Wolf–Deer Movement After the 2021 Wolf Harvest, by North-South Gradient and Proximity to Towns using a Five-County Camera Record (2018–2025)

Damian Anthony Vraniak

Previous analysis of a wolf-enriched extract of the Snapshot Wisconsin camera network for five northwestern counties (Barron, Burnett, Bayfield, Douglas, and Sawyer; 525 cameras, 294,031 camera-days, 634,214 detections, 2018–2025; 1,856 wolf detections on 197 cameras) established patterns in the way gray wolves (Canis lupus) follow, select which class of deer to follow, and how strongly they displace white-tailed deer (Odocoileus virginianus). Using the same directional-following and return-interval methods, I asked whether the wolf–deer relationship differs (i) before versus after the February 2021 Wisconsin wolf harvest, (ii) near versus far from human development (per-camera Global Human Modification, GHM), and (iii) along a north–south latitude gradient. Then I asked how long a wolf delays returning to a camera after a person. Wolves arrived after deer 55% of the time before the harvest (n = 40), indistinguishable from a day-reassignment null (P = 0.30), and 85% after (n = 76, P = 0.002). The difference survived a joint model holding development and latitude fixed (odds ratio 4.8, P = 0.001) and a control that subsampled post-harvest wolves to the pre-harvest count (follow fraction 0.85, 95% band 0.75–0.94; no draw reached 0.55). Deer displacement by wolves was strongest in wild country and weakest near towns (wild ≈ 2.4×, most-developed cameras 1.5×; rank correlation P = 0.036). This was a nonlinear, threshold-shaped effect carried mainly by does (1.65× near vs 2.50× far, P = 0.058). Bucks stayed away from wolves more strongly than does, and did so everywhere, near towns and in wild country alike, so the easing near towns was a doe effect, not a buck one. Latitude was not an independent driver for the deer. The wolf’s own avoidance of people was the largest of the three main relationships (2.78× its baseline revisit interval, 89 cameras, P < 0.0001) — greater than the deer’s delay after a person (1.83×) or a wolf (2.24×). The wolf’s avoidance of people was stronger after the harvest (joint model ×2.6, P = 0.009), independent of nearness to human development, and stronger in the southern counties (×0.6 per standard-deviation northward, P = 0.008). Actual following times were short and stable across every split (medians 3.8–7.0 min, no significant difference), indicating that what varied was how often wolves followed and how long deer stayed away, not the timing of the following. I interpret these patterns cautiously given the modest number of wolf encounters, and propose that analysis of a statewide wolf-enriched extract is the natural next step.