In order to effectively manage mule deer herds, wildlife biologists must know in what habitats the mule deer prefer to live. In any given area, animals will select some habiat types while avoiding others.
The focus of our project will be to evaluate summer habitat availability vs. preference for the Paunsaugunt Plateau mule deer herd in Southern Utah. More specifically we wanted to test the null hypothesis: Paunsaugunt mule deer are utilizing vegetational habitat in proportion to it's availability, against the alternative hypothesis: Paunsaugunt mule deer are not utilizing vegetational habitat in proportion to it's availability.
The Paunsaugunt Plateau lies within the boundaries of Utah Division of Wildlife Resources Herd Unit 52. This unit consists of 953,103 ha within Garfield and Kane counties in southwestern Utah (Fig. 1). Thirteen percent of the acreage consists of private land. The rest of the land is owned and/or managed by the state of Utah (6.4%), National Park Service (3.4%), US Forest Service (61.0%), and the Bureau of Land Management (BLM) (13.7%).
Figure 1. The Paunsaugunt Plateau in Southern Utah, which is the study area for this project.
The Paunsaugunt Plateau mule deer herd unit summer range within Utah consists of sagebrush Artemesia spp., bitterbrush Purshia tridentata, snowberry Symphoricarpos spp., rabbitbrush Chrysothamnus spp., aspen Populus tremuloides, wax current Ribes cereum, and curlleaf mountain mahogany Cerocarpus ledifolius. Sagebrush is the dominant component of the summer range at lower altitudes (Fig. 2), while aspen and other deciduous and coniferous trees comprise most of the vegetation at higher altitudes
(Fig. 3). Irrigated alfalfa (Medicago sativa) located within the Paunsaugunt deer unit is 939 ha of which 680 ha occurs on the Paunsaugunt Laundowners Wildlife Associations' land. Alfalfa is the primary forage grown on private land. The balance of privately owned land is dry land containing range seedings or native range.
Figure 2. Typical low altitude mule deer habitat on the Paunsaugunt Plateau in Southern Utah. (Note: high percentage of sagebrush present)
Figure 3. Typical high altitude mule deer habitat on the Paunsaugunt Plateau in Southern Utah. (Note: high percentage of coniferous trees present)
Eighteen randomly selected radio-equipped mule deer were chosen for intensive, daily observation (Fig. 4). Animals were located for approximately 3 months for a total of between 40-50 locations. Animals were located at different times of the day using a latin-square design. All animals were visually located, using handheld antenna's and a TR-2 telonics Inc. reciever (Fig. 5). A Universal Transverse Mercator (UTM) reading was taken where the animal was first observed. UTM's were determined using a hand-held Global Positioning System (GPS) unit which allows for immediate and accurate locations. The UTM's were then input into the software "Calhome" which then calculated the home range for each animal. Home ranges were calculated using a 95% contour interval.
Figure 4. One of eighteen collared mule deer that were monitored on a daily basis.
Figure 5. The author visually monitoring a collared mule deer with a handheld antenna, reciever, and Global Positioning System (GPS) unit.
Chart a: Flow chart of data analysis.
Individual home ranges were superimposed onto a vegetation map of Utah (GAP). Then a query was preformed in order to determine how many hectares of each habitat type were in each home range. The actual point locations, that were used to calculate the home range, were then superimposed onto the GAP data to determine how many points fall in each habitat type. If an animal is found in a habitat more than expected, it was assumed it was selecting that habitat. For example, if sage occurs 20% in the home range and the animal was located in sage 30% of the time, we say that the animal was in the sage habitat more than expected, thus selecting it. A habitat availability/preference index for each individual animal was developed.
Summer home ranges of 18 mule deer, that were located on the Paunsaugunt Plateau, were identified. These home ranges were calculated by inputting UTM locations into the software "CalHome". Calhome calculated the configuration of each home range using between 40-50 individual locations. Both the home range configuration and point data (individual locations) were copied to our workspace.
The GAP vegitation data was aquired from CD-Rom and the study area was over layed onto it using IMAGINE. Using an AOI, the study area, and corresponding vegitation, was clipped and loaded to our file. The initial files using Calhome, had two methods of homerange calculations. One was the minimum convex polygon (MCP) method (straight lines) and the other was the adaptive kernel method (oval) (Fig. 6). Using ARC/INFO and Arcedit, we removed the MCP information, leaving the adaptive kernel polygon, which was to be used for final analysis (Fig. 7).
Figure 6. Example of initial Calhome file with both, the MCP and Adaptive Kernel methods being used to calculate homerange.
Figure 7. 'Usable' file with only the adaptive kernel method present.
Individual home ranges were then dropped down onto the GAP vegitation coverage to determine the amount of each habitat type present in each home range (Fig. 8). Figure 9 illustrates a 'close-up' of one homerange that was dropped down onto GAP data plus the corresponding attribute table.
Figure 8. Example of 1 of 18 home range areas overlayed onto GAP vegetation data. (deer 148.465)
Figure 9. One of eighteen homeranges (deer 148.465) dropped onto GAP vegetation data (close up view) and corresponding attribute table.
All 18 homeranges were created and transposed onto the GAP vegitation data. Next, the quantity of habitat available in each homerange was determined. All eighteen homeranges were layed onto the GAP vegetation layer using the 'SUBSET' commmand as well as other commands. See Figure 10 for example of 4 homeranges overlayed on to GAP vegetation data. Homeranges were also clipped in order to preform vegetative analysis (Fig. 11)
Figure 10. Four homerange configurations super-imposed onto GAP vegetation data layer.
Figure 11. Example of clipped homerange used for vegetation analysis.
The attributes that accompanied the original GAP vegetation layer were copied over to each individual home range. Using the 'REPORT' command we determined the amount (ha) of each vegetation/habitat type that is present in each homerange; 18 in total (Fig. 12). We saved all files in 'TEXTEDIT' for easier retrieval and manipulation. Actual percentages of each habitat type were calculated manually and recorded in the textedit file.
Figure 12. Example of table summary of vegetation/habitats present, and quantity in each homerange; 18 total.
Our next step was to overlay the actual location/point files on to the homerange and query so as to discover the percentage of points that are found in each habitat type thus producing a habitat availability/preference index for each individual animal. For statistical reasons, similar vegetation types, in the GAP vegetation database, were combined thus we reduced 36 original classes to 11 classes. We changed point files to grid files and home range image files to grid files using ARC/Info. Then we overlaid each of the grid onto the other (ie. deer 1050 points onto deer 1050 homerange image ect. We then used the 'SAMPLE' command in Arc to determine in which habitat the points were located (Fig. 13). The major class of vegitation from bottom left to bottom right are as follows: mtn. fir, juniper, aspen, ponderosa pine, sagebrush, mtn. shrub, wetlands, spruce, barrens, grassland, and other.
Figure 13. Chart of vegetation utilization for mule deer of the Paunsaugunt Plateau. X axis = Vegetation class, Y axis = Frequency.
Using our vector point file of recorded UTM deer locations (vector) and our homerange vegetation coverages (raster), we were able to query how frequently deer were located in each vegetation class. From our queried data, we performed statistical analysis. After creating a frequency distribution table our hypotheses were tested with a Chi-Square Goodness of Fit test.
RESULTS & CONCLUSION
Our Chi-Square Goodness of fit test results are summarized in the following table.
Our initial hypotheses were as follows:
Ho: Paunsaugunt mule deer are utilizing vegetational habitat in proportion to it's availability.
Ha: Paunsaugunt mule deer are not utilizing vegetational habitat in proportion to it's availability.
X observed = 473
X critical = 18.3
Based on the Chi-Square critical value, we reject the null hypothesis (Ho:) and accept the alternative (Ha:).
Our overall observations show that Paunsaugunt mule deer are not utilizing vegetational habitat in proportion to it's availability. Based on figure 13 and the results table, we can infer that they are selecting some types of vegetation more than expected and avoiding others. Vegetation types that are being selected more than expected include Mtn. Fir, Aspen, Ponderosa Pine, Sagebrush, Wetlands, Grasslands and Other. Vegetation types that appear to be selected less than expected consist of Juniper, Mtn. Shrub, Spruce and Barrens. Some vegetation types are obviously avoided more than other types (juniper) while others are decidedly being selected for (ponderosa pine, sagebrush). It is, however, not as clear for other classes, in which observed and expected values are similar. In such cases further statistical evaluation is needed to determine the presence or absence of statistical significance. However based on this project, which is an initial attempt to quantify habitat availability/preference, we conclude that areas on the Paunsaugunt Plateau, that are characterized by heavy stands of juniper, be modified so as to stimulate increased growth of sagebrush and associated vegetation. In the past, this has been accomplished by techniques such as chaining or prescribed burning. In addition, areas that have a large percentage of ponderosa pine should be preserved as mule deer appear to be utilizing this vegetation type at least 50% more than it is available.