Monday, October 27, 2014

An invasive species drives rapid evolution in a native

Anolis carolinensis male, dewlapping. Photo by Ambika Kamath.

In 1956, W.L. Brown and E.O. Wilson proposed the following eco-evolutionary process: two closely-related species come into contact, interact strongly (usually over food and other resources), and thereby experience natural selection to diverge from one another--ecology influences evolution. Then, if such divergence resulted in sufficient resource partitioning, the species’ population dynamics would stabilize and the two (or more) species would coexist--evolution influences ecology.

They called this particular eco-evolutionary process character displacement. During my dissertation, I spent several years investigating character displacement in Anolis carolinensis, culminating in publication last week. Thus, E-E-E-E regular Andrew Hendry asked me to describe the study for Eco-Evo-Evo-Eco.

Anolis carolinensis, marked 47 with Sharpie during a capture-mark-recapture study. Photo by Todd Campbell.
Anolis carolinensis is the only anole native to the southeast United States. Adults are about 3-5cm in length, weigh 2-5 grams, eat arthropods, and are active during the day. Males are territorial and use a dewlap along with push ups and headbobs to communicate during territorial and mating interactions.  

The sister species of A. carolinensis, called A. porcatus and A. allisoni, live in multi-species anole assemblages in Cuba today. Likely because of interspecific interactions, they are restricted in their habitat-use to high parts of tree trunks and the tree canopy. Thus, when the ancestor of A. carolinensis, also from Cuba, arrived to the US 2-3 million years ago, it probably experienced ecological release: with no other anole competitors, it shifted its habitat use towards low parts of the tree trunk and the ground. At least, that's where we find it today.

In the late 1940s/early 1950s, Anolis sagrei arrived to south Florida, likely as a stowaway in agricultural and other pre-embargo trade shipments between the US and Cuba. Like A. carolinensis, this lizard is 3-5cm long, 3-6 grams heavy, active during the day, eats arthropods, and maintains territories. Moreover, like A. carolinensis in Florida, this species likes to use the ground and low parts of the tree trunks.

Thus, where they overlap, the two species are likely to interact strongly with one another, setting the stage for character displacement.

Shortly after its colonization, A. sagrei established and spread northwards quickly. Today, it is well into southern Georgia and has even founded populations in Louisiana, Texas, and Hawaii (such jump-dispersal from Florida/Georgia is likely facilitated by horticultural traffic - the best place to find A. sagrei anoles in Houston a few years ago was Home Depot's garden department).

Anolis sagrei male, dewlapping. Photo by Adam Algar.
With its colonization and spread throughout Florida, A. sagrei has become arguably the most abundant vertebrate by biomass in the state and it must therefore affect its close relative, A. carolinensis. My colleagues and I tested two predictions for how: one ecological prediction, and one evolutionary prediction.

Prediction 1: Ecology
By the time of our study, Collette (1961) and others had noted that A. carolinenis tended to perch higher in the canopy whenever it was in sympatry with A. sagrei in Florida. Collette predicted that A. sagrei was responsible for this habitat-use shift in A. carolinensis, but the definitive evidence remained elusive, as there were many alternative explanations. A field experiment was need, and in 1995, my colleague and co-author, Todd Campbell found the place to do it: Mosquito Lagoon.

Dredge-spoil islands in Mosquito Lagoon, viewed from Oak Hill, looking south towards Cape Canaveral and the Kennedy Space Center. Note the houseboat in the foreground for scale; larger islands are about a hectare in size. There are about 80 islands in this part of the lagoon. Photo by Todd Campbell.
In the 1950s, the US Army Corps of Engineers dredged a channel through the lagoons that line the Atlantic coast of the United States. This channel, the Intracoastal Waterway, was meant to provide a sheltered lane for shipping traffic. The dredging machines (nicknamed 'clinkers' for the sound they made) would suck material from the bottom of the lagoon and then deposit that material alongside the channel. As a byproduct (spandrel?) of dumping the dredge spoil material, an island was formed, and this process was repeated regularly every 50 meters or so for hundreds of kilometers of coastline. These islands provide a replicated system excellent for testing the effects of A. sagrei introduction on A. carolinensis.

It took until the late 1980s for A. sagrei to arrive to the mainland bordering Mosquito Lagoon, which is about halfway up the Atlantic coast of Florida. At that time, the dredge-spoil islands there had established plant communities (mostly cabbage palm, eastern red cedar, buttonwood, and mangroves) and supported multiple arthropod species. The islands also had A. carolinensis on them.

Todd wished to know what the demographic impact of A. sagrei introduction would be on those A. carolinensis island populations. In 1995, he picked three islands (one small, one medium, and one large) as experimental islands and three similar islands as controls (using a random blocked design). That May, he conducted capture-mark-recapture studies for A. carolinensis on all six islands. Then he introduced A. sagrei (collected from the nearby mainland) to the three experimental islands.* He followed the populations with capture-mark-recapture surveys for the rest of that summer as well as summers of 1996, 1997, and 1998 - a colossal effort.

At the same time, and crucial for our story here, Todd also monitored perch heights in A. carolinensis. Thus, he could ask, is there a perch height shift in A. carolinensis following A. sagrei invasion? Here's what he found.

Within a few months, A. carolinensis perched higher in the presence of A. sagrei. This provided compelling experimental evidence confirming Collette's prediction: that interactions with A. sagrei would drive a habitat use shift by A. carolinensis.

Prediction 2: Evolution
Collette also made a second prediction; he had observed that A. carolinensis, in regions with A. sagrei, had larger toepads with more specialized scales, called lamellae. Thus, Collette predicted that a habitat shift in A. carolinensis, driven by A. sagrei, would result in the evolution of larger toepads. (Anoles that have larger toepads with more lamellae are better at clinging to surfaces. Across the ~400 species of anoles, those species that live higher in the canopy tend to have larger toepads. Together, this suggests that it is adaptive to have larger toepads when living higher in the canopy.)

The right hindfoot of an A. carolinensis male. Note the expanded scales on the distal portion of the toes. These are the lamellae. Photo by Yoel Stuart.
In 2010, I went to Mosquito Lagoon to test this prediction. My plan, as proposed to my thesis committee, was to go back and compare A. carolinensis toepads on the experimental versus control islands from Todd's demographic study. This would be great experimental evidence for character displacement.

Unfortunately, this wasn't feasible, primarily because A. sagrei had naturally colonized Todd's control islands. In fact, my surveys in 2010 and surveys by Nathan Turnbough (at UT Knoxville) a few years prior showed that A. sagrei had reached all but five islands in the lagoon - more than 70 natural colonizations.

Being unable to use the experimental islands was a disappointment to be sure, but a nice long conversation between myself, Todd, and co-author Jonathan Losos convinced me that we weren't sunk yet. (We had this conversation in the field, which is I think the best place to plan such work, as you really do get a sense for the challenges you'll be facing).

We decided that though I couldn't look for evolution in a true experiment, I did have a replicated setting where I could compare A. carolinensis toepads on un-invaded islands to toepads on islands with the invader. Moreover, I could bound the invasions in time. In preparation for the 1995 study, Todd surveyed most of the islands in the lagoon for both species, finding that most islands had just A. carolinensis. My 2010 survey showed that these islands had been invaded sometime in the intervening 15 years, setting the amount of generations A. carolinensis could have been evolving with A. sagrei on those islands to approximately 20. It wasn't a true experiment, but this natural experiment was pretty darn close.

We collected our data as follows. My field help and I headed out in a small boat every morning, landing on an island as the sun was coming up. We walked through the islands slowly until we saw a lizard that was undisturbed by our presence. We noted the perch for perch height measurements and then tried to capture the lizard so that we could measure its toepads. To catch the lizards, we used extendable fishing poles with fishing-line lassos tied to the end--get that lasso around the head, give a little tug, and you had your anole. (The lizards are very light, so this doesn't injure them). In the afternoons, we returned to our lodging and collected toepad images using a flatbed document scanner available for purchase in any office supplies store, ramped up to 4800 dpi (see image above). During that process, the lizards were anesthetized; after scanning, we let them wake up and recover overnight, and then put them back where we caught them the next day. We often wondered if their friends believed their abduction stories.

Stuart (foreground) and Campbell (background) pursuing lizards on Hook Island. Note the fishing pole pointing upward from Stuart's hand. The lizard must have been high up, as the pole is nearly fully extended.
Once I had the perch height and toepad data, first, I wished to double check that A. carolinensis on invaded islands did indeed perch higher than on un-invaded islands. They did, confirming the experimental result.

Then I tested whether A. carolinensis from invaded islands had larger toepads with more lamellae. Consistent with prediction, they did, suggesting that A. carolinensis was adapting to its arboreal lifestyle!

As noted above, I knew that A. sagrei couldn't have gotten to the islands earlier than 15 years, or about 20 generations, before 2010. I calculated the rate of divergence in haldanes over those 20 generations, and found that populations on invaded islands were diverging from populations on un-invaded islands at about 0.08 standard deviations per generation for each trait. To put that in human perspective: the average height of the American male is about 5'9". If American male height were increasing at 0.08 standard deviations for 20 generations, the average American male would be 6'4", or the size of an NBA shooting guard (assuming basketball was still around). This divergence was substantial and fast.

However, because the evidence for toepad divergence was observational, there were several alternative hypotheses, other than evolution, that might explain it instead. In the interests of space, I'll just say that: (1) we used a common garden experiment to show that there is an evolved genetic component to the observed divergence; (2) we conducted random habitat surveys to show no appreciable differences in environment between invaded and un-invaded islands; and (3) we used RAD-seq data to show that the islands were evolving independently of one another, so that the observed divergence wasn't the result of ecological sorting. This was a huge amount of work to collapse to a single paragraph; one paragraph doesn't suffice to give enough credit to co-authors Graham Reynolds, Liam Revell, Paul Hohenlohe, and Jonathan Losos for all the work they did here.

Alright, we're nearing the end. Let me sum up. With experimental and comparative evidence, we showed that the arrival of A. sagrei results in a perch height shift in A. carolinensis, suggesting a strong interaction between the two species. We then ruled against many alternative hypotheses, allowing us to say confidently that this habitat use shift resulted in rapid, morphological evolution by A. carolinensis, likely as adaptation to maneuver better on small, thin, and slippery arboreal perches during feeding, mating, and anti-predator behaviors. The major step now, and a focus of my future work, will be to determine exactly what kind of interaction is happening between the two species. It's likely that they compete for food and space resources. They may also interact agonistically. Moreover, adult male A. sagrei will eat hatchlings of A. carolinensis, so perhaps there is also some intraguild predation at work here, not to mention the possibility of indirect interactions through shared predators and parasites.

Nevertheless, regardless of the nature of the interaction between the two species, the invasion of A. sagrei is driving the rapid evolution of character displacement by A. carolinensis.
The exhaust trail of the space shuttle Endeavour drifting in the skies over Mosquito Lagoon after lift-off on July 15, 2009 (STS-127). The shuttle returned on July 31, accompanied by its typical double sonic boom. Photo by Yoel Stuart.
Y.E. Stuart, T.S. Campbell, P.A. Hohenlohe, R.G. Reynolds, L.J. Revell, and J.B. Losos. 2014. Rapid evolution of a native species following invasion by a congener. Science 346: 463-466 
DOI: 10.1126/science.1257008

* Over the last week or so, I've had a number of folks inquire about the ethics of introducing invasive species. This is a very valid concern; the spread of invasive species should be limited as much as possible. In this case, however, by 1995, A. sagrei was already highly abundant on the mainland and was starting to get to the spoil islands (which, recall, are man-made and didn't exist 40 years prior). Our opinion was that the lizards were going to get to the rest of the islands eventually anyways, so we might as well learn something from the invasions in a controlled, experimental framework. In retrospect, that almost all the islands in the lagoons have A. sagrei on them today substantiates our reasoning. However, we were also lucky that there weren't any unintended consequences (unlikely as those seemed at the time)--I doubt permission from the local permitting agencies would be granted today.


  1. Hi Yoel, great work! Would you be willing to explain what you meant by "divergence wasn't the result of ecological sorting"? Doesn't the presence of both lizard species preclude competitive exclusion? This is fascinating work.


  2. Hi Kelvin,
    Sorry to take so long to find this comment. It's a good question. What we meant is that sagrei's spread to the islands wasn't biased in some way by differences in carolinensis morphology or habitat use across populations. In other words, maybe sagrei is on the islands it's on because carolinensis were perching higher and had larger toepads before the invasion on those very islands. Thus, sagrei and carolinensis were able to coexist there, but not on the islands with just carolinensis, where they perch lower. The genetic data support that each island population of carolinensis is evolving independently and observational/experimental data support that sagrei colonizes every island it reaches. This suggests that there weren't any pre-existing differences among carolinensis populations and that character displacement is evolving in situ on every island that sagrei is reaching, rather than being an outcome of ecological sorting or a single evolutionary event. Hope that helps. Thanks for the question and the kind words! - Yoel