Abstract
Rapid increases in sika deer (Cervus nippon) populations have recently begun to disrupt vegetation in wetland ecosystems in Japan. The Mizorogaike wetland in Kyoto, Japan, has sustained repeated invasions by sika deer. Research has indicated that the impacts of these invasions on wetland vegetation may be evaluated through the detection of deer traces by aerial photography. The use of unmanned aerial vehicles (UAVs), capable of taking repeated high-resolution aerial photographs in an inexpensive manner, would be a novel way to assess the consequences of sika deer invasion on wetland vegetation, which is often found in remote locations. To determine changes in deer traces over time, I analysed and compared aerial photographs taken by a UAV in the Mizorogaike Wetland on two dates: March 6, 2015, and February 26, 2016. Through object-based image analysis, I generated objects from the RGB colour values of orthochromatic images and a digital surface model (DSM), and then categorised these objects into deer traces and other zones. I calculated the density of deer traces by kernel density estimation, and determined the difference in deer trace density between 2016 and 2015, finding that deer trace density had increased toward the western periphery of the wetland. This result substantiated the effectiveness of this simple UAV-based method for the early detection of detrimental effects of deer inhabitation on wetland vegetation. We hope that the addition of UAV-based aerial photography to the library of ecosystem information will contribute to improved ecosystem management.
