Monitoring Wild Ring-necked Pheasant Population Restoration in Pennsylvania

Restricted (Penn State Only)
Author:
Williamson, Lacey Taylor
Graduate Program:
Wildlife and Fisheries Science
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
November 16, 2017
Committee Members:
  • W. David Walter, Thesis Advisor
  • Duane Diefenbach, Thesis Advisor
  • Matt Marshall, Committee Member
Keywords:
  • density estimation
  • detection probability
  • Pennsylvania
  • Phasianus colchicus
  • restoration
  • ring-necked pheasant
  • habitat modeling
  • pheasant
  • habitat
Abstract:
Ring-necked pheasants (Phasianus colchicus) are a non-native species that has become naturalized and a popular game bird in the United States. The population in Pennsylvania has been declining since the 1970s despite stocking and habitat restoration efforts. One of the management objectives of the Pennsylvania Game Commission is to provide quality pheasant hunting which requires restoration of the wild pheasant population so that it is naturally reproducing and able to withstand hunting pressure. The current best potential habitat for pheasants within the state was identified and 4 wild pheasant recovery areas (WPRAs) were created to monitor pheasant restoration efforts. At these areas, it was illegal to release stocked pheasants, hunt and harvest either sex, or train dogs. Wild-trapped pheasants from South Dakota and Montana were released at the study areas to ensure there would be an adequate founding population for restoration purposes. The Pennsylvania Game Commission set objectives for a density of 3.86 female pheasants/km2 and that would be adequate for maintaining a sustainable population with hunting pressure. To assess the success of the project and aid in future management decisions, we explored methods of estimating the density by incorporating multiple detection probabilities and a model for predicting potential female pheasant density based on micro-habitat data. As opposed to requiring multiple years of monitoring to obtain population trends using indices of abundance, we used crowing counts and adjusted for detection probabilities to estimate density at each of 12 study areas from 2013 to 2016. Our density estimates were adjusted for the probability a male pheasant crowed in a 3-minute survey period, the probability an observer was able to detect a pheasant given that it crowed, and the probability of flushing a male pheasant. We found the probability a male crowed in the survey period to decrease linearly during the breeding period (21 April–23 May) from 0.659 to 0.464. The probability of detecting a crowing pheasant at 0.80 km was >0, indicating that there was no distance at which it was reasonable to assume no birds could be detected. Instead, the effective area was used and is robust to choice of radius. Because the male pheasants are recorded during the crowing counts and female densities are required to meet objectives, we estimated the probability of flushing a male pheasant to be 0.495 yielding an almost 1:1 sex ratio. Only one of 12 study areas achieved the female density goal of 3.86 females/km2 from 2013 to 2016. The methods used to obtain these density estimates simplified the crowing count protocol and can easily be used and adjusted for other species, detection probabilities, or survey areas to estimate density. To assess habitat or landscape composition at the WPRAs, we conducted a micro-habitat analysis by identifying all vegetation types within a 0.56 km radius of the survey point (hereafter referred to as survey circle) resulting in digitized maps of micro-habitat within a survey circle. Our objective was to create a model that would be able to predict potential female pheasant densities based on proportion of habitat type in a survey circle. We had 37 vegetation types that were combined into 8 habitat types a priori and used these as independent variables in our predictive models. We used our methods of estimating pheasant densities (Chapter 2) to estimate female density at the survey circle to be used as the dependent variable. For the model, we used 2 years of micro-habitat data (2013–2014) and density estimates (2014–2015). We estimated that female pheasant densities were influenced most by the proportion of idle grasses and forest. We found the proportion of idle grass to positively influence pheasant densities, while forest had a negative influence. This model will be important for improving habitat to meet a desired pheasant density goal by allowing managers to make recommendations for quantity and proportions of habitat needed to achieve desired pheasant density goals. Only one study area, Washingtonville West, was successful in meeting the density goal of 3.86 female pheasants/km2, while all other study areas did not exceed a density of 2 females/km2. Washingtonville West had the lowest average proportion of forest in a survey circle of any study area. From 2013 to 2015, the average proportion of forest ranged from 4.3% to 9.3% at Washingtonville West, compared to an overall average at all the WPRAs ranging from 15.6% to 17.6%. Given that roughly 60% of the overall landscape in Pennsylvania is forested, the WPRAs represent a small section of the state that is potentially suitable pheasant habitat. Even at this suitable habitat, we only achieved a self-sustainable pheasant population that can withstand hunting pressure at one study area which had <10% forest. Because forest has a negative effect on pheasant densities and so much of the Pennsylvania landscape is forested, it will be difficult to locate enough suitable habitat to support sustainable pheasant populations in order to provide hunting opportunities across the state.