First a question a bit on the side: I would assume that when tracking wildlife range is important. Why are you not using a sub-1GHz device like CC1120 that has higher output power and better sensitivity?
8m is very short but it is a bit difficult to say why without doing measurements. It could be that a desert is different RF wise than other places but I doubt it. The output stage of the PA in CC2500 runs on regulated power (the LNA runs on the same regulator) ensuring a fairly constant output power and sensitivity as long as the device is supplied with a voltage in the range 1.8V to 3.6V.
Is it in some way possible to get at least an indication on the TX and RX performance on the device in the field? If you have a kit controlled from SmartRF Studio connected with a SMA cable from SMA contact to SMA contact, do you get a RSSI that indicate that the TX side with your code send with the expected power? If you use the kit in parallel with your RX, which RSSI do you read? Is this consistent with a very low range?
