Comparing methods for predicting the impacts of invasive species

Abstract

With the accelerating spread of invasive species around the globe, there is an urgent need to predict where invaders are likely to have the largest impacts and what those impacts will be. Predicting the impacts of individual invasive species a priori has proven to be a problematic endeavor, and several possible methods for doing this have been proposed. In this study we use a model system to compare the different predictive methods that have been proposed to determine which provides the most accurate predictions. Community composition in this system was measured at two separate sites in the Gulf of Maine 20 years ago, just before the arrival of the invasive Asian shore crab Hemigrapsus sanguineus. We repeated the community assessment at each site and determined the overall impacts of this invader over the last two decades. We then predicted the likely impacts based on the historical community composition using 3 methods: the invasion history of this species at sites further south along the mid-Atlantic, scaling up per capita consumption rates that have been measured on individual species in the recipient community, and using mesocosm experiments that used a range of experimental approaches. We show that the invader has had clear, but moderate impacts on the invaded sites since its arrival. We further show that for this omnivorous consumer, predictions based on invasion history and on long term field mesocosm experiments performed much better than predictions based on scaling up from consumption on individual species or based on short term mesocosm experiments conducted in the lab or field. These results highlight potential biases that can be produced by some mesocosm approaches and by scaling up from per capita consumption on individual prey species measured in isolation to predict consumer impacts under more natural conditions.