My pal
Julie Stewart tags Humboldt squid. She catches squid, attaches little recording devices to them, then drops them back in the ocean and waits for the tag to pop off a few days later. When it pops off, it's supposed to chirp out a satellite signal. That's Julie's cue to hop in a boat, pick up the tag and (hopefully) decode all the tag data to learn about the day-to-day lives of the squid.
That's the ideal model, and it doesn't always work out. Tags malfunction and the ocean is unpredictable, and a certain number are lost at sea. It happens.