Overlooked but not forgotten: the first new extant species of Hawaiian land snail described in 60 years, Auriculellagagneorum sp. nov. (Achatinellidae, Auriculellinae)

Abstract Recent surveys of Oahu’s Waianae Mountains uncovered a small, previously undescribed species of Auriculella that is conchologically similar to the three members of the A.perpusilla group all of which are endemic to the Koolau Mountain Range. However, sequence data demonstrate that the perpusilla group is not monophyletic. Moreover, the new species is not closely related to A.perpusilla or A.perversa, the only extant members of the group, but instead is sister to A.tenella, a species from the high spired A.castanea group. A neotype is designated for A.auricula, the type species of Auriculella; all members of the conchologically similar perpusilla group are anatomically redescribed; and lectotypes designated for A.minuta, A.perversa, and A.tenella. The new species is described and compared to the type of the genus, members of the perpusilla group, and the genetically similar species A.tenella.


Introduction
Pacific Island land snails are among the most threatened faunas in the world, with more recorded extinctions since 1600 than any other group of animals (Régnier et al. 2009). Of the more than 25,000 islands spread across the Pacific, few have been extensively surveyed in modern times for their invertebrate fauna, and the estimates of extinction are probably a vast underestimate. Of the few islands and archipelagos that have been studied, like Hawaii, extinctions have been shown to be extensive (Régnier et al. 2015;. For example, as much as 93% of the endemic family Amastridae has been lost, and the other 12 families of land snails represented in Hawaii are not fairing much better . Critical to understanding and slowing the rate of extinction is accurate and updated systematics and biogeography of land snails, and other understudied groups (Cardoso et al. 2011).
The Pacific Island family Achatinellidae is the second most diverse land snail family in the Hawaiian Islands with 209 species divided into five subfamilies, two of which, the Achatinellinae Gulick, 1873 andAuriculellinae Odhner, 1922, are endemic (Cooke and Kondo 1960). Historically the large and colorful Achatinellinae have garnered much attention and the lion's share of molluscan conservation attention in Hawaii (Gulick 1872;Hadfield et al. 1993;Thacker and Hadfield 2000;Holland and Hadfield 2002, 2004, 2007Erickson and Hadfield 2008;Hadfield and Saufler 2009;O'Rorke et al. 2015;Price et al. 2015Price et al. , 2016aPrice et al. , b, 2018Sischo et al. 2016), and include the only Hawaiian land snail species protected under the US Endangered Species Act (1981,2013). However, the smaller, less colorful Auriculellinae, comprising 31 species in the genus Auriculella Pfeiffer, 1854 and one species in the genus Gulickia Cooke in Pilsbry & Cooke, 1915 have remained understudied and unprotected since the last revisions more than a century ago (Pilsbry andCooke 1914-1916). Although fossils (Solem 1977;Severns 2009) and extinct species (Severns 2011) of Hawaiian land snails have continued to be described, no new extant species of native Hawaiian land snails have been described in more than 60 years. The last described extant Hawaiian land snail species was an achatinellid in the subfamily Tornatellidinae Cooke & Kondo, 1960, Philopoa singularis Cooke & Kondo, 1960 and the most recently described Auriculella species is A. lanaiensis Cooke in Pilsbry & Cooke, 1915. Cooke and Kondo (1960) arranged Auriculella into four conchologically distinct groups: the cerea group from the southeastern islands of Hawaii, Lanai, Maui, and Molokai; and the auricula, castanea, and perpusilla groups which are all endemic to Oahu (Pilsbry andCooke 1914-1916;Cooke and Kondo 1960). The perpusilla group (A. perpusilla Smith, 1873, A. minuta Pilsbry in Pilsbry &Cooke, 1915, andA. perversa Cooke in Pilsbry &Cooke, 1915) contains the smallest species; all 6 mm or less in adult shell height. In addition to their small size these species have thin shells with 5 strongly convex whorls with low spires and weakly reflected apertures distinguishing them from the many-whorled, high-spired castanea group and the larger, thicker shelled auricula and cerea groups (Pilsbry andCooke 1914-1916).
In addition to their morphological similarity, the three species in the perpusilla group are all endemic to Oahu's eastern Koolau range (Fig. 1B-D). Recent collecting in the island's western Waianae range uncovered a previously undescribed species with features of shell size and shape that would place it in this group. The two mountain ranges are separated by a relatively dry, low elevation saddle 22 km long and 8 km wide and few land snail species have distributions in both ranges (Pilsbry andCooke 1914-1916;Cowie et al. 1995). Specimens of the undescribed species were also found in samples collected prior to 1940, which were housed in the Bishop Museum (BPBM) and labelled by Y. Kondo as a potentially new species.
Within Auriculella, intraspecific shell morphology varies and may often overlap interspecifically, making species delineation based on conchology alone difficult (Pilsbry and Cooke 1914Cooke -1916. As such, additional morphological and molecular data (e.g., DNA and RNA sequences) are necessary to distinguish among closely related species. The reproductive anatomy of only a few Auriculella species is known. Pilsbry and Cooke (1915 on plate 22) figured A. pulchra Pease, 1868 (figs 1, 2); A. cerea (Pfeiffer, 1855) (fig. 3); and A. armata (Mighels, 1845) as A. westerlundiana Ancey, 1889 (fig. 6). The reproductive anatomy of the type species of the genus, Auriculella auricula (Férussac, 1821), was figured and described by Cooke and Kondo (1960: figs 113a-d, 114a-c) who also dissected 22 other species but figured only A. castanea (Pfeiffer, 1853) (Cooke and Kondo 1960: fig. 114d). The reproductive anatomy of the other species has never been figured or described, including all members of the perpusilla group. As part of a broader project whose aim is to fully revise the systematics of the Achatinellidae, we use an integrative approach using data from conchology, radula, reproductive system, and DNA sequences, to redescribe A. auricula, the type species of the genus and all members of the perpusilla group (A. perpusilla, A. minuta, and A. perversa). We also describe a new species, A. gagneorum sp. nov., based on recently collected material and from lots housed in the Bishop Museum. Relationships of the taxa traditionally relegated to the perpusilla group, and of the conchologically similar A. gagneorum sp. nov., are explored with a mitochondrial and nuclear gene dataset. To enhance the stability of the nomenclature, we designate a neotype for A. auricula and lectotypes for members of the perpusilla group.

Materials and methods
As part of a long-term study of extant Hawaiian land snails, our team has surveyed more than 1000 sites across the six largest Hawaiian Islands (Kauai, Oahu, Maui, Molokai, Lanai, Hawaii). The targeted locations were those that historically supported snail populations, as well as more remote areas with remnant native vegetation that were often accessible only by helicopter. Surveys followed Durkan et al. (2013) and consisted of leaf litter sampling and hand collecting for at least one-person-hour by a minimum of two experienced malacologists in quadrats of at least 10 m 2 , but up to 100 m 2 , terrain permitting. GPS coordinates were collected at every survey site and coordinates were estimated for historical BPBM specimen records using locality, field notes, maps, and other descriptions. The precise locations (e.g., GPS coordinates) for material listed are not provided here for conservation purposes but are kept in the State of Hawaii Department of Land and Natural Resources Snail Extinction Prevention Program and Bishop Museum Malacology databases. Distributional maps were created using QGIS v3.8.2 (QGIS 2019) and used to show historical and current distributions of the species treated herein.
Newly collected material was photographed, flash boiled (Fukuda et al. 2008), and then fixed in 95% ethanol, after which a small piece of foot tissue was removed for DNA extraction. The remaining soft tissues were preserved in 80% ethanol, and dissections were performed on preserved specimens submerged in 75% ethanol. Shells and reproductive anatomy were photographed with digital single-lens reflex cameras (e.g., Cannon EOS 7D) attached to a dissecting microscope. Photographs of reproductive anatomy were traced in Photoshop to produce line drawings. Shell measurements were made using an ocular micrometer and each measurement was repeated three times and averaged for 50 specimens per species. Shell measurements, shell height (H), shell width (W), aperture height (AH), aperture width (AW), and number of whorls (WH) were made following Slapcinsky and Kraus (2016: 30, fig. 1). All pertinent type and comparative material were examined and photographed. Locality and collector information of materials examined were listed as verbatim. Materials examined for the new species is provided in the text and all others can be found in Suppl. material 1. Museum collections are abbreviated: Radulae were tissue-digested in 180 µL of T1 lysis buffer (Macherey-Nagel) containing 20 mg/mL of Proteinase-K and rinsed in de-ionized water. Cleaned radulae were mounted directly on carbon adhesive tabs attached to aluminum stubs, which were then coated with 25-30 nm gold/palladium (60/40) and photographed using an Apreo scanning electron microscope (FEI Company) at the National Museum of Natural History, Washington.
Total genomic DNA (gDNA) was extracted from an approximately 1 mm 3 piece of foot tissue using the Macherey-Nagel NucleoSpin Tissue Kit following the manufacturer's instructions, with the exception that elution was with 60 µl of elution buffer supplied with the kit, and gDNA stored at -20 °C prior to amplification via the polymerase chain reaction (PCR).
Portions of two mitochondrial genes, 16S ribosomal DNA (rDNA) and cytochrome c oxidase subunit I (COI), and the nuclear encoded 28S rDNA were amplified using primers listed in Table 1. Reactions were carried out in 25 µl volumes containing 1-2 µl template DNA and a final concentration of 1 U of MangoTaq DNA polymerase (Bioline), 1X reaction buffer, 0.2 mM each dNTP, 2.5 mM MgCl 2 and 0.75 µM of each primer, 10 µg BSA, and 0.5% DMSO. Cycling parameters were one cycle of 5 min at 95 °C, 1 min at 44-48 °C, 2 min at 72 °C, followed by 34 cycles of 95 °C, 46-50 °C,  Table 2) Electropherograms were checked for errors, edited, and assembled using Geneious Prime 2019 (http://www.geneious.com/). Sequences of COI were unambiguously aligned using MAFFT ver. 7.388 with the iterative refinement method E-INS-I (Katoh and Standley 2013) implemented in Geneious Prime 2019. Alignments where checked against amino acid sequences as references. Ribosomal genes were aligned using MAFFT and refined using Gblocks ver. 0.91b (Castresana 2000). Refinement of the 16S and 28S alignments in Gblocks removed regions of ambiguous homology created by the addition of gaps during initial alignment and the hypervariable nature of some regions. Phylogenetic analyses were done with and without these regions to evaluate their impact. Sequence alignments were concatenated in Geneious Prime and exported as phylip files for phylogenetic analysis.
Phylogenetic reconstruction was conducted using maximum likelihood (ML) in IQ-TREE ver. 1.6.12 (Nguyen et al. 2015). The best-fit partitioning scheme and the most appropriate substitution model for each partition were estimated using the integrated ModelFinder algorithm (Kalyaanamoorthy et al. 2017) and partition models (Chernomor et al. 2016). Nodal support was estimated with 5,000 ultra-fast bootstrap replicates (Hoang et al. 2018).
To corroborate species delineation based on conchological and anatomical analyses and phylogenetic reconstruction, we used the DNA barcode-based species identification method implemented in SpeciesIdentifier ver. 1.8 (Meier et al. 2006).
Museum catalog numbers for specimens used in DNA analysis with numbers of specimens from which shell measurements, reproductive anatomy, and radular morphology were obtained, are listed in Table 2.

Results
Recent surveys recorded extant populations of two of the three species within the perpusilla group: A. perpusilla and A. perversa (Fig. 1C, D, respectively) and a new spe- Table 2. Museum catalog numbers for specimens used in genetic analysis with numbers of specimens from which shell measurements, reproductive anatomy and radular morphology were obtained. Catalog numbers (BPBM) are for lots from which specimens were sequenced (N = number of individuals sequenced from each lot).

Auriculella ambusta
Molokai BPBM 285842 (1) MT519859 MT519878 MT519913 ---cies with similar shell morphology, A. gagneorum sp. nov. (Fig. 1F). No populations of A. auricula (type species of the genus, Fig. 1A) or A. minuta (Fig. 1B) were recorded in our surveys and both species may be extinct. The 104 snails representing ten Auricullela species and two outgroup taxa (Tornatellaria sp. and Tornatellides sp.) sequenced for this study produced 53 COI haplotypes, 19 and 35 sequences for 16S and 28S, respectively. Alignments for each locus were 654 bp for COI, 464 bp for 16S and 539 bp for 28S, making the concatenated dataset of 53 individuals 1657 bp with 223 parsimony informative sites. Sixteen individuals were represented by all three loci, while three individuals had only COI and 16S, 19 with COI and 28S, and 15 with only COI. The best-fit partitioning scheme used distinct models for each locus with the best-fit models being K3Pu+F+I+G4, TPM2u+F+G4, and TIM3+F for COI, 16S, and 28S respectively.
The ML tree constructed from the concatenated dataset produced a well-resolved tree with all conchologically defined taxa recovered in strongly supported clades (Fig. 2). None of the groupings suggested by Cooke and Kondo (1960) based on gross shell morphology were recovered in the ML tree. As such, A. perpusilla and A. perversa, previously referred to the perpusilla group were recovered in unrelated clades with each as sister to much larger shelled species, A. ambusta and A. montana, respectively. Similarly, the new species Auriculella gagneorum sp. nov. was recovered as sister to A. tenella and not close to A. perpusilla or A. perversa with which it was previously confused.
The best match/best close match criteria (Meier et al. 2006) applied to all 53 COI haplotypes successfully matched all sequences in the correct conspecific clusters within a 3-4% threshold consistent with conchologically and phylogenetically recognized clades. Correct identifications with both approaches was 94.33%, with the other 5.66% (three sequences) lacking any conspecific sequences with which to cluster. These included the two outgroup taxa and A. montana, all of which were represented by a single sequence.  (Gulick 1873).

Systematics
Diagnosis. Small to moderately sized Achatinellidae, 4 to 12 mm in adult shell height. Shells either dextral or sinistral, taller than wide, with a strong parietal la- mella. Juvenile shells have two columellar lamellae, one or both of which are lacking in adults. Phallus with an epiphallus and a nearly apical appendix. Phallus retractor muscle inserted apically on the epiphallus and not secondarily attached to the appendix. Members of Auriculella are the only achatinellids known to have an epiphallus. All Auriculella species are oviparous (Pilsbry andCooke 1914-1916). Figures 1A, 3A Diagnosis. Shell. Shell dextral or sinistral with flat-sided whorls and an obtuse apex, H = 8.0 ± 0.4 mm, W = 4.3 ± 0.2 mm, WH 6.0 ± 0.2, AH = 4.1 ± 0.2 mm, AW = 3.1 ± 0.2 mm (N = 50; Table 2). Columella with a single strong lamella and without an axial ridge. Parietal lamella is strong and smooth and not undulate, extending 0.3 to 0.7 whorls into the aperture. Shell color is tan, brown, or yellowish, often with a single narrow brown or white band (Fig. 3B). White bands are sometimes bordered by two darker brown bands and apical whorls are often darker brown. Lip reflected, thickened, white or brown in color.

Auriculella auricula (Férussac, 1821)
Reproductive system. Phallus retractor muscle relatively long, attached apically to a short but well-defined epiphallus (Fig. 4A). Appendix is longer than the phallus and about ⅔ the diameter of the phallus at its attachment. The appendix narrows abruptly at ⅕ its length and remains narrow to its terminus. Phallus is broad, narrowing only slightly apically and basally. Atrium is relatively short and broad. Vagina is about ⅓ the length of the phallus.
Radula. Radula with an irregular rachidian flanked on either side by rastriform marginal teeth, as diagnostic of the family (Fig. 5A). Each tooth has a long narrow base that expands slowly for ¾ of the length of the tooth before reaching the forward curving cusps, which comprise the remaining ¼ of the tooth. There are three long cusps at mesocone, endocone, and ectocone positions with two or more alternating larger and smaller cusps intercalated between them. Number of teeth per row range from 177 to 183 (N = 6; Table 2).
Distribution and ecology. Auriculella auricula is endemic to Oahu's Koolau Mountains (Fig. 1A), historically found across the range at elevations from 61 m to 305 m. The species is arboreal and found on vegetation including: Cordyline sp. Freycinetia arborea, Metrosideros polymorpha, Canna sp. (BPBM 34025, 49056, 51405), Aleurites moluccanus, Psychotria sp., Zingiber sp., Psidium cattleyanum, Musa sp., Asplenium sp., and unspecified ferns and shrubs. The species has also been recorded on the ground under stones, logs and dead leaves. Live specimens recorded in the BPBM collection were last collected by Y. Kondo in 1946 from Palolo Valley; the species has not been recorded in recent surveys and is considered here possibly extinct.

Remarks.
In the original description, Férussac (1821) provided measurements for a single shell of three lines (6.8 mm) in height and 1¾ lines (4.0 mm) in width. His collection is housed in MNHN where there are two lots labelled A. auricula that are attributed to Férussac. The first (MNHN IM-2000-34306, 34307, 34308) is from Férussac's collection but does not contain original labels. The three dextral shells are identified as A. auricula from the Mariana Islands, but they are not A. auricula and instead appear to be a gerontic adult and two juveniles similar to Auriculella ambusta, a species not found on the same mountain range as A. auricula. The other lot (MNHN IM-2014-7009) is from the Deshayes collection. Its source is unknown but probably came from Férussac whose specimens Deshayes used to complete Férussac's "Histoire naturelle des mollusques terrestres et fluviatiles" after Férussac's death in 1836. The lot contains six specimens of Auriculella pulchra, two of which are sinistral and all of which are larger than 6.8 mm. The two lots are not consistent with Férussac's description, and we exclude these lots as possible syntypes of A. auricula. We have not located any other type material of A. auricula and we consider the types to be lost. Stabilizing the nomenclature of this species is important because it is the type species of the genus Auriculella, a genus with many similar but conchologically variable and poorly resolved species, nearly all of which are highly endangered. We designate BPBM 18709 (Fig. 3A) from Tantalus, Oahu as neotype of Auriculella auricula to stabilize the taxonomic status and type locality of the species as well as the genus Auriculella. The neotype matches Férussac's original description in having an acute ovoid shell with an obtuse apex, strong parietal lamella, and single columellar lamella. The color of the neotype is more tan than yellowish as described in the original description but the species is known to be polymorphic for shell color and pattern as well as chirality. The shell used in Férussac's description was sinistral while the neotype is dextral. We chose a dextral specimen with slightly different coloration because it was used by Cooke and Kondo (1961) to describe the nervous system and reproductive anatomy of Auriculella auricula thus clearly defining the species as well as the genus. The other four specimens from BPBM 189709 are re-cataloged as BPBM 285783. One of these is a broken shell presumably corresponding to the animal dissected by Cooke and Kondo (1961). Figures 1B, 3C, D, 4B, 5B
Reproductive system. Phallus retractor muscle relatively long, attached apically to a short but well-defined epiphallus (Fig. 4B). Appendix is nearly equal in length to the phallus. Appendix the diameter of the phallus at its attachment, narrowing abruptly at ⅓ its length and remaining narrow to its terminus. Apical ¾ of the phallus is broad, basal ¼ narrows abruptly remaining narrow to the junction with the moderately long atrium. Vagina is long and nearly half the length of the phallus.
Radula. Radula with an irregular rachidian flanked on either side by rastriform marginal teeth, as diagnostic of the family (Fig. 5B). Each tooth has a long narrow base that expands slowly for ¾ of the length of the tooth before reaching the forward curving cusps, which comprise the remaining ¼ of the tooth. There are three long cusps at mesocone, endocone, and ectocone positions with two or more alternating larger and smaller cusps intercalated between them. There are roughly 105 teeth per row (N = 5; Table 2).
Distribution and ecology. Auriculella minuta is endemic to Oahu's Koolau Mountain Range (Fig. 1B), found predominantly in the southern portion of the range with a few historical records from the southern edge of the northern Koolau Mountains. No elevational range information is available with these historical specimen records. The species is arboreal and found on vegetation, including Cordyline fruticosa, Dioscorea alata, Freycinetia arborea, Kadua affinis, Lobelia sp., Psidium guajava, and Touchardia latifolia. Live specimens recorded in the BPBM collection were last collected by Olaf Oswald in Waiahole in 1931 and is considered herein extinct.
Remarks. A holotype was not designated in the original description and the type series came from two different localities: Nuuanu collected by Cooke, and Palolo collected by both Cooke and Lyman (Pilsbry and Cooke 1915: 90). Five figures were provided with the original description (Pilsbry and Cooke 1915: pl. 25, figs 5-9) from Nuuanu, which according to the figure caption were based on specimens from BPBM and ANSP. The figure caption did not indicate which museum lots the figured specimens came from but the BPBM ledger in Cooke's handwriting lists: BPBM 42377 "holotype", figs 5, 9 (see note for ANSP 91816 below); BPBM 42378, "cotype", 73037. The ledger also indicated that two specimens were given to Dautzenberg whose collections were obtained by RBINS. Two ANSP lots 91816 and 113294 were received by Pilsbry from Cooke. The original label for ANSP 91816 is marked "cotype" and the source for fig. 9 in the description. Because the caption for figs 5-9 states that at least one of the figured specimens is from ANSP we believe this to be the source for fig. 9 rather than BPBM 42377 as stated in the BPBM ledger, although we do believe BPBM 42377 is the source for fig. 5. Johnson (1996) lists lot BPBM 42377 as the holotype citing the original BPBM specimen labelling. However, the species description is clearly based on multiple specimens all of which should be considered syntypes.
In addition to the specimens from Nuuanu, the material from Palolo collected by both Lyman and Cooke are also part of the type series. There is only one lot of Auriculella minuta (BPBM 12808) collected by Lyman from Palolo and although it is not labelled as being part of the type series it is likely the lot collected by Lyman that was mentioned in the species description. A second lot, BPBM 16435, lacks information on the collector but may be the lot collected by Cooke. We here designate lot BPBM 42377 as the lectotype, restricting the type locality to Nuuanu. Unlike the other species traditionally placed in the perpusilla group, the shell of A. minuta is dextral rather than sinistral. The columella does not bear an axially oriented ridge like the one found in A. perversa. The palatal lamella is smooth and not undulate unlike that of A. gagneorum sp. nov. The epiphallus is short and well defined similar to A. gagneorum sp. nov., but unlike the long epiphallus of A. perpusilla or the poorly defined epiphallus of A. perversa. The appendix narrows abruptly at approximately ⅓ its length unlike the gently tapered appendix of A. gagneorum sp. nov. Figures 1C, 3E, F, 4C, 5C
Reproductive system. Phallus retractor muscle relatively short, attached apically to a long epiphallus, which is nearly ⅓ the length of the phallus (Fig. 4C). Appendix is nearly equal in length to the phallus. Appendix slightly over half the diameter of the phallus at its attachment, narrowing abruptly at ⅓ its length and remaining narrow to its terminus. Apical ⅔ of the phallus is broad, basal ⅓ narrows abruptly and remains narrow to the junction with the short atrium. Vagina is long and nearly half the length of the phallus.
Radula. Radula with an irregular rachidian flanked on either side by rastriform marginal teeth, as diagnostic of the family (Fig. 5C). Each tooth has a long narrow base that expands slowly for ¾ of the length of the tooth before reaching the forward curving cusps, which comprise the remaining ¼ of the tooth. There are three long cusps at mesocone, endocone, and ectocone positions with two or more alternating larger and smaller cusps intercalated between them. There are roughly 127 teeth per row (N = 5; Table 2).
Remarks. No holotype was designated in the original description which included a single figure and provided a single set of measurements: height 4 mm width 2 ⅔ mm. The shell donated by Gulick is MCZ 39912 and is labeled holotype. Pilsbry and Cooke (1915: 91) indicated that only a single shell existed; "The single specimen collected by Mr. Gulick and described by Mr. Smith, is unfortunately broken." Consequently, MCZ 39912 is the holotype by monotypy.
Unlike A. minuta, A. perpusilla is sinistral and the columella does not bear an axially oriented ridge like the one found in A. perversa. The palatal lamella is smooth and not undulate like A. gagneorum sp. nov. The epiphallus is long unlike the poorly defined epiphallus of A. perversa or the short but well-defined epiphallus of A. minuta and A. gagneorum sp. nov. The appendix narrows abruptly at approximately ⅓ its length unlike A. gagneorum sp. nov. Figures 1D, 3G, H, 4D, 5D
Reproductive system. Phallus retractor muscle relatively short attached apically to a short and poorly defined epiphallus (Fig. 4D). Appendix is as long as the phallus and a bit over half the diameter of the phallus at its attachment, narrowing abruptly at ⅓ its length and remaining narrow to its terminus. Phallus is broad, narrowing only slightly at the junction with the short atrium. Vagina is short.
Radula. Radula with an irregular rachidian flanked on either side by rastriform marginal teeth, as diagnostic of the family (Fig. 5D). Each tooth has a long narrow base that expands slowly for ¾ of the length of the tooth before reaching the forward curving cusps, which comprise the remaining ¼ of the tooth. There are three long cusps at mesocone, endocone, and ectocone positions with two or more alternating larger and smaller cusps intercalated between them. There are roughly 127 teeth per row (N = 4; Table 2).
Distribution and ecology. Auriculella perversa is endemic to Oahu's southern Koolau Mountain Range (Fig. 1D), recorded from 61 m to 914 m elevation. Auriculella perversa is arboreal and found on Clermontia sp., Cordyline fruticosa, Dubautia laxa, Freycinetia arborea, Metrosideros polymorpha, Musa sp., Pritchardia sp., Psidium guajava, and unspecified ferns, tree trunks, and dead leaves. Prior to our recent surveys the last live specimens were collected in 1939 by O.H. Emerson, E.H. Bryan Jr., and D. Anderson on Kulepeamoa Ridge in the southern Koolau Mountain Range, and the only known extant population recorded occurs in Tantalus.

Remarks.
A holotype was not designated in the original description and the type series came from two different localities: Nuuanu collected by Cooke, and Kuliouou collected by Thaanum. Two figures were provided with the original description (Pilsbry and Cooke 1915: pl. 25, figs 3, 4) for material from Nuuanu at BPBM. However, the figure caption does not indicate type status or lot numbers. The BPBM ledger in Cooke's handwriting lists: BPBM 42384 "holotype", figs 3, 4; BPBM 42385, "paratypes". The BPBM ledger documents that BPBM 42385 was also the source of MCZ 7034 and SMF 7090 (Zilch, 1962: 78). The ANSP online catalog list additional specimens from BPBM and labeled as syntypes: ANSP 163411, 91817, Nuuanu; ANSP 163399 Kuliousu [sic]. ANSP 108272 Kuliouou was collected by D. Thaanum. Johnson (1996: 193) stated that the "holotype" was BPBM 42384 based on its specimen label. However, it is clear that the original description was based on multiple specimens which should be considered syntypes. We here designate BPBM 42384 as the lectotype. As a result of this lectotype designation the type locality is restricted to Nuuanu. Unlike A. minuta, the shell of A. perversa is sinistral. The columella bears an axially oriented ridge unlike all other species in the perpusilla group. The palatal lamella is smooth and not undulate like A. gagneorum sp. nov. The reproductive system includes a short and poorly defined epiphallus and an appendix that narrows abruptly at ap-proximately ⅓ its length. The epiphallus is short and poorly defined unlike the long epiphallus of A. perpusilla or the short but well-defined epiphallus of A. minuta and A. gagneorum sp. nov. Figures 1E, 3I, J, 4E, 5E
Reproductive system. Phallus retractor muscle relatively long attached apically to a short but well-defined epiphallus (Fig. 4E). Appendix ⅓ longer and about half the diameter of the phallus at its attachment, narrowing abruptly at ⅓ its length and remaining narrow to its terminus. Phallus is broad, narrowing by half at the junction with the short atrium. Vagina is of moderate length.
Radula. Radula with an irregular rachidian flanked on either side by rastriform marginal teeth, as diagnostic of the family (Fig. 5E). Each tooth has a long narrow base that expands slowly for ¾ of the length of the tooth before reaching the forward curving cusps, which comprise the remaining ¼ of the tooth. There are three long cusps at mesocone, endocone, and ectocone positions with two or more alternating larger and smaller cusps intercalated between them. There are roughly 129 teeth per row (N = 3; Table 2).
Distribution and ecology. Auriculella tenella is endemic to Oahu's Waianae Mountains, historically found throughout the range between 518 and 1227 m in elevation (Fig. 1E). This species is arboreal and found on Broussaisia sp., Cordyline sp., Freycinetia arborea, Lantana sp., Pelea sp., Sadleria cyatheoides, Bidens sp., Coprosma sp., Euphorbia sp., Metrosideros sp., Psychotria sp., Ilex sp., Philodendron sp., and unspecified ferns, grasses, tree trunks, and small plants on stream banks. Occasionally, this species has been recorded on the ground on stones, dead leaves, and bark. The last live specimens in the BPBM collection were recorded in 1948. Our recent surveys documented the species in only three locations in the southern Waianae range.
Remarks. A holotype was not designated in the original description, however, the type locality is listed as "Waianae" and collected by Baldwin. Ancey provided measurements in the original description, "Long., 6; diam., 3; alt. ap., 2 2/3 millim.", which agree well with the designated lectotype. The ledger entry for BPBM 18943 lists four "types" collected by Baldwin from Waialae [sic]. However, only three specimens were found. The material probably came from Paul Geret who acquired Ancey's collection after his death and subsequently sold it. Much of Ancey's Hawaiian land and freshwater material was purchased by BPBM in 1908 (Johnson, 1996) but some was sold to other buyers. Both NMW 1955.158.24126 and RBINS 10591 (accession number) have Geret "cotype" labels (Wood and Gallichan 2008:88). Tomlin, the source of the NMW lot, had a sales list confirming purchase from the Ancey collection.
The shell of A. tenella has approximately seven nearly flat-sided whorls unlike A. auricula, A. minuta, A. perpusilla and A. perversa, which have approximately five whorls, and are inflated in all but A. auricula. Auriculella tenella is sinistral unlike A. minuta and does not bear an axially oriented columellar ridge like A. perversa or an undulating palatal lamella like A. gagneorum sp. nov. The epiphallus is short and well defined unlike the long epiphallus of A. perpusilla or the poorly defined epiphallus of A. perversa. The appendix narrows abruptly at approximately ⅓ its length unlike A. gagneorum sp. nov.    Table 2). Columella in juveniles with a strong lamella that is reduced and covered by a thickening of the inner edge of the lip in adults. Adults do not show a short projection or angular edge where the columellar lamella was located. Parietal lamella is often undulate, usually with three peaks, extending 0.2 to 0.5 whorls into the aperture. Shell color is white, pale tan or dark brown, with or without irregularly placed axial bands of brown, or with a single peripheral band of pale tan or dark brown. Specimens occasionally pale tan with two poorly defined dark bands on either side of a pale tan peripheral band.

Auriculella gagneorum
Reproductive system. Phallus retractor muscle long, attached apically to a short but well-defined epiphallus (Fig. 4F). Appendix slightly longer than the phallus. Appendix ⅔ the diameter of the phallus at its attachment, tapering gently to ⅓ its length, then remaining narrow to its terminus. Apical ¾ of the phallus is broad, tapering slightly both apically and basally, basal ¼ narrows slightly above junction with the short atrium. Vagina is short.
Radula. Radula with an irregular rachidian flanked on either side by rastriform marginal teeth, as diagnostic of the family (Fig. 5F). Each tooth has a long narrow base that expands slowly for ¾ of the length of the tooth before reaching the forward curving cusps, which comprise the remaining ¼ of the tooth. There are three long cusps at mesocone, endocone, and ectocone positions with two or more alternating larger and smaller cusps intercalated between them. Number of teeth per row range from 135 to 153 (N = 3; Table 2).
Distribution and ecology. Auriculella gagneorum sp. nov. is endemic to Oahu's Waianae Mountain Range and was recorded as a potentially new species primarily from the southern Waianae Mountain Range, with several populations in the northern part of the range (Fig. 1F). The species is arboreal and has been found on Antidesma platyphyllum, Broussaisia arguta, Lantana sp., Melicope anisate, Myrsine lessertiana, and occasionally on unspecified ferns and dead leaves. The last known record of this species prior to recent surveys was by Yoshio Kondo, T. Maa, George F. Arnemann, and Peter Char in 1960. From 2013 to 2018 we recorded extant populations of this species from three locations in the southern Waianae Mountains.
Remarks. The shell is sinistral unlike A. minuta and the columella does not bear an axially oriented ridge like the one found in A. perversa. The palatal lamella is often undulate unlike all other members of the A. perpusilla group. The epiphallus is short but well defined similar to A. minuta but unlike the long epiphallus of A. perpusilla or the poorly defined epiphallus of A. perversa. The appendix tapers gently unlike the appendices of A. auricula, A. minuta, A. perpusilla, A. perversa, and A. tenella which all narrow abruptly.
Etymology. Named in honor of Betsy and Wayne Gagne for their indefatigable efforts advocating for the conservation of Hawaii's unique and highly endangered biota.

Discussion
The Auriculella perpusilla species group (A. perpusilla, A. perversa, A. minuta) was defined as having species with small, thin, relatively low spired shells of approximately five inflated whorls. Auriculella gagneorum sp. nov., shares these shell characteristics. These four species can be distinguished from one another using a suite of morphological features including shell chirality (only A. minuta is dextral); presence of axially oriented ridge of the columella (only present in A. perversa); appearance of the palatal lamella (undulated only in Auriculella gagneorum sp. nov.); length of the epiphallus (those of both Auriculella gagneorum sp. nov. and minuta are short and well-defined); and development of the appendix (tapers gently in Auriculella gagneorum sp. nov. and narrow abruptly in others). The DNA data corroborate the difference seen in anatomy and conchology. In contrast to expectations based on shell morphology alone, the perpusilla group is not monophyletic and Auriculella gagneorum sp. nov. is not closely related to either A. perpusilla or A. perversa, the only other extant members of the group for which DNA data are available (Fig. 2). Instead, A. gagneorum sp. nov. clusters with A. tenella, a high spired and tightly coiled species from the castanea group, which also occurs in the Waianae Mountains. Similarly, A. perpusilla and A. perversa are more closely related to species with highly dissimilar shell morphologies, A. ambusta and A. montana, respectively (Fig. 2). The latter two have large, thick shells and are usually placed in the auricula group with other robust species. Patterns of relatedness recovered in our phylogenetic analyses indicate these gross shell characters, which are unlikely to be independent of one another, are insufficient for delineating taxa or characterizing relationships within the genus. Multiple instances of convergence in shell morphology across the genus may be explained by adaptation to similar microhabitats, or non-adaptive diversification combined with constraints on shell morphospace (Gittenberger 1991;Cowie 1995;Rundell and Price 2009; Chiba and Cowie 2016; Gillespie et al. 2018). Disentangling the processes responsible for these patterns will require additional studies of the functional morphology, ecology, and behavior of Auriculella species.
Historically, all four species treated here once had much larger geographic ranges, with multiple populations recorded in the last century (Fig. 1). Like nearly all land snail species across Hawaii, Auriculella spp. numbers have declined dramatically with an estimated 45% of the species considered extinct, and many historical populations extirpated as a result of habitat destruction, invasive species, and possibly climate change. Despite the grim statistics, there remain a number of species that can yet be saved from extinction, but only with a clear understanding of their systematics, biogeography and ecology. For example, A. tenella, A. gagneorum sp. nov., A. perversa, and A. perpusilla, are now known from only three locations for each of these species. These data combined with knowledge of reproduction and population growth rates can be used to better manage these imperiled species.
Low reproductive and growth rates are often characteristic of species that have evolved on isolated oceanic islands (MacArthur and Wilson 2001; Covas 2011), and Auriculella spp. are probably no exception. Two laboratory reared adults of Auriculella gagneorum sp. nov. produced 33 eggs in 250 days between 17 May 2018 and 23 January 2019 (Fig. 6A-C). The delicate nature of the eggs of this imperiled species permitted the measurement of only three eggs, which had an average diameter of 0.99 ± 0.05 mm. These large eggs, relative to the size of the animal, take approximately 58 days to hatch (Lindsay Renshaw, pers comm.). Such low fecundity in combination with extreme range reduction decreases the chances of long-term species and population persistence (Bick et al. 2018), particularly in the face of predation by introduced predators (Chiba and Cowie 2016).
Updated and comprehensive assessments of the systematics, biogeography, and ecology of taxa are necessary for effective management and development of long-term recovery plans. Additional surveys to locate remaining species and persisting populations are needed now, while there is still an opportunity to prevent or slow the rate of species loss (Solem 1990;. These surveys provide important opportunities to study and preserve species and develop populations for captive rearing, which in turn can be repatriated to protective enclosures in natural habitats with the goal of ultimately reintroducing species back into the wild (Natural Area Reserves Program 2016; Yeung and Hayes 2018). Our surveys have recovered species not recorded alive since the 1950s (e.g., Auriculella perpusilla, A. perversa, A. tenella) and others feared extinct . They have also uncovered several previously undescribed species, indicating that there is still much to learn about this highly imperiled fauna, and still hope that we might save some of it for future generations (Solem 1990).