Introduction

Traditional methods for identifying earthworm species and their phylogenetic relationships (i.e., the study of their morpho-anatomical features) have been limited by high levels of homoplasy. The structural simplicity of earthworms, the low degree of variability and the overlap of diagnostic characters among species, the absence of a fossil record and their adaptation to life in the soil, are the principal factors responsible for the difficulties in recognizing species. DNA sequence data has however facilitated the distinction of closely related species and may be the solution to understanding the true level of biodiversity within morphologically-difficult groups, such as earthworms.

Some degree of controversy has arisen on how to describe and delimit species, but discrete morphological features remain the most used criterion. Others are in favour of molecular-based descriptions (e.g., Cook et al. 2010) who justify species descriptions based solely on DNA sequences, even ignoring morphological data, but they also recognize that in cases with incomplete molecular databases – as for most taxa –, this alternative alone is not viable. Species descriptions including both morphological and DNA-based data are imperative for a more universal taxonomy. There are many authors in favor of this integrative taxonomy, consisting in a multidisciplinary approach including morphological, molecular, ecological and geographical data. This type of approach can include complex procedures therefore using multi-gene genetic distances, analyses such as General Mixed Yule-Coalescent (GMYC) models or Automatic Barcode Gap Discovery (ABGD) analyses and weighting of the established hypotheses with complementary data such as morphological, geographical or ecological (see Puillandre et al. 2012 and included references).

Hormogastridae includes middle to large-sized earthworms, currently comprising 27-29 species and subspecies that are exclusively distributed in the western Mediterranean (Díaz Cosín et al. 1989, Cobolli Sbordoni et al. 1992, Blakemore 2004, 2008), where they play a very important ecological role (Bouché 1972). The highest abundance of species seems to be located in the NE Iberian Peninsula, where more than a dozen species have been described.

The taxonomy of this group, as in other earthworm families, has been based until now solely on morphological features. The first species described are Hormogaster redii Rosa, 1887 and Hormogaster pretiosa Michaelsen, 1899. Subsequently, other species were added to the group by different authors, including Cognetti (1914), Zicsi (1970), Bouché (1970), Álvarez (1971, 1977), Díaz Cosín et al. (1989), and Rota (1993) but most were described by Qiu and Bouché (1998), including eleven new species from Spain presenting very subtle morphological differences. The known species are grouped in four genera, Hormogaster Rosa, 1887 (22-24 species and subspecies), Hemigastrodrilus Bouché, 1970 (one or two subspecies), Vignysa Bouché, 1970 (two species) and Xana Díaz Cosín et al., 1989 (one species).

Omodeo (1956) provided the first revision of the family in 1956, and later on Omodeo and Rota (2008) presented additional considerations on their evolution in an article including different Mediterranean areas. Cobolli Sbordoni et al. (1992) provided the first phylogenetic hypothesis of the family using allozymes, but that seminal work lacks a comprehensive sampling in NE Spain, where most of the hormogastrid diversity concentrates (Qiu and Bouché 1998). More recently, Novo et al. (2009, 2010) used DNA sequence data from multiple markers to detect cryptic diversity within Hormogaster elisae in the central area of the Iberian Peninsula. These studies highlight, among other aspects, the morphological stasis present in this group, whose anatomy seems to have adapted to the dry soils of this region.

During a collecting trip for the phylogenetic study of Novo et al. (2011), 22 hormogastrid specimens were collected near Sant Joan de les Abadesses (Girona, Spain). The specimens were assigned to Hormogaster, but were thought to represent a new species here described as Hormogaster abbatissae sp. n. Its description, including its relationship to other the closely related hormogastrid species are the initial objectives of this paper. The description we provide is complemented with a molecular analysis of different genes in the closest species, GMYC species delimitation and ecological data. This study resembles the first example to describe a new earthworm species by combining all these different data sources (Blakemore and Kupriyanova 2010, see also Blakemore 2010, Blakemore et al. 2010 and for lumbricids Blakemore and Grygier 2011 and Blakemore 2012) and also other studies on different taxa (e.g. Chullasorn et al. 2011, Heethoff et al. 2011, Hart et al. 2012). However, this is the first work to do so for an hormogastrid. Contradictory results between morphological and molecular data are found, and whether a particular data set should be favored over the remaining sources is discussed.

We expect that this example, combining molecular and morphological data and including ecological features, goes beyond the specific interest of a new earthworm species description and could be applied to other groups with comparable taxonomic problems.

Material and methods

Specimens were collected by hand and fixed in the field in ca. 96% EtOH, with subsequent alcohol changes. Once in the laboratory, specimens were preserved at -20 °C.

The studied material includes 22 specimens (five mature specimens, four semi-mature specimens with tubercula pubertatis and/or clitellum draft and 13 immatures or fragments) collected between Ripoll and Sant Joan de les Abadesses, road C26, km 210 in a little forest near the Ter river (42°13'30.0"N, 2°14'57.5"E). Mean annual temperature is 14.3 °C and mean annual precipitation is 724 mm, as indicated by the nearest weather station (in the airport of Girona, 55km away: http://www.aemet.es/es/serviciosclimaticos/datosclimatologicos/valoresclimatologicos?l=0367&k=cat)

Specimens have been deposited in the Oligochaete Cryo collection of the Departamento de Zoología y Antropología Física, Universidad Complutense de Madrid (DZAF, UCM), Spain.

Specimens of nearly all other hormogastrid species were examined for comparison (list of specimens in Novo et al. 2011). Morphological characters include those features traditionally used for hormogastrids and other earthworms. Only the distantly related species Hormogaster lleidana Qiu and Bouché, 1998 and Hormogaster multilamella Qiu & Bouché, 1998 were not examined, and thus their information was limited to the published descriptions (Qiu and Bouché 1998). All the specimens are deposited in the earthworm criocollection of Complutense University of Madrid (DZAF, UCM).

Molecular data generation follow Novo et al. (2011, 2012). Phylogenetic inference and GMYC analyses discussed here are based on data published in those papers. Nine molecular regions of specimens SAN1, 2, 3, 4, 7, 8, 9, 10 were included: mitochondrial regions of cytochrome c oxidase subunit I (COI), 16S rRNA and tRNA Leu, Ala, and Ser, two nuclear ribosomal genes (complete 18S rRNA and a fragment of 28S rRNA) and two nuclear protein-encoding genes (histones H3 and H4). GeneBank accession numbers for the paragenetypes, following Chakrabarty (2010) for the mitochondrial markers, analyzed here are shown in Table 1.

We constructed networks with SplitsTree4 v.4.11.3 (Huson and Bryant 2006) for the mitochondrial genes (16S-tRNA, COI), including the phylogenetically closest species of Hormogaster abbatissae sp. n., in order to visualize in more detail the relationships and genetic distances among them. Default settings were used. We analysed 41 sequences of each gene including hormogastrids close to Hormogaster abbatissae sp. n.and Hormogaster elisae Álvarez, 1977 from Siguero and Aporrectodea trapezoides (Dugés 1828) as more distant outgroups (see Table 2). Uncorrected pairwise differences were calculated between these species with Arlequin 3.5 (Excoffier et al. 2005).

Results Taxonomic results

Phylum Annelida Lamarck, 1802

Subphyllum Clitellata Michaelsen, 1919

Class Oligochaeta Grube, 1850

Order Haplotaxida Michaelsen, 1900

Family Hormogastridae Michaelsen, 1900

Genus Hormogaster Rosa, 1887

Type-species Hormogaster redii Rosa, 1887

Discussion

Most species within the genus Hormogaster are very similar morphologically, with the clitellum, tubercula pubertatis, spermathecae and typhlosole, in addition to size or colour, being the key morphological characters traditionally used for species diagnosis. Table 5 includes a comparison of the characters of Hormogaster abbatissae sp. n. with those of its closest congeners, showing a large degree of overlap in the distribution of these characters and their states. In this case we have a species that appears the closest morphologically, Hormogaster riojana, collected in Alesanco, a locality ca. 420 km from Sant Joan de les Abadesses, that can be distinguished by the body and clitellum colour, shape of the tubercula pubertatis and the number of spermathecae (although Hormogaster riojana specimens with three pairs of spermathecae have been reported by Novo 2010). This could lead to consider Hormogaster abbatissae sp. n. a variety of Hormogaster riojana. Nevertheless, as shown by the phylogenetic and phylogeographic analyses of molecular data (see Figure 2), Hormogaster riojana appears distantly related to Hormogaster abbatissae sp. n.

The sister group of Hormogaster abbatissae sp. n. is Hormogaster sylvestris (Figure 2), collected in Montmajor, 50 km away from Sant Joan de les Abadesses. These two species, closely related phylogenetically and biogeographically, are easilydistinguished by their tubercula pubertatis (generally starting in more anterior segments and finer in Hormogaster abbatissae sp. n), clitellum (shorter and saddle shaped in Hormogaster abbatissae sp. n. and annular in Hormogaster sylvestris), spermathecae (three pairs in Hormogaster abbatissae sp. n. and two pairs in Hormogaster sylvestris) and typhlosole (15 lamellae in Hormogaster abbatissae sp n. and 13 in Hormogaster sylvestris). To these characters we can add other more variable characters such as colour, length, weight and number of segments (Hormogaster sylvestris is longer, heavier and with a higher number of segments). Of all these characters, the presence of three pairs of spermathecae in Hormogaster abbatissae sp. n. is the most conspicuous trait. It is therefore the combination of the morphological information and the phylogenetic position of the species, as derived from the molecular data, which aids in the global taxonomy of the group and serves to assess the degree of homoplasy in characters thought to be of taxonomic importance.

Some characters, such as the presence of Morren’s glands or the existence of a posterior gizzard, can be difficult to observe and of subjective interpretation. Morren’s glands seem to be absent because although an enrichment of blood vessels is detected in the oesophageal wall of some segments 10 – 14, the lamellae that define this organ were never observed. Likewise, the presence of a posterior gizzard is difficult to determine, as the gut thickens in segments 15 – 19 in the members of some species. However, in Hormogaster abbatissae sp. n. there is neither strong musculature, nor the thickening and hard covering of the lumen as observed in the gizzards of earthworms.

Regarding the molecular characters, Novo et al. (2009, 2010) proposed the presence of five cryptic species within the Hormogaster elisae complex, which resulted to be separated by genetic divergences between 9.41 – 18.31% for cytochrome c oxidase subunit I (Kimura 2-parameter distances, whose values are slightly higher than the uncorrected distances, used here). Also Hebert et al. (2003) reported comparable divergences for the same marker between 11.3%, for congeneric species of various animal groups and 15.7% between annelid species. It is evident, though that strict phenetic distances cannot be used for delimiting taxonomic boundaries, as other studies have shown that the same marker may have within species divergences much larger than the ones proposed by Hebert et al. (2003) (e.g., Barber et al. 2006, Boyer et al. 2007). This has been debated for earthworms by Chang and James (2011), who proposed that differences among species are indeed clade-specific, but they propose the existence of a consensus for COI (Kimura-corrected) distances: values under 9% normally indicate the same species, while values above 15% most probably indicate different species and values between 9-15% can be ambiguous. The species pairs Hormogaster abbatissae sp. n. and Hormogaster sylvestris as well as Hormogaster riojana and Hormogaster ireguana present COI uncorrected divergences within this ambiguity range (11.71% and 9.53%, respectively). The latest species were described by Qiu and Bouché (1998) based on morphology. Therefore, it seems that in this case distances need to be treated cautiously, thus reinforcing the critiques of their use for species delimitation (DeSalle et al. 2005, Hickerson et al. 2006, Whitworth et al. 2006). However, distances seem to be conservative in hormogastrid’s case and our data suggest that divergence below the level proposed by Chang and James (2011) may correspond to different species. Anyway it is clear the necessity of morphological data to verify the status of two lineages that present a divergence value within this range. Moreover, in the present case, these species are known to appear only in their type locality and therefore barcoding gap (ABGD) cannot be calculated with accuracy. Species delimitation with GMYC has been recently implemented in earthworms by Fernández et al. (2012) and particularly in hormogastrids by Novo et al. (2012). In both cases an overestimation of the species number, when compared with morphology, was detected. The marked genetic structure and scarce dispersion capacity of the studied earthworms could be the cause for this overestimation, being these factors particularly evident in Hormogaster elisae’s case, with various cryptic species. A GMYC analysis shows Hormogaster abbatissae sp. n. as a different entity from Hormogaster sylvestris. Whether the GMYC method is overestimating in this particular case is unknown but Hormogaster abbatissae sp. n., Hormogaster riojana and Hormogaster sylvestris are well-separated when combining morphology, phylogenetic analyses and network information.

After examining its morphology, phylogenetic placement and additional data such as GMYC and soil characteristics, it is evident that Hormogaster abbatissae sp. n. constitutes a new hormogastrid taxon not phylogenetically related to those species that show closest morphological similarities. Morphological and molecular data supply different signals thus clashing in the case of Hormogaster abbatissae sp. n. The question arising in this case is what should taxonomists do when different data sources provide conflict? The answer to this question is not straightforward. On the one hand, these animals can present a morphological stasis, as shown in Hormogaster elisae (Novo et al. 2009, 2010). On the other hand, molecular techniques rely on limited information, in this case based on a group of specific genes and depend on specific algorithms. This decision should thus be based on a profound knowledge of the morphological variability and peculiarities of the studied group, and an understanding of the strengths and weakness of the applied molecular analyses (used genes, sampling scheme, algorithms, etc.) that could lead to different decisions depending on the study case.

In this particular case, phylogeny is robust because it is based on a great amount of data, combining mitochondrial and nuclear genes (COI, 16S-tRNA, H3, H4, 28S, 18S) with different phylogenetic signal and including individuals representing most of the species in the family. Also we know that living conditions in the soil induce cryptic speciation processes in earthworms (King et al. 2008, Novo et al. 2009, 2010, James et al. 2010, Buckley et al. 2011, Dupont et al. 2011 – but see rebuttals of these “cryptic” genetic cladograms by Blakemore et al. 2010 and Blakemore 2010, 2011) and in many occasions the most important morphological characters used for earthworm species delimitations overlap showing a poor discrimination capacity (Fernández et al. 2012). Therefore, morphological characters should be applied cautiously by earthworm taxonomists in case of conflict with other data source.

Regarding ecological factors, some important differences are detected for texture and organic matter among soils of Hormogaster abbatissae sp. n., Hormogaster sylvestris and Hormogaster riojana. However, it should be considered that a single locality is known per species and that the discovery of other populations may show a higher ecological range.

In summary, this study evidences the need of complementing the morphological data with molecular characters data in taxonomy, especially in groups with limited morphological characters and rampant convergence in their functional morphology, perhaps due to strong selective pressure due to habitat restriction. This study also proves that in case of rather small genetic divergence (within the range of uncertainty), morphology can be also helpful to conclude complementing molecular sources. We propose to establish the new species Hormogaster abbatissae sp. n. Given the existence of species closely-related phylogenetically (Hormogaster sylvestris) and an unrelated but morphologically similar species (Hormogaster riojana), a more exhaustive sampling effort in NE Spain could provide new diversity to help evaluate this situation. As indicated by Sites and Crandall (1997), species descriptions are not facts, but hypothesis established when certain criteria available in a specific moment are fulfilled and they can be accepted or rejected when new data are available.