonata page 90
90 Dennis Paulson
A small number of endophytically ovipositing tropical odonate species, in both Zygoptera (e.g., Mecistogaster and Megaloprepus, Fincke 1992; Cora, Pritchard 1996) and Anisoptera (e.g., Gynacantha, Fincke 1992; Tetracanthagyna, Orr 2003), oviposit primarily in woody tissue above the water, and forest cover should be essential to such species. A female Micrathyria dictynna oviposited epiphytically on the underside of the tip of a palm leaf well above a forest stream (Förster 1998). In this large genus, M. dictynna is one of the forestbased species, perhaps because of this unusual oviposition. Unfortunately, the oviposition habits have been described for extremely few neotropical species.
CONSTRAINTS IMPOSED BY FOREST LIFE Forests are not all good for Odonata; afforestation of open streams by invasive tree species has affected odonate species negatively in South Africa (Kinvig & Samways 2000). Even within established forests, some of the same factors that make this environment favorable for odonates may have aspects that reduce their fitness. There is presumably a high diversity of dragonfly predators and competitors within the neotropical wet forest, the most biodiverse habitat and region on Earth. There should be a high diversity of prey species as well, in all these cases more diversity than in nearby open areas. However, nothing is known about how these differences would affect adult odonates. The constraints presented by the physical environment are more obvious. The most obvious constraint is the much lowered light levels within the forest, even near the canopy. Light levels just below the canopy drop rapidly below the canopy, falling to 1% or less of ambient by the time the light reaches the forest floor (Richards 1966). Blocking of the sun’s rays keeps the temperature lower within the forest, and this could have an effect on dragonfly activity, especially early in the day or during cooler periods at the edge of the tropics. Sunflecks are present only during midday. The reduction in temperature and absence of direct sunlight should also contribute to a reduction in prey activity, as many other insects are as dependent on sunlight for warmth as dragonflies are. Finally, vision is the most important sense of adult odonates, and the absence of the sun might contribute to a reduction in visual acuity.
TROPICAL HELIOPHILY Most odonates need to be warm for activity; thus those that live in forested environments may use whatever sunlight is available to elevate their body temperatures. Many forest species seek out sunflecks in order to regu-
CyanMagentaYellowBlack Odonata page 91
The Importance of Forests to Neotropical Dragonflies
91
late their temperature at optimal levels for flight, predator avoidance, foraging, and reproductive behavior. Sunflecks are moving targets as the Earth rotates, but because odonates fly and are thus highly mobile, they can move with them. They thus have a choice of micro-environment—sunlight or shade, the best of both worlds—not available to open-country species. They are like lizards in using this ability for behavioral thermoregulation (Pianka and Vitt 2003) but are even more mobile than lizards. In warm-temperate woodlands in Japan, Lestes sponsa used sunflecks to maintain an elevated body temperature (Watanabe & Taguchi 1993). Platycnemis echigoana did the same and defended the sunflecks for short time intervals against conspecifics; some pairs also rendezvoused for mating at them (Watanabe et al. 1987). In Brunei, Libellago hyalina used sunflecks as feeding territories and defended them as well (Orr 2004). As foraging was often done from shaded perches, Orr thought the sunflecks were attractive because flying prey in the sun was more visible than that in the shade. Notiothemis robertsi would alternate territorial presence at shaded forest pools with “sun flights” into the canopy, presumably to raise body temperature (Clausnitzer 1996). Although I cannot present a complete list, some species encountered on forest trails at Explorer’s Inn, Tambopata Reserve, Madre de Dios, Peru, were consistently seen in sunflecks, from which and in which they presumably foraged. I learned to walk quickly from sunfleck to sunfleck to find greater numbers of odonates. Only one of these species, Uracis siemensi, was thought to be using the sunflecks as mating rendezvous sites in the dry forest in the dry season when the study was conducted. Shelly (1982) studied two zygopterans on Barro Colorado Island, Panama, that differed in their choice of habitat. Argia difficilis usually foraged in the sun, had thoracic temperatures elevated 4-8° C. above ambient, and foraged five times more frequently than Heteragrion erythrogastrum, which foraged in the shade. A. difficilis also made longer foraging flights. Clearly either of these modes is an appropriate one for forest-dwellers, and these two species are representative of the different modes that characterize their large genera (unpubl. data).
TROPICAL SCIOPHILY In contrast to the sun lovers, many species of numerous genera are active in shade. These include the dusk-flying aeshnids (Gynacantha, Neuraeschna, Triacanthagyna) and libellulids (Tholymis) but also a wide variety of genera in many families that appear to be quite independent of sunlight for their daily activities (Table 5, which lists known but surely not all examples). Some
CyanMagentaYellowBlack Odonata page 92
92 Dennis Paulson
Table 5. Species of at least 24 neotropical genera of 12 families are considered shadeadapted, not needing sunlight for their daily activities (unpubl. data). FAMILY
GENUS
Polythoridae Amphipterygidae Calopterygidae Lestidae Perilestidae Megapodagrionidae
Polythore Amphipteryx Mnesarete Archilestes Perissolestes Heteragrion Oxystigma Paraphlebia Palaemnema Forcepsioneura Psaironeura Leptobasis Metaleptobasis Gynacantha Neuraeschna Triacanthagyna Agriogomphus Anatya Cannaphila Micrathyria Oligoclada Perithemis Tholymis Uracis
Platystictidae Protoneuridae Coenagrionidae Aeshnidae
Gomphidae Libellulidae
genera are so thoroughly adapted to tropical rain forest that they are active only during rain (Corbet 1999)! Neotropical examples of “rainflies” include two species of Oxystigma (Megapodagrionidae) and Acanthagrion egleri (Coenagrionidae).
DISPERSAL ABILITY AND DISTRIBUTION OF FOREST ODONATES In general, forest species appear to have more restricted distributions than those of open habitats (Figure 3a). Orr (2001) reached the same conclusion about the Odonata of Brunei. The Zygoptera typically have the smallest ranges (Figure 3b), the Libellulidae the largest (Figure 3c). Some forest genera have particularly small ranges and are apparently among the poorest at dispersing. Of the 43 species of Palaemnema in the New World Tropics, 30
CyanMagentaYellowBlack Odonata page 93
The Importance of Forests to Neotropical Dragonflies
93
50
NUMBER OF SPECIES
45 40 35 30 nonforest species forest species
25 20 15 10 5 0
1
2 3 4 5 6 7 EXTENT OF DISTRINUTION
8
Fig. 3a. Breadth of geographic distribution of the Odonata species occurring in Costa Rica. Each species was given a number as an indicator of the breadth of its distribution. All species were assigned ‘1’ for their occurrence in Costa Rica, then numbers were added to this based on both their northern and southern distributional extent. A ‘1’ was added for Nicaragua to the north and Panama to the south, so a species occurring from Nicaragua to Panama would receive a ‘3’ score. A ‘2’ was added for Guatemala or Mexico to the north and Ecuador or Venezuela to the south, a ‘3’ for the United States to the north and Brazil to the south, a ‘4’ for Argentina to the south. The maximum score would thus be 8. Note that forest odonates are less widely distributed than those of open country.
NUMBER OF SPECIES
30 25 20 nonforest species forest species
15 10 5 0
1
2 3 4 5 6 7 EXTENT OF DISTRINUTION
8
Fig. 3b. Breadth of geographic distribution of the Zygoptera species occurring in Costa Rica. Note that zygopterans are less widely distributed than odonates on the average, and the widely distributed species are almost all nonforest species. Numbers as in Figure 3a.
CyanMagentaYellowBlack Odonata page 94
94 Dennis Paulson
NUMBER OF SPECIES
25
10
20 nonforest species forest species 15
5
0
1
2 3 4 5 6 7 EXTENT OF DISTRINUTION
8
Fig. 3c. Breadth of geographic distribution of the species of Libellulidae occurring in Costa Rica. note that libellulids are the most widely distributed odonates, and even forest libellulids are relatively widely distributed, but the most widely distributed libellulids also tend to be nonforest species. Numbers as in Figure 3a.
(70%) are known from only one country (unpubl. data). It would be interesting to know why, then, P. nathalia is distributed from southern Mexico to Panama. It may be adapted to more open habitats. The similarly wide-ranging P. domina, which occurs from Arizona to southern Mexico, is found on wooded streams in desert habitats (Hoekstra and Garrison 1999), perhaps indicating the ability to disperse across open country.
THE CONSEQUENCES TO ODONATA OF FOREST DESTRUCTION To date, no forested neotropical site that has been cleared has had a survey of Odonata before and after; in fact, no such site that has had an Odonata survey has been cleared. This would be considered good news if it weren’t for the fact that such a tiny handful of sites have been surveyed. All of the neotropical species listed by Paulson (2004) as of priority for conservation efforts because of their limited range or taxonomic uniqueness are forest species. Forest fragmentation, the most common method of habitat loss in the New World tropics, has been found to have a negative effect on species diversity and complete extirpation of populations in neotropical mammals (Schwarzkopf and Rylands 1989, Bentley et al. 2000), birds (Robinson 1999, Stratford and Stouffer 1999), and beetles (Didham et al. 1998). However, butterfly,
CyanMagentaYellowBlack Odonata page 95
The Importance of Forests to Neotropical Dragonflies
95
frog, and small mammal diversity at study sites in the Brazilian Amazon basin were usually enhanced by fragmentation (Brown and Hutchings 1997, Tocher et al. 1997, Malcolm 1997), because of the added diversity of edge species, and representative assemblages of forest-understory species were present in very small fragments. Road clearings prevent many forest-interior vertebrates from dispersing across them (Goosem 1997, Laurance et al. 2004), and it may be that small, forest-interior odonates are similarly inhibited. In fact, wide gaps such as clearings, rivers, and roads may prove effective barriers to some forest odonates, based on research by Ola Fincke (pers. comm.). In Southeast Asia, Singapore has had 95% of its forest cleared, and it has lost at least 32% of its land birds (Castelletta et al. 2000); almost all of the extirpated species were forest-dependent. That 5% of the forests still hold 68% of the birds is good news for odonates, which are smaller and may persist in smaller forest patches. Nevertheless, the logical end point in forest fragmentation is total clearing, which will eliminate all forest odonates.
FURTHER RESEARCH Paulson (2004: 177) wrote: “Although without quantitative data, it is common knowledge among those who study tropical odonates that the upland vegetation is just as crucial as the aquatic situation.” Tropical forest as upland vegetation must be crucial for forest-based species, but there are still no hard data to support this. Knowledge must replace speculation. For example, why exactly do so many species need forest? Why are the forest-based species apparently unable to succeed in open country (at least they are never found there)? What happens when you take the forest away or alter it by logging? In the north temperate zone, wetlands associated with undisturbed forest habitats supported more species of odonates than wetlands associated with logged forests (Rith-Najarian 1998, Sahlén 1999), and the same was true for wetlands at a site in northern Argentina (von Ellenrieder 2000), but the extent of our knowledge is based on very few studies. Studies of experimental forest fragmentation are available for neotropical birds, other vertebrates, and butterflies (Laurance and Bierregard 1997, Sekercioglu et al. 2002) but not for dragonflies. It is fortunate that none of the few well-studied neotropical sites has been deforested, but odonate researchers should take advantage of fragmentation experiments that are carried out to learn about other organisms. The sampling methods that have been used for insects in such studies (e.g., Sekercioglu et al. 2002) do not even record Odonata. Are there actual differences between forest and open dragonflies, for example in cuticle thickness (thinner cuticle within the protected forest environment) or optimal body temperature (lower in the shaded forest environment)?
CyanMagentaYellowBlack Odonata page 96
96 Dennis Paulson
Are forest dragonflies likely to be K-selected, open-country dragonflies r-selected? Do the former lead longer lives? The record longevity for an odonate so far (Corbet 1999) is held by Megaloprepus caerulatus (Fincke 1984), a neotropical forest species. No data are available to compare clutch size in tropical forest and open dragonflies; in fact, there appears to be clutch-size information for only one neotropical odonate, again Megaloprepus caerulatus (Fincke and Hadrys 2001). The clutch size of a few hundred eggs in that species is not appreciably different from those of temperate-zone coenagrionids (Corbet 1999). There appear to be color differences in forest and open odonates, with many forest species being dull-colored, presumably for camouflage while roosting. De Marco (1998) noted that a large proportion of anisopterans feeding within a low forest in Brazil were well-camouflaged females. Species of Acanthagrion at the Tambopata Reserve, Peru, fall into two color modes: males of three species that live in forest swamps (A. apicale, A. obsoletum, A. n. sp. A) have an orange thorax, and males of five species that live in open oxbow lakes (A. aepiolum, A. gracile, A. minutum, A. viridescens, A. n. sp. B) have a blue thorax (unpubl. data). Based on morphology, these color groups do not correspond to clades, and a molecular phylogeny of the genus would be of interest to test the hypothesis that lighting influences color. One blue species, A. phallicorne, lives in swamps, but it appears to seek out sunflecks so is lighted like the lake species. One of the major questions to be asked about tropical-forest Odonata is how they are using the forest canopy. Thus far no one has systematically watched, or watched for, odonates from one of the many places where canopy access has been made possible. True forest odonates may or may not ascend to the sunlit heights of the upper canopy, but we should attempt to determine whether they do. It would be of equal interest to know if open-country Odonata were present up there. Rehfeldt (1986) reported a flight of several species of libellulids from the understory into the canopy at midday in Panama. My only contribution to the resolution of these questions is the observation in Peru and Venezuela of numerous libellulids of several genera (Diastatops, Erythrodiplax, Micrathyria, Zenithoptera) roosting on twigs up to at least halfway to the canopy, apparently foraging in the more open spaces above the understory layer.
FORESTS AND ODONATE BIODIVERSITY Forest species tend to have smaller ranges than those of open country, and this indicates they have a lower propensity and/or ability to disperse. Thus they are especially threatened by forest fragmentation. The low dispersal potential of forest species in many genera has presumably promoted speciation and thus high overall biodiversity in tropical forest regions. The most speciose neotropical genera – Argia (112 species), Progom-
CyanMagentaYellowBlack Odonata page 97
The Importance of Forests to Neotropical Dragonflies
97
phus (67), Erythrodiplax (58), Phyllogomphoides (46), Micrathyria (46), Heteragrion (42), Palaemnema (42), and Acanthagrion (40) – all include species that are very restricted in distribution. Forest species are often quite rare and poorly known, and a second locality recorded for a species may be very far from the first (Paulson 2004), indicating they are more widespread than was first believed. The apparent rarity may have several explanations: (1) some species may be extreme habitat specialists and will only be found where their specialized habitat occurs; (2) some species may be active only at dusk, or only during rains, and are therefore not often encountered even where common; (3) some species may spend much of their time in the forest canopy or feeding well above the ground, as is known for neotropical aeshnids (Paulson 1994), gomphids (Belle 1966a, 1966b), and libellulids (Belle 1984).
NEOTROPICAL DRAGONFLIES MUST HAVE FORESTS From the number of species that are not usually found away from it, forest cover must be important for adults of the majority of neotropical Odonata. Every effort should be made to preserve forests and forest wetlands in this most biodiverse region of the Earth.
ACKNOWLEDGEMENTS I thank Adolfo Cordero Rivera for organizing the forest Odonata symposium in Pontevedra, Spain, at which this paper was presented. I also thank Max Günther and the staff of Explorer’s Inn for their assistance and courtesy while I studied the Odonata of that rich and beautiful spot, and I thank Netta Smith especially for accompanying me into the forest.
REFERENCES BELLE, J. 1966a. Surinam dragon-flies of the genus Progomphus. Studies on the Fauna of Suriname and other Guyanas 8: 1-28. BELLE, J. 1966b. Surinam dragon-flies of the Agriogomphus complex of genera. Studies on the Fauna of Suriname and other Guyanas 8: 29-60. BELLE, J. 1984. Nothodiplax dendrophila, a new genus and a new species from Surinam (Odonata: Libellulidae). Entomologische Berichten 44: 6-8. BENTLEY, J. M., C. P. CATTERALL & G. C. SMITH. 2000. Effects of fragmentation of Araucarian vine forest on small mammal communities. Conservation Biology 14: 1075-1087.
CyanMagentaYellowBlack Odonata page 98
98 Dennis Paulson
BROWN, K. S., JR. & R. W. HUTCHINGS. 1997. Disturbance, fragmentation, and the dynamics of diversity in Amazonian forest butterflies. In: W. F. Laurance and R. O. Bierregard, Jr. (eds), Tropical forest remnants, ecology, management, and conservation of fragmented communities, pp. 91-110, University of Chicago Press, Chicago. CASTELLETTA, M., N. S. SODHI & R. SUBARAJ. 2000. Heavy extinctions of forest avifauna in Singapore: lessons for biodiversity conservation in Southeast Asia. Conservation Biology 14: 1870-1880. CLAUSNITZER, V. 1996. Territoriality in Notiothemis robertsi Fraser (Anisoptera: Libellulidae). Odonatologica 25(4): 335-345. CORBET, P. S. 1999. Dragonflies: behavior and ecology of Odonata. Cornell University Press, Ithaca, NY. CORDERO RIVERA, A., C. UTZERI & S. SANTOLAMAZZA CARBONE. 1999. Emergence and adult behaviour of Macromia splendens (Pictet) in Galicia, northwestern Spain (Anisoptera: Corduliidae). Odonatologica 28: 333-342. DE MARCO, P. The Amazonian Campina dragonfly assemblage: patterns in microhabitat use and behaviour in a foraging habitat (Anisoptera). Odonatologica 27: 239-248. DE MARMELS, J. 1988. A five year survey of an odonate community inhabiting a north Venezuelan mountain stream. Odonatologica 27: 189-199. DE MARMELS, J. 1990. Nota sobre dos “formas” en Acanthagrion fluviatile (De Marmels, 1984) y una descripción de la náyade (Odonata: Coenagrionidae). Boletin de Entomología Venezolana N.S. 5: 116-122. DIDHAM, R. K., P. M. HAMMOND, J. H. LAWTON, P. EGGLETON & N. E. STORK. 1998. Beetle species responses to tropical forest fragmentation. Ecological Monographs 68: 295-323. DUNKLE, S. W. 1989. Dragonflies of the Florida Peninsula, Bermuda and the Bahamas. Scientific Publishers, Gainesville, Florida. FINCKE, O. M. 1984. Giant damselflies in a tropical forest: reproductive biology of Megaloprepus coerulatus with notes on Mecistogaster (Zygoptera: Pseudostigmatidae). Advances in Odonatology 2: 13-27. FINCKE, O. M. 1992. Behavioural ecology of the giant damselflies of Barro Colorado Island, Panama (Odonata; Zygoptera: Pseudostigmatidae). In: D. Quintero & A. Aiello (eds), Insects of Panama and Mesoamerica, selected studies, pp. 102-113, Oxford Univ. Press, Oxford. FINCKE, O. M. & H. HADRYS. 2001. Unpredictable offspring survivorship in the damselfly, Megaloprepus coerulatus, shapes parental behavior, constrains sexual selection, and challenges traditional fitness estimates. Evolution 55: 762-772. FÖRSTER, S. 1998. Oviposition high above water in Micrathyria dictynna Ris (Anisoptera: Libellulidae). Odonatologica 27: 365-369. FORSYTHE, A. & K. MIYATA. 1984. Tropical nature. Charles Scribner’s Sons, New York. FRASER, F. C. 1957. A reclassification of the order Odonata. Royal Zoological Society of New South Wales, Handbook No. 12: 1-133.
CyanMagentaYellowBlack Odonata page 99
The Importance of Forests to Neotropical Dragonflies
99
GLOBAL FOREST WATCH. 2005. (http://www.globalforestwatch.org/english/about/ faqs.htm#) GONZÁLEZ SORIANO, E. 1997. Odonata. In: E. González Soriano, R. Dirzo & R. C. Vogt (eds), Historia natural de los Tuxtlas, pp. 245-255, Universidad Nacional Autónoma de México. GOOSEM, M. 1997. Internal fragmentation: the effects of roads, highways, and powerline clearings on movements and mortality of rainforest vertebrates. In: W. F. Laurance & R. O. Bierregard, Jr. (eds), Tropical forest remnants, ecology, management, and conservation of fragmented communities, pp. 241255, University of Chicago Press, Chicago. HOEKSTRA, J. D. & R. W. GARRISON. 1999. Range extension of Palaemnema domina Calvert (Odonata: Platystictidae) to southeastern Arizona, U. S. A.: a new odonate family for the United States. Proceedings of the Entomological Society of Washington 101: 756-759. KINVIG, R. G. & M. J. SAMWAYS. 2000. Conserving dragonflies (Odonata) along streams running through commercial forestry. Odonatologica 29: 195-208. KRICHER, J. 1997. A Neotropical Companion. Princeton University Press, Princeton. LAURANCE, S. G. W., P. C. STOUFFER & W. E. LAURANCE. 2004. Effects of road clearings on movement patterns of understory rainforest birds in Central Amazonia. Conservation Biology 18: 1099-1109. MALCOLM, J. R.. 1997. Biomass and diversity of small mammals in Amazonian forest fragments. In: W. F. Laurance & R. O. Bierregard, Jr. (eds), Tropical forest remnants, ecology, management, and conservation of fragmented communities, pp. 207-221, University of Chicago Press, Chicago. MAY, M. L. 1976. Thermoregulation and adaptation to temperature in dragonflies (Odonata: Anisoptera). Ecological Monographs 46: 1-32. MAY, M. L. 1979. Lista preliminar de nombre y clave para identificar los Odonata (caballitos) de la Isla de Barro Colorado. Cuadernos de Ciencias No. 1, Smithson. Tropical Research Institute & La Editorial Universitaria, Panama. MIYAKAWA, K. 1994. Autumnal migration of mature Sympetrum frequens (Selys) in western Kanto Plain, Japan (Anisoptera: Libellulidae). Odonatologica 23: 123-132. MORTON, E. S. 1977. Ecology and behavior of some Panamanian Odonata. Proc. Ent. Soc. Wash. 79: 273. ORR, A. G. 2001. An annotated checklist of the Odonata of Brunei with ecological notes and descriptions of hitherto unknown males and larvae. International Journal of Odonatology 4: 167-220. ORR, A. G. 2003. A guide to the dragonflies of Borneo, their identification and biology. Natural History Publications (Borne), Kota Kinabalu. ORR, A. G. 2004. Territorial behaviour associated with feeding in both sexes of the tropical zygopteran, Libellago hyalina (Odonata: Chlorocyphidae). International Journal of Odonatology 7: 493-504. OSBORN, R. & M. J. SAMWAYS. 1996. Determinants of adult dragonfly assemblage patterns at new ponds in South Africa. Odonatologica 25: 49-58.
CyanMagentaYellowBlack Odonata page 100
100 Dennis Paulson
PAULSON, D. R. 1994. Two new species of Coryphaeschna from Middle America, and a discussion of the red species of the genus. Odonatologica 23: 379-398. PAULSON, D. R. 2004. Critical species of Odonata in the Neotropics. International Journal of Odonatology 7: 163-188. PIANKA, E. R. & L. J. VITT. 2003. Lizards: windows to the evolution of diversity. University of California Press, Berkeley. PRITCHARD, G. 1996. The life history of a tropical dragonfly: Cora marina (Odonata: Polythoridae) in Guanacaste, Costa Rica. Journal of Tropical Ecology 12: 573-581. REHFELDT, G. E. 1986. Verteilung und Verhalten von Segellibellen (Odonata: Libellulidae) während der Trockenzeit in Regenwäldern Panamas. Amazoniana 10: 57-62. REHN, A. C. 2003. Phylogenetic analysis of higher-level relationships of Odonata. Systematic Entomology 28: 181-239. RICHARDS, P. W. 1966. The Tropical Rain Forest. Cambridge University Press, Cambridge. RITH-NAJARIAN, J. C. 1998. The influence of forest vegetation variables on the distribution and diversity of dragonflies in a northern Minnesota forest landscape: a preliminary study (Anisoptera). Odonatologica 27: 335-351. ROBINSON, W. D. 1999. Long-term changes in the avifauna of Barro Colorado Island, Panama, a tropical forest isolate. Conservation Biology 13: 85-97. RODERICK, G. K. & R. G. GILLESPIE. 1998. Alien arthropods in Hawaii: trouble in paradise. Aliens 8: 2-3. SAHLÉN, G. 1999. The impact of forestry on dragonfly diversity in Central Sweden. International Journal of Odonatology 2: 177-186. SAMRAOUI, B., S. BOUZID, R. BOULAHBAL & P. S. CORBET. 1998. Postponed reproductive maturation in upland refuges maintains life-cycle continuity during the hot, dry season in Algerian dragonflies (Anisoptera). International Journal of Odonatology 1: 118-135. SCHWARZKOPF, L. & A. B. RYLANDS. 1989. Primate species richness in relation to habitat structure in Amazonian rainforest fragments. Biological Conservation 48: 1-12. SEKERCIOGLU, C. H., P. R. EHRLICH, G. C. DAILY, D. AYGEN, D. GOEHRING & R. F. SANDI. 2002. Disappearance of insectivorous birds from tropical forest fragments. Proceedings of the National Academy of Sciences 99: 263-267. SHATTUCK, S. O. & N. J. BARNETT. 2001. Distribution patterns of ants within Australia. In Australian Ants Online (http://www.ento.csiro.au/science/ants/ distribution_patterns.htm). SHELLY, T. E. 1982. Comparative foraging behavior of light- versus shade-seeking adult damselflies in a lowland Neotropical forest (Odonata: Zygoptera). Physiological Zoology 55: 335-343. STRATFORD, J. A. & P. C. STOUFFER. 1999. Local extinctions of terrestrial insectivorous birds in a fragmented landscape near Manaus, Brazil. Conservation Biology 13: 1416-1423. TOCHER, M. D., C. GASCON & B. L. ZIMMERMAN. 1997. Fragmentation effects on a Central Amazonian frog community: a ten-year study. In: W. F. Laurance &
CyanMagentaYellowBlack Odonata page 101
The Importance of Forests to Neotropical Dragonflies
101
R. O. Bierregard, Jr. (eds), Tropical forest remnants, ecology, management, and conservation of fragmented communities, pp. 124-137, University of Chicago Press, Chicago. VON ELLENRIEDER, N. 2002. A phylogenetic analysis of the extant Aeshnidae (Odonata: Anisoptera). Systematic Entomology 27: 437-467. WALTER, H. 1973. Vegetation of the Earth in Relation to Climate and the Ecophysiological Conditions. Springer-Verlag, New York. WATANABE, M., H. MATSUOKA & M. TAGUCHI. 2004. Habitat selection and population parameters of Sympetrum infuscatum (Selys) during sexually mature stages in a cool temperate zone of Japan (Anisoptera: Libellulidae). Odonatologica 33: 169-179. WATANABE, M., N. OHSAWA & M. TAGUCHI. 1987. Territorial behaviour in Platycnemis echigoana Asahina at sunflecks in climax deciduous forests (Zygoptera: Platycnemididae). Odonatologica 16: 273-280. WATANABE, M. & M. TAGUCHI. 1993. Thoracic temperatures of Lestes sponsa (Hansemann) perching in sunflecks in deciduous forests of the cool temperate zone of Japan (Zygoptera: Lestidae). Odonatologica 22: 179-186. WHITMORE, T. C. 1998. An Introduction to Tropical Rain Forests. Oxford University Press, Oxford. WILSON, E. O. & R. W. TAYLOR. 1967. The ants of Polynesia (Hymenoptera: Formicidae). Pacific Insects Monograph 14, Bishop Museum Press, Honolulu.
CyanMagentaYellowBlack Odonata page 102
This page intentionally left blank
CyanMagentaYellowBlack Odonata page 103
Use ofAdolfo ForestCordero and Tree Species, Dispersal by Giant Damselflies Rivera (ed)and 2006 Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 103-125. © Pensoft Publishers
Sofia–Moscow
Use of Forest and Tree Species, and Dispersal by Giant Damselflies (Pseudostigmatidae): Their Prospects in Fragmented Forests Ola M. Fincke Department of Zoology, University of Oklahoma, Norman, OK 73019 USA
[email protected]
ABSTRACT Phytotelmata, or water-filled plant containers, provide an important aquatic habitat in tropical forests typically depauperate of permanent ponds and lakes. As top predators in these microhabitats, species of Pseudostigmatidae offer a rare opportunity to measure the effects of forest plant species on the abundance and distribution of their aquatic occupants. Like the specialists of bromeliads, which depend on the presence of a small group of epiphytic plants, odonates ovipositing in water-filled tree holes and fruit husks require a non-random assortment of tree sizes and species. The size and density of microhabitats ultimately affect larval abundance, although, for most species, it remains unclear how closely adult recruitment tracks larval survivorship. Within its geographic range, Megaloprepus relies more heavily on primary forests than do species of Mecistogaster that are adapted to dryer conditions and hence are more tolerant of secondary and highly disturbed forests. An experiment with Megaloprepus revealed that it exhibited relatively low flight endurance, particularly in females, which rarely dispersed across open areas. Recent comparative work challenges the status of Megaloprepus as a monospecific genus, and suggests that many pseudostigmatid populations may be highly structured genetically. The larval ecology and adult behavior of Megaloprepus suggest that its populations should be more higly structured than those of the more vagile tree-hole aeshnids. Collectively, the data reviewed here suggest that forest fragmentation, exacerbated by less predictable threats from global warming, may pose a greater threat to Mega-
103
CyanMagentaYellowBlack Odonata page 104
104 Ola M. Fincke
loprepus and similar species such as Microstigma rotundatum, than to species of Mecistogaster. The fate of all pseudostigmatids is intimately tied to that of the plant species harboring them. As predators of phytotelm mosquito larvae, some of which are disease vectors, the demise of pseudostigmatids may affect not only forest food chains, but also human health.
HABITATS AND HABITS OF PSEUDOSTIGMATIDS Phytotelmata, or water-filled plant containers, provide an important microhabitat in tropical forests (Lounibos 1980; Frank & Lounibos 1983; Kitching 2000; Greeney 2001), areas depauperate of permanant ponds and lakes, which in temperate regions were often formed during past glaciation events. The odonate species that depend on phytotelmata are limited to tropical forests of Central and South America, Africa, Asia, and Australia (reviewed by Corbet 1983). Their dependence on a relatively small group of plant species for larval survivorship makes these odonate predators well-suited for measuring the variables limiting the local abundance and geographic distribution of closely related species. Because, as both adults and larvae, the species are differentially adapted to drying conditions, changes in their abundance and distribution could serve as indicators of forest conversion and/or climate change. Although this review focuses primarily on the Pseudostigmatidae, because phytotelmata generally, and tree holes in particular, impose similar selective pressures regardless of the continent on which they occur (e.g. Orr 1994), general trends found in pseudostigmatids are likely to be relevant for the phytotelmata niche throughout the tropics. Of the 20 species attributed to the family Pseudostigmatidae by Davies & Tobin (1984), three, Mecistogaster amazonica, M. garleppi and Microstigma calcipennis are considered synonyms of Mecistogaster buckleyi and Microstigma anomalum, respectively (Lencioni 2005; D. Paulson, pers. comm.). Because of similarities in its morphology, feeding habits, and larval habitat, the monospecific Coryphagrion grandis of east Africa has long been suspected to be a closely related taxon, but recent genetic work places the African species squarely within the Pseudostigmatidae, close to Mecistogaster (Groeneveld et al. 2006). The latter finding changes the way we view the evolutionary history of the family, as it suggests that the last common ancestor of Coryphagrion and the New World clades is quite ancient (see Clausnitzer and Lindeboom 2002). To date, the genus Megaloprepus is considered monospecific; M. caerulatus is here referred to by its genus name only. Other odonates that regularly co-exist in tree holes with pseudostigmatids are the aeshnid dragonflies, Gynacantha membranalis and Triacanthagyna dentata (DeMarmels, 1992; Fincke 1992a, Fincke 1998). Epiphytic
CyanMagentaYellowBlack Odonata page 105
Use of Forest and Tree Species, and Dispersal by Giant Damselflies
105
bromeliads harbor an even greater array of small coenagrionid species (reviewed in Corbet 1983), although little is known about odonate interactions within bromeliads. Of the phytotelmata available to odonates, bromeliads and tree holes provide the two most distinctive, persistent, and widespread larval niches. Despite our ignorance about 6 species (i.e. Anomisma abnorme, Mecistogaster asticta, M. buckleyi, M. lucretia, M. martinezi and M. pronoti), the majority of pseudostigmatids depend on tree holes rather than bromeliads (Table 1), contrary to Calvert’s (1911) prediction that most members of the family probably depend on bromeliads. Calvert speculated that the phytotelm habit originated in flooded forests of the Amazon, where, at high water, low hanging bromeliads might have been mistaken for aquatic vegetation by ovipositing females. Similarly, the origin of tree hole use might have arisen from oviposition ‘mistakes’ in holes of trees that fall over or along streams or other water bodies, not an uncommon occurance Table 1. Larval habitats of pseudostigmatids. Larval habitat: T= treehole, B = bamboo, F = fruit husk, BR = bromeliad. ? = larvae has yet to be found. Mecistogaster n sp. is very similar to M. jocaste (J. Louton, pers. comm.) Genus
species
Anomisma Coryphagrion Mecistogaster
abnorme grandis amalia asticta buckleyi jocaste linearis lucretia martinezi modesta
Megaloprepus Microstigma
Pseudostigma
n sp. ornata pronoti caerulatus anomalum maculatum rotundatum aberrans accedans
Habitat: Reference T B F BR ? x . x x . . . ? .
. . .
Clausnitzer and Lindeboom (2002) F.A.A. Lencioni (pers. comm.) F.A.A. Lencioni (pers. comm.)
x . x .
. .
. .
Machado and Martinez 1982 Fincke 1984, 1992a, 1998;
. .
? .
. .
. x
. x x .
. .
. .
F.A.A. Lencioni (pers. comm.) Calvert 1911; Melnychuk and Srivastava 2002 Louton et al. 1996 Fincke 1984, 1992a
x . x x
. . . .
. x . x
. . . .
x . x .
. .
. .
?
?
? Young 1980; Fincke 1984, 1992a; Caldwell, 1993 A.A. Lencioni (pers. comm.) De Marmels 1989, Santos 1981, S. Yanoviak (pers. comm.) Fincke 1998 Fincke 1992a
CyanMagentaYellowBlack Odonata page 106
106 Ola M. Fincke
in Panama. Evidence in support of this hypothesis comes from northern Venezuela, where Libellula herculea breeds in quiet pools of mountain streams, but, when there is a flood, holes and crevices in boulder stones in or at the margin of streams fill up and later contain L. herculea larvae (J. De Marmels, pers. comm.). Whereas tree-hole species may also be found in fruit husks, and possibly bamboo internodes, they have seldom been found in bromeliads. One exception is Gynacantha membranalis, found in Peru with unidentified Mecistogaster larvae in an Aechmea sp. of bromeliad that held over 100 liters of water (J. Hoffmann, pers. comm.). Similarly, bromeliad species have not been reported in tree holes or other microhabitats. This is probably due to larval adaptations to dissolved oxygen, which is much higher in bromeliads (Laessle 1961) than in the more anoxic tree holes, where pH also varies across forests (Fincke 1998, 1999; Clausnitzer and Lindeboom 2002). Behavioral differences among larvae offer further evidence that odonates have specialized on tree holes or bromeliads, but not both. For example, larvae of the tree hole species Megaloprepus, Mecistogaster linearis, and M. ornata seldom if ever climb out of rearing jars. In contrast, larvae of M. modesta, a bromeliad species, typically crawl out of the same jars (Fincke 1998). The latter behavior would be adaptive in their native bromeliads, as growing larvae move from smaller to larger leaf axils. Similar behavior was described by Machado for the protoneurid, Roppaneura beckeri, which lives in leaf axils of Eryngium floribundum (see Corbet 1983). Of note is the habit of Coryphagrion grandis larvae, which float upside down in a conspicuous manner (Clausnitzer and Lindeboom 2002). To my knowledge, that behavior distinguishes them from any neotropical phytotelm species. In forests where tree holes are rare, some pseudostigmatids may oivposit in fallen fruit husks that fill with water. In the Brazilian forest where Microstigma anomalum occupied the fruit husks of Brazil nut trees, Bertholletia excelsa (Caldwell 1993), tree holes were uncommon (J. Caldwell pers. comm.). In Kenya, larvae of Coryphagrion grandis were found both in tree holes and water-filled coconut husks (Clausnitzer and Lindeboom 2002). In Panama, on Barro Colorado Island (BCI), where tree holes are common, I have seen only one pseudostigmatid, Mecistogaster ornata, oviposit in a fruit husk (Tonelea ovalifolia), but I have never found any larvae in husks. In the lowland Atlantic forest of La Selva field station in Costa Rica, odonate larvae were never found in husks of Lecythis costaricensis, despite repeated checks. There, fruit husks dry out more readily than tree holes and risk being overturned by animals (Fincke 1998). Fallen palm bracts are the least stable of the phytotelmata available in the tropics; to date, sampling of this microhabitat has not revealed any odonate larvae (e.g. Fincke 1998; Greeney 2001).
CyanMagentaYellowBlack Odonata page 107
Use of Forest and Tree Species, and Dispersal by Giant Damselflies
107
One as yet unidentified pseudostigmatid is known to use bamboo internodes (Louton et al. 1996a), and two other small species, Mecistogaster asticta and M. martinezi are also suspected to do so (F.F. Lencioni, pers. comm.). The unidentified larvae found in Peruvian bamboo were never found in exposed internodes that tended to quickly dry out. Rather, females used only internodes that were accessed via small perforations created by beetles (Louton et al. 1996a), suggesting that this species might be a true bamboo specialist, unlike the asian non-pseudstigmatids known to use both tree holes and bamboo (reviewed by Corbet 1983). Females were similar to M. jocaste but smaller, and laid their eggs endophytically (J. Louton, pers. comm.). M. jocaste females have been described as ‘shooting’ into a tree hole eggs subsequently found floating on the water surface (Machado & Martinez 1982). Note that Machado, in an unpublished record, later described the species as M. martinezi (J. Louton, pers. comm.). However, I suspect that the female was merely hitting the water surface with her abdomen to make sure that the hole held water, similar to the behavior of female M. linearis before they perch to lay their eggs endophytically (Fincke 1992b). When reared, eggs floating on the water surface of Peruvian tree holes were always those of Toxorhynchites mosquitos (J. Louton, pers. comm). To my knowledge, all other observations of pseudostigmatid egg-laying, as well as the structure of the ovipositor, suggest endophytic oviposition. Work on a molecular phylogeny of the Pseudostigmatidae is underway, and should help resolve the above discrepancies and permit us to determine whether bamboo use has evolved multiple times, independently (Hadrys, Fincke, Lencioni, unpublished data). Within phytotelmata, odonates are top predators whose prey include the larvae of mosquitos, ceratopogonids, chironomids, tipulids, other odonates, and the tadpoles of several anuran genera (Fincke 1992a, 1998, 1999; Caldwell 1993; Yanoviak 2001; Haugen 2002; Melnychuck and Srivastava 2002). Kitching (2000) offers a general summary of phytotelm food webs. However, the effects of specific odonates on the dynamics of prey species probably varies. Megaloprepus, for example, is a more voracious predator than co-occuring Mecistogaster larvae, and seems to have a disproportionately great impact on its prey (Fincke 1998). Although both pseudostigmatids and tree hole aeshnids can greatly reduce the percent of mosquitos that successfully emerge from tree holes, the smaller Mecistogaster larvae appear to be the least effective in doing so (Fincke et al. 1997, Fincke, unpublished data). Some phytotelm prey have evolved responses to their odonate predators. In Jamaica, female crabs kill any odonate larvae in epiphytic bromeliads before laying their own eggs there (Diesel 1992). With over a hundred species of frogs also breeding in tree holes, bamboo, fruit husks, and brome-
CyanMagentaYellowBlack Odonata page 108
108 Ola M. Fincke
liads (Lehtinen 2004), one expects similar adaptations among phytotelm anurans, but few have been documented. Female Dendrobates pumilio search the tree hole or bromeliad before laying their eggs there (Summers 1989), a tactic that apparently reduces the risk of odonate predation on tadpoles (Fincke 1998). In Amazonian Peru, a female laughing frog, Osterocephalus planiceps lays hundreds of eggs into a single bromeliad leaf axil (Haugen 2002), and the resulting viscousity of the egg mass suffocates any odonate larva present (L. Haugen pers. comm.). As adults, Coryphagrion, Mecistogaster, Megaloprepus, Microstigma, and Pseudostigma have all been observed to prey on small orb-weaving spiders, which are plucked from their webs (Calvert 1911, 1923; Young 1980; Fincke 1992b; Clausnitzer & Lindeboom 2002; F.A.A Lencioni, pers. comm.). Although they specialize on taking spiders, pseudostigmatids are known occasionally to take the wrapped prey in spider webs (Stout 1983; Young 1980). In my experience, typical spider prey are < 6 mm in total length. Within the forest, the damselflies forage on spiders in sun flecks and sunny gaps created by large branch or tree falls. These high-light environments enable the damselflies to detect spider webs; in low light they risk becoming entangled in the webs, which is one reason why it is difficult to keep pseudstigmatids alive for very long in small, outdoor insectaries (Fincke unpublished data). The UV-reflectant wing and abdominal signals of Megaloprepus and M. linearis, respectively, are also adapted to high-light conditions (Schultz and Fincke, unpublished data). Although there may be some habitat partitioning with respect to foraging height among co-exisiting adults (Fincke 1992b), the abundance of their spider prey is unlikely a factor limiting the number of adult pseudostigmatids. More significant limiting factors are the seasonal duration of tree holes and the abundance of that larval resource, coupled with the dynamics of cannibalism and intraguild predation in the larval stage (Fincke 1992a,c, 1994).
SIZE AND DENSITY OF MICROHABITATS All the original data presented herein were collected from the lowland, seasonally moist forest of Barro Colorado Island (BCI), Panama (for its general ecology, see Leigh et al. 1982; Leigh 1999). Over the past two decades, I have tagged and censused several hundred individual tree holes used as habitats (i.e. containing > 100 ml water) by treehole odonates to quantify the distribution and species interactions of 5 species of odonates, 3 species of tadpoles, and the predatory mosquito larva, Toxorhynchites (Fincke 1984, 1992a, 1994, 1998). Occupancy by odonates never exceeded 70% of tree holes sampled frequently.
CyanMagentaYellowBlack Odonata page 109
Use of Forest and Tree Species, and Dispersal by Giant Damselflies
109
Tree holes, particularly those formed when a tree falls and indentations in the trunk fill with water, harbor as much as 50 l of water (Fig. 1a, b). However, the majority of tree holes hold less than one liter (Fig. 1 c, d). Persistence of a hole from one year to the next depends primarily on whether it is in a live or dead tree. Due to the rapid rate of decomposition on BCI, only 42% of the holes in fallen trees and 56% of those in dead, upright trees held water for more than two years. In contrast, only about 3% of the holes in live, upright trees failed to hold water from one year to the next (Yanoviak and Fincke 2005). Although the limiting nature of such larval habitats predisposes them to be defended by adults, variation in the mating system of pseudostigmatids indicates that their value to adults varies across species. Whereas some species defend large microhabitats (e.g. Megaloprepus Fincke 1992c, Microstigma rotundatum, De Marmels 1989, Mecistogaster modesta, Srivastava et al. 2005), others briefly defend light gap areas where matings occur (M. linearis, Fincke 1984) or merely mate in sun flecks and gaps where feeding occurs (M. ornata, Fincke 1984). Such differences in mating tactics reflect differences among species in the competitive ability of their larvae; there is minimal niche partitioning based on the characteristics of tree holes used by each species (Fincke 1992a). Within occupied phytotelmata, larval density depends on priority effects and the species’ propensity for cannibalism and intraguild predation (Fincke 1992a, 1999). In Megaloprepus, females oviposit in a wider assortment of tree hole volumes and shapes than are defended by males. Females lay more eggs in large tree holes than small ones. But regardless of the size, females typically lay many more eggs than could ever survive in any given hole. For example, clutch size of Megloprepus ranges from about 50 to 500, but larval density is reduced via cannibalism and obligate siblicide (Fincke 1994, Fincke & Hadrys 2001, Fincke unpublished data) to about one larva /l (Fincke 1992a). Most Megaloprepus recruited to the next generation are produced in the largest holes, which can produce three cohorts per season, or an estimated few dozen individuals surviving to emergence (Fincke 1992b, 1998). Tree holes under one liter rarely produce more than one adult at a time, and rarely more than two adults per wet season on BCI. Similary, the density of Microstigma in fruit husks is usually one per husk, the volume of which, in one study, never exceeded 245 ml. (Caldwell 1993). Preliminary results from a census of water-filled tree holes under 2 m in height, in a tract of primary forest in Panama indicated an estimated density of 13.87 usable holes per hectare (Fincke unpublished data). Similarly, in a west African forest on the Ivory Coast, tree hole habitats occured at a mean density of 8.3 tree holes/ha, the highest density being 23/ha (Rödel et al. 2004). Such measures underestimate the true density of available larval microhabitats be-
CyanMagentaYellowBlack Odonata page 110
110 Ola M. Fincke
cause tree holes occur from the ground level to the forest canopy (Yanoviak 1999), above the level at which holes can be monitored easily. But even if we assume the density to be twice the above estimates, they would still be an order of magnitude less than the density of bromeliad microhabitats. The largest epiphytic bromeliads typically hold at most a few liters of water in a central tank-like container with surrounding leaves (Laessle 1961; Richardson 1999). Because they are divided into multiple leaf axil compart-
a
c
b
d
e
Fig. 1. Variation in trees and their tree holes a) large pan hole in a fallen Platypodium elegans, BCI that was a habitat for two years before rotting through b) large hole (30-50 l) in Ceiba pentandra tree, known to be consistently defended by Megaloprepus from 1982 until the tree fell in 1990 c) bowl-shaped hole (< 1 l) in live tree d) tiny hole in a root buttress, too small to support an odonate to emergence e) Bursera semiruba at Los Tuxlas, Mexico, a species in which I have never found tree holes.
CyanMagentaYellowBlack Odonata page 111
Use of Forest and Tree Species, and Dispersal by Giant Damselflies
111
ments, bromeliads support a higher density of odonates per ml of water than do tree holes and fruit husks. In the wet forest at Pitilla Biological Station in Guanacaste, Costa Rica, where Mecistogaster modesta was the only odonate occupant, larval density in occupied bromeliads ranged from 1-2 per plant in the primary forest canopy, where most plants contained less than one liter of water (Melnychuck and Srivastava 2002), and 1-26 larvae/plant in the secondary forest at the same site (Srivastava et al. 2005). Although they are typically smaller in volume than the largest tree holes, epiphytic bromeliads can occur at much higher density of microhabitats per hectare. In the above secondary forest, the density of bromeliads above 40 cm diameter, roughly the minimum size for odonate development, was about 680 per hectare (Melnychuk and Srivastava 2002). Here, in early May, adults were commonly seen at bromeliads and 73% of those inspected contained larvae in Oct-Nov. (Srivastava et al. 2005). This compared with about 420 suitable bromeliads/hectare in primary forest, where only 23% had larvae (Srivastava et al. 2005). In the lowland, aseasonally wet forest at the La Selva station in Costa Rica, only about a third of bromeliads sampled in June and July held larvae (Fincke 1998). The inability of females to find all of the oviposition sites scattered about a forest may explain these intermediate levels of microhabitat occupancy. A critical question for species conservation is how larval abundance translates into adult abundance. Based on an apparently high density of bromeliads relative to tree holes in wet forests such as La Selva, one might predict that the smaller Mecistogaster modesta should outnumber the larger Megaloprepus, but this does not seem to be the case. In a pooled sample of 64 pseudostigmatid adults caught over 18 days between Sept. 1966 and Aug. 1967, 56% were Megaloprepus, 39% were Mecistogaster modesta, and 4% were M. linearis (D. Paulson, pers. comm.). At the same site, another sample of 93 marked adults (i.e. to avoid recounts of the same individual) between June and July, 1991 gave: 70% Megaloprepus, 18% Mecistogaster modesta, 2% each of M. linearis and Pseudostigma aberrans (Fincke 1998). The sample differences may reflect seasonal trends in the abundance of M. modesta (see Hedström and Sahlén 2001), but they don’t explain the dominance of Megaloprepus.
EFFECTS OF TREE SIZE AND SPECIES The abundance and geographic distribution of pseudostigmatids depends not only on their adaptation to abiotic conditions of tropical forests and the suite of odonates present, but on a non-random array of plant species. Among bromeliads, only a subset of species such as those in the genera Aechmea, Guzmania, and Vriesia retain sufficient water to provide suitable habitats for
CyanMagentaYellowBlack Odonata page 112
112 Ola M. Fincke
odonates (Corbet 1983; Melnychuk and Srivastava 2002). Even fewer species (e.g. Bertholletia excelsa, Lecythis costaricensis, Tonelea ovalifolia ) produce fruits whose fallen husks collect water. In addition to particular species of bamboo, the use of that microhabit may also require the presence of a katydid to create the initial hole (Louton et al.1996a). Among tropical forests, the presence of tree holes is expected to vary as a function of tree-species composition, which can vary considerably among geographic areas (e.g. Gentry 1990). Because the formation of holes depends on upon features such as indentations in the tree bole or in buttresses, and the propensity for burls or holes to form, species with very smooth boles, such as Bursera simarouba, are unlikely ever to have tree holes (Fig. 1e). In my experience, the density of pseudostigmatids is relatively high on BCI, where tree-hole species such as Platypodium elegans and Ficus spp. are common (Fincke 1992a, herein). The abundance of Megaloprepus is also relatively high at La Selva, where an estimated 1/3 of all stems are the woody tree, Pentaclethra macroloba (Hartshorn and Hammel, 1994), the trunk morphology of which provides multiple holes in a single tree (Fincke 1998). In contrast, at Aceer station in northern Peru, the forest is depauperate of tree holes, and Microstigma rotundatum is relatively scarce (personal observation). We have only begun to quantify the tree species that provide tree hole habitats. Here I report on a subset of 110 tree holes on BCI for which the tree species were known. The forest there comprises 409 woody tree species, providing a canopy height of 25-35 m featuring a few emergents over 40 m (Foster and Brokaw 1982). Some of the trees holes sampled were located in 90-yr-old secondary forest, but most were coincidentally located in the Hubbell-Foster 50hectare plot. This is an intensively studied plot of primary forest on the central plateau of the island, where every stem over 2 cm diameter at breast height (dbh) has been tagged and identified to species as part of an ongoing study of forest dynamics (Hubbell and Foster 1983). The plot contains 75% of the island’s woody tree species > 1 cm dbh, and 317 woody species > 100 mm dbh. Treeholes in this study were located primarily along trails or in treefall gaps, and my sample was not designed to be representative of the plot itself. All tree holes in my sample were below 2 m. in height, above which it was impractical to monitor tree holes (see Yanoviak and Fincke 2005 for sampling methods). Data from an ongoing census of the 50-hectare plot, once completed, will be reported elsewhere. All variables were log transformed for statistical analysis. For the 52 trees in my sample for which tree dbh, species identity and tree hole volume were known, dbh was positively correlated with tree hole volume (Fig. 2) and the total water held in holes (r = 0.94, P < 0.0001), but not with the number of holes per tree (r = 0.08, P = 0.52). The three largest trees were all Ceiba pentandra, which collectively accounted for 47.5 litres of water, 36% of the total 132 liters held by the 52 trees in the sample. The
CyanMagentaYellowBlack Odonata page 113
Use of Forest and Tree Species, and Dispersal by Giant Damselflies
113
smallest tree with a tree hole over 0.1 l was a Eugenia oerstedeana whose dbh was only 45 mm. My results suggest that tree hole volume increases with the age of a tree species, at least up to a certain age, after which the number of holes may not increase with dbh. r = 0.38 P = 0.005 n = 52
Lo g Me a n Vo lu m e (l)
2 1.5 1 0.5 0 -0.5 -1 -1.5 1.5
2
2.5
3
3.5
4
Log DBH (mm)
Fig. 2. Tree-hole volume as a function of host tree dbh on BCI, Panama.
16
Percent of tree holes
14 12 10 8 6 4 2 0 1
3
5
7
9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Species rank
Fig. 3. Species rank, by percentage contribution to the total holes sampled (n= 110) on BCI, Panama. The top 11 species are Platypodium elegans, Quaraibea asterolepis, Ficus trigonata, Trichilia tuberculata, Ceiba pentandra, Alseis blackiana, Gustavia superba, Randia armata, Eugenia oerstedeana, Hirtella triandra, Brosimum alicastrum.
CyanMagentaYellowBlack Odonata page 114
114 Ola M. Fincke
The 110 tree holes in my sample were found in a total of 35 tree species in 32 genera (Table 2). Of these, only 23% are known to be colonizing species (see Table 2. Family, genus and species, mean holes per tree (range), and volume per hole. N = 110 tree holes known to be larval habitats. * indicates colonizing species (see Condit et al. 1996); † includes 1 fallen tree. Family
Genus
Species
Mean ± s.e. Mean ± s.e. holes/tree liters/hole
Annonaceae Apocynacea Bignoniaceae Bombacaceae
Guatteria Aspidosperma Jacaranda Ceiba Quararibea Dipteryx Cordia Hirtella
dumetorum cruenta copaia * pentandra asterolepis panamensis bicolor * americana triandra terniflora triloba * acuparium versicolor inermis oleifera elegans * rohrii arborea * whitei superba * pallida tuberculata alicastrum costaricana trigonata sebifera oerstedeana standleyanum blackiana foliacea glabrescens * armata stipitata membranacea seemannii *
1 1 1 1.7 (1-3) 1.6 (1-3) 2.5 (2-3) 1 1 1 2 1 1 1 1 1 3.7 (1-10) † 1 1 1 1 1 1 1 1 2.67 (1-6) 1 1 1 1.3 (1-3) 2 1 1.3 (1-2) 2 1 1
Boraginaceae Chrysobalanacea
Sloanea Adelia Sapium Fabaceae: Caesal. Tachigali Fabaceae: Papil. Andira Dipteryx Platypodium Pterocarpus Flacourtiaceae Casearia Lauraceae Ocotea Lecythidacea Gustavia Meliaceae Trichilia Elaeocarpaceae Euphorbiaceae
Moraceae
Brosimum Ficus
Myristicaceae Myrtaceae Nyctaginaceae Rubiaceae
Virola Eugenia Guapira Alseis Guettarda Macrocnemum Randia Pouteria Apeiba Luehea
Sapotaceae Tiliaceae
0.6 0.5 0.6 9.5 ± 5.9 0.66 ± 0.6 1.15 ± 1.6 0.3 0.8 0.47 ± 0.08 0.9 . . 0.35 ± 0.15 0.5 0.7 2.0 ± 2.5 0.1 0.1 0.8 0.17 ± 0.06 0.8 ± 0.40 0.42 ± 0.11 0.3 . 1.0 ± 0.40 0.1 0.13 ± 0.04 0.1 9.2 ± 12.6 0.1 0.35 ± 0.17 0.62 ± 0.6 0.25 ± 0.18 0.3 0.2
CyanMagentaYellowBlack Odonata page 115
Use of Forest and Tree Species, and Dispersal by Giant Damselflies
115
Condit et al. 1996), and hence, common in secondary forests (R. Condit pers. comm.). Over half the holes were accounted for by eight species (Fig. 3). For trees with holes, there was no relationship between tree species abundance rank and its rank by tree hole number (r = -0.04, P = 0.83). However, of the species that represented half the individuals in the respective growth forms of canopy (6 species) or midstorey (9 species) trees on BCI (Hubbell & Foster 1992), three mid-storey (Eugenia oerstedeana, Hirtella triandra, Virola sebifera) and three canopy trees (Alseis blackiana,Tachigali versicolor, and Trichilia tuberculata) provided water-filled tree holes in my sample. And of these, A. blackiana and T. tuberculata are the two most common canopy species on BCI. Nevertheless, most species were represented by only one individual, and hence my analysis of species nested within genus indicated no effect of species identity on hole volume or total water volume; even the effect of genus on volume was insignificant (F1,30 = 2.06, P = 0.16). Preliminary results from censusing all tree holes under 2 m in 13 of the 50 hectares indicated that several tree species provided more tree holes than expected from their abundance (Ceiba pentandra, Dipteryx oleifera, Platypodium elegans, Fincke unpublished data). Of these, P. elegans is a colonizing species, and where present, offers considerable potential for providing larval habitats in secondary forests.
USE OF PRIMARY VERSUS SECONDARY FORESTS Given the increasing rate of tropical deforestation, the survival of primary forest odonates will depend on their ability to survive in conditions less than pristine. Pseudostigmatids are differentially susceptible to dry conditions, and may thus differ in their ability to avoid local extinction after forest clearing and succession. Within its geographic range, Megaloprepus is notably absent from tropical dry forests, where Mecistogaster ornata are common (Hedström and Sahlén 2001). Several lines of evidence suggest generic differences in physiological tolerance to drying. Whereas both M. linearis and M. ornata can be found flying throughout the dry season on BCI, Megaloprepus siccatates during that time (Fincke 1992b). And even in the wet season, adults avoid flying across the large laboratory clearing, and do not frequent extremely large natural gaps that offer no shaded perches (O.M. Fincke personal observation). Yanoviak (1999) found larvae of M. linearis and M. ornata, but not Megaloprepus, in canopy tree holes on BCI. In the high-light environment of open canopy, water temperature was higher and pots (i.e. plastic surrogate tree holes) were more likely to dry out than those in understorey. As adults, pseudostigmatid genera also use the forest differently. On BCI, where the primary forest (> 400 yrs) is contiguous with 90-yr-old secondary forest, adult Megaloprepus colonized tree holes in both forest types. Nevertheless, despite continuous canopy over both
CyanMagentaYellowBlack Odonata page 116
116 Ola M. Fincke
forest types, Megaloprepus were more abundant in primary forest, whereas two co-occurring Mecistogaster (linearis and ornata) were more common in the secondary forest. Similarly, in a forest of the Nairi reserve in Limón province on the Caribbean slope of Costa Rica, Megaloprepus failed to colonize pots adjacent to recently logged forests. In constrast, co-existing Mecistogaster linearis colonized pots in both forest types (Fincke and Hedström, in review). These results were similar to data from Amazonian Peru, where Microstigma rotundatum colonized bamboo pots in successional and primary forests, but not in small-scale subsistence clearings (Yanoviak et al. 2006). Although Mecistogaster modesta and the bromeliads on which they depend rely on relatively wet forests, they are found in both primary and secondary forests. In Costa Rica where the high-light environment of a secondary forest favored a great abundance of bromeliads, Srivastava et al. (2005) predicted abundance of M. modesta to be a remarkable 36x higher than in a nearby primary forest. It is not known whether differences in adult abundance were of similar magnitude. No adults were found in the primary forest during the surveys, and were rarely seen at that site, suggesting that this species has a preference for secondary forest.
DISPERSAL ABILITY OF MEGALOPREPUS In contrast with tree-hole aeshnids, pseudostigmatids are not strong flyers. From the canopy tower on BCI, I have often watched dragonflies foraging in the canopy, but these never included pseudostigmatids, probably because the high wind velocity of the canopy is not conducive to the hovering flight needed to forage on web-building spiders there (see Rüppell and Fincke 1989). However, pseudostigmatids can and do disperse considerable distances in forest understorey. Studies of marked individuals within forests indicate that Mecistogaster linearis and M. ornata can travel several km. in a few weeks, and Megaloprepus can travel 3.5 km in less than a week (Fincke 1984, unpublished data). A factor critical to their conservation is the extent to which treehole species disperse across an unforested landscape. As a first step towards answering this question, in January 1997, I measured the maximum flight duration of individual Megaloprepus released over Gatun Lake, Panama (see map in Leigh et al., 1982). Flight was measured in early morning, starting shortly after sunrise on clear, sunny, days with little or no wind. Individuals were collected from the field during the previous afternoon and held individually in small cages overnight. The next morning, each individual was marked if it wasn’t already, weighed, and scored for age, based on wing wear, from 1-3, 3 being old and 1 being young. The damselflies were put in a small cage and transported by motor boat north of the island to a marker buoy
CyanMagentaYellowBlack Odonata page 117
Use of Forest and Tree Species, and Dispersal by Giant Damselflies
117
in Lake Gatun. Upon release, an individual was followed by boat until it fell into the water, at which time it was retrieved. The compass direction of its intial flight and any change in direction was recorded. Total flight time was recorded, and the individual’s final position was estimated relative to marker buoys in the lake. The distance flown was then estimated using nautical maps showing the position of canal buoys. This method ignores actual distance flown if the individual circled. Except for one male (of unknown territorial status) that died after falling into the water after 212 sec. of flight, all individuals were released in the BCI forest less than 2 hrs after the start of the experiment. Variables were log transformed for analysis. Means are reported ± s.e. In total, 8 males and 3 females were tested. Of the males, five were age 1 and three were age 2; three were current territory residents at defended sites, two were satellite males, and the remainder were not collected at a territory. Despite a general trend for males of this species to be larger than females (Fincke 1992c), the males used for testing did not differ significantly from the 3 females in abdomen (t = -1.9, P = 0.09) or wing length (t = -1.06, P = 0.32); nor were they significantly heavier (t= -1.56, P = 0.17). Consequently, there was no difference in wing loading between the sexes in this sample (t = -1.3, P = 0.23). There was no trend in the direction the damselflies took upon release over water. Three headed east, two south, two west, two north, and two circled. As shown in Fig. 4, males flew significantly longer (¯x = 554.8 sec, range = 114-968 sec) than did females (¯x = 65.0 ± 27.5 sec, range 10-94 sec, t = -4.0, P = 0.004). There was no difference between satellite and nonterritorial males in flight duration or distance flown, so the two groups were pooled for analysis. Relative to males not known to be territorial, territorial 900 800 700 600 Territorial
500
Nonterritorial Female
400 300 200 100 0 Seconds
Meters flown
Fig. 4. Flight duration and distance traveled by 8 male and 3 female Megaloprepus over Gatun Lake, Panama.
CyanMagentaYellowBlack Odonata page 118
118 Ola M. Fincke
males flew for a longer time (t = -2.70, P = 0.05), but they frequently circled. Consequently, they covered less distance than did the non-territorial males (t = 2.68, P = 0.04). The longest distance traversed was an estimated 969 m by a non-resident male that flew for 968 sec, equivalent to a speed of 1 m/sec. The fastest rate was 4.9 m/sec, by a non-resident male that flew for 114 sec. There was no effect of male age on flight duration (rs = -0.11, P = 0.79) or distance flown (rs = -0.23, P = 0.59). Males flew farther (¯x = 391.2 ± 106.4 m) than females (¯x = 8.3 ± 6.0 m, P < 0.01). The female flying the longest (94 sec) circled frequently, and traversed only 20 m. All 3 females and the 7 remaining males flew off with no apparent difficulty when later released into the forest. Although none of the Megaloprepus flew from BCI across the canal to the mainlaind on the north, the flight time of males would have been sufficient for males to cross the relatively more narrow stretch of water separating BCI from Gigante Peninsula on the mainland on the south side of the island. Because I found Megaloprepus larvae in tree holes on small Pepper Island in the canal, this species may also reach the mainland by island hopping. The most difficult result to explain is the behavior of the females. Despite the small sample size, females were strikingly different from the males in their reluctance to fly over open water. This suggests that the sexes may differ in their effective dispersal among forest patches. That the test females had the ability to disperse is evident from their immediate flight upon being released in the forest.
GEOGRAPHIC VARIATION IN MEGALOPREPUS POPULATIONS Dispersal ability directly affects gene flow among populations. I expect, based on differences in flight endurance, that gene flow would be greater among the faster-flying treehole aeshnids (see Wikelski et al. 2006) than among pseudostigmatids. Species differences in mating systems, patterns of female oviposition, and degree of cannibalism and intraguild predation among larvae all have consequences for the effective population size Ne, the number of adults that contribute genes to the next generation (for the formula, see Futuyma 1998). Using microsatellite genetic markers (Hadrys et al. 2005) to quantify parentage of larvae successfully emerging from holes, Fincke and Hadrys (2001) demonstrated that its resource-defense mating system, coupled with high larval cannibalism, reduced Ne of Megaloprepus well below the number of sexually mature adults. For non-territorial species or those whose larvae cannibalize only when hungry, Ne should more closely approach the number of mature individuals. Because genetic variation should decrease as Ne and dispersal decrease (see Futuyma 1998), I expect populations of Megaloprepus to be more highly structured genetically than those of Mecistogaster ornata or the more vagile tree-hole aeshnids.
CyanMagentaYellowBlack Odonata page 119
Use of Forest and Tree Species, and Dispersal by Giant Damselflies
119
Initial comparative work on Megaloprepus offers some support for the prediction that its populations are highly structured genetically (i.e. among population genetic variance >> than the genetic variation within populations). First, populations differ in body size. At La Selva in Costa Rica and at Los Tuxtlas field station in Veracruz, Mexico, males and females were larger than on BCI in Panama (Fincke 1998, unpublished data). Secondly, at Los Tuxtlas, as on the Osa Peninsula on Costa Rica’s Pacific coast (D. Paulson, pers. comm.), males lack the broad white wing band that characterizes the sexually dimorphic species in Panama and throughout most of its range (Hadrys and Fincke unpublished ms.). On BCI, it is the female’s white wing tip that reveals her sex to males. A male will take in tandem a male whose wing tips have been painted white, whereas he will try to fight with a female whose white wing tips have been artificially darkened (Fincke, unpublished ms.). Cues to sexual recognition in the more sexually monomorphic populations remain unknown. Preliminary analysis of the Los Tuxtlas and BCI populations indicated that their genetic divergence is nearly as great as it is between some other odonate species (Hadrys and Fincke unpublished ms.). Such strong population divergence may reflect past or current geological barriers. However, current populations are becoming increasingly isolated due to the rapid deforestation that has occurred throughout Central America (Brown and Hutchings 1997). Los Tuxtlas has become an island of tropical forest in a sea of cow pastures and agricultural lands. Based on the station’s collection of Megaloprepus and my own experience with this population since 1994, it appears to be in decline (also, E. González Soriano, pers. comm.). Preliminary results from a dispersal experiment suggested that M. caerulatus does not readily colonize secondary forest patches one km away from primary forest (Fincke and Haalboom, unpublished data). Subspecies status of populations within M. jocaste, M. linearis, M. lucretia, M. modesta, and M. ornata (R. Garrison 2004) are also indicative of genetically structured populations.
FUTURE PROSPECTS FOR PSEUDOSTIGMATIDS Much of the work on the effects of forest fragmentation on the fauna of primary tropical forests has focused on birds or mammals, there being relatively little work on insects (e.g., Pimm and Raven 2000; Frankie and Matta 2004). Because of their visually conspicuous adults and discrete larval habitats in a small subset of plant species, phytotelm odonates are well suited for documenting the effects of forest conversion across species that differ in vagility and habitat requirements. Collectively, the data reviewed here suggest that Megaloprepus, and by analogy, Microstigma rotundatum, are reliable
CyanMagentaYellowBlack Odonata page 120
120 Ola M. Fincke
indicators of primary forest, and consequently more vulnerable to forest fragmentation than some species of Mecistogaster, a genus which also occurs in tropical dry forests. In a survey of Peruvian odonates, Louton et al. (1996b) found that of five tree-hole odonates, only Gynacantha membranalis and Mecistogaster linearis were present at all three of their collecting sites, which spanned a distance of about 1,000 km. Those data support my conclusion that the ability of a phytotelm odonate to persist in secondary forests will depend both on its tolerance of such sites, and its ability to disperse across open areas. Hence, one might expect the distributions of species such as Microstigma rotundatum and Megaloprepus to become increasingly disjunct, unless dispersal routes via riparian or forested corridors are maintained to connect primary forest habitats. Ongoing work focuses on quantifying the genetic structure of pseudostigmatid populations in comparison with treehole aeshnids to test such predictions (Hadrys and Fincke, unpublished data). Global warming is also likely to affect the future of the guild of phytotelm odonates, but in less predictable ways. El Niño effects may provide some clues (see Curtis and Adler 2003). In one experiment I conducted on BCI during 1997-1998, when, due to El Niño effects, the dry season was greatly extended, tree holes dried a full month earlier than usual, and larvae took nearly twice as long to emerge from experimental pots that were kept filled with water (Fincke, unpublished). This was a curious result because there was no lack of mosquito prey. Elevated water temperature might have played a role. Deforestation is already reducing rainfall in some tropical areas (e.g. Rand & Rand, 1982). Global warming may exacerbate the trend, threatening tree hole aeshnids, which require a minimum of 4-5 months to develop (Fincke 1992a), or species in forests where the wet season is already relatively short (e.g. Los Tuxtlas, González Soriano 1997). As important natural predators of larval mosquitos (Fincke et al. 1997), the fate of pseudostigmatids in tropical forests will have consequences reaching beyond their effects on species assemblages of phytotelm habitats. Several tree-hole mosquitoes are important disease vectors (e.g. Galindo et al. 1955; Theiler and Downs 1973; Pecor et al. 2000; Jones et al. 2004). Given that larvae of Megaloprepus and the tree hole aeshnids are more voracious predators than are those of Mecistogaster (Fincke unpublished ms.), the elimination of the former may have a greater impact on mosquito populations than the disappearance of the latter. Documenting the effects of forest clearing on the abundance of phytotelm species requires long-term data. Unfortunately, given the unrestrained growth of the human species, increasing stress will be imposed on many pseudostigmatid species, whose future, like so much of the world’s biodiversity, remains uncertain.
CyanMagentaYellowBlack Odonata page 121
Use of Forest and Tree Species, and Dispersal by Giant Damselflies
121
ACKNOWLEDGEMENTS I am grateful to the Smithsonian Tropical Research Institute, which has provided logistical support of my research over the past two decades. I am indebted to S.P. Hubbell and R.B. Foster for access to data on dbh and species identity of trees in the 50-hectare plot, and to R. Condit who confirmed tree species found in secondary forests. I thank J. McFarland for assistance in the field, and A. Cordero for organizing the symposium on forest odonates. The ms. was improved by comments from D. Srivastava and an anonymous reviewer. The original studies herein were supported by grant IBN-9408143 from the National Science Foundation.
REFERENCES BROWN, K.S., JR. & R.W. HUTCHINGS. 1997. Disturbance, fragmentation, and dynamics of diversity in Amazonian forest butterflies. In: W.F. Laurance & R.O. Bierregaard, Jr. (eds), Tropical Forest Remnants: Ecology, Management, and Conservation of Fragmented Communities, pp. 91-110, University of Chicago Press, Chicago. CALDWELL, J. P. 1993. Brazil nut fruit capules as phytotelmata: interactions among anuran and insect larvae. Canadian Journal Zoology 71: 1193-1201. CALVERT, P. P. 1911. Studies on Costa Rican Odonata. II. The habits of the plantdwelling larva of Mecistogaster modestus. Entomological News 22: 402-411. CALVERT, P. P. 1923. Studies of Costa Rican Odonata. X. Megaloprepus, its distribution, variation, habits and food. Entomological News 34: 168-174. CLAUSNITZER, V. & M. LINDEBOOM. 2002. Natural history and description of the dendrolimnetic larvae of Coryphagrion grandis (Odonata). International Journal of Odonatology 5: 35-50. CONDIT R., S. P. HUBBELL & R. B. FOSTER. 1996. Changes in tree species abundance in a Neotropical forest: impact of climate change. Journal of Tropical Ecology 12: 231-256. CORBET, P.S. 1983. Odonata in phytotelmata, In: J.H. Frank & L. P.Lounibos (eds), Phytotelmata: terrestrial plants as hosts for aquatic insect communities, pp. 29-54, Plexus Publishing, Inc., Medford, N.J. CURTIS, S. & R. F. ADLER. 2003. Evolution of El Niño-precipitation relationships from satellites and gauges, Journal of Geophysical Research 108: 4153. DAVIES, D.A.L & P. TOBIN. 1984. A synopsis of the dragonflies of the world: a systematic list of the extant species of Odonata. Vol.1. Zygoptera, Anisozygotera. Societas Internationalis Odonatologia Rapid Communications (Supplements), No. 3, Utrecht. DE MARMELS, J. 1989. Odonata or dragonflies from Cerro de la Neblina I. Adults. Academia de las Ciencias Fisicas, Matematicas y Naturales. Vol. 25.
CyanMagentaYellowBlack Odonata page 122
122 Ola M. Fincke
DE MARMELS, J. 1992. Dragonflies (Odonata) from the Sierras of Tapirapeco and Unturan, in the extreme south of Venezuela. Acta Biologica Venezeuelica 14: 57-78. DIESEL, R. 1992. Maternal care in the bromeliad crab, Metopaulias depressa: protection of larvae from predation by damselfly nymphs. Animal Behaviour 43: 803-812. FINCKE, O.M. 1984. Giant damselflies in a tropical forest: reproductive biology of Megaloprepus coerulatus with notes on Mecistogaster (Zygoptera: Pseudostigmatidae). Advances in Odonatology 2: 13-27. FINCKE, O.M. 1992a. Interspecific competition for treeholes: consequences for mating systems and coexistence in neotropical damselflies. American Naturalist 139: 80-101. FINCKE, O.M. 1992b. Behavioral ecology of the giant damselflies of Barro Colorado Island, Panama (Odonata: Zygoptera: Pseudostigmatidae). In: D. Quintero & A. Aiello (eds), Insects of Panama and Mesoamerica: Selected Studies, pp. 102-113, Oxford University Press, Oxford. FINCKE, O.M. 1992c. Consequences of larval ecology for territoriality and reproductive success of a Neotropical damselfly. Ecology 73: 449-462. FINCKE, O.M. 1994. Population regulation of a tropical damselfly in the larval stage by food limitation, cannibalism, intraguild predation and habitat drying. Oecologia 100: 118-127. FINCKE, O.M. 1998. The population ecology of Megaloprepus coerulatus and its effect on species assemblages in water-filled tree holes. In: J.P. Dempster & I.F.G. McLean (eds), Insect Populations: In Theory and in Practice, pp. 391416, Chapman and Hall, London. FINCKE, O.M. 1999. Organization of predator assemblages in Neotropical tree holes: effects of abiotic factors and priority. Ecological Entomology 24: 13-23. FINCKE, O.M. & H. HADRYS. 2001. Unpredictable offspring survivorship in the damselfly Megaloprepus coerulatus shapes parental strategies, constrains sexual selection, and challenges traditional fitness estimates. Evolution 55: 653-664. FINCKE, O.M. & I. HEDSTRÖM. Differential forest use by predatory tree-hole damselflies (Pseudostigmatidae): implications for forest conversion. Studies on Neotropical Fauna and Environment (in review). FINCKE, O.M., S.P. YANOVIAK & D.R. HANSCHU. 1997. Predation by odonates depresses mosquito abundance in water-filled tree holes in Panama. Oecologia 112: 244-253. FOSTER, R.B. & N.V.L. BROKAW. 1982. Structure and history of the vegetation of Barro Colorado Island, In: E.G. Leigh, A.S. Rand & D.M. Windsor (eds), The Ecology of a Tropical Forest: Seasonal Rhythms and Long-term Changes, pp. 67-82, Smithsonian Institution Press, Washington D.C. FRANKIE, G.W. & A. MATA (eds.). 2004. Biodiversity Conservation in Costa Rica. University of California Press, Berkeley. FRANK, J.H. & L. P.LOUNIBOS (eds.). 1983. Phytotelmata: Terrestrial Plants as Hosts for Aquatic Insect Communities. Plexus Publishing, Inc. Medford. FUTUYMA, D.J. 1998. Evolutionary Biology, 3rd ed. Sinauer Associates, Sunderland.
CyanMagentaYellowBlack Odonata page 123
Use of Forest and Tree Species, and Dispersal by Giant Damselflies
123
GALLINDO, P., S.J. CARPENTER & H. TRAPIDO. 1955. A contribution to the ecology and biology of tree hole breeding mosquitoes of Panama. Annals of the Entomological Society of America 48: 158-164. GARRISON, R.W. 2004. New World Odonata List: A synonymic list of the New World Odonata. (http://www2.ups.edu/biology/museum/NewWorldOD.html) GENTRY, A.H. 1990. Floristic similarities and differences between southern Central America and upper and central Amazonia. In: A.H. Gentry (ed.), Four Neotropical Rainforests, pp. 141-157. Yale University Press, New Haven. GONZÁLEZ SORIANO, E. 1997. Odonata. In: E. González Soriano, R. Dirzo & R.C. Vogt (eds), Historia Natural de los Tuxtlas, pp. 245-255. Universidad Nacional Autónoma de México, México. GREENEY, H.F. 2001. The insects of plant-held waters: a review and bibliography. Journal of Tropical Ecology 17: 241-260. GROENEVELD, L.F., V. CLAUSNITZER & H. HADRYS. 2006. Convergent evolution of gigantism in damselflies of Africa and South America: evidence from nuclear and mitochondrial sequence data. Molecular Phylogenetics and Evolution (in press). HADRYS H., W. SCHROTH, B. STREIT, B. SCHIERWATER & O.M. FINCKE. 2005. Noninvasive isolation of polymorphic microsatellites from the neotropical damselfly Megaloprepus caerulatus: Use of tree hole odonates as environmental monitors in fragmented forests. Conservation Genetics 6: 481-483. HARTSHORN, G. & B. HAMMEL. 1994. An introduction to the flora and vegetation of La Selva, In: L.A McDade, K.S. Bawa, H.A. Hespenheide & G.S. Hartshorn (eds), La Selva: Ecology and Natural History of a Neotropical Rainforest, pp. 73-89, University of Chicago Press, Chicago. HAUGEN, L. 2002. Privation and uncertainty in the small nursery of Peruvian laughing frogs: larval ecology shapes the parental mating system. Ph.D. dissertation, University of Oklahoma, Norman. HEDSTRÖM I. & G. SAHLÉN. 2001. A key to the adult Costa Rican “helicopter” damselflies (Odonata, Pseudostigmatidae), with notes on their phenology and life zone preference. International Journal of Tropical Biology and Conservation 49: 1037-1056. HUBBELL, S.P. & R. B. FOSTER. 1983. Diversity of canopy trees in a neotropical florest and implications for the conservation of tropical trees. In: S.J. Sutton, T.C. Whitmore & A.C. Chadwick (eds), Tropical Rain Forest: Ecology and Management, pp. 25-41, Blackwell, Oxford. HUBBELL, S.P. & R. B. FOSTER. 1992. Short-term dynamics of a neotropical forest: why ecological research matters to tropical conservation and management. Oikos 63: 48-61. JONES, J. W., M. J. TURELL, M. R. SARDELIS, D. M. WATTS, R. E. COLEMAN, R. FERNANADEZ, F. CARBAJAL, J. E. PECOR, C. CALAMPA & T. A. KLEIN. 2004. Seasonal distribution, biology, and human attraction patterns of culicine mosquitoes (Diptera: Culicidae) in a forest near Puerto Almendras, Iquitos, Peru. Journal of Medical Entomology 41: 349-360. KITCHING, R.L. 2000. Food Webs and Container Habitats: the Natural History and Ecology of Phytotelmata. Cambridge University Press, Cambridge.
CyanMagentaYellowBlack Odonata page 124
124 Ola M. Fincke
LAESSLE, A.M. 1961. A microlimnological study of Jamaican bromeliads. Ecology 42: 499-517. LEHTINEN, R.M. 2004. Ecology and evolution of phytotem-breeding anurans. Miscellaneous Publications of the University of Michigan 193. LEIGH, E.G. JR. 1999. Tropical Forest Ecology: A View from Barro Colorado Island. Oxford University Press, Oxford. LEIGH, E.G. JR., A.S. RAND & D.M. WINDSOR (eds). 1982. The Ecology of a Tropical Forest: Seasonal Rhythms and Long-term Changes. Smithsonian Institution Press, Washington D.C. LENCIONI, F.A.A. 2005. Damselflies of Brazil: An Illustrated Identification Guide I: Non-Coenagrionidae Families. All Print Editora, São Paulo. LOUNIBOS, L.P. 1980. Larval Odonata in water-containing treeholes at the Kenya coast. Notulea Odonatologicae 1: 99-100. LOUTON, J, J. GELHAUS & R. BOUCHARD. 1996a. The aquatic macrofauna of waterfilled bamboo (Poaceae: Bambusoideae: Guadua) internodes in a Peruvian lowland tropical forest. Biotropica 28: 228-242. LOUTON, J.A., R.W. GARRISON & O.S. FLINT. 1996b. The Odonata of Parque Nacional Manu, Madre de Dios, Peru: Natural history, species richness and comparisons with other Peruvian sites. In: D.E. Wilson & A. Sandova (eds), The Biodiversity of Southeastern Peru, pp. 431-449, Smithsonian Institution Press, Washington, DC. MACHADO, A.B.M. & A. MARTINEZ. 1982. Oviposition by egg-throwing in a zygopteran, Mecistogaster jocaste Hagen, 1869 (Pseudostigmatidae). Odonatologica 11: 15-22. MELNYCHUK, M.C. & D.S. SRIVASTAVA. 2002. Abundance and vertical distribution of a bromeliad-dwelling zygopteran larva, Mecistogaster modesta, in a Costa Rican rainforest (Odonata: Pseudostigmatidae) with notes on their phenology and life zone preferences. International Journal of Odonatology 5: 81-97. ORR, A.G. 1994. Life histories and ecology of odonate breeding in phytotelmata in a Bornean rainforest. Odonatologica 23: 365-377. PECOR, J. E, J. JONES, M. TURELL, R. FERNANDEZ, F. CARBAJAL, M. O’GUINN, M. SARDELIS, D. WATTS, M. ZYZAK, C. CALAMPA & T. KLEIN. 2000. Annotated checklist of the mosquito species encountered during arboviral studies in Iquitos, Peru. Journal of the American Mosquito Control Association 16: 210-218. PIMM, S. & P. RAVEN. 2000. Extinction by numbers. Nature 403: 843-858. RAND, A.S. & W.M. RAND. 1982. Variation in rainfall on Barro Colorado Island. In: E.G. Leigh, A.S. Rand & D.M. Windsor (eds), The Ecology of a Tropical Forest: Seasonal Rhythms and Long-term Changes, pp. 47-59, Smithsonian Institution Press: Washington, D.C. RICHARDSON. B. A. 1999. The bromeliad microcosm and the assessment of faunal diversity in a Neotropical forest. Biotropica 31: 321-336. RÖDEL, M.-O., V.H.W. RUDOLF, S. FROHSCHAMMER & K.E. LINSENMAIR. 2004. Life history of a West African tree-hole breeding frog, Phrynobatrachus guineensis GUIBÉ & Lamotte, 1961 (Amphibia: Anura: Petropedetidae). In: R.M. Lehtinen (ed.), Ecology and evolution of phytotelm-breeding anurans.
CyanMagentaYellowBlack Odonata page 125
Use of Forest and Tree Species, and Dispersal by Giant Damselflies
125
Miscellaneous Publications of the Museum of Zoology, pp. 31-44, University of Michigan, Ann Arbor, No. 193. RÜPPELL, G. & O.M. FINCKE. 1989. Megaloprepus coerulatus (Pseudostigmatidae) Flug- und Fortpflanzungs verhalten (Flying and reproductive behaviour). Publikationen Zu Wissenschaftlichen Filmen Sektion Biologie, Serie 20, Nr 10/E 2976. SANTOS, N.D. 1981. Odonata. In: S.H. Hulbert, G. Rodriquez & N.D. Santos (eds), Aquatic biota of tropical South America part 1: Arthropoda, pp. 64-85, San Diego State University, San Diego. SRIVASTAVA D.S., M.C. MELNYCHUK & J.T. NGAI. 2005. Landscape variation in the larval density of a bromeliad-dwelling zygopteran, Mecistogaster modesta (Odonata: Pseudostigmatidae). International Journal of Odonatology 8: 67-79. STOUT, J. 1983. Megaloprepus and Mecistogaster (Gallito Azul, Helicopter Damselfly), In: D.H. Janzen (ed.), Costa Rican Natural History, pp. 734-735, University of Chicago Press, Chicago. SUMMERS, K. 1989. Sexual selection and intra-female competition in the green poison-dart frog. Behavioral Ecology and Sociobiology 27: 307-13. THEILER, M. & W.G. DOWNS. 1973. The Arthropod-borne Viruses of Vertebrates. Yale University Press, New Haven. WIKELSKI, M., D. MOSKOWITZ, J.S. ADELMAN, J. COCHRAN, D.S. WILCOVE & M.L.MAY. 2006. Simple rules guide dragonfly migration. Biology Letters doi:10.1098/rsbl.2006.0487. YANOVIAK, S.P. 1999. Community structure in water-filled tree holes of Panama: effects of hole height and size. Selbyana 20: 106-115. YANOVIAK, S.P. 2001. The macrofauna of water-filled tree holes on Barro Colorado Island, Panama. Biotropica 33: 110-120. YANOVIAK, S.P. & O.M. FINCKE. 2005. Sampling methods for water-filled tree holes and their analogues. In: S.R. Leather (ed.), Insect Sampling in Forest Ecosystems, pp. 168-185, Blackwell Publishing, London. YANOVIAK, S.P., L.P. LOUNIBOS & S.C. WEAVER. 2006. Land use affects macroinvertebrate community composition in phytotelmata in the Peruvian Amazon. Annals of the Entomological Society of America (in press). YOUNG, A. M. 1980. Feeding and oviposition in the giant tropical damselfly Megaloprepus coerulatus (Drury) in Costa Rica. Biotropica 12: 237-239.
CyanMagentaYellowBlack Odonata page 126
This page intentionally left blank
CyanMagentaYellowBlack Odonata page 127
How can forest influence species composition, diversity and speciation in tropical Odonata? Adolfo Cordero Rivera (ed) 2006 Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 127-151.
127
© Pensoft Publishers
Sofia–Moscow
Thoughts from Africa: how can forest influence species composition, diversity and speciation in tropical Odonata? Klaas-Douwe B. Dijkstra & Viola Clausnitzer National Museum of Natural History Naturalis, PO Box 9517, NL-2300 RA Leiden, The Netherlands.
[email protected] Gräfestr. 17, 06110 Halle/Saale, Germany.
[email protected]
ABSTRACT We introduce tropical African forests and their Odonata, and speculate how climatic oscillations and associated large-scale habitat shifts may have governed speciation across the forest-savanna ecotone, presenting several hypothetical scenarios. Ecological traits of forest species and possible reasons for their disappearance when forest is opened up are discussed. We believe that low insolation in forest habitats and interspecific competition are key factors segregating forest and non-forest species. While openland species cannot cope with low insolation inside the forest, forest species have evolved a slow lifestyle to cope with the forest environment, but are out-competed by more aggressive non-forest species beyond forest borders. Casual field observations support this hypothesis, although the reality is likely to be more complex. Phylogenetic reconstruction of groups that straddle the habitat divide, linked to ecological observations, may elucidate evolutionary reactions to landscape change. The reaction of odonate assemblages to forest loss is studied easily in Africa’s imperilled forests. Because many of these forests are believed to be relatively young and highly forest-adapted species may have very low dispersal capacities, comparative ecological research of ‘forest-dependent’ odonate assemblages inside and outside ancient forest refugia is recommended. Key words: Afrotropical, dragonflies, forest, competition, speciation, biodiversity, biogeography
CyanMagentaYellowBlack Odonata page 128
128 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
INTRODUCTION The mechanisms which keep odonate species inside or outside forests are essential for understanding present-day distribution patterns and speciation in relation to historical landscape change. Knowledge of these mechanisms in tropical dragonflies is limited. We aim to discuss characteristics of Odonata and their habitats which may influence their distribution in the landscape (species assemblages, distribution patterns) and their development in time (speciation). Our basis is largely anecdotal, relying on our experience in tropical continental Africa, rather than on quantitative data. This paper must therefore be read as an essay of ideas, rather than an in-depth analysis of the available literature and data. Observations which we believe are supportive of our ideas are presented in text boxes. We investigate four related issues: 1 What is the geographic and historical setting of Africa’s forest odonate diversity? 2. Which characteristics of forest odonates do we observe at a larger scale of time and space? Their diversity and distribution is governed by geography and history. How might patterns be explained and where did the present diversity originate? 3. Which patterns do we observe at a smaller scale? Deforestation and the replacement of indigenous vegetation by exotic species offer a means to test the role of habitat structure. Differential dispersal capacities or environmental tolerances of species further affect assemblages, providing additional ‘natural experiments’ which help assess interspecific effects of assemblages. Which factors determine the composition of these assemblages? 4. How can formulated hypotheses be implemented in research? The importance of forest conservation is not addressed explicitly, but becomes obvious in the context. An overview on conservation issues of African forests is given by Clausnitzer (2003b, 2004a). African forests The term ‘forest’ is often used indiscriminately to describe dense stands of trees. White (1983, p. 44-55) defines forest, as opposed to woodland, as closedcanopy stands with a more or less complete shading of the ground, hindering the development of a grass layer (see also Clarke 2000). In considering forests of continental tropical Africa, two major categories can be distinguished, which are also represented by the distribution of forest odonates (Fig. 1). Central and
CyanMagentaYellowBlack Odonata page 129
How can forest influence species composition, diversity and speciation in tropical Odonata?
129
western Africa are dominated by Guineo-Congolian lowland rainforest, which is almost continuous from Uganda to Sierra Leone (Figs. 2, 3), and which has a broad transition (mainly of riverine forests) into peripheral areas. Eastern and southern Africa harbour diverse but fragmented forests, restricted to areas of higher precipitation, especially at greater elevations but also on the coast. The former is often referred to as the Afromontane archipelago (Fig. 4). The history of tropical Africa is governed by strong climatic oscillations with associated habitat shifts across a relatively continuous land surface. “Climate is always changing, but fluctuations have been particularly marked in tropical Africa […] during the 2.43 million years which have passed since the first major glaciation in the northern hemisphere” (Hamilton 1992). The Quarternary has seen a general trend towards increasing aridity and more pronounced oscillations. The severest recent forest contraction was 21-14,000 BP, at the height of the last major world glaciation, restricting tropical forest in Africa to a few relatively small refugia, the latest maximum extent of forest was 8-7,000 BP (Hamilton 1992).
+ up to three Chlorocnemis species
up to seven Chlorocnemis species
up to three Chlorocnemis species
+
+ ++
A. leucosticta
Fig. 1. Distribution of Chlorocnemis and Allocnemis leucosticta, an Afro-endemic group restricted to shaded forest habitat. Chlorocnemis (including Isomecocnemis) is widespread in the continuous forests of central and western Africa, with up to three species occurring together, although up to seven species co-exist in the Cameroon highlands. C. abbotti (squares), C. montana (circles) and C. marshalli (triangles) represent the genus in the Afromontane archipelago, the related A. leucosticta replaces it in the southernmost extent of the archipelago. Afromontane sites where representatives are apparently absent are marked with crosses.
CyanMagentaYellowBlack Odonata page 130
130 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
Fig. 2. Swampy rainforest with natural glade, Budongo Forest, Viola Clausnitzer, April 1999.
Fig. 3. Rainforest of the Congo Basin, Klaas-Douwe B. Dijkstra, November 2004.
African forest Odonata As elsewhere in the tropics, the greatest diversity of Odonata in Africa is in its forests (Boxes 1-2). Most publications on Afrotropical dragonflies have a taxonomic or regional focus, giving at most only a little general habitat information (e.g. Pinhey 1970; Consiglio 1978; O’Neill & Paulson 2001). Few authors have provided more specific habitat requirements (e.g. Neville 1960; Pinhey 1984; Legrand & Couturier 1985; Lempert 1988; Miller 1993, 1995; Clausnitzer 1999; Vick 1999) and even fewer have attempted to quantify the available Fig. 4. Afromontane forest, Mt. Kenya, 2200 m, Viola Clausnitzer, March 1993.
CyanMagentaYellowBlack Odonata page 131
How can forest influence species composition, diversity and speciation in tropical Odonata?
131
information (Clausnitzer 2003a; Dijkstra & Lempert 2003). The latter task is difficult because the data are heterogeneous and often apply only to fractions of species’ ranges. There are barely any studies on Afrotropical odonate larval assemblages and none comparing forest and non-forest habitats.
Box 1. Diversity of Odonata in subsaharan African
85 93
42
76
19
58
59 74
88
103
70 195 198 208
219
229 226
261
10 8
211 221
283 219
223
227
257
64 175
323 195 218 168 15 7
240 166 135
174
155
161
Fig. 5. Figures indicate the estimated number of species in each country, based on literature and our own data; doubtful literature records were omitted. Data are of varying quality and therefore records were interpolated: ‘gaps’ between countries where species occurred were filled, e.g. a species recorded from Ghana, Benin and Cameroon was also added to Nigeria and Togo. This may ‘over-correct’ species with truly disjunct distributions, but we believe this effect is minimal. Some smaller states were omitted for lack of data. Similarity of country faunas was established with a cluster analysis (distance measure: Sorensen; linkage method: group average). The most dissimilar groups of countries are separated by solid lines, less dissimilar groups by dashed lines, revealing three main regions of diversity: Madagascar and the southern and eastern countries are both regions where national diversity generally lies between 100 and 200 species. The greatest diversity (generally over 200 species per country) is found in the western and central countries (compare Fig. 1). The Sahel countries have impoverished faunas, mostly with fewer than 100 species per nation. In northern Africa (largely not on map) Afrotropical influence decreases further and Palaearctic species predominate.
CyanMagentaYellowBlack Odonata page 132
132 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
Box 2. Correlations between distribution and habitat preference in eastern African dragonflies G4
100
50 S.fo
Axis 2
A8
A.lo
G5
N5
A.losu
0
E5
A7
N6 P4
E6 G6
P7
-50 E4
-100 -100
-50
0 Axis 1
50
100
Fig. 6. The distributions and habitat requirements of 467 species of eastern African dragonflies (the area from Sudan to Zimbabwe, east of 22°E) were analysed to assess general correlations of biogeography and ecology. The information on habitat requirements is based on literature surveys (mainly Pinhey 1984; Lempert 1988; Vick 1999) and personal experience. Since this is unsatisfactory for many species, we have used a coarse classification of habitats and ranges. Habitat parameters used were current (running; standing; 2), insolation (shady; sunny; 3), size (small; large; 1), aquatic vegetation (present; absent; 1), landscape (forest; open; 3) and altitude (lowland; montane; 4), extreme values and the number of intermediate value recognised are indicated in brackets. Distribution parameters used to categorise the ranges of eastern African dragonflies are provided in the figure legend. Only one value can be assigned to each species for each parameter. A Detrended Correspondence Analysis (DCA, Hill & Gauch 1980) was performed with distribution as main matrix and habitat parameters as secondary matrix. The species distributions were mainly explained by the habitat parameters forest cover and altitude, although even these vectors were not strong (Axis 1 describes 26% and Axis 2 11% of the variance). Most range types grouped together along the lowland and submontane vector. This is not surprising, as few African dragonfly species are montane. The only disparate range types were G4, characterised by numerous western
CyanMagentaYellowBlack Odonata page 133
How can forest influence species composition, diversity and speciation in tropical Odonata?
133
and central Africa forest species confined to the Congo basin in eastern Africa, and E4, which contains endemics of the highlands and coastal forest of eastern Africa. A Mantel Test was performed to test for the similarity of the distribution and habitat parameter matrices. The result (randomisation, Monte Carlo Test) shows a significant positive association (p=0.001) between the two matrices (observed z slightly greater than average z from randomised runs), but still the similarity is quite low (r=0.08026). Legend. – S.fo: forest cover; A.lo: lowland (0-1000 m a.s.l.); A.losu: lowland and submontane (0-2000 m a.s.l.); A: widespread in Afrotropics (A7: endemic; A8: not endemic); peripheral, predominantly with Asian affinities (P4: Sahara and Arabia; P7: widespread in Asia, just reaching Africa); E: eastern and southern Africa (E4: Angola, NE South Africa to SE Kenya; E5: as previous and to Uganda and E DRC; E6: as previous and to Ethiopia and Sudan); G: centred on Guineo-Congolian biome (G4: western Africa to Congo Basin; G5: as previous and to C Uganda and W Kenya; G6: as previous and to W Ethiopia and N Malawi); N: northern and western Africa (N5: W Africa to NE Congo; N6: as previous and to N Uganda, NW Kenya and W Ethiopia). Values. – r² = 0.05, vector scaling = 500%, total variance (‘inertia’) in the species data: 11.00%.
BIOGEOGRAPHY OF AFROTROPICAL FOREST ODONATE DIVERSITY Although extinctions may have resulted from fluctuations in forest cover and forest types related to climatic change, climatic fluctuations must also have created opportunities for rapid evolution for some forest organisms. Climatic change will have resulted in repeated isolation and connection of populations of some species. Speculation on patterns of speciation related to forest history suggests that some groups […] have had complicated histories and it can be difficult to describe evolutionary connections of modern taxa from their present distribution and morphological similarities alone. This is a field in which considerable progress will soon be made with the application of […] DNA analyses. These words by Hamilton (1992) neatly summarize the scenarios and research opportunities for Afrotropical forest biogeography. Although the greatest diversity of African Odonata is in forests (see Box 1, Fig. 5), the fauna is impoverished in comparison to tropical America and Asia. The small and isolated African highlands functioned poorly as forest refugia during drier periods, unlike the long mountain chains of South America and southern
CyanMagentaYellowBlack Odonata page 134
134 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
Asia, while the proximity of many forests to the sea moderated aridity in tropical Asia. Thus climate-induced habitat shifts posed a greater threat to African forest biota, demanding a greater tolerance to change. Scenarios of the origin and diversification of African Odonata are still wholly speculative. Nonetheless several patterns are apparent that offer some clues to the history of today’s diversity and an incentive for future investigation. History of diversity: old and new Africans Statements on the origin and age of the Afrotropical forest Odonata can only be made by comparing the continental and Madagascan faunas with each other and with tropical faunas elsewhere. Especially Madagascar can be considered as a ‘time capsule’ that provides an impression of Africa’s ancient forest odonates, because it was separated from the mainland long ago and has harboured rainforest ever since. The best examples are endemic Afrotropical taxa conserved both here and in climatically relatively stable areas on the continent. Such probably ‘old Africans’ are Nesolestes (also found in the Cameroon highlands together with the closely related Neurolestes), Metacnemis (also in Cape region), phyllogomphine Gomphidae (Isomma on Madagascar, Phyllogomphus in equatorial Africa and Ceratogomphus in southern Africa) and libellulid genera like Malgassophlebia and Neodythemis (both also equatorial Africa). Notable is the impoverishment of ‘ancestral’ families, such as Megapodagrionidae (Fig. 7). Synlestidae are largely confined to South Africa (Chlorolestes, Ecchlorolestes), and Corduliidae (excluding macromiines and Hemicordulia) are represented only by Idomacromia and Neophya in equatorial Africa, Libellulosoma and Nesocordulia in Madagascar and Syncordulia in South Africa. The isolated presence of Nubiolestes and Pentaphlebia in the Cameroon highlands and Coryphagrion on the East Coast, suggest that the otherwise exclusively neotropical Perilestidae, rimanelline Amphipterygidae and Pseudostigmatidae were once more widespread in Africa (Figs 7, 8). Of mysterious relations are several monotypic genera now placed in Megapodagrionidae and Platycnemididae with small, probably relict ranges: Amanipodagrion (Usambara Mts), Leptocnemis (Seychelles), Oreocnemis (Mt Mulanje in Malawi), Paracnemis (Madagascar) and Stenocnemis (Cameroon highlands). One can only speculate if obligate rainforest dwelling families like Platystictidae ever inhabited Africa. None of the ‘old Africans’ are nowadays dominant in the forests of continental Africa. The ‘new Africans’ are in families that are absent on Madagascar, or only represented by one adaptable species that probably colonised the island recently from the mainland. Examples are Calopterygidae (e.g. Umma), Chlorocyphidae (Chlorocypha), Protoneuridae (Elattoneura), lindeniine Gomphidae (Diastatomma) and macromiine Corduliidae (Phyllomacromia). The genera Platycnemis and Pseudagrion are important elements in
CyanMagentaYellowBlack Odonata page 135
How can forest influence species composition, diversity and speciation in tropical Odonata?
135
Ne’olestes (2) Amphipterygidae (2) + Perilestidae (1)
Allolestes (1) Amanipodagrion (1) Pseudostigmatidae (1)
?
Ne’olestes (16)
Tatocnemis (10) + Protolestes (8) Synlestidae (9)
Fig. 7. Approximate distribution of relict damselfly families in tropical Africa, indicating probable centres of climatic stability. The number of species in each group/area is indicated. Genera classified in Megapodagrionidae are marked with a black line, other families with grey shading. Nesolestes and Neurolestes (= Ne’olestes; indicated with broken line) are combined because they are closely related.
Pentaphlebia stahli
Neurolestes trinervis
Coryphagrion grandis
Amanipodagrion gilliesi
Nubiolestes diotima
Oreocnemis phoenix
Chlorolestes elegans
Stenocnemis pachystigma
Fig. 8. Male appendages of eight relict Afrotropical damselflies in dorsal view. Classified in six different families (see text), all share the plesiomorphic focipate cerci.
CyanMagentaYellowBlack Odonata page 136
136 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
both the forests of Madagascar and continental Africa, but each group is highly distinctive and does not appear closely related (Fig. 9). All mentioned continental groups are related to more diverse faunas in tropical Asia and are characterised by many relatively similar species. This and their absence on Madagascar suggest rapid radiations after a comparatively recent arrival from the east. A few of these arrivals were probably trans-oceanic. Hemicordulia and Teinobasis are dominant groups in (parts of) Indonesia, New Guinea and Australia, which hardly occur on the Asian mainland, but range across to eastern Africa through the Seychelles, Mascarenes and Madagascar (e.g. Clausnitzer 2003c; see also Fig. 9). A similar scenario may apply to the bispinagroup of Gynacantha (Dijkstra 2005). What is the age of these faunas? The oldest fossils of typical Anisoptera and Zygoptera date from around the Triassic-Jurrasic boundary, 210 million years ago (Mya). Neotropic affinities may date back to the Cretaceous, before Africa and South America separated 100 Mya (Goldblatt 1993). Madagascar “apparently slid south along the east African coast for most of the Cretaceous
sikassoensis sikassoensis sikassoensis
guttifera rufipes
?
congolensis
?
?
nyansana
?
? spec. nov.
melana agrioides o ot other insular species
Fig. 9. Approximate distributions of Afrotropical Platycnemis species; insular (black) and continental (grey) groups. Especially the general southern limits and contact zones of continental species are unclear; range P. sikassoensis (at least partly) incorporates ranges of other continental species. The continental group is remarkably similar to the Oriental Copera, but the relationship with the insular fauna appears more distant. Transoceanic dispersal over hundreds of kilometres is the only plausible scenario for the presence of insular species on the Comores and Pemba. From: Dijkstra et al. (in press).
CyanMagentaYellowBlack Odonata page 137
How can forest influence species composition, diversity and speciation in tropical Odonata?
137
rather freely interchanging plant and animal taxa with the mainland at least until 90 Mya and perhaps intermittently thereafter” (Gentry 1993). The temperate faunas of Patagonia and Australia were still linked through Antarctica during the Paleocene and Eocene (65-35 Mya), but Africa probably separated much earlier. In conclusion, the ‘old Africans’ are probably of mesozoic age. The ‘new Africans’ must be largely cenozoic, expanding between the separation of Madagascar and the end of the trans-African rainforest belt and the tropical connection with Asia. The latter was probably in the Oligocene or Miocene, 25-5 Mya. The current fauna was probably largely shaped by strong climatic fluctuations of the Quaternary, roughly in the last two million years. Trans-oceanic colonisations probably also took place so recently. In conclusion, although Odonata are an ancient group of insects, the Afrotropical fauna is relatively young with a ‘broad but shallow’ diversity, being rich in species but poor at higher taxonomic levels (especially families). Instability favours adaptable taxa, but eliminates those which adapt slowly. Extinctions may have created ecological ‘vacuums’ into which adaptable taxa radiated during the more favourable (i.e. hot and wet) periods that followed. This may explain notably speciose genera like Chlorocypha, Pseudagrion, Elattoneura, Phyllomacromia, Orthetrum and Trithemis, and the dominance of Coenagrionidae and Libellulidae. The two families are considered as the largest and evolutionary most advanced in Odonata, and are adapted better to temporary conditions than any other. Geography of speciation The image of forest species as ‘habitat hermits’ —poorly dispersing specialists that are confined by the limits of their ancestral habitat (see below)— implies that their distribution and evolution is governed by the geography and history of their forest home. Present-day diversity suggests that habitat change not only leads to extinctions, but also to adaptation and ultimately speciation. How did these species evolve and where? Did savanna species evolve from forest species, or vice versa? How may a climatic cycle drive speciation? A rainfall decrease will result in the gradual contraction of wet habitat (e.g. forest streams) and the corresponding intensification of selection on adaptations for a drier habitat (e.g. faster larval development), especially at the periphery. Peripheral habitats become fragmented, with genetic isolation of dry-adapted populations and the possible rise of new species. While at the height of an arid period wetadapted species are restricted to wet refugia, where they speciate in allopatry (see Fig. 11), dry-adapted species can expand. A rainfall increase will induce the expansion of wet habitat and fragmentation of dry habitat, and further speciation of the dry-adapted species may then take place. Moreover it may induce a ‘reverse scenario’, in which non-forest populations adapt to increas-
CyanMagentaYellowBlack Odonata page 138
138 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
ingly wet circumstances, becoming isolated in open enclaves in a forested world. A succession of wet and dry periods may function as a ‘species pump’. What sort of distributional patterns does this scenario predict? For instance, might broader or more structured gradients between forest and open habitats generate and conserve more species? That could explain the greater odonatological richness south of the rainforest belt than north of it. The wet, large, central and connected Congo Basin is an obvious focus for odonate diversity (Figs 5, 10). Ever-shifting rivers, swamps, forests, woodlands and savannas, created a mosaic of habitats in time and space. With its position in the heart of the continent, the basin has always been on a cross-roads, between the forests to the west and east in wet periods, and between the savannas to the north and south in drier times. Moreover, sensitive species could survive in forest refugia west and east of the basin and along its rivers. Kingdon (1989) postulated that the basin is an “evolutionary whirlpool” of species
Porpax asperipes
Porpax asperipes, P. garambensis & P. sentipes
Fig. 10. Distribution of Porpax species. Two species are endemic to the Congo Basin, a third extends to the Lower Guinea. P. bipunctus (black squares) occurs in four disjunct populations, which coincide with important rainforest refugia. This is the clearest example of refugial disjunction found in Afrotropical Odonata so far. The pattern is supported by distinctive coloration in each region (suggesting allopatric speciation in progress) and the absence of intermittent records. Other Porpax species have been collected more widely and are geographically uniform. The species’ isolation may be linked to the ephemeral nature (probably flooded areas in stream beds) of its reproductive habitat, confining it to areas with perennial and predictable rainfall. P. risi (open squares) inhabits highland swamps, as is shown by its archipelago-like distribution. From: Dijkstra (2006).
CyanMagentaYellowBlack Odonata page 139
How can forest influence species composition, diversity and speciation in tropical Odonata?
139
evolution, conservation and dispersal, leading to high endemism and diversity. This should be especially true for Odonata, because of their strong ties to freshwater and vegetation structure. Endemic species in the eastern African highlands with affinities to the Guineo-Congolian rainforests (Umma declivium, Chlorocnemis abbotti, Nepogomphoides stuhlmanni, Micromacromia miraculosa) are testimony of former links between these areas. While the ‘oscillating landscape’ has stranded these species on forest islands, the reverse may apply to open-land dragonflies ‘trapped’ in forest. Several widespread non-forest species are represented in the Congo Basin by larger and darker forms or sister species: the dark forms of Gomphidia bredoi, G. quarrei (Müller et al. 2005) and Gynacantha manderica (Dijkstra 2005), Trithemis congolica (sister species of T. aconita; see Box 3, Fig. 11) and Zygonoides occidentis (sister species of Z. fraseri and Z. fuelleborni). These populations were apparently sufficiently isolated from the periphery and interconnected with each other to develop, but their ecology remains unknown. Many of the dominant African genera have species exclusively in either forest, savanna or intermediate woodland habitats. Box 3 presents two possible scenarios of how a (group of) non-forest species may have radiated out of an ancestral rainforest stream habitat, but perhaps most intriguing is the
Box 3. Hypotheses of speciation
Fig. 11. Each map shows the distribution of a morphologically uniform subgroup of the basitincta-group of Trithemis, which appear as concentric layers focused on a forest core. All species inhabit running waters, but their shadiness differs between subgroups. Adaptation and range expansion during two or more phases of forest regression may have given rise to the two outer layers, while allopatric speciation in forest fragments could have split the pairs in the first two subgroups. Forest expansion may have separated the fourth pair and isolated T. congolica from T. aconita in open pockets in the Congo Basin. Endemic to Príncipe, T. nigra represents the dispersal of (something near) T. aconita to this volcanic island. Its smaller size, wing shape (less suitable for sustained flight) and bold yellow and black coloration (replacing pruinosity) are adaptations to its insular rainforest environment.
CyanMagentaYellowBlack Odonata page 140
140 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
id
it y
pe tit ion ?
ar
co m
r x
p
r
aa p
p
p
fo re st
r
a a
r
i i
ar id it y
f f
co
ld
f f
f
f
f
f
Fig. 12. Compared with their Oriental relatives, the over 40 species of Afrotropical Chlorocyphidae are uniform in their venation, morphology and the lack of coloured wings. This and their absence on Madagascar suggests that they diversified rapidly quite recently from Asian stock. While lowland forest streams appear to be the ancestral habitat of Chlorocyphidae, Platycypha species are more extreme: P. fitzsimonsi and P. amboniensis inhabit submontane streams, P. caligata exposed rivers and even lakeshores. The genus also possesses an especially elaborate set of colour signals, most notably their expanded legs and strong colour contrasts (Figs 13-14). Perhaps Platycypha developed from Chlorocypha-like stock that extended east when tropical forest
Fig. 13. Platycypha undescribed species, K.-D.B. Dijkstra.
Fig. 14. Platycypha caligata, K.-D.B. Dijkstra.
CyanMagentaYellowBlack Odonata page 141
How can forest influence species composition, diversity and speciation in tropical Odonata?
141
was continuous, but when forest shrunk became fragmented into several relict species in eastern African lowland forest. Some members spread south to the Cape, possibly becoming isolated in highland forests as the climate became relatively dry or hot. Change along temporal, altitudinal and climatic gradients may have favoured selection for tolerance to non-forested habitats. The currently most abundant eastern African chlorocyphid, P. caligata, may be the epitome of that development, which spread more than any other in a ‘vacuum’ of potential habitat. It has expanded almost as far as its ecology allows, bounded by unsuitable habitat on all sides of its range. It barely overlaps with other chlorocyphids and these are confined to more sheltered habitats where they co-occur (e.g. C. consueta). The only ‘escape’ is the savanna corridor to the west, but perhaps there competition with C. curta is limiting. The relict population of C. curta in Sudan’s Jebel Marra (Dumont 1988) shows that this species is equally adapted to non-forest habitats and that the ‘front’ against P. caligata was once broader. Being adapted to rather extreme conditions, P. caligata was the first and only chlorocyphid to colonise lakeshores. This scenario demonstrates a remarkable parallel with mankind: in geologically recent times, an ‘enhanced’ savanna species developed from a forest group and conquered an unprecedented habitat array. Interestingly, here too evolution is paired with a strong development of sexual characters. Legend. – Open circles: relict species of lowland occurrence (A: P. auripes; P: P. picta; R: P. rufitibia; X: undescribed species); black circles: relict species of submontane occurrence (A: P. amboniensis; F: P. fitzsimonsi; P: P. pinheyi); squares: P. lacustris; dark shading: P. caligata (possible factors limiting its spread westwards indicated); triangle: morphologically deviant lacustrine population of P. caligata: paler shading: main range of other Chlorocyphidae (areas with 2 or more species), mostly Chlorocypha, outside that of Platycypha; grey line: probable extreme northern limit of Chlorocypha (including isolate of C. curta in W Sudan).
evolution of Pseudagrion, the largest odonate genus in Africa and one of the largest in the world. Some 140 species are known, two-thirds occur in Africa and Madagascar, the remainder ranges across southern Asia into Australia. The genus has occupied all freshwater habitats in tropical Africa, dominating damselfly communities from pools in the Kalahari to alpine streams on the Kilimanjaro. Diverse assemblages inhabit equatorial rainforests, while relict populations survive in the Saharan mountains, Morocco and the Levant. In Africa, the genus is subdivided on morphological and ecological grounds: the A- and B-group dominate on the continent, while Madagascar hosts a third group (Pinhey 1964). The dark-bodied A-group generally inhabits cooler habitats: mostly running waters, often shaded or montane. It includes many rainforest species with small ranges, confined to deeply shaded habitats. Some
CyanMagentaYellowBlack Odonata page 142
142 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
species develop extensive pruinosity on the body, and occur in more open habitats. The reflective pruinosity may be an adaptation to increased insolation. The latter category includes species with restricted ranges at considerable altitude, but also P. kersteni, the most widespread and strongly pruinose species. The pale-bodied B-group generally inhabits rather hot habitats: running and standing waters, sunny and often exposed, generally at low altitudes. Possibly forest streams are the ancestral habitat of Pseudagrion and the A- and B-groups diversified separately in non-forest habitats, as these expanded during periods of forest regression. Judging from similarities with the Madagascar group, the presence of the A-group in tropical Africa is relatively ancient. Perhaps the pruinose species evolved in highlands and were pre-adapted to invade open lowland habitats, such as grassland streams. The B-group possibly arrived later (from Asia?) and radiated into warmer habitats left unoccupied by A-group members.
FOREST AND THE ECOLOGY OF SPECIES ASSEMBLAGES Habitat selection is strongly dependent on structural characteristics, like shading and vegetation structure. […] Numerous interrelated variables determine the characteristics of [river] habitats from the source to the mouth […] and therefore a strong turnover of odonate species and assemblages is expected as one goes downstream. Similar change may also be expected as human disturbance increases. Deforestation and damming, for instance, will result in a more open character of running water sites. Associated changes may be a decrease of flow velocity and permanence of water, and an increase of silt load. This may result in a habitat shift and the accompanying change of the odonate assemblage. A dammed forest stream, for instance, could result in a sunny, slow-flowing stretch of water with species typical of a small river. A river that has been cleared of forest, resulting in increased erosion and an irregular discharge of water, may lead to a fauna of temporary pools. This prediction by Dijkstra & Lempert (2003) for West African rainforest, characterizes how the relation of forest dragonflies and their habitat may be perceived. Many forest species appear to be stenotopic and sensitive to disturbance. As discussed, this may have a profound influence on their speciation and diversity. But why are some Odonata restricted to forest? The decrease of shading is an obvious change for odonate assemblages as forest is opened up, but other factors must also have an impact. Interactions between species, by predation and competition, further complicate assemblage composition (Suhling & Lepkojus 2001). All factors must interact, but only shadiness is discussed further here, as it seems to determine odonate assemblages most strongly.
CyanMagentaYellowBlack Odonata page 143
How can forest influence species composition, diversity and speciation in tropical Odonata?
143
Forest structure and species assemblages Clausnitzer (2003a) demonstrated that as coastal forests in Kenya deteriorated, habitats were colonized by widespread and generalistic dragonflies. Although this resulted in a species increase per site, the localized specialists disappeared. This agrees with studies on amphibian communities in pristine and disturbed forests in Madagascar. Although the diversity and abundance were not significantly affected by selective logging, Vallan et al. (2004, p. 416) concluded that “species typical of pristine rainforests (specialists) are […] replaced by species adapted to secondary habitats (often generalists).” To the human observer, the balance of shade and sunlight (degree of shadiness) is an obvious habitat characteristic, that coincides with a marked faunal break between streams and rivers (Dijkstra & Lempert 2003), and forested and deforested sites (Clausnitzer 2003a). The importance of this balance is illustrated by the ‘shade paradox’: in South Africa Phaon iridipennis, Pseudagrion sjoestedti and Trithemis aconita are characteristic of heavily shaded parts of the Sabie River, while they favour open sites in West Africa (Clark & Samways 1996, Dijkstra & Lempert 2003). This can be explained by the environmental context. In the highly exposed South African savanna these species can perhaps only cope under the most forest-like conditions, the reverse applies in the West African rainforest. “The structure and appearance (the “architecture”) of the plants or plant communities rather than individual plant species are likely to serve as cues for biotope and habitat recognition” (Corbet 1999, p. 13). The degree of shadiness seems to be the principal cue for dragonflies to select a forest habitat, but the forest type (species composition) and “the age of the forest bordering running water sites is of little influence for the odonate assemblage there, as long as the required cover is present” (Dijkstra & Lempert 2003, p. 409). The importance of structure can be seen in the response of assemblages to the replacement of indigenous vegetation by exotic species. Kinvig & Samways (2000) concluded that for Odonata in South African Pinus groves “it does not matter whether the trees are exotic or indigenous, […] but whether the architecture permits the right combination of sunlight and shade. This right combination is necessary for thermal balance and for encouraging low bushes under the tree canopy, which are necessary for perching and ovipositing.” This was highlighted by the response of two endemic Chlorolestes species to exotic wattles Acacia mearnsii overgrowing streams. The shade-seeking C. fasciatus is abundant at such sites, while the sun-loving C. apricans disappears. Observations in East Africa supplement the hypothesis that if general habitat structure remains unaltered, dragonflies tolerate a degree of exotic vegetation, but if these severely alter habitat architecture, this can have a detrimental effect on dragonfly populations (Box 4).
CyanMagentaYellowBlack Odonata page 144
144 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
Box 4. Habitat structure change and specialized species. Teinobasis alluaudi is known from the Seychelles, Madagascar, Zanzibar, Pemba and three localities in mainland Africa, where it inhabits seasonal swamp forest with a dense pinnatifolious understorey. The habitat on Zanzibar has been largely transformed into exotic plantations, mainly of Teak (Tectona grandis), but T. alluaudi was found in huge numbers in the dense fern thickets in the undergrowth (Clausnitzer 2003c). The treehole-breeder Hadrothemis camarensis (Fig. 15) oviposits into introduced bamboo in Uganda (Corbet 1983). Coryphagrion grandis larvae were found in nearly every water-filled stump of exotic bamboo in the Amani Botanical Garden in Tanzania. The species inhabits pristine coastal forests of Kenya and Tanzania, but also coastal thickets and forests where all large overstorey trees have been logged: dense, shady vegetation with phytotelmata (even half coconut husks lying on the ground) are the most important habitat attributes for C. grandis (obs. & Clausnitzer 2004b). Oreocnemis phoenix (Fig. 16) is endemic to the isolated Mulanje plateau in S Malawi, having been found both in forest and open habitats at 82% of the 51 sites investigated, with an average density of 5.8 individuals per 100 m stream. Of six sites surveyed in pine plantations only two males were present at a single 100 m stretch. Unlike natural streams, plantation streams sometimes contained algae and silt, or were clogged up with wood and pine needles (obs. K.-D.B. Dijkstra).
Fig. 15. Hadrothemis camarensis, D. Motshagen.
Fig. 16. Oreocnemis phoenix, K.-D.B. Dijkstra.
Insolation and competition as key factors for distribution Two energy components in forest habitats are limited: insolation and nutrients. For dragonflies these components are interrelated, because they need a sufficient body temperature to fly and forage. Corbet (1999, p. 382) concluded that “foraging incurs a high opportunity cost, and its energy cost may constitute a large proportion of total somatic effort”. Reproductive be-
CyanMagentaYellowBlack Odonata page 145
How can forest influence species composition, diversity and speciation in tropical Odonata?
145
haviour costs additional energy. We see a major factor segregating forest and non-forest species in this conflict between energy demand and supply. Despite high ambient temperatures in the tropics, most forest dragonflies need direct sunlight to maintain a sufficiently high body temperature. Species with a shaded rendezvous on the forest floor frequently return to sunny clearings or the canopy to bask, and even leave the rendezvous when it becomes overcast (e.g. Lempert 1988; Miller 1993; Clausnitzer 1998; Clausnitzer & Dijkstra 2005b). In some rainforest species (e.g. Corduliidae, Gomphidae), only females appear to venture into the cool understorey to oviposit, suggesting that they mate and forage elsewhere, probably in the forest canopy. The forest dragonfly Notiothemis robertsi holds small territories in a shade-sun mosaic near an oviposition site. These territories are defended against all intruders, including males of other species (Clausnitzer 1998). Intruding males never took over an occupied territory, but the chance for them to establish a territory later without energy-costly fights was high (Clausnitzer 1996, 1998). This strategy is suitable for dragonflies of small water bodies on the forest floor. Once the forest opens up, whether by human impact or the natural widening of water bodies, more aggressive and opportunistic non-forest species can invade the habitat, diminishing the chances for N. robertsi males to obtain and hold a territory long enough for successful reproduction. The defensive territoriality of rainforest dragonflies is considered to be an adaptation to save energy in the shady environment (Shelly 1982), but it becomes a disadvantage when an increase of sunlight allows intrusion of more competitive species. A long adult lifetime and/or a long-term territoriality has been observed in many rainforest dragonflies, e.g. in Polythoridae (Fraser & Herman 1993), Pseudostigmatidae (Fincke 1992; Clausnitzer 2004b), Protoneuridae (Lempert 1988) and ‘tetrathemistine’ Libellulidae (Clausnitzer & Lempert 1998). We postulate that many forest dragonflies have developed a ‘slow’ lifestyle to cope with the low insolation and low nutrient levels of their environment, not investing in territorial clashes or fast mating success. This low-energy lifestyle is compensated by longevity, ensuring reproductive success in the long run, enabling a ‘first come, best served’ strategy: forest species can wait to become first, but in direct competition with the more aggressive non-forest species their chance to gain and maintain a territory is negligible. Competition may be the most important factor that keeps forest dragonflies inside the forest, while the low levels of nutrients and insolation inside the forest prevents the non-forest species from living their high-energy lifestyles there. Examples of microclimatic differences in tropical areas between forest, forest edge and non-forest are provided by Turton & Freiburger (1997) and, in relation to stream temperature, by Benstead & Pringle (2004). Interspecific competition is frequently asymmetrical, where one species is (almost) completely unaffected (Begon et al. 1996).
CyanMagentaYellowBlack Odonata page 146
146 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
Of course the competition hypothesis is only applicable within a certain range of habitat parameters. Even without competition many open habitats will be unsuitable for forest dragonflies, but their fundamental niche is considered to be much larger than the realized niche. The credibility of our hypothesis can be tested in areas where numbers of strong competitors are constrained, e.g. by elevation (Box 5). These may, however, be the ‘wrong’ examples. Highland species like most in Box 5 must have evolved under naturally fluctuating circumstances. This predicts a certain tolerance to ecological vicissitudes (drought, temperature), disturbance and probably competition. ‘Real’ forest species that evolved and remained within stable rainforest, like Madagascan Nesolestes or Cameroonian Pentaphlebia, may respond to deforestation and competition much more strongly. A ‘reverse scenario’ in which one forest species pushes another into an open habitat would contradict our hypothesis. This may be the case with Chlorocypha tenuis, which occurs on rainforest streams at many sites in Uganda, but is replaced in that habitat in Bwindi Impenetrable Forest by the closely related C. molindica (hybrids have been found on two occasions). Here C. tenuis occurs on streams outside the forest, a habitat not occupied elsewhere (obs. K.-D.B. Dijkstra). Our hypothesis and observations also have implications for conservation. Locally reduced degrees of competition may leave species like Ethiopia’s endemics less vulnerable to forest destruction, but this may be only temporary as other species gradually invade the altered landscape.
Box 5. Habitat and competition Chlorocnemis abbotti occurs at small forest streams in Tanzania and S Kenya, but was found on Mt Kasigau along a tiny stream in largely waterless thornbush. Almost no non-forest species are found in this part of S Kenya due to the unreliable presence of water (obs. V. Clausnitzer). Pseudagrion bicoerulans is endemic to Kenya and adjacent Tanzania and Uganda. It inhabits forest streams, mainly between 2000 and 3000 m a.s.l. (Fig. 4), but also ventures into open moorlands higher up. Similar habitats at the low end of its elevational range are occupied by the more competitive P. spernatum. Below 2000 m P. bicoerulans is only found rarely, but always in dense forest, where P. spernatum is scarce. Competition might explain the observations, but maybe P. bicoerulans is just restricted to a certain low temperature regime found only in deep shade low down (obs. V. Clausnitzer). The Mulanje plateau (see Box 4) is at most 24 km wide, and was probably once completely covered by Afromontane forest. Frequent fires now maintain large areas of grassland and bracken. Oreocnemis phoenix was found at 82% of the 51 stream sites investigated, although only a quarter of all sites was
CyanMagentaYellowBlack Odonata page 147
How can forest influence species composition, diversity and speciation in tropical Odonata?
147
(partially) forested. Average densities at (partially) forested stream sites (8.9 individuals per 100 m, n = 13) were higher than at open sites (4.6, n = 29). Only five additional zygopteran species have been recorded on the plateau (only two commonly), but thirteen occur in the surrounding lowlands, including three families that are absent on the plateau (obs. & Dijkstra 2004). The endemic species of the Ethiopian highlands occur in completely deforested habitats (Pseudagrion kaffinum, Elattoneura pasquinii (Fig. 17)) or seem tolerant of forest disturbance (Pseudagrion guichardi (Fig. 18), Notogomphus ruppeli). Nonetheless the original vegetation of the highlands and the species’ taxonomic affinities suggest they originally all inhabited forest. Densities of many ubiquitous Afrotropical species in the highlands appear comparatively low (Clausnitzer & Dijkstra 2005a). Streams near the Sempaya hot springs in W Uganda are well shaded by rainforest, but were inhabited by Platycypha caligata and Pseudagrion sublacteum, rather than by typical forest species. Similar streams away from the springs were inhabited by the usual forest species. The two species are always found in largely exposed streams elsewhere. Possibly a higher water temperature allows them to displace the forest species (obs. K.-D.B. Dijkstra).
Fig. 17. Elattoneura pasquinii, K.-D.B. Dijkstra.
Fig. 18. Pseudagrion guichardi, K.-D.B. Dijkstra.
RECOMMENDATIONS Our African hypotheses suggest several directions of research that will shed light on the question why forests have such rich and special dragonfly faunas. We see three main components in which we can deepen our knowledge: 1. The ecosystem component. What are the ecological, behavioural and physiological aspects that determine survival inside or outside the forest environment? Are competition and insolation the main factors explaining the loss of forest species by deforestation, or are other factors equally detrimental,
CyanMagentaYellowBlack Odonata page 148
148 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
such as the loss of prey (types) or the change of substrate in the larval habitat (due to erosion, loss or increase of detritivore organisms)? How do these factors interact? 2. The geographic component. Comparison of the composition and changetolerance of odonate assemblages in areas with distinct climatological histories, may clarify how absolute is the relationship of the present dragonfly fauna with the forest. The assemblages in stable rainforest cores (refugia), such as in SW Cameroon and E Madagascar, may prove to be much more sensitive than those in areas where forest is relatively young. 3. The evolutionary component. Phylogenetic reconstructions for groups that have radiated on both sides of the ‘sunlight-shade divide’ (see Box 3) may clarify the effect and direction of landscape structure and change in speciation.
ACKNOWLEDGEMENTS Thanks are due to the following institutions for assistance: International Centre for Insect Physiology & Ecology (ICIPE), Kenya; National Museums of Kenya (NMK); Kenya Wildlife Service (KWS); Ugandan National Council for Science and Technology (UNCST); Uganda Wildlife Authority (UWA); Tanzanian Commission for Science and Technology (COSTECH) and the National Museum of Natural History Naturalis, The Netherlands. The studies were supported by grants from the German Federal Ministry of Science (BMBF, BIOLOG Programme, 01LC0025 and 01LC0404), the International Dragonfly Fund (IDF) and by SYS-RESOURCE infrastructure of the European IHP Programme. Mike Parr and an anonymous referee gave valuable comments to earlier versions of the manuscript.
REFERENCES BEGON, M., J.L. HARPER & C.R. TOWNSEND. 1996. Ecology. Blackwell Science, London. BENSTEAD, J.P. & C.M. PRINGLE. 2004. Deforestation alters the resource base and biomass of endemic stream insects in eastern Madagascar. Freshwater Biology 49: 490-501. CLARKE, G.P. 2000. Defining the eastern African Coastal Forests. In: Burgess, N.D. & G.P. Clarke (eds.), Coastal Forests of Eastern Africa, pp. 9-26, IUCN, Gland, Switzerland & Cambridge, UK. CLARK, T.E. & M.F. SAMWAYS. 1996. Dragonflies (Odonata) as indicators of biotope quality in the Kruger National Park, South Africa. Journal of Applied Ecology 33: 1001-1012.
CyanMagentaYellowBlack Odonata page 149
How can forest influence species composition, diversity and speciation in tropical Odonata?
149
CLAUSNITZER, V. 1996. Territoriality in Notiothemis robertsi Fraser (Anisoptera: Libellulidae). Odonatologica 25: 335-345. CLAUSNITZER, V. 1998. Territorial behaviour of the rain forest dragonfly Notiothemis robertsi Fraser, 1944: proposed functions of specific behavioural patterns (Odonata: Libellulidae). Journal of Zoology, London 245: 121-127. CLAUSNITZER, V. 1999. Dragonfly (Odonata) records or Kakamega Forest, Western Kenya, with notes on the ecology of rain forest species. Journal of East African Natural History 88: 17-24. CLAUSNITZER, V. 2003a. Dragonfly communities in coastal habitats of Kenya: indication of biotope quality and the need of conservation measures. Biodiversity and Conservation 12: 333-356. CLAUSNITZER, V. 2003b. Odonata of African humid forests - a review. Cimbebasia 18: 173-190. CLAUSNITZER, V. 2003c. Teinobasis alluaudi Martin, 1896 from mainland Africa: Notes on ecology and biogeography (Zygoptera: Coenagrionidae). Odonatologica 32: 321-334. CLAUSNITZER, V. 2004a. Critical species of Odonata in Eastern Africa. In: Clausnitzer, V. & R. Jödicke (eds.) “Guardians of the Watershed. Global status of dragonflies: critical species, threat and conservation”. International Journal of Odonatology 7: 189-206. CLAUSNITZER, V. 2004b. Ecology and biogeography of the dendrolimnetic Coryphagrion grandis (Odonata). In: Breckle, S.-W., B. Schweizer & A. Fangmeier (eds.), Results of worldwide ecological studies. Proceedings of the 2nd Symposium of the A.F.W. Schimper-Foundation, pp. 243-256, Günther Heimbach, Stuttgart. CLAUSNITZER, V. & K.-D.B. DIJKSTRA. 2005a. The dragonflies (Odonata) of Ethiopia, with notes on the status of endemic taxa and the description of a new species. Entomologische Zeitschrift 115: 117-130. CLAUSNITZER, V. & K.-D.B. DIJKSTRA. 2005b. Honouring Nobel Peace Prize winner Wangari Maathai: Notogomphus maathaiae spec. nov., a threatened dragonfly of Kenya’s forest streams. International Journal of Odonatology 8: 177-182. CLAUSNITZER, V. & J. LEMPERT. 1998. Preliminary comparative approach of the reproductive behaviour of African Tetratheminae (Anisoptera: Libellulidae). Journal of African Zoology 112: 105-107. CONSIGLIO, C. 1978. Odonata collected in Ethiopia by the expeditions of the Accademia Nazionale Dei Lincei. II Introduction and the Zygoptera. Problemi attuati di sciencza di cultura (III) 243: 27-51. CORBET, P.S. 1983. Odonata in Phytotelmata. In: Frank, J.H. & L.P. Lounibos (eds.), Phytotelmata: terrestrial plants as hosts for aquatic insect communities, pp. 304, Marlton, New Jersey. CORBET, P.S. 1999. Dragonflies: Behaviour and Ecology of Odonata. Harley Books, Colchester. DIJKSTRA, K.-D.B. 2004. Dragonflies (Odonata) of Mulanje, Malawi. IDF Report, Newsletter of the International Dragonfly Fund 6: 23-29.
CyanMagentaYellowBlack Odonata page 150
150 Klaas-Douwe B. Dijkstra & Viola Clausnitzer
DIJKSTRA, K.-D.B. 2005. Taxonomy and identification of the continental African Gynacantha and Heliaeschna (Odonata: Aeshnidae). International Journal of Odonatology 8: 1-33. DIJKSTRA, K.-D.B. 2006. Taxonomy and biogeography of Porpax, a dragonfly genus centred in the Congo Basin (Odonata, Libellulidae). Tijdschrift voor Entomologie 149: 71-88. DIJKSTRA, K.-D.B., V. CLAUSNITZER, & A. MARTENS, in press. Tropical African Platycnemis damselflies and the biogeographical significance of a new species from Pemba Island, Tanzania (Odonata: Platycnemididae). Systematics & Biodiversity. DIJKSTRA, K.-D.B. & J. LEMPERT. 2003. Odonate assemblages of running waters in the Upper Guinean forest. Archiv für Hydrobiologie 157: 397-412. DUMONT, H.J. 1988. On the composition and palaeoecological significance of the odonate fauna of the Darfur, Western Sudan. Odonatologica 17: 385-392. FINCKE, O.M. 1992. Interspecific competition for tree holes: Consequences for mating systems and coexistence in neotropical damselflies. The American Naturalist 139: 80-101. FRASER, F.C. & T.B. HERMAN. 1993. Territorial and reproductive behaviour in a sympatric species complex of the neotropical damselfly Cora Sélys (Zygoptera: Polythoridae). Odonatologica 22: 411-429. GENTRY, A.H. 1993. Diversity and floristic composition of lowland tropical forest in Africa and South America. In: Goldblatt, P. (ed.), Biological Relationships between Africa and South America, pp. 500-547, Yale University, Dexter, Michigan. GOLDBLATT, P. 1993. Biological Relationships between Africa and South America, Yale University, Dexter, Michigan. HILL, M.O. & H.G. GAUCH. 1980. Detrended correspondence analysis, an improved ordination technique. Vegetatio 42: 47-58. HAMILTON, A.C. 1992. History of Forests and Climate. In: Sayer, J.A., C.S. Harcourt & N.M. Collins, The conservation atlas of tropical forests, pp. 17-25, Macmillan, London. KINGDON, J. 1989. Island Africa. Princeton University Press, Princeton. KINVIG, R.G. & M.J. SAMWAYS. 2000. Conserving dragonflies (Odonata) along streams running through commercial forestry. Odonatologica 29: 195-208. LEGRAND, J. & G. COUTURIER. 1985. Les Odonates de la forêt de Taï (Côte d’Ivoire). Premières approches: faunistique, répartition écologique et association d’espèces. Revue d’Hydrobiologie tropicale 18: 133-158. LEMPERT, J. 1988. Untersuchungen zur Fauna, Ökologie und zum Fortpflanzungsverhalten von Libellen (Odonata) an Gewässern des tropischen Regenwaldes in Liberia, Westafrika. Diplomarbeit an der Friedrich-Wilhelms Universität, Bonn. MILLER, P.L. 1993. Some dragonflies of the Budongo Forest, Western Uganda (Odonata). Opuscula zoologica fluminensia 102: 1-12. MILLER, P.L. 1995. Some dragonflies of forests near Kampala, Uganda, with notes on their ecology and behaviour (Odonata). Opuscula zoologica fluminensia 136: 1-19.
CyanMagentaYellowBlack Odonata page 151
How can forest influence species composition, diversity and speciation in tropical Odonata?
151
MÜLLER, O., V. CLAUSNITZER, K. GRABOW, G. VICK & F. SUHLING. 2005. Description of the final stadium larvae of African Gomphidia (Odonata: Gomphidae). International Journal of Odonatology 8: 233-241. NEVILLE, A.C. 1960. A list of Odonata from Ghana, with notes on their mating, flight and resting sites. Proceedings of the Royal Entomological Society of London 35: 124-128. O’NEILL, G. & D.R. PAULSON. 2001. An annotated list of Odonata collected in Ghana in 1997, a checklist of Ghana Odonata, and comments on West African odonate biodiversity and biogeography. Odonatologica 30: 67-86. PINHEY, E. 1964. A revision of the African members of the genus Pseudagrion Selys (Odonata). Revista de Entomologia de Moçambique 7: 5-196. PINHEY, E. 1970. Monographic study of the genus Trithemis Brauer (Odonata: Libellulidae). Memoirs of the entomological Society of southern Africa 11: 1-159. PINHEY, E. 1984. A checklist of the Odonata of Zimbabwe and Zambia. Smithersia 3: 1-64. SHELLY, T.E. 1982. Comparative foraging behaviour of light- versus shade-seeking damselflies in a lowland neotropical forest (Odonata: Zygoptera). Physiological Zoology 55: 335-343. TER BRAAK, C.J.F. 1986. Canonical correspondance analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67: 1167-1179. SUHLING, F. & S. LEPKOJUS. 2001. Differences in growth and behaviour influence asymmetric predation among early-instar dragonfly larvae. Canadian Journal of Zoology 79: 854-860. TURTON, S.M. & H.J. FREIBURGER. 1997. Edge and aspect effects on the microclimate of a small tropical forest remnant on the Atherton Tableland, northeastern Australia. In: Laurance, W.F. & R.O. Bierregaard (eds.), Tropical forest remnants, pp. 45-54, The University of Chicago Press, Chicago & London. VALLAN, D., F. ANDREONE, V.H. RAHERISOA & R. DOLCH. 2004. Does selective wood exploitation affect amphibian diversity? The case of An’Ala, a tropical rainforest in eastern Madagascar. Oryx 38: 410-417. VICK, G.S. 1999. A checklist of the Odonata of the South-West province of Cameroon, with the description of Phyllogomphus corbetae spec. nov. (Anisoptera: Gomphidae). Odonatologica 28: 219-256. WHITE, F. 1983. The vegetation of Africa. A descriptive memoir to accompany the UNESCO / AETFAT / UNSO vegetation map of Africa. UNESCO, Paris.
CyanMagentaYellowBlack Odonata page 152
This page intentionally left blank
CyanMagentaYellowBlack Odonata page 153
Specialists vs. generalists the Odonata the importance of forest environments AdolfoinCordero Rivera– (ed) 2006 Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 153-179. © Pensoft Publishers
Sofia–Moscow
Specialists vs. generalists in the Odonata – the importance of forest environments in the formation of diverse species pools Göran Sahlén Ecology and Environmental Sciences, Halmstad University, P. O. Box 823, SE-30118 Halmstad, Sweden.
[email protected]
ABSTRACT In Scandinavia more Odonate species occur in forested environments than in agricultural areas. Some authors attribute the high number of forest species to extensive river and wetland networks. But because there are also fewer species in some agricultural areas with numerous wetlands, there must be another explanation. It is known that forestry practices affect species composition. Remove the trees, the environment changes and some species disappear. The time elapsed after logging affects species survival. While undisturbed forest habitats support the greatest number of species, partivoltine species decrease during the first 5-10 years after disturbance. Univoltine species are not affected - in fact the univoltine species present here are also part of the species pool of agricultural areas; they are true generalists. A discriminant analysis comparing the species composition of lakes in different seral stages during forest regrowth gave more than 90% separation between the stages. Moreover, an even better separation was achieved when the investigation was combined with an analysis of (semi)aquatic plant communities along the shoreline, or when dragonfly density was taken into account. Plants and odonates are interconnected; the insects respond to the habitat’s form and structure rather than to water chemistry (e.g., acidity or nutrient levels) or other ecological parameters. Forestry thus affects the very structures needed for survival. What kind of structures are we dealing with? A classification of species according to habitat preferences in a comparison between agricultural and
153
CyanMagentaYellowBlack Odonata page 154
154 Göran Sahlén
forested areas showed that in treeless habitats there were fewer specialists as well as generalists in constructed wetlands compared to older ponds and lakes. The latter habitats, however, had fewer species than were present in the adjacent forested lakes. An investigation of constructed wetlands under 10 years of age showed that those close to forest habitats (even small clumps of trees) had, on average, more than twice as many breeding species than those in more open areas. Trees are obviously important to Odonata species, at least during some stage of their life. All species would probably survive in the waters of open areas, yet certain species do not survive unless a forest habitat occurs at a moderate distance from their breeding waters. Larval as well as adult habitat is relevant; egg-laying substrates must be included. Forests thus seem to possess what agricultural areas do not — the maintenance of a high diversity of Odonata within a landscape depends on several seral stages and many different wetlands, surrounded by a diverse matrix of plants, including trees. All this adds up to one general rule: forests harbour specialists, while open landscapes are the playgrounds of generalists. Key words: Dragonfly, Odonata, forestry, species richness, specialist species, generalist species, diversity, habitat structures, species pool.
INTRODUCTION It is not surprising that forests are critical habitats for many dragonfly species. Such species may well be called forest species, as opposed to species living in open areas. In this paper I argue that forest species are frequently specialists, while the inhabitants of open areas are generalists. My focus is on the temperate forests of northern Europe, but I occasionally choose a somewhat more global perspective. Specialised forest dragonflies live in the tropics (e.g. Clausnitzer 2003a), while in the European taiga, few species seem to occur solely in forests (ecological descriptions in Robert 1958, Askew 1988, Fogh Nielsen 1998 and others). One obvious difference between tropical and temperate forest environments lies in the types of water bodies available. Rivers occur in all forests, whereas lakes and bogs are common in many temperate areas but rare in tropical forests. The ecological preferences of forest-living dragonflies vary accordingly. In tropical forests river species predominate, while lake species, if present, will use any standing water for breeding. These standing-water habitats are often artificial ones, such as dams and reservoirs. In temperate forests there is a more pronounced separation between standing and running waters (the lotic and lentic waters of limnologists), with numerous species favouring either or both of these environments. In Northern Europe and parts of Asia and North America the water, whether running or standing, is integrated into the temperate forest,
CyanMagentaYellowBlack Odonata page 155
Specialists vs. generalists in the Odonata – the importance of forest environments
155
forming a mosaic with large areas of swampy forest displaying a high level of biodiversity (cf., Hörnberg et al. 1995). As pointed out by Sahlén (1999), all the wetlands within the forest itself, with their special flora and fauna, must be treated as an integral part of the forest ecosystem, because changes in the wetlands affect the surrounding forest and vice versa. Temperate forests lack dragonfly species that breed in water-filled holes and other containers, and hence bromeliad dwellers (e.g. Fincke 1992, DeMarco & Furieri 2000) have no counterparts in the northern fauna. Temporary water bodies in temperate forests are vernal or autumnal pools in which only a few dragonfly species are reported to breed, among them the genera Lestes, and Sympetrum (e.g. Robert 1958; Fig. 1). Somatochlora alpestris also breeds in these kinds of waters in the north (Johansson & Nilsson, 1991). The first two genera are univoltine and overwinter as eggs, which makes them typical generalists - only demanding that water is present during a sufficient time span from spring into summer. The last species is, however, always partivoltine in northern Europe and must therefore overwinter at least once as a larva; which is difficult in a temporary pool.
Fig. 1. Sympetrum flaveolum, a young female after emergence from a small forest pool in central Sweden. The members of this genus are univoltine in most northern boreal forests. They are generalists, capable of using any kind of water, including temporary pools, for reproduction; cf., the habitat of Lestes sponsa in Fig. 8. Photo by the author.
CyanMagentaYellowBlack Odonata page 156
156 Göran Sahlén
As previously stated, there are no specialised forest dragonflies in temperate forests. Many, if not most, of the species occurring in forest habitats may also occur in environments other than forests. Even so, temperate forests tend to support a rich odonate fauna, whether it is in the southern hemisphere, e.g., in Patagonia (Muzon 1997) or in the northern hemisphere. In the Bialowieza Forest and its surroundings in Poland the high number of species are dependent on the river network and on the abundance of small-scale sandpits with water bodies in various seral stages (Theuerkauf & Rouys 2001). The number of species present is dependent on both the latitude and the diversity of the environment. A forest in France (Notre-Dame) supported 31 species (Le Calvez 1998), a diversity comparable to that of central Sweden, but higher species numbers have been found elsewhere. According to Samways & Steytler (1996) and Stewart & Samways (1998), odonate communities in disturbed habitats will often be less species rich and consist of many widespread generalists. Their studies, of course, relate to warmer climates than European ones. Another example from the tropics is the study by Machado et al. (1991) in Brazil where species with a wide geographical range, i.e. the common species, were predominant in open savannah, while those with a more restricted distribution were predominant
Fig. 2. Mårdsjön, a Sphagnum-lined lake in a central Swedish boreal forest which supports 19 species and is used as a reference lake with high biological values in Sahlén & Ekestubbe (2001). The lake is fish-free and among the odonate species breeding there are many uncommon ones, e.g., Coenagrion armatum, C. johanssoni, C. lunulatum, Leucorrhinia albifrons, L. caudalis and L. pectoralis. Photo by the author.
CyanMagentaYellowBlack Odonata page 157
Specialists vs. generalists in the Odonata – the importance of forest environments
157
in forests. In central Sweden Sahlén (1999) and Sahlén & Ekestubbe (2001) surveyed 74 lakes in a southern boreal forest containing some deciduous trees mixed with spruce and pine; examples of lake habitats are shown in Figs 2 and 3). Although 34 species were encountered (Table 1), a maximum of 19 species co-existed in the same lake (if the five univoltine species present but not previously counted are added to the number of partivoltine species presented in the paper). In agricultural areas in southern Sweden surveyed by the author and Linda Birkedal between 2001 and 2002, also using data from Svensson et al. (2004), the species pool was about as rich as in the forested areas — 30 species in all (Table 1) but with a lower maximum of 13 species per lake. This implies that the open, anthropogenically changed agricultural areas would be comparable to the forests in Scandinavia in terms of species diversity.
Fig. 3. An average forest lake in the boreal forest of central Sweden. The shoreline is partly filled with large areas of reed, Phragmites australis or sedges, Carex spp. Some scattered floating plants (e.g., Nymphaea alba) are also present. The species composition varies, but often consists of the following nine species: Lestes sponsa, Coenagrion hastulatum, Aeshna grandis, A. juncea, Cordulia aenea, Libellula quadrimaculata, Leucorrhinia dubia, L. rubicunda and Sympetrum danae. Photo by the author.
+ + + + + + + + + + + + + + + +
Calopteryx splendens (Harris) Calopteryx virgo (Linnaeus) Lestes dryas Kirby L. sponsa (Hansemann) Erythromma najas (Hansemann) Pyrrhosoma nymphula (Sulzer) Coenagrion armatum (Charpentier) C. hastulatum (Charpentier) C. johanssoni (Wallengren) C. lunulatum (Charpentier) C. puella (Linnaeus) C. pulchellum (Vander Linden) Enallagma cyathigerum (Charpentier Ischnura elegans (Vander Linden) Aeshna caerulea (Ström) A. cyanea (Müller) A. isosceles (Müller) A. juncea (Linnaeus) + + + + + + + + + + + + +
AG FO
Species + + + + + + + + + + +
A + + + + + + + + + + + +
B + + + -
C + + + + + -
D + + + + + + + + + + + + + +
E + + + + + + + + + + + + +
F + + + + + + + + + + + + + + + +
G
+ + + + + -
H
+ + + + -
I
+ + + + + + + + -
J
2 3 10 10 4 6 5 4 4 4 6 6 5 5 5 4 2 5
N
SPE SPE GEN GEN INT INT INT INT INT INT INT INT INT INT INT INT SPE INT
Classification
Table 1. The species pool in Swedish agricultural (AG) vs. forest (FO) areas in the south and central parts of the country. Sampling in wetlands, ponds and lakes only; no running waters. Species are classified according to occurrence in 10 defined habitat types: A) waters dominated by peat, B) waters of bog character with Sphagnum moss, C) vegetation free waters, often more or less disturbed, D) shady waters, often surrounded by dense stands of trees, E) waters with low nutrient levels, F) waters with high nutrient levels, G) waters with emergent vegetation, H) waters with floating vegetation, I) waters more or less covered with emergent and/or floating vegetation, J) inlets and outlets in waters with the characteristics of running water. Occurrence is indicated with a +, absence with a -. N is the number of habitats per species. Generalists (GEN) occur in ≥ 7 habitats, intermediates (INT) in 4-6, specialists (SPE) in 1-3. At the bottom are numbers of species found in each area and numbers of species found in each habitat type. Note that the habitat preferences are derived from the areas surveyed; in other areas they may be different.
CyanMagentaYellowBlack Odonata page 158
158 Göran Sahlén
+ + + + + + + + + + + + + + + 31
A. grandis (Linnaeus) A. mixta Latreille A. osiliensis Mierzejewski A. subarctica Walker A. viridis Eversmann Anax imperator Leach Brachytron pratense (Müller) Cordulia aenea (Linnaeus) Somatochlora arctica (Zetterstedt) S. flavomaculata (Vander Linden) S. metallica (Vander Linden) Libellula depressa Linnaeus L. quadrimaculata Linnaeus Orthetrum coerulescens (Fabricius) Leucorrhinia albifrons (Burmeister) L. caudalis (Charpentier) L. dubia (Vander Linden) L. pectoralis (Charpentier) L. rubicunda (Linnaeus) Sympetrum danae (Sulzer) S. flaveolum (Linnaeus) S. sanguineum (Müller) S. striolatum (Charpentier) S. vulgatum (Linnaeus) + + + + + + + + + + + + + + + + + + + + + 34
AG FO
Species + + + + + + + + + + + + + + + + + 28
A + + + + + + + + + + + + + + + + + 29
B + + + + + + + 10
C + + + + + + 11
D + + + + + + + + + + + + + + + + + + 31
E + + + + + + + + + + + + + + + + + 30
F + + + + + + + + + + + + + + + + + + + + + 37
G + + + + + + + + + + + + 17
H + + + + + + + + + 13
I + + + + + + + + 15
J 8 2 2 3 2 4 3 6 4 5 6 2 7 4 5 5 5 5 4 10 10 10 10 10
N GEN SPE SPE SPE SPE INT SPE INT INT INT INT SPE GEN INT INT INT INT INT INT GEN GEN GEN GEN GEN
Classification
CyanMagentaYellowBlack Odonata page 159
Specialists vs. generalists in the Odonata – the importance of forest environments
159
CyanMagentaYellowBlack Odonata page 160
160 Göran Sahlén
The Scandinavian forests surveyed were not remote primary forests but ecosystems used by humans; in fact they are a type of tree plantation with a turnover time of about 100 years. One might then ask: why do agricultural areas have almost the same number of species as forests? A look at the distribution pattern within the surveyed lakes gives a few clues. In the forest lakes (data used in Sahlén & Ekestubbe 2001; univoltine species added) an average of 8.1 species was found in each lake (range 2-19) while in southern Sweden the average was only 3.0 species (range 0-13). In fact almost 33% of the southern lakes supported no odonates at all, and only 22% harboured five or more species. This is a totally different pattern from that found in the forests, where all surveyed lakes had at least one dragonfly species, and 85% had five or more species. Thus, in Scandinavia more species occur in the water bodies of forested environments than in agricultural areas. The southern agricultural areas have fewer lakes than the forests, but owing to recent wetland constructions (cf., Svensson et al. 2004) the number of possible breeding sites for dragonflies is relatively high. Hence, the difference in occupancy must be owing to factors other than the mere number of water bodies. Larval surveys such as those of Sahlén (1999), Sahlén & Ekestubbe (2001) and Svensson et al. (2004) are still rare. Most surveys consider only adults, and adults migrate. Hence, species found in forests may not breed in forests at all. A river running through a forested area certainly has a given set of dragonfly species, but these species might also have occurred there had the river run through other habitats, such as fields or even urban areas. To some species forest habitats are important only during certain stages of the adult life. For instance, Sympetrum infuscatum in Japan uses forest gaps for foraging only (Watanabe et al. 2004). Many riverine species found in forested areas use the forest for foraging (Kirkton & Schultz 2001). Other stream-dwelling species such as Calopteryx aequabilis and C. maculata in North America feed in forests and subsequently return to streams to breed (Jonsen & Taylor 2000). At least some of the individuals move on a more or less daily basis between the forests and streams. These species never leave the close vicinity of the stream if there is no forest cover. These examples show that odonate species may depend on forests even if they do not actually breed within them.
FORESTRY AND SPECIES COMPOSITION Human impact on forests will affect all odonates present to some extent. The use of pesticides in forest management will, of course, have adverse effects on all species living in the streams and wetlands concerned (Poirer & Surgeoner 1987). However, in boreal forests, at least in Scandinavia, this is no longer a problem because all major forestry companies are desperately trying
CyanMagentaYellowBlack Odonata page 161
Specialists vs. generalists in the Odonata – the importance of forest environments
161
to live up to their new environment friendly, non-pesticide-using image. Instead, it is the logging itself that affects the presence or absence of the species. Timber management is a major industry, and it undoubtedly has an impact on the dragonfly fauna of the wetlands/forest mosaic (Fig. 4). The first author to point this out was Rith-Najarian (1998) who studied the anisopteran fauna along the upper Mississippi River in Minnesota. Using a species pool of 39 species she found that recently cleared areas contained the lowest number of species and the lowest species diversity. The highest diversity was found in old-growth forest areas while regrowth areas had an intermediate diversity (Figs 5, 6, 7). Sahlén (1999) found the same pattern in central Sweden, where the logging of forests in the vicinity of small lakes caused a decline in partivoltine species while univoltine species remained unaffected. The changes were slow and took several years to appear. Sahlén (1999) also showed that the species richness of the odonate fauna rebounded more than 15 years after the local deforestation. The recovery of odonate diversity after logging was deemed slower in Minnesota, taking several decades after the disturbance (Rith-Najarian 1998). None of these studies, however, investigated the species composition in detail. It is easy enough to state that diversity rose again a number of years after the disturbance, but what species had returned? Was it the displaced species that had come back? The
Fig. 4. The typical mosaic landscape of lakes, moors and forest patches dominating large areas of central and northern Scandinavia. In this picture no logging has taken place due to a dominance of moorland. In other areas with larger, drier and denser patches of forest, most old growth has been cut down and replaced with plantations of spruce and pine. Photo by the author.
CyanMagentaYellowBlack Odonata page 162
162 Göran Sahlén
Fig. 5. Schoolcraft River, a tributary to the Mississippi River headwaters in Minnesota. Here logging has not occurred since the early 20th century. The Mississippi River is designated as having “wild river” status in some State Forests along the headwaters. This means that logging is not supposed to occur within 300 m of the river where a buffer zone should always be present. But in reality there are many places where logging has occurred well within 150 m of the shoreline. In such areas, logging runoff affects water quality, and there is a sharp decline in dragonfly diversity at sample sites in the corridor close to the logging site. See Fig. 10 for a Swedish comparison. Photo by Janet Rith-Najarian.
Fig. 6. Pachydiplax longipennis, an indicator for recovering forests in Minnesota. It is currently expanding its range. Scan by Janet Rith-Najarian.
CyanMagentaYellowBlack Odonata page 163
Specialists vs. generalists in the Odonata – the importance of forest environments
163
Fig. 7. Typical Minnesotan forest near Wolf Ridge. Photo by the author.
current monoculture of secondary forest plantations in Scandinavia gave rise to a certain fear that the recovered biodiversity might be composed of only the most common species. In 1999 I asked if the re-established fauna was really identical to the original fauna and I therefore decided to subject the 1999 data to discriminant analyses to ascertain whether or not this was the case. As the critical point of the 1999 paper was that univoltine species were not affected by logging, I expected that the composition of these species in the lakes would not differ among the three forestry classes used: class 1) no logging in the area for the past 15 years or more; class 2) recent logging activities (0-5 years ago); class 3) logging undertaken 6-15 years ago. As predicted, the univoltine species formed no characteristic groups corresponding to the classes. Only 59.4% of the cases grouped correctly. This shows that the univoltine species in the area disperse randomly without reacting to water quality, shading or other factors affected by logging. They are true generalists and opportunists. A man-made depression that the following year formed the larval habitat of the generalist Lestes sponsa is shown in Fig. 8. Norling (1984 and pers. comm.) studied the life cycles of several dragonfly species in central Sweden. When the forest surrounding one of his study areas was logged, the larval development of certain species was shortened by a whole year (from 3-4 years to 2-3 years) because of increased solar radiation and subsequent heating of the water. He also noted a significant increase in population size owing to the release of nutrients into the water caused by the logging (cf., Sahlén 1999). It is therefore a reasonable assumption that any opportunistic (univoltine) species present would also increase their popula-
CyanMagentaYellowBlack Odonata page 164
164 Göran Sahlén
Fig. 8. Tracks of a clear-felling harvester in a central Swedish boreal mixed forest. The pools formed in the depressions are frequented by egg-laying Libellula depressa and L. quadrimaculata. Next season these tracks became larval habitats for Lestes sponsa. Photo by the author.
CyanMagentaYellowBlack Odonata page 165
Specialists vs. generalists in the Odonata – the importance of forest environments
165
tion sizes, and that any subsequent decline in the partivoltine species may be caused by interspecific competition from univoltine opportunists. This connection has not yet been demonstrated, but should be investigated. A further discriminant analysis, using presence/absence data on partivoltine species showed that 93.8% of all cases grouped in the predicted way. There was a clear separation into the three forestry classes, showing that there was a difference in species composition among the different seral stages. The separation was mainly based on the presence and absence of 14 species, presented in table 2. We can, for instance, note that the stenotopic Leucorrhinia species are found in class 1 only, but that the equally sensitive Aeshna viridis is found in the lakes of the class 3 recently logged areas. Interestingly, the two cases that did not group in the predicted way belonged to class 1, which included both old-growth forests and secondary growth over 15 years after the disturbance. This was intrinsically a heterogeneous group, and the 15-year time perspective used in the 1999 paper was obviously too short (cf., Rith-Najarian 1998). A better time-span might be 30-50 years, which would correspond to 30-50% of the tree turnover time in this managed area, which is approximately 100 years between each felling and replanting. To find further clues to the changes imposed on the species composition by forestry, I used the presence and abundance (number of larvae in proportion to amount of sampling effort) of all odonates found in the lakes in a third discriminant analysis where, interestingly, all cases split in the predicted way (Fig. 9). The plant composition of the lakes also showed a 100% separation of the three classes. In Figure 9, function 1 corresponds to the drastic changes that forestry imposes on the species composition, whereas function 2 clearly demonTable 2. Species uniquely present or absent in lakes of three different forestry classes. See text for further explanation. Forestry class
species present
species absent
1
Erythromma najas Brachytron pratense Epitheca bimaculata Leucorrhinia albifrons L. pectoralis
Coenagrion lunulatum
2
Aeshna osiliensis Somatochlora metallica
3
Coenagrion armatum Aeshna caerulea
Coenagrion johanssoni Aeshna viridis Cordulia aenea Orthetrun cancellatum
CyanMagentaYellowBlack Odonata page 166
166 Göran Sahlén 6
4
+
1
2 2
0
-2
-4
Function 2
3 -6 -8 -10
0
10
20
30
Function 1
Fig. 9. Diagram showing the distribution of the examined localities according to the first two Canonical Discriminant Functions using presence/absence and abundance for larvae. The symbols indicate the different forestry classes: 1) no forestry in the area for the past 15 years or more; 2) recent forestry activities (0-5 years ago); 3) forestry measures undertaken 6-15 years ago. Function 1 corresponds to changes imposed by forestry on the species composition. Function 2 demonstrates that as time passes the species composition changes again, not back to what it was but into something different. Forest lakes disturbed by forestry need a long time to recover, but the present data do not show whether this recovery is actually taking place.
strates that, as time goes by, the species composition again starts to change. However, it does not return to its previous condition, but develops into something completely different. My 1999 paper indicated that the species diversity, defined as the mere number of species, did return to the original state after more than15 years, but Figure 9 sends another message — it indicates that all classes are widely separated. Whatever species composition occurs in class 3, it is very different from that in class 1. Unfortunately, the material is not extensive enough to separate class 1 into old-growth and old regrowth areas, which might have clarified the matter. Anyway, to sum it up: logging affects both the species composition and the abundance of odonates, as well as the plant composition of the lakes. The association between odonates and plants, described by Buchwald (1992), is worth mentioning at this stage. In 1999 I showed that the lakes with
CyanMagentaYellowBlack Odonata page 167
Specialists vs. generalists in the Odonata – the importance of forest environments
167
the highest number of dragonfly species also harboured the highest number of aquatic plants. It seems reasonable to assume that an odonate species selective in its choice of breeding water is more likely to find its favoured plant composition at least somewhere in a lake rich in plant species (Sahlén 1999). Hence, the odonate community of a species-rich lake should have more resilience and greater ability to survive, or recover from, local logging (Figs 10, 11,
Fig. 10. A central Swedish lake two years after a logging event some 100 m behind the shoreline to the left in the picture. It is clear where the logged area is as reed (Phragmites) has started to grow near the shore due to influx of nutrients from the deforested area. This is an example of a more prudent logging; often the vegetation is cut down almost to the shoreline leaving only a thin stand of trees. Leaving a narrow “corridor” is currently part of the forestry companies’ measures for a better environment. Prudent or not — in this lake a population of Coenagrion armatum disappeared two years after logging occurred, but other populations live in speciesrich lakes nearby, so recolonisation is possible. Photo by the author.
Fig. 11. A male Aeshna caerulea resting on a pine tree near a lake in northern Sweden. Photo by the author.
CyanMagentaYellowBlack Odonata page 168
168 Göran Sahlén
12). The local populations facing the highest risk of extinction are those from lakes that are already depleted and species-poor when logging occurs. As I put it in 1999: The lakes with the highest number of dragonflies and plants were situated in old woodland where logging had occurred more than 15 years ago, if at all. Apparently, an undisturbed forest habitat is necessary to maintain a high diversity of dragonflies in boreal forests. Given that selective (rare) species will breed exclusively in the lakes and rivers of old-growth forest areas, the implications of forestry impact on these habitats are severe. The number of sites with a wide array of microhabitats for dragonflies is likely to decrease, which will make dispersal difficult for species dependent on a particular plant composition or other factors necessary for oviposition or larval development (Sahlén 1999, and cf., Rith-Najarian 1998). In other parts of the world many studies have pointed in the same direction. Specialists are replaced by generalists when forests disappear. Some studies claim that the “new” generalist species invade from other (more open) areas, but they may well be present in the forest already, albeit in low numbers, as is the case in Sweden. A study on assemblages of adult odonates in Liberia and Ghana (Dijkstra & Lempert 2003) showed that the anthropogenic opening of stream habitat by deforestation or damming resulted in an invasion of species from more open habitats downstream, and in the disappearance of upstream (dense forest) species. In Panama, Rehfeldt (1986) observed that the number of odonate individuals, as well as the number of
Fig. 12. A male Aeshna caerulea resting on the ground in a recently logged area in northern Sweden. Some species are able to persist in logged areas by changing their behaviour. Photo by the author.
CyanMagentaYellowBlack Odonata page 169
Specialists vs. generalists in the Odonata – the importance of forest environments
169
species, increased with the size of the logged areas. This might be consistent with the pattern observed by Dijkstra & Lempert (2003) — an increase in open area generalists at the expense of forest specialists. Samways (2003) stated that forest species breeding in running waters were dependent on the forest, whereas pool species were tolerant to deforestation, and thus present also in deforested areas. In East Africa the forest species are often confined to the remaining coastal forests (Clausnitzer 2003b) and, as in northern Europe, they are sensitive to disturbances. When the habitat is changed the species richness initially increases, but most of the colonisers are common and widely distributed species (Clausnitzer 2003b). When drastic habitat changes occur (e.g., removal of natural forests) species will disappear, or at least become rare, because of habitat fragmentation (Samways 1999). In South Africa even the presence of rare species in nature reserves does not necessarily guarantee their survival (Samways 1999). Also forest plantations may be detrimental to some species. Kinvig & Samways (2000) showed that commercial afforestation along rivers in South Africa caused a drop in diversity where the water became shaded by a closed canopy, be it from natural forest or from exotic trees. Ormerod et al. (1990) showed that the larval habitat of Cordulegaster boltoni (Fig. 13) was eroded
Fig. 13. Cordulegaster boltoni, a riverine species, in Scandinavia confined to small streams in coniferous forest. However, in the south it may switch to deciduous forests or more open habitats, such as heather heaths. The species has been shown to react negatively to conifer plantations along stream margins. Photo by A. Cordero.
CyanMagentaYellowBlack Odonata page 170
170 Göran Sahlén
in forest streams in Wales where conifers were planted. Forestry thus affects the very structures in the water that odonates need for survival: substrate, plant composition and various faunal elements.
WHAT IS THIS GOOD THING ABOUT TREES? I have shown that forest habitats in Scandinavia and elsewhere have relatively large species pools, but that agricultural areas may also contain many species. The difference is that the average number of species per lake or wetland is higher in forests than in open areas. This means that there must be something about trees that attracts more species to forests. Or is it really the trees themselves that affect the species composition? I analysed the data from the agricultural areas in southern Sweden to find to what extent odonate communities in constructed wetlands (age <15 years), semi-natural ponds (age >30 years) and natural lakes differed, and how each habitat type contributed to the regional species pool. Data from 60 communities (37 constructed wetlands, 13 seminatural ponds and 10 lakes) were used. These data showed a distinct grouping among the three habitats (canonical discriminant analysis, 96.7% certainty). I found that the dragonfly communities in the constructed wetlands consisted mainly of generalist species, while a higher proportion of specialists occurred in ponds and lakes. Apart from Libellula depressa, a primary coloniser, all of the species found in the constructed wetlands also lived in the ponds and lakes. To separate generalists from specialists I used the habitats in which each species is known to occur (Table 1). Species occurring in few habitats were named specialists, e.g., Aeshna subarctica, which was only found in Sphagnum-dominated lakes and ponds in the area as well as Calopteryx virgo and C. splendens (Fig. 14), which require currents with emergent vegetation. At the other end of the spectrum were the generalists, the extremes being certain Sympetrum and Lestes species, which occur in almost all kinds of habitat, from wetlands devoid of vegetation to the diverse lakes. Species in the middle of the scale were classified as intermediate. In a GLM I used total number of species as response, the number of generalist and specialist species and their interraction with wetland type as predictors, assuming poisson distribution. The total number of species did not vary between wetland types (p>0.05), but there were clear associations between wetland type and number of generalist (p=0.031) and specialist (p=0.012) species, with a significantly lower number in constructed wetlands compared to the other two habitats. Constructed wetlands also had fewer intermediates when checked separately (ANOVA; F = 34.85; p < 0.001). Interestingly, the constructed, younger environments had less of everything compared to the lakes and ponds, which are both older
CyanMagentaYellowBlack Odonata page 171
Specialists vs. generalists in the Odonata – the importance of forest environments
171
Fig. 14. Calopteryx splendens, a specialist species in lakes. Because it normally breeds in running waters, lakes are only accepted if they possess certain qualities such as running water, which may be found near the outlet or the inlet of the lake and emergent vegetation that is required for larval habitat and adult territorial display. These qualities are examples of complex structures in and around the waters that are often present in forests but rarely in open, agricultural habitats.
environments. It is reasonable to assume that the older habitats have a higher structural complexity, which is what many species — in particular the specialists — seek. A comparison between the average number of species in the most species-rich communities in the agricultural and the forested areas (5.9 and 8.1, respectively; cf., above) reveals that, irrespective of the age and structural complexity of the agricultural wetlands, the diversity will not reach the levels we see when trees are present. In southern Sweden, Svensson et al. (2004) showed that a positive but weak regression existed between the amount of forest surrounding wetlands and the number of dragonfly species breeding (Regression; r2 = 0.16; F=20.3; p < 0.0005). Looking more closely at the areas adjacent to the wetlands, these authors found that the presence of forest (in this case at least one stand of some 10-20 trees) less than 20 metres from the shoreline resulted in a higher number of dragonfly species in the wetland than if trees were not present (ANOVA; F=5.75; p = 0.018). In fact, the number of species in wetlands with trees was more than twice as high as in those without. These results emphasize that trees per se, or something associated with trees, are beneficial to dragonfly diversity, at least in agricultural areas. In North America Rith-Najarian (1998) showed that the amount of trees in the runoff area of a northern Minnesota river is correlated with the number of anisopteran species present. In addition, she found that some species exhibit
CyanMagentaYellowBlack Odonata page 172
172 Göran Sahlén
specific preferences for old-growth trees or other structural parameters while others are affected by anthropogenic disturbance (pers. comm.). She affirms that adult dragonflies are useful bioindicators of disturbances in terrestrial and terrestrial/aquatic ecotone habitats. Her data show that some dragonfly species respond negatively to structural alteration of the vegetation immediately adjacent to their aquatic habitats, and many stenotopic species appear to be sensitive to the forest fragmentation that accompanies timber management or development within the larger landscape. This terrestrial disturbance may directly impact the viability of sensitive adult dragonflies, but it also affects larval viability by influencing the quality of adjacent aquatic habitats. Clearly, trees themselves are important to odonate species, but all the life stages of Odonata must be considered. Most species in a forest could probably live in waters in open areas, but some will not survive unless forest habitat is available at a moderate distance from their breeding waters. We must consider the characteristics of adult habitats such as perches, feeding areas, suitable egg-laying substrates and protection from predators and those of larval habitats including the right climbing or hiding substrate and the structures needed by their prey.
INDICATORS OF UNDISTURBED FOREST WATERS Are there specific indicator species among the dragonflies that might help us identify undisturbed forest waters? In this context I refer to waters in forests that are, to some degree, disturbed (i.e. most of the forests remaining today) rather than to waters in completely undisturbed forests. In the latter case no aquatic indicators are needed to prove what is already known, namely, that the forest is pristine. Palmer (1995) states that species richness (number of species) is one of the most important components of biodiversity, but he also stresses the problem of obtaining complete species lists. It is usually extremely time-consuming, if not impossible, to determine organisms to the species level. Because dragonflies are among the exceptions to this rule, the Odonata is a typical target taxon (Kremen, 1994) along with vascular plants (Lapin & Barnes 1995). Sahlén & Ekestubbe (2001) showed that these two target taxa are coupled in that a high diversity in dragonflies reflects a high diversity in vascular plants. In conservation work the establishment of such relationships is important in order to avoid the setbacks encountered by workers such as Faith and Walker (1996) and Lawton et al. (1998), where the diversity of different groups were not at all interconnected. The selected taxon/taxa should provide information that is useful when taking conservation decisions; a taxon merely indicating “change” is of low practical value. Some examples of ideal criteria when selecting target taxa are:
CyanMagentaYellowBlack Odonata page 173
Specialists vs. generalists in the Odonata – the importance of forest environments
173
1) Easy identification and well-known taxonomy 2) Simple sampling and preparation techniques required. 3) High sensitivity to the changes studied. 4) At least moderate abundance, in order to facilitate monitoring. Dragonflies fulfil these criteria (cf., Kremen et al. 1993, McGeogh 1998, Simberloff 1998), and they have been shown to respond to environmental changes in an easily monitored way (Brown 1991, Sutton & Collins 1991, Clark & Samways 1996, Rith-Najarian 1998, Sahlén 1999, Sahlén & Ekestubbe 2001, Clausnitzer 2004). Dragonflies can therefore be used in many ways as indicators in forest habitats. Counting the numbers of species present, as in Rith-Najarian (1998), Sahlén (1999) and Sahlén & Ekestubbe (2001), is one approach. Here a high number of species is indicative of old-growth/undisturbed forest. A more precise method is to seek out certain species whose presence indicates a high diversity (Sahlén & Ekestubbe 2001, cf., also Suhling et al.,2005, for a selection method used in dry and not-so-forested environments). In Sahlén & Ekestubbe (2001) the species sought were indicative of general species richness in Swedish boreal forests. The method was simple and dependent on a species pool appearing in nested subsets at the sampling sites. This may be determined through an analysis of nestedness (Patterson and Atmar 1986). In brief; when organisms occur in a nested distribution pattern, the species-poor biota are non-random parts of more species-rich biota (Cutler 1994). This kind of distribution pattern is common in dragonflies as well as in a wide range of other organisms (Sahlén & Ekestubbe 2001, Worthen 2003, Kadoya et al. 2004 and others). It should, however, be stressed that forests occur in which the odonate distribution is not nested. I analysed the species distribution of Odonata at 12 sites along seven rivers in Costa Rican tropical rainforest between 1998-2004; the unpublished data indicates that the species pool there is not nested. This may result from the high mobility of the species present (Atmar & Patterson 1993), but obviously also from other factors. Nevertheless, the important point is that nestedness is not something that exists in all forest systems, and that all forest data therefore can not be analysed in this way; a clear limitation of the usefulness of this method. Hence, before choosing indicator species from the middle portion of the nestedness matrix (Sahlén & Ekestubbe 2001), one must ensure that a nested system is actually being studied. If that is the case, the “moderately common” species in the middle portion of the nestedness matrix are suitable as indicators as they are neither too rare nor too common, and because they indicate that all of the common species in the matrix should occur at the locality surveyed. See Sahlén & Ekestubbe (2001) for a detailed description of this selection method. Having selected our indicator species we must also consider them from a regional perspective. Sahlén and Ekestubbe (2001) showed that the selection
CyanMagentaYellowBlack Odonata page 174
174 Göran Sahlén
must be made in each specific area where the indicators are to be used, because ecological preferences and life cycles may vary regionally. I will give a few examples of this, using some species demonstrated to be suitable as indicators in the forests of central Sweden in 2001, but which are not at all useful in the southernmost part of Sweden: Three species of Leucorrhinia — L. albifrons, L. caudalis and L. pectoralis — were shown to be indicators of species richness in central Sweden. They were selective with regard to breeding waters without being extremely uncommon, which means that they were possible to find without excessive effort in the field. During field work in the southern provinces of Halland and Skåne in 2001-2005, these species were, however, rare. This is not surprising as they are mainly forest and bog dwellers, and such habitats are either more rare or more degraded in these parts of Sweden. Hence, while these three species are suitable as diversity indicators in central Sweden, the mere effort required to find them in the south renders their use as indicators there impossible. In central Sweden Enallagma cyathigerum was a useful indicator. In this region it was selective in its choice of breeding sites, occurring in only 15% of the lakes surveyed. In southern Sweden this species showed the opposite pattern as compared to the Leucorrhinia species: it was too common, occurring in more than 90% of all waters surveyed, and hence not suitable as an indicator in that area. Somatochlora flavomaculata (Fig. 15) was included among the indicators in the 2001 paper since it was difficult to find as a larva. The collec-
Fig. 15. Somatochlora flavomaculata, once an indicator of species richness in the boreal forests of central Sweden. Today the species is becoming more and more common and should no longer be considered selective in its choice of breeding waters. Photo by the author.
CyanMagentaYellowBlack Odonata page 175
Specialists vs. generalists in the Odonata – the importance of forest environments
175
tion of a larva in a lake was said to indicate that the locality was thoroughly surveyed. Considering that one may, by chance, find any rare species in the first sample, this was probably not a valid reason for including it among the indicators in the first place. Furthermore, this species has become more and more common in recent years, either by extending its range northwards in Scandinavia or by adapting to a wider range of habitats; it is too early to be precise about the reasons for this change. In the last few years I have found the larva of this species in several waters, including newly constructed wetlands in southern Sweden — wetlands surveyed only briefly by students. Thus, this species evidently has changed ecologically, becoming less selective than it was five years ago, and should no longer be regarded as a selective indicator species. Among the 11 indicators of general species richness proposed for central Sweden, only two are usable in southern Sweden: Pyrrhosoma nymphula and Brachytron pratense seem to show the same selective distribution in both regions, while the other nine species do not. Therefore, indicators of species richness in forests are 1) normally only usable in the area where they were selected, because their distribution patterns, ecology and abundance differ among geographical areas, and 2) sometimes short-lived as diversity patterns change over time (including potential changes resulting from global warming). In reality, perhaps the most efficient way of investigating an environmental or ecological issue is to abstain from using any indicators at all. The effort in finding them often overshadow their usefullness both in a geographical and time perspective. It is normally more efficient to clearly formulate what issues needs to be investigated, and do the survey accordingly. But to answer the question posed in the first sentence of this section: yes, there are definitely indicator species for undisturbed forest waters, but the pattern is complex. If one has a real urge to work on indicators, they should be selected locally and with great care. Extrapolation to other areas and other forest types is not possible.
TO SUM IT UP Forests seem to have what agricultural areas do not. Trees are one thing, but more important are the complex structures in and around the waters, structures used by larvae and adults at various stages of their life history (Fig 16). These structures increase the number of niches available and make it possible for specialised species to colonise the water body. Forestry measures taken today affect these structures, and we should keep in mind that to maintain a high diversity within a forest landscape, several seral stages and many different wetlands, surrounded by a diverse matrix of plants (including
CyanMagentaYellowBlack Odonata page 176
176 Göran Sahlén
Fig. 16. An example of complex structures in the water formed by thin-leafed water plants. Plants of various species will increase the number of niches available for the dragonfly larvae.
trees) must be present. The general rule seems to be that forests harbour specialists, while the open landscape is the playground of the generalists.
ACKNOWLEDGEMENTS I would like to express my gratitude to Janet Rith-Najarian who kindly allowed me to mention results from a paper in preparation and also lent me photographs of dragonflies and key habitats in the Minnesotan boreal forest. Dr Hans Mejlon at the Museum of Evolution, Uppsala, managed to find me a slide scanner when all other possibilities of bringing old slides into the digital age were depleated. My wife Anna has — as usual — saved the day by making sure that my English is actually readable to others than myself. I also wish to thank two anonymous reviewers whose comments greatly improved the structure and content of this chapter.
REFERENCES ASKEW, R.R. 1988. The dragonflies of Europe. Harley Books, Colchester. ATMAR, W. & PATTERSON, B. D. 1993. The Measure of Order and Disorder in the Distribution of Species in Fragmented Habitat. Oecologia 96: 373-382.
CyanMagentaYellowBlack Odonata page 177
Specialists vs. generalists in the Odonata – the importance of forest environments
177
BROWN JR., K. S. 1991. Conservation of neotropical environments and their habitats: insects as indicators. In: Collins, N. M. & Thomas, J. A. (eds), The Conservation of Insects and their Habitats, pp. 350-404, Academic Press, London. BUCHWALD, R. 1992. Vegetation and dragonfly fauna - characteristics and examples of biocenological field studies. Vegetatio 101: 99-107. CLARK, T. E. & SAMWAYS, M. J. 1996. Dragonflies (Odonata) as indicators of biotope quality in the Kruger National Park, South Africa. Journal of Applied Ecology 33: 1001-1012. CLAUSNITZER, V. 2003a. Odonata of African humid forests - a review. Cimbebasia 18: 173-190. CLAUSNITZER, V. 2003b. Dragonfly communities in coastal habitats of Kenya: Indication of biotope quality and the need of conservation measures. Biodiversity and Conservation 12: 333-356. CLAUSNITZER, V. 2004. Guardians of the watershed. In: Clausnitzer, V. & R. Jödicke (eds): Guardians of the watershed. Global status of dragonflies: critical species, threat and conservation. International Journal of Odonatology 7: 111. CUTLER, A. 1994. Nested biotas and biological conservation: metrics, mechanisms and meaning of nestedness. Landscape and Urban Planning 28: 73–82. DIJKSTRA, K.-D. B. & LEMPERT, J. 2003. Odonate assemblages of running waters in the Upper Guinean forest. Archiv für Hydrobiologie 157: 397-412. DEMARCO JR, P. & FURIERI, K. S. 2000. Ecology of Leptagrion perlongum Calvert, 1909: A bromeliad-dweller odonate species. Boletim do Museu de Biologia Mello Leitao Nova Serie 2000: 135-148. FAITH, D. P. & WALKER, P. A. 1996. Environmental diversity: on the best-possible use of surrogate data for assessing the relative biodiversity of sets of areas. Biodiversity and Conservation 5: 399–415. FINCKE, O. M. 1992. Interspecific competition for tree holes: consequences for mating systems and coexistence in Neotropical damselflies. American Naturalist 139: 80-101. FOGH NIELSEN, O. 1998. De danske guldsmede. Danmarks Dyreliv, bind 8. Apollo Books Aps., Stenstrup, Denmark HÖRNBERG, G., OHLSON, M. & ZACKRISSON, O. 1995. Stand dynamics, regeneration patterns and long-term continuity in boreal old-growth Picea abies swampforests. Journal of Vegetation Science 6: 291-298. JOHANSSON, F. & NILSSON, A. N. 1991. Freezing tolerance and drought resistance of Somatochlora alpestris Selys larvae in boreal temporary pools (Anisoptera, Corduliidae). Odonatologica 20: 245-252. JONSEN, I. D. & TAYLOR, P. D. 2000. Fine-scale movement behaviors of calopterygid damselflies are influenced by landscape structure: An experimental manipulation. Oikos 88: 553-562. KADOYA, T., SUDA, S. & WASHITANI, I. 2004. Dragonfly species richness on manmade ponds: effects of pond size and pond age on newly established assemblages. Ecological Research 19: 461-467. KINVIG, R. G. & SAMWAYS, M. J. 2000. Conserving dragonflies (Odonata) along streams running through commercial forestry. Odonatologica 29: 195-208.
CyanMagentaYellowBlack Odonata page 178
178 Göran Sahlén
KIRKTON, S. D. & SCHULTZ, T. D. 2001. Age-specific behavior and habitat selection of adult male damselflies, Calopteryx maculata (Odonata: Calopterygidae). Journal of Insect Behavior 14: 545-556. KREMEN, C. 1994. Biological inventory using target taxa: a case study of the butterflies of Madagascar. Ecological Applications 4: 407–422. KREMEN, C., COLWELL, R. K., ERWIN, T. L., MURPHY, D. D., NOSS, R. F. & SANJAYAN, M. A. 1993. Terrestrial arthropod assemblages: their use in conservation planning. Conservation Biology 7: 796–808. LAPIN, M. & BARNES, B. V. 1995. Using the landscape ecosystem approach to assess species and ecosystem diversity. Conservation Biology 9: 1148–1158. LAWTON, J. H., BIGNELL, D. E., BOLTON, B., BLOEMERS, G. F., EGGLETON, P., HAMMOND, P. M., HODDA, M., HOLT, R. D., LARSEN, T. B., MAWDSLEY, N. A., STORK, N. E., SRIVASTAVA, D. S. & WATT, A. D. 1998. Biodiversity inventories, indicator taxa and effects of habitat modification in tropical forest. Nature 391: 72–76. LE CALVEZ, V. 1998. Les odonates de la forêt domaniale de Notre Dame (Départements du Val-de-Marne et de Seine-et-Marne). Martinia 14: 137-145. MACHADO, A. B. M., MESQUITA, H. G. & MACHADO, P. A. R. 1991. Contribution to the knowledge of Odonata of the Maraca Ecological Station, Roraima. Acta Amazonica 21: 159-173. MCGEOGH, M.A. 1998. The selection, testing and application of terrestrial insects as bioindicators. Biological Reviews 73, 181-201. MUZON, J. 1997. Odonata (Insecta) from Patagonia: Species richness and distributional patterns. Biogeographica 73: 123-133. NORLING, U. 1984. Life-history patterns in the northern expansion of dragonflies. Advances in Odonatology 2: 127-156. ORMEROD, S. J., WEATHERLEY, N. S. & MERRETT, W. J. 1990. The influence of conifer plantations on the distribution of the golden ringed dragonfly Cordulegaster boltoni (Odonata) in upland Wales, UK. Biological Conservation 53: 241-252. PALMER, M. W. 1995. How should one count species? Natural Areas Journal 15: 124–135. PATTERSON, B. D. & ATMAR, W. 1986. Nested subsets and the structure of insular mammalian faunas and archipelagos. Biological Journal of the Linnean Society 28: 65–82 POIRIER, G. G. & SURGEONER, G. A. 1987. Laboratory flow-through bioassays of four forestry insecticides against stream invertebrates. The Canadian Entomologist 119: 755-764. REHFELDT, G. 1986. Distribution and behavior of libellulid dragonflies (Odonata, Libellulidae) during the dry season in Panamanian tropical forests. Amazoniana - Limnologia et Oecologia Regionalis Systemae Fluminis Amazonas 10: 57-62. RITH-NAJARIAN, J. C. 1998. The influence of forest vegetation variables on the distribution and diversity of dragonflies in a northern Minnesota forest landscape: A preliminary study (Anisoptera). Odonatologica 27: 335-351. ROBERT, P. -A., 1958. Les Libellules (Odonates). Delachaux et Niestlé, Neuchâtel and Paris.
CyanMagentaYellowBlack Odonata page 179
Specialists vs. generalists in the Odonata – the importance of forest environments
179
SAHLÉN, G. 1999. The impact of forestry on dragonfly diversity in Central Sweden. The International Journal of Odonatology 2: 177-186. SAHLÉN, G. & EKESTUBBE, K. 2001. Identification of dragonflies (Odonata) as indicators of general species richness in boreal forest lakes. Biodiversity and Conservation 10: 673-690. SAMWAYS, M. J. 1999. Diversity and conservation status of South African dragonflies (Odonata). Odonatologica 28: 13-62. SAMWAYS, M. J. 2003. Conservation of an endemic odonate fauna in the Seychelles archipelago. Odonatologica 32: 177-182. SAMWAYS, M. J. & STEYTLER, N. S. 1996. Dragonfly (Odonata) distribution patterns in urban and forest landscapes, and recommendations for riparian management. Biological Conservation 78: 279-288. SIMBERLOFF, D. 1998. Flagships, umbrellas, and keystones: is single-species management passé in the landscape era? Biological Conservation 83: 247–257. STEWART, D. A. B. & SAMWAYS, M. J. 1998. Conserving dragonfly (Odonata) assemblages relative to river dynamics in an African savanna game reserve. Conservation Biology 12: 683-692. SUHLING, F., SAHLÉN, G., MARTENS, A., MARAIS, E. & SCHÜTTE, C. 2005. Dragonfly assemblages in arid tropical environments: a case study from western Namibia. Biodiversity and Conservation, In press. SUTTON, S. L. & COLLINS, N. M. 1991. Insects and tropical forest conservation. In: Collins, N. M. & Thomas, J. A. (eds), The Conservation of Insects and their Habitats, pp. 350-404, Academic Press, London. SVENSSON, J. M., J. STRAND, G. SAHLÉN & S. WEISNER, 2004. Rikare mångfald och mindre kväve. Utvärdering av våtmarker skapade med stöd av lokala investeringsprogram och landsbygdsutvecklingsstöd. Swedish Environmental Protection Agency, Report No 5362. THEUERKAUF, J. & ROUYS, S. 2001. Habitats of Odonata in the Bialowieza Forest and its surroundings (Poland). Fragmenta Faunistica 44: 33-39. WATANABE, M., MATSUOKA, H. & TAGUCHI, M. 2004. Habitat selection and population parameters of Sympetrumin fuscatum (selys) during sexually mature stages in a cool temperate zone of Japan (Anisoptera: Libellulidae). Odonatologica 33: 169-179. WORTHEN, W. B., 2003. Nested-subset structure of larval odonate assemblages in the Enoree River basin, USA. International Journal of Odonatology 6: 79-89.
CyanMagentaYellowBlack Odonata page 180
This page intentionally left blank
CyanMagentaYellowBlack Odonata page 181
Dragonfly distributional predictive models Japan: relevance of land cover ... Adolfo Cordero Rivera (ed)in2006 Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 181-205. © Pensoft Publishers
Sofia–Moscow
Dragonfly distributional predictive models in Japan: relevance of land cover and climatic variables Yoshitaka Tsubaki & Nobuyuki Tsuji National Institute for Environmental Studies, Tsukuba, 305-8506 Japan
ABSTRACT We constructed dragonfly distributional models (logistic regression models) based on occurrence records collected in the national recording scheme of Japan. Such occurrence records have several shortcomings in that they only record what is present and not what is absent, and sampling efforts are highly variable among recording grid-squares (about 10x10 km). Moreover, the accuracy of logistic regression models is strongly influenced by the presence/absence prevalence. We developed two data screening methods to select ‘reliable’ species presence/absence data sets from presenceonly species assemblage records: exclusion of grid-squares without enough survey efforts, and exclusion of grid-squares out of temperature range in each species. Then we tried to find out landcover-occurrence relationships within the temperature range based on logistic regression models. We obtained statistically significant models for 98 species among all dragonflies inhabiting the main four islands of Japan (128 species). Goodness-offit tests showed that some landcover types significantly affected the occurrence of each species. Area of broad-leaved forests within a grid-square (10x10 km) had positive effects on the occurrence of 57 species, indicating that at least 50% of dragonflies depend on forests. Our analysis also showed that landcover heterogeneity (Shannon-Wiener’s H’) had positive effects on the occurrence of most species (73 among 98 species). We showed three examples of habitat maps generated by the logistic model together with actual occurrence records. We discussed how the model performance might change in relevance to the data screenings we applied.
181
CyanMagentaYellowBlack Odonata page 182
182 Yoshitaka Tsubaki & Nobuyuki Tsuji
INTRODUCTION A crucial step in conservation is determining where animals and plant species occur. This applies to any wildlife including dragonflies. However, conducting complete field inventories of animal occurrences is generally infeasible. Then, animal-habitat models based on environmental surrogate measures are often used to predict species occurrence, absence, or relative abundance (review in Guisan and Zimmermann, 2000; Samways 2004). The first step for generating animal-habitat models is to detect a correlation between a species’ distribution and the attributes common to the habitats that might constitute ecological requirements. Although our knowledge of dragonfly habitat use is limited, our experience to date indicates that species respond to the features of their habitats in a hierarchical manner, from the biotope (e.g., woodland, marsh), through the larval habitat (e.g., pond, stream), to the oviposition site (living macrophytes, rotting wood) (Wildermuth, 1994). Therefore, processes underlying habitat selection of a given species are rather complex (Corbet, 1999). The main purpose of our analyses is to detect correlations between landcover characteristics and the occurrence of a given species in 10 km grid-squares. Therefore, our analysis may reflect mainly biotope level habitat preference of the species. We used dragonfly occurrence records collected in the national recording scheme (National Survey on the Natural Environment). Records reported by the network of volunteer recorders provided, to some extent, comprehensive coverage of the country. These are immensely valuable for determining how well or not species are doing over time, as well as the extent of the geographical ranges of species. The outcome has been the production of an atlas (Japan Integrated Biodiversity Information System), which provides an immediate visual overview of present geographical ranges. These types of maps, based on information in about 10x10 km squares (about 100 km2), have been used to analyze gross range changes of butterflies of Britain, for example, and to predict future ranges (Hill et al, 2002), as well to determine other landscape effects (Warren et al. 2001). However, there are shortcomings with these “record maps”. Firstly, the records are accumulated in an ad hoc manner, resulting in geographically biased records (Dennis and Hardy, 1999). Secondly the data only record what is present and not what is absent. Thirdly, they do not recognize recorder effort that can bias results (Dennis et al., 1999). Fourthly, abundance is neglected though it gives important survival implications for populations. We report here our recent efforts to overcome these shortcomings inherent to the national recording schemes. We have developed a method to
CyanMagentaYellowBlack Odonata page 183
Dragonfly distributional predictive models in Japan: relevance of land cover ...
183
obtain species presence/absence data sets from presence-only species assemblage records. Based on the data sets we tried to find out suitable surrogate measures for the dragonfly-habitat models for all species. The results were used to categorize the diversity of habitat selection in dragonflies and to generate potential habitat maps of each species.
DATA SOURCES Dragonfly records Historical occurrence records of dragonflies in Japanese national recording scheme between 1900 and 1999 consist of 107,717 records, which include 205 (sub) species, though most of the records (>90%) were collected after 1980 (Biodiversity center of Japan, 2002). Each dragonfly record includes information of species, grid code, year and month of collection or sighting. We limited our analysis to the four main islands (Hokkaido, Honshu, Shikoku and Kyushu) in order to avoid island effects (effects of small land area and distance from mainland). The number of 10x10 km gridsquares covering Japan’s mainland is 3,961 and the number of grid-squares in which at least 1 species is recorded is 3,083. Therefore, one fourth of grids have no records. The number of records at each grid-square ranges from 0 to 1400 with an average of 18, and the number of species ranges from 1 to 70 with an average of 7. These figures suggest that occurrence records provide incomplete species lists for most grids, though some of them may provide almost complete lists of species, particularly when the number of occurrence records is large. Historically, 148 species have been recorded within the main islands of Japan. Among them, we used 128 inhabitants for the analysis excluding seasonal migrants and apparent vagrants. Climate data Temperature data were obtained from “Mesh Climate Data 2000” (Japan Meteorological Business Support Center, 2002) that was released from the Japan Meteorological Agency. This dataset includes 1x1 km grid-square temperature data covering the whole of Japan, which was averaged for 30 years between 1970 and 1999. We calculated the average temperature for each 10x10 km mesh and used it for analysis. Landcover data We used a vegetation data set derived from the National Survey on the Natural Environment (Biodiversity Center of Japan, 1999). In this dataset, area of vegetation and land use types (about 358,200 km2) are
CyanMagentaYellowBlack Odonata page 184
184 Yoshitaka Tsubaki & Nobuyuki Tsuji
described with vector data (polygon-shaped), and categorized into 326 types mainly based on the plant community structures within each polygon. In order to simplify our analysis, however, we re-categorized them into 9 landcover types: broad leaved forests including evergreen and deciduous forests (BLF), coniferous forests including cypress and cedar plantations (CF), grassy land (GL), wetland vegetation (WL), bamboo or sasaplant vegetation (BS), paddy field (PF), agricultural land use other than paddy field (AF), urban area including residential area, factories and architecture (UR), and others. Table 1 shows the area and the proportion of each landcover types of main lands and that of selected 361 grids (see below). In addition, we used Shannon-Wiener’s H’ as a measure of landcover heterogeneity, H’ = - ΣPi (log2 Pi), where Pi is a proportion of a given landcover type within a given square-grid.
Table 1. Proportion of landcover types in Japan’s main four islands and selected gridsquares for analysis (see text). Contingency table analysis showed that the composition of landcover types of selected grid-squares are significantly different to that of Japan’s main four islands (G=6730.5, P<0.001). Higher proportion of UR in selected grids probably reflects that it is easy to approach, and lower proportions of CF and BS reflects that recorders usually take little interest in such landcover types. Land cover type Area in main islands BLF CF GL WL BS PF AF UR others Total
123,397 106,639 23,130 3,396 21,174 42,345 28,007 20,841 21,850 390,779 km2
% 31.6 27.3 5.9 0.9 5.4 10.8 7.1 5.3 5.6 100.0
Area in selected 361 grids 10,054 6,860 1,339 404 488 5,848 3,048 5,281 1,383 34,704 km2
% 29.0 19.8 3.9 1.2 1.4 16.9 8.8 15.2 4.0 100.0
Abbreviations: BLF, broad leaved forests including evergreen and deciduous forests; CF, coniferous forests including cypress and cedar plantations; GL, grassy land; WL, wetland vegetation; BS, bamboo or sasa-plant vegetation; AF, agricultural land use other than paddy field; PF, paddy field; UR, urban area including residential area, factories and architecture
CyanMagentaYellowBlack Odonata page 185
Dragonfly distributional predictive models in Japan: relevance of land cover ...
185
ANALYTICAL PROCEDURE Temperature range of each species At a broad scale, the main controlling factors of species’ geographic ranges are probably climatic conditions (temperature). It is not meaningful to analyze the relationship between landcover types and species occurrence where temperature conditions are not suitable. Therefore, we estimated temperature limits of a given species based on the annual average temperatures of all grid-squares where the species was recorded. As a rule, the temperature range of each species was defined as the range where 95% of grids are included. Therefore, minimum and maximum temperature were at 2.5% and 97.5% points respectively. This procedure is probably effective in excluding latitudinal and altitudinal outliers of the distribution. In cases where the target species occurs further south from the main islands, maximum temperature was defined separately as the highest temperature within the main islands (17.8°C). The results of temperature range estimation are shown in Appendix Table 1. It should be noted that the occurrence probability might vary even within the temperature range of each species. Selection of grid-squares for analysis It is expected that the more occurrence records of any species we have within a grid-square, the more its species list will become complete. Let’s suppose that we make several inventory efforts over the years within a grid-square. During the first inventory we may obtain a list of some proportion of species living within this grid-square. In the next inventory we may add some new species into this list, but the list of new species is likely to be smaller than the previous one. The list of new species will become gradually smaller as we repeat this procedure (Fig. 1). The relationship is often described using a negative exponential function relating the number of species (Sr) to the number of records chronologically accumulated (r). This relationship is given by Sr = Smax[1 – exp(-br)], where Smax, the asymptote, is the estimated total number of species in a given grid-square and b is a fitted constant that controls the shape of curve (e.g., Gotelli and Colwell 2001; Lobo & Martin-Peira 2002; Colwell et al. 2004). The curvilinear function was fitted by the quasi-Newton method using Mathematica (v.5.1). We also calculated the 95% confident interval of Sr and determined the adequacy of records in each grid-square. Where the number of species recorded was within the confidence interval, we assumed that the
CyanMagentaYellowBlack Odonata page 186
186 Yoshitaka Tsubaki & Nobuyuki Tsuji
40
Number of species (S)
35
Smax
30 25 20 15 10
Sr = Smax[1 – exp(-br)]
5 0 Cumulative number of records (r)
Fig. 1. A typical relationship between the cumulative number of dragonfly records and the cumulative number of dragonfly species in a grid. As the species list becomes larger, new species are less likely to be added in the list. Number of species present in the grid was estimated by fitting a negative exponential equation (see text).
records made up an almost complete species list. As a result, we selected 361 grids for analysis. The selected grids cover about 10% of the whole area, and the proportion of each land cover type in the selected grids was roughly the same as that of whole land area except for UR, CF and BS (Table 1). Contingency table analysis showed that the composition of landcover types of selected gridsquares was significantly different to that of Japan’s main four islands (G=7369.4, P<0.001). Higher proportion of UR in selected grids is probably due to the easiness of approaches, and lower proportions of CF and BS reflects that dragonfly recorders usually take little interest in such landcover types. This kind of bias is difficult to avoid when we deal with records reported by the network of volunteer recorders. The ability of the model to detect the effects of UR may be stronger and that of CF or BS may be weaker than other variables. Therefore we should be careful in interpreting the results. However, some preliminary analyses reducing the number of UR-rich gridsquares showed that such effects were not large. The selection of grids reduced the number of species for statistical analysis, because some species are recorded only once or less among the selected grids. We therefore excluded these species and consequently we analyzed 126 out of 128 species.
CyanMagentaYellowBlack Odonata page 187
Dragonfly distributional predictive models in Japan: relevance of land cover ...
187
Sensitivity
Construction of logistic models and evaluation Using the selected datasets of 361 grid-squares, we analyzed the effects of landcover type on the occurrence of each dragonfly species using multiple logistic regression models coupled with a stepwise variable selection procedure (JMP v.6.0, SAS). Because logistic regressions results tend to be influenced by extreme prevalence scores, it is necessary to use the same number of presence and absence records. As this condition is not generally satisfied even if we excluded grid-squares out of temperature range, we used weighing methods suggested by King and Zeng (2000) using the prevalence value calculated in the later section. The discrimination ability of logistic regression models was quantified by calculating statistics from a confusion matrix of predictions and observations (Fig. 2) (e.g., Edwards et al. 1996; Boone and Krohn 1999). A species was predicted to be present or absent at a grid-square based on whether the predicted probability for the grid is higher or lower than a specified threshold probability. We used the relative operating characteristic (ROC) curve (Fielding and Bell 1997; Manel et al. 1999; Guisan 2002) to find out a suitable threshold probability. An ROC curve is a plot of the specificity and false positive values of sensitivity obtained by considering a large number of threshold probability values. We show the ROC plot of a logistic model for Calopteryx cornelia as an example (Fig. 3). For a given threshold, sensitivity is the proportion of occupied grids correctly classified by the model as occupied. We used sensitivityCalopteryx cornelia 1.0 specificity sum maximization approach (Cantor et al. 1999, Manel et al. 2001) to deter0.8 mine threshold cut-off to predict distribution. We also calculated overall prediction success 0.6 rate (OPS) [(a+d)/(a+b+c+d)], sensitivity 0.4 (a/a+c) and specificity (d/b+d) as accuracy indices of predictions. 0.2 AUC=0.835
0.0 0.0
Prediction Present
Absent
Present
a
b
Absent
c
d
Observation
Fig. 2. The confusion matrix comparing observed to predicted numbers of present and absent data.
0.2
0.4
0.6
0.8
1.0
1-Specificity Fig. 3. Graphic representation of discrimination performance (ROC curve) of a model for Calopteryx cornelia. AUC: area under the ROC curve. We used sensitivity-specificity sum maximization approach to determine threshold cut-off to predict distribution (see text).
CyanMagentaYellowBlack Odonata page 188
188 Yoshitaka Tsubaki & Nobuyuki Tsuji
A more universal accuracy measure should describe the accuracy of the whole model and not just its performance for a given threshold value. One such measure is the area under the ROC curve. The area under this curve (AUC), expressed as a proportion of the area yielded by a model with perfect accuracy, provides a measure of discrimination ability. This area is equivalent to the Mann-Whitney-Wilcoxon statistic (Hanley and McNeil 1982), and ranges between 0.5 and 1.0, with 0.5 indicating discrimination performances equivalent to a random model and 1.0 indicating complete discrimination for occupied and unoccupied grids.
RESULTS Prevalence of species Among selected grid-squares, we classed them within and out of the range of temperature conditions for a given species. Appendix Table also shows the total number of grids within the temperature range for each dragonfly among selected grids (Nt), and the number of grids in which the dragonfly was actually observed (Np). Np/Nt ratio is the positive prevalence of species. Np/Nt value ranged between 0.0 and 0.9, indicating a wide spectrum in prevalence among species. Model performances Fig. 4a shows the frequency distribution of AUC. Pearce and Ferrier (2000) provide guidelines for interpreting the 0.5-1 ranges. They suggest that values greater than 0.9 indicate an excellent level of discrimination. Values between 0.7 and 0.9 indicate a reasonable level of discrimination, while values between 0.5 and 0.7 indicate poor to marginal discrimination ability. Based on this criteria, models for 73 species among 98 species showed acceptable (AUC greater than 0.7) levels of discrimination, while models for 25 species showed poor levels of discrimination. Fig. 4b,c and d show frequency distributions of OPS, sensitivity and specificity, respectively. Average OPS was 0.697 (± 0.066 s.d.). Average sensitivity and specificity were 0.744 (± 0.101 s.d.) and 0.666 (± 0.101 s.d.), respectively. Effects of landcover type on dragonflies Results of model evaluations are summarized in Appendix table. We could obtain 98 statistically significant models among 128 dragonfly species. As results of goodness-of-fit tests, we could identify which landcover types explain and how strongly each landcover type (positively or negatively) is associated with the occurrence of each species, as well as to assess the
CyanMagentaYellowBlack Odonata page 189
30
Frequency (%)
Frequency (%)
Dragonfly distributional predictive models in Japan: relevance of land cover ...
a 20 10 0
0.5
0.6
0.7
0.8
0.9 1.0
0.0
0.2
0.4
0.6
Sensitivity
0.8 1.0
0.2
0.4
0.6
0.8 1.0
Overall prediction success
Frequency (%)
Frequency (%)
c
b
0.0
Area under ROC curve 35 30 25 20 15 10 5 0
35 30 25 20 15 10 5 0
189
35 30 25 20 15 10 5 0
d
0.0
0.2
0.4
0.6
0.8 1.0
Specificity
Fig. 4. Discriminative performance of logistic models applied to 98 dragonfly species. Model-building data: grid-squares out of species temperature range were excluded from 361 well-surveyed grid-squares. (a) Frequency distribution of area under ROC curve (AUC), (b) Overall prediction success (OPS), (c) Sensitivity, and (d) Specificity.
temperature effects on it (Table 2 and 3). We will briefly describe general features of these parameters Temperature had positive effects (P<0.05) on 55 species out of 98 species, while on 24 species it had a negative effect, indicating that most of dragonflies prefer warmer climate conditions. For the remaining 19 species temperature had no significant effect. These species may have alternative responses to temperature conditions, including unimodal or uniform effects of temperature, or it is merely due to small sample size. Broad leaved forests (BLF) had positive effects (preference) on 57 out of 98 species, and negative effects (avoidance) on 8 species. All Calopterygidae preferred BLF, and most of Aeshnidae, Gomphidae and Corduliidae also preferred BLF. On the other hand, the preference to BLF is variable within Coenagrionidae and Libellulidae. Coniferous forests (CF) had positive effects on 43 species and negative effects on 26 species. All Calopterygidae preferred CF, and most of Gomphidae also preferred CF. Most of species that showed preference to CF also preferred BLF (39 out of 43 species). These results probably indicate that forest dragonflies generally prefer BLF to CF, but only a few of them show clear distinction between BLF and CF. Grassy land (GL) had positive and negative effects on 6 and 18 species respectively, and it had no significant effects on 74 species, suggesting that this kind of habitat was not a critical habitat for most dragonfly species.
1 2 3 4 5 8 10 11 12 13 14 16 18 19 21 22 24 25 27 29 30 31 32 33 34 35 40 41 42 43
species
Calopteryx japonica Calopteryx atrata Calopteryx cornelia Mnais pruinosa Mnais costalis Lestidae Lestes sponsa Lestes japonicus Lestes temporalis Indolestes peregrinus Sympecma paedisca Platycnemididae Platycnemis foliacea sasakii Copera annulata Coenagrionidae Agriocnemis femina oryzae Mortonagrion selenion Ceriagrion melanurum Ceriagrion nipponicum Aciagrion migratum Ischnura senegalensis Ischnura asiatica Cercion calamorum calamorum Cercion sieboldii Cercion sexlineatum Cercion hieroglyphicum Cercion plagiosum Coenagrion terue Coenagrion lanceolatum Epiophlebiidae Epiophlebia superstes Petaluridae Tanypteryx pryeri Aeshnidae Oligoaeschna pryeri Boyeria maclachlani
Calopterygidae
No family
XX XXX
X OOO XXX XXX OOO OOO OOO
XXX XXX XXX
XXX XXX
OO XXX
XXX X XXX O OOO
CF
XXX XXX
XX
XXX X XX
XXX
XXX X
OOO
X
OO
OO
XX
X XXX
X O OOO OOO OOO
OOO
OOO
XXX XXX XXX
X XXX
X
XXX XXX X X XXX X
Landcover heterogenity, H’
OOO XXX
UR
XXX XXX XXX XX XXX XXX XXX XXX XXX XXX
AF
XXX XXX XXX XXX XXX OOO XXX OOO XX XXX XXX XX OO X XXX OOO XXX O XXX XX XXX X
Landcover type GL WL BS PF
XXX XXX OOO XXX XXX XXX OO XXX OO O OOO OOO OO OOO XXX OOO O XXX XXX OO XXX X XXX XXX OOO XXX XXX XXX O OO OO OO XXX O X X XXX XXX O OOO OOO OOO OOO O O X XXX O OOO OO XXX OO XXX OOO OOO XXX O OO OOO XXX O O XXX XXX XXX XXX O O XXX XXX XXX XXX OOO O
XXX XXX XXX XXX XXX
Temperature BLF
Table 2. Effects of temperature and landcover types on the presence/absence of 98 dragonfly species estimated by logistic models.
CyanMagentaYellowBlack Odonata page 190
190 Yoshitaka Tsubaki & Nobuyuki Tsuji
Planaeschna milnei Aeschnophlebia longistigma Aeschnophlebia anisoptera Gynacantha japonica Polycanthagyna melanictera Aeshna juncea juncea Aeshna nigroflava Anaciaeschna martini Anax parthenope julius Anax nigrofasciatus nigrofasciatus Anisogomphus maacki Gomphus postocularis Asiagomphus melaenops Asiagomphus pryeri Davidius nanus Davidius fujiama Davidius moiwanus moiwanus Lanthus fujiacus Trigomphus citimus tabei Trigomphus melampus Trigomphus interruptus Trigomphus ogumai Stylogomphus suzukii Sinogomphus flavolimbatus Nihonogomphus viridis Onychogomphus viridicostus Sieboldius albardae Ictinogomphus pertinax Sinictinogomphus clavatus Anotogaster sieboldii
44 45 46 47 48 49 50 54 55 56 57 Gomphidae 61 62 63 64 65 66 67 68 69 70 71 72 74 75 76 77 78 79 81 Cordulegastridae OOO OO XXX OOO XXX XXX XXX XXX XXX XXX XXX XXX XXX XXX
XXX X XXX
XXX XX XXX XXX XXX OOO OOO XXX XXX XXX XXX XXX XXX XXX XXX XXX
O
O
UR
XXX O OOO OOO X XX XXX OO O X XXX
XXX
AF
XXX XXX XXX XXX XXX X XXX X XXX XXX XXX XXX O XX XXX XXX O XX XXX XXX XXX XXX OO OOO O OOO XXX OO OOO XXX XX OO X OOO OOO O XXX XXX O XXX OOO XXX XXX XXX OO XX XXX XXX OOO XXX X XXX XXX XXX XXX XXX XXX XXX XX XXX XXX XXX XXX OO XXX OOO OOO XX XX XXX XXX X XX XXX
OO
O
XXX
O
OO X
Landcover type GL WL BS PF
X
CF
XXX XXX OOO OO XXX XXX XXX
Temperature BLF
species
No family
Table 2. Continued.
X XX XXX XXX XXX XXX XXX XXX XXX XXX XXX XXX XXX
XX
X XXX X XXX XXX XXX XXX XXX
XXX XXX
XXX XX
Landcover heterogenity, H’
CyanMagentaYellowBlack Odonata page 191
Dragonfly distributional predictive models in Japan: relevance of land cover ...
191
Epophthalmia elegans elegans Macromia daimoji Macromia amphigena amphigena Epitheca marginata Epitheca bimaculata sibirica Cordulia aenea amurensis Somatochlora arctica Somatochlora uchidai Somatochlora viridiaenea Somatochlora clavata Lyriothemis pachygastra Libellula quadrimaculata asahinai Libellula angelina Orthetrum albistylum speciosum Orthetrum japonicum japonicum Orthetrum triangulare melania Nannophya pygmaea Deielia phaon Crocothemis servilia mariannae Sympetrum pedemontanum elatum Sympetrum darwinianum Sympetrum frequens Sympetrum striolatum imitoides Sympetrum eroticum eroticum Sympetrum kunckeli Sympetrum parvulum Sympetrum risi Sympetrum infuscatum Sympetrum baccha matutinum Sympetrum gracile
82 Corduliidae 83 84 85 86 87 88 92 93 94 95 Libellulidae 97 98 101 102 103 104 105 106 107 108 109 110 111 112 113 116 117 118 119 XXX
XXX XXX OOO
OOO XXX OOO
XXX OO
OO OOO XXX X OOO XXX XXX X XXX
XXX XXX XXX XXX OOO OOO
CF
AF
UR
XXX
XXX OO
OOO XXX XX
XX
X XXX XXX XXX O XXX
XX OOO OOO OOO X OO XXX XXX XXX O XXX XXX O
XX X OOO OO XXX XXX OOO
OO
XX XXX
XXX XX XXX XXX
OOO OO
XXX XXX XXX XXX XX X XXX OOO XXX XXX
Landcover type GL WL BS PF
O XXX XXX XXX XX XXX XXX XXX XXX XXX XXX XXX XXX XXX OO X XX XXX XX O XXX XXX X XX XXX XXX XXX XXX XXX O OOO OOO OOO XXX XXX X XX XX X OOO X OOO OOO O XXX XXX OOO OOO X XXX XXX OOO OO OOO OOO XXX OO XXX
Temperature BLF
species
No family
Table 2. Continued.
XX XXX
XXX XX
XXX
XX XXX XX XX
XXX XXX
XXX XXX XXX XXX XXX
XXX XX O
Landcover heterogenity, H’
CyanMagentaYellowBlack Odonata page 192
192 Yoshitaka Tsubaki & Nobuyuki Tsuji
Sympetrum maculatum Sympetrum speciosum speciosum Sympetrum croceolum Sympetrum uniforme Leucorrhinia dubia orientalis Pseudothemis zonata Rhyothemis fuliginosa Tramea virginia
120 121 122 123 124 126 127 128
OOO XXX OOO XXX O XXX XXX XXX
CF
X XX O O
X
XXX OOO XX XXX XXX O
Temperature BLF
AF
UR
XX OOO O XX OOO OOO
X X OOO OOO
XXX
OOO OOO XXX OOO XXX X OOO X OOO X XXX OOO XXX
Landcover type GL WL BS PF
XX XXX XXX
XXX XXX XXX XXX
Landcover heterogenity, H’
55 19 24
Temperature 57 33 8
BLF 43 29 26
CF 6 74 18
GL
23 57 18
WL
5 72 21
BS
49 41 8
PF
9 58 31
AF
52 40 6
UR
73 24 1
H’
Dragonfly distributional predictive models in Japan: relevance of land cover ...
Abbreviations, same as Table 1. H’: Landcover heterogeneity calculated as Shannon-Weaner’s index.
Positive effects Non significant Negative effects
Landcover
Table 3. Number of species (N=98) on which temperature and landcover types showed positive and negative effects on their occurrence (P<0.05).
Abbreviations, same as Table 1. H’: Landcover heterogeneity calculated as Shannon-Weaner index.
Note: Pearson’s chi-square test for likelihood-ratio P values: Positive effects; P < 0.05 = X, P < 0.01 = XX, P < 0.001 = XXX Negative effects; P < 0.05 =O, P < 0.01 = OO, P < 0.001 = OOO.
species
No family
Table 2. Continued.
CyanMagentaYellowBlack Odonata page 193
193
CyanMagentaYellowBlack Odonata page 194
194 Yoshitaka Tsubaki & Nobuyuki Tsuji
Wetland vegetation (WL) had positive and negative effects on 23 and 18 species respectively, and it had no significant effects on 57 species. Most of Coenagrionidae, Corduliidae and Libellulidae preferred WL, while some Calopterygidae, Lestidae, Aeshnidae and Gomphidae avoided WL. Bamboo vegetation (BS) had positive and negative effects on 5 and 21 species respectively, and no significant effects on 72 species. No odonate family showed consistent preference or avoidance to BS, however, BS was generally avoided by most dragonfly species. Paddy field (PF) had positive and negative effects on 49 and 8 species respectively. Most of Lestidae, Coenagrionidae, Aeshnidae, Corduliidae and Libellulidae showed preference to PF, reflecting that paddy fields are important habitat for various dragonfly species. However, some Gomphidae avoided PF. Agricultural field (AF) excluding paddy field had positive and negative effects on 9 and 31 species respectively. No particular dragonfly family showed consistent preference or avoidance to AF. Urban area (UR) had positive and negative effects on 52 and 6 species respectively. This was an unexpected result to us, however this result might reflect that UR area usually consisted of small but various types of landcover: i.e., private gardens with small ponds, cemetery parks, parks with ponds, small forests, school grounds. Sampling bias toward UR might increased the number of significant results, but it is still apparent that many dragonflies are inhabitants of urban areas. Landcover heterogeneity (H’) had positive effects on 73 species. Although one species showed a marginally significant negative effect (Epitheca bimaculata sibirica), the results suggest that dragonflies generally require multiple landcover types within about 10x10 km size of habitat. Generation of habitat maps Based on logistic models constructed using presence/absence data from 361 selected square-grids, together with landcover data for the whole area (3961 grids), we could extrapolate occurrence probabilities even for gridsquares without any dragonfly records. We show predicted habitat maps for three species in Fig. 5: Calopteryx cornelia, Ischnura senegalensis and Orthetrum albistylum speciosum, which are representatives of three different levels of prevalence (Np/Nt = 0.405, 0.572, 0.963, respectively). Dark squares in each predicted habitat map represent occurrence probability higher than the threshold cut-off. A grid map of occurrence records is shown immediately to the right of the predicted map. Even on the commonest species (O. a. speciosum), it is suggested that there are many grids in which the species is expected to occur but not recorded yet.
CyanMagentaYellowBlack Odonata page 195
Dragonfly distributional predictive models in Japan: relevance of land cover ...
195
a) Calopteryx cornelia
Model prediction Present Absent
Actual records Present
b) Ischnura senegalensis
Model prediction Present Absent
Actual records Present
c) Orthetrum albistylum speciosum
Model prediction Present Absent
Actual records Present
Fig. 5. Habitat maps predicted for three dragonflies: (a) Calopteryx cornelia, (b) Ischnura senegalensis and (c) Orthetrum albistylum speciosum. Dark squares in each predicted habitat map represent grids with occurrence probability higher than the threshold cut-off estimated (see Fig. 2). Grid maps of occurrence records are shown immediately to the right of the predicted maps.
CyanMagentaYellowBlack Odonata page 196
196 Yoshitaka Tsubaki & Nobuyuki Tsuji
DISCUSSION
Frequency (%)
Frequency (%)
Frequency (%)
Predicting species distributions is an important step for environmental conservation and biodiversity management. For this purpose, many modeling techniques to predict species presence/absence have been developed (e.g., Fielding and Bell 1997, Mantel et al. 1999, Austin 2002). Model prediction is largely influenced by the prevalence of model70 building data, and several researchers tested the 60 a 50 effects of the prevalence on assessing indices for 40 model performance (e.g., King and Zeng 2000; 30 20 Liu et al 2005). However, there are some more 10 practical problems before building models. 0 0.0 0.2 0.4 0.6 0.8 1.0 Any wildlife distributional predictive modPseudo positive prevalence els require an accurate presence/absence 25 dataset. However, it is not always easy to ob20 b tain a high quality model-building data. One of 15 the serious problems in obtaining an accurate 10 presence/absence data comes from characteris5 tics of “absence” records. Absence records are 0 0.0 0.2 0.4 0.6 0.8 1.0 almost always less reliable than present records, Positive prevalence because we can be confident about the pres25 ence of species if we observed one or more indi20 c 15 viduals within a local area but we are generally 10 less confident about absence even if we did not 5 see any individual during many times of sur0 veys within the same area. Reliability of ab0.0 0.2 0.4 0.6 0.8 1.0 Positive prevalence sence records is expected to increase only by Fig. 6. Frequency distributions increasing survey efforts, although perfect conof positive prevalence estimat- fidence of absence will not be obtained practied for 98 dragonfly species. (a) cally. We believe that the method used here is pseudo positive prevalence an efficient way to select well-surveyed squarewithout any data screening (all grids. Fig. 6a shows the frequency distribution the available presence-absence of pseudo-positive prevalence for all the availinformation). (b) positive prev- able presence-absence information (without any alence calculated using 361 data screening). While Fig. 6b shows the frewell-surveyed grid-squares. (c) quency distributions of positive prevalence for positive prevalence calculated well-surveyed square-grids without considering using well-surveyed gridsquares excluding grids out of upper and lower temperature limits of each spedistributional temperature cies. It is apparent that model building without any data screening is misleading. Therefore range for each species.
CyanMagentaYellowBlack Odonata page 197
Dragonfly distributional predictive models in Japan: relevance of land cover ...
197
30
Frequency (%)
Frequency (%)
the inclusion of poorly surveyed grids into model-building data always lead to underestimation of positive prevalence. Several accuracy indices derived from a confusion matrix are usually used to assess the model prediction (for example, OPS (operational sex-ratio), Sensitivity, Specificity). However, some values are sensitive to prevalence (p). For example, OPS is affected by the prevalence since OPS = p · sensitivity – (1-p) · specificity (Ruttiman 1994), where sensitivity is the ratio of correctly predicted positive cases to the total number of positive cases and specificity is the ratio of correctly predicted negative cases of the total number of negative cases. Liu et al. (2005) examined the effects of prevalence of model-building data on indices of model predictive ability (including OPS, sensitivity, and specificity) in relation to threshold determining approaches (including sensitivity-specificity sum maximization approach that is used in this study). They confirmed that OPS is highly sensitive to very low or very high prevalence, while sensitivity and specificity are less sensitive to prevalence when sensitivity-specificity sum maximization approach was used. Our second data screening procedure was to discard meaningless records from model-building data by establishing upper and lower temperature limits, which aimed to balance the numbers of presence and absence records as much as it is possible. We will now examine the effects of this data screening on AUC and on three indices of model performance (OPS, sensitivity and specificity). Fig. 7 shows frequency distributions of AUC, OPS, sensitivity and specificity of
a 20 10 0
0.5
0.6
0.7
0.8
0.9 1.0
0.0
0.2
0.4
0.6
Sensitivity
0.8 1.0
0.2
0.4
0.6
0.8 1.0
Overall prediction success
Frequency (%)
Frequency (%)
c
b
0.0
Area under ROC curve 35 30 25 20 15 10 5 0
35 30 25 20 15 10 5 0
35 30 25 20 15 10 5 0
d
0.0
0.2
0.4
0.6
0.8 1.0
Specificity
Fig. 7. Discriminative performance of logistic models applied to 98 dragonfly species. Model-building data: 361 well-surveyed grid-squares. (a) Frequency distribution of area under ROC curve (AUC), (b) Overall prediction success (OPS), (c) Sensitivity, and (d) Specificity.
CyanMagentaYellowBlack Odonata page 198
198 Yoshitaka Tsubaki & Nobuyuki Tsuji
well-surveyed grids when upper and lower temperature limits were not taken into account. Average AUC was 0.773 (± 0.076 s.d.), which was slightly larger than that of well-surveyed grids with establishments of temperature limits (Fig. 4; 0.754 ± 0.068 s.d.), but there was no significant difference between them (ttest, P=0.09). Average OPS was 0.701 (± 0.082 s.d.) that was almost the same as Fig. 4b, and there was no difference between them (t-test, P=0.730). Average sensitivity of Fig. 7c was 0.786 (± 0.121 s.d.) and was significantly larger than that of Fig. 4c (t-test, P=0.009). Average specificity of Fig. 7d was 0.661 (± 0.118 s.d.) and was not different to that of Fig.4d (t-test, P=0.769). In conclusion, the exclusion of grid-squares out of species temperature range did not generally increase the model predictive ability measured as OPS, sensitivity or specificity. However, the effect of the establishment of temperature range was variable from species to species. This may be because this treatment reduced the size of model-building data. The difference in sensitivity between models with and without temperature limitations was negatively correlated with the difference in specificity between models (Fig. 8). Therefore, it is a matter of choice which model-building data we should use in the conservation practice. For example, if a model is used to find grids likely be present, a modelbuilding data that gives high sensitivity may be preferred. On the other hand, if a model is used to find grids likely be absent, a model-building data that gives high specificity may be preferred. We have shown that relatively simple logistic models have the ability to describe habitat preferences and produce habitat maps for dragonfly species after careful data screening. The logistic regression models we used for the analysis were successful in describing landcover-habitat relationships for 98 species, with acceptable levels of model performance for 73 species. However,
Increase in specificity
0.5
r 2=0.414 P<0.0001
0.4 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3 -0.4 -0.5 -0.4
-0.2
0.0
0.2
0.4
Increase in sensitivity Fig. 8. Relationship between the increases (or decreases) in sensitivity and specificity caused by the exclusion of grid-squares out of species temperature range from modelbuilding data.
CyanMagentaYellowBlack Odonata page 199
Dragonfly distributional predictive models in Japan: relevance of land cover ...
199
we still have 25 species with poor levels of model performance and some more species not analyzed properly. One way to improve the model performance is to increase the number of reliable gird-squares. However, it is not always an efficient procedure, because most species have a small geographic range. It might be useful to change the grid size, for example from 10km (as used here) to 1 km or smaller, and make sampling in a more confined area. Cowley et al. (1999) found, by mapping at the fine scale, British butterflies to be declining faster than conventional coarse-scale maps suggested. We have not used river data for our analyses, because digital cartography of the small (first order) streams is not available at this moment. This means that we assumed there are no 10 km grids without streams. Although this is generally correct, it is not always the case. Incorporation of river data may improve our predictions particularly for stream dwelling dragonflies (mainly Calopterygidae and Gomphidae). By analyzing habitat relational models independently on all species, we can give insight to habitat preference of species in a quantitative manner, as well as rank species according to their sensitivities to environmental change. When models for all (or most) species are constructed, the difference in species composition between a list predicted from the models and a list from actual records would be useful as an indicator of environmental conditions of the grid.
ACKNOWLEDGEMENTS We thank Kaoru Imaizumi-Jonathan and Chihiro Kato for their help with data collection, analysis, and modeling. Comments by two anonymous reviewers improved this paper very much. This work was partly supported by the Global Environment Research Fund of the Ministry of the Environment of Japan.
REFERENCES AUSTIN, M.P. 2002. Spatial prediction of species distribution: an interface between ecological theory and statistical modeling. Ecological Modelling 157:101-118. BIODIVERSITY CENTER OF JAPAN. 1999. The national survey on the natural environment: Vegetation map. CD-version. Ministry of the Environment, Japan (in Japanese) BIODIVERSITY CENTER OF JAPAN. 2002. The national survey on the natural environment: Report of the distributional survey of Japanese animals (Dragonflies). Ministry of the Environment, Japan (in Japanese) BOONE R. B. & KROHN W. B. 1999. Modeling the occurrence of bird species: are the errors predictable? Ecological Applications 9: 835-848.
CyanMagentaYellowBlack Odonata page 200
200 Yoshitaka Tsubaki & Nobuyuki Tsuji
CANTOR, S.B., SUN, C.C., TORTOLERO-LUNA, G., RICHARDS-KORTUM, R. & FOLLEN, M. 1999. A comparison of C/B ratios from studies using receiver operating characteristic curve analysis. Journal of Clinical Epidemiology, 52: 885-892. CORBET P. S. 1999. Dragonflies, behaviour and ecology of Odonata. Harley Books, Essex. COWLEY, M. J. R., THOMAS, C. D., THOMAS, J. A. & WARREN, M. S. 1999. Flight areas of British butterflies: assessing species status and decline. Proceedings of the Royal Society of London, B 266: 1587-92. DENNIS, R. L. H. & HARDY, P. B. 1999. Targeting squares for survey: predicting species richness and incidence of species for a butterfly atlas. Global Ecology and Biogeography Letters 8: 443-454. DENNIS, R. L. H., SPARKS, T. H. & HARDY, P. B. 1999. Bias in butterfly distribution maps: the effects of sampling efforts. Journal of Insect Conservation 3: 33-42. EDWARDS T. C., DESHLER E.T., FOSTER D. & MOISEN G.G. 1996. Adequacy of wildlife habitat relation models for estimating spatial distribution of terrestrial vertebarates. Conservation Biology 10: 263-270. FIELDING A. H. & BELL J. F. 1997. A review of methods for the assessment of prediction errors in conservation presence/absence models. Environmental Conservation 24: 38-49. GOTELL N.J. & COLWELL R.K. 2001. Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness. Ecology Letters 4: 379-391. COLWELL, R. K., C. X. MAO & J. CHANG. 2004. Interpolating, extrapolating, and comparing incidence-based species accumulation curves. Ecology 85: 27172727. GUISAN A. 2002. Semiquantitative response models for predicting the spatial distribution of plant species. In: Scott, J.M., Heglund, P.j., Samson F., Haufler, J. Morrison M., Raphael M. & Wall B. (eds.), Predicting species occurrences: issues of accuracy and scale, pp. 315-334, Island Press, Covelo, California. GUISAN, A. & ZIMMERMANN, N. E. 2000. Predictive habitat distribution models in ecology. Ecological Modelling 135: 147-186. HANLEY, J.A. & MCNEIL, B.J. 1982. The meaning and use of the area under receiver operating characteristic (ROC) curve. Radiology 143: 29-36. HILL, J. K., THOMAS, C. D., FOX, R., TELFER, M. G., WILLIS, S. G., ASHER, J. & HUNTLEY, B. 2002. Responses of butterflies to twentieth century climate warming: implications for future ranges. Proceedings of the Royal Society of London, B 269: 2163-2171. JAPAN METEOROLOGICAL BUSINESS SUPPORT CENTER. 2002. Mesh climatic data of Japan. CD version. Japan Meteorological Agency. KING, G. & ZENG, L. 2000. Logistic regression in rare events data. The Global Burden of Disease 2000 in Aging Populations, Research Paper No. 2. LIU C., BERRY, P.M., DAWSON, T.P. & PEARSON R.G. 2005. Selecting thresholds of occurrence in the prediction of species distributions. Ecography 28: 385-/393.
CyanMagentaYellowBlack Odonata page 201
Dragonfly distributional predictive models in Japan: relevance of land cover ...
201
LOBO, J.M. & MARTÍN-PIERA, F. 2002. Searching for a predictive model for species richness of Ibeerian dung beetle based on spatial and environmental variables. Conservation Biology 16: 158-173. MANTEL, S., DIAS, J.-M. & ORMEROD, S.J. 1999. Comparing discrimination analysis, neural networks and logistic regression for predicting species distributions: a case study with a Himalayan river bird. Ecological Modelling 120: 337-347. MANTEL, S., WILLIAMS, H.C. & ORMEROD, S.J. 2001. Evaluating presence-absence models in ecology: the need to account for prevalence. Journal of Applied Ecology 38: 921-931. PEARCE, J. L. & FERRIER, S. 2000. Evaluating the predictive performance of habitat models developed using logistic regression. Ecological Modeling 133: 225-245. RUTTIMAN, U.E. 1994. Statistical approaches to development and validation of predictive instruments. Critical Care Clinics 10: 19-35. SAMWAYS M. J. 2004. Insect diversity conservation. Cambridge University Press, Cambridge. WARREN, M. S., HILL, J. K., THOMAS, J. A., et al. 2001. Rapid responses of British butterflies to opposing forces of climate and habitat change. Nature 414: 65-69. WILDERMUTH 1994. Habitatselektion bei Libellen. Advances in Odonatology 6: 223-257.
Calopterygidae
species
Calopteryx japonica Calopteryx atrata Calopteryx cornelia Mnais pruinosa Mnais costalis MegapodaRhipidolestes aculeatus grionidae Rhipidolestes hiraoi Lestidae Lestes sponsa Lestes dryas Lestes japonicus Lestes temporalis Indolestes peregrinus Sympecma paedisca Platycnemididae Platycnemis foliacea sasakii Platycnemis echigoana Copera annulata Copera tokyoensis Coenagrionidae Agriocnemis femina oryzae Mortonagrion selenion Mortonagrion hirosei Ceriagrion melanurum Ceriagrion nipponicum Ceriagrion auranticum ryukyuanum 24 Aciagrion migratum 25 Ischnura senegalensis 26 Ischnura elegans elegans
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
No family
specificity
15.5 16.4 16.1 16.3 15.2 16.8 16.1 16.3 7.1 15.9 16.0 17.8 17.8 15.4 12.3 16.2 15.0 17.6 15.9 16.3 16.6 16.5 17.8 16.2 17.8 9.4
10.9 11.4 4.7
2
8.8 9.5 7.8 8.7 4.9 10.0 9.0 3.2 1.6 8.8 7.7 7.7 4.7 9.8 7.6 6.6 12.8 14.7 8.3 10.7 7.7 12.2 14.6
1
1600 1201 1263
2145 1658 2160 1935 2679 2036 2308 3224 995 2333 2209 2247 3338 1825 1368 2542 775 528 2329 1845 2334 1295 572
3
187 533 4
133 600 506 552 477 50 44 429 12 65 487 543 246 87 14 457 11 24 158 32 461 91 16
4
Main four islands of Japan (3961 grid-squares)
307 297 31
302 325 333 332 303 318 328 361 20 323 332 337 356 282 69 347 190 94 330 310 337 274 105
5
83 170 1
37 221 135 154 140 4 18 193 5 33 210 205 115 28 4 198 10 9 72 13 185 44 2
6
8 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 1.000 0.974 <0.0001 1.000 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 1.000 <0.0001 0.997 <0.0001 <0.0001 1.000 <0.0001 <0.0001 1.000
0.759 0.719 0.835 0.788 0.759 . . 0.690 . 0.732 0.736 0.686 0.673 0.804 . 0.685 . 0.878 0.669 . 0.590 0.817 .
9
0.270 <0.0001 0.783 0.572 <0.0001 0.816 0.032 0.9991 .
0.123 0.680 0.405 0.464 0.462 0.013 0.055 0.535 0.250 0.102 0.633 0.608 0.323 0.099 0.058 0.571 0.053 0.096 0.218 0.042 0.549 0.161 0.019
7
selected 361 grid squares
0.487 0.570 .
0.464 0.498 0.529 0.493 0.532 . . 0.576 . 0.423 0.472 0.472 0.508 0.407 . 0.582 . 0.506 0.545 . 0.480 0.516 .
10
146 104 .
169 68 153 123 112 . . 135 . 210 77 70 148 159 . 124 . 65 192 . 69 96 .
11
78 23 .
96 36 45 55 51 . . 101 . 80 45 62 93 95 . 25 . 20 66 . 83 56 .
12
21 53 .
9 63 23 31 36 . . 33 . 16 52 59 37 4 . 90 . 0 31 . 54 88 .
13
62 117 .
28 158 112 123 104 . . 92 . 17 158 146 78 24 . 108 . 9 41 . 131 97 .
14
Performance of lositic models
0.678 0.744 .
0.652 0.695 0.796 0.741 0.713 . . 0.629 . 0.703 0.708 0.641 0.635 0.649 . 0.669 . 0.787 0.706 . 0.593 0.704 .
15
0.7470 0.6882 .
0.7568 0.7149 0.8296 0.7987 0.7429 . . 0.7360 . 0.5152 0.7524 0.7122 0.6783 0.8571 . 0.5455 . 1.0000 0.5694 . 0.7081 0.5243 .
16
0.6518 0.8189 .
0.6377 0.6538 0.7727 0.6910 0.6871 . . 0.5720 . 0.7241 0.6311 0.5303 0.6141 0.6260 . 0.8322 . 0.7647 0.7442 . 0.4539 0.6316 .
17
1 - lower temperture limit; 2 - higher temperature limit; 3 - Number of grid-squares within temperature range; 4 - Number of grid-squares with at least 1 record; 5 - Number of gridsquares within temperature range, Nt; 6 - Number of grid-squares with records, Np; 7 - Positive prevalence , Np/Nt; 8 - P of goodness-of-fit test; 9 - AUC (Area under ROC curve); 10 - Threshold cut-off to predict distribution; 11 - true negative; 12 - false positive; 13 - false negative; 14 - true positive; 15 - CCR Correct classification rate; 16 - sensitivity; 17 -
Appendix table. Temperature range estimated from speices distributional records, number of grid-squares within temperature range within main four islands of Japan, number of grid-squares within temperasture range among selected 361 grid-squares (see text), and number of those with distributional records, and performance of logistic models for 128 dragonfly species inhabiting main four islands of Japan.
CyanMagentaYellowBlack Odonata page 202
202 Yoshitaka Tsubaki & Nobuyuki Tsuji
Ischnura asiatica Enallagma boreale circulatum Cercion calamorum calamorum Cercion sieboldii Cercion sexlineatum Cercion hieroglyphicum Cercion plagiosum Coenagrion terue Coenagrion lanceolatum Coenagrion ecornutum Coenagrion hylas Erythromma humerale Nehalennia speciosa Epiophlebia superstes Tanypteryx pryeri Oligoaeschna pryeri Boyeria maclachlani Planaeschna milnei Aeschnophlebia longistigma Aeschnophlebia anisoptera Gynacantha japonica Polycanthagyna melanictera Aeshna juncea juncea Aeshna nigroflava Aeshna subarctica subarctica Aeshna mixta soneharai Anaciaeschna jaspidea Anaciaeschna martini Anax parthenope julius Anax nigrofasciatus nigrofasciatus Anisogomphus maacki Stylurus oculatus Stylurus nagoyanus Stylurus annulatus Gomphus postocularis Asiagomphus melaenops
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 60 61 62
Gomphidae
Epiophlebiidae Petaluridae Aeshnidae
species
No family
Appendix table. Continued.
7.1 0.9 2.6 7.5 11.4 6.8 10.0 3.5 2.4 1.7 1.6 3.5 2.9 4.3 7.4 6.9 9.6 8.5 8.7 12.8 9.8 10.3 2.3 2.9 1.9 5.5 15.0 10.5 7.1 8.8 5.4 8.3 8.4 13.0 3.5 9.7
1 15.9 12.0 16.4 16.0 16.9 16.0 15.0 13.2 13.7 7.5 6.1 5.2 12.1 15.4 15.6 16.3 15.8 16.0 16.2 16.3 16.7 16.3 14.1 14.6 7.1 13.9 15.6 16.5 16.6 16.2 15.6 16.2 15.8 16.2 15.6 16.4
2 2232 2361 2649 2451 1596 2591 1660 2646 2829 1061 778 390 2392 3102 2402 2611 1882 2205 2259 1047 1869 1779 2893 2920 986 2426 247 1778 2168 2025 2924 2538 2422 1029 3025 1780
3 588 87 727 276 122 316 21 89 174 53 15 4 30 313 260 290 242 390 170 92 329 271 304 403 12 61 2 223 773 505 217 20 35 10 195 458
4
Main four islands of Japan (3961 grid-squares)
343 82 361 335 297 346 258 151 197 24 14 9 81 320 322 346 350 329 331 251 322 315 229 268 20 207 263 315 342 331 307 334 324 230 342 323
5 250 23 295 137 62 164 12 28 50 15 2 2 13 69 89 142 114 152 100 57 153 137 76 127 5 24 1 119 286 223 73 7 13 4 91 182
6 0.729 0.280 0.817 0.409 0.209 0.474 0.047 0.185 0.254 0.625 0.143 0.222 0.160 0.216 0.276 0.410 0.326 0.462 0.302 0.227 0.475 0.435 0.332 0.474 0.250 0.116 0.004 0.378 0.836 0.674 0.238 0.021 0.040 0.017 0.266 0.563
7
selected 361 grid squares
<0.0001 0.5021 <0.0001 <0.0001 <0.0001 <0.0001 0.0122 <0.0001 <0.0001 1.000 1.000 1.000 1.000 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.731 0.717 1.000 <0.0001 <0.0001 <0.0001 <0.0001 0.2829 0.333 1.000 <0.0001 <0.0001
8 0.715 . 0.730 0.639 0.780 0.730 0.934 0.835 0.910 . . . . 0.804 0.701 0.679 0.730 0.780 0.771 0.698 0.697 0.679 0.840 0.791 . . . 0.649 0.860 0.679 0.691 . . . 0.674 0.790
9 0.481 . 0.492 0.504 0.528 0.542 0.652 0.380 0.438 . . . . 0.478 0.581 0.504 0.504 0.535 0.542 0.479 0.448 0.452 0.341 0.373 . . . 0.427 0.471 0.594 0.526 . . . 0.545 0.493
10 61 . 46 118 174 135 219 88 122 . . . . 172 182 125 144 135 171 121 86 88 100 79 . . . 83 41 54 151 . . . 167 99
11 32 . 20 80 61 47 27 35 25 . . . . 79 51 79 92 42 60 82 83 90 53 62 . . . 113 15 54 83 . . . 84 42
12 79 . 99 53 20 65 1 2 7 . . . . 14 45 47 22 44 34 15 28 31 8 16 . . . 21 42 58 27 . . . 32 44
13 171 . 196 84 42 99 11 26 43 . . . . 55 44 95 92 108 66 42 125 106 68 111 . . . 98 244 165 46 . . . 59 138
14
Performance of lositic models
0.676 . 0.670 0.603 0.727 0.676 0.891 0.755 0.838 . . . . 0.709 0.702 0.636 0.674 0.739 0.716 0.649 0.655 0.616 0.734 0.709 . . . 0.575 0.833 0.662 0.642 . . . 0.661 0.734
15 0.6840 . 0.6644 0.6131 0.6774 0.6037 0.9167 0.9286 0.8600 . . . . 0.7971 0.4944 0.6690 0.8070 0.7105 0.6600 0.7368 0.8170 0.7737 0.8947 0.8740 . . . 0.8235 0.8531 0.7399 0.6301 . . . 0.6484 0.7582
16
0.6559 . 0.6970 0.5960 0.7404 0.7418 0.8902 0.7154 0.8299 . . . . 0.6853 0.7811 0.6127 0.6102 0.7627 0.7403 0.5961 0.5089 0.4944 0.6536 0.5603 . . . 0.4235 0.7321 0.5000 0.6453 . . . 0.6653 0.7021
17
CyanMagentaYellowBlack Odonata page 203
Dragonfly distributional predictive models in Japan: relevance of land cover ...
203
species
Asiagomphus pryeri Davidius nanus Davidius fujiama Davidius moiwanus moiwanus Lanthus fujiacus Trigomphus citimus tabei Trigomphus melampus Trigomphus interruptus Trigomphus ogumai Stylogomphus suzukii Stylogomphus ryukyuanus ryukyuanus 74 Sinogomphus flavolimbatus 75 Nihonogomphus viridis 76 Onychogomphus viridicostus 77 Sieboldius albardae 78 Ictinogomphus pertinax 79 Sinictinogomphus clavatus 80 Cordulegastridae Chlorogomphus brunneus brunneus 81 Anotogaster sieboldii 82 Corduliidae Epophthalmia elegans elegans 83 Macromia daimoji 84 Macromia amphigena amphigena 85 Epitheca marginata 86 Epitheca bimaculata sibirica 87 Cordulia aenea amurensis 88 Somatochlora arctica 89 Somatochlora alpestris 90 Somatochlora japonica 91 Somatochlora graeseri 92 Somatochlora uchidai 93 Somatochlora viridiaenea 94 Somatochlora clavata 95 Libellulidae Lyriothemis pachygastra Lyriothemis elegantissima 96 97 Libellula quadrimaculata asahinai
63 64 65 66 67 68 69 70 71 72 73
No family
Appendix table. Continued.
16.3 15.7 15.6 13.3 14.8 15.8 14.8 15.9 15.9 15.8 17.8 15.7 15.7 15.7 16.0 17.1 16.1 17.3 16.2 16.2 15.7 16.2 16.3 12.2 11.7 7.1 0.7 7.4 9.4 15.6 15.9 16.4 16.6 17.5 17.0
8.7 9.1 9.1 2.6 12.6 8.5 12.1 2.6 5.8 10.1 2.6 9.9 2.6 2.1 1.9 -1.1 1.6 0.7 4.1 3.6 4.9 8.5 14.6 2.6
2
12.0 7.6 5.7 2.6 5.5 11.5 2.6 12.1 10.7 9.8 14.7
1
2196 2026 1997 2610 1208 2188 1489 2483 2456 1844 2576 1871 2423 2216 974 10 1039 1592 3013 3466 3375 2145 578 3314
1346 2212 2851 2727 2524 1392 2951 1188 1685 1767 543
3
145 222 268 554 102 300 32 938 478 58 512 227 40 65 21 3 40 34 460 223 97 443 6 413
116 435 206 98 284 142 264 138 122 276 9
4
Main four islands of Japan (3961 grid-squares)
320 318 318 357 257 332 276 361 349 300 361 321 92 71 20 0 22 35 344 354 354 333 102 361
283 327 333 158 276 285 284 271 306 311 92
5
48 86 108 199 49 155 10 299 226 23 194 113 19 21 6 0 9 11 168 39 45 189 2 182
65 126 57 19 68 67 90 72 65 101 2
6
0.150 0.270 0.340 0.557 0.191 0.467 0.036 0.828 0.648 0.077 0.537 0.352 0.207 0.296 0.300 0.000 0.409 0.314 0.488 0.110 0.127 0.568 0.020 0.504
0.230 0.385 0.171 0.120 0.246 0.235 0.317 0.266 0.212 0.325 0.022
7
selected 361 grid squares
<0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 1.000 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.0469 . 0.0716 1.000 <0.0001 <0.0001 <0.0001 <0.0001 0.424 <0.0001
<0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 1.000
8
0.835 0.788 0.789 0.786 0.791 0.711 . 0.738 0.693 0.846 0.787 0.757 0.855 0.803 0.815 . . . 0.686 0.774 0.687 0.631 . 0.683
0.707 0.848 0.809 0.848 0.743 0.794 0.769 0.743 0.808 0.766 .
9
0.462 0.559 0.523 0.566 0.414 0.437 . 0.511 0.468 0.410 0.531 0.512 0.552 0.349 0.351 . . . 0.506 0.534 0.529 0.524 . 0.493
0.499 0.430 0.469 0.536 0.436 0.438 0.521 0.514 0.641 0.469 .
10
184 166 140 119 131 91 . 43 64 182 117 145 64 26 12 . . . 108 226 206 86 . 105
126 141 184 114 116 142 140 127 205 132 .
11
88 66 70 39 77 86 . 19 59 95 50 63 9 24 2 . . . 68 89 103 58 . 74
92 60 92 25 96 76 59 72 36 78 .
12
6 24 27 58 7 24 . 79 49 2 50 32 5 2 2 . . . 45 11 16 73 . 60
18 13 8 5 9 8 18 17 23 15 .
13
42 62 81 141 42 131 . 220 177 21 144 81 14 19 4 . . . 123 28 29 116 . 122
47 113 49 14 59 59 54 55 42 86 .
14
Performance of lositic models
0.706 0.717 0.695 0.728 0.673 0.669 . 0.729 0.691 0.677 0.723 0.704 0.848 0.634 0.800 . . . 0.672 0.718 0.664 0.607 . 0.629
0.611 0.777 0.700 0.810 0.634 0.705 0.683 0.672 0.807 0.701 .
15
0.8750 0.7209 0.7500 0.7085 0.8571 0.8452 . 0.7358 0.7832 0.9130 0.7423 0.7168 0.7368 0.9048 0.6667 . . . 0.7321 0.7179 0.6444 0.6138 . 0.6703
0.7231 0.8968 0.8596 0.7368 0.8676 0.8806 0.7500 0.7639 0.6462 0.8515 .
16
0.6765 0.7155 0.6667 0.7532 0.6298 0.5141 . 0.6935 0.5203 0.6570 0.7006 0.6971 0.8767 0.5200 0.8571 . . . 0.6136 0.7175 0.6667 0.5972 . 0.5866
0.5780 0.7015 0.6667 0.8201 0.5472 0.6514 0.7035 0.6382 0.8506 0.6286 .
17
CyanMagentaYellowBlack Odonata page 204
204 Yoshitaka Tsubaki & Nobuyuki Tsuji
species
Libellula angelina Orthetrum sabina sabina Orthetrum poecilops miyajimaensis Orthetrum albistylum speciosum Orthetrum japonicum japonicum Orthetrum triangulare melania Nannophya pygmaea Deielia phaon Crocothemis servilia mariannae Sympetrum pedemontanum elatum Sympetrum darwinianum Sympetrum frequens Sympetrum striolatum imitoides Sympetrum eroticum eroticum Sympetrum kunckeli Sympetrum parvulum Sympetrum flaveolum flaveolum Sympetrum danae Sympetrum risi Sympetrum infuscatum Sympetrum baccha matutinum Sympetrum gracile Sympetrum maculatum Sympetrum speciosum speciosum Sympetrum croceolum Sympetrum uniforme Leucorrhinia dubia orientalis Leucorrhinia intermedia ijimai Pseudothemis zonata Rhyothemis fuliginosa Tramea virginia
No family
98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128
Appendix table. Continued.
12.6 11.0 14.3 5.2 5.9 6.6 6.4 8.9 8.9 2.6 7.9 2.6 4.8 2.6 7.0 6.7 2.9 0.6 5.6 2.6 5.3 12.2 9.1 7.2 4.7 9.5 0.3 3.5 9.4 10.0 10.4
1 16.8 17.8 14.3 17.8 16.3 16.4 16.3 16.4 16.5 15.6 16.3 15.9 16.6 16.4 16.5 16.0 6.0 9.0 16.4 16.0 16.5 15.9 15.5 16.4 16.0 16.0 7.5 5.9 16.3 16.5 17.4
2 1373 1802 1 3409 2509 2340 2876 2116 1889 3100 2018 2588 3375 2565 2598 2584 674 1440 2822 2646 3002 1226 2158 2585 3195 2147 1059 588 1763 1770 1935
3 120 14 1 1135 641 724 204 373 604 448 730 1059 133 829 315 354 12 49 433 875 373 82 23 320 218 48 57 9 588 356 85
4
Main four islands of Japan (3961 grid-squares)
279 305 361 353 350 346 347 332 332 343 337 359 354 361 344 341 13 31 346 357 352 269 302 341 352 321 23 13 327 319 315
5 48 2 0 340 240 270 75 182 254 142 271 316 48 282 148 158 2 10 200 281 148 47 12 119 108 29 10 4 260 170 34
6 0.172 0.007 0.000 0.963 0.686 0.780 0.216 0.548 0.765 0.414 0.804 0.880 0.136 0.781 0.430 0.463 0.154 0.323 0.578 0.787 0.420 0.175 0.040 0.349 0.307 0.090 0.435 0.308 0.795 0.533 0.108
7
selected 361 grid squares 8 <0.0001 1.000 . <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 1.000 1.000 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.003 1.000 <0.0001 <0.0001 <0.0001
0.777 . . 0.859 0.740 0.703 0.667 0.812 0.722 0.700 0.659 0.746 0.748 0.721 0.718 0.667 . . 0.704 0.791 0.642 0.731 0.831 0.724 0.692 0.764 0.908 . 0.807 0.778 0.835
9 0.452 . . 0.286 0.549 0.477 0.445 0.498 0.434 0.456 0.445 0.422 0.439 0.501 0.506 0.479 . . 0.552 0.446 0.519 0.493 0.369 0.519 0.475 0.531 0.660 . 0.604 0.543 0.569
10 153 . . 9 82 49 121 111 45 110 34 26 197 48 139 116 . . 110 52 126 137 178 147 140 206 12 . 56 106 227
11 78 . . 4 28 30 151 39 33 91 32 17 109 31 57 67 . . 36 24 78 85 112 75 104 86 1 . 11 43 54
12 9 . . 42 85 67 14 43 51 32 70 74 11 80 54 63 . . 82 67 59 12 1 37 25 10 2 . 99 52 9
13 39 . . 298 155 203 61 139 203 110 201 242 37 202 94 95 . . 118 214 89 35 11 82 83 19 8 . 161 118 25
14
Performance of lositic models
0.688 . . 0.870 0.677 0.728 0.524 0.753 0.747 0.641 0.697 0.747 0.661 0.693 0.677 0.619 . . 0.659 0.745 0.611 0.639 0.626 0.672 0.634 0.701 0.870 . 0.664 0.702 0.800
15 0.8125 . . 0.8765 0.6458 0.7519 0.8133 0.7637 0.7992 0.7746 0.7417 0.7658 0.7708 0.7163 0.6351 0.6013 . . 0.5900 0.7616 0.6014 0.7447 0.9167 0.6891 0.7685 0.6552 0.8000 . 0.6192 0.6941 0.7353
16
0.6623 . . 0.6923 0.7455 0.6203 0.4449 0.7400 0.5769 0.5473 0.5152 0.6047 0.6438 0.6076 0.7092 0.6339 . . 0.7534 0.6842 0.6176 0.6171 0.6138 0.6622 0.5738 0.7055 0.9231 . 0.8358 0.7114 0.8078
17
CyanMagentaYellowBlack Odonata page 205
Dragonfly distributional predictive models in Japan: relevance of land cover ...
205
CyanMagentaYellowBlack Odonata page 206
This page intentionally left blank
CyanMagentaYellowBlack Odonata page 207
Threat levels to odonate assemblages from invasive alien tree canopies
CONSERVATION
AND
BEHAVIORAL ISSUES
207
CyanMagentaYellowBlack Odonata page 208
208 Michael J. Samways
This page intentionally left blank
CyanMagentaYellowBlack Odonata page 209
Threat levels odonate Rivera assemblages from invasive alien tree canopies AdolfotoCordero (ed) 2006 Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 209-224.
209
© Pensoft Publishers
Sofia–Moscow
Threat levels to odonate assemblages from invasive alien tree canopies Michael J. Samways Department of Conservation Ecology and Entomology, University of Stellenbosch, P/Bag X1, Matieland 7602, South Africa. samways@sun ac za
ABSTRACT Dragonflies are well-known to be sensitive to light conditions, with the various species having a range of light conditions that they prefer. When these conditions are changed, such as by human removal of the tree canopy, the odonate assemblage changes accordingly, with forest species being replaced by species preferring sunlit habitats. Most of the South African species, including the national and local endemics, are mostly species that inhabit sunlit habitats, especially those fringed with indigenous grasses and bushes. During the 20th century, many of the South African riparian corridors became invaded and radically transformed by alien trees, especially Acacia spp. As these trees are a threat to hydrological processes, a massive national ‘Working for Water Programme’ was started to clear riparian zones of these alien trees. These trees were also posing a major threat to local biodiversity, especially endemic odonates. Some odonate species were even on the verge of extinction as a result of shading of their habitats by the alien trees as well as from various synergistic impacts such as over-abstraction of water and damage to the banks by domestic livestock. The recovery of some of these odonate species as a direct result of alien tree removal has been absolutely remarkable, and is a strong message in support of genuinely effective and positive conservation action involving removal of alien trees. Key words: Threats from invasive alien trees, conservation, Odonata
CyanMagentaYellowBlack Odonata page 210
210 Michael J. Samways
INTRODUCTION Dragonflies are sensitive to habitat quality (Corbet 1999). Furthermore, species abundance and assemblage composition at any one locality can change over time (Moore 1991, 2001). This change is due to migration between sites (Conrad et al. 1999), natality and mortality and variation in vegetation structure and composition, along with associated hydrological changes (Suh & Samways 2005). There may also be a strong response to major stochastic events such as floods (Samways 1989a). Assemblage composition can also change as a result of anthropogenic impact, which can be severe and rapid (Samways 2006b, Samways & Steytler, 1996). Among these impacts are invasive alien organisms, which are becoming recognized as the second-largest threat, after habitat loss, to biodiversity (Walker & Steffen 1999; Wilcove et al. 1998) and a major cause of extinctions globally (Clavero & García-Berthou 2005). However, evidence supporting a general and primary role for invasive aliens in extinctions remains limited (Gurevitch & Padilla 2004). Nevertheless, invasive alien trees (IATs) can dominate riparian systems and change hydrology (Le Maitre et al. 1996, Görgens & van Wilgen 2004). Although little is known of the impact of IATs on general stream ecology, IATs are clearly the most major threat to globally Red Listed dragonflies in South Africa (Samways & Taylor 2004). Interestingly, the threat posed to nationally Red Listed species (excluding the globally Red Listed species) is not from IATs but mostly from natural weather factors such as floods and drought (Samways 2006) associated with biogeographical marginality (Samways 2003a). Two of the world’s 25 biodiversity hotspots are South Africa (Myers et al. 2000). IATs are supposedly a major threat to this biodiversity, although little quantitative evidence is actually available (Richardson & van Wilgen 2004). Knowing that Odonata are sensitive to habitat change suggested the need for research into the impact of IATs on their assemblage composition and population levels. This research became even more imperative as a massive South African nationwide rehabilitation programme (Working for Water Programme) had begun. The aim of this programme is to remove IATs, principally for improvement of stream flow and social upliftment. In turn, this programme has provided an ideal opportunity for large-scale experimental studies where odonate assemblages in natural vegetation can be compared with those in infested and cleared areas.
CyanMagentaYellowBlack Odonata page 211
Threat levels to odonate assemblages from invasive alien tree canopies
211
RESPONSE OF SOUTH AFRICAN ODONATA TO CANOPY COVER In a savanna river in the Kruger National Park, certain habitat variables predicted with 82% success the odonate species present. These variables were flow rate, permanency of water body, reed cover, shade cover, and aquatic macrophyte cover (Clark & Samways 1996). Flow rate is a well-known variable for Odanata assemblage composition, with the broad categories of ‘lotic’ and ‘lentic’. However, in the South African context, there needs to be some recognition of spatial scale. Ischnura senegalensis (Rambur), for example, at the small scale of still water but at the larger scale of the landscape, can occur in a variety of water bodies from rivers to marshes. A more apt description would be that it is an inhabitant of ‘deposition zones’. Permanency of water body is also a well-known determinant of odonate species composition, with pools that maintain a constant water level being richer than those with varying levels (Osborn & Samways 1996). In turn, reed cover as from Typha spp., can be impoverishing, with few species being able to survive within a dense patch of these plants. Again, it is the widespread, vagile species that are present in this microhabitat e.g. Trithemis arteriosa (Burmeister). However, spatial scale is again important. Often, adjacent to a stand of Typha, are patches of sedges with aquatic macrophytes which can be home to more localized species, such as Agriocnemis falcifera Pinhey (Plate 1) and Africallagma elongatum (Martin) (Samways et. al. 1996). Taking ‘reed cover’ and ‘permanence’ together, it is not surprising that artificial pools, such as farm reservoirs, are rapidly colonized by odonates. However, these species are generally common and opportunistic, with endemic species usually preferring only permanent, natural water bodies. While these artificial water bodies generally increase the area of occupancy, particularly of common generalists, they rarely increase the extent of occurrence, especially of endemic specialists (Samways 1989b). In the instances where endemic species have colonized artificial water bodies, the reservoir is permanent, with a constant water level, shallow and wellvegetated with macrophytes, as well as having an abundance of marginal vegetation such as sedges (low canopy) and indigenous trees (high canopy) (Samways et al. 1996). Indeed, high canopy can be critical for some species, such as Pseudagrion hageni Karsch and Notiothemis jonesi Ris, which perch in shady habitats. Other species, such as Lestes virgatus (Burmeister), L. uncifer Karsch and Aciagrion cf zambiense Pinhey (Plate 2) require stands of trees in which to mature or to move back and forth with changing age and weather conditions. Aquatic macrophytes present an interesting situation for the larvae of many species that require perches and even hiding places from predators.
CyanMagentaYellowBlack Odonata page 212
212 Michael J. Samways
Some invasive alien macrophytes, although having a major adverse impact on some indigenous ecosystems, can, in the African savanna, actually benefit some already widespread and common species (Stewart & Samways 1998; Niba & Samways 2006). This anthropogenic disturbance, like artificial water bodies, tends only to benefit the already common species, by increasing their area of occupancy. The importance of shade cover is a complex issue because it has two facets: structural (level of shade i.e. percentage cover) and compositional (plant species involved in creating the shade). As odonates are predators, it would seem that perhaps both these features are unimportant, and rather, the overriding determinant is the density of prey irrespective of local landscape features, as is the case with ladybirds (Magagula & Samways 2001). However, in any one locality there are distinct sun-loving species (e.g. Trithemis furva Karch, T. arteriosa, T. stictica (Burmeister), Nesciothemis farinosa (Förster) and I. senegalensis) and shade-preferring species (e.g. Chlorolestes tessellatus (Plate 3) (Burmeister) and P. hageni) (Steytler & Samways 1995). Indeed, C. tessellatus (Samways & Steytler 1996) and Allocnemis leucosticta Selys (Plate 4), both South African endemics, are excellent indicators of undisturbed, shady conditions, so long as there are sunflecks.
TREE CANOPIES AND SOUTH AFRICAN SPECIES South Africa has few extensive forested areas, and few odonate species that are truly forest species. There are some tropical forest species that are marginal in South Africa, such as Parazyxomma flavicans (Martin), Zyxomma atlanticum Selys and N. jonesi. Those forest species that are nationally extensive in distribution tend to subspecific status, or distinct morphological forms, or are endemic. Those that are endemic or near-endemic include some highly local species such Chlorolestes umbratus Selys (Plate 5), C. elegans Pinhey, Ecchlorolestes nylephtha Barnard, as well as A. leucosticta. C. tessellatus additionally, has variation in wing colour, with a banded wing form in the Eastern Cape but usually clearwinged in KwaZulu-Natal (Samways 2006b). Forest insularization has thus led to population distinctiveness and separate evolutionarily significant units. The dependence on shade can be highly localized, with C. umbratus even seeking single trees in an open-stream habitat. Similarly, Ecchlorolestes peringueyi (Ris), although perching principally on large, linchen-covered boulders, appears to prefer a stream bordered with a thin canopy of indigenous trees (Plate 6). The remaining species are largely sun-loving, with some, such as the opportunist Crocothemis erythraea (Brullé), even benefiting from removal of indigenous trees (Samways & Steytler 1996). The ‘ideal indigenous state’ for maximal South African odonate species richness is given in Figure 1A.
CyanMagentaYellowBlack Odonata page 213
Threat levels to odonate assemblages from invasive alien tree canopies
213
A – Enriching effects (the indigenous state) Indigenous tree canopy IMPACT OF INVASIVE ALIEN Sun Sunfleck
Sunlit bank2
Bush1 Sedges
TREES
Boulder2
Lily pads
Shallow aquatic macrophytes
Tall grass
Permanent water body
1. Maturation zone and oviposition site 2. Basking and emergence site
B – Impoverishing effects (the invasive alien state) Invasive alien tree canopy
Shade
Shading of bank and water and loss of bushes
Synergistic impact4 Loss of shallow aquatic macrophytes and lily pads
Loss of tall grass
Fluctuating water level3
Invasive alien fish
Loss of basking and emergence sites
3. From altered hydrology 4. e.g. Cattle trampling at water’s edge
Fig. 1. Two opposite scenarios on South African odonate diversity. A. In the ideal enriching indigenous state there is much sunlight penetrating indigenous vegetation and there is a hydrological regime that is not impacted by invasive alien trees. Most odonate species benefit from this sunlit environment, even those of indigenous forests that require sunflecks. There is an abundance of oviposition sites, larval microhabitats, perching and basking sites, and bushes and long grass in which to retreat at night or in which to mature. B. In the impoverishing invasive alien state, sunlight is excluded, and the hydrological regime is disturbed by the water demands of the alien trees. The loss of sunlit sites deprives the indigenous odonates of oviposition sites and larval microhabitats through loss of understorey vegetation and aquatic macrophytes as well as loss of perching and basking sites. The impact of invasive alien trees tends to be adversely synergistic with other anthropogenic impacts such as cattle trampling of banks. Invasive alien fish can add extra pressure to the larval population.
CyanMagentaYellowBlack Odonata page 214
214 Michael J. Samways
Shade from trees not only reduces light but can also reduce temperature, which, in turn, and among other factors, affects egg and larval development (Corbet 1999) and adult behaviour (McGeoch & Samways 1991). Thus, sunshine probably acts indirectly as well as directly way, signaling other important conditions that are not necessarily recognizable to the adults at the time (Wingfield Gibbons & Pain 1992). Invasive alien trees can dominate the riparian zone, virtually to the exclusion of indigenous vegetation. These alien trees can exclude all but the most shade-loving odonate species. But even these species seem to prefer some indigenous bushes present under the alien tree canopy. In other words, this alien canopy must not be too dense. When it is very dense, and indigenous bushes are excluded, perching sites are lost. In the case of synlestids, oviposition sites are also lost. This indirect affect is additional to the direct affect on the adult behaviour (Stewart & Samways 1998). Low-growing alien plants such as Solanum mauritianum and Rubus cuneifolius do not affect the odonate assemblages in the same way as dense, high-canopy IATs such as Pinus spp. and Acacia mearnsii (Kinvig & Samways 2000). However, indigenous trees can also have the same effect when dense. This leads to the conclusion that it is the intensity of shade from the upper canopy that reduces odonate species richness and abundance. Sunflecks among this shade canopy, whether in natural forest or under alien trees, provides encouragement for establishment (Kinvig & Samways 2000). In the case of the South African species even distinctly dense-forest species such as E. nylephtha, need some sunshine. In short, it is principally vegetation structure rather than vegetation composition that drives the composition of the odonate assemblage in South Africa. The effect of shade on odonate presence extends beyond the drip zone of the trees and can extend into the shadow-zone, some 30 m beyond the tree. This, combined with a general inhibitory effect of alien shade trees means that even where there is a suitable sunny area along a IAT-invaded riparian corridor, the local odonates nevertheless are excluded or do not find this isolated, local, salubrious spot. In other words, landscape context can be important. The exclusion effect of dense IATs in South African mirrors the inhibitory effect of bridges on the European zygpoteran Calopteryx splendens (Harris) (Schutte et al. 1997). The ‘ameliorating impoverishing state’ leading to loss of South African odonate diversity is given in Figure 1B.
COSERVATION IMPLICATIONS IAT canopies can be an extinction risk for some species. Chlorolestes apricans Wilmot (Plate 7), which is Endangered, is threatened principally from the impact of Acacia mearnsii which can locally exclude virtually all local biota
CyanMagentaYellowBlack Odonata page 215
Threat levels to odonate assemblages from invasive alien tree canopies
215
(Samways 2002). Indeed, evidence to date suggests that no alien trees should be planted within 30 m of the edge of the water (Kinvig & Samways 2000). However, the problem is not simply about tree canopies. The impact of shade is often synergistic with other effects. C. apricans, for example, is suffering from multiple stressors. As well as A. mearnsii shading the banks, clumps of these trees encourage the gathering of shade-seeking cattle next to the water. The result is an eroded bank under the shade and in general along the riparian zone. Other impacts include that of alien trout (Onchorhychus mykiss) and the effect of detergent from hand washing of clothes, which is a major disturbance for endemic Comoro odonates (Samways 2003b). Often several endemic species occur in the same area, and in the case of C. apricans and Metacnemis valida Hagen in Sélys (Plate 8) (which is also Endangered) both occur sympatrically and largely suffer the same impacts. These threats do not mean that South African odonate conservation aims only at open habitats. As many national species are also forest species, the aim is to have spatial heterogeneity with various successional stages of vegetation. The variety of plant structures provides a wide range of microhabitats. Even the most localized of endemics have remarkable search behaviour, readily colonizing new ponds (Osborn & Samways 1996; Samways et al. 1996; Steyler & Samways 1996; Suh & Samways 2001). It is likely that not all individuals move across the land mosaic to the same extent, as shown for British odonates (Conrad et al. 1999). Of the threatened habitats, the most impacted over a wide scale in South Africa are pools in river braids (‘Kuile’) (Samways & Taylor 2004). IATs are particularly invasive of this habitat. Interestingly, the flexibility of certain odonate populations to lifting of the threat is remarkable (Samways et al. 2005). Three species in particular ((Pseudagrion newtoni Pinhey (Plate 9), Metacnemis angusta Selys (Plate 10), Proischnura polychromaticum (Barnard) (Plate 11)), which were feared extinct and having not been recorded for several decades, appeared soon after IATs were removed. Today they are only known from these sites. Of course, source populations must have existed, despite searches, but the overall population level must be very small to have escaped detection. The aim now is to keep the riparian zones that have been cleared of IATs free from IAT regrowth and germination. This is being achieved through close co-operation with the Working for Water Programme as well as with the Provincial conservation bodies. An important corollary in dragonfly conservation is to increase awareness in the eyes of the public, particularly as dragonflies ‘stand in’ for many other components of biodiversity (Moore 1997). In South Africa, this is being done by developing ‘dragonfly awareness trails’ which are designed not so much for sitespecific conservation per se, but to raise the profile of dragonflies among the public in general, resulting in indirect conservation (Suh & Samways 2001, 2005).
CyanMagentaYellowBlack Odonata page 216
216 Michael J. Samways
Some of the South African forest species (e.g. P. flavicans, Z. atlanticum and Gynacantha usambarica Sjöstedt) of the north-east of the country are tropical species whose geographical ranges extend to just across the national border. These forests are being increasingly disturbed as human pressure increases along the eastern seaboard (Samways 1999). Similarly, in East Africa, forest species are under increasing threat from forest loss (Clausnitzer 2004), a situation mirrored, for example, in Sri Lanka (Bedjaniˇc 2004), Hong Kong (Wilson 2004), Thailand (Hämaläinen 2004) and other parts of Southeast Asia (Orr 2004) and South America (Paulson 2004). The impact of IATs and plantation trees on odonate populations outside South Africa has been little explored, although pine plantations in Wales have an adverse impact on populations of Cordulegaster boltoni (Donovan) (Ormerod et al. 1990). On the Indian Ocean island of Mayotte (Samways 2003b) and Seychelles (Samways 2003c) there is a tendency towards two sets of odonate assemblages: the vagile, lentic species on the open flatlands, and range-restricted, often endemic, lotic species on the forested slopes. The endemic Seychelles fauna, in particular, is remarkably tolerant of the IAT canopy invasion, probably because these species are highly tolerant of shade and because there is still a high proportion of indigenous trees. The point is that this illustrates that the impact of IATs is a combination of behavioural responses of the local species in the context of the intensity of the impact. Some species, such as C. umbratus, are tolerant of the low levels of IATs. But once the IATs establish to become impenetrable stands, they then become a major local adverse impact to this species. As IATs cannot easily be partially removed (‘thinned’), the solution is an all-or-nothing management option, where biodiversity recovery and extinction reprieve means total removal of both the IAT stand and its seed bank (Samways et al. 2005). This means that follow up clearing of IATs is as critical as removal of the first generation. Thus, maintenance of the rare, endemic South African odonates is a long-term commitment. In all likelihood, many other components of endemic biodiversity are also likely to benefit. This long-term commitment need not be everywhere all at once but in strategic locations and for particular habitats (e.g. river braids). Red Listing would feed into this prioritization process, which would provide a highly targeted and effective conservation strategy for odonates primarily threatened by IATs.
ACKNOWLEDGEMENTS I thank Mrs. Marlene Isaacks for processing the manuscript, and the Working for Water Programme for financial support towards the original research. Special thanks to Professor Adolfo Cordero Rivera for the opportu-
CyanMagentaYellowBlack Odonata page 217
Threat levels to odonate assemblages from invasive alien tree canopies
217
nity to present this paper, along with financial support from the Universidade de Vigo, The Ministerio de Educadión y Ciencia, and the Xunta de Galicia.
REFERENCES BEDJANICˇ , M. 2004. Odonata fauna of Sri Lanka: research and threat status. International Journal of Odonatology 7: 279-294. CLARK, T.E. & M.J. SAMWAYS. 1996. Dragonflies (Odonata) as indicators of biotope quality in the Kruger National Park, South Africa. Journal of Applied Ecology 33: 1001-1012. CLAUSNITZER, V. 2004. Critical species of Odonata in Eastern Africa. International Journal of Odonatology 7: 189-206. CLAVERO, M. & E. GARCÍA-BERTHOU. 2005. Invasive species are a leading cause of animal extinctions. Trends in Ecology and Evolution 20: 110. CONRAD, K.F., K.H. WILLSON, I.F. HARVEY, C.J. THOMAS & T.N. SHERRATT. 1999. Dispersal characteristics of seven odonate species in an agricultural landscape. Ecography 22: 524-531. CORBET, P.S. 1999. Dragonflies: Behaviour and Ecology of Odonata. Harley, Colchester, UK. GÖRGENS, A.H.M. & B.W. VAN WILGEN. 2004. Invasive alien plants and water resources in South Africa: current understanding, predictive ability and research challenges. South African Journal of Science 100: 27-33. GUREVITCH, J. & D.K. PADILLA. 2004. Are invasive species a major cause of extinction? Trends in Ecology and Evolution 19: 470-474. HÄMÄLÄINEN, M. 2004 Critical species of Odonata in Thailand and Indochina. International Journal of Odonatology 7: 305-310. KINVIG, R. & M.J. SAMWAYS. 2000. Conserving dragonflies (Odonata) along streams running through commercial forestry. Odonatologica 29: 195-200. LE MAITRE, D.C., B.W. VAN WILGEN, R.A. CHAPMAN & D.H. MCKELLY. 1996. Invasive plants and water resources in the Western Cape Provinces, South Africa: modeling the consequences of lack of management. Journal of Applied Ecology 33: 161-172. MAGAGULA, C.N. & M.J. SAMWAYS. 2001. Maintenance of ladybeetle diversity across a heterogenous African agricultural/savanna land mosaic. Biodiversity and Coservation 10: 209-222. MCGEOCH, M.A. & M.J. SAMWAYS. 1991. Dragonflies (Odonata: Anisoptera) and the thermal landscape: implications for their conservation. Odonatologica 20: 303-320. MOORE, N.W. 1991. The development of dragonfly communities and the consequences of territorial behaviour: a 27 year study on small ponds of Woodwalton Fen, Cambridgeshire, UK. Odonatologica 20: 203-231. MOORE, N.W. (Compiler) 1997. Dragonflies: Action Plan. IUCN, Gland, Switzerland.
CyanMagentaYellowBlack Odonata page 218
218 Michael J. Samways
MOORE, N.W. 2001. Changes in the dragonfly communities at the twenty ponds at Woodwalton Fen, Cambridgeshire, United Kingdom, since the study of 19621988. Odonatologica 30: 289-298. MYERS, N., R.A. MITTERMEIER, C.G. MITTERMEIER, G.A.B. DA FONSECA & KENT. 2000. Biodiversity hotspots for conservation priorities. Nature 403: 853-858. NIBA, A.S. & M.J. SAMWAYS. 2006. Remarkable elevational tolerance in an African dragonfly (Odonata) larval assemblage. Odonatologica 35: 255-270. ORMEROD, S.J., N.S. WEATHERLEY & W.J. MERRETT. 1990. The influence of conifer plantations on the distribution of Cordulegaster boltoni (Odonata) in upland Wales. Biological Conservation 53: 241-251. ORR, A.G. 2004. Critical species of Odonata in Malaysia, Indonesia, Singapore and Brunei. International Journal of Odonatology 7: 371-384. OSBORN, R. & M.J. SAMWAYS. 1996. Determinants of adult dragonfly assemblage patterns at new ponds in South Africa. Odonatologica 25: 49-58. PAULSON, D.R. 2004. Critical species of Odonata in the Neotropics. International Journal of Odonatology 7: 163-188. RICHARDSON, D.M. & B.W. VAN WILGEN. 2004 Invasive alien plants in South Africa: how well do we understand the ecological impacts? South African Journal of Science 100: 45-52. SAMWAYS, M.J. 1989a. Insect conservation and the disturbance landscape. Agriculture, Ecosystems and Environment 27: 183-194. SAMWAYS, M.J. 1989b. Farm dams as nature reserves for dragonflies (Odonate) at various attitudes in the Natal Drakensberg mountains, South Africa. Biological Conservation 48: 181-187. SAMWAYS, M.J. 1999. Diversity and conservation status of South African dragonflies (Odonata). Odonatologica 28: 13-26. SAMWAYS, M.J. 2002. Red Listed Odonata species of Africa. Odonatologica 31: 117-128. SAMWAYS, M.J. 2003a. Marginality and national Red Listing of species. Biodiversity and Conservation 12: 2523-2525. SAMWAYS, M.J. 2003b. Threats to the tropical island dragonfly fauna (Odonata) of Mayotte, Comoro archipelago. Biodiversity and Conservation 12: 1785-1792. SAMWAYS, M.J. 2003c. Conservation of an endemic odonate fauna in the Seychelles archipelago. Odonatologica 32: 177-182. SAMWAYS, M.J. 2006a. Open and banded wings: hypotheses on damselfly wing position (Zygoptera: Odonata). Odonatologica 35: 67-73. SAMWAYS, M.J. 2006b. National Red List of South African dragonflies (Odonata). Odonatologica, in press. SAMWAYS, M.J., P.M. CALDWELL & R. OSBORN. 1996. Spatial patterns of dragonflies (Odonata) as indicators for design of a conservation pond. Odonatologica 25: 157-166. SAMWAYS, M.J. & STEYTLER. 1996. Dragonfly (Odoanta) distribution patterns in urban and forest landscapes and recommendations for riparian corridor management. Biological Conservation 78: 279-288. SAMWAYS, M.J. & S. TAYLOR. 2004. Impacts of invasive alien plants on Red-Listed South African dragonflies (Odonata). South African Journal of Science 100: 78-80.
CyanMagentaYellowBlack Odonata page 219
Threat levels to odonate assemblages from invasive alien tree canopies
219
SAMWAYS, M.J., S. TAYLOR & W. TARBOTON, W. 2005. Extinction reprieve following alien removal. Conservation Biology 19: 1329-1330. SCHUTTE, G., M. REICH & H. PLACHTER. 1997. Mobility of the rheobiont damselfly Calopteryx splendens (Harris) in fragmented habitats (Zygoptera: Calopterygidae). Odonatologica 26: 301-327. STEWART, D.A.B. & M.J. SAMWAYS. 1998. Conserving dragonfly (Odonata) assemblages relative to river dynamics in an African savanna game reserve. Conservation Biology 12: 683-692. STEYTLER, N.S. & M.J. SAMWAYS. 1995. Biotope selection by adult male dragonflies (Odonata) at an artificial lake created for insect conservation in South Africa. Biological Conservation 72: 381-386. SUH, A.N. & M.J. SAMWAYS. 2001. Development of a dragonfly awareness trail in an African Savanna game reserve. Conservation Biology 12: 683-692. SUH, A.N. & M.J. SAMWAYS. 2005. Significance of temporal changes when designing a reservoir for conservation of dragonfly diversity. Biodiversity and Conservation 14: 165-178. WALKER, B.M. & W.L. STEFFEN. 1999. Interactive and integrated effects of global change on terrestrial ecosystems. In: B. Walker, W.L. Steffen, J. Canadell & J. Ingram (eds), The Terrestrial Biosphere and Global change Implications for Natural and Managed Ecosystems, Cambridge University Press, Cambridge. WILCOVE, D.S., D. ROTHSTEIN, J. DUBOW, A. PHILLIPS & E. LOSOS. 1998. Quantifying threats to imperiled species in the United States. BioScience 48: 607-615. WILSON, K.D.P. 2004. Critical species of Odonata in China. International Journal of Odonatology 7: 409-422. WINGFIELD GIBBONS, D. & D. PAIN. 1992. The influence of river flow rate on the breeding behaviour of Calpopteryx damselflies. Journal of Animal Ecology 61: 283-289.
CyanMagentaYellowBlack Odonata page 220
220 Michael J. Samways
Plate 1. Agriocnemis falcifera, a South African endemic species that inhabits sedges at the edge of pools.
Plate 2. Aciagrion cf. zambiense recently recorded in South Africa, and which shuttles from marshy habitats to forested ones with changing weather.
CyanMagentaYellowBlack Odonata page 221
Threat levels to odonate assemblages from invasive alien tree canopies
221
Plate 3. Chlorolestes tessellatus, a South African endemic, indicative of high-quality natural forest conditions.
Plate 4. Allocnemis leucosticta, a South African endemic, characteristic of small streams in indigenous forest.
CyanMagentaYellowBlack Odonata page 222
222 Michael J. Samways
Plate 5. Chlorolestes umbratus, a South African endemic which, as its name suggests, seeks shade.
Plate 6. Ecchlorolestus peringueyi, a rare Western Cape endemic, that perches on linchen-covered boulders in small, upland streams.
CyanMagentaYellowBlack Odonata page 223
Threat levels to odonate assemblages from invasive alien tree canopies
223
Plate 7. Chlorolestes apricans, a South African endemic, highly threatened by invasive alien trees.
Plate 8. Metacnemis valida, another South African endemic, dicing with extinction from the impact of alien trees.
CyanMagentaYellowBlack Odonata page 224
224 Michael J. Samways
Plate 9. Pseudagrion newtoni, last seen in 1961, makes a limited comeback with the removal of invasive alien trees.
Plate 10. Metacnemis angusta, last seen in 1920, and previously a taxonomic mystery, was recently recorded where alien invasive trees were removed from river braids.
Plate 11. Proischnura polychromaticum, a minute Cape endemic, last seen in 1962 and 1936, was recently recorded among sedges in a river pool (‘Kuil’) where invasive alien trees had been removed.
CyanMagentaYellowBlack Odonata page 225
Movement behaviours of a forest odonate in two heterogeneous landscapes Adolfo Cordero Rivera (ed) 2006 Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 225-238.
225
© Pensoft Publishers
Sofia–Moscow
Movement behaviours of a forest odonate in two heterogeneous landscapes Philip D. Taylor Atlantic Cooperative Wildlife Ecology Research Network Department of Biology, Acadia University, Wolfville, NS. B4P 2N5
ABSTRACT The results from an empirical survey of C. maculata along streams in both a largely forested landscape and a more open, agricultural landscape are compared so simple measures of landscape structure, and the output from a behavioural simulation model based on a set of simple rules that govern how C. maculata accesses resources in the two landscapes. In the more open landscape, only proximity of the forest to the stream explains the empirical pattern of distribution, but in the more forested landscape, only simulated use of streams does. Further, populations are aggregated at broader spatial scales in the more open landscape. Collectively, the results suggest that C. maculata move more extensively when compared to the more closed, forested landscape, which has implications for landscapescale population structure.
INTRODUCTION Humans can influence the structure of forested landscapes over very large spatial scales. Activities such as farming permanently remove forest habitat and replace it with open pasture or crops. Forest harvest removes standing timber which is replaced by new forest that will have a younger age structure and may also have a radically different species composition. In both cases, the underlying structure can also vary through time. Most importantly, human-induced processes such as these typically occur at very broad spatial
CyanMagentaYellowBlack Odonata page 226
226 Philip D. Taylor
scales; scales that may encompass typical population-level processes such as dispersal and mating, for a variety of taxa. For many odonates, forest is an essential habitat – a home for species that typically are not or can not be found anywhere else (Corbet 1999). But for many odonates, forests are just one of several habitats where important components of the life-cycle can occur. For example, forests may serve as ‘resting’ habitat for tenerals, so they can mature away from areas where they may otherwise be harassed by adults or subjected to greater predation (Corbet 1999). For yet more species, forests provide foraging and daily roosting sites that are distinct from oviposition and mating sites (Corbet 1999). In these latter situations, daily or seasonal movements must occur so that individuals can access not only patches of forest habitat but other habitats where oviposition and mating take place. Population persistence is the key issue for management and conservation of many species. In heterogeneous landscapes, persistence of populations depends not only on the presence of a complete suite of resources patches in a particular area, but further, on the collective ability of individuals (a population) within that area to move and access all of those resources. For taxa that have life stages in both terrestrial and aquatic environments (such as odonates) improving our understanding of their ability to move through landscapes and ‘string together’ sets of resources in order to survive, is key to gaining an understanding of how their populations persist in human-modified landscapes. Studies of landscape connectivity (Taylor et al. 1993; Belisle 2005) aim to do just that. Considerable research has tackled the issue of how individual animals respond to patterns of altered forest structures arising from human activities. In odonates, such studies have typically focussed on behaviours at fine or medium spatial and temporal scales (e.g. Pither & Taylor 1999). Fine scales as those where behaviours occur within a single day and over moderate distances (up to 500 m for example for many odonate species). Activities such as territorial defence and daily foraging are included in this set. Medium scales are broader (500-2000 m), and encompass movement behaviours that typically only occur every day or second day, for example decisions regarding movement from oviposition to resting sites. Such movements are typically thought of as within population movements. Finally, landscape scales (> 2000 m), are scales over which population-level processes such as dispersal or migration occur. Such movements may occur only once or twice in an individual’s lifetime. Of interest in these fine- and medium-scale studies are questions about how processes (for example, the movement of individual damselflies between forested and non-forested habitat) differ depending on the distribution and amount of habitat at the landscape scale (e.g. Jonsen & Taylor 2000a) and how the behaviours translate into patterns of distribution at
CyanMagentaYellowBlack Odonata page 227
Movement behaviours of a forest odonate in two heterogeneous landscapes
227
landscape scales (e.g. see Levey 2005). These are questions that are difficult to tackle directly, because of the large spatial and long temporal scales involved. As a consequence, it is useful to explore alternate means by which we might gain insight into these broad-scale processes that might then suggest new avenues of inquiry. In this paper, I present an example of one such means to explore links between fine-scale behaviours and landscape-scale patterns. The system I use involves the damselfly Calopteryx maculata (Odonata, Calopterygidae). The species is typically found along forested streams, but can also be found in partially forested landscapes where it makes frequent flights traversing open areas to access nearby patches of forest (Taylor & Merriam 1995). It is known that fine-scale movement processes in this species (such as travel between forest habitats used for foraging, and oviposition sites) differ depending on the arrangement of those habitats (the distance between them) in the landscape. For example, Jonsen and Taylor (2000a) showed that C. maculata had higher rates of net displacement in landscapes that were partially or non-forested compared to those that were forested. Moreover, Jonsen & Taylor (2000b) suggested, using a simulation modelling approach, that the medium-scale patterns of dispersion in the species arose from finescale movement behaviours that were not only related to within-habitat characteristics, but also to characteristics of the broader landscape. They showed that, at least for C. maculata, that landscape ‘context’ influenced fine-scale behavioural decisions suggesting a kind of cross-scale interaction (Allen & Holling 2002) in behaviour. I hypothesize that such cross-scale interactions will produce patterns of distribution unique to particular landscape types, and could also influence population structure. Consider three types of landscapes, all containing a network of streams separated from one-another by distances of 1-2 km. All streams are inhabited by C. maculata but the landscapes differ in the amount of forest between the streams. In landscapes with large amounts of forest between the streams, individuals will venture off the streams frequently to forage in the forest, but will rarely move from one stream to another. In landscapes with little forest, individuals will rarely move off the streams, and so again, will rarely move from one stream to another. However, in landscapes where the space between streams is partially covered with forest, individuals will move over broader spatial extents (i.e. they will have expanded ecological neighbourhoods; Addicott et al. 1987) and therefore interact more frequently with more distant populations. In the two extreme cases (completely forested and largely non-forested landscapes) it might be hypothesized that the population structure in a forested landscape will be more like a true metapopulation (sensu Harrison 1994) whereas in landscapes with partial forest cover, there will be more of a ‘patchy population’ (sensu Harrison 1994).
CyanMagentaYellowBlack Odonata page 228
228 Philip D. Taylor
If these fundamental processes were indeed correct, then we could infer that landscape-scale patterns of distribution of odonates might relate to landscape-scale patterns of forest structure through the influence of that structure on fine-scale processes. Such ‘cross-scale effects’ are of interest because they may, because of thresholds and non-linearities, alter the broad-scale structure or even behaviour of systems in unexpected ways (Allen & Holling 2002). Such effects are of particular importance to landscape managers who may, for various reasons, continue to view systems as stable or linear, even when confronted with evidence that they are unstable and non-linear (Walters 1997). A clearer understanding of such complexity in empirical systems is invaluable. However, even if they were common, such effects may be difficult to identify simply because they will likely be manifested over broad-spatial or longer temporal scales. In spite of this, it may be possible to detect the signatures of such effects through combinations of statistical models of distributions at large scales, and simulation models that relate patterns to processes across scales. Here, using a data set on the distribution of C. maculata within two landscapes dominated by different types of forest landscape structure, I use a simple simulation model to explore whether and how fine-scale behaviours may be translated across scales in ways consistent with the hypothesis that landscape-scale patterns of distribution of a forest damselfly are a function of fine-scale habitat attributes, meso-scale habitat, and interactions between fine-scale behaviours and meso-scale habitats. I first sampled the distribution of C. maculata along most streams in two landscapes that differed in their overall pattern of forest cover and land use. I then derived two measures that could potentially explain the observed pattern of distribution of the species: 1) the distance between the stream and the edge of the nearest forest patch (of three different size classes) and 2) an index of the frequency of use by individuals of different streams within each landscape derived from a model that simulated, using a small set of very simple rules, the movement of individual damselflies within the landscape in relation to the same patches of forests and streams. I used statistical models to explore how the empirical pattern of stream occupancy related to both variables. Of particular interest was whether the measures of simulated stream use was better at explaining the empirical distribution of damselflies when compared to the simple measure of distance between stream and forest, and whether the importance of each measure differed between the two landscape types. If simulated numbers of individuals did predict incidence, it would suggest that fine-scale individual behaviour interacted with broader-scale landscape structure (sensu Taylor et al. 1993) and that in these situations empirical distributions could not be predicted simply from knowledge of the scale of a set of particular behaviours.
CyanMagentaYellowBlack Odonata page 229
Movement behaviours of a forest odonate in two heterogeneous landscapes
229
METHODS Calopteryx maculata primarily inhabit streams that flow through forest. Two resource patches are required by the species: the flowing streams (along which they mate and oviposit, and within which the larvae develop) and forested areas (within which adults roost and forage for prey) (Johnson 1962). The species is also found in fragmented-forested landscapes, where populations are present along streams that run through pastures. Individuals are capable of moving between forest and stream resource patches on a daily basis (Pither & Taylor 1999; Jonsen & Taylor 2000a). Regional distribution. The regional distribution of C. maculata was surveyed in 1991 in the area around Ottawa, Ontario, Canada. All road-stream intersections in two landscapes (one north and one south of the Ottawa River; Figure 1) were visited during the middle of the day (1000-1500) during the month of July (the period of day and month when the species is most active).
Fig. 1. Classified Landsat image of study area showing open areas (light gray), forest (medium gray) and water (dark grey). The image represents an area of approximately 100 km by 200 km near the City of Ottawa, Ontario, Canada. The major river in approximately the middle of the image is the Ottawa river which roughly divides the northern and southern landscapes.
CyanMagentaYellowBlack Odonata page 230
230 Philip D. Taylor
The two landscapes contrasted one another; the northern landscape was primarily forested, with small intermittent areas of agricultural land whereas the southern landscape was more open, with a more patchy distribution of forest. At each intersection, I recorded the number of male and female C. maculata present within 50 m of the intersection and whether the stream at that location was flowing. 85 stream/road intersections were surveyed in the northern landscape and 147 were surveyed in the southern landscape. A Landsat image (25 m pixel size) of the Ottawa-Hull region was classified into three classes: forest, large bodies of water and non-forest (Figure 1). The area of each patch of contiguous forest was calculated, and three maps were created, one map containing all forest patches greater than 100 m2, a second containing all patches greater than 1000 m2 and a third map containing all patches greater than 10,000 m2. The first map therefore included all patches on the second and third maps; and the second map included all patches on the third. Within these maps, I reclassified all forest below the threshold sizes as non-forest. For each of these three maps I then also created a further set of maps that showed, for each 25 × 25 m pixel, the distance and direction to the edge of the nearest forest patch. Similarly, all streams in the region were digitized from topographic maps (1:50000 NTS series) and I created a fourth ‘map’ giving the distance and direction of the nearest stream for each pixel. The complete set of maps thus allowed me to determine the presence of, distance and direction to three sizes of forest patch and streams for every 25 × 25 m pixel in the landscape. Simulation model of landscape connectivity. I simulated movement of many individual female C. maculata within the landscape at a fine scale (i.e. movement in non-forest vs. movement in forest) through a single generation (a 30 day period). The simulation model was made deliberately simple since I was not interested in recreating the precise dynamics of the species’ movement, but rather, in exploring how relatively simple measures of movement influenced distribution across scales. Within the model, each female lived for 30 days (approximately the maximum lifetime for adults). Reflecting the general biology of the species, animals spent the first 8 days of their life feeding, then subsequently alternated between feeding and oviposition states. While in a feeding state, animals moved towards nearby patches of forest; while in the oviposition state, animals moved toward streams. Individuals moved in single pixel steps (~25 m) up to a maximum that depended on their state. Individuals could detect streams or forest from distances of up to ~1000 m; when further than 1000 m from either, or in forest, individuals moved according to a correlated random walk (Turchin 1998). While in a feeding state, and not in forest, individuals could move a minimum of 1500 m plus an additional amount drawn from a lognormal distribution (median distance of 300 m) in an attempt to find
CyanMagentaYellowBlack Odonata page 231
Movement behaviours of a forest odonate in two heterogeneous landscapes
231
forest. While in an oviposition state, individuals were effectively allowed to move until they detected a stream and then, after reaching a stream, individuals could move up or down the stream in a random direction for a maximum of 500 m. If an individual could not find the appropriate habitat within the time/distance constraints noted, it persisted in the same state for a subsequent ‘day’. All individuals survived regardless of habitat. The specific parameters outlined above were estimated based on observations of the biology of the species in the area (Taylor 1993); their suitability within the model was assessed by graphically comparing the simulated distribution of damselflies in forest and non-forest sites with local-scale measures of density obtained from 750 m transects perpendicular to streams at two sites. The patterns of movement of C. maculata were then simulated by populating the simplified forest/stream landscapes discussed above, with 1 individual per 50 m of stream and aggregating all points visited by all damselflies across the landscape over 30 simulated days. I term this ‘simulated stream use’. The purpose of the simulation was to determine, when subjected to the specific behavioural rules outlined above, on which streams’ individuals spent most of their time. The model makes no attempt to include additional factors that might influence behaviour including: density dependence, competition with con-generics or other factors associated with streams that are known to be important to the species. As such, it is highly simplified, but also overly conservative, since, if incorporated into the model, such additional elements would most likely improve the overall probability of detecting an effect. Statistical models. All analyses were done using the R statistical package (v 2.1.0; R Development Core Team 2005). First, the spatial point pattern (random vs. clumped) was determined by calculating Ripley’s K across a series of distance classes using the Kfn function in the MASS package (Venables and Ripley 2002). The values were compared to randomized distributions (1000 replications) based on the underlying distribution of sample points. Logistic regression models were fit using the glm function with binomial errors and logit link. Three models were fit for each of the northern and southern landscapes, one for each of the three categories of forest patch size (100 m2; 1000 m2; 10,000 m2). Each model included a term indicating whether the stream was flowing at each empirical sample point, and pairs of the terms for proximity of forest of the three minimum size classes, and an index of the relative use of the stream at the sample point by simulated individuals (simulations run with forest patches at that scale). Fitting the term for simulated stream use after fitting the term for proximity of forest is a conservative test of the effect of simulated stream use. Such a conservative test is important, since the two terms will clearly be correlated. Overall goodness of fit of the models was assessed by examining residual plots and the dispersion parameter, and terms in models were considered of interest if their value was > 2 times their estimated standard error.
CyanMagentaYellowBlack Odonata page 232
232 Philip D. Taylor
RESULTS
1 .0
Female C. maculata were present at 32% (27/84) points in the forested northern landscape and at 12% (17/145) points in the more open southern landscape. Across both landscapes, individuals were present at ~50% of survey points where large patches of forest were closer than 750 m versus 18% of other points. At fine scales, the empirical distribution of female C. maculata in the forested northern landscape was clumped at distances of ~250 m compared to the open southern landscape, where it showed a clumped distribution at scales of ~500 m (Figure 2). Beyond a scale of ~750 m, the distribution of female C. maculata in the southern landscape was consistently and significantly clumped whereas in the northern landscape, the distribution was not significantly different from random.
0 .0
0 .2
0 .4
0 .6
0 .8
South of the Ottawa River North of the Ottawa River
0
500
1000
1500
2000
2500
3000
3500
Distance (m)
Fig. 2. The extent to which the spatial point pattern for occupied points in the southern and northern study areas are clumped, random or uniform. Shown is the deviation of the Ripley’s K value calculated from points occupied by Calopteryx maculata from the expected value calculated from all points surveyed (y axis) compared across distance classes. Points near the top of the plot (above the horizontal line) suggest a clumped distribution; points near the bottom suggest that points are uniformly distributed.
CyanMagentaYellowBlack Odonata page 233
Movement behaviours of a forest odonate in two heterogeneous landscapes
233
Stream flow was positively correlated with the presence of female C. maculata for all models at all scales (Table 1) reflecting the fundamental importance of that resource type to the species (flowing streams are used for oviposition). In both landscapes proximity of forest or simulated stream use Table 1. Parameter estimates (estimated standard error in brackets) from generalized linear models (binomial errors; logistic regression) fitting the presence/absence of female Calopteryx maculata to three variables: stream flow at the sampling site, distance to the nearest large (> 10,000 m2) forest patch and the density of animals from the simulation model. Only parameter estimates significantly different from zero (t-test) are shown. South
2.90 (1.10) ns 0.02 (0.01)
2.70 (1.00) -0.05 (0.02) ns
0 .4
0.8
0.8
0 .0
0 .4 0 .0
400
600
800
1000
200
400
600
800
1000
20
40
60
80
100
20
40
60
80
100
0.4
0 .4
0 .8
0 .8
200
0 .0
0 .0
Pre s e nc e o f fe ma le C. ma c ula ta
Pre s e nc e o f fe ma le C. ma c ula ta
Flow Distance to large forest patches Simulated density
North
Distance to large patches of forest (m)
0
Simulated density of female C. maculata
Fig. 3. Presence of female C. maculata in relation to Distances to large patches of forest, and to simulated stream use. Lines are locally-weighted regressions showing the probability of incidence as a function of the distance to forest patches (10,000 m2; left panels) and simulated stream use of individuals (right panels). The top two panels show results from the northern landscape; the bottom two show results from the southern landscape.
CyanMagentaYellowBlack Odonata page 234
234 Philip D. Taylor
100 80 60 40 20
Simulated stream use in northern landscape
of damselflies were only found to be important when considering large forest patches (>10,000 m2). In the open, southern landscape, incidence increased with increasing proximity of large forest patches but there was no additional effect of simulated stream use. In the more forested northern landscape, there was no effect of distance to forest, but incidence increased with the index of simulated stream use (Table 1; Figure 3). Quantile-quantile plots comparing the distribution of simulated stream use in the northern and southern landscapes show that, across the entire distribution, simulated stream use is higher in the northern landscape (points above the line in Figure 4). Since streams were populated with the same initial densities of individuals, this result demonstrates that, with the given set of behavioural rules and at the points sampled empirically, simulated damselflies spend more time along streams in the northern landscapes than in the southern landscape.
0
20
40
60
80
100
Simulated stream use in southern landscape Fig. 4. Quantile-Quantile plot comparing the distributions of numbers of C. maculata visiting simulated streams in the southern versus the northern landscape. The straight line is the expected pattern when the distributions are the same. Points lying above the line indicate a larger numbers of individuals in the northern landscape compared to the southern landscape.
CyanMagentaYellowBlack Odonata page 235
Movement behaviours of a forest odonate in two heterogeneous landscapes
235
DISCUSSION Collectively, the results are consistent with the scenario that in the more open, southern landscape, where resources are likely more patchily distributed, female C. maculata move over greater spatial extents to access resources compared to the more forested northern landscapes. Three lines of evidence support this proposition: 1) population incidence was correlated over broader spatial scales in the south; 2) proximity of large forest patches was only an important predictor of incidence in the south, and 3) quantile plots of simulated stream use showed that individuals spent less time at streams in the south than in the north. Importantly, the results also demonstrate that a measure of landscape use, based on a small, simple set of behavioural rules, adequately explains variation in the incidence of the species in the northern (forested), but not the southern (open) landscape. The result also suggests that the presence of the species along streams in the northern landscape results from a more complex function of resource use than simple availability or proximity of resources. The converse was true for the southern landscape; proximity of the resource was the only predictor, and there was no need for additional information of how behaviour interacted with landscape structure to result in a given pattern of distribution. If females interact with landscape structure at broader scales, different fine-scale behaviours would lead to altered patterns of distribution, and potentially, different kinds of spatial structuring of populations. One can envision that if landscapes with population structures that were similar to metapopulations (in the strict sense; Harrison 1994) gradually had resources removed from them (making them more highly fragmented) that those populations might begin to show increased flow of individuals between sub-populations and begin to behave more like patchy populations. The completely random spatial point pattern exhibited in the forested northern landscape contrasts sharply with the clumped pattern in the more open southern landscape. Such a pattern is consistent with a difference in overall population structure between the two landscapes. If animals must move over larger spatial extents to access resources, then population incidence will be spatially correlated over broader scales (Turchin 1998). Within the more forested northern landscape, most streams are situated within 400 m of a patch of large forest. It is known that the species is readily able to access resources over these scales (Pither & Taylor 1999), however, the positive association between the simulated numbers of animals using the stream and the presence of female C. maculata in the forested northern landscape
CyanMagentaYellowBlack Odonata page 236
236 Philip D. Taylor
implies that key elements of their behaviour are captured within the simple simulation rules (essentially, moving between forest and streams up to ~1000 m apart) within that landscape. This also implies that behaviour in the more open southern landscape is not captured by those same rules. Assuming that movement is likely occurring over broader spatial scales in the more open landscape, then it suggests that our fine-scale behavioural rules should include ‘context’ effects (a difference in behaviour that is a function of the overall landscape context; Jonsen & Taylor 2000b). Such an interaction is an example of a cross-scale effect (Allen & Holling 2002). Additional detailed explorations of the simulation model, coupled with additional empirical investigations could be used to more precisely tease apart the specific interactions between landscape structure and behaviour that gave rise to the different patterns observed in the two landscapes. In addition to possible cross-scale context effects of landscape on fine-scale movement and distribution, there are likely effects of the intervening matrix that may enhance or diminish the observed patterns (Ricketts 2001). For example, experiments on movement abilities of C. maculata have been done largely within landscapes dominated by forage crops or pasture, and less frequently in landscapes with open areas dominated by scrub or food-crops. In a different system involving heterogeneous boreal forest in western Newfoundland, McPherson (2003) showed that the Libellulid, Leucorrhinia hudsonica is less abundant in peatlands surrounded by forest than in peatlands surrounded by scrub or harvested forest. The implication (at least partially supported by direct measurements of movement; Chin 2006) is that movement is facilitated where the matrix between peatlands is more open, thus facilitating inter-peatland movement which has the effect of enhancing populations. Comparable results have been observed in other systems such as in the alpine butterfly system in Alberta, Canada (Roland et al. 2000). Assuming that matrix does generally influence movement, we might expect additional non-linearities within such systems to arise from the combined effects of human activity reducing overall resource abundance, changing its positioning within landscapes, and through changes to the non-resource matrix itself, that will influence movement processes. The results also suggest that studies of a species’ ‘habitat’ need to incorporate measures from multiple scales. Such multi-scale studies are now relatively commonplace for some taxa such as birds (e.g. Betts et al. 2006) but could be even more prevalent in odonate and other insect studies (e.g. see Cronin & Reeve 2005). Insight into the relative importance or even existence of underlying processes that lead to patterns of distributions will be enhanced by exploring the relationships between correlates of those processes and the patterns of distributions using simple statistical relationships.
CyanMagentaYellowBlack Odonata page 237
Movement behaviours of a forest odonate in two heterogeneous landscapes
237
Finally, populations of many species inhabiting human-dominated landscapes may be able to cope with moderate amounts of broad-scale habitat change, simply by altering how they link together resource patches through the process of movement. In forest landscapes, this suggests that managers must first pay attention to the amounts of different resource patches that remain on the landscape (that is, that all requisite resources must be present for a given species to survive, but then consider how such resource patches are tied together – the concept of landscape connectivity; Taylor, With and Fahrig in press). Managers need to be alert to the fact that altering broad-scale landscape structure may influence population structure in unexpected and possibly non-linear ways. REFERENCES ADDICOTT, J. F., J. M. AHO, M. F. ANTOLIN, D. K. PADILLA, J. S. RICHARDSON & D. A. SOLUK. 1987. Ecological neighborhoods: scaling environmental patterns. Oikos 49: 340-346. ALLEN, C. R. & C. S. HOLLING. 2002. Cross-scale structure and scale breaks in ecosystems and other complex systems. Ecosystems 5: 315-318. BETTS, MATTHEW G., G.J. FORBES, A.W. DIAMOND & P.D. TAYLOR. 2006 Independent effects of habitat amount and fragmentation on songbirds in a forest mosaic: an organism-based approach. Ecological Applications 16: 1076-1089. BÉLISLE M. 2005. Measuring landscape connectivity: the challenge of behavioral landscape ecology. Ecology 86: 1988–1995 CHIN, K. 2005. Within peatland spatial structuring and the influence of the matrix on inter-peatland movement of the dragonfly, Leucorrhinia hudsonica in Western Newfoundland. MSc. Thesis, Acadia University, Wolfville, NS. CRONIN, J.T. & J.D. REEVE. 2005. Host-parasitoid spatial ecology: a plea for a landscape-level synthesis. Proceedings of the Royal Society – B. 272: 2225-2235. CORBET, P. S. 1999. Dragonflies: Behavior and Ecology of Odonata. Cornell University Press, Ithaca. HARRISON, S. 1994. Metapopulations and conservation. In: Edwards, P.J., R.M. May & N.R. Webb, Large-scale ecology and conservation biology. Blackwell, Oxford. JOHNSON, C. 1962. Breeding behaviour and oviposition in Calopteryx maculatum (Beauvois) (Odonata: Calopterygidae). American Midland Naturalist 68: 242-247. JONSEN, I.D. & P.D. TAYLOR. 2000a. Landscape structure and fine-scale movements of Calopterygid damselflies. Oikos 88: 553-562 JONSEN, I & P.D. TAYLOR. 2000b. Calopteryx damselfly dispersions arising from multi-scale responses to landscape structure. Conservation Ecology 4(2): 4. [online] URL: http://www.consecol.org/vol4/iss2/art4 KRAWCHUK, M.A. & P.D. TAYLOR. 2003. Changing importance of habitat structure across multiple spatial scales for three species of insects. Oikos 103: 153-161.
CyanMagentaYellowBlack Odonata page 238
238 Philip D. Taylor
MCPHERSON, A. M. 2003. Between-patch movements of a peatland dragonfly (Leucorrhinia hudsonica), and the influence of landscape structure on its distribution and abundance in western Newfoundland. M.Sc. Acadia University, Wolfville, Canada. PITHER, J. & P.D. TAYLOR. 1998. An experimental assessment of landscape connectivity. Oikos 83: 166-174. R DEVELOPMENT CORE TEAM (2005). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org. ROLAND, J., N. KEYGHOBADI & S. FOWNES. 2000. Alpine Parnassius butterfly dispersal: effects of landscape and population size. Ecology 81: 1642-1653. RICKETTS, T. H. 2001. The matrix matters: effective isolation in fragmented landscapes. The American Naturalist 158: 87-99. TAYLOR, P.D. & G. MERRIAM. 1995. Wing morphology of a forest damselfly is related to landscape structure. Oikos 73: 43-48. TAYLOR, P. D., L. FAHRIG, HENEIN K & MERRIAM G. 1993. Connectivity is a vital element of landscape structure. Oikos 68: 571-573. TAYLOR, P. D., L. FAHRIG & K. A. WITH. Landscape connectivity: back to the basics. In: K. Crooks & S. Muttulingam (eds), Maintaining Connections for Nature, Cambridge University Press, Cambridge, UK. TURCHIN, P. 1998. Quantitative analysis of movement. Sinauer, Mass. VENABLES, W. N. & B. D. RIPLEY. 2002. Modern Applied Statistics with S, 4th edition. Springer, New York, USA. WALTERS, C. 1997. Challenges in adaptive management of riparian and coastal ecosystems. Conservation Ecology [online] 1(2): 1. Available from the Internet. URL: http://www.consecol.org/vol1/iss2/art1/
CyanMagentaYellowBlack Odonata page 239
The structure of the CoenagrionAdolfo mercuriale populations in the Cordero Rivera (ed) 2006New Forest, southern England Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 239-258.
239
© Pensoft Publishers
Sofia–Moscow
The structure of the Coenagrion mercuriale populations in the New Forest, southern England David J. Thompson1 & Phillip C. Watts School of Biological Sciences, The Biosciences Building, Crown Street, University of Liverpool, Liverpool L69 7ZB, UK 1 Corresponding Author: same address.
[email protected]
ABSTRACT The damselfly Coenagrion mercuriale (Charpentier) is a poor disperser and susceptible to habitat fragmentation/loss. It is protected by European legislation. An analysis of Capture-Mark-Recapture (CMR) data indicated that the population network on Beaulieu Heath in the New Forest comprised some 40 000 individuals. A nationwide genetic study indicates that the New Forest is a principal reservoir of genetic diversity for UK C. mercuriale. The New Forest is, however, presently best characterised as five genetic units. We found that several small, isolated populations of C. mercuriale in the New Forest showed substantial genetic differentiation from the principal populations on Beaulieu Heath, Setley Plain and Mill Lawn Brook. Isolation is bought about by preventing dispersal across intervening areas of unsuitable habitat such as forest, farmland or road. Although habitat loss is a principal concern for the persistence of this species, the pattern of limited movement to proximate sites highlights the need for a network of suitable habitat patches. This will also help to slow the rate of genetic erosion at peripheral sites. Key Words: Odonata, dragonfly, capture-mark-recapture, population size, population structure, microsatellite
INTRODUCTION Coenagrion mercuriale (Charpentier) is one of Europe’s highest profile damselfly species from a conservation perspective. This paper is concerned
CyanMagentaYellowBlack Odonata page 240
240 David J. Thompson & Phillip C. Watts
Fig. 1. Location of the New Forest in Great Britain; forested areas are highlighted green.
with the population structure of this species in its UK stronghold, the New Forest of southern England (Fig. 1). This species has a somewhat fragmented population structure throughout its range and this is even more apparent at its range margin in the UK. Some preliminary results from a multi-site capture-mark-recapture (CMR) project are discussed together with genotype data at 14 microsatellite loci. Together, our results give an indication of the likely structure of the New Forest populations both from short-term (ecological) and historical perspectives. They also point the way towards resolving potential conservation problems in the medium to long term.
The New Forest It is appropriate to be discussing the odonate fauna of England’s New Forest in 2005 as on 1st March the New Forest was designated as the UK’s eighth National Park. The name New Forest does not reflect its new status as a National Park. The area has been known as the New Forest since it was created as a medieval royal deer hunting area in 1079 by William I, William the Conqueror. The New Forest is situated in southern England in the county of Hampshire (Fig. 1) and is the UK’s smallest National Park in area, but with the highest human population density of all the National Parks (Table 1). The inhabitants retain some of the rural practices conceded by the Crown in medieval times such as the pasturing of ponies, cattle, pigs and donkeys in the open forest. These rights will be maintained in the Forest’s new role as a
CyanMagentaYellowBlack Odonata page 241
The structure of the Coenagrion mercuriale populations in the New Forest, southern England
241
Table 1. Some parameters of the New Forest National Park, southern England, UK. Human population Area Area covered by the ‘Perambulation’, within which commoners’ rights apply. Area of national or international importance for nature conservation Scheduled ancient monuments Listed buildings Conservation areas
34 400 57 086 ha 37 677 ha (66 %) 31 968 ha (56 %) 61 634 18
National Park. It is these commoner’s rights that are the key to the survival of Coenagrion mercuriale in the New Forest in that watercourses are seldom undergrazed. Coenagrion mercuriale The European distribution of Coenagrion mercuriale (Askew 1988, Fig. 2) is restricted at both global and national level. It is mainly limited to the south and west of Europe and has populations of unknown status in northern
Fig. 2. Distribution of C. mercuriale in Europe (adapted from Askew 1988).
CyanMagentaYellowBlack Odonata page 242
242 David J. Thompson & Phillip C. Watts
Africa. Populations in Italy and northern Africa consist of different subspecies (C. m. castellani and C .m. hermeticum respectively) to those found in the rest of Europe (Askew 1988). C. mercuriale is protected within Europe as a whole and several European countries have taken complementary legislative measures for protection at a national or regional level (Table 2). The UK distribution of C. mercuriale is shown in Fig. 3. There are population strongholds in the New Forest and Test and Itchen Valleys (all situated in Hampshire), the heathlands of Dorset and the Preseli hills of Pembrokeshire, with isolated populations in Anglesey, the Gower, Oxfordshire, the east Devon pebble beds and Dartmoor. The species has suffered a 30% decline in UK distribution since 1960. It has been lost from Cornwall, some Devon and Dorset sites and from St. David’s peninsula in Pembrokeshire. Even within the New Forest it has disappeared from some sites in the last ten years. However intensive research effort has led to the discovery of some other sites. Coenagrion mercuriale occurs in three biotopes in the UK. Most populations occur in heathland streams or boggy runnels emanating from base-rich substrate. The Test and Itchen populations occur in the water meadow ditch systems surrounding these chalk streams. The third biotope is fen in which there is some water flow (e.g. the Anglesey and Oxfordshire sites). There are several threats to the continued long term persistence of Coenagrion mercuriale in the UK (Table 3), the most important of which is likely to be undergrazing (Thompson et al. 2003). Table 2. Protection measures for C. mercuriale (adapted from Thompson et al. 2003). Legislation or convention
Level
Listed on the Bonn Convention for the conservation of Migratory Species of Wild Animals. Listed on Appendix II of the Bern Convention on the Conservation of European Wildlife and Natural Habitats (1979), which outlaws the collection and possession of listed species. Listed on Annex II of the European Community Habitat and Species Directive (1992), which requires the designation of Special Areas of Conservation (SACs) for animal and plant species of community interest. Listed on Schedule 5 of the Wildlife and Countryside Act (1981), which protects against damage and killing of individuals, and damage or destruction of habitat, and protects biotopes in localities designated as Sites of Special Scientific Interest (SSSIs). Listed as rare (Category 3) in the British Red data Book, and also features on the red lists of other European countries (Grand 1996). Subject of the UK Biodiversity Action Plan (HMSO 1994).
International International
Europe
UK
UK (and Europe) UK
CyanMagentaYellowBlack Odonata page 243
The structure of the Coenagrion mercuriale populations in the New Forest, southern England
243
Fig. 3. The distribution of C. mercuriale throughout the UK. The colours represent different centres of population and are retained in later analysis (see Fig. 8). Table 3. Threats to long-term persistence of Coenagrion mercuriale in the UK (after Thompson et al. 2003). Threat Fragmentation and population size Undergrazing
Channel silting and management
Water abstraction
Eutrophication
Many British populations are isolated from other populations by tens of kilometres. Isolated populations are more prone to extinction. Changes in grazing regimes on some sites has undoubtedly played a major role in the decline of C. mercuriale. Grazing by heavier animals such as cattle and horses creating poaching of water course margins is preferred. Ditches in the flood plains of chalk streams are more prone to silting than sites with a more natural hydrology. Overzealous clearance of channel vegetation can be almost as serious a threat if it is not undertaken with the life cycle of C. mercuriale in mind. Water abstraction from aquifers or springs feeding sites is a potential threat to some sites on the northern side of the Mynydd Preseli sites in Pembrokeshire. Few British sites are directly threatened by nutrient runoff, but, significantly, several of them are small, isolated populations (e.g. Anglesey).
CyanMagentaYellowBlack Odonata page 244
244 David J. Thompson & Phillip C. Watts
The distribution of Coenagrion mercuriale throughout the New Forest is shown in Fig. 4. The New Forest consists of a series of gravel terraces deeply eroded by river valleys. These terraces are highest in the north which makes the river valleys that drain west into the River Avon most prominent. In the south these gravel deposits have been eroded away, removing most of the acidic strata and exposing more of the base-enriched Headon Beds. Most Coenagrion mercuriale sites have their origins in springs that emerge from the Headon Beds. The aim of this paper is to provide a general overview into the extensive population research on this species in its UK population stronghold. For this, we discuss two data sets (i) a CMR study at Beaulieu Heath in the New Forest, and (ii) one subset of data on genetic variation in the UK’s Coenagrion mercuriale population.
Fig. 4. Locations of C. mercuriale sites in the New Forest National Park, southern England; forested areas are marked in green. Also shown are the predominant proportions of membership to one of five model clusters (see Fig. 9) that were defined by a Bayesian genetic analysis (see text for details). Full sample location names are provided in Appendix 1.
CyanMagentaYellowBlack Odonata page 245
The structure of the Coenagrion mercuriale populations in the New Forest, southern England
245
MATERIALS AND METHODS Fieldwork Coenagrion mercuriale on Beaulieu Heath (50°47.8’N, 01°29.9’W) inhabits a network of small flushes and runnels that may be subdivided into seven central areas and four peripheral sites (Fig. 4). At each sub-site around Beaulieu Heath a pair of research assistants captured mature damselflies with a kite net and recorded the location using a Global Positioning System (GPS) calibrated to the UK Ordnance Survey. Animals were marked by writing a unique alphanumeric code on the left hindwing in waterproof ink and by putting a small dab of paint on the thorax. Locations were recorded on all subsequent sightings using a GPS. We sampled every day for five weeks from 11th June 2002, except during bad weather when adult damselflies are not active, which was timed to coincide with the peak flight period in this area. Estimation of population size Daily population sizes were calculated using a full Jolly-Seber model for open populations (Pisces Conservation Ltd. 2003). The numbers of damselflies present outside our sampling (and also for days when no CMR was undertaken) was estimated from a logistic growth (or decline) trend based on the increasing (or declining) daily population estimates and a zero population on the first (or last) date of on which Coenagrion mercuriale have been recently sighted in southern England (6th May and 25th September in 2004, D.K. Jenkins pers. comm.). Population estimates at Beaulieu Heath on 23 rd June, 4th July and 6th July appeared downwardly biased (Fig. 5), probably because poor weather reduced capture efficiency, and were replaced with census sizes estimated from the logistic growth/decline trend. Although the genetic odonate sex determination mechanism should lead to an equivalent number of each sex (Kiauta 1969), the encounter rate of female C. mercuriale is significantly lower than that of males. Conversely, note that Purse & Thompson (2003) recorded a slight male bias of emerging C. mercuriale. Rather than using all CMR data and underestimate population size, censuses were made using the encounter rates of males only, and then doubled to account for (unobserved) females. Total population size was calculated by dividing the sum of all daily census estimates by the average mature adult lifespan, which was estimated to be the average number of days between first and last captures.
CyanMagentaYellowBlack Odonata page 246
246 David J. Thompson & Phillip C. Watts 6,000
Daily numbers of males
5,000
4,000
3,000
2,000
1,000
0 8-May 22-May
5-Jun
19-Jun
3-Jul
17-Jul
31-Jul 14-Aug 28-Aug 11-Sep 25-Sep
Date
Fig. 5. Daily estimates (± standard errors) of male population size of C. mercuriale on Beaulieu Heath, New Forest, southern England. Estimates were made using a full Jolly-Seber model. Open circles and solid line are estimated data extrapolated from the daily estimates.
Genotype data collection and analysis We took tissue samples from up to 90 individuals from all of the UK’s Coenagrion mercuriale populations. One hind leg per individual was taken and stored in 100 % ethanol until analysis. See Appendix I for full details of sample dates and sample sizes. Genomic DNA was extracted using a high salt protocol (Sunnucks & Hales 1996). We examined allelic variation in 14 unlinked microsatellite loci (LIST4-002, LIST4-023, LIST4-024, LIST4-030, LIST4-031, LIST4-034, LIST4-035, LIST4-037, LIST4-042, LIST4-060, LIST4062, LIST4-063, LIST4-066 and LIST4-067) described by Watts et al. (2004a, b). Full details of the PCR and genotyping procedures using an automated sequencer are given by Watts et al. (2004a, b, c). We have used the microsatellite data in two ways in this paper. First, by principal component analysis (PCA) in attempt to explain the multivariate data set by a few linear combinations of the original variables whilst still retaining nearly as much of the total variation between samples (Johnson & Wichern 1992). A plot of the sample scores (eigenvectors) of significant principal components offers a convenient representation of the overall spatial variation in data as long as the principal components still
CyanMagentaYellowBlack Odonata page 247
The structure of the Coenagrion mercuriale populations in the New Forest, southern England
247
account for a significant amount of the total between-sample variation. Rather than attempt to describe the spatial pattern of individual allele frequencies (127 separate variables, i.e. alleles) the multi-allelic variation between damselfly samples was reduced to two-dimensions by a PCA of the sample allele frequencies using PCA-GEN v.1.2.1 (Goudet 1999). The significance of each principal component was assessed from 5 000 randomisations of genotypes. Second, the population genetic structure of the New Forest samples was assessed in more detail using the model-based clustering approach implemented by STRUCTURE v. 2.0 (Pritchard et al. 2000). This approach simultaneously identifies clusters (populations) and assigns individuals to populations using a Bayesian approach. Briefly, STRUCTURE models K populations (where K may be unknown) that are characterised by a set of allele frequencies at each locus. Individuals are (probabilistically) assigned to populations (or jointly if their genotypes indicate that they are admixed) on the basis of their multilocus genotypes, assuming unlinked loci and Hardy-Weinberg equilibrium conditions within populations. The number of distinct populations (K) may be estimated from the value of K that maximises the posterior probability of the data for a given posterior probability distribution Pr(K|X) that is calculated from the posterior distribution of Pr(X |K) (where X is the genotypes of sampled individuals). It is worth noting that rather than determine the actual number of populations this approach provides a heuristic guide to the models that are most consistent with the data set. Independent runs of structure were carried out for the total data set for K=1 to K=9 using the admixture model. All model runs were based on 500 000 iterations after an initial burn-in period of 50 000 iterations. Five independent runs were made for each value of K to assess consistency of the results.
RESULTS Estimates of population size We sampled adults during the peak flight season, as is evident from the estimates of daily population size (±SE) at Beaulieu Heath (Fig. 5). The numbers of males reached a maximum of some 5-6 000 males per day (during late June). Using a mean mature adult lifespan of 5.93 days provided an estimate the total number of individuals on Beaulieu Heath during the summer of 2002 of 39 913. This calculation is based on the 10 259 (4 158) individuals actually marked (and recaptured) during the study. The relative population sizes at each Beaulieu Heath sub-site (Fig. 6) were estimated as the proportion of marked animals at each site. The smallest populations were at the peripheral sites Bagshot Moor, Greenmoor and Hatchet Stream.
CyanMagentaYellowBlack Odonata page 248
248 David J. Thompson & Phillip C. Watts
Fig. 6. Movement of Coenagrion mercuriale between sub-sites on Beaulieu Heath, New Forest, southern England. The diameter of the circles represents the estimated population sizes of the sites. The arrows indicate the direction and the number of individuals that moved. Full sample location names are provided in Appendix 1.
Pattern of movement In the CMR study we were looking primarily at the dispersal potential of Coenagrion mercuriale in heathland. The overall pattern of movement between the sub-sites on Beaulieu Heath (Fig. 6) resulted in a limited interchange among most pairs of populations, except among the three Peaked Hill sites and Lower Crockford. Interchange was limited to neighbouring areas in almost all cases. The large population at Roundhill (NW of Beaulieu Heath) and the next most northerly population at Hatchet Stream proved to be isolated, at least during the present study. The central sites on Beaulieu Heath are bisected by a road (Fig. 6) that did not prevent movement. This finding was in agreement with Purse et al. (2003) who also recorded movement across the road. However, dispersal was limited to a single individual and only in the direction indicated. Single damselflies were observed moving in and out of the small, isolated populations at Greenmoor and Bagshot Moor. Purse et al. (2003) indicated that C. mercuriale moved very little in its mature adult lifetime, but their study provided no data on the tail of the distribution. Fig. 7 shows net lifetime movement (defined as the dis-
CyanMagentaYellowBlack Odonata page 249
The structure of the Coenagrion mercuriale populations in the New Forest, southern England
249
2,000
Frequency
1,500
1,000
500
2 3 1
1
2
1
1
92 5 10 00 10 75 11 50 12 25 13 00
3
85 0
3
77 5
70 0
55 0
47 5
40 0
32 5
25 0
17 5
10 0
25
62 5
2 1
0
Distance (m)
Fig. 7. Net lifetime movement in mature adult Coenagrion mercuriale Beaulieu Heath in the New Forest, southern England.
tance between first and last sighting) of mature adult (both sexes) C. mercuriale on Beaulieu Heath in the New Forest. Seventy per cent of mature adults moved less than 50 m in their mature adult lifetimes and 85 % moved less than 100 m. However, five individuals (0.12 %) moved more than 1 km, with 1.25 km the greatest distance moved in this study. In a parallel study in the more linear habitat of water meadow ditch systems the pattern was generally similar with the longest recorded distance being 1.79 km (Watts et al. 2004c). Population genetic structure The first two principal components (Fig. 8) account for 24 % and 17 % of the variation within the data and are significant (P < 0.001 for each axis). The PCA plot is based on allele frequencies, that is, shared or similar alleles. It has little real ‘genetic interpretation’ other than that more closely related populations might be expected to share alleles. The New Forest populations generally occur in the centre of the plot because they contain more genetic variation than other populations. Those from Dorset are the closest in allele frequencies to the majority of the New Forest populations and there is some overlap. Also falling in the ‘New Forest’ region is one population from Normandy in France (Saint-Sulpice-de-Grimbouville) that is comparable to the New Forest in terms of levels of genetic variation and some of the most common alleles (although the French sample does possess some unique alleles, data not shown). In general, populations from similar geographical areas have, for the most part, clustered together (Fig. 8). For example, the Pembrokeshire populations are grouped in the top left quad-
CyanMagentaYellowBlack Odonata page 250
250 David J. Thompson & Phillip C. Watts
rant, while small, isolated populations fall furthest away from the New Forest, for example, with the Anglesey population (2002 and 2003 datasets) falling in the top right of the top right quadrant. There are, however, some exceptions whereby some New Forest populations are separate, notably Acres Down, Shobley, Common Moor and Kingston Great Common, while at least one isolated population, Oxfordshire (2002 and 2003 datasets), is positioned with the main New Forest cluster (Fig. 8). With respect to the New Forest itself, the lowest posterior probability of the data (PPD) indicate that the New Forest appears to contain five distinct genetic ‘clusters’ (average Ln PPD = -30,991 for K=5, Figs. 4 & 9). The three ‘best defined’ clusters (with regard to the proportion of membership of individuals) include samples identified by PCA (Fig. 8) as being quite distinct: Acres Down, Shobley and Common Moor (Fig. 9). Also similar to the latter sample are other populations that drain into Mill Lawn Brook (the River Oberwater) plus Stony Moors. The fourth cluster
Fig. 8. Principal component analysis plot showing spatial pattern of allele frequencies in the UK C. mercuriale populations. The coloured symbols reflect different centres of population (see Fig. 3). One French population (from Saint-Sulpice-de-Grimbouville, Normandy - SSG) is also plotted. Two Devon populations (Moortown Gidleigh Common and Aylesbeare Common), the Anglesey and Oxfordshire populations have two points representing sampling across two years. Full sample location names are provided in Appendix 1.
CyanMagentaYellowBlack Odonata page 251
The structure of the Coenagrion mercuriale populations in the New Forest, southern England
251
comprises populations at Setley Plain, Three Beech Bottom, Widden Bottom, and also Kingston Great Common. The final cluster includes samples from Gypsy Hollies and Foulford and almost ‘by default’, a poorly defined group comprising all Beaulieu Heath sites that were included in the CMR study. Future analyses will determine whether the northern populations flowing into Millersford Brook and Latchmore Brook (Fig. 4) proves to be distinct or not. With caution, these data may be summarized by the proportion of membership of individuals from each of the pre-defined populations to each of the five model clusters (Fig. 4). Again, the most distinct populations are Acres Down, Common Moor and Shobley where 82 %, 78 % and 67 % of individuals respectively are assigned to a particular cluster. Individuals from three sites (Mill Lawn, Stag Brake Bog, Stony Moors) near Common Moor are also predominantly assigned (26-37 %) to the ‘Common Moor’ cluster, while those from Setley Plain form the fourth group whereby 42-48 % of individuals from the samples are assigned to that cluster. In contrast, both Foulford and Kingston Great Common sites show genetic differences to nearby populations. The Beaulieu Heath samples are similar in that they all show no strong affinity to any of the five model clusters; hence, while two Peaked Hill sites appear to be similar to the ‘Acres Down’ cluster this simply reflects some 3 % of the sample (ca. 1 individual) clustering with Acres Down rather than within the ‘Foulford-Kingston Great Common’ group.
Fig. 9. Clustering results for all populations of Coenagrion mercuriale sampled from the New Forest (2002) for K=5. Every individual is represented by a vertical bar that is partitioned into 5 coloured segments, the lengths of which are proportional to the individual’s estimated membership in each of the 5 clusters. Solid black lines separate different samples (labelled below the figure) that are arranged in a clockwise direction from Acres Down through Beaulieu Heath up to Gypsy Hollies (see Fig. 4 for sample locations).
CyanMagentaYellowBlack Odonata page 252
252 David J. Thompson & Phillip C. Watts
DISCUSSION Population size These data, the first published estimates of population size in Coenagrion mercuriale, suggest that the New Forest Population is relatively healthy and certainly above, for example, than the 50/500 limit suggested to retain reproductive fitness or evolutionary potential (Frankham et al. 2002). A recent meta-analysis sparked a debate about the significance of genetic factors in precipitating species extinction (Spielman et al. 2004; DeSalle 2005) but, unfortunately, there are too few relevant studies to establish whether poor sample size was the reason behind the lack of evidence for a genetic impact upon insect taxa (in contrast to plants and large vertebrates). At least one well-documented instance of inbreeding depression in an insect (Saccheri et al. 1998) argues against this. However, it is likely that habitat loss/degradation, rather than genetic factors per se, pose a more immediate threat to the persistence of this species in the New Forest (see also Thompson et al. 2003). While we have data on Beaulieu Heath, there are still no more than qualitative estimates of population size for Coenagrion mercuriale elsewhere. There is a clear need for future work to correlate these estimates of population size with standardized transect counts so that the population demography of this species may be monitored with some quantitative meaning. Movement Coenagrion mercuriale is a species that occurs in an even more fragmented landscape than most other damselfly species because of its rather particular habitat requirements (Thompson et al. 2003). It is one of the smallest of the blue damselflies and body size has been correlated with dispersal capability in some odonates (Conrad et al. 1999; Angelibert & Giani 2003). From the present study and that of Purse et al. (2003) it is clear that most individuals do not move more than 100 m during their mature adult lifespans. There was relatively little movement between many of the patches of suitable habitat connected by the same stream (which provided a corridor for movement), and where movement was observed, it was almost exclusively between adjacent sites (Fig. 6). Given that many sites are separated by more than several kilometres of unsuitable (forested) habitat, we would expect to find a large number of more or less isolated populations within the New Forest and this is supported by genetic analysis (Figs. 4 & 9). On the other hand, although most individuals do not move far, a small percentage does move up to about 1.2 km; if these animals breed then gene flow between sites separated by 1-2 km seems assured. Watts et al. (2004c) in a parallel study to the present one,
CyanMagentaYellowBlack Odonata page 253
The structure of the Coenagrion mercuriale populations in the New Forest, southern England
253
but in the second biotope of C. mercuriale (water meadow ditch system surrounding chalkstreams), found similar patterns of movement. Hunger & Röske (2001) also observed limited movement by adult C. mercuriale. Genetic structure Analysis of microsatellite variation across the UK revealed that similar geographical regions tended to share similar alleles, with the New Forest samples forming a large central cluster indicative of the large amount of genetic variation in this region and hence its importance for biodiversity conservation. In general, increased levels of genetic differentiation from the main New Forest sites (i.e. outlying samples in Fig. 8) possess some combination of (i) geographic separation (e.g. Anglesey) that presumably correlates with an increased temporal separation, (ii) habitat isolation (e.g. Acres Down, Mariners Meadow), and (iii) small population size (e.g. Acres Down, Anglesey, Aylesbeare) where the effect of genetic drift is more pronounced. This pattern of differentiation is a feature of C. mercuriale population-genetic structure (Watts et al. 2004c, 2005). A more detailed look at the New Forest highlights the effects of genetic drift, but at a more localized scale. The Beaulieu Heath sites, separated by several kilometres, were not all linked during the CMR study but are indistinguishable genetically, thus indicating that this population is behaving like a metapopulation with the strong central sub-sites providing a source for the smaller peripheral sites. It is important to recognise that apparently separated populations will not show substantial genetic divergence when there is gene flow between intermediate populations. Appropriate management of streams (cutting down trees and shrubs) so that ponies can get closer to graze, at further peripheral sub-sites, is likely to lead to re-establishment of C. mercuriale there. The population at the apparently isolated site of Roundhill does not show substantial genetic differentiation likely because it is large and also as there has been insufficient time for substantial genetic drift. We do not exclude the possibility (more so for Hatchet Stream) that there is occasional immigration from the main Beaulieu Heath populations. The Setley Plain and Mill Lawn clusters probably behave in a similar way. The two populations that seem not to resemble any others, Acres Down and Shobley are particularly interesting. Acres Down is a small isolated population, at a higher altitude (70 m a.s.l.) than any of the other New Forest populations. The site is small, never likely to have held a large population, and was probably founded by a few individuals and seldom replenished genetically, if at all. We do not know whether Coenagrion mercuriale in this or even other small, isolated sites (Watts et al. 2005) suffer from inbreeding depression, but if so its long-term survival would probably be enhanced by translocation of individuals from the nearest, genetically similar populations
CyanMagentaYellowBlack Odonata page 254
254 David J. Thompson & Phillip C. Watts
at Beaulieu Heath. The Shobley population is more ‘problematic’. Although no CMR study or monitoring work has ever been carried out there (the site was only discovered in 2002), it is a large population. It is less than 1 km from the Foulford site (also discovered in 2002) but separated by a long high ridge carrying the main trunk road through the New Forest. The Shobley and Foulford sites are genetically dissimilar. Some combination of the ridge and road are evidently a barrier to movement between these two sites. Here, the effect of the road as a barrier contrasts with movement observed on the Crockford stream (see also Watts et al. 2004c) where water flow was still maintained between ‘separated’ sites by a bridge. Also different to surrounding sites are the populations at Common Moor and Kingston Great Common. Genetic differentiation at Common Moor is another example of anthropogenic isolation - here the surrounding land is farmed - which has been observed in this species in Devon (Watts et al. 2005). Kingston Great Common likely show genetic divergence because it is a weak population and susceptible to rapid divergence by genetic drift. To summarise, in the UK C. mercuriale exists as a number of isolated population fragments at the northern edge of its distribution. This species is a relatively sedentary odonate, a characteristic that combined with specialised habitat requirements makes C. mercuriale susceptible to the detrimental effects of habitat loss and fragmentation. An extensive genetic survey has demonstrated that allele frequencies in isolated populations can rapidly drift apart from the closest neighbouring populations. Despite concerns about its conservation, CMR data reveal the UK’s New Forest to sustain a large population of this species that has sustained similar levels of gene diversity as a (more central) European population but whether this is because the French population is isolated requires further study. Bayesian genetic analysis provides evidence that the New Forest stronghold is subdivided into several distinct genetic units and this needs to be considered for future biodiversity management.
ACKNOWLEDGEMENTS The work described in this paper was funded by the Natural Environment Research Council (grant no. NER/A/S/2000/01322) and the Environment Agency. Coenagrion mercuriale is protected under Schedule 5 of the 1981 Wildlife and Countryside Act in the UK. All work was carried out under licence from English Nature and DIREN Normandie.
CyanMagentaYellowBlack Odonata page 255
The structure of the Coenagrion mercuriale populations in the New Forest, southern England
255
REFERENCES ANGELIBERT, S. & GIANI, N. 2003. Dispersal characteristics of three odonate species in a patchy habitat. Ecography 26: 13-20. ASKEW, R. R. 1988. The dragonflies of Europe. Harley Books, Colchester. CONRAD, K. F., WILLSON, K. H., HARVEY, I. F., THOMAS, C. J. & SHERRATT, T. N. 1999. Dispersal characteristics of seven odonate species in an agricultural landscape. Ecography 22: 524-531. DESALLE, R. 2005. Conservation genetics - Genetics at the brink of extinction. Heredity 94: 386-387. FRANKHAM, R., BALLOU, J. D. & BRISCOE D. A. 2002. Introduction to Conservation Genetics. Cambridge University Press, Cambridge. GOUDET, J. 1999. PCA-General for Windows, ver. 1.2. www.2.unil.ch/izea/ softwares/pcagen.html. GRAND, D. 1996. Coenagrion mercuriale (Charpentier, 1840). In: van Helsdingen, P. J., Willemse, L. & Speight, M. C. D. (eds.), Background information on invertebrates of the Habitats Directive of the Bern Convention – Part II: Mantodea, Odonata, Orthoptera and Arachnida. Nature and Environment, Volume. 80, Council of Europe, Strasbourg. HUNGER H. & RÖSKE W. 2001. Short-range dispersal of the southern damselfly (Coenagrion mercuriale: Odonata) defined experimentally using UV fluorescent ink. Zeitschrift für Okologie und Naturshutz 9: 181-187. JOHNSON, R. A. & WICHERN, D. W. 1992. Applied Multivariate Statistical Analysis. Prentice Hall, Englewood Cliffs, NJ. KIAUTA, B. 1999. Sex chromosomes and sex determining mechanisms in Odonata, with a review of the cytological conditions in the family Gomphidae, and references to the karyotypic evolution of the order. Genetica 40: 127-157. PISCES CONSERVATION LTD. 2003. Simply Tagging. www.pisces-conservation.com. PRITCHARD, J. K., STEPHENS, M. & DONNELLY, P. 2000. Inference of population structure using multilocus genotype data. Genetics 155: 945-959. PURSE, B. V., HOPKINS, G.W., DAY, K. J. & THOMPSON, D. J. 2003. Dispersal characteristics and management of a rare damselfly. Journal of Applied Ecology 40: 716-728. PURSE, B. V. & THOMPSON, D. J. 2003. Emergence of the damselflies, Coenagrion mercuriale (Charpentier) and Ceriagrion tenellum (Villers) (Odonata: Coenagrionidae), at their northern range margins, in Britain. European Journal of Entomology 100: 93-99. SACCHERI, I. J., KUUSSAARI, M., KANKARE, M., VIKMAN, P., FORTELIUS, W. & HANSKI, I. 1998. Inbreeding and extinction in a butterfly metapopulation. Nature 392: 491-494. SPIELMAN, D., BROOK, B. W. & FRANKHAM R. 2004. Most species are not driven to extinction before genetic factors impact them. Proceedings of the National Academy of Sciences, USA 101: 15261-15264.
CyanMagentaYellowBlack Odonata page 256
256 David J. Thompson & Phillip C. Watts
SUNNUCKS, P. & HALES, D. F. 1996. Numerous transposed sequences of mitochondrial cytochrome oxidase I-II in aphids of the genus Sitobion (Hemiptera: Aphididae). Molecular Biology and Evolution 13: 510-524. THOMPSON, D. J., ROUQUETTE, J. R. & PURSE, B. V. 2003. Ecology of the Southern Damselfly, Coenagrion mercuriale. Conserving Natura 2000 Rivers Ecology Series No. 8. English Nature, Peterborough. WATTS, P. C., ROUQUETTE, J. R., SACCHERI, I. J., KEMP, S. J. & THOMPSON, D. J. 2004c. Molecular and ecological evidence for small-scale isolation by distance in an endangered damselfly, Coenagrion mercuriale. Molecular Ecology 13: 2931-2945. WATTS, P. C., THOMPSON, D. J. & KEMP, S. J. 2004a. Cross-species amplification of microsatellite loci in some European zygopteran species (Odonata: Coenagrionidae). International Journal of Odonatology 7: 87-96. WATTS, P. C., WU, J. H., WESTGARTH, C., THOMPSON, D. J. & KEMP, S. J. 2004b. A panel of microsatellite loci for the Southern Damselfly, Coenagrion mercuriale (Odonata: Coenagrionidae). Conservation Genetics 5: 117-119. WATTS, P. C., KEMP, S. J., SACCHERI, I. J. & THOMPSON, D. J. 2005. Conservation implications of genetic variation between spatially and temporally distinct colonies of the endangered damselfly Coenagrion mercuriale. Ecological Entomology 30: 541-547.
New Forest River Itchen
Acres Down Allington Manor Aylesbeare Bagshot Moor Bishopstoke Brynberian Cefn Bryn Clum Maen Colaton Raleigh Common Common Moor Corfe East Cors Tewgyll Deep Moor Dolau Isaf Dry Sandford Pit Foulford Gors Fawr Greenmoor Gypsy Hollies Hatchet Stream Highbridge Itchen Valley Country Park - Lower Itchen Valley Country Park - Middle Itchen Valley Country Park - Upper River Test (King’s Somborne) Kingston Great Common New Forest Preseli Hills New Forest New Forest New Forest River Itchen River Itchen River Itchen River Itchen River Itchen New Forest
Preseli Hills New Forest Preseli Hills
New Forest
Preseli Hills
New Forest River Itchen Preseli Hills
Region
Sample site Hampshire Hampshire Devon Hampshire Hampshire Pembrokeshire Gower Pembrokeshire Devon Hampshire Dorset Pembrokeshire Hampshire Pembrokeshire Oxfordshire Hampshire Pembrokeshire Hampshire Hampshire Hampshire Hampshire Hampshire Hampshire Hampshire Hampshire Hampshire
County ACD ALM AYL BAG BIS BRY CEB CLM COR COM COE COT DEM DOI DRS FOU GOF GRE GYH HAT HIG ITL ITM ITU KIS KGC
Code
49 48 48 48 39
47
2001
32
48 48 49 20 45 48 48 49 46 50 36 48 49 47 50
48 48
48
49
50
48
48
48
48
48
48
48
48
Sample Size / Year 2002 2003 2004
Appendix 1. Summary of sample sites, their locations and sample sizes used for genotype analysis.
CyanMagentaYellowBlack Odonata page 257
The structure of the Coenagrion mercuriale populations in the New Forest, southern England
257
New Forest River Itchen New Forest
Lower Crockford Stream Mariner’s Meadow Mill Lawn Moortown Gidleigh Common Nant Isaf Orchard Outflow Pantithel Peaked Hill - East Peaked Hill - West Rhossili Down Roundhill Setley Plain Shipton Bottom Shobley Saint-Sulpice-de-Grimbouville Stag Brake Bog Stony Moor Three Beech Bottom Two Bridges Bottom Twyford Moors Upper Crockford Stream Waun Fawr Waun Isaf North Waun Maes West Horton Widden Bottom Forest Forest Forest Forest
New Forest New Forest New Forest New Forest River Itchen New Forest Preseli Hills Preseli Hills Preseli Hills River Itchen New Forest
New New New New
Preseli Hills New Forest New Forest
Region
Sample site Hampshire Hampshire Hampshire Dartmoor Anglesey Dorset Dorset Pembrokeshire Hampshire Hampshire Gower Hampshire Hampshire Hampshire Hampshire Normandy Hampshire Hampshire Hampshire Hampshire Hampshire Hampshire Pembrokeshire Pembrokeshire Pembrokeshire Hampshire Hampshire
County LOCD MAM MIL MGC NAI ORC OUT PAN UPR UPL RHD ROU SEP SHB SHO SSG SBB STM TBB TWB TWM UPCD WAF WAN WAM WEH WIB
Code
48
30
83
2001
50
48 44 48 48
48 52 46 50
48 51 52 48 48 48 18 47 48 47 48 33 90
52
49
48
48 49
48
48
21
48
48
48 48
48
Sample Size / Year 2002 2003 2004
CyanMagentaYellowBlack Odonata page 258
258 David J. Thompson & Phillip C. Watts
CyanMagentaYellowBlack Odonata page 259
Mate Location and Competition for Mates Relation Adolfo Cordero Riverain(ed) 2006 to Sunflecks of Forest Floors Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 259-268. © Pensoft Publishers
Sofia–Moscow
Mate Location and Competition for Mates in Relation to Sunflecks of Forest Floors Mamoru Watanabe Graduate School of Life and Environmental Sciences, University of Tsukuba, Ibaraki 305-8572, Japan
ABSTRACT Although most forest odonate species have a maiden flight away from water and sexually immature adults stay in the forests foraging for food, mature males of some species (e.g. Platycnemis echigoana, Lestes sponsa and hyaline-winged males of Mnais pruinosa costalis) tend to remain in forests. To locate females in the forests, males mainly perch in sunflecks (a sunlit site in the forest floor) and adopt a sit-and-wait tactic. Some of them try to occupy perching sites. Territorial behaviour of males of the damselfly, P. echigoana, is described at sunflecks in climax deciduous forests. Males showed patrolling flight along the periphery of the sunfleck, and hovering flight above it, suggesting that such flight was a display associated with the occupation of the sunfleck. Flight behaviour of the damselfly, L. sponsa, in the forest floor also showed male-male interference and the existence of a lek-like mating system is discussed. Some solitary males interfered in copulation in the forest floor, while others were also observed on the shoreline of the pond throughout the day, but they did not harass pairs ovipositing in tandem. Although hyaline-winged males of M. pruinosa costalis adopt sneak tactics, a male that failed in occupying a perching site to intercept females entering the territory is called an ‘opportunist’, which moves around forest floor with sunflecks to search females. The longest copula duration was observed in the opportunists, suggesting that the entire sperm displacement must be occurred. These observations point to functional relationships with habitat selection and thermoregulation. Perching behaviour under direct sunlight at sunflecks was shown to result in considerable variation in thermoregulatory properties. The relationships of thermoregulation to mate location strategy are different among
259
CyanMagentaYellowBlack Odonata page 260
260 Mamoru Watanabe
species. A male that has been able to perch in direct sunlight will gain an advantage over an individual that has not, and this advantage may manifest itself in fights with other males. Forest structures with sunflecks are discussed from the viewpoint of habitat selection acting on female choice. These relationships are also relevant to other behaviours, particularly oviposition behaviour by water. Adults that showed mating behaviour in the forests oviposit in tandem by water. The importance of sunflecks in the forest floor is discussed in relation to the life history strategies of the damselfly species inhabiting forests.
INTRODUCTION Odonata exhibit a broad range of mate location tactics and mating systems. Reviews characterizing these systems have often examined factors that affect the distribution of receptive females. The number and duration of available females during the flight season (Corbet, 1999), the distribution of oviposition substrates (Tsubaki & Ono, 1986), and the behaviour of males and females at encounter sites (d’Aguilar et al., 1986), have been proposed as determinants of mating system structure. Receptive females are rarely randomly distributed, so that some locations are expected to be more profitable to searching males than others. Although such encounter sites are most likely to be oviposition sites, topographic features such as spots providing high visibility (Watanabe et al., 1987), or spots containing favourable thermal properties (Watanabe & Taguchi, 1993) have been reported. Mate location strategies can be divided into two main categories (cf. Corbet, 1999) relating to the situation in which the sexes meet, which can be described by the terms “perching” and “flying.” Flying males typically spend the majority of the time available for mate searching on the wing, patrolling in search of females. Perching males typically sit and wait at some vantage point and fly up to inspect any object remotely similar to a conspecific female that comes within their field of vision. The goal of this review is to describe several aspects of the mate location behaviour of damselflies inhabiting forests, and to consider the importance of the forests as habitats.
TERRITORY IN SUNFLECKS While many damselflies are non-territorial (e.g. Lutz, 1968; Robertson, 1982; Fincke, 1985; Banks & Thompson, 1985), some species show territorial behaviour near water (e.g. Johnson, 1962; Ueda, 1976; Utzeri et al., 1983). Corbet (1980) described territories as including a settling base from which
CyanMagentaYellowBlack Odonata page 261
Mate Location and Competition for Mates in Relation to Sunflecks of Forest Floors
261
the occupant surveys the territory, makes sallies towards intruders, and engages in intermittent patrolling flights. Few observations on territoriality of damselfly species away from water have been made, except for a white-legged damselfly, Platycnemis echigoana (Watanabe et al., 1987). In this species, most of the sexually mature males spend much time in the climax deciduous forests containing their pond oviposition sites, and perch in sunflecks. They defend such sunflecks from conspecific males, but never show territorial behaviour by the ponds (Watanabe & Ohsawa, 1984). The typical mating system of territorial damselflies is known as resource defence (e.g. Johnson, 1964). Males of many species defend oviposition sites and thereby gain access to receptive females that come to the area to lay their eggs (e.g. Corbet, 1980). The territorial behaviour shown by P. echigoana is a special case, in which males defend forest sunflecks preferentially visited by females (Watanabe & Ohsawa, 1984; Watanabe et al., 1987). However, males of many damselfly species do not have territories in forests, and aggregate in the reproductive areas of shallow ponds where they attempt to find their mates (e.g. Ueda, 1976). Oviposition in tandem has been observed in many non-territorial damselflies (e.g. Gower & Kormondy, 1963). The size of sunfleck is generally small (ca. 30X30 cm), and then, as a rule, a single male of P. echigoana occupies one sunfleck in the climax deciduous forests (Watanabe et al., 1987). The mean duration of occupation at a single sunfleck by a male was about 30 min in the morning when mating activity was high. Males encountered either a conspecific male or another damselfly species if moving to another sunfleck occupied. Males at a sunfleck showed some behaviours that could be easily differentiated: movement, predation, attack to intruders, disregard, and so on. Flight behaviour was limited at sunflecks. Males sometimes hovered just above the perching point, and such flight seemed to be not for direct defence, but for display to demonstrate occupation of the sunfleck, like the territorial song of birds. During the flights within a single sunfleck, the resident males of P. echigoana did not always encounter conspecific males (Watanabe et al., 1987). Most encounters with conspecific males (or females or other insects) occurred when the male was perching at the sunfleck. When another non-territorial male came into the sunfleck, the resident rapidly dashed to the intruder and usually chased him away. Such high territorial activity in the morning is related to the mating behaviour (copulation, tandem flight and oviposition) around noon. Once females visit the sunflecks, copulations take place on the shrub layer of the forest near the sunfleck. Following copulation the pair formed a tandem, and flew fast toward water where the female oviposited into floating stems, still in tandem.
CyanMagentaYellowBlack Odonata page 262
262 Mamoru Watanabe
LEK-LIKE SYSTEM IN THE FOREST FLOOR In Lestes species, the habitats for the sexually immature and mature stages are the same (Watanabe & Matsunami, 1990). The forest floor is also known to be the site for mating behaviour, and is a site which is not based on any resource for the damselflies. Campanella & Wolf (1974) pointed out that such a mating system was a kind of lek. Although many works on the behaviour and the ecology of L. sponsa have been conducted using the observation of mature individuals (e.g. Ueda, 1978), little attention has been given to the activity of both immature males and females staying on the forest floor. They perch mainly on the stems of Sasa paniculata, which is dominant in the deciduous forest floor in Japan. The distribution of perching sites in mature adults was similar to that of the immature ones. In the sexually immature stage of L. sponsa, most of the interference was seen when another male invaded the perching male’s space, which was bordered by patrolling flights. The intruder was driven away. The immature perching males attacked females as well as males. However, no perching females chased members of the same sex. The distribution of the mature males of L. sponsa in the forest floor was clumped, suggesting that the damselflies seemed to prefer sunflecks rather than shaded areas. There was a diurnal fluctuation in male-male interaction with a peak at noon. Whenever a female entered a male’s space during patrolling flights, the male dashed against her, and flew with her into nearby bushes. When they were in tandem, sperm transfer was observed. They then coupled. Copulations were restricted to within the forest floor, where the tandem couplings were frequently harassed by solitary males. They then flew rapidly to the shoreline to oviposit. Few solitary males along the shoreline disturbed the oviposition behaviour of tandems, in spite of frequent near misses. The occurrence of male and female L. sponsa together in the same patches of forest floor raises the possibility of some inhibition of unfavourable mating behaviour at certain times. Few abnormal couplings, such as mature males with immature females, were observed, because of their relatively synchronized emergence. Imprudent male attacks were overcome through a diurnal rhythm for reproductive behaviour. The difference between behavioural activities in immature and mature stages has been noted in many damselfly species (e.g. Lutz, 1968; Ueda, 1978; Corbet, 1980). L. sponsa males become more aggressive with ageing when they can acquire their mates only through competition with their rivals. The highest amount of interference by mature males in the forest floor was observed around noon, before reproductive behaviour took place.
CyanMagentaYellowBlack Odonata page 263
Mate Location and Competition for Mates in Relation to Sunflecks of Forest Floors
263
In non-territorial damselflies, the aggregation of solitary males at available oviposition sites has been explained as a mate-finding tactic (e.g. Banks & Thompson, 1985). Males of some species which search for females arriving at the oviposition site interfere with each other (e.g. Fincke, 1985). In L. sponsa, there is also an excess of solitary males at the oviposition site on the shoreline. Few of them mated because most females arrived in tandem from the forest floor and then oviposited. However, pairs in tandem were seldom disturbed by solitary males staying at the oviposition site. Such non-interference by solitary males at the oviposition site favours the males aggregating in the forest floor, where they have the chance to encounter their mates (Stoks, 2000). Encounter flight in L. sponsa has resulted in the rather constant male density in the forest floor. Consequently, the male density in the forest floor was relatively low, while male density in the oviposition site increased. Since all of the immature males were found in the forest floor, and not in the oviposition site, such a distribution in the forest floor might be due to interference flights of mature males. Since the forest floor was consistently used for encounters between mates, mature adults of L. sponsa were in reality forming a lek similar to those of the swallowtail butterfly, Papilio zelicaon (Shields, 1967). No report has been published proposing a lek or an arena in Odonata except for Plathemis lydia (Campanella & Wolf, 1974). However, such an odonate lek is not a true lek, defined as being removed from the nesting and feeding areas (e.g. Wilson, 1975).
THERMOREGULATION IN THE FOREST FLOOR Damselflies that stay in the understory of deciduous forests throughout their adult lives perch all day long with a few intermittent periods of flight. Most of the perching sites are in sunflecks, where they are exposed to direct sunlight. Radiant heat load might effectively contribute to an increase of the body temperature of damselflies perching in direct sunlight (Watanabe, 1991). The ability of dragonflies to maintain a relatively constant body temperature is related functionally to their body size and behaviour (Corbet, 1999). Although thermoregulation in some Odonata species inhabiting temperate zones has been investigated (Heinrich & Casey, 1978), data on the body temperature of damselflies inhabiting forests have so far been lacking. Insects, which retain much heat in their thorax during flight, require the highest muscle temperatures in order to maintain sufficient power output to continue flight (Heinrich, 1974). Essentially everything that an active adult damselfly does depend on being able to fly instantly and quickly: it catches prey in flight, takes mates on the wing and so on (Watanabe & Matsunami, 1990). Physiologically, the advantage of maintaining a high thoracic temper-
CyanMagentaYellowBlack Odonata page 264
264 Mamoru Watanabe
ature probably lies in permitting active flight (May, 1977). Although there is no direct evidence in Odonata, flight efficiency probably continues to increase with body temperature (Gibo, 1981). However, the length of time on the wing in damselflies inhabiting the forest floor is too short to increase body temperature. Alternatively, wing-beating produces so much heat that they may be unable to continue on the wing for a long time. L. sponsa was termed a “percher” (Corbet, 1999), which may display a wide range of behaviour, possibly related to thermoregulation (Heinrich & Casey, 1978), because they perch in sunshine. In some species, such temperatures are achieved either by basking (May, 1976) or by both metabolic heat production (shivering) and basking (May, 1977). However, no basking and shivering of L. sponsa adult have been observed at cool temperatures, particularly early in the morning or at dusk when the sun is visible but solar radiation is low. Wing beating and consequently heat production was not found, except when they shifted perching sites, fought against conspecifics or showed mating behaviour. However, every adult perching in sunflecks of the forest floor is capable of immediate and rapid flight without prior warm-up after a long duration of perching, even in the morning with relatively low ambient temperature.
WANDERING AROUND THE FOREST FLOOR FOR OPPORTUNIST TACTICS In eastern Japan, the damselfly Mnais pruinosa costalis shows male wing dimorphism. Orange-winged males usually defend territories for mating, while hyaline-winged males sneak copulations with females (Nomakuchi & Higashi, 1985; Watanabe & Taguchi, 1990). These latter males showed two kinds of mating tactics: satellite and wanderer. The former involves perching at the periphery of the territory of orange-winged males to intercept females entering the territory. Watanabe & Taguchi (1990, 1997) observed a series of competitive interactions between hyaline-winged males for a suitable perching site near the territory of orange-winged males. The expelled hyaline-winged males can be observed to hold no regular perching sites and are given the name ‘wanderers’. The territory occupied by orange-winged males includes substrates for oviposition. The probability of females being mated by the satellite hyaline-winged males was higher than that by orange-winged males (Watanabe & Taguchi, 1990). However, the hyaline-winged males failed to fertilize the eggs, because females could be re-mated with orange-winged residents after mating with the hyaline-winged males (Siva-Jothy & Tsubaki, 1989). In spite of female re-mating with orange-winged males, aggressive behaviour exhibited by hyaline-winged males at the perching site was directed
CyanMagentaYellowBlack Odonata page 265
Mate Location and Competition for Mates in Relation to Sunflecks of Forest Floors
265
predominantly towards, not the orange-winged males, but rather to other hyaline-winged males, which consequently became wanderers. Corbet (1980) stated that the nature and intensity of aggressive behaviour vary within species according to the density of males or to the distance from the centre of the territory. However, the limited number of perching sites elicited much more severe competition among the hyaline-winged males than among the orange-winged males for territory (Watanabe & Taguchi, 1997). Especially when sperm displacement occurs during copulation (Siva-Jothy & Tsubaki, 1989), reproductive success clearly results in strong sexual selection in favour of the orange-winged males being able to maintain a territory. Most females that copulated with a territorial orange-winged male began to oviposit within his territory. Accordingly, the hyaline-winged males were willing to perch at the periphery of a territory, where they tried to intercept arriving females. For the hyaline-winged males, however, the risk of take-over is great and guarding is not preferred. Thus their reproductive success might be low. Because wanderers usually stayed, not by water, but in the deciduous forests or edges of the forests, it was difficult to completely observe their behaviours. Watanabe (1991) pointed out that hyaline-winged damselflies preferred sunflecks in deciduous forests. It was clear that when a wanderer encountered a mate the copulation duration was long and complete sperm displacement might completely occur. Although most females that copulated with wanderers re-copulated with orange-winged males at the oviposition sites, there is a possibility that not all the sperm of wanderers was removed, since the copulation duration of territorial orange-winged males was relatively short (Watanabe & Taguchi, 1990)
CONCLUSION In general, immature individuals of most dragonfly species fly away from water (Corbet, 1999), and remain in the forest gaps generally use a sit-andwait foraging tactic in order to catch prey throughout the entire day. The size of gap is 15 m X 15 m (Watanabe et al., 2005), because a gap is typically generated by a few fallen trees. Higashi (1973) observed the feeding behaviour of Sympetrum frequens in coniferous forests where there was an excess of females. Since no mating behaviour is observed in forest gaps, the habitat for such species may be a feeding site, as well as, presumably, a roosting site. Watanabe et al. (2004) suggested that the large number of females of S. infuscatum in the forest gaps might develop clutches of eggs cyclically. All the adult males tended to simply perch, and did not interfere with the females. On the other hand, there is a number of sunflecks in the climax deciduous forest which is used by the damselflies. Although the abundance of the
CyanMagentaYellowBlack Odonata page 266
266 Mamoru Watanabe
sunflecks depends upon forest spatial structures, the size is very small due to the tiny space of the crown in the tree layer. The sunfleck itself moves with the movement of the sun from morning to evening. The occupation process at the sunfleck in the white-legged damselflies may be similar to the territoriality of the speckled wood butterfly, Pararge aegeria (Davies, 1978). The damselflies inhabiting forests also required time for maturation of teneral adults away from water in the adjacent forests, but did not return to water. Instead they stayed in their forests and showed mating behaviour there. Thus, the sites for mating (forest floor), oviposition (by water) and roosting (forest crown) are all important in the habitat of the damselfly inhabiting forests.
REFERENCES J., J.-L. DOMMANGET & R. PRECHAC. 1986. A Field Guide to the Dragonflies of Britain, Europe and North Africa. Collins, London. BANKS, M. J. & D. J. THOMPSON. 1985. Lifetime mating success in the damselfly Coenagrion puella. Animal Behaviour 33: 1175-1183. CAMPANELLA, P. J. & L. L. WOLF. 1974. Temporal leks as a mating system in a temperate zone dragonfly (Odonata: Anisoptera). I. Plathemis lydia (Drury). Behaviour 51: 49-87. CORBET, P. S. 1980. Biology of Odonata. Annual Review of Entomology 25: 189-217. CORBET, P. S. 1999. Dragonflies: behaviour and ecology of Odonata. Cornell University Press, New York. DAVIES, N. B. 1978. Territorial defence in the speckled wood butterfly (Pararge aegeria): the resident always wins. Animal Behaviour 26: 138-147. FINCKE, O. M. 1985. Alternative mate-finding tactics in a non-territorial damselfly (Odonata: Coenagrionidae). Animal Behaviour 33: 1124-1137. GIBO, D. L. 1981. Some observations on slope soaring in Pantala flavescens (Odonata: Libellulidae). Journal of the New York Entomological Society 89: 184-187. GOWER, J. L. & E. J. KORMONDY 19630 Life history of the damselfly Lestes rectangularis with special reference to seasonal regulation. Ecology 44: 398-402. HEINRICH, B. 1974. Thermoregulation in endothermic insects. Science 185: 747-756. HEINRICH, B. & T. M. CASEY. 1978. Heat transfer in dragonflies: ‘Fliers’ and ‘perchers’. Journal of experimental Biology 74: 17-36. HIGASHI, K. 1973. Estimation of the food consumption for some species of dragonflies. I. Estimation by observation for the frequency of feeding flights of dragonflies. Reports of the Ebino biological Laboratory, Kyushu Univ., (1): 119-129.-[Jap., with Engl. summary] JOHNSON, C. 1962. A description of territorial behavior and a quantitative study of its function in males of Hetaerina americana (Fabricius) (Odonata: Agriidae). Canadian Entomologist 94: 178-190. JOHNSON, C. 1964. The evolution of territoriality in the Odonata. Evolution 18: 89-92. D’AGUILAR,
CyanMagentaYellowBlack Odonata page 267
Mate Location and Competition for Mates in Relation to Sunflecks of Forest Floors
267
LUTZ, P. E. 1968. Life-history studies on Lestes eurinus Say (Odonata). Ecology 49: 576-579. MAY, M. L. 1976. Thermoregulation and adaptation to temperature in dragonflies (Odonata: Anisoptera). Ecological Monographs 46: 1-32. MAY, M. L. 1977. Thermoregulation and reproductive activity in tropical dragonflies of the genus Micrathyria. Ecology 58: 787-798. NOMAKUCHI, S. & K. HIGASHI. 1985. Patterns of distribution and territoriality in the two male forms of Mnais pruinosa pruinosa Selys (Zygoptera: Calopterygidae). Odonatologica 14: 301-311. ROBERTSON, H. M. 1982. Courtship displays and mating behaviour of three species of Chlorocyphidae (Zygoptera). Odonatologica 11: 53-58. SHIELDS, O. 1967. Hilltopping. Journal of Research on the Lepidoptera 6: 69-178. SIVA-JOTHY, M. T. & Y. TSUBAKI. 1989. Variation in copulation duration in Mnais pruinosa pruinosa Selys (Odonata: Calopterygidae). 1. Alternative mate-securing tactics and sperm precedence. Behavioral Ecology and Sociobiology 24: 39-45. STOKS, R. 2000. Components of lifetime mating success and body size in males of a scrambling damselfly. Animal Behaviour 59: 339-348. TSUBAKI, Y. & T. ONO. 1986. Competition for territorial sites and alternative mating tactics in the dragonfly, Nannophya pygmaea Rambur (Odonata: Libellulidae). Behaviour 97: 234-252. UEDA, T. 1976. The breeding population of damselfly, Cercion calamorum Ris (Odonata: Zygoptera). I. Daily movements and spatial structure. Physiological Ecology 17: 303-312. UEDA, T. 1978. Geographic variation in the life cycle of Lestes sponsa. Tombo 21: 27-34. UTZERI, C., E. FALCHETTI & G. CARCHINI. 1983. The reproductive behaviour in Coenagrion lindeni (Selys) in central Italy (Zygoptera: Coenagrionidae). Odonatologica 12: 259-278. WATANABE, M. 1991. Thermoregulation and habitat preference in two wing color forms of Mnais damselflies (Odonata: Calopterygidae). Zoological Science 8: 983-989. WATANABE, M. & E. MATSUNAMI. 1990. A lek-like system in Lestes sponsa (Hansemann), with special reference to the diurnal changes in flight activity and mate-finding tactics (Zygoptera: Lestidae). Odonatologica 19: 47-59. WATANABE, M., H. MATSUOKA, K. SUSA & M. TAGUCHI. 2005. Thoracic temperature in Sympetrum infuscatum in relation to habitat and activity (Anisoptera: Libellulidae). Odonatologica 34: 271-283. WATANABE, M., H. MATSUOKA & M. TAGUCHI. 2004. Habitat selection and population parameters of Sympetrum infuscatum during sexually mature stages in a cool temperate zone of Japan (Anisoptera: Libellulidae). Odonatologica 33: 169-179. WATANABE, M. & N. OHSAWA. 1984. Flight activity and sex ratios of a damselfly, Platycnemis echigoana Asahina (Zygoptera, Platycnemididae). Kontyu 52: 435-440. WATANABE, M., N. OHSAWA & M. TAGUCHI. 1987. Territorial behaviour in Platycnemis echigoana Asahina at sunflecks in climax deciduous forests (Zygoptera: Platycnemididae). Odonatologica 16: 273-280.
CyanMagentaYellowBlack Odonata page 268
268 Mamoru Watanabe
WATANABE, M. & M. TAGUCHI. 1990. Mating tactics and male wing dimorphism in the damselfly, Mnais pruinosa costalis Selys (Odonata: Calopterygidae). Journal of Ethology 8: 129-137. WATANABE, M. & M. TAGUCHI. 1993. Thoracic temperatures of Lestes sponsa (Hansemann) perching in sunflecks in deciduous forests of the cool temperate zone of Japan (Zygoptera: Lestidae). Odonatologica 22: 179-186. WATANABE, M. & M. TAGUCHI. 1997. Competition for perching sites in the hyalinewinged males of the damselfly Mnais pruinosa costalis Selys that use sneaky mate securing tactics (Zygoptera: Calopterygidae). Odonatologica 26: 183-191. WILSON, E. O. 1975. Sociobiology. The Belknap Press of Harvard University press, Cambridge.
CyanMagentaYellowBlack Odonata page 269
Differences in immune ability in Adolfo forest habitats varying quality: CorderoofRivera (ed) 2006 dragonflies as study models Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 269-278.
269
© Pensoft Publishers
Sofia–Moscow
Differences in immune ability in forest habitats of varying quality: dragonflies as study models Alex Córdoba-Aguilar* & Jorge Contreras-Garduño *Author for correspondence:
[email protected] Laboratorio de Ecología de la Conducta de Artrópodos, Instituto de Ecología, Universidad Nacional Autónoma de México, Apdo. P. 70-275, Circuito Exterior, Ciudad Universitaria, 04510, Coyoacán, Distrito Federal, México
ABSTRACT In this chapter we review the potential use of dragonflies for testing current ideas of differences in immune ability related to habitat quality. It is known that immune ability in insects can be affected by a number of biotic and abiotic factors. We briefly review these factors in dragonflies. Given the fact that the same species of dragonfly may live in forests of varying quality (e.g. food abundance), this can lead to immune ability differences among dragonfly populations. We examine the literature regarding this, in particular studies of varying parasite burden and immune ability to advance the hypothesis that forest quality can be assessed using immune ability. One particular trait that may be used for this is male wing pigmentation. Current knowledge suggests that this trait is sexually selected (the more pigmentation, the more successful the male is in leaving more offspring), sensitive to environmental stress (such as food supply) and an indicator of immune ability. These conditions make pigmentation ideal to see the response of dragonflies to forest quality and environmental stress.
INTRODUCTION Recently, evolutionary ecologists have centered their interest to the study of the immune system, particularly in insects (reviewed in Rolff & Siva Jothy 2003; Schmid-Hempel 2003, 2005). The basic question under investigation is:
CyanMagentaYellowBlack Odonata page 270
270 Alex Córdoba-Aguilar & Jorge Contreras-Garduño
what are the causes that underlie variation in immune response? (SchmidHempel 2003, 2005). Several possible causes include the cost of the use and maintenance of the immune system, differential response to different pathogens, sexual differences (for example, females having a better immune system than males; Rolff, 2002) and/or differences in life history (how reproduction, growth and lifetime affect the immune response; Rolff & Siva Jothy 2003; Schmid-Hempel 2003, 2005). One way to analyze these causes is to compare populations of the same species (Zuk & Stoehr 2002) to see the effect of genetic and environmental factors on immune ability (understood as a quantifiable way for assessing how good an individual is in defending itself against infections and/or maintaining the necessary physiological processes to keep this ability; the surrogates of immune ability most frequently used are different components of immune defense which include phenoloxidase activity and melanization [see below for an explanation of how these components work]). Despite this utility, there has been little research on the comparison of immune ability between populations, and firm conclusions have not been reached since the results are contradictory. Some research has been carried out on beetles and crustaceans. It was found that populations of the beetle Popillia japonica that are resistant to the parasite Metarhizium anisoplie showed greater phenoloxidase activity compared to less resistant populations (Tucker & Stevens 2003). A similar result was observed in Daphnia magna (Mucklow & Ebert 2003). In this species, clones of different populations were used to test resistance to the bacteria Pasteuria ramosa. It was observed that the most resistant populations had higher levels of phenoloxidase compared to populations with reduced resistance. However, another study of the same species and three populations similar to those of the above study, did not find population differences in phenoloxidase expression when faced with four different parasite species (Mucklow et al. 2004). In this chapter we propose that dragonflies are good models for investigating differences in immune response caused by environmental factors using the basis that the same species is able to live in contrasting forest environments which would vary in quality. Our goal here is to explore the role of environmental elements to propose that dragonflies can be used to assess the impact of different habitats (interpreted as forests with varying quality) on immunity. In particular we pay attention to the fact that since dragonfly male wing pigmentation is a trait sensitive to environmental changes and is largely correlated with immune ability (see below), this can be one trait to test this idea.
CyanMagentaYellowBlack Odonata page 271
Differences in immune ability in forest habitats of varying quality: dragonflies as study models
271
THE IMPORTANCE OF HABITAT OF DRAGONFLIES AND ITS RELEVANCE TO IMMUNE ABILITY The dragonfly habitat must contain sites for larval development, emergence, foraging, mate search, mating, oviposition and nocturnal perches (Corbet 1999). Availability and variation in quality of these sites can vary from one population to another (Corbet 1999) which means that there is a great variety of biotic and abiotic factors that could differ among populations. To cite one example, males of Calopteryx haemorrhoidalis in an Italian population, unlike other European populations (e.g. Northwest Spain, CórdobaAguilar 2000), show no precopulatory courtship to females presumably by the lack of territories to defend which impede courtship displays (Cordero Rivera & Andrés 2002). These factors that vary across populations may have a direct or indirect effect on immune ability as this seems very sensitive to environmental stress (reviewed by Schmid-Hempel 2005). For example, high temperature in temperate places favours high values of immune ability (Robb & Forbes 2005). Factors that have been correlated with changes in immunity in dragonflies appear listed in table 1. Similar to other insects (reviewed by Schmid-Hempel 2005), these factors strongly suggest that dragonflies are largely influenced by environmental variables affecting their immune ability. All these environmental factors are faced by dragonflies rather frequently which means Table 1. Factors related to immune ability in Odonata. Factor
Species
Reference
Age Food level
Mnais costalis Mnais costalis Ischnura verticalis Mnais pruinosa Lestes congener L. dryas L. forcipatus L. unguiculatus Coenagrion puella Coenagrion puella Lestes viridis
Hooper et al. 1999 Hooper et al. 1999; Tsubaki & Hooper 2004 Leung et al. 2001 Siva-Jothy et al. 1998 Yourth et al. 2001, 2002
Copulation Seasonality
Parasitism risk Predation risk Reduced time for development Year changes in Calopteryx parasite burden xanthostoma Temperature Lestes forcipatus Territorial defence Hetaerina americana
Joop & Rolff 2004 Joop & Rolff 2004 Rolff et al. 2004 Rolff & Siva-Jothy 2004 Robb & Forbes 2005 Contreras-Garduño et al. 2006
CyanMagentaYellowBlack Odonata page 272
272 Alex Córdoba-Aguilar & Jorge Contreras-Garduño
that immune variation is a common phenomenon in nature. Different environments which differ in these factors may be safely said to vary in quality as they will affect the dragonfly condition. Given that odonates frequently live in distinct environments and this also occurs in the same species, different populations are frequently facing distinct environmental regimes. For example, variation in food availability would be expected in populations that inhabit contrasting environments. In this case, some habitats may be more stressful, such as dry forests, compared to relatively high quality forests such as temperate or rain forests. Although no research has been done on how much food varies among habitats, it is likely that dragonflies may suffer from food shortage. The conditions that can lead to this are, for example, poor weather conditions or impoverished forests that inhibit development of prey used by dragonflies. Other elements related to environmental status are predation and parasitism risk, parasite burden, temperature, and fighting behaviour, depending on the availability of sites to defend (Table 1). These elements may vary from one habitat to another and as long as these are occupied by the same species, these elements may have an effect on immune ability and, hence, parasite defense. Related to this, some population differences in parasite burden have been found in the zygopteran Calopteryx maculata in which both sexes were more parasitized in a forest than in a pasture landscape (Taylor & Merriam 1996). The kind of parasite is a gregarine, a protozoan whose oocytes are ingested when the dragonfly preys upon small flies (Åbro 1976). Once ingested, the parasite attaches to the posterior gut where it grows using the food the host has consumed. Distinct pieces of evidence in several dragonfly species have found that the parasite has a negative impact on different fitness components such as survival (Córdoba-Aguilar 2002, Córdoba-Aguilar et al. 2003; CanalesLazcano et al. 2005), male mating success (Siva-Jothy 1999, 2000; CórdobaAguilar 2002), egg production (Córdoba-Aguilar et al. 2003; Canales-Lazcano et al. 2005), accumulation of fat reserves (Siva-Jothy & Plaistow 1999) and, possibly, flying performance (Marden & Cobb 2004). Very little is known about the ecology of the parasite except for the fact that once released through the odonate faeces, it gets attached to the fly’s legs (Åbro 1976). There are several interpretations for the parasite differences in C. maculata. Authors suggest that is the encounter rate between the dragonfly and the parasite that is different in both habitats being higher in the forest landscape (Taylor & Merriam 1996). This explanation assumes that gregarines are more common in the forest habitat perhaps because there are more flies that are carrying the gregarine or there are more gregarines per fly. A second explanation is that gregarines are similarly common in either habitat (or the encounter rate between dragonflies and parasites) but that the dragonfly immunity is more negatively affected in the forest because resources necessary for immune defense are in
CyanMagentaYellowBlack Odonata page 273
Differences in immune ability in forest habitats of varying quality: dragonflies as study models
273
shortage. One resource can be food although some others, as those outlined in Table 1, may also play a role. Whatever the reason, the effects of a fragmented habitat with these characteristics is that the host-parasite relationship is different from what occurs in the forest landscape. This clearly deserves further investigation particularly for studies on conservation (the implications of habitat type on population dynamics and status) and evolutionary ecology (the causes and evolutionary consequences of the difference in parasite burden).
SEXUAL SELECTION, IMMUNITY AND DRAGONFLIES The theory of sexual selection explains the origin and maintenance of ornaments that are typically present in males, where the only apparent advantage for their carrier is to attract potential reproductive mates and/or to avoid male competitors (Darwin 1871). Examples of these ornaments are the horns in ungulates and insects, the plumage color in birds or singing in crickets, among others. There is enormous variation in ornament expression (Andersson 1994) and sexual selection suggests that this implies a high cost in development and maintenance of ornaments, and that only those males in good condition can afford highly developed traits (Zahavi 1975; Iwasa & Pomiankowski 1999). Given the strong influence of parasites on their hosts, it has been proposed that a good way to assess male condition is via the resistance to pathogens (Hamilton & Zuk 1982). This idea predicts that males with more elaborate ornaments resist pathogens better than those with less elaborate ornaments (Folstad & Karter 1992) and allows the study of the effect of environmental conditions on ornament expression and reproductive ability (Iwasa & Pomiankowski 1999). Few traits have been admitted as ornaments in dragonflies. The only well documented case is that of wing pigmentation in Calopterygidae (CórdobaAguilar & Cordero Rivera 2005). Male calopterygids can be distinguished by the lack of a pterostigma while females show wing white spots called pseudoterostigmas. Male adults defend riverine habitats against conspecific males, where females arrive to for copulation. Pigmentation is expressed soon after emergence and is probably affected by the food consumed by the newly emerged adult (Hooper et al. 1999). Melanin is the base of this color, and its precursor is tyrosine (Riley 1997). An experiment that subjected males to different diets showed that the amount of food consumed during the teneral stage (when the pigmentation is formed) affects pigmentation degree (Hooper et al. 1999). Given this reason, it has been suggested that pigmentation in Calopterygidae is costly to produce and that only males in better condition will show this and defend a territory (Hooper et al. 1999). In addition, pigmentation is related to parasitism and immune response. In Calopteryx haemor-
CyanMagentaYellowBlack Odonata page 274
274 Alex Córdoba-Aguilar & Jorge Contreras-Garduño
rhoidalis it has been proposed that more pigmented males resist parasites better compared to less pigmented males (Córdoba-Aguilar 2002). Territorial males have more pigment, survive for longer, copulate more often and have fewer parasites compared to less pigmented males (Córdoba-Aguilar 2002). In this species, as in Mnais costalis (Hooper et al. 1999), it has been suggested that melanin is the link between wing pigmentation and immune response (Córdoba-Aguilar 2002 for other species see Siva-Jothy 2000). In agreement with this idea, more pigmented males showed greater phenoloxidase activity than less pigmented males, as has been demonstrated in Calopteryx splendens (Siva-Jothy 2002). Also, more pigmented males produce more melanin when confronted with an immune challenge compared to less pigmented males as has been observed in Calopteryx virgo (Koskimäki et al. 2004) and Hetaerina americana (Contreras-Garduño et al. 2006). Melanin and phenoloxidase are fundamental factors during immune response in insects. After an infection, phenoloxidase triggers pathogen melanization and promotes phagocytic activity by secreting enzymes that destroy the pathogen’s membrane (Cerenious & Söderhäll 2004). When phenoloxidase is secreted, haemocytes are activated and produce melanin (Riley 1997). Melanin is secreted around the pathogen and forms a capsule (Cerenious & Söderhäll 2004). Also, during melanin production and encapsulation, free radicals are produced such as thioles (-SH) or amino groups (-NH2) that help to have a more effective elimination of the pathogen (Riley 1997; Guillespie & Kanost 1997; Nappi & Ottaviani 2000). Since immunity and pigmentation are linked, immunity is also affected by diet which has been corroborated experimentally for the case of encapsulation (Hooper et al. 1999). Given the strong input of diet on the production of immune response and, therefore, pigmentation in dragonflies, one can speculate that if environments differ in quality, this may have an effect on immunity and the expression of pigmentation. This quality may be interpreted as a difference in either quality and/or quantity of food that odonates may use, which can cause physiological adaptations or restrictions to particular places (for a similar rationale see Tynkkynen et al. 2004). That there are population differences in immune response has been already documented (e.g. Calopteryx splendens, Siva-Jothy 2000; Rantala et al. 2000). In fact, our recent results in Hetaerina americana have provided grounds to the relation between forest quality and immunity (Córdoba-Aguilar & Contreras-Garduño unpub. data). In two populations inhabiting two forest types (a desert and a rainforest) which show differences in the number of prey available for damselflies, there are differences in wing pigmentation area, melanization response and fat reserves. As supposed, males in the population with more resources (the rainforest) had higher values of these parameters than males of the population with fewer resources (the desert). We expect that these differences will be common in other species as long as popu-
CyanMagentaYellowBlack Odonata page 275
Differences in immune ability in forest habitats of varying quality: dragonflies as study models
275
lations differ in the quality of habitats where they live. An alternative explanation for these differences is that there is a genetic component; one experiment to unravel both explanations would be to exchange recently emerged males from different populations and see whether pigmentation, fat reserves and immune response change once the animal has reached sexual maturity (having fed in each habitat during the pre-reproductive period). This experiment will not clarify which variable could explain the observed differences but certainly it will be the onset to look for environmental or genetic differences. If the environment is playing a role, one would expect that males coming from the poor environment will have equivalent pigmentation, immunity and fat reserves to the males of the rich environment. In summary, we postulate that pigmentation can be a trait that can be used as an indicator to assess habitat quality, as pigmentation has a strong link with immunity. This can be looked in those species that show this trait. This may be the case in calopterygids, a family whose members are well distributed in different habitats and in all continents except for Australia and New Zealand.
CONCLUSIONS We have reviewed some environmental factors that may cause differences in immune ability and pathogen resistance in dragonflies. These factors can be present in forest habitats of varying quality. The impact of these habitats can be reflected, therefore, on immune ability. One trait that can be used for assessing forest quality is male wing pigmentation in those species that show this trait. This can lead to a promising research line on how forest quality can affect dragonfly populations via physiological restrictions of immunity.
ACKNOWLE\DGEMENTS To one anonymous reviewer for very valuable comments and to Adolfo Cordero for his patience. To PAPIIT (IN-230603) and the Doctorado en Ciencias Biomédicas (both from the Universidad Nacional Autónoma de México) for financial support.
REFERENCES ÅBRO, A. 1976. The mode of gregarine infection in Zygoptera (Odonata). Zoologica Scripta 5: 265-275. ANDERSSON, M. 1994. Sexual Selection. Princeton University Press, Princeton.
CyanMagentaYellowBlack Odonata page 276
276 Alex Córdoba-Aguilar & Jorge Contreras-Garduño
CANALES-LAZCANO, J., CONTRERAS-GARDUÑO, J. & CÓRDOBA-AGUILAR, A. 2005. Fitness-related attributes and gregarine burden in a nonterritorial damselfly Enallagma praevarum Hagen (Zygoptera: Coenagrionidae). Odonatologica 34: 123-130. CERENIUS, L & SÖDERHÄLL, K. 2004. The prophenoloxidase-activating system in invertebrates. Immunological Reviews 198: 116-126. CONTRERAS-GARDUÑO, J., CANALES-LAZCANO, J. & CÓRDOBA-AGUILAR, A. 2006. Wing pigmentation, immune ability and fat reserves in males of the rubyspot damselfly, Hetaerina americana. Journal of Ethology 24: 165-173. CORBET, P. S. 1999. Dragonflies. Behavior and Ecology of Odonata. Harley Books, Colchester. CORDERO RIVERA, A. & ANDRÉS, J. A. 2002. Male coercion and convenience polyandry in a Calopterygid damselfly (Odonata). Journal of Insect Science 2: 14. Available online at http://www.insectscience.org/2.14. CÓRDOBA-AGUILAR, A. 2000. Reproductive behaviour in the territorial damselfly Calopteryx haemorrhoidalis asturica Ocharan (Zygoptera: Calopterygidae). Odonatologica 29: 297-307. CÓRDOBA-AGUILAR, A. 2002. Wing pigmentation in territorial male damselflies, Calopteryx haemorrhoidalis: a possible relation to sexual selection. Animal Behaviour 63: 759-766. CÓRDOBA-AGUILAR, A., SALAMANCA-OCAÑA, J. C. & LOPEZARAIZA, M. 2003. Female reproductive decisions and parasite burden in a calopterygid damselfly (Insecta: Odonata). Animal Behaviour 66: 81-87. CÓRDOBA-AGUILAR, A. & CORDERO RIVERA, A. 2005. Evolution and ecology of Calopterygidae (Zygoptera: Odonata): status of knowledge and research perspectives. Neotropical Entomology 34: 861-879. DARWIN, C. 1871. The Descent of Man and Selection in Relation to Sex. John Murray, London. FOLSTAD, I. & KARTER, A. J. 1992. Parasites, bright males and the immunocompetence handicap. American Naturalist 139: 604-622. GUILLESPIE, J. P. & KANOST, M. R. 1997. Biological mediators of insect immunity. Annual Review of Entomology 42: 611-643. HAMILTON, W. D. & ZUK, M. 1982. Heritable true fitness and bright birds: a role for parasites ? Science 218: 384-386. HOOPER, R. E., TSUBAKI, Y. & SIVA-JOTHY. M. T. 1999. Expression of a costly secondary sexual trait is correlated with age and condition in a damselfly with two male morphs. Physiological Entomology 24: 364-369. IWASA, Y. & POMIANKOWSKI, A. 1999. Good parent and good genes models of handicap evolution. Journal of Theoretical Biology 200: 97-109. JOOP, G. & ROLFF, J. 2004. Plasticity of immune function and condition under the risk of predation and parasitism. Evolutionary Ecology Research 6: 1015-1062. KOSKIMÄKI, J., RANTALA, M. J., TASKINEN, J., TYNKKYNEN, K. & SUHONEN, J. 2004. Immunocompetence and resource holding potential in the damselfly, Calopteryx virgo L. Behavioral Ecology 15: 169-173. LEUNG, B., FORBES, M.R. & BAKER, R. 2001. Nutritional stress and behavioural immunity of damselflies. Animal Behaviour 61: 1093-1099.
CyanMagentaYellowBlack Odonata page 277
Differences in immune ability in forest habitats of varying quality: dragonflies as study models
277
MARDEN, J. H. & COBB, J. R. 2004. Territorial and mating success of dragonflies that vary in muscle power output and presence of gregarine gut parasites. Animal Behaviour 68: 857-865. MUCKLOW, P. T. & EBERT, D. 2003. Physiology of immunity in the water flea Daphnia magna: environmental and genetic aspects of phenoloxidase activity. Physiological and Biochemical Zoology 76: 836-842. MUCKLOW, P. T., VIZOSO, D. B., JENSEN, K. H., REFARDT, D. & EBERT, D. 2004. Variation in phenoloxidase activity and its relation to parasite resistance within and between populations of Daphnia magna. Proceedings of the Royal Society of London, ser. B 271: 1175-1183. NAPPI, A. J. & OTTAVIANI, E. 2000. Cytotoxicity and cytotoxic molecules in invertebrates. BioEssays 22: 469-480. RANTALA, M. J., KOSKIMÄKI, J., TASKINEN, J., TYNKKYNEN, K. & SUHONEN, J. 2000. Immunocompetence, developmental stability and wingspot size in the damselfly Calopteryx splendens L. Proceedings of the Royal Society of London, ser. B 267: 2453-2457. RILEY, P. A. 1997. Molecules in focus. International Journal of Biochemistry and Cell Biology 29: 1235-1239. ROBB, T. & FORBES, M. R. 2005. On understanding seasonal increases in damselfly defense and resistance against ectoparasitic mites. Ecological Entomology 30: 334-341. ROLFF, J. 2002. Bateman’s principle and immunity. Proceedings of the Royal Society of London, ser. B 269: 867-872. ROLFF, J. & SIVA-JOTHY, M. T. 2003. Invertebrate ecological immunology. Science 301: 472-475. ROLFF, J. & SIVA-JOTHY, M. T. 2004. Selection on insect immunity in the wild. Proceedings of the Royal Society of London, ser. B 271: 2157-2160. ROLFF, J., VAN DE MEUTTER, F. & STOKS, R. 2004. Time constraints decouple age and size at maturity and physiological traits. American Naturalist 164: 559-565. SCHMID-HEMPEL, P. 2003. Variation in immune defense as a question in evolutionary ecology. Proceedings of the Royal Society of London, ser. B 270: 357-366. SCHMID-HEMPEL, P. 2005. Evolutionary ecology of insect immune defenses. Annual Review of Entomology 50: 529-551. SIVA-JOTHY, M. T. 1999. Male wing pigmentation may affect reproductive success via female choice in a calopterygid damselfly (Zygoptera). Behaviour 136: 1365-1377. SIVA-JOTHY, M. T. 2000. A mechanistic link between parasite resistance and expression of a sexually selected trait in a damselfly. Proceedings of the Royal Society of London, ser. B 267: 2523-2527. SIVA-JOTHY, M. T., TSUBAKI, Y. & HOOPER, R. E. 1998. Decreased immunocompetence as a proximate cost of copulation and oviposition in a damselfly. Physiological Entomology 23: 274-277. SIVA-JOTHY, M. T. & PLAISTOW, S. J. 1999. A fitness cost of eugregarine parasitism in a damselfly. Ecological Entomology 24: 465-470. TAYLOR, P. D. & MERRIAM, G. 1996. Habitat fragmentation and parasitism of a forest damselfly. Landscape Ecology 11: 181-189.
CyanMagentaYellowBlack Odonata page 278
278 Alex Córdoba-Aguilar & Jorge Contreras-Garduño
TSUBAKI, Y., HOOPER, R. E. & SIVA-JOTHY, M. T. 1997. Differences in adult and reproductive lifespan in the two male forms of Mnais pruinosa costalis (Seyls) (Odonata: Calopterygidae). Research in Population Ecology 39: 149-155. TSUBAKI, Y. & HOOPER, R. E. 2004. Effects of eugregarine parasites on adult longevity in the polymorphic damselfly Mnais costalis Selys. Ecological Entomology 29: 361-366. TUCKER, T. M. & STEVENS, L. 2003. Geographical variation and sexual dimorphism of phenoloxidase levels in Japanese beetles (Popillia japonica). Proceedings of the Royal Society of London, ser. B 270: S245-S247. TYNKKYNEN, K., RANTALA, M. J. & SUHONEN, J. 2004. Intersexual agression and caracter displacement in the damselfly Calopteryx splendens. Journal of Evolutionary Biology 17: 759-767. YOURTH, C. P., FORBES M. R. & SMITH B. P. 2001. On understanding variation in immune expression of the damselfles Lestes spp. Canadian Journal of Zoology 79: 815-821. YOURTH, C. P., FORBES, M. R. & SMITH, B. P. 2002. Immune expression in a damselfly is related to time of season, not fluctuating asymmetry or host size. Ecological Entomology 27: 123-128. ZAHAVI, A. 1975. Mate selection- a selection for a handicap. Journal of Theoretical Biology 53: 205-214. ZUK, M. & STOEHR, A. M. 2002. Immune defense and host life history. American Naturalist 160: S9-S22.
CyanMagentaYellowBlack Odonata page 279
The present role and future promise of conservation Adolfo Cordero Rivera (ed) 2006 genetics for forest Odonates Forests and Dragonflies. Fourth WDA International Symposium of Odonatology, Pontevedra (Spain), July 2005, pp. 279-299. © Pensoft Publishers
Sofia–Moscow
The present role and future promise of conservation genetics for forest Odonates Heike Hadrys
1,*
, Viola Clausnitzer
2
& Linn F. Groeneveld 1,3
1
*
ITZ, Ecology & Evolution, TiHo, D-30559 Hannover, Germany Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT, 06520-8104, USA 2 Department of Geography, Philipps-Universität Marburg, D-35032 Marburg, Germany 3 current address: DPZ, Behavioral Ecology and Sociobiology, D-37077 Göttingen, Germany
ABSTRACT Although the history of conservation genetics as a discipline dates back more than two centuries, odonates have only recently entered the scene. This is highly unfortunate since – especially in tropical forests – odonates may serve as prime examples for the application and potential of conservation genetic research. Faced with the same conservation problems as the forests themselves, they epitomize the difficulties of maintaining biodiversity in tropical forests. To date, no data exist on population structures, dynamics, viabilities or histories of afrotropical forest odonates. Below, a case study is introduced that demonstrates the application of population genetic research to three African damselfly species of the genus Pseudagrion. The three species selected represent a habitat gradient ranging from open habitats in Namibia to isolated mountain forests in Kenya and Tanzania. The results of mitochondrial (ND1) sequence analyses revealed strong inter- and intraspecific differences in the population structures of all three species, reflecting their habitat adaptations and demographic distribution. Mean genetic diversity and genetic isolation patterns increased with habitat specificities and restricted distributional range of the species. The two species with a wider distributional range, Pseudagrion massaicum, and P. kersteni displayed similar low genetic diversities in Namibia but showed considerable differences in population sub-structures between Namibian
279
CyanMagentaYellowBlack Odonata page 280
280
Heike Hadrys, Viola Clausnitzer & Linn F. Groeneveld
and East African populations. The third species, P. bicoerulans, an endemic of high-elevated mountain forests in Kenya and Tanzania, shows a multifold higher genetic diversity and complete genetic isolation between populations. The comparison with divergence values of true species suggests, that speciation in this species is well advanced. Given that the strong divergence patterns are neither correlated with geographic distance nor with the differences in morphological traits, the results provide a good example on how genetic data can provide information about conservation units and cryptic speciation processes. Future challenges in conservation genetic research for tropical forest species should focus on establishing as many genetic species profiles of current conditions as possible. Those data sets are valuable snapshots of the current conditions and may serve as calibration points for future conservation work.
OVERVIEW Present day patterns of biodiversity in ecosystems have arisen over time through a variety of natural and anthropogenic factors. Understanding these factors may allow predictions on the effects of future environmental changes (Soulé 1986, Spellerberg 1997). A basic requirement to understand biodiversity patterns and to predict conservation priorities is to study population structures, dynamics, viabilities and histories. Achieving the above in highly mobile organisms, such as dragonflies, is a difficult task. This is one of the reasons why despite the long and extensive history of odonate research, and despite their importance as indicators for evaluating freshwater ecosystems, studies of population structures and dynamics are still at an exploratory stage (e.g. Freeland et al. 2003, Watts et al. 2004, Hadrys et al. 2005, Giere & Hadrys 2006). Unfortunately, this aspect is especially true for species confined to tropical forest habitats. Afrotropical forests harbour a high number of rare and endemic dragonfly species, many of which exist in small, highly fragmented and isolated populations. In a checklist for East Africa Clausnitzer (2004) suggests 95 odonate species for inclusion into a global red list including all species that are endemic or strictly confined to the different types of forest areas. Studying the biodiversity of tropical forest species one finds the complete array of complex diversity patterns and evolutionary processes that lead to speciation or to extinction (Bouchier et al. 2000). In terms of evolutionary processes, tropical forests harbour recently radiated as well as evolutionary old species groups. In terms of diversity patterns, continuous deforestation leads to newly fragmented isolated populations versus old and naturally isolated mountain forest populations. For example, the once contiguous Eastern Arc and Coastal Forest belt of Kenya and Tanzania has been reduced to some
CyanMagentaYellowBlack Odonata page 281
The present role and future promise of conservation genetics for forest Odonates
281
250 small, isolated patches that differ in size, microclimate, light scheme, and other abiotic factors. The areas of the Kilimanjaro and Usambara Mountains show a history of intense human land use leading to the loss of the lowland forests at the base of these mountains (Burgess & Clarke 2000). The Eastern Arc forests belong to the 14 most threatened tropical forest diversity centres and represent – together with the Coastal Forests – the biodiversity hotspot with the highest number of endemic species on an exceptionally small area (Lovett 1998, Myers et al. 2000). In short, research on afrotropical forest odonates is almost forced to focus on detection and consequences of fragmentation, delineation of conservation units and detection of speciation processes. This focus leads straight into the field of conservation genetics where the study of population structures and dynamics via estimates of gene flow and genetic diversity become a sine qua non (Frankham et al. 2002, Avise 2004). Our comparative study on three damselfly species provides a first example on how genetic research can be applied to conservation questions in afrotropical odonates.
FROM OPEN HABITATS TO FORESTS: GENETIC POPULATION STRUCTURES AND HABITAT PATTERNS OF THREE AFRICAN PSEUDAGRION SPECIES The genus Pseudagrion is recognized as the largest modern genus of the Odonata. Pinhey (1964) lists 123 species worldwide. Fifty of these are included in an identification key by Clausnitzer & Dijkstra (in prep.) for East Africa (defined as the region of tropical Africa enclosed by 22 °N in the North, the northern borders of Namibia, Botswana and South Africa in the South, 22 °E in the West, and the Indian Ocean in the East). The same authors suggest a split of the African Pseudagrion complex into two major groups based on one morphological trait: “apex of abdominal segment ten with or without black denticles”. This split may not only reflect the morphological synapomorphy of the tenth segment, but also an ecological shift related to the preferred habitat types of the two taxonomic groups. Species without black denticles (group A) are taxa largely confined to different types of forest habitats, while those with black denticles (group B) mainly inhabit open areas such as savannah or bushland (Pinhey 1964). However, in contrast to the morphological split, there is no clear boundary between habitat adaptations of the two groups. In East Africa some species of group A also colonize more open habitats along rivers and streams with gallery forest, and thus might represent a link between the two groups. Recent phylogenetic analyses suggest that this subdivision is not only based on ecological or morphological traits but is supported by close genealogical relationships within both groups (Groeneveld & Hadrys, unpublished data).
CyanMagentaYellowBlack Odonata page 282
282
Heike Hadrys, Viola Clausnitzer & Linn F. Groeneveld
For the present study a typical group B species (open habitats, with black denticles; P. massaicum) and two group A species (forested habitats, no black denticles; P. kersteni and P. bicoerulans) were selected. The three species not only show a gradient in habitat specificity but also differ in their demographic distribution across Africa. While P. massaicum inhabits grassy or reedy pools and streams – rarely in swamps – with a distribution ranging from the Cape Province northwards to Somalia and Eritrea, westwards to Ghana, P. kersteni is abundant in most habitats, including open – not dense – forest areas and is distributed across continental Africa. Pseudagrion bicoerulans is the most specialized species with respect to habitat requirements and distributional range. It is an endemic species of mountain forest streams with a range restricted to Kenya, East Uganda, Tanzania and possibly Malawi (no confirmed record). In this species, different-coloured postocular spots have been observed in populations of different mountain regions (Clausnitzer & Dijkstra, in prep.). The study aims to provide first insights into population genetic structures of the three species and will discuss causal mechanisms underlying the observed patterns. A basic question to ask is whether population genetic data reflect differences in habitat specificities, morphological traits, and demographic distribution of the species.
MATERIAL AND METHODS Habitat monitoring A total of 214 localities were visited in Kenya, Tanzania, Ethiopia and Uganda in different seasons during 1999 to 2003. For the calculation of the habitat diagrams of a given species, all localities within the species specific altitudinal and geographic range have been sampled. The presence of the given species was then monitored for each locality with the total number of sites visited in a certain habitat category amounting to 100 percent. Two habitat categories were scored for each locality; one categorizes the type of water body (stream, river, pool, or lake) and the second the type of vegetation cover (forest, gallery/secondary forest, bush, or open habitat). Tissue sampling, DNA extraction and ND1 sequencing Tissue samples for genetic analyses were collected during October 2001 through September 2002. Full details of the sample sizes and localities are given in Table 1. The sampling method was non-destructive by removing a middle leg per specimen (Fincke & Hadrys 2001). For P. kersteni 40 individuals from eight localities and for P. massaicum 38 individuals from six local-
CyanMagentaYellowBlack Odonata page 283
The present role and future promise of conservation genetics for forest Odonates
283
Table 1. Locality, sample size (n), haplotypes found at the respective locality, country and geographical coordinates for three Pseudagrion species; a) P. kersteni (PK), b) P. bicoerulans (PB), c) P. massaicum (PM) a) Locality
n
Haplotypes*
Country
Latitude
Longitude
Naukluft Ongongo Baynes Mts. Pemba River Kiboko River Mzima Springs Rufiji Delta E. Usambara Mts.
3 3 6 5 5 5 5 5
PK3 PK4 PK4 PK10, PK11 PK1, PK5, PK10 PK2, PK6, PK8, PK10 PK1, PK5, PK6, PK10 PK6, PK7, PK9; PK10
Namibia Namibia Namibia Kenya Kenya Kenya Tanzania Tanzania
24° 19° 17° 04° 02° 02° 08° 05°
16° 13° 12° 39° 37° 36° 38° 38°
Locality
n
Haplotypes*
Country
Latitude
Longitude
Mt. Kenya Mt. Elgon Aberdare Mts. Kilimanjaro
5 5 5 5
PB1 PB2, PB3 PB4 PB7, PB8, PB9, PB10
Kenya Kenya Kenya Tanzania
00° 01° 00° 03°
37° 34° 36° 37°
n
Haplotypes*
Country
Latitude
Longitude
PM1 PM1 PM1 PM1 PM1, PM2, PM3, PM4 PM4, PM5
Namibia Namibia Namibia Namibia Kenya Kenya
24° 22° 19° 22° 04° 02°
16° 16° 13° 16° 39° 37°
15' 08' 01' 11' 15' 58' 15' 05'
43''S 25''S 00''S 55''S 00''S 51''S 00''S 43''S
14' 49' 39' 24' 32' 01' 37' 37'
05''E 00''E 00''E 59''E 00''E 15''E 00''E 43''E
b)
09' 02' 31' 10'
44''S 00''N 00''S 00''S
07' 46' 43' 13'
50''E 00''E 00''E 00''E
c) Locality
Tsauchab River 4 Van-Bach-Dam 5 Palmwag 6 Kuiseb River 4 Pemba River 7 Kiboko River 12
30' 00' 53' 40' 11' 15'
14''S 54''S 11''S 10''S 55''S 00''S
06' 57' 56' 37' 24' 32'
41''E 11''E 15''E 02''E 59''E 00''E
*For GenBank accession numbers please contact the first author
ities were sampled in Kenya, Tanzania and Namibia. Due to the restricted distributional range of P. bicoerulans specimens were sampled only at four localities in East Africa (Figure 1). Twenty-four individuals were sampled and the colour of postocular spots was determined for each individual. DNA extraction followed the protocol of Hadrys et al. (1992). A 610 bp long mitochondrial fragment, spanning partial 16S, tRNALeu and partial NADH dehydrogenase region 1 (ND1), was amplified using the primers described in Abraham et al. (2001). All reactions were carried out in a Gene Amp PCR System 9700 (Applied Biosystems) using the following reaction conditions: 25 µL reaction mix, containing 1X amplification buffer (20 mM tris-HCl, pH 8.4; 50 mM KCl), 2.5 mM MgCl2, 0.05 mM dNTPs, 0.5 pmol/µl each primer, and 0.03 U/µl Taq DNA polymerase (Invitrogen). PCR-profiles were as follows:
CyanMagentaYellowBlack Odonata page 284
284
Heike Hadrys, Viola Clausnitzer & Linn F. Groeneveld
Fig. 1. Map with sampling localities in Namibia, Kenya and Tanzania (from GIS data). Species occurrence at the respective locality is indicated by the abbreviation of each species’ name (P. kersteni – PK, P. massaicum – PM and P. bicoerulans – PB) in the legend. Colours correspond with Figure 4.
1 min initial denaturation at 95 °C, followed by 30 cycles of 94 °C 30 s, 48 °C 30 s, 72 °C 1 min and 6 min extension at 72 °C. All PCR products were purified with Microcon-PCR Centrifugal Filter Devices (Millipore) following the manufacturer’s instructions. Sequencing reactions were carried out using the ABI PRISM BigDye Terminator Cycle Sequencing Ready Reaction Kit and products were subsequently purified over Sephadex columns (Sigma). Sequencing was performed on an ABI PRISM 310 Genetic Analyzer (Applied Biosystems). Genetic data analyses Sequences were edited manually using SeqManII (DNASTAR, Inc.) and aligned with SeaView (Galtier et al. 1996). Different ND1 haplotypes were defined by eye. Pairwise genetic distances were calculated by using the Simple Matching Coefficient (uncorrected Hamming distance) as implemented in PAUP, vers. 4.0b 10 (Swofford 2002). A hierarchical analysis of molecular variance (AMOVA, implemented in Arlequin, vers. 2.000; Schneider et al. 2000) was conducted to detect genetic substructuring within and among populations via
CyanMagentaYellowBlack Odonata page 285
The present role and future promise of conservation genetics for forest Odonates
285
a refined Fst approach. AMOVA separates and tests tiers of genetic diversity at different levels, (i) diversity among groups of populations, (ii) diversity among the populations within groups, (iii) diversity among individuals within a populations. Hereby the Fst value is a measurement for partitioning of the genetic diversity among the tested groups (Hartl 2000, Frankham et al. 2002). A correlation between genetic and geographic distance was tested using a Mantel test, implemented in IBD, vers. 1.52 (Bohonak 2002). To estimate the genealogical relationships between the haplotypes and populations, mutational networks based on statistical parsimony were computed, using TCS, vers. 1.20 (Clement et al. 2000). This, in contrast to traditional methods of estimating phylogenetic relationships, had the advantage that recombination, low levels of genetic divergence and the presence of non-bifurcating genealogical information were incorporated in the output. All individual sequences generated for a species were collapsed into haplotypes by TCS and haplotype frequencies were also incorporated into the analyses. This way, an ancestral haplotype was calculated using the predictions of coalescent theory (Clement et al. 2000). The phylogenetic relationships among all haplotypes of the three species were assed by using the Maximum Likelihood (ML), Maximum Parsimony (MP) and Neighbour Joining (NJ) algorithms, implemented in PAUP (vers. 4.0b10, Swofford 2002).
RESULTS Habitat diagrams The habitat requirements of the three species are presented in the form of spiderwebs in Figure 2. P. massaicum was recorded at 19 of the 73 localities checked, P. kersteni was found at 35 of the 159 possible localities and P. bicoerulans was present at all 12 high-elevated localities checked. Regarding the habitat categories, P. massaicum was present in all types of freshwater habitats, in 34 % of the “stream” localities, in 29 % of all “river” localities, in 14 % of the “pool” and in 33 % of all “lake” localities visited. In terms of vegetation requirements, the species was not found in any of the 21 sampled “forest” localities, but was present in the other three vegetation categories: in 36 % of the “gallery/secondary forest” sites, in 29 % “bush” localities and in 44 % of the “open” localities (Figure 2a). In contrast to P. massaicum, P. kersteni was not recorded in any of the 54 “pool” and “lake” localities. The species was only present in 15 % of the “river” and in 39 % of the “stream” localities. In terms of vegetation cover, P. kersteni was not present in any of the 30 localities scored as “open”, but was found in 9 % of the “forest” sites, in 35 % of the “gallery/secondary forest” sites and in 30 % of the sampled “bush” localities (Figure 2b). P. bicoerulans, was only recorded at stream
CyanMagentaYellowBlack Odonata page 286
286
Heike Hadrys, Viola Clausnitzer & Linn F. Groeneveld
a) P. massaicum
b) P. karsteni
c) P. bicoerulans
Fig. 2. Habitat diagrams of P. kersteni , P. massaicum and P. bicoerulans representing two categories of scored habitat parameters; water bodies and vegetation cover (details of sampling method and localities scored/species are given in the text).
CyanMagentaYellowBlack Odonata page 287
The present role and future promise of conservation genetics for forest Odonates
287
localities and none of the 12 high elevated sites was categorized as “open” (Figure 2c). Haplotype distribution patterns and genetic distances The lowest number of variable nucleotide positions and haplotypes was detected in the ND1 fragment of P. massaicum. Only six variable nucleotide positions defining five haplotypes were found across six localities (Table 2b). All Namibian populations shared a single haplotype, which also was present Table 2. Variable nucleotide positions (vnp) for the 483 bp long mitochondrial ND1 fragment defining eleven haplotypes in a) P. kersteni (PK); b) five haplotypes in P. massaicum (PM); and c) ten haplotypes in P. bicoerulans (PB); here differently coloured haplotypes correlate with the four mountain forests and correspond to colours used in Figure 3c. a) v n p PK1 PK2 PK3 PK4 PK5 PK6 PK7 PK8 PK9 PK10 PK11
b) 0 1 1 2 2 2 3 3 3 3 3 4 4 4 4 4 4 5 0 4 5 2 3 4 6 7 1 3 5 5 6 8 7 5 1 9 8 7 3 4 0 8 3 5 7 TTACCTATGCTGTTG . . . . . C . . . . . . . . . A. . . . C . . . . . . . . . . . . . TC . C . TCA . CA . . . . . . . . . . . . C . . . . . . . .G. . . . . . . . . .G. . .G. . . . . . . . . . . . . . . . A. . . . . . . . . . T . . . A. . . . . . . C . . . . . . . . . . . . . . C . T . . . . . . . . . . .
v n p PM1 PM2 PM3 PM4 PM5
2 2 2 2 4 4 3 4 5 9 3 4 8 9 3 2 3 9 CTCGCT . . . . T . T . . . . . T . TATC TCTATC
c) v n p
0 0 0 0 0 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 0 0 5 6 8 1 1 4 6 7 8 8 9 9 0 0 1 3 5 8 9 1 5 7 7 7 7 8 9 0 1 3 3 3 4 4 5 5 5 7 3 7 0 0 4 4 5 6 9 0 1 2 6 9 2 5 1 8 9 3 5 3 5 3 6 7 9 6 8 6 5 3 4 6 2 4 1 4 7 2
PB1
T T T C C A TGC CGAGC A C A TGGT A C A T A T T T C CGTGT T AAG T
PB2
. . . . . . . . T . . . . . . . . . AA . . . . . . . . . . . . C . . . . . . .
PB3
. . . . . . . . T . . . . . . . . . AA . . . . . . . . . . . . C . . . G . . .
PB4
. . . . . . . . . . A . AT . TG . AAC . TGCGC . . TTC . A . . . . A .
PB5
. . . . . . . . . . A . ATG . G . AAC . TGCGC . . TTC . A . . . . A .
PB6
. . . . . . . . . . A . ATGTG . AAC . TGCGC . . TTT . A . . . . A .
PB7
. C . T T T AA T A TG . . . . GT A C CGT T CGC C C . T A . A C AGGA C
PB8
A C . T T T AA T A TG . . . . GT A C CGT T CGC C C . T A . A C AGGA C
PB9
. C . T T T AA T A TG . . . . GT A T CGT T CGC C C . T A . A C AGGA C
PB10
. C C T T T AA T A TG . . . . GT A T CGT T CGC C C . T A . A C AGGA C
CyanMagentaYellowBlack Odonata page 288
288
Heike Hadrys, Viola Clausnitzer & Linn F. Groeneveld
in one Kenyan population. For P. kersteni 15 variable nucleotide positions were detected, resulting in the definition of 11 haplotypes distributed across the eight localities (Table 2a). Interestingly, one haplotype (PK4), which was exclusively present in the Namibian Ongongo and Baynes Mountains populations, counted for the majority of variable nucleotide positions (n = 8). The highest number of variable nucleotide positions, defining ten haplotypes across four populations, was detected in P. bicoerulans. Here 40 nucleotide exchanges were distributed across the four East African mountain populations, displaying a very distinct pattern (Table 2c). Haplotype distributions and localities are listed in Table 1 and Figure 1. Another significant correlation between the habitat gradient (open to forest) and genetic patterns was seen when the overall genetic variability within the three species was compared. The lowest mean intraspecific genetic variability was found in P. massaicum (0.5 %) followed by a slightly higher value in P. kersteni (0.8 %). A multifold higher variability of 3.6 % was present in P. bicoerulans. When comparing the genetic distances between populations within each species the same patterns became even more obvious. In P. massaicum there were no genetic differences between the four Namibian populations (genetic distance = 0 %). Even the two Kenyan populations, Pemba and Kiboko River, displayed an average genetic distance of only 0.2 % and 1.1 %, respectively, relative to the Namibian populations (Table 3b). Genetic distances between the P. kersteni populations were slightly higher. They ranged from 0 % (Baynes Mts. / Ongongo, Namibia) to 1.9 % (Baynes Mts., Namibia / Pemba River, Kenya) (Table 3a). For P. bicoerulans the genetic distances among the different populations were higher (from 0.9 % between the Mt. Kenya / Mt. Elgon populations to 6.7 % separating the Mt. Kenya from the Kilimanjaro populations; Table 3c). Geographic patterns AMOVA analyses revealed strong sub-structuring between populations and between groups of populations in different demographic areas in all three species. For P. massaicum the Namibian and East African populations appeared to be genetically differentiated from each other with restricted gene flow (Fst = 0.87; p = 0 when grouping Namibian populations against East African populations). The same result is obvious in P. kersteni (Fst = 0.86; p = 0 when grouping the Namibian populations against the East African populations). In contrast to the former species P. kersteni also display strong sub-structuring within Namibia. In both species the Mantel test confirmed that population sub-structuring increased with geographical distance, P. kersteni (r = 0.4261, one-sided, p = 0.01) and P. massaicum (r = 0.5476, onesided, p = 0.048).
CyanMagentaYellowBlack Odonata page 289
The present role and future promise of conservation genetics for forest Odonates
289
Table 3. Average genetic distances based on the ND1 fragment (483 bp) among the populations of a) P. kersteni (PK), b) P. massaicum (PM), and c) P. bicoerulans (PB) a)
Naukluft Ongongo Baynes Mts. Pemba River Kiboko River Mzima Springs Rufiji Delta E. Usambara Mts.
Naukluft Ongongo Baynes Pemba Mts. Riv.
Kiboko Mzima Riv. Spr.
Rufiji Delta
E. Usam. Mts.
0.0000 0.0166 0.0166 0.0066 0.0058 0.0054 0.0058 0.0070
0.0021 0.0032 0.0022 0.0036
0.0037 0.0031 0.0038
0.0029 0.0036
0.0050
Pemba Riv.
Kiboko Riv.
0.0037 0.0088
0.0008
0.0000 0.0000 0.0190 0.0182 0.0178 0.0182 0.0186
0.0000 0.0190 0.0182 0.0178 0.0182 0.0186
0.0008 0.0017 0.0037 0.0025 0.0037
b) Tsauch. V.-B. -Dam Tsauchab Van-Bach-Dam Palmwag Kuiseb Pemba River Kiboko River
Palmwag Kuiseb
0.0000 0.0000 0.0000 0.0000 0.0021 0.0109
0.0000 0.0000 0.0000 0.0021 0.0109
0.0000 0.0000 0.0021 0.0109
0.0000 0.0021 0.0109
Mt. Kenya
Abe. Mts.
Mt. Elgon
Kilimanjaro
0.0000 0.0376 0.0086 0.0671
0.0030 0.0380 0.0550
0.0007 0.0647
0.0029
c)
Mt. Kenya Aberdare Mts. Mt. Elgon Kilimanjaro
The four populations of P. bicoerulans appeared to be genetically isolated. Here two hierarchical models were tested which result in identical Fstvalues. Model one tested the geographically close Aberdare Mts. and Mt. Kenya against each remaining population (Fst = 0.96; p = 0), while model two tested the geographically more separated, but genetically closer (see genetic distances, Table 2c), Mt. Elgon and Mt. Kenyan populations against each remaining population (Fst = 0.966; p = 0). Although the population substructuring in P. bicoerulans was found to be highly distinct it was, in contrast to the two other species, not correlated with geographical distance (r = 0.4555, one-sided p = 0.3290; Mantel test).
CyanMagentaYellowBlack Odonata page 290
290
Heike Hadrys, Viola Clausnitzer & Linn F. Groeneveld
Genealogical patterns Information about intraspecific genealogical relationships between populations was revealed by mutational network analyses. The most parsimonious network for P. massaicum depicts the ancestral haplotype (PM1) in all populations except in one of the two Kenyan populations (Kiboko River). It was the only haplotype found in the four Namibian populations and was also present in one of the Kenyan population (Shimba Hills, Figure 3a). While the Shimba Hills population represented the genealogical link between Namibia and the second Kenyan population, it also displayed the highest number of haplotypes found of all populations analysed (four out of five, Figures 1 and 3a). The network for P. kersteni displayed two distinct features: (i) a strong split between North and South Namibian populations, and (ii) a close relationship (many shared haplotypes) between all Eastern Arc and Coastal Forest populations in Kenya and Tanzania (Figures 1 and 3b). North Namibian populations from Baynes Mts. and Ongongo shared a single and exclusive haplotype (PK4), which was well separated (seven mutational steps) from its closest related haplotype in the southern Namibian Naukluft population. The five East African populations shared the ten remaining haplotypes which were all closely related (Figure 3b). For P. bicoerulans TCS analyses resulted in three separate networks (Figure 3c): (i) the populations from Mt. Kenya and Mt. Elgon; (ii) the Aberdare Mts. population; (iii) the Kilimanjaro population. Surprisingly, the geographically close populations from the Aberdare Mts. and Mt. Kenya were separated by 16 mutational steps (ms), while the geographically more distant Mt. Kenya and Mt. Elgon populations displayed a much closer genealogical relationship (3 ms). The Kilimanjaro population was by far the most separated of all populations. It was split by 30 ms from the Mt. Elgon / Mt. Kenyan populations and by 26 ms from the Aberdare Mts. population. Phylogenetic patterns Phylogenetic analyses performed with MP, ML and NJ algorithms all showed congruent tree topologies with similarly high bootstrap values (e.g. MP tree in Figure 4). The species specific haplotypes grouped into three distinct clades with a basal position for P. massaicum relative to the two sister clades P. kersteni and P. bicoerulans (bootstrap support of 77 % and 100 %, respectively, Figure 4). This preliminary result mirrored the morphological and ecological split between forest and open area species and has recently been verified by a more detailed phylogenetic analyses so far including 16 Pseudagrion species (Groeneveld & Hadrys, data not shown). No phylogenetic separation could be detected between populations of P. massaicum. In P. kersteni, the North Namibian haplotype PK4 appeared in a sister group relationship to all remaining haplotypes. In contrast, the haplo-
CyanMagentaYellowBlack Odonata page 291
The present role and future promise of conservation genetics for forest Odonates
291
Fig. 3. Mutational haplotype network based on statistical parsimony. Haplotypes considered to be ancestral are depicted as rectangles, all other haplotypes as circles. Missing mutational steps connecting haplotypes are represented by small non-coloured circles. Haplotypes connected by a single line differ in one mutational step. The size of the rectangle and circles correlates with haplotype frequency within each network. a) Genealogical relationships between five haplotypes in six populations of P. massaicum; b) between eleven ND1 haplotypes in eight populations of P. kersteni; c) between ten haplotypes in four populations of P. bicoerulans. Note, that the split into three separate networks, under the confidence limit of 95 %, displays the high number of mutational steps between haplotypes of different localities. The connections between sub-networks are indicated by dashed lines, with the number of mutational steps separating the networks in bold.
types of P. bicoerulans grouped as three distinct sister clades confirmed by high bootstrap support (Figure 4). The first clade consisted of the haplotypes from the Mt. Kenya and Mt. Elgon populations. The second clade included the three haplotypes found in the Aberdare Mts. population, which formed a sister clade to the haplotypes found on Mt. Kilimanjaro.
CyanMagentaYellowBlack Odonata page 292
292
Heike Hadrys, Viola Clausnitzer & Linn F. Groeneveld
Fig. 4. Maximum Parsimony tree inferred from ND1 sequence data of the 26 haplotypes detected for P. massaicum (PM, yellow box), P. kersteni (PK, pink box) and P. bicoerulans (PB, blue box) using the heuristic search option. Characters (483 total, including 109 parsimony informative) were unweighted and unordered. Gaps were treated as missing data. Starting trees were obtained using random stepwise addition with 20 replicates; branchswapping algorithm: tree-bisection-reconnection (TBR). Indicated bootstrap values are based on 500 replicates. Note, that no root was defined for the purpose of this study.
DISCUSSION The data presented here represent the first published detailed genetic information on damselfly population structures in Africa. The comparison of (i) three species with different distributional ranges across (ii) three different geographical regions of high conservation value proved to be highly informative: Region I, Namibia, one of the most arid countries; Region II, the Eastern Arc and Coastal Forests of Kenya and Tanzania, one of the most anthropo-
CyanMagentaYellowBlack Odonata page 293
The present role and future promise of conservation genetics for forest Odonates
293
genically disturbed forest regions; and Region III, the mountain forests in Kenya and Tanzania, four - over evolutionary time scales naturally isolated and high elevated - mountains of volcanic origin (Burgess & Clarke 2000). (I) Namibia: Pseudagrion massaicum and Pseudagrion kersteni Of the three species only P. massaicum and P. kersteni are found in Namibia. Both species show similar low level of genetic diversity compared to populations in the Eastern Arc and Coastal Forests of Kenya and Tanzania. It seems obvious that both P. massaicum and P. kersteni are close to their natural distribution boundary in terms of climate in the Namibian region. P. kersteni as a species confined to permanent water bodies, stream and river habitats with bushy or forested bank vegetation, has been found exclusively at the three sites described in this study. Gene flow among the few and geographically well separated populations is presumably more restricted than in P. massaicum. This is illustrated by the genetic isolation between the North and South Namibian populations. With no preferred type of freshwater habitat and a preference for more open sites with no dense vegetation P. massaicum seems to adapt better to arid regions. There are more examples of desert-like areas where P. massaicum is recorded, while P. kersteni is absent (e.g. in Northern Kenya, Clausnitzer, pers. observation). The two species belong to a genus that radiated extensively, but no species adapted to the desert conditions. Only 13 Pseudagrion species are found in Namibia, in contrast to approximately 50 species recorded for Eastern Africa (Clausnitzer & Dijkstra, in prep.). Although Namibia, and with respect to insects particularly central Namibia, is known for high rates of endemism no endemic desert odonates are known (Barnard et al. 1998). The radiation potential for the genus Pseudagrion to evolve in response to environmental change seems to be low in arid regions compared to the afrotropical forest areas of East Africa. Both Pseudagrion species may exist only in small populations displaying low genetic variability, possibly caused by inbreeding depression and/or genetic drift. Both mechanisms lead to loss of genetic variability and reduced population viability (Ralls et al. 1988, Waples 1989, Davis 1992, Moritz 1994, Losos & Glor 2003). It should be noted that these effects are size-dependent and become more obvious when mtDNA sequences are studied (Frankham et al. 2002). Although P. massaicum is more widely distributed, no genetic variability was detected in Namibia, suggesting again possible inbreeding depression due to small effective population sizes. The species might be even more susceptible to local bottlenecks or even extinction than P. kersteni. A natural cause for bottlenecks in an arid country like Namibia is drought, which can result in the drying out of normally permanent water bodies (Jacobson et al. 1995). Barnard et al. (1998) state: “Probably the most threatened (endemic)
CyanMagentaYellowBlack Odonata page 294
294
Heike Hadrys, Viola Clausnitzer & Linn F. Groeneveld
insect taxa in Namibia are those confined to aquatic and riverine vegetation”. This is especially true for damselfly species with their endophytic oviposition and – compared to dragonflies – generally lower dispersal potential, one of the most important parameters in arid and/or fragmented regions (e.g. Conrad et al. 1999, Watts et al. 2004). The above scenario can be applied to P. kersteni as well, with the exception that the Baynes Mts. and Ongongo populations in North Namibia must have been isolated from the remaining populations over a longer time scale. These populations show similarly high genetic distances when compared to the South Namibian and East African populations (1.7 % to 1.9 %, respectively). Studies in other odonate species, e.g. in the two anisopteran species, Trithemis arteriosa and Anax speratus, have also shown that the Baynes Mountains area is genetically rather distant from other sampling sites (Giere & Hadrys, unpublished data). Further studies have to prove its possible importance as a separate conservation unit. (II) Eastern Arc and Coastal Forest region of Kenya and Tanzania: Pseudagrion massaicum and Pseudagrion kersteni In contrast to the Namibian populations, the populations in East Africa display a much higher level of genetic variability. In P. massaicum the three East African populations count for the majority of haplotypes. In P. kersteni the narrow genealogical haplotype network reflects close genealogy and gene flow between all populations. Obviously, P. kersteni as an open forest species is less affected by deforestation than species confined to dense forest areas. The dispersal potential of P. kersteni seems to be quite high, with many shared haplotypes between all sampling sites, especially between the southernmost Rufiji Delta and the remaining populations. Interestingly, the Rufiji Delta (Tanzania) with its catchment and forests may constitute a separate biota, especially for species strictly confined to forests. The Rufiji Delta population of the tree hole breeding giant damselfly Coryphagrion grandis, for example, shows progressed speciation at this site. Coryphagrion grandis is (in contrast to P. kersteni) a species strongly confined to forest areas with continuous canopy cover and therefore highly affected by deforestation (Clausnitzer & Lindeboom 2002, Groeneveld et al., in press). Forest odonates, such as C. grandis, may serve as positive indicators for different types of natural forests, whereas species of more open areas, which colonize forest habitats only after degradation, may serve as negative indicators for different degrees of forest disturbance (Clausnitzer 2003, Hadrys et al. 2005, Fincke, herein). Under this scheme, the high dispersal of P. kersteni in the study region could be seen as the result of ongoing destruction of the once contiguous forest areas.
CyanMagentaYellowBlack Odonata page 295
The present role and future promise of conservation genetics for forest Odonates
295
(III) The mountain forests in Kenya and Tanzania: Pseudagrion bicoerulans The observed genetic picture for the forest species P. bicoerulans is the result of a particular situation, which is not comparable with the above scenarios. P. bicoerulans is not found in Namibia nor in the Eastern Arc and Coastal Forest region. The species’ range is restricted to high-elevated forest streams and has its endemic distribution across isolated mountains in Kenya, Tanzania, East Uganda, and Malawi (no confirmed record). Genetic analyses suggested complete genetic isolation between populations in three of the four mountain forests. The natural isolation of the mountain forests resulted in radiation of P. bicoerulans in at least three significant units of conservation. The genetic distances between the populations are about as high (3.8 % to 6.7 % except for the Mt. Kenya/Mt. Elgon sites; 0.9 %) as those between the 16 “true” Pseudagrion species (3.6 % to 10.6 %; Groeneveld & Hadrys, unpublished data). Given this calibration range, the distance values suggest widely progressed speciation and call for scrutinizing the presence of cryptic species. Several interesting questions relate to the observation that despite wide geographical distance, the Mt. Kenya and the Mt. Elgon populations are genetically the closest related populations. What are the mechanisms that led to the constellation of two geographically close, but genetically very distant populations, while vice versa two geographically isolated populations are genetically more closely related? Interestingly, the geographically close populations, Mt. Kenya and the Aberdares, share one postocular spot colour while the other two populations each have their own specific colour. Three different postocular spot colours are found: green on Kilimanjaro, blue on Mt. Elgon and orange on Mt. Kenya and the Aberdares. Based on this morphological trait, the Mt. Kenya and the Aberdare Mts. populations group together, which correlates with their close geographic distance. Genetically, however, the Mt. Kenya and the Mt. Elgon populations show a much higher degree of similarity than the Mt. Kenya and the Aberdares populations. The opposite had been expected not only because of the shared postocular spot colour but also because of the fact that many shared Mt. Kenya and Aberdare endemic species are known. Three hypotheses are conceivable based on these observations: (i) the Mt. Kenya and the Mt. Elgon populations have changed at different rates than the Aberdares and the Kilimanjaro populations, (ii) time since isolation differs between the populations or (iii) an ecological rather than genetically fixed adaptation has resulted in allopatric polychromatism. The first hypothesis seems unlikely under the scheme of a uniform rate of evolution for the ND1 region, which has been suggested for insects (Pashley & Ke 1992). However, factors like small population size, inbreeding depression and genetic drift result in slower radiation and may explain the low genetic variability within the Mt. Kenya and
CyanMagentaYellowBlack Odonata page 296
296
Heike Hadrys, Viola Clausnitzer & Linn F. Groeneveld
Aberdare Mts. populations. The second hypothesis is conceivable, but the phylogeographic history in East Africa provides no obvious clue. The third hypothesis – i.e. the distribution of the postocular spot colour is the result of an ecological adaptation – assumes that the trait evolved after the lineages had split. For odonates it is known that allopatric polychromatism can correlate with region and habitat (Corbet 1999). This hypothesis would be convincing if the habitats on Mt. Kenya and the Aberdares displayed the same environmental factors leading to the morphological differences. In summary, genetic analyses in P. bicoerulans detected two unexpected cryptic patterns: (i) genetic distance does not reflect morphological and geographical patterns; (ii) radiation into three endemic species is well progressed, which has important consequences for conservation management.
CONCLUSION: CHALLENGES OF GENETIC TECHNOLOGIES IN CONSERVATION OF ODONATES William Conway, a senior conservationist from the Wildlife Conservation Society of the US, forecasts, “if deforestation remains at the level from 1979 through 1989, the last tropical forest tree will fall in 2045, but - the rate is increasing”. Thus odonate species confined to tropical forests may already be on the road to extinction. To determine an endangered status of a species requires the combination of knowledge in all fields of conservation biology. Hereby the role of genetic factors is not just reflective, indeed in a landmark paper Spielman et al. (2004) already have shown that they directly impact most species on the road to extinction (Brook et al. 2002, DeSalle & Amato 2004). Above we presented a first effort to determine population genetic patterns in damselflies from regions of high conservation interest. The comparison of three Pseudagrion species revealed strong inter- and intraspecific differences in population genetic patterns. The observed patterns (i) reflected ecological, morphological and distributional patterns, (ii) provided information on the consequences of different types of habitat fragmentation, and (iii) helped to detect conservation unit boundaries and cryptic speciation processes (Davis 1992, Moritz 1994, Losos & Glor 2003). An increase of relevant population genetic studies in insects is needed to further support that genetic patterns can be relevant to and should be included into landscape-based approaches to conservation decision-making (Moritz et al. 2000, Neel & Cummings 2003). In the future an important role of conservation genetic research will be to establish as many population genetic and systematic profiles for as many forest odonate species as possible (Baker et al. 2003, Moritz et al. 2000, Neel et al. 2003). These data sets are snapshots of the current conditions and will serve as calibration points for future conservation work (Haila et al. 1997,
CyanMagentaYellowBlack Odonata page 297
The present role and future promise of conservation genetics for forest Odonates
297
Morin 2000). In addition recent efforts to form centralized DNA barcode databases (unique species identification keys) and DNA registries may be crucial in establishing fast and long-term monitoring systems (Stoeckle 2003, Tautz 2003, Rach et al. 2006).
ACKNOWLEDGEMENTS We are grateful to Bernd Schierwater, Ola Fincke and Dave Thompson for helpful comments. We thank Dietmar Zinner for help with GIS and Frank Suhling for sample collection in Namibia. This work was supported by the BMBF BIOLOG program BIOTA South and BIOTA East to HH and VC.
REFERENCES ABRAHAM, D., RYRHOLM, N., WITTZELL, H. & HOLLOWAY, J.D. 2001. Molecular phylogeny of the subfamilies in Geometridae (Geometroidea: Lepidoptera). Molecular Phylogenetics and Evolution 20: 65-77. AVISE, J.C. 2004. Molecular Markers, Natural History and Evolution. Sunderland Sinauer Associates, MA. BAKER, S., DALEBOUT, M.L., LAVERY, S. & ROSS, H.A. 2003. www.DNA-surveillance: applied molecular taxonomy for species conservation and discovery. Trends in Ecology and Evolution 18: 271-272. BARNARD, B., BETHUNE, S. & KOLBERG, H. 1998. Biological Diversity in Namibia. In: Barnard, P. (ed.), Biodiversity of terrestrial and freshwater habitats, Namibian National Biodiversity Task Force. Windhoek. BOHONAK, A.J. 2002. IBD (Isolation By Distance): A program for analyses of isolation by distance. Journal of Heredity 93: 154-155. BOUCHIER, C., BOYLE. T. & YOUNG, A. 2000. Forest Conservation Genetics. CSIRO, Australia. BROOK, B.W., TONKYN, D.W., O’GRADY, J.J. & FRANKHAM, R. 2002. Contribution of inbreeding to extinction risk in threatened species. Conservation Ecology 6: 6-28. BURGESS, N.D. & CLARKE, G.P. 2000. Coastal Forests of Eastern Africa, IUCN, Gland & Cambridge. CLAUSNITZER, V. & LINDEBOOM, M. 2002. Natural history and description of the dendrolimnetic larvae of Coryphagrion grandis (Odonata). International Journal of Odonatology 5: 29-44. CLAUSNITZER, V. 2003. Dragonfly communities in coastal habitats of Kenya: indication of biotope quality and the need of conservation measures. Biodiversity and Conservation 12: 333-356. CLAUSNITZER, V. 2004. Critical species of Odonata in eastern Africa. International Journal of Odonatology 7: 189-206.
CyanMagentaYellowBlack Odonata page 298
298
Heike Hadrys, Viola Clausnitzer & Linn F. Groeneveld
CLEMENT, M., POSADA, D. & CRANDALL, K.A. 2000. TCS: a computer program to estimate gene genealogies. Molecular Ecology 9: 1657-1659. CONRAD, K.F., WILLSON, K.H., HARVEY, I.F., THOMAS, C.J. & SHERRATT, T.N. 1999. Dispersal characteristics of seven odonate species in an agricultural landscape. Ecography 22: 524-531. CORBET, P.S. 1999. Dragonflies: Behaviour and Ecology of Odonata. Harley Books, Colchester. DAVIS, J.I. & NIXON, K.C. 1992. Populations, genetic variation, and the delimitation of phylogenetic species. Journal of Systematic Biology 41: 421-435. DESALLE, R. & AMATO, G. 2004. The Expansion of Conservation Genetics. Nature Reviews Genetics 5: 702–722. FINCKE, O.M. & HADRYS, H. 2001. Unpredictable offspring survivorship in the damselfly, Megaloprepus coerulatus, shapes parental behavior, constrains sexual selection and challenges traditional fitness estimates. Evolution 55: 762-772. FRANKHAM, R., BALLOU J.D. & BRISCOE D.A. 2002. Introduction to Conservation Genetics, Cambridge University Press. Cambridge, U.K. FREELAND, J. MAY, M.L., LODGE, R. & CONRAD, K.F. 2003. Genetic diversity and widespread haplotypes in a migratory dragonfly, the common green darner (Anax junius). Ecological Entomology 28: 413-421. GALTIER, N., GOUY, M. & GAUTIER, C. 1996. SeaView and Phylo_win, two graphic tools for sequence alignment and molecular phylogeny. Computer Applications in the BioSciences 12: 543-548. GIERE, S. & HADRYS, H. 2006. Polymorphic microsatellite loci to study population dynamics in a dragonfly, the libellulid Trithemis arteriosa (Burmeister, 1839) Molecular Ecology Notes, in press. GROENEVELD, L.F., CLAUSNITZER, V. & HADRYS, H. 2006. Convergent evolution of gigantism in Damselflies of Africa and South America: Evidence from nuclear and mitochondrial sequence data. Molecular Phylogenetics and Evolution, in press. HADRYS, H., BALICK, M. & SCHIERWATER, B. 1992. Applications of random amplified polymorphic DNA (RAPD) in molecular ecology. Molecular Ecology 2: 79-87. HADRYS, H., SCHROTH, W., STREIT, B., SCHIERWATER, B. & FINCKE, O.M. 2005. Noninvasive isolation of polymorphic microsatellites from the neotropical damselfly Megaloprepus caerulatus: Use of tree hole odonates as environmental monitors in fragmented forests. Conservation Genetics 6: 481-483. HAILA, Y., COMER, P.J. & HUNTER, M. 1997. A ‘natural’ benchmark for ecosystem function. Conservation Biology 11: 300-305. HARTL, D.L. 2000. A Primer of Population Genetics. Sinauer Associates, Inc., 3rd edition. Sunderland, MA. JACOBSON, P.J., JACOBSON, P.M. & SEELY, M.K. 1995. Ephemeral Rivers and their Catchments: Sustaining People and Development in Western Namibia. Desert Research Foundation of Namibia, Windhoek. LOSOS, J.B. & GLOR, R.E. 2003. Phylogenetic comparative methods and the geography of speciation. Trends in Ecology and Evolution 18: 220-227. LOVETT, J.C. 1998. Eastern Tropical African Centre of Endemism: a candidate for World Heritage status? Journal of East African Natural History 87: 359-366.
CyanMagentaYellowBlack Odonata page 299
The present role and future promise of conservation genetics for forest Odonates
299
MYERS, N., MITTERMEIER, R.A., MITTERMEIER, C.G., FONSECA, G.A.B. & KENT, J. 2000. Biodiversity hotspots for conservation priorities. Nature 405: 853-858. MORIN, P.A. 2000. Genetic resources: opportunities and perspectives for the new century. Conservation Genetics 1: 271-275. MORITZ, C. 1994. Defining ‘evolutionarily significant units’ for conservation. Trends in Ecology and Evolution 9: 373-375. MORITZ, C., PATTON, J.L., SCHNEIDER, C.J. & SMITH, T.B. 2000. Diversification of rainforest faunas: an integrated molecular approach. Annual Review of Ecology and Systematics 31: 533-563. NEEL, M. & CUMMINGS, M.P. 2003. Genetic consequences of ecological reserve design guidelines: an empirical investigation. Conservation Genetics 4: 427-439. PASHLEY, D.P. & KE, L.D. 1992. Sequence evolution in mitochondrial ribosomal and ND1-genes in Lepidoptera: Implications for phylogenetic analyses. Molecular Biology and Evolution 9: 1061-1075. PINHEY, E. 1964. A Revision of the African Members of the Genus Pseudagrion Selys (Odonata). Revista de Entomologia de Mozambique 7: 5-196. RACH, J., SAKAR, I.N., DESALLE, R., SCHIERWATER, B. & HADRYS, H. 2006. Character-based DNA barcoding for identifying conservation units in Odonata. The DNA Barcode Data Analysis Initiative (DBDAI):(Developing Tools for a New Generation of Biodiversity Data. National Museum of Natural History, Paris. RALLS, K., BALLOU, J.D. & TEMPLETON, A. 1988. Estimates of lethal equivalents and the cost of inbreeding in mammals. Conservation Biology 2: 40-56. SCHNEIDER, S., ROESSLI, D. & EXCOFFIER, L. 2000. Arlequin: A software for population genetics data analysis. Vers. 2.000, Genetics and Biometry Laboratory, Dept. of Anthropology, University of Geneva. SOULÉ, M.E. 1986. Conservation biology: the science of scarcity and diversity. Sinauer Associates, Sunderland, MA. SPELLERBERG, I. 1997. Conservation Biology. Longman Group, Ltd, Harlow. SPIELMAN, D., BROOK, B.W. & FRANKHAM, R. 2004. Most species are not driven to extinction before genetic factors impact them. Proceedings of the National Academy of Sciences 101: 15261-15264. STOECKLE, M. 2003. Taxonomy, DNA, and the barcode of life. BioScience 53: 2-3. SWOFFORD, D.L. 2002. PAUP* Phylogenetic Analysis Using Parsimony (*and other Methods). Vers. 4, Sinauer Associates, Sunderland, MA. TAUTZ, D., ARCTANDER, P., MINELLI, A., THOMAS, R.H. & VOGLER, A.P. 2003. A plea for DNA taxonomy. Trends in Ecology and Evolution 18: 70-74. WAPLES, R.S. 1989. A generalized approach for estimating effective population size from temporal changes in allele frequency. Genetics 121: 379-391. WATTS, P.C., ROUQUETTE, J.R., SACCHERI, I.J., KEMP, S.J. & THOMPSON, D.J. 2004. Molecular and ecological evidence for small-scale isolation by distance in an endangered damselfly, Coenagrion mercuriale. Molecular Ecology 13: 2931-2945.
CyanMagentaYellowBlack Odonata page 300