DISCUSSION
The extent to which diversity-related processes (diversity-invasibility hypothesis 1 ), abiotic factors (environmental-matching hypothesis 2 ) and habitat heterogeneity (invasion paradox hypothesis 3 ) influence the spread of non-native aquatic macrophytes in space and time in European lakes has received little empirical attention. We use E. canadensis as a model system in which to address this general question using survey and palaeolimnology data to partial out the significance of different diversity signatures and abiotic predictors. This enabled us to test three competing hypotheses and to demonstrate that habitat heterogeneity has played a defining role in driving variation in E. canadensisabundance over space and time.
We found a nested spatial dependence between E. canadensisabundance and the location of sampling points in each lake (Fig. 4). This spatial structure in the data suggests that besides water clarity and depth there might be other unmeasured within-lake abiotic factors (e.g. surface area and basin morphology) that also influence E. canadensis abundance. The influence of this spatial structure in determining E. canadensis abundance diminished with lake isolation. Such a spatial pattern is in line with our previous studies of the ULE system showing that eutrophication has spread unevenly across the lakes (Salgado et al. 2018a; Salgado et al. 2019). In particular, native macrophyte assemblages in the more eutrophic and isolated lakes of Group 3 have become more homogenous over time, unlike those lakes with a higher degree of connectivity to the central lake (Salgado et al. 2018a). Macrophyte assemblages undergoing such homogenisation may have reduced effects on E. canadensisabundance arising from any unique lake properties.
After accounting for spatial autocorrelation in the data we found that variation in water clarity and diversity measures among the lakes together explained other significant portions of variation in E. canadensis abundance. In lakes with greater hydrological connectivity (Groups 1 and 2 ), E. canadensis abundances were less influenced by water clarity than by beta diversity. This pattern may be explained by higher prevalence of zebra mussels in these more connected lakes (and consequently improved water clarity; Salgado et al. 2019a), and source sink dynamics that maintain heterogeneous plant associations at the local scale (Salgado et al. 2018a,b). In contrast, E. canadensis abundances in more isolated headwaters (Group 3 ) were strongly impacted by low water clarity, floating plant cover and native Shannon diversity (mostly driven by Digh Lough). These drivers are consistent with biotic homogenisation in these more degraded lakes where species-sorting processes previously dominated (Salgado et al. 2018a).
Much of the variation in E. canadensis abundance over space and time was related to the interplay of beta diversity and plant cover. These two native plant community attributes have been found to better capture macrophyte ecological change dynamics in other human-dominated landscapes than species richness alone (Capers et al. 2007; Fu et al. 2019). The consistently positive correlation between E. canadensis abundance and native species richness in each lake over space and time suggests that in the ULE system, native species richness does not confer invasibility resistance. Instead, it suggests that native and non-native species respond similarly to within-lake and regional habitat heterogeneity. This overwhelms any negative effects of local diversity on E. canadensis, thus native and non-native species coexist across the landscape (Levine 2000; Jiang & Morin 2004; Davis et al. 2005). Although causal relationships are difficult to discern from our observational data, we propose that locations favourable for native macrophytes, may generally therefore also favourE. canadensis (Melbourne et al. 2007).
While the presence of diverse native macrophyte species increased the probability of occurrence of E. canadensis within the lakes, relationships with macrophyte beta diversity and plant cover suggest that E. canadensis abundance is be limited by high plant cover. This pattern was also revealed through time for lake Groups 1 and2 and in regional-scale plant invasion studies (Cleland et al. 2004; Capers et al. 2007; Davies et al. 2007). The positive association between plant cover and E. canadensisthrough time in Group 3 lakes, likely reflects the overall responses of macrophyte communities in these lakes to more advanced eutrophication. Different processes may therefore control the establishment vs. proliferation of E. canadensis . For example, spatial and temporal environmental heterogeneity may facilitate establishment and coexistence by providing a range of opportunities forE. canadensis to invade (Davis et al. 2005; Clark et al. 2013). However, high native plant cover could lower resource availability for a non-native colonist, thereby reducing opportunities to proliferate (Cleland et al. 2004; Davis et al. 2007).
Overall, our data support hypotheses 3. Despite observingE. canadensis in all but one site, abundances in lakes were generally low to moderate. This finding, coupled with a positive relationship between E. canadensis and speciose communities over space and time, indicate sufficient environmental heterogeneity within and among-lakes to enable coexistence of native and non-native macrophytes (Clark et al. 2013) and hence, reduced the often described macrophyte homogenisation impacts of non-native species (Muthukrishnan & Larkin 2020). Habitat heterogeneity was also related to macrophyte beta diversity and plant cover variation.
Inferring the history of E. canadensis invasion
Palaeolimnological data reveal that at the time, that E. canadensis colonised the ULE system in the late 1800s, macrophyte communities were similar to those currently observed in Group 1lakes (Salgado et al. 2019). Simpson (1984) reported a cycle of local colonization by E. canadensis involving establishment over a three-year period and a subsequent rapid increase in abundance. Given the extensive interconnection by winter flooding E. canadensisprobably spread rapidly to many sites. Following its widespread establishment, E. canadensis probably therefore persisted at moderate abundances for a long period, co-existing with a reported high diversity of other submerged species across the lakes (Salgado et al. 2018a). Subsequently, post-1950s, paleoecological data indicate gradual biotic changes associated with more eutrophic conditions that intensified after the 1980s (Battarbee 1986) but with differential local nutrient concentrations influencing biota (Salgado et al. 2019). These post-1950s biotic shifts involved gradual increases in floating plant cover and dominance of fine-leaved Potamogeton species (Fig. 5), although some sites, such as Castle Lough and Mill Lough, have maintained high macrophyte species diversities and abundances. Reductions in E. canadensis abundance in the most degraded lakes of Group 3 and negative associations with plant cover suggest a gradual decline in abundance across the ULE system over the last three to four decades leading to its current status of being widespread but seldom very abundant.
Reconstructing E. canadensis abundance over time based on survey and sediment core data may have limitations. For instance, some species may have been unrecorded and detection in sediment cores may be biased by preservation issues and under-representation of rare or distantly located macrophyte taxa (Zhao et al. 2006; Clarke et al. 2014). Our assessments of native macrophyte richness variation over space and time probably favour the more abundant taxa. Unique lake histories could have also introduced some discrepancies between the observed currentE. canadensis occurrence and the inferred past abundance (Bennion et al. 2018). Nevertheless, analyses of both palaeo- and contemporary data showed a consistent positive relationship betweenE. canadensis abundance and native plant richness in most lakes. Plant cover, beta diversity, and Shannon diversity similarly emerged as the main predictors in explaining the temporal variation of E. canadensis abundance. These lines of evidence coupled with the history of Elodea spread in the British Isles (Simpson 1984) thus allow us to hypothesize what the general long-term patterns of E. canadensis spread in the ULE system would as described above.
Conclusions and perspectives
E. canadensis is commonly reported in the invasion literature to dominate over native submerged species once well established, and to exert strong negative ecosystem engineering effects (Rørslett et al. 1986; Schwarz et al. 2015). Conditions considered to favourE. canadensis include high availability of nutrients, suitable carbon sources, and silty substrates (Schwarz et al. 2015). Such conditions generally prevailed across our study sites (Salgado et al. 2019) and would support the environmental-matchinghypothesis 2 . However, our data provide a novel demonstration that habitat heterogeneity enables the coexistence of native and non-native plant species at landscape scales. We also found evidence for invasion resistance in stressful environments or when native plant cover of native species is high.
Predicting future trajectories of E. canadensis distribution and abundance is challenging. E. canadensis has spread in the British Isles by asexual growth, from one giant male clone (Simpson 1984). It is therefore possible that conditions (e.g. disease) may eventually challenge the persistence of clonal populations. Furthermore, with globalization, unexpected and novel invasion dynamics are becoming ever likelier (Pyšek et al. 2020). Meanwhile, like many other shallow lake systems, the ULE system is declining through advancing eutrophication, which, if unabated, will eventually override positive regional species storage effects (Salgado et al. 2018a). These impacts are likely to accelerate in forthcoming decades due to climate-driven magnification of eutrophication effects in lakes (Birk et al. 2020). In addition, the sibling invasive species, Elodea nuttallii, is rapidly spreading across the ULE system and outcompeting E. canadensisunder high nutrient-enrichment conditions (Kelly et al. 2015). Quantifying the dynamics of these two invasive species at both landscape and temporal scales is critical, therefore, if invasion processes are to be better understood.