Streams are coupled with their riparian area. Emerging insects from streams can be an important prey in the riparian area. Such aquatic subsidies can cause predators to switch prey or increase predator abundances. This can impact the whole terrestrial food web. Stressors associated with agricultural land use can alter insect communities in water and on land, resulting in complex response patterns of terrestrial predators that rely on prey from both systems. This thesis comprises studies on the impact of aquatic nsects on a terrestrial model ecosystem (Objective 1, hapter 2), the influence of agricultural land use on riparian spiders’ traits and community (Objective 2, Chapter 3), and on the impact of agricultural land use on the contribution of different prey to spider diet (Objective 3, Chapter 4). In chapter 2, I present a study where we conducted a mesocosm experiment to examine the effects of aquatic subsidies on a simplified terrestrial food web consisting of two types of herbivores (leafhoppers and weevils), plants and predators (spiders). I focused on the prey choice of the spiders by excluding predator immigration and reproduction. In accordance with predator switching, survival of leafhoppers increased in the presence of aquatic subsidies. By contrast, the presence of aquatic subsidies indirectly reduced weevils and herbivory. In chapter 3, I present the results on the taxonomic and trait response of riparian spider communities to gradients of agricultural stressors and environmental variables, with a particular emphasis on pesticides. To capture spiders with different traits and survival strategies, we used multiple collection methods. Spider community composition was best explained by in-stream pesticide toxicity and shading of the stream bank, a proxy for the quality of the habitat. Species richness and the number of spider individuals, as well as community ballooning ability, were negatively associated with in-stream pesticide toxicity. In contrast, mean body size and shading preference of spider communities responded strongest to shading, whereas mean niche width (habitat preference for moisture and shading) responded strongest to other environmental variables. In chapter 4, I describe aquatic-terrestrial predator-prey relations with gradients of agricultural stressors and environmental variables. I sampled spiders, as well as their aquatic and terrestrial prey along streams with an assumed pesticide pollution gradient and determined their stable carbon and nitrogen signals. Potential aquatic prey biomass correlated positively with an increasing aquatic prey contribution of T. montana. The contribution of aquatic prey to the diet of P. amentata showed a positive relationship with increasing toxicity in streams. Overall, this thesis contributes to the emerging discipline of cross-ecosystem ecology and shows that aquatic-terrestrial linkages and riparian food webs can be influenced by land use related stressors. Future manipulative field studies on aquatic-terrestrial linkages are required that consider the quality of prey organisms, fostering mechanistic understanding of such crossecosystem effects. Knowledge on these linkages is important to improve understanding of consequences of anthropogenic stressors and to prevent further losses of ecosystems and their biodiversity.

The world wide decline of insects is often associated with loss of natural and semi-natural habitat caused by intensified land-use. Many insects provide important ecosystem services to agriculture, such as pest control or pollination. To efficiently promote insects on remaining semi-natural habitat we need precise knowledge of their requirements to non-crop habitat. This thesis focuses on identifying the most important semi-natural habitats (forest edges, grasslands, and semi-open habitats) for pollinators and natural enemies of crop pests with respect to their food resource requirements. Special attention is given to floral resources and their spatio-temporal distribution in agricultural landscapes. Floral resource maps might get closer at characterizing landscapes the way they are experienced by insects compared to classical habitat maps. Performance of the two map types was compared on the prediction of wild bees and natural enemies that consume nectar and pollen, identifying habitats of special importance in the process. In wild bees, influences of spatio-temporal floral resource availability were analysed as well as habitat preferences of specific groups of bees. Understanding dietary needs of natural enemies of crop pests requires additional knowledge on prey use. To this end, ladybird gut contents have been analysed by means of high-throughput sequencing for insight into aphid prey-use. Results showed, that wild bees were predicted better by floral resource maps compared to classical habitat maps. Forest edge area, as well as floral resources in forest edges had positive effects on abundance and diversity of rare bees and important crop pollinators. Similar patterns were retained for grassland diversity. Especially early floral resources seemed to have positive effects on wild bees. Crops and fruit trees produced a resource pulse in April that exceeded floral resource availability in May and June by tenfold. Most floral resources in forest edges appeared early in the season, with the highest floral density per area. Grasslands provided the lowest amount of floral resources but highest diversity, which was evenly distributed over the season. Despite natural enemies need for floral resources, classical habitat maps performed better at predicting natural enemies of crop pests compared to floral resource maps. Classical habitat maps revealed a positive effect of forest edge habitat on the abundance of pest enemies, which translated into improved aphid control. Results from gut content analysis reveal high portions of pest aphid species and nettle aphids as well as a broader insight into prey spectra retained from ladybirds collected from sticky traps compared to individuals collected by hand. The aphid specific primer designed for this purpose will be helpful for identifying aphid consumption by ladybirds in future studies. Findings of this thesis show the potential of floral resource maps for understanding interactions of wild bees and the landscape but also indicate that natural enemies are limited by other resources. I would like to highlight the positive effects of forest edges for different groups of bees as well as natural enemies and their performance on pest control.

Carabids, which are frequently distributed in agricultural landscapes, are natural enemies of different pests including slugs. Semi-natural habitats are known to affect carabids and thus, their potential to support natural pest control. The impact of semi-natural habitats was investigated on carabids and slugs within different non-crop habitats (chapter 2). Most carabids and Deroceras reticulatum showed preferences for herbaceous semi-natural habitats, while Arion spp. occured mainly in woody habitats. An increase of predatory carabid abundance, which was linked to an inclining amount of semi-natural habitats in the landscape, and a decrease of Arion spp. densities, indicated a high potential for slug control in structural rich landscapes. Effects of semi-natural habitats were investigated on predatory carabids and slugs in 18 wheat fields (chapter 3). Predatory carabid species richness was positively affected by the increasing amount of semi-natural habitats in the landscape, whereas predatory carabid abundance was neither influenced by adjacent habitat type nor by the proportion of semi-natural habitats in the landscape. The target pest species showed divergent patterns, whereas Arion spp. densities were highest in structural poor landscapes near woody margins. D. reticulatum was not affected by habitat type or landscape, reflecting its adaptation to agriculture. Results indicate an increased control of Arion spp. by carabids in landscapes with a high amount of semi-natural habitats. Effects of semi-natural habitats and the influence of farming system was tested on carabid distribution within 18 pumpkin fields (chapter 4). Carabid species richness generally increased with decreasing distance to the field margins, whereas carabid abundance responded differently according to the adjacent habitat type. Farming system had no effect on carabids and landscape heterogeneity only affected carabids in organic pumpkin fields. Slug and slug egg predation of three common carabid species was tested in single and double species treatments in the laboratory (chapter 5). Results show additive and synergistic effects depending on the carabid species. In general, semi-natural habitats can enhance the potential of slug control by carabids. This counts especially for Arionid slugs. Semi-natural habitats can support carabid communities by providing shelter, oviposition and overwintering sites as wells as complementary food sources. Therefore, it is important to provide a certain amount of non-crop habitats in agricultural landscapes.