The genus Arion Férussac, 1819 is the most species rich genus of the terrestrial slug family Arionidae (Mollusca, Pulmonata, Gastropoda). It comprizes approximately 40 species, grouped into four subgenera, viz. Arion s.s. Férussac, 1819, Kobeltia Seibert, 1873, Carinarion Hesse, 1926 and Mesarion Hesse, 1926. Species of the subgenus Mesarion (type species: Limax subfuscus Draparnaud, 1805) are characterized by 1) a medium body-size (up to 75 mm when extended), 2) an orange to dark brown dorsum, 3) two dark bands on the sides of the mantle, 4) (usually) yellow to orange body mucus, and 5) an enlarged free-oviduct with a long and V-shaped ligula (Kerney et al. 1983). Many Mesarion species are highly polymorphic with respect to body colour and genital anatomy. As a consequence, the species limits and phylogenetic relationships of taxa within this subgenus have been debated for decades (e.g. Garrido et al. 1995, Castillejo 1997, 1998, Pinceel et al. 2004, 2005a, b, Quinteiro et al. 2005). Arion subfuscus (Draparnaud, 1805) (type locality: Montagne Noire, France) is probably the most problematic “species” within Mesarion as it shows an overwhelming amount of variation in body pigmentation, genital anatomy, and reproductive behavior [see Garrido et al. (1995) and the references listed in their table 1]. This variation often has been interpreted as indicating reproductive isolation between geographically isolated populations, and Arion subfuscus thus is considered a species complex (Wiktor 1973, Waldén 1976, De Winter 1986, Backeljau 1989, Altonaga et al. 1994, Backeljau et al. 1994, Garrido et al. 1995). Especially in the Pyrenees and the coastal regions of Spain there are local, morphologically diverged populations (e.g. Garrido et al. 1995, Castillejo 1998). Several of these have been described as endemic species on the basis of where the epiphallus, oviduct and pedunculus of the bursa copulatrix open into the atrium, in combination with differences in the relative lengths of the vas deferens and the epiphallus (e.g. Castillejo 1998, Garrido et al. 1995, Quintana Cardona 2007). Two of these endemic taxa occur in the eastern coastal region of Spain or the Balearic Islands, viz. Arion gilvus Torres Mínguez, 1925 and Arion ponsi Quintana Cardona, 2007.

Arion ponsi (Figure 1) was described from Menorca (Balearic Islands, type locality: Barranc d’Algendar). The species has a medium body size (range: 54–66 mm), an orange to beige dorsal body colour with dark lateral bands that can be blurry in the posterior parts, a foot sole that is cream coloured with a greyish hue, and a transparent body mucus (Quintana Cardona 2007). Its genital anatomy is very similar to that of Arion gilvus, Arion iratii Garrido, Castillejo & Iglesias, 1995, Arion molinae Garrido, Castillejo & Iglesias, 1995 and Arion lizarrustii Garrido, Castillejo & Iglesias, 1995, but its epiphallus is shorter than the vas deferens (as in Arion molinae) and opens into the genital atrium in between the oviduct and the pedunculus of the bursa copulatrix (unlike in Arion molinae, where the pedunculus is positioned in between the epiphallus and oviduct) (figures 3–5 in Quintana Cardona 2007).

Arion gilvus (Figure 2) was described from ‘Mandol’ in the Spanish Province of Tarragona. However, the toponym ‘Mandol’ seems to be erroneous (e.g. Bech 1990) and therefore Castillejo (1990) assigned eight specimens with an Arion gilvus morphology from Serra de Pandóls near Gandesa (Province of Tarragona) as topotypes [see also Castillejo and Rodríguez (1991)]. Subsequently, Arion gilvus was redescribed by Garrido (1992). Afterwards, the species has also been found in the Provinces of Valencia, Teruel and Albacete [Borredà (1994), figure 15 in Castillejo (1997), figure 1 in Quinteiro et al. (2005)]. Arion gilvus reaches a length of up to 65 mm when extended. It has a yellowish to brown dorsum that gets lighter downwards at the sides and dark lateral bands that have a yellowish grey line on their upper side (Figure 1). The sole is white or evenly yellowish and the mucus is pale yellow (Torres Mínguez 1925, Bech 1990, Garrido 1992, Castillejo 1997). The epiphallus, the pedunculus of the bursa copulatrix, and the free oviduct join the atrium on a single line with the pedunculus of the bursa copulatrix in the middle, as in Arion molinae, but in contrast to the latter, the epiphallus is longer than the vas deferens (Torres Mínguez 1925, Borredà 1994, Castillejo 1997, and figures 26–28 in Garrido et al. 1995).

As illustrated by Arion ponsi and Arion gilvus, the alleged species-specific genital differences among the Iberian species of the Arion subfuscus complex are very subtle and little is known about their intraspecific variation. Moreover, genital differences among arionid taxa do not necessarily imply reproductive isolation (Dreijers et al. 2013). Hence, if alleged species-specific phenotypic differences in arionids are to be interpreted under a phylogenetic species concept, then their correlation with reproductive isolation should be corroborated by molecular data. Molecular markers have been very effective in this respect (e.g. Pinceel et al. 2005a, b, Quinteiro et al. 2005, Geenen et al. 2006, Jordaens et al. 2010). As such, Quinteiro et al. (2005) investigated the taxonomic affinities of Iberian Mesarion species using DNA sequence data. Their analysis of the nuclear ribosomal internal transcribed spacer 1 region (ITS1) showed a polytomy of Mesarion species, yet, the analysis of the mitochondrial NADH dehydrogenase I (ND1) gene suggested a strongly bootstrap supported group of Iberian Mesarion species with a continental-Mediterranean distribution (Arion paularensis, Arion baeticus, Arion urbiae, Arion anguloi, Arion wiktori, and Arion gilvus), and an unsupported group of species with an Atlantic distribution (Arion lusitanicus, Arion nobrei, Arion fuligineus, Arion hispanicus and Arion flagellus). In addition, the positions of three Pyrenean species (Arion lizarrustii, Arion iratii, Arion molinae) remained unresolved. More specifically, the ND1 data placed Arion gilvus as sister taxon of Arion urbiae and Arion anguloi. Quinteiro et al. (2005) did not study individuals from the Balearic Islands and thus probably did not include Arion ponsi.

Because DNA sequence data do not only provide phylogenetic information, but can also serve as DNA barcodes for species identification (Hebert et al. 2003, 2004), we here expand on the work of Quinteiro et al. (2005) by 1) characterizing Arion gilvus and Arion ponsi using mitochondrial COI and 16S rDNA gene fragments, and the larger part of the nuclear ITS1 region, 2) exploring the phylogenetic affinities of Arion gilvus and Arion ponsi within the subgenus Mesarion, and 3) providing diagnostic COI barcodes for both species.

Information on the species and specimens included here is provided in Table 1. In total, we screened 45 specimens (Table 1). DNA was extracted from small parts of the foot using a NucleoSpin Tissue Kit (Macherey-Nagel, Düren) following the manufacturer’s instructions. PCR reactions were done in 25 µl reaction volumes that contained 1.5 mM MgCl₂ in 1 × PCR buffer (Qiagen), 0.2 mM of each dNTP, 0.2 µM of each primer and 0.5 units of Taq polymerase (Qiagen). A fragment of the mitochondrial COI and 16S genes was amplified using primer pairs LCO1490 and HCO2198 (Folmer et al. 1994) and 16Sar and 16Sbr (Palumbi 1996), respectively. The nuclear ITS1 region (except the ± first 30 bp) was amplified using the primer pair ITS1L and 58C (Hillis and Dixon 1991). The PCR profile was an initial denaturation step of 5 min at 95 °C, followed by 35 cycles of 45 s at 95 °C, 45 s at an annealing temperature of 40 °C (COI), 42 °C (16S) or 55 °C (ITS1) and 1.5 min at 72 °C, and ending with a final extension step of 5 min at 72 °C. PCR products were purified using the GFX PCR DNA Purification Kit (GE Healthcare) following the manufacturer’s instructions. Purified DNA was diluted in 15 µl of sterile water. PCR-products were bidirectionally sequenced using the ABI PRISM BigDye® Terminator v1.1 Cycle Sequencing Kit and run on a ABI3130xl Genetic Analyzer. Sequences were assembled in SeqScape v2.5 (Life Technologies) and inconsistencies were checked by eye on the chromatogram. Sequences were submitted to GenBank under accession numbers KF305196–KF305225 for COI, KF356212–KF356245 for 16S and KF385449–KF385469 for ITS1. These datasets were supplemented with DNA sequences from GenBank [including a few species of the other Arion subgenera (Table 1)]. We used those of Carinarion as outgroup.

Sequences were aligned in ClustalW (Thompson et al. 1994) with default settings and without subsequent manual adjustments. In each alignment sequences were trimmed to equal length. The final alignments had a length of 504 bp (COI), 408 bp (16S) and 587 bp (ITS1), and of 1499 bp after concatenating the three fragments. The COI sequences were translated to amino acid sequences to check for stop codons (but none were found). The ITS1 sequences were also analysed together with those of Quinteiro et al. (2005). In this way we could extend our taxon coverage to Arion hispanicus Simroth, 1886, Arion fuligineus Morelet, 1845 and Arion nobrei Pollonera, 1889 (Table 1). Because Quinteiro et al. (2005) used other ITS1 primers, we had to trim this dataset to a length of 378 bp. For each gene fragment, and for the concatenated dataset, we constructed Neighbour-Joining (NJ) trees (Saitou and Nei 1987) using the Kimura 2-parameter (K2P) model in MEGA v5 (Tamura et al. 2011) with complete deletion of insertions and deletions (indels). Branch support was evaluated with 1000 bootstrap replicates (Felsenstein 1985). Only bootstrap values ≥ 70% were considered as indicating strong support (Hillis and Bull 1993). Uncorrected p-distances (hereafter simply referred to as p-distance) were calculated in MEGA v5 (Tamura et al. 2011). For these calculations we considered the following Molecular Operational Taxonomic Units (MOTUs): 1) the five 16S rDNA clades of Arion subfuscus (S1 to S5) defined by Pinceel et al. (2005a), 2) Arion anguloi and Arion urbiae jointly as a single MOTU (Backeljau et al. 1994, Quinteiro et al. 2005), 3) Arion wiktori and Arion paularensis jointly as a single MOTU (Backeljau et al. 1996, Quinteiro et al. (2005), and 4) Arion lusitanicus from Portugal vs. Arion lusitanicus from elsewhere as two different MOTUs (Davies 1987, Castillejo 1998, Quinteiro et al. 2005). Standard errors of mean p-distances among taxa and MOTUs were calculated on 1000 bootstrap replicates.

The NJ-tree of the concatenated dataset confirms the major outcomes of previous phylogenetic studies, viz. 1) a strong support for the monophyly of the subgenus Carinarion (Geenen et al. 2006), 2) a clade of Arion s.s. and non-Portuguese Arion lusitanicus (Quinteiro et al. 2005), 3) Arion wiktori clustering with Mesarion species, in particular with Arion paularensis (Quinteiro et al. 2005) instead of with Kobeltia species (Castillejo 1998), and 4) the strong differentiation within Arion subfuscus s.s. that consists of, at least, five phylogenetic species (Pinceel et al. 2005a). It therefore seems that the analysis of COI, 16S and ITS1 DNA sequences yields relevant taxonomic information with respect to the characterisation of arionid species that have been described under the morphospecies concept.

Because Arion gilvus and Arion ponsi were originally described on morphological grounds they are to be interpreted as morphospecies. This phenetic morphological distinction, however, correlates well with a phenetic separation based on mtDNA distances. Indeed, the overall mean p-distance among the Mesarion MOTUs (excluding Arion ponsi and Arion gilvus) dealt with in this study is 16.7% for COI and 13% for 16S. As such, the mean p-distances between Arion ponsi and Arion molinae (COI: 14.2%; 16S: 16.2%) or between Arion gilvus and Arion urbiae (COI: 14.5%; 16S: 13.4%) are perfectly comparable with the mean p-distances among the other MOTUs and morphospecies in Mesarion. Hence, with respect to mtDNA divergence, both Arion ponsi and Arion gilvus, behave as other Mesarion species or putative species-level MOTUs.

Obviously, the strong COI differentiation among Mesarion taxa, and of Arion ponsi and Arion gilvus in particular, suggests that DNA barcoding may be a suitable identification tool for these animals. Yet, this may be a too simplistic conclusion, since stylommatophorans may show extremely high intraspecific mtDNA divergences of sometimes up to 27% (K2P-distances, but note the uncorrected p-distances are almost similar) (Thomaz et al. 1996, Chiba 1999). In addition, Davison et al. (2009) showed that in the Stylommatophora the mean interspecific K2P-distances (± 3%) can be substantially lower than the mean intraspecific K2P-distances (± 12%). Under these conditions, it becomes very difficult to define generally applicable thresholds that distinguish between intra- and interspecific sequence divergences. Such thresholds are normally associated with DNA barcoding gaps (Hebert et al. 2003), but Davison et al. (2009) were unable to detect DNA barcoding gaps in the taxa they studied. Nevertheless, Davison et al. (2009) suggested a pragmatic 4% threshold to separate intra- and interspecific values, but at the same time they also concluded that DNA barcoding in itself is insufficient to identify and/or detect stylommatophoran species. Unfortunately, our sample sizes were too small to explore eventual DNA barcoding gaps in Mesarion.

Because DNA barcoding on its own may be unreliable for identifying and detecting species-level taxa in stylommatophorans, it its necessary to backup this sort of data with, amongst others, phylogenetic analyses. As such, our phylogenetic trees of the DNA sequence data show that the morphospecies Arion ponsi and Arion gilvus, also represent phylogenetic species, since both form well-supported clades that are “significantly” associated with well-defined, but morphologically different sister species. For Arion ponsi, the sister species appears to be Arion molinae, the distribution range of which is located in NE continental Spain (Castillejo 1997), i.e. north of, and facing, the Balearic Islands. Conversely, the sister taxon of Arion gilvus is the “tandem” of Arion urbiae and Arion anguloi, two species that have been synonymized by Backeljau et al. (1994) and that jointly should be referred to as Arion urbiae. Our DNA sequence data on COI, 16S and ITS1 (e.g. Figure 3), as well as those on ND1 and ITS1 of Quinteiro et al. (2005) are in line with this. As such, the distribution range of Arion urbiae is situated northwest of, and probably adjacent to, that of Arion gilvus. Thus, for both the species pairs Arion ponsi / Arion molinae and Arion gilvus / Arion urbiae, the distribution ranges appear at least consistent with the suggested sister group relationships.

In conclusion, the present work shows that Arion ponsi and Arion gilvus clearly differ from Arion subfuscus or any other currently recognized arionid species. As such, former records of Arion subfuscus from Menorca (e.g. Gasull and van Regteren Altena 1970, Mateo 1993, Beckmann 2007) almost certainly refer to Arion ponsi. Similarly, probably all reports of Arion subfuscus in the regions of Valencia and Albacete involve Arion gilvus (e.g. Borredà 1994, Borredà and Collado 1996). Finally, Borredà (1994) wondered about the eventual relationship between Arion subfuscus from Menorca and Arion gilvus. The current data confirm unambiguously that these are two different species, with the former being Arion ponsi. Yet, the overall phylogenetic relationships within Mesarion and many other Arion subfuscus-like taxa remain to be resolved. In this context, one of the main questions is whether Mesarion in its present use is a monophyletic taxon. At the same time one may wonder about the relationships with the subgenus Arion s.s., with which Mesarion seems to form a well-supported clade (Figure 3).