Comparison of Birds Detected from Roadside and Off-Road Point Counts in the Shenandoah National Park

Roadside point counts are generally used for large surveys to increase the number of samples. We examined differences in species detected from roadside versus off-road (200-m and 400-m) point counts in the Shenandoah National Park. We also compared the list of species detected in the first 3 minutes to those detected in 10 minutes for potential species biases. Results from 81 paired roadside and off-road counts indicated that roadside counts had higher numbers of several edge species but did not have lower numbers of nonedge forest species. More individuals and species were detected from roadside points because of this increase in edge species. Sixtyfive percent of the species detected in 10 minutes were recorded in the first 3 minutes. The USDI Fish and Wildlife Service's Breeding Bird Survey (BBS) has been monitoring bird populations from roadside routes since the 1960's. Each survey route is a series of 50 3-minute point counts conducted at 0.8-km intervals along secondary roads. The BBS now has over 3000 routes distributed across North America. The large geographic coverage of this survey would not be possible without using roads to expedite travel between points. However, using secondary roads for survey routes has several potential biases. Geographic biases might occur because road density differs among regions. Large, undeveloped areas with few roads will be undersampled compared to more populated agricultural and urban areas. Road placement may also bias the habitats sampled by the route. For example, roadside surveys are less likely to sample marshes and bogs because roads are not easily constructed in these areas. These potential geographic and habitat biases must be considered when interpreting monitoring data based on road counts (Temple and Wiens 1989). But these sampling biases can be defined and may be addressed by adding routes in specific habitats or areas. A more insidious bias of roadside counts may occur if the species detected from roads differ from those that would be encountered in the same habitat away from the road. The presence of a road in a forested area often creates a break in the canopy and a forest/road edge. Are we less likely to detect or even miss some forest birds and more likely to detect forest edge species from roadside points? Will roadside surveys give an adequate assessment of the bird community in the habitats encountered? To address these questions, we compared point counts of breeding birds conducted on and off the road in the Shenandoah National Park. We also examined any potential biases in the species sampled by staying 1 An abbreviated version of this paper was presented at the Workshop on Monitoring Bird Population Trends by Point Counts, November 6-7, 1991, Beltsville, Maryland. 2 Research Wildlife Biologists, Patuxent Wildlife Research Center, USDI National Biological Service, Laurel, MD 20708 USDA Forest Service Gen. Tech. Rep. PWS-GTR-149. 1995 3 minutes versus 10 minutes at a count to guide future monitoring efforts using point counts. Study Area and Methods The Shenandoah National Park is located in the Blue Ridge Mountains of Virginia and is composed primarily of second growth forest that has developed on farmland abandoned in the early 1900's. Skyline Drive is a two-lane paved road with grassy margins 1-25 m wide and runs 169 km through the Park along the crest of the Blue Ridge Mountains. We selected sites where three point count stations could be placed in a line perpendicular to the road with one point on the roadside and one each at 200 m and 400 m from Skyline Drive or any other road. Sites were constrained to areas where the same forest type occurred from the road to at least 500 m from the road, as delineated on the Park vegetation map (Teetor 1988), and the elevation change was not greater than 60 m among the 3 points. Sites had to be at least 0.8 km from another site. Eighty-one locations meeting these conditions were found along Skyline Drive, and we sampled all of these. At each site, the same observer consecutively sampled the roadside at 200-m and 400-m points, reversing the order of coverage at alternating sites. Observers noted all birds detected in 3 minutes and continued recording for a full 10 minutes to enable comparisons of the number of species detected by increasing the length of the count. Counts were conducted only once at each point. Thus, within a site, the three point counts have the same forest type, elevation, observer, and general time-of-day. For each species, the number counted from the roadside point was compared to the number counted from the 200-m point and the 400-m point in two paired-sample, one-sided, t-tests (Zar 1984). Because the two tests both use the roadside count and are thus not independent, our interpretation of statistical significance was conservative, and we used the Bonferroni adjustment (which halves the critical P value) and consequently accepted P < 0.025 to indicate significant differences between a species' abundance at roadside and off-road points (Miller 1981:15). The paired-sample t-tests were conducted using only the sites where the species was present on at least one of the counts. Results and Discussion Roadside and Off-Road Abundance of Species Thirty-five species were observed on 10 or more sites. Greater numbers of species and individuals were detected at roadside points than at their paired 200-m or 400-m off-road points (table 1). Roadside counts had significantly more Chipping Sparrows (Spizella passerina), Indigo Buntings (Passerine cyanea), American Crows (Corvus brachyrhynchos), Gray Catbirds (Dumetella carolinensis), American