Skip to main content
PMC home page
UKPMC Funders Author Manuscripts logoLink to UKPMC Funders Author Manuscripts
. Author manuscript; available in PMC: 2016 Aug 3.
Published in final edited form as: Ecol Lett. 2015 Dec 22;19(3):219–229. doi: 10.1111/ele.12556

What it takes to invade grassland ecosystems: traits, introduction history and filtering processes

Marta Carboni 1,2, Tamara Münkemüller 1,2, Sébastien Lavergne 1,2, Philippe Choler 1,2, Benjamin Borgy 3,4, Cyrille Violle 3, Franz Essl 5, Cristina Roquet 1,2, François Munoz 6,7, DivGrass Consortium, Wilfried Thuiller 1,2
PMCID: PMC4972145  EMSID: EMS69435  PMID: 26689431

Abstract

Whether the success of alien species can be explained by their functional or phylogenetic characteristics remains unresolved because of data limitations, scale issues and weak quantifications of success. Using permanent grasslands across France (50,000 vegetation-plots, 2000 species, 130 aliens) and building on the Rabinowitz’ classification to quantify spread, we showed that phylogenetic and functional similarities to natives were the most important correlates of invasion success compared to intrinsic functional characteristics and introduction history. Results contrasted between spatial scales and components of invasion success. Widespread and common aliens were similar to co-occurring natives at coarse scales (indicating environmental filtering), but dissimilar at finer scales (indicating local competition). In contrast, regionally widespread but locally rare aliens showed patterns of competitive exclusion already at coarse scale. Quantifying trait differences between aliens and natives and distinguishing the components of invasion success improved our ability to understand and potentially predict alien spread at multiple scales.

Keywords: Alien species, Functional and phylogenetic similarity, Invasibility, Invasion success, Invasiveness, Rabinowitz

Introduction

The repeated introduction of alien species is a major component of ongoing global changes and a major threat to global biodiversity (Walther et al. 2009; Vila 2013). Understanding and predicting which introduced species will spread, naturalize and finally become invasive requires two major steps: 1) characterizing invasion success and 2) identifying its driving forces. Traditionally, invasion success was studied by focusing either on intrinsic biological characteristics conferring invasiveness, or on characteristics of the resident communities favoring invasion (invasibility; Richardson & Pyšek 2006). Additionally, historical factors, such as time since introduction or introduction pathways, also play a role (Wilson et al. 2007; Mahoney et al. 2015). Nevertheless, attempts to link the degree of success across species to intrinsic traits, community invasibility and introduction history in a combined way have been rare and have identified few generalities (e.g. Thuiller et al. 2006). One reason is that our ability to test the importance of complementary factors on invasion processes hinges on the development of integrative frameworks to better quantify and measure the invasion success of alien species (Colautti et al. 2014).

Studying the success of alien species requires detailed quantification of both their local performance and geographic spread. Most studies attempting to identify alien species of greatest concern have focused on frequencies of species occurrence. However, focusing only on occurrence data tends to over-estimate the success of species that occur in low abundances but are widespread, while under-estimating the success of species that dominate where present but are not widely distributed (Colautti et al. 2014). Here we propose to account for different components of invasion success by applying for the first time the Rabinowitz’ classification of rarity and abundance (Rabinowitz 1981, Fig. 1) to alien species. Combining frequency of occurrence with geographic range, local abundance and niche breadth gives a more comprehensive measure of invasion success in a given area. In this framework, aliens can be considered most successful when they are regionally widespread, locally abundant and have broad ecological preferences.

Figure 1.

Figure 1

Open in a new tab

Conceptual visualization of the Rabinowitz framework. Shades of green indicate different environments, blue dots indicate a species’ occurrence, and the size of the dots indicates its local abundance. Depending on their geographical distribution (narrow vs. widespread), local abundance (abundant vs. sparse) and niche breadth (specialist vs. generalist) species can be divided into eight distinct classes (A through H in the figure). Only species in class A, geographically widespread, locally abundant and generalists, can be considered common. We follow the same rationale to argue that alien species that fulfill all of these characteristics have high invasion success. At the opposite extreme, aliens in class H are unsuccessful. In between, alien species common for only one criterion have low success, while species common for at least two criteria have intermediate success. From the three criteria a continuous measure of invasion success can also be derived based on a dimension reduction analysis.

Pinpointing intrinsic functional traits associated with invasiveness to profile successful invaders has been a long lasting quest in invasion ecology. For plants, a number of traits have been consistently found to facilitate invasion (Richardson & Pyšek 2006; Van Kleunen et al. 2010). Small seed sizes, prolific seed production and short juvenile periods regularly emerge as important determinants of invasiveness, as they correlate with long dispersal distances and high competitive potential at the juvenile stage (Richardson & Pyšek 2006). Fast relative growth rates are thought to favor highly invasive species in frequently disturbed or early-successional ecosystems (Rejmanek & Richardson 1996; Dawson et al. 2011b; Ordonez & Olff 2013). Finally, studies highlighted that invasive plant species tend to be larger than non-invasive ones, probably in relation with the ability to capture light (Van Kleunen et al. 2010). Overall, the finding that successful alien plants share sets of functional characters is also supported by studies showing that certain phylogenetic lineages have higher proportions of invasive aliens than expected by chance (Lambdon 2008). Hence, it seems possible to screen successful alien plant species based on trait syndromes, such as the leaf-height-seed (LHS) spectrum (Westoby 1998), or based on their phylogenetic origins, which will capture biogeographic history and potential broad bioclimatic adaptations.

While traits conferring invasiveness can favor invasion, they do not guaranty its success. The abiotic and biotic characteristics of the recipient communities also determine invasion opportunities (invasibility, Thuiller et al. 2010). For instance, if environmental filtering is the main process determining alien species spread, the most successful aliens should share similar ecological adaptations to natives (Diez et al. 2008). If instead certain trait values confer greater competitive ability, the most successful aliens should be the ones that rank higher in the trait hierarchy (Kunstler et al. 2012; Gallien et al. 2014b). Conversely, if we assume more intense competition for resources between ecologically-similar species, local communities are likely to be resistant to aliens that have very similar traits or are closely related to the resident natives (Darwin 1859; Elton 1958). Based on these assumptions, specific metrics of functional and phylogenetic similarity have been proposed in an attempt to infer the processes driving invasion (Thuiller et al. 2010; Gallien et al. 2014a). The application of these metrics has highlighted that invasion dynamics can broadly vary depending on local ecosystem properties, spatial scale, and the amount of available resources (Diez et al. 2008, Carboni et al. 2013, Gallien et al. 2014b).

However, to date very few studies have explicitly investigated how multiple, interacting drivers of invasion success vary among alien species (Pellock et al. 2013) and fewer still have considered different components of invasion success at the same time (Lemoine et al. 2015). If specific processes are important for regionally widespread, locally abundant and generalist alien species compared to less successful ones, identifying processes linked to invasion success seems of utmost importance. Ultimately, the identification of these crucial processes can help anticipating the most concerning invasions.

Here we tackled the challenge of relating invasion success of alien species to intrinsic traits conferring invasiveness and to processes determining invasibility of native communities, while accounting for species' introduction history. First, we developed a framework for quantifying invasion success based on each alien species’ invaded range area, local abundance and realized niche breadth in native communities. Second, we asked whether functional- and phylogenetic-based measures of community invasibility are good predictors of the degree of invasion success of aliens compared to their intrinsic traits and introduction history. While our initial aim was to pinpoint the processes determining invasion success, we also tested our framework in a predictive context. We did so in the context of temperate, C3 dominated permanent grassland communities, based on an extensive dataset spanning the whole of France. Investigating invasion success is of high importance in these grasslands that are already severely threatened by urban development and the abandonment of traditional agricultural practices (Wesche et al. 2012; Essl & Dirnböck 2008).

Methods

Study system, habitats and vegetation plots

Our study focuses on permanent grasslands (hereafter “grasslands”) defined as semi-natural herbaceous ecosystems supporting livestock farming systems. They are usually characterized by low external inputs (e.g. feed, fertilizer, energy). In France the area of grasslands is declining since the 1970s, but still covers ca. 10 million hectares (Violle et al. 2015).

We used a comprehensive database compiled as part of the DIVGRASS project (Violle et al. 2015) and reporting the composition of more than 50,000 vegetation plots throughout France, geo-referenced at a precision of around 5 km on average, for a total of 4280 species. The plant communities sampled consist of semi-natural grasslands (mesic, ruderal and trampled grasslands) created and maintained by mowing and grazing, and natural grasslands (mountain and dry calcareous grasslands) covering wide ecological gradients. The data consist of estimated relative cover of all species present within an area ranging from 25 to 100 m2 (Appendix S1).

Each plot was assigned to one of four broadly defined grassland habitats based on an analysis of modularity of the overall species-site network (Carstensen et al. 2013) and using the algorithm of Blondel et al. (2008). The four habitats were: (i) dry calcareous grasslands, (ii) mountain grasslands, (iii) mesic grasslands and (iv) ruderal and trampled grasslands.

Alien species and introduction history

We focused on 160 herbaceous alien species defined as naturalized in France (Table S2) within the DAISIE database (www.europe-aliens.org), which provides an up-to-date inventory of species alien to European countries (Pyšek et al. 2009). Both neophytes (new alien species introduced > c. 1500) and few archeophytes (old alien species, introduced < c. 1500) were listed in this database.

Species’ introduction history (year of first record and main introduction pathway) in France was taken from DAISIE. If no record was available for France, we used the information from bordering countries as an approximation. We removed species with no information on introduction history leading to a final set of 127 alien naturalized species. To quantify minimum residence time, we determined the year of first record as the earliest year of all the records available for each alien species. Introduction pathways were based on the DAISIE hierarchical classification system (Table S1). We summarized this information by identifying the most common introduction pathway among “Agriculture”, “Ornamental and horticulture”, “Unintentional through transport” and “Other”.

Phylogeny and traits

We reconstructed a molecular and dated genus-level phylogeny for the entire species pool based on DNA sequences available in GenBank (Roquet et al. 2013; Appendix S2). To calculate distance-based phylogenetic metrics (see below), we extracted the cophenetic distance between all species-pairs from the phylogenetic tree.

We collated information on four functional traits: Specific Leaf Area (SLA; the ratio of leaf area to dry mass), plant maximum height at maturity (Height), seed mass (SM) and growth form. These trait data were extracted from a number of databases and local datasets (see Appendix S3 and Violle et al. 2015 for details). The leaf-height-seed (LHS) scheme captures variation in resource acquisition, establishment and competitive ability among species (Westoby 1998). SLA reflects species relative growth rate (faster growing species having higher SLA values), nutrient economy and water use efficiency. Seed mass reflects dispersal capability, seed production and longevity on the one hand, and species survival at the seedling stage on the other. Height captures species’ ability to dominate the vegetation layer in a given habitat and to intercept light. Finally, we characterized growth form as either graminoid, forb or dwarf shrub. We kept only species for which we had information on at least two traits (2930 species from the initial 4280) and estimated the remaining missing values based on multivariate imputation by chained equations (van Buuren & Groothuis-Oudshoorn 2011; Appendix S3). To calculate distance-based functional metrics (see below), we compiled a pair-wise species dissimilarity matrix for all traits combined using the Gower distance (function “daisy” in the “cluster” package), as a multivariate measurement of functional dissimilarity.

Determining invasion success

We characterized invasion success based on the three components of the Rabinowitz scheme (Fig. 1). First, to quantify regional distribution we calculated the species’ frequency of occurrence, a measure that correlated strongly also with geographical spread (Fig. S1). Second, local abundance was taken from the average percentage cover of each alien species within the plots where it occurred. Third, niche breadth was calculated using a hypervolume approach based on three bioclimatic variables and one productivity measure hypothesized to be important for the naturalization and spread of alien species. Climatic variables were derived from the French meteorological model Aurelhy (Bènichou & Le Breton 1987), based on interpolated measurements downscaled at 100 m resolution and summarizing climatic information over the last 30 years (1971–2000). We focused on average annual precipitation, average daily minimum temperature in winter (December-February) and average daily maximum temperature in summer (June-August). We used the Normalized Difference Vegetation Index (NDVI), summed over the whole year as a proxy of gross primary productivity. Data were obtained from the Moderate Resolution Imaging Spectro-radiometer (MODIS) for the years 2000-2014 and processed as described in Choler (2015). Based on these variables, we used multivariate kernel density estimation to construct a 4-dimensional ecological hypervolume for each alien species (Blonder et al. 2014). All variables were rescaled between 0 and 1 (by subtracting the minimum and dividing by the range) for comparability. Hypervolumes were constructed using a Silverman bandwidth estimator and a 0% quantile threshold (Blonder et al. 2014). We then used the calculated volumes in ecological space as proxies of niche breadth for each alien species.

Based on these three components of invasion success (regional distribution, local abundance and niche breadth), we then used two complementary approaches to characterize alien success: 1) we separated them into eight classes (corresponding to the classical Rabinowitz classes –Fig. 1), and 2) we synthesized the three components into a synthetic continuous index of success. The first approach categorizes aliens into invasion classes that might require specific management strategies, while the second ranks species in terms of their overall invasion success with the aim of identifying the features that can best explain it.

Figure 2.

Figure 2

Open in a new tab

Panel on the left: Numbers of alien species in the eight Rabinowitz classes. Red: common based on three criteria; Dark orange: common for two criteria; Light Orange: common for one criterion; Yellow: rare. Panel on the right: Concordance between the continuous invasion success measure (based on the first axis of a PCA) and the Rabinowitz classes for the 127 alien species. The least widespread class is on the left of the graph (class H - small regional distribution, locally not abundant, small niche breadth) and the most successful aliens are on the right (class A - regionally widespread, locally abundant, large niche breadth).

For the first (categorical) approach, we first split species into common vs. rare groups for each Rabinowitz' criterion independently (number of plots, local abundance and niche breadth), based on the sum of squares from species to group centers. Then the species were assigned to one of the eight Rabinowitz classes based on the cross-tabulation of the groups obtained for the three criteria (Figs. 1 and 2).

For the second (continuous) approach, we calculated a synthetic index representing the species scores on the first axis of the principal component analysis (PCA) performed on the three components of invasion success. This PCA axis explained roughly 60% of the variability in the data (Figure S1). It was most strongly linked to regional distribution, but it also captured niche breadth and local abundance. This "invasion success” indicator (or PCA1) was then used as a response variable in subsequent analyses to identify the most important drivers of invasion success.

Phylogenetic and functional dissimilarity of alien species

We calculated a set of functional and phylogenetic similarity metrics to native species (invasibility metrics, Thuiller et al. 2010) based on pairwise co-occurrences at two scales (Appendix S4). We first identified the native species that most often co-occurred with each alien species by using the V-score as a measure of species co-occurrence (Lepš & Šmilauer 2003, modified to account for species abundances). V-scores were calculated at both the plot (local community) and the habitat scale (defined by the set of plots belonging to the same habitat). Then, for each alien species we calculated 1) the (absolute) Mean Distance to the Most often Co-occurring Species (MDMCS), and 2) whether it had higher or lower values than the natives for each trait (i.e. its hierarchical position on each trait gradient, Kunstler et al. 2012). MDMCS was based on both phylogenetic and functional distances, whereas the hierarchical index was calculated for each trait independently (Height.hier, SLA.hier and SM.hier). Overall these metrics measure the relative invasibility of native grassland assemblages at multiple scales (Table 1, Appendix S4). In the framework of the hierarchical filtering model of invasion biology (Richardson & Pysek 2006), the habitat-level analyses provide insights into environmental filtering, whilst plot-level analyses allow recovering signals of biotic filtering.

Table 1.

Summary of distance-based invasibility metrics used in this study, identifying the type of distance (Functional vs. phylogenetic), the spatial scale of reference and the underlying assumptions. The hierarchical indices are reported only for the plot scale for brevity.

Index Distance Scale Description Assumptions
MDMCS.Hab Functional Habitat Mean functional distance to the most often co-occurring species within habitats The overall similarity of species niches is well captured by a subset of measured traits;
Environmental filtering is most captured by similarity to the native species with which the invader comes most often into contact at the habitat level
MDMCS.Hab Phylogenetic Habitat Mean phylogenetic distance to the most often co-occurring species within habitats The overall similarity of species niches is best captured by their phylogenetic relatedness;
Environmental filtering is most captured by similarity to the native species with which the invader comes most often into contact at the habitat level
MDMCS.Plot Functional Plot Mean functional distance to the most often co-occurring species within plots The overall similarity of species niches is well captured by a subset of measured traits;
Biotic resistance is driven by the native species with which the invader comes most often into contact within plots
MDMCS.Plot Phylogenetic Plot Mean phylogenetic distance to the most often co-occurring species within plots The overall similarity of species niches is best captured by their phylogenetic relatedness;
Biotic resistance is driven by the native species with which the invader comes most often into contact within plots
Height.hier Height Plot Mean height difference to the most often co-occurring species within plots Competitive ability can be represented on a trait hierarchy of height (eg. light interception);
Biotic resistance is driven by the native species with which the invader comes most often into contact within plots
SM.hier Seed Mass Plot Mean seed mass difference to the most often co-occurring species within plots Competitive ability can be represented on a trait hierarchy of seed mass (eg. colonization);
Biotic resistance is driven by the native species with which the invader comes most often into contact within plots
SLA.hier SLA Plot Mean SLA difference to the most often co-occurring species within plots Competitive ability can be represented on a trait hierarchy of seed mass (eg. growth rate);
Biotic resistance is driven by the native species with which the invader comes most often into contact within plots
Open in a new tab

Disentangling the factors explaining invasion success in grasslands

Modeling species independently of each other may bias analyses if closely related species have similar success. We tested for a phylogenetic signal of our continuous indicator of invasion success (i.e. PCA1) on the phylogenetic tree of alien species using Blomberg’s K statistic (Blomberg & Garland 2002) and found no significant signal (Appendix S5; Fig. 3; K = 0.074, PIC variance p-value = 0.931). Hence, we considered alien species as phylogenetically independent in further analyses of invasion success.

Figure 3.

Figure 3

Open in a new tab

Phylogenetic tree of alien species in 50,000 French grassland plots with the corresponding commonness class. Color scheme follows the one in the contingency table in Fig. 1.

We used boosted regression trees (BRT) to model invasion success of alien species as a function of introduction history (pathway and residence time), functional traits related to invasiveness (SLA, height, seed mass and growth form) and community invasibility metrics (functional and phylogenetic MDMCS at habitat and plot scales and trait-hierarchy metrics). BRTs allow handling different types of predictor variables, can fit nonlinear relationships, are robust to outliers, automatically handle interaction effects between predictors and are able to extract robust estimations of variable importance (Elith et al. 2008). We estimated variable importance for explaining invasion success based on the reduction of squared error attributable to each variable, normalized to sum up to 100 (relative influence). We estimated the goodness-of-fit of the overall model by assessing the R-square of fitted values against observed values. Models were fitted in R (R-Development-Core-Team 2014), using the ‘dismo’ package (Elith et al. 2008).

Finally, to test whether our approach for understanding invasion success was also able to predict which species from a given pool might become successful, we conducted a set of cross-validations through 1) repeated split-sampling and 2) jackknifing, using several goodness-of-prediction statistics to assess performance (Appendix S6 for details).

Results

Invasion success of alien plant species in French grasslands

Based on our three components of invasion success, we classified alien species into the eight abundance classes of the Rabinowitz framework (Fig. 2, Table S2). In particular, alien species were considered to be rare when: (1) they occurred in less than 13 plots, (2) covered in average less than 2.5% of total cover in the plots where they occurred, and (3) occupied an ecological hypervolume of less than 0.008 sd4.

Most alien species in French grasslands were rare, with almost 1/3 of species being in the rarest category (narrow geographic range, sparse abundance and small niche breadth). Forty-four aliens were classified as abundant for at least two criteria (e.g. Ambrosia artemisiifolia) and 20 alien species fell into the most successful alien species category (class A, see Table S2 for the list of species).

The synthetic index of invasion success gave very consistent results with the categorical classification of invasion success (Fig. 2, right panel). The rarest alien species had the lowest values of the invasion success indicator, species that were common for only one criterion had low to medium values, while species common for at least two criteria had medium to high success indicator values. This result supports the dimension reduction approach through PCA as a mean of synthesizing the Rabinowitz multi-dimensional scheme into a one-dimensional continuous success indicator.

Disentangling the factors explaining invasion success

Community invasibility metrics were collectively more important for explaining the invasion success of alien species, than either intrinsic trait values or introduction history (Fig. 4). In particular, the dissimilarity to species co-occurring in the same habitat type was by far the most important predictor of success both from a functional and phylogenetic perspective. Specifically, the most successful alien species were phylogenetically close and functionally similar to the native species co-occurring in the same habitat (Fig. 5A-B), but dissimilar to those co-occurring in the same plots (Fig. 5E). Minimum residence time, a classical feature for explaining invasion success, was also important, but only in third place. Further, having higher or lower seed mass compared to the natives co-occurring within plots (SM.hier.plot) was also determinant. Thereby, the most successful alien species were introduced long ago and had smaller seeds, greater SLA and lower height than the natives co-occurring within plots (Fig. 5C,D,F; Fig. S2). Importantly, while the relative trait hierarchy of alien and native species contributed to invasion success, intrinsic trait values of aliens species did not. Overall, this set of historical, functional and invasibility variables explained well the invasion success of alien species in French grasslands (R2: 0.571).

Figure 4.

Figure 4

Open in a new tab

Variable importance in the Boosted Regression Trees- analysis. Bar colors indicate symmetric invasibility metrics (black), hierarchical competition metrics (dark grey), species traits (light grey) and introduction history (white). MDMCS.Habitat.phylo = Mean phylogenetic distance to the most often co-occurring species at habitat level; MDMCS.Habitat.func = Mean functional distance to the most often co-occurring species at habitat level; MDMCS.plot.func = Mean functional distance to the most often co-occurring species at plot level; MDMCS.plot.phylo = Mean phylogenetic distance to the most often co-occurring species at plot level; SM.hier.plot = mean (signed) difference in seed mass at plot level; Height.hier.plot = mean (signed) difference in height at plot level; SLA.hier.plot = mean (signed) difference in SLA at plot level; SM.hier.Hab = mean (signed) difference in seed mass at habitat level; Height.hier.Hab = mean (signed) difference in height at habitat level; SLA.hier.Hab = mean (signed) difference in SLA at habitat level. Also see Table 1.

Figure 5.

Figure 5

Open in a new tab

Response curves of invasion success to the five most important variables in the BRT. All variables are rescaled between zero and one (apart for the hierarchical indices to highlight positive vs. negative trait differences). Invasion success is expressed by the first axis of a PCA. Negative values relate to low invasion success while positive values relate to high invasion success. Legend: MDMCS.hab.phylo = Mean phylogenetic distance to the most often co-occurring species within habitats MDMCS.hab.func = Mean functional distance to the most often co-occurring species within habitats; MDMCS.plot.func: Mean functional distance to the most often co-occurring species within plots; SM.hier.plot = mean (signed) difference in seed mass at plot level; Height.hier.plot = mean (signed) difference in height at plot level.

Moreover, contrasting phylogenetic similarity patterns at the habitat scale suggests distinct underlying drivers for the different components of success (Fig. 5A). When considering variation across the Rabinowitz’ classes (Fig. S3), we found that different strategies could favor different components of invasion success. In congruence with the BRT results, alien species closely related to natives became both widespread and abundant (classes A and E; Fig. S3, left panel). By contrast, alien species that were phylogenetically most distant from natives became widespread regionally but not locally abundant (class B). In other words, the species in class A were those indicating high invasion success of close relatives to the natives in the BRT response curve (left part of Fig. 5A), while species in class B corresponded to the moderate invasion success obtained by distantly related species (right part of Fig. 5A). Finally the species with very low invasion success (rare aliens) had intermediate phylogenetic distances to the natives (central part of Fig. 5A).

Predicting invasion success

While our framework is mainly designed for for identifying the factors driving the success of invasions in grassland ecosystems, we found that our model also had a reasonable predictive performance to estimate invasion success of newly introduced species (Pearson R = 0.56 and R-squared = 0.33; Tables S3, Appendix S6). In particular, frequent, abundant and generalist species were on average correctly predicted as having greater invasion success than the rare ones (Fig. S11). The highest prediction errors and under-predictions occurred for the invaders of class C (regionally scarce, but locally abundant and generalist) and for some of the species of class A (Fig. S12).

Discussion

By explicitly considering different components of invasion success we were able to unravel the drivers of plant invasions in French permanent grasslands. We found support for many classical hypotheses relating invasion success to species' traits and introduction history. Moreover, we were able to identify signals of competition at large spatial extent and for highly diverse communities. But more importantly, we showed that functional and phylogenetic similarities to natives were by far the most important features, suggesting that they are excellent proxies for processes of environmental filtering and competition driving invasion success. Specifically, environmental conditions filtered aliens similar to the natives at the habitat scale, while a combination of hierarchical competition and limiting similarity led to greater success of shorter aliens with smaller seeds and faster growth rates than the natives at fine scale.

We confirmed that most alien species introduced in French grasslands remained rare and failed to establish large and successful populations (Richardson & Pyšek 2006). Interestingly, certain aliens considered a threat in other habitats or countries, (e.g. Cortadeira selloana, one of the 100 worst invasive species in Europe; Pyšek et al. 2009), appear to still be rare in French grasslands demonstrating the importance of the native community and of the historical context. Still, we identified twenty alien species as highly successful (Fig. 2). Several of them, including Erigeron annuus, Oenothera biennis and Paspalum dilatatum, are already known to be invasive elsewhere, causing negative environmental impacts in a number of different regions and habitats (Weber & Gut 2004; Pyšek et al. 2009; Kumschick et al. 2015). A few species, such as Avena sativa and Linum usitatissimum, were instead species that are, or have formerly been, cultivated, suggesting that they profited from high propagule pressure, but are unlikely to cause negative impacts.

In accordance with earlier work we found that invasion success was highest for species with a long residence time, which was the third most important factor predicting alien species success (Hamilton et al. 2005; Wilson et al. 2007; La Sorte & Pysek 2009; Dawson et al. 2011a). For example, Veronica persica, an archeophyte in Europe, turned out to be among the most successful alien species. Residence time correlates with a number of factors such as propagule pressure or the overcoming of an initial lag-phase in the process of invasion (Richardson & Pyšek 2006; Wilson et al. 2007), particularly by allowing time for acclimatization and local adaptation (Lavergne & Molofsky 2007). Conversely, we found no strong relationship between invasion success and introduction pathways. This is a surprising result given that most of the naturalized alien plants considered in this study (ca. 50%) have been introduced as ornamental or horticultural plants. These are often suggested to spread faster than other alien species (Wilson et al. 2007) due to selective breeding and hybridization in the cultivation process, high propagule pressure and repeated introduction of new genetic material through the ornamental trade (Lavergne & Molofsky 2007). However, our finding may result from repeated introductions diluting adaptations to new environments.

The most interesting results from our study are those linking invasion success to similarity patterns between plant invaders and native communities. Many studies have already demonstrated the existence of non-random similarity patterns in invasions (Diez et al. 2008; Carboni et al. 2013; Pellock et al. 2013; Gallien et al. 2014b; Kraft & Ackerly 2010). However, here we showed that even when accounting for traits and introduction history, phylogenetic and functional similarity metrics were by far the most important correlates of plant invasion success in French grasslands.

In accordance with the filtering model of invasion biology (Richardson & Pysek 2006), whether successful aliens co-occurred with similar or dissimilar natives depended on the spatial scale (habitat vs. plot). Specifically, the most successful alien species (both widespread and locally abundant) were those sharing habitat preferences with close relatives and functionally similar species in the native species pool (Cadotte et al. 2009; Lemoine et al. 2015). This finding suggests that alien species share ecological adaptations with natives when they successfully spread within a particular habitat type. It also implies that widespread, locally abundant and generalist aliens are not competitively limited at the habitat scale. The reason may be either that these very successful species are better competitors than natives, or that they avoid co-occurrence with closely related native competitors at fine scales. The fact that the most successful aliens tended to co-occur with functionally distinct natives at the plot scale (having high MDMCS.plot values, smaller seeds and lower height, Fig. 5) supports the latter hypothesis. These results corroborate previous findings showing that filtering mechanisms differ with the scale of observation (Thuiller et al. 2010, Carboni et al. 2013, Kraft & Ackerly 2010). Here we find evidence of multiple hierarchical filters at an unprecedented spatial extent, thanks to the use of three dimensions of invasion success, rather than just based on dichotomous contrasts of aliens vs. natives or invasive vs. naturalized species.

Interestingly, we did not find strong support for intrinsic functional traits defining an overall successful invader (Lemoine et al. 2015). By contrast, all hierarchical indices emphasizing trait differences to the natives at the plot level were much more important than the trait values per se. In particular, successful aliens were shorter, had smaller seeds and greater SLA than natives co-occurring within plots. Small seed mass generally correlates with prolific seed output, wider dispersal (often through wind), long persistence in the seed bank, short juvenile periods and high relative growth rate of seedlings (Moles & Westoby 2006; Rejmanek & Richardson, 1996; Van Kleunen et al., 2010). These characteristics are likely to favor rapid colonization by alien species in sites where natives are not as fast-growing, particularly in response to grassland disturbance. Competition for light does not seem to be the main process filtering successful aliens in grasslands, as shorter species were more successful. Shorter species can spread laterally in disturbed habitats, thereby being superior competitors for space and/or soil nutrients, and are also likely to be favored in grazed habitats, due to both grazing avoidance and better ability to deal with higher soil bulk density associated with trampling. Therefore, our data suggest that in grasslands, faster growth rates than the surrounding natives, specifically at the juvenile stage (small seed size), together with lateral spread and good dispersal ability to reach disturbed patches are more important characteristics than high competitive ability for light interception (competition-colonization trade-off; Kneitel & Chase, 2004).

Based on the Rabinowitz framework, we also highlighted distinct ecological processes driving different components of invasion success. In contrast with the most successful alien species of class A (widespread, generalists and abundant), aliens of class B were widespread and generalists but remained locally sparse. Aliens of this latter group were very distantly related to the natives not only at the plot scale but also at the habitat scale. One possible explanation is that these species are poor competitors in most grasslands (e.g. they were mostly species with low height and SLA values). They may only colonize grassland sites when resources become available through resource pulses which weaken competitive exclusion (Davis et al. 2000) or due to continuous propagule pressure and dispersal from adjacent, more suitable habitats. Finally, the alien species in the rare categories are neither very similar nor dissimilar to co-occurring natives at the habitat scale, suggesting that rather neutral processes determine their presence in these grasslands. We argue that processes driving invasion are not the same across alien species and that specific processes are important for widespread species compared to less successful ones. Future studies should acknowledge that the relative importance of filtering processes affecting the success of alien species is likely to change along the classical introduction-naturalization-invasion continuum (sensu Richardson & Pyšek 2006).

Finally, through a cross-validation procedure we highlighted that our model based on similarity to the natives also had the potential to estimate invasion success of newly introduced species. Predictive performance was generally reasonable (Appendix S6), though we found high prediction errors and under-predictions particularly for the invaders of class C. These species are locally abundant and generalist, but they are at the moment restricted to a narrow extent, potentially close to introduction sources (i.e. they may represent recent introductions). The invasion success of these species, which are likely not to be at equilibrium, is the most difficult to predict. However, it is encouraging that, while invasion success of species in class A was not always correctly identified, the most well-known invaders in this class (e.g O. biennis, E. annuus, B. inermis, J. tenuis) were indeed very well predicted by the model (Appendix S6). Such a framework has thus the potential to be used in other types of systems and in other regions.

Conclusions

Using French grasslands as a study system, we showed that complementary measures of invasion success are useful to identify the driving forces of invasion across spatial scales and environmental gradients. The Rabinowitz’ components originally created for assessing different forms of species’ rarity represent a straightforward framework for categorizing alien species and, in turn, for identifying the most successful and potentially harmful ones. In doing so, this framework provides guidance for differentiating and prioritizing alien species for management. Further, our use of complementary similarity metrics based on co-occurrence patterns allowed the identification of the relevant processes of invasion at two spatial scales. The most successful alien species are specifically adapted to the habitat they invade the most, but they then express different traits than co-occurring native species at fine scale within these habitats. Together with residence time and dispersal ability, these characteristics make up the winning alien species in permanent grasslands. Moreover we showed that these characteristics in a modeling framework are likely to give hints on the invasion success that can be attained by newly introduced species. Taken together, the set of approaches and metrics we have proposed here can be used in other ecosystems and biogeographic regions, not only to anticipate invasions but also to gain a better understanding of the processes shaping invasion success worldwide.

Supplementary Material

Supplementary Materials

Acknowledgements

We thank J. Renaud for his help with the preparation of the environmental data, and R. Douzet and three anonymous reviewers for their insightful comments. The research leading to these results received funding from the European Research Council under the European Community's Seven Framework Programme FP7/2007-2013 Grant Agreement no. 281422 (TEEMBIO). MC was supported by a Marie Curie Intra European Fellowship within the European Community’s Seventh Framework Program (FP7-PEOPLE-2012-IEF; IASIMOV 327616). MC, WT, SL, TM and F.E. acknowledge support by the ERA-Net BiodivERsA, with the national funders ANR (ANR-13-EBID-0004), DFG and FWF, part of the 2013 BiodivERsA call (project “WhoIsNext”). The LECA is part of Labex OSUG@2020 (ANR10 LABX56). CV was supported by the European Research Council (ERC) Starting Grant Project “Ecophysiological and biophysical constraints on domestication in crop plants” (Grant ERC-StG-2014-639706-CONSTRAINTS). All data were extracted from the DIVGRASS database supported by the French Foundation for Research on Biodiversity (FRB; www.fondationbiodiversite.fr) in the context of the CESAB project "Assembling, analysing and sharing data on plant functional diversity to understand the effects of biodiversity on ecosystem functioning: a case study with French Permanent Grasslands" (DIVGRASS).

Footnotes

Statement of authorship: MC, WT, TM and SL conceived the ideas. PC, CV, FM, BB and FE contributed and processed data. CR built the phylogeny. MC performed the analyses. All authors contributed to the writing, which was led by MC.

References

  1. Bènichou P, Le Breton O. Prise en compte de la topographie pour la cartographie des champs pluviomÈtriques statistiques. La Mètèorologie. 1987;7:23–34. [Google Scholar]
  2. Blomberg SP, Garland T. Tempo and mode in evolution: phylogenetic inertia, adaptation and comparative methods. Journal of Evolutionary Biology. 2002;15:899–910. [Google Scholar]
  3. Blondel VD, Guillaume J-L, Lambiotte R, Lefebvre E. Fast unfolding of communities in large networks. Journal of Statistical Mechanics-Theory and Experiment. 2008 [Google Scholar]
  4. Blonder B, Lamanna C, Violle C, Enquist BJ. The n-dimensional hypervolume. Global Ecology and Biogeography. 2014;23:595–609. [Google Scholar]
  5. Cadotte MW, Hamilton MA, Murray BR. Phylogenetic relatedness and plant invader success across two spatial scales. Diversity and Distributions. 2009;15:481–488. [Google Scholar]
  6. Carboni M, Munkemuller T, Gallien L, Lavergne S, Acosta A, Thuiller W. Darwin's naturalization hypothesis: scale matters in coastal plant communities. Ecography. 2013;36:560–568. doi: 10.1111/j.1600-0587.2012.07479.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Carstensen DW, Dalsgaard B, Svenning J-C, Rahbek C, Fjeldsa J, Sutherland WJ, et al. The functional biogeography of species: biogeographical species roles of birds in Wallacea and the West Indies. Ecography. 2013;36:1097–1105. [Google Scholar]
  8. Choler P. Growth response of temperate mountain grasslands to inter-annual variations of snow cover duration. Biogeosciences. 2015;12:3885–3897. [Google Scholar]
  9. Colautti R, Colautti R, Parker JD, Cadotte MW, Pyšek P, Brown CS, et al. Quantifying the invasiveness of species. NeoBiota. 2014;21:7–27. [Google Scholar]
  10. Darwin C. The Origin of Species. John Murray; London: 1859. [Google Scholar]
  11. Davis MA, Grime JP, Thompson K. Fluctuating resources in plant communities: a general theory of invasibility. Journal of Ecology. 2000;88:528–534. [Google Scholar]
  12. Dawson W, Burslem D, Hulme PE. The comparative importance of species traits and introduction characteristics in tropical plant invasions. Diversity and Distributions. 2011a;17:1111–1121. [Google Scholar]
  13. Dawson W, Fischer M, van Kleunen M. The maximum relative growth rate of common UK plant species is positively associated with their global invasiveness. Global Ecology and Biogeography. 2011b;20:299–306. [Google Scholar]
  14. Diez JM, Sullivan JJ, Hulme PE, Edwards G, Duncan RP. Darwin's naturalization conundrum: dissecting taxonomic patterns of species invasions. Ecology Letters. 2008;11:674–681. doi: 10.1111/j.1461-0248.2008.01178.x. [DOI] [PubMed] [Google Scholar]
  15. Elith J, Leathwick JR, Hastie T. A working guide to boosted regression trees. Journal of Animal Ecology. 2008;77:802–813. doi: 10.1111/j.1365-2656.2008.01390.x. [DOI] [PubMed] [Google Scholar]
  16. Elton CS. The ecology of invasions by plants and animals. Methuen & Co. Ldt.; London: 1958. [Google Scholar]
  17. Essl F, Dirnböck T. Diversity of native and alien vascular plant species of dry grasslands in central Europe. Applied Vegetation Science. 2008;11:441–450. [Google Scholar]
  18. Gallien L, Carboni M, Muenkemueller T. Identifying the signal of environmental filtering and competition in invasion patterns - a contest of approaches from community ecology. Methods in Ecology and Evolution. 2014a;5:1002–1011. [Google Scholar]
  19. Gallien L, Mazel F, Lavergne S, Renaud J, Douzet R, Thuiller W. Contrasting the effects of environment, dispersal and biotic interactions to explain the distribution of invasive plants in alpine communities. Biological Invasions. 2014b:1–17. doi: 10.1007/s10530-014-0803-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hamilton MA, Murray BR, Cadotte MW, Hose GC, Baker AC, Harris CJ, et al. Life-history correlates of plant invasiveness at regional and continental scales. Ecology Letters. 2005;8:1066–1074. [Google Scholar]
  21. Kneitel JM, Chase JM. Trade-offs in community ecology: linking spatial scales and species coexistence. Ecology Letters. 2004;7:69–80. [Google Scholar]
  22. Kraft NJB, Ackerly DD. Functional trait and phylogenetic tests of community assembly across spatial scales in an Amazonian forest. Ecological Monographs. 2010;80:401–422. [Google Scholar]
  23. Kumschick S, Bacher S, Evans T, Marková Z, Pergl J, Pyšek P, et al. Comparing impacts of alien plants and animals in Europe using a standard scoring system. Journal of Applied Ecology. 2015;52:552–561. [Google Scholar]
  24. Kunstler G, Lavergne S, Courbaud B, Thuiller W, Vieilledent G, Zimmermann NE, et al. Competitive interactions between forest trees are driven by species’ trait hierarchy, not phylogenetic or functional similarity: implications for forest community assembly. Ecology Letters. 2012;15:831–840. doi: 10.1111/j.1461-0248.2012.01803.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. La Sorte FA, Pysek P. Extra-regional residence time as a correlate of plant invasiveness: European archaeophytes in North America. Ecology. 2009;90:2589–2597. doi: 10.1890/08-1528.1. [DOI] [PubMed] [Google Scholar]
  26. Lambdon PW. Is invasiveness a legacy of evolution? Phylogenetic patterns in the alien flora of Mediterranean islands. Journal of Ecology. 2008;96:46–57. [Google Scholar]
  27. Lavergne S, Molofsky J. Increased genetic variation and evolutionary potential drive the success of an invasive grass. Proceedings of the National Academy of Sciences of the United States of America. 2007;104:3883–3888. doi: 10.1073/pnas.0607324104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Lemoine NP, Shue J, Verrico B, Erickson D, Kress WJ, Parker JD. Phylogenetic relatedness and leaf functional traits, not introduced status, influence community assembly. Ecology. 2015;96:2605–2612. doi: 10.1890/14-1883.1. [DOI] [PubMed] [Google Scholar]
  29. Lepš J, Šmilauer P. Multivariate Analysis of Ecological Data using CANOCO. Cambridge University Press; Cambridge: 2003. [Google Scholar]
  30. Mahoney PJ, Beard KH, Durso AM, Tallian AG, Long AL, Kindermann RJ, et al. Introduction effort, climate matching and species traits as predictors of global establishment success in non-native reptiles. Diversity and Distributions. 2015;21:64–74. [Google Scholar]
  31. Moles AT, Westoby M. Seed size and plant strategy across the whole life cycle. Oikos. 2006;113:91–105. [Google Scholar]
  32. Ordonez A, Olff H. Do alien plant species profit more from high resource supply than natives? A trait-based analysis. Global Ecology and Biogeography. 2013;22:648–658. [Google Scholar]
  33. Pellock S, Thompson A, He KS, Mecklin CJ, Yang J. Validity of Darwin's naturalization hypothesis relates to the stages of invasion. Community Ecology. 2013;14:172–179. [Google Scholar]
  34. Pyšek P, Lambdon PW, Arianoutsou M, Kühn I, Pino J, Winter M. Handbook of alien species in Europe. Springer: 2009. Alien vascular plants of Europe; pp. 43–61. [Google Scholar]
  35. R-Development-Core-Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing; Vienna, Austria: 2014. [Google Scholar]
  36. Rabinowitz D. Seven forms of rarity. In: Synge H, editor. The Biological Aspects of Rare Plant Conservation. Riley; 1981. pp. 205–217. [Google Scholar]
  37. Rejmanek M, Richardson DM. What attributes make some plant species more invasive? Ecology. 1996;77:1655–1661. [Google Scholar]
  38. Richardson DM, Pyšek P. Plant invasions: merging the concepts of species invasiveness and community invasibility. Progress in Physical Geography. 2006;30:409–431. [Google Scholar]
  39. Roquet C, Thuiller W, Lavergne S. Building megaphylogenies for macroecology: taking up the challenge. Ecography. 2013;36:13–26. doi: 10.1111/j.1600-0587.2012.07773.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Thuiller W, Gallien L, Boulangeat I, de Bello F, Munkemuller T, Roquet C, et al. Resolving Darwin's naturalization conundrum: a quest for evidence. Diversity and Distributions. 2010;16:461–475. [Google Scholar]
  41. Thuiller W, Richardson D, Rouget M, Proches S, Wilson J. Interactions between environment, species traits, and human uses describe patterns of plant invasions. Ecology. 2006;87:1755–1769. doi: 10.1890/0012-9658(2006)87[1755:ibesta]2.0.co;2. [DOI] [PubMed] [Google Scholar]
  42. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate Imputation by Chained Equations in R. Journal of Statistical Software. 2011;45:1–67. [Google Scholar]
  43. Van Kleunen M, Weber E, Fischer M. A meta-analysis of trait differences between invasive and non-invasive plant species. Ecology letters. 2010;13:235–245. doi: 10.1111/j.1461-0248.2009.01418.x. [DOI] [PubMed] [Google Scholar]
  44. Vila M. Invasive species. What everyone needs to know. Science. 2013;342:424–424. [Google Scholar]
  45. Violle C, Choler P, Borgy B, Garnier E, Amiaud B, Debarros G, et al. Vegetation ecology meets ecosystem science: Permanent grasslands as a functional biogeography case study. Science of The Total Environment. 2015;534:43–51. doi: 10.1016/j.scitotenv.2015.03.141. [DOI] [PubMed] [Google Scholar]
  46. Walther GR, Roques A, Hulme PE, Sykes MT, Pysek P, Kuhn I, et al. Alien species in a warmer world: risks and opportunities. Trends in Ecology & Evolution. 2009;24:686–693. doi: 10.1016/j.tree.2009.06.008. [DOI] [PubMed] [Google Scholar]
  47. Weber E, Gut D. Assessing the risk of potentially invasive plant species in central Europe. Journal for Nature Conservation. 2004;12:171–179. [Google Scholar]
  48. Wesche K, Krause B, Culmsee H, Leuschner C. Fifty years of change in Central European grassland vegetation: Large losses in species richness and animal-pollinated plants. Biological Conservation. 2012;150:76–85. [Google Scholar]
  49. Westoby M. A leaf-height-seed (LHS) plant ecology strategy scheme. Plant and Soil. 1998;199:213–227. [Google Scholar]
  50. Wilson JRU, Richardson DM, Rouget M, Proches S, Amis MA, Henderson L, et al. Residence time and potential range: crucial considerations in modelling plant invasions. Diversity and Distributions. 2007;13:11–22. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Materials
NIHMS69435-supplement-Supplementary_Materials.docx (4.3MB, docx)

RESOURCES