Molecular phylogeny of the Orthalicoidea land snails: Further support and surprises

The superfamily Orthalicoidea comprises approximately 2,000 species of terrestrial gastropods, mostly concentrated in the Neotropics but also present in southern Africa and Oceania. We provide a multi-marker molecular phylogeny of this superfamily, reassessing its family- and genus-level classification. We exclude two families from the group, Odontostomidae and Vidaliellidae, transferring them to Rhytidoidea based on their phylogenetic relationships as recovered herein. Two new families are recognized herein as members of Orthalicoidea, Tomogeridae and Cyclodontinidae fam. nov. The family Megaspiridae and the subfamily Prestonellinae are paraphyletic but are retained herein for taxonomic stability. The subfamily Placostylinae is synonymized with Bothriembryontinae. The new genera Alterorhinus gen. nov. and Sanniostracus gen. nov. containing some Brazilian species are described here to better reflect the phylogeny. The fossil record and paleobiogeographic history of the group is explored under the new phylogenetic framework.


Introduction
The superfamily Orthalicoidea Martens, 1860 contains a diverse assemblage of mostly neotropical snails, conservatively estimated at around 2,000 species [1]. Generally referred to as "tree snails", their centre of diversity is in the Americas, where they are distributed from southern USA to Patagonia. One lineage is present in Oceania, with most representatives in Australia and several additional species found on the southern Pacific Islands, including Aotearoa New Zealand, the Solomon Islands, Fiji and Vanuatu. Finally, there is also a lineage in southern Africa, with few representatives.
Presently, Orthalicoidea contains seven families [2]: Amphibulimidae Fischer, 1873, Bothriembryontidae Iredale, 1937 (the only one with representatives outside the Americas), Bulimulidae Tryon, 1867, Megaspiridae Pilsbry, 1904 manufacturer's standard protocol, with the addition of a repetition of the final step to increase yield. The molecular markers targeted for this study were the same as in [7]: (1) the barcoding fragment of the COI mitochondrial gene, circa 650 bp long, using the primers LCO/HCO [18]; (2) a fragment of the nuclear H3 gene (histone 3), circa 270 bp long, using the primers H3pulF and H3pul3 [19]; (3) a continuous fragment of nuclear DNA including the 3 0 end of the 5.8S rRNA gene, the ITS2 region, and the 5 0 end of the 28S rRNA gene, totalling circa 1,300 bp long and amplified in two fragments using the primers LSU-1/LSU-3 and LSU-2/LSU-5 [20,21].
The PCR amplification protocols were as follows. COI: initial denaturation at 95˚C for 3 min; 35 cycles of denaturation at 95˚C for 30 s, annealing at 48˚C for 1 min, and extension at 72˚C for 2 min; final extension at 72˚C for 5 min. COI: initial denaturation at 95˚C for 3 min; 40 cycles of denaturation at 95˚C for 30 s, annealing at 57˚C for 30 s, and extension at 72˚C for 40 s; final extension at 72˚C for 5 min. ITS2+28S: initial denaturation at 95˚C for 3 min; 40 cycles of denaturation at 95˚C for 30 s, annealing at either 50˚C (ITS2 section) or 45˚C (28S section) for 1 min, and extension at 72˚C for either 5 min (ITS2 section) or 2 min (28S section); final extension at 72˚C for 4 min.
The success of the PCR was visually assessed via agarose gel electrophoresis. The PCR products were cleaned with ExoSAP-IT™ (Affymetrix Inc.) following the manufacturer's protocol. Samples were prepared and sent to Massey Genome Service (Massey University, Palmerston North, New Zealand) to be Sanger sequenced. Sequences were quality-proofed and de novo assembled in Geneious Prime (v.2020.2.2, Biomatters Ltd.). Consensus sequences were uploaded to GenBank (Table 1).

Phylogenetic analysis
Sequences were aligned using the MUSCLE plugin [22] in Geneious Prime with default settings (i.e., optimized for accuracy). The resulting alignments were visually proofed for inconsistencies and then run through Gblocks [23] with the least restrictive settings to eliminate poorly-aligned or data-deficient positions that could introduce noise into the analysis. The resulting post-Gblocks alignments were concatenated for a single phylogenetic analysis, with all three genetic markers treated as individual partitions.
Bayesian inference phylogenetic analysis was performed using MrBayes (v.3.2.7, [24]) through the CIPRES Science Gateway (v. 3.3, [25]). Two concurrent analyses, each with 4 Markov chains of 200 million generations (the first 20% discarded as 'burn-in'), were run with the default priors, nst = 6, rates = invgamma, temperature parameter = 0.1, sampling every 1,000 generations. Substitution model parameters were unlinked across the markers (COI, 16S, and ITS+28S). MCMC convergence was assessed by examining the standard deviation of split frequencies (~0.01) and the potential scale reduction factor, PSRF (~1.0), as well as the trace plots in Geneious [26].  To check for inconsistencies between markers, two further trees were built using only the H3 or the ITS2+28S alignments (species lacking sequences of one of these markers were excluded from both trees; trees not shown). In these cases, only 20 million generations were enough. The COI alignment was not used due to the known low resolution this marker provides to elucidate family level relationships within Orthalicoidea (e.g., [10,12]). Almost no inconsistencies were found, but some interesting differences are discussed below.
Finally, given the unexpected position of two genera (Odontostomus and Pilsbrylia) outside of Orthalicoidea, a further phylogenetic analysis was conducted to better asses their relationship to other stylommatophorans. This analysis includes a small subset of Orthalicoidea from the analysis above, as well as further taxa chosen to accommodate the potential Helicina superfamilies that these two genera could belong to ( Table 3). The same methodology and parameters from above apply, apart from the number of generations, which was set to 80 million.

Nomenclatural acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/". The LSID for this publication is: urn:lsid:zoobank.org:pub:B5736A88-9939-4D49-B151-8670B2402F2C. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.
The trees built with only one marker (H3 or ITS2+28S) to test for inconsistencies show some loss of resolution, which is to be expected considering these sequences are much shorter and the large number of taxa involved. The loss of resolution is more marked in some specieslevel relationships (inside Bulimulus and Drymaeus) and in the appearance of a few familylevel polytomies. Nevertheless, almost no inconsistencies were found in relation to the total- evidence tree. The few (and interesting) differences are as follows: (1) in the ITS2+28S tree, the node from which Odontostomus branches is more basal than the one from which Catracca does; (2) in the H3 tree, there was a better separation of the genera Leiostracus and Simpulopsis inside family Simpulopsidae, with the former genus being monophyletic, albeit with low support.

Outgroup
Most noticeably, in our phylogeny the genera Odontostomus and Pilsbrylia are placed outside of Orthalicoidea. Thus, an additional tree (with further species representing other stylommatophoran lineages) was built to further test their phylogenetic position. This second tree contained 30 terminal taxa, representing 28 species (Fig 7). The concatenated sequences (post trimming with Gblocks) included 1920 bp (COI: 641 bp; H3: 267 bp; ITS2+28S: 1012 bp). Despite the polytomies on the family-level relationships of Rhytidoidea (Fig 7), the family groups with the taxa of interest are all maximally supported (PP = 1). Both Odontostomus and Pilsbrylia are found not to be members of Orthalicoidea. The former genus clusters with species of Anthinus (Strophocheilidae) in a polytomy, while the latter is sister to a representative of the Clausiliidae (PP = 1 in both cases). The † indicates sequences from [30]. https://doi.org/10.1371/journal.pone.0288533.t003 Strophocheilidae (PP = 1) is resolved as paraphyletic. Rhytidoidea is also found to be paraphyletic but recent studies have shown that this superfamily (Rhytidoidea sensu lato) is presently an arrangement of convenience rather than an actual monophyletic lineage (e.g., [29]). Considering that Strophocheilidae is presently classified within Rhytidoidea sensu lato [2], we provisionally interpret Odontostomidae as also belonging to this group; however, further studies are needed to clarify the classification of Odontostomidae and Strophocheilidae.

Orthalicoidea
The superfamily Orthalicoidea is a maximally supported clade (PP = 1; Figs 1 and 7). It is composed of two main clades: a maximally supported clade containing two families, Orthalicidae and Amphibulimidae (PP = 1), and a weakly supported clade containing all other families (PP = 0.69).
In the clade (Orthalicidae + Amphibulimidae), both families are monophyletic although without sufficient nodal support (PP = 0.55 each; Figs 1 and 2). However, the relationships within Orthalicidae remained essentially unresolved while those in Amphibulimidae were partly poorly supported. Most genera that were represented by more than one species were recovered as monophyletic (PP = 1, except for Orthalicus: PP = 0.92). The exception was Plekocheilus, which decayed into two deeply divergent clades, one of these containing P. nebulosus. However, Plekocheilus is a rather speciose genus, and an increased taxon sampling is needed before systematic conclusions should be drawn. The semi-slugs Amphibulima and the slugs Gaeotis and Pellicula cluster together in a polytomy, albeit without sufficient nodal support (PP = 0.54).
The second main clade, containing the rest of the Orthalicoidea, lacks sufficient nodal support (PP = 0.69). In this clade, Paeniscutalus crenellus forms the sister group of all remaining Orthalicoidea. However, the monophyly of these remaining Orthalicoidea is maximally supported (P = 1). In this clade, Megaspira is the sister lineage of all other taxa; a position that renders Megaspiridae as currently delineated non-monophyletic (Figs 1 and 3). The successively next more derived offshoot containing Thaumastus and Kora is also not well supported (PP = 0.68).
The sister group to the "Megaspiridae" clade Thaumastus+Kora is strongly supported (PP = 0.99). It contains a well-supported family-level clade (PP = 0.99), sister to the remainder Orthalicoidea, that includes the genus Anostoma as sister to a clade formed by Hyperaulax and Tomigerus (Figs 1 and 3), all of which are strongly supported (PP = 1). This clade is an assemblage of "orphaned" taxa that were previously classified in Odontostomidae and is herein recognized as a new family within Orthalicoidea. The name Tomogeridae is available for this clade (see Discussion below).
The next node (PP = 0.97) has Bothriembryontidae (PP = 1; Figs 1 and 4) as sister to the remainder Orthalicoidea. Interestingly, a species previously identified as Rhinus taipuensis, from northeast Brazil, is the sister taxon to all other bothriembryontids. Rhinus is member of the Simpulopsidae, so a new genus is described below to allocate this species (see below). In the clade comprising the remainder of Bothriembryontidae, there is a branch formed by the African Prestonella (PP = 1), sister to a polytomous clade that includes the South American Plectostylus and Discoleus and the bothriembryontids from Oceania. This renders the subfamily Prestonellinae paraphyletic. The Australasian Bothriembryontinae is monophyletic (though containing Placostylinae; PP = 0.94 ; Figs 1 and 4), containing the Australian Bothriembryon (recovered as paraphyletic) and a well-supported Pacific clade (PP = 1). The latter contain one branch from the Solomon Islands (PP = 1, with both genera being polyphyletic) and a branch with low support (PP = 0.7) comprising species from Vanuatu, New Caledonia, and New Zealand.

Crown Orthalicoidea
The crown Orthalicoidea is a maximally-supported clade (PP = 1; Fig 5), sister to Bothriembryontidae, restricted to the Americas and containing three families, one of which is the extremely diverse Bulimulidae. The most basal node in crown Orthalicoidea has Simpulopsidae (PP = 1) as sister to the rest. Simpulopsidae contains the genera Rhinus, Leiostracus and Simpulopsis (Fig 5); the latter two are non-monophyletic. The type species of Leiostracus (L. vittatus) is represented in our tree and its clade apparently excludes at least L. perlucidus from the genus. The type species of Simpulopsis (S. sulculosa) is also represented, belonging to a clade in a derived position, far removed from other supposed Simpulopsis spp.
The sister clade to Simpulopsidae (PP = 1) is formed by Bulimulidae and a new family. This new family has good support (PP = 0.95) and is composed of several genera that are "orphans" of Odontostomidae (Fig 5). There are no names available for this clade, so it is described below as Cyclodontinidae fam. nov. This family is seemingly composed of two main branches: the  The final clade is a weakly supported Bulimulidae (PP = 0.80), although support is much stronger (PP = 0.95) after the first internal node (Fig 6; sister group to the clade Otostomus + Cochlorina). The relationship between the other more-encompassing monophyletic clades (of subfamily level) within Bulimulidae is more uncertain, although most of them have good support (Fig 6). The first of these groups is formed by the genera Anctus, Oxychona and Leiostracus carnavalescus (PP = 1). The latter clearly does not belong to Leiostracus (Simpulopsidae), and a new genus is erected to house it (see below).
The next group in the Bulimulidae is the subfamily Bostrycinae (PP = 1), composed solely of the genus Bostryx in a strict sense, as it contains the type species B. solutus. The genus, as understood until the present, is a wastebasket taxon and further studies will show if many of its species must now be either classified elsewhere or whether some of the multitude of synonyms for Bostryx [31] will need to be elevated to genus level within this subfamily. The final two groups are potentially sister taxa (unsupported, PP = 0.54; Fig 6) and consist in the subfamilies Bulimulinae and Peltellinae.
Bulimulinae is strongly supported (PP = 1) and includes the genera Naesiotus, Protoglyptus, Rabdotus, and Bulimulus, as well as a few species previously classified in Bostryx. The genera

PLOS ONE
Naesiotus and Protoglyptus as currently understood are polyphyletic and include species previously classified in Bostryx, but the other genera are monophyletic (PP = 1). Bulimulus tenuissimus, often thought to be a species complex in southern Brazil, forms a tight clade with virtually no genetic distance (Fig 6).
Peltellinae is largely unsupported (PP = 0.70), so this group could represent a non-monophyletic assemblage. In our results, however, this subfamily contains the genera Scutalus, Pseudoxychona, Neopetraeus, and Drymaeus. Drymaeus, long thought to be a wastebasket taxon like Bostryx, is potentially monophyletic, pending the exclusion of D. vexillum and the inclusion of the slug genus Peltella. Inside Drymaeus, there are three distinct clusters representing some of its purported subgenera: (1) a clade with moderate support (PP = 0.94) containing D. elongatus, type species of subgenus Mesembrinus; (2) a strongly supported clade (PP = 1) containing D. inusitatus, type species of Antidrymaeus; and (3)  Diagnosis: Shell typically medium-sized, elongated fusiform. Teleoconch smooth or sculptured by axial ribs or irregular striae. Aperture typically displaying a columellar lamella plus a varying number (that can be zero) of teeth and knobs. Peristome typically reflexed, often thickened. Parietal callus sometimes thickened.
Discussion: The family Cyclodontinidae Salvador & Breure, fam. nov. is here established, containing a monophyletic assemblage of genera that were "orphaned" from Odontostomidae. The new family is understood as the sister taxon to the Bulimulidae (Fig 5). The taxon chosen as type genus was Cyclodontina (type species C. inflata (Wagner, 1827)), which is the oldest available generic name within the family.
Etymology: From 'alter' (Latin for 'alternate, other') and Rhinus, the genus where the species were previously classified. The connective particle 'o' is used to aid in pronunciation. Grammatical gender: masculine.
Diagnosis: Shell bulimoid, multiwhorled, relatively slender to wide (expanded body whorl), with low whorls and well-marked suture. Protoconch (~1½ whorl) sculptured with fine wrinkled (zig-zag appearance) granulated striae that become less regular and increasingly granulated towards end of protoconch; first ½ whorl of protoconch may appear initially smooth; transition to teleoconch clearly marked. Aperture ovate-elongated; peristome reflexed, including the columellar region over the umbilicus. Thin columellar fold may be present.
Discussion: The present phylogeny places the specimens of Rhinus taipuensis as the sister taxon to all other Bothriembryontidae (Figs 1 and 4), far removed from Rhinus in Simpulopsidae. As such, the new genus Alterorhinus is proposed here to house it. Furthermore, other species previously classified in Rhinus [32] are here transferred to this new genus, considering that they (1) share a very similar shell morphology to the type species, including the diagnostic characters [31,[33][34][35][36]; (2) have been considered related to the type species [34]; and (3) do not present any of the diagnostic morphological features of Rhinus (i.e., wide rotund shell, more fragile shell walls, periostracal hairs, a white spiral line on the medial region of the whorl).
Alterorhinus constrictus is known from Colombia, Venezuela, Guyana, and the Brazilian state of Roraima [31,33,[35][36][37][38][39]. The records of A. constrictus from northeastern Brazil [32,40] are likely mistaken, belonging instead to one or more of its congeners, which are all restricted to that geographical area. A. ovulum is known only from Pernambuco state [32]; A. rochai is known from Pará to Pernambuco states in Brazil [32]; A. suturalis is known from Ceará to Bahia states [32,41]; and A. taipuensis is known from Ceará and Rio Grande do Norte states [32]. Therefore, the genus is distributed from Colombia to northeastern Brazil.
Family  . Etymology: From 'sannio' (Latin for 'harlequin') and 'ostrakon' (Greek for 'shell'), referring to the typical red, white, and black coloration of the shell of the type species. Grammatical gender: masculine. Vernacular name: harlequin snails. Diagnosis: Shell bulimoid, medium-sized, multi-whorled (whorls growing regularly in size), with acuminated apex. Shell base colour white, but overall colour pattern widely varied, with multiple combinations of solid spiral lines and dotted spiral lines in the colours red, orange, yellow, brown, and black. Protoconch (~1½ whorl) sculptured by fine closely-packed sinuous axial riblets in upper portion and numerous fine spiral striae in lateral portion; transition to teleoconch clear. Whorl profile slightly convex; keel absent. Aperture medium-sized, oval. Peristome reflexed, partially covering narrow umbilicus.
Discussion: The present phylogeny places specimens of Leiostracus carnavalescus within Bulimulidae, in a clade containing representatives of the genera Oxychona (sister taxon) and Anctus (Fig 6), far removed from Leiostracus in Simpulopsidae. As such, the new genus Sanniostracus is proposed here to house it.
In their study describing Sanniostracus carnavalescus [42], the authors noted the morphoanatomical similarities between their new species and member of the family Bulimulidae. Nevertheless, other similarities with Leiostracus spp. led them to classify their species in the latter genus. Under the light of the present phylogeny, those can now be interpreted as superficial similarities (e.g., shell shape, black and white body pattern) and the differences to Leiostracus become clearer, the most prominent of which is the protoconch sculpture, which in Leiostracus consists of sinuous subsutural axial riblets that give way to spiral cordlets towards the median region of the whorl [31,43].
Likewise, the shared characters between Sanniostracus and Oxychona become starkly clear, even though the species in the latter genus all have wide conical and strongly keeled shells. The soft body is of similar shape, white in colour, with a wide black band on each side of the head, positioned immediately below the eye stalks and that can end in the median region of the foot or extend itself toward the "tail" [43,44]. The base colour of shells is white, and the colour pattern has variable solid and dotted lines of multiple earth-toned colours [32,43,44]. The radular teeth are similar in shape, with a small blunt rachidian tooth, spatula-like lateral teeth, and bicuspid marginal teeth bearing small acute ectocones posteriorly positioned at their base [31,42].
Remarks: S. carnavalescus is highly polymorphic in shell colour patterns [42]. However, the holotype and paratype (red morph) show a reasonable genetic distance between them (Fig 6), which might be indicative that they potentially represent two different sympatric taxa. Further molecular investigation using a series of additional specimens (also including the black morph; [42]) is advised.

Discussion
The results of the present phylogenetic analysis allow that several modifications (some major, some minor) be made to the classification of the Orthalicoidea. These are explained in the section that follows. Furthermore, the new framework also brings new biogeographical insights and has implications on how to interpret the fossil record; these topics are discussed in detail further below.

Classification
Here are delineated all the implications the present study has for the phylogeny of the Orthalicoidea. As in the Results section above, the taxa are treated in the order of the nodes starting from the root (Figs 1-6).
Pilsbrylia. The genus Pilsbrylia contains three species found in southeastern Brazil and northern Argentina [45]. While it bears superficial similarity in shell shape to crown orthalicoid genera such as Cyclodontina and Clessinia, our results showed that it does not belong in Orthalicoidea (Figs 1 and 4), as already suggested by [7]. Instead, it is closely related to the door-snails (Clausiliidae). While clausiliids haver narrow turreted shells of which Pilsbrylia is reminiscent, there is a major difference: the shells of door-snails are sinistral, while Pilsbrylia is dextral like the majority of the Stylommatophora. The exact placement of this genus in relation to the door-snails needs to be further investigated, particularly if it will cluster with South American Clausiliidae.
Odontostomidae. The representative of Odontostomus in our analyses forms a group with the Strophocheilidae genus Anthinus in a polytomy with the remainder Strophocheilidae (Fig 7). The Strophocheilidae are paraphyletic, pending the exclusion of Anthinus. To solve this, here we transfer the genus Anthinus to Odontostomidae; furthermore, we classify Odontostomidae within Rhytidoidea sensu lato.
This leaves the former "odontostomid" genera in Orthalicoidea without a family name. Those genera are divided into two unrelated branches within Orthalicoidea (see the entries below for Tomogeridae and Cyclodontinidae fam. nov.).
We have sequenced a single species of Odontostomus (O. paulistus) in the present analysis, which is not the type species of the genus (O. odontostomus (Sowerby, 1824)). Nevertheless, based on close morphological similarity we are confident that O. paulistus is a congener of O. odontostomus and thus, a good representative of the genus and family. We are likewise confident that it belongs in a different group than the Orthalicoidea, even though this result might seem unexpected at first sight. The shells of Odontostomus are unlike any other Orthalicoidea, including those previously classified as odontostomids. Rather, Odontostomus have sturdy shells reminiscent of Strophocheilidae and particularly similar to Anthinus [32,46]. In particular, the shells are medium to large sized, with strong walls, often with bulging whorls, a coarse teleoconch sculpture, a thick expanded peristome that typically covers the umbilicus, and an ochre to brown periostracum bearing in some cases an "camouflaged" pattern. The shells of members of the Strophocheilidae genus Gonyostomus (of which Anthinus was a subgenus) also share these features [32,46] and is, therefore, likewise here transferred to Odontostomidae.
Ongoing molecular studies by members of our group focusing on the Strophocheilidae will now also include the Odontostomidae to further investigate the relationship of these two families (and other Rhytidoidea sensu lato). Hopefully we will be able to elucidate if they are sister taxa, consecutive branches of the Rhytidoidea, or even synonymous. A recent morphological analysis [46] has placed Anthinus inside Strophocheilidae, but it did not include Odontostomus or Gonyostomus.
Orthalicidae & Amphibulimidae. Each of these families have no support, though together they form a strongly supported clade (Figs 1 and 2). As such, there is the possibility that one or both families are not monophyletic and further studies focusing on them are necessary. If that is confirmed, Amphibulimidae should be considered part of (and thus a synonym) of Orthalicidae. Breure & Romero [7] also recovered these families as sister taxa in the base of Orthalicoidea, although a monophyletic Orthalicidae was likewise unsupported.
By the confusion within the genus Orthalicus (Fig 2), it can be seen that this genus would benefit from a revision. Likewise, the paraphyletic genus Plekocheilus is also in need of a revision, and some subgenera (and/or synonymized names) might need to be elevated to genera. The Paleogene monotypic genus Cortana from Rio de Janeiro, Brazil, was interpreted to be related to Eudolichotis Pilsbry, 1896 (presently considered a synonym of Plekocheilus and represented by P. lacerta in our phylogeny; Fig 2), but classified in the Bulimulidae [4,47]. Thus, Cortana is here transferred to Amphibulimidae.
The semi-slugs Amphibulima and the slugs Gaeotis and Pellicula seem to form a monophyletic group, although with no support, with a polytomy (Fig 2). This group, however, makes sense from an evolutionary perspective, considering that it is more parsimonious that the shell was reduced only once and then internalized in Gaeotis and Pellicula, which might have happened once or twice depending on how the polytomy is solved. Furthermore, the species Gaeotis flavolineata, which is typically considered a junior synonym of G. nigrolineata (e.g., [48]), was shown to be genetically distant from the latter (Fig 2). This result supports G. flavolineata as a distinct species.
"Megaspiridae". Despite this being a paraphyletic assemblage (Figs 1 and 3), it remains a useful group, so we will retain it here for stability. This result is not unexpected, as this family unites species with very distinct shell morphologies (most notably the turreted Megaspira). Breure & Romero [7] observed a monophyletic Megaspiridae, albeit with low support, and suggested that this group was a basal relict group of Orthalicoidea, which is not exactly the case given the position of Orthalicidae + Amphibulimidae (Fig 1).
We confirmed the placement of Paeniscutalus within this family, as provisionally suggested by [49]. The eastern Brazilian genus Kora, previously classified in Bulimulidae [50], is shown here to be closely related to Thaumastus, and is thus transferred to family "Megaspiridae".
Tomogeridae. This branch is one of the two monophyletic groups in the phylogeny that is an "orphan" of Odontostomidae, where its genera and species were previously classified (Figs  1 and 3). [7] had no representatives of this group in their study.
Given its position in the tree, this clade should be recognized as a family-level taxon. Fortunately, there is already a name available in the literature for it: Tomogeridae Jousseaume, 1877. Jousseaume [51] proposed the family among the "Bulimus" to allocate the genera Tomogeres Montfort, 1810 and Tomogerina Jousseaume, 1877 (both synonyms of Anostoma Fischer von Waldheim, 1807).
The genus Anostoma, despite its strikingly different morphology with an apically bent body whorl, was later conflated with other Orthalicoidea whose shells had apertural barriers (the Odontostomidae, as understood until now). The present molecular phylogeny supports Jousseaume's assertion [51] that Anostoma lineage represents a separate and unique family. Our study also includes two other genera in this family: Tomigerus Spix, 1827 and the island endemic Hyperaulax Pilsbry, 1897. The species in both genera have similarities in their shells that indicated their affinity and their sister-taxa relationship proposed by [10] is fully supported here (Fig 3). Likewise, we propose here that the genera Biotocus Salgado & Leme, 1990 and the cave endemic Clinispira Simone & Casati, 2013, which have shells closely resembling Tomigerus [32,52,53], also belong in this resurrected family.
Bouchet et al. [2] argued that a petition to the ICZN was necessary to conserve the younger name Odontostomidae Pilsbry & Vanatta, 1898 over Tomogeridae Jousseaume, 1877. However, that is not necessary, as it has been shown here that the two families are two completely distinct lineages (see also the entry for Odontostomidae above).
Bothriembryontidae. Bothriembryontidae is monophyletic (Figs 1 and 4), and its branches closer to the family's basal node comprise South American and African species: the non-monophyletic subfamily "Prestonellinae", which we maintain here because it is a useful grouping. The subfamily Bothriembryontinae is paraphyletic pending the including of Placostylinae; as such, here we consider Placostylinae a synonym of Bothriembryontinae.
The paraphyletic Prestonellinae was also observed in [7], although those authors had monophyletic Bothriembryontinae and Placostylinae (with low support), likely due to the smaller sample of taxa.
The presence of this family in South America was until now considered relict, with just the genera Plectostylus and Discoleus in the southern and western regions of the continent. However, Alterorhinus gen. nov. was revealed to be the sister taxon to all other Bothriembryontidae; its species presently inhabit a large area in northern and eastern South America, showing that the family is widespread in the continent and hinting that it was potentially more diverse in the past.
The Australian genus Bothriembryon, as currently understood, is not monophyletic and would benefit from a revision. The two genera from the Solomon Islands, Eumecostylus and Placocharis are polyphyletic. Considering that there are no consistent diagnostic characters separating these two genera [54], here we consider Placocharis a junior synonym of Eumecostylus.
Maoristylus (from New Zealand and Lord Howe Island) has been considered either as a subgenus of New Caledonian Placostylus or its synonym, but it was recently shown to be a distinct genus [55]. These two genera have thus been considered as sister taxa, albeit in works with low taxon coverage (e.g., [56]). The inclusion of Vanuatuan Diplomorpha in the present study has shown that is not necessarily the case, as it clusters (with low support) with the New Caledonian Placostylus.
Simpulopsidae. This is a well-supported family and most of its internal relationships are resolved with good support (Fig 5). This result confirms the findings of [7], who redefined the family Simpulopsidae and established it as the sister to the remainder crown Orthalicoidea.
Still, the genera Leiostracus and Simpulopsis are not monophyletic and need a thorough revision. Two further genera, even though not represented in our study, likely belong in this family as well: Lopesianus Weyrauch, 1958 and Eudioptus Martens, 1860 (see discussion in [57]). Simpulopsidae is neotropical, with only a few species present in the sub-tropical areas of South America (e.g., [32,58]).
Cyclodontinidae fam. nov.. This newly established family has a somewhat resolved internal structure, although most clades have low support (Fig 5). While some genera were recovered as monophyletic (Clessinia, Plagiodontes and Bahiensis), the type genus Cyclodontina as presently understood is polyphyletic, including the genera Moricandia and a likewise polyphyletic Burringtonia. The type species of Cyclodontina, C. inflata, could not be included in our analysis; C. inflata is conchologically distinct from other congeners present in the tree, so we cannot speculate where it would be placed. As such, the restructuring of the genus Cyclodontina (and Moricandia and Burringtonia) will remain for a future study. Cyclodontinidae is endemic to South America and restricted to Bolivia, Brazil, Paraguay, Uruguay, and Argentina [49,54].
Notably, [7] recovered a similar clade (they named it Odontostomidae, following the then current practice and considering that they did not have a representative of Odontostomus), albeit with the genus Bahiensis being place outside of it and as sister taxon to Bulimulidae. In our tree (Fig 5), with a larger sampling, Bahiensis is placed within Cyclodontinidae; its long branch is likely due to the lack of COI sequences for both its species (Table 2).
Bulimulidae. The Bulimulidae are weakly supported when considering that it includes the branch formed by Otostomus and Cochlorina (Fig 6), two genera defined by their shell's barrel-like body whorl and highly modified aperture. Given their morphological similarities (e.g., [32]), their close relationship in the phylogeny makes immediate sense. Excluding this branch, the remainder of Bulimulidae is well supported (Fig 6).
The first branch includes Anctus, Oxychona and Sanniostracus gen. nov. As shown above, despite the obvious difference in shell shape, the latter two share many conchological and morphoanatomical features. The next branch of Bulimulidae is the subfamily Bostrycinae, containing only the newly circumscribed genus Bostryx.
The Bulimulinae includes Naesiotus, Protoglyptus, Rabdotus, and Bulimulus. Naesiotus and Protoglyptus, as currently understood, are non-monophyletic and need further investigation. We could not include their type species Naesiotus nux (Broderip, 1832) and Protoglyptus pilosus (Guppy, 1871) in our analysis, so the actual placement and composition of these genera remain uncertain. Nevertheless, considering that we can better define Bostryx based on its type species as explained above, we preliminarily transfer some of those "Bostryx" spp. within Bulimulinae to the genus Naesiotus (Table 4).
Peltellinae contains the genera Scutalus, Pseudoxychona, Neopetraeus and Drymaeus. The close relationship of Pseudoxychona to Drymaeus was expected considering their similarly reticulated protoconchs, despite the similarities of the shells between Pseudoxychona and Leiostracus (Simpulopsidae). Considering this new finding, one species previously described by two of the present authors [59] is herein transferred to Pseudoxychona: Pseudoxychona faerie (Salvador & Cavallari, 2014) comb. nov.
Drymaeus is potentially monophyletic, though it might need the exclusion of a few species such as D. vexillum. A revision of this genus with more representatives (and its type species, D. hygrohylaeus (d'Orbigny, 1835)) is advised, considering that some of its subgenera and/or synonymized names (e.g., Leptodrymaeus) might have phylogenetic reality and help to achieve a better classification. For instance, the monophyletic clades representing subgenera Mesembrinus (type species D. virgulatus) and Antidrymaeus (type species D. inusitatus) are elevated here to the genus level (Fig 6).
The genus Peltella of bright-colored slugs is contained within Drymaeus (as already suggested by [7]) and thus considered its synonym. It is an interesting case of limacization; only two instances of such phenomenon occurred in Orthalicoidea (the other being the Amphibulimidae clade Amphibulima + Gaeotis + Pellicula; Fig 6).
The relationships between Bostrycinae, Bulimulinae and Peltellinae are unclear, as the support values are low (even though the subfamilies Bostrycinae and Bulimulinae are each strongly individually supported). Overall, they have a similar arrangement to the Bulimulidae recovered by [7], although the sister-taxon relationship between the groups varies. Of the three subfamilies, Peltellinae has the lowest support (including its inner nodes), which hints at a potential non-monophyletic group. In any event, it is uncertain whether a division in subfamilies is really needed in Bulimulidae, particularly considering that Bostrycinae is monotypic at present. Furthermore, that could raise the need of the other branches to be recognized as further subfamilies, which would cause unnecessary taxonomic inflation.
Vidaliellidae: As mentioned above, the placement of the extinct Vidaliellidae in the Orthalicoidea [3] was based on superficial and incomplete conchological comparison and without fully considering the biogeographical history and chronology of the taxa involved.
The size and shape of the shells, their multiwhorled spire that is comparatively short when compared to the expanded body whorl, the aperture shape, the strongly reflexed peristome, and the presence of a strong parietal callus, are all features shared with the strong-shelled Rhytidoidea such as the southern African Dorcasiidae and Malagasy Clavatoridae (and even the South American Strophocheilidae) [54,60]. In particular, the fossils are strongly reminiscent of Clavator Martens, 1860 and Leucotaenius Martens, 1860, to the point that several Vidaliellidae species were previously classified in those genera ( [3] and references therein).
The link to the Malagasy Clavatoridae was considered biogeographically unlikely by [3]; yet those authors suggested a classification in Orthalicoidea, which is even more unlikely, from biogeographical, chronological, and conchological standpoints. As such, here we provisionally classify the Vidaliellidae in the Rhytidoidea sensu lato. By considering Vidaliellidae part of an African lineage of rhytidoids instead of having a South American ancestry, the presence of this family in the Paleogene of northern Africa becomes less aberrant. The European fossils of Vidaliellidae, however, need to be reassessed. While some may in fact represent a branch of this family that extended into Europe when the climate was warmer [3], others seem more closely related to typical European taxa such as the Filholiidae (e.g., [61,62]).

Paleobiogeography
The superfamily Orthalicoidea likely originated in South America and was already present there in the Late Cretaceous [4,5,7]. Notably, the family Bothriembryontidae is the only truly Gondwanan lineage known so far, with representatives in South America, southern Africa, and Australia (and later New Zealand and a few other Pacific Islands). Thus, the origin of Orthalicoidea might lie in older times in the Cretaceous.
However, the Cretaceous orthalicoid fossils need to be revised considering the new phylogenetic framework and classification proposed herein. Several of those fossils have been assigned to modern genera in crown Orthalicoidea such as Bulimulus and Bahiensis [4,5], but they likely belong to lineages that appeared earlier, such as Orthalicidae and "Megaspiridae". For instance, the monotypic genus Cortana from the Late Paleocene [4,47] is now considered a member of the Amphibulimidae rather than Bulimulidae, as explained above.
By the Late Paleocene, all families of Orthalicoidea were already established, as evidenced by the fossils of Cyclodontinidae and Bulimulidae from Itaboraí Basin, Rio de Janeiro, Brazil [4,47]. In all likelihood, the different branches of this superfamily (Orthalicidae, Amphibulimidae, Simpulopsidae, and Bulimulidae) spread to Central America and the Caribbean long before the Isthmus of Panama formed around 3 million years ago [63] and physically connected the continents.

Conclusion
Based on the present results, we propose the following revised classification of superfamily Orthalicoidea Martens, 1860 (Fig 8). Quotation marks are used to indicate non-monophyletic https://doi.org/10.1371/journal.pone.0288533.g008 (sub)family-level assemblages. We list only those genera that are represented in our analysis or that have been otherwise discussed above. A summary of species transferred to other genera and the new combinations can be seen in Table 4.