Ecological communities often exhibit greater resistance to biological invasions when these communities consist of species that are not closely related. The effective size of this resistance, however, varies geographically. Here we investigate the drivers of this heterogeneity in the context of known contributions of native trees to the resistance of forests in the eastern United States of America to plant invasions. Using 42,626 spatially referenced forest community observations, we quantified spatial heterogeneity in relationships between evolutionary relatedness amongst native trees and both invasive plant species richness and cover. We then modelled the variability amongst the 91 ecological sections of our study area in the slopes of these relationships in response to three factors known to affect invasion and evolutionary relationships –environmental harshness (as estimated via tree height), relative tree density and environmental variability. Invasive species richness and cover declined in plots having less evolutionarily related native trees. The degree to which they did, however, varied considerably amongst ecological sections. This variability was explained by an ecological section’s mean maximum tree height and, to a lesser degree, SD in maximum tree height (R²_GLMM = 0.47 to 0.63). In general, less evolutionarily related native tree communities better resisted overall plant invasions in less harsh forests and in forests where the degree of harshness was more homogenous. These findings can guide future investigations aimed at identifying the mechanisms by which evolutionary relatedness of native species affects exotic species invasions and the environmental conditions under which these effects are most pronounced.

Environmental harshness, environmental variability, evolutionary divergence, forests, invasive plants, phylogeny, relative tree density

Observations across large geographic areas reveal considerable spatial heterogeneity in the degree to which ecological communities are invaded by non-native species (Stohlgren et al. 2006, Liebhold et al. 2013, Iannone et al. 2015). Spatial variability in the degree to which various ecological factors affect invasions partially explains this heterogeneity (e.g. Riitters et al. 2018). Identifying what drives this heterogeneity is of importance to both basic ecological understanding and land management aimed at controlling invasive species. Analyses of increasingly available large-scale forest data have revealed that forests exhibit large-scale spatial heterogeneity in both the number and dominance of invasive plant species that are present within them (Iannone et al. 2015, Oswalt et al. 2016). Understanding what causes this heterogeneity will improve basic understanding of forest plant invasions and help to protect the many ecosystem services that forests provide from threats posed by invasive plants (Gonzalez et al. 2005, Martin et al. 2009, Pejchar and Mooney 2009).

Analyses of the same large-scale forest data have also revealed evidence that native tree communities contribute to the ability of forests to resist plant invasions in general, i.e. regardless of invader growth form, and that these contributions vary spatially (Guo et al. 2015, Iannone et al. 2016). The characteristics of native tree communities most negatively related to plant invasions (i.e. invasive plant species richness and dominance) include aboveground native tree biomass and the degree to which native tree species are evolutionarily related (Iannone et al. 2016). The detected contribution of native tree biomass to invasion resistance likely reflects the fact that trees are a leading component of forest biomass, primary productivity and resource uptake (Muller 2003). The contribution of evolutionary relatedness of native trees to invasion resistance likely reflects the assumption that communities containing less evolutionarily related species exhibit greater diversity in both measurable functional traits and unidentified phenotypic traits and, thus, a greater breadth in niche usage (Faith 1992, Webb et al. 2002, Lososová et al. 2015, Tan et al. 2015, but see Gerhold et al. 2015).

The objective of this investigation was to determine the degree to which environmental characteristics of forests drive spatial heterogeneity in the effects that native tree evolutionary relatedness has on overall forest plant invasions. This investigation was conducted in the forests of the eastern United States of America (USA). We pursue this objective in the context of three separate characteristics of evolutionary relatedness: how divergent (sensuTucker et al. 2017) species are relative to one another across a phylogenetic tree (i.e. phylogenetic species clustering [PSC] and phylogenetic species variability [PSV]; [Helmus et al. 2007]), the amount of evolutionary history contained within these phylogenetic relationships (i.e. phylogenetic diversity [PD]; [Faith 1992]) and the evenness in abundance at which species occur across a phylogenetic tree (i.e. phylogenetic species evenness [PSE]; [Helmus et al. 2007]). Regarding environmental characteristics, we investigate environmental harshness, as defined by Marks et al. (2016) and relative tree density, both of which can affect the establishment and growth of invasive plants (Theoharides and Dukes 2007, Guo et al. 2015). We also investigate the role of environmental variability, as it can accelerate evolutionary radiation amongst native species via allopatric speciation (Qian and Ricklefs 2000) and can reflect niche availability for invasions (Davies et al. 2005).

We found evidence that environmental harshness and, to a lesser degree, variability in environmental harshness drive spatial heterogeneity in the contribution of phylogenetic divergence (PSC and PSV) of native trees to biotic resistance to overall plant invasions in eastern USA forests. While spatial heterogeneity in the contribution of native trees to biotic resistance to forest plant invasions was previously noted (Iannone et al. 2016), this investigation is the first to our knowledge that identifies potential drivers of this large-scale heterogeneity. This detected contribution of phylogenetic divergence to invasion resistance illustrates how variation in localised processes by which evolutionary relatedness is hypothesised to limit invasions, e.g. niche overlap/competitive exclusion (Procheş et al. 2008, Thuiller et al. 2010), can contribute to complex large-scale invasion patterns. In addition, the spatial heterogeneity revealed also illustrates the utility of applying a community-level framework when investigating large-scale invasion patterns (Shea and Chesson 2002). Finally, our findings support the value of considering biotic and environmental filters to invasion in unison in order to understand overall invasion patterns (Pearson et al. 2018).

In contrast to phylogenetic divergence, we found no evidence that the amount of evolutionary history (i.e. PD) or the evenness at which these native tree species occur across a given community’s phylogenetic tree (i.e. PSE) contribute to biotic resistance across macroscales. The consistently positive associations that PD and PSE shared with both invasive richness and cover suggests these metrics to be more indicative of niche availability than biotic resistance, at least at the spatial scale of our investigation. The weak magnitudes of these associations also revealed them to be of little value in predicting macroscale invasion patterns. The positive association between PD and invasion was not surprising, as this metric is strongly related to native species richness, which itself is positively related to invasive richness and cover at large spatial scales (Stohlgren et al. 2006, Fridley et al. 2007, Iannone et al. 2016). Causes of the positive association between phylogenetic species evenness (PSE) and invasion are less clear. Even though species evenness can contribute to biotic resistance at small spatial scales in grasslands (Wilsey and Polley 2002), this tells us little about how species evenness affects large-scale invasions in forests. Thus, studies to determine how variability in spatial scale and ecosystem type affects the contribution of species evenness to biotic resistance are needed.

By revealing factors that affect the strength of relationships between phylogenetic divergence and invasion, our study revealed conditions under which phylogenetic divergence of native tree communities likely contributes most to invasion resistance in forest ecosystems. Standardised slope estimates revealed that mean maximum tree height was the explanatory variable having the greatest effect. This factor was negatively correlated with slope values for relationships between both PSC and PSV and both invasion richness and cover. Given that maximum tree height is an inverse measure of environmental harshness (Marks et al. 2016), these negative relationships reveal a greater potential for more phylogenetically divergent tree communities to limit invader establishment and dominance under less harsh environmental conditions. This effect may reflect the long-recognised decline in the ecological importance of competition in more harsh environments (Bertness and Callaway 1994). That is, an invader’s ability to deal with stress may be more important than its ability to deal with competition when invading harsher environments, regardless of the niche space occupied by the native community. Determining the component(s) of environmental harshness that most affect relationships between phylogenetic relatedness and invasion will increase our understanding of the causes of the patterns revealed here. Investigating factors both related to tree height and that directly affect plant growth, e.g. soil productivity, drought, cold etc. (Marks et al. 2016), may be fruitful.

Increased environmental variability across large geographic areas can increase species richness (Qian and Ricklefs 2000), which, in some situations, is positively related to certain metrics of evolutionary relatedness (Potter and Woodall 2014). Increased environmental variability is also positively associated with invasions (Davies et al. 2005). For these reasons, we expected factors indicative of environmental variability to affect the magnitude of relationships between metrics of phylogenetic divergence and invasion. We found such evidence for the SD of maximum tree height, but not for SD of relative tree density or mean Jaccard’s distances. The positive correlation of SD of maximum tree height to slope values for relationships between invasive richness and both PSC and PSV reveals phylogenetically divergent tree communities to limit invader establishment more in ecological sections that are less variable regarding environmental harshness. However, smaller slope and larger p-value estimates for SD vs. mean maximum tree height reveal overall environmental harshness to be more influential.

The decline in the effects of phylogenetic divergence on invasion in ecological sections having more variable maximum tree height may reflect an increased number of locations having more harsh environments. That is, it indicates an increase in the number of locations where phylogenetic divergence affects invader establishment less. An increase in the number of locations experiencing canopy disturbance (natural or human) may also contribute to our finding regarding SD in maximum tree height given that increased light facilitates forest plant invasions (Eschtruth and Battles 2009, Iannone et al. 2013). However, given that SD in maximum tree height and mean stand age are only weakly related, it is likely that increased SD in maximum tree height reflects more than increased rates of canopy disturbance. Declines in the effects of phylogenetic divergence on invasion in ecological sections having more variable maximum tree height may also reflect the previously noted positive associations between invasion and environmental variability (Davies et al. 2005). Considering components of forest plant communities other than, or in addition to, native trees, e.g. native understorey plants, soils etc., is likely to improve understanding of how environmental variability affects large-scale forest plant invasions.

We utilised two different measures of invasion — invasive species richness and cover — as both provide different perspectives on invasion patterns. Invasive richness is an indicator of invader establishment and invasive cover is an indicator of invader dominance. Prior theoretical and empirical investigations reveal the deeper understanding that can be gained by considering multiple invasion metrics simultaneously within the same investigation (Kennedy et al. 2002, Stohlgren et al. 2003, Guo and Symstad 2008, Catford et al. 2012, Iannone et al. 2015). For instance, we found evidence that the effects of phylogenetic divergence on both invader establishment and dominance (i.e. richness and cover, respectively) declines with increased environmental harshness. In contrast, variability in environmental harshness only influenced the ability of phylogenetic divergence to affect invasive plant establishment and not invasive plant dominance. Based on our findings, we would predict that, even if phylogenetic divergence does not prevent invader establishment in a given area due to that area exhibiting greater spatial variability in environmental harshness, it might still limit invader dominance so long as the overall/mean environmental harshness of the area is low. This insight is of practical importance given that invasive cover may reflect invader impacts more than invasive species richness, potentially making it of greater relevance to invasive plant management (Guo and Symstad 2008, Hillebrand et al. 2008).

Identifying the factors that affect the ability of PSC and PSV to contribute to invasion resistance may also help to reveal how characteristics of evolutionary relationships (e.g. divergence, branch lengths and species evenness) emerge under different environmental conditions. For instance, both PSC and PSV limit invader establishment (as estimated by invasive richness) to greater degrees in forests that are less harsh. Therefore, the levels of environmental harshness found in a forest, by indirectly affecting invader establishment, have the potential to affect future PSC and PSV values for that forest. Therefore, our findings illustrate how knowing the phylogenetic relationships within an ecological community can help to understand the conditions from which these relationships emerge, i.e. the “phylogenetic-patterns-as-results utility” of known phylogenetic relationships (Gerhold et al. 2015).

We found evidence that environmental harshness and, to a lesser degree, spatial variability in environmental harshness, affect the ability of more phylogenetically divergent native tree communities to limit the establishment and dominance of invasive plants. Therefore, through indirect pathways, these factors may affect short-term invasion patterns and community-level change and, in so doing, affect the long-term characteristics of evolutionary relationships. Nevertheless, these factors did not explain all variability in phylogenetic-invasion relationships. Therefore, future investigations are needed. Considering how other known drivers of invasion patterns, such as propagule pressure and plant functional traits, as well as understorey native plant communities and forest soils, contribute to variability in the degree to which phylogenetic divergence of native trees contributes to invasion resistance, will likely be beneficial. Also needed is the determination of the component(s) of environmental harshness most contributing to our findings and the mechanisms by which phylogenetic divergence amongst native trees affects overall plant invasion. Controlled experiments replicated across our study region would greatly help in this regard. Such considerations will increase our understanding of how the evolutionary relatedness amongst species in a given community affects invasions and community change.

Thanks to the many Forest Inventory and Analysis workers who collected the data used in this study and to Beth Schulz, Andrew Gray and Chris Witt for helping to compile data. This study was supported by National Science Foundation Macrosystems Biology grants #1241932 and #1638702 and by Cost Share Agreement 14-CS-11330110-042 between the United States Department of Agriculture Forest Service and North Carolina State University.