River barriers such as hydropower dams and weirs can negatively affect river ecosystems by disrupting connectivity and reducing biodiversity. However, such barriers could also limit the spread of invasive species. Here, we used a spatial population genetics approach to test whether river barriers act as a hindrance to gene flow in the invasive round goby (Neogobius melanostomus Pallas, 1814). We sampled gobies from four different rivers across their invasive range in Central Europe (the Danube, Dyje, Morava, and Rhine rivers), with locations on either side of eight major river barriers. Using microsatellite genotyping, we found that round goby populations were differentiated with increasing number of river barriers and with increasing distance between sampling sites, depending on the river system in focus. We found significant population differentiation across three individual barriers, but no clear indication that this was related to barrier type as barriers were highly diverse. We also found reduced genetic diversity in populations that were more recently established. Our findings suggest that successive river barriers can sometimes slow the spread of round goby. Further research on the features of barriers that hinder round goby movement will help to design barrier passage solutions that will both limit spread of this invasive species and maintain connectivity for the native fauna.

Aquatic habitat invasion, dispersal, genetic differentiation, hydropower dam, invasive species, Neogobius melanostomus, river connectivity

Aquatic invasive species are a significant driver of global biodiversity loss, especially in freshwaters (Reid et al. 2018; Pyšek et al. 2020; WWF 2020). Invasive species can negatively impact native species via increased predation pressure, heightened resource competition, habitat destruction, and the introduction of diseases and/or parasites that were not previously present in the environment (Pyšek et al. 2020). Invasive species can also have a significant economic cost in invaded regions, especially if they impact the productivity of native ecosystems (e.g., fisheries Marbuah et al. 2014) or damage infrastructure (Booy et al. 2017). It is difficult to eradicate invasive species after their introduction and establishment, and therefore management strategies often involve actions that try to limit the spread of invasive species into new areas. One means to achieve this is through the use of barriers (Krieg and Zenker 2020). Barriers can be particularly effective in riverine systems where an entire invasion corridor can be blocked to limit further spread. While some barriers can be explicitly designed to prevent the expansion of an invasive species (e.g., electrical grids or chemical barriers), others such as hydropower dams and weirs can secondarily act as a means to slow the spread of invasive species – or be equipped with additional barriers to impede the spread of invasive species (Wiegleb et al. 2022). Despite this, river barriers themselves are another contributor towards biodiversity loss in freshwater river systems because they alter environmental flows, fragment previously connected habitats, and disrupt migratory movements for native biota (Mueller et al. 2011; Thieme et al. 2023). Consequently, measures to improve connectivity, such as adding passage solutions or even complete barrier removal, are being implemented, but thereby may also inadvertently facilitate the dispersal of aquatic invasive species. It is therefore important to understand how existing river barriers affect species movements to better inform new developments. Here, we studied the extent to which various riverine barriers restricted the dispersal of a widespread invasive species, the round goby (Neogobius melanostomus Pallas, 1814).

The round goby is considered a prime model of a highly invasive species (Cerwenka et al. 2023) with ranges that continue to expand in both Europe (e.g., the Baltic Sea, the Danube, Rhine, and Elbe rivers and surrounding tributaries) and North America (e.g., the St. Lawrence River, all five Laurentian Great Lakes, and surrounding tributaries). The round goby is native to the Ponto-Caspian region of Europe and was first documented in its introduced ranges in Europe and North America in the late 1980’s and early 1990’s (Kornis et al. 2012; Ojaveer et al. 2015). Its long-distance dispersal to new locations has been mediated by human translocation via ship ballast water discharges (Corkum et al. 2004). Then, over more local spatial scales, range expansions continue primarily via short-distance movement dispersal (Šlapanský et al. 2020) and stochastic transport events, presumably facilitated by human activities (e.g., boating, bait fish) (Bronnenhuber et al. 2011). The round goby was originally thought to have a limited capacity for natural, non-human assisted dispersal due to its small body size, lack of swim bladder, and small home range. However, more recent research indicates that some individuals can disperse longer distances, ranging from 10 to 27 river-km/year (Brownscombe and Fox 2012; Brandner et al. 2013; Cerwenka et al. 2018; Christoffersen et al. 2019; Andres et al. 2020). Moreover, recent work has also demonstrated their ability to climb vertical surfaces and withstand fast flowing waters (Pennuto and Rupprecht 2016; Bussmann and Burkhardt-Holm 2020). It is currently unknown to what extent round goby have the capacity to bypass river barriers, and to what degree river barriers slow their expansion. Some initial work found that upstream range expansion was faster in a river without weirs compared to a river with weirs (Šlapanský et al. 2017). This knowledge is especially relevant given the current global recognition and push for increased river connectivity. Indeed, the European Union’s (EU) Water Framework Directive requires that rivers harnessed for hydropower production need to mitigate the damages that barriers cause to their ecosystems, through for example, implementing passage solutions like fishways (Geist 2021).

To assess the extent to which river barriers restrict round goby dispersal, we have used a spatial population genetics approach and estimated pair-wise population differentiation as a proxy of gene flow. This approach captures the effects of migration (or lack thereof) over long time periods, given that cross-barrier movements are probably too infrequent to be tracked in real-time for individual fish. We sampled round goby populations at sites throughout their invasive range in Central Europe where round goby already had established populations on either side of river barriers. We then related genetic population differentiation to geographic distance and the number of river barriers (e.g., weirs, hydropower dams). We predicted that population structure would follow a pattern of isolation by resistance, which would be evidenced by higher levels of population differentiation across multiple river barriers relative to barrier-less river stretches. This would indicate either a limited capacity for the fish to bypass barriers of their own accord and/or limited human-facilitated transfer (e.g., bait fish dumping, boat traffic). The barriers studied here from river systems in the wild are diverse, and they vary in type and potential impact on fish dispersal. We can therefore only preliminarily explore whether certain river barrier types appear to be more impactful. Finally, this data allowed us to investigate if more upstream populations showed reduced genetic variation in relation to downstream populations, as expected when gene flow towards the invasion front is restricted.

Our MRM models indicated that population pair-wise F_ST in the Danube was marginally associated with geographic distance between sampling sites (P = 0.05; Fig. 1A), and significantly associated with the number of river barriers separating sites (P = 0.001, Fig. 1B). In the Dyje River, F_ST was significantly associated with barrier number (P = 0.04, Fig. 1B), but not geographic separation (P = 0.16, Fig. 1A). In the Rhine, F_ST was not clearly associated with either geographic separation (P = 0.10, Fig. 1A) or the number of river barriers (P = 0.09, Fig. 2B).

Figure 2. 

Pairwise F_ST values plotted against A distance between sampling sites, or B number of river barriers between sampling sites, paneled by river system. * shows a significant (P < 0.05) result from the MRM models. Slopes derived from a linear regression to aid in visualizing the MRM model results, coloured data points show raw values. Note, the Morava river was not included in the MRM analyses because only two sites were sampled.

Significant population differentiation was detected across three out of the eight barriers: Poikam hydropower dam in the Danube, Birsfelden hydropower dam in the Rhine river, and Břeclav weir in the Dyje river (Fig. 2; Table 2).

Within-population genetic diversity represented by H_e declined as we sampled further upstream (LMM; est. ± std. error = –0.0014 ± 0.0005, z = –2.84, P = 0.0046), but was not clearly associated with the number of barriers between the sampling site and the most downstream site (est. ± std. error = –0.010 ± 0.0086, z = –1.18, P = 0.24, Fig. 3). When we remove the two far right data points, the effect of distance is still significant (est. ± std. error = –0.0059 ± 0.0025, z = –2.34, P = 0.019, Suppl. material 2: fig. S1). The modified Garza-Williamson indices suggested that all populations had experienced population size bottlenecks (M ranged from 0.19–0.23 in the Rhine, 021–0.28 in the Danube, 0.19–0.20 in the Dyje, and 0.20–0.21 in the Morava Rivers).

Figure 3. 

Model prediction (fitted line with 95% confidence interval) showing how the expected heterozygosity (H_e) of round goby populations decreases as sampling progressed upriver and approached the various rivers’ invasion fronts. * shows a significant (P < 0.05) result for geographic separation from the LMM. Coloured data points show raw H_e values.

We investigated whether river barriers slowed the dispersal of invasive round goby using a spatial population genetics approach, sampling round goby populations on either side of barriers in four different rivers in their invasive range in Europe. We found that populations were significantly differentiated over increasing distances and increasing numbers of barriers, depending on the river system. Starting broadly, round goby population differentiation was marginally associated with distance and significantly associated with barrier number in the Danube River (Germany); with increasing number of barriers, but not with distance in the Dyje River (Czech Republic); and with neither distance nor number of barriers in the Rhine River (Switzerland). Our findings that populations differentiate with distance (albeit marginally after accounting for barrier number) are in line with previous work at similar spatial scales. For instance Bronnenhuber et al. (2011) showed genetic differentiation in round goby populations as they invaded up river tributaries in the North American Great Lakes, with populations differentiating after approximately 50 km in one river. Similarly, Björklund and Almqvist (2010) found that round goby populations in the Baltic Sea differentiated over distances of only 30 km (though the authors used a different miscrosatellite marker set as Bronnenhuber et al. (2011)). Brandner et al. (2018) documented the progression of the round goby population spreading up the Danube River in Germany, and recorded a 30 km movement of the invasion front upstream over the course of four years (between 2010 and 2014). Overall, our findings suggest that with increasing numbers of river barriers, upstream round goby dispersal becomes progressively more difficult, a pattern that is likely to be mirrored in other fish species navigating these rivers.

River barriers are highly diverse, ranging in size and structure and differing in their implementation of fish passage solutions (Table 1). However, even with fish passage solutions in place, upstream fish passage efficiency is still low for non-salmonid fishes (averages ranging from 18–51%, Sun et al. 2023), and how efficient fish are at passing each barrier depends on characteristics the passage type (e.g., attraction, design, slope, flow) and of the migratory behaviour and swimming performance of the species (Noonan et al. 2012; Sun et al. 2023). Fish with migratory life-histories like salmonids generally pass more efficiently than non-migratory species (Noonan et al. 2012; Sun et al. 2023). Besides upstream migration of the invasion front, many fish have pelagic larval phases that may facilitate downstream dispersal in river systems. Although round goby are benthic and have no strictly pelagic larval phase, round goby larvae have been sampled nocturnally in the pelagic zones in lakes (Hensler and Jude 2007; Hayden and Miner 2009) and sampled in drift nets in European rivers (Janáč et al. 2013; Borcherding et al. 2016). However, no downstream drift distances have been reported yet for this species, which is partly because round goby can display plasticity in spawning times and reproductive output across different habitats making it challenging to get such estimates (Klarl et al. 2024). Furthermore, it is also unknown how well round goby larvae would pass river barriers with different fish passage solutions while passively drifting downstream.

We observed significant upstream-downstream differentiation across three individual barriers, one within each of the three river systems. The rivers and barriers studied here were highly diverse, reflecting the diversity of barrier types that have been installed in European river systems. Therefore, we explored our data to investigate whether there were any consistent patterns of differentiation associated with one barrier type. However, given the low level of replication within barrier types in our dataset, this investigation should be treated as preliminary. Overall, however, we saw no clear indiciation that any particular river barrier type (e.g., weir, dam – with or without fish passage solutions) was related to round goby population differentiation. In two of the cases where we observed significant cross-barrier differentiation, it occurred at small spatial scales with less than 2 km between the upstream and downstream sites. The Poikam dam in the Danube has no fish passage solution (e.g., fish ladders), which may hinder round goby dispersal across the barrier. However, at the same time there is a large lock system allowing ships to pass by the dam, which fish can presumably use as well. Meanwhile, the Birsfelden dam in the Rhine also showed genetic differentiation and has a vertical slot fish passage. Vertical slot passages may be more challenging for round goby to pass if the flow is high, the passage is steep, or the passage contains no bottom structure on which the goby can anchor itself (Pennuto and Rupprecht 2016; Wiegleb et al. 2023). However, the Birsfelden dam also has a boat lock, and gobies are regularly observed at high densities in the lock (pers. obs. PB). Birsfelden in Basel, Switzerland, is also located between two commercial harbours, which thereby increases the chances at this site that round goby from different geographic and genetic backgrounds have been introduced on either side of this barrier in particular (Adrian-Kalchhauser et al. 2016). The final barrier with significant differentiation was at the Břeclav weir in the Dyje river, which was equipped with a boulder ramp-style fish passage. This type of passage might be more easily passable for round goby, but it should be noted that the up- and downstream sampling sites were also separate by a comparatively larger spatial distance than at the Poikam or Birsfelden dams (~22 km).

It should be noted that several mechanisms can contribute to our observed population structures, including stochasticity in the genetic makeup of the founder population, gene flow, genetic drift (particularly in small populations), mutations, and different selective regimes across populations. Our measures of population differentiation, however, ought to be most strongly influenced by founder effects, gene flow (or hinderances to gene flow) and somewhat less by genetic drift, as our study populations are extremely young (founded within the past 8–15 years) (Sefc et al. 2007). Indeed, we found that the Garza-Williamson indices calculated for each sampling site suggested that the populations had all gone through recent population size declines. These findings are consistent with the round goby populations in these rivers all having been recently founded.

We also found that round goby showed less genetic variation in the furthest upstream—and therefore presumably most recently invaded—sampling sites relative to the downstream sites within the same river system. Lower genetic diversity at the invasion front can occur in newly invaded areas and diversity can increase later as alleles accumulate over time (Roman and Darling 2007). Newly invaded areas can therefore show evidence of founder effects, i.e. a reduced representation of genetic diversity in the novel population compared to the source population combined with further loss of diversity due to strong genetic drift while the novel population is small. Recently, Green et al. (2023) showed that round goby populations were less genetically diverse with increasing inland invasion off the Baltic Sea. However, in contrast, Bronnenhuber et al. (2011) found that genetic diversity was similar and relatively high among lake and recently colonized river populations in North America. It is important to note that we did not sample any round goby source populations or older, established populations within the European range, as this was not the primary focus of our current work. Population differentiation estimates between novel and source populations are likewise affected by the stochastic representation of alleles by the founders and the subsequent genetic drift in the novel population. Following upstream range expansion, substantial downstream gene flow (via pelagic larvae) may even carry the genetic founder effects taking place at the expansion front, and bring these effects back to the source populations altering their genetic make-up and opposing genetic differentiation. Given the observed patterns in genetic diversity and differentiation, the range expansion of round goby in the investigated river systems seems to proceed in the face of contraints to gene flow and these are apparently insufficient to prevent the spread of the species.

In conclusion, we have used a spatial genetics sampling approach to test the hypothesis that river barriers would slow round goby dispersal as indicated by population differentiation. This work was motivated by an increasing recognition of the benefits of river connectivity, which may also come with the cost of increased spread of invasive species, like the round goby (Mclaughlin et al. 2013; Thieme et al. 2023). In the Baltic Sea, round goby continue to expand northward, and they have recently been documented in inland freshwater tributaries (Verliin et al. 2017). There is concern that round goby will negatively affect native fish populations in these freshwater habitats via competition and egg predation, and this is of special concern for rivers hosting spawning grounds for species with eggs left to develop unguarded, such as salmonids (Verliin et al. 2017; Wallin Kihlberg et al. 2024) or other non-caregiving fish (e.g., Common barbel, Barbus barbus Linnaeus, 1758). Our work suggests that barriers, such as hydropower dams or weirs, appear to have an additive effect slowing the spread of round goby. Whilst restoration of river systems including their connectivity should stay a key priority (Geist and Hawkins 2016), understanding the mechanisms of how barriers affect dispersal of undesired invasive species is useful for their management. Since round goby ultimately benefit from artificial rip-rap structures and other anthropogenic modifications of river systems (Brandner et al. 2018; Roche et al. 2021), conservation or restoration of functionally intact river habitats is key in limiting their population densities and impacts on native fauna (Pander et al. 2016; Ramler and Keckeis 2020). Although mitigation options are presently quite limited for most established invasive species, future work assessing a greater number of barriers and detailed descriptions of their features (e.g., flow, incline, substrate) will add to a growing body of literature optimizing the design of barriers and fish passage solutions that potentially limit round goby (and other invasive fish) passage.

ESM: Conceptualization, Funding acquisition, Investigation, Writing original draft, Writing – review & editing Visualization; KMS: Formal analysis, Writing - review & editing; PB: Investigation, Resources, Writing – review & editing; TB: Resources; KB: Investigation, Resources, Writing – review & editing; AF: Funding acquisition, Writing – review & editing; JG: Investigation, Resources, Validation, Writing – review & editing; MJ: Investigation, Resources, Validation, Writing – review & editing; PJ: Investigation, Resources, Validation, Writing – review & editing; JP: Investigation, Resources, Validation, Writing – review & editing; JMM: Investigation, Validation, Writing – review & editing; APHB: Conceptualization, Funding acquisition, Formal Analysis, Investigation, Writing original draft, Writing – review & editing, Visualization.

This research was supported by funding from the Swedish Environmental Protection Agency (Naturvårdsverket, 2020-00051 to A.F. and E.S.M.) the Kempe Foundation (JCK22-0043 to E.S.M. and A.B.), and the Swedish Research Council Formas (2022-02796/2023-01253 to J.M.M). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

We sampled round goby from four European rivers in three countries, and all field collections were carried out under the appropriate licenses issued within each country (Czech Republic: JMK 153264/2020; Switzerland: authorizations for special fishing catches 01/02/2022 by Authority for Environment and Energy; Germany: electrofishing activities carried out under the license number 31-7562 issued by the district office of Freising, Bavaria, Germany).

Species georeferenced records are available at the European Alien Species Information Network: https://easin.jrc.ec.europa.eu/easin/RJD/Download/5c243fea-47d6-489d-8d37-129fb35ff91b.

Besides the first, second, and last authors, the remaining authors are ordered alphabetically and contributions should be considered relatively equal. We would like to thank the anonymous reviewers for the feedback on this publication.