Tarebia granifera Lamarck, 1816

Veeravechsukij, Nuanpan, Krailas, Duangduen, Namchote, Suluck, Wiggering, Benedikt, Neiber, Marco T. & Glaubrecht, Matthias, 2018, Molecular phylogeography and reproductive biology of the freshwater snail Tarebiagranifera in Thailand and Timor (Cerithioidea, Thiaridae): morphological disparity versus genetic diversity, Zoosystematics and Evolution 94 (2), pp. 461-493 : 468-478

Tarebia granifera Lamarck, 1816

Type material.

4 syntypes (MHNG 1093/72/1-4).

Type locality.

Originally given as “Timor” by Lamarck (1822). This island, of which the western part is today a province of Indonesia (the eastern part, in contrast, forms the recently independent state of East Timor, or Timor Leste), was an important stop-over for major expeditions of discovery in the Indo-West-Pacific and Australia in particular (see Glaubrecht 2002). However, at that time and the time of collecting, around 1800, all expeditions known to us have anchored at the natural harbor of Kupang. Thus, we here restrict the type locality on this island to the vicinity of its western part (see Fig. 1). Nevertheless, we regard material collected recently by Vince Kessner elsewhere on this island of Timor and used in the present study as reference and for comparison as to qualify as topotypical material.

Taxonomy.

Lamarck (1816) depicted for the first time shells of this thiarid, creating the name Melania granifera , however without any further description. Later, Lamarck (1822) described this new species and its shell morphology in more detail; see also Mermod (1952: 75, fig. 137). Adams and Adams (1854) transferred Melania granifera to its own genus Tarebia . Many subsequent authors, though, referring to Lamarck (1822) continue to use the generic allocation as " Melania " granifera ; see e.g. Brot (1874-1879) in his widely used monography that was followed by most authors for nearly a century. However, the generic allocation remains vage, as e.g. Benthem-Jutting (1937) either used Thiara while she later employed Melanoides (see Benthem-Jutting 1956). Starmühlner (1976), in his thorough faunistic revision, provided an extensive list of synonyms for this taxon.

In addition, in the past some authors employed " Melania " lineata for shells found to exhibit spiral ridges and/or dark bands on its body whorls. Accordingly, Rensch (1934) divided Tarebia into two subspecies, namely " Melania " granifera granifera and " Melania " granifera lineata . In contrast, for Thailand, Brandt (1974) considered and employed Tarebia granifera as the only congeneric species to exist there; as was also done by Glaubrecht (1996).

Biogeography

The distributional range of Tarebia granifera (Fig. 1) extends from mainland Southeast Asia, with Thailand and Vietnam at its northern most margin, to the island of Taiwan and the Philippines. It also comprises, from the Malay Peninsula south and east, the region of the entire Sunda shelf area, with occurrences on the larger Sunda Islands Sumatra, Java and Borneo, as well as the islands of Nusa Tenggara (or Lesser Sunda islands), i.e. from Bali east to Timor. The species is also abundant in Wallacea, i.e. on Sulawesi and on several islands of the Moluccas (e.g. Halmahera, Ceram, Ambon). From there, it extends east into the Indo-West Pacific, with occurrences in western and eastern New Guinea and the Bismarck Archipelago.

In Thailand, this species occurs in most lentic and lotic water bodies ranging throughout the various regions, provinces and river systems. There, T. granifera was found in both natural and artificial water bodies on a variety of substrata, such as e.g. sand, mud, rock (and, alternatively, concrete bridge foundations, concrete walls), on bottoms of reservoirs, irrigation canals and ornamental ponds. This species is usually found together with other thiarids, most often with M. tuberculata and Mieniplotia scabra . We were not able to correlate any consistent ecological features that clearly distinguish either at particular locations or specific habitat and/or populations where T. granifera was found to occur. Thus, the ecological requirements of this taxon, in particular contrasting those to that of other thiarids, remain insufficiently known.

Material examined

In the following we document here in detail the geographical origin of material studied from Thailand, in comparison with the syntypes as well as topotypical material from Timor as reference (see above). Data on other localities indicated in Fig. 2 to depict the extension of the entire distribution range of the species, will be provided and analysed elsewhere (Glaubrecht et al., in prep.).

Thailand:

Pai drainage (Salween river system): Mae Hong Son province: Pang Mapha district, Huai Pa Hung, 19°22 ’20” N, 098°26 ’36” E, 435 m (SUT 0515083, 03. V. 2015); Mueang Mae Hong Son district, Huay Nam Kong, 19°28 ’34” N, 098°07 ’02” E, 425 m (SUT 0515081, 03. V. 2015); Tham Pla, 19°25 ’31” N, 097°59 ’27” E, 300 m (SUT 0515077, 02. V. 2015); Pai River, 19°21 ’55” N, 097°58 ’11” E, 215 m (SUT 0515078, 02. V. 2015); Huay Sua Tao, 19°15 ’32” N, 097°54 ’45” E, 235 m (SUT 0515079, 02. V. 2015).

Moei drainage (Salween river system): Tak province: Tha Song Yang district, check point near Moei River, 17°13 ’23” N, 098°13 ’34” E, 130 m (SUT 0515075, 02. V. 2015); Mae Salit Luang harbour, 17°26 ’05” N, 098°03 ’33” E, 110 m (SUT 0515076, 01. V. 2015); Mae Sot district, Ban Wang Takhian, 16°42 ’39” N, 098°30 ’22” E, 195 m (SUT 0515073, 30. IV. 2015); Thong Dee harbour, 16°41 ’39” N, 098°31 ’04” E, 205 m (SUT 0515072, 30. IV. 2015); Ban Huay Muang, 16°40 ’58” N, 098°31 ’07” E, 200 m (SUT 0515074, 30. IV. 2015).

Ping drainage (Chao Phraya river system): Chiang Mai province: Chom Thong district, Mae Soy bridge, 18°17 ’23” N, 098°39 ’04” E, 270 m (SUT 0514054, 24. VI. 2014); Ban Huay Phang, 18°17 ’09” N, 098°39 ’17’’ E, 260 m, SUT 0514050, 25. VI. 2014; Ban Mae Suai Luang, 18°17 ’04” N, 098°39 ’15” E, 270 m (SUT 0514051, 25. VI. 2014); Ban Mai Saraphi, 18°16 ’26” N, 098°38 ’54” E, 275 m (SUT 0514052, 25. VI. 2014); Tak province: Mueang Tak district, Ban Pak Huay Mae Tho, 16°52 ’29” N, 099°07 ’14” E, 105 m (SUT 0516126, 10. III. 2016).

Wang drainage (Chao Phraya river system): Lampang province: Chae Hom district, Wang river, 18°56 ’01” N, 099°38 ’55” E, 375 m (SUT 0514045, 23. IV. 2014); Ban Thung Hang stream, 18°52 ’48” N, 099°40 ’01” E, 375 m (SUT 0514044, 23. IV. 2014); Huay MaeYuak, 18°46 ’40” N, 099°38 ’39” E, 350 m (SUT 0514046, 22. IV. 2014); km. 40 + 075 bridge, 18°42 ’15” N, 99°35 ’32” E, 330 m (SUT 0516124, 09. III. 2016).

Yom drainage (Chao Phraya river system): Phayao province: Chiang Muan district, Thansawan waterfall, 18°51 ’22” N, 100°11 ’09” E, 230 m (SUT 0516119, 08. III. 2016); Phrae province: Mueang Phrae district, Mae Nam Saai km 9/457 bridge, 18°05 ’03” N, 100°13 ’00” E, 170 m (SUT 0516108, 07. III 2016); Sung Men district, Mae Marn reservoir, 18°00 ’51” N, 100°08 ’23” E, 205 m (SUT 0516113, 07. III 2016); Sukhothai province: Si Satchanalai district, Tat Duen waterfall, 17°33 ’16” N, 099°29 ’48” E, 135 m (SUT 0516103, 06. III. 2016).

Nan drainage (Chao Phraya river system): Nan province: Bo Kluea district, Wa river, 19°11 ’30” N, 101°12 ’13” E, 715 m (SUT 0515090, 11. VI. 2015); Ban Luang district, Huay Si Pun reservoir, 18°51 ’45” N, 100°28 ’37” E, 430 m (SUT 0516114, 08. III. 2016); Uttaradit province: Tha Pla district, Kaeng Sai Ngam, 17°52 ’20” N, 100°18 ’02” E, 255 m (SUT 0516112, 07. III 2016); Kaeng Wang Wua, 17°52 ’30” N, 100°18 ’26” E, 230 m (SUT 0513019, 28. VI. 2013); Huai Nam Re Noi, 17°52 ’51” N, 100°16 ’15” E, 270 m (SUT 0513023, 28. VI. 2013); Laplae district, Mae pool waterfall, 17°43 ’42” N, 099°58 ’50” E, 125 m (SUT 0516109, 07. III. 2016).

Khek drainage (Chao Phraya river system): Phitsanulok province: Nakhon Thai district, Huai Nam Sai, 17°01 ’08” N, 100°55 ’36” E, 215 m (SUT 0515086, 20. V. 2015); Ban Kaeng Lat, 16°57 ’21” N, 100°55 ’31” E, 325 m (SUT 0515087, 20. V. 2015); Wang Thong district, Kaeng Sopha, 16°52 ’13” N, 100°50 ’17” E, 415 m (SUT 0516118, 08. III. 2016); Poi waterfall, 16°50 ’36” N, 100°45 ’16” E, 200 m (SUT 0515067, 08. II. 2015); Khao Kho district, Kaeng Wang Nam Yen, 16°37 ’24” N, 100°54 ’01” E, 710 m (SUT 0516121, 09. III. 2016); Rajapruek resort, 16°36 ’01” N, 100°54 ’30” E, 705 m (SUT 0516120, 09. III. 2016); Phetchabun province: Khao Kho district, Huai Sa Dao Pong, 16°34 ’24” N, 100°59 ’24” E, 320 m (SUT 0516123, 10. III. 2016); Kaeng Bang Ra Chan, 16°32 ’52” N, 100°54 ’03” E, 600 m (SUT 0515088, 21. V. 2015).

Pa Sak drainage (Chao Phraya river system): Phetchabun province: Lom Sak district, Than Thip waterfall, 16°39 ’46” N, 101°08 ’10” E, 375 m (SUT 0516130, 11. III. 2016); Khao Kho district, Samsipkhot waterfall, 16°32 ’26” N, 101°04 ’58” E, 385 m (SUT 0516129, 11. III. 2016); Wichian Buri district, Ban Wang Ta Pak Moo 13, 15°47 ’54” N, 101°14 ’08” E, 120 m (SUT 0514041, 27. VI. 2014); Huai Leng, 15°47 ’52” N, 101°13 ’54” E, 115 m (SUT 0514042, 27. VI. 2014); Ban Wang Tian, 15°47 ’30” N, 101°13 ’31” E, 120 m (SUT 0514040, 27. VI. 2014); Huay Range reservoir at Ban Wang Ta Pak, 15°47 ’19” N, 101°15 ’07” E, 140 m (SUT 0514043, 27. VI. 2014); Lop Buri province: Phatthan Nikhom district, Suanmaduea waterfall, 14°55 ’12” N, 101°13 ’11” E, 135 m (SUT 0516132, 26. IV. 2016); Sara Buri province: Muak Lek district, Dong Phaya Yen waterfall, 14°44 ’06” N, 101°11 ’31” E, 155 m (SUT 0516133, 26. IV. 2016). Nakhon Sawan province: Mueand Nakhon Sawan district, Bungboraped, 15°41 ’00” N, 100°14 ’59” E, 30 m (SUT 0516127, 10. III. 2016).

Loei drainage (Mekong river system): Loei province: Phu Ruea district, Pla Ba waterfall, 17°23 ’25” N, 101°22 ’27” E, 665 m (SUT 0515068, 07. II. 2015); Phu Luang district, km. 50/350 at Loei River, 17°04 ’38” N, 101°29 ’21” E, 675 m (SUT 0516125, 10. III. 2016); Tatkoktup waterfall, 17°03 ’04” N, 101°31 ’39” E, 690 m (SUT 0516128, 10. III. 2016).

Chee drainage (Mekong river system): Khon Kaen province: Mueang Khon Kaen district, Bueng Thung Sang, 16°34 ’46” N, 102°50 ’23” E, 170 m (SUT 0515064, 05. II. 2015).

Moon drainage (Mekong river system): Nakhon Ratchasima province: Pak Thong Chai district, Lamphraphloeng reservoir, 14°35 ’32” N, 101°50 ’30” E, 260 m (SUT 0516131, 22. III. 2016).

Khwae drainage (Mae Klong river system): Kanchanaburi province: Thong Pha Phum district, Hindad hot spring, 14°37 ’26” N, 098°43 ’41” E, 160 m (SUT 0515091, 27. VI. 2015); Sai Yok district, Sai Yok Yai waterfall, 14°26 ’03” N, 098°51 ’15” E, 105 m (SUT 0515092, 27. VI. 2015); Sai Yok Noi waterfall, 14°14 ’28” N, 099°03 ’56” E, 115 m (SUT 0515093, 27. VI. 2015).

Phachi drainage (Mae Klong river system): Kanchanaburi province: Dan Makham Tia district, Ban Thung Makham Tia, 13°54 ’18” N, 099°23 ’08” E, 45 m (SUT 0515061, 17. III. 2015); Ban Ta Pu, 13°51 ’18” N, 099°22 ’59” E, 55 m (SUT 0515060, 17. III. 2015); Ban Nong Phai, 13°46 ’45” N, 099°25 ’27” E, 70 m (SUT 0515059, 17. III. 2015); Ratchaburi province: Chom Bueng district, Phachi River bridge, 13°45 ’01” N, 099°26 ’27” E, 65 m (SUT 0515058, 17. III. 2015); Ban Dan Thap Tako, 13°41 ’28” N, 099°29 ’08” E, 80 m (SUT 0515057, 17. III. 2015); Ban Pa Wai, 13°37 ’00” N, 099°24 ’37” E, 75 m (SUT 0515056, 17. III. 2015); Suan Phueng district, Lum Nam Phachi, 13°32 ’54” N, 099°21 ’42” E, 110 m (SUT 0515070, 23. I. 2015); Huai Ban Bor, 13°32 ’07” N, 099°20 ’32” E, 135 m (SUT 0515071, 23. I. 2015); Huay Nueng, 13°32 ’52” N, 099°17 ’34” E, 155 m (SUT 0515069, 23. I. 2015); Suan Phueng district, Ban Purakom, 13°19 ’29” N, 099°14 ’22” E, 275 m (SUT 0515066, 23. I. 2015).

Mae Klong river system: Nakhon Pathom province: Mueang Nakhon Pathom district, pond on campus of Silpakorn University, 13°49 ’01” N, 100°02 ’28” E, 80 m (SUT 0515055, 13. I. 2015).

Gulf of Thailand: Rayong province: Mueang Rayong district, Mae Rumphueng beach (Mae Rumphueng canal), 12°37 ’50” N, 101°20 ’35” E, 10 m (SUT 0516135, 28. IV. 2016); Phetchaburi province: Cha-am district, Khlong Cha-am (Cha-am canal), 12°48 ’03” N, 099°58 ’53” E, 20 m (SUT 0513032, 16. X. 2013); Prachuap Khiri Khan province: Mueang Prachuap Khiri Khan district, Khlong Bueng reservoir, 11°55 ’29” N, 099°42 ’40.9” E, 70 m (SUT 0516146, 11. V. 2016); Huai Yang district, Khlong Huai Yang (Yang canal), 11°36 ’50” N, 099°40 ’08” E, 55 m (SUT 0514037, 23. XI. 2014); Bang Saphan district, Kar on waterfall, 11°26 ’14” N, 099°26 ’33” E, 55 m (SUT 0514038, 23. XI. 2014); Chumphon province: Tha Sae district, Krapo waterfall, 10°44 ’29” N, 099°12 ’55” E, 75 m (SUT 0511149, 2. VII. 2011); Surat Thani province: Tha Chang district, Khlong Tha Sai (Takhoei canal), 09°12 ’40” N, 099°11 ’56” E, 10 m (SUT 0516147, 04. VI. 2016); Phunphin district, Ban Tung Ao (Ta Khoei canal), 09°12 ’26” N, 099°12 ’26” E, 5 m (SUT 0516148, 04. VI. 2016); Don Sak district, Vibhavadi waterfall (Tha Thong canal), 09°08 ’07” N, 099°40 ’32” E, 25 m (SUT 0516142, 09. V. 2016); Ban Na San district, Dat Fa waterfall, 08°52 ’19” N, 099°25 ’59” E, 80 m (SUT 0514048, 22. XI. 2014); Khlong Klai (Nong Noi canal), 08°48 ’07” N, 099°26 ’45” E, 110 m (SUT 0516137, 9. V. 2016); Nakhon Si Thammarat province: Nopphitam district, Khlong Prong (Klai canal), 08°47 ’23” N, 099°38 ’13” E, 100 m (SUT 0516139, 09. V. 2016); Krung Ching waterfall, 08°43 ’17” N, 099°40 ’15” E, 195 m (SUT 0516145, 09. V. 2016); Phatthalung province: Si Banphot district, Khlong Tha Leung (Tha Nae canal), 07°42 ’48” N, 099°51 ’34” E, 70 m (SUT 0516138, 08. V. 2016); Songkhla province: Singhanakhon district, Khlong Sathing Mo (Songkhla lake), 07°13 ’37” N, 100°31 ’42” E, 10 m (SUT 0516144, 08. V. 2016); Khlong Hoi Khong district, Khlong La reservoir, 06°52 ’29” N, 100°19 ’48” E, 60 m (SUT 0516141, 07. V. 2016); Khlong Cham Rai reservoir, 06°49 ’30” N, 100°19 ’50” E, 55 m (SUT 0516143, 07. V. 2016).

Andaman Sea: Krabi province: Mueang Krabi district, Khlong Sai (Khlong Sai canal), 08°10 ’20.8’’ N, 098°47 ’38’’ E, 25 m (SUT 0515097, 30. X. 2015); Wang Than Thip (Wang Than Thip canal), 08°09 ’49” N, 098°47 ’51” E, 20 m (SUT 0515098, 30. X. 2015); Trang province: Yan Ta Khao district, Khlong Palian (Palian canal), 07°22 ’11” N, 099°40 ’48” E, 20 m (SUT 0515095, 29. X. 2015).

Timor Leste: Manatuto district, W bank of Laclo river near Condae, ca. 4 km WSW of Manatuto, 08°31 ’32” S, 125°58 ’50” E, 35 m (ZMH 119364, 21 VI. 2012); south coast, 3.8 km N of Nancuro beach, 4.7 km SE of Natarbora, 09°00 ’31” S, 126°03 ’45” E, 20 m (ZMH 119359, 13. XI. 2011); 3.4 km N of Nancuro beach, 5 km SE of Natarbora, 09°00 ’45” S, 126°03 ’49” E, 20 m (ZMH 119358, 13. XI. 2011); 2.5 km N of Nancuro beach, 5.7 km SE of Natarbora, 09°01 ’11” S, 126°03 ’58” E, 15 m (ZMH 119354, 13. XI. 2011); Baucau district, NE of Baucau, Watabo beach, 08°26 ’36” S, 126°28 ’11” E, 20 m (ZMH 119357, 9. XI. 2011); Lautem district, Ira-Ara village, Lutu-Ira, 08°20 ’32” S, 127°01 ’08” E, 100 m (ZMH 119356, 23. V. 2011); near the Baucau/Lautem district border marker, 11.8 km NE of Laga, 08°25 ’35” S, 126°41 ’43” E, 5 m (ZMH 119353, 10. XI. 2011); Bobonaro district, north coast, 0.5 km from the mouth, Large seasonal stream in Batugade, 08°56 ’47” S, 124°58 ’28” E, 10 m (ZMH 119362, 20. V. 2012); Viqueque district, Ossu subdistrict, near village Usu Decima, Wai-eu-Lau, 08°44 ’36” S, 126°22 ’50” E, 670 m (ZMH 119355, 13. V. 2011); spring in the village, Loihuno, 08°47 ’05” S, 126°22 ’32” E, 255 m (ZMH 119360, 11. XI. 2011); spring in the village, Loihuno, 08°47 ’05” S, 126°22 ’32” E, 255 m (ZMH 119363, 17.V. 2012); Manufahi district, south coast, Fatuhcahi village, Wetetefuik creek, 09°02 ’00” S, 125°59 ’36” E, 30 m (ZMH 119361, 12 XI. 2011).

Phylogenetic analyses

The final alignment of the cox1 sequences had a length of 658 base pairs (bp) and that of the 16S sequences 781 bp. Genetic p-distances for cox1 sequences of specimens determined as T. granifera from Thailand ranged from 0% to 14.7%, whereas all cox1 sequences obtained from specimens from Timor Leste were identical.

For 16S sequences, p-distances among specimens from Thailand ranged from 0% to 10.4% and for Timor Leste, pairwise p-distance between specimens were very low, ranging from 0% to 0.1%.

All three phylogenetic analyses recovered two deeply divergent clades of specimens assigned to T. granifera (clades A and B, Fig. 4), with high to very high support (clade A, PP: 1.00, BS (ML): 95, BS (MP): 100; clade B, PP: 1.00, BS (ML): 90, BS (MP): 100). Genetic p-distances between these two clades were distinctly higher than p-distances within either clade A or clade B, 13.8% for cox1 and 10% for 16S sequences. Genetic p-distances within clade A were with 0% to 3.34% for cox1 and 0% to 1.44% for 16S sequences rather low.

All specimens from Timor Leste were included in clade A together with specimens mostly from the southern to southern-central parts of Thailand (Fig. 4), viz. those from the provinces Songkhla, Trang, Krabi, Nakhon Si Thammarat, Surat Thani, Chumphon, Prachuap Khiri Khan, Phetchaburi, Ratchaburi, Kanchanaburi, Nakhon Pathom, Sara Buri and Nakhon Sawan. But this clade included also specimens from the northern part of the country, viz. Chang Mai, Lampang, Phrae and Phitsanulok, and specimens from Nakhon Ratchasima and Rayong in northeast to eastern Thailand. Within clade A, relationships among specimens were generally not well-supported (Fig. 4). However, there is a general pattern that Thai specimens of T. granifera assigned to clade A were more frequent in the southern part of the country.

In contrast, specimens of T. granifera assigned to clade B were more frequent in the northern part of Thailand, i.e. the majority of specimens in this clade originate from the northern to north east Thai provinces, such as Chang Mai, Mueang Mae Hong Son, Phayao, Lampang, Nan, Uttaradit, Tak, Sukhothai, Phitsanulok, Phetchabun and Loei, while only few specimens in this clade are from the southern-central Thai provinces Phatthalung, Nakhon Si Thammarat, Surat Thani, Ratchaburi, Kanchanaburi and Lop Buri. Almost all specimens assigned to clade B were placed in a polytomy in the tree shown in Fig. 4. Corresponding to the results of the phylogenetic analyses, genetic p-distances within clade B were very low, with 0% to 0.46% for cox1 and 0% to 0.52% for 16S sequences.

When analysed by drainage systems, we found that all specimens from the north-western part of Thailand, which is drained through the Salween river system into the Andaman Sea, were included in clade B. Likewise, specimens from the headwaters of the Ping, Wang, Yom and Nan rivers belonging to the Chao Phraya system, with few exceptions, were assigned to clade B in the phylogenetic analyses. In the lower courses of northern to northern-central Thai drainages, such as e.g. the Chao Phraya and Mae Klong drainages that run into the Gulf of Thailand, specimens assigned to both clades are present.

Similarly, specimens belonging to both mitochondrial clades are present in the Mekong drainage, whereas specimens assigned to clade A predominate in the smaller rivers in the Thai parts of the Malay Peninsula to the north and south of the Isthmus of Kra that either drain into the Gulf of Thailand or the Andaman Sea (Fig. 4). Noteworthy are a few populations from the somewhat more elevated parts of the provinces Surat Thani (SUT 0516137), Nakhon Si Thammarat (SUT 0516139) and Phatthalung (SUT 0516138) on the Malay Peninsula that were assigned to clade B (Fig. 4).

In contrast to this geographical pattern in Tarebia granifera , with broadly speaking an essentially southern clade A and an essentially northern clade B, we found no correspondence of specimens from the three morphotypes with the two genetically differentiated clades as outlined above as all morphs were present in both clades (data not shown).

Haplotype networks, molecular species delimitation and dating

Evolutionary relationships among haplotypes were inferred applying a median-joining network approach that showed the two mitochondrial clades A and B to be separated by> 60 steps (cox1 and 16S; Fig. 5a, b), while within these clades haplotypes were separated by usually only a few steps (Fig. 5a, b).

The ABGD approach suggested that the T. granifera clades A and B could be classified as two species for prior intraspecific divergences (d) of the combined cox1 and 16S data set of d ≥ 0.0077. The bGMYC analysis (Fig. 5c) recovered a probability of conspecifity of less than 0.05 for specimen pairs belonging to both, the mitochondrial clades A and B. For specimen pairs assigned to clade A in the phylogenetic analyses a probability of conspecifity of more than 0.7 was recovered, with most pairs having a probability of conspecifity of more than 0.95. All specimen pairs assigned to clade B in the phylogenet ic analyses were assigned a probability of conspecificity of more than 0.95 in the bGMYC analysis.

The results of the BEAST analysis assuming a strict molecular clock and a divergence rate of 1% per million years (Fig. 5d) suggests, following the split of Tarebia ( granifera) from Thiara ( amarula ) at about 7.1 million years ago (Mya), a separation of the mitochondrial clades A and B at about 5.3 Ma BP (95% highest posterior density interval (HPD): c. 6.5-4.0 Mya). The diverification within clade A is suggested to have started c. 0.65 Mya (95% HPD: 0.95-0.45 Ma BP), while the slitting within clade B occurred presumably c. 0.33 Mya (95% HPD: 0.50-0.25 Mya).

Shell morphology

The shells of Tarebia granifera (Fig. 2), which are often of greenish or brownish colour, are medium-sized, with 12 to 44 mm, of elongately ovate-conoidal or turreted shape, much shorter than Melanoides and rather thick, the body whorl being greater in length than half the entire length of the shell. The spire is usually sharp, the whorls are not much convex, almost flat in the spire. The sculpture consists of spiral grooves and tubercles on the whorl. The shape of the aperture is oval with sharp peristome and curved columella; the umbilicus is closed.

As shown in Fig. 2 Tarebia granifera exhibits a wide phenotypical spectrum of shell morphology, which varies with respect to size and shape and in particular in sculpture and colouration including banding patterns. We separated, based on superficial “Gestaltwahrnehmung” of morphologically distinct shells, three groups called morphs A, B and C here, without implying morphotypes in the sense of species under a respective species concept, but for convenience only and to fasciliate further research into the potential correlation of phenotypical and genetic proprinquity.

Starting off from the type series of T. granifera from Timor (Fig. 2a) and comparing to topotypical material collected in Timor Leste (Fig. 2 t–y) we distinguished based on phenotype only three major morphologies, comprising a combination of several distinct features, which taken together allows to differentiate the three morphs. The first (morph A) is similar to and characteristic by shell features also visible in the Timor types (Fig. 2 b–g), with shell shape ovate-conoidal to moderately turreted and rather thick; the apex is pointed and often eroded; the colour is highly variable, ranging from yellowish-brown to dark brown and even nearly black. The number of whorls is mostly between 3 and 7, with a high spire and regularly increasing size. The body whorl is large and measures about half the length of the shell. The sculpture consists of spiral grooves and tubercles on the whorl, the suture is shallow. Next we separated those shells as morph B which agree to features similar to the description of T. lineata (Gray, 1828), as shown in Fig. 2(h-m), with the shell being moderately thick and elongately or ovate-conoidal, with 3-9 whorls and the body whorl being two-thirds of the shell. The colour is mostly yellowish-brown to dark brown. The sculpture of these shells were found to have small brown spiral ridges on the whorl, sometimes built as rows of tubercles. Morph C is represented by shells which combine features from both of the former morphs, but were differentiated here primarily due to the pronounced banding pattern (Fig. 2 n–s).

We were not able to find any correlation of shell morphology with molecular genetic clusters as described above, or any other geographical or ecological factor matching these distinct phenotypes in Tarebia granifera .

Biometry

For ranges and mean values of measured shell parameters for the different predefined groups, i.e. shell morphs/geographic groups or genetic clades, see Table 2. For all but one of the shell parameters tested, at least one group was present that was not normally distributed (Shapiro-Wilk-test, p <0.05). The exception was the length of the last three whorls (l3w). Here, normally distributed data was found in every tested shell morph/geographic group (Shapiro-Wilk-test, p> 0.05). Hence, we conducted an ANOVA, scoring significant (p <0.05) followed by a Bonferroni-corrected LSD-test. The latter found significant differences (p <0.025) between the means of morph B and C. For all other parameters we performed a Kruskal-Wallis-rank sum test, significant (p <0.05) for shell height and width, but not for the index of lw3/w (p> 0.05). Hence, the latter was found to contain no differences between groups. For shell height, a subsequent Bonferroni-corrected Dunn-test identified significant differences between the means of morph A and B (p <0.025). The same test identified significant differences of means in shell width between morph B and C (p <0.025). It has to be noted, however, that the ranges of all measured shell parameters widely overlap and, therefore, do not qualify as diagnostic characteristics (see boxplots in Fig. 6 a–d).

Between genetic clades at least one of the groups was found to be not normally distributed (Shapiro-Wilk-test, p <0.05) for shell width and l3w/w. By contrast normal distribution was found for lw3 and shell height. Subsequent Levene-testing identified the height and l3w data sets as homoscedastic (p> 0.05), hence a two-sample t-test was performed, identifying significant differences (p <0.025) between the means for the two clades for lw3 and no significant differences for shell height. For shell width and l3w/w a Wilcoxon signed rank test was performed, revealing significant differences (p <0.025) for the mean of both shell parameters. However, similar to the situation when comparing the different shell morphs/geographical groups, it has to be noted that the ranges of all measured shell parameters widely overlap and, therefore, do not allow to derive diagnostic characteristics for the two main clades found in the phylogenetic analyses (see boxplots in Fig. 7 a–d).

Geometric morphometrics

A principal component analysis (PCA) identified the first six major axes to account for a relevant proportion of variance (p> 0.05) (PC1: 0.303; PC2: 0.181; PC3: 0.117; PC4: 0.090; PC5: 0.058; PC6: 0.052), explaining a cumulative proportion of 0.801 of variance.

Principal components (PC) 1-6 had all at least one group that proved to be not normally distributed (Shapiro-Wilk-test, p <0.05). Subsequent Kruskal-Wallis-testing was significant (p <0.05) in PC1-5 and not significant in PC6. Hence, no further testing was done for PC6. The Bonferroni-corrected Dunn-test identified the mean value for specimens from Timor to be significantly different (p> 0.025) from all other morphs on PC1. By contrast, examining PC2 and PC4 with the same test, proved morph A and B to be the only groups not significantly different (with regard to mean values) from one another. Finally, on PC3 and PC5 the Bonferroni-corrected Dunn-Test revealed the mean value of morph C not to be significantly different from all other groups, but the means of morph A and B to be significantly different to that of the specimens from Timor.

Finally, when morph C was integrated into morph B (since these were only differentiated on the basis of slight differences in banding pattern), PC1-5 supported only the group consisting of specimens from Timor to have significantly different means from all other specimens (data not shown). The scatter plot in Fig. 6e shows the distribution of PC1 vs. PC2, illustrating that all predefined groups widely overlap, which indicates that a clear separation is not possible on the basis of shell shape.

For PC1 and PC3-6 at least one of the groups (clade A/clade B) was not normally distributed (Shapiro-Wilk-test, p> 0.05). Hence, we conducted Wilcoxon singed rank tests for all these PC, with none showing significant differences between groups (p> 0.05). By contrast, in PC2 both groups showed normally distributed data. Therefore, Levene-testing based on deviations from the mean followed and was found significant (p <0.05). Accordingly, we conducted Welch’s two sample t-test, revealing significant differences between the means of the two clades on PC2. The scatter plot in Fig. 7e shows the distribution of PC1 vs. PC2, illustrating that the clusters of specimens assigned either to clade A or clade B widely overlap, which indicates that a clear separation is not possible on the basis of shell shape.

Brood pouch content

Females of Tarebia granifera were found to contain embryos and shelled juveniles in their “marsupium”, or subhemocoelic brood pouch, situated in the neck region as in other thiarids studied so far. They usually release crawling juveniles with shells comprising several whorls that are built before hatching from the brood pouch. In this study, we found the snails to possess brood pouches filled with all ontogenetic stages, ranging from early to late embryos and six additional size classes of juveniles, with shells measuring between less than 0.5 to more than 3 mm (see Figs 8-10a).

The frequency of these different size classes in the subhemocoelic brood pouch of the total of n = 1,007 dissected females of Tarebia granifera from a total of 107 populations from Thailand (n = 95) and Timor Leste (n = 12) is shown as to their geographic occurrence for the two mitochondrial clades A (n = 42) and B (n = 53) as well as the predefined morphs A, B and C in Figs 8 and 9 a–c. Although the content of the brood pouch varied considerably among individuals and populations, no geographic pattern could be observed, neither for the populations within Thailand nor for those from Timor Leste. We were also unable to find any specific pattern in the distribution of the eight ontogenetic stages in correlation with the two genetic clades A and B or for the different predefined shell morphs (Figs 8, 9).

In all examined populations, the number of early and late embryonic stages was above 50%, in most cases even above 75%; see Fig. 10a for the composition of the brood pouch contents according to the three morphs A-C, and see Fig. 10c for those of the two mitochondrial clades. Nevertheless, in nearly all populations shelled juveniles of the size between less than 0.5 to more than 3.0 mm were present in the female’s brood pouches; with the only exception for females (n = 1 and 9) from two populations of morph A and C, both in locations in the south in streams draining to the Gulf of Thailand (see Fig. 9a, b).

When considering the overall distribution of different size classes in the different morphs/geographic clusters or mitochondrial clades, the resulting histograms (Fig. 10a, c) all show essentially the same composition of ontogenetic stages, which suggests the presence of the same reproductive strategy in all investigated groupings. The overall ratio of non-gravid vs. gravid specimens was 164:943 (= 17.4%). Among the 255 dissected specimens assigned to morph A, 21 snails were found to be non-gravid (= 8.2%), while among the 652 dissected snails assigned to morph B, in 123 of these no offspring was observed (= 18.9%). For morph C, the ratio of non-gravid vs. gravid specimens was 11:128 (= 8.6%) and that ratio for specimens from Timor Leste was 9:72 (= 12.5%) (Fig. 10b). Considering the two main mitochondrial clades, similar values were observed (Fig. 10d), with the proportion of gravid females well above 85%.

We also compared the size class composition of offspring in the subhemocoelic brood pouches of Tarebia populations from different drainage systems. Although considerable variation was present among the rivers and streams of the 17 drainage systems in Thailand (Fig. 11a), clear differences could not be observed. There is, however, one possible exception, i.e. females of T. granifera from the Moei River in the Northwest of Thailand, where a very low amount of early embryonic stages and less later embryonic stages were found, while there was the largest proportion of larger shelled juveniles. Also, there is a slight trend for populations in streams and rivers in the south of Thailand, both draining into the Gulf of Thailand and the Andaman Sea, to exhibit higher proportions of the earliest embryonic stages.

The distribution of gravid vs. non-gravid specimens according to the 17 rivers systems exhibits some variation (Fig. 11b), albeit with usually (far) more gravid specimens present in all populations; but again with the exception of females from populations in the Northwest of Thailand, in particular from the rivers Moei, Ping and Pai. The populations in Moei River are in this respect exceptional because only there we found more non-gravid than gravid specimens. Conversely, all females from populations in the rivers Chao Phraya, Loei, Chee, Moon, Khwae, Mae Klong and from streams of the Andaman Sea were found to be gravid, with no non-gravid specimens at all detected in our samples.

Whether reproduction is seasonal, or whether there is any influence of the month of collecting on our data, can currently not be answered with certainty. In an attempt to correlate reproduction (i.e. the frequency of gravid vs. non-gravid females) with climatic effects such as, for example, rainy season resulting in high water levels in rivers and streams, we have used published meteorological data (e.g. minimum/maximum temperature and precipitation) for stations representing the different climatic regions of Thailand, viz. Chiang Mai for northern inland region, Ko Samui for the Gulf of Thailand and Phuket for the Andaman Sea localities (see map in Fig. 8 for these locations). As is evident from Fig. 12, specimens collected in populations from inland places were to a high proportion gravid females at the end of winter ( January–February) and into the summer season ( March–June). During this first half of the year the proportion of gravid females somehow reflect percipitation in so far, as there is a trend to be high when it is dry (see Fig. 12a); also the proportion of non-gravid females increases towards the rainy season in the North of Thailand (April/May). At localities in the Gulf of Thailand region, high numbers of specimens with brood pouch content were found both during the little ( May–June) and great ( Oct–Nov.) rainy season; however, we lack sufficient collecting data for the dry season (Fig. 12b). For the Andaman Sea region, only specimens collected during the rainy season were available, reflecting in general the picture from the Gulf region, though; with ~25% non-gravid specimens at the beginning and only gravid specimens shortly after the peak of the rainy season (Fig. 12c).