|dc.description.abstracteng||Agricultural landscapes account for approximately 40% of global terrestrial area and this proportion tends to grow due to the increasing demand for fuel, food and other agricultural products. For successful crop production in the future, it is crucial to preserve the ecosystem services performed by biodiversity in these areas. Understanding how land use and landscape composition affect species distribution and ecosystem functions is key to achieve this objective. Organisms respond to habitat heterogeneity at different spatial scales, which needs to be considered in landscape studies. The temporal scale can also be meaningful, given that agricultural landscapes are highly dynamic environments and the frequent changes in land cover may affect ecological processes.
In this thesis we study local and landscape effects on plant-insect interactions, local diversity and community structure in changing landscapes. We focus on two landscape components that potentially influence insect communities and interactions: semi-natural habitats and oilseed rape fields (Brassica napus L.). Semi-natural habitats are stable environments that provide food and nesting resources constantly. Oilseed rape is a mass flowering crop that offers huge amounts of food resources in a short period of time.
We sampled landscapes using a grid design replicated in space and time. This multiscale approach allowed us to investigate:
(1) How different sampling designs can affect the estimate of biodiversity patterns.
(2) How the effect of landscape composition on a herbivore-parasitoid trophic interaction changes with time
(3) How landscape composition affects spatial community similarity at the landscape scale.
Chapter 1 introduces the topics covered in this thesis.
Chapter 2 shows how sampling whole landscapes can improve biodiversity estimates. Our aim was to determine how the predictability and stability of these estimates are affected by restricting sampling to only one habitat or to only few sampling points per area. We found that sampling organisms at a small spatial scale can influence the results of ecological studies when they use resources at large scales. Our results show that (i) the number of samples per study area affects the precision of parameter estimation and (ii) the selection of just one habitat type for sampling may generate biased estimates of species richness. Moreover, we observed that the spatial scale of best predictions of the species richness of organisms, which is the landscape sector experienced by organisms, changes with the sampled habitat type. We conclude that estimates obtained by sampling limited to one habitat type or with few replicates per area will often not be representative of the landscape-wide population or community. Therefore, caution should be taken when generalizing too widely from such restricted studies.
In Chapter 3, we apply multi-level generalized path analysis to understand how the dynamics of agricultural landscapes affects the tritrophic interaction between pollen beetles (Brassicogethes aeneus F), their host plant oilseed rape and their parasitoids. We investigate how the effect of landscape composition (percentage of oilseed rape fields and non-crop areas) on herbivore and parasitoid abundance depends on the temporal scale of observation and whether system dynamics showed interannual carry-over effects. We found that the effect of oilseed rape fields on beetle abundance changed with time from negative to positive. Parasitism had a negative effect on the number of pollen beetles, but only in areas with a low proportion of oilseed rape. Interestingly, our path analysis revealed that landscape composition affected herbivore abundance one or two years later, mediated by changes in parasitism. These results indicate an interannual carry-over effect on plant-herbivore-parasitoid interactions, as the insects are affected by landscape composition and top-down effects in previous years.
In Chapter 4, we study the effect of landscape composition on spatial community turnover of pollinators. We compared solitary bee and hoverfly communities, both of which provide important pollination services, yet often show contrasting responses to landscape context. Our grid design allowed us to sample whole landscapes and provided insights on the exchange of individuals between habitats. While the proportion of oilseed rape did not explain spatial community turnover for either taxon, the flowering period influenced the effect of semi-natural habitats, which promoted the homogenization of bee communities during oilseed rape flowering and of hoverfly communities after flowering. The transience of the effect indicates that this pattern was caused by increased movement of individuals between habitats. This spatial homogenization of the community can be important to stabilize pollination in crop fields and to promote community resilience after disturbances, which is of socio-economic importance in agricultural landscapes.
In summary, the different habitat types that compose heterogeneous agricultural mosaic landscapes can play different roles for tritrophic interactions and the structure of insect communities. Larger proportions of semi-natural habitats may increase stability of pollination and biological control by benefiting solitary bees, hoverflies and parasitoids. Mass flowering crops can influence insect communities and population densities, affecting ecosystem services both positively and negatively. This effect may be temporary, limited by the flowering period, or may extend to the following years, particularly for specialized insects, such as the pollen beetle and its parasitoids. This diversity of effects from different landscape components can be best investigated when all habitat types are sampled. Results originating from only one habitat type cannot be extrapolated to the whole landscape. In conclusion, landscape planning should take into consideration the movement of organisms between habitat types and through time in order to guarantee conservation of ecosystem services and crop yield.||de