Introduction

Water dams represent barriers on streams that attenuate seasonal and inter-annual hydrological fluctuation in river floodplains, thereby altering the natural dynamics of an entire riverine landscape (Poff et al. 2007). The alteration of flow regimes is often claimed to be the heaviest threat to ecological equilibrium in rivers and their associated floodplain wetlands (Naiman et al. 1995; Sparks 1995; Lundqvist 1998; Ward et al. 1999). While the obvious and often irreversible impact of large impoundments on the longitudinal connectivity of river habitats is now well-known, there is also a growing awareness about the key role of the lateral connectivity and flow regime dynamics as the fundamental drivers maintaining the natural functioning riverine ecosystem (Junk et al. 1989; Poff et al. 1997; Richter et al. 1997; Puckridge et al. 1998; Hart and Finelli 1999; Poff et al. 1997; Bunn and Arthington 2002). Flow dynamics provide specific structure of riverine landscape (Scott et al. 1997), creating mosaics of contrasted patches with different levels of connectivity to the stream and disturbance regimes (Tockner et al. 1999; Ward 1998; Ward and Tockner 2001). Natural flood disturbances are an integral component of active floodplain ecosystems (Ward 1998). It is well-known that artificially altering flow regimes negatively affects the biodiversity not only in the stream but also in the entire associated riverine landscape. (Hart and Finelli 1999; Adis and Junk 2002; Poff and Zimmermann 2009). For the key environmental determinants that affect the taxonomical and functional composition of the floodplain, terrestrial mollusks are considered flow regime and its extremes (flooding and scouring), groundwater level and soil moisture, as well as local climate conditions (Čejka et al. 2008). River floodplains are therefore colonized by species that are well-adapted to the disturbances represented by flooding events and the fluctuation in soil moisture (Henle et al. 2006). Due to low active mobility, mollusks only slowly recolonize lost habitats, and therefore, the mollusk community composition is determined by both current and past habitat conditions (Wallace 1990; Castella et al. 1994).

For the last 150 years, the human impact on the Danube River has had an increasing influence on its drainage basins and their watercourses. In Slovakia, the regulated Danube as it is known today, is essentially the result of a complex system for river regulation, designed by E. Lanfranconi at the end of the nineteenth century. The main objective of the regulation was to improve navigation conditions and to facilitate the quickest ice break-up in order to protect adjacent areas against ice floods (Pišút and Timár 2007). This intervention led to the end of free hydrological processes and consequently initiated a chain of far-reaching changes that led to the construction of the Gabčíkovo barrage system (1977–1992). The hydrological regime of the Danube River was further altered by the operation of a hydroelectric power project—the Gabčíkovo Waterworks, which went into operation in October 1992. It was predicted that this large engineering system on the Danube (WWF 1997) would have a substantial, long-lasting impact on the aquatic and terrestrial environment of the entire Danube inland delta (Holčík et al. 1981). Thus, in order to evaluate the impact of the Gabčíkovo Waterworks on the Danube floodplain ecosystem, long-term environmental monitoring was started in 1990. Several papers have been published in the past on the results of long-term monitoring of other terrestrial biota (e.g., Kalúz 1994; Krumpálová 1997; Petrášová-Šibíková et al. 2017; Šustek 1995). Monitoring of terrestrial molluscan fauna is an integral part of this complex environmental project, but the comprehensive results for more than 20 years of monitoring have not yet been published. In this study, we aimed (1) to determine the trajectory in changes of the mollusk community composition in the area impacted by the Gabčíkovo barrage system; (2) to identify key events in succession of species richness and taxonomic and functional composition of the land snail assemblages; (3) to characterize correlation between changes of land snail community and the alterations of the soil moisture as the consequence of the Gabčíkovo Waterworks operation.

Methods

Study area and sites

The study area extends from Bratislava to the village of Kľúčovec in southwest Slovakia, along the river line between the coordinates 48.058509°N, 17.166943°E and 47.785831°N, 17.678931°E (Fig. 1). The sites represent two of the most common and clearly recognized plant associations (Jurko 1958a): (1) the softwood floodplain forests of the Salici-Populetum association, (2) the hardwood floodplain forests of the Fraxino pannonicae-Ulmetum.

Fig. 1
figure 1

Map of the Gabčíkovo barrage system with monitored sites indicated

Full size image

Monitoring plots (MP) were situated (1) above the by-passed section (MP_2, MP_3)—sites not impacted by a decreasing level in groundwater. The level of groundwater mainly increased there after the Čunovo reservoir was filled; (2) inside the by-passed section (MP_6, MP_9, MP_10, MP_14)—monitoring plots under the influence of decreased groundwater level (mostly by a so-called drainage effect caused by the former main channel of the river); (3) beyond the dam operation impact (reference site, MP_18)—it is located downstream below the tailrace channel (Table 1).

Table 1 Monitoring plots and their characteristics
Full size table

Sampling method

Sampling was carried out seasonally, three times per year (spring, summer, autumn) in the years 1991–2014. Each monitoring plot (ca. 400-m2 area) was chosen to represent a typical part of the surrounding vegetation, without edges or a transition zone to other habitats. At each plot, four quadrants (area of 25 × 25 cm) were sampled. From the quadrants, all litter, twigs, vegetation, and loose soil (ca. up to 5 cm) were sieved through a sieve with a mesh size of 10 mm. Coarse fraction was also removed in the field. The samples from the sieve were bagged and taken to the laboratory (for details concerning the extraction of mollusks from litter samples, see Čejka et al. 2008). In the spring and autumn, the samples were enriched by snail samples taken by individual sampling that lasted for 1 h (Cameron and Pokryszko 2005). During the entire monitoring period, a total of 23 samples were obtained from each site. Thus, each sample from 1 year consisted of 12 pooled quadrants with a total area of 0.75 m2 and a collection representing two person-hours. Soil moisture was measured by the neutron probe method approximately once every 14 days (Matečný and Bedrna 2014). Measurements were taken at a depth of 0.1 m at sites MP_6, MP_9, MP_10, MP_14, and MP_18 from the year 1990, and at sites MP_2 and MP_3 from the year 1995.

Data analysis

The trend lines of change in species richness—alpha, beta, gamma diversity (Whittaker 1972) throughout the entire monitoring period were modeled by simple regression analysis. A Bray-Curtis dissimilarity matrix based on log(x + 1) transformed abundance was used in metric multidimensional scaling (MDS) for identification of position site centroids, as well as to evaluate the variability in the molluscan community composition at each sampling site. For testing of homogeneity of multivariate dispersions, a permutation test was used (number of permutations = 999) with post hoc pairwise comparisons with Bonferroni-corrected p values. In addition, Constrained Analysis of Principal Coordinates (CAP) based on Bray-Curtis similarity matrices and mean annual values of the soil moisture at a depth of 0.1 m was employed to create a two-dimensional map of the temporal shift of the community at each monitoring site. Generalized additive models with a thin plate spline were used to predict and plot the surface of the soil moisture gradient on the CAP ordination maps. For each species, soil moisture optimum was determined (weighted mean ± bootstrapped 95% confidence interval), moisture tolerance (weighted mean ± standard deviation), and coefficient of variation (CV). Species trait composition was analyzed using the fuzzy principal component analysis (FPCA) on the trait × site matrix. The trait matrix consisted of three ecological species traits, i.e., habitat specialization (forest specialist, forest generalist, open habitat preferred, indifferent), life history strategy (k/r selected species), and soil moisture preference (euhygric, mesohygric, and polyhygric species and species with dry habitat preference) (Čejka and Hamerlík 2009; Lisický 1991). The ordinations were performed using the “ADE4” (Dray and Dufour 2007) and “Vegan 2.0.10” (Oksanen et al. 2012) packages in the R 3.1.0 (R Core Team 2014) software environment. The bootstrapped 95% confidence interval for the weighted mean was performed using “boot” package (Canty and Ripley 2016) in the R 3.1.0 (R Core Team 2014) software environment.

Results

In total, 39 species were recorded at 7 monitoring plots from 1991 to 2014. Forest hygrophilous species (Arianta arbustorum, Monachoides incarnatus, Cochlicopa lubrica, Fruticicola fruticum, and Cochlodina laminata) with a preference of mesohygric conditions were most frequent at the study area during the entire study period. Local diversity of molluscan communities had a very similar decreasing trend at the plots above (MP_2, MP_3) and inside (MP_6) the by-passed section, where the highest species loss was observed between 1991 and 1995 (Fig. 2). Just the opposite trend in molluscan diversity was observed at the other plots of by-passed section (MP_9, MP_10, MP_14), where the increase in species number occurred between 2003 and 2004. The values of global and between habitats diversity were almost constant during the entire monitored period, similarly as it was at the reference site MP_18.

Fig. 2
figure 2

Changes of molluscan species richness (local, between-habitat, and global diversity) in the Danube floodplain over the studied period (1991–2014). Diversity trend lines and their statistical significance are shown. Grey box indicates the sites above the by-passed section, black box indicates the sites within a by-passed section, white box indicates the site out of the dam operation impact

Full size image

The ordination maps of the non-metric multidimensional scaling (nMDS) are shown in Fig. 3, where the centroids and temporal variation of molluscan community at each sampling site are displayed. The position of site centroids at the first nMDS axis clearly indicates the difference in molluscan community composition between the sites, both inside and outside the by-passed section. In the permutation test, we identified significant differences in the homogeneity of multivariate dispersions between the sites (F = 2.87, permutation number = 999, p < 0.05). During the studied period, the higher values of the multivariate dispersion (lower similarity of species composition) was recorded at the sites situated inside the by-passed section, especially at the sites MP_9 and MP_14 (Table 2). The CAP ordination of the snail communities produced reasonable ordinations with explained variation at the first ordination axis representing the soil moisture gradient from 5.7 to 13.5% (Fig. 4). At the monitoring plots situated upstream of the by-pass section (MP_2, MP_3), a gradual increase of soil moisture occurred, while at the plots inside the by-pass section (MP_9, MP_10, MP_14), a clear gradual decrease of soil moisture linked to the operation of the waterworks was observed. Changes in soil moisture were reflected in a marked change of the mollusk taxonomical composition, especially in the first 5 years after the waterworks went into operation. In Table 3, moisture optimum and tolerance for each species are shown. Results of the FPCA showed that the gradients reflected in the first two axes represented between 81.9 and 93.4% of total variability in the molluscan community composition at the studied sites (Fig. 5). At the plots situated outside the by-passed section (MP_2, MP_3), there was an obvious shift from the community created by euryhygric species with an open habitat preference, through a community of polyhygric forest specialists to a community with a dominance of mesohygric to xeric forest generalists with r-selection life strategy (e.g., Alinda biplicata, C. laminata, Helix pomatia) was observed. At plot MP_6 situated inside the by-passed section and out of reach of the artificial flooding, a rapid one-way change of the community occurred, i.e., only a few years after the construction of the Gabčíkovo barrier, the original polyhygric community with species preferring opened habitats changed to community predominantly created by forest generalists that prefer a dry habitat. A different trajectory of the community succession occurred at two sites (MP_9, MP_10) inside the by-passed section, where the original polyhygric community with a dominance of k-strategists changed to meso- and euryhygric community with a dominance of forest specialists in the second half of the 1990s. In 1999, a dominance of polyhygric species was even observed. During the following decade, a significant shift to a xeric community with dominance of species preferring open and dry habitat occurred. The return to mesohygric community with a dominance of forest generalists was observed again from 2010. This turnover was determined mainly by increase in the abundance of Trochulus hispidus, Trochulus striolatus, Fruticicola fruticum, Cochlicopa lubrica, Zonitoides nitidus. A different trend in the land snail community was observed at site MP_14, where a polyhygric community with higher abundance of species preferring open habitats changed to mesohygric with a dominance of forest specialists between 2000 and 2004. In the following period, a gradual increase in the abundance of forest generalists preferring dry condition (e.g., A. biplicata, C. laminata, Vallonia costata) was observed. Nevertheless, only slight change of soil moisture was observed at reference site MP_18, the community succession occurred there, but it had a totally different trend as at the previous sites. At the beginning of the monitoring period (1991–1995), species with a preference to euryhygric conditions and open habitats dominated. In the following years, a succession passed through the community with a dominance of forest generalists preferring dry conditions (period 1996–1999) to community created by meso- to polyhygric species inhabiting opened habitat (period 2000–2014).

Fig. 3
figure 3

nMDS (non-metric multidimensional scaling) ordination plots based on log-transformed quantitative species data. The centroid positions and temporal variation of community are shown at each studied site. The ellipses represent confidence intervals for multivariate dispersion at each site. Grey box indicates the sites above the by-passed section, black box indicates the sites within a by-passed section, white box indicates the site out of the dam operation impact

Full size image
Table 2 Pairwise comparisons of homogeneity of the multivariate dispersions based on temporal changes of molluscan communities at selected sites (permuted p value below diagonal)
Full size table
Fig. 4
figure 4

CAP ordination maps of molluscan communities at 7 studied sites over a 23-year period. Contour lines display the smooth surface of soil moisture (in 10 cm) fitted using generalized additive models (GAMs). Grey box indicates the sites above the by-passed section, black box indicates the sites within a by-passed section, white box indicates the site out of the dam operation impact. Abbreviations of taxa names are given in Table 3

Full size image
Table 3 Species preference and tolerance to the soil moisture
Full size table
Fig. 5
figure 5

Fuzzy principal component analysis of molluscan community changes based on three ecological species traits at selected sites over the studied period. Grey box indicates the sites above the by-passed section, black box indicates the sites within a by-passed section, white box indicates the site out of the dam operation impact. Curve represents the direction of successional changes of community (based on Euclidean distance) to the point where the change is the highest compared to previous break point

Full size image

Discussion

At six monitoring plot, we recorded 55% of the entire terrestrial species pool known from the Slovak part of the Danubian Plain (71 species, Čejka 2019). The relatively low number of species compared with Čejka et al. (2008) was caused by the absence of some semi-synanthropic species (e.g., Arion distinctus (Mabille, 1868), A. fasciatus (Nilsson, 1823), A. vulgaris (Mabille, 1868)) and some xeric/xenocoenous species (e.g. Cecilioides acicula (O. F. Müller, 1774), Chondrula tridens (O. F. Müller, 1774), Cochlicopa lubricella (Porro, 1858), Granaria frumentum (Draparnaud, 1801), Helicopsis striata (O. F. Müller, 1774) and Xerolenta obvia (Menke, 1828)), as well as some rare polyhygric species (e.g., Vallonia enniensis (Gredler, 1856) and Vertigo moulinsiana (Dupuy, 1849)) in the monitored floodplain area. Over the last 50 years, a number of studies dealing with molluscan fauna have been published from the floodplains of European rivers (e.g., Frank 1984; Obrdlík et al. 1995; Horsák 2000; Ilg et al. 2009; Horáčková et al. 2014). Most of these studies is only descriptive, without any findings of the key environmental drivers explaining the current state of diversity and taxonomical composition of the molluscan fauna.

Soil moisture is one of the most important environmental factors that affect the local diversity of soil fauna (Silvan et al. 2000; Morecroft et al. 2002) and has often been considered as the key determinant responsible for the differences in taxonomical and functional richness of land snail assemblages between alluvial habitats (Wäreborn 1969; Gleich and Gilbert 1976; Wardhaugh 1995; Martin and Sommer 2004; Horáčková et al. 2014). Some of the previous studies documented a significant response of land snail diversity to the change in soil moisture, i.e., increasing species richness in wetter habitats (Wäreborn, 1969; Martin and Sommer 2004). Despite several adaptations to habitat drying (e.g., estivation in the shell, production of temporary epiphragma, or mechanical burrowing into the top layers of soil), most species preferred a certain range of soil moisture (Čejka and Hamerlík 2009; Horáčková et al. 2014). However, it appears that this relationship is habitat-specific and probably relates also to synergic effects of soil moisture with other environmental variables important for snails as well, such as vegetation type, soil pH, and calcium content in topsoil. The poorest mollusk assemblages usually occur in the alder and willow-poplar softwood forests (Horáčková et al. 2014; Čejka et al. 2008) at the beginning of succession phases, representing the moistest and most flood-affected biotopes in river floodplains. On the contrary, the richest land snail fauna inhabits the terminal stages of softwood and transitional forests. A slight decrease of mollusk diversity is observed towards drier hardwood forests. Martin and Sommer (2004) found significant increase of species richness and total density of malacocenosis with increasing soil alkalinity only in intermediate moist and the wettest forest habitats. No significant relationship was also found between soil pH and mollusk diversity in dry grassland habitats (Martin and Sommer 2004). Nevertheless, the previous studies analyzed change of mollusk diversity at the broad scale of the soil moisture; taxonomic composition of molluscan fauna reflects also fine changes in this parameter (Hettenbergerová et al. 2013). From the point of view of the Gabčikovo waterworks impact on the soil moisture conditions in Danube floodplain, no gradual changes in soil moisture occurred at the monitoring plots situated upstream of the by-pass section (MP_2, MP_3); while at the plots inside the by-pass section (MP_9, MP_10, MP_14), a clear gradual decrease of groundwater level and soil moisture was observed. At the monitoring plots upstream of the by-pass section, land snail assemblages consisted of just small populations of polyhygric species (Carychium minimum, Z. nitidus, Succinea putris) during the first few years after completion of construction and operation of the waterworks. Those populations quickly disappeared after a dry period between 1999 and 2001 (Zuzulová and Šiška 2017). Although the groundwater level in the area increased by 2 to 3 m after the Danube damming and filling of the reservoir (Matečný and Bedrna 2014), the former hygrophilic communities had previously been extinct there. Despite the re-increase of groundwater level, no close immigration sources for the possibility of either passive (drifting by floods) or active spreading of autochthonous gastropod populations had occurred there (Čejka 2006). At MP_6 situated inside the by-pass section and out of reach artificial floods, the level of groundwater before the construction of the waterworks had fluctuated between 0.5–2.5 m below ground level (Hlavatý et al. 1999). After the construction of the waterworks, the level of groundwater was in depth of 1.5–2.5 m, and due to the soil profile occurred there, the capillary water reached the topsoil only at a high groundwater level (Matečný and Bedrna 2014). After 1993, the value of topsoil moisture mainly depended in precipitation quantity (Jedlička et al. 1999). This fact was fully reflected in the drying up of the site and destruction of wetland habitats during the drier period in 1997 to 1999 (Zuzulová and Šiška 2017). During the whole monitored period, the molluscan community had been gradually changed there from a polyhygrophilic type (with key species, e.g., S. putris, C. minimum, and Vitrea crystallina) to the mesohygrophilic type with dominance of forest generalist species. (e.g., Aegopinella nitens and M. incarnatus). The succession was also observed in vegetation type there, when the originally softwood floodplain forests had showed a hint of transition to hardwood floodplain forests with significant decrease of hydrophytic plant species (Matečný and Bedrna 2014). Taxonomical and functional change of land snail communities in relation to the lack of surface flooding and decrease of the soil moisture was documented at the monitoring plots situated in the by-passed section (MP_9, 10, and 14). After the waterworks operation, the highest decrease of soil moisture was observed at the site MP_9, that was the consequence of the so-called drainage effect of the former Danube river bed (Jedlička et al. 1999). In the first few years after construction, the rapid soil moisture decrease was linked by the increase of number of the mesohygric forest specialists, less demanding on higher values of soil moisture (e.g., Clausilia pumila, A. arbustorum, and T. striolatus), while polyhygrophilic species (e.g., C. minimum, S. putris and Z. nitidus) totally disappeared there. After 1999, when regular artificial flooding was stopped, gradual drying up of this site was reflected in diversity and abundances of xerophilous snail species. At the monitoring plot MP_10, the absence of large floods from 1991 resulted in strong increase of abundance of the mesohygric forest snails. This type of the snail community was probably kept through periodic simulated floods of branch system (Table 4). Since 1999, the inland delta has been flooded irregularly (Krno et al. 2018), which resulted in strong drying up of this site and dominance of xerophilous forest generalists in the next years. At MP_14, originally polyhygric snail community gradually moved towards the dry type, typical of drier types of soft to transient floodplain forests with snail species like A. biplicata, C. laminate, and V. costata. Until 1992, the MP_14 belonged to the strongly hygric habitats with softwood floodplain forest (Jurko 1958b). After the Danube damming, the shallow depressions filled by water were reduced and succession process started toward xeroseries there (Lisický 1995). The extensive clear cutting of the poplar monocultures also contributed to the drying up of this habitat (Jedlička et al. 1999). At MP_18, which was situated downstream from the tailrace channel, changes in the structure of the land snail community was also observed. According to the hydrological and climatic conditions, a succession of land snail community passed from a polyhygric to xeric type in the dry period between 1997 and 1999 (Zuzulová and Šiška 2017), followed by a return to the polyhygric type again. According to Ward (1998), a repetitive succession of the communities is the fundamental attribute of healthy and functional floodplain ecosystems in a riverine landscape.

Table 4 Changes of hydrological regime in the Danube River in years 1990–2014 after putting Gabčíkovo Waterworks into operation. Data received from the Slovak Hydrometeorological Institute, Bratislava
Full size table

Conclusions

Construction of the Gabčíkovo Waterworks, which is similar to other river regulations (Rood et al. 1994; Kopeć et al. 2014; Vale et al. 2015), caused a huge impact on the dynamic and natural succession of the floodplain environment. Alteration in soil moisture caused by the Gabčíkovo Waterworks operation led to the slow degradation of former hygric habitats and resulted in significant changes of taxonomical and functional structure of the land snail community, i.e., a decrease in the number of polyhygric softwood floodplain forest species and an increase in diversity and density of generalists that prefer dry biotopes. Our results also showed that artificially induced and repetitive floods can partially stop the gradual degradation of originally hygric habitats and polyhygric communities in the Danube inland delta, and can even lead to their partial recovery as was shown at sites MP_9 and MP_10 in the late 1990s. We also found that supplying only isolated localities with a recharge system is insufficient for the preservation of original species composition. For more effective protection of the ecosystem of softwood floodplain forests, application of supply systems in extensive areas would be appropriate.