2.4. Evaluation of female behaviour in small laboratory populations
To evaluate potential differences between the first occurrence
of stridulation and the first occurrence of mounting or mating,
respectively, those females that stridulated and allowed mounting
and mating were analysed in a generalized estimating equations
(GEE) model. The GEE allows repeated observations of the same
individuals nested in a hierarchy and the entry of not normally
distributed data. The data were fitted to a Poisson distribution with
a log-link function. Over- or underdispersion was eliminated by
choosing a robust estimation of covariance. In the GEE, female
behaviour (first mounting, first mating, first stridulation) was
included as a within subject factor with a first-order autoregressive
working correlation matrix. Female identity nested within an
experimental group (one of the five populations) was entered as a
subject effect.