Table 1: Results of generalized linear mixed-effects model testing the effects of ant abundance treatment (10, 20 or 30 ants), larval age (d) and time of observation (morning or afternoon) on the probability that ants would be tending Miami blue butterfly larvae. β is the coefficient describing the shape of the logistic curve (Figure 2). The table below presents the tests of contrasts among levels of ant abundance (the -value of the 10 ant treatment is from a test against the null of β = 0 and the -values associated with the 20 and 30 ant treatments indicate the difference between those treatments and the 10 ant treatment). The interaction term between ant abundance treatment and age of larvae was not significant so was removed to more accurately assess the main effects.