Development of a GIS-based model of yellow-billed cuckoo breeding habitat within the Lower Colorado River Multi-Species Conservation Area, San Pedro River and Verde River, AZ

Call Number:

URL:

Keywords:

Abstract:

This report summarizes the results of a project to develop a GIS-based model of breeding habitat for the yellow-billed cuckoo (Coccyzus americanus occidentalis) within the Lower Colorado River Multi-Species Conservation Area, the San Pedro River, and the Verde River in Arizona, a project funded by the Lower Colorado River Multi-Species Conservation Program (LCR MSCP). The goal of the project was to provide management agencies with the knowledge necessary to conserve and enhance yellow-billed cuckoo breeding habitat along the lower Colorado River and its tributaries. Specific objectives to accomplish this goal were to 1) characterize and map yellow-billed cuckoo (cuckoo) breeding habitat inside the LCR MSCP boundaries, 2) develop a GIS-based probability model of cuckoo breeding habitat, and 3) extrapolate and verify the cuckoo model inside (LCR MSCP) and outside (Verde and San Pedro rivers) the project area. The authors hypothesized that cuckoo breeding locations are typically associated with broad floodplains, moderately rough terrain, and large, vigorous riparian vegetation dominated by cottonwood and/or willow trees. For habitat characterization and logistic-regression model construction, cuckoo breeding locations collected in 2006 were used, as well as vegetation data and floodplain features obtained from satellite imagery, orthorectified digital aerial photography, and a digital elevation model (DEM). Multivariate logistic regression was used to identify significant associations among cuckoo occurrences and biophysical features and to test hypotheses. Two different types of models were tested: Satellite-based and Map-based (Aerial). The Satellite model characterized vegetation features surrounding cuckoo breeding locations off of Landsat5 Thematic Mapper imagery, identifying vegetation density and structure with the Normalized Difference Vegetation Index [NDVI]). In contrast, the Aerial model characterized vegetation types (aerial cover of species, communities) surrounding cuckoo breeding locations off of a vegetation map obtained from orthorectified digital aerial photography. Both techniques were examined because each provided unique information and had different strengths and weaknesses. The authors hoped a dual approach to modeling would result in the most useful information for the management and conservation of cuckoo breeding habitat. They tested the accuracy of each model’s predictions with an independent set of cuckoo locations obtained the following year (2007). The Satellite and Aerial models provided information about several important features associated with cuckoo breeding habitat. First, a core area of dense cottonwood/willow within a 120-m radius (4.5 ha) of a location increased the chances of cuckoo occurrence. The likelihood of cuckoo occurrence continued to increase if the core area was surrounded by a large, native forest (480 m radius/72 ha) that contained lots of structural diversity. Third, a landscape of moderate topographic roughness (i.e., the Bill Williams NWR) further increased the odds of cuckoo occurrence. The odds of cuckoo occurrence decreased rapidly when too much tamarisk surrounded the site. When the authors challenged the 2006 model with 2007 data, accuracies between 75 and 80% were obtained inside the LCR MSCP boundary. A correlation analysis found the Satellite and Aerial models’ predictions were significantly correlated, indicating the Satellite model indirectly characterized structural features and vegetation types important to breeding cuckoos. An important finding of this project is that there were higher numbers of cuckoo detections in areas that the habitat models predicted to be most suitable (i.e., higher-probability classes), making them useful to management agencies for conservation and restoration purposes. The Satellite model appeared more accurate on reaches with large floodplains and broad riparian patches, while the Map model produced better results on reaches with smaller floodplains and riparian patches. The accuracy of the Satellite model was mixed and somewhat ambiguous outside the project area, so the authors recommend it be used with caution outside the LCR MSCP boundary, and only with field verification.

Notes:

ELECTRONIC FILE - Zoology

[Cited elsewhere as the following: Johnson, M.J., J.R. Hatten, J.A. Holmes, and P.B. Shafroth. 2012. Development of a GIS-based model of yellow-billed cuckoo breeding habitat within the Lower Colorado River Multi-Species Conservation Area, San Pedro River, and Verde River, AZ. Admin. Report to Lower Colorado River Multi-Species Conservation Program, Bureau of Reclamation, Boulder City, NV. 53 pp.]