Abstract
<jats:p>Recreation and routine land management can alter wildlife behavior even where they do not reduce abundance, and the acoustic component of human activity is difficult to isolate from human presence itself. Within a single restored prairie–savanna, the same trail network is worked on foot, on a silent electric utility vehicle (UTV), and behind gas engines (a gas ATV and a mower), with all vehicle use confined to daytime. This design holds time-of-day and location roughly constant and allows engine noise to be separated from human presence as a disturbance cue. Using a steward-conducted full-frame camera census (4,248, 5,787 and 1,785 labeled frames in 2024, 2025 and January–June 2026), I measured the time for white-tailed deer (Odocoileus virginianus) to return to a camera after a human pass of each conveyance type, expressed as a ratio to each deer population's own baseline inter-visit interval. Across all three years the direction was consistent: deer returned fastest after the electric UTV (pooled ratio 0.83, median 7.18 h), slowest after a gas engine (0.99, 8.49 h), with foot traffic between (1.02). The contrast was statistically significant in the single best-sampled year (2025: Mann–Whitney p = 0.0161) but not in the pooled three-year test under strict independent-pass counting (p = 0.1433). The methods are standard camera-trap tools; what is uncommon is the controlled within-property electric-versus-gas natural experiment and the multi-year full-frame census that supports it. I present the result as a management-relevant lead — quiet, electric stewardship appears less disturbing to the most abundant animal on the land — while being candid about its statistical fragility, single-site scope, and the limits of relative-abundance metrics.</jats:p>