Introduction to Models of Continuous-Character Evolution

Overview of Standard Models of Continuous-Character Evolution

Michael R. May

Last modified on September 16, 2019

Overview: Models of Continuous-Character Evolution

In these tutorials we will focus on a few types of models that are commonly used to study how continuous characters evolve over a phylogeny. All of the models we describe are Gaussian models, meaning that character evolution follows a normal distribution, as in the Brownian motion and Ornstein-Uhlenbeck models. Models that include stochastic jumps (Landis and Schraiber 2017) are not covered here.

Models of continuous-character evolution are used to understand many interesting evolutionary phenomena, including: the rate of character evolution, the mode of character evolution, how these vary over time or among lineages, and understanding the biotic and abiotic factors that contribute to this variation.

Brownian Motion models

The simplest models of continuous-character evolution assume that the character evolves under a Brownian motion model (missing reference). Under this model, the expected amount of character change along a branch is zero, and the variance in character change is proportional to time and a rate parameter, $\sigma^2$. These tutorials focus on estimating the rate parameter, $\sigma^2$, as well as how it varies over time and among lineages.

(1) Estimating rates of evolution

What is the (global) rate of evolution for my continuous character?

The simplest Brownian-motion model assumes that the continuous character evolves at a constant rate over time and among lineages. While this model is limited in terms of the evolutionary questions it is able to address, it provides a useful introduction to modeling continuous-character evolution in RevBayes.

You can find examples and more information in the Simple Brownian Rate Estimation tutorial.

(2) Detecting variation in rates of evolution through time

Do rates of evolution vary over time?

Many hypotheses can naturally be addressed by asking whether rates of evolution vary over time, for example: are rates of evolution increased early in an adaptive radiation (Harmon et al. 2010), do rates of evolution increase over time (Blomberg et al. 2003), or do rates vary in less predictable ways?

Work in Progress

(3) Detecting variation in rates of evolution among lineages

Is there evidence for variation in the rate of evolution across the branches of my phylogeny?

Identifying the number, location, and magnitude of shifts in rates of continuous character evolution can illuminate many evolutionary questions. Relaxing the assumption that rates are constant requires specifying a model that describes how rates vary (a ‘‘relaxed morphological clock’’). Many such models have been proposed (Eastman et al. 2011; Venditti et al. 2011); we provide an example of relaxing the morphological clock using a ‘‘random local clock’’ model, as described in Eastman et al. (2011), in the Relaxed Brownian Rate Estimation tutorial.

(4) Detecting state-dependent rates of evolution

Are rates of evolution correlated with a discrete variable on my phylogeny?

If rates of evolution are found to vary across branches, a natural question to ask is if some focal variable is associated with the rate variation. For example, we can test whether changes in rates are associated with habitat type in reef and non-reef dwelling fishes (Price et al. 2013; May and Moore 2020).

We provide an example of fitting the state-dependent model to discrete- and continuous-character data in the State-Dependent Brownian Rate Estimation tutorial.

(5) Detecting correlated evolution

Are characters X and Y evolutionarily correlated?

Often we would like to know whether two (or more) continuous characters are correlated, and if they are, how strongly (Felsenstein 1985). Additionally, we may view correlations among continuous characters as nuisance parameters: perhaps we are interested in estimating how rates of evolution vary among lineages or over time from multiple characters, and are concerned that failing to model correlations among characters will mislead us (Adams et al. 2017).

In the Multivariate Brownian Motion tutorial, we will provide examples for working with multivariate Brownian motion models to test hypotheses about correlations, and to estimate the strength of correlations among many continuous characters.

Ornstein-Uhlenbeck models

Another major class of Gaussian models are the Ornstein-Uhlenbeck models, sometimes referred to as the Hansen models (Hansen 1997; Butler and King 2004). These models can describe the evolution of a continuous character under stabilizing selection. These tutorials focus on estimating the optimum parameter, $\theta$, as well as how it varies among lineages.

(1) Estimating evolutionary optima

What is the optimal value for the continuous character?

The simplest Ornstein-Uhlenbeck model assumes that the continuous character evolves at a constant rate, and is drawn toward an optimal value, $\theta$, that is assumed to be the same over time and across branches.

You can find examples and more information in the Simple Ornstein-Uhlenbeck Models tutorial.

(2) Detecting shifts in evolutionary optima

Have optimal phenotypes changed over evolutionary history?

Optimal traits may change as species evolve over the adaptive landscape. Relaxing the assumption that the optimal value is fixed across the tree requires specifying a model that describes how theta varies.

In the Relaxed Ornstein-Uhlenbeck Models tutorial, we provide an example of relaxing the assumption that optima are homogeneously across branches using a ‘‘random local clock’’ model, which is spiritually similar to the one described in Uyeda and Harmon (2014).

  1. Adams D.C., Korneisel D., Young M., Nistri A. 2017. Natural history constrains the macroevolution of foot morphology in European plethodontid salamanders. The American Naturalist. 190:292–297.
  2. Blomberg S.P., Garland Jr T., Ives A.R. 2003. Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution. 57:717–745.
  3. Butler M.A., King A.A. 2004. Phylogenetic comparative analysis: a modeling approach for adaptive evolution. The American Naturalist. 164:683–695.
  4. Eastman J.M., Alfaro M.E., Joyce P., Hipp A.L., Harmon L.J. 2011. A novel comparative method for identifying shifts in the rate of character evolution on trees. Evolution. 65:3578–3589.
  5. Felsenstein J. 1985. Phylogenies and the comparative method. The American Naturalist.:1–15. 10.1086/284325
  6. Hansen T.F. 1997. Stabilizing selection and the comparative analysis of adaptation. Evolution. 51:1341–1351.
  7. Harmon L.J., Losos J.B., Jonathan Davies T., Gillespie R.G., Gittleman J.L., Jennings B.W., Kozak K.H., McPeek M.A., Moreno-Roark F., Near T.J., Purvis A., Ricklefs R.E., Schluter D., Schulte II J.A., Seehausen O., Sidlauskas B.L., Torres-Carvajal O., Weir J.T., Mooers A.Ø. 2010. Early bursts of body size and shape evolution are rare in comparative data. Evolution. 64:2385–2396.
  8. Landis M.J., Schraiber J.G. 2017. Pulsed evolution shaped modern vertebrate body sizes. Proceedings of the National Academy of Sciences. 114:13224–13229. 10.1073/pnas.1710920114
  9. May M.R., Moore B.R. 2020. A bayesian approach for inferring the impact of a discrete character on rates of continuous-character evolution in the presence of background-rate variation. Systematic Biology. 69:530–544.
  10. Price S.A., Tavera J.J., Near T.J., Wainwright P.C. 2013. Elevated rates of morphological and functional diversification in reef-dwelling haemulid fishes. Evolution. 67:417–428.
  11. Uyeda J.C., Harmon L.J. 2014. A novel Bayesian method for inferring and interpreting the dynamics of adaptive landscapes from phylogenetic comparative data. Systematic Biology. 63:902–918.
  12. Venditti C., Meade A., Pagel M. 2011. Multiple routes to mammalian diversity. Nature. 479:393–396.