A biodiversity project might reasonably begin with consideration of species distributions in the project area. We acquired atlas data for as many plant and animal taxa as possible. Several datasets were digitized from hardcopy sources, while others were acquired in digital form. Species locations were generally recorded as town-level records. We standardized town names where necessary and translated other data types to town-level records. One dataset was based on smaller polygonal sampling units, which were used in preference to towns. Species records were then summarized to generate species totals by town. For each atlas we classified these "richness" data into 10 quantiles and related them to a polygon coverage of Vermont political boundaries. The resulting maps show the distribution of town-level atlas records across the state.

Atlas maps revealed problems or inadequacies in the data more often than they presented an accurate depiction of species distributions in the state. Few atlases originated with a scientifically-defensible sampling scheme. Many atlases suffered from incomplete coverage of the state, showing little or no data for remote, sparsely-populated towns. Other atlas datasets betrayed sampling biases often characterized by intensive sampling near population centers, educational institutions, or areas holding a high degree of ecological interest.

Stoms (1994) noted that because species richness was dependent upon the scale at which observations were collected, choice of sampling unit size should be based on the objectives of the intended application. Knowledge of which species occurred in a particular town may be of little value without knowing which areas of a town were occupied. None of the atlases we considered were designed for use in a detailed biodiversity assessment and some were incomplete at the time of acquisition. These atlases represented valuable baseline data useful for a number of worthwhile applications including partial verification of biodiversity hotspots identified by the Vermont Biodiversity Project. Nevertheless, we decided that an overwhelming majority of the atlases were inadequate for delineating areas of high biodiversity at the desired level of resolution.


An atlas of vascular plants was created specifically for use in the Vermont Biodiversity Project. Frank C. Seymour first compiled these herbarium records, many of which were collected by the Vermont Botanical and Bird Club. We digitized town-occurrence records for 654 species from a subset of 10 families. Families selected were Compositae, Cruciferae, Cyperaceae, Equisetaceae, Gramineae, Liliaceae, Lycopodiacea, Orchidaceae, and Rosaceae. Several towns contained 100 or more species. However, approximately a dozen towns were absent from the database, and 70% contained 35 or fewer species.

We extracted data from a Forest Inventory and Analysis (FIA) to create an atlas of woody trees in Vermont. This dataset had the advantage of point locations of the 625 sample points. The disadvantage of FIA data was that the sampling scheme was designed to provide a regional perspective of species distributions (Hansen et al. 1992). The data were meaningful at the county level in Vermont, but did not accurately represent tree distributions when points were grouped at the town level. At least 40 towns had no records at all, and several densely-forested towns listed only a few species. The FIA atlas contained over 24,000 records representing 60 of Vermont's tree species.

The bird atlas was unusual in several regards. One of the few published atlases, the breeding bird atlas also was one of the few atlases with a scientifically-defensible sampling design. This atlas also stood out for the relatively fine scale of its sample units. Breeding birds were sampled in blocks created by subdividing a U. S. Geological Survey 7.5' quadrangle grid into 6 equal rectangles per quadrangle. Our digital version of the atlas includes 188 of the 195 species thought to be breeding in the state at the time. The only substantial disadvantage of the breeding bird atlas was the age of this dataset, which was published in 1985 (Laughlin and Kibbe 1985).

We acquired a digital version of Cyrus McQueen's (1992) unpublished compilation of herbarium citations, "The Bryophytes of Vermont". Featuring 484 species, this atlas represented only 143 towns. The majority of the 2,255 records were spread across west-central Vermont, although some towns in other parts of the state were also heavily-sampled.

Ferns were also depicted in a separate atlas. Covering 218 towns, this atlas covered more of the state than did the Bryophyte atlas, but the distribution map again revealed uneven sampling efforts across the state. Eighty-six species were documented in this unpublished database of 1977 records (Barrington 1988).

We created a fish atlas from data collected by the Vermont Department of Environmental Conservation as part of an aquatic community classification. Records of 62 species were collected at reference-level and moderately-impacted sites on reach-representative sections of wadeable streams during the period 1984-1997 (Langdon et al. 1998). Acquisition of latitude and longitude at each sample site provided better locational accuracy for this atlas than for most other atlases. However, sampling effort was highly biased toward the Lake Champlain Valley. When locations were generalized to the town level only 62 towns were represented in the 2941 records of this database.

An atlas project for reptiles and amphibians was in progress when we acquired species distributional data. Jim Andrews at Middlebury College provided us with almost 20,000 town-occurrence records collected through 1998. Nearly all towns were represented at this point, although many towns had few species records. Every town in Vermont was expected to have at least 15 species (Andrews 1999). The distribution map showed 175 towns well below this level, indicating a substantial gap between the true distribution of species and current knowledge.

We constructed a mammal atlas from a variety of sources including museum citations, harvest and mortality data, and a small mammal atlas created as part of a PhD dissertation. A database of museum citations was acquired from the Carnegie Museum of Natural History in Philadelphia, Pennsylvania, with permission for their use in the Vermont Biodiversity Project. The 1874 records in this database represent 43 species and 55 towns. The Vermont Department of Fish and Wildlife supplied us with approximately 18 years of data on furbearer tag records, black bear harvest data, and non-hunter moose mortality records. In these databases 3 furbearers are represented in 243 towns, bear in 223 towns, and moose in 182 towns. Another dataset we included was a small mammal atlas compiled from museum records, literature review, and trapping data. Eighteen small mammal species were represented in the 767 records of this dataset (Chipman 1994). Together these sources provide data for all of Vermont's 255 towns. However, 70% of towns recorded fewer than 8 species.


Andrews. 1999. Unpublished draft map.

Barrington, D. S. 1988. Atlas of Vermont pteridophytes. Draft. Pringle Herbarium, University of Vermont, Burlington.

Chipman. 1994. Distribution, relative abundance, and habitat use by small mammals in Vermont. PhD dissertation, University of Vermont, Burlington.

Hansen, M. H., T. Frieswyk, J. F. Glover, and J. F. Kelly. 1992. The eastwide forest inventory data base: users manual. General tech. rep. NC-151, U.S. Dept. of Agric., For. Serv., St. Paul, Minnesota, USA.

Langdon, R., J. Andrews, K. Cox, S. Fiske, N. Kamman, and S. Warren. 1998. A classification of the aquatic communities of Vermont. Unpublished report for The Nature Conservancy and the Vermont Biodiversity Project.

Laughlin and Kibbe. 1985. The atlas of breeding birds of Vermont. University Press of New England, Hanover, New Hampshire, USA.

McQueen, C. B. 1992. The bryophytes of Vermont. Unpublished atlas.

Stoms, D. M. 1994. Scale dependence of species richness maps. Professional Geographer 46:346-358.