Predicting a species’ distribution can be helpful for evaluating management actions such as critical habitat designations under the U.S. Endangered Species Act or habitat acquisition and rehabilitation. Whooping Cranes (Grus americana) are one of the rarest birds in the world, and conservation and management of habitat is required to ensure their survival. We developed a species distribution model (SDM) that could be used to inform habitat management actions for Whooping Cranes within the state of Nebraska (U.S.A.). We collated 407 opportunistic Whooping Crane group records reported from 1988 to 2012. Most records of Whooping Cranes were contributed by the public; therefore, developing an SDM that accounted for sampling bias was essential because observations at some migration stopover locations may be under represented. An auxiliary data set, required to explore the influence of sampling bias, was derived with expert elicitation. Using our SDM, we compared an intensively managed area in the Central Platte River Valley with the Niobrara National Scenic River in northern Nebraska. Our results suggest, during the peak of migration, Whooping Crane abundance was 262.2 (90% CI 40.2−3144.2) times higher per unit area in the Central Platte River Valley relative to the Niobrara National Scenic River. Although we compared only 2 areas, our model could be used to evaluate any region within the state of Nebraska. Furthermore, our expert-informed modeling approach could be applied to opportunistic presence-only data when sampling bias is a concern and expert knowledge is available.
detection data with all covariates
The file ‘detection data with all covariates.csv’ includes probabilities of detection elicited from the experts in our study and the associated predictors. The column ‘p.det’ is the probability of detection provided by the expert. The column ‘group.size’ is the whooping crane group size the probability was provided for. The column ‘sample’ indicates if the whooping crane group location was used in the training (75%) or test (25%) partition of the data. The column ‘water.500’ is the percentage of water within a 500 m radius buffer around the whooping crane group location the expert provided. Names for other covariates of similar form to ‘water.500’ indicate the type of habitat and the buffer size (water, development, barren, forest, grassland, pasture, crop, woodywetland, and emergentwetland (see Appendix S4 for details); 500 m, 1,000 m, 2,500 m, 5,000 m, and 10,000 m). The column ‘expert’ is a numeric value that identifies the anonymous source of the data (composed of seven reporting experts). The columns ‘day’, ‘month’ and ‘year’ identify the date provided by the expert. The columns ‘lat’ and ‘long’ is the latitude and longitude of the point provided by the expert.
niobrara covariates
The file ‘niobrara covariates.csv’ includes the percentage of nine habitat types (water, development, barren, forest, grassland, pasture, crop, woodywetland, emergentwetland; see Appendix S4 for details) within a buffer (of radius 500 m, 1,000 m, 2,500 m, 5,000 m, and 10,000 m) from the 2006 National Land Cover Database at 99,995 locations (equally spaced) in the Niobrara National Scenic River area. For example, the column ‘water.500’ is the percentage of water within a 500 m radius buffer. The columns ‘lat’ and ‘long’ indicate the latitude and longitude of the location where the predictor was calculated.
program covariates
The file ‘program covariates.csv’ includes the percentage of nine habitat types (water, development, barren, forest, grassland, pasture, crop, woodywetland, emergentwetland; see Appendix S4 for details) within a buffer (of radius 500 m, 1,000 m, 2,500 m, 5,000 m, and 10,000 m) from the 2006 National Land Cover Database at 99,997 locations (equally spaced) in the Platte River Recovery Implementation Program associated habitat area. For example, the column ‘water.500’ is the percentage of water within a 500 m radius buffer. The columns ‘lat’ and ‘long’ indicate the latitude and longitude of the location where the predictor was calculated.
state covariates
The file ‘state covariates.csv’ includes the percentage of nine habitat types (water, development, barren, forest, grassland, pasture, crop, woodywetland, emergentwetland;see Appendix S4 for details) within a buffer (of radius 500 m, 1,000 m, 2,500 m, 5,000 m, and 10,000 m) from the 2006 National Land Cover Database at 100,044 locations (equally spaced) in Nebraska, USA. For example, the column ‘water.500’ is the percentage of water within a 500 m radius buffer. The columns ‘lat’ and ‘long’ indicate the latitude and longitude of the location where the predictor was calculated.
whooping crane data with all covariates
The file ‘whooping crane data with all covariates.csv’ includes information and predictors for 407 whooping crane group locations in Nebraska from 1988‒2012. The file also contains 10,000 Monte Carlo integration points with predictors calculated from the 1992, 2001, and 2006 National Land Cover Database (NLCD) for a total of 30,000 points. The column ‘ID’ is the United States Fish and Wildlife Service unique identifier for the whooping crane group. The column ‘y’ indicates if the row is a record of a whooping crane group (y = 1) or a Monte Carlo integration point (y = 0). The column ‘group.size’ is the number of whooping cranes in each group. The column ‘sample’ indicates if the whooping crane group location was used in the training (75%) or test (25%) partition of the data. The column ‘water.500’ is the percentage of water within a 500 m radius buffer around whooping crane group locations and Monte Carlo integration points. Names for other covariates of similar form to ‘water.500’ indicate the type of habitat and the buffer size (water, development, barren, forest, grassland, pasture, crop, woodywetland, and emergentwetland (see Appendix S4 for details); 500 m, 1,000 m, 2,500 m, 5,000 m, and 10,000 m). The columns ‘day’, ‘month’ and ‘year’ identify the date the group was first observed on or a date associated with the Monte Carlo integration point. The column ‘population’ is the estimated number of whooping cranes in the Aransas-Wood Buffalo population. The columns ‘lat’ and ‘long’ refer to the latitude and longitude of where the whooping crane group or Monte Carlo integration point was located. The column ‘accuracy’ is the accuracy of the whooping crane group location (2 = ‘1/2-Section’, 3 = ‘1/4-Section’, 5 = ‘section’, 7 = ‘GPS’, and 10 = ‘Landmark’; see Appendix S1 of manuscript for more information and Hefley et al. (2014) citation in manuscript). The column ‘NLCDclass’ refers to the NLCD that was used to calculate the predictors (e.g., 1992). Note that Monte Carlo integration points (y=0) will have NA recorded in the columns ‘ID’, ‘group.size’, ‘sample’, and ‘accuracy’.
To obtain the original whooping crane group location records and associated information for Nebraska please contact the author (Trevor Hefley) or the Nebraska Field Office of the United States Fish and Wildlife Service (http://www.fws.gov/nebraskaes/).
Niobrara shapefiles
A shapefile of a portion of the Niobrara National Scenic River, Nebraska.
Niobrara.zip
associated habitat area shapefile
A shapefile of the Platte River Recovery Implementation Program associated habitat area along the central Platte River, Nebraska.
associated habitat area.zip
Nebraska
A shapefile of Nebraska, USA.
