Shared decision-making allows subordinates to lead when dominants monopolize resources

Subordinates lead when dominants monopolize.


Dominance hierarchy
We found a clear dominance hierarchy within vulturine guineafowl groups (Fig. 1, Fig. S2). The hierarchy of the habituated groups was very steep; the probability of the higher-ranked individual winning an agonistic interaction was 0.9 when the rank difference was only 1.

Movement initiations from non-monopolisable resources
All group members could initiate movement, but adult males higher in the dominance hierarchy were slightly more likely to be followed (Table S2). The reason why dominant individuals appeared to be more often among the initiators could be that males are 20-30% larger in body size, and therefore likely to be more motivate to lead. The observed differences in initiating group movements are more strongly defined in the Male-Female dichotomy rather than along dominance lines, because even individuals halfway down the male dominance hierarchy could initiate successfully. In addition, one female initiated frequently, and this female was the oldest member of the group. While we have no historical data on this individual, her leadership could reflect differences in her knowledge of the environment. Thus, while we found that there is stratification in leadership in this group, the differences do not suggest deviation away from shared decision-making, since all adult individuals were able to initiate movement successfully.

Time spent on the monopolisable patches
The mean duration of departure process from the periphery of the patch, from the onset of the initiation to the last individual was 1.23 minutes (SD= ±1.37, min= 0.85min, max= 4.78min) and the group members stayed on the patch for 8.16 minutes (sd=2.85).

The relation between dominance hierarchy and initiations from monopolisable patches
We found no consistency in the identity of the initiator of the movement away from the patch across initiation attempts (Table S4). That is, neither the dominant individual, nor other highlyranked individuals, were responsible for most of movement initiations. This clearly differs from the expectations of despotic leadership.

The relation between agonistic interactions on monopolisable patches and departure orders
The ordinal logistic regression model revealed that the probability for an individual to occupy the first orders in a departure increases significantly if this individual had been the loser of an agonistic interaction on the patch (Fig. 3, Fig. S6, Table S5). The departure is not a direct effect of being the loser of an agonistic interaction, as shown by the distribution of the time lags between losing in an interaction (i.e. being displaced) and subsequently departing (Fig. 4a).    The probability of leaving the patch right after having being displaced as a function of the timing of a session. (C)

Supplementary Figures
The distribution of the events of rejoining the patch after having being displaced as a function of the timing of a session.   Group cohesion 3 x S1 13 days

14
Where the group goes after departing from the patch 3 x S7

departures 15
Distance and speed of the last individuals leaving the patch from the rest of the group's centroid 3 x S6 1 Following on foot and voice recording 2 Camera tracking and voice recording 3 GPS tracking 4 All members were colour banded Usually observed in chicks, but also between adults when on the move and accompanied by a "crying call" and sometimes repeated in zig-zag. (eliciting little response from B) 9 SUB Ind A submissive caress around the chest of individual B (often triggered by an agonistic behaviour by B) Table S3. Overview of agonistic interactions. We identified 9 different types of agonistic interactions that guineafowl presented. In all but "8" and "9", individual A was regarded as the winner of the interaction and B as the loser.    But, because the data are difficult to interpret in this way, we present just the raw data in the main text in Fig. 3 Table S7. Individuals at the periphery of the monopolisable patch consumed significantly less food than individuals that remained on the patch. The results of the GLM testing the pecks per second for the birds that first accessed the patch (0) compared to birds that remained on the patch in the middle (0.5), at the end of the session (1) and in the periphery of the patch. Data come from HG2 and HG3.

Caption for Movie S1
The visualization of the GPS data for one day (25.03.2019) when HG3 visited the monopolisable patch following the process described in Fig.2 of the main text. The last individual that leaves the patch to catch up with the rest of the group (light green) is the alpha male.