Comparing phylogenetic and statistical classification methods for DNA barcoding
Résumé
Intraspecific variation in mtDNA results from a genealogical process. We can therefore address similar questions as were raised when cladistics methods were introduced in species systematics: should our classification methods account for the characteristics of the evolutionary process in producing patterns of taxonomic diversity? Should we use either phylogenetic methods, or statistical methods that are not based on evolutionary biology models, or both? The difference is that the old cladistic-classification debate was concerned with distantly related taxons, whereas the main issue for the barcode is to correctly interpret individual differences around the speciation threshold. The aim of our ongoing research is to compare the efficiency of genealogical methods (neighbour-joining and maximum likelihood methods) and statistical methods (supervised classification). To this end, we use both empirical and simulated data following given scenarios. We partition the data into two subsets: (1) some individuals make up a reference sample of known specific status, having each a sequence and a species name; (2) some individuals are known only from their DNA sequence, and the method is used to assign them to the right species. We compare the efficiency of the two series of methods in interpreting the dataset. This can be assessed exactly for the simulated sets. We will present the design and the preliminary results of this investigation on the empirical and simulated data sets .
Origine : Fichiers éditeurs autorisés sur une archive ouverte
Loading...