(Gastropoda: Stylommatophora) Abstract

Transcription

1 Arch. Molluskenkunde 141 (1) figures Frankfurt am Main, Support and surprises: molecular phylogeny of the land snail superfamily Orthalicoidea using a three-locus gene analysis with a divergence time analysis and ancestral area reconstruction (Gastropoda: Stylommatophora) Abraham S.H. Breure & Pedro E. Romero Abstract Phylogenetic relationships of Orthalicoidea, a highly diverse and dominant element in the Neotropics, were studied using nuclear and mitochondrial DNA sequences (ITS2/28S, CO1, H3). Specimens of various locations from the Southern Hemisphere (South America, Africa, Australia, New Zealand, Solomon Islands) were analysed (74 taxa, representing 30 genera). Our results support previously presented hypotheses, but also give surprises in terms of unexpected topologies. Phylogenetic trees were estimated using maximum likelihood and Bayesian inference, and compared with traditional classifications. Phylogenetic estimations using three loci gave a strong support for monophyly of Orthalicoidea, as well as for some clades within this group (Bulimulidae, Bothriembryontidae, Orthalicidae, Amphibulimulidae), but not for others (Odontostomidae and Megaspiridae). In the resulting revision of the classification scheme of the Orthalicoidea, the tribe Simpulopsini is raised to family rank. One new subfamily is recognized, the Bostrycinae. The family Bulimulidae Tryon 1867 is retained under ICZN Art. 35.5, despite the senior synonymy of Peltellinae Gray Our analysis supports an origin of the Orthalicoidea in South America, with subsequent radiations into other parts of the Neotropics and the Southern Hemisphere. The hypothesis that the distribution on the southern continents may be explained by vicariance due to break-up of Gondwana is only partially supported by divergence time analysis using fossil calibrations. Ancestral area reconstruction suggests various independent dispersals out of South America into Central America and the West Indies, and possibly two independent dispersals to explain the remaining relations between groups of taxa on the southern continents. Divergence time analysis further shows that the major diversification of extant taxa within the superfamily may have started around the Cretaceous-Paleogene boundary. Keywords: Mollusca, taxonomy, ribosomal RNA, histone 3, cytochrome oxidase I. Introduction The land snail group Orthalicoidea is a highly diverse and dominant element in the Neotropics, with currently approximately 68 genera (Schileyko 1999) and more than 1700 available taxon names (Richardson ). Moreover, members of the group occur in South Africa with the relict genus Prestonella (Herbert 2007), in Australia with the genus Bothriembryon (Stanisic & Solem 1998), and in New Zealand and several Australasian island groups with the family Placostylidae (Delsaerdt 2010; Neubert et al. 2009). Phylogenetic data on this group are still scarce, due to the paucity of suitable samples in museum collections. Author s addresses: Dr Abraham S.H. Breure, Netherlands Centre for Biodiversity Naturalis, P.O. Box 9517, NL-2300 RA Leiden, The Netherlands; ashbreure@gmail.com. Pedro E. Romero, Museo de Historia Natural, Apartado , Lima-14, Peru; quipu.romero@gmail.com. E. Schweizerbart sche Verlagsbuchhandlung (Nägele u. Obermiller), 2012, ISSN DOI /arch.moll/ /141/

2 2 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea Fig. 1. Bayesian phylogeny for the Orthalicoidea as a result of a BEAST analysis, based on a 2079 bp multi-locus data set of mitochondrial and nuclear DNA. Posterior probabilities of 0.9 or above shown at left side of nodes. Scale bar in substitutions/site. Summary of clades recognized in the present study (in parentheses corresponding ones of Breure et al., 2010): A (A) genera Bostryx, Bulimulus, Drymaeus, Naesiotus, Peltella, Rabdotus, and Scutalus; B (B) genera Bahiensis, Clessinia, Cyclodontina, Plagiodontes, and Spixia; C (-) genera Leiostracus and Simpulopsis; D (C) genera Bothriembryon, Discoleus, Eumecostylus, Placostylus, Placocharis, Plectostylus and Prestonella; E ( ) genera Megaspira and Thaumastus; F (D1) genera Corona, Orthalicus, Porphyrobaphe, and Thaumastus; G (D2) genera Gaeotis and Plekocheilus. * denotes the basal taxon Thaumastus crenellus. Wade et al. (2001) were the first to present a phylogenetic tree based on ITS2/28S gene markers, with one species of Orthalicoidea represented (Placostylus, then classified with the Bulimulidae; Richardson 1995). They showed that this taxon was included in their non-achatinoid clade with Leucotaenius (Acavidae) and Elasmognatha (Succinea, Athoracophorus) as sister group. In a more extensive study, Wade et al. (2006) added five additional species of the Orthalicoidea, covering the genera Placostylus, Bulimulus, Drymaeus and Gaeotis. The latter genus was placed in the Amphibulimidae, the others in the Orthalicidae. In their tree the Orthalicoidea appear to be monophyletic with support values of 86% bootstrap in Neighbour-Joining (NJ) analyses. Gaeotis branched off basally and the Orthalicidae had support of 49% bootstrap in NJ; Placostylus appears basal to Bulimulus and Drymaeus. Herbert & Mitchell (2009) added two species of the African genus Prestonella and one species of Australian Bothriembryon to a subset of the results of Wade et al. (2006). In their resulting tree Gaeotis is basal to the rest of the Orthalicoidea, with the Prestonella species as a strongly supported sister-group (posterior probability (PP) = 1.0) to a clade of Bothriembryon and Placostylus. Bulimulus and Drymaeus appear as a strongly supported (PP = 0.99) sister-group of the clade with Prestonella, Bothriembryon and Placostylus. In a study of Peruvian Bostryx and Scutalus, Ramírez et al. (2009) used a different genetic marker (16S rrna). Their results indicate that Placostylus is more basal than the two Peruvian genera. In a recent study using ITS2/28S, Breure et al. (2010)

3 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea 3 Fig. 2. Likelihood mapping of the combined-gene data set showing the distribution of quartet analyses, with fully resolved quartets in the corners of the triangle, unresolved quartets in the centre and partially resolved quartets (as a result of conflict between alternate topologies) along the sides. The data have a high phylogenetic signal, with 90.4% of quartets fully resolved. added 20 more species, covering all major groups within the Orthalicoidea. Their results showed five monophyletic groups, corresponding to the Amphibulimulidae, Orthalicidae, Placostylidae (s.l.), Odontostomidae and Bulimulidae. They concluded that the Coelociontidae, postulated to be part of the Orthalicoidea by Uit de Weerd (2008), had to be excluded from this group. The Megaspiridae, another possible candidate for inclusion in this superfamily, could not be analysed due to lack of data. Other studies using phylogenetic methods in the Orthalicoidea focused on the relationships at species level. Ponder et al. (2003) studied the relationships between Placostylus species from Lord Howe Island and New Zealand. Parent & Crespi (2006) investigated the radiation in Galápagos Bulimulus [sic, Naesiotus]. Trewick et al. (2009) studied the diversity in New Caledonian Placostylus. They all used CO1 and 16S as genetic markers. The classification of this group is still unstable, especially at the (sub)familiar level, and is hitherto based on shell morphology and anatomy. Orthalicoid shells are generally elongate-ovate, but shell shape is rather diverse and several extreme examples are known showing cylindrical to fusiform shells (Spartocentrum and Berendtia in Baja California: Christensen & Miller 1975; some Bostryx species: Breure 1978, 1979) or discoid shells (Breure 2008). Breure (1979) used a limited number of morphological traits in a cladistic phylogenetic reconstruction of the Orthalicoidea (then classified as Bulimulidae; see Breure et al. 2010). Neubert et al. (2009) questioned this character selection and the polarity (as plesio- and apomorphic) used. They called for a re-evaluation using more precise anatomical data. In a recent, preliminary morphological cladistic study of part of the family, Cuezzo et al. (2010) corroborated hypotheses of both Breure (1979) and Herbert & Mitchell (2009) about the Gondwanan relationships of a number of genera. In this study we will present the results of ongoing phylogenetic research on this superfamily, which so far yielded both support for previously presented hypotheses as well as surprises in terms of unexpected topologies (Breure et al. 2010). The resulting molecular phylogenetic framework is compared to traditional morphology-based classifications, providing new insights in the monophyly at several levels within the Orthalicoidea and shedding new light on the position of some taxa. Our central hypothesis, to be tested in this paper, is that the distribution of this group of Gondwanan genera (Bothriembryon, Placostylidae sensu Neubert et al. (2009), Discoleus, Plectostylus, and Prestonella) can be explained by vicariance caused by the break-up of that continent. This leads to the following derived hypotheses (see further Discussion): (H1) The time of divergence between the New Zealand and Melanesian Placostylidae sensu Neubert et al. (2009), and other taxa of the group is around 80 Ma; (H2) The time of divergence between Australian Bothriembryon and the South American Discoleus and Plectostylus is around 30 Ma; (H3) The time of divergence between African Prestonella and the South American taxa is 90 Ma or more. The hypotheses of Breure (1979) for the relationships of different groups of genera in other parts of the superfamily remain, however, untested as yet. Taxon sampling Material and methods Material sequenced in this study was partly collected during a field trip in Peru in March 2010; additional material was kindly supplied by colleagues. All material is listed in Table 1 with accession numbers for museum vouchers, localities and GenBank accession numbers. All tissue samples were taken from the tail of snail feet and fixed in 96% ethanol. The total number of Orthalicoid taxa represented is 74 (compared to 30 in Breure et al. 2010), representing 30 genera. Taxon sampling was not evenly distributed, but all the families recognized by Breure et al. (2010) were represented (including Megaspiridae). Species identifications were made using the shell morphology and generic assignments follow Breure (1979) and Breure & Schouten (1985). Type species were included whenever possible, representing nine (sub)genera. DNA extraction, amplification and sequencing Whole genomic DNA was extracted from tail tips of snail feet with a DNeasy kit (Qiagen, Inc.) following the manufacturer s protocol for animal tissues.

4 4 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea Fig. 3. Maximum clade credibility tree from relaxed phylogenetic analysis in BEAST. The timescale is in millions of years before present. Circles represent geographical areas of ancestral taxa as inferred by maximum likelihood (ML) reconstruction. The size of the slice indicates the ML probability of that state if it is less than Colours refer to area codes: AAC, Atlantic Forest, Amazonia, and Chaco of Brazil, Paraguay, and Argentina (orange); ASA, Andean South America, including Guayana Highlands (green); AUS, Australia (grey); CAC, Central America, Caribbean (dark blue); NZM, New Zealand and Melanesia (grey-blue); SAF, South Africa (purple); SSA, Southern South America (Patagonia and Chile south of 25 o S; red). For nodes I-VIII see section on Molecular age and divergence times (below). Three gene fragments were amplified in 25 μl reactions. Fragments of mitochondrial cytochrome oxidase 1 (CO1) were amplified using the Folmer primers (Folmer et al. 1994). Fragments of histone 3 (H3) were amplified using primers H3pulF 5 -GGAGGCAAGGCCCCACG- TAARCA -3 and H3pul3 5 -TTGGCGTGGATGGCG- CACARG -3 (Uit de Weerd, in prep.). Fragments of the nuclear ITS2 and 28S rrna (hereafter 28S) were amplified using the primers LSU1 3 and LSU 2 5 (Wade & Mordan 2000). Reactions consisted of 2.5 μl of Qiagen PCR buffer, 0.5 μl of 10 mm dntps, 1 μl each of forward and reverse 10 mm primers, 0.25 μl Taq DNA Polymerase (Qiagen Inc.), 1 μl of template DNA, and water to 25 μl. Reaction conditions included an initial denaturation step at 94 C for 3 min, followed by cycles of 94 C for 15 s, an annealing of 50 C (COI / 28S) or 57 C (H3) for 30 s and 72 C for 40 s. PCR products were purified, and then sequenced under BigDye terminator cycling conditions, purified by ethanol precipitation, and run on an Applied Biosystems 3730xl sequencer by Macrogen. All sequences were checked for contamination using a BLAST search. Some sequences have been previously published (Wade & Mordan 2000; Herbert & Mitchell 2009; Breure et al. 2010); new sequences obtained during the present study are indicated in Table 1. Phylogenetic analysis Forward and reverse sequences of each gene were assembled and edited with Sequencher 4.7 (GeneCodes Co., Ann Arbor, USA). Accession numbers of all sequences are listed in Table 1. Alignments were obtained with CLUSTALX (Larkin et al. 2007) and manually corrected with MacClade 4.08 (Maddison & Maddison, Sinauer Associates Inc.). Poorly aligned positions and divergent regions were eliminated by submitting the aligments to the Gblocks server (v. 0.91b, Castresana 2000; Talavera & Castresana 2007).

5 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea 5 Fig. 4. Maximum clade credibility (MCC) chronogram inferred using the non-autocorrelated model of rate evolution in BEAST, with most clades collapsed to highlight the Gondwanan clade. A G refer to the clades mentioned in the text; * denotes the basal taxon Thaumastus crenellus. Colours at right-hand side refer to areas: green, South America; light-blue, Africa; purple, Australia; orange, New Zealand and Melanesia. Time is given in millions of years before present. The vertical dashed red line indicates the K Pg boundary. The red-shaded bar indicates the separation of Zealandia from the remainder of Gondwana (based on Crisp et al., 2011); the yellow-shaded bar indicates separation of Australia, Antarctica and South America. Blue error bars represent 95% posterior credibility intervals and are only given for nodes that were present on more than 50% of the posterior sampled trees. See text for further explanation. Phylogenetic reconstructions were executed with Maximum Likelihood (ML) and Bayesian methods (Mr- Bayes (MrB) and BEAST (B) respectively), both on individual gene partitions as well as on the combined partitions set (CPS). Bayesian inference was used to obtain posterior probabilities for the nodes in the trees, using MrBayes (v , Ronquist & Huelsenbeck 2003). All sequences were analyzed using the nucleotide substitution model GTR+Gamma+Invar selected by jmodeltest (Posada 2008), with four categories. MCMC runs were performed with a chain length of 10 7, and two chains (two heated t=0.1, 0.3, one cold), sub-sampling frequency 10,000, a burn-in length of 100,000, and unconstrained branch lengths. The results of MrBayes were used to explore the CPS data with BEAST (see below) to obtain an improved topology. In all analyses, an outgroup was chosen consisting of two Megalobulimus species (Megalobulimidae), Coelocion australis (Coelociontidae), and Leucotaenius proctori (Acavidae), with the latter as root in 28S analyses and M. perfragilior in the CO1 and H3 analyses.

6 6 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea Fig. 5. Maximum-likelihood phylogeny for the Orthalicoidea, based on ITS2/28S ribosomal RNA varying in length between 1098 and 1154 bp. Bootstrap values of 90 and above are presented to the left of the nodes. Scale bar in substitutions/site. Phylogenetic signal detection The amount of phylogenetic signal present in a data set is reflected by the percentage of fully resolved quartets, as opposed to unresolved or partially resolved quartets, with the use of quartet puzzling and likelihood mapping (Schmidt et al. 2002; Strimmer & von Haeseler 1996). This method, as implemented in Tree-Puzzle v5.2 (Schmidt et al. 2002), investigates the support of internal branches without forcing a tree and graphically visualizes the phylogenetic signal present in a data set. By assigning sequences to clusters, likelihood mapping allows for exploration of possible relationships and alternative topologies (Schmidt & von Haeseler 2007; Strimmer & von Haeseler 1997). Overall trees were tested using estimation of maximum likelihood branch lengths and likelihood values, with Expected Likelihood Weights (Strimmer & Rambaut 2002) implemented in Tree-Puzzle v5.2. This program uses a one-side Kishino & Hasegawa (KH) test, with pairwise Shimodaira & Hasegawa (SH) tests, to correct for the fact that the SH-test takes into account that any tree in the test set could be the maximum likelihood tree. Ancestral area reconstruction The combined partitions set (CPS) was used during BEAST v1.6.1 (Drummond & Rambaut 2007) analyses to estimate divergence times, using only nucleotide data. The input file was created using BEAUti v1.6.1 (Drummond & Rambaut 2007) using the same model as in MrBayes (GTR+Gamma+Invar) and the gamma distribution modeled with four categories and estimated base frequencies. A relaxed molecular clock was used

7 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea 7 with rates for each branch drawn independently from a lognormal distribution (Drummond & Rambaut 2007). The tree model used a normal distribution for the birthdeath meangrowthrate and relativedeathrate priors (initial = 1), and a randomly generated starting tree. After several short runs and a run sampling from priors only, the MCMC operators were tuned to the suggested output diagnostics. Finally, two BEAST runs were performed of each 50,000,000 generations. Convergence was checked using Tracer v1.5 (Drummond & Rambaut 2007). After discarding of a 10% burn-in, the trees and parameter estimates were combined and down-sampled to 20,000 trees using LogCombiner v1.6.1 (Drummond & Rambaut 2007). The samples from the posterior were summarized using TreeAnnotator v1.6.1 (Drummond & Rambaut 2007), using a maximum clade credibility (MCC) tree, a posterior probability limit of 0.5, and mean node height as settings. FigTree v1.2.3 (available at was used for visualization. After BEAST analysis, the fully-resolved MCC tree was used for Maximum Likelihood (ML) ancestral area reconstruction (AAR), using MESQUITE (Maddison & Maddison 2006) with Ancestral State Reconstruction package installed and the Markov k-state one-parameter model (Lewis 2001) implemented. Areas were coded as multistate character in seven different states: AAC, Atlantic Forest, Amazonia, and Chaco of Brazil, Paraguay, and Argentina; ASA, Andean South America, including Guayana Highlands; AUS, Australia; CAC, Central America, Caribbean; NZM, New Zealand and Melanesia; SAF, South Africa; SSA, Southern South America (Patagonia and Chile south of 25 o S). Molecular age and divergence times The use of multiple, well-supported fossil dates for calibration is ideal, particularly if these fossils are sequential along a lineage (Hug & Roger 2007). The earliest fossils that have been referred to the other families within the Orthalicoidea date from the Tertiary (Breure 1978; Brito 1967; Miquel & Bellosi 2010; Parodiz 1969; Roth & Megaw 1989). They are classified with the extant genera Thaumastus, Plagiodontes, and Rabdotus, and the fossil genera Paleobulimulus and Itaborahia. Fossils attributed to the three extant genera all originate from as early as the Eocene, implying that nodes I, VI, and VIII (Fig. 3) should be placed in or before that period. Paleobulimus and Itaborahia species have been dated as Eocene and Miocene respectively. The former genus has been related to Lissoacme (Parodiz 1969), which has been considered a synonym of Bostryx s.l. by Breure (1979); it cannot be associated with any of the species in this study. Itaborahia species show a protoconch sculpture that could be associated with either Odontostomidae (e.g., Spixia) or Bulimulidae (e.g., Naesiotus). It is therefore difficult to assign this genus to a specific branch in the current tree. Calibration dates for the minimum ages of the fossils were set to 45 Ma for the clades (corresponding to Fig. 3 nodes I, VI, and VIII) derived from analyses in the previous BEAST runs; these taxon sets were constrained to be monophyletic in BEAUti, and the stem was defined not to be included in the tmrca. BEAST was run for 25,000,000 generations, the conversion was checked in Tracer v1.5 (Drummond & Rambaut 2007) and the result was visualized with FigTree v Results Sequences For 71 taxa new sequences were obtained (52 CO1, 60 H3, and 49 28S additional to those obtained by Breure et al. 2010). CO1 sequences were 654 bp long, no indels found; H3 sequences were 267 bp long, and ribosomal RNA varied in size between 1098 and 1157 bp. The CPS was 2079 bp in length (ungapped length mean 1649 bp, standard deviation 424.2), with 595 identical sites (28.6%) and a pairwise identity of 76.5%. Base frequencies as percentage of non-gaps were: A 22.1%, C 25.3%, G 28.4%, and T 24.2%; GC 53%. Phylogenetic analyses: single gene datasets 28S, CO1, and H3 were analysed as single datasets. They were partially poorly resolved at the lower taxonomic levels, although the Orthalicoidea as a superfamily are well supported in most analyses under Maximumlikelihood (28S: 95; CO1: <50; H3: 94; Figs. 5, 7 and 8 respectively). In 28S analysis, all clades (A D1&D2, herein re-numbered A B, D, F G) obtained by Breure et al. (2010) could be recognized, plus two additional clades (C and E) due to increased taxon sampling (See summary of clades in Fig. 1). No fundamental changes in support values could be observed between the Bayesian analyses (Fig. 6 and Breure et al. 2010: fig. 1). Analyses of CO1 as single gene analysis (with a reduced taxon set) proved to be rather uninformative, especially under Bayesian inference, when most taxa appeared as a single polytomy (not shown). Under ML, several groups of taxa could be recognized, partially according with the clade structure mentioned above (Fig. 7). However, the support values were generally very low. Analyses of H3 also used a reduced taxon set. The results show the same pattern as in the 28S analysis, but generally with lower support values (Fig. 8).

8 8 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea Fig. 6. Bayesian phylogeny for the Orthalicoidea as a result of a MrBayes analysis, based on the same dataset as shown in Fig. 5. Posterior probabilities of 0.95 or above shown at left side of nodes. Scale bar in substitutions/site. Phylogenetic analyses: combined dataset All analyses of the CPS gave a strong support for monophyly of the Orthalicoidea (ML: 99, MrB: 1, B: 0.94; Figs 9 10 and Fig. 1 respectively). The topology followed generally those obtained in the single gene analyses of 28S (exceptions, see below), with the same order for the different clades. The following is based on the BEAST analysis only. Within clade A three subclades were obtained, comprising species hitherto classified within Bulimulus, Naesiotus, Rabdotus and Bostryx (A1), Drymaeus, Neopetraeus and Scutalus (A2), and Bostryx (A3); posterior probabilities 1, 1, and 1 respectively. Clade B is weakly (posterior probability: 0.55) supported in this analysis. This clade is clearly demarcated in two subclades: Clessinia and Spixia (B1) and Plagiodontes, Cyclodontina and Bahiensis (B2). In the former subclade, both genera appear to be paraphyletic. Clade C is strongly supported and comprises both Leiostracus and Simpulopsis species. Also in this clade, paraphyletic relations appear. The root of clade A C is strongly supported. Also clade D is strongly supported and shows an assemblage of two subclades. Placocharis, Eumecostylus and Placostylus are strongly supported as subclade D1. Bothriembryon, Prestonella, and the combination of Plectostylus and Discoleus are, however, weakly supported as subclade D2 (posterior probability: 0.68; see also Discussion). Clade E is moderately supported (posterior probability: 0.85) and consists of Megaspira and two species hitherto classified as Thaumastus s.str. Clade F comprises Corona, Porphyrobaphe, Orthalicus and Thaumastus (Kara), but is only weakly supported (posterior probability: 0.74). The support for clade G, comprising Plekocheilus and Gaeotis, is strong. It should

9 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea 9 Fig. 7. Maximum-likelihood phylogeny for the Orthalicoidea, based on 654 bp cytochrome oxidase I mitochondrial DNA. Bootstrap values of 90 and above are presented to the left of the nodes. Scale bar in substitutions/site. be noted, however, that both clade F and G are poorly sampled in terms of taxa described from these groups. Finally, in this analysis Thaumastus (Paeniscutalus) crenellus is basal in the Orthalicoidea tree. The ML and MrB analysis (Figs 9 10 respectively) follow the pattern described above, with some notable exceptions. The positions of subclades A1 and A2 were swapped in the ML and MrB analyses, with a weak support (ML: 50) for the node between the two subclades in the ML analysis. In both the ML and MrB analyses, Bahiensis appears paraphyletic to the other members of clade B. In clade D the topology differs from that described above. In the ML analysis Bothriembryon is basal within the clade. In the MrB analysis, the Neotropical genera Plectostylus and Discoleus appear basal, while the other members of the clade form a polytomy. The root of clades F and G is strongly supported in both the ML and MrB analyses (ML: 98, MrB: 1). The topology differs, however, in both analyses due to the position of Thaumastus (Kara) thompsonii (in ML analysis within clade G, support < 50; in MrB analysis within clade F, support 0.64). In the ML analysis, Thaumastus (Paeniscutalus) crenellus is the sister-group of clades F G but with very weak support (< 50); in the MrB analysis the basal part of the Orthalicoid tree is a polytomy. Finally, Pilsbrylia paradoxa appears in all analyses (including the single gene analyses) among the outgroup species. Hitherto, Pilsbrylia has been classified within the Odontostomidae. Phylogenetic signal detection Likelihood mapping of the CPS revealed that the data are phylogenetically informative, with 90.4% of the quartets being fully resolved. The distribution of the quartet analysis is shown in Fig. 2, with the fully

10 10 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea Fig. 8. Maximum-likelihood phylogeny for the Orthalicoidea, based on 267 bp histone 3 nuclear DNA. Bootstrap values of 90 and above are presented to the left of the nodes. Scale bar in substitutions/site. recovered quartets depicted in the corners of the triangle. Conflicting topologies account for 5.9% (partially resolved quartets along the sides); 3.7% of the quartets are unresolved and shown in the centre. Ancestral area reconstruction Proportional probabilities inferred with maximum likelihood, representing the mapping confidence given the current area of the extant taxa, are presented in Fig. 3. The uncertainty about the biogeographical conclusion is highest in three parts of the tree, where one-state likelihood is less than This is markedly the case within clade D (Placostylidae sensu lato), which may be due to the limited taxon sampling in combination with a complex geological and biogeographical history of the Gondwana area. Parts of clades A and G suggest three independent dispersals out of South America into Central America and the West Indies. However, it may be noted that one-state likelihood of several nodes indicate uncertainty about the timing, and further studies on this group are clearly needed. Divergence time analysis Divergence time estimates based on our molecular data (Fig. 4) place the origin of the crown group of the Orthalicoidea at approximately 114 Ma (time of most recent common ancestor, tmrca; 95% posterior credibility interval (CI): Ma). During the late Cretaceous the ancestal proto-orthalicids arose at 72 Ma (CI: ) from which derived the Orthalicidae and Amphibulimidae (clades F and G, respectively). Around the same time (tmrca 88 Ma, CI: ) the proto-bulimulids split into the Gondwana clade (D) and all remaining families (clade A C, E). The major diversification within the superfamily, however, started around the Cretaceous Paleogene (K Pg) boundary (Fig.

11 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea 11 Fig. 9. Maximum-likelihood phylogeny for the Orthalicoidea, based on a 2079 bp multi-locus sites of mitochondrial and nuclear DNA. Bootstrap values of 90 and above are presented to the left of the nodes. Scale bar in substitutions/site. 4, Table 2), when as many as 50% of land-dwelling species went extinct (Bailey et al. 2005; Benton 1997). With fossil calibration points in clades A, B, and E, the resulting tmrca for clade D is markedly recent (38.6 Ma). The Placostylidae sensu Neubert et al. (2009) appear basal in this clade, with a tmrca at 28.9 Ma for the node between New Zealand and Solomon Island taxa (Fig. 4; CIs in Fig. 11). The node between Bothriembryon and the remaining three genera (Prestonella, Discoleus and Plectostylus) has a tmrca of 31.4 Ma (CI: ). The node between the African Prestonella and South American taxa is dated as 25.9 Ma (CI: ). Discussion Ancestral area reconstruction, divergence times, and geological evidence Recently, Crisp et al. (2011) have argued for applying testable hypotheses in biogeography, and to assess processes like vicariance and long-distance dispersal and establishment (LDDE). They also argue that AAR is an inductive and narrative approach. Although we concur with them that testing alternative hypotheses to explain current disjunctions is a more sound approach, we see still an added value for AAR in stimulating the generating of new hypotheses. Our preliminary analysis supports an origin for this group in (what is now south-eastern) South America,

12 12 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea Fig. 10. Bayesian phylogeny for the Orthalicoidea as a result of a MrBayes analysis, based on the same dataset as shown in Fig. 9. Posterior probabilities of 0.9 or above shown at left side of nodes. Scale bar in substitutions/site. with subsequent radiations into other parts of this continent and into different parts of Gondwana. With clade D as the only one within the Orthalicoidea which is distrubuted on three southern continents, the origin of this group (Fig. 3, nodes II III) may be dated as early as the Cretaceous (Boger 2011; Franzese & Spalletti 2001; McLoughlin 2001). The split between Placostylidae sensu Neubert et al. (2009) and Bothriembryon (node III; see below) was postulated to be related to the division of Antarctica in a Western and Eastern part (Breure 1979); however, Boger (2011: fig. 13) dates the Terra Australis Orogen that divides these parts at Ma, which pre-dates the Cretaceous. The split between Zealandia and Gondwana is positioned in the Cretaceous (Laird & Bradshaw 2004; Trewick et al. 2007). Within the Placostylinae, the split between species from the Solomon Islands and those from New Zealand and New Caledonia (node IV) is difficult to define from geological data, as the geology of this region is complex (Trewick et al. 2007; Neall & Trewick 2008). However, a land connection between New Zealand and Solomon Island probably never existed (Trewick, pers. commun.). The data presented herein only support one of the hypotheses, viz. H2: the split between Australian Bothriembryon and South American Discoleus and Plectostylus, node V, is according to the known geological data of break-up of Gondwana. As Crisp et al. (2011) point out, tests of vicariance ought to be two-tailed. Vicariance is rejected if the divergence between the taxa is too young (post-dates the origin of the barrier) or too old (pre-dates the origin of the barrier). In our data the CI of node V overlaps the time bar of the break-up between Australia and South America. Therefore, vicariance cannot be rejected. The two other hypotheses (H1, split between

13 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea 13 Fig. 11. Maximum clade credibility (MCC) chronogram inferred using the non-autocorrelated model of rate evolution in BEAST, with all nodes shown. Scale bar represents 20 Ma. Blue error bars represent 95% posterior credibility intervals and are only given for nodes that were present on more than 50% of the posterior sampled trees. See text for further explanation. Placostylidae and others, respectively H3, split between Prestonella and Discoleus/Plectostylus) are rejected because the divergence time and error bars do not overlap with the known geological timing. Commonly, dispersal (in this case LDDE) is inferred as the alternative explanation of a biogeographical disjunction of a vicariance hypothesis (Crisp et al. 2011). Land snails are generally thought not to be dispersed over long distances, despite a growing evidence of the contrary (Gittenberger et al. 2006, Greve et al. 2010; Hoekstra & Schilthuizen 2011; Rowson et al. 2011). In this group two instances of LDDE would need to be hypothesized and made testable using independent evidence; however, this is beyond the scope of the present paper and a topic for further research. Topologies and systematics The division within the Orthalicoidea into different clades of which some are monophyletic, is strongly supported in both the ML and BI analyses. These clades correspond to the Bulimulidae (clade A; but see below), Odontostomidae (clade B), Placostylidae sensu lato (clade D), Orthalicidae (clade F) and Amphibulimidae (clade G) of Breure et al. (2010) respectively. The Bulimulidae is made up by three clades, A1 A3, which may each be given subfamily rank. Clade A1 comprises the genera Bulimulus, Rabdotus, Naesiotus, and partly the genus Bostryx (sensu lato; Breure 1979). As the genus Bulimulus is represented in the clade, it may be called Bulimulinae Tryon Further analyses need to show which of the subgenera of Bostryx may be assigned to this subfamily; their taxonomic status has to be re-evaluated accordingly. In clade A2 the genera Drymaeus, Peltella, Neopetraeus, and Scutalus are represented. The placement of Peltella in this clade is surprising at first sight, since it is a semi-slug which has been classified with other semi-slugs in the Amphibulimidae (Breure 1974; Peltellinae sensu Schileyko 1999). Semislugs, however, may be seen as adaptations to humid environments and are known to occur in various families of otherwise full-shelled taxa (Breure 2010; Cuezzo 1997;

14 14 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea Naranjo-Garcia et al. 2000). Given the placement of Peltella in this clade, it may be called Peltellinae Gray The third clade (A3) consists of Bostryx s.str. species (including the type species B. solutus); it may be called subfamily Bostrycinae Breure subfam.nov. Applying Art of the International Code of Zoological Nomenclature, the name of the family may be retained as Bulimulidae Tryon 1867, despite the senior synonymy of Peltellinae Gray Bahiensis species appear as a paraphyletic group in the ML analyses, but as sister-group to all other Odontostomidae (clade B) in Bayesian analyses. In the BEAST analyses, however, Bahiensis appears as a member of the Odontostomidae; in the present study this genus is retained as a member of this family. Clade C is composed of Leiostracus and Simpulopsis species, and corresponds to the Simpulopsini of Schileyko (1999); it is here raised to family rank to reflect its paraphyletic position to the Odontostomidae. Clade D is the only clade that shows a Gondwanan distribution pattern. Differences in topologies were only observed in the ML and MrB analyses; in the latter there remained a polytomy. In the BEAST analysis a topology was obtained that is partly consistent with the topology presented by Herbert & Mitchell (2009): Prestonella as sister-group to Bothriembryon, and Bothriembryon as sister-group to Placostylus. However, the inclusion of Discoleus and Plectostylus in the current analyses suggests that Prestonella is equally closely related to these South American genera. The support values in the molecular analyses for this group are relatively low and may need studies with increased taxon sampling. As this clade D appears to be monophyletic, the oldest available family name i.e. Bothriembryontidae Iredale 1937 should be used for it. This replaces the use of Placostylidae Pilsbry 1946 (Bouchet et al. 2005), which may be retained at subfamily rank for these snails from New Zealand and Melanesia (clade D1) since Neubert et al. (2009) showed their anatomical distinctness. Clade E, comprising Thaumastus (Thaumastus) and Megaspira species, appears as paraphyletic to the previous one in all analyses. This clade may provisionally be called the Megaspiridae; increased taxon sampling should show if this grouping may be retained, as the support value is only moderate. Clades F and G correspond to the Orthalicidae, respectively Amphibulimidae (Breure et al. 2010). On the basis of this analysis Breure (2011) treated Thaumastus (Kara) as a separate genus. The paraphyletic and basal position of the monotypic Paeniscutalus crenellus suggests that this species is a relict of an older group; the species was hitherto classified as Thaumastus (Paeniscutalus); its present status is in accordance with Parodiz (1962). It is surprising that Pilsbrylia paradoxa, hitherto thought to belong to the Odontostomidae, in all analyses is grouped with the outgroup species. However, only part of the genera currently considered to belong to the Odontontomidae have been studied phylogenetically. The current classification within the superfamily Orthalicoidea is summarized in Table 3. Conclusion This analysis supports in many aspects the results of previous studies, summarized by Breure et al. (2010). However, also a number of surprising new results were obtained which alter the classification of this group as currently understood. For a better understanding of the deeper phylogenetic relationships within the Orthalicoidea, it would be interesting to have a better taxon sampling, especially of the Bothriembryontidae (Bothriembryon in Australia, Placostylus in Fiji and Vanuatu, and Plectostylus in Chile), Orthalicidae (several genera in northern South America, Central America, and the Caribbean), and Amphibulimidae (Plekocheilus, and various Caribbean semi-slugs). This would probably also shed light on the dispersals, now supposedly independent, into the Caribbean, and could lead to a well-sampled phylogenetic hypothesis that would allow further biogeographical hypotheses and a framework for future evolutionary studies. Acknowledgements We are indebted to the following persons who generously donated material for sequencing: A.C. Breure (Oegstgeest), M.G. Cuezzo (Tucumán), A. Delsaerdt (Aarschot), B. Hausdorf (Hamburg), D. Herbert (Pietermaritzburg), A. Hovestadt (Amersfoort), V. Mogollón (Lima), G. Montalván (Lima), I. Richling (Stuttgart), D.G. Robinson (Philadelphia), C. Schizzi (Buenos Aires), L.R.L. Simone (São Paulo). For assistance with laboratory work we are most grateful to D.S.J. Groenenberg. M. Schilthuizen provided useful suggestions on a previous draft of the paper. We also like to thank V. Mogollón for all his support during field work in Peru. For helpful suggestions with prior settings and calibration points in BEAST we are grateful to J. Miller and S. Williams. During his stay in the Netherlands, P.R. was supported by a Martin fellowship of NCB Naturalis. Comments from B. Rowson and an anonymous reviewer helped to improve the manuscript and is here thankfully acknowledged.

15 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea 15 References Bailey, J.V., Cohen, A.S. & Kring, D.A. (2005): Lacustrine fossil preservation in acidic environments: implications of experimental and field studies for the Cretaceous-Paleogene boundary acid rain trauma. Palaios, 20: doi: /palo.2003.p03 Benton, M.J. (1997): Vertebrate paleontology, 2nd edition: 452 pp. London (Chapman & Hall). Boger, S.D. (2011): Antarctica Before and after Gondwana. Gondwana Research, 19: doi: / j.gr Bouchet, P., Fryda, J., Hausdorf, B., Ponder, W., Valdés, A. & Warén, A. (2005): Working classification of the Gastropoda. Malacologia, 47: Breure, A.S.H. (1974): Notes on the genus Gaeotis Shuttleworth, 1854 (Mollusca, Gastropoda, Bulimulidae). Netherlands Journal of Zoology, 24: Breure, A.S.H. (1978): Notes on and descriptions of Bulimulidae (Mollusca, Gastropoda). Zoologische Verhandelingen Leiden, 164: Breure, A.S.H. (1979): Systematics, phylogeny and zoogeography of Bulimulinae (Mollusca). Zoologische Verhandelingen Leiden, 168: Breure, A.S.H. (2008): Carination strikes the eye: extreme shell shapes and sibling species in three Andean genera of the Orthalicidae (Gastropoda, Stylommatophora). Zoologische Mededelingen Leiden, 82: Breure, A.S.H. (2010): The rediscovery of a semi-slug: Coloniconcha prima Pilsbry, 1933 (Gastropoda, Pleurodontidae) from Hispaniola. Basteria, 74: Breure, A.S.H. (2011): Annotated type catalogue of the Orthalicoidea (Mollusca, Gastropoda) in the Royal Belgian Institute of Sciences, Brussels, with descriptions of two new species. ZooKeys, 101: Breure, A.S.H., Groenenberg, D.S.J. & Schilthuizen, M. (2010): New insights in the phylogenetic relations within the Orthalicoidea (Gastropoda, Stylommatophora) based on 28S sequence data. Basteria, 74: Breure, A.S.H. & Schouten, J.R. (1985): Notes on and descriptions of Bulimulidae (Mollusca, Gastropoda), II. Zoologische Verhandelingen Leiden, 216: Brito, I.M. (1967): Gastrópodos continentais do Paleoceno do Estado do Rio de Janeiro, Brasil. Boletim de Geología, Rio de Janeiro, 1: Castresana, J. (2000): Selection of conserved blocks from mutiple alignments for their use in phylogentic analyses. Molecular Biology and Evolution, 17: Christensen, C.C. & Miller, W.B. (1975): Genital anatomy and phylogeny of the snails, Berendtia Crosse & Fischer and Spartocentrum Dall (Stylommatophora). The Nautilus, 89: Crisp, M.D., Trewick, S.A. & Cook, L.G. (2011): Hypothesis testing in biogeography. Trends in Ecology and Evolution, 26: doi: / j.tree Cuezzo, M.G. (1997): Cryptostrakon corcovadensis, a new species of semislug from Costa Rica (Helicoidea: Xanthonychidae) with comments on the systematic position of the genus. American Malacological Bulletin, 14: 1 8. Cuezzo, M.G., Valdovinos, C. & Breure, A.S.H. (2010): Discoleus Breure, 1978: Is there a pre-gondwanan ancient stock of Orthalicidae? (Mollusca, Stylommatophora) Abstracts VI Southern Connection congress, February 2010, Bariloche, Argentina: 128. Delsaerdt, A. (2009 [2010]): Land snails on the Solomon Islands, Vol. 1. Placostylidae: Ancona (L Informatore Piceno). Drummond, A.J. & Rambaut, A. (2007): BEAST, Bayesian evolutionary analysis by sampling trees. BMC Evolution Biology, 7: 214. doi: / Folmer, O., Black, M., Hoeh, W., Lutz, R. & Vrijenhoek, R. (1994): DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates. Molecular Marine Biology and Biotechnology, 3: Franzese, J.R. & Spalletti, L.A. (2001): Late Triassic-early Jurassic continental extension in southwestern Gondwana: tectonic segmentation and pre-break-up rifting. Journal of South American Earth Sciences, 14: doi: /s (01) Gittenberger, E., Groenenberg, D.S.J., Kokshoorn, B. & Preece, R.C. (2006): Molecular trails from hitchhiking snails. Nature, 439: 409. doi: / a Greve, C., Hutterer, R., Groh, K., Haase, M. & Misof, B. (2010): Evolutionary diversification of the genus Theba (Gastropoda: Helicidae) in space and time: A land snail conquering islands and continents. Molecular Phylogentics and Evolution, 57: doi: /j.ympev Herbert, D.G. (2007): Revision of the genus Prestonella (Mollusca: Gastropoda: Orthalicoidea: Bulimulidae s.l.): a distinctive component of the African land snail fauna. African Invertebrates, 48: Herbert, D.G. & Mitchell, A. (2009): Phylogenetic relationships of the enigmatic snail genus Prestonella: the missing African element in the Gondwanan superfamily Orthalicoidea (Mollusca: Stylommatophora). Biological Journal of the Linnean Society, 96: doi: /j x Hoekstra, P. & Schilthuizen, M. (2011): Phylogenetic relationships among isolated populations of the limestonedwelling microsnail Gyliotrachela hungerfordiana

16 16 Breure, A.S.H. & Romero, P. E.: molecular phylogeny of the land snail superfamily Orthalicoidea (Gastropoda: Vertiginidae). Journal of. Zoological Systematics and Evolutionary Research, 49: doi: /j x. Hug, L.A. & Roger, A.J. (2007): The impact of fossils and taxon sampling on ancient molecular dating analyses. Molecular Biology and Evolution, 24: doi: /molbev/msm115 Jobb, G., von Haeseler, A. & Strimmer K. (2004): Treefinder: a powerful graphical analysis environment for molecular phylogentics. BMC Evolution Biology, 4: 18. doi: / Jobb, G. (2008): Treefinder version of October Munich, Germany. Distributed by author, Laird, M.G. & Bradshaw, J.D. (2004): The break-up of a longterm relationships: the Cretaceous separation of New Zealand from Gondwana. Gondwana Research, 7: doi: /s x(05) Larkin, M.A., Blackshields, G., Brown, N.P., Chenna, R., McGettigan, P.A., McWilliam, H., Valentin, F., Wallace, I.M., Wilm, A., Lopez, R., Thompson, J.D., Gibson, T.J. & Higgins, D.G. (2007): Clustal W and Clustal X version 2.0. Bioinformatics, 23: doi: /bioinformatics/btm404 Lewis, P.O. (2001): A likelihood approach to estimating phylogeny from discrete morphological character data. Systematic Biology, 50: doi: / Maddison, W.P. & Maddison, D.R. (2006): Mesquite: a modular system for evolutionary analysis. Version Available from McLouglin, S. (2001): The breakup history of Gondwana and its impact on pre-cenozoic floristic provincialism. Australian Journal of Botany, 49: doi: /bt00023 Miquel, S.E. & Bellosi, E.S. (2010): Middle Eocene Oligocene gastropods of the Sarmiento Formation, central Patagonia. In: Madden, R., Carlini, A. & Guiomar, M. (eds.), The paleontology of Gran Barranca: evolution and environmental change through the middle Cenozoic of Patagonia: New York (Cambridge University Press). Naranjo-Garcia,E., Polaco, O.J. & Pearce, T.A. (2000): A new genus and species of semi-slug from southern Chiapas, Mexico (Gastropoda: Pulmonata: Xanthonychidae). Archiv für Molluskenkunde, 128: Neall, V.E. & Trewick, S.A. (2008): The age and origin of the Pacific Islands: a geological overview. Philosophical Transactions of the Royal Society B, 363: doi: /rstb Neubert, E., Chérel-Mora, C. & Bouchet, P. (2009): Polytypy, clines, and fragmentation: The bulimus of New Caledonia revisited (Pulmonata, Orthalicoidea, Placostylidae). In: Grandcolas, P. (ed.), Zoologia Neocaledonica 7. Biodiversity studies in New Caledonia. Mémoires du Muséum national d Histoire naturelle, 198: Parent, C.E. & Crispi, B.J. (2006): Sequential colonization and diversification of Galápagos endemic land snail genus Bulimulus (Gastropoda, Stylommatophora). Evolution, 60: Parodiz, J.J. (1962): New and little-known species of South and Central American land snails (Bulimulidae). Proceedings of the U.S. national Museum, 113: Parodiz, J.J. (1969): The Tertiary non-marine Mollusca of South America. Annals of the Carnegie Museum, 40: Ponder,W., Colgan, D., Gleeson, D. & Sherley, G The relationships of Placostylus from Lord Howe Island. Molluscan Research, 23: doi: / MR03001 Posada, D. (2008): jmodeltest: Phylogenetic model averaging. Molecular Biology and Evolution, 25: doi: /molbev/msn083 Ramírez, R., Ramírez, J., Romero, P., Chumbe, A. & Ramírez, P. (2009): Posición evolutiva de caracoles terrestres peruanos (Orthalicidae) entre los Stylommatophora (Mollusca: Gastropoda). Revista Peruana Biología, 16: Richardson, C.L. (1993): Bulimulacea: catalog of species. Amphibulimidae, Anadromidae, Grangerellidae, Odontostomidae, Orthalicidae. Tryonia, 27: Richardson, C.L. (1995): Bulimulidae: catalog of species. Tryonia, 28: Ronquist, F. & Huelsenbeck, J.P. (2003): MrBayes 3: Bayesian phylogentic inference under mixed models. Bioinformatics, 19: doi: /bioinformatics/btg180 Roth, B. & Megaw, P.K.M. (1989): Early Tertiary land mollusks (Gastropoda: Pulmonata) from Sierra Santa Eulalia, Chihuahua, Mexico, and the origins of the North American arid-land mollusk fauna. Malacalogical Review, 22: Rowson, B., Tattersfield, P. & Symondson, W,O,C. (2011): Phylogeny and biogeography of tropical carnivorous land-snails (Pulmonata: Streptaxoidea) with particular reference to East Africa and the Indian Ocean. Zoologia Scripta, 40: Schileyko, A.A. (1999): Treatise on Recent terrestrial pulmonate molluscs, 3. Partulidae, Aillyidae, Bulimulidae, Orthalicidae, Megaspiridae, Urocoptidae. Ruthenica, Supplement, 2: Schmidt, H.A. & von Haeseler, A. (2007): Maximum-Likelihood analysis using TREE-PUZZLE. In: Baxevanis, A.D., Davison, D.B., Page, R.D.M., Stormo, G. & Stein, L. (eds.) Current Protocols in Bioinformatics (Supplement 17), Unit 6.6: New York (John Wiley and Sons). doi: / bi0606s17 Schmidt, H.A., Strimmer, K., Vingron, M. & von Haeseler, A. (2002). Tree-Puzzle: maximum likelihood phylogenetic analysis using quartets and parallel computing.