Applying GIS and Remote Sensing to Avian Research: A Case Study on the Ouachita National Forest

Preliminary first year results showed that a total of 3,201 birds were encountered representing 61 species, of which 31 species were neotropical migrants. Habitat data collected at the 524 bird census points were used along with more current multi-temporal Landsat TM imagery to classify the study area into actual vegetation categories. The spatial arrangement of these categories was used to examine previously over or under sampled areas and to modify bird census points for the 1996 sampling season. In addition, analyses of avian communities as they relate to stand and landscape level habitat parameters were initiated that incorporated the spatial modeling features of GIS.

The Ecosystem Management Research Study originated during a field trip by Arkansas Senator David Pryor and USDA Forest Service Chief Dale Roberston in the Ouachita National Forest in 1990. The trip resulted in the elimination of clearcutting in the Ouachita National Forest in favor of other silviculture methods (Mersmann et al., 1994). In 1990, the USDA Forest Service established a three-phase ecosystem research study on the Ouachita National Forest that also included some areas on the Ozark National Forest (Baker, 1994; Wilson and Guldin, 1990). The three-phase study involved researchers, managers, and citizens using the National Forests.

The approach of the first phase (Phase I) was to establish demonstration stands as soon as possible; the second phase (Phase II) was to take a statistical approach at the stand level to look at alternative silviculture treatments; and the third phase (Phase III) was to address questions at a larger watershed scale. Objectives of the Phase III avian research project are to: (1) characterize avian abundance, species richness, and diversity at both the stand and landscape/watershed levels, (2) develop and validate models for predicting breeding season avian community composition (presence/absence and abundance) based on stand-level habitat characteristics and landscape parameters, (3) relate abundance, distribution, and frequency of occurrence of brown-headed cowbirds and avian nest predators (crows and jays) to habitat and landscape characteristics, and (4) establish and collect data from roadside bird survey routes to obtain baseline data for long-term monitoring purposes and compare these data to those obtained from an off-road, stratified sampling design. This paper will describe how GIS and remote sensing technologies have been utilized for planning and designing this study, and how these technologies will be utilized to analyze and interpret data collected throughout the course of this long-term project.

Study Area

Figure 1. Location of the four Phase III watersheds near Hot Springs, Arkansas.

Sampling Design and Planning

These GIS layers were used to determine the location of more than 72 miles of transects and 524 plots for the first year measurements. The transect line locations were established to bisect as many spectral and geophysical types as possible within the four watersheds. These transects were mapped and used to establish the transect and plot locations in the field.

Bird Surveys

Figure 2. A location map of the bird censusing points in the North Alum Creek, South Alum Creek and Bread Creek watersheds.

Figure 3. A location map of the bird censusing points in the Little Glazypeau watershed.

Bird surveys were conducted from sunrise to 10:30 hrs with each plot visited by three individuals from May 9 to June 8, 1995. Surveys consisted of individuals walking transects, stopping at each point, and recording birds seen or heard within the plot, respective to number, gender (sight or song), and vocalization (singing or not singing), during a 5-minute sampling period. This was recorded on a bird census data sheet. Birds flying over plots or observed between plots were not included in the census, but their presence was recorded within the watershed. Timing of point surveys and the order in which points where surveyed varied between observes so as to diffuse sampling for a specific point across time of day and season.

Habitat Sampling

Collection of habitat data started in May and was completed by August. Habitat features were quantified at the center of each bird plot and on three 3 meter-subplots located 25 meters from the bird census point. Each subplot was individually marked and arranged as detailed in Figure 4. Location of subplots was adjusted at times to avoid bodies of water.

Figure 4. Habitat sampling scheme for breeding bird census points.

Habitat measurements at the bird census points included slope, aspect and basal area by tree species. Additionally, height, DBH, type (hardwood or conifer), and decay class of snags > 2 meters tall and > 10 cm in d.b.h. that was within 25 meters of the bird census point were tallied. Percent of semi-permanent water (absent only in summer of dry years) covering each bird plot (50 meter radius) was also recorded.

At each subplot habitat structure was assessed by measures of basal area, canopy heights, and canopy coverage for both conifers and hardwoods. Percentage of crown closure was measured at each subplot for both conifers and hardwoods using a spherical densiometer. Heights of residual density and vertical profiles were also recorded for each subplot. Vegetation density and vertical profiles were also recorded for each subplot. Vegetation density was measured at 0.25, 1.25, and 2.25 meters using a density board placed at the center of the subplot and read at 15 meters in direct line with the bird point. Vegetation profiles where taken at four points (90 degrees from one another) around the perimeter of each subplot using a 10 meter extending pole. As the pole was extended the presence or absence of vegetation touching the pole at 1 meter increments up to 10 meter was recorded for conifers and deciduous (including herbaceous) vegetation.

Of the 524 bird survey points, 227 were used to census herpetofauna populations and an additional 26 plots were established as aquatic sites for herpetofauna surveys. Habitat measures at herptofauna plots included the above measures and estimates of ground cover. Two 2 meter by 2 meter quadrants, each containing a 1 meter by 1 meter quadrant was placed in each subplot (Figure 4). Vegetative cover < 1 meter tall was recorded for trees, shrubs, and vines within each 2 meter by 2 meter quadrant. Additionally, percent cover of down wood 1 - 10, 10 - 25, and > 25 cm in diameter was recorded; coverage of 10 - 25 and > 25 cm diameter groups were estimated by decay class. Within each 1 meter by 1 meter quadrant, cover of herbaceous vegetation, rock, moss, bare ground, and litter was estimated. Litter depth was also recorded.

The number of neotropical migrants observed within a watershed appeared directly related to intensity of management with South Alum Creek having the fewest birds (220), and Bread Creek (305 birds), North Alum Creek (429 birds), and Little Glazypeau (627 birds) with greater numbers. Pooling the data by watershed, species diversity was calculated for each watershed using the Shannon-Weaver Index. Species diversity was highest on the Little Glazypeau watershed (3.19), followed by North Alum Creek (3.09), Bread Creek (2.82), and South Alum Creek (2.40). Species richness (number of species recorded) exhibited a similar trend with highest richness on the Little Glazypeau watershed (53), followed by North Alum Creek (50), Bread Creek (27), and South Alum Creek (26).

Preliminary statistical analyses of bird populations between watersheds were accomplished by pooling individual observations (n = 3) by plot, and then calculating number of individuals, species richness, and species diversity. The Little Glazypeau watershed had the highest (P < .05) mean number of individuals, species richness, and species diversity per plot for both neotropical migrants and resident birds compared to the other three watersheds. North Alum Creek had more individuals per plot, higher mean species richness, and higher mean diversity compared to Bread Creek and South Alum Creek (P < .05) for neotopical migrants. For resident birds, North Alum Creek, Bread Creek, and South Alum Creek had similar (P > .05) mean number of individuals, species richness, and species diversity per plot.

Results of Integrating Remote Sensing and GIS in Avian Research, Now and in the Future

Current multi-temporal Landsat TM imagery (February 21, 1995; April 26, 1995: and July 15, 1995) was used along with the 524 vegetation plots to develop a detailed vegetation classification over the four watersheds. The classification is based on linking spectral and geophysical characteristics to canopy vegetation species, basal area by species, percent canopy closure, and understory species composition. This is being used to assist in the establishment of approximate 500 new bird census points in the spring of 1996 and as a base layer to examine spatial relationships between vegetation types and bird species. Also because the road layer was outdated, a new road geographic layer was developed by locating more than 100 miles of roads with a GPS in and around the four watersheds.

In the future analyses will be done using the 30+ geographic layers to measure landscape properties surrounding each bird census plot, as well as across watersheds. Bird data from individual plots will be linked though GPS-obtained coordinates with landscape parameters such as spectral vegetation class size, perimeter, and perimeter/area ratios. Distances from bird census plots to specific habitats (e.g., the nearest mature forest, edge, or riparian area) within the surrounding forest matrix will be computed. Using GIS to buffer around bird census plots, the proportion of different vegetation classes within specific distances of the plots will also be measured. Overall watersheds will be characterized using measures such as mean vegetation class size, perimeter, and perimeter/edge ratios. In addition, sums of induced and inherent edges, and proportions of vegetation classes will be computed. Most of these measures relate to habitat size, isolation, and connectivity, thus serving as indices to habitat fragmentation. Listed below in Table 1 are some of the measures that will be examined.

Table 1. Spatial measures of landscape structure.
_______________________________________________________________

Related to Individual Bird Census Plots

Population declines in migratory birds in eastern North America. Current Ornithology 7:1-57.

An overview of Stand-Level Ecosystem Management Research in the Ouachita/Ozark National Forests. IN: Baker, J. comp. Proceedings of the Symposium on Ecosystem Management Research in the Ouachita Mountains: Pretreatment Conditions and Preliminary Findings. October 26-27, 1993; Hot Springs, AR. Gen. Tech. Rep. SO-112. New Orleans, LA: USDA, Forest Service, Southern Forest Experiment Station. p 18-28.

Proceedings of the Symposium on Ecosystem Management Research in the Ouachita Mountains: Pretreatment Conditions and Preliminary Findings. October 26-27, 1993; Hot Springs, AR. Gen. Tech. Rep. SO-112. New Orleans, LA: US Dept. of Agriculture, Forest Service, Southern Forest Experiment Station. 259 pp.

Population ecology, habitat requirements, and conservation of neotropical migratory birds. U.S. For. Serv. Tech. Rep. RM-205. 26 p.

Biological diversity, ecological integrity, and neotropical migrants: new perspectives for wildlife management. Pages 24-31 In: D.M. Finch and P.W. Stangel, eds. Status and Management of Neotropical Migratory Birds. Gen. Tech. Rep. RM-229. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Forest and Range Experiment Station. 422 p.

Implementing Ecosystem Management Research: Bringing Researchers, Managers, and Citizens Together. IN: Baker, J. comp. Proceedings of the Symposium on Ecosystem Management Research in the Ouachita Mountains: Pretreatment Conditions and Preliminary Findings. October 26-27, 1993; Hot Springs, AR. Gen. Tech. Rep. SO-112. New Orleans, LA: USDA, Forest Service, Southern Forest Experiment Station. p 10-17.

Habitat area requirements of breeding forest birds of the middle Atlantic states. Wildlife Monograph 103:1-34.