Life history, patchy distribution, and patchy taxonomy in a shallow-water invertebrate (Mollusca: Polyplacophora: Lepidopleurida)

Things without names are difficult to rationalise, and so species that go without names are difficult to conserve or protect. This is a case study in resolving conflicts in historical taxonomy and ‘real’ species (identifiable and evolutionarily relevant groupings) using an approach including population genetics, natural history, and pragmatism. We report the observation that populations of a shallow-water chiton species from Washington and British Columbia demonstrate extremely high site fidelity and patchy distribution. Their limited dispersal potential and isolation could be explained by a brooding life history. This stands in direct contrast with the supposedly wide distribution of this “species”, Leptochiton rugatus (Carpenter in Pilsbry, 1892) sensu lato, from the Sea of Japan to Baja California. But this lineage has previously been suggested to comprise several cryptic species. Indeed, a haplotype network analysis using 61 individual sequences of the cytochrome oxidase c subunit I gene for L. rugatus s.l. revealed four discrete clusters which correspond to different parts of the geographic range. We infer these to represent four distinct species, at least two of which are likely novel. Leptochiton rugatus sensu stricto is herein reinterpreted as restricted to California and Baja California, and the new name L. cascadiensis sp. nov. is established for the lineage with a distribution in the Cascadia coastal bioregion from the panhandle of Alaska to Oregon. There are minor morphological differences among these species in the L. rugatus species complex, but genetic data or morphological observations alone would not have been sufficient to definitively recognise these groups as species-level lineages. The observation that different species within the complex may have different life history strategies provides important support for interpreting different populations as genuinely separate species.

This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No H2020-MSCA-IF-2014-655661.

Introduction
Genetic information has repeatedly revealed higher levels of first-order diversity than traditional morphological assessment. In most cases, genetic differences prompt a re-examination of previously overlooked morphological variation, which reveals "pseudocryptic" assemblages differentiated by these emergent features (Knowlton 1993). But molecular evidence, like some morphological evidence, may not be sufficient in isolation to make a taxonomic determination, especially when dealing with one gene region or one character (Riedel et al. 2013). Overlooked differences are found to correlate with historical hypotheses about species, representing additional taxonomic names that have been consigned to obscurity as junior Communicated by V. Urgorri This article is registered in ZooBank under urn:lsid:zoobank.org: pub:A5CD7021-089C-4494-B0CA-875FB64F72ED synonyms. These synonymies have been flagged as a key bottleneck in quantifying global species richness (Costello et al. 2013a), a problem that needs to be addressed through concerted efforts by taxonomic experts. It is presently unclear to what extent the number of available names in synonymy balances the number of global marine species left undescribed.
Polyplacophoran molluscs (chitons) have a reputation as a taxonomically "difficult" group. In particular, the genus Leptochiton GRAY, 1847 is well known to be paraphyletic; although key features clearly separate more than 120 accepted living species, no characters have been identified that can be used to diagnose supraspecific groups (Sigwart et al. 2011). Many species of the genus Leptochiton are known from deepsea habitats, but several species are known from shallow water, including the North Pacific Leptochiton rugatus (Carpenter in Pilsbry, 1892). This species (complex) ranges from the Sea of Japan across to Alaska, down the coast of North American to Baja California (Kaas and Van Belle 1985). It is anecdotally recognised as representing potentially multiple distinct species (e.g. Vendrasco et al. 2012). Previously published population genetics results recovered two distinct 'populations' (Kelly and Eernisse 2007), but these were never applied to taxonomic revision, and taxonomic features suitable for distinguishing these apparently cryptic species have not been reported.
One often overlooked aspect of the identification of novel species is the importance of basic natural history observations. Life history information is often unavailable or undocumented for marine invertebrate taxa, even those in the accessible intertidal (Knowlton 1993). Yet key features of reproduction and feeding can provide important insight, particularly into the connectivity of natural populations.
We report an example of a cryptic species complex, describing one new species and noting the presence of two additional lineages that remain unresolved. In this case, morphological or molecular differences alone would not have been sufficient to recognise species-level lineages, but new observations of brooding behaviour in one population and the inferred limitation on dispersal in the group shed new light on these animals.

Materials and methods
Leptochiton rugatus sensu lato lives in low intertidal and subtidal depths, typically on the underside of stones partially buried in sand. Previous observations of populations in Washington, USA, and British Columbia, Canada, indicated that local populations of this species in that part of its range have extremely high site fidelity, and small patches (<1 m 2 ) can reliably be re-collected over a span of at least a decade. Such patches appear sparsely distributed in the intertidal, but their true density is not well understood and is likely more abundant in the subtidal. For example, a population at a site in Sooke, British Columbia, adjacent to a specific prominent intertidal boulder, has been home to a population continuously since at least 1986 (JDS, pers. obs. and museum records, last observation 2013). Field observations herein were recorded from specimens collected on 17 July 2012 from a population in False Bay, San Juan Island, Washington, USA (48.57°N 123.17°E).
To examine the potential genetic differences among the Leptochiton aff. "rugatus" species complex, we assembled all available published COI sequences attributed to Leptochiton rugatus sensu lato and species that resolved within the L. rugatus clade in previous analyses (Sigwart et al. 2011;Sigwart 2016), and added sequences from two intermediate populations in Washington and Oregon, USA. These 61 sequences (Table 1) represent specimens from Russia, Alaska, British Columbia, Washington, Oregon, California, and Mexico. We used TCS version 1.21 software (Clement et al. 2000) to reconstruct statistical parsimony haplotype networks from a 362-bp alignment of the COI gene. The connection probability was set to 95%, and sequences differing only by ambiguous characters were treated as the same haplotype.
Museum specimens of Leptochiton rugatus sensu lato from the collections in Naturalis (Leiden, the Netherlands; RMNH) were prepared by dissecting valve elements and radula, soaking in bleach to remove tissue, and rinsing thoroughly in dH 2 O before mounting on carbon adhesive and imaged using a scanning electron microscope (JEOL JSM-6480, Naturalis) at 15 kV.   Etymology "From Cascadia", in reference to the biogeographical region and possible proposed country that constitutes the range of this species.

Distribution
North America: coastal waters of Alaska, British Columbia, Washington, and Oregon.

Description
Holotype ∼7.5 mm long. Shell not carinated, evenly rounded, moderately elevated (elevation ratio 0.36 in valve III of the holotype), valves not beaked. Colour tegmentum white to cream-coloured, some specimens with small patches of dark mineral deposit (Fig. 2). Head valve slightly less than semicircular, wider than tail valve. Intermediate valves rectangular, lateral areas not inflated, anterior margins nearly straight, except margin of valve II convex, posterior margins straight. Tail valve with mucro anterior of centre, postmucronal slope straight or slightly convex.
Tegmentum of central areas and antemucronal area with subtle fine lines of fused granules (approx. 50 × 40 μm, less elongate in lateral areas), about 30 ribs per side in intermediate valves. Ribs becoming slightly more widely spaced toward jugum; on distal pleural areas, ribs continue straight and terminate at lateral area.
Each granule with one megalaesthete (6 μm diameter) surrounded by 4-6 slightly smaller micraesthetes (4 μm), regularly arranged. The posterior pair of micraesthetes may be obscured below the granule viewed from above (Fig. 1e). Valves in the type series and other specimens seem prone to erosion of the distal tegmentum.
Gills seven per side, without interspace, extending posteriorly to the anus.

Remarks
The basic taxonomic features of L. cascadiensis sp. nov., summarised above, are only subtly differentiated from those of L. rugatus s.s. The similarity of members in this species complex was noted by Ferreira (1979), who considered a number of available names to be junior synonyms of L. rugatus. However, minor differences between the two species can be observed: the granulation of valve sculpture in L. cascadiensis sp. nov. is less dense than in L. rugatus (Fig. 1a, d) and the granules less prominent (Fig. 1e), though these are features that may be plastic to individual environmental conditions. In addition, the valves are somewhat thinner (antero-dorsally), and the apophyses are larger, in L. cascadiensis sp. nov. compared to L. rugatus (Fig. 1a, d). Our conclusion that L. cascadiensis sp. nov. represents a valid species and does not correspond to any available names is dependent on these distinctions as part of the total evidence gained from morphology, natural history, and molecular genetic evidence, summarised below.
Leptochiton cascadiensis sp. nov. is most similar to L. rugatus s.s., while examination of the type material of L. alascensis (Thiele, 1909) shows several distinct features (B. Sirenko, pers. comm.). Leptochiton alascensis has carinated valves, while those of L. cascadiensis sp. nov. are rounded. In L. alascensis, the granules on lateral areas of the tegmentum are arranged in a more random manner (compared to clear radial lines in both L. rugatus and L. cascadiensis sp. nov.), the aesthete bulbs have around three pores (vs. groups of five or seven pores in L. rugatus and L. cascadiensis sp. nov.), and dorsal scales are narrower, with 5-8 ribs (vs. 13-16 in L. rugatus and around 12 in L. cascadiensis sp. nov.).

Discussion
Specimens of Leptochiton cascadiensis sp. nov. were found to be apparently harbouring eggs in the pallial cavity (Fig. 2). We are not able to definitively confirm that the eggs observed were from the chitons, but the circumstantial evidence is nonetheless interesting, so we present it here as a very preliminary observation and a hypothesis subject to further testing. The eggs had a smooth hull, which would be typical of the clade; however, they were around 100 μm in diameter, which is very small compared to the usual egg size of at least 200 μm reported for other chitons, including other species of the genus Leptochiton. This, and a lack of definitive developmental or genetic evidence to tie the eggs to the adult, leaves us with doubt about their origin. However, the hypothesis is worthy of consideration.
Leptochiton cascadiensis sp. nov., like almost all chitons, is dioecious. As several but not all of the co-occurring specimens we observed had eggs, we could infer that only female individuals were brooding eggs. The eggs were found only in the pallial cavity, not elsewhere on the body of the chiton, which is the mode in which brooding chitons keep their eggs. The eggs were not clustered in the form of an egg mass from a parasitic copepod (Avdeev and Sirenko 1991 ; Fig. 2). Brooding behaviour is known from 41 chitons, including five confirmed species of Leptochiton (Luizzi and Zelaya 2013;Ituarte and Arellano 2016;Sirenko 2015), and may be present in many additional species. It was not apparent whether the eggs were fertilised, and none of them were observed to progress in development. Indeed, they may have been released prematurely in reaction to handling stress on collection. Or, given the preliminary and circumstantial nature of the observation, it may be coincidence.
Brooding behaviour does present a very plausible explanation for previous observations of site fidelity in Their patchy distribution is speculatively correlated with very low dispersal potential and larval recruitment, staying primarily in the immediate vicinity of the parent population. The species apparently depends on specific habitat conditions: under-boulder conditions that are stable and undisturbed on a decadal scale (JDS, pers. obs.). Leptochiton cascadiensis sp. nov. lives under rocks that are not tumbled in annual storm events, but also not so embedded as to block chiton-sized access. There are large stretches of the high-energy, exposed coast of Oregon which probably completely lack such habitat, and this would represent a major genetic break in the species complex that correlates to observed patterns (Fig. 3).
Our population genetic analysis identified four distinct non-overlapping haplotype groups corresponding to distinct geographical ranges within the total distribution of the species complex (Fig. 3). This expands upon other studies that have used subsets of the same data, here identifying L. rugatus s.s., L. cascadiensis sp. nov., and two additional populations. All of the sequences included are shallow-water specimens, predominantly collected in the intertidal. The haplotype network also provides some additional evidence as to potential life history strategies of these different species.
The haplotype network of the group corresponding to Leptochiton cascadiensis sp. nov., from southern Alaska, British Columbia, Washington, and Oregon (n = 33), contained only four haplotypes. The dominant haplotype was represented in 23 individuals, indicating little genetic variation and haplotype diversity in the populations. The two individuals from Haida Gwaii, British Columbia, shared a haplotype which differed from the dominant haplotype by one step. This homogeneity is not inconsistent with the observation that they are brooders.
On the other hand, the group from Baja California and California (n = 24), corresponding to L. rugatus s.s., contained as many as 15 sampled haplotypes, of which 13 were singletons, and a further seven inferred haplotypes. The dominant haplotype was found in nine individuals. No clustering pattern could be seen in haplotype distribution according to locality (Fig. 3); individuals from the three localities in California as well as those form Baja California were well mixed, indicating  Fig. 3 Haplotype network of the Leptochiton rugatus species complex, reconstructed based on the COI gene. The size of circles and numbers within denote sampled frequency; lack of a number indicates a single specimen. Colours correspond to sampling localities, the approximate positions of which are shown on the map with matching colours (Sigwart et al. 2011: fig. 2). The specimens we have examined in the L. rugatus species complex present morphological differences that might be considered within the realm of plasticity if compared between individuals. Other coastal invertebrate species, including chitons, from the northeast Pacific are known to be panmictic across very large ranges (e.g. Doonan et al. 2012;Marko et al. 2010). The new insight into potential brooding behaviour in L. cascadiensis sp. nov. and the inferred limitation on its dispersal potential underscores the probable division of species lineages, upheld by strong genetic separation. Total evidence of dispersal limitation, morphology, and molecular data indicates that these populations are separate species.
Despite speculative limitations on dispersal and local retention, we see little evidence of isolation or drift at small scales within the range of L. cascadiensis sp. nov.; however, the sample density presently available is not sufficient for investigating fine structure patterns. The obvious differences in genetic structure between L. cascadiensis sp. nov. and L. rugatus may yet indicate differential life history strategies, and a much lower dispersal potential for local populations of L. cascadiensis sp. nov., despite its overall broad range from Oregon to Alaska. The local-scale isolation and patchiness of L. cascadiensis sp. nov. raises concern that any further intensive destructive sampling could inadvertently obliterate a local population. Occasional events (e.g. storms) that transport individuals between adjacent populations could be sufficient to maintain genetic mixing, even if these occur at multi-generational scales (Bryan et al. 2012). But if a population is removed, this could create a significant barrier to longer-term long-distance dispersion.
These chiton species are relatively common, broadly distributed members of the low intertidal to subtidal communities in the northeast Pacific coast. Leptochiton cascadiensis sp. nov. is relatively difficult to find, only because of their extreme site fidelity and patchy distribution, now understood to be probably related to their unusual life history. This case study presents another example where a study of familiar intertidal species reveals a cryptic species complex, and a conservation imperative to recognise separate regional species. and Dr. Ken Takai (JAMSTEC) for facilities that supported analysis and writing.

Compliance with ethical standards
Funding This research was supported by the European Commission award H2020-MSCA-IF-2014-655661 to JDS, and additional data obtained via the SYNTHESYS access to infrastructure programme for JDS' travel to RMNH Naturalis and to RBINS.