The populations of two related Daphnia species from the same mountain range strongly differed in their degree of clonal diversity, consistently with assumptions of their overwintering strategy. The differences in population structure between the two groups agree well with predictions of a physiologically-structured model of Daphnia life-history , which assessed the impact of clonal erosion on the genetic structure as observed by neutral markers.
Actively overwintering populations of D. galeata, which undergo a prolonged period of clonal selection, had much lower levels of clonal diversity, being almost completely dominated by a single clone in two cases. Moreover, the single individuals in these lakes that exhibited different multilocus genotypes differed only slightly from the dominant clones, being identical at a number of heterozygous loci. It is therefore likely that these individuals did not represent other hatchlings from sexually-produced dormant eggs but the observed differences were rather due to PCR artefacts or possibly somatic mutations of the dominant clones. This is particularly likely for Nižné Žabie Bielovodské, in which the rare MLG differed from the dominant MLG by a single allele of one locus.
In contrast with the situation in D. galeata, intrapopulation variation of D. longispina was very high, similar to what could be expected in an obligately sexual species (as demonstrated by the analysis of datasets simulating sexual reproduction in D. galeata; see also ). Although we only have data on the absence of individuals under ice for one of the three studied D. longispina populations, we can assume that all three populations most likely overwinter as dormant eggs, as suggested by similar genetic variation patterns. High clonal diversity suggests the limited influence of clonal erosion and the strong impact of sexual reproduction, which is consistent with the yearly re-establishment of the active population from dormant egg banks .
The observed patterns of clonal variation suggest that most populations we studied (apart from Štrbské Lake) represent extreme cases in a continuum of overwintering strategies. The contribution of individuals surviving the winter period to the next season's population may substantially differ, both among different localities (as observed by us) and among years within the same locality . The overwintering clones may belong to the most successful ones within the population. The persistence of several genotypes during two years was observed for Daphnia longispina in an artificial German lake ; two such MLGs were the most common in both studied years (ranging from 11 to 16% of all individuals in a particular sample). We assume that this represents an intermediate situation in which both overwintering clones and new hatchlings substantially contribute to the genetic structure, a pattern probably similar to the one observed for D. galeata in Štrbské Lake.
A wide range of clonal diversities were observed for multiple populations of the D. longispina complex from various European habitats analysed by Thielsch et al. . The pattern was congruent with our observations - the majority of D. longispina populations were characterised by very high clonal richness and diversity, but two out of three populations of D. galeata included in the analysis showed reduced variation. One of these was from a Dutch lake, in which we assume Daphnia may indeed persist year-round due to the relatively mild oceanic climate. However, low MLG diversity does not always correspond to long-term or strong clonal erosion. For example, the lowest diversity among D. longispina populations was reported from a small Tatra Mountain lake, Vyšné Satanie . In that case, however, it was most likely due to low allelic richness and low heterozygosity (and thus limited ability to differentiate between different MLGs), as it is unlikely that harsh winter conditions in this lake allow overwintering of cladocerans in active stages. The low allele richness in that particular population more probably results from an introduction bottleneck when the population became re-established after a period of severe anthropogenic acidification [21, 24].
Although we studied two different species, the overwintering strategy in the D. longispina complex is not necessarily species-specific. Whether or not Daphnia successfully survive the winter period is rather influenced by the ability to adapt to local conditions in a particular lake [18, 25, 26]. For example, populations of the same two species co-occurring in Lake Constance showed the opposite pattern to that observed in the Tatra Mountains: D. longispina (labelled D. hyalina in previous studies) overwintered in Lake Constance and did not invest in sexual reproduction, while D. galeata produced dormant eggs and disappeared from the water column in winter . This was also reflected in MLG variation (based on two allozyme loci) being lower in overwintering D. longispina than in D. galeata .
The fact that two coexisting Daphnia taxa in Lake Constance differed in their overwintering strategy within a single lake suggests that this life history trait has a genetic component. Although the conspecific Tatra Mountain populations analysed here showed a similar genetic structure, this is not necessarily a consequence of genetic relatedness. The patterns of mitochondrial variation suggest that each of these species colonised the Tatra Mountains multiple times, and in only two of the studied lakes from the West Tatras, Dolné Roháčské and Vyšné Jamnícke, did they likely originate from the same source [24, 27]. This pattern is well reflected in the Factorial Correspondence Analysis plot (Figure 2), in which individuals from these lakes are most similar. Furthermore, the overwintering strategy of D. galeata tends to be similar despite the fact that some important selection factors, such as predation pressure, strongly differ among the lakes. In two of them (Morskie Oko and Štrbské Lake), fish predation on Daphnia is high [24, 29], while there are no fish in the third studied lake (Nižné Žabie Bielovodské).
The strategy of daphnids in the Tatra Mountain lakes is most likely directly influenced by the size and depth of lakes. In this mountain range, D. galeata inhabits several relatively large and deep lakes, while D. longispina is found in smaller ones . In deeper lakes, the likelihood of successful overwintering of Daphnia individuals is higher, as they may survive in the deep refuge . In addition, investing into dormant eggs is much less efficient, as ephippia that sink to deep parts of the lakes are much less likely to hatch . The studied D. galeata populations from the Tatra Mountains apparently regularly overwinter, as observed in previous studies (Table 1). In fact, investment in sexual reproduction and ephippia production seems to be absent or extremely low in the deep lake Morskie Oko, where no ephippia-bearing females or males were observed during an intensive one-year study of this population .
Apparently, prolonged clonal reproduction is highly advantageous for D. galeata populations in the studied lakes. Gliwicz et al.  even suggested that Morskie Oko Daphnia are obligately asexual. Indeed, a higher frequency of obligate parthenogenetic Daphnia has been observed at high altitudes [e.g., [2, 30], and populations identified as D. pulicaria inhabiting alpine lakes in the Tatra Mountains are strictly asexual . Interestingly, even asexual lineages coexisting in a single lake may differ in overwintering strategy . However, we consider it unlikely that D. galeata populations in the Tatra Mountains are obligate parthenogens, in the sense that they produce dormant eggs parthenogenetically (see  for an overview). To our knowledge, this reproduction mode has never been convincingly shown in the D. longispina complex, and it is unlikely that it would have arisen at least twice independently in the Tatra Mountain lakes that were likely colonized by D. galeata from different sources [24, 27]. It seems rather that local D. galeata populations remain in the parthenogenetic life cycle phase only, and have stopped investing in sexual reproduction altogether.
Moreover, we observed reduced levels of clonal diversity in the D. galeata population from Štrbské Lake, a locality in which the Daphnia history in recent decades is relatively well known. Štrbské Lake was colonized by D. galeata in the second half of the 20th century, and subsequently gradually replaced the resident population of D. longispina [24, 32]. From this locality, we have direct evidence for the production of ephippia in D. galeata, as well-preserved dormant eggs of this species extracted from various sediment layers from this lake were analysed genetically . Interestingly, D. longispina also could be found in Štrbské Lake in winter at the beginning of the 20th century , which suggests generally favourable conditions for overwintering at this locality. Historical data also confirm that D. galeata was already present in Morskie Oko a century ago [24, 33]. It is possible that the higher clonal diversity observed in D. galeata from Štrbské Lake is due to the much younger age of this population, as a result of which clonal selection has not yet eroded genetic diversity to the same extent as in the two other populations.
Alternatively, this population may have a generally higher tendency to recruit from dormant eggs than the other two studied D. galeata populations, either due to local environmental factors or due to its genetic background. In particular, it is possible that the local population was influenced by introgression from D. longispina during the period of species replacement (see ), as also suggested by the intermediate position of some MLGs from Štrbské Lake in the Factorial Correspondence Analysis (Figure 2). However, we presume that environmental conditions are more likely driving the Daphnia overwintering strategy than the taxonomic composition of the populations.
Apart from reductions of clonal diversity, another predicted consequence of prolonged clonal erosion is deviation from the Hardy-Weinberg equilibrium. Simulation data  suggest that the typical result of clonal selection is a heterozygote excess even in the absence of any selection advantage for heterozygotes. Indeed, we observed strong heterozygote excess in all three studied D. galeata populations that survive winter as active animals, in contrast to D. longispina populations. The third consequence of clonal erosion may be among-population differentiation . This was high for both studied species, but D. galeata did show higher differentiation (D
= 0.42, in comparison with 0.33 for D. longispina), which would conform to the prediction. In this case, however, factors other than clonal selection are probably more important, in particular persistent founder effects [4, 35]. Patterns of mitochondrial DNA variation suggest multiple independent colonisations of Tatra Mountain lakes within each species [24, 27], and the high among-population differentiation is likely a direct consequence.
Our data show that the genetic diversity of a Daphnia population may be strongly influenced by the choice of reproductive strategy during unfavourable periods, and can show greatly different patterns among populations living in close proximity to one another, depending on local environmental conditions. What are the benefits and costs of an active overwintering strategy? Females surviving under the ice are ready to start reproduction as soon as conditions improve in spring, which is a great advantage against genotypes hatching from dormant eggs . If seasonal changes in a particular lake are sufficiently predictable over a long period, overwintering genotypes may eventually prevail in the population if no selection force acts in a negative frequency-dependent manner (e.g., microparasites; see ). Furthermore, immigrating genotypes less adapted to local conditions would be highly unlikely to establish , and even if so, they would not succeed in successful interbreeding with the locally adapted clones.
In extreme cases, long-term clonal erosion may result in the overwhelming dominance of a single clone, as observed in two of the three studied D. galeata populations. However, if these clones invest little or nothing into the production of dormant eggs, this may be an evolutionary trap. Ephippia produced earlier in the population are effectively buried in the sediment, and re-establishment from the dormant egg bank may be difficult (but see ). A low effective population size may then prevent microevolutionary changes necessary for adaptation to new selective forces, such as the introduction of new predators or parasites.