Various alternative topologies are compared to the ML topology using an AU test. For each alternative topology (rows of the table), the lnL of the alternative topology is given along with the percentage of occurrences of the alternative topology in the unconstrained bootstrap analysis (BV), and the P-value of the AU test on a larger set of trees. On the first data line, the Bayesian tree is compared to the ML tree. In this case, the null hypothesis of the AU test is that the ML tree is not significantly more likely than the BI tree. In the middle part of the table, each of the non-monophyletic orders is listed along with the lnL of the topology in which the order is constrained to be monophyletic. In this case, the null hypothesis of the AU test is that unconstrained and constrained topologies are equally likely. In the bottom part of the table, the possibility that the poorly resolved regions represent hard polytomies is tested. The listed lnL are for the trees in which one of the poorly resolved region was collapsed, and in this case the null hypothesis of the AU test is that uncollapsed and collapsed topologies are equally likely. The lnL of the unconstrained, uncollapsed topology is -185,569.97.