I don't know about GPS trackers, but my experience with scientific sensors in general has been that they tend to be quite expensive when compared to similar commodity items. They need to be in robust waterproof housings, compatible with other proprietary equipment and software, and accessible to non-programmers and newbs. I have seen things less fancy then a GPS tracker, like temperature or water level sensors with a few kb of storage and a serial port, that easily exceed $800 cost. This is frustrating to scientists, and many use substitute consumer equipment, roll their own when they can, or substitute cheap labor for overpriced automation. Sometimes an undergrad research assistant with sneaker-net is cheaper then then sensors' proprietary networking module. Most of this stuff has a "stuck in the 90's" kind of feel to it. 9600b serial ports, ncurses looking vendor software and propriety radio instead of wifi or cellular. Which would be cool if it was free and open, but it's not.

I think it's low-volume, high margin market, and not very competitive. An opportunity perhaps?