Species interactions and their role in niche models: A study on two sympatric mustelids in the Northeast U.S
Ecological niche models (ENMs) are commonly used to predict species range limits and how they shift over time. Most ENMs use abiotic variables to predict suitability, and only include biotic effects implicitly. This approach can produce inaccurate models if interspecific interactions have strong spatial effects on the species’ distributions. We consider the distributions of two North American mustelids, the American marten (Martes americana), a species of conservation concern in much of the northeast U.S., and the fisher (Martes pennanti), a mostly sympatric and larger-bodied congener that competes with and preys upon the marten. Leveraging a large New York State Department of Environmental Conservation dataset, we plan to test three different approaches to incorporate biotic effects into abiotic ENMs: 1) conduct a spatial overlay of abiotic models for the two species, 2) add the interactor’s ENM as a model predictor, and 3) build a joint model that separates shared environmental suitability from residual co-occurrence patterns, a process that aims to elucidate the strength of biotic effects. Here, we present preliminary Maxent models for New York State for the first approach and outline the future steps to be taken.
Preliminary models find increasing relative suitability for marten with increasing elevation. We found a number of areas exhibiting a rapid degradation in relative suitability for marten and a rapid rise for fisher over elevational transitions, mostly in the southwestern Adirondacks and High Peaks regions. Our current interpretation is that this phenomenon may indicate strong local species interactions. In subsequent years, such areas will be field validated for each approach via transect sampling to estimate the models’ ability to predict both occurrence and relative abundance. After validation, we will project the best performing model to contrasting climate change scenarios, and as a final step, make conservation recommendations for marten based on the above results and corridor analyses of modeled suitability surfaces that estimate connectivity of suitable areas through time. We think that this multi-step approach that targets areas with sharp suitability gradients for validation will result in more refined models that take better account of the key ecological drivers shaping the system, and therefore should prove very useful for conservation management.