Received: 1 November 2024 | Revised: 24 September 2025 | Accepted: 27 September 2025 DOI: 10.1002/wsb.1642 S P E C I A L S E C T I O N Decreased female survival may help explain wild turkey population decline Marcus A. Lashley 1 | M. Colter Chitwood 2 | Anna K. Moeller 2 | William D. Gulsby 3 | Alex D. Potash 1 | Kelly O'Neil 1 | Mark Turner 1 1 Department of Wildlife Ecology and Conservation, University of Florida, 1745 McCarty Drive, Gainesville, FL 32611, USA 2 Department of Natural Resource Ecology and Management, Oklahoma State University, 320 G Agricultural Hall, Stillwater, OK 74078, USA 3 College of Forestry, Wildlife and Environment, Auburn University, 602 Duncan Drive, Auburn, AL 36849, USA Correspondence Dr. Marcus A. Lashley, Department of Wildlife Ecology and Conservation, University of Florida, 1745 McCarty Drive, Gainesville, FL 32611, USA. Email: marcus.lashley@ufl.edu Present address Mark Turner, Department of Natural Resource Ecology and Management, Oklahoma State University, Stillwater, OK 74078, USA. Funding information National Wild Turkey Federation, Grant/Award Numbers: AWD15616, AWD17584 Abstract Recent declines in wild turkey ( Meleagris gallopavo ) populations have prompted extensive research efforts and adjustments to state hunting regulations across the range of wild turkeys. Research comparing historical and modern vital rates is needed to identify demographic factors that may explain declines. Based on previously published matrix models, adult female wild turkey survival has the greatest effect on the intrinsic rate of growth. Thus, we hypothesized that changes in annual female survival may explain contemporary declines in populations if it has decreased concomitant with population declines. We searched peer ‐ reviewed literature, student theses and disser- tations, and government reports for empirical estimates of annual female survival across the range of wild turkeys. We then used matrix models to evaluate population growth under observed survival rates. Of the 51 resulting estimates of annual female survival across all subspecies, most were ≥ 0.68 through 2004, but 15 out of 23 estimates post ‐ 2004 were lower, consistent with observed declining trends in population tra- jectories. We used matrix models to illustrate that the observed change in annual female survival estimates indeed predicts a declining population trajectory, even when other important vital rates are at the upper end of the published range of process variation. If published annual female wild Wildlife Society Bulletin 2025;49(S1):e1642. wileyonlinelibrary.com/journal/wsb | 1 of 16 https://doi.org/10.1002/wsb.1642 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2025 The Author(s). Wildlife Society Bulletin published by Wiley Periodicals LLC on behalf of The Wildlife Society. turkey survival estimates represent those experienced more broadly in populations, our meta ‐ analysis and matrix models indicate that decreases in survival may help explain declines. The causative factor(s) associated with the apparent decreases in female survival are unknown and little research has eval- uated management strategies to increase annual female sur- vival. Accordingly, we encourage wild turkey researchers to focus efforts on producing robust estimates of annual female survival, cause ‐ specific mortality, and rigorous evaluation of management strategies to improve annual female survival. K E Y W O R D S matrix model, Meleagris gallopavo , population dynamics, population trajectory, process variation Wild turkey ( Meleagris gallopavo ) populations have recently declined throughout their range (Casalena et al. 2015, Ericksen et al. 2015, Chamberlain et al. 2022). Londe et al. (2023) estimated a 9% annual population decline in the eastern subspecies ( M. g. silvestris ) based on a matrix model populated with vital rates published during 1970 – 2021. Declines have catalyzed a widespread increase in wild turkey research and prompted several state agencies to adjust wild turkey hunting seasons and test hypotheses to explain observed declines (Quehl et al. 2024). Although the causative mechanisms explaining population declines remain unclear, hypotheses include changes in predator context, habitat, climate, disease, and hunter harvest (Casalena et al. 2005). For wild turkeys, vital rates associated with the female segment of the population typically have the greatest influence on population trajectory. Accordingly, female ‐ only matrix models often include estimates of adult and yearling survival, nesting rate, nest survival, re ‐ nesting rate, clutch size, hatchability, and poult survival (Rumble et al. 2003, Pollentier et al. 2014, Londe et al. 2023). Several states in the conterminous United States have reported low poult ‐ per ‐ hen ratios, suggesting low productivity may be negatively influencing populations (Byrne et al. 2015). Although nest and poult survival can influence population growth rates (Roberts et al. 1995, Pollentier et al. 2014), adult female survival commonly has the greatest effect on the intrinsic rate of growth (Pollentier et al. 2014, Lehman et al. 2022, Londe et al. 2023). As such, population growth rates of wild turkeys are sensitive to even small changes in adult female survival (Wakeling 1991, Vangilder and Kurzejeski 1995, Alpizar ‐ Jara et al. 2001, Londe et al. 2023, Tyl et al. 2023). Even though annual female survival plays an important role in population growth rate, we are not aware of any attempts to determine if annual female survival has changed over time across the range of wild turkeys. Additionally, we are unaware of any attempts to determine whether decreases in annual female survival over time could explain observed declines in wild turkey populations. Thus, we performed a comprehensive review of the literature to compile all estimates of adult female annual survival, including peer ‐ reviewed publications, gov- ernment reports, and theses. We hypothesized that annual female survival has declined over time, particularly during the past 2 decades, mirroring the pattern observed in population declines. To explore how any observed changes in annual female wild turkey survival estimates might affect population trajectory, we also populated female ‐ only, stage ‐ based matrix projection models to estimate the asymptotic population growth rate ( λ ), pairing low, medium, and high annual female survival estimates with low, medium, and high reproduction rates detailed in Londe et al. (2023). 2 of 16 | LASHLEY ET AL METHO DS Literature review We used published data from the literature across the distribution of eastern, Rio Grande ( M. g. intermedia ), and Merriam's ( M. g. merriami ) wild turkey subspecies, including data from 27 states and Canadian provinces (Figure 1). We compiled literature in 4 ways. First, we downloaded the supplementary file provided by Londe et al. (2023) in their recent review of wild turkey vital rates from 1970 – 2021. Second, we conducted a literature search in Web of Science using the keywords “ meleagris ” and “ survival ” . Third, we downloaded all National Wild Turkey Symposia Proceedings through May 2024 and searched for manuscripts that may have included female survival estimates. Finally, in all studies including female survival estimates, we reviewed the literature cited for additional sources not found through other search methods. We screened over 2,000 manuscripts, theses, dissertations, and final reports for results related to annual or seasonal adult female survival. We only included known ‐ fate studies estimating female survival from very ‐ high ‐ frequency (VHF) or Global Positioning System (GPS) radiotags, given their increased certainty relative to banding studies (Wightman et al. 2024). Data analysis To assess trends in female wild turkey survival over time, we used a meta ‐ regression framework (Mengersen et al. 2013). To standardize data from studies with different reporting formats, we calculated one annual female survival F I G U R E 1 Location of studies that reported annual survival of female wild turkeys by region from 1978 – 2023. FEMALE WILD TURKEY SURVIVAL META ‐ ANALYSIS | 3 of 16 estimate for each independent study or population by pooling data across all years. As papers may have reported results for multiple wild turkey study areas, our total sample size was greater than the number of studies reviewed. We handled data extraction according to how survival estimates and sample sizes were presented. For studies that reported a single survival estimate for all study years (i.e., a pooled estimate) or had only one year of data, we recorded the provided estimate and sample size. When studies presented annual survival rates and sample sizes for each year, we extracted values for each individual year. In cases where annual survival rates were given but a total sample size for the entire study period was reported, we assumed the sample size was distributed evenly across all years. Due to varying approaches for reporting error across studies, we calculated the variance for pooled annual female survival estimates, which were proportions, as: p p n σ = (1 − ) 2 where σ 2 = variance, p = estimated annual female survival and n = sample size. For annual estimates derived from studies with multiple years, we applied the formula for each year p p n σ = (1 − ) i 2 i i i where i was the observation year. We then pooled annual estimates into a single survival estimate for each study or population using inverse ‐ variance weighting: x ˆ = ∑ ∑ pooled i=1 k x σ i=1 k 1 σ i i 2 i 2 Where x i = the survival estimate for year i and k was the number of years in the study. We calculated the associated pooled variance as: σ = 1 ∑ pooled 2 i=1 k 1 σ i 2 All survival estimates were assigned to the final year of the study. We also recorded the state or province where each study was conducted, which we collapsed into 4 mutually exclusive regions: North, Midwest, Southeast, and Southwest (Figure 1). We grouped studies by region based on landscape composition, climate, and data availability, primarily to explore whether any trends were consistent across regions. In the interest of being thorough, we performed identical models using a subspecies grouping instead of region (Figure S1, available in Supporting Information). We conducted a meta ‐ regression using annual survival as the response variable and Study Year and Region as predictor variables. We weighted all regression models using the inverse variance from each study, which we normalized by dividing by the mean inverse variance of all studies, giving greater weight to more precise study results, which is the strength of a meta ‐ analytical framework (Mengersen et al. 2013). Because annual survival estimates were proportions (i.e., between 0 and 1), we assumed that each observed value was drawn from a beta distribution (Douma and Weedon 2019). We used beta regression to model the expected value of survival as a function of predictor variables, accounting for the bounded nature of the data. We fit a suite of candidate models using the fixed effect Study Year and either Region or Subspecies. However, we could not include both Region and Subspecies in the same model as they were not independent of each other. We therefore used a 2 ‐ step model fitting procedure. First, we fit 2 models, one including Region as the only fixed effect and the other including only Subspecies. We compared the 2 models using Akaike Information Criterion for small sample sizes (AIC c ) and fit additional models using only the parameter from the model with the lower AIC c 4 of 16 | LASHLEY ET AL value. Recognizing that ecological patterns are often nonlinear, we modeled Study Year as a linear term as well as second ‐ and third ‐ order polynomials (Guthery and Bingham 2007). We included Region or Subspecies both as an additive fixed effect and an interaction term with each Study Year term. Additionally, we fit models with only Study Year or only Region or Subspecies as fixed effects, along with a null model containing no fixed effects. We ranked all models by AIC c and considered those with Δ AIC c < 2.0 as competitive models (Burnham et al. 2011). If we found a significant effect of a polynomial term on female survival, we identified the inflection point when annual female survival began to change by calculating the first derivative of our top model with respect to time, setting it equal to zero, and solving for time yielding the local maximum of the survival curve, indicating the inflection point after which survival began to decrease. We fit all models using package glmmTMB in Program R v4.4.1 (Brooks et al. 2017, R Core Team 2024). To further investigate female survival over time, we subset our data to only states with repeated measures of female survival at least 10 years apart. To the resulting subset, we fit 3 candidate regression models using Study Year as a linear term and as an additive second ‐ and third ‐ order polynomial. Each model included State as a random intercept effect and data points were inversely weighted by study variance. We compared models using AIC c to identify the most parsimonious fit for repeated measures of female survival. Matrix models After testing for a shift in adult female wild turkey survival over time, we quantified the effect of differential adult female survival on population growth. Our meta ‐ analysis revealed that female survival estimates reported in the literature began to decline around 2005. Thus, we created matrix models based on early (1978 – 2004) and late (2005 – 2023) estimates of female survival. We specifically used 2 matrix model approaches: deterministic and stochastic. We used the 2 ‐ stage (after ‐ second ‐ year [ASY], second ‐ year [SY]), female ‐ based, prebirth pulse, Lefkovitch matrix model from Londe et al. (2023) as the basis for both approaches. First, we used a determi- nistic matrix model to compare effects of survival while reproduction was held constant. We calculated asymptotic growth rate ( λ ) for every pairwise combination of low, medium, and high survival and reproduction (0.25, 0.5, and 0.75 quantiles, respectively). The survival quantiles were calculated from all published survival rates during each period (early and late) collected in this review, regardless of subspecies and without con- sideration of how previous authors separated or combined age classes of females. The reproduction quantiles (ASY: low = 0.236, medium = 0.328, high = 0.431; SY: low = 0.094, medium = 0.160, high = 0.239) were taken from the process distribution published by Londe et al. (2023), which incorporated published data from 1970 – 2021 for adults of the eastern subspecies. The reproduction equation from Londe et al. (2023; equation 2) included the following for first and second nests: incubation initiation, apparent nest success, clutch size, hatching rate, apparent poult survival to 28 days, and youth survival from 29 – 365 days. Second, we used a stochastic matrix model approach to assess population growth potential under the full process distributions of survival (by period; this study) and reproduction (from Londe et al. [2023]). For each of 10,000 iterations, we selected values of survival and reproduction from their process distributions, using normal distributions trun- cated at 0 and 1. Across the 10,000 iterations, we quantified the proportion of model runs that projected a stable or increasing population (i.e., λ ≥ 1). R E S U L T S We identified 46 publications that reported female survival, which resulted in 51 estimates of female survival, spanning the years 1978 – 2023 (Table S1, available in Supporting Information). Twenty ‐ five of these publications were included in the Londe et al. (2023) review. The remaining 21 were added via our literature review because FEMALE WILD TURKEY SURVIVAL META ‐ ANALYSIS | 5 of 16 they either were published after Londe et al. (2023) or excluded from Londe et al. (2023) because they had different subspecies or were published in reports, theses, or dissertations. Sixteen estimates (31%) came from the Midwest, 10 (20%) from the North, 14 (28%) from the Southeast, and 11 (22%) from the Southwest. Pooled estimates of annual female wild turkey survival ranged from 21.9 – 85.8%. The mean inverse ‐ variance weighted annual survival across all studies was 61.1% (SE = 0.6%). The AIC c value for the model using Region ( − 82.33) was lower than the AIC c value for the model fit with Subspecies ( − 78.22). Therefore, we fit all additional models using Region (but see Figure S2 in Supplemental Information for results grouping by subspecies). Our top ‐ ranked model indicated that the change in annual survival over time was nonlinear (Table 1). The top ‐ ranked model included Region as an additive fixed effect and showed that annual female survival was lower ( β = − 0.53, SE = 0.16, P < 0.001) in the North compared to the Midwest, which was our reference region (Table 2). Annual female survival rates in the Southeast and Southwest were similar to the Midwest ( P = 0.25, P = 0.36, respectively). The top ‐ ranked model also included Study Year ( β = − 1.12, SE = 0.45, P = 0.01), Study Year 2 ( β = − 0.53, SE = 0.49, P = 0.28), and Study Year 3 ( β = − 1.22, SE = 0.47, P = 0.009; Figure 2). We ran an identical model including only survival from adult T A B L E 1 Model selection table providing the number of parameters ( K ), log ‐ likelihood (logLik), Akaike's Information Criteria for small sample sizes (AIC c ), Δ AIC c , and model weight ( w i ) for beta regression models investigating the impacts of time and region and annual female wild turkey survival in North America. Models are ranked by AIC and only models with Δ AIC c < 2 were considered competitive models. Model K logLik AIC c Δ AIC c w i Survival ~ Year + Year 2 + Year 3 + Region 8 52.666 − 85.9 0 0.592 Survival ~ Year + Region 6 48.51 − 83.1 2.79 0.146 Survival ~ Region 5 47.156 − 83.0 2.93 0.137 Survival ~ Year + Year 2 + Region 7 49.195 − 81.8 4.12 0.075 Survival ~ 1 (Null model) 2 41.991 − 79.7 6.17 0.027 Survival ~ Year 3 42.394 − 78.3 7.63 0.013 Survival ~ Year * Region 9 49.583 − 76.8 9.13 0.006 Survival ~ Year * Region + Year 2 * Region 13 55.588 − 75.3 10.57 0.003 Survival ~ Year * Region + Year 2 * Region + Year 3 * Region 17 56.805 − 61.1 24.84 0 T A B L E 2 Parameter estimates, standard errors (SE), and P ‐ values for the top model describing changes in annual female wild turkey survival in North America from 1978 – 2023. Model Term Estimate SE P Intercept 0.53 0.08 <0.001 Study Year − 1.12 0.45 0.013 Study Year 2 − 0.53 0.49 0.283 Study Year 3 − 1.22 0.47 0.01 Region: North − 0.53 0.16 <0.001 Region: Southeast − 0.15 0.13 0.25 Region: Southwest 0.17 0.16 0.295 6 of 16 | LASHLEY ET AL females which yielded the same pattern (Figure S3, available in Supporting Information). Analysis of the first order derivative showed that female wild turkey survival began declining in 2005. Six states (Alabama, Arizona, Missis- sippi, South Dakota, Texas, Wisconsin, USA) and 1 Canadian province (Ontario, Canada) included multiple measures of female survival that were >10 years apart. Female survival declined between repeated measures in Alabama, Mississippi, South Dakota, Texas, and Ontario, remained stable in Arizona, and increased in Wisconsin (Figure 3). We selected a single model describing female survival over repeated measures, which included a linear effect of Study Year ( β = − 0.006, SE = 0.002, P = 0.014) on the log ‐ odds of survival, corresponding to an approximate 0.5% F I G U R E 2 Model results from meta ‐ regression showing estimated marginal means of annual female wild turkey survival from 1978 – 2023. The top figure illustrates the estimated marginal means of annual female survival across all studies, while the lower figures display the estimated marginal means of annual female survival across 4 different regions of North America. The dashed black line shows the inflection point when annual female survival began to decline. Each circle represents an individual study, and the circle size reflects the weight of each study in the meta ‐ regression, calculated as the standardized inverse variance. FEMALE WILD TURKEY SURVIVAL META ‐ ANALYSIS | 7 of 16 decline in the odds of female survival per year. Notably, regardless of how we grouped studies with repeated measures (based on subspecies, region, or state), annual female turkey survival always showed a declining trend over time, other than the Wisconsin time series. Matrix models Using the 0.25, 0.5, and 0.75 quantiles (low, medium, and high, respectively) of survival from each of the early and late periods for our deterministic matrix model approach, we determined that even survival from before 2005 (i.e., early period) was insufficient to support a stable or growing population (i.e., λ ≥ 1; Table 3, Figure 4). When holding survival at its median value for the early period, reproduction needed to be in the 81 st percentile to maintain a stable population (i.e., λ = 1), and 83 rd percentile for the late period to achieve a stable population. When comparing λ between the early and late periods (i.e., after 2004) for each combination of low and medium survival with low, medium, and high reproduction, λ decreased 1 – 12% (Figure 5). When comparing λ between the early and late periods for each combination of high survival with low, medium, and high reproduction, λ increased 0.5 – 0.6% (Figure 4), which was driven by greater variation in survival rates during the late period (i.e., standard deviation of F I G U R E 3 Model results from meta ‐ regression showing annual female wild turkey survival estimates from 1978 – 2023 for North American states and provinces with repeated temporal measures that were at least 10 years apart. Point size indicates weight of each study in the meta ‐ regression, which was calculated as the standardized inverse variance. 8 of 16 | LASHLEY ET AL 0.095 in the late period versus standard deviation of 0.037 in the early period). Using the stochastic matrix model approach, we determined that mean λ decreased from 0.90 to 0.83 between the early and late periods (Figure 5). Additionally, 85% of model runs in the early period showed a decreasing population ( λ < 1), and 92% of runs in the late period showed a decreasing population (Figure 5). T A B L E 3 Annual survival of female wild turkey by quantile (as well as mean and standard deviation [SD]) for 2 time periods (1978 – 2004 and 2005 – 2023) in North America. Period Quantile Annual Survival 1978 – 2004 0.25 0.605 0.5 0.614 0.75 0.627 Mean (SD) 0.624 (0.037) 2005 – 2023 0.25 0.516 0.5 0.604 0.75 0.632 Mean (SD) 0.560 (0.095) F I G U R E 4 Asymptotic growth rate ( λ ) for wild turkey from the deterministic matrix model under low, medium, and high survival rates (0.25, 0.5, and 0.75 quantiles, respectively) from the early period (1978 – 2004) and late period (2005 – 2023), matched with low, medium, and high reproduction rates (0.25, 0.5, and 0.75 quantiles, respectively) from 1970 – 2021 (from Londe et al. [2023]). Percentages shown on the bars are the percent decrease or percent increase in λ between the early (orange) and late (blue) periods for each respective pairing. FEMALE WILD TURKEY SURVIVAL META ‐ ANALYSIS | 9 of 16 D I S C U S S I O N Our results indicated that temporal trends in female wild turkey survival may help explain contemporary wild turkey population declines. The change in female survival over time is compelling and has implications for population performance. Adult female survival is the most important factor influencing wild turkey population growth rates (Londe et al. 2023). Small changes to female survival strongly influenced λ . Indeed, our matrix models confirmed that the reported changes in published annual female survival estimates would correspond with a declining pop- ulation growth rate, even when other important vital rates are at the upper end of their published range of the process distribution. Though wild turkeys are shorter lived than most ungulates, demographic research in relatively long ‐ lived species has shown that greater variation inherent in some vital rates (e.g., offspring survival) gives them a disproportionate effect on population trajectory (Gaillard et al. 1998, 2000; Raithel et al. 2007; Chitwood et al. 2015), especially when it comes to management. The natural variability in vital rates is the primary determinant of spatial and temporal variation in population growth (Raithel et al. 2007). Therefore, management actions should be more effective at moving a vital rate when it has more variation with which to work. Continued research into how management can improve wild turkey reproduction (the more variable rate) will be critical, but adult females must survive for reproduction to matter. Several aspects of our meta ‐ analysis give us confidence that the studies included represent state ‐ and regional ‐ level population trends. First, the declines in survival we reported over time generally matched trends in population proxies in time and space (Casalena et al. 2015, Ericksen et al. 2015, Chamberlain et al. 2022). For example, Casalena et al. (2015) found that wild turkey male harvest decreased by 25% in Mid ‐ Atlantic states since the early 2000s which they believe was associated with a population decline. Male harvest is an imperfect measure of abundance because of changes in hunter numbers, but among states that consistently reported wild turkey abundance, there was an estimated 3% decline in population size between 2014 and 2019 (Chamberlain et al. 2022). Moreover, hunter numbers have not consistently declined in most states indicating declines in harvest likely F I G U R E 5 Estimated asymptotic growth rate ( λ ) for wild turkey using survival from 2 time periods: 1978 – 2004 and 2005 – 2023. Dashed lines represent mean λ from 10,000 iterations of the stochastic matrix model (0.90 for the early period and 0.83 for the late period); dotted lines represent 95% confidence intervals. Shaded gray areas indicate values representing a declining population ( λ < 1), and unshaded areas indicate values representing an increasing population ( λ > 1). In the early period, 85% of model runs indicated a declining population, and in the late period, 92% of model runs indicated a declining population. 10 of 16 | LASHLEY ET AL reflect population trends more so than hunter effort (Steele and Lashley 2025). Second, in most cases where multiple reports from the same state were available, survival was greater in the earlier study (Figure 5). We only found one instance where an historical and modern female survival estimate was reported from the same study site. In that South Dakota population, annual adult female survival decreased from 0.67 during 1999 – 2001 (Shields and Flake 2006) to 0.49 during 2017 – 2019 (Tyl et al. 2023). Thus, we believe the negative temporal trend in adult female survival we reported is likely representative of real demographic change over time. Historically, managers have influenced female wild turkey survival through adjusting harvest regulations such as bag limits and season length (Vangilder and Kurzejeski 1995). Conservative fall harvest rates of ≤ 10% have been recommended to maintain population growth; however, these harvest rates were based on historical female sur- vival rates ≥ 56% (Vangilder and Kurzejeski 1995, Alpizar ‐ Jara et al. 2001). In recognition of the role of hunter harvest in female survival, combined with declining wild turkey populations, many states currently allow limited or no female harvest. Thus, most managers lack the flexibility to increase female survival through regulatory mech- anisms. Given the declines in female survival in the literature, we agree with Londe et al. (2023) that female harvests may not be sustainable in areas where wild turkey populations are declining. Mechanisms to explain recent declines in female wild turkey survival are unknown, but we believe several factors are plausible. Negative density dependence may contribute to population declines, as suggested by Byrne et al. (2015) who identified possible density dependence in southeastern wild turkey populations, though they did not link it to female survival. However, Byrne et al. (2015) included only 4 estimates from studies after 2005 (the year our analysis indicated that female survival began declining), so it appears the negative trend in female survival became more evident after most of their time series ended. Food availability could influence adult survival. There have been well ‐ documented declines in both insect and insectivorous bird populations; the spatial and temporal extent of insect declines align with declines in wild turkey populations (Van Klink et al. 2020, Wagner 2020, Tallamy and Shriver 2021). Furthermore, acorns are an important fall and winter food resource for wild turkeys (Meanley 1956, Hurst 1992). Overstory oaks ( Quercus spp.) have been declining in many areas (McShea et al. 2007, Haavik et al. 2015, Alexander et al. 2021). It also is possible that baiting and supplemental feeding, which only recently became widespread practice in many states, could negatively affect female survival (Arkansas Game and Fish Commission 2019, South Carolina Department of Natural Resources 2020, Huang et al. 2022). Feeding may expose females to greater predation risk during nesting, while also posing a health threat by increasing disease and toxin exposure (Cooper and Ginnett 2000; Huang et al. 2022 a , b ). There has also been increased concern over several diseases that could negatively affect female survival (MacDonald et al. 2022); however, little information is available to understand how exposure to diseases has changed over space and time. Changes in predation risk associated with changes in predator abundance or communities is also a plausible mechanism for the trend we reported. Female survival is generally lowest during the spring reproductive period (e.g., Vander Haegen et al. 1988, Yarnall et al. 2020, Tyl et al. 2023). Multiple studies point to predators as the primary source of cause ‐ specific mortality (Kurzejeski et al. 1987, Miller et al. 1998, Hubbard et al. 1999, Nguyen et al. 2003, Moore et al. 2010). There are examples of a changing wild turkey predator context across our study period. For example, coyotes ( Canis latrans ) are a common predator of female wild turkeys (Miller et al. 1995, Delahunt 2011, Pollentier et al. 2014). The timeline of coyote range expansion into the eastern United States coincides with wild turkey population declines in some areas (Hody and Kays 2018). Moreover, coyote abundance was negatively correlated with wild turkey abundance in Mississippi (Wang et al. 2023). By contrast, coyote range expansion seems insufficient to explain wild turkey survival decreases rangewide, as recent changes to female survival also have occurred within the historical range of coyotes (e.g., Shields and Flake 2006, Tyl et al. 2023). Based on recent reductions in landscape coverage of herbaceous plant communities, along with the nesting cover they provide (Keyser et al. 2019, Crawford et al. 2021, Bernath ‐ Plaisted et al. 2023), another hypothesis is that predation on adult female wild turkeys may be increasing in response to recent reductions in landscape coverage of nesting and brooding cover. Indeed, occupancy probability for mammalian turkey predators was FEMALE WILD TURKEY SURVIVAL META ‐ ANALYSIS | 11 of 16 decreased in burned areas as compared to non ‐ burned areas in a recent study (Boone et al. 2024). Climate also may influence survival, as increased precipitation can increase female and nest predation, perhaps because mammalian predators are more effective at locating nesting females via olfaction during or immediately following rain events (Roberts et al. 1995, Lehman et al. 2008, Webb et al. 2012, Yarnall et al. 2020; but see Boone et al. 2023). Broad scale changes in climate and land use could drive patterns in demographic rates given they are associated with broad scale patterns in relative abundance (Sibiya et al. 2025). Overall, it is likely that several of the proposed mechanisms outlined above have interactive or additive effects that contributed to the temporal decline in annual adult female wild turkey survival we reported. Specifically, reduced food availability and disease could decrease female body condition, survival, and ability to escape predators, while changes in baiting and feeding practices, habitat, and climate have simultaneously increased predation risk. Although other mechanisms may play a role, we suggest these hypotheses are worth exploring as mechanisms to explain declines in female survival. One potential limitation of our meta ‐ analysis approach is that the studied populations may not be repre- sentative of the broader trends in unstudied populations. For example, much of the research conducted on wild turkeys in the past was associated with recently restocked populations that were stable or increasing (e.g., Vander Haegen et al. 1988, Roberts et al. 1995, Shields and Flake 2006). Whereas, more contemporary work may have focused on declining populations or those perceived as declining by biologists (Zenas 2018, Lehman et al. 2022, De Filippo 2024). Methods have also been inconsistent, as several studies only provided survival estimates for a combination of adult and juvenile females (Vangilder and Kurzejeski 1995, Lehman et al. 2000, Nguyen et al. 2003). More recent studies commonly separate analyses by age class (Pollentier et al. 2014, Little et al. 2016, Tyl et al. 2023). We note studies included in our analyses reported no difference in adult and juvenile female survival (Hennen and Lutz 2000, Kane et al. 2007, Delahunt 2011, Yarnall et al. 2020, Tyl et al. 2023), but at least one study did (Vander Haegen et al. 1988). We also conducted analyses excluding studies that combined age groups; the resulting patterns over time mirrored the analysis of the full dataset (Figure S3, available in Supporting Information). Changes in technology also may have influenced our results, as earlier studies primarily used VHF transmitters, whereas several modern studies used both GPS and VHF transmitters (Niedzielski and Bowman 2015, Zenas 2018, De Filippo 2024). Nonetheless, VHF and GPS data should provide similar survival estimates, especially given that studies employing GPS transmitters often rely on a VHF beacon for mortality checks (Latham et al. 2015, Niedzielski and Bowman 2015). Moreover, we excluded banding studies in favor of known ‐ fate radiotag studies as a pre- cautionary measure to avoid biases associated with monitoring methods. Previous research and management to increase wild turkey populations focused on improving productivity, but our results indicate focusing efforts on practices that increase female survival may be just as important. Pollentier et al. (2014) suggested that continued restoration and enhancement of high ‐ quality habitat with early successional vegetation has the greatest potential to positively influence population growth by improving nest and poult survival. Such improvements to nesting and brood ‐ rearing areas also could improve female survival by reducing mortality risks while incubating eggs and raising poults, as greater female predation rates may occur during nesting if cover is limited (Hohensee 1999, Pollentier et al. 2014). Similar improvements also could reduce the risk of females being killed by farm equipment in regions where the best available nesting cover is located in fields likely to be hayed or harvested during nesting (Tyl et al. 2023). Similar habitat manipulations were suggested for Merriam's turkeys in the West, where vegetation treatments directed at improving body condition of females through improved abundance and distribution of food would likely increase nesting rates (Wakeling 1991, Wakeling and Rodgers 1995, Rumble et al. 2003). However, there is limited information on whether vegetation influences female survival, especially during incubation. Lohr et al. (2020) reported no effect of vegetation height on survival of incubating females, but multiple vegetation covariates such as density and visual obstruction may interact to influence survival (Fuller et al. 2013). Even if vegetation structure and composition has limited influence on nest survival, it could have a large impact on population growth if vegetation conditions reduce female predation rates during nesting (Keever et al. 2023). Importantly, there have been no manipulative studies to determine the effect size of any management strategies on female survival, other than those associated with hunter harvests. 12 of 16 | LASHLEY ET AL CONSER VATION I MPLICATIONS Our results suggest a couple of opportunities for future research and management of wild turkey populations. First, any management actions, harvest ‐ related or otherwise, that improve female survival will have positive effects on population performance. Restricting or removing harvest of females (bearded or not) from spring and fall seasons is an easy first step in areas where female harvest is currently legal and declines are apparent. Second, mechanistic understanding about how management can move the needle on reproductive vital rates is critical. Upper ‐ end reproductive values can offset reduced female survival but only if those values are sustained. Based on current data and anecdotal reports across the country, periodic booms in reproduction will not be sufficient to stall reported declines. Therefore, we recommend renewed focus on the landscape variables most likely to affect female nutri- tional condition, nest success, and poult survival. Consideration for conditions promoting female survival during nesting are of particular interest, as these conditions may simultaneously increase multiple vital rates. Where possible, experimental design should be prioritized to help isolate variables affecting wild turkey vital rates. Here, we recommend the wild turkey research community to develop rigorous, manipulative studies and take advantage of natural experiments, when possible, to isolate causative factors linked to female survival and reproductive success and to evaluate the effectiveness of corrective management actions. CO NFL I CT OF INTERES T S T ATEME NT The authors declare no conflicts of interest. ETHICS STATEME NT This research did not involve the use of animals since it was a metanalysis of existing published studies. Thus, no oversight committees or permits were secured for the purpose of the work. Please see cited original works for information on their specific permits worked under. D A TA A V A I L A B I L I T Y S T A T E M E N T All data used in the meta ‐ analysis herein are provided and cited for their location within the table in the manuscript. ORCID Marcus A. Lashley https://orcid.org/0000-0002-1086-7754 M. Colter Chitwood https://orcid.org/0000-0001-7240-7430 William D. Gulsby https://orcid.org/0000-0001-8327-2391 Alex D. Potash https://orcid.org/0000-0002-6343-9390 Mark Turner https://orcid.org/0000-0003-1990-7422 REFERENCES Alexander, H. D., Siegert, C., Brewer, J. S., Kreye, J., Lashley, M. A., McDaniel, J. K., A. K. Paulson, H. J. Renninger, and J. M. Varner. 2021. Mesophication of oak landscapes: evidence, knowledge gaps, and future research. BioScience 71: 531 – 542. Alpizar-Jara, R., E. N. Brooks, K. H. Pollock, D. E. Steffen, J. C. Pack, and G. W. Norman. 2001. An eastern wild turkey population dynamics model for Virginia and West Virginia. Journal of Wildlife Management 65:415 – 424. Arkansas Game