Species identification of Collembola by means of PCR-based marker systems

Abstract

The MOLART project on which this PhD thesis bases deals with the development and standardisation of molecular methods for safe and efficient identification of species which can not - or only with great effort and expertise - be distinguished by morphological analysis. These aims were exemplified by two problematic species groups within the Collembola as examples of taxa that combine ecological importance and controversial species status, the Protaphorura armata group (Onychiuridae, Poduromorpha) and the Isotoma viridis group (Isotomidae, Entomobryomorpha). For this purpose, DNA sequencing, PCR-RFLP, RAPD, AP-PCR and AFLP were applied and evaluated according to their diagnostic power, their efficiency for different taxa, the effort of time and the experience required. It was aimed to extract information from individual specimens to avoid mixing of genetic information. We could show that DNA can be extracted successfully from single specimens. Based on identification by morphological keys, molecular markers were tested for their accordance with morphological taxonomy and molecular identification tools were designed based on the results. For the I. viridis group, the mitochondrial COII sequence was used to infer a molecular phylogeny that is in accordance with morphological findings. Molecular traits were found suitable to provide a fast and more robust discrimination compared to morphological characters. A fourth member of this group recently proposed by morphological hints, I. caerulea, could be confirmed to represent a distinct species. Genetic differences were used to establish a set of restriction enzymes that will provide a fast and efficient identification system for I. viridis group from different European countries. For the P. armata group, sequences were obtained from three gene loci, mitochondrial COI, COII and nuclear ITS1, in order to resolve the systematics of this group. Concordant patterns for nuclear and mitochondrial markers suggest that the P. armata group consists of more than one species. Especially P. fimata and P. armata seem to represent distinct species. Differences were found between phylogenies derived from the three gene loci, most of them due to the small number of individuals tested in some species. In the phylogenetic trees, morphologically intermediate forms were not found to represent distinct species. The ITS1 sequence confirmed the findings of both mitochondrial gene regions but did not provide a more detailed taxonomic resolution. Obviously a detailed resolution of a species group consisting of about 40 species will require a larger molecular marker set, about 10 to 20 specimens per 'species' from different sampling sites and several independent marker sets to check for possible inconsistencies. Both RAPD and AP-PCR methods did not reveal reproducible results and yielded fragment patterns rather on the level of populations than of species. Whereas TE-AFLP failed to produce visible bands, AFLP yielded reproducible fragment patterns in both species groups as the only method that includes two amplification steps, indicating a low template concentration to be the reason for failure of amplification or reproducibility. Contamination by alien DNA, especially from gut content, was found to be a severe problem when investigating small-sized taxa by molecular methods, and a long-time storage in ethanol further decreases the extraction yield due to DNA degradation. Based on these investigations, standardised methods are proposed to be used in future molecular taxonomy, especially in cooperation with traditional morphological taxonomy, for optimising DNA yield and purity for better characterisation of small-sized specimens and for establishing reference collections and future identification keys. As implications for future molecular taxonomy, it is recommended to use live specimens that can be cleaned from alien DNA prior to extraction and to use species-specific primers that bear a lower risk than universal primers to amplify unwanted contaminations. When sequencing single gene loci, mitochondrial protein-coding genes seem to be most appropriate since they are present in high copy numbers and fixed differences suitable for discriminating closely related taxa are more likely to be found in faster evolving mitochondrial genes than in nuclear sequences. Methods known to be influenced easily by reaction conditions, like RAPD and AP-PCR, should be tested for their reproducibility prior to use. Among fingerprinting methods, AFLP seems to be the best choice, since it produces a reproducible and complex information pattern that, if desired, can be reduced to a level suitable for faster taxon identification.