Using multi-level generalized path analysis to understand herbivore and parasitoid dynamics in changing landscapes.

Context: In patchy environments, such as agricultural landscapes, both spatial and temporal scales of habitat heterogeneity can affect population dynamics and trophic interactions. As a result of crop rotation, landscapes and local resource availability may change dramatically within and between years. Objectives: We used a tritrophic interaction constituted by pollen beetles, their host plant oilseed rape, and their parasitoids, as a model system to investigate how the effect of landscape composition on insect abundance changes with time and whether system dynamics showed carry-over effects of previous years. We employ path analysis models that allow us to study whole networks of hypotheses rather than univariate cause-effect relationships. Methods: We exposed pan traps in a 5×5 grid design within 10 landscapes in June 2011 (after oilseed rape flowering) and May 2012 (at peak oilseed rape flowering). Additionally, we assessed parasitism rates of pollen beetle larvae in May 2011 and measured changes in landscape composition. Results: The effect of the oilseed rape proportion on beetle abundance changed with time from negative (during flowering) to positive (after flowering). Parasitism had a negative effect on the number of newly emerged pollen beetles, but only in landscapes with a low proportion of oilseed rape. Interestingly, our path analysis showed that landscape composition affected herbivore abundance 1 or 2 years later, mediated by changes in parasitism. Conclusions: Our results suggest that plant-herbivore-parasitoid interactions in dynamic agricultural landscapes can show interannual carry-over effects, as they are affected by landscape composition and top-down effects in previous years.