Some of our field research activities and locations, including the French Polynesia, Thailand, and the Mexican Caribbean – all marine work shown here, but we’re open to all study systems! :).


How does environmental change shape ecosystems? Our fundamental understanding of nature and our ability to sustain critical ecosystem services rest upon the answer to this question. Growing evidence suggests that the answer may hinge on animal behavior. Animals can rapidly shift their behavior in response to environmental change and the actions of others, at timescales far shorter than demographic timescales. But when even routine behaviors, such as foraging, are correlated across individuals in space or time, they can drive species interaction rates that govern the structure and function of the greater system. I develop high-throughput data collection systems in the field to rigorously and objectively identify animal behaviors and their drivers, and I combine these methods with statistical, analytical, and computational models to produce data-driven answers to the question: how does animal behavior shape ecological responses to environmental change? Under this overarching question, I ask:

1) How do spatial patterns of disturbance affect the behavior and ecological role of organisms?

2) How do social interactions among organisms affect ecological responses to environmental change?

1) How do spatial patterns of disturbance affect the behavior and ecological role of organisms?

Natural and anthropogenic environmental changes can drastically alter the structure and function of ecosystems. However, the ecological consequences of such disturbances can depend acutely on how their patterns of occurrence in space interact with consumer behavior, a relationship I explored in my dissertation research. Nutrient enrichment, often fueled by pollution, affects coastal ecosystems worldwide by stimulating weedy algae that can outcompete foundation species (e.g., corals, seagrasses). Experiments at small spatial scales, including ones I have led, have given rise to the paradigm that herbivorous fishes, when not overharvested, can prevent harmful ecological effects of enrichment, by eating away the excess biomass of weedy algae (e.g., Gil et al., Coral Reefs 2016). However, in an analytical modeling study, my colleagues and I found that the ability of herbivores to mitigate effects of enrichment can be explained by their movement behavior; herbivores can easily move into small enriched areas (i.e., experimental scales) and control primary producers, but they fail to do so at larger scales (e.g., widespread pollution from runoff; Gil et al., Oecologia 2016). This finding suggests that the scale of disturbance and the behavior of consumers can interact to determine ecosystem states and their resilience.

Remotely spying on the seafloor in coral reefs doesn’t just reveal powerful insights about ecologically fundamental fish behaviors, but you also get see exciting sides of wilderness when humans aren’t around. (Videos collected as part of Gil et al. Ecology 2017)

Due to natural and anthropogenic disturbances, habitat used as refuge by consumer species can exhibit extensive heterogeneity within consumer home ranges. How such ‘landscapes of safety’ affect consumer function remains an open question that can inform our understanding of how animal behaviors shape trophic ecology and can be used to enhance environmental forecasting. Using field experiments video recorded by camera arrays, my colleagues and I tested for interactive effects of the density of algae and isolation from refuge habitat (coral structures) on control of algae by a fish community in situ (Gil et al. Ecology 2017). We fitted a stochastic death process to time series of algal loss, which revealed that isolation from coral structures weakened herbivory, but less so for high-density patches of algae. These findings suggest that reef habitat loss fuels perceived risk, which inhibits consumers from controlling weedy algae, but that perceived reward can partially counteract this effect. In addition, 120 hrs of video of >7,000 bites of algae revealed that individual fish can tune their behavior to a dynamic landscape, consuming resources at higher rates in riskier (sensu ‘hazardous duty pay’) or more rewarding patches. We also saw that fish that formed groups tolerated greater risk. This may drive spatial complementarity in consumer function, which may generally respond to the largely unexplored effects of the social landscape.

2) How do social interactions of organisms affect ecological responses to environmental change?

Like human societies, communities of wildlife function as networks of individuals, linked by flows of information. Through actions as routine as feeding or fleeing from a predator, organisms produce ‘social information’ that can alert nearby individuals to resources or danger. My colleagues and I recently used dynamic state variable modeling to isolate the effects of social information from effects inherent to group formation. Our work revealed that social information, alone, incentivizes grouping and provides significant fitness advantages across energetic states, environmental contexts, and species, as long as individuals share one or more needs (Gil et al. American Naturalist 2017). We further found that information sharing often favors the grouping of different species that share predators. These findings suggest that social information could generally influence the function and demography of species and, thus, highlight a pressing unanswered question: how does social information affect ecology?

Using simple PVC scaffolding and trusty cinderblocks, we construct large-scale remote observatories, within which we hoist arrays of video cameras that allow us to record precisely how aggregations of ecologically critical herbivorous fishes behave in coral reefs. (Videos from experiments that led to Gil and Hein PNAS 2017, Gil et al., PNAS 2020)

I developed novel video camera arrays in situ to record the behavior of many individuals over ecologically relevant spatial scales. In a recent experiment, we used this approach to record >4,000 events, in which reef fish enter or exit foraging grounds rich in resources but exposed to predators. Using a novel stochastic process to analyze our data, we revealed that social information can determine the ecological function of fish and the flow of materials and energy through ecosystems (Gil & Hein, PNAS 2017). In particular, we showed that non-schooling fish from multiple species follow one another into foraging grounds and stay there longer, eating more algae, when there are more fish around. In turn, information-mediated positive density dependence could account for over 60% of the herbivory of the fish community. This work connects widespread effects of social information to fundamental ecological rates and motivates my new research on how animal behavior scales up to affect ecosystems.

My colleagues and I have since further examined both the mechanism and consequences of these findings. With the same setup as above, we revealed that behavioral rules, conserved across reef fish species, drive these patterns: using computer vision to reconstruct the visual fields of fish in situ, we found that fish flee less when they observe more fish between themselves and a perceived threat, and they make this assessment instantaneously (in less than half a second) (Hein et al. PNAS 2018). We also used a literature synthesis and theoretical models to evaluate how social information can drive population and community function, dynamics, coexistence, and extinction (Gil et al. TREE 2018; Gil et al. Ecology 2019; Gil et al., PNAS 2020). Our findings demonstrate that combining high-throughput data collection with analytical and computational tools can reveal animal behaviors and quantify their contribution to ecological dynamics.

Tracking what fish do and see in a coral reef using a ray-casting algorithm (originally developed for classic video games, like Wolfenstein): we used an underwater tablet computer to present a looming stimulus to fish (shown with red rays projected onto the fish’s retina) that can get information from surrounding fish (white rays), and were able to reverse engineer (describe with simple math) their fundamental decisions to feed or flee. (Videos from supplemental material in Hein et al. PNAS 2018)

The future

One major focus of my lab in the near term will be to merge questions 1 and 2 to examine the role that social information plays in determining the relationship between biodiversity and ecosystem function. Specifically, we will measure the degree to which effects of social information on ecological function: 1) depend on the identity, stage, and state of focal individuals, as well as the density, diversity, and species composition of surrounding individuals that may differ extensively in their provision and use of social information, 2) depend on environmental change, including human impacts (e.g., harvest, habitat loss, pollution, warming), and 3) determine the function, dynamics and resilience of populations, communities, and ecosystems.

Our work on collecting ‘Big Data’ on animal behavior in the wild is expanding in exciting new directions, including AI-driven video processing (training computers to track the position and body posture of each recorded fish in the wild) and mobile video capture to track large fish aggregations. And there’s sooo much more on the horizon, now that we’ve started teaming up with tech startups in the computer vision and embedded sensor spaces.

I am developing new empirical and quantitative approaches to test the context dependence of ecological effects of animal social interactions and the generality of their evolutionary origins. I recently developed a mobile multi-camera system that allows us to track landscape-scale behavior of individual fish across social contexts and to map this behavior onto high-resolution 3D models (generated using photogrammetry) of entire study areas. We are also using machine learning architectures to automate the precise reconstruction of movements and behaviors of individual fish in groups in situ. These advances in ‘Big Data’ collection at landscape scales will motivate statistical models and simulations that allow us to rigorously characterize, for the first time, socially-mediated functional roles of species and how these roles shift in space and time.

Furthermore, I am developing evolutionary models to measure the generality of ecological effects of social information by identifying environmental contexts that select for social facilitation and its density dependence. My lab will use this modeling framework and meta-analysic methods to investigate the role of social facilitation in shaping and maintaining diversity, and, concordantly, we will measure the effect of phylogeny on the frequency and nature of social interactions in the wild.

In addition, our lab will build on previous and ongoing collaborative works on the somewhat tangential but nonetheless fascinating topic of marine plastic pollution (see Gil and Pfaller 2016, Pfaller and Gil 2016, Pfaller et al. 2020) to develop new empirical experiments and quantitative models to understand the ecological consequence thereof.

I am excited to collaborate with and mentor graduate and undergraduate students and postdocs in theoretical and empirical research. Our lab will address basic research questions with clear management links in coral reefs and in other systems (including those in Colorado). If you’re interested in graduate school, and some/all of this sounds exciting, learn more about how to join the Gil Lab here.

%d bloggers like this: