A major challenge in agroecosystems is how to manage the systems so that it reduces crop pests and enhances natural pest control. This thesis investigates patterns of crop pests and top-down effects of birds and arthropod predators in relation to land-use composition across spatial scales. In paper (I) I examined the crop distribution and land-use types in relation to the crop raiding patterns in 15 transectsin sites close to and far from forests along with a questionnaire survey at household level. I found severe crop raiding close to forests, but it had no impact on crop composition growing between the two sites. In paper (II) I examined the effect of forest and tree cover, at local and landscape scales, on the abundance of arthropod predators by collecting specimens from 40 home gardens. My result showed higher abundance of arthropod predators when either the home garden or the surroundings had a high tree-cover, compared to when tree-cover at both scales was similarly either high or low. In paper (III) I investigated the effect of excluding birds and arthropod predators on leaf damage on rape seed in 26 home gardens. I found stronger top-down impacts from arthropod predators on crop pests in tree-poor gardens than in tree-rich gardens. There was no effect of birds. In paper (IV) I explored the effect of landscape complexity on bird and arthropod predation using plasticine caterpillars in 36 home gardens across landscapes. The rate of arthropod predation on caterpillars was higher in simple than in complex landscapes. The rate of bird predation did not vary between complex and simple landscapes. In simple landscapes, arthropod predation was higher than that of birds. The overall results suggest that simplified gardens/landscapes still have enough habitat heterogeneity to support arthropod predators for the significant top-down controlling effect on crop pests. However, I did not find clear effect of complexityon the top-down effect of birds. / <p>At the time of the doctoral defence the following papers were unpublished and had a status as follows: Paper 1: Manuscript; Paper 3: Manuscript; Paper 4: Manuscript</p>

To show how predaceous fish and dragonflies affect benthic community structure, we conducted enclosure experiments in the littoral zone of Bays Mountain Lake, Tennessee. A 'natural' benthic assemblage was subjected to all combinations of 2 densities of 3 predator treatments - small sunfish, 0 or 4/m2; large dragonfly larvae, 0 or 15/m2; and dragonfly eggs at 2 densities, 90 or 900/m2. Treatments were assigned randomly in each of 6 spatiotemporal blocks. Net effects of predation over 4 mo show that small sunfish had 'strong' effects (>50% reduction of densities) on triclads, large daphnid cladocerans, and snails, and 'moderate' effects (50% > reduction > 25%) on small dragonflies and ostracods; all these effects were statistically significant, except the one on ostracods. Large dragonflies had moderate non-significant effects on triclads and ostracods. There were no significant increases in prey density associated with fish predation; but chydorid cladocerans and midge larvae showed 'weak' non-significant increases that might be caused by 'indirect effects' of fish predation on invertebrate predators or grazers. There were only 2 significant interaction terms indicative of 'higher-order interactions': Fish X Dragonfly on the large daphnid Simocephalus, and Fish X Egg Density on snails, which were associated with relatively low probabilities. In both cases, the net effect of dragonflies was to reduce prey densities more when fish were present.