The social network of coral reef fish: Q&A with ecologist Mike Gil

Unless you’re thinking of the colorful characters in the movie Finding Nemo, the first word you think of for coral reef fish is probably not “social.” But after spending a lot of hours underwater, ecologist Mike Gil discovered something unexpected: the coral reef fish he swam with, especially the large algae-munching species who ventured into dangerous open areas, seemed to be paying very close attention to their neighbors.

When Gil and his colleagues set up an array of video cameras to spy on these fish, they found that reef fish were actually taking cues on when to eat — and when to flee — by watching the other fish around them. If they were surrounded by lots of other fish, regardless of the species, they were also less likely to run from predators like sharks.

“Which we can relate to, as humans,” said Gil, who is now a chancellor’s postdoctoral fellow in the Department of Ecology and Evolutionary Biology at the University of Colorado Boulder. “When you go into a haunted house, it’s nice to go with your group of friends rather than to go solo.”

His latest research found something even more surprising. By simulating these reef ecosystems in statistical models, Gil discovered that social behavior makes reefs more sensitive to overfishing than previously thought. Even more fascinating, his models suggest that fish behavior makes reef ecosystems particularly sensitive to the pace of fishing — removing the same amount of fish overall, but over a longer period of time, was less likely to cause an ecosystem collapse.

His discovery could help communities that depend on reefs for food use them more sustainably.

“We live in a world where it would be ignorant to not acknowledge that we need to use these ecosystems. It would be convenient if that weren’t the case, but it’s not,” he said. “What I think we all want is to use these systems in a sustainable way — so that we don’t just get the immediate benefits they offer, but we ensure that they are going to provide them for generations to centuries to come.”

Mongabay reached Gil by phone to talk social fish, Lord of the Rings characters, and how computer models might be missing a key piece of what makes an ecosystem tick. The interview has been lightly edited for clarity.

Mongabay: How did you first become interested in studying the ocean?

Mike Gil: I was always interested in nature as a little kid. I loved watching the Discovery Channel and National Geographic documentaries, and I always thought I wanted to get involved in conservation in some capacity. But I actually never thought I would be a scientist. You could say I thought science was kind of boring and unrelatable as a kid. But that was just me not knowing what science actually was.

When I finally got the opportunity to experience science for myself, and learn it is this creative process, where you can ask whatever you want of the world, and then rigorously collect evidence to try and answer that question in as an objective a manner as possible — that sent me down the science road in a very powerful way. And I never really looked back.

When you were thinking about this career, I imagine you pictured yourself diving all day, spending most of your time in the water. Did you ever envision how big a role technology and computers would come to play?

Honestly, I had no idea what I was getting into, except that I loved the idea of just being able to come up with whatever questions I wanted, and then come up with how you answer them. I have been pleasantly surprised by just how much cutting-edge technology can inform how scientists — and I can speak to field biologists in particular — go about making plans for how to answer these questions nobody has been able to answer yet. It’s been really fun being a field biologist right now, during this confluence of incredible technological innovation, just readily available both in the form of remote data collection systems but also on the computational side.

That lets us basically code up really incredible mathematical models and simulations that let us make sense of our data, and then really extrapolate far beyond these short periods of time and short areas in space that we can observe. They help us really understand what nature is doing on a grander scale. On a scale that allows us to make informed decisions about how to better manage these systems.

Let’s talk about the experiment that led to your TED talk in 2017. How did you get this idea to first “spy” on reef fish?

I’m an ecologist; that’s my background, and that’s how I was trained as a graduate student. And in ecology, we do a lot of counting of animals and plants in nature, and we try to understand how different amounts of plants and animals, how populations of these organisms change over time. We tend to not pay that much attention to behavior. Individual animals’ decision-making, that’s often not a focus for an ecologist.

But I had spent so much time working underwater in coral reefs that I couldn’t help but notice that coral reef fish, that I would have assumed before are kind of stupid — they’re kind of just bumbling around and doing their thing — what I noticed was these fish, they seemed to be paying really close attention to one another.

The fish I was particularly interested in are these large-bodied herbivorous fish that eat algae. And these fish, we know, are really important, because the algae that these fish eat can kill coral if left unchecked. I noticed most of the species in the main ecosystem as a grad student, in French Polynesia, these fish don’t form schools. What they do instead is they seem to form these clumps, with lots of other species.

It made me think: these guys are paying a lot of attention to one another, even though we tend to think they’re not that sophisticated. What would happen if we could watch exactly what they’re doing over long periods of time and over large spatial scales? Could we understand what makes these guys tick, and a little more specifically: are they playing close attention to one another, are they influenced by one another’s behavior like we humans are?

So, I came up with this idea, to do something that seems almost boneheadedly obvious, which is: why don’t we don’t put a bunch of cameras on a giant frame, so that we could look from above and see everything these fish do on a moment-to-moment basis?

To be completely honest with you we didn’t really think this was going to work. It was a crazy idea and nobody had done anything like this, but we thought, let’s give it a shot. And we were able to get just incredible volumes of data.

Sort of like the social media revolution of today has produced unprecedented volumes of data on human behavior, and in part on human social behavior, and we have learned so much from that big data set — we can do something similar with nature if we can figure out ways to collect vast quantities of data, which is what this really odd camera contraption allowed us to do.

So, you had these cameras out there for a long period, gathering a lot of information. Tell me about the process of sorting through and processing all of the data that resulted.

There are multiple experiments we conducted with these same frames, which we affectionately refer to as fear frames. We were looking at fish that are inherently fearful because they’re feeding in open areas in reefs, which are accessible to predators like sharks. These open areas also get a ton of sunlight, and because these fish eat algae that are primary producers, they’re a great place to go to fill your belly. But they’re really open to predators.

We put the frames out on three different summer field seasons and they were out for months at a time. We would go out and do behavioral trials where we would video record with the cameras on the frames themselves usually for four-and-a-half-hour periods.

The downside to getting tons of data is that you have to figure out what to do with those data, right? We had tons of video, and then we had to figure out how we wanted to process it.

I actually brought together a team of undergrad students from the University of Florida, that was where I was doing my Ph.D., and the team along with myself and my collaborator Andrew Hein, we manually scored when exactly those fish enter and exit these foraging areas. We ended up with over 4,000 observations of fish entering and exiting, exactly when they happened, and then we know how many fish were around at any time.

That gave us a ton of data to basically turn fish behavior into math. We fit statistical models to the data, which is a fancy way to say, we tried to figure out what mathematical equation might predict when these fish decide they’re going to enter and when they’re going to exit. And obviously the incentive to enter is you gotta eat, the incentive to exit might be you don’t want to get eaten.

We tried many different types of models, and we did find one model that in part does a really nice job of predicting our actual data set.

The insight that we get is that fish tend to follow other fish when they enter these dangerous feeding areas, and fish tend to follow other fish when they exit these dangerous feeding areas. Which isn’t that shocking. As I said before, it seems like these fish are paying attention to one another. Lo and behold, these data point to the answer yes, they are.

The thing we did not expect and which ended up being really important, and drove a really critical pattern, was that when these fish were surrounded by more fish, they were less likely to leave the dangerous feeding area. And so it seems like these fish perceive safety in numbers — which we can relate to as humans. When you go into a haunted house, it’s nice to go with your group of friends rather than to go solo.

In a follow-up experiment, we actually perturbed the system by putting a threat stimulus on an iPad. It showed a video to fish of what we call a looming stimulus: an itty-bitty shape that gets really big quickly. So that mimics what it would look like for an object to come right at you. We found those same kinds of fish are far less likely to flee from a scary thing when they specifically see lots of fish between themselves and that scary thing.

In your more recent study using this data, you turned from diving on reefs to playing with a simulated reef on a computer. Why make that switch?

So, what we have now is this rich data set, and we found some cool behavior patterns, where it seems like fish essentially give each other the confidence to eat more in dangerous parts of coral reefs. On its face that’s a cool, interesting finding, but doesn’t tell us a heck of a lot about the ecosystem.

But what’s really cool is we have such a robust pattern from this data. So, we can take this simple pattern of fish following each other around, and use that pattern in mathematical simulations of coral reefs. They’re very similar to video games, and like video games you can play around with the ecosystem. You can impose different amounts of fish, you can change its predators, to see how the ecosystem then responds.

Have you seen The Lord of the Rings, with the tree Ents? They’re slow-growing, and they move very slowly. Nature in general is very similar to a tree Ent. It’s very different from observing a mouse in a lab. If you want to understand how ecosystems work, you have to widen your timescale in a way that’s not practical for humans to study.

But the tradition in these models is to assume that subtleties in the way that wild animals behave probably don’t matter very much — because of course, animal behavior happens very rapidly relative to the decades, centuries, or millennial timescales we’re simulating in these ecosystems. The classic thing to do is treat animal populations like ideal gases: that these animals are behaving and interacting with each other the same way gas particles randomly collide in space. This essentially precludes the decision-making, the subtleties and very simple patterns in animal behavior.

We took that incredibly rich behavioral oversight we gained and we included it, for the first time, in a simulation of an entire coral reef ecosystem. What we find is we could be missing a critical piece to the code that dictates nature.

Tell me more about that, and in particular how you were able to extrapolate to how this makes reefs sensitive to human activities.

What we specifically see is that when we allow fish to be the social creatures that our research has revealed them to be, the system becomes hypersensitive. You can cause the ecosystem to collapse with far less fishing than we would expect if we assumed these fish were randomly colliding in space and time.

We also showed something that’s never been shown before: that the social behavior makes the ecosystem sensitive to the precise pace that we change the ecosystem.

If we were to fish a coral reef at some target level — say, fish up to 20% of the fish biomass per year — if we approach that fishing level quickly, we can cause the entire fish population and the entire coral population to collapse. But if we instead reach that exact same target level but slower, and even in some cases very slightly slower, you can preserve the fish and coral population and the ecosystem at large.

And this, by the way, is in the absence of global climate change. That’s for future directions.

What is it about social behavior that specifically makes reefs so sensitive to this?

The hypersensitivity that we’re seeing is being caused by the fact that these fish don’t just eat algae. We know now that if you’re removing these fish, you’re also removing the social influence that that fish had on the other fish in its social network. You’re leaving the surviving fish with less information that they should be getting from the removed fish, that would tell them when and where to go out and eat this algae.

When you fish the system quickly, this feedback loop gets intensified.

It’s pretty simple in that when you remove fish from a coral reef, you’re reducing the density of fish. So, for a given area of fish there will be fewer fish around. When there’s fewer fish around, our experiments show each fish would eat less, because it has less cues from any given fish that it’s safe to go out in these areas to eat and control algae.

How can people — say, fisheries managers for countries with tropical reefs — use these results to make policies that are more reef safe?

I think what this tells us is that when we are thinking about specific fisheries management, we need to consider not just a target amount of harvest that we’re willing to allow the system to experience, but also the exact path to that target. Our findings suggest we should try to slowly ramp up to target levels of change.

Of course, this is a very coarse first step, and to model a specific coral reef for a government body you have to be much more specific with the parameters of the model. But our findings generally suggest we could make catastrophic mistakes with management policies if we focus, myopically, on only the magnitude of change.

And coral reefs are only one of many ecosystems we use a similar approach to study. In fact it’s possible in coral reefs and in other ecosystems far beyond reefs, and for environmental changes far beyond fishing pressure, that the pace of environmental change — and humans are causing all different changes around the world — could determine whether or not we will collapse or sustain ecosystems we have come to rely heavily on.

In my opinion, this work speaks really well to one of the misconceptions of environmental policy, which is that people think it will stop us from using natural resources altogether. When in fact, usually what we need to do is just do so a little more carefully. What are your thoughts on that?

One hundred percent. We live in a world where it would be ignorant to not acknowledge that we need to use these ecosystems. It would be convenient if that weren’t the case, but it’s not. We use coral reefs for all sorts of stuff. They provide fish that provide the primary protein source for billions of people on the planet. These ecosystems keep the economies of many developing countries going, through ecotourism. And if that’s not enough evidence that we depend on reefs, they pump hundreds of billions of U.S. dollars into the global economy annually. We have to use these systems, we can’t get around that.

What I think we all want is to use these systems in a sustainable way — so that we don’t just get the immediate benefits they offer, but we ensure that they are going to provide them for generations to centuries to come.

There is a lot to be said about coral reefs at this unprecedented moment in human history, where we’re dealing with global climate change at magnitudes and scales we haven’t experienced before. We know global climate change is the grandest threat that coral reefs face. So yeah, we need to use these systems and keep them around. I could not agree with that point more.

But global climate change poses an incredibly challenging question: how do you accomplish that sustainable use and management of these critical ecosystems, when we’re dealing with human-driven pressures at the global scale, and which requires global action to be mitigated?

That’s the elephant-in-the-room question that we’re going to have to face. We can create these mathematical simulations all day, and we will continue to reveal insights to the subtleties about how systems work. But at the end of the day, science is powerless to keep coral reefs sustainable. We need to use the insights we gain and implement them at a broad global scale.

With that in mind, what’s next? What more do you want to learn?

We are at the tip of the iceberg in our understanding of how incorporating decision-making by wild animals can change the way we understand ecosystems to work. I’m really excited about this work, because there are so many directions to expand into, and these bigger questions about how social networks interface with other environmental changes beyond fishing. That’s inclusive of global climate change, the warming of ocean, pollution, and nutrient pollution, which affects the algae that these fish eat.

There are so many ways we can not just build on the power of fish social networks to determine decision-making, but how that then combines with these other known environmental changes happening in coral reefs all around the planet.

My team and I, we have a lot of exciting stuff in the pipeline. We can teach computers to not only measure where fish are and what they see moment to moment, but we’re developing tools that will let us develop 3D simulations of how these fish move through and interact with the 3D structures of coral themselves. We can test how protection and safety can play a role in how these ecosystems are changing not only in time, but in space. Coral is essentially growing rocks that provide a little fortress, and we know that these growing rocks can get destroyed in space and in time — so it can be very variable.

And we will continue to develop machine learning. The times of having armies of incredible volunteer undergrads, those times are done. We have machines doing the heavy lifting so we can really do justice to the volumes of data we’re getting. We are truly in the golden age of big data, and I’m happy to report that that has filled in to field biology. So, we have this confluence of technological opportunities that offer unprecedented opportunities to truly understand what’s going on in the wild.