Planning for Landscape Management and Adaptation
Evaluation of Climate Change Impacts for the Southwest U.S. and Northern Mexico


About the Maps

The development of sophisticated species distribution modeling techniques provides an opportunity to examine the potential effects of future changes in the global climate and landscape to bird communities. We used species distribution modeling techniques to relate bird location data to environmental layers. This allowed us to generate projections of current and future species occurrence. We provide a brief summary on our methods below. Please contact us for more details on our modeling methods.

Avian Models

We ran bird habitat suitability models with the program Maxent v3.3.3k (Phillips et al. 2006) using climate and vegetation type as inputs. We acquired bird presence locations from several sources throughout our area of interest.  We then filtered locations for spatial and temporal accuracy. We used "boosted regression tree" models to run current and future projections of broad vegetation types. Climate variables included total precipitation, mean temperature, and temperature range for species specific breeding windows in time (e.g. March through June).

Climate Data

We acquired contemporary climate data from WorldClim at a spatial resolution of approximately 1km. We acquired future climate projections from Conservation International. The future projections were taken from the World Climate Research Programme's (WCRP's) Coupled Model Intercomparison Project phase 3 (CMIP3) multi-model dataset which was used for the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report. These data were  at a spatial resolution of approximately 5km for the time period averaged across 2041-2060 using the A2 emissions scenario. We selected five general circulation climate models including:

Vegetation Models

We used boosted regression trees to model the current and future distribution of vegetation with climate, soil,and geophysical variables as input. For vegetation data we used the Global Land Cover 2000 dataset for North America (Latifovic et al., 2004) which was derived from the VEGETATION instrument aboard the SPOT-4 satellite. Land cover is separated into 29 coarse classes of vegetation and land use type at a resolution of approximately 900m over the study area. Climate data consisted of bioclimatic variables from the sources described above and soil variables from the Harmonized World Soil Database (FAO et al., 2012), as well as slope, solar radiation, and distance to stream.