The wood thrush is a long distance Neotropical migrant that has become the poster child for songbird declines due to forest fragmentation. Due to its general abundance, ease of nest location and monitoring, and area sensitivity, the wood thrush is easy to study and a large body of knowledge exists on this bird (Roth and others 1996). Wood thrush are common in deciduous and mixed forests but rarely occur in pure evergreen stands (Roth and others 1996). Mesic, upland forests with a moderate density of midcanopy trees and shrubs for nesting and an open understory with abundant leaf litter for foraging are generally considered optimal (Roth and others 1996). Closed overstory canopies are commonly used (Roth and others 1996, Bell and Whitmore 2000).
The wood thrush displays area sensitivity in productivity but not occupancy of habitats. Wood thrushes nest in forest fragments as small as 0.3 ha (albeit at low densities; Weinberg and Roth 1998) and in narrow (<150 m wide) riparian strips (Sargent and others 2003); however, nest predation and parasitism rates in fragments <80 ha and riparian buffers <530 m wide are extremely high (Donovan and others 1995, Hoover and others 1995, Peak and others 2004). Landscapes with greater amounts of forest cover (particularly unfragmented forest) mediate some of these effects in small woodlots (Donovan and others 1997, Driscoll and Donovan 2004, Driscoll and others 2005). Nest success is better predicted by the amount of forest in the landscape than the structural characteristics of microhabitat around nests (Hoover and Brittingham 1998, Driscoll and others 2005).
The wood thrush density model contains seven parameters:
- successional age class
- forest patch size
- percent forest in the local (1-km radius) landscape
- small (<2.5 cm d.b.h.) stem density
- canopy cover
The first suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that defines unique combinations of these classes
. We directly assigned suitability index scores to these combinations based on habitat associations reported in Hamel (1992) but made minor adjustments to increase suitability scores for sapling stands based on Thompson and others (1992).
Although wood thrush will occupy small forest fragments, their density may be lower within them. Therefore, we included forest patch size (SI2) in the model of wood thrush habitat suitability. We fit a logarithmic function
to data from Robbins and others (1989), Kilgo and others (1998), and Tilghman (1987) that documented changes in relative occurrence with changes in forest patch size
. Nevertheless, the suitability of a forest patch is influenced not only by its size but also by its landscape context (SI3). In predominantly forested landscapes, small forest patch sizes that may not be otherwise utilized provide habitat due to their proximity to large forest blocks (Rosenberg and others 1999). Assuming landscapes with <30 percent forest provided poor habitat (suitability index score ≤ 0.100) and landscapes with >80 percent forest were excellent habitat (suitability index score ≥ 0.900), we developed a logistic function
to predict habitat suitability from the percentage of forested landcover in the local (1-km radius) landscape
. We used the maximum suitability index score from either SI2 or SI3 to increase the suitability of small patches in heavily forested landscapes.
The wood thrush forages in leaf litter on the forest floor and is most common in stands containing an open understory. We included small stem density (SI4) in the model as a proxy to understory cover. We fit an inverse logistic function
to small stem density numbers that discounted habitat suitability in habitats with high small stem numbers and presumably dense understories
The wood thrush is also associated with closed-canopied forests. Therefore we included canopy cover (SI5) as a model parameter and fit a logistic function
to data from Annand and Thompson (1997) and Hoover and Brittingham (1998) to predict suitability index scores from canopy cover values
To calculate the overall suitability index, we determined the geometric mean of suitability index scores for forest structure attributes (SI1, SI4, and SI5) and then calculated the geometric mean of this value and the maximum of either suitability index scores from either forest patch size or percent forest in the landscape (Max(SI2,SI3)).
Overall SI = ((SI1 * SI4 * SI5)0.333 * Max(SI2, SI3))0.500