Automated Vertical Photography for Detecting Pelagic Species in Multitaxon Aerial Surveys

Abstract

Marine aerial surveys are designed to maximize the potential for detecting target species. Collecting data on different taxa from the same platform is economically advantageous but normally comes at the cost of compromising optimal taxon-specific scanning patterns and survey parameters, in particular altitude. Here, we describe simultaneous visual and photographic sampling methods as a proof of concept for detecting large whales and turtles from a single aircraft, despite very different sighting cues. Data were collected for fishing gear, fish, sharks, turtles, seals, dolphins, and whales using two observers and automated vertical photography. The photographic method documented an area directly beneath the aircraft that would otherwise have been obscured from observers. Preliminary density estimates were calculated for five species for which there were sufficient sample sizes from both methods after an initial year of data collection. The photographic method yielded significantly higher mean density estimates for loggerhead turtles, ocean sunfish, and blue sharks (p < 0.01), despite sampling a substantially smaller area than visual scanning (less than 11%). Density estimates from these two methods were not significantly different for leatherback turtles or basking sharks (p > 0.05), two of the largest species included in the analysis, which are relatively easy to detect by both methods. Although postflight manual processing of photographic data was extensive, this sampling method comes at no additional in-flight effort and obtains high-quality digital documentation of sightings on the trackline. Future directions for this project include automating photographic sighting detections, expanding the area covered by photography, and performing morphometric measurement assessments.