Use of drones to study the behavior of buffaloes in a production system in the Eastern Amazon

TL;DR

This study shows how drones can be used to monitor buffalo behavior in Brazil, helping improve farming practices with minimal stress to the animals.

Contribution

The study introduces a low-cost drone-based method for noninvasive behavioral monitoring of Murrah buffaloes in different production systems.

Findings

Drones effectively identified 17 buffalo behaviors with minimal stress at 15 m altitude.

Buffaloes in pastures showed more diverse behaviors than confined buffaloes.

Multivariate analysis confirmed significant behavioral differences between production systems.

Abstract

This study aimed to analyze the use of drones for behavioral monitoring of Murrah buffaloes in extensive and intensive production systems in Amapá, Brazil, contributing to the implementation of more accurate and sustainable breeding practices and reducing the need for labor. Ethograms were constructed at a height of 15 m, allowing safe and noninvasive identification of behaviors. The distribution of the data was analyzed for normality using the Shapiro–Wilk test (W = 0.803; p < 0.05) and homoscedasticity (F = 0.345; p = 0.558), which was shown to be homogeneous. Animal reactivity was evaluated using Spearman’s correlation coefficient, and the environmental effects on the response variables were evaluated using PERMANOVA. PCoA was used to explore the spatial distribution of the data. After 104 h of image storage, 17 behavioral types were identified. These results validated the use of the DJI Mini 2 drone for minimally invasive, effective, and low-cost aerial monitoring. The reactivity of the buffaloes to the drone decreased with increasing altitude, with 15 m being ideal for monitoring because it minimized stress and behavioral changes. Confined buffaloes (area 2) showed greater reactivity and spent more time in alert and tense states than buffaloes in pasture (area 1), which showed less reactivity due to more environmental stimuli. Multivariate analysis and PERMANOVA confirmed significant differences between the areas, with area 1 showing greater behavioral diversity (12 types) than area 2 (eight types).

Figures