INTRODUCTION
The ability of a non-native plant species to invade depends on resource availability, natural enemies and the physical environment (Shea & Chesson 2002). Accordingly, three hypotheses have been proposed to explain invasion success (Jeschke 2014): (1) The diversity-invasibility hypothesis (Elton 1958; Kennedy et al. 2002; Levine et al. 2004) posits that species- and functionally-rich communities limit establishment opportunities for invaders by reducing access to resources; (2) The environmental-matching hypothesis(Mack et al. 2000; Shwartz et al. 2009) postulates that, irrespective of local native biota, plant species will have higher invasion success if introduced to environments similar to those in their natural optimal range; (3) The invasion paradox (Levine & D’Antonio 1999; Fridley et al. 2007) proposes that invasive species spread over time through native metacommunities (communities linked by dispersal) in a heterogeneous landscape, leading to the regional coexistence of both native and non-native plant species. We subsequently refer to these as invasion hypotheses 1, 2 and 3 , respectively.
Processes affecting biotic interactions and species distributions are scale-dependent (McGill 2010). At the landscape scale, both habitat quality and heterogeneity (defined here as the spatial and temporal variation in biotic and abiotic conditions) within and between lakes may facilitate species coexistence, regardless of their native status (Shea & Chesson 2002; Davies et al. 2005). Habitat heterogeneity therefore has the potential to concurrently increase invasion probability whilst reducing invasion impact by promoting coexistence in space and time that might be precluded in a more homogenous setting (Melbourne et al. 2007; Clark et al. 2013). Thus, biological assemblages in a given landscape may include species that are extirpated at some sites but present at others through spatial and/or temporal storage effects, provided there is sufficient regional connectivity and spatiotemporal habitat heterogeneity (Melbourne et al. 2007).
In lake landscapes, macrophyte species distributions may be determined by the hydrological network (Salgado et al. 2018a). For example, communities in hydrologically-connected lakes may be particularly influenced by repeated colonisation events; i.e. mass effects (Capers et al. 2010) while local environmental factors may dictate community structure through species sorting according to habitat optima in more isolated lakes (Salgado et al. 2019). Habitat heterogeneity may therefore be spatially autocorrelated with hydrological connectivity. Although, there have been attempts to quantify how spatial autocorrelation affects invasibility dynamics (e.g. Davies et al. 2005), disentangling the simultaneous effects of abiotic factors, biodiversity and spatially-structured dynamics on invasion processes in nature has proved challenging (von Holle 2013; Nunez-Mir et al. 2017). There is evidence that temporal variation in environmental stress factors and dispersal-related mechanisms may promote the co-existence of both native and non-native species in lake landscapes (Clark et al. 2013), and that this may explain a positive correlation between native diversity and exotic macrophyte abundance at any one time (Capers et al. 2007). Systems comprising multiple sites that span environmental gradients offer the possibility of more explicitly quantifying how spatial autocorrelation affects invasibility dynamics and hence scope to disentangle factors that contribute to regional coexistence.
Canadian pondweed (Elodea canadensis ) is considered amongst the most widespread non-native plant species in Europe (Hussner 2012; Nentwig et al. 2018). It was first recorded in Great Britain in 1836 (Simpson 1984). Thereafter it spread rapidly, reaching the maximum extent of its distribution in Great Britain and Ireland by the middle of the twentieth century (Simpson 1984). The rapid colonization and spread of this species is commonly attributed to a high capacity for vegetative propagation (Barrat-Segretain 1996) and tolerance of a broad range of physical-chemical conditions, including low illumination, enabling growth at a wide range of water depths and under eutrophication-induced shade (Abernethy et al. 1996; Zehnsdorf et al. 2015). Once established E. canadensis can quickly replace native submerged macrophytes by forming a dense, closed canopy (Howard-Williams et al. 1987; de Winton & Clayton 1996; Zehnsdorf et al. 2015). Indeed, the propensity for encountering E. canadensis in eutrophic isolated temperate lakes has promoted the view that its spread and colonisation across Britain and Ireland is attributable to environmental–matching; i.e. hypothesis 2 above (O’Hare et al. 2012). However, few studies have investigated the patterns of spread of E. canadensisover space and time [in a multivariate environmental context].
Here we examine the drivers of E. canadensis abundance in space and time by focusing on the macrophyte-rich (> 40 submerged and floating-leaved water plant species), and environmentally-heterogeneous lowland shallow lake complex of the Upper Lough Erne (ULE) system in Northern Ireland. Present-day and historical data derived from surveys and sediment core analyses enable us to address the following questions:
To what extent do diversity-related factors (invasion hypothesis 1 ), abiotic factors (invasion hypothesis 2 ) and a regional gradient in lake habitat quality (invasion hypothesis 3 ) contribute to variation in E. canadensis abundance?
Does habitat heterogeneity (including spatial autocorrelation)per se promote both E. canadensis invasibility and coexistence with native macrophyte communities in space and time (decades-centuries)?
Answering these questions provides novel demonstration that both habitat heterogeneity and habitat quality influence the coexistence of native and non-native plants at landscape scales. We also provide evidence that invasion resistance is linked to stressful environments and when high native plant cover limits opportunities for E. canadensis to proliferate.