Predicting current and future suitability for intermediate snail hosts of urogenital and intestinal schistosomiasis in a floodplain of Malawi

Predicting current and future suitability for intermediate snail hosts of urogenital and intestinal schistosomiasis in a floodplain of Malawi | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Predicting current and future suitability for intermediate snail hosts of urogenital and intestinal schistosomiasis in a floodplain of Malawi Clinton Nkolokosa, James Chirombo, Christopher M. Jones, Eggrey Aisha Kambewa, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5908499/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 29 Aug, 2025 Read the published version in Parasites & Vectors → Version 1 posted 7 You are reading this latest preprint version Abstract This paper presents the first species distribution models (SDMs) for intermediate snail hosts for urogenital and intestinal schistosomiasis in the Lower Shire Valley (LSV), Malawi. The SDMs are specific to the Bulinus africanus group and Biomphalaria pfeifferi . The former transmits urogenital schistosomiasis ( Schistosoma haematobium ), and the latter transmits intestinal schistosomiasis ( Schistosoma mansoni ), both of which affect nearly 240 million people globally. This study addresses the following questions: 1. Where are the most suitable habitats for intermediate host snails in the LSV? 2. Which environmental factors influence the geographical distribution of such snails in the LSV? 3. How will climate change shape future schistosomiasis transmission risk? Consistent with expectations, the SDMs reveal the following: 1) currently, Bu. africanus not only has a wide distribution across central Chikwawa and eastern Nsanje but is also concentrated in floodplains, and the LSV has few habitats that can support Bi. pfeifferi , 2) vegetation cover is the most important predictor of Bu. africanus distribution, whereas precipitation variables are most important for Bi. pfeifferi in the LSV, and 3) future projections indicate a moderate increase (4.4%) and east-ward shift in Bi. pfeifferi distribution, with patchy spatial coverage, and a significant expansion (46%) of suitable habitats for Bu. africanus in LSV. Understanding the spatial and temporal distributions of these snails is important for controlling and eliminating schistosomiasis. Habitat suitability species distribution modelling schistosomiasis Lower Shire Valley ensemble machine learning climate change Figures Figure 1 Figure 2 Figure 3 Figure 4 Highlights • We present a species distribution model (SDM) for and snails to predict potential habitats for the transmission of urogenital and intestinal schistosomiasis, respectively. • The SDM comprises environmental predictors and an ensemble model of random forest, support vector machine and multilayer perceptron. • The current suitability map reveals high suitability for in large wetland areas and a low suitability for in small cropland areas within the Lower Shire Valley (LSV), each showing distinct habitat ranges. • The future suitability map forecasts a significant increase (46%) in suitable habitats for and a modest gain (4.4%) in suitable habitats for in LSV. • Areas with a high probability of snail presence could be the first priority for both human surveillance and snail control. Background Relating the geographical distribution of freshwater snails (vectors) to local environmental attributes offers value for understanding the epidemiological landscape of schistosomiasis transmission in a changing environment. Schistosomiasis—both urogenital and intestinal—causes significant human suffering. Schistosomiasis, which affects approximately 240 million people globally, is a neglected tropical disease (NTD) caused by parasitic flatworms and is associated with human water contact behaviour ( 1 ). The disease is endemic in Malawi. Approximately 40–50% of the Malawi population are at risk of being infected, with school-aged children being the highly infected and the most infected group ( 1 – 3 ). If schistosomiasis is not treated, it can result in severe health complications, including infertility, anaemia, malnutrition, abdominal pain, enlarged or damaged liver, haematuria, and blood in the stool ( 1 , 4 ). In rare cases, convulsions, paralysis, or spinal cord inflammation may occur when eggs released by a pair of adult worms are found in the brain or spinal cord ( 1 , 3 ). Stigma, social exclusion and poor educational outcomes aggravate the suffering from schistosomiasis ( 1 ). The parasitic worms (trematodes) that cause the disease are of the genus Schistosoma : Schistosoma mansoni and Schistosoma haematobium ( 5 , 6 ). The latter is responsible for intestinal schistosomiasis, and the former is responsible for urogenital schistosomiasis. The transmission chain depends on compatible snails. In Malawi, S. mansoni and S. haematobium parasites are transmitted by freshwater-intermediate host snails of the genera Biomphalaria (Planorbidae) and Bulinus (Bulinadae), respectively ( 5 , 7 ). Larval schistosomes (cercariae) from infected freshwater snails penetrate human skin, causing (re)infection. This occurs in water bodies, for example, during routine agricultural, domestic, occupational and recreational activities, such as irrigating, washing, wading, bathing or swimming. Therefore, human–surface water interactions and the presence of intermediate host snails predicate the spatial distribution of schistosomiasis prevalence. This is particularized in Malawi, where the consequence of proximity to surface water (lake, river, wetland, canal, dam, pond) is generally a high risk and prevalence of schistosomiasis ( 2 , 5 , 7 – 11 ). Among the different approaches used to control the spread of schistosomiasis, snail control is essential for interrupting the parasite life cycle ( 6 ). Eliminating snail hosts is considered an important, effective and convenient strategy for schistosomiasis prevention in endemic areas ( 12 , 13 ). Accordingly, the WHO recommends prevention and treatment: molluscicidal (chemical) control, ecological control (sanitation and environmental modification) and health education for behavioural change ( 14 ). Identifying the habitats where intermediate host snails occur could actively inform NTD control programs to address schistosomiasis, which is evidently an important public health challenge in Lower Shire Valley (LSV), southern Malawi ( 2 , 15 ). The Malawi Neglected Tropical Diseases Master Plan (2023–2030) recognizes that a changing climate is a threat and challenge to schistosomiasis control. However, in the absence of evidence, endemicity dynamics and transmission risk resulting from anthropogenic, climatic and ecological changes remain unclear. Revealing the current and future schistosomiasis risk is vital for opportune control efforts and resolving the elusive climate change challenges, holding back national and global efforts to eliminate NTDs in resource-constrained settings, such as Malawi. To date, several studies across Africa have investigated the abundance, distribution and spread of freshwater snails at broad scales via a species distribution model (SDM). The task of an SDM is to determine the probability of a species occurring in a particular habitat as a function of a set of environmental conditions. For example, in Kenya, potential habitats of freshwater snails were mapped via an SDM based on maximum entropy (Maxent) ( 16 ). This study incorporated a set of environmental variables, including land surface temperature, soil pH, and vegetation greenness, to predict environmental suitability ( 16 ). In a related study, Maxent was also applied to forecast the distribution of suitable habitats for Bu. globosus and Bi. pfeifferi in South Africa ( 17 ). In Malawi, Reed et al. ( 10 ) provided a first step toward understanding the spatial risk of intestinal and urogenital schistosomiasis within under sampled areas via 2D mean Gaussian process prediction. In the present study, a machine learning classifier is utilized in potential distribution modelling over a mosaic habitat. Thus, the importance of this study is by no means restricted to intestinal and urogenital schistosomiasis in LSV but also potentially to other NTDs and forms of schistosomiasis of veterinary importance in other geographical settings. Therefore, with respect to LSV, the aim of this study is twofold: 1) developing and parameterizing an SDM for Bi. pfeifferi and Bu. africanus snails in LSV, and 2) using the parameterized model to now- and forecast the distribution of Bi. pfeifferi and Bu. africanus in LSV under current and future climate conditions. Methods Study area and species occurrence data Data of Bi. pfeifferi and Bu. africanus occurrence (absence/presence) samples were collected in the floodplains of the LSV between May and June 2023. The LSV study area (latitude: 14°25′ S and 16°55′ S and longitude: 35°16′ E and 35°12′ E) is the southern subregion of Malawi, which is situated in the lower Shire floodplain and consists of two districts: Chikwawa and Nsanje. The study area, covering approximately 6833 km 2 , is characterized by a wide range of topography (medium-altitude plain 750–1300 m and floodplain 35–105 m) and ecological zones (freshwater habitats, tropical grasslands, savannas and shrublands, montane forest-grassland mosaics, and flooded grasslands) ( 18 , 19 ). The climate is subtropical: a warm-wet season from November to April and a hot-dry season from May to October ( 18 ). According to the National Statistics Office, Chikwawa and Nsanje had estimated population densities of 128 people/km 2 and 168 people/km 2 in 2023, respectively ( 20 ). Across the valley, schistosomiasis is endemic and focal, with district-level prevalence considered low for S. mansoni and moderate for S. haematobium ( 21 ). For example, in 2017, the schistosomiasis prevalence rates in Chikwawa and Nsanje were 15.8% (95% CI: 10.9%, 22.4%) and 25.4% (95% CI: 15.3%, 38.9%), respectively ( 21 ). This study surveyed a total of 70 sampling sites across the LSV: Chikwawa ( n = 35) and Nsanje ( n = 35) (see Fig. 1 ). The inclusion criteria were as follows: 1) confirmed cases of intestinal or urogenital schistosomiasis in the area and 2) the presence of surface water (wetland, river, dams, ponds, canals). Here, a malacologist and three trained field collectors who adhered to the WHO sampling design protocol ( 14 ) collected the survey data and used a field guide to African freshwater snails to identify the snail intermediate hosts. Standardized sampling effort was achieved by setting the sampling time to 15 minutes per site. The collected dataset includes information on the geographic coordinates of the sampled sites, site number, data and time of collection, all freshwater snails encountered, habitat type, human- and animal-water contact, aerial photograph of the site, and geographical elevation. A hand-held global positioning system (GPS) device (Garmin Montana 700 GPS, US) was used to locate and map the sites. Using stratified random sampling, georeferenced occurrence records of Bi. pfeifferi and Bu. africanus groups were generated along the Shire River, Elephant Marsh and irrigation canals by foot, car and boat (Fig. 1 ). From the malacological survey, a total of 597 Bi. pfeifferi and 1994 Bu. africanus samples were collected from four and 26 sites, respectively. Typical of studies determining the distribution of freshwater snails, field sampling was designed on the basis of the knowledge that such snails occur in lentic and lotic ecosystems: standing water and running water habitats, respectively ( 10 , 22 – 24 ). Therefore, the survey targeted representative aquatic habitats, such as lakes and marshlands, ponds and pools, rivers and canals, especially where human–water contact occurs (fishing, gardening, bathing, swimming, washing, etc.). Environmental predictors The present study uses abiotic (climatic, topographic, soil and proximity) and biotic (vegetation cover) factors to determine the influence of environmental factors and gradients on Bu. africanus and Bi. pfeifferi distributions (Additional file: Table 1 ). More specifically, temperature, precipitation, geographical elevation, terrain slope, soil type, distance to a water body and normalized difference vegetation index (NDVI) variables were included in the snail distribution modelling. Previous studies have shown that abiotic factors, such as rainfall ( 25 , 26 ), air temperature ( 25 , 27 ), and altitude ( 28 , 29 ), influence the distribution and habitat preference of freshwater snails of medical importance. For example, it has been demonstrated that temperature affects the mortality, fecundity, and growth rate of Bu. africanus and Bi. pfeifferi snails ( 26 , 30 ), and precipitation and elevation gradients are negatively correlated with freshwater snail distribution and abundance ( 29 , 31 – 33 ). This study uses slope as a direct factor regulating water velocity, which is a key determinant of intermediate host snail habitat preference. The velocity of water tends to have a positive relationship with the slope gradient ( 34 , 35 ). The importance of water velocity on freshwater snails has been previously established. Generally, Bulinus snails and Biomphalaria snails prefer stagnant or slow flow (velocity ≤ 0.3 m/s) ( 36 , 37 ). The argument is that rapidly flowing water impedes the establishment of reproductive colonies and displaces the snail population ( 26 ). As demonstrated by Min et al. ( 26 ), water velocity is a critical variable influencing snail occurrence. Not surprisingly, Bu. africanus and Bi. pfeifferi tend to inhabit stagnant water or slow-flowing water. However, the effect of slope on the distribution of Bu. africanus and Bi. pfeifferi across the LSV has yet to be determined. The distance to water or surface water proximity variable was included because the distribution of freshwater snails is determined by surface water (permanent or ephemeral). In accordance with previous studies, this study posits that surface water influences the composition of habitat variables relevant to Bu. africanus and Bi. pfeifferi . For example, vegetation composition and soil characteristics promote or inhibit (micro)biological processes ( 26 , 31 , 38 , 39 ). Notably, typically Bu. africanus and Bi. pfeifferi snails are frequently found close to water‒land edges ( 10 ). Hence, the presence of and proximity to a waterbody were predicted to be positively associated with Bu. africanus and Bi. pfeifferi presence. It was also expected that soil properties would have an intermediate and interactive causal influence on Bu. africanus and Bi. pfeifferi occurrence, resulting in spatial associations of particular soil group(s) with the presence or absence of snails ( 10 ). This is because several lines of evidence show that soil properties - physical (texture, porosity, colour) and chemical (mineralogy, organic matter content, acidity and alkalinity) -significantly govern soil water dynamics, ultimately affecting plant assemblages and the population growth of intermediate snail hosts ( 40 , 41 ). The bioclimatic data (19 Bioclim variables) were retrieved from the WorldClim database ( https://www.worldclim.org/data/bioclim.html ). Future changes in habitat suitability (2021–2040) were predicted using bioclimatic data under the shared socioeconomic pathway 585 (SSP585) scenario, which represents a high-emissions future scenario ( 42 ). The inclusion of future climate projections enabled an assessment of potential shifts in habitat suitability under changing climatic conditions. The Bioclim variables were downloaded using the R package geodata (version 0.6–2, Hijmans et al., 2024). The elevation and soil group datasets were obtained from the Malawi Spatial Data Platform (MASDAP, http://www.masdap.mw/ ). On MASDAP, the soil data are available as a vector, and were converted to a categorical raster the Rasterize tool in QGIS 3.22.1. Based on the national Soil and Terrain database for Malawi, a total of ten soil groups were identified for the study area: Luvisols, Arenosols, Cambisols, Solonetz, Gleysols, Leptosols, Lixisols, Phaeozoms, Solonetz and Vertisols. The slope (Shuttle Radar Topography Mission-based) and distance to water bodies (OpenStreetMap-based) data were obtained from WorldPop ( https://hub.worldpop.org/project/categories?id=14 ). The NDVI, which is a quantitative measure of vegetation greenness, was computed from Sentinel-2 imagery in the Google Earth Engine. This study aggregated the seasonal NVDI time series for 2022–23 (May–June) to derive the mean annual NDVI. The purpose of this was to identify how vegetation density and changes in greenness, as captured in a satellite image, affect the distribution of Bu. africanus and Bi. pfeifferi in LSV. This is because the removal or establishment of vegetation cover, particularly invasive macrophytes and hydrophytes, alters snail abundance and human schistosome transmission. For example, hydrophyte cover by water hyacinth is highly correlated with snail abundance and increases the total production of human-infectious cercariae sixfold ( 43 ). Species distribution modelling To predict the suitable habitat areas for Bi. pfeifferi and Bu. africanus using a suite of environmental covariates, an SDM model was developed based on the ensemble (consensus) method comprising the following machine learning (ML) models: random forest (RF), support vector machine (SVM) and multilayer perceptron (MLP); see Table 1 (Naimi & Araújo, 2016; Singleton et al., 2024). RF, SVM and MLP are nonparametric ML models and were used to classify the presence or absence of Bi. pfeifferi and Bu. africanus without making assumptions about the occurrence data. RF is an ensemble learning method that constructs multiple decision trees to enhance predictive performance and reduce overfitting ( 44 ). SVM is a classification method that finds the hyperplane that best separates the data into different classes ( 45 ). The MLP is an artificial neural network method that is composed of multiple layers of interconnected nodes, each performing a weighted sum of inputs followed by a nonlinear activation function ( 46 ). There is growing recognition of the importance of combining models, often referred to as ensemble modelling, with the aim of capturing complementary strengths from individual algorithms. This approach enhances predictive performance, robustness, and generalizability by leveraging diverse model architectures and perspectives ( 47 ). Thus, the present study uses a weighted model averaging over all predictions from three fitted models: RF, SVM and MLP. Potential multicollinearity (nonindependence of predictor variables) was evaluated by calculating variance inflation factors (VIFs) ( 48 ). The VIF is a measure used to detect the severity of multicollinearity. A VIF value above 10 indicates high collinearity; however, a threshold of 5 is not uncommon ( 48 ). This study used the vifstep function available from the sdm package to detect and exclude variables with the highest VIF (> 10 threshold). The collinearity test is particularly useful in the SDM because if the predictor variables are highly correlated, the individual effect of each variable is difficult to delineate. A major problem with this approach is that it can cause inflation of standard errors on estimates and wrong identification of relevant variables in a model. Consequently, erroneous extrapolation can occur. Table 1 Fitted snail SDMs. Model Description Definition RF Ensemble of decision trees with randomization. Classification: \(\:{\text{Ĉ}}_{rf}^{B}\left(x\right)=majority\:vote{\left\{{\text{Ĉ}}_{b}\right(x\left)\right\}}_{1}^{B}\) Where: \(\:{\text{Ĉ}}_{rf}^{B}\left(x\right)\) is the predicted class for an input \(\:x\) based on RF, \(\:{\text{Ĉ}}_{b}\left(x\right)\) is the predicted class for \(\:x\) from the \(\:bth\) tree in the forest, \(\:B\) is the total number of trees in the forest, and \(\:majority\:vote\) is the class that appears most frequently among the predictions ( 44 ). SVM Finds optimal separation hyperplane in feature space. Classification rule: \(\:h\left(x\right)=\:\sum\:_{i=1}^{n}{\alpha\:}_{i}{y}_{i}K\left({X}_{i},X\right)+b\) Where: \(\:{\alpha\:}_{i}\) is the Lagrange multiplier found during optimization, \(\:{y}_{i}\) are class labels (+ 1 or -1), \(\:K\left({X}_{i},X\right)\) is the Kernel function measuring similarity between data points, \(\:n\) is the number of support vectors, and \(\:b\) is the bias term (offset) that shifts the decision boundary ( 45 ). MLP A type of neural network composed of multiple layers of nodes. Prediction: \(\:ŷ=f({W}^{\left(L\right)}\bullet\:f\left({W}^{\left(L-1\right)}\bullet\:\cdots\:f\left({W}^{\left(1\right)}\bullet\:x+{b}^{\left(1\right)}\right)+{b}^{\left(L-1\right)}\right)+{b}^{\left(L\right)})\) Where: ŷ is the predicted output of the MLP, in this case class label (presence or absence), \(\:f(\bullet\:)\) is the activation function applied elementwise at each layer, \(\:L\) is the total number of layers in the network (excluding the input layer), \(\:x\) is the input vector (features of the data), \(\:{W}^{\left(i\right)}\) is the weight matrix for the \(\:i\) -th layer, \(\:{b}^{\left(i\right)}\) is the bias vector for the \(\:i\) -th layer, and \(\:·\) is the matrix multiplication between layers ( 46 ). Model performance evaluation Several performance metrics were applied to assess the classification accuracy and robustness of the models presented in the study. The receiver operating characteristic (ROC)-area under the curve (AUC) was used as the primary metric for measuring a model's discrimination ability. This statistic evaluates the trade-off between sensitivity (the proportion of correctly predicted presences) and specificity (the proportion of correctly predicted absences) ( 47 , 49 ). An AUC score of 1 represents perfect classification, whereas a score of 0.5 suggests that the model performs no better than random classification. In addition to the ROC-AUC, the true skill statistic (TSS) and correlation coefficient (COR) were also calculated to provide a more comprehensive assessment of model performance. Here, COR is used to evaluate the strength and direction of the relationship between observed values (true outcomes) and predicted values (model predictions) ( 48 ), with a high positive COR (close to 1) indicating strong alignment and good performance, a COR near 0 indicating poor performance, and a negative COR indicating inverse correlation and poor performance. The TSS is a balanced measure of both sensitivity and specificity ( 49 ). The TSS values range from − 1 to 1, with values closer to 1 indicating better model performance. A TSS value of 0 or less indicates that a model's predictive capacity is no better than random. By applying the ROC-AUC, COR and TSS, the present study ensured a robust evaluation of the models’ ability to accurately predict Bi. pfeifferi and Bu. africanus snail distribution based on environmental variables. Relative variable importance This analysis evaluated the relative importance of each environmental variable included in the ensemble of predictive models. This study implemented the permutation-based variable importance function available in the sdm package ( 47 ). The purpose of this was to identify the most important variable(s) determining the geographic distribution of Bu. africanus and Bi. pfeifferi in the LSV floodplain. Two complementary evaluation metrics, AUC and COR (Pearson correlation) were used to quantify the contribution of each variable. This dual evaluation allowed for a more comprehensive understanding of each variable’s contribution to the model’s predictive performance. Results and discussion Predicted current landscape of intestinal and urogenital schistosomiasis risk Using an ensemble of RF, SVM and MLP models, the distributions of Bi. pfeifferi and Bu. africanus in LSV was predicted (Fig. 2 ). The resulting prediction maps reveal marked differences in range and habitat suitability between the snails. The ensemble model for Bi. pfeifferi showed that a substantial area in LSV is an unsuitable habitat for Bi. pfeifferi. Broadly, contemporary environmental conditions appear harsh for the snail. However, at the fine scale, Chikwawa contains pockets of suitable conditions. As a result, a few areas in Chikwawa—namely, Nchalo, the eastern bank of the Shire River, Chipwaila, and Ngabu—are identified as suitable habitats for Bi. pfeifferi . Similarly, in Nsanje, greater suitability was predicted in a few areas around Manjolo, Ngabu and bordering Chikwawa South. Thus, in both districts, the majority of areas were characterized by a very low to low probability of presence, including Chileka, Chikwawa Boma, Therere, Thekerani, Muona, Bangula, Tengani, Nsanje Boma, Chimombo, and Marka. Furthermore, moderate suitability - albeit scanty - was nowcast along the Mwanza River near Misomali, Nchalo and Bangula. Turning now to the ensemble model for Bu. africanus , what stands out in the prediction map is the dominance of suitable areas in most areas of Chikwawa and Nsanje, particularly along the Mwanza River, Shire River and marshland. The map clearly shows that the areas highly suitable for Bu. africanus is present in riverine and wetland areas. This is particularly evident in floodplain areas with inundated vegetation and vegetation cover. For example, Thekerani, Muona, Bangula, Tengani, Nsanje Boma, Chimombo and Marka presented the widest distributions of snails in both districts. The Bu. africanus distribution map also indicates a large ecologically unsuitable habitat in northern Chikwawa (areas surrounding the Majete Wildlife Reserve) and a small unsuitable habitat in western Nsanje. The reports on the prevalence of urogenital schistosomiasis in Chikwawa and Nsanje, combined with malacological surveys, indicate a high prevalence rate and wide species range encompassing most parts of the valley. For example, epidemiological surveys in 2024 revealed a high burden for urogenital schistosomiasis among school-aged children in Chikwawa (prevalence: 35.0%, 95% CI: 33.6–36.5%), whereas snail surveys revealed a wide distribution and high abundance. On the other hand, the prevalence of intestinal schistosomiasis was low at 1.9% (95% CI: 1.4–2.3%). The low burden of intestinal schistosomiasis could be attributed to the rarity and restricted range of Bi. pfeifferi , evidently an elusive species, as noted by Poole et al. ( 50 ). This study detected Bi. pfeifferi at only four out of 70 sites, indicating that it is a rare species with few positive observations (see Fig. 1 a). The implication of this finding is that while the study area can be considered low risk for intestinal schistosomiasis, the disease remains highly focalized. In general, these results show that the modelled distribution of Bu. africanus group shows large areas of highly suitable habitat in central Chikwawa and eastern Nsanje. That is, both districts have the most suitable habitats for Bu. africanus . In contrast, the predicted distributions of Bi . pfeifferi shows a constrained range and few suitable habitats that can support the snail. Interestingly, the marshlands (Elephant Marsh, Ndindi Marsh in Nsanje District) appear to be completely unsuitable habitats for Bi . pfeifferi . The most suitable habitats for Bi. pfeifferi are sparse, fragmented, and do not overlap with those of Bu. africanus . These findings reinforce the idea that Bulinus host snails are generalist species with higher survival rates and broad tolerances to varying environmental conditions (e.g., ( 51 ),( 52 ) than Biomphalaria sp. in the bioclimatic areas of Africa. Predicted future landscape of intestinal and urogenital schistosomiasis risk The results obtained from the forecasting of the distribution of Bu. africanus and Bi. pfeifferi in the LSV under the worst-climate change scenario are presented in Fig. 3 a and b. For Bi. pfeifferi , Fig. 3 a shows a noticeable expansion of suitable habitats (brown areas: 298 km 2 ) in eastern Chikwawa and northern Nsanje, whereas considerable areas across Chikwawa and western Nsanje will experience habitat loss (green zones: 791 km 2 ). Conversely, Bu. africanus (Fig. 3 b) reveals a significant widespread pattern of habitat gain (brown zones: 3120 km 2 ) across the entire LSV, with only minor suitability loss (green zones: 154 km 2 ), mainly along the Mwanza and Shire riverine areas in Chikwawa. A comparison of Fig. 3 a and b highlights the differential impacts of climate change on the two intermediate host species of schistosomiasis. The more pronounced habitat gains for Bu. africanus suggest that this species may benefit from the warmer and wetter conditions predicted under the SSP585 scenario. This expansion could heighten the risk of urogenital schistosomiasis transmission in areas previously unaffected, creating new public health challenges in LSV. In contrast, the prominent habitat loss and marginal habitat gains predicted for Bi. pfeifferi suggests a potential retraction and shift in the risk areas for intestinal schistosomiasis. However, the persistence of suitable habitats in central Chikwawa (Nchalo Sugar Estate), the only known site with abundant Bi. pfeifferi ) suggests that the area may continue to serve as a transmission hotspot. These findings highlight the complex interplay between climate change and schistosomiasis ecology, emphasizing the need for tailored control measures for both intestinal and urogenital schistosomiasis. For Bu. africanus , environmental monitoring in areas predicted to gain suitability is essential, coupled with epidemiological surveys to prevent disease spread. For Bi. pfeifferi , the focus should be on sustaining control efforts in the core habitats that are projected to remain suitable, particularly in the agricultural irrigated land of Nchalo Sugar Estate. Multicollinearity Among the 24 input variables, 15 variables were identified with collinearity problems, namely, bio2, bio4, bio5, bio8, bio9, bio10, bio11, bio12, bio16, bio17, bio18, bio19, bio14 and elevation. This indicates that these variables are highly correlated with each other, which can cause inflated standard errors and less reliable coefficient estimates in the prediction model. The variables that remained are bio1, bio3, bio6, bio13, bio15, slope, soil, distance to water and the NDVI (Additional file: Fig. 1 ). The remaining variables showed acceptable levels of multicollinearity, as indicated by their VIF values, all of which are below the threshold of 10 (Table 2 ). This indicated a well-balanced set of variables with negligible multicollinearity. The minimum linear correlation coefficient was 0.004 between bio15 and bio6. This indicates a very weak negative correlation between these two variables. The maximum linear correlation coefficient was 0.822 between bio15 and bio13, indicating a strong positive correlation between these two variables. Table 2 VIFs of the remaining variables. Variable VIF bio 1 3.67 bio 3 5.09 bio 6 4.43 bio 13 9.19 bio 15 5.27 Slope 2.54 Soil 2.18 Distance to water body 1.61 NDVI 1.79 Relative variable importance The variable importance plot for Bu. africanus ensemble (Fig. 4 a) shows that the most important variable in the model is the NDVI (COR = 0.59, AUC = 0.37). The next two important variables are the mean annual temperature (bio1, COR = 0.13, AUC = 0.18) and slope (COR = 0.97, AUC = 0.11). The other variables had lower importance: distance to the water body (COR = 0.04, AUC = 0.03), soil type (COR = 0.03, AUC = 0.05), precipitation seasonality (bio15, COR = 0.05, AUC = 0.03), precipitation in the wettest month (bio13, COR = 0.02, AUC = 0.05), minimum temperature in the coldest month (bio6, COR = 0.03, AUC = 0.05) and isothermality (bio3, COR = 0.02, AUC = 0.03). Turning now to the variable importance plot for Bi. pfeifferi (Fig. 4 b), precipitation seasonality (COR = 0.32, AUC = 0.18) was the most important variable, followed by precipitation in the wettest month (COR = 0.17, AUC = 0.15). Isothermality showed moderate importance (COR = 0.16, AUC = 0.07). The least important variables in order of magnitude from relatively low importance to very low importance are the minimum temperature of the coldest month (COR = 0.07, AUC = 0.03), the distance to the waterbody (COR = 0.11, AUC = 0.04), the mean annual temperature (COR = 0.05, AUC = 0.04), the soil type (COR = 0.05, AUC = 0.02), the NDVI (COR = 0.03, AUC = 0.02), and the slope (COR = 0.02, AUC = 0.01). The results highlight the importance of climatic factors over topographic and vegetation factors in determining Bi. pfeifferi distribution. Although broadly consistent with Monde et al. ( 53 ), this outcome is contrary to that of Ponpetch et al. ( 40 ), who reported that soil properties and elevation are key factors in the distribution of Bi. pfeifferi in Ethiopia. Notably, this study has methodological limitations due to the difficulty in obtaining spatial coverage for proximal environmental predictors, such as physicochemical water parameters (e.g., pH and conductivity). Here, ecological processes (e.g., predators, competitors, and dispersal barriers such as rapids and mountain ranges), anthropogenic changes (e.g., agricultural expansion) and disturbances (e.g., floods and droughts) were not included in the SDM. Therefore, this study acknowledges the potential influence of other critical environmental predictors that remain unidentified and excluded. For example, the LSV is an agro-landscape currently undergoing large-scale agricultural expansion driven by the Shire Valley Transformation Project. In this geographical context, the expansion of cropland has been shown to significantly alter natural habitats, leading to habitat loss and fragmentation ( 54 ). The loss of habitat and biodiversity can subsequently trigger cascading effects within ecological communities, impacting species abundance, distribution, and interactions ( 55 ). Unlike environmental conditions, biotic processes are rarely integrated into species distribution models, underscoring the need for further attention ( 56 ). Here, a far-reaching consequence is that relying on only available data (often referred to as "weak" data) instead of more comprehensive data ("hard" data) can result in skewed or inaccurate predictions, potentially leading to an incomplete understanding of snail habitat suitability. Model performance and evaluation Table 3 provides a performance summary of the predictive models used to predict the distributions of Bi. pfeifferi and Bu. africanus in the LSV. The AUC, COR and TSS values underscore the model’s capacity to delineate suitable snail habitats from unsuitable areas. For Bu. africanus , the SVM model achieved the highest performance (AUC = 0.74, COR = 0.51, TSS = 0.56, deviance = 0.85), followed by RF (AUC = 0.73, COR = 0.33, TSS = 0.57, deviance = 0.97), both of which demonstrated high prediction accuracy. In contrast, the MLP showed weak performance (AUC = 0.56, COR = 0.07, and TSS = 0.44) and the highest deviance (2.04), indicating high prediction error. This suggests that the MLP had poor classification performance for Bu. africanus , a species that is widely distributed and abundant in the LSV. For Bi. pfeifferi model, MLP (AUC = 0.83, COR = 0.42, TSS = 0.75, deviance = 0.81), outperformed RF (AUC = 0.77, COR = 0.34, TSS = 0.69, deviance = 0.77) and SVM (AUC = 0.73, COR = 0.38, TSS = 0.64, deviance = 0.78). The MLP model struggled to predict Bu. africanus distribution; here, the model performed reasonably well, showing very good predictive accuracy and model fit for Bi. pfeifferi , a species that is rare and patchy in the LSV. This highlights significant variability in performance for the MLP and more stable performance and consistent results for the RF and SVM. Overall, the RF, SVM and MLP models provided acceptable results, with SVM being the top performer across metrics. Table 3 Model performance of Bu. africanus and Bi. pfeifferi distribution models across ML model types on the test dataset generated using partition. Species Model type AUC COR TSS Deviance Bu. africanus RF 0.73 0.33 0.57 0.97 SVM 0.74 0.51 0.56 0.85 MLP 0.56 0.07 0.44 2.04 Bi. pfeifferi RF 0.77 0.34 0.69 0.77 SVM 0.73 0.38 0.64 0.78 MLP 0.83 0.42 0.75 0.81 Conclusion Three conclusions can be drawn from this study. First, the habitat suitability map for Bu. africanus and Bi. pfeifferi in LSV reveals that the distribution of suitable habitats for urogenital and intestinal schistosomiasis transmission is not uniform. Bu. africanus habitat is the most dominant and abundant, indicating a widespread and high risk of urogenital schistosomiasis across the valley. In contrast, environmental conditions, in climatic terms, appear harsh for Bi. pfeifferi . Nonetheless, while Bi. pfeifferi habitat is patchy and scarce, indicating a lower risk of intestinal transmission, there is still reason to suggest that the valley has a nonnegligible risk profile. Second, in LSV, vegetation cover is the most important predictor of Bu. africanus distribution, whereas precipitation variables are most important for Bi. pfeifferi . Third, under the SSP585 scenario, Bu. africanus habitats are projected to expand, increasing the risk of urogenital schistosomiasis in new areas. On the other hand, Bi. pfeifferi habitats show moderate gains with a limited shift, indicating the potential persistence of intestinal schistosomiasis in Chikwawa and a low risk in Nsanje. This spatial understanding is crucial for targeted snail control and mass drug administration with praziquantel. Abbreviations SDM Species distribution model LSV Lower Shire Valley km Kilometre NDVI Normalized difference vegetation index SSP585 Shared Socioeconomic Pathway 585 ML Machine learning RF Random forest SVM Support vector machine MLP Multilayer perceptron VIF Variance inflation factor ROC Receiver Operating Characteristics AUC Area Under the Curve TSS True Skill Statistic COR Correlation Declarations Ethics approval Approval for this study was received from the College of Medicine Research Ethics Committee (COMREC) (Protocol number: P.02/23/3989), Chikwawa and Nsanje District Health Office research committees and the ILLOVO Nchalo Estate. Consent for publication Not applicable. Authors' contributions CN: Conceptualization of this study, funding acquisition, investigation (lead), analysis, project administration, visualization, method, writing-original draft preparation, review and editing. JC: Supervision, writing – review and editing. CMJ: Supervision, writing-review and editing. EAK: Investigation, data collection and review. JAT: Supervision, writing, review and editing. JRS: Supervision, investigation, data collection, resources, writing-review and editing. All the authors read and approved the final manuscript. Data availability The datasets used and/or analysed during the current study are available from the corresponding author upon reasonable request. Historical and future climate and elevation data are provided by WorldClim and are available at https://www.worldclim.org/data/bioclim.html. Competing interests The authors declare that they have no competing interests. Funding statement This work was supported by the National Institute for Health Research (NIHR) (using the UK’s Official Development Assistance (ODA) Funding) and Wellcome [223660/Z/21/Z] under the NIHR-Wellcome Partnership for Global Health Research. JAT is funded by the MRC Skills Development Fellowship [MR/T031743/1]. This UK-funded award is part of the EDCTP2 programme supported by the European Union. The views expressed are those of the authors and not necessarily those of Wellcome, the NIHR or the Department of Health and Social Care. Acknowledgements We would like to thank the communities of Chikwawa and Nsanje for welcoming us and allowing this study. CN would like to thank Chimwemwe Jamali, Peter Makaula, Gladys Namacha, Frank Mbalume, Chisomo Lwanda, Dalitso Damiano, Joel Kanyanga and Calisto Moda for their assistance during data collection. References WHO. Schistosomiasis (bilharzia) [Internet]. Geneva: World Health Organization; 2024 [cited 2024 Oct 25]. Available from: https://www.who.int/health-topics/schistosomiasis#tab=tab_1 Chiepa B, Mbewe R, Stanton MC, Kapumba B, Kambewa E, Kaunga L, et al. A baseline epidemiological survey for malaria and schistosomiasis reveals an alarming burden in primary schools despite ongoing control in Chikwawa District, southern Malawi. Curr Res Parasitol Vector-Borne Dis. 2024;5:100183. Nelwan ML. Schistosomiasis: Life Cycle, Diagnosis, and Control. Curr Ther Res. 2019;91:5–9. Gryseels B, Polman K, Clerinx J, Kestens L. Human schistosomiasis. The Lancet. 2006 Sep;368(9541):1106–18. Cetron MS, Chitsulo L, Sullivan JJ, Pilcher J, Wilson M, Noh J, et al. Schistosomiasis in Lake Malawi. The Lancet. 1996 Nov;348(9037):1274–8. Zheng L, Deng L, Zhong Y, Wang Y, Guo W, Fan X. Molluscicides against the snail-intermediate host of Schistosoma: a review. Parasitol Res. 2021 Oct;120(10):3355–93. Alharbi MH, Condemine C, Christiansen R, LaCourse EJ, Makaula P, Stanton MC, et al. Biomphalaria pfeifferi Snails and Intestinal Schistosomiasis, Lake Malawi, Africa, 2017–2018. Emerg Infect Dis. 2019 Mar;25(3):613–5. H.N Mtethiwa A. Prevalence and Intensity of Schistosomiasis in Communities around Water Reservoirs in Malawi. J Trop Dis [Internet]. 2016 [cited 2024 Nov 14];04(01). Available from: http://www.esciencecentral.org/journals/prevalence-and-intensity-of-schistosomiasis-in-communities-around-water-reservoirs-in-malawi-2329-891X-1000183.php?aid=65296 Makaula P, Sadalaki JR, Muula AS, Kayuni S, Jemu S, Bloch P. Schistosomiasis in Malawi: a systematic review. Parasit Vectors. 2014 Dec;7(1):570. Reed AL, Al-Harbi MH, Makaula P, Condemine C, Hesketh J, Archer J, et al. A geospatial analysis of local intermediate snail host distributions provides insight into schistosomiasis risk within undersampled areas of southern Lake Malawi. Parasit Vectors. 2024 Jun 27;17(1):272. Wiselka M, Robinson M, Clipsham K, Weddon S. The epidemiology of schistosomiasis in central Malawi. 1988; Souza CPD. Molluscicide control of snail vectors of schistosomiasis. Mem Inst Oswaldo Cruz. 1995 Apr;90(2):165–8. WHO. Field use of molluscicides in schistosomiasis control programmes: an operational manual for programme managers. World Health Organization, Geneva; 2017. WHO. Diagnostic target product profiles for monitoring, evaluation and surveillance of schistosomiasis control programmes. World Health Organization, Geneva; 2021. Ministry of Health, Republic of Malawi. Malawi Neglected Tropical Diseases Master 2015 to 2020; Master Plan for Malawi. Ministry of Health, Lilongwe; 2014. Boitt MK, Suleiman MK. Mapping of Freshwater Snails’ Habitat—A Source of Transmitting Bilharzia in Mwea Sub-County, Kenya. J Geosci Environ Prot. 2021;09(10):130–50. Manyangadze T, Chimbari MJ, Gebreslasie M, Ceccato P, Mukaratirwa S. Modelling the spatial and seasonal distribution of suitable habitats of schistosomiasis intermediate host snails using Maxent in Ndumo area, KwaZulu-Natal Province, South Africa. Parasit Vectors. 2016 Dec;9(1):572. FAO. Malawi Country Profile: Physiography, climate and soils [Internet]. Food and Agriculture Organization; 2002 [cited 2024 Feb 16]. Available from: https://www.fao.org/3/ac468e/ac468e03.htm#bm03 Thieme M, Abell R, Stiassny M, Skelton P. Freshwater Ecoregions of Africa and Madagascar: A Conservation Assessment. Bibliovault OAI Repos Univ Chic Press. 2005 Jan;1–431. National Statistical Office. Malawi 2018 Population and Housing Census Main Report [Internet]. Malawi Government; 2019. Available from: https://malawi.unfpa.org/en/publications/malawi-2018-population-and-housing-census-main-report Ministry of Health, Republic of Malawi. Malawi Neglected Tropical Diseases Master Plan 2023 - 2030 [Internet]. Available from: https://espen.afro.who.int/system/files/content/resources/Malawi%20NTD%20Master%20Plan%202023-2030.pdf Briers RA. Range limits and parasite prevalence in a freshwater snail. Proc R Soc Lond B Biol Sci [Internet]. 2003 Nov 7 [cited 2024 Oct 27];270(suppl_2). Available from: https://royalsocietypublishing.org/doi/10.1098/rsbl.2003.0046 Krauth SJ, Wandel N, Traoré SI, Vounatsou P, Hattendorf J, Achi LY, et al. Distribution of intermediate host snails of schistosomiasis and fascioliasis in relation to environmental factors during the dry season in the Tchologo region, Côte d’Ivoire. Adv Water Resour. 2017 Oct;108:386–96. Magero VO, Kisara S, Wade CM. Geographical Distribution of Biomphalaria snails in East Africa [Internet]. 2021 [cited 2024 Oct 26]. Available from: http://biorxiv.org/lookup/doi/10.1101/2021.11.04.467236 Glidden CK, Singleton AL, Chamberlin A, Tuan R, Palasio RGS, Caldeira RL, et al. Climate and urbanization drive changes in the habitat suitability of Schistosoma mansoni competent snails in Brazil. Nat Commun. 2024 Jun 19;15(1):4838. Min F, Wang J, Liu X, Yuan Y, Guo Y, Zhu K, et al. Environmental Factors Affecting Freshwater Snail Intermediate Hosts in Shenzhen and Adjacent Region, South China. Trop Med Infect Dis. 2022 Dec 9;7(12):426. Walz Y, Wegmann M, Dech S, Raso G, Utzinger J. Risk profiling of schistosomiasis using remote sensing: approaches, challenges and outlook. Parasit Vectors. 2015 Dec;8(1):163. Stanton MC, Adriko M, Arinaitwe M, Howell A, Davies J, Allison G, et al. Intestinal schistosomiasis in Uganda at high altitude (>1400 m): malacological and epidemiological surveys on Mount Elgon and in Fort Portal crater lakes reveal extra preventive chemotherapy needs. Infect Dis Poverty. 2017 Dec;6(1):34. Stensgaard AS, Utzinger J, Vounatsou P, Hürlimann E, Schur N, Saarnak CFL, et al. Large-scale determinants of intestinal schistosomiasis and intermediate host snail distribution across Africa: does climate matter? Acta Trop. 2013 Nov;128(2):378–90. Malone JB, Yilma JM, McCarroll JC, Erko B, Mukaratirwa S, Zhou X. Satellite climatology and the environmental risk of Schistosoma mansoni in Ethiopia and east Africa. Acta Trop. 2001 Apr;79(1):59–72. Olkeba BK, Boets P, Mereta ST, Yeshigeta M, Akessa GM, Ambelu A, et al. Environmental and biotic factors affecting freshwater snail intermediate hosts in the Ethiopian Rift Valley region. Parasit Vectors. 2020 Dec;13(1):292. Shi Y, Qiu J, Li R, Shen Q, Huang D. Identification of Potential High-Risk Habitats within the Transmission Reach of Oncomelania hupensis after Floods Based on SAR Techniques in a Plane Region in China. Int J Environ Res Public Health. 2017 Aug 30;14(9):986. Simoonga C, Utzinger J, Brooker S, Vounatsou P, Appleton CC, Stensgaard AS, et al. Remote sensing, geographical information system and spatial analysis for schistosomiasis epidemiology and ecology in Africa. Parasitology. 2009 Nov;136(13):1683–93. Veerappan R, Sayed S. Urban flood susceptibility zonation mapping using evidential belief function, frequency ratio and fuzzy gamma operator models in GIS: a case study of Greater Mumbai, Maharashtra, India. Geocarto Int. 2020 Mar 4; Wu S, Yu M, Chen L. Nonmonotonic and spatial‐temporal dynamic slope effects on soil erosion during rainfall‐runoff processes. Water Resour Res. 2017 Feb;53(2):1369–89. Ezeugwu SMC, Obiamiwe BA. Application of biostatistics to the effects of water velocity on the distribution of Bulinus globosus and Bulinus forskali in an irrigation scheme in Borno State, Nigeria. Afr Sci. 2005;6(3). Ntonifor H N, Ajayi JA. Studies on the ecology and distribution of some medically important freshwater snail species in Bauchi State, Nigeria. 2007. 2007;1(2):121–7. Coelho PRS, Ker FTO, Araújo AD, Guimarães RicardoJPS, Negrão-Corrêa DA, Caldeira RL, et al. Identification of Risk Areas for Intestinal Schistosomiasis, Based on Malacological and Environmental Data and on Reported Human Cases. Front Med. 2021 Aug 6;8:642348. Huang Q ni, Tang L li, Jiang X guang, Chen Z, Zhou X nong. [Retrieving eco-environment factors relevant to Oncomelania snail distribution based on QuickBird image]. Zhongguo Ji Sheng Chong Xue Yu Ji Sheng Chong Bing Za Zhi. 2007 Aug;25(4):304–9. Ponpetch K, Erko B, Bekana T, Kebede T, Tian D, Yang Y, et al. Environmental Drivers and Potential Distribution of Schistosoma mansoni Endemic Areas in Ethiopia. Microorganisms. 2021 Oct 14;9(10):2144. Rodrigues PMS, Schaefer CEGR, Silva JDO, Ferreira Júnior WG, Dos Santos RM, Neri AV. The influence of soil on vegetation structure and plant diversity in different tropical savannic and forest habitats. J Plant Ecol. 2016 Dec 26;rtw135. Riahi K, Van Vuuren DP, Kriegler E, Edmonds J, O’Neill BC, Fujimori S, et al. The Shared Socioeconomic Pathways and their energy, land use, and greenhouse gas emissions implications: An overview. Glob Environ Change. 2017 Jan;42:153–68. Desautels DJ, Hartman RB, Weber ME, Jacob N, Sun A, Civitello DJ. Experimental water hyacinth invasion and destructive management increase human schistosome transmission potential. Ecol Appl. 2023 Mar;33(2):e2767. Breiman L. Random Forests. Mach Learn. 2001;45(1):5–32. Vapnik VN. The Nature of Statistical Learning Theory [Internet]. New York, NY: Springer New York; 1995 [cited 2024 Dec 31]. Available from: http://link.springer.com/10.1007/978-1-4757-2440-0 Rumelhart DE, Hinton GE, Williams RJ. Learning representations by back-propagating errors. Nature. 1986 Oct;323(6088):533–6. Naimi B, Araújo MB. sdm: a reproducible and extensible R platform for species distribution modelling. Ecography. 2016 Apr;39(4):368–75. Dormann CF, Elith J, Bacher S, Buchmann C, Carl G, Carré G, et al. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography. 2013 Jan;36(1):27–46. Allouche O, Tsoar A, Kadmon R. Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). J Appl Ecol. 2006 Dec;43(6):1223–32. Poole H, Terlouw DJ, Naunje A, Mzembe K, Stanton M, Betson M, et al. Schistosomiasis in preschool-age children and their mothers in Chikhwawa district, Malawi with notes on characterization of schistosomes and snails. Parasit Vectors. 2014 Dec;7(1):153. Diakité NR, Koffi PB, Konan CK, Bassa FK, Chamberlin AJ, Ouattara M, et al. Variability of biological traits of Bulinus truncatus and Biomphalaria pfeifferi, the intermediate host snails of schistosomiasis, from three climatic zones of Côte d’Ivoire. Front Environ Sci. 2023 Dec 7;11:1193239. Van Der Deure T, Maes T, Huyse T, Stensgaard A. Climate change could fuel urinary schistosomiasis transmission in Africa and Europe. Glob Change Biol. 2024 Aug;30(8):e17434. Monde C, Syampungani S, Van Den Brink PJ. Natural and human induced factors influencing the abundance of Schistosoma host snails in Zambia. Environ Monit Assess. 2016 Jun;188(6):370. Nkolokosa C, Stothard R, Jones CM, Stanton M, Chirombo J, Tangena JAA. Monitoring and simulating landscape changes: how do long-term changes in land use and long-term average climate affect regional biophysical conditions in southern Malawi? Environ Monit Assess. 2023 Oct;195(10):1247. Senadeera M. Effects of land use on species diversity and ecosystem functioning in agricultural landscapes. 2023 Apr 24; Austin MP. Spatial prediction of species distribution: an interface between ecological theory and statistical modelling. Ecol Model. 2002 Nov;157(2–3):101–18. Additional Declarations No competing interests reported. Supplementary Files AdditionalfileNkolokosaetal.odt Cite Share Download PDF Status: Published Journal Publication published 29 Aug, 2025 Read the published version in Parasites & Vectors → Version 1 posted Editorial decision: Revision requested 25 Feb, 2025 Reviews received at journal 22 Feb, 2025 Reviewers agreed at journal 02 Feb, 2025 Reviewers invited by journal 31 Jan, 2025 Editor assigned by journal 29 Jan, 2025 Submission checks completed at journal 29 Jan, 2025 First submitted to journal 26 Jan, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5908499","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":408651344,"identity":"69180995-11e5-4720-9540-9c17f2522c72","order_by":0,"name":"Clinton Nkolokosa","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAt0lEQVRIiWNgGAWjYFACNgaGBwwJQAbzASCDWC0JYC1sEIoULTwGxGnhn92W+CChIk1Ovv3M1w2JbQzRBgcIaJG4c+ywQcKZHGODM7nbbgC15G4gpMVAIr1NIrGtInEDA0jLNiBJtJb5/W+eEasl7RhQS05iw40cNuK0SNxISwb6Jc3Y4MYzsxuJ/yRy9xPSwj8jzfDBh4pkOfn+5Gc3PpyxyZ3ZQEALhq0kqh8Fo2AUjIJRgBUAAPUlSQemgdyyAAAAAElFTkSuQmCC","orcid":"","institution":"Malawi-Liverpool-Wellcome Programme","correspondingAuthor":true,"prefix":"","firstName":"Clinton","middleName":"","lastName":"Nkolokosa","suffix":""},{"id":408651347,"identity":"22f313aa-6ced-41b9-90fb-99b48e51181b","order_by":1,"name":"James Chirombo","email":"","orcid":"","institution":"Malawi-Liverpool-Wellcome Programme","correspondingAuthor":false,"prefix":"","firstName":"James","middleName":"","lastName":"Chirombo","suffix":""},{"id":408651348,"identity":"e5125833-020d-4201-b060-c96ec1422cea","order_by":2,"name":"Christopher M. Jones","email":"","orcid":"","institution":"Liverpool School of Tropical Medicine","correspondingAuthor":false,"prefix":"","firstName":"Christopher","middleName":"M.","lastName":"Jones","suffix":""},{"id":408651349,"identity":"979f3edf-4583-4dd4-9c4b-17afe15aa074","order_by":3,"name":"Eggrey Aisha Kambewa","email":"","orcid":"","institution":"Malawi-Liverpool-Wellcome Programme","correspondingAuthor":false,"prefix":"","firstName":"Eggrey","middleName":"Aisha","lastName":"Kambewa","suffix":""},{"id":408651350,"identity":"9a508dc2-f552-4654-9065-cb35e3bce261","order_by":4,"name":"Julie-Anne Akiko Tangena","email":"","orcid":"","institution":"Liverpool School of Tropical Medicine","correspondingAuthor":false,"prefix":"","firstName":"Julie-Anne","middleName":"Akiko","lastName":"Tangena","suffix":""},{"id":408651351,"identity":"c04d774e-27af-41cc-873b-b445f671d076","order_by":5,"name":"J. Russell Stothard","email":"","orcid":"","institution":"Liverpool School of Tropical Medicine","correspondingAuthor":false,"prefix":"","firstName":"J.","middleName":"Russell","lastName":"Stothard","suffix":""}],"badges":[],"createdAt":"2025-01-26 22:53:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5908499/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5908499/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s13071-025-06952-3","type":"published","date":"2025-08-29T15:57:47+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":75169416,"identity":"06695cc7-8664-4d95-93b8-6b3d378d1db9","added_by":"auto","created_at":"2025-01-31 14:04:43","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1123896,"visible":true,"origin":"","legend":"\u003cp\u003eStudy area: the sampled locations and abundances of (a) \u003cem\u003eBu. africanus \u003c/em\u003eand (b) \u003cem\u003eBi. pfeifferi \u003c/em\u003ein Chikwawa and Nsanje during the 2023 malacological survey mapped using QGIS 3.34. The insert map shows the geographical location of LSV in the context of Malawi.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-5908499/v1/7709d0d57cadd4bb97a21f92.png"},{"id":75169420,"identity":"2be4b42e-213b-4451-a1dc-2ed25541b7ae","added_by":"auto","created_at":"2025-01-31 14:04:43","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":3199475,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted current (2023) environmental suitability for intestinal and urogenital schistosomiasis transmission in LSV mapped using QGIS 3.34. Warmer colours (red, orange) indicate very high to high habitat suitability, whereas green to light green colours indicate very low to low habitat suitability.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-5908499/v1/e16e7f3bc257491e83701237.png"},{"id":75169413,"identity":"12cc6fb6-cb90-4779-aeac-fc0c1a33fa7d","added_by":"auto","created_at":"2025-01-31 14:04:43","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":4444072,"visible":true,"origin":"","legend":"\u003cp\u003eFuture changes in habitat suitability for a) \u003cem\u003eBi. pfeifferi\u003c/em\u003e and b) \u003cem\u003eBu. africanus\u003c/em\u003e across the LSV as predicted using future climate data under the SSP585 scenario mapped using QGIS 3.34. Lost suitability (brown) indicates habitats projected to become unsuitable for snails due to degraded or unfavourable habitat conditions (i.e., retraction). No change (yellow) indicates regions where habitat suitability is projected to remain stable. Gained suitability (green) indicates areas where changing environmental conditions are expected to create new suitable habitats (i.e., expansion).\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-5908499/v1/3ad8465a555a6e5f2ca075ba.png"},{"id":75169421,"identity":"19c47d93-7279-4b80-8436-1aa320299e7e","added_by":"auto","created_at":"2025-01-31 14:04:43","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":311830,"visible":true,"origin":"","legend":"\u003cp\u003eVariable importance plots for a) \u003cem\u003eBu. africanus \u003c/em\u003eand b) \u003cem\u003eBi. pfeifferi\u003c/em\u003eensemble models.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-5908499/v1/15d8ce1179041746e0da5e5c.png"},{"id":90344894,"identity":"08c231ab-13b7-4944-b9b5-eeef96af3058","added_by":"auto","created_at":"2025-09-01 16:07:20","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":13172921,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5908499/v1/5a55aebd-3f05-4915-bfe7-f93a3ec66246.pdf"},{"id":75170268,"identity":"a1cfaa4d-4bac-415b-a13c-5c62c4861970","added_by":"auto","created_at":"2025-01-31 14:12:43","extension":"odt","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":2404697,"visible":true,"origin":"","legend":"","description":"","filename":"AdditionalfileNkolokosaetal.odt","url":"https://assets-eu.researchsquare.com/files/rs-5908499/v1/3c6b042edb2c405fe3f82863.odt"}],"financialInterests":"No competing interests reported.","formattedTitle":"Predicting current and future suitability for intermediate snail hosts of urogenital and intestinal schistosomiasis in a floodplain of Malawi","fulltext":[{"header":"Highlights","content":"\u003cp\u003e\u0026bull; We present a species distribution model (SDM) for and snails to predict potential habitats for the transmission of urogenital and intestinal schistosomiasis, respectively.\u003c/p\u003e\u003cp\u003e\u0026bull; The SDM comprises environmental predictors and an ensemble model of random forest, support vector machine and multilayer perceptron.\u003c/p\u003e\u003cp\u003e\u0026bull; The current suitability map reveals high suitability for in large wetland areas and a low suitability for in small cropland areas within the Lower Shire Valley (LSV), each showing distinct habitat ranges.\u003c/p\u003e\u003cp\u003e\u0026bull; The future suitability map forecasts a significant increase (46%) in suitable habitats for and a modest gain (4.4%) in suitable habitats for in LSV.\u003c/p\u003e\u003cp\u003e\u0026bull; Areas with a high probability of snail presence could be the first priority for both human surveillance and snail control.\u003c/p\u003e"},{"header":"Background","content":"\u003cp\u003eRelating the geographical distribution of freshwater snails (vectors) to local environmental attributes offers value for understanding the epidemiological landscape of schistosomiasis transmission in a changing environment. Schistosomiasis\u0026mdash;both urogenital and intestinal\u0026mdash;causes significant human suffering. Schistosomiasis, which affects approximately 240\u0026nbsp;million people globally, is a neglected tropical disease (NTD) caused by parasitic flatworms and is associated with human water contact behaviour (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). The disease is endemic in Malawi. Approximately 40\u0026ndash;50% of the Malawi population are at risk of being infected, with school-aged children being the highly infected and the most infected group (\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). If schistosomiasis is not treated, it can result in severe health complications, including infertility, anaemia, malnutrition, abdominal pain, enlarged or damaged liver, haematuria, and blood in the stool (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). In rare cases, convulsions, paralysis, or spinal cord inflammation may occur when eggs released by a pair of adult worms are found in the brain or spinal cord (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). Stigma, social exclusion and poor educational outcomes aggravate the suffering from schistosomiasis (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). The parasitic worms (trematodes) that cause the disease are of the genus \u003cem\u003eSchistosoma\u003c/em\u003e: \u003cem\u003eSchistosoma mansoni\u003c/em\u003e and \u003cem\u003eSchistosoma haematobium\u003c/em\u003e (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). The latter is responsible for intestinal schistosomiasis, and the former is responsible for urogenital schistosomiasis. The transmission chain depends on compatible snails. In Malawi, \u003cem\u003eS. mansoni\u003c/em\u003e and \u003cem\u003eS. haematobium\u003c/em\u003e parasites are transmitted by freshwater-intermediate host snails of the genera \u003cem\u003eBiomphalaria\u003c/em\u003e (Planorbidae) and \u003cem\u003eBulinus\u003c/em\u003e (Bulinadae), respectively (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). Larval schistosomes (cercariae) from infected freshwater snails penetrate human skin, causing (re)infection. This occurs in water bodies, for example, during routine agricultural, domestic, occupational and recreational activities, such as irrigating, washing, wading, bathing or swimming. Therefore, human\u0026ndash;surface water interactions and the presence of intermediate host snails predicate the spatial distribution of schistosomiasis prevalence. This is particularized in Malawi, where the consequence of proximity to surface water (lake, river, wetland, canal, dam, pond) is generally a high risk and prevalence of schistosomiasis (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan additionalcitationids=\"CR8 CR9 CR10\" citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAmong the different approaches used to control the spread of schistosomiasis, snail control is essential for interrupting the parasite life cycle (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). Eliminating snail hosts is considered an important, effective and convenient strategy for schistosomiasis prevention in endemic areas (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). Accordingly, the WHO recommends prevention and treatment: molluscicidal (chemical) control, ecological control (sanitation and environmental modification) and health education for behavioural change (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). Identifying the habitats where intermediate host snails occur could actively inform NTD control programs to address schistosomiasis, which is evidently an important public health challenge in Lower Shire Valley (LSV), southern Malawi (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). The Malawi Neglected Tropical Diseases Master Plan (2023\u0026ndash;2030) recognizes that a changing climate is a threat and challenge to schistosomiasis control. However, in the absence of evidence, endemicity dynamics and transmission risk resulting from anthropogenic, climatic and ecological changes remain unclear. Revealing the current and future schistosomiasis risk is vital for opportune control efforts and resolving the elusive climate change challenges, holding back national and global efforts to eliminate NTDs in resource-constrained settings, such as Malawi.\u003c/p\u003e \u003cp\u003eTo date, several studies across Africa have investigated the abundance, distribution and spread of freshwater snails at broad scales via a species distribution model (SDM). The task of an SDM is to determine the probability of a species occurring in a particular habitat as a function of a set of environmental conditions. For example, in Kenya, potential habitats of freshwater snails were mapped via an SDM based on maximum entropy (Maxent) (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). This study incorporated a set of environmental variables, including land surface temperature, soil pH, and vegetation greenness, to predict environmental suitability (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). In a related study, Maxent was also applied to forecast the distribution of suitable habitats for \u003cem\u003eBu. globosus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e in South Africa (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). In Malawi, Reed et al. (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e) provided a first step toward understanding the spatial risk of intestinal and urogenital schistosomiasis within under sampled areas via 2D mean Gaussian process prediction. In the present study, a machine learning classifier is utilized in potential distribution modelling over a mosaic habitat. Thus, the importance of this study is by no means restricted to intestinal and urogenital schistosomiasis in LSV but also potentially to other NTDs and forms of schistosomiasis of veterinary importance in other geographical settings. Therefore, with respect to LSV, the aim of this study is twofold: 1) developing and parameterizing an SDM for \u003cem\u003eBi. pfeifferi\u003c/em\u003e and \u003cem\u003eBu. africanus\u003c/em\u003e snails in LSV, and 2) using the parameterized model to now- and forecast the distribution of \u003cem\u003eBi. pfeifferi\u003c/em\u003e and \u003cem\u003eBu. africanus\u003c/em\u003e in LSV under current and future climate conditions.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eStudy area and species occurrence data\u003c/p\u003e \u003cp\u003eData of \u003cem\u003eBi. pfeifferi\u003c/em\u003e and \u003cem\u003eBu. africanus\u003c/em\u003e occurrence (absence/presence) samples were collected in the floodplains of the LSV between May and June 2023. The LSV study area (latitude: 14\u0026deg;25\u0026prime; S and 16\u0026deg;55\u0026prime; S and longitude: 35\u0026deg;16\u0026prime; E and 35\u0026deg;12\u0026prime; E) is the southern subregion of Malawi, which is situated in the lower Shire floodplain and consists of two districts: Chikwawa and Nsanje. The study area, covering approximately 6833 km\u003csup\u003e2\u003c/sup\u003e, is characterized by a wide range of topography (medium-altitude plain 750\u0026ndash;1300 m and floodplain 35\u0026ndash;105 m) and ecological zones (freshwater habitats, tropical grasslands, savannas and shrublands, montane forest-grassland mosaics, and flooded grasslands) (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). The climate is subtropical: a warm-wet season from November to April and a hot-dry season from May to October (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). According to the National Statistics Office, Chikwawa and Nsanje had estimated population densities of 128 people/km\u003csup\u003e2\u003c/sup\u003e and 168 people/km\u003csup\u003e2\u003c/sup\u003e in 2023, respectively (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). Across the valley, schistosomiasis is endemic and focal, with district-level prevalence considered low for \u003cem\u003eS. mansoni\u003c/em\u003e and moderate for \u003cem\u003eS. haematobium\u003c/em\u003e (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). For example, in 2017, the schistosomiasis prevalence rates in Chikwawa and Nsanje were 15.8% (95% CI: 10.9%, 22.4%) and 25.4% (95% CI: 15.3%, 38.9%), respectively (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThis study surveyed a total of 70 sampling sites across the LSV: Chikwawa (\u003cem\u003en\u003c/em\u003e\u0026thinsp;=\u0026thinsp;35) and Nsanje (\u003cem\u003en\u003c/em\u003e\u0026thinsp;=\u0026thinsp;35) (see Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The inclusion criteria were as follows: 1) confirmed cases of intestinal or urogenital schistosomiasis in the area and 2) the presence of surface water (wetland, river, dams, ponds, canals). Here, a malacologist and three trained field collectors who adhered to the WHO sampling design protocol (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e) collected the survey data and used a field guide to African freshwater snails to identify the snail intermediate hosts. Standardized sampling effort was achieved by setting the sampling time to 15 minutes per site. The collected dataset includes information on the geographic coordinates of the sampled sites, site number, data and time of collection, all freshwater snails encountered, habitat type, human- and animal-water contact, aerial photograph of the site, and geographical elevation. A hand-held global positioning system (GPS) device (Garmin Montana 700 GPS, US) was used to locate and map the sites. Using stratified random sampling, georeferenced occurrence records of \u003cem\u003eBi. pfeifferi\u003c/em\u003e and \u003cem\u003eBu. africanus\u003c/em\u003e groups were generated along the Shire River, Elephant Marsh and irrigation canals by foot, car and boat (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). From the malacological survey, a total of 597 \u003cem\u003eBi. pfeifferi\u003c/em\u003e and 1994 \u003cem\u003eBu. africanus\u003c/em\u003e samples were collected from four and 26 sites, respectively. Typical of studies determining the distribution of freshwater snails, field sampling was designed on the basis of the knowledge that such snails occur in lentic and lotic ecosystems: standing water and running water habitats, respectively (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan additionalcitationids=\"CR23\" citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). Therefore, the survey targeted representative aquatic habitats, such as lakes and marshlands, ponds and pools, rivers and canals, especially where human\u0026ndash;water contact occurs (fishing, gardening, bathing, swimming, washing, etc.).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eEnvironmental predictors\u003c/p\u003e \u003cp\u003eThe present study uses abiotic (climatic, topographic, soil and proximity) and biotic (vegetation cover) factors to determine the influence of environmental factors and gradients on \u003cem\u003eBu.\u003c/em\u003e africanus and \u003cem\u003eBi. pfeifferi\u003c/em\u003e distributions (Additional file: Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). More specifically, temperature, precipitation, geographical elevation, terrain slope, soil type, distance to a water body and normalized difference vegetation index (NDVI) variables were included in the snail distribution modelling. Previous studies have shown that abiotic factors, such as rainfall (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e), air temperature (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e), and altitude (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e), influence the distribution and habitat preference of freshwater snails of medical importance. For example, it has been demonstrated that temperature affects the mortality, fecundity, and growth rate of \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e snails (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e), and precipitation and elevation gradients are negatively correlated with freshwater snail distribution and abundance (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan additionalcitationids=\"CR32\" citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e). This study uses slope as a direct factor regulating water velocity, which is a key determinant of intermediate host snail habitat preference. The velocity of water tends to have a positive relationship with the slope gradient (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e). The importance of water velocity on freshwater snails has been previously established. Generally, \u003cem\u003eBulinus\u003c/em\u003e snails and \u003cem\u003eBiomphalaria\u003c/em\u003e snails prefer stagnant or slow flow (velocity\u0026thinsp;\u0026le;\u0026thinsp;0.3 m/s) (\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e). The argument is that rapidly flowing water impedes the establishment of reproductive colonies and displaces the snail population (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e). As demonstrated by Min et al. (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e), water velocity is a critical variable influencing snail occurrence. Not surprisingly, \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e tend to inhabit stagnant water or slow-flowing water. However, the effect of slope on the distribution of \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e across the LSV has yet to be determined.\u003c/p\u003e \u003cp\u003eThe distance to water or surface water proximity variable was included because the distribution of freshwater snails is determined by surface water (permanent or ephemeral). In accordance with previous studies, this study posits that surface water influences the composition of habitat variables relevant to \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e. For example, vegetation composition and soil characteristics promote or inhibit (micro)biological processes (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e). Notably, typically \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e snails are frequently found close to water‒land edges (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e). Hence, the presence of and proximity to a waterbody were predicted to be positively associated with \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e presence. It was also expected that soil properties would have an intermediate and interactive causal influence on \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e occurrence, resulting in spatial associations of particular soil group(s) with the presence or absence of snails (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e). This is because several lines of evidence show that soil properties - physical (texture, porosity, colour) and chemical (mineralogy, organic matter content, acidity and alkalinity) -significantly govern soil water dynamics, ultimately affecting plant assemblages and the population growth of intermediate snail hosts (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe bioclimatic data (19 Bioclim variables) were retrieved from the WorldClim database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.worldclim.org/data/bioclim.html\u003c/span\u003e\u003cspan address=\"https://www.worldclim.org/data/bioclim.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e).\u003c/span\u003e Future changes in habitat suitability (2021\u0026ndash;2040) were predicted using bioclimatic data under the shared socioeconomic pathway 585 (SSP585) scenario, which represents a high-emissions future scenario (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e). The inclusion of future climate projections enabled an assessment of potential shifts in habitat suitability under changing climatic conditions. The Bioclim variables were downloaded using the R package geodata (version 0.6\u0026ndash;2, Hijmans et al., 2024). The elevation and soil group datasets were obtained from the Malawi Spatial Data Platform (MASDAP, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.masdap.mw/\u003c/span\u003e\u003cspan address=\"http://www.masdap.mw/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e).\u003c/span\u003e On MASDAP, the soil data are available as a vector, and were converted to a categorical raster the Rasterize tool in QGIS 3.22.1. Based on the national Soil and Terrain database for Malawi, a total of ten soil groups were identified for the study area: Luvisols, Arenosols, Cambisols, Solonetz, Gleysols, Leptosols, Lixisols, Phaeozoms, Solonetz and Vertisols. The slope (Shuttle Radar Topography Mission-based) and distance to water bodies (OpenStreetMap-based) data were obtained from WorldPop (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://hub.worldpop.org/project/categories?id=14\u003c/span\u003e\u003cspan address=\"https://hub.worldpop.org/project/categories?id=14\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e).\u003c/span\u003e The NDVI, which is a quantitative measure of vegetation greenness, was computed from Sentinel-2 imagery in the Google Earth Engine. This study aggregated the seasonal NVDI time series for 2022\u0026ndash;23 (May\u0026ndash;June) to derive the mean annual NDVI. The purpose of this was to identify how vegetation density and changes in greenness, as captured in a satellite image, affect the distribution of \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e in LSV. This is because the removal or establishment of vegetation cover, particularly invasive macrophytes and hydrophytes, alters snail abundance and human schistosome transmission. For example, hydrophyte cover by water hyacinth is highly correlated with snail abundance and increases the total production of human-infectious cercariae sixfold (\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eSpecies distribution modelling\u003c/p\u003e \u003cp\u003eTo predict the suitable habitat areas for \u003cem\u003eBi. pfeifferi\u003c/em\u003e and \u003cem\u003eBu. africanus\u003c/em\u003e using a suite of environmental covariates, an SDM model was developed based on the ensemble (consensus) method comprising the following machine learning (ML) models: random forest (RF), support vector machine (SVM) and multilayer perceptron (MLP); see Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e (Naimi \u0026amp; Ara\u0026uacute;jo, 2016; Singleton et al., 2024). RF, SVM and MLP are nonparametric ML models and were used to classify the presence or absence of \u003cem\u003eBi. pfeifferi\u003c/em\u003e and \u003cem\u003eBu. africanus\u003c/em\u003e without making assumptions about the occurrence data. RF is an ensemble learning method that constructs multiple decision trees to enhance predictive performance and reduce overfitting (\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e). SVM is a classification method that finds the hyperplane that best separates the data into different classes (\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e). The MLP is an artificial neural network method that is composed of multiple layers of interconnected nodes, each performing a weighted sum of inputs followed by a nonlinear activation function (\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e). There is growing recognition of the importance of combining models, often referred to as ensemble modelling, with the aim of capturing complementary strengths from individual algorithms. This approach enhances predictive performance, robustness, and generalizability by leveraging diverse model architectures and perspectives (\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e). Thus, the present study uses a weighted model averaging over all predictions from three fitted models: RF, SVM and MLP.\u003c/p\u003e \u003cp\u003ePotential multicollinearity (nonindependence of predictor variables) was evaluated by calculating variance inflation factors (VIFs) (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e). The VIF is a measure used to detect the severity of multicollinearity. A VIF value above 10 indicates high collinearity; however, a threshold of 5 is not uncommon (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e). This study used the vifstep function available from the sdm package to detect and exclude variables with the highest VIF (\u0026gt;\u0026thinsp;10 threshold). The collinearity test is particularly useful in the SDM because if the predictor variables are highly correlated, the individual effect of each variable is difficult to delineate. A major problem with this approach is that it can cause inflation of standard errors on estimates and wrong identification of relevant variables in a model. Consequently, erroneous extrapolation can occur.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eFitted snail SDMs.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDescription\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003eDefinition\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eEnsemble of decision trees with randomization.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eClassification:\u003c/p\u003e \u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\text{Ĉ}}_{rf}^{B}\\left(x\\right)=majority\\:vote{\\left\\{{\\text{Ĉ}}_{b}\\right(x\\left)\\right\\}}_{1}^{B}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e \u003cp\u003eWhere: \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\text{Ĉ}}_{rf}^{B}\\left(x\\right)\\)\u003c/span\u003e\u003c/span\u003e is the predicted class for an input \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:x\\)\u003c/span\u003e\u003c/span\u003e based on RF, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\text{Ĉ}}_{b}\\left(x\\right)\\)\u003c/span\u003e\u003c/span\u003e is the predicted class for \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:x\\)\u003c/span\u003e\u003c/span\u003e from the \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:bth\\)\u003c/span\u003e\u003c/span\u003e tree in the forest, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:B\\)\u003c/span\u003e\u003c/span\u003e is the total number of trees in the forest, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:majority\\:vote\\)\u003c/span\u003e\u003c/span\u003e is the class that appears most frequently among the predictions (\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e).\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eFinds optimal separation hyperplane in feature space.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eClassification rule:\u003c/p\u003e \u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:h\\left(x\\right)=\\:\\sum\\:_{i=1}^{n}{\\alpha\\:}_{i}{y}_{i}K\\left({X}_{i},X\\right)+b\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e \u003cp\u003eWhere: \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\alpha\\:}_{i}\\)\u003c/span\u003e\u003c/span\u003e is the Lagrange multiplier found during optimization, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{y}_{i}\\)\u003c/span\u003e\u003c/span\u003e are class labels (+\u0026thinsp;1 or -1), \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:K\\left({X}_{i},X\\right)\\)\u003c/span\u003e\u003c/span\u003e is the Kernel function measuring similarity between data points, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:n\\)\u003c/span\u003e\u003c/span\u003e is the number of support vectors, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:b\\)\u003c/span\u003e\u003c/span\u003e is the bias term (offset) that shifts the decision boundary (\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e).\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMLP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eA type of neural network composed of multiple layers of nodes.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePrediction:\u003c/p\u003e \u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:ŷ=f({W}^{\\left(L\\right)}\\bullet\\:f\\left({W}^{\\left(L-1\\right)}\\bullet\\:\\cdots\\:f\\left({W}^{\\left(1\\right)}\\bullet\\:x+{b}^{\\left(1\\right)}\\right)+{b}^{\\left(L-1\\right)}\\right)+{b}^{\\left(L\\right)})\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e \u003cp\u003eWhere: ŷ is the predicted output of the MLP, in this case class label (presence or absence), \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:f(\\bullet\\:)\\)\u003c/span\u003e\u003c/span\u003e is the activation function applied elementwise at each layer, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:L\\)\u003c/span\u003e\u003c/span\u003e is the total number of layers in the network (excluding the input layer), \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:x\\)\u003c/span\u003e\u003c/span\u003e is the input vector (features of the data), \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{W}^{\\left(i\\right)}\\)\u003c/span\u003e\u003c/span\u003e is the weight matrix for the \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:i\\)\u003c/span\u003e\u003c/span\u003e-th layer, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{b}^{\\left(i\\right)}\\)\u003c/span\u003e\u003c/span\u003e is the bias vector for the \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:i\\)\u003c/span\u003e\u003c/span\u003e-th layer, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\u0026middot;\\)\u003c/span\u003e\u003c/span\u003e is the matrix multiplication between layers (\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e).\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eModel performance evaluation\u003c/p\u003e \u003cp\u003eSeveral performance metrics were applied to assess the classification accuracy and robustness of the models presented in the study. The receiver operating characteristic (ROC)-area under the curve (AUC) was used as the primary metric for measuring a model's discrimination ability. This statistic evaluates the trade-off between sensitivity (the proportion of correctly predicted presences) and specificity (the proportion of correctly predicted absences) (\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e, \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e). An AUC score of 1 represents perfect classification, whereas a score of 0.5 suggests that the model performs no better than random classification. In addition to the ROC-AUC, the true skill statistic (TSS) and correlation coefficient (COR) were also calculated to provide a more comprehensive assessment of model performance. Here, COR is used to evaluate the strength and direction of the relationship between observed values (true outcomes) and predicted values (model predictions) (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e), with a high positive COR (close to 1) indicating strong alignment and good performance, a COR near 0 indicating poor performance, and a negative COR indicating inverse correlation and poor performance. The TSS is a balanced measure of both sensitivity and specificity (\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e). The TSS values range from \u0026minus;\u0026thinsp;1 to 1, with values closer to 1 indicating better model performance. A TSS value of 0 or less indicates that a model's predictive capacity is no better than random. By applying the ROC-AUC, COR and TSS, the present study ensured a robust evaluation of the models\u0026rsquo; ability to accurately predict \u003cem\u003eBi. pfeifferi\u003c/em\u003e and \u003cem\u003eBu. africanus\u003c/em\u003e snail distribution based on environmental variables.\u003c/p\u003e \u003cp\u003eRelative variable importance\u003c/p\u003e \u003cp\u003eThis analysis evaluated the relative importance of each environmental variable included in the ensemble of predictive models. This study implemented the permutation-based variable importance function available in the sdm package (\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e). The purpose of this was to identify the most important variable(s) determining the geographic distribution of \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e in the LSV floodplain. Two complementary evaluation metrics, AUC and COR (Pearson correlation) were used to quantify the contribution of each variable. This dual evaluation allowed for a more comprehensive understanding of each variable\u0026rsquo;s contribution to the model\u0026rsquo;s predictive performance.\u003c/p\u003e"},{"header":"Results and discussion","content":"\u003cp\u003ePredicted current landscape of intestinal and urogenital schistosomiasis risk\u003c/p\u003e \u003cp\u003eUsing an ensemble of RF, SVM and MLP models, the distributions of \u003cem\u003eBi. pfeifferi\u003c/em\u003e and \u003cem\u003eBu. africanus\u003c/em\u003e in LSV was predicted (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The resulting prediction maps reveal marked differences in range and habitat suitability between the snails. The ensemble model for \u003cem\u003eBi. pfeifferi\u003c/em\u003e showed that a substantial area in LSV is an unsuitable habitat for \u003cem\u003eBi.\u003c/em\u003e pfeifferi. Broadly, contemporary environmental conditions appear harsh for the snail. However, at the fine scale, Chikwawa contains pockets of suitable conditions. As a result, a few areas in Chikwawa\u0026mdash;namely, Nchalo, the eastern bank of the Shire River, Chipwaila, and Ngabu\u0026mdash;are identified as suitable habitats for \u003cem\u003eBi. pfeifferi\u003c/em\u003e. Similarly, in Nsanje, greater suitability was predicted in a few areas around Manjolo, Ngabu and bordering Chikwawa South. Thus, in both districts, the majority of areas were characterized by a very low to low probability of presence, including Chileka, Chikwawa Boma, Therere, Thekerani, Muona, Bangula, Tengani, Nsanje Boma, Chimombo, and Marka. Furthermore, moderate suitability - albeit scanty - was nowcast along the Mwanza River near Misomali, Nchalo and Bangula.\u003c/p\u003e \u003cp\u003eTurning now to the ensemble model for \u003cem\u003eBu. africanus\u003c/em\u003e, what stands out in the prediction map is the dominance of suitable areas in most areas of Chikwawa and Nsanje, particularly along the Mwanza River, Shire River and marshland. The map clearly shows that the areas highly suitable for \u003cem\u003eBu. africanus\u003c/em\u003e is present in riverine and wetland areas. This is particularly evident in floodplain areas with inundated vegetation and vegetation cover. For example, Thekerani, Muona, Bangula, Tengani, Nsanje Boma, Chimombo and Marka presented the widest distributions of snails in both districts. The \u003cem\u003eBu. africanus\u003c/em\u003e distribution map also indicates a large ecologically unsuitable habitat in northern Chikwawa (areas surrounding the Majete Wildlife Reserve) and a small unsuitable habitat in western Nsanje. The reports on the prevalence of urogenital schistosomiasis in Chikwawa and Nsanje, combined with malacological surveys, indicate a high prevalence rate and wide species range encompassing most parts of the valley. For example, epidemiological surveys in 2024 revealed a high burden for urogenital schistosomiasis among school-aged children in Chikwawa (prevalence: 35.0%, 95% CI: 33.6\u0026ndash;36.5%), whereas snail surveys revealed a wide distribution and high abundance. On the other hand, the prevalence of intestinal schistosomiasis was low at 1.9% (95% CI: 1.4\u0026ndash;2.3%). The low burden of intestinal schistosomiasis could be attributed to the rarity and restricted range of \u003cem\u003eBi. pfeifferi\u003c/em\u003e, evidently an elusive species, as noted by Poole et al. (\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e). This study detected \u003cem\u003eBi. pfeifferi\u003c/em\u003e at only four out of 70 sites, indicating that it is a rare species with few positive observations (see Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea). The implication of this finding is that while the study area can be considered low risk for intestinal schistosomiasis, the disease remains highly focalized. In general, these results show that the modelled distribution of \u003cem\u003eBu. africanus\u003c/em\u003e group shows large areas of highly suitable habitat in central Chikwawa and eastern Nsanje. That is, both districts have the most suitable habitats for \u003cem\u003eBu. africanus\u003c/em\u003e. In contrast, the predicted distributions of \u003cem\u003eBi\u003c/em\u003e. \u003cem\u003epfeifferi\u003c/em\u003e shows a constrained range and few suitable habitats that can support the snail. Interestingly, the marshlands (Elephant Marsh, Ndindi Marsh in Nsanje District) appear to be completely unsuitable habitats for \u003cem\u003eBi\u003c/em\u003e. \u003cem\u003epfeifferi\u003c/em\u003e. The most suitable habitats for \u003cem\u003eBi. pfeifferi\u003c/em\u003e are sparse, fragmented, and do not overlap with those of \u003cem\u003eBu. africanus\u003c/em\u003e. These findings reinforce the idea that \u003cem\u003eBulinus\u003c/em\u003e host snails are generalist species with higher survival rates and broad tolerances to varying environmental conditions (e.g., (\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e),(\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e) than \u003cem\u003eBiomphalaria\u003c/em\u003e sp. in the bioclimatic areas of Africa.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003e Predicted future landscape of intestinal and urogenital schistosomiasis risk\u003c/h3\u003e\n\u003cp\u003eThe results obtained from the forecasting of the distribution of \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e in the LSV under the worst-climate change scenario are presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea and b. For \u003cem\u003eBi. pfeifferi\u003c/em\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea shows a noticeable expansion of suitable habitats (brown areas: 298 km\u003csup\u003e2\u003c/sup\u003e) in eastern Chikwawa and northern Nsanje, whereas considerable areas across Chikwawa and western Nsanje will experience habitat loss (green zones: 791 km\u003csup\u003e2\u003c/sup\u003e). Conversely, \u003cem\u003eBu. africanus\u003c/em\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb) reveals a significant widespread pattern of habitat gain (brown zones: 3120 km\u003csup\u003e2\u003c/sup\u003e) across the entire LSV, with only minor suitability loss (green zones: 154 km\u003csup\u003e2\u003c/sup\u003e), mainly along the Mwanza and Shire riverine areas in Chikwawa. A comparison of Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea and b highlights the differential impacts of climate change on the two intermediate host species of schistosomiasis. The more pronounced habitat gains for \u003cem\u003eBu. africanus\u003c/em\u003e suggest that this species may benefit from the warmer and wetter conditions predicted under the SSP585 scenario. This expansion could heighten the risk of urogenital schistosomiasis transmission in areas previously unaffected, creating new public health challenges in LSV. In contrast, the prominent habitat loss and marginal habitat gains predicted for \u003cem\u003eBi. pfeifferi\u003c/em\u003e suggests a potential retraction and shift in the risk areas for intestinal schistosomiasis. However, the persistence of suitable habitats in central Chikwawa (Nchalo Sugar Estate), the only known site with abundant \u003cem\u003eBi. pfeifferi\u003c/em\u003e) suggests that the area may continue to serve as a transmission hotspot.\u003c/p\u003e \u003cp\u003eThese findings highlight the complex interplay between climate change and schistosomiasis ecology, emphasizing the need for tailored control measures for both intestinal and urogenital schistosomiasis. For \u003cem\u003eBu. africanus\u003c/em\u003e, environmental monitoring in areas predicted to gain suitability is essential, coupled with epidemiological surveys to prevent disease spread. For \u003cem\u003eBi. pfeifferi\u003c/em\u003e, the focus should be on sustaining control efforts in the core habitats that are projected to remain suitable, particularly in the agricultural irrigated land of Nchalo Sugar Estate.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eMulticollinearity\u003c/p\u003e \u003cp\u003eAmong the 24 input variables, 15 variables were identified with collinearity problems, namely, bio2, bio4, bio5, bio8, bio9, bio10, bio11, bio12, bio16, bio17, bio18, bio19, bio14 and elevation. This indicates that these variables are highly correlated with each other, which can cause inflated standard errors and less reliable coefficient estimates in the prediction model. The variables that remained are bio1, bio3, bio6, bio13, bio15, slope, soil, distance to water and the NDVI (Additional file: Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The remaining variables showed acceptable levels of multicollinearity, as indicated by their VIF values, all of which are below the threshold of 10 (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). This indicated a well-balanced set of variables with negligible multicollinearity. The minimum linear correlation coefficient was 0.004 between bio15 and bio6. This indicates a very weak negative correlation between these two variables. The maximum linear correlation coefficient was 0.822 between bio15 and bio13, indicating a strong positive correlation between these two variables.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eVIFs of the remaining variables.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eVIF\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ebio 1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.67\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ebio 3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5.09\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ebio 6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.43\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ebio 13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9.19\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ebio 15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5.27\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSlope\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.54\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSoil\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.18\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDistance to water body\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.61\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNDVI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.79\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eRelative variable importance\u003c/p\u003e\u003cp\u003eThe variable importance plot for \u003cem\u003eBu. africanus\u003c/em\u003e ensemble (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea) shows that the most important variable in the model is the NDVI (COR\u0026thinsp;=\u0026thinsp;0.59, AUC\u0026thinsp;=\u0026thinsp;0.37). The next two important variables are the mean annual temperature (bio1, COR\u0026thinsp;=\u0026thinsp;0.13, AUC\u0026thinsp;=\u0026thinsp;0.18) and slope (COR\u0026thinsp;=\u0026thinsp;0.97, AUC\u0026thinsp;=\u0026thinsp;0.11). The other variables had lower importance: distance to the water body (COR\u0026thinsp;=\u0026thinsp;0.04, AUC\u0026thinsp;=\u0026thinsp;0.03), soil type (COR\u0026thinsp;=\u0026thinsp;0.03, AUC\u0026thinsp;=\u0026thinsp;0.05), precipitation seasonality (bio15, COR\u0026thinsp;=\u0026thinsp;0.05, AUC\u0026thinsp;=\u0026thinsp;0.03), precipitation in the wettest month (bio13, COR\u0026thinsp;=\u0026thinsp;0.02, AUC\u0026thinsp;=\u0026thinsp;0.05), minimum temperature in the coldest month (bio6, COR\u0026thinsp;=\u0026thinsp;0.03, AUC\u0026thinsp;=\u0026thinsp;0.05) and isothermality (bio3, COR\u0026thinsp;=\u0026thinsp;0.02, AUC\u0026thinsp;=\u0026thinsp;0.03).\u003c/p\u003e \u003cp\u003eTurning now to the variable importance plot for \u003cem\u003eBi. pfeifferi\u003c/em\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb), precipitation seasonality (COR\u0026thinsp;=\u0026thinsp;0.32, AUC\u0026thinsp;=\u0026thinsp;0.18) was the most important variable, followed by precipitation in the wettest month (COR\u0026thinsp;=\u0026thinsp;0.17, AUC\u0026thinsp;=\u0026thinsp;0.15). Isothermality showed moderate importance (COR\u0026thinsp;=\u0026thinsp;0.16, AUC\u0026thinsp;=\u0026thinsp;0.07). The least important variables in order of magnitude from relatively low importance to very low importance are the minimum temperature of the coldest month (COR\u0026thinsp;=\u0026thinsp;0.07, AUC\u0026thinsp;=\u0026thinsp;0.03), the distance to the waterbody (COR\u0026thinsp;=\u0026thinsp;0.11, AUC\u0026thinsp;=\u0026thinsp;0.04), the mean annual temperature (COR\u0026thinsp;=\u0026thinsp;0.05, AUC\u0026thinsp;=\u0026thinsp;0.04), the soil type (COR\u0026thinsp;=\u0026thinsp;0.05, AUC\u0026thinsp;=\u0026thinsp;0.02), the NDVI (COR\u0026thinsp;=\u0026thinsp;0.03, AUC\u0026thinsp;=\u0026thinsp;0.02), and the slope (COR\u0026thinsp;=\u0026thinsp;0.02, AUC\u0026thinsp;=\u0026thinsp;0.01). The results highlight the importance of climatic factors over topographic and vegetation factors in determining \u003cem\u003eBi. pfeifferi\u003c/em\u003e distribution. Although broadly consistent with Monde et al. (\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e), this outcome is contrary to that of Ponpetch et al. (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e), who reported that soil properties and elevation are key factors in the distribution of \u003cem\u003eBi. pfeifferi\u003c/em\u003e in Ethiopia. Notably, this study has methodological limitations due to the difficulty in obtaining spatial coverage for proximal environmental predictors, such as physicochemical water parameters (e.g., pH and conductivity). Here, ecological processes (e.g., predators, competitors, and dispersal barriers such as rapids and mountain ranges), anthropogenic changes (e.g., agricultural expansion) and disturbances (e.g., floods and droughts) were not included in the SDM. Therefore, this study acknowledges the potential influence of other critical environmental predictors that remain unidentified and excluded. For example, the LSV is an agro-landscape currently undergoing large-scale agricultural expansion driven by the Shire Valley Transformation Project. In this geographical context, the expansion of cropland has been shown to significantly alter natural habitats, leading to habitat loss and fragmentation (\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e). The loss of habitat and biodiversity can subsequently trigger cascading effects within ecological communities, impacting species abundance, distribution, and interactions (\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e). Unlike environmental conditions, biotic processes are rarely integrated into species distribution models, underscoring the need for further attention (\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e). Here, a far-reaching consequence is that relying on only available data (often referred to as \"weak\" data) instead of more comprehensive data (\"hard\" data) can result in skewed or inaccurate predictions, potentially leading to an incomplete understanding of snail habitat suitability.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eModel performance and evaluation\u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e provides a performance summary of the predictive models used to predict the distributions of \u003cem\u003eBi. pfeifferi\u003c/em\u003e and \u003cem\u003eBu. africanus\u003c/em\u003e in the LSV. The AUC, COR and TSS values underscore the model\u0026rsquo;s capacity to delineate suitable snail habitats from unsuitable areas. For \u003cem\u003eBu. africanus\u003c/em\u003e, the SVM model achieved the highest performance (AUC\u0026thinsp;=\u0026thinsp;0.74, COR\u0026thinsp;=\u0026thinsp;0.51, TSS\u0026thinsp;=\u0026thinsp;0.56, deviance\u0026thinsp;=\u0026thinsp;0.85), followed by RF (AUC\u0026thinsp;=\u0026thinsp;0.73, COR\u0026thinsp;=\u0026thinsp;0.33, TSS\u0026thinsp;=\u0026thinsp;0.57, deviance\u0026thinsp;=\u0026thinsp;0.97), both of which demonstrated high prediction accuracy. In contrast, the MLP showed weak performance (AUC\u0026thinsp;=\u0026thinsp;0.56, COR\u0026thinsp;=\u0026thinsp;0.07, and TSS\u0026thinsp;=\u0026thinsp;0.44) and the highest deviance (2.04), indicating high prediction error. This suggests that the MLP had poor classification performance for \u003cem\u003eBu. africanus\u003c/em\u003e, a species that is widely distributed and abundant in the LSV.\u003c/p\u003e \u003cp\u003eFor \u003cem\u003eBi. pfeifferi\u003c/em\u003e model, MLP (AUC\u0026thinsp;=\u0026thinsp;0.83, COR\u0026thinsp;=\u0026thinsp;0.42, TSS\u0026thinsp;=\u0026thinsp;0.75, deviance\u0026thinsp;=\u0026thinsp;0.81), outperformed RF (AUC\u0026thinsp;=\u0026thinsp;0.77, COR\u0026thinsp;=\u0026thinsp;0.34, TSS\u0026thinsp;=\u0026thinsp;0.69, deviance\u0026thinsp;=\u0026thinsp;0.77) and SVM (AUC\u0026thinsp;=\u0026thinsp;0.73, COR\u0026thinsp;=\u0026thinsp;0.38, TSS\u0026thinsp;=\u0026thinsp;0.64, deviance\u0026thinsp;=\u0026thinsp;0.78). The MLP model struggled to predict \u003cem\u003eBu. africanus\u003c/em\u003e distribution; here, the model performed reasonably well, showing very good predictive accuracy and model fit for \u003cem\u003eBi. pfeifferi\u003c/em\u003e, a species that is rare and patchy in the LSV. This highlights significant variability in performance for the MLP and more stable performance and consistent results for the RF and SVM. Overall, the RF, SVM and MLP models provided acceptable results, with SVM being the top performer across metrics.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel performance of \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e distribution models across ML model types on the test dataset generated using partition.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpecies\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eModel type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCOR\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTSS\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eDeviance\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eBu. africanus\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.97\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.51\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.85\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMLP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.44\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2.04\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eBi. pfeifferi\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.34\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.69\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.77\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.64\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMLP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.83\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThree conclusions can be drawn from this study. First, the habitat suitability map for \u003cem\u003eBu. africanus\u003c/em\u003e and \u003cem\u003eBi. pfeifferi\u003c/em\u003e in LSV reveals that the distribution of suitable habitats for urogenital and intestinal schistosomiasis transmission is not uniform. \u003cem\u003eBu. africanus\u003c/em\u003e habitat is the most dominant and abundant, indicating a widespread and high risk of urogenital schistosomiasis across the valley. In contrast, environmental conditions, in climatic terms, appear harsh for \u003cem\u003eBi. pfeifferi\u003c/em\u003e. Nonetheless, while \u003cem\u003eBi. pfeifferi\u003c/em\u003e habitat is patchy and scarce, indicating a lower risk of intestinal transmission, there is still reason to suggest that the valley has a nonnegligible risk profile. Second, in LSV, vegetation cover is the most important predictor of \u003cem\u003eBu. africanus\u003c/em\u003e distribution, whereas precipitation variables are most important for \u003cem\u003eBi. pfeifferi\u003c/em\u003e. Third, under the SSP585 scenario, \u003cem\u003eBu. africanus\u003c/em\u003e habitats are projected to expand, increasing the risk of urogenital schistosomiasis in new areas. On the other hand, \u003cem\u003eBi. pfeifferi\u003c/em\u003e habitats show moderate gains with a limited shift, indicating the potential persistence of intestinal schistosomiasis in Chikwawa and a low risk in Nsanje. This spatial understanding is crucial for targeted snail control and mass drug administration with praziquantel.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eSDM\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eSpecies distribution model\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eLSV\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eLower Shire Valley\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ekm\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eKilometre\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eNDVI\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eNormalized difference vegetation index\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eSSP585\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eShared Socioeconomic Pathway 585\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eML\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eMachine learning\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eRF\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eRandom forest\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eSVM\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eSupport vector machine\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMLP\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eMultilayer perceptron\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eVIF\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eVariance inflation factor\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eROC\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eReceiver Operating Characteristics\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eAUC\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eArea Under the Curve\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eTSS\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eTrue Skill Statistic\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eCOR\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eCorrelation\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eApproval for this study was received from the College of Medicine Research Ethics Committee (COMREC) (Protocol number: P.02/23/3989), Chikwawa and Nsanje District Health Office research committees and the ILLOVO Nchalo Estate.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCN:\u0026nbsp;\u003c/strong\u003eConceptualization of this study, funding acquisition, investigation (lead), analysis, project administration, visualization, method, writing-original draft preparation, review and editing. \u003cstrong\u003eJC:\u0026nbsp;\u003c/strong\u003eSupervision, writing \u0026ndash; review and editing. \u003cstrong\u003eCMJ:\u0026nbsp;\u003c/strong\u003eSupervision, writing-review and editing. \u003cstrong\u003eEAK:\u0026nbsp;\u003c/strong\u003eInvestigation, data collection and review. \u003cstrong\u003eJAT:\u003c/strong\u003e Supervision, writing, review and editing.\u003cstrong\u003e\u0026nbsp;JRS:\u0026nbsp;\u003c/strong\u003eSupervision, investigation, data collection, resources, writing-review and editing. All the authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used and/or analysed during the current study are available from the corresponding author upon reasonable request. Historical and future climate and elevation data are provided by WorldClim and are available at https://www.worldclim.org/data/bioclim.html.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the National Institute for Health Research (NIHR) (using the UK\u0026rsquo;s Official Development Assistance (ODA) Funding) and Wellcome [223660/Z/21/Z] under the NIHR-Wellcome Partnership for Global Health Research. JAT is funded by the MRC Skills Development Fellowship [MR/T031743/1]. This UK-funded award is part of the EDCTP2 programme supported by the European Union. The views expressed are those of the authors and not necessarily those of Wellcome, the NIHR or the Department of Health and Social Care.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe would like to thank the communities of Chikwawa and Nsanje for welcoming us and allowing this study. CN would like to thank Chimwemwe Jamali, Peter Makaula, Gladys Namacha, Frank Mbalume, Chisomo Lwanda, Dalitso Damiano, Joel Kanyanga and Calisto Moda for their assistance during data collection.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eWHO. Schistosomiasis (bilharzia) [Internet]. Geneva: World Health Organization; 2024 [cited 2024 Oct 25]. Available from: https://www.who.int/health-topics/schistosomiasis#tab=tab_1\u003c/li\u003e\n\u003cli\u003eChiepa B, Mbewe R, Stanton MC, Kapumba B, Kambewa E, Kaunga L, et al. A baseline epidemiological survey for malaria and schistosomiasis reveals an alarming burden in primary schools despite ongoing control in Chikwawa District, southern Malawi. Curr Res Parasitol Vector-Borne Dis. 2024;5:100183.\u003c/li\u003e\n\u003cli\u003eNelwan ML. Schistosomiasis: Life Cycle, Diagnosis, and Control. Curr Ther Res. 2019;91:5\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003eGryseels B, Polman K, Clerinx J, Kestens L. Human schistosomiasis. The Lancet. 2006 Sep;368(9541):1106\u0026ndash;18.\u003c/li\u003e\n\u003cli\u003eCetron MS, Chitsulo L, Sullivan JJ, Pilcher J, Wilson M, Noh J, et al. Schistosomiasis in Lake Malawi. The Lancet. 1996 Nov;348(9037):1274\u0026ndash;8.\u003c/li\u003e\n\u003cli\u003eZheng L, Deng L, Zhong Y, Wang Y, Guo W, Fan X. Molluscicides against the snail-intermediate host of Schistosoma: a review. Parasitol Res. 2021 Oct;120(10):3355\u0026ndash;93.\u003c/li\u003e\n\u003cli\u003eAlharbi MH, Condemine C, Christiansen R, LaCourse EJ, Makaula P, Stanton MC, et al. \u003cem\u003eBiomphalaria pfeifferi\u003c/em\u003e Snails and Intestinal Schistosomiasis, Lake Malawi, Africa, 2017\u0026ndash;2018. Emerg Infect Dis. 2019 Mar;25(3):613\u0026ndash;5.\u003c/li\u003e\n\u003cli\u003eH.N Mtethiwa A. Prevalence and Intensity of Schistosomiasis in Communities around Water Reservoirs in Malawi. J Trop Dis [Internet]. 2016 [cited 2024 Nov 14];04(01). Available from: http://www.esciencecentral.org/journals/prevalence-and-intensity-of-schistosomiasis-in-communities-around-water-reservoirs-in-malawi-2329-891X-1000183.php?aid=65296\u003c/li\u003e\n\u003cli\u003eMakaula P, Sadalaki JR, Muula AS, Kayuni S, Jemu S, Bloch P. Schistosomiasis in Malawi: a systematic review. Parasit Vectors. 2014 Dec;7(1):570.\u003c/li\u003e\n\u003cli\u003eReed AL, Al-Harbi MH, Makaula P, Condemine C, Hesketh J, Archer J, et al. A geospatial analysis of local intermediate snail host distributions provides insight into schistosomiasis risk within undersampled areas of southern Lake Malawi. Parasit Vectors. 2024 Jun 27;17(1):272.\u003c/li\u003e\n\u003cli\u003eWiselka M, Robinson M, Clipsham K, Weddon S. The epidemiology of schistosomiasis in central Malawi. 1988;\u003c/li\u003e\n\u003cli\u003eSouza CPD. Molluscicide control of snail vectors of schistosomiasis. Mem Inst Oswaldo Cruz. 1995 Apr;90(2):165\u0026ndash;8.\u003c/li\u003e\n\u003cli\u003eWHO. Field use of molluscicides in schistosomiasis control programmes: an operational manual for programme managers. World Health Organization, Geneva; 2017.\u003c/li\u003e\n\u003cli\u003eWHO. Diagnostic target product profiles for monitoring, evaluation and surveillance of schistosomiasis control programmes. World Health Organization, Geneva; 2021.\u003c/li\u003e\n\u003cli\u003eMinistry of Health, Republic of Malawi. Malawi Neglected Tropical Diseases Master 2015 to 2020; Master Plan for Malawi. Ministry of Health, Lilongwe; 2014.\u003c/li\u003e\n\u003cli\u003eBoitt MK, Suleiman MK. Mapping of Freshwater Snails\u0026rsquo; Habitat\u0026mdash;A Source of Transmitting Bilharzia in Mwea Sub-County, Kenya. J Geosci Environ Prot. 2021;09(10):130\u0026ndash;50.\u003c/li\u003e\n\u003cli\u003eManyangadze T, Chimbari MJ, Gebreslasie M, Ceccato P, Mukaratirwa S. Modelling the spatial and seasonal distribution of suitable habitats of schistosomiasis intermediate host snails using Maxent in Ndumo area, KwaZulu-Natal Province, South Africa. Parasit Vectors. 2016 Dec;9(1):572.\u003c/li\u003e\n\u003cli\u003eFAO. Malawi Country Profile: Physiography, climate and soils [Internet]. Food and Agriculture Organization; 2002 [cited 2024 Feb 16]. Available from: https://www.fao.org/3/ac468e/ac468e03.htm#bm03\u003c/li\u003e\n\u003cli\u003eThieme M, Abell R, Stiassny M, Skelton P. Freshwater Ecoregions of Africa and Madagascar: A Conservation Assessment. Bibliovault OAI Repos Univ Chic Press. 2005 Jan;1\u0026ndash;431.\u003c/li\u003e\n\u003cli\u003eNational Statistical Office. Malawi 2018 Population and Housing Census Main Report [Internet]. Malawi Government; 2019. Available from: https://malawi.unfpa.org/en/publications/malawi-2018-population-and-housing-census-main-report\u003c/li\u003e\n\u003cli\u003eMinistry of Health, Republic of Malawi. Malawi Neglected Tropical Diseases Master Plan 2023 - 2030 [Internet]. Available from: https://espen.afro.who.int/system/files/content/resources/Malawi%20NTD%20Master%20Plan%202023-2030.pdf\u003c/li\u003e\n\u003cli\u003eBriers RA. Range limits and parasite prevalence in a freshwater snail. Proc R Soc Lond B Biol Sci [Internet]. 2003 Nov 7 [cited 2024 Oct 27];270(suppl_2). Available from: https://royalsocietypublishing.org/doi/10.1098/rsbl.2003.0046\u003c/li\u003e\n\u003cli\u003eKrauth SJ, Wandel N, Traor\u0026eacute; SI, Vounatsou P, Hattendorf J, Achi LY, et al. Distribution of intermediate host snails of schistosomiasis and fascioliasis in relation to environmental factors during the dry season in the Tchologo region, C\u0026ocirc;te d\u0026rsquo;Ivoire. Adv Water Resour. 2017 Oct;108:386\u0026ndash;96.\u003c/li\u003e\n\u003cli\u003eMagero VO, Kisara S, Wade CM. Geographical Distribution of \u003cem\u003eBiomphalaria\u003c/em\u003e snails in East Africa [Internet]. 2021 [cited 2024 Oct 26]. Available from: http://biorxiv.org/lookup/doi/10.1101/2021.11.04.467236\u003c/li\u003e\n\u003cli\u003eGlidden CK, Singleton AL, Chamberlin A, Tuan R, Palasio RGS, Caldeira RL, et al. Climate and urbanization drive changes in the habitat suitability of Schistosoma mansoni competent snails in Brazil. Nat Commun. 2024 Jun 19;15(1):4838.\u003c/li\u003e\n\u003cli\u003eMin F, Wang J, Liu X, Yuan Y, Guo Y, Zhu K, et al. Environmental Factors Affecting Freshwater Snail Intermediate Hosts in Shenzhen and Adjacent Region, South China. Trop Med Infect Dis. 2022 Dec 9;7(12):426.\u003c/li\u003e\n\u003cli\u003eWalz Y, Wegmann M, Dech S, Raso G, Utzinger J. Risk profiling of schistosomiasis using remote sensing: approaches, challenges and outlook. Parasit Vectors. 2015 Dec;8(1):163.\u003c/li\u003e\n\u003cli\u003eStanton MC, Adriko M, Arinaitwe M, Howell A, Davies J, Allison G, et al. Intestinal schistosomiasis in Uganda at high altitude (\u0026gt;1400 m): malacological and epidemiological surveys on Mount Elgon and in Fort Portal crater lakes reveal extra preventive chemotherapy needs. Infect Dis Poverty. 2017 Dec;6(1):34.\u003c/li\u003e\n\u003cli\u003eStensgaard AS, Utzinger J, Vounatsou P, H\u0026uuml;rlimann E, Schur N, Saarnak CFL, et al. Large-scale determinants of intestinal schistosomiasis and intermediate host snail distribution across Africa: does climate matter? Acta Trop. 2013 Nov;128(2):378\u0026ndash;90.\u003c/li\u003e\n\u003cli\u003eMalone JB, Yilma JM, McCarroll JC, Erko B, Mukaratirwa S, Zhou X. Satellite climatology and the environmental risk of Schistosoma mansoni in Ethiopia and east Africa. Acta Trop. 2001 Apr;79(1):59\u0026ndash;72.\u003c/li\u003e\n\u003cli\u003eOlkeba BK, Boets P, Mereta ST, Yeshigeta M, Akessa GM, Ambelu A, et al. Environmental and biotic factors affecting freshwater snail intermediate hosts in the Ethiopian Rift Valley region. Parasit Vectors. 2020 Dec;13(1):292.\u003c/li\u003e\n\u003cli\u003eShi Y, Qiu J, Li R, Shen Q, Huang D. Identification of Potential High-Risk Habitats within the Transmission Reach of Oncomelania hupensis after Floods Based on SAR Techniques in a Plane Region in China. Int J Environ Res Public Health. 2017 Aug 30;14(9):986.\u003c/li\u003e\n\u003cli\u003eSimoonga C, Utzinger J, Brooker S, Vounatsou P, Appleton CC, Stensgaard AS, et al. Remote sensing, geographical information system and spatial analysis for schistosomiasis epidemiology and ecology in Africa. Parasitology. 2009 Nov;136(13):1683\u0026ndash;93.\u003c/li\u003e\n\u003cli\u003eVeerappan R, Sayed S. Urban flood susceptibility zonation mapping using evidential belief function, frequency ratio and fuzzy gamma operator models in GIS: a case study of Greater Mumbai, Maharashtra, India. Geocarto Int. 2020 Mar 4;\u003c/li\u003e\n\u003cli\u003eWu S, Yu M, Chen L. Nonmonotonic and spatial‐temporal dynamic slope effects on soil erosion during rainfall‐runoff processes. Water Resour Res. 2017 Feb;53(2):1369\u0026ndash;89.\u003c/li\u003e\n\u003cli\u003eEzeugwu SMC, Obiamiwe BA. Application of biostatistics to the effects of water velocity on the distribution of Bulinus globosus and Bulinus forskali in an irrigation scheme in Borno State, Nigeria. Afr Sci. 2005;6(3).\u003c/li\u003e\n\u003cli\u003eNtonifor H N, Ajayi JA. Studies on the ecology and distribution of some medically important freshwater snail species in Bauchi State, Nigeria. 2007. 2007;1(2):121\u0026ndash;7.\u003c/li\u003e\n\u003cli\u003eCoelho PRS, Ker FTO, Ara\u0026uacute;jo AD, Guimar\u0026atilde;es RicardoJPS, Negr\u0026atilde;o-Corr\u0026ecirc;a DA, Caldeira RL, et al. Identification of Risk Areas for Intestinal Schistosomiasis, Based on Malacological and Environmental Data and on Reported Human Cases. Front Med. 2021 Aug 6;8:642348.\u003c/li\u003e\n\u003cli\u003eHuang Q ni, Tang L li, Jiang X guang, Chen Z, Zhou X nong. [Retrieving eco-environment factors relevant to Oncomelania snail distribution based on QuickBird image]. Zhongguo Ji Sheng Chong Xue Yu Ji Sheng Chong Bing Za Zhi. 2007 Aug;25(4):304\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003ePonpetch K, Erko B, Bekana T, Kebede T, Tian D, Yang Y, et al. Environmental Drivers and Potential Distribution of Schistosoma mansoni Endemic Areas in Ethiopia. Microorganisms. 2021 Oct 14;9(10):2144.\u003c/li\u003e\n\u003cli\u003eRodrigues PMS, Schaefer CEGR, Silva JDO, Ferreira J\u0026uacute;nior WG, Dos Santos RM, Neri AV. The influence of soil on vegetation structure and plant diversity in different tropical savannic and forest habitats. J Plant Ecol. 2016 Dec 26;rtw135.\u003c/li\u003e\n\u003cli\u003eRiahi K, Van Vuuren DP, Kriegler E, Edmonds J, O\u0026rsquo;Neill BC, Fujimori S, et al. The Shared Socioeconomic Pathways and their energy, land use, and greenhouse gas emissions implications: An overview. Glob Environ Change. 2017 Jan;42:153\u0026ndash;68.\u003c/li\u003e\n\u003cli\u003eDesautels DJ, Hartman RB, Weber ME, Jacob N, Sun A, Civitello DJ. Experimental water hyacinth invasion and destructive management increase human schistosome transmission potential. Ecol Appl. 2023 Mar;33(2):e2767.\u003c/li\u003e\n\u003cli\u003eBreiman L. Random Forests. Mach Learn. 2001;45(1):5\u0026ndash;32.\u003c/li\u003e\n\u003cli\u003eVapnik VN. The Nature of Statistical Learning Theory [Internet]. New York, NY: Springer New York; 1995 [cited 2024 Dec 31]. Available from: http://link.springer.com/10.1007/978-1-4757-2440-0\u003c/li\u003e\n\u003cli\u003eRumelhart DE, Hinton GE, Williams RJ. Learning representations by back-propagating errors. Nature. 1986 Oct;323(6088):533\u0026ndash;6.\u003c/li\u003e\n\u003cli\u003eNaimi B, Ara\u0026uacute;jo MB. sdm: a reproducible and extensible R platform for species distribution modelling. Ecography. 2016 Apr;39(4):368\u0026ndash;75.\u003c/li\u003e\n\u003cli\u003eDormann CF, Elith J, Bacher S, Buchmann C, Carl G, Carr\u0026eacute; G, et al. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography. 2013 Jan;36(1):27\u0026ndash;46.\u003c/li\u003e\n\u003cli\u003eAllouche O, Tsoar A, Kadmon R. Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). J Appl Ecol. 2006 Dec;43(6):1223\u0026ndash;32.\u003c/li\u003e\n\u003cli\u003ePoole H, Terlouw DJ, Naunje A, Mzembe K, Stanton M, Betson M, et al. Schistosomiasis in preschool-age children and their mothers in Chikhwawa district, Malawi with notes on characterization of schistosomes and snails. Parasit Vectors. 2014 Dec;7(1):153.\u003c/li\u003e\n\u003cli\u003eDiakit\u0026eacute; NR, Koffi PB, Konan CK, Bassa FK, Chamberlin AJ, Ouattara M, et al. Variability of biological traits of Bulinus truncatus and Biomphalaria pfeifferi, the intermediate host snails of schistosomiasis, from three climatic zones of C\u0026ocirc;te d\u0026rsquo;Ivoire. Front Environ Sci. 2023 Dec 7;11:1193239.\u003c/li\u003e\n\u003cli\u003eVan Der Deure T, Maes T, Huyse T, Stensgaard A. Climate change could fuel urinary schistosomiasis transmission in Africa and Europe. Glob Change Biol. 2024 Aug;30(8):e17434.\u003c/li\u003e\n\u003cli\u003eMonde C, Syampungani S, Van Den Brink PJ. Natural and human induced factors influencing the abundance of Schistosoma host snails in Zambia. Environ Monit Assess. 2016 Jun;188(6):370.\u003c/li\u003e\n\u003cli\u003eNkolokosa C, Stothard R, Jones CM, Stanton M, Chirombo J, Tangena JAA. Monitoring and simulating landscape changes: how do long-term changes in land use and long-term average climate affect regional biophysical conditions in southern Malawi? Environ Monit Assess. 2023 Oct;195(10):1247.\u003c/li\u003e\n\u003cli\u003eSenadeera M. Effects of land use on species diversity and ecosystem functioning in agricultural landscapes. 2023 Apr 24;\u003c/li\u003e\n\u003cli\u003eAustin MP. Spatial prediction of species distribution: an interface between ecological theory and statistical modelling. Ecol Model. 2002 Nov;157(2\u0026ndash;3):101\u0026ndash;18.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"parasites-and-vectors","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"parv","sideBox":"Learn more about [Parasites \u0026 Vectors](http://parasitesandvectors.biomedcentral.com/)","snPcode":"13071","submissionUrl":"https://submission.nature.com/new-submission/13071/3","title":"Parasites \u0026 Vectors","twitterHandle":"@bugbittentweets","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Habitat suitability, species distribution modelling, schistosomiasis, Lower Shire Valley, ensemble machine learning, climate change","lastPublishedDoi":"10.21203/rs.3.rs-5908499/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5908499/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis paper presents the first species distribution models (SDMs) for intermediate snail hosts for urogenital and intestinal schistosomiasis in the Lower Shire Valley (LSV), Malawi. The SDMs are specific to the \u003cem\u003eBulinus africanus\u003c/em\u003e group and \u003cem\u003eBiomphalaria pfeifferi\u003c/em\u003e. The former transmits urogenital schistosomiasis (\u003cem\u003eSchistosoma haematobium\u003c/em\u003e), and the latter transmits intestinal schistosomiasis (\u003cem\u003eSchistosoma mansoni\u003c/em\u003e), both of which affect nearly 240\u0026nbsp;million people globally. This study addresses the following questions: 1. Where are the most suitable habitats for intermediate host snails in the LSV? 2. Which environmental factors influence the geographical distribution of such snails in the LSV? 3. How will climate change shape future schistosomiasis transmission risk? Consistent with expectations, the SDMs reveal the following: 1) currently, \u003cem\u003eBu. africanus\u003c/em\u003e not only has a wide distribution across central Chikwawa and eastern Nsanje but is also concentrated in floodplains, and the LSV has few habitats that can support \u003cem\u003eBi. pfeifferi\u003c/em\u003e, 2) vegetation cover is the most important predictor of \u003cem\u003eBu. africanus\u003c/em\u003e distribution, whereas precipitation variables are most important for \u003cem\u003eBi. pfeifferi\u003c/em\u003e in the LSV, and 3) future projections indicate a moderate increase (4.4%) and east-ward shift in \u003cem\u003eBi. pfeifferi\u003c/em\u003e distribution, with patchy spatial coverage, and a significant expansion (46%) of suitable habitats for \u003cem\u003eBu. africanus\u003c/em\u003e in LSV. Understanding the spatial and temporal distributions of these snails is important for controlling and eliminating schistosomiasis.\u003c/p\u003e","manuscriptTitle":"Predicting current and future suitability for intermediate snail hosts of urogenital and intestinal schistosomiasis in a floodplain of Malawi","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-01-31 14:04:38","doi":"10.21203/rs.3.rs-5908499/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-02-25T15:22:51+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-02-22T14:33:04+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"246939426462940586790627671631816481070","date":"2025-02-02T10:06:48+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-01-31T09:41:20+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-01-29T13:17:43+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-01-29T12:57:25+00:00","index":"","fulltext":""},{"type":"submitted","content":"Parasites \u0026 Vectors","date":"2025-01-26T22:47:29+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"parasites-and-vectors","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"parv","sideBox":"Learn more about [Parasites \u0026 Vectors](http://parasitesandvectors.biomedcentral.com/)","snPcode":"13071","submissionUrl":"https://submission.nature.com/new-submission/13071/3","title":"Parasites \u0026 Vectors","twitterHandle":"@bugbittentweets","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"c33f8cc4-8401-4191-833d-11311a89fa3a","owner":[],"postedDate":"January 31st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-09-01T16:01:43+00:00","versionOfRecord":{"articleIdentity":"rs-5908499","link":"https://doi.org/10.1186/s13071-025-06952-3","journal":{"identity":"parasites-and-vectors","isVorOnly":false,"title":"Parasites \u0026 Vectors"},"publishedOn":"2025-08-29 15:57:47","publishedOnDateReadable":"August 29th, 2025"},"versionCreatedAt":"2025-01-31 14:04:38","video":"","vorDoi":"10.1186/s13071-025-06952-3","vorDoiUrl":"https://doi.org/10.1186/s13071-025-06952-3","workflowStages":[]},"version":"v1","identity":"rs-5908499","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5908499","identity":"rs-5908499","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}