Received: 21 July 2021 | Accepted: 16 September 2021 DOI: 10.1002/wsb.1268 H U M A N D I M E N S I O N S Connecting hunt outcomes to the demographics, behaviors, and experiences of wild turkey hunters in Mississippi Adam B. Butler 1 | Guiming Wang 2 1 Mississippi Department of Wildlife, Fisheries, and Parks, 1505 Eastover Drive, Jackson, MS 39211, USA 2 Department of Wildlife, Fisheries, and Aquaculture, Mississippi State University, Box 9690, Mississippi State, MS 39762, USA Correspondence Adam B. Butler, Mississippi Department of Wildlife, Fisheries, and Parks, Jackson, MS 39211, USA. Email: Adam.Butler@wfp.ms.gov Funding information Federal Aid in Wildlife Restoration Funds; MDWFP; Department of Wildlife, Fisheries, and Aquaculture Abstract The ability to evaluate population trends through time is essential to successful conservation of wildlife. Directly estimating abundance of eastern wild turkeys ( Meleagris gallopavo silvestris ) is difficult, with few practical options for wide ‐ scale population monitoring. As a result, managers often rely on abundance indices derived from hunter harvest. Catch ‐ per ‐ unit ‐ effort (CPUE) is a commonly used metric which standardizes harvest totals and has been shown to track turkey abundance, but its use is predicated on an assumption of spatially and temporally consistent hunter efficiency. From 2015 – 2018, we surveyed avid turkey hunters in Mississippi to evaluate mechanisms driving harvest efficiency. We ques- tioned each hunter about their demographics and behaviors, collected observations from their hunts, and used 2 ‐ stage hurdle models to relate these variables to individual CPUE. Hunter attributes had variable effects on hunt outcomes. Hunter age had a significant positive effect on the probability of harvesting at least one turkey (an approximate 2% gain in success per year of age) yet was only weakly associated with CPUE of successful hunters. When considering only successful hunters, the percentage of hunts in which a turkey was heard, and the standardized index of total turkey observations were related to CPUE. Additionally, we determined the tendency for hunters to miss a turkey during the hunting season decreased with increasing hunter age. Our results indicate individual characteristics subject to temporal change play a role in determining CPUE and thereby have the potential to mask Wildlife Society Bulletin 2022;46:e1268. wileyonlinelibrary.com/journal/wsb © 2022 The Wildlife Society. | 1 of 13 https://doi.org/10.1002/wsb.1268 trends in abundance derived from CPUE data. Moreover, evidence suggested relative hunter efficiency was amplified when turkey encounters were fewer. Collectively, our results urge caution in use of CPUE as a surrogate for abundance estimates. K E Y W O R D S Avid hunter survey, catch ‐ per ‐ unit ‐ effort, citizen science, eastern wild turkey, hunter behavior, hunter demographics, hunting success, hurdle model, Meleagris gallopavo silvestris , 2 ‐ stage model The evaluation of population change through time is a primary underpinning of successful wildlife conservation. Without the ability to measure population change, managers cannot accurately assess the state of wildlife resources nor measure the response of those resources to management. Abundance of eastern wild turkeys ( Meleagris gallopavo silvestris ; hereafter, turkeys), like many wildlife species, can be notoriously difficult to estimate (Cobb et al. 2001), with few practical options available to directly assess population status at scales to which most policy decisions are directed (Kurzejeski and Vangilder 1992, Leopold and Cummins 2016). Consequently, those tasked with the management of this resource often rely on indirect abundance indices as indicators of population health and sustainability (Kurzejeski and Vangilder 1992, Healy and Powell 2000, Butler and Godwin 2017). Catch ‐ per ‐ unit ‐ effort (CPUE) is a commonly used index that can be cheaply and easily obtained from hunter harvest data. Trends in CPUE are a reasonable approximator of relative abundance if the relationship between effort and harvest changes linearly and in proportion to abundance (Lancia et al. 2005). When this assumption holds, CPUE removes a source of bias from raw harvest totals by accounting for differences due to changing levels of hunter effort. The relationship between CPUE and turkey population abundance has been established from band ‐ recovery modeling (Lint et al. 1995), and variations in CPUE can be predicted by recruitment surveys (Butler et al. 2015). In addition, catch ‐ per ‐ unit ‐ effort has been used in the development of turkey abundance estimates (Clawson et al. 2015, Stevens et al. 2020) and to demonstrate synchrony across metapopulations (Wang 2018). Moreover, collection of turkey hunter CPUE has been recommended to state wildlife agencies to ensure harvest trends provide reasonable inference about population status (Byrne et al. 2015). Despite its use in the management of turkeys and other games species, the underlying assumptions of CPUE data are rarely examined. A primary assumption is consistency of the catchability coefficient; the proportion of the population harvested by a single unit of effort must remain constant for inference based on CPUE data to be valid (Healy and Powell 2000, Lancia et al. 2005). In this view, catchability can be understood to be the mean harvest vulnerability of individual animals across the population. For a stable population with constant removal effort, variations in CPUE would then only result from variations in animal vulnerability (Arreguín ‐ Sánchez 1996). Vulnerability to harvest may be influenced by several factors including animal behaviors (Rose and Kulka 1999, Gross et al. 2015, Hayato 2017), population density (Willebrand et al. 2011), natural and human ‐ made landscape features (Bonaudo et al. 2005, Breisjøberget et al. 2018), naivety of the harvested population (Hilborn and Walters 2013), or hunter efficiency (Bowyer et al. 1999, Guthery et al. 2004). Any of the aforementioned factors can affect CPUE if their mean values change across space or time. Hunter efficiency is a relative measure of the likelihood for individual hunter success, regardless of population abundance of the prey species. Put differently, at a fixed prey density, more efficient hunters would be expected to be successful more often than less efficient hunters. Potential differences in individual hunter efficiency are 2 of 13 | BUTLER AND WANG 23285540, 2022, 2, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1268 by University Of Florida, Wiley Online Library on [10/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License acknowledged (Maunder et al. 2006, Parent et al. 2015), but for many widespread applications of CPUE estimates, mean efficiency is assumed to remain unchanged (Healy and Powell 2000, Asmyhr et al. 2012). Nonetheless, the techniques (Allen et al. 2020), strategies (Vajas et al. 2020), and tools (Weckerly et al. 2005) employed by hunters can influence success and may vary over time. Furthermore, hunter demographics are not static; mean hunter age is increasing across North America (Casalena et al. 2011, Porter et al. 2011) and hunter motivations and preferences often change with age, which, in turn, influence harvest tendencies (Bhandari et al. 2006, Watkins et al. 2018). The likelihood of a hunter accurately hitting their target is a component of hunter efficiency, and variables associated with the hunter or hunting scenario may influence accuracy (Aebischer et al. 2014). Divergence in hunter demographics, behaviors, and/or technologies may therefore potentially affect hunter efficiency and create spatial or temporal heterogeneity in the catchability coefficient of mean hunter CPUE. Such heterogeneity may bias interpretations of CPUE derived abundance trends if the influence of altered hunter efficiency on catchability is not acknowledged. In their analysis of turkey population trends in the midwestern United States, Parent et al. (2015) acknowledged this possibility by specifically noting that inference from their results could be compromised if turkey hunters were collectively becoming more efficient at harvesting turkeys through time. Given concerns regarding perceived declines in turkey populations (Byrne et al. 2015, Casalena et al. 2015, Eriksen et al. 2015, Butler and Godwin 2017), it is critical for resource managers relying on CPUE as a surrogate for population abundance to ensure CPUE trends are directly reflective of trends in population abundance rather than changes in catchability. Although catchability may be influenced in various ways, herein, we focused on evaluating hunter efficiency by attempting to decouple individual ‐ hunter variables that may influence efficiency from the turkey ‐ population variables that are naively assumed to drive CPUE. To do this, we examined data gathered from a survey of avid turkey hunters in Mississippi in which participants recorded observations and harvest from their hunting trips, in conjunction with attributes of their demographics, behaviors, and tools. The combination of 2 differing sources of information — observational – harvest data linked to hunter behavioral data — provided a unique opportunity to examine the relationship between hunter ‐ specific traits and behaviors, the observations and experiences of those same hunters while afield, and their hunt outcomes. Our primary hypothesis was individual CPUE would be principally predicted by the abundance and observability of the turkey population and would be unaffected by the demographics, behaviors, and tools employed by individual hunters. We also investigated factors influencing the likelihood turkey hunters would shoot at and miss turkeys, allowing us to test a second hypothesis regarding factors influencing how shooting accuracy influences hunter efficiency. We hypothesized misses would be random events, unrelated to the hunter's demographics, behaviors, or tools, and thereby inconsequential to CPUE trends. STUDY AREA The Mississippi Department of Wildlife, Fisheries, and Parks (MDWFP) divides Mississippi into 5 turkey management regions based upon physiographic (Pettry 1977) and political boundaries. The northeast region consisted of 21 counties in north ‐ central Mississippi primarily composed of oak ‐ hickory ( Quercus spp. ‐ Carya spp.) and oak ‐ pine ( Pinus spp.) forests, row crop agriculture, and pasture. The 10 counties in the northwest, adjacent to the Mississippi River, comprised the Delta region. Forests of bottomland hardwood species formed a long, narrow band of contiguous turkey habitat along the Mississippi River's margin, whereas extensive row crop agriculture dominated the region's interior. The east ‐ central region was composed of 21 counties, which were primarily comprised by loblolly ‐ shortleaf pine ( P. taeda ‐ P. echinata ) and oak ‐ pine forests. The 12 counties comprising the southwest region were heavily wooded, primarily by oak ‐ hickory forests. The southeast region consisted of 18 counties, which were also heavily forested and dominated by loblolly, longleaf ( P. palustris ), and slash pine ( P. elliottii ). CONNECTING HUNT OUTCOMES TO DEMOGRAPHICS, BEHAVIOR, AND EXPERIENCES | 3 of 13 23285540, 2022, 2, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1268 by University Of Florida, Wiley Online Library on [10/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Mississippi had a spring, male ‐ only turkey season that ran from 15 March until 1 May each year. The spring season was preceded by a week ‐ long special season for hunters ≤ 15 years of age. The seasonal bag limit was 3 adult males or males which showed characteristics of adulthood (beard length ≥ 15.24 cm, fully rounded tail fan, or other characteristics of adult male turkeys). Hunters aged 15 years and younger could harvest 3 males of choice, including juvenile males. METHO DS Data collection Data for our study was collected as part of the MDWFP's annual avid hunter survey, which was initiated in 1995 to gather supplemental information on statewide turkey populations, hunter effort, and harvest. We used survey data gathered from 2015 – 2018. We provided hunters with data collection materials prior to the start of the spring turkey season and asked them to return data to the MDWFP at the season's conclusion. We did not choose survey participants at random. Instead, the MDWFP advertised the opportunity in various outlets and allowed all hunters who expressed an interest the opportunity to participate. As a nonrandom sample, data gathered may not be representative of the average hunter; however, any biases associated with the dataset were likely consistent throughout the period of study. Participants reported 2 unique information sets. First, turkey observations, gobbling activity, and harvests were recorded from each individual hunt outing. Hunters specifically reported the hunt location (private or public land), date, hunt start and end time, county, number of unique males and total gobbles heard, number of turkeys seen (adult males, juvenile males, females, and unidentifiable turkeys), and qualitative information about harvested males. We asked survey participants to complete a questionnaire of their personal information and preferences. From this questionnaire, we developed metrics that were specific to each annual spring hunting season. We assigned each respondent to one of the MDWFP's 5 wild turkey management regions based on the county in which they hunted most frequently. We determined each participant's age and the number of years they previously hunted spring turkeys. We asked participants the degree to which they used turkey decoys while hunting (transformed into a binary variable based on whether decoys were frequently ‐ always or rarely ‐ never used). A categorical variable with 4 levels was created corresponding to responses about the hunting approach each participant reported. Categorical levels were (a) hunter actively covers large areas looking for turkeys, (b) hunter remains stationary preferring to wait for turkeys to reveal themselves, (c) hunter employs a combination of the 2, or (d) if turkeys are not quickly apparent in the hunting area, hunter concludes their hunt. We asked each hunter to gauge the hunting pressure on the areas they most frequently hunted. Responses were transformed into a categorical variable with 3 levels corresponding to (a) light pressure (mean of 1 – 2 hunter outings per ~200 ha of property/week), (b) moderate pressure (mean of 3 – 6 hunter outings per ~200 ha of property/week), heavy pressure ( ≥ 7 hunter outings per ~200 ha/week). A categorical variable with 5 levels was created to correspond to the weapon (10 ‐ , 12 ‐ , or 20 ‐ gauge shotgun, other shotgun, or archery) hunters primarily used. We asked hunters the effective range they felt comfortable shooting at turkeys, which was summarized into a categorical variable with 6 levels (Table 1). The within ‐ year CPUE values of individual hunters were our primary response variables and were calculated as each hunter's total harvest across all outings within a year standardized by 100 hours hunted. Likewise, we standardized total number of turkeys observed by each individual hunter, within years, per 100 hours hunted. We calculated a continuous variable for the within ‐ year percentage of hunts in which a hunter heard a male wild turkey. Hunters were asked whether they shot at and missed one or more turkeys within a hunting season. From their responses, we created a binary variable to describe whether a hunter shot at and missed a turkey within a hunting season (Table 1). The structure of our dataset reflected annual summations of individual hunter values from within a given hunting season; CPUE for each individual hunter was based on a standardized harvest per 100 hours within a year. 4 of 13 | BUTLER AND WANG 23285540, 2022, 2, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1268 by University Of Florida, Wiley Online Library on [10/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License T A B L E 1 Descriptive statistics of variables collected from 618 avid turkey hunters in Mississippi, USA to relate hunter demographics, behaviors, and tools to both catch ‐ per ‐ unit ‐ effort and the probability of missing a turkey within a spring season, 2015 – 2018. Variables Symbol Obs. Mean Std. dev. Min. value Max. value Dependent variables Turkey harvests per 100 hours hunted CPUE 1211 4.1 9.9 0 200 Binary variable for missing a turkey within a season MISS 1211 (1 = 251) 0 1 Random variable Unique survey participants HUNTER 618 Independent variables Hunter's age AGE 1211 48.8 14.8 16 75 Hunter's years of turkey hunting experience EXPERIENCE 1211 25.1 13.8 1 70 Hunter's approach to finding a tom: APPROACH (a) = 132 0 1 (a) Covers lots of ground. (b) = 234 (b) Remains stationary. (c) = 642 (c) Combination of a & b. (d) = 203 (d) Tries another day. Hunting pressure: PRESSURE (a) = 568 0 1 (a) Light. (b) = 503 (b) Moderate. (c) = 140 (a) (c) Heavy. Hunter's choice of weapon: WEAPON (a) = 37 0 1 (a) 10 gauge shotgun. (b) = 1075 (b) 12 gauge shotgun. (c) = 89 (c) 20 gauge shotgun. (d) = 3 (d) Other gauge shotgun. (e) = 7 (e) Archery. Hunter's maximum comfortable rage: RANGE (a) = 77 0 1 (a) <30 m. (b) = 235 (b) 30 – 35 m. (c) = 409 (c) 35 – 40 m. (d) = 293 (d) 40 – 45 m. (e) = 149 (e) 45 – 50 m. (f) = 48 (f) >50 m. (Continues) CONNECTING HUNT OUTCOMES TO DEMOGRAPHICS, BEHAVIOR, AND EXPERIENCES | 5 of 13 23285540, 2022, 2, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1268 by University Of Florida, Wiley Online Library on [10/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License We included continuous covariates in the dataset as within ‐ year means or summations. Binary or categorical covariates were year specific. Statistical analysis Over 50% of individual hunter CPUE values included zeros. Therefore, we used 2 ‐ stage, hurdle gamma mixed models to dually predict the influence of covariates on producing a successful hunt (a non ‐ zero value for CPUE), and for successful hunters, positive CPUE. Some individual hunters participated in the survey across multiple years, so we used hunter identity as a random effect to account for repeated observations of the same hunter in both the binary zero component ( hu ) and positive observation component of the hurdle gamma mixed models. Fixed effects on positive CPUE included survey year, region, hunter age, years of turkey hunting experience, decoy use, hunting approach, percentage of hunts in which a male turkey was heard, and total turkey observations per 100 hours hunted. We used generalized linear mixed models (GLMMs) with a Bernoulli distribution for the observed binary response of missing turkeys. As with CPUE models, hunter identity was treated as a random effect to account for repeated measures, and survey year, region, hunter age, years of turkey hunting experience, decoy use, weapon choice, and effective range were fixed effects. For both sets of models, we conducted a backward step selection of the full models, which included all predictors of fixed effects, using Akaike information criterion corrected for small sample size (AIC c ). We calculated model ‐ averaged coefficients and their 95% confidence intervals (CIs) for all variables included in the candidate set based on Akaike weight and Δ AIC c < 4.0. We used the R package glmmTMB to build the GLMMs in the R environment v.3.5.3 (Brooks et al. 2017, R Core Team 2019). Backward step selection and calculation of model ‐ averaged coefficients were done with the R package MuMln (Barton 2015). Model ‐ averaged 95% CIs and p ‐ values were the measure of parameter estimation. We tested for multicollinearity of all the covariates included in the full models using the R package performance (Lüdecke et al. 2021). Values of variance inflation factor (VIF) were less than 2.0, indicating low multicollinearity among covariates. R E S U L T S From 2015 – 2018, we documented the experiences of 618 unique turkey hunters across 20,755 trips afield, which yielded 1,211 seasonal data summaries (Table 1). Across all years, hunters harvested a mean of 1.4 (standard deviation [SD] = 1.6) males per season or 4.1 (SD = 9.9) per 100 hours hunted. Hunters averaged 48.8 (SD = 14.8) years of age with 25.1 (SD = 13.8) years of turkey hunting experience. About 34.6% of hunters always or frequently used turkey decoys while hunting. Slightly over half (53%) of hunters employed a combination of both actively covering large areas looking for turkeys and periodic waiting for turkeys to reveal themselves while hunting. T A B L E 1 (Continued) Variables Symbol Obs. Mean Std. dev. Min. value Max. value Mostly or always use a decoy? DECOY 1 = 420 0 1 Percentage of hunts in which a male turkey was heard %HEARD 1211 0.567 0.242 0 1 Total turkey observations per 100 hours hunted OBSERVATIONS 1211 81.0 142.3 0 4000 6 of 13 | BUTLER AND WANG 23285540, 2022, 2, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1268 by University Of Florida, Wiley Online Library on [10/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Most participants hunted on lightly (46.9%) or moderately (41.5%) pressured properties. Twelve ‐ gauge shotguns were the most ‐ used weapon (88.6%) for hunting turkeys, and 59.5% of hunters believed their effective range was ≤ 40 m. Hunters averaged 17.1 (SD = 12.1) outings per year, mean outing length was 3.1 (SD = 1.2) hours, and hunters heard 1.3 (SD = 0.9) male turkeys per outing. Hunters reported hearing at least one male turkey on 56.7% (SD = 24.2) of trips afield and observed 81.0 (SD = 142.3) total turkeys per 100 hours hunted. Thirty ‐ one percent of individual hunters missed a turkey at least once during the study and 20.7% of individual ‐ hunter seasonal summaries contained a miss (Table 1). Models of CPUE suggested the probability of harvesting at least one turkey (producing a non ‐ zero CPUE) increased with hunter age (Table 2). Likelihood of success increased approximately 2% for each year hunter age increased. In the prediction of positive CPUE values, the candidate model set ( Δ AIC c < 4.0) contained variables for survey year, hunter age, years of experience, decoy use, hunting pressure, hunting approach, the percentage of hunts in which a male turkey was heard, and total turkey observations per 100 hours hunted. The model ‐ averaged coefficients for both the percentage of hunts in which a male turkey was heard and total turkey observations per 100 hours hunted were positive and excluded zero (Figure 1, Table 2). Catch per unit effort increased throughout years of our study and was significantly reduced for hunters using moderately pressured properties. For models predicting the tendency to shoot at and miss turkeys, survey year, hunter age, years of experience, and decoy use were included in the candidate model set with Δ AIC c <4.0 (Table 3). Of these, the effect of hunter age was negative, suggesting the likelihood of missing declined with increasing age; however, the upper limit of the variable's 95% CIs encompassed zero. The model ‐ averaged covariates for experience and decoy were not T A B L E 2 Model mean predictors of top hurdle gamma mixed models ( Δ AIC c < 4.0) for catch ‐ per ‐ unit ‐ effort (CPUE), measured as turkey harvests per 100 hours hunted by avid spring turkey hunters in Mississippi, USA, 2015 – 2018. The binary logistic component ( Zi ) represents the probability of non ‐ zero CPUE. Gamma distribution was used for non ‐ zero, positive CPUE. Estimate Standard error LCL a UCL a Z ‐ value Pr(>|z |) Intercept 0.742 0.181 0.388 1.096 4.106 <0.001 AGE − 0.004 0.003 − 0.009 0.001 1.526 0.127 YEAR (2016) 0.093 0.073 − 0.050 0.235 1.274 0.202 YEAR (2017) 0.197 0.073 0.054 0.341 2.697 0.007 YEAR (2018) 0.230 0.072 0.089 0.371 3.194 0.001 PRESSURE (b) − 0.136 0.062 − 0.257 − 0.015 2.210 0.027 PRESSURE (c) − 0.096 0.103 − 0.297 0.105 0.936 0.350 %HEARD 1.035 0.139 0.763 1.306 7.460 <0.001 OBSERVATIONS 0.002 0.000 0.002 0.002 8.336 <0.001 DECOY 0.056 0.066 − 0.074 0.185 0.845 0.398 EXPERIENCE − 0.001 0.003 − 0.007 0.006 0.174 0.862 APPROACH (b) 0.031 0.109 − 0.184 0.245 0.282 0.778 APPROACH (c) − 0.109 0.092 − 0.289 0.071 1.184 0.236 APPROACH (d) − 0.078 0.112 − 0.297 0.142 0.691 0.490 Zi Intercept − 2.040 0.356 − 2.738 − 1.342 5.729 <0.001 Zi AGE 0.024 0.007 0.011 0.037 3.638 <0.001 a LCL and UCL correspond to lower and upper 95% confience limits, respectivley. CONNECTING HUNT OUTCOMES TO DEMOGRAPHICS, BEHAVIOR, AND EXPERIENCES | 7 of 13 23285540, 2022, 2, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1268 by University Of Florida, Wiley Online Library on [10/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License significant (Table 2). There was no evidence weapon choice or effective shooting range influenced the tendency to miss. D I S C U S S I O N In our study, use of 2 ‐ stage models revealed variables associated with both hunter characteristics and turkey abundance influenced CPUE of a subset of Mississippi turkey hunters. In contrast to our primary hypothesis, a key individual demographic (hunter age) influenced CPUE by exerting a significant positive effect on the probability of harvesting at least one turkey. Interestingly, hunter age was only weakly associated with subsequent differences in CPUE amongst successful hunters. For that group, variables that may be attributes of turkey population abundance, including both a standardized measure of the number of turkeys observed and the percentage of hunts in which a male turkey was heard, became the most significant drivers of CPUE. In opposition to our second hypothesis, we determined the likelihood of missing a turkey within a season decreased with increasing hunter age. Taken collectively, our results suggest hunter efficiency is nonrandom and may be influenced by variables that could be subject to mean temporal change. Increasing age has been shown to positively influence the hunting or foraging efficiency of several taxa (Wunderle 1991, Rutz et al. 2006), including within primitive human cultures (Walker et al. 2002, Gurven et al. 2006). Hunting proficiency may entail mastery of certain skill sets, thereby increasing efficiency with age, or may require a minimum, age ‐ linked competency to ensure consistent success. Age may also influence motivations for hunting, which could directly affect harvest frequency. In Tennessee, turkey hunters for whom consumptive ‐ oriented motivations were most important tended to be older than the mean (Watkins et al. 2018). In Missouri, as hunter age and experience increased over time, hunters became more selective in their harvests, possibly due to confidence gained through increased competence (Isabelle and Reitz 2015). Interestingly, although age was an important predictor of hunting success in our study, years of turkey hunting experience was not. Years of turkey hunting experience has been previously linked to success rates (Isabelle and Reitz 2015). However, it should be noted most Mississippi turkey hunters do not begin their overall hunting careers as turkey hunters (Duda et al. 2003); therefore, age may be a better surrogate for overall hunting acumen than years of experience specific to hunting turkey. Work elsewhere has also shown turkey hunters to have fewer turkey ‐ specific years of experience as compared to other forms of hunting (Swanson et al. 2005, Casalena et al. 2011), and overall F I G U R E 1 Predicted response of catch per unit effort (CPUE) to (A) the percentage of hunts in which a male turkey was heard and (B) total turkey observations per 100 hours hunted for turkey hunters in Mississippi with positive CPUE, 2015 – 2018. Shaded areas represent 95% confidence intervals. 8 of 13 | BUTLER AND WANG 23285540, 2022, 2, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1268 by University Of Florida, Wiley Online Library on [10/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License hunting knowledge may outweigh more explicit short ‐ term familiarity in the determination of hunting success (Asmyhr et al. 2012). Furthermore, age may subtly influence CPUE in ways beyond experience. Casalena et al. (2011) reported the proportion of turkey hunters in Pennsylvania who were retired increased from 1% in 1995 to 24% in 2008. Thirty ‐ four percent of respondents in our sample were at least 65 years old, a common age ‐ related marker of retirement potential. With fewer work obligations and greater flexibility, retired hunters may be more selective with their hunt outings, preferring to hunt days when weather conditions are more favorable for increased gobbling activity (Palumbo et al. 2019). Such considerations could translate into increased likelihood of eventual success. When only considering those hunters who successfully harvested at least one turkey, variables associated with the turkey populations became more relevant. The percentage of hunts in which male turkeys were heard and the standardized index of total turkey observations were both significant positive drivers of CPUE for successful hunters. Previous work in Mississippi has shown gobbling activity is linked to availability of adult males (Miller et al. 1997, Palumbo et al. 2014) and hunting success (Palmer et al. 1990). Likewise, standardized observations by hunters or others can track estimates of turkey abundance (Welsh and Kimmel 1990, McJunkin et al. 2005, Butler et al. 2015). Nonetheless, gobbling activity can be suppressed by hunting pressure (Lehman et al. 2007, Wightman et al. 2018) and influenced by weather (Palumbo et al. 2019), which may mask its relationship with abundance. Similarly, detectability differences must be considered when using observation data (Healy and Powell 2000). These potential sources of bias notwithstanding, our results link CPUE of successful hunters to factors naively assumed to be a function of turkey abundance (gobbling activity and total turkey sightings). Based on our results, hunters become less likely to miss turkeys as they age. To our knowledge, this relationship has not been previously documented, yet complements the positive relationship we documented between age and the likelihood of turkey hunting success. Social sciences have repeatedly demonstrated risk tolerance changes with age; specifically, older individuals take fewer risks within recreational pursuits (Rolison et al. 2013). In the context of turkey hunting, this tendency likely manifests by hunters becoming less willing to risk a negative outcome (e.g., missing or crippling a turkey) as they age. In turn, they may become more selective with their shots, thereby reducing misses, which would increase their overall hunting success by ensuring harvest opportunities are not squandered when presented. Hunters in our survey only averaged 1.4 harvests per season, meaning any miss could, on average, become the difference between an individual hunter finding success or not within a given year. The data we collected allowed for an estimation of CPUE across a range of values for covariates that served as naive measures of turkey abundance (i.e., the percentage of hunts in which a male turkey was heard and T A B L E 3 Mean effects of hunter age (AGE), years of turkey hunting experience (EXPERIENCE), and decoy use (DECOY) on the probability by avid turkey hunters in Mississippi, USA, to miss a tom within a single spring season, 2015 – 2018. The coefficient estimates were the model means of the candidate model set based on delta Akaike information criterion adjusted for small sample size (AIC c ) < 4.0. Estimate Standard error LCL a UCL a Z ‐ value Pr(>|z |) Intercept − 1.110 0.367 − 1.828 − 0.039 3.029 0.003 AGE − 0.013 0.007 − 0.025 0.000 1.933 0.053 YEAR (2016) 0.148 0.234 − 0.310 0.606 0.635 0.525 YEAR (2017) 0.166 0.238 − 0.301 0.633 0.696 0.487 YEAR (2018) − 0.137 0.236 − 0.599 0.325 0.582 0.561 DECOY − 0.224 0.181 − 0.578 0.130 1.240 0.215 EXPERIENCE 0.002 0.009 − 0.015 0.019 0.258 0.796 a LCL and UCL correspond to lower and upper 95% confience limits, respectively. CONNECTING HUNT OUTCOMES TO DEMOGRAPHICS, BEHAVIOR, AND EXPERIENCES | 9 of 13 23285540, 2022, 2, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1268 by University Of Florida, Wiley Online Library on [10/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License standardized total turkey observations). The relationship of CPUE to both these variables was not proportional. For instance, declines in the percentage of hunts in which a male turkey was heard did not translate into equitable declines in predicted CPUE. As fewer turkeys were encountered, CPUE flattened, suggesting the mean relative efficiency of turkey hunters may increase at lower levels of turkey abundance. Similar relationships have been demonstrated for other game birds (Guthery et al. 2004, Willebrand et al. 2011) due to behavioral aggregations by the harvested species (Erisman et al. 2011) or a tendency for participation by less skilled hunters to decline when game populations are low (Bowyer et al. 1999, Guthery et al. 2004). In either case, mean CPUE of the more avid hunters who remain may be bolstered. Spring, male ‐ only turkey hunting potentially fits both scenarios given male turkeys are likely to be clumped across the landscape during spring (Pollentier et al. 2019) and harvests tend to be concentrated amongst a small percentage of overall turkey hunters (Godwin et al. 1997) who may not necessarily be distributed in patterns corresponding to turkey abundance (Clawson et al. 2015, Stevens et al. 2020). Given male turkeys proclivity to vocalize, even in sparse, low ‐ density populations skilled hunters can target the limited individual male turkeys available, thereby reinforcing CPUE even as the overall availability of males drops. Our work offers 2 sources of concern for the use of CPUE as an index to turkey abundance. First, we demonstrated hunter age may influence hunt success. The hunting populace in North America, including turkey hunters, is aging (Casalena et al. 2011, Porter et al. 2011, Isabelle and Reitz 2015). If this general pattern affects mean success, as our results suggest, then turkey population declines could be concealed when unadjusted CPUE is the primary monitoring data employed. Furthermore, CPUE of successful hunters in our study did not appear to be linearly and proportionally related to other naive measures of abundance. Catch ‐ per ‐ unit ‐ effort tended to flatten as both the number of turkeys heard and observed declined, offering additional evidence of CPUE's potential shortcomings as an abundance index, particularly at low turkey densities. A limitation of our study was its reliance on a subset of avid hunters. Avid hunters, by their nature, are not representative of the average hunter in their region (Hurst et al. 1982), so the findings of our work may not hold across a wider stratum of individuals. Additional work is needed to determine if our results are more widely applicable, particularly as they pertain to more broad ‐ scale datasets. Furthermore, although models suggested naive indices of turkey abundance (gobbling and hunter observations) were more closely tied to successful hunter CPUE rather than mean hunter CPUE, our work did not directly test the relative performance of these measures against one another. We propose future studies should measure CPUE across a range of study areas in which male turkey density has been accurately estimated from more robust methodology. This would allow for an assessment of our results and potentially provide a means to calibrate CPUE based on known population densities. M A N A G E M E N T I M P L I C A TI O N S The lack of reliable methodologies to easily estimate turkey abundance both spatially and temporally will continue to hinder management of the species. As calls from constituents in response to alleged population declines intensify (Casalena et al. 2015), managers and policy makers will be increasingly pressed to identify means to accurately judge the legitimacy of public concerns and evaluate their own management actions. Our work indicates managers using CPUE as a surrogate to turkey population abundance should approach inferences from these datasets cautiously and with a clear understanding of their potential biases. Given their long ‐ term application by management agencies (Healy and Powell 2000), abandonment of CPUE is probably unlikely. Nonetheless, our findings cast enough doubt on their underlying assumptions that we recommend managers seek other supporting sources of information and not solely rely on CPUE as their only measure of turkey population trends. 10 of 13 | BUTLER AND WANG 23285540, 2022, 2, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1268 by University Of Florida, Wiley Online Library on [10/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License A C KN O W L E D G M E N T S Thanks to L. Taylor for ensuring survey data collection and entry ran smoothly. We are indebted to the legions of Mississippi turkey hunters who voluntarily collect observations for the Spring Gobbler Hunting Survey; their respect for the wild turkey is greatly admired. We thank C. Moorman (Guest Editor ‐ in ‐ Chief), B. Wakeling (Associate Editor), A. Knipps (Editorial Assistant), A. Tunstall (Copy Editor), J. Levengood (Content Editor) and 2 anonymous reviewers for their comments, which improved the manuscript. This manuscript is a contribution of the Mississippi State University Forest and Wildlife Research Center. ETHICS STATEME NT Human Subjects Protocol or Institutional Review Board approval are not required for the survey conducted in this research. REFERENCES Aebischer, N. J., C. J. Wheatley, and H. R. Rose. 2014. Factors associated with shooting accuracy and wounding rate of four managed wild deer species in the UK, based on anonymous field records from deer stalkers. PloS ONE 9(10):e109698. Allen, M. L., N. M. Roberts, and J. M. Bauder. 2020. Relationships of catch ‐ per ‐ unit ‐ effort metrics with abundance vary depending on sampling method and population trajectory. PloS ONE 15(5):e0233444. Arreguín ‐ Sánchez, F. 1996. Catchability: a key parameter for fish stock assessment. Reviews in Fish Biology and Fisheries 6: 221 – 242. Asmyhr, L., T. Willebrand, and