The theory of adaptive habitat selection suggests resource selection by animals should reflect underlying quality, such that individual selection confers an adaptive advantage via increased fitness. Using resource selection functions and nest survival models, we demonstrated that visual obstruction at the nest site was adaptively significant but timing of nest initiation had the greatest effect on nest survival for eastern wild turkeys (Meleagris gallopavo silvestris). Predation risk is a selective pressure, and if individuals can perceive predation risk, they may respond by altering selection of nest site characteristics based on prior experience. We evaluated patterns in nest site selection of 387 wild turkeys and consequences of selection on reproductive success across the southeastern United States from 2014-2019. We monitored 549 nest sites and found that nest initiation date had the strongest effect on daily nest survival rates, wherein adult females at our earliest nest initiation date were ~ 4 times more likely to successfully nest than females at our latest nest initiation dates. Selection of nest sites with greater visual obstruction also increased daily nest survival rates as females were 1.17 (1.100 – 1.234; 95% CI) and 1.37 (1.258 – 1.486; 95% CI) times more likely to select sites for every 10 cm increase in visual obstruction and maximum vegetation height, respectively. Collectively, our results indicate that nest initiation date is likely the critical parameter driving wild turkey nest success, whereas vegetative conditions play a lesser role influencing nest success. Females nesting earlier may be in better body condition and show increased nest attentiveness, which may mediate nest success more than vegetation conditions around nest sites. Our work indicates that increasing the reproductive success of wild turkeys may hinge on females being able to nest as early as possible within the reproductive season.

We captured female turkeys using rocket nets from January–March 2014–2019. We radio-tagged each female with a backpack-style, mortality-sensitive GPS transmitter with VHF capabilities (Biotrack Ltd., Wareham, Dorset, UK; Guthrie et al. 2011). We programmed transmitters to record hourly locations from 0500-2000 and one nightly location at 2359 for the life of the unit or until the unit was recovered (Cohen et al. 2018). Turkey capture, handling, and marking procedures were approved by the Institutional Animal Care and Use Committee at the University of Georgia (Protocol No. A2014 06008Y1A0, A343701, and A2016 04-001-R1) and the Louisiana State University Agricultural Center (Protocol No. A2014-013, A2015-07, and A2018-13). We downloaded GPS locations from each female ≥ 1 time per week.

We conducted vegetation surveys at each nest site at expected date of hatch regardless of nest fate (McConnell et al. 2017) and measured vegetation conditions immediately surrounding the nest (Yeldell et al. 2017a, Wood et al. 2019). We used conditional logistic regression in a matched-pairs case-control design (Compton et al. 2002, Manly et al. 2002) to estimate a resource selection function and explain patterns in nest site selection by female turkeys. Cases were the initiated nest site locations of individual female turkeys, and controls were random sites paired with the nest site locations. Paired-random sites were located 100-500 m away ( ) at a random direction from the nest site, and we sampled vegetation using identical methods described above for nest site locations. We used logistic exposure models in a Bayesian framework (Schmidt et al. 2010) to examine the effect of covariates influencing nest site selection on daily nest survival rates. We fit models using Markov Chain Monte Carlo (MCMC) methods implemented in JAGS (Plummer 2003) via program R (v4.0.3, R Core Team 2020) and the R2jags package (Su and Yajima 2020). We used uninformative priors for all parameters in the model. We ran 3 independent Markov chains for 100,000 iterations each, discarded the first 10,000 iterations as a burn-in period, and used a thinning rate of 5.

The csv data files can be opened in Microsoft Excel or similar program that reads csv files.