© A. Szlauer-Łukaszewska et al., Hosted by EDP Sciences 2021

1 Introduction

Environmental factors are crucial in shaping assemblages of animals, including aquatic invertebrates. The invertebrate community structure is a result of a continuous sorting process through environmental filters ranging from regional or catchment-wide processes, involving speciation, geological history and climate, to small-scale characteristics of individual patches, such as the local predation risk, substratum porosity and current velocity (Malmqvist, 2002). Lowland springs are often characterized by stable habitats, with modest changes in hydrological, thermal, chemical, and biological characteristics (De Luca et al., 2014; Rossetti et al., 2020). However, they may be affected by destabilizing factors, such as drought, flooding, and human impact (Rossetti et al., 2005, 2020; Pieri et al., 2007; Pakulnicka et al., 2016; Zawal et al., 2018). Individual springs can support a mosaic of microhabitats which results in a high species richness, for which reason springs are sometimes referred to as biodiversity hotspots (Cantonati et al., 2012; Rosati et al., 2014). Springs can be considered distributional islands due to their spatial isolation, with dispersal between springs expected to be limited (Stutz et al., 2010; Rosati et al., 2014). Despite the wide range of environmental factors which can be taken into account when studying springs in the context of ostracod assemblage structure controls, few researchers have undertakes a holistic ecological study. Examples include Külköylüoğlu and Yılmaz (2006) as well as Sarı and Külköylüoğlu (2008) who considered only physical and chemical water properties as environmental factors. In contrast, the holistic approach is represented by Rosati et al. (2014), Rossetti et al. (2005), Zhai et al. (2015), Mezquita et al. (1999), as well as Roca and Baltanás (1993). The research potential of springs situated in the valley of the River Krąpiel has been noted by Pakulnicka et al. (2016) who studied aquatic beetles, by Zawal et al. (2018) who investigated ecological determinants of the water mite occurrence, and by Bankowska et al. (2015) who discovered new and rare water mite species.

The traditional classification of springs is based on flow velocity at the source: in limnocrenic springs the water forms a nearly still pool, with the water flowing out at the bottom. Helocrenic springs form a swampy zone, with the water flowing out − relatively sluggishly − over a relatively large area, where as in rheocrenic springs the water flows out rapidly, at a single point (Steinmann, 1915; Thienemann, 1922). However, the distinction between different types of springs is not always so clear, intermediate types being frequent. The nature of a spring is strongly influenced by its surroundings. The presence of trees in the close vicinity of the spring greatly enriches it with organic material from fallen leaves and can cause strong shading, which may reduce or even eliminate spring vegetation. Highly important is the presence of other water bodies from which fauna can migrate and thereby substantially alter the nature of the spring fauna (Pakulnicka et al., 2016; Stryjecki et al., 2016; Zawal et al., 2017). On the other hand, as small, isolated habitats, springs are highly susceptible to degradation associated with pollution, drainage, and land use in the catchment (Rossetti et al., 2005, 2020; Pieri et al., 2007). Springs in a river valley may be flooded by the river when it overflows, resulting in exchange of fauna between these habitats as well as in alteration of the spring water chemistry. The fauna of springs is also influenced by their location with respect to the river; a close proximity enhances migration of river fauna (Buczyńska et al., 2016; Zawal et al., 2018). Under the current climate change, the groundwater level is often lowered, causing some springs to become ephemeral and thus facilitating survival (e.g. via resting eggs or dispersal) of species adapted to such conditions.

The aim of this study was to determine the structure of ostracod assemblages in springs situated in the valley of a small lowland river, to identify environmental factors (and the extent of their influence) shaping these assemblages, and to investigate the potential of ostracods as indicators of spring habitats.

2 Materials and methods

2.1 Study sites

The River Krąpiel is a small, lowland river, about 60 km in length, located in north-western Poland. The study area and specifically the Krąpiel, as well as the water bodies in the river valley, were described by Stryjecki et al. (2016) and Zawal et al. (2017). The samples for the present study were taken from springs situated in the Krąpiel valley. Six sampling sites along the entire length of the valley, 4.5–15 km away from one another, at locations with the highest numbers of springs, were visited. The sites were designated Z1 (53_28010.6300N 15_21041.7900E), Z2 (53_27036.9700N 15_16033.200E), Z3 (53_27041.4700N 15_12022.9400E), Z4 (53_2106.400N 15_1105.2300E), Z5 (53_ 20029.5600N 15_9015.0400E), and Z6 (53_19058.1400N 15_7057.5400E) (Fig. 1). The springs at a given site were located less than 50 m apart and fed a common stream. The number of springs sampled depended on the habitat diversity at each particular locality; two springs were sample dat Z4, three at Z1, and four each at Z2, Z3, Z5 and Z6. For each spring, its type, distance from the river, depth, flow type (permanent vs temporary), predominant sediment type, surrounding and submerged vegetation, flooding status (present or absent), and dryness (drying up vs. not drying up) were recorded.

Fig. 1 Location of the sampling sites. (A) Rivers. (B) Lakes and fish ponds. (C) Forests. (D) Locations of springs (Z1–Z6).

2.2 Fauna sampling

In 2011, when the study was conducted, the Krąpiel water level was a long-term average, so in each season the valley was flooded or dry to an extent typical of that resulting from long-term observations (A. Zawal and A. Szlauer- Łukaszewska, pers. obs., 2008–2012). The samples were taken during and after flooding, except that no samples were collected if the spring was completely dried up. The samples were taken in May, July, and November 2011, but the springs were monitored throughout the seven-month period from May to November to find out whether they were flooded or dried up. To avoid destruction of the springs, which were very small, only a single sample was taken from an area of about 0.25 m² in each spring, using a hand net, at least at 3 spots in the spring to capture its microhabitat diversity. The spring surface area ranged from 1 to 10 m², with the predominance of smaller springs. A total of 42 samples were collected (one sample from each of the 14 springs × 3 sampling events). Due to thevery small surface area of the springs, the numbers of ostracods collected were unusually low. Therefore, the material should be regarded as representing a ‘general community’ rather than a statistical sample from this community, which justifies the data processing approach adopted. Therefore, extrapolating the findings to other locations should be done, if at all, with a great caution.

2.3 Environmental parameters

The following environmental variables characterizing the springs were measured: water temperature (temp., °C), insolation (insolati, %), water pH (pH), conductivity (cond., μS/cm), total hardness (hardness, mg CaCO₃/L), oxygen content (O₂, mg/L), ferric ions (Fe₃C, mg/L), phosphates (PO₃, mg/L), ammonia nitrogen (NH₄, mg/L), nitrate nitrogen (NO₃, mg/L), concentration of solids (mg/L), BOD₅ (O₂, mg/L),mineral sediment contribution (mineral, %), organic sediment contribution (organic, %), mean sediment grain size (M, mm), sediment sorting (W, mm), and density of aquatic vegetation (plants, on a scale from 0 to 5, where 0 indicates the complete absence of plants and 5 means total overgrowth). The water parameters, i.e. temperature, pH, electrolytic conductivity and dissolved oxygen content, were measured using an Elmetron CX-401 multiparametric sampling probe; the flow rate was determined with a SonTek acoustic FlowTracker flowmeter (m/s); BOD₅ was determined using Winkler's method; insolation was measured with a CEM DT-1309 light meter, and the remaining parameters were determined with a Slandi LF205 photometer. On each sampling occasion, three measurements were taken and the median was used for further analysis. The hydrological status of the Krąpiel valley was assessed by performing the standard River Habitat Survey (RHS) at each site (Z1-Z6), which ensures that the results can be compared with those of other studies (Szoszkiewicz and Gebler, 2012). For the purpose of this study, the RHS methodology was modified in that the assessments were performed along 100-m sections of the river rather than along the standard 500-m long ones. The following indices were calculated from the data (Szoszkiewicz and Gebler, 2012): river habitat modification (RHM), habitat modification score (HMS), river habitat quality (RHQ), and habitat quality assessment (HQA).

2.4 Data analysis

The data were analyzed with the STATISTICA ver. 10 INC StatSoft, PAST ver. 2.17c (Hammer et al., 2001), and CANOCO v. 4.5 (TerBraak and Šmilauer, 2002) software packages.

Diversity metrics, i.e. the number of taxa (S), dominance (D), density, Simpson index (1–D), Shannon index (H), and Buzas and Gibson's evenness index (e^H/s), were calculated for different types of springs. Significance of differences in the diversity metrics between the spring types was tested with the Kruskal-Wallis test (Sokal and Rohlf, 1995).

The non-metric multidimensional scaling (NMDS), with Simpson's similarity coefficient, was used to look for patterns in spring grouping and correlation in the species' occurrence. The data for the multivariate analyses were not transformed, and the taxa which appeared in the samples once or twice were excluded.

The CANOCO v. 4.5 software was used to analyse relationships between ostracod species composition and habitat variables. The Canonical Correspondence Analysis (CCA) (TerBraak, 1986) was run to analyse patterns of species distribution in relation to environmental variables; it was preceded bythe Detrended Correspondence Analysis (DCA) which defined the structure of the data (Jongman et al., 1987). The stepwise variable selection was used to determine the significance (p ≤ 0.05) of the effect of each environmental variable on the species diversity. The Monte-Carlo permutations test with 499 permutations was applied to determine which variables were significant above a given threshold. The proportion of variation in the ostracod composition explained by the type of spring, substrate type, and hydromorphological parameters was expressed as the ratio of the sum of all canonical eigenvalues to the value of the total variance (total inertia), converted to a percentage. The proportion of variation in the ostracod composition explained by individual variables was calculated from the ratio of Lambda A to the total variance (total inertia), converted to a percentage.

Spearman's rank correlation coefficient was used to express the extent of correlation between the abundance of individual ostracod taxa and the habitat type, chemical water variables, spring type, and substrate variables.

3 Results

The materials collected from the Krąpiel valley springs yielded a total of 34 ostracod taxa, including 30 species. The numbers of individuals in the springs were very low, averaging 100 per sample. The ostracods were dominated by juvenile Candona sp., juvenile Psychrodromus sp., and the eurytopic species Cypria ophtalmica and Cypridopsis vidua. The high proportion of Cyprois marginata was due to its extremely high numbers in just two springs in samples collected in May (Tab. 1).

The mean number of taxa present in the springs was very low (5), which was reflected in the low values of other metrics based on the taxon richness (Tab. 2).

The two-dimensional NMDS plot (Fig. 2) shows three clusters, with area 1 grouping limnocrenes (denoted by L) and area 2 grouping mainly helocrenes (denoted by H). Area 3 groups springs flooded by the river overflow (F).

In the species-based NMDS plot (Fig. 3), the taxa known to be associated with springs are grouped in areas 1 and 2, whilethe eurytopic species and those associated with smallwater bodies, usually with mud deposits, are grouped in between areas 1 and 2.

Positive Spearman's rank correlation coefficients were found between spring species pairs: Candonopsis scourfieldi and Psychrodromus fontinalis (k = 0.49), Eucypris pigra and Potamocypris zschokkei (k = 0.44), Eucypris pigra and Fabaeformiscandona brevicornis (k = 0.52), and P. zschokkei and Psychrodromus olivaceus (k = 0.32). Scottia pseudobrowniana and Cryptocandona vavrai were not significantly correlated with any other spring species.

The DCA results revealed the gradient length represented by the first ordination axis to be higher than 3 SD, for which reason the direct CCA ordination was performed. The cumulative percent variance of the species data on the first and second axis explains 27% of the total variance in the species composition (Tab. 3).

Fig. 2 Two-dimensional NMDS plot for each spring. Coding: Two digits − spring number; L, H or R − limnocrene, helocrene or rheocrene, respectively; D − drying up, C − presence of Cardamine amara, L − presence of leaf litter, F − flooded.

Fig. 3 Two-dimensional NMDS plot for species. For abbreviations see Table 1.

3.1 Types of springs and substrates

The CCA performed on the ostracod assemblages and spring types showed the variables used in the ordination to explain 28% of the total variance. The stepwise selection of environmental variables showed only three out of the six environmental parameters to statistically significantly (p < 0.05) explain the moderate range of total explained variance in the species occurrence. The presence of limnocrenes, flooding, and leaf litter were responsible for 9.4, 6.9, and 3.5% of the variance in the ostracod assemblages, respectively. As shown by the CCA ordination plot, the presence of leaf litter is related to the area 1 species association in which Fabaeformiscandona brevicornis, Psychrodromusolivaceus, Scottia pseudobrowniana, Cryptocandona vavrai, Eucypris pigra, Potamocypris zschokkei, and Candonopsis scourfieldi are known for being associated with springs. Area 2 brings together species associated with limnocrenes (Fig. 4, Tab. 1).

A summary of the ostracod occurrence in individual spring types is given in Table 4. The diversity metrics showed no significant differences between limnocrenes, helocrenes, and rheocrenes. Significant were the differences in the Evennesse^H/S and the number of individuals between springs with and without leaf organic matter as well as between drying-up and not drying-up springs. Significant correlations were found between the total number of ostracods and springs without leaf litter (k = −0.41, p < 0.05), flooded springs (k = 0.34), and springs drying up in the warm part of the year (k = 0.4, p < 0.05).

The water flow rate in the springs was low, ranging from 0 to 0.181 m/s. The insolation depended primarily on the season and was linked to shading by trees. The plant cover consisted mainly of Cardamine amara and Carex acutiformis. The bottom sediment was mainly mineral. The plant cover correlated positively with the sediment organic matter content and sorting coefficients M and W (Tab. 5).

The CCA run on the ostracod assemblages and the substratevariables showed the variables used in the ordination to explain 17.7% of the total ostracod variance. The stepwise selection of environmental variables showed only two of the seven environmental parameters proved significant (p < 0.05) in explaining the moderate range of the total variance in the species occurrence. Substrates dominated by the mineral and organic content accounted for 7.2 and 7.4% of the variance in the ostracod assemblages, respectively. As shown by the CCA plot, the presence of mineral sediment is related to the occurrence of species typical of springs (Fig. 5, Tab. 1). Organic matter-dominated substrates were related only to juvenile Psychrodromus sp. individuals and, to a lesser extent, to Candona fragilis and C. candida.

Fig. 4 Ordination plot for species and environmental variables (spring type) on the first two CCA axes for samples from the Krąpiel valley springs. For abbreviations see Table 1.

Fig. 5 Ordination plot for species and environmental variables (substrate) on the first two CCA axes for samples from the Krąpiel valley springs. For abbreviations see Table 1.

3.2 Physical and chemical water properties

The oxygen content in some of the springs was low due to the low water flow rate and a high leaf litter supply. The leaf litter content was negatively correlated with the mineral content, the correlations with BOD₅ and flow rate being positive. The water pH fluctuated between neutral and acidic. The mineral and organic matter contents varied extensively between the springs (Tab. 6).

The abundance of Candona candida correlated positively with PO₄ (k = 0.43, p < 0.05), Fe (k = 0.66), and turbidity (k = 0.43); Cypridopsis vidua correlated positively with PO₄ (k = 0.39) and NH₄ (k = 0.62); Eucypris pigra correlated positively with BOD₅ (k = 0.34); F. brevicornis correlated positively with conductivity (k = 0.33) and NO₃ (k = 0.52); juvenile Psychrodromus sp. correlated positively with PO₄ (k = 0.6), NH₄ (k = 0.83), and turbidity (k = 0.38). On the other hand, reverse correlations were found between the abundances of Psychrodromus olivaceus, Scottia pseudobrowniana, and F. fragilis and pH (k = −0.37, k = −0.41, k = −0.35, respectively). The total number of ostracods showed significant correlations with the contents of iron (k = 0.41) and organic matter (k = 0.49).

The results of CCA for ostracod assemblages and hydromorphological parameters indicated that the variables used in the ordination explained 22% of the total variance in Ostracoda. Stepwise selection of environmental variables showed that three of the four environmental parameters statistically significantly (p < 0.05), explained the moderate range of total variance in the occurrence of species. The river habitat modification index (RHM) was responsible for 6% of the variance in ostracod assemblages, the habitat modification score (HMS) for 6%, and the river habitat quality index (RHQ) for 5.5%. In the CCA ordination diagram the RHM is related to juvenile Psychrodromus sp., and HMS is related to the species in area 2. The species concentrated in area 1 are no trelated to any of the hydromorphological parameters, which may indicate that the transformations of the river habitat are not severe enough to affect the assemblages of these ostracods in the springs (Fig. 6).

Fig. 6 Ordination plot for species and environmental variables (hydromorphological factors) on the first two CCA axes for samples from the Krąpiel valley springs. For abbreviations see Table 1.

3.3 Taxonomic remarks

According to Meisch (2000), Cypria ophtalmica has two forms: C. ophtalmica f. ophtalmica and C. ophtalmica f. lacustris. Morphological features, coloration and the habitat of occurrence showed the form lacustris to be represented in our materials. Cypria ophtalmica f. lacustris occurs almost exclusively in springs (Meisch et al., 2019).

4 Discussion

The 22 springs sampled in this study yielded 30 ostracod species, including 8 known from the literature to prefer springs or to be found in water bodies associated with springs (Tab. 1). In addition, there were also two stygobiontic species: Schellencandona belgica and Cryptocandona vavrai (Meisch, 2000). The remaining species were mainly either eurytopic or associated with small water bodies or wetlands. For comparison, Rossetti et al. (2005) found 12 species (including 3 spring ones) in 31 lowland springs of Northern Italy, while Rossetti et al. (2020) recorded 19 (5) taxa from 50 springs in the same region. The spring species in out study accounted for 30% of the total species richness, about 25% being reported by the studies cited. In our study, the number of taxa per spring ranged from 2 to 9 and averaged 5. This is consistent with results of Rossetti et al. (2020) who reported the mean number of taxa per site to be 2.5–2.2.

In this study, the dominants and/or the most frequent taxa were Cypria ophtalmica, Cypridopsis vidua, Candona candida, Cyclocypris laevis, juvenile Candona sp., juvenile Psychrodromus sp., and juvenile Pseudocandona sp. Literature data on the distribution of ostracods in springs concern mainly from mountain areas. Despite the differences in the altitude, the mountain spring ostracod assemblages were dominated by the same species as those listed above (Roca and Baltanás, 1993; Scharf et al., 2004; Külköylüoğlu and Yılmaz, 2006; Rosati et al., 2014; Zhai et al., 2015). In limnocrenes of lowland Italy, the most common species were C. ophtalmnica, Cyclocypris leavis (Rossetti et al., 2005), and C. ophthalmica, Cypridopsis vidua, Prionocypris zenkeri (Rossetti et al., 2020). Crenophilic and spring-related species showed usually a low prevalence and low frequency. One of such species, Scottia pseudobrowniana, may also be semi-terrestrial (Meisch, 2000; Rossetti et al., 2020). Differences in the ecological preferences of S. pseudobrowniana and the stygobiontic C. vavrai were confirmed by the correlation analysis which showed those species not to be significantly correlated with other spring species. A very similar pattern among crenobionts was reported by Rossetti et al. (2020) who found single such species at individual sites (1–7).

4.1 Effects of spring type and substrate

With respect to the spring type, the present study showed C. ophtalmica to be the most common taxonin both helocrenes and limnocrenes; while C. laevis, C. ovum, and C. vidua were most common in helocrenes, C. candida and N. neglecta being most common in limnocrenes. C. ophtalmica, C. laevis, and C. ovum are remarkably tolerant of a wide range of environmental condition (Meisch, 2000). C. vidua is often associated with a plant-derived substrate (Meisch, 2000), present in 54% of the helocrenes. The dominance of adult specimens of C. candida and N. neglecta in limnocrenes may have been related to the fact that these springs did not dry up, for which reason the species could complete their life cycles there; on the other hand, helocrenes, highly prone to drying-up, showed numerous Candona spp. larvae. No comparative literature data could be found on lowland helocrenes. In the case of limnocrenes, Rossetti et al. (2005) indicated C. ophtalmica and C. laevisas frequently occurring and C. vidua as very numerous there. Their recent research (Rossetti et al., 2020) showed C. ophthalmica, C. vidua, and Prionocypris zenkeri as the most frequent species.

The average species richness in our study was 5 per spring. Rosati et al. (2014) reported 4.7 and 2.1 species for helocrenes of the Palearctic rheo-limnocrenic springs. Rossetti et al. (2020) found 3.6, 2.5, 2.0, and 2.2 species in 2004, 2015 (in Lombardy), 2001, and 2015–2016 (in Emilia-Romagna), respectively. Thus, the species richness data collected so far show the Krąpiel valley springs to support quite high numbers of aquatic invertebrate species, which may be taken as evidence of the springs' high ecological status.

In no other ostracod study have the spring habitat properties such as the presence of leaf litter, drying-up in the warm part of the year, and flooding been taken into account. In our study, the number of ostracod species in springs without leaf litter was twice as high as that in springs with leaf-derived organic material. The species richness in flooded springs was higher than in not flooded ones by a factor of 3. In addition, the number of species in springs that were intermittently drying up was higher than in those not affected by drying by a factor of 3 as well. Consequently, the high number of ostracod species was associated with the absence of leaf litter, with flooding, and with intermittent drying up. A large amount of leaf litter generates oxygen deficiency in water with low flow rate, which clearly did not act in favor of the ostracod fauna. Spring dry-up, in turn, enhances sediment-bound organic matter mineralization and sediment oxygenation. In addition, desiccation is often necessary for ostracod resting eggs to develop. Flood water can remove the leaf litter burden from a spring and facilitate faunal exchange between the spring and the river bed.

4.2 Effects of physical and chemical water properties

It is interesting to find out whether the differences in species richness could be associated with differences in physical and chemical water properties. The species richness in our study was twice that reported by Rossetti et al. (2020). The oxygen status of the springs studied by Rossetti et al. (2020) was much lower and hypoxia was often observed. The temperature ranges in both studies were similar, particularly with respect to the Alpine springs. While our springs showed a slightly lower mineralization, their phosphate and nitrate levels were 10 and 3 times higher, respectively, that the springs studied by Rossetti et al. (2020). The latter were alkaline, while ours were neutral or slightly acidic. This comparison allows to infer that the oxygen content could have affected the number of ostracod species, while the mineral salt content was irrelevant. However, Rossetti et al. (2020) concluded that it was the water temperature and mineralization level that were the variables most important for structuring the ostracod communities. Mezquita et al. (1999) found ostracod assemblages to be mainly affected by water chemistry, with organic content and oxygen concentration playing a secondary role. Ostracod assemblages studied by Zhai et al. (2015) were significantly influenced by the mineral content and TOC. In this study, the total number of ostracods correlated significantly with the contents of iron and organic matter. The number of ostracods was probably not affected by the iron content in itself, and the correlation between the iron content and that of organic matter was more important. However, the association between the organic matter and ostracod abundance may be due to the ostracod feeding on i.a. dead plant fragments, including leaf litter, where by the litter decomposition is facilitated and contributes to the dissolved organic matter pool.

There is no literature evidence on the relationship between ostracod assemblages and the hydrological assessment of a river valley (River Habitat Survey, RHS). Such analyses have been performed for aquatic beetles (Pakulnicka et al., 2016) and water mites (Zawal et al., 2018) in the Krąpiel valley. In the case of beetles, hydrological parameters were responsible for 13.3% of the total variance in species composition, while anthropogenic transformation of the river valley (RHM and HMS) had a positive influence on the crenobiontic water mite fauna. For ostracods of this study, environmental variables explained 22% of the total variance in the species composition, and parameters defining anthropogenic transformations also significantly influenced the species assemblages (12% of the variance). It appears, however, that this group of environmental factors was more important for eurytopic than for most of the typical spring species.

5 Conclusions

Ostracod assemblages were most affected by the spring type (particularly limnocrenes), spring flooding by the river, the presence of leaf litter and fine particulate organic matter, a high content of NH₄, BOD₅, conductivity, pH and Fe. The total number of ostracods correlated significantly with the contents of iron and organic matter. The ostracod abundance in springs without leaf litter was twice that in springs supporting leaf-derived coarse organic matter. In flooded springs, the abundance was three times that found in not flooded springs. The springs subject to periodic dry-out supported ostracod abundances three times higher than those in springs that did not dry up. Similarly, a higher number of species was associated with the absence of leaf litter, with flooding, and with drying up. Springs with a substrate dominated by mineral material rather than fine organic matter, with a higher pH, the presence of leaf litter, and the absence of flooding provided apparently more suitable habitat conditions for spring species. The hydrological factors within the river valley (River Habitat Survey, RHS) showed no effect on the spring species.

Acknowledgements

The research was supported by grant No. N 305 222537 from the Polish Ministry of Science and Higher Education.

References

All Tables

All Figures

	Fig. 1 Location of the sampling sites. (A) Rivers. (B) Lakes and fish ponds. (C) Forests. (D) Locations of springs (Z1–Z6).
In the text

	Fig. 2 Two-dimensional NMDS plot for each spring. Coding: Two digits − spring number; L, H or R − limnocrene, helocrene or rheocrene, respectively; D − drying up, C − presence of Cardamine amara, L − presence of leaf litter, F − flooded.
In the text

	Fig. 3 Two-dimensional NMDS plot for species. For abbreviations see Table 1.
In the text

	Fig. 4 Ordination plot for species and environmental variables (spring type) on the first two CCA axes for samples from the Krąpiel valley springs. For abbreviations see Table 1.
In the text

	Fig. 5 Ordination plot for species and environmental variables (substrate) on the first two CCA axes for samples from the Krąpiel valley springs. For abbreviations see Table 1.
In the text

	Fig. 6 Ordination plot for species and environmental variables (hydromorphological factors) on the first two CCA axes for samples from the Krąpiel valley springs. For abbreviations see Table 1.
In the text