Аннотация:It is long known that indices of non-parametric bootstrapping significantly underestimate the probability of existence of a topological element, particularly, in cases when lineages analysed differ in rates of character substitution. Results of conventional bootstrapping often indicate low support for deeper nodes of the tree or suggest unresolved phylogenies. True topological elements that may be present in some of equal trees are lost upon building a consensus. Among discrete methods of phylogenetic inference, parsimony is the fastest and most widely used but also most susceptible to this problem. Particularly, if the amount of homoplasy at the level of individual characters is high enough, it may graft taxa erroneously on the basis of homoplastic characters while searching for most parsimonious trees during additional sequence replicates. Ultimately, heuristic search may stall within a local tree island, which is isolated from the global optimum by branch swappings that can not be implemented given a current combination of taxa and thus will fail to find the true tree. Here we present an easy method of estimating reliability of individual nodes of the tree and screening the equal tree space for reliable topological elements. We analysed a complete SSU rDNA alignment of 36 nematode and 6 other metazoan groups, which contained clear disparities in rates of nucleotide substitution across taxa. To test the performance of parsimony under different signal/noise ratio in the data, variable positions were gradually removed from the alignment according to the amount of change assigned to each position on the most parsimonious tree, which resulted in a series of 24 subalignments. Screening of equal tree spaces produced by subsequent processing of the subalignments with parsimony (PAUP* 4.0beta10 package) revealed a number of nodes matching those present in the tree found on the basis of the initial alignment. However, composition of other nodes varied depending on the amount of positions removed and in most cases matched neither those retained in the consensus of the corresponding parsimony run, nor those present in the initial tree. To elaborate a measure of reliability of a node, we computed values of homoplasy index (HI) for each character considered by the algorithm synapomorphic for a node. Analysis confirmed that HI values for characters supporting recurrent nodes are close to or equal zero. Further, it revealed other nodes not present in the initial tree and yet reconstructed on the basis of low homoplastic characters. Some of them correspond to higher nematode taxa, which monophyly has already been substantiated with independent evidence (morphology or SSU rRNA secondary structure), and some were novel. However, this approach does not allow one to resolve alternative nodes supported by characters with almost equally high HI values. To study this case, we analysed the total amount of homoplasy generated by a topology through computing pairwise homoplasy distances for trees inferred in a series of subsequent reconstructions. Homoplasy matrices thus obtained and visualised as "trees" give an idea about how much homoplasy is generated between taxa by a particular topology. The analysis showed that lineages joined in recurrent nodes contain less homoplasy with respect to each other and are thus situated close on tree representations of the matrices. This was not the case for nodes reconstructed on the basis of highly homoplastic characters. The actual content of such nodes is in discord with patterns of pairwise homoplasy distance distribution, i.e. taxa joined by the algorithm generate high level of homoplasy in the node and so do not cluster together in homoplasy “trees”. Alternatively, some of local combinations of taxa did correspond to the pattern of homoplasy distance distribution and, thus, are likely to represent natural monophyletic clades. The minimum homoplasy requirement was applied to screening the analytic tree space pooled over the series of reconstructions. Individual nodes thus selected were compiled in a phylogeny found in neither of heuristic searches. We tested this phylogenetic hypothesis by constraining combinations of taxa, which either were supported by low homoplastic characters or were in accord with corresponding homoplasy distance distribution patterns, and conducted a constrained parsimony heuristic search on the basis of the initial alignment. None of the equal trees found fully coincided with the compiled phylogeny, although it was assigned the highest likelihood score by Shimodaira-Hasegawa one-tailed test using RELL bootstrap.