D. A. Baum & N. I. Cacho 2009 Botany 563: Phylogenetic Analysis of Molecular Data Lab12: Testing correlated evolution of discrete characters Learning objective: Familiarize yourself with different approaches to test two hypotheses regarding correlated evolution of discrete traits. Goal: We will make use of two different approaches to test two hypotheses regarding correlated evolution of discrete traits, one in Asterids, and one in Primates. For the first study case we will use Maddison's Concentrated-Changes Test implemented in MacClade (Maddison and Maddison 2001); for the second one, we will use a program called BayesTraits (Pagel and Meade 2004). TASK 1. Do single seeds evolve more often in fleshy-fruited lineages than in dry-fruited ones? Data set: Asteridae_fruits.nex (kindly provided by Diane Ferguson). The data set we will explore is relevant to understanding fruit evolution in the angiosperm subclass Asteridae. The tree is an rbcL/parsimony tree for 137 taxa of Asteridae and two outgroups (Dillenia and Vitis). There is information on two characters: 1) Fruit (dry vs. fleshy), and 2) fruit seed number (one vs. many). We will first look at character 1 (dry vs. fleshy fruit) to see what sort of information on its evolution can be extracted by looking at its distribution on the tree. Then we will look for evidence of correlated evolution between these two characters. 1.1) Open the file Asteridae_fruits.nex in MacClade. In order to see the character and character-state definitions of a character, you can select the data editor (windows menu) and then hit "state names & symbols" (characters menu); or you can use the following display option: Display>Data Matrix Styles>wide w/state names. 1.2) What is the number of steps for character 1? ________ And for character 2? __________ This information can be obtained under the tree window (windows menu) by selecting “character list” (character menu). Also, if you choose to “trace” a character (that is, to paint it onto the tree), you can see the number of steps of character 1 by scrolling through the characters in the character legend window until character 1 (dry vs. fleshy) is selected. 1.3) How many most-parsimonious reconstructions are there for character 1? ________ For this you need to trace the character of interest, in this case, character 1. Go to “resolving options” (trace menu), and select “show all most parsimonious states at each node”. This will turn on the “number of MPRs” option under the trace menu, which is what you want. Repeat the procedure for character 2: __________ 1.4) On the tree, what is the minimum and maximum number of gains of (fleshy –> dry) and losses (dry –> fleshy) for character one? Select "state change & stasis" (chart menu). Then look at minimum and maximum values for the traced character only. The area of these circles is proportional to the maximum and minimum number of changes. You need to click on the area to get the values. Alternatively, you can go select “view table” (display menu). GainsMin= GainsMax= LossesMin= LossesMax=D. A. Baum & N. I. Cacho 2009 1.5) To perform a test to see if the number of steps for character one on this tree is lower than would be expected by chance, you need to do the following: a) Select character one in the data editor. Use the “replicate” command (edit menu) to generate 100 copies of the character. b) Highlight characters 2-101 (the duplicates but not the original) and shuffle the states between taxa (i.e., permute) using the “shuffle” command (utilities menu). c) To chart out number of steps for each character (x-axis) against the number of characters, go to tree window, and select "character steps/etc” (chart menu). d) How many of your permuted characters have as few or fewer steps than the original character? What do you make of this result? What do you think of the test? 1.6) Now we wish to test the hypothesis that single seeds evolve more often in fleshy-fruited lineages than in dry-fruited lineages. That is to say we are looking to see if there is a pattern of correlation in the evolution of fruit fleshiness (character one) and fruit seed number (character two). This will be done using Maddison's Concentrated-Changes Test using character one as an independent variable. You might want to delete the 100 replicates you created before to keep things tidy. a) Set to ACCTRAN optimization (trace menu > resolving options). What is the total number of gains (one–>many) and losses (many–>one) of character TWO (seed number)? Gains: Losses: b) To save you time we have counted-up the number of gains and losses of character two that take place on branches whose end-states of character one are fleshy (i.e., on yellow branches). Gains (GF ): 3 Losses (LF ): 6 c) The Concentrated-Changes Test determines the probability of various numbers of gains and losses in the dependent character (here, seed fleshiness) given the number of gains and losses observed in a certain clade (here, GF and LF). The independent character is the traced character (in this case character one), and the null hypothesis is that gains and loses are randomly distributed. To perform this test in MacClade, you need to trace the character that is your independent variable (i.e., character one, fruit fleshiness), and select the “test correlation” tool from the tool palette (the name of a tool will appear as you click on it). Then select the base of the Asteridae (excluding the two outgroups: Dillenia and Vitis). Ask MacClade to do 1000 simulations of the Gains and Losses found in 1.6a to see how many typically occur on fleshy branches. The “distinguished branches” are those having fleshy state, which is coded as zero. Set the ancestral state to "either" and use "actual changes." To find the probability of various gains/losses scenarios, simply select or de-select the appropriate boxes. What is the probability of GF or fewer changes on fleshy branches? What is the probability of exactly GF changes on fleshy branches? What is the probability of LF or more changes on fleshy branches? What is the probability of GF or fewer and LF or more changes on fleshy branches? d) Is there evidence that single seeds evolved more often in fleshy-fruited lineages than in dry-fruited ones?D. A. Baum & N. I. Cacho 2009 TASK 2. Do female primates tend to prominently advertise their estrus when living in a multimale mating