Where Do Deer Go Under Pressure? Shifting Ranges, Protected Cover, and Places Hunters Skip

Does pressure send deer elsewhere? We cannot settle that question with our new analysis of South Carolina collar records. Locations beyond the property boundary were excluded. What remains describes shifts among deer that stayed: from pre-rut to post-rut, range centers changed by a median 90 meters for does and 685 for bucks. Legally protected refuges raise a different issue. The two studies we located reached different conclusions about benefits to surrounding hunters.

Locations underlying the 2022 research by Stewart and colleagues at Brosnan Forest are available in a public research archive. Every 30 minutes, collars recorded a location, starting 23 August and continuing to 23 November. Each location carries a period label: pre-rut begins 23 August and ends 18 September; rut begins 19 September and ends 28 October; post-rut begins 29 October and ends 23 November. The property's hunting season opened 15 September and closed 1 January. That places 23 pre-rut days ahead of opening day and 4 afterward, while the entire post-rut period falls during hunting season. Hunting effort was low, at approximately 5 hours per 100 hectares per week.

The archive contains 70 animal-seasons, with each record tracking a deer through a single fall: 43 does plus 27 bucks across eight seasons, spanning 2009 to 2018. We outlined a 95% kernel range separately for each animal and period. This smoothed boundary contains 95% of estimated use calculated from locations recorded during that period. We calculated the distance separating the pre-rut and post-rut centers, then the proportion of pre-rut area contained within post-rut area. To establish a comparison for small changes, we repeated those calculations after randomly dividing each animal's pre-rut locations into two halves.

Doe pre-rut ranges had a median 91% overlap with post-rut ranges, nearly the 92% overlap produced by randomly dividing a single period. At least half the earlier area remained inside the later range for 42 of 43 does. Median displacement between centers reached 90 meters; random splits yielded 17 meters.

Bucks had median overlap of 52%, compared with 92% after random splitting. Their median center displacement measured 685 meters, or roughly 0.4 mile, with a bootstrap 95% interval of 430 to 811 meters. Seventeen of 27 bucks shifted farther than 500 meters. Only 2017 and 2018 had data for both sexes; in those seasons, range centers shifted 606 meters for 8 bucks and 115 meters for 9 does. Median center displacement measured 460 meters between pre-rut and rut, followed by 335 meters between rut and post-rut. Each value is a separate median, so combining them does not produce the 685-meter shift.

Interpret these measurements as the distance a range shifts over one fall on hunted ground; do not attribute those shifts to hunting. Both rut activity and opening day occur between the comparison periods. There is no unhunted comparison area, and collars were largely deployed on does and bucks in different years.

The guide to what collar studies show hunting pressure does reviews research from Oklahoma, Mississippi, and Georgia. Only results concerning ranges and refuges appear in the table.

Both collar papers describe deer remaining within ranges they already occupied. Marantz and coauthors reported continued site fidelity during hunting, while Karns and coauthors reported that observed disturbances did not force deer beyond their home-range boundaries. The Pennsylvania report documented deer survival and hunter locations; its GPS collars did not work.

Lonnie Hansen and Charles Nixon described tagging research in a summary for the Illinois wildlife agency. In Piatt County, Illinois, researchers followed a 1,500-acre protected area over 5 years. The other project examined a large area of north-central Missouri containing two refuges over 6.

Forest covered 36% of the immediate Illinois study area, compared with just 2% across the county. Deer were more numerous within the refuge. Without animals arriving from protected ground, surrounding deer numbers dropped close to zero annually. Dispersal out of the refuge increased the nearby herd. Computer modeling using those findings indicated that, at the hunting pressure then in effect, keeping the herd outside the refuge stable would require the refuge to contribute about 40% of the female and buck fawn population. The account leaves unclear whether 40% applies separately to each group or jointly to both.

Forest occupied about 18% of the Missouri site. Deer numbers grew both within and beyond the refuges, with slower growth outside. Across that entire study area, merely reducing the rate of population growth would have required hunters to increase harvest by about 20%.

The authors offer this rough guideline: in areas with sparse, scattered woodland occupying less than about 5% of the ground, a refuge seems to benefit hunter success outside its boundaries. With more forest and fewer hunters clustered together, refuges frequently complicate herd management. The summary leaves unexplained how its 5% threshold applies to the Illinois study area: forest occupied 36% locally, within a county whose forest share was 2%. Allowing hunting in a refuge can leave neighboring hunters without the deer that previously dispersed onto their land.

These percentages come from the authors' account of research in two states. We did not locate the original study papers.

In Pennsylvania, some ground remained lightly hunted despite having no legal protection. This report does not document deer dispersing into surrounding hunted land. Instead, it records hunter locations and collared-deer survival. Keenan, Rosenberry and Wallingford tracked does wearing collars and plotted hunter locations on a pair of state forests.

On Sproul's public land, 73% of hunters confined their hunting to ground within 600 meters of a road, which accounts for 60% of the land. Slopes below 8 degrees accounted for 57% of the land and were where 70% of hunters hunted. Much of Sproul recorded hunting mortality below 2%; the authors suggested these places could function as refugia in practice. Yearly survival on Sproul's public acreage reached 90%, compared with 72% on private acreage. For deer occupying public slopes of 10-degree steepness, the report's model estimated hunting mortality at 6.4% beside roads and 3.2% at 1,000 meters away.

The pattern differed at Tuscarora. Its public land had a higher hunter density, with relatively low hunter numbers in just a few small areas. Deer survival was 60% on public land compared with 79% on private land. For Sproul, the report reasoned that extending seasons and issuing additional antlerless permits might accomplish little where refugia persist. It described expanded hunter access to steep terrain and places distant from roads as the sole remedy.

A property's distance to roads will not always reveal a refuge. Brosnan Forest's roads totaled 153 kilometers across 2,593 hectares. The nearest road was a median 55 meters from a collar location, and 98.9% of locations fell no farther than 200 meters away. Median distance during daylight was 63 meters, compared with 50 at night. The day-night difference measured 17 meters ahead of 15 September and 12 afterward, showing no increase when hunting opened.

Each outline used a 95% fixed-kernel estimate and enclosed 95% of estimated animal use. We applied a reference bandwidth and a grid spaced at 25-meter intervals. Overlap measures how much pre-rut area is contained within post-rut area. Using a minimum convex polygon drawn around the locations produces the same overall result: does showed 91% overlap and bucks 46%. The file includes no more than 400 sampled locations for each animal in each period, excluding any recorded beyond the study boundary. That restriction confines these measurements to shifts within the property, prevents detection of departures, and produces areas below published home-range sizes. The scope is a single, lightly hunted 2,593-hectare property, with observations beginning 23 August and ending 23 November. The paper lists 42 females plus 28 males; the file instead contains 43 and 27. We left that discrepancy unresolved.

Evaluate a sanctuary's hunting access along with the surrounding habitat. Before treating a quiet stand as evidence of a refuge, consider three questions. First, does a closure prohibit hunting, or does such limited hunting occur that mortality approaches zero? If neither applies, that spot is cover on hunted land. Second, are nearby woods scarce, like those in the Illinois county, or more plentiful, like those at the Missouri site? Third, are deer moving across the boundary? Without those crossings, neighboring hunters gain no benefit from a refuge.

The collar results also caution against assuming deer departed because a stand went quiet. In a single fall at Brosnan Forest, the typical buck's range center shifted 685 meters; 17 of 27 had shifts exceeding 500. Those files still cannot establish whether any deer departed the property. Use our guide on reading public-land pressure by effort to identify overlooked pockets of public ground.