I'm trying to learn more about working with autocorrelated data. I'm wondering about using an autocovariate (neighborhood y values used as predictor of focal y value) and a correlated error structure simultaneously. The standard equation fitted for correlated errors includes a weight function on all Y and X values. It seems in this case that adding an autocovariate might be redundant because the weighting of the Ys might be adjusted to account for this in the stand-alone correlated error equation. One way to check would be to run both and see if the same predictions are given. I thought someone here wiser than I may have a quick answer and intuition or tell me a place to look for such.