Is It Necessary to Kill Marine Fishes and Invertebrates in Order to Count Them?
Posted by Milton Love | , United States
There are a number of extremely valuable bottom fisheries throughout the world. In many places, the United States being a prime example, the assessments of these fish populations are carried out by bottom trawling. In this process, hundreds of net tows are made, the organisms are brought to the surface, and these dead and moribund animals are measured and identified. These surveys kill hundreds of thousands of fishes and invertebrates annually and, arguably, can damage sea floor habitats.
But is there a non-destructive way to carry out these surveys? One possibility may be to use Autonomous Underwater Vehicles (AUVs) — camera-carrying, untethered vehicles that fly above the sea floor. These vehicles can be programmed to travel a predetermined path, at a predetermined height above the bottom, and take photographs at preset times. A stereoscopic camera feature allows the lengths of organisms to be determined. In this way, both fishes and invertebrates can be identified, their sizes estimated, and their densities determined all without damaging these organisms. In addition, we are working with engineers here at UCSB to create image recognition software that will identify fishes and invertebrates to lowest possible taxon and thus will semi-automate the laboratory processing.
Thus, our question: Might it be possible to replace destructive trawl surveys with ones carried out by AUVs?
In order to determine whether this is a feasible approach it is necessary to compare the results of surveys conducted by government trawling surveys with those conducted in the same general vicinity by an AUV. This would tell us 1) how comparable are the species assemblages determined by each technique, 2) how comparable are the densities of these species, 3) how comparable are their size frequencies, and 4) how many and how long would AUV transects have to be to replicate trawl surveys. In addition, we would use the images gathered to train the image-recognition software to semi-automate the identification of marine organisms.