Northern bobwhites have experienced population declines in Colorado and range-wide. Estimating vital rates can provide clues to factors limiting population growth rate. Although recent estimates of breeding season survival in the northwest corner of the northern bobwhite range are available, there have been no recent studies on nonbreeding season survival. We used radio-telemetry to estimate nonbreeding season (Oct–Mar) survival of northern bobwhites at two study sites in northeastern Colorado during winter 2019–2020 and 2020–2021. Based on our sample of 157 bobwhites, we found that survival was highly variable between years and was negatively affected by colder daily minimum temperatures and deeper snow depths. Seasonal (6-month) survival during the first year was 0.219 (SE = 0.040) and during the second year was 0.006 (SE = 0.005). We found no evidence that sex, age, or study site influenced survival, and very weak support for an effect of body mass. During our study, there were two extreme winter weather events, during which we found unusually high numbers of non-predation mortality. Overall, northern bobwhite nonbreeding season survival in the northwest corner of their range appears to be generally similar to other regions, except during extreme winter weather events, which resulted in high mortality. We encourage managers to create or maintain vegetation characteristics that will provide shelter from winter weather while also providing abundant food in close proximity.

Capture and marking

We trapped bobwhites using baited walk-in traps (Stoddard 1931, Smith et al. 1981, Behney et al. 2020) from 6 September until 17 October 2019 at 288 trap locations, and 1 September to 23 October 2020 at 493 trap locations on Tamarack SWA and Dune Ridge SWA. Trap locations were spread throughout each property to achieve representation of all areas of the properties. Traps were moved if areas became saturated with radio-marked bobwhites or if no bobwhites were being caught. During the second season, we also used targeted night-lighting, where we tracked previously radio-marked birds at night to locate coveys and used spotlights and a hoop net to capture untagged birds.

Bobwhites were affixed with ≤ 6.5-g necklace-style VHF radio transmitters equipped with an 8-hour mortality sensor (American Wildlife Enterprises, Monticello, FL), which have been used frequently for bobwhites (Burger et al. 1995, DeMaso et al. 1997, Taylor et al. 1999). We did not deploy transmitters on quail weighing less than 130 g to keep the transmitter mass less than 5% of the bird’s body mass. These transmitters have been reported to not affect survival on bobwhites when making up <5% of individuals’ mass (Palmer and Wellendorf 2007, Terhune et al. 2007, Wann et al. 2020). We captured 25 birds in year one and three birds in year two that were deemed lighter than the allowable weight to be fitted with transmitters. All captured bobwhites received a numbered aluminum leg band (National Band & Tag Company, Newport, Kentucky, USA) and were weighed using a Pesola 300 g spring scale. Sex and age (subadult: hatched previous breeding season, adult: hatched prior to previous breeding season) were determined based on plumage. After processing, birds were released at the site of capture. All trapping, handling, and marking procedures were consistent with the guidelines of the University of Nebraska, Institutional Animal Care and Use Committee (Project ID #1844).

Survival monitoring

We attempted to assess status (live or dead) of all radio-marked bobwhites four or five times each week from October – March. When the transmitter’s mortality sensor indicated a dead bird, we retrieved the transmitter and assessed cause of mortality (i.e., mammal, avian, hunter cripple, etc.). In the event a bird was no longer able to be detected, we extensively searched the site and continued to monitor for it until the end of the season.

Statistical analysis

Deaths determined to be caused by the radio-collar (n=1) were excluded from analysis. We defined the nonbreeding season as a 6-month period (26 weeks) from 1 October to 31 March based on our observations during previous research of when nesting behaviors began and ended. We limited our analysis to birds alive at the start of the nonbreeding season (1 October) and censored birds after they went missing from the site or transmitters were believed to be no longer functional. We used the RMark package (Laake 2013) in R (R Core Team 2023) to construct known fate models for program MARK (White and Burnham 1999) to estimate and assess variation in weekly survival. We included year, study site, age, sex, body mass at capture, weekly mean of daily minimum temperatures, and weekly mean snow depth as predictor variables. We used Akaike’s Information Criterion corrected for sample size (AICc) to rank the models (Burnham and Anderson 2002). We began by comparing single-variable models including individual-specific predictors (age, sex, year, site, and mass). We then combined variables that appeared in a model that performed better than the null model (lower AIC_c) with weather variables (minimum temperature and snow depth). Minimum temperature and snow depth were correlated (r = -0.6), so we did not include them in the same model. The two winters during our study had very different weather conditions, so we included year (1: 2019–2020 and 2: 2020–2021) in interactive relationships with the weather variables. We also included mass in interactions with weather variables because lighter individuals may be more sensitive to harsh weather than heavier individuals. Variables appearing in interactions were also included as main effects.