Building urban predator-prey networks using camera traps

Abstract

Loss of interspecific interactions often precedes extinction events. Therefore, knowledge of species interactions is important to inform conservation strategies aimed at maintaining biodiversity in a changing world. Collecting data on species interactions can, however, be logistically challenging and costly. Hence, alternative data collection and processing methods are needed, and camera traps potentially represent a cost-effective way to identify the predator-prey interactions needed to generate trophic networks. Here, we compare two types of datasets, derived from camera traps, and test their effectiveness to build bipartite predator-prey networks: one based on predation events (prey observed within a predator's jaws), and one based on potential encounter events (i.e., predator and prey species detected within a defined time threshold). Our analyses focused on two predators, coyotes (Canis latrans) and red foxes (Vulpes vulpes), in an urbanized area (Toronto, Canada). Our findings suggest the two datasets provided complementary information on the bipartite networks' structure: predation events underestimated large mammal prey, and potential encounter events underestimated small mammal prey. Together, the predator-prey interactions that we found reflected previous literature on dietary analysis. To improve the information provided by camera traps on trophic interactions, we suggest taking into account site-specific conditions (e.g., presence of birdfeeders or microhabitat differences), as these may alter both predation events and potential encounter events detected. Furthermore, to improve the accuracy of the predator-prey interactions obtained with potential encounter events, time interval threshold and detection sequence (i.e., prey or predator detected first) should be taken into account when defining a potential encounter event. Potential encounter events provide a promising complementary method for assessing predator-prey interactions from camera traps, in particular when predator photos from camera traps are not available. Future research should devise strategies to estimate and include predator-prey predation likelihood in bipartite networks derived from potential encounter events.