How to cope with negative estimates of components of variance in ecological field studies