CHANGES IN APE VERSION 5.8-1 NEW FEATURES o write.nexus() has a new option, digitis = 10, similar to the one with the same name in write.tree(). The help page of the latter has been clarified (see PR #130 on GitHub). o kronoviz() has two new options, 'direction' and 'side', to add more flexibbility to the plot. BUG FIXES o plot.phylo() shifted the coordinates when the tree was plotted left- or downwards (see PR #119 on GitHub). o Newick strings with a single tip and multiple nodes were not correctly parsed. Also, if a root edge was present it was not correctly decoded (see issue #124 on GitHub). o dist.nodes() returned a (meaningless) result for trees with no branch lengths (sometimes a segfault occurred). This function has now the option 'fail.if.no.length = FALSE' so that branch lengths, if absent in the tree, are fixed to one (with a warning). An error is returned if this option is set to FALSE. OTHER CHANGES o Plotting circular trees has been improved (see PR #129 on GitHub). o axisPhylo(): the new default is axisPhylo(side = NULL) instead of axisPhylo(side = 1). CHANGES IN APE VERSION 5.8 NEW FEATURES o dist.topo() has a new option 'mc.cores'. With the default PH85 distance, the gain of using several cores is visible with 200+ trees. The internal code for the score distance has been improved and is twice faster with a single core, and 4-5 times faster with mc.cores = 2. o keep.tip() gains a '...' argument which is passed to drop.tip(). The behaviour of the latter is changed so that node labels, if available, are used to create new tip labels when subtree = TRUE (see PR #90 on GitHub). o nj() can now handle distances stored in long vectors. There was previously an implicit limit to 46,341 observations (see PR #97 on GitHub). o del.gaps(), del.colgapsonly(), and del.rowgapsonly() now accepts objects of class "AAbin". Besides, the class of the input object is always respected (it used to be "DNAbin" only even when it had several elements). o makeNodeLabel() is now generic with methods for the "phylo" and "multiPhylo" classes. o unroot() gains two options, 'collapse.singles' and 'keep.root.edge', to handle some complicated situations (see issue #103 on GitHub). o read.GenBank() has a new argument 'type' to download AA sequences. The default is type = "DNA", and the other choice is type = "AA". BUG FIXES o rtt() was fixed when dates include NAs (issue #73 on GitHub). o njs(), bionj(), bionjs(), fastme.bal(), and fastme.ols() now check that there are at least three observations in the distance matrix (issue #84 on GitHub). o nodelabels(, frame = "circle") used to ignore the argument 'adj'. This is fixed (issue #85 on GitHub). o write.tree() ignored tree names. This bug was introduced in ape 5.7 (see issue #101 on GitHub). o read.tree() now reads correctly trees with many nodes but a single tip such as "(());" (see issue #104 on GitHub). o bitsplits() failed with large trees. There are now checks on memory requirements; some explanations have been added in the help page. o comparePhylo() failed with two large trees. This is fixed. OTHER CHANGES o .uncompressTipLabel() now returns an object of the same class than the input object (previously the returned object was always of class "multiPhylo"). o The help page of is.binary() has been clarified that branch lengths are ignored (issue #83 on GitHub). o dist.nodes() has an improved internal C code with unlimited tree size. The code is three times faster with small trees; twice faster with ~1000 tips (the gain is less with bigger trees). o ace() now uses node labels, if available, in the output. o comparePhylo() now restores the graphical window in its previous state (i.e., after calling layout() if plot = TRUE). o read.tree() now gives a warning (still returning NULL) if there is (are) no semicolon(s) [end of tree(s)] in the Newick string. o write.tree() now gives a warning if the length of the node labels does not match the number of nodes given by '$Nnode'. CHANGES IN APE VERSION 5.7-1 BUG FIXES o Some very unbalanced trees resulted in a segfault (see issue #70 on GitHub). OTHER CHANGES o ape 5.7 introduced an implicit dependence on R >= 4.1. This has been corrected and this version should work with R >= 3.4 (and certainly with R >= 3.2). o All examples should now write files in the TMP directory (some still used to do this in the user's working directory). CHANGES IN APE VERSION 5.7 NEW FEATURES o The new function muscle5 calls MUSCLE5. There are two other new functions that depend on this program, efastats and letterconf, to compute summaries for a set of alternative alignments and display the confidence in each site in HTML, respectively. o print.LargeNumber() gains the options 'latex' (FALSE by default) and 'digits' (depends on 'latex'). o drop.tip() and keep.tip() are now generic functions with methods for the classes "phylo" and "multiPhylo". BUG FIXES o A bug was fixed in multi2di() and di2multi() for compressed "multiPhylo" objects causing an error if the object had to be reordered internally (see PR #51 on GitHub). o unroot() now accepts different "order" of the "phylo" object (see PR #52 on GitHub). o write.tree(), ladderize(), and balance() failed with very unbalanced trees (issues #53 and #54 on GitHub). o PGLS models used a wrong ordering of data if 'species' was a factor in corBrownian(, form = ~species) or other correlation structure functions. o A bug was fixed in the calculations of the variance of the TN93 distance (dist.dna). o consensus() did not use postprocess.prop.part() and might scramble nodelabels (see PR #65 on GitHub). o is.binary.phylo() used to return TRUE with trees such as ((a:1):1,b:2,c:2):1;. This test now conforms with what is written in the documentation. OTHER CHANGES o reconstruct() gains the options 'low_alpha = 0.0001' and 'low_alpha = 1' to allow estimation of values of alpha greater than 1. o src/me.c and src/me.h are now OK with -Wstrict-prototypes flag. o makeNodeLabel(method = "md5sum") now uses the function digest from the package of the same name. The package tools has been removed from the Imports list. CHANGES IN APE VERSION 5.6-2 NEW FEATURES o plot.phylo() can now plot "tidy" trees with the option type = "t[idy]" (see: van der Ploeg, 2014, Drawing non-layered tidy trees in linear time. J. Software: Practice and Experience 44:1467). BUG FIXES o is.monophyletic() didn't work correctly with duplicated labels (issue #50 on GH). o birthdeath()'s 'if' conditions should be now all of length 1. CHANGES IN APE VERSION 5.6-1 BUG FIXES o axisPhylo() misplaced the scale with some trees drawn with plot(type = "fan"). o The internal code of bitsplits() has been improved and can now handle all types of unrooted trees. CHANGES IN APE VERSION 5.6 NEW FEATURES o The new generic function degree returns the degrees of all nodes in a tree or network. There are methods for "phylo" and "evonet" objects. o The new function solveAmbiguousBases replaces R, Y, W, ... by A, G, C, or T using columnwise base frequencies. o There is a new method as.phylo.default() for any object inheriting the class "phylo". o read.dna() and read.FASTA() can now read gzipped files. o where() now works also with objects of class "AAbin". 'NULL' is now returned, instead of an error, if the sequence is shorter than the searched pattern (useful with lists with some short sequences; issue #28 on GH). o comparePhylo() has a new '...' argument for options passed to plot.phylo(), and two new options: 'commons' to specify whether to show the splits in common, and 'location' for the legend (PRs #41 and #42 on GH). o The code of dnds() has been rewritten which fixes some minor bugs, is more robust, and gains the options 'details' to print the numbers of TS and TV for each degeneracy category, and 'return.category' to return the degeneracy categories of the original data (see ?dnds for details). BUG FIXES o A bug was introduced in read.tree() in ape 5.5 when reading trees with quoted labels (issue #14 on GH). o nj() did not duplicate the distances if originally stored in a "dist" object (issue #12 on GH). o The double-centered matrix computed by pcoa() is now more strictly symmetric, thus avoiding complex eigenvalues in some cases (issue #17 on GH). o chronos() failed with discrete models when branch lengths were extremely heterogeneous leading to badly initial rate values. o multi2di() sometimes resulted in an ultrametric tree to become non-ultrametric (see issue #23 on GH). o read.nexus() failed to read correctly files with labels containing spaces in the TRANSLATION bloc (PR #29 on GH). o ace(, type = "d") now woks with n = 2. o drop.tip() used to remove the root edge of a tree if the option 'root.edge = 0' was set. Since the latter is the default, the side-effect was to possibly turn a rooted tree into an unrooted one (see issue #32 on GH). The root edge is now removed only if the option 'rooted = TRUE' is used. The help page has been clarified on this point. o as.hclust.phylo() now returns the correct order of leaves so that plot() does not cross lines (PR #33 on GH). o unique.multiPhylo() failed with a single tree in the list or with an empty list (PR #38 on GH). o trans() failed with a matrix with a single codon. A similar bug affected as.character.AAbin(). o rTraitDisc() now uses expm::expm unless this package is not installed in which case ape::matexpo is used. OTHER CHANGES o multi2di() is faster as it avoids calling which() repeatedly. o dist.nodes() now gives an error with an explicit error message if the tree is too large (i.e., more than 46,340 tips + nodes). o The default values of the options 'edge.color', 'edge.width', and 'edge.lty' of plot.phylo() are now taken from the graphical parameters: par("fg"), par("lwd"), and par("lty"), respectively. o The visual aspect of circular trees (type = "f" in plot.phylo) has been improved. o ace() now accepts trees with some branch lengths equal to zero. o write.tree() now accepts simple lists of trees. o chronos(, model = "clock") is a short-cut to specifying model = "discrete" and a single category of rates in the settings. CHANGES IN APE VERSION 5.5 NEW FEATURES o The new functions rtopology and rmtopology generate random trees with equal frequencies of *labelled* topologies. rtree(, equiprob = TRUE) now gives equal frequencies of *unlabelled* topologies. These features are described in the new vignette "RandomTopologies". o dist.dna()'s internal code has been improved: it now accepts up to 2.1 billion sequences each up to 2.1 Gb (see CHANGES in APE VERSION 4.0 below). o nj()'s internal code has been improved and is slightly more efficient: combined with the improved dist.dna(), boot.phylo() with an NJ tree is now about 10% faster. o howmanytrees() can now compute (approximately) the number of binary, (un)rooted, labelled topologies beyond the standard resolution of numeric values (i.e., > 10^308) thanks to the new utility function LargeNumber. BUG FIXES o bind.tree() sometimes returned badly conformed trees. o write.FASTA() made R crash if the output file was in a non-existing directory (thanks to Richel Bilderbeek for the report). o ltt.plot.coords() now handles correctly trees with singleton nodes. This also fixes functions calling this one (e.g., ltt.plot). o estimate.dates() now works with non-bifurcating trees. o root(phy, outgroup, resolve.root = TRUE) failed to place correctly the new root if 'outgroup' was of longer two or more and paraphyletic (although monophyletic once 'phy' was unrooted). o pcoa() failed when a single axis was returned. OTHER CHANGES o node- and tiplabels(pie = ) now accept data frames (or a single-column matrix/data frame). o boot.phylo() now returns integer(0) with a warning if there are less than 4 rows in the data (argument x). In case of unrooted trees, the first support value is now NA (it used to be B, the number of bootstrap replications). o prop.part() and is now faster thanks to some internal improvements. This also affects dist.topo() which is twice faster for small trees, and up to 30 times faster for trees with 1000 tips. o bitsplits() now gives an error with rooted trees CHANGES IN APE VERSION 5.4-1 NEW FEATURES o rtree() gains several improvements: * The new option 'equiprob = TRUE' generates all topologies in equal proportions; 'equiprob = FALSE' generates unbalanced topologies in higher proportions (as in previous versions of ape). The latter is the default because several packages use rtree() with set.seed() in their examples. * Rooted trees with n = 1 tip can now be simulated. * Vector of tip labels supplied by the user are now checked and a warning is issued if its length does not match 'n'. * The internals of the code have been rewritten and are now more efficient (particularly if 'equiprob = TRUE'). o multi2di() gains an option 'equiprob' similar to rtree(). o The new function nexus2DNAbin is a helper to convert outputs from read.nexus.data(). BUG FIXES o is.monophyletic() might return the wrong answer if the tips were given as labels not alphabetically sorted. o boot.phylo() returned bootstrap values all equal to 0. CHANGES IN APE VERSION 5.4 NEW FEATURES o The new function getAnnotationsGenBank reads annotations from GenBank and returns them in data frame(s). o read.GenBank() gains two options: 'chunk.size' to set the number of sequences downloaded together, and 'quiet' to display the download progress (will also display the name of the temporary file where the sequences were saved). The options 'gene.names' has been removed. o The new convenience function latag2n (leading and traling alignment gaps to N) substitutes leading and trailing gaps in DNA alignments into N's. The gaps in the middle of the sequences are left unchanged. o There are new cbind() and rbind() methods for the class "AAbin". o seg.sites() gains two options: 'strict' and 'trailingGapsAsN' to give more flexibility in the presence of ambiguous bases and/or alignment gaps. o read.nexus.data() now handles polymorphisms in discrete characters (however with no distinction between ambiguous or true polymorphic characters). o trans() now supports genetic codes 1 to 6 (dnds() still supports only genetic codes 1 and 2). o chronos() has been improved with contributions by Santiago Claramunt: * Some errors were fixed in the calculation of the log- likelihood of the "discrete" models with more than one category of rates. * Calculations are now faster with "discrete" models. * A general convergence diagnostic has been added in the output. Some other contributions by Guillaume Louvel: * Improved algorithm to find initial dates. * There is now a parameter 'epsilon' to check convergence. * The number of iterations is output in the final results. o Phylogenetic correlation structures (functions with class "corPhyl") can now specify grouping factors via the arguments 'form' and 'covariates' (which are no more ignored; see examples in ?ape::corClasses). This can be used, for instance, to specify species names to match the tree with the covariates. BUG FIXES o A bug was fixed in pcoa() (thanks to Jesse Connell). o A bug was fixed in countBipartitions(). o unroot() did not handle node labels correctly. o label2table() now keeps all levels in labels (not only the first three ones. o A bug was fixed in drop.tip(, root.edge = ). o rphylo() failed with time-dependent parameters. o is.monophyletic() now returns an error if some labels are not found in the tree (thanks to David Winter for the fix). o read.tree() and read.nexus() crashed R if the numbers of left and right parentheses in a Newick string were not equal: this is fixed and an error is now returned. OTHER CHANGES o The option 'tbr' in fastme.bal() has been disabled. CHANGES IN APE VERSION 5.3 NEW FEATURES o The new function dnds calculates pairwise dN/dS ratios. o alview() has a new option 'showpos' to display the positions of the sites (can be a subset of these). Previously, only the position of the last site was printed. BUG FIXES o mantel.test() returned NA if the second matrix had its diagonal with at least one NA. o Several bugs were fixed in plot.popsize(). It gains the options 'xlab', 'ylab', and 'log' for more flexible plotting. OTHER CHANGES o NAMESPACE has now explicit exports. The present version has 325 exported objects. Most internal functions are not exported, as well as several methods. o checkLabel() is now documented. o The running times of write.tree() now scales with N (number of trees) instead of N^2: writing 2 million trees with 9 tips now takes 8 mins instead of 28 hrs with ape 5.2. CHANGES IN APE VERSION 5.2 NEW FEATURES o The new function keep.tip does the opposite operation than drop.tip(). o The new function rDNAbin generates random DNA sequences. o The new generic function updateLabel changes some (or all) labels in an object giving two vectors 'old' and 'new' of labels. There are methods for the following classes: DNAbin, AAbin, phylo, evonet, data.frame, and matrix. o read.gff() has a new option 'GFF3' (TRUE by default) to set correctly the column names of GFF2 (aka GTF) files. o Xplor() has the new option from = "HOME" which is passed to Xplorefiles(). o ace() can take state uncertainty for discrete characters into account: this should be coded with R's NA. o plotTreeTime() has a new option, color (TRUE by default). o as.phylo.formula() has a new option, collapse, to add a single node if a given taxonomic level has only one representative (contributed by Eric Marcon). BUG FIXES o read.FASTA(, type = "AA") failed. o image.DNAbin() failed with a single sequence. o write.nexus.data() wrote the binary representation of DNA or AA sequences. o read.tree() failed to read correctly single-quoted node labels unless some tip labels were also single-quoted. o node.height(, clado.style = TRUE) returned the wrong vector (it used to return the same result than node.depth(). o tiplabels() did not align labels properly if the tree was plotted with plot(, type = "fan", align.tip.label = TRUE). OTHER CHANGES o print.DNAbin() now prints nicely the total number of bases. o drop.tip(tr, 1:Ntip(tr), trim.internal = FALSE) now returns a tree with all the terminal branches of tr removed, unless it has less than three nodes. CHANGES IN APE VERSION 5.1 NEW FEATURES o The new function write.FASTA is much more efficient than write.dna(, format = "fasta") for very big data set. It can also write "AAbin" objects into FASTA files. o The new function comparePhylo compares two trees with optional plot showing the differences. o There are three new methods as.AAbin() to convert from BioConductor for the classes "AAString", "AAStringSet", and "AAMultipleAlignment". o read.FASTA() can now read amino acid sequences; it has a new option type = "DNA" by default. o write.nexus.data() can now handle continuous and standard data thanks to a contribution by Thomas Guillerme. o muscle(), clustal(), and clustalomega() have new arguments to do tree-guided or progressive alignment, and a new option 'file' to save the alignment in a file (see ?details). These functions now handle "AAbin" sequences. o mantel.test() can now analyse asymmetric matrices following a suggestion by Andrzej Galecki. o as.prop.part.bitsplits() gains an option 'include.trivial = FALSE', and the generic as.prop.part() gains a '...' argument. BUG FIXES o as.hclust.phylo() failed with two-tip trees. o drop.tip(, subtree = TRUE) failed when dropping a single tip. Thanks to Thomas Sibley for the fix. o nj() now checks that there are at least three observations. o chronos() better finds initial dates in difficult situations, particularly when the root age is unknown. o chronos() now exits with the current estimates if the plogLik is NA/NaN. Thanks to Daniel Lang for the fix. o plot.phylo(phy, type = "p", edge.color = x) failed if 'phy' had singleton nodes and 'x' was a vector of length > 1. o bind.tree() failed if the trees had unordered edges. Thanks to Veronika Boskova for the fix. o cbind.DNAbin() now checks that the rownames of each matrix are not duplicated (unless check.names = FALSE). OTHER CHANGES o branching.times() now returns a zero-length vector and a warning if the tree has no branch length (it used to return an error). o igraph is now a suggested package. o The as.DNAbin() methods to convert from BioConductor are more efficient. CHANGES IN APE VERSION 5.0 NEW FEATURES o ape now supports all types of phylogenetic trees and nerworks (see details below). o The new functions read.evonet and write.evonet read and write files in Newick extended format (Cardona et al. 2008. BMC Bioinformatics 9:532). o Four new functions contributed by Franz Krah help to manipulate AA sequences: methods c(), as.list(), and as.matrix() for the class "AAbin", and as.AAbin() for lists. o The new function has.singles tests if a tree has single nodes: such trees can now be plotted with plot.phylo(). o There are two new convenience functions cladewise() and postorder() which are short-cuts to reorder(, index.only = TRUE). o The new function Xplor shows all data files on the local machine in a Web browser with clickable links to the directories and files; it is based on the Xplorefiles() function. o image.DNAbin() and image.AAbin() gain two options, grid and show.base (or show.aa), to display a grid and the symbols of the bases (or amino acids), and three options to control the aspect of the latter: base.cex, base.font, and base.col (or aa.cex, aa.font, and aa.col). o image.AAbin() now shows the polymorphic positions. o Several internal improvements were made to pcoa(). biplot.pcoa() gains an optional argument 'main' to add a title. o tiplabels() has a new option 'offset' to position the labels with respect to the tips of the tree. This can be used in combination with 'adj' and works for all types of trees (except unrooted ones). o phydataplot(type = "mosaic", continuous = FALSE) allows to define the colours more flexibly thanks to a suggestion by Elizabeth Purdom; this can be done if the function passed with 'funcol' returns a named vector. Also, the values are now sorted alphabetically in the legend. o phydataplot(type = "mosaic", legend = "none") does not display the legend. o read.tree() and read.nexus() can now read trees with singleton nodes (i.e., nodes of degree 2) thanks to Klaus. o read.tree() can now read labels with special characters within single quotes. o There are reorder() and Nedge() methods for objects of class "evonet", and a new generic as.evonet() with a method for the class "phylo". o LLT() gains a '...' argument (e.g., to use log = "y"). o drop.tip() and extract.clade() gain an option 'collapse.singles' (TRUE by default). BUG FIXES o CADM.*(... make.sym = FALSE) failed: this is now fixed. o phydataplot(x, , type = "mosaic") now works if x is a vector or a one-column matrix. o as.igraph.phylo() has been improved to work with igraph >= 1.0. o reorder.multiPhylo() failed with compressed lists. o arecompatible() returned FALSE when comparing two identical splits. o as.hclust.phylo() failed if internal branches near the root have length zero, so sorting the branching times failed due to rounding errors. o plot.phylo() failed when branch lengths were very long compared to the tip labels: use of the plotting space has been improved. OTHER CHANGES o getMRCA() is now much faster (thanks to Joseph Brown and Klaus). o read.tree() and read.nexus() are now based on C code (thanks to Klaus) and should be several times faster. o The default read.tree(, comment.char = "#") has been changed to comment.char = "" so that extended Newick files can be read. o The internal code of prop.part() and of reorder.phylo() have been rewritten in C++ and should be several times faster. collapse.singles(), drop.tip(), and extract.clade() are also much faster. o dist.topo() is much faster with the default distance. o write.tree() and write.nexus() used to replace multiple underscores or dashes in labels by single ones: they are now left unchanged. The help page ?write.tree has been clarified on how tip and node labels are checked before being printed. CHANGES IN APE VERSION 4.1 NEW FEATURES o The new function read.fastq reads FASTQ files returning a "DNAbin" object with an attribute "QUAL". o The new functions Xplorefiles, bydir, and edit.file.extensions help to find files on the local disk. o read.FASTA() can now read connections thanks to a contribution by RJ Ewing: the help page shows how to read *.zip, *.gz, or *.bz2 files. This function can now read remote files through secure connections (with HTTPS or FTPS). These also work for read.dna(format = "fasta"). o read.nexus() has a new option, force.multi (FALSE by default), to always return an object of class "multiPhylo" even if the file contains a single tree (to keep the tree name which is lost in the object of class "phylo"). BUG FIXES o read.FASTA() crashed when reading a file with no sequence or a compressed file not identified as such (see above about the support for connections). o read.GenBank() failed when at least one accession number was wrong: it now returns the sequences eventually read and prints a warning with the accession numbers not read. o extract.clade() failed with trees previously rooted with root(resolve.root = TRUE). o A bug was fixed in estimate.dates(). o A bug was fixed in multi2di.multiPhylo(). o drop.tip() failed when dropping all tips but one. o nj() now checks for infinite distances (it used to return a badly conformed tree). OTHER CHANGES o chronos() now checks the presence of branch lengths in the tree. o print.DNAbin() now prints base frequencies until 10 million bases. The display of labels has been improved. CHANGES IN APE VERSION 4.0 NEW FEATURES o Two new functions, estimate.mu and estimate.dates, contributed by Bradley Jones, estimate the mutation rate and the dates for a rooted phylogenetic tree with dated tips. o The function plotTreeTime plots a non-ultrametric tree together with the dates of the tips. o The new function read.gff reads GFF files. o The new method all.equal.DNAbin compares two sets of DNA sequences mainly to compare alignments with a graphical display of the differences. o boot.phylo() has two new options: 'jumble' (TRUE by default) to randomize the order of the rows of the data matrix (and avoid artificially too large bootstrap values associated with very short branches), and 'mc.cores' to perform parallel bootstraps. o reconstruct() now computes ancestral states under directional models (Brownian with trend and OU; see: Royer-Carenzi, M. and Didier, G. 2016. A comparison of ancestral state reconstruction methods for quantitative characters. J Theor Biol, 404:126-142). o root() has the new option 'edgelabel' to treat node labels as edge labels so they are associated with the correct edges when using drawSupportOnEdges (edgelabel = FALSE by default). o read.GenBank() is faster and more flexible: it can read records such as contigs or scaffolds (based on a code available on GitHub for a while). o is.ultrametric() has a new argument, option, to change the criterion used to test ultrametricity. o .compressTipLabel() has a new option (ref = NULL) to constrain the order of the tips. BUG FIXES o plot.phylo() used to fail with NAs in branch lengths: they are now ignored with a warning. o plot.phylo(, type = "unrooted", label.offset = 1) used to have no effect (unless lab4ut = "a" was used). o checkValidPhylo() failed with trees with a single node. o unroot() returned badly conformed trees in some situations. o richness.yule.test() failed when the speciation rate(s) is high: the code has been improved with better starting values for the optimization. o dist.topo() now checks that the trees have the same labels. The code is ~100 times faster for comparing 100 trees with 100 labels. o as.character.AAbin() striped matrix rownames (thanks to Shaun Wilkinson for the report). OTHER CHANGES o clustal(), clustalomega(), muscle(), and tcoffee() now give a more explicit error message when they fail to find the executable. The help page has been completed and gives some recommendations. o "DNAbin" objects larger than 2.1 billion bases (exactly 2^31-1) can now be handled by most functions in ape. The limitations are the same than in base R: 2^52 (~4.5*10^15) elements, and the numbers of rows and columns of matrices cannot exceed 2^31-1. read.dna() can now read FASTA files larger than 2.1 gigabytes. Two functions are still limited to 2.1 Gb: dist.dna and DNAbin2indel (the limit applies to the product of the numbers of sequences by the sequence length since they work with matrices). o As a side-effect of the above, several functions are slightly faster as several calls to .C have been replaced by .Call. o The following functions are now generic with methods for the classes "phylo" and "multiPhylo": di2multi, multi2di, Ntip, Nnode, Nedge, reorder, is.ultrametric, root, unroot, is.rooted, and is.binary (new function name, see below). o c.phylo() and c.multiPhylo() have now recursive = TRUE by default and try to return a list of single trees unless some objects are not trees or lists of trees. o image.DNAbin() and image.AAbin() now order rows from top to bottom. o read.GenBank() now uses HTTPS in place of HTTP. o ape now requires R 3.2.0 (or higher). o boot.phylo() now checks for duplicated labels. o checkAlignment() now prints the gap lengths that are on the borders of the alignment. DEPRECATED & DEFUNCT o is.binary.tree() is now obsolete and will be removed soon. It currently calls is.binary(). o The data sets landplants.newick and opsin were removed. CHANGES IN APE VERSION 3.5 NEW FEATURES o Eleven new functions are provided to work with the new class "AAbin": - trans() translates DNA to AA sequences; - complement() returns the reverse complement of DNA sequences; - Six methods to manipulate and display objects of class "AAbin": print, [, labels, as.character, as.phyDat, and image (alview() works also on AA sequences); - New generic as.AAbin() with a method for the class "character"; - dist.aa() computes pairwise Hamming distances; - AAsubst() returns the indices of polymporphic sites (similar to seg.sites() for "DNAbin"). o The new function checkAlignment performs diagnostics on a DNA alignment. o The new functions plotBreakLongEdges and drawSupportOnEdges help to plot and annotate trees. o del.colgapsonly() gains two options: threshold (1 by default) to delete columns with less than 100% of gaps, and freq.only (FALSE by default) to return only the number of gaps for each column. o The new function del.rowgapsonly does the same operation than del.colgapsonly but on the rows (with the same new options). BUG FIXES o root() did not calculate the basal branch length correctly (thanks to Liam Revell for the report). o clustal() failed with long labels. o as.DNAbin.character() now works with uppercase letters as well. o prop.clades() now returns a sensible results whatever the value of its option 'rooted' (fixed by Klaus). The code is also faster. o root(resolve.root = TRUE) failed in some situations. OTHER CHANGES o boot.phylo(phy, ....) has now by default rooted = is.rooted(phy). CHANGES IN APE VERSION 3.4 NEW FEATURES o The new function checkValidPhylo checks the internal structure of a "phylo" object. o The new function clustalomega calls Clustal-Omega from R. o The new function DNAbin2indel recodes DNA sequences with gaps as numeric values. o The new function alview prints an alignment in a user-friendly way in the console or a file. o Three new functions, label2table, stripLabel and abbreviateGenus, help to manage taxa labels. o phydataplot() now accepts style="boxplot" and style="dotchart", and has a new option 'scaling' when the data are on a very different scale than the branch lengths of the tree. o phydataplot() now accepts style="mosaic" to plot discrete or continuous variables in matrix form, and gains three options (width, continuous, funcol) that work with this option. o plot.phylo() has a new option (align.tip.label) to align the labels of the tips even when the tree is not ultrametric. BUG FIXES o parafit() did not work with data frames or "dist" objects: this is now fixed. o plot.phylo() did not manage space correctly when show.tip.label = FALSE and some tip labels were long (thanks to David Bapst for the report and to Liam Revell for the fix). o collapse.singles() failed in some situations, returning trees that could make R crash. The function was completely re-written. o In clustal(), the default path and executable names were not correctly set under Windows. o balance() failed after reordering a tree (bug reported by G. Valiente in January 2008). o seg.sites() failed to find some segregating sites in some specific situations: gaps and ambiguous nucleotides are now handled correctly. The code is slightly faster. o rtt() didn't find the optimal root if located on a long branch. o plot.phylo(, type = "f") sometimes failed when colouring the edges of a tree with multichotomies (thanks to François Michonneau for the fix). o rphylo() now returns better conformed trees. o drop.tip() sometimes shuffled node labels. o edgelabels() did not put the labels in the correct place with "fan" trees (fixed by Klaus). o reorder.phylo(, index.only = TRUE) sometimes returned the tree if it has a single node. o as.hclust.phylo() returned node heights divided by two. o ladderize() did not work with trees not in "cladewise" postorder. o dist.dna(, model = "indelblock") has been fixed. o as.hclust.phylo failed when trees had ties in their branching times. o root() failed with some trees returned by phangorn. OTHER CHANGES o Updated NAMESPACE file for r-devel. o root() gives a more explicit error message when the outgroup is not among the labels of the tree. CHANGES IN APE VERSION 3.3 NEW FEATURES o Two new functions contributed by Anthony Ives, binaryPGLMM and binaryPGLMM.sim, fit phylogenetic generalized linear mixed models to binary data and simulate such data. o The new function corphylo, contributed by Anthony Ives, calculates Pearson correlation coefficients for multiple continuous traits that may have phylogenetic signal. o The new function rphylo simulates trees under any time-dependent model of diversification using the method from Stadler (2011, Syst Biol 60: 676) which conditions the simulation on a fixed number of living species at present time. o The new function del.colgapsonly removes the columns of a DNA alignment that contain only gaps (useful when a small matrix is extracted from a large alignment). o There is a sort() method for objects of class "bitsplits". o plot.phylo(type = "fan", root.edge = TRUE) now draws the root edge. Besides, the root edge is now coloured like the other edges if they are all of the same colour (for all tree types). o dist.topo() now accepts lists of trees. BUG FIXES o prop.clades() failed with compressed lists of trees. o read.tree() did not ignore comments inside straight brackets (contrary to what is written in ape's FAQ) potentially resulting in wrongly conformed "phylo" objects. o dbdTime(x = 0, ...., conditional = TRUE) did not return 0. o parafit() required a fix when F2.stat could not be calculated. o write.nexus.data() did not write "DATATYPE=" correctly so that the output NEXUS file could not be read by some other programs (thanks to Robin Cristofari for the fix). o seg.sites() made R crash when all sequences ended with N's. o root(phy, node = Ntip(phy) + 1, resolve.root = TRUE) now returns an explicit error. OTHER CHANGES o plot.prop.part() has been improved: it now uses image() internally, and gains a new option (col). It is now much more efficient to display very large numbers of bipartitions. o as.character.DNAbin() and as.DNAbin.character() are much faster (contributed by Klaus Schliep). DEPRECATED & DEFUNCT o node.height.clado() has been removed. CHANGES IN APE VERSION 3.2 NEW FEATURES o The new function reconstruct, contributed by Manuela Royer-Carenzi and Gilles Didier, does ancestral character state reconstruction using new algorithms based on matrix computations. o The new function vcv2phylo, contributed by Simon Blomberg, transforms a variance-covariance matrix into a phylogenetic tree (ultrametric or not). o The new function def helps to (re)define attributes for plotting and annotating trees using taxon names, labels, or other vectors of character strings. o There are four new as.DNAbin functions (S3 methods) to convert objects storing DNA sequences in BioConductor. The supported (S4) classes are: DNAString, DNAStringSet, DNAMultipleAlignment, and PairwiseAlignmentsSingleSubject (the examples show how to convert from the class DNAStringSetList). o The new function nodepath finds paths of nodes in a tree. o axisPhylo() gains two options: root.time and backward (similar to ltt.plot). o write.nexus.data() now accepts matrices (it accepted only lists in previous versions). BUG FIXES o dist.dna() did not compute distances correctly if gamma = TRUE. o read.nexus() sometimes failed on files with no TRANSLATION block. o read.nexus() failed to read correctly tree names (thanks to Graham Gower for the fix). o multi2di() used to return trees with elements as doubles instead of integers (this was not a problem for most applications). o In the output of phymltest(), the model F84 and HKY were swapped (thanks to Luiz Max Fagundes de Carvalho for the fix). OTHER CHANGES o plot.phylo() now saves the element 'root.time' of the plotted tree (if present) in the environment .PlotPhyloEnv. o write.nexus.data() now writes correctly all "DNAbin" objects (matrices and lists). The help page has been clarified. CHANGES IN APE VERSION 3.1-4 BUG FIXES o A bug was fixed in which.edge(). OTHER CHANGES o This file is no more distributed with the sources of ape and can be found at: http://ape-package.ird.fr/NEWS. CHANGES IN APE VERSION 3.1-3 NEW FEATURES o The new function rtt contributed by Rosemary McCloskey roots a tree with dated tips in the location most compatible with the assumption of a strict molecular clock. BUG FIXES o root(, resolve.root = TRUE) did not manage node labels correctly in the case where 'outgroup' was already an outgroup of the tree. o read.dna(file, format = "fasta") works (again) with URIs (i.e., file = "http://..." or "ftp://..."). o rotate() made R crash if the tree was not in cladewise order. o which.edge() did not work correctly for trees not in cladewise order. o read.GenBank failed to read older GenBank records. CHANGES IN APE VERSION 3.1-2 NEW FEATURES o plot.phylo: the option 'open.angle' now works also when type = "radial". The option 'lab4ut' now works also for type = "radial" or "fan"; the default is now NULL since it will behave differently depending on 'type'. o ace(, type = "discrete") gains an option 'marginal' (FALSE by default). The default is now to do a joint reconstruction of ancestral states. BUG FIXES o chronos: the gradients were not correctly computed when the age of the root was unknown. o prop.part() did not work with objects of class composite like c("chronos", "phylo"). (Thanks to Steve Walker for the fix.) o bionjs: a bug was fixed in the C code. o read.GenBank() failed if a line starts with "ORIGIN" in the description of the sequence (thanks to Sofia Sal Bregua for the fix). o cophyloplot() works better with length.line = 0. o The internal function clado.build used to return trees with the 'edge' matrix stored as double instead of integer. OTHER CHANGES o base.freq() now checks the class of the data. CHANGES IN APE VERSION 3.1-1 BUG FIXES o rTraitCont: a bug was introduced in the previous release. OTHER CHANGES o ape now requires R 3.0.0 or higher. CHANGES IN APE VERSION 3.1 NEW FEATURES o Two new functions, bitsplits and countBipartitions, handle bipartitions (aka splits) more efficiently. o The new generic function as.prop.part helps to convert among classes of bipartitions. o Two new functions, phydataplot and ring, helps to graphically annotate trees. The help page has many examples. o The new function LLT draws the theoretical LTT-plot under specified values of speciation and extinction rates together with a prediction interval. The function has several options for flexible plotting. o predict.compar.gee() gains a new option, newdata, to predict values from new observations of the predictors. This works like most predict() methods. BUG FIXES o root: the fix introduced in the previous version was not correct. o compar.gee: using a user-defined link failed. o plot.phylo: the tip labels were often outside the plotting area with direction = "downward". o unique.multiPhylo() did not work correctly with compressed lists. This function now returns a vector of integers giving the correspondance among similar trees. OTHER CHANGES o add.scale.bar() now draws a longer bar by default. o In boot.phylo(), the data 'x' must be a matrix-like object (lists are no more accepted). o boot.phylo() uses by default (ie, if rooted = FALSE) the new function countBipartitions, and so should be much faster even with small sample sizes and especially if the number of bootstrap replicates is large. o plot.phylo(, type = "fan") now colours the arcs in the same way than the default type = "p". o All instances of DUP=FALSE in calls of .C have been removed. CHANGES IN APE VERSION 3.0-11 BUG FIXES o branching.times() now reorders the tree if needed. o root(, resolved.root = TRUE) sometimes misplaced the root when 'outgroup' was of length one. o nodelabels() didn't work with versions of R older than 3.0.0. OTHER CHANGES o The code of branching.times() is now ca. 3 times faster. CHANGES IN APE VERSION 3.0-10 NEW FEATURES o The new function rotateConstr rotates internal branches given a constraint on the order of the tips. o plot.phylo() has a new option, node.depth, to specify the depths of nodes when the tree is plotted without branch lengths. node.depth() has a new option, method, with the same effect. o ace() has a new option, use.eigen, used when type = "d" to avoid computing matrix exponentials. With other coding improvements, the function is now about five times faster. BUG FIXES o node.height() and node.height.clado() returned wrong values. o pie charts drawn by nodelabels(), and other functions, used to draw a "3-o'clock" segment when only one proportion was equal to one (fixed by Klaus). o plot(phy, main = "title...") now works again. OTHER CHANGES o Improved DESCRIPTION and NAMESPACE files. o The C routines are now registered. o The generic function as.igraph has been removed from ape as it is now defined in the package igraph. o node.height() has now an option 'clado.style' (FALSE by default). node.height.clado() will be removed soon. o The code of several functions has been improved. DEPRECATED & DEFUNCT o mst() has been moved to pegas. CHANGES IN APE VERSION 3.0-9 NEW FEATURES o getMRCA() is now documented. BUG FIXES o ace(, type = "d") now checks for zero or negative branch lengths. o ace(, type = "d") now catches the error if the SEs of the rates cannot be calculated (usually due to a poor model fit). o reorder.phylo() made R crash with badly conformed "phylo" trees. o drop.tip(phy, interactive = TRUE) did not work. o drop.tip() now returns NULL (with a warning) if all tips are dropped. o mixedFontLabel() failed when passed only one vector of labels. It also failed when labels included - or '. In some cases, spaces in labels were not treated correctly. o root(, resolve.root = TRUE) wrongly placed the new root node when the outgroup has more than one tip. This is fixed, and now the root node is placed at the MRCA of the outgroup (and not of the ingroup). o where() failed with matrices. CHANGES IN APE VERSION 3.0-8 NEW FEATURES o The new function ewLasso tests whether an incomplete set of distances uniquely determines the edge weights of a given unrooted topology using the 'Lasso' method by Dress et al. (2012, J. Math. Biol. 65:77). o ace() gains a new option 'use.expm' to use expm() from the package of the same name in place of matexpo(). BUG FIXES o read.dna(, "fasta") may add '\r' in labels: this is fixed. o prop.clades() returned wrong numbers when the tip labels of 'phy' are not in the same order than the list of trees (thanks to Rupert Collins for the report). o CADM.post() displayed "1" on the diagonal of the matrix of Mantel p-values. It now displays "NA" on the diagonal, indicating that no test of significance is computed between a distance matrix and itself. o rtree(n, rooted = FALSE) returned trees with an 'edge' matrix stored as doubles instead of integers for n > 4. OTHER CHANGES o The files CDAM.global.R and CDAM.post.R have been renamed CADM.global.R and CADM.post.R. o ace() has a new default for its option 'method': this is "REML" for continuous characters and "ML" for discrete ones. o ape does not import gee anymore so the latter doesn't need to be installed. CHANGES IN APE VERSION 3.0-7 NEW FEATURES o The new function chronos estimates chronograms by penalised likelihood and maximum likelihood with a completely reworked code and interface. There is a new function makeChronosCalib to set the calibration points easily. chronos() will eventually replace chronopl(). o The new function 'where' searches patterns in DNA sequences. o pic() gains an option 'rescaled.tree = FALSE' to return the tree with its branch lengths rescaled for the PIC calculation. o clustal(), muscle(), and tcoffee() gain an option 'original.ordering = TRUE' to ease the comparisons of alignments. o plot.phylo() has a new option, open.angle, used when plotting circular trees. o The new function read.FASTA reads FASTA files much faster and more efficiently. It is called internally by read.dna(, "fasta") or can be called directly. BUG FIXES o drop.tip() shuffled node labels on some trees. o axisPhylo() now works correctly with circular trees, and gives a sensible error message when type = "r" or "u". OTHER CHANGES o .compressTipLabel() is 10 times faster thanks to Joseph Brown. o base.freq() is now faster with lists. o as.matrix.DNAbin() should be faster and more efficient with lists; it now accepts vectors. CHANGES IN APE VERSION 3.0-6 NEW FEATURES o reorder.phylo() has a new order, "postorder", and a new option index.only = TRUE to return only the vector of indices (the tree is unmodified, see ?reorder.phylo for details). o The three new functions node.depth.edgelength, node.height, and node.height.clado make some internal code available from R. See ?node.depth (which was already documented) for details. BUG FIXES o reorder(, "pruningwise") made R crash if the rows of the edge matrix are in random order: this is now fixed. o drop.tip() sometimes shuffled node labels (thanks to Rebecca Best for the report). o drop.tip(phy, "") returned a tree with zero-length tip labels: it now returns the tree unchanged (thanks to Brian Anacker for the report). o plot.phylo() made R crash if the tree has zero-length tip labels: it now returns NULL (thanks again to Brian Anacker). OTHER CHANGES o dist.nodes() is now 6 to 10 times faster. o reorder(, "cladewise") is now faster. The change is not very visible for small trees (n < 1000) but this can be more than 1000 faster for big trees (n >= 1e4). o The attribute "order" of the objects of class "phylo" is now strongly recommended, though not mandatory. Most functions in ape should return a tree with this attribute correctly set. o dbd() is now vectorized on both arguments 'x' (number of species in clade) and 't' (clade age) to make likelihood calculations easier and faster. CHANGES IN APE VERSION 3.0-5 BUG FIXES o ace() should better catch errors when SEs cannot be computed. OTHER CHANGES o write.dna(format = "fasta") now conforms more closely to the FASTA standard thanks to François Michonneau. o print.DNAbin() does not print base compositions if there are more than one million nucleotides. CHANGES IN APE VERSION 3.0-4 BUG FIXES o read.dna() failed to read Phylip files if the first line used tabulations instead of white spaces. o read.dna() failed to read Phylip or Clustal files with less than 10 nucleotides. (See other changes in this function below.) OTHER CHANGES o read.dna() now requires at least one space (or tab) between the taxa names and the sequences (whatever the length of taxa names). write.dna() now follows the same rule. o The option 'seq.names' of read.dna has been removed. o The files ape-defunct.R and ape-defunct.Rd, which have not been modified for almost two years, have been removed. o The C code of bionj() has been reworked: it is more stable (by avoiding passing character strings), slightly faster (by about 20%), and numerically more accurate. o The C code of fastme.*() has been slightly modified and should be more stable by avoiding passing character strings (the results are identical to the previous versions). o The file src/newick.c has been removed. CHANGES IN APE VERSION 3.0-3 BUG FIXES o birthdeath() now catches errors and warnings much better so that a result is returned in most cases. OTHER CHANGES o Because of problems with character string manipulation in C, the examples in ?bionj and in ?fastme have been disallowed. In the meantime, these functions might be unstable. This will be solved for the next release. CHANGES IN APE VERSION 3.0-2 NEW FEATURES o The new function alex (alignment explorator) zooms in a DNA alignment and opens the result in a new window. BUG FIXES o compute.brtime() did not completely randomized the order of the branching times. o write.nexus() did not work correctly with rooted trees (thanks to Matt Johnson for the fix). o mltt.plot(, backward = FALSE) did not set the x-axis correctly. o A bug was introduced in prop.clades() with ape 3.0. The help page has been clarified relative to the use of the option 'rooted'. o mantel.test() printed a useless warning message. o plot.phylo(, direction = "downward") ignored 'y.lim'. o is.monophyletic() did not work correctly if 'tips' was not stored as integers. o prop.part() could make R crash if the first tree had many multichotomies. o njs(), bionjs(), and mvrs() now return an error if 'fs < 1'. o SDM() did not work correctly. The code has also been generally improved. OTHER CHANGES o The DESCRIPTION file has been updated. o The option 'original.data' of write.nexus() has been removed. o The files bionjs.c, mvr.c, mvrs.c, njs.c, triangMtd.c, and triangMtds.c have been improved which should fix some bugs in the corresponding functions. o dist.gene() now coerces input data frame as matrix resulting in much faster calculations (thanks to a suggestion by Markus Schlegel). CHANGES IN APE VERSION 3.0-1 NEW FEATURES o dist.dna() has two new models: "indel" and "indelblock". o bind.tree() now accepts 'position' > 0 when the trees have no banch length permitting to create a node in 'x' when grafting 'y' (see ?bind.tree for details). BUG FIXES o cophyloplot( , rotate = TRUE) made R hanged after a few clicks. Also the tree is no more plotted twice. o read.GenBank() has been updated to work with EFetch 2.0. o unroot() on trees in "pruningwise" order did not respect this attribute. CHANGES IN APE VERSION 3.0 NEW FEATURES o The three functions dyule, dbd, and dbdTime calculate the density probability (i.e., the distribution of the number of species) for the Yule, the constant and the time-dependent birth-beath models, respectively. These probabilities can be conditional on no extinction and/or on a log-scale. o plot.phylo() has a new option 'rotate.tree' to rotate unrooted, fan, or radial trees around the center of the plot. o boot.phylo() and prop.clades() have a new option rooted = FALSE. Note that the behaviour of prop.part() is unchanged. o edgelabels() has a new option 'date' to place labels on edges of chronograms using the time scale (suggestion by Rob Lanfear). BUG FIXES o In chronopl(), the code setting the initial dates failed in complicated settings (several dates known within intervals). This has been generally improved and should result in faster and more efficient convergence even in simple settings. o mantel.test() sometimes returned P-values > 1 with the default two-tailed test. o extract.clade() sometimes shuffled some tip labels (thanks to Ludovic Mallet and Mahendra Mariadassou for the fix). o clustal() should now find by default the executable under Windows. OTHER CHANGES o The code of yule() has been simplified and is now much faster for big trees. o The code of mantel.test() has been adjusted to be consistent with common permutation tests. o The C code of base.freq() has been improved and is now nearly 8 times faster. o The option 'original.data' of write.nexus() is now deprecated and will be removed in a future release. o The code of is.ultrametric() has been improved and is now 3 times faster. o The code of vcv.phylo() has been improved and is now 10 or 30 times faster for 100 or 1000 tips, respectively. Consequently, fitting models with PGLS will be faster overall. CHANGES IN APE VERSION 2.8 NEW FEATURES o Twelve new functions have been written by Andrei-Alin Popescu: additive, ultrametric, is.compatible, arecompatible, mvr, mvrs, njs, bionjs, SDM, treePop, triangMtd, triangMtd*. o A new class "bitsplits" has been created by Andrei-Alin Popescu to code splits (aka, bipartition). o There is a new generic function as.bitsplits with a method to convert from the class "prop.part" to the class "bitsplits". o The new function ltt.coplot plots on the same scales a tree and the derived LTT plot. o ltt.plot() has two new options: backward and tol. It can now handle non-ultrametic trees and its internal coding has been improved. The coordinates of the plot can now be computed with the new function ltt.plot.coords. BUG FIXES o prop.part() crashed if some trees had some multichotomies. CHANGES IN APE VERSION 2.7-3 NEW FEATURES o The new function compute.brtime computes and sets branching times. o mantel.test() has a new argument 'alternative' which is "two-sided" by default. Previously, this test was one-tailed with no possibility to change. o ace() can now do REML estimation with continuous characters, giving better estimates of the variance of the Brownian motion process. BUG FIXES o Branch lengths were wrongly updated with bind.tree(, where = , position = 0). (Thanks to Liam Revell for digging this bug out.) o Simulation of OU process with rTraitCont() did not work correctly. This now uses formula from Gillespie (1996) reduced to a BM process when alpha = 0 to avoid division by zero. The option 'linear' has been removed. o Cross-validation in chronopl() did not work when 'age.max' was used. o consensus(, p = 0.5) could return an incorrect tree if some incompatible splits occur in 50% of the trees (especially with small number of trees). o c() with "multiPhylo" did not work correctly (thanks to Klaus Schliep for the fix). o root() failed in some cases with an outgroup made of several tips. The help page has been clarified a bit. CHANGES IN APE VERSION 2.7-2 NEW FEATURES o There is a new class "evonet" to code evolutionary networks, with a constructor function evonet(), a print() and a plot() methods, and four conversion methods to the classes "phylo", "networx", "network", and "igraph". o The new function rTraitMult does multivariate traits simulation with user-defined models. o plot.phylo() has a new option 'plot = TRUE'. If FALSE, the tree is not plotted but the graphical device is set and the coordinates are saved as usual. o diversity.contrast.test() gains a fourth version of the test with method = "logratio"; the literature citations have been clarified. o add.scale.bar() has two new options, 'lwd' and 'lcol', to modify the aspect of the bar. o boot.phylo() now displays a progress bar by default (can be off if 'quiet = TRUE'). o There is a new predict() method for compar.gee(). BUG FIXES o bionj() made R crash if distances were too large. It now returns an error if at least one distance is greater than 100. o drop.tip() returned a wrong tree if 'tip' was of zero length. o read.nexus.data() failed with URLs. o boot.phylo() returned overestimated support values in the presence of identical or nearly identical sequences. OTHER CHANGES o The data bird.families, bird.orders, cynipids, and woodmouse are now provided as .rda files. CHANGES IN APE VERSION 2.7-1 NEW FEATURES o The new function trex does tree exploration with multiple graphical devices. o The new function kronoviz plots several rooted (dated) trees on the scale scale. o identify.phylo() has a new option 'quiet' (FALSE by default). BUG FIXES o A bug was introduced in read.nexus() in ape 2.7. o image.DNAbin() did not colour correctly the bases if there were some '-' and no 'N'. o .compressTipLabel() failed with a list with a single tree. o identify.phylo() returned a wrong answer when the x- and y-scales are very different. o write.nexus() failed with lists of trees with compressed labels. OTHER CHANGES o identify.phylo() now returns NULL if the user right- (instead of left-) clicks (an error was returned previously). o read.nexus() should be robust to commands inserted in the TREES block. CHANGES IN APE VERSION 2.7 NEW FEATURES o There is a new image() method for "DNAbin" objects: it plots DNA alignments in a flexible and efficient way. o Two new functions as.network.phylo and as.igraph.phylo convert trees of class "phylo" into these respective network classes defined in the packages of the same names. o The three new functions clustal, muscle, and tcoffee perform nucleotide sequence alignment by calling the external programs of the same names. o Four new functions, diversity.contrast.test, mcconwaysims.test, richness.yule.test, and slowinskiguyer.test, implement various tests of diversification shifts using sister-clade comparisons. o base.freq() gains an option 'all' to count all the possible bases including the ambiguous ones (defaults to FALSE). o write.nexus() now writes tree names in the NEXUS file if given a list of trees with names. BUG FIXES o prop.part() failed in some situations with unrooted trees. o read.nexus() shuffled node labels when a TRANSLATE block was present. o varCompPhylip() did not work if 'exec' was specified. o bind.tree() shuffled node labels when position > 0 and 'where' was not the root. OTHER CHANGES o BaseProportion in src/dist_dna.c has been modified. o A number of functions in src/tree_build.c have been modified. o The matching representation has now only two columns as the third column was redundant. CHANGES IN APE VERSION 2.6-3 NEW FEATURES o rTraitCont() and rTraitDisc() gains a '...' argument used with user-defined models (suggestion by Gene Hunt). BUG FIXES o as.hclust.phylo() now returns an error with unrooted trees. o as.hclust.phylo() failed with trees with node labels (thanks to Jinlong Zhang for pointing this bug out). o read.dna(, "fasta") failed if sequences were not all of the same length. o plot.phylo() did not recycle values of 'font', 'cex' and 'tip.color' correctly when type = "fan" or "radial". o plot.phylo() ignored 'label.offset' when type = "radial", "fan", or "unrooted" with lab4ut = "axial" (the placement of tip labels still needs to be improved with lab4ut = "horizontal"). OTHER CHANGES o In drop.fossil() the default tol = 0 has been raised to 1e-8. o The help command ?phylo now points to the man page of read.tree() where this class is described. Similarly, ?matching points to the man page of as.matching(). CHANGES IN APE VERSION 2.6-2 NEW FEATURES o Two new functions, pic.ortho and varCompPhylip, implements the orthonormal contrasts of Felsenstein (2008, Am Nat, 171:713). The second function requires Phylip to be installed on the computer. o bd.ext() has a new option conditional = TRUE to use probabilities conditioned on no extinction for the taxonomic data. BUG FIXES o write.tree() failed to output correctly tree names. o dist.nodes() returned duplicated column(s) with unrooted and/or multichotomous trees. o mcmc.popsize() terminated unexpectedly if the progress bar was turned off. o prop.part(x) made R frozen if 'x' is of class "multiPhylo". o Compilation under Mandriva failed (thanks to Jos Käfer for the fix). o drop.tip() shuffled tip labels with subtree = TRUE or trim.internal = FALSE. o Objects returned by as.hclust.phylo() failed when analysed with cutree() or rect.hclust(). o write.tree() did not output correctly node labels (thanks to Naim Matasci and Jeremy Beaulieu for the fix). o ace(type = "discrete") has been improved thanks to Naim Marasci and Jeremy Beaulieu. CHANGES IN APE VERSION 2.6-1 NEW FEATURES o The new function speciesTree calculates the species tree from a set of gene trees. Several methods are available including maximum tree and shallowest divergence tree. BUG FIXES o A bug introduced in write.tree() with ape 2.6 has been fixed. o as.list.DNAbin() did not work correctly with vectors. o as.hclust.phylo() failed with trees with node labels (thanks to Filipe Vieira for the fix). CHANGES IN APE VERSION 2.6 NEW FEATURES o The new functions rlineage and rbdtree simulate phylogenies under any user-defined time-dependent speciation-extinction model. They use continuous time algorithms. o The new function drop.fossil removes the extinct species from a phylogeny. o The new function bd.time fits a user-defined time-dependent birth-death model. It is a generalization of yule.time() taking extinction into account. o The new function MPR does most parsimonious reconstruction of discrete characters. o The new function Ftab computes the contingency table of base frequencies from a pair of sequences. o There is now an 'as.list' method for the class "DNAbin". o dist.dna() can compute the number of transitions or transversions with the option model = "Ts" or model = "Tv", respectively. o [node|tip|edge]labels() gain three options with default values to control the aspect of thermometers: horiz = TRUE, width = NULL, and height = NULL. o compar.gee() has been improved with the new option 'corStruct' as an alternative to 'phy' to specify the correlation structure, and calculation of the QIC (Pan 2001, Biometrics). The display of the results has also been improved. o read.GenBank() has a new option 'gene.names' to return the name of the gene (FALSE by default). BUG FIXES o extract.clade() sometimes shuffled the tip labels. o plot.phylo(type = "unrooted") did not force asp = 1 (thanks to Klaus Schliep for the fix) o dist.dna(model = "logdet") used to divide distances by 4. The documentation has been clarified on the formulae used. OTHER CHANGES o rTraitCont(model = "OU") has an option 'linear = TRUE' to possibly change the parameterisation (see ?rTraitCont for details). o pic() now returns a vector with the node labels of the tree (if available) as names. o write.tree() and read.tree() have been substantially improved thanks to contributions by Klaus Schliep. CHANGES IN APE VERSION 2.5-3 NEW FEATURES o The new function mixedFontLabel helps to make labels with bits of text to be plotted in different fonts. o There are now replacement operators for [, [[, and $ for the class "multiPhylo" (i.e., TREES[11:20] <- rmtree(10, 100)). They possibly check that the tip labels are the same in all trees. o Objects of class "multiPhylo" can be built with c(): there are methods for the classes "phylo" and "multiPhylo". o The internal functions .compressTipLabel and .uncompressTipLabel are now documented. BUG FIXES o bind.tree(x, y, where, position = 0) did not work correctly if 'y' was a single-edge tree and 'where' was a tip. o rTraitCont() did not use the square-root of branch lengths when simulating a Brownian motion model. CHANGES IN APE VERSION 2.5-2 NEW FEATURES o There is now a print method for results from ace(). o There is a labels() method for objects of class "DNAbin". o read.dna() has a new option 'as.matrix' to possibly force sequences in a FASTA file to be stored in a matrix (see ?read.dna for details). BUG FIXES o as.phylo.hclust() used to multiply edge lengths by 2. o A minor bug was fixed in rTraitDisc(). o ace() sometimes failed (parameter value was NaN and the optimisation failed). DEPRECATED & DEFUNCT o evolve.phylo() and plot.ancestral() have been removed. o chronogram(), ratogram(), and NPRS.criterion() have been removed. OTHER CHANGES o nj() has been improved and is now about 30% faster. o The default option 'drop' of [.DNAbin has been changed to FALSE to avoid dropping rownames when selecting a single sequence. o print.DNAbin() has been changed to summary.DNAbin() which has been removed. CHANGES IN APE VERSION 2.5-1 NEW FEATURES o The new function stree generates trees with regular shapes. o It is now possible to bind two trees with x + y (see ?bind.tree for details). o drop.tip(), extract.clade(), root(), and bind.tree() now have an 'interactive' option to make the operation on a plotted tree. o cophyloplot() gains two new arguments 'lwd' and 'lty' for the association links; they are recycled like 'col' (which wasn't before). BUG FIXES o rTraitDisc() did not use its 'freq' argument correctly (it was multiplied with the rate matrix column-wise instead of row-wise). o [node|tip|edge]labels(thermo = ) used to draw empty thermometers with NA values. Nothing is drawn now like with 'text' or 'pch'. The same bug occurred with the 'pie' option. o A bug was fixed in compar.ou() and the help page was clarified. o bind.tree() has been rewritten fixing several bugs and making it more efficient. o plot.phylo(type = "p") sometimes failed to colour correctly the vertical lines representing the nodes. o plot.phylo(direction = "l", x.lim = 30) failed to plot the branches in the correct direction though the tip labels were displayed correctly. OTHER CHANGES o The c, cbind, and rbind methods for "DNAbin" objetcs now check that the sequences are correctly stored (in a list for c, in a matrix for the two other functions). CHANGES IN APE VERSION 2.5 NEW FEATURES o The new function parafit by Pierre Legendre tests for the coevolution between hosts and parasites. It has a companion function, pcoa, that does principal coordinate decomposition. The latter has a biplot method. o The new function lmorigin by Pierre Legendre performs multiple regression through the origin with testing by permutation. o The new functions rTraitCont and rTraitDisc simulate continuous and discrete traits under a wide range of evolutionary models. o The new function delta.plot does a delta plot following Holland et al. (2002, Mol. Biol. Evol. 12:2051). o The new function edges draws additional branches between any nodes and/or tips on a plotted tree. o The new function fancyarrows enhances arrows from graphics with triangle and harpoon heads; it can be called from edges(). o add.scale.bar() has a new option 'ask' to draw interactively. o The branch length score replaces the geodesic distance in dist.topo. o Three new data sets are included: the gopher-lice data (gopher.D), SO2 air pollution in 41 US cities (lmorigin.ex1, from Sokal & Rohlf 1995), and some host-parasite specificity data (lmorigin.ex2, from Legendre & Desdevises 2009). BUG FIXES o add.scale.bar() drew the bar outside the plotting region with the default options with unrooted or radial trees. o dist.topo() made R stuck when the trees had different sizes (thanks to Otto Cordero for the fix). OTHER CHANGES o The geodesic distance has been replaced by the branch length score in dist.topo(). CHANGES IN APE VERSION 2.4-1 NEW FEATURES o rtree() and rcoal() now accept a numeric vector for the 'br' argument. o vcv() is a new generic function with methods for the classes "phylo" and "corPhyl" so that it is possible to calculate the var-cov matrix for "transformation models". vcv.phylo() can still be used for trees of class "phylo"; its argument 'cor' has been renamed 'corr'. BUG FIXES o bind.tree() failed when 'y' had no root edge. o read.nexus() shuffled tip labels when the trees have no branch lengths and there is a TRANSLATE block. o read.nexus() does not try to translate node labels if there is a translation table in the NEXUS file. See ?read.nexus for a clarification on this behaviour. o plot.multiPhylo() crashed R when plotting a list of trees with compressed tip labels. o write.nexus() did not translate the taxa names when asked for. o plot.phylo(type = "fan") did not rotate the tip labels correctly when the tree has branch lengths. o ace(type = "continuous", method = "ML") now avoids sigma² being negative (which resulted in an error). o nj() crashed with NA/NaN in the distance matrix: an error in now returned. CHANGES IN APE VERSION 2.4 NEW FEATURES o base.freq() has a new option 'freq' to return the counts; the default is still to return the proportions. BUG FIXES o seg.sites() did not handle ambiguous nucleotides correctly: they are now ignored. o plot(phy, root.edge = TRUE) failed if there was no $root.edge in the tree: the argument is now ignored. o add.scale.bar() failed when 'x' and 'y' were given (thanks to Janet Young for the fix). OTHER CHANGES o Trying to plot a tree with a single tip now returns NULL with a warning (it returned an error previously). o The way lines representing nodes are coloured in phylograms has been modified (as well as their widths and types) following some users' request; this is only for dichotomous nodes. o The argument 'adj' in [node][tip][edge]labels() now works when using 'pie' or 'thermo'. o A more informative message error is now returned by dist.dna() when 'model' is badly specified (partial matching of this argument is done now). o Deprecated functions are now listed in a help page: see help("ape-defunct") with the quotes. DEPRECATED & DEFUNCT o The functions heterozygosity, nuc.div, theta.h, theta.k and theta.s have been moved from ape to pegas. o The functions mlphylo, DNAmodel and sh.test have been removed. CHANGES IN APE VERSION 2.3-3 BUG FIXES o add.scale.bar() always drew a horizontal bar. o zoom() shuffled tips with unrooted trees. o write.nexus() failed to write correctly trees with a "TipLabel" attribute. o rcoal() failed to compute branch lengths with very large n. o A small bug was fixed in compar.cheverud() (thanks to Michael Phelan for the fix). o seg.sites() failed when passing a vector. o drop.tip() sometimes shuffled tip labels. o root() shuffled node labels with 'resolve.root = TRUE'. CHANGES IN APE VERSION 2.3-2 BUG FIXES o all.equal.phylo() did not compare unrooted trees correctly. o dist.topo(... method = "PH85") did not treat unrooted trees correctly (thanks to Tim Wallstrom for the fix). o root() sometimes failed to test for the monophyly of the outgroup correctly. o extract.clade() sometimes included too many edges. o vcv.phylo() did not work correctly when the tree is in "pruningwise" order. o nj() did not handle correctly distance matrices with many 0's. The code has also been significantly improved: 7, 70, 160 times faster with n = 100, 500, 1000, respectively. CHANGES IN APE VERSION 2.3-1 NEW FEATURES o The new function is.monophyletic tests the monophyly of a group. o There is now a c() method for lists of class "DNAbin". o yule.cov() now fits the null model, and its help page has been corrected with respect to this change. o drop.tip() has a new option 'rooted' to force (or not) a tree to be treated as (un)rooted. BUG FIXES o dist.gene() failed on most occasions with the default pairwise.deletion = FALSE. o read.tree() failed to read correctly the tree name(s). o boot.phylo() now treats correctly data frames. o del.gaps() did not copy the rownames of a matrix. o A small bug was fixed in CDAM.global(). o ace() failed with large data sets. Thanks to Rich FitzJohn for the fix. With other improvements, this function is now about 6 times faster. o write.tree() failed with objects of class "multiPhylo". o drop.tip(, subtree = TRUE) sometimes shuffled tip labels. OTHER CHANGES o [.multiPhylo and [.DNAbin now respect the original class. o Instances of the form class(phy) == "phylo" have been replaced by inherits(phy, "phylo"). o rcoal() is now faster. DEPRECATED & DEFUNCT o klastorin() has been removed. CHANGES IN APE VERSION 2.3 NEW FEATURES o The new functions CADM.global and CADM.post, contributed by Pierre Legendre, test the congruence among several distance matrices. o The new function yule.time fits a user-defined time-dependent Yule model by maximum likelihood. o The new function makeNodeLabel creates and/or modifies node labels in a flexible way. o read.tree() and write.tree() have been modified so that they can handle individual tree names. o plot.phylo() has a new argument 'edge.lty' that specifies the types of lines used for the edges (plain, dotted, dashed, ...) o phymltest() has been updated to work with PhyML 3.0.1. BUG FIXES o drop.tip() shuffled tip labels in some cases. o drop.tip() did not handle node.label correctly. o is.ultrametric() now checks the ordering of the edge matrix. o ace() sometimes returned negative values of likelihoods of ancestral states (thanks to Dan Rabosky for solving this long lasting bug). OTHER CHANGES o The data set xenarthra has been removed. CHANGES IN APE VERSION 2.2-4 BUG FIXES o The bug fix in read.nexus() in version 2.2-3 was wrong: this is now fixed. (Thanks to Peter Wragg for the fix!) o A warning message occurred for no reason with ace(method="GLS"). OTHER CHANGES o There is now a general help page displayed with '?ape'. CHANGES IN APE VERSION 2.2-3 NEW FEATURES o The new function extract.clade extracts a clade from a tree by specifying a node number or label. o fastme.bal() has two new options 'spr' and 'tbr' to perform tree operations of the same names. o dist.dna() can now return the number of site differences by specifying model="N". BUG FIXES o chronopl() did not work with CV = TRUE. o read.nexus() did not work correctly in some situations (trees on multiple lines with different numbers of lines and/or with comments inserted within the trees). o ltt.plot(), ltt.lines(), and mltt.plot() did not count correctly the number of lineages with non-binary trees. OTHER CHANGES o ape has now a namespace. o drop.tip() has been improved: it should be much faster and work better in some cases (e.g., see the example in ?zoom). CHANGES IN APE VERSION 2.2-2 NEW FEATURES o dist.gene() has been substantially improved and gains an option 'pairwise.deletion'. o cbind.DNAbin() has a new option 'fill.with.gaps' and is now more flexible. BUG FIXES o prop.part() failed with a single tree with the default option 'check.labels = TRUE'. o summary.DNAbin() failed to display correctly the summary of sequence lengths with lists of sequences of 10,000 bases or more (because summary.default uses 4 significant digits by default). o read.nexus() failed to read a file with a single tree with line breaks in the Newick string. o del.gaps() returned a list of empty sequences when there were no gaps. OTHER CHANGES o phymltest() has been updated for PhyML 3.0 and gains an option 'append', whereas the option 'path2exec' has been removed. o rbind.DNAbin() and cbind.DNAbin() now accept a single matrix which is returned unchanged (instead of an error). o The data sets bird.orders and bird.families are now stored as Newick strings; i.e., the command data(bird.orders) calls read.tree(). CHANGES IN APE VERSION 2.2-1 NEW FEATURES o The new function makeLabel() helps to modify labels of trees, lists of trees, or DNA sequences, with several utilities to truncate and/or make them unique, substituting some characters, and so on. o The new function del.gaps() removes insertion gaps ("-") in a set of DNA sequences. o read.dna() can now read Clustal files (*.aln). BUG FIXES o root() failed with 'resolve.root = TRUE' when the root was already the specified root. o Several bugs were fixed in mlphylo(). o collapsed.singles() did not propagate the 'Nnode' and 'node.labels' elements (thanks to Elizabeth Purdom for the fix). o read.nexus() failed to remove correctly the comments within trees. o read.nexus() failed to read a file with a single tree and no translation of tip labels. o read.nexus() failed to place correctly tip labels when reading a single tree with no edge lengths. o A bug was fixed in sh.test(). OTHER CHANGES o unique.multiPhylo() is faster thanks to a suggestion by Vladimir Minin. o The option 'check.labels' of consensus() and prop.part() is now TRUE by default. o write.dna() now does not truncate names to 10 characters with the Phylip formats. CHANGES IN APE VERSION 2.2 NEW FEATURES o Four new functions have been written by Damien de Vienne for the graphical exploration of large trees (cophyloplot, subtrees, subtreeplot), and to return the graphical coordinates of tree (without plotting). o The new functions corPagel and corBlomberg implement the Pagel's "lambda" and Blomberg et al.'s "ACDC" correlation structures. o chronopl() has been improved and gains several options: see its help page for details. o boot.phylo() has now an option 'trees' to possibly return the bootstraped trees (the default is FALSE). o prop.part() has been improved and should now be faster in all situations. BUG FIXES o read.dna() failed if "?" occurred in the first 10 sites of the first sequence. o The x/y aspect of the plot is now respected when plotting a circular tree (type = "r" or "f"). o Drawing the tip labels sometimes failed when plotting circular trees. o zoom() failed when tip labels were used instead of their numbers (thanks to Yan Wong for the fix). o drop.tip() failed with some trees (fixed by Yan Wong). o seg.sites() failed with a list. o consensus() failed in some cases. The function has been improved as well and is faster. CHANGES IN APE VERSION 2.1-3 BUG FIXES o A bug in read.nexus() made the Windows R-GUI crash. o An error was fixed in the computation of ancestral character states by generalized least squares in ace(). o di2multi() did not modify node labels correctly. o multi2di() failed if the tree had its attribute "order" set to "cladewise". CHANGES IN APE VERSION 2.1-2 NEW FEATURES o There three new methods for the "multiPhylo" class: str, $, and [[. o root() gains the options 'node' and 'resolve.root' (FALSE by default) as well as its code being improved. o mltt.plot() has now an option 'log' used in the same way than in plot.default(). BUG FIXES o mltt.plot() failed to display the legend with an unnamed list of trees. o nodelabels() with pies now correcly uses the argument 'cex' to draw symbols of different sizes (which has worked already for thermometers). o read.nexus() generally failed to read very big files. OTHER CHANGES o The argument 'family' of compar.gee() can now be a function as well as a character string. o read.tree() and read.nexus() now return an unnamed list if 'tree.names = NULL'. o read.nexus() now returns a modified object of class "multiPhylo" when there is a TRANSLATE block in the NEXUS file: the individual trees have no 'tip.label' vector, but the list has a 'TipLabel' attribute. The new methods '$' and '[[' set these elements correctly when extracting trees. CHANGES IN APE VERSION 2.1-1 NEW FEATURES o The new function rmtree generates lists of random trees. o rcoal() now generates a genuine coalescent tree by default (thanks to Vladimir Minin for the code). BUG FIXES o nuc.div() returned an incorrect value with the default pairwise.deletion = FALSE. OTHER CHANGES o The internal codes of bionj(), fastme.bal(), and fastme.ols() have been improved so that they are stabler and faster. o R packages used by ape are now loaded silently; lattice and gee are loaded only when needed. CHANGES IN APE VERSION 2.1 NEW FEATURES o The new function identify.phylo identifies clades on a plotted tree using the mouse. o It is now possible to subset a list of trees (object of class "multiPhylo") with "[" while keeping its class correct. o The new function as.DNAbin.alignment converts DNA sequences stored in the "alignment" format of the package seqinr into an object of class "DNAbin". o The new function weight.taxo2 helps to build similarity matrices given two taxonomic levels (usually called by other functions). o write.tree() can now take a list of trees (class "multiPhylo") as its main argument. o plot.correlogram() and plot.correlogramList() have been improved, and gain several options (see the help page for details). A legend is now plotted by default. BUG FIXES o dist.dna() returned some incorrect values with `model = "JC69"' and `pairwise.deletion = TRUE'. This affected only the distances involving sequences with missing values. (Thanks to Bruno Toupance for digging this bug out.) o write.tree() failed with some trees: this is fixed by removing the `multi.line' option (trees are now always printed on a single line). o read.nexus() did not correctly detect trees with multiple root edges (see OTHER CHANGES). OTHER CHANGES o The code of mlphylo() has been almost entirely rewritten, and should be much stabler. The options have been also greatly simplified (see ?mlphylo and ?DNAmodel for details). o The internal function nTips has been renamed klastorin_nTips. o The code of is.ultrametric() contained redundancies and has been cleaned-up. o The code of Moran.I() and of correlogram.formula() have been improved. o read.tree() and read.nexus() now return an error when trying to read a tree with multiple root edges (see BUG FIXES). The correction applied in previous version did not work in all situations. o The class c("multi.tree", "phylo") has been renamed "multiPhylo". DOCUMENTATION o There is now a vignette in ape: see vignette("MoranI", "ape"). DEPRECATED & DEFUNCT o as.matching() and as.phylo.matching() do not support branch lengths. o correlogram.phylo() and discrete.dist() have been removed. CHANGES IN APE VERSION 2.0-2 NEW FEATURES o The new function matexpo computes the exponential of a square matrix. o The new function unique.multi.tree removes duplicate trees from a list. o yule() has a new option `use.root.edge = FALSE' that specifies to ignore, by default, the root edge of the tree if it exists. BUG FIXES o which.edge() failed when the index of a single terminal edge was looked for. o In diversi.time(), the values returned for model C were incorrect. o A bug was fixed in yule() that affected the calculation of the likelihood in the presence of ties in the branching times. o There was a bug in the C function mat_expo4x4 affecting the calculations of the transition probabilities for models HKY and GTR in mlphylo(). o A small bug was fixed in as.matrix.DNAbin (thanks to James Bullard). o rtree() did not `shuffle' the tip labels by default, so only a limited number of labelled topologies could be generated. CHANGES IN APE VERSION 2.0-1 NEW FEATURES o The three new functions bionj, fastme.ols, and fastme.bal perform phylogeny estimation by the BIONJ and fastME methods in OLS and balanced versions. This is a port to R of previous previous programs done by Vincent Lefort. o The new function chronoMPL performs molecular dating with the mean path lengths method of Britton et al. (2002, Mol. Phyl. Evol. 24: 58). o The new function rotate, contributed by Christoph Heibl, swaps two clades connected to the same node. It works also with multichotomous nodes. o The new `method' as.matrix.DNAbin() may be used to convert easily DNA sequences stored in a list into a matrix while keeping the names and the class. BUG FIXES o chronopl() failed when some branch lengths were equal to zero: an error message is now returned. o di2multi() failed when there was a series of consecutive edges to remove. CHANGES IN APE VERSION 1.10-2 NEW FEATURES o plot.phylo() can now plot circular trees: the option is type = "fan" or type = "f" (to avoid the ambiguity with type = "c"). o prop.part() has a new option `check.labels = FALSE' which allows to considerably speed-up the calculations of bipartitions. As a consequence, calculations of bootstrap values with boot.phylo() should be much faster. BUG FIXES o read.GenBank() did not return correctly the list of species as from ape 1.10: this is fixed in this version o Applying as.phylo() on a tree of class "phylo" failed: the object is now returned unchanged. CHANGES IN APE VERSION 1.10-1 NEW FEATURES o The three new functions Ntip, Nnode, and Nedge return, for a given tree, the number of tips, nodes, or edges, respectively. BUG FIXES o read.nexus() did not set correctly the class of the returned object when reading multiple trees. o mllt.plot() failed with objects of class c("multi.tree", "phylo"). o unroot() did not work correctly in most cases. o reorder.phylo() made R freeze in some occasions. o Plotting a tree in pruningwise order failed. o When plotting an unrooted tree, the tip labels where not all correctly positioned if the option `cex' was used. CHANGES IN APE VERSION 1.10 NEW FEATURES o Five new `method' functions have been introduced to manipulate DNA sequences in binary format (see below). o Three new functions have been introduced to convert between the new binary and the character formats. o The new function as.alignment converts DNA sequences stored as single characters into the class "alignment" used by the package seqinr. o read.dna() and read.GenBank() have a new argument `as.character' controlling whether the sequences are returned in binary format or as character. BUG FIXES o root() failed when the tree had node labels: this is fixed. o plot.phylo() did not correctly set the limits on the y-axis with the default setting: this is fixed. o dist.dna() returned a wrong result for the LogDet, paralinear, and BH87 models with `pairwise.deletion = TRUE'. OTHER CHANGES o DNA sequences are now internally stored in a binary format. See the document "A Bit-Level Coding Scheme for Nucleotides" for the details. Most functions analyzing DNA functions have been modified accordingly and are now much faster (dist.dna is now ca. 60 times faster). CHANGES IN APE VERSION 1.9-4 BUG FIXES o A bug was fixed in edgelabels(). o as.phylo.hclust() did not work correctly when the object of class "hclust" has its labels set to NULL: the returned tree has now its tip labels set to "1", "2", ... o consensus could fail if some tip labels are a subset of others (e.g., "a" and "a_1"): this is now fixed. o mlphylo() failed in most cases if some branch lengths of the initial tree were greater than one: an error message is now issued. o mlphylo() failed in most cases when estimating the proportion of invariants: this is fixed. CHANGES IN APE VERSION 1.9-3 NEW FEATURES o The new function edgelabels adds labels on the edge of the tree in the same way than nodelabels or tiplabels. BUG FIXES o multi2di() did not handle correctly branch lengths with the default option `random = TRUE': this is now fixed. o A bug was fixed in nuc.div() when using pairwise deletions. o A bug occurred in the analysis of bipartitions with large numbers of large trees, with consequences on prop.part, prop.clades, and boot.phylo. o The calculation of the Billera-Holmes-Vogtmann distance in dist.topo was wrong: this has been fixed. CHANGES IN APE VERSION 1.9-2 NEW FEATURES o The new function ladderize reorganizes the internal structure of a tree to plot them left- or right-ladderized. o The new function dist.nodes computes the patristic distances between all nodes, internal and terminal, of a tree. It replaces the option `full = TRUE' of cophenetic.phylo (see below). BUG FIXES o A bug was fixed in old2new.phylo(). o Some bugs were fixed in chronopl(). o The edge colours were not correctly displayed by plot.phylo (thank you to Li-San Wang for the fix). o cophenetic.phylo() failed with multichotomous trees: this is fixed. OTHER CHANGES o read.dna() now returns the sequences in a matrix if they are aligned (interleaved or sequential format). Sequences in FASTA format are still returned in a list. o The option `full' of cophenetic.phylo() has been removed because it could not be used from the generic. DEPRECATED & DEFUNCT o rotate() has been removed; this function did not work correctly since ape 1.9. CHANGES IN APE VERSION 1.9-1 BUG FIXES o Trees with a single tip were not read correctly in R as the element `Nnode' was not set: this is fixed. o unroot() did not set correctly the number of nodes of the unrooted tree in most cases. o read.GenBank() failed when fetching very long sequences, particularly of the BX-series. o A bug was introduced in read.tree() with ape 1.9: it has been fixed CHANGES IN APE VERSION 1.9 NEW FEATURES o There are two new print `methods' for trees of class "phylo" and lists of trees of class "multi.tree", so that they are now displayed in a compact and informative way. o There are two new functions, old2new.phylo and new2old.phylo, for converting between the old and new coding of the class "phylo". o dist.dna() has three new models: Barry and Hartigan ("BH87"), LogDet ("logdet"), and paralinear ("paralin"). o compute.brlen() has been extended: several methods are now available to compute branch lengths. o write.dna() can now handle matrices as well as lists. BUG FIXES o cophenetic.phylo() sometimes returned a wrong result with multichotomous trees: this is fixed. o rotate() failed when a single tip was specified: the tree is now returned unchanged. o ace() did not return the correct index matrix with custom models: this is fixed. o multi2di() did not work correctly when resolving multichotomies randomly: the topology was always the same, only the arrangement of clades was randomized: this is fixed. This function now accepts trees with no branch lengths. o The output of diversi.gof() was blurred by useless prints when a user distribution was specified. This has been corrected, and the help page of this function has been expanded. OTHER CHANGES o The internal structure of the class "phylo" has been changed: see the document "Definition of Formats for Coding Phylogenetic Trees in R" for the details. In addition, the code of most functions has been improved. o Several functions have been improved by replacing some R codes by C codes: pic, plot.phylo, and reorder.phylo. o There is now a citation information: see citation("ape") in R. o write.tree() now does not add extra 0's to branch lengths so that 1.23 is printed "1.23" by default, not "1.2300000000". o The syntax of bind.tree() has been simplified. This function now accepts trees with no branch lengths, and handles correctly node labels. o The option `as.numeric' of mrca() has been removed. o The unused options `format' and `rooted' of read.tree() have been removed. o The unused option `format' of write.tree() has been removed. o The use of node.depth() has been simplified. CHANGES IN APE VERSION 1.8-5 NEW FEATURES o Two new functions read.nexus.data() and write.nexus.data(), contributed by Johan Nylander, allow to read and write molecular sequences in NEXUS files. o The new function reorder.phylo() reorders the internal structure of a tree of class "phylo". It is used as the generic, e.g., reorder(tr). o read.tree() and read.nexus() can now read trees with a single edge. o The new data set `cynipids' supplies a set of protein sequences in NEXUS format. BUG FIXES o The code of all.equal.phylo() has been completely rewritten (thanks to Benoît Durand) which fixes several bugs. o read.tree() and read.nexus() now checks the labels of the tree to remove or substitute any characters that are illegal in the Newick format (parentheses, etc.) o A negative P-value could be returned by mantel.test(): this is now fixed. CHANGES IN APE VERSION 1.8-4 NEW FEATURES o The new function sh.test() computes the Shimodaira- Hasegawa test. o The new function collapse.singles() removes the nodes with a single descendant from a tree. o plot.phylo() has a new argument `tip.color' to specify the colours of the tips. o mlphylo() has now an option `quiet' to control the display of the progress of the analysis (the default is FALSE). BUG FIXES o read.dna() did not read correctly sequences in sequential format with leading alignment gaps "-": this is fixed. o ace() returned a list with no class so that the generic functions (anova, logLik, ...) could not be used directly. This is fixed as ace() now returns an object of class "ace". o anova.ace() had a small bug when computing the number of degrees of freedom: this is fixed. o mlphylo() did not work when the sequences were in a matrix or a data frame: this is fixed. o rtree() did not work correctly when trying to simulate an unrooted tree with two tips: an error message is now issued. OTHER CHANGES o The algorithm of rtree() has been changed: it is now about 40, 100, and 130 times faster for 10, 100, and 1000 tips, respectively. CHANGES IN APE VERSION 1.8-3 NEW FEATURES o There are four new `method' functions to be used with the results of ace(): logLik(), deviance(), AIC(), and anova(). o The plot method of phymltest has two new arguments: `main' to change the title, and `col' to control the colour of the segments showing the AIC values. o ace() has a new argument `ip' that gives the initial values used in the ML estimation with discrete characters (see the examples in ?ace). This function now returns a matrix giving the indices of the estimated rates when analysing discrete characters. o nodelabels() and tiplabels() have a new argument `pie' to represent proportions, with any number of categories, as piecharts. The use of the option `thermo' has been improved: there is now no limitation on the number of categories. BUG FIXES o mlphylo() did not work with more than two partitions: this is fixed. o root() failed if the proposed outgroup was already an outgroup in the tree: this is fixed. o The `col' argument in nodelabels() and tiplabels() was not correctly passed when `text' was used: this is fixed. o Two bugs were fixed in mlphylo(): parameters were not always correctly output, and the estimation failed in some cases. o plot.phylo() was stuck when given a tree with a single tip: this is fixed and a message error is now returned. o An error was corrected in the help page of gammaStat regarding the calculation of P-values. o Using gls() could crash R when the number of species in the tree and in the variables were different: this is fixed. CHANGES IN APE VERSION 1.8-2 NEW FEATURES o The new function mlphylo() fits a phylogenetic tree by maximum likelihood from DNA sequences. Its companion function DNAmodel() is used to define the substitution model which may include partitioning. There are methods for logLik(), deviance(), and AIC(), and the summary() method has been extended to display in a friendly way the results of this model fitting. Currently, the functionality is limited to estimating the substitution and associated parameters and computing the likelihood. o The new function drop1.compar.gee (used as, e.g., drop1(m)) tests for single effects in GEE-based comparative method. A warning message is printed if there is not enough degrees of freedom. BUG FIXES o An error message was sometimes issued by plot.multi.tree(), though with no consequence. CHANGES IN APE VERSION 1.8-1 NEW FEATURES o There is a new plot method for lists of trees (objects of class "multi.tree"): it calls plot.phylo() internally and is documented on the same help page. BUG FIXES o A bug was fixed in the C code that analyzes bipartitions: this has impact on several functions like prop.part, prop.clades, boot.phylo, or consensus. o root() did not work correctly when the specified outgroup had more than one element: this is fixed. o dist.dna() sometimes returned a warning inappropriately: this has been corrected. o If the distance object given to nj() had no rownames, nj() returned a tree with no tip labels: it now returns tips labelled "1", "2", ..., corresponding to the row numbers. OTHER CHANGES o nj() has been slightly changed so that tips with a zero distance are first aggregated with zero-lengthed branches; the usual NJ procedure is then performed on a distance matrix without 0's. CHANGES IN APE VERSION 1.8 NEW FEATURES o The new function chronopl() estimates dates using the penalized likelihood method by Sanderson (2002; Mol. Biol. Evol., 19:101). o The new function consensus() calculates the consensus tree of a list of trees. o The new function evolve.phylo() simulates the evolution of continuous characters along a phylogeny under a Brownian model. o The new plot method for objects of class "ancestral" displays a tree together with ancestral values, as returned by the above function. o The new function as.phylo.formula() returns a phylogeny from a set of nested taxonomic variables given as a formula. o The new function read.caic() reads trees in CAIC format. o The new function tiplabels() allows to add labels to the tips of a tree using text or plotting symbols in a flexible way. o The new function unroot() unroots a phylogeny. o multi2di() has a new option, `random', which specifies whether to resolve the multichotomies randomly (the default) or not. o prop.part() now returns an object of class "prop.part" for which there are print (to display a partition in a more friendly way) and summary (to extract the numbers) methods. o plot.phylo() has a new option, `show.tip.label', specifying whether to print the labels of the tips. The default is TRUE. o The code of nj() has been replaced by a faster C code: it is now about 10, 25, and 40 times faster for 50, 100, and 200 taxa, respectively. o write.nexus() now writes whether a tree is rooted or not. BUG FIXES o Two bugs have been fixed in root(): unrooted trees are now handled corretly, and node labels are now output normally. o A bug was fixed in phymltest(): the executable couldn't be found in some cases. o Three bugs have been fixed in ace(): computing the likelihood of ancestral states of discrete characters failed, custom models did not work, and the function failed with a null gradient (a warning message is now returned; this latter bug was also present in yule.cov() as well and is now fixed). o pic() hanged out when missing data were present: an error is now returned. o A small bug was fixed in dist.dna() where the gamma correction was not always correctly dispatched. o plot.phylo() plotted correctly the root edge only when the tree was plotted rightwards: this works now for all directions. OTHER CHANGES o dist.taxo() has been renamed as weight.taxo(). o dist.phylo() has been replaced by the method cophenetic.phylo(). o Various error and warning messages have been improved. CHANGES IN APE VERSION 1.7 NEW FEATURES o The new function ace() estimates ancestral character states for continuous characters (with ML, GLS, and contrasts methods), and discrete characters (with ML only) for any number of states. o The new function compar.ou() fits the Ornstein-Uhlenbeck model of directional evolution for continuous characters. The user specifies the node(s) of the tree where the character optimum changes. o The new function is.rooted() tests whether a tree (of class "phylo") is rooted. o The new function rcoal() generates random ultrametric trees with the possibility to specify the function that generates the inter-nodes distances. o The new function mrca() gives for all pairs of tips in a tree (and optionally nodes too) the most recent common ancestor. o nodelabels() has a new option `thermo' to plot proportions (up to three classes) on the nodes of a tree. o rtree() has been improved: it can now generate rooted or unrooted trees, and the mathematical function that generates the branch lengths may be specified by the user. The tip labels may be given directly in the call to rtree. The limit cases (n = 2, 3) are now handled correctly. o dist.topo() has a new argument `method' with two choices: "PH85" for Penny and Henny's method (already available before and now the default), and "BHV01" for the geometric distance by Billera et al. (2001, Adv. Appl. Math. 27:733). o write.tree() has a new option, `digits', which specifies the number of digits to be printed in the Newick tree. By default digits = 10. The numbers are now always printed in decimal form (i.e., 1.0e-1 is now avoided). o dist.dna() can now compute the raw distances between pairs of DNA sequences by specifying model = "raw". o dist.phylo() has a new option `full' to possibly compute the distances among all tips and nodes of the tree. The default is `full = FALSE'. BUG FIXES o Several bugs were fixed in all.equal.phylo(). o dist.dna() did not handle correctly gaps ("-") in alignments: they are now considered as missing data. o rotate() did not work if the tips were not ordered: this is fixed. o mantel.test() returned NA in some special cases: this is fixed and the function has been improved and is now faster. o A bug was fixed in diversi.gof() where the calculation of A² was incorrect. o cherry() did not work correctly under some OSs (mainly Linux): this is fixed. o is.binary.tree() has been modified so that it works with both rooted and unrooted trees. o The documentation of theta.s() was not correct: this has been fixed. o plot.mst() did not work correctly: this is fixed. CHANGES IN APE VERSION 1.6 NEW FEATURES o The new function dist.topo() computes the topological distances between two trees. o The new function boot.phylo() performs a bootstrap analysis on phylogeny estimation. o The new functions prop.part() and prop.clades() analyse bipartitions from a series of trees. OTHER CHANGES o read.GenBank() now uses the EFetch utility of NCBI instead of the usual Web interface: it is now much faster (e.g., 12 times faster to retrieve 8 sequences, 37 times for 60 sequences). BUG FIXES o Several bugs were fixed in read.dna(). o Several bugs were fixed in diversi.time(). o is.binary.tree() did not work correctly if the tree has no edge lengths: this is fixed. o drop.tip() did not correctly propagated the `node.label' of a tree: this is fixed. CHANGES IN APE VERSION 1.5 NEW FEATURES o Two new functions, as.matching.phylo() and as.phylo.matching(), convert objects between the classes "phylo" and "matching". The latter implements the representation of binary trees introduced by Diaconis and Holmes (1998; PNAS 95:14600). The generic function as.matching() has been introduced as well. o Two new functions, multi2di() and di2multi(), allow to resolve and collapse multichotomies with branches of length zero. o The new function nuc.div() computes the nucleotide diversity from a sample a DNA sequences. o dist.dna() has been completely rewritten with a much faster (particularly for large data sets) C code. Eight models are available: JC69, K80, F81, K81, F84, T92, TN93, and GG95 (the option `method' has been renamed `model'). Computation of variance is available for all models. A gamma-correction is possible for JC69, K80, F81, and TN93. There is a new option, pairwise.deletion, to remove sites with missing data on a pairwise basis. The option `GCcontent' has been removed. o read.GenBank() has a new option (species.names) which specifies whether to return the species names of the organisms in addition to the accession numbers of the sequences (this is the default behaviour). o write.nexus() can now write several trees in the same NEXUS file. o drop.tip() has a new option `root.edge' that allows to specify the new root edge if internal branches are trimmed. BUG FIXES o as.phylo.hclust() failed if some labels had parentheses: this is fixed. o Several bugs were fixed in all.equal.phylo(). This function now returns the logical TRUE if the trees are identical but with different representations (a report was printed previously). o read.GenBank() did not correctly handle ambiguous base codes: this is fixed. OTHER CHANGES o birthdeath() now returns an object of class "birthdeath" for which there is a print method. CHANGES IN APE VERSION 1.4 NEW FEATURES o The new function nj() performs phylogeny estimation with the neighbor-joining method of Saitou and Nei (1987; Mol. Biol. Evol., 4:406). o The new function which.edge() identifies the edges of a tree that belong to a group specified as a set of tips. o The new function as.phylo.phylog() converts an object of class "phylog" (from the package ade4) into an object of class "phylo". o The new function axisPhylo() draws axes on the side of a phylogeny plot. o The new function howmanytrees() calculates the number of trees in different cases and giving a number of tips. o write.tree() has a new option `multi.line' (TRUE by default) to write a Newick tree on several lines rather than on a single line. o The functionalities of zoom() have been extended. Several subtrees can be visualized at the same time, and they are marked on the main tree with colors. The context of the subtrees can be marked with the option `subtree' (see below). o drop.tip() has a new option `subtree' (FALSE by default) which specifies whether to output in the tree how many tips have been deleted and where. o The arguments of add.scale.bar() have been redefined and have now default values (see ?add.scale.bar for details). This function now works even if the plotted tree has no edge length. o plot.phylo() can now plot radial trees, but this does not take edge lengths into account. o In plot.phylo() with `type = "phylogram"', if the values of `edge.color' and `edge.width' are identical for sister-branches, they are propagated to the vertical line that link them. BUG FIXES o Repeated calls to as.phylo.hclust() or as.hclust.phylo() made R crashing. This is fixed. o In plot.phylo(), the options `edge.color' and `edge.width' are now properly recycled; their default values are now "black" and 1, respectively. o A bug has been fixed in write.nexus(). OTHER CHANGES o The function node.depth.edgelength() has been removed and replaced by a C code. CHANGES IN APE VERSION 1.3-1 NEW FEATURES o The new function nodelabels() allows to add labels to the nodes of a tree using text or plotting symbols in a flexible way. o In plot.phylo() the arguments `x.lim' and `y.lim' can now be two numeric values specifying the lower and upper limits on the x- and y-axes. This allows to leave some space on any side of the tree. If a single value is given, this is taken as the upper limit (as before). CHANGES IN APE VERSION 1.3 NEW FEATURES o The new function phymltest() calls the software PHYML and fits 28 models of DNA sequence evolution. There are a print method to display likelihood and AIC values, a summary method to compute the hierarchical likelihood ratio tests, and a plot method to display graphically the AIC values of each model. o The new function yule.cov() fits the Yule model with covariates, a model where the speciation rate is affected by several species traits through a generalized linear model. The parameters are estimated by maximum likelihood. o Three new functions, corBrownian(), corGrafen(), and corMartins(), compute the expected correlation structures among species given a phylogeny under different models of evolution. These can be used for GLS comparative phylogenetic methods (see the examples). There are coef() and corMatrix() methods and an Initialize.corPhyl() function associated. o The new function compar.cheverud() implements Cheverud et al.'s (1985; Evolution 39:1335) phylogenetic comparative method. o The new function varcomp() estimates variance components; it has a plot method. o Two new functions, panel.superpose.correlogram() and plot.correlogramList(), allow to plot several phylogenetic correlograms. o The new function node.leafnumber() computes the number of leaves of a subtree defined by a particular node. o The new function node.sons() gets all tags of son nodes from a given parent node. o The new function compute.brlen() computes the branch lengths of a tree according to a specified method. o plot.phylo() has three new options: "cex" controls the size of the (tip and node) labels (thus it is no more needed to change the global graphical parameter), "direction" which allows to plot the tree rightwards, leftwards, upwards, or downwards, and "y.lim" which sets the upper limit on the y-axis. BUG FIXES o Some functions which try to match tip labels and names of additional data (e.g. vector) are likely to fail if there are typing or syntax errors. If both series of names do not perfectly match, they are ignored and a warning message is now issued. These functions are bd.ext, compar.gee, pic. Their help pages have been clarified on this point. CHANGES IN APE VERSION 1.2-7 NEW FEATURES o The new function root() reroots a phylogenetic tree with respect to a specified outgroup. o The new function rotate() rotates an internal branch of a tree. o In plot.phylo(), the new argument "lab4ut" (labels for unrooted trees) controls the display of the tip labels in unrooted trees. This display has been greatly improved: the tip labels are now not expected to overlap with the tree (particularly if lab4ut = "axial"). In all cases, combining appropriate values of "lab4ut" and the font size (via "par(cex = )") should result in readable unrooted trees. See ?plot.phylo for some examples. o In drop.tip(), the argument `tip' can now be numeric or character. BUG FIXES o drop.tip() did not work correctly with trees with no branch lengths: this is fixed. o A bug in plot.phylo(..., type = "unrooted") made some trees being plotted with some line crossings: this is now fixed. CHANGES IN APE VERSION 1.2-6 NEW FEATURES o Six new functions (Moran.I, correlogram.formula, discrete.dist, correlogram.phylo, dist.taxo, plot.correlogram) have been added to implement comparative methods with an autocorrelation approach. o A new data set describing some life history traits of Carnivores has been included. BUG FIXES o A fix was made on mcmc.popsize() to conform to R 2.0.0. OTHER CHANGES o When plotting a tree with plot.phylo(), the new default of the option `label.offset' is now 0, so the labels are always visible. CHANGES IN APE VERSION 1.2-5 NEW FEATURES o The new function bd.ext() fits a birth-death model with combined phylogenetic and taxonomic data, and estimates the corresponding speciation and extinction rates. OTHER CHANGES o The package gee is no more required by ape but only suggested since only the function compar.gee() calls gee. CHANGES IN APE VERSION 1.2-4 NEW FEATURES o Four new functions (mcmc.popsize, extract.popsize, plot.popsize, and lines.popsize) implementing a new approach for inferring the demographic history from genealogies using a reversible jump MCMC have been introduced. o The unit of time in the skyline plot and in the new plots can now be chosen to be actual years, rather than substitutions. CHANGES IN APE VERSION 1.2-3 NEW FEATURES o The new function rtree() generates a random binary tree with or without branch lengths. o Two new functions for drawing lineages-through-time (LTT) plots are provided: ltt.lines() adds a LTT curve to an existing plot, and mltt.plot() does a multiple LTT plot giving several trees as arguments (see `?ltt.plot' for details). BUG FIXES o Some taxon names made R crashing when calling as.phylo.hclust(): this is fixed. o dist.dna() returned an error with two identical DNA sequences (only using the Jukes-Cantor method returned 0): this is fixed. OTHER CHANGES o The function dist.phylo() has been re-written using a different algorithm: it is now about four times faster. o The code of branching.times() has been improved: it is now about twice faster. CHANGES IN APE VERSION 1.2-2 NEW FEATURES o The new function seg.sites() finds the segregating sites in a sample of DNA sequences. BUG FIXES o A bug introduced in read.tree() and in read.nexus() with version 1.2-1 was fixed. o A few errors were corrected and a few examples were added in the help pages. CHANGES IN APE VERSION 1.2-1 NEW FEATURES o plot.phylo() can now draw the edge of the root of a tree if it has one (see the new option `root.edge', its default is FALSE). BUG FIXES o A bug was fixed in read.nexus(): files with semicolons inside comment blocks were not read correctly. o The behaviour of read.tree() and read.nexus() was corrected so that tree files with badly represented root edges (e.g., with an extra pair of parentheses, see the help pages for details) are now correctly represented in the object of class "phylo"; a warning message is now issued. CHANGES IN APE VERSION 1.2 NEW FEATURES o plot.phylo() has been completely re-written and offers several new functionalities. Three types of trees can now be drawn: phylogram (as previously), cladogram, and unrooted tree; in all three types the branch lengths can be drawn using the edge lengths of the phylogeny or not (e.g., if the latter is absent). The vertical position of the nodes can be adjusted with two choices (see option `node.pos'). The code has been re-structured, and two new functions (potentially useful for developpers) are documented separately: node.depth.edgelength() and node.depth(); see the respective help pages for details. o The new function zoom() allows to explore very large trees by focusing on a small portion of it. o The new function yule() fits by maximum likelihood the Yule model (birth-only process) to a phylogenetic tree. o Support for writing DNA sequences in FASTA format has been introduced in write.dna() (support for reading sequences in this format was introduced in read.dna() in version 1.1-2). The function has been completely re-written, fixing some bugs (see below); the default behaviour is no more to display the sequences on the standard output. Several options have been introduced to control the sequence printing in a flexible way. The help page has been extended. o A new data set is included: a supertree of bats in NEXUS format. BUG FIXES o In theta.s(), the default of the option `variance' has been changed to `FALSE' (as was indicated in the help page). o Several bugs were fixed in the code of all.equal.phylo(). o Several bugs were fixed in write.dna(), particularly this function did not work with `format = "interleaved"'. o Various errors were corrected in the help pages. OTHER CHANGES o The argument names of as.hclust.phylo() have been changed from "(phy)" to "(x, ...)" to conform to the definition of the corresponding generic function. o gamma.stat() has been renamed gammaStat() to avoid confusion since gamma() is a generic function. CHANGES IN APE VERSION 1.1-3 BUG FIXES o base.freq() previously did not return a value of 0 for bases absent in the data (e.g., a vector of length 3 was returned if one base was absent). This is now fixed (a vector of length 4 is always returned). o Several bugs were fixed in read.nexus(), including that this function did not work in this absence of a "TRANSLATE" command in the NEXUS file, and that the commands were case-sensitive. CHANGES IN APE VERSION 1.1-2 NEW FEATURES o The Tamura and Nei (1993) model of DNA distance is now implemented in dist.dna(): five models are now available in this function. o A new data set is included: a set of 15 sequences of the cytochrome b mitochondrial gene of the woodmouse (Apodemus sylvaticus). BUG FIXES o A bug in read.nexus() was fixed. o read.dna() previously did not work correctly in most cases. The function has been completely re-written and its help page has been considerably extended (see ?read.dna for details). Underscores (_) in taxon names are no more replaced with spaces (this behaviour was undocumented). o A bug was fixed in write.dna(). CHANGES IN APE VERSION 1.1-1 BUG FIXES o A bug in read.tree() introduced in APE 1.1 was fixed. o A bug in compar.gee() resulted in an error when trying to fit a model with `family = "binomial"'. This is now fixed. CHANGES IN APE VERSION 1.1 NEW FEATURES o The Klastorin (1982) method as suggested by Misawa and Tajima (2000, Mol. Biol. Evol. 17:1879-1884) for classifying genes on the basis of phylogenetic trees has been implemented (see the function klastorin()). o Functions have been added to convert APE's "phylo" objects in "hclust" cluster objects and vice versa (see the help page of as.phylo for details). o Three new functions, ratogram(), chronogram() and NPRS.criterion(), are introduced for the estimation of absolute evolutionary rates (ratogram) and dated clock-like trees (chronogram) from phylogenetic trees using the non-parametric rate smoothing approach by MJ Sanderson (1997, Mol. Biol. Evol. 14:1218-1231). o A summary method is now provided printing a summary information on a phylogenetic tree with, for instance, `summary(tree)'. o The behaviour of read.tree() was changed so that all spaces and tabulations in tree files are now ignored. Consequently, spaces in tip labels are no more allowed. Another side effect is that read.nexus() now does not replace the underscores (_) in tip labels with spaces (this behaviour was undocumented). o The function plot.phylo() has a new option (`underscore') which specifies whether the underscores in tip labels should be written on the plot as such or replaced with spaces (the default). o The function birthdeath() now computes 95% confidence intervals of the estimated parameters using profile likelihood. o Three new data sets are included: a gene tree estimated from 36 landplant rbcL sequences, a gene tree estimated from 32 opsin sequences, and a gene tree for 50 BRCA1 mammalian sequences. BUG FIXES o A bug was fixed in dist.gene() where nothing was returned. o A bug in plot.mst() was fixed. o A bug in vcv.phylo() resulted in false correlations when the option `cor = TRUE' was used (now fixed). CHANGES IN APE VERSION 1.0 NEW FEATURES o Two new functions, read.dna() and write.dna(), read/write in a file DNA sequences in interleaved or in sequential format. o Two new functions, read.nexus() and write.nexus(), read/write trees in a NEXUS file. o The new function bind.tree() allows to bind two trees together, possibly handling root edges to give internal branches. o The new function drop.tip() removes the tips in a phylogenetic tree, and trims (or not) the corresponding internal branches. o The new function is.ultrametric() tests if a tree is ultrametric. o The function plot.phylo() has more functionalities such as drawing the branches with different colours and/or different widths, showing the node labels, controling the position and font of the labels, rotating the labels, and controling the space around the plot. o The function read.tree() can now read trees with no branch length, such as "(a,b),c);". Consequently, the element `edge.length' in objects of class "phylo" is now optional. o The function write.tree() has a new default behaviour: if the default for the option `file' is used (i.e. file = ""), then a variable of mode character containing the tree in Newick format is returned which can thus be assigned (e.g., tree <- write.tree(phy)). o The function read.tree() has a new argument `text' which allows to read the tree in a variable of mode character. o A new data set is included: the phylogenetic relationships among the orders of birds from Sibley and Ahlquist (1990). CHANGES IN APE VERSION 0.2-1 BUG FIXES o Several bugs were fixed in the help pages. CHANGES IN APE VERSION 0.2 NEW FEATURES o The function write.tree() writes phylogenetic trees (objects of class "phylo") in an ASCII file using the Newick parenthetic format. o The function birthdeath() fits a birth-death model to branching times by maximum likelihood, and estimates the corresponding speciation and extinction rates. o The function scale.bar() adds a scale bar to a plot of a phylogenetic tree. o The function is.binary.tree() tests whether a phylogeny is binary. o Two generic functions, coalescent.intervals() and collapsed.intervals(), as well as some methods are introduced. o Several functions, including some generics and methods, for computing skyline plot estimates (classic and generalized) of effective population size through time are introduced and replace the function skyline.plot() in version 0.1. o Two data sets are now included: the phylogenetic relationships among the families of birds from Sibley and Ahlquist (1990), and an estimated clock-like phylogeny of HIV sequences sampled in the Democratic Republic of Congo. DEPRECATED & DEFUNCT o The function skyline.plot() in ape 0.1 has been deprecated and replaced by more elaborate functions (see above). BUG FIXES o Two important bugs were fixed in plot.phylo(): phylogenies with multichotomies not at the root or not with only terminal branches, and phylogenies with a single node (i.e. only terminal branches) did not plot. These trees should be plotted correctly now. o Several bugs were fixed in diversi.time() in the computation of AICs and LRTs. o Various errors were corrected in the help pages.