ORIGINAL PAPER Latitude and daily-weather effects on gobbling activity of wild turkeys in Mississippi Matthew D. Palumbo 1,2 & Francisco J. Vilella 3 & Guiming Wang 1 & Bronson K. Strickland 1 & Dave Godwin 4,5 & P. Grady Dixon 6 & Benjamin D. Rubin 7 & Marcus A. Lashley 1 Received: 31 July 2018 / Revised: 23 March 2019 / Accepted: 3 April 2019 / Published online: 25 April 2019 # ISB 2019 Abstract Weather has been recognized as a density independent factor influencing the abundance, distribution, and behavior of vertebrates. Male wild turkeys ’ ( Meleagris gallopavo ) breeding behavior includes vocalizations and courtship displays to attract females, the phenology of which can vary with latitude. State biologists design spring turkey-hunting season frameworks centered on annual vocalization patterns to maximize hunter engagement. The Mississippi Department of Wildlife, Fisheries, and Parks has tradi- tionally instituted a statewide, 7-week, spring harvest season. However, hunters routinely argue that different peaks in gobbling activity across the state exist. The objective of this study was to determine whether differences in peak gobbling activity existed across a latitudinal gradient of Mississippi and assess the effect of weather on gobbling. During 2008 and 2009, we conducted a statewide gobbling survey. We used generalized additive mixed models to describe the probability and frequency of gobbling activity within northern and southern regions of the state. We also investigated the effect of daily weather conditions on gobbling activity. Our results revealed an approximate 10 – 14-day difference in peak gobbling activity between southern and northern Mississippi. The majority of all gobbling activity occurred within the current spring harvest framework. Perhaps more impor- tantly, gobbling activity was more prevalent on days of regionally dry conditions (i.e., less humid) according to the Spatial Synoptic Classification. Our results provide information on gobbling activity phenology relative to hunting-season dates and weather-response information. Our approach may be particularly applicable in states with relatively shorter seasons or highly variable daily weather conditions that moderate gobbling frequency. Keywords Call counts . Generalized additive mixed model . Phenology . Road survey . Spatial Synoptic Classification Introduction Across the globe natural resource managers strive to under- stand the effects of weather on wildlife populations to imple- ment effective conservation and game species management (Malan et al. 1993; R ō del and Dekker 2012; Gunnarsson et al. 2012; Notaro et al. 2016). During the breeding season weather can influence reproductive behavior and timing (e.g., effort, success, phenology) which thereby influence fitness (Moss 1986; Wilson and Martin 2010). Males of many bird * Matthew D. Palumbo Matthew.Palumbo@dec.ny.gov 1 Department of Wildlife, Fisheries and Aquaculture, Box 9690, Mississippi State University, Mississippi State, MS 39762-9690, USA 2 Present address: New York State Department of Environmental Conservation, 625 Broadway, Albany, NY 12233-4754, USA 3 U.S. Geological Survey, Mississippi Cooperative Fish and Wildlife Research Unit, Department of Wildlife, Fisheries and Aquaculture, Mississippi State University, Mississippi State, MS 39762-9690, USA 4 Present address: Mississippi Forestry Association, 620 North State Street, Suite 201, Jackson, MS 39202-3398, USA 5 Mississippi Department of Wildlife, Fisheries, and Parks, 1505 Eastover Drive, Jackson, MS 39211, USA 6 Department of Geosciences, Fort Hays State University, Hays, KS 67601, USA 7 Department of Biology, Western University, London, ON N6A 3K7, Canada International Journal of Biometeorology (2019) 63:1059 – 1067 https://doi.org/10.1007/s00484-019-01720-2 species sing during the breeding season to attract potential mates and establish and defend territories from competitors. Male wild turkeys ( Meleagris gallopavo ) attract females dur- ing the breeding season through a combination of vocal (e.g., gobbling) and physical (e.g., strutting) courtship displays (Williams 1984). Wild turkey gobbling behavior is influenced by weather conditions that can differ regionally (Bevill Jr 1975; Kienzler et al. 1996; Miller et al. 1997b). Spring hunting seasons allow hunters to benefit from the male turkey ’ s pro- pensity to vocalize during the breeding seasons as gobbling can aid hunters to locate males and hunters typically attempt to lure male turkeys into range by mimicking receptive females. Therefore, understanding regional differences in a behavior that is influenced by weather conditions is important to the hunters that pursue them and the biologists that manage the species. Spring hunting seasons for male wild turkeys in the USA are typically designed in response to the phenology of repro- ductive behaviors, and also tradition, which can influence hunting opportunity (Kurzejeski and Vangilder 1992; Whitaker et al. 2005). Timing of the spring hunting season is based on two peaks in gobbling activity; the first corresponds to disaggregation of winter flocks and the second to a peak in egg incubation by female turkeys (Baily and Rinell 1967; Bevill Jr 1975; Porter and Ludwig 1980; Miller et al. 1997a). For some US states (e.g., Iowa), timing of the spring hunting season is designed to capture this second peak (Kurzejeski and Vangilder 1992; Kienzler et al. 1996) under the assumption that if most female turkeys are incubating eggs then unintended harvest of females will be unlikely (Bevill Jr 1975; Hoffman 1990; Kurzejeski and Vangilder 1992; Kienzler et al. 1996). Overall, wild-turkey managers schedule the spring hunting season taking into account hunter satisfac- tion, which can depend on gobbling activity of male turkeys (Oleson and He 2004). The ability to meet this goal, in turn, requires reliable knowledge of spring turkey gobbling patterns. Regional gobbling patterns can be variable and influenced by a variety of factors. Kienzler et al. (1996) reported that turkey gobbling activity in Iowa (USA) decreased in the pres- ence of hunters. Furthermore, Miller et al. (1997a) document- ed a single peak in gobbling activity that was not associated with peak incubation and concluded that gobbling activity was influenced by several biological variables. Miller et al. (1997b) reported that age structure (i.e., proportion of 2- year-old gobblers in the population) could be another factor influencing gobbling activity. In addition to population char- acteristics, researchers have suggested that environmental fac- tors such as precipitation, wind speed, light intensity, humid- ity, and temperature may influence gobbling activity (Bevill Jr 1975; Kienzler et al. 1996; Miller et al. 1997b). The spring turkey-hunting season in Mississippi, USA, is approximately 7 weeks long and is designed to capture the majority of spring gobbling activity (Palumbo 2010). However, turkey hunters in the state have argued that the timing of the spring hunting season does not coincide with peak gobbling activity in their respective regions. The com- mon claim is that peak gobbling activity in southern Mississippi tends to occur before the season starts while hunters in northern Mississippi claim that the spring gobbling peak occurs very late in the season (personal communication, Dave Godwin, Mississippi Department of Wildlife, Fisheries, and Parks, Wild Turkey Program Coordinator). Both situa- tions would supposedly reduce the amount of time when vo- calizing male turkeys would be available for harvest, with potential impact to hunter success. The claim by hunters, that there is a latitudinal gradient in peak gobbling activity, agrees with the common biological hypothesis that seasonal breeding behavior is triggered with an increase in photoperiod (Healy 1992), which also follows a latitudinal gradient. Therefore, latitude can influence the onset of the wild turkey breeding season while weather conditions can affect within season be- haviors (i.e., propensity to gobble). Mississippi has a rich tradition of wild-turkey hunting, an activity that generates approximately $90 million annually in direct, indirect, and induced economic expenditures in the state (Henderson et al. 2010). As a valuable natural resource, it is important for managers to understand the spatio-temporal variability of biological processes that affect the decision- making process for creating spring wild-turkey harvest sea- sons. We instituted this study to examine these concerns and developed a large-scale, statewide approach to assess varia- tions in gobbling activity. Here we report spatial and temporal variations of gobbling activity in Mississippi, and weather conditions associated with gobbling activity. Materials and methods We divided the state of Mississippi into three regions to ad- dress the issue of latitudinal differences in spring gobbling. The northern region is the area north of US Highway 82, which extends westward from the city of Columbus to the city of Greenville. The central region is south of US Highway 82 and north of US Highway 84, which extends westward from the city of Waynesboro to the city of Natchez. The southern region is south of US Highway 84 to the coast (Fig. 1). We selected the northern and southern regions of the state for our study; the central region was not surveyed for gobbling activity. We divided each survey region into a grid of 250-km 2 cells, and we randomly selected nine cells for each respective re- gion. Within each selected cell, available harvest data and expert opinion provided by Mississippi Department of Wildlife, Fisheries, and Parks (MDWFP) biologists and area managers were used to delineate areas that supported B high, ^ 1060 Int J Biometeorol (2019) 63:1059 – 1067 B medium, ^ and B low ^ abundance of turkeys. We then ran- domly selected an area identified as high or medium based on turkey abundance for each selected cell. Finally, inside each of the selected areas, we randomly selected a county road as the starting point of each survey route. Each survey route consisted of 10 points spaced at 1.6-km intervals on county roads. We used GIS to initially locate survey routes across each region. After verifying locations on the ground, if a route was considered to be in a poor location (i.e., heavy traffic or residential area), it was replaced by another randomly selected county road to be the start of the survey route. As a result of the selection process, we relocated one route and removed three routes (one in the northern region and two in the southern region). Our route selection process resulted in eight routes in the northern region and seven routes in the southern region. The selected set of routes was based on location, lo- gistical survey support provided by MDWFP staff, and initial assessment of turkey habitat while verifying survey routes. Gobbling-call count surveys were conducted at least 2 days per week from 11 February to 31 May 2008 and 15 February to 30 May 2009. These periods represented approximately 1 month prior to and 1 month past the current spring hunting season in Mississippi. Surveys began 30 min before sunrise. Observers listened for 4 min at each stop and recorded the number of individual gobblers and gobbles heard (Bevill 1975; Lint et al. 1995; Miller et al. 1997b). The average Esri, HERE, Garmin, © OpenStreetMap contributors, and the GIS user community Fig. 1 Map of Mississippi illustrating northern and southern regions used for wild turkey gobbling phenology surveys during 2008 and 2009 Int J Biometeorol (2019) 63:1059 – 1067 1061 amount of time to travel between survey points was 3 min and 37 s. Therefore, an average survey lasted approximately 73 min. Observers always started each route at one of its terminal points. During inclement weather (rain or wind ≥ 12.1 km h − 1 ), surveys were postponed until the next possible day due to the potential of observers ’ detections being influ- enced by the rain or wind (Healy and Powell 2000). All observers had experience or training in detection and identification of gobbling activity as well as in the survey protocol before each survey season (Healy and Powell 2000). Observers were a combination of hired technicians and MDWFP staff. Because observers had varying experience levels in detecting individual gobblers but were either trained or had experience in identifying turkey vocalizations (i.e., gobbling), we used B number of gobbles heard per day ^ to reduce bias associated with identifying individuals. Due to logistical constraints of surveying across the latitudinal ex- tremes of the state and the need to meet certain weather criteria to conduct a survey, there was a potential that multiple routes or no routes were surveyed on any given day per our survey design. We assessed daily variation in weather conditions from categories generated from the Spatial Synoptic Classification (SSC) system (http://sheridan.geog.kent.edu/ssc.html). The SSC uses surface-based data to establish air-mass types for an area (Sheridan 2002). The SSC categories are dry moderate (DM), dry polar (DP), dry tropical (DT), moist moderate (MM), moist polar (MP), moist tropical (MT), and transitional (TR). Dry moderate is typically mild and dry to dry and warm. Dry polar is cool or colder dry air with northerly winds with little cloud cover. Dry tropical is the hottest, sunniest, driest classification. Moist polar conditions are cool, cloudy, and humid with moist moderate having similar conditions but are relatively warmer or more humid. Moist tropical condi- tions are warm and very humid. Transitional conditions are when one weather type is changing to another. The weather types are geographically and temporally relative, so a DP day in Mississippi is very different from a DP day in more north- ern latitudes, and at a given location an MT day in January is quite different from one in July. Nevertheless, these weather types were designed to be particularly useful for biometeorol- ogy applications, and the SSC has been a widely used tool for such purposes (Dolney and Sheridan 2006; Senkbeil et al. 2007; Davis et al. 2012; Hondula et al. 2013; Dixon et al. 2016). There were four SSC measuring stations within or near the study area (Memphis, Tennessee; Jackson, Mississippi; Meridian, Mississippi; and New Orleans, Louisiana). We used the nearest SSC station to assess the weather type for each survey route (Fig. 2.) One survey route in the northern region of Mississippi was not within 150 km of a SSC station and was therefore not included in the analysis. Also, 26 route- survey days were eliminated from the remaining routes because SSC classifications were not available at the corre- sponding station. Therefore, our total sample was 755 route survey days from 7 northern and 7 southern survey routes. We used generalized additive mixed models (GAMMs) to represent two response variables: the presence or absence of gobbling for a given survey route on a given day and the number of gobbles counted at a given route on a given day. GAMMs are semi-parametric and suited to address non-linear relationships that are not easily described by polynomial re- gression or addressed through data transformations. Zuur et al. (2009) provide an introduction to the application of GAMMs in ecology, and Wood (2006) provides a more technical de- scription. These models extend generalized linear models to include smooth functions of the dependent variables and ran- dom effects that can accurately include data from hierarchical study designs. Modeling employed maximum likelihood pa- rameter estimates based on Laplace approximation and was implemented using the gamm4 package (Wood and Scheipl 2014) of R version 3.2.3 (R Core Team 2015). The presence or Fig. 2 Map of locations of the Spatial Synoptic Classification System stations (black dots) used to assess the influence of climatic conditions on gobbling activity data collected from survey route points (white dots) in Mississippi during 2008 and 2009. Rings around the Spatial Synoptic Classification station locations are in 50-km intervals 1062 Int J Biometeorol (2019) 63:1059 – 1067 absence of gobbling was modeled as a binary variable with a logit link. The count of gobbles was over-dispersed relative to a Poisson distribution (primarily due to the large number of zeros, 468 of 755 route days). We therefore modeled it using a negative binomial distribution with a natural log link. In gamm4, negative binomial regression requires estimating an initial value of the θ parameter (Venebles and Ripley 2002). We estimated θ based on the fit of a generalized additive model without random effects (as suggested by Zuur et al. 2009). The random components of our statistical models were route-nested within each SSC station because data from a single SSC station were used to represent conditions at several routes (Fig. 2). We aggregated SSC categories based on hu- midity (SSC.M) or temperature (SSC.T) and included each as a categorical fixed effect in the models. The three humidity categories of SSC.M were B dry ^ (DM, DP, DT), B moist ^ (MM, MP, MT), and B transitional ^ (TR). The four tempera- ture categories of SSC.T were: B polar ^ (DP, MP), B moderate ^ (MM, DM), B tropical ^ (MT, DT), and transitional (TR). It was not possible to include both aggregations in the same GAMM because the transitional category present in each variable caused the model to be rank-deficient and gamm4 would not produce parameter estimates. Therefore, we started model se- lection with two full logistic models each of which included one of the aggregations of SSC types and two corresponding full negative binomial models. Other fixed effects included in the full models were a thin- plate regression spline (TPRS) for Julian day and two TPRS applied to subsets of the data to detect non-linear effects of region and year. TPRS are the default smoothing method in the R package mgcv (Wood 2011), which is called by gamm4. They are appropriate for our analysis due to their computation efficiency and because they do not use knots and are thus not sensitive to knot number or location (Wood 2006, 2011). After running initial models, non-significant fixed effects (approxi- mate p values based on χ 2 statistics, α = 0.05) were removed until no further simplification was possible. We also report the effective degrees of freedom (edf.) for each smoothed term. An edf. increasing from two indicates increasing variability (i.e., B wiggliness ^ ) compared to a linear relationship (Zuur et al. 2009). We assessed the goodness of fit of the final models using adjusted R 2 and by inspecting plots of normal- ized residuals (Wood 2006; Zuur et al. 2009). Results We recorded 672 gobbles in the northern region of Mississippi during 2008. Of these 672 gobbles heard dur- ing the 2008 sample period (11 February 2008 – 31 May 2008), 68% occurred during the hunting season (16 March 2008 – 3 May 2008). During 2009, we recorded 542 gobbles. Of these 542 gobbles recorded during the 2009 sample period (15 February 2009 – 30 May 2009), 86% occurred during the hunting season (15 March 2009 – 2 May 2009). In the southern region of the state, we record- ed 458 gobbles during 2008. Of the 458 gobbles observed during the 2008 sample period (17 February – 31 May), 67% occurred during the hunting season (16 March – 3 May). In 2009 we recorded 458 gobbles (15 February – 30 May), and 71% occurred during the hunting season (15 March 2009 – 2 May 2009). We conducted 47% of our surveys within the hunting season during 2008 and 49% within the hunting season during 2009. In the full logistic model, the SSC.T weather aggregation was not significantly ( P ≥ 0.2 for all categories) related to the probability of hearing a gobble on a given day. Therefore, we continued further model simplification with the full logistic model containing SSC.M. For this model, the smooth term B Julian day by year ^ was not significant (edf. = 2, χ 2 = 1.7, P = 0.43); thus, we pooled years for further analysis. After the removal of the year effect, all model terms were significant and adjusted R 2 was 0.11. Plots of resid- uals against fitted values and all predictor variables indi- cate generally strong fit, but they revealed two data points with high residuals. Removal of these two points did not substantively change any model output, so the results re- ported include them. The smooth term of B Julian day ^ was significantly (edf. = 4.2, χ 2 = 68.86, P < 0.0001) relat- ed to the probability of hearing a turkey gobble, as was the smooth term of B Julian day by region ^ (edf. = 2, χ 2 = 18.14, P < 0.0001), indicating differences in regional dis- tributions (Fig. 3). When comparing the SSC.M weather categories, the odds of hearing a turkey gobble vary de- pending on each weather type. Relative to the odds of hearing a turkey gobble on dry days (the reference condi- tion in our model), the odds of hearing gobbling on moist days was reduced by approximately 60% ( P < 0.0001, 95% confidence interval 40 to 73% reduction in odds). On transitional days the estimated reduction in odds was 39% relative to dry days and was not statistically signif- icant ( P = 0.08, 95% confidence interval − 8 to 74%). For all SSC.M categories the predicted peak in the proba- bility of hearing a turkey gobble was 13 April in the northern region and 30 March in the southern region (Fig. 3). The random factor of B route ^ had a variance of 1.14, the random factor of B station ^ had a variance of 0.03, and based on inspecting normal QQ plots, both appeared to be approximate- ly normal. For the negative binomial models we estimated θ to be 0.23. As with the logistic model, SSC.T weather aggregations were not significantly ( P ≥ 0.2 for all categories) related to the count of gobbles on a given day. Therefore, we continued further model simplification with the full negative binomial model containing SSC.M. Again, for this model, the smooth Int J Biometeorol (2019) 63:1059 – 1067 1063 term Julian day by year was not significant (edf. = 2, χ 2 = 1.48, P = 0.33); thus, we pooled years for further analysis. Plots of normalized residuals showed strong positive skew, an indication that the model did not completely compensate for the over-dispersion in counts. Adjusted R 2 was 0.06. The smooth term of Julian day was significantly (edf. = 4.39, χ 2 = 96.88, P < 0.0001) related to the count of gobbles, as was the smooth term of Julian day by region (edf. = 2, χ 2 = 10.54, P < 0.005), indicating differences in regional distributions (Fig. 4). Relative to the number of gobbles heard on dry days (the reference condition in our model), the number heard on moist days was reduced by approximately 47% ( P < 0.0001, 95% confidence interval 5 to 69% reduction). On transitional days, the estimated reduction was 36% relative to dry days ( P = 0.03, 95% confidence interval 24 to 54%). For all SSC.M categories, the predicted peak in the number of gobbles was 17 April in the northern region and 7 April in the southern region (Fig. 4). The random factor of route had a variance of 1.19, the random factor of station had a variance of 0.01, and based on inspecting normal QQ plots, both appeared to be approximately normal. Discussion Gobbling activity appears to occur relatively more during dry conditions. The SSC aggregates allowed us to exam- ine several weather conditions that otherwise are correlat- ed and can be difficult to use simultaneously. During the time of year studied, the SSC.M aggregated class of dry was typically representative of days with low humidity (i.e., dry conditions) and winds from the north or west. Such days are commonly behind (i.e., to the north or west) recently passed cold fronts. Conversely, the aggre- gated class of moist is typically representative of the op- posite: high humidity, winds from the south or east, and little influence from significant surface frontal boundaries. Transitional days represent a shift from one class to the other, which is typically caused by the passage of a front and accompanied by shifts in dew point, cloud cover, and wind throughout the day. Our results indicate that gob- bling activity was more prevalent during days of low hu- midity, similar to results of previous studies (Bevill Jr 1975; Kienzler et al. 1996). Miller et al. (1997b) reported Fig. 4 The predicted count and 95% confidence intervals of turkey gobbles in the northern and southern regions of Mississippi during 2008 and 2009 survey periods combined. Categories of B dry, ^ B transitional, ^ and B moist ^ represent the three humidity categories aggregated from the available Spatial Synoptic Classifications. The x axis and spring hunting season are based on 2009 dates Fig. 3 The predicted probability and 95% confidence intervals for hearing a turkey gobble in the northern and southern regions of Mississippi during 2008 and 2009 survey periods combined. Categories of B dry, ^ B transitional, ^ and B moist ^ represent the three humidity categories aggregated from the available Spatial Synoptic Classifications. The x axis and spring hunting season are based on 2009 dates 1064 Int J Biometeorol (2019) 63:1059 – 1067 that dewpoint influences the number of males heard but that the relationship between humidity and number of gobbles and temperature and numbers of gobbles was not detected. Our findings did not suggest that tempera- ture was a significant predictor of gobbling activity while the SSC aggregates of humidity were more correlated. This distinction provides further insight into the relation- ship of humidity, temperature, and day of the year, and may be a result of how the two sources of weather con- ditions were gathered. Ultimately, our models ’ weak ex- planatory power suggests that there may be other influen- tial biotic (e.g., age structure) or abiotic (e.g., hunting pressure) factors that we could not record but have been documented in other studies (Bevill Jr 1975; Palmer et al. 1990; Kienzler et al. 1996; Miller et al. 1997b). Our results indicate an approximate 10 – 14-day differ- ence in peak gobbling activity, defined as either the prob- ability of hearing a turkey gobble or the count of individ- ual gobbles, between northern and southern regions of Mississippi. However, while there was a difference in gobbling between the northern and southern regions, both regional peaks occurred within the current spring harvest- season framework. Moreover, the greatest frequency of gobbling activity for both regions occurred during the spring hunting season rather than before or after, contrary to anecdotal hunter observations. Miller et al. (1997a), using 12 years of data from one site in central Mississippi, reported a yearly difference in median dates in peak gobbling activity with dates ranging from 23 March to 23 April for calls heard, while observing one peak in gobbling activity. The single peak in gobbling activity is similar to our results of 2 years of pooled data (Figs. 3 and 4). But since we sampled across two regions, our results suggest that managers observing a single peak in gobbling activity based on multi-year data sets should be aware that there may be regional or latitudinal differ- ences in these peaks. Our sampling was conducted over broad geographic regions with a limited number of routes, and conse- quently, there was a large amount of spatial, temporal, and environmental heterogeneity within regions unac- counted for by our models. Nonetheless, we documented the temporal peaks in gobbling activity despite the var- iation in vegetation types, land use, hunting pressure, nesting behavior, and population sizes. Previous research documented a multitude of factors that may influence gobbling activity and, consequently, account for tempo- ral and spatial variation in gobbling peaks (Hoffman 1990; Lint et al. 1995; Miller et al. 1997a, 1997b). These factors most likely contribute to our models ’ low explanatory power for probability of hearing a tur- key gobble ( R 2 = 0.11) and especially for the number of gobbles ( R 2 = 0.06). Conclusions The mating system of the wild turkey is considered a male-dominant polygyny. Spring gobbling is critical in assessing a male ’ s ability to attract flocks of receptive females (Buchholz 1997). Consequently, it is important for state turkey programs to understand how geographic, environmental, and anthropogenic factors influence this critical component of their life cycle. Monitoring pro- vides managers with the ability to assess the relationship between gobbling peaks and the structure of the spring harvest season (Healy and Powell 2000). This under- standing of the phenology of gobbling in relation to how state agencies design their harvest seasons is partic- ularly relevant for states with short seasons, as is the case of many states in the northeastern USA (e.g., 3 – 4- week seasons), but determining how these factors influ- ence gobbling activity may be logistically difficult and costly. Future research could utilize portable weather stations to collect site-specific conditions simultaneously with spring gobbling surveys to verify our observations and improve model fit and estimates. Autonomous recording units can be integrated in sampling regimes to eliminate the logistical constraints of needing personnel to conduct surveys (Colbert et al. 2015). Designed experiments that concomitantly monitor hunting pressure and gobbling ac- tivity will assist in determining effects of hunting pressure on gobbling activity. Furthermore, the meteorological conditions identified in our study can inform turkey hunters making decisions to hunt on a particular morning and can be used by state wildlife agencies as an explana- tion for temporal patterns in gobbling activity. The dispar- ity between our findings and purported hunter experiences could be partially due to hunters experiencing decreased gobbling activity on moist days. Finally, our study high- lights the value of using high-quality weather data to val- idate or refute arguments that may negatively impact man- agement and conservation prescriptions. Similarly, the ap- plication of high-quality weather data in ecological stud- ies can be a useful tool understanding the mechanisms that influence animal populations in a multitude of ways including physiology (Cartar and Morrison 1997), move- ment (Juang et al. 2002; Boone and Hobbs 2004), and diseases (Morgan et al. 2006). Acknowledgments We thank MDWFP staff, especially S. Edwards, R. Seiss, and A. Butler, and project technicians for the logistical support. We also would like to thank Darren Miller, an anonymous reviewer, and the editor-in-chief for their review of a previous version of this manuscript and for their helpful comments. Conflict of interest The authors declare that there is no conflict of interest. Int J Biometeorol (2019) 63:1059 – 1067 1065 Funding information This project was funded by the Federal Aid in Wildlife Restoration Funds through the Mississippi Department of Wildlife, Fisheries, and Parks (MDWFP; Project W-48-45, Study 58). References Baily RW, Rinell KT (1967) Events in the turkey year. In: Hewitt OH (ed) The wild turkey and its management. The Wildl. Society, The Wildl. Society, Washington, D. C., USA, pp 73 – 91 Bevill WV Jr (1975) Some factors influencing gobbling activity among wild turkeys. Proc of the Annual Conference of Southeast Association of Game and Fish Commissioners 27:62 – 73 Boone RB, Hobbs NT (2004) Lines around fragments: effects of fencing on large herbivores. Afr J Range Forage Sci 21:147 – 158 Buchholz R (1997) Male dominance and variation in fleshy head orna- mentation in wild turkeys. J Avian Biol 28:223 – 230. https://doi.org/ 10.2307/3676973 Cartar RV, Morrison RG (1997) Estimating metabolic costs for homeotherms from weather data and morphology: an example using calidridine sandpipers. Can J Zool 75:94 – 101 Colbert DS, Ruttinger JA, Streich M, Chamberlain M, Conner LM, Warren RJ (2015) Application of autonomous recording units to monitor gobbling activity by wild turkey. Wildl Soc Bull 39:757 – 763. https://doi.org/10.1002/wsb.577 Davis RE, Rossier CE, Enfield KB (2012) The impact of weather on influenza and pneumonia mortality in New York City, 1975 – 2002: a retrospective study. PlosOne 7(3). https://doi.org/10.1371/journal. pone.0034091 Dixon PG, Allen M, Gosling SN, Hondula DM, Ingole V, Lucas R, Vanos J (2016) Perspectives on the synoptic classification system and its role in interdisciplinary research. Geogr Compass 10:147 – 164. https://doi.org/10.1111/gec3.12264 Dolney TJ, Sheridan SC (2006) The relationship between extreme heat and ambulance calls for the city of Toronto, Ontario, Canada. Environ Res 101:94 – 103. https://doi.org/10.1016/j.envres.2005.08. 008 Gunnarsson G, Waldström J, Fransson T (2012) Direct and indirect ef- fects of winter harshness on the survival of mallards Anas platyrhychos in Northwest Europe. Ibis 154:307 – 317. https://doi. org/10.1111/j.1474-919X.2011.01206.x Healy WM, (1992) Behavior. In: Dickson JG (ed) The wild turkey, biol- ogy and management. Stackpole Books, Harrisburg, Pennsylvania, USA, pp 46 – 65 Healy WM, Powell SM (2000) Wild turkey harvest management: biolo- gy, strategies, and techniques. U. S. Fish and Wildl. Service Biolo gical Technical Pub lication, BTP-R500 1 – 1 999, Shepherdstown, West Virginia, USA Henderson JE, Grado SC, Munn IA, Jones WD (2010) Economic impacts of wildlife and fisheries associated recreation on the Mississippi economy: an input-output analysis. For. and Wildl. Res. Center, Res. Bull. FO398. Mississippi State University, Mississippi State, USA Hoffman RW (1990) Chronology of gobbling and nesting activities of Merriam ’ s wild turkeys. - Proc. of the Natl. Wild Turk. Symposium 6:25 – 31 Hondula DM, Davis RE, Knight DB, Sitka LJ, Enfield K, Gawtry SB, Stenger PJ, Deaton ML, Normile CP, Lee TR (2013) A respiratory alert model for the Shenandoah Valley, Virginia, USA. Int J Biometerol 57:91 – 105. https://doi.org/10.1007/s00484-012-0537-7 Juang P, Oki H, Wang Y, Martonosi M, Peh LS, Rubenstein D (2002) Energy-efficient computing for wildlife tracking: design tradeoffs and early experiences with ZebraNet. ACM SIGARCH Computer Architecture News 30:96 – 107 Kienzler JM, Little TW, Fuller WA (1996) Effects of weather, incubation, and hunting on gobbling activity in wild turkeys. Proc Natl Wild Turk Symp 7:61 – 68 Kurzejeski EW, Vangilder LD (1992) Population management. In: Dickson JG (ed) The wild turkey, biology and management. Stackpole Books, Harrisburg, Pennsylvania, USA, pp 165 – 184 Lint RJ, Leopold BD, Hurst GA (1995) Comparison of abundance index- es and population estimates for wild turkey gobblers. Wildl Soc Bull 23:164 – 168. https://doi.org/10.2307/3782783 Malan G, Little RM, Crowe TM (1993) The effects of hunting effort and weather on hunting success and population dynamics of Namaqua sandgrouse . S Afr J Wildl Res 23:107 – 111 Miller DA, Hurst GA, Leopold BD (1997a) Chronology of wild turkey nesting, gobbling, and hunting in Mississippi. J Wildl Manag 61: 840 – 845. https://doi.org/10.2307/3802192 Miller DA, Hurst GA, Leopold BD (1997b) Factors affecting gobbling activity on wild turkeys in central Mississippi. Proc. of the Annual Conference of Southeast. Assoc Fish Wildl Agencies 51:352 – 361 Morgan ER, Lundervold M, Medley GF, Shaikenov BS, Torgerson PR, Milner-Gulland EJ (2006) Assessing risks of disease transmission between wildlife and livestock: the Saiga antelope as a case study. Biol Conserv 131:244 – 254 Moss R (1986) Rain, breeding success and distribution of Capercaillie Tetrao urogallus and black grouse Tetrao tetrix in Scotland. Ibis 128:65 – 72. https://doi.org/10.1111/j.1474-919X.1986.tb02093.x Notaro M, Schummer M, Zhong Y, Vavrus S, Van Den Elsen L, Coluccy J, Hoving C (2016) Projected influences of changes in weather se- verity on autumn-winter distributions of dabbling ducks in the Mississippi and Atlantic flyways during the twenty-first century. PLoS One 11(12):e0167506. https://doi.org/10.1371/journal.pone. 0167506 Oleson JJ, He CZ (2004) Space-time modeling for the Missouri turkey hunting survey. Environ Ecol Stat 11:85 – 101. https://doi.org/10. 1023/B:EEST.0000011366.68489.d8 Palmer WE, Hurst GA, Lint JR (1990) Effort, success, and characteristics of spring turkey hunters on Tallahala Wildl. Management area, Mississippi. Proc Natl Wild Turk Symp 6:208 – 213 Palumbo MD (2010) Influence of latitudinal and climate variation, and field observations, on spring gobbling phenology of wild turkey in Mississippi. Thesis, Mississippi State University Porter WF, Ludwig JR (1980) Use of gobbling call counts to monitor the distribution and abundance of wild turkeys. Proc Natl Wild Turk Symp 4:61 – 68 R Core Team (2015) R: a language and environment for statistical com- puting. R Foundation for Statistical Computing, Vienna, Austria. URL: https://www.R-project.org/ R ō del HG, Dekker JJA (2012) Influence of weather factors on population dynamics of two lagomorph species based on hunting bag records. Eur J Wildl Res 58:923 – 932 Senkbeil JC, Rodgers JC, Sheridan SC (2007) The sensitivity of tree growth to air mass variability and the Pacific Decadal Oscillation in coastal Alabama. InterNatl J Biometerol 51:483 – 492. https://doi. org/10.1007/s00484-007-0087-6 Sheridan SC (2002) The redevelopment of a weather-type classification scheme for North America. InterNatl J Climatology 22:51 – 68. https://doi.org/10.1002/joc.709 Venebles WN, Ripley BD (2002) Modern applied statistics with S, 4th edn. Springer, New York Whitaker DM, Pack J, Norman GW, Stauffer DF, Klopfer SD (2005) A range-wide meta-analysis of wild turkey nesting phenology and spring season opening dates. Proc Natl Wild Turk Symp 9:351 – 360 Williams LE (1984) The voice and vocabulary of the wild turkey. Gainesville, Florida, USA Wilson S, Martin K (2010) Variable reproductive effort for two ptarmigan species in response to springe weather in a northern alpine 1066 Int J Biometeorol (2019) 63:1059 – 1067 ecosystem. J Avian Biol 41:319 – 326. https://doi.org/10.1111/j. 1600-048X.2009.04945.x Wood SN (2006) Generalized additive models: an introduction with. R. Chapman and Hall/CRC, Boca Raton Wood SN (2011) Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. J R Statist Soc(B) 73:3 – 36. https://doi.org/10.1111/j.1467-9868.2010. 00749.x Wood SN, Scheipl F (2014) gamm4: generalized additive mixed models using mgcv and lme4. R package version 0.2-3.URL: https://CRAN. R-project.org/package=gamm4 Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith GM (2009) Mixed effects models and extensions in ecology with R. Springer, New York Publisher ’ s note Springer Nature remains neutral with regard to jurisdic- tional claims in published maps and institutional affiliations. Int J Biometeorol (2019) 63:1059 – 1067 1067