Morphological and molecular analyses of Anodontinae species (Bivalvia, Unionidae) of Lake Baikal and Transbaikalia

The diversity and taxonomy of anodontine species in Lake Baikal and Transbaikalia region has been contentious since it is based on a typological species concept, the so called “Comparatory Method”. Using this method, six Comparatory anodontine species have been described for the study area as belonging to the genus Colletopterum. This genus was separated from Anodonta based on shell characteristics and further split into two subgenera, i.e. Colletopterum sensu stricto and Colletopterum (Piscinaliana). However, many authors do not recognize this separation maintaining all Colletopterum forms within Anodonta. The current study clarifies the taxonomy and systematics of Anodontinae in this region, using a combination of molecular, morphological and anatomical data. All previously recognized Comparatory forms are here recognized as a single species, i.e. Anodonta anatina.


Introduction
It is well established that the high intraspecific phenotypic shell variability of freshwater mussels (Bivalvia: Unionida) is intrinsically connected with environmental variables [1], [2], [3], [4], [5], [6]. Additionally, other factors, such as sex and ontogeny may also influence shell shape [7], [8], [9], [10]. Due to these reasons, the use of shell shape is limited to identify taxonomic units in freshwater mussels. In fact, this has been a persistent and contentious problem that resulted in exaggerating the extant number of freshwater mussel species [11], [12]. The Russian taxonomy for freshwater bivalves, the Typological Species Concept with the Comparatory Method' (TCS-CM) [13], [14], uses mainly the arc of maximal convexity of the valve's outline (AMCVO) and the index of convexity (ratio of shell width to shell height) for species delimitation [15], [16]. This system is known to inflate the number of species, especially those with high morphological plasticity of shells, which only reflect different ecophenotypes. In a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 summer of 2004 and autumn of 2016. (Fig 1). All specimens were measured to the nearest 0.1 mm including shell length (L), shell width (B), maximal shell height (H m ), and distance from umbo apex to the anterior shell margin (l). Additionally, the dimensions and shell characteristics of 29 specimens from European Russia were extracted from Zhadin [5] and Bogatov & Kijashko [33] for morphological and statistical analyses. Since freshwater mussels (Colletopterum sp.) are not rare, endangered or protected species, no permits are required for the collection of these mussels, in the study region. Additionally, field work did not involve any territory inside national parks or other protected areas. The map of the study area and collection site maps were built using QGIS 3.0 (Fig 1).

Morphometry
Standard morphometric shell characters, i.e. B/H, H/L and l/L ratios were calculated according to Bogatov et al. [16] (Panels 1-3 in Fig 2). TCS-CM species identification was based on shell shape, B/H and l/L indexes, and standard curves of the Arc of Maximal Convexity of the Valve's Outline (AMCVO) [16], [36] (Panels 4-6 in Fig 2). The age of each specimen was estimated by counting annual growth rings. Statistical analysis was performed using Microsoft Excel, 2010.

DNA extraction, PCR, sequencing and species identification
A total of 24 specimens collected from the study area, including at least 2 individuals per site and all previously identified TCS-CM species, were sampled for genetic analyses (Table 2). For comparison, six specimens from Ukraine were also included ( Table 2). In detail, a small piece of foot tissue was clipped and stored individually in 96% ethanol. DNA extraction, PCR and sequencing conditions of F-type mtDNA cytochrome c oxidase subunit I (COI) followed Klishko et al. [20] (Annealing temperature 48-50ºC). Sequences were assembled using Chro-masPro 1.7.4 (Technelysium, Tewantin, Australia). Alignment was performed in Bioedit 7.2.5 [37] including all newly sequenced individuals and all sequences from the most recent and Map of the study area with anodontine specimens sampling sites (red circles). A-location of Lake Baikal in Russia territory, B-Map of Lake Baikal and Transbaikalia region; Lake Baikal: 1 -Cherkalov Sor and 2 -Chyvyrkuy Bay; Baikal Basin (3 -River Selenga, 4 -Lake Torma, 5 -Lake Schuchje, 6 -Gusinoye, 7 -River Khilok); River Lena basin (8 -Lake Bol'shoe Eravnoe); Baikal-Lena basin (9 -Ivan-Arachley lake system including Lakes Shaksha, Arachley, Kergendu, Ivan and Tasey), closed lake-refuges (10 -Lake Kenon, 11 -Lake Arey). https://doi.org/10.1371/journal.pone.0194944.g001 comprehensive COI dataset for Anodonta anatina from Froufe et al. [25], with a total of 125 COI sequences. A haplotype network was constructed using TCS 1.21 [38] with a threshold of 95% and the output file was run through tcsBU [39]. Sequence divergences within the European haplogroup (uncorrected p-distance) were calculated using MEGA 7.0 [40]. Two distinct tools were used for this COI dataset to detect the number of Molecular Operational Taxonomic Units (MOTUs), and therefore the number of potential species. First, the Cluster Sequences tool implemented in BOLD 4 [41] was used and the generated BINs recognized as MOTUs. For the second, the 95% statistical parsimony connection limit using TCS 1.21 [38], was also applied.

General morphological features
Shell shape of collected specimens is highly variable, from ovate-rectangular and ovatequadrangular to ovate-elongated with a curved or straight noticeable wing, on the dorsal margin (Table 3; S1-S6 Figs). Shell convexity varied from flattened to strongly convex. Periostracum is usually yellow-brown or green-yellow-brown; the nacre is white, yellowish or light blue. The anterior muscle scar is well imprinted and the posterior adductor scar is weakly visible, pseudocardinal and lateral teeth are absent. The umbo is wide and often eroded and does not project above the dorsal shell margin, its position from the anterior shell margin varies from 0.18 to 0.36% of the total shell length. Umbo sculpture of all examined specimens is double-looped [42] (Panels 1-4 in Fig 3) or with almost straight bars (Panels 5 and 6 in Fig 3), sometimes with discrete lines (Panels 7 and 8 in Fig 3). Umbo sculpture is strongly pronounced in some animals but weakly marked in others, regardless of age.
Soft body anatomy (or internal anatomy) is not dependent of shell shape and is similar in all specimens from the Lake Baikal, Baikal drainage and outside (Panels 1-3 in Fig 4). The
(P.) from the studied region may also be identified as other TCS-CM species of Colletopterum s. str. from the European distribution (Table 3).

Morphometric analyses
Analyses of morphometric shell indexes B/H, H/L and l/L have shown significant variability and overlap in Colletopterum species from different localities of Lake Baikal, Transbaikalia and European Russia. This overlap can be substantial, even almost complete, in the TCS-CM species C. anatinum and C. piscinale ( Table 3). The lower values of B/H were characteristic for all young individuals from all species and for C. sorensianum (including C. subcirculare and C. baeri). These values increased in C. nilssonii (including C. milaschewichi and C. ostiarium) and C. piscinale (including C. covexum) ( Table 3; Panel 1 in Fig 5), with the higher values being obtained in C. ponderosum and C. rostratum (Table 3; Panel 1 in Fig 5). The values of H/L decrease in the reverse order (Panel 2 in Fig 5). Following the same order from C. sorensianum to C. rostratum, the position of the umbo (l/H) in the same species increases from C. sorensianum until C. anatinum and then decreased to C. rostratum (Panel 3 in Fig 5). The age of all examined specimens varied from < 1 year to 12 years, with shell length from 18 to 125 mm. The B/H showed a low variability range with an increasing trend with age (R 2 = 0.996) (Panel 4 in Fig 5). Individuals from distinct locations showed different growth rates, with the mean growth of individuals from all populations being shown in Panel 5 in Fig 5. Age is more closely connected with the B/H index values (R 2 = 0.996) than with shell length (R 2 = 0.928) with its higher accuracy reflecting its logarithmic dependence (R 2 = 0.964). In addition, B/H values increase with shell length in groups that have been identified as separate species (Fig 6). Shell shape, wing development and shell convexity change with shell length, corresponding to six different size-age group, in the analyzed TCS-CM species ( Table 3). All of the Comparatory species described from the study area correspond to at least another distinct TCS-CM species with a disjunct range (Table 3; S1-S6 Figs). While for C. piscinale and C. rostratum only a single additional species can be recovered (Table 3; S3 and S6 Figs), two species for C. sorensianum and C. nilssonii (Table 3; S1 and S2 Figs), three for C. ponderosum and even four additional species can be recovered in the case of C. anatinum (Table 3; S4 and  S5 Figs). Additionally, specimens identified as C. anatinum, C. piscinale, and C. ponderosum can all be recognized as C. convexum with a total overlap of the curvatures of the AMCVO ( Table 3; S3-S5 Figs). Furthermore, specimens described as C. anatinum can be assigned as C. ponderosum and vice-versa (Table 3; S4 and S5 Figs).

Molecular analyses
The aligned COI dataset presented a mean length of 578 bp with 21 polymorphic and 14 parsimony informative sets. No indels and no unexpected codons were observed in the corresponding amino-acid translation. TCS produced a single network shown in Fig 7. The 25 newly sequenced individuals from the study area, previously identified as six distinct TCS-CM species, resulted in five haplotypes (Dark Purple : Fig 7) all clustering within the previously recognized European A. anatina mitochondrial haplogroup (Purple: Fig 7) presenting a low intraspecific genetic diversity (p-distance = 0.6%). Eighty per cent of the individuals share a common haplotype with the remaining 20% representing 5 single private haplotypes. Moreover, the common haplotype is also shared with individuals from Poland and Sweden. From the six newly sequenced individuals from Ukraine, we retrieved four haplotypes also clustering within the European haplogroup (Fig 7). Both molecular species delineation methods applied in the COI dataset (Table 2) resulted in the identification of a single MOTU and therefore of a single species.

Discussion
The present study represents the first comprehensive investigation of the TCS-CM Colletopterum species (Anodontinae) from Lake Baikal and adjacent territory of Transbaikalia. The integrative results here presented clearly show that all collected specimens belong to the species Anodonta anatina.
Molecular results additionally show that all analyzed specimens cluster within the European Anodonta anatina previously described mtDNA clade [25], with most individuals sharing haplotypes with specimens from Europe (Fig 7). Furthermore, no genetic subdivision was detected within the specimens from the studied region, discarding further division in subgenera and species. Both molecular species delimitation methods indicate a single MOTU, corroborating the existence of a single species, i.e. A. anatina. These results are further confirmed by the morphological and morphometric analyses.
According to Bogatov et al. [16], Colletopterum was separated from Anodonta by changes in shell size, umbo sculpture and shell surface. In the present study, all specimens were undoubtedly identified as a single species. The shell differences that A. anatina presents compared to other anodontine species within Anodonta are very weak. These slight differences do not support the separation of A. anatina and other Anodonta species (i.e., Anodonta cygnea and Western North American Anodonta) into distinct genera and should correspond to infrageneric differences. In fact, since anodontine species are known to present a high shell plasticity, these differences were even overlooked by the most comprehensive classification system carried by Haas [11], combining A. anatina with A. cygnea forms under a single species. Additionally, A. anatina clusters with A. cygnea and Anodonta nuttalliana in a monophyletic clade in the most recent molecular phylogeny [23].
According to Bogatov et al. [16], Colletopterum was separated into two subgenera based on shell characteristics, i.e presence/absence of a clear wing, but the same author noted that in mature or strongly convex specimens, this wing may be reduced. The present study revealed a considerable variability on the wing expression (S1-S6 Figs), without any correspondence to the two putative Colletopterum subgenera. Again, the data here presented discards the existence of a subdivision within the putative Colletopterum species, i.e within Anodonta anatina.
The main characteristics used to delimit each of the 12 TCS-CM Colletopterum species are the arc of maximal convexity of the valve's outline (AMCVO) and the index of convexity (ratio of shell width to shell height) (Panels 2-4 in Fig 2) [16]. The present study shows that these features are not species specific. In fact, each TCS-CM species might be identified by the AMCVO as two-three, or even four different species (Panel 12-in S3 Fig) from both putative subgenera (S1-S6 Figs). No significant differences were found with regard to double-looped umbonal sculpture, soft part anatomy, and larval and adult shell morphology that supports the distinction of Colletopterum in two distinct subgenera (Panels 1-3 in Fig 4) [43].
It is well established that shell shape of unionoids is strongly influenced by both biotic and abiotic factors such as age, size, sex of the mussels, and temperature, hydrological, hydrochemical and trophic conditions [5], [6], [10]. Furthermore, many unionid and especially anodontine mussels, exhibit considerable intraspecific shell shape variation, which is often related with their occurrence in distinct habitats, shifts of metabolism at sexual maturity or even by changes in allometric growth and other physiological characteristics [10], [44], [45], [46], [47]. This influence in shell shape is also shown here for A. anatina, where B/H values increase with shell length. These values may be divided in six classes (Fig 6), each including 2-3 TCS/CM species that present an overlap of the B/H values. The same classes can also be recovered using the other two morphometric indexes H/L, and l/L values. However, these morphometric classes are composed of TCS-CM species from distant geographic ranges. Since no other independent morphological, anatomical or molecular character exhibit the same pattern, it clearly shows that these indexes are being influenced by environmental factors and should not be used to divide species. Furthermore, all collected young individuals were included in the C. sorensianum class group which indicates that age is clearly influencing not only the B/H but also the H/L, and l/L values (Figs 5 and 6). In fact, using all specimens across all TCS-CM species, the B/H index show very little variability with age, supporting the idea that shell shape is closely linked with growth (Fig 6). Furthermore, the sex of the animal might also influence shell shape in Anodonta [10], and consequently influencing the B/H, H/L, l/L morphometric indexes and AMCVO curves. According to Zieritz & Aldridge [10] A. anatina is dimorphic, with females being more inflated than males from the same size-age, due to the presence of swollen marsupia filled with glochidia during gravidity.