Continuous Manual - Mark Pagel
Contents
1. sl s2 s3 s4 t3 t4 The pag input format for this tree is PAG 43 sl 1 tl datal data2 data3 s2 1 t2 datal data2 data3 s3 2 t3 data1 data2 data3 s4 3 t4 data1 data2 data3 1 2 t5 2 3 t6 The second line 4 3 is similar to Phylip format and describes first the number of species followed by the number of variables The variables are the comparative data measured across species and should be kept to about ten or fewer although this is not a strict requirement Read the third line of the input file as species 1 goes to node 1 over time or length t1 with data 1 2 3 the fourth line as species 2 goes to node 1 over time or length t2 with data 1 2 3 and so on Data points must be real numbers Missing data are coded as 66399 Beginning with the seventh line of the file the connections among the internal nodes are described node 1 goes to node 2 over time or length 4 and node 2 goes to node 3 over time or length t6 Nodes do not have data and the branch lengths can be any real number Branch lengths can be any units but units of time and genetic distance operational time are especially useful Only tips species have data and the tree should be bifurcating If you have so called polytomies in your tree it is best to resolve them to bifurcations although the program will handle some polytomies Trees must be rooted although the root itself is not de2. 690 opossum 15 096190 6 420 1 474 wallaroo 15 081150 33 920 12 460 y 2y sOLL 770 3 008890 5 027240 5 6 oS z 2055220 017280 7 043700 217 L1190 0 9 082200 19 022010 11 053360 18 033600 17 027100 17 034110 22 110190 15 16 073490 16 23 029620 17 18 018380 18 19 017810 19 20 036300 20 21 025510 21 22 032440 22 23 081790 os ON OMAATNA oO FWD PRPRPPR t WNE O Sc T 14 The phylogeny as described in the above tree file baboon gibbon gorilla homosapiens panpaniscus pantroglodytes pongo pyg pongo abelli armadillo cow fin whale blue whale cat harb seal grey seal white rhino indian rhino horse donkey mouse rat platypus opossum wallaroo 0 1 Analyses of the brain size and body size data using log transformed values log likelihood of Model A with trait covariances set to zero 83 78 Model A allowing covariance to be non zero 62 87 LR test 17 84 Model B allowing covariance to be non zero 61 42 no evidence for Model B Estimated correlation brain and body size 0 93 slope 0 62 Estimated ancestral states brain size 3 02 logged data 20 grams anti logged body size 1 65 logged data 5 2 kg anti logged Scaling Parameters estimated on both traits simultaneously non zero covariance kappa 0 85 95 CI 0 09 1 5 log likelihood 62 77 not different from 62 87 delta 2 07 0 73 4
3. 05 log likelihood 61 75 lambda 0 99 0 88 1 0 log likelihood 62 86
4. correlations The degrees of freedom for such tests are always a function of n where here n is the number of species Scaling Parameters in other analyses In a likelihood ratio test the maximum likelihood values of the scaling parameters can differ under the null and alternatuve hypothesis tests of some other feature such as the correlation To ensure that tests are nested see LR tests above scaling parameters can be separately estimated under thenull and alternative hypothesis and then fixed in both analyses at the value least supportive of the hypothesis Lamba values may differ for two traits analysed separately but whose correlation one wishes to estimate Here it is appropriate to estimate the value of lambda simultaneously with estimating the correlation VI Graphing and Display Continuous has some basic display and graphing options The Show Likelihood Curves box allows one to view a plot of the likelihood surface associated with the scaling parameters These plots are informative by showing both the 12 maximum likelihood value of the parameter and by showing its 95 confidence intervals If these intervals exlcude 0 or 1 this is equivalent to having performed an LR test The Set Graph Options and Show Graph option allow one to choose the nature of a display and then view it Currently one can plot traits against total path length or against each other These are useful for identifying outliers that may unduly influ
5. User s Manual for Continuous copyright M Pagel Mark Pagel School of Animal and Microbial Sciences University of Reading Reading RG6 6AJ UK email m pagel rdg ac uk www ams reading ac uk zoology pagel Continuous Continuous is a computer program that implements a generalised least squares GLS model for the across species analysis of comparative data The method is described in two papers Pagel 1997 1999 I would be grateful if users would cite these papers as appropriate when they use the method The application program can be used to e test for correlated evolution between pairs of characters e to find ancestral states e to examine random walk versus directional change models e to investigate the tempo and mode of trait evolution e and to assess whether and to what extent the a phylogenetic correction is required in the data This manual describes how to use Continuous and its hypothesis testing capabilities The manual is intended to provide enough of an introduction to use the program although it does not attempt to describe all of the things one might use the program for Some features of Continuous are not yet implemented in the program even though they appear in the menus The computer interface was designed and programmed by Dr Andrew Rambaut The GLS approach makes it possible to analyse and display comparative data across species without the need to calculate independent contrasts or other phyl
6. a beta path length V Hypothesis Testing Hypotheses are tested using the likelihood ratio statistic The likelihood ratio statistic compares the log likelihood of the null hypothesis model to that of the alternative hypothesis model Continuous automatically calculates the log likelihood of whatever model is chosen in the Set Model box and displays this likelihood in the text window The likelihood ratio LR test compares the goodness of fit of a model to the data with that of a simpler model that lacks one or more of the parameters The LR statistic is then defined as LR 21og H H where H represents the simpler null model and H the alternative model containing the parameters representing the evolutionary processes one wishes to estimate If the simpler model is a special case of the more complicated one the LR statistic is asymptotically distributed as a chi squared variate with degrees of freedom equal to the difference in the number of parameters between the two models i e LR y v where v is the number of degrees of freedom One test is a special case of another if it is possible to collapse the more complicated model to the simpler model by setting some parameters to zero or to other fixed values For example the model in which a correlation is estimated collapses to the null hypothesis model by forcing the correlation to be zero the zero correlation or null model is a special case of the alternative hypothesis
7. a into Continuous Input files must be text files in the format as described above If the input file has been saved by a word processor such as Microsoft Word be sure to choose the text only option Sometimes word processors can insert invisible characters that can interfere with the input file being read in If you suspect this open the file in a word processor and save it as some other file choosing a text only format If the treefile loads properly a picture of the phylogeny will be represented in the open window The traits are listed on the left of the open window Traits with checks by them are those that will be analysed Clicking on a trait holding down the option key selects or de selects it from analysis Double clicking on a trait brings up a box that allows you to view the data and to transform it logarithmically This box can also be selected from the Data menu The Show Data Matrix option in the Data menu shows all of the species and their data points Unselected traits are in grey Species with missing data are in grey Double clicking on a species removes it from any of the analyses that follow Clicking on it again re inserts it Options in the Tree menu are not implemeted Additional windows can be opened simultaneously by selecting the New command followed by Import IV Analysing Data Continuous can be used to characterise and test hypotheses about the evolution of single traits to find ancestral states and to test f
8. ence the results With a likelihood surface or plot displayed selecting Export from the File menu produces a text file of the raw data used to make the plot This file can be used as input into plotting routines and is useful for making graphs and figures VII General Tips and Known Problems Likelihood solutions can be unstable especially when the number data points per parameter estimated drops below a ratio of about ten i e ten data points per parameter It is often a good idea to fit a model several times Outlying data points can exert large effects on the estimates of parameters especially on delta It is best to test nested hypotheses for discussion see Pagel 1994 or 1997 as these are approximately chi squared distributed If the program returns an error message about the determinant being zero it usually means that the phylogeny either default or as implied by choice of scaling options has at least two species that are effectively perfectly correlated Removing one of them or changing the scaling can treat this problem An effective way to alter the scaling in this circumstance is to change the lambda parameter from its default value of 1 0 to a value of 0 9999 VIII References Pagel M Detecting correlated evolution on phylogenies a general method for the comparative analysis of discrete characters Proceedings of the Royal Society B 255 37 45 1994 Pagel M Inferring evolutionary processes from phylogenies Zoo
9. estimated for each Continuous will automatically incorporate the maximum likelihood values of the three scaling parameters into its calculations or they can be fixed to predetermined values using the dialog boxes in the Set Model box A maximum of any two parameters can be simultaneously estimated although this can be time consuming because the likelihood surface is large Three scaling parameters and their interpretation when applied to trait evolution on a phylogeny Parameter Action 0 lt l 1 gt l Assess contribution star phylogen hylogenetic history has default not defined lambda Poe y phy gen y of phylogeny species minimal effect phylogeny independent Scale branch punctuational stasis in longer branches default longer branches x kappa lengths in tree evolution gradualism more change delta Scale total path not defined temporally early change default temporally later root to tip in tree important adaptive gradualism change species radiation specific adaptation Set trait covariances to zero When checked this box forces the covariances among pairs of traits to be zero when calculating the likelihood It is used to test whether the correlation between two traits is greater than zero The test is described below Show Tree Matrix This command displays a box with a number of options Most are repeated in the main menu The Traits option displays the raw data on species The Tree option displays the tree ma
10. logica Scripta 26 331 348 1997 Pagel M Inferring the historical patterns of biological evolution Nature 401 877 884 1999 IX Disclaimer The routines have been tested and are correct to the best of my knowledge However I cannot take responsibility for anyone else s use of the program 13 Appendix Example data set This data set includes values of brain size grams and body size kg on 24 mammal species It is in the correct pag file input format for Continuous Some of the trait data are missing The phylogeny is based on Mt DNA pag format input file for tree shown below Traits are in order of brain size body size PAG 24 2 pan pan 1 011750 295 500 33 000 pan trog 1 011040 410 300 48 000 h sap 2 022490 1250 000 65 000 gorilla 3 029830 505 900 126 500 pong pyg 4 027020 413 300 53 000 pong abelli 4 024970 gibbon 6 054450 107 700 5 500 baboon 7 126260 142 500 15 450 fin whale 8 012300 8325 000 59394 000 blue whale 8 014020 6800 000 58059 000 cow 9 053440 450 000 900 000 harb seal 10 007580 297 200 84 833 grey seal 10 008250 320 000 163 400 cat 11 052040 20 700 2 213 white rhino 12 021870 655 000 764 000 indian rhino 12 028610 horse 13 007820 712 000 484 000 donkey 13 006680 405 000 150 000 armadillo 20 115520 12 000 3 700 mouse 14 055170 450 024 rat 14 051110 2 380 339 platypus 16 142190
11. model In such circumstances the two models are often referred to as being nested If two different models are chosen in the Set Models box in succession their log likelihoods can be compared by hand to perform a LR test Alternatively the Likelihood Ratio Tests option in the Continuous Analysis menu allows one to define two different models simultaneously and Continuous then fits the models and automatically performs a likelihood ratio test on them A p value is displayed from the appropriate chi squared distribution on the assumption that the tests are nested i e subsets of each other Continuous is reasonably good at recognising tests that are not nested but may not be infallable The Likelihood Ratio Tests option displays separate Set Model boxes for the null and alternative hypotheses By choosing the models appropriately one can test a range of hypotheses The list that follows gives an idea about how to perform some general hypothesis tests in the LR framework Example Likelihood Ratio tests Pagel 1999 gives examples of several of the tests listed below 10 Correlations and Regressions compare a model A or B in which the set trait covariances to zero box is checked for the Null hypothesis model but not for the Alternative hypothesis model this is a nested test This tests the hypothesis of p 0 0 The correlation can be found from the Continuous Analysis menu The regression coefficient can be found by calculating the
12. o see if default or scaled gradualism is better supported Adaptive Radiation versus Species Adaptation If delta is lt 1 0 on a LR test adaptive radiation is suggested Delta gt 1 0 suggests that later or species specific adaptation has been dominant Other Analysis Issues Species Regressions A species regression can be approximated i e result is very nearly equivalent by setting lambda to 0 0 in the Set Model box Setting lambda to 0 0 and delta to 0 001 approximate gives a result virtually indistinguishable from a species regression Independent Contrasts Model A with no scaling transformations returns results equivalent to those obtained from independent contrasts analyses The estimate of alpha root in this cased is equivalent to that obtained from squared change parsimony Binary Data Continuous will accept binary data If a single binary trait is significantly correlated with a quantitative trait the interpetation is that the mean value of the quantitative trait for the group coded zero on the binary trait differs from the mean of the group coded 1 If all of the variables are binary a different model should be employed see Pagel 1994 and computer program Discrete Partial Correlation and Regression Continuous is not set up to calculate regressions on more than a pair of traits However partial correlations and regressions can be calculated by hand using the correlation matrix and standard formulae for partial
13. ogenetic corrections independent contrasts turn out to be a special case of the GLS approach see below In the GLS approach non independence among the species is controlled for internally by reference to a matrix of the expected covariances among species This means that data can be plotted across species and interpreted using the correlations and regressions obtained from Continuous The GLS approach as implemented in Continuous also makes it possible to transform and scale the phylogeny to test the adequacy of the underlying model of evolution to assess whether phylogenetic correction of the data is required and to test hypotheses about trait evolution itself for example is trait evolution punctuational or gradual is there evidence for adaptive radiation is the rate of evolution constant I Running Continuous This version of the program runs on a PowerPc Macintosh and prefers System 8 or later II Data Input Format Continuous requires a well resolved phylogeny and data on species It uses its own input format pag format to describe the phylogenetic tree and the data The attraction of the tree format is its transparency although it is not as compact as Phylip Newick or Nexus formats We hope to make available software for converting among these three formats The input format is simple Consider the tree below of four species and three internal nodes 1 2 and 3 where node 3 is also the root Trees must be rooted
14. ogresses Seen this way delta is a parameter that detects differential rates of evolution over time and re scales the phylogeny to a basis in which the rate of evolution 1s constant The parameter lambda reveals whether the phylogeny correctly predicts the patterns of covariance among species on a given trait This important parameter in effect indicates whether one of the key assumptions underlying the use of comparative methods that species are not independent is true for a given phylogeny and trait If a trait is in fact evolving among species as if they were independent this parameter will take the value 0 0 and indicate that phylogenetic correction can be dispensed with A lambda value of 0 0 corresponds to the tree being represented as a star or big bang phylogeny If traits are evolving as expected given the tree topology and the random walk model lambda takes the value of 1 0 Values of lambda 1 0 are consistent with the constant variance model sometimes called Brownian motion being a correct representation of the data Intermediate values of lambda arise when the tree topology over estimates the covariance among species The value of lambda can differ for different traits on the same phylogeny If the goal is to estimate the correlation between two traits then lambda should be estimated while simultaneously estimating the correlation If on the other hand the goal is to characterise traits individually a separate lambda can be
15. or evidence of correlations among traits All phylogenetic correction of the data is done automatically All parameters are found as maximum likelihood estimators Some are found analytically others must be found by searching a likelihood surface for the value that maximises the likelihood of observing the data given the value of the parameter the phylogenetic tree and the model of evolution All likelihoods are expressed as log likelihoods The Continuous Analysis menu provides a number of ways to investigate trait evolution and correlations The next section describes some of the options How to test these options for significance is described in a later section Set Model command Models of evolution Two models of evolution are available Model A corresponds to the standard constant variance random walk model It has a single parameter the instantaneous variance of evolution Model B is a directional random walk This model has two parameters the variance of evolution parameter as in Model A plus the directional change parameter This parameter effectively measures the regression of trait values across species against total path length from the root of the tree to the tips It detects any general trends towards a dominant direction of evolutionary change have species got bigger smaller faster longer and so on The estimates of the parameters for Model A and Model B from a given set of data can be viewed by selecting the Parameters o
16. ption described below Model A and B can be compared via their log likelihoods If Model B fits the data better larger log likelihood closer to zero than Model A it says that a directional trend exists This is important in its own right and has implications for other estimators For example estimates of ancestral states at the root of the tree will differ greatly between Model A and Model B when Model B is significant and the Model B estimates are to be preferred N B Model B can only be fitted to trees that have some variation in the total path length from the root to species Model B cannot be used with ultrametric trees Transformations A useful feature of Continuous are the three scaling parameters that can be estimated for a given data set and phylogeny They are shown in the Set Model box where it is possible to fix them to a specified value leave them at their default values or estimate them by maximum likelihood The parameters are denoted kappa i lambda A and delta 5 These scaling parameters allow tests of the tempo mode and phylogenetic associations of trait evolution All three take the value 1 0 by default These values correspond to assuming that the phylogeny and its branch lengths accurately describe the constant variance random walk model A or B However if trait evolution has not followed the topology or the branch lengths these values will depart from 1 0 When they do incorporating them into the analy
17. ratio of the covariance between the two traits to the variance of the X axis character These values are in the Var Covar display The regression line can be drawn on a plot of the two variables using the calculated regression coefficient and the formula y y bx where y bar and x bar are the means in the raw data yin 1s the y axis intercept and b is the regression coefficient from Continuous If the correlation is significant so is the regression coefficient and they have identical p values Drift versus Directional Models of Evolution Compare Model A to Model B using the LR test If done for a single trait this test shows whether there is a dominant direction of change to a trait If Model B is significant the maximum likelihood estimate of the root node can be reconstructed to lie outside of the range of values observed in the data see Pagel 1999 for an example Random walk or Model A methods always reconstruct the ancestor within this range Model B can be used when estimating the correlation between two traits If Model B is significant then the correlation between the traits will represent that variation in the two traits that is independent of their respective relationships with path length In effect the correlation coefficient estimated under Model B is a correlation of residuals from the line relating each trait to time or distance from the root Constant variance so called Brownian motion model The adeq
18. scribed but inferred by the program from the input format The species names can be any alphanumeric character but should start with a letter Internal nodes must be integers and must go in ascending order from the tips to the root Thus a descendant node must always have a smaller number than its ancestral node It may often be easiest to number the species from left to right beginning with species s1 to species sn Then label the nodes as n 1 n 2 and so on until you reach the root Every node in a bifurcating tree must have two and only two descendants The root does not go to any other node as is not described any further Items are separated by commas At the moment Continuous does not give useful information when it tries to read in a treefile that is wrong in some respect The most common errors are failing to separate items by commas or inserting more than one comma having too few or too many descendants of a node specifying the wrong descendant or incorrectly specifying the number of species or variables The best way to de bug a treefile that is not working is to print it out and compare it line by line to a picture of the phylogeny An example data file is included in the Appendix III Loading and Manipulating the Data If there is no open window when Continuous is started up select New from the File menu The treefile is loaded into Continuous via the Import command in the File menu At present this is the only way to load dat
19. sis of the data e g when estimating the correlation between two traits significantly improves the fit of the data to the model The kappa parameter differentially stretches or compresses individual phylogenetic branch lengths and can be used to test for a punctuational versus gradual mode of trait evolution Kappa gt 1 0 stretches long branches more than shorter ones indicating that longer branches contribute more to trait evolution as if the rate of evolution accelerates within a long branch Kappa lt 1 0 compresses longer branches more than shorter ones In the extreme of Kappa 0 0 trait evolution is independent of the length of the branch Kappa 0 0 is consistent with a punctuational mode of evolution The parameter delta scales overall path lengths in the phylogeny the distance from the root to the species as well as the shared path lengths It can detect whether the rate of trait evolution has accelerated or slowed over time as one moves from the root to the tips and can find evidence for adaptive radiations If the estimate of Delta lt 1 0 this says that shorter paths earlier evolution in the phylogeny contribute disproportionately to trait evolution this is the signature of an adaptive radiation rapid early evolution followed by slower rates of change among closely related species Delta gt 1 0 indicates that longer paths contribute more to trait evolution This is the signature of accelerating evolution as time pr
20. trix which is proportional to the default variance covariance matrix among species as implied by the phylogeny Elements of this matrix are the sum of the branch lengths on the main diagonal entries corresponding to a species and the sum of shared branch lengths on the off diagonal entries corresponding to the co variance between two species The values in this matrix are proportional to the variances and covariances It is this matrix that delta acts on Its effects can be seen by setting delta to different values and viewing the resulting tree matrices The Var Covar option shows the calculated variances of the traits and the covariances between traits as estimated from the model of evolution that has been assumed Model A or B and assumed or estimated values for kappa lambda and delta If the set trait covariances to zero box has been checked the covariances will be zero The Correlations option displays the correlations among traits subject to the same model options as described for the Var Covar option The Parameters option shows the values of up to two parameters and their standard errors The alpha parameter is the estimated root of the tree The beta parameter is the directional change parameter of Model B It can be interpreted as a regression coefficient of the trait values against total path length The combination of alpha and beta can be used to predict the values of the trait for any given total path length according to alph
21. uacy of this model as a descriptor of the data can be tested by asking whether all three scaling parameters are either 1 0 or not significantly different from 1 0 as their ML values Molecular clock If the sampling times are available for the tips of the tree and the branch lengths are in units of genetic distance the comparison of Model B versus Model A tests whether time and genetic distance are linearly related A significant Model B does not imply a strict clock but if kappa or delta do not differ from 1 0 this interpretation is strengthened The Model B directional coefficient gives the realtonship of time to genetic distance Its reciprocal is the rate of evolution see Pagel 1999 for an application to Influenza evolution Influence of Phylogeny Compare a model in which lambda is set to 1 0 with one in which lambda is allowed to take its maximum likelihood ML value If the LR test is significant it says that lambda lt 1 0 Alternatively test a model in which lambda 0 0 to one in which it takes its ML value This tests whether lambda gt 0 0 If it is significant it says that some sort ofphylogenetic correction is required Punctuational and Gradual Trait Evolution Perform a LR test of kappa 0 0 null to kappa ML value If kappa is not significantly different from 0 0 then trait evolution is 11 consistent with a punctuational mode of change Kappa gt 0 0 implies some form of gradualism Test whether kappa lt 1 0 t
Download Pdf Manuals
Related Search