Ecological Minimums for Greater Sage-grouse across their Historic Range in Western North America
Dates
Publication Date
2015-07-25
Time Period
2015-07-25
Citation
Crist, M.R., Knick, S.T., and Hanser, S.E., 2017, Raster digital data sets identifying a range-wide network of priority areas for greater sage-grouse: U.S. Geological Survey data release, https://doi.org/10.5066/F7DB7ZZK.
Summary
We partitioned a Mahalanobis D2 model of sage-grouse habitat use into separate additive components each representing independent combinations of species-habitat relationships to identify and map range-wide ecological minimums for sage-grouse. We assumed the states delineations of priority areas capture higher quality habitat and larger numbers of sage-grouse populations across our study area. We randomly selected 1669 points from the priority areas and corresponding variables GIS datasets to calibrate models. We estimated distributions of our variables from 1000 iterative samples created by bootstrapping the calibration data. To better incorporate conditions in both large and small priority areas, we restricted the contribution from [...]
Summary
We partitioned a Mahalanobis D2 model of sage-grouse habitat use into separate additive components each representing independent combinations of species-habitat relationships to identify and map range-wide ecological minimums for sage-grouse. We assumed the states delineations of priority areas capture higher quality habitat and larger numbers of sage-grouse populations across our study area. We randomly selected 1669 points from the priority areas and corresponding variables GIS datasets to calibrate models. We estimated distributions of our variables from 1000 iterative samples created by bootstrapping the calibration data. To better incorporate conditions in both large and small priority areas, we restricted the contribution from each priority area in a sample to a random selection of a maximum of 25 points. We then performed a PCA on each of the 1000 iterative samples. The final model was created by subsequently averaging the PCA output after correcting for sign ambiguity (Bro et al. 2008) across all iterations. We selected 14 out of 23 partitions that meet our criteria of having an eigenvalue < 1 (Table 1). We rescaled the D2 to range continuously from 0 to 1; a value of 1 indicates environmental conditions identical to the mean habitat vector, whereas a value near 0 indicates very dissimilar conditions (Fig 2). To evaluate model performance, we calculated the area under the curve (AUC) for a receiver operating characteristic (ROC) to assess sensitivity (fraction of habitat points correctly classified) and specificity (fraction of non-habitat points predicted as habitat) (Fielding and Bell 1997). We derived our presence/absence datasets using sage-grouse breeding densities (Doherty et al. 2011). For our presence data, we overlaid the 100 percent sage-grouse breeding densities (Doherty et al. 2011) representing spatial locations of all known sage-grouse populations with our habitat map and selected all habitat values that fall within the density boundaries. For our absence data, we selected all habitat values that fall outside of the breeding density boundaries. To calculate the AUC, we randomly sampled 5000 data points from the presence data set and 20,000 data points from the absence data set. We created a null presence/absence dataset by iteratively random sampling from the habitat map 1000 times. For each iteration, we divided the resulting sample into two datasets (null presence and null absence) based on a proportion of the total rows and columns, we then sampled from the two datasets and computed an AUC score. We calculated a mean AUC score and distribution from all null samples. To test for significance, we used a t-test to compare our AUC to the null AUC.
Click on title to download individual files attached to this item.
ecomin_23_p14.tif.xml Original FGDC Metadata
View
17.16 KB
application/fgdc+xml
ecomin_23_p14.tif.zip
11.69 MB
application/zip
Purpose
Data provides a range-wide assessment of habitat for greater sage-grouse. Map reflects coarse-spatial scales and likely do not capture fine-scaled habitat used by sage-grouse. Map should be used at the scale the analysis was implemented or cautiously at finer scales.
