On the asymptotic equivalence of two ranking methods for K-sample linear rank statistics