Pesticide Risk Assessment in Rice Paddies: Theory and Practice

Preface In December 2002, the General Assembly of the United Nations declared 2004 as the International Year of Rice. This reflects the importance of rice as a primary food source for over 3 billion people, most of whom are living in Asia . In addition, this declaration stresses the environmental importance of the complex ricebased production systems that are directly linked to issues of global concern including food safety, poverty alleviation, preservation of cultural heritage and sustainable development . In the immediate future, the main challenges associated with rice production include the coverage of the increasing demand for rice from Asian population, the limited possibility for expanding rice-cultivated area, declining of rice yield and low returns from rice production . While these problems may be partially or fully resolved using modern agricultural technologies and marketing strategies, some socio-economic aspects related to rice quality traits, health and environmental concerns may have an important impact on rice development. These concerns have arisen by the potential negative effects of pesticides, extensively used in rice production, on environment and food quality. Pesticides are considered an integral part of modern agriculture ; more than 2 million tons of pesticides derived from 900 active ingredients are used annually worldwide . Pesticide contamination may entail a risk for the quality of different environmental compartments including soil, water and air and may adversely affect the health of non-target organisms . The extensive application of pesticides in rice cropping systems in combination with inappropriate agricultural practices may result in the contamination of environmental resources thus entailing a risk for human health (operator, consumer and residential population) . In the last few years, a comprehensive set of guidelines and tools have been developed in USA and Europe on how pesticide risk assessment should be performed for regulatory purposes. However, these are not applicable to rice cropping systems due to its particular cultivation conditions (submersion) . Individual research efforts during the last few years by experts in the area of ecotoxicology, pesticide environmental fate and agronomists working with rice have created a robust platform of knowledge for the development of pesticide risk assessment procedures and tools in rice. This book has gathered this dispersed knowledge on pesticide risk assessment in rice and aims to fulfil the gap of limited knowledge in the topic of pesticide risk assessment in rice ecosystems . The first chapter introduces the agronomy of rice crop. The second and third chapters focus on pesticide regulatory issues with particular reference to rice . The fourth chapter is offering to the reader an up-todate picture of the extent of contamination of water resources in rice-cultivated zones by pesticides . Ecotoxicological issues in rice cropping ecosystems are presented in chapter five . The next four chapters provide a detailed description of guidelines and procedures which have been developed in Europe, USA and Japan for pesticide risk assessment in rice . Lower and higher tier assessment procedures and tools are described and their use and implementation are explained in



x

Preface

detail . Finally a socio-economic aspect of rice cropping systems is presented in the closing chapter of the book . This book will be a useful tool for most new and senior researchers working in the area of pesticide risk assessment and most particularly to people directly involved in the pesticide regulatory scheme including regulators, stakeholders and industrial bodies. The Editors Ettore Capri and Dimitrios Karpouzas

Pesticide Risk Assessment in Rice Paddies: Theory and Practice E. Capri and D.G. Karpouzas (editors) © 2008 Elsevier B.V. All rights reserved

Chapter 1

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices Aldo Ferrero and Antorsio Tinarelli Dipartimento Agronomia, Selvicoltura e Gestione del Territorio, Universita, di Torino, Italy Contents 1. Introduction 2. Rice ecosystems 3. Rice cultivation in Europe 4. Ecological conditions 5. Agronomic practices 5.1. Soil preparation 5.2. Seeding 5.3. Crop rotation 5.4. Irrigation and water management in the rice field 5.5. Fertilization 6. Weeds and weed management 7. Rice diseases 8. Invertebrate pests 9. Ripening and harvesting 10. Paddy drying process and product preservation 11. Crop management and environment References

1 2 4 6 8 8 9 11 11 13 14 20 20 21 22 22 23

1. INTRODUCTION Rice, after wheat, is the second largest cereal crop and the most widely consumed staple food grain. Globally, rice occupies about 145 million ha, a surface which constitutes one-tenth of the arable land, while in the majority of Asian countries, it comprises one-third or more of the planted area (FAOSTAT, 2006). India, China, Indonesia, Bangladesh, Vietnam and Thailand are the largest rice producing countries. The rice production originated from all these countries together accounts for more than three quarters of the total world production. Rice is grown in at least 114 countries with a total production of about 610 million metric tonnes (FAOSTAT, 2006). It has been estimated that rice has fed more people over a longer period than any other crop. Even now, one-half of the world population and virtually all of East and Southeast Asia is entirely dependent upon rice. In Bangladesh, Myanmar and Vietnam, for instance, the annual rice consumption per person ranges from 150 to 200 kg and accounts for most caloric and protein intake (Maclean et al., 2002). Some hundred millions of small farmers and landless workers derive their income from rice production. In most Asian

2

A. Ferrero and A. Tinarelli

countries, rice cultivation provides main employment to the majority of people, as it requires up to 250 person-days per hectare. Rice shows an incredible capacity to adapt itself to a great variety of soil and climatic conditions. It is cultivated in areas such as those in the Upper Sind in Pakistan, where during the rice season temperature averages 33ºC, or in temperate climate areas, like those in Northern Italy, where the mean temperature during the growing season is about 18ºC. Rice is grown under a wide range of water availability; in areas with mean annual rainfall of about 100 mm (Al Hasa Oasis in Saudi Arabia) to areas with mean annual rainfall of 5000 mm (Myanmar’s Arakan Coast). The crop is cultivated at very different altitudes, from the sea level, in river delta, to the Himalayan slopes of Nepal, at 2600 m (Ferrero, 2005). After World War II, rice-cultivated surface increased by almost 70% and the total production roughly tripled. During this period of time, most traditional rice importing countries which faced severe rice security problems (e.g. India, Vietnam, Philippines) achieved self-sufficiency in rice. The greatest achievements were obtained mainly from the early 1960s, with the so-called Green Revolution, which led to the cultivation of new lands to rice or the shift into rice from other crops and to the introduction of higher-yielding varieties or increased cropping intensity. Since 1990 the rice harvested area expanded with a pace of about 0.4% a year, and the average yield of the crop increased by about 2%. If this trend and the estimates of the Asian population growth rate are maintained over the next decades, rice production will fail to keep pace with population growth. In major rice producing and consuming countries in Asia, population is forecasted to grow more than 1% per year. 2. RICE ECOSYSTEMS At present rice is grown in all continents under various environmental conditions, which according to IRRI can be separated into four main ecosystems including irrigated, rainfed lowland, upland and flood prone (Maclean et al., 2002) (Figure 1). The irrigated ecosystems account for about half of the world’s rice harvested area and provide 75% of total rice production. Irrigated rice covers most of the America, Australia and Europe cultivation area and a significant part of Asia and Africa. In relation to rainfall availability, irrigation ecosystems are divided into irrigated wet season and irrigated dry season systems. Irrigated wet season includes ecosystems where rice is grown during the wet season and irrigation water is provided with the purpose to supplement rainfall. Irrigated dry season is typical of the ecosystems where during rice cultivation rainfall is usually very low, and water is mostly provided by irrigation. These areas are characterized by a markedly high solar radiation and evapotranspiration, low pest incidence and high yield potential. Average paddy yield can vary from 4 to 10 t ha⫺1 according to input levels that are adopted. Highest values are recorded in America, Australia and Europe where in most situations yield exceeds 8 t ha⫺1, with only one cultivation in a year. Irrigated rice is grown in leveled and bunded fields, where irrigation water is maintained in a layer ranging from 2.5 to 15.0cm depending on the availability of water. Rice is directly seeded or transplanted in

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices

Irrigated

Rainfed lowland

Upland

Flood-prone

3

- 48% of world’s harvested area - 75% of total production - bunded leveled fields - water control (5-15 cm depth) - water from irrigation (from rivers, lakes, pumped, etc.) and rainfall - average paddy yield: 4-10 t ha−1 - high input supply

- 32% of world’s harvested area - 19% of total production - level or slightly sloping bunded fields - no water control (0 - 30 cm depth) - water from rainfall - average paddy yield: 1-3 t ha−1 - medium-low input supply

- 13% of world’s harvested area - 4% of total production - non-leveled unbunded fields - slightly or highly sloping fields (0-40%) - no water control (0 cm or variable) - water from rainfall - average paddy yield: 1-4 t ha−1 - low input supply

- 7% of world’s harvested area - 2% of total production - unbunded fields - level or slightly sloping fields - no water control (0.5 - 5 metres depth) - water from rainfall or tides - average paddy yield: 1-1,5 t ha−1 - low input supply

Fig. 1. Main rice ecosystems in the world.

puddled soil. Rice varieties that are suitable for irrigated ecosystems are usually short in size, highly responsive to N supply, resistant to many pests and with high yield potential. Constraints to sustainable rice production in irrigation ecosystems include losses from weeds, pest and diseases, inadequate water management, environment degradation because of excessive inputs. This ecosystem is losing land due to urbanization and industrialization. Rainfed lowland ecosystems are characterized by slightly leveled and bunded fields where water is provided by rainfall and can have variable depth. In these

4

A. Ferrero and A. Tinarelli

conditions, there is no control over the water level and during the same season rice is severely exposed to the risk of deep flood and drought, with an alternation of anaerobic and aerobic environment (Garrity et al., 1986). Worldwide, about 54 million ha of rice, mostly concentrated in Asia and Africa, are grown under rainfed conditions. These ecosystems are frequent in areas with adverse climate and poor soil, which prevent growers from applying costly and modern technologies. Rice grown in rainfed ecosystems includes mostly traditional photoperiodicsensitive varieties, which are often poorly responsive to fertilizer supply. For all these reasons rice yields are variable, but in most cases quite low. Major constraints are related to the high degree of uncertainty in water availability, soil fertility and pest and weed aggressiveness and a low access of rice growers to modern inputs. In upland ecosystems rice is grown in fields located from level valley bottoms to steep mountainous lands with slopes ranging from 0% to more than 40%, respectively. Upland soils are aerobic and are rarely flooded. Upland rice is grown in rainfed fields prepared and seeded when dry, much like wheat or maize. The major risk in such cultivating conditions associates with high runoff and lateral water movement. These ecosystems accounts for about 13% of the world’s harvested rice area, but only 4% of total production. Upland rice is mostly grown in Asia (India and Bangladesh), Africa (on wet hills of West Africa) and Latin America (on gently rolling lands of Brazil). Rice fields are not edged by levees except in India where they are sometimes bunded to save scarce water. In most areas, upland rice is a subsistence crop with a yield ranging from 1 t ha⫺1 in lowinput systems, to 3–4 t ha⫺1 in areas with supplementary irrigation water and fertilizer supply. Nearly 100 million people depend on upland rice as their daily staple food. Upland ecosystems have several heavy constraints, mainly related to weed and disease aggressiveness and low soil fertility. Flood prone ecosystems include rice fields subject to uncontrolled flooding for a period of time ranging from 10 days to 5 months. Water depth may range from 0.5 to 5m during crop growth. In these conditions rice plants, called floating rice, have to excessively elongate their stems in order to reach the water surface. Deepwater rice is mainly grown in Asia and Africa covering about 7% of the world’s ricecultivated land. Flood prone fields are usually located in the deltas of rivers such as the Ganges and Brahmaputra of India and Bangladesh, the Mekong of Vietnam and Cambodia, the Chao Phraya of Thailand and the Niger of West Africa (Greenland, 1997a). Deepwater rice is also grown in level coastal areas of India, Bangladesh, Vietnam, Indonesia and West Africa, which are subject to daily tidal inundation. Flood prone systems suffer from excess water and soil salinity and sudden increase of water (flash floods). Rice yield is extremely variable due to unpredictable alternation of heavy flood and drought and averages about 1.5tha⫺1. Major constraints of deepwater rice are mainly related to the severe environmental stresses, which overwhelms most of the positive effects of the cultivation inputs. 3. RICE CULTIVATION IN EUROPE The introduction of rice in the Mediterranean European countries has long been debated. This plant came first in the North African regions from the East through

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices

5

an unusual series of leaps and bounds, thanks to its incredible capacity to adapt itself to different and variable environmental conditions. It is quite certain that rice has been introduced in the European countries during the early period of Arab domination of southern European regions, at first in the Seville area (Spain) in the eighth century and later in Sicily (Italy) and Camargue (France). In the European Union rice is at present cultivated in about 410,000 ha, mostly located in the Mediterranean countries. It is mostly grown in concentrated areas such as the Po valley in Italy, the Rhone delta in France, the Thessaloniki area in Greece. In Spain, rice cultivation is scattered in several areas such as the Aragon area, Ebro delta, Valencia Albufera, Guadalquivir valley unlike Portugal where the rice cultivation is concentrated in two regions: Tejo and Mondego valleys. Submerged rice cultivation is fundamental for a sustainable management of the wet ecosystems (estuaries, river basins, albuferas, etc.) where the crop is grown (Ferrero and Nguyen, 2004). Many European rice fields are located in natural parks or environmentally protected areas. In these areas, rice allows a diversification of agricultural production and landscape maintenance. The top two rice producers which together contribute more than 80% of the total rice production in Europe are Italy (227,000 ha) and Spain (116,000 ha) (Table 1). In the period 1995–2005, some slight variations in the harvested area have been recorded in each country of cultivation, in relation to the market price or water availability. In 2005, the average crop yield has been 6.43 t ha⫺1 and was quite variable among the different countries, as it ranged from 4.80 t ha⫺1 in Portugal to 7.25 t ha⫺1 in Greece. In the last decade, Greece’s average yield showed a significant reduction because of the unfavorable environmental conditions recorded in 2005. Rice is not a staple food for most of the European population; nevertheless, rice consumption in the continent has increased in the last few years due to immigration and diversification of the diet of the Europeans. In the last two decades, rice consumption has significantly increased in all European countries, whether rice producers (Southern Europe) or non-rice producers (Northern Europe). It is presumable that this trend will continue in the next years, particularly in northern European countries (CEC, 2002). In 2005, milled rice Table 1. Evolution of rice area (ha) and yield (t ha-1) in Europe and Mediterranean countries in 1995-2005 (FAOSTAT, 2006) Country

Italy Spain France Greece Portugal Totala – Weighted meanb Variations 1995–2005 a b

Total area (x 1000 ha) Weighted mean of the yield (t ha⫺1)

Area (x 1000 ha)

Yield (t ha⫺1)

1995

2005

1995

2005

223 105 18 22 24 392a

227 116 19 23 27 412a

6.15 6.80 5.70 7.52 6.04 6.37b

6.17 7.23 5.72 7.25 4.80 6.43b

⫹5.1%

⫹0.9%

6

A. Ferrero and A. Tinarelli

consumption in rice-growing countries ranged from 5.4 to 8.8 kg /capita /year, except in Portugal, where the individual consumption was much higher reaching 17.8kg/capita/year. In southern European countries, about 80% of the rice consumed belongs to japonica varieties (mainly medium and long A type) and 20% to indica varieties. Very often, local and “specialty” varieties are gaining a significant importance for small and medium farms in local markets. For example, varieties like “Carnaroli” and “Vialone nano” in Italy or “Senia”, “Bomba” and “Bahia” in Spain have a good appreciation in local markets, also thanks to the promotion of their quality, through the attribution of APO (Appellation of Protected Origin) and the direct selling of rice by rice growers, who directly process their own rice production on the farm with small rice milling plants. From 1980 to 2004 the number of farms dramatically decreased in all European rice producing countries. For example, the total number of rice farms decreased to one-half in Italy and to one-fifth in the Valencia area of rice cultivation. In the same period the mean surface area per farm showed an increase roughly proportional to the reduction of the number of farms (from 20 to 47 ha in Italy and from 1.9 to 4.7 ha in Valencia area) (Finassi and Ferrero, 2004). In Italy, the area managed by a single worker can range from 40 to 60 ha and the farm size is then usually a multiple of that value. Many farmers, aged from 25 to 45 are graduates. The good returns obtained from rice cultivation during the period from 1970 to 2000 encouraged young farmers to continue their parental enterprise. The level of mechanization and cultivation technology applied in a small farm is not much different from that applied in a big farm, because of the high cost of labor and shortage of manpower. On an average there is a tractor for every 12 ha and a harvester combine for every 60 ha. Most of the tractors bought during the last three years have more than 100kW power (Finassi and Ferrero, 2007). The rice-growing equipments belonging to the typical rice farm includes also a wide set of farm implements: plows, harrows, furrowers, milling-machines, scrapers, manure-spreaders, weeding-machines and many more. The adoption of external services to carry out the cultivation operations is not so frequent, but is constantly increasing. 4. ECOLOGICAL CONDITIONS Rice fields show a wide pedological, edaphic and structural variability and a quite differentiated natural flora between the different areas, because of the old geological genesis and evolution of the soils. A wide variability is also recorded in the climatic parameters, as shown by the relatively broad temperature range in the fields closer to the mountains and the wide difference in rainfall intensity among various rice areas. A frequent alternation between clayey and alluvial soils, even in short distances is a frequent phenomenon in rice-cultivated zones. Some areas close to the river estuaries are peaty due to ancient sedimentation of marshes. Fine textured and poorly drained soils with impervious hardpan or claypan layers are well suited to rice production, since that soils are not very suitable for cultivating other crops. The ideal soil types, according to Mikkelsen and Evatt (1966) are

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices

7

those that conserve water. Usually clay and clay loams, silty clay loams or silt loams are considered most desirable for rice cultivation. Soils with a high clay and silt content are characterized by slow water percolation. Rice soils should be capable of easy surface drainage also, since many aspects of mechanization require removal of the surface water. In the Valencia area (Spain) rice is cultivated in marshlands of a coastal lagoon. Other lands of autochthonous origin nearest to the Alps (known as Baragge), which were in past woods or untilled fields, are cultivated nowadays only with rice, as they are very firm and compact, with little organic matter content and low fertility. The soils cultivated with rice show a very wide range of pH. In Italy they are usually acidic, with a pH ranging from 4.5 to 6.8. In France (Camargue area) the soils are rich in calcareous material (15–45%), thus having high pH ranging from 7.5 to 8.5 (Audebert and Mendez del Vilar, 2007). The organic matter content of soils from different areas of rice cultivation averages to about 2.0% and it can range from a minimum of 0.5% to a maximum of about 14% in peaty soils. In coastal areas (e.g. Camargue in France, Ebro delta in Spain, Po delta in Italy), soils are characterized as saline. The subsoil in the southern half of the Camargue area – recently reclaimed from the sea – has a high salt content. Some areas in the southern half are below sea level and the ground water sometimes has higher salt concentrations than the sea water (36 g l⫺1). According to FAO (1996) classification, in most European areas the primary climate of rice production is temperate-continental, with a cold winter and warm summer and main rainfall occurring during either in the first stages of the crop growth (April–June) or during the harvesting period (September–October) (FAO, 1996). In the Mediterranean countries the climate is subtropical (Mediterranean climate) with a dry summer, with warm, dry, clear days and long growing season, longer than in areas with temperate-continental climate. Average temperatures range from 10 to 12ºC during rice germination and from 20–25ºC during crop flowering. This climate is favorable for high photosynthetic rates and high rice yields, while its low relative humidity throughout the growing season reduces the development, severity and importance of rice diseases. For example, in northwest Italy a minimum and a maximum yearly thermal mean of 8ºC and 19ºC can be recorded, respectively. The Camargue area is constantly swept by strong winds resulting from the meeting of Mediterranean and Atlantic air masses (Barbier and Mouret, 1992). Examination of the wind system shows that the Mistral (a north wind) is clearly the prevailing wind. It accentuates the Mediterranean character of the Rhône delta. Blowing more than 200 days a year, it increases evaporation and contributes to the overall water deficit in the plain of the delta. Drops in temperature can result from cold air streams or night irradiation or heavy storms during summer. Drops in temperature occurring during the first months of the growing season are not likely to seriously damage rice, thanks to the thermostatic function of flooding water, which protects seedlings from injuries of their vegetative organs. Conversely, serious damages from flower sterility and caryopsis abort or poor caryopsis filling can result when drops in temperature occur during flowering.

8

A. Ferrero and A. Tinarelli

5. AGRONOMIC PRACTICES 5.1. Soil preparation Rice field formation starts with the creation of the rice cultivation basins. Their shape does not follow the contour line, as it used to be in the earlier times of ricegrowing; nowadays they are usually made with a rough rectangular shape. Basin size is usually established according to the specific land slope, as well as the soil texture and structure. In areas where the cultivated land is flat, soils are light and water leaching in the subsoil is favored, basins can exceed 10 ha in size. In contrast, in areas where soils are clayey and compact, the most efficient basin size is about 5 ha. Small furrows with a depth of 20–25 cm, and a distance from each other of 18–20 m, are often dug inside the basin in the same direction as the plowing. Several more perpendicular small furrows are set up in order to improve water drainage. The perimeter embankments, or levees, of each plot are permanent, but sometimes they are remade every autumn or spring, in order to minimize the risk of collapse during rice cultivation. To withstand the water loss caused by rodents and other animals that usually try to pierce embankments every year, on both sides of the embankment a slice of ground is cut away, then replaced and pressed with a special straightening roller pulled by the farm tractor. Every two or three years a careful land leveling is carried out in order to form a sowing surface as flat and even as possible. This is done using a scraper with a laser guided leveling blade. In this circumstance, and especially in very well drained soils, the blade of the scraper, at the same time, can be a useful instrument to press the ground and to reduce water loss in the subsoil after submersion. Leveling allows to maintain a uniform and shallow layer of water, with a maximum level variability of ⫾2cm in each cultivation basin. This results in better rice germination after flooding and reduced weed growth. Leveling is a major practice for rice-growing, so that when a laser device is not available, a broad metal board is pulled by a tractor on every wrinkled zone. This implement is employed, paying attention to keep a slight layer of water working as a soil and water level in the basin. Plowing depth usually does not exceed 20 cm. A higher depth would be useless for fertility and it would limit the efficiency of tractors in the submerged rice fields during the operation of fertilization and herbicide spraying. Plowing of very compact, clayey and poorly drained soils is usually carried out during autumn, as the climate variables from autumn to spring supports organic matter degradation and ground structure building by the action of winter chill. In order to prevent faster degradation of organic matter, spring plowing is preferred for shallow soils, which usually have a moderate content of organic matter; the successive tillage practices and sowing should be carried out quickly. Well timed sowing could be important to minimize the competition between emerging rice plants and noxious weeds. Soil tillage is then completed by using rotary harrow which can be useful even on wet soil conditions. To avoid water

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices

9

losses, soil is often compacted through several passes with tractors equipped with special rollers and cage-shaped wheels, during tillage operation (puddling). More and more frequently plowing operations are replaced by minimum tillage. This trend in soil tillage is mainly due to: – labor cost reduction; – improvement of the efficiency of dry seeding technique in non-submerged soils, particularly in smooth soils which limits the operative capacity of machinery after submersion; – application of the false seeding technique through stimulation, before rice seeding, of the germination of weed seeds, especially those of weedy rice. Reduced tillage operations are normally carried out using chisels suitable to operate at 20cm depth followed by rotary harrows for smoothing soil clods. In case of dry seeding, tillage is carried out at 5–10cm to avoid an excessive descent of the rice seeds in the ground. Minimum tillage and no-tillage practices, if repeated for more than two or three years, can cause a yield reduction especially in less fertile soils. For this reason, when minimum or no-tillage practices are used for a long time, it will be necessary to pay particular attention to the fertilization programs modulating carefully the rate and the period of intervention. 5.2. Seeding Seeding periods range from the beginning of April to the end of May and are related to water availability for the field submersion, the choice of the variety (early or late) and climatic conditions (Figure 2). Early varieties are usually seeded in the second week of May after the destruction of the weeds grown before crop seeding. Sometimes a late seeding is applied with early varieties due to unfavorable weather conditions that do not allow the application of the most appropriate agronomical practices.

Fig. 2. Scheme of main agronomical practices carried out during rice cultivation.

10

A. Ferrero and A. Tinarelli

The most popular seeding method consists of broadcast application of rice seeds in fields submerged with a 5–10 cm water layer. Seeding is done by using centrifugal spreading devices, consisting of two rotating disks, which are also used for the distribution of mineral fertilizers. To promote a faster rice germination and prevent seeds from floating and drifting, rice seeds are usually soaked into cold water, in containers or in irrigation ditches for two or three days. Seeds are then drained for 10–12 h and they are broadcasted in the field when seeds show the first signs of germination (the root tip starts to emerge from the caryopsis). Most of the problems regarding seed germination and seedling development are essentially related to the rice field temperature conditions, as drops in temperature to below 10ºC stop the germination process. Such sudden drops in temperature are frequent in rice-cultivated basins. Problems might also arise by non precise leveling of the soil surface or by not appropriate fertilization interventions, that might stimulate weed growth and the development of algal scum covering the rice seedlings. In about 40,000 ha, mostly in Italy, seeds are drilled into dry soil in rows. The rice is generally managed as a dry crop until it reaches the 3–4 leaf stage (Ferrero and Nguyen, 2004). After this period, the rice is flooded continually, as in the conventional systems. In these conditions, rice has no competitive growth advantage over weeds, which can compete with the crop from the beginning of stand establishment. Rice drill seeding on a non-submerged soil is especially suitable for less compact and more permeable soils. Rice seeding is done at a 30cm distance between the rows and at a 1–2cm depth, with the same seeding-machine utilized for wheat. After the rice seedlings have reached the 3 leaf stage (about 30 days after seeding), fields are flooded, as is done the conventional water-seeded rice. The main benefits of drill seeding are water saving, which is linked to the reduced needs for irrigation water during the first period of cultivation, prevention of algal propagation, seeding before water availability (even at the end of March) and the prevention of seed drift due to wind during seeding. Compared to rice plants seeded in flooded soil, those seeded in dry soil commonly show several advantages, including better rooting, higher resistance to lodging and to pests and diseases. The most important disadvantages are related to a lower tillering rate, a greater competition by the weeds and a risk of nitrogen loss due to the important nitrification process that can occur during the pre-flooding period. Considering weeds, dry soil reduces or delays the growth of weed species that require an aquatic environment (e.g. Heteranthera species) but increases the development of non-aquatic weeds such as Panicum dichotomiflorum, Digitaria sanguinalis, Polygonum species. It should be also pointed out that dry seeding is a slower process, as dry seeders complete the seeding of 2 ha h⫺1, while the centrifugal seeding equipment is able to complete the seeding of 5ha h⫺1. An old dry seeding technique, which includes seeding on water saturated soil, is becoming more popular in a few farms. According to this technique, soaked rice seeds are seeded on the top of soil shaped out by a special device of the seeder. This system also allows for the application of mechanical devices to control weeds that grow between the rows.

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices

11

About 70% of the European rice area is cultivated with long- and mediumgrain japonica varieties (Nguyen and Ferrero, 2006). The remaining area is cultivated with long-grain indica varieties. The cultivation of indica-type and early-maturing varieties has significantly increased over the last years in all European rice-growing countries. Most indica-type varieties are short with a low growth and moderate competitive capacity towards weeds. The features of these varieties certainly gave an important contribution to the spread of weedy rice infestations. Early varieties usually have a cultural cycle of 130–145 days (about 20–30 days shorter than regular ones). The growing interest for these varieties is mainly due to the need of escaping negative effects of the low temperatures of April and August months, during which the delicate phases of emergence and flowering, respectively occur. This need favored the selection of high yielding varieties that are seeded in mid-May and begin to flower before August. The availability of short-cycle varieties allows the application of the weedy rice control to be carried out before rice planting. 5.3. Crop rotation In the past, when the rice farm economy was relying on animal and vegetable productions, in most European rice systems, meadows (ryegrass, clover, alfalfa) were put in rotation with gramineus crops such as rice and wheat (and sometimes barley) or rapeseed. Meadow seeding was carried out during August–September during the maturation of rice plants, a little before harvesting, in order to have an already moist but no longer wet soil when climatic and temperature conditions were suitable for germination. Otherwise, only clover was seeded in the springtime. Rice was then transplanted during the following months of May–June, after cutting and haymaking. However, during the 1960s, the industrial revolution resulted to a remarkable reduction of the manpower available for agriculture and the abandoning of animal breeding, which in turn resulted in the abandoning of meadows. In a few years the single-crop farming system became dominant. Rice transplanting and crop rotation disappeared, together with rice varieties that were disease-resistant especially when cultivated with the transplanting method. This trend was also favored by the agricultural policies developed at national and European level. At the present time, in most European areas rice is grown as a single-crop system although this practice may impose several problems. In Camargue area the main cropping system is a rotation consisting of rice – durum wheat or rice, where rice allows the leaching of the excess of salt and wheat which is used to break the weed cycle in rice. 5.4. Irrigation and water management in the rice field In all European countries, rice fields are permanently flooded during most of the crop growth cycle, with some drainings applied during cultivation for enhancing crop rooting, fertilization and herbicide spraying. In rice fields, flooding is used as a mean of meeting plant water needs, regulating temperature conditions and

12

A. Ferrero and A. Tinarelli

limiting weed growth. Most of the irrigation water for European rice fields originates from rivers (Po river in Italy, Ebro river in Spain, Rhone river in France, Tejo river in Portugal, etc.) and lakes, while less than 5% of rice irrigation water is pumped from wells. Through the centuries, a dense network of canals has been built for carrying riverine water. These canals deliver water to the rice fields while at the same time allowing for correct draining, so as to prevent soils from becoming marshy. Once it has been dammed by weirs, water flows by gravity into the canals and then into the fields, commonly applying a flow-through system where water flows from the most uphill to the most downhill basin. In such systems it is estimated that the irrigation water is used at least twice. In some regions of Spain, Italy and Portugal water may be supplied to each basin through a head ditch. Water requirements for rice cultivation can be very different, according to the soil conditions. Total seasonal volumes of irrigation water supplied can vary from about 17,000 m3 ha⫺1 in heavy soils to 42,000 m3 ha⫺1 in sandy soils. The average consumption measured out in clayey and compact soils during the cultivating season is about 1.2–1.5 Ls⫺1 ha, whereas it settles to about 2.5–3.5 L s⫺1 ha⫺1 in loamy soils and reaches values of 5–6 L s⫺1 ha⫺1 in sandy soils because of faster water percolation, although most of the percolated water usually returns into the irrigation network through downhill springs. Water requirements can be reduced by compacting the soil through several passes with tractors equipped with special rollers and cage-shaped wheels, during tillage operations. Submersion brings about important changes in a soil, such as a decrease in redox potential, an increase in pH and specific conductance, the disapprearance of nitrates, an accumulation of ammonia, the generation of a variety of organic substances, an increase in solubility of iron, manganese, phosphate, silica and the displacement of cations into the soil solution (Ponnamperuma, 1965). In rice fields, the water-bearing stratum moves from the depth of almost 2 m during the winter season to a depth of 0.4–1 m during the cultivating season. Also evapotranspiration, which is high during summer, plays an important role in increasing water-consumption. Its average value in rice fields, calculated on a period of 150 days of flooding, is about 4500 m3 year⫺1 ha⫺1 . In each cultivation area, irrigation associations are responsible for the measurement, delivery and management of water flow, in order to meet the needs of the rice farms. Water management has changed quite remarkably over the years. This should be attributed to different growth behavior and features of the new rice varieties (different vegetative vigor during the early stages, tall or short size of the plants, tillering degree, etc.) and the specific needs of the weed control techniques that are applied. In the past, rice field flooding was employed before seeding, keeping water at a level of 10–15 cm or more, up to maturation. At the present time, farmers can manage water flow according to the varying water needs of the crop during the growing season. Seeding is commonly done after flooding the rice field with 5–10 cm of water, and this level is maintained for 15–20 days, until the seedlings reach the 2–3 leaf stage. Afterwards, the water is drained. The moment of this first draining depends on many parameters including: formation of algal scum (that can menace seedling survival); wind (that uproots seedlings); the presence of worms, crustaceans or insects that can uproot or damage the

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices

13

small rice plants; the need to apply herbicides. Most herbicides introduced into the market in the last few years are mainly characterized by foliar absorption, and their application requires a wide exposure of the plant surface to the herbicide spray. Water draining can sometimes result in the death of weaker seedlings, nevertheless this effect is preferable to the possible negative consequences related to the water presence. In case of the drill seeding on dry soil, flooding is done at the 3–4 leaf stage of the crop and just after herbicide application. In almost every situation, it would be important to fertilize the soil with nitrogen at the end of the first draining period (or before flooding, in the case of drill seeding). Further drainage events are sometimes required for stimulating crop rooting or tillering or to allow for herbicide or fertilizer applications. 5.5. Fertilization Fertilization of the soil mostly aims to restore the main plant nutrients removed by crops. Crop nutrition depends on soil fertility, the choice of the rice variety, type of fertilizer, timing and method of fertilization. Rice plants need 16 different elements for their growth and among these nitrogen, phosphorus and potassium are the most important. In fields where soils are acidic and almost calcium-free such as in most Italian rice areas, calcium is also an indispensable element of fertilization programs. The abandoning of crop rotation and animal breeding in rice farms has shifted attention to the use of mineral and mineral⫺organic fertilizers and the management of crop residues. One of the most discussed aspects concerning rice management concerns the fate of the straw after crop harvesting. Some rice farms burn crop residues in the autumn or the following spring, before starting a new rice cultivation. Since 1980s more and more farms bury rice straw, turning them into the soil during plowing operations, usually in the autumn in compact clayey soils or in spring in sandy or half-mixed soils. Spring burial does not allow enough time for complete degradation of rice crop residues because of a series of abnormal fermentations caused by the permanent water stagnation. The straw-burying practice has to be preferred particularly in less anoxic soils, since in the middle or long term, it may result in a better yield response. A way to improve the effects of the straw-burying method in many situations can be extra supply of calcium⫺magnesium and nitrogen (organic or mineral) before burying rice straws, not later than 30 days before the initiation of rice field flooding. The first fertilization intervention is carried out immediately before or after plowing and usually involves the application of simple, complex, organic or mixtures of organic and mineral fertilizers depending on estimated needs. The use of animaloriginated organic substances, byproducts of the sugar industry or organic⫺mineral compounds is becoming more and more interesting, as they simultaneously provide the most important nutrients such as nitrogen, phosphorus, potassium or calcium. To avoid nutritional losses, industry has also provided some fertilizers with extra components that delay nitrification, but at the present, their use remains quite limited because of their high cost compared to that of the conventional fertilizers.

14

A. Ferrero and A. Tinarelli

Notwithstanding the results of many agronomic studies which recommend burying fertilizers during plowing, fertilization is usually carried out after plowing, during harrowing operations. Uneven or incomplete fertilizer burial facilitates the development of algal scum during the early stages of rice growth and favors a faster growth of weeds with subterranean organs of propagation, which can start their vegetative activity in springtime much earlier than rice. Crop requirements vary according to the genetic traits of the planted varieties and the different physiological needs occurring during the crop cycle and the natural sensitivity to diseases. In the case of early varieties, less nitrogen is usually provided because of their shorter cycle and lower resistance to plant diseases, compared to late varieties. Due to the flooding conditions applied in rice cultivation, nitrogen is primarily absorbed by plants in its ammonium form. This is also the more stable form, less prone to be dispersed in the environment through leaching or runoff. Nitrogen is absorbed by rice plants during the early stages of their growth and is primarily utilized by plants for producing straw rather than grains. On the contrary, if nitrogen is supplied to the plants at their later growth stages, it is usually utilized for grain production. The first fertilization intervention, usually provides a nitrogen⫺phosphorus⫺ potassium complex. It is usually carried out before field flooding, because of the easier access to the rice fields when they are still dry. For this reason, most of the fertilizer required by the crop is applied in this period, thus supplying an excessive amount of nutrients at such an early growth stage. The second fertilization intervention, usually includes addition of nitrogen or nitrogen⫺potassium and occurs when rice plants are at the 3–5 leaf stage, at the beginning of tillering. The second fertilization should be carried out when flooding conditions are restored after the drainage often carried out for the herbicide distribution. This should help to reduce nitrogen loss and promote tillering. The number of tillers increases following the temperature rise, with an optimum around 22–25ºC. Low temperatures during tillering may reduce nutrient absorption, and this in turn may result in a significant reduction of the allocation of carbohydrates and other nutrients in the growing organs. A third fertilization can sometimes occur during panicle initiation and is mainly aimed at stimulating panicle development. Late interventions are particularly effective in increasing grain filling and weight. As a general rule, nitrogen is supplied at 80–120kgha⫺1, 50% as pre-planting and 50% as post-planting treatment, using urea or other ammonium-based fertilizers (Bocchi, 1996; Audebert and Mendez del Vilar, 2007). Phosphorus and potassium are supplied at the pre-planting stage at rates of 50–70 and 100–150kgha⫺1, respectively. When rice follows a rotational crop such as soybean or other leguminous species, only phosphorus and potassium nutrients are supplied. 6. WEEDS AND WEED MANAGEMENT In European rice fields weeds are considered the most noxious organisms affecting rice production. It has been estimated that without weed control, at a yield

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices

15

level of 7 to 8 t ha⫺1, yield loss can be as high as about 90% (Oerke et al., 1994; Ferrero and Vidotto, 2007). Herbicides account for more than 80% of the total consumption of pesticides utilized for crop protection, with a total spending of about ¤ 110 million year⫺1. All crop management practices may have a significant influence on the competitive relationships between weeds and rice. For instance, the shift from transplanting to direct seeding and the abandoning of the weed hand-picking which occurred in European rice fields in the early 1950s, stimulated infestations by weeds such as Echinochloa species, Alisma species, sedges and weedy rice. Management of the same weeds was furthermore complicated because of the occurrence of an ecological environment more favorable to their growth after the introduction of short-stature rice varieties and the practice of shallow water in the fields during early stages. The rice field ecosystem is notably complex. It is characterized by numerous weed species, both C4 and C3, characterized by particular morphophysiological traits. C4 plants are mainly present in dry-seeded fields, while C3 species tend to dominate in the submerged rice crop (Bayer, 1991). These plants are associated differently according to the specific ecological conditions and anthropic pressure, which lead in time to the appearance and spread of some species and disappearance of others. The major weed problems in the European rice fields are aquatic species (Table 2) (Batalla, 1989; Ferrero and Vidotto, 2007). In relation to the practices which are adopted to control them, main weeds can be grouped in the following way: – – – – –

Echinochloa species; Heteranthera species; Alisma species and cyperaceae weeds (sedges); various weedy rice biotypes; weeds of the drill seeded fields.

The main Echinochloa species are E. crus-galli, E. crus-pavonis, E. oryzoides, E. erecta and E. phyllopogon. These species are major weeds in rice cropping systems worldwide, both in water and dry-seeded rice (Holm et al., 1977; Ferrero et al., 2002). These plants show a high variability in morphological and competition-related traits, such as plant size, tillering ability, seed dimensions and germination behavior (Barret and Wilson, 1983; Tabacchi and Ferrero, 2003). This variability makes field identification of different species difficult and uncertain. The control of Echinochloa species was carried out in the past by manual weeding. Nowadays they are primarily controlled with various residual and foliar herbicides such as molinate, thiocarbazil, clomazone, propanil, quinclorac, cyhalofop-butyl, profoxydim, azimsulfuron and bispyribac-sodium. Major Heteranthera species present in the European rice fields now are H. reniformis, H. rotundifolia and H. limosa. These weeds are exotic plants that were first reported in Italy in 1962 (Pirola, 1968). In some areas of rice cultivation, these plants have had an enormous diffusion during the last years.

16

A. Ferrero and A. Tinarelli

Table 2. Average diffusion (as a percentage of the infested area on the total area cultivated with rice) and coverage (as an average of the soil coverage determined in the areas were the plants were present) of the main weeds in Italian rice fields Species

Diffusion %

Coverage %

Echinochloa species group Echinochloa crus-galli Echinochloa crus-pavonis Echinochloa hostii Echinochloa colonum Echinochloa phyllopogon

100 95 65 65 54

53 15 7 4 3

Heteranthera group Heteranthera reniformis Heteranthera rotundifolia Heteranthera limosa

80 80 60

75 75 45

Alisma and cyperaceae (sedges) group Alisma plantago-aquatica Alisma lanceolata Schoenoplectus (Scirpus) mucronatus Bolboschoenus (Scirpus) maritimus Butomus umbellatus

79 35 61 48 12

72 48 57 50 8

Weedy rice group Oryza sativa

75

30

5 3 3 2 1 1

15 6 4 5 3 3

Others Cyperus serotinus Eleocharis species Sparganium erectum Alopecurus geniculatus Ammania coccinea Bidens cernua

Pre-seeding application of oxadiazon is the main weed control strategy that rice farmers apply to control Heteranthera species. It is estimated that oxadiazon is now applied to about 80% of the rice areas infested by this weed. Weed infestations of dry-seeded rice fields include Alisma plantago-aquatica and A. lanceolatum, Cyperus difformis, Bolboschoenus maritimus, Schoenoplectus mucronatus and Butomus umbellatus. These plants are considered together, as they are normally subjected to common control programs and are often sensitive to the same herbicides. Weed management currently relies on strategies based on a combination of herbicide application with appropriate agronomic practices, such as soil leveling, accurate water management, variety choice. Neither chemicals nor cultural practices alone are adequate to give satisfactory weed control (Bayer and Hill, 1993). Precision land leveling, obtained with laser-directed equipment, is the agronomic practice that has greatly contributed to weed management in European rice production. Regular slopes within basins enable appropriate water management,

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices

17

leading to both limiting weed growth and assuring a uniform emergence of weeds, which consequently results in better control, thanks to the use of herbicides. Adoption of short-cycle varieties allows the application of stale seed bed technique for the control of weedy rice before rice planting by applying systemic herbicides. The technologically advanced equipment for herbicide distribution fit to spray herbicides with low water volume and pressure has allowed to improve significantly the efficiency of the applied products and limiting the risk of environmental pollution. Besides, the higher width of the boom sprayers has permitted a reduction of the number of passes in the basins with tractors equipped with toothed wheels, thus diminishing the cost of spraying and the late weed germinations that occur along the wheel tracks. Numerous herbicides are at present available to control major rice weeds (Table 3) (Ferrero et al., 2002). In the last year much effort has gone into developing herbicide strategies that maximize the use of the commercial products while reducing the number of treatments. The programs of weed control in rice are principally established on the basis of the degree of weedy rice infestation and the conditions of crop seeding (water or dry seeding). In case of infestations characterized by the weedy rice presence together with other species, weed control programs are principally aimed at weedy rice control. These practices are mostly performed in rice pre-planting either with an antigerminative herbicide such as flufenacet or pretilachlor applied about one month before rice planting, or with mechanical means (harrows or cage wheels) or systemic graminicides (dalapon, cycloxydim, glyfosate) to destroy weedy rice seedlings grown after stale seedbed application. In both cases, a second treatment is usually required to control Heteranthera species in rice pre-planting, and at least a third treatment with a mixture of specific herbicides, to control Echinochloa species, sedges and other weeds, 25–40 days after rice planting. When the pre-planting treatments against weedy rice are not sufficiently effective, rice growers frequently carry out an additional intervention against this weed at rice flowering time. With limited weedy rice infestation, the control of the weed is usually done manually. In the case of high infestations, the weed is devitalized by applying systemic herbicides (glyphosate) with wiping bars, provided that the weed plants are taller than those of the crop. Particularly promising appears for weedy rice control the Clearfield technology based on the planting of a rice variety tolerant to imazamox, an imidazolinone herbicide with a wide spectrum of activity that also includes weedy rice plants. In case of absence of weedy rice and with infestations characterized by the presence of most common weeds such as Echinochloa, Heteranthera and cyperaceae species, two or three treatments are usually required. In general, one is done in pre-emergence mainly against Heteranthera species and one or two from 10 to 40 days after crop emergence. The first treatment is principally carried out with oxadiazon (at 0.3–0.4 kg a.i. ha⫺1) combined with a graminicide effective against Echinochloa species (e.g. thiobencarb or molinate) and sometimes with an ALS inhibitor applied at 1/2 or 1/3 of their normal dosage rate, against sedges and other weeds.

18

A. Ferrero and A. Tinarelli

Table 3. Rate and application timing of the herbicides applied in Italy against main weeds of rice Target weeds and active ingredients

Rates (kg a.i. ha⫺1)

Target weeds: Echinochloa species Molinate 3.0–4.5 3.0–4.5 Thiobencarb 3.0–4.0 3.0–4.0 Quinclorac 0.5–0.6 Propanil 3.5–4.0 ⫹3.5–4.0 Bispyribac-sodium 0.02 Profoxydim 0.1 Azimsulfuron 0.02 Cyhalofop-butyl 0.2–0.3 Penoxsulam 0.04 Imazamox 0.05 Pendimetalin

1.0–1.3

Clomazone

0.2

Target weeds: Alismataceae and Cyperaceae Bensulfuron-methyl 0.06 Cinosulfuron 0.06–0.08 Ethoxysulfuron 0.06 Bensulfuron-methyl + 0.05 ⫹0.002 Metsulfuron-methyl Metosulam 0.06–0.08 Azimsulfuron 0.02 MCPA 0.4–0.6 Triclopyr 0.3–0.4 Bentazone 1.2–1.6 Penoxsulam 0.04 Imazamox 0.05

Target weeds: Heteranthera species Oxadiazon 0.2–0.4 Pretilachlor 1.0–1.1 Triclopyr 0.3–0.4 Imazamox 0.05

Target weeds: weedy rice Flufenacet Pretilachlor

0.4 1.0

Application timing

pre-seeding post-emergence pre-seeding early post-emergence post-emergence late post-emergence early post-emergence early post-emergence early post-emergence early post-emergence early post-emergence post-emergence of rice (in combination with tolerant rice varieties) pre-emergence (only in dry-seeded rice) pre-emergence (only in dry-seeded rice) post-emergence post-emergence post-emergence late post-emergence post-emergence post-emergence post-emergence late post-emergence post-emergence early post-emergence post-emergence of rice (in combination with tolerant rice varieties) pre-seeding early post-emergence late post-emergence post-emergence of rice (in combination with tolerant rice varieties) pre-seeding (30 days before seeding) pre-seeding (30 days before seeding)

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices

19

Table 3. (Continued) Target weeds and active ingredients

Rates (kg a.i. ha⫺1)

Target weeds: weedy rice Cycloxydim Dalapon Glufosinate-ammonium Glyphosate

0.4 12.0–15.0 1.0 1.0–1.2

Imazamox

0.05

Application timing

pre-seeding (after stale seed bed) pre-seeding (after stale seed bed) pre-seeding (after stale seed bed) pre-seeding (after stale seed bed) post-emergence of rice (wick bars) post-emergence of rice (in combination with tolerant rice varieties)

These products are applied 3–4 days before planting and soil flooding or 5–6 days before planting on flooded soil. The second and, if necessary, a third treatment, are frequently necessary to control late emergence of Echinochloa species, sedges, alismataceae and other species. When rice is directly seeded in dry soil, in general two or three treatments are required: – The first one is carried out in rice pre-emergence with pendimethalin (1.0– 1.3 kg a.i. ha⫺1) or clomazone (0.2 kg a.i. ha⫺1) to control Echinochloa species, P. dichotomiflorum and other weed grasses; with a presence of Heteranthera species oxadiazon 0.2–0.4 kg a.i. ha⫺1 is usually added to these two herbicides; – The second one is carried out from 10 to 30 days after rice emergence, with propanil in combination with an ALS inhibitor to control sedges, B. umbellatus, A. plantago-aquatica and Echinochloa plants which escaped the pre-emergence treatment; – The third one, if necessary, is carried out just before flooding, against the weeds which were not controlled by previous treatments or emerged late; sometimes this intervention is performed after field flooding, by applying the same products used in rice seeded in flooded field. Herbicides are an essential component of rice weed control programmes, but their improper use may result in the appearance of resistant species, cause environmental pollution and risk disrupting the precarious balance of the natural enemies to pests. Several cases of weed resistance to herbicides have been reported in the European rice areas. Main resistant populations belong to S. mucronatus, A. plantago-aquatica, C. difformis and Echinochloa species (Busi et al., 2006). Herbicide potential to contaminate rice floodwater and groundwater is an environmental concern that should always be addressed (Ferrero et al., 2001). Circulation of treated floodwater should be stopped until the herbicides is dissipated. This practice allows to reduce contamination of non target waters and maximize herbicide effect on weeds.

20

A. Ferrero and A. Tinarelli

7. RICE DISEASES Rice is vulnerable to several diseases. Most of the diseases in the European rice fields are caused by parasitic fungi including: Pyricularia oryzae (Magnaporthe grisea), the agent of blast disease, Bipolaris oryzae (Cochliobolus miyabeanus), the agent of the brown spot disease, Fusarium moniliforme (Gibberella fujikuroi), the agent of the bakanae disease (Luppi et al., 2000; Aguilar Portero, 2001). Other fungi may also infest the rice plants, as for example, Sclerotium oryzae, the agent of the stem rot and those that cause seedling damping off. In most European growing areas symptoms of rice diseases are frequent, however epidemics and heavy yield losses are uncommon. Most of the pathogenic agents are seed- or soil-borne pathogens and can proliferate through the stubbles. Fungicide seed treatments help to significantly diminsh the occurrence of fungal diseases and of seedling damping off, in particular. Accurate application of nitrogen fertilizers, minding the correct timing and rate may diminsh the occurrence of certain pathogens in rice plants. Soil analysis facilitates the selection of the correct combination of fertilizer for supplying all the important nutritional elements. Particular attention has to be paid to possible lack of potassium, in order to avoid imbalances in cellular concentration, on which plant disease resistance also depends. On the whole, the best guarantee for productive performances and for a good protection against diseases relies on strong natural soil fertility, on a correct C/N ratio and on a high cation-exchange capacity. Rice farmers are usually well aware of symptoms of fungal diseases. An important sign of a high sensitivity to a disease can be the display of an early and exaggerated growth during the vegetative phase: in this case, draining should be carried out immediately and rice fields should be kept dry as long as possible. When pathogen attack causes the first leaf spots, it is necessary to timely apply specific fungicides. Especially against the blast disease, two treatments are usually necessary: one when the first leaf spots appear, at the panicle formation (or a little later); and a second one at the booting stage, a little before panicle exertion (heading). Sometimes it is suggested to only do the second treatment. Most fungicides applied against blast disease in Europe are: tricyclazole, azoxystrobin, pyroquilon, isprothiolane, prochloraz, tebuconazole. In these last few years particular emphasis has been given to the development of rice varieties exhibiting resistance to major rice pathogens. 8. INVERTEBRATE PESTS Invertebrate pests are not considered a major threat for rice production in Europe. Some invertebrates, such as crustaceans, insects and worms can affect the establishment of the crop. The most dangerous among crustaceans is Triops longicaudatus (tadpole shrimp) which feeds on young rice seedlings, uprooting them and making water muddy (Aguilar Portero, 2002). Water turbidity results in a lower photosynthetic activity and growth reduction in the younger plants. Immediate seeding after flooding is the best agronomic system to prevent damages from tadpole shrimps. The main insect pest of rice is Hydrellia griseola (rice leafminer) which is dangerous in the early crop stages. It is a fly which can injure rice plants when their leaves are floating on the water. Damages are related to the tunnels the

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices

21

insect can make on the leaves and the consequent growth reduction of the young plant. Rice leafminers can be controlled by lowering the water level to favor plant growth. Among the insects which can attack rice plants during their late vegetative stages are Chilo suppressalis (stem borer) and Eusarcoris inconspicuus (called “pudenta” in south of Spain). The most visible symptom of the stem borer is the so-called “white head”. In the area of higher risks of attack (Estremadura, Spain) this pest is mainly controlled by using lures in traps or mating disruption products. Eusarcoris inconspicuus which causes mainly damages to the rice grain during its formation is mostly controlled by means of aerial ULV applications of insecticides (tebufenozide) (Batalla, 1989; Aguilar Portero, 2001). 9. RIPENING AND HARVESTING Between mid-to late-August, rice fields are drained. In this period, rice plants have passed the caryopsis formation stage and have entered the dough grain stage. The water supply from irrigation canals is interrupted so as to dry up the smallest ditches feeding the fields. The aim is to dry rice fields as quickly as possible before starting harvesting operations, for which heavy machines are usually used. The best climatic conditions for rice ripening occur when the temperature difference between day and night is around 10ºC and when daytime temperatures are around 24ºC. Temperatures higher than 30ºC result in a faster cellular respiration rate. In the exceptional case of persistence of high daytime temperatures (over 34–35ºC) for a long time, the higher respiration and transpiration rates might cause a shortening of the duration of the ripening phase and production of caryopsis with lower weight and smaller dimensions, which results in yield reduction. Frequent and prolonged rains can slow down ripening and, when combined with minimum temperatures lower than 10ºC, they can completely hamper the maturation of rice grains which are at the panicle base (the ones that were formed late or partially). In addition to yield loss, the presence of these grains in the harvested paddy usually results in reduction of milling quality. Rice yield may be strongly influenced by air temperature during flowering: exposure for a period of 3–5 days at temperatures of 12–13ºC or 8–10 days at temperatures of 15–18ºC may induce pollen sterility. A reduction of risk from exposure to low temperatures during panicle formation and flowering may be achieved by maintaining the water level in the paddy fields to at least 15cm. If air temperatures fall below 3–4ºC for 1–2 days while the ripening is ending, the caryopsis maturation might stop and a subsequent temperature increase will not restart ripening. Harvest time is determined by the average grain water content, that has to reach an optimum average of 21–22%, in the case of the indica varieties, and an average of 24–25%, in the case of japonica varieties (Luppi et al., 2000). When harvest is carried out too early or too late, a reduction of the whole grain yield can occur. If harvest is done too early, grains that have not completed maturation appear chalky. If harvest is carried out too late, the almost ripened grains can suffer sun-cracking: grains absorb dew drops at night and this water, under the strong action of the sun during the day, evaporates very quickly with an increase of pressure inside the rice grain and a consequent grain cracking. This results in grain breaking during milling operations.

22

A. Ferrero and A. Tinarelli

Harvesting is usually carried out by combines equipped with caterpillars replacing the front wheels. Combines used for rice harvesting are designed to keep their unitary weight under 300g cm⫺2 in order to avoid damaging the field leveling. Besides that, many of them are specially equipped to cut the straw immediately when coming out of the straw-walkers. In relation to the dimension of their thresher cylinder, combines have a different hourly productivity, whose value ranges between 6 and 10 t h⫺1. Grain-losses should be below 1–2%. 10. PADDY DRYING PROCESS AND PRODUCT PRESERVATION Most rice farms are equipped with their own drying systems. Dryers can be static or dynamic. In both cases, they operate with ambient or hot air obtained from oil-fired or methane-fired burners. The basic principle of the drying process is to reduce paddy humidity under a maximum level of about 14%. Hot air should be kept at a temperature that can generate an optimum condition of heating in the inside of the paddy mass with temperatures around 30ºC. The drying process ends after 2h of air flowing with burners off, in order to prevent moist air from penetrating rice grains. Drying-equipments based on an infrared system (infrared dryers) guarantees lower installation and managing costs, while offering the same results. Such equipments have already been tested in the USA and in Taiwan and they are now going to be installed in Italy. Every rice farm has its own storehouse to keep rice until the market is favorable for selling. In order to preserve rice from fermentations and from insect attacks, many farms have equipped their storehouses with refrigerating devices that create optimum temperature (10–15ºC) and humidity conditions for storing rice. Paddy rice processing is carried out mainly by the numerous mills existing throughout the cultivation area. A few rice-growing farms have their own milling plants for processing their production and are organized for the direct selling of the processed rice. 11. CROP MANAGEMENT AND ENVIRONMENT The agricultural practices, which are usually adopted in flooded rice cultivation, may have a significant impact in the environment. Plowing and puddling often result in the development of a dense layer below the cultivated topsoil. Puddling decreases the macroporosity of clay soils and increases their microporosity, thereby causing an increase in water-holding capacity of the puddled soil (Sharma and De Datta, 1986). After submersion soil pores become totally saturated by water, clods swell up and are gradually dispersed. Structure breaks up with a rapid interruption of the gas exchanges between soil and atmosphere (Mikkelsen and De Datta, 1991). Oxygen content gradually reduces from the top to the deep layers. A thin oxdized soil layer still remains in the 1–2 top mm due to the oxygen present into water and photosynthetic activity of the aerobic organisms which are in this layer (Figure 3) (Aguilar Portero, 2001). An oxdized zone is also normally present around the rice

Rice Cultivation in the E.U. Ecological Conditions and Agronomical Practices

Water layer Aerobic soil layer (1-2 mm)

Aerobic layer

Plough pan

Anaerobic layer

23

Fig. 3. Aerobic and anaerobic layers in the rice fields (from Aguilar Portero, 2001).

roots due to the transport of oxygen through the aerenchyma of the rice plant to the rice surface (Liu et al., 1990). Soil submergence, no matter how prolonged it is, makes rice soil much different from all arable soils (Greenland, 1997b). Flooded topsoil, due to the reducing conditions, favor the mobilization and accumulation of Fe and Mn. In such conditions, permeability of the soil is low, thus facilitating a horizontal rather than a vertical water movement through the topsoil (Grant, 1964). The fate of a chemical during and following an application is an environmental issue that should always be addressed. REFERENCES Aguilar Portero, M. (2001). Cultivo del arroz en el sur de España. Lince Artes Graficas, 192. Audebert, A. and Mendez del Vilar, P. (2007). Characterization of rice crop systems and rice sector organization in Camargue-France. In: A. Ferrero and F. Vidotto (Eds), Agro-Economical Traits of Rice Cultivation in Europe and India (pp. 278). Edizioni Mercurio, Vercelli, Italy. Barbier, J. M. and Mouret, J. C. (1992). Le riz et la Camargue. Inra mensuel 64, 39–51. Barret, S. C. H. and Wilson, B. F. (1983). Colonizing ability in the Echinochloa crus-galli complex (barnyard grass) ⫺ II: Seed biology. Can. J. Bot. 61, 556–562. Batalla, J. A. (1989). Malas hierbas y herbicidas en los arrozales españoles. Phytoma 8, 36–42. Bayer, D. E. (1991). Weed management. In: B.S. Luh (Ed), Rice Production (vol. I, pp. 267–309). AVI book, Van Nostrand Reinhold, New York. Bayer, D. and Hill, J. (1993). Weeds. In: M. Flint (Ed), Integrated Pest Management for Rice (vol. 3280, pp. 32–55). University of California and Division of Agricultural and Natural Resources, Oakland, California. Bocchi, S. (1996). Principi di fertilizzazione del riso – 1. L’azoto nell’agro-sistema della risaia sommersa. Ente Nazionale Risi. Quaderno n. 13. Busi, R. Vidotto, F., Fischer, A. J., Osuna, M., De Prado, R. and Ferrero, A. (2006). Patterns of resistance to ALS herbicides in smallflower umbrella sedge (Cyperus difformis) and ricefield bulrush (Schoenoplectus mucronatus). Weed Technology, 20, 1004–10014. CEC (Commission of the European Communities) (2002). Rice, Markets, CMO and Medium Term Forecast. Commission Staff Working Paper. SEC (2202) 788. FAO (1996). Groups and Types of World Climates. Map. FAO, Rome, Italy. FAOSTAT (2006). http://faostat.fao.org/. Ferrero, A. (2005). Preface. In: A. Ferrero and M. Scansetti (Eds), Rice Landscapes of Life (pp. 207). Edizioni Mercurio, Vercelli, Italy. Ferrero, A. and Nguyen, V. N. (2004). The sustainable development of rice-based production systems in Europe. Proc. of the FAO Rice Conference “Rice is Life”, vol. 53, pp. 115–124.

24

A. Ferrero and A. Tinarelli

Ferrero, A., Tabacchi, M. and Vidotto, F. (2002). Italian rice field weeds and their control. In: J. E. Hill and B. Hardy (Eds), Proc. 2nd Temperate Rice Conference, 13–17 June 1999, Sacramento, CA, USA. Los Baños (Philippines): International Rice Research Institute, pp. 535–544. Ferrero, A. and Vidotto, F. (2007). Weeds and weed management in Italian rice fields. In: A. Ferrero and F. Vidotto (Eds), Agro-Economical Traits of Rice Cultivation in Europe and India (p. 278). Edizioni Mercurio, Vercelli, Italy. Ferrero, A., Vidotto, F., Gennai, M. and Nègre, M. (2001). Behaviour of cinosulfuron in paddy surface waters, sediments, and ground water. J. Environ. Qual. 30, 131–140. Finassi, A. and Ferrero, A. (2004). Outline of the Italian farm structure. In: A. Ferrero and F. Vidotto (Eds), Proc. of the Conference “Challenges and Opportunities for Sustainable RiceBased Production Systems”. Torino, Italy, 13–15 September 2004 (pp. 583–596). Edizioni Mercurio, Vercelli, Italy. Finassi, A. and Ferrero, A. (2007). Mechanization and labor organization of rice farms in the Vercelli area. In: A. Ferrero and F. Vidotto (Eds), Agro-Economical Traits of Rice Cultivation in Europe and India (pp. 278). Edizioni Mercurio, Vercelli, Italy. Garrity, D. P., Oldeman, L. R., Morris, R. A. and Lenka, D. (1986). Rainfed lowland rice ecosystems: Characterization and distribution. In: Progress in Rainfed Lowland Rice (pp. 3–23). IRRI, Los Baños, Philippines. Grant, C. J. (1964). Soil characteristic associated with the wet cultivation of rice. In: The mineral nutrition of the rice plant. Proc. of a Symposium at The International Rice Research Institute (p. 494). The Johns Hopkins Press, Baltimore, Maryland. Greenland, D. J. (1997a). Rice Farming Today. In The Sustainability of Rice Farming (pp. 43–74). Cab International in Association with the International Rice Institute, Wallingford Oxon, UK. Greenland, D. J. (1997b). The Biophysical Basis of the Sustainability of Rice Farming. In the Sustainability of Rice Farming (pp. 65–102). Cab International in Association with the International Rice Institute, Wallingford Oxon, UK. Holm, L. G., Pucknett, D., Pancho, J. V. and Heberger, J. P. (1977). The World’s Worst Weeds (p. 610). The University Press of Hawaii, Honolulu, USA. Liu Z., Shi W. and Fan X. (1990). The rizosphere effects of phosphorous and iron in soils. Transactions 14th International Congress Soil Science 2, 147–152. Luppi, G., Finassi, A. and Cavallero, A. (2000). Riso (Oryza sp.pl.). In: R. Baldoni and L. Giardini (Eds), Coltivazioni erbacee. Cereali e proteaginose (233–285). Patron Editore, Bologna, Italy. Maclean, J. L., Dawe, D. C., Hardy, B., Hettel, G. P. (Eds) (2002). Rice almanac. Los Baños (Philippines): International Rice Research Institute; Bouaké (Côte d’Ivoire): West Africa Rice Development Association: Cali (Colombia): International Center for Tropical Agricolture: Rome (Italy): Food and Agricolture Organization (pp. 11–29). Mikkelsen, D. S. and De Datta, S. K. (1991). Rice culture. In: Bor S. Luh (Ed), Rice ⫺ I: Production (p. 439). Van Nostrand Reinhold, New York. Mikkelsen, D. S. and Evatt, N. S. (1966). Soils and Fertilizers. In: Rice in the United States: Varieties and production. Agricultural Research Service U.S. Department Agriculture Handbook No. 289. Nguyen, V. N. and Ferrero, A. (2006). Meeting the challenges of global rice production. Paddy and Water Environment 4, 1–9. Oerke, E. C., Dehene, H. V., Schoenbeck, F. and Weber, A. (1994). Rice Losses. Crop Production and Crop Protection. Estimated Losses in Major Food and Cash Crops (p. 808). Elsevier Science B.V., Amsterdam. Pirola, A. (1968). Heteranthera reniformis Ruitz et Pavon (Pontederiaceae) avventizia delle risaie pavesi. Il Riso 4, 15–21. Ponnamperuma, F. N. (1965). Dynamic aspects of flooded soils. The mineral nutrition of the rice plant. Proc. of a Symposium at The International Rice Research Institute (p. 494). The Johns Hopkins Press, Baltimore, Maryland. Sharma, P. K. and De Datta, S. K. (1986). Physical properties and processes of puddled soils. Adv. Soil Sci. 5, 140–178. Tabacchi, M. and Ferrero, A. (2003). Morphological traits related to Echinochloa spp. infesting Italian rice fields. Third International Temperate Rice Conference, Punta del Este, Uruguay, 10–13, March.

Pesticide Risk Assessment in Rice Paddies: Theory and Practice E. Capri and D.G. Karpouzas (editors) © 2008 Elsevier B.V. All rights reserved

Chapter 2

Regulatory Aspects of Pesticide Risk Assessment Andrew Craven1, James Garratt2 and Laura Padovani3 1Independent

Consultant, 23 Wenlock Drive, Escrick, York YO19 6JB, UK Nanotechnology Centre, Herschel Building, Newcastle University, NE1 7RU, UK 3European Food Safety Authority, Largo N. Palli 5/A, I-43100 Parma, Italy

2Enviresearch,

Contents 1. Introduction 2. The legislative framework 2.1. Directive 91/414/EEC concerning the placing of plant protection products on the market 2.2. Maximum residue levels 3. The scientific assessment of active substances 3.1. Risk assessment for new active substances 3.1.1. Physico-chemical properties 3.1.2. Potential toxicity in humans 3.1.3. Dietary intake and risk assessment 3.1.4. Exposures and risk assessment for operators, other workers and bystanders 3.1.5. Environmental fate and behaviour 3.1.6. Ecotoxicology and environmental risk assessment 3.1.7. Efficacy and risk to following crops 3.2. Reviews of pesticides that are already approved 3.2.1. The precautionary principle in pesticide regulation 3.2.2. Checking for possible adverse effects of pesticide usage 3.2.3. Re-approval of a pesticide 3.2.4. Revocation of a pesticide approval 4. The process of pesticide approval and registration 4.1. The approvals process 4.2. EU organisations involved with pesticide approval 4.2.1. The European Commission 4.2.2. The Council of the EU 4.2.3. The European Parliament 4.2.4. The European Food Safety Authority (EFSA) 4.2.5. Regulatory Committees 4.2.6. Plant Protection Products Working Groups 4.3. MS organisations involved with pesticide approval 4.3.1. MS government departments 4.3.2. MS pesticide registration authorities 4.3.3. MS expert advisory committees 5. Current developments in the pesticide regulatory process in the EU 5.1. Technological advances 5.2. Scientific advances in risk assessment 5.3. Public opinion and expectations 5.4. Administrative and legislative changes References

26 26 27 28 28 29 29 29 31 31 32 33 33 34 34 34 35 36 36 36 37 37 38 39 39 39 39 39 39 40 40 40 40 41 42 43 43

26

A. Craven et al.

1. INTRODUCTION Pesticides can make a significant contribution to a modern agricultural production system. For example, they enable crops to be produced more efficiently, they reduce the contamination of food by toxic fungi and they are used to control insects that spread human diseases. Since they are designed to be biologically active, pesticides also have the potential to harm humans and other species that are not their intended target. Moreover, by their action on a target organism, they may alter the broader balance of nature, and this may be undesirable. It is therefore important to control the use of pesticides, carefully assessing the risks they may pose and ensuring that they are capable of delivering the benefits claimed for them. In controlling potential environmental hazards such as pesticides, it is important not only that the right balance is drawn between benefits and risks, but also that the rationale for decisions is made clear and accessible to the public. If the regulatory system is opaque and not open to scrutiny, it is liable to be distrusted. Over recent years, steps have been taken, particularly within the EU context, to increase openness. Under the system for approval of pesticides within the EU, each product is classified as agricultural or non-agricultural, depending on its intended use. Agricultural pesticides include products used in agriculture, horticulture and forestry, weedkillers for use in and around watercourses and lakes and weedkillers for use on non-crop land such as roads and railways. Because pesticides may work in ways other than killing pests (e.g. by pest repellence or growth regulation), the popular title of ‘agricultural pesticide’ is more accurately changed within the EU to ‘plant protection product’ (PPP). Non-agricultural pesticides are usually called ‘biocides’ in the EU. These products include those used in wood preservation; disinfectants; surface treatment (e.g. fungicides applied to masonry); public hygiene insecticides used in domestic and other areas; anti-fouling products (e.g. paints for application to boats). This chapter explains how pesticides are currently regulated in the EU and its constituent Member States (MS), what information is used to assess the risks that they might pose and the roles of the various organisations that participate in the regulatory process. This chapter concentrates on agricultural pesticides (PPPs), but many of the principles discussed are also relevant for biocides. 2. THE LEGISLATIVE FRAMEWORK The legislative framework is designed with the aim that: (a) pesticides should only be approved for use if they are effective; (b) no-one should develop any serious illness through the use of pesticides; (c) no-one should be harmed or made ill by the presence of pesticide residues in food or drink; and (d) when pesticides are used according to the conditions of their approval, any adverse effects on wildlife or the environment are sufficiently small to be deemed acceptable.

Regulatory Aspects of Pesticide Risk Assessment

27

The EU and its constituent MS have an extensive range of legislative and administrative controls over the regulation of PPP. These controls contribute to the overall EU and MS Government policy of providing effective means of pest and disease control, consistent with protecting the safety of people, wild creatures and the environment. The main laws that apply in this area are Directive 91/414/EEC (concerning the placing of PPPs on the market) and legislation surrounding maximum residue levels (MRLs) in produce (Regulation (EC) No. 396/2005 of the European Parliament and of the Council; 23 February 2005). 2.1. Directive 91/414/EEC concerning the placing of plant protection products on the market At present there are two parallel systems for the approval of PPP in the different MS of the EU. Under the first (older) system, the scientific evaluation of PPP is carried out entirely at a national (MS) level. However, this is gradually being replaced by a second system in which a major part of the scientific evaluation is organised by the European Commission. This transition was introduced first for PPP, and is now being extended to biocidal products, with separate legislation (Directive 98/8/EC). Under the newer European system, a committee of MS assesses the active substances in PPP and if they are shown to be acceptable they are entered on a positive list of such substances approved at EU level, called Annex I of Directive 91/414/EEC. Once an active substance has been listed in this way on Annex I, applications can be made to have products that contain it approved in individual MS for specified uses; and in responding to such an application, the MS government concerned would be expected to draw upon the scientific assessment that had already been agreed for the active substance at EU level. Thus, although the final say in whether a product is approved remains at a national level, unnecessary duplication of activities between MS is reduced. The evaluation of new and existing (already on the market in EU MS in July 1993) active substances used in PPPs begins with an evaluation of the scientific data by a rapporteur MS, to ensure that the active substance fulfils all the Directive’s requirements concerning safety to man and the environment. The draft assessment report (DAR) on this initial assessment provided by the MS is then submitted to a peer review by experts from MS and scientists of the European Food Safety Authority (EFSA). Within 12 months, a final report on the risk assessment of the substance is submitted to the European Commission with recommendations regarding inclusion in Annex I. The recommendations are considered by all MS through the European Commission’s Standing Committee on the Food Chain and Animal Health (SCFAH). Once an active substance has been listed on Annex I, individual MS may decide whether to authorise PPPs containing the substance within their national boundaries. National MS government regulations will continue to apply to all other active substances until they have been reviewed at European Community level for possible listing on Annex I. The details of this system are described later in Section 4 of this chapter. Under the national and European systems, a range of levels of approval is recognised, as described in Box 1.

28

A. Craven et al.

Experimental approval Some of the scientific data that are needed to support the commercial use of a pesticide (e.g. on its efficacy and on the levels of residues in treated crops) can only be generated by use in ‘real life’ situations. Experimental approvals allow for development work on new pesticides (or new uses of existing pesticides) to be carried out on a limited scale so that data can be produced in support of a future approval for commercial usage. Experimental approval is usually given for a limited period. Provisional approval Provisional approval allows commercial use of a pesticide (including its sale) for a stipulated period whilst it is being evaluated under the European system by MS, or whilst specific scientific data are being generated. Full approval Full approval is usually granted for a period of 10 years where all the data requirements necessary to support a pesticide’s use have been met. Such approvals may, however, be reviewed at any time if new evidence calls into question their safety or new data requirements are thought appropriate. Emergency approval In special circumstances an emergency approval may be given (for a limited period) for the sale, supply and limited and controlled use of an otherwise unapproved product. An emergency approval will only be considered where an unforeseeable pest, disease or weed problem has arisen that cannot be contained by other means.

Box 1. Levels of approval for active substances under the national and European review programme.

2.2. Maximum residue levels Pesticide residue levels in food are controlled at national and European levels by specification of many thousands of MRLs for different crops and commodities. Further MRL are planned to implement a European programme to harmonise these across all EU MS. MRLs are defined as the maximum concentration of pesticide residue (expressed as milligrams of residue per kilogram of food/feeding stuff ) likely to occur in or on food and feeding stuffs after the use of pesticides according to good agricultural practice (GAP). MRLs are intended primarily as a check that GAP is being followed and to assist international trade in treated produce. MRLs are not safety limits, and exposure to residues at levels exceeding MRLs does not necessarily imply a risk to health. Health risks are covered under 91/414/EEC (see below). 3. THE SCIENTIFIC ASSESSMENT OF ACTIVE SUBSTANCES With the establishment of the EFSA, risk assessment was separated from risk management. In compliance with this requirement, EFSA is responsible with MS for the risk assessment of active substances contained in PPPs which includes existing active substances already on the market in July 1993 as well as new active

Regulatory Aspects of Pesticide Risk Assessment

29

substances submitted by applicants. For the risk assessment of existing pesticides and following notification procedures to establish those substances with sufficient support for continued use, Commission Regulations (EC) No. 451/2000, No. 1490/2002 and No. 2229/2004 assign to EFSA the task of conducting risk assessments for the remaining substances of stages 2 (52 substances), 3 (144 substances) and 4 (249 substances) of the EU review programme for active substances contained in PPP. 3.1. Risk assessment for new active substances The responsibility for demonstrating efficacy and safety rests with the company intending to market the PPP. To this end, companies seeking provisional or full approval for a new product are required to submit an extensive package of scientific data, satisfying the data requirements of Directive 91/414/EEC. The components of the package vary to some extent according to the nature of the pesticide and the uses to which it will be put, but as far as possible they are standardised. Such standardisation helps to ensure that important aspects of risk assessment are not overlooked, and also that products are assessed consistently and equitably. The main components of the data package that typically would be required for a new pesticide fall into the following seven areas. 3.1.1. Physico-chemical properties

The applicant is required to specify the chemical composition of the product, its active substance and any significant impurities that it may contain. Information must also be supplied on the physico-chemical properties of the active substance (e.g. solubility and vapour pressure), and on methods by which it can be detected and measured, for example in foodstuffs and water. 3.1.2. Potential toxicity in humans

Data on potential toxicity are required for the active substance, the representative product as a whole, and also any important metabolites of the active substance to which humans might be exposed. These data are derived largely from tests in laboratory animals, and care is taken to ensure that all use of laboratory animals is the minimum strictly necessary. If reliable information can be obtained by other means, these are used in preference. An important objective of the toxicological assessment is to establish ‘No observed adverse effect levels’ (NOAELs) for any adverse effects that might occur. A NOAEL is the highest dose in an investigation that does not cause ill-effects. The data that are required to assess potential human toxicity cover: ●

How the active substance is metabolised, absorbed, distributed and excreted in mammals.

30 ●

●

●

●

●

●

●

●

●

A. Craven et al.

The acute toxicity of a single high dose of the active substance and of the product by oral, dermal and inhalation exposure. The sub-acute and chronic toxicity of the active substance when administered to animals over periods of several weeks or longer. The potential of the active substance to cause cancer when it is administered over a lifetime. The genotoxicity of the active substance – i.e. its potential to damage the genetic material in cells. The developmental toxicity of the active substance – i.e. whether it can cause foetal death or malformations when administered to female animals during pregnancy. The toxicity of the active substance when it is administered to at least two successive generations of animals over the course of their lifetime. The potential of the active substance and product to irritate the skin or eyes. The potential of the active substance and product to cause allergies (sensitisation). Further tests may be required if there is a need to understand effects better, for example on particular organ systems such as the nervous, immune or endocrine systems.

On the basis of these data, a decision is made as to whether the product requires labelling as a hazard (e.g. irritant, harmful, toxic). In addition, acceptable levels of exposure or reference doses may be defined as described in Box 2.

Acceptable daily intake (ADI) This is the amount of a chemical which can be consumed every day for a lifetime in the practical certainty, on the basis of all known facts, that no harm will result. It is expressed in milligrams of the chemical per kilogram bodyweight of the consumer per day. The starting point for the derivation of the ADI is usually the relevant NOAEL that has been observed in animal studies of toxicity. This is then divided by an uncertainty factor (most often 100) to allow for the possibility that animals may be less sensitive than humans and also to account for possible variation in sensitivity between individuals. The studies from which NOAELs and hence ADIs are derived take into account any impurities in the pesticide active substance as manufactured, and also any toxic breakdown products of the pesticide. Acute reference dose (ARfD) The definition of the ARf D is similar to that of the ADI, but it relates to the amount of a chemical that can be taken in at one meal or on one day. It is normally derived by applying an appropriate uncertainty factor to the relevant NOAEL in studies that assess acute toxicity or developmental toxicity. Acceptable operator exposure level (AOEL) This is intended to define a level of daily exposure that would not cause adverse effects in operators who work with a pesticide regularly over a period of days, weeks or months. Depending on the pattern of usage of the pesticide, it may be appropriate to define a short-term AOEL (i.e. for exposures over several weeks or on a seasonal basis), long-term AOEL (i.e. for repeated exposures over the course of a year) or both. AOELs are normally derived in a manner analogous to the ADI.

Box 2. Definitions of levels of exposure of pesticides in food.

Regulatory Aspects of Pesticide Risk Assessment

31

3.1.3. Dietary intake and risk assessment

One of the ways by which a pesticide might cause harm to humans is through its presence as a residue in food. A pesticide can only be approved if the potential exposure of consumers to its residues in food results in a residue intake that is within the ADI and (where relevant) the ARf D for that pesticide. A particular concern is the potential for residues in the food derived directly from any crops to which it is applied, but residues may also occur in other foods by indirect routes. For example, they might arise in the meat of animals that have been fed on a treated crop. Furthermore, the particular product that is being evaluated may not be the only source of the pesticide in the diet. The same chemical may also be a constituent of other products that are already on the market in the EU or in other countries from which we import food. In assessing the risks from residues of a pesticide in foods, therefore, it is necessary to identify and take account of all foodstuffs in which significant residues might occur, including those resulting from the use of other products that contain the same active substance. To check whether the proposed use of a pesticide might cause unacceptable long-term dietary exposures, an estimate is made of the maximum intake that an individual would be expected to incur over a prolonged period. This is based on the distribution of measured residues of the pesticide in foods derived (directly or indirectly) from treated crops, and data on the national patterns of consumption for different foods from official government surveys. The effect of any overestimation of potential dietary intakes is to err on the side of safety. The long-term dietary exposure to a pesticide, calculated in this way, is compared with the ADI, which is set as described in Box 2. If the ADI is exceeded, the proposed use of the pesticide will not be acceptable. 3.1.4. Exposures and risk assessment for operators, other workers and bystanders

The other circumstance in which human exposure to pesticides commonly occurs is in the course of their application or through contact with crops or other materials that have been treated with them. For example, an operator might be exposed when mixing or applying a pesticide; a passer-by or neighbour might be exposed inadvertently to droplets that drift when a pesticide is being sprayed; and a worker harvesting a crop that has been treated might handle foliage that is coated with residues of a pesticide. Most often, the uptake of pesticides from such sources will be by absorption through the skin (dermal absorption), but exposure by inhalation is also possible when products are sprayed or the active substance vaporises. Estimating the profile of exposure in operators, other workers and bystanders is complex and must take into account many factors. These include: ● ●

●

the physical form of the pesticide (e.g. liquid or granules); the way in which it is used (e.g. sprayed with a tractor mounted sprayer or painted with a brush); the circumstances in which exposure occurs (e.g. during mixing and application or through contact with a treated surface);

32 ● ● ●

A. Craven et al.

the use of any personal protective equipment such as gloves or a face mask; the extent to which the pesticide penetrates the skin; patterns of use (including frequency and duration).

Sometimes it is necessary to carry out field studies to establish the likely extent of exposures. Often, however, a first step is to estimate the exposure by use of a mathematical model. Models have been constructed for various exposure scenarios such as the tractor-mounted spraying of cereal crops and the application of pesticidal surface coatings by brush. The models are based on representative measurements, and tend to err on the side of over-estimating exposures. For example, in the absence of data to the contrary, they assume that a relatively high proportion of the pesticide is absorbed when it is in contact with the skin. Once an exposure has been estimated, it is compared with the AOEL described above, or with the NOAELs from relevant toxicity studies. If the AOEL is not exceeded or the ratio of NOAELs to the estimated exposure is sufficiently high, the exposure is considered acceptable. Where the exposure, when estimated by a model, is greater than the AOEL, it may be possible to refine the calculation through use of additional experimental data. For example, experimental assessment of skin absorption might show that it is lower than the relatively high rate assumed by the model. However, the use of a new product cannot be approved until exposure is shown to be below the AOEL. 3.1.5. Environmental fate and behaviour

In order to assess the potential impact of a pesticide on the environment, it is necessary to establish what happens to it once it has been applied – where it gets to; how fast it is degraded and by what mechanisms; and whether any of its degradation products might occur at levels sufficient to pose a risk. In particular, information is needed about the concentrations of the pesticide and any relevant breakdown products that will occur in soil, water and air, and the persistence of such pollution. Predicted environmental concentrations (PECs) are derived, and are used to assess: ● ● ● ●

exposure of non-target species in soil and water; possible contamination of drinking water supplies; possible contamination of groundwater and other natural waters; the potential for effects on, or residues in, following crops.

The distribution and breakdown of pesticides in the environment depends on many factors including the physical and chemical properties of the pesticide, the climatic conditions following use and the pattern of usage. The rate of breakdown of a pesticide is usually summarised by a half-life value, which represents the time it takes for half of the pesticide amount to degrade. The ease with which a pesticide can be washed out of the soil into other environmental compartments like groundwater, is usually termed its ‘mobility’ and a general impression of this

Regulatory Aspects of Pesticide Risk Assessment

33

can be gained from a Koc value (organic carbon soil sorption coefficient), which gives a measure of how well the pesticide adsorbs (sticks) to soil. The mobility and degradation of a specific pesticide can vary in different soils and can also be influenced by rainfall and temperature. The application rate, frequency of application and overall pattern of usage can all affect the concentrations of the pesticide present in the various environmental compartments of soil, groundwater, surface water, aquatic sediment and air, and must be taken into account. 3.1.6. Ecotoxicology and environmental risk assessment

The other major determinant of a pesticide’s environmental impact is its toxicity to wildlife or non-target species. The environmental risk assessment focuses upon possible effects of the pesticide on non-target organisms including: birds, wild mammals, fish, aquatic invertebrates and plants, insects (including bees) and other non-target arthropods, earthworms and soil micro-organisms. The risk to non-target organisms is generally assessed by the use of standard laboratory tests that give an indication of the toxicity of a particular active substance or formulation. These are used to derive LD50/LC50 values (the doses/concentrations at which 50% mortality occurs in an acute toxicity study) and NOEL/NOEC values (the levels or concentrations at which no effect is observed in longer term studies). The LD50/LC50 and NOEL/NOEC values are set alongside information on the PEC of a pesticide. From this an estimate of the relationship between its toxicity and the likely exposure is obtained. This is compared with internationally agreed ‘trigger values’ used by the European Commission to determine whether the risk is ‘acceptable’. If an ‘acceptable’ relationship between toxicity and likely exposure cannot be demonstrated at the initial assessment further higher tier data may be required or risk management strategies may be considered to reduce exposure to an acceptable level. These might include, for example introduction of a buffer zone adjacent to watercourses. 3.1.7. Efficacy and risk to following crops

Consideration of product efficacy is an integral part of the risk assessment process. At the present time in the EU, product efficacy is only considered at national MS level, rather than at EU level. Approval of a pesticide is only recommended if there are discernible benefits from the application of that pesticide. Data must be available to demonstrate the efficacy of the pesticide against target organisms when it is used in accordance with the label instructions. Data are now also required to demonstrate that the dose recommended is the minimum necessary to achieve the desired effect. In addition, the application of pesticides (especially herbicides) to a crop may pose a risk to the crop itself or to adjacent or following crops. This risk is assessed through the evaluation of both laboratory and field trial data to ensure that there are no unacceptable effects from the permitted uses.

34

A. Craven et al.

3.2. Reviews of pesticides that are already approved 3.2.1. The precautionary principle in pesticide regulation

Scientific risk assessment always entails an element of uncertainty that must be taken into account when managing risks. In recognition of this requirement, the 1992 Rio Conference on the Environment and Development set out the ‘Precautionary Principle’ that ‘Where there are threats of serious or irreversible damage, lack of full scientific certainty should not be posed as a reason for postponing cost effective measures to prevent environmental degradation’. Subsequently, the principle has been extended to cover threats to health as well as to the environment, and has been widely embraced by governments and agencies internationally. Effectively, the precautionary principle is applied as routine in the regulation of new pesticides. As has already been explained, new products are not authorised unless there is adequate scientific evidence that their use will not pose unacceptable risks to health and to the environment. In other words, uncertainties in the risk assessment must have been sufficiently eliminated before a new product is approved. Despite the precautionary approach taken with risk assessment, a measure of uncertainty remains and science can never give a cast-iron guarantee of zero risk. Therefore, once a pesticide has been approved, it is important to look for any unexpected adverse effects that were not predicted by the initial risk assessment. The position when reviewing pesticides that are already on the market is more complex, since there may be risks associated with withdrawal as well as with continued use. For example, loss of an agent for the control of cockroaches might threaten health through its impact on hygiene in buildings. In this situation, it could be argued that the precautionary stance would be to maintain the status quo and continue approval. 3.2.2. Checking for possible adverse effects of pesticide usage

Several systems are in place in the EU MS to monitor pesticide usage and the occurrence of possible adverse effects. ●

Notification by approval holders

Companies holding approval for pesticides are required by law to notify the registration authorities if at any stage they become aware of new data or information that raise concerns about the safety of their product. When a notification is received, it is evaluated, and the need for regulatory action or the generation of further data to clarify the problem is determined. ●

Food residues

Extensive monitoring is carried out of both MS-produced and imported food for pesticide residues. The purpose of this monitoring is threefold: to back up the statutory approvals process for pesticides by checking that no unexpected residues

Regulatory Aspects of Pesticide Risk Assessment

35

are occurring; to check that residues do not exceed statutory MRLs (see above); and to check that human dietary intakes of residues are within acceptable levels. ●

Pesticides in water

Many environment agencies in the MS monitor for pesticides in environmental waters, sediment and aquatic life. This programme is largely dependent on requirements set out in a number of European Directives. Additional non-statutory monitoring tailored to local pesticide usage patterns is also undertaken. Any occurrences of aquatic pollution incidents caused by pesticides are used to inform the risk assessment process for product approval and review.

3.2.3. Re-approval of a pesticide

Where information from one of the sources above suggests a previously unsuspected hazard it may be necessary to review the risk assessment for the pesticide. In addition, reviews may be carried out even where no new adverse data have been reported. Over the years the regulation of pesticides has become progressively more precautionary and we now seek much more evidence that human exposures and environmental impacts will be acceptable than was customary in the past. It follows that the scientific data supporting older products are often less extensive than would now be required for a new approval, and it is therefore necessary to bring the data packages up to modern standards. There are many older pesticides on the market, and each review is timeconsuming. Therefore the programme for reviewing older pesticides must be spread over a number of years, and the order in which products are considered must be prioritised. This ordering is based on various criteria, including the level of scientific concern about possible risks and the extent to which the product is used. Sometimes, one aspect of the risk assessment, such as the potential toxicity to humans, is a special concern and is examined ahead of other elements. The scientific data sought at reviews are similar to those used in the evaluation of new pesticides, but in addition epidemiological and other data on human health effects are assessed when available. For example, there might be reports of skin sensitisation in those handling the product, even though this did not occur when it was tested on animals. Also, once a pesticide is used on a commercial scale, large numbers of people are exposed to it, and epidemiological studies of their health may be possible. Such studies can provide direct evidence of any risks in humans, without the need for extrapolation from observations in animals. Where the risk assessment fails to provide adequate reassurance that human exposures and risks to wildlife are acceptable, it may be necessary to restrict the use of the product in order to overcome the problem, or if this is not possible, to revoke its approval. Sometimes when new data are first generated to bring the scientific information about a product up to modern standards, they do not provide the level of reassurance about safety that is now required, but nor do they clearly indicate that use of the product is unacceptable. For example, initial application of an

36

A. Craven et al.

operator exposure model may not demonstrate that exposures are acceptably low, but it is possible that the acceptability of exposures could be established through the generation of additional experimental data (e.g. on skin absorption rates), if these showed that the model was erring unnecessarily on the side of safety. New laboratory studies may have to be carried out additional to those that were evaluated when approval was granted. In these circumstances it may be judged acceptable for usage to continue while the additional data are produced to refine the risk assessment, and a decision on this is made on a case-by-case basis. 3.2.4. Revocation of a pesticide approval

Sometimes the company that holds an approval for a product that requires review may make a commercial decision not to invest in generating the additional scientific data that would be needed by today’s standards to support continued use. It may be, for example that the product has only minor uses and generates little profit. In these circumstances the approval for the product is withdrawn and after time it is revoked. Where revocation of an approval occurs, the time-scale for removal of the product from the supply chain will depend on the reason that it is being withdrawn. In most cases, the indication for withdrawal is a failure to provide adequate reassurance of safety rather than positive evidence that adverse effects are occurring. In this situation, the safest way to dispose of stocks that are already in the supply chain is usually through their use in the normal manner. Where there are no obvious safety concerns or the product is being withdrawn for commercial reasons, the approval holder will usually be allowed to market the product for a stated period (e.g. 1 year). A further period may then be allowed for persons other than the approval holder to sell, supply or use the product. This approach also has the advantage that users may have time to develop alternative methods for controlling the target pest. If, however, revocation is required because a product is shown to have harmful effects, a more rapid elimination from the supply chain may be necessary.

4. THE PROCESS OF PESTICIDE APPROVAL AND REGISTRATION 4.1. The approvals process Around 850 active substances used in PPPs were on the EU market in 1993. In order to comply with Council Directive 91/414/EEC concerning the placing on the market of PPPs, risk assessments for these substances should be completed by 2008. The work of registering PPP in Europe is carried out by the European Commission with the support of the (currently) 27 MS under the coordination of EFSA. The various MS pesticide registration authorities play an important role in Europe. They represent the MS in the European process for legislation of new PPPs and for the review of existing ones. The MS authorities also act as the regulatory bodies for national approvals which follow from EU decisions.

Regulatory Aspects of Pesticide Risk Assessment

37

Once most or all MS are content with the draft proposal, it is moved to the meeting of the Working Group on Pesticide Legislation for discussion. This group is more political and one where 27 sets of national interests come into play. The Commission in practice tries to ensure that there is agreement to the proposal at this stage before it is put to the SCFAH for an ‘opinion’. Armed with this, and with input from advisory committees, the Commission makes the final decision. For the evaluation of the 89 active substances of the first stage of the EU Review Programme, a peer review based on consultation of experts of the MS was developed. This peer review procedure and its outcome have widely been accepted by the MS. However, the peer review as it is set up for the first stage substances, where all comments received are dealt with, would overload the system considering the workload and timeframe of the second stage. For the peer review of the active substances of the second stage, the newly developed procedure includes the following basic elements (Figure 1): ● ●

● ●

DARs are distributed to all MS and the notifier company(ies); A period for all MS, the notifier company(ies) and other public interested parties to comment on the DARs; Evaluation of the comments received in time by the Rapporteur MS; Discussion of RMS evaluation and consideration of possible options to proceed by the EFSA with MS during an evaluation meeting: – rapid finalisation of the risk assessment without further detailed consultation for substances without problems identified (straight ‘green light’ fast track); – rapid finalisation of the risk assessment for substances without problems identified but with the need for minor bilateral consultations (lagged ‘green light’ fast track); – rapid finalisation of the risk assessment without further detailed consultation for substances with high risk or major essential data gaps (‘red light’ fast track); – further evaluation in expert meetings for substances with major problems identified, but focussing on the issues of concern (‘yellow light’ track).

The outcome would then be considered at further evaluation meetings to finalise the EFSA conclusion on the peer review.

4.2. EU organisations involved with pesticide approval 4.2.1. The European Commission

The European Commission is one of the main investigators in the preparation, formulation, implementation and monitoring of binding decisions taken by the EU. It presents proposals to the European Council and the Parliament. The Council and Parliament can only enact legislation on the basis of a proposal from the Commission.

38

A. Craven et al. Procedure of peer review of active substances Draft Assessment Report (DAR) RMS Updated dossier

Basic check of files

Preparatory phase

Phase 1

MSs/

1 wk

EFSA

public version

Draft Assessment Report (DAR) Comments of MS/EFSA/appl.

Public comments

min. 11 wks

Evaluation by RMS/(Co-RMS)

Phase 2

6 wks Rep. table / consultation rep. 25 MS (+ applicant)

Written procedure

Phase 3

EFSA

No data gaps, no open points RMS + MS

7 wks

Evaluation of comments by RMS / Co-RMS / MS / EFSA / notifier (rep. table)

Phase 4

25 MS

Completion of eval table by RMS * PPR Panel

12 wks High risk, major data gaps

ExpertMeetings

8 wks COM

Preparation of draft EFSA conclusion

2-3 wks

RMS comments

1-2 wks

Evaluation meeting (discussion of EFSA conclusion) EFSA conclusion

(*) Notifier may submit further requested data to RMS

4 wks

COM

1-2 wks Rev. 15 June 2006

Fig. 1. Procedure of the peer review of active substances of the second stage of the EU Review programme. Source: EFSA. (See Colour Plate Section, page 253.)

4.2.2. The Council of the EU

The Council is the organ which represents the MS at EU level. It has executive powers, makes decisions and performs the role of a legislative chamber. Among other things, the Council is consulted on new legislation, for example establishing a new scheme or regime (Directive 91/414/EEC on the authorisation of PPPs went through Council procedure), or where amendments are being proposed to the main legislation. Since the Amsterdam Treaty came into force in 1999, the

Regulatory Aspects of Pesticide Risk Assessment

39

Council and the European Parliament jointly decide on legislation (the ‘codecision’ procedure). 4.2.3. The European Parliament

The European Parliament is the parliamentary institution of the EU, directly elected since 1979. Its role was originally limited to advising the Council and to monitoring the activities of the Commission. More recently its powers, influence and involvement in a wide range of EU business have been increased. 4.2.4. The European Food Safety Authority (EFSA)

The EFSA is the keystone of EU risk assessment regarding food and feed safety. Whereas the task of evaluating the PPP dossiers is still distributed among the competent authorities of the MS, the EFSA was assigned the task of organising the peer review of the DARs prepared by the MS before they are submitted to the SCFAH. In close collaboration with national authorities and in open consultation with its stakeholders, EFSA provides independent scientific advice and clear communication on existing and emerging risks. 4.2.5. Regulatory Committees

Proposals are drawn up in specialised committees and working groups which are made up of MS, independent experts, pressure groups or Commission officials. Then in the case of pesticides legislation, regulatory committees consider them. Regulatory committees comprise representatives of the MS dealing with a specific area. They give their opinion on decisions about the regulations to apply in general areas such as food law, common veterinary or plant health standards etc. The regulatory committee reporting on pesticide matters is the SCFAH. The SCFAH considers data for approval of new PPP active substances and for the review of existing PPP active substances. 4.2.6. Plant Protection Products Working Groups

The proposals on active substances in the PPPs regime are based on reports from EFSA which in turn are informed by expert consideration of evaluations by MS at EFSA peer review co-ordination meetings (former EPCO meetings). The Evaluation Working Group is largely a technical meeting and all MS are represented. 4.3. MS organisations involved with pesticide approval 4.3.1. MS government departments

Responsibility for the approval of pesticides in most MS rests with the relevant departments for environment, food, rural affairs, health, agriculture etc., as appropriate according to how the government is organised in each MS.

40

A. Craven et al.

4.3.2. MS pesticide registration authorities

In each MS there are competent authorities responsible for the registration of pesticides: these are generally an agency of a government department, and are usually charged with the co-ordination and administration of the pesticide regulatory process in the respective MS. In addition to the evaluation and processing of applications for approval of pesticides, the various registration authorities may also be responsible for advising government ministers on the development and enforcement of pesticide policy and legislation, and advising respective ministers on all aspects of pesticides approvals policy. This may include data protection, public access to information and better regulation. 4.3.3. MS expert advisory committees

Some MS have independent committees, which provide advice in such areas as (a) regulations which government ministers contemplate making; (b) approvals which ministers contemplate giving, revoking or suspending; and (c) conditions to which ministers contemplate making approvals subject. These independent expert committees can also contribute to discussions on continuous development of means to protect the health of humans, creatures and plants, safeguarding the environment, securing safe, efficient and humane methods of controlling pests and making information about pesticides available to the public. Most of the members of such committees are selected to provide expertise in scientific disciplines relevant to the evaluation of pesticides. 5. CURRENT DEVELOPMENTS IN THE PESTICIDE REGULATORY PROCESS IN THE EU 5.1. Technological advances One obvious area of development is in the pesticides themselves. In recent years a number of ‘lower risk’ products have been developed using a range of technologies, such as: ● ●

●

chemicals that change insect behaviour, such as pheromones; microbial pesticides, such as insecticidal bacteria (especially Bacillus thuringiensis), viruses and fungi; preparations of nematode worms.

Such products are attractive in that they may appear to offer a more natural and targeted method of pest control. Use of such products will not necessarily lead to complete control of the pest in question, but will help to ensure that the population of the pest does not lead to excessive loss of agricultural yield. The mode of application will in some cases be via traditional sprayers, but in other

Regulatory Aspects of Pesticide Risk Assessment

41

cases will be released into the atmosphere by a controlled delivery mechanism. Their effectiveness may be more dependent on ecological and management factors than traditional chemical control. There is often very little prospect of the product reaching either surface water or groundwater, and so theoretically it should be unnecessary to perform a full assessment of the environmental fate and behaviour. However, there may be concerns that are not present for chemical control, for example the products might carry a risk of infection to people or animals. Current methods of assessing the toxicity of chemical pesticides are not entirely suitable for this type of product, and special risk assessments are therefore required. The technological advances are currently ahead of the regulatory advances, and ‘low-risk’ preparations are subject to broadly the same data-heavy assessments as chemical pesticides in the EU. Advances also occur in the ways in which pesticides are formulated, packaged and applied. In general these have the effect of reducing the exposure of operators or levels of environmental contamination, and as they become established, they allow tighter controls to be applied to existing as well as new products. A further area of development is in the way in which pesticides are used. Currently experimental work is being carried out to assess the use of herbicides on genetically modified herbicide tolerant crops as a means of weed control. In theory this might have ecological benefits by reducing the frequency with which herbicides are applied to the crops and perhaps allowing untreated ‘islands’ to be left in a field without a risk of uncontrollable spread of weeds to other parts of the crop. However, in assessing whether this type of herbicide usage will be acceptable on a large scale, it will be important to establish what effects it has on wildlife in practice. 5.2. Scientific advances in risk assessment Changes in the way in which pesticides are regulated may also arise from improvements in the science of risk assessment. Surveillance programmes have detected residues of more than one pesticide in the same food sample. During the approval of pesticides, active substances are normally assessed singly for their potential impact on human health. However, there are families of pesticides that work toxicologically through the same mechanism and, hence, it is possible that interactions between substances may result in a greater toxic effect than predicted during the approval process. Consumer groups have been concerned for some time about the possible implications of interactions between the components of mixtures of chemicals. As a response to such concerns, we can critically review what is known about the science of mixtures and consider the implications for the risk assessment process. Risk assessment is developing in the environmental area too. One interesting special case is the issue of pest control in rice. Products used on rice have a much higher chance of contaminating water than products used on other arable fields. The nature of the arable environment is more like a wide ditch than a field for substantial periods of the year. The drainage systems used in rice production

42

A. Craven et al.

would allow contaminated water to flow directly into streams if not managed correctly, and the saturated conditions can lead to leaching of pesticides to ground water. Therefore, to approve a PPP for use on rice, it is important to have information on the timing of use in relation to the hydrological management of the paddy, and the behaviour of the compound within the paddy environment. This might require the use of specialised models that describe the paddy system, or additional experimental work. Another area of increasing concern is biodiversity. Methods to predict the indirect effects of pesticides on terrestrial and aquatic biodiversity are improving, through the use of monitoring programmes and food chain modelling. However, it is not straightforward to include this aspect in the approvals process, as the effects may be due to the combination of pesticides applied over the course of a season, which is outside the scope of current risk assessment. Probabilistic risk assessment is likely to form a large part of the risk assessment process in future, and there are substantial research efforts to support this (e.g. EUFRAM, www.eufram.com). Such methods rely on distributions of data, and aim to account for natural variability in exposure and effects of pesticides, and uncertainty in our knowledge. In principle, these methods are applicable to all areas of risk assessment, both human and environmental, and the results would be presented as the probability that an ecological or health effect would occur. 5.3. Public opinion and expectations The regulation of pesticides depends not only on scientific assessment of their efficacy and of any risks that they might pose, but also on the acceptability of the risks that are identified. Judgements about the acceptability of risks are not simply a matter of science and require broader public input. Within the regulatory system this is achieved most obviously through the part played by ministers of the elected government. Ultimately it is the elected government that has the final say in which pesticides should be approved and with what conditions. Public attitudes to environmental hazards have changed over the years and the regulatory system must respond to these changes. One requirement has been for greater openness of the decision-making process. In the past there was a tendency not to publicise risk assessments widely for fear that they would be misunderstood and generate unnecessary anxiety. Now, however, the public are likely to be worried more by a failure to disclose information, fearing that a problem is being covered up. Over recent years, there has also been growing pressure for more precautionary approaches to the management of environmental hazards. Many people worry about the pace of scientific and technological developments and their potential to do harm as well as good. They therefore call for positive evidence that new technologies will be acceptably safe rather than simply assuming adequate safety in the absence of evidence to the contrary. As described above, the regulation of pesticides is now more precautionary than in the past, and it is likely that this trend will continue.

Regulatory Aspects of Pesticide Risk Assessment

43

5.4. Administrative and legislative changes The other major factor that is altering the EU system for regulation of pesticides is the movement towards greater co-ordination within the European Community. As described above, an important part of the risk assessment for pesticides used as PPPs is now being undertaken at a European level, and legislation is now in place which establishes a parallel scheme for non-agricultural pesticides (biocidal products). Other chemicals (including non-active ingredients of pesticide formulations) will be assessed at the European level under the REACH legislation (Registration, Evaluation and Authorisation and Restriction of Chemicals, EC Regulation 1907/2006). The Water Framework Directive will increasingly affect pesticide risk assessment through the goal of compliance with Environmental Quality Standards at the river basin scale as compared with the ‘field-edge’ assessments considered under 91/414/EEC. Finally, a major revision of 91/414/EEC is being planned, in which a major addition will be the regulation of pesticide usage, which is not currently covered by 91/414/EEC. REFERENCES Commission Regulations (EC) No. 451/2000, No. 1490/2002 and No. 2229/2004 on conducting the risk assessments for the PPP active substances of Stages 2, 3 and 4 respectively, of the EU Review Programme. Commission Regulations (EC) No. 396/2005 on maximum residue levels of pesticides in or on food and feed of plant and animal origin. Commission Regulations (EC) No. 1907/2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) and establishing a European Chemicals Agency. EU Directive 91/414/EEC, concerning the placing of plant protection products on the market, and EU Directive 97/57/EEC, establishing Annex VI (Uniform Principles for decision making) to Directive 91/414. EU Directive 98/8/EC concerning the placing of biocidal products on the market. EU Directive 2000/60/EC of the European Parliament and of the Council establishing a framework for the Community action in the field of water policy.

253

Colour Plate Section Procedure of peer review of active substances

• • • MSsl EFSA

Preparatory phase

1 wk

Phase 1

min. 11 wks Phase 2 6wks

25 MS (+ applicant)

Phase 3

Written procedure

7wks

EFSA

12 wks

RMS

8 wks

+ MS

Phase 4

2-3 wks

1-2wks

25 MS

Evaluation meeting (discussion of EFSA conclusion)

4wks

1-2 wks (*) Notifier may submit further requested data to RMS

Rev. 15 June 2006

Plate 1. Procedure of the peer review of active substances of the second stage of the ED Review programme. Source: EFSA. (See also page 38 of this book.)

Pesticide Risk Assessment in Rice Paddies: Theory and Practice E. Capri and D.G. Karpouzas (editors) © 2008 Published by Elsevier B.V.

Chapter 3

Will Risk Assessment Help Risk Management? Mark H. M. M. Montforts Expertise Centre for Substances, National Institute for Public Health and the Environment, Bilthoven, the Netherlands Contents 1. Introduction 2. The knowledge transfer process 2.1. The playground and the rules 2.2. The players and the game 3. Management by methodology 3.1. The benchmark for the assessment: criteria 3.2. Uncertainty factors 3.3. Extrapolation 3.4. Testing and data handling 4. A gradient from science to management References

45 46 46 48 51 51 53 54 54 55 56

1. INTRODUCTION This chapter will introduce the concepts of risk management and risk assessment. I will use the risk assessor’s viewpoint to clarify this. Science is what risk assessors do. Regulation is what risk managers do. The work of the risk assessor helps the risk manager to make sound decisions. Both actors co-exist together in the bigger picture of risk analysis and regulation. They have different roles in the risk analysis process. Because of differences in views, backgrounds and interests, they may also have different expectations of what the other does and should do. There is also a general public with its own opinion about risk. These are very important facts that have an impact on the work of the risk assessor. Every risk assessor will most likely agree with the following assertion: risk assessment is a scientifically based process, and shall be based on the available scientific evidence and undertaken in an independent, objective and transparent manner. However, from the public perspective, science may not be regarded as impartial and objective, but as a tool that is only used to suggest that risks are absent, to delay the decision making process, or to distract attention from other political factors. The reason for this discrepancy is that risk perception (within societies) does not always operate via the same mechanisms as scientific risk analysis (within expert communities). Risk communication and the way risk assessment is implemented need to consider how risk is perceived by the audience. Risk assessors live with the fact that although science should be objective, in reality science may be observed as partial. This is most prone in fields where there is

46

M. H. M. M. Montforts

great uncertainty combined with the possibility of irreversible harm. Given these uncertainties, the contribution of science lies with elucidating this uncertainty much more than providing an opinion on likelihoods and consequences. Regulators can only tackle these communication problems through open participation of societal stakeholders aided by the transparent input of scientific data. Recent risk paradigms call for a participatory procedure, in which the different stakeholders are involved early in the risk analysis process (Wagner, 1995; De Marchi and Ravetz, 1999; Heyvaert, 1999b; Amendola, 2001; Chess, 2001; Jones, 2001; Slovic, 2004). Chapter 2 in this book already commented more specifically on the public awareness concerning pesticides. Risk assessors play an important role in the risk analysis and in the communication of risk, uncertainty and connected mitigation measures between managers and the public. The assessors need to realise that both risk managers and the public may misinterpret facts. Even when provided with high quality risk information, decision making in the regulatory context is complicated and often moves away from the scientific perspective. Once risk assessments have made it to the public, the use of the information with selective citations, or without contextual sources, may still lead to subjective interpretations (Vlek and Keren, 1992; Hassenzahl, 2006; IET, 2006; HoC, 2006a). The following questions are addressed in the following sections: – Where ends risk assessment and where starts risk management? – What are the perceptions of risk assessors and of risk managers about this connection? – What are value judgements within risk assessment? – How much science is needed by risk management in interpreting and considering risk assessment opinions? Not just in the field of food and pesticides, but in many fields risk assessment are thought to improve the policy making process. Examples will be given from the area of marine aquaculture, environmental impact assessment, human medicines, pesticides, chemicals, biocides, climate change, natural hazards, toxicology, identity cards, criminal law and veterinary medicines. 2. THE KNOWLEDGE TRANSFER PROCESS The borderline between risk assessment and risk management is shaped by the rules of procedure that are laid down in legislation, but also by the perceptions of both risk assessors and risk managers. 2.1. The playground and the rules The design of the regulatory context, together with the resources available to participants to actually employ science, do influence the regulatory process (Halffman, 2003). Basic principles of EU governance are based on the separation

Will Risk Assessment Help Risk Management?

47

between science and politics. The way the European Commission has organised the use of science for food, clearly deals with some earlier critiques on the use of risk assessment in the regulatory process of chemicals (Heyvaert, 1999b). Let us consider this organisation for pesticides. In the European Union, regulation and policy making for pesticides rely heavily on science. The scientific assessment of pesticides is part of the European Food Safety Agency (EFSA). The EFSA was legally established by a European Parliament and Council Regulation adopted on 28 January 2002 (Regulation EC No 178/2002). The regulation lays down the basic principles and requirements of food law and stipulates that EFSA is an independent source of scientific advice, information and risk communication in the area of food and feed safety. To be more specific, article 22 of this Regulation stipulates that pesticide authorisation is within the remit of EFSA, because it relates to plant health. The Regulation clearly defines different aspects of risk. ●

●

●

‘Risk analysis’ means a process consisting of three interconnected components: risk assessment, risk management and risk communication. ‘Risk assessment’ means a scientifically based process consisting of four steps: hazard identification, hazard characterisation, exposure assessment and risk characterisation. ‘Risk management’ means the process, distinct from risk assessment, of weighing policy alternatives in consultation with interested parties, considering risk assessment and other legitimate factors, and, if need be, selecting appropriate prevention and control options.

The Regulation defines that risk assessment is a scientifically based process, and demands in article 6 that it shall be based on the available scientific evidence and undertaken in an independent, objective and transparent manner. Three main tasks of EFSA (be aware that there are more) are: 1. To provide scientific opinions; 2. To promote and coordinate the development of uniform risk assessment methodologies; and 3. To provide support to the Commission, also on the interpretation and consideration of risk assessment opinions. As risk assessors, we can see that science should be objective and free from politics. The scientific opinion should be taken into account when coming to decisions and the assessor can help explain and understand the scientific opinion. While EFSA advises on risk, the responsibility for risk management remains with the Commission, supported by the Standing Committee on the Food Chain and Animal Health. This Standing Committee consists of the Member States. Considering the tasks of EFSA, i.e. providing opinions, and in particular of promoting and coordinating uniform risk assessment methodologies, there is a clear need for risk communication between risk assessors and risk managers. Still, the risk managers are allowed to act without scientific information. According to the European Court of Justice, pesticides are “per se” dangerous

48

M. H. M. M. Montforts

products,1 therefore the Commission does not always have to establish danger (scientifically proven) before taking action (Haibach, 1999; Heyvaert, 1999a). The Directive 91/414/EEC on placing on the market of plant protection products, sets the stage for the authorisation of pesticides (Anonymous, 1991). The reader is referred to Chapter 2 for more details.

2.2. The players and the game The perception of assessors and managers is increasingly important since legislation is generally prescriptive on the procedures, and seldom on the approaches or on how to make the connection between the measured, estimated and/or extrapolated results on the one hand, and a risk recommendation on the other (Heyvaert, 1999b). The way risk assessment and risk management are in reality organised is important. At EU level, assessors are separated from managers. In Member States and at the regional level, this may be different. Risk assessors and risk managers may be unified in one person. They may be different actors within one institute, or in different institutes. Lets find some clues how assessors and managers communicate, and what their expectations are. Knowledge transfer between science and management is crucial. To start with, scientists and regulators, hence assessors and managers, may have different risk perceptions. But let us take one step backwards. How many science-based judgements are possible among risk assessors? Depending on the institution, and its primary concern, the viewpoints on risks already differ between scientists. For example, in a study into perception of risk associated with salmon aquaculture, government based scientists perceive risks on average as higher, but the uncertainties involved as lower, than their counterparts who work in industry, consultancy or academia. Likewise, within the pharmaceutical field one will find different judgements on environmental risk priorities when travelling from the US to Europe, or when moving from industry to government. This phenomenon may well influence decisions made in qualitative statements on relevance of certain missing factors and sources of uncertainty, or on priorities for research (Doerr-MacEwen and Haight, 2006; McDaniels et al., 2006). In the field of natural hazards and disasters, research has identified four factors affecting the knowledge transfer process. It is interesting to see if and how these factors are also operational in other regulatory fields. The factors are: 1. 2. 3. 4.

1

culture, institutions, links, and interactions.

The concept of “per se” dangerous substances was introduced by the ECJ in a judgement concerning residues in foodstuffs: Case 94/83 Albert Heijn BV [1984] ECR 3263.

Will Risk Assessment Help Risk Management?

49

Culture: Science and policy are two distinct cultures. Researchers do not know what a regulator really does, and regulators do not understand the responsibilities of scientists. Second, the two cultures use different jargon and may use words, or conventions of expressing oneself, that are understood differently by the other party. Institutions: Even within the science realm, there are considerable variations on the ideas of knowledge transfer processes. Institutions refers to all the restraints, rewards and boundaries, that accompany work in a particular institution and system, and thus determine the way information is communicated, or how information can be assimilated. Culture and institutions further promote the creation of the third factor, that of links. Links: Links are all the entities that serve as linkages between the two groups, like people, projects, workshops and information centres. Both researchers and policy makers agree that information exchange between them is important, and both groups agree that there are too few people, or links, who exchange the information for them. Interactions: Finally, interpersonal interaction between scientists and regulators was identified as a factor that actually promotes knowledge transfer (Fothergill, 2000; Lehtonen and Peltonen, 2006). Illustrative of the expectations that risk managers have from science is the information provided by Dr Andrew Wadge2 on the question: “how do we interpret that science once we get the science?” (HoC, 2006b). For that very purpose a checklist of questions was drawn up (FSA, 2006). There are several questions in the list that specifically focus on aspects of science that need to be understood by the manager in order to value the merits of the science: 1. What are the facts underpinning the risk assessment? What are the assumptions? 2. Has an assessment been made of the likely impact and probability of occurrence? 3. Are all key scientific uncertainties highlighted? Has any indication been given about the degree of uncertainty or consensus involved? 4. Are significant gaps in the current evidence base noted? 5. How have the areas of uncertainty been handled when reaching final conclusions and how do they impact on the advice? No referral was made to possible decisions to be made within the assessment by the risk manager. It thus seems that science is not influenced by management. The impression is that science works according to its own judgements. The question is whether that is true, but also whether that would be justifiable. In many frameworks the assessor is bound to rules how to apply expert judgement, but at the same time he has a relative freedom in deciding if, and when to apply expert judgement (Hamilton et al., 2006). These questions indicate that the understanding of facts, assumptions, likelihood, uncertainties and the way these have been handled, determines the way the 2

Director of Food Safety Policy and Acting Chief Scientist, of the UK Food Standards Agency.

50

M. H. M. M. Montforts

result of the assessment should be handled by the risk manager. Looking at all these pieces of information one can see that there are limits to the knowledge transfer process. Hardly ever are all relevant factors accounted for through their full operating range, and hardly ever is all variability and uncertainty fully expressed. Very often the problem is concerned with discrete variables, which means that it is not possible to give the scientific opinion in terms of a holistic distribution of results following a gliding scale of assumptions and uncertainty. Instead, choices must be made. This means that the implications of basic choices and the meaning of missing information cannot be valued in the same way as the implication of a given probability on an enumerated effect. To solve this gap, the risk manager has no choice but to enter the risk assessment domain, by making choices. At the same time, the risk assessor will enter the risk management domain, where he has to explain the implications of these choices in such a way, that the risk manager can decide. At least, that is my theory. In practice, one may very well find that managers and assessors act differently. In a response to the UK House of Commons Science & Technology Committee’s Inquiry into Government handling of scientific advice and risk, one organisation expressed a concern that it is only possible for government to handle risk and science appropriately if it has a sufficiently expert and critical in-house capability to allow it to formulate the questions it needs to ask of external experts. Second, it is necessary to identify accurately the subjects on which research evidence is needed (CaSE, 2006). It is clear that there is an iterative process of risk identification and risk management that involves a co-operation between science and regulators. Risk management must make decisions on priorities and measures, and in both cases needs scientific input to make informed decisions. There is however one difficulty. Inherent of scientific research is that it may result in wrong outcomes. This is addressed by the type I, II and III errors. A type I error is a false positive; a type II error is a false negative. Type I and II errors are closely related. The further type I errors are avoided; the more likely it is that type II errors are made. Therefore, it is important to know, if causality must be demonstrated, or that presence of adverse effects is avoided. In the former case type I errors must be avoided, and in the latter case type II errors must be avoided. Type III errors are independent of type I and II errors. When the research does not answer to the question, one speaks of a type III error. The science may be correct, but the answer is not useful. Making sound decisions about managing risks necessarily involves relying on judgements by technical specialists informed by the best available scientific evidence. However, for example, on the topic of pharmaceuticals in the environment, an academic signalled the existence of type III errors (thus a focus on the wrong problems) within the scientific community (Doerr-MacEwen and Haight, 2006). Those regulatory processes that have a certain degree of repetition or continuity, and similarity between cases, may benefit from this feature. The process should give rise to opinions that can be easily applied without extensive and specific support on interpretation of each scientific opinion. A very efficient way to accomplish this is the use of risk assessment methodology and guidance documents.

Will Risk Assessment Help Risk Management?

51

3. MANAGEMENT BY METHODOLOGY Risk assessment methodologies support the transparency and uniformity of the scientific process, thus basic principles of good management (Read et al., 2001). The guidance that accompanies the methodology should give rise to opinions that can be easily applied without extensive and specific support on interpretation of each scientific opinion. In order to achieve this, the consideration of model design and of risk assessment options should be dealt with before and during the establishment of the risk assessment methodologies. Since the bulk of scientific opinions in the regulatory framework is routinely constructed with the aid of guidance documents and models, these risk assessment methodologies must excel in both the scientific quality and the regulatory applicability. However, whilst risk assessment methodology offers transparency and consistency, by consolidating scientific principles and consensus, it also departs from the cutting edge of science to its own routine. To deal with this drawback, mechanisms that monitor the state-of-art of science and intervene when changes are called for should be in place. The general principles of the risk assessment methodology for pesticides, and in particular for rice crop, have been described in Chapter 4. In the process of creating methodology, there are several moments where risk assessors need feedback from managers on value judgements. Perhaps because these judgements seem to be purely scientific, or perhaps due to a lack of links between assessors and regulators, in reality discretionary powers have been transferred to the scientists. We will look at what appears to be typically scientific challenges, but basically harbour value judgements: Protection goals: when should one be concerned; how to deal with extrapolation and with uncertainty; what is the problem we try to demonstrate; and what is the information we have actually telling us? 3.1. The benchmark for the assessment: criteria The risk assessment process needs to be helpful in decision making. To be helpful, it should be clear what is to be addressed, and what risk or impact is acceptable and what is not. The Directive 91/414/EEC on the placing on the market of plant protection products, explicitly describes a normative conclusion for the risk assessment in its Annex VI on criteria for decision making. If the toxicity-exposure-ratio (TER) ratio is above the specified triggers, risk assessors must conclude that the substance is of no concern, and consequently no further information can be required that could feed the scientific research. If the TER ratio is below the benchmark, then no authorisation is granted unless it is demonstrated that under field conditions the proposed use of the product causes no unacceptable impact. The legislation does not clarify what unacceptable is. In Table 1 for a number of pesticides the endpoint value used for the aquatic risk assessment is used. The data are from the public assessment reports of the Dutch Board for the Authorisation of Pesticides (http://www.ctb-wageningen.nl/) and the

52

M. H. M. M. Montforts

Table 1. Endpoints from aquatic (semi)field studies used for authorisation Substance Atrazin Azoxystrobin Carbendazim Chlorpyrifos Chlorothalonil Deltamethrin Esfenvaleraat Kresoxim-methyl Lambda-cyhalothrin Linuron Maneb Mancozeb Metamitron Metiram Metolachlor Metoxuron Metribuzin Terbuthylazin Thiacloprid Thiram Tolylfluanid Triazamate Trifloxystrobin aSF

Assessment EAC class 3 NOEAEC EAC class 1 NOEC EAC class 1–2 EAC class 3 EAC class 1–2 EAC class 2 SF 2a EAC class 2 EAC class 3 EAC class 3 EAC class 3 EAC class 2 EAC class 3 NOEC NOEC EAC class 2 EAC class 2 SF 2a EAC class 3 EAC class 3 Not derived; sensitive taxa absent Less reliable study Not explained

⫽ Safety factor.

information on the classifications used can be found in the Guidance Document on Aquatic Toxicology (ECCO, 2002). This guidance document actually suggests that effects that disappear within eight weeks, are deemed acceptable (Class 3 effects). However, to the opinion of the EFSA Panel on plant protection products and their residues, the way the risk assessment for single products is carried out, does not allow for integrating recovery periods into the decision making (EFSA PPR, 2006). When looking at the assessments that gave rise to these endpoints, we can see that no less than seven different criteria were applied: Environmental Acceptable Concentration (EAC) class 1, 2 or 3; or EAC class 2 with a safety factor of 2; a No Observed Effect Concentration (NOEC); and a No Observed Ecological Adverse Effect Concentration. The reasoning behind one value was not traceable. For another substance the fact that sensitive taxa were absent was a reason not to select the value. One should be aware that this reasoning also applies to other substances, like chlorothalonil, but was not made in practice. This lack of a definition of acceptability of effects is recognised as a problem by all stakeholders: industry, risk assessors and regulators. Because of this lack of definition in the legislation, it is unclear what – critical – effect values should be assessed in field studies. Despite the extensive documentation on field study

Will Risk Assessment Help Risk Management?

53

performance (see Chapter 5), the decision making is not based on publicly shared values, but on opinions of scientists in combination with the technical limitations of the test design (Montforts and De Jong, 2007). Not all product legislation specifies protection goals. The legislation on the authorisation of veterinary medicines merely talks about the environment as a protection goal (Anonymous, 2004). In those circumstances, all relevant environmental protection goals are to be addressed in the methodology. It appeared that this had been insufficiently explored in the drafting of the methodology for veterinary medicines, since not all relevant protection goals were targeted. The picture emerged that policy makers and scientists had not engaged themselves in a reconnaissance of regulatory goals, assessment scales and model approaches, and the decision were left to the scientists (Montforts and De Knecht, 2004). Also in the arena of ecological risk assessments in the US the same problem was identified earlier. Although the guidelines recommend assessment of several criteria, like the nature of the effects, intensity of the effects, spatial scale, temporal scale and potential for recovery, they do not provide specific standards for judging adversity (Henning and Shear, 1998). 3.2. Uncertainty factors The abovementioned TER values also provide a fine example of risk management decisions that are intertwined with risk assessment: the TER value itself (5, 10, 100) expresses a degree of uncertainty (in the exposure assessment or in the effect assessment, or in both) that apparently needs to be accounted for. The use of uncertainty factors, or safety factors, is not a strictly a regulatory, nor strictly a scientific issue. It has been noted that the risk assessment methodologies within different regulatory frameworks apply different factors where it essentially concerns identical risk models, the same substances and the protection of the same environment (SSC, 2003). From a scientific point of view there is no reason to choose diverging safety factors. For the risk manager there may be reasons to accept less certainty, or more risk, depending on the benefit of the products involved. Whether or not it is expedient to calculate this risk management decision in the assessment methodology, and whether or not this was the reason for the currently diverging assessment methodologies, remains to be seen. When we look at the risk assessment for earthworms at the authorisation for pesticides, then we can see that the assessment factor applied to the LC50 determined in a laboratory test is 10 (Anonymous, 1997). The analysis of laboratory LC50 results and field data made by Heimbach (1992) reveals that “medium” effects, or 40–75% effect on abundance of earthworms, can be expected when the LC50/10 is exceeded. Should the exposure in the agricultural soil be higher than this value of LC50/10, then for the higher tier assessment, the earthworm field study will not reveal statistically significant effects smaller than 50% (De Jong et al., 2006). Is this level of protection a deliberate choice because we are dealing with in-crop assessment? At the authorisation for biocides (where it often concerns the same substances), the assessment factor on the LC50 is 1000. Here

54

M. H. M. M. Montforts

the agricultural field is not the treated area, but the area to be protected from contamination. However, for pesticides, the endpoint for chronic exposure (NOEC) is also used, leading to an assessment that is almost (NOEC/5) as strict as for biocides (NOEC/10). It is therefore not straightforward that the choice for assessment factors has been guided by a clear notion of acceptable impact. 3.3. Extrapolation Methods to extrapolate, whether from high test results to low test results, or from models to society, have both scientific and political traits. Two examples are given here: one on extrapolation to levels-of-no-concern for carcinogens, and one on distribution models of species sensitivity. The choice of the extrapolation model for carcinogenity is characterised by the US EPA as a policy decision (Shere, 1995). One would expect that this regulatory decision has been based on scientific support for the models the administrator could choose from. Not only do different models yield different results, the validity of concepts underlying models may be challenged (Kitchin and Drane, 2005). In ecotoxicology, the concept of species-sensitivity distributions is applied to extrapolate to acceptable concentrations using data on multiple species. Wheeler et al. (2002) proposed that the selection of the best approach to generate these species-sensitivity distributions is to be based on the best fit on the data, especially in the lower tail, based on both the coefficient of regression and visual inspection. They also recommended two methods, based on data availability, thus allotting the choice of the extrapolation model to the scientist, primarily driven by empirical factors. Once such an approach would be applied for regulatory purposes, should then the risk manager not accept the proposed method first? In the field of coastal aquaculture, a discussion document states that high levels of uncertainty imply the need for an informed political, rather than purely scientific decision making process. A lack of guidance on the analysis and characterisation of uncertainty and variability is seen as a main weakness of risk assessment (Hambrey and Southall, 2002; Thompson, 2002). Recently, EFSA has published guidance on uncertainty analysis and communication in risk assessment (EFSA, 2006). 3.4. Testing and data handling Risk characterisation and effect assessment for chemicals are typically based on test data instead of epidemiological studies. Laboratory or field-testing methods contain decisions on design and analysis that harbour subjective decisions. The degree of certainty by which an effect is determined to be a result of an exposure, has to be set. Science typically strives at minimising false positives. Generally, the statistical significance of an effect has to meet the condition of p ⫽ 0.05, i.e. that the chance is less than 1 in 20 that an erroneously significant difference is established.

Will Risk Assessment Help Risk Management?

55

Depending on the test design and variability in the control groups, effects may not be seen as significant. Hence, the degree of statistical significance applied to a given test design, directly affects the outcome of the risk assessment. The reciprocals of false positives, i.e. the chance that a real effect may be overlooked, could however be of higher concern for the risk manager, allowing for higher p-values, e.g. p ⫽0.1. Also the statistical method chosen to do the analysis, or the specific endpoint derived from the study (for example, a NOEC of an ECx) will influence the results. Depending on how results of testing are used, there needs to be understanding between risk managers and risk assessors on the power of a test, and the chance on false positives and false negatives (Hoekstra, 1993; Heyvaert, 1999b; Sanderson and Petersen, 2002; De Jong et al., 2006). Another example of the importance of testing and data handling is the algal growth inhibition test. A parameter, called the EbC50 (the concentration that causes 50% reduction in biomass) was guided to be calculated by comparing absolute changes in cell counts instead of relative changes, compared to the starting conditions. Nyholm (1985) had already concluded that in this approach the absolute change depends on test-dependent factors besides the true growth rate. Tests with different inoculum densities will yield different EbC50 values for the same substance. The ISO guidance document on statistical analysis of ecotoxicological data ignores this way of assessing effect and deals with relative growth (ErC50) only (ISO, 2004)). This approach is now also recommended in the OECD Test Guideline (2006). Depending on the value judgement made when choosing the endpoints, the result of the risk assessment may have been different. Once the assessor has the data that were needed, there are still stages in the scientific process where the assessor determines the results. Pitfalls in the selection of data from public literature, leading to non-existing data, have been reported by Pontolillo and Eganhouse (2001). The selection of endpoints for a certain assessment from a database, and the subsequent assessment, will differ between assessors depending on the availability of guidance and the skills of the assessor (Montforts, 2006). The influence of the evaluator’s subjectivity on the selection of endpoints from raw data for use in exposure models has been demonstrated by Boesten (2000). These factors that cannot be controlled to the full extent, heavily influence the information that is provided to the risk manager. 4. A GRADIENT FROM SCIENCE TO MANAGEMENT Chapter 10 has illustrated the difficulties in assigning values to negative impacts on environmental and health assets in the broader analysis of the socio-economic aspects of rice cultivation, where the environmental risk assessment is but one aspect. Thus it becomes clear that a scientific assessment of risk and impact associated with pesticide use in rice crops, taking appropriate scales of assessment into account, and differentiating between in-crop and off-crop, and local assessment and regional assessment where relevant, offers choices for risk management. Above a collection of information taken from a wide range of research on environmental risk assessment, risk management and risk communication was

56

M. H. M. M. Montforts

presented. The chapter has addressed these intertwined aspects of risk analysis from different angles, thus offering a plethora of information on the perception of risk analysis, the responsibility of actors within the risk analysis process, and pitfalls in the management of the process. Building a robust and effective risk assessment methodology, along the lines presented in Chapter 5, depends on the science used, but also to a large extent on the way the scientists, the end-users (managers) and stakeholders (the public) are involved during the process. During the process of actual risk assessment, there is a constant need for communication amongst assessors, and between assessors and risk managers. REFERENCES Amendola, A. (2001). Recent paradigms for risk informed decision making. Saf. Sci. 40, 17–30. Anonymous (1991). Council Directive 91/414/EEC of 15 July 1991 concerning the placing of plant protection products on the market. Anonymous (1997). Council Directive 97/57/EG establishing Annex VI to Directive 91/414/EEC. Anonymous (2004). Directive 2004/28/EC of the European Parliament and of the Council of 31 March 2004 amending Directive 2001/82/EC on the Community code relating to veterinary medicinal products. Boesten, J. J. T. I. (2000). Modeller subjectivity in estimating pesticide parameters for leaching model using the same laboratory data set. Agric. Water Manage. 44, 389–409. CaSE (2006). Independent expertise to assess risk and scientific evidence. Response to the House of Commons Science & Technology Committee’s Inquiry into Government handling of scientific advice and risk. Campaign for Science and Engineering in the UK, 2006, London. Chess, C. (2001). Organizational theory and the stages of risk communication. Risk Anal. 21, 179–188. De Jong, F. M. W., Van Beelen, P., Smit, C. E. and Montforts, M. H. M. M. (2006). Guidance for summarising earthworm field studies. Bilthoven, the Netherlands. RIVM. 90-6960-154-0. De Marchi, B. and Ravetz, J. R. (1999). Risk management and governance: A post-normal science approach. Futures 31, 743–757. Doerr-MacEwen, N. A. and Haight, M. E. (2006). Expert stakeholders’ views on the management of human pharmaceuticals in the environment. Environ. Manage. 38, 853–866. ECCO (2002). Guidance Document on Aquatic Ecotoxicology; E1 – Plant Health Sanco/3268/ 2001 rev.4 (final) 17 October 2002, DG Sanco, Brussels. Available at: http://europa.eu.int/ comm/food/fs/ph_ps/pro/wrkdoc/wrkdoc10_en.pdf EFSA (2006). Guidance of the scientific committee on a request from EFSA related to uncertainties in dietary exposure assessment. The EFSA Journal 438, 1–54. EFSA PPR (2006). Opinion of the scientific panel on plant protection products and their residues on a request from EFSA on the Final Report of the FOCUS Working Group on landscape and mitigation factors in ecological risk assessment. The EFSA Journal 437, 1–30. Fothergill, A. (2000). Knowledge transfer between researchers and practitioners. Natural Hazards Review, May 2000, 91–98. FSA (2006). The Governance of Science. Food Standards Agency, London. FSA 06/02/07 AGENDA ITEM 9, 9 February 2006. Haibach, G. (1999). Council decision 1999/468 – A new comitology decision for the 21st century? EIPASCOPE 3, 1–9. Halffman, W. (2003). Boundaries of regulatory science, Universiteit van Amsterdam, Amsterdam. Hambrey, J. and Southall, T. (2002). Environmental risk assessment and communication in coastal aquaculture, Nautilus Consultants, Edinburgh, UK. Hamilton, J. D., Daggett, D. A. and Pittinger, C. A. (2006). The role of professional judgment in chemical hazard assessment and communication. Regul. Toxicol. Pharmacol. 46, 84–92. Hassenzahl, D. M. (2006). Implications of excessive precision for risk comparisons: Lessons from the past four decades. Risk Anal. 26, 265–276.

Will Risk Assessment Help Risk Management?

57

Heimbach, F. (1992). Correlation between data from laboratory and field tests for investigating the toxicity of pesticides to earthworms. Soil Biol. Biochem. 24, 1749–1753. Henning, M. and Shear, N. (1998). Regulatory perspectives on the significance of ecological changes as reported in ecological risk assessments. Hum. Ecol. Risk Assess. 4, 807–814. Heyvaert, V. (1999a). The changing role of science in environmental regulatory decision making in the European Union. Law and European Affairs 9, 426–443. Heyvaert, V. (1999b). Reconceptualizing risk assessment. RECIEL 8, 135–143. HoC (2006a). Identity Card Technologies: Scientific Advice, Risk and Evidence. House of Commons Science and Technology Committee, London 2006a. Sixth report of Session 2005–06 HC 1032. HoC (2006b). Scientific Advice, Risk and Evidence: How the Government Handles Them. Wednesday 10 May 2006 DAME DEIRDRE HUTTON and DR ANDREW WADGE Evidence heard in Public Question 575–670. House of Commons, London, 2006b. CORRECTED TRANSCRIPT OF ORAL EVIDENCE To be published as HC 900-vi. House of COMMONS, MINUTES OF EVIDENCE TAKEN BEFORE Science and Technology Committee. Hoekstra, J. A. (1993). Statistics in ecotoxicology. Thesis. Free University, Amsterdam. IET (2006). Inquiry into scientific advice, risk and evidence – how government handles them: IEE comments to the House of Commons Science and Technology Select Committee. The Institution of Engineering and Technology, Hertfordshire, UK. (S)742, 20 January 2006. ISO (2004). ISO/CD 20281 Water quality – Guidance document on the statistical analysis of ecotoxicity data. Final draft 3_2004. International Organization for Standardization. Jones, R. N. (2001). An Environmental Risk Assessment/Management Framework for Climate Change Impact Assessments. Natural Hazards 23, 197–230. Kitchin, K. T. and Drane, J. W. (2005). A critique of the use of hormesis in risk assessment. Hum. Exp. Toxicol. 24, 249–253. Lehtonen, S. and Peltonen, L. (2006). Risk communication and sea level rise: Bridging the gap between climate science and planning practice. Special Paper of the Geological Survey of Finland 41, 61–69. McDaniels, T., Keen, P. and Dowlatabadi, H. (2006). Expert judgments regarding risks associated with salmon aquaculture practices in British Columbia. J. Risk Res. 9, 775–800. Montforts, M. H. M. M. (2006). Assessment of persistency and bioaccumulation in pesticide registration frameworks within the Organization for Economic Cooperation and Development. Integrated Environmental Assessment and Management 2, 21–31, e1–e6. Montforts, M. H. M. M. and De Jong, F. M. W. (2007). Field studies in pesticide registration; questioning the answers. Integr. Environ. Assess. Manage. 3, 150–153. Montforts, M. H. M. M. and De Knecht, J. A. (2004). European medicines and feed additives regulation are not in compliance with environmental legislation and policy. In: D. Dietrich, S. F. Webb and T. Petry (Eds), Hot Spot Pollutants: Pharmaceuticals in the Environment. Elsevier Inc., San Diego. Nyholm, N. (1985). Response variable in algal growth rate inhibition tests – biomass or growth rate? Water Res. 3, 273–279. OECD (2006). OECD guidelines for the testing of chemicals. Freshwater Alga and Cyanobacteria, Growth Inhibition Test. OECD201 2006. Pontolillo, J. and Eganhouse, R. P. (2001). The search for reliable aqueous solubility (Sw) and octanol-water partition coefficient (Kow) data for hydrophobic organic compounds: DDT and DDE as a case study. US Department of the Interior, US Geological Survey, Reston, Virginia, USA, 2001. Water-Resources Investigations Report 01-4201. Read, P. A., Fernandes, T. F. and Miller, K. L. (2001). The derivation of scientific guidelines for best environmental practice for the monitoring and regulation of marine aquaculture in Europe. J. Appl. Ichthyology 17, 146–152. Sanderson, H. and Petersen, P. (2002). Power analysis as a reflexive scientific tool for interpretation and implementation of the precautionary principle in the European union. Env. Sci. Pollut. Res. 9, 221–226. Shere, M. E. (1995). The myth of meaningful environmental risk assessment. Harvard Environ. Law Rev. 19, 409–492. Slovic, P. (2004). Risk as analysis and risk as feelings: Some thoughts about affect, reason, risk and rationality. Risk Anal. 24, 311–322.

58

M. H. M. M. Montforts

SSC (2003). The future of risk assessment in the European Union. The second report on the harmonisation of risk assessment procedures. Scientific Steering Committee, Health and Consumer Protection Directorate-General, European Union, Brussels. Thompson, K. M. (2002). Variability and uncertainty meet risk management and risk communication. Risk Anal. 22, 647–654. Vlek, C. and Keren, G. (1992). Behavioral decision theory and environmental risk management: Assessment and resolution of four “survival” dilemmas. Acta Psychol. 80, 249–278. Wagner, W. E. (1995). The science charade in toxic risk regulation. Columbia Law Rev. 95, 1613–1723. Wheeler, J. R., Grist, E. P. M., Leung, K. M. Y., Morritt, D. and Crane, M. (2002). Species sensitivity distributions: Data and model choice. Mar. Pollut. Bull. 45, 192–202.

Pesticide Risk Assessment in Rice Paddies: Theory and Practice E. Capri and D.G. Karpouzas (editors) © 2008 Published by Elsevier B.V.

Chapter 4

Water Resource Contamination in Italian Paddy Areas Ettore Capri Istituto di Chimica Agraria ed Ambientale, Università Cattolica del S. Cuore, via Emilia Parmense, 84, I-29100 Piacenza, Italy Contents 1. Introduction 2. Potential non-point contamination 3. Defining vulnerable areas 4. Human risk References

59 60 62 64 66

1. INTRODUCTION Rice is considered a high-value crop that is cultivated in Europe under submerged conditions. The requirements of high volumes of water for maintaining flooding conditions throughout the cultivating season have led to the establishment of rice growing areas within river basins. Such rice-cultivating areas are located close to large urban areas, including the Axios river basin in northern Greece and the Po Valley basin in northern Italy, whereas others are situated within zones of high ecological value such the National Park of Donana and the Lagoon of Valencia in Spain (Ramos et al., 2000). Weeds have been recognized as the most significant crop protection constraint in rice cultivation in Europe, although many insecticides have been measured in other continents (Schulz, 2004). However, their intensive use in rice cultivation has resulted in the frequent detection of herbicides in adjacent surface and groundwater systems. A monitoring study employed in the Albufera Lake in Spain, which receives and provides water for rice irrigation, revealed maximum concentrations of 1.78, 10.3 and 0.31
g /L for the herbicides molinate, bensulfuron-methyl and benthiocarb, respectively (Tarazona et al., 2003). Similar results have been reported in the main rice cultivation areas in Greece (Papadopoulou-Mourkidou et al., 2004), Italy (Readman et al., 1993; Capri et al., 1999) and Portugal (Cerejeira et al., 2003), raising concerns about the risk of environmental and human exposure. Overseas similar results have been reported in Asia, Australia and America as reported in other chapters of this book and in the review of Schulz (2004) and FOCUS (2001). On the other hand, the potential environmental risk is high because the environmental fate assessment at landscape level is scarce and, because monitoring studies are difficult and expensive due to the high vulnerability of the area cropped under flooding conditions. However, in the recent years with the challenges in

60

E. Capri

legislation towards a higher precautionary principle, only those pesticides performing well in laboratory studies, i.e. low persistence in the soil–water system, may be authorized for agronomic use. So, if the misuse of pesticides is excluded as well the estimation of cumulative residue and transformation products, the contamination of the water resources at level of risk for humans is unlikely as reported in the following paragraphs. 2. POTENTIAL NON-POINT CONTAMINATION Italy represents an interesting case study because pesticides may contaminate water resources, both in the surface and in the ground due to the geological conditions of the areas where rice is cropped (e.g. shallow groundwater, sandy soil and permeable soil). Italy is the most important rice producing country in Europe, accounting for 660,000 of the 1,140,000 metric tons of milled rice produced by the European Community each year. Ninety-nine percent of rice grown in Italy is grown in Northern Italy on an area totalling 232,800 ha, located mainly along the basin of the Po River (Figure 1). The area under cultivation has increased rapidly due to the high profitability and the cultural and traditional relevance of this crop. Expansion has occurred even in areas traditionally less suitable for rice production, where the presence of irrigation water makes cultivation possible. Rice is grown mostly under flooded (paddy) conditions on a range of soils, where organic matter content can vary from 0.5 to 20% and where clay and sand contents can be greater than 50%. Groundwater ranges in depth from 0 to 50 m. Due to these properties, it is expected that agronomic practices can have a positive or negative impact on water quality in this rice growing area. Water quality can

Rice fields Water courses and water bodies

MI

VC

MN

PV

Adriatic sea Po river

N W

E S

Tyrrhenian sea

50000

0

50000

100000 150000

200000 Meters

Fig. 1. Schematic representation of the main area cropped with rice in Italy and belonging the province of Mantua (MN), Pavia (PV), Milan (MI) and Vercelli (VC).

Water Resource Contamination in Italian Paddy Areas

61

be negatively affected because of different factors: (a) irrigation increases the likelihood of pesticides leaching to groundwater or reaching surface water via drainage; (b) in most cases, the surface water in flooded paddies saturates the soil so increasing the potential for groundwater contamination; (c) the cultivation of rice throughout the Po River Valley is very intensive and so the potential loading of agrochemicals, such as fertilizers, trace elements and pesticides, is high. When the investigation is carried out at landscape level all these factors can negatively affect both the shallow groundwater and the surface water bodies that are interconnected with main water resources. In the last years pesticides are the main environmental issue in Europe so many studies have been conducted to determine the processes of pesticide dissipation in rice paddies. Capri et al. (1999) reviewed most of the published data on the pesticide monitoring programme carried out in the paddy areas of Italy. Results showed a diffuse pollution of both surface and groundwater bodies with pesticides at concentrations varying from 0.1 to 30
g L⫺1. Amongst the pesticides found, the authors reported specific pesticides used in rice crops (propanil, molinate, bethiocarb, bensulfuron-methyl) as well as pesticides used in other rotational crops like maize and soya (alachlor, atrazine, metolachlor and bentazone). In some provinces of Italy, water contamination is so persistent that the use of pesticides has been limited. For example, the use of molinate and bentazone has been prohibited since 1987 (Ordinanza Ministeriale, 1987a,b). In paddy area the surface water is more vulnerable to contamination compared with groundwater due to its water continuum with the treated field. Trevisan et al. (1991) in their survey of propanil residues demonstrated, in a worst-case situation, that proximity to the application date, for such a mobile but not persistent pesticide represents a hazard for surface water instead of groundwater. When only the surface water is investigated the type of water bodies, however, seems not to influence the distribution of the pesticide within the watershed as demonstrated by the high variability of data measured in the different monitoring studies. An example for surface water using tricyclazole as test compound in large monitoring study at ditch level the mean was 1.57 ⫾ 3.80, against 0.19 ⫾ 0.55, 0.19 ⫾ 0.55, 0.43 ⫾ 1.77, 0.08 ⫾ 0.16, 0.15 ⫾ 0.41, 0.25 ⫾ 0.50
g L⫺1 in different Italian surface water bodies (canale, roggia, stream, river, cavo, rio respectively). The sediment samples were sporadically contaminated by pesticide residues. 6.8% of the sample showed concentration higher than the detection limit with a 95th percentile equal to 0.03 mg kg⫺1. No statistical difference was found between the classification factors used in the analysis. The results showed that, during the tricyclazole application period (July–August), most of the sampling points in the paddy fields (target area) and in the surrounding water bodies (non target area) contained tricyclazole residues. Contamination also for such a lipophilic compound was limited to the water phase, particularly in the target area (paddy fields), and sediments are rarely contaminated. Residues measured in the total water samples collected from all the sampling points (173) showed a 50th percentile concentration below the LOD (⬍0.05
g L⫺1), with 95th percentile concentration of 10.1
g L⫺1 at the farm scale, 1.1
g L⫺1 at the catchment scale and 1.0
g L⫺1 at the basin scale. Contamination was much lower a few weeks after harvesting (November–December). This is mainly due

62

E. Capri

to pesticide dissipation and the mitigation of contamination by the rice paddy system as occurs for other contaminants such as trace elements, fertilizer and gaseous pollutants (Padovani et al., 2005). The above examples demonstrate that contamination can be restricted to target areas, such as the paddy field and the surrounding farm drainage channels, that are not considered ecologically relevant in Italian legislation; the contamination is restricted to the period of application; the groundwater resources affected by contamination are shallow groundwater already contaminated by other pollutants and not used for irrigation and drinking purposes. The contamination of the water resources from pesticide use in rice is mainly related to the intrinsic vulnerability of each area (e.g. paddy field concentration per unit of area). In this area the pesticide use should be controlled and measured for a sustainable use of pesticide. To reduce the risk, landscape management at the farm and regional level should be developed, including the use of crop rotation and other agronomical practices such as reducing the use of water and irrigation systems based on continuous flooding. 3. DEFINING VULNERABLE AREAS Vulnerability of the rice cropped is an intrinsic characteristic of the landscape system (intrinsic vulnerability) that interacts with the properties of individual contaminants or contaminant groups and human activities (specific vulnerability) mitigating or enhancing the phenomena, for example the contamination. In general, as vulnerability of the water resources to contamination we consider the tendency or likelihood for pesticides to reach a specified position in the water resource after introduction at a location (e.g. the paddy field or a farmyard in case of point source) above the uppermost aquifer or surrounding surface water bodies such as drainage channel, stream, river. Various physical, chemical and biological processes determine the environmental fate of pesticides. The rates and importance of each of these processes are, in turn, affected by different factors. For examples in rice cultivation the main factors influencing the intrinsic vulnerability within the application area are: (1) the land use and land management (pesticide application rate and timing, tillage, irrigation practice); (2) the soil properties (organic matter content, texture, structure); (3) the crop properties (sensitivity to pest, plant sorption and uptake); (4) the climate (timing of first rainfall after pesticide application, temperature, potential evapotranspiration); (5) the subsoil (chemical, physical and biological properties); (6) the vadose zone (thickness); (7) the groundwater (flow, dilution). Off-site the main factors are the characteristics of the surrounding surface water bodies (networks, pH, flow, temperature, chemical properties, presence of macrophyte, biofilms). The specific vulnerabiltiy is function of the pesticide properties mainly their adsorption to the sediment and suspended organic matter, the biotic and abiotic degradation, the volatilization. The land cropped with rice may be analyzed using maps of various scales, moving from large scale, territorial analysis (national and regional) to smaller scales (province, commune, farm, field). Geographical material on paper as well

Water Resource Contamination in Italian Paddy Areas

63

as digital maps in either raster or vectorial form, processed with a Geographic Information Systems (GIS) application, can be utilized. An initial observation is made of the hydro-agronomic characteristics throughout the national territory on a geographical map, scale 1:1,000,000. Once rice growing regions are identified, a regional analysis is made on a scale of 1:100,000 in order to identify areas at a higher risk of contamination. Data on scales of 1:50,000–1:60,000 are used to define hydrologic characteristics within the higher risk areas. Sampling sites, if necessary, may then be identified on 1:10,000 scale maps. Three aspects are considered relevant for the identification of areas at a higher hazard and risk of pesticide contamination: (1) the extent and location of rice cropping areas throughout the national territory; (2) the quantities of pesticide applied in those areas; and (3) the geographical distribution of abiotic factors that make the area susceptible to contamination. For example Padovani et al. (2005) developed two indicators to estimate the hazard entailed by tricyclazole and performing his environmental monitoring. The first hazard indicator represented the extent of rice grown in the various administrative areas:

RI ⫽

SRC  SC  ⫻  1⫺ SAUC  ∑ SCR 

(1)

where C is the administrative area (e.g. administration 1–100 km2 in size); RI the rice index calculated for each single area; SRC the surface area rice cropped (ha); SAUC the total agricultural surface in the area potentially used in agriculture (ha); SC the surface area (ha);  SCR the sum of surface areas in Italy where rice is grown (ha). A second indicator can be overlapped to quantify the spatial use pattern of the pesticide used within each area. PI ⫽

BC SRp

(2)

where p is the province; PI the pesticide index calculated for administrative area (kg ha⫺1); BC the pesticide sales (kg); SR the surface area where rice is grown (ha). These hazard indicators may then be illustrated in a map (Figure 2). A more deterministic model can be used to allow the vulnerability assessment towards an integrated risk assessment that couples exposure and effect predictions. Such models need sophisticated database as large as the details required. Recently the Soil Survey Service in Lombardy, one of the Italian regions with the highest rice cropped surface area in Italy, developed a Decision Support System called SUSAP (Jantunen et al., 2005) as an instrument for planning sustainable use of pesticides in agriculture. The software combines soil and meteorological maps from the area with the rice crop and its pests and a database that covers the physicochemical and ecotoxicological properties, application plan and efficacy of 215 active substances and pesticide products that contain them. The system can be applied on the regional, local or farm level, utilizing basic soil data on either 1:250,000 or 1:50,000 scale. Twenty “macroareas” characterized

64

E. Capri

Fig. 2. Hazard index calculated for the Milano province.

by one soil and climate were defined: representative soil profiles were characterized by field and laboratory analysis (pedotransfer functions that were found to best calculate the measured data were used for producing field capacity, wilting point and bulk density to add to the basic soil data), and weather data were collected from 125 meteorological stations and interpolated with geostatistical methods in order to produce rainfall and temperature maps for the area. PELMO model is used for simulating the pesticide fate in unsaturated zone, and RICEWQ for paddy areas and surrounding surface water bodies are run for a 12-year simulation period, including a 2-year warm-up, in order to assess the vulnerability. On the local and regional level, the results are given in the form of GIS vulnerability maps that visualize the 80th percentiles of pesticide concentration in soil water at 1m depth inside areas that are similar in reference to climate and soil. At farm level, PECs of pesticides in groundwater are calculated for use in the calculation of an environmental risk indicator that helps the farmer choose a suitable pesticide. 4. HUMAN RISK Control of the overall quality of the water resources in Italy is carried out by government institutions dealing with the environmental and human health. The Italian national agency for the environmental protection (APAT) coordinates the overall monitoring plans (technical protocols, data processing, statistical assessment and yearly report), while each region (through its environmental agency) applies the monitoring plan that lasts 3 years. The general framework for ranking the pesticide monitored is based on national uses, the list of priority chemicals, bans at EU

Water Resource Contamination in Italian Paddy Areas

65

level, reported contamination in previous monitorings and the EPA-California model which selects the pesticides to be monitored on the basis of water solubility (⬎3 mg L⫺1), soil adsorption coefficient (Koc ⬍1900 cm3 g⫺1), hydrolysis halflife (⬎14 days), aerobic soil metabolism half-life (⬎610 days) and anaerobic soil metabolism half-life (⬎9 days). For the year 2004 (APAT, 2005), 1992 sampling points in GW generated 3529 samples and 65,383 measures of pesticides. Pesticide residues were detected in 22.5% of the sampling points and 19.6% of the samples. Out of the 188 compounds analyzed, 34 are detected in ground water with a median of 1.7 molecule per sample. Herbicides (and their metabolites) are the group most frequently detected (97% of the detections). The 10 molecules most frequently detected were desethylatrazine (10.5% of the 1882 samples), atrazine (10.2% of the 3228 samples), desethylterbuthylazine (9.6% of the 1892 samples), 2,6-dichlorobenzamide (8.2% of the 404 samples), terbuthylazine (8.0% of the 3128 samples), bentazone (7.1% of the 798 samples), esazinone (6.2% of the 1063 samples), simazine (4.6% of the 3238 samples), oxadiazon (3.0% of the 1185 samples) and metolachlor (2.1% of the 2463 samples). The exposure level for GW indicates that 7.4% of the samples exceed the threshold value of either 0.1 µg L⫺1 per substance or 0.5 µg L⫺1 for the sum of all substances detected. The above situation also applies in rice cultivated areas where in addition to the above figures more frequent detection of molinate occurs and, less often, of bentazone. This is evidence that the contamination is not due to the crop itself but it is due to the intrinsic vulnerability of the soil and to the characteristics of the pesticides themselves (specific vulnerability), plus other factors including misuse. Even when the contamination is diffuse it does not represent a risk for human health because the overall quality of water, chemical and biological, is so low that it is unlikely to be used for drinking purposes. In fact, the historical use of these resources has been only for irrigation. At present most of the drinking water is extracted from deep groundwater wells and collected from springs in the mountains. To verify safe conditions of the water resources for drinking use in the year 2005, a survey of 20 drinking water supplies was conducted by the Health Ministry indicating that, even though at least one pesticide or metabolite is found in every sample, the World Health Organization (WHO) RMA threshold values were never reached. Atrazine (2), desethylatrazine (5), bentazone (2), terbuthylazine (2), desethylterbuthylazine (2) and oxhadiazon (3) occurred few times (in brackets) at concentration higher than 0.1
g L⫺1. In few samples more than one pesticide was detected whose sum was higher than 0.5
g L⫺1. In two samples 7 parent compounds and 10 metabolites at concentrations of 0.83 and 1.04
g L⫺1 were found, respectively. In any case the values were below the WHO threshold such as for atrazine, 2
g L⫺1; simazine, 2
g L⫺1; terbuthylazine, 7
g L⫺1; molinate, 6
g L⫺1; alachlor, 20
g L⫺1; metolachlor, 10
g L⫺1 (Funari and Bottoni, 2007). These available data at national scale give a good picture of the status of the quality of the water resources demonstrating that the application of the new EU directives for the authorization of the pesticide and the good agricultural practices followed by the farmers are safe for the water resources used as food. These data also confirm the estimation carried out with modelling and vulnerability assessment of the soil.

66

E. Capri

On the other hand, the monitoring plan identifies gaps resulting from the fact that (i) sampling and measurement are still not derived following a harmonized plan, (ii) laboratories are very much heterogeneous in detection limits, analytical methods and pesticides analyzed and (iii) as a consequence an overall national picture of the monitoring is still lacking. For this reason to confirm the safe use for human health in the future the coupling of vulnerability maps and fate models are needed to address better the monitoring survey and to complement the stewardship activities planned in rice cropped area. In developed countries like Italy safety issues related to rice as well to other crops still remain open: the cumulative risk assessment of the metabolites and pesticides detected at low concentration in the diet basket of the consumer. In the near future such issues should emphasize the importance to carry out further work to develop a methodology to take into account cumulative and synergistic effects of pesticides to human health. Up to now there is no generally agreed framework/approach on combined risk assessment of pesticides. However at European and International levels there are some ongoing activities concerning approaches for cumulative risk assessment of pesticides which have a common mode of action (EFSA, 2007). REFERENCES APAT. (2005). Piano nazionale di controllo degli effetti ambientali dei pesticidi. Residui di pesticidi nelle acque. Rapporto annuale dati 2004. APAT, Roma, p. 77. Capri, E., Cavanna, S. and Trevisan, M. (1999). Ground- and surface water bodies contamination by pesticide use in paddy field. Proc. Workshop on Environmental Risk Parameters for Use of Plant Protection Products in Rice, Cremona, Italy, September 16, pp. 48–71. Cerejeira, M. J., Viana, P., Batista, S., Pereira, T., Silva, E., Valerio, M. J., Silva, A., Fereira, M. and Silva-Fernandes, A. M. (2003). Pesticides in Portuguese surface and ground waters. Water Res. 37, 1055–1063. EFSA. (2007). Colloquium 7: Cumulative risk assessment of pesticides to human health: The way forward, in press. FOCUS. (2001). “FOCUS Surface Water Scenarios in the EU Evaluation Process under 91/414/EEC”. Report of the FOCUS Working Group on Surface Water Scenarios, EC Document Reference SANCO/4802/2001-rev.2 final (May 2003), 245 pp. Funari, E. and Bottoni, P. (2007). Relazione finale piani triennali di monitoraggio. Istituto Superiore della Sanità, Roma, in press. Jantunen, A. P. K., Trevisan, M. and ja Capri, E. (2005). Computer models for characterizing the fate of chemicals in soil: Pesticide leaching models and their practical applications. In: J. Álvarez Benedí and R. Muñoz-Carpena ja (Eds), Soil-Water-Solute Process Characterization: An Integrated Approach (pp. 715–756). CRC Press, USA. Ordinanza Ministeriale. (1987a). n. 135. Gazzetta Ufficiale della Repubblica Italiana, n. 80. Ordinanza Ministeriale. (1987b). n. 217. Gazzetta Ufficiale della Repubblica Italiana, n. 127. Padovani, L., Capri, E., Padovani, C., Puglisi, E. and Trevisan, M. (2005). Monitoring Tryciclazole residues in rice paddy watersheds. Chemosphere 62, 303–314. Papadopoulou-Mourkidou, E., Karpouzas, D. G., Patsias, J., Kotopoulou, A., Milothridou, A., Kintzikoglou, K. and Vlachou, P. (2004). The potential of pesticides to contaminate the groundwater resources of the Axios river basin. Part II. Monitoring study in the south part of the basin. Sci. Total Environ. 321, 147–164. Ramos, C., Carbonell, G., Garcia Baudin, J. M. and Tarazona, J. V. (2000). Ecological risk assessment of pesticides in the Mediterranean region. The need for crop-specific scenarios. Sci. Total Environ. 247, 269–278.

Water Resource Contamination in Italian Paddy Areas

67

Readman, J. W., Albanis, T. A., Barcelo, D., Galassi, S., Tronczynski, J. and Gabrielides, G. P. (1993). Herbicide contamination of Mediterranean estuarine waters: Results from a MED POL pilot survey. Mar. Pollut. Bull. 26, 613–619. Schulz, R. (2004). Field studies on exposure, effects, and risk mitigation of aquatic nonpointsource insecticide pollution: A review. J. Environ. Qual. 33, 419–448. Tarazona, C., Carrasco, J. M. and Sabater, C. (2003). Monitoring of rice pesticides in an aquatic system of natural park of Albufera, Valencia, Spain. Hazard evaluation. Proc. XII Symposium of Pesticide Chemistry: Pesticides in Air, Plant, Soil and Water System, Piacenza, Italy, July 4–6, pp. 727–735. Trevisan, M., Montepiani, C., Ghebbioni, C. and Del Re, A. A. M. (1991). Evaluation of potential hazard of propanil to groundwater. Chemosphere 22(7), 637–643.

Pesticide Risk Assessment in Rice Paddies: Theory and Practice E. Capri and D.G. Karpouzas (editors) © 2008 Elsevier B.V. All rights reserved

Chapter 5

Ecotoxicology of Rice Pesticides Jose V. Tarazona1 and G. Peter Dohmen2 1Department

of the Environment, Spanish National Institute for Agriculture and Food Research and Technology (INIA), 28040 Madrid, Spain 2BASF, Agricultural Center Limburgerhof, Ecotoxicology, D-67114 Limburgerhof, Germany

Contents 1. Introduction 2. Environmental and ecotoxicity data for risk assessment of pesticides 3. Specific issues associated with paddy rice 4. Ecotoxicological assessment of rice pesticides 4.1. Identification of relevant non-target populations and communities 4.2. Assessing the effects on the paddy community 4.3. Effects on associated communities and populations 4.3.1. Effects on bird populations 4.3.2. Effects on other terrestrial vertebrates and the use of mammalian toxicity data in ecotoxicological assessment 4.3.3. Effects on associated wetlands 4.4. Proposals for testing strategies and future needs 5. Conclusions References

69 71 74 77 80 81 84 84 85 85 87 87 88

1. INTRODUCTION Ecotoxicology can be defined as the study of the effects of substances (and physical agents according to some authors) on ecosystems. This covers the assessment, monitoring and diagnosis of effects on populations, communities and ecosystems including their structure and function and interactions with the physical and chemical environment, but also includes the development of methods for assessing the level of impact considered to not endanger the sustainability of ecosystems, the determination of conditions for the safe use of chemicals or agents and their management in order to achieve safe uses. Pesticides are biologically active substances intended to be effective against certain groups of organisms. Accordingly, some side effects next to the wanted activity can usually not be excluded. Regulatory decisions and management practices should limit such unwanted impact. However, as many other anthropogenic stressors on ecosystems, rice pesticides have caused unwanted and unpredicted ecological effects due to lack of knowledge or misuse. These effects have been reported and reviewed by other authors (Tejada et al., 1995; Abdullah et al., 1997). The objective of this chapter is to review the use of ecotoxicological methods for assessing the risk of rice pesticides and to explore new developments. As risk

70

J. V. Tarazona and G. P. Dohmen

assessments offer the scientific basis for regulatory processes, this revision has been based on the current status of pesticide registration in the European Union. It should be noted that although almost all countries have established registration programmes for pesticides, there are significant differences in the requirements and capacities between developed and developing countries (Castillo et al., 1997). One of the main applications of ecotoxicology is to predict the potential environmental effects within the risk assessment context. This assessment is usually carried out in two subsequent steps, starting with the identification of the hazards potentially associated to a chemical followed by the establishment of dose/ response relationships (Pugh and Tarazona, 1998). In a subsequent step, the potential environmental concentrations of the substance are estimated and/or measured and compared to the toxicity (hazard) of the compound resulting in a prediction of risk. The protection goal in ecotoxicology is generally the sustainability of ecosystems and populations and not the individual organism, in contrast to human toxicology where effects on individuals are the main focus of the evaluation. As a consequence, the advance of ecotoxicology has required the development of new methodologies – starting from subcellular enzymatic bioindicator tests to complex community studies – and conceptual models for covering issues such as ecosystem redundancy and resilience, the assessment of indirect effects, or the potential for recovery (e.g. USEPA, 1998). Figure 1 summarises the ecotoxicological assessment process.

ECOSYSTEM

EFFECT MAGNIFICATION

ABIOTIC FACTORS COMMUNITY

ENVIRONMENTAL FACTORS

EFFECT MITIGATION

COMMUNITY +

POPULATIONS ENVIRONMENTAL FACTORS INDIVIDUALS

GEOGRAPHICAL DIMENSION Fig. 1. The ecotoxicological effect assessment requires the extrapolation of effects observed on individuals to consequences on populations, communities and ecosystems.

Ecotoxicology of Rice Pesticides

71

As a matter of principle, the toxicological effects represent the response of the individual to the chemical exposure, while the goal of the assessment is to identify the consequences on populations and communities or on their functions. Thus, ecotoxicological approaches can use individual responses, and then estimate its consequences on populations and communities, or use directly higher tier assays measuring endpoints closer to ecology than to classical toxicology. Both approaches are complementary, and most risk assessment protocols are established on the basis of tiered levels. The initial or lower tier approaches are often based on single-species bioassays, resembling those developed by classical toxicology. In general, the lower tier tests are conducted under laboratory conditions; they are designed to provide clear, reproducible results with limited input of time and resources. At the same time, they only poorly reflect the reality. Higher tier tests are often designed to be closer to the real situation. This makes such studies often more complicated and expensive and reduces at the same time their replicability. Finally, field tests and monitoring studies provide insight into the real situation. However, the results are often only exactly true for this situation, although extrapolation to other situations is generally possible and valid, if certain limitations are considered. Such a tiered approach is applied frequently nowadays, when it is important to discover any significant, sometimes subtle impact relevant under environmental conditions, which may be small compared to other natural or anthropogenic impacts. For example, normal agricultural practice like ploughing, preparing the seed bed and harvesting often have a much stronger impact on soil organisms then the impact of pesticide use. During the 1960s and 1970s, the interpretation of the observed effects in hot spots, exploring the potential causal-effect relationships between measured concentrations and the observed ecological disasters, was rather straightforward. The clear evidence of pollution related effects, associated to a technological development with low environmental concern, expanded the role of ecotoxicology, and the 1970s and 1980s can be characterised by the development of ecological quality standards and objectives, setting maximum acceptable concentrations expected to be of insignificant impact to the environment. The methodology for setting these standards has been further developed and still represents a main request from the regulatory arena to the scientific community (Bro-Rasmussen et al., 1994) and for scientific debate (Tarazona, 1997; CSTEE, 2004). However, decision makers have to balance potential negative environmental impact, environmental costs, with the benefits of certain human activities. To do this the risk based approaches are the most appropriate tool. The decision-making process is only partly a scientific question, and depends mainly on societal, economic and political considerations. 2. ENVIRONMENTAL AND ECOTOXICITY DATA FOR RISK ASSESSMENT OF PESTICIDES Pesticides are mostly recognised as plant protection products, although other uses should be considered in some cases. In the particular case of rice pesticides, the use for vector control is particularly relevant in certain areas. Within this

72

J. V. Tarazona and G. P. Dohmen

chapter, the assessment will focus on the use of rice pesticides as plant protection products. From a scientific perspective, the principles for the ecotoxicological assessment of other uses do not differ significantly; however, it should be recognised that the testing strategies, risk management and particularly, the cost/benefit assessment and decision-making, is directly related to the intended use, and may result in large differences in terms of the level of ecological risk considered to be acceptable. The environmental risk assessment of pesticides as plant protection products poses several particularities. Pesticides are designed to be biologically active compounds that can affect certain biota. Furthermore, the release of the chemical into the agricultural environment is intentional and the purpose is to affect organisms that are competitors, pathogens or pests for the crop. The intended release may also have consequences to organisms not intended to be affected – side effects on non-target organisms – and on adjacent and remote areas, where the pesticide is transported by/through air, water, soil or even living organisms. Both processes should be considered, but the approaches, protection aims and methodologies are different. In the off-crop area, exposure should be considered as a contamination event, resulting in the unintended presence of a potentially toxic chemical in environmental compartments. Exposure mitigation measures are feasible in some cases and welcomed. The consequences associated to the exposure and the ecotoxicological assessment should be based in the generic goal to avoid significant effects on the structure and functioning of the ecosystem. The in-crop assessment, however, must consider that the application of the pesticide is planned and intentional with the goal of altering the system and interspecies relationships in a way that is favourable to the crop, i.e. by reducing competition, herbivory and diseases (and thus intentionally reducing biodiversity). Thus, the assessment cannot be done on the basis of preventing biological/ ecological changes, but on restricting those changes to the intended ones. In addition, the assessment is not conducted on a (semi-)natural ecosystem but on human-managed agrobiosystems. The landscape, the abiotic compartments and the biodiversity of the biological community are heavily modified for improving crop productivity and reducing cost. Agricultural practices modify soil conditions and require severe adaptations of the soil community. The consequences for the above ground compartment are even higher, as the vegetation cover is replaced by the crop. Nevertheless, the agrobiosystems represent a relevant habitat for a large number of species. The relationships between the crop area and wild species is highly variable; in some cases, the crop area represent the predominant habitat for the species, while in others it is just a sporadic feeding zone. These reasons justify an in-crop effect assessment not based on the protection of the structure and function of undisturbed ecosystems. On the other hand, although being effective biological agents, undesired side effects should be minimised. As a direct consequence, pesticide effect assessment protocols require a new problem formulation for the in-crop assessment, which is based on the identification of non-target organisms. This identification is relatively easy in some cases. For example, some insecticides target very specific pests which can be identified at the species level for each crop. In these cases all other species,

Ecotoxicology of Rice Pesticides

73

including other insect species, should be considered as non-target organisms. The assessment is not so clear in other cases, and a regulatory decision is required; for example, despite of their selectivity for some plant species, the European assessment consider that all plants within the crop area are target for herbicides (DGSANCO, 2002a). The situation can be even more complex, e.g. for fumigants and soil sterilisers, where the efficacy may be directly related not just to a reduction of a limited number of harmful species on the crop agrobiosystem but to intense structural changes. The in-crop level of protection also incorporates aspects associated to the beneficial effect of some species and taxa for soil protection and crop productivity. In fact, the protection of beneficial organisms used to be a main aim within the environmental safety assessment of pesticides. Although, nowadays, the goal has been extended in most regulatory schemes for covering all non-target groups independently of their agronomic implications, it must be recognised that the protection of well-known beneficial species is still a relevant aspect within the in-crop environmental assessment. Above ground invertebrates represent the most typical example. Although the final aim is theoretically established on ground, foliar and pollinator invertebrates as a whole, the reality in most cases is an assessment for honey bees and a few predatory and parasitoid species, which are considered sensitive and also posse agronomic relevance (Candolfi et al., 2001). There is also an intermediate level, which could be defined as an edge of field assessment, covering the non-crop areas surrounding the field. The level of agronomic impact in this zone is not as high as within the crop area, but it is not negligible and a pure ecological assessment is not possible. The first 1 m next to the field is often heavily disturbed by agricultural practices (in paddy rice, for example, the banks surrounding the field). The following 2 m (i.e. up to 3 m from the crop zone) are still disturbed by agricultural practices (e.g. turning areas for machinery) or removing vegetation (mowing). These areas may represent the transition between the agricultural land and the off-crop zone, and the level of protection should be established at least at the functional level, including as an ecological function the capability for re-colonisation of the crop area either between agronomic seasons or after abandoning the agricultural activity. These three levels have been summarised in Table 1. Using the European assessment scheme, taxonomic groups have been allocated to each area. Although some divergences may exist among the schemes developed in different world regions, the basis for the assessment are of generic nature and therefore worldwide applicable with just some differences related to regulatory settings or agricultural practices. The consequences of these different levels of protection are addressed in the methodology employed for the assessment but also on the margin of safety, or using a more proper definition the uncertainty factor, employed in the legislation. Larger factors are usually applied when the extrapolation is required for covering a wider set of taxonomic groups and/or ecological functions. The revision of the EU guidelines conducted by the Task Force for Harmonisation of Risk Assessment Procedures under the European Steering Scientific Committee offers several considerations on how these differences have been addressed within several European regulations (SSC, 2003).

74

J. V. Tarazona and G. P. Dohmen

Table 1. Local assessment scenarios normally employed in the ecotoxicological assessment of pesticides Off-crop assessment Aquatic organisms including sediment and associated terrestrial systems

Edge of field

In-crop assessment

Terrestrial plants – ground and foliar invertebrates (non-target arthropods)

Birds and mammals – pollinators, non-target arthropods, soil dwelling organisms, soil function Protection level based on the Protection level based on Protection level based structure and functioning populations/communities on direct effects on of ecosystems including the potential non-target populations for recovery including the beneficial aspects of some species and groups, soil function and sustainability of the system

Regarding the methodological issues, the main difference in the lower tier assessment is the selection of the test system. The screening assessment of ecologically relevant effects is based on a simplistic assumption: the protection of the structure guarantees the protection of the ecosystem functions. The test systems should cover all relevant groups and, as discussed above, the uncertainty factor must be large enough for covering the non-tested groups. When the problem formulation excludes the target species, the selection of the proper non-target species in the lower tier screening assessment is required, and factors such as potential benefits (e.g. in the selection of bees and specific foliar arthropods) or citizen perception (e.g. in the selection of birds instead of reptiles) becomes more apparent. Probably, the main differences appear in the application of higher tier assessment, based on multispecies experiments, semi-field and field studies. Indirect effects are always relevant for ecosystem-based assessments, while they must be excluded in some population/community evaluations in the context of an agriculturally intended reduction of biodiversity with the aim to benefit the crop. For example, the use of non-selective herbicide will provoke in all cases a dramatic reduction of foliar arthropod biodiversity within the treated area. This effect is not intended, but is the unavoidable consequence of the effects on the target vegetation. Thus, while the potential direct toxicity of the herbicide for arthropods is included in the assessment as this could be considered an avoidable side effect, the indirect consequences are excluded.

3. SPECIFIC ISSUES ASSOCIATED WITH PADDY RICE The assessment of any particular crop requires some adaptations of the assessment scenario. The environmental conditions of rice paddies are so unique, however, that a different conceptual model is required for addressing the problem formulation. It should be noticed that in some areas rice is grown on dry arable

Ecotoxicology of Rice Pesticides

75

soil instead of paddies; obviously, in those circumstances the assessment of rice pesticides becomes similar to the one for other cereals. There are four main issues associated to the paddy that should be considered and will be described below: ● ● ● ●

the specific conditions in rice semi aquatic/terrestrial ecosystem, fate considerations, the presence of specific communities, and the specific ecological value and in some regions the close connection to natural, ecologically important nature reserves.

An additional specificity of rice paddies is the use of pesticides for vector control instead of for crop protection. These assessments offer fundamental differences when compared to conventional pesticide evaluations. The main obvious difference focus on risk/benefit considerations, as Public Health, not farmer economy, is involved in this case. In addition, the efficacy of these measures requires a different geographical dimension, which must cover a significant part of the vector distribution area and do not distinguishes between in-crop and off-crop assessments. Furthermore, taxonomic groups traditionally considered as “non-target” standard species, such as chironomids, the most typical sediment dwelling organisms in ecotoxicity test design at the OECD level, can be the target for pesticide applications in rice paddies (Stevens and Warren, 1995). Focusing on plant protection products, as a matter of principle, the basic concept of in-crop and edge of field assessments based on terrestrial organisms and an off-crop assessment for aquatic ecosystems is not applicable to rice. Rice paddies are certainly agrobiosystems and not wetland ecosystems, but the artificial flooding of rice paddies creates a sequence of aquatic and terrestrial communities, which obviously is not observed in other crops. As a consequence, the in-crop assessment for rice paddies requires the definition of non-target effects for the paddy community, as well as a reconsideration of the relevance of the potential effects on pollinators, foliar and ground arthropods and on soil dwelling organisms. The second relevant group of differences between rice paddies and other crops are those regarding the fate processes. These aspects have received particular attention in other chapters within this volume. In the following the ecotoxicological consequences associated to these fate differences are highlighted. For the in-crop assessment, the application patterns, flooded or drainage conditions; will determine how and when the aquatic paddy community will be exposed, as well as the sequence of soil/sediment dwelling communities that should be considered in the assessment. Although heavily modified by the agricultural practices, the below ground community should be assimilated to that observed in Mediterranean wetlands, which, in a simplified approach could be described as a succession of soil, flooded soil, sediment, drained sediment and back to soil dwelling communities. The exposure assessment for mammals and birds also requires some adaptations with respect to the exposure patterns and in particular the food items to be considered. Regarding the off-crop assessment the most significant exposure difference is related to the routes regulating the transfer of the applied chemical from the

76

J. V. Tarazona and G. P. Dohmen

treated area to adjacent aquatic ecosystems. Drift during application is obviously as relevant for rice as for other crops, while the relevance of the drainage to surface water and leaching to groundwater processes mostly depends on the specific paddy conditions. In addition a new emission route should be considered: the direct discharge of paddy water into associated water bodies. This route appears as a consequence of the need for maintaining a water flow in the paddy system and it may constitute the most relevant route in some cases. The third and fourth specific aspects are connected and directly linked to the biology of the paddy system and associated environments. The paddies constitute a very rich agrobiosystem in terms of biodiversity and large numbers of species are expected in these systems (Hidaka, 1998). Rice cultivation in Europe and other parts of the world can be divided in two main subgroups, showing clear differences in term of the associated biological community. The first subgroup corresponds to paddies associated to wetlands, mostly in coastal areas such as those of La Albufera, the Ebro delta or the Guadalquivir in Spain or the Camargue in France. The paddy community in these systems is directly related to the wetland community. Birds represent a particularly relevant group. Elphick (2000) studied Californian paddies and concluded that flooded fields apparently provide equivalent foraging habitat to semi-natural wetlands to bird species; similar situations have been observed in Europe (Hafner et al., 1986; Fasola et al., 1996). The second subgroup corresponds to inland arable land transformed into rice paddies through irrigation practices. The rice area in the Po valley represents a typical example similar to that observed in Zaragoza and Badajoz in Spain. These conditions create new habitats to be colonised by adapted species. The suitability of rice paddies as foraging habitats for aquatic bird species is not restricted to paddies associated to wetlands. Perez-Chiscano (1975) studied the changes in avifauna that followed the transformation of agricultural land into rice paddies in Badajoz (Spain). He found a dramatic change in the avifauna composition and distribution, with 60 birds species directly associated to the paddies and several others associated to the irrigation channels; most of them not reported in that area before. Maeda (2001) studied the avifauna of rice paddies in Japan and recorded 19 waterbird species and 31 landbird species over the study year. Waterbird occurrence was largely restricted to the flooded season, confirming the direct relationship between rice agricultural practices and the associated avian community. Other particularly relevant species expected in rice paddies are fish, amphibians and reptiles among the vertebrates, and crayfish among the macroinvertebrates (Hidaka, 1998). In summary, the same vertebrate and invertebrate groups are relevant for all paddies, associated to natural or to artificial (irrigated) wetlands systems. However, from an ecological perspective the value in term of biodiversity and nature protection may be very different and thus also the protection goals. In fact, wetland associated paddies are sometime part of the natural wetland ecosystems and in several areas are directly related to nature reserves. The actual situation is obviously relevant for site-specific assessments. Therefore, in addition to the generic assessment, such as required for national registrations of commercial products, a further

Ecotoxicology of Rice Pesticides

77

site-specific assessment (e.g. within the Biodiversity Management Plans) for the use of pesticides in those protected areas with special concerns may be appropriate. This should include also a range of other management practices of importance for the protection of biodiversity and ecological values in protected wetlands, such as the paddy water regime, fertilisation, management of associated canals, etc. A frequently applied practice is to drain paddies before pesticide application. This management enhance the performance of the product, and also may reduce the risk to aquatic organisms in off-crop situations. On the other hand, draining will have a major impact on the animals living within the paddy, which usually depend on water, including for example, amphibia. Thus, in these areas good agricultural practice should be specifically designed for minimising the effects, and the overall consequences of the different alternatives should be evaluated in a holistic approach. Depending on regional demands, special emphasis could be on management plans to enhance the protection of, wetland plants, amphibia, fish, birds or plant species. 4. ECOTOXICOLOGICAL ASSESSMENT OF RICE PESTICIDES When the assessment is conducted within a regulatory framework, the test requirements, at least for the initial level, are generally specified, allowing some flexibility specifically at higher levels. As for other chemicals or pesticides, the ecotoxicological assessment of rice pesticides should be conducted through a tiered testing strategy. The initial lower tier testing is done using single-species laboratory bioassays under worst case condition. If the results of these studies indicate a risk, then additional, higher tier tests can be performed, which should produce relevant information and thus reduce uncertainties associated with the risk assessment process based on a limited set of data. They include studies to reflect more realistic exposure conditions, additional species testing and at the highest tier microcosms, mesocosms and field studies. The lower tier testing is mostly conducted on a set of “typical” laboratory species and based as much as possible on standardised testing conditions. The selected species should cover a variety of taxonomic groups representing the key ecological receptors. The endpoints are selected through a combination of ecological relevance, appropriateness and feasibility. The standardisation of testing conditions by international and/or national organisations provides the level of reproducibility required for a regulatory process and the exchange of data at the international level. Table 2 offers a summary of the main regulatory ecotoxicological requirements in the European Union under Directive 91/414/EC and in USA under FIFRA Subdivision E, Part 158; test are conducted according to the OECD, EU or USA guidelines. As shown in the Table 2, the selected species and assays present large similarities in both regulatory assessments. Mammals are not included as the mammalian toxicity assays requested for the health assessment are used for the environmental assessment. The OECD runs harmonisation programmes for the guidelines and for the mutual recognition of ecotoxicological results. OECD guidelines include

78

J. V. Tarazona and G. P. Dohmen

Table 2. Standard ecotoxicity tests required for the registration of pesticides in the EU and USA, OECD guideline are generally relevant in both regions (and worldwide) Ecotoxicity test Avian acute oral Avian dietary Avian reproduction-chronic toxicity Simulated or actual field test mammals and birds Freshwater fish acute (two species) Freshwater invertebrate acute Estuarine/marine fish, shellfish, shrimp acute Chronic toxicity test on juvenile fish (growth) Freshwater or marine/estuarine fish early life stage chronic toxicity Freshwater invertebrate life cycle chronic toxicity Effects on sediment dwelling organisms Full fish life cycle Simulated or actual aquatic field study Bioconcentration in fish Effects on algal growth Tier I seed germination – single dose Tier I seedling emergence – single dose Tier I vegetative vigor – single dose Tier I aquatic plant growth – single dose Tier II seed germination – multi-dose Tier II seedling emergence – multi-dose Tier II vegetative vigor – multi-dose Tier II aquatic plant growth – multi-dose Honey bee acute contact LD50 Honey bee acute oral LD50 Honey bee toxicity of residues on foliage Bee brood feeding test Other arthropods Acute toxicity for earthworms: artificial soil test. Chronic toxicity for earthworms: growth, reproduction and behaviour Soil non-target microorganisms Effects on other non-target organisms (flora and fauna) believed to be at risk Effects on biological methods for sewage treatment

EU guideline

USA guideline

SETAC, 1995 OECD 205 OECD 206

71-1 71-2 71-4 71-5

OECD 203, EU C1 OECD 202, EU C2

72-1 72-2 72-3

OECD 215 OECD 210

72-4a

OECD 211

72-4b

OECD 218, 219

OPPTS 850.1790 ASTM E83-93 72-5 OPPTS 850.1500

Protocol to be discussed case-by-case

72-7 OECD 305E OECD 201, EU C3

Not relevant OECD 208 OECD 201, 221 OECD 214, EPPO 170 OECD 213, EPPO 170

122-1 122-1 122-1 122-2 123-1 123-1 123-1 123-2 141-1 141-2

ICPBR Guidance documents OECD 207, EU C OECD 222 OECD 216, 217, SETAC Not defined Not defined

Ecotoxicology of Rice Pesticides

79

several of the assays for which the current EU regulation does not present a specific own guideline. In practice, most studies are principally based on OECD guidelines. However, the assays conducted under equivalent specific guidelines (e.g. USA guidance for European submissions) are usually accepted and the guidance documents (e.g. DGSANCO, 2002a, b) offer additional information on which guidelines are comparable and proposals on appropriate methodologies for conducting those assays for which a guideline is not specified in the regulation. As expected, the species selection was done under generic aspects and do not focus on the particularities of rice paddies. Nevertheless, the overall picture offers a reasonable selection of key taxonomic groups relevant for a standard first tier paddy and wetland assessment, may be with two main exceptions: aquatic macroinvertebrates and wetland plants, although there are discrepancies in this point. There are various options for higher tier testing depending on the specific concerns or uncertainties that need to be addressed. Studies can be performed simulating more realistic exposure conditions and covering processes such as degradation and dissipation of the substance. Time to event studies may help to more realistically address short-term exposure. Additional species testing reduces the uncertainty with respect to sensitivity differences between species. Population studies cover different life stages of a species. All these tests are usually performed with a single species in the laboratory. Community studies are more complex and cover a range of species. Usually such studies are conducted as micro–mesocosm or as (semi-)field studies. These studies provide the most realistic and complex results including information on indirect effects, interactions and recovery, but they are also least reproducible. Such approaches are in principle similar in the different compartments, aquatic and terrestrial. Obviously, aquatic microcosms, mesocosms and semi-field/field studies are usually employed for covering aquatic ecosystems. Higher tier soil laboratory studies with communities focus on soil microcosms with two main approaches, the use of soil cores with its biological community, such as the terrestrial model ecosystem (TME) approach (Knacker et al., 2004), and the use of artificial soil assemblages, such as the MS·3 approach, (Boleas et al., 2005a,b; Fernández et al., 2005). The next step moves directly to semi-field/field studies which may also cover other terrestrial groups, such as bees, foliar arthropods, birds and mammals. The appropriateness of these designs for covering the specific conditions of rice paddies is limited. Aquatic micro and mesocosms may be relevant for the assessment of the adjacent water bodies, but its biological structure is clearly different from that expected for the paddy. On the other hand, the information obtained in mesocosm studies with a large diversity of species will also provide relevant data on potential effects in paddies if the specific fate and climatic conditions of paddies are considered. Due to the significant proportion of rice paddies located in coastal areas, estuaries may become particularly relevant, and estuarine mesocosms have been used for higher tier testing of rice pesticides (Wirth et al., 2004). The soil microcosms are obviously of low value, as the specific paddy conditions are very different from that observed in pasture areas or arable land. The peculiarity of rice paddies conditions and the economic, social, public health and ecological relevance of this crop has conferred a significant

80

J. V. Tarazona and G. P. Dohmen

interest to the paddy situations, and specific higher tier test designs are available (Dennet et al., 2003; Sanchez et al., 2006; Sanchez-Bayo and Goka, 2006). The general principles for assessing these higher tier studies, such as those recommended in the Society of Environmental Toxicology and Chemistry (SETAC) workshops HARAP and CLASSIC (Campbell et al., 1999; Giddings et al., 2002) are generally applicable, although the specific conditions of the rice paddy should be considered. Some issues are addressed below. 4.1. Identification of relevant non-target populations and communities The use of pesticides in rice paddy may have two main objectives, as plant protection products, for improving rice yield and control pests and diseases, or for controlling vectors and pests in public health programmes. Accordingly, the problem definition or problem formulation of the risk assessment can differ significantly. They determine which organisms should be considered as target, non-target and also the protection goals of the environmental risk assessment. In this chapter, we will focus mostly on the use of rice pesticides as plant protection products. As mentioned above, the ecotoxicological assessment for vector control programmes is not significantly different in the lower tier level assessment. The main difference may be that the treatment should cover all the wetland area, not just the rice paddies, and therefore the transfer of the pesticide from the paddy to the adjacent water bodies becomes irrelevant as these water bodies will be also directly treated. As a consequence, the distinction between the paddy community and the associated wetland communities of the paddy neighbourhood is useless. At the higher tier level, significant differences may arise due to the particularities of other parts of the risk assessment protocol, basically, the efficacy assessment and the risk characterisation. Vector control programmes usually require a long-term planning approach to be effective, and the geographical scale is usually wider than for plant protection products. The assessment of long-term effects in the case of persistent pesticides and the consequences on biodiversity for repeated applications become essential. In addition, the public health benefits expected from this control are so relevant that the question is usually not whether to use a pesticide but which one should be applied at least until a significant level of control is achieved and/or other control measures are implemented. Thus, the risk characterisation tends to be a comparative risk characterisation, analysing the different alternatives and the risk/benefits of each option. Finally, in most countries these programmes are handled by public authorities, not as a private decision from the farmers, and require site-specific impact and risk assessments adapted to the local particularities (e.g. Stevens and Warren, 1995). Going back to the use of rice pesticides as plant protection products, the “classical” in-crop versus off-crop assessment is applicable, as all intended effects are related to the paddy itself and any effect outside the paddy should be considered as undesired. Some regulatory guidelines have conferred a low relevance to the paddy community (e.g. MEDRice, 2003) as heavily disturbed in-crop area, but from an ecological perspective, the paddy community is clearly relevant, as presented above, and this circumstance requires an additional consideration of the

Ecotoxicology of Rice Pesticides

81

level of protection required for the paddy and the associated communities, ensuring efficient crop production while avoiding unacceptable levels of effect. 4.2. Assessing the effects on the paddy community The complex paddy rice community is subjected to continuous changes. The typical paddies located around wetlands should be considered as modified wetland communities, typical for non-permanently flooded systems, where the vegetation cover is replaced by the crop. Rice pesticides are mostly applied during the flooding period, and therefore the in-paddy effect assessment should focus mostly on the aquatic paddy community, including sediment. Sometimes, the paddies are also drained before pesticide application with the loss of a large part of the aquatic community, modifying the risk assessment (i.e. reducing the risk of the outflow and increasing the effects on the paddy community). In addition to the contribution of the in-crop paddy community to the overall biodiversity, a major role is to serve as food for other organisms with particular relevance, for example, to many bird species. The main groups to be considered for the in-paddy assessment are phytoplankton, zooplankton, macroinvertebrates, sediment/soil microorganisms and birds. Within the crop, other wetland plants are generally considered weeds, and therefore potential targets for the pesticide, however, for endangered and/or endemic species a site-specific assessment is required. The situation is more complex with respect to phytoplankton. This is the main basis for the aquatic food chain; in addition some blue-green algae are useful due to their nitrogen fixation capability. On the other hand, from an agronomic perspective, algae compete for nutrients with the rice plant and can impede the water management; thus rice yield is usually better if strong algal growth (particularly that of large filamentous species) is avoided. Not surprisingly, the main risks for aquatic plants and algae are expected from the use of herbicides. The effects assessment usually considers that all plants within the crop area are target species for herbicides. On the other hand, endangered species are frequently found in wetland areas and paddies and, depending on the water management, a protection at the paddy level may be essential when designing the species conservation strategy. As a corollary, a generic assessment may assume a low relevance for in-paddy wetland plants, while a site-specific assessment may be required for endangered species in the prevalent areas. Some herbicides are also expected to be particularly toxic for phytoplankton, while, not surprisingly insecticides tend to be particularly risky for zooplankton species. Both communities play a key role in paddy systems but, at the same time, present a large variability regarding the special and temporal composition due to, among other factors, agronomic water management. Typical characteristics of these communities are their very high potential for rapid colonisation and/or for biological cycles with resistance forms or other mechanisms allowing the species to remain inactive during the dry season and recover quickly as soon as the conditions allow a proper development. In addition, the water inflow brings new populations, which can develop very rapidly under the conditions

82

J. V. Tarazona and G. P. Dohmen

typically prevailing in paddy systems, such as high temperature and sufficient nutrients. These characteristics also reduce the impact of rice agricultural practices where the water level is artificially controlled and the paddies are flooded or drained as needed. As a consequence, the effect assessment of pesticides within the system should consider the high potential for recovery/re-colonisation of the paddy community. Populations of copepods, cladocerans and ostracods fluctuate during the paddy-growing season in response to flooding, field drainage, ploughing and other practices (Tejada et al., 1995; Abdullah et al., 1997). There is also a tendency within taxonomic groups with communities dominated initially by phytophagous species, gradually giving way to predators and scavengers (Sanchez-Bayo and Goka, 2006). Under these circumstances it can be assumed that the ecological relevance of the plankton communities should focus on their function, while structural changes are expected to be acceptable within the crop if the function is maintained. The main consequence for that assumption is the reduction in the uncertainty/assessment factor. Among macroinvertebrates, crayfish are considered particularly relevant in rice paddies and have received a significant attention. Part of this relevance is related to its role as food source not only for wild vertebrates but also for humans. In certain areas, rice paddies are used for a commercial cultivation of crayfish, stocking crayfish in the paddy when rice reaches the green ring stage (Huner and Barr, 1991). Figure 2 shows a comparison of the acute toxicity of several pesticides for daphnids and crayfish. The information available for crayfish is too limited for allowing a range estimation for this group, while for most pesticides, several daphnid species have been tested. The figure presents for each pesticide

Acute toxicity crayfish

1000000

10000

100

1

0.01 0.01

1

100

10000

1000000

Acute toxicity daphnids Insecticides

Other pesticides

Fig. 2. Comparison of the acute toxicity of several pesticides to crayfish versus daphnids. The graph gives one value for crayfish, which is either the only available result or a geometric average. Daphnids data are presented as the lowest value (symbol) and the range (line).

Ecotoxicology of Rice Pesticides

83

the reported acute toxicity data for crayfish and the lowest and highest acute EC50 for daphnids, combining information from several databases (USEPA, EU, INIA, etc.). The figure confirms the highest toxicity of insecticides and shows that crayfish are in most cases, but not always, less sensitive than daphnids. Part of this apparent difference simply reflects the fact that the database is larger for daphnids than for crayfish. Differences of one and even two orders of magnitude in the acute toxicity of the same pesticide for daphnids are relatively frequent as observed by the bars. There are several insecticides in use at present, which are potentially toxic to fish, crayfish or which may have a negative impact on birds feeding in the paddy. Those will need a very careful evaluation and eventually a risk/benefit analysis to determine whether, when and where their use can be considered acceptable or not. The bioaccumulation of pesticides in crayfish and other aquatic invertebrates is receiving a significant attention (Chaton et al., 2002; Klosterhaus et al., 2003; Maenpaa et al., 2003). Although there are not sufficient data for sound conclusions yet; a conceptual comparison of the bioaccumulation mechanisms in fish and invertebrates suggests two potential issues that require further assessment on the role of invertebrates in bioaccumulation and transfer of pesticides into the food chain, particularly for chemicals with a relatively fast dissipation from the water column: the highest bioaccumulation potential of invertebrates compared to fish for short-term exposures due to lower depuration rates (e.g. for chemicals metabolised through cytochrome P450 related routes) and the potential for accumulation related not to the liphophilicity of the chemical but to reactivity with exoskeleton structures. The latest is an adsorption-like phenomena that although does not affect the internal body burden may contribute to the overall exposure of those predators swallowing the whole crayfish. An interesting aspect of the paddy non-target community is its beneficial role for controlling pests and vectors (Dennett et al., 2003); which provides additional arguments for protecting the aquatic community. This assessment may be compared to the beneficial role of some non-target terrestrial arthropods in other crops. A distinction between the protection goals to be applied to in-crop and off-crop non-target arthropods has been suggested (Candolfi et al., 2001) and accepted in the regulatory arena (DGSANCO, 2002a). The ecotoxicological assessment of the sediment/soil microbial community is based on functional parameters related to the nutrient cycles: respiration and nitrification. These roles are essential for a sustainability assessment. However, the evaluation is more difficult in the specific paddy situation, as the flooding conditions by themselves, independently of the potential pesticide effects, may interfere with these processes. The current guidelines (OECD, 2004a,b) and the decision criteria for agricultural soils accept a significant effect if a recovery is observed in 100 days. Similar criteria are considered acceptable for the paddy assessment while specific conditions on paddy soil should be used if become available (MEDRice, 2003). In summary, the relevant ecotoxicological effects to be specifically highlighted for the assessment of the paddy community are those related to the plankton function and the role as food source for birds. The plankton assessment can be covered through the standard ecotoxicological tests as the selected organisms

84

J. V. Tarazona and G. P. Dohmen

may be considered representative. The protection of function alone, allowing structural changes, allows a reduction of the uncertainty/assessment factors. It has been suggested that in a first tier assessment the lack of effects on representative species such as green algae and daphnids may be enough for the required level of protection and, therefore, the EC50 (alga) and NOECs (fish, Daphnia) from the standard tests without additional factors could be sufficient for the protection of plankton function (Tarazona and Sanchez, 2006). The main goal for the in-crop assessment should be the potential for recovery of the community. As a difference to the arable situation, where recovery has to be demonstrated until the following season, and considering the ecological importance of such artificial wetlands the recovery should preferably occur within the same season. It must also be considered that in some tropical countries fishes are cultured in the rice paddy, and therefore the effects on fish populations should be also covered (Abdullah et al., 1997). The evaluation of the potential effects of pesticides on fish or crayfish populations may be done on the basis of similar methodologies than those employed for the ecotoxicological assessment. 4.3. Effects on associated communities and populations The complement of the paddy community assessment must cover the “users” of the paddy environment and the surrounding ecosystems that may be exposed during or after the application. These assessments cover birds and mammals feeding in the paddy and the aquatic bodies and wetland areas that may receive the spray drift or the paddy water drainage. 4.3.1. Effects on bird populations

As explained previously, the assessment of the potential effects on birds is a critical aspect in the assessment of rice pesticides. Due to animal welfare and ethical issues the possibilities for toxicity testing on birds are limited. Typical testing strategies for birds cover two species from different families, usually Bobwhite quail (Colinus virginianus) and Mallard duck (Anas platyrhynchos). A single dose acute lethality tests, a short-term dietary test with oral exposure from food during five days and reproduction tests are presently required for covering the risk assessment of acute mortality and reproduction effects that may impair populations. This data set is considered the minimum for getting acceptable risk estimations. The endpoints to be measured in avian reproduction studies and the interpretation and extrapolation of the observed results is receiving significant attention nowadays. Several proposals have been published suggesting better alternatives for data handling (Mineau, 2005) and for using the data in risk assessments (Bennett et al., 2005; Shore et al., 2005). The bird assessment in the European Union is moving to the identification of key species, representative of specific crops and to focus the assessment in a species-specific risk characterisation for the selected species (compare DGSANCO, 2002b). The application of this approach to complex systems, such as rice paddies, with hundred of species with very different feeding habits seems to be problematic with the current

Ecotoxicology of Rice Pesticides

85

level of information. The worst case assumption of a significant number of individual birds feeding almost exclusively in the rice paddies seems to be realistic (Hafner et al., 1986; Fasola et al., 1996; Elphick, 2000), and the feeding behaviour is very variable and clearly controls the environmental exposure of birds to pesticides, with large differences even within the same family (Albanis et al., 1996). As a consequence, the environmental fate properties of the pesticide will trigger which species will receive the highest exposure. The suitability of the generic residue estimations in plants, seeds, arthropods, etc., used for other crops (e.g. DGSANCO, 2002b) must be re-evaluated depending on the pesticide application practice. For example, the generic assessment employed for a pesticide applied in drained paddies which are re-flooded immediately after treatment should be checked carefully. The direct consumption of paddy soil/sediment and the exposure via aquatic invertebrates including crayfish should be also considered. A potential concern based on the avian toxicity studies and the expected level of residues should require a further assessment, considering relevant exposure routes, such as the crop, soil/ sediment, other plants, algae, invertebrates and fish. As already mentioned, the bioaccumulation assessment should consider not only fish but also algae, zooplankton, crayfish and sediment dwelling invertebrates. 4.3.2. Effects on other terrestrial vertebrates and the use of mammalian toxicity data in ecotoxicological assessment

Birds and mammals toxicity data are usually employed for a generic assessment of all vertebrates, including amphibians and reptiles. Both groups have a greater relevance in rice paddies than mammals. The decline of amphibian populations all over the world is creating a significant concern. Depending on the species and developmental status, the main exposure route for amphibians may be water, either through the respiratory system or by dermal exposure, food or a combination of both. The basic assumption that an assessment based on fish for water exposure and birds and mammals for oral exposure is also protective for amphibians and reptiles has been demonstrated to be true for the limited number of cases in which information on data-rich chemicals has allowed a comparison. With the current level of knowledge and before specific assays are developed and standardised this assumption seems to be the only pragmatic solution to be applied also for the assessment of rice pesticides. Nevertheless, it should be clear that this is a pragmatic solution for the time being, not sufficiently scientifically established, and that this should be revisited as soon as additional information may be provided. As a direct consequence, a generic risk assessment for mammals based on worst case exposure estimations should be conducted for rice pesticides, regardless of the relevance of mammals feeding on the paddy systems, as a method for covering the oral exposure of vertebrates in general. 4.3.3. Effects on associated wetlands

In several areas of Europe and elsewhere, rice paddies are directly connected to wetland areas of high ecological value. These areas include some of the most significant biodiversity reservoirs, receiving the highest possible level of protection.

86

J. V. Tarazona and G. P. Dohmen

A direct relationship between the use of pesticides in rice paddies and the contamination of these natural reservoirs has been described (Manosa et al., 2001). In addition to drift and drainage, the direct discharge of paddy water into the surrounding water bodies constitutes the main exposure route for these wetland ecosystems. As these systems mostly comprise low-flow systems, the dilution is limited and the concentrations expected at the discharge sites may reach levels that clearly require a specific assessment, as depending on the pesticide fate properties and the management situations this route may be much more relevant than spray drift. A standard risk assessment for off-crop areas focusing on the effects on both structure and function of the ecosystem is appropriate for these systems. This is the same situation addressed for other crops and their adjacent water bodies; thus, the same methodologies may be applied. With respect to nontarget plant testing this may need some adaptation to cover wetland plants, too. Nevertheless, there is a specific aspect with paddy rice concerning its relevance for rare and/or threatened species, which is not covered in a generic assessment for registration purposes. The use of pesticides in areas where endangered and/or rare species may be at risk requires an additional specific evaluation for those particular species in the form of an impact or a risk assessment in addition to the generic regulatory evaluation. In fact, this is a common practice in rice cultivations located in the vicinity of ecological reserves; such assessments on “endangered species” in specific areas is nowadays also practiced in the USA. The effect assessment for wetland plants should be at least partially covered in the generic risk assessment. The assessment of pesticide risks for non-target plants has traditionally differentiated aquatic and terrestrial systems. The aquatic compartment has been covered by algae and one aquatic vascular plants, Lemna spp., as species selected for the standard ecotoxicity tests in the first tier. The risk assessment for terrestrial plants has been introduced later on, with the development of specific proposals. Basically, the assessment is only relevant for herbicides, as other pesticides show in general low toxicity for plants. The generic assessment usually considers spray drift in the vicinity of the crop as the major relevant exposure route. The possibility for medium-term transport related to volatility and subsequent atmospheric deposition in other areas has also been considered in certain cases. Both routes may be relevant in rice paddies, however, in most cases the already mentioned discharge of contaminated paddy water into the wetland would be the predominant path of potential contamination. This route of exposure may be reduced by spraying the herbicide on the drained paddy, which is often done anyway to increase efficacy of the treatment. There are no standard ecotoxicity tests, other than those on Lemna spp., for covering this potential risk, however the evaluation of the efficacy, selectivity and mode of action of the herbicide may provide enough information for this assessment. Basically, the required information is a set of toxicity thresholds or concentration/response relationships for a set of wetland relevant species covering a wide range of taxonomic groups and considering exposures from spray/drift, from water and from soil/sediment when relevant, which may produce enough information even for the application of probabilistic risk assessments.

Ecotoxicology of Rice Pesticides

87

As explained above, if endangered wetland species are present in the area, a specific assessment at the local level should be conducted and methods have been developed for this ecotoxicological evaluation (e.g. Luo and Ikeda, 2005). 4.4. Proposals for testing strategies and future needs Due to the complexity of rice paddies, the ecotoxicological assessment of rice pesticides and its further development into a risk characterisation can be highly benefited by the development of a specific conceptual model. A methodology based on versatile conceptual models which are adapted to the characteristics of each chemical were initially developed for terrestrial systems (Tarazona et al., 2002; Tarazona and Vega, 2002) and has been adapted to the paddy situation for the assessment of rice pesticides (Tarazona and Sanchez, 2006). The option covers a full set of possibilities, maximising the use of first tier data whenever possible. This proposal has been completed with the development/adaptation of specific testing proposals, including off-site toxicity testing of treated paddy water and rice paddy mesocosms-type studies (Sanchez et al., 2006). The use of experimental rice paddies, either in ad hoc facilities or using compartments build within rice paddies, offers large possibilities for conducting higher tier ecotoxicity tests (Dennet et al., 2003; Sanchez-Bayo and Goka, 2006) and it is feasible to combine fate and ecotoxicological parameters in semi-field paddy assays measuring simultaneously realistic exposure estimations and the effects on the paddy community (Sanchez et al., 2006). The development of some guidance for conducting ecotoxicological assays on experimental paddies should be considered for facilitating the regulatory use of these tools. In addition, one of the main needs for the future is the development of scenarios and exposure models for birds feeding on the paddy, reviewing the extensive literature available on the ecology of most of the relevant species.

5. CONCLUSIONS The ecotoxicological assessment of rice pesticides is particularly complex due to the specificities of this crop where aquatic and terrestrial communities are mixed in a human-managed wetland-type agrobiosystem. Any assessment must cover at least three main aspects, potential effects on the paddy community, potential effects on associated wetlands and water bodies, and the potential risks for vertebrates, particularly birds, feeding on the paddy. The initial lower tier assessment may be conducted using the generic approaches derived for other crops. For a higher tier risk assessment it is recommended to develop a specific conceptual model for identifying the key elements that should be addressed. Proposals for constructing these models on the basis of the characteristics of the pesticide are available and may allow a proper selection of the essential combinations of: compartment → exposure route → ecological receptor

88

J. V. Tarazona and G. P. Dohmen

These combinations can be then prioritised and should be considered in the refinement. Specific tools, and in particular, experimental paddies, which can be easily created in rice fields, offer a perfect higher tier experimental tool for assessing exposure, including bioaccumulation, and effects under realistic conditions. Further site-specific considerations and management practices may become relevant with respect to nature conservation issues or where dual use of paddies for (cray)fish production is intended. REFERENCES Abdullah, A. R., Bajet, C. M., Matin, M. A., Nhan, D. D. and Sulaiman, A. H. (1997). Ecotoxicology of pesticides in the tropical paddy field ecosystem. Environ. Toxicol. Chem. 16, 59–70. Albanis, T. A., Hela, D., Papakostas, G. and Goutner, V. (1996). Concentration and bioaccumulation of organochlorine pesticide residues in herons and their prey in wetlands of Thermaikos Gulf, Macedonia, Greece. Sci. Total Environ. 182, 11–19. Bennett, R. S., Dewhurst, I. C., Fairbrother, A., Hart, A. D., Hooper, M. J., Leopold, A., Mineau, P., Mortensen, S. R., Shore, R. F. and Springer, T. A. (2005). A new interpretation of avian and mammalian reproduction toxicity test data in ecological risk assessment. Ecotoxicology 14, 801–815. Boleas, S., Alonso, C., Pro, J., Babín, M. M., Fernández, C., Carbonell, G. and Tarazona, J. V. (2005a). Effects of sulfachlorpyridazine in MS·3-arable land: A multispecies soil system for assessing the environmental fate and effects of veterinary medicines. Environ. Toxicol. Chem. 24, 811–819. Boleas, S., Alonso, C., Pro, J., Fernández, C., Carbonell, G. and Tarazona, J. V. (2005b). Toxicity of the antimicrobial oxytetracycline to soil organisms in a multispecies-soil system (MS·3) and influence of manure co-addition. J. Hazard. Mater. 122, 233–241. Bro-Rasmussen, F., Calow, P., Canton, J. H., Chambers, P. L., Silva Fernandes, A., Hoffmann, L., Jouany, J. M., Klein, W., Persoone, G., Scoullos, M., Tarazona, J. V. and Vighi, M. (1994). EEC water quality objectives for chemicals dangerous to aquatic environment (List 1). Rev. Environ. Contam. Toxicol. 137, 83–110. Campbell, P. J., Arnold, D. J. S., Brock, T. C. M., Grandy, N. J., Heger, W., Heimbach, F., Maund, S. J. and Streloke, M. (Eds), (1999). Guidance Document on Higher-tier Aquatic Risk Assessment for Pesticides (HARAP). SETAC-Europe Publications, Brussels, Belgium. Candolfi, M. P., Barrett, K. L., Campbell, P. J., Forster, R., Grandy, N., Huet, M. C., Lewis, G., Oomen, P. A., Schmuck, R. and Vogt, H. (Eds), (2001). Guidance document on regulatory testing and risk assessment procedures for plant protection products with non-target arthropods. From the ESCORT 2 workshop (p. 46). SETAC, Pensacola. Castillo, L. E., De La Cruz, E. and Ruepert, C. (1997). Ecotoxicology and pesticides in tropical aquatic ecosystems of Central America. Environ. Toxicol. Chem. 16, 41–51. Chaton, P. F., Ravanel, P., Tissut, M. and Meyran, J. C. (2002). Toxicity and bioaccumulation of fipronil in the nontarget arthropodan fauna associated with subalpine mosquito breeding sites. Ecotoxicol. Environ. Saf. 52, 8–12. CSTEE (2004). Opinion on the Setting of Environmental Quality Standards for the Priority Substances included in Annex X of Directive 2000/60/EC in Accordance with Article 16 thereof (p. 32). CSTEE, SANCO C7/GF/csteeop/ WFD/280504 D(04). Brussels. Dennett, J. A., Bernhardt, J. L. and Meisch, M. V. (2003). Operational note effects of fipronil and lambda-cyhalothrin against larval Anopheles quadrimaculatus and nontarget aquatic mosquito predators in Arkansas small rice plots. J. Am. Mosq. Control Assoc. 19, 172–174. DGSANCO (2002a). Guidance Document on Terrestrial Ecotoxicology under Council Directive 91/414/EEC. SANCO, 10329/2002rev.2 final. DGSANCO (2002b). Guidance Document on Risk Assessment for Birds and Mammals Under Council Directive 91/414/ EEC. SANCO/4145/2000rev.6 – final. Elphick, C. S. (2000). Functional equivalency between rice fields and semi-natural wetland habitats. Conserv. Biol. 14, 181–191.

Ecotoxicology of Rice Pesticides

89

Fasola, M., Canova, L. and Saino, N. (1996). Rice fields support a large portion of herons breeding in the Mediterranean region. Colon. Waterbirds 19, 129–134. Fernández, M. D., Cagigal, E., Vega, M. M., Urzelai, A., Babín, M., Pro, J. and Tarazona, J. V. (2005). Ecological risk assessment of contaminated soils through direct toxicity assessment. Ecotoxicol. Environ. Saf. 62, 174–184. Giddings, J. M., Brock, T. C. M., Heger, W., Heimbach, F., Maund, S. J., Norman, S. M., Ratte, H. T., Schäfers, C. and Streloke, M. (Eds), (2002). Community-Level Aquatic System Studies— Interpretation Criteria (CLASSIC). SETAC-Europe Publications, Brussels, Belgium. Hafner, H., Dugan, P. J. and Boy, V. (1986). Use of artificial and natural wetlands as feeding sites by little egrets (Egretta garzetta) in the Camargue, southern France. Colon. Waterbirds 9, 149–154. Hidaka, K. (1998). Biodiversity conservation and environmentally regenerated farming system in rice paddy fields. Jpn. Ecol. 48, 167–178. Huner, J. V. and Barr, J. E. (1991). Red Swamp Crawfish: Biology and Exploitation. LSA-T-90003. The Louisiana Sea Grant College Program, Baton Rouge, LA, USA. Klosterhaus, S. L., DiPinto, L. M. and Chandler, G. T. (2003). A comparative assessment of azinphosmethyl bioaccumulation and toxicity in two estuarine meiobenthic harpacticoid copepods. Environ. Toxicol. Chem. 22, 2960–2968. Knacker, T., van Gestel, C. A. M., Jones, S. E., Soares, A. M. V. M., Schallnas, H.-J., Förster, B. and Edwards, C.A. (2004). Ring-testing and field-validation of a Terrestrial Model Ecosystem (TME) – An instrument for testing potentially harmful substances: Conceptual approach and study design. Ecotoxicology 13, 9–27. Luo, X. Y. and Ikeda, H. (2005). Effects of four rice herbicides on seed germination and seedling growth of a threatened vascular plant Penthorum chinense Pursh. Bull. Environ. Contam. Toxicol. 75, 382–389. Maeda, T. (2001). Patterns of bird abundance and habitat use in rice fields of the Kanto Plain, central Japan. Ecol. Res. 16, 569–585. Maenpaa, K. A., Sormunen, A. J. and Kukkonen, J. V. (2003). Bioaccumulation and toxicity of sediment associated herbicides (ioxynil, pendimethalin, and bentazone) in Lumbriculus variegatus (Oligochaeta) and Chironomus riparius (Insecta). Ecotoxicol. Environ. Saf. 56, 398–410. Manosa, S., Mateo, R. and Guitart, R. (2001). A review of the effects of agricultural and industrial contamination on the Ebro delta biota and wildlife. Environ. Monit. Assess. 71, 187–205. MEDRice (2003). Guidance Document for Environmental Risk Assessments of Active Substances used on Rice in the EU for Annex I Inclusion (108 pp.). Document prepared by Working Group on MED-Rice, EU Document Reference SANCO/1090/2000 – rev.1, Brussels, June 2003. Mineau, P. (2005). A review and analysis of study endpoints relevant to the assessment of “long term” pesticide toxicity in avian and mammalian wildlife. Ecotoxicology 14, 775–799. OECD (2004a). Technical guidance 216: Soil Microorganisms, Nitrogen Mineralization Test. OECD, Paris. OECD (2004b). Technical guidance 217: Soil Microorganisms, Carbon Mineralization Test, OECD, Paris. Perez-Chiscano, J. L. (1975). Avifauna de los cultivos de regadios del Guadiana (Badajoz). Ardeola 21, 753–794. Pugh, D. M. and Tarazona, J. V. (Eds), (1988). Regulation for Chemical Safety in Europe: Analysis, Comment and Criticism. Environment & Policy Series (vol. 15, p. 210). Kluwer Academic Publishers, Dordrecht/Boston/London. Sanchez, P., Kubitza, J., Dohmen, P. G. and Tarazona, J. V. (2006). Aquatic risk assessment of the new rice herbicide profoxydim. Environ. Pollut. 142, 181–189. Sanchez-Bayo, F. and Goka, K. (2006). Ecological effects of the insecticide imidacloprid and a pollutant from antidandruff shampoo in experimental rice fields. Environ. Toxicol. Chem. 25, 1677–1687. Shore, R. F., Crocker, D. R., Akcakaya, H. R., Bennett, R. S., Chapman, P. F., Clook, M., Crane, M., Dewhurst, I. C., Edwards, P. J., Fairbrother, A., Ferson, S., Fischer, D., Hart, A. D., Holmes, M., Hooper, M. J., Lavine, M., Leopold, A., Luttik, R., Mineau, P., Moore, D. R., Mortenson, S. R., Noble, D. G., O’Connor, R. J., Roelofs, W., Sibly, R. M., Smith, G. C., Spendiff, M., Springer, T. A., Thompson, H. M. and Topping, C. (2005). Case study part 1: How to calculate

90

J. V. Tarazona and G. P. Dohmen

appropriate deterministic long-term toxicity to exposure ratios (TERs) for birds and mammals. Ecotoxicology 14, 877–893. SSC (2003). Second report on the harmonization of risk assessment procedures. Scientific Steering Committee, DGSANCO, EC, Brussels. Stevens, M. M. and Warren, G. N. (1995). Control of chironomid larvae (Diptera: Chironomidae) in establishing rice crops using starch-based chlorpyrifos pellets. J. Am. Mosq. Control Assoc. 11, 206–210. Tarazona, J. V. (1997). The identification of thresholds of acceptability and danger: The biological route. Arch. Toxicol. Sup. 19, 137–146. Tarazona, J. V. and Sanchez, P. (2006). Development of an innovative conceptual model and a tiered testing strategy for the ecological risk assessment of rice pesticides. Paddy Water Environ. 4, 53–59. Tarazona, J. V., Hund, K., Jager, T., S-Salonen, M., Soares, A. M. V. M., Skaare, J. U. and Vigui, M. (2002). Standardizing chemical risk assessment, at last. Nature 415, 14. Tarazona, J. V. and Vega, M. M. (2002). Hazard and risk assessment of chemicals for terrestrial ecosystems. Toxicol. 181–182, 187–191. Tejada, A. W., Varca, L. M., Calumpang, S. M. F., Ocampo, P. P., Medina, M. J. B., Bajet, C. M., Paningbatan, E. B., Medina, J. R., Justo, V. P., Habito, C. L., Martinez, M. R. and Magallona, E. D. (1995). Assessment of the environmental impact of pesticides in paddy rice production. In: P. L. Pingali and P. A. Roger (Eds), Impact of Pesticides on Farmer Health and the Rice Environment (pp. 149–179). Kluwer, Norwell, MA, USA. USEPA (1998). Guidelines for Ecological Risk Assessment. EPA/630/R-95/002F. Washington D.C. Wirth, E. F., Pennington, P. L., Lawton, J. C., DeLorenzo, M. E., Bearden, D., Shaddrix, B., Sivertsen, S. and Fulton, M. H. (2004). The effects of the contemporary-use insecticide (fipronil) in an estuarine mesocosm. Environ. Pollut. 131, 365–371.

Pesticide Risk Assessment in Rice Paddies: Theory and Practice E. Capri and D.G. Karpouzas (editors) © 2008 Elsevier B.V. All rights reserved

Chapter 6

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis S. Cervelli1 and R. Jackson2 1Institute

of Environmental Studies, CNR, Pisa, Italy 2Dow Agro Sciences, UK

Contents 1. Introduction 2. Environmental fate data requirements 3. Definition of scenarios for exposure assessment 4. Predicted environmental concentrations in surface water and sediment 4.1. Step 1a (no degradation, no adsorption) 4.2. Step 1b (degradation, no adsorption) 4.3. Step 1c (degradation, adsorption) 5. Predicted environmental concentrations in groundwater 6. Predicted environmental concentrations in soil 6.1. Application to the flooded field 6.2. Application to the drained field 7. Step 2 exposure assessment 7.1. Surface water, sediment and soil 7.1.1. Closed rice system 7.1.2. Flooding period 7.2. Groundwater 7.3. Sensitivity analysis 7.4. Example calculations 7.5. Validation 8. Conclusions References

91 92 92 94 96 97 97 99 101 101 102 102 102 102 105 109 112 115 120 122 123

1. INTRODUCTION In order to evaluate the potential environmental risk from the use of pesticides in rice, it is necessary to understand the fate of the pesticide under the unique environmental conditions used in rice culture. From this understanding, a scheme can then be developed to determine the magnitude of the exposure that non-target organisms will be subjected. Under the direction of the European Commission, a working group was set up in 1999 to develop a Guidance Document for the environmental risk assessment of plant protection products (PPP) in rice in Europe. The aim was to develop a first tier of assessment to allow decision making for inclusion in Annex I of the Council Directive 91/414/EEC. The Guidance Document was released in June 2003 (MED-Rice, 2000). In this chapter, the main aspects of this guidance document are summarised.

92

S. Cervelli and R. Jackson

2. ENVIRONMENTAL FATE DATA REQUIREMENTS A conventional assessment of the environmental fate of a pesticide involves an evaluation of the route and rate of degradation in soil and aquatic environments and the potential to be transported from the site of application to other “offtarget” areas (e.g. groundwater, surface water, air). In essence, the same principles apply to rice pesticides but the main difference is that the flooded conditions of rice culture can lead to very different behaviour compared to other agricultural crops. The flooded conditions can alter the physico-chemical and microbial composition of the soil leading to differences in both route and rate of degradation of pesticides. A major effect of flooding is that the field becomes a compartmentalised system with varying degrees of anaerobic and aerobic conditions existing in the water and soil layers (Schnell et al., 2000). Furthermore, these conditions will vary throughout the rice growing season as flooding and draining are carried out as part of normal agronomic practices (e.g. many herbicides are applied after draining the rice field to expose weeds). The presence of flooded saturated conditions will also increase the potential for the pesticide to move laterally away from the treated field to surface water, particularly in response to controlled water management procedures. In addition to chemical and microbial degradation, for some pesticides, photolysis can be more important in rice than in other crops because of the aqueous environment and the strong solar irradiance in rice growing regions of the world. The guidelines developed by SETAC (SETAC, 1995), and the OECD (OECD, 2002), for aerobic and anaerobic transformation in soil can be adapted to provide information in flooded soil. The OECD guideline recommends an aerobic flooded soil degradation study to address the degradation of pesticides under paddy field conditions. Following this protocol, a soil study using a representative rice growing soil should be conducted. Additionally, small-scale or full-scale outdoor dissipation studies may give useful information for certain compounds (e.g. where photolysis may be important). 3. DEFINITION OF SCENARIOS FOR EXPOSURE ASSESSMENT Rice in Europe is grown under a Mediterranean climate characterised by warm, dry, clear days and a long growing season (a more extensive and deeper description of agronomic and weather conditions has been given in Chapter 1). Rice is grown mostly on fine-textured, poorly drained soils (clays, silty clays and silty clay loams) but some sandy soils are also used for rice cultivation in areas of Italy and Spain. From a survey of agronomic conditions in different European countries, two European standard scenarios were developed to evaluate contamination of surface waters and leaching to ground water. The scenarios consider impermeable soils with high clay content on the one hand, and more permeable sandy and low organic matter soils on the other. The essential parameters of the two scenarios are shown in Table 1.

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

93

Table 1. Definition of the two Med-Rice scenarios (MED-Rice, 2000) Characteristic Soils: texture % clay % o.m. (% o.c.) pH Water level Water velocity: outflow (L s
1 ha
1) field (L s
1 ha
1) Flooding conditions Time of closure of field (days) Depth of drainage channel (m) Crop rotation Infiltration (leakage) rate (mm day
1) Evapotranspiration rate (mm day
1) Usage of outflow water Temperature (ºC) Conditions in soil

Scenario 1

Scenario 2

Clayey 30 3 (1.8) 8 10 cm

Sandy 5 1.5 (0.9) 6 10 cm

0.5 1.8 May–August 5 1 No 1 10 No 20 Aerobic

0.5 2.8 May–August 5 1 No 10 10 No 20 Aerobic

The first scenario represents more vulnerable conditions with regard to surface water exposure, whereas the second scenario represents conditions vulnerable to leaching and groundwater contamination. These two scenarios are characterised by different infiltration rates: 1 mm day
1 for the clayey scenario and 10 mm day
1 for the sandy scenario. Both scenarios are intended to represent the two extremes of actual situations (realistic worst case), and most real situations will be in between these two. In common with other risk assessment procedures for pesticides, a stepwise approach was developed starting with a relatively simple and conservative step, up to a more sophisticated and realistic assessment (Figure 1). The first step is based on simple assumptions regarding the geometry and size of rice fields and adjacent surface water bodies as well as the distribution and dissipation of the product in the environment. Three environmental compartments are considered, paddy water (including soil), surface water (including sediment) and groundwater (Figure 2). At higher steps, modelling with more complex simulation models or site specific considerations are proposed. The need for further assessment at each step will depend upon the results of the risk assessment for relevant non-target organisms and groundwater contamination. For the three compartments, the predicted environmental concentrations (PECs) are calculated, and these values are then used with ecotoxicity data in the risk assessment. The various environmental fate studies must provide sufficient information regarding the degradation and sorption of the active substance in different environmental matrices.

94

S. Cervelli and R. Jackson

Step 1 Loading based on total dose

Use

yes

safe?

no Step 2 Model calculations for paddy rice scenario using advanced mathematical models

no further work

yes

Use safe?

no

Step 3 Refined leaching modelling incl. site specific considerations and/or geographical /statistical approaches

Fig. 1. Generalised stepwise approach for the risk assessment.

4. PREDICTED ENVIRONMENTAL CONCENTRATIONS IN SURFACE WATER AND SEDIMENT It is important to distinguish between the “target” and “non-target” compartments in rice culture. The term “surface water” refers to water in non-target areas such as irrigation/drainage canals and other water bodies whereas the term “paddy water” refers to water in the cropped field (target area). Analogously, the term “sediment” refers to sediment in non-target area and the term “soil” is used only for the cropped field. Groundwater is defined as water in the saturated condition 1 m below the soil surface (FOCUS, 2000). A simplified scheme of the proposed step 1 approach for calculating PEC values is shown in Figure 3. It is divided into three sub-steps which differ in the exclusion or inclusion of degradation and sorption.

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis paddy water soil

drained water

ground water

canal water sediments

canal

95

paddy field

Fig. 2. A schematic representation of the rice environment.

Using the information in Table 1, the scenario used for the calculation of PEC values at step 1 assumes 1 ha square field with a canal on one side separated by a 1 m wide levee. Pesticide loadings to the surface water body (canal) can occur via spray drift at the time of application and outflow of water from the field five days after application (five days is considered to be a normal holding period for water following pesticide application). The following definitions are used for the calculations: a. b. c. d. e. f. g. h. i. j.

area of rice field: depth of water in field: depth of receiving canals: dosage: fraction deposited on paddy water: fraction drift to adjacent surface waters: fraction drift to adjacent surface waters: dilution factor (DF) time of field closure: mixing in surface water:

area  104 m2 depthwater  0.1 m depthcanal  1 m Dose, according to the label (g ha
1) fdep  1 (conservative assumption) fdrift  0.0277 fdrift  0.332 (for aerial application) fdilution  10 tclose  5 days complete.

The values for drift are based on the most recent data issued by the FOCUS group on surface water scenarios (FOCUS, 2001). Note that the deposition fraction ( fdep) can be adjusted to account for crop interception, depending on the growth stage of the crop. The values used in the FOCUS groundwater guidance for cereals may be appropriate. All water concentrations are expressed as
g L
1, and all soil and sediment concentrations are expressed as
g kg
1.

96

S. Cervelli and R. Jackson

START

STEP 1 a

Loading based on outflow and drift No degradation, no adsorption

yes Use Safe no

STEP 1 b

Loading as in step 1 a plus degradation, no adsorption

yes

Use Safe

No Further Work

no

STEP 1 c

Loading as in step 1 a plus degradation and adsorption

yes

Use Safe no

Further Steps

Fig. 3. Flow chart for step 1 calculations.

4.1. Step 1a (no degradation, no adsorption) Step 1a represents the simplest case. Only initial concentrations are derived and neither degradation nor adsorption are considered. Using the previous assumptions, the predicted initial concentration in paddy water (PECpw,initial) is PECpw, initial  0.1⋅

fdep ⋅ Dose depth water

.

(1)

The initial concentration in the receiving canal due to spray drift at the time of application (PECsw,drift,initial) is PECsw, drift, initial  0.1 ⋅

Dose ⋅ fdrift . depth canal

(2)

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

97

After discharging of water from the field to canal the estimated concentration is PECsw, initial 

PECsw, drift, initial ⋅ fdilution  PECpw, initial 1 fdilution

(3)

.

4.2. Step 1b (degradation, no adsorption) During the 5-day holding period for water in the field, degradation of the pesticide will reduce its concentration in the paddy water. A reduction will also take place in the canal. The degradation in the paddy water is described using an appropriate “half-life” (i.e. the time taken for 50% of the initial amount of pesticide to degrade or DT50) value from the aerobic flooded soil degradation or field dissipation studies. Degradation in the canal is described using the DT50 from water/sediment studies. If no separate DT50 values in water and sediment are reported, then the total DT50 is used. Assuming first order dissipation kinetics, the concentration in the paddy water at the end of the 5-day closure time is described by
tclose ⋅ln( 2 ) / DT50 pw

PECpw (tclose )  PEC pw, initial ⋅ e

.

(4)

During the same period of time, the concentration in the canal (resulting from drift) will have decreased and is described by
tclose ⋅ln( 2 ) / DT50 sw

PECsw, drift (tclose )  PECsw, drift , initial ⋅ e

(5)

.

Note that the water flow in the canal is not considered as a dissipation process at this step. At the end of the 5-day closure period, the concentration in the canal due to drift and outflow is given by PECsw (tclose ) 

PECsw, drift (tclose ) ⋅ fdilution  PECpw (tclose ) 1  fdilution

.

(6)

4.3. Step 1c (degradation, adsorption) In step 1c, adsorption to soil and sediment is taken into consideration. At this step it is necessary to use the actual degradation half-life in water, in order not to double account for the dissipation from water due to partitioning to the soil or sediment. In many cases, this degradation rate is not known, so step 1b should be used with the “dissipation” DT50 value. Adsorption on the sediment is assumed to be an instantaneous process determined by the adsorption coefficient KD (dm3 kg
1) CS  K D ⋅ C W

(7)

where CS is the amount of compound adsorbed on sediment (
g kg
1) and CW the concentration in the water phase (
g L
1). Due to the dependence of the

98

S. Cervelli and R. Jackson

adsorption on the amount of water and soil or sediment, the following equations are used Fdissolved 

depth water depth water  depth soil / sed ⋅ BD ⋅ K D Fadsorbed  1
Fdissolved

(8) (9)

where Fdissolved is the mass fraction dissolved in the water phase and Fadsorbed is the mass fraction adsorbed on the solid phase. BD is the bulk density (kg dm
3) and depthsoil/sed is the depth of the soil or sediment (5 cm). The dissolved material will be subject to outflow into the receiving canals, while the adsorbed material will remain in the paddy. The initial concentration in paddy water is calculated by PECpw, initial  0.1⋅

fdep ⋅ Dose ⋅ Fdissolved depth water

(10)

The partitioning between water and sediment is taken into account for the spray drift entry into surface water. For the surface water PECsw, drift , initial  0.1⋅

fdrift ⋅ Dose ⋅ Fdissolved depth canal

(11)

fdrift ⋅ Dose ⋅ Fadsorbed depth sed ⋅ BDsed

(12)

and for the sediment PECsed , drift , inital  0.1⋅

Taking into account the degradation in the water and in the soil during the closure time of the field, for the paddy water
tclose ⋅ln( 2 ) / DT50 pw

PECpw (tclose )  PEC pw, initial ⋅ e

(13)

Similarly, for the surface water
tclose ⋅ln( 2 ) / DT500 sw

PECsw, drift (tclose )  PECsw, drift , initial ⋅ e

(14)

and for the sediment
tclose ⋅ln( 2 ) / DT50 sed

PECsed , drift (tclose )  PECsed , drift , initial ⋅ e

(15)

The resulting total concentration in the surface water due to the contribution of outflow and drift is PECsw (tclose ) 

PECsw, drift (t close ) ⋅ fact dilution  PECpw (tclose ) 1 fact dilution

(16)

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

99

and the resulting total concentration in the sediment PECsed (tclose )  PECsed ,drift (tclose ) 

PEC pw (tclose ) ⋅ depth water ⋅ Fsorbed fact dilution ⋅ depth sed ⋅ BDsed

(17)

At the end of the 5-day closure period, the PECsw and PECsed will be equal to PECsw ( tclose ) and PECsed ( t ) , respectively. For ecological risk assessment, time close weighted average concentrations (TWA) are often required. The general formula to calculate TWAs in the case of first order kinetics for the water phase is
T ⋅ln( 2 ) / DT50 sw

TWA sw (T ) 

PECsw, initial ⋅ (1
e

)

(18)

T ⋅ ln(2) / DT50 sw

and for the sediment phase TWA sed (T ) 

PECsed , initial ⋅ (1 − e

− T ⋅ln( 2 ) / DT50 sed

T ⋅ ln(2) /DT50 sed

)

.

(19)

5. PREDICTED ENVIRONMENTAL CONCENTRATIONS IN GROUNDWATER In Europe, rice is mainly grown on fine-textured, poorly drained soils and, therefore, the transport of pesticides into groundwater is often limited. A simple screening calculation has therefore been developed for the calculation of PECs in groundwater, which should lead to a conservative assessment of leaching potential. Following application to the paddy water surface, it is assumed that the applied pesticide partitions instantly between the water phase and the soil phase of the paddy field, according to the partition coefficient KD (equation (7)). Only the amount dissolved in the water phase is available for leaching, whereas the adsorbed fraction will remain in the upper soil layer. The dissolved fraction is subjected to both degradation and adsorption processes. The initial concentration in paddy water is given by equation (10) and the concentration at the end of the 5-day closure period is given by equation (4). During this period, the mass of the pesticide available for leaching is M leak , field 

TWA pw ⋅ tclose ⋅ leakage 100

(20)

where
tclose ⋅ln( 2 ) / DT50 pw

TWA pw  PECpw, initial ⋅

(1
e

) ⋅ DT50 pw

tclose ⋅ ln(2)

(21)

During the time period between the end of the 5-day closure period and the time when the paddy is permanently drained prior to harvest (called the flooding

100

S. Cervelli and R. Jackson

period) the decrease in concentration is determined both by degradation and the rate of water flowing out of the paddy. Therefore, at the end of the flooding period  ln( 2 ) 
tflood ⋅   outflowrate  DT50 pw 

PECpw ( end _ flood )  PECpw (tclose ) ⋅ e

(22)

where tflood is the total time of flooding (assumed to be 120 days for this calculation), and outflowrate  (86,400 outflow)/1,000,000, as there are 86,400 s per day and 1,000,000 L of paddy water per hectare. Finally, the mass of the product potentially available for leaching during the flooding period will be M leak , flood 

TWA pw, flood ⋅ tflood ⋅ leakage 100

(23)

where

TWA pw, flood  PECpw (tclose ) ⋅

 ln( 2 ) 
tflood ⋅  outflowrate  DT50 pw 

1
e .  ln(2)  tflood ⋅   outflowrate  DT50 pw 

(24)

The total mass available for leaching may then be calculated M leak  M leak , field  M leak , flood .

(25)

In the next step of the calculation, the retardation due to adsorption and degradation in subsoil layers is considered. This is done by using a retardation factor given by Ri  1

BD ⋅ K D ⋅ bioi


(26)

where i is the horizon number of the soil, bioi is a biofactor to account for the decrease in organic matter, and hence adsorption, with depth (1 in the 0–300 mm layer, 0.5 in the 300–600 mm layer and 0.3 in the 600–1000 mm layer) and
is the saturated volumetric water content of the soil (m3 m
3). The value of
for a clayey soil (Scenario 1) is 0.44, and for a sandy soil (Scenario 2) is 0.39. The residence time of water in a horizon (tres) can be calculated from the leakage rate and the horizon depth, given in mm, according to the equation

ti ,res  Ri ⋅

depth i ⋅
. leakage

(27)

The residence time is the time during which the pesticide degrades in a horizon. The remaining amount of pesticide is then transferred to the next horizon. The amount left in the third horizon in g ha
1 is then converted into a

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

101

percolate concentration by dividing by the amount of percolate water per year. The mass of pesticide moving from the 0–300 mm horizon to the 300–600 mm horizon is
t1,res ln( 2 ) / DT50 soil

M leak ( 0 − 300 )  M leak ⋅ e

(28)

where Mleak is calculated from equation (25). The mass of pesticide moving from the 300–600 mm horizon to the 600–1000 mm horizon will be given by
t2 , res ln( 2 ) / DT50 soil ⋅0.5

M leak ( 300 − 600 )  M leak ( 0 − 300 ) ⋅ e

(29)

and the mass of pesticide moving below the 1000 mm horizon will be
t3 , res ln( 2 ) / DT50 soil ⋅0.3

M leak (1000 )  M leak ( 300 − 600 ) ⋅ e

(30)

Finally, the average concentration in the groundwater after one year is computed PECpgw (t  365) 

M leak (1000 ) ⋅ 100 365 ⋅ leakage

(31)

.

6. PREDICTED ENVIRONMENTAL CONCENTRATIONS IN SOIL For soil, at step 1 two different situations can arise, one where the application is made to the flooded field and one where the application is made to a drained field which is then re-flooded at a later date. The latter case is common for many foliar-acting herbicides where it is necessary to expose as much of the weed as possible to obtain maximum efficacy.

6.1. Application to the flooded field In this step, an instantaneous partition between soil and water is assumed, as described for step 1c of the surface water calculation, and the initial concentration in soil is given by PECsoil, initial  0.1⋅

fdep ⋅ Dose ⋅ Fadsorbed depth soil ⋅ BDsoil

.

(32)

At any time t, the concentration is
t ⋅ln( 2 ) / DT50 soil

PECsoil (t )  PECsoil, initial ⋅ e

(33)

102

S. Cervelli and R. Jackson

and the time weighted concentration
t ⋅ln( 2 ) / DT50 soil

TWA soil (t ) 

PECsoil, initial ⋅ (1
e

t ⋅ ln(2) /DT50 soil

)

.

(34)

6.2. Application to the drained field When the substance is applied to the drained field the assumptions are the same as for other crops, and the assessment of exposure will follow the FOCUS Group recommendations. Therefore, in this case the initial PECsoil will be PECsoil, initial  0.1⋅

fdep ⋅ Dose depth soil ⋅ BDsoil

.

(35)

7. STEP 2 EXPOSURE ASSESSMENT As shown in Figure 1, a more realistic step 2 analysis is required if step 1 indicates that there is concern for the ecological integrity of receiving surface water bodies or the potential for groundwater contamination. Several different models were suggested in the MED-Rice Guidance Document (MED-Rice, 2000). Recently, a new mechanistic model called SWAGW, last release 1.095, was developed and tested for estimation of PECs in surface water, sediment, groundwater and soil in the paddy and canal environment (Cervelli, et al., 2004, 2005; Redolfi et al., 2004; Karpouzas et al., 2006). SWAGW is currently used by the Consultant Commission for Pesticide Compounds of the Italian Ministry of Health for regulatory purposes. The major improvement in the assumptions of step 2 calculations for paddy and surface waters is a more realistic representation of the pesticide adsorption process. The amount of substance adsorbed on the solid phase (soil and sediment) is in constant equilibrium with the amount present in the water phase (paddy field and canal). For groundwater, the model assumes a miscible displacement behaviour of the pesticide, a constant soil moisture content corresponding to saturation, and a constant addition of the pesticide corresponding to its TWA in paddy water, both during the period of paddy closure and the flooding period. 7.1. Surface water, sediment and soil 7.1.1. Closed rice system

The chemical–physical model of the rice cropping system during the period of paddy closure is shown in Figure 4. The equation for the conservation of the mass of the substance in the paddy field is given by the following equation: (36) ATotal
ALL  AL  AS  BL

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

103

Before Opening Paddy

Environment

Canal

Paddy Water Paddy Soil AS

Canal

Water

K2 A*L

AL kinf

Environment

K1

Sediments A*S

k1W

k2W

B*L

BL

Fig. 4. A schematic representation of the chemical–physical model of the rice–paddy and canal environments during paddy closure. AL and AS are the amount of pesticide in the water and soil * compartment, respectively. BL is the residue amount in water. AL* , AS* and BL are the corresponding pesticide amounts in the water/sediment system of the canal. K1 and K2 are the adsorption ratios in canal and paddy respectively, kinf is a pseudo first order leaching constant, and k1w and k2w are the first order degradation rate constants in the canal and paddy, respectively.

where ATotal, ALL, AL, AS and BL (all expressed as
g) are the total amount of the substance added, the amount leached through soil, the amount present in water, the amount adsorbed on soil and the amount degraded, respectively. After differentiating equation (36) with respect to time and considering the following elementary equations: dALL  kinf AL dt

(37)

dBL  k2W AL dt

(38)

dAS dA  K2 L dt dt

(39)

k  kinf dAL AL 
2W dt 1 K 2

(40)

we have

where kinf is a pseudo first order degradation rate constant to take into account the leaching (day
1), k2w is the first order degradation rate constant in paddy water (day
1) and K2 is the adsorption ratio. We also have K 2  K 2d 2

L2S L2W

(41)

104

S. Cervelli and R. Jackson

where K2d is the soil adsorption constant (dm3 kg
1), 2 the soil bulk density (kg dm
3), L2S the depth of the soil (cm) and L2W (cm) the level of the water on top of the soil. After integration of equation (40) with the initial condition t  0 → AL  A0

(42)

where A0 (
g) is the initial amount of substance in the water phase following the adsorption equilibrium, the solution is AL  A0 e
(( k2W kinf ) /(1K2 )) t .

(43)

The amount of substance adsorbed onto soil, AS (
g), is AS  K 2 AL

(44)

The TWA, in the time interval 0  t  T1 , where T1 is the closing time period, is calculated according to

TWA 

t2 1 f (t )dt ∫ t2
t1 t1

(45)

and therefore it will be TWA AL 

A0 (1 K 2 ) (1
e
(( k2W kinf ) /(1K2 )) t ). t (k2W  kinf )

(46)

For the canal, following the scheme and symbols of Figure 4 (the superscript “*” indicates any compound present in the canal system), analogously to the paddy but without any infiltration we have A*L  A*0 e
( k1W /(1K1 )) t

(47)

where K1  K1d 1

L1S L1W

(48)

where K1 is the adsorption ratio in the canal, k1d the sediment adsorption constant (dm3 kg
1), 1 the sediment bulk density (kg dm
3), L1S the depth of the sediments (cm) and L1W (cm) the level of the water in the canal. The amount of substance adsorbed onto sediment, AS* (
g), is AS*  K1 AL*

(49)

In the time interval 0  t  T1 , where T1 is the closing time period, the corresponding TWA, according to equation (45), will be TWA A*  L

A*0 (1 K1 ) tk1W

(1
e
( k1W /(1K1 )) t ).

(50)

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

105

After Opening

Paddy

Environment

Canal Environment CanalWater

Paddy Water Paddy Soil

K2

AS

AL

koutflow

A*L

k2W kinf

k1W

K1

Sediments

A*S

B*L

BL

Fig. 5. A schematic representation of the chemical–physical model of the rice–paddy and canal environments after opening the paddy field. AL, AS, BL, AL* , AS* , BL* , K1, K2, kinf, k1w and k2w are defined as in Figure 4. kinf, and koutflow are pseudo first order rate constants associated with leaching and outflow.

7.1.2. Flooding period

The chemical–physical model after opening the rice field (i.e. after the 5-day closure period) and allowing water to flow in and out of the paddy is presented in Figure 5. Thus, the relationship for the conservation of the mass in the paddy environment is ATotal
ALL
AL  AL  AS  BL

(51)

where ALL (
g) is the amount of AL lost by leaching through soil and AL (
g) the amount of AL flowing out of the paddy to the canal. After differentiating equation (51) with respect to time and considering the elementary equations dALL  kinf AL dt

(52)

d AL  kout AL dt

(53)

dAS dA  K2 L dt dt

(54)

dBL  k2W AL dt

(55)

k  kinf  kout dAL AL 
2W dt 1 K 2

(56)

we have

106

S. Cervelli and R. Jackson

where kout (day
1), that takes into account the water outflow, is equal to koutflow/Vp, where koutflow is the volume of water leaving the paddy environment during the time (L day
1), and Vp the volume of the paddy water (L). The ratio K2 has been previously defined in equation (41). The solution of equation (56), with the initial condition t  T1 → AL  A0º

(57)


where A0 (
g) is the amount of substance at the end of the closing time T1 (day), is


AL  A e º 0

k2W kinf kout ( t
T1 ) 1K 2

(58)

.

The amount of pesticide AS (
g) which is adsorbed onto soil is given by equation (44). The TWA in the time interval T1  t  T2 , where T1 (day) is the end of the closing time and T2 (day) the permanent drain time, according to equation (45) will be k  kinf  kout  ( t
T1 ) 
2W A0
(1 + K 2 ) 1  K2 TWA AL   1
e  (t
T1 )(k2W  kinf  kout )  

(59)

Following the opening of the paddy field, water and substance flow to the canal and, following the scheme of Figure 5, the relationship for the mass conservation in this environment will be * ATotal  AL  AL*  AS*  BL*

(60)

* where ATotal (
g) and AL (
g) are the amount of substance in the canal due to drift and outflow, respectively. During the flow, a DF of 10 is considered. After differentiating with respect to time and taking into account the equations

dBL*  k1W AL* dt

(61)

dAS* dA*  K1 L dt dt

(62)

d AL kout AL  dt DF

(63)

we have
k A* kout dAL* A0
e 
1W L  dt 1 K1 (1 K1 )DF

k2W kinf kout ( t
T1 ) 1K 2

.

(64)

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

107

The constant K1 has been previously defined in equation (48). The solution of equation (64) with the initial conditions t  T1 → AL  A0


(65)

t  T1 → A*L  A0
*

(66)

is


AL* 

e

k1 w k k k (t
T1 )
2 w 1infK out (tT1 ) 1  K1 ( 2)

DF 
kinf (1 K1 )  k1w (1 K 2 )
(1 K1 ) ( k2 w  kinf  kout ) 2 ( k  k  kout ) k1 w   k2 w 1 kinfK  kout (t  T1 ) T1 (t
T1 )  2 w 1  inf K2 1  K1  2  
A0  e
e   

 (1 K 2 ) kout
A0* DF e

  

(67)

k2 w  kinf  kout (t  T1 ) 1  K2

   kinf (1 K1 )
k1w (1 K 2 )  (1 K1 ) ( k2 w  kinf  kout )  

and the TWA, according to equation (45), for the time interval T1  t  T2 , where T1 (day) is the end of the closing time and T2 (day) the permanent drain time, is TWAA*  L

 k (k  kinf  kout )  t 2 T
 1w  2 w 1) ( 1 K 2  1 K1 

(t
T1 ) (k2 w  kinf  kout )

e


kinf (1 K1 )  k1w (1 K 2 )
(1 K1 ) ( k2 w  kout )

  (k2 w  kinf  kout ) (t  T1 )  k1w (1K2 )(1K1 )(k2 w  kinf  kout ) T1 1K1    1  o* 1K 2  A0 DF  e DF k1w     k1 w k2 w  kinf  kout t  (t  2 T1 )  1K 2
e 1K1  kinf (1 K1 )
k1w (1 K 2 )  (1 K1 ) ( k2 w  kout ) 

(

) (68)

  k1w t  3 (k2 w  kinf  kout ) T1 1K 2 k1w (1 K 2 )  (1 K1 ) ( k2 w  kinf  kout )  A0 kout (1 K 2 ) 
e 1K1  

e



e

(k2 w  kinf  kout ) (tT1 )

k1 w (1K 2 )(1K1 )(k2 w  kinf  kout ) 1K1

T1

(1 K1 ) (k2 w  kinf  kout )

1K 2

k1 w k  kinf  kout t  2w (t  2 T1 ) 1K1 1K 2

(
k (1 K ) k (1 K )
(1 K ) (k inf

1

1w

2

1

2w

     

)

 kout )

108

S. Cervelli and R. Jackson

The amount As* (
g) adsorbed on sediment is given by the equation (49). The concentrations in paddy and canal water are calculated by dividing the amounts AL (
g) and AL* (
g) by the volume of paddy and canal water (L), respectively. The amounts of substance adsorbed per weight units of soil and sediments are calculated by dividing As (
g) and As* (
g) by the weight of the soil and sediment (kg), respectively. Recently (Karpouzas et al., 2006), an improvement has been implemented into SWAGW to take into account the time dependent partitioning of pesticides between paddy water and soil. In this case equation (44) will be changed to AS  K 2 (1
e
t ) AL

(69)

where  is a constant (day
1). When t → equation (69) is equal to equation (44). During paddy closure, using the equation (69) the differential equation (40) becomes k  kinf  K 2 e
t dAL AL 
2W dt 1 K 2 (1
e
t )

(70)

the solution of which is t

 k k 
 1 2W inf    (1K 2 ) 

t

AL  A0 e [
K 2  e (1 K 2 )]

(71)

and, according to equation (45), TWA is 

k2W kinf 


 1 A0  (1K 2 )  [
K 2  et (1 K 2 )]  TWA AL  t (k2W  kinf )  k k    1 2W inf    (1K 2) 
t t    K 2
e (1 K 2 )
( K 2
e (1 K 2 ))     

(72)

After opening of the paddy, the differential equation (56) becomes (k  kinf  kout  K 2 e
t ) dAL AL 
2W dt 1 K 2 (1
e
t )

(73)

the solution of which is
t 0

t

 k k k 
 1  2W inf out   (1K )  

AL  A e [
K 2  e (1 K 2 )]

2

(74)

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

109

and, according to equation (45), TWA is k k k 
2W inf out A0
T1  (1K 2 ) TWA AL  (
K 2  e (1 K 2 )) (t
T1 )(k2W  kinf  kout ) 



(
K 2  et (1 K 2 ))

k2W kinf kout  (1K 2 )

 . 

(75)

7.2. Groundwater The complete chemical–physical model assumed for the development of the step 2 approach for calculating pesticide leaching is illustrated in Figure 6. The equations used to calculate PECs at different times and depths are derived from the general equation for the conservation of mass (Jury et al., 1991) ∂J ∂C T 
S  ∂t ∂x

(76)

PADDY SOIL COLUMN k2w 10 cm

AL

Kads 0-30 cm

k AL

AS

K'ads 30-60 cm

k'

K''ads AS

BL

AL

AS

60-100 cm

BL

BL

k'' AL

BL

Fig. 6. A schematic representation of the chemical–physical model of the rice–paddy leaching system. AL, AS and BL are defined as in Figure 2. Kads, K ads ′ and K ads ′′ are the soil adsorption constants, and k, k and k ′′ are the first order degradation rate constants for the different soil layers, taking into account the respective biofactors.

110

S. Cervelli and R. Jackson

where CT (
g L
1) is the total concentration, JS (
g cm
2 day
1) the flow of the substance, t the time (day), x the depth (cm) and  a source–sink term (
g L
1 day
1). We have also (Hutson and Wagenet, 1992) CT  K ads c 
c J S 

Di (
, q )

(77)

∂c  qc ∂x

(78)

q


(79)

Di (
, q )  DW aeb
+   

kreact c

(80)

where  (kg dm
3) is the soil bulk density, Kads (dm3 kg
1) the substance adsorption constant, c (
g L
1) the substance concentration,
(cm3 cm
3) the volumetric water content, q (cm day
1) the water flux, Di (cm2 day
1) the dispersion coefficient, DW (cm2 day
1) the diffusion coefficient in pure water, a and b are two empirical constants reported to be approximately b  10 and 0.005  a  0.01 (Hutson and Wagenet, 1992),  (cm) the dispersivity factor, having the value: 0.2    8 (Hutson and Wagenet, 1992) and kreact (day
1) the first order degradation rate constant. After substitution of equations (77)–(80) in equation (76), we obtain the following equation: ∂c ∂2 c ∂c  Disp 2
v
kc ∂t ∂x ∂x

(81)

q
 DW aeb
    R


(82)

where Disp 

where DW aeb
is the effective diffusion coefficient (Kemper and Van Schaik, 1966), and q R

(83)


kreact R

(84)

v k

R 
 K ads

(85)

where v (cm day
1) is the pore water velocity and R (dimensionless) the retardation coefficient. If Disp (cm2 day
1) and v are constant and the substance is

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

111

added at a constant concentration for a time T, according to the following initial and boundary conditions c  C0

x0 0t T

(86)

c0

x0 t T

(87)

c0

x 0 t0

(88)

c0

x → t 0

(89)

equation (81) can be solved analytically (Misra et al., 1974), and its solution is c( x, t )  P ( x, t ) → 0  t  T

(90)

c( x, t )  P ( x, t )
P[ x,(t
T )] → t  T

(91)

where T is the time of the substance addition, with  x
t v 2  4 Disp k   x  1   P ( x, t )  C0  exp  (v
v 2  4 Disp k )  erfc  2  2 Disp 4 Disp t       (92)  x  t v 2  4 Disp k    x  2   exp  (v  v  4 Disp k )  erfc  4 Disp t     2 Disp  where exp is the natural exponential and erfc is the error function complementary. During the period of paddy closure, the constant concentration of the added substance is assumed to be the TWA AL given by equation (46) in paddy water for 0  t  T1, while during the period of paddy opening, the same concentration is assumed to be the TWA AL given by equation (59) for T1  t  T2. As in step 1, according to the assumptions introduced by the FOCUS Ground Water Scenario Workgroup (FOCUS, 2000), some biofactors can be taken into account. Since it is not possible to use equations (90) and (91) while k and Kads are changing along the soil profile, it is assumed that the degradation and adsorption constants to be used in groundwater calculations can be expressed as weighted averages along the soil depth according to the following equations

kreact  K ads 

1 ∑ i xi

∑k x

i i

(93)

1 ∑ i xi

∑K x

(94)

i

i i

i

where ki  bioi k2

(95)

K i  bioi K 2d

(96)

112

S. Cervelli and R. Jackson

and bioi and xi (mm) are the biofactors and the horizon depths, respectively. Finally, the TWAs are computed according to the following equation: TWA 

1 ∑ t i

∑ f ( t ) t i

i

i

(97)

i

where ti is the elementary time interval (day) and fi(t) are the functions (90) or (91). 7.3. Sensitivity analysis While the degradation rate constants and adsorption coefficients, used in the paddy and surface water, the sediment and the soil subroutines of SWAGW are experimentally determined as part of the regulatory requirements for inclusion of a pesticide in Annex I of the Council Directive 91/414/EEC, the constant DW necessary in the groundwater subroutine is not provided. Furthermore, the values of the constants  and , are not accurately known, even though their role in the pesticide percolation prediction is only minor. Therefore, the sensitivity analysis of SWAGW was limited to the groundwater subroutine. The sensitivity analysis was carried out for the most vulnerable rice scenario 2 (sandy soil) for leaching using a dummy pesticide, the physical–chemical properties of which were preliminarily found to be the most critical for inclusion in Annex I (DT50 in paddy and canal waters is 30 days and KOC for soil and sediments is 100 dm3 kg
1). The effect of the variation of the input parameters q (water flow rate), Diff (effective diffusion coefficient),  (dispersivity factor) and the parameter a of the equation (79), on TWAs of groundwater was explored running a number of simulations for 120 days. The predicted TWAs were found to be significantly affected by the dispersivity factor (Figure 7) and the water flow rate (Figure 8). In contrast, variations of the values of the effective diffusion coefficient (Figure 9) and the constant a (Figures 10 and 11) did not significantly affect the TWA predictions of the model. Therefore, the dispersivity factor  and the water flow q, which both vary in different soils, and their values are not well known in Mediterranean rice growing soils, have the largest influence on the uncertainty of the groundwater predictions of the SWAGW model. However, while according to the works of Greppi et al. (1998) and Greppi (1999) on Italian paddy soils an approximately constant water flow of 1 cm day
1 can be used as realistic worst case, the lack of reliable dispersivity factor data may severely affect TWAs prediction. The results of some simulations carried out using different dispersivity factors are reported in Table 2. Reduction of the dispersivity factor from 1 cm to 0.2 cm, after 120 days of simulation, resulted in a TWA decrease of about 20%, while increasing the dispersivity factor from 1 cm to 8 cm resulted in a TWA increase of more than 100%. The FOCUS Workgroup for Groundwater (FOCUS, 2000) parameterised the chromatographic models PRZM, PELMO and PEARL according to nine

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

113

Fig. 7. The effect of the dispersivity factor parameter,  (cm) on the prediction of groundwater TWAs (
g L
1), after a simulation of 120 days.

Fig. 8. The effect of the water flow rate parameter, q (cm day
1) on the prediction of groundwater TWAs (
g L
1), after a simulation of 120 days.

representative European scenarios. The results of the different simulations revealed that the groundwater PECs delivered by the different models for the same scenario significantly differ in several cases. These differences were attributed to the different default dispersivity factors used by the three different models. Recently, Boesten (2004) examined the influence of the dispersion length parameter on the leaching predictions of the FOCUS groundwater models, and found that the dispersivity factor significantly influence the groundwater PECs of the models. The opinion of the Scientific Panel on Plant Health, PPP and their Residues (EFSA, 2004) on the comparability of the three groundwater

114

S. Cervelli and R. Jackson

Fig. 9. The effect of the effective diffusion coefficient parameter, Diff (cm2 day
1) on the prediction of groundwater TWAs (
g L
1), after a simulation of 120 days.

Fig. 10. The effect of the variation of the parameter a of the Kemper and Van Schaik equation (Kemper and Van Schaik, 1966) on the prediction of groundwater TWAs (
g L
1), after a simulation of 120 days.

models, equally stressed the same point. Therefore, the data reported in Table 2 are in agreement with these findings (FOCUS, 2000; Boesten, 2004; EFSA, 2004) and show that the correct parameterisation of the dispersivity factor parameter plays a major role in obtaining realistic leaching predictions. For the model SWAGW, according to the work of Perfect et al. (2002), the values of 0.5 and 5 cm were chosen as defaults for the sandy and the clayey scenarios, respectively.

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

115

a & lamda 2.4 2.2

λ = 8.0 cm

TWA (µg/ L)

2

λ = 4.1 cm

1.8 1.6 1.4 1.2 1 0.004

λ = 0.2 cm 0.005

0.006

0.007

0.008

0.009

0.01

0.011

a

Fig. 11. The effect of the variation of the parameter a of the Kemper and Van Schaik equation (Kemper and Van Schaik, 1966), for different values of the dispersivity factor  (cm), on the prediction of groundwater TWAs (
g L
1), after a simulation of 120 days.

Table 2. The uncertainty of groundwater TWAs computations due to the dispersivity factor , accounted as percent variation from a reference value (  1 cm) Dispersivity factor () (cm) 0.2 0.5 1.0 2.0 4.1 6.0 8.0

 TWA (%)
21.6
12.0 0.0 20.7 52.0 74.7 108.4

7.4. Example calculations In this section, some example calculations using theoretical (dummy) pesticides with a range of different properties are given using the Med-Rice calculator and the SWAGW model. The main advantage of SWAGW is both its mechanistic approach and the low number of input parameters required for its parameterisation. In fact, only the agronomic scenario, the pesticide application rate, the DT50 and the KOC values are required for its parameterisation. Furthermore, the latest version of the SWAGW model allows the user to operate outside the two fixed Med-Rice scenarios. Therefore, the user is allowed to select values other than the default ones for certain parameters including the period of paddy closure and simulation, by

116

S. Cervelli and R. Jackson

Fig. 12. An example of the input choices available in the SWAGW parameterisation.

simply choosing no-default running. An example of all possible input choices is presented in Figure 12. A total of six simulations were performed with both the Med-Rice calculator and the SWAGW model with three dummy compounds which had variable DT50 and KOC values. These values were selected to cover a large number of pesticide characteristics. The TWAs in paddy water, surface water and groundwater produced using the Med-Rice calculator and SWAGW model are reported in Tables 3 and 4 for the clayey and the sandy scenarios, respectively. All dummy compounds were added at the rate of 100 g ha
1. Both clayey and sandy scenarios were used for model simulations, and other default parameters for computation are reported in Table 5. It should be noted that for the Med-Rice PECgw calculation for the dummy compound 2 in the sandy soil scenario, the mass of pesticide leaching below 1 m (Mleak) exceeded the amount of pesticide applied (Table 4). The reason for this discrepancy is the default high infiltration rate of 10 mm day
1 used for the sandy soil scenario coupled with the assumption that the level of water in the paddy remains constant without any dilution taking place (due to inflow of fresh water to replace the infiltrated water). This leads to an overestimation of the mass of pesticide available for leaching. The graphical representation of the dissipation of the concentration of the Dummy 2 compound in paddy water and surface water are reported in Figures 13 and 14. The graphical representation of leaching of the compound Dummy 2 (sandy soil scenario) under the same conditions as shown in Figures 13 and 14, is shown in Figure 15. The DT50 of 30 days for the Dummy compound 2, in both paddy and canal water, and the KOC of 100 mL g
1 for both soil and sediment have been chosen to magnify the influence of soil texture on the predicted TWA at 1 m depth. In this example, after about 100 days the TWAgw exceeded the trigger value of 0.1
g L
1.

Compound

DT50

KOC

TWApw 5 days

Dummy 1 Dummy 2 Dummy 3

3 30 100

10 100 1000

TWAsw 120 days

5 days

SWAGW

Med-Rice

SWAGW

Med-Rice

54.1 40.1 6.6

52.2 40.2 6.8

1.1 10.4 5.3

0.8 4.8 1.1

120 days

SWAGW 1.6  10
1 2.4  10
1 1.2  10
1

5.3  10
3 2.4  10
1 1.6  10
1

TWAgw

Step 1c max

Max

120 days

Med-Rice

SWAGW

SWAGW

Med-Rice

2.6 3.6 0.72

2.4  10
1 2.4  10
1 1.3  10
1

1.9  10
3 0.01 0.01

0.01 0.01 0.01

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

Table 3. TWA concentrations (
g L
1) of three dummy compounds for the clayey scenario computed for the period of paddy closure (5 days) and the whole simulation period (120 days). TWAs in paddy water (TWAspw), surface water (TWAssw) and groundwater (TWAsgw). TWAsgw have been computed at the depth of 100 cm according to FOCUS (2000)

117

118

Table 4. TWA concentrations (
g L
1) of three dummy compounds for the sandy scenario computed for the period of paddy closure (5 days) and the whole simulation period (120 days). TWAs in paddy water (TWAspw), surface water (TWAssw) and groundwater (TWAsgw). TWAsgw have been computed at the depth of 100 cm according to FOCUS ( 2000) Compound

DT50

KOC

TWApw 5 days

Dummy 1 Dummy 2 Dummy 3 aNote

3 30 100

10 100 1000

TWAsw 120 days

5 days

SWAGW

Med-Rice

SWAGW

Med-Rice

47.6 49.9 12.5

55.5 56.4 12.7

4.9  10
3 3.6 4.8

0.9 6.7 2.1

120 days SWAGW

1.6  10
1 2.3  10
1 1.2  10
1

4.2  10
3 1.4  10
1 1.7  10
1

TWAgw

Step 1c max

Max

120 days

Med-Rice

SWAGW

SWAGW

2.8 5.0 1.2

2.2  10
1 2.4  10
1 1.3  10
1

1.7  10
2 1.7  10
3 8.6  10
1 0.493a 0.01 0.0156

Med-Rice

that total mass leached exceeds the total amount applied.

S. Cervelli and R. Jackson

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

119

Table 5. The values of the input variables used, in addition to the ones previously reported in Table 1, for the parameterisation of the SWAGW model for the two scenarios, clayey and sandy soil Parameter Rate of application (g ha
1) Time of water flooding after paddy opening (day) Amount of Dummy reaching the canal by drift (%) Groundwater depth (cm) Bulk density of the paddy soil (kg L
1) Bulk density of the canal sediment (kg L
1) Dilution factor Volumetric soil water content at saturation (m3 m
3) Organic carbon in sediment (%) Constant a Constant b Dispersivity factor (cm) Flow velocity of water (cm day
1) Diffusion coefficient in water (cm2 day
1) DT50 (day) KOC (mL g
1) Rate constant of outflow (day
1)

Clayey soil 100 115 2.8 100 1.5 1.5 10 0.44 1.6 0.008 10.0 5 0.1 0.32 3, 30, 100 10, 100, 1000 0.043

Sandy soil 100 115 2.8 100 1.5 1.5 10 0.39 1.6 0.008 10.0 0.5 1.0 0.32 3, 30, 100 10, 100, 1000 0.043

Fig. 13. PECs and TWAs (
g L
1) of Dummy compound 2 in the paddy water (sandy soil). At day 5 the paddy was opened and the water flowed to the canal.

120

S. Cervelli and R. Jackson

Fig. 14. PECs and TWAs (
g L
1) of Dummy compound 2 in the surface water of canal (sandy soil). At day 5 the paddy was opened and the water flowed to the canal.

Fig. 15. PECs and TWAs (
g L
1) for the Dummy compound 2 in the groundwater at the depth of 100 cm.

7.5. Validation A preliminary validation of SWAGW was performed with selected field datasets (Redolfi et al., 2004; Karpouzas et al., 2006). A first set of data (Redolfi et al., 2004) was provided by different agrochemical companies (DuPont, Bayer and Dow AgroSciences), and among the datasets provided one compound, here

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

121

named XA, was selected for validation of the paddy water subroutine. The rate of application of the compound was 25.8 g ha
1, DT50 in paddy water was 12.2 days and KOC was 70 mL g
1. The experimental data were obtained in a clayey and a sandy soil. 15 days after pesticide application no pesticide residues were detected in the paddy water. As shown in Table 6, the SWAGW model fitted well the measured concentrations of the compound, and the difference between observed and predicted values were not significant at the 0.05 level of probability. That was true for both the sandy soil and the clayey soil (Snedecor and Cochran, 1967). A second set of data was obtained from field dissipation studies employed in northern Italy with the herbicides cinosulfuron for the period 1997–1998 and pretilachlor for the period 2001–2002. These datasets were previously used for validating the ability of the SWAGW model to predict the dissipation of these two herbicides when parameterised outside the framework of the two standard Med-Rice scenarios (Karpouzas et al., 2006). Due to the time dependence of the adsorption of the two pesticides on soil, a new subroutine using the equations (69), (71) and (74) was developed. The values used for the parameterisation of the SWAGW model are listed in Table 7 and a summary of the statistical analysis of the results is reported in Table 8. As shown in Table 8, the predicted concentrations of cinosulfuron for the year 1997 fitted well the measured concentrations of the herbicide and no significant difference between measured and predicted concentrations was observed at the 0.05 level of probability. For pretilachlor, a good correspondence between measured and predicted concentrations in the paddy water in both 2001 and 2002 was Table 6. Correlations between the measured concentrations and the PECs computed by SWAGW for the compound XA in the paddy water of two different soils. The levels of probability have been determined for degrees of freedom corresponding to the number of couples of data minus two Soil texture Clayey Sandy

Correlation coefficient

Level of probability

0.7556 0.7801

0.05 0.05

Table 7. Values used for parameterisation of the SWAGW model Parameter Rate of application (g ha
1) Degradation rate in paddy water (day
1) Degradation rate in paddy soil (day
1) Adsorption coefficient KOC (mL g
1) Constant (day
1) Percolation rate (cm day
1) Time of paddy closure after pesticide application (day) Organic carbon content (%)

Cinosulfuron 70 0.0355 0.0346 115 0.085 0.23 14/22 1.3

Pretilachlor 1125 0.1023/0.1473 0.06923 542.3 0.013 0.07/0.27 23/18 1.3

122

S. Cervelli and R. Jackson

Table 8. Correlations between pesticide observed and predicted concentrations by SWAGW for cinosulfuron and pretilachlor in the paddy water and paddy soil. The levels of probability have been determined for degrees of freedom corresponding to the number of couples of data minus two Pesticide

Cinosulfuron Pretilachlor

Year

1997 1998 2001 2002

Correlation coefficient

Level of probability

Water

Soil

Water

0.5622 0.2291 0.9608 0.9772

– – 0.3421 0.7147

0.05 0.5 0.05 0.05

Soil – – 0.1 0.05

evident since there was no significant difference between measured and predicted concentrations at the 0.05 level of probability. Similarly, no significant differences between measured and predicted concentrations of pretilachlor in paddy soil in the year 2002 were evident at the level of 0.05 level of probability. However, a significant difference at the 0.1 level of probability was evident between predicted and measured concentrations of pretilachlor in paddy soil for the year 2001. No data of groundwater concentrations were available in the two datasets, and therefore no validation of the subroutine has been possible. 8. CONCLUSIONS A series of calculation methods have been presented to generate exposure estimates in aquatic and soil environments for pesticides that are used in rice cultivation. In general, the current fate and behaviour studies that are conducted for registration purposes can provide the input data necessary for such calculations. In special cases, further studies may be necessary to evaluate the fate under more realistic outdoor conditions, particularly where there are competing routes of degradation such as microbial degradation and photolysis. Also, the particular use pattern of the pesticide needs to be considered. For example, the exposure potential for a herbicide applied early in the season may be considerably different to that of a fungicide applied later in the season to a maturing crop. For the first tier Med-Rice calculator, it should be noted that some anomalous results can be obtained particularly for PECgw values in sandy soil for certain combinations of degradation and sorption values. However, in general the results can be useful to give a conservative estimate of exposure for different pesticides. For the model SWAGW, a preliminary validation was attempted with the data available for three pesticides, XA, cinosulfuron and pretilachlor. However, no groundwater data were available. According to the model characteristics, in the groundwater environment the computation of PECs are highly affected by the dispersivity factor , which plays the most important role in determining the pesticide dispersion during leaching. The very few data available for paddy water and soil, and the lack of data for groundwater imply the necessity to obtain enough experimental data to validate the model for reliable predictions of PECs and TWAs.

Pesticide Exposure Assessment in Rice Paddies: The Lower-Tier Analysis

123

REFERENCES Boesten, J. T. I. (2004). Influence of dispersion length on leaching calculated with PEARL, PELMO and PRZM for FOCUS groundwater scenarios. Pest Manag. Sci. 60, 971. Cervelli, S., Cervelli, F. and Cervelli, L. (2004). A Step 2 approach to compute water, soil and sediment PECs and TWAs for the inclusion of rice pesticides in Annex I of the Council Directive 914/414/EEC. In: A. Ferrario and F. Vidotto (Eds), Proc. of the Conference Challenges and opportunities for sustainable rice-based production systems (p. 431). Edizioni Mercurio, Turin, Italy. Cervelli, S., Cervelli, F. and Cervelli, L. (2005). La Bonifica dei Siti Contaminati, Gruppo Scientifico Italiano Studi e Ricerche, Milano, 41. EFSA (2004). Opinion of the Scientific Panel on Plant Health, Plant Protection Products and their Residues on a Request of EFSA related to FOCUS groundwater models. The EFSA Journal 93, 1. FOCUS (2000). FOCUS groundwater scenarios in the EU plant protection product review process. Report of the FOCUS Groundwater Scenarios Workgroup, EC Document Reference Sanco/321/2000. FOCUS (2001). FOCUS surface water scenarios in the EU evaluation process under 91/414/EEC. Report of the FOCUS Working Group on surface water scenarios, EC Document Reference SANCO/4802/2001-rev.1. Greppi, M. (1999). In Environmental Risk Parameters for Use of Plant Protection Products in Rice, Workshop, Cremona, 10. Greppi, M., Silvestri, S., Anselmetti, M. P., Vietti, L. and Raviolo, F. (1998). Giornata mondiale dell’acqua, Roma. Hutson, J. L. and Wagenet, R. J. (1992). LEACHM (Leaching Estimation and Chemistry Model). A process-based model of water and solute movement, transformations, plant uptake and chemical reactions in the unsatured zone. NY State College of Agriculture and Life Sciences, Cornell University, Ithaca, New York, 14853. Jury, W. A., Gardner, W. R. and Gardner, W. H. (1991). Soil Physics. John Wiley & Sons, New York. Karpouzas, D. G., Cervelli, S., Watanabe, H. and Capri, E. (2006). Pesticide exposure assessment in rice paddies: A comparative study of existing mathematical models. In publication on Pest Manage. Sci. (in press). Kemper, W. D. and Van Schaik, J. C. (1966). Diffusion of salts in clay-water systems. Soil Sci. Soc. Am. Proc. 30, 534. MED-Rice (2000). Guidance Document for environmental risk assessments of active substances used on rice in the EU for Annex I inclusion. Document prepared by Working Group on MEDRice, EU Document Reference Sanco/1090/2000 – rev. 1, Brussels, June 2003, (http:// www. ime.fraunhofer.de/download/FOCUS/rice_report/). Misra, C., Nielsen, D. R. and Bigger, J. W. (1974). Soil Sci. Soc. Am. Proc. 38, 300. OECD (2002). OECD Guideline for the Testing of Chemicals, Aerobic and Anaerobic Transformation in Soil. OECD Test Guideline 307, OECD. Perfect, E., Sukop, M. C. and Haszler, G. R. (2002). Soil. Sci. Soc. Am. J. 66, 696. Redolfi, E., Azimonti, G., Auteri, D. and Cervelli, S. (2004). Validation of a new model (SWAGW) to estimate PECs and in paddy fields. In: A. Ferrario and F. Vidotto (Eds), Proc. of the Conference Challenges and opportunities for sustainable rice-based production systems (p. 485). Edizioni Mercurio, Turin, Italy. Schnell, S., Liesack, W. and Revsbech, N. P. (2000). FEMS Microb. Reviews 24, 625. SETAC (1995). Procedures for assessing the environmental fate and ecotoxicity of pesticides. In: M. R. Lynch (Ed). SETAC-Europe, ISBN 90-5607-002-9. Snedecor, G. W. and Cochran, W. G. (1967). Statistical Methods. The Iowa State University Press, Ames, Iowa.

Pesticide Risk Assessment in Rice Paddies: Theory and Practice E. Capri and D.G. Karpouzas (editors) © 2008 Elsevier B.V. All rights reserved

Chapter 7

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective D. G. Karpouzas1 and Z. Miao2 1Department

of Biochemistry-Biotechnology, Ploutonos 26 & Aiolou, 41221 Larisa, University of Thessaly, Greece 2McGill University, Biology Department, 1205 Docteur Penfield, Montreal, Quebec, Canada H3A 1B1

Contents 1. Introduction 2. Exposure assessment at tier 3: paddy field scale level 2.1. Description of the RICEWQ model 2.2. Parameterization, uncertainty and sensitivity analysis of RICEWQ model 2.2.1. Meteorological and agronomic parameters 2.2.2. Water management parameters 2.2.3. Paddy soil/sediment parameters 2.2.4. Physicochemical pesticide parameters 2.2.5. Parameters controlling pesticide dissipation 2.2.6. Simulation of pesticide metabolites with the RICEWQ model 2.2.7. Parameterization of the VADOFT sub-model 2.3. Tier-3 scenarios at member-state level 2.3.1. Scenarios development for Greece 2.3.2. Scenarios development for Italy 2.4. Other available models 3. Exposure assessment at tier 4: basin-scale level 3.1. Description of the RIVWQ model 3.2. Description of the combined use of RICEWQ – RIVWQ models 3.3. Parameterization of RICEWQ, RIVWQ models using basin-scale scenarios 3.3.1. Basin-scale scenario for Greece 3.3.2. Basin-scale scenario for Italy 3.4. Landscape risk assessment in rice paddy areas based on geographical information systems (GIS) 3.4.1. Landscape exposure assessment in a rice cultivated watershed in Lombardy 4. Concluding remarks and the way forward Acknowledgments References

125 128 128 129 129 131 131 133 134 137 138 139 139 140 142 145 145 148 149 150 155 156 157 160 162 162

1. INTRODUCTION Rice cultivation in Europe is limited to the southern European countries with Italy, Spain, Greece, France and Portugal being the main rice producers. From an environmental point of view, the requirement for large amounts of irrigation water used for rice cultivation increases the likelihood for contamination of receiving

126

D. G. Karpouzas and Z. Miao

surface water (SW) bodies with pesticides via overflow or controlled drainage and of groundwater (GW) bodies via leaching if rice paddies are not confined by impermeable layers. Numerous monitoring studies in Europe have provided evidence that rice cultivation may be responsible for SW contamination with pesticide concentrations exceeding 0.1
g/L (Readman et al., 1993; Batista et al., 2002; Cerejeira et al., 2003; Tarazona et al., 2003; Papadopoulou-Mourkidou et al., 2004). Among the pesticides detected, herbicides constitute the major group of pesticides detected in SWs adjacent to rice cultivated land, with molinate and propanil being the most frequently detected compounds (Batista et al., 2002; Tarazona et al., 2003). As monitoring programs and field studies are money and time-consuming, validated mathematical models in Europe have been integrated into the regulatory process for estimating the predicted environmental concentrations (PECs) in SW and GW bodies. PECs are then used as a tool for assessing the risk for potential environmental and human exposure. Although, detailed guidelines for the proper use of mathematical models are now available in Europe, these are not applicable to rice cultivation due to the constant flooding conditions used in most of the European countries. Currently, PECs of pesticides in rice paddies at tier 1 are calculated using a rather simple spreadsheet which is parameterized according to two conservative pan-European scenarios (Med-Rice, 2003). Recently, an improved mechanistic model based on the tier 1 Med-Rice spreadsheet called SW and GW model (SWAGW) was developed by Cervelli et al. (2004), and was proposed to be used as a tier-2 tool for calculating PECs of pesticides in rice paddies. Although, a more realistic representation of the processes involved in pesticide dissipation in rice paddies is provided by the SWAGW, this model is still considered an improved screening lower tier tool which, for regulatory purposes, should operate within the Med-Rice tier-1 scenarios (Redolfi et al., 2004). According to the tiered approach used for risk assessment in rice paddies (Figure 1), when a potential risk is identified at the lower tiers (tiers 1 and 2), a more sophisticated mathematical model should be used to more realistically assess the potential exposure at paddy field scale (tier 3) (Karpouzas and Capri, 2004). However, only few models are currently available for this purpose. The pesticide paddy field model (PADDY) and the Predicting Concentrations in Paddy Field (PCPF-1) model were both developed in Japan for simulating the environmental fate of pesticides in rice paddy systems (Inao and Kitamura, 1999; Watanabe and Takagi, 2000a). Recently, a modeling exercise with all the available higher tier models specific to rice cultivation, showed that PCPF-1 needs some adjustments and calibration before being used as a higher tier model in rice paddies under European conditions (Karpouzas et al., 2006a). Another mathematical model which could be used as a higher tier model for simulating pesticide fate in rice paddies is the Rice Water Quality (RICEWQ) model. RICEWQ was initially developed in USA for providing pesticide exposure assessment in rice paddies (Williams et al., 1999). Validation studies with European datasets revealed that RICEWQ could be an effective higher tiermodeling tool for calculating PECs in rice fields in Europe (Miao et al., 2003a; Karpouzas et al., 2005a,b). Further improvements in the water management

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

127

Tier 1 PECs calculated based on Med-Rice recommendation

Use Safe

Yes

No Tier 2 PECs calculated using the SWAGW model S

No further assessment Yes

Yes Use Safe No Tier 3 PECs calculated using sophisticated modeling at paddy field scale including realistic site specific consideration

Use Safe

No

Tier 4 PECs calculated at basin scale

Fig. 1. The standard tiered exposure assessment scheme applied for pesticides used in rice paddies.

routines of RICEWQ facilitated its parameterization for better describing the different water management practices applied in rice cultivated areas in Europe (Karpouzas et al., 2006a). Higher tier exposure assessment in rice paddies includes, tier 3 based on representative scenarios developed at member-state level for estimating PECs at paddy field scale, and tier 4 including landscape risk assessment using welldefined basin-scale scenarios, where mitigation processes could also be considered (Karpouzas and Capri, 2004). Rice is commonly cultivated in Europe in large river basins where artificial (e.g. drainage canals) and natural SW bodies (e.g. streams, rivers) constitute a unique ecosystem hosting birds, arthropods and other living organisms. Such basins could be located close to urban areas such as the Axios river basin in northern Greece, or within zones of high ecological value such as the National Park of Donana and the Lagoon of Valencia in Spain (Ramos et al., 2000). Therefore, it will be more relevant to consider pesticide risk assessment for rice crop at watershed scale. Recent studies by Miao et al. (2003b), revealed that the combination of RICEWQ with the River Water Quality (RIVWQ) 2.02v model provides a realistic estimation of PECs in SW bodies associated with treated rice paddies. This chapter aims to discuss all the recent developments in higher tier exposure assessment in rice paddies in Europe and identify key issues regarding modeling paddy systems at field scale and basin scale.

128

D. G. Karpouzas and Z. Miao

2. EXPOSURE ASSESSMENT AT TIER 3: PADDY FIELD SCALE LEVEL 2.1. Description of the RICEWQ model RICEWQ 1.6.1v was the initial version of the model which could calculate chemical dissipation within the paddy and SW releases but could not simulate pesticide leaching beneath the top 5 cm of paddy soil. To address this problem, a new RICEWQ 1.6.2v was developed by coupling the standard RICEWQ 1.6.1v with the Vadose zone Flow and Transport model (VADOFT). The latter is a vadose zone transport sub-model contained within the Pesticide Root Zone Model (PRZM) and it is used to simulate the fate of pesticides in the soil layers beneath the root zone (Carsel et al., 1998). Validation of the RICEWQ 1.6.2v model under European conditions showed that it could be an effective tool for higher tier exposure assessments in rice fields (Miao et al., 2003a; Karpouzas et al., 2005a,b). However, further improvements to its water management routines were required since it did not allow irrigation and drainage to occur concurrently. Therefore, the RICEWQ 1.6.2v was using a parameterization compromise to represent the continuous flow-through systems of irrigation /drainage which are common in rice-cultivating areas in Europe (Capri and Miao, 2002; Karpouzas et al., 2005b). In order to address this deficiency, an improved RICEWQ 1.6.4v was developed which allows irrigation and drainage to occur concurrently and also distinguishes the different degradation processes (hydrolysis, photolysis, microbial degradation) involved in pesticide dissipation in paddy water and soil. Recent validation studies with European datasets illustrated that RICEWQ 1.6.4v was equally efficient in simulating the environmental fate of pesticides in rice paddies as its previous version RICEWQ 1.6.2v (Karpouzas et al., 2006a). RICEWQ using daily time steps, simultaneously tracks mass balance of the chemical in the rice foliage, the water column and the paddy soil/sediment. RICEWQ considers all the major processes controlling the environmental fate of a pesticide applied in rice paddies including chemical and microbial decay in water, paddy soil/sediment and foliage, pesticide loss through leaching, overflow or controlled drainage and volatilization. Chemical partitioning between paddy water and soil occurs through direct partitioning, diffusion, settling of chemical adsorbed to suspended paddy soil and re-suspension of adsorbed paddy soil. RICEWQ was linked to the VADOFT sub-model in order to provide GW PECs at selected soil depths. RICEWQ and VADOFT were integrated by transferring water and pesticide flux predicted as seepage by RICEWQ as prescribed boundary loadings into VADOFT (Figure 2). The top 5 cm of the soil profile is represented by the active soil layer in RICEWQ. The remainder of the soil profile is represented as multiple compartments in VADOFT. The bottom of the active soil layer is the interface between the two sub-systems represented by the two models. When irrigation and precipitation exceeds the depth of the paddy outlet overflow occurs. When soil moisture in the paddy exceeds field capacity, percolation to VADOFT commences. As the paddy dries, soil moisture can decrease to the wilting point through evapotranspiration. It should be clarified that the term seepage refers to water and chemical percolating from paddy sediment into the

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

129

RICEWQ runoff/controlled drainage RICEWQ-VADOFT interface

paddy water leaching

active sediment layer

VADOFT

subsoil horizon

Ground water

Fig. 2. A schematic representation of the modeled system of RICEWQ 1.6.2v with a build-in interface between RICEWQ and VADOFT models.

vadose zone and leaching refers to downward movement of water and chemical within the vadose zone or from vadose zone to GW. Recent improvements of the RICEWQ model included an improved irrigation/ drainage routine and also distinguishes between the different degradation processes, including abiotic and biotic degradation, controlling pesticide loss in paddy water and soil. A detailed description of the model is given in the user’s manuals for RICEWQ (Williams et al., 1999) and elsewhere (Miao et al., 2003a; Karpouzas et al., 2005b). 2.2. Parameterization, uncertainty and sensitivity analysis of RICEWQ model The variables required for RICEWQ parameterization could be grouped into five categories including parameters describing meteorological data, crop practices, water management practices, paddy sediment characteristics and physicochemical and environmental fate parameters of pesticides (Table 1). 2.2.1. Meteorological and agronomic parameters

The model operates with a separate meteorological input file which provides daily precipitation and evapotranspiration. The model uses these daily meteorological inputs to calculate water depth in the paddy field and thus to predict possible water and pesticide losses via overflow or vertical infiltration. The dates of the major agronomic practices applied in rice paddy are also required

130

D. G. Karpouzas and Z. Miao

Table 1. Input parameters required for RICEWQ parameterization Data Meteorological data Crop practices

Paddy water

Paddy sediment

Pesticides

Parameters Daily precipitation (cm) Daily evapotranspiration (cm) Seeding, emergence and maturation date Maximum crop coverage at maturation stage (fraction) Deposition of pesticide residues at harvest Surface area of paddy (ha) Paddy water depth at treatment (cm) Depth of paddy outlet (cm) Paddy water depth at which irrigation commences (cm) Paddy water depth at which irrigation terminates (cm) Date and type of water practice applied (irrigation or drainage) Irrigation and drainage rate (cm/day) Seepage rate (cm/day) Depth of active sediment (cm) Wilting point of sediment (cm3/cm3) Bulk density of sediment (t/m3) Concentration of suspended sediment (mg/L) Mixing depth for direct partitioning to sediment, VBIND (cm) Settling velocity (m/day) Mixing velocity, VMIX (m/day) Water/sediment partition coefficient (ml/g) Microbial degradation rate in water (day⫺1) Photolysis rate in water (day⫺1) Hydrolysis rate in water (day⫺1) Microbial degradation rate in sediment (day⫺1) Abiotic degradation rate in sediment (day⫺1) Degradation rate in foliage (day⫺1) Wash off rate per cm of precipitation Volatilization coefficient (m/day) Pesticide water solubility (mg/L) Number of pesticide applications Pesticide application rate (g/ha) Pesticide application date Application efficiency (%)

by the RICEWQ model including seeding, emergence, maturation and harvest date. It is perceived that crop emergence occurs 5–7 days after seeding (DAS) depending on the temperature of irrigation water. Rice maturation usually coincides with the final drainage of the paddy fields and occurs 15–20 days before harvest. The maximum crop coverage at the maturation stage along with emergence and maturation dates are used by the model to calculate the amount of pesticide intercepted by the crop for post-emergent pesticide applications. When the model is used for regulatory purposes a crop coverage value of 90% should be used in accordance with FOCUS GW recommendations for grasses (FOCUS, 2000).

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

131

2.2.2. Water management parameters

Water management in rice paddies is very critical for crop development but it also strongly influences the environmental fate of pesticides in paddy systems. The RICEWQ could adequately simulate the various pesticide and water management practices applied in different rice cultivated areas in Europe. Parameterization of the RICEWQ water management routine requires the dates of commencement and termination of paddy irrigation or drainage, the depth of water maintained in the paddy field during irrigation/drainage periods, seepage, irrigation and drainage rates and finally the depth of paddy outlet. Variation in the values given to the latter parameter allows the user to select if intermittent or continuous flow – through irrigation system will be simulated. For example, setting the depth of paddy outlet at 20 cm and the depth of paddy water at which irrigation will cease at 10 cm prevents any overflow to occur during this intermittent irrigation period. In contrast, setting the depth of paddy outlet to 0 cm and the paddy water depth at which irrigation will commence and cease at 9 and 10 cm respectively, will allow irrigation and drainage to occur concurrently at the given irrigation and controlled drainage rates, while maintaining the water depth in the paddy field between 9 and 10 cm. The model also requires a seepage/percolation rate of paddy water through the soil/sediment, which cannot be varied during the simulation period. Previous validation studies have identified seepage rate as a key parameter influencing the leaching predictions of the RICEWQ model (Miao et al., 2003a). Sensitivity analysis of the model have indicated that higher seepage rates have led to higher pesticide concentration leaching to the deeper soil layers and to the vadose zone (Miao et al., 2003a). 2.2.3. Paddy soil/sediment parameters

Certain parameters related to the characterization of paddy sediment control the amount of pesticide partitioning to paddy sediment and thus the amount of pesticide which is available for leaching to the deeper soil layers. The depth of active sediment (DACT), which constitutes the interface between RICEWQ and VADOFT models, significantly affect leaching predictions. DACT is a user-defined parameter which is not measurable in the field. It is parameterized based on the depth of sampling used during sediment collection in field dissipation studies and usually varies between 1 and 5 cm (Fajardo et al., 2000; Ferrero et al., 2001; Vidotto et al., 2004; Watanabe et al., 2006). It has been argued that the top 1 cm of the paddy soil/sediment is always at the aerobic state unlike the immediately below soil zones (2–10 cm) which can be considered anaerobic (Fajardo et al., 2000). However, practical difficulties in the collection of only the top 1-cm of the paddy sediment in dissipation studies identified that 5 cm is a good approximation of the actual depth of the active paddy sediment layer. Sensitivity analysis tests with DACT and RICEWQ revealed that leaching predictions are very sensitive to DACT with lower values leading to higher pesticide concentrations in the deeper soil layers (Karpouzas et al., 2005a). Lower values of DACT represent a thinner active sediment layer; a shorter time of interaction between the incoming pesticide and sediment and,

132

D. G. Karpouzas and Z. Miao

consequently, lower pesticide losses due to adsorption and degradation in the active sediment layer. Another group of parameters, which are all user-estimated and not measurable, determine pesticide partitioning between paddy water and soil/sediment and include concentration of suspended soil /sediment (CSS), settling velocity of suspended soil/sediment (VSETL), mixing velocity by molecular diffusion (VMIX) and mixing depth to allow direct partitioning to paddy soil/sediment (VBIND). A sensitivity analysis of the model against these parameters revealed that variation of VSETL and CSS did not significantly influence model predictions (Karpouzas et al., 2005b). In contrast, leaching predictions of the model were very sensitive to input parameters like VBIND and VMIX (Figure 3). Higher VBIND and VMIX values resulted in higher amounts of pesticides partitioning

25000 (a) 20000

Total pesticide massentering the vadose zone (mg/ha)

15000

10000

5000

0 0

0.1

0.2

0.3

0.4

0.5

VBIND values 25000 (b) 20000 15000 10000 5000 0 0.00

0.05

0.10

0.15

0.20

0.25

0.30

VMIX values

Fig. 3. RICEWQ 1.6.2v sensitivity to (a) VBIND and (b) VMIX input parameters controlling partitioning of pesticide between paddy water and sediment (adapted from Karpouzas et al., 2005b).

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

133

to paddy sediment and, thus, higher pesticide leaching. The model was very sensitive to VMIX values in the range of 0.01–0.1 m/day and no sensitivity was shown when VMIX values higher than 0.1 m/day were given (Figure 3b). VMIX controls pesticide partitioning between paddy water and sediment via molecular diffusion. VMIX should be given high values in cases where pesticide is directly applied onto drained paddies followed by field submersion within the next 24–48 h. In these cases, high VMIX values allow repartitioning of pesticide from paddy soil/sediment to paddy water which floods the paddy field. The model was very sensitive to the broad range of VBIND values tested in the study (0–0.5 cm), although values up to 0.3 cm are considered realistic for model calibration (Williams et al., 1999) (Figure 3a). Uncertainty analysis of the RICEWQ model revealed that VBIND was one of the parameters seriously contributing to model uncertainty, thus, it should be parameterized carefully in order to avoid erroneous results (Miao et al., 2003c). Recently, Karpouzas et al. (unpublished data) using eight different datasets from field dissipation studies employed in paddy fields in Europe (Ferrero et al., 2001; Vidotto et al., 2004), Japan (Watanabe and Takagi, 2000b; Watanabe et al., 2006) and USA (Ross and Sava, 1986) identified a positive correlation between VBIND and the pesticide adsorption coefficient Kd which could be used as a preliminary guide for model parameterization. VBIND is a measure of how quickly a pesticide partitions onto paddy sediment. Generally, higher VBIND values should be used for pesticides showing high sorption affinity, although a quick calibration should be performed in cases where monitoring data are available. For example, when dissipation studies in water/sediment systems have revealed that the major part of the applied pesticide rapidly partitions onto paddy sediment, then a high VBIND value should be given in order to adequately describe pesticide partitioning between paddy water and soil/sediment. Another parameter, which along with VBIND is a major contributor to model’s uncertainty as far as leaching and runoff predictions are concerned, is the adsorption coefficient Kd (Miao et al., 2003a,c). Ideally Kd values for model parameterization should be derived from laboratory adsorption studies employed with at least two characteristic paddy soils. However, in cases where no such studies are available, Kd could be calculated by literature Koc values using the following formula: K d ⫽ K oc ⫻ Foc

(1)

where Foc is the fraction of soil organic carbon. Generally, all guidelines provided by the FOCUS GW group for the selection of adsorption variables for model parameterization are also applicable here (FOCUS, 2000). 2.2.4. Physicochemical pesticide parameters

The model also requires the parameterization of a set of variables associated with the physiochemical characteristics of the pesticide including water solubility, pesticide application date, application rate, application efficiency and times of

134

D. G. Karpouzas and Z. Miao

application per year. All the above data could be obtained from the good agricultural practice (GAP) document for each pesticide which is an indispensable part of the dossier. 2.2.5. Parameters controlling pesticide dissipation

Another group of variables control the processes involved in pesticide dissipation in paddy systems and include abiotic and biotic degradation rates of pesticides in paddy water, sediment and foliage, volatilization coefficient, wash-off coefficient and pesticide adsorption coefficient. RICEWQ allows the separation of the different abiotic and biotic degradation processes, thus, the user is able to input separately the rates of photolytic, hydrolytic and microbial degradation of pesticide in paddy water. The same applies to paddy sediment where the user is allowed to input separately the rates of pesticide degradation under wet and dry soil conditions. When the model is used for regulatory purposes such data could be directly obtained from the appropriate sections of Annexes II and III which provide laboratory derived DT50s for all the above processes. The flexibility of the RICEWQ model to distinguish between the various dissipation processes allows its application for predicting the environmental fate of all type of pesticides including those that are particularly labile to hydrolytic or photolytic degradation in paddy water. For example, RICEWQ could be used for pesticides like pendimethalin which is considered labile to photolysis. It should be noted that the approach used to derive degradation parameters from experimental datasets must be consistent with the methodology used in the simulation model. RICEWQ model assumes that all degradation processes conform with first order kinetics (Williams et al., 1999), thus, parameters estimated using non-linear equations are usually inappropriate (Beulke and Brown, 2001). It is recommended that the degradation rate of pesticides in paddy water to be derived from the DT50water of the water/sediment study, while the corresponding degradation rate in paddy soil/sediment to be obtained by the DT50 value of the aerobic flooded soil study in the laboratory. Alternatively, the DT50 value of the whole system could be used for calculating the degradation rate of pesticide in paddy soil/sediment. In cases where, the DT50water and DT50whole system values derived from an aerobic water/sediment study are similar it can be assumed that the contribution of degradation in paddy soil/sediment is minimal and thus the DT50whole system could be used for calculating the degradation rates in both paddy water and sediment. In contrast, for compounds which are more persistent in paddy soil/sediment than in paddy water separate degradation rates should be calculated and used for model parameterization. It is quite common, that in the absence of specific degradation data describing the contribution of photolysis, hydrolysis or abiotic degradation in paddy water and paddy sediment, lumped DT50s from field dissipation studies are commonly used. In that case the DT50 reflect the time for 50% loss of the pesticide due to a number of dissipation processes including volatilization, photolysis and plant uptake. If these values are used for modeling, other dissipation subroutines such as volatilization, photolysis or hydrolysis should be disabled. An advantage of field DT50s over laboratory data is that they are determined under conditions

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

135

specific for the field and thus may closely match the situation which is to be modeled (Beulke and Brown, 2001). An example of how to select degradation parameters for model parameterization is shown in Table 2. It should be mentioned here that all the guidelines given by the FOCUS GW and SW groups for the selection of degradation parameters when more than one degradation value is given are also applicable here (FOCUS, 2000, 2001). Table 2. An example of how degradation data should be selected for a realistic parameterization of the RICEWQ model Studies

Values (days)

DT50 – aerobic soil (20ºC)

20

DT50 – aerobic flooded soil (20ºC)

28

DT50 – anaerobic flooded soil (20ºC)

159

DT50 – field dissipation studies

35

Hydrolytic DT50 Photolytic DT50 Water/sediment study DT50water

No hydrolysis No photolysis 5

Water/sediment study DT50whole system Volatilization

35 35% loss field study

Choice This value could be used to calculate the degradation rates in the deeper soil layers, beneath the top anaerobic soil layer, which are generally considered aerobic. The biodegradation factors suggested from FOCUS GW group should be included in the calculation of degradation rates in the deeper soil zones This value would be used for the calculation of the degradation rate of pesticide in paddy sediment. The paddy sediment is considered aerobic and the higher DT50whole system compared with DT50water suggests that the pesticide is more persistent in paddy sediment than in paddy water This value would be used for calculating the degradation rate in the first soil layer beneath the active sediment layer which is generally considered anaerobic DT50field could be used instead of laboratory data; in that case the degradation rate in paddy water will be calculated by the DT50field and the volatilization routine of the model will be disabled Subroutines set to 0 Subroutines set to 0 This value would be used for the calculation of the degradation rate in paddy water Could be disabled if field dissipation data are used for model parameterization

136

D. G. Karpouzas and Z. Miao

For the degradation of pesticide in crop foliage, according to the EC 91/414 Directive it is not essential for agrochemical companies to provide data on pesticide degradation in crop foliage, thus, it is common that no such information is available. In that case the default DT50folliage of 10 days established by the FOCUS GW group could be utilized for model parameterization (FOCUS, 2000). It should be noted that input degradation rate of pesticide in crop foliage is meaningful only for pesticides applied post-emergence since the model considers that crop interception is the only route for pesticide partitioning in rice plants, thus it does not consider any plant uptake. To summarize, the selection of the most appropriate degradation data used for the correct model parameterization should be done after a thorough consideration of all the available degradation values and the particularities of the studied compound regarding its vulnerability to certain degradation processes. For most uses of the RICEWQ up to date, its volatilization routine was set to zero as relevant input data are lacking and lumped half-life values including volatilization and photodegradation were used for model parameterization (Miao et al., 2003a,b; Karpouzas et al., 2005a,b, 2006a). Since volatilization has been identified as a significant process determining the fate of certain pesticides in rice paddies (Ross and Sava, 1986; Seiber et al., 1986), improvement of the existing volatilization routine of RICEWQ model is indispensable. RICEWQ considers that volatilization occurs only by the water dissolved fraction of the pesticide and calculates the mass volatilized through an empirical formula:  F ⫻ Mw  M volat ⫽  K volat DW  dt Dw 

(2)

where Mvolat is the mass of pesticide volatilized (mg), Kvolat an empirical coefficient which is used by the model for the description of pesticide volatilization (m/day); FDW the fraction of pesticide mass in dissolved phase (%); Mw the pesticide mass in water (mg); Dw the depth of paddy water (m) at each time step dt (day) (Williams et al., 1999). The model requires parameterization of the coefficient Kvolat which is not readily measurable and easy to use, in contrast to most of the other pesticide fate models like PELMO, PEARL and PRZM which use physicochemical parameters like vapor pressure or Henry’s law constant as inputs to simulate volatilization (FOCUS, 2000). Therefore, a simple approach was taken in order to identify any possible correlation between the empiric coefficient Kvolat and physicochemical pesticide parameters like vapor pressure or Henry’s law constants (Hk) (Ferrari et al., 2005). For this purpose, a dataset derived from a field study was used, where the volatilization of five pesticides applied in an artificial paddy field was measured (Ferrari et al., 2005). The RICEWQ model was used to simulate the fate of pesticides in the artificial paddy system and the Kvolat showing the best fit to the measured values was derived for all pesticides. When these Kvolat values were used as inputs for RICEWQ, the model simulated with good accuracy the dissipation of pesticides from paddy water (Ferrari et al., 2005). By simply plotting the values of Kvolat for the five test compounds against their corresponding vapor pressure or Hk values a good regression correlation was

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

137

Volatilization Coefficient (m day-1)

0.005

0.004 All Compounds

0.003

y = 0.0022x0.3505 R2 = 0.9149

Chlorpyrifos omitted 0.002

y = 0.0016x

0.001

0.3147

R2 = 0.975 0.000 0.0

0.2

0.4

0.6

0.8 3

1.0

1.2

-1

Henrys Law Constant (Pa m mol )

Fig. 4. Regression correlation between the volatilization coefficient Kvolat of the RICEWQ model and Henry’s law constant (adapted from Ferrari et al., 2005).

identified only with the latter (Figure 4). This was expected since Hk is a more relevant parameter for water applied pesticides since it considers water solubility as well as vapor pressure (Kubiak et al., 2003). An improved regression correlation between Kvolat and Hk was evident when chlorpyrifos, one of the five pesticides tested was ignored (Figure 4). Exclusion of chlorpyrifos was justified since the simulation of its volatilization behavior by the model was poor (Ferrari et al., 2005). The regression equation derived (Kvolat ⫽ 0.0016 ⫻ Hk0.3147) is a preliminary attempt to improve the volatilization routine of the RICEWQ model. However, this regression equation should be further validated for consistency with other datasets for pesticides with Hk within the range of values of the pesticides used in this study. Similar regression equations between substance properties and volatilization losses have been used in the past to estimate pesticide emissions from soil and plant surfaces (Hassink et al., 2003). However, a more complex approach might be necessary for further improvement of the volatilization routine of the model. Such an approach could be the implementation of a well-validated volatilization routine of a SW model like EXAMS into the RICEWQ model. 2.2.6. Simulation of pesticide metabolites with the RICEWQ model

RICEWQ could simulate, apart from the parent compound, the fate of up to three metabolic products formed in any of the three environmental compartments, water, soil and foliage, of the paddy system. As input parameters for the metabolites the model requires values for all the variables requested for the parent compound, thus, degradation rates in water, sediment, foliage, wash-off rate, volatilization coefficient, VBIND, VMIX, VSETL and water solubility. In addition, the model requires the data on the fraction of the parent compound which

138

D. G. Karpouzas and Z. Miao

is transformed to the metabolic product in each of the environmental compartments of the paddy systems. The ability of RICEWQ to simulate the fate of pesticide metabolites in rice paddy systems is particularly important in higher tier risk assessment since several compounds are rapidly transformed into more persistent or toxic metabolites whose fate should be studied as well. For example, cyhalofop-butyl, a post-emergence rice herbicide, is rapidly transformed within less than 12 h to its corresponding acid and subsequently to cyhalofop amide and cyhalofop diacid which are more persistent chemicals than the parent compound (Jackson et al., 1999). Similarly, propanil hydrolysis in water leads to the formation of 3,5-dichloroaniline which possess higher toxicity than the parent compound (Pothuluri et al., 1991). RICEWQ is at the moment the only available model, specific for rice cultivation, which could simultaneously simulate the fate of parent compound and metabolic products. 2.2.7. Parameterization of the VADOFT sub-model

Apart from the RICEWQ input file, another indispensable part of the model which should be parameterized, is the input file of the VADOFT sub-model which simulates pesticide fate in the deeper soil layers beneath rice paddy. The main hydrological and pesticide parameters required for the flow and transport modules of the VADOFT sub-model are shown in Table 3. Generally, most of the hydrological parameters are difficult to measure. The model allows the division of the soil horizon into texturally different soil zones, and hydrological and physicochemical characteristics of each soil zone could be derived or calculated from tabulated values based on their soil texture. The model requires input of the degradation rate of the pesticide in each soil zone. It is suggested that the DT50 value derived from the anaerobic soil degradation study is the most appropriate value to be used for the calculation of the pesticide degradation rate in the first soil layer (0–20 cm) immediately below the active sediment layer of the paddy system (see example in Table 2). For the deeper soil layers (20–100 cm) it is believed that they are generally in an aerobic state and therefore the degradation rates could be calculated by the DT50 values derived from the aerobic soil degradation study. However, the degradation rates in the deeper soil zones should be adjusted for the reduction in the microbial degradation of pesticides with soil depth using the biodegradation factors suggested by the FOCUS GW group (FOCUS, 2000). The retardation coefficient, which is a measure of the pesticide sorption affinity, should be calculated and input for each soil zone (Table 3). The adsorption coefficient of the pesticide for each soil zone could be calculated by its corresponding Koc value as shown before. A modeling parameter which significantly influences the leaching predictions of the VADOFT sub-model is the number of nodal layers that will be used for separating the simulated soil horizon. Previous studies with the RICEWQ 1.6.2v (RICEWQ – VADOFT linked model) and the herbicides cinosulfuron and propanil revealed that the lower the number of nodal layers used, the highest is the leaching concentrations at the bottom of the simulated horizon (Miao et al., 2003a; Karpouzas et al., 2005a). The reason is that with a lower number of nodal layers

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

139

Table 3. Physicochemical and hydraulic parameters required for the parameterization of the VADOFT model Parameters Number of nodal layers Saturated hydraulic conductivity (K ) (cm/day) Saturated water content (
s) (cm3/cm3) Residual water content (
r) (cm3/cm3) Effective porosity () (dimensionless) Residual water phase saturation (Swr) (dimensionless) Leading coefficient of the saturation versus capillary head () (cm⫺1) Power index of saturation versus capillary head relationship () (cm⫺1) Power index of the saturation versus capillary head relationship () (cm⫺1) Longitudinal dispersion (L) (cm) Retardation coefficient (R) (dimensionless) Darcy velocity (q) (cm/day) Pesticide degradation rate (day⫺1)

Comments – Tabulated values from Carsel et al. (1998) depending on soil texture Tabulated values from Carsel et al. (1998) depending on soil texture Tabulated values from Carsel et al. (1998) depending on soil texture  ⫽
s ⫺
r
Swr ⫽ r
s Tabulated values from Carsel et al. (1998) depending on soil texture Tabulated values from Carsel et al. (1998) depending on soil texture Tabulated values from Carsel et al. (1998) depending on soil texture – K  R ⫽ 1⫹ d
s Hi q ⫽ Km ∑ Li –

the soil column is divided into fewer and wider layers. Numerical dispersion is increased, so the distance traveled by the leading edge of the pesticide increases. 2.3. Tier-3 scenarios at member-state level A major conclusion of the Med-Rice group of experts was that there is a need for the development of several higher tier scenarios at member-state level, which would be utilized in the pesticide risk assessment for registration purposes (Med-Rice, 2003). The need for several national rice scenarios instead of one or few pan-European scenarios was dictated by the considerable differences in the crop, water and pesticide management practices used in the different rice-cultivating European countries. 2.3.1. Scenarios development for Greece

Rice is cultivated under submerged conditions in specific river basins of northern Greece, the Axios river basin and the Strimonas river basin. The southern part of the former basin is considered the main rice-cultivating area of Greece and constitutes over 70% of the total area cultivated with rice in Greece (Figure 5).

140

D. G. Karpouzas and Z. Miao

Fig. 5. A map of Greece where the main rice cultivation areas are indicated

A survey of the current agronomic practices applied in the Axios river basin and reports from recent studies employed in the specific area formed the basis for the development of a national scenario (Ntanos, 1997, 2001; Ntanos et al., 2000). Insecticides and fungicides are only occasionally applied and herbicides constitute the main group of pesticides used in rice. Molinate and propanil have been the most used herbicides in rice paddies of the area for the control of barnyardgrass (Echinochloa crus-galli) and red rice (Oryza sativa) for the last 30 years (Ntanos, 2001). Molinate is applied as a pre-emergence treatment in flooded fields unlike propanil which is usually applied as post-emergence treatment in drained fields. The extended use of these herbicides in the area was documented by the presence of molinate at high concentrations in the rainfall water collected from the area (Charizopoulos and Papadopoulou-Mourkidou, 1999), but also by their frequent detection in the phreatic horizon, the drainage canals and the adjacent Axios and Loudias rivers (Papadopoulou-Mourkidou et al., 2004). However, in recent years these herbicides have been partly replaced by novel substances like cyhalofop, azimsulfuron and penoxsulam (Eleftherohorinos and Dhima, 2002). The scenario developed for the Axios river basin is briefly presented in Table 4. The RICEWQ model was parameterized using the scenario developed and validated against data from a field study employed during 1992–1994 in specific paddies located in the southern part of the Axios river basin (Papadopoulou-Mourkidou et al., 2004). The model was only validated for its capacity to simulate leaching of molinate and propanil. After extended calibration, the model predicted with good accuracy the concentrations of herbicides leaching to the deeper soil layers (Karpouzas et al., 2005a). 2.3.2. Scenarios development for Italy

The whole rice cultivated area in Italy was analyzed using maps of various scales. Rice is grown throughout Italy including south (Calabria), center (Lazio), north (Lombardia), northwest (Piemonte), northeast (Emilia Romagna) and islands like

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

141

Table 4. A summary of the scenario developed for rice cultivation in Greece Parameters Meteorological data Agronomic Practices

Pesticide

Hydrology Soil Properties

Values Precipitation (cm) Evapotranspiration (cm) Rice cropping season (flooding) Seeding date Emergence date Crop maturation Crop harvest Irrigation–drainage practices Number of applications per year Application rate (kg/ha) Application date Adsorption coefficient (cm3/g) Solubility in water (mg/L) Volatilization rate (m/day) Degradation rate in water, sediment, foliage (day⫺1) Paddy water depth (cm) Groundwater level Surface area of paddies (ha) Soil texture pH Organic matter (%)

aNumbers

A 3-year (1992–1994) daily data from local meteorological station Calculated with RadEst 3.00v May–September Beginning of May 7 DAS Mid September Mid October Intermittent irrigation GAP documents GAP documents GAP documents Pesticide dependent Pesticide dependent Pesticide dependent Pesticide dependent 9–11 Shallow 2 Clay (55 ⫾10.8)a; sand (35.8 ⫾7.1); silt (9.2 ⫾7) 9.38 ⫾0.51 2.30 ⫾0.92

in the brackets represent the mean content of soils in clay, sand and silt ⫾ the standard

deviation.

Sardegna. Once rice growing regions were identified, a regional analysis to a scale of 1:100,000 was performed in order to identify areas entailing higher risk for pesticide contamination. Subsequently, a more detailed analysis to a scale of 1:50,000–60,000 was employed in order to better define the hydrological profile of those areas. The main criterion for the identification of areas more vulnerable to pesticide contamination was the total surface area cultivated with rice (relative and absolute distribution). Consequently, four areas were selected including provinces of Vercelli (VC), Mantova (MN), Pavia (PV) and Milano (MI), all located in the Po river basin that represents more than 95% of the whole area cultivated with rice in Italy (242,325 ha) (Figure 6). The sites vary considerably in their main characteristics such as rice varieties cropped, soil properties, hydrological characteristics of respective soil horizons, water quality, type of irrigation,

142

D. G. Karpouzas and Z. Miao

Rice fields Water courses and water bodies

MI

VC

MN PV Adriatic sea Po river

N W

E S

Tyrrhenian sea

50000

0

50000

100000

150000 200000

Meters

Fig. 6. A map of the main rice-cultivated basin in Italy, the Po river basin.

size of receiving SW bodies and their interconnections (Table 5). In contrast, similar agronomic practices are applied in all selected areas and their application time sequence is shown in Figure 7. The above scenario has been successfully evaluated by Miao et al. (2003b). 2.4. Other available models No other model which could be used for higher tier exposure assessment in rice paddies has been validated in Europe. PADDY is a mathematical model developed in Japan for predicting pesticide concentrations in paddy water and soil (Inao and Kitamura, 1999). The PADDY model is executed using Microsoft® Visual Basic® based on Windows®. The model assumes that the paddy system consists of two interrelated compartments (1) a surface layer composed of paddy field water and an active soil layer (5 cm) which comprises pore water and soil compartments under flooded conditions and (2) a sub-surface soil layer which is also composed of soil particles and pore water and the thickness of each layer is 0.5 cm. Each of the compartments is a completely mixed reactor and the driving force of mass transfer is the pesticide concentration gradient between the different compartments (Inao and Kitamura, 1999). PADDY model considers all the basic processes controlling the fate in paddy systems including pesticide dissolution from granules, volatilization, degradation and pesticide adsorption/desorption to paddy sediment and also leaching to the deeper soil layers. However, the model does not consider daily fluctuations in paddy water depth and ignores the contribution of photodegradation in overall pesticide dissipation. Daily fluctuations in paddy water depth are closely associated with water management practices, evapotranspiration and precipitation and is an important factor in calculating

Region Mantova

Area (ha) 1092

Pavia

11,879

Milano

79,558

Vercelli

67,347

aContinuous:

Irrigationa

Ground waterb

Surface waterb

Organic matter (%)

Clay (%)

Sand (%)

Continuous

Shallow

28.2 (0.3 ⫼86.9)c

9.9 (0.2 ⫼ 15.2)

58.1 (30.7 ⫼86.7)

Intermittent/ continuous Intermittent/ continuous Intermittent

Shallow

Natural/artificial bodies Natural/artificial bodies Natural/artificial bodies Natural/artificial bodies

15.2 (1.0 ⫼38.9)

2.2 (0.0 ⫼8.5)

85.0 (54.8 ⫼99.9)

21.1 (2.7 ⫼36.9)

7.2 (0.0 ⫼ 28.0)

73.6 (23.4 ⫼99.4)

19.6 (1.3 ⫼ 38.5)

4.2 (0.1 ⫼ 10.1)

72.8 (24.4 ⫼99.5)

Shallow Shallow

water flows continuously from one paddy to the other for most of the growing season; intermittent: the outflow gate of the paddy is often closed. both GW and SW the risk for pesticide contamination is high. cNumber within brackets represent the variation of the specific properties in the area. bFor

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

Table 5. Different representative scenarios of rice cultivation in Italy

143

144

D. G. Karpouzas and Z. Miao irrigation

crop emergence

seeding

fungicide application

herbicide application

harvest

maturation period

15-20 cm paddy water March

April

May

paddy water June

July

August

10-11 cm September

October

Fig. 7. A schematic representation of the different agronomic, irrigation and pesticide practices applied during the cultivating season in rice paddies in Italy.

water and mass balance. The PADDY model has been successfully validated and applied in a number of cases in Japan (Takagi et al., 1998; Inao and Kitamura, 1999; Inao, 2003). However, several of its routines should be adjusted in order to be readily applicable in rice cultivation practices in Europe. Another model which was also developed in Japan for describing pesticide fate in rice paddy systems is called PCPF-1 (Watanabe and Takagi, 2000a). It is a lumped parameter model which simulates the fate and transport of pesticides in the two compartments of paddy fields: paddy water and paddy soil (Watanabe and Takagi, 2000a). The paddy water compartment was assumed to be a completely mixed reactor having variable water depths. The paddy soil is also assumed to be a completely mixed compartment but with a constant depth of usually 1.0 cm where pesticide dissipation and transport processes occur under oxidative flooded condition. Both compartments are assumed to be homogeneous, and having uniform and unsteady chemical concentrations (Watanabe and Takagi, 2000b). PCPF-1 could not simulate leaching of pesticides below the top 5 cm of the active paddy soil layer. However, only recently an interface between the PCPF-1 model and the SWMS 2D, which is the open fortran code of Hydrus 2D model, enabled the simulation of pesticide fate and transport in the soil profile beneath rice paddies (Tournebize et al., 2004, 2006). Water balance in rice field is determined by the following components: irrigation, precipitation, overflow/controlled drainage, evapotranspiration, lateral seepage and vertical percolation. PCPF-1 provides daily concentrations of pesticide in paddy water considering that the dominant processes controlling pesticide dissipation in paddy water are: pesticide dissolution from granular formulation, pesticide desorption from paddy soil to water, volatilization, microbial and photochemical degradation, dilution or concentration of the pesticide dissolved in paddy water by precipitation, irrigation or evapotranspiration. In the paddy soil layer, PCPF-1 also provides daily pesticide concentrations in paddy soil considering that adsorption/ desorption, microbial degradation and leaching of pesticides to the subsurface soil beneath the surface soil layer are the major processes controlling its dissipation in paddy soil. PCPF-1 model-program was coded using Visual Basic® for

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

145

applications in Microsoft Excel®. The Macro program calculates and automatically creates output data and figures in a Microsoft Excel® file. In comparison with PADDY, PCPF-1 has incorporated a routine to account for the daily fluctuation of paddy water depth and also for pesticide losses due to photolysis (Watanabe and Takagi, 2000a). PCPF-1 has been validated with field datasets obtained under Japanese conditions (Watanabe et al., 2006; Watanabe and Takagi, 2000b). Only recently a slightly modified version of the original PCPF-1 model was tested for its ability to effectively simulate the environmental fate of pesticides applied in rice paddies under European conditions (Karpouzas et al., 2006a). The original PCPF-1 algorithm was modified so as to more realistically represent the processes involved in pesticide fate after spraying and not after granular application. The pesticide dissolution was assumed to proceed only in the paddy water compartment. For the soil compartment, it was assumed that the transfer of pesticide from paddy water to paddy soil was achieved mainly by the vertical percolation of paddy water. The results of these validation studies of PCPF-1 with European datasets revealed that PCPF-1 has the potential to be used under European conditions but further calibration and adjustments of certain parameters is needed in order to more effectively predict pesticide exposure in rice paddies in Europe. 3. EXPOSURE ASSESSMENT AT TIER 4: BASIN-SCALE LEVEL Rice is commonly cultivated in large river basins in Europe, where high loads of applied herbicides have resulted in the contamination of related SW and GW systems. Therefore, it is relevant to consider pesticide risk assessment for rice crop at basin-scale level. According to the tiered risk assessment scheme (Figure 1), tier-3 assessment includes site-specific consideration and calculation of PECs at paddy-field scale, and tier-4 assessment includes landscape risk assessment using basin-scale scenarios, where mitigation processes could also be included. Previous studies have shown that RICEWQ 1.6.2v overestimates PECs in SW systems adjacent to rice paddies (Miao et al., 2003b). It was suggested that RICEWQ model when used in connection with a SW model like RIVWQ, TOXic Substances in Surface Waters (TOXSWA) or Exposure Analysis Modeling System (EXAMS) could provide realistic estimates of pesticide concentrations in surrounding SWs. Such modeling studies revealed that the combination of RICEWQ with the RIVWQ 2.02v model could provide a realistic estimation of PECs in SW bodies associated with treated rice paddies (Miao et al., 2003b; Warren et al., 2004; Karpouzas and Capri, 2006; Karpouzas et al., 2006b). 3.1. Description of the RIVWQ model RIVWQ 2.02. simulates the transport of organic chemicals in tributary stream systems based on the theory of constituent mass balance. The system geometry is represented using a link-mode approach in which the water body is divided into a number of discrete junctions (nodal points and headwaters) connected by flow channels. Dynamic constituent transport occurs between nodal junctions via links

146

D. G. Karpouzas and Z. Miao

and is a balance between river-driven flows and dispersion processes. RIVWQ can accommodate trapezoidal or rectangular cross-sections of SW bodies and geometric rating curves are used to calculate cross-sectional areas and water depths as functions of flow. Chemical transformation occurs within each node including dilution, volatilization, partitioning between water and sediment, decay in water and sediment and re-suspension from bed sediments. Boundary conditions for the RIVWQ model include: optional inflow of paddy water and compound mass as runoff or controlled drainage in different nodal points (corresponding to floodgates of paddy fields), optional incremental inflows and stream features such as river sediment sorption, dead storage (DS) and base water flow along the nodal network. As base flow is considered a constant flow in the river regardless of other inflows. A detailed description of the RIVWQ model is given in the user’s manuals (Williams et al., 2004), and also in previous work (Miao et al., 2003b). The variables required for the parameterization of RIVWQ can be categorized in four groups including parameters describing the network of SW bodies, physicochemical and pesticide fate parameters, sediment parameters and parameters describing the hydrological characteristics of the SW bodies simulated. A summary of the parameters in each category are shown in Table 6. Simulation of a system of ditches, canals which are interlinked and discharge their water into a larger natural SW body could be performed using RIVWQ. The modeled system of SW bodies is divided into a number of interlinked nodal layers and inflow is coming from one or several selected headwaters depending on the complexity of the system simulated. The length of each nodal layer could be varied depending on the SW body simulated but large variations, e.g. 5–2000 m should be avoided in order to precluded model instability. RIVWQ could simulate the fate of one chemical and up to three metabolites at the same time. This is quite useful in cases where pesticides are rapidly transformed to one or more persistent metabolites, whose fate is more environmentally relevant than the fate of the parent compound itself (FOCUS, 2001). The RIVWQ model parameters describing pesticide chemistry and environmental fate are similar to the parameters used in RICEWQ including pesticide decay rate in sediment and water, adsorption coefficient, volatilization coefficient and water solubility. Pesticide degradation rates in paddy water and sediment could be derived from a laboratory water/sediment aerobic study which should be provided for all compounds seeking registration. Depending on the nature of the study used to obtain the pesticide decay rates in water, volatilization coefficient could be disabled as has been described before for RICEWQ. The same group of parameters required for sediment characterization in RICEWQ are also included in RIVWQ model, such settling velocity, mixing velocity, depth of active sediment, porosity, bulk density and the concentration of suspended sediment. Parameterization of these variables in RIVWQ follows the same principals described for RICEWQ. The most important set of parameters controls the hydrology of the flow and transport of the simulated SW systems. In RIVWQ, the user is allowed to describe SW bodies with irregular or rectangular cross-sections using either geometric rating curves or the Manning’s equation. In the first instance, cross-sectional areas

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

147

Table 6. A summary of the main parameters used for the parameterization of RIVWQ Modeling parameters Number of nodal junctions Number of headwaters Pesticide parameters Degradation rate in water (day⫺1) Degradation rate in sediment (day⫺1) Adsorption coefficient (ml/g) Volatilization coefficient (m/day) Water solubility (mg/L) Sediment parameters Settling velocity (m/day) Mixing velocity (m/day), VMIX Mixing depth for direct partitioning to sediment (cm), VBIND Depth of active sediment (cm), DACT Porosity (dimensionless) Bulk density (g/ml) Concentration suspended sediment (mg/L), CSS SW body parameters Channel geometry format (1: rectangular channel flow rating curve; 2: variable channel based on flow rating curve; 3: rectangular channel Manning equation) Bottom width of channel (m) Qa coefficient in stage-flow rating curve Qb exponent in stage-flow rating curve Qc coefficient in stage-flow rating curve Qd exponent in stage-flow rating curve Qe coefficient in stage-flow rating curve Qf exponent in stage-flow rating curve Slope of channel downstream of junction (m/m) Manning roughness coefficient Dispersion coefficient (m2/s) Base flow (m3/s/km2 of drainage area) Dead storage (m, depth) Muskingum K coefficient (s) Muskingum X coefficient (fraction)

and water depths are calculated as a function of flow, unlike in the latter case where cross-sectional areas and water depth are calculated as a function of flow and slope. When geometric rating curves are used for SW bodies with rectangular cross-sections only Qc and Qd coefficients and the bottom width of the canal are required as inputs for the model. When geometric rating curves are used for SW bodies with irregular cross-section all Qa, Qb, Qc, Qd, Qe and Qf are required for model parameterization. In cases where the Manning’s equation is used, the bottom width, the slope of the SW body and the Manning roughness coefficient are required for model parameterization. Other parameters which are necessary for the parameterization of the flow routine of the RIVWQ model include the dispersion coefficient and the incremental drainage area which is associated with each junction point of the system.

148

D. G. Karpouzas and Z. Miao

RIVWQ also includes the Muskingum routing option for flood routing through a reach. A linear model for Muskingum routing flow in a stream could be written as S ⫽ K [ XI ⫹ (1⫺ X )Q]

(3)

where S is the water storage of one node at a certain time (m3), Q the water flow (m3/s), I the water inflow (m3/s) from the previous node and K and X are Muskingum coefficients. Generally, typical values of the Muskingum coefficient X vary between 0.1 and 0.3 depending on the shape of the modeled wedge storage. The Muskingum coefficient K represents the time of travel of the flood wave through a stream reach and is usually estimated by the ratio of the SW body segment to the water velocity in the SW body simulated. The Muskingum option should be parameterized when large dynamic SW systems like rivers are simulated. In contrast, the Muskingum option should be disabled in cases where smaller artificial or natural ditches or canals are to be modeled. RIVWQ model allows the user to attribute a constant or base flow to the modeled SW system which might change downstream by incremental additions of water at selected nodal junctions. Another parameter which influences water flow in the modeled SW systems is the DS. This parameter is usually utilized when small artificial or natural canals are simulated, where an obstacle like an animal’s nest could create a temporary storage of water hampering water flow and also molecular diffusion of any compound in the water. It should be noted that the DS parameter is commonly not used when large dynamic SW systems like rivers are simulated, since such obstacles creating DS could not significantly influence water flow in a dynamic SW body like a river. RIVWQ is compatible with pesticide runoff models including PRZM and RICEWQ. Incremental additions of water and pesticide inflow at selected junction points along the route of the modeled SW system are provided by external input files like the mass loading file (mas file) and the lateral inflow file (hyd file). The former input file provides daily summaries of pesticide mass loadings (mg) at selected junction points of the simulated SW body, compared with the latter input file which provides daily summaries of water volume loadings (m3) at the corresponding junction points. Such input files (mas and hyd) could be created manually by the output zzt file of the RICEWQ model which provides daily summaries of the pesticide mass (mg) and water volume releases (m3) from a rice paddy field due to overflow or controlled drainage. RIVWQ requires that the water volume loadings are presented as m3/s and not as m3/day which is the form provided as an output from RICEWQ. This unit conversion could be either done externally in a manual mode or internally by RIVWQ which contains a parameter for converting the values in the hyd file to the appropriate units. 3.2. Description of the combined use of RICEWQ – RIVWQ models The combined use of RICEWQ with RIVWQ has been utilized in order to simulate the fate of rice pesticides in the SW systems associated with rice paddies, including drainage ditches, canals and larger river systems (Miao et al., 2003b;

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

149

Karpouzas and Capri, 2006; Karpouzas et al., 2006b). The models are loosely coupled and daily based pesticide and water releases generated by RICEWQ due to overflow or controlled drainage are considered as water and chemical inputs for RIVWQ at selected junction points along the length of the SW system. Drift inputs to drainage canals adjacent to treated paddies could be also considered and added manually in the mass loading file at the day of application. However, drift inputs should be considered depending on the mode of application. For example, drift inputs were not considered for granular application of molinate, unlike for propanil where a 2.77% of the application rate should be considered as drift load onto adjacent canals. 3.3. Parameterization of RICEWQ, RIVWQ models using basin-scale scenarios Basin-scale scenarios which are considered representative for rice cultivation in certain European countries could be formed and utilized in the regulatory processes for exposure refinement when assessment at lower tiers (tiers 1–3) have indicated high risk. In addition, such scenarios could include mitigation measures which should be used in order to diminish exposure of associated SW systems to rice pesticides. A preliminary attempt to develop a simple watershed level scenarios using a combination of RICEWQ and RIVWQ models was first presented by Miao et al. (2003b). The authors developed a three-stage modeling exercise in order to evaluate which individual model or model combination is suitable for modeling pesticide fate at the paddy field scale and at the watershed level. The results of this exercise showed that RICEWQ is a useful tool for predicting pesticide fate at the paddy field scale but it significantly overestimated PECs when used for simulating the fate of pesticides in SW bodies associated with paddy fields. In contrast, when RICEWQ was only used for simulating the fate of pesticide in paddy fields and then RIVWQ simulated the fate of pesticide loaded in the receiving SW bodies, the PECs provided were within the measured range (Miao et al., 2003b). The simple watershed scenario included a paddy cultivation area of varying surface (20, 40 and 200 ha) which was all treated at the same day with the highest recommended dose of the fungicide tricyclazole (Figure 8). The fate of pesticide in the paddy area was simulated with RICEWQ and the overflow or controlled drainage predicted by the model constituted the water and pesticide input at the top-most point (headwater) of a 10 km-river system. Another incremental inflow of clean water without fungicide in the river was assumed at the 4 km point downstream of the headwater (Figure 8). The top width of the river system (0.5, 1, 5 m) and the headwater constant flow of the river (0 and 0.88 m3/s) were varied. The fate of the pesticide in the river testing all combinations of paddy surface area, river width and constant river flow was predicted using RIVWQ. The results illustrated that PECsw were negatively correlated with channel size and the distance from the pesticide inflow point but were positively correlated with the size of the pesticide-treated area. Inclusion of constant base flow in the river system significantly reduced PECsw due to increasing dilution of incoming pesticide

150

D. G. Karpouzas and Z. Miao

Fig. 8. A schematic representation of the simple watershed scenario developed by Miao et al. (2003b) and the combination of models used to conceptualize the scenario (adapted from Miao et al. 2003b).

mass. However, high constant base flow values accelerate pesticide transport in the river and minimize the time available for pesticide dissipation along the route of the river. Generally, topographic and hydrologic characteristics of the studied watershed should be well studied before used for model parameterization since RIVWQ is rather sensitive to these parameters. 3.3.1. Basin-scale scenario for Greece

Scenario development. A basin-scale scenario was developed based on the agricultural, water management and pesticide practices used in the Axios river basin in Greece. Soils of the rice-cultivated area in the Axios river basin are mainly heavy clay, clay loams with poor infiltration (0.1 cm /day) (PapadopoulouMourkidou et al., 2004). The simulated system was an intensively rice-cultivated watershed of 2000 ha consisting of 500 paddy fields (4 ha each). Paddy fields were grouped in 10 management blocks of 50 paddies. Each of the paddy blocks was associated to a drainage canal receiving pesticide mass and water releases as predicted by the RICEWQ 1.6.2v model. Drainage canals subsequently discharged their water into a large river system (Figure 9, Photograph 1). The fate of pesticides in each one of the paddy fields of the basin was simulated using the

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

drainage canal

4 ha rice paddies

4 ha rice paddies

4 ha rice paddies

4 ha rice paddies

4 ha rice paddies

drainage canal drainage canal drainage canal drainage canal

151

drainage canal drainage canal

4 ha rice paddies

4 ha rice paddies

4 ha rice paddies

4 ha rice paddies

4 ha rice paddies

R

RIVE

drainage canal drainage canal

Fig. 9. A schematic representation of the basin-scale scenario developed for the Axios river basin in Greece (adapted from Karpouzas et al., 2006).

RICEWQ model. Daily discharge from paddy fields, as overflow and controlled drainage, were used as loadings for the receiving drainage canals. The fate of pesticide in the drainage canals and in the receiving river system of the simulated basin was simulated in one run using the RIVWQ model. GW risk assessment was performed by the RICEWQ outputs. Parameterization of RICEWQ. A simulation period of 26 years for RICEWQ 1.6.2v was considered using daily meteorological data from local meteorological stations (FOCUS, 2000). A series of 10 different seeding dates during a 20-day period, from 20 April to 7 May of every year, were selected and were assigned to each of the 500 paddy fields of the simulated basin following a completely randomized block design. This period represents the main establishment period for rice crop in Greece. Subsequent agronomic practices including emergence, maturation and harvest date were applied 7, 110 and 130 DAS. The pesticides included in the study were the rice herbicides propanil and molinate. All paddies of the simulated basin were treated once a year with the maximum recommended dose; 35 and 10 DAS for propanil and molinate, respectively. The mode of pesticide application to rice paddies is different for propanil and molinate and it was modeled as described in previous sections. For parameterization of the VADOFT sub-model, the depth of the GW table was set to 1 m, and the soil horizon was subdivided into three soil zones (Med-Rice, 2003). The degradation rate inputs

152

D. G. Karpouzas and Z. Miao

Photo. 1. Receiving drainage canals adjacent to flooded rice paddies in the southern part of the Axios river basin.

for each soil layer were selected following the criteria and principals described earlier in the RICEWQ parameterization section (see Table 2). Parameterization of RIVWQ. For the parameterization of RIVWQ, a clay sediment was considered for both the drainage canals and the river system (Med-Rice, 2003). Hydrological characteristics and dimensions, including flow velocity and water discharge were derived by local observations. Drainage canals had a rectangular shape, with a total length of 5 km, depth and width of 1.5 and 4 m, respectively and a flow velocity of 0.1 m/s. The river systems simulated in both

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

153

scenarios had a rectangular shape with a total length of 10 km, depth and width of 3 and 60 m, respectively and a flow velocity of 0.3 m/s. Nodal segments with a length of 100 m for the drainage canals and 500 and 1000 m for the river system were selected for modeling purposes. Pesticide mass and water volume discharges from drainage canals into the river system occurred at 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5 and 5.5 km along the length of the river. Application of mitigation measures. An integral part of the tier-4 risk analysis is the inclusion of possible mitigation measures in order to diminish pesticide exposure in SW bodies. The prolongation of paddy closure after application of rice pesticides has been suggested as a possible mean of controlling pesticide releases in receiving SW systems (Crepeau and Kuivila, 2000; Vu et al., 2004; Watanabe et al., 2005). The effect of closure time applied to rice paddies after application of molinate, on the magnitude of its PECs in the related drainage canals and river was examined as a possible mitigation strategy. Three different closure periods: 5, 10 and 20 day after pesticide application were used for the parameterization of the RICEWQ model and the predicted pesticide mass and water volume lost by rice paddies were used as loadings into the RIVWQ model. GW risk assessment. The risk for GW contamination by the use of molinate and propanil was assessed by calculating their annual average concentrations leaching below the 1-m soil zone for the 20-year simulation and the 80th percentile value was considered as relevant GW PEC. The 80th percentile values for the two herbicides were well below 0.1
g/L suggesting low risk for GW contamination (Figure 10).

1.6

GW PECs propanil (fg L-1)

1.4 1.2 1.0 0.8 0.6 -1

0.4

0.32 fg L

0.2 0 1

2

3

4

5

6

7

8

9

10 11 12

13 14 15 16 17 18 19 20

Simulation years

Fig. 10. Annual average GW PECs of propanil for a 20-year simulation period. The shaded bar represent the 80th percentile value (adapted from Karpouzas et al., 2006).

154

D. G. Karpouzas and Z. Miao

SW risk assessment and mitigation effects. The 90th percentile value of all the yearly maximum daily PECs derived at specific points (1, 2, 3, 4 and 5 km) along the length of each of the 10 drainage canals for all simulation years was used as relevant PEC for risk assessment. Similarly, the 90th percentile value of all the yearly maximum daily PECs obtained at all junction points between drainage canals and river and also at specific points downstream of the last junction (7, 8 and 9 km) were used as relevant PECs for the calculation of acute toxicity exposure ratios (TERacute) for aquatic organisms. Simulation of this rice-cultivated basin scenario predicted higher SW PECs for molinate than for propanil in the simulated drainage canals and river (Table 7). Application of longer closure periods after application of molinate as a possible mitigation strategy resulted in significantly lower SW PECs in the simulated drainage canals and river (Table 7). The longer period of paddy closure after application of molinate, allows longer time for pesticide dissipation and diminishes the likelihood for the contamination of adjacent SW systems with high pesticide concentrations. Therefore, the duration of paddy closure after pesticide application should be considered in relation to pesticide persistence and efficacy. A longer paddy closure should be used as an effective mitigation strategy after the application of rather persistent pesticides. This would allow the dissipation of pesticide from paddy water during the closure time and thus diminish the mass of pesticide released to drainage canals at the end of the closure period. A similar mitigation strategy was recommended to be used in Japan after monitoring results and modeling evidences. It was suggested that the duration of paddy closure after application of pesticides in rice paddies should be pesticide-specific and determined according to pesticide persistence and particularly their DT90 values (Watanabe et al., 2005). The aquatic risk assessment suggested that propanil poses low risk for aquatic organisms in both drainage canals and river of the simulated basin. The aquatic risk assessment suggested that molinate may pose a risk for certain aquatic organisms, including Daphnia magna, alga, fish and Gamarus species in drainage canals. However, it should clarified that the drainage canals in river basins are not Table 7. Maximum daily PECs (
g/L) and measured concentrations of propanil and molinate in drainage canals and river Statistical Indices

Median 90th Percentile 95th Percentile

Propanil Canals

1.171 1.960 3.120

River

0.164 1.149 (1.0) 2.464

Molinate 5 Days closure time

10 Days closure time

20 Days closure time

Canals

River

Canals

River

Canals River

8.42 26.7

0.161 2.358

4.40 11.6

0.073 1.760

1.26 6.07

37.4

3.049

19.7

2.270

9.56

0.024 0.394 (0.3) 1.159

Note: Values in brackets represent yearly maximum concentrations of propanil and molinate measured in the Axios river in 1994.

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

155

permanent SW bodies, which only occasionally maintain a low water level during winter, when they are not operative. Therefore, considering the risk for fish in drainage canals might not be relevant, especially at higher tier analysis, where the relevance of each organism should be considered on a case-by-case basis. Molinate appears to pose low risk for any aquatic organism in the river when the closure time after application of molinate was higher than 10 days. Scenario validation. An indispensable part of the scenario development is its validation against measured data in order to verify that the developed scenario represents as much as possible realistic situations and lead, when modeled, to the calculation of PECs which are within the range found in environmental samples. The validation of the basin scale scenario developed for the rice-cultivated basin in Greece was validated by comparison of the 90th percentile of the yearly maximum daily PECs for propanil and molinate at point 9 km along the length of the simulated river with the maximum yearly concentration of the two herbicides at the most downstream point of the Axios river (Papadopoulou-Mourkidou et al., 2004). A good agreement between measured and predicted concentrations of propanil was evident in the simulated river (Karpouzas et al., 2006b). However, the predicted concentrations of molinate were markedly higher from the measured values for the 5 and 10-day closure period but were in good agreement with the measured values when a closure period of 20 days was applied for model parameterization (Table 7). In Greece, a closure period of 5–20 days is recommended after application of molinate depending on the visual efficacy of the herbicide against weeds. In the application of the developed scenario for molinate and propanil all simulated paddies were annually treated with molinate and propanil, which is a rather conservative assumption. A recent survey employed in the Axios river basin showed that more than 70% of local rice farmers utilize molinate and/or propanil alone or as a tank mixture with other herbicides (pretilachlor, bentazone) as a standard herbicide treatment practice (DowHelanco, personal communication, 2002). However, the measured data used for validation purposes had been obtained from a monitoring study employed in the period 1993–1994 when probably the entire rice cultivated area of the Axios river basin was treated with propanil and molinate (Papadopoulou-Mourkidou et al., 2004). In the developed scenario, the surface area of the simulated basin, which will be hypothetically treated with a studied pesticide, could be adjusted according to local information on the acreage treated with the tested pesticide. Therefore, for herbicides like azimuslfuron, pretilachlor or cyhalofop-butyl which are used less frequently in rice fields in Greece, the percentage of paddy area, treated with these pesticides should be adjusted to lower values accordingly. A more detailed description of the scenario development and validation is provided elsewhere (Karpouzas et al., 2006b). 3.3.2. Basin-scale scenario for Italy

Warren et al. (2004) have developed a similar watershed level scenario to simulate the environmental fate of bensulfuron-methyl, a rice herbicide, in the SW systems of a rice-cultivated basin in northern Italy. The modeled system comprised

156

D. G. Karpouzas and Z. Miao

a 2000 ha watershed which was divided into 10 management blocks of 200 ha each. Paddies (2 ha) were draining into ditches, then to canals and finally to a large river system. Timing of key agronomic and water management practices were identical for each management block relative to the day of seeding which was randomly varied among the blocks over a 3-week period from late April to mid May. Five meteorological datasets from France, Spain, Italy, Greece and Portugal were used for model simulations. Bensulfuron-methyl application occurred at all paddies 30 DAS with a foliar interception value of 50%. A total of 25% of the watershed received treatment which is consistent with the market distribution of the product. Spray drift of 2.77% to ditches adjacent to treated paddies was assumed at the time of application. Degradation and adsorption values for model parameterization were obtained from field dissipation studies (Molinari et al., 1999). Soil and hydrology of the modeled system were based on the clay scenario from Med-Rice (2003) and also guidance from local experts. PECsw and PECsed in paddies were generated with RICEWQ. PECsw and PECsed in receiving ditches, canals and river were generated using RIVWQ. Pesticide risk assessment and scenario validation. Daily PECs from one treated field, one ditch node, one canal node in each management block and daily PECs from the river node at the exit of the watershed were summarized and used to calculate indices for risk assessment. The model outputs were summarized by calculating the average of the maximum daily PECsw and PECsed from each simulation year, for each sample point (10 paddies, ditch, canal, river) and for each meteorological scenario (5 datasets). This average value provided a conservative PECsw and PECsed which could be used in risk assessment for TERs calculation. The magnitude of PECsw was directly correlated with the average monthly rainfall, with higher PECsw in years with high precipitation from April to June. This was also evident with monitoring data suggesting that the modeled system provides a close representation of the real situation. Another risk assessment indicator used was the number of days in each simulation year that the PECsw exceeded selected acute toxicological endpoint for aquatics. The overall average exceedence time was 9 days for paddies, 3 days for ditches and 1 day in the canals whereas no exceedences were observed in the river system. A good agreement between predicted and observed concentrations of bensulfuron-methyl in paddy fields and also in the exit of the receiving river were evident providing a further confirmation that the developed scenario offers a good representation of bensulfuron-methyl fate in rice cultivated basins at European conditions. 3.4. Landscape risk assessment in rice paddy areas based on geographical information systems (GIS) Pesticide risk assessment is mainly based on conservative approaches which include numerous assumptions with the aim to provide a realistic worst case scenario for the prediction of exposure (Med-Rice, 2003; FOCUS, 2000, 2001). However, in cases where a potential risk for a pesticide has been identified at the

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

157

tier 1–3 level, the registrants should either provide refined modeling results or suggest effective mitigation practices which should be backed up by supporting studies, thus, moving to tier-4 analysis. Refined modeling is usually identified as a mean of representing as much as possible the real agricultural situation by gradually removing conservative assumptions included in previous risk assessment steps. This should be supported by evidence showing that these assumptions do not really represent the real world and overestimate the risk for non-target organisms. A common method for establishing tier-4 analysis is to include landscape considerations from areas relevant to the uses of pesticide under question. Such landscape information is currently provided by remote sensing (RS) and GIS. Only recently, a new group called FOCUS working group on landscape and mitigation factors in ecological risk assessment completed its report which included a summary of the existing tools on landscape risk analysis for pesticides and a preliminary guide on what mitigation strategies are available and how they can be implemented (FOCUS, 2007). GIS tools can use multiple datasets such as slope, soil and hydrology parameters and land cover and analyze them in such a manner as to evaluate their co-occurrence (Kay, 1998). Integration of GIS as a pre- or post-processor of data for environmental models provide a great variety of options for higher tier risk assessment. Recently, great progress has been made in predicting spatial and temporal concentrations of pesticides at the watershed level using mathematical modeling in combination with GIS in almost all crop situations (Verro et al., 2002; Padovani et al., 2004). However, no attempts have been made so far in Europe regarding the combined use of GIS and mathematical modeling in rice paddy areas. As previously mentioned, rice is cultivated at large basins where, paddy fields, artificial and natural SW bodies create a unique ecosystem which should be studied as a whole. Therefore, rice cultivated basins offer an ideal example where landscape risk assessment including mathematical modeling and GIS information could be combined in order to provide realistic higher tier risk analysis. Application of landscape risk analysis for pesticides applied in rice paddies includes parameterization of RICEWQ and RIVWQ models using GIS datasets from rice cultivated zones, which are considered representative of wider rice cultivated areas. Until now, the possibility of using landscape information from GIS systems to simulate the fate of rice pesticides at watershed-scale level has not been explored due to several other requirements. A detailed scouting of the studied area at the period of rice cultivation, on-site monitoring of the dimensions, shape and water flow of the receiving canals, information on annual pesticide use and pesticide monitoring data including data collected from the most upstream point of the SW body simulated are certain prerequisites which are needed before risk analysis at landscape level should be performed in rice cultivated zones. 3.4.1. Landscape exposure assessment in a rice cultivated watershed in Lombardy

Recently, a risk assessment study in the Region of Lombardy, northern Italy developed a landscape approach to predict pesticide exposure in target SW bodies associated with rice fields of the area (Figure 11a). The area selected was a small

158

(e) (d) (a) Paddy Pesticide Treatments Molinate Propanil Propanil & Molinate Te )

nte

rre

To

io(

Loamy Sand Sandy

Untreated

pp

Loam

rdo

Soil Texture

Sandy Loam

(c)

Meteo Station

X

R

river terdoppio streams drainage canals

Fig. 11. A map of Lombardy (a), the main rice cultivated zone in Lombardy, the province of Pavia and the community of Tromello (b), the network of receiving surface water systems of the simulated watershed in Tromello (c), the soil texture of the rice cultivated land in the community of Tromello (d) and the simulated rice-cultivated watershed with the pesticide treatments assigned to each digitized paddy (e). (See Colour Plate Section, page 254.)

D. G. Karpouzas and Z. Miao

(b)

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

159

rice-watershed of 467 ha, comprising 200 rice paddies, which was located in the community of Tromello, at the center of the province of Pavia which is the main rice cultivated area of Lombardy (Figure 11b). The community of Tromello is intercrossed by the river Terdoppio whose total length in Lombardy is 60 km. However, pesticide exposure was estimated only for the part of the river Terdoppio which passes through the community of Tromello comprising a length of 5.3 km. A series of drainage canals, which receive outflow water and drift from rice paddies and discharge their water into streams, which in turn discharge their water into river Terdoppio constitute a network of SW bodies which were selected as targets for predicting pesticide exposure (Figure 11c). Spatial datasets at the scale of 1:50,000 for the community of Tromello were provided by Ente Regionale per i Servizi all’ Agricoltura e alle Foreste (ERSAF) Lombardy and included land use, soil physicochemical properties, hydrology and SW topology along with aerial photographs and a technical card of the area (Figure 11d). Other data needed for model parameterization including SW body dimensions, water flow velocity, discharge, flow direction and shape of cross section were all derived by on-site observations. Information on the common agronomic practices applied in rice cultivation in the area including seeding, harvest and maturation time, irrigation and drainage schedule were collected by local agronomic experts. Information on the type of pesticide used in the paddies of the simulated watershed were gathered after personal interviews with the rice farmers of the area and discussion with local experts (Figure 11e, Table 8). Exposure in the watershed was calculated for two rice herbicides, molinate and propanil, which were commonly used in the simulated watershed during the study period. Pesticide properties, degradation and adsorption data required for model parameterization were derived from literature. Soil and hydrological data also required for the parameterization of RICEWQ but also for the parameterization of the VADOFT sub-model were derived from the relevant spatial datasets. A reason for the selection of the specific water body and watershed for the study was the availability of monitoring data from specific points along the route of the river Terdoppio upstream and downstream of the community of Tromello allowing the validation of the simulation results. Monitoring data were provided by the analytical laboratory of the Province of Pavia which perform regular monitoring studies in the primary SW bodies of the province. Meteorological data for Table 8. Pesticide application scheme in the paddies within the simulated watershed Treated area (ha)

No. of paddies

Mode of application

No. of applications

Molinate

73

31

1 per year (4 kg/ha)

Propanil

243.4

85

Granular application on flooded fields Sprayed on drained fields

Treated with other herbicides

150.6

85

2 per year with a 7 day-interval (4.5 kg/ha)

160

D. G. Karpouzas and Z. Miao

performing the simulations were obtained from a local meteorological station for a 10-year period (1991–2000). A modeling approach similar to the one reported in Section 3.2 was followed with the combined use of the models RICEWQ and RIVWQ. The fate of propanil and molinate in the selected paddies was simulated using RICEWQ which was parameterized using all the information described above. RICEWQ simulations produced PECs for the two pesticides at the depth of 2 m, which was considered as relevant depth for GW risk assessment. The annual average PEC for the year 1999 (the reference year for the spatial data) was used for assessing the risk for GW contamination. In addition, RICEWQ simulations produced pesticide mass and paddy water loadings as daily outflow from paddies. These loadings were used as pesticide mass and water inputs for the receiving drainage canals and streams. In addition, a drift loading of 2.77% of the application rate of propanil was considered for drainage canals adjacent to propanil-treated paddies. The subsequent fate of the pesticide into the network of drainage canals, streams and the river Terdoppio was simulated by RIVWQ. PECsw at a selected point along the route of the river Terdoppio, downstream of the community of Tromello, were compared with measured concentrations of propanil and molinate obtained from a sampling point close to the simulated point selected. GW PECs for propanil suggested low risk for GW contamination unlike molinate whose PECs exceeded 0.1
g/L in 7 of the 31 rice paddies treated with this herbicide. Comparison of PECsw for the year 1999 at a selected point in the river Terdoppio with measured values for molinate indicated a relatively good agreement although the peak predicted concentrations were higher than the measured value (Figure 12a). Similarly, comparison of measured with predicted values for propanil indicated a good agreement although the measured concentrations of propanil exceeded the limit of detection only once in 1999 (Figure 12b). These results suggest that landscape analysis of pesticide exposure in rice cultivated watersheds using GIS tools in combination with mathematical modeling could provide a realistic prediction of exposure at spatial scale. This will be a pioneering study for the future incorporation of landscape risk assessment as an acceptable method for refining risk in rice paddy areas. 4. CONCLUDING REMARKS AND THE WAY FORWARD Methodologies and implementation of higher tier risk assessment of pesticides has rapidly improved over the last few years. The outcome of the various FOCUS groups have provided clear guidelines and modeling tools which are relatively easy to use even for people with no extensive modeling experience (FOCUS, 2000, 2001). A similar initiative called Med-Rice produced general guidelines for how risk assessment should be performed at tier 1 using simple spreadsheets (Med-Rice, 2003). At the moment, registration of rice pesticides is based on Med-Rice tier-1 scenarios and only limited attention has been given to the possibility of future requirements for the development of guidelines and modeling tools for higher tier risk assessment.

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

161

0.8 (a)

Measured Simulated

0.6

0.4

Pesticide Concentrations in water (µg/l)

0.2

0 0

50

100

150

200

250

300

350

0.6 (b) Simulated 0.4 Measured

0.2

0.0 0

50

100

150

200

250

300

350

Julian days

Fig. 12. Simulated and measured concentrations of molinate (a) and propanil (b) in a selected point along the simulated length of the river Terdoppio.

The progress made so far towards the development of a higher tier assessment approach in rice paddies is attributed to isolated research efforts and not to concerted actions. These efforts have shown that RICEWQ is a well-validated model which can handle complex situation and the different pesticide and water management practices applied in rice cultivation in Europe. In addition, it can be easily parameterized and used within the context of member-state scenarios for rice cultivation. Therefore, RICEWQ could be considered as a well-developed modeling tool for higher tier risk assessment in rice paddies. Other higher tier models specific for rice cultivation like PCPF-1 or PADDY should be validated under European conditions before considered for use in the regulatory scheme. When further modeling refinement is needed the combined use of RICEWQ with the SW model RIVWQ could be utilized for calculating pesticide exposure at receiving SW bodies. The development and validation of representative basin-scale scenarios for

162

D. G. Karpouzas and Z. Miao

the main rice-cultivation areas in Greece and Italy constitutes a preliminary step towards the establishment of a more thorough pesticide risk-assessment at the basinscale level. In accordance with the progress made for all the other crops first steps have been made for the implementation of GIS tools in regulatory modeling for rice. Future research in the area of higher tier risk assessment in rice paddy areas should focus on the following areas: ● ●

●

●

●

Development of guidance on how risk assessment should be performed at tier 3. Preparation of user-friendly versions of validated higher tier models like RICEWQ and preparation of guidelines on how to parameterize these models in accordance with FOCUS GW and SW. Validation of other higher tier models as candidates for implementation in higher tier risk assessment scheme. The Japanese PCPF-1 model is a userfriendly model which could be a possible candidate if and when certain routines will be modified to better describe rice cultivation practices in Europe. Co-operation between scientists from rice-cultivating countries in Europe with the aim to develop new rice scenarios at member-state level since up to now such scenarios are only available for Italy and Greece. Further development of more advanced methods for the combined use of GIS with models like RICEWQ and SW models like RIVWQ or TOXSWA for calculating realistic pesticide exposure levels in rice cultivated watersheds.

ACKNOWLEDGMENTS This work was produced within the framework of a Marie Curie individual fellowship “Environmental risk analysis leading to simulate a sustainable ecosystem management in rice area” held by Dr. D. Karpouzas (QLK5-CT-200251598). The authors would like to thank Mr. C. Riparbelli and Mr. M. Pastori from ERSAF Lombardia for supplying us with the spatial data of the region of Lombardy. REFERENCES Batista, S., Silva, E., Galhardo, S., Viana, P. and Cerejeira, M. J. (2002). Evaluation of pesticide contamination of ground water in two agricultural areas of Portugal. Int. J. Environ. Anal. Chem. 82, 601–609. Beulke, S. and Brown, C. D. (2001). Evaluation of methods to derive pesticide degradation parameters for regulatory modeling. Biol. Fertil. Soils 33, 558–564. Capri, E. and Miao, Z. (2002). Modelling pesticide fate in rice paddy. Agronomie 22, 363–371. Carsel, R. F., Imhoff, J. C., Hummel, P. R., Cheplick, J. M. and Donigian Jr., A. S. (1998). PRZM – 3, a model for predicting pesticide and nitrogen fate in the crop root and unsaturated soil zones: Users Manual for release 3.0. National Exposure Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Athens, GA, USA. Cerejeira, M. J., Viana, P., Batista, S., Pereira, T., Silva, E., Valerio, M. J., Silva, A., Fereira, M. and Silva-Fernandes, A. M. (2003). Pesticides in Portuguese surface and ground waters. Water Res. 37, 1055–1063. Cervelli, S., Cervelli, F. and Cervelli, L. (2004). A Step 2 approach to compute water, soil and sediment PECs and TWAs for the inclusion of rice pesticides in Annex I of the EU Council Directive

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

163

91/414/EEC. In: A. Ferrero, F. Vidotto (Eds), Proc. of the Conference challenges and opportunities for sustainable rice-based production systems, Edizioni Mercuri, Torino, Italy, pp. 431–443. Charizopoulos, E. and Papadopoulou-Mourkidou, E. (1999). Occurrence of pesticides in rain of the Axios River Basin. Environ. Sci. Technol. 33, 2363–2368. Crepeau, K. L. and Kuivila, K. M. (2000). Rice pesticide contamination in the Colusa basin drain and the Sacramento river, California, 1990–1993. J. Environ. Qual. 29, 926–935. Eleftherohorinos, I. G. and Dhima, K. V. (2002). Red rice (Oryza sativa) control in rice (O. sativa) with pre-emergence and post-emergence herbicides.Weed Technol. 16, 537–540. Fajardo, F., Takagi, K. and Usui, K. (2000). Dissipation of mefenacet and pretilachlor in paddy soils under laboratory oxidative and reductive conditions. J. Weed Sci. Techol. 45, 250–253. Ferrari, F., Karpouzas, D. G., Trevisan, M. and Capri, E. (2005). Measuring and predicting environmental concentrations of pesticides in air after application to paddy water systems. Environ. Sci. Technol. 39, 2968–2975. Ferrero, A., Vidotto, F., Gennari, M. and Negre, M. (2001). Behavior of cinosulfuron in paddy surface waters, sediments and ground water. J. Environ. Qual. 30, 131–140. FOCUS. (2000). Groundwater scenarios in the EU plant protection product review process. Report of the FOCUS Groundwater Scenarios Workgroup, European Commission Document Reference Sanco 321, Brussels, Belgium. FOCUS. (2001). Surface water scenarios in the EU evaluation process under 91/414/EEC. Report of the FOCUS Working Group on surface water scenarios, European Commission Document Reference SANCO/4802/2001, Brussels, Belgium. FOCUS. (2007). Landscape and mitigation factors in aquatic risk assessment. Volume 1. Extended summary and recommendations. Report of the FOCUS Working Group on Landscape and Mitigation Factors in Ecological Risk Assessment, EC. Hassink, J., Guth, J. A., Reischmann, F. J., Allen, R., Arnold, D., Leake, C. R., Skidmore, M. and Reeves, G. L. (2003). Vapour pressure and volatile losses of plant protection products from plants and soils. In: A. A. M. Del Re, E. Capri, L. Padovani and M. Trevisan (Eds), Proc. of the XII Symposium Pesticide Chemistry, Piacenza, Italy, pp. 359–366. Inao, K. (2003). Development of a simulation model (PADDY) for predicting pesticide behavior in rice paddy fields. J. Pestic. Sci. 28, 322–323. Inao, K. and Kitamura, Y. (1999). Pesticide paddy field model (PADDY) for predicting pesticide concentrations in water and soil in paddy fields. Pestic. Sci. 55, 38–46. Jackson, R. and Douglas, M. (1999). An aquatic risk assessment for cyhalofop-butyl: A new herbicide for control of barnyard grass in rice. In: A. A. M. Del Re, C. Brown, E. Capri, G. Errera, S. P. Evans and M. Trevisan (Eds), Proc. of the XII Symposium of Pesticide Chemistry, Piacenza, Italy, pp. 345–354. Karpouzas, D. G. and Capri, E. (2004). Higher tier risk assessment for pesticides applied in rice paddies: Filling the gap at European level. Outl. Pest Manage. 15, 36–41. Karpouzas, D. G. and Capri, E. (2006). Risk analysis of pesticides applied in rice paddies using RICEWQ 1.6.2v and RIVWQ 2.02 models. Paddy Water Environ. 4, 29–38. Karpouzas, D. G., Capri, E. and Papadopoulou-Mourkidou, E. (2005a). Application of the RICEWQ-VADOFT model to simulate leaching of propanil in rice paddies in Greece. Agron. Sustain. Dev. 25, 35–44. Karpouzas, D. G., Ferrero, A., Vidotto, F. and Capri, E. (2005b). Application of the RICEWQVADOFT model for simulating the environmental fate of pretilachlor in rice paddies. Environ. Toxicol. Chem. 24, 1007–1017. Karpouzas, D. G., Capri, E. and Papadopoulou-Mourkidou, E. (2006b). Basin-scale risk assessment in rice paddies: An example based on the Axios river basin in Greece. J. Vadose Zone 5, 273–282. Karpouzas, D. G., Cervelli, S., Watanabe, H., Capri, E. and Ferrero, A. (2006a). Pesticide exposure assessment in rice paddies in Europe: A comparative study of existing mathematical models, Pest Manage. Sci. 62, 624–636. Kay, S. A. (1998). Use of GIS and Remote Sensing in agrochemical regulatory strategies. CSI Visions 1. Kubiak, R., Burkle, W. L., Cousins, I., Hourdakis, A., Jarvis, T., Jene, B., Koch, W., Kreuger, J., Maier, W. M., Millet, M., Reinert, W., Sweeny, P., Tournayre, J. C. and Van de Berg, F. (2003). FOCUS-Air: Remits and first results. In: A. A. M. Del Re, E. Capri, L. Padovani and M. Trevisan (Eds), Proc. of the XII Symposium Pesticide Chemistry, Piacenza, Italy, pp. 473–485.

164

D. G. Karpouzas and Z. Miao

Med-Rice. (2003). Final report of the working group MED-RICE prepared for the European Commission in the framework of Council Directive 91/414/EEC. European Commission Document Reference Sanco 1092, Brussels, Belgium. Miao, Z., Cheplick, J. M., Williams, W. M., Trevisan, M., Padovani, L., Gennari, M., Ferrero, A., Vidotto, F. and Capri, E. (2003a). Simulating pesticide leaching and runoff in rice paddies with RICEWQ-VADOFT model. J. Environ. Qual. 32, 2189–2199. Miao, Z., Padovani, L., Riparbelli, C., Ritter, A. M., Trevisan, M. and Capri, E. (2003b). Prediction of the environmental concentration of pesticide in paddy field and surrounding surface water bodies. Paddy Water Environ. 1, 121–132. Miao, Z., Trevisan, M., Capri, E., Padovani, L. and Del Re, A. A. M. (2003c) An uncertainty assessment of the RICEWQ model. In: A. A. M. Del Re, E. Capri, L. Padovani and M. Trevisan (Eds), Proc. of the XII Symposium Pesticide Chemistry, Piacenza, Italy, pp. 545–554. Molinari, G. P., Cavanna, S., Bonacini, F., Giammarrusti, L. and Barefoot, A. C. (1999). Bensulfuron-methyl and metsulfuron-methyl dissipation in water and soil of rice fields. In: A. A. M. Del Re., C. Brown, E. Capri, G. Errera, S. P. Evans and M. Trevisan (Eds), Proc. of the XI Symposium Pesticide Chemistry, Cremona, Italy, pp. 45–50. Ntanos, D. (2001). Strategies for rice production and research in Greece. In: J. Chataigner (Ed), Research strategies for rice development in transition economies, CIHEAM-IAMM, Montpellier, France, pp. 115–122. Ntanos, D. A. (1997). Rice production and research in Greece. In: J. Chataigner (Ed), Mediterranean rice research activities, CIHEAM-IAMM, Montpellier, France, pp. 127–153. Ntanos, D.A., Koutroubas, D. and Mavrotas, C. (2000). Barnyardgrass (Echinochloa cruss-galli) control in water-seeded rice (Oryza sativa) with cyhalofop-butyl. Weed Technol. 14, 383–388. Padovani, L., Capri, E. and Trevisan, M. (2004). Landscape-level approach to assess aquatic exposure via spray drift for pesticides: A case study in a Mediterranean area. Environ. Sci. Technol. 38, 3239–3246. Papadopoulou-Mourkidou, E., Karpouzas, D. G., Patsias, J., Kotopoulou, A., Milothridou, A., Kintzikoglou, K. and Vlachou, P. (2004). The potential of pesticides to contaminate the groundwater resources of the Axios river basin. Part II. Monitoring study in the south part of the basin. Sci. Total Environ. 321, 147–164. Pothuluri, J. V., Hinson, J. A., Cerniglia, C. E. (1991). Propanil: Toxicological characteristics, metabolism and biodegradation potential in soil. J. Environ. Qual. 20, 330–347. Ramos, C., Carbonell, G., Garcia Baudin, J. M. and Tarazona, J. V. (2000). Ecological risk assessment of pesticides in the Mediterranean region. The need for crop-specific scenarios. Sci. Total Environ. 247, 269–278. Readman, J. W., Albanis, T. A., Barcelo, D., Galassi, S., Tronczynski, J. and Gabrielides, G. P. (1993). Herbicide contamination of Mediterranean estuarine waters: Results from a MED POL pilot survey. Mar. Pollut. Bull. 26, 613–619. Redolfi, E., Azimonti, G., Auteri, D. and Cervelli, S. (2004). Validation of a new model (SWAGW) to estimate PECs in paddy fields. In: A. Ferrero and F. Vidotto (Eds), Proc. of the Conference challenges and opportunities for sustainable rice-based production systems, Edizioni Mercuri, Torino, Italy, pp. 485–489. Ross, L. J. and Sava, R. J. J. (1986). Fate of thiobencarb and molinate in rice fields. J. Environ. Qual. 15, 220–225. Seiber, J. N., McChesney, M. M., Sanders, P. F. and Woodrow, J. E. (1986). Models for assessing the volatilization of herbicides applied to flooded rice fields. Chemosphere 25, 127–138. Takagi, K., Fajardo, F. F., Inao, K. and Kitamura, Y. (1998). Predicting pesticide behavior in a lowland environment using computer simulation. Rev. Toxicol. 2, 269–286. Tarazona, C., Carrasco, J. M. and Sabater, C. (2003). Monitoring of rice pesticides in an aquatic system of natural park of Albufera, Valencia, Spain. Hazard evaluation. In: A. A. M. Del Re, E. Capri, L. Padovani and M. Trevisan (Eds), Proc. of the XII Symposium of Pesticide Chemistry, Piacenza, Italy, pp. 727–736. Tournebize, J., Watanabe, H., Takagi, K. and Nishimura, T. (2004). New coupled model of pesticide fate and transport in paddy field. In: A. Ferrero and F. Vidotto (Eds), Proc. of the Conference challenges and opportunities for sustainable rice-based production systems, Edizioni Mercurio, Torino, Italy, pp. 497–506.

Higher Tier Exposure Assessment in Rice Paddy Areas: A European Perspective

165

Tournebize, J., Watanabe, H., Takagi, K. and Nishimura, T. (2006). The development of a coupled model (PCPF-SWMS) to simulate water flow and pollutant transport in Japanese paddy fields. Paddy Water Environ. 4, 39–51. Verro, R., Galliera, M., Maffioli, G., Auteri, D., Sala, S., Finizio, A. and Vighi, M. (2002). GISBased system of surface water risk assessment of agricultural chemicals. 1. Methodological approach. Environ. Sci. Technol. 36, 1532–1538. Vidotto, F., Ferrero, A., Bertoia, O., Gennari, M. and Alessandro, C. (2004). Dissipation of pretilachlor in paddy surface water and sediment. Agronomie 24, 473–479. Vu, S. H., Ishihara, S., Watanabe, H., Ueji, M. and Tanaka, H. (2004). Monitoring pesticide fate and transport in surface water in Japanese paddy field watershed. In: A. Ferrero and F. Vidotto (Eds), Proc. of the Conference challenges and opportunities for sustainable rice-based production systems, Edizioni Mercuri, Torino, Italy, pp. 509–521. Warren, R. L., Ritter, A. M. and Williams, W. M. (2004). A rice herbicide Tier 2 exposure assessment for European rivers based on RICEWQ/RIVWQ. In: A. Ferrero and F. Vidotto (Eds), Proc. of the Conference challenges and opportunities for sustainable rice-based production systems, Edizioni Mercuri, Torino, Italy, pp. 523–533. Watanabe, H. and Takagi, K. (2000a). A simulation model for predicting pesticide concentrations in paddy water and surface soil. I. Model development. Environ. Technol. 21, 1379–1391. Watanabe, H. and Takagi, K. (2000b). A simulation model for predicting pesticide concentrations in paddy water and surface soil II. Model validation and application. Environ. Technol. 21, 1393–1404. Watanabe, H., Takagi, K. and Vu, S. H. (2006). Simulation of mefenacet concentration in paddy field by improved PCPF-1 model. Pest Manage. Sci. 62, 20–29. Watanabe, H., Vu, S. H., Tournebize, J., Nguyen, M. H. T., Komany, S., Phong, T. K., Ishihara, S. and Takagi, K. (2005). Monitoring and modeling of pesticide fate and transport in paddy fields; challenges for reducing environmental risk. In: Proc. of the 2nd International Conference of Japan Korea Research Cooperation, Impact assessment of farm chemicals runoff from paddy field and biodiversity conservation, Tsukuba, Japan, pp. 69–82. Williams, W. M., Ritter, A. M., Cheplick, J. M. and Zdinak, C. E. (1999). RICEWQ: Pesticide runoff model for rice crops – user’s manual and program documents version 1.6.1. Waterborne Environment Inc., S.E. Leesburg, VA. Williams, W. M., Zdinak, C. E., Ritter, A. M., Cheplick, J. M. and Singh, P. (2004). RIVWQ: Chemical transport model for riverine environments – user’s manual and program documentation version 2.02. Waterborne Environment Inc., S.E. Leesburg, VA.

'" CJ'1

.,I:::a.

(d)

()

o

5" c..., "U

Plate 2. A map of Lombardy (a), the main rice cul!ivated zone in Lombardy, the province of Pavia and the community of Tromello (b), the network of receiving surface water systems of the simulated watershed in Tromello (c), the soil texture of the rice cultivated land in the community of Tromello (d) and the simulated rice-cultivated watershed with the pesticide treatments assigned to each digitized paddy (e). (See also page 158 of this book.)

or CD

(j)

CD

Q.

0" :::J

Pesticide Risk Assessment in Rice Paddies: Theory and Practice E. Capri and D.G. Karpouzas (editors) © 2008 Elsevier B.V. All rights reserved

Chapter 8

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective H. Watanabe,1 K. Inao,2 S. H. Vu,1 T. K. Phong,1 S. Ishihara,2 K. Takagi2 and J. Tournebize3 1Tokyo

University of Agriculture and Technology, 3-5-8, Saiwaicho, Fuchu, Tokyo 183-8509, Japan 2National Institute of Agro-Environmental Science, 3-1-3 Kannondai, Tsukuba, Ibaraki, Japan 305-8604 3Cemagref, Parc de Tourvoie, BP 44, 92163 Antony Cedex, France Contents 1. Introduction 2. Pesticide use in Japan 3. Pesticide regulation in Japan 4. Tier system approach in the new pesticide registration scheme 4.1. Rationale 4.2. General procedure 4.3. PECs determination 4.3.1. Basic concept 4.3.2. Environmental model and standard scenarios used for PECs calculation 4.3.3. Calculation of PECs for tier 1 4.3.4. Determination of PECs for tier 2 4.3.5. Determination of PECs for tier 3 4.4. Determination of acute effect concentration (AEC) 5. Monitoring pesticide risk in monsoon Asian climate 5.1. Monitoring the fate and transport of rice pesticides in the paddy plots 5.1.1. Monitoring setup and water balance in the paddy plots 5.1.2. Monitoring pesticide concentration in paddy water and soil 5.2. Monitoring the fate and transport of rice pesticides using lysimeters 5.2.1. An experimental facility for studying the fate and transport of agrochemicals 5.2.2. Application of lysimeter experiments for developing a mitigation technique for controlling herbicide runoff from paddy field using rice husk charcoal powder 5.3. Monitoring the fate and transport of rice pesticides at a watershed scale 5.4. Monitoring pesticide concentrations in a river basin 6. Modeling the fate and transport of pesticides at paddy field and watershed scale 6.1. Simulation of pesticide fate and transport in the paddy rice environment using PADDY, PADDY-2 and PADDY-Large 6.1.1. PADDY model 6.1.2. PADDY-2 model 6.1.3. PADDY-Large model 6.2. Simulation of pesticide fate and transport in paddy fields by the PCPF-1, PCPF-SWMS, and PCPF-C 6.2.1. PCPF-1 model 6.2.2. PCPF-SWMS model 6.2.3. PCPF-C model

168 169 170 171 171 173 174 174 174 177 177 179 179 180 180 180 181 182 183 185 186 189 192 192 192 195 197 199 200 202 204

168

H. Watanabe et al.

7. Risk management for reducing pesticide losses into aquatic environments in Asian monsoon climate 7.1. EWSD and WHP 7.2. Good agricultural practices for reducing pesticide losses from the paddy fields 8. Summary and conclusions References

206 206 209 209 211

1. INTRODUCTION Located in the east end of the temperate Asian monsoon area, Japan has an annual precipitation of 1750 mm. Currently, agricultural land in Japan shares only about 13% of the total land area. Of the total agricultural land of 4.8 million hectares, paddy fields (including non-rice paddy) account for about 54%, followed by 25% of upland field, 13% of pasture, and 7% of orchards in 1997 (Hosoya, 2001). The rice cultivation in Japan has a history of more than 2400 years. After World War II, paddy rice production increased drastically in order to overcome the food crisis. Along with the rapid economic growth, the modern rice production has been promoted through the national land consolidation project that started in 1963 (Hasegawa and Tabuchi, 1995). This project has restructured paddy field, irrigation, and drainage facilities while rice cultivation including land preparation, transplanting, and harvesting has been mechanized in order to increase production efficiency. In 2000, 1.48 million hectares of rice paddies were cultivated by 1.74 million farmers equaling to a ratio of 0.85 ha/farmer (MAFF, 2005). In Japan, paddy fields are mainly composed of alluvial soils (Gray lowland soil and Gley soils). The typical water budget for well-managed Japanese paddy fields is as follows: the evapotranspiration (ET) ranges from 0 to about 8 mm day⫺1; the percolation varies widely from about 5 mm day⫺1 to 30 mm day⫺1, depending on soil type and its management; the average surface drainage, including runoff, can be about 35 mm day⫺1; and the average water requirement as irrigation can be about 15 mm day⫺1 (Mizutani, 1995). Almost all rice paddies in Japan are grown by transplanting of rice seedlings. The land preparation starts around April and June, and the harvest season lasts from August to October depending on the region and the rice varieties. Most of the rice paddies are treated with herbicides and other crop protection products, such as fungicides and insecticides that are applied during the crop season accordingly. Newly developed insecticides and fungicides are also applied during seedbed preparation. Since paddy field shares more than half of the agricultural land in Japan, the rice production seems to be the major non-point source of pesticide pollution. The typical paddy field in Japan is susceptible to herbicide runoff since the chemical is applied directly onto paddy water. Pesticide losses from paddy field range from a few percent to more than 50% of the applied amount depending on the water management (Maru, 1991; Sudo et al., 2002). Monitoring of pesticide concentrations in river systems in Japan detected a number of herbicides commonly used in paddy fields (Inoue et al., 2002; Okamura et al., 2002; Sudo et al., 2002; Nakano

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

169

et al., 2004), and these concentrations appear to have adverse effects on the aquatic ecosystem (Okamura et al., 2002). In 2005, the Ministry of Environment of Japan imposed a new pesticide registration program in order to meet the public expectation for high environmental quality as well as to meet the regulatory standards of EU and USA. In this chapter, an overview of the Japanese regulatory issues regarding pesticide use in rice paddies and an introduction of the new pesticide registration program are provided. In addition, the experience of the environmental monitoring of pesticides and the modeling approaches used for the calculation of predicted environmental concentrations (PECs) in surface water and ground water systems adjacent to rice paddies in Japan will be discussed. Then, the mitigation techniques that are pointed out in the monitoring and modeling studies will be summarized. The good agricultural practices (GAPs) for controlling pesticide losses from paddy fields will also be discussed with the objective of reducing and managing the pesticide risks associated with paddy rice production, especially in the Asian monsoon region. 2. PESTICIDE USE IN JAPAN Before World War II in Japan, pesticides originated mainly from natural substances such as (Dalmatian) pyrethrum and some inorganic substances. After the war, organically synthesized pesticides such as DDT, BHC, 2,4-D, and parathion were introduced in late 1940s to early 1950s and the domestic production of these active ingredients started. However, organochlorine insecticides including DDT, BHC were banned in 1969 because of their high bioaccumulation potential and subsequently the organophosphorus insecticides such as diazinon, ENP, PAP, and MEP were introduced in the market (Hosoya, 2001). The production of pesticides as active ingredients in Japan increased after the war until the mid 1980s and then the annual production has been maintained stable at about 90,000 tons (Figure 1). However, product shipment of pesticides peaked in the mid 1980s and decreased steadily thereafter. This mainly resulted from the reduction of the pesticide shipment for rice paddies, which decreased from 401,000 tons in 1984 to 152,000 tons in 1998 (Hosoya, 2001). This is probably attributable to the reduction of the number of pesticide applications, the development of lightweight granule and liquid formulations, and finally the reduction of the total paddy field area. The reduction in the number of pesticide applications was due to the recent development of pesticide products with multiple ingredients. Formulations of rice herbicides (one-shot or single-application herbicide) that usually contain three or more active ingredients can cover a wide spectrum of activity against various weeds so that farmers only need to apply them once in the growing season. This type of formulation was used in about 30% of the total treated paddy field area in 2000 (Takeshita and Noritake, 2001). While the number of registered pesticide products decreased since early 1990s the number of registered active ingredients gradually increased (Table 1) (Green Japan, 2005). However, the number of registered products is still well over 4000,

170

H. Watanabe et al.

Fig.1. Annual pesticide production (ton of active ingredients) in Japan (Hosoya, 2001).

Table 1. Number of registered pesticide products and active ingredients (Green Japan, 2005) Year

1992

1994

1996

1998

2000

2002

2004

Registered products Annual registration of new products Registered active ingredients Annual registration of new active ingredients Withdrawal of active ingredients

6037 271

5780 380

5434 287

5369 304

5309 226

5059 208

4781 256

451

465

493

520

531

553

544

8

15

25

26

15

11

9

8

11

5

7

12

12

21

and every year there are more than 200 new products registered. This is because in addition to the different combinations of active ingredients, many other formulations are available including solid type granules, liquid type emulsion concentrates and suspension concentrates, large packed herbicides which can be thrown into the paddy field without entering the plots, and so on (Takeshita and Noritake, 2001). Pesticide development and their use for laborsaving rice production in Japan were discussed in detail by Takeshita and Noritake (2001). 3. PESTICIDE REGULATION IN JAPAN In 1948, the Agricultural Chemicals Regulation Law was first issued and it was amended in 1963 in order to set the standards for agricultural chemicals registration and to regulate pesticides that are toxic for aquatic organisms (Hoshino,

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

171

2004). In general, the issues related to pesticides are handled by the Ministry of Agriculture, Forestry and Fishery (MAFF), the Ministry of Environment (MOE), and the Ministry of Health, Labor and Welfare (MHLW) according to the Agricultural Chemicals Regulation Law. For pesticide registration, the general procedure requires the submission of a dossier of the formulated product and the active ingredient to MAFF. The required test data in the dossier include data illustrating pesticide activity and the negative effects on crops, mammalian toxicity data (acute and chronic carcinogenicity, toxicity to carp daphnia and algae) environmental fate data in crop, soil, and water, and the toxicity data for non-target beneficial organisms such as silkworm, bees, and birds. According to the submitted data, the Food Safety Commission of the Cabinet Office, Japanese Government calculated the acceptable daily intake (ADI). The Pharmaceutical Affairs and Food Sanitation Council of the MHLW sets the Maximum Residue Limits (MRLs) of pesticides on food on the basis of individual ADI and pesticide residue data in crops. Meanwhile, the Central Environmental Council of the MOE sets the standards for granting or rejecting (withholding) the registration to a pesticide concerning its persistency in crops and soil, water pollution, and toxicity to aquatic organisms. Table 2 shows the standards established for the pesticide registration and regulation together with the responsible authorities (Hoshino, 2001). The registration of a product is valid for three years and the re-registration is required at the end of this period. 4. TIER SYSTEM APPROACH IN THE NEW PESTICIDE REGISTRATION SCHEME 4.1. Rationale The environmental risk associated with pesticide use was evaluated according to the standards for rejecting the pesticide registration (Table 2). Concerning toxicity to aquatic organisms, in the previous pesticide registration scheme, only acute toxicity for carp was considered. The registration of the pesticide was suspended when both conditions (1) and (2) below were met: 1. (a) In case the application rate of the active ingredient is lower than 1 kg ha⫺1; LC50 (48 h for carp) ⱕ 0.1 ppm (b) In case the application rate of the active ingredient is higher than 1 kg ha⫺1;

LC50 (48 h for carp in ppm) ⱕ 0.1 Applied mass(ai ) per 1 ha( kg) 2. The dissipation period that the concentration decreases below the mortality level for carp is longer than 7 days for the standard pesticide use.

172

H. Watanabe et al.

Table 2. Standard levels established for pesticide registration (Hoshino, 2001) Regulatory standards (The imposed ministry) Standards for the rejection (withholding) of the pesticide registration according to the Agricultural Chemicals Regulation Law (MOE)

Maximum Residue Limits (MRLs) in food according to the Food Sanitation Law (MHLW) Environmental quality standards according to the Basic Environmental Law (MOE)

Water quality advisory level (MOE)

Provisional guidelines for prevention of water pollution by agricultural chemicals used at golf links (MOE) Drinking water quality standard according to the Water Supply Law (MHLW)

Contents (1) Crop residue standard (Allowable concentration in crop for animals and human intake, same as MRLs) (2) Standard for persistence in soil (half-life >180 days) (3) Water quality standard (Allowable concentration in public water for affecting animals and humans) (4) Standard for toxicity to aquatic organisms (Minimum concentration for acute effect on fish, crustacean, and algae) Allowable pesticide concentration in food (1) Standard values for water pollution (Standard concentrations in public water concerning the human health) (2) Standard values for the soil contamination (3) Standard values for the ground water contamination Advisory concentration level for protecting water quality in public water Guideline targets which are the concentrations of 45 target pesticides in the drainage from golf links Allowable concentration in drinking water

This previous regulation scheme only evaluates the acute toxicity for carp. The acute toxicity value has been set as the hazard level for pesticide registration. However, the pesticide risk assessment is completed by comparing the hazard level with the exposure level or the PECs as in the case of EU and the US. Such a risk assessment scheme has never been applied before in Japan. In addition, this evaluation was required for rice pesticides but not for pesticides used in orchards or only upland fields. The basic plan with regard to the environmental conservation (authorized at the cabinet meeting on December 22, 2000) in the Basic Environmental Law considers the value of the ecosystem and sustainable human activity. In 2002, the MOE Working Group on ecotoxicological assessment of pesticides suggested

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

173

amendments on the previous pesticide registration scheme for evaluating the influence on the aquatic ecosystem (Hayakawa, 2005). In this section, the new pesticide registration scheme imposed in Japan from 2005 is presented. Important sections were extracted and translated from the report on the amendments of the standards for pesticide registration, which was published by the Japanese Plant Protection Association (JPPA, 2006) and MAFF (2006a). 4.2. General procedure The objective of the amendments was mainly to reduce the effects of pesticides on natural ecosystems such as the surface water systems. Therefore, the new registration scheme was specifically reviewed and amended from the following viewpoints: (a) Increase the type of organism used for risk assessment; (b) Adapt the evaluation method which compares toxicity with exposure; (c) Evaluate not only pesticides applied to rice but also all other pesticides used in orchards and upland fields. Figure 2 shows the new pesticide registration scheme in Japan. The scheme focuses only on the acute toxicity, and the test organisms used for the evaluation are considered as representative species of algae, crustacean, and fish. The evaluation is performed by comparing the acute effect concentration (AEC) of the pesticide with PECs calculated using environmental models for the Acute effect concentration (AEC)

Predicted environmental concentration (PEC) con The first stage (Tier1 PEC)

The acute toxicity test

Prediction by numerical computation PEC>AEC The second stage (Tier 2 PEC) * Rice pesticides: Water pollution assessment * Upland field pesticides: Run-off assessment Or drift assessment

The third stage (Tier 3 PEC)

* Fish toxicity test (AECf ) * Daphnia toxicity test (AECd) * Algae toxicity test (AECa)

No

Register

More realistic method of assessment from other countries may be applied

Ye Yes

* Rice pesticides: Field study of pesticide concentration in paddy water or drift monitoring

Reject registration

Fig. 2. General scheme for the new Japanese pesticide registration (JPPA, 2006).

174

H. Watanabe et al.

pesticide concerned. A tiered system is used for PEC computation for the efficient assessment procedure. Registration is suspended when PEC exceeds AEC value. However, even if the PEC value is lower than the AEC, based on the result of the risk assessment, it might be necessary to display the limitation of the application method and put them on the product label, as well as to implement an environmental monitoring for the pesticide under assessment. It is important to perform risk assessment even after registration is granted using environmental monitoring and then to employ studies for identifying risk management measures in case there is a risk. For currently registered pesticides, the same risk assessment should be performed but the computed PECs should be replaced by results obtained from monitoring studies in the public environment or the river basin in the area where the pesticide is used. 4.3. PECs determination 4.3.1. Basic concept

In this new registration scheme, the determination of PECs for rice pesticides in water comprises three stages (tiers): the first tier of preliminary calculation by numerical computation, the second tier using lysimeter tests, and third tier using plot-scale experiments for assessing pesticide runoff and drift (Nakamura, 2005; MAFF, 2006b). For upland field pesticides, risk assessment is composed of two stages: the first stage includes a preliminary calculation of PECs using numerical computation and the second stage with the use of data from pesticide runoff and drift experiments (Table 3). Table 3 presents the data source for PECs calculations for the new registration scheme. The calculation of PECs takes into account both runoff and drift. 4.3.2. Environmental model and standard scenarios used for PECs calculation

In Japan, most of the surface drainage of the cropland flows into public waters. Considering the geographical conditions in Japan, the environmental model was set to include both the agricultural field and the river as shown in Figure 3. The model watershed of 100 km2 includes the main river, tributaries, paddy fields, and upland fields. The ratios of the paddy fields and the upland fields over the watershed area were derived from the total crop area in Japan (500 ha for rice paddy and 750 ha for upland field). The prescribed river area was set to be 2.0 km2 considering that the area ratio of the river over the watershed was set to be equivalent to the ratio of the national landscape. The main river accounts for 60% and its tributaries for 40% of the total river area. The main river represents a class A river having discharges at the ordinary water level (50% frequency level) of 3.0 m3 s⫺1, at the low water level (75% frequency level) of 1.9 m3 s⫺1, and an average discharge of 5.0 m3 s⫺1. It is assumed that if a pesticide is being applied in one field, no simultaneous spraying of the same pesticide can be performed in other fields. The fraction of the area where the target pesticide is applied over the total area was set to be 10% for rice pesticides and 5% for upland field pesticides. Due to variations in the

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

175

Table 3. Data sources for PEC calculation in tiered approach (JPPA, 2006) Exposure pathway

Treatment

Runoff (seepage)

Paddy field

Tier 1

Drift over drainage (paddy field)

Tier 3

Numerical calculation Constant (0.02%)

Lysimeter test

Field test

Surface runoff test

–

Paddy field (Ground application) Upland field (Ground application) Aerial application

Drift table

Drift table

Field test

Drift table

Field test

–

Drift table

Drift table

Drift table

Ground application Aerial application

Drift table

Drift table

Drift table


a


a


a

Upland field Drift over river

Tier 2

Note: If the result of the risk assessment in tier 1 conforms to the registration standard, tier 2 is not required. The same applied for tier 2. a
is equal to the percentage of drainage area to the total treated area.

10km x 10km Watershed Main river (10km long 120m wide)

Tributary river (50km long 16m wide) Paddy field (1km x 5km)

Up land field (1km x 7.5km) Evaluation point

Fig. 3. The environmental scenario used for PEC calculation (JPPA, 2006).

176

H. Watanabe et al.

Table 4. Prescribed scenarios for the pesticide application (JPPA, 2006) Site Application method

Paddy Ground application 500 5.0

Upland Aerial application

Ground application 750 7.5

Aerial application

Area of application (ha) Length of the tributaries along the field edge (km) Use ratio (%) 10 5 Treated area (ha) 50 37.5 Application period (day) 5 1 5 1 Length of applied field (5.0 km ⫻ 0.1)/ (5.0 km ⫻0.1)/ (7.5 km ⫻0.05)/ (7.5 km ⫻0.05)/ along tributaries per day 5 ⫽ 100 m 1 ⫽ 500 m 5 ⫽ 75 m 1 ⫽ 375 m

Table 5. The standard transport scenarios of the applied pesticide (JPPA, 2006) Pesticide used in paddy field

The paddy field runoff flow in the river is set to be at a steady state. The water holding period may be set according to the pesticide label. Case of pesticide drift entering the river. (1) Constant drift from paddy field to the tributaries. (2) Spraying drift to the drainage canal that flows into the river. (3) Over-spraying or off-target spraying on the drainage canal (in case of aerial application).

Pesticide used in upland field

Runoff is generated after heavy rainfalls and flows into the river. Drift as in case (1) for paddy field pesticides.

Pesticide used in both rice and upland fields

PEC shall be calculated for both paddy field and upland field scenarios.

field management, it is assumed that the pesticide application for the entire area takes five days for ground application and one day for aerial application. Detailed information is also presented in Table 4. For the paddy field, the runoff water contains pesticide flows into the river at a fixed rate. The drift is generated at the time of spraying and enters directly into the tributaries of the river. For upland pesticides, the drift is generated at the time of spraying and then enters into the tributaries while runoff is generated after significant rainfall and flows into the tributaries. However, the PECs of the pesticide will be calculated separately because the pesticide is not sprayed at the time of rainfall (Table 5). The calculation of pesticide drift is based on “drift tables” which were prepared according to Ganzelmeier et al. (1995). The method

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

177

for calculating PECs of pesticides used in the paddy field is introduced below for each tier. 4.3.3. Calculation of PECs for tier 1

The PEC of the rice pesticide in the river at the first tier is calculated by the following formula: PECtier 1 ⫽

M runoff ⫹ M Dr ⫹ M Dd 3 ⫻ 86, 400 ⫻ Te

(1)

where, PECtier 1 is the first tier PEC of the rice pesticide in the river (gm⫺3), Mrunoff the maximum runoff amount (g), MDr the amount of the pesticide drift on the event day (g), MDd the amount of the pesticide entering the river via drainage on the event day (g), and Te is the duration of the toxicity test (day). Mrunoff, MDr, and MDd are calculated by the following equations: M runoff ⫽ I ⫻

Rp 100

⫻ Ap ⫻ fp

(2)

M Dr ⫽ I ⫻

Driver ⫻ Z river ⫻ N drift 100

(3)

M Dd ⫽ I ⫻

Dditch ⫻ Z ditch ⫻ N drift 100

(4)

where I is the application rate of the pesticide at the single-application event (gha⫺1), Rp the runoff ratio from paddy field (%), Ap the surface of treated area (ha), fp the adjustment factor for runoff ratio according to the application method, Driver is the drift ratio on river (%), Zriver is the drift area per day (ha day⫺1), Dditch is the drift ratio on ditch (%), Zditch is the drift area per day (ha day⫺1), and Ndrift is the number of drift event (day). Table 6 lists the values of those parameters, which are included in the above equations (JPPA, 2006; MAFF, 2006b). 4.3.4. Determination of PECs for tier 2

In the second tier, PECs should be obtained by lysimeter studies. The pesticide concentration in paddy water is determined by measurements derived from paddy field lysimeter tests and the PECs at the evaluation point in the environmental model are calculated by the prescribed equations. The lysimeter can be made of concrete with an area of 1m2 (1m × 1m) or more and the percolation rate in the lysimeter should be adjustable. The paddy soil is filled homogeneously by the wet filling method, and the depth of the soil layer might be set to about 50 cm. The lysimeter should be installed in such a place that fully reflects the climatic conditions of the area. Lined lysimeters may be preferable to prevent excessive rainwater intake but care should be taken not to disturb natural climatic conditions such as wind and sunlight.

178

H. Watanabe et al.

Table 6. Parameter values applied for the estimation of PEC in tier 1 (JPPA, 2006) Parameter (unit) Ap (ha) Rp (%)

Ground application Te ⫽ 2 days Te ⫽ 3 days Te ⫽ 4 days

Driver (%) Zriver (ha day⫺1) Dditch (%) Zditch (ha day⫺1) Ndrift Te ⫽ 2 days Te ⫽ 3 days Te ⫽ 4 days fp (⫺)

50 15.6 22.4 29.1 0.3 0.16 4 0.07 1 2 2 1 (irrigation-application) 0.5 (foliage-application) 0.2 (seedling box app.)

Aerial application 50 19.0 27.1 34.4 1.9 0.8 100 0.33 1 1 1 0.3 (foliage application) 1 (others)

During the test period, daily percolation is set to about 1–2 cm, and the water level is maintained at about 5 cm. Drainage of ponded water and spillover irrigation should be avoided. The plant grown in the test lysimeter should be the specified target crop grown by the conventional cultivation method as mentioned in the registration documents. There should be at least two replications for each experiment. The test substance is appropriately applied according to the specification mentioned in the registration documents. Due care should be taken to avoid runoff at the time of application or soon after this event. Water samples should be collected just before, immediately after (1–3 hours after the application) and 1, 3, 7, and 14 days after the application from more than four points at 2–3 cm water depth of the paddy water. When residues of the test substance are detected 14 days after the application, the sampling should continue until the detected concentrations fall below 1/10 of the peak value. When the water holding period (WHP) is provisioned, samples are collected on the final day of the WHP. More details and brief analytical procedure for the chemical analysis can be found in the guidelines of MAFF (2006b). Based on the test results, the pesticide concentration in paddy water between 0 to 14 days after the pesticide application is determined by the pesticide dissipation curve in the paddy water for each lysimeter. The PEC value at the evaluation point is calculated considering pesticide mass discharge through surface drainage and runoff, seepage, pesticide drift, and pesticide adsorption onto the stream bed sediments and the PECtier 2 (g m⫺3) are calculated by the following formula: PECtier 2 ⫽

M runoff ⫹ Mseepage ⫹ M Dr ⫹ M Dd ⫺ Mse 3 ⫻ 86, 400 ⫻ Te

(5)

where Mrunoff is the mass of pesticide discharged in the river from the paddy field due to runoff (g), Mseepage the mass of pesticide discharged in the river via

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

179

seepage (g), MDr the pesticide mass entering the river (g), MDd the pesticide mass entering the river via surface drainage (g), Mse the pesticide mass adsorbed onto the bed sediment in the tributary (g), and Te the duration of the toxicity test (day). In the second tier, the PEC of the pesticide in the paddy field is calculated separately for the following cases: (1) WHP is not set; (2) WHP is set; and (3) the pesticide is rapidly degraded in paddy water. In the latter case, PEC is computed considering hydrolysis and photolysis of the pesticide in the water. Evaluation uses the data that present the worst-case values among replications. Detailed procedures including equations are presented in the report of JPPA (2006) and MAFF (2006a). 4.3.5. Determination of PECs for tier 3

The principle of this third tier evaluation is the execution of the second tier test using data from real outdoor paddy field studies (MAFF, 2006b). The test should be carried out in a well-managed paddy field having sufficient area (0.05–0.3 ha) where water losses via percolation and seepage are small. Seepage outflow through the levee should be prevented as much as possible and the average daily decrease of paddy water depth during the test period should be about 1 cm. The conventional paddy rice cultivation of the region should be performed in the test field. Chemical interactions between the test pesticide and other agricultural chemicals used in the field or the influence of other chemicals in the irrigation water should be avoided. In the test paddy field, the water-holding practice after pesticide application is implemented by maintaining the average depth of water at about 5 cm uniformly by managing field operation and irrigation. The use of the test pesticide should be in accordance with its registration documents. Samples are collected from various points (at least ten points for 0.1 ha) in order to include local variability. The sampling will be carried out just before, immediately after (1–3 h after the application) and 1, 3, 5, 7, 10, and 14 days after the application. Samples should be collected at the same hour in each sampling date. In addition, the weather data (temperature, precipitation, and evaporation if it is possible), water depth, and pH should be monitored for the whole test period. Moreover, the occurrence of runoff from the test field should be reported. Replications of two field plots should be implemented, and PECs are calculated as in the second tier test. More details are given in the guidelines of MAFF (2006b). 4.4. Determination of acute effect concentration (AEC) The assessment organisms used were included in order to cover three organism groups: fish representing the second consumer group, crustacean representing the first consumer group, and algae representing the main producers in the aquatic ecosystem. The selected assessment organisms can be typical species of their groups. The actual selection of test organisms is as follows. 1. Fishes: Cyprinodont (Oryzias latipes) or carp (Cyprius carpio) 2. Crustacean: Daphnia (Daphnia magna)

180

H. Watanabe et al.

3. Algae: Green alga (Pseudokirchneriella subcapitata, known as Selenastrum capricornutum) In addition, any organism whose susceptibility is higher than those mentioned above can be selected as assessment organism. The guideline of the toxicity test is available from MAFF (2006b) based on the OECD test guideline for chemical substances. The end point of the acute toxicity for fish is the concentration that is observed to cause the mortality in half of the population of the test organism, denoted as “LC50”. Similarly, the end points of the acute toxicities for crustacean and algae are indicated by “EC50”, the concentration causing swimming obstruction in the crustacean test or growth inhibition in the algae test, in 50% of the population of the test species. Because of the uncertainties in the chemical sensitivity of the individual, the duration of pesticide exposure and the growth stages of the test organisms, an uncertainty coefficient is applied when calculating the AEC. Values between 1 and 10 are commonly used as uncertainty coefficients depending on the test species. 5. MONITORING PESTICIDE RISK IN MONSOON ASIAN CLIMATE The fate of pesticides in paddy fields and their transport into the aquatic environment is a complex phenomenon. The chemical and physical conditions of paddy water and soil significantly affect the fate of the pesticide in the paddy field. In addition, farmers’ practices including land preparation and water management are important factors influencing the transport of pesticides into ground and surface water systems. In order to elucidate the importance of those factors influencing the fate and transport of pesticide in/from the paddy fields, selected monitoring studies at multiple scales, such as field plots, lysimeter, watershed, and river basin studies, should be undertaken. Such studies employed in Japan will be presented in this section. 5.1. Monitoring the fate and transport of rice pesticides in the paddy plots 5.1.1. Monitoring setup and water balance in the paddy plots

The pesticide fate and transport monitoring in a single paddy field includes the water balance monitoring and the pesticide concentration monitoring in paddy water or/and paddy soil at desired soil profiles. Measuring water balance is one of the key elements of the pesticide fate and transport monitoring because “water” is the major carrier phase of the dissolved chemicals in the paddy field. Typical water balance components in a paddy plot can be precipitation, irrigation, surface runoff/drainage, lateral seepage through plot borders such as levees and concrete bunds, vertical percolation, ET, and paddy water depth. A typical setup for the water balance monitoring in paddy plots can be found elsewhere (Watanabe et al., 2006a). Precipitation can be measured by a rain gauge or such data can be obtained from a local weather station. The volume of irrigation is monitored by any type of flow meter in the pipeline or flumes in case of open channel irrigation

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

181

(Replogle et al., 1983), or simply by measuring the rise in paddy water depth during irrigation. Surface drainage can be monitored by using a weir (a V-notch shape or flat weir) and measuring paddy water depth in the vicinity of the weir. The relationship between paddy water level and outflow can be calculated by appropriate hydraulic equations (Rao and Muralidhar, 1963). A covered PVC ring (without ET) can be used in order to measure the daily vertical percolation of the soil. However, considering spatial and temporal variation, multiple periodical measurements such as at the earlier period after puddling and also at later periods are recommended. For a typical Japanese paddy field, the percolation rate ranges from 5 to 30 mm day⫺1 depending on the soil type (Nakagawa, 1967), and it highly depends on farmer’s practice. The ET can be monitored by daily measurements of water level in a lysimeter box (e.g. 30 cm high, 35 cm ⫻ 55 cm) with equivalent plant density of four growing rice plants planted in about 20 cm deep soil that is placed in a plot. The ponding water depth in a paddy field can be monitored by a water depth sensor (LSP-100, UIJIN Co. Ltd, Tokyo) or by hook gauge at each event on irrigation and precipitation. Daily water balance can be estimated using a mass conservation equation for the measured data as given below. hi⫹1 ⫽ hi ⫹ RAIN i ⫹ IRR i ⫺ DRAINi ⫺ LSEEPi ⫺ PERCi ⫺ ETi

(6)

where hi and hi⫹1 are the paddy water depth (cm) at day i and day i ⫹1, RAINi is the depth of rainfall during day i to i ⫹ 1. Similarly, IRRIi, DRAINi, LSEEPi, PERCi, and ETi are the depth of irrigation, drainage, lateral seepage, vertical percolation, and ET during days i to i ⫹1, respectively. Watanabe et al. (2006a) monitored paddy plots managed by intermittent automatic irrigation (AI) and continuous irrigation–drainage (CI). The CI plot maintained relatively deeper paddy water and significant discharges during the monitoring period in response to the continuous irrigation scheme. However, the paddy water depth of AI plot oscillates less frequently with larger amplitude (deferential depth) reflecting the intermittent irrigation schedule and periodic rainfall events. The heights of the drainage gates were 7.5 cm for AI plot and 4.0 cm for CI plot. Mizutani (1995) reported that typical water balance in Japanese paddy fields presents daily depths of precipitation, irrigation, surface drainage, deep percolation, lateral seepage, and ET during an irrigation period of 120 days of 0.75, 1.50, 0.55, 0.30, 0.90, and 0.50 cm, respectively. A similar water balance was observed in a paddy field study in Tsukuba, Japan where daily averages of irrigation, sum of drainage, seepage and percolation, and precipitation and ET were 0.97, 1.17, 0.47, and 0.28 cm, respectively (Watanabe and Takagi, 2000a). Discussion on other important factors influencing pesticide fate phenomena such as solar radiation, pH, Eh, and temperature can be found in Takagi et al. (1998). 5.1.2. Monitoring pesticide concentration in paddy water and soil

A typical schedule for water sampling after pesticide application could be 6 h, 24 h, and 3, 7, 14, 21, 28, and 35 days (Watanabe and Takagi, 2000a). However, for monitoring the dissolution of pesticide in paddy water during the initial period,

182

H. Watanabe et al.

an hourly sampling interval can be selected (Okamoto et al., 1998; Inao et al., 2001). Water samples can be taken as composite samples from 5 to 10 spots in order to represent average concentrations in paddy water. Herbicide concentration near the irrigation pipe is subject to dilution by the entering irrigation water whereas samples taken from spots near the drainage gate are expected to have higher concentrations. Watanabe et al. (2006b) reported that bensulfuron-methyl concentrations in the paddy field ranged from below 0.03 µg L⫺1 near the irrigation pipe to about 1 µg L⫺1 near the drainage gate. This was attributed to the uneven chemical distribution in the paddy field as affected by the irrigation practice. Soil samples can be taken from the corresponding profiles; however, the most important layer controlling pesticide fate in paddy soil is the top 0–1 cm layer. According to Takagi et al. (1998), the submerged top 1 cm layer of paddy soil is aerobic such that pesticide adsorption/desorption and microbial degradation occurs under oxidative conditions. Watanabe and Takagi (2000b) defined the 1.0 cm-thick paddy surface soil layer (PSL) to be the profile governing pesticide desorption and dissipation processes under oxidative flooded conditions. Samples from 0 to 1 cm of the top soil can be taken by driving a 25 cm diameter PVC ring into the surface soil layer, removing surface water from the ring, and scraping the 0–1 cm surface soil layer (ca. 50 g). Composite samples consisting of five or more spots can be taken with the same procedure as for the water samples (Watanabe and Takagi, 2000a; Watanabe et al., 2006a). Figure 4 illustrates typical rice herbicide dissipation in the paddy water and 0–1 cm surface soil layer in the AI and CI plots mentioned in the previous section (Watanabe et al., 2006a). The maximum herbicide concentration occurred at the first day after treatment and then rapidly declined during the next two weeks after herbicide application. During the monitoring period, mefenacet concentrations in paddy water ranged from 660 to 1.1 µg L⫺1 for AI plot and from 540 to 3.1 µg L⫺1 for CI plot. In the same plot of CI, mefenacet concentrations in the drainage water exhibited higher values as compared to its average concentration in the water of the plot because of the spatial variation in the pesticide concentration affected by the irrigation practice as discussed above. Mefenacet concentrations in AI plot during the monitoring period were comparable with the concentrations observed in pesticide monitoring studies at the National Institute for Agro-Environmental Sciences (NIAES) in Tsukuba, Japan (Fajardo et al., 2000; Ishii et al., 2004) when considering that in the presented monitoring study the application rate was a quarter lower than the rate used in the NIAES study. The CI scheme in CI plot resulted in a 38% loss of the applied mefenacet and most of the losses occurred during the first two weeks. Herbicide concentrations in 0–1 cm paddy soil showed a maximum of 22.9 mg kg⫺1 a few days after pesticide application and it exponentially decreased thereafter. 5.2. Monitoring the fate and transport of rice pesticides using lysimeters Lysimeters can be a powerful tool for monitoring pesticide fate and transport. Lysimeter monitoring has many advantages such as multiple replications, data acquisition system, and controllable environmental conditions with relatively

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective 1.2

183

40

30 0.8

MF-AI

0.6

MF-CI

20

MF-drain 0.4

MF-loss

10

Loss (% applied)

MF conc. (mgl-1)

1

0.2 0

0

MF conc. (mgkg-1)

100 MF-AI MF-CI 10

1 0

10

20

30

40

DAHA

Fig. 4. Mefenacet (MF) concentrations in paddy water in AI and CI plots and in drainage water of the CI plot, and its loss (above) and MF concentrations in 0–1 cm paddy soil in AI and CI plot (below) (Watanabe et al., 2006a).

low cost. In addition, application of specific conditions and scenarios can be possible depending on the lysimeter design. In this section, a lysimeter facility and an experimental study using lysimeters at the NIAES are presented. 5.2.1. An experimental facility for studying the fate and transport of agrochemicals

An experimental facility for studying the fate and transport of agrochemicals was established at NIAES in 2000. This research facility has been developed by Takagi et al. (2001) in order to analyze fate and transport mechanisms of agrochemical and endocrine disrupting substance (EDS) in the soil environment, and to evaluate computer simulation models. The facility includes eight box-type packed lysimeters and two dry-well type lysimeters (Figure 5). The box-type lysimeters (2 m ⫻2 m ⫻2 m) consist of four model-rice paddies and four model-upland fields for different soil types including Andisols (Koroboku)-loam, alluvial clay loam, loam, and sand. The soil properties of packed materials in the lysimeters are listed in Table 7. Each lysimeter has an

184

H. Watanabe et al.

Fig. 5. Overview of the lysimeter facility at NIAES.

Table 7. Physicochemical properties of the soil used for packing the lysimeters at NIAES Paddy lysimeter

Unit

Soil profile Soil type pH (H2O) pH (KCl) Coarse sand Fine sand Silt Clay Total carbon content Total nitrogen content

(cm)

1 and 3a

2b

4c

% % % % (%)

0–60 LiC 6.5 5.0 6.5 19.5 35.0 39.0 2.06

60–200 LiC 6.3 5.0 7.8 22.3 36.1 33.8 1.76

0–200 L 6.2 4.8 11.6 50.7 22.7 15.0 0.87

0–60 LiC 5.8 4.8 5.3 23.9 33.5 37.2 4.41

60–200 LiC 6.1 5.4 6.2 31.7 26.4 35.7 1.04

(%)

0.15

0.13

0.08

0.31

0.09

a Gray lowland soil was sampled from the upper and lower layer of a paddy field in Mizukaido and filled for each layer (0–60 cm and 60–200 cm) of the lysimeter. b Brown lowland soil was sampled from an abandoned non-paddy field next to the Tone river and filled the entire lysimeter. c Surface humic Andisols was sampled from the upper and lower layer of an upland field in Ushiku, and filled for each profile (0–60 cm and 60–200 cm) of the lysimeter.

automated irrigation system responding to paddy water depth fluctuations for the paddy lysimeter and to soil water potential fluctuations for upland lysimeters. Monitoring devices installed every 50 cm in the lysimeter soil can measure soil water potential, temperature, pH, and Eh. Ports for soil water sampling are also

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

185

available. The lysimeters are equipped with subsurface drainage system, which drains the water infiltrating through the soil matrix and edge flow along the concrete surfaces separately. The infiltration through the soil matrix is collected in a bottom pan (1.90 m ⫻1.90 m) and the edge flow along the concrete surfaces is collected outside of the bottom pan; then these outflows are drained through separate pipes equipped with tipping-bucket flow meters (Takagi et al., 2001). The dry-well type lysimeters consist of a model paddy field and an upland field (Figure 5). For the paddy field, a dry well of 1.5 m diameter and 3.0 m deep was installed at the center of a 3.0 m ⫻ 4.0 m model paddy field. The model paddy field was divided into two parts consisting of a 60 cm deep Andisols-loam soil on one side and an alluvial clay loam soil with the same depth on the other side. For the model-upland field, a dry well with 1.5 m diameter and 5.5 m depth was prepared in the same way as in the model paddy field but with the soil type of Andisols-loam and loam. Each lysimeter has an automated irrigation system as in the box-type lysimeters above. Chambers equipped with monitoring sensors for measuring soil matric potential, temperature, pH, Eh, and soil water sampling ports are mounted every 50 cm on the spiral coordinate around the well wall. This facility is also equipped with a microclimate monitoring system consisting of air temperature, rain gauge, solar radiation, UV-A, and UV-B radiation sensors. All the monitored data are stored through the online monitoring and data acquisition system as indicated in Table 8. Also, the online monitoring and data acquisition system controls the irrigation systems for both paddy and upland fields. In the dry-well lysimeter, gas-tight chambers are installed in the spiral coordinate every 1 m for soil gas monitoring. The soil gas for O2 and CO2 in the chamber are also automatically sampled and analyzed (Takagi et al., 2004).

5.2.2. Application of lysimeter experiments for developing a mitigation technique for controlling herbicide runoff from paddy field using rice husk charcoal powder

This study aimed to develop a technique for reducing herbicide runoff from paddy fields by restraining the desorption of herbicides from surface soil and also enhancing herbicide adsorption by using rice husk charcoal powder (grain size: 0.25–0.5 mm) as an herbicide-adsorbing material (Takagi and Takanasi, 2003). Plot experiments were conducted by applying charcoal powder in two paddy field lysimeters (2 m ⫻2 m ⫻2 m, replenished with gray lowland soil) assigned to be PL1 (treated plot) and PL3 (untreated plot). Five days after rice transplanting, 12 g (3 kg per 10 a) of Hayate® granules containing 1.5% of pretilachlor (PTC), 0.3% of imazosulfuron (IMS), and 0.2% of dimethametryn (DMT), were applied in both plots. After 24 h, 200 g (50 g/m2) of rice husk charcoal powder was evenly applied in the PL1. After the herbicide application, paddy water and runoff water were regularly sampled in both plots over a period of five weeks to determine the herbicide concentrations in paddy water. With the application of rice husk charcoal powder, the runoff of PTC, IMS, and DMT was reduced by 46%, 46%, and 64%, respectively; as compared with the runoff losses observed in the untreated plot where the charcoal powder was

186

H. Watanabe et al.

Table 8. Environmental variables monitored in the paddy field lysimeters No.

Measured variables

1

Paddy water depth

2

Precipitation

3

Leaching and edge flow

4

Soil matric potential

5

Soil pH

6

Soil Eh

7

Soil temperature

8

Solar radiation

9

UV-A radiation

10

UV-B radiation

11

Soil water sampler

12

Data logger

Materials Water level sensor (HM920-02-KT, Fujiwara Scientific Co. Ltd., Tokyo) Tipping-bucket rain gauge (MES-1705B, Koito Co. Ltd., Tokyo) Tipping-bucket water gauge (F-W100-KT, Fujiwara Scientific Co. Ltd., Tokyo) Tensiometer (DIK-3032-KT, Daiki Rika Kogyo Co. Ltd., Saitama) pH sensor (ELCP13L-KT, Fujiwara Scientific Co. Ltd., Tokyo) Eh sensor (ELCP63L-KT, Fujiwara Scientific Co. Ltd., Tokyo) Type T thermocouple (F-T1-3.2-300, Fujiwara Scientific Co. Ltd., Tokyo) Solar radiation sensor (ISK-37, Koito Co. Ltd., Tokyo) UV-A radiometer (MS-210A, EKO instruments trading Co. Ltd., Tokyo) UV-B radiometer (MS-210W, Eko instruments trading Co. Ltd., Tokyo) FV-429-500-KT, Fujiwara Scientific Co. Ltd., Tokyo DIK-9420, Daiki Rika Kogyo Co. Ltd., Saitama

Note: Time-averaged data recorded every 10 min for No. 1, 4, 8, 9, 10; hourly data recorded for No. 5, 6, and 7.

not applied (Figure 6). Comparing the cumulative values for the three herbicides, the total loss was reduced by 49% compared to the control plot due to the decreased herbicide concentration in the paddy water after the application of the rice husk charcoal powder (Takagi and Takanasi, 2003). 5.3. Monitoring the fate and transport of rice pesticides at a watershed scale Monitoring the pesticide fate and transport of pesticides at a watershed scale consists of monitoring water balance and pesticide concentrations in water and soils. Although the scale of a watershed ranges from several to thousands of hectares, a monitoring project on a rice-cultivated watershed of about 100 ha is presented in this section (Vu et al., 2004, 2006). The monitoring of hydrological balance and pesticide concentrations in surface water was carried out in a watershed of 97 ha in the Sakura river basin in Tsukuba, Japan (Figure 7). The monitoring was conducted at three different scales: (1) paddy

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective 1.6

140

1.4

120 100

1 80 0.8 60 0.6 40

0.4

20

0.2 0

Runoff Water (L)

PTC Loss (mg)

1.2

187

0

10 20 Days after herbicide application

30

0

Runoff water from PL1

Runoff water from PL3

PTC runoff from PL1

PTC runoff from PL3

Fig. 6. Cummulative pretilachlor (PTC) losses and runoff water volume from plots treated with (PL1) and without charcoal powder (PL3).

plot, (2) farm blocks, and (3) watershed. Precipitation was monitored by a rain gauge placed at Station 11 (S11). Water balance variables (irrigation, runoff/ drainage, and ponding water depth) were monitored at a selected paddy plot (plot 1) of 0.3 ha in the watershed. Water balance in selected farm blocks 1 and 2 of about 5 ha each was also estimated from measured data of surface drainage, seepage from field plots, and water balance in corresponding canal sections. For the inflow– outflow balances, discharges in the canal networks were monitored by water level sensors with periodical flow measurements at each station in the watershed (S1 through S10). The monitoring period lasted for 45 days, from rice transplanting up to the midterm drainage period. The selected paddy plot was irrigated using an intermittent irrigation and drainage practice. The height of the drainage gate had been set to 5 cm, which is similar to typical water management in Japan (Watanabe, 1999). However, water management depends on the farmer. According to the field survey conducted in 2005, in 113 over a total 296-surveyed plots the overflow drainage scheme is used. In more than 60% of the paddy plots the paddy water level was kept 1 cm below the top of the drainage gate. This setup resulted in a high potential for pesticide runoff when significant rainfall occurs but also in case of strong wind if the drainage gate is located along the wind direction. Therefore, increasing the excess water storage depth (EWSD), defined as the depth between the paddy water level and the top of the drainage gate that can store the excess rainfall, is important for the control of runoff from paddy fields. Paddy water runoff and

188

H. Watanabe et al.

Fig. 7. A map of the studied watershed and monitoring stations (Vu et al., 2004).

discharge from the watershed significantly increased after rainfall events of more than about 1.5 cm day-1 (Vu et al., 2004). The studied watershed had a cyclic irrigation system meaning that the paddy drainage is pumped up and reused for irrigation. The monitoring data for the hydrological balance in the watershed during a 30-day period from May 1st to 31st in 2003 showed that about 3.8 cm drainage water was estimated to be reused for irrigation, which corresponds to 10.2% of watershed discharge monitored at S10. Besides increasing the water use efficiency, cyclic irrigation system may play an important role in reducing pollutant runoff from the watershed. The monitoring of pesticide concentration in surface water was conducted on a weekly basis for paddy water and drainage canal water. During the monitoring period, pesticide discharge from the watershed was also monitored at the outlet of the watershed (S10). In total 16 herbicides were detected in the stream water. Herbicides detected at high concentrations included dymron, mefenacet, IMS, and PTC with maximum concentrations ranging from 2 to 65 µg L⫺1. The maximum concentrations occurred mostly during May 1st to 22nd corresponding to the period during or shortly after herbicide application. The total concentration in stream water, which is the summed concentrations of the detected herbicides, peaked around May 14th (Figure 8). The maximum concentration of mefenacet, one of the most commonly used herbicides in Japan, in the secondary drainage canal (S5 and S7) was 20–30 µg L⫺1 which is 2 to 3 times higher than the water quality advisory level of 9.0 µg L⫺1 for mefenacet in surface water (Ministry of

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

S7 (2003)

80 60 40 20

100 Conc.(µg L-1)

Conc.(µg L-1)

100

0 4/30 5/7 5/14 5/21 5/28 6/4 6/11

60 30

.

0 4/30 5/7 5/14 5/21 5/28 6/4 6/11 Oxazic lomefone Dimethametryn Pyributicarb Cafenstrole Bensulfuron-methyl Dymron

Conc.(µg L-1)

Conc.(µg L-1)

120 90

S10 (2003)

80 60 40 20

0 4/30 5/7 5/14 5/21 5/28 6/4 6/11

150 S7 (2004)

189

60 S10 (2004) 50 40 30 20 10 0 4/30 5/7 5/14 5/21 5/28 6/4 6/11

Symetryn Pretilachlor Pentoxazone Molinate Imazosulfuron Total concentration

Esprocarb Pyriminobac-methyl (E) Mefenacet Thiobencarb Pyrazosulfuron-ethyl

Fig. 8. Measured herbicide concentrations in stream water at stations S7 and S10 in 2003 and 2004 (Vu et al., 2004).

the Environment, 2006) issued by the MOE. Significant daily herbicide losses corresponded well with the peak outflow after significant rainfall events. More than 70% of the total loss of detected herbicides was discharged from the watershed from May 7th to May 22nd responding to the pesticide application period (Figure 9). This finding implies that the pesticide runoff is strongly affected by the timing and the amount of rainfall events relatively to the timing of pesticide application. From this monitoring study it can be concluded that the recommended strategies for reducing pesticide runoff at paddy plot level are the holding of paddy water during and shortly after pesticide application, and a water management practice such that the paddy water runoff following significant rainfall is prevented by increasing the EWSD. At watershed level, cyclic irrigation system can reduce herbicide discharge from the watershed to public water systems. The combination of water management practices in the rice fields as indicated above and small-scale water cycling at the watershed level could be a good management practice to reduce pollutant runoff into river in a paddy watershed. 5.4. Monitoring pesticide concentrations in a river basin Most of the rice-cultivated areas in Japan are treated with herbicides. About 50 different active ingredients (herbicides) are now registered in Japan. The number of registered rice herbicides almost doubled in the last 20 years but the

190

H. Watanabe et al. 8.0

2000

Daily loss (g)..

6.0

1200 4.0 800

Cum. loss (kg)..

2003 1600

2.0

400 0

0.0 8.0

1000

Daily loss (g)..

800

6.0

600 4.0 400 2.0

200 0

5/1

5/8

5/15

5/22

Daily loss (g)

5/29

6/5

6/12

Cum. loss (kg)..

2004

0.0

Cum. loss (kg)

Fig. 9. Herbicide losses from watershed monitored at S10 into Sakura river in 2003 and 2004 (Vu et al., 2004).

introduction of low-dose high-potency herbicides such as sulfonylurea herbicides helped to reduce the sales of rice herbicides by volume by almost a third. Therefore, some rice herbicides are detected frequently in river water, although at low levels (in the order of
g L-1), even a couple of months after rice transplanting (Maru, 1991; Nakamura, 1992; Okamoto et al., 1998; Shiraishi et al., 1998; Mitobe et al., 1999; Sudo et al., 2002). Their adverse effects on aquatic ecosystems have been demonstrated by several studies (Hatakeyama et al., 1992, 1994, 1997a,b). In this section, behavior of rice herbicides at a typical rural river will be presented as an example of pesticide monitoring in a river basin in Japan. Monitoring the concentrations of rice herbicides was carried out at the Sakura river basin and the lake Kasumigaura in Japan (Figure 10). The lake Kasumigaura is the second largest lake in Japan, and is located 60 km northeast of Tokyo. The Sakura river is one of the main rivers which flows into the lake Kasumigaura and the river basin drains about 81.6 km2 of paddy fields accounting for about 25% of total basin area (Figure 10). Paddy fields were spread evenly along the river line. Rice herbicide concentrations were monitored at seven stations in 2001. The S1 was located along the hillside of Mount Tsukuba and one of the headstreams of the Saka river, which is the tributary of the Sakura river. There are neither paddy fields nor houses above this point. The S2 is located downstream of the Saka river, which

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

191

Fig. 10. Sampling stations at the Sakura river basin and the lake Kasumigaura (Ishihara et al., 2005).

flows through an extensive paddy field area. The sampling point S3 is located at the midstream of Sakura river, and the S4 is located at the downstream of Sakura river. The S5 is located at the river mouth, and S6 is located in Tsuchiurairi Bay. Finally, S7 is located at the center of the lake Kasumigaura. The water-sampling period was extended from March 20th to September 19th in 2001. Sixteen rice herbicides (cafenstrole, dimepiperate, DMT, esprocarb, mefenacet, molinate, pentoxazone, PTC, pyributicarb, pyriminobac-methyl, simetryn, thiobencarb, bensulfuron-methyl, dymron, IMS, and pyrazosulfuronethyl) were analyzed. The highest concentrations of rice herbicides were observed in Sakura river during mid May to early June and these maximum detected concentration levels at S3 ranged from 0.12
g L⫺1 for pyriminobac-methyl to 8.8
g L⫺1 for dymron. The concentrations of the rice herbicides in the Sakura river were declined as they enter the lake Kasumigaura and diluted with lake water. The chemical concentrations in the lake Kasumigaura were about 3 to 17 times lower than those in the Sakura river (Figure 11). Recently, the detected levels of herbicides in surface waters tend to decrease in Japan. Especially during 1990s, this tendency became obvious. This phenomenon is probably due to the reduction of rice-cultivated areas. Also, the development/application of low-dose high-potency herbicides whose application favor the decrease in herbicide levels in the surface water systems has advanced during this period. The concentration levels of rice herbicides observed in lake Kasumigaura did not exceed the EC50’s for fresh water algae. However, the concentrations of some rice herbicides at the Sakura river stations exceeded the EC50 for fresh water algae (Ishihara et al., 2005). Therefore, the results imply that more precise risk assessment of these herbicides on primary producers of rural rivers will be required in the future.

192

H. Watanabe et al.

Total conc. (µg/L)

20

15

10

5

0 Mar.

Apr.

May St. 1 St. 5

June St. 2 St. 6

July

Aug.

St. 3 St. 7

× St. 4

Sep. (2001)

Fig. 11. Seasonal variation in the total herbicide concentration in water samples from the Sakura river basin and the lake Kasumigaura in 2001 (Ishihara et al., 2005).

6. MODELING THE FATE AND TRANSPORT OF PESTICIDES AT PADDY FIELD AND WATERSHED SCALE 6.1. Simulation of pesticide fate and transport in the paddy rice environment using PADDY, PADDY-2 and PADDY-Large In this section, a pesticide paddy field model called PADDY will be presented. The PADDY model has been used for predicting pesticide concentrations in a paddy field and the amount of pesticide runoff to adjacent water bodies (Inao and Kitamura, 1999). A modified version of the PADDY model (PADDY-2) was later developed for the calculation of daily water balance in paddy field with sitespecific conditions (Inao et al., 2001). This section also covers a simulation model (PADDY-Large) for predicting pesticide concentration in main drainage canals and rivers (Inao et al., 2003). The PADDY model series has been primarily developed for assisting the pesticide regulation process by predicting environmental concentrations of pesticides. 6.1.1. PADDY model

The PADDY model calculates pesticide concentration in paddy water and surface soil by considering pesticide behavior and average water balance in a paddy field as shown in Figure 12 (Inao and Kitamura, 1999). The paddy field is a compartment system, which consists of surface and subsurface layers. The surface layer is composed of the paddy water compartment and the surface soil phase, which comprises pore water and soil solid compartments under flooded conditions. The thickness of the surface soil phase is set to be 5 mm (active soil layer), and the pore water is included in the paddy water. Each compartment is assumed

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

193

Fig. 12. Conceptual water balance and pesticide fate and transport in paddy field of the PADDY model (Inao et al., 2001).

to be in a completely mixed condition and the driving force of mass transfer between compartments is the gradient of pesticide concentration in each compartment. The pesticide behavior in the system is controlled by the interaction of the following processes: the dissolution of pesticide from granules into paddy water, the adsorption and desorption of pesticide between paddy water and soil, the runoff, the leaching and the volatilization of pesticides from paddy water, and the degradation of pesticide in paddy water and soil. The mass balance equations in paddy water and surface soil compartments are expressed in terms of the kinetics of fate and transport processes. In the model, steady state was applied in water balance (i.e. water depth, runoff, and percolation rate are constant). The model program was coded using Microsoft Visual Basic based on Windows. Methods for measuring certain pesticide parameters were also developed for assisting model simulation (Inao and Kitamura, 1999). Measurement of the dissolution rate of pesticide from granules was performed using the batch method. Distilled water corresponding to about 5 cm depth of water was put into a cylindrical glass jar (15 cm ⫻ 20 cm ID), and the granule formulation representing the recommended normal dose was added to the jar. At various time intervals (e.g. 0.5, 1, 3, 6, 24, and 48 h after treatment) pesticide concentration in water was measured and the dissolution rate was obtained from a concentration–time profile. Desorption rate was measured by use of the soil column method. Pesticide aqueous solution penetrates slowly through a soil column. The leachate is analyzed for pesticide concentration, and the amount of pesticide adsorbed onto the soil is estimated. Then distilled water, without the test pesticide, is added to the column and surface water is collected after 6 h. Amounts of pesticide desorbed

194

H. Watanabe et al.

from soil are analyzed, and pesticide concentrations in soil are estimated. These operations are repeated at various intervals (e.g. 24, 51, and 75 h). Desorption rate is obtained from a concentration–time profile in the soil. This test is conducted in case the depth of the soil column varied from 5 to 50 mm. To validate this model, a field experiment was carried out in a paddy field in Tokyo from June 24 through July 22 in 1991 using two herbicides (molinate and simetryn) (Inao and Kitamura, 1999). The concentrations of these pesticides were measured in paddy water and surface soil. The model input data for environmental conditions in paddy field and pesticide parameters are listed in Tables 9 and 10, respectively. Water depth, outflow rate, and percolation rate were taken from average values obtained during the experimental period. Table 9. Conditions of experimental paddy field in the validation study for the model PADDY (Inao and Kitamura, 1999) Parameter

Unit

Value

Area Water depth Depth of surface soil Thickness of each subsurface layer Dry bulk density at the 0–2 cm soil layer Dry bulk density at the 2–4 cm soil layer Porosity at the 0–2 cm soil layer Porosity at the 2–4 cm soil layer Outflow rate of water Percolation rate of water

m2 m m m ton m⫺3 ton m⫺3 m3 m⫺3 m3 m⫺3 m3 day⫺1 m3 day⫺1

1000 0.032a 0.005 0.005 0.60 0.71 0.56 0.46 17.7a 9.5a

aAverage

value during experimental period.

Table 10. Model input parameters for molinate and simetryn for PADDY (Inao and Kitamura, 1999) Parameter

Unit

Physicochemical properties Molecular weight Water solubility Vapor pressure

g mol⫺1 mg L⫺1 mm Hg

Molinate

Simetryn

187.3 800 (20ºC) 5.7 ⫻ 10⫺3 (25ºC)

213.3 450 (20ºC) 7.1 ⫻ 10⫺7 (20ºC)

Equilibrium constants Henry’s constant ⫺ Adsorption coefficient L kg⫺1 [Freundlich exponent (1/n)] ⫺

7.2 ⫻ 10⫺3 7.9 0.96

1.8 ⫻ 10⫺8 13.1 0.75

Rate constants Dissolution Adsorption Desorption Volatilization Degradation in water Degradation in soil

3.1 ⫻ 10⫺2 2.9 ⫻ 10⫺1 2.9 ⫻ 10⫺1 1.6 ⫻ 10⫺2 1.9 ⫻ 10⫺2 1.7 ⫻ 10⫺2

9.6 ⫻ 10⫺3 1.5 ⫻ 10⫺1 1.5 ⫻ 10⫺1 3.8 ⫻ 10⫺6 4.7 ⫻ 10⫺3 1.4 ⫻ 10⫺2

day⫺1 day⫺1 day⫺1 m day⫺1 day⫺1 day⫺1

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

195

The predicted results by the model were in a good agreement with the experimental ones. The PADDY model can also estimate the contribution of the fate and transport processes in pesticide dissipation. The major dissipation processes estimated by the model under the experimental conditions in 1991 were runoff and volatilization for molinate, and was runoff for simetryn. The PADDY model was constructed using the assumption that the thickness of the surface soil phase was 5 mm. As a result of desorption rate measurement, there was no difference in the total amounts of desorbed pesticides when the soil thickness was changed from 5 to 50 mm. It was suggested that the thickness of the active soil layer, where the adsorption and desorption of pesticide occurs, is about 5 mm or less.

6.1.2. PADDY-2 model

An improved version of the PADDY model (PADDY-2) was evaluated to estimate the behavior of pesticides more accurately by considering daily water balance in paddy field with site-specific environmental conditions (Inao et al., 2001). For the water movement processes in paddy field, the PADDY-2 model considers inflow (irrigation), precipitation, evaporation, transpiration, outflow (runoff ), and both vertical and horizontal percolation as shown in Figure 12. In the PADDY-2 model, daily water depth and outflow rate can be calculated by the water balance equation. These values when substituted in the pesticide mass balance equations can be used for the calculation of pesticide concentrations in paddy water and soil. To validate the PADDY-2 model, field studies were performed in the experimental paddy fields (40 m2) at NIAES under two different water management conditions in 1996 using molinate and simetryn. In the first treatment, the water depth was maintained at about 4 cm by occasionally supplying irrigation water without outflow. The other treatment involved continuous irrigation where the outlet of the paddy field was set equal to the water level of 4 cm and irrigation water was supplied at a fixed flow rate; thus, excess water overflowed from the outlet. The concentrations of two pesticides in paddy water, water depth, volume of irrigation water, and the fluctuation of paddy water depth resulted from vertical percolation and ET were measured during the experimental period. The average vertical and horizontal percolation rates of water during the experimental period were 0.4 and 0.3 cm day⫺1, respectively. For the continuous irrigation method, water was supplied at an average rate of 0.92 m3 day⫺1, which was equivalent to a 2.3 cm day⫺1 increase in the water level. Precipitation (observed), discharge (simulated), and water depth (observed and simulated) are shown in Figure 13. In the field where continuous irrigation was applied, the fraction of discharged water over the total volume of paddy water was about 30% per day under no rainfall condition. At 7–9 days after the application, daily precipitation exceeded 20 mm day⫺1, and runoff was also generated on the field with water-holding management practice. Calculated depth of water in the field with the water-holding management practice agreed well with the trend of measured values (Figure 13).

196

H. Watanabe et al. 60 a

Precipitation (mm)

50 40 30 20 10

Discharge (mm/day)

0 80 b

Continuous irrigation

60

Water holding management 40 20

Water depth (mm)

0 70 c

60

simulated

50

measured

40 30 20 10 0 7/1

7/8

7/15

7/22

7/29

8/5

1996

Fig. 13. Changes in environmental conditions: (a) daily rainfall; (b) discharge simulated by PADDY-2; (c) water depth in the field with water-holding management (Inao et al., 2001).

For both pesticides the concentrations in paddy water were the same during the first day after the application under both water management conditions. From the second day, pesticide concentrations in the field with continuous irrigation were lower than those in the field with water-holding management. This difference was mainly attributed to higher runoff losses of pesticide in the plot with continuous irrigation. Good agreement between measured and simulated values was obtained for the two pesticides by considering the water management condition and rainfall. Simulated and measured concentrations of molinate in paddy water and soil are shown in Figure 14. Runoff losses of pesticides were also calculated by the PADDY-2 model over the experimental period of 32 days. The cumulative runoff losses of molinate and simetryn were 40.5% and 60.4% of the initial applied mass, respectively, under the continuous irrigation management. From the viewpoint of reducing the environmental pollution of pesticides due to surface

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

197

10

Concentration (mg/l)

(a)

10

1

Water-holding management

simulated measured

-1

Continuous irrigation

simulated measured

10-2 10-3 10-4 0

5

10

15

20

25

30

35

30

35

Time after application (days) 10

Concentration (mg/kg)

(b) 1

10-1 0 - 2.5 cm 10-2 2.5 - 5 cm 10-3

0

5

simulated measured simulated measured

10 15 20 25 Time after application (days)

Fig. 14. Comparison between simulated and measured concentrations of molinate (a) in paddy water; (b) in soil of water-holding management plot (Inao et al., 2001).

runoff, it is important to regulate paddy water depth using the water-holding management practice. 6.1.3. PADDY-Large model

A landscape-scale simulation model (PADDY-Large) based on the PADDY and PADDY-2 models was developed for predicting pesticide concentrations in main drainage canals and rivers due to runoff from paddy fields (Inao et al., 2003). Depending on the irrigation systems, a rice-producing area was classified into four levels as “individual field plot (30 a)”, “farm block comprising 20 field plots and branch canals (6 ha)”, “district with a main drainage canal”, and “river basin” (Figure 15) and pesticide behavior was estimated focusing on the main drainage canals.

198

H. Watanabe et al.

Mountain, Forest

District

Main canal (stream)

River basin

River Branch canal Field plot (30 a) Farm road

Branch canal

Farm block (6 ha)

Fig. 15. Standard scenario for evaluating pesticide behavior in a rice-producing area (Inao et al., 2003).

Pesticide concentrations in paddy drainage/runoff from the field plot can be calculated using PADDY or PADDY-2 models. When pesticides are used as ground applications, generally they are not applied all at once in a farm block. In the model, it is assumed that the distribution of application dates follows a normal distribution function, and the number of field plots where pesticides are applied at a time in a farm block was estimated by considering the amount of actual pesticide used and the timing of application in a district. For predicting pesticide concentration in a main canal of a district, a CSTR (continuous stirredtank reactor) model concept was employed (Neely, 1976). A main canal can be visualized as a series of continuous stirred flow compartments. The main canal consists of surface water and a sediment solid compartment and the canal is divided longitudinally into a number of segments. The mass balance equations for pesticides in these compartments are expressed in terms of the kinetics of fate and transport processes. To validate the model, a pesticide monitoring was conducted in a rice-producing area in the southern part of Ibaraki. A main drainage canal was located in the center of the district with a catchment area of 271 ha. Rice paddy fields were spread along the basin of the canal and the total planted area was 55 ha. Rice was transplanted from late April to early May, and harvested in the middle of September. One-shot herbicides were mainly applied at 5–15 days after transplant. The

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

199

25

Concentration (
g / l )

20 simulated 15

measured

10

5

0 4/24 5/1

5/8 5/15 5/22 5/29 6/5 6/12 6/19 6/26 7/3 7/10 7/17 7/24

1997

Fig. 16. Measured and simulated concentrations of mefenacet in the standard scenario for evaluating pesticide behavior in a rice-producing area (Inao et al., 2003).

surveillance was carried out from late April to late August in both 1996 and 1997. Herbicide concentrations in the canal increased in early May, reached a maximum in mid May, and declined to below detection limits by early July. Concentrations of the herbicide mefenacet were higher than those of the other herbicides, because of the wide use of mefenacet covering 61% of the area. The maximum concentrations of mefenacet in 1996 and 1997 were 6.8 and 11.4
g L⫺1, respectively. Agreement between simulated and measured concentrations of the mefenacet in the main canal was obtained by considering actual pesticide use and environmental conditions in the rice-producing area (Figure 16). Therefore, the PADDYLarge model has the potential to predict pesticide concentrations in rivers accurately for the ecological risk assessment. 6.2. Simulation of pesticide fate and transport in paddy fields by the PCPF-1, PCPF-SWMS, and PCPF-C In this section, the simulation model series of Pesticide Concentrations in Paddy Field (PCPF) will be introduced for simulating the fate and transport of pesticides in paddy fields using PCPF-1, in paddy soil profiles using PCPF-SWMS, and in the surface water of a paddy watershed using PCPF-C. The PCPF model series has been used mainly for developing and evaluating field management practices such as irrigation and drainage control and soil management in order to minimize pesticide runoff and seepage losses. The simulation model PCPF-1 was used for predicting pesticide concentrations in paddy water and surface soil (Watanabe and Takagi, 2000a,b; Karpouzas et al., 2006; Watanabe et al., 2006c). PCPF-1 applications for evaluating the water management practices for pesticide runoff control have been discussed previously (Watanabe and Takagi, 2000c;

200

H. Watanabe et al.

Watanabe et al., 2005); and importance in mitigation practice is also discussed through PCPF-1 simulations in later section. The simulation of pesticide transport in paddy soil is performed by the PCPF-SWMS model (Tournebize et al., 2004, 2006) in which PCPF-1 and a finite element model for solute transport simulation are coupled. PCPF-C simulates pesticide fate and transport at a catchment scale and provides pesticide concentrations in drainage canals and streams as well as its losses from paddy fields. PCPF-C has been applied for the probabilistic assessment of the best water management practices for controlling pesticide runoff from paddy fields. 6.2.1. PCPF-1 model

The PCPF-1 model simulates the pesticide fate and transport in two compartments: the paddy water and 1 cm paddy soil layer (PSL). The paddy water compartment is assumed to be a completely mixed reactor having variable water depths. The PSL compartment is also assumed to be a completely mixed reactor, but with a constant depth of 1.0 cm. Both compartments are assumed to be homogeneous having uniform, unsteady chemical concentrations. The PSL is defined as 1.0 cm-thick conceptual surface paddy soil layer governing the pesticide desorption and dissipation processes under oxidative flooded conditions. Considering the paddy field environment, the conceptual pesticide fate scenario used for the model is shown in Figure 17. For water balance, the model considers the changes in paddy water depth, precipitation, irrigation, drainage, vertical percolation, lateral seepage, and ET. In case there are no ET data, the model has an internal module for estimating daily ET using the FAO Penman–Monteith procedure when other weather data are available (Vu et al., 2005). Evapotranspiration

Precipitation Volatilization

Drainage Irrigation

Photochemical degradation Biochemical degradation Dissolution

Desorption Lateral Seepage

Granule pesticide

Percolation

Paddy water compartment

Adsorption Biochemical degradation Pesticide Desorption layer compartment

Fig. 17. Conceptual pesticide fate and transport in paddy rice field of the PCPF-1 model (Watanabe and Takagi, 2000b).

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

201

% applied mass

For the chemical mass balance, chemical processes in each compartment are considered. During the simulation, the pesticide concentration in the paddy water is controlled by its dissolution upon application, the pesticide transfer by desorption from the PSL, the dilution of the dissolved pesticide by precipitation and irrigation, the concentration by ET, the pesticide dissipation by volatilization, and the biochemical and photochemical degradation. In the PSL compartment, pesticide concentration is controlled by pesticide desorption into paddy water, percolation of pesticide from paddy water, adsorption of pesticide on to PSL, leaching to the subsurface soil below PSL, and biochemical degradation. Chemical processes including dissolution, desorption, and biochemical and photochemical degradation were all assumed to follow first-order kinetics. The model program is coded using Visual Basic for Applications in Microsoft Excel®. The input data consist of 23 measured parameters, the daily water balance of the paddy water, and local meteorological data. Upon execution of the Macro program, the Macro performs the model calculations and automatically creates output data and figures in a Microsoft Excel file. The current version is also equipped with a pesticide mass balance sheet in order to facilitate further data analysis. Detailed explanation for the model development and evaluation are given elsewhere by Watanabe and Takagi (2000a,b) and Watanabe et al. (2006c). The PCPF-1 model was evaluated using the results from a field monitoring study with the herbicide mefenacet, which was carried out in the experimental paddy field at NIAES from May 13th to July 15th in 1998 (Watanabe et al., 2006c). Mefenacet concentration in paddy water and PSL were monitored for the study period and used for model validation. Measured daily data for the water balance calculation, and ET simulation during 64 days were fed in the data worksheets of the PCPF-1 model. Parameter values were determined through laboratory experiments conducted in NIAES (Takagi et al., 1998; Watanabe and Takagi, 2000a). The PCPF-1 model successfully simulated concentrations of mefenacet in paddy water and soil with good accuracy (Watanabe et al., 2006c). Figure 18 shows mefenacet mass balance in paddy field calculated by the PCPF-1 model. In this mass balance, seepage loss was not considered and all applied herbicide was assumed to dissolve in paddy water. At the end of the 63 days Paddy water

70 60 50 40 30 20 10 0

Pa Paddy soil Runoff Percolation 0

10 20 30 40 50 Time after herbicide application (days)

60

Dissipation

Fig. 18. Distribution of mefenacet in paddy water and paddy surface soil layer (PSL), and pesticide losses by runoff/drainage, percolation, and dissipation (including photochemical degradation, biochemical degradation, and volatilization) during the monitoring period (Neely, 1976).

202

H. Watanabe et al.

of simulation, mefenacet residues in paddy water, and 1 cm PSL were calculated to be 0.04 and 0.27%, respectively, and losses by surface drainage or runoff and percolation were 41.9 and 6.4%, respectively. The chemical dissipation losses including photochemical and biochemical degradation, and volatilization losses from paddy water and biochemical degradation from PSL accounted for 57.3% (Watanabe et al., 2006c).

6.2.2. PCPF-SWMS model

A new simulation model was recently developed for simulating pesticide fate and transport in the deeper layers beneath paddy fields (Tournebize et al., 2004, 2006). PCPF-SWMS model is a coupled model based on the PCPF-1 model described above and the SWMS-2D model. The main objective of the development of this coupled model was to assist the analysis of the comprehensive functioning of paddy field for the whole soil profile. Also, the model could be the basis for developing a simplified module dedicated to pesticide transport in the paddy system. SWMS-2D is an open-source Fortran-coded model in HYDRUS-2D, a Windows-based modeling environment for the analysis of water flow and solute transport in variably saturated porous media (Simunek et al., 1999). The program solves the Richards’ equation for saturated–unsaturated water flow and a Fickian-based advection–dispersion equation for solute transport including provisions for linear equilibrium adsorption, zero-order production, and first-order degradation. The governing equations are solved using a Galerkin type linear finite element scheme. Specifically for the pesticide, the degradation processes are modeled by a first-order kinetic with half-life coefficient and the sorption processes by soil/water partitioning coefficient (Kd). The coupling of the two models was carried out by linking percolation flux, which was induced by ponded water depth, and the predicted chemical concentration at soil surface between paddy water compartment and paddy soil compartment using PCPF-1 and SWMS-2D (Figure 19). Interaction between these two model compartments for water movement and solute exchange could be summarized as follows: PCPF-1 provides paddy water depth as a top boundary condition in SWMS (pressure prescribed data) and SWMS-2D determines the vertical percolation which is an input data in the PCPF water balance equation. For the solute transport, PCPF-1 provides the top boundary solute concentration which, in association with the percolation rate, determines the input solute flux for the SWMS-2D simulation. Practically, a water balance equation is first incorporated in the main program of SWMS-2D and then a new subroutine called PCPF is added which is the Fortran version of the PCPF-1 code. The calculation is carried out in a way that daily ponded water depth imposes prescribed pressure conditions and then Watflow subroutine calculates percolation rate, which is then used in water balance for the next simulated day. For the solute transport, the boundary concentration (cBnd) value is replaced by the predicted value by PCPF subroutine. The model parameters of PCPF-SWMS were calibrated using the observed soil water potentials and Cl⫺ ion tracer in the paddy soil profiles (Tournebize et al., 2006).

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

203

PTC (ng/ml)

Fig. 19. Conceptual model compartment of the PCPF-SWMS model (Tournebize et al., 2006).

0.10 0.09

Simul (15cm)

0.08 0.07

Obs SW9 (15cm)

Obs SW6 (15cm)

0.06 0.05

Mid Term Drainage

0.04 0.03 0.02 0.01 0.00 0

10

20

30

40

50

60

70

80

90 100 110 120 130

DAHA

Fig. 20. Observed pretilachlor concentrations in soil water at two spots (SW6 and SW9) and corresponding simulated concentrations at 15 cm depth in the experimental paddy plot in 1998 (Neely, 1976).

The coupled model was applied for simulating PTC transport in an experimental paddy plot at NIAES (Tsukuba, Japan) in 1998 (Watanabe and Takagi, 2000a). Figure 20 shows the observed and simulated concentrations of PTC by the PCPF-SWMS model in paddy soil water at 15 cm depth at two spots in the experimental plot. The maximum concentrations of PTC in soil water ranged

204

H. Watanabe et al.

from 0.06 µg L⫺1 at 15 cm depth to 0.02 µg L⫺1 at 45 and 85 cm depths. In the puddled layer, at 15 cm depth, the pesticide concentration peaked at 29 days after the application. The residential time in this layer was evaluated to be 41 days. Below hardpan layer up to 50 cm of subsoil layer, pesticide concentration ranged mostly below 0.02 µg L⫺1 during the monitoring period. These concentrations were obtained considering no degradation of PTC in the subsoil due to its low organic matter content. An amount of 6% of the total applied PTC was calculated to remain in the soil below the hardpan layer at the end of crop season. The new coupled PCPF-SWMS model can be a beneficial tool to simulate pesticide transport in the soil profile beneath rice paddies. The model has the potential to contribute to the surface and ground water quality management. The main drawback of this model is the time-consuming procedures for calibration of the model parameters (Tournebize et al., 2004).

6.2.3. PCPF-C model

As an expansion of the PCPF model series, a new model called PCPF-C was developed for simulating pesticide fate and transport in paddy areas at the catchment scale. The PCPF-C model algorithms were developed to analyze and simulate more realistic situations of water management and pesticide application procedure in rice fields in order to evaluate the Best Management Practices (BMP) for reducing pesticide discharge at the catchment scale. The model simulates pesticide runoff from a paddy catchment which ranges from several hectares (farm block scale) to few dozens of hectares (small watershed scale). The catchment consists of paddy block compartment and canal compartment. The paddy block compartment is divided into Pesticide Treatment Groups (PTGs). Each PTG is treated with one pesticide active ingredient. The model PCPF-1 (Watanabe and Takagi, 2000a,b) is used for simulating fate and transport of pesticide in paddy water within each PTGs. The canal compartment is assumed to be a completely mixed reactor having uniform and unsteady concentration of pesticide (Completely Mixed Tank Model). The processes of pesticide dissipation in the canal water, such as biochemical and photochemical degradation, pesticide sorption onto bed sediment, and volatilization, are neglected. The governing equations consist of a water balance equation in paddy field and river compartment, and pesticide mass balance equations in paddy water and PSL for PTGs. Input data required for PCPF-C consist of parameters for describing the physical and chemical properties of the pesticide, pesticide application procedure, dimensions of the paddy field and canal compartments, and factors for water management in the fields (Table 11). The date of pesticide application in a PTG is assumed to be random and follows a normal distribution with a standard deviation  and a mean . Datasheet for water balance in the paddy compartment considers precipitation, irrigation, drainage, percolation, lateral seepage, ET, and paddy water depth. To create water management scenarios, two factors are simulated including the WHP after pesticide application and the average EWSD. The average EWSD concept may also be mentioned as the excess water storage capacity (EWSC) (Vu et al., 2006).

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

205

Table 11. Parameters used for the PCPF-C model validation Parameter

Unit

Valuea

Simulation period

days

46

g m⫺2 ha

0.105 Uniform (Obs. ±10%)c 0.43 Point 6/5/2004 Point

g m⫺2 ha

0.045 1.18

Pesticide application Mefenacet (MF) Application rate Treated area Application date Pretilachlor (PTC) Application rate Treated area Application date (Normal distribution) + Mean + Standard deviation + Day of start application

Distributionb

Uniform (Obs. ±10%)c Point

2.76 Point 0.95 Point 8/5/2004 Point

Pesticide parameter

Uniform (Obs. ±10%)c

Rainfall

Uniform (Obs. ±10%)c

Water management in paddy field Hwmind Hwmaxd PERCd LSEEPd ETd Hgated WHPd

cm cm cm cm cm cm days

Paddy catchment and hydrology in canal Total paddy field area Length of received canal Width of the canal Discharge in canal Water depth in canal

ha m m L s⫺1 m

Uniform (3, 4) Uniform (5, 6) Uniform (Obs. ±10%)c Uniform (Obs. ±10%)c Uniform (Obs. ±10%)c Normal (5.2, 14.1) Uniform (3, 4) 5.32 320 0.5

Point Point Point Uniform (Obs. ±10%)c Uniform (Obs. ±10%)c

aThe

observed data were used in the deterministic model validation. were used in MCS for the probabilistic model validation. To describe the distributions, the minimum and maximum values are used for uniform distribution, mean, and standard deviation for normal distribution. cMaximum and minimum values equal observed data plus and minus 10%, respectively. dH max and H min are the maximum and minimum paddy water depths; PERC, LSEEP, and ET w w are the average value of percolation, lateral seepage, and ET in paddy plot; Hgate is the height of the drainage gate of the paddy plot from paddy soil; and WHP is the duration of water holding period. bDistributions

The model program which is coded by Visual Basic for Application in Microsoft Excel provides the pesticide losses into the canal water through surface drainage and lateral seepage and also daily pesticide concentrations in canal water. For further environmental risk assessment, Monte Carlo simulation (MCS), a widely used method for probabilistic assessment and uncertainty analysis in pesticide fate modeling (Dubus et al., 2002), is incorporated into the PCPF-C modeling process. The method involves random sampling from the

206

H. Watanabe et al.

distribution of inputs and successive model runs until the model obtains a stable statistical distribution of outputs. Therefore, the PCPF-C simulation can provide an insight into the level of confidence which should be given to model prediction and can also provide useful information for the pesticide risk assessment and the decision making for the application of possible mitigation practices for minimizing pesticide discharges from the paddy field to the adjacent surface water systems. 7. RISK MANAGEMENT FOR REDUCING PESTICIDE LOSSES INTO AQUATIC ENVIRONMENTS IN ASIAN MONSOON CLIMATE 7.1. EWSD and WHP This section will focus on the mitigation aspects which control pesticide runoff from paddy fields. Since the typical monsoon climate in Japan is similar to that of several other Asian countries, the practices recommended in this section may be applicable to paddy field management in other Asian countries such as Korea and Taiwan. As indicated previously, pesticide concentrations are considerably higher at the earlier period after pesticide application and the management of surface water in this period is crucial for controlling pesticide runoff (Watanabe and Takagi, 2000a; Inao et al., 2003; Vu et al., 2004; Watanabe et al., 2006a,b,c). As reported in those studies, the CI scheme, which is also referred to as spillover irrigation or flow-through irrigation, seems to result in significant pesticide losses especially in the earlier period after the pesticide application. Although this CI scheme is often practiced for saving labor for manually irrigated paddy fields, farmers should be discouraged to use this irrigation practice due to increased environment risks. Table 12 indicates from previous studies that pesticide losses when CI schemes were applied ranged from 12% to 60% of the total applied pesticide mass depending on the pesticide used. Note that the amount of the irrigation input differs among the different experiments. As indicated in the previous section by PCPF-C simulation, EWSD in a paddy plot is an important factor controlling pesticide discharge through paddy water Table 12. A summary of pesticide losses from paddy fields where continuous irrigation and drainage managements were applied Pesticide Mefenacet Bensulfuron methyl Mefenacet Simetryn Molinate Thiobencarb Simetryn

Year of study 2001 2001 2003 2003 1996 2003 1996

Reference Watanabe et al. (2006a) Watanabe et al. (2006a) Watanabe et al. (2007) Watanabe et al. (2007) Inao et al. (2001) Watanabe et al. (2007) Inao et al. (2001)

Pesticide loss as % applied 38 49 35 37 41 12 60

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

207

100%

40 Frequency 35

80%

Frequency

30

Cumulative probability %

25

60%

20 40%

15 10

20%

5 0 -5

-4

-3

1 2 -2 -1 0 Excess Water Storage Depth (cm)

3

4

5

0%

Fig. 21. Frequency distribution of excess water storage depth (EWSD) in 296-surveyed paddy plots in a 100 ha watershed.

runoff following significant rainfall events. The results of a field survey regarding the EWSD conducted in 2005 are shown in Figure 21. The average value of EWSD was 0.5 cm, and the negative values result in paddy water runoff. Actually, the runoff was observed in 113 plots over a total of 296 surveyed plots. The percentage of the plot which had the EWSD of less than 1 cm was about 69% and that of less than 2 cm was more than 90% of the total. Smaller EWSD leads to more paddy water runoff hence pesticide runoff following rainfall events. In order to assess the importance of the EWSD for controlling pesticide runoff losses, prescribed scenarios for CI scheme and water managements with different EWSD were evaluated using PCPF-1 simulations for the pesticide losses from paddy fields as shown in Figure 22 (Watanabe et al., 2005). The height of the drainage gate was assumed to be 6 cm. Intermittent irrigation and drainage was practiced to keep water depth between 2–4 cm (i.e. irrigation initiates at the water level of 2 cm and ceases at 4 cm, EWSD = 2 cm), 3–5 cm (EWSD = 1 cm), and 4–6 cm (EWSD = 0). Pesticide fate parameters used for these scenarios were the same as in the case of mefenacet as reported by Watanabe and Takagi (2000c). Paddy water management practice from 2 to 4 cm corresponding to the water storage depth of 2 cm gave less herbicide runoff since it could store water for rainfall events up to 2 cm day⫺1 (Figure 22). Therefore, water management with 2 cm EWSD appears to be a good water management in the paddy field for preventing excessive pesticide runoff. Although practicing intermittent irrigation and drainage scheme, significant pesticide losses may occur without appropriate excess water storage following major rainfall events (Figure 22). Water holding after the pesticide application is one of the most effective management practices for preventing pesticide runoff from paddy fields into aquatic environments. In California, rice growers are required to hold paddy water following the pesticide application for various periods (up to 30 days in case of

208

H. Watanabe et al. 7 6

8

5 6

4

4

3

Rain (cm)

( % applied)

Herbicide runoff

10

2 2

1

0

0 1

8

15

22

29

36

43

50

57

64

Days after application Rain

EWSD = 0cm

EWSD = 1cm

EWSD = 2cm

35

7

30

6

25

5

20

4

15

3

10

2

5

1

0 1

8

15

Rain WH-10

22 29 36 43 Days after application

50

57

64

Rain ( cm)

( % applied)

Herbicide runoff

Fig. 22. Herbicide loss by different water management schemes with EWSD of 0, 1, and 2 cm (Watanabe et al., 2005).

0

Con. Dr

WH-4

WH-21

WH-30

Fig. 23. Herbicide loss in the paddy field corresponding to different WHPs of 0 (Con.Dr), 4, 10, 21, and 30 days (Watanabe et al., 2005).

granular thiobencarb) depending on the active ingredient (Newhart, 2002). For standard safe use of herbicide application in Japan, paddy WHP of 3–4 days is recommended. Paddy water depth during the WHP was suggested to be 5 cm. However in Japan, the problem seems to lie on the fact that water management instructions are available for farmers only in the pesticide label without any effective extension program. PCPF-1 simulations were used again to elucidate the importance of the management of the WHP. The effect of the duration of the WHP after pesticide application on pesticide losses from paddy fields via runoff are presented in Figure 23 (Watanabe et al., 2005). Prescribed scenarios with the same mefenacet parameters as above were assigned for CI scheme with WHP of 0 day (Con.Dr), 4 days (WH-4),

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

209

10 days (WH-10), 21 days (WH-21), and 30 days (WH-30). Herbicide loss was more than 35% of the applied mass when continuous irrigation and drainage without WH was applied. In case of WH-4, which is the standard scenario in Japan, herbicide loss was still 25% of the applied mass. Obviously, the longer the WHP, the smaller the herbicide loss becomes. In the case of WH-30, a scenario similar to the scenario found in California, the total herbicide loss was minimized to 3% of the applied mass. 7.2. Good agricultural practices for reducing pesticide losses from the paddy fields In order to control pesticide runoff from the paddy fields, several GAPs have been derived from previous studies. Watanabe et al. (2006a) demonstrated that the intermittent irrigation scheme requires less irrigation water and reduces the paddy field runoff. In addition, increasing excess water storage with higher paddy drainage gate (7.5 cm from paddy soil) prevented paddy water runoff after heavy rainfall events. Also the above simulation exercise with PCPF-1 has demonstrated the benefit of higher drainage gate to increase the excess water storage. Using AI system has been proven to be the recommended agricultural practice for both saving water and maintaining water quality (Watanabe et al., 2006a). The above simulations also supported the conclusion that the longer the WHP, the smaller the herbicide losses become due to runoff. As indicated in Figure 23, current Japanese WHP instruction may lead to significant pesticide losses and application of longer WHP is recommended. Dissipation of the herbicides in paddy water exhibits half-lives (DT50) of 1.9– 4.5 days and DT90 (90% mass dissipation) of 7.8–11.3 days (Watanabe et al., 2006a). Ishii et al. (2004) suggested that increasing WHP from 3 days to 7 days would reduce herbicide concentrations from 1/2 to 1/10 of the maximum concentrations. Therefore, a longer holding period, based on DT90 up to 10 days, is advisable for controlling herbicide runoff. With proper extension efforts and regulations through the water holding requirement for the rice pesticide management program, the state of California reduced rice pesticide concentrations in surface water systems to nearly no detectable level (Newhart, 2002). In conclusion, the recommended GAPs for controlling pesticide runoff from the paddy field are: (1) application of the intermittent irrigation scheme with high drainage gate and considerable excess water storage (2 cm or more) and (2) application of longer WHP, at least 10 days. These GAPs are recommended to be put into action for the extension project and also mentioned in the design of the irrigation–drainage system and the agricultural land consolidation and maintenances for the paddy field in Japan. 8. SUMMARY AND CONCLUSIONS In this chapter, pesticide risk assessment in rice paddy areas from a Japanese perspective has been discussed. Earlier sections provided an introduction into Japanese pesticide use and the pesticide registration procedures which were

210

H. Watanabe et al.

recently revised. Later part of the chapter focused on monitoring and modeling aspects of pesticide fate and transport in rice paddies as well as on the pesticide risk management for controlling pesticide losses from paddy fields. Section 1 to 4. Pesticide use in Japan and the new pesticide registration scheme: Since paddy fields cover more than half of the agricultural land in Japan, rice production seems to be the major non-point source of pesticide pollution. The recent annual production has been maintained at about 90,000 tons. As far as rice herbicides are concerned, the industry has been developing pesticide products for laborsaving rice production. The popular herbicide formulations have been used as one-shot or single-application products containing multiple active ingredients covering a wide spectrum of the various weeds so that farmers can apply them only once in the growing season. In addition, other formulations are available such as solid type granules, liquid type emulsion concentrates and suspension concentrates, large packed herbicides, and so on for laborsaving purposes. For assuring human health as well as environmental safety, pesticides are regulated by the MAFF, the MOE and the MHLW. From 2005, a new pesticide registration scheme was imposed in order to perform a realistic evaluation. The risk assessment is performed by comparing the AEC for three groups of organisms (fish, crustacean, and algae) and PECs for rice pesticides obtained at three tiers (numerical computation, lysimeter test, and plot-scale examination). Section 5. Monitoring pesticide fate and transport: Pesticide fate and transport monitoring specifically in monsoon Asian climate using paddy lysimeters, paddy plot experiments, watershed- and basin-scale monitoring are presented. Typical procedures included monitoring water balance such as rainfall, irrigation, drainage, percolation, seepage, paddy water depth, and hydrology in canal–river networks and also monitoring of pesticide concentrations in paddy water and in canal–river network for watershed and basin level. Also, the impact of important factors such as timing and amount of rainfall events, EWSD of the fields and WHP to pesticide runoff from paddy plot and watershed are examined. Section 6. Modeling pesticide fate and transport: Mathematical models have been developed for the simulation of fate and transport of rice pesticide at the paddy scale and watershed scale in Japan. Two series of simulation models based on PADDY and PCPF models were presented in this section. PADDY model was primarily developed for assisting pesticide regulation and registration process by predicting environmental concentrations of pesticides. In contrast, the PCPF model has been used mainly for developing and evaluating field management practices, such as irrigation and drainage control and the soil management in order to control pesticide runoff and seepage losses. The PADDY model series consists of PADDY, PADDY-2 and PADDY-Large, in which the former two provide PECs for pesticides in paddy water and soil at paddy field scale while the latter provides PECs in streams at the watershed scale. These three models have been validated with observed data from laboratory experiments and a field monitoring study. The PCPF model series includes PCPF-1, PCPF-SWMS, and PCPF-C models. PCPF-1 simulates pesticide dissipation in paddy water and surface soil, and

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

211

PCPF-SWMS simulates pesticide dissipation in the paddy soil profile. PCPF-C simulates pesticide fate and transport in a catchment scale and provides pesticide concentrations in drainage canals and streams as well as its losses from paddy fields. PCPF-1 has been used for evaluating water management practices, and PCPF-C has been applied for the probabilistic assessment of the best water management practices for controlling pesticide runoff from paddy fields. Section 7. Risk management for controlling pesticide losses from paddy fields to the aquatic environment in Asian monsoon climate: This section discussed applicable management practices for controlling pesticide losses associated with paddy rice production. Good Agricultural Practices (GAPs) for pesticide runoff control have been evaluated through monitoring results and PCPF model simulations. The recommended GAPs for controlling pesticide runoff from paddy fields are (1) application of intermittent irrigation scheme with high drainage gate and considerable excess water storage (2 cm or more) in the paddy field and (2) application of longer WHP, at least 10 days. These GAPs are recommended to be put into action for the extension project and also to be mentioned in the design of the irrigation–drainage system and the agricultural land consolidation as well as the maintenances for the paddy field system in Japan. In conclusion, Japan has been one of the major pesticide users in the world. Although having a long history in rice cultivation, the pesticide exposure assessment for paddy rice production received less attention compared with EU and US. Applications of up-to-date techniques and the development of realistic assessment procedures under specific climatic conditions as well as mitigation management practices for controlling pesticide contamination are important for an environmental-friendly rice production. Through the international cooperation and research exchanges, advances in pesticide risk assessment for rice paddies in Asian region and other rice-growing areas in the world would contribute to sustainable rice production. REFERENCES Dubus, I. G., Brown, C. D., Beulke, S. and Turner, N. L. (2002). Uncertainty and probabilistic approaches to pesticide fate modeling, The Department for Environment, Food & Rural Affairs, DEFRA project PL0548, No. JA3755E. Fajardo, F., Takagi, K., Ishizaka, M. and Usui, K. (2000). Pattern and rate of dissipation of pretilachlor and mefenaset in plow layer and paddy water under lowland field conditions – A 3 year study. J. Pest. Sci. 25, 94–100. Ganzelmeier, H., Rautmann, D., Spangenberg, R., Streloke, M., Herrmann, M., Wenzelburger, H. J. and Walter, H. F. (1995). Studies on the spray drift of plant protection products; results of a test program carried out throughout the Federal Republic of Germany. Blackwell Wissenschafts – Verlag BmbH, Berlin. Green Japan, Recent situation of the registered pesticides (Nouyaku no touroku joukoyou), (2005). http://www.greenjapan.co.jp/greenjapan.htm, Access date November 20, 2006. Hasegawa, S. and Tabuchi, T. (1995). Well facilitated paddy fields in Japan. In: S. Hasegawa and T. Tabuchi (Eds), Paddy Fields in the World. The Japanese Society of Irrigation, Drainage and Reclamation Engineering, Tokyo. Hatakeyama, S., Fukushima, S., Kasai, F. and Shiraishi, H. (1992). Assessment of the overall herbicides effects on algal production in the river. Jap. J. Limnol. 53(4), 327–340.

212

H. Watanabe et al.

Hatakeyama, S., Fukushima, S., Kasai, F. and Shiraishi, H. (1994). Assessment of herbicide effects on algal production in the Kokai River (Japan) using model stream and Selenastrum bioassay. Ecotoxicol. 3, 143–156. Hatakeyama, S., Fukushima, S., Kasai, F., Shiraishi, H. and Uno, S. (1997a). Joint effects on algal production in rivers. Ecological. Chem. 6, 45–52. Hatakeyama, S., Shiraishi, H. and Uno, S. (1997b). Overall pesticide effects on growth and emergence of two species of Ephemeroptera in a model stream carrying pesticide-polluted river water. Ecotoxicol. 6, 67–180. Hayakawa, Y. (2005). Amendment of pesticide registration program considering eco-toxicological risk assessment. Abstract of the 30th annual meeting of the Pesticide Science Society of Japan, Tokyo, Japan. p. 31. Hoshino, T. (2001). Pesticide registration program. In: N. Motoyama (Ed), Nouyaku-gaku-jiten (p. 31). Asakura Shoten Co. Ltd., Tokyo, in Japanese. Hoshino, T. (2004). Environmental regulation. In: M. Ueji, A. Katayama, K. Nakamura, T. Hoshino and H. Yamamoto (Eds), Frontiers of Environmental Pesticide Science. Soft Science, Inc., Tokyo. Hosoya, S. (2001). Pesticide production. In: N. Motoyama (Ed), Nouyaku-gaku-jiten (p. 31). Asakura Shoten Co. Ltd., Tokyo. Inao, K., Ishii, Y., Kobara, Y. and Kitamura, Y. (2001). Prediction of pesticide behavior in paddy field by water balance on the water management using pesticide paddy field model (PADDY). J. Pestic. Sci. 26, 229–235. Inao, K., Ishii, Y., Kobara, Y. and Kitamura, Y. (2003). Landscape-scale simulation of pesticide behavior in river basin due to runoff from paddy fields using pesticide paddy field model (PADDY). J. Pestic. Sci. 28, 24–32. Inao, K. and Kitamura, Y. (1999). Pesticide paddy field model (PADDY) for predicting pesticide concentrations in water and soil in paddy fields. Pestic. Sci. 55, 38–46. Inoue, T., Ebise, S., Numabe, A., Nagafuchi, O. and Matsui, Y. (2002). Runoff characteristics of particulate pesticides in a river from paddy fields. Water Sci. Technol., 45(9), 121–126. Ishihara, S., Horio, T., Kobara, Y., End, S., Ohtsu, K., Ishizaka, M., Ishii, Y. and Ueji, M. (2005). ACS Symposium Series 899 Environmental Fate and Safety Management of Agrochemicals (p. 112). Washington, USA. Ishii, Y., Inao, K. and Kobara, Y. (2004). Dissipation of some herbicide in a flooded rice field and increase of water holding time after application of herbicide. Bulletin of National Institute of Agro-Environmental Science, Japan, No. 23, pp. 15–25 (in Japanese with English summary). JPPA (Japan Plant Protection Association) (2006). Report of test procedure for PEC evaluation, http://www.env.go.jp/water/dojo/noyaku/n_kentoukai/index.html. Access date November 20, 2006. Karpouzas, D. G., Cervelli, S., Watanabe, H. and Capri, E. (2006). Pesticide exposure assessment in rice paddies: A comparative study of existing mathematical models. Pest Manag. Sci. 62, 624–636. MAFF (2005). Annual agricultural statistics by the Ministry of Agriculture, Forestry, and Fishery, http://www.maff.go.jp/census/past/data/04-02/nsr/nsr02/nsr0225.xls. Access date November 20, 2006. MAFF (2006a) Japan, Regarding data to be appended to applications for registration of agricultural chemicals (provisional translation), http://www.acis.go.jp/eng/shinsei/13-3987.pdf. Access date November 20, 2006. MAFF (2006b). Guidelines for preparation of study results submitted when applying for registration of agricultural chemicals, http://www.acis.go.jp/eng/shinsei/annex.htm. Access date November 20, 2006. Maru, S. (1991). Study on the behavior and fate of pesticide in aquatic environment. Special Bulletin of the Chiba Prefecture Experimental Station, No. 18, in Japanese with English summary. Ministry of the Environment (MOE), Japan. (2006). Water quality advisory level for pesticides in surface water, http://www.env.go.jp/water/dojo/noyaku/law_data/f406kansuido0086.htm. Access date November 20, 2006, in Japanese. Mitobe, H., Ibaraki, T., Tanabe, A., Kawata, K., Sakai, M. and Kifune, I. (1999). Change of pesticide concentrations in river water surrounding paddy areas. J. Environ. Chem. 9, 311–320.

Pesticide Exposure Assessment in Rice Paddy Areas: A Japanese Perspective

213

Mizutani, M. (1995). Water utilization. In: Rice Post-Harvesting Technology, The Food Agency (p. 79). Ministry of Agriculture, Forestry and Fisheries, Tokyo. Nakagawa, S. (1967). Suiden-Yosuiro-Chosakeikakuhou (Survey and planning method of water requirement in paddy fields). Hatachinogyo-Shinkokai (Association for Promotion of Upland Agriculture), Tokyo, in Japanese. Nakamura, K. (1992). Studies on Behavior and Fate of Pesticides in Soil and Aquatic Environment of Agricultural Lands (p. 46). Bulletin of the Saitama agricultural experiment station, Kumagaya, Japan. Nakamura, K. (2005). Calculation and problems of Predicted Environmental Concentration (PEC) for aquatic ecological risk assessment (p. 32). Abstracts of the 30th annual meeting of the Pesticide Science Society of Japan, Tokyo, Japan. Nakano, Y., Miyazaki, A., Yoshida, T., Ono, K. and Inoue, T. (2004). A study on pesticide runoff from paddy fields to a river in rural region-1: Field survey of pesticide runoff in the Kozakura River, Japan. Water Res. 38(13), 3017–3022. Neely, W. B. (1976). Mathematical models predict. Concentration-time profiles resulting from chemical. Spill in a river. Environ. Sci. Technol. 10, 72–76. Newhart, K. (2002). Rice pesticide use and surface water monitoring 2002, California Environmental Protection Agency, Department of Pesticide Regulation, Sacramento CA. Okamoto, Y., Fisher, R. L., Armbrust, K. L. and Peter, C. J. (1998). Surface water monitoring survey for bensulfuron-methyl applied in paddy fields. J. Pestic. Sci. 23, 235–240. Okamura, H., Piao, M., Aoyama, I., Sudo, M., Okubo, T. and Nakamura, M. (2002). Algal growth inhibition by river water pollutants in the agricultural area around Lake Biwa, Japan. Environ. Pollut. 117, 411–419. Rao, N. S. G. and Muralidhar, D. (1963). Discharge characteristics of weir of finite – crest width. La Houille Blanche - Revue Internationale de l’Eau. 5, 537–545. Replogle, J. A., Merriam, J. L., Swarner, L. R. and Phelan, J. T. (1983). Farm water delivery systems. In: M. E. Jensen (Eds), Design and Operation of Farm Irrigation System (p. 317). The American Society of Agricultural Engineers, St. Joseph, Michigan, USA. Shiraishi, H., Pura, F., Otsuki, A. and Iwakuma, T. (1998). Behaviour of pesticides in Lake Kasumigaura, Japan. Sci. Total Environ. 72, 29–42. Simunek, J., Sejna, M. and v. Genuchten, M. T. (1999). The HYDRUS-2D software package for simulating two-dimensional movement of water, heat, and multiple solutes in variably saturated media, Version 2.0, International Ground Water Modeling Center, Colorado School of Mines, Golden, Colorado. Sudo, M., Kunimatsu, T. and Okubo, T. (2002). Concentration and loading of pesticide residues in Lake Biwa Basin (Japan). Water Res. 36, 315–329. Takagi, K., Fajardo, F. F., Inao, K. and Kitamura, Y. (1998). Predicting pesticide behavior in a lowland environment using computer simulation. Rev. Toxicol. 2, 269–286. Takagi, K., Si, Y. and Iwasaki, A. (2004). Pedo-climate variations and microbial activity in relation to pesticide degradation in unsaturated subsoils of Japanese Andosol. Final Program and Abstracts of the 4th International Symposium on Environmental Aspects of Pesticide Microbiology, Thessaloniki, Greece. p. 79. Takagi, K. and Takanasi, S. (2003). Proc. of 3rd International Conference on Contaminants in the Soil Environment in the Australasia-Pacific Region (p. 50). Beijing, China. Takagi, K., Watanabe, H. and Ishizaka, M. (2001). Leaching of two amide herbicides in andosol upland by using large scale lysimeter with an edge flow collection device (p. 130). Abstracts of the 26th annual meeting of the Pesticide Science Society of Japan, Osaka, Japan, in Japanese. Takeshita, T. and Noritake, K. (2001). Development and promotion of laborsaving application technology for paddy herbicides in Japan. Weed Biol. Manag. 1, 61–70. Tournebize, J., Watanabe, H., Takagi, K. and Nishimura, T. (2004). New coupled model of pesticide fate and transport in paddy field. The proceeding of the conference on challenges and opportunities for sustainable rice-bases production system (pp. 497–507). Torino, Italy. Tournebize, J., Watanabe, H., Takagi, K. and Nishimura, T. (2006). Coupled PCPF-1 and SWMS2D Model to simulate pesticide fate and transport in paddy field: 1. Tracer Validation. Paddy Water Environ. 4, 39–51.

214

H. Watanabe et al.

Vu, S. H., Ishihara, S. and Watanabe, H. (2006). Exposure risk assessment and evaluation of the best management practice for controlling pesticide runoff from paddy fields. Part 1: Paddy watershed monitoring. Pest Manag. Sci. 62(12), 1193–1206. Vu, S. H., Ishihara, S., Watanabe, H., Ueji, M. and Tanaka, H. (2004). Monitoring pesticide fate and transport in surface water in Japanese paddy field watershed. Proc. of the conference on challenges and opportunities for sustainable rice-bases production system (pp. 509–521). Torino, Italy. Vu, S. H., Watanabe, H. and Takagi, K. (2005). Application of FAO-56 for evaluating evapotranspiration in simulation of pollutant runoff from paddy rice field in Japan. Agr. Water Manag. 76, 195–210. Watanabe, H., Kakegawa, Y. and Vu, S. H. (2006a). Evaluation of the management practice for controlling pesticide runoff from paddy fields using intermittent and spillover irrigation schemes. Paddy Water Environ. 4, 21–28. Watanabe, H., Nguyen, M. H. T., Komany, S., Vu, S. H., Asami, Y., Phong, T. K. and Tournebize, J. (2006b). Application of ELISA in pesticide monitoring to control runoff of bensulfuronmethyl and simetryn from paddy fields. J. Pestic. Sci. 31, 123–129. Watanabe, H., Nguyen, H. M., Komany, S., Vu, H. S., Thai, K. P., Tournebize, J. and Ishihara, S. (2007). Effect of water management practice on pesticide behavior in paddy water. Agr. Water Manag. 88, 132–140. Watanabe, H. and Takagi, K. (2000a). A simulation model for pesticide concentrations in paddy water and surface soil. I. Model validation and application. Environ. Technol. 21, 1393–1404. Watanabe, H. and Takagi, K. (2000b). A simulation model for pesticide concentrations in paddy water and surface soil. I. Model development. Environ. Technol. 21, 1379–1391. Watanabe, H. and Takagi, K. (2000c). Simulation of pesticide concentrations in paddy field by pcpf-1 model - in case of mefenacet and its control of runoff and leaching (p. 36). Proc. of the XIV Memorial CIGR congress. Tsukuba, Japan. Watanabe, H., Takagi, K. and Vu, S. H. (2006c). Simulation of mefenacet concentrations in paddy field by improved PCPF-1 model. Pest Manag. Sci. 62, 20–29. Watanabe, H., Vu, S. H., Tournebize, J., Nguyen, M. H. T., Komany, S., Phong, T. K., Thai, H. Q., Ishihara, S. and Takagi, K. (2005). Monitoring and modeling of pesticide fate and transport in paddy field; challenges for reducing environmental risk (pp. 69–82). Proc. of the 2nd International Conference of Japan–Korea Research Cooperation. Tsukuba, Japan. Watanabe, T. (1999). Irrigation water requirement. In: M. Mizutani and S. Hasegawa (Eds), Advanced Paddy Field Engineering (p. 31). The Japanese Society of Irrigation, Drainage and Reclamation Engineering, Tokyo, Japan.

Pesticide Risk Assessment in Rice Paddies: Theory and Practice E. Capri and D.G. Karpouzas (editors) © 2008 Elsevier B.V. All rights reserved

Chapter 9

Higher Tier Exposure Assessments in Rice Paddy Areas: An American Perspective Amy M. Ritter and W. Martin Williams Waterborne Environmental, Inc., 897-B Harrison Street, Leesburg, VA, USA Contents 1. Overview of rice production in the United States 2. United States Environmental Protection Agency tiered process for risk assessment 3. Tier 1 assessment 4. Tier 2 assessment (RICEWQ/ EXAMS) 4.1. RICEWQ/ EXAMS model processes and algorithms 4.1.1. Rice paddy model 4.1.2. Pond bayou receiving water model 4.2. Description of scenarios 4.2.1. Location 4.2.2. Agronomic practices 4.2.3. Weather 4.2.4. Soils 4.2.5. Receiving water 4.3. Probabilistic assessment 5. Tier 3 assessment (RICEWQ/ RIVWQ watershed) 5.1. RIVWQ model processes and algorithms 5.2. RICEWQ/ RIVWQ watershed scenario 5.2.1. Characterization of rice paddies 5.2.2. Application 5.2.3. Loadings to receiving water 5.2.4. Characterization of receiving water 5.3. Probabilistic assessment 6. Tier 4 assessment (monitoring/mitigation) 6.1. Approach and example 6.2. Monitoring 6.3. Validation 6.4. Mitigation 7. Discussion 7.1. Tier 2 7.2. Tier 3 8. Disclaimer References

215 217 218 219 219 219 220 221 221 222 222 224 224 225 226 227 228 228 228 228 228 229 229 229 230 231 231 233 233 234 235 235

1. OVERVIEW OF RICE PRODUCTION IN THE UNITED STATES Rice production began in the United States (U.S.) in 1685 (USA Rice Federation, 2006) and has grown to be one of the highest harvested crops in the U.S. In 2002, 3,197,641 acres of rice were harvested making it the 11th highest harvested crop in the U.S. (USDA-NASS, 2004a,b). The six states with the highest amount of

216

A. M. Ritter and W. M. Williams

Table 1. States with harvested acres of rice Harvested rice acreage State Arkansas Louisiana California Mississippi Texas Missouri Florida

2002

1997

1,506,615 538,518 531,314 233,447 204,069 167,716 14,108

1,404,942 583,734 514,281 235,283 287,822 124,258 10,691

Fig. 1. Rice harvested acres in the United States in 2002. Source: USDA-NASS (2004b).

harvested rice include: Arkansas, Louisiana, California, Mississippi, Texas, and Missouri (Table 1, Figure 1). As with other crops, rice fields are vulnerable to weeds, insects (such as rice water weevil and rice leafminer), and diseases (e.g. aggregate sheath spot of rice, rice blast, or stem rot). In 1997, 25 different pesticides were used on rice as herbicides, insecticides, and fungicides (USGS, 1997). Rice production presents a unique problem with respect to agrochemical exposure because of water management and proximity of cropland to surface water bodies. Rice agriculture also

Higher Tier Exposure Assessments in Rice Paddy Areas: An American Perspective 217

provides a unique environment for many species of birds, mammals, amphibians, and reptiles and can provide a wetland habitat and critical flyways during the winter months for migratory birds and shorebirds (USEPA, 2006). Winterflooded rice fields can provide rest and forage for waterfowl. Protecting these “nontarget” species from toxic concentrations from the use of pesticides is important to the U.S. Environmental Protection Agency (USEPA). 2. UNITED STATES ENVIRONMENTAL PROTECTION AGENCY TIERED PROCESS FOR RISK ASSESSMENT The USEPA uses a tiered process for conducting risk assessments for crop protection chemicals. Early tiers involve the use of conservative assumptions and readily available data. Higher tiers integrate additional information and sophistication as needed to address variability and reduce uncertainty in the risk assessment (Figures 2 and 3). The tiered process allows continual refinement in problem formulation and directs resources expenditure only to those areas that warrant additional evaluation. The tiered process has been adopted and recommended for risk assessments by scientific and regulatory communities in both the U.S. and Europe (FOCUS, 1997; ECOFRAM, 1999; Morris and Kendall, 2000; USEPA, 2000; Hart, 2001). For most crops, Tier 1 consists of using the GENEEC2 and FIRST models to evaluate ecological risk and drinking water risk, respectively (USEPA, 2001a,b). If the pesticide does not pass the Tier 1 risk assessment, then Tier 2 is conducted. Tier 2 involves running the Pesticide Root Zone Model (PRZM) with the Exposure Analysis Modeling System (EXAMS) with Environmental Fate and Effects Department (EFED) standard crop and standard water body scenarios

Fig. 2. Progressive refinement of exposure analysis. Source: ECOFRAM (1999).

218

A. M. Ritter and W. M. Williams Realistic (Predictive)

Unachievable

Tier 4

Tier 3

Tier 2

Conservative (Protective)

Undesirable

Tier 1

Simple (Data Scarce)

Complex (Data Rich)

Fig. 3. Tiered process (step process). Source: USEPA (2000a).

(USEPA, 2003). For Tiers 1 and 2 ecological and drinking water risk assessments, the exposure concentrations in aquatic environments are simulated using USEPA’s standard pond and standard index reservoir scenarios, respectively. The standard pond is configured as having a 1-ha surface area and a depth of 2 m. The standard index reservoir is configured as having a 5.26-ha surface area and 2.74-m depth. For Tier 2, residue loadings into the pond and reservoir are derived from predicted mass in runoff and erosion from PRZM simulations for a 10-ha field and a 172.8-ha field, respectively. Spray drift is also included as a mass loading to the EXAMS model. However, these models are not appropriate for rice and at this time, USEPA only has Tier 1 for evaluation of risk of pesticides used on rice crops. 3. TIER 1 ASSESSMENT Currently, USEPA does not have a standard Tier 1 rice model. However, two separate approaches have been found in Reregistration Eligibility Decisions (REDS). One approach assumes that the pesticide is applied uniformly to a rice paddy. The model calculates an estimated environmental concentration (EEC) in the water column that could be released from the rice paddy. Adsorption is included in the model (1 cm of sediment). However, degradation is not calculated in the model. The calculation for estimating pesticide concentrations in water from pesticide application to rice paddies is the following equation: EEC ⫽

10 9 ⫻ M T VT ⫹ msed ⫻ K d

in which, MT is the total mass of pesticide applied in kg/ha, VT the volume of water in the paddy (1,067,000 L/ha for a paddy 10.16 cm deep and pore space in

Higher Tier Exposure Assessments in Rice Paddy Areas: An American Perspective 219

a 1 cm interaction zone), msed the mass of the sediment in the top 1 cm in L/kg, and Kd the sorption coefficient in kg/L, and 109 is the conversion factor from kilograms to micrograms (USEPA, 2004a). There is also a more refined screening level model for ecological risk assessment developed by EFED that includes degradation and holding times in the rice paddy. This version has three scenarios (Arkansas, California, and Louisiana) with added site-specific characteristics for rice agriculture (USEPA, 2004a, 2006). The refined model incorporates adsorption to sediment, a pre-flood aerobic degradation period, and a post-flood degradation period. A single EEC that represents acute and chronic exposure is calculated for each scenario. The EEC approximates the concentration at the point of release into a receiving water body and is used for risk assessment. The concentration does not represent exposure in downstream water bodies. 4. TIER 2 ASSESSMENT (RICEWQ/EXAMS) 4.1. RICEWQ/EXAMS model processes and algorithms An analogous approach to the PRZM/EXAMS standard pond has been used (USEPA, 2006) linking the Rice Water Quality Model, RICEWQ (Williams et al., 1999) with the EXAMS (Burns, 2004). 4.1.1. Rice paddy model

Existing pesticide transport models (e.g. PRZM) are not designed to simulate the flooding conditions, overflow, and controlled releases of water that are typical under rice production (Suárez, 2005). One of the most significant modeling challenges was characterizing the surface-water hydrology and water management associated with rice production. Therefore, the fate and transport model, RICEWQ, was used to simulate water and chemical mass balance associated with these unique governing processes. The water balance algorithms in RICEWQ account for precipitation, evaporation, seepage, irrigation, releases and overflow from various paddy outlet configurations, and controlled drainage prior to harvest (Williams et al., 1999). Important governing processes for pesticides use on rice include: foliage interception during application, dilution, partitioning between water and sediment, and degradation in water, foliage, and sediment (Figure 4). Degradation is represented by first-order decay rates for metabolism, hydrolysis, and photolysis in water and decay rates for saturated and unsaturated soil. Crop growth algorithms are similar to those in PRZM (Suárez, 2005) and allow for pesticide interception, decay on foliage, and washoff. Chemical loss to drift and volatilization can also be represented in the model. RICEWQ operates under a sub-daily time step. RICEWQ is able to simulate different draining, flooding, and post-harvest practices. RICEWQ is also capable of simulating metabolites, if needed. The model also contains an algorithm to allow partitioning of chemical residues between the water column and bed sediment.

220

A. M. Ritter and W. M. Williams Application

Volatilization Drift

Crop Interception Wash off

Outflow (overflow, drainage)

Biological Decay, Chemical Decay, Transformation,

c at i qu e A as Ph

and Sorption to Suspended Sediments Seepage (loss)

Resuspension Diffusion Settling Adsorption

Biological and

t en im d Se ase Ph

Chemical Decay

Fig. 4. Schematic of RICEWQ processes.

The model has been endorsed by the European community (Med-Rice, 2003) and has been validated with a number of field and watershed applications (Capri and Miao, 2002; Miao et al., 2003a,b; Warren et al., 2004). USEPA has been evaluating this model as a potential tool for Tier 2 assessments. 4.1.2. Pond bayou receiving water model

EXAMS was selected to evaluate the fate of pesticide residues in aquatic settings because of preferences for its use by USEPA’s Office of Pesticide Programs (USEPA, 2004b). EXAMS combines a chemical fate and transport model with a steady-state hydraulic model to simulate the following processes: advection,

Higher Tier Exposure Assessments in Rice Paddy Areas: An American Perspective 221

dispersion, dilution, partitioning between water, biota, and sediment, and degradation in water, biota, and sediment (Burns, 2004). Model geometry is based on the segment/compartment approach in which the simulated system is divided into a number of discrete volumes that are connected by advective and dispersive fluxes. RICEWQ output was modified to allow the creation of EXAMS transfer files. 4.2. Description of scenarios 4.2.1. Location

Candidate regions for modeling were identified from the agricultural census for rice production as shown in Figure 1. Three regions were chosen based on geographical clustering of rice production and distinctions in agronomic practices in each region. Rice production in California is concentrated in the Sacramento and San Joaquin Valleys. The Gulf Coastal Plain includes southeastern Texas and southern Louisiana. The Mississippi Alluvial Plain consists of eastern Arkansas, western Mississippi, and northeastern Louisiana. The Gulf Coastal Plain and the Mississippi Alluvial Plain have similar rainfall, but different planting practices. Dry seeding is the predominant planting method in the Mississippi Alluvial Plain, while aerial water seeding is practiced in the Gulf Coastal Plain and California. The U.S. has been subdivided into Land Resource Regions (LRRs) based on similar climate, soils, and land use activities including vegetation and crop types. Many Major Land Resource Areas (MLRAs) make up LRRs. LRRs and MLRAs are important in statewide agricultural planning as well as interstate, regional, and national planning (USDA-NRCS, 2006). Using the three regions shown in Figure 1 and the agronomic practices, the predominant MLRAs were identified for rice to evaluate variability in climate and soil runoff potential as shown in Figure 5 (USDA-NRCS, 2006). Predominant Rice-Producing Regions in the United States

Selected Regions for Rice Scenarios Central California Mississippi Alluvial Plain Gulf Coastal Plain

Fig. 5. Location of Tier 2/3 rice scenarios.

N E

W S

222

A. M. Ritter and W. M. Williams

4.2.2. Agronomic practices

Paddies are characterized as “typical” in terms of spillway depths and irrigation requirements. Simulations are started without water in the paddies. The paddies are flooded the day before planting for the wet seed regions (Gulf Coastal Plain and California) and 1 month after planting for the dry seed region (Mississippi Alluvial Plain). Irrigation is regulated to maintain initially 2.5–5.0 cm (1–2 in.) depth in the paddy in the Gulf Coastal Plain then drained after 2 days; the paddy is then reflooded after 7 days to maintain a depth of 5.0–10.2 cm (2–4 in.). Once flooded, irrigation is regulated to maintain 5.0–15.2 cm (2–6 in.) depth in the Mississippi Alluvial Plain paddy. Irrigation is regulated to maintain initially 2.5–5.0 cm (1–2 in.) depth in the paddy in California and then raised to maintain a depth of 5.0–10.2 cm (2–4 in.) after 2 weeks. Outlet (overflow) depths are specified as 15.2 cm (6 in.), 20.3 cm (8 in.), and 40.6 cm (16 in.) in the Gulf Coastal Plain, Mississippi Alluvial Plain, and Central California, respectively. The outlet depths are set to 5 cm (2 in.) higher than the maximum irrigation level, except in California in order to mimic discharge paddies in California. In California the outlet depths were set higher because California may not allow any water to overflow into the receiving waters without first being held in a holding pond for 30 days after application. A brief description of the agronomic practices is shown in Table 2. 4.2.3. Weather

Selected MLRAs for the three regions were evaluated by critical season rainfall, which is defined herein as normal rainfall during the 3- to 4-month period following planting. Critical season rainfall was determined by examining monthly normal precipitation records for the weather station contained within the USEPA’s PIRANHA databases (Bird et al., 1992) associated with each identified MLRA. This information is presented in Table 3. The table provides the regional description, corresponding MLRA numbers, meteorological station number and

Table 2. Agronomic practice for the three regions Rice seeding practice

Agronomic practices description

Wet seed

Flood, seed, drain within 2 days, wait 7 days for germination, hold flood for approximately 90 days, drain Seed, wait 1 month, hold flood for about 90 days, drain Flood, seed, hold flood for approximately 90 days or more, drain

Dry seed Wet seed

Region

MLRA

Gulf Coastal Plain

150A, 150B, 152B, 151, 134

Mississippi Alluvial Plain Sacramento and San Joaquin Valleys

131, 134 17, 15, 18, 16

Region

MLRA

Weather station

Southeastern Texas and 150A, 150B, Port Arthur W12917 Southern Louisiana: 152B, 151, 134 Gulf Coastal Plain Eastern Arkansas, 131, 134 Little Rock W13963 Western Mississippi, and Northeastern Louisiana: Mississippi Alluvial Plain Central California: 17, 15, 18, 16 Sacramento W23232 Sacramento and San Joaquin Valleys Note: Shaded areas are the planting and harvest months.

Normal rainfall (cm) April

May

June

July

Aug.

Sept.

Oct.

10.6

11.7

11.4

14.2

13.6

14.6

9.5

13.4

13.9

9.6

8.7

8.3

10.0

7.9

3.6

1.0

0.3

0.1

0.2

0.8

2.5

Higher Tier Exposure Assessments in Rice Paddy Areas: An American Perspective 223

Table 3. Seasonal weather for rice modeling scenarios

224

A. M. Ritter and W. M. Williams

name, and normal monthly rainfall for the critical season. Data for April through October were included to correspond to the typical rice-growing season in the different regions. The highlighted (shaded) months represent the months of planting and harvest. 4.2.4. Soils

The National Resource Inventory (NRI) was used to identify candidate soils for the MLRA that best represents each region (USDA, 1994a). NRI is a survey of land use, prime farmland soils, soil erosion, and conservation practices. “Rice” (land use code 113) was used as a land use category to identify candidate rice soils. Soils were then ranked for decreasing runoff/overflow potential. Soil properties used to assess runoff potential were obtained from the U.S. Department of Agriculture’s soil property database, SOILS5 (USDA, 1994b). Soil properties representing the “midpoint” values were calculated as the average of the low and high values reported in SOILS5 for each series. Hydrologic Soil Group (USDA, 1972) was used as the primary ranking criteria, in which “D” soils were ranked above “C” soils, etc. Average organic matter was used for secondary ranking (lowest to highest). Average bulk density (low to high), pH (low to high), and acreage (high to low) were used for subsequent ranking. High runoff/overflow potential soils were identified for each of the three regions by selecting the 5th percentile soil in terms of cumulative acreage for the region (Table 4). 4.2.5. Receiving water

Characterization of pond bayou receiving water. Exposure concentrations in aquatic environments were predicted for receiving water systems by transferring discharge from the RICEWQ paddies to the EXAMS model. A rice “bayou” receiving water body was created using characteristics of USEPA’s standard pond. Table 4. Rice soils selected for rice modeling scenarios Region

MLRA

Southeastern Texas and Southern Louisiana: Gulf Coastal Plain Central

152B

Evadale

Silt loam

D

Eastern Arkansas, Western Mississippi and Northeastern Louisiana: Mississippi Alluvial Plain

134

Henry

Silt loam

D

San Joaquin

Sandy loam

D

California: Sacramento 17 and San Joaquin Valley

Series name

Surface texture

Hydrologic soil group

Higher Tier Exposure Assessments in Rice Paddy Areas: An American Perspective 225

The pond bayou receiving water body has a surface area of 2 ha and a depth of 1m (rather than the standard pond dimensions of 1 ha surface area and 2 m deep). The volume of the pond bayou is consistent with the volume of the standard pond. Each rice paddy simulation has an area of 10 ha, thus preserving the typical ratio of watershed area to receiving water volume used in USEPA Tier 2 standard scenario PRZM/EXAMS modeling simulations. Loadings to the pond bayou receiving water. EXAMS pesticide mass loadings were derived from RICEWQ simulations and spray drift. Five or 1% drift was simulated for aerial or ground applications, respectively. Water in the pond bayou is maintained at a constant depth of 1m, but a different base flow is simulated for each location. Average daily flow from the receiving water body was determined by modeling PRZM with local meteorological data following the same procedure used to calculate flows in index reservoir scenarios (USEPA, 2000b). Water from rice paddies is typically discharged into a permanent water body (stream or bayou) having a base flow rate. Base flows of 5.63, 4.76, and 0.89 m3/h were calculated for Mississippi Valley, Gulf Coast, and California scenarios, respectively. These flows are conservative because they are not affected by rainfall or discharge events. 4.3. Probabilistic assessment The USEPA uses the results of the 1-in-10 year probabilities in the risk assessments from the Tier 2 modeling. In order to calculate the upper 10th percentiles for each exposure duration concentration, frequency analyses are performed that evaluate how often specific levels of a pesticide concentration may be expected to occur. The analyses are conducted using the Weibull plotting distribution (Haan, 1977). Time weighted average concentrations are calculated for six exposure duration periods for each day (maximum initial, 96-h, 21-day, 60-day, 90-day, and annual average). The highest exposure duration concentration for each year is determined and these 30 values per duration period are ranked in descending order. Along with the Weibull plotting positions for each maximum annual exposure duration, the associated recurrence intervals for each year’s ranking are also derived. The Weibull plotting position represents the probability that a specific event will be equaled or exceeded in any given year. The equation is the following: Weibull plotting position ⫽

Rank ⫻100 n ⫹1

where rank is the plotting position (1–30 for 30 years of climate), n the number of years (30). After conducting the frequency analysis, the 10% probability values are determined for each exposure duration. USEPA regulates on the upper 10th percentile EECs (USEPA, 2004b). The 10th percentile EECs are expected once in every

226

A. M. Ritter and W. M. Williams 0.080 5-Yr

10-Yr

2-Yr

Concentration (ppb)

0.060

0.040

0.020

0.000 0

20

60

40

80

100

Exceedence Probability (%) Max. Initial

96-Hour

21-Day

60-Day

90-Day

Annual

Fig. 6. Frequency analysis on the annual maximum series.

10 years based on the 30 years of daily values generated from the modeling. In other words, the EECs have a 10% probability (10-year recurrence interval) of being equaled or exceeded in any given year. Figure 6 shows an example of the results of a frequency analysis on the annual maximum exposure duration concentrations. For chronic assessments, the USEPA typically uses the 21-day average EEC for aquatic invertebrates and the 60-day average is typically used for fish. 5. TIER 3 ASSESSMENT (RICEWQ/RIVWQ WATERSHED) The Tier 3 scenario described herein introduces additional refinements that are a more realistic representation of an ecosystem with multiple paddies draining along a dynamic and more realistic receiving water. Figure 7 shows a schematic of the watershed and Riverine Water Quality Model (RIVWQ) channel system. The model scenarios assume a 2400-ha watershed with 50% of the watershed treated with pesticide (over a 5-day window). The rice paddies drain into a 3.5 ha by 0.6 m deep stream with a 0.5 m/s velocity. The chemical loadings to the stream are assumed to occur from spray drift (1% for ground or 5% for aerial applications), paddy overflow, and draining the paddy. The base depth of the receiving water ranges from 0.52 m to 0.67 m and the width ranges from 1.8 m to 6 m. A watershed model combining the RICEWQ and the RIVWQ are suggested for Tier 3 ecological assessments. The RICEWQ model was described in the previous section. RICEWQ mass loadings are linked to the RIVWQ model to

Higher Tier Exposure Assessments in Rice Paddy Areas: An American Perspective 227

40 ha 40 ha

40 ha

40 ha

200 ha Rice Paddies

200 ha Rice Paddies 40 ha 40 ha

40 ha

40 ha 200 ha Rice Paddies

40 ha

40 ha

40 ha

40 ha

40 ha

40 ha

40 ha

Flow 200

100

40 ha

40 ha

500

400

300

600

40 ha

40 ha

40 ha 40 ha

40 ha

700

40 ha

40 ha 200 ha Rice Paddies

200 ha Rice Paddies 40 ha

40 ha

200 ha Rice Paddies 40 ha

40 ha

40 ha

40 ha Note: Represents a node. 1. 100 2. 40 ha Represents 20% of 200 ha. 3. Total Watershed Drainage Area = 2400 ha

Fig. 7. Model schematic of rice receiving water channel system.

predict concentrations in a receiving water body downstream from the rice paddies. To better simulate rice agriculture, scenarios were developed for each region (Gulf Coastal, Mississippi Alluvial Plain, and California) to represent a watershed. 5.1. RIVWQ model processes and algorithms In receiving streams, water and chemical mass balances were simulated with the river water quality model, RIVWQ (Williams et al., 2004). RIVWQ can accommodate tributary systems, non-uniform flow, and mass loadings anywhere along the model system. Model geometry is based on the link-node approach in which the simulated system is divided into a number of discrete volumes (nodes or junctions), which are connected by flow channels (links). Steady-state hydraulics are assumed that can change from time period to time period (i.e., any change in the hydraulic regime is assumed to be instantaneous throughout the system), which allows for flows to be calculated using the continuity equation alone and to neglect the momentum equation. Dynamic constituent transport is a combination of advective flows and dispersion processes. Dispersion processes, including constituent mixing as a result of backwater and flow reversals, are lumped together into a single diffusion coefficient. RIVWQ includes transformation of the parent to metabolites and the degradation of the metabolites. RIVWQ operates under a user-specified time step that must satisfy certain stability criteria.

228

A. M. Ritter and W. M. Williams

RIVWQ utilizes the same chemical algorithms as RICEWQ. A sediment transport routine that contains the ability to simulate bed scour is also included in this model. RIVWQ was selected over other water quality models such as EXAMS (Burns, 2004) and WASP-5 (Ambrose et al., 1991) because of its efficiency in data setup, computation, file management, and ability to simulate relevant governing processes at an adequate time step and scale for the watersheds being studied. 5.2. RICEWQ/RIVWQ watershed scenario 5.2.1. Characterization of rice paddies

The locations, soils, weather, and agronomic practices in RICEWQ are the same as the Tier 2 scenarios. The Tier 2 rice scenario assumes a single rice paddy draining or overflowing into a bayou. The Tier 3 rice scenario assumes a 2400-ha watershed with 50% of the watershed treated with pesticide. The rice watershed for each region consisted of six 400-ha paddy blocks each containing 200-ha of treated rice paddies. Agronomic events (planting, irrigating, draining, maturation, harvest, and pesticide applications) were assumed to occur over a 5-day window (40-ha rice paddies per 200-ha treated paddies in a paddy block). This assumption was made because it would be unrealistic for all farmers within a watershed to plant their rice on the same day. 5.2.2. Application

For each region, the application is assumed to occur over a 5-day period under the assumption that the total treated 1200 ha in the watershed would not be realistic (i.e., application lagged such that 20% of the area planted in rice would be applied on day 1, 20% would be applied on day 2, etc. through day 5). This amounts to 240 ha in the watershed treated per day for 5 days. 5.2.3. Loadings to receiving water

Pesticide releases from overflow, drainage, and drift are input as loadings to the receiving water. Additional mass due to drift may be added depending on the pesticide type of application. If the application is aerial then 5% drift or 1% drift if ground application. The drift loading is calculated from the total application rate times the surface area of the channel segments (i.e., applied rate, kg/ha ⫻ 0.05 drift fraction ⫻ 3.51 ha ⫽ drift loading, kg). Loadings to the receiving water predicted using RICEWQ are simulated separately for each application scenario. 5.2.4. Characterization of receiving water

All receiving waters are characterized as “typical” in terms of depths and widths based on previous experiences by the authors with rice agriculture, hydrology, and geomorphology. The receiving water channel is simulated to be a rectangular channel with a velocity of 0.5 m/s and a base depth ranging from 0.52 m (1.7 ft.)

Higher Tier Exposure Assessments in Rice Paddy Areas: An American Perspective 229

in the most upstream segment to 0.67 m (2.2 ft.) in the most downstream segment. When the paddies are drained, overflow, or runoff from precipitation, the channel increases in depth and starts to flow. Water discharged from the paddies is simulated using RICEWQ. Water runoff for non-paddy areas is simulated using PRZM. A “generic” soil series representative of hydrologic soil groups of C, D, and C are simulated for MLRAs 152B, 131, and 17, respectively, because these are the most predominant soil hydrologic groups for each MLRA. These hydrologic soil groups are used in the PRZM input files. The receiving water channel is divided into six segments, each 1500 m in length. The width of the segments ranges from 1.8 m in the headwater segment to 6 m in the most downstream segment. 5.3. Probabilistic assessment Upper 10th percentile EECs are calculated for various points in the watershed for the six exposure durations using procedures discussed in Section 4.3. Concentrations are generated for the most downstream point of the river and at the point where the rice paddies drain into the stream. 6. TIER 4 ASSESSMENT (MONITORING/ MITIGATION) ECOFRAM (1999) describes Tier 4 as generally involving broad reaching experimental or monitoring programs designed to definitively characterize key aspects of the toxicity or exposure profiles. Examples of Tier 4 programs include: (1) widespread monitoring; (2) detailed investigation of the efficacy of mitigation; (3) highly refined watershed evaluations and modeling; (4) benchmark modeling relative to existing chemical data; (5) modeling of population or ecosystem dynamics; and (6) microcosm or mesocosm studies. Options selected at Tier 4 depend entirely on the risk assessment issues that remain after Tier 3. Both Tiers 3 and 4 are intended to be highly flexible. Consultation between registrants and regulators is essential at this stage because of the extraordinary cost associated with the programs. 6.1. Approach and example The Tier 4 example herein incorporates aquatic monitoring studies of pesticide use on rice crops and in the receiving water along with modeling best management practices (BMPs). The approach to Tier 4 is shown in the schematic below (Figure 8). Aquatic monitoring data may be acquired by conducting a field study or using data collected by an agency such as the U.S. Geological Survey. Pesticide use can be obtained from surveying farmers, analyzing sales data, or state agency databases such as the CA Department of Pesticide Regulation’s Pesticide Use Reporting (PUR) database. After the environmental fate and transport models are calibrated, the models are used to estimate potential concentrations of residues under a range of climatological conditions at the study sites. BMPs may be

230

A. M. Ritter and W. M. Williams

Aquatic Monitoring

Pesticide Use Survey

Paddy Overflow Tributary Concentrations

Paddy Size Application Rates Application Dates

Model Setup/Sensitivity/Calibration/Verification RICEWQ In-paddy Dissipation Pesticide Runoff

RIVWQ Receiving Water Dissipation Watershed-wide Impacts

Model Application POST-PROCESSING Statistical Data Reduction Exposure Duration Concentrations

SCENARIOS Baseline Use Rate Mitigation

Fig. 8. Schematic of Tier 4 approach.

modeled to determine which mitigation option may be the most effective on lowering pesticide concentrations. An example of this higher tier type of study is presented in this section. The field studies in the example involve monitoring the application of the pesticide to paddies in rice growing regions such as Arkansas or Louisiana and monitoring the dissipation of the chemical in water and sediment samples collected from within the paddies, downstream ditches, and receiving streams. RICEWQ and RIVWQ models are used to estimate potential concentrations of residues under a range of climatological conditions at the study sites. 6.2. Monitoring The field studies involved monitoring the application and dissipation of residues at nine different sites in Louisiana and Arkansas where rice is grown. Each study system included a rice paddy and associated drainage ditch and receiving stream. Applications were made to each paddy according to label directions. Water and sediment samples were collected from within the paddies, ditches, and receiving streams periodically from the day prior to the first application until at least 30 days after the last application or until residues are no longer detected. Figure 9

Higher Tier Exposure Assessments in Rice Paddy Areas: An American Perspective 231

S-2 S-5 S-1

S-3 S-4

S-6

S-7 S-8

Rice paddies treated with pesticide

Fig. 9. Schematic of a watershed with rice and sampling points.

shows a schematic of the sampling points within the watershed along with the rice fields and streams. Along with collecting samples, it is important to collect weather data, soils data, paddy depths, and water management practices. Stream flows, velocities, and geometries at the sampling points are also valuable information to be able to model afterwards. 6.3. Validation The watersheds were selected for monitoring based on having high-density rice production in three geographically distinct areas – central Arkansas, eastern Arkansas, and southern Louisiana. Runoff and receiving water models (RICEWQ and RIVWQ, respectively) were selected based on their ability to simulate governing hydrologic and physicochemical processes associated with pesticide dissipation. Models were calibrated and verified using results from the field studies. The ability to reproduce the hydrology, water management, and pesticide dissipation (in water and sediment) for six rice paddies and three watershed systems indicated that the models would be effective in evaluating other climatological conditions in these systems. Results are not presented herein to preserve data confidentiality of the study sponsor. 6.4. Mitigation After model validation, additional simulation modeling was conducted to evaluate changes in water quality resulting from four different use scenarios in the three watersheds systems. The example mitigation scenarios represent a range of

232

A. M. Ritter and W. M. Williams

environmental conditions potentially associated with pesticide use in the Mississippi delta. The mitigation scenarios can address important factors such as climatological conditions, density of rice production and product use, pounds of pesticide applied, application dates, water management practices, and receiving water characteristics. The scenarios defined below considered variations in application rates, water storage, and irrigation control in the rice paddies: ●

●

●

●

Baseline scenario. Results of the validated model were used as the Baseline scenario. The Baseline scenario was designed to estimate concentrations of residues in receiving waters resulting from actual rates of pesticide applied during the growing season of the year monitored for each watershed. These conditions were then run under a range of climatological conditions; Low Use scenario. This scenario was designed to estimate concentrations of residues in receiving waters under lower application rates that might be applied under conditions of reduced disease pressure; High Use scenario. The High Use scenario reflects concentrations of residues in receiving waters estimated to occur at much higher rates than Baseline rates throughout the watershed; and Water Storage scenario. This scenario simulated the Baseline scenario application rates under water management practices designed to maximize water storage and minimize paddy overflow.

Each scenario was modeled using 20–36 years of historical climatology, representative of the region in which each watershed resides. Probability analyses were conducted on the maximum average 96-h, 21-day, and 100-day concentrations of residues predicted for each climatological year. 21-day concentrations in water (ppb) in downstream permanent stream areas predicted for 2, 5, and 10-year recurrence intervals are shown in Figure 10. SCENARIO RESULTS 21-Day Duration Concentration (mass/vol.)

10-Yr

0

5-Yr

2-Yr

20

40

60

80

Percent Greater Than Low Use

Baseline

High Use

Water Management

Fig. 10. Comparison of scenario 21-day concentrations.

100

Higher Tier Exposure Assessments in Rice Paddy Areas: An American Perspective 233 16

% of Applied A.I.

14 12 10 8 6 4 2 0 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 Year Overflow

Drainage

Fig. 11. Pesticide releases from the rice paddy due to overflow versus drainage.

For this example, it was shown that, in general, environmental concentrations resulting from reduced disease pressure (Low Use) are approximately half of those predicted under the Baseline scenario. High Use concentrations are roughly twice as high as those under the Baseline scenario. The Water Storage scenario has the net result of reducing Baseline concentrations by up to 34%. However, these benefits were less pronounced in years with higher levels of precipitation in which the storage capacity of the paddies were frequently exceeded (Figure 11).

7. DISCUSSION Examples presented in this chapter demonstrate a tiered process for modeling rice in the U.S. The tiers presented are a stepwise increase in representing processes in a more realistic manner and each tier is designed to be protective. However, model results must be approached with an understanding of potential deficiencies and therefore certain caveats pertain to the analyses. Tier 1 is the most conservative and is designed as a screening process. Tier 4 uses actual pesticide use and aquatic monitoring data. Discussed in this section are caveats for Tiers 2 and 3.

7.1. Tier 2 Tier 2 is similar to the standard PRZM/EXAMS approach used by USEPA for most crops. However, the RICEWQ model is used instead of PRZM due to its ability to simulate rice crops and the EXAMS environment is a bayou instead of the standard pond. Tier 2 results are meant to provide conservative estimates of

234

A. M. Ritter and W. M. Williams

environmental concentrations and comparisons of the relative magnitudes between scenarios. Caveats pertaining to this tier are as follows: ●

●

● ●

Chemical degradation is assumed to be identical between regions when in reality, differences would occur when exposed to different biological communities. Agronomic practices and environmental systems vary spatially and temporally: planting and application dates would likely be staggered, irrigation scheduling would be different, and rice density and percentage of rice treated would vary. Simulations are conducted at maximum label rates. The standard bayou represents an extreme exposure environment because it has a high intensity of product use relative to water body size; because it simulates chemical inflow in paddy releases without the additional dilution that in reality would be accompanied by the inflow of paddy water; and because it is a closed bayou system that does not generate outflow, regardless of the size of overflow/drainage events. Simulation of other more typical receiving water systems would result in lower concentrations.

7.2. Tier 3 The Tier 3 scenarios represent a realistic vulnerable ecosystem in rice producing areas of the country. RIVWQ was selected over EXAMS based on its ability to simulate time-varying flow and pesticide loadings at any point within the channel system. Even though this tier is designed to be more realistic, it is still conservative as noted below. ●

●

●

●

●

First, the scenarios simulated 50% of the watershed-contained rice that was treated with pesticide. The percent of the 8-digit Hydrologic Unit Code (HUC-8) in rice production ranges from 0.01% to approximately 36%. The percent of county area with rice productions ranges from approximately 0.15% to 29%. Second, physicochemical properties for pesticides are selected to be realistic based on laboratory environmental fate studies, but on the conservative side. Particularly if only one laboratory value is reported. The value is multiplied by three to account for uncertainties. Third, the receiving water is more representative of a stream containing aquatic organisms. The scenario was developed to provide conservative exposure estimates for headwater systems for ecological risk assessments. Fourth, the receiving water (channel) scenario is a relatively slow moving water. The treated drainage area to normal capacity ratio (DA/NC) is very conservative, approximately 550 m⫺1 as compared with the DA/NC of 5 for the USEPA standard pond. Fifth, typically the rice farmer will turn over the paddy after harvest which would bury the residues, making them unavailable for the following year. The simulation assumed that the paddies were not turned over and the residues were allowed to build up.

Higher Tier Exposure Assessments in Rice Paddy Areas: An American Perspective 235 ●

Sixth, natural and anthropogenic landscape features exist in larger drainage basins to attenuate pesticide residue transport to water supply sources (vegetative filters, holding ponds, wetlands, etc.).

8. DISCLAIMER At this time USEPA has not endorsed nor rejected RICEWQ or RIVWQ models nor the tiered approaches and scenarios presented herein. It should be disclosed that the authors of this chapter are the authors of the RICEWQ and RIVWQ models. REFERENCES Ambrose, R. B., Wool, T. A., Martin, J. L., Connolly, J. P. and Schanz, R. W. (1991). WASP5.x, a hydrodynamic and water quality model – Model theory, user’s Manual, and programmer’s guide. Athens Environmental Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency. Bird, S. L., Cheplick, J. M., Carsel, R. F. and Fendley, M. J. (1992). Pesticide and Industrial Chemical Risk Analysis and Hazard Assessment (PIRANHA), PRZM Input Collator (PIC) – Version 2.0. Environmental Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Athens, GA. Burns, L. (May 2004). Exposure Analysis Modeling System (EXAMS) – User’s manual and system documentation. Ecosystems Research Division, U.S. Environmental Protection Agency, Athens, GA. EPA/600/R-081, September 2000, Revision G. http://www.epa.gov/ceampubl/swater/ exams/index.htm. Capri, E. and Miao, Z. (2002). Modeling pesticide fate in rice paddy. Agronomie 22, 363. Ecological Committee for FIFRA Risk Assessment Methods (ECOFRAM). (1999). ECOFRAM Aquatic Report, Sponsored by the U.S. Environmental Protection Agency, Office of Pesticide Programs (draft). Forum for the co-ordination of pesticide fate models and their use (FOCUS). (1997). Surface water models and EU registration of plant protection products. Final report of the work of the Regulatory Modelling Working Group on surface models. 24-2.97. Haan, C. T. (1977). Statistical Methods in Hydrology (p. 135). Iowa State University Press, Ames, IA. Hart, A. (2001). Probabilistic risk assessment for pesticides in Europe: Implementation and research needs, Report of the European workshop on Probabilistic Risk Assessment for the Environmental Impacts of Plant Protection Products (EUPRA). Med-Rice. (2003). Final report of the working group MED-RICE prepared for the European Commission in the framework of Council Directive 91/414/EEC. European Commission Document Reference Sanco 1092, Brussels, Belgium. Miao, Z., Cheplick, J. M., Williams, W. M., Trevisan, M., Padovani, L., Gennari, M., Ferrero, A., Vidotto, F. and Capri, E. (2003a). Simulating pesticide leaching and runoff in rice paddies with the RICEWQ-VADOFT model. J. Environ. Qual. 32, 2189. Miao, Z., Padovani, L., Riparbelli, C., Ritter, A. M., Trevisan, M. and Capri, E. (2003b). Prediction of the environmental concentration of pesticide in paddy field and surrounding surface water bodies. Paddy Water Environ. 1, 121. Morris, L. E. and Kendall, R. J. (2000). FIFRA Scientific Advisory Panel Meeting, April 5–6, Session I – A Set of Scientific Issues Being Considered by the Environmental Protection Agency Regarding: Implementation Plan for Probabilistic Ecological Assessment, SAP Report No. 2000-02A (August 2, 2000). Suárez, L. A. (2005). PRZM-3, a model for predicting pesticide and nitrogen fate in the crop root and unsaturated soil zones – User’s manual version 3.12.2. EPA/600/R-05/111, U.S. Environmental Protection Agency, National Exposure Research Laboratory, Athens, GA.

236

A. M. Ritter and W. M. Williams

U.S. Department of Agriculture (USDA). (1972). National engineering handbook, Section 4, Hydrology: Soil Conservation Service, p. 71. U.S. Department of Agriculture (USDA). (1994a). 1992 National resources inventory: Soil Conservation Service. U.S. Department of Agriculture (USDA). (1994b). Soil property database, SOILS5: Soil Conservation Service. U.S. Department of Agriculture – National Agricultural Statistics Service (USDA-NASS). (2004a). http://www.nass.usda.gov/Census_of_Agriculture/index.asp. U.S. Department of Agriculture – National Agricultural Statistics Service (USDA-NASS). (2004b). http://www.nass.usda.gov/research/atlas02/atlas-crops.html. U.S. Department of Agriculture, Natural Resources Conservation Service (USDA-NRCS). (2006). Land resource regions and major land resource areas of the United States, the Caribbean, and the Pacific Basin. U.S. Department of Agriculture Handbook 296, Washington, D.C. U.S. Environmental Protection Agency (USEPA). (2000). Technical progress report of the implementation plan for probabilistic ecological risk assessments: Aquatic systems, U.S. Environmental Protection Agency, Office of Pesticide Programs, Washington, D.C. U.S. Environmental Protection Agency (USEPA). (2000). Drinking waters screening level assessment, Part A: Guidance for use of the index reservoir in drinking water exposure assessments. U.S. Environmental Protection Agency, Office of Pesticide Programs, Washington, D.C. (draft September 1, 2000). U.S. Environmental Protection Agency (USEPA). (2001). GENEEC (Gen)eric (E)stimated (E)nvironmental (C)oncentration Model, Tier one screening model for pesticide aquatic ecological exposure assessment – User’s manual version 2.0. U.S. Environmental Protection Agency (USEPA). (2001). FIRST (F)QPA (I)ndex (R)eservoir (S)creening (T)ool, Tier one screening model for drinking water pesticide exposure – User’s manual version 1.0. U.S. Environmental Protection Agency (USEPA). (2003). Pesticide root zone model field and orchard crop scenario metadata. http://www.epa.gov/oppefed1/models/water/przm_scenario_metadata.wpd (May 22, 2003). U.S. Environmental Protection Agency (USEPA). (2004). Environmental Fate and Effects Division’s Risk assessment for the reregistration eligibility document for 2,4-dichlorophenoxyacetic acid (2,4-D) (draft December 1, 2004). U.S. Environmental Protection Agency (USEPA). (2004). Overview of the Ecological Risk Assessment Process in the Office of Pesticide Programs, U.S. Environmental Protection Agency. Endangered and Threatened Species Effects Determinations. Office of Prevention, Pesticides, and Toxic Substances, Office of Pesticide Programs, Washington, D.C. (January 23, 2004). U.S. Environmental Protection Agency (USEPA). (2006). Amendment to reregistration eligibility decision (RED) for propanil (March 2006) and the propanil RED (September 2003), Docket ID: EPA-HQ-OPP-2003-0348, Amendment to the Propanil RED (March 7, 2006), EPA-HQ-OPP2003-0348-0024, Propanil RED (September 30, 2003): EPA-HQ-2003-0348-0002. http:// www.epa.gov/oppsrrd1/REDs/propanil_red_combined.pdf. U.S. Geological Survey (USGS). (1997). National water quality assessment, pesticide national synthesis project. http://ca.water.usgs.gov/pnsp/crop/rice.html (1997, last update). USA Rice Federation. (2006). http://www.usarice.com/industry/communication/factsheet.html. Warren, R. L., Ritter, A. M. and Williams, W. M. (2004). A rice herbicide Tier 2 exposure assessment for European rivers based on RICEWQ/RIVWQ. In: A. Ferrero and F. Vidotto (Eds), Proc. of the Conference Challenges and Opportunities for Sustainable Rice-Based Production Systems, Edizioni Mercurio, Torino, Italy, p. 523. Williams, W. M., Ritter, A. M., Cheplick, J. M. and Zdinak, C. E. (1999). RICEWQ: Pesticide runoff model for rice crops – User’s manual and program documents version 1.6.1. Waterborne Environment Inc., S.E. Leesburg, VA. Williams, W. M., Zdinak, C. E., Ritter, A. M., Cheplick, J. M. and Singh, P. (2004). RIVWQ: Chemical transport model for riverine environments – User’s manual and program documentation version 2.02. Waterborne Environment Inc., S.E. Leesburg, VA.

Pesticide Risk Assessment in Rice Paddies: Theory and Practice E. Capri and D.G. Karpouzas (editors) © 2008 Elsevier B.V. All rights reserved

Chapter 10

Socio-Economic and Environmental Cost-Benefit Analysis of Rice Cultivation G. Canali Istituto di Economia agro-alimentare, Università Cattolica del S. Cuore, via Emilia Parmense, 84, I-29100 Piacenza, Italy Contents 1. Introduction: externalities, public goods and economic value of the resource-environment system 2. Towards a general framework for a social cost-benefit analysis of rice cultivation 3. Economic tools and techniques for the economic evaluation of costs and benefits 4. Quantitative assessment of specific aspects of rice cultivation 5. Major issues and perspectives for future work References

237 239 243 244 245 247

1. INTRODUCTION: EXTERNALITIES, PUBLIC GOODS AND ECONOMIC VALUE OF THE RESOURCE-ENVIRONMENT SYSTEM The focus of this chapter is on the social evaluation of costs, risks and benefits of rice cultivation, considered both from a theoretical and, to some extent, also from an empirical point of view. This general issue is very complex and entails the more specific one of economic assessment of pesticide risk. However it seems of crucial importance to understand the overall complex economic interactions between the product, rice in this case, different production techniques, risks of health and environmental damages, to assess actual or potential, negative or positive, “external” effects of rice cultivation on the environment as well as on other economic activities and on social welfare. Moreover, the economics of rice production largely depends on many different domestic agricultural policies, including regulation of pesticide use, as well as international policies and trade agreements, both at the regional (European Union, North American Free Trade Area, etc.) and world level (e.g. General Agreement on Tariffs and Trade before 1995 and World Trade Organization after that date, World Bank and Food and Agricultural Organization). All these factors interact, in the real world, in order to define the actual production and trade levels in different countries and geographic areas, and, directly or indirectly, the effective adoption of specific agronomic techniques, and, among these, the use of different types of pesticides, in terms of quantity used, distribution modalities, timing, safety regulations and so on. In order to develop the following economic analysis, first of all it is important to recall few definitions like the one of externalities, public good (vs. private

238

G. Canali

goods) and “economic value” of the resource-environment system. All these definitions are of crucial importance for the assessment of a social cost-benefit analysis of rice cultivation and its implementation in different geographical areas. Natural resources, such as land, forests, water, etc., as well as environmental attributes, like air and water quality, landscape amenities, soil erosion, etc., play an important role in increasing and/or decreasing the overall welfare of a society in a given geographical place (and all over the globe). Sometimes it is even very difficult to assess the area interested by some changes of environmental attributes or resource level (e.g. acid rain, sea pollution, area covered by the rainforest, etc.). Externalities arise when the production or consumption activity of an economic agent (e.g. rice producers), affects, in a positive or negative way the economic activity (consumption and/or production) of other economic agents, without any economic transaction on the market, i.e. without any “mediation” of the market; the economic effect, in this case, is external from the market. Therefore we may have both positive and/or negative externalities. This is perhaps one of the more common causes of “market failure”, i.e. the market is not able, in this case, to work “properly” even if many other economic conditions do apply, like product homogeneity and competitive structure of the market itself; in this case the market is not able to promote the best use of available finite resources in order to maximize social welfare. This is a typical text-book case for market failure and the more common examples are the ones referred to pollution. Most of the times, when we analyze natural resources and environmental services, we discover that two conditions do not apply to them: exclusivity and rivalry. With the first term we simply recognize the possibility to prevent people and/or firms from the use of a given product (if exclusivity holds), while with the second one we state that if one uses a given good, anybody else cannot do the same at the same time (if rivalry holds). Goods possessing both properties are known as private goods while if none of them apply, i.e. the good is non-exclusive and nonrival, the good is (strictly) public. Therefore, once that such good is produced (or destroyed), no one can be excluded from benefits (or from suffering for the decrease of welfare, in the opposite case). Clearly almost all natural resources and all environmental attributes have, at least to some extent, one or both these characteristics. Air and groundwater as well as sea, oceans, biodiversity and many other resources, are clearly public goods. Air and groundwater or surface water pollution, for example, are clear examples of negative externalities that can be generated also by some agricultural activity (like rice production). Therefore, the externalities and the public-good character of many environmental services are responsible for the failure of the market system in allocating and pricing resources and environmental services correctly. This situation creates the need for economic measure of the true monetary values in order to improve the allocation of public as well as private money (and therefore also production activities) (Johansson, 1993; Cropper Maureen and Oates Wallace, 1992). In other words, if we consider both private costs and benefits derived from rice production as well as social cost, risks and (perhaps) benefits, taking into account negative and positive externalities, and other effects on supply and demand of public goods, complexity is clearly the main characteristic of rice production activity. To some extent the same is true in general also for other agricultural products even if in this case, larger

Socio-Economic and Environmental Cost-Benefit Analysis of Rice Cultivation

239

differences in production techniques characterize different geographical areas, and the role of water is absolutely more important than in any other agricultural product. It should be clear, at this point, that what is really needed is a broader but correct resource evaluation, including all types of costs and benefits, in order to describe adequately the cost benefit evaluation applied to a given production activity and moreover in order to analyze the effects of a given technology when compared with another one. If one considers that a comprehensive analysis of this kind has been made only sporadically, one can easily understand both the difficulty of this task and its relevance in terms of economic efficiency and in terms of social welfare; policy implications, when clearly identified, are also of critical importance. Due to this complexity, we try to proceed first by identifying all possible aspects needing specific economic evaluation; in this way we try to develop a complete approach towards an economic evaluation, also taking into account interaction among different aspects. This is the aim and, to some extent also the structure, of this chapter. First, a tentative theoretical framework for a comprehensive analysis of all costs, risks and benefits, direct and indirect, of short and long run, private and public is developed. Later, in order to make this theoretical attempt empirically possible, we try to suggest different techniques for the evaluation of various possible effects of rice cultivation in different areas of the world, on human health, on the environment and its various qualities, on the quality of life in general, with emphasis on farmers themselves and population living nearby rice paddies. At this point the chapter focus is on available economic tools and techniques for the evaluation in monetary terms of the above costs and benefits, since in many cases they are not directly available from market information. Later, in paragraph 10.4 the chapter presents few quantitative assessments available in recent literature, with respect to specific aspects of the general framework, with an attempt to identify and quantify, from an empirical point of view, different results with respect to different cultivation areas and techniques around the world. In the last paragraph a number of relevant and still open issues are described and some possible policy implications are drafted. 2. TOWARDS A GENERAL FRAMEWORK FOR A SOCIAL COST-BENEFIT ANALYSIS OF RICE CULTIVATION In this section we try to develop a general framework for the assessment of the social cost-benefit analysis of rice cultivation. The framework could apply also to different agricultural products but the emphasis here is on this specific crop. First of all we only need to recall that the first step of a complete analysis may be simply the economic evaluation of the private production cost and revenue due to the main agricultural product and by its by-products. Among production costs we must include labour, land, water, pesticides, fertilizers, energy, mechanization, buildings, other capital, insurance services and other goods and services including extension and technical assistance. Even if most of the time this fact is not clear enough, different production technologies may be convenient in different ways in diverse geographical areas, due

240

G. Canali

to relative factor abundance; if labour is relatively abundant and capital scarce, among all theoretically possible production techniques farmers will use the one relatively more labour intensive and not capital intensive (Van der Eng, 2004). On the other hand this implication of the Heckscher-Ohlin-Samuelson theorem of international trade, may apply also to other production factors like, land, of course when we analyze agricultural production we must consider not only labour and capital but also land. Beside other possible problems in the empirical evaluation of this theory, it is now clear that agricultural trade policies may play a major role in determining the degree of factor abundance of a specific crop, by modifying the comparative advantage in a distorted way. For example, it is fairly easy to verify, looking at the data shown in Table 1 and in Graph 1, that rice yields reach the highest level, among the most important world producers, if compared with main European producers and Japan, just in these last countries where land is relatively scare and trade has been limited through different economic tools. Spain, Japan, Italy and also China are, in the last 6 years, the countries with the highest yields. The Chinese case is somehow different as in this case land is abundant and at a first glance this could make us think that production should be land intensive and therefore not so labour intensive as it seems to be looking at the very high average yield. However we must recall that what is crucial in the above-mentioned theory, is the relative factor abundance; it is well known that China is by far the largest country in terms of population and therefore it is fairly easy to understand that this country is relatively labour abundant. These yields seem coherent with this situation. Another way of looking intuitively the same issue could be based upon the theoretical need of rice for food in these countries, and this fact has certainly played a major role at least in explaining protectionist agricultural trade policies for this basic food both in Japan and in China. The need of domestic food and the political importance of being self-sufficient or almost so, for this basic product, has pushed policy makers in these countries, to support rice production in different ways. The result has been the development of a very labour- and capital-intensive production technologies for rice easily visible through the high yield. Of course production technologies do differ also in terms of pesticides-intensity, water-intensity and also energy-intensity (Rigg Jonathan, 1985). But together with these technologies, we may also find easily different degrees of pollution intensity and therefore different types and levels of negative externalities (Huang et al., 1997). Pollution can be of different types: first of all it can be due to (excessive or inefficient) nitrogen fertilization, as well as use (or misuse) of other fertilizers (mostly potassium and phosphorus), even if to a minor extent; the second source of negative externalities in terms of pollution is due to agro-chemicals, i.e. insecticides, herbicides and other pesticides. In both cases there are a number of different actual and potential negative effects. These chemical products may contaminate surface water as well as groundwater and negative effects can be generated for human health, other economic activities using water for their production processes, the environment and other natural resources (fishes, other animals, plants, different ecological equilibria, etc.). See, for example, the paper by Phuong (2002) dealing with negative effects of pesticide use in rice production on aquaculture in the Mekong Delta, where an interesting quantitative and dynamic model has been developed and used.

Socio-Economic and Environmental Cost-Benefit Analysis of Rice Cultivation

241

Table 1. Rice: area, yield and production in major world producing countries: 2000–2005 Data Area (.000 ha) India China Indonesia Bangladesh Thailand Vietnam Myanmar Philippines Brazil Japan Italy Spain Yield (t/ha) India China Indonesia Bangladesh Thailand Vietnam Myanmar Philippines Brazil Japan Italy Spain Production (.000 t) India China Indonesia Bangladesh Thailand Vietnam Myanmar Philippines Brazil Japan Italy Spain

2000

2001

2002

2003

2004

2005

44.712 30.301 11.793 10.801 9.891 7.666 6.302 4.038 3.655 1.770 220 117

44.900 29.144 11.500 10.661 10.125 7.493 6.413 4.065 3.143 1.706 218 116

41.200 28.509 11.521 10.771 9.988 7.504 6.381 4.046 3.146 1.688 219 113

42.500 26.780 11.477 10.824 10.193 7.452 6.528 4.006 3.181 1.665 219 118

42.100 28.616 11.923 10.369 9.200 7.444 6.000 4.127 3.733 1.701 230 122

43.400 29.270 11.801 11.100 10.200 7.340 6.270 4.000 3.936 1.706 221 117

2.85 6.26 4.40 3.48 2.61 4.24 3.38 3.07 3.03 6.70 5.58 7.07

3.12 6.15 4.39 3.40 2.62 4.29 3.42 3.19 3.24 6.64 5.85 7.58

2.61 6.19 4.47 3.49 2.61 4.59 3.42 3.28 3.32 6.58 6.31 7.22

3.11 6.06 4.54 3.61 2.65 4.64 3.55 3.37 3.25 5.85 6.41 7.28

3.04 6.31 4.54 3.62 2.59 4.82 3.95 3.51 3.56 6.42 6.63 7.41

3.01 6.26 4.57 3.70 2.65 4.95 3.91 3.65 3.34 6.65 6.40 7.23

127.400 189.814 51.898 37.628 25.844 32.530 21.324 12.389 11.090 11.863 1.230 827

139.900 179.305 50.461 36.269 26.523 32.108 21.916 12.955 10.184 11.320 1.273 876

107.600 176.342 51.490 37.593 26.057 34.447 21.805 13.271 10.457 11.111 1.379 819

132.200 162.304 52.138 39.090 27.038 34.569 23.146 13.500 10.335 9.740 1.402 855

128.000 180.523 54.088 37.548 23.860 35.888 23.700 14.497 13.277 10.912 1.523 900

130.513 183.354 53.985 41.104 27.000 36.341 24.500 14.615 13.141 11.342 1.413 846

Source: Data on areas and productions are from FAO, Faostat, while data on yields have been obtained by own calculation.

Negative health effects may hit farmers and/or agricultural workers, living close to the production sites, and/or other people living quite far from production activities but connected with them through water, land or air quality. Another difficult task is the need to identify and measure not only possible acute negative health

242

G. Canali

8.00

India

7.00

China Indonesia

6.00

Bangladesh

5.00

Thailand

4.00

Viet Nam Myanmar

3.00

Philippines 2.00

Brazil

1.00

Japan

0.00

Italy 2000

2001

2002

2003

2004

2005

Spain

Graph 1. Rice yield (t/ha) in major producing countries: 2000–2005. (See Colour Plate Section, page 255.)

effects due to pollution, but also the chronic ones. Quite often it is also very difficult to identify the exact cause–effect relationship between some kind of pollution even the one associated with rice production, and health problems. Moreover there are few other possible causes of negative externalities and, perhaps, also some positive ones: relationship between rice cultivation, land use and erosion, on one side, and methane emission from paddy rice production on the other. The need of land use for producing rice has often pushed farmers to deforestation, to intensive agricultural activity on poor land and later to its abandonment. Of course such kind of agricultural production technology has strong negative effects in terms of “consumption” of good or fertile land, and soil erosion. On the other hand, sometimes traditional cropping methods in several geographic areas, are highly protective in terms of erosion and land conservation (e.g. terracing). At the same time some specific cropping activities in specific areas may have positive effects on social welfare also due to its positive impact on landscape creation and conservation. Moreover, paddy rice production may have an important effect in terms of methane production; the issue has been addressed, in the paper by Boisvert et al. (2004), also with respect to the evaluation of possible techniques for reducing and managing this kind of emission, as well as cost related to it. Finally, rice production has a strong and strict connection with water use (and abuse); of course it is true that all agricultural crops require and use, in different ways and intensity – water, but it is clear that the role of water in this case is crucial. But water is a natural resource with the character of public good, in general, and therefore without a market price able to drive its allocation efficiently among alternative uses (Chirwa Wiseman, 2002; Teague Mark et al., 1995). For this reason, water use in rice production may create negative overall effects on the economy being possibly excessive or insufficient. Of course water quality is also important, as already mentioned, and rice production may generate externalities with respect to this point.

Socio-Economic and Environmental Cost-Benefit Analysis of Rice Cultivation

243

In order to evaluate global implications, i.e. economic implication in terms of social cost-benefit analysis, of a given rice production technology in a given environment, specific evaluation methods need to be applied to every single case since the relative importance of different possible negative as well as positive externalities may differ substantially from one country to another, and from one geographical area to another. Of course this evaluation is very difficult and absolutely unusual but the final result of this non-evaluation is probably, an inefficient resource allocation and possibly an excessive rice production activity at least in some specific areas where these negative externalities are more important, and, finally, an excessive pollution level. As we will see in next paragraphs, these negative effects could also be the result of an inefficient and/or insufficient technological assistance activity as well as extension activity. Very often a good cropping technology generates high level of risks and/or pollution or other negative effects, mostly due to an inadequate ability and professionalism of farmers and agricultural workers. Finally it is also important to note that this kind of approach is more difficult in developing countries than in developed ones. Developing countries, in fact, tend to place more emphasis on economic development than on environmental protection (and sometimes also health protection) and because of this fact they may tend to develop a comparative advantage in pollution-intensive activities; this may also be the case for more polluting agricultural productions, and rice is among these. Moreover if we also consider the widespread need, in many countries, for a quantitatively adequate production of rice, these negative effects possibly associated with its production may be even more evident and widespread. Finally, in this context, it seems even more important to develop efficient extension and technical assistance services in order to help farmers at least to apply properly available and well known technologies, according to a market equilibrium, if not a social optimum equilibrium. 3. ECONOMIC TOOLS AND TECHNIQUES FOR THE ECONOMIC EVALUATION OF COSTS AND BENEFITS The economic and monetary evaluation of all different sorts of positive and negative externalities is very difficult in general and, of course, also in this specific case. In general, one of the most important problem is the one of measurement of the benefits and costs of pollution control. While the costs of pollution control are fairly easy to measure, the benefits are generally more difficult. To this aim economists have developed both direct and indirect methods. In order to do this one needs to apply appropriate techniques for measuring the economic value of a number of environmental (public) goods and resources which may be negatively affected, directly or indirectly, by rice production. At this point the economic evaluation of changes of the level of environmental goods or their quality is feasible. Among economic tools used for direct evaluation there are already many different kinds of contingent valuation (CV) methods. These methods allow to evaluate, at least theoretically, a large number of public goods by developing, through interviews or questionnaires, a way for making the willingness

244

G. Canali

to pay for a given level of an environmental good or quality, for example, to emerge from a representative sample of people. With reference to indirect evaluation methods, instead, we attempt to infer from actual choices and behaviours the value that people recognize to the environmental good. Examples of these methods are the travel cost methods, and especially a number of different hedonic methods, based on choices among different goods bought (e.g. houses), different places where to live, different jobs (hedonic wage), etc. Sometimes costs and/or value of reduced health problems are needed in order to fully develop a social cost-benefit analysis and in this case specific difficulties do arise. First of all, most of the times no specific medical information exists with reference to the specific relation between a given type and intensity of pollution, in a given geographical contest, but these informations are absolutely essential. Therefore, the first step in order to assess these costs is to develop these studies, when not available. They can require quite a long time, and they can be costly. Of course the choice itself to require or not a given kind of scientific study of this kind before allowing a given technology to be used is a crucial point which is subject to a previous, often “informal” evaluation: if someone “think” that some sort of health problems may not be important, these analysis may not be required from the beginning. Since every study and every evaluation is costly, some sort of initial informal evaluation is often the real world starting point of an eventual global evaluation process. From this point, often only some presumed critical issues are sometimes assessed in a scientific way and also through an economic evaluation. The overall evaluation, therefore, in the real world is absolutely not usual. All these economic methods have been used and tested in the last few decades in many different geographical situations and with reference to diverse environmental goods or services. Many papers and books already contain detailed description of these methods and therefore it seems unnecessary to present and discuss them again here. On the other hand, as we will see in the next paragraph, only very few applications have been made with specific reference to rice production. 4. QUANTITATIVE ASSESSMENT OF SPECIFIC ASPECTS OF RICE CULTIVATION As previously mentioned, rice is probably one of the crop which may generate greater impacts on human health as well as on the environment and on other production activities. Scientific production assessing issues of this kind, at least from an economic point of view, is very limited if we exclude paper available only in Chinese. Here we decide to analyze, by means of some case studies, major issues already studied with reference to rice in recent years. The first issue is the effect of some cropping techniques on environmental degradation and soon after, on grain production itself; this aspect has been studied, by Rozelle et al. (1997) with reference to grain production in China in the period 1975–1990. After a long period of growing intensity in the activity of bringing marginal land into cultivation during the years of the reform period in China, and after an

Socio-Economic and Environmental Cost-Benefit Analysis of Rice Cultivation

245

increase in the availability and more intensive use of major farm inputs which reached the maximum level at the end of the 1980s, both yields and gross production of grains in this country have decreased. The paper discovers and measures this problem which is finally attributed to a rapid and quantitatively important problems of environmental degradation due to cropping techniques in former marginal areas. This is a very interesting case of important negative externalities which are able to generate visible effects on the agricultural production in a relatively short number of years; land degradation has played a major role in explaining the negative evolution of production. In this paper an estimation of real costs of land degradation has been developed and applied to the specific case. As stated in the paper, “China’s land and water resources experienced serious environmental stress, particularly in areas where marginal land was brought into production”. Again in this specific case study, erosion, salinization, soil exhaustion and the increase of land stocks prone to natural disaster, have been identified as the major causes of environmental damage, together with bad deforestation practices, and degradation of grasslands. This case study is not strictly focused on rice production but is useful to identify a way for dealing with this set of environmental problems. Another very interesting issue is the one of health risk due to use (and/or misuse) of pesticides in rice production. In a paper by Pingali et al. (1994), agricultural economists and medical doctor worked together in order to identify, evaluate from a medial point of view first and then evaluate from an economical point of view, specific effects of use of different types of pesticides. Econometric analysis showed the magnitude of acute and chronic health effects and health costs to be directly related to prolonged pesticide exposure, among other factors. The somehow amazing implication is that when health effects were explicitly included in the overall economic evaluation, the net benefits of insecticide use in rice cultivation were negative. According to this empirical research, the value of the rice crop lost to pests is invariably lower than the cost of treating agrochemical-caused diseases. In this case, when these costs are accounted for, the natural control (“do nothing”) option is the most profitable from the private (farmer’s) point of view. However these results are strongly dependent on agro-chemical management practices, the key factors in determining strong negative effects on farmers’ and agricultural workers’ health are inadequate storage, unsafe handling practices, improper sprayer maintenance, short re-entry intervals, and so on. Finally, after a deeper analysis, it seems that investments in farmer and agricultural worker training and information campaigns on proper agro-chemical management could reduce some health risks and therefore possibly change also the final cost-benefit equilibrium of agro-chemical use. 5. MAJOR ISSUES AND PERSPECTIVES FOR FUTURE WORK The major aim of the present work has been to try to develop a generalized framework for assessing risks, costs and benefits of rice cultivation from a general point of view, taking into account not only private (i.e. farmers’) costs and benefits, and not only market values, but also the economic value of non-market

246

G. Canali

goods and services that the agricultural land and the environment as a whole, provides to the entire society of a given area (province, region or country). What seems clear is that without a proper economic evaluation of both positive and negative externalities associated with rice cultivation, the final overall economic result for the society of a given production activity and a specific technology may vary dramatically with respect to the “simple” market balance. The first major non-market issue is the one connected to the economic evaluation of the economic impact of prolonged pesticide use on farmers and agricultural workers’ health. In order to evaluate these aspects specific studies must be realized in order to take into account specific local land, climate, groundwater characteristics, among others, as well as specific cropping techniques and availability of services such as extension and technical assistance. Little evidence is available from recent literature with respect to rice production, but few implications may be obtained from them. (1) Social cost-benefit analysis is very difficult to perform but it seems reasonably important especially for rice production if one considers the number and intensity of possible interactions between this particular agricultural production, health risks, water use, possibility of pollution, effects on soil erosion and landscaper amenities, etc. (2) The analysis must be performed at the national or regional level in order to identify specific production and environmental conditions that may generate a wide range of possibly different technical and economic interactions and effects (Barrett Christopher et al., 2004). (3) The evaluation of a complete social cost-benefit analysis could be too costly to be performed and benefits from the evaluation of minor external costs of rice production can be smaller than their cost. Therefore it seems reasonable to develop, first, an overall draft vision of all positive and negative interactions among rice production, the environment, human health, other production activities, etc. (4) Only major critical values of social (i.e. both private and strictly social) costs and benefits need to be identified and measured for introducing and supporting specific and useful policy tools. (5) A major role in reducing negative externalities seems to rely on the availability of effective extension and technical assistance services: with small expenditure, large social cost reduction can be achieved. (6) The role of technical assistance is important especially in developing countries for reducing social costs of rice production as well as for improving private (and social) benefit of this production. Developing and developed countries face, to some extent, different issues and have, in general, different political priorities. This may initially push developing countries to under-evaluate environmental risks and cost more than developed countries. A risk may also exist, for this reason, to increase the risk that developing countries may become, even without an explicit choice in this direction, a sort of “pollution heaven” when protectionist policies in some countries for rice will start to decrease in a significant way.

Socio-Economic and Environmental Cost-Benefit Analysis of Rice Cultivation

247

On the other hand, also in developing countries there is a growing knowledge and concern of health risks and negative implication of the overall economic performance of negative externalities of rice production, these factors can play a role in increasing the attention of policy makers to a social cost-benefit analysis, at least partial if not yet global. One last issue relates to WTO negotiation: one of the unresolved problem in this arena is the relation between Multilateral Environmental Agreements and trade negotiation, and rice could be one of the more interesting cases with reference to agricultural products. A global approach to this relationship does not seem very probable but few particular provisions could be made at least with respect to homogeneity of environmental regulations in order to reduce overall negative environmental effects of production activities, while increasing competition. Growing competition and freer trade can be even dangerous, not only from a strictly environmental point of view, but also in strictly economic terms, on a global as well as national basis, if negative externalities associated with some productions are not accounted for and if they are not included in the overall economic evaluation first and in the environmental regulation next. REFERENCES Barrett, C. B., Christine, M., McHugh, O. V. and Barison, J. (2004). Better technology, better plots, or better farmers? Identifying changes in productivity and risk among Malagasy rice farmers. Am. J. Agric. Econ. 86(4), 869–888. Boisvert, R., Chang, H., Barker, R., Levine, G., Matsuno, Y. and Molden, D. (2004). Refining the positive and negative externalities of Taiwanese paddy-rice production, in “Accounting of Agricultural and Nonagricultural Impacts of Irrigation and Drainage Systems – A Report of Research in Taiwan and Sri Lanka in 2003”, Working Paper no. 68, International Water Management Institute, Colombo, Sri Lanka. Chirwa Wiseman, C. (2002). Land use and extension services at Wave Rice Scheme, Malawi. Dev. South. Africa 19(2), 307–327. Cropper Maureen, L. and Oates Wallace, E. (1992). Environmental economics: A survey. J. Econ. Lit. XXX(June), 675–740. Food and Agriculture organization of the UN, Statistic Division (2006). Faostat, http://faostat.fao.org/site/340/default.aspx. Huang, J., Qiao, F., Zhang, L. and Rozelle, S. (1997). Farm pesticide, rice production, and human health, Research Report, International Development Research Centre, Ottawa, Canada. Johansson, Per-Olov (1993). Cost-benefit analysis of environmental change, Cambridge University Press, Cambridge, Great Britain. Phuong, D. M. (2002). The impacts of pesticide use in rice production on aquaculture in the Mekong Delta: A dynamic model, Working Paper, EEPSEA. Pingali Prabhu, L., Marquez Cynthia, B. and Palis Florencia, G. (1994). Pesticides and Philippine rice farmer health: A medical and economic analysis. Am. J. Agric. Econ. 76(August), 587–592. Rigg Jonathan, D. (1985). The role of the environment in limiting the adoption of new rice technology in Northeastern Thailand. Trans. Inst. Br. Geogr. 10, 481–494. Rozelle, S., Veeck, G. and Huang, J. (1997). The impact of environmental degradation on grain production in China, 1975–1990. Econ. Geogr. 73(1), 44–66. Teague Mark, L., Bernardo Daniel, J. and Mapp Harry, P. (1995). Meeting environmental goals efficiently on farm-level basis. Rev. Agric. Econ. 17(1), 37–50. Van der Eng, P. (2004). Productivity and comparative advantage in rice agriculture in South-East Asia since 1870. Asian Econ. J. 18(4), 345–370.

Colour Plate Section

255

8.00

-+-India ----China

7.00 6.00

•

•

Indonesia

~

~

~

5.00 4.00

•

•

"\L

3.00

~

-----------....

2.00

... '"

----

Bangladesh

--*- Thailand ~

---- Viet Nam -+-Myanmar

~

- - Philippines --Brazil

1.00

Japan

0.00

Italy 2000

2001

2002

2003

2004

2005

Spain

Plate 3. Rice yield (t/ha) in major producing countries: 2000-2005. (See also page 242 of this book.)

Subject Index Acceptable daily intake, 30, 171 Acceptable operator exposure level, 30 Active substances, 25, 27-29, 36-39, 41, 63, 69 Acute reference dose, 30 Biocides, 26, 46, 53-54 Climate, 2, 4, 7-8, 46, 62, 64, 92, 167-168, 180,206,210-211,221,225,246 Crustaceans, 12, 20 Cultivated area, 65, 127, 131, 140, 150, 155, 157, 159, 189, 191 Dummy pesticides, 112, 115 Ecotoxicological methods, 69 Emergency approval, 28 European commission's standing committee on the food chain and animal health, 27 European food safety authority, 25, 27, 39 European standard scenario, 92 Experimental approval, 28 Full approval, 28-29 Fungi, 20, 26, 40 Geographic information systems, 63 Good agricultural practice, 28, 65, 77, 134, 168-169,209,211 Hardpan, 6, 204 Hazard indicator, 63 Herbicides, 13, 15-19, 33, 41, 59, 65, 73, 81, 86, 92, 101, 121, 126, 138, 140, 145, 151, 153,155,159,168-170,185-186,188-191, 194,198-199,209-210,216,240 In-crop, 53, 55, 72-75, 80-81, 83-84 Insects, 12, 20-21, 26, 33, 216

Mineral fertilization, 10, 13 Monitor pesticide usage, 34 Multilateral environmental agreement, 247 No observed adverse effect levels, 29 Off-crop, 55, 72-75, 77, 80, 83, 86 Organic fertilization, 13 Pedology, 6 Peer-review coordination meetings, 39 Plant protection products, 25, 27, 39, 48, 51-52,71-72,75,80,91 Plowing, 8-9, 13-14, 22 Predicted environmental concentration, 32, 91, 93-94,99,101,126,169,173 Probablistic risk assessment, 42, 86 Protection goal, 51, 53, 70, 76, 80, 83 Provisional approval, 28 Puddling, 9, 22, 181 Reach, 4, 12, 21, 43, 62, 86, 148, 240 Social cost-benefit analysis, 237-239, 243-244,246-247 Sowing, 8 Straw-burying, 13 SWAGW, 102,108,112,114-122,126-127 Tiered exposure assessment, 127 Tillage, 8-9, 12, 62 Trade negotiation, 247 Transplanting, 11, 15, 168, 185, 187, 190 Varieties, 2-4, 6, 9, 11-12, 14-15, 17-21, 141, 168

Land levelling, 8, 16

Water framework directive, 43 Water requirements, 12 Weeds, 1, 3, 8-11, 14-19, 41, 59, 81, 92, 155, 169, 210, 216 Worms, 12, 20, 40

Mechanization, 6-7, 239 Med-rice scenario, 93, 115, 121

Yield, 2-5, 9, 13-15, 20-21, 40, 54-55, 80-81,240-242

Japanese pesticide registration, 173