I was heavily influenced by a handful of papers that were published during my first few years of graduate school. Some of these — Fine et al. (2004) on how herbivores promote habitat specialization in trees, Rooney et al. (2006) on food-web structure and stability — resonated because I could connect them to problems that I was working on. Others, such as Schmitz et al. (2004) on the ecological importance of predator-avoidance behavior, made an impression because they seemed to herald seismic shifts in the outlook of community ecology. And then there was a set of papers that captivated me with their sheer elegance: beautifully designed and executed field experiments that inspired me and made me jealous.
A string of papers by Tom Schoener, Dave Spiller, and Jonathan Losos belonged to this last category. There was a new one each year, with titles like “Predator-induced behaviour shifts and natural selection in field-experimental lizard populations” (2004) and “Island biogeography of populations: an introduced species transforms survival patterns” (2005). These studies used tiny cays in the Bahamas as arenas for a simple yet powerful experiment. On some islands, the investigators had introduced top predators — curly-tailed lizards (Leiocephalus carinatus), which occurred naturally on larger islands just a few stone-throws away from the experimental islands. A major aim of this work was to understand how the introduced predators affected populations of brown anoles (Anolis sagrei), which were native to the experimental islands.
The results were dramatic. Curly-tailed lizards are stocky, ground-dwelling animals, and they devastated brown-anole populations. The brown anoles that survived did so by climbing into the vegetation, beyond the reach of the curly-tails, and this behavioral shift was associated with natural selection on hindlimb length — an evolutionary consequence of predator-avoidance behavior. When I was a kid, we used to play a game called ‘the floor is lava’; if your feet touched the ground, you were dead. It seemed kind of like that for the brown anoles on these islands.
When I started a post-doc at the Harvard Society of Fellows in 2009, I met Losos and we started discussing ideas for a new experiment. I thought that a minor innovation on the earlier experiments could open up new conceptual territory. Losos said that he’d been wanting to do the same thing for years. To wit: if we introduced not just curly-tailed lizards, but also a second species of anole — green anoles (Anolis smaragdinus) — then we could ask questions about predation, competition, and the interaction between the two. Among other things, this design would enable us to test classic ideas about how predators affect the ability of competing prey species to coexist.
The risky thing about this idea was that so much of it had already been done to one degree or another. Previous work had painted a rich picture of the interaction between curly-tailed lizards and brown anoles — our odds of discovering something new on that front were low. And there were dozens of studies about competition between sympatric Anolis lizards. The novelty of our approach hinged on the interaction between predation and competition, which was a thin thread on which to hang such a massive undertaking. But I felt supremely confident that the experiment would work. Todd Palmer, Rowan Barrett, and a raft of other collaborators must have been confident too, because they joined me in setting up and monitoring the experiment.
By 2013, two full years into the study, my confidence was giving way to panic. I had started a job as an assistant professor in 2012; I was anxious about my professional survival, and I had ploughed large amounts of time and money into an experiment that did not seem to be working after all. The introduced curly-tailed lizards were firmly established in their new homes, and the brown anoles were responding by becoming more arboreal, as previous work had indicated they would. But the introduced green-anole populations seemed to be struggling. It looked as if they might die out, in which case our experiment would amount to a very expensive confirmation of the earlier work by Schoener, Spiller, and Losos. Our project had some original twists — Tyler Kartzinel was spearheading an effort to monitor the lizards’ diets using DNA metabarcoding — but it wasn’t at all clear that we would discover anything new or noteworthy.
Our break came in 2014, when it became clear that the green-anole populations were indeed thriving on some islands — just not on any of the islands with curly-tailed lizards. When we returned to the Bahamas in 2015, buoyed by the emerging results and freshly funded by the US National Science Foundation, we found that green anoles had disappeared on one island with curly-tailed lizards (the largest such island). Sometime during 2016, a second green-anole population vanished, this time from the smallest island with curly-tailed lizards. That left just two islands where green anoles still persisted in the presence of curly-tails, and one of those populations looked like it might soon join the list of casualties. Was this because the curly-tailed lizards were simply eating the green anoles to extinction on those islands? Probably not. The green anoles were highly arboreal; they rarely descended to the ground and instead moved by scampering through the twigs and leaves in the canopy. The chunky curly-tailed lizards, by contrast, lumbered across the ground, rarely climbing higher than 50 centimeters — and then only on the thickest of tree trunks. Indeed, the curly-tails didn’t seem to be eating many lizards of any kind. We saw them feasting on cockroaches, and occasionally snacking on fallen fruits and dead hermit crabs, but it wasn’t until 2016 that we finally saw one eat a small female brown anole. Isotopic analysis revealed that curly-tailed lizards actually occupied a slightly lower trophic position than did either anole species, which suggested that the top predator was subsisting more on insects than on other lizards.
The more plausible explanation for our results was that the presence of curly-tailed lizards intensified competition between the two anole species within the predator-free arboreal refuges, and that this competition — not direct predation — was the primary reason why the introduced green-anole populations failed to increase on islands with curly-tailed lizards. Molecular analysis of fecal samples subsequently reinforced this impression. DNA metabarcoding produced evidence that curly-tailed lizards exacerbated the competition between brown and green anoles for insect prey. And a quantitative PCR assay — conducted by Charles Xu at the behest of one of the four very thoughtful reviewers for Nature — detected the DNA of brown and green anoles in just 4% of the curly-tailed lizards that we sampled. Curly-tailed lizards really were the top predators; they just didn’t catch anoles very often.
We concluded that indirect effects of the top predator destabilized the coexistence of competing prey species. In the landscapes of fear created by curly-tailed lizards, the clear niche partitioning exhibited by brown and green anoles on predator-free islands — a product of adaptive radiation — was no longer evident. Instead, these species were trapped together in the top story of the small islands, competing for the same space and food, afraid of getting burned by the hungry predators on the ground. Green anoles, despite being better adapted to arboreal life, got the shorter end of the stick (both literally and figuratively). This might be because brown anoles, as the incumbents on the islands, had greater strength in numbers. If we had introduced both brown and green anoles at identical starting numbers, would the green anoles have come out on top? Or, now that the combination of competition and predation has greatly diminished brown-anole populations, might green anoles stage a comeback? In 2018, we reintroduced green anoles on the two islands where they had been extirpated, with the hope of answering this question.
In any event, our findings ran counter to one of the motivating hypotheses of the project. Early studies, notably Bob Paine’s classic experiment in the rocky intertidal habitats of Makah Bay, suggested that predators tend to ameliorate competition between species at lower trophic levels by preventing any one species from becoming too abundant and excluding the others. Many ecologists, myself included, love this idea of ‘keystone predation’. Not only is it an elegant concept, but it also validates top predators as linchpins of ecological integrity. But when can we expect predators to play this role? In rocky intertidal communities, where keystone predation is a powerful force, sea stars feed on sessile invertebrates; but prey that are attached to the substrate have a limited ability to escape predators in space. In predator-prey interactions involving fast-moving prey that can rapidly adjust their behavior to avoid predators, I would expect keystone predation (sensu stricto, as opposed to the broader concept of ‘keystone species’) to be infrequent, and competition for enemy-free space to be both frequent and strong.
It has now been almost a decade from the conception to the publication of this work. What started out as a post-doc project has become an enduring annual ritual, and one that I have (usually) been able to enjoy thanks to a talented group of collaborator-friends: Palmer, Barrett, Kartzinel, and Xu, along with Tim Thurman, Kena Fox-Dobbs, Matt Hutchinson, Tyler Coverdale, Josh Daskin, Dominic Evangelista, Kiyoko Gotanda, Naomi Man in ‘t Veld, Hanna Wegener, and Jason Kolbe — and, of course, Schoener, Spiller, and Losos.
The interdisciplinarity of this team enabled what is to me the most satisfying feature of our work. We were fortunate to have access to a replicated set of small cays on which to manipulate species composition. That is a rare opportunity and would have made for a nice study in itself. But by also integrating molecular assays (to quantify diet composition and intraguild predation) and stable-isotope analyses (to quantify trophic position and food-chain length), we were able to gain deeper insight into the mechanisms underlying the population dynamics. Indeed, without these additional assays, our suppositions about the relative importance of consumptive and non-consumptive effects would have been equivocal at best. Molecular techniques have fully entered the mainstream of ecology over the past decade, yet they are still rarely paired with the kind of manipulative field experiments that so inspired me as a first-year graduate student. The fusion of sound natural history, rigorous experimentation, and forensic mechanistic exploration offers tremendous power to resolve the kind of messy complexity that has long frustrated ecologists.