The slug within the bivalve: molecular and morphological systematics of the family Juliidae (Gastropoda: Panpulmonata: Sacoglossa)

INTRODUCTION

The family Juliidae E. A. Smith, 1885 (Mollusca, Gastropoda, Panpulmonata: Sacoglossa) is morphologically unique among gastropods as the only group with a bivalved shell (Kawaguti & Baba 1959;Keen & A. G. Smith 1961;Kay 1968;Jensen 1996). Juliids were originally described from shell material, and have been historically and consistently classified as members of the Bivalvia (Gould 1862;Deshayes 1863;E. A. Smith 1885;Cossmann 1887;Beets 1944;Habe 1952), except for a few authors who hesitantly assigned single valves from Juliidae to various heterobranch sea slug groups (Carpenter 1857;Crosse 1875;Hedley 1912). However, the discovery of the first living animals by Kawaguti & Baba (1959) made evident their taxonomic placement in the Sacoglossa (Cox & Rees 1960;Burn 1960a), resulting in the re-assignment of previously described taxa (Cox & Rees 1960;Keen 1960b;Kay 1962b;Boettger 1962;Le Renard 1980). Placement of Juliidae within the shelled sacoglossans was confirmed by later phylogenetic studies (Jensen 1996;Händeler & Wägele 2007;Krug et al. 2015).

Depending on the classification scheme (see Keen & A. G. Smith 1961;Boettger 1962;Kay 1968;Le Renard 1980, 1989;Le Renard et al. 1996;Burn 1998Burn , 2006;;Schneider et al. 2008), Juliidae includes 3-8 valid genera represented in the fossil record (Julia Gould, 1862;Berthelinia Crosse, 1875;Anomalomya Cossmann, 1887;Ludovicia Cossmann, 1887;Hemiplicatula Deshayes, 1861; Saintia de Raincourt, 1877;Namnetia Cossmann 1905;Gougerotia Le Renard, 1980;and Candinia Le Renard, Sabelli & Taviani, 1996), as well as 2-5 genera containing extant species (Julia; Berthelinia; Tamanovalva Kawaguti & Baba, 1959;Edenttellina Gatliff & Gabriel, 1911;and Midorigai Burn, 1960). Conflicting classifications resulted primarily from the incongruent use of traits in taxonomic decisions as well as inconsistencies between the neontological and paleontological literature, which underscore the need for taxonomic revision in Juliidae.

Early descriptions of Recent and fossil species relied exclusively on shell morphology (Gould 1862;Cossmann 1887;Crosse 1887;Gatliff & Gabriel 1911;Hedley 1920); characteristics such as the color, thickness, size, and overall shape of the shell and hinge were considered important to characterize and delineate species (Beets 1944(Beets , 1949;;Howard 1951;Habe 1952;Keen & A. G. Smith, 1961;Boettger 1962). More recently, after the discovery of living Juliidae, the emphasis has shifted toward the of traits of the reproductive and digestive systems for taxonomic classification, including the radulae (Kawaguti & Baba 1959;Edmunds 1963;Kawaguti 1981;Jensen 1996Jensen , 1997aJensen , 1997bJensen , 2015)). Descriptions of fossil taxa can only be based on shells because internal anatomical details are rarely preserved, but paleontologists have focused on conchological features such as the shell hinge, the adductor muscle scars, and the protoconch, in addition to overall shell shape (Boettger 1962;Bałuk & Jakubowski 1968;Le Renard 1980, 1989;Le Renard et al. 1996;Schneider et al. 2008).

These different emphases have created two parallel taxonomies within Juliidae, one based exclusively on shell morphology (for fossils and early descriptions of Recent species) and another based on both shell and internal morphology (for contemporary descriptions of Recent species). To make matters worse, the shells in species of Juliidae are difficult to compare as they do not possess many discrete distinguishing characteristics at the species level, apart perhaps from protoconch morphology and the shape and location of the adductor muscle scar (Kawaguti & Baba, 1959;Baba 1961b;Jensen 1997a). At the genus level, shell differences are more apparent; for example, the Recent genera Julia and Berthelinia are easily distinguishable by the overall shape and thickness of the shell. However, continuous traits are notoriously difficult to quantify, and subjective interpretations of those traits have resulted in authors considering Berthelinia as a single genus (Jensen 2011(Jensen , 2015) ) or splitting this group into four genera (Burn 1998(Burn , 2006)).

While molecular approaches can address outstanding taxonomic issues in Recent Juliidae, two of the Recent genera are also represented in the fossil record. An integrative approach to resolve the systematics of this group must therefore evaluate the taxonomy of extinct as well as Recent taxa in a quantitative framework. In this paper, we provide a revised classification of extinct and extant Juliidae at the genus level using a combination of morphological and molecular data.

Molecular analyses were conducted using two mitochondrial genes (CO1, 16S) and three nuclear genes (H3, 18S, 28S). Because of the lack of discrete and distinctive traits in the shells of Juliidae, we use geometric morphometrics to quantify differences among shells, along with more traditional characters from radular morphology and other soft body traits. In the course of this study, several species-level Recent taxa were also identified. However, the basis for identifying the recognized species of the genus Julia, and descriptions of new species, will be presented in a separate monographic work.

DNA extraction, amplification, and sequencing

All tissue samples were obtained from specimens in a sterile environment using flamed forceps and razor blades. Specimens were placed under a dissecting microscope and the shell gently pried open; 0.5 mg of tissue was removed from the mantle or foot, depending on the size of the specimen. DNA extractions were initially performed using a hot Chelex® (Bio-Rad, Hercules, CA, USA) protocol as follows. Tissue samples were macerated and placed in 1 ml of TE buffer (Tris:EDTA). Tubes were capped and placed onto a rotator block for ≥20 min, mixed by vortexing for 15 sec, and centrifuged for 3 min at 21130.2 × g; 975 µL buffer was removed without disturbing the pellet and 175 µL of 10% Chelex ® added. Tubes were then heated on a water bath at 56 ºC for ≥20 min. Tube contents were mixed by vortexing for 15 sec, heated dry for 8 min at 100 ºC, mixed by vortexing for 5 sec and centrifuged for 3 min at 21130.2 × g. If DNA yield was low or if specimens were older than 10 years, a Qiagen DNeasy ® (Qiagen, Germantown, MD, USA) kit was used for DNA extraction following the manufacturer's protocols.

The Polymerase Chain Reaction (PCR) was performed to amplify the two mitochondrial (CO1, 16S) and three nuclear (H3, 28S, 18S) gene regions using universal primers (Table 2). The PCR master mix volume was 50 µL per reaction, containing: 33.75 µL of H2O, 0.25 µL of DreamTaq, 5 µL of 10× DreamTaq Buffer, 5 µL of 2 mM dNTPs, 3 µL of 25 nM MgCl2, 1.5 µL of each forward and reverse primers, and 2 µL of DNA template. Reaction conditions for CO1 were: initial 95 ºC denaturation for 3 min, 35 cycles of 94 ºC for 45 sec (denaturation), 45 ºC for 45 sec (annealing), 72 ºC for 45 sec (elongation), followed by a final elongation step at 72 ºC for 10 min. Reaction conditions for 16S and H3 were: initial 94 ºC denaturation for 2 min, 35 cycles of 94 ºC for 30 sec (denaturation), 50 ºC for 30 sec (annealing), 72 ºC for 1 minute (elongation), followed by a final elongation step at 72 ºC for 7 min. Reaction conditions for 28S were: initial 95 ºC denaturation for 2 min, 38 cycles of 94 ºC for 30 sec (denaturation), 52.5 ºC for 30 sec (annealing), 72 ºC for 90 sec (elongation), followed by a final elongation step of 72 ºC for 10 min. Reaction conditions for 18S were: initial 95 ºC denaturation for 2 min, 40 cycles of 94 ºC for 30 sec (denaturation step), 45 ºC for 30 sec (annealing step), 72 ºC for 1 minute (elongation step), followed by a final elongation step at 72 ºC for 7 min. PCR products were purified using Genejet purification kits and sent to Source Bioscience (Santa Fe Springs, CA, USA) for Sanger sequencing. Sequences were assembled and edited in Geneious Pro R8 (Kearse et al. 2012).

Species delimitation analysis

Automatic Barcoding Gap Discovery (ABGD) species delimitation analysis was performed on CO1 and 16S haplotypes after collapsing identical sequences. Genetic distances among individuals were calculated in MEGA 11 (Tamura et al. 2021) using the model (COI) or K2P + G (0.5) model (16S). Each distance matrix was then analyzed using the ABGD webtool (Puillandre et al. 2012) with default settings except for a reduced relative gap width (X = 1.2). Results from both initial and recursive partitions were compared with support for species hypotheses from morphological data. The 16S analysis was run after changing the default pmax value to 0.5. Analyses performed on each genus separately and on the combined dataset yielded the same outcomes.

A second method of species delimitation analyzed the same distance matrices by Assemble Species by Automatic Partitioning or ASAP (Puillandre et al. 2021), as implemented by the webtool (https://bioinfo.mnhn.fr/abi/public/asap/asapweb.html). The number of species supported by the highest ASAP score and the associated barcode gap was compared with the number supported by the barcode gap closest to that recovered by ABGD.

A third method used the Bayesian implementation of the Poisson Tree Processes (bPTP) algorithm (Zhang et al. 2013). An unrooted COI guide tree was generated in RAxML using separate GTR + Γ models for the first two versus third codon positions. The tree was then uploaded to the bPTP webtool (https://species.h-its.org/ptp/) and MCMC simulations run for 10 5 generations, discarding 25% as burn-in. The acceptance rate was 0.13. Outputs of both ML and BI partitions were examined to assess support for all species (see Results). Groups supported by one or more methods of delimitation were considered candidate species and their status further evaluated using morphological data.

Morphological data

At least two preserved specimens of each candidate species of Edenttellina recovered in the ABGD analysis were examined morphologically. The two valves were separated exposing the soft parts of the animal, which were photographed using a Leica EZ4D microscope with a builtin digital camera. The anterior portion of the digestive system and the reproductive organs were examined and photographed. The penis of each specimen was dissected and photographed using a Nikon Eclipse E400 microscope with an attached Nikon D70 digital camera. In a few cases, penises were submerged in hexamethyldisilazane for 24 hours until all the liquid evaporated; the resulting hardened structure was mounted for SEM imaging.

To isolate the radula, the buccal mass of 2-3 specimens were placed in 10% sodium hydroxide (NaOH) for ≥48 hr. Radulae were then rinsed in deionized water, placed on an SEM stub, and coated with either an Emitech K550X Sputter Coater (NHMLA) or a Denton Vacuum Desk V HP Sputter Coater (California State Polytechnic University, Pomona). Samples were imaged with a Hitachi S-3000N variable pressure SEM (NHMLA) or a Jeol JSM-6010 variable pressure SEM (California State Polytechnic University, Pomona).

The external surface of the uncoated shell of at least three specimens of each candidate species was photographed using a Leica EZ4D microscope with a built-in digital camera. At least one of the shells was also examined by SEM, including the outer side of the left valve (to illustrate the sculpture), the inner side of the right valve (to illustrate the adductor muscles), the protoconch, and the hinges of both valves.

Species delimitation

There was no clear barcoding gap in the distribution of pairwise distances for either COI or 16S (Fig. 2). The ABGD analysis for COI distances recovered 10 species entities with a barcoding gap of 13.5% for the initial partition, which corresponded to the morphologically distinctive species described herein (see Systematics) with three exceptions: E. babai was lumped with E. singaporensis; E. chloris was lumped with E. pseudochloris; and J. mishimaensis was split into two species. The same result was obtained from the recursive partition with a prior maximal distance of 4%, except E. chloris and E. pseudochloris were recovered as distinct (Table 4, Fig. 2). ABGD results for 16S were the same for the initial partition and recursive partitions using a prior maximal distance of roughly 1%. A barcoding gap of 3.7% was supported, under which described and candidate morphospecies were recovered as separate entities with the following exceptions: E. caribbea and Julia sp. 1 were both split into two cryptic species; E. chloris was lumped with E. pseudochloris; and J. exquisita was lumped with Julia sp. 3 (Table 4).

ASAP analysis of the COI data yielded a significantly favored partition with 16 species under a threshold distance of ~7%. All traditional and candidate species were separated, while E. caribbea, J. mishimaensis and J. zebra were split into two entities, and Julia sp. 3 was split into three entities (Table 4). Only a partition with a threshold distance of 12% recovered the 10 species supported by the more conservative ABGD analysis. For 16S, the most favored ASAP partition had a threshold distance of 1.65% and supported 21 species: E. caribbea, J. burni and Julia sp. 1 were each split into two cryptic entities, while J. mishimaensis, J. exquisita, and J. zebra were split into three entities apiece. Results comparable to ABGD were only obtained with a less favored partition that recovered a 4.25% threshold distance.

The bPTP analysis of the COI guide tree supported a mean of 18 species (range: 14 to 21 OTUs).

As above, the majority of morphospecies were split into two (E. caribbea, E. pseudochloris, J. mishimaensis, J. exquisita, J. zebra) or three (Julia sp. 3) cryptic species entities. No species were lumped by bPTP (Table 4).

Geometric morphometrics

In the Principal Component Analysis (PCA) of all examined Juliidae shell material (Table 3), PC1 explained the majority of the variance (85%) while PC2 explained 8% of the variance. On the score plot, five clusters of taxa were broadly separated into distinct quadrats (Fig. 3). All Julia spp. had positive scores for PC2, and fell into three clusters separated by PC1 scores.

Cluster 1 comprised one fossil, the type specimen of Prasina borbonica Deshayes, 1863, and one specimen of an extant species identified as J. japonica (but no molecular data were available to confirm its phylogenetic placement). Cluster 1 was differentiated from other Julia and Candinia specimens by large, negative PC1 scores. Cluster 2 included specimens of the fossil genus Candinia and two Recent Julia: J. thecaphora and one specimen identified as J. exquisita.

Within cluster 2, more negative PC1 scores separated fossil specimens of C. krachi from extant Julia spp. Cluster 3 (positive PC1 scores) included the remaining extant Julia spp.: J. mishimaensis, J. zebra, and specimens provisionally identified as J. exquisita, Two other clusters were separated from Julia and Candinia by negative PC2 scores, with the exception of one of three specimens of Berthelinia elegans with a slightly positive PC2 score but the highest PC1 score (Fig. 3). Cluster 4 included all Recent specimens that were originally classified as Midorigai, Edenttellina, Tamanovalva, and Berthelinia; all 1854(accepted as Gaimardia Gould, 1852), and the bivalved sacoglossans Prasina Deshayes, 1863and Julia Gould, 1862. Subsequently, E. A. Smith (1885) proposed the family name Juliidae for Julia. Because Prasina is a synonym of Julia (see remarks on Julia), the family names Juliidae and Prasinidae are also synonyms. Beets (1949) introduced the family name Berthelinidae (for Berthelinia Crosse, 1875) for the first time but without a diagnosis, making it invalid (ICZN 1999: Article 13.1). Thus, the first valid introduction of the name Berthelinidae was by Keen & A. G. Smith (1961), who provided a description. Kawaguti & Baba (1959) introduced the family Tamanovalvidae for Tamanovalva Kawaguti & Baba, 1959. Most authors have supported a classification scheme for this group within a single family, Juliidae, and two Recent subfamilies, Juliinae E.A. Smith, 1885and Berthelininae Keen & A. G. Smith, 1961(Kay 1968) with the later addition of the fossil subfamily Gougerotiinae Le Renard 1980(Le Renard et al. 1996). However, as Valdés & Bouchet (2005) pointed out, Prasininae has priority over Juliinae and Tamanovalvinae over Bertheliniinae. However, Valdés & Bouchet (2005) continued using the names Juliinae and Bertheliniinae for taxonomic stability, until a formal proposal is submitted to the International Commission on Zoological Nomenclature. Therefore, Juliinae and Bertheliniinae are used herein.

Members of the family Juliidae were generally classified in the class Bivalvia until the discovery of the first live bivalved sacoglossans by Kawaguti & Baba (1959). Thus, morphological descriptions prior to 1959 often assumed that the narrower end of the shell corresponded to the anterior part of the body as in bivalve convention (see Coan et al. 2000). However, as shown by Kawaguti & Baba (1959) and subsequent papers cited herein, the head of juliids emerges from the wider end of the shell, and therefore the convention in modern Juliidae literature is to refer to that as the anterior end of the shell. This convention is used in the present paper.

The results of the five-gene phylogenetic analyses recovered two Recent clades in Juliidae, one corresponds to the traditional use of the name Julia and the other with the traditional use of the name Berthelinia. Additionally, geometric morphometrics analyses indicate that these two clades are morphologically distinct. Therefore, the traditional classification with two subfamilies is here maintained. Below we provide systematic descriptions for all genera and species, and for completeness we also list fossil groups not examined in this study. Generic diagnoses are based on Keen & A. G. Smith (1961), Le Renard (1980Renard ( , 1989)), Le Renard et al. (1996) and Jensen (2011). SUBFAMILY JULIINAE E. A. SMITH, 1885 GENUS JULIA GOULD, 1862 Julia Gould 1862: 283. Type species: Julia exquisita Gould, 1862, by monotypy. Prasina Deshayes 1863: 25-29. Type species: Prasina borbonica Deshayes, 1863, by monotypy. Parasina Habe 1951: 122 [error for Prasina Deshayes 1863]. Diagnosis Shell, thick, heavy, opaque, typically colored; oval in lateral view, cordate in anterior/posterior views; anterior margin rounded, narrowing posteriorly, deeply excavated at umbo, with sharp posterior peripheral margin; shell tallest point near mid-length; valves equal in size; hinge with pronounced condyloid (or cardinal) tooth on left valve, right valve with fossette-like hinge socket; protoconch on left valve, towards posterior of shell; adductor muscle scar subcentral;

shell typically green with white, brown, or black spots or lines; gill continues across mid-dorsal line to left side; heart on left side near the posterior end of gill; short segment of intestine visible dorsally on surface of digestive gland; mantle cavity extends across mid-dorsal line of shell; radular teeth elongate, blade-shaped, inflated, with row of hair-like denticles along each side; teeth cusps with bifid tips.

Remarks

Carpenter ex. Nuttall (1857) introduced the species name Smaragdinella thecaphora Carpenter (ex. Nuttall), 1857 from Mazatlán, Mexico, and it constitutes the first description of a species now in the genus Julia. Carpenter (1857) considered this new species a member of the Gastropoda, primarily because he had access to only one valve. However, when describing the apical morphology of the shell, he noted that it looked remarkably like the ligamental cavity of Anatinella Sowerby G. B. II., 1833 (Bivalvia). Gould (1862) described the genus Julia based on the single species Julia exquisita Gould, 1862, the type species by monotypy. Although Gould (1862) unequivocally included Julia in Bivalvia, he was unsure of the taxonomic placement within this group and discussed its affinities with other bivalve genera including Vulsella Röding, 1798 [Family Pteriidae Gray, 1847 (1820)],

Pedum Bruguière, 1792 (Family Pectinidae Rafinesque, 1815), or even members of the family Veneridae Rafinesque, 1815. However, Gould (1862) also recognized the affinities between Julia and heterobranch sea slugs, by indicating, "A beautiful little shell, colored like Smaragdinella A. Adams, 1848 from the same islands. The sudden and profound depression in front of the beaks is quite striking, and thus produces a very pointed somewhat upturned anterior end. A single valve looks much like some of the Bullidae, Chelidonura, for instance." Gould's (1862) short description of Julia in Latin summarizes the main diagnostic traits, which are consistent with the modern usage of the name. Deshayes (1863) introduced the genus name Prasina for the species Prasina borbonica Deshayes, 1863, the type species by monotypy, collected in Réunion, Indian Ocean. Deshayes (1863) examined a series of single valves collected by Louis Maillard, and because of their green color, brilliant internal surface, and shell projection, initially concluded they belonged to the genus Smaragdinella. However, by assembling the shells together, Deshayes (1863) realized they were bivalve shells, regular and symmetrical. Although Deshayes (1863) found the external characteristics of these shells very peculiar, he found other traits in the interior of the valves "of much greater importance," such as the unique shape of the hinge, the insertion of the lunula, and the position of the muscle scars. Thus, Deshayes (1863) introduced the new genus name Prasina for these shells, but as in the case of Gould (1862) and Julia, Deshayes (1863) was unsure of the taxonomic placement of Prasina. Deshayes (1863) proposed that Prasina had intermediate characteristics between the monomyarians (bivalves with a single posterior adductor muscle) and the dimyarians (bivalves with two adductor muscles, one anterior one posterior), but proposed it was closer to the family Mytilidae Rafinesque, 1815, of which it could constitute a subfamily.

Prophetically, Deshayes (1863) suggested this species could perhaps be, "destined to play a more important role, for it may be that the animal offers in its organization facts of greater zoological value than the shell," and hoped that this small shell would be collected and observed alive. Two additional species of Juliidae were described in the genus Prasina, Prasina cornuta De Folin in De Folin & Périer, 1868 (Recent, Mauritius) and Prasina lecointreae Dollfus & Dautzenberg 1901 (Miocene, Touraine, France). Semper (1865) was the first to note the similarities between Julia and Prasina, indicating that it was nearly impossible to separate these two genera. E. A. Smith (1885) examined shells of Julia exquisita from the Hawaiian Islands and introduced the family name Juliidae for it, although without a taxonomic placement beyond confirming its membership in Bivalvia. One year later, Fischer (1886) formally considered Julia as a subgenus of Prasina and suggested that Berthelinia Crosse, 1875 could also belong to the Prasinidae or Aviculidae Goldfuss, 1820

[=Pteriidae]. Fischer (1886) also discussed the similarities between Prasinidae, which he considered to be bivalves, and Mytilidae. E. A. Smith (1885) placed P. borbonica in the synonymy of J. exquisita without further explanation, and Dall (1898) formally synonymized these two genera. Some authors maintained Prasina as a synonym of Julia but considered J. borbonica as a distinct species and pointed out that specimens examined by E. A. Smith (1885) from the Hawaiian Islands belonged to J. borbonica (Beets 1944(Beets , 1949;;Howard 1951;Boettger 1962). Other authors agreed to maintain Prasina as a synonym of Julia (e.g., Kay 1968;Kawaguti & Yamasu 1982). Kay (1968) went further and indicated that all Julia shells she examined from the Indian and Pacific Oceans fell within the range of variation of the Hawaiian specimens and suggested that J. japonica and J. equatorialis may represent geographic subspecies. Kawaguti & Baba (1959) described the first live specimens of Juliidae, which they taxonomically placed in the Sacoglossa (Gastropoda) (see remarks on Edenttellina). However, these authors did not immediately make a connection between their newly discovered species and the extensive Recent and fossil literature on this group (Cox & Rees 1960;Keen 1960b). Keen (1960b) suggested that when specimens of Julia were found alive, they would prove to be gastropods. Just two years later, Kay (1962b) reported the first known live animals within the genus Julia, confirming they unequivocally belonged to Sacoglossa.

Additional Recent species of Julia have been described from Japan: Julia japonica Kuroda & Habe, 1951 [also reported from Pleistocene deposits (Mimoto 1984)], Julia zebra Kawaguti, 1981 Julia mishimaensis Kawaguti & Yamasu, 1982, and an unnamed species from Okinawa (Mizofuchi & Yamasu 1987); India: Julia burni Sarma, 1975;and [=Serravallian] in the original description, but subsequently re-assigned to the Late Oligocene by Schneider et al. 2008), Julia gardnerae Woodring, 1925 (Miocene, Jamaica), and Julia borneensis Boettger, 1962 (Late Miocene, East Borneo).

Based on the molecular analyses presented here, Julia is clearly distinct from other Juliidae genera described to date, and therefore is considered valid. However, the geometric morphometrics analysis recovered substantial morphological diversity that will be examined in more detail in a separate monograph. Several species regarded as Julia herein cluster in different groups (Clusters 1-3, Fig. 3) and it is particularly noteworthy that J. japonica and J. borbonica (Cluster 1, Fig. 3) are substantially distant from the rest of Julia. In this study we regard J. borbonica as a synonym of J. exquisita (see remarks on J. exquisita), which is also highly variable morphologically and represented in two different groups of the PCA. Unfortunately, we did not have access to material of J. japonica suitable for molecular analysis to determine its phylogenetic placement. Additionally, representatives of the fossil genus Candinia are clustered with other members of Julia in the PCA, casting doubt on the validity of the former genus.

Further work is necessary to resolve these outstanding issues. Currently, Julia includes Recent and fossil species, which are listed below in chronological order.

Julia equatorialis

† Julia gardnerae Woodring 1925: 87-88, pl. 10, figs. 15-16. Type locality: Bowden, Jamaica (Middle Miocene).

Julia japonica Kuroda & Habe 1951: 86. Type locality: Nada, Wakayama Prefecture, Japan.

† Julia borneensis Boettger 1962: 422-423, fig. 3. Type locality: Gunung Mendong, East Borneo (Late Miocene).

† Julia borbonica erythraea Selli 1973: 201-203, pl. 30, figs. 8A-C. Type locality: Massawa, Erytraea (Pleistocene). Julia burni Sarma 1975: 12-15, figs. 1-5. Type locality: Port Blair, Andaman Islands. Julia zebra Kawaguti 1981: 9-13, figs. 1A, 2-4. Type locality: Mishima, Hagi City and Tsunoshima, Yamaguchi Prefecture, Ishigaki Is., Okinawa Prefecture, Japan. Julia mishimaensis Kawaguti & Yamasu 1982: 29-31, figs. 1-4. Type locality: Tsunoshima, Yamaguchi Prefecture and Ishigaki Is., Okinawa Prefecture, Japan. In this study, three additional, probably undescribed species are recognized. As mentioned above the taxonomy of species of Julia and the formal description of these new species will be addressed in a separate monograph. GENUS CANDINIA LE RENARD, SABELLI & TAVIANI, 1996 † Candinia Le Renard, Sabelli & Taviani 1996: 231-232. Type species: Candinia pliocaenica Le Renard, Sabelli & Taviani, 1996, by original designation. Diagnosis Shell thick, heavy, oval-trigonal in lateral view, cordate in anterior/posterior views; anterior margin rounded, narrowing abruptly posteriorly into sharp triangular margin; shell tallest point near anterior end; right valve larger than left valve; hinge with anterior, posterior parts interrupted beneath apex, condyloid (or cardinal) tooth on right valve, left valve with fossettelike socket; protoconch on left valve, towards posterior third of shell; adductor muscle scar subcentral.

† Berthelinia krachi

In this study, morphological evidence indicates that Berthelinia cannot be used for Recent species. The PCA analysis recovered Recent and fossil species of Berthelliniinae in two distinct clusters, indicating they are morphologically distinct. Therefore, the genus Berthelinia is used herein only for the extinct taxa listed below. Renard (1980) described the new Juliidae species Anomalomya (Namnetia?) sphaerica Le Renard, 1980. Le Renard (1980), based on this interpretation of the original description of Namnetia discoides by Cossmann (1905b), suggested that Namnetia could be a subgenus of Anomalomya and belong to the family Juliidae. This placement has not been confirmed with the examination of actual specimens. Le Renard (1989) examined additional specimens of this group and using morphometric data confirmed the placement of Namnetia in Juliidae. Le Renard (1989) also considered Anomalomya sphaerica Le Renard, 1980 as a possible synonym of Anisodonta sphaericula Cossmann, 1886, and proposed both as members of Namnetia. Finally, Le Renard (1989) pointed out the similarities between Namnetia and Anomalomya but preferred to maintain them as separate genera. Gatliff & Gabriel (1911) described the genus Edenttellina based on Recent specimens collected in Port Phillip, Australia. Gatliff & Gabriel (1911) designated their new species Edenttellina typica Gatliff & Gabriel, 1911 as the type, but mentioned that a second species collected by C.

Species list

Hedley also belonged to the same genus and was going to be described at a later date. Gatliff & Gabriel (1911) placed Edenttellina among other bivalve species but did not discuss the taxonomic placement of this group or compare it with other previously described taxa in Bivalvia. Hedley (1912) suggested that Edenttellina was not a bivalve (or pelecypod) but the internal shell of a tectibranch gastropod (obsolete term for some shelled sea slugs). Hedley Gould 1862, Hedley (1920) proposed the placement of Edenttellina in the family Juliidae and related to the Chamacea [= Chamoidea Lamarck, 1809, Bivalvia]. Dall (1918) reported a third Recent species of this group based on two left shells collected in Magdalena Bay, Baja California Sur, Mexico that he described as ?Scintilla chloris. Because of the limited material available (only two valves with no protoconch), Dall (1918) did not make the connection between his specimens and previous records by Dautzenberg (1895) and Gatliff & Gabriel (1911), thus he suggested his specimens should probably be placed in the bivalve genus Scintilla Deshayes, 1856 (Family Galeommatidae Gray, 1840) until more material became available. Kawaguti & Baba (1959) discovered the first live specimens of Sacoglossa with a bivalved shell, for which they proposed the new genus Tamanovalva and the new species T. limax. Kawaguti & Baba (1959) were unaware of any other species of sacoglossans with such an unusual morphology and therefore introduced the new family Tamanovalvidae and the new suborder Tamanovalvacea for this group. Kawaguti & Baba (1959) compared the shell of Tamanovalva with that of Edenttellina, which they considered a member of Bivalvia, and concluded they were different. However, Cox & Rees (1960) stressed the morphological similarities between Tamanovalva and Edenttellina as well as to the fossil genus Ludovicia, and suggested they could be synonyms. Keen (1960a) agreed with Cox & Rees (1960) but proposed that the fossil genus name Berthelinia could be a more senior synonym for this group. Burn (1960a), in a brief note, described for the first time live specimens of Edenttellina typica in Torquay, Victoria, Australia. Burn (1960a) also mentioned collecting a second species of the same group belonging to a different genus. In a second paper the same year, Burn (1960b) reemphasized that Tamanovalva is a synonym of Edenttellina, but also proposed that these two genera are synonyms with the fossil genus Berthelinia from the Eocene of the Paris Basin, as suggested by Keen (1960a). Burn (1960b) confirmed that the species he collected from Victoria belonged to E. typica (renamed Berthelinia typica) and that E. corallensis was a synonym, but that the specimens described by Verco (1916) as E. typica constituted a second, distinct species. Burn (1960b) suggested that T. limax described by Kawaguti & Baba (1959) was also a synonym of B. typica, and ?Scintilla chloris Dall, 1918 (from Baja California), although larger, was probably another synonym of B. typica. Finally Burn (1960b) considered the second species he collected in Victoria to belong to a new genus of that he named Midorigai, based on the new species Midorigai australis Burn, 1960. According to Burn (1960b), the main differences between Midorigai and Berthelinia were the more swollen shape of the shell, the size and arrangement of the protoconch (in a horizontal plane across the hinge line), the presence of two adductor muscles, the foot narrower than the neck, and some other anatomical details.

Julia exquisita

In subsequent years, a number of papers described additional species of bivalved gastropods and provided alternative assessments of the taxonomy of this group. First, Keen (1960b) formally synonymized Tamanovalva with Edenttellina and Berthelinia and suggested a close relationship with Julia, which was not yet known from live animals. This opinion was followed by Ludbrook & Steel (1961) in the description of the new species Berthelinia burni Ludbrook & Steel, 1961 [not Julia burni Sarma, 1975]. Baba (1961a) provided a comprehensive revision of the morphological characteristics of Tamanovalva and compared them with those of Edenttellina and other shelled sacoglossans. Baba (1961a) also mentioned that Tamanovalva was different from Edenttellina in shell and radula morphology, and that Edenttellina seems more similar to Ludovicia than to Tamanovalva, which appears closer to Berthelinia. Baba (1961a) concluded that it is difficult to compare Recent and fossil taxa and more data are needed to establish a definitive taxonomy for the group, but should all these genera become synonyms, Berthelinia would have priority. In a second paper the same year, Baba (1961b) concluded that Berthelinia, Ludovicia, Edenttellina, Tamanovalva, Midorigai, and possibly Anomalomya were synonyms and provided a detailed review of the radular and conchological characteristics of this group. The same year, Keen & A. G. Smith (1961) described the new subspecies Berthelinia chloris belvederica and revisited the taxonomy of the bivalved sacoglossans formally proposing that both Julia and Berthelinia were members of the family Juliidae. Keen & A. G. Smith (1961) Boettger (1962) followed a similar classification scheme as proposed by Keen & A. G. Smith (1961), but regarded Midorigai as a distinct genus from Berthelinia, instead of a subgenus, and proposed the new subgenus name Cossmannella to replace Ludovicia, which he erroneously thought it was preoccupied (see remarks of Berthelinia). Boettger (1962) also described the new fossil species Julia borneensis.

Additional records and new species from the Hawaiian Islands (Kay 1962a(Kay , 1962b(Kay , 1964) -Berthelinia pseudochloris, and the Caribbean (Edmunds 1962(Edmunds , 1963) -Berthelinia caribbea, largely followed the classification scheme proposed by Keen & A. G. Smith (1961). However, Burn (1965) indicated that he considered Berthelinia an exclusively fossil genus and that the Recent genus Tamanovalva was distinct and included the species T. limax, E. corallensis and S. chloris. Burn (1965) also recognized that among the paratypes he described as M. australis (Burn 1960b), there were some specimens that belonged to the true E. typica as described by Gatliff & Gabriel (1911), a point already made by Baba (1961a). Burn (1965) concluded that the specimens he described in Burn (1960b) as E. typica were different from those in the original description by Gatliff & Gabriel (1911) and belonged to the genus Tamanovalva, thus he introduced the new name Tamanovalva babai for them. Finally, although not clearly stated, Burn (1965) appeared to also consider Edenttellina as a valid genus and distinct from both Tamanovalva and Berthelinia. Burn (1966) described a new species of Tamanovalva from Fiji, T. fijiensis, and clearly stated that he considered Edenttellina and Tamanovalva as distinct genera but did not provide distinguishing characters. Burn (1966)

also distinguished Berthelinia and

Tamanovalva because of the presence of one and a half whorls in the protoconch of Tamanovalva instead of two, as in the protoconch of Berthelinia. Kay (1968) proposed a classification, informed by that of Keen & A. G. Smith (1961) and Boettger (1962), in which Midorigai was a subgenus of Berthelinia (because it possessed a uniquely swollen shell and two adductor muscle scars) and the subgenus Berthelinia could only be applied to fossils. This classification scheme was followed by some authors (e.g. Ganapati & Sarma 1972;Sarma 1975), but Jensen (1993Jensen ( , 1997aJensen ( , 1997bJensen ( , 2015) ) synonymized all the subgenera under Berthelinia. Burn (1998Burn ( , 2006) ) accepted Berthelinia and Tamanovalva as synonyms, but retained Edenttellina and Midorigai as separate genera and continued to use Tamanovalva. Importantly, conchological traits often used to distinguish these genera are variable. For example, Edmunds (1963) found that the position of the protoconch varies within B. caribbea and Jensen (1993) noted that the overall shape of the shell and the angle of the protoconch varies within B. rottnesti, with small specimens having a fairly erect protoconch, and larger specimens having an almost horizontal one. Thus, it seems due to ontogenetic and/or intraspecific variability, these characters are not taxonomically informative.

In this study, molecular phylogenetic analyses revealed that the type species of Edenttellina, Tamanovalva, and Midorigai belong in the same clade. Therefore, we find no compelling reasons to maintain these as different genera and we agree with Jensen's (1993) proposal to synonymize Edenttellina, Tamanovalva, and Midorigai. However, geometric morphometric analyses revealed that fossil and Recent members of this group clustered in different groups, with significant morphological differences. Thus, we propose to maintain Berthelinia as a valid genus for fossil taxa and retain Edenttellina (the oldest available name) for Recent species. Edenttellina typica Gatliff & Gabriel, 1911: 190, pl. 46 Berthelinia pseudochloris Kay 1964: 191-193, fig. 1, pl. 9, figs. 1, 4. Type locality: Near Koloa Landing, Koloa, Kaua'i, Hawaiian Islands. Dautzenberg (1895) described the shell as uniformly white, oval, slightly trapezoid, with a rounded anterior end shorter than the posterior end. The holotype is untraceable and the limited description of the animal does not allow for a reliable identification of this species. Ganapati & Sarma (1972) reported juvenile shell specimens from the Andaman Islands as B.

schlumbergeri. Ganapati & Sarma (1972) also indicated that these shells were similar to the Madagascar specimen described by Dautzenberg (1895) but recognized that their species could not be determined until more material of fully grown individuals became available. In a later paper, Sarma (1975) reported two new species of Berthelinia from mainland India but made no reference to B. schlumbergeri. Gosliner (1987) and Ono (1999) reported and illustrated photographs of specimens identified as B. schlumbergeri from Sodwana Bay, South Africa and Okinawa, Japan, respectively. Both photographs show green animals with some white spots on the head and neck, white rhinophoral tips, a green mantle, with the edge of the shell surrounded by alternating white and dark bands. Gosliner (1987) confirmed that his specimens from South Africa fed on Caulerpa racemosa (Forsskål) J. Agardh, 1873. The morphological characteristics of B. schlumbergeri are consistent with those of E. pseudochloris here examined, a species that also feeds on C. racemosa, and has planktotrophic development. Therefore, we regard the records of B. schlumbergeri by Gosliner (1987) and Ono (1999) to be E. pseudochloris.

Although it is possible that B. schlumbergeri is an older name for E. pseudochloris this is impossible to determine with certainty based on the original description.

Because the original description of B. schlumbergeri is based on juvenile specimens and lacks details of the internal anatomy, we are unable to confirm the validity of this species. Therefore, B. schlumbergeri is here regarded as a nomen dubium until more information becomes available.

Edenttellina typica Gatliff & Gabriel 1911: 190, pl. 46, figs. 5-6. Type locality: Portsea, Port Phillip, Victoria, Australia.

Description

Body elongate, up to 6.1 mm in length, completely retractable inside of shell. Body color pale green, with minute white speckles on dorsal side of head (Figs. 5A-B). Head elongate, with eye spots located on dorsal swelling near center, surrounded by white pigment. Rhinophores enrolled, green, with or without few white speckles. Oral tentacles short, green. Foot lighter than rest of animal. Mantle visible through shell, pale green, with few longitudinal light brown irregular lines, variable in size; edge surrounded by conspicuous, alternating cream patches composed of densely arranged speckles. Foot not extending beyond posterior end of shell.

Shell up to 4.5 × 3.4 mm in size, tallest point slightly anterior to center, widest point near center; shell shape ovoid, dorsal and ventral margins regularly curved; anterior margin convex, regularly curved, posterior margin narrowing gradually, also regularly curved (Figs. 6C-D). Protoconch on left valve of teleoconch, ~110 µm long, with 1.5 whorls (Fig. 6D). Hinge on dorsal margin of shell, formed by flattened, corrugated, nearly straight area, margin on both valves; small, oval condyloid tooth on right valve at posterior end of hinge, and triangular, fossette-like hinge socket on left valve, at posterior end of hinge (Figs. 6E-F). Shell translucent, with no visible markings or spots on shell surface, soft parts of body visible through it (Figs. 6A-B).

Adductor muscle located closer to anterior end of shell, slightly below widest point of shell (Fig. 7), connected to narrow and elongate head retractor muscle. Adductor scar visible on shell (Figs.

6D). Gill large, occupying almost height of body, posterior to adductor muscle, covering anterior portion of digestive gland. Penis not observed.

Radula with 19 teeth in descending limb and 5 fully formed teeth + 1 ghost tooth in ascending limb, in 2.5 mm long specimen from Victoria, Australia (MV F21539) (Fig. 8A). Active tooth ~70 μm long, with bifid tip; blade elongate, lacking denticles; base short, slightly curved (Fig. 8B). Ascus containing several disorganized elongate pre-radular teeth.

Remarks

Gatliff & Gabriel (1911) described Edenttellina typica Gatliff & Gabriel, 1911 based on a right and left valve of similar size, but not from the same specimen, collected in Victoria, Australia. Gatliff & Gabriel (1911) described the shells as translucent yellow, small, thin and fragile, fattened, broadly ovate, with the posterior side [anterior end] longer and broader, and hinge short, without teeth. Gatliff & Gabriel (1911) also reproduced photographs of the two valves from inner views, with the right valve having a protoconch attached. They also mentioned additional larger valves that were not included in the original description because of their poorer preservation, as well as specimens of a second species from North Queensland, to be described by C. Hedley (Gatliff & Gabriel 1911).

In 1911, Verco (1911) reported E. typica from South Australia that he described in more detail in 1916 (Verco 1916). In the interim, Hedley (1912) commented on the original description of E. typica suggesting it may represent the internal shell of a sea slug, but if this animal were to be a bivalve, as suggested by Gatliff & Gabriel (1911), Edenttellina could be a synonym of Ludovicia. Verco (1916) described several right and left valves and one complete shell collected in Guichen Bay, South Australia and provided details on its protoconch and teleoconch morphology. Hedley (1920) described the specimens mentioned by Gatliff & Gabriel (1911) from North Queensland as the new species Edenttellina corallensis Hedley, 1920 based on shells collected in Hope Island, and for comparison, illustrated the specimens of E. typica reported by Verco (1911Verco ( , 1916)). Hedley (1920) also provided an additional record for E. typica from King George Sound, Western Australia. Cotton & Godfrey (1938) described the presence of an anterior tooth in the hinge of E. typica and suggested the name Edenttellina was inappropriate.

Subsequently, Burn (1960a) reported collecting specimens of E. typica alive in Torquay, Victoria, Australia, which he confirmed to be sacoglossans, probably congeneric with the recently described species from Japan, Tamanovalva limax Kawaguti & Baba, 1959. Additionally, Burn (1960a) reported a second species of bivalved sacoglossan belonging to a different genus from the same area. In a second paper the same year, Burn (1960b) proposed the synonymy between the genera Tamanovalva and Edenttellina as well as with the fossil genus Berthelinia. Burn (1960b) also suggested that T. limax and E. typica are synonyms of ?Scintilla chloris Dall, 1918, originally described from Baja California, Mexico, but not with the specimens identified as E. typica by Verco (1911Verco ( , 1916) ) and illustrated in Hedley (1920).

However, Burn (1960b) treated Berthelinia typica as a valid name (despite the fact older names should have priority) and did not use the generic names consistently with his proposed synonymy. Moreover, Burn (1960b) listed several differences between B. typica and T. limax, although he re-emphasized that T. limax and B. typica are synonyms, and also synonyms of E. corallensis and ?S. chloris, arguing this latter species was larger but remarkably similar. Burn (1960b) provided a re-description of the shell and radula of B. typica based on the specimens collected in Torquay, and described the second species reported in Burn (1960a) as Midorigai australis Burn, 1960. Baba (1961a) examined additional specimens from Japan as well as a paratype of M. australis sent by Burn. Baba (1961a) noted that this paratype belonged to a species different from M. australis as it had smooth radular teeth. Baba (1961a) listed some morphological differences between the shell and radula of Edenttellina and Tamanovalva but indicated these two genera could be synonyms of Berthelinia along with Ludovicia and Midorigai; a synonymy that was later confirmed by Baba (1961b). Burn (1965) recognized that the paratype of M. australis sent to Baba for examination was a juvenile mistakenly placed in the paratype series of M. australis, and he now identified this specimen as belonging to the true E. typica. Burn (1965) also recognized the specimens he previously identified as E. typica belonged to an undescribed species that he named Tamanovalva babai Burn, 1965. Revising his previous assessment, Burn (1965) recognized E. corallensis and ?S. chloris as distinct species belonging to the genus Tamanovalva. According to Burn (1965) all three species of bivalved sacoglossans found in Victoria, Australia were easily distinguishable as follows: 1) E. typica has a uniformly green shell with horizontal parallel black lines on the mantle and smooth and bifid radular teeth (lacking denticles); 2) T. babai has a completely green shell and denticulate radular teeth; and 3) M. australis has a green shell with numerous yellow patches, a mantle with yellow to creamcolored large rounded spots, and denticulate radular teeth. Burn (1989Burn ( , 2015) ) depicted all three species alive. Burn (1965) did not provide further clarification on the specimens reported by Verco (1911Verco ( , 1916) ) or illustrated in Hedley (1920), but in 1966 (Burn 1966) reported locating and re-examining those specimens. Based on their shell morphology, Burn (1966) suggested they could constitute a third species with intermediate characteristics between T. babai and E. typica.

In this study, we sequenced one specimen with the characteristic horizontal parallel black lines on the mantle described for this species, which was genetically distinct from all other species of Juliidae here examined. We also examined the syntypes of Edenttellina typica (Figs. 4A-B) and several additional specimens from temperate Australia, which were morphologically similar.

Therefore, we propose that E. typica is a distinct and valid species. Additionally, phylogenetic analyses place E. typica in the same clade as all other species previously assigned to Berthelinia.

Based on geometric morphometrics analyses (see above) Berthelinia, which was described based on fossil shells, is distinct from Recent species. Therefore, Recent species, including E. typica are here transferred to Edenttellina, which is the oldest available name for this clade. Keen & A. G. Smith 1961: 53-61, figs. 18-19, 21-24, 27-32, pl. 5, lower 1971-53.15). Punta Espinosa, Isla Fernandina, Galapagos Islands, Ecuador, 25 Jan 1972, 1 shell 9 mm long, leg. R. M. Lease (Searcher stn. 333) (NHMLA 1972-197.8); Nov 1992, 1 specimen 4 mm long (shell), leg. T Gosliner (CASIZ 087112). Bahía Urbina, Isla Isabela, Galapagos Islands, Ecuador, 1-3 m depth, 24 Jan 1972, 2 valves 9 mm long, leg. S. Earle (NHMLA 1972-192.1).

Description

Body elongate, up to 10.5 mm in length, completely retractable inside of shell. Body color vibrant dark green, with minute white speckles throughout (Figs. 5C-D). Head elongate, with eye spots located on dorsal swelling near center, surrounded by white pigment. Rhinophores enrolled, green, with small white speckles often forming a ring at the apex, cluster of white speckles on dorsal side. Oral tentacles short, green with white speckles on long edge. Foot lighter than rest of animal, with small spots along edge. Mantle visible through shell, vibrant dark green, covered with few whitish to pale green spots, variable in size; edge surrounded by conspicuous alternating opaque white patches composed of densely arranged speckles. Foot extending to posterior end of shell, forming small triangular projection.

Shell up to 9 × 5.6 mm in size, tallest point near center, widest point near ventral margin; shell shape ovoid, dorsal margin regularly curved, ventral margin more flattened; anterior margin convex, irregularly curved, slightly more flattened dorsally, posterior margin narrowing gradually, forming a nearly triangular, elongate shape, with round end (Figs. 9G-H). Protoconch on left valve of teleoconch, ~110 µm long, with 1.5 whorls (Fig. 9K). Hinge on dorsal margin of shell, formed by flattened, corrugated, nearly straight area, margin on both valves; small, oval condyloid tooth on right valve at posterior end of hinge, and triangular, fossette-like hinge socket on left valve, at posterior end of hinge (Figs. 9I-J). Shell translucent, with no visible markings or spots on shell surface, soft parts of body visible through it (Figs. 9A-F).

Adductor muscle located closer to anterior end of shell, slightly below widest point of shell (Fig. 10A), connected to narrow and elongate head retractor muscle. Adductor scar visible on shell (Figs. 9H). Gill large, occupying almost height of body, posterior to adductor muscle, covering anterior portion of digestive gland. Anterior half of body in preserved specimens with pair of elongate pharyngeal appendages visible between head retractor muscle and gill (Fig 10B ), connecting to the pharyngeal bulb dorsally (Fig. 10C). Penis elongate; distal end pointed, with no stylet visible; proximal end wider, with two strong retractor muscles attached and long, tubular deferent duct (Fig. 10D).

Radula with 29 teeth in descending limb and 9 fully formed teeth + 1 ghost tooth in ascending limb, in 9 mm long specimen from Baja California Sur, Mexico (NHMLA 1974-33.1) (Fig. 11A), and 25 teeth in descending limb and 7 fully formed teeth + 2 ghost teeth in ascending limb, in 8 mm long specimen from Sonora, Mexico (NHMLA 186690) (Fig. 11C). Active tooth ~130 μm long, with sharp, pointed tip; blade elongate with central row of numerous long, delicate denticles; base elongate, slightly curved (Fig. 11B, D). Ascus containing a rod-shaped preradular tooth plus about 5 intermediate teeth (Fig. 11C).

Range

Eastern Pacific Ocean: Baja California (Dall 1918;Keen & A. G. Smith 1961;present paper), Costa Rica (Camacho-García et al. 2005), and Galapagos Islands (Sphon & Mulliner 1972; present paper). Dall (1918) introduced the name ?Scintilla chloris Dall 1918 based on two shells collected in Magdalena Bay, Baja California Sur, Mexico. He described the shells as pale green, translucent, with paler rays, the posterior end [anterior] wider, longer, and more rounded than the anterior end [posterior], which was shorter with a hinge, a single noduliform denticle, and a faint lateral lamella (Dall 1918). Dall (1918) interpreted the two shells as left valves and suggested they could belong to the bivalve genus Scintilla Deshayes, 1856. Keen & A. G. Smith (1961) examined manuscript illustrations of the type material of this species as well as additional specimens from the Pacific coast of the Baja California Peninsula and concluded that the two shells examined by Dall (1918) were in fact opposite valves. This has been corroborated by recent examination of the type material (USNM 218179) (Figs. 4C-D). Keen & A. G. Smith (1961) placed this species in Berthelinia and described the subspecies B. chloris belvederica based on additional specimens from the Gulf of California coast of Baja California Sur, which were morphologically different from the type material of B. chloris chloris. Specifically, Keen & A. G. Smith (1961) noted that the syntype of B. chloris chloris they examined was larger, more inflated, and had a more prominent lamellar tooth in the hinge than their specimens of B. chloris belvederica. Boettger (1962) considered the differences between B. chloris belvederica and B. chloris chloris to fall within the normal range of variation of a species and therefore regarded them as synonyms. This opinion is generally accepted, and more recent records refer to B. chloris belvederica as B. chloris (Sphon & Mulliner 1972;Camacho-García et al. 2005). We obtained the holotype of B. chloris belvederica (Figs. 4E-F) and did not find any distinguishing differences between it and B. chloris chloris here examined. Therefore we consider both to be B.

Description

Live animals not examined, illustrated by Abe (1964: pl. 6, fig. 20) and Hamatani (2000: pl. 379;2017: pl. 405) as having an elongate body, up to 10 mm in length. Body color pale green, lacking white speckles. Head elongate, with eye spots located on dorsal swelling near center.

Rhinophores enrolled, green, with few white speckles. Oral tentacles short, green. Foot lighter than rest of animal. Mantle visible through shell, dark green, covered with large, whitish-beige patches, variable in size; edge surrounded by alternating opaque white patches composed of densely arranged speckles; adductor muscle visible through shell as white patch. Foot forming small triangular projection, not extending beyond posterior end of shell.

Shell up to 6.5 × 4.4 mm in size, tallest point near anterior margin, widest point near ventral margin; shell shape ovoid to quadrangular, dorsal margin regularly curved with distinct apex, ventral margin more flattened; anterior margin convex, irregularly curved, slightly more flattened dorsally, posterior margin shorter, narrowing gradually (Figs. 12G-H). Protoconch on left valve of teleoconch, ~150 µm long, with 1.5 whorls (Fig. 12K). Hinge on dorsal margin of shell, formed by flattened, corrugated, nearly straight area, margin on both valves; low, oval condyloid tooth on right valve, triangular, fossette-like hinge socket on left valve, at posterior end of hinge (Figs. 12I-J). Shell translucent, with no visible markings or spots on shell surface, soft parts of body visible through it (Figs. 12A-F).

Adductor muscle located in line with highest point of shell, slightly below widest point (Fig. 13A), connected to narrow and elongate head retractor muscle. Adductor scar visible on shell (Figs. 12H). Gill large, occupying almost height of body, posterior to adductor muscle, covering anterior portion of digestive gland. Penis short, triangular; distal end pointed, with conical stylet (Fig. 13C); proximal end of penis wider, with two strong retractor muscles attached and tubular deferent duct (Fig. 13B). Pharyngeal bulb as wide as tall (Fig 13D ).

Radula with 28 teeth in descending limb and 6 fully formed teeth + 1 ghost tooth in ascending limb, in 4.5 mm long specimen from Tamano Bay, Japan (CASIZ 201957) (Fig. 14A). Active tooth ~135 μm long with sharp, pointed tip; blade elongate with central row of numerous long, delicate denticles; base short, curved (Fig. 14B). Ascus containing one long, rod-shaped preradular tooth plus 5 intermediate teeth (Fig. 14C).

Remarks

Kawaguti & Baba (1959) described for the first time live animals of Juliidae under the new name Tamanovalva limax Kawaguti & Baba, 1959. For the original description, Kawaguti & Baba (1959) had access to more than 200 specimens collected near the former Tamano Marine Laboratory, Okayama University, Japan. Unfortunately, they did not designate a holotype and we have been unable to locate the type series. Kawaguti & Baba's (1959) description included accounts of the characteristics of the live animals as well as features of the internal anatomy and the shell, all illustrated in great detail. Kawaguti & Baba (1959) described the shell as translucent yellowish white, thin, elongate-oval, narrowing posteriorly, with the umbo behind the midlength, hinge with no teeth. The live animals' color was described as deep green with opaque white dots on the rhinophores and head, mantle margin sparsely spotted with white and dark brown, and the mantle's deep green color showing through the shell. Other characteristics included having stout, grooved rhinophores, small foot corners, a raised area where the eyes are located, and a short posterior end of the foot, not extending beyond the shell. The radula was formed of 35 blade-like teeth finely denticulate on both edges.

Prabhakara Rao (1965) reported four specimens of bivalved gastropods from India found on Caulerpa racemosa. The animals were described as leaf green with opaque white spots with a few dark green spots on the digestive gland. The illustration of the specimens (Prabhakara Rao 1965: fig. 1) and the host alga, are more consistent with those of E. pseudochloris and therefore are here assigned to this species. Subsequently, Ganapati & Sarma (1972) reported recently metamorphosed juvenile specimens also found on C. racemosa in India and assigned them to T. limax. Again, based on the host alga, this record is here considered to be of E. pseudochloris.

We examined several specimens from Japan that are morphologically consistent with the characteristics of T. limax, some of which were collected and identified by S. Kawaguti. These specimens have a conical stylet in the penis, which is absent in other specimens of Juliidae here examined. Unfortunately, we were unable to obtain sequence data from those specimens, thus we cannot confirm that they are genetically distinct. However, these Japanese specimens are morphologically distinct from other species here recognized and therefore we regard T. limax as a valid species, probably endemic to Japan. Finally, because T. limax is morphologically coherent with other recent species here considered to be members of Edenttellina, it is here transferred to this genus.

EDENTTELLINA AUSTRALIS (BURN, 1960) (Figs. 4J-K, 5E-F , 15-17) Midorigai australis Burn 1960b: 46, figs. 8-14. Type locality: Torquay, Victoria, Australia.

Description

Body elongate, up to 5.6 mm in length, completely retractable inside of shell. Body color pale to dark green, with numerous minute white speckles throughout (Figs. 5E-F). Head elongate, with eye spots located on dorsal swelling near center, surrounded by white pigment. Rhinophores enrolled, green, with small white speckles forming two rings. Oral tentacles short, green with white speckles at base. Foot lighter than rest of animal, with small spots along edge. Mantle visible through shell, dark green, covered with large, whitish-beige patches, variable in size; edge surrounded by alternating opaque white patches composed of densely arranged speckles.

Foot not extending to posterior end of shell, forming small triangular projection.

Shell up to 4 × 2.6 mm in size, tallest point near anterior margin, widest point near ventral margin; shell shape ovoid to quadrangular, dorsal margin regularly curved with distinct apex, ventral margin more flattened; anterior margin convex, irregularly curved, slightly more flattened dorsally, posterior margin shorter, narrowing gradually (Figs. 15E-F). Protoconch on left valve of teleoconch, not observed. Hinge on dorsal margin of shell, formed by flattened, corrugated, nearly straight area, margin on both valves; no distinct condyloid tooth on right valve, but elongate protuberance at posterior end of hinge, triangular, fossette-like hinge socket on left valve, at posterior end of hinge (Figs. 15G-H). Shell translucent, with no visible markings or spots on shell surface, soft parts of body visible through it (Figs. 15A-D).

Adductor muscle located in line with highest point of shell, slightly below widest point (Fig. 16A), connected to narrow and elongate head retractor muscle. Adductor scar visible on shell (Figs. 15F). Gill large, occupying almost height of body, posterior to adductor muscle, covering anterior portion of digestive gland. Anterior half of body in preserved specimens with penis visible above adductor muscle (Fig. 16B). Penis elongate; distal end pointed, with very elongated, hollow stylet (Fig. 16D); proximal end of penis wider, with two strong retractor muscles attached and long, tubular deferent duct (Fig. 16C). A single, reduced pharyngeal appendage observed, connected to the dorsal side of the pharyngeal bulb (Fig 16E).

Radula with 35 teeth in descending limb and 6 fully formed teeth + 1 ghost tooth in ascending limb, in 4.1 mm long specimen from Victoria, Australia (MV F194029) (Fig. 17A). Active tooth ~260 μm long with sharp, pointed tip; blade elongate with central row of numerous long, delicate denticles; base short, slightly curved; juvenile teeth bicuspid (Fig. 17B). Ascus containing one long, rod-shaped pre-radular tooth and 3 (possibly 4) intermediate teeth (Fig. 17C).

Remarks

In addition, Burn (1960b) introduced the new name Midorigai australis Burn, 1960 for the second species mentioned in the previous paper. Burn (1960b) described the live animals of M. australis as dark green, with large yellowish patches on the rhinophores and the rest of the body, including the mantle, which is densely spotted with large, rounded, yellowish-cream patches and has square-shaped yellow patches along the edges. Burn (1960b) described the shell as translucent olive-green with darker rays of green, squarer than in Berthelinia, with the left valve more convex and higher than the right valve, hinge with no teeth.

For this paper, we have studied several specimens collected in Philip Island, Victoria, Australian (~75 km east of Torquay), characterized by having a green body color with numerous, large, rounded yellowish green to opaque white spots, larger on the mantle. We obtained nuclear DNA sequence data for two of them. We also examined the holotype of M. australis (Figs. 4J-K) but it is too damaged to draw any conclusions. These animals match the original description of M. australis and are here regarded as members of this species. Both the five-gene sequence data and the geometric morphometrics analyses confirm these animals belong in the genus Edenttellina, but are genetically distinct from all other species. Thus, we regard M. australis as a distinct and valid species of Edenttellina.

EDENTTELLINA CARIBBEA (EDMUNDS, 1963) (Figs. 5G-H Type material Berthelinia caribbea, Holotype, complete specimen, not examined (NHMUK 1962261), six paratypes (NHMUK 1962262, NHMUK 1962263).

Description

Body elongate, up to 6.5 mm in length, completely retractable inside of shell. Body color bright green, with numerous minute white speckles throughout, but more densely arranged on the dorsal side of head (Figs. 5G-H). Head elongate, with eye spots located on dorsal swelling near center, surrounded by white pigment; two parallel brown lines run between base of rhinophores and eye spots, then merge into single brown line running backwards to pericardium; brown spots scattered over rest of head in some specimens. Rhinophores enrolled, green, with high concentration of small white and brown speckles. Oral tentacles short, green with white speckles at base. Foot lighter than rest of animal. Mantle visible through shell, dark green, covered with large, irregular whitish-beige patches, variable in size and numerous transverse white and brown lines; edge surrounded by alternating opaque white patches composed of densely arranged speckles and dark brown patches. Foot not extending to posterior end of shell, forming small triangular projection.

Shell up to 4.8 × 3.5 mm in size, tallest point near anterior margin, widest point near ventral margin; shell shape ovoid, dorsal margin regularly curved with distinct apex, ventral margin more flattened; anterior margin convex, irregularly curved, slightly more flattened dorsally, posterior margin shorter, narrowing gradually (Figs. 18G-H). Protoconch on left valve of teleoconch, ~110 µm long, with 1.5 whorls (Fig. 18K). Hinge on dorsal margin of shell, formed by flattened, corrugated, nearly straight area, margin on both valves; large, rounded condyloid tooth at posterior end of hinge on right valve, triangular, fossette-like hinge socket on left valve, at posterior end of hinge (Figs. 18I-J). Shell translucent, with no visible markings or spots on shell surface, soft parts of body visible through it (Figs. 18A-D).

Adductor muscle in line with highest and widest points of shell (Fig. 19A), connected to narrow and elongate head retractor muscle. Adductor scar visible on shell (Figs. 18H). Gill large, occupying almost height of body, posterior to adductor muscle, covering anterior portion of digestive gland. Anterior half of body in preserved specimens with pair of elongate pharyngeal appendages visible above adductor muscle (Fig 19B), connecting to the pharyngeal bulb posteriorly (Fig. 19C). Penis elongate; distal end pointed, with no stylet visible; proximal end wider, with strong retractor muscle and long, tubular deferent duct (Fig. 19D).

Radula with 25 teeth in descending limb and 11 fully formed teeth + 1 ghost tooth in ascending limb, in 3 mm long specimen from Abaco, Bahamas (NHMLA 011629) (Fig. 20A). Active tooth ~110 μm long (Fig. 20B), with sharp, harpoon-tip shaped, pointed tip; blade elongate with central row of numerous short, delicate denticles (Fig. 20D); base short, curved. Ascus containing 3 disorganized elongate pre-radular teeth (Fig. 20C).

Remarks

For this paper, we examined and sequenced several specimens from the Caribbean region and Brazil matching the original description of B. caribbea. Geometric morphometrics analyses confirmed the shells of these specimens are morphologically similar to those of other species of Edenttellina. Additionally, molecular data confirm B. caribbea is sister to the rest of Edenttellina. Based on those two lines of evidence we transfer B. caribbea to Edenttellina and confirm that E. caribbea is a valid species name.

The ABGD species delimitation analyses based on the 16S gene split E. caribbea into two different groups, but this split was not recovered in the COI and H3 analyses. Moreover, there are no obvious morphological differences between these two groups, thus they are here maintained in the same species. The two groups recovered have different geographic ranges, one including Bahamian specimens and the other Caribbean-proper specimens, suggesting a certain degree of genetic isolation between the two populations. Further research including a larger sample size may yet support cryptic diversity in E. caribbea. EDENTTELLINA PSEUDOCHLORIS (KAY, 1964) (Figs. 4G-H, 4L, 5I-L, 21-23)

Berthelinia pseudochloris Kay 1964: 191-193, fig. 1, pl. 9, figs. 1, 4. Type locality: Near Koloa Landing, Koloa, Kaua'i, Hawaiian Islands. Burn 1966: 54-55, pls. 15-19. Type locality: Nukulau Island, Viti Levu, Fiji. Sarma 1975: 16-20, figs. 6-13, 28-29. Type locality: Visakhapatnam, India.

Type material

Berthelinia pseudochloris, Holotype, complete specimen, not examined (BM 8903). Paratype, 1 complete specimen, 3.7 mm long (shell) (CASIZ 018374).

Tamanovalva fijiensis, Holotype, complete specimen, shell dissolved, 1.5 mm long (animal) (MV F25647).

Berthelinia ganapatii, Holotype and paratypes at the Department of Zoology, Andhra University, Waltair, India, not examined. pointed tip; blade elongate with central row of numerous long, delicate denticles; base short, curved (Fig. 23B, D). Ascus spirally coiled, with one long, rod-shaped pre-radular tooth, plus several (>5) intermediate teeth (Fig. 23C). Kay (1964) reported Berthelinia pseudochloris occurs on dense masses of Caulerpa racemosa var. turbinata [= Caulerpa chemnitzia] with little intrusion of other algal species, in areas with constant and frequently heavy surf. Burn (1966) reported collecting B. fijiensis on Caulerpa racemosa var. turbinata [= Caulerpa chemnitzia]. Sarma (1975) found Berthelinia ganapatii fairly commonly on the Visakhapatnam foreshore during the period December to May among Caulerpa racemosa. Sarma (1975) described the egg masses and development of B. ganapatii; the number of eggs in egg masses fluctuated from 500-3520, with eggs 50 µm in diameter and capsules 90 µm, indicating this species is probably planktotrophic (see Krug et al. 2015). Both C. racemosa and C. chemnitzia are closely related species and members of the C. racemosa-peltata complex (see Belton et al. 2014).

Range

Hawaiian Islands (Kay 1964; present paper), Japan (Ono 1999), Fiji (Burn 1966), Philippines (present paper), Papua New Guinea (present paper), Vanuatu (present paper), New Caledonia (present paper), India (Prabhakara Rao 1965;Ganapati & Sarma 1972;Sarma 1975), Madagascar (Legendre 1965), Mozambique (present paper), Réunion (Joannot & Vendel 2011), South Africa (Gosliner 1987), and possibly Easter Island (Rehder 1980). Kay (1962a) reported a single specimen of a live bivalved sacoglossan collected in the Hawaiian Islands. Kay (1962a) argued that this animal was morphologically different from other species described to date and probably constituted a distinct species. Kay (1964) formally described the species as Berthelinia pseudochloris Kay 1964, based on several specimens collected Near Koloa Landing, Kaua'i, Hawaiian Islands. Kay (1964) described the shell of this species as small, thin, translucent green, ovate-triangular, rounded anteriorly, narrower posteriorly, with the white protoconch on the left valve, and the hinge with a strong tooth, and frequently a secondary weaker tooth posteriorly. The live animals, subsequently illustrated by Kay (1979), were described as leaf-green, with minute opaque milk-white spots on the oral tentacles, rhinophores and foot; rhinophores with one or two larger white blotches, head and neck are microscopically speckled with brown, mantle with red bands alternating with white patches, visible through the shell ventrally (Kay 1964). We examined several specimens from the Hawaiian Islands matching the characteristics enumerated in the original description of B. pseudochloris as well as the shell morphology of the paratype (Figs. 4G-H); these animals are anatomically and genetically distinct from other species here examined and constitute a distinct species. Therefore, we here regard B. pseudochloris as a valid species. Additionally, geometric morphometrics and molecular analyses of specimens here examined from the Hawaiian Islands confirm this species is distinct and a member of the Recent genus Edenttellina. Burn (1966) introduced Tamanovalva fijiensis Burn, 1966 based on a single specimen collected from Viti Levu, Fiji. The live holotype was described as pale green with cream speckling on the slender, auriculate, and abruptly truncate rhinophores (Burn 1966). The characteristics of the pale-yellow shell included a steeply angled and rather straight anterior margin, a broadly rounded posterior margin, and a small, white protoconch, slightly inclined to the left, anterior to the second third of the shell length. According to Burn (1966) the radula had 5 teeth in the ascending and 25-30 in the descending limbs; radular teeth rather stout, strongly rounded above the simple tip and laterally finely denticulate. Burn (1966) placed this species in Tamanovalva because of having one and a half whorls, not two, in the protoconch. Burn (1966) recognized that the holotype of T. fijiensis had a similar shell outline to that of E. pseudochloris (see Kay1964: fig. 4; Burn 1966: figs. 15-16). Also, both feed on Caulerpa chemnitzia, have a white protoconch with 1.5 whorls, and short and wide radular teeth with numerous denticles (Fig. 23; Burn 1966: figs. 18). Given these similarities, we propose that T. fijiensis is a synonym of E. pseudochloris.

Remarks

A photograph of the holotype of T. fijiensis included here (Fig. 4L) confirms the morphological similarities with E. pseudochloris. Sarma (1975) described Berthelinia ganapatii Sarma, 1975 based on live specimens collected in Visakhapatnam, India. The color of the rhinophores, neck and foot were described as uniformly yellow, with the tips of the rhinophores speckled with white spots (Sarma 1975); the mantle was deep yellowish-green, with two bright shining yellow patches, all visible through the transparent shell. Sarma (1975) described the shell as deep green but transparent [sic.] with yellow rays on the surface, fragile, ovate trigonal in outline, with the anterior margin rounded and abrupt, and the posterior margin longer. The small, erect protoconch was situated at the third quarter of the shell length and the hinge had a weak tooth in the left valve and a strong tooth posteriorly on the right valve. Sarma (1975) described the radula as having blade-like teeth bearing fine denticulations on the sides and having simple tips, with 7 teeth in the ascending row and 35 in the descending row. Sarma (1975) recognized that B. ganapatii was close to E. pseudochloris but distinguishable by having a more abrupt anterior margin, a longer posterior margin and an erect protoconch. Similarities between these two species are remarkable: they both feed exclusively on species of the C. racemosa-peltata complex; they have a hinge with teeth, and fairly similar external coloration. The radula of B. ganapatii contains short and broad teeth with numerous denticles (Sarma 1975: fig. 13), very similar to the teeth of E. pseudochloris here illustrated (Fig. 23). Finally, both species have planktotrophic development. Because of the similarities between the original descriptions of B. ganapatii and B. pseudochloris, we regard these two names as synonyms.

Records of Berthelinia spp. from several localities in the Indo-Pacific region, including Madagascar (Legendre 1965), India (Prabhakara Rao 1965;Ganapati & Sarma 1972), South Africa (Gosliner 1987), and tropical Japan (Ono 1999) were based on animals very similar to the specimens of E. pseudochloris here examined and are here regarded as members of this species.

A record of E. pseudochloris from Easter Island by Rehder (1980) could not be verified with certainty. translucent, with no visible markings or spots on shell surface, soft parts of body visible through it (Figs. 25A-F).

Adductor muscle slightly posterior to highest point of shell, in line with widest point (Fig. 26A), connected to narrow and elongate head retractor muscle. Adductor scar visible on shell (Figs.

25H). Gill large, occupying almost height of body, posterior to adductor muscle, covering anterior portion of digestive gland. Anterior half of body in preserved specimens with pair of elongate pharyngeal appendages visible below adductor muscle (Fig 26B), connecting to the pharyngeal bulb posteriorly (Fig. 26C). Penis elongate; distal end pointed, with a short stylet (Fig. 26E); proximal end wider, with strong retractor muscle and long, tubular deferent duct (Fig. 26D).

Radula with 23 teeth in descending limb and 5 fully formed teeth + 1 ghost tooth in ascending limb, in 3 mm long specimen from Victoria, Australia (MV F112386) (Fig. 27A) and 22 teeth in descending limb and 5 fully formed teeth + 1 ghost tooth in ascending limb, in 4.5 mm long specimen from Queensland, Australia (lot 94-5E) (Fig. 27D). Active tooth ~100-150 μm long (Fig. 27B, E), with sharp, bifid or pointed tip; blade elongate with central row of numerous elongate, delicate denticles; base short, curved. Ascus containing one long, rod-shaped preradular tooth, 4-5 intermediate teeth (Fig. 27C, F). Burn (1989)

Remarks

We have been unable to obtain specimens matching the original description of B. waltairensis, which remains as an uncertain species. However, because of its morphological similarities to other species of Edenttellina, it is here transferred to this genus. brownish, with mostly radiating, lines, including a row of alternating white and brown spots along the mantle edge, not quite reaching the umbo (Jensen 1993). Jensen (1993) noted that the shell and the position of the protoconch were variable, and described the radula as formed of blade-shaped teeth with fine lateral denticles on both sides of the blade, including 4-6 teeth (plus 1 incompletely formed ghost tooth) in the ascending limb, and 22-23 teeth in the descending limb. Based on the observation of a newly metamorphosed specimen, Jensen (1993) suggested that B. rottnesti appeared to have direct development. Jensen (1993) indicated that B. rottnesti is anatomically similar to B. babai; both species have bifid denticulate teeth and a similar penis; but according to Jensen (1993) the denticles of B. rottnesti are shorter than those of B. babai; additionally, the morphology of the pharynx is different between these two species. Jensen (1993) also compared B. rottnesti with B. limax and suggested that they could represent ecotypes of one species. According to Jensen (1993), the small differences between B. rottnesti and B. limax could be explained by reduced gene flow due to the direct development mode in the two species.

According to Burn (2006) B. rottnesti, is the western cognate of, if not identical with, Midorigai australis as both species have an obligate association with the green alga Caulerpa simpliciuscula. However, the external shell coloration of these two species is very different.

Midorigai australis is completely covered with pale round spots and B. rottnesti is nearly uniformly green. Wells & Bryce (1993) illustrated a live animal possibly belonging to B. rottnesti from Western Australia that was green with white spots on the head and neck, more densely covering the rhinophores and the mantle edged by a line of white dots.

Based on the description by Jensen (1993) B. rottnesti appears to be different from other species here examined. Unfortunately, we had no access to specimens from Western Australia and therefore could not confirm the validity of this species with molecular data. Therefore, we maintain B. rottnesti as a valid species until more material becomes available. Although we did not have molecular data for B. rottnesti, this species is morphologically similar to other species within Edenttellina here examined. Therefore, B. rottnesti is provisionally transferred to this genus.

In this study, we examined specimens from Tasmania with an external morphology and anatomy very similar to those in the original description of B. rottnesti. For example, the radular teeth and the penis here examined are very similar to the description of those organs by Jensen (1993). Jensen (1993: fig. 5A) described and illustrated the radular teeth of B. rottnesti as having bifid radular teeth with a short row of denticles near the apical end, very similar to the teeth here illustrated (Fig. 55C), no other species of Edenttellina here examined has similar characteristics. Jensen (1993;fig. 6A) described the penis of B. rottnesti as elongate and lacking a stylet, which is also similar to the material here examined; all other species from southern Australia studied to date have a penial stylet. Additionally, the pharyngeal bulb of the material here examined, with a dorsal pharyngeal appendage is very similar to the descriptions and illustrations by Jensen (1993: fig. 4). Because of the morphological similarities between the Tasmania specimens and the original description of B. rottnesti we tentatively assign our specimens to this species. However, due to the geographic distance between Tasmania and Rottnest Island (the type locality of B. rottnesti) and the lack of genetic data for B/E. rottnesti, we cannot dismiss that the Tasmania specimens could represent an undescribed species.

EDENTTELLINA DARWINI (JENSEN, 1997) Berthelinia darwini Jensen 1997a: 170-175, figs. 6-9. Type locality: Lee Point, Darwin, Australia.

Remarks

Jensen (1997a) introduced the new species Berthelinia darwini Jensen, 1997 based on several specimens collected in Darwin, Australia. Jensen (1997a) described the shell as nearly elliptical, rounded at both ends and not particularly narrow posteriorly; protoconch about 200 µm long, located just anterior to the posterior third of the shell length, with a variable angle of insertion.

Live animals were described as different shades of green with white rhinophoral tips, white mottling scattered over the head and mantle, and brown pigment along the mantle margin, alternating with white patches (Jensen 1997a). The radula had 6-7 fully formed teeth (and one ghost tooth) in the ascending limb, 20-30 in the descending limb, and up to 10 teeth, plus the rod-shaped pre-radular tooth in the coiled ascus; the teeth were blade-shaped with fine lateral denticles and pointed tips. Jensen (1997a) compared the anatomy of B. darwini with that of B. typica and B. australis (the other two species whose anatomy was known) and concluded the penial morphology and the radular teeth were all different. According to Jensen (1997a) the relatively large protoconch of B. darwini indicates that this species has direct development.

Additionally, this species deposited an egg mass with just 15 eggs with "large" capsules (Jensen 1997a), potentially indicating low dispersal ability. Therefore, it is likely that B. darwini may constitute a northern Australia endemic species. Unfortunately, we had no access to specimens from this region. Although we did not have molecular data for B. darwini, this species is morphologically similar to other species in Edenttellina here examined. Therefore, B. darwini is provisionally transferred to this genus.

Remarks

Jensen (2015) introduced the name Berthelinia singaporensis Jensen, 2015 based on several specimens collected in Singapore. Jensen (2015) described the shell as broadly rounded anteriorly, more narrowly rounded posteriorly, with the highest point in the anterior half, in front of protoconch; protoconch having 1.5 whorls and located in posterior half of left valve, tilted towards right valve; hinge with a fork and a cardinal tooth on each valve. The live animal was described as uniformly bright green, protoconch white, rhinophoral tips doted in white, some specimens having white dots continuing down the rhinophores (Jensen 2015). Jensen (2015) indicated the radula was composed of 10-11 fully formed teeth (and one ghost tooth) in ascending limb and 26-30 teeth in descending limb plus 1 rod-shaped preradular tooth; the radular teeth were blade-shaped with a row of hair-like denticles along each side, tips with a small knob but not bifid. Finally, the penis was described as having a flexible stylet with peculiar lateral flanges.

In this study, we examined specimens from localities in Malaysia adjacent or very close to Here we revise the systematics of family Juliidae based on shell morphological data, including new names for Recent and fossil genera. Analyzing shell morphology of both fossil and Recent taxa by geometric morphometrics resolved five distinct clusters of taxa Juliidae, and informed the generic placement of extinct versus extant species. While PCAs did not include exactly the same specimens as molecular analyses (notably, fossil specimens), the results of both analyses support our revisionary framework for the family.

The PCA of Juliidae recovered three clusters corresponding to subfamily Juliinae and two corresponding to subfamily Bertheliniinae. The subfamily Gougerotiinae was not examined due to the absence of material suitable for morphometrics analysis. Within Juliinae, two clusters correspond to Recent taxa and one included both fossil and Recent taxa. One of the clusters of Recent taxa included specimens identified as Julia japonica and J. borbonica, the type species of the genus Prasina. Based on morphometric analyses, Prasina may need to be resurrected for this morphologically distinct group. Unfortunately, we were unable to obtain molecular data for J. japonica or J. borbonica, which is needed to confirm whether Prasina is a distinct genus. The mixed fossil/Recent cluster recovered specimens identified as Candinia krachi, which clearly supports the placement of Candinia in the subfamily Juliinae. The other cluster of Recent Julia taxa included specimens identified as J. exquisita as well as J. zebra, J. sp. 1, J. mishimaensis, and J. burni. The Bayesian and ML analyses also recovered these species as a monophyletic group that included Julia exquisita, the type species of Julia; we therefore retain the name Julia for this Recent group if Prasina is found to be valid.

The PCA analysis divided Berthelininae into two clusters, one cluster including Recent taxa and the other only fossil taxa. The fossil cluster comprised specimens identified as Berthelinia elegans, B. elegans elata, and Gourgerotia orthodontia. Because of the morphological similarities between these specimens, Gourgerotia orthodontia could be a member of the genus Berthelinia (type species Berthelinia elegans) but further analysis is needed to confirm this hypothesis. The cluster of Recent taxa comprised specimens that were previously identified as

Midorigai australis, Berthelinia caribbea, Berthelinia singaporensis, Berthelinia pseudochloris, Tamanovalva babai, Edenttellina cf. rottnesti, and Edenttellina typica. Due to the morphological distinction between Recent and fossil taxa, Berthelinia is herein deemed inappropriate for Recent taxa. We propose that Edenttellina (type Edenttellina typica), the oldest genus name available for Recent taxa within the Berthelinia lineage, should be reinstated for this group. Supporting the shell morphometric analysis, molecular analyses also recovered this Recent genus as monophyletic. Based on the results from the Bayesian and ML analyses, Edenttellina includes E. caribbea, E. pseudochloris, E. limax, E. chloris, E. australis, E. singaporensis, E. typica, E. cf. rottnesti, and E. babai.

Biodiversity of Recent taxa

The molecular results of this study recovered undescribed species of Julia, some morphologically similar to previously described species, indicating that morphology-based systematics has been overly conservative and that there are cryptic species awaiting integrative taxonomic study. However, lecithotrophic development (common in Juliidae; Edmunds, 1963;Grahame, 1969) may produce highly subdivided populations that are recovered as distinct entities in species delimitation analyses. For example, specimens of E. caribbea that were delimited as distinct species using some molecular datasets were considered to be conspecific by morphological analysis. A high degree of population genetic differentiation can confound delimitation analyses based on the multispecies coalescent model, so it is not surprising that taxa with non-dispersive larval development were oversplit in some of our delimitation analyses (Sukumaran & Knowles 2017). However, the barcoding gaps needed to recover traditionally recognized species in Juliidae (>12% for COI, >4% for 16S) are much higher than those that supported morphologically distinct species in other sacoglossan genera (Krug et al. 2018a(Krug et al. , 2018b;;McCarthy et al. 2019;Medrano et al. 2019;Rodriguez & Krug 2022;Martín-Hervás et al. 2021, 2023;Moreno et al. 2023Moreno et al. , 2025)). Thus, detailed morphological study of potential cryptic species, as well as the candidate species documented herein, are warranted to fully resolve the extent of unrecognized biodiversity in Juliidae.

Biogeography and evolution of Recent taxa

Species distributions are notably different in Julia versus Edenttellina, suggesting distinct evolutionary histories of lineage diversification, dispersal and extinction. Julia is restricted to the tropical Indo-Pacific but species have wide ranges, frequently spanning the Hawaiian Islands and west Pacific localities; as a result, sister species often co-occur. In contrast, Edenttellina also occurs in temperate Australia, the Caribbean the Eastern Pacific, so is more globally distributed at the genus level. However, individual species typically have more restricted distributions than Julia species. Sister species were generally restricted to nonoverlapping regions in Edenttellina, consistent with predictions of allopatric speciation. A number of Edenttellina species are lecithotrophic, producing non-feeding larvae with shorter expected planktonic periods, and hence more limited dispersal potential (Table 5). Reduced dispersal ability may contribute to the smaller geographic range sizes recovered for Edenttellina species, but the larval development mode of Julia remains unknown. The larger ranges of Julia spp. suggest planktotrophy is the dominant mode of development, but the more limited biogeography of the genus is surprising given greater apparent connectivity among populations.

It remains unclear whether restricted dispersal increases or decreases net diversity for marine invertebrates over evolutionary timescales, a form of species-level selection. There are few radiations of lecithotrophic lineages in Sacoglossa, in which species selection generally favors planktotrophs (Krug et al., 2015). As our understanding of diversity in Juliidae is refined and more data on development mode becomes available, this family will be important for testing current hypotheses of trait-dependent diversification. Notably, if there are clades of lecithotrophs in this group, juliids may represent an exception to the trend of greater diversification for planktotrophic lineages. Further work clarifying species boundaries and development is needed to delineate radiations of endemic species and better understand how larval type affects species richness in marine heterobranchs (Moreno et al., 2023(Moreno et al., , 2025)).

Diversity in Juliidae may also be influenced by ecological factors such as algal host specialization, as in other sacoglossan clades (Rodriguez and Krug, 2022;Moreno et al., 2023;2025). Niche partitioning is expected when ecologically similar species co-occur, and may reflect ecological speciation when sympatric sister taxa both feed and mate on distinct hosts. Species of Edenttellina are usually found living in physical association with algal hosts, and a few are highly specialized: E. caribbea on C. verticillata J. Agardh, 1847 (Clark 1994), and E. pseudochloris on the C. chemnitzia-racemosa complex (Kay, 1964;Sarma, 1975). However, many species feed widely on a range of Caulerpa spp., including more than one morphotype, e.g.

"feather" (C. serrulata, C. mexicana, C. sertularioides, C. scalpelliformis, C. taxifolia), "sea grape" (C. chemnitzia, C. lentillifera, C. racemosa, C. sedoides) or "pipecleaner" (C. C. simpliciuscula) types. Other sacoglossan genera often specialize within morphotypes or on one species of Caulerpa, Edenttellina notably diverse in its diet (Berriman et al. 2018). In other sacoglossans, preference for Caulerpa species covaries with levels of secondary metabolites (caulerpenyne, oxytoxin) produced by different algae (Baumgartner et al. 2009).

Future studies could examine whether Edenttellina species have conserved preferences for particular metabolite levels, which may limit host breadth to a subset of available Caulerpa species. However, there is less evidence that speciation via host shifting may drive diversification in Edenttellina compared with many other sacoglossan groups, given their relatively wide niche breadth among Caulerpa-feeders.

Julia appears even less likely to speciate through host associations, as specimens are not typically found feeding on Caulerpa. Without physical host association, it is unclear how ecological speciation could operate in this system (Krug 2011). The feeding behavior of most Julia species has not been verified experimentally, but two studies reported Julia species feeding on Caulerpa species that are components of microalgal communities growing in and on 'live rock': C. ambigua for J. japonica (Kawaguti & Yamasu 1966), and C. fastigiata for an unidentified Julia species (Mizofuchi & Yamasu 1987). Recent work has highlighted extensive cryptic diversity in green algal communities comprising minute, filamentous taxa, including Caulerpa species (Wade & Sherwood 2023). Future work on the specialization of Julia on minute Caulerpa species is needed to understand their adaptations to this distinctive niche, and the mechanisms allowing sympatric species of Julia to form and coexist.

Based on biogeographic theory, we would hypothesize that the more widespread genus (Edenttellina) would accumulate greater diversity, as having lineages in many different regions acts as a hedge against extinction risk (Kiessling and Aberhan, 2007). We would also expect higher diversity Edenttellina due to the larger proportion of local endemic species, and the potential for host-dependent ecological speciation due to physical association with preferred algal substrates. However, overall species richness between Julia and Edenttellina was generally comparable despite their differences in distribution, range size and patterns of host use. One possibility is that the relatively high success of Julia is due to its exploitation of an underutilized niche, feeding on microalgal species of Caulerpa that are largely inaccessible to most herbivores.

Other potential mechanisms diversification in this group, including longer-lived planktotrophic lineages or sexual selection promoting speciation, await comparative study.

Future work

This study integrated morphological traits with molecular data to produce an updated classification for Juliidae. A quantitative morphometrics approach resolved the relationships of some fossil and extant genera that long challenged systematists. Our updated classification should guide further investigations of the fossil record, informing our understanding of how Juliidae has evolved and providing valuable calibration points for molecular phylogenetics.

Many features of their biology make species of Julia and Edenttellina interesting model systems, notably the convergent evolution of a bivalved shell in Gastropoda and Bivalvia; this innovation is unique among and presents opportunities to study developmental and biomineralization pathways involved in the transition from a single-shelled larva to a bivalved adult. Unusual patterns of host use and the high proportion of lecithotrophic species in Juliidae also differ from the dominant trends in Sacoglossa and also warrant further study to better understand trait-dependent diversification in heterobranchs. Table 1). Colored dots on branch indicate the geographic of the specimen. B-C. 3. Species legend [fossil specimens indicated by white circles, Recent species indicated by solid black circles]: (1) Julia thecaphora (Carpenter, 1857), (2) Julia exquisita Gould, 1862, (3) Julia exquisita Gould, 1862, (4) Julia exquisita Gould, 1862, (5) Julia exquisita Gould, 1862, (6) Julia borbonica (Deshayes, 1863), ( 7) Julia japonica Kuroda & Habe, 1951, (8) Julia zebra Kawaguti, 1981, (9) Julia mishimaensis Kawaguti & Yamasu, 1982, (10) Julia mishimaensis Kawaguti & Yamasu, 1982, (11) Candinia (= Berthelinia) krachi Bałuk & Jakubowski, 1968, (12) Candinia (= Berthelinia) krachi Bałuk & Jakubowski, 1968, (13) Candinia (= Berthelinia) krachi Bałuk & Jakubowski, 1968, (14) Berthelinia elegans elata Cossmann, 1887, (15) Berthelinia elegans Crosse, 1875, (16) Berthelinia elegans Crosse, 1875, (17) Berthelinia elegans Crosse, 1875, (18) Gougerotia orthodonta Le Renard, 1980, (19) Edenttellina typica Gatliff & Gabriel, 1911, (20) from Victoria, Australia (MV F21539), left valve hinge (E), right valve hinge (F). G. Protoconch of a specimen from Victoria, Australia (MV F21539). Abbreviations: ams, adductor muscle scar; ct, condyloid tooth; hs, hinge socket. G-H. SEMs of the right and left valves of the same specimen from Baja California Sur, Mexico (NHMLA 1966-28.30), outer view left valve (G), inner view right valve (H). I-J. SEMs of the hinge of the right and left valves of two different specimens from Baja California Sur, Mexico (NHMLA 1966-28.30), left valve hinge (I), right valve hinge (J). K. Protoconch of a specimen