Abstract/Summary

Insects are integral to many ecosystem services, including pollination of crops by insect pollinators and control of pests by natural enemies (i.e. predators and parasites). These beneficial insects are suffering widespread declines, partly due to agricultural expansion and intensification. We must therefore change the management of agricultural landscapes to preserve these ecosystem services and the resilience of the systems that depend on them. To make such changes, we first need to understand the relationships between landscapes, beneficial insect communities and the delivery of the services they provide. This thesis focuses on analysing novel combinations of large-scale spatial datasets to explore these relationships. Firstly, a ten-year time-series of wheat yield data was analysed in conjunction with satellite�derived land cover data to explore national-scale relationships between crop yield resilience and landscape structure. Whilst relative yield was highest in landscapes dominated by arable land, stability and resistance were promoted by semi-natural habitats. Secondly, data from biological recording schemes were used to construct potential plant-pollinator networks across Great Britain and explore relationships between network structure and land cover. Networks were most robust to simulated extinctions in highly agricultural landscapes, because they supported distinctive, generalist pollinator communities. Finally, data on land cover and cropping patterns were used to develop a range of scenarios of agricultural change. These were linked to beneficial insect richness and functional diversity using species distribution models based on biological records. Scenarios involving restoration of semi-natural grasslands increased the richness and functional diversity of beneficial insects, even if cropped land remained intensive. The findings presented in this thesis demonstrate the value of combining and modelling spatial data in exploring insect-delivered ecosystem services. Whilst there is much scope for further work, including integration with experimental data, spatial modelling remains key to providing the large-scale evidence required by policy makers and agricultural land managers.