The Chinese pond mussel Sinanodonta woodiana demographically outperforms European native mussels

Unionid mussels are essential for the integrity of freshwater ecosystems but show rapid worldwide declines. The large-sized, thermophilic Chinese pond mussel Sinanodonta woodiana s.l., however, is a successful global invader, spread with commercially traded fish encysted with mussel larvae; its negative impacts on native mussels are expected. Here, we exploit a natural experiment provided by a simultaneous introduction of S. woodiana and four species of native unionids for water filtration to a pond in north-eastern Poland. Sinanodonta woodiana established a self-sustaining population and persisted for 19 years in suboptimal thermal conditions (mean annual temperature, 7.4 °C; mean temperature of the coldest month, − 3.7 °C, 73-day mean yearly ice-formation), extending the known limits of its cold tolerance. Over four study years, its frequency increased, and it showed higher potential for population growth than the native mussels, indicating possible future dominance shifts. Outbreaks of such sleeper populations are likely to be triggered by increasing temperatures. Additionally, our study documents the broad tolerance of S. woodiana concerning bottom sediments. It also points to the importance of intentional introductions of adult individuals and the bridgehead effect facilitating its further spread. We argue that S. woodiana should be urgently included in invasive species monitoring and management programmes.

www.nature.com/scientificreports/ spatial extent of the invaded habitats, their broad connectivity and inaccessibility. Here we exploit an exceptional opportunity provided by a natural experiment inadvertently set up in north-eastern Poland. In 2000, a pond fed by groundwater was made in the edge zone of a fen complex in the Biebrza River basin (N53.7517, E23.3102, Fig. 1) for water retention and recreation. In the same year, to improve water quality, the owners of the pond introduced native unionid mussels: Anodonta anatina, A. cygnea, Unio pictorum and U. tumidus originating from local populations. Additionally, several individuals ("a bucketful") of S. woodiana were brought purposely from southern Poland (a straight-line distance of over 500 km) as this species is known to be particularly effective at water filtration. The area of origin of these individuals harbours well-established populations of S. woodiana 29 .
The pond was regularly stocked with fish: Alburnus alburnus, Ctenopharyngodon idella, Cyprinus carpio, Esox lucius, Rutilus rutilus, Hypophthalmichthys molitrix, Perca fluviatilis, and Tinca tinca originating from local sources; fish from hatcheries with heated water were not introduced. The pond is located in the coldest region of the combined natural and invasive range of established S. woodiana populations, excluding its presence in heated water effluents (climatic characteristics of the study area are given in the "Methods" section). Following the discovery of this site in 2016, we carried out a comparative study of the structure and dynamics of the S. woodiana and native unionid populations over four consecutive years. We addressed the following hypotheses: (i) S. woodiana would persist and successfully reproduce despite the harsh climatic conditions; (ii) population structure would differ among species, with more dynamic recruitment in S. woodiana than the native mussels; (iii) spatial distribution would differ among species reflecting their habitat preferences, and S. woodiana would reach higher densities at sites with a sandy than a muddy bottom. Additionally, our study points to intentional introductions of adult S. woodiana individuals as an important route of dispersal of this invasive species.

Results
Species composition and changes in relative frequencies. Sinanodonta woodiana co-occurred with native unionids: Anodonta anatina, A. cygnea, Unio pictorum, and U. tumidus. No other invasive bivalves were observed. Anodonta anatina was a dominant species, with a frequency ranging between 42 and 49%. The frequency of A. cygnea ranged between 11 and 15%, and U. pictorum between 26 and 40%. The frequency of U. tumidus did not exceed 3% in any of the study years. The frequency of S. woodiana was 1.7% in 2016, 4.5%  Shell-length distributions. Consistently over the study period, smaller-sized mussels contributed a higher proportion of individuals in S. woodiana than in the native mussels (Fig. 3). Except for 2016, when the S. woodiana sample consisted of only six individuals, differences in the shell-length distribution between S. woodiana and the native mussels were statistically significant (Kolmogorov-Smirnov test, P < 0.0001 in all comparisons, Bonferroni corrected significance level P = 0.0083). In A. anatina and U. pictorum, mean shell-lengths and shell-length distributions changed over time towards an increasing dominance of larger individuals, with statistically significant differences between the consecutive years (Tables 1 and 2). In A. cygnea, statistically significant changes towards the dominance of larger-sized mussels were observed between 2017 and 2018 and between 2017 and 2019. In S. woodiana, changes in shell-size distribution between study years were not statistically significant despite the annual removal of collected individuals, carried out in adherence to the general guidelines on dealing with invasive species. Individual shell-length measurements are given in Supplementary Table S1 online.

Species distribution and densities in relation to bottom sediments. Sinanodonta woodiana, A.
anatina, A. cygnea, and U. pictorum were present at all sampling sites, and within sites, no clustering or spatial separation among species was observed. Summary statistics of mussel densities at sites differing in the bottom substrate are given in Table 3. Overall mussel densities did not differ significantly between the study years (two-way ANOVA, F 3 = 1.43, P = 0.24) or bottom types (F 1 = 0.39, P = 0.53), but there were significant differences among species (F 3 = 14.00, P < 0.0001). Interaction between species and bottom type also had a significant effect (F 3 = 9.54, P < 0.0001). The bottom type had a significant effect in A. cygnea (F 1 = 14.67, P < 0.001) and U. pictorum (F 1 = 9.56, P < 0.01): in A. cygnea, densities were higher at sites with a muddy bottom, and in U. pictorum at sites with a sandy bottom (Tamhane's T2 pairwise comparisons, P < 0.001 and P < 0.01, respectively). The effect of bottom type was not significant in A. anatina (F 1 = 1.40, P = 0.25) and S. woodiana (F 1 = 0.19, P = 0.67), although both species occurred at higher densities at sites with a sandy bottom. Year and interaction between bottom type and year did not have a significant effect in any of the species.
Observations on the behaviour of S. woodiana individuals. This study provided an opportunity for preliminary observations on the behaviour of S. woodiana. Compared to the native unionids, S. woodiana was often burrowing much deeper into the sediments, and some individuals were found 20 cm beneath the surface of the sandy substrates. At a low water level during the hot and dry summer in 2018, we observed S. woodiana individuals crawling over distances of up to 10 m in one day.

Discussion
This study contributes to the understanding of the population dynamics of S. woodiana and its native counterparts during the early stages of invasion. It documents a self-sustaining population of S. woodiana in an area with cold and long winters and extends the known limits of its thermal tolerance. Comparison of demographic profiles shows a more favourable population structure in S. woodiana than in the native mussels, indicating possible future dominance shifts. This study also shows that S. woodiana is a habitat generalist concerning bottom sediments, and points to intentional introductions of adult individuals as an important and underappreciated route of dispersal of this invasive species.  www.nature.com/scientificreports/ collected in 2019). While the populations of native mussels might have been augmented with glochidia attached to the stocking fishes, this was less likely in S. woodiana. The species was not recorded in the vicinity before 13 , and given the distance of over 500 km over which the founding individuals were transported, their local availability was unlikely. In any case, as the stocking fishes originated only from local sources, which did not include heated-water hatcheries, any S. woodiana glochidia would have also been from locally-adapted populations. Winters in the study area are relatively cold and long. In 2000-2019, the mean temperature of the coldest month was − 3.7 °C, and the lowest mean monthly temperature was − 10.8 °C. The absolute minimum temperature was − 31.1 °C. Ice formed each year, on average for 73 days, with a maximum of 104 days. As far as we know, these are the most extreme climatic conditions in which an established population of this species was documented to date. Sinanodonta woodiana has been reported from Sweden, but no reproduction was observed there 30 . The populations in the Yenisei and Ob River basins inhabit heated water effluents 9,31 . The other population in northern Poland in a thermally-unpolluted water body is in a milder climate 24 . Thus, our study extends the known limits of cold tolerance of S. woodiana, indicating a shift in its realized niche 32 or an ongoing in situ adaptation 11 . Furthermore, with the ongoing climate change, the abiotic conditions in the invaded range of S. woodiana increasingly match its physiological optimum 33 . In our study area during the time since mussel introduction, the mean annual temperature increased by 0.8 °C, the number of days with ice formation decreased by 21, and the number of days with temperatures over 15 °C (coinciding with the production of ripe glochidia by S. woodiana 15 ) increased by nine. Sinanodonta woodiana survives at water temperatures up to 38 °C 34 and has a higher tolerance to thermal stress than the native mussels 21 . In heated water bodies it reproduces throughout the year 18 and occupies habitats with higher temperature ranges than the native unionids 14 , indicating that climate warming will increase its competitive advantage. Additionally, high mobility of S. woodiana and its tendency to burrow deeply into the sediments may help it better survive during drought episodes. www.nature.com/scientificreports/ As shown in our study, in suboptimal thermal conditions, S. woodiana can persist at low abundances for decades. Outbreaks of such sleeper populations (sensu 35 ) are likely to be triggered by changes in the environment, e.g., rising temperatures.
Population structure of S. woodiana in relation to native unionids. Over four study years, the relative frequency of S. woodiana increased from 2 to 9%. A comparison of shell-length distributions, approximating population age-structure, shows that smaller-sized mussels contributed a higher proportion of individuals in S. woodiana than in the native mussels in all study years, and this difference was increasing over time. This increase over time was possibly related to the removal of S. woodiana individuals, as hand-sampling tends to be biased towards larger individuals. On the other hand, the high mobility of this species and its striking burrowing behaviour, which lowers its detectability, might have counterbalanced this effect, as illustrated by the largest S. woodiana individual, with a shell length of 22.5 cm, found in the last study year. Nevertheless, it is possible that without the removal of individuals, the size structure would also shift towards larger sizes in S. woodiana, and its relative abundance at the study sites would increase even faster. Interestingly, a higher contribution of smaller- www.nature.com/scientificreports/ sized individuals in S. woodiana than in the native mussels was also observed in 24 , where no mussels were removed before the study. Thus, in both these studies, S. woodiana not only established viable populations but also showed higher potential for population growth than the native mussels. This is not surprising given that S. woodiana grows faster, matures earlier and produces more glochidia per female than the native unionids [17][18][19]36 . At increasing relative frequencies, its direct effects on the native unionids: competition for food, bottom space and host fish, filtering out sperm and larvae, and transmission of diseases 6 will play an increasing role, and a dominance shift can be expected. This, in turn, is expected to affect ecosystem functioning, including changes in water transparency and nutrient availability 25,37 , benthic habitat modification 38,39 , and reduction in the condition of fish 40 . Additionally, S. woodiana invasion threatens the endangered European bitterling Rhodeus amarus 16 , and its massive die-offs negatively impact water quality and reverberate to terrestrial ecosystems 41,42 . The increasing prevalence of S. woodiana in invaded areas 17,23,43,44 supports its predicted ability to effectively compete with native mussels. Our present study shows that demographic profiles of co-occurring mussel populations can indicate future dominance shifts already at initial invasion stages. However, as in many alien species 45,46 , the time-lag between the establishment of S. woodiana and the expression of its impacts can last decades, explaining why, despite its striking body-size ("a football-sized invasive mussel" 47 ), the threats from its invasion are largely underestimated.

Tolerance of S. woodiana for bottom sediment type. Despite a large number of studies documenting
the spread of S. woodiana (for a recent summary, see, e.g., 11,48 ), not much is known on its preferences concerning bottom sediments. Sinanodonta woodiana is mainly reported from ponds and reservoirs, which suggests its preference for muddy sediments. However, its presence in these habitats is related to its mode of dispersal rather than habitat preferences. Basing on a study in a heated lakes system with various habitats, Kraszewski and Zdanowski 14 suggested a preference of S. woodiana for sandy bottom substrates. The patchy distribution of sandy and muddy bottom substrates allowed us to test this hypothesis in the present study.
According to expectations, based on the known preferences of the native species 49 , A. cygnea occurred predominantly at sites with a muddy bottom, U. pictorum at sites with a sandy bottom, and A. anatina occurred at similar densities on both bottom types. Contrary to expectations, however, S. woodiana did not show a preference for either bottom type. Although its overall density was higher at sites with a sandy than a muddy bottom, this difference was not significant. Sinanodonta woodiana can utilize a broad range of host-fish species [15][16][17] and survive in a broad range of water-body types 13 . Our study indicates that it is also a habitat generalist concerning bottom sediments and adds to the suit of the known tolerances of this species.
Intentional human-mediated dispersal. The global spread of S. woodiana is primarily due to the trade in freshwater fish 7,9 . Our study points to intentional introductions for water filtration as an additional route of dispersal of this species. Large individual sizes and arguably beautiful colouration of S. woodiana add to its perceived attractiveness, and some people are willing to undertake considerable efforts to obtain individuals of this species. Occasional long-distance translocations can cause the bridgehead effect 46,50 , in which the establishment of populations in new locations facilitates the further dispersal of the species and leads to a self-accelerating invasion process. The way humans interact with invasive species is one of the main determinants of their spread  www.nature.com/scientificreports/ and establishment 51,52 . Our local interviews indicate that individuals from the study pond have already been transferred to nearby water bodies, and their filtering ability is highly appreciated. The propensity of people to acquire and translocate Sinanodonta mussels has been noted before 13,17,24,[53][54][55] and is probably more important than previously appreciated.

Management implications. Eradication of established invasive bivalve populations is extremely difficult 6 .
An apparently successful attempt to eradicate S. woodiana from invaded fish ponds involved lowering the water level and poisoning the fish and mussels 10,47 , but usually such measures cannot be applied. An alternative is the removal of individuals by hand harvesting. To be effective, however, it should cover the whole surface of the invaded water body and be repeated regularly. A related, commonly used practice in field research on invasive species is to remove the collected individuals from the study area. Our study shows that at least in S. woodiana, this is not likely to have any practical effect. We took out all individuals collected during four annual surveys from collection sites covering approximately 8% of the surface area of the pond. The relative frequency of S. woodiana increased while its densities and shell-length distributions remained unchanged. This was not unexpected, given a small proportion of the population sampled, combined with the high reproduction rates and mobility of this species. As sampling rarely includes more than 10% of the studied populations, alternatively to removing individuals from a study area, long-term studies involving marking and releasing them back might be considered. Knowledge of the biology of S. woodiana in the wild (e.g., growth rates, longevity, behavioural responses) is scarce, limiting our ability to manage and reduce its further spread. The priority, however, is to prevent introductions of S. woodiana to non-invaded water bodies. Fish trade remains its dominant dispersal route, so effective biosecurity measures are necessary. Well-coordinated monitoring programmes are needed for evidence-based management decisions 56 . Public participation is key to successful management of invasive species. Publicly accessible educational programmes explaining the problems of invasive species and increasing the appreciation of the native ones are required, especially when the invasive species elicit favourable reactions from people 51 , as is the case with S. woodiana.
Sinanodonta woodiana does not yet have the status of a recognized pest. For example, it is not included in the list of invasive alien species of European Union concern 57 and there are no regulations concerning this species in most countries. Our study documents the potential of S. woodiana to demographically outcompete native unionids. Combined with its recognized impacts and rates of spread, it highlights the need to urgently call the attention of policymakers and the public to the threats posed by S. woodiana to the integrity of freshwater ecosystems.

Conclusions
This study documents the potential of S. woodiana for long-term persistence at low abundances in suboptimal thermal conditions. Sinanodonta woodiana established a self-sustaining population and persisted for almost 20 years in an area with cold and long winters. Outbreaks of such sleeper populations (sensu 35 ) are likely under the rising temperatures scenario. Over four study years, the frequency of S. woodiana increased from 2 to 9%. Already at this low-abundance invasion stage, it showed a more favourable population structure than the cooccurring native mussels, indicating possible future dominance shifts. The tolerance of S. woodiana concerning bottom substrates adds to the suit of the broad tolerances of this species that contribute to its invasive potential.
Sinanodonta woodiana is usually spread with fish infested with its larvae. In this study, several adult individuals were transported from over 500 km away, purposely introduced for water filtration and subsequently translocated to other water bodies in the vicinity. This illustrates the importance of how people interact with invasive species and points to the underappreciated role of intentional introductions of adult S. woodiana that can lead to the bridgehead effect facilitating its further spread.
The potential of S. woodiana to demographically outcompete native mussels, combined with its recognized impacts and rates of spread, justify its urgent inclusion in monitoring and management programmes. Publically accessible educational programmes are needed to increase the awareness of the problems produced by this species and an appreciation of native bivalves.

Methods
Climatic conditions. To characterise climatic conditions, we used temperature and precipitation records provided by the Polish Institute of Meteorology and Water Management collected at the meteorological stations in Biebrza and Suwałki, which lie at a straight-line distance of 50 and 49 km from the study area within the same isotherm values. In 2000-2019, i.e. during the time since the introduction of S. woodiana till the end of our study, the mean (± SD) ambient temperature was 7.4 ± 0.6 °C, ranging from 6.4 °C (2010) to 8.7 °C (2019), (Fig. 4). The mean temperatures of January and July were − 3.7 ± 2.9 °C and 18.5 ± 1.5 °C, respectively. The lowest mean monthly temperature was − 10. The mean (± SD) pH was 7.84 ± 0.08, the mean specific conductance was 393 ± 30.2 µS cm −1 , and the mean dissolved oxygen content was 5.5 ± 0.8 mg dm 3 . The concentration of N-NO 3 was 0.64 ± 0.07 mg dm 3 , of N-NH 4 was 0.42 ± 0.12 mg·dm −3 , and of P-PO 4 3− was below the quantification limit (< 0.05 mg dm 3 ). The concentration of dissolved organic carbon (TOC) did not exceed 40 mg·dm −3 . The concentrations of Ca 2+ and Mg 2+ were 64.6 ± 16.1 mg dm 3 and 14.7 ± 1.0 mg dm 3 , respectively.
Mussel collection and documentation. During the study period, the bottom of the pond consisted of interspaced sandy and muddy patches. In 2016 we established seven sampling sites: four with a sandy bottom and three with a mud layer of approximately 30 cm. In 2017 and 2018, we included three additional sites (together, 5 with a sandy bottom and 5 with a muddy bottom). In 2019, parts of the pond dried out during a hot and dry summer, and sampling was possible at only three sandy sites and four muddy sites. Mussels were handcollected by wading, snorkelling or scuba diving, depending on water depth, from areas delineated with metal chains placed on the bottom. The collection area covered approximately 30% of a patch with a uniform bottom substrate (20 m 2 at sites S01, S05, S06, S07, S08, S09, and 10 m 2 at sites S02, S03, S04, S10). In unionid mussels, age determination based on external growth annuli is often unreliable 58 and was ambiguous in our study area, so we used shell-length distributions to approximate the age-structure of the populations. We measured shell-lengths of all individuals to the nearest 1 mm in 2016 and 2017 and the nearest 0.1 mm in 2018 and 2019. In 2017, due to adverse weather conditions, we measured only 30 randomly chosen individuals from the largest subsamples of A. anatina (collected at sites S01, S03, S09) and U. pictorum (S01, S07, S09); together in 2017, we measured 245 individuals of A. anatina out of collected 500, and 161 individuals of U. pictorum out of 480. We released the native mussels at sites of collection. In adherence to the commonly adopted practice in research on invasive species, we removed the collected S. woodiana from the study area. In 2018, we dissected 37 individuals of this species to determine their reproductive status. We used the opportunities provided by this study to document some observations on the behaviour of S. woodiana. We estimated the distance covered by individual mussels by measuring the length of the traces left on the sediment surface, clearly visible www.nature.com/scientificreports/ at a low water level during the hot and dry summer in 2018. On observing the striking burrowing behaviour of some S. woodiana individuals, we measured the depth at which they were found with a ruler.

Data analysis.
We included all mussel species in the analysis of the relative frequencies. In the analyses of shell-length distributions and mussel densities, we omitted U. tumidus, which occurred in low numbers throughout the study period. Changes in relative frequencies were tested with the χ 2 homogeneity test and Fisher's exact test. Differences in mean shell lengths between study years were compared using the Kruskal-Wallis test followed by Dunn's pairwise comparisons with a Bonferroni corrected significance level. Differences in cumulative shell-length distributions between species and study years were compared using the Kolmogorov-Smirnov test with Bonferroni-corrected significance level. To account for shell-size differences among species, we rescaled shell-length measurements from 0 to 100 in each species over measurements from the whole study period using min-max normalisation. Because of a generally low detectability of small mussels during hand-sampling, we set a minimum shell-length at 0 mm in each species to avoid accidental differences among them; we omitted this value in the analyses and graphical presentations. Differences in mussel densities were tested using the two-way analysis of variance (ANOVA) with species, bottom type and study year as explanatory variables, followed by Tamhane's T2 multiple pairwise comparisons. Density data were log 10 transformed to meet assumptions of normality. All analyses were carried out with XLStat 2020.

Data availability
The datasets generated and analysed during this study are included in this published article and its Supplementary Information files.