The dynamic nature of cereal food webs challenges the suitability of snapshot sampling for assessing ecosystem services

Current theory of food web dynamics is based on mid- to long-term data. However, the short-term dynamics are still poorly understood, as the technology to study them is not readily available. We addressed this gap, by sampling invertebrate food webs bi-weekly, with generalist and specialist predators, herbivores and detritivores as prey, in barley fields over two years. We measured weighted connectance, community composition and link rewiring dissimilarity, as well as species-level specialisation of predators. We expected the connectance to be lowest, and both dissimilarities to be highest during the mid-season, whereas specialisation was predicted to follow prey abundance. We found differing trends across years, with connectance declining over time, and community and rewiring dissimilarity fluctuating more in the second year than the first, when an aphid infestation occurred. Specialisation varied across the season, with aphids and cereal leaf beetles being consumed by more predator species as their abundances peaked, and by fewer before and after that, while springtails were universally consumed. Our results show that agro-ecosystems are highly dynamic, and that food webs undergo continuous restructuring even over very short timeframes; potentially leading to a considerable systematic under- or overestimation of ecosystem functioning assessment.