Evolution of Pycnogonida

Abstract

Pycnogonids (sea spiders) form a phylum of Arthropods unrelated to true spiders. They have a global distribution, with particular abundance in Antarctic waters. Due to their obscure nature, pycnogonids have historically been little studied, and this paucity of data has been reflected in the wide range of taxonomies put forward. Recently, however, several studies, including the 2004 BioRoss expedition to Antarctica, have added a considerable volume of new genetic data, and these data are analysed in this study. The gene sequences used are from several genes, both nuclear and mitochondrial, and both supermatrices and supertrees are used to combine the sequences. Two alignment algorithms, implemented by the programs Clustalw2 and PRANK, are used to generate the nucleotide alignments. Clustalw2 works better than PRANK for large alignments, as aligning the sequences using PRANK appears to cause a loss of phylogenetic information, resulting in trees with lower posterior clade probabilities, and an inability for these trees to reject constrained topology tests. This does not appear to be the case for data sets with fewer taxa, for which both alignment methods work equally well. The phylogenies for the individual genes, as well as the combined supermatrix data sets, were created using the program MrBayes, which performs a Bayesian analysis. Constrained topology tests were performed using MrBayes to test the monophyly of traditional ycnogonid families, as well as some clades consisting of two or more families. The constrained topology tests were implemented using Bayes’ factors. All families tested were monophyletic, after the family Ammotheidae has been divided into Ascorhynchidae and Ammotheidae; and Callipallenidae into Pallenopsidae and Callipallenidae. The families Callipallenidae and Nymphonidae are shown to be closely related, as are Colossendeidae and Pycnogonidae. Endeidae may also be associated with this latter clade. The position of Austrodecidae is uncertain: it often appears at the base of phylogenies; however, the constrained topology tests allow it to fall in a clade with Colossendeidae and Pycnogonidae. The supertree analysis was performed using the program Clann, with three optimisation criteria: most similar supertree, maximum quartet fit and maximum split fit. Of these three methods, the maximum split fit appears to be the most powerful. Data sets with more input trees give better results when creating supertrees than smaller data sets. The results for the supertrees using the maximum split fit method are consistent with those from the supermatrices.