Modeling Species Niches and Distributions:
A particularly important tool for our research is the use of species occurrence localities (especially museum/herbarium records), environmental data, and Geographic Information Systems (GIS) to model species geographic distributions (Anderson 2012). The niche-based nature of these models allows for synthetic studies of evolutionary ecology and biogeography. Additionally, these techniques are also relevant to research on global climate change, invasive species management, zoonotic diseases, and conservation biology.
The lab has used these approaches in two areas: biogeography and conservation/management.
First, the lab is working to integrate niche/distributional modeling into evolutionary and biogeographic studies. This work is supported by a grant from the National Science Foundation entitled “Testing species limits, phylogeographic concordance, and niche evolution in Madagascar's endemic small mammals." This work is in collaboration with
Predictive model of areas suitable for Bradypus variegatus in the Neotropics; from Phillips et al., 2006; locality data from Anderson and Handley, 2001.
Link Olson (University of Alaska), Sharon Jansa (University of Minnesota), and Steven Goodman. Specifically, we are combining niche models (applied to reconstructed conditions at glacial maxima) with phylogeographic analyses that characterize genetic differences among populations across the range of each species. This project continues a similar line of research, at a higher taxonomic level, funded by a previous grant from NSF “Integrating systematics and GIS modeling: biogeography of spiny pocket mice (Heteromyidae) in South America." In that research, we collaborated with Duke Rogers and his students at Brigham Young University, who sequenced mitochondrial and nuclear DNA and conducted phylogenetic analyses.
In our Neotropical work, we have studied ecological issues with important effects on both biogeographic and conservation-related research. One project studied the relevance of local habitat heterogenetiy on estimates of species niches and suitable areas (Soley-Guardia et al. 2014). Complementarily, another empirical study yielded conceptual models of how changes in the distributions of key biotic interactors can affect species ranges and alter population connectivity (Gutiérrez et al. 2014).
For direct conservation applications, our contributions are aimed at studying the present ranges of rare species and making projections of the possible effects of climate change. Our current work in this area is being led by Maria Gavrutenko and Beth Gerstner. They are conducing studies for montane species in Madagascar and the northern Andes, leading toward conservation assessments that include the effects of both deforestation and climate change.
For many years, Rob Anderson and the lab have worked on the development of a technique from machine learning (Maxent; Maximum entropy) to model species niches and distributions. Rob collaborated with Steven Phillips then at AT&T-Research and Robert Schapire at Princeton (and in coordination with Miroslav Dukík; then also at Princeton, now at Microsoft) in early development of Maxent (Phillips et al. 2006; download software). Maxent is among the highest-performing techniques (Elith et al. 2006) and is being used by myriad researchers worldwide (Google scholar).
Subsquently, Maxent has received substantial clarifications and advancements, and the Anderson lab has participated in this. In doing so, we have made contributions also relevant to other techniques for modeling species niches and distributions. For example, we have conducted methodological studies to increase the robustness of Maxent to biases in sampling (Anderson and Gonzalez 2011; Boria et al. 2014), aid in quantifying performance and estimating optimal model complexity (Shcheglovitova and Anderson 2013; Radosavljevic and Anderson 2014), and clarify principles for selecting the appropriate study region (Anderson and Raza 2010).
Rob also was involved in a collaborative book project synthesizing the field of ecological niche modeling led by A. Townsend Peterson. It was published in 2011 at Princeton University Press (see Publications). Rob and the other authors hope that those years of effort led to a product that proves helpful to diverse readers.