Introducing MERGANSER: A Flexible Framework for Ecological Niche Modeling

Abstract:

Ecological Niche Modeling (ENM) is a collection of techniques to find a ”fundamental niche”, the range of environmental conditions suitable for a species’ survival in the absence of inter-species interactions, given a set of environmental parameters. Traditional approaches to ENM face a number of obstacles including limited data accessibility, data management problems, computational costs, interface usability, and model validation. The MERGANSER system, which stands for Modeling Ecological Residency Given A Normalized Set of Environmental Records, addresses these issues through powerful data persistence and flexible data access, coupled with a clear presentation of results and fine-tuned control over model parameters. MERGANSER leverages data measuring 72 weather related phenomena, land cover, soil type, population, species occurrence, general species information, and elevation, totaling over 1.5 TB of data. To the best of the authors’ knowledge, MERGANSER uses higher-resolution spatial data sets than previously published models. Since MERGANSER stores data in an instance of Apache SOLR, layers generated in support of niche models are accessible to users via simplified Apache Lucene queries. This is made even simpler via an HTTP front end that generates Lucene queries automatically. Specifically, a user need only enter the name of a place and a species to run a model. Using this approach to synthesizing model layers, the MERGANSER system has successfully reproduced previously published niche model results with a simplified user experience. Input layers for the model are generated dynamically using OpenStreetMap and SOLR’s spatial search functionality. Models are then run using either user-specified or automatically determined parameters after normalizing them into a common grid. Finally, results are visualized in the web interface, which allows for quick validation. Model results and all surrounding metadata are also accessible to the user for further study.