Cerulean warbler territory habitat selection related to operational silviculture prescriptions in West Virginia

Background/Question/Methods

We are quantifying breeding habitat selection by cerulean warblers (Setophaga cerulea) at three sites in West Virginia during the 2013-2015 breeding seasons to examine how silvicultural practices affect habitat use of an interior-forest passerine. Different silvicultural treatments (e.g., shelterwoods, clearcuts) are integrated into a mosaic of harvests at each site. The goal of the study is to recommend ways to improve breeding habitat for cerulean warblers through operational silviculture. We are mapping territories on 7 plots to evaluate changes in territory density and territory habitat selection, pre- and post-harvest.  We then will examine habitat selection between actual territories and randomly generated territories for landscape-scale metrics (aspect, slope position, and landform) and for local-scale vegetation metrics (canopy tree basal area, tree species composition, and relative tree density). We will also compare vegetation metrics measured within cerulean territories to random vegetation plots to evaluate use versus availability across the territory mapping plots. We hope to explain how ceruleans select territories within an implemented harvest mosaic that offers structural diversity to the birds and how the overall songbird community may be managed under the umbrella of cerulean breeding habitat management.

Results/Conclusions

In 2013 and 2014, average annual territory density was 0.38/ha at harvested sites and 0.26/ha at unharvested sites. Ceruleans placed territories on all available aspects; of 71 territories, 25.3% were on east-facing slopes and 19.7% were on northeast- or southeast-facing slopes (respectively). Ceruleans placed 76.1% of 71 territories on middle, side slopes and 15.5% on upper slopes. To evaluate territory habitat selection, we generated within ArcMap an equal number of random territories as actual territories for each territory mapping plot. All random territories are within the size range of actual territories at our sites (0.05-1.55 ha). Preliminary chi-square tests indicated that there were no differences in selection of any landscape metric between actual and generated territories (P-values range from 0.078 for landform within post-harvest territories to 0.919 for Beers aspect within all territories, regardless of harvest status). We will present analyses comparing vegetation data from cerulean territories to random vegetation plots within territory mapping plots to look for differences in use versus availability in relation to forest stand characteristics. Data from the 2015 breeding season will be incorporated into the analyses and presentation.