954 Environmental Toxicology and Chemistry, Vol. 24, No. 4, pp. 954–965, 2005  2005 SETAC Printed in the USA 0730-7268/05 $12.00 + .00 ANALYZING EFFECTS OF PESTICIDES ON INVERTEBRATE COMMUNITIES IN STREAMS MATTHIAS LIESS* and PETER CARSTEN VON DER OHE UFZ—Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany ( Received 1 December 2003; Accepted 25 September 2004) Abstract—The aim of this investigation was to find patterns in aquatic invertebrate community composition that are related to the effects of pesticides. Investigations were carried out in 20 central European streams. To reduce the site-specific variation ofcommunity descriptors due to environmental factors other than pesticides, species were classified and grouped according to their vulnerability to pesticides. They were classified as species at risk (SPEAR) and species not at risk (SPEnotAR). Ecological traits used to define these groups were sensitivity to toxicants, generation time, migration ability, and presence of aquatic stages during time of maximum pesticide application. Results showed that measured pesticide concentrations of 1 : 10 of the acute 48-h median lethal concentration (LC50) of Daphnia magna led to a short- and long-term reduction of abundance and number of SPEAR and a corresponding increase in SPEnotAR. Concentrations of 1 : 100 of the acute 48-h LC50 of D. magna correlated with a long-term change of community composition. However, number and abundance of SPEAR in disturbed stream sections are increased greatly when undisturbed stream sections are present in upstream reaches. This positive influence compensated for the negative effect of high concentrations of pesticides through recolonization. The results emphasize the importance of considering ecological traits and recolonization processes on the landscape level for ecotoxicological risk assessment. Keywords—Field effect Pesticides Recovery Cumulative risk assessment INTRODUCTION The European Union uniform principles for the assessment of pesticides require that if the preliminary risk characteriza- tion indicates potential concerns, registration cannot be granted unless it can be demonstrated that ‘‘under field conditions no unacceptable impact on the viability of exposed organisms occurs.’’ To date, such assessments have been made by con- ducting higher-tier studies, which have included a range of laboratory and semifield experiments. Therefore, it still is not clear to what extent pesticides change population dynamics and community structures in the field. Recently, some studies have quantified pesticide exposure, adverse effects on aquatic life, and recovery of these invertebrate communities in the field. Mortality of six mayfly species in an Australian river was linked to endosulfan contamination due to runoff [1]. Other investigations also found a link between mortality of invertebrate species and insecticide concentrations in streams [2,3]. Several invertebrate species that declined in abundance due to pesticides were found to recover within a year [2]. Nevertheless, most existing studies lack sufficient numbers of investigations in various streams to evaluate the frequency of potentially harmful events in a specific region, evaluation of long-term effects on invertebrate communities, and quantifi- cation of the recovery of impacted communities due to re- colonization from undisturbed stream sections. The inclusion of habitat quality may put the risks resulting from contami- nation in context with other stressors. According to these open questions, the aim of the present investigation was to find patterns in community composition that were related to the effect of pesticides. It is challenging in field investigations to reveal the importance of a specific * To whom correspondence may be addressed (matthias.liess@ufz.de). environmental factor, because other environmental factors may mask possible effects. Therefore, to tackle this problem a new approach that aims at reducing variability in community char- acterization is presented. METHODS Study area The study area is located around Braunschweig, Lower Sax- ony, Germany. The dominant land use is agricultural (field 61%, forest 34%, pasture 5%; Fig. 1). The most common crops in the catchments are sugar beets, winter barley, and winter wheat. The investigation was carried out in an area where sites had a risk of runoff ranging from very low (level 0) to high (level 5), on a scale ranging from level 0 to level 6 (very high), defined for German agricultural areas [4]. Description of streams Twenty sites, located on first-order streams, were selected to match the following requirements: All-year water flow; no dredging in the years before and during the investigation; no pollution from other than agricultural nonpoint sources; var- ious pesticide loads, stemming from differences in the per- centage of adjacent arable land. Nine streams were monitored for one year; six streams were monitored for two years; and five streams were monitored for three years (11 sites in 1998, 11 sites in 1999, and 14 sites in 2000). The streams investigated for three years spanned the entire range of measured toxicity. Data on streams that were investigated for several years were pooled to avoid temporal pseudoreplication. Physical and chemical standard parameters were measured monthly in April (period before the application of insecticides), May, and June. Oxygen, pH, and temperature were recorded with instruments made by WTW (Weilheim, Germany). Concentrations of ni- trate, nitrite, ammonium, and phosphate were determined in