Likelihood-based statistical tests of   competing evolutionary hypotheses (tree topologies) have been available for   approximately a decade. By far the most commonly used is the Kishino-Hasegawa   test. However, the assumptions that have to be made to ensure the validity of   the Kishino-Hasegawa test place important restrictions on its applicability.   In particular, it is only valid when the topologies being compared are   specified a priori. Unfortunately, this means that the Kishino-Hasegawa test   may be severely biased in many cases in which it is now commonly used: for   example, in any case in which one of the competing topologies has been   selected for testing because it is the maximum likelihood topology for the   data set at hand. We review the theory of the Kishino-Hasegawa test and   contend that for the majority of popular applications this test should not be   used. Previously published results from invalid applications of the   Kishino-Hasegawa test should be treated extremely cautiously, and future   applications should use appropriate alternative tests instead. We review such   alternative tests, both nonparametric and parametric, and give two examples   which illustrate the importance of our contentions.

基于最大似然法检验不同的进化假说（树的拓扑结构）已问世十多年。到目前为止，最常用的是Kishino-Hasegawa检验。然而，为确保检验有效性而必须作出的假设使其适用性受到诸多重大限制。特别是，只有当备择拓扑结构被预先指定时，它才有效。不幸的是，这意味着Kishino-Hasegawa检验在许多现在普遍使用的情况下可能会有严重的偏差：例如，在任何情况下，都会选择一个候选拓扑结构作为标的进行检验，因为它是现有数据集的最大似然树。我们回顾了Kishino-Hasegawa检验的理论，并认为对于大多数通常的应用，不应使用此检验。先前公布的Kishino-Hasegawa检验无效应用结果应谨慎对待，未来的应用应使用适当的替代检验。我们回顾了非参数检验和参数检验，并给出了两个例子来说明本争论的重要性。