An empirical, field-based test of extinction cascades in food webs

Background/Question/Methods

The extinction of species at low trophic levels is predicted to have a cascading effect through food webs, with negative effects on species at higher trophic levels. For example, simulating the sequential loss of prey species results in secondary extinctions; the extent to which secondary extinctions occur is described as food web ‘robustness’. However, these simulations make assumptions about the responses of individuals to the loss of food resources, for example, prey-switching by predators. Depending on the assumptions being made, the estimates of robustness can dramatically vary but these assumptions have not been empirically evaluated in ‘natural’ food webs, partly due to the difficultly of manipulating experimental systems in the field at relevant ecological scales.

In this study I considered birds feeding at bird tables as a model food web. The bird tables were stocked with a wide range of food sources, presented in a range of different ways. Removing food sources was analogous to the extinction of prey, and in doing this I created many potential extinction cascades, thus empirically evaluating the robustness of this food web.

Results/Conclusions

I found that most species of bird were specialised on a few of the food sources available at the bird tables, but with the loss of their preferred food they switched to poorly-utilised food resources. For some species, the overall number of visits remained constant but for others the number of visits reduced rapidly with the sequential loss of preferred foods. The realised ability of birds to switch prey was therefore species- and context-specific. This has important implications for the interpretation of ‘robustness’ from empirically-derived food webs. Of course, one criticism is that birds visiting bird tables is only a subset of an ecosystem, although this is a criticism that is applicable to some degree to most food web studies. However, the easy of studying and manipulating the ‘food web’ of birds visiting bird feeders makes it an excellent system with which to test theoretical simulations estimating food web robustness.