Bachman’s sparrows occupy two main habitats in the Southeast: mature (>80 year old) pine stands that are frequently burned (<3 year burn interval) and recently cutover areas (<5 year old; Dunning and Watts 1990). However, productivity is lower in these latter habitats (1 vs. 3 offspring/pair/year; Liu and others 1995, Perkins and others 2003a). Based on this and the poor colonizing ability of this species – suitable clearcut habitats >3 km from a source population generally remained unoccupied in South Carolina (Dunning and others 1995) – Tucker and others (2004) considered Bachman’s sparrow to be endemic to mature longleaf pine stands.
In all studies of Bachman’s sparrow habitat, two features are repeatedly identified: a dense grass understory and an open overstory, both of which are maintained through frequent fires (Haggerty 1998, Plentovich and others 1998, Tucker and others 2004, Wood and others 2004). Stands managed for red-cockaded woodpeckers via prescribed burning typically provide excellent habitat for Bachman’s sparrows as well because the fires are frequent enough to suppress dense woody understories and maintain sparse canopies (Wilson and others 1995, Plentovich and others 1998, Provencher and others 2002, Wood and others 2004).
Our habitat suitability model for Bachman’s sparrows contains six parameters:
- Landcover type
- Successional age class
- Forest patch size
- Canopy cover
The first suitability function combines landform, landcover type, 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 data from Hamel (1992) on the relative quality of different vegetation types in different successional stages for Bachman’s sparrows.
We also included forest patch size (SI2) as a factor because of the relatively large home range size for this species (mean = 2.5 ha; Haggerty 1998). Home range size varied among regions and habitat types (reviewed in Mitchell 1998), being slightly larger in evergreen stands (4.8 ha) than in ephemeral early successional habitats (2.2 ha). We fit a logistic function
through these data points, assuming the former represented a stand area which sparrows could reliably occupy and the latter value was a minimum below which sparrows would not occur
We included canopy cover (SI3) as a third suitability function to satisfy the two-fold requirement for open canopies and dense understories, two habitat components often well-correlated
. Haggerty (1998) observed an average canopy cover of 9.5 percent at sites occupied by Bachman’s sparrows and 40 percent canopy cover at unoccupied sites. Wood and others (2004) observed twenty times more Bachman’s sparrows in habitats with 25-50 percent canopy cover than sites with 50-75 percent canopy cover. We fit an inverse logistic function to these data to extrapolate values between these known points
As a resident species restricted to a specialized habitat, the occupancy of a site by Bachman’s sparrows is affected by the ability of dispersers to colonize it. This ability is directly affected by the connectivity (or conversely the isolation) of habitat patches (SI4). Birds are unable to colonize clearcuts >3 km away before succession renders the habitat conditions within them unsuitable (Dunning and others 1995). Although isolation may also affect the occupancy of mature evergreen stands, the habitat conditions within them are less ephemeral. Therefore, Bachman’s sparrows have a potentially longer time birds to colonize these stands. To compensate for this differential temporal window in accessibility, we doubled the 3 km distance threshold applicable to clearcuts and fit a longer tail to the function relating connectivity of patches to their suitability as Bachman’s sparrow habitat
. We also assumed source populations were restricted to mature evergreen forest stands with a preliminary overall suitability index score (calculated from SI1, SI2, and SI3) >0.8.
To calculate the overall suitability index score, we calculated the geometric mean of the two suitability indices related to forest structure (SI1 and SI3) and landscape attributes (SI2 and SI4) separately and then the geometric mean of these values together.
Overall SI = ((SI1 * SI3)0.500 * (SI2 * SI4)0.500)0.500