
C HAPTER 10 PHYLOGENETIC APPROACHES FOR RESEARCH IN COMPARATIVE COGNITION Evan L. MacLean and Charles L. Nunn Comparative studies have the potential to address be treated as independent observationsASSOCIATION in statisti- a wide range of questions about how, when, and cal analyses because of patterns of inheritance on a why different traits have evolved. The compara- bifurcating phylogeny (Felsenstein, 1985; Martins & tive approach refers to studies in which variation in Garland, 1991). To deal with this nonindependence, traits of different species (or different populations) comparative biologists incorporate information is used to test specifc hypotheses or to generate new about the evolutionary relationships between species. hypotheses about evolutionary phenomena. In many Phylogenetic comparative methods are a set of sta- cases, comparative methods are used to investigate tistical approaches designed for exactly this purpose how two or more traits covary. Comparative meth- (Garamszegi,PSYCHOLOGICAL 2014; Garland, Bennett, & Rezende, ods are also used to reconstruct evolutionary history 2005; Nunn, 2011; Rezende & Diniz-Filho, 2012), or to assess how traits infuence patterns of diver- a toolkit that is becoming increasingly useful in the sifcation (speciation and extinction; Nunn, 2011). feld of comparative psychology. In addition to being In the context of comparative psychology, these applied in a broader range of evolutionary contexts, methods have been used, for example, to investigate comparative methods themselves are evolving, and associations between aspects of neuroanatomyAMERICAN and many classical techniques are rapidly being replaced executive function (MacLean et al., 2014;© Shultz & by newer and more fexible approaches. Dunbar, 2010; see also Chapter 24, this volume), In this chapter, we introduce the reader to a to explore how life history traits infuence temporal range of phylogenetic comparative methods that decision making (Stevens, 2014; see also Volume 2, can be used to address fundamental questions in Chapter 24, this handbook),PROOFS and to make inferences comparative psychology, including some recent about how life in complex societies has shaped pri- methodological advances that will create powerful mate cognitive evolution (Amici, Aureli, & Call, opportunities for future research. We illustrate these 2008; Burkart et al., 2014; MacLean et al., 2013; see methods by using a combination of simulated data also Chapters 12 and 13, this volume). and analyses of published datasets. Given the rapid Understanding evolutionary relationships—or growth of research on comparative cognition, we phylogeny—among the species of interest is critical highlight the utility of comparative methods for the for effective inference in comparative research. In a study of cognitive evolution. However, the concepts UNCORRECTEDvery real sense, phylogeny is the scaffolding on which and statistical approaches described in this chapter one investigates evolutionary change in traits and the can be similarly applied in other areas of compara- factors that lead to these changes. This is important tive psychology. conceptually, and it is also critically relevant in statis- At the outset, we want to emphasize that tical analyses. For example, in the context of studying some traits are measured on a quasi-continuous how traits covary, data on different species cannot scale (e.g., percentage of correct responses on an http://dx.doi.org/10.1037/XXXXX-XXX APA Handbook of Comparative Psychology: Vol. 1. Basic Concepts, Methods, Neural Substrate, and Behavior, J. Call (Editor-in-Chief) 1 Copyright © 2017 by the American Psychological Association. All rights reserved. BK-APA-HCM_V1-160213-Chp10.indd 1 04/07/16 6:13 PM MacLean and Nunn experimental task), and other traits are measured terms and concepts that are foundational to all phy- on a discrete scale (e.g., presence or absence of mir- logenetic comparative methods. ror self-recognition). These two types of data often require different statistical approaches. For brevity, PHYLOGENETIC TREES we present a single variant of each method below, yet we also direct the reader to relevant literature on As just noted, phylogenies represent the evolutionary other approaches throughout and to several recent relationships between taxa and are frequently reviews (Garamszegi, 2014; Garland et al., 2005; visualized as trees with a branching pattern (see Nunn, 2011; Rezende & Diniz-Filho, 2012). We Figure 10.1A). Phylogenetic trees consist of nodes and begin with a brief introduction to additional key branches. Nodes indicate speciation events where an (A) ASSOCIATION PSYCHOLOGICAL (B) ABCDEFGH A41000000 B 1 4 AMERICAN000000 C 00© 4 32111 D 0034 2 111 E 0022 4 111 F PROOFS00111 4 32 G 001113 4 2 H 00111224 FIGURE 10.1. A phylogenetic tree and variance–covariance matrix representing the evolutionary relationships between species. A: The root of the tree is at the bottom with the tips (terminal branches) extend- UNCORRECTEDing upward. Uppercase letters refer to extent taxa at the tips of the tree. Nodes are indicated by lowercase letters enclosed in circles, and branch lengths are shown to the left of each branch in the tree. The scale bar indicates 1 million years (MY). B: The variance–covariance matrix repre- sents the total age of the phylogeny along the diagonal, with the extent of shared evolutionary history between pairs of species indicated on the off-diagonals. 2 BK-APA-HCM_V1-160213-Chp10.indd 2 04/07/16 6:13 PM Phylogenetic Approaches for Research in Comparative Cognition ancestral lineage gave rise to two (or more) descendent PHYLOGENETIC SIGNAL species. Nodes are connected by branches, which are typically drawn to be proportional to evolutionary As a result of shared evolutionary history, closely time. Figure 10.1A shows a phylogenetic tree for eight related species tend to resemble one another more so species with circles enclosing the internal nodes. Node than less closely related taxa; this tendency is termed r is located at the root of the phylogeny and repre- phylogenetic signal (Blomberg & Garland, 2002; sents the oldest bifurcation in the tree, which in this Blomberg, Garland, & Ives, 2003). Interestingly, the example occurred 4 million years ago. The internal extent to which traits are associated with phylogeny branches of the tree (i.e., branches connecting nodes, varies widely from one trait to another, with morpho- rather than ending at a tip) represent the time that logical traits tending to exhibit higher levels of phylo- species have shared evolutionary history, whereas the genetic signal than behavioral, cognitive, or ecological terminal branches (leading to a tip) represent the time variables (Blomberg et al., 2003; Kamilar & Cooper, that each extant species has evolved independently of 2013; MacLean et al., 2012). Quantitative estimates of phylogenetic signal in continuous traits can be other taxa in the tree (of course keeping in mind that ASSOCIATION there are often many extinct lineages that are not rep- obtained by using a variety of different approaches resented on a phylogeny of extant species; see Nunn, (for reviews, see Kamilar & Cooper, 2013; Münke- 2011). müller et al., 2012; Nunn, 2011). Here we focus on The extent of shared evolutionary history one commonly used metric, Pagel’s lambda (Freckle- between pairs of species in a phylogeny can be rep- ton et al., 2002; Pagel, 1999a). resented as a variance–covariance matrix (see Figure Lambda is a continuous parameter that ranges 10.1B; Cunningham, Omland, & Oakley, 1998; from 0 to 1, with a value of 0 indicating that trait Freckleton, Harvey, & Pagel, 2002). The diagonal covariances are independent of phylogeny and a of the matrix (gray background) represents the vari- valuePSYCHOLOGICAL of 1 indicating that variation among spe- ances, or the total time from the root to the tips of cies approximates expectations from a Brownian the tree (4 million years). The off-diagonals of the motion model of evolution (i.e., a random walk in matrix represent the covariances, or the amount of which trait variance accumulates proportionally to time that pairs of species have shared evolutionary evolutionary time). The lambda parameter scales history since the root of the tree. In this example, the internal branches of a phylogeny by multiply- the covariance between species C and D AMERICANis 3 mil- ing the internal branch lengths (the off-diagonals in lion years, refecting the time between© nodes r and the variance–covariance matrix) by lambda while d, whereas the covariance between species C and retaining the original variances along the diagonal. H is 1, refecting the time between nodes r and b. Thus, when λ = 0, the internal structure of the tree Variance–covariance matrices play an important role is entirely eliminated (all internal branch lengths in many phylogenetic comparativePROOFS methods; hence, are 0, yielding a “star phylogeny”; see Felsenstein, we revisit this concept throughout the chapter. 1985). In contrast, the internal branch lengths (off- A frst step in most comparative analyses is to diagonals in the variance–covariance matrix) remain obtain a phylogeny for the species of interest. For- unchanged when λ = 1. Lambda may take any value tunately, digital phylogenies are widely available between 0 and 1 (and even slightly higher than 1, and can be downloaded from sites such as 10ktrees subject to constraints determined by characteristics (http://10ktrees.fas.harvard.edu; Arnold, Mat- of the tree). The lambda value
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages16 Page
-
File Size-