The North American Breeding Bird Survey (BBS), which is coordinated by the Unoted States Geological Survey and Canadian Wildlife Service, is a primary source of population trend and distribution information for most species of North American birds. Several recent summaries of the BBS have been published (e.g., Peterjohn and Sauer 1992, Droege 1990, Droege and Sauer 1988, 1989), hence we will only briefly review the operational details of the survey. The survey unit is a roadside route, which is 39.4 km (24.5 miles) long. An observer surveys the route once each year during the peak of the breeding season, primarily during June although routes in desert regions and some southern states are surveyed during May. The observer stops at 0.8 km (0.5 mile) intervals, and records all birds seen or heard within a 0.4 km radius circle of each stop during a 3-min sampling period. The starting point and direction of each route is randomly located within a degree block of latitude and longitude (Robbins et al. 1986, Droege and Sauer 1990).
After experimental surveys were conducted in Maryland and Delaware in 1965, the BBS has expanded to cover the continental United States and southern Canada. The survey was initiated in different years in different parts of its range. BBS routes were run only on routes in the United States east of the Mississippi River and in Quebec and the maritime provinces of Canada in 1966. In 1967, the BBS extended to the Central United States, with a few routes in Ontario and Manitoba. By 1968, all of the continental United States was covered, and routes were run across southern Canada. Routes in Alaska, Yukon, and Northwest Territories were added during the early 1980s.
Many geographic patterns appear in BBS data. Some of these patterns are simply a result of deficiencies in sampling design. For example, BBS routes near human population centers tend to be surveyed consistently but remote routes are not surveyed every year, which causes regional variation in the efficiency of the survey (Robbins et al. 1986). Although all States and Provinces vary in coverage over time, some consistent regional patterns occur in BBS coverage. Coverage in the Northeastern United States (excluding ME) is most extensive, with the highest density of BBS routes (up to 16 per degree block of latitude and longitude) occurring in MD, DE, and NJ (See also Table 1). Routes are much less densely distributed elsewhere with maxima of 4-5 routes / degree block in a number of states while NV and portions of some Canadian Provinces have a minimum of 1 route / degree block. Route densities are less than 1 route / degree block only in arctic and boreal regions. In Canada, the BBS is largely restricted to the south, and our summary maps truncate the range of each species to indicate the region with sufficient BBS data.
Superimposed on the regional variation in route densities is temporal variation in the number of routes run in many states. Most states show a general pattern of increasing routes over time. In some cases, this pattern is extreme, with states such as ME having few routes until the early 1970's, when numbers started to increase. Some states (e.g., UT) have had consistently low numbers of routes over most of the survey period, while other states (VT, SC, SD) have had periods when the number of routes decreased. For some States and Provinces (e.g., Alaska, Yukon, PEI and NFD), the small time period (years) over which routes were run clearly invalidates any trend analyses from the data.
We evaluated overall sampling efficiency of the BBS for each bird species by examining sampling attributes over the entire range of the BBS, and categorized them as (1) not sampled by the BBS, (2) small sample-size, (3) highly variable, or (4) low relative abundance. Possession of one of these attributes does not necessarily eliminate the species from trend analyses. These species can be well surveyed by the BBS within portions of their breeding range or during certain time periods. However, long-term regional or survey-wide trend estimates for these species may be less accurate. Numbers presented below were taken from an unpublished manuscript by Sauer and Droege, and use trend data from 1966 - 1990.
Species not sampled by the BBS-North American bird species that were seen on no more than 1 BBS route fall into broad categories of northern breeders (note that Alaskan BBS routes are not included in standard BBS analyses), tropical or Mexican resident species, pelagic or coastal species, accidental species, and exotics.
Small sample-size species-These species have trend estimates with fewer than 14 degrees of freedom, ([number of routes] - [number of strata within states in which the species occurred]) < 14). The 99 species seen on few routes of the BBS fall into several general categories, most of which are related to breeding ranges or habitats of the species. Many northern-breeding species and coastal colonial species are seen on some BBS routes, but must be considered accidental occurrences on the BBS. Many tropical and Mexican species are seen on routes in southern Florida and near the Mexican border. Several species of owls and species with northern ranges are only infrequently recorded on BBS routes. Although some species listed by or under review for Endangered Species status are occasionally detected on BBS routes, the BBS provides little information of use for evaluating their population status. None of the BBS guilds contain a high proportion of species with extremely low sample sizes, and overall, 20% of all species seen on BBS routes are sampled with < 14 degrees of freedom.
Highly variable species-Eighty-four species had trends with large variances. Several heron species, many duck species, waterbirds, a few species with northern breeding ranges, several warblers, some sparrow species, and some blackbirds occur in this list, among other species. These species can be characterized as birds with specialized habitats or limited distributions in the BBS range, spruce-budworm species, or colonial nesting species. Overall, 22% of the species in the survey were in this category, with hunted species, coniferous forest nesting species, waterfowl, and wetland nesting species guilds having > 22% of their species with high variances.
Species with low relative abundance-Ninety-five species had rangewide average counts of < 0.5 birds/route in the BBS. This list has similar species categories as the highly variable and low sample size lists, including the widely distributed nocturnal birds, rails, hawks, ducks, and species whose breeding ranges only partially overlap with regions surveyed by the BBS. Several species that are widely distributed but sampled at low numbers on the BBS occur in this list, such as Wood Ducks, American Woodcock, and Great Horned Owl. Overall, 20 % of the species contained in the BBS data had low relative abundances, and the only guilds with > 20 % of species in this category were the hunted and primary cavity nesting guilds.
Overall, 204 (41%) of the 504 species we analyzed for 25-year trends were in at least one of the categories. The guild with the lowest proportion of species in the lists was urban species (8%), while hunted, wetland nesting, and waterfowl guilds had > 41% of their species in 1 of the 3 categories.
Other potential biases in the BBS cannot be documented from analysis of the survey. These biases have been discussed elsewhere (e.g., Bystrak 1981, Droege 1990), and we will only mention them here to remind BBS users of the potential problems. They include:
Proportion of range in the survey area-All BBS analyses incorporate data only from BBS routes. Analysis of survey data cannot tell us the proportion of the individuals of a species that is breeding outside the range of the survey. Species that are recorded only on the margins of the surveyed area are often of low sample size or are highly variable, but many species (e.g., Canada Goose) may have substantial populations within the survey area. Trends are always specific to the areas surveyed.
Roadside biases-The BBS is a roadside survey, and a major criticism of the survey has been that habitat changes along roadsides may not be representative of regional habitat changes. Trends from the BBS may therefore reflect only populations along roads rather than regional bird population changes.
Habitat biases-Within the range of the BBS, many habitats are not well covered, and species that specialize in those habitats are poorly sampled. Wetland birds and species occupying alpine tundra habitats are examples of groups thought to be poorly represented in the survey.
Even with all of the sampling and other biases discussed above, the BBS represents a unique attempt at a survey of breeding populations of birds in North America. Enormous amounts of data have been collected that provide the only information on regional population trends and breeding distributions of birds. The challenge for the future is to maintain and augment survey effort, and to identify and minimize deficiencies in the survey.
Bystrak, D. 1981. The North American Breeding Bird Survey. Pp.
34-41 in C. J. Ralph and J. M. Scott, eds. Estimating numbers
of terrestrial birds. Studies in Avian Biol. No. 6.
Droege, S. 1990. The North American Breeding Bird Survey. Pp. 1-4
in J. R. Sauer and S. Droege, eds. Survey designs and
statistical methods for the estimation of avian population
trends. U.S. Fish Wildl. Serv. Biol. Rep. 90(1).
-----, and J. R. Sauer 1989. North American Breeding Bird
Survey annual summary 1988. U. S. Fish. Wildl. Serv.,
Biol. Rep. 89(13):1-16.
-----, and -----. 1990. North American Breeding Bird
Survey annual summary 1989. U. S. Fish. Wildl. Serv.,
Biol. Rep. 90(8):1-22.
Peterjohn, B. G., and J. R. Sauer. 1993. North American
Breeding Bird Survey annual summary 1990-1991. Bird
Populations 1:1-15.
Robbins, C. S., D. Bystrak, and P. H. Geissler. 1986. The
breeding bird survey: its first fifteen years, 1965-1979.
U.S. Fish Wildl. Serv. Resour. Publ. 157. 196pp.