Simpsonian ‘Evolution by Jumps’ in an Adaptive Radiation of Anolis Lizards

Simpsonian ‘Evolution by Jumps’ in an Adaptive Radiation of Anolis Lizards
Jonathan M. Eastman, Daniel Wegmann, Christoph Leuenberger, Luke J. Harmon
(Submitted on 18 May 2013)

In his highly influential view of evolution, G. G. Simpson hypothesized that clades of species evolve in adaptive zones, defined as collections of niches occupied by species with similar traits and patterns of habitat use. Simpson hypothesized that species enter new adaptive zones in one of three ways: extinction of competitor species, dispersal to a new geographic region, or the evolution of a key trait that allows species to exploit resources in a new way. However, direct tests of Simpson’s hypotheses for the entry into new adaptive zones remain elusive. Here we evaluate the fit of a Simpsonian model of jumps between adaptive zones to phylogenetic comparative data. We use a novel statistical approach to show that anoles, a well-studied adaptive radiation of Caribbean lizards, have evolved by a series of evolutionary jumps in trait evolution. Furthermore, as Simpson predicted, trait axes strongly tied to habitat specialization show jumps that correspond with the evolution of key traits and/or dispersal between islands in the Greater Antilles. We conclude that jumps are commonly associated with major adaptive shifts in the evolutionary radiation of anoles.

1 thought on “Simpsonian ‘Evolution by Jumps’ in an Adaptive Radiation of Anolis Lizards

  1. We enjoyed reading this paper. One of the interesting things to think about, as these models that include jumps become more and more popular, is the distinction between the various models that have been offered. As we see it, there are two emerging classes of models: 1) event-based models, e.g. jumps in trait values may occur at cladogenesis (Bokma 2008); and 2) event-free models, e.g. jumps in traits according to a stochastic process (Landis, Schraiber, and Liang 2013).

    The authors’ work falls into category 1: jumps co-occur with ecomorphological and geographical transitions. We believe their approach explores a new region of event-based jump models, which pairs wonderfully with data-rich systems such as Anolis lizards. Through model testing, they find ecomorphological transitions are better indicators for the phylogenetic locations of jumps in continuous trait evolution when compared with geographical transitions and unrestricted jump locales. Models of pure gradual evolution do not fit the data well. It is striking that geographical events explain the distribution of jumps so poorly, since colonization events are thought to stimulate rapid evolution due the selection pressures in the new environment or small founding population size.

    There’s no doubt morphology plays a central role in a species’ ecological interactions. For the Anolis example, the ecomorphological classes are correlated with the continuous traits of interest. On one hand, their event-based jump models effectively filter the possible jump locations in their free-jump model. Since the stochastic mapping occurs before the continuous trait evolution analysis, this could be nicely formalized as an informative prior distribution. On the other hand, we’re unsure whether it’s appropriate to treat the ecomorphotype evolution as independent of the continuous trait evolution when they are not (by definition). That is to say, we think this the general approach to this model is very clever, but fear using ecomorphological transitions as bins for jumps may introduce some circularity into the analysis. If this concerns the authors, it could be readily addressed via simulation.

    In general, their approach seems to work well, but we have some concerns. The prior distribution applied to the jumps captures the intuitive notion that jumps are rare (in assigning a substantial amount of mass to a jumpless model), without explicitly modeling a rate of jumps. However, allowing only one jump per branch seems to disregard the effect that branch length should have under any reasonable biological model: a long branch is likely to have had more speciation events and more environmental shifts. Along the same lines, it seems to us that longer branches are more likely to have at least one jump when compared to shorter branches. As we interpret the prior, the jumps occur on a branch chosen uniformly at random, which contradicts this intuition.

    On a more technical level, we would have liked to see the authors outline more clearly the MCMC procedure for adding and removing jumps from branches. It seems to require a reversible-jump move and those are notoriously easy to foul up. By including a description of the reversible jump move (or explaining why one isn’t necessary), the readers will be able to understand the algorithm better.

    Josh Schraiber
    Michael Landis


    Bokma, F. “Detection of “punctuated equilibrium” by Bayesian estimation of speciation and extinction rates, ancestral character states, and rates of anagenetic and cladogenetic evolution on a molecular phylogeny.” Evolution 62.11 (2008): 2718-2726.

    Landis*, MJ, Schraiber*, JG, and Liang, M. “Phylogenetic Analysis Using Lévy Processes: Finding Jumps in the Evolution of Continuous Traits.” Systematic Biology 62.2 (2013): 193-204.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s