The Swainson’s warbler is locally distributed across the Southeast (Brown and Dickson 1994). Once believed to be restricted to canebrakes in bottomland hardwood and swamp forests of the Atlantic and Gulf Coastal Plains, this species has now been documented breeding at low densities in regenerating clearcuts in Texas and rhododendron-mountain laurel thickets in the southern Appalachians (Graves 2002). Within these sites, Swainson’s warbler may be area-sensitive. Territory size is large for a wood-warbler (3.2 ha; Brown and Dickson 1994), and the species demonstrates area-sensitivity. Swainson’s warblers are not observed on tracts <350 ha in Illinois (Eddleman and others 1980).
Swainson’s warblers do not use canopy height, basal area, successional age class, or species composition as habitat cues (Eddleman and others 1980, Graves 2002), but rather select habitat based on understory characteristics. Dense thickets are required, and stem densities ~35000 stems/ha are optimal (Graves 2002). Canopy gaps are important for encouraging this dense growth, and canopy cover is typically high (70-80 percent) but rarely closed (>90 percent; Eddleman and others 1980, Graves 2001, Somershoe and others 2003). Understory vegetation is primarily woody; herbaceous cover is typically sparse (<25 percent; Eddleman and others 1980, Brown and Dickson 1994). Leaf litter is abundant and provides an important foraging substrate (Graves 2001, Somershoe and others 2003).
Hydrology is a critical factor influencing the habitat suitability for this warbler as well. In bottomland and floodplain habitats, birds select areas that are typically drier than surrounding sites (Graves 2001, Somershoe and others 2003). Inundation of otherwise suitable habitat from March–September negatively affects the quality of an otherwise suitable site (Graves 2002). The species is occasionally found in xeric uplands with appropriate understory characteristics (Carrie 1996).
The habitat model for Swainson’s warbler contains six parameters:
- successional age class
- forest patch size
- proportion of forest in a 1-km radius
- small (<2.5 cm d.b.h.) stem density
The first suitability function combines landform, landcover, and successional age class into a single matrix (SI1) defining unique combinations of these classes
. We adjusted the relative habitat quality rankings of Hamel (1992) for Swainson’s warbler vegetation and successional age class associations to maximize habitat suitability in woody wetland habitats along floodplains and to ensure transitional sapling stands that may be utilized in the WGCP were assigned suitability index scores (Carrie 1996).
We included forest patch size (SI2) in our model because of the preference of Swainson’s warblers for interior sites within large forest tracts. We assumed forest patch sizes ≥350 ha were optimal (suitability index score = 1.000) for Swainson’s warblers but that birds would occupy significantly smaller tracts
. We based a logistic function on these assumptions to predict the impact of forest patch size on habitat suitability
. Nevertheless, the suitability of a specific forest patch size is also influenced by its landscape context (SI3). In predominantly forested landscapes, small forest patches that may not be otherwise suitable may be occupied due to their proximity to large forest blocks (Rosenberg and others 1999). We assumed landscapes with <30 percent forest were non-habitat (suitability index score = 0.000) and landscapes with 100 percent forest were optimal (suitability index score = 1.000;
. We fit a logistic function
to a dataset based on these assumptions and utilized the maximum score from either SI2 or SI3 to account for the higher suitability small patches in predominantly forested landscapes have relative to their size alone.
Swainson’s warblers breed in dense thickets and stem densities ~35000 stems/ha are optimal (suitability index score = 1.000; Graves 2002). Because stem densities can be even higher in early successional bottomland hardwoods (>200000), though, the relationship between Swainson’s warbler habitat suitability and stem density is not linear. Therefore, we fit a quadratic function
to data from Graves (2002) that captured the effect of varying stem density on habitat suitability
To calculate the overall suitability index, we determined the geometric mean of suitability index scores for forest structure (SI1 and SI4) and multiplied that by the maximum suitability index score for either forest patch size (SI2) or percent forest in the 1-km landscape (SI3) and finally calculated the geometric mean of that product.
Overall SI = ((SI1 * SI4)0.500 * Max(SI2, SI3))0.500