Figure 3.

The relationship between the completion of the nuptial moult and the relative contribution to population growth, p_t(i). The curves were estimated using a conventional linear model using REML (continuous line), a generalized additive model (GAM) based on a cubic spline (broken line), and by dividing the explanatory variable into the five categories from figure 1 (filled circles). For the latter case, sample sizes are the number of males in each category for this analysis, the value of nuptial plumage is the average for that class of individuals, and the vertical bar denotes twice the average standard error of the difference (s.e.d.) in the REML analysis.