DNA barcoding and morphological characters were used to identify adult snails belonging to the genus Biomphalaria from 17 municipalities in the state of São Paulo, Brazil. The DNA barcode analysis also included twenty-nine sequences retrieved from GenBank. The final data set of 104 sequences of the mitochondrial cytochrome oxidase I (COI) gene was analyzed for K2P intraspecific and interspecific divergences, through tree-reconstruction methods (Neighbor-Joining, Maximum Likelihood and Bayesian inference), and by applying different models (ABGD, bPTP, GMYC) to partition the sequences according to the pattern of genetic variation. Twenty-seven morphological parameters of internal organs were used to identify specimens. The molecular taxonomy of Biomphalaria agreed with the morphological identification of specimens from the same collection locality. DNA barcoding may therefore be a useful supporting tool for identifying Biomphalaria snails in areas at risk for schistosomiasis.

Biomphalaria , COI, DNA barcoding, morphological taxonomy, schistosomiasis, species identification

Brazil contains one of the richest faunas of freshwater snails of the genus Biomphalaria (Agostinho et al. 2005, Scholte et al. 2012). The state of São Paulo, in southeast Brazil, is of enormous epidemiological importance, as all the three Neotropical intermediate hosts of Schistosoma mansoni (Sambon, 1907), Biomphalaria glabrata (Say, 1818), B. straminea (Dunker, 1848), B. tenagophila (d’Orbigny, 1835), are distributed in streams, ponds, dams and reservoirs in this municipality. Biomphalaria occidentalis (Paraense, 1981), B. peregrina (d’Orbigny, 1835), B. intermedia (Paraense & Deslandes, 1962), B. oligoza (Paraense, 1975) and B. schrammi (Crosse, 1864) are also distributed in São Paulo state (Vaz 1989; Teles 2005; Ohlweiler et al. 2010).

Identification of Biomphalaria specimens to the species level and analysis of infection by S. mansoni are key elements of surveillance strategies for schistosomiasis control and elimination (PAHO 1968, WHO 2013). Shell morphology is of limited use for identifying different species of snails in this genus (Paraense 1966; Jarne and Théron 2001), and therefore the anatomical characters described by Paraense (1961, 1975, 2001) are used instead. However, identification of Biomphalaria solely based on morphological characters is constrained by phenotypic plasticity, the limited descriptions of cryptic species, and the difficulty in applying species-diagnostic characters to juvenile specimens (Carvalho et al. 2008; Teodoro et al. 2010). The issue of how useful molecular tools may be in the identification of Biomphalaria snails has become particularly important in recent years as there is consensus among malacologists that morphological identification using internal anatomical parameters is susceptible to error, especially when the snails being analyzed belong to complexes of morphologically similar species (Paraense 1972, 1974, 1988; Spatz et al. 1999; Vidigal et al. 2000). To overcome these limitations and difficulties associated with traditional taxonomy, various methodologies based on molecular markers have been developed.

PCR-RFLP analysis of mitochondrial and nuclear genes (Spatz et al. 1999; Vidigal et al. 1998, 2000; Caldeira et al. 2000, 2009), fingerprinting techniques using nonspecific primers (Abdel-Hamid et al. 1999; Al-Quraishy et al. 2014) and sequence analysis of COI and r16RNA genes and ITS-1 and ITS-2 sequences (Woolhouse and Chandiwana 1989; Langand et al. 1998; Vidigal et al. 2000; Campbell et al. 2000; DeJong et al. 2003; Wethington et al. 2007; Tuan and Santos 2007; Tuan et al. 2012) have all produced results that allowed significant genetic differences in species and populations to be identified.

When used in conjunction with bioinformatics tools and sequence databases, DNA barcoding routinely facilitates the identification of biological species (Ratnasingham and Hebert 2007; Casiraghi et al. 2010). This technique is based on the polymorphism of a short region (approximately 600 bp long) of the mitochondrial cytochrome c oxidase 1 (COI) gene (Hebert et al. 2003). DNA barcode includes a series of strategies for delimiting species into molecular operational taxonomic units (MOTUs) using a combination of laboratory and bioinformatics methods (Fontaneto et al. 2013). The most important strategies for identifying MOTUs include analysis of intraspecific and interspecific genetic distances, and analyses based on population and phylogenetic models. These approaches include (ABGD) (Puillandre et al. 2012) and the barcode index number (BIN) system (Ratnasingham and Hebert 2013), which use algorithms based on the partition of molecular data according distance methods, and the generalized mixed Yule coalescent (GMYC) method (Fujisawa and Barraclough 2013) and Bayesian Poisson Tree Processes (bPTP) method (Zhang et al. 2013).

DNA barcoding has been used to augment morphological identification of Bulinus in Africa (Kane at al. 2008; Stothard et al. 2013; Standley et al. 2014), and yielded better results than identifications based on shell characters. Although there are over 500 COI sequences in GenBank from snails of the genus Biomphalaria found in African and Neotropical regions, most DNA barcoding studies use African species. There is therefore a dearth of knowledge about the effectiveness of DNA barcoding in taxonomic identification of Neotropical species of Biomphalaria (Standley et al. 2011; Tuan et al. 2012).

Here, we investigate the utility of analysis of distributions of intraspecific and interspecific COI divergences based on genetic distances, tree reconstruction methods based on Bayesian inference, Maximum Likelihood (ML), and K2P-Neighbor-Joining (NJ) grouping of sequences, and the ABGD, GMYC and bPTP methods for delimitation of Biomphalaria species in conjunction with schistosomiasis field surveys.

This study sought to determine the utility of DNA barcoding in delimiting species in freshwater snails of the genus Biomphalaria. The Bayesian, ML and NJ analyses (Fig. 3, Suppl. material 1) yielded trees with well-supported internal branches (≥90), resolving six out of the seven taxa as monophyletic groups.

The assessment of the potential of DNA barcode for species differentiation in Biomphalaria essentially revolves around the comparison of results of the morphological and molecular analysis of closely similar or taxonomically ambiguous species. In the case of the three taxa in the B. tenagophila complex, one character that is normally effective for specific identification is the vaginal pouch, which is present in B. tenagophila and B. t. guaibensis but not in B. occidentalis. (The anatomical features of these three taxa were illustrated by Tuan et al. 2012). Although we did not observe this in our material, in some specimens of B. occidentalis there is a slight projection of the ventral wall of the vagina (Paraense 1981), which raises questions regarding the distinctness of this taxon.

The intraspecific genetic distance within B. tenagophila showed values with a range from 0 to 3% (Table 4, Figs 2, 3). A high level of genetic divergence within this species was obtained for sequences associated with specimens collected in Juquiá (650,651,653), Itariri (535), Embu das Artes (524,535) and São Paulo (549, 551). Due to these values we could not assign a clear barcode gap between B. tenagophila and B. occidentalis and B. t. guaibensis (Fig. 2 b, d, f). However, the Bayesian tree inferred from COI data (Fig. 3), as well as the ABGD and both bPTP and GMYC analyses recovered these close related taxa as distinct groups. We suggest that in geographical areas where B. tenagophila species complex have the same geographical distribution.

The application of the 3-4% cutoff value for maximum intraspecific divergence may be appropriate for our dataset as 36% of the intraspecific comparisons reached this value (Table 4). The highest values for intraspecific divergence (>3%) do not appear to be a consequence of geographic distance given that the greatest divergence in B. tenagophila was between closely proximal localities in São Paulo state (Fig. 3).

Biomphalaria glabrata and B. tenagophila, are differentiated by the renal ridge, which is present in the former and absent in the latter. Paraense and Deslandes (1959) described a false ridge that runs obliquely to the renal tube and is attached to the pneumostome, in specimens of B. tenagophila from Macaé, RJ. The presence of this false ridge in B. tenagophila may lead to incorrectly identify this species, particularly in juvenile specimens or specimens that have not been properly fixed. Five specimens of B. tenagophila in our study (three from São Lourenço da Serra and two from São Paulo) had a membrane on the renal tubes similar to that described by Paraense and Deslandes. The genetic distance of 9% between B. glabrata and B. tenagophila observed with both genetic distance and tree-based approaches show that DNA barcoding is an important tool for identifying these closely similar taxa.

The ABGD analysis partitioned B. glabrata into five distinct groups, while the GMYC analysis yielded a more cohesive group. Despite the pronounced COI divergence within B. glabrata, in all the specimens analyzed here the renal ridge has been considered a robust and consistent taxonomical character, suggesting that morphology is more effective than DNA barcode in this case. However, bBTP analysis and phylogenetic reconstruction supported B. glabrata as single and well supported MOTU, a result congruent with the morphological identification.

Another group of morphologically similar and frequently misidentified congeners includes B. intermedia and B. straminea; the latter a natural intermediate host of S. mansoni. Of the seventeen diagnostic characters common to B. straminea and B. intermedia, the degree of corrugation in the dorsal wall of the vagina has been used to these taxa as a species complex (Paraense 1975) . The vaginal corrugation, which is markedly wavy in B. straminea appears as swollen in B. intermedia. The large genetic divergence between B. straminea and B. intermedia, which was 9% greater than the intraspecific values in both species, indicates that these two species can be identified by DNA barcode. Note, however, that our study only included two of the three species in the B. straminea complex, as B. kuhniana does not occur in São Paulo state (Paraense 1988, Teodoro et al. 2010). In addition, we were unable to collect many specimens of B. intermedia owing to its rareness in São Paulo state.

Our findings show Biomphalaria species delimitation by phylogenetic approaches and bPTP yielded the same groups identified by traditional taxonomy. The use of DNA barcode to identify species in conjunction with Biomphalaria surveys requires the application of both evolutionary and bioinformatics criteria, making it a time-consuming approach that is dependent on specialist knowledge. Morphological identification also requires specialist knowledge. However, as shown in this study DNA barcoding can identify subtle (genetic) differences between intraspecific populations that are not detectable by traditional morphological study.

Furthermore, morphological identification of Biomphalaria species depends on subjective interpretation of anatomical variations, as these are measured in terms of relative rather than absolute sizes. We therefore agree with Hebert and Gregory (2005, p. 853), who stated that by reversing the logic of standard taxonomic approaches that “operate in an a priori fashion—seeking…morphological discontinuities”, DNA barcoding may, as “a posteriori approach”, direct the study of morphological variation in genetically divergent groups of Biomphalaria.

This work was supported by a grant from the Superintendency for the Control of Endemic Diseases (SUCEN) (ref. no. 0375/2012) to R. Tuan and M. C. A. Guimarães. The acquisition of equipment for the molecular analysis was funded by FAPESP grants to R. Tuan.