Introduction

The spread of species to areas where they are not native is increasing globally because of an increasingly connected world and the rise in human population size (Pyšek et al., 2020). The effects of invasive species have been regarded particularly problematic in aquatic ecosystems because of threats to biodiversity and human needs for water resources (Havel et al., 2015). In freshwater habitats, invasive filter-feeding species can have marked effects as they link the water column and benthic habitats and modify nutrient cycling (Atkinson et al., 2010; Vanni, 2021). Invasive species can also have direct effects on biota through displacement of native species (Darrigran, 2002) or indirect biotic effects involving more species (White et al., 2006).

Some of the most well-documented cases of invasion effects in freshwater systems are from the two dreissenids Zebra mussel (Dreissena polymorpha, Pallas 1771) and quagga mussel (Dreissena rostriformis bugensis, Andrusov 1897). They have spread widely throughout Europe and North America, having a major impact on the colonised ecosystems in both lakes and rivers as well as on drinking water treatment systems and electric power generation facilities (Aldrige et al., 2004; Connelly et al., 2007; Carlton, 2008). The dreissenids often become abundant, and as they are effective suspension feeders, the importance of the benthic-pelagic coupling is greatly enhanced (Karatayev et al., 2010; Nalepa & Schloesser, 2014). Because of their profound effects on the nutrient cycling, physical conditions and species interactions, the mussels function as ecological engineers (Strayer & Malcom, 2006; Higgins & Vander Zanden, 2010; Vanderploeg et al., 2017). The term “benthification” has been suggested to describe the process of ecosystem change after the invasion of dreissenids, implying increased water clarity and light penetration that modify the physical habitat and redirect energy from the pelagic to the benthic habitat (Mayer et al., 2014). The invaded systems can change quickly, and it has been suggested that zebra mussels reach their maximum density about 2–3 years after the populations become large enough to be detected (Burlakova et al., 2006).

The many ecological effects of dreissenid invasions include impacts on phytobenthos and meio- and macrobenthic communities (Radziejewska et al., 2009; Spear et al., 2022) as well as on the overall biodiversity (Salgado et al., 2019). The phytoplankton community is affected through nutrient excretion and selective grazing on specific taxa and size ranges (Naddafi et al., 2007). Nutrient dynamics can be impacted directly by changes in nutrient circulation or mussel excretion and indirectly through improved light conditions at the sediment–water interface, which may impact the internal nutrient release and recycling (Li et al., 2021). Indirectly, mussels also affect the growth potential of light-limited submerged macrophytes, particularly in eutrophic shallow lakes where large areas of the lake bottom can be colonised when the water becomes clearer (Zhu et al., 2006; Leisti et al., 2012). Changes in submerged macrophyte abundance can have significant cascading and structuring effects on other biotas such as fish and zooplankton (Jeppesen et al., 1998). The spatial structure of the fish community may change considerably after dreissenid invasion due to changes in trophic links, interspecific competition and alterations in food sources (Smirnov et al., 2020; Morrison et al., 2021; Karatayev & Burlakova, 2022).

The literature on dreissenid invasion is comprehensive, and effects of dreissenid invasion have been documented in many studies, see for example reviews by van der Velde et al. (2010), Mayer et al. (2014) and Nalepa & Schloesser (2014). There are, however, also reports showing no effects on key lake variables such as water clarity, chlorophyll a and/or nutrient concentrations, suggesting that its effect on the community structure and biomass at multiple trophic levels and their cascading effects are complex and maybe system specific (Jones & Montz, 2020; Lovejoy et al., 2023; Rantala et al., 2023), and further studies are therefore needed. Furthermore, most studies have focussed on the effects on specific ecosystem components and/or single ecosystems, and long-term studies of the ecosystem impact of invasive species are still relatively rare (Karatayev et al., 2023).

Zebra mussels were first reported in Denmark in 1840 (Kerney & Morton, 1970), whereas the presence of quagga mussel is so far restricted to a recently reported single location. During the past decades, zebra mussels have expanded their geographical distribution and been observed in many rivers and lakes in Denmark. Recently, it has invaded the largest watercourse system in Denmark, the River Gudenå, where it was first reported in 2006 (Gudenåkomiteen, 2009). This watercourse system includes many lakes, most of which are eutrophic due to high external nutrient loading but have different sizes and water depths.

The aim of this study was to elucidate the overall ecological effects recorded in a group of lakes in the River Gudenå system from which data are available on the first 10–15 years after zebra mussel invasion. We wanted to document how eutrophic lake ecosystems change when they become clearer, possibly due to increased grazing of phytoplankton by the mussels. We hypothesised that the invasion would not only lead to higher water clarity due to mussel filtration of the water, it would also exert a strong cascading effect on the entire ecosystem dynamics by affecting key structuring elements such as phytoplankton, macrophytes and fish (Miehls et al., 2009). The effects might, however, depend on the morphology and trophic state of the lakes and are strongest in shallow and eutrophic lakes (Strayer et al., 2011; Karatayev et al., 2021).

Methods

Study lakes

Our study included eight lakes that are all part of the River Gudenå system flowing from the south (Lake Mossø) to the north (Lake Tange). Four of the lakes (Mossø, Julsø, Brassø and Tange) have direct inflows from and outflows to the River Gudenå, whereas the other four lakes (Knud, Vejlsø, Hinge and Alling) are connected to the River Gudenå via channels or rivers (Fig. 1). The median maximum hydraulic loading is 17 m3/s upstream Lake Mossø (Voervadsbro) and 32 m3/s upstream (Tvillumbro) and near Lake Tange (Larsen & Ovesen, 2021). Generally, nutrient concentrations in Danish river systems have experienced significant reductions, particularly from 1970 to 1990, mainly due to improved wastewater treatment (Kronvang et al., 2005). Since 2000 and after the invasion of zebra mussels, the overall external nutrient loading reductions have been only minor, and the external P loading has changed only slightly between the last two monitoring periods (Table 1). Some of the lakes still suffer from internal loading of phosphorus (P) from a pool accumulated in the sediment when the external P loading was high (Søndergaard et al., 1999). Zebra mussels were recorded for the first time in 2006 in Lake Knud, but already in 2008 they were observed in most parts of the River Gudenå system (Orbicon, 2009), including all our study lakes.

Fig. 1
figure 1

Sites of the eight study lakes. The small inserted map with the square shows their location in Denmark

Full size image
Table 1 Physical and estimated phosphorus loading data on the eight study lakes
Full size table

Sampling and analyses

We used physical, chemical and biological data obtained following the standards of the Danish national lake monitoring programme (NOVANA) and collected by regional or national environmental authorities. NOVANA uses well-defined and comparable sampling techniques and analytical procedures (Svendsen et al., 2005). An overview of the available data before and after the invasion of zebra mussels is given in Table 2. From Lake Hinge, we have more detailed data that include phytoplankton and more sampling years than the available data on the other lakes. The physical data comprise lake area, mean water depth and Secchi depth, and the chemical data encompass total nitrogen (TN), total phosphorus (TP), total alkalinity (TA) and chlorophyll a (Chla). Each lake was sampled at least four times during summer from a central position, and a mean summer (1 May to 30 September) concentration was calculated. For Lake Hinge, data on inorganic nutrient concentrations: orthophosphate (PO4), nitrate + nitrate (NO3) and ammonium (NH4) were also available (see Søndergaard et al. (2005) for more detailed descriptions). When using seasonal data, we calculated a mean value based on monthly averages from May to September.

Table 2 Overview of years with data availability from the eight study lakes
Full size table

Phytoplankton (Lake Hinge only) was sampled from the same position as the water chemistry samples. The samples were fixed in Lugol’s iodine, and the phytoplankton was identified to genus or sometimes species level. Mean summer phytoplankton biovolumes were calculated by fitting each species/genus to simple geometric forms (Johansson & Lauridsen, 2017).

For submerged macrophytes, sampling was conducted at maximum abundance between 1 July and 15 August. The sampling method applied is described in more detail by Søndergaard et al. (2005). We used compiled data on maximum colonisation depth (Cmax, the deepest point where submerged macrophytes occurred), total number of taxa (excluding filamentous algae), mean coverage (RPA, relative plant covered area as % of the total lake area) and plant-filled volume (PVI as % of the total lake volume) from 150 to 375 observation points along a number of transects in each lake, the number of transects being lake size dependent. The transects covered the whole lake area and all depth zones but with main emphasis on areas with vegetation. Areas in deep parts of deep lakes where macrophytes were not likely to occur were not investigated. Observations were based on the use of a water glass or a rake or by a diver in deep lakes. For non-investigated areas beyond Cmax, coverage was assumed to be zero. The calculations of RPA, PVI and Cmax did not encompass filamentous algae or floating-leaved species. A list of total taxa numbers was obtained based on transect observations supplemented with observations of additional taxa in selected areas. The macrophyte taxa included floating-leaved species of which submerged forms were observed.

As for fish, two fish monitoring methods were used. Until 2003, gillnetting was conducted in both the littoral and offshore zones using monofilament nylon nets. Each net was 1.5 m deep and 42 m long and consisted of 14 units of 3 m length with different mesh sizes (6.25–75 mm) placed in random order. In each lake, benthic nets were set in the littoral and open water zone and in deep lakes also in the pelagic zone. The total number of nets used per lake varied between 12 and 56 depending on maximum depth and lake size (for details see Mortensen et al., 1990; Menezes et al., 2015). After 2004, stratified random sampling was conducted in different lake strata (Lauridsen et al., 2008) following the CEN standard (CEN, 2005). The sampling procedure was based on stratified random sampling (Appelberg, 2000). With both net settings, the samplings were conducted between 15 August and 15 September when underyearling fish had reached a size where they were likely to be caught in the nets. Nets were set in late afternoon and retrieved early next morning after approximately 18 h. In the analyses of fish data, we include only the three by far most dominant species using catch per unit effort and body size data (Menezes et al., 2015): roach Rutilus rutilus (Linnaeus, 1758), bream Abramis brama (Linnaeus, 1758) and perch Perca fluviatilis (Linnaeus, 1758). We only have before and after data from three lakes, and for two of these only one set from before (Table 2).

Statistical tests were performed for lakes and parameters with at least 3 years data before and 3 years data after zebra mussel invasion. This includes water chemistry data from Lake Hinge, Lake Brassø and Lake Knud and macrophyte and phytoplankton biovolume data from Lake Hinge. Monthly differences in nutrient concentrations, Chla and Secchi depth were tested from Lake Hinge. We used the Wilcoxon signed-rank test (two-sided) for all tests to test for differences between before and after invasion years and for differences between monthly samples in Lake Hinge. A nonparametric test was used because we cannot assume that the rather few data from before and after the invasion were normally distributed. We used SAS (proc npar1way Wilcoxon, P values from the Exact Test Two-Side) to run the analyses. It should be noted, however, that integrating fragmented data that are not collected for the objective in this study is a challenge in evaluating the results and conducting statistical tests.

Results

Nutrients and water chemistry

For most of the lakes, marked changes in lake water chemistry were recorded after zebra mussel invasion (Fig. 2; Table 3). In seven of the lakes (excepting Lake Knud), TP was reduced to 37–64%, TN to 49–77% and Chla to 18–62% of pre-invasion concentrations. Secchi depth increased by 38–171%. In Lake Hinge and Lake Brassø where sufficient data were available for testing, all changes were statistically significant. In the deep and less nutrient-rich Lake Knud, nutrient and Chla concentrations also decreased, and Secchi depth increased, but the changes were modest and not statistically significant. In some of the lakes such as Lake Hinge, a decreasing trend in nutrient and Chla concentrations was seen before the zebra mussel invasion, but this decrease became more abrupt and larger just after the invasion.

Fig. 2
figure 2

TP, TN, Secchi depth and Chla level in the eight study lakes from 1989 to 2021. The vertical grey bars indicate the year of zebra mussel invasion

Full size image
Table 3 Selected chemical and biological data before (B, before 1995–2005) and after (A, after 2006) on the eight study lakes
Full size table

On a seasonal scale, changes in Lake Hinge were seen during most of the year for TP, TN, Chla and Secchi depth (Fig. 3). For TP, the reduction was statistically significant for all months except December, for TN the reduction was statistically significant for all months except October–December and for Chla all months except February–April. Correspondingly, Secchi depth increased significantly all months except January–March. PO4-P varied less after the invasion, but the concentrations did not differ significantly from the before period for any of the months. NO3-N declined in winter and spring after the invasion, which largely explains the TN reductions. Seasonal NH4-N exhibited the same pattern before and after the invasion.

Fig. 3
figure 3

Seasonal changes in Chla, Secchi depth and nutrient concentrations in Lake Hinge before (1997–2005, red boxes) and after (2007–2021, blue boxes) zebra mussel invasion. The boxes and whiskers represent 10%, 25%, 75% and 90% fractiles. Tests for monthly differences were performed using Wilcoxon signed-rank test and significance levels marked with ***: P < 0.001, **: 0.001–0.01, *: P = 0.01–0.05

Full size image

Submerged macrophytes

Except from deep Lake Knud, the abundance of submerged macrophytes increased markedly after the mussel invasion in all lakes with available macrophyte data (Fig. 4; Table 3). RPA increased by a factor of 10 or more, Cmax increased from 0.1 to 5.4 m, and the taxa number rose in three of the lakes from only 1 to 7–9 taxa. In Lake Hinge, where data allow statistical tests, the increase in RPA from 1.1 to 13.3% was statistically significant as was the increase in taxa number from 7.2 to 10.4, whereas the increase in Cmax from 1.46 to 1.74 m was not statistically significant. The number of taxa in Lake Hinge and Lake Knud remained almost unchanged, whereas it increased from 7 to 42 in Lake Mossø. Data on lakes including multiple sampling years shortly after mussel invasion, e.g. Lake Brassø and Lake Hinge, suggest that there were only a few years’ delay in macrophyte recovery.

Fig. 4
figure 4

Macrophyte data showing RPA, RPV, Cmax and number of taxa. No before and after data were available on Lake Tange and Lake Alling. The vertical grey bars indicate the year of zebra mussel invasion

Full size image

The recovery of macrophytes included 16 different taxa in the four lakes on which detailed macrophyte data are available (Fig. 5). Different taxa appeared in the different lakes, and only Elodea canadensis (Michx.) made up a considerable RPA in all four lakes. Potamogeton was represented by several species and dominated in two of lakes in the last sampling year. Myriophyllum spicatum (L.) and Nitella flexilis (Agardh.) were dominant in two lakes.

Fig. 5
figure 5

Changes in RPA of the dominant macrophytes in four of the study lakes on which detailed taxa data were available. Only taxa with RPA > 0.5% are included in the figure. “0” indicates that this taxon was not recorded that year. Years 2004 and 2005 represent data from before zebra mussel invasion. Note different scales on the x-axis

Full size image

Phytoplankton

We only have phytoplankton data from Lake Hinge. Here, the mean summer biovolume declined from 13.3 mm3/l before (1994–2004) the invasion to 7.4 mm3/l after the invasion (2006–2018), but the biovolume varied considerably, particularly before zebra mussel invasion (Fig. 6). Thus, prior to invasion, the phytoplankton community was dominated by diatoms and some cyanobacteria, but after the invasion diatoms were less dominant, and the phytoplankton community became more diverse.

Fig. 6
figure 6

Phytoplankton from Lake Hinge showing summer means of total biovolume (A) and the relative importance of selected groups (B) from 1994 to 2018. The vertical grey bar indicates the year of zebra mussel invasion

Full size image

Fish

While roach CPUE (by weight) showed a mixed pattern by decreasing in two lakes and increasing in one, bream CPUE overall decreased, and perch CPUE increased (pairwise T-test of means of after minus before as percentage of before with all three lakes included, P < 0.05). Moreover, the average body size of perch increased (Fig. 7, pairwise T-test as above, P < 0.05), while no clear changes in body size were found for the other two species (data not shown). The proportion of perch CPUE of the total CPUE of the three species also increased (pairwise T-test as above, P < 0.05).

Fig. 7
figure 7

Fish populations in three of the study lakes based on before and after data. A–C Catch per unit effort (by weight) of roach, bream and perch, D individual perch body mass and E perch%, calculated as the total catch per unit effort of perch/(roach + bream)*100. Data show means + SE before and after zebra mussel invasion

Full size image

Discussion

The data from our eight study lakes confirmed the profound effects that zebra mussel invasion can have on many levels of lake ecosystems. The effects can be so substantial that lakes classified as having moderate or worse ecological state according to the European Water Framework Directive (European Union, 2000) have a much higher chance of achieving the goal of at least good ecological state. The ultimate implication of this could be that the management plans for these mussel-invaded lakes should be changed from “external nutrient loading reduction needed” to “no action needed”. In lake management, it should, however, not be forgotten that control of the external nutrient loading is highly important to avoid undesired eutrophication effects. Here, dreissenids may block the usually strong relationship between nutrient concentrations and water clarity as the clearer water may, in part, owe to zebra mussel invasion rather than to nutrient level declines (Zhu et al., 2006), giving a wrong impression of water quality improvement (Zaiko & Daunys, 2015). Consequently, even though high zebra mussel abundances, as shown in our study, have the capacity to counteract eutrophication and provide clearer water conditions and higher biodiversity—at least in the short term—these positive effects only last as long as the zebra mussels remain abundant, and use of dreissenids to conserve biodiversity and improve ecological status is therefore risky (Salgado et al., 2019). Furthermore, dreissenid invasion may negatively impact the local and original benthic filtrators, and dreissenids are therefore regarded as an undesired new fauna element (Geist et al., 2023). We have no data on the benthic fauna in the eight lakes to confirm this.

Nutrient concentrations were affected in all the study lakes, suggesting changed internal nutrient dynamics. Reduced TP and TN but unchanged inorganic P and N concentrations after mussel invasion, following phytoplankton biomass reduction, indicate that high mussel nutrient excretion rates, as for example recorded by Conroy et al. (2005), did not appear in our study lakes or may have been neutralised by the improved light conditions at the sediment–water interface, increasing the redox-sensitive retention of P (Zhang et al., 2022). Effects of improved light conditions at the sediment surface on the internal loading of both nitrogen and P, probably due to increased benthic primary production, have been seen in other shallow Danish lakes after biomanipulation and the establishment of clearer water (Søndergaard et al., 2017). Reduced TP concentrations, particularly during summer as seen in Lake Hinge, suggest that zebra mussels can reduce the otherwise high internal P loading characterising eutrophic Danish lakes (Søndergaard et al., 2005). TN was also reduced for the greater part of the year and during winter and spring due to less nitrate as well as during summer, most likely due to reduced phytoplankton biomass. Bruesewitz et al. (2009) suggested that NO3- limitation of denitrification can be reduced by zebra mussel activity that may translate into altered seasonal N cycling patterns in localised areas of lakes where the mussels are particularly abundant.

We cannot exclude the possibility that the decreasing P concentrations in the lakes after 2006 partly reflect a steadily reduced internal P loading, as seen in many Danish lakes (Søndergaard et al., 2013a). However, the very abrupt changes in P concentrations coinciding with lower Chla and improved light conditions immediately after the zebra mussel invasion in 2006 indicate that the decline is mainly caused by the changed trophic structure of the lakes. Unfortunately, we do not have any estimates of the biomass of zebra mussels in the eight lakes, but direct uptake of P in their tissues could be an important factor. A major part of the plankton filtered by the mussels will end up as faecal pellets or pseudofaeces that will settle out of the water to be incorporated into the benthic system, potentially changing the internal nutrient cycling. In the lower four Great Lakes, USA, Li et al. (2021) found that quagga mussels are now the primary regulator of the P cycling, and via their enormous biomasses they sequester large quantities of P in their tissues and dramatically intensify the benthic P exchange. They concluded that P availability is now regulated by the dynamics of mussel populations, while the role of external P inputs is suppressed.

At ecosystem level, markedly reduced Chla and phytoplankton biovolume were among the profound effects from the high abundance and filtration capacity of zebra mussels. Thus, in most lakes, Chla was reduced to summer mean values below 25 or 12 µg/l, which are the boundary concentrations used in Denmark by the Danish Environmental Protection Agency to discriminate between moderate and good ecological quality when using Chla as the only indicator (Søndergaard et al., 2013a, b). Our data on phytoplankton composition are limited, but the available data from one lake, Lake Hinge, suggest a more diverse phytoplankton community after mussel invasion. In this lake, there were, however, no indications that zebra mussels modified the proportion and concentration of nutrients in favour of the appearance of cyanobacteria blooms, as occasionally observed (Bykova et al., 2006).

Reduced phytoplankton biomass ensures better light conditions as reflected in the improved Secchi depth in all our study lakes. As expected, this promoted the growth of submerged macrophytes whose abundances increased in all the six lakes on which macrophyte data exist. This is in accordance with e.g. Leisti et al. (2012), who found a Secchi depth increase from 1.3 to 1.9 m after zebra mussel invasion, an increase in Cmax from 1.6 to 3.5 m as well as enhanced wet macrophyte biomass (by a factor 19) in some parts of Bay of Quinte, Canada. Similarly, in Oneida Lake, USA, Zhu et al. (2006) found that the maximum depth of macrophyte colonisation increased from 3.0 m before zebra mussel invasion to 5.1 m afterwards. Zhu et al. (2006) also found an increase in macrophyte species richness and argued that the composition of the macrophyte community changed from low-light tolerant species to those tolerating a wide range of light conditions, and they moreover revealed that increased light penetration alone explained these changes in macrophyte distribution and diversity. Likewise, Ibelings et al. (2007) and Salgado et al. (2019) coupled increased macrophyte species richness in lakes with enhanced zebra mussel abundance and high filtration capacity because both factors help to maintain clear water. In our study lakes, the number of taxa also increased considerably, particularly in the lakes with low richness before the invasion. In lakes with multiple samplings conducted shortly after the mussel invasion, such as Lake Brassø and Lake Hinge, macrophyte recovery time was relatively short regarding both abundance and species richness; thus, there was only a few years’ delay in recovery after the mussel invasion and the improved light conditions that benefitted the growth potential of the submerged macrophytes. This indicates that even a few years with clear water conditions induced by zebra mussels might be enough to increase biodiversity and trigger a shift in primary producers from phytoplankton dominance to macrophyte dominance. The shallow nature of many Danish lakes means that a relatively small increase in Secchi depth suffices to expose a large part of the lake bottom to light with subsequent suitable growth conditions for and expansion of submerged macrophytes (Søndergaard et al., 2017). The relatively fast response after zebra mussel invasion supports the finding by Karatayev et al. (2023) of the strongest ecosystem impacts after 5–10 years.

We have only few data on fish, but the changes in fish community composition relative to species dominance also illustrate the improved lake conditions (Jeppesen et al., 2000); thus, we found an increase in biomass CPUE and the body mass of the potential predator perch, while the CPUE of the coarse fish bream and roach decreased, increasing the perch percentage of CPUE. Several studies have shown that perch is a superior predator within vegetation and under clear water conditions (Svärdson, 1976; Persson, 1983; Persson et al., 1988), while roach and bream are better competitors in open water in eutrophic lakes (Lessmark, 1983; Persson, 1983). The latter two species reinforce turbid conditions, not least in shallow lakes, due to sediment disturbance when feeding on benthic invertebrates and by preying on zooplankton, whereby they release phytoplankton from their grazers. In contrast, large-sized perch are piscivores and predate on coarse fish, thereby contributing to clearer conditions (Persson et al., 1988). The observed enhanced average body weight of perch is also to be expected after reduced eutrophication accompanied by better feeding conditions for perch, reflecting both reduced competition with cyprinids, elimination of stunted populations of small perch and increased coverage of submerged macrophytes (Persson et al., 1988; Jeppesen et al., 2000). Predation on zebra mussels may also have contributed to the increase in the dominance and body weight of perch as seen for yellow perch (Perca flavescens) in lakes in North America (Zhu et al., 2006). Moreover, Idrisi et al. (2001) found no evidence that zebra mussels had a negative impact on the growth, biomass or production of young yellow perch in Lake Oneida despite a major Chla reduction. Combined, these results indicate a shift towards a fish community characterising that of less eutrophic systems and with higher capacity of exerting top-down control in the lake ecosystem.

All our eight study lakes responded to the invasion of zebra mussels but to a variable extent, and in particular Lake Knud behaved differently. Lake Knud is considerably deeper (mean depth = 13.7 m) than the other study lakes, meaning that the presence of habitats suitable for mussel invasion is relatively small. Lake morphometry has been suggested to determine the invasion dynamics of zebra mussels, and Karatayev et al. (2021) predicted that the long-term reduction in primary producers would be lower in deep than in shallow lakes due to thermal stratification and the fact that a smaller proportion of the epilimnion is in contact with the bottom. Furthermore, Lake Knud was by far the most nutrient poor of the lakes and exhibited low (9 µg/l) Chla concentrations already before the invasion, which might suggest food limitation for the mussels as found in some studies (James et al., 2001; Strayer et al., 2011) and thus overall fewer changes from pre- to post-invasion. The second deepest lake in our study, Lake Mossø, responded strongly to the invasion of zebra mussels, but this lake is more eutrophic, indicating that zebra mussels have the lowest impact in deep and not nutrient-rich lakes.

To conclude, zebra mussel invasion alters lake ecosystem structure and functions in many and overall positive ways seen from a water quality perspective. These effects might be enough to change lakes into a good ecological state as required in the European Water Framework Directive. Although the zebra mussel is regarded as an invasive and unwanted species in Denmark, we should not ignore its significant benefits with respect to water clarification, mitigation of eutrophication effects, creation of biogenic reef structures and increased biodiversity as well as the supply of other ecosystem services with potentially important economic impacts (Burlakova et al., 2023). A warning signal is, however, that although the effects of invasion have lasted up to 15 years so far in our eight study lakes, a return to turbid and phytoplankton-dominated systems is still a risk because of the continuously high external nutrient loading. The occurrence of high submerged macrophyte abundances might help to ensure stable clear conditions even if the mussels disappear again, but not if the external nutrient loading is too high. Reduction of the external loading is therefore still crucial for all lakes to ensure good quality and high biodiversity. As we must anticipate that dreissenid mussels and other invasive species will spread further in Denmark, and probably also in other countries, in the coming years, we must be prepared to include their presence in the management of lakes.