Home Range Analysis

Project Overview

Use coyote radio collar location data near Dugway, UT to visualize individual home ranges and calculate home range areas
Implement a Python script in ArcGIS Pro to find core and general home ranges using kernel density
Determine coyote preferences based on resource availability

Data download and a tutorial for this exercise by the Utah Geospatial Consortium can be found here. The coyote data was provided by Dr. Eric Gese at Utah State University.

All maps are displayed in WGS 1984 Zone 12N. Spatial calculations were also performed in WGS 1984 Zone 12N.

Photograph of a coyote, courtesy of NPS.

Home Range Areas

Figure 1 - Radio collar data for seven coyotes near Dugway, UT was used to create a kernel density raster of coyote locations. This raster was then reclassified using the 50th and 95th percentile kernel values to generate the core and home ranges, respectively. The same process was repeated using data for male and female coyotes separately. A higher resolution version of this figure can be found here.

Key observations:

Female ranges are more compact and clustered than male ranges (Figure 1).
Male ranges are split between the northwest and southeast (Figure 1). This could indicate that males prefer not to overlap territories with other males, but this would need to be verified with individual data.
Male ranges are larger than the ranges for all coyote data and for female coyote data (Table 1), which could indicate male coyotes generally walk farther distances than their female counterparts. However, this hypothesis would need to be tested using more than seven coyote individuals.

Table 1 - Areas for estimated core and home ranges based on male, female, and all coyote radio collar data.

Figure 2 - Core and home ranges for individual coyotes near Dugway, UT. Radio collar data for seven coyotes was loaded to a Python script which created core and home ranges using the 50th and 95th percentile kernel density values, respectively, for each coyote. A higher resolution version of this map can be found here.

Key observations:

There is high variation in the home and core ranges for individual coyotes (Figure 2).
The two coyotes with the largest home and core ranges are C07 and C14, which are also the only male coyotes of the seven individuals. This supports the hypothesis based on Figure 1 that the male coyotes in this area tend to avoid the territory of the other male coyote.

Preference Ratios

Table 2 - Coyote landcover preferences based on Landfire.gov vegetation cover class. Preference ratio was calculated by dividing used point count (observed values) by expected count (based on available cell counts). Values over 1 indicate the landcover class is highly preferred by coyotes; values below one indicate the landcover class is underutilized by coyotes.

Key Observations

The landcover class with the highest preference ratio, Inter-Mountain Basins Semi-Desert Shrub-Steppe, had a preference ratio more than 3 times the next highest landcover class, Developed-Medium Intensity. However, this landcover class had low used and expected counts, so this conclusion is not based on the most robust data. In reality, the most utilized land cover class by coyotes in this area is likely Inter-Mountain Basins Big Sagebrush Shrubland, which has a preference ratio over 2 and higher used and expected counts (Table 2).
The barren landcover class was utilized the least (Table 2). This is likely a solid conclusion for this group of coyotes since the expected count (56) was much higher than the used point count (8).
Several cover classes have very low used and expected counts (<3). These preference ratios should be interpreted with caution as they might not be representative of the group as a whole.

Mastery Observations

Since coyotes22 dataset only provides X and Y values, Arc needs to be explicitly told the spatial reference for those values. The instructions said that the data set was collected in WGS 1984 UTM Zone 12N, so I decided to use that projection when I ran XY Table to Point. I also used that projection for the rest of the exercise.
I decided to used km2 for all area calculations and used planar calculations because the coordinate system was projected.
I verified that all calculated density kernel rasters ran properly by overlaying the original coyote points data.
Initially, I uploaded the coyotes22 data set as a .csv, but the Python script did not run with the coyotes22 dataset in that format. I saved the .csv as a .xlsx file, reloaded the dataset to Arc, and the Python tool ran just fine.

Calculations

Group:

50th percentile: 984/2 = 492 --> 7.08 raster value
95th percentile: 984*0.95 = 935 --> 0.56 raster value

Female:

50th: 768/2=384 --> 9.9 raster value
95th: 768*0.95=730 --> 1.0 raster value

Male:

50th: 216/2=108 --> 0.94 raster value
95th: 216*0.95=205 --> 0.19 raster value

Learned how to use batch calculations today. I ended up batch processing Calculate Geometry Attributes to calculate range areas and Raster to Polygon to convert kernel density rasters to range polygons. This is a really helpful feature!
I tried many different approaches to join landcover counts to my frequency table and it just would not join properly. The "Count" column would be created but be populated by Null values. To complete the exercise, I just manually matched counts to their corresponding classnames in Excel. This approach may not always be possible, so I would like to know what happened so that I can avoid this issue in the future.

Page updated

Google Sites

Report abuse