The Crash That Hasn't Come: CWD Prevalence vs. the Wisconsin Herd
The models are unambiguous: once chronic wasting disease saturates a herd, the herd should shrink. Wisconsin's southern core is the best place in the country to test that prediction: prevalence there has climbed to one deer in three. So we asked the two most independent field measures we have (deer-vehicle collisions and total harvest) whether the core herd is falling faster than the rest of the state. Through 2024 it isn't, which is not the same thing as reassurance. It is a clock.
- In Wisconsin's eleven highest-prevalence counties, the deer-vehicle collision rate (per billion vehicle-miles) fell a median of just 6% from 2010–12 to 2022–24, while lower-prevalence counties fell 30%.
- Total harvest in the core fell a median 7%, versus 3% elsewhere: a real but modest gap, not a collapse.
- Put plainly: even at one-in-three prevalence, the core herd is not disappearing faster than the rest of Wisconsin in either independent signal. In crashes, it held up better.
- The survival science still holds (infected deer die young), but the population effect lags prevalence by years. The core is where that lag runs out first, which makes it the herd to watch, not the herd to write off.
IThe prediction
Long-running Wisconsin research put a number on it: once CWD prevalence among adult females passes roughly 29%, infected does stop surviving long enough to replace themselves, and the herd is expected to begin shrinking. Several southern-core counties are now at or beyond that line in the current-year data. If the threshold model is right and fast, the core should already be losing deer measurably faster than the clean north. The point of this paper is to check that against signals the model does not control.
IITwo honest signals
Harvest is a tempting herd gauge and a treacherous one: it moves with permit levels, hunter effort, and regulation as much as with deer. Deer-vehicle collisions have the opposite weakness (they move with traffic), but normalized to vehicle-miles they become a clean, involuntary count of deer on the landscape that no season structure can bias. The two are confounded by different things, so when both move together the signal is trustworthy. Here is what they say about the core.
IIIWhat the core actually shows
County by county the core is a scatter, not a slide. Richland (the state's highest prevalence) did see its collision rate fall 46%, the kind of number the model predicts. But Iowa County's rate rose 50%, Walworth's 51%, Grant's 27%, all while carrying 20-plus-percent prevalence. Harvest is similarly mixed: down 44% in Iowa but up in Rock, Walworth, and Juneau. There is no coherent fingerprint of a herd being emptied. If CWD were thinning the core the way the threshold model warns, eleven of the worst counties would not look like this.
| County | 2025 prev % | Crash-rate Δ | Harvest Δ |
|---|---|---|---|
| Richland | 35.7 | −46% | −12% |
| Sauk | 33.0 | +1% | −20% |
| Lafayette | 31.0 | −21% | −34% |
| Columbia | 27.4 | −41% | −6% |
| Green | 26.6 | −6% | −7% |
| Iowa | 23.5 | +50% | −44% |
| Rock | 23.1 | +20% | +1% |
| Grant | 20.7 | +27% | −6% |
| Walworth | 20.0 | +51% | +4% |
| Dane | 19.3 | −16% | −10% |
| Juneau | 16.5 | −25% | +2% |
Per-county change, 2010–12 to 2022–24. Median crash-rate change −6%, median harvest change −7% — and the signs are mixed, the opposite of a uniform collapse.
IVA clock, not an all-clear
This is where the paper has to be careful, because it is easy to misread. The finding is not that CWD is harmless: infected deer demonstrably die young, and prevalence is still climbing. The finding is that the population-level crash the models predict has not yet shown up in the two field signals best positioned to catch it, because that effect lags prevalence by years to decades. The core is the herd to watch, not the herd to write off. It is where the lag will expire first. When these two lines finally bend together in southern Wisconsin, it will be the realest herd-decline signal CWD has yet produced anywhere.
The models predicted a crash. Wisconsin's core hasn't delivered one, not because CWD is harmless, but because the fuse is this long. Watch the core; that's where it burns down first.
- Core = 11 Wisconsin counties with ≥15% CWD prevalence in the 2025 sampling year (≥50 deer analyzed). Comparison group = counties under 5% prevalence.
- Crash rate = deer-involved collisions per billion vehicle-miles (WisTransPortal / WI DNR Deer Metrics), which removes traffic-volume growth. Harvest = total county deer harvest (WI DNR deermetrics). Both are 3-year averaged at each endpoint (2010–12 vs. 2022–24) to damp single-year noise.
- Each signal is independently confounded — harvest by regulation and effort, crashes by road and traffic change — which is why an accelerated core decline must appear in both to count. Here it appears in neither.
- This is a population-level statement, not a claim about individual-deer risk; CWD's lethality to infected animals is established (see the companion paper on the Wisconsin epicenter and the 29% threshold).
- Absence of a decline signal through 2024 does not forecast later decades; the lag between prevalence and population effect is the subject, not a caveat to it.
Fully Drawn Research is an independent data analysis desk, not affiliated with the Illinois Department of Natural Resources, the Illinois Department of Transportation, or any mapping or outfitting provider. Figures trace to Wisconsin DNR and WisTransPortal public records, retrieved June–July 2026.