Ecology of Elodea canadensis Michx. and Elodea nuttallii (Planch.) H. St. John—Insights from National Water Monitoring in Croatia

Elodea canadensis Michx. (common waterweed) and Elodea nuttallii (Planch.) H. St. John (Nuttall’s waterweed), two invasive aquatic plants from North America, have coexisted in European water bodies since the early 20th century. New localities for both species in Croatia continued to be discovered during a study that ran from 2016 to 2023 as a part of the annual implementation of Water Framework Directive monitoring that covered the entire territory of Croatia (786 sampling points in total). Based on these data, the distribution and ecology of both species were analysed. Elodea canadensis was found at 30 sampling points, mostly in rivers, and E. nuttallii at 15 sampling points, mostly in artificial canals. Nearly three-quarters (72.5%) of all elodea sampling points were in the Pannonian Ecoregion. Elodea canadensis was discovered for the first time in the Continental—Dinaric and Mediterranean—Dinaric Subecoregions. To study the ecology of the species, for each sampling point, vegetation relevés were performed and monthly measurements of physico-chemical parameters were collected. The most common accompanying species for both elodeas are presented, and the difference in species assemblages between the sites with E. canadensis and E. nuttallii was confirmed with the ANOSIM test. Furthermore, Indicator Species Analysis revealed eight species characteristic of E. canadensis sites and eleven species characteristic of E. nuttallii sites. Fitting multivariate models (CCA and NPMR) to species abundance revealed the ecological reaction of E. canadensis and E. nuttallii to environmental descriptors. The most strongly contributing environmental descriptors that influence the distribution of both Elodea species are biochemical oxygen demand, electrical conductivity and total phosphorus. In Croatia, the replacement of E. canadensis with E. nuttallii was observed in several water bodies with high nutrient loads.


Introduction
In the light of the ongoing spread of alien invasive species, more and more research is focused on revealing the ecology of such plants.Invasive alien aquatic species have numerous traits, such as allelopathy, phenotypic plasticity and propagule pressure, that make them more competitive than native species and can be considered their mechanisms of invasion [1].When successfully occupying a habitat, invasive plants become a threat to the stability of ecosystems by causing a loss of biodiversity, modifying the trophic structure and even changing overall habitat properties.When quantified, the management and damage costs of invasive alien species in Europe in the last 60 years add up to over a hundred billion euros [2].In comparison to terrestrial plants, an increased proportion of aquatic plants have ecological and economic impacts on their habitats [3].In Europe, the most widespread alien aquatic plant family is Hydrocharitaceae, with Elodea canadensis being the most widespread and E. nuttallii being the fourth most widespread species [4].
The genus Elodea consists of nine accepted species of submerged, freshwater perennials spread throughout the world, with records from all continents except Antarctica.All species are native to either North or South America.Out of the three species that have been introduced to Europe, E. canadensis and E. nuttallii are both native to temperate North America [16] and express invasive character all over the continent [17].One of the main drivers of the spread of waterweeds in the introduced areas is their vegetative reproduction in the absence of male plants.After fragmentation, the detached stem parts quickly develop adventitious roots and continue the expansion of the species.Both species have also been recorded in Croatia-E.canadensis was first recorded near the city of Sisak by Josip Schlosser, presumably before 1883, as the collector died in 1882 [18]; E. nuttallii was first recorded in 2006, in the drainage canals of Kopački Rit [9].
To date, research on waterweeds in Croatia has been very limited.Košćec [19] reported a record of E. canadensis in the surroundings of the city of Varaždin in 1909, the first published record of the species (Northern Croatia).He was, however, probably unaware of the two unpublished herbarium specimens stored in the Herbarium Croaticum-ZA collection, previously collected by Josip Schlosser around Sisak (Central Croatia) (ZA152740) and by Stjepan Gjurašin, who recorded the species in Ješkovo Pond (Northern Croatia) (ZA152740) [20].In his paper, Košćec [19] gave a detailed overview of the habitat and accompanying species.He assumed that E. canadensis was introduced into Croatian waterways from the northwest via the Drava River, since it was found in one of its tributaries.After this paper, E. canadensis was sporadically reported from the waterways and ponds of the Pannonian Ecoregion [21] for over a century, after which a new, Elodea-centred research paper was published by Kočić et al. [9] following the first discovery of E. nuttallii in Croatia in 2006.The distribution and morphological variations of both waterweeds were presented by the authors, and the further spread of E. nuttallii in Croatia was anticipated.In 2018, the first record of the previously congeneric Egeria densa Planch (syn.Elodea densa (Planch.)Casp.) in Croatia was published by Rimac et al. [22] updated by Vuković et al. [23], who presented the results of extensive mapping, showing a large number of localities spread across a much wider area than first reported.Finally, Piria et al. [24] classified E. nuttallii as very high risk and E. canadensis as high risk in terms of their invasiveness under current climate conditions in the Pannonian and Mediterranean regions of Croatia, as determined using risk-screening tools.
This paper gives the latest overview of the distribution of both waterweeds in Croatia and the first analyses of their ecology in the area.Resulting from the ecological studies that include the species samples from the whole of the country, this research contributes to the overall knowledge of the species' environmental preferences.
The aims of this research were to do the following: (a) present the latest observations of E. canadensis and E. nuttallii in Croatia; (b) analyse both species' preferences in terms of habitat characteristics; (c) analyse both species' response to water chemical and physicochemical parameters; and (d) present the most common accompanying species.

Distribution
During the period 2016-2023, waterweeds were recorded at 40 sites (Figure 1) out of the 786 sampled-24 rivers, 12 canals and 4 reservoirs.Elodea canadensis and E. nuttallii were exclusively recorded at 25 and 10 sampling sites, respectively.Furthermore, at 5 sampling sites, they were both recorded, but in different years, never accompanying one another.Over the years, E. nuttallii replaced E. canadensis in three reservoirs on the Drava River in Northern Croatia (sites number [33][34][35][36][37][38] and in two canals of the Drava River Basin (sites number 31,32,43,44).Overall, waterweeds were mostly recorded in rivers and canals (90% of records).Elodea canadensis was recorded in 22 rivers, 4 canals and 4 reservoirs, while E. nuttallii was recorded in 2 rivers, 10 canals and 3 reservoirs.Most records are in the Pannonian Ecoregion (72.5% of records), followed by the Continental-Dinaric Subecoregion (15.0%) and the Mediterranean-Dinaric Subecoregion (12.5%).During the study, E. canadensis was found for the first time in the Dinaric Ecoregion: first in the Continental Subecoregion in 2016, followed by a record in the Mediterranean Subecoregion in 2017.On the other hand, E. nuttallii was recorded for the first time in the Sava River Basin and is still present only in the Pannonian Ecoregion.Out of 40 sampling sites where waterweeds were recorded, only four are represented by a true lentic habitat in the form of a reservoir.
(sites number 31,32,43,44).Overall, waterweeds were mostly recorded in rivers and canals (90% of records).Elodea canadensis was recorded in 22 rivers, 4 canals and 4 reservoirs, while E. nuttallii was recorded in 2 rivers, 10 canals and 3 reservoirs.Most records are in the Pannonian Ecoregion (72.5% of records), followed by the Continental-Dinaric Subecoregion (15.0%) and the Mediterranean-Dinaric Subecoregion (12.5%).During the study, E. canadensis was found for the first time in the Dinaric Ecoregion: first in the Continental Subecoregion in 2016, followed by a record in the Mediterranean Subecoregion in 2017.On the other hand, E. nuttallii was recorded for the first time in the Sava River Basin and is still present only in the Pannonian Ecoregion.Out of 40 sampling sites where waterweeds were recorded, only four are represented by a true lentic habitat in the form of a reservoir.

Habitat Characteristics
Assessment of certain abiotic characters of the habitat (average water depth, turbidity, flow velocity, illumination/canopy coverage and substrate size) showed the preference of both waterweeds for water bodies with an average water depth of over 30 cm, clear to moderately turbid water, stagnant to moderately turbulent flow velocity and full illumination, with finer substrate (particles smaller than 2 cm) (Table 1).Furthermore, the difference between E. canadensis and E. nuttallii can be observed in their preferences with respect to flow velocity and substrate size.Elodea canadensis showed a preference for moderately turbulent watercourses (57% of records), whereas E. nuttallii was recorded more frequently in stagnant water (56% of records).Furthermore, E. nuttallii grew predominately in watercourses with clay and mud (81% of records), while E. canadensis grew on

Habitat Characteristics
Assessment of certain abiotic characters of the habitat (average water depth, turbidity, flow velocity, illumination/canopy coverage and substrate size) showed the preference of both waterweeds for water bodies with an average water depth of over 30 cm, clear to moderately turbid water, stagnant to moderately turbulent flow velocity and full illumination, with finer substrate (particles smaller than 2 cm) (Table 1).Furthermore, the difference between E. canadensis and E. nuttallii can be observed in their preferences with respect to flow velocity and substrate size.Elodea canadensis showed a preference for moderately turbulent watercourses (57% of records), whereas E. nuttallii was recorded more frequently in stagnant water (56% of records).Furthermore, E. nuttallii grew predominately in watercourses with clay and mud (81% of records), while E. canadensis grew on clay and mud (36%), sand and small pebbles (37%) and pebbles (17%).A slight difference in preference for average water depth can be seen as well-E.nuttallii was recorded evenly in water bodies with average depths of 30-100 cm and over 100 cm, while E. canadensis preferred water bodies with an average water depth of more than 100 cm in 71% of records.All of the characteristics were tested using the Mann-Whitney test.Here, substrate sizes were tested separately because, in each site, the substrate was evaluated as a fraction of each size category.The test showed that the sites with E. canadensis and E. nuttallii significantly differed in flow velocity (p = 0.023) and the content of the smallest substrate size (clay and mud) (p = 0.007).
Table 1.An analysis of average water depth, turbidity, flow velocity, illumination and substrate size in water bodies where waterweeds were recorded.A significant difference between sites with E. canadensis and E. nuttallii is indicated with an asterisk (Mann-Whitney test, p = 0.023 for flow velocity and p = 0.007 for clay and mud).

Accompanying Species
A total of 80 species were recorded alongside the two waterweeds during our study.Vegetation relevés are given in Table S1.Elodea canadensis was accompanied by 68 species and E. nuttallii by 46 species.Only three species frequently accompany both E. canadensis and E. nuttallii (in over 30% of relevés)-Ceratophyllum demersum, Myriophyllum spicatum and Sparganium erectum.Elodea canadensis is also frequently accompanied by Potamogeton nodosus, P. perfoliatus, P. pectinatus, P. crispus, Berula erecta, Mentha aquatica and Nuphar lutea.Elodea nuttallii, on the other hand, is most frequently accompanied by the free-floating Lemna minor, L. trisulca, Hydrocharis morsus-ranae and Spirodela polyrhiza alongside the helophytes Glyceria maxima, Phragmites australis and Alisma plantago-aquatica (Figure 2).Bryophytes were recorded in 19 relevés almost exclusively alongside E. canadensis, except for two relevés with only one bryophyte, Riccia fluitans, accompanying E. nuttallii.The most frequent bryophytes were Fontinalis antipyretica (24%) and Cinclidotus fontinaloides (13%), while the rest were recorded in less than 9% of relevés.Algae were represented by five stonewort species (Charales) in eight relevés, each with a frequency of up to 5%, and exclusively accompanying E. canadensis in our study (Table S1).

Water Chemical and Physico-Chemical Parameters
The results of the Mann-Whitney test showed a significant difference (p < 0.05) between sites of Elodea nuttallii and E. canadensis sites for six water parameters: total phosphorus, orthophosphates, ammonium, pH and chemical (COD) and biochemical (BOD) oxygen demands (Figure 3).The sites did not differ regarding total nitrogen (Ntot), nitrates (NO₃⁻), nitrites (NO₂⁻), dissolved oxygen (DO), oxygen saturation (sO₂), temperature (T) and conductivity (EC).Elodea nuttallii occurred in water with much higher BOD and COD, within ranges 1.5-6.6 mgO2/L and 1.9-8.7 mgO2/L respectively, whereas the median for sites of E. canadensis for BOD and COD lie approximately at the minimum of those values for E. nuttallii (1.07 mgO2/L for BOD and 2.03 for COD mgO2/L).Regarding pH, E. nuttallii showed a broader range but had a much lower median (7.66) in comparison with E. canadensis, whose median value of 7.96 exceeded the upper quartile of values for E. nuttallii.In terms of nutrient concentrations, sites of E. canadensis showed significantly lower values for ammonium, orthophosphate and, consequently, total phosphorus values.Elodea nuttallii was recorded in more eutrophic water, tolerating extreme values of total phosphorus up to 0.699 mgP/L and ammonium of 6.065 mgN/L (the latter value for ammonium was removed from the boxplot as an extreme outlier for visual purposes).All averages of environmental parameters of sample sites are given in Table S3.

Water Chemical and Physico-Chemical Parameters
The results of the Mann-Whitney test showed a significant difference (p < 0.05) between sites of Elodea nuttallii and E. canadensis sites for six water parameters: total phosphorus, orthophosphates, ammonium, pH and chemical (COD) and biochemical (BOD) oxygen demands (Figure 3).The sites did not differ regarding total nitrogen (Ntot), nitrates (NO 3 − ), nitrites (NO 2 − ), dissolved oxygen (DO), oxygen saturation (sO 2 ), temperature (T) and conductivity (EC).Elodea nuttallii occurred in water with much higher BOD and COD, within ranges 1.5-6.6 mgO 2 /L and 1.9-8.7 mgO 2 /L respectively, whereas the median for sites of E. canadensis for BOD and COD lie approximately at the minimum of those values for E. nuttallii (1.07 mgO 2 /L for BOD and 2.03 for COD mgO 2 /L).Regarding pH, E. nuttallii showed a broader range but had a much lower median (7.66) in comparison with E. canadensis, whose median value of 7.96 exceeded the upper quartile of values for E. nuttallii.In terms of nutrient concentrations, sites of E. canadensis showed significantly lower values for ammonium, orthophosphate and, consequently, total phosphorus values.Elodea nuttallii was recorded in more eutrophic water, tolerating extreme values of total phosphorus up to 0.699 mgP/L and ammonium of 6.065 mgN/L (the latter value for ammonium was removed from the boxplot as an extreme outlier for visual purposes).All averages of environmental parameters of sample sites are given in Table S3.Plants 2024, 13, 1624 7 of 16

Results of Ordination Analyses
Results of the CCA (Figure 4) revealed the distribution pattern of waterweeds and accompanying species.Forward selection of the 13 measured environmental variables (physico-chemical and chemical parameters; listed in Table S4) revealed the six most contributing variables explaining the species distribution in different sample sites-total nitrogen, total phosphorus, electrical conductivity, chemical oxygen demand, dissolved oxygen and pH.The main compositional gradient, which can be seen along the first axis, clearly follows the increase in nutrient loads (total nitrogen, total phosphorus, electrical conductivity) and chemical oxygen demand, as well as a decrease in pH value.Hence, the less disturbed sites of lower trophy levels (high dissolved oxygen and pH values) on the left half of the plot are in contrast to the nutrient-rich, eutrophic sites on the right.This analysis clearly separates E. canadensis and E. nuttallii in different ecological niches, at least as observed in Croatian watercourses.Elodea canadensis shows an affinity for less eutrophic waters, while E. nuttallii grows in nutrient-rich watercourses with a high trophy level.Accompanying species of waterweeds are distributed accordingly.Furthermore, COD highly correlated with BOD (rs = 0.83, p < 0.001), while BOD also correlated with NH 4 + (rs = 0.72, p < 0.001), PO 43− (rs = 0.72, p < 0.001) and Ptot (rs = 0.79, p < 0.001).Dissolved oxygen was highly correlated with SO 2 (rs = 0.96, p < 0.001).

Results of Ordination Analyses
Results of the CCA (Figure 4) revealed the distribution pattern of waterweeds and accompanying species.Forward selection of the 13 measured environmental variables (physico-chemical and chemical parameters; listed in Table S4) revealed the six most contributing variables explaining the species distribution in different sample sites-total nitrogen, total phosphorus, electrical conductivity, chemical oxygen demand, dissolved oxygen and pH.The main compositional gradient, which can be seen along the first axis, clearly follows the increase in nutrient loads (total nitrogen, total phosphorus, electrical conductivity) and chemical oxygen demand, as well as a decrease in pH value.Hence, the less disturbed sites of lower trophy levels (high dissolved oxygen and pH values) on the left half of the plot are in contrast to the nutrient-rich, eutrophic sites on the right.This analysis clearly separates E. canadensis and E. nuttallii in different ecological niches, at least as observed in Croatian watercourses.Elodea canadensis shows an affinity for less eutrophic waters, while E. nuttallii grows in nutrient-rich watercourses with a high trophy level.Accompanying species of waterweeds are distributed accordingly.Furthermore, COD highly correlated with BOD (rs = 0.83, p < 0.001), while BOD also correlated with NH₄⁺ (rs = 0.72, p < 0.001), PO₄ 3 ⁻ (rs = 0.72, p < 0.001) and Ptot (rs = 0.79, p < 0.001).Dissolved oxygen was highly correlated with SO₂ (rs = 0.96, p < 0.001).The eigenvalues for the first and second axes equalled 0.41 and 0.14, respectively.Overall analysis was statistically significant (p < 0.002), which was confirmed by the Monte Carlo test (499 permutations).For environmental variables abbreviations, see the caption of Figure 3.For abbreviations of species' names, see Table S1.

Nonparametric Multiplicative Regression Analysis
NPMR model identified the three most contributing predictors of the distribution of E. canadensis and E. nuttallii-biochemical oxygen demand, conductivity and total phosphorus (Figure 5, Table 2).Increasing values of the most relevant factor, total phosphorus, leads to a gradual increase in the relative abundance (RA) of E. nuttallii and a decrease in the RA of E. canadensis.Elodea nuttallii becomes relatively more abundant above a value Plants 2024, 13, 1624 8 of 16 of approximately 0.17 mgP/L of total phosphorus, which corresponds to hypereutrophic water bodies [25].A similar relationship of the species' RA can be seen with the increasing values of BOD.Interestingly, E. canadensis shows a bimodal response to conductivity, thriving in habitats with conductivity values of around 400 and 700 µS/cm but decreasing in abundance at 500-600 µS/cm and even more drastically after 700 µS/cm.The RA of E. nuttallii, on the other hand, increases gradually, with an increase in conductivity from 400 µS/cm onwards, and after 700 µS/cm it surpasses E. canadensis in abundance.

Nonparametric Multiplicative Regression Analysis
NPMR model identified the three most contributing predictors of the distribution of E. canadensis and E. nuttallii-biochemical oxygen demand, conductivity and total phosphorus (Figure 5, Table 2).Increasing values of the most relevant factor, total phosphorus, leads to a gradual increase in the relative abundance (RA) of E. nuttallii and a decrease in the RA of E. canadensis.Elodea nuttallii becomes relatively more abundant above a value of approximately 0.17 mgP/L of total phosphorus, which corresponds to hypereutrophic water bodies [25].A similar relationship of the species' RA can be seen with the increasing values of BOD.Interestingly, E. canadensis shows a bimodal response to conductivity, thriving in habitats with conductivity values of around 400 and 700 µS/cm but decreasing in abundance at 500-600 µS/cm and even more drastically after 700 µS/cm.The RA of E. nuttallii, on the other hand, increases gradually, with an increase in conductivity from 400 µS/cm onwards, and after 700 µS/cm it surpasses E. canadensis in abundance.

Discussion
Elodea canadensis was found for the first time in Croatia in the Pannonian Ecoregion (near the city of Sisak) in the late 19th century.The distribution of E. canadensis was further explored in the following century, but more sporadically than systematically [26][27][28][29][30].It was not until the 2010s when E. canadensis was targeted as an invasive species that its distribution was more thoroughly investigated as a part of the inventory of invasive species [31] in Croatia.This information was deepened with the implementation of the WFD and the corresponding monitoring.Elodea canadensis was recorded across the Pannonian Ecoregion-in the basins of the rivers Sava, Drava and Danube.It was also found in the Continental-Dinaric Subecoregion for the first time in 2016, in the rivers Mrežnica and Dobra, both belonging to the Sava River Basin.The population of E. canadensis in these two sampling sites seems to be stable as has been confirmed for three and four consecutive years, respectively, in low abundances.Furthermore, in 2017, E. canadensis was also recorded in a canal connected to the Cetina River (Mediterranean Subecoregion) and subsequently in four locations along the river.Since there are not many tributaries, the species was presumably introduced in the Cetina either by human (disposal of aquarium contents) or bird activity.
A similar absence of E. nuttallii and isolated, scattered occurrences of E. canadensis in the Mediterranean biogeographical region of Italy was recorded by Buldrini et al. [32] who suggest that E. nuttallii is the less thermophilous of the two.A survey from neighbouring Serbia [33] found that most waterweed records have been documented in surface running waters as opposed to standing waters.This is also the case in our survey, even though in other parts of Europe, waterweeds are commonly recorded and studied in lakes [6,10,34,35].The aforementioned new records of E. nuttallii in the Drava River Basin from 2019 are especially interesting since they match the monitoring sites where, only two years before, E. canadensis was observed.In these sites, E. nuttallii seemingly displaced E. canadensis, a possible scenario Kočić et al. [9] had predicted earlier for some water bodies.These sites were the reservoirs and two canals.Interestingly, E. canadensis is dispersed more widely in Europe and has been the more dominant species since it was introduced approximately a century before E. nuttallii [16], however, multiple research projects [11,14,[35][36][37] showed that E. nuttallii spreads rapidly and consequently displaces E. canadensis in some areas.Assumed mechanisms of this displacement will be discussed further as these two species were first defined as ecologically and functionally redundant by having similar biological trait combinations and similar ecological responses, precisely the growth rate, to various environmental conditions and their combinations [38].
A study from Slovenia [39] found that E. canadensis was absent from the surveyed standing water bodies, whereas E. nuttallii, present in all water body types, dominated the ponds.Elodea nuttallii is poorly resistant to mechanical stress induced by water turbulence [40] and shows its invasive nature more in parts of the Drava River away from the main water flow [41].Kuhar et al. [7] reported that E. canadensis is absent from water bodies with the most dynamic flow and prefers low-velocity currents in Slovenia and Thiébaut [13] reported similarly from north-eastern France.This is in accordance with our findings that E. nuttallii prefers stagnant flow when in running waters, and that neither of the two elodeas inhabit fast-flowing rivers.
Both waterweeds grow in water bodies with more sediment of smaller size; however, E. nuttallii seems to prefer water bodies with finer sediment in contrast to E. canadensis, which is equally found in those with coarser substrates such as sand and gravel.Accordingly, Kuhar et al. [7] report the preference of E. canadensis for a mixture of gravel, sand and silt and Grudnik et al. [41] assume that the deposition of silt might contribute to the further expansion of E. nuttallii in reservoirs.Similarly, Crane et al. [35] report on a higher proportional cover of E. nuttallii on finer substrate, while Thiébaut [13] found E. nuttallii in habitats with high sand sedimentation.In Ukraine, in the central part of the Dnipro River Basin, E. nuttallii grows on sandy and muddy sediment, whereas E. canadensis grows on various substrates-sand, muddy sand, silt and rocky sediment [42].In contrast, Bubíková et al. [43] recorded both elodeas mostly on coarse and sandy substrate as opposed to gravel in the surface waters of Slovakia.
Not many studies examined the relationship between water depth and Elodea occurrence/abundance.The reason why E. canadensis was found in waterbodies with deeper water might lie in the fact that it was more often found in rivers as opposed to canals which are usually shallower.Barrat-Segretain and Cellot [44] found that E. nuttallii tolerates droughts for longer periods than E. canadensis.This might result in E. nuttallii inhabiting shallower water bodies than E. canadensis as shallower water bodies (in the same ecoregions) are more prone to drawdowns during droughts.However, contrasting results come from Prokopuk and Zub [42] who recorded E. canadensis in somewhat shallower water (0.1-1.0 m) than E. nuttallii (0.2-1.5 m).
In the Greek part of Lake Prespa, E. canadensis was found accompanying Myriophyllum spicatum, Trapa natans, Potamogeton perfoliatus, Ceratophyllum demersum, etc. [45].Such a list of accompanying species is in line with the high frequency of E. canadensis within the stands belonging to the vegetation of the class Potamogetonetea, as documented in Czechia [43].Hence, it is not surprising that alongside E. canadensis, the most commonly recorded species in our survey were Potamogeton spp., and that P. berchtoldii, P. crispus, P. nodosus, P. perfoliatus were indicative of the sites with E. canadensis according to IAS analysis.On the other hand, the most common accompanying species of E. nuttallii were free-floating species, e.g., Lemna spp., Spirodela polyrhiza, Hydrocharis morsus-ranae inhabiting canals characterised by very slow flow, finer sediment and higher anthropogenic influence.These species were singled out as indicative of sites with E. nuttallii, alongside some less frequent species, such as Salvinia natans and Nymphoides peltata, and helophytes Glyceria maxima, Phragmites australis and Typha latifolia.It is widely known that free-floating plants belonging to vegetation of the class Lemnetaea often inhabit agricultural ditches and canals and are driven by nutrient enrichment, both nitrogen and phosphorus [46].Our study confirmed that such habitats are also preferred by E. nuttallii in Croatia.Bryophytes and algae were not very frequent in the relevés, however, their distribution along the environmental gradient showed that they preferred sites with more oxygenated water, with higher pH and lower nutrient values, i.e., the sites with E. canadensis in our study.One of the bryophyte species, Fontinalis antipyretica was even shown to be indicative of sites with E. Canadensis within our study.The only species that accompanied E. nuttallii was Riccia fluitans.Rimac et al. [47] showed that among the water bryophytes in Croatia, only R. fluitans and Leptodyctium riparium had an affinity for hypereutrophic water with neutral pH, high electrical conductivity and organic matter content, although having wide niches considering these environmental parameters.Vanderpoorten et al. [48] also found that L. riparium exhibits a wide ecological range and is thus not a reliable indicator regarding trophy levels.Interestingly, in our study, L. riparium was not associated with hypereutrophic and eutrophic sites, but rather with oligo-and mesotrophic water of E. canadesis sites.Similarly, the stoneworts, known to have an affinity for oligotrophic conditions, which they help maintain by controlling nutrient cycles [49], were exclusively found on E. canadensis sites within our study.
Canonical correspondence analysis separated E. nuttallii and E. canadensis with their respective co-occurring species in terms of environmental niches.Elodea canadensis showed an affinity for more oxygenated water bodies with higher pH, as did its accompanying species.By contrast, E. nuttallii and its accompanying species showed a preference for water bodies with higher electrical conductivity, greater nutrient load (Ptot, Ntot) and higher chemical oxygen demand.The presence of E. canadensis, rather than of E. nuttallii, in sites with higher water pH might be explained by the fact that the photosynthesis rate of E. nuttallii reduces at pH > 7 [50].This could also be the reason why E. nuttallii is not recorded in the oligo-mesotrophic and oxygenated waters of the Dinaric Ecoregion where the pH is more alkaline due to the karstic matrix.Even though both waterweeds show high growth rates and high tolerance to wide ranges of environmental variables, along with other ecological mechanisms of invasion (enemy release, allelopathy, phenotypic plasticity) [1] and are considered ecological equals, it has become more evident that E. nuttallii outcompetes E. canadensis in more eutrophic conditions, and the reasons behind that phenomenon are becoming more and more clear.Other than the previously mentioned higher tolerance to drawdowns [44], E. nuttallii has a higher relative growth in nitrogen-enriched water as shown by Ozimek et al. [34] and grows quicker in shadier and hypertrophic conditions, which allows it to outgrow and outcompete other species [12].NPMR analysis supports the CCA by indicating that the most important environmental predictors of the distribution of waterweeds are the amount of total phosphorus, biochemical oxygen demand and water electrical conductivity, with high values indicating eutrophy.
In our study, the displacement of E. canadensis by E. nuttallii was documented in five sampling sites, while the complete corresponding water chemistry was available only for two sites.These were characterized by high concentrations of orthophosphates and total phosphorus and ammonium, nitrates and total nitrogen (Table S3).These findings might suggest that E. nuttallii is more competitive in hypereutrophic situations and in general in those with a greater nutrient load.Furthermore, the possible displacement of E. canadensis by E. nuttallii was observed in our study on three reservoirs on the Drava River which might be due to the better tolerance of E. nuttallii to the drawdowns as well as the fast growth early in vegetation season underpinning its successful competition with other submerged species for light and nutrients.This was already suggested by Kunii [51], who found that E. nuttallii can grow even in the winter if the water temperature is above 4 • C. The invasion of E. nuttallii was documented earlier in reservoirs in the upper reach of the Drava in Slovenia, situated upstream from Croatia [41].Here, the species developed large biomass in parts of the watercourse unexposed to the main current in warmer years with high winter and spring water temperatures.Furthermore, Wang et al. [52] found that E. nuttallii showed superior advantages in terms of growth, length and an increase in shoot number in late winter and spring over other alien species in China, such as Egeria densa.Both Elodea species are recorded alongside a number of native species, and they do not dominate the vegetation relevés in the watercourses.However, in the artificial impoundments, the contrary was noticed and these habitats could become major expansion hotspots [41].Since E. nuttallii is listed as an invasive alien species of Union concern as of 2017 [53], future trends of its spread in Croatia should closely be monitored.

Materials and Methods
Elodea canadensis and E. nuttallii were recorded alongside other macrophytes within the national surface water monitoring scheme conducted from 2016 to 2023.Macrophyte vegetation is monitored to assess the ecological status of water bodies, as required by the Water Framework Directive (WFD) [54].Sampling was performed at 786 sites according to the national methodology for macrophyte sampling [21].Watercourses are surveyed along 100-metre-long transects from the banks or by zigzagging across the riverbed if possible.Standing water bodies are surveyed from a boat and also by walking along the banks.If unattainable by eye or hand, macrophytes are sampled with rakes at the end of a long pole or tied to a rope.Distribution maps were created using QGIS Desktop 3.4.9software.Vegetation relevés are recorded following the extended Braun-Blanquet scale (r = one individual, + = up to 5 individuals, 1 = up to 50 individuals, 2 m = over 50 individuals, coverage < 5%, 2a = coverage 5-15%, 2b = coverage 15-25%, 3 =25-50%; 4 = coverage 50-75%; 5 = coverage over 75%) [55][56][57].The nomenclature follows Euro + Med [58] for vascular plants, Hodgetts et al. [59] for bryophytes and AlgaeBase [60] for algae.
According to the typology developed as a basis for the monitoring of surface waters [21], the territory of Croatia (56,594 km 2 ) is divided into two hydrological and biogeographical regions-the Pannonian and the Dinaric Ecoregion, the latter being subdivided into Continental and Mediterranean Subecoregions.The Pannonian Ecoregion is situated in the continental part of Croatia between three large rivers-Sava, Drava and Danube, so its watercourses belong exclusively to the Black Sea Basin.It is composed of low-altitude alluvial and diluvial plains in between low, solitary mountain massifs.Geologically and lithologically, the Pannonian Ecoregion is characterised by silicate Quaternary deposits, while limestone is found only in the highest mountain areas.The climate is temperate, without a dry season, with warm summers in most of the territory (Cfb) and hot summers predominantly in the eastern part (Cfa).The Dinaric Ecoregion is characterised by the Dinarides, the largest uninterrupted karst system in Europe, so it is predominantly built of limestone and dolomite bedrock.Many of its rivers exhibit partly subterranean courses, and some of them in the Mediterranean-Dinaric Subecoregion also periodically dry out over the summer season.This subecoregion is characterized by temperate Mediterranean climate, with dry and hot summer months (Csa), while the Continental-Dinaric Subecoregion is characterised by a continental climate (Cfb) and constant river discharge levels [61,62].Information on each sampling site (Table S5) also comprised water chemical and physico-chemical parameters of water, as well as habitat characters of the water body, listed in Table 3.Chemical and physico-chemical parameters were measured monthly by an accredited laboratory (Central Water Management Laboratory, Zagreb) and their annual average was used in the analyses.Dissolved oxygen, oxygen saturation, temperature, pH and electrical conductivity were measured in situ with a Hach HQ40D Portable Multi Meter (Hach, Loveland, CO, USA) under standard conditions, while chemical parameters were analysed in the laboratory from water samples.Habitat characters were assessed in situ only for running waters (rivers and artificial canals).Basic descriptive statistics of these habitat characters were calculated.The distribution of E. canadensis and E. nuttallii along the gradient of chemical and physico-chemical parameters was shown with box-plot graphs made using Past 4.16 software [63].Furthermore, a significant difference between the parameters was tested for each environmental variable with the Mann-Whitney pairwise post hoc test.Non-parametric Mann-Whitney test was chosen because the majority of the variables did not have a normal distribution, previously tested with the Shapiro-Wilk test.Both tests were performed in Past 4.16 software.
The frequency of accompanying species was calculated for the sites with E. canadensis and those with E. nuttallii to explore the differences.Furthermore, a non-parametric ANOSIM test (Analysis of Similarities) was used to test the significant difference of sites with E. canadensis and E. nuttallii regarding the total species composition and their abundances.Indicators Species Analysis (ISA) was performed to identify species characteristic of each group of sites, i.e., the species group with E. canadensis and that with E. nuttallii.The statistical significances of the species indicator values were estimated by 9999 random reassignments (permutations) of sites across groups.Both analyses were performed in Past 4.16 software.
To explore the relationship between the environmental variables and patterns in the distribution of waterweeds and accompanying species, a direct ordination method, canonical correspondence analysis (CCA), was used.After removing the outliers, vegetation and environmental data from 50 localities were included in the analysis.CCA was selected because the response data were compositional with a 3.6 SD unit-long gradient [64].A step-forward selection procedure in CANOCO 5 [64,65] was used to identify the most contributing and nonredundant subset of environmental predictors influencing the investigated species.Six variables with the highest conditional effect and with a 5% significance cut level (p < 0.05; Monte Carlo test, 499 permutations) were included.Before the analysis, rare species were down-weighted.Ecological niches of the species regarding the measured water parameters were explored using nonparametric multiplicative regression analysis (NPMR) performed in Hy-perNiche V2.3 [66].To create multiple best models, a local mean with Gaussian weighting for both the response (species relative abundance) and the predictors (13 measured physicochemical parameters) was used.A stepwise free search was employed to automatically find the best models with different combinations of environmental variables and by adopting default values for all remaining search criteria.For validation of the model, the predictive quality of the cross-R2 (×R2) was calculated based on the residual sum of squares (RSS), divided by the total sum of squares (TSS).The value of ×R2 ranges from 0 to 1, indicating no relation and perfect fit, respectively.The best model for each species was selected on account of the additional predictor (i.e., environmental) variable resulting in only a minimal increase (< 5%) of ×R2.The relative importance of a particular predictor within the model was evaluated using sensitivity analysis.The sensitivity of the model was estimated by calculating the tolerance as a proportion of variable range.Sensitivity values range from 0 to 1 and greater sensitivity indicates a higher influence of a particular variable in the model [67].The significance of the selected models was tested using a Monte Carlo randomization test with 100 runs to calculate the probability value [67].

Conclusions
In Croatia, E. canadensis has historically been present much longer and is more widespread than E. nuttallii.It inhabits more sites in more water bodies in all three (sub)ecoregions.It shows an affinity for illuminated mesotrophic to eutrophic, moderately turbulent rivers, where it grows on different substrates in various depths.Most often, it grows with macrophytes such as Potamogeton spp., Myriophyllum spicatum, Berula erecta, etc., whereas E. nuttallii accompanies free-floating species such as Lemna spp., Spirodela polyrhiza, Hydrocharis morsus-ranae and submerged Ceratophyllum demersum.E. nuttallii is still poorly recorded in Croatia.It mostly inhabits the shallower, eutrophic, stagnant to slow-flowing canals and impoundments characterised by soft sediment in the Pannonian Ecoregion.
The replacement of E. canadensis with E. nuttallii was observed in several water bodies with high nutrient loads, which may be related to the latter species greater tolerance of nutrient-rich and oxygen-depleted conditions.Future monitoring of these two species is expected to reveal further expansion of E. nuttallii in the eutrophic water bodies of the Pannonian Ecoregion of Croatia, especially in the case of further eutrophication.

Figure 4 .
Figure 4. CCA ordination plot.The eigenvalues for the first and second axes equalled 0.41 and 0.14, respectively.Overall analysis was statistically significant (p < 0.002), which was confirmed by the

Figure 4 .
Figure 4. CCA ordination plot.The eigenvalues for the first and second axes equalled 0.41 and 0.14, respectively.Overall analysis was statistically significant (p < 0.002), which was confirmed by the Monte Carlo test (499 permutations).For environmental variables abbreviations, see the caption of Figure3.For abbreviations of species' names, see TableS1.

Table 2 .
Summary of NPMR models for Elodea species to environmental predictors of highest explanatory power.

Table 3 .
A list of measured chemical and physico-chemical parameters and assessed habitat characters with accompanying abbreviations and measurement units or categories.