AN EVALUATION OF THE PENNSYLVANIA SEX-AGE-KILL MODEL FOR WHITE-TAILED DEER

Open Access
Author:
Norton, Andrew Scott
Graduate Program:
Wildlife and Fisheries Science
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
April 22, 2010
Committee Members:
  • Duane R Diefenbach, Thesis Advisor
Keywords:
  • temporal variability
  • spatial variability
  • harvest rate
  • bootstrapping
  • Sex-Age-Kill
  • ecology
  • wildlife
  • population modeling
  • white-tailed deer
  • Odocoileus virginianus
  • deer
  • sensitivity
  • robustness
  • Pennsylvania
  • precision
Abstract:
Accurately monitoring population trends is an essential component of game species management. This is especially true for white-tailed deer (Odocoileus virginianus) in Pennsylvania where the Pennsylvania Sex-Age-Kill (PASAK) model is used to estimate annual abundance. Historically, the PASAK model only provided point estimates of abundance. However, statistically valid estimates of precision are needed to quantify uncertainty. Additionally, assumption violations can incorporate bias into estimates, the extent of which is unknown. I estimated survival and harvest rates based on data from 1,131 radio- or GPS-collared white-tailed deer and obtained harvest data (number killed, age-sex structure, etc.) to (1) evaluate assumptions of the PASAK model, (2) develop a method to estimate precision, and (3) evaluate model sensitivity and robustness. The PASAK model input parameters for the mature (¡Ý 2.5 years old) male harvest rate, yearling (1.5 years old) male harvest rate, and percent yearling females in the adult (¡Ý 1.5 years old) female population did not meet required assumptions. Mature male harvest rates differed by as much as 32.8% among wildlife management units (WMUs) studied. However, this spatial variability could be reliably modeled using a hunting effort index as a covariate. I found difficulties in the current method used to estimate yearling male harvest rates. However, direct yearling male harvest rate estimates from radiocollar data, which differed by < 16.5% among WMUs studied, provided a suitable alternative estimate. In addition, variation in hunting effort was related to spatial variability in yearling male harvest rates. Yearling female deer were nearly 50% more likely to be harvested than mature female deer, which caused the PASAK model to underestimate population size. Mean coefficient of variation (CV) estimates by WMU, 14.1% in year t, were slightly above benchmarks recommended for managing game species populations (¡Ü 12.8%). Doubling reporting rates by hunters or doubling the number of deer checked by personnel in the field would improve mean CVs in year t to recommended levels. The PASAK model was robust to variability and bias in estimates for mature male harvest rate, female harvest rate, and non-harvest mortality. However, model estimates were sensitive to variability and bias in yearling male harvest rates. Consequently, yearling male harvest rate estimates had the greatest influence on the accuracy of population estimates. Very few studies have monitored spatial and temporal variability in male white-tailed deer harvest rates. My results support the hypothesis that the majority of variability in mature male white-tailed deer harvest rates can be explained by variation in hunting effort. Similar variability exists in yearling male harvest rates, however, antler point restrictions (APRs) in Pennsylvania limit the extent of variability. Although minimized by APRs, yearling male harvest rate variability has profound influences on PASAK model estimates. Because harvest rates in the PASAK model are assumed constant across years, potential influences of regulation changes or shifts in hunting practices that alter male harvest rates will have profound influences on model performance. Yearling male harvest estimates are similarly important, and research to evaluate their accuracy will improve reliability of PASAK model estimates. Increasing harvest reporting rates by hunters may be the most efficient approach to improve precision of harvest estimates, whereas similar improvements by increasing the number of deer checked and aged may be more difficult or expensive to achieve. Independent estimates for the percent of yearling males in the adult male population will reduce the importance of yearling male harvest rate and harvest estimates.