Southern Indiana bobwhites had higher estimated winter survival when they more strongly favored roosts close to shrubs and tree saplings over nearby random locations. At the highest level of preference the model put survival at 0.60 with a 95% credible interval (CrI) spanning 0.42 to 0.77 for 181 days between October and March. We take that as a reason to look over patches of shrubs and saplings on similar quail ground, although this study never tested where hunters locate coveys.
IWinter Roosts
Stronger Preference for Woody Cover Went With Better Estimated Survival
In Hill and colleagues' 2026 study in Oecologia, 0.60 comes from fitting the model at the far end of the selection range where the value reached -0.42. That figure does not count actual birds. It says 60 birds out of every 100 at that selection value would survive the stretch, with a broad interval. Survival averaged 0.31 and 0.40 for the full groups in the separate winters, but those group averages cannot be compared with a fitted estimate at the range's far end. The relationship was measured in one area, and the authors want experiments that alter habitat before it is treated as proof of cause.
IITracking Birds
Radio Tags Followed Bobwhites Through Two Winters at an Indiana Site
Goose Pond Fish and Wildlife Area in Greene County was the study site on southern Indiana public ground. The Indiana Department of Natural Resources managed and owned it while the research took place. Built wetlands occupied the middle, bordered by native warm-season grasses and fields under row-crop rotation. According to Hill and colleagues the quail hunt used quotas and a draw, with a unit closing early when hunters killed 30% of its estimated population. Current hunting rules were outside our check.
Most woody cover grew in narrow bands. Strips measured 1 to 10 meters in width beside drainages, fence rows, filter strips and field edges. Small patches also grew in grass left unburned for more than one year. Plants usually stood 1 to 3 meters high and included ash, black cherry, sweetgum and red maple.
Trapping took place during September and October in 2020 and 2021. Birds over 129 grams carried 6-gram necklace radio transmitters. Tracking covered 169 bird-winters split into 86 and 83, with 3 birds appearing in both years. Between October and February crews located each bird after dark from roughly 25 to 30 meters away one to three times per week. The paper lists 720 distinct roost locations divided into 351 and 369, plus 70 covey ID codes.
For every roost the model used 20 randomly picked locations no farther than 180 meters (about 590 feet) away, matching the average distance a bird moved between successive roosts. Five of the six measurements described distance to grassland, woody cover, wetland, woodland and row-crop fields. The sixth variable measured the proportion of the area classified as disturbed ground. This category included land burned the previous spring or summer, land left fallow the previous or current year, habitat-improvement areas and dove fields.
Bird survival was tracked between October 1 and March 31. Removing birds that hunters or vehicles killed left 132 birds divided into 63 and 69. Among those birds 82 died, and an estimated 52 deaths from avian predators made that the leading cause. An additional 16 birds had censored survival records.
IIICover Choices
Roosting Birds Favored Woody Cover Along With Grass and Disturbed Ground
Woody cover drew roosting birds. The model put probability of use at 0.56 at a distance of 0 meters from that cover (95% CrI of 0.54 to 0.58). Probability of use decreased by 0.19 (0.17 to 0.21) for each additional 250 meters (about 820 feet) from woody cover. Birds favored disturbed ground and grass as well, while showing only weak avoidance of wetland edges. Results for all the cover types appear in the table.
Neither woodland nor row-crop fields had a clear effect on where birds roosted in this study. The authors think plants and avoiding predators likely matter more in choosing a roost than food in surrounding fields. They also suggest danger from predators drives the preferences they found. Probability of use weighs a chosen location against 20 randomly picked spots. We read it as a comparison that does not tell you a bird's chance of sleeping at a given spot on a particular night.
Swipe table to see every column
| Cover variable | Result | Probability of use at reference point | Rate of change |
|---|---|---|---|
| Distance to woody cover | Selected, closer was higher | 0.56 (0.54 to 0.58) at 0 m | Fell 0.19 (0.17 to 0.21) per 250 m farther away |
| Distance to grassland | Selected, interior was higher | 0.63 (0.58 to 0.69) at 228 m inside | Fell 0.05 (0.03 to 0.07) per 250 m away from the interior |
| Proportion disturbed ground | Selected | 0.45 (0.44 to 0.46) at a proportion of 0 | Rose 0.06 (0.05 to 0.07) per 0.20 proportion |
| Distance to wetland | Weak avoidance | 0.45 (0.41 to 0.49) at 0 m | Rose 0.03 (0.01 to 0.06) per 250 m farther away |
| Distance to row-crop field | No clear effect, interval overlapped zero | Not reported | Not reported |
| Distance to woodland | No clear effect, 49% of the posterior overlapped zero | Not reported | Not reported |
IVSurvival Connection
Woody Cover Alone Had a Clear Relationship Between Selection and Survival
Each bird's average roost selection went into a second model that estimated daily survival. The resulting survival estimates are reported for the 181-day period. Only woody cover had a credible interval excluding zero among the landcover variables. On the logit scale its coefficient came to -0.27 (95% CrI from -0.54 to -0.02), with stronger preference for woody cover represented by a more negative selection value. The interval's upper end is just 0.02 under zero. The authors express minimal concern about their interpretation because selection and survival both showed strong signals. That paragraph also says uncertainty around individual selection values was left out of the model, so we would keep an eye on that -0.02 boundary.
No cover type in the study paired preferred roosts with poorer survival, leaving the authors without evidence of an ecological trap. Birds favored disturbed ground and grass without a clear connection to survival. The authors suggest suitable roost patches may be evenly scattered through those covers, which would make predation random there. Survival also showed no clear difference between winters, averaging 0.31 (95% CrI of 0.21 to 0.46) in the first and 0.40 (0.27 to 0.55) in the second.
Swipe table to see every column
| Covariate | Coefficient | 95% credible interval | Interval excludes zero |
|---|---|---|---|
| Woody cover selection | -0.27 | -0.54 to -0.02 | Yes |
| Woodland selection | -0.18 | -0.38 to 0.03 | No |
| Wetland selection | 0.03 | -0.25 to 0.30 | No |
| Row-crop selection | 0.12 | -0.12 to 0.37 | No |
| Grassland selection | -0.02 | -0.30 to 0.27 | No |
| Disturbed ground selection | -0.14 | -0.36 to 0.07 | No |
| Year (2021 to 2022 versus 2020 to 2021) | 0.21 | -0.38 to 0.80 | No |
VManaging Cover
The Authors Pointed to Gaps Between Woody Patches
The authors suggest that limited spacing of woody cover may limit the quality of winter habitat at similar sites, where woody cover is often narrow and linear. They suggest adding woody cover at regular intervals through usable habitat could improve survival, particularly where little exists. They also say the benefit probably hinges on how much cover there is and how it is arranged, neither of which this study tested.
We think winter scouting on public quail ground should include noting fence rows and patches of shrubs and saplings, because tagged birds favored roosts nearer that cover compared with nearby random locations. Whether hunters find more coveys by hunting near those features was not tested. For bobwhite counts over the long haul, read FDR's state-by-state June bobwhite count paper.
Our rule is to chart shrub and sapling patches on your quail ground in winter and mark long gaps between them. Think twice before clearing a patch by dormant-season burning, mowing or disking. Removing patches with those methods was never tested here, so that advice is our judgment.
VIResearch Scope
One Public Area and Two Winters With Cover Amounts Untested
The birds all lived on one public site followed through two winters, and the survival model included 132 birds. Narrow strips made up most woody cover at that site, so findings may change on other ground. This was observational work, and the authors say identifying cause requires experiments that change habitat.
Each bird's average selection went into the survival model without the uncertainty surrounding that estimate. Roost microsite traits were not measured by the authors. With hunter and vehicle deaths excluded, hunting mortality is beyond what the study can tell us. Neither the authors nor we identify a best acreage, spacing or shrub species. The authors examined birds roosting only during fall and winter, and they suggest looking at other stages of life such as raising broods.
NOTES AND SOURCES
- Source: Hill and colleagues 2026, Oecologia 208, open access, northern bobwhite roost site selection and survival during the non-breeding season.
- Scope: Radio-tagged northern bobwhites on Goose Pond Fish and Wildlife Area, Greene County, Indiana, trapped in 2020 and 2021 and followed from October through March. 169 bird-winters, 720 roost sites and 132 birds in the survival model.
- Method: Figures are as printed. Each roost was compared with 20 random points within 180 meters. Survival values are fitted model estimates over 181 days with 95% credible intervals. Conversions from meters to feet are ours.
- Limits: One public area, two winters, 132 birds in the survival model. Observational association, not an experiment.
- Limits: Survival model used each bird's mean selection and ignored its uncertainty. The woody cover interval ends at -0.02.
- Limits: Hunting and vehicle deaths were removed, so the study says nothing about hunting mortality.
- Limits: No acreage, spacing or shrub species was tested.
Fully Drawn Research is an independent data analysis desk, not affiliated with any state wildlife agency, transportation agency, mapping provider or outfitter. Fully Drawn Research is an independent data and analysis desk. This paper summarizes published research and adds FDR's reading of what it means for hunters. Figures are quoted from the studies linked in the text, and figures we computed from them are labeled as ours.
Keep reading
FDR Scout puts research like this on the map of the ground you hunt. Open Scout →See the plans →