 So thanks very much to the organizers for inviting me. I'm really excited to talk to you guys today about some work that I've been doing looking at collective movement in animal societies. And so one of the fundamental tasks that many animals in groups that live and move together as a group have to solve is the task of coming to consensus. And consensus decision making has to take place in order for the group to stay together, despite the fact that animals in groups often may not agree on where the group, where and when the group should move. And so people have been really interested in this topic of how groups come to consensus and in particular the distribution of influence within groups for a very long time. And to prove this to you, I have a quote from Aristotle from History of Animals where he says that some animal groups obey a leader and others are anarchical and gives some examples of Stork in the B being the leader class and the ant being the other anarchical class. So as is usually the case with Aristotle, he was a little bit off on the details and not quite right, but it does just go to show that this is something that people have been thinking about for a very long time. And up until, and when we think about consensus decision making in animal groups today, we often think about it as falling along a spectrum where at one end of the spectrum, decisions can be completely shared, meaning that all individuals have equal say in the decision outcome, whereas at the other end of the spectrum, decisions can be completely unshared, meaning that one individual has all the say in what the outcome is and everyone else is just following that individual. And of course, this isn't so binary, we can fall also along this spectrum and there also can be different mechanisms by which animal groups can arrive anywhere along this spectrum. And so people interested in this topic has continued from the 4th century BC up until the present day. Here's just a couple of examples of recent review papers looking at this topic. But in the past, this has been a really actually very difficult thing to study in the wild. And the reason is because the traditional way of studying animals in the wild was of course to go out and watch them. So often, especially when we're thinking about stable social groups, what you will do is you'll go out and as an observer and sort of habituate the group to your presence so they no longer mind you and then you can kind of take behavioral observations. But as an observer in the field, it's very tricky to look at these collective processes. And the reason for that I think is nicely put in this quote by Richard Byrne in his chapter in On the Move, whoop, off the bottom of the screen there, but in the quote he said that this looking at collective decision making is an observational task of daunting dimensions to attempt to record the actions of many potential decision maker animals at once. And so in order to be able to tackle this daunting task, some recent tools are becoming quite helpful. So in particular, GPS tracking technology, which is what I'm gonna be talking a lot about today, has really revolutionized our ability to actually look at the behavior of many individuals in groups simultaneously and in the wild. And so today I'm gonna be telling you three sort of short stories about looking at collective movement and collective decision making in animal groups. The first one is about baboons, which is some of my previous work that I did during my PhD. And then these latter two are some more recent work that I've just begun looking at now. So first for the baboons, whoops, and this is a project that is in collaboration with Meg Krofutt, Damien Freen and Ian Cousin. And so why are baboons an interesting system in which to study collective decision making? So just first a brief bit of information on baboons. They live in stable groups that stay together throughout the course of most of their lives. And those groups move around together throughout the day foraging. And that means that they are constantly having to come to consensus about where and when they're going to move. But groups are also very heterogeneous. So you have males and females of different sizes. You have different age classes. And you also have a complex social structure where individuals will maintain long-term relationships with one another over the course of years. And in addition, as we already heard in the last talk, they're characterized by dominance hierarchies. So in general, the males are dominant over the females and the troop. And also within each sex you have a sort of linear dominance hierarchy or pecking order that has to do with who gets access to resources, who can displace others and who gets access to mating. And so baboons, being mainly terrestrial and widely abundant, are convenient to study. And they have been very well studied from the perspective of understanding their social structure. And so we really understand a lot about this system, but what's been tricky to study for the reasons I mentioned before has been the process by which these groups collectively decide where to move throughout the day. And so to tackle this observational task of daunting dimensions, my collaborator, Meg Crowfoot, went out and she put GPS collars on most of the baboons within a troop. So here I'm showing you some of the data. Each dot represents a baboon. The colors represent the age sex class of the baboon, which you don't need to really worry about here. And as you can see, here they are. They were at their sleep site in the morning and they headed off across this land over to that other river on the right. And so we have the positions of these baboons every second over the course of a couple of weeks. And we want to understand something about the dynamics of how they're making decisions. And so in order to do that, we came up with this kind of operational definition of influence that I like to call pulling. So it's a really simple idea. If you have two baboons and one baboon moves off and the other one follows it, we can pull that out and call it a successful pull. And if you have the same two individuals, one goes out, this one doesn't follow and the red one subsequently comes back, we would call that a failed pull. And so we can extract these pulls automatically from the data. So here on the left, I'm showing you an example of a successful pull and on the right, an example of a failed pull. And I think it's important to note here that of course this is observational data. So we don't really know why this blue one followed. We don't necessarily know that there was a causal relationship between these two things. But what we can do is look at patterns across many, many of these events and see whether it can tell us, suggest some ideas about how the process of decision making is taking place. So one thing that is very characteristic of baboon societies is their dominance hierarchy. And so one thing that we might expect is that because dominant individuals have control over all these sort of social things, they may also have control over where the group goes. So we looked at the probability that an individual was followed in one of these pulling events as a function of their dominance. And we were actually surprised to find that there was no effect of the individual's dominance. So even so the most dominant individuals were not followed more than the subordinates. And there was also no effect of age sex class on this probability of being followed. But of course it's not always the case that you just have individuals going out. You can also have instances where there's multiple groups going out in different directions. And so then you can ask the question, as a follower, what do you do if there are multiple groups going out in these two directions that we clustered here into cluster one and cluster two? So we extracted some of these events and what I'm gonna show you on the y-axis is the probability that a baboon follower chooses cluster one as a function of the numerical difference between the number of baboons in cluster one and the number in cluster two. And so what we find is this sigmoidal shaped curve here where if say you have three more baboons in cluster one than cluster two, but a baboon follower will follow them about 80% of that cluster, about 80% of the time. And this is sort of a classic shape of a curve when you have a majority rule. So in general, the baboons tend to follow the majority and the greater the numerical difference between the majority and the minority, the greater the effect. So to us, this suggested that there may be less of a leader in the group and more of a shared decision-making process. But one thing that we are really missing here is the role of the environment because these baboons are not just moving around in a featureless area, they actually are interacting with things like trees and roads and everything else. So what I'm gonna show you to drive this point home is two videos. The first one is gonna be the same, it's gonna be some trajectory data from the baboons shown on a blank green background. So you can try and draw some conclusions from watching this. So you can see they're initially in kind of a line and then up here, they seem to split a bit. These guys go up right and then eventually rejoin the rest of the group over there. So we get some information from that, but now I'm gonna show you the same video but it's gonna be plotted on where it actually took place in the world on Google Earth images. So now you can see that when they were in this really straight line, they were actually aligned with this road here and the place that they split happens to be exactly the place where the road split. I'm sure it's a complete coincidence. So this I think just goes to show that clearly the environment is also an important driver of movement decisions and so we can't just ignore this big effect. So in order to try and take the environment into account, we went back out to Kenya where the baboons live and we flew this drone, this fixed wing drone, only one, not a flock of them, over the area and collected overlapping images which allowed us to then construct a three-dimensional representation of the environment. And what we can then do is basically put the baboons back into their environment. So here's the baboon trajectories overlaid on this three-dimensional map of the environment and this then allows us to try and understand and disentangle social from environmental influences on movement. And so the general approach that we used here without going into too much detail was to model the movement decisions of individuals using a step selection approach which is an approach from coming out of movement ecology and basically the idea is that you wanna build a model to predict the movement decisions, probabilistically, of individual animals based on the features of their environment and you do so by comparing locations that they chose to locations that they could have gone to but didn't and you try and figure out what are the features of those chosen locations over the non-chosen locations. And what we did here was to not just incorporate features of the habitat but also social predictors of movement into those models. And so what we can then do is from those models we can extract what are the features that are in most of the good models are sort of the highest weighted predictors of baboon movement decisions. And we can look at both these habitat features over here as well as these social features over here. And so what we found in general was that both habitat and social features came out as important predictors in these models. And in particular we found some very interesting effects here like if we look at the social factors actually the current positions of other baboons was not very predictive of where a baboon goes but rather the past predictions of where the troupe mates have gone in the past five minutes which was also a parameter we could fit in the model was the most important predictor of baboon movement decisions. So in a sense they're kind of following in the footsteps of others which again is aligned with this idea of majority rule but it shows that not only are they moving in the direction of the majority but also moving through the same specific areas. And we also got a couple of habitat features which came out as very important for one thing the sleep site location which is not surprising is they need to leave there and come back to there in the morning and the evening. And we also found that the baboons as you saw in the first video really like to follow roads and what I thought was kind of fun was that it turns out that they follow roads much more in the morning than in the evening than in the midday. So I like to think they have that their little daily commute as well. And so putting this all together we can see that both social and habitat factors are driving movements of baboons. And in particular on the social side we think from based on the data that the decision making process seems to be more shared than rather than unshared. And that also there's habitat factors are important for coming into play to understand why they move where they do. But of course we're still here only looking at movement and I think you could really criticize this by saying hey you've taken this really complicated animal that can be doing a lot of different stuff and you've kind of just reduced it to this dot that can move around. And of course these other things that this animal is doing and these other factors of its life may also have a really important impact on the process of collective decision making. And so just a couple of examples of things that also might be going on to influence movement decision making. We already talked about the role of the environment but there's also the role of behavioral context. So what are these animals actually doing? Are they grooming each other? Are they fighting with each other? And not just the sort of local timescale stuff but also the more long term social bonds between these individuals can have an important impact on their decision making process. And in addition we've looked at only how baboons are moving relative to the movements of others but we haven't looked at the fact that they may actually be producing signals. So either gestures or also vocalizations that could influence the decision making processes within the group. And so in order to kind of try and build up this multi-layer picture and bring some of these other factors in what I've been doing in the past of the while has been trying to develop collaborations with long term field studies that are able to bring in some of these other dimensions. So basically to try and apply some of these high resolution tracking methods but within the context of a long term study where we have lots of this information on the system. And so and in particular I've been focusing on the role of signaling. So despite the fact that active signaling is not required necessarily for group coordination we do see signals very commonly used in animal groups for the coordination of movement. And in today I'm gonna be talking mainly about vocal signals. So for example we say lots of species in which you have signal vocalizations that are given during movement to kind of allow animals to keep track of where others are in the group and these are called cohesion calls. You also have instances where individuals will give specific calls to initiate more rapid or directed group travel. And finally you can also have calls that are used to recruit others to join in some kind of collective action. For example mobbing a predator or defending your clan territory against an intruder. And so what I'm gonna now tell you about very briefly is just two studies where I've been trying to to start incorporating this role of signaling to understand how it interacts with movement dynamics. So first I'm gonna tell you a little bit about this project that I've been working on in collaboration with Marta Manzer at the University of Zurich and the Kalahari Meerkat Project. And so this is looking at communication and collective movement in Meerkat groups. So first why Meerkat? So very similar to baboons in some ways but Meerkats live in stable groups that stay together most of their lives and move around together foraging throughout the day. So again they're making these consensus decisions all the time. And they also have dominance hierarchies. In this case they mainly have a dominant male and a dominant female who are monopolized the breeding within the group. And then they have a bunch of subordinate helpers that are their children that actually help to raise the younger siblings essentially. It's called cooperative breeding system. And one thing that makes them good for looking at the interplay between signaling and movement is that Meerkats are very vocal and they use a lot of different acoustic signals. Some of which are associated with coordinating movement. And based on the work of Marta Manzer my collaborator over the past really more than 20 years their communication system has been very well characterized. So they have over 30 different distinct signals that they give, vocal signals that they give and we know mostly what they all mean. So this is a really good opportunity to start integrating this into understanding how signals and movement interact. And so this project is done in collaboration with the Kalahari Meerkat Project which is a long-term study in the Kalahari Desert in South Africa that's been ongoing for more than 20 years headed by Marta Manzer and Tim Cluttenbrook. And they monitor anywhere between six to over a dozen Meerkat groups at any one time. And what they do is they have volunteers out there on a daily basis following these groups around. And also recording behavioral data. So there's really a lot, a lot of background information on these individuals in these groups and in addition through a very painstaking process that takes many years they actually have also managed to habituate all of these groups so that you can actually stand there next to the Meerkats and they don't freak out which they would if they were wild or if they were not habituated. And what's also quite cool about this is that they also using a bit of water as enticement or a bit of boiled egg they can actually coax the Meerkats onto these portable scales and then they can weigh them multiple times a day to be able to get a measure of their foraging success for example. So basically a lot of information on the behavior of these animals. So this is an example of what Meerkats do throughout most of the day. Let's see if the audio works here. Yeah, well the audio is not super important for this video but so you have the Meerkats here and they spend most of the time digging in the sand for prey. They eat some invertebrates and small vertebrates and so they spend like a good portion of the day sort of with their heads down and this means that they can't constantly be looking around to coordinate with each other. And so in order to sort of keep track of where the other group mates are they use these little cohesion calls that you might be able to hear. You'll hear them again soon, don't worry. So they're just these little trilled calls that are like and they give these kind of every couple of seconds to help coordinate with each other while they're foraging. And so if you want to think about what are the movement rules that individuals are using that will ultimately scale up to allowing the group to decide where to go. If you think about it from the level of the individual Meerkat the Meerkat has to decide of course where it's going to move which it might do in relation to the position the relative positions of its group mates but not just where they are but also what they're saying. So what vocalizations they're producing and then at the same time that individual also has to be deciding itself what it's going to say and that can also be driven by both of these things. So what we have here is really this complex interplay between movement and vocalizations. And so the goal of this project is to try and pick apart this interplay and understand how individuals are making decisions about both of these levels movement and vocalizations and how these things interact to ultimately allow the group to come to consensus. And to do this we're using simultaneous GPS and audio recordings from a whole group whole groups of Meerkats. And so here's our small collars that we put on the Meerkats. They weigh 24 grams which is about the size of a weight of a AA battery. And they record high resolution GPS at one fix per second as well as continuous audio. And so as I mentioned before these Meerkats are very habituated and if you use the most habituated group that they have there at the site and if they're relaxed and sunning outside their burrow in the morning you can actually come up to them and sort of tickle them around the neck a little bit groom them and then very carefully put the collars on. If they're in the mood for it. And so using this technique and especially with the help of Chinni Gal who is an expert Meerkat collarer we managed to collar a whole group. In this case it was a relatively small group of seven individuals. And so here are some of the data that come off of the collars on the audio side. So you can hear kind of a lot of scraping sound which is them digging and then you also hear occasionally these trill vocalizations. Those are those cohesion calls that I mentioned before. And then occasionally that's what you get most of the time and then occasionally you also get some more exciting stuff. So these more sort of longer tonal calls here are move calls which are related to initiating rapid group movement. And so then what we can do is basically put these things together. And so here I'm showing you each dot represents a Meerkat and if you can see there's also a bunch of small dots or small markers that are coming up around these dots and these represent the vocalizations that that Meerkat is giving. So there's a lot of these close calls that are coming up as gray circles that I mentioned before. And then you can also see there's a lot of complexity of other different types of calls that they are giving. And so there for example was an alarm call from this one and then they all ran over there probably to a bolt hole to hide. And so there's really of course a lot going on here and one big challenge with this data set is to take go from the raw audio data to some meaningful vocalizations. So I'm trying to use machine learning to identify the calls automatically but this is still in progress. So what I can show you now is really just some preliminary results without having all of the data there yet. So just to do a very simple analysis here imagine that you have two Meerkats. This one is the Approacher and this one is the Approachee. And we just look at how often does this Meerkat, what percentage of the time does this Meerkat move more towards that one than away from it as a function of the initial distance between them. And so what we find if we just look across all the data is that you get this kind of curve shape where in general this is above 50%. So they do tend to attract each other which of course they have to otherwise they wouldn't stay together as a group. And the farther away you are the more likely you are to approach. So this shape of this curve is not necessarily that interesting per se but what I think is more interesting is to look at how different things can affect the shape of this curve. For example, the identity of the Approacher and the Approachee and also the calling behavior of the Approachee. And so looking at the identity what we find here is that there's actually quite a lot of differences between individuals and how often they are approached. So this red curve up here is the dominant female who in the group and she seems to have the highest level the highest percentage of being approached relative to the others. And if we look at the approach now subset by the Approacher we find that the dominant female as well as to a lesser extent the dominant male are also less likely to approach others within the group. And now looking at the calling behavior we might predict that if these cohesion calls are functioning as expected that if you have a high call rate this curve will kind of shift up whereas if you have a low call rate the curve will shift down. And so what we find is this does seem to mostly be the case where if you have a high call rate the percentage of approach goes up as compared to a low call rate but interestingly there's one individual for whom this is not the case and that is the dominant female. Actually with her when you have a high call rate you actually the percentage of approach goes down relative to the low call rate. And as I said this is quite preliminary but I think one thing that may be going on here is that if you look at the overall frequency with which individuals in the group give cohesion calls the dominant female really does not give very many relative to the others. Whereas if you look at the frequency with which they're giving these move and lead calls which as I mentioned before are initiating rapid group travel you see that the dominant female gives way way more of these calls. So this is of course still quite preliminary but what I think we're already starting to get a sense of is that there's something particular going on with the dominant female here. She's more likely to be followed by others. She's less likely to follow others and she's giving a lot of these move and lead calls that are associated with initiating rapid movement. So this to me suggests that we're more likely to be following on the kind of unshared side of the decision making spectrum at least for this preliminary data set. And so in the future I'm gonna look at a lot more of the things with this data in particular trying to look at do some modeling of what predicts where Meerkat moves. And I'm also particularly interested in the flexibility of individual influence and whether individuals can alter their influence by changing their call rate. Okay, so I'm just gonna tell you very briefly about one more project if I have time. Okay, so this is a project that's in collaboration with Kay Holkamp at MSU along with two collaborators Andy Gersik and Franz Jensen looking at long-distance communication and collective defense in a dispersed society of spotted hyenas. So hyenas in contrast to the oops, in contrast to the two examples that I just gave of the baboons and the Meerkats, they don't stay together as a group. So even though they live in stable clans where the individuals will be a member of the same plan over many years and they'll all interact with each other at some point, what they do is that they tend to split up and form small subgroups that split and merge over time. This is known as a fission fusion social structure. And they live in large cooperatively protected communal plant territories. And one thing that really characterizes these groups is that they have a lot of competition with both with neighboring clans, but also with lions. So they can have these instances where they will compete with lions over prey items. So this is an example of one of these competitions that I did a video taking for my collaborator, Andy Gersig. And you can see here that there's a lot of vocalizations going on in this process. And here you'll see the lions. So it seems that the lions are kind of mostly winning in this case here because they're over there with the carcass. But in general, if you have one lion versus one hyena, the lion is definitely going to win. But what often is the case is that the hyenas often have the numbers advantage compared to the lions. So this is how they can effectively compete. So this sets up a bit of a collective action problem for them. And one challenge is that even though the clans are relatively large, the subgroup size where they're actually moving around is relatively small, usually between one and three individuals. And so to solve this sort of, or they have this sort of fission fusion dilemma where within group competition tends to drive them to hunt alone, but then extra clan competition tends to drive them together to defend resources. And so the way that they solve this fission fusion dilemma is by making use of these long distance recruitment calls. So this is some work from my collaborator, Andy Gersig. So if you've ever been to the African Savannah, you've undoubtedly heard these calls. They carry very long distances and you can hear them very spookily at night. And what these calls do is that they, if an individual gives a call, other individuals may hear it and come to that location. And this is the way that they managed to recruit others to engage in collective action. And so in order to understand both this resolution of this fission fusion dilemma as well as the dynamics of information transfer during these recruitment call events, I've been collaborating as I mentioned before with the Masaimara hyena project with K-Hole Camp. And this is a long-term study of spotted hyenas that has been going on for over 30 years now where they have lots of behavioral and demographic information on these groups of hyenas. And so what we're doing now is putting high resolution GPS and audio callers on these hyenas as well as accelerometer and some other bells and whistles that we can manage to do because the hyenas are a bit heavier than the meerkats so they can carry more. And this is designed by Mark Johnson and Franz Jensen, these callers. And the ultimate goal is to be able to put these callers on everyone in the clan. But for the moment what we've done is just a pilot phase where we managed to collar six adult females, especially due to the efforts of Benson, who is a field assistant who is a real expert at darting and collaring hyenas. And then so these callers recorded data for about a month and a half and then afterwards they have an automatic release so that you don't have to dart the hyena again. And so what we get out of this is first just long-term data on where all of the hyenas are moving. So here is the communal den and you can see them moving throughout the evening and mostly actually resting during the day. So we can use this to potentially look at fission fusion dynamics once we have more individuals collared. And what we can also do is some manipulative, so in addition to also observing these long-range recruitment of earth events as they happen naturally, we can also do some manipulative experiments where we simulate threats by playing back whoop calls to resting hyenas and looking at their reactions. And this type of a playback experiment is a very common way of addressing decision-making in these animal groups. And usually what you can do is you can then look at the short-term response. Basically however long you can follow the hyena for in your car is how long you can see the response. But what's kind of cool about having these callers on at the same time is that you can actually also look at the long-term response. So here's the short-term response. This is sitting here in the car, this video doesn't have any sound, but there's a speaker kind of up this way and this hyena is sitting here under the tree, under the shrub and it hears the call played back from one of its clan members, one of those whoop calls and you can see she kind of walks off in that direction. And if you look carefully in the upper right hand corner you'll see another hyena come in from the side just about now, there. And that was another one that just happened to be sitting somewhere nearby who also heard the call and came running. And as I mentioned, this is the short-term response but we can also look at the long-term response. So here is a video where this black box represents the speaker and these are the two hyenas you saw before. And this blue one was the one we were watching. So she ran, sort of went towards the box and then eventually kind of lost interest and goes back to her resting spot. But this orange one up here, she actually got really excited and went for a huge journey. This is actually near the clan border. So she's potentially having a longer-term response to this stimulus. And so to kind of wrap up here, one thing that I'm really trying to do with all of these studies is to understand the interplay between collective movement with social structure and communication in animal societies. And I think one thing that really characterizes all of these studies is that they require a lot of people and a lot of interdisciplinary and collaborative approach. And so with that in mind, I think this is a really good way to go forward in understanding these systems. And with that, I will end and say thank you again to my collaborators all listed here. And I will take questions. Thank you.