Spatio-statistical Predictions of Vernal Pool Locations in Massachusetts: Incorporating the Spatial Component into Ecological Modeling

Abstract

Vernal pools are small, isolated, depressions that experience cyclical periods of inundation and drying. Many species have evolved strategies to utilize the unique characteristics of vernal pools; however, their small size, seasonal nature, and isolation from other, larger water bodies, suggest increased risk of damage or loss by development. The objectives of this research were to statistically determine physical predictors of vernal pool presence, and subsequently, to represent the output cartographically for use as a conservation tool. Logistic regression and Classification and Regression Tree (CART) methods were used to identify important predictors of 405 known vernal pools across northeastern Massachusetts. The CART models performed most favorably, achieving map accuracies as high as 97 percent and providing a set of rules for vernal pool prediction. It is important to note that we observed significant discrepancies between model accuracy and map accuracy, illustrating the pitfall of relying on statistical metrics alone (e.g., R2 values) to assess the quality of spatial analyses.

Department

Natural Resources and the Environment

Publication Date

1-1-2013

Journal Title

Photogrammetric Engineering and Remote Sensing

Publisher

American Society for Photogrammetry and Remote Sensing

Digital Object Identifier (DOI)

10.14358/PERS.79.1.25

Document Type

Article

Rights

© 2013 American Society for Photogrammetry and Remote Sensing

Link to Full Text

Share

COinS