SOURCES OF DATA -- AVIAN DISTRIBUTION Breeding Bird Atlases Breeding Bird Atlas (BBA) data from the New England states are a major component of Gap Analysis vertebrate range data. These atlas data are the result of statewide, grid-based surveys conducted during the breeding season. Observed breeding activity of a species in a grid block results in a species-by-block occurrence record in the state's atlas database. Observations are categorized as possible, probable, or confirmed. Atlas survey methods are derived from similar projects conducted in Europe, where standard codes were established in 1971 (Laughlin and Kibbe 1985). Although many North American atlas projects predate the publication of standard methods in North America, the New England atlases closely follow standards derived from the European efforts. With few exceptions, atlas surveys are based on a standard grid consisting of 5-km square blocks. A block is one-sixth of a USGS 7.5-minute survey map. Atlas survey teams search for breeding birds in each block until a predetermined level of coverage is met. States usually base the standard coverage level on an expected number of breeding birds per block (Bevier 1994). Observers categorize breeding activity in accordance with standard codes. These standard codes are further categorized as possible, probable, or confirmed. The confirmed category contains nine codes which correspond to such direct evidence of breeding as an occupied nest, a nest with eggs, recently fledged young, etc. Examples of the six codes in the probable category are courtship behavior and territory defense. Two codes corresponding to observation of a bird in suitable habitat and to observation of a singing male in suitable habitat indicate possible breeding. An additional code is used for birds observed in the block but believed not to be breeding. Massachusetts used the standard grid and these eighteen standard codes while conducting the first North American state-wide atlas project beginning in 1974 (Bevier 1994). The Massachusetts data have not been published, but were acquired from and used with permission (given verbally to Curt Griffin) from Mass Audubon. Other New England states adopted these methods with only minor modifications. The Vermont BBA comes from surveys done from 1976 to 1981. The Vermont atlas data represent roughly one-sixth of the state's grid blocks (the maximum number that could be adequately surveyed with available resources). In addition to a randomly selected block from each 7.5-minute quad, the block total includes 24 blocks containing areas of unique and fragile habitat. The Vermont coverage standard requires 75 species per block (35 of which must be confirmed) for a block to be adequately surveyed. This standard is based on expected species richness of a block as determined during the first two seasons (Laughlin and Kibbe 1985). The Rhode Island BBA covers breeding seasons from 1982 to 1987. All 165 grid blocks are represented in the database. The codes and coverage standard are only slightly modified. The number of species per block varies enough that each block has been given an estimated number. The block is considered adequately surveyed when 75 percent of this number is recorded with 50 percent of these at the confirmed level (Enser 1992). Connecticut's BBA also began in 1982 and concluded in the 1986 season. Grid blocks and codes follow the Massachusetts example with a coverage standard of fifty species per block. All 596 blocks in the state are represented (Bevier 1994). The New Hampshire BBA resembles Vermont's in that only a sample of the state's grid blocks are represented in the database. New Hampshire's 178 atlas blocks result from a random sampling of 7.5-minute quads. The survey includes data collected from 1981 to 1986 with a coverage standard comparable to that used in Vermont. Breeding codes correspond closely to the standard codes (Carol Foss, Univ. of Maine, pers. commun.). Unlike the atlases of other New England states, the Maine BBA consists of data surveyed on a grid of full 7.5-minute quads. The Maine atlas covers breeding seasons from 1978 to 1983 (Adamus 1987). We created a coverage of atlas grid blocks for New England by combining data for all six states. These state-specific coverages have various origins. The block coverage for Maine was acquired as a quadindex from the Maine Natural Heritage Program. Similarly, the Massachusetts and Connecticut block coverages came from their respective Natural Heritage Programs. We generated coverages of sampled blocks in Vermont and New Hampshire from x,y coordinates. The Rhode Island coverage was also created in our lab by subdividing a quadindex obtained from Rhode Island. We acquired atlas databases in digital format from all New England states except for Rhode Island. Subsequent conversion to ARC/INFO format where necessary allowed display of the data with the block coverage. The Rhode Island BBA was available only in published format and had to be loaded into an INFO file by manually entering data for each species from the published maps. Blocks coded with the same breeding status on a published map were selected on-screen from the Rhode Island block coverage. Corresponding block numbers, along with the currently-selected species and breeding-status codes were then written to a textfile. An AML automated this process which was repeated three times (once each for confirmed, probable, and possible records) for each species. This resulting textfile of block-occurrences was then converted to an INFO datafile. Observations of breeding activity are translated to presence/absence data for use in Gap Analysis. Species with atlas records in a particular block are coded as present in a county or EMAP hexagon when any portion of that block lies within the county or hexagon. At present, all codes (possible, probable, and confirmed) translate to presence. Breeding Bird Survey Another substantial source for Gap Analysis vertebrate range data is the Breeding Bird Survey (BBS). BBS data consist of bird counts conducted at roadside stops during the breeding season. A single route contains fifty stops spaced at half-mile intervals. Observers record all birds seen or heard during a three-minute period at each stop. Ten consecutive stops are compiled into one page of data. The term 'page' here refers to this compilation. Consequently, a full 50-stop route results in five pages of data. We acquired BBS data for the last eleven years (1983 - 1993) from the National Biological Survey (NBS). Compiled as page totals, these data can be graphically associated with their respective stops on a coverage of BBS routes. We constructed a GIS coverage of New England BBS routes by digitizing copies of road maps that had been acquired from NBS (Larry Master, The Nature Conservancy). In most cases these maps were photocopied sections of 7.5-minute USGS survey maps on which the BBS route was drawn in with a highlighter. Using these maps, we digitized the routes on-screen by selecting the corresponding road sections from 1:100,000 USGS digital line graphs (DLG). We acquired the DLGs from the UVM Bailey-Howe library, converted them from DLG to ARC/INFO format, and edge-matched them before using them to digitize the BBS routes. Arcs from the DLG coverage were selected on-screen by visually comparing the DLG and BBS maps. This process of identifying the corresponding roads in the DLG was not difficult; however, errors may have occurred in some cases where the DLG and BBS road patterns disagreed, or where the highlighted route on the maps was confusing or sloppy. Although such errors will have minimal, if any, effect on county- or hexagon-coding, a textfile of notes was compiled during the process of route-digitizing so that possible errors could be tracked down if necessary. Occasionally, the DLG did not contain a road section needed to create an unbroken BBS route. In these cases we digitized in a freehand fashion an approximate route section to bridge the gap. Selected arcs from the DLG were built into a line coverage and unsplit so that each complete route would consist of a single arc. Routes were coded with section and route numbers and were then subjected to ARC/INFO dynamic segmentation processing to divide them into fifths. The one-fifth segments correspond to the ten-stop pages of BBS data. Page totals were then associated with corresponding route segments. Birds included in a particular page total were coded as present in a county or hexagon when a portion of the route segment containing that species fell within the county or hexagon. Unlike the Atlas data, the BBS data contain no categories of confidence. Species observed at a stop are simply recorded as present at that location. Consequently, occurrences in the BBS data are not as strong an indication of breeding status as a confirmed observation in the BBA data. As with the Atlas data, any occurrence of a species in a county or hexagon translates to presence of that species. However, we did eliminate one-time occurrences from our analyses, i.e., a species was recorded as present in a county or polygon if it occurred at least two times on a page total in the eleven- year period. LITERATURE CITED Adamus, P.R. 1987. Atlas of breeding birds in Maine, 1978-1983. Maine Department of Inland Fisheries and Wildlife, Augusta, Maine. 366pp. Bevier, L. R. ed. 1994. The Atlas of Breeding Birds of Connecticut. State Geological and Natural History Survey of Connecticut. 461 pp. Enser, R. W. 1992. The Atlas of Breeding Birds in Rhode Island. Rhode Island Department of Environmental Management. 206 pp. Lauglin, S. B. and D. P. Kibbe. 1985. The Atlas of Breeding Birds in Vermont. University Press of New England, Hanover, Connecticut. 456 pp.