Rotenone is a substance that occurs naturally in the roots of several plants (one of which is the commonly eaten jícama plant). It has actually been used for 1000s of years by indigenous tribes as a method of catching fish for food. It's a poison, but according to Wikipedia, it's only weakly absorbed through skin and the GI tract. So it's harmless to mammals. For fish, on the other hand, it's readily taken up through the gills and kills them quite quickly. What is being used here is rotenone in powder form. It gets mixed with water and a splash of dish soap (as an emulsifier) to make a sort of dough or thick batter (one of the researchers does this with his bare hands so it's some indication of how certain he is that it won't harm him). Then it gets double-bagged and is ready to be deployed.
The bags are taken down to a selected reef (usually on SCUBA), opened, and slowly shaken empty in the area. Since corals don't have gills, they are completely unaffected. The fish, however, die quickly. It has sort of a shotgun effect, killing everything with gills, whether a sample is needed or not. This is what made it so troubling for me. I don't like seeing things die for no reason. Fortunately, it quickly dilutes and is rendered inert by sunlight, so it doesn't have long lasting or widespread effects. It also mostly only kills small fishes. Anything larger than about 5 or 6 inches can withstand it long enough to swim to safety. Sadly, though, a large fish doesn't mean a large brain. Some big fish, mostly Yellowtail Snappers but also one large Yellowfin Grouper, were lured back into the toxic cloud by the prospect of an easy meal. They see the dead little fish and come swimming back in to gobble them up. If they do it enough times, they get a lethal dose themselves.
After the cloud settles out enough for us to see, we swim in and scoop up the dead fish lying on the ground. Those that are easily identifiable and which have already been collected, are left behind and, eventually, eaten by scavengers (which is harmless even to fish because the toxin is only effective when absorbed through the gills). The affected area is eventually repopulated from neighboring reefs or from unaffected portions of the same reef (the size of rotenone stations being done here is only about 20 or 30 feet). Again, its a matter of viewing it from an ecosystem-wide level or on an individual level - the impact is minimal on the ecosystem, but the impact on individual fish is dramatic.
Its understandable, though, why rotenone is such an important tool for this project - many species just can't be captured in other ways. As you can see from the pictures of the lab, a great diversity is turned up: