Figure 1
(a) Representative examples of four small simulated networks evolving from 1, 5, 10, and 30 seasons of 66 pairwise contests between 12 competitors (‘everyone against everyone’). Arrows indicate winners and losers of the last contests (winner → loser). Colors indicate accumulated number of tokens (red: > 20 tokens, pink: 10–20 tokens, gray: 3–9 tokens, light blue 0–1 token). ‘Null variant’: Each competitor starts with the same number of tokens (“homogeny” of the initial resource-holding-power). Odds of winning are (¼), losing (¼) or drawing (½). Outcomes of previous contests are not taken into account. ‘Memory5 variant’: Competitors “remember” the last 5 contests. ‘Winner-loser variant’. Competitors “remember” the outcome of all previous contests. Winners steadily accumulate, whereas losers progressively lose tokens. These networks turn into centralized ‘star’ networks with one or a few centrally situated individuals. ‘Keystone variant’: Keystone network with at least one individual with a priori higher resource holding power. In this simulation, the individual remained in the group of three central individuals. (b) Triad motif patterns (mean values, 2nd and 3rd quartiles, and 95% confidence intervals) in 10 repetitions. ‘Null variant’ games remain random, whereas ‘Winner-loser variant’ and ‘Keystone variant’ games quickly centralize. dd double dominant, ds double subordinate, pa pass-along, tr transitive, cy cyclic triads.
