Chamaeleolis, the un-anole anole. Does this evolutionary one-off demonstrate that island anole radiations are not convergent? Photo by Veronika Holanova and more on this snail crunching anole in a previous post.
Stephen Jay Gould famously proclaimed that if the tape of life were rewound and played again, a different evolutionary outcome would result. He argued that chances events–a lightning strike, a particular mutation–would send evolution careening down a different path. Gould’s idea has been controversial and provocative, but also untestable beyond the confines of the laboratory (you actually can replay the tape of life with microbes, but that’s another story).
The next closest possibility is to look at closely related species evolving in similar environmental settings. These “natural experiments” are in some ways a test of Gould’s hypothesis–will evolution from a similar starting point, in similar environments, unfold in similar ways? Gould would have said no, but what do the lizards say?
The evolutionary radiations of Anolis lizards on the islands of the Greater Antilles are renowned for their convergence, with the same set of “ecomorphs” (i.e., habitat specialists) evolving repeatedly on each island. The existence of the ecomorphs would seem to indicate that, in fact, evolution is deterministic–put an ancestral anole on a Caribbean island and let it evolve, and you get the same ecomorphs each time. But there’s a catch, two of them, actually. First, not all ecomorphs occur on all four Greater Antillean islands. Grass-bush anoles are missing from Jamaica and trunk anoles from both Jamaica and Puerto Rico. Moreover, second, there are some habitat specialists that are unique to a single island, with no ecomorphological counterpart on the other islands. The leaf-litter dwelling Anolis (Chamaelinorops) barbouri from Hispaniola or the Chamaeleolis clade–snail-crunching chameleon döppelgangers–from Cuba, pictured above. These quite distinctive habitat specialists–and a number of others–are evolutionary one-offs, with no convergent counterparts elsewhere.
So, convergence isn’t complete across the Greater Antilles. And that raises the question of whether the island radiations are actually replicated or not. Sure, there are some particular instances of convergence, but it’s always possible that these are instances embedded in a larger sea of non-convergence. Is island radiation truly deterministic? Are the evolutionary outcomes on the four islands actually more similar than one might expect by chance?
Average distance of species to their nearest neighbor on another island (gray line) is considerably less than would be expected by chance (gray blocks representing the results of many simulation trials; see papers for details on how these simulations were conducted).
A paper just published by Luke Mahler et al. in Science asks exactly that question (disclosure: I’m one of the authors). The paper took two approaches, using data on the morphology of each species (e.g., limb length, head length, number of toepad scales). First, the morphological similarity of each species to all others was quantified, based on these measurements. In this way, the morphological distance (in a statistical sense) to the nearest species on another island could be calculated. We then asked whether species had a nearest neighbor on another island that was closer (more similar) than would be expected by chance. The result is clear, and illustrated by the figure above. The average distance to nearest neighbor is much smaller than expected by chance. In other words, species tend to have convergent counterparts on other islands.
The number of convergent shifts to the same adaptive peak is substantially greater than expected by chance.
The second approach used a new method developed by Travis Ingram and Luke Mahler which estimates the underlying macroevolutionary landscape. To make a long story short, based on phylogenetic relationships of species and their morphology, the method identifies the minimum number of adaptive peaks across all islands. Convergent evolution would be indicated by unrelated species evolving to occupy the same peaks. The results indicate that there is far more convergence than expected by chance. This result indicates that the underlying macroevolutionary landscape on the islands is similar on the four islands, thus driving evolutionary diversification to occur in similar ways. The idea of the macroevolutionary landscape traces back to famed paleontologist George Gaylord Simpson, and this study is perhaps the first to explicitly test the idea that similarity in the macroevolutionary landscape is what leads to replicated adaptive radiations.
So, Gould seems to have gotten it wrong–let anoles diversify in similar settings, and the outcome will be very similar. Not identical–perhaps revealing the importance of his so-beloved historical contingencies–but pretty darn close, much more so than one would expect by chance. Now, in Gould’s defense, there is reason to believe (because he once told a reporter so) that he didn’t mean his tape rewinding metaphor to apply to the minor dabblings of insignificant little lizards; rather,he was concerned about grand patterns, deep in evolutionary time, when evolution was much less constrained and life was still exploring vast horizons and testing its limits.
Regardless, there is one last Gouldian twist. Anoles have been diversifying on these islands for tens of millions of years. So, the fact that the placement of the adaptive peaks is so similar suggests that there is a long-term stability to the macroevolutionary landscape. Such a long term arrangement would lead to lack of evolutionary change. As we all know, evolutionary stasis was another of Gould’s favorite ideas, so it would seem that SJG went one-for-two with anoles, a .500 average good enough for the all-star team in many leagues.
If you want to more on this paper, there’s a nice article/interview of Luke Mahler by Ed Yong on the Phenomena: Not Exactly Rocket Science website.
Recently, we had a post on the 
