In evolutionary ecology, we model survival--or any other metric of fitness--as a function of a phenotypic trait (body size, size of beak, etc.). In general, such a univariate "fitness surface" can show to shape of phenotypic (natural) selection acting on a trait. In many cases, we regress fitness onto two traits and generate 3-dimensional fitness surfaces. I would like to calculate a measure of skew (away from multivariate normal) for each surface. Any suggestions on how to do so? Ideally I would like a solution (or hints at a solution) in program R.