+ CHANGES IN APE VERSION 2.8-1
+
+
+BUG FIXES
+
+ o mantel.test() could return 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).
+
+
+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.
+
+
+
+ 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 = <tip>,
+ 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.
+
+
+