|
Article Excerpt
The vast boreal region of North America encompasses a quarter of the world's remaining intact forests and is estimated to support 50% or more of the entire breeding population of at least 98 species of birds (Blancher and Wells 2005). A substantial amount of avian research has been conducted in southern boreal forests of Canada during the past decade, much of it focused on understanding the composition of landbird communities relative to forest structure and ecological processes (e.g., Kirk et al. 1996, Schmiegelow et al. 1997, Hobson and Schieck 1999, Imbeau et al. 1999, Whitaker and Montevecchi 1999, Hobson and Bayne 2000). This region is experiencing increasing pressure from forestry and other industrial development, and there is concern over potential effects on bird populations from habitat loss, fragmentation, and changes in forest structure and composition (Schmiegelow et al. 1997, Hobson and Schieck 1999, Schmiegelow and Mrnkkrnen 2002). There is also recent evidence that breeding bird distributions may be shifting in response to global climate change, particularly at northern latitudes (Thomas and Lennon 1999, Brommer 2004, Hitch and Leberg 2007) with unknown effects on population dynamics.
Little is known about the distribution, abundance, and habitat associations of many terrestrial birds in the northwestern portion of the continent's boreal forest region (Machtans and Latour 2003). Much of this region is remote and difficult to access, and available information is largely restricted to areas near the few existing roads or along river corridors (e.g., Bishop 1900, Osgood 1909, Kessel and Springer 1966, White and Haugh 1969, West and DeWolfe 1974, Ritchie and Ambrose 1992, Paton and Pogson 1996, Benson et al. 2000, Cotter and Andres 2000, Sinclair et al. 2003). Studies that have examined avian abundance relative to ecological characteristics have been based largely on habitat-specific surveys with few replicate samples within the region of interest (e.g., Theberge 1976, Spindler and Kessel 1980, Kessel 1998, Matsuoka et al. 2001, Machtans and Latour 2003).
Rosenberg and Blancher (2005) recently identified the need for setting numerical population objectives to aid in conservation efforts for landbird species in North America (cf. Blancher 2003, Rich et al. 2004). They developed methods (Blancher et al. 2007) to derive estimates of regional and continental population sizes based on data from the North American Breeding Bird Survey (BBS), a volunteer-based roadside survey originally designed to monitor population trends.
Rosenberg and Blancher (2005) encouraged refinement of their methodology and Thogmartin et al. (2006) subsequently emphasized the need to: (1) measure detection probabilities to estimate densities more accurately (cf. Thompson 2002), (2) account for potential biases due to non-representative sampling of habitats, and (3) estimate densities in the boreal region, for which there is a paucity of survey data.
We conducted a 2-year study of birds of Yukon-Charley Rivers National Preserve, Alaska, a large, roadless conservation unit at the northwestern extent of North America's boreal zone in east-central Alaska. Our primary objectives were to: (1) document the distribution of birds during the breeding season relative to ecological landforms in the Preserve, and (2) estimate densities and population sizes of passerines with distance-sampling techniques that incorporate detection probabilities. Ours is the first study specifically designed to estimate densities and population sizes of terrestrial species of birds across a broad natural landscape of the boreal forest zone in North America.
STUDY AREA
Yukon-Charley Rivers National Preserve is a 10,194-[km.sup.2] National Park Service unit in the subarctic boreal forest of east-central Alaska (65[degrees]N, 143[degrees]W; Fig. 1). Elevations in the Preserve range from 180 m along the Yukon River to 2,010 m in the mountains of the Yukon-Tanana Uplands. Based on weather data for Eagle, Alaska (elevation 256 m), mean annual precipitation at low elevations in the Preserve is 30 cm and mean monthly temperatures range from 16[degrees]C in July to -25[degrees]C in January (Western Regional Climate Center 2007).
The complex geology, climate conditions, natural fire regime, hydrology, and discontinuous permafrost have produced a mosaic of different-age taiga and tundra plant communities in the Preserve. Forest communities with tree canopy cover of 25% or more include coniferous forest dominated by white spruce (Picea glauca) or black spruce (P. mariana); deciduous forest dominated by paper birch (Betula papyrifera), quaking aspen (Populus tremuloides), or balsam poplar (P. balsamifera); and mixed spruce/deciduous forest. White and black spruce dominate the woodland communities that have a sparser tree canopy cover of 10-24%. Tall and low shrub communities consist primarily of willow (Salix spp.), dwarf birch (Betula nana), and alder (Alnus spp.) while dwarf shrub communities are dominated by dwarf ericaceous shrubs (family Ericaceae) and mountain-avens (Drvas spp.). The wet sedge/graminoid community is mainly tussock tundra formed by tussock cottongrass (Eriophorum vaginatum).
[FIGURE 1 OMITTED]
Swanson (2001) classified the Preserve into 29 ecological units relative to patterns in potential natural communities, soils, hydrology, landforms and topography, lithology, climate, and natural processes (cf. Cleland et al. 1997). We used this classification system to design our inventory and subsequently grouped similar units into five landforms for analysis: Floodplain and Terrace, Lowland, Hill and Bluff', Subalpine Mountain Slope and Valley, and Alpine Mountain Dome and Ridge (Fig. 1).
Floodplain and Terrace.--This landform encompasses 81,683 ha of floodplains and terraces along the Yukon River and lower segments of its major tributaries (elevations 180-365 m; 8.0% of the Preserve). We did not sample the waters of the Yukon River or its small, frequently flooded alluvial bars and islands, which cover an additional 14,609 ha (1.4%) of the Preserve. Ponds, thermokarst lakes, and sloughs are scattered throughout this landform. Soils are often wet and permafrost occurs in all but active floodplain and drier upper terraces that have been burned. Floodplains are generally highly disturbed areas with varied vegetation including tall and low shrub, spruce forest, deciduous forest, and mixed forest. This landform contains the highest percent of tall shrub and wet sedge/graminoid vegetation in the Preserve. Wet sedge/graminoid communities (tussock tundra) and low shrubs occur on wet terraces along with sparse spruce forest and woodland. White and black spruce forest, black spruce woodland, and deciduous forest are on slopes of high terraces. About 29% of the landform has burned during the past 50 years and the fire-return interval is about 150 years (D. K. Swanson, pers. comm.). Fire occurs infrequently throughout most of the landform because of the low inherent flammability and the presence of natural firebreaks (sloughs, lakes, and other waterways). The high terraces are drier and susceptible to frequent fire events.
Lowlands.--The 79,104 ha of gently sloping basins and plains that define the Lowlands landform (elevations 300-915 m; 7.8% of the Preserve) have wet, fine-grained soils underlain by permafrost, which may recede locally to several meters below the surface due to fires. This landform is primarily covered with black spruce woodland and spruce forest. It contains the highest percent of black spruce woodland (50%) of all the landforms in the Preserve. Deciduous forest, mixed forest, and low shrubs occur on previously burned areas and steep slopes. Wet sedge/graminoid meadows (often with tussocks) and low shrubs occur in low areas. An estimated 42% of the landform has burned during the past 50 years (fire-return interval of about 90 years), but fire frequency is highly variable among different areas (D. K. Swanson, pets. comm.).
Hill and Bluff--The Hill/Bluff landform encompasses 431,227 ha of gentle to moderately steep hills (elevations 210-1,160 m; 42.3% of the Preserve) with mostly dry and rocky soils, except on north slopes, lower slope positions, and nearly flat summits, where permafrost and wetter soils occur. Hillsides are generally covered with deciduous forest in previously burned areas and mixed forest where unburned, although black spruce forest and woodland occur on north-facing and lower elevation slopes. Wet sedge/graminoid and low shrub communities dominate the summits. Steep slopes and high ridge tops often are covered with rock scree or knobs. Bluffs in these units are steep west- and south-facing slopes rising directly from the Yukon River dominated by steppe communities of grasses and sagebrush (Artemisia spp.), deciduous forest, mixed forest, scree, and exposed rock. Fires occur frequently in this relatively dry landform (fire-return interval of about 85 years) and about 51% has burned during the past 50 years (D. K. Swanson, pets. comm.).
Subalpine Mountain Slope and Valley.--The Subalpine landform includes 164,859 ha of sub-alpine basins, slopes, and valleys (elevations 550-1,465 m; 16.2% of the Preserve). Soils range from dry and rocky on higher slopes to wet with permafrost under lower slopes, valleys, and basins. Plant communities consist of wet sedge/ graminoid (including tussocks) and low shrub thickets of willow and dwarf birch in valley bottoms, spruce forest or woodland on mountain slopes, tall shrub (primarily alder) thickets in ravines, and low to tall shrub thickets in riparian areas. Fire frequency is low and only 3% of the landform has burned during the past 50 years (D. K. Swanson, pers. comm.).
Alpine Mountain Dome and Ridge.--The Alpine landform (elevations 790-2,010 m; 24.3% of the Preserve) encompasses 247,937 ha that extends from rounded mountains covered with low or dwarf shrubs and some wet sedge/graminoid meadows (often with tussocks) on wetter substrates to steep mountain peaks with patches of herbaceous plants or dwarf shrubs among considerable rock rubble and exposed bedrock. Soils are generally dry and rocky. Only 2% of the landform has burned during the past 50 years as there is insufficient combustible material to fuel fires (D. K. Swanson, pets. comm.).
METHODS
Field Procedures.--We stratified the Preserve by ecological unit (Swanson 2001) and used the existing systematic grid of survey townships mapped by the U. S. Public Land Survey System as potential sampling units across the Preserve (126 blocks of nominal size 9.7 x 9.7 km). We assigned each block to the ecological unit that occurred at the block center and randomly selected 40 blocks (32% of total blocks) in proportion to the area in the Preserve that each ecological unit comprised. We mapped three transects with 12 sampling points each in all selected blocks prior to fieldwork. Transects were oriented across major landscape gradients (i.e., up mountain slopes or perpendicular to the river on floodplains) to sample the range of variability of habitats and avifauna within the landforms. Sampling points were placed [greater than or equal to] 400 m apart in open habitats and [greater than or equal to] 200 m apart in forested and tall-shrub habitats to minimize double-counting of individuals. We allocated sample points proportionally to the area of each ecological unit within those blocks that encompassed more than one unit. Half of the blocks within each stratum were allocated randomly to 1999 and half to 2000 with results pooled across years for analysis.
We surveyed breeding birds using point-transect techniques (Buckland et al. 2001). Our survey dates of 5-28 June encompassed peak territory establishment and display for most migrant species. Surveys began between 0100 and 0200 hrs AST and ended at 0900 hrs, the diurnal period of peak passerine activity in the Preserve (Swanson and Nigro 2003). We sampled as many points as possible along preselected transects within this time frame but did not survey when inclement weather (rain, wind, fog, etc.) impaired detection of birds. We estimated distances from each survey point to all birds seen or heard during an 8-min count (recorded in 2 intervals, 0-5 min and >5-8 rain) in 10-m intervals to 100 m, in 25-m intervals from 100 to 150m, and as >150m. We recorded all species detected between survey points or at other times of day that had not been previously seen in the sampling block. We used a shotgun microphone and portable tape recorder to record unusual bird vocalizations for later verification. We classified the vegetation within a 50-m radius around each sample point into one of 24 vegetation communities following Viereck et al. (1992) and recorded the type of ecological unit at the site.
Four two-person crews were trained for 3 weeks in identifying birds by sight and sound, estimating distance to birds, classifying habitat, and cataloging point locations. All crew leaders had extensive prior experience conducting point counts in Alaska and participated in both 1999 and 2000.
Species Richness.--We constructed rarefaction curves to compare observed and predicted species richness of the avian communities within the five landforms. We used program EstimateS Version 8 (Colwell 2006) to compute the species accumulation function Mao [tau] rescaled by mean number of individuals detected using all detections from the 8-min surveys; this provides a smoothed curve of number of species observed in each community relative to sampling effort (Gotelli and Colwell 2001, Colwell et al. 2004). We used the first-order jackknife method (Burnham and Overton 1979) in EstimateS Version 8 (Colwell 2006) to estimate predicted species richness; this adjusts for bias due to species being missed during sampling (Nichols et al. 1998). We compared 95% confidence limits of the point estimates given sampling efforts of 1,000 individuals per landform to test for differences among landforms in species richness.
We used the Chao-Jaccard incidence-based estimator (Chao et al. 2005) to evaluate similarity of species composition among landforms. This estimator is based on incidence (presence-absence) data using spatial (or temporal) replicates, and incorporates the effect of undetected shared species. We used this incidence-based estimator because we had density estimates for only the most common species and similarity indices are sensitive to rarity....
|
