Dutch roll is initiated by oscillation in yaw. The yaw roll coupling of a swept wing (acts like dihedral) then causes an oscillation in roll. So putting a gyro on ailerons stops the roll but still leaves the fishtailing. Putting a gyro on rudder stops the fishtailing which stops the rolling too. That's why full size primarily use yaw dampers when no roll dampers are fitted. However some models with highly swept wing can have a very high gain of yaw to roll, meaning the very slight yaw that a gyro must allow may translate to a tiny and just noticeable wing rock. (A rate mode gyro can not prevent rotation, for some rotation must exist in order for the gyro to sense it, what the gyro does is stop the rotation after a much smaller angle than a human reactions can) Therefore sometimes for the sake of aesthetics it is better to have the gyro on ailerons to stop the wing rock and just accept the very little, barely noticeable fishtailing. The fishtailing will actually be reduced by the aileron gyro because in Dutch roll there is feedback from the roll angle causing sideslip which causes a yawing. Ideally you would have both yaw and roll gyros, setting the yaw gyro first and then the aileron gyro to remove any last bit of wing rock that is left.

As mentioned by John Wright, there are gyros such as the ACT Fuzzy Pro that have 2 channels in and 2 out to cope with ailerons and elevons. I use the ACT Fuzzy Pro for the elevons of my Eurofighter Typhoon. I think Futaba and JR also have dual channel gyros. One thing to beware of is that elevons often have much smaller aileron travel than elevator travel, so the aileron gyro will typically only need the lowest of gains. If you set up your Tx gyro gain travels to the normal defaults so that you can set up the gyro in flight, it will be very coarse and may go into over-gain too easily, so if the ail travel of your elevons is much less than the ele travel, turn down the gyro max gain travel in the tx to get a finer control. Another snag with elevons is the gyro's fade out is based on full elevon travel. So if the ele travel is greater than the ail travel, instead of fading out to 0% gyro at full aileron, it will only be down to say 50% fade out and you get a very reduced roll rate. Therefore after flight testing to establish the correct gain, I mix the aileron stick to the gyro gain channel and program a multi-point curve so that the aileron stick controls the gain and makes the gyro fade out properly. The gyro's action has an exponential effect so you can expect to reduce or even cancel Tx expo.

H