Differentiation of Bulinus senegalensis and Bulinus forskalii Snails in West Africa Using Morphometric Analysis

Purpose Accurate identification of medically important intermediate host and vector species is crucial for understanding disease transmission and control. Identifying Bulinus snails which act as intermediate host species for the transmission of schistosomiasis is typically undertaken using conchological and genital morphology as well as molecular methods. Methods Here, a landmark-based morphometric analysis of shell morphology was undertaken to determine its utility to distinguish the closely related and morphologically similar sister species Bulinus senegalensis and Bulinus forskalii. The method was developed to increase the accuracy of conchological morphology methods to identify Bulinus species in the field. Both species are found in West Africa, but only B. senegalensis is implicated in the transmission of urogenital schistosomiasis. Results We found when scaled down to the same length, 3-whorl and 4-whorl (juvenile) B. senegalensis shells had a longer spire, narrower body whorl and shorter aperture than B. forskalii. In contrast, 5-whorl (adult) B. senegalensis had a shorter spire, but still had a shorter aperture and narrower body whorl than B. forskalii. Canonical Variate Analysis (CVA) showed minimal overlap between B. senegalensis and B. forskalii for 3-whorl and 4-whorl shells, with a clear separation for 5-whorl shells. Overall, B. senegalensis had a consistently shorter aperture size and narrower body whorl than B. forskalii for all development stages. Spire length was variable depending on the stage of development, with 3-whorl and 4-whorl shells having the opposite trends of adult shells. Conclusions Our study demonstrates the applicability of landmark-based morphometrics in distinguishing the medically important, Bulinus senegalensis from its morphologically similar sister species, Bulinus forskalii. We recommend using measurements based on spire length, penultimate whorl length, body whorl width and aperture size to differentiate B. senegalensis and B. forskalii, when used with the appropriate information for each shell’s development stage. Supplementary Information The online version contains supplementary material available at 10.1007/s11686-024-00830-1.


Introduction
Bulinus (Gastropoda: Bulinidae) [1,2] is a genus of tropical freshwater snails with a sinistral shell, found throughout Africa and in neighbouring tropical islands, South West Asia, Arabia and the Mediterranean region [3].The genus is of significant medical (and veterinary) importance as its species act as intermediate hosts for the causative agents of urogenital schistosomiasis (Schistosoma haematobium) and intestinal schistosomiasis (S. guineensis and host for the human parasites Schistosoma intercalatum and S. guineensis [5][6][7].They are less prevalent than S. haematobium and only occur in specific areas of the Lower Guinea region and Democratic Republic of the Congo [5,7]. In West Africa, B. senegalensis and B. forskalii often coinhabit the same permanent (irrigation canals, rice fields and swamps) and seasonal (laterite pools) habitats [12].Given the shell morphological similarity of these two species (Fig. 1), it is often exceedingly difficult to determine which species of Bulinus is present in a locality and therefore what risk a Bulinus population may pose to a local community.The ability to identify medically and veterinary relevant species is important.Bulinid snail species are commonly categorised and identified using a combination of genital, radula and shell morphology [13], with taxonomic guides traditionally using morphological characters to identify species [1,[14][15][16][17][18][19].Genital morphology is the most dependable as complementary copulatory organs are essential for intraspecies mating [20].However, identifying snails using genital morphology requires time and expertise, as the genitals need to be extracted from the snail, prepared, and mounted under a microscope to be observed and measured.Molecular identification methods (e.g., COI barcoding) provide a reliable alternative to traditional morphological identification methods but are time consuming, expensive and require training and facilities.A good identification method is one with high reliability (like genital morphology and COI barcoding), but with less requirement for training, equipment, and time.
Shell morphological identification by itself does not meet all of these criteria as shell morphology is a variable characteristic in snails and can be affected by environmental factors such as water flowrate, temperature, predation and parasitism [21][22][23][24][25][26][27][28].As a result, taxonomists using only shell morphology typically overestimate the number of species within a genus due to the variety of shell morphotypes exhibited in one species, with these taxonomical mistakes only being clarified using more advanced techniques such as genital morphology or COI barcoding.However, the simplicity of shell morphological identification, its direct applicability and low cost make it a desirable option for the identification of snails.For example, a simplistic morphological characteristic for the differentiation of B. senegalensis and B. forskalii shells is the presence of a shoulder angle on the early whorls of B. forskalii shells; shoulder angles are absent in B. senegalensis [3,19].However, due to shell plasticity, the shoulder angles are not always present in B. forskalii shells or can be obscured by detritus.Moreover, as Bulinus snails develop from juveniles to adults their shells grow, the number of whorls increase from three to five and the shoulders angles become more prominent [29].Therefore, the stages of development (or number of whorls) can affect how similar B. senegalensis and B. forskalii look, with three whorl juveniles having underdeveloped characteristics and being harder to distinguish from each other.One way to address the issues of using morphological characteristics for Bulinus identification is to employ geometric, landmarkbased morphometric techniques [30,31].The addition of geometric, landmark-based morphometrics could improve how reliably and accurately one can distinguish snail species that share similar shell morphologies.
Here we investigate the use of landmark based morphometric methods for the differentiation of Bulinus senegalensis and B. forskalii using B. truncatus as a comparative outgroup.Specimens were examined from multiple locations across The Gambia, a country endemic with both B. senegalensis and B. forskalii, with additional museum specimens also examined to add variation from other African countries.

Sample Collection, Preparation, and Photography
Bulinus senegalensis, B. forskalii and B. truncatus specimens were collected from permanent and seasonal sites across The Gambia over the course of two years (2017-2019) [32].In total, 15 sites, each with only a single Bulinus species present, were selected for this study (Table 1).Additional B. senegalensis and B. forskalii samples were provided by the London Natural History Museum (NHM), with dry shell collections provided by Jonathan Ablett and wet collections provided by Dr Aidan Emery.These NHM Bulinus collections were from multiple African countries, with B. senegalensis available from three countries and B. forskalii available from 16 countries (Table 2).Bulinus samples from The Gambia (Table 1) were identified using  [32].Bulinus samples from the NHM (Table 2) were identified using both genital and shell morphology upon collection by their respective collectors (D. S. Brown; G. Mandahl-Barth; O. F. Müller; C. A. Wright & S. R. Smithers).The Gambian samples were stored in absolute ethanol, while the NHM samples were kept either as dry collections or preserved in IMS (wet collection).The NHM samples were not suitable for molecular analysis since only shells were available for specimens in the dry collection and specimens in the wet collection were preserved in IMS and not amplifiable in PCR.
Samples with no (or minimal) damage to their shells were selected for photography.This limited the number of viable shells collected from The Gambia due to their fragility as a result of ethanol preservation.Photographs were taken using a dissection microscope with a 64MegaPixel mobile phone camera attached.All shells were positioned  [27].This coordinate data was stored in a TPS file, and each sample was scaled and had a unique specimen ID, which displayed species (Bsen, Bfor or Btru), number of whorls (3W, 4W or 5W) and location (e.g., ID = Bsen-3W-Changai-1).The TPS file was then imported into the MorphoJ v1.07 program [35].The data was treated as symmetrical due to the bilateral symmetry for some of the landmarks (e.g., LM4&5, LM7&8, LM9&10, LM11&12 and LM13&14; Fig. 2).Each specimen was grouped by species (B.senegalensis, B. forskalii or B. truncatus), with the groups being defined by genetic (Gambian dataset) or genital morphology (NHM dataset) identification methods.The specimens were then partitioned based on the number of whorls they had (3-whorl, 4-whorl and 5-whorl).
Each of the whorl datasets had a Procrustes' fit analysis performed and shell shape was assessed using a Canonical Variate Analysis (CVA) across all landmarks using 10,000 permutations.
A Canonical Variate Analysis (also known as Canonical Correlation Analysis or Linear Discrimination Analysis) is a statistical method used to examine the relationship between two sets of variables.In our scenario, the landmark coordinate data is the independent variable and the species groups are the dependent variables.Our CVA is multivariate analysis that extracts crucial information (named canonical variables) from complicated datasets and is able to uncover and photographed the same way, with a 1 mm, 5 and 10 mm scalebar present; shell length (L1-L3) and shell width (L7-L8) were measured digitally using the software Digimizer v5.7.2 [33].

Morphometric and Data Analysis
To minimise errors associated with shell plasticity, we took several precautionary measures in our morphometric analysis: (i) using a comparative outgroup; (ii) incorporating multiple individuals from diverse populations to average out the plastic variation within the dataset; (iii) separating adult and juveniles specimens by whorl number to reduce the morphological disparities between shells at different stages of development; (iv) using a Procrustes' fit analysis to eliminate undesirable effects of translation, rotation, and scaling from our landmark placement; and (v) utilising outlier detection to exclude individuals with significant morphological deviations from the final analysis.Shell photographs were imported into the tpsDig2 v2.31 program [34] and each image was digitised using 16 landmarks comprised of ten fixed landmarks (anatomically meaningful) and six semi-landmarks (non-anatomically meaningful) to collect two-dimensional coordinate data (Fig. 2).The placement of our 16 landmarks was inspired by the Bulinus shell measurements used by Stothard et al. (1997) [17] and the landmarks  1 and 2).
When the samples were scaled down to the same shell length and the distance between specific landmarks were measured, we found clear differences in shell morphology.The 3-whorl shells had eight significant differences between landmark measurements.Bulinus senegalensis had a longer distance from L1 to L2, L2 to L11 and L2 to L12 of the spire; a longer distance from L14 to L16 of the penultimate whorl; a narrower distance from L5 to L6 and L6 to L8 of the body whorl; a shorter distance from L4 to L6 and L6 to L7 of the aperture than B. forskalii (Fig. 4a).These were the only significant measurements found that can distinguish 3-whorl B. senegalensis from B. forskalii.The 4-whorl shells had five significant differences between landmark measurements; B. senegalensis had a longer distance from L1 to L2, L2 to L11 and L2 to L12 of the spire; a longer distance between L13 to L15 of the penultimate whorl and a shorter distance from L3 to L10 of the aperture than B. forskalii (Fig. 4b).Finally, the 5-whorl shells had ten significant differences between landmark measurements, though some of the trends were the opposite to 3-whorl and 4-whorl shells.Adult, 5-whorl B. senegalensis had a shorter distance from L1-L2, L2-L11 and L2-L12 of the spire; a shorter distance from L2 to L13 and L2 to L14 of the penultimate whorl than B. forskalii.However, B. senegalensis still had a narrower distance from L2 to L16 of the body whorl and a shorter distance from L3 to L9, L3 to L10, L4 to L6 and L6 to L7 of the aperture than B. forskalii (Fig. 4c).

Discussion
Shell morphology is a useful parameter for the identification of snail species due to its simplicity and low cost.However, if morphological identification is to be used in instances where morphologically similar looking species co-inhabit patterns not immediately apparent in the raw data.These canonical variables are linear combinations of the original variables and are selected based on their efficacy in explaining the variation between the two datasets.The first canonical variable (CV1) explains the most variation, while the following canonical variable (CV2) explains the second most.A canonical variate analysis was preferred over alternative multivariate statistical methods, such as a Principal Component Analysis (PCA), due to its optimisation for classifying and discriminating groups within large datasets.
In addition to the morphometric analysis, the mean distances between informative landmarks were taken for both B. senegalensis and B. forskalii shells using Digimizer v5.7.2 [33].However, on average the B. senegalensis shells were larger than the B. forskalii shells.Therefore, the shell length for both species was first scaled down to a standardised length depending on development stage (3-whorl = 2 mm; 4-whorl = 4 mm; 5-whorl = 6 mm).The informative landmarks selected were informed by the CVA analysis (Supplementary Fig. 1).A Mann-Whitney U test was chosen as the distance data was interval and had a non-normal distribution.The Mann-Whitney U test was performed in SPSS v26 (IBM, Armonk, USA) [36] and was used to see whether there was a significant difference in the landmark distances between the two species.
There were no noticeable differences between the mean length and widths of B. senegalensis and B. forskalii shells (Fig. 3a/3b), though the Gambian samples showed on Chikungunya & Lymphatic filariasis) [38], Triatominae bugs (Trypanosomiasis) [39], and even other medically important gastropods, Biomphalaria (intestinal schistosomiasis) [28].Bulinus shell morphology has been previously investigated using shell measurements and morphometrics the same environment (as in the case of B. senegalensis and B. forskalii), more accurate techniques are required [30,31].Morphometrics has been previously used to help distinguish other medically important invertebrates such as Calliphoridae (Myiasis) [37], Culicidae (Malaria, Dengue, the same length (2 mm), the distance between landmarks showed clear differences.Juvenile 3-whorl B. senegalensis had a longer distance from L1 to L2, L2 to L11 and L2 to L12 of the spire; a longer distance from L14 to L16 of the penultimate whorl; a narrower distance from L5 to L6 and L6 to L8 of the body whorl; a shorter distance from L4 to L6 and L6 to L7 of the aperture than B. forskalii.Through the use of photomicrography, these significant measurements can be used to distinguish 3-whorl B. senegalensis from B. forskalii (Fig. 6).
The next development stage, 4-whorl shells, had fewer significant measurements than 3-whorl shells.When scaled down to the same length (4 mm), 4-whorl B. senegalensis had a longer distance from L1 to L2, L2 to L11 and L2 to L12 of the spire; a longer distance from L13 to L15 of the penultimate whorl and a shorter distance from L3 to L10 of the aperture than B. forskalii.The observation of fewer [14,15,[16][17][18]27].However, this is the first study that investigates the morphological differences between species, specifically the B. forskalii group.In this study, we used landmark based morphometrics, to distinguish two sister species of Bulinus (B.senegalensis and B. forskalii) at different developmental stages.
Canonical variate analysis showed varying levels of separation.3-whorl juvenile shells showed some separation with minimal overlap.This separation becomes clearer in 4-whorl and 5-whorl shells which show complete separation between adult B. senegalensis and B. forskalii.This emphasizes the importance of stage of development on accuracy when trying to distinguish between B. senegalensis from B. forskalii.
Juvenile 3-whorl Bulinus senegalensis and B. forskalii were more similar in morphology than 4-whorl and 5-whorl shells.However, when the shells were scaled down to  6).These measurements are sufficient to distinguish B. senegalensis and B. forskalii when used with the appropriate information for each development stage.More information about key landmark measurements expressed as ratios using shell length (L1-L3) for B. senegalensis and B. forskalii shells can be found in supplementary Table 3.

Conclusion
We show that landmark based, morphometric analysis provides an effective and viable method for the differentiation of Bulinus senegalensis and B. forskalii.The accuracy of identifying a Bulinus species is affected by the stage of development (3-whorl, 4-whorl and 5-whorl).When scaled down to the same length, B. senegalensis shells consistently had a significantly shorter aperture size and body whorl width than B. forskalii.While spire length and penultimate whorl length were dependent on the development stage, with 3-whorl and 4-whorl B. senegalensis having a longer significant measurements for 4-whorl shells compared to the 3-whorl shells indicates that as B. senegalensis and B. forskalii progress from 3-whorl to 4-whorl, the shells become more similar.Developmental stage can affect how similar B. senegalensis and B. forskalii look [29].However, spire length and aperture size can still be used to distinguish B. senegalensis from B. forskalii at this stage of development.
Adult 5-whorl shells had the most significant measurements of all stages.However, some of the trends of the previous two stages were the opposite.When scaled down to the same length (6 mm), 5-whorl B. senegalensis had a shorter distance from L1 to L2, L2 to L11 and L2 to L12 of the spire; a shorter distance from L2 to L13 and L2 to L14 of the penultimate whorl than B. forskalii.However, B. senegalensis still had a shorter distance from L2 to L16 of the body whorl and a shorter distance from L3 to L9, L3 to L10, L4 to L6 and L6 to L7 of the aperture than B. forskalii.
To summarise, B. senegalensis consistently had a significantly narrower body whorl and shorter aperture size than B. forskalii, regardless of whorl number.Spire length and penultimate whorl length can also be used to differentiate B. senegalensis and B. forskalii.However, our morphological technique has several limitations such as the need of a microscope due to the small size of the shells; the need to identify the developmental stage of the snail before identifying it; and our sampling only covers a small number of populations from specific African countries.Overall, landmark based morphometrics is helpful in improving the

Fig. 1
Fig. 1 Shell morphologies of adult Bulinus senegalensis (left) and B. forskalii (right).Arrows indicate the shoulder angles found only in adult B. forskalii shells

Fig. 2
Fig. 2 Example of the fixed landmarks (black) and semi-landmarks (grey) placed on Bulinus forskalii and B. truncatus shells (not to scale).Additionally, adult Bulinus truncatus shells were also included as a comparative outgroup.L1-L3 were used to measure shell length (or height) and L7-L8 were used to measure shell width.Landmark placements, L1: tip of the Apex; L2: middle of the Body whorl and Teleconch suture (aligned with L1); L3: bottom of Basal lip (aligned vertically with L1 and L2); L4: meeting point between the Aperture and Body whorl; L5: aligned horizontally on the right of L4; L6: