Reconciling museum records and ecological surveys in biogeographic analyses of New England’s ant fauna

Background/Question/Methods

Museum records represent a large, untapped resource for biogeographic and ecological analysis, but they are difficult to use with many statistical methods (such as niche envelope modeling) because historical specimen records usually are lacking precise geographic coordinates for the site of collection. To incorporate historical records of ant occurrence in New England into a quantitative analysis, we aggregated >27,000 museum records by state county (n = 77 New England counties). Museum records do not include fine-scale measurements of vegetation or consistent habitat designations, but they provide large sample sizes and good geographic coverage. We also compiled climatic variables, interpolated to the geographic centroid of each New England county. Regression analyses at the county level allow for some statistical control of county area and sampling effort and reveal influences of latitude, elevation, and temperature on species richness. We compared the results to analyses of a more standardized ecological survey of forest and bog habitats in a limited number of sites in Vermont and Massachusetts.

Results/Conclusions

The earliest collections were from April 1868 in New Haven, Connecticut, and the most recent were from November 2011 in Boston, Massachusetts. The temporal pattern of specimen accumulation expanded from base locations in urban centers with museums and universities in the late 1800s, through larger field sampling programs in the 1970s, 1990s, and 2000s. Although the accumulated record includes representatives from all counties, there is a persistent bias towards over-sampling of three eco-regions: the Northeastern Coastal Zone, Maine's Acadian Plains and Hills, and the Atlantic Coastal Pine Barrens. A total of 131 species of ants (including 13 non-native species, 7 of which are tropical tramps) have been recorded in New England. The Chao1 asymptotic estimator is 149 species. Of the estimated 18 undetected species, 11 species have been recorded from comparable habitats in adjacent counties in New York and in Québec. Regression analyses gave qualitatively similar results for museum records and standardized ecological samples. Our results suggest that aggregation of museum records to the county level is a valid method for ecological analysis, although care must be taken to control for surveys of specialized habitats, and to properly aggregate multiple records from large ecological surveys to avoid pseudoreplication.