Hydrobiologia (2006) 553:129–142 Springer 2006 DOI 10.1007/s10750-005-0968-9

Primary Research Paper Spatio-temporal distribution of freshwater species in relation to migration and environmental factors in an irrigated area from Morocco

Ghita Chlyeh1, Marine Dodet2, Bernard Delay2, Khalid Khallaayoune3 & Philippe Jarne2,* 1De´partement de Zoologie, Rabat Institut, IAV Hassan II, 6202, BP, Morocco 2CEFE-CNRS, 1919 route de Mende, 34293, Montpellier Cedex 5, France 3De´partement de Parasitologie, Rabat Institut, IAV Hassan II, 6202, BP, Morocco (*Author for correspondence: E-mail: [email protected])

Received 20 September 2004; in revised form 6 May 2005; accepted 2 June 2005

Key words: freshwater , irrigation area, environmental factors, human diseases, demography

Abstract Nine sites were sampled 19 times over 2 years in an irrigation system in Morocco in order to study species abundance in a snail community in relation to environmental parameters (including human activities) and migration (geographic distance) among sites. Each site was made of a sink and the first meters of the downstream canal. The snail community included four species (Bulinus truncatus, Lymnaea truncatula, Mercuria similis and Physa acuta). Strong spatial variation in species occurrence and abundance was detected which might be partly due to variation in water availability. However abundance in sinks and canals in which water availability differs were correlated. There was, as predicted, limited evidence in favor of isolation by distance which might be due to fast water current. Dispersal might therefore be an important factor structuring this community. On the other hand, the temporal variation was much more limited. This is consistent with the analysis of individual size distributions in B. truncatus, since no clear-cut cohorts were detected. The environmental parameters recorded (e.g. temperature, occurrence of macrophytes or cleaning of sinks) were extremely variable in time and space, except temperature. Analyzing their association with species through multidimensional methods indicated that P. acuta is ubiquitous and B. truncatus positively associated with macrophytes. These two species were associated in sinks. Less clear trends were detected for the two other species. Annual cleaning of sinks affected all species, but population recovery was fast in B. truncatus and P. acuta.

Introduction agricultural purposes in dry areas creates favor- able habitats for communities of freshwater One of the goals of freshwater ecology is to organisms (see e.g. Brown, 1994). Studying such understand how communities of freshwater species systems is interesting in several respects. (i) The are structured in space and time, and how envi- structure of habitats in irrigation areas is generally ronmental factors affect their distribution. Studies much simpler than that of natural environments have generally been conducted in natural envi- (e.g. concrete canals versus rivers), and habitats are ronments, such as rivers (e.g. Leland & Fend, much more clearly delimited and more readily 1997; Lewis & Magnuson, 2000; Lods-Crozet accessible to sampling. As a consequence, the et al., 2001; Stewart et al., 2002; Paavola et al., number of species per site is lower than in natural 2003). However, the development of irrigation for habitats, and it should be easier to evaluate species 130 distribution and how they are affected by envi- various environmental parameters, including ronmental factors. (ii) In species spending their water availability, which should lead to differences whole life-cycle in water (e.g. snails), migration in community structure. (ii) Previous work on the among populations occurs either passively or population genetics and dynamics of one of the actively in water only – note though that migration species considered here (Bulinus truncatus; mediated by birds is possible. Migration routes can Audouin, 1827) showed high migration rate more easily be determined in irrigation areas than among populations, with snails moving for more in natural environments. This allows explicitly than 100 m per reproductive life (Chlyeh et al., taking into account the role of space (geographic 2002, 2003). Migration should synchronize species distances among sites) in the structure of fresh- abundance among sites, at least at short geo- water communities, a point which has been high- graphic distances. Moreover, given that the geo- lighted in ecology from a theoretical perspective graphic scale is extremely limited, environmental (e.g. Borcard et al., 1992; Keitt et al., 2002). parameters, such as temperature, should have the Geographically close sites can indeed be more same effect. We therefore predicted that there similar than sites separated by larger distances, should be limited isolation by distance. In other that is that there could be some isolation by dis- words, the similarity in abundance should not tance. A likely cause for such a pattern is migra- decrease with geographic distance among sites. tion among populations (references in Clobert This was further evaluated in two species et al., 2001). (B. truncatus and Lymnaea (Galba) truncatula; We focus here on the snail component of a Mu¨ller, 1774) based on the analysis of the tem- freshwater community from an irrigation area in poral distribution of individual size: cohorts Morocco. Several studies have monitored the ini- should also be synchronized. (iii) Species commu- tial expansion or analyzed the subsequent distri- nities and their environment can evolve relatively bution of snails in such areas (e.g. review in rapidly in time which promoted repeated temporal Brown, 1994; Sturrock et al., 2001). One reason is sampling in studies analyzing communities of that some snail species are involved in the trans- benthic organisms (e.g. Leland & Fend, 1997; mission of water-dependent diseases in tropical Lods-Crozet et al., 2001; Kanandjembo et al., and arid areas (see e.g. Brown, 1994; Dillon, 2000), 2001). However the time scale retained is generally and the development of irrigation areas has of the order of several months or years, and might favored their development (Brown, 1994; Sturrock therefore not be appropriate for evaluating short et al., 2001; Boelee, 2003). However, these studies time-scale variation (e.g. among seasons). Such focused on vector snails, and they generally did variation has indeed been largely documented not take into account the fact that these snails through population biology approaches of single belong to communities of benthic organisms, species (see, for examples in snails, Brown, 1994; including other snails. The irrigation system stud- Sturrock et al., 2001). ied was built about 50 years ago for agricultural We also monitored several environmental development (e.g. orange orchards). It is consti- parameters that can be evaluated simply (e.g. tuted by a series of concrete sinks which are con- water temperature, water current or occurrence of nected by concrete aerial canals, and the maximum macrophytes) and which are known to affect the distance between two points is about three km. structure of snail communities (review in Dillon, The environment is therefore very simple, and 2000). Sinks are annually cleaned out by farmers, populations are well localized (essentially in sinks in an effort to control populations of B. truncatus. and parts of canals neighboring sinks). Nine sites, This species is indeed involved in the transmission separated by less than 3 km, were surveyed of urinary bilharziasis, and the area studied is a monthly over 2 years. In each site, both the sink transmission focus (see Laamrani et al., 2000). and the first meters of the downstream canal were Time after cleaning was considered in the analysis sampled. Our goal was to analyze how the snail of environmental parameters. The relationship of community (four species) is structured in space (i) these parameters with community structure was between sinks and canals and (ii) among sites, as analyzed through multidimensional analyses. We well as (iii) in time. (i) Sinks and canals differ in predicted that water temperature should have a 131 strong influence on the abundance of all species, schematic map of the irrigation system is presented since it has been shown to strongly affect snail in Figure 1. Water is pumped from ground water populations and communities (Dillon, 2000), as at several stations, and directed towards irrigated well as a collapse of the whole community after fields and orchards through a series of primary and cleaning of sinks. secondary concrete aerial canals. Canals are indeed half-pipes (diameter: 0.4 m), in which water runs by gravitation along a gentle slope, for dis- Materials and methods crete periods of time according to a schedule defined by farmers. Pairs of squared-section sinks The sites studied (1 m), 0.5–2 m deep, are located about every 100 m along these canals. Sinks from a pair are The study was conducted within the irrigation separated by about 10 m and connected by system of Oulad Sid’cheikh, 20 km South of underground pipes. Sinks are paired to allow Marrakech, located in the Western part of the access to agricultural parcels. At most sinks, water Central Haouz region (3137¢ N, 811¢ W). This is can be redirected towards fields and orchards a semi-arid area with average annual precipitation though se´guias (clay canals). As water is running of 250 mm, and no precipitation between May at 0.4 m/s in concrete canals, water turbulence and October. Air temperature at noon varies within the first sink of a pair is much higher than between 10 C in January and 44 C in August. A in the second one. Some of the sites studied (see

Figure 1. Schematic diagram map of the Oulad Sid’cheikh irrigation scheme, located along the Marrakech – Essaouira road, giving the location of the four sites (numbered 1P–4P) and sinks studied, the connections between sinks (empty squares) through canals and the movements of water (arrows within canals). Filled squares are pumping stations. Two to three sinks, and downstream canals (see text for details) were studied per site (numbered 1–3). Note that the diagram is not of scale, and the largest distance along the Marrakech – Essaouira axis is 3 km. The actual distances (in m) between pairs of sites are 1P–2P: 300; 1P–3P: 1200; 1P–4P: 2700; 2P–3P: 900; 2P–4P: 2400; 3P–4P: 1500. 132 below) are directly connected by running water, canals. As water depth never exceeded 40 cm, allowing for potential downstream dispersal fol- sampling of canals was performed by hand. Snails lowing the water current (Chlyeh et al., 2002). were sampled irrespective of size and species over Populations of snails are found in sinks in 10 min by a single person as done by Madsen which water is available year round. However (1982). Three species were recorded: L. truncatula, populations of some species can be found along M. similis and P. acuta. L. truncatula individuals the canals, at least temporarily. This is especially were measured to the nearest 0.1 mm. As previ- true of the first meters of canals below the sinks ously, all individuals were put back in the canals in where water can stagnate long enough for snails to which they were sampled. Note that sampling in settle. We therefore studied snail populations in both sinks and canals was performed by G. Chlyeh sinks and downstream canals. Four groups of sites only. were studied (Sidi Mansour, Bahja, Boukrich and Hadj Laarbi; 1P to 4P in Fig. 1). Each group Environmental parameters consisted of two to three sites (Fig. 1; the first site of 1P was 1P1 and so on), a site being defined as a Several environmental parameters were recorded sink and the first meters of the downstream canal. in sinks, including temperature, substrate thick- Sites of 4P are not connected to other sites. The ness on walls (made of algae and silt), pH, water study was conducted from November 1997 to depth, water current (presence/absence), and speed November 1999 and in April 2000. Sampling took of water current (in canals, downstream of sinks). place about once a month, but no sample is The speed was measured using a chronometer and available for the July–August period. On the a piece of cork. Sinks are cleaned out by farmers whole, the irrigation area was visited 19 times. once a year (in May), and this is followed by molluscicide treatment by Marrakech SIAAP Snail sampling (Service d’Infrastructure et d’Action Ambulatoire Pre´fectoral). The goal is to control populations of Snail sampling was performed in sinks using a the vector-snails B. truncatus and L. truncatula. metallic dredge with a nylon mesh (mesh Time since last cleaning out was registered. The size = 0.8 mm; description in Khallaayoune & two sinks in group 2 (Bahja) were not cleaned out Laamrani, 1992). The dredge blade (20 cm wide) during our study. We mentioned above that sinks was applied once on each of the four walls are paired, and that water turbulence is much (from bottom to water surface), as well as on the higher in the first than in the second sink of a pair bottom, of sinks. Snails were collected together with which influences snail settlement. The position of algae and macrophytes covering sink walls and sinks in pairs was recorded (Upd). Macrophyte bottom. Individuals were sorted per species and and algae species were found in the irrigation area counted. Four species were recorded: B. truncatus, (listed in Chlyeh, 2002). Their abundance was L. truncatula, Mercuria similis (confusa) (Forbes, encoded in four classes (0, 5–25, 25–50, >50%) 1838) and Physa acuta (Draparnaud 1805). More taking into account wall covering (for algae) and information on the distribution of these species in occurrence on bottom substrate. The occurrence of Morocco can be found in Ghamizi (1998), and arthropods (e.g. dragonflies, mosquito larvae) and Brown (1994) gives their distribution, except that vertebrates (frogs, toads and snakes) was recorded, of M. similis, in Africa. B. truncatus and L. but not analyzed quantitatively (see Chlyeh, 2002). truncatula individuals were measured to the nearest Environmental parameters were also recorded 0.1 mm using a digital calliper. Except for a few in canals, some of which were identical to those samples of B. truncatus and L. truncatula that were estimated in sinks: water presence (absent, humid, stored in 85 alcohol for genetic analyses running), water depth, substrate thickness on (Chlyeh et al., 2002), all individuals were put walls, algae and macrophyte abundance. Water back in the sinks in which they were sampled. temperature was recorded when water was present Snails were also sampled in the canals down- in the canals. However it cannot be used in the stream of the sinks considered above, except for analyses since data were missing when canals have sink 1P3. They were sampled in the first 2 m of dried out (this also holds for water depth). We 133 instead used a season parameter, and samples were was 1 for sinks from the same pair (within a site), 2 attributed to four classes corresponding to the four for sites separated by a canal (about 100 m), 3 for seasons. This parameter was indeed correlated sites separated by several sinks, and 4 when the with the average water temperature (estimated for two sites did not belong to the same part of the each season and sink) across sinks (Spearman irrigation area (i.e. 4P versus other groups). For rank-order correlation coefficient, rs = 0.386, the frequency-based distance, the analysis was n = 72, p <10)5). performed for sinks only, since there were not enough data for canals. The correlation between Statistical analysis the biological and geographical distances was tes- ted using a Mantel test. The first part aimed at evaluating the variation of The spatio-temporal homogeneity of environ- the snail community in space and time. The snail mental parameters (temperature, depth, substrate distribution was analyzed using presence/absence thickness, macrophyte and algal abundance) was of species per site (sinks and canals) and sampling tested among sites using Friedman two-way date. This allowed drawing curves of occurrence ANOVA. The relationship of environmental (proportion of sites occupied) as a function of parameters with snail abundance (log(1+x)- sampling date for each species. Curves were com- transformed data) was investigated using factorial pared between sinks and canals for each species correspondence analysis on instrumental variables using Spearman rank-order correlation coefficient. (FCAIV; Lebreton et al., 1988), a procedure which The spatio-temporal homogeneity of abundance is very similar to canonical correspondence anal- was tested using Friedman two-way analysis ysis (Ter Braak, 1986). Four analyses were con- of variance (ANOVA) for each species on ducted as described above for the FCAs. Note that log(1+x)-transformed data. The homogeneity of the number of environmental variables varied with snail distribution was further analyzed using fac- analyses (N = 7, 8, 7 and 4) since different sets of torial correspondence analysis (FCA; Lebart et al., variables were recorded in sinks and canals. 2000) on log(x+1)-transformed abundance data. A note of caution should be introduced here. Four datasets were analyzed: sinks (all data), sinks Even if the surfaces sampled in sinks and canals after cleaning out (i.e. a restricted dataset com- (see Materials and Methods) were comparable pared to the first one), canals, and sinks and canals. (about 1 m2 in sinks 1 m deep vs. 0.8 m2 in For each of these datasets, coordinates on the first canals), they were not identical. The sampling FCA axis were extracted and used in ANOVA in technique also differed between sinks and canals. order to test for the influence of the time (sampling This resulted in different capture probability of date), group and site (nested within group) factors. snails, and comparisons of snail numbers between Isolation by distance was examined by com- sinks and canals should therefore be considered paring biological and geographical distances cautiously. This is of concern in joined analyses of between pairs of sites. Two biological distances sinks and canals, though not when data were were used. The first is based on presence/absence analyzed separately (most analyses). Moreover we of a given species. It was 0 when a given species were more interested in the spatial and temporal was either present or absent at the two sites con- variation in snail numbers, rather than in the sidered, and 1 otherwise. It was then averaged over actual values, and the log-transformation of data species and sampling date. Secondly, a Euclidean prior to analysis tends to erase differences in actual distance based on frequencies was used (David numbers. et al., 1997): The analysis of cohort data was restricted to three sites (2P1, 2P2 and 3P2), since not enough rXffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 data were available for the other sites. From the djk ¼ ðCij CikÞ distribution of individual size classes, it is some- i times possible to infer the number of cohorts where j and k are two sites and cij the proportion occurring in a population at a given time based on of species i in site j. This value was then averaged the number of modes (e.g. Ebert, 1999). The over sampling dates. The geographical distance temporal variation of this distribution allows 134 inferring demographic events such as recruitment Similar results were found when considering of new cohorts (see more details in Charbonnel species abundance with significant heterogeneity et al., 2002). However such histograms are often among sites for each species, both in sinks and ca- hard to interpret when size classes overlap over nals (Friedman two-way ANOVA: p < 10)3). cohorts and sample size is limited. Persat & Examples of temporal variation in abundance in Chessel (1989) developed a method based on cor- two sinks are shown in Figure 3. This was further respondence analysis adapted to such situations. investigated using FCA. In all analyses, about 80% Briefly, the analysis is performed on a table in of the variation was distributed on the first two axes. which cells give the number of individuals col- When sinks only were considered, B. truncatus and lected according to sampling date (row) and indi- P. acuta behaved very similarly, as shown by the vidual size class (column). The results provide a projection of data barycenters on the first axis representation of factorial coordinates and eigen- (Fig. 4). On the second axis, P. acuta is more closely values, and the smoothing of histograms by an associated with L. truncatula. Using the coordinates appropriate statistical transformation, thus facili- on the first FCA axis as data points, an ANOVA tating the interpretation of temporal data. More showed significant variation for the group factor details and an example in snail populations are only (Table 1). Very similar results were obtained given in Charbonnel et al. (2002). when post-cleaning data only were considered. A The analyses were conducted using Biomeco highly significant effect of sampling date was (Lebreton et al., 1990) and PrAxis (Reboulet et al., 1995) for multidimensional analyses, Genepop (Raymond & Rousset, 1995) for isolation by dis- (a) Frequency tance, and Jump (Sall & Lehman, 1996) for all 1 other analyses. 0.8 0.6

0.4 Results 0.2 Snail species 0 1 5 91317 The four species collected showed variation in their Time spatial and temporal distributions (see Fig. 2). P. acuta were detected at more than 60% of the (b) Frequency sampling sites in the study. B. truncatus was also 1 extremely frequent in sinks, but was never found in canals. On the other hand, M. similis occurred at 0.8 the 4P sites, but in autumn and winter only. 0.6 L. truncatula was found more often, but was almost 0.4 absent in 1999. The analysis of species presence/ absence indicated variability in mean occurrence 0.2 per site between species, both in sinks and canals )3 0 (Friedman two-way ANOVA: p <4· 10 ; 1 5 91317 Fig. 2). Far fewer individuals were found in canals Figure 2. Frequency of snail species (proportion of sites where than in sinks, but species occurrence in canals ap- a species was collected) in time (sampling date) in (a) sinks pears as a subset of that in sinks. When considering (N = 9) and (b) downstream canals (N = 7). The irrigation all sampling dates, the frequency of occurrence area was visited about monthly over two years (19 visits; see over all sites was significantly correlated between text). Dates 1–8: November 1997–June 1998; dates 9–16: Sep- sinks and canals for L. truncatula (r = 0.733, tember 1998–June 1999; dates 17–18: September–November )3 )3 1998; date 19: April 1999. Dotted parts of lines indicate gaps p <10 ) and M. similis (r = 0.701, p <10 ), and between successive samples of more than a month. Diamonds: the correlation was almost significant for P. acuta Bulinus truncatus; squares: Lymnaea truncatula; triangles: Physa (r = 0.376, p = 0.10). acuta; crosses: Mercuria similis. Arrows refer to sink cleaning. 135

Number of because no snails were detected in many occur- individuals 3P2 rences. When using distances based on presence/ 400 absence, isolation by distance was detected in sinks (p = 0.003), but not in canals (p = 0.250). 350 300 Environmental variables 250 200 All environmental variables showed significant 150 temporal variation, as exemplified by temperature 100 and depth in four sinks in Figure 5. There was also 50 spatial variation, as shown for depth in Figure 5, 0 except for temperature (Fig. 5). Accordingly the p-values associated with Friedman two-way 1 5 9 13 17 ANOVA were highly significant (p < 0.001), ex- Sampling date cept for temperature (p = 0.404), in both sinks and canals. Note that the water pH and speed of Number of water current (when water was running in canals) individuals did not vary among sites and sampling dates 4P2 400 (pH = 7; speed = 0.4 m/s) and are not further 350 considered. 300 FCAIV were conducted on four sets of data. The percentage of explained variance on the first two 250 axes of these analyses was higher than 90%. (i) Sinks 200 (all data): the analysis was conducted on four spe- 150 cies and seven environmental variables. The asso- 100 ciation of environmental variables on the first two 50 axes is shown in Figure 6. Note the strong associa- tion between macrophyte and algae, and to a lesser 0 15 9 13 17 extend between depth and substrate thickness. The abundance of B. truncatus was associated with high Sampling date values of algae, macrophytes and substrate param- Figure 3. Number of individuals per species as a function of eters, that of L. truncatula with temperature and sampling date in sinks 3P2 and 4P2. Dates 1–8: November Upd, that of M. similis with low values of algae, 1997–June 1998; dates 9–16: September 1998–June 1999; dates 17–18: September–November 1998; date 19: April 1999. Species macrophyte and temperature, while P. acuta did not are depicted as in Figure 2, dotted lines have the same meaning show very specific associations. (ii) Sinks post than in Figure 2, and arrows refer to sink cleaning. cleaning out: B. truncatus was still associated with algae and macrophytes. It is noticeable that the detected in addition to that of group (Table 1). abundance of both B. truncatus and P. acuta was When considering canals only, the ANOVA model positively associated with time post-cleaning out, did not indicate significant variation on the first while the reverse was observed for the two other FCA axis, and the influence of factors was not fur- species. The abundance of P. acuta was also posi- ther considered. The FCA analysis including all tively associated with Upd. (iii) Canals: the abun- data opposed B. truncatus and P. acuta to the two dance of L. truncatula was positively associated with other species on the first axis, but these two species macrophytes, that of P. acuta with substrate thick- can be differentiated on the second axis. All effects ness, algae and water presence, and M. similis with tested were significant (Table 1). depth and Upd. (iv) When using the whole dataset, Isolation by distance was not detected when B. truncatus was associated with high values of algae using pairwise distances based on species fre- and macrophytes (the reverse holds for M. similis), quencies (Mantel test, 1000 permutations, L. truncatula with substrate thickness and P. acuta p = 0.168). The test was not conducted in canals with depth. 136

(a) 1.5 1 0.5 0 F1 -1 -0.5-0.5 0 0.5 1 1.5 2 2.5 3 -1 F2

(b) 2

1

0 F1 -1 -0.5 0 0.5 1 1.5 2 2.5 3 -1 F2 Figure 4. Distribution of species barycenters (a) and of individual sampling points (b; N = 171) in sinks on the first two axes of the factorial correspondence analysis. Species are depicted as in Figure 2.

Table 1. Results of ANOVA on individual coordinates on the first axis of the factorial analysis for four sets of data

Dataset Fit Date Group

Sinks F 1.44 0.94 4.34 1.51 p 0.09 0.53 0.01 0.19 df 26 & 145 18 & 153 8 & 162 4 & 163 Sinks post- cleaning out F 3.74 5.95 3.15 0.55 p <10)4 <10)4 0.03 0.74 df 26 & 55 18 & 62 8 & 72 4 & 73 Canals F 0.81 - - - p 0.72 df 24 & 109 Sinks and canals F 6.72 4.15 4.70 15.05 ) ) ) ) p <10 4 <10 4 2 · 10 4 <10 4 df 31 & 235 18 & 247 6 & 263 6 & 254

The effects of the sampling date, group of sites and sites are reported. F, p and df refer to the F-values of ANOVA and the associated probability and degrees of freedom. ‘Fit’ refers to the fit of the whole model. – = not done since the simplest model (without explanatory variables) was as good as more complex models.

Cohort analysis 2P2 with more than 40 individuals in the most numerous size classes at most sampling dates The occurrence of L. truncatula was too limited in (Figure 7). However it was not possible to detect both time and space to allow the analysis of cohorts. In 3P2, snail abundance increased from cohorts. For the same reason, the analysis in November 1997 to May 1998 without significant B. truncatus was limited to sink 2P1, 2P2 and 3P2. evolution in individual size. A peak of small-sized In 2P1, snails were absent or rare until May 1998, individuals in October 1998 (3 mm) can be and a spring cohort can be inferred. The distri- followed up to March 1999 (7 mm), a time at bution of individuals afterwards did not allow any which the largest individuals disappeared, possibly inference. Much more individuals were detected in as a consequence of cleaning out. 137

(a) Temperature 30 25 20 15 10 5 0 1 5 9 13 17 Sampling date

(b)

Depth 3 2.5 2 1.5 1 0.5 0 1 5 9 13 17 Sampling date Figure 5. Variation of water temperature (a) and water depth (b) as a function of time (19 sampling visits) in four sinks. Diamonds: 1P1; squares: 2P1; triangles: 3P1; crosses: 4P1. Dotted lines have the same meaning than in Figure 2, and arrows refer to sink cleaning.

Discussion example, Costil et al. (2001) found one to seven species per site in an area harboring 14 species in The irrigation area studied harbors a simple French Brittany. Studies over 200 sites in Guade- community of molluscs made of four species which loupe (Lesser Antilles) similarly returned values of represents a subset of the nine commonest species 0–10 species per site in a mollusc community of 25 found in an extensive survey of the Haouz area species (P. David, P. Jarne, J.-P. Pointier, unpub- (Ghamizi, 1998). These species were also detected lished data). Other examples can be found in by Laamrani (1999) in Attaouia, an irrigation area Dillon (2000, chapter 9). The spatio-temporal located about 50 km from the one studied here, distribution of occurrence of the four species was and are classical members of very uneven, as shown by the homogeneity tests. communities in the Western Mediterranean area M. similis was limited to two sites, and was never (e.g., Gallardo et al., 1994). However, these four found in spring. L. truncatula occurred more often, species were never all found together, since zero to which is consistent with its widespread distribution three species were collected per sampling site and in Morocco. More surprising is the fact that it date in sinks (zero to two in canals). Although it is disappeared for several months in 1998. B. trunc- difficult to compare with studies conducted in less atus and P. acuta were common at all sites and artificial habitats, far fewer species are generally times, and exhibited very similar distributions in found at single sites than at several sites (local vs. sinks. These two species both have extremely wide regional diversity; see e.g. Hubbell, 2001). For distributions -P. acuta is thought to be the most 138

(a) 0.8 0.6 0.4 0.2 F1 0 -1.5 -1 -0.5 -0.2 0 0.5 1 1.5

-0.6 F2 -0.8

(b) 0.6 0.4 0.2 F1 0 -0.4 -0.2-0.2 0 0.2 0.4 0.6 0.8 -0.4 -0.6 -0.8 -1 F2

(c) 15

10 5 F1 0 -8 -6 -4 -2 0 2 4 -5

-10 -15 F2 Figure 6. Distribution of (a) species barycenters, (b) barycenters of environmental parameters and (c) individual sampling points on the first two axes of the factorial correspondence analysis on instrumental variables conducted on data from sinks. Species in (a) are depicted as in Figure 2. Environmental parameters in (b) are: diamonds: temperature; squares: depth; triangles: substrate thickness; crosses: algae; stars: macrophytes; circle: water running; plus: Upd. cosmopolitan freshwater mollusc- and often significant factors in the PCA including all data. exhibit large population size (Brown, 1994; Dillon, This might partially be due to differences between 2000). sinks and canals (see discussion below), since the Our analyses were also based on the number of group factor only was significant when considering individuals per species. Wide variation in abun- sinks alone and no significant variation in abun- dance was observed in occupied sites with up to dance was found in canals. We now discuss the 675 B. truncatus individuals in 2P2 on May 25 factors that might explain this variation, namely (i) 1998. However, counts were more generally in the environmental parameters (meaning here all tens of individuals. Assuming that about one-fifth recorded factors besides snail abundance) and of the sink walls were sampled with our technique, time, (ii) geographical distance and (iii) coloniza- and that the capture probability if of the order of tion rate. 0.3 (Chlyeh et al., 2002, 2003), the number of (i) The first environmental parameter is the individuals per species in sinks were in the thou- distinction between sinks and the canals located sands, with a peak at 104. The variation observed immediately downstream them. The snail com- appeared in the results of ANOVAs on the first munity in canals is a sub-sample of that in sinks, in axis of the PCAs. Date, group and site indeed were both number of species and individuals per species. 139

Figure 7. Size distributions of Bulinus truncatus in sink 2P2 for the 19 sampling dates. Dates 1–8: November 1997–June 1998; dates 9– 16: September 1998–June 1999; dates 17–18: September–November 1998; date 19: April 1999. Each graduation on the x-axis corre- sponds to a 2-mm interval, and graduations begin at 1 mm. 140

This result is not extremely surprising since canals The other environmental parameters included can dry out and sinks cannot. Water availability is in our analysis showed variation, but it is difficult indeed an important determinant of both the to clearly point out specific species – environ- population biology (Charbonnel et al., 2002) and mental parameters associations in FCAIVs. community ecology (Lodge et al., 1987) of aquatic Maybe the most significant results where that snails. Sinks and canals also differ in their physical B. truncatus was associated with high values of structure and ecological parameters. For example, algae and macrophyte (the reverse for M. similis) the water column is much higher in sinks than in and that P. acuta did not show very specific canals which affects water temperature. This dif- associations. Little is known about the physiologic ference might also explain why B. truncatus was requirements of these species, and it is therefore never recorded in canals. On the other hand, the difficult to offer mechanistic explanations to these occurrence of the three other species was positively results. However, those obtained in P. acuta are correlated between sinks and canals. These results consistent with the wide distribution of this spe- suggest that the relationship between sinks and cies, and the idea that it exhibits wide phenotypic canals is unilateral, sinks serving as source of plasticity (Henry, 2002). On the whole, although individuals to canal populations or communities some species are associated with environmental which cannot maintain themselves when canals are parameters, whether biotic or abiotic, no strong drying out. A second factor that might be impor- correlations was found, as in a similar snail com- tant is time, as it generally strongly affects the munity (Gallardo et al., 1994). On the other hand, abundance of freshwater snails, for example in significant association have been found by other relation with the life-cycle of species (see e.g. in authors (e.g. Leland & Fend, 1997; Lods-Crozet Brown, 1994; Dillon, 2000). The date factor was et al., 2001). It remains possible that a less crude indeed significant in the ANOVA on all data, but approach to environmental parameters would this might be accounted for by differences between have detected stronger associations. sinks and canals (see above). However it was not One environmental factor which influence we significant in sinks, in line with visual inspection of wish to discuss separately is cleaning of sinks; this time distributions which did not show marked is performed for controlling populations of vector differences. This was confirmed by cohort analysis snails occupying the irrigation area (see Laamrani of B. truncatus in the three sinks in which enough et al., 2000). Our results suggest that cleaning is data were gathered. An autumn cohort was punctually very efficient (populations of all species detected in one sink, but none in the two others. were almost completely eliminated in the sinks This suggests that there is no seasonal trend in the surveyed). However, populations of both demography of this species, in contrast to what is B. truncatus and P. acuta recovered quickly, and often found in snail populations under temperate their density even increased as a function of time to sub-tropical conditions (see Brown, 1994, Dil- after cleaning. One reason why they recover lon, 2000; Charbonnel et al., 2002). For example, quickly might be local survival in underground Goumghar et al. (2001) found two generations per pipes connecting sinks, followed by dispersal, year in Middle Atlas populations of L. truncatula. allowing fast recolonization (Chlyeh et al., 2002, A possible explanation of our result is the 2003). In terms of B. truncatus control, we previ- smoothing role of water temperature which ously suggested targeting only those large popu- exhibits little variation by comparison to air tem- lations serving as source of migrants (Chlyeh perature. We note also that winter temperatures et al., 2002, 2003). Results from the present study are not extremely low. In other words, snails of the indicate that targeting sinks harboring large four species studied might be reproductively active amounts of macrophytes and algae might consti- all year long. Data on the distribution of egg tute an appropriate strategy, since these factors are capsules would be useful here. Capsules of related to the distribution of B. truncatus. On the B. truncatus were indeed sampled and eggs coun- other hand, cleaning of sinks is associated with a ted; although their number was too low to allow lasting effect on populations of L. truncatula. This for statistical analysis, capsules were indeed found suggests more efficient control of this vector snail. year-round (Chlyeh, 2002). However this species also occupies leaking parts of 141 the irrigation area (outside sinks and canals), the role of extinction and colonization is in where concrete is damaged, from which recolon- principle possible, but would require much more ization is possible. Such sites can neither be easily extensive surveys in terms of number of sites detected, nor be cleaned. studied. (ii) Isolation by distance (IBD) was detected in sinks when using distances based on pre- sence/absence, but not with distances based on Acknowledgements frequency. No signal was detected when using data from canals. There is therefore limited evidence for This project was supported by funds from FICU IBD in the distribution of snail species. IBD is (99/PAS/29 to G.C.) and from bilateral exchange expected when migration among populations is a programs between Morocco CNR and France decreasing function of distance (see e.g. Clobert CNRS (PICS program). The authors thank T. et al., 2001; Rousset, 2001). A previous study Beqqali for drawing maps, D. El Ouardi and F. showed that strong migration occurs among sink M’siid for technical help, N. Berday for deter- populations of B. truncatus (Chlyeh et al., 2002). mining algae species, J.-D. Lebreton and A.-M. For example, the probability that a given indi- Reboulet for help with the statistical analyses, and vidual spend its whole life in a sink was only 0.3. an anonymous reviewer for very helpful com- Moreover there is limited evidence of population ments. genetic structure in this species: at best, two pop- ulations occur in the irrigation area studied (group 1, 2 and 3 vs. group 4; Chlyeh et al., 2002). Under the hypothesis that the colonization rate from References outside the irrigation area is negligible and that other species disperse in the same way as B. Boelee, E., 2003. Malaria in irrigated agriculture. Irrigation and truncatus, IBD should therefore not be observed. Drainage 52: 65–69. This is compatible with our results which might Borcard, D., P. Legendre & P. Drapeau, 1992. Partialling out suggest that migration strongly affects community the spatial component of ecological invasions. Ecology 73: 1045–1055. structure at the scale studied. From the discussion Brown, D. S., 1994. Freshwater Snails of Africa and Their in point (ii) above, it is unlikely that environmental Medical Importance. Taylor & Francis Ltd, London. parameters affect IBD through an homogenizing Charbonnel, N., M. Quesnoit, R. Rasatavonjizay, P. Bre´mond, effect. Another factor that might play is popula- C. Debain & P. Jarne, 2002. A spatial and temporal tion history. For example, M. similis which was approach of microevolutionary forces affecting population biology in the freshwater snail Biomphalaria pfeiffer. Amer- detected in populations of group 4 only might have ican Naturalist 160: 741–755. colonised the irrigation area only recently, and not Chlyeh, G., 2002. Dynamique et ge´ne´tique des populations have had enough time to colonise the whole area. de mollusques hoˆ tes interme´diaires de parasitoses eau- It is also likely that the number of sites studied was de´pendantes (bilharzioses et fascioloses). The` se de doctorat not large enough to unravel the influence of space d’e´tat, Institut Agronomique et Ve´te´rinaire Hassan II, Rabat, Maroc. 121 pp. and environmental parameters. A study conducted Chlyeh, G., P.-Y. Henry & P. Jarne, 2003. Spatial and temporal at several spatial scales would be useful in this variation of life-history traits documented using capture- respect, allowing to fully account for the influence mark-recapture methods in the vector snail Bulinus truncates. of space on the distribution of species (see Borcard Parasitology 127: 243–251. et al., 1992; Keitt et al., 2002). Chlyeh, G., P.-Y Henry, P. Sourrouille, B. Delay, K. Kha- llaayoune & P. Jarne, 2002. Population genetics and (iii) The low number of species detected might dynamics at short spatial scale in Bulinus truncatus, the simply be constrained by the colonization rate of intermediate host of Schistosoma haematobium, in Morocco. the irrigation area from outside sources, but it Parasitology 125: 349–357. might also affect the spatio-temporal variation in Clobert, J., E. Danchin & A. DhondtNichols, (eds.) 2001. both occurrence of species and number of indi- Dispersal. Oxford University Press, Oxford. Costil, K., G. B. J. Dussart & J. Daguzan, 2001. Biodiversity of viduals per species. Similarly, extinction of local aquatic gastropods in the Mont St. Michel basin (France) in populations, due to other factors than snail con- relation to salinity and drying of habitats. Biodiversity and trol, might affect the local dynamics. Assessing Conservation 10: 1–18. 142

David, P., P. Berthou, P. Noel & P. Jarne, 1997. Patchy Lebreton, J.-D., M. Roux, G. Banco & A.-M. Bacou, 1990. recruitment patterns in marine invertebrates: a spatial test of Biomeco (Biome´trie-Ecologie), version 4.2. Software of the density-dependent hypothesis in the bivalve Spisula Statistical Ecology for PC. CEFE/CNRS. Montpellier, ovalis. Oecologia 111: 331–340. France. Dillon, R. T., 2000. The Ecology of Freshwater Molluscs. Leland, H. V. & S. V. Fend, 1997. Benthic invertebrate distri- Cambridge University Press, Cambridge. butions in the San Joaquim river, California, in relation to Ebert, T. A., 1999. Plant and Populations – Methods in physical and chemical factors. Canadian Journal of Fisheries demography. Academic Press, San Diego. and Aquatic Sciences 55: 1051–1067. Gallardo, A., J. Prenda & A. Pujante, 1994. Influence of some Lewis, D. B. & J. J. Magnuson, 2000. Landscape spatial pat- environmental factors on the freshwater macroinvertebrates terns in freshwater snail assemblage across Northern High- distribution in two adjacent river basins under mediterra- land catchments. Freshwater Biology 43: 409–420. nean climate. Archiv fu¨r Hydrobiologie 131: 449–463. Lodge, D. M., K. M. Brown, S. P. Klosiewski, R. A. Stein, Ghamizi, M., 1998. Les mollusques des eaux continentales du A. P. Covich, B. K. Leathers & C. Bronmark, 1987. Distri- Maroc: syste´matique, bioe´cologie et malacologie applique´e. bution of freshwater snails: spatial scale and the relative The` se de doctorat, Universite´de Marrakech, Maroc. 548 pp. importance of physicochemical and biotic factors. American Goumghar, M. D., P. Vignoles, D. Rondelaud, G. Dreyfuss & Malacological Bulletin 5: 73–84. M. Benlemlih, 2001. Relationships between the annual gen- Lods-Crozet, B., E. Castella, D. Cambin, C. Ilg, S. Knispel & erations of the snail Lymnaea truncatula ( Gastro- H. Mayor-Simeant, 2001. Macroinvertebrate community poda : Lymnaeidae), altitude, and the type of its habitats in structure in relation to environmental variables in a Swiss central Morocco. Revue de Me´decine Ve´te´rinaire 152: 457– glacial stream. Freshwater Biology 46: 1641–1661. 462. Madsen, H., 1982. Ecologie de gaste´ropodes. 1. Me´thodologie. Henry P.-Y., 2002. Dynamique des populations et e´volution du Danish Bilharziasis Laboratory. 49 pp. syste` me de reproduction. Etude chez le gaste´ropode her- Paavola, R., T. Muotka, R. Virtanen, J. Heino & P. Kreivi, maphrodite Physa acuta (pulmone´). The` se de doctorat, 2003. Are biological classifications of headwater streams Universite´Montpellier II, France. concordant across multiple taxonomic groups?. Freshwater Hubbell, S. P., 2001. The Unified Neutral Theory Of Biodi- Biology 48: 1912–1923. versity And Biogeography. Princeton University Press, Persat, H. & D. Chessel, 1989. Typologie de distributions en Princeton. classes de taille : inte´reˆ t dans l’e´tude des populations de Kanandjembo, A. N., M. E. Platell & I. C. Potter, 2001. The poissons et d’inverte´bre´s. Acta Oecologica 10: 178–195. benthic macroinvertebrate community of the upper reaches Raymond, M. & F. Rousset, 1995. GENEPOP (version 1.2): of an Australian estuary that undergoes marked seasonal population genetics software for exact tests and ecumeni- changes in hydrology. Hydrological Processes 15: 2481– cism. Journal of Heredity 86: 248–249. 2501. Reboulet, A.-M., P. Lespinasse & J.-D. Lebreton, 1995. PrAxis, Keitt, T. H., O. N Bjornstad, P. M. Dixon & S. Citron-Pousty, version 2.0. Software for Multivariate Data Analysis for PC. 2002. Accounting for spatial patterns when modeling CEFE / CNRS. Montpellier, France. organism-environment interactions. Ecography 25: 616–625. Rousset, F., 2001. Inferences from spatial population genetics. Khallaayoune, K. & H. Laamrani, 1992. Seasonal patterns in In Balding, D. M. Bishop, & C. Cannings (eds.) Handbook the transmission of Schistosoma haematobium in Attaouia, of Statistical Genetics. Wiley, Chichester: 239–269. Morocco. Journal of Helminthology 66: 89–95. Sall, J. & A. Lehman, 1996. JMP – Start statistics. Belmont, Laamrani, H., 1999. Environmental control of schistosomiasis USA. in irrigation schemes in Morocco. PhD thesis, University of Stewart, T. W., T. L. Shumaker & T. A. Radzio, 2002. Linear Copenhagen, Copenhagen, Denmark. 201 pp. and nonlinear effects of habitat structure on composition Laamrani, H., J. Mahjour, H. Madsen, K. Khallaayoune & B. and abundance in the macroinvertebrate community of a Gryseels, 2000. Schistosoma haematobium in Morocco: large river. American Midland Naturalist 149: 293–305. moving from control to elimination. Parasitology Today 16: Sturrock, R. F., O.-T. Diaw, M. Niang, J.-P. Piau & 257–260. A. Capron, 2001. Seasonality in the transmission of schis- Lebart, L., A. Morineau & M. Piron, 2000. Statistique Explo- tosomiasis and in populations of its snail intermediate hosts ratoire Multidimensionnelle, 3rd edn. Dunod, Paris. in and around a sugar irrigation scheme at Richard Toll, Lebreton, J.-D., D. Chessel, R. Prodon & N. Yoccoz, 1988. Senegal. Parasitology 123: S77–S89. L’analyse des relations espe` ces-milieux par l’analyse cano- Ter Braak, C. J. F., 1986. Canonical correspondence analysis: a nique des correspondances. Acta Oecologica – Oecologia new eigenvector technique for multivariate direct gradient Generalis 9: 53–67. analysis. Ecology 69: 69–77.