New vegan version was submitted to CRAN today, much earlier than scheduled. The only real change is a fix in adonis permutation tests. This is a critical change, and all adonis users should upgrade vegan and re-analyse adonis models.

The problem only concerns models with several explanatory variables: single-variable models are OK. THe problem was that permutation tests used different statistics than was reported. Therefore the tests were inconsistent, and much too high P-values were reported to variables after the first one.

In addition, this corrects two minor problems: the degrees of freedom did not take into account aliased variables in rank-deficit models, and unused factor levels were not dropped.

The fixes are also included in the devel version 1.12-11 and later in R-Forge.