SEARCH

SEARCH BY CITATION

Keywords:

  • Biological invasions;
  • canopy height;
  • ecological strategies;
  • exotic/introduced ranges;
  • plant traits;
  • SLA;
  • seed mass;
  • trait conservatism;
  • trait consistency;
  • invasions

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References
  10. Supporting Information

Aim

Invasion biologists have made an extensive exploration of ways to identify the characteristic traits of invasive plants based on the assumption that plant attributes remain similar when plants become invasive. However, this assumption needs to be critically evaluated when predicting successful future introductions.

Location

Global.

Methods

Using a global database of plant functional traits (i.e. specific leaf area, maximum canopy height and individual seed mass) encompassing 129 different species three questions were evaluated using a meta-analytical approach. (1) Do traits of introduced aliens change between native and introduced areas? (2) Do the responses show directionality, indicating that traits of aliens are either consistently higher or lower in their introduced range? (3) Are there smaller differences in species between native and introduced areas (within-species multitrait variation) than between two random species (between-species multitrait variability)?

Results

Mean trait differences (measured as log-response ratios) between native and introduced (invasive + naturalized), native and naturalized, or native and invasive areas showed no significant differences across evaluated species, even after controlling for invasion status and growth form. This pattern was the result of aliens showing both higher (Alienarea > Nativearea) and lower (Alienarea < Nativearea) trait values in their introduced areas. Furthermore, multitrait differences between populations in the native and introduced areas were significantly less than differences in the mean trait composition between species (as determined by contrast with two different null models).

Main conclusions

The similarity of evaluated traits between areas, in combination with a smaller within-species than among-species trait variability, are fundamental results for conservation and nature management efforts. The results presented in this study validate the assumption that mean trait measurements in the native areas are likely to be reasonably representative of trait mean values in the invaded areas, and support the use of trait-based prediction methods to evaluate the potential of an introduced plant to become established.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References
  10. Supporting Information

An important aim in the study of invasions has been the prediction of which introduced non-native species are likely to expand in a new geographic area (Rejmanek & Richardson, 1996) and which communities are more prone to invasions (Richardson & Pyšek, 2006; Olyarnik et al., 2009). Invasion biologists have searched extensively for methods and indicators that could determine species invasiveness (Rejmanek & Richardson, 1996; Richardson & Rejmanek, 2004) or the invasibility of a community (as reviewed by Daehler & Carino, 2000). Successful understanding of the mechanisms facilitating biological invasions would improve the accuracy of invasive species screening programmes and help to target recently naturalized species in time before they become serious problems.

Two implicit assumptions are at the core of the methods used most often for assessing the invasive potential of an introduced plant (discriminant functions, Rejmanek & Richardson, 1996; or scoring check lists, Daehler & Carino, 2000): (1) species characteristics are on average constant within each species (there is small variability across the entire genetic range and the environmental region in which it occurs), and (2) within-species trait variation is lower than between-species variability (so that trait-based methods of predicting invasiveness can be used to distinguish between successfully introduced, unsuccessfully introduced and native plants). Based on these two ideas, trait measurements in the native area can be used to predict future successful introductions, as they will reflect the traits of the species under the new conditions.

Little is known about how population characteristics and traits of introduced species do, or do not, change between their native and their invasive areas. Furthermore, there are various reasons why traits in the native and introduced areas might not be comparable. For example, novel introductions may originate from a few individuals or genotypes (i.e. founder effects) that, due to chance, may not represent the average trait values of the species in the native area (Hawkes, 2007; van Kleunen et al., 2010). Also, the novel conditions encountered in the introduced area may lead to phenotypic adjustments due to plasticity beyond the mean or variation found in the native area (Maron et al., 2007; Alexander & Edwards, 2010; van Kleunen et al., 2010). Additionally, during the expansion phase of invasions, traits may change as a consequence of novel selection pressures (Maron et al., 2004a), causing a shift in the association between trait values in native and introduced areas. As a result, it is quite possible that traits of introduced species differ between native and introduced areas with consequences for predicting the invasiveness of species.

Evaluating the degree of trait consistency in non-natives is also necessary for understanding which communities are most susceptible to invasions (Olyarnik et al., 2009). Consistency of traits between native and introduced areas would support the assumption that success of non-natives will be highest under conditions similar to those found in the native area (Daehler, 2003), provided that these conditions are realized in the introduced area. Alternatively, the differentiation in traits between native and introduced areas would support hypotheses in which phenotypic adaptation (Maron et al., 2007; van Kleunen et al., 2010; Hulme & Barrett, 2013) or rapid evolutionary processes (i.e. founder effects or selection; Maron et al., 2004b; Müller-Schärer et al., 2004; Prentis et al., 2008; Colautti et al., 2010) enhance geographical expansion in the introduced area and the invasion of novel types of habitats.

To date, only a few studies have addressed the question of whether traits of successful non-natives are similar between their native and introduced areas. Most of these are based on a single species, using relatively small databases and without considering an invasiveness gradient (native to naturalized to invasive). Moreover, previous studies have mainly focused on only one set of attributes – these being either qualitative vegetative (growth form, presence/absence of nitrogen fixation; Bradshaw, 1991; Thompson et al., 1995), reproductive attributes (seed size/number, germination rates and seedling survival; Buckley et al., 2003; Hawkes, 2007; Mason et al., 2008; Hierro et al., 2009) or proxies for competitive ability (Thebaud & Simberloff, 2001; Blumenthal & Hufbauer, 2007; Hawkes, 2007; Maron et al., 2007; Colautti et al., 2009, 2010; Parker et al., 2013). As a consequence of all this variability, these works have yielded different results, with many studies supporting the consistency of traits (Thebaud & Simberloff, 2001; Buckley et al., 2003; Hinz & Schwarzlaender, 2004; Maron et al., 2004b, 2007; Mason et al., 2008) while many others show between-area shifts in life-history attributes (Siemann & Rogers, 2001; Jakobs et al., 2004; Hawkes, 2007; Lavergne & Molofsky, 2007; Hierro et al., 2009; Parker et al., 2013). The heterogeneity across these results calls for a systematic evaluation of the consistency of species traits between their native and introduced areas, using a biogeographical approach, considering a large number of species and focusing on a diversity of traits (e.g. vegetative and reproductive among others).

Using a global-scale meta-analysis, this study examines quantitative changes in species traits between their native and introduced areas under similar ecological conditions. Three fundamental questions were asked: (1) Do traits of an introduced plant change between its native geographical extent and those areas where it has been successfully introduced? (2) When species are compared between areas, are responses within a trait consistently higher, consistently lower or show no direction across species? For example most species are expected to have higher specific leaf area (SLA), larger maximum plant height and lower individual seed mass (SWT) in the introduced areas. (3) Are multitrait between-areas differences within species more variable or less variable than multitrait differences across species?

To answer these questions, a database containing 129 non-native species with trait measurements from both their native and introduced areas was compiled. The database includes species from 56 plant families, encompassing all continents and most biomes across the world. The species in this study represent a spectrum of growth forms and lineages: 116 dicot species, 8 monocots, 3 gymnosperms and 2 pteridophytes (ferns and fern allies). Given the spatial and taxonomic coverage of the dataset used, this work is one of the most comprehensive evaluations of the within-species trait differences between native and introduced areas. Additionally, the use of multiple species and different non-native categories allows extrapolation of the results beyond a single species and invasion category (introduced versus naturalized versus invasive). This extrapolation is a much needed improvement upon most works on invasion biology, which often limit their scope to just the worst plant invaders.

Data and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References
  10. Supporting Information

Database compilation and trait selection

Publications and online databases were reviewed through keyword searches for references in Web of Science (1945–2009), examination of the references in these citations and direct communication with researchers. The keywords used were ‘plant traits’, ‘SLA’, ‘LMA’, ‘leaf size’, ‘leaf nutrients’, ‘plant height’, ‘seed size’, ‘seed weight’, ‘seed mass’, ‘seed production’, ‘plant attributes’, ‘LHS’, ‘plant physiology’, ‘weed’, ‘weeds’, ‘naturalized’, ‘invasive’, ‘exotic’, ‘noxious’, ‘introduced’, ‘alien’, ‘foreign’, ‘non-native’. Full details of the compilation of the database are presented in Appendix S1 in the Supporting Information.

Plant status was defined, following the native–naturalized–invasive continuum defined by Richardson et al. (2000). Henceforth, alien species are considered to be intentionally or accidentally introduced as a result of human activity. Naturalized plants are the subset of all aliens that sustain viable populations without direct intervention by humans. Invasive plants are the subset of all naturalized aliens that produce large numbers of reproductively viable offspring at considerable distances from parent plants. For each location/study, plant status (i.e. native, naturalized or invasive) was established using national, regional and global databases of invasive species (as explained in Appendix S1). Evaluated species could be classified as invasive in a region (or study) and naturalized in another, hence having values for both naturalized and invasive status; but no single study reported the same species as both naturalized and invasive. This classification determined whether differences exist between native versus naturalized and native versus invasive trait contrasts.

This study focuses on three traits which are often found to be among the attributes most related to invasiveness: specific leaf area (SLA; in cm2 g–1), maximum plant height (Hmax; in m) and individual seed mass (SWT; in mg). The SLA of a plant indicates how much light-capturing area it produces for each gram of leaf tissue. The SLA is negatively correlated with leaf thickness, lifespan, toughness and sensitivity to herbivory and positively correlated with mass-based measurements of leaf nitrogen content, photosynthetic capacity and plant relative growth rate, all of which are relevant to plant performance (Wright et al., 2004). Hmax is the simplest measure of plant growth form. This measure represents the balance between the gains from access to light, the cost of structural support (given the disturbance regime), water transport and sensitivity to biomass loss from mechanical disturbances. Hmax is also a proxy for other key traits indicating plant growth (Falster & Westoby, 2003). Lastly, SWT represents the balance between the number and size of offspring, which is often related to age-dependent survival probability (Westoby et al., 2002). SWT is often negatively correlated with dispersal distance (in the case of wind-dispersed species) and positively correlated with probability of seedling survival under light- or water-limited conditions (Leishman et al., 2000; Moles & Westoby, 2006).

SLA, Hmax and SWT, can be seen as proxies for the position of a species along three independent ecological strategy dimensions underpinned by various optimization trade-offs (Westoby et al., 2002). For instance, position along the SLA–Hmax–SWT spectrum differentiates plants that invest heavily in structural support and high seedling survival (k-selected) from those that invest in large seed output with little energy investment in structural support (r-selected). Typically there is little correlation between SLA, Hmax and SWT among co-occurring species and more correlation within species (Westoby et al., 2002), enabling the assessment of patterns of trait consistency between native and introduced areas [i.e. alien (naturalized + invasive), naturalized or invasive].

The compiled database comprises data from 129 different known successful alien species. These aliens had data for at least one of the three traits mentioned earlier. The data were compiled from their native area, their introduced area (naturalized and/or invasive) or both their native and introduced areas. Of these, 55 species contained information for at least two traits and 24 species contained data for all three traits from both areas. In most cases, traits were measured in more than one location, in either the native or introduced region (ranging from 1 to 6 for introduced and 1 to 8 for native populations). Traits were summarized per species per area (using the geometric mean) as the trait mean across all populations within a biome (biomes specified using the WWF Terrestrial Ecoregions of the World database; Olson et al., 2001). This allowed the statistical detection of between-area variation across different environments where the species are known to be successful. For those cases in which a species had both measured trait values in the naturalized and invasive areas, only the invasive value was used in comparisons of native versus alien areas. Summarizing species in this way assumes that trait means are representative of the true population (native or introduced) trait value in a particular biome and represents a summary across the evaluated local population trait means (as argued in the literature on intraspecific variability; Albert et al., 2010; Messier et al., 2010). This assumption was validated by the trait convergence across sites of similar environmental conditions, and the lower between-site versus within-site trait variability, in relation to the within-site trait variability, reported for different regions of the world (Reich et al., 1997). A summary of the database is presented in Appendix S2.

Univariate trait differences between native and introduced areas

Differences in trait means between native and introduced areas were assessed using log-response ratios (Hedges et al., 1999). This metric measures the proportional difference of traits in the introduced area relative to traits in the native area (i.e. effect size), thus controlling for those differences introduced by other covariates (e.g. growth form, habitat type, life zone or metric). The use of log-response ratios provided a standardized measurement of mean trait differences in native and introduced areas, where the 95% confidence interval defines the significance of the mean log-response ratio (the 95% CI does not overlap zero).

All effect size calculations followed the formulations of Hedges et al. (1999) and were implemented in R (R Development Core Team, 2009). The effect sizes for the evaluated conspecific comparisons were defined as the log-ratio between the mean trait value for native and introduced areas, and estimated using a flexible meta-analytic procedure (Nakagawa et al., 2007). The method is based on a linear mixed model (LMM) approach, with a restricted maximum likelihood method optimization (REML; nlme package in R; Pinheiro et al., 2009). The approach uses species as a grouping random factor and weights individual log-response ratios by the inverse variance of the corresponding effect size. This method allows comparison of the non-independent log-response ratios (differences across species) without merging the log-ratios within categories prior to a meta-analysis, so there is no loss of statistical power. Heterogeneity in the effect sizes among species within a trait, and between native versus naturalized and native versus invasive comparisons, were evaluated using the QREML statistic (Hedges et al., 1999; Nakagawa et al., 2007). The QREML statistic is the equivalent of residual heterogeneity in the random-effects models (that is the residual sum of squares) and is tested against a chi-squared distribution with degrees of freedom equal to one less than the number of classes (in this case species).

Given that the observed pattern of trait consistency might be influenced by the phylogenetic relatedness among contrasted species, the strength of the phylogenetic signal was evaluated using the Pagel (1992) phylogenetic dependence estimator (λ). No significant phylogenetic structure was found (Pagel λ indistinguishable from zero) for evaluated between-areas differences for all analysed traits, so there was no need to account for the phylogenetic relatedness amongst contrasted species.

Directionality of univariate trait differences

The mean log-ratio sign indicates the direction of the mean between-area trait differentiation: positive values indicate that traits in introduced areas are larger than traits in native areas while negative values indicate that traits in introduced areas are smaller than in native areas. Although the spread of effect sizes is a factor that strongly affects the significance of the measured differences between two groups, the directionality of the compared effect size can show a signal that is not captured by the global effect size (i.e. an overall tendency to positive or negative effect sizes). In fact, the classic principle of competitive exclusion (Gause, 1934) can be interpreted such that a very small but consistent trait differentiation (either positive or negative) would lead to competitive advantages or novel ecological configurations.

Because of this, the proportion of positive and negative effects sizes was statistically tested using the chi-squared test (one degree of freedom). The objective of this test was to determine whether trait values tended to be larger or smaller in the introduced geographical area. There are hardly any data on in the literature on how large a trait difference should be in order to make an ecological difference (e.g. a fitness increase between areas). For this reason, evaluating the directionality of the compared effect sizes is essential if we are to evaluate differences that might have ecological significance.

Multitrait differences in within- and between-species variability

A multitrait analysis allowed us to determine if the trait composition varied more or less than expected based on a random sampling of species. Differences between native and introduced areas [alien (naturalized + invasive), naturalized or invasive] were measured in a multitrait space using the Euclidean distance in a tridimensional trait means space. For this analysis, only taxa with information about all three traits in both native and introduced areas were considered. As traits are measured in different units, between-areas differences in this multitrait space were evaluated using standardized (mean = 0, SD = 1) log10-transformed traits. Standardizations were done using a global weighting procedure (as suggested by Cornwell et al., 2006) in which each trait was scaled relative to a global mean and variance derived from Ordonez et al. (2010). This ensures that trait measurements are equivalent world-wide. The means for multitrait between-areas differences were then evaluated for introduced, naturalized and invasive categories using a flexible meta-analytic procedure (i.e. an intercept-only mixed model of between-area multitrait Euclidean distances that use species as a grouping random factor).

Given that Euclidean distances are always positive, the significance of between-area differences has to be evaluated using null models (i.e. the difference in observed differences from a null distribution of between-areas differences derived from random communities of identical size). If trait differences within species tend to be smaller than those among species (i.e. observeddifferences < nulldifferences), this will support the use of trait-based prediction methods of invasiveness. On the contrary, if trait differences within species are larger than those among species (i.e. observeddifferences > nulldifferences), there would be little value in developing trait-based prediction methods of invasiveness as the within-species trait variability would mask any difference between co-occurring native and alien species.

The first null model randomized the trait values in the introduced area across evaluated species, restricting the randomization to species with the same growth form. This null model determined if the observed between-areas differences would differ from those expected from random between-areas differences (comparing the traits of two random native species versus introduced species). The second null model randomly compared two species of the same growth form, extracting the trait values from the native community. This null model determined if the observed between-area trait differences were higher (i.e. observed differences were more dissimilar than null) or lower (i.e. observed differences were less dissimilar than null) than those expected from random interspecific differences (comparing the traits of two random species). In both null models, 10,000 independent randomizations were made, resulting in 10,000 expected between-area trait differences. The mean and 95% confidence interval of null differences were used for comparison with the observed trait differences in alien (naturalized + invasive), naturalized and invasive areas.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References
  10. Supporting Information

Univariate trait differences between native and introduced areas

Effect sizes of conspecific comparisons between native and alien (naturalized + invasive) regions overlapped zero (Aliens, Fig. 1), indicating that the evaluated trait means of introduced species do not change between areas. In comparison, trait contrasts between native and naturalized areas showed marginally significant between-areas differences for Hmax and SWT (Naturalized, Fig. 1), with species in the naturalized area having, on average, taller Hmax (positive effect sizes) and smaller SWT (negative effect sizes). Lastly, trait comparisons between native and invaded areas did not show significant differences in trait means (Invasive, Fig. 1). Statistical power tests determined that the observed convergent patterns between native versus alien and native versus naturalized contrasts (log-ratios indistinguishable from zero) were not an artefact of sample size. However, native versus invasive contrasts had a low statistical power due to a low sample size, making it difficult to detect a difference between these two regions. Furthermore, the estimated log-response ratio values for each factor should be interpreted with care as there was a large variability in the magnitude and direction of these across species, pointing to other moderator variables as factors determining the directionality of the observed between-areas differences.

figure

Figure 1. Mean log-response ratios (circles, triangles and squares) and 95% confidence intervals (whiskers) of species trait differences between their native versus alien, native versus naturalized or native versus invasive areas. Evaluated traits are specific leaf area (SLA; in cm2 g–1) (circles), maximum canopy height (Hmax; in m) (triangles) and individual seed mass (SWT; in mg) (squares). Log-ratios refer to the species ratio between native and alien, naturalized or invaded areas. The mean log-response ratios were calculated using a linear mixed effect model with a restricted maximum likelihood method optimization and show no significant between-areas differences (95% confidence interval overlap zero). Numbers of records in each category: Aliens SLA = 77, Hmax = 108, SWT = 147; Naturalized SLA = 55, Hmax = 84, SWT = 119; Invasive SLA = 32, Hmax = 35, SWT = 40.

Download figure to PowerPoint

Comparisons between native versus naturalized and native versus invasive differences show homogeneous mean log-ratios between categories for both SLA (QREML(1) = 5.703, P = 0.983) and Hmax (QREML(1) = 5.703, P = 0.983). In comparison, SWT showed heterogeneous effect sizes (QREML(1) = 0.001, P < 0.001) as contrasts between native versus naturalized areas for this trait are significantly different and have a smaller variability. The overall homogeneity in native versus naturalized and native versus invasive log-response ratios indicated that variability in between-areas differences is similar for both contrasts (overlapping native versus naturalized and native versus invasive comparisons 95% CI; Figs 1 & 2). Nonetheless, the limited statistical power of native versus invasive contrast could be masking differences between these groups.

figure

Figure 2. Mean log-response ratios (circles, triangles and squares) for woody and non-woody species and 95% confidence intervals (whiskers) of species trait differences between their native versus alien, native versus naturalized or native versus invasive areas. Evaluated traits are specific leaf area (SLA; in cm2 g–1) (circles), maximum canopy height (Hmax; in m) (triangles) and individual seed mass (SWT; in mg) (squares). The mean log-response ratios were calculated using a linear mixed effect model with a restricted maximum likelihood method optimization and show no significant between-areas differences (95% confidence interval overlap zero). Number of records for woody species: Aliens SLA = 43, Hmax = 59, SWT = 92; Naturalized SLA = 35, Hmax = 49, SWT = 79; Invasive SLA = 10, Hmax = 12, SWT = 17. Numbers of records for non-woody species: Aliens SLA = 33, Hmax = 49, SWT = 51; Naturalized SLA = 20, Hmax = 35, SWT = 36; Invasive SLA = 21, Hmax = 23, SWT = 23.

Download figure to PowerPoint

Comparisons controlling for growth form showed no significant between-areas trait differences for native versus alien, native versus naturalized or native versus invasive contrasts (Fig. 2). As in the case of uncontrolled contrasts of growth form, native versus naturalized and native versus invasive comparisons showed homogeneous mean log-ratios between categories for both woody and non-woody plants for SLA (woody, QREML(1) = 0.442, P = 0.494; non-woody, QREML(1) = 7.851, P = 0.995) and Hmax (woody, QREML(1) = 3.182, P = 0.926; non-woody, Hmax QREML(1) = 1.064, P = 0.698), but not in the case of SWT (woody, QREML(1) = 0.001, P < 0.00; non-woody, QREML(1) = 0.001, P < 0.001).

Directionality of univariate trait differences

The fraction of species showing no change (SLA 1%, Hmax 16%, SWT 48%), an increase (SLA 39%, Hmax 50%, SWT 22%) or a decrease (SLA 30%, Hmax 34%, SWT 30%) in mean trait value changed across evaluated traits. Although there appears to be some signal of plastic or rapid evolution towards lower SLA, greater height and smaller seeds among naturalized species in their introduced areas (Fig. 1), the proportion of effect sizes with either positive or negative values was statistically indistinguishable for SLA (native versus alien, χ(1) = 3.482, P = 0.062; native versus naturalized, χ(1) = 5.4, P = 0.02; native versus invasive, χ(1) = 1.059, P = 0.303) and Hmax (native versus alien, χ(1) = 0.009, P = 0.926; native versus naturalized, χ(1) = 0.044, P = 0.833: native versus invasive, χ(1) = 1.324, P = 0.25) supporting the between-areas trait consistency hypothesis. However, the native versus alien and native versus naturalized contrast of SWT showed a statistically dissimilar proportion of effect sizes with either positive or negative values (native versus alien, χ(1) = 47.848, P < 0.001; native versus naturalized, χ(1) = 49.869, P = 0.001; native versus invasive, χ(1) = 3.429, P = 0.064), with most between-areas comparisons for this trait having negative log-response ratios (i.e. SWTalien-area < SWTnative-area).

When controlling for growth form, the proportion of positive and negative effects for SLA and Hmax showed no statistical differences for woody and non-woody plants for both native versus naturalized and native versus invasive contrasts. As for the uncontrolled contrasts, the proportion of positive and negative effects for SWT was statistically dissimilar for native versus naturalized comparisons, with most of the log-response ratios being negative (woody, χ(1) = 35.561, P < 0.001; non-woody, χ(1) = 11.111, P = 0.001). This was not the case for native versus invasive comparisons, as the proportion of positive and negative effects showed no differences for woody (χ(1) = 2.579, P = 0.108) and non-woody plants (χ(1) = 1.087, P = 0.297).

Multitrait differences in within- and between-species variability

Relative to the null models, differences in mean traits within species (observed multitrait differences) tend to be smaller than those among species (null multitrait differences), indicating that plasticity is a relatively minor player in the success of introduced plants. Observed multitrait differences for native versus alien, native versus naturalized and native versus invasive areas were significantly lower than the null model expectation of comparing two random native versus introduced area plants or two random species (all species comparisons, Table 1). Along the same lines, comparisons for woody and non-woody species showed, that relative to the null models, observed multitrait differences were significantly lower than those expected from null model 1 (random native versus alien areas comparison, Table 1) and null model 2 (random species comparison, Table 1).

Table 1. Change in multitrait trait composition of species between their native versus alien, native versus naturalized or native versus invasive areas. The significance of between-areas differences was determined using null models (null model 1, random native versus alien areas comparison; null model 2, random species comparison). Null model between-area trait differences were estimated from 10,000 independent randomizations. Significance values were determined as the total number of Observed > Null cases divided by 10,000
Growth formCompared areasnObserved multitrait distanceNull model 1 multitrait distancePNull model 2 multi-trait distanceP
Mean [95% CI]Mean [95% CI]Mean [95% CI]
All speciesNative versus alien540.315 [0.003, 1.137]1.126 [0.946, 1.316]< 0.0011.096 [0.904, 1.293]< 0.001
Native versus naturalized350.331 [0.017, 1.167]1.119 [0.899, 1.340]< 0.0011.089 [0.845, 1.330]< 0.001
Native versus invasive290.328 [0.002, 1.190]0.923 [0.735, 1.115]< 0.0010.868 [0.664, 1.077]< 0.001
Woody speciesNative versus alien230.266 [0.003, 0.711]1.283 [0.988, 1.617]< 0.0011.265 [0.935, 1.612]< 0.001
Native versus naturalized170.261 [0.011, 0.712]1.405 [0.990, 1.829]< 0.0011.376 [0.941, 1.837]< 0.001
Native versus invasive80.283 [0.008, 0.749]0.644 [0.375, 0.891]< 0.0010.604 [0.264, 0.910]< 0.001
Non-woody speciesNative versus alien300.363 [0.040, 1.169]0.914 [0.773, 1.078]< 0.0010.888 [0.726, 1.060]< 0.001
Native versus naturalized180.397 [0.057, 1.175]0.843 [0.679, 1.029]< 0.0010.820 [0.604, 1.025]< 0.001
Native versus invasive200.362 [0.046, 1.200]0.917 [0.740, 1.128]< 0.0010.844 [0.620, 1.054]< 0.001

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References
  10. Supporting Information

In general, this study shows how individual mean traits and the average trait composition of species does not change between native and introduced [alien (naturalized + invasive), naturalized or invasive] areas. This was demonstrated by the mean log-response ratios for individual traits overlapping zero, the variability in the directionality of between-areas differences, and multitrait differences being indistinguishable from the expectations of two different null models.

These results are consistent with those of other smaller scale meta-analyses reporting no changes in the height of species between their native and introduced areas (Thebaud & Simberloff, 2001); consistent plant vigour between areas, due to increased intraspecific competition in the new range (Hinz & Schwarzlaender, 2004); similar seed mass and seed production of invasive species between their introduced and original areas (even after growth form is taken in to consideration; Mason et al., 2008); and analogous fitness, size, growth rate, shoot allocation, leaf-area allocation and physiology of invasive species when compared with their native area counterparts (van Kleunen et al., 2010).

Based on Richardson et al.'s (2000) classification of introduced species, it was shown that mean trait consistency did not increase with introduction status (i.e. plants classified as invasive are as dissimilar from native species as natives are from those classified as naturalized). This supports a recent study, showing that invasive species express large performance variability between ranges, with some species performing better in the introduced area while others show no between-areas differences (Parker et al., 2013). Furthermore, the transition between naturalized and invasive does not imply that traits differentiating naturalized from invasive species are necessarily the same as the traits that determine whether a plant will be successful once introduced. Nonetheless, it is necessary to emphasize that the lack of significance in mean trait differences between native and invaded areas could be an artefact of a small sample size for invasive areas (in native versus invasive contrasts statistical power is limited to determining differences from native versus naturalized contrasts) or reflect the limitations of using mean trait values instead of measures of within-species trait variation (within-site intraspecific trait variability could change in response to different environmental constraints; Albert et al., 2010).

The observed between-area similarities in traits, could be considered as the result of stabilizing selection (Ackerly, 2003; Ackerly, 2009; Alexander & Edwards, 2010), habitat filtering (Kraft et al., 2007), niche conservatism (Pearman et al., 2008; Wiens et al., 2010) or analogous phenotypic adaptations to predators and parasites in the introduced range (Hawkes, 2007; Verhoeven et al., 2009). Alternatively, population genetic constraints (e.g. founder effects; Ackerly, 2003; Cavender-Bares et al., 2004), new environmental restrictions and novel selection pressures (Maron et al., 2004a; Wilson et al., 2009; Alexander & Edwards, 2010) can potentially cause a shift in the relationship between trait values in native and introduced ranges. As a result, traits that are less affected by environmental and ecological factors (e.g. wood density or leaf life span, height, seed mass) will show a smaller change in their means between areas (regardless of founder effects or other mechanisms) than those that are highly influenced by introduction history, biotic, ecological or environmental gradients (e.g. canopy transpiration, rhizome re-sprouting and relative growth rate).

It cannot be suggested that the consistency in mean traits between native and introduced areas reported in this study is the only possible, or even the most likely, outcome when a plant is introduced. For example, it has been shown that the reproductive (Buckley et al., 2003; Hawkes, 2007; Hierro et al., 2009), defensive (Hawkes, 2007) and performance attributes (Blumenthal & Hufbauer, 2007; Hawkes, 2007; Maron et al., 2007; Colautti et al., 2010) of non-native species change between areas; particularly when changes in environmental conditions are taken into consideration (Maron et al., 2007; Colautti et al., 2009, 2010). However, these examples are based on experimental contrasts aimed at understanding the invasion of a particular species at a particular local field site, rather than generalization across a broad taxonomic spectrum. When several species are considered, as is the case in this study and other meta-analyses (Thebaud & Simberloff, 2001; Hinz & Schwarzlaender, 2004; Mason et al., 2008; van Kleunen et al., 2010; Parker et al., 2013), attributes of introduced species show no change between areas, and the direction of these differences shows no significant pattern.

The reported mean trait similarity between native and introduced areas indicates the conservation of functional responses of non-natives in space, something that can potentially contribute to the success of an introduced species (Scheffer & van Nes, 2006). Given the limitations of the database used, determining the influence of founder effects and the temporal dimension of these patterns was not possible as information on the introduction history, pressure or how long the evaluated aliens have occupied the new introduced areas is currently not available. However, there is no strong evidence that rapid evolution in novel areas would be of wide importance in nature as an explanation for the spatial and temporal dynamics of introduced plants (Webb, 1988; Williams et al., 2004). Further work is needed to evaluate trait shifts of co-occurring species in native and novel areas, as well as large-scale genetic screening to differentiate populations.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References
  10. Supporting Information

The fact that mean traits did not change between areas and less trait variation was recorded within than among species is vital information for conservation and nature management efforts. These results validate the assumption that trait measurements in the native areas (i.e. mean traits) are likely to be reasonably representative of traits in the invaded area, and hence could be used to assess the establishment potential of an introduced alien. Furthermore, within-species multitrait differences tend to be smaller than those among species, supporting the use of trait-based prediction methods to assess the establishment potential of an introduced alien.

Consistency of mean traits between areas provides support to those hypotheses based on phenotypic attraction in the case of naturalized species, where a plant will be successful when introduced to an area with environmental conditions similar to those in its native area due to pre-adaptation of its traits. Additionally, the results shown here are pertinent to the evaluation of screening mechanisms, particularly discriminant functions (e.g. discriminant analysis Z-scores; Rejmanek & Richardson, 1996) or classification checklists (regional screening systems; Daehler & Carino, 2000), as it has been shown that the use of attributes of species in their native area is a good representation of traits in an introduced area. Nevertheless, the facts that the directionality of between-areas trait differences was species specific and that no differences were found between native versus naturalized and native versus invaded areas are important for the development of a comprehensive trait-based screening system for potentially invasive species.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References
  10. Supporting Information

The author thanks J. C. Hopcraft, H. Olff, N. Bhola, the managing editor, and three anonymous referees for useful comments and discussions during the elaboration of this manuscript. The LEDA project, the Royal Botanic Gardens Kew Seed Information Database, and I. J. Wright are acknowledged for contributing data. The University of Groningen (Netherlands) Ubbo Emmius scholarship, and the HISTFUNC project (ERC Starting Grant 310886) supported A.O. during the writing of this manuscript.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References
  10. Supporting Information

Alejandro Ordonez is a post-doctoral research associate at the University of Aarhus (Denmark). His research interests focuses on past and future global change phenomena (climate change, species invasions) and their impact on species ranges, community assembly and functional diversity.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
geb12123-sup-0001-si.pdf76K

Appendix S1 Description of the protocol used to build the native versus introduced area database.

geb12123-sup-0002-si.pdf328K

Appendix S2 Database of the trait values for the native and introduced areas.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.