The Mississippi Department of Wildlife, Fisheries, and Parks utilize data from turkey hunter observations and brood surveys from across the state to manage wild turkey Meleagris gallopavo populations. Since 1995, hunters have collected gobbling and jake observation data, while the Mississippi Department of Wildlife, Fisheries, and Parks' personnel and cooperating wildlife managers of several natural resource agencies throughout the state have collected brood survey data. Both sources of data serve to forecast poult recruitment and gobbling activity. The objective of this study was to evaluate if these data can serve as a viable predictor of gobbling activity. We used three mixed models to investigate the relationship between the number of jakes observed per hour of hunting 1 y prior and the total number of poults per hens 2 y prior (model 1), number of gobblers heard per hour of hunting and the number of jakes observed per hour of hunting 1 y prior (model 2), the number of gobblers heard per hour of hunting and the total number poults per total hens observed 2 y prior (model 3) using data from 1995 to 2008 among five wild turkey management regions encompassing the state. We incorporated region as a random effect to account for spatial variation. We found the number of jakes observed per hour of hunting 1 y prior correlated with the total number of poults per total hens observed 2 y prior. We also found the number of gobblers heard per hour of hunting correlated with the number of jakes observed per hour of hunting 1 y prior. Additionally, we found that the total poults per total hens observed 2 y prior was correlated to the number of gobblers heard per hour of hunting. Our results show promise for using indices of gobbling activity, jake observations, and brood surveys to estimate gobbling activity.
Many states index wildlife population abundance by collecting harvest data or use data gathered during hunting seasons to estimate harvest, the number of hunters, and how many days of hunting (Strickland et al. 1994). For example, Reynolds and Sauer (1991) examined the correlation between annual changes of mallard Anas platyrhynchos breeding populations and harvest rates determined from hunter harvest data, in addition to rates of productivity from breeding population data. However, many of these types of estimates vary greatly, making yearly and regional comparisons difficult (Connelly et al. 2005). Thus, wildlife agencies also rely upon observations from field personnel to index wildlife populations. For some states, recruitment estimates for wild turkey Meleagris gallopavo are indexed through brood surveys conducted by volunteers or staff on “routine duties” during summer (Shultz and McDowell 1957; Wunz and Shope 1980).
The Mississippi Department of Wildlife, Fisheries and Parks (MDWFP) relies on hunter harvest data and field observations (from hunters, staff, and cooperating agencies) to index wild turkey populations throughout the state. A voluntary Spring Gobbler Hunting Survey (SGHS) provide data for indexing harvest rates, population size, age structure, and gobbling activity on an annual basis across the state. Additionally, MDWFP, Mississippi Forestry Commission, U.S. Forest Service, U.S. Army Corps of Engineers, and Weyerhaeuser Company conduct brood surveys and collect data to index reproduction and estimate productivity of wild turkey populations statewide. The MDWFP uses a combination of data from the SGHS and brood surveys to predict the age structure and population levels that spring hunters can expect. The MDWFP also attempts to predict gobbling activity based on the brood survey of 2 y prior and subadult males (jakes) observed 1 y prior from the SGHS.
Previous research on wild turkeys focused on monitoring broods to index population parameters, such as productivity (Speake et al. 1969; Gardner et al. 1972; Glidden 1977; Speake 1980; Bartush et al. 1985). Miller et al. (1997b) reported that nest success 2 y prior was correlated to the number of gobblers heard, but not to the number of calls heard. The authors hypothesized that the proportion of 2-y-old birds in a local population could be indexed by nest success 2 y earlier and could contribute to the potential of hearing an individual turkey gobble. An objective of our study was to evaluate arguments by Miller et al. (1997b) whereby nest success 2 y prior could be an adequate indicator of hearing an individual turkey gobble at a regional scale using the MDWFP brood survey dataset.
Furthermore, on average, adult male turkeys (gobblers) gobble more than subadults (Bevill 1973; Hoffman 1990) and 2- and 3-y age class males are dominant over older male age classes (Watts 1968). This dominance can be displayed through increased gobbling (Bevill 1973). But gobbling activity is extremely variable and influenced by a variety of factors (Hoffman 1990; Lint et al. 1995; Kienzler et al. 1996; Miller et al. 1997a, 1997b). However, the MDWFP assumes the 2-y-old age gobbler class is the most vocal age group of male turkeys (D. Godwin, A. Butler, and J. Koloski, MDWFP, unpublished report). Furthermore, the MDWFP spring harvest regulations protect juvenile males (i.e., jakes) from harvest by mandating that only birds with a 3-inch or longer beard may be legally harvested. The beard length rule supports jake recruitment from the subadult age class into the adult (2-y) age class. We acknowledge that other studies (Hoffman 1990; Lint et al. 1995; Kienzler et al. 1996; Miller et al. 1997a, 1997b) describe the variability in gobbling activity, but since the MDWFP attempts to support recruitment to the 2-y age class and previous research by Miller et al. (1997b) demonstrated significant correlations of nest success with gobbling activity of the 2-y age class we designed our analysis of MDWFP's data to focus on gobblers in the 2-y age class.
Since 1995, turkey hunters have contributed data to MDWFP on the number of gobblers heard per hour of hunting and number of jakes observed per hour of hunting. Further, MDWFP staff and cooperating agencies document number of poults per hen from brood surveys in the five turkey management regions of Mississippi. Our objective was to assess the relationship between counts of jakes observed and poults per hen and subsequent gobbling activity. A relationship among these counts would make predictions of gobbling activity more reliable and strengthen current strategies for turkey harvest recommendations.
In the SGHS, individual hunters voluntarily report throughout the spring season (typically mid-March to 1 May) location of hunt (i.e., whether the hunt was on private or public land); date; start and end time of hunting; county of hunt; number of gobblers heard; number of gobbles heard; number of gobblers, jakes, hens, and unknowns observed; time a gobbler was harvested; and beard and spur lengths (D. Godwin, A. Butler, and J. Koloski, Mississippi Department of Wildlife, Fisheries and Parks, unpublished report). From June to August while conducting daily activities, brood survey participants record number of poults, hens with and without poults, gobblers, and “unknown” (age and sex not discernible) observed, as well as location (county and/or wildlife management area). We assumed that all observations for the SGHS and brood survey were independent.
Sample sizes were the product of the number of years for which we had available data to generate an index, and the number of wild turkey management regions across the state. We calculated mean number of jakes observed per hour of hunting, total poults per total hens observed, and mean number of gobblers heard per hour of hunting from 1995 to 2008 for each turkey management region (Tables S1 and S2, Supplemental Material). Differences in physiography and location of logistical support delineated the five turkey management regions (Figure 1).
We assumed the mean number of gobblers heard per hour of hunting as a conservative index of gobbling activity (gobbling activity index, GAI) since the propensity to gobble may differ among birds. We assumed that the number of jakes observed per hour of hunting would be an adequate index of recruitment of 1-y-old males to the 2-y age class, when gobblers are assumed to be most vocal (jake recruitment index [JRI]). The total number of poults per total hens (brood survey index [BSI]) served as an index of fecundity and recruitment (Glidden 1977; Seiss 1989), and we assumed that BSI from 2 y prior would index abundance of 2-y-old gobblers.
We tested three separate linear mixed models to determine the relationships among BSI, JRI, and GAI from 1996 to 2008 using the SAS 9.2 software package (SAS Institute, Cary, NC). We used the mean count of GAI, JRI, and BSI for each region and year for analysis (Table 1). In model 1 we compared JRI to BSI to determine if poult recruitment was a reliable predictor of jakes observed while hunting the following year (1997–2008). In model 2 we compared GAI to JRI to determine if the mean number of jakes observed while hunting was a reliable predictor of gobbling activity the subsequent year (1996–2008). Finally, with model 3 we compared GAI to BSI to determine if brood survey observations were a reliable predictor of gobbling activity 2 y later (1996–2008). We assessed the influence of region on the relationships between poult, jake, and gobbling counts by incorporating region as random effects in our models. For each of the three models we evaluated the inclusion of region as a random effect and compared it to the same model without region as a random effect and selected the appropriate model structure by comparing resultant AICc values (Littell et al. 2006). From each model, we output parameter estimates and associated errors for use in predictive models. Additionally, to gauge model fit we compared the predicted values from our mixed models to the observed values for JRI in model 1 and GAI in models 2 and 3 using Pearson's correlation coefficient. We acknowledge this is not a true correlation between the response and explanatory variables, but we provide this correlation statistic as a simple measure of model fit to augment the F- and P-values for each model.
We found relationships between current and prior turkey counts in all three models (Figure 2). In model 1, the random effects of region did not improve model fit and was eliminated. The JRI was related to BSI 1 y prior (n = 60, F = 18.95, P < 0.001) with a correlation of r = 0.50 between observed and predicted JRI counts. In model 2, inclusion of region as a random effect resulted in the best model fit. The GAI was related to JRI 1 y prior (n = 65, F = 4.81, P < 0.001) with a correlation of r = 0.73 between observed and predicted GAI counts. Finally, in model 3, inclusion of region as a random effect also resulted in the best model fit. The GAI was related to BSI 2 y prior (n = 60, F = 9.21, P = 0.04) with a correlation of r = 0.74 between observed and predicted GAI counts (Table 2).
Our results at the statewide level suggest that the BSI is related to gobbling activity similar to the relationship that Miller et al. (1997b) determined for local estimates of nest success at one wildlife management area to local gobbling activity. Miller et al. (1997b) did not have to account for the large degree of spatial variation, differences in survey effort, and lack of standardization in terms of monitoring turkey nest success compared to the brood survey framework that MDWFP used. We attempted to address the lack of spatial congruence by incorporating region as a random effect. Despite the coarseness of our data, this approach resulted in detecting statewide correlations for all three models.
We determined the index used (JRI) to track the proportion of juvenile male turkeys in a population that will potentially be recruited to the 2-y-old age class was related to the number of gobblers heard during a hunting season. This relationship suggests an index for the gobbler population 1 y prior can indicate the potential of hearing an individual turkey gobble at the statewide scale. Additionally, the correlation between BSI and JRI suggests it may be useful as a tool for tracking wild turkey population size at the regional scale.
Due to the logistical constraints of collecting statewide data, the MDWFP has used the SGHS and brood survey data together with underlying assumptions. Assumptions include the following: observations were independent; observations were spatially replicated each year; effort was consistent across management regions and years; and observers were equally capable of detecting jakes, gobblers, and broods at the same rate. We also made the assumption that the BSI was an indicator of nest success to evaluate the argument that nest success 2 y prior is an adequate indicator of hearing an individual turkey at the regional scale (Miller et al. 1997b).
The SGHS is a useful tool for examining long-term trends in wild turkey populations across the state. Nevertheless, the amount of spatial and sampling variation between sources of data (i.e., gobbling activity, brood surveys, and jake encounters) may result in models of gobbling activity with less explanatory ability. Although our analysis shows that there is a relationship between all of the indices currently used to track gobbling activity, we recommend the MDWFP should improve the protocols to ensure consistency for brood, jake, and gobbling call count surveys to reduce these sources of variation. Observer detection ability has been recognized as a source of variation in many studies (Diefenbach et al. 2003). Therefore, unbiased observers (i.e., not hunters) consistently assigned to specific survey areas may further reduce variation. Finally, attempting to predict a behavioral response (gobbling) influenced by such a wide variety of environmental and ecological factors may consistently result in a high level of inaccuracy. The indices we investigated may serve to predict regional gobbling activity, recognizing that there is a large amount of natural variation associated with this mating behavior. Therefore, we recommend further research into potential sources of variation (i.e., observers, habitat, hunter effort and success, brood survey effort, weather, population dynamics), which may help improve understanding of these sources of variation and ultimately allow more accurate relationships to be determined. Improvements will benefit state agency efforts to monitor long-term trends in turkey populations and enhance the information of annual forecasts to the hunting public.
Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.
Table S1. Data from the wild turkey (Meleagris gallopavo) Spring Gobbler Hunter Surveys used for the mixed model analyses from 1995–2008 for each of the five Mississippi Department of Wildlife, Fisheries, and Parks wild turkey management regions.
Found at DOI: 10.3996/032013-JFWM-023.S1 (13.7 MB XLSX)
Table S2. Data from the wild turkey (Meleagris gallopavo) brood surveys used for the mixed model analyses from 1995 to 2008 for each of the five Mississippi Department of Wildlife, Fisheries, and Parks wild turkey management regions.
Found at DOI: 10.3996/032013-JFWM-023.S2 (1069 KB XLSX)
We greatly appreciate the efforts of Mississippi's sportsmen and the various natural resource agencies that have gathered wild turkey observations. We thank S.L. Edwards, R.S. Seiss, A. Butler, and J. Koloski for their support. Funding for this project was provided by the Mississippi Department of Wildlife, Fisheries, and Parks through project W-48-56, Study 58 of the Federal Aid in Wildlife Restoration Program. We also would like to thank the Subject Editor and anonymous reviewers for their insightful comments and recommendations during the review of the manuscript.
Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Palumbo MD, Vilella FJ, Strickland BK, Wang G, Godwin D. 2014. Brood surveys and hunter observations used to predict gobbling activity of wild turkeys in Mississippi. Journal of Fish and Wildlife Management 5(1):151–156; e1944-687X. doi: 10.3996/032013-JFWM-023
The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.