4.2 Limits of molecular taxonomy and eDNA perspectives
Our results are still potentially biased by the uncertainties of molecular taxonomy. It is not clear whereas our species delimitations, mainly based on mitochondrial DNA (mtDNA) markers, actually reflect species boundaries.
First, do the mtDNA markers used to characterize species reflect species boundaries? This question has already been addressed for French Naiad species by Prié and Puillandre (2014), but the status of the species within the “pictorum” group, i.e. for France U. mancusLamarck, 1819 and U. pictorum (Linnaeus, 1758), and within the“crassus” group, i.e. for France U. crassus crassusPhilipsson, 1788 and U. c. courtillieri (Hattemann, 1859) remains unsolved. From our database U. mancus and U. pictorum , as defined by mtDNA molecular markers, seem to largely overlap geographically - and are not discriminable neither from shell shape nor anatomy where both lineages are sympatric. Their mtDNA divergence may only result from ancestral polymorphism retention, and these two species may be synonyms. The least that can be said is that eDNA large scale sampling results in France, together with field experience, do not argue in favor of two distinct species.
Regarding the Thick-Shelled River Mussel U. crassus (a species having Doubly Uniparental Inheritance, Significance Box 1), mtDNA female marker would argue in favor of distinct species, namely U. crassus and U. courtillieri . Mioduchowska, Kaczmarczyk, Zajac and Zajac (2016) showed that an independent marker, identified in male mtDNA, also reveals two distinct lineages with a significant divergence. But for now, there is no evidence that these two independent markers tell the same story in any sampled specimen: these two lineages may largely interbreed and mt DNA may just bear the scars of a complex phylogeography with multiple species contractions and expansions throughout the glacial ages. We decided not to distinguish these two taxa in our eDNA protocol.
Hybridization and the hazards of molecular taxonomy have been largely documented for the Asiatic Clams Corbicula spp. (Pfenninger, Reinhardt, & Streit, 2002; Pigneur et al., 2011). Here, we did not distinguish the three species described for France (C. fluminea(O.F. Müller, 1774), C. fluminalis (O.F. Müller, 1774) andC. leana Prime, 1867, sensus Pigneur et al. 2011), although one haplotype of C. fluminalis seems to be proper to this species. eDNA large-scale analysis makes it possible to map the different haplotypes of this species complex, and maybe provide insights for a better understanding of invasive species colonization routes.
Second, the haplotype diversity of each species is not known. We have only a few sequences for most of the species, sometimes only one specimen studied for the 16S gene (eg. the Iridescent Pea MusselE . pulchella (Jenyns, 1832) or the Oval Orb MusselS. ovale (A. Férussac, 1807), see Supporting information). And yet, some of the short fragments we amplify with our primers are haplotypes shared between different species (Prié et al., 2020). Therefore, it is not sure if the various haplotypes detected reflect species distribution or are shared between unknown species. For example,E. pulchella is known from only one 16S sequence, which is closely related to the haplotypes of the Short-ended Pea Mussel E. subtruncata (Malm, 1855) (seven known haplotypes of our short 16S fragment). If the haplotype referenced as E. pulchella in our reference library was also a rare haplotype of E. subtruncata , the distribution given here for E. pulchella could be heavily biased. What we present here is haplotypes distribution.
But eDNA remains an interesting tool to investigate genetic diversity in the field. In that respect, the samples performed in Morocco and Italy allowed characterizing male mtDNA of The Moroccan River MusselUnio foucauldianus Pallary, 1936 and the Italian River MusselU. elongatulus C. Pfeiffer, 1825, and in Morocco, many previously unknown haplotypes of the Caserta Pea Mussel E. casertana (Poli, 1791) were amplified.