Digging experiment and attacks on the introduced predator: results of two poisson GEEs with log-link, testing for fixed effects of the treatment, status and their interaction on the digging plus sand carrying effort after the breeding shelter was covered with sand (<i>n</i> = 117) and attacks on th

Note that for practical reasons effort was measured for all individuals per status lumped, so the number of individuals observed per status was entered as a covariate to arrive at per capita estimates. Results were corrected for random group effects (repeated measures of 24 groups for digging and 33 groups for attacks, entered as subjects in the GEEs), the scaling parameter was adjusted using the deviance method in each model. aGroup member status was divided in five classes (n = 24 in each case): small helper (15.5–25.0 mm SL, mean individuals per group ± s.d., range: 2.58±1.35, 0–6), medium helper (25.5–35 mm SL, 2.33±1.58, 1–8), large helper (>35 mm SL, 2.33±1.17, 1–6), breeder female (1.04±0.36, 0–2) or breeder male (0.96±0.20, 0–1), so in total 222 individuals were observed. Small helpers were missing for one group with a medium predator, and both the breeder female and breeder male were missing from another group with a medium predator. bGroup member status was divided in five classes (n = 33 in each case): small helper (15.5–25.0 mm SL, mean individuals per group ± s.d., range: 2.74±2.03, 1–10), medium helper (25.5–35 mm SL, 2.45±1.58, 1–8), large helper (>35 mm SL, 4.88±2.83, 2–17), breeder female (1.09±0.29, 1–2) or breeder male (1.00±0.00, 1–1), and since these groups were observed multiply we had 803 cases in total. However, since small helpers were never seen to attack the introduced predators, they were omitted from the GEE and this reduced the sample size to 664 (n = 33 groups with 8 to 36 measurements per group).