Many recent field experiments have examined plant responses to global change factors such as climate warming, elevated atmospheric CO2, increased nitrogen addition or altered precipitation simulated at the plot level, yet the mechanisms underlying these responses can be difficult to isolate. One concern has been that the infrastructure used in these experiments can restrict the access of influential herbivores, detritivores or pollinators to the plots, and the absence of these animals is confounded with the treatment effects. However, in this paper we describe why free access by animals to experimental plots does not ensure realistic animal densities in response to global change treatments. On the contrary, much like moths swarm around streetlamps, animals that prefer the local conditions in treated plots may congregate at artificially high densities, or conversely, those that are repelled by the treatments may choose to avoid them. Therefore, animal densities or herbivore damage in the plots of global change experiments may grossly exaggerate or underestimate the contributions of animals to primary productivity or plant species composition under future environmental conditions. We describe how these potential animal congregation and avoidance artifacts may have been overlooked in the interpretation of results from many plot-level global change field experiments. We also provide suggestions for how to best interpret the results of these experiments and how to isolate the effects of animal density artifacts.