Recipient Organization
UNIVERSITY OF ARKANSAS
(N/A)
FAYETTEVILLE,AR 72703
Performing Department
Forestry And Natural Resources
Non Technical Summary
Wild turkeys (Meleagris gallopavo silvestris) are economically and culturally important game bird species in Arkansas and throughout the southeastern United States (Tapley et al. 2011). Despite intensive habitat and harvest management efforts in recent years, trends continue to indicate declining populations of turkeys state-wide. Habitat management in the Ozark and Ouachita Mountain regions of north Arkansas consist of timber logging operations and clearing of ridgetops to create wildlife openings. Wildlife openings are often planted to nature grasses and forbs for foraging ungulates and brood rearing game birds, such as wild turkeys. These clearings also provide open habitats that are importing for male turkeys during the breeding season. Prescribed forest fire plays a critical role in ecosystem management of this pine-hardwood mixed forest ecosystem. The U.S. Forest Service and their partners burn thousands of acres annually in the region. In addition to habitat improvement, harvest regulations can be modified by the regulating agency by altering season structure and bag limits to improve sex and age-class ratio of the population. In recent years, policy changes have occurred without the data to support the regulatory change. It is the intension of my research over the next 5-years to facilitate both habitat and harvest management decisions be investigating their effects of wild turkey population dynamics and space use.Demographic parameters of a population are vital pieces of information for management agencies, particular for wild turkey in which harvest management policy can be adapted in times of demographic distress. In addition, management agencies can gauge the impacts of forest management and wildlife openings by monitoring population demographics. Currently, spring gobbler surveys and hunter success rates are used by the Arkansas Game and Fish Commission (AGFC) to monitor trends in wild turkey populations. From these data, the AGFC interprets declining population trends of wild turkey across most of the state of Arkansas. In response, the AGFC authorized change in hunting regulations during 2011 that restricted harvest of jake turkeys to only youth hunters during the spring hunting season. The No-Jake Harvest Policy aimed to reduce overall harvest mortality on jakes, while subsequently increasing the rate of recruitment into the 2-year old age class. Consequently, the current gobbler surveys and total harvest does not provide estimates of abundance and can be misleading indicators of population trends particular when assessing the effects of policy change on population abundance. Thus, we aim to monitor survival and harvest rates of male turkeys in northern Arkansas. These data can be used to gauge the impacts of hunting mortality and non-hunting mortality on the population and will be used to inform management.With increasing concern of high harvest rates in the Ozarks and Ouachita Mountain regions, the logical next step for improving our understanding of population status in Arkansas is to generate parameter estimates for local abundance and population density. Spatial-capture recapture (SCR) models (Royle et al. 2013a) are a recent and emerging advancement in the field of population abundance and density estimation. SCR is an extension of capture-recapture models that make use of encounter location data to study spatial aspects of animal populations including spatial variation in density, resource selection, and animal movements (Royle et al. 2013a). The models incorporate the geographic location of each capture event, thereby explicitly accounting for unequal detection probabilities among individuals due to their unique location in relation to the camera trap (i.e., individuals living in close proximately have higher probability of detection; Royle and Young 2008, Borchers 2012). Point process models are used in order to identify an individual activity center (si = activity center or home range center for individual i ), and then computes the number of activity centers across a precisely defined area containing the trap array. The series of SCR models are evolving daily and have varying levels of application (i.e., parameter estimates from camera-only studies (both spatial and non-spatial approaches, and with and without paired telemetry data). These models have been used across a range of species from elusive carnivores (e.g., Sollmann et al. 2013) to forest-dependent songbirds (Chandler et al. 2011). SCR data are collected using an array of trap sites, in our case trail cameras. The SCR approach has numerous benefits over traditional mark-recapture methods including (1) less invasive, (2) less expensive, (3) avoids assumption of perfect detection probability, (4) avoids assumption of geographic closure, (5) does not require random installation of camera traps, (6) it can model for spatial variation in density, and (7) it can be used to infer density within unmarked- and partially-marked populations.This technique could be a practical, cost-effective, and long-term survey design for an agency to implement that will require only minimal staff time. In fact the NWTF has estimated a series of national research priorities, of which one includes "Evaluate techniques for estimating turkey population across large spatial scales". The purpose of this proposed research is to determine the utility of trail camera surveys and SCR models to derive parameter estimates of abundance and density for individual WMAs in the Ozark Mountains in northern Arkansas. WMA specific estimates of density will facilitate effective harvest and habitat management decisions. In particular, agencies can annual estimate population abundance in accordance with estimated harvest rates in the region to regulate turkey harvest through adaptive harvest management and a regulated permit hunting system. Lastly, modeling spatial variation in density as it relates to forest harvest operations, wildlife openings, and prescribed forest to investigate direct impacts of forest management on wild turkey populations to ensure they are achieving their desired goals.
Animal Health Component
100%
Research Effort Categories
Basic
(N/A)
Applied
100%
Developmental
(N/A)
Goals / Objectives
The purpose of this proposal is to outline my five-year plan to obtain the information needed to achieve science-based harvest and habitat management decisions for wild turkey across forested landscapes in the Ozark and Ouachita Mountain regions of north Arkansas. The following projects will be conducted during time frame of 1 Oct 2017 - 30 Sept 2022. I discuss in detail multiple aspects of this research and cover in detail 6 measurable objectives, the methods for evaluating objectives, an anticipated timeline for completion, and a projected budget. Area of research and specific objectives for this project are as follows:Estimate annual and seasonal survival rates of adult and juvenile male turkeysDetermine harvest rate for adult male turkeys during spring hunting seasonDerive spatially-explicit estimates of population abundance and density of wild turkeysDetermine spatial variation in density associated with forest composition and prescribed fireModel effects of prescribe forest fire management on seasonal habitat use of male turkeysIndividual tree characters and landscape factors influencing nightly roost sites of male turkeys
Project Methods
Capture and Marking--. During 2017-2019, we will live-trap turkeys in the Ozarks and Ouachita Mountains using rocket nets at camera-monitored bait stations during January-March, annually. The age-class of captured turkeys will be determined based on feather characteristics of the ninth and tenth primaries as defined by Pelham and Dickson (1992). All male turkeys will be fitted with a uniquely numbered aluminum rivet leg band. Rivet bands will be used because retention rates of aluminum butt-end bands are as low as 50% on wild turkeys (Diefenbach and Vreeland 2010). We will attach backpack style, battery-powered PTT satellite transmitters (North Star Science and Technology, Inc., King George, VA) to as many as 15 gobblers and 15 jake turkeys annually in each of the Ozark and Ouachita Mountain regions (n = 60 PTTs). The PTT duty cycles will be programed to collect 5 GPS location points per day throughout the life the PTT unit. PTTs will transmit data through satellite on a schedule of 8 h on, 74 h off. All PTTs will be equipped with a 4-hour, motion-sensitive mortality sensor. Location data for all marked turkeys will be downloaded from Argos (www.agros.gov) every 3 days (i.e., 72 hour transmission cycle) and monitored for potential mortality events (multiple location points from a single location for an individual turkey).Survival and Harvest Estimates--. We will model survival rates for adults and jakes using the Kaplan-Meier staggered entry approach (Kaplan and Meier 1958, Pollock et al. 1989). The staggered entry approach is most appropriate given that all marked turkeys will not be released into the population concurrently, thus allowing us to estimate survival for the duration of the study period. Individuals that survival <1 week will be assumed to have exhibited capture myopathy and thus will be censored from the data set to prevent potential bias resulting from handling and increased stress levels. Survival will be modeled as a function of age, year, and season within year covariates, and will be tested against estimates of constant survival probabilities. The most parsimonious or best-fit a priori model explaining survival rates will be determined by ranking Akaike's information criterion adjusted for small sample sizes (AICc). Model averaging will be conducted to determine the effects size of the most influential covariates that appear in those models within <2.0 ΔAICc of the top model, given that the covariates are not serving as uninformative parameters (Arnold 2010).Habitat Selection--.Seasonal home range estimates will be derived using a kernel density estimator (Kernohan et al. 2001). These data will be used to determine the influence of forest management actions including timber harvest operations and prescribed forest fire on space use. To investigate patch-level and landscape-level influences, we will model macro-habitat characteristics by measuring forest structure and composition at the stand-level and at the landscape-level. Field work will consist of measuring general forest structure and composition attributes that are hypothesized to influence turkey habitat use and survival. In addition, landscape cover data will be downloaded from the National Land Cover Data Set (NLCD 2012), and a grid of random points will be ground-truthed for accuracy. Generalized linear models will be used to rank a priori models to determine the influence of these covariates of survival, movement rate, and home range of jakes and gobblers.Spatial-Capture Recapture models--. We will apply a spatial-capture recapture (SCR) model augmented with telemetry data using camera-trap and telemetry data sets for wild turkey in Arkansas (Fiske and Chandler 2011, Sollmann et al. 2013). SCR uses a point process model to delineate the distribution of animal activity centers (si = activity center or home range center for individual i ) across the study area by evaluating the number of photographs (i.e., detections) collected at each camera trap. The number of activity centers will be determined across the area of interest (i.e., the area within the sampling grid). The SCR model assumes a decreasing function of distance (-dist) between trap locations (x) and the activity center (s), in other words the number of detections for an individual will decrease with increasing distance from the activity center. The model estimates encounter probably using the function [p(x, s) = p0 *exp(-dist (x,s)2/σ2]. In addition to estimating density by identifying si, we can also determine how auxiliary covariates(x) influence density (D(s)) by fitting a regression model with the following function [log(D(s)) = beta0 + beta1*factorx (s)] (Borchers & Efford 2008, Royle et al. 2013b). The assumption lies in the fact that the encounter probability of an individual at a trap location is affected by how that individual uses space (i.e., resource selection). From this, density surface maps can be produced, also in Program R (see example in Figure 3; Royle et al 2013b), which effectively will illustrate spatially explicit density estimates as a function of covariates in the model (Royle et al. 2013b). In the case of this proposed study we are interested in habitat patches of prescribed fire, wildlife openings created from timber harvesting operations, among other management activities.