The results of our study demonstrate that the mature reservoir system on the Ruhr River in the west of Germany provides an excellent environment for the development of species-rich and abundant trematode communities in the first intermediate hosts and this was in line with our expectations based on bird abundance. The purely faunistic by-product is nevertheless striking since the total of 15 species of trematodes we detected as a result of a limited sampling of the two mollusc hosts represents half the diversity registered by Faltýnková & Haas  in more than 6,000 molluscs belonging to 28 species studied in the southeast of Germany. Furthermore, the faunal richness we observed in the four Ruhr River reservoirs is similar to the 17 species recorded by these authors in the Aischgrund lowland area. Although their data are pooled among the 26 localities sampled and thus not directly comparable to our results based on distinct snail population samples, another remarkable difference is worth noting. Faltýnková & Haas  registered a poorer trematode fauna in R. auricularia compared with L. stagnalis (4 vs 10 species). Faltýnková  and Żbikowska  reported a similar relationship (3 vs 10 spp. and 1 vs 13 spp., respectively) in small fish ponds near Ceske Budejovice (Czech Republic) and a variety of water bodies in Poland. On the contrary, we observed the reverse faunal richness: 12 vs 6 species in R. auricularia and L. stagnalis, respectively. These data tend to support the hypothesis of Wesenberg-Lund  that in lakes R. auricularia plays a role in the life-cycles of trematodes similar to that of L. stagnalis in ponds. In contrast to Wesenberg-Lund's suggestion, Adam & Lewis  reported high trematode richness (11 spp.) in a population of R. auricularia during the process of its colonisation of a gravel-pit near Wraysbury (UK), whereas the sparse populations of L. stagnalis studied in the same area were uninfected. However, the sample sizes of the two snail hosts were not comparable.
Concerning the rates of infection, although data for Asian populations of R. auricularia indicate that parasitism may affect large proportions of the snail population (e.g. overall prevalence 18.2% at Lam Tsuen River, Hong Kong ), surprisingly little is known for levels of larval parasitism in European populations of this snail host. Adam & Lewis'  study depicting prevalence patterns in R. auricularia over three years in a single snail population appears so far unique. Notably, their data show high monthly prevalences (typically above 20%) occasionally peaking up to c. 60% (in at least three out of 11 distinct monthly samples, estimated from Figure 1 in Adam & Lewis ) in association with the high prevalence of the bird parasite E. recurvatum (> 30%). Although overall prevalences observed in communities in R. auricularia in the Ruhr reservoirs showed a wider range of variation there were important infection foci, e.g. sites B3, He2 and He3, the latter two showing extremely high infection rates. In contrast, the overall prevalences in L. stagnalis (max. 14%) were low compared with those for both R. auricularia (; this study) and communities of L. stagnalis studied in the same season in two small fishponds in the Czech Republic (up to 53 and 65%, Kostadinova, unpublished data). Thus, our data tend to support the "reverse role" hypothesis of Wesenberg-Lund  for the two snail hosts in lakes.
There were clear differences in snail population densities associated with reservoir characteristics. Thus the 'large-lake adapted' species R. auricularia dominated over L. stagnalis especially at Hengsteysee where the macrophyte cover at the sampling sites was poor and all R. auricularia were found aggregated on stones. The spread and abundance of R. auricularia in the studied reservoir system could also be related to the fact that Radix spp. in general can withstand severe water fluctuation . Interestingly, L. stagnalis is considered a "calciphile species" not normally found reproducing in waters with less than 20 mg/L calcium . This sensitivity to external calcium concentration can explain the distinctly higher density of this species at Baldeneysee which is characterised by highest mean calcium content of the sediments (10.5 mg/kg vs slightly above 4.0 mg/kg at Hengsteysee and less than 4.0 mg/kg in the other two reservoirs ) and the higher abundance of the more "softwater" R. auricularia at Hengsteysee. Further, although all studied reservoirs show backwater effects as evidenced by the perpetual removal of sediment deposits during the last 20 years, Baldeneysee is characterised by the lowest water flow velocity both in spring and summer (0.07 and 0 m/s, respectively ), thus providing a range of lentic habitats suitable for L. stagnalis.
Our ability to detect parasites with low prevalence decreases at low population densities since sample sizes also decrease and this is particularly true for mollusc-parasite systems characterised by a remarkable consistency in prevalences (usually within the 5-10% range, see Esch et al. ). Although this has inevitably affected our observations, we believe that infection levels may well have reflected real variation among the Ruhr reservoirs and the logistic regression results tend to support this. To the best of our knowledge, this is the first study relating environmental variables with the outcomes of trematode transmission at the level of individual snails in a natural setting. In fact, studies on the impact on the survival and infectivity of parasite transmission stages of environmental variables have rarely been undertaken in laboratory and then only in the case of a few trematodes (reviewed by Pietrock & Marcogliese ). In addition to the well-known strong association between infection and snail size (reviewed in ), our results indicate that oxygen content may play an important role for trematode survival and/or infectivity especially in areas with poor water column oxygen conditions (e.g. L. stagnalis dataset, predominantly sampled at Baldeneysee) and, therefore, be of importance for the distribution of larval trematode infections among reservoirs which differed significantly with respect to this environmental variable. The strong correlation between oxygen content and conductivity explains the exclusion of the latter variable from the models. Further, although hydrogen ion concentration varied within a narrow range, pH was also identified as an important factor with a negative effect on the probability of infection.
We expected a substantial homogenisation of trematode distribution accross reservoirs which would reflect an even distribution of the limited number of bird species at the relatively small spatial scale of the study (Table 1, Additional file 1). However, we observed at Baldeneysee and Hengsteysee, in addition to differential snail densities, the highest trematode richness and this was in sharp contrast with the other two reservoirs. Trematode diversity and prevalence are directly associated with bird diversity and abundance (e.g.[30–32]) and both bird species richness and maximum densities increase with nutrient load . Therefore, the diversity and composition of larval trematodes in the bird-parasite dominated (87% of all species and 83% of the dominant species) snail systems studied by us appear to reflect an advanced eutrophication at Baldeneysee and Hengsteysee in line with the predictions for parasites in eutrophic water bodies in the classical study of Wisniewski . In support of our findings, Podraza et al.  considered the total phosphorus content of the sediments of Hengsteysee characteristic for eutrophic lakes and recorded more than twice as high levels at Baldeneysee. Furthermore, the two infection foci at Hengsteysee are located in a shallow stagnant bay (sites He2 and He3, see map in Additional file 1) affected predominantly by the inflow from the Lenne River which includes the inflow from a nearby wastewater treatment plant of a paper mill.
However, the two snail-trematode systems depicted different prevalence patterns among reservoirs. The sharp differences in the overall prevalence and the prevalence of three out of five dominant species in R. auricularia were in contrast with the more homogeneous overall prevalence between reservoirs in L. stagnalis. The pattern observed in the latter host-parasite system is probably associated with the substantially lower overall trematode richness and the higher prevalence of the only dominant species, D. pseudospathaceum in Harkortsee and Kemnader See. In spite of the variability observed, the domination of parasites maturing in birds with respect to both richness and prevalence, is a common feature of the two snail-trematode systems studied in the Ruhr reservoirs. Although we have selected a very high threshold prevalence value for dominance status, the number of dominant species appears unprecedented in view of the typically low prevalences of larval trematodes in molluscs .
Remarkably, the life-histories of five out of the six dominant species appear to depict two aspects of progressive eutrophication in this mature reservoir system. First, the relatively homogenous prevalences of the two Diplostomum spp. completing their life-cycles in gulls and the specific parasite of cormorants P. radiatum, indicate increased densities of their most suitable intermediate hosts, cyprinids, which are characteristic for eutrophic conditions. Although no records are available for P. radiatum, increased infections of Diplostomum spp. in fish were related to progressive eutrophication in the Lake Constance (Germany) . The rates of infection with the three species associated with cyprinids and fish-eating birds in the Ruhr reservoirs studied are within the upper ranges reported so far (typically from pond systems; see [9, 21, 36, 37]).
The high infection rates of the second group of dominant species, E. recurvatum and T. franki, which mature in anatid birds, suggest parasite proliferation associated with another aspect of lake eutrophication, i.e. the dominance of anatids and Anas platyrhynchos in particular . High eutrophication of water reservoirs, accompanied by colonisation by snails and nesting ducks, is considered as one of the most important factors for the increase of outbreaks of cercarial dermatitis worldwide [38, 39]. Although Trichobilharzia spp. infection rates in lymnaeids are typically low, ranging between 0.3 and 5.2% in Europe , two of the most diverse sites in our study were also important foci of T. franki showing prevalences as high as the maxima recorded (22% and 26% in the Czech Republic  and South Germany , respectively). It has been suggested that these high infection rates may have resulted from disturbances in the ecological balance of the lakes leading to increased snail densities [39, 42] and our data tend to support this. An added factor in our system is the presence of dense mats of E. nuttallii on water surface attracting large numbers of plant-grazing anatids and cyprinids and thus facilitating trematode transmission (see bird aggregations outlined in Additional file 1). It is worth noting that cases of cercarial dermatitis have been recorded at Hengsteysee.
One important result of our study is that component community composition and structure provided evidence that larval trematodes in R. auricularia, and to a much lesser degree in L. stagnalis, may have reflected spatial bird aggregations on the small-scale within-reservoir study especially with respect to the focal occurrence of parasites. Thus, at least three foci exist in Baldeneysee and Hengsteysee that support diverse communities involving up to four different groups of vertebrate final hosts (Table 4). The small-scale patterns of variation of component communities in the two snail hosts also indicated differentiation between host-parasite associations in the two study reservoirs. Thus, infection levels in both snail hosts from Baldeneysee were somewhat homogenous and this was in contrast with the substantial variation in communities in R. auricularia at Hengsteysee; unfortunately no representative samples were available for L. stagnalis from the latter. One possible explanation is that the larger size of Baldeneysee levels down the environmental instability e.g. water level and temperature variation which is significant at Hengsteysee as it serves as a lower reservoir for a pumped-storage hydropower plant. The bird sanctuary Heisingen at Baldeneysee also supports important numbers of breeding birds. These factors in combination may tend to homogenise trematode communities in molluscs. We are tempted to suggest that the contrasting patterns of variation in communities in R. auricularia may be associated with the more advanced eutrophication at Baldeneysee thus offerring support for another insightful prediction of Wisniewski i.e. that the distribution of hosts and parasites in an eutrophic water body "leads, as a rule, to the possibility of the hosts being invaded in the same degree at all points at the lake". A wider small-scale sampling at the four Ruhr reservoirs would be influential in testing this prediction.