Majorca is the largest island in the Balearic archipelago, located in the western Mediterranean. Its geographical situation has made it a strategic point for trade and exchange with the rest of the Mediterranean since the Phoenician colonies in the third century BC (Aubet 2001), and since the mid-20th century tourism has played a fundamental role in its economy. The archipelago is considered one of the main tourist destinations in Europe, and more than 800 cruise ships and 44000 merchant’ ships dock annually in its four main international ports (Palma, Alcudia, Ibiza and Mahón). In addition, Majorca’s 30 marinas have over 28000 private docks for pleasure boats (APB 2017). All this maritime traffic has likely caused the involuntary dispersion of adult specimens attached to ship hulls (Zibrowius 1991) or larvae in ballast waters (Hewitt et al. 2009).

Seven localities were sampled around the island between July 2017 and July 2019 (Fig.1), including the international ports of Palma and Alcudia and the leisure or fishing ports of Sóller, Sa Rápita and Portitxol. In addition, two localities in the bay of Palma but outside the harbour environments were also selected: Sa Porrassa and Cala Blava, in the Marine Reserve of the Bay of Palma. Several samples were taken at each locality, maximizing habitat diversity (exposure, orientation, traffic, type of substrate, etc., Table 1). Modified qualitative rapid assessment surveys (similar to Ulman et al. 2017) were undertaken with a rectangular scraper equipped with a 1 mm diameter mesh fabric at one end and a 2 m stick at the other. At each station, artificial substrates were scraped off on the shore to determine the absence or presence of targeted species. Finally, 3 to 5 L sediment samples were collected with a van Veen grab from both anthropogenic and natural environments.

Quantitative samples were also taken by scuba divers (Fig. 2A, B), who scraped a standard surface of 30×30 cm on hard substrates (both artificial and natural) at depths of 0.5 to 7 m (Fig. 2A, B). In addition, 33×33 cm fouling plates were placed at depths of between 0.5 and 5 m for 3 to 6 months in order to study settlement and colonization processes (Fig. 2D). The surfaces were also scraped after this time. Target species from all samples were sorted in the laboratory and stored at 4°C to 6°C in 96% ethanol. Additional specimens of H. norvegica, fixed and preserved in 96% ethanol, were obtained from Norway for genetic comparison with the morphologically similar H. elegans.

The aggregation or colonies present at each locality were considered as populations. A total of 52 specimens of Hydroides and 21 specimens of Ficopomatus were selected from the collected samples for molecular work.

All specimens were examined with a stereo and compound light microscope and identified to morphospecies following original and updated descriptions (e.g. Zibrowius 1971, Fauvel 1923, Bastida-Zavala and ten Hove 2002). For scanning electron microscopy (SEM), specimens were dehydrated in a series of mixtures of absolute ethanol and hexamethyldisilazane (HMDS) with the following ratios 3:1, 2:2, 1:3, and 1:1, 1:3, and then into pure HMDS. The prepared samples were mounted on holders, sputter-coated with gold (10 nm thickness) and examined with a HITACHI S-3400N scanning electron microscope at the University of the Balearic Islands. Vouchers were deposited at the Museo Nacional de Ciencias Naturales (MNCN, Madrid; Table 2).

A small portion, 1 to 2 mm, of each specimen’s thorax or a few radioles were taken for molecular work. DNA was extracted from Ficopomatus specimens using the Quick-gDNA Miniprep Kit (Zymo) according to the manufacturer’s instructions. DNA was extracted from Hydroides specimens using QuickExtract (Epicentre). Cytochrome c oxydase subunit 1 (COI) and Cytochrome b (Cytb) were amplified by PCR. The amplification reaction of COI for Ficopomatus contained 10.8 μl of water, 1.2 μl of 50 mM MgCl₂, 2 μl of buffer 10×, 0.4 μl of Biotaq 5 U μl^-1, 2 μl of dNTPs mix at 10 mM, 0.8 μl of each primer jgLCO1490/jgHCO2198 (Geller et al. 2013) at 10 mM and 2 μl of DNA (4-20 ng μl^-1). PCR reactions for amplification of Cytb and COI for Hydroides contained 4.5 μl of water, 1 μl of each primer jgLCO1490/jgHCO2198, Hydro-COIF/Hydro-COIR or Cytb424F/cobr825, 7.5 μl of MyTaq Red Mix (Bioline) and 1 μl of DNA (4-20 ng μl^-1). Primer sequences and cycling conditions are given in Table 3. PCR products were run on a 1% agarose gel containing ethydium bromide for 30 min at 80 V and visualized with UV light. Amplified PCR fragments were of around 660 bp, while Cytb fragments were of around 430 bp. Successful PCR products were cleaned using microCLEAN for PCR clean-up (Microzone) or ethanol/sodium acetate precipitation. For some samples, cycle sequencing was performed on both strands by Eurofins Genomics DNA Sequencing Department (Ebersberg, Germany). The rest of the samples were terminated using BigDye Terminator v3.1 (ThermoFisher) and sequenced on an ABI3130 sequencer. Forward and reverse reads were merged into consensus sequences and edited using Geneious v.7 (Kearse et al. 2012).

Additionally, 57 COI and 42 Cytb sequences of F. enigmaticus and Hydroides spp. were downloaded from GenBank (Benson et al. 2008, see Supplementary Material).

Sequences were aligned with MAFFT 7 online version (Katoh and Standley 2013) and alignments were checked in Aliview 1.25 (Larsson 2014). Flanking regions with missing data were removed using Gblocks 0.91b (Castresana 2000) with the softest parameters (allow for smaller final blocks, gap positions within the final blocks and less strict flanking positions).

Best-fitting models and partition schema for each marker were selected using PartionFinder 2.1.1 (Lanfear et al. 2016, Guindon et al. 2010) with the Bayesian information criterion. The number of variable and parsimony-informative sites was calculated with MEGA X 10.0.5 (Kumar et al. 2018). Two datasets were created for phylogenetic analyses: one containing unique COI sequences and one containing unique Cytb sequences.

Bayesian inference (BI) analyses were performed on both datasets to obtain ultrametric trees compatible with downstream species delimitation analyses. BI analyses were conducted using BEAST2 (Bouckaert et al. 2014) for both datasets (COI and the concatenated dataset) using the nucleotide substitution model mentioned above. A strict clock was assumed for both datasets. A Yule model was used as tree prior with a default Γ distribution as birth rate prior. A lognormal distribution with M=1.0 and S=1.25 was used for the kappa parameter prior (Drummond and Bouckaert 2015). All analyses were run with a chain length of 50000000. Convergence of each run and parameter was checked using Tracer 1.7.1 (Rambaut et al. 2018), i.e. ESS > 200. A maximum clade credibility tree was obtained with Treeannotator (Bouckaert et al. 2014) after discarding 25% of the trees as burnin. All phylogenetic analyses were performed on Cypres Science Gateway (Miller et al. 2010). Trees were visualized and edited using FigTree 1.4.4 (Rambaut 2014) and later in LibreOffice Draw 5.1.6.2.

The general mixed Yule-coalescent model (GMYC) (Pons et al. 2006, Fujisawa and Barraclough 2013) and the multi-rate Poisson tree process (mPTP) (Kapli et al. 2017) were used to delimit the number of molecular species in the datasets. GMYC and mPTP 0.2.4 were used on both BI trees with a single threshold. The methods were applied separately on the Ficopomatus and Hydroides groups for the Cytb datasets. GMYC was implemented in R (R Core Team 2015) with the packages ape 5.3 (Paradis and Schliep 2018), MASS 7.3-45 Venables and Ripley 2002), Paran 1.5.2 (Dinno 2018) and splits 1.0-19 (Ezard et al. 2017). mPTP was applied through its webserver (https://mptp.h-its.org).

Nucleotide divergence (K2P) over sequence pairs within and between the well supported lineages after the phylogenetic analyses and species delimitation analyses were estimated in MEGA X 10.0.5 (Kumar et al. 2018). Paired positions containing gaps and missing data were removed.

The presence of F. enigmaticus was confirmed in two of the seven sampled areas: the port of Palma and Portixol (Fig. 2C). In the port of Palma, it was only found on one of the deployed fouling plates. Scrapings in Portixol recovered aggregations of at least 280 individuals/m² at some sites, especially those closed to the torrent mouth, right at the end of the port. Four species of Hydroides were identified morphologically from the ports of Palma, Portixol, Alcudia and Sóller: H. dianthus, H. elegans, H. dirampha and H. nigra (Figs. 2D, 3). These five serpulid species are recorded for the first time in Majorca. Juvenile or damaged specimens without an operculum could not be identified to the species level. The most common and abundant species was H. elegans, which was found in the port of Palma at densities of up to 270 individuals/m². No Hydroides specimens were found in the ports of Sa Rápita and Cala Blava and only one specimen of one species, H. nigra, was collected at the Islet of Sa Porrassa (donated to Sun et al. 2017b). Specimens identified as H. norvegica from Norway were compared morphologically with H. elegans to confirm that the main morphological differences are in the collar chaetae, with subdistal spines in the latter absent in the former. The wide variation observed in the opercular morphology of members of both species indicates that this is a less reliable character for species discrimination.

Despite our efforts to amplify and sequence both markers, only 10 COI and two Cytb sequences of F. enigmaticus were obtained (GenBank accession numbers (AC): MT044486-MT044495, MT215014-MT215015; Table 2), as well as six COI and five Cytb sequences of Hydroides (AC: MT044496-MT044501, MT215008-MT215013; Table 2). The COI dataset contained 68 sequences and was 379 bp long, with 228 variable sites, 212 of them parsimony-informative. The Cytb dataset contained 50 sequences and was 264 bp long, with 190 variable sites, 175 of them parsimony-informative.

Specimens of F. enigmaticus split into three distinct and supported clades, and species delimitation analyses (GMYC and mPTP) confirmed the presence of three molecular species of the F. enigmaticus species complex in Majorca. For two species, F. enigmaticus Clade 4 and 5 (Fig. 4), only COI sequences are available and do not match any existing records. Therefore, these species could be new cryptic species within the F. enigmaticus complex, or else could match any of the clades previously assessed in Australia (Styan et al. 2017). For the third species, F. enigmaticus Clade 1 (Fig 4, 5), both COI and Cytb sequences are available, and this species has previously been recorded from distant bioregions such as Australia, New Zealand, California and Portugal and northern Spain (e.g. Styan et al. 2017, Yee et al. 2019, Oliva et al. 2020). K2P distances within these species range from 0.3% to 4.7%, and K2P distances between them range from 10.5% to 24.7% (Tables 4, 5).

Among specimens identified as Hydroides elegans, two species are recovered after phylogenetic and species delimitation analyses. For one species, H. elegans Clade E2 (Fig 4), only COI sequences are available, and do not match any existing records. Therefore, this species appears to be a new cryptic species within the H. elegans species complex. For the second species, H. elegans Clade E1 (Figs 4, 5), both COI and Cytb sequences are available (albeit from different specimens). This species has previously been recorded from distant bioregions such as Australia, California and China. K2P distances within these species range from 0.07% to 1.4%, and K2P distances from other Hydroides species range from 9.2% to 46.3% (Tables 4, 5).

Among the specimens identified as Hydroides dirampha, one species was recovered with one COI sequence available (Fig. 4). This species has previously been recorded from distant bioregions such as Australia and Panama. K2P distances from other Hydroides species range from 21.9% to 31.3% (Tables 4, 5).

Among the specimens identified as Hydroides dianthus, two species were recovered. For one species, H. diantus Clade A (Fig. 4), one COI sequence is available. This species has been previously identified from the Mediterranean, as well as from distant bioregions such as China, Brazil and the east coast of the US. For the other species, H. dianthus Clade B (Fig. 5), one Cytb sequences was available. This species has previously been recorded from distant bioregions such as Ukraine and Texas. K2P distances within these species range from 0.08% to 1.7%, and K2P distances from other Hydroides species range from 6.7% to 34.2% (Tables 4, 5).

One Cytb sequence was obtained from the specimen identified as H. nigra, which does not cluster with any available sequences, including that of another specimen identified as H. nigra from Majorca (Fig. 5). K2P distances from other Hydroides species range from 21.7% to 46.3% (Table 5). This result may indicate that either of the two specimens identified as H. nigra from Majorca actually belongs to a different species, with morphological features similar to the diagnostic features of this species.

Assessing cryptic annelid diversity after analyses of DNA sequences is now a common procedure (Nygren 2014). Previous analyses of Cytb sequences showed that F. enigmaticus was in fact a species complex (Styan et al. 2017, Yee et al. 2019), gathering at least two cryptic species (understood as morphologically identical but separately evolving metapopulation lineages).

Clade 1 is currently reported as widely distributed (e.g. California, Australia and Majorca) (Styan et al. 2017, Yee et al. 2019, present study). In the case of two of the lineages found in the present study, Clade 4 and Clade 5, there are no previous records of these species elsewhere, and they are for now only known from Majorca. Considering that the F. enigmaticus species complex is not originally from Mediterranean waters, they have probably been introduced, and it is also expected that members of these clades can be found in other regions worldwide. It is interesting to note that Clade 2 reported by Styan et al. (2017) and Yee et al. (2019) is recovered as two separate species in the present analyses. While we did not include all the data available for this clade in the species delimitation analyses, this suggests that molecular species delimitation is not straightforward in this group.

The situation is similar with species of Hydroides found in Majorcan waters. Both H. operculata and H. dianthus have previously been shown to be species complexes (Sun et al. 2017a,b). Hydroides dianthus Clade A has already been reported from the eastern and southern USA, Brazil, east China, Japan, Turkey and Italy. Its distribution area now includes Majorca. Although H. dianthus was originally from the USA (Verrill 1873), it is argued that it may originate from the Mediterranean (Sun et al. 2017a). If this is confirmed, although the species has not been reported in the Balearic Islands before, it would involve reconsidering their status as invasive in Mediterranean localities. The situation is different for Hydroides cf dianthus Clade B, which has been recorded from only two places, Texas (Caribbean) and Ukraine (Black Sea), and never before in the same region as Hydroides dianthus clade A. It was suggested that this species has been introduced via the Mediterranean from an American population to the Black Sea, and our record provides more evidence in support of this hypothesis. However, previous studies of H. cf. dianthus Clade B were done after analyses of COI sequences, and our record uses a Cytb sequence, which limits further interpretations. (The link of the identity of sequences from these two markers was made possible by later sequencing of a voucher for more global phylogenetic purposes - see Sun et al. 2018.)

Genetic data has a great potential for detecting invasive species (Muñoz-Colmenero et al. 2018), determining the source of the invasion and understanding the routes taken by the species (Geller et al. 2010, Yee et al. 2019). When encountering cryptic diversity, it is also particularly important to distinguish between the potentially invasive and native lineages. Though it was one of the aims of the present study, despite the recent progress in the COI DNA barcoding of Hydroides species (Sun et al. 2012), we encountered many difficulties in sequencing COI and Cytb from Hydroides and Ficopomatus. While the data we obtained are sufficient to detect the presence of the species complexes and identify the species present in Majorcan waters, they are not sufficient to properly assess their population structure or the sources of the invasion.

The presence of F. enigmaticus (sensu lato) has been observed in the ports of Palma and Portitxol and near the mouth of torrents. These environments are characterized by low hydrodynamism (expect for after a large rainfall), eutrophic waters and changes in salinity. This kind of habitat has been observed previously in other studies dealing with this species (Yee et al. 2019) and conforms to the ecology of the optimal habitat for the proliferation of F. enigmaticus (sensu lato), which consists of enclosed environments with murky waters. The wide ranges of tolerance to salinity and temperature give members of this species complex the ability to establish themselves in these variant environments, unlike other serpulid species (Dittmann et al. 2009).

During the summer, there was a greater proliferation and density of F. enigmaticus populations, as previously observed in other studies (Vuillemin 1958, Dixon 1981). According to the literature, the minimum temperature to reproduce ranges between 14°C and 18°C, which in the Balearic Islands corresponds to the month of May. The main episode of settlement and growth is between spring and summer, which correlates with an increase in biomass and carbonated production (Fornós et al. 1997). Knowing that complete maturation takes four months (Obenat and Pezzani, 1994), large aggregations may appear during the summer. The addition of the intensification of the international maritime traffic at this period increases the potential for dispersion.

Species diversity showed an uneven presence and abundance between the locations sampled (Table 1). The most common species were H. dianthus and H. elegans, which were present at more sites within the ports of Palma, Portixol and Alcudia. Hidroides dirampha occurred in the ports of Palma, Alcudia and Sóller, and H. nigra only in Portixol (Table 1). Accordingly, the port environments with the greatest species diversity were in descending order the ports of Palma, Alcudia, Portixol and Sóller. There seems to be some correlation between the ports with the highest volume of maritime traffic (Palma, Alcudia and Portixol) and the greatest diversity of invasive species, which would conform to the expected results, because a greater flow of maritime transport could favour the translocation of these species associated with fouling communities embedded on the hulls of the boats (e.g. Zibrowius 1971, ten Hove and Kupriyanova 2009, Çinar 2012). In the present study, the preference of Hydroides for artificial substrates is verified (as in Pawlik 1992, Kupriyanova et al. 2001), with four species being present in the ports of Palma and Portixol, while none were recorded on rocky substrates in nearby anthropized areas, such as Cala Blava, and only one specimen in Sa Porrasa Islet (Sun et al. 2017b).