Abstract - Monitoring for Chronic Wasting Disearse in Mule Deer and White-tailed Deer at Wind Cave National Park: Investigating an Emerging Epizootic
Schuler, Krysten L. 2006. Monitoring for Chronic Wasting Disease in Mule Deer and White-tailed Deer at Wind Cave National Park: Investigating an Emerging Episootic. PhD Dissertation. South Dakota State University. 155 p.
Chronic wasting disease (CWD) has emerged as a disease of concern in cervid populations. Testing, transmission, and detection of CWD are areas of interest for managing populations of North American deer (Odocoileus). This dissertation addresses these areas with research from data collected on mule deer (O. hemionus) and white-tailed deer (O. virginanus) in Wind Cave National Park, South Dakota, USA, between February 2003 and December 2005. The objectives for this research were to: 1) assess performance of 2 sampling approaches with tonsillar biopsy for CWD testing; 2) evaluate home-range overlap and direct contacts for potential transmission of CWD; 3) determine probabilities of detection and death related to CWD from deer mortalities; and 4) examine sample size and smoothing parameters on home ranges from GPS collars on home ranges for mule deer and coyotes (Canis latrans). For objective 1, we collected data between January 2003 and January 2005 on white-tailed deer in southeast and southwest Minnesota and white-tailed and mule deer in Wind Cave National Park, South Dakota through biopsy sampling with dorso-lateral and ventral-medial approaches for collecting tonsillar follicles. We obtained significantly more follicles using the dorso-lateral (median number of follicles = 19) than the ventral-medial (median number of follicles = 5.5) approach. No differences were observed in collection of tonsillar follicles that were related to sex, age class, or species of deer. We recommend the dorso-lateral approach for assessing CWD prevalence in deer populations. Regarding objective 2, horizontal transmission (i.e., transmission between potentially unrelated animals) is suspected of serving as the predominate source of infection for CWD epizootics. An indirect form of horizontal infection includes environmental contamination, which has recently been confirmed experimentally to be a source of infection. We radiocollared and tested 4 white-tailed deer and 60 mule deer for CWD in Wind Cave National Park, South Dakota, USA. Our study determined if home-range overlap was an accurate predictor of contact rates between deer. We examined contact and home-range parameters between species, mule deer sexes, and CWD-infected versus non-infected mule deer. Home-range overlap was not a good predictor of contact between deer pairs unless home ranges overlapped by ?70%. Moreover, interspecific contact was unlikely despite home-range overlap. Intraspecific contact between mule deer resulted from same-sex associations for most of the year, but reproductive status (e.g., lactation, rut) influenced association of sexes. Deer infected with CWD had slightly decreased contact rates and increased home-range overlap compared with non-infected deer. We hypothesize that indirect transmission through environmental contamination is a more likely route of interspecific disease spread than direct horizontal contact because of overlapping home ranges with limited social contact. Thus, models of CWD transmission in deer should include home-range overlap because of the possibility of environmental contamination. For objective 3, estimates of mortality rates are essential for understanding population dynamics of ungulates. Currently, there are few data on how CWD affects deer populations or the likelihood that CWD-infected deer are tested. Detection rates for CWD are important because in some instances, deer positive with CWD at time of death cannot be tested for the disease because of availability or condition of appropriate tissues. We evaluated survival, mortality, transition, and detection rates for deer infected with CWD residing in Wind Cave National Park. We included 34 monthly encounters of deer resightings and mortalities for 67 deer (4 white-tailed deer, 63 mule deer). Mortalities were tested for CWD and resulted in prevalence rates of 2.9% in 2003, 9.1% in 2004, and 15.0% in 2005 for a pooled prevalence rate of 9.0% over 3 years. We assessed survival and influence of covariates, such as age, sex, species, collar type, and capture, using known-fate analysis in Program MARK as a precursory evaluation for multi-state modeling. We created multi-state models within Program MARK using a capture-recapture framework with 3 states: 1) 1= alive; 2) A = dead with CWD; and 3) B = dead, non-CWD. Transition probabilities were fixed to 0 for nonsensical transitions. We performed a suite of models and determined that mortality for non-infected deer was 2 times higher than CWD-infected deer, but non-infected deer were 7 times more likely to be tested. As many as 24% of infected deer may not be tested, indicating failures in detection that may be related to mortality from predators. Our multistate model with constant encounters over time and time-dependent transitions indicated CWD did not function as a fully additive source of mortality in this deer population. Nevertheless, there may be deer on the landscape infected with CWD that are being removed by predators and thereby go undetected. For our last objective, estimation of home ranges of animals is a common practice in wildlife science, but a variety of factors, such as choice of smoothing parameter and number of locations necessary for an accurate estimate, are under debate. Methods of data collection, such as the use of global positioning system (GPS) collars versus traditional radiotelemetry, can influence estimates as a result of sampling protocols. We evaluated effect of sample size and smoothing parameter on 2 species (coyotes and mule deer) that likely have variable utilization distributions using location data collected from GPS collars. Data were collected from GPS collars on mule deer in Wind Cave National Park and coyotes in Badlands National Park. We calculated home ranges using 95% and 50% adaptive kernel contours using reference (h_ref) and ad hoc (h_ad hoc) smoothing parameters using ArcGIS 9.1 with Home Range Tools . The ad hoc approach was designed to reduce bias commonly associated with reference bandwidths that use a bivariate normal distribution. We calculated home ranges with both smoothing parameters for random subsamples of 5, 10, 25, and 50% of GPS locations to evaluate necessary sample size and compared area estimates with those obtained for 100% GPS locations. For mule deer, we also conducted visual relocations using conventional radiotelemetry, which resulted in fewer locations than from GPS collars (< 20%). Area was overestimated at smaller sample sizes, but an interesting pattern was noted with greater bias at 60-100 locations than at sample sizes <50 locations. This bias may have resulted from increasing autocorrelation, but it was most likely a result of random sampling error. Examining relative differences produced from home ranges created through subsampling provided a means to estimate variability by smoothing parameter and sample size. Minimum number of points to obtain bias <10% varied by smoothing method, species, study duration, and volume contour (95% or 50%), indicating there may not be a universal recommendation for all studies and species. While h_ad hoc performed consistently better over most sample sizes, we suggest researchers examine point distributions, choose smoothing parameters with the least bias variance, justify and report choices for home-range analysis based on study objectives. Key words: autocorrelation, Canis latrans, capture-recapture, chronic wasting disease, contact rates, coyote, environmental contamination, global positioning system (GPS), home range, known-fate analysis, mule deer, multistate models, Odocoileus hemionus, O. virginianus, prevalence rates, Program MARK, radiotelemetry, smoothing parameter, subsampling, tonsillar biopsy, transmission, white-tailed deer.