I think I get where you're coming from on this, so I'll add a bit more to the discussion.
I've seen some incredibly passionate/heated arguments over the subject, but I'll throw this out there anyway. I don't think we can completely rule out that the fishes body doesn't come into play to some degree, especially with more sophisticated units now on the market, just that it's the air bladder that is the primary thing responsible for us seeing fish on our units. With that said then, it's not the actual thermocline (temperature gradient) we're picking up in most cases, but more often the things living and suspended in and around the thermocline instead (plankton, detritus, silt, micros, etc.). Let me explain...
Water at its most dense (4 C) has a density of 1000 g/cm3. At 20 C, water is only slightly less dense at 0.9982 gm/cm3, about a .18% difference. But, 4 C to 20 C would make for a pretty significant thermocline. However, the average fish body has a density of 1080 g/cm3 (range 1040-1090 gm/cm3 @ 20 C), which is a significantly greater difference in density than the temperature difference mentioned above. As such, if it's not the fishes body we're picking up mostly on sonar returns, then it's also likely not the temperature/density gradient we're picking up in thermoclines either. The change in density (whether talking fish or temperature) changes the reflectivity of the water (the speed of sound increasing or decreasing with density differences, as well as the slight bending of the signal through a different density medium), and sensitive units might be able to detect some of this occurring, but in most cases with typical lower powered units, it's more likely the "stuff" in the thermocline we're picking up, "giving away" the presence of the actual thermocline itself.
Hopefully that makes some sense.