Table 1.
Relationship benefit index (RBI) of all subject – subject dyads in the North group of Taï chimpanzees.
Figure 1.
Predicted relationships between the bystander (B) and both of the opponents (A: aggressor, V: victim) for each of the three hypotheses are shown separately for bystander affiliation with the victim (top) and with the aggressor (bottom).
Solid lines show the observed interactions (conflict: double line; bystander affiliation: single line), discontinuous lines show the predicted relationship (dashed: good relationship; dotted: bad relationship), while no line indicates no predicted relationship.
Table 2.
Bystander affiliation with the victim: Mean relationship benefit index (RBI) of bystanders (N = 12) with the aggressor and with the victim and the percentage of friendly interactions following bystander affiliation.
Table 3.
Bystander affiliation with the aggressor: Mean relationship benefit index (RBI) of bystanders (N = 15) with the aggressor and with the victim and the percentage of friendly interactions following bystander affiliation.
Table 4.
Results of the Generalized Linear Models (GLZ) investigating the effect of bystander's ID, sex and RBI with the two opponents on whether or not successive interactions are friendly.
Figure 2.
Percentage of friendly interactions during the second post-conflict interaction by the recipient of affiliation, following bystander affiliation with the victim (left) or with the aggressor (right).
This is dependant on the RBI of the bystander with either the recipient of affiliation (left) or the recipient's opponent (right).
Figure 3.
Latency of friendly second post-conflict interaction among the opponents following bystander affiliation.
The data are divided in categories depending on the bystander's RBI with the recipient's opponent (RBI: 1 = no, 2 = weak, 3 = good friendship). The latency is presented in relation to the average inter-interaction time between individuals' friendly interactions (relative latency >1 indicates that post-affiliation interaction has a longer than average latency). Error lines represent the 90% confidence interval (CI) of the distribution calculated with bootstrap sampling. In case the CI excludes the value 1, opponents are significantly (p = 0.05) less tolerant with each other than under normal conditions.