大山蝸牛屬及台灣山蝸牛屬之種化事件與山蝸牛科之系統發育學研究

全文

(1)國立臺灣師範大學生命科學系博士論文. 大山蝸牛屬及台灣山蝸牛屬之種化事件與山蝸牛科之系統發育學研究 Cyclophoridae phylogeny and the speciation events of Cyclophorus and Cyclotus taivanus ssp.. 研究生：李彥錚 Yen-Chen Lee. 指導教授：呂光洋 Kuang-Yang Lue 巫文隆 Wen-Lung Wu. 中華民國九十七年六月.

(2) 96.11. 版學位論文授權書. 國立臺灣師範大學學位論文授權書本授權書所授權之論文為授權人在國立臺灣師範大學研究所. 生命科學. 96 學年度第 2 學期取得博士學位之論文。. 論文題目：大山蝸牛屬及台灣山蝸牛屬之種化事件與山蝸牛科之系統發育學研究指導教授：呂光洋‧巫文隆授權事項： □同意非專屬無償授權本校及國家圖書館將上列論文資 □不同意料以微縮、數位化或其他方式進行重製，並可上載網路收錄於本校博. 一、授權人. 碩士論文系統、國家圖書館全國博碩士論文資訊網及臺灣師範校院聯合博碩士論文系統，提供讀者基於個人非營利性質之線上檢索、瀏覽、下載、傳輸、列印或複印等利用。二、論文全文電子檔上載網路公開時間：【第一項勾選同意者，以下須擇一勾選】 □ 即時公開 □ 自＿ 2009. 年＿＿8. 月＿＿31. 授權人姓名：學. 日始公開。（請親筆正楷簽名）. 號：891430048. 註：1. 本授權書須列印並簽署兩份，一份裝訂於紙本論文書名頁，一份繳至圖書館辦理離校手續。 2. 授權事項未勾選者，分別視同「同意」與「即時公開」。. 中. 華. 民. 國. 97 年. 6 月. 27 日.

(3)

(4)

(5) Content Abstract………………………………………………………………………....………..III Chapter 1 General introduction ……………………………………...…………..………..1 Chapter 2 Molecular phylogeny of the family Cyclophoridae (Gastropoda: Architaenioglossa) in East Asia…………………………………..……………3 Chapter 3 Ring speciation and morphological adaptations of genus Cyclophorus in Taiwan………………………………………………………………………...37 Chapter 4 The phylogenetic evolution and morphological adaptations in Cyclotus taivanus ssp.………………………………………...…………..…………….63 Chapter 5 The cyclophorids fauna of Taiwan……………..……..………………………91 Chapter 6 Summary and conclusion……………………………………………………151 References………………………………………………………………………………155 Acknowledgment…………………………………………………………...…………..163. I.

(6) II.

(7) Abstract. Cyclophoridae consists of four subfamilies and about 300 species currently arranged in 38 genera, occuping varies habitats, with great morphological diversity allover the world. However, the relationship among Cyclophoridae is thus far not clear. In order to investigate this, I sampled cyclophorid snails around Taiwan and its adjacent areas, and then sequenced part of the mitochondrial COI (cytochrome oxidase subunit I) and the 16S rRNA gene from 32 species of 10 cyclophorid genera to establish a phylogenetic tree of Cyclophoridae. Phylogenetic relationships based on mtDNA sequences suggest that Cyclophorus, Cyclotus, Leptopoma, and Platyrhaphe are monophyletic while the traditional genus Japonia is polyphyletic, and the previous J. zebra should be placed into a new genus Pilosphaera. In addition, Pilosphaera yentoensis n. sp. and Japonia boonkioensis n. sp. will also be described as new species. Members of Cyathopoma are tiny white cyclophorid snails occurring in East Asia, Madagascar and the Seychelles. Phylogenetic relationships of Cyathopoma are uncertain. Combined with COI and radular data, I conclude that Cyathopoma and Cyclotus are only distantly related. Cyathopoma iota has been considered to be a controversial member of this group. Through molecular and radular data, I found C. iota to be closer to C. taiwanicum than to C. micron, and concluded that C. micron, C. ogaitoi, C. iota and C. taiwanicum all belong to Cyathopoma. There are 10 genera and 29 cyclophorid species in Taiwan. Among them, the most interesting taxa are Cyclophorus and Cyclotus, both sharing similar ecological niches and representing by a north and south form in morphology. In order to clarify their relationship, I have to find out their sister group as out groups to compare with the members among Cyclophorus and Cyclotus. The gene trees of Cyclophoridae indicate that Japonia and Pterocyclus are sister group of Cyclophorus and Cyclotus, respectively. The former two will be used as the out groups of Cyclophorus and Cyclotus in their phylogenetic studies. Both COI and 16S rRNA gene trees of Taiwan Cyclophorus show prominent geographic structure. The Mantel test showed significant positive correlation between fixation index (FST) and cumulative geographic anti-clockwise distance (origin in the region around Tainan, anti-clockwise pass through south cape, Taidung, Hualian, Iran, Taipei, Taichung and meet the original populations in Jia-yi). There are finite gene flew between adjacent populations. And there are series of clines around the Central Range. Cyclophorus of Taiwan is a proposed “ring species”. In the morphology and environmental variables correlation study, I found the currently shell morphology may III.

(8) be caused by the adaptations of recent long term climate. In traditional classification, Cyclotus taivanus consists of five subspecies, with clear morphological diversity. The molecular phylogenetic relationships of this group have never been discussed before. I sequenced part of the mitochondrial COI (cytochrome oxidase subunit I) and the 16S rRNA gene from 27 sampling sites. I also measured 9 shell traits to investigate the relationships between C. taivanus ssp. Even though the morphology PCA revealed a more or less continuous distribution of individuals in morph-space, the two highly divergent haplotype clades in COI and 16S rRNA analysis indicated the presence of two independently evolving lineages. The sequence divergence between two clades was almost as high as between other Cyclophoridae species. Therefore C. adamsi should be a considered valid species. For the environmental analysis, temperature may be a limiting element to the distribution of C. adamsi and C. taivanus group. The ecological divergence probably is the ruling force of speciation in my case. The PLS analysis results indicate, that phenotypic plasticity may be a key element of variable shell in C. taivanus group. The ecological divergence probably appears rule of speciation in C. taivanus ssp. case. The speciation process may be incomplete among C. t. dilatus, C. t. diminutus, C. t. peraffinis, and C. t. taivanus, and the adaptation of climatic pressure continue being a rule of speciation process. This study provides an opportunity to understand that no matter how similar two taxonomic groups are, occupying similar niche, undergoing the same geology history, with morphological adaptation to the same long term climate, they may have different speciation model.. IV.

(9) Chapter 1. General introduction. Kobelt (1902) had been introduced cyclophorids in 1902, there were only 7 genera and 13 species in Taiwan. After Kobelt, Pilsbry and Hirase (1906) reported 7 genera and 21 cyclophorid species of Taiwan. There were 9 genera and 19 Taiwan cyclophorid species in Kuroda’s report of 1941 (Kuroda 1941). After these pioneers, Lee and Wu reported 9 genera and 29 species of Cyclophoridae in Taiwan (Lee & Wu, 2001). However, the phylogeny of cyclophorid is not clear. Besides, some new species and new records were found recently. Cyclophorids are very common snails in Taiwan, the genus Cyclophorus and Cyclotus are the most mysterious groups. The Cyclophorus moellendorffi found in southern Taiwan was considered a subspecies of C. formosaensis found in northern Taiwan by some authors (Kuroda, 1941; Chang, 1984; Lai, 1990; Higo & Goto, 1993). Besides, in the eastern side of Central Mountain Range, C. friesianus and C. latus were also believed to share a subspecies relationship by some authors (Kuroda, 1941; Chang, 1984; Lai, 1990; Higo & Goto, 1993; Hsieh et al., 2006). It is interesting that all the southern Cyclophorus species possess keeled shells, but the northern species possess round shells. Cyclotus taivanus ssp. are another interesting group. Pilsbry and Hirase (1905, 1906) reported C. t. peraffinis and C. t. adamsi as a subspecies of C. taivanus by their glossy surface and tall spire, respectively. Lee & Wu (2001) reported C. t. dilatus and Cyclotus t. diminutus as subspecies of C. taivanus by their wide out lip and very small shell, respectively. Like Cyclophorus, Cyclotus taivanus ssp. could divide north form (C. t. adamsi) and south from (C. t. dilatus, C. t. diminutus and C. t. peraffinis), roughly. The north form possess tall spire, the south form possess low spire. It is interesting that both Cyclophorus and Cyclotus taivanus ssp. occupy similar habitat, have two morphologic forms. Do Cyclophorus and Cyclotus taivanus ssp. undergo similar speciation process? The objectives of this study were (1) to establish the Molecular phylogeny of the family Cyclophoridae, (2) to understand the speciation model morphological adaptations of genus Cyclophorus, (3) to understand the speciation model morphological adaptations of genus Cyclotus, (4) to discuss the biogeography and morphology of Taiwan Cyclophoridae. 1.

(10) 2.

(11) Chapter 2. Molecular phylogeny of the family Cyclophoridae (Gastropoda: Architaenioglossa) in East Asia. Abstract Cyclophoridae consists of four subfamilies and about 300 species currently arranged in 38 genera, occupying varies habitats, with great morphological diversity. The molecular phylogenetic relationships of this group have never been discussed before. In order to investigate the relationships between cyclophorids, I sequenced part of the mitochondrial COI (cytochrome oxidase subunit I) and the 16S rRNA gene from 32 species of 10 genera of cyclophorid (including Cyclophoridae of China and Japan). I constructed phylogenetic trees using neighbor joining, Bayesian and maximum likelihood analyses on part of COI and 16S rRNA gene data set comprising. Phylogenetic relationships based on mtDNA sequences suggest that Cyclophorus, Cyclotus, Leptopoma, and Platyrhaphe are monophyletic while the traditional genus Japonia is polyphyletic and the previous J. zebra should be placed into a new genus Pilosphaera. In addition, Pilosphaera yentoensis n. sp. and Japonia boonkioensis n. sp. will also be described as new species. Besides, members of Cyathopoma are tiny white cyclophorid snails occurring in East Asia, Madagascar and the Seychelles. Phylogenetic relationships of Cyathopoma are uncertain. Combined with COI and radular data, I conclude that Cyathopoma and Cyclotus are only distantly related. Cyathopoma iota has been considered to be a controversial member of this group. Through molecular and radular data, I found C. iota to be closer to C. taiwanicum than to C. micron, and concluded that C. micron, C. ogaitoi, C. iota and C. taiwanicum belong to Cyathopoma. In addition we provide the first report of C. ogaitoi from Guei-Jou Province.. Introduction Recent studies on phylogenetic relationships within the molluscan class Gastropoda have involved morphological (Kay et al. 1998), ultrastructural (Healy 1996), anatomical (Kantor 1996) and molecular (e.g., McArthur & Koop 1999, Lydeard et al. 3.

(12) 2002, Remigio & Hebert 2003) approaches. These investigations have provided new insights into gastropod affinities and classification and have enabled a rigorous testing of taxonomic schemes for the group. While gastropod phylogeny has received much recent attention (Tillier et al. 1992, Ponder & Lindberg 1997, Rosenberg et al. 1997, Thollesson 1999, Remigio & Hebert 2003), relationships within some land snail clades are still poorly understood. One of this is Cyclophoridae which consists of 4 subfamilies and about 300 species currently arranged in 34 genera. The most generally accepted system of classification today, partitioning the diverse Cyclophoridae into four subfamilies (Vaught 1989; Millard 1996). The Cyclophorinae are extremely diverse. By contrast, the other three, Spirotomatinae, Alycaeinae, and Pterocyclinae are less diverse. Although some investigators treat Spirotomatinae and Alycaeinae as independent families (Higo & Goto 1993, Azuma 1982), their subordinate taxa are uncontentious. However, the traditional classification is based on shell morphology, but part of the shell morphology may be subject to convergent evolution and thus may hamper a straightforward taxonomy (Lee et al. 2008). The phylogenetic knowledge of Cyclophoridae is limited and thus there is great interest in resolving their phylogenetic issues. Furthermore, the classification of tiny white shell cyclophorids, genus Cyathopoma W. & H. Blanford, 1861 and Nakadaella Ancey, 1904 have long been controversial. Cyathopoma micron (Pilsbry, 1900) was the first record of this tiny cyclophorid from East Asia. Cyathopoma iota (Pilsbry & Hirase 1904), C. taiwanicum Pilsbry & Hirase, 1906, C. taiwanicum degeneratum Pilsbry & Hirase, 1906, C. nishinoi Minato, 1980, C. ogaitoi (Minato, 1988) were subsequently reported but their generic placements were doubtful. For example, C. micron was initially placed in Cyclotus (Pilsbry 1900) before transferring to Nakadaella by Ancey (1904) and Cyathopoma (Pilsbry & Hirase, 1906) but the generic status of C. micron remained controversial. Afterward, whether this species should belong to Cyathopoma, Cyclotus, or Nakadaella was a huge controversy. It is interesting to understand the relationship within this group and with other cyclophorid genus. Dentition of the radula has long been established as providing informative characters for the taxonomy of gastropods (Trochel 1856-1863, Habe 1942, Ponder & Lindberg 1996). Radulae differ fundamentally among different gastropod groups. However, radular morphology must be used with care when inferring relationships because there is marked convergence in the radulae of taxa that belong to different clades but occupy similar adaptive zones (Lindberg & McLean 1981). Cuspidal features, number of teeth as well as teeth shape play an important role in classification and have proved to be of particular value at generic level (Kilburn 1988). Traditional classification of these tiny cyclophorids was based on shell and operculum morphology, while little attention has been paid to the radular morphology of cyclophorids, and the 4.

(13) Cyathopoma radula has not been previously described. Over the past decade molecular approaches have proven their value not only in clearing up relationships among taxa, but also in providing a sense of the time scales of evolutionary divergence. Among the 13 protein-coding genes within the mitochondrial genome, COI has gained particular popularity for resolving affinities among species (Hebert et al. 2004a, b; Hogg & Hebert 2004; Barrett & Hebert 2005; Smith et al. 2006). In the case of possible cryptic species identification and identifying morphologically difficult species (Paquin & Hedin 2004), COI provided a high degree of taxonomic resolution. Besides, 16SrRNA the more rapidly evolving mitochondrial gene has generally been employed to infer relationships among groups with a more recent ancestry i.e. at the species and population level (e.g. Chiba 1999; Pinceel et al. 2004). However, 16SrRNA gene can also provide insights concerning deeper divergences (e.g. Thollesson 1999; Remigio & Hebert 2003). COI and 16S data could be suitable to infer the affinities at the genus level of Cyclophoridae. The objectives of this study were (1) to test whether the genus of Cyclophoridae is monophyletic, (2) if part one is negative, to rearranged its classification or created a new genus, (3) to describe new species and provide its distributional and ecological data, (4) to resolve the relationship of the Cyathopoma species, (5) to understand the affinity between Cyathopoma species and other cyclophorid genus, (6) to provide Cyathopoma radular microstructure data.. Materials and methods DNA preparing and sequencing All samples of 32 species representing 9 genera and including members from 3 major subfamilies (Cyclophorinae, Alycaeinae and Pterocyclinae) were collected from 116 sites (Fig. 2.1) (Table 2.1), and separated their shells and soft parts at library. Shells were well cleaned for identification, soft parts were stored at -80℃, except for very tiny species (e.g. C. micron) which were placed in pure ethanol until DNA extraction. DNA was extracted from columellar muscle; the whole animal was used in the case of tiny species. I extracted DNA from separate individuals using TEK-based protocol (Jiang et al. 1997) with minor modifications. Tissue was placed in TEK buffer (12.5mM Tris-Cl pH 7.3, 2.5mM EDTA, 0.4% KCl), then ground with glass pestle and incubated at 57℃ with 20µl of proteinase K (20mg/ml) more than 2 hrs. The tissue extract was extracted at least twice with phenol and chloroform. 400µl DNA extract was precipitated by adding 1000µl pure ice-cold ethanol, and was then placed in -20℃ for 20 min. DNA was pelleted by centrifugation for 30 min. After 70% ethanol rinse, DNA was resuspended in distilled water and stored at -80℃ for DNA 5.

(14) amplification. An approximately 800bp mitochondrial 16S rRNA gene was amplified by PCR using primers 16SRT (5’–ACA TAT CGC CCG TCA CTC TC–3') and 16SL900 (5’–AAA TGA TTA TGC TAC CTT TGC–3'), exactly 531bp COI using primer LCO1490 (5'–GGT CAA CAA ATC ATA AAG ATA TTG G–3') and HCO2198 (5'–TAA ACT TCA GGG TGA CCA AAA AAT CA–3') (Williams et al. 2003). For genus Chamalycaeus, I design primer 16SRPL (5’–TTT TGC ATC ATG GTT TAG CAA G–3') and 16SLS (5’–ATG CTA CCT TAG CAC AGT CA–3') to amplify approximately 530bp 16S rRNA gene. PCR reactions contained template DNA 10–50ng/µl, 10pmol of each primer, 5µl 10× reaction buffer (10mM Tris-HCl, pH9.0, 50Mm KCl, 1.5Mm MgCl2, 0.1% gelatin, 1% Trinton X-100), 0.4µl 25mM/µl dNTP, 0.2µl 50mM Mg2+ and 0.4µl Taq polymerase (5unit/µl) in a total volume of 50µl. Thermal cycling for 16S rRNA was performed with an initial denaturation for 5min at 95℃, followed by 30 cycles of 30 sec. at 95℃, 45 sec. at 57℃, 50 sec. at 72℃ and ultimate extension at 72℃ for 10min, final hold in 4℃. Thermal cycling for COI gene was performed similarly, changing only its annealing temperature to 47℃. PCR products were purified using a purification kit (AMP PCR purification, Beckman) and then sequenced using an ABi 3700 autosequencer. Phylogenetic analyses The COI sequences were combined with data of the out-groups (Littoraria scabra Linnaeus and Littoraria undulata (Gray) accession No. are AJ488637 & AJ488635 respectively) from GenBank. Sequences were assembled and edited using Bioedit 5.0.9 (Hall 1999). All alignments employed Clustal X (Thompson et al. 1997) and were manually proofread. Codon positions within COI were tested using the incongruence length difference (ILD) test (Farris et al. 1995), as implemented by the partition homogeneity test in PAUP 4.0b10 (Swofford 1998) (100 replicates). COI sequence data were divided into two partitions, first and second codon positions in one and third codon positions in the other. Two parts of COI gene were congruent and all codon positions were combined and used in the following analysis. In 16S rRNA, Truncatella guerinii A. & J.B. Villa was used as out-group. Regions where the alignment was ambiguous were excluded from the analyses. All data sets were subject to Neighbor-joining (NJ) using PAUP 4.0b10, to the Maximum likelihood (ML) analyses using PHYML 3.0 (Guindon & Gascuel 2003), to the Bayesian analysis using MrBayes 3.1.2 (Ronquist & Huelsenbeck 2003). The substitution model used for the COI data set corresponded to the General time reversible model, and included invariable sites, and rate variation among sites (GTR+I+G). The substitution model used for the 16S rRNA data set corresponded to the transitional model, include invariable sites, rate variation among sites (TIM+I+G). 6.

(15) These were the best models found using Modeltest 3.06 (Posada & Crandall 1998). Before model fitting, the full-length sequences were tested to confirm that there was no significant heterogeneity in base frequencies across taxa (in COI: X2=237.21, df=249, P=0.6939; in 16S rRNA: X2=97.31, df=165, P=1). NJ bootstraps consisted of 1000 iterations. Reliability of ML trees were estimated by the approximate likelihood ratio test (aLRT) (using custom define model, base frequencies: A = 0.3184, C = 0.1131, G = 0.1239, T = 0.4327 in COI and A =0.3664, C = 0.0720 G = 0.1351, T = 0.4266 in 16S) using PHYML 3.0 (Guindon & Gascuel 2003). In the Bayesian analysis was run for 2,000,000 generations, with a sample frequency of 100. The first 2000 trees were discarded, so that the final consensus was based on 18,000 trees. Support for nodes was expressed as posterior probabilities (calculated by MrBayes). For constraint analyses, I conducted parsimony heuristic searches to find the best trees and using the Kishino-Hasegawa test to evaluate the resulting trees and trees which are consistent with traditional taxonomy. Environmental Scanning electron microscope (ESEM) microstructure observation The empty shell was glued to the SEM specimen stub for further observation. Radulae were removed from snails and soaked in 0.5% NaOH solution to remove organic tissue adhering to the radula, then fixed using 90% ethanol. The radula was glued to the SEM specimen stub and one side of the marginal teeth was unfolded for further observation. A final step prior to examination with the SEM, the specimens were coated, in vacuum, with gold-palladium. The specimens were then observed and photographed using Environmental Scanning Electron Microscope (FEI Quanta 200).. Results Phylogenetic analysis The aligned 531 bp COI gene data matrix, including 270 variable sites of which 250 (92.59%) were parsimony informative. No length difference from the out-group was detected among members of three subfamilies, the Cyclophorinae, Alycaeinae, and Pterocyclinae. The average p-distance of Cyclophoridae haplotype was =0.197. Sequence divergence among the haplotypes ranged from 0.002 to 0.277, intra generic ranges were from 0.066 to 0.156, and inter generic ranges from 0.193 to 0.268. The alignment of the 16S rRNA gene fragment data yielded 592 characters of which 444 could be unambiguously aligned. Of these 444 positions, 300 (67.6%) were parsimony informative. The average p-distance among haplotypes was 0.239. Sequence divergence among the haplotypes ranged from 0.002 to 0.352. The COI and 16S rRNA sequence divergence among species of the same genus ranged from 0.028 7.

(16) to 0.172 and 0.041 to 0.166, respectively (Table 2.2). The inferred phylogenetic trees between the haplotypes of COI gene are shown in Figs. 2.2, 2.3 & 2.4, of 16S rRNA gene are shown in Figs. 2.5, 2.6 & 2.7. Three monophyletic groups, Cyclophorus, Cyclotus, and Leptopoma, are present, with support values higher than 97% using NJ, Bayesian and ML methods. Although with lower support, members of genus Platyrhaphe also had a monophyletic relationship. The relationship between subfamily Cyclophorinae, Alycaeinae and Pterocyclinae were uncertain because of the low bootstrap support. Cyathopoma Phylogeny It was controversial whether micron, ogaitoi, and iota belong to Cyathopoma, Cyclotus, or Nakadaella in the previous literatures. However, all cladograms in this study indicated these species were all in the same clade as C. taiwanicum instead of genus Cyclotus. Particularly, C. iota was clustered with C. taiwanicum in a subclade, and without resolution. In order to test the position of these species in classification, I performed two likelihood analyses between topologies constrained to match given mutually exclusive hypotheses. The constraints were designed to test these cyclophorid snails as the different contentions described in the previous literatures. I performed the Kishino-Hasegawa likelihood evaluation to test these contentions. Using the constraints option in PAUP, I conducted parsimony heuristic searches (specifics same as below) to find the best trees that were consistent and inconsistent with the monophyly of these clades. The sets of trees consistent and inconsistent with the constraint were then compared using the Kishino-Hasegawa test (Kishino & Hasegawa 1989). For (micron, ogaitoi, iota, and genus Cyclotus) clade issue, the COI data significantly reject the monophyly of this clade, six “best” trees supporting monophyly were significantly worse than the 48 genuinely best trees (P ≤ 0.0008). For (micron, ogaitoi, and genus Cyclotus) clade issue, all 112 “best” trees were significantly worse (P ≤ 0.0025) than the 48 trees which did not match the constraint. Some investigators considered iota is a subspecies of micron (Higo & Goto 1993, Lee & Wu 2001). I also performed the Kishino-Hasegawa likelihood evaluation to test this issue. For this issue, 48 “best” trees were significantly worse (P ≤ 0.0044) than the 48 trees which did not match the constraint. Furthermore, if (micron, ogaitoi) and (iota, taiwanicum) were separated groups, the divergence among these two groups were 0.151 which was moderately lower than among other genus (Table 2.3). I conclude that micron, ogaitoi, and iota had closer relationship with C. taiwanicum than with members of genus Cyclotus, while micron, ogaitoi, iota and taiwanicum belong to genus Cyathopoma.. 8.

(17) Japonia & Pilosphaera Phylogeny Interestingly, the genus Japonia appeared polyphyletic for both the COI and 16S rRNA data. The zebra group, traditionally placed within the genus Japonia, differs conchologically from all other members of the genus Japonia in having red-brown longitudinal stripes on the shell and a reflected outer lip. In order to compare our results with traditional classifications, we performed two likelihood analyses between topologies constrained to match given mutually exclusive hypotheses. The constraints were designed to test the genus Japonia as described in the literature (Pilsbry & Hirase 1906; Kuroda 1941; Higo & Goto 1993). We performed the Kishino-Hasegawa likelihood evaluation to test the monophyly of the traditional classification of Japonia clades (Japonia formosana Pilsbry & Hirase, 1906, Japonia boonkioensis n. sp., Japonia lanyuensis Lee & Wu, 2001, Pilosphaera zebra (Pilsbry & Hirase, 1906), Pilosphaera yentoensis n. sp.). Using the constraints option in PAUP, we conducted parsimony heuristic searches (specifics same as above) to find the best trees that were consistent and inconsistent with the monophyly of these clades. The sets of trees consistent and inconsistent with the constraint tree were then compared using Kishino-Hasegawa test (Kishino and Hasegawa 1989). COI data significantly rejected the monophyly of Japonia. All 16 “best” trees supporting monophyly were considerably worse than the ten genuinely best trees (P < 0.0001). For 16S rRNA data, one “best” tree supporting monophyly of Japonia was significantly worse (P < 0.0001) than the two genuinely best trees. Accordingly we consider the monophyly of Japonia to be suspected. Cyathopoma ESEM microstructure Radular morphology was very similar between C. iota and C. taiwanicum. Both possessed the same number of teeth cuspids and general shape. They had a scoop-shaped central tooth, 5 cuspids on the convex side, 7–9 irregular tiny cuspids on the convex side; inner lateral teeth possessed 5 cuspids and outer lateral teeth 6–7 cuspids within the same individual; marginal teeth with 6 small cuspids on the inner side and 3 large cuspids on the outer side (Fig. 2.8D, 2.9D). The radulae of C. micron and C. ogaitoi were similar to the above, but the cuspids on the teeth were shorter and broader (Fig. 2.10D, 2.11D). Further, the outer lateral teeth had fewer cuspids (5–6 in number). The protococh of C. iota and C. taiwanicum exhibited a granular surface (Fig. 2.8B, 2.9B), but C. micron and C. ogaitoi were relatively smooth (Fig. 2.9B, 2.11B). The opercula of C. iota, C. micron, C. ogaitoi and C. taiwanicum were almost the same shape (Fig. 2.8C, 2.9C, 2.10C, 2.11C). In addition, we examined four subspecies of Cyclotus taivanus H. Adams, 1870. The radulae within the Cyclotus we examined were similar to each other. Namely 5 9.

(18) cuspids on the central tooth, 4 cuspids on inner lateral teeth, 4 cuspids on outer lateral teeth and 3 cuspids on marginal teeth (Fig. 2.12). The only distinct difference was the shape of marginal teeth. The marginal teeth of Cyclotus taivanus adams Pilsbry et Hirase, 1905 were sickle-shaped, while the other three species exhibited hook shaped marginal teeth.. Discussion Although the molecular relationship between the three major subfamilies of cyclophorids are uncertain, our results show that many taxa in traditional classifications of cyclophorids are nonmonophyletic. Because Cyclophorinae, Spirotomatinae (not included in this study), Alycaeinae and Pterocyclinae in a Linnean system are considered to be equal entities, problems associated with ranks and synonymy arise. I conclude that COI gene is useful in unraveling cyclophorid phylogeny but need to be combined with other data (such as morphological and anatomical data) to fully clarify the evolutionary relationships. Cyathopoma were known as an Indian endemic before 1900 (Pilsbry 1900). After Pilsbry’s report, C. iota, C. nishinoi, C. ogaitoi, and C. taiwanicum were found occurring in Japan and Taiwan. Prior to our discovery of several C. ogaitoi specimens in Lei Gong Shan, Guei-Jou Province in July 2006 there were no reports indicating that Cyathopoma occurs in the area between East Asia and India. This is the first report of Cyathopoma from this area. C. micron was initially placed in the genus Cyclotus (Pilsbry 1900), then placed in the genus Nakadaella by Ancey in 1904. Pilsbry reversed his conclusion and treated micron as genus Cyathopoma with coauthor in 1906 (Pilsbry & Hirase 1906). Subsequently, Kuroda (1941) and Chang (1984) recognized this species as belonging to the genus Cyathopoma, while Lai (1990) and Lee and Wu (2001) placed it in the genus Cyclotus. C. iota was considered a closer relative to C. micron than to C. taiwanicum (Pilsbry & Hirase 1904, Higo & Goto 1993, Lee & Chen 2003). However, based on our Kishino-Hasegawa test result, molecular phylogenetic and radular data, C. iota is closer to C. taiwanicum than to C. micron. Ancey (1904) proposed a genus Nakadaella for micron, but this species does not differ from typical forms of Jerdonia W. & H. Blanford, 1861 except for the absence of spiral striation (Pilsbry & Hirase 1906), and genus Nakadaella was retained as a subgenus for the smooth species (Pilsbry & Hirase 1906). However, our results indicate that the smooth shell may be a plesiomorphic character of these species, and Nakadaella was a synonym for Jerdonia. In all cladograms C. iota and C. taiwanicum were clustered without resolution. All C. 10.

(19) iota were reported from mountain regions higher than 1000m altitude (Lee & Chen 2003), except the types, and we suspect C. iota may be an ecotype of C. taiwanicum. Comparison of the radulae between these species and Cyclotus, the serrate cuspids on the convex side of central teeth in Cyathopoma are not present in Cyclotus. In addition, the fine serrate cuspids on the inner side of marginal teeth are not present in Cyclotus. Thus the radulae of Cyathopoma are clearly distinctly from those of Cyclotus. Based on molecular and radula data, I conclude that Cyathopoma and Cyclotus are distant relatives. However, the Indian species were not included in this study. Obviously, further studies are needed to include the Indian species and to understand the phylogenetic relationship of these interesting snails. The genus Japonia was established by Gould in 1859. He indicated that the group characterized by the thin paucispiral opercle with thinned edgeds, the globose conic form, free umbilicus, nearly circular peristome which barely touches the preceding whorl, and the projecting lamellar striae of growth decussating with revolving ridges in some cares furnished with epidermal barbs (Gould 1859). Japonia zebra Pilsbry & Hirase, 1906 was named by Pilsbry and Hirase and placed in genus Japonia. However, the beautiful red brown longitudinal stripes and reflexed outerlip exhibited by the zebra group are not present in the members of genus Japonia. Furthermore, in the molecular characters, the haplotype distance between zebra group and the other members of Japonia is almost twice distance within groups (0.207–0.226 between group; 0.026–0.164. within zebra group, 0.002–0.122 within Japonia group) (Table 2.4). Compared to other genera of Cyclophoridae, including the zebra group in the genus Japonia presents the highest p-distance value (Table 2.5). Base on molecular evidence and shell morphology, zebra and its analogous from riverside of Nan-Xi river in Zhejiang province are polyphyletic relationship with Japonia. I would like to establish a new genus for this group here and will describe it below.. Pilosphaera new genus Type species: Japonia zebra Pilsbry & Hirase, 1906 Diagnosis: Shell small, turbinate, conical-globe shape, with convex shell whorls. Shell is always festucine color with reddish brown longitudinal stripes. Surface is sculptured with some spiral cords and irregular growth lines, above these furnished with several row regular periostracum hairs. Umbilicus opened. The aperture is circular, outer lip reflexed. The operculum is ceratoid, translucent, a little concave, multispiral type with very thin pellucid edge. Etymology: Pilo (grow hairy) + sphaera (ball).. 11.

(20) Pilosphaera yentoensis n. sp. (Fig. 2.13A–F) Description: Shell small, 5.17–5.27mm in length and 5.27–5.49mm in width. Shell turbinate and conical-globe shape, with moderately convex whorls 5–5.25 in number. Shell is festucine color with reddish brown longitudinal stripes. Surface sculptured with several indistinct spiral cords, covered with festucine color dull periostracum and regular periostracum lamella, interval with irregular fine growth lines. There are 3 row regular periostracum hairs between the sutures on penultimate whorl. There is no periostracum hair and with polish periostracum under the peripheral line. The periostracum hairs are sometimes entirely lost, perhaps in old shells such as holotype. Umbilicus opened. The aperture is circular. The outer lip is reflexed. The operculum is translucent ceratoid, a little concave center, multispiral type with very thin pellucid edge. There is an orange red proboscis between two purplish gray tentacles on the head. Dark gray food covers with two pieces of dark gray lobe, which has pale color edge. The lobes join at the tail and come into being a pale color groove. (Fig. 2.15A–B) Notes: the present species differs from its only know congener Pilosphaera zebra in dark gray soft body color (Fig. 2.15A–C) and fewer periostracum hairs (Fig. 2.13A–I, 2.16A–B). The later has 5–6 rows regular periostracum hairs between the sutures on penultimate whorl which is 3 on the new species. The later shell base with 6–7 rows periostracum hairs, but not present on the new species (Fig. 2.13). The later has 4–7 rows tiny periostracum hairs on the position just under the suture (Fig. 2.16B), which are not present on the new species. In COI gene data, the average distance among this new species and Pilosphaera zebra was = 0.160, which was close to the average distances among cyclophorids species (among species was = 0.198, with in species was = 0.061). Measurement and type depository Holotype: SL: 5.17mm, SW: 5.27mm; APL: 2.82mm, APW: 2.82mm; NMNS5635-001, National Museum of Natural Science, Taiwan. Paratype: SL: 5.27mm, SW: 5.49mm; APL: 3.10mm, APW: 2.93mm; NMNS5635-002, National Museum of Natural Science, Taiwan. Etymology: The name is after the Yen-To town, primary habitat of this species. Type locality: Yen-To town near Nan-Xi river in Zhejiang province, China. Gathered from grass slope under leaves. Japonia boonkioensis n. sp. (Fig. 2.14A–C) Description: Shell small, 4.47mm in length and 4.7mm in width. Shell turbinate and 12.

(21) conical-globe shape, with five moderately convex whorls. Shell is red brown color, somewhat pale at peripheral. Surface sculpture of 6–10 indistinct spiral veins crossed by finer growth lines, rendering them somewhat crispate and the interstices minutely plicaulate. Shell surface covers with red brown slight shining periostracum and irregular periostracum lamella. There are 2 row regular periostracum hairs between the sutures, one furnished on the position of shoulder, one on the peripheral site. There are 7 rows periostracum hairs under the peripheral, they longer its length from umbilicus side to the peripheral side. The shoulder and peripheral periostracum hairs are 3–4 times longer than basal periostracum hairs. The periostracum hair is tapering tip. Umbilicus opened. The aperture is circular. The outer lip is not reflexed. The operculum is translucent ceratoid, a little concave center, multispiral type with very thin pellucid edge. There is an orange red proboscis between two bluish gray tentacles on the head, gray food cover with two pieces of dark gray lobe. (Fig. 2.15D) Notes: This new species is similar to its sympatric species Japonia formosana (Fig. 2.14G–I). But the later is festucine color, and having spoon like periostracum hairs (Fig. 2.16D), instead tapering tip (Fig. 2.16C). Japonia lanyuensis (Fig. 2.14D–F) is another analog, but with rougher periostracum and longer basal periostracum hairs. Measurement and type depository Holotype: SL: 4.47mm, SW: 4.70mm; APL: 2.21mm, APW: 2.33mm; NMNS5636-001, National Museum of Natural Science, Taiwan. Etymology: The name is after the boonkio (old toponym of Fen-chi-hu), primary habitat of this species. Type locality: Fen-chi-hu in Jia-i County, central Taiwan, 1400 meters in altitude, gathered from grass slope under leaves.. 13.

(22) Table 2.1 List of species included in the molecular analysis, sampling sites, and GenBank accession numbers Subfamily Genus and Species. Sampling sites/references. GenBank accession number. COI. 16S. Alycaeinae Chamalycaeus C. varius. 1. Ming-jyu Shan, Nei-hu, Taipei City. EU219770. C. varius. 2. Da-chi-jiau, Shin-dian City. EU219771. C. varius. 3. Yn-her-donq, Taipei County. EU219792. C. varius. 4. Nei-gou, Nei-hu, Taipei City. EU219791. Dioryx D. swinhoei. 5. Wu-Lai, Taipei County. EU219759. D. swinhoei. 6. Jang-hu, Jia-yi County. D. swinhoei. 7. Yue-mei waterful, San-diau-ling, Taipei County. D. swinhoei. 8. Ren-tzer, Iran County. D. swinhoei. 9. Li-jia mountain road, Taidung County. EU219758. D. swinhoei. 10. Gu-lu mountain road, Iran County. EU249291. C. iota. 11. Lake of Mandarin Ducks, Iran County. EU219766. C. iota. 12 Fwu-shan, Taipei County. EU249284. C. iota. 13. Bai-liing, Iran County. EU219764. C. iota. 14. Bai-liing, Iran County. EU249288. C. iota. 15. Bai-liing, Iran County. EU249289. C. micron. 16. li-shyng mountain road, Nan-tou County. EU219768. C. micron. 17. li-shyng mountain road, Nan-tou County. EU249275. C. micron. 18. li-shyng mountain road, Nan-tou County. EU249276. C. micron. 19. Fwu-shan, Taipei County. EU249283. C. micron. 20. Fwu-shan, Taipei County. EU249285. C. micron. 21. Gu-lu mountain road, Iran County. EU219769. C. ogaitoi. 22. Lei Gong Shan, Guei-Jou Province. EU249292. C. taiwanicum. 23. San-diau-ling, Taipei County. C. taiwanicum. 24. San-diau-ling, Taipei County. EU249282. C. taiwanicum. 25. San-diau-ling, Taipei County. EU219767. C. taiwanicum. 26. Chu-shuei-shi mountain road, Iran County. EU219765. EU219819 EU219760 EU219820 EU219821. Cyclophorinae Cyathopoma. 14. EU219793.

(23) C. taiwanicum. 27. Da-chi-jiau, Shin-dian City. EU249295. Cyclophorus C. formosaensis. 28. Charn-guang Temple, Taroko valley. EU219801. C. formosaensis. 29. Charn-guang Temple, Taroko valley. EU249274. C. formosaensis. 30. Sheau-jiau-shi, Iran County. EU249278. C. formosaensis. 31. Nan-an, Hualian County. EU219743. C. formosaensis. 32. Nan-an, Hualian County. EU219739. C. formosaensis. 33. Hong-ye, Hualian County. EU219740. C. formosaensis. 34. Torng-men, Hualian County. C. formosaensis. 35. Chorng-der, Hualian County. EU219744. C. formosaensis. 36. Hong-ye, Hualian County. EU219741. C. friesianus. 37. Shan-ping, Kaohsiung County. EU219745. C. friesianus. 38. Provincial Highway No.20, 49k, Nan-huah,. EU219799 EU219802. EU219800. EU219808. Tainan City C. friesianus. 39. Provincial Highway No.20, 49k, Nan-huah,. EU219747. Tainan City C. friesianus. 40. San-dih-men, Ping-dung County. C. friesianus. 41. San-dih-men, Ping-dung County. C. latus. 42. County highway No.100, Dung-biann village,. EU219746 EU219809 EU219737. Taichung County C. latus. 43. Dar-Guan Mountain, Tauyuan County. EU219742. C. latus. 44. Gu-guan, Taichung County. EU219738. C. latus. 45. Li-leeng mountain road, Her-ping township,. EU249281. EU219795. EU219796. Taichung County C. latus. 46. Mei-feng, Nan-tou County. EU219797. C. latus. 47. Sheau-jiau-shi, Iran County. C. latus. 48. Dong-shan, Iran County. C. martensianus. 49. Jing Shan Zoo, Wen Zhou. C. moellendorffi. 50. Mu-dan township, Ping-dung County. C. moellendorffi. 51. Shuang-liou Wood Park, Ping-dung County. EU219752. C. moellendorffi. 52. East Coastal Mountain, Taidung County. EU219749. C. moellendorffi. 53. Taidung County. EU249280. C. moellendorffi. 54. Chair-shan, Kaohsiung City. EU249286. C. moellendorffi. 55. Chair-shan, Kaohsiung City. EU219750. C. moellendorffi. 56. Chair-shan, Kaohsiung City. EU219748. C. moellendorffi. 57. Jy-been, Taidung County. C. moellendorffi. 58. Da-wu, Taidung County. EU249287 EU219794 EU219756. EU219813 EU219806. EU219804. EU219807. EU219803 EU219751. 15.

(24) C. moellendorffi. 59. Da-wu, Taidung County. EU219805. C. pyrostoma. 60. Jian Feng Ling, Hainan. EU219755. EU219814. C. subcarinatus. 61. Hong Kong. EU219757. EU219812. C. t. angulatus. 62. Okinawa. EU219754. C. t. angulatus. 63. Okinawa. EU219811. C. t. radians. 64. Iriomote Island. EU219810. C. t. radians. 65. Iriomote Island. C. cf. turgidus. 66. Provincial Highway No.149, Yun-lin County. EU219798. C. hirasei. Lydeard et al. 2002. AY010505. C. adamsi. 67. Guan-in Mountain, Taipei County. EU219838. C. adamsi. 68. Fwu-shan, Taipei County. C. adamsi. 69. The Cao-ling Historic Trail , Iran County. C. adamsi. 70. Ren-tzer, Iran County. EU249290. C. taivanus dilatus. 71. Charn-guang Temple, Taroko valley. EU249273. C. t. dilatus. 72. Chorng-der, Hualian County. C. t. dilatus. 73. Chorng-der, Hualian County. C. t. dilatus. 74. Chin-heng Bridge, Taroko valley, Hualian County. EU219841. C. t. diminutus. 75. Yongsing, Lanyu Island, Taidung County. EU219843. C. t. diminutus. 76. Yongsing, Lanyu Island, Taidung County. EU219790. C. t. peraffinis. 77. Iriomote Island. EU219789. EU219844. C. t. taivanus. 78. Jang-hu, Jia-yi County. EU249269. EU219839. C. t. taivanus. 79. Shuang-liou Wood Park, Ping-dung County. EU249270. C. t. taivanus. 80. County highway No. 136, 39.5K, Taichung County. EU249271. C. t. taivanus. 81. Liou-guei, Kaohsiung County. EU249272. C. t. taivanus. 82. Shan-ping, Kaohsiung County. EU219788. C. t. taivanus. 83. County Highway No.129, 20.5k, Jia-yi County. EU249293. C. t. taivanus. 84. Provincial Highway No.20, 79k, Bau-Lai,. EU249294. EU219753. Cyclotus. EU219786 EU219837. EU219840 EU249279. EU219842. Kaohsiung County C. t. taivanus. 85. Provincial Highway No.20, 79k, Bau-Lai,. EU219787. Kaohsiung County C. t. taivanus. 86. South Cross-Island Highway, Ping-dung County. EU249296. J. boonkioensis. 87. Fen-chi-hu in Jia-i County. EU219762. J. formosana. 88. County Highway No. 126, 26k, Miau-li County. EU219763. J. lanyuensis. 89. Yongsing, Lanyu Island, Taidung County. EU219761. Japonia. Leptopoma. 16. EU219815.

(25) L. nitidum. 90. Sha-ka-dang pavement, Taroko valley. EU219783. L. nitidum. 91. Okinawa, Nakagusuku-jo. EU249277. L. nitidum. 92. Ishigaki Island. EU219823. L. nitidum. 93. Iran County. EU219822. L. tigris. 94. Lanyu Island, Tai-dong County. L. tigris. 95. Liu-dau, Tai-dong County. EU219784. EU219825 EU219824. Pilosphaera P. yentoensis. 96. Nan-Xi river in Zhejiang province, China. EU219772. EU219818. P. zebra. 97. Nei-gou, Nei-hu, Taipei City. EU219773. EU219816. P. zebra. 98. Wu-yu sentry post, Wu-Lai, Taipei County. EU219817. Pterocyclinae Platyrhaphe P. lanyuensis. 99. Lanyu Island, Tai-dong County. EU219781. P. lanyuensis. 100. Round-the-island highway, 10-14K, Liu-dau,. EU219782. EU219834. Tai-dong County P. minutus. 101. Bei-Nan, Taidung County. EU219830. P. minutus. 102. Chair-shan, Kaohsiung City. EU219833. P. minutus. 103. Liouciou Township. EU219779. P. minutus. 104. Fashing Buddhist temple, Kaohsiung City. EU219780. P. sunggangensis. 105. Cycas taitungensis nature reserve, Taidung County. EU219778. P. s. depressus. 106. Dah-dong-shan, Jai-yi County. EU219827. P. swinhoei. 107. County highway No.100, Dung-biann village,. EU219828. EU219832. Taichung County P. swinhoei. 108. Jang-hu, Jia-yi County. EU219774. EU219826. P. swinhoei. 109. Shin-liau waterfall, Iran County (2). EU219777. EU219831. P. swinhoei. 110. Provincial Highway No.20, 49k, Nan-huah,. EU219829. Tainan City P. swinhoei. 111. County highway No.175, 3k, Tainan County. EU219775. P. swinhoei. 112. County highway No. 106B, Shih Ding, Taipei County. EU219776. P. wilsoni. 113. Frong Stone, Sin-Xu County. EU219785. P. wilsoni. 114. Frong Stone, Sin-Xu County. EU219835. P. wilsoni. 115. Wu-yu sentry post, Wu-Lai, Taipei County. EU219836. Pterocyclos. Out group Littoraria scabra. Williams et al. 2003. AJ488637. L. undulata. Williams et al. 2003. AJ488635. Truncatella guerinii. 116. Kaan-ding, Ping-dung County. 17. EU233814.

(26) Table 2.2 The range of p-distance among Cyclophoridae species of the same genus Genus. COI. 16S rRNA. Alycaeinae Chamalycaeus Dioryx. ＃＃. ＃＃. Cyclophorinae Cyathopoma Cyclophorus Cyclotus Japonia Leptopoma Pilosphaera Ptychopoma. 0.028–0.152 0.083–0.162 0.066 0.107–0.121 0.148 0.156 ＃. ＃ 0.041–0.158 0.081 ＃ 0.153 0.104 ＃. Pterocyclinae Platyrhaphe 0.103–0.172 ＃indicates only one species contained in analysis. 0.085–0.166. Table 2.3 Average pairwise differences between genera of Cyclophoridae of East Asia using COI data set. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 1. Chamalycaeus 2. Dioryx. 0.241. 3. (micron, ogaitoi). 0.227. 0.243. 4. (iota, taiwanicum). 0.212. 0.225. 0.151. 5. Cyclophorus. 0.221. 0.218. 0.216. 0.212. 6. Cyclotus. 0.220. 0.268. 0.222. 0.221. 0.221. 7. Japonia. 0.224. 0.248. 0.203. 0.204. 0.196. 0.220. 8. Leptopoma. 0.236. 0.255. 0.227. 0.218. 0.230. 0.211. 0.213. 9. Pilosphaera. 0.246. 0.266. 0.247. 0.235. 0.234. 0.248. 0.219. 0.231. 10. Platyrhaphe. 0.207. 0.229. 0.207. 0.190. 0.200. 0.197. 0.193. 0.207. 0.217. 11. Ptychopoma. 0.213. 0.241. 0.235. 0.206. 0.229. 0.206. 0.227. 0.211. 0.238. 18. 0.193. 11.

(27) Table 2.4 Pairwise differences between Japonia and Pilosphaera using COI data set. Species 1. Japonia boonkioensis. 1. 2. 3. 4. 5. 6. 7. 8. 9. 0.107 0.113 0.111 0.113 0.111 0.218 0.215 0.218. 2. Japonia formosana 57 0.122 0.121 0.119 0.121 0.218 0.220 0.226 3. Japonia lanyuensis 60 65 0.002 0.004 0.002 0.213 0.207 0.224 4. Japonia lanyuensis 59 64 1 0.002 0.004 0.213 0.207 0.222 5. Japonia lanyuensis 60 63 2 1 0.006 0.213 0.207 0.220 6. Japonia lanyuensis 59 64 1 2 3 0.211 0.205 0.222 7. Pilosphaera yentoensis n. sp. 116 116 113 113 113 112 0.026 0.156 8. Pilosphaera yentoensis n. sp. 114 117 110 110 110 109 14 0.164 9. Pilosphaera zebra 116 120 119 118 117 118 83 87 Below diagonal: total character differences. Above diagonal: p-distance. Shade areas indicate pairwise between different genera.. Table 2.5 The average p-distance of haplotype within genus of Cyclophoridae Genus. COI. 16S rRNA. Alycaeinae Chamalycaeus Dioryx. 0.034＃ 0.056＃. 0.005＃ 0.035＃. Cyclophorinae Cyathopoma Cyclophorus Cyclotus Japonia (include zebra group) Japonia (exclude zebra group). 0.106 0.111 0.066 0.181 0.113. 0.000＃ 0.083 0.056 0.185 0.071. Leptopoma Ptychopoma. 0.099 0.000＃. 0.083 0.019＃. Pterocyclinae Platyrhaphe 0.120 ＃indicates only one species contained in analysis. 19. 0.087.

(28) Fig. 2.1 Map of East Asia, with sampling sites of Cyclophoridae indicated. Solid square and asterisk indicated the sampling sites of Cyathopoma and Cyclotus respectively. Solid and open triangle indicated the sampling sites of Japonia and Pilosphaera respectively. Open circles indicated sampling sites of the others. 20.

(29) Fig. 2.2 Molecular phylogeny of Cycloporidae produced by Neighbor-joining (NJ) analysis of individual gene sequence data from COI. Relative branch support indices are given as bootstraps test consisted of 1000 iterations. Different numbers after scientific name indicate sampling site number show in table 2.1 and fig. 2.1.. 21.

(30) Fig. 2.3 Molecular phylogeny of Cycloporidae produced by maximum likelihood (ML) analysis of individual gene sequence data from COI. Relative branch support indices are given as approximate likelihood ratio test. (aLRT). Different numbers after scientific name indicate sampling site number show in table 2.1 and fig. 2.1. 22.

(31) Fig. 2.4 Molecular phylogeny of Cycloporidae produced by Bayesian analysis of individual gene sequence data from COI. Relative branch support indices are given as Bayesian probability. Different numbers after scientific name indicate sampling site number show in table 2.1 and fig. 2.1. 23.

(32) Fig. 2.5 Molecular phylogeny of Cycloporidae produced by Neighbor-joining (NJ) analysis of individual gene sequence data from 16S rRNA. Relative branch support indices are given as bootstraps test consisted of 1000 iterations. Different numbers after scientific name indicate sampling site number show in table 2.1 and fig. 2.1.. 24.

(33) Fig. 2.6 Molecular phylogeny of Cycloporidae produced by maximum likelihood (ML) analysis of individual gene sequence data from 16S rRNA. Relative branch support indices are given as approximate likelihood ratio test. (aLRT). Different numbers after scientific name indicate sampling site number show in table 2.1 and fig. 2.1.. 25.

(34) Fig. 2.7 Molecular phylogeny of Cycloporidae produced by Bayesian analysis of individual gene sequence data from 16S rRNA. Relative branch support indices are given as Bayesian probability. Different numbers after scientific name indicate sampling site number show in table 2.1 and fig. 2.1.. 26.

(35) Fig. 2.8 ESEM photograph of Cyathopoma iota (Pilsbry & Hirase, 1904) from Bai-liing, Iran County (24.525722N, 121.516083E), A: shell lateral view; B: Protoconch; C: operculum (outer view); D: radula, cm = central tooth, Ilt = inner lateral teeth, Olt = outer lateral teeth, mt = marginal teeth.. 27.

(36) Fig. 2.9 ESEM photograph of Cyathopoma taiwanicum Pilsbry & Hirase, 1906 from Da-chi-jiau, Shin-dian City (24.957972N, 121.571833E), A: shell lateral view; B: Protoconch; C: operculum (outer view); D: radula, cm = central tooth, Ilt = inner lateral teeth, Olt = outer lateral teeth, mt = marginal teeth.. 28.

(37) Fig. 2.10 ESEM photograph of Cyathopoma micron (Pilsbry, 1900) from li-shyng mountain road, Nan-tou County (24.067833N, 121.159722E), A: shell lateral view; B: Protoconch; C: operculum (outer view); D: radula, cm = central tooth, Ilt = inner lateral teeth, Olt = outer lateral teeth, mt = marginal teeth.. 29.

(38) Fig. 2.11 ESEM photograph of Cyathopoma ogaitoi (Minato, 1988) from Lei Gong Shan, Guei-Jou Province (26.404N, 108.214E), A: shell lateral view; B: Protoconch; C: operculum (outer view); D: radula, cm = central tooth, Ilt = inner lateral teeth, Olt = outer lateral teeth, mt = marginal teeth.. 30.

(39) Fig. 2.12 ESEM photograph of Cyclotus species, A1: shell lateral view of Cyclotus taivanus adams Pilsbry et Hirase, 1905, A2: radula of Cyclotus t. adams; B1: shell lateral view of Cyclotus t. dilatus Lee et Wu, 2001, B2: radula of Cyclotus t. dilatus; C1: shell lateral view of Cyclotus t. peraffinis Pilsbry et Hirase, 1905, C2: radula of Cyclotus t. peraffinis; D1: shell lateral view of Cyclotus t. taivanus H. Adams, 1870, D2: radula of Cyclotus t. taivanus. 31.

(40) Fig. 2.13 a-c: Lateral, apex and basal view of Pilosphaera shell. Holotype of Pilosphaera yentoensis n. sp. NMNS5635-001 (28°18’32.7”N; 120°32’30”E); d-f: Paratype of Pilosphaera yentoensis n. sp. NMNS5635-002 (28°18’32.7”N; 120°32’30”E); h-j: Pilosphaera zebra from Wu-Lai, Taipei, Taiwan (24° 50’57.1”N; 121°34’11.5”E). Black bar is 1mm.. 32.

(41) Fig. 2.14 a-c: Lateral, apex and basal view of Japonia shell. Holotype of Japonia boonkioensis n. sp. NMNS5636-001 (23°29’31.1N”; 120°41’59.4”E); d-f: Japonia lanyuensis from Lan-Yu, Taiwan (22°1’41.6”N; 121°34’46.8”E); g-h: Japonia formosana from Sin-Xu, Taiwan (24°34’57.1”N; 120°52’30.0”E). Black bar is 1mm.. 33.

(42) Fig. 2.15 A-B: Pilosphaera yentoensis n. sp. from Yen-To town near Nan-Xi river in Zhejiang province, China (28°18’32.7”N; 120°32’30”E); C: Pilosphaera zebra from Nei-Hu, Taipei, Taiwan (25°5’19.1”N; 121°37’42.4”E); D: Japonia boonkioensis n. sp. NMNS5636-001 from Fen-chi-hu in Jia-i County, central Taiwan (23°29’31.1”N; 120° 41’59.4”E).. 34.

(43) Fig. 2.16 Environmental Scanning Electron Microscope (FEI Quanta 200) photographs of periostracum hairs on Pilosphaera and Japonia shell. A: Paratype of Pilosphaera yentoensis n. sp. NMNS5635-002 (28°18’32.7”N; 120 °32’30”E); B: Pilosphaera zebra from Nei-Hu, Taipei, Taiwan (25°5’19.1”N; 121°37’42.4”E); C: Holotype of Japonia boonkioensis n. sp.; D: Japonia formosana from in Sin-Xu, Taiwan (24°34’57.1”N; 120°52’30.0”E). 35.

(44) 36.

(45) Chapter 3. Ring speciation and morphological adaptations of genus Cyclophorus in Taiwan. Abstract The samples of Cyclophorus span two critical links in the chain of morphologically distinct units: the transition from the round and keeled peripheral shell types from north to south Taiwan. I examined COI and 16S rRNA mitochondrial DNA of the five taxa of Taiwan Cyclophorus (see chapter 1, p10–16) and found both COI and 16S rRNA gene trees show strong geographic structures. The Mantel test show significant positive correlation between fixation index (FST) and cumulative geographic anti-clockwise distance (origin in the region around Tainan, anti-clockwise pass through south cape, Taidung, Hualian, Iran, Taipei, Taichung and meet the original populations in Jia-yi). There are finite gene flew between adjacent population. And there are series clines around the Central Range. I believe Cyclophorus of Taiwan is a proposed “ring species”. In the morphology and environmental variables correlation study, I found the currently shell morphology may be caused by the adaptation of recent long term climate.. Introduction Cyclophorids are very common snails in Taiwan, the genus Cyclophorus is the most abundant group. Koblet (1902) was the first to revise the world Cyclophorus, but no major review of the Taiwan Cyclophorus has been completed till Lee and Wu’s study (2001). There are five Cyclophorus taxa in Taiwan, C. formosaensis, C. friesianus, C. latus, C. moellendorffi and C. cf. turgidus, all of them have yellowish solid shell, with reddish brown zigzag interlard with several brown bands. It is interesting that all the northern Cyclophorus possess peripheral keeled shells and the southern Cyclophorus possess peripheral round shells. However, the northern and southern terminal forms are connected with gradual geographical variation (Lee & Wu 2001). The evolutionary processes shaping the phenotypic variation among and within relative landsnail have become a focus of research interest recently (Pfenninger & Magnin 37.

(46) 2001, Teshima et al. 2003, Pfenninger et al. 2003). There are many suggestions on the peripheral keel evolved in previous studies. Gould (1971) suggested that the peripheral keel evolved as the retention of a juvenile character. Cook and Pettitt (1979) considered that keeled shells may be more resistant to cruching than round shell. The other authors suggested the form of shells were associated with habitat types (Solem & Climo 1985, Alonso et al. 1985, Pfenninger & Magnin 2001). But it is hard to test whether an association really exists between phenotype and environment. And I am unaware of any phylogeny controlled comparative studies that have tested whether an association exists. However, since Carles R. Darwin (1859) the idea of the relation between phenotype and environment had been widely described by evolutionists. Ring species are a distinct cline where the geographical distribution is circular in shape, so that the two ends of the cline overlap with one another. The two adjacent populations rarely interbreed due to the cumulative effect of the many changes in phenotype along the cline. The populations elsewhere along the cline interbreed with their geographically adjacent populations as in a standard cline. The ring species teach us about speciation and reconstruct the pathway of speciation. Therefore, finding a potential ring species is a great interest to evolutionists. There are many cases about ring species (Jackman & Wake 1994, McKnight 1995, Irwin 2000, Irwin et al. 2001, Knijff et al. 2001); however few of the cases have characteristics of ideal ring species (Irwin et al. 2001). The topography of Taiwan provides a good chance to find the ring species because of the Central Mountain Chain. In traditional classification, there are five Cyclophorus species around mountain area of Taiwan. However, there are some intermediate forms between these Cyclophorus species (Lee & Wu 2001). These Cyclophorus species are potential ring species. The objectives of this study were using the genetic information from COI and 16S rRNA haplotype analysis as a background to study the historical biogeographical changes and quantitative morphologic differences of all described Cyclophorus taxa in Taiwan. In particular, I focus on five questions: (1) How do currently Cyclophorus biogeographical pattern cause by? (2) Is Taiwan Cyclophorus a “ring species”? (3) Does the observed phenotypic variation correspond to different evolutionary lineages? (4) How is the phenotypic variation distributed within and among populations in relation to the COI and 16S variation? And (5) Can I identify environmental variables that co-vary with population differences in morphology?. Materials and methods Populations sampled 38.

(47) In total, 57 populations of Cyclophorus were examined in Taiwan (Fig. 3.1) (Table 3.1). I do not get any living specimen (sampling site 31, 32, 33, 34, 35, 38, 49, 53) and adult individual (sampling site 18, 27, 28, 29, 46) in some sampling sites. Morphological analysis was performed on adult individuals of a subset from 52 locations, molecular analysis using 214 individuals from 46 locations. In molecular analysis, it seems hard to estimate the parameters using cumulative geographic anti-clockwise distance for all populations simultaneously. Therefore, I pooled the populations within 30 km and representing distinct geographical regions in groups. The centeral site of the groups is calculated using ArcGIS 9.2 (2006). Morphology analysis Morphological analysis was performed on adult individuals of a subset from 52 locations (205 individuals). The data set for morphological and genetic analysis overlapped in 153 individuals for sampled populations. Individuals were photographed through a Canon D1 digital camera from the aperture view. The paper prints were digitalized with a resolution of 300 pixel/inch. Eight shell morphology characters were measured by digital image analysis: W (width), H (height), h (height to periphery keel or mid-point of periphery), AW (aperture width), AH (aperture height), BW (body whorl width), BH (body whorl height) and PW (penultimate whorl width) (Fig. 3.2). H/W, h/H, PW/W, AW/W, AW/AH, BH/BW were calculated as indices of tallness/flatness, keel position, sharp/dull spire, aperture proportion, aperture shape and flat degree of body whorl, respectively. It was not possible to measure accurately the angle of the keel, so I chose three qualitative categories (Fig. 3.3). The digital images were then scored by 10 persons for ‘angularity’ and the mean used as an index of angularity. These shell morphology variables were used to perform a principal component analysis (PCA) using PCA option of the package XLSTAT (2007). Partial least square (PLS) analysis was performed using the PLS option of the package XLSTAT (2007) to assess the correlations between the quantitative shell traits and environmental variables for each population (Lin 1990). DNA preparing and sequencing The shells and soft parts were separated at laboratory. Shells were well cleaned for identification and measuring its shell characters, soft parts were stored at -80℃ until DNA extraction. DNA was extracted from columellar muscle. I extracted DNA from separate individuals using TEK-based protocol (Jiang et al. 1997) with minor modifications. Tissue was placed in TEK buffer (12.5mM Tris-Cl pH 7.3, 2.5mM EDTA, 0.4% KCl), then ground with glass pestle and incubated at 57℃ with 20µl of 39.

(48) proteinase K (20mg/ml) more than 2 hrs. The tissue extract was extracted at least twice with phenol and chloroform. 400µl DNA extract was precipitated by adding 1000µl pure ice-cold ethanol, and was then placed in -20℃ for 20 min. DNA was pelleted by centrifugation for 30 min. After 70% ethanol rinse, DNA was resuspended in distilled water and stored at -80℃ for DNA amplification. An approximately 860bp mitochondrial 16S rRNA gene was amplified by PCR using primers 16SRT (5’– ACA TAT CGC CCG TCA CTC TC–3') and 16SL900 (5’–AAA TGA TTA TGC TAC CTT TGC–3'), exactly 531bp COI using primer LCO1490 (5'–GGT CAA CAA ATC ATA AAG ATA TTG G–3') and HCO2198 (5'–TAA ACT TCA GGG TGA CCA AAA AAT CA–3') (Williams et al. 2003). PCR reactions contained template DNA 10–50ng/µl, 10pmol of each primer, 5µl 10× reaction buffer (10mM Tris-HCl, pH9.0, 50Mm KCl, 1.5Mm MgCl2, 0.1% gelatin, 1% Trinton X-100), 0.4µl 25mM/µl dNTP, 0.2µl 50mM Mg2+ and 0.4µl Taq polymerase (5unit/µl) in a total volume of 50µl. Thermal cycling for 16S rRNA was performed with an initial denaturation for 5min at 95℃, followed by 30 cycles of 30 sec. at 95℃, 45 sec. at 57℃, 50 sec. at 72℃ and ultimate extension at 72℃ for 10min, final hold in 4℃. Thermal cycling for COI gene was performed similarly, changing only its annealing temperature to 47℃. PCR products were purified using a purification kit (AMP PCR purification, Beckman) and then sequenced using an ABi 3700 autosequencer. Phylogenetic analyses Both COI and 16S rRNA sequences were combined with data of the out-group Japonia formosana Pilsbry et Hirase, 1905. Sequences were assembled and edited using Bioedit 5.0.9 (Hall 1999). All alignments employed Clustal X (Thompson et al. 1997) and were manually proofread. Codon positions within COI were tested using the incongruence length difference (ILD) test (Farris et al. 1995), as implemented by the partition homogeneity test in PAUP 4.0b10 (Swofford 1998) (100 replicates). COI sequence data were divided into two partitions, first and second codon positions in one and third codon positions in the other. Two parts of COI gene were congruent and all codon positions were combined and used in the following analysis. All data sets were subject to Neighbor-joining (NJ) using PAUP 4.0b10, to the maximum likelihood (ML) analyses using PHYML 3.0 (Guindon & Gascuel 2003), to the Maximum parsimony (MP) analysis using MEGA 4.0 (Tamura, Dudley, Nei, and Kumar 2007). The substitution model used for COI data set corresponded to the Tamura-Nei model, and included invariable sites, and rate variation among sites (TrN+I+G); for 16S rRNA data set corresponded to the two transversion-parameters model, and included invariable sites, and rate variation among sites (K81uf+I+G). These were the best models found using Modeltest 3.06 (Posada & Crandall 1998). 40.

(49) Before model fitting, the full-length sequences were tested to confirm that there was no significant heterogeneity in base frequencies across taxa (in COI: X2=63.78, df=246, P=1; in 16S rRNA: X2=31.66, df=282, P=1). NJ and MP bootstraps consisted of 1000 iterations. Reliability of ML trees were estimated by the approximate likelihood ratio test (aLRT) (using custom define model, base frequencies: A = 0.2653, C = 0.1357, G = 0.1822, T = 0.4168 in COI and A =0.3779, C = 0.0903 G = 0.1291, T = 0.4027 in 16S) using PHYML 3.0 (Guindon & Gascuel 2003). Gaps (from insertions/deletions) were treated as missing data. Neutrality was tested using Tajima's D (COI: D=-0.40890, P > 0.10; 16S rRNA: D= -1.25632, P > 0.10) and Fu & Li's F statistic (COI: F= -1.06768, P > 0.10; 16S rRNA: F= -2.12474, 0.10 > P > 0.05) (Tajima, 1989; Fu & Li, 1993). There are no selective effects on COI and 16S rRNA gene. The divergent time was estimated using MEGA 4.0 (Tamura, Dudley, Nei, and Kumar 2007). Associations between genetic, morphological, pairwise geographic distance and Cumulative geographic anti-clockwise distance between populations were explored through Mantel-tests in XLSTAT (2007) on the pairwise population fixation indices derived from AMOVA analysis (using Arlequin 3.1) (Excoffier et al. 2005), pairwise shell characters similarity (using Euclidean distance calculated by Primer 5.1.2) (2000), pairwise geographic distance and Cumulative geographic anti-clockwise distance to test for parallel evolution between genetic and morphological data, and isolation by distance, respectively.. Results Morphology analysis The first three principal components account for 34.11%, 22.83% and 17.00 % of the total variation in the matrix. The loadings of characters on the principal components reflect distinct groups of shell traits: sharp/dull spire (PC1), flat degree of body whorl (PC2), and aperture shape (PC3). Most of five taxa are complete overlap (Fig 3.4). However, the north group (sampling sites 22–57) and south group (sampling sites 1–21) are mostly separated, but show partial overlap (Fig 3.5). Phylogenetic analysis The aligned 531 bp COI gene data matrix, included 225 variable sites of which 187 (83.11%) are parsimony informative. No length difference from the out-group was detected among members of Cyclophorus. The average p-distance among Taiwan and Okinawa haplotypes is = 0.137. Average sequence divergence among Taiwan haplotypes is 0.092, ranged from 0.002 to 0.177. The alignment of the 16S rRNA 41.

(50) gene fragment data yielded 852 characters of which 407 are variable sites. Of these 407 positions, 357 (87.7%) are parsimony informative. The average p-distance among Taiwan and Okinawa haplotypes is 0.120. Average sequence divergence among Taiwan haplotypes is 0.077, ranged from 0.001 to 0.192. The inferred phylogenetic trees between the haplotypes of COI gene are shown in Figs. 3.7, 3.8 & 3.9 of 16S rRNA gene are shown in Figs. 3.10, 3.11 & 3.12. Both COI and 16S rRNA cladograms construct using different tree building method present similar topology. The cladograms indicate that the Cyclophorus populations around Tainan (sampling sites 1–11) are the origin, because they are placed nearest the root. All trees show Taichung, Nan-tou and Jia-yi populations (sampling sites 45–57) are latest divergency. The paleogeographical study indicates Taiwan Island was emerged in late Miocene orogeny approximate 6 million years ago (Chen & Wang 1996). I consider the divergent time of Taiwan Cyclophorus would not earlier than 6 million years ago. I use the timing of Taiwan emerged as a calibration point, the COI and 16S rRNA evolutionary rate of Cyclophorus are 2.28% and 2% per million years, respectively. Mantel test Pairwise FST increases significantly with Cumulative geographic anti-clockwise distance, despite the strong variation within groups (COI/ Cumulative geographic anti-clockwise distance Mantel test, r = 0.217, P = 0.003; 16S rRNA/ Cumulative geographic anti-clockwise distance Mantel test, r = 0.255, P = 0.00025) (Fig. 3.6). The not particularly high r-value, resulting from a scatter around the linear regression line, might be caused by slightly different rates of increase of FST in different spatial directions because of heterogeneity in population densities. Mantel test was also performed between molecular P-distance and Cumulative geographic anti-clockwise distance matrix. P-distance increases significantly with Cumulative geographic anti-clockwise distance and with high correlation coefficient (COI: r = 0.611, P < 0.0001; 16S rRNA: r = 0.563 P < 0.0001). I also performed Mantel test to explore associations between fixation index, molecular P-distance and pairwise geographic distance. However, the correlation coefficient is relative lower than those associated with cumulative geographic anti-clockwise distance (Table 3.3). PLS analysis PLS analysis detected significant correlation between shell traits and environmental variables for the respective sampling site (Table 3.2). Populations with a high mean temperature tend to be composed of individuals showing a tall shell with sharp spire 42.

(51) and lower keel position. Populations at high altitude tend to have flat shell with dull spire, small aperture and high keel position (Fig. 3.13) (Table 3.2). Shell angularity revealed positive correlation with warm season mean temperature, and negative correlation with cold season amount of precipitation and annual range of monthly mean temperature.. Discussion The gene trees show strong geographic structure, and suggest that the origin of Taiwan Cyclophorus is in the region around Tainan, the divergency pathway anti-clockwise pass through south cape, Taidung, Hualian, Iran, Taipei, Taichung and meet the original populations in Jia-yi. The hypothesis of possible biogeographical history matches that of Taiwan geologic history. Base on COI and 16S rRNA sequences clock and using the timing of Taiwan emerged as a calibration point. The split of east populations occurred in 1.79–2.93 million years ago. Comparing with the geologic history, Coastal Range and East Rift Valley were emerged in late Pleistocene (1.8–2million years ago) (Chen & Wang 1996, Ho 1982). This suggests the Cyclophorus has had a long history in the eastern part of its present range when Coastal Range and East Rift Valley were emerged. It is possible that the phylogeographic structure developed in a continuous isolation by distance model in the absence of any geographic break. However, the alternative, that there was geographical separation of the northern and southern parts of the Cyclophorus from east Taiwan and relatively recent contact between them, cannot presently be rejected. The correlation coefficient between fixation index, molecular P-distance and pairwise geographic distance is relative lower than those associated with cumulative geographic anti-clockwise distance. It suggests that currently populations seem to be connected throughout the ring. However, the correlation coefficient between shell traits and pairwise geographic distance is relative higher than associated with cumulative geographic anti-clockwise distance. It is because that some shell traits (e.g. peripheral keel) of Jia-yi populations are similar to unrelated, but not related taxa. The topology of the mitochondrial DNA tree also indicates that there is little association between shell peripheral type (angular or circular) and mitochondria1 DNA divergence and strongly supports the hypothesis of multiple origins of angularity. For instance, morphologically distinct angular form from Jia-yi County (sampling sites 54–57) clustered together with circular form from Taichung and Nan-tou County (sampling sites 45–52). The mitochondrial DNA evidence implies that this shell peripheral type may have arisen by parallel or convergent evolution. There are several explanations as to why shells might be keeled. One is paedomorphy. 43.