Marine interconnectivity among nations has risen in recent decades, a trend that will further increase and that is favoring species introductions all over the world (Levine and D’Antonio 2003; Seebens et al. 2016; Carrasco et al. 2017; Sardain et al. 2019). As introduced species lack natural regulators in their new environment (Papacostas et al. 2017), their effects may be unpredictable. Some invaders may completely restructure ecosystems, potentially leading to the loss of biodiversity and ecosystem services (Pejchar and Mooney 2009; Johnston et al. 2015; Walsh et al. 2016), causing local extinctions (Blackburn et al. 2019), and high economic impacts (Lovell et al. 2006; Olson 2006; Jardine and Sanchirico 2018; Diagne et al. 2021). As an example, direct damages as well as costs generated from combating introduced species accumulated to over $29 billion in Europe alone (see supplementary material table S1 of Diagne et al. 2021). These damages on ecosystems and the economy do not even require high abundances of an introduced species to be tangible (Blackburn et al. 2011). For these reasons, invasive species (sensuBlackburn et al. 2011) are regarded as a crucial global biosecurity issue, and prevention and early management of these species constitutes the best strategy to minimize their impact (Lovell et al. 2006; Olson 2006; Pyšek et al. 2020). This may prove difficult in phyla with few or small morphological identification criteria, that form cryptic species complexes (Mackie et al. 2006; Mackie et al. 2012; Vieira et al. 2014; Mastrototaro et al. 2020; Salonna et al. 2021), or for which taxonomic expertise is rare, such as for many marine invertebrates. Identifying the precise species at a certain location is however important for subsequent evaluation of spread and invasion (Vieira et al. 2014; Golo et al. 2023).

Especially when morphological criteria are lacking to identify a species, genetic methods have been increasingly used to help with the identification of introduced species. DNA barcoding of individual specimens, metabarcoding of communities in bulk (from sediments, scrapings, or other substrates) and more recently environmental DNA (eDNA) are now all regarded as useful techniques for detecting introduced species. However, the efficiency of molecular species identification highly depends on the quality of reference sequences, particularly how well species identification was carried out for the reference sequences before being submitted to public databases (Couton et al. 2022). Incorrect species identification of reference sequences can have important effects on subsequent studies which will base molecular identifications on imprecise species names.

The morphology of the bryozoan genus Watersipora is notoriously complicated, and the redescription of the genus by Vieira et al. (2014) has reattributed individuals of introduced populations in multiple localities to Watersipora subatra (Ortmann, 1890) (identified as “Watersipora subtorquata” (d’Orbigny, 1852) in Mackie et al. 2012; Ryland et al. 2009) and others to W. subtorquata (identified as “W. subvoidea” (d’Orbigny, 1852) in Mackie et al. 2006, 2012). This has led to confusion regarding the identity of populations of several localities around the world, most notably in Europe. Many species identifications of reference sequences in Genbank have not been updated since the redescription by Vieira et al. (2014). Nevertheless, these sequences continue to be used for e-DNA monitoring. As an example, individuals that were resolved as W. subatra are still listed as “W. subtorquata” on GenBank (accessed June 2022, NCBI, Benson et al. 2013). This probably induces subsequent errors in the literature, falsely detecting W. subatra as “W. subtorquata”, hindering the accurate e-DNA detection of several Watersipora species.

In the Mediterranean Sea, the presence of W. subtorquata has been reliably confirmed (Vieira et al. 2014) and this species is the most frequently described introduced Watersipora species in this area (Vieira et al. 2014; Harmelin et al. 2016; Rosso and Di Martino 2016; Tempesti et al. 2020), even though Watersipora arcuata Banta, 1969 seems to be spreading rapidly in recent years (Ulman et al. 2017, 2019; Reverter-Gil and Souto 2019). Here however, we identified W. subatra as the dominant Watersipora species in several French Mediterranean harbors, with only anecdotal presence of W. subtorquata (and no observations of W. arcuata). This points towards an inherent identification problem, as the high abundance and persistence on many artificial substrates makes it impossible to miss the species (see Fig. 1a) and suggests it has been present some time already.

The present study has a two-fold objective. Firstly, we wish to declare the first formal record of W. subatra as an already well-established introduced species in the French Mediterranean Sea. Secondly, we provide a phylogenetic analysis of existing COI sequences of Watersipora from Genbank, including new sequences from individuals that were carefully identified according to morphological criteria, in order to improve molecular identification and detection of non-indigenous Watersipora species, particularly the spread of W. subatra in the Mediterranean or elsewhere.

At least five species of Watersiporidae have been reported from the Mediterranean Sea, including Watersipora cucullata (Busk, 1854), W. souleorum, Terwasipora complanata and the introduced W. subtorquata and W. arcuata (Vieira et al. 2014; Ulman et al. 2017; Reverter-Gil and Souto 2019). Watersipora subatra, despite being the most dominant introduced Watersiporidae in the north-eastern Atlantic, has only been recorded recently and sporadically in the Mediterranean Sea (Fernández-Romero et al. 2021; Ramalho and Caballero-Herrera 2022; Gauff et al. 2023a, b)). The individuals examined within our study unambiguously correspond to the morphological description of W. subatra in Vieira et al. (2014). This is further validated by the genetic analysis that cluster our Mediterranean individuals within the clade regrouping individuals of “W. subtorquata” from Mackie et al. (2006, 2012), that were resolved as W. subatra (Vieira et al. 2014), individuals identified as W. subatra from the French Atlantic (Couton et al. 2019) and our own W. subatra from the French Atlantic. We thus can confidently report the presence of this species in the French Mediterranean Sea. Out of the 13 accepted Watersipora spp., all four species that are spreading throughout the world (W. subtorquata, W. souleorum, W. arcuata and W. subatra) are thus now present and established in the Mediterranean Sea. Watersipora subatra is an invasive fouling species, mostly dispersed by ship traffic, that has recently spread from the north-east Atlantic towards the south of the Iberian Peninsula, suggesting that its introduction into the Mediterranean Sea likely occurred through the straits of Gibraltar (Reverter-Gil and Souto 2019). France possesses an Atlantic and Mediterranean coast, harboring numerous introduced species (Massé et al. 2023). National commerce and exchange (such as shellfish culture) could favor species transfers between those two provinces (Bachelet et al. 2004; Fernández-Rodríguez et al. 2022). Direct ship traffic between the naval bases of Toulon and Brest (Atlantic), where W. subatra is common (see Gauff et al. 2022; Rondeau et al. 2022) might further increase this risk of introduction. It is, therefore, not unexpected to find this species in a large Mediterranean harbor like Toulon.

More troubling is the high abundance of this species, suggesting that it is well established and has been present for some time already. This species seems to have been present for at least four years in the Mediterranean, as a previous study in Marseilles has tentatively identified the species in 2019 (Gauff et al. 2023b). Watersipora subatra seems to be the most dominant Watersipora species in the French Mediterranean as we did not detect any W. subtorquata despite this species being identified in the Toulon Darse Nord Marina in past studies Gauff et al. 2023a)). This might suggest that the species has been misidentified as W. subtorquata for several years. The comparatively low number of taxonomic experts still in full activity in this area might explain such misidentifications/absence of identifications (Ferrario et al. 2018). Taxonomic expertise requires much time and rigor (Caley et al. 2013; Coleman and Radulovici 2020). Currently, there is an increasing tendency to substitute taxonomy with time-efficient genetic tools to either confirm species identifications by barcoding (Liu et al. 2017; Kenworthy et al. 2018) or to detect species in an area via metabarcoding (Leray and Knowlton 2015; Miralles et al. 2016; Ardura and Planes 2017; Couton et al. 2019, 2022; Azevedo et al. 2020). These tools can be quite powerful to detect NIS, especially when combined with morphological analyses (Azevedo et al. 2020; Couton et al. 2022), however the efficiency of the method depends on the quality of reference sequences (Viard et al. 2019; Couton et al. 2022). Species identification based on molecular methods still require detailed morphological identifications to be carried out when reference sequences are generated. Integrative taxonomy, where genetics and morphology are both carefully considered, is required in order to ensure accurate species identifications (Dayrat 2005). Monitoring programs that use eDNA or other molecular approaches require reliable databases, with reference sequences generated from specimens that have been carefully identified or updated once errors are detected. The lack of genetic references for some species can result in missing or misidentified NIS and other species using metabarcoding (Couton et al. 2022). Even more problematic however, are reference sequences with misidentifications, as they provide a false sense of certitude to authors without taxonomic expertise (Viard et al. 2019; Cognato et al. 2020). Vieira et al. (2014) first noted that high numbers of Watersipora identifications (ex. Mackie et al. 2006; Mackie et al. 2012) were erroneous and our phylogeny has since revealed that 49% of Watersipora sequences on Genbank were incorrectly identified, a percentage that rises to 65% when excluding sequences identified only to the genus level. Most misidentifications concern W. subatra being listed as “W. subtorquata”. This causes subsequent identification errors in publications using these reference sequences (ex: Duncan et al. 2022; Miralles et al. 2016; Portas et al. 2022) This problem with non-updated sequences has already been pointed out in the past, as it prevents clear conclusions on species identity and origin (Miralles et al. 2018). This might explain why W. subatra has not been detected for a long time in the Mediterranean Sea. One must note that a recent checklist of NIS in France (Massé et al. 2023) includes W. subatra in the Atlantic and Mediterranean as an established species, but the absence of W. subtorquata in this list suggests that both species are potentially synonymized. Our updated genetic reference list could be used as a guide by authors having deposited Watersipora sp. sequences to update their species identity.

Our phylogeny includes two problematic species identifications: “Watersipora sp. sensu Mackie et al. (2012)” and “W. edmondsoni”. The first can be attributed to the genetic description of a novel species from Mackie et al. (2012). This species however lacks a morphological description and could thus be an already described species that simply lacks a corresponding genetic sequence for now. It may be Watersipora atrofusca, as it co-occurs with “Watersipora sp. sensu Mackie et al. (2012)” in California. However a record of “Watersipora sp. sensu Mackie et al. (2012)” identified as “W. subtorquata” in Korea (HQ896194, Lee et al. 2011) suggests one of three alternatives: either W. atrofusca is also present in Korea, another species corresponds to “Watersipora sp. sensu Mackie et al. (2012)”, or “Watersipora sp. sensu Mackie et al. (2012)” does indeed constitute an undescribed species. Without morphological identification of the individuals, inference on these three options remains speculative. “Watersipora edmondsoni” constitutes a similar problem. The holotype of W. edmondsoni Soule & Soule, 1968 was reexamined by Vieira et al. (2014) and has been reattributed as synonymous to W. subtorquata. We here however note a genetically distinct clade containing the individuals identified as “W. edmondsoni” by two separate studies (Mackie et al. 2006; Suppl. material 1: table S1 MW277712). Vieira et al. (2014) suggested that specimens reported as W. edmondsoni in Soule and Soule (1975) could indeed include one or more species. The identified sequences could thus potentially be attributed to W. edmondsonisensuSoule and Soule (1975) (non Soule & Soule, 1968). Further specimens and sequences are required to resolve the species status of specimens reported as “W. edmondsoni”.

Due to the high damages of NIS on ecosystems and the economy (Blackburn et al. 2019; Diagne et al. 2021), they constitute a key global biosecurity issue (Lovell et al. 2006; Olson 2006; Pyšek et al. 2020). Proper monitoring of their spread is thus crucial for preventing or mitigating their impact (Pyšek et al. 2020). This however requires correct identification, as the species identity may impact how we evaluate NIS invasion patterns (Vieira et al. 2014; Viard et al. 2019; Golo et al. 2023). Here we show that some authors may have been misled due to misidentifications of Watersipora spp. (prior to its’ redescription) in the reference literature and in genetic reference banks. We suggest that authors should maintain genetic references in accordance with new research by including the name originally used on their research and potential changes to their ID after reexamination. However, this would be very time consuming. A new way of genetic database management, similar to the WoMRS database, might compensate for the time-consuming nature of follow-up corrections on sequences. Taxonomists and geneticists dedicated to a family/genus could have a right to modify scientific names associated with sequences following the recommendations of recent peer-reviewed papers. Changes (by whom, references, etc.) should be tracked for transparency. This could help avoid the perpetuation of errors and improve the monitoring of both the spread of NIS and species distributions in general. Our new sequences, as well as the table updating the identity of almost all existing Watersipora sequences (Suppl. material 1: table S1) may help with future identification of different Watersipora species.

This study was supported by the Région Bretagne through the CoEcoDigue project. (Ref/Région n° 2017-01, Ref/Ifremer n°21/1001756) and by the Chambre de Commerce et d’Industrie (CCI) du Var. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Robin P. M. Gauff: research conceptualization, methodology, investigation and data collection, data analysis and interpretation, writing – original draft; Marc Bouchoucha: research conceptualization, investigation and data collection, funding provision, writing – review & editing; Amelia Curd: research conceptualization, funding provision, writing – review & editing; Gabin Droual: research conceptualization, investigation and data collection, data analysis and interpretation, writing – review & editing; Justine Evrard: investigation and data collection, data analysis and interpretation, writing – review & editing; Nicolas Gayet: investigation and data collection; Flavia Nunes: research conceptualization, methodology, investigation and data collection, data analysis and interpretation, funding provision, writing – review & editing.