It isn't easy to get accurate fish counts in the wild. Every method is going to be flawed to some degree. The use of nets is the most accurate that I can think of. Nets provide some constants in the sampling (net area, mesh size, location, orientation, depth and time deployed). It still leaves a lot of variables, but significantly fewer than other sampling methods.
I don't like the fact that they have to use destructive sampling methods, but, given the environment and the nature of the research subject, I can't think of any other way to achieve results that would be as accurate.
If you read through this report
The Nearshore Fish Assemblage of the Scripps Coastal Reserve you will see that similar sampling methods were used right here in La Jolla for this 2002 report. They used otter trawl, gill net, diver survey and ichthyocide sampling and collection methods. Judging by the comments on the other forum, I think that some people are not aware that gill netting and other destructive sampling methods are allowed under scientific take permits. This is not new and it is not illegal.