Species distribution models (SDMs) or, more specifically, ecological niche models (ENMs) are a useful and rapidly proliferating tool in ecology and global change biology. ENMs attempt to capture associations between a species and its environment and are often used to draw biological inferences, to predict potential occurrences in unoccupied regions, and to forecast future distributions under environmental change. USGS scientist Robert Fisher and colleagues at the University of California, San Diego, used an economically important and ecologically damaging invasive species, the Argentine ant, to quantify how issues relating to region and spatial scale affect ENM predictions. Their study, published recently in Global Ecology and Biogeography, offers basic methodological and conceptual guidelines for appropriate sampling and scale matching.

The authors used two independently collected, high-resolution presence/absence data sets from two adjacent and climatically similar regions of southern California, where Argentine ants are widespread. They systematically analyzed how the accuracy of model predictions and the inferences drawn from them hinge on sufficient sampling, independently collected data from different regions, and the spatial scale of environmental variables. They found that model predictions were strongly influenced by the thoroughness of sampling. Moreover, they found that variables changed in importance at different spatial resolutions: climatic factors became more important at coarser resolutions, while habitat variables became less important. These results were due in part to changes in the sampling distribution and in the performance of variables at different spatial resolutions.

In light of these findings, the authors recommended that efforts to model species distributions: 1) use both presence and absence data when appropriate, 2) sample across the environmental tolerance of a species, 3) sufficiently sample the environmental parameter space of the region into which predictions will be made, 4) test model predictions in a distinct region with independently collected data, 5) use variables at the appropriate spatial scale, and 6) make model predictions at the same spatial resolution as model parametrization.

Management Implications

• Insufficient sampling of environmental parameter space can cause large errors in predicted distributions and biological interpretation.
• Models perform best when parametrized with data that have sufficiently sampled environmental parameter space.
• Altering the spatial scale of analysis changes the relative importance of different environmental variables.
• Exercising appropriate levels of caution in efforts to model species distributions will increase the likelihood that the interpretations and predictions of SDMs are biologically meaningful.

Menke, S. B., D. A. Holway, R. N. Fisher, and W. Jetz. 2009. Characterizing and predicting species distributions across environments and scales: Argentine ant occurrences in the eye of the beholder. Global Ecology and Biogeography 18:50–63.

Download this publication brief in pdf format