Efficiency of the DNA-barcode expertise and its relationships with the practice of alpha-taxonomy illustrated by the study of the genus Eumunida (Decapoda). Sarah Samadi (IRD) UMR 7138, Dept Syst & Evol, MNHN, Paris. Nicolas Puillandre, Enrique Macpherson, Josie Lambourdière, Marie-Catherine Boisselier

The context > The constitution of DNA barcode dataset on a given set of organisms must follow the requirements of specimens sampling of the taxonomic research. Given the estimated amount of biological species that needs to be described  Offering an efficient expertise tool is an important challenge for the taxonomy of the 21st century. > The taxonomic expertise directly lies on the taxonomic knowledge which continually evolves with the new data provided by the taxonomic research. > New expertise tools must be directly linked to taxonomic research. Our positions

> What is a species ? > How in practice do we delimit species ? > How names are linked to these hypotheses ? TAXONOMIC EXPERTISE = TAXONOMIC RESEARCH =  To propose hypotheses of species delimitations  to link a specimen to a species name

A species is a permanently diverging lineage Samadi & Barberousse, 2006 De Quieroz, 1998 A species is a piece of genealogical network that is definitively divergent from others part of the network Delimiting species =  recovering the structure of the genealogical network

(i) Similarity in heritable characters (ii) Interfecondity, gene flows (iii) Share a common history using heritable characters (i) alpha-Taxonomy (morphology or molecules) (phenetic criteria) (ii) Populations genetics and biology (biological criteria) (iii) Phylogenetic reconstruction (phylogenetic criteria) Samadi & Barberousse, 2006  a species is a reproductive community  Divergence with others such species is definitive How in practice do we resolve the structure of the network ? Delimiting species =  resolving the structure of the genealogical network

DNA offers many polymorphic and heritable characters on which the different criteria can be applied … > In addition : The analysis of type-specimens permits to link the DNA-barcode to the names > Phylogenetically close species must be included (the evolutionary context) > Sampling must cover the known geographic distribution > to measure similarity or to estimate gene flows sampling must include several specimens But with some sampling requirements

COI is DNA fragment that offers many polymorphic and heritable characters > unlinked molecular characters (nuclear versus mitochondrial) > Morphological characters > Life histroy traits > Ecological data (biotic and abiotic environnement …) But … other data should be integrated to test the primary hypotheses

Many type specimens available in the collections of the MNHN (Paris) Well studied by « classical » taxonomy Collections preserved in ethanol The genus Eumunida Many specimens available for each species (MUSORSTOM cruises) Among the 26 described species we were able to extract and sequence DNA from specimens of the 17 species available in the collections of the MNHN, that cover the known distribution area Type specimens were available for 13 species 9 holotypes & 24 paratypes

> 229 specimens > 17 species / 26 described > 9 holotypes et 24 paratypes (13 species) Distribution of pairwise genetic distances for COI In black, the distribution of the distances between pairs of type specimens > COI (« barcode fragment», 658pb) > 28S (C1’-D2, 867pb) The genus Eumunida

> 229 specimens > 17 species / 26 described > 9 holotypes et 24 paratypes (13 species) > COI (« barcode fragment», 658pb) > 28S (C1’-D2, 867pb) The genus Eumunida For some old specimens we did not obtain the 28S sequence COI was obtained for all except one specimens 28S fragment is less variable than COI but the genetic clusters are the same COI data are more easily obtained

The type specimens are used to give names to the genetic clusters When no type specimens is included in a genetic cluster the morphological determination key is used E. sternomaculata E. similor E. spinosa E. annulosa (holotype) E. marginata E. minor E. bispinata E. keiji E. treguieri E. squamifera E. multineata E. laevimana E. capilata E. picta E. funambulus > 229 specimens > 17 species / 26 described > 9 holotypes et 24 paratypes (13 species) > COI (« barcode fragment», 658pb) > 28S (C1’-D2, 867pb) The genus Eumunida

Overall, the species recognized using the traditional approach are recovered … but E. annulosa E. multineata E. laevimana E. capilata E. picta E. sternomaculata E. similor E. spinosa E. annulosa (holotype) E. marginata E. minor E. bispinata E. keiji E. treguieri E. squamifera E. funambulus > 229 specimens > 17 species / 26 described > 9 holotypes et 24 paratypes (13 species) > COI (« barcode fragment», 658pb) > 28S (C1’-D2, 867pb) E. parva + E. karubar + E. smithii One cluster correspond to three names (including the types of two of them) E. annulosa (holotype) The genus Eumunida The morphological diagnostic for one cluster correspond to a name already attributed, using the position of the holotype, to another cluster

E. annulosa E. annulosa (holotype) Spines on the carpus of the chelipeds One cryptic species is detected This character is diagnostic only for adult specimens  Such characters are often used in determination keys In such a case the DNA- barcode expertise is more efficient than the traditional morphological approach Moreover it works for all the stages of the life cycle The genus Eumunida

E. parva E. karubar E. smithii S S P P P P P P P Ph K P K P K H P K K Three species correspond to a unique genetic cluster Type specimens and localities are mixed for the three species The hypothesis of synonymy is supported by the 28S data The genus Eumunida

This sampling and the associated molecular data set allow us to : (i) Support most of the primary species hypotheses based on morphology using the phenetic criterion on molecular data … but also to bring up new hypotheses (ii) For some of the species to show that gene flows occurs among populations over the geographic range of the species (Biological criterion) (… more details for two of them in Samadi et al, Mar Biol 2006) (iii) Using a close out-group outside the genus, to reveal that each of the proposed species has it own evolutionary history (Phylogenetical criterion) > 229 specimens > 17 species / 26 described > 9 holotypes et 24 paratypes (13 species) > COI (« barcode fragment», 658pb) > 28S (C1’-D2, 867pb) The genus Eumunida

When the taxonomy is well established The DNA-barcode can be used as a determination key When sampling is adequate such data should enhance taxonomy and bring up new hypotheses But sampling must cover the intra and inter specific variability, the geographic distribution and when possible type specimens But only if we go back to specimens and morphology Thus, DNA-barcode is an efficient expertise tool only if sampling is adequate and directly linked to taxonomical research.

acknowledgments The crew of the R/V Alis The staff of the ‘Service de systématique moléculaire’ at the ‘MNHN, Paris’ Régis Cleva (curator of crustacean collections) The ‘Consortium national de recherche en génomique’, Genoscope