Figure 1.
Focal subordinate behaviour depending on the predator treatments.
Depicted are means ± s.e.m. per 15 minutes observation time, except (A) per minute not hiding, of behaviour depending on the treatments (white circles: control; black triangles: medium predator; black squares: large predator) and helper size (small symbols and thin lines: medium helpers; large symbols and bold lines: large helpers). For statistics see Table S1. (A) Feeding rate per minute not hiding, (B) total food intake, (C) digging frequency, (D) territory defence frequency (excluding against introduced predator), (E) within-group conflicts, (F) within-group social contacts. Sample sizes are n = 42 for each symbol, except for medium helpers large predator treatment (n = 41 due to one sample lost).
Figure 2.
Focal subordinate feeding rate was closely related to spacing behaviour.
Intake rate increased with average distance moved from protective shelter in both (A) large helpers and (B) medium helpers. Depicted are hyperbolic curve fits (distance/[a+b x distance]) from non-linear regressions for large helpers (n = 126, coefficients ± s.e.): a = 1.1572±0.2642, b = 0.0257±0.0064 (F2,124 = 229.7, p<0.001, R2 = 0.23); and for medium helpers (n = 125): a = 0.4575±0.0910, b = 0.0342±0.0042 (F2,123 = 226.6, p<0.001, R2 = 0.30).
Figure 3.
Effect of covering the breeding shelter with sand on subsequent digging and carrying sand behaviour from this shelter by the different group members in the three predator treatments. Depicted are means ± s.e.m. and sample sizes (number of groups). For statistics see Table 1.
Table 1.
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 (n = 117) and attacks on the introduced predator (n = 664) separately.
Figure 4.
The per capita frequency of aggression against the introduced predator.
(A) Shows the introduced predator on the left (L. elongatus) and defending group members of N. pulcher on the right. Depicted are means ± s.e.m. of aggression and sample sizes (number of groups) for the five different types of groups members in the (B) medium and (C) large predator treatment (Sh: Small helpers, Mh: Medium helpers, Lh: Large helpers, Bf: breeding females, Bm: breeding males). Note that small helpers were never seen to attack the medium and large predators (see text). For statistics see Table 1.
Table 2.
Results of Spearman Rank Correlations between the per capita frequency of aggression against the predators by the different group members and the number of adults living in the group (number of breeders and large helpers), for the medium (above diagonal) and large predator treatments (below diagonal), separately.
Figure 5.
The ratio of (A) helping effort and (B) submission shown to the breeders, per received aggression from the breeders depending on the predator treatments (white circles: control; black triangles: medium predator; black squares: large predator) and helper size (small symbols and thin lines: medium helpers; large symbols and bold lines: large helpers). Depicted are means ± s.e.m (ratios = [frequency helping or submission square-root +3/8 transformed]/[frequency breeder aggression received square-root +3/8 transformed]) with sample sizes (number of observations). For statistics see text. Note that the sample size for medium helpers in the large predator treatment was n = 40 due to one sample lost and one sample the helper was hiding 100% of the time (gives 42–2 = 40).
Table 3.
Expected effects of the predator treatments under the three hypotheses mentioned in the introduction and the observed differences (-: predator treatments < control treatment, 0: predator treatments = control treatment, +: predator treatments > control treatment).