Understanding the role of feeding behaviour in the rewiring of food webs

It is widely held by theoretical ecologists that ecological network structure explains the dynamics and functioning of communities and ecosystems (Bohan et al. 2013 ; Ings et al. 2009). Understanding the structure of ecological networks, such as food webs, is therefore expected to lead to better prediction and resilience of the functions we want to support, such as biological control in agricultural situations. Considerable work has been done to construct food webs for economically important systems. What is often missed in the apparent complexity of food webs is that these structures are essentially static snapshots of a particular community. There is little dynamic interaction in these food web representations and therefore little of the trophic behaviour that has been extensively studied in ecology.
Carabid beetles are an abundant and diverse family of the Coleoptera, which in arable farmland play an important role in the regulation of weeds, via the predation of weed seed, slugs and aphids (Bohan et al. 2000, 2011 ; Symondson et al. 2002). While laboratory feeding studies have shown that carabid species include specialists on these food groups, the majority of carabids are generalists or omnivores. These non-specialists have marked preferences for certain types of food items, but will switch to others depending upon their availability. Sandrine Petit and colleagues have extensively studied these choices and feeding behaviours that carabids have for different species of weed seeds and have suggested that these preferences may be due to seed traits (Petit et al. 2014). More recently the group at Dijon have incorporated competition and predation risk into the explanation of food item choice. This demonstrated ‘choosiness’ for food items, which is carabid species specific and contingent on the predation and competition context of feeding trial (Charalabidis et al. 2017, 2019). For example, individuals of Harpalus affinis takes seed food items more quickly, irrespective of the weed species and its quality, when subjected to a risk of predation by other carabid species; it becomes less choosy (Charalabidis et al. 2017). These phenomena would suggest a hypothesis that carabid and weed seed food webs should be highly dynamic, with the carabids switching between species of weeds depending on their preference, on weed species availability and on a level of choosiness that depends on the other carabid species within the network. This context-dependent variation in network structure is known as network rewiring.
When replicated food webs of carabids and weed seeds are examined, it becomes clear that the network structure and weed regulation are related. We find that weed seed regulation (i.e. the biological control of weeds), when measured as a change in the weed seedbank across the year, is related to carabid and weed seed interaction frequencies within these food webs. This is consistent with network rewiring with the different contexts producing different interaction combinations. However, this replicate-based comparison is not a test of switching or a concrete demonstration of rewiring. For that it would be necessary to show that the same individuals, when subjected to different contexts, switch to other prey items and rewire the food web.
A manipulative study to demonstrate rewiring would be extremely difficult to construct in ‘real-world’ agriculture. Our approach for this PhD is to use a combination of network modelling that has been developed within the EU PREAR project (Pocock et al. 2018, submitted) and laboratory and literature expectations of the feeding behaviour of carabids (Honek et al. 2003, 2007, 2011 ; Petit et al. 2014) to predict feeding-behaviour explicit and dynamic food webs of carabids and weeds. These expectations will be tested against fieldwork data for carabids and weed seeds that are replicated in time and come from existing databases and data collected in the C-IPM BioAWARE project. Predictions of weed-carabid network could also be tested against molecular DNA data of carabid gut contents as probably the highest resolution data on carabid feeding when this becomes available.