Appendix A. Justification for using both open and shrub-covered foraging sites to determine the relationship between quitting-harvest rates and population densities.
We regressed the log-transformed GUDs for both open and shrub-covered foraging trays against population density. We treated open and shrub-covered foraging sites as independent observations in these regressions because use of each type of site depends on the quality and risk of that site relative to all others available. So use of covered trays, for example, is (mostly) independent of the use of open ones. But the data for each site were collected at the same time, at identical densities, and perhaps from identical foragers. We could compensate for this quasi-independent effect by using the mean value for both types of sites in the regression. The regression equation would be identical in both approaches, but not the degrees of freedom. The degrees of freedom in our test are inflated. But the degrees of freedom for the alternative are too low. So though our test may be too liberal, the alternative is likely too conservative. Regardless, the relationship is so tight (Fig. 4), and absent of density-dependent bias (Fig. 5), that either approach will yield the same outcome.