Vernal pool occurrence in southeastern Vermont as determined by soil classification and topographic parameters

McLane, Elisabeth
Jon Atwood, PhD
Department of Environmental Studies
2000
Because vernal pools in the northeastern United States are often small and hidden by forest cover, improved methods for locating them would help considerably with conservation efforts. This study sought to develop a quantitative, predictive model for locating vernal pools based on existing GIS data. Eighty-one natural pools known to support breeding amphibians and 72 randomly-selected sites were located with GPS, and their locations integrated with GIS data layers showing soil and topographic features. Additional soil and topographic data were field collected at each site. Of the 72 random sites, 16 (22.5%) contained vernal pools with definite potential for amphibian breeding. Forty-five percent of pools were located under hemlock-dominated cover which would have made their detection unlikely from aerial photos. With soil categories grouped according to depth and drainage features, pools were significantly correlated with soil groupings. The majority of pools occurred on soils with shallow or moderate depths to bedrock. Logistic regression analysis was used to examine the relative significance of 14 field collected and GIS-generated variables. Two GIS-generated variables showed significant relationships to vernal pool presence/absence: slope (from topographic maps) and minimum depth to bedrock (from soil maps). Four of the field collected variables showed significance. The model created using only GIS-generated data was inadequate, primarily because of its inability to predict sites where pools were absent. Adding the four significant field-collected variables increased the model’s predictive value, but negated the original objective of modeling vernal pool presence on the basis only of regionally-available GIS data layers. The addition of bedrock coverages, which might describe site features missing in the topographic and soils data, could possibly increase the effectiveness of a GIS-based model.

Read Full Text