The Influence of Community Feeders and Commercial Food Outlets on the Spatial Distribution of Free-Roaming Dogs—A Photographic Capture and Recapture Study

**Background:** Free-roaming dogs are a significant public health and animal welfare concern in urban environments, particularly in Brazil where the canine population is estimated at 54.2 million. While it is known that food availability influences dog distribution, the specific roles of direct (community feeders) versus indirect (commercial food outlets) food sources have been poorly quantified. Understanding these spatial relationships is critical for designing effective population control, zoonosis prevention, and welfare improvement strategies.

**Methods:** The study was conducted in Divinópolis, Minas Gerais, Brazil (population ~242,505), across eight neighborhoods with ~7,600 residents. Over five sampling campaigns between September 2018 and September 2019, researchers drove a fixed route at 20 km/h on three consecutive mornings per campaign, photographing all free-roaming dogs (defined as animals not accompanied by an owner) at a maximum distance of 10 m using a Canon PowerShot SX60HS camera. At least three photographs per dog were taken from different angles. Geographical coordinates were recorded for each sighting, each community feeder (food provided by households in containers or on the ground), and each commercial food outlet (fast-foods, grocery stores, bars, restaurants, bakeries, butchers, food markets). Two independent researchers classified photographs with high inter-rater agreement (Kappa index = 0.86, p < 0.01). Spatial clustering was assessed using nearest-neighbor distance analysis and Kernel density estimation (100 m search radius, quantile normalization). Bivariate Ripley's K function evaluated spatial correlations between dog distributions and food source locations up to 500 m. Median distances were compared using the Mann–Whitney test (α = 0.05).

**Key Results:** A total of 1,207 photographic captures/recaptures were recorded, identifying 554 unique free-roaming dogs. Males comprised 62.6% (347/554) of captured dogs and 64.7% (279/431) of recaptures. Small dogs predominated (57.7%, 320/554), and black was the most common coat color (42.0%, 233/554). The number of new captures decreased across campaigns (178 → 101 → 124 → 94 → 57), while recaptures generally increased (0 → 59 → 71 → 88 → 81). A total of 151 community feeders and 124 commercial food outlets were identified across all campaigns. Nearest-neighbor analysis confirmed that dogs were distributed in a clustered (non-random) pattern (p < 0.01). Kernel maps showed agglomerations consistently near food sources across all campaigns and for both sexes. Ripley's K function revealed positive spatial autocorrelations between dog distribution and food sources up to 500 m. The median distance from dogs to community feeders was 1.2 km (IQR = 0.5–2.6 km), significantly shorter than the median distance to commercial food outlets of 1.4 km (IQR = 0.5–2.7 km; p < 0.01).

**Clinical Implications:** The study provides quantitative evidence that community feeders exert a stronger pull on free-roaming dog populations than commercial food outlets, likely because feeders offer a more continuous and stable nutritional source with less human presence. This has direct implications for urban planning and animal management: relocating community feeders to designated areas with reduced vehicle and pedestrian traffic could decrease accident and bite risks while maintaining animal welfare. The findings also underscore that without addressing the underlying human behavior of feeding and abandonment, population control efforts (e.g., sterilization alone) will be insufficient. The photographic capture-recapture method proved effective and safe for urban canine ecology studies, though limitations include morning-only sampling and potential missed mobile animals. The results are likely generalizable to other Brazilian and global urban settings where community feeding of free-roaming dogs is common.