Using drones to detect and quantify wild pig damage and yield loss in corn fields throughout plant growth stages

Abstract

Abstract Presently, there are an estimated 6.9 million wild pigs (Sus scrofa) in the U.S., which cause over US$1 billion in damage to agriculture, environmental impacts, and control costs. However, estimates of damage have varied widely, creating a need for standardized monitoring and a method to accurately estimate the economic costs of direct wild pig damage to agriculture. The goal of our study was to integrate remotely sensed imagery from drones and crop harvest data to quantify wild pig damage in corn fields. We used drones with natural color (red, green, blue) cameras to monitor corn fields at different growth stages in an agricultural matrix in Delta County, Texas, USA, during 2019–2020. We flew 36 drone missions and classified wild pig damage in 18 orthomosaics by a combination of manually digitizing and deep‐learning algorithms. We compared estimates of damage from drone imagery to those derived from ground‐based transect surveys, to verify pig damage. Finally, we compared damaged areas of fields to maps of collected real‐time yields at harvest to estimate yield loss. All classified drone orthomosaics of pig damage had >80% overall accuracy for all growth stages. Ground transect surveys, which subsampled 2.6–4.1% of the field, were found to miss damage compared to the complete field coverage provided by drone imagery. Most damage occurred in latter growth stages, when corn ears were maturing, seed was most nutritious, and producers had already invested in the majority of annual crop inputs. Wild pigs damaged up to 9.2% of a single monitored field, which resulted in a mean loss of 3,416 kg of corn/ha and a direct cost to producers of US$17.18–48.24 per ha of damage. Drone imagery, when combined with spatially‐ explicit, harvest yield data, provides an accurate assessment of crop damage and yield loss due to wild pigs in the currency required for the cost‐benefit evaluation of management actions.

Keywords

• crop damage
• deep‐learning
• drones
• harvest yield
• Sus scrofa
• wild pigs