Assessing the effects of land use on birds in the southern Appalachian Mountains through occupancy modeling

Background/Question/Methods

The southern Appalachian Mountains have experienced dramatic rates of land use / land cover (LULC) change over the last 50 years. In general, the LULC changes can be characterized as: (1) increases in forest because of declines in timber harvest, (2) declines in agriculture, (3) increases in urban land cover, and (3) increased fragmentation of native habitat through amenity-driven exurbanization. The southern Appalachian Mountains are also an area of high biodiversity, making changes in LULC a potential concern for species conservation. As birds are known to be sensitive to landscape changes and are readily detected and identified by their songs, they are a suitable focal taxon. Macon County, North Carolina, the location of the Coweeta Long Term Ecological Research site, is a microcosm in which to study the effects of LUCL on birds. We conducted point counts during the breeding season at 112 sites across a range of LULCs and elevations in Macon County with a double-observer sampling design and collected detection data for 72 songbird and woodpecker species. We built a candidate set of occupancy models based on a priori hypotheses relating LUCL and elevation to avian occupancy and detection. We discriminated among competing models using posterior model probabilities.    

Results/Conclusions

We detected 64 species during the 2010 breeding season. We identified a plausible set of occupancy models and obtained model-averaged parameter estimates. The covariates included in these models identified natural and human-associated attributes that affected avian occupancy and detection, and we interpreted effect size from the sign and magnitude of parameter estimates. In particular, we identified key LULC features affecting species of conservation concern and differentiated guild responses to LULC characteristics and elevation. We also found influential LULC attributes at multiple scales – site, local land use derived from aerial photographs, and landscape land cover derived from satellite imagery. From these occupancy models, we predicted species-specific occupancy probabilities across a range of LULCs and elevations. Models will be validated and updated using data gathered during the 2011 breeding season. In future work, model results will be compared to local knowledge of avian habitat needs and effects of land use, as determined through stakeholder surveys. Results from the occupancy modeling and stakeholder input can be used to develop alternatives to manage local development pressure and avian conservation.