Integrative taxonomy and biogeographic affinities of the first freshwater sponge and mollusc association discovered in tropical Asia

2.1 Data sampling

In total, we morphologically examined five specimens of the freshwater sponge Corvospongilla ultima (Annandale, 1910) [=Corvospongilla siamensis Manconi & Ruengsawang, 2012 syn. nov.] (Demospongiae: Spongillidae) [two specimens were sequenced]; 18 specimens of the freshwater clam Limnoperna siamensis (Morelet, 1866) [five specimens were sequenced]; four specimens of the freshwater clam Limnoperna fortunei (Dunker, 1857) [all of which were sequenced] (Bivalvia: Mytilidae); and 28 specimens of the freshwater mussel Hyriopsis khoratensis Pfeiffer et al., 2021 (Bivalvia: Unionidae) [eight specimens were sequenced].

The samples of freshwater sponge and mollusc association were collected from a deep pool (depth ca. 5 m) of the Mun River near Tha Tum town (15.3297°N, 103.6821°E) by a local free diver (Table 1 and Figure 1). The images of living specimens were taken using Canon EOS 7D digital camera (Canon Inc.) with Canon EF 100 mm f/2.8L IS USM macro lens (Canon Inc.). Samples of additional mollusc taxa were collected from rivers of Thailand, South Korea, and Vietnam, that is, H. khoratensis and L. fortunei (Table 1).

2.2 Morphological studies

2.3 Molecular analyses

2.4 Phylogenetic analyses, divergence dating, and biogeographic reconstructions

3.1 Discovery of the Mekong's freshwater sponge-mollusc association

In the Mun River, we found an epifaunal association of a freshwater sponge (Corvospongilla, Spongillidae) and two species of freshwater bivalve molluscs (Figures 1-4). The freshwater mussel species (Hyriopsis, Unionidae) serves as the substrate for a sponge, while the byssus-attaching clam (Limnoperna, Mytilidae) mostly uses the sponge's crust surface as a substrate for its attachment (Figure 2a). The freshwater mussel individuals seem to be negatively threatened by the sponge encrustation that hampers opening of the shell and, hence, in some cases, may reduce water intake by the mussel due to partly blocked siphons (Figure 2b).

3.2 Taxonomic position of the freshwater sponge species

Based on morphological survey, the sponges collected in the Mun River in Thailand belong to C. siamensis Manconi & Ruengsawang, 2012. Our collection site is situated within the same basin but only 150 km away from its original type locality, the Pong River. Furthermore, we found that C. siamensis from the Mun River is nearly identical to C. ultima from the Kaladan River in Myanmar based on sequences of five molecular markers, that is, the COI (no substitutions; p-distance = 0%), 28S (no substitutions; p-distance = 0%), and ITS1 + 5.8S + ITS2 (one substitution; p-distance = 0.2%) fragments. These values are within the range of intraspecific variation in freshwater sponges (e.g., Addis & Peterson, 2005).

Our 28S phylogeny indicated that this species takes a distant position in the order Spongillida being a sister lineage to a few other available taxa in this group (Figure 5). The same pattern was found on the basis of the ITS1 + 5.8S + ITS2 phylogeny (Figure 6). These findings indicated that Corvospongilla is a valid and phylogenetically distant genus of freshwater sponges. Whether this genus is a member of the Spongillidae as it was thought until recently based on morphological data or belongs to other family, cannot be understood from our phylogenies. Both the 28S and ITS1 + 5.8S + ITS2 consensus phylogenetic trees clearly demonstrated that the family Spongillidae in its current understanding is a polyphyletic group (Figures 5 and 6). Members of smaller families such as Lubomirskiidae, Metaniidae, Metschnikowiidae, and Potamolepidae were found to be subclades/lineages within the Spongillidae.

Our PTP species delimitation modeling returned a number of MOTUs, most of which correspond well to certain species such as Spongilla lacustris (Linnaeus, 1759), Ephydatia fluviatilis (Linnaeus, 1759), Ephydatia muelleri (Lieberkühn, 1856), and Radiospongilla cerebellata (Bowerbank, 1863) (Figure 6). Conversely, C. siamensis from the Mun River and C. ultima from the Kaladan River were recorded within a single MOTU with a high probability value (p < 0.001). The lack of clear genetic differences between these nominal taxa together with the results of our species delimitation modeling strongly supported the conclusion that C. siamensis and C. ultima are conspecific. Furthermore, the range of this species (i.e., C. ultima with its junior synonym C. siamensis syn. nov.) is broader than it was assumed earlier and extends from India through Myanmar to the Mekong River Basin.

3.3 Taxonomic position and evolutionary biogeography of the byssus-attaching clam

The byssus-attaching clam from the sponge and mollusc association from the Mun River is a member of the genus Limnoperna Rochebrune, 1882 (Figure 4). This genus was thought to contain several marine taxa but one widespread freshwater species, that is, the golden mussel L. fortunei (see section 4 and section 5 for detail and references). Conversely, our samples from the Mun River were found to be a sister species-level lineage to L. fortunei (Figure 7). Based on available taxonomic names, this species should be considered L. siamensis stat. rev. that was described from the Mekong Basin (see section 4 for detail).

Our multi-locus time-calibrated phylogeny revealed that Limnoperna is a sister clade to Sinomytilus Thiele, 1934, other freshwater mytilid genus from tropical Asia (Figure 7). The origin of the most recent common ancestor (MRCA) of these clades was placed in the Early Miocene (mean age of the crown group = 19.1 Ma, 95% HPD 9.9–29.5 Ma). The ancestral habitat reconstruction indicated that this MRCA was a freshwater species (probability = 83.5%; Figure 7). Hence, freshwater Mytilidae in Asia represents a monophyletic group originated via a single colonization event from marine to freshwater environment followed by a subsequent radiation in continental water bodies. The divergence event between L. siamensis and L. fortunei most likely occurred at the Miocene—Pliocene boundary (mean age = 5.3 Ma, 95% HPD 2.7–8.1 Ma). The MRCA of these species was almost certainly a freshwater mollusc (probability = 99.3%).

The Limnoperna + Sinomytilus clade was recovered as a sister group to the endolithic species Leiosolenus lischkei Huber, 2010 [=Lithophagus curtus Lischke, 1874 (unavailable name, secondary homonym)] (Figure 7).

3.4 Taxonomic position of the freshwater mussel species

The freshwater mussel species serving as a substrate for the Mekong's epifaunal association of a freshwater sponge and byssus-attaching clam belongs to the genus Hyriopsis Conrad, 1853 (Figure 8). Based on sequences of three DNA markers (COI, 16S, and 28S), our samples must be considered H. khoratensis, a recently described species endemic to the Mun River basin (Pfeiffer et al., 2021).

4.1 Family Spongillidae Gray, 1867 (Porifera: Demospongiae: Spongillida)

Figures 2 and 3.

Material examined. THAILAND: Surin Province, Mekong River Basin, Mun River near Tha Tum town, 15.3297°N, 103.6821°E, deep pool with clay and stony bottom substrate, depth ca. 5 m, March 07, 2018, 3 specimens, local collector leg. MYANMAR: Kaladan River, 21.0094°N, 92.9813° E, submerged siltstone rocks with boreholes, 01.05.2015, 2 specimens, Bolotov, Vikhrev, Aksenova, Konopleva, & local villagers leg.

Re-description. Growth form from thick crusts to massive roundish cushion on shells or rocks. Color light brown, yellowish, greenish. Consistency extremely hard, stony, firm. Basal spongin plate present. Surface flat or notably irregular and conulose, with conules supported by short ascending pauci- to multispicular tracts. Skeleton as an alveolar (isotropic) network of megascleres with paucispicular polygonal meshes and ascending tracts supporting conules. Megascleres oxeas or strongyles (201–306 × 17–29.2 μm) covered with single to moderately abundant microspines, hardly seen even under SEM. Microscleres micropseudobirotules with smooth bent shaft (25–48 × 2–2.5 μm) bearing distal smooth pseudorotules with 4–8 hooks. Gemmules large (up to 1000 μm), hemispherical, sole or grouped, strictly adhering to the substratum. Pneumatic layer of fixed gemmules reduced. Gemmular cage easily detachable, constituted by megascleres and smaller strongyles of various length (12–162 μm) covered with spines or tubercles. The cage forms a dark brown mat extending to 1–5 gemmules. Gemmuloscleres short strongyles with bent spines (30–55.2 × 6.2–22.5 μm). Foramen single, lateral, porus tube prominent (about 60 × 40 μm).

Distribution. India (Annandale, 1910, 1912; Jakhalekar & Ghate, 2016; Soota, 1991), Myanmar (Bolotov, Aksenova, et al., 2018), and Thailand (Ruengsawang et al., 2012; this study). Inhabits solid substrata in rivers, ponds, and reservoirs.

Comments. The description of morphology and measurements given here is based not only on the original material, but also on samples studied by other authors. Measurements of spicules for specimens from India, Myanmar, and Thailand are presented in Table 3.

We have compared a sponge found in the Mun River to another sponge previously found in the Kaladan River, Myanmar (Bolotov, Aksenova, et al., 2018). Morphological analysis showed that these two sponges are distinct according to the current taxonomic system (Manconi & Pronzato, 2019; Ruengsawang et al., 2012). The sponge from the Mun River was identified as C. siamensis due to microspined strongyles as megascleres, pseudobirotules as microscleres, and spined strongyles as gemmuloscleres. The sponge from Myanmar was identified as C. ultima due to microspined oxeas as megascleres, pseudobirotules as microscleres, and spined strongyles as gemmuloscleres (Annandale, 1910; Bolotov, Aksenova, et al., 2018; Jakhalekar & Ghate, 2016; Manconi & Pronzato, 2004). Both sponges highly fit the original and consequent descriptions. Identification of these species was also confirmed by the location of the sampling sites. Corvospongilla siamensis from the Mun River was found relatively close to the type locality of this nominal species, which is also situated within the Mun Basin (Ruengsawang et al., 2012). In its turn, our sample of C. ultima was collected from a costal river of the Bay of Bengal in northwestern Myanmar housing a derivative of the Indian fauna (Bolotov, Aksenova, et al., 2018). This finding aligns with the Indian origin of the type specimen (Annandale, 1910).

A notable difference between these two sponges occurs in the shape of megascleres: C. ultima has oxeas (see suppl. figure 3 in Bolotov, Aksenova, et al., 2018), while C. siamensis has strongyles. However, our comprehensive genetic study revealed that the sponges share identical sequences of the COI and 28S fragments. Furthermore, these two samples share a single nucleotide substitution in the ITS1 + 5.8S + ITS2 region (uncorrected p-distance = 0.2%). The DNA sequence analysis and the PTP species delimitation model unambiguously indicate that the samples from Myanmar and Thailand belong to a single species and that C. siamensis should be considered as a junior synonym of C. ultima.

4.2 Family Mytilidae (Mollusca: Bivalvia: Mytilida)

Figures 2 and 4.

Material examined. THAILAND: Surin Province, Mekong River Basin, Mun River near Tha Tum town, 15.3297°N, 103.6821°E, deep pool with clay and stony bottom substrate, depth ca. 5 m, attached to freshwater mussel and sponge crusts, 07 March 2018, 5 specimens [RMBH biv0463, including biv0463_1 and biv0463_2 being sequenced], local collector leg.; Surin Province, Mekong River Basin, Mun River near Tha Tum town, 15.2966°N, 103.5965°E, attached to petrified wood and hard clay fragments, 09.iii.2018, 13 specimens [RMBH biv0719, including biv0719_1, biv0719_2, and biv0719_3 being sequenced], local collector leg.

Diagnosis. Apparently, there are no reliable diagnostic features to separate L. siamensis from L. fortunei based on morphological characters. However, L. siamensis differs from its sister species by 36 and 21 fixed nucleotide substitutions in the COI and 16S gene fragments, respectively. The diagnostic substitutions in the COI gene fragment are as follows: 17 G, 23 A, 29 A, 77 G, 83 G, 98 G, 101 C, 122 A, 137 C, 161 G, 191 A, 206 G, 207 C, 227 C, 230 A, 260 T, 284 G, 296 A, 302 A, 317 G, 332 A, 350 G, 371 C, 374 G, 422 A, 431 T, 437 A, 473 A, 479 G, 494 A, 503 A, 518 G, 549 T, 563 G, 582 C, and 638 T. The diagnostic substitutions in the 16S gene fragment are as follows: 44 A, 48 G, 108 G, 114 T, 138 C, 227 A, 233 G, 242 T, 248 A, 249 C, 250 G, 255 A, 280 G, 296 G, 301 Del, 303 T, 321 A, 336 G, 343 T, 361 G, and 408 T. The mean uncorrected p-distances are 6.6% and 4.0% for the COI and 16S gene fragments, respectively. Intraspecific differences in the 28S gene fragment of L. siamensis and L. fortunei were not recorded. Finally, these species share strictly allopatric distribution that can also be used as a supplementary feature for species-level diagnostics, although the rapid human-mediated range expansion of L. fortunei may change this pattern in the near future.

Re-description. The shell length of adult individuals up to 20 mm, the shell height up to 10 mm (larger specimens can probably be found). The shell shape typical for Mytilidae, rounded-triangular, elongated wedge-shaped, pointed anteriorly, and rounded posteriorly, with almost straight or concave ventral margin and obtuse keel on dorsal side (Figure 4a,b). In ventral margin, there is narrow chink for the byssus. The shell is attached to the substratum by an array of byssal threads (Figure 4-8). The umbo is placed anteriorly. The ligament internal and relatively long, situated between the hinge plates but visible externally. The hinge plate narrow, without teeth. The outer surface of the shell of adult individuals dark brown to black, with olive-green lines between coarse growth lines (Figure 4a,b). The shell of small individuals yellowish-white on the ventral margin and umbo, and burgundy or reddish-brown on the dorsal aspect (Figure 4-8). Internally, the shell of L. siamensis nacreous. The internal shell layer slightly lustrous, darker postero-dorsally, and lighter antero-ventrally. The nacre purple or gray-blue above and white below the keel at the anterior margin (Figure 4c,d). The shell varies somewhat in size, shape, and coloration. The internal architecture of the shell is illustrated in Figure 4c–c4,d–d1. Apical septum absent. Large posterior adductor muscle scar and two byssal retractor muscle scars present (Figure 4-8). There is also a small antero-ventral anterior adductor muscle scar and an anterior pedal retractor muscle scar underneath the ligament (Figure 4c1,c4), and ctenidial attachment muscle scars (Figures 4c3,d). The valves thin and brittle. The mantle is fused on the dorsal side and between the exhalant siphon and the inhalant aperture (Figure 4g). The ctenidial-labial palp junction typical of the Mytiloidea in general (Figure 4-8). Foot finger-shaped, with byssus.

Distribution. Mekong Basin and, probably, some adjacent freshwater systems such as the Chao Phraya River. Numerous occurrences of L. fortunei from the Mekong Basin (e.g., Morton & Dinesen, 2010; Ng et al., 2020) belong to L. siamensis.

Comments. Earlier authors listed L. siamensis and several other taxa described from the Mekong and Chao Phraya as valid species (e.g., Brandt, 1974). Conversely, Morton (1973) assumed that L. siamensis could be synonymous with L. fortunei based on the similarity of conchological and anatomical characters. The conspecificity of these two taxa was widely accepted until recently (Morton et al., 2020; Ng et al., 2020), with a plethora of names listed as synonyms of a single widespread species, L. fortunei (e.g., Morton, 2015; Morton & Dinesen, 2010). Here, we revise the status of the two species and propose L. siamensis stat. rev.

Two peculiar nominal species of Limnoperna were described from rapids and waterfalls, that is, L. depressa Brandt & Temcharoen, 1971 from the Mekong River in Laos, and L. supoti Brandt, 1974 from the Chao Phraya Basin in Thailand (Brandt, 1974; Brandt & Temcharoen, 1971). Interestingly, the latter species was described as a dwarf clam that exclusively inhabits eroded tips of Brotia shells (Gastropoda: Pachychilidae) (Brandt, 1974). Here, we tentatively assign both L. supoti and L. depressa to be probable synonyms of L. siamensis, although these remarkable “white-water” populations need future DNA-based studies of topotypes.

The nominal taxon Modiola taprobanensis Preston, 1915 [type locality: Ceylon] cannot be linked to the freshwater Mytilidae of Southeast Asia (i.e., Limnoperna and Sinomytilus taxa) as several authors assumed (Beu, 2006: 191) because it was described from marine waters of Sri Lanka (Preston, 1915). Furthermore, this nominal taxon was recorded only once since its original description, that is, from the saline Ennur backwater, Madras [currently, the Ennore Creek, 13.2313°N, 80.3239°E, Chennai, Tamil Nadu, South India] (Preston, 1916: 35). There are no doubts that it is a saltwater mytilid bivalve, which is nothing to do with Limnoperna.

4.3 Family Unionidae (Mollusca: Bivalvia: Unionida)

Genus Hyriopsis Conrad, 1853

Type species: Unio delphinus Gruner, 1841 [=Hyriopsis bialatus Simpson, 1900]

Figure 2.

Material examined. THAILAND: Surin Province, Mekong River Basin, Mun River near Tha Tum town, 15.3297°N, 103.6821°E, deep pool with clay and stony bottom substrate, depth ca. 5 m, 07.iii.2018, one specimen [RMBH biv0459, being sequenced], local collector leg.; Surin Province, Mekong River Basin, Mun River near Tha Tum town, 15.3575°N, 103.6637°E, 07.iii.2018, eight specimens [RMBH biv0465, including biv0465_1 and biv0465_2 being sequenced], local collector leg.; Surin Province, Mekong River Basin, Mun River near Tha Tum town, 15.2966°N, 103.5965°E, 07.iii.2018, five specimens [RMBH biv0469, including biv0469_1 and biv0469_2 being sequenced], local collector leg.; Maha Sarakham Province, Mekong Basin, Chi River, 16.2158°N, 103.2526°E, 11.iv.2014, 14 specimens [RMBH biv0130, including biv0130_1, biv0130_2, and biv0130_3 being sequenced], Bolotov and Vikhrev leg.

Diagnosis. Conchologically, H. khoratensis can be distinguished from Hyriopsis bialata by having less elongate and more ovate shell (Pfeiffer et al., 2021). Both these species share a large, triangular dorsal wing.

Distribution. Endemic to the Mun River watershed within the Khorat Plateau (Pfeiffer et al., 2021).

Comments. In earlier revisions, Hyriopsis samples collected from the Mun and Chi rivers in Thailand were listed as H. bialata Simpson, 1900, which was thought to be widespread throughout Southeast Asia (e.g., Brandt, 1974). Bolotov, Kondakov, et al. (2017) and Bolotov, Vikhrev, et al. (2017) found that a separate Hyriopsis species inhabits the Chi River and that this species is distinct phylogenetically from the true H. bialata Simpson, 1900 from Malaysia. A year later, Do et al. (2018) suggested that the name H. gracilis Haas, 1910 could be used for the Mekong's species, which was historically identified as H. bialatus Simpson, 1900. Finally, Pfeiffer et al. (2021) resolved this taxonomic puzzle. These authors revised the entire genus Hyriopsis based on a comprehensive DNA sequence dataset and described H. khoratensis as a species new to science (Pfeiffer et al., 2021).

5.1 Sponge-mollusc associations in fresh waters of Asia and South America

To the best of our knowledge, we report on the first discovery of a freshwater sponge-mollusc association in Asia. Furthermore, the association described here is the only example discovered by us during a long-term survey of freshwater bivalves throughout tropical and temperate Asia in 2011–2020 (Myanmar, Laos, Thailand, Vietnam, Malaysia, Korea, and the Russian Far East) (Bolotov, Aksenova, et al., 2018; Bolotov, Klass, et al., 2019; Bolotov, Kondakov, et al., 2017, 2020; Bolotov, Pfeiffer, et al., 2018; Bolotov et al., 2014; Bolotov, Vikhrev, et al., 2017; Konopleva, Pfeiffer, et al., 2019). At first glance, it could be assumed that such associations are not so common in the region.

Conversely, our data can be biased by sampling approaches because we primarily collected samples by wading and snorkeling from smaller water bodies and shallow places of larger rivers and lakes (e.g., Bolotov, Konopleva, et al., 2020). Volkmer-Ribeiro et al. (2019) noted that their team was able to collect large series of freshwater sponge-mollusc associations from the Xingu River by bottom trawling and surface-supplied diving at depth 5–30 m, and that these associations were not discovered there based on traditional sampling approaches. Our examples from the Mun River were collected by a local free diver in a deep pool with stony bottom, while we did not find such associations in shallower places of the river examined ourselves.

Generally, the freshwater sponge-mollusc associations discovered in the Xingu River also comprise a freshwater sponge (Spongillidae), a freshwater mussel (Hyriidae), and a byssus-attaching clam (Dreissenidae) (Volkmer-Ribeiro et al., 2019). Hence, the Mekong's association can be considered analogous to those discovered in South America. Based on the classification scheme of Volkmer-Ribeiro et al. (2019), the Mekong's association belongs to the BSA type, that is, sponges encrusted on the posterior region of bivalve shell surface around siphonal area.

5.2 Note on Corvospongilla taxonomy: does a spicule shape reflect species limits?

These findings are not surprising given the intricate taxonomy of Corvospongilla. The genus comprises 19 species to date (Van Soest et al., 2020). It can be distinguished from other taxa by having a specific combination of spicule types: microscleres shaped as pseudobirotules and gemmuloscleres shaped as stout spiny strongyles or oxeas (Manconi & Pronzato, 2002, 2004). The species-level concept of this genus is based exclusively on morphological features, some of which are quite inconspicuous. For example, Corvospongilla micramphidiscoides Weltner, 1913 and Corvospongilla victoriae Annandale, 1914 share similar spicules (Manconi & Pronzato, 2004, 2009), and their taxonomic status as separate species is supported only by different proportion of oxeas in the megasclere composition.

The ability of widely used morphological traits to mark true species borders should be carefully checked. For instance, C. siamensis differs from Corvospongilla burmanica (Kirkpatrick, 1908) only by the presence of microspines on the megasclere surface (Ruengsawang et al., 2012), while Corvospongilla loricata (Weltner, 1895) differs from C. burmanica by the presence of spiny megascleres supplementing smooth strongyles (Kirkpatrick, 1908; Manconi & Pronzato, 2004). Currently, the usability of megasclere microspines for Corvospongilla taxonomy can be questioned, since the majority of holotype descriptions lack SEM photographs. These spines can be extremely small and totally invisible under LM, as those in a C. ultima specimen described here (Figure 3), or hardly visible, as those in a specimen from Myanmar (Bolotov, Aksenova, et al., 2018). Moreover, in Spongillida the number of spines often varies from spicule to spicule, for example, in E. fluviatilis (Manconi & Pronzato, 2002), and from sponge to sponge, for example, in Spongilla alba Carter, 1849 (our unpublished data) or Metschnikowia tuberculata Grimm, 1877 (Sokolova et al., 2020). Many spicules should be meticulously checked under SEM to be sure that none of the spines was missed.

The general shape of megascleres is also considered informative in current taxonomy of the genus (Manconi & Pronzato, 2004). Thus, C. ultima has sharpened spicules (oxeas), while C. siamensis has blunt ones (strongyles) (Annandale, 1910; Ruengsawang et al., 2012). Shape of megascleres in freshwater sponges is, however, not as stable as one would like. For example, Rezvoj (1925) described a new species, Eunapius rotundacuta Rezvoj, 1925, based on the megasclere strongyles that were unusual for the known representatives of the genus. Later, more specimens were investigated revealing different strongyles-to-oxeas ratio in certain individuals, and this taxon was reconsidered as a subspecies of Eunapius carteri (Bowerbank, 1863) (Rezvoj, 1926). Same unstable strongyles-to-oxeas ratio was observed in specimens of S. alba living in adjacent waterbodies but having identical DNA sequences (our unpublished data). Furthermore, many Corvospongilla representatives comprise an additional megasclere type besides the main type (Manconi & Pronzato, 2004, 2019), which also impedes understanding of the character's reliability. Finally, the developmental stage of a spicule also influences its tips shape. Megasclere shape seems therefore not so useful in species-level identification. Our genetic analysis suggests its total inconsistency at least within Corvospongilla, since a sponge with exclusive oxeas has been proved to have identical DNA sequences to a sponge with exclusive strongyles.

The structure of gemmuloscleres, being a relevant diagnostic character in other freshwater sponges, shows some stability in Corvospongilla species (Manconi & Pronzato, 2004, 2009, 2019; Ruengsawang et al., 2012). Gemuloscleres are represented by spiny strongyles in the majority of species (10 species), but five species have also spiny oxeas or oxeas-strongyloxeas: Corvospongilla caunteri Annandale, 1911, Corvospongilla lemurensis Manconi & Pronzato, 2019, Corvospongilla thysi (Brien, 1968), Corvospongilla seckti Bonetto & Ezcurra de Drago, 1966, and Corvospongilla mesopotamica Manconi & Pronzato, 2004. Two species, that is, Corvospongilla novaeterrae (Potts, 1886) and Corvospongilla bhavnagarensis Soota, Pattanayak & Safena, 1984, have oxeas as the only gemmuloscleres. Gemmules are absent in the type material of Corvospongilla sodenia Brien, 1969 and Corvospongilla zambesiana (Kirkpatrick, 1906) (see Manconi & Pronzato, 2004), which decreases the credibility of these species.

Microscleres seem to be the most stable character within Corvospongilla. Pseudobirotules are rare within freshwater sponges but always occur in Corvospongilla taxa (Manconi & Pronzato, 2002). Three species, that is, C. mesopotamica, C. victoriae, and C. micramphidiscoides, possess spiny oxeas or strongyles as microscleres in addition to the pseudobirotules, but this feature was not stressed in the current classification scheme.

Spicule complement of the gemmular cage in Corvospongilla usually includes megascleres, gemmuloscleres, and plenty of spicules of intermediate types (Manconi & Pronzato, 2004; Ruengsawang et al., 2012; Jakhalekar & Ghate, 2016; Bolotov, Aksenova, et al., 2018: suppl. figure 3 AND figure 4e2). In identification keys, these spicules do not attract much attention (Manconi & Pronzato, 2004, 2016; Soota, 1991), and some of species descriptions lack information on the cage composition.

Finally, our results call into question the taxonomic potential of the morphological trait “shape of megascleres” in Corvospongilla. We obviously need genetic data to reveal which morphological traits are informative in every genus of freshwater sponges. It is also important to describe spicules from more specimens when reporting on new findings, since extremely little data on natural variability of this group exist.

5.3 Taxonomy and evolutionary biogeography of the freshwater Mytilidae

Morton et al. (2020) revealed that the septate Sinomytilus harmandi and the aseptate L. fortunei are closely related taxa and that the MRCA of these two species colonized Asian freshwaters relatively recently. Our statistical modeling results align with these earlier findings. In particular, our time-calibrated phylogeny suggests a Miocene age for the MRCA of Sinomytilus and Limnoperna (mean age = 19.1 Ma, 95% HPD 9.9–29.5 Ma), while the ancestral habitat reconstruction indicates that this ancestral mollusc most likely was a freshwater species (probability = 83.5%). The origin of Novaculina Benson, 1830 (Bivalvia: Pharidae), another secondary freshwater genus from Asia, was also placed in the Miocene, with the MRCA being a freshwater taxon (Bolotov, Vikhrev, et al., 2018).

It was suggested that Limnoperna and Sinomytilus could be placed in a separate subfamily, the Limnoperninae Scarlato & Starobogatov, 1979 (Lee et al., 2019; Morton et al., 2020). Since the DNA-based subfamily-level classification scheme of the Mytilidae is still not developed, we would agree with this preliminary placement. The Limnoperna + Sinomytilus clade sisters to the endolithic Leiosolenus (Lee et al., 2019 [as “Lithophaga”]; Morton et al., 2020; this study), the latter genus could be transferred from the Lithophaginae Adams & Adams, 1857 to Limnoperninae. If Sinomytilus comprises freshwater taxa only (Brandt, 1974; Morton & Dinesen, 2010), the genus Limnoperna appears to harbor a plethora of fossil and recent marine species (e.g., Beu, 2006: 191). Although molecular sequences of these saltwater species are not available, our phylogenetic reconstruction raises some doubts that they can belong to the freshwater Limnoperna + Sinomytilus clade, which seem to colonize fresh waters since the Early Miocene.

The golden mussel L. fortunei attracts the full attention of scientists and stakeholders as an invasive species rapidly spreading in South America, Japan, and Hong Kong, and causing large economic and environmental impacts (Morton, 1973, 2015; Pessotto & Nogueira, 2018; Ricciardi, 1998). Although a number of molecular studies were already focused on this species, including the whole genome sequencing (De Paula et al., 2020; Uliano-Silva et al., 2018), any DNA sequence data on its populations from the Mekong Basin were not available. It was thought that the genus Limnoperna mostly contains brackish and marine taxa but one freshwater species, L. fortunei, which is widespread from the Mekong Basin to China (Beu, 2006; Graf, 2013; Morton, 1973, 2015; Morton & Dinesen, 2010; Morton et al., 2020). Hence, the presence of two vicariate species of Limnoperna, that is, L. siamensis in the Mekong Basin and L. fortunei in East Asia, is an unexpected finding of the present study that agrees with available data on distribution of other freshwater animals. The allopatric ranges of the razor clam species Novaculina siamensis Morlet, 1889 (Mekong, Bang Pakong, and Pa Sak rivers) and Novaculina chinensis Liu & Zhang, 1979 (Yangtze River) seem to reflect the same biogeographic pattern (Bolotov, Vikhrev, et al., 2018). It was shown that freshwater mussels assemblages of the Sundaland (Mekong, Chao Phraya, Mae Klong and the drainages of the Malay Peninsula and Greater Sunda Islands) and East Asian (rivers east and north of the Mekong Basin up to the Russian Far East) subregions are clearly different even at the subfamily and tribe levels (Bolotov, Pfeiffer, et al., 2018). Three subfamilies, that is, Parreysiinae, Gonideinae (with large radiations of the Contradentini, Rectidentini, and Pseudodontini), and the monotypic Modellnaiinae, prevail in the Sundaland Subregion, while the Unioninae and Gonideinae (with Lamprotulini, Gonideini, and Chamberlainiini) form the basis of the East Asian fauna (Bolotov, Konopleva, et al., 2020; Bolotov, Pfeiffer, et al., 2018). However, one member of the East Asian Unioninae is known to occur in the Mekong Basin, that is, Cristaria cf. plicata (Leach, 1814) (Bolotov, Pfeiffer, et al., 2018; Zieritz et al., 2018). Among freshwater fish assemblage, Carassius praecipuus Kottelat, 2017 was described from central Laos as the unique Mekong's representative of this East Asian genus (Kottelat, 2017a). Several other fish taxa of East Asian affinities such as Vanmanenia orcicampus Kottelat, 2017 and Rhinogobius milleri Chen & Kottelat, 2003 are known to occur in the Mekong's tributaries in Laos (Kottelat, 2017a,2017b). In summary, there were several faunal exchanges between the Mekong and East Asian rivers, most likely via ancient stream captures. The timing of these freshwater connections and subsequent vicariance events needs future studies using time-calibrated phylogenies of various freshwater taxa. As for the L. fortunei—L. siamensis species pair, it indicates a divergence event at the Miocene—Pliocene boundary (mean age = 5.3 Ma).