INTRODUCTION
Urinary schistosomiasis, is endemic in much of subSaharan Africa where suitable habitats for snails of the genus Bulinus, the intermediate hosts of Schistosoma haematobium are abundant (Jordon and Webbe, Reference Jordon, Webbe, Jordon, Webbe and Sturrock1993; Brown, Reference Brown1994). Along coastal Kenya, Bulinus globosus and Bulinus nasutus species are the predominant intermediate hosts of S. haematobium. In Msambweni, where B. nasutus is the only intermediate host, 52% of school-aged children are infected with S. haematobium (Clennon et al. unpublished obsevations), despite chemotherapy programmes targeting school children during the 1980s and 1990s (King et al. Reference King, Muchiri, Ouma and Koech1991; Muchiri et al. Reference Muchiri, Ouma and King1996; Satayathum et al. Reference Satayathum, Muchiri, Ouma, Whalen and King2006). Concurrent control measures directed at the intermediate host in the region may be necessary for sustained suppression of human infection.
Bulinids are aquatic snails found in streams, ponds and temporary water pools (O'Keefe, Reference O'Keefe1985; Kariuki et al. Reference Kariuki, Clennon, Brady, Kitron, Sturrock, Ouma, Ndzovu, Mungai, Hoffman, Hamburger, Pellegrini, Muchiri and King2004). Control of these snail populations is difficult because they can auto-reproduce and aestivate (McCullough et al. Reference McCullough, Eyakuze, Msinde and Nditi1968). Aestivation, when snails may be found under vegetation or burrowed into the soil in a condition of reduced metabolic activity, is triggered by the drying of water pools, (Brown, Reference Brown1994). When rainwater fills the pond beds and creates water pools, the snails emerge from aestivation and repopulate the water bodies (Webbe and Msangi, Reference Webbe and Msangi1958; Donnelly et al. Reference Donnelly, Appleton and Schutte1983; Sturrock et al. Reference Sturrock, Kinyanjui, Thiongo, Tosha, Ouma, King, Koech, Siongok and Mahmoud1990). Live snails are limited to locations with standing water or with enough moisture for survival of aestivating snails, while shells can also be found along dry sections of the drainage corridors that connect such locations.
Infected B. globosus snails are considered more efficient shedders of cercariae than B. nasutus (Hamburger et al. Reference Hamburger, Hoffman, Kariuki, Muchiri, Ouma, Koech, Sturrock and King2004): however, these findings do not consider differential snail survival under adverse weather conditions, such as dry periods. This is an important consideration since B. globosus inhabits stable aquatic habitats while B. nasutus is found in less stable habitats (Stothard et al. Reference Stothard, Loxton, Rollinson, Mgeni, Khamis, Ameri, Ramsan and Savioli2000, Reference Stothard, Mgeni, Khamis, Seto, Ramsan and Rollinson2002a).
As environmental factors (e.g. rainfall, water drainage and vegetation cover) fluctuate temporally, they influence transmission patterns by changing the distribution and density of competent snail hosts (Kariuki et al. Reference Kariuki, Clennon, Brady, Kitron, Sturrock, Ouma, Ndzovu, Mungai, Hoffman, Hamburger, Pellegrini, Muchiri and King2004). How population dynamics of B. nasutus are affected by fine-scale spatial heterogeneity remains largely undetermined in dynamic environments such as coastal Kenya. The patterns of snail dispersion and dispersal within and between aquatic habitats are determined by habitat suitability and connectivity (Utzinger et al. Reference Utzinger, Mayombana, Smith and Tanner1997). Since dispersal is important for maintaining local and meta-populations following droughts and floods (Marti et al. Reference Marti, Tanner, Degremont and Freyvogel1985; Mavarez et al. Reference Mavarez, Amarista, Pointier and Jarne2002), the connectivity of suitable breeding and transitory aquatic habitats needs to be considered for implementation of effective control measures targeting snail populations.
The predominant method for determining snail habitat connectivity has long been capture-mark-recapture (Thomas and Tait, Reference Thomas and Tait1984; Woolhouse, Reference Woolhouse1992) and, more recently, molecular-based studies (e.g. polymerase chain reaction (PCR)) to examine the genetic relatedness of aquatic snail populations (Stothard et al. Reference Stothard, Bremond, Andriamaro, Sellin, Sellin and Rollinson2001; Chlyeh et al. Reference Chlyeh, Henry, Sourrouille, Delay, Khallaayoune and Jarne2002; Mavarez et al. Reference Mavarez, Amarista, Pointier and Jarne2002). In this study, we used fine-resolution satellite imagery and field data from aquatic habitats to detect potential dispersal pathways and probable locations of snail aestivation and develop a hydrologically based dispersal model. We applied indicators used commonly by palaeontologists to study processes leading to fossilization of organisms (taphonomy) in order to reconstruct prior habitat conditions and to validate the dispersal model. Our objectives were to evaluate landscape corridors that connect fragmented patches of habitats suitable for breeding of B. nasutus and enable its dispersal, and to determine habitat suitability for snail aestivation throughout our study area in Msambweni Division in coastal Kenya.
MATERIALS AND METHODS
Research setting
The study area is Msambweni Division, Kwale District, Coast Province, Kenya (Fig. 1), where prevalence of S. haematobium remains high despite long-term intensive chemotherapy campaigns. The climate is tropical with low temperatures of 23·5°C in July to 26·9°C in March–April and highs of 26·7°C in August to 36·3°C in February. Annual precipitation fluctuates widely, averaging 1200 mm, with 2 rainy seasons, the long rains (usually March–July) and the short rains (October–November). However, since 2001 and through 2005, the normal precipitation pattern in the area has been interrupted by a severe drought. Located on the coastal plains, the land cover is a mosaic of grasslands, rice and maize plots, and mango, palm and cashews trees, some in plantations. The soil composition ranging from clay loam to extremely sandy soils over crystalline and coralline limestone that generally provide good drainage heading southeast through the area towards the Indian Ocean (Mzuga et al. Reference Mzuga, Tole, Ukakuwun and Hoorweg1998).
Fig. 1. Aquatic habitats located in Msambweni Division, Kwale District, Kenya.
Water bodies include several rain-filled ponds (Marigiza, Kiziamkala, Maridzani, Bovo, Nimbodze, Mwachiangwa, Mwamagongo), and the Koromojo Reservoir (Fig. 1). Spring-fed waterways include the perennial Mukurumudzi River, and the seasonal Mwabanda River and Lukungwi Stream. The water levels in the rain-filled ponds and seasonal waterways fluctuate with rainfall and dry out during periods of prolonged drought. Fortnightly and monthly snail scooping was conducted at these water sources during March 1984–July 1987 (Sturrock et al. Reference Sturrock, Kinyanjui, Thiongo, Tosha, Ouma, King, Koech, Siongok and Mahmoud1990) at 40 human water contact sites and February 2001–August 2004 (during the drought period) at 24 human water contact water sites (Kariuki et al. Reference Kariuki, Clennon, Brady, Kitron, Sturrock, Ouma, Ndzovu, Mungai, Hoffman, Hamburger, Pellegrini, Muchiri and King2004). When the ponds and various locations along the waterways contain water, various snail species can be collected, including B. nasutus (Stothard et al. Reference Stothard, Llewellyn-Hughes, Griffin, Hubbard, Kristensen and Rollinson2002b), the only intermediate host of S. haematobium in the study area. Of all open water sources in the area, Nimbodze Pond, Maridzani Dam and Kiziamkala Dam (Fig. 1) have been the most important transmission foci with the highest numbers and proportion of B. nasutus shedding human cercariae (Sturrock et al. Reference Sturrock, Kinyanjui, Thiongo, Tosha, Ouma, King, Koech, Siongok and Mahmoud1990; Clennon et al. Reference Clennon, King, Muchiri, Kariuki, Ouma, Mungai and Kitron2004; Kariuki et al. Reference Kariuki, Clennon, Brady, Kitron, Sturrock, Ouma, Ndzovu, Mungai, Hoffman, Hamburger, Pellegrini, Muchiri and King2004; Clennon et al. unpublished observations).
Dispersal modelling: landscape characterization
LandSat 5 Thematic Mapper (TM) imagery (courtesy of the U.S. Geological Survey) (28·5 m multispectral) taken in June 1986 (a non-drought period) during the long rains was used to identify the most likely corridors for B. nasutus to transverse during dispersal by applying a normalized difference moisture stress index (NDMSI) using bands 4 (near-infrared, 0·7–1·3 microns) and 5 (middle-infrared, 1·3–2·5 microns) (Rock et al. Reference Rock, Vogelmann, Williams, Vogelmann and Hoshizaki1986). Pan-sharpened, LandSat 7 Enhanced Thematic Mapper (ETM+) imagery (courtesy of the U.S. Geological Survey) (14 m panchromatic, 28·5 m multispectral) taken during the drought (January 20, 2003) were used to detect suitable locations for aestivation (in soil or under vegetation) by applying the NDMSI and performing unsupervised classification. Multispectral Ikonos imagery (acquired on 4 March 2001) was obtained, and then pan-sharpened (1 m resolution) for detailed mapping of the Msambweni landscape and for verifying that no major landscape changes influencing the water drainage had occurred since 1986. All imagery was processed in Imagine 8.7 software (ERDAS, Atlanta, GA), and georectified to Universal Transverse Mercator (UTM) zone 37S projection, 1984 datum.
LandSat 5 TM satellite imagery (which preceded LandSat 7 ETM+) from 1986 was used because obtaining relatively cloud-free imagery for the Msambweni study area during the long rainy season is very difficult and few images are available. Thus, the 1986 imagery allowed us to determine where flooding occurs, while the 2003 drought imagery taken in the dry season during the drought allowed us to assess the soils for suitability for aestivation.
Dispersal modelling: delineation of drainage pathways
To determine the major pathways of water drainage through the study area, Shuttle Radar Topography Mission (SRTM) level 1 Digital Terrain Elevation Data (DTED) (courtesy of the U.S. Geological Survey) of 3-arc seconds (∼90 m) horizontal resolution and 1 m vertical resolution for the study area (path 165, row 63, World Geographic System 2) were obtained (Fig. 2A). The relative vertical accuracy is ±6 m (90% vertical error), but approaches 3 m in flat areas with little vegetation; relative horizontal accuracy is ±15 m (90% circular error). Elevation values represent the reflective surface, be it vegetation (e.g. trees), soil or manmade structures (e.g. houses). The advantage of using multi-frequency, multi-polarization radar system is the ability to collect terrain data through clouds, that often obscure the south coast of Kenya.
Fig. 2. Hydrological model processing steps. (A) Digital elevation model; (B) flow direction; (C) drainage network; (D) 1986 LandSat in false true colour; (E) moisture stress; (F) least cost dispersal routes between 2 source habitats to 2 sink habitats:
.
The DTED was initially processed in Imagine 8.7 and imported into ESRI ArcGIS 9.0 (Redlands, CA), using the Terrain Analysis Using Digital Elevation Models extension (TauDem; Tarboton, Utah State University, 2005) to characterize the landscape's terrain and model water flow. Smoothed elevation data were used to calculate slope per pixel, that subsequently allowed the determination of flow direction and flow accumulation, the variables needed to identify major drainage lines (Fig. 2B and C). Flow direction was estimated based on steepest slope on a triangular facet (Tarboton, Reference Tarboton1997) and the flat area flow direction method (Garbrecht and Martz, Reference Garbrecht and Martz1997). Flow accumulation was calculated as the number of upslope cells flowing into a cell. The initial drainage network model did not account for the irrigation system that transverses the area, as the irrigation canal is merely 2 m wide, so it was added to the network. In addition, a topographical wetness index was estimated as the ratio between slope and the area for a specific catchment to predict areas of water concentration. A visual accuracy assessment comparing a priori knowledge (known stream and river routes), the 1986 LandSat imagery and the unclassified 2001 Ikonos imagery with the predicted major drainage pathways to ensure the correctness of the calculations.
Dispersal modelling: dispersal least-cost analysis
Populations of snails are needed upstream along a drainage corridor for dispersal, but they do not need to be permanent. Permanent/semi-permanent populations were chosen for our model starting points, because they allowed us to use transient population locations to validate our model. Additionally, dispersal may at times require individual snails as well as their progeny to cross the Msambweni drainage system to other permanent/semi-permanent aquatic habitats.
To determine the least cost surface for B. nasutus snails to disperse across the Msambweni landscape, a weighted NDMSI map for habitat suitability along drainage corridors was created using 1986 LandSat imagery (Fig. 2D and F). Then, least-cost dispersal route corridors between known locations of B. nasutus breeding habitats were determined. The path of least cost maximizes the amount of soil/vegetation moisture (a measure of dispersal suitability) encountered along the water drainage network between source and destination, while also minimizing the distance travelled. This is done by using a weighted-distance function where the cost per cell is equal to the moisture stress index multiplied by the cell resolution (e.g. 28·5 m). The final accumulative cost is the sum of all movements between the source and sink habitat.
Model validation
To validate the spatial dispersal model and to better understand the underlying dispersal process of B. nasutus distribution in Msambweni, we investigated aquatic snail shell distributions and taphonomy prior to model creation and, then, examined fortnightly and monthly temporal sequences of B. nasutus collection data from water sites previously not considered to be connected.
Snail and shell characteristics and distribution. Two methods, exploratory surveying and transect sampling, were used to determine potential water drainage corridors and presence/absence of snails prior to model creation. Exploratory surveying (August 2003 and June–July 2004) consisted of walking along probable water drainage pathways, collecting aquatic snail shells and live B. nasutus, and taking global positioning system (GPS) readings (Trimble, GeoExplorer II GPS). While the distribution of B. nasutus shells was our focus, shells of other aquatic snails were also collected, as they also indicate previous water flow through an area. Transect sampling for aquatic snail shells was conducted along 4 drainage corridors during June–July 2004 (Fig. 3). Printed maps of the Ikonos imagery (1 m resolution) were used for precise mapping of sampling locations. All shells from 4 m2 plots were collected along these transects. Different sections along the drainage valleys vary in width, so sampling plots were selected every 2–10 m starting from the lowest point of the valley depression and outward on each side for at least 50 m or until a defined slope was reached. Rice fields were sampled across the rice field grids, or with verbal consent of the owner. Locations where exploratory collections were conducted prior to plot sampling were noted. Each sampling plot was described in terms of distance to the valley centre, soil type and percentage of vegetation cover (ranked by quantiles as <25%, 26–50%, 51–75%, >75%).
Fig. 3. Maps of drainage corridors and species compositions (by percentages and numbers) of aquatic snail shells per drainage corridor collected using transect sampling.
Distribution patterns and taphonomic characteristics of shells (operculum presence or articulation, dissolution, rounding, breakage and concentration, size) were considered in order to assess the environmental conditions (e.g. water energy levels and possible post-mortem transportation) of each drainage corridor.
RESULTS
Survey sampling
Aquatic snails (or shells) of 8 species (B. nasutus, Lanistes purpureus, Melanoides tuberculata (2 morphotypes), Cleopatra spp., Bulinus forskalii, Pila ovata, Lymnaea natalensis and Neritina [pulligera] knorri) were collected (Fig. 3). We also found snails and shells of an amphibious snail, Lymnaea columella.
Exploratory survey
During the initial exploratory survey (August 2003), B. nasutus were found consistently nearly every 10 m between Mwamagongo and Maridzani dams, with the exception of the 100 m leading to Maridzani Pond, where the drainage corridor widened and was covered by dense vegetation. In 2004, live B. nasutus snails were found along the Mwabanda River and at the culverts northwest of Koromojo Reservoir, and B. nasutus shells (and occasionally live snails) were found repeatedly along an abandoned irrigation channel connecting Koromojo Reservoir in the north to Lukungwi Stream in the south. Bulinus nasutus shells were also found east of Koromojo Reservoir by the Mukurumudzi River. Along the Lukungwi Stream flood plain, shells of L. natalensis, M. tuberculata and B. nasutus were collected (Fig. 3).
Transect sampling
Concentrations of shells (including B. nasutus) were found across the rice fields north Nimbodze and Lukungwi Stream drainage lines and not only on the sides of the valleys as would be expected following deposition by flowing waters (Fig. 3). Shell concentrations (including shells of B. nasutus) were scattered across the northern portion of the valley leading to Mwamagongo and Maridzani ponds, and were found more evenly distributed in the extreme south. Shells were generally most abundant in sandy loam or heavily silted soil, regardless of percentage of vegetation cover.
The transects along the Maridzani Dam drainage line had more shell breakage and generally uniform shells in terms of size and shape, whereas shells found along transects leading to Nimbodze Pond exhibited low levels of shell breakage and a greater diversity in size, indicative of relatively calmer environments (Fig. 3).
Lymnaea natalensis shells were collected just outside a sampling plot at a culvert near Nimbodze Pond and in the Maridzani Dam corridor (Fig. 3). They were also found along the northern portion of the irrigation scheme, some 300 m south of Koromojo Reservoir. The presence of L. natalensis suggests that permanent water sources, flowing or stagnant, are connected to these areas (Brown, Reference Brown1994; Utzinger and Tanner, Reference Utzinger and Tanner2000). Notably, several shells of N. knorri, a snail that inhabits rivers and streams were collected (with intact operculums) in the drainage corridor just past the culverts south of Nimbodze Pond.
Dispersal least-cost analysis
A drainage network model was developed from digital elevation model (DEM) (Fig. 2C). The results of the least-cost analysis for dispersal from potential reservoir B. nasutus populations upstream of known transmission sites are shown in Fig. 2F. The drainage line calculations revealed 3 defined dispersal pathways connecting different B. nasutus breeding habitats. One path connects the Mwabanda River in the northwest to Marigiza Pond and then all the way to Maridzani Dam. Two other paths lead to Nimbodze pond, one begins along the irrigation system just west of Nimbodze Pond, and one along Lukungwi Stream.
Landscape characterization
The 1986 flood imagery revealed considerable amounts of pooled water in the Lukungwi and Nimbodze drainage corridors where we found shells in high abundance. Of particular importance, it shows a series of temporary habitats from Lukungwi Stream through a drainage corridor south of Nimbodze pond that are probably suitable not only for survival but also for reproduction of B. nasutus. Indeed, live snails or shells of B. nasutus were found in locations where pooling of water was predicted based on topographic wetness values and where soil moisture was moderate or high (Fig. 4).
Fig. 4. Moisture levels at predicted locations for water pooling and at sites where Bulinus nasutus have been collected.
Temporal malacological data
During 1984–1987 the spatial distribution of B. nasutus followed a temporal trend starting with snails collected along Lukungwi Stream (Site A) followed by increasing numbers of snails collected downstream at a road culvert (Site B) leading to Nimbodze Pond and, finally, at Nimbodze Pond (Site C) (Fig. 5).
Fig. 5. Biweekly number of Bulinus nasutus collected at three sites in the Msambweni, Kenya study area. March 1984–March 1987.
Sites: A, ——— B, - - - - C
Time lags of differing lengths between water presence at habitats and B. nasutus presence were also observed during the drought period of 2001–2004. In 2001, the B. nasutus snail population fluctuations at Bovo Pond were similar to those upstream at Mwamagongo Pond, but snail activity was higher at Bovo during 2002 and 2003. In fact, a population peak in September 2003 associated with flooding coincided with flooding and subsequent appearance of B. nasutus after an 8·5 month absence at Maridzani Dam, just south of Bovo. Since January 2002, there have been no sightings of B. nasutus at Kiziamkala Dam which is connected to the Maridzani Dam water drainage pathway, although water was present during 8 collection periods.
In the southwest portion of the area, the Lukungwi Stream connects the rice field site with Nimbodze Pond and 2 stream sites (Fig. 5). There were only 2 occasions when water was present in the rice field site and no B. nasutus were collected. Nimbodze Pond, a known epicentre for S. haematobium transmission, was found to have B. nasutus snails from February 2001 to 2002. Although water returned to the pond in May 2003, a lag of 4 months occurred before any B. nasutus were found again.
DISCUSSION
Environmental and meteorological data derived from coarse satellite imagery (spatial resolution of 1 km or more) have been used to develop risk maps for schistosomiasis at country, regional and continental scales for human schistosomes (Malone et al. Reference Malone, Huh, Fehler, Wilson, Wilensky, Holmes and Elmagdoub1994; Bavia et al. Reference Bavia, Hale, Malone, Braud and Shane1999, Reference Bavia, Malone, Hale, Dantas, Marroni and Reis2001; Bergquist et al. Reference Bergquist, Malone and Kristensen2000; Abdel-Rahman et al. Reference Abdel-Rahman, El-Bahy, Malone, Thompson and El Bahy2001; Zhou et al. Reference Zhou, Malone, Kristensen and Bergquist2001). Such risk models allow for general predictions of presence of snail intermediate hosts over wide areas, but do not take into account local heterogeneity (Kitron, Reference Kitron2000). Fine spatial resolution imagery (30 m or less: LandSat, SAR, Ikonos) can be used to study ecological processes operating on a local level (Kitron et al. Reference Kitron, Clennon, Gürtler, King, Cecere, Vázquez-Prokopec, Thornhill, Beck, Confalonieri and Marinho2006). While such modelling efforts have been initiated for the amphibious intermediate host of S. japonicum in China (Zhou et al. Reference Zhou, Malone, Kristensen and Bergquist2001, Reference Zhou, Dandan, Huiming, Honggen, Leping, Guojing, Qingbiao, Brown and Malone2002; Guo et al. Reference Guo, Vounatsou, Cao, Utzinger, Zhu, Anderegg, Zhu, He, Li, Hu, Chen and Tanner2005; Zhang et al. Reference Zhang, Xu, Zhou, Zhou and Liu2005), they have not been applied to bulinid snails whose distribution is restricted to aquatic habitats.
In a study of S. japonicum in China, the spatial distribution of the miracidia and cercariae was associated with a survey-based hydrological model of a water drainage network (Maszle et al. Reference Maszle, Whitehead, Johnson and Spear1998). More recently, Shaman and colleagues (Reference Shaman, Stieglitz, Stark, Le Blancq and Cane2002, Reference Shaman and Day2005) applied a hydrological model based on topography and soil moisture to predict mosquito abundance within watersheds. Hydrological modelling is highly suitable for understanding the connectivity of spatially distributed breeding and aestivation habitats of bulinid snails, and for determining the repopulation potential of habitats following mollusciciding.
Bulinus nasutus populations can tolerate the dynamic drought-flood process occurring in Msambweni. Coupled with low S. haematobium patent infection rates, this may sustain B. nasutus metapopulations while facilitating S. haematobium transmission. The patches where B. nasutus live snails and shells were found included dispersal pathways (which are only very transiently occupied), temporary habitats and breeding habitats. Dispersal pathways in Msambweni are generally characterized as having little vegetation, with predominantly light, sandy soil along water drainage routes where water often moves relatively swiftly. Temporary habitats with dark, sandy soil and a certain degree of shade (suitable for aestivation) are present along these pathways.
Following sufficient rains, B. nasutus snails have been found in pools along the drainage corridors not only persisting but, in habitats with relatively stagnant waters over less sandy, dark soil, at times also breeding. Koromojo Reservoir has not been found to maintain B. nasutus populations, but rather appears to be part of the aquatic route along which snails disperse. Indeed, culverts in the area have been found to provide very suitable habitats for B. nasutus.
Two main routes, the drainage systems from the Lukungwi stream and from the rice fields northwest of the pond, have now been identified as leading to Nimbodze Pond, and both routes include breeding sites for B. nasutus. The rice field site is particularly important in that it is not only linked with the irrigation scheme, but is also connected to Lukungwi Stream that, in turn, links it to other breeding habitats. An old irrigation system west of the human population centres continues to impact the study area not only through water drainage but also by acting as a well-defined dispersal corridor for B. nasutus snails.
These recent findings suggest that within Msambweni, snails, specifically B. nasutus, are not merely travelling through flood plains from sources to transmission sites (e.g. Nimbodze Pond). Rather, they are using well-defined dispersal corridors along the irrigation scheme and streams/rivers. The primary route of snail dispersal is not necessarily the shortest route linking source to destination, but rather the route with the better defined corridor and lower levels of moisture stress. This is in agreement with recognized routes of spread for snails elsewhere (Marti et al. Reference Marti, Tanner, Degremont and Freyvogel1985), and highlights the role of anthropogenic changes to the landscape (construction of irrigation canals and road culverts) that increase the availability and connectedness of aquatic habitats suitable for intermediate host snails (Steinmann et al. Reference Steinmann, Keiser, Bos, Tanner and Utzinger2006). In Msambweni, the likely result is a more continuous input of host snails into areas of S. haematobium transmission and less dependence on floods and heavy rains.
Nimbodze Pond, where transmission of S. haematobium has been most intense in our study area, exemplifies the importance of connectivity. Given that Nimbodze Pond was dry for a prolonged period (19 months in 2002–2003) and repeated burnings of vegetation, it is unlikely that B. nasutus have survived there continuously through aestivation. Rather, Nimbodze Pond receives water from a northern hilly area and from rice fields and a stream, where B. nasutus have been found, and that following a heavy rain, remain connected with Nimbodze for several days.
The primary landscape change that occurred between 1986 and 2000 was a shift from sugar to rice agriculture in the western portion of Msambweni. In our study, we used 1986 LandSat imagery to determine wetness levels along potential dispersal routes, because of the difficultly of getting imagery with low levels of cloud cover during the long rains. A comparison between the 1986 LandSat and 2001 Ikonos revealed a high level of similarity between the boundaries of areas where water has often flowed or accumulated. If anything, the conversion to rice agriculture may have increased the potential for snail dispersal, through additional water pooling in those fields and increased connectivity to Nimbodze Pond.
Aquatic sites that have now been demonstrated to be linked with Nimbodze Pond, have been found to contain not only uninfected B. nasutus, but also B. nasutus shedding S. haematobium cercariae (Sturrock et al. Reference Sturrock, Kinyanjui, Thiongo, Tosha, Ouma, King, Koech, Siongok and Mahmoud1990). One implication is that measuring S. haematobium water contamination by humans using snail-based methods (Hamburger et al. Reference Hamburger, Hoffman, Kariuki, Muchiri, Ouma, Koech, Sturrock and King2004) at a particular pond, may be confounded by infected snails entering the habitat from elsewhere.
It has been documented that B. nasutus due to its superior aestivation strategy is more tolerant than B. globosus of the dynamic drought-flood patterns that are found in Msambweni (Rollinson et al. Reference Rollinson, Stothard and Southgate2001). While B. globosus snails may be dispersed in the same manner (Marti et al. Reference Marti, Tanner, Degremont and Freyvogel1985), their ability to survive long distances using very temporary habitats is questionable. Snails with patent schistosome infections are less likely to survive into aestivation than those with pre-patent infections (Sturrock, Reference Sturrock, Jordan, Webbe and Sturrock1993), and their reproductive output is curtailed as infection nears patency, at which point they stop laying eggs entirely (Gracio, Reference Gracio1988). Of B. nasutus collected in Msambweni, the vast majority of those infected with S. haematobium have pre-patent rather than patent infections (Hamburger et al. Reference Hamburger, Hoffman, Kariuki, Muchiri, Ouma, Koech, Sturrock and King2004). While B. nasutus snails may not shed cercariae as profusely as B. globosus, they are more than adequate S. haematobium intermediate hosts, as is manifested in more than half of school-aged children in Msambweni being infected (Clennon et al. unpublished observations). Thus, although B. globosus may be a superior intermediate host in sustained aquatic habitats, B. nasutus is an efficient intermediate host under dynamic conditions with transient aquatic habitats, as are found in Msambweni.
High concordance between the distribution of live snails and of snail shells in relatively unperturbed freshwater systems allows for reconstruction of past environmental conditions (Cummins, Reference Cummins1994). Understanding the processes affecting snail distribution and shell condition, is a less expensive alternative to molecular profiling, and one that can be pursued during dry periods to study snail dispersal in an area. Phylogenetic analyses can confirm whether the snails in the sinks/destinations are more closely related to snails in various source habitats.
After flooding or sufficient rains, individual snails and potentially snail populations can traverse environmental boundaries between different aquatic habitats within the study area. Dispersal routes throughout the study area (some better defined than others) connect many of these habitats. When considering dispersal routes, the directionality of dispersal along a route needs to be taken into account (e.g. knowing the source of dispersal to optimize mollusciciding or to construct mechanical barriers).
Our findings have applications for S. haematobium transmission control programmes targeting B. nasutus snail populations along the southern coast of Kenya. Highly connected dispersal routes allow for reintroduction of bulinid snails and re-emergence of S. haematobium transmission within the area following human chemotherapy campaigns and mollusciciding. Determining the spatial and temporal patterns of connectivity between aquatic habitats allows for identification of potential dispersal pathways of B. nasutus snails, and needs to be considered in S. haematobium transmission control programmes targeting snail populations.
This research was supported by the National Institutes of Health/National Science Foundation Ecology of Infectious Disease programme grant (AI45473) and by the Fogarty International Center grant (TW/ES01543) awarded to Dr Charles H. King (CWRU). This project was supported in part by Grant Number 1 C06 RR16515 to the UIUC College of Veterinary Medicine from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NCRR or NIH. We thank the residents of Milalani, Nganja, Vidungeni, Maragiza, Mabatani, Sawa Sawa, Bomani, Mwaembe, Kisimachande, Vingujini and Vidungeni villages in Msambweni Division, Kwale District, Kenya for their gracious cooperation. We would also like to thank Idi Masemo, Joyce Bongo, Robinson Ireri and Mwanasha Chuo who helped with the shell surveys. Cindy Schmidt, Brad Lobitz and Louisa Beck from NASA-CHAART provided imagery processing training. Dr Robert Sturrock participated in helpful discussions. Publication of this paper is done with the kind permission of the Director of Medical Services, Ministry of Health, Kenya. This research was conducted in partial fulfilment of Ph.D. degree requirements by Julie Clennon.
