The loss of biodiversity is recognised on a global scale and also in the anthropogenic landscapes used for agriculture, now covering almost 50% of the global terrestrial land surface. In agriculture pesticides, biologically active chemicals are deliberately distributed to control pests, disease and weeds in the cropped areas. The quantification of remaining semi-naturals structures such as field margins and hedges is a prerequisite to understand the impact of pesticides on biodiversity, since these structures represent habitats for many organisms in agricultural landscapes. The presence of organisms in these habitats and crops is required to obtain an estimate of their potential pesticide exposure. In this text I provide studies on animal groups so far not addressed in risk assessment procedures for the regulation of pesticides such as amphibians, moths and bats. For all groups it becomes apparent that they are present in agricultural landscapes and potentially coincide with pesticide applications indicating a risk. Risk quantification also requires data on the sensitivity of organisms and here data for plants, amphibians and bees are presented. Effects translating to community level were studied for herbicide, insecticide and fertiliser effects in a natural system. After three years the treatments resulted in simplified plant communities with lower species numbers and a reduction in flowering plants. This reduction of flowers is used as an example for an indirect effect and was especially obvious for the effect of an herbicide on the common buttercup. Sublethal herbicide effects for a plant translated in an impact on feeding caterpillars, indicating a reduction in food quality. Insecticide inputs realistic for field margins also reduced moth pollination of white champion flowers by 30%. These indirect effects by distortions of food web characteristics are playing a critical role to understand declines in organism groups, however so far are not accounted for in pesticide risk assessment schemes. The current intense use of pesticides in agriculture and their inherent toxicity may lead to a chemical landscape fragmentation, where populations may not be connected anymore. Source-sink dynamics are important ecological processes and as a final result not only population size but also genetic population structure might be affected. Including potential pesticide impacts as costs in a model for amphibians migrating to breeding ponds in vineyards in Rhineland-Palatinate indicated the isolation of investigated populations. A first validation by analyzing the population structure of the European common frog confirmed the model prediction for some sites. For the regulation of pesticides in Europe a risk assessment is required and for the organisms of the terrestrial habitat a multitude of guidance documents is in place or is recently developed or improved. The results of the presented research indicate that wild plants and especially their reproductive flower stage are highly sensitive and risks are underestimated. Population recovery of arthropods needs a reevaluation at landscape scale and the addition of amphibian risk assessment in regulation procedures is suggested. However, developing or adopting risk assessment procedures and test systems is a time consuming task and therefore the establishment of risk management options is a pragmatic alternative with immediate effects. Artificial wetlands in the agricultural landscape proved to be important foraging sites for bats and their creation could mitigate negative pesticide effects. The integration of direct and indirect effects in a risk assessment scheme for all organism groups addressing also landscape scale and pesticide mixtures requires a long developing time. The establishment of model landscapes where management options and integrated pest management are applied on a larger scale would allow us to study pesticide effects in a realistic scenario and to develop an approach for the agriculture of the future.

The estimation of the potential risk of pesticide entries into streams - and therefore the potential risk for the ecosystems - is an important requirement for the planning of risk mitigation strategies. Especially on the landscape level the required event triggered sampling methods are conjuncted with considerable efforts with regard to input data, time and personnel. To circumvent these problems simulation models form a reasonable alternative. The aims of this work were (A) the development of a simulation tool for the estimation of pesticide entries into surface waters on the landscape level, and (B) the application of the simulator for an exposure- and risk-assessment as well as the assessment of negative effects of pesticides on aquatic communities. Section 1 - Exposure-, Risk- and Effects In sections 1.1 and 1.2 the simulation model was applied to a multitude of small and medium sized streams in an agricultural impacted study area around the city of Braunschweig, Germany. Section 1.3 gives an overview of the simulators field of application and the general system structure. Section 1.1 - Scenario based simulation of runoff-related pesticide entries into small streams on a landscape level (English publication, p. 27): In this paper we present a simulation tool for the simulation of pesticide entry from arable land into adjacent streams. We used the ratio of exposure to toxicity (REXTOX) model proposed by the OECD which was extended to calculate pesticide concentrations in adjacent streams. We simulated the pesticide entry on the landscape level at 737 sites in small streams situated in the central lowland of Germany. The most significant model parameters were the width of the no-application-zone and the degree of plant-interception. The simulation was carried out using eight different environmental scenarios, covering variation of the width of the no-application-zone, climate and seasonal scenarios. The highest in-stream concentrations were predicted at a scenario using no (0 m) buffer zone in conjunction with increased precipitation. According to the predicted concentrations, the risk for the aquatic communities was estimated based on standard toxicity tests and the application of a safety factor. Section 1.2 - Linking land use variables and invertebrate taxon richness in small and medium-sized agricultural streams on a landscape level (English publication, p. 50): In this study the average numbers of invertebrate species across an arable landscape in central Germany (surveys from 15 years in 90 streams at 202 sites) were assessed for their correlation with environmental factors such as stream width, land use (arable land, forest, pasture, settlement), soil type and agricultural derived stressors. The stress originating from arable land was estimated by the factor "risk of runoff", which was derived from a runoff-model (rainfall induced surface runoff). Multivariate analysis explained 39.9% of the variance in species number, revealing stream width as the most important factor (25.3%) followed by risk of runoff (9.7%). Section 1.3 - Informationssystem zur ökotoxikologischen Bewertung der Gewässergüte in Bezug auf Pflanzenschutzmitteleinträge aus der Landwirtschaft - Systemaufbau und Anwendungsmöglichkeiten (German publication, p. 61): Section 1.3 contains a short overview of the simulation tool, the field of application and some examples of use, covering the effects of the width of the buffer zone as well as the creation of risk maps on the landscape level. Section 2 - The simulation tool An important aspect for the employment of a simulation model in the context of risk assessment is the applicability in practice: the accessibility of the needed input data, the conversion of the mathematical model into a software application that can be run on any current personnel computer and also an appropriate end-user documentation of the system. Section 1.4 - Informationssystem zur ökotoxikologischen Bewertung der Gewässergüte in Bezug auf Pflanzenschutzmitteleinträge aus der Landwirtschaft - Simulationsmodell und Systemaufbau (German report, p. 67): In this section a general overview of the simulation model as well as the schematic system structure given. Section 1.5 - Benutzerhandbuch (German report, p. 71): The user manual contains details concerning the installation of the system, generation of the required input data and the general use of the system. Moreover it presents some application examples (what-if analyses). Section 1.6 - Technical documentation (German report, p. 104): The technical documentation describes internal structures and processes of the simulation system. Section 1.6 provides information regarding the required structure of input/output tables.