What makes a good leader?
Traditionally, leadership in horses has been understood as a function of dominance, embodied in the myth of the “lead mare”: the highest-ranking individual in a group will be the one leading it.
Thanks to a growing body of scientific evidence, we now know that this is incorrect. Studies have looked at manifestations of leadership in domestic, feral, and wild horses, and the general conclusion is that “leadership”—defined as the ability to initiate movement among fellow group members—is a shared responsibility.
Any horse in a group, regardless of rank, can initiate movement and be followed by others, with a complex, context-specific, and communal decision-making process preceding the joint departure.
Interestingly, while rank doesn’t affect the likelihood that a horse will be followed by its group members, reliability might.
A recent study by Valenchon et al. (2022), ‘Does a high social status confer greater levels of trust from groupmates? An experimental study of leadership in domestic horses’, found that if a horse is unsuccessful in leading herdmates to a resource, they will eventually stop following, particularly if the unsuccessful horse has a low social status in the herd.
To investigate this, they studied three groups of six mares (mixed herds of Haflingers, Dartmoors, and Shetlands, ages varied between 1–23 years) at a French stud farm. Four horses from the first two groups were included in the third group, which was studied one year later. All groups had been stable for six months prior to the experiments, and lived outside 24/7.
Each group was filmed for six hours/day in the four weeks preceding the experiments. From these observations, social status for each horse in each group was calculated based on a combination of dominance and centrality.
Dominance was calculated based on agonistic encounters where one horse was clearly avoided by another horse.
Centrality was calculated based on how much time two horses spent within one metre of each other, for all possible combinations in each group. (It can be understood as a proxy measure for a horse’s social connection—a high centrality score means that more horses choose to spend time close to that particular individual, and a low score means that few horses choose to spend time close to that individual.)
In the table above, you can see the dominance (David’s score) and centrality index for each individual in each of the three groups A, B, and C—a higher number means a higher score.
There are some interesting things happening here, which I want to mention even though they are not part of the conclusions in the paper:
- There seems to be no clear relation between dominance and centrality. In group B, for example, the horse (ARM) with the lowest dominance score has the highest centrality score, indicating that even though she frequently avoids others in agonistic encounters, she also gets to spend a lot of time close to them.
- Dominance rank and centrality scores seem to be context-specific. The scores for the four horses from groups A and B that were also part of group C (ARM, OTO, SER, and TAK) changed when they became part of a new social group. Note, for example, the dramatic and counter-intuitive change for ARM: while her dominance score rose markedly in the new group, her centrality score fell.
These findings support that social interactions in horses are complex and context-specific, and that an individual’s dominance rank will change depending on the group composition.
But that was a side note, now on to the results of the study! Valenchon et al. did two separate experiments:
First, they tested which horses would be followed by their herdmates to a hidden resource. They removed one horse from the group, led her to a hidden pile of carrots, then released her back into the group. Then, they opened the enclosure so that the informed horse could go back to the carrots, and noted which of the other horses followed her.
They repeated this for every horse in every group, and “found no variation in the number of followers according to the initiator’s identity since a large majority of the trials was followed by the whole group”. This is consistent with previous research, which has shown that leadership is shared by all horses in a herd.
Second, they tested whether individuals with a higher social status were more likely to be followed even after an unsuccessful attempt. This experiment was only performed on group C, as it was the only group with a linear dominance hierarchy and a correlation between dominance and centrality scores, which enabled a clear assignment of highest and lowest social status (to SER and TAK, respectively).
The test setup was the same as in the first experiment, only this time they removed the carrots before allowing the informed horse to lead the group to the location, effectively making her an unreliable leader.
They did the test with the highest and the lowest scoring individuals. In addition, they did control trials with two mid-scoring individuals where the carrots were left in place, making them reliable leaders.
Interestingly, the number of followers decreased over consecutive trials in both unreliable leaders, regardless of their social status. There was no significant difference between the two, although the graphs (see below) show a more dramatic drop in followers for the horse with the lowest status, and no recovery of followers.
Looking at the graphs, we can also see that the two reliable leaders lost some followers, but that the majority of the group still followed them in most trials. The only statistically significant difference was between the unreliable leader with the lowest status and both reliable leaders.
A limitation of this study is the small sample size (only one group of six horses), but it does seem to indicate that reliability is a factor when horses decide who to follow: an unreliable leader who fails to lead the group to a resource is less likely to be followed the next time around. There also seems to be an association between perceived reliability and social status, in that a horse with a higher social status may be more likely to be followed even after a number of unsuccessful attempts.
The lack of any clear statistical significance means the results need to be interpreted cautiously for now. It’s also important to bear in mind potential learning effects—horses are very quick at associating places with food rewards, so unless the test location was varied between trials (this is not specified in the study as far as I can tell), it’s possible that the horses went to the location independently once they had discovered the presence of carrots.
Whether social status has a definitive effect on how likely a horse is to be followed, and which aspects of social status are the deciding factor, remains to be confirmed in future studies.
However, there is an ethical consideration when designing these experiments. Is it right for us to manipulate a horse’s reliability in this way? Given how complex equine social interactions are, we may well be disrupting individual relationships and group cohesion.
With these caveats, it is a neat little study that confirms the complexity of horses’ social behaviour. Most importantly, it indicates that “leadership” is less about any innate characteristics in “leaders”, and more about perceptions among “followers.”
This article was originally published on theequineethologist.substack.com and reprinted here with permission.