Invasion science, ecology and economics: seeking roads not taken

Research on post-establishment evolution in nonnative plant populations has focused almost exclusively on testing the Evolution of Increased Competitive Ability (EICA) hypothesis, which posits that the lack of specialized herbivores in the invaded range drives evolution in nonnative plant populations. Fifteen years of conflicting EICA test results suggest that selection pressures other than specialized herbivory are important in driving post-establishment evolution in invasive species. Alternative hypotheses, such as the Evolution of Reduced Competitive Ability (ERCA) hypothesis, have been proposed but have received little attention or testing. We argue that the lack of consensus across studies that test EICA may be due in part to the lack of consistent definitions and varying experimental design parameters, and that future research in this field would benefit from new methodological considerations. We examined previous work evaluating post-establishment evolution and evaluated the range of study systems and design parameters used in testing the EICA hypothesis. Our goal was to identify where different uses of ecological terms and different study parameters have hindered consensus and to suggest a path forward to move beyond EICA in post-establishment evolution studies. We incorporated these methods into a design framework that will increase data harmony across future studies and will facilitate examinations of any potential selection pressure driving evolution in the invaded range.


Introduction
As members of the editorial board of Neobiota who, for various reasons, didn't get our names on the original editorial , we would like to add a coda to it. Even though there were 38 bullet points listing areas in invasion science where more work is needed, we would like to mention additional areas that we hope would be addressed in future issues of Neobiota. Like the other editors, we would like this innovative and exciting new journal to lead the way in all areas of invasion science. As the graphs in Gurevitch et al. (2011) and  show, the literature on invasions has been increasing almost exponentially since the early 1980s and so we cannot expect any list of areas of interest to stay complete and up to date for very long.
Three areas that we would like to stress are the interaction between invasion science and economics and the role that invasion science should play in advancing pure ecology in two areas, population dynamics and ecosystem ecology. Neither ecology nor economics appears as a word in the original bullet list, but many of the topics are obviously ecological while none are obviously economic. For economics, we want to point out its relevance to invasion science and the feedback between the two disciplines, particularly in a rapidly changing world with powerful new emerging economies. For ecology, we want to emphasise not what ecology tells us about invasions but what invasions reveal about ecology and evolution at two scales.

economics
There are two recent multi-author books that show the extent and variety of the interaction of economics with invasion science (Keller et al. 2009, Perrings et al. 2010) and also the variety of approaches to tackling these problems, though in this field models are almost always important. Both books also cover management and policy. As any manager knows, finance, either explicit in income or implicit from volunteers etc., determines what can be done and which problems can be tackled. But economics is about much more than costs. Economic analysis and theory are important in developing policy for dealing with invasive problems and serve as motivators for both the public and private sector to take action.
The ecological and economic dimensions of the problem of invasive species are connected at different levels. Many of the changes that lead ecosystems to be more vulnerable to the impact of invasives (e.g. fragmentation, disturbance, loss of diversity, pollution) are direct consequences of economic behaviour. The ecological mechanisms affecting invasives, such as functional diversity and dispersion, are correlated with trade, transport and travel. The consequences of the reduction in ecosystem functionality and the ability to provide ecological services have direct implications for the value of the output and ecological capital of the system.
At every level, the ecological impacts of economic activities are incidental to and usually ignored by the actors concerned. These impacts are externalities of the market transactions; they are not taken seriously by those making the transactions perhaps because they are not held legally responsible for the impacts nor are the markets directly affected by these impacts. Instead these impacts are often borne by those who receive little or no benefit from the market transactions. In addition, quantifying some ecosystem services (and disservices) is difficult and approaches to do so vary necessarily by scale, type of service, and region (Meyerson et al. 2005). Therefore, the major economic problems of invasions are first to understand the nature of invasive species externalities, second to evaluate the consequences they have for well-being, and third to develop policies and instruments for their internalisation.

ecology
The intimate relation between ecology and invasion science is well known. Less recognised is that invasions throw light on some ecological processes that can be more difficult to study in uninvaded systems. Many ecosystems are close to equilibrium or are following a moving equilibrium from seasonal or longer changes which makes detecting significant changes a long-term prospect and out of sync with two to four year funding cycles. Invasions supply ample examples of unintended experiments with systems well away from equilibrium, often over relatively short time periods. The resultant changes are informative both for the population dynamics of individual species and for the coevolution of communities.
In population dynamics, the growth and spread of populations are natural aspects to study in invaded systems. We will just mention two aspects, lag and the pattern of spread.
There is much misunderstanding of lag. It occurs when a population is not growing in numbers at all. When there are, as so often, casuals, i.e. individuals not producing population growth, it can be difficult to be sure whether the population is growing or not. A common problem is to mistake the early stages of logarithmic or quadratic growth with lag. The quickest solution is usually to plot transformations, e.g. log or square root, of the species counts. Too many statisticians want arithmetic plots which frequently conceal the behaviour of a population. Lags are important for management as they result in invasive species that appear to be harmless, sleeper weeds and such, leading to a lack of action when it would still be relatively cheap and easy to control or even eradicate a population. The lengths of lags are surprisingly variable and some can be quite long. Williamson (2010) found lags from 7 to 154 years in some beautiful Czech plant data, though 22 of 50 species showed no lag. The median lag was 41 years, a period long enough to make a serious management problem. We would emphasize though that lag is an important and little understood aspect of population growth in every sort of species.
It is easy to suggest causes for lag, such as the wrong habitat at introduction, the wrong genotype first introduced or Allee effects, but we know of very few cases when the cause can even be guessed let alone demonstrated. One such is in Epilobium ciliatum, a willow herb native to North America, first established in England in the midlands before 1891 but which didn't spread though it did establish. Another introduction into south-east England in the 1920s spread rapidly throughout Britain, overrunning the midlands. It is the plant species with the fastest known spread in Britain so far (Williamson 2011). In that case, the failure of the 19 th century introduction to spread can be ascribed with some confidence to genotype. Not that that helps much without knowing the genes and what characters they affect and why those characters prevent spread in England. Lags, though important, are difficult to study and much neglected by ecologists.
Conversely, spread is often quite easy to study though here again some biologists have made an elementary mistake, namely regarding the increase in records as an increase in population size, while nearly always it is only a measure of population range. Possibly this comes from models of spread involving population parameters such as the intrinsic rate of natural increase. Long term records of the ranges of invading species show much variation in the rate and pattern of spread, phenomena complicated by the heterogeneity of natural systems (Williamson 2010). Again, practically nothing is known of the causes of such variation. Yet such knowledge would be most useful in predicting the behaviour of new invaders. It would also, importantly, advance our understanding of basic ecology, and perhaps also provide insights into range expansions and spread under global climate change. The quantitative study of both lag and spread depends on long term, good quality and consistent data. So invasions can be of benefit to ecological science by strengthening the case for long term studies.
The other ecological topic we would like to mention is that species invasions may help us to better understand the mechanisms which generally govern ecosystems. How does co-evolutionary history among species shape the diversity, functioning and stability of ecosystems? Although this topic is somewhat related to the bullet point "Ecoevolutionary feedback between invasive traits and ecosystem function" in the original list, it focuses on different processes.
Interactions among plants, herbivores and microbes influence ecosystem functions (Stein et al. 2009(Stein et al. , 2010 and are commonly thought to be shaped by joint evolutionary history (Thompson 1994); exotic species are disconnected from such coevolved relationships and may encounter evolutionarily naïve communities. This in turn may affect competitive outcomes among species (Thorpe et al. 2011) and lead to the disruption of species interactions in the invaded ecosystem (Stinson et al. 2006). It is therefore reasonable to suggest that ecosystem properties are shaped by the coevolutionary history among species, but this hypothesis has been tested only in a very few studies. Recently, Wilsey at al. (2009) set up experimental communities either composed of native plants or composed of exotics which were phylogenetically related to the natives, and revealed that exotic communities declined faster in diversity, but showed higher productivity than the native communities. These findings demonstrate how fine-tuned species interactions within communities are, and suggest that not only the invasive traits of species, but also the novelty of exotic species per se may affect ecosystems. In another study, Maron and Marler (2008) showed that the diversity-productivity relationship was even reversed when native plant communities were experimentally invaded by exotics. Perhaps, the dislocation from co-evolved relationships may not only contribute to unifying theories explaining invasions (e.g. Hallett 2006), but also to a predictive framework for the ecosystem impact of invaders.

Introduction
It has been commonly observed that life-history traits of nonnative plant species vary across habitats in native and introduced ranges, most conspicuously as either increased growth (Crawley 1987;Thebaud and Simberloff 2001) or decreased growth (Crawley 1987;Siemann and Rogers 2003a;Bossdorf et al. 2004). While trait variations across habitats may in some cases be plastic (i.e., non-genetic) responses to local environmental conditions (e.g., Fowler et al. 1996;Willis et al. 2000;Keane and Crawley 2002), a growing number of studies have examined this variation from an evolutionary perspective: the characteristics of the invaded habitat amount to a suite of novel selection pressures that drive post-establishment evolution in invading populations (e.g., Blossey and Nötzold 1995;Cody and Overton 1996; Table 1). To date research regarding postestablishment evolution has been heavily focused on testing the Evolution of Increased Competitive Ability (EICA) hypothesis. This hypothesis posits that nonnative plant populations may evolve a modified energetic tradeoff in which vegetative growth is increased while herbivore defense is decreased due to the lack of specialized herbivores in the invaded range. The potential for increased reproductive effort was also discussed, though not tested (Blossey and Nötzold 1995). Increases in vegetative growth were interpreted broadly by Blossey and Nötzold (1995) as "increased competitive ability." Because research regarding post-establishment evolution has been so tightly focused on testing the EICA hypothesis, much of the literature in this field evaluates a single reduced selection pressure: the lack of specialized herbivores in the invaded range. However, the inconclusive support found for the EICA hypothesis suggests that factors other than herbivore release may drive post-establishment evolution in some systems (Bossdorf et al. 2005;Fornoni 2010). To this end, researchers have called for an expanded consideration of habitat characteristics that may act as selection pressures in the invaded range (Callaway and Maron 2006;Whitney and Gabler 2008). As this field expands, however, the approaches to testing hypotheses focused on post-establishment evolution merit careful consideration so that results across studies can be more easily used to form generalized conclusions. In addition, transitioning the focus from "competitive ability" (which is not well defined in this context) would enhance future research by more broadly focusing on evolved trait modifications that impact survival and fecundity.
There have been a handful of studies evaluating selection pressures other than specialized herbivory, although they have not received the attention and scrutiny given to the EICA hypothesis. Maron et al. (2004), for example, demonstrated morphological adaptations associated with latitude in nonnative populations of Hypericum perforatum, and Van Grunsven et al. (2009) found that a lack of soil pathogens resulted in increased growth for nonnative populations of Carpobrotus edulis. More broadly, Bossdorf et al. (2004) proposed the Evolution of Reduced Competitive Ability (ERCA) hypothesis, which posits that post-establishment evolution can be driven by the amount of plant competition found in the invaded habitat. Specifically, plant populations invading areas with fewer competitors should evolve an energetic tradeoff in which vegetative growth is decreased while other traits such as reproductive effort or herbivore defense are increased (Bossdorf et al. 2004). One potential implication of this hypothesis is that intraspecific competitive interactions may be reduced in monotypic or near-monotypic stands, increasing stand-level fitness while decreasing individual growth. A test of this hypothesis in any organism or geographic system has not yet been published.
Expanding our knowledge of post-establishment evolution beyond evaluations of the EICA hypothesis would address the omission of evolutionary potential from invasive species weed risk assessments, which attempt to predict the impact of an invading species on a given habitat based on the combination of species traits and habitat characteristics (Whitney and Gabler 2008). A better understanding of how the traits of invading species may rapidly change over time via natural selection would facilitate predictions regarding the immediate and long-term impacts of a given invasion (Daehler et al. 2004;Whitney and Gabler 2008). Such an understanding requires evaluations of not only those selection pressures posited by EICA or ERCA, but any number of habitat characteristics that vary across the native and invaded ranges of a given species.
We examined previous work regarding post-establishment evolution in order to better understand why outcomes across studies have been inconsistent. Because this literature has focused primarily on testing the EICA hypothesis, we focused our analysis within the field of evolution and invasive species by specifically examining tests of the EICA. Unlike previous reviews, we specifically evaluated study systems and methodologies in order to identify design parameters that will allow better synthesis across research on postestablishment evolution. We also focused on the use of the term "competitive ability" and how its definition varied based on the context of experimental designs. We searched for relevant literature published since the introduction of EICA that explicitly tested the predictions of the EICA hypothesis in a common garden or reciprocal transplant design, resulting in 58 studies. We focused on common garden and reciprocal transplant designs because of their frequent use and their ability to minimize the effects of phenotypic plasticity in the examination of evolved trait differentiation. We reviewed each study and table 1. Published hypotheses related to post-establishment evolution.

Reference
Description Blossey and Nötzold 1995 Evolution of Increased Competitive Ability (EICA): Evolved increase in growth, decrease in defense associated with lack of herbivores in invaded range. Sexton et al. 2002 Invading populations benefit first from plasticity, then from local adaptation Maron et al. 2004 Latitudinal clines drive local adaptations in nonnative populations. Bossdorf et al. 2004 Evolution of Reduced Competitive Ability (ERCA): Low amounts of plant competition result in an evolved decrease in growth, increase in reproduction, and defense.

Roman 2006
Multiple introductions and hybridization increase invasion success through increased genetic variability.

Blumenthal 2006
Resource-Enemy Release Hypothesis (R-ERH): Resource availability effects how enemy release drives plastic and genetic trait variation. Van Grunsven et al. 2009 The absence of soil pathogens results in an evolved increase in growth. recorded information regarding study systems and design parameters, including whether abiotic and biotic variables were reported for seed collection sites, the inclusion of introduction history, sample size, the traits measured and metrics used, and the incorporation (or lack thereof) of competition in experimental manipulations. We also noted whether each study ultimately found support for the predictions of the EICA hypothesis (Table 2).
Based on our results, we developed a framework that can be used to move beyond EICA to evaluate a broad range of habitat characteristics that can act as selection pressures driving post-establishment evolution. Our intent is to facilitate future research that expands the consideration of potential selection pressures and encourages integration of results across study species and organisms.

Methodological variability in evaluating selection pressures: lessons from tests of the eICa hypothesis
The variation in sample sizes of both native and nonnative populations used in common garden research is one of the most readily apparent differences among studies of post-establishment evolution ( Table 2). The logistical difficulties of obtaining individuals from both the introduced and invaded ranges likely contribute to the small number of populations used to represent each range in most studies. Though a handful of studies used ten or more populations in each range (e.g., Blossey 1996;Blair and Wolfe 2004;Güsewell et al. 2006), most tests of the EICA hypothesis have used a small number of populations (e.g., <10) and 13 of the 58 studies we reviewed used just one population to represent a native or invaded range (e.g., Blossey and Nötzold 1995;Lankau et al. 2004;Rogers and Siemann 2005;Franks et al. 2008).
Most of the studies we examined measured only one trait despite the fact that hypotheses focused on post-establishment evolution generally discuss multiple traits related to one another through energetic tradeoffs. The EICA hypothesis makes two predictions that were explicitly tested by Blossey and Nötzold (1995): that nonnative populations will demonstrate increased growth and will also demonstrate decreased herbivore defense. Of the 58 studies we reviewed, only 15 examined differences in both growth and defense in the same species. Of these, only eight found support for both predictions of the EICA hypothesis ( Table 2). Twelve of the 58 studies identified reproductive effort as a third energetic "sink" in evolved tradeoffs, though only three studies quantified all three of these traits in the same species (Fig. 1, Table 2).
Common garden designs also varied by study, either as outdoor gardens or greenhouse benches. Those studies that used outdoor gardens diverged further in whether plants were grown in pots (e.g., Willis and Blossey 1999) or were planted directly into the ground (e.g., Siemann and Rogers 2001). Each approach has its advantages and disadvantages: while greenhouse studies allow researchers to better control environmental variables such as temperature or photoperiod, some researchers have argued that growth under these conditions may be too artificial for evaluating growth in wild populations (Gibson et al. 1999). Design parameters of studies testing the EICA hypothesis. In the column labeled "Abiotic data," "1" indicates that researchers tried to incorporate a variety of abiotic environments in seed collection sites while "2" indicates that researchers tried to utilize similar environments. In the column "Traits," "G" = vegetative growth, "T" = herbivore tolerance, and "R" = reproductive effort. The column labeled "Comp." indicates whether or not competition was incorporated into experimental manipulations. For "Metrics," growth metrics are denoted by "B" = biomass, "H" = height, "NL" = number of leaves, " "LA" = leaf area, "BA" = basal area; tolerance metrics are denoted by "HM" = herbivore mass, "HA" = herbivore abundance, "LD" = leaf damage, "DC" = defense chemicals, "TD" = trichome density; reproductive metrics are denoted by "RM" = reproductive mass, "ReMR" = reproductive mass ratio, and "NF" = number of fruits. In the column "Sample Size," the values correspond to the number of native and nonnative populations used, respectively, unless the sample size was not differentiated by range. The column "EICA" indicates whether support was found, with "Partial" indicating that multiple traits were tested but not all results supported predictions, and "1" indicating that support was found, but only when plants were grown in the absence of competition.

Article
Abiotic data The term "competitive ability" had multiple interpretations across the studies that we reviewed. In studies where individuals are grown alone (e.g., Blossey and Nötzold 1995;Maron et al. 2004), "competitive ability" appeared to be interpreted as the ability of an individual to secure resources for itself regardless of the presence of neighbors. Alternatively, studies in which individuals are grown alongside competitors (e.g., Leger and Rice 2003;Vilà et al. 2003;Bossdorf et al. 2004) appeared to interpret "competitive ability" as the ability of an individual to preempt resources for itself specifically in the presence of neighbors. The distinction is an important one: Blumenthal and Hufbauer (2007) found that biomass was higher in nonnative populations than native populations when grown in a common garden, but only when individuals were grown alone. When exposed to actual competition, there was no difference in biomass between nonnative and native populations. Of the 58 studies we examined, only  et al. 2008 No Yes G T Yes B, LD 2&2 Yes ten exposed individuals to actual competition by growing study individuals alongside competing neighbors. Of these, only one found support for both predictions of the EICA hypothesis ( Fig. 1, Table 2). These different definitions of competitive ability have also muddled our understanding of the energetic tradeoffs being examined in these studies. The focus on testing EICA has resulted in the general adoption of the term "competitive ability" as synonymous with "vegetative growth," as it was interpreted by Blossey and Nötzold (1995). This is an oversimplification since EICA links multiple traits that are influenced by competition through energetic tradeoffs, including growth, defense, and, in more recent work, reproduction (Herms and Mattson 1992; Bazzaz and Grace 1997; Table 2). It is possible that, as predicted by the ERCA hypothesis, nonnative plant populations could produce decreased total biomass but increased fruit mass, resulting in increased fecundity in future generations (Bossdorf et al. 2004). This would be considered reduced competitive ability despite the increased success of the nonnative plant population.
In evaluating specialized herbivory as a selection pressure, studies of post-establishment evolution have generally neglected other habitat characteristics that can act as confounding variables across seed collection sites. Because comparisons of native and nonnative populations often use seeds collected on different continents, abiotic characteristics such as photoperiod or climate may vary significantly across the study area. For example, of the 58 studies we reviewed, only 15 identified differences in abiotic conditions (e.g., climate and/or photoperiod) between the native and invaded ranges. None of the studies we reviewed recorded information regarding the biotic characteristics of the collection site (e.g., plant community composition) other than differences in herbivore assemblages (Fig. 1, Table 2). This is particularly significant in light of a study by Williams et al. (2008), which found that the results of common garden experiments change depending on the geographic location and habitat characteristics of the common garden.
Furthermore, the 15 studies that did report abiotic characteristics often disagreed as to whether consistency in abiotic factors across the sampling range was a desired component of the experimental design, despite having similar aims in evaluating evolution in nonnative populations. Several studies noted that seeds utilized in a common garden were intentionally collected from a wide variety of habitat types in order to incorporate environmental heterogeneity across the distribution of the species (e.g., Siemann and Rogers 2001; Vilà et al. 2003;Cano et al. 2008;Cripps et al. 2009). Cripps et al. (2009, for example, chose populations "covering a wide geographic range to obtain representative samples" across the distribution of Lepidium draba. Conversely, other studies noted that seeds for common gardens were collected from very similar habitats in order to minimize the potential for confounding habitat variables (e.g., Leger and Rice 2003;Buschmann et al. 2005;Blumenthal and Hufbauer 2007). Erfmeier and Bruelheide (2005), collected seeds only from populations of Rhododendron ponticum growing on north-facing, forested slopes with an aspect of 10-20˚. This variation in experimental design demonstrates a fundamental difference in how researchers assess the potential for confounding variables and leads to varying degrees of robustness across studies.
Introduction history and subsequent spread of an invader are potentially influential factors often missing in reports of post-establishment evolution. This history is fundamental to understanding the selection pressures to which a species has been exposed. For example, Maron et al. (2004) used genetic analyses to determine that multiple introductions of Hypericum perforatum had occurred in the US, thereby influencing the amount of genetic variability in nonnative populations. However, detailed introduction histories are often unavailable and genetic analyses can be time-and costintensive, likely contributing to the omission of such information from many studies. Of the 58 studies we examined, 44 studies reported some data regarding introduction history, though often this information was limited to the date of first introduction to a country or continent and as such does not provide a detailed description. The remain- ing 14 studies made no mention of introduction history. Only two studies conducted genetic analyses to assess variability among populations ( Fig. 1, Table 2). The broad range of approaches used in testing EICA may be one factor that limits consensus among the collective results. Twenty-four of the 58 studies we examined found support for the predictions of the EICA hypothesis, while 27 did not. Five studies found partial support, and two found support only when individuals were not exposed to competition (Fig. 1, Table 2). While EICA is likely an accurate predictor of evolutionary changes in some (but not all) of the species on which it has been tested, the fundamental differences in experimental designs and the use of loosely defined terms such as "competitive ability" in these studies may have hindered a more complete understanding of the applicability of the EICA hypothesis and of post-establishment evolution in general. While it is possible that meta-analyses could be used to make generalizations across methodologies (e.g., Colautti et al. 2009), the ecological relevance of results might be easier to interpret if some fundamental design parameters, such as the incorporation of competition into experimental manipulations, were similar across studies.
Post-establishment evolution is an area of research that has attracted substantial attention since the introduction of the EICA hypothesis, but the tests to date have not  Table 2. "*" Indicates that support for the EICA hypothesis was found only when plants were grown in the absence of competition. yet provided a meaningful consensus. One approach to facilitate progress in this field is to move towards standard definitions and comparable approaches that will more specifically evaluate potential selection pressures beyond the predictions of the EICA hypothesis.

Moving forward: a framework for designing evaluations of selection pressures in post-establishment evolution
Based on our analysis, we developed a framework for future research on selection pressures potentially driving evolution. We focused our recommendations on experimental designs that reduce the potential for confounding factors and increase the ability to integrate results among studies (Table 4).

Consider abiotic and biotic conditions across ranges
Studies that examine a potential selection pressure should account for other habitat characteristics that may confound results. Ideally, the study system used for testing a post-establishment evolution hypothesis would use two ranges that are as similar as possible for all factors except for the characteristic being evaluated as a potential selection pressure. In particular, researchers can use seed collection sites at similar latitudes to partially control for photoperiod and climate (e.g., Blumenthal and Hufbauer 2007). Whether a study incorporates seed collections from separate continents or uses a smaller study area within a continent, a description of the environmental characteristics across the study system, including plant community composition, would inform comparisons across studies (e.g., Ebeling et al. 2008).

Choose study species with appropriate life-history strategies
Study species selected to test post-establishment evolution hypotheses are ideally those that have a high potential for rapid evolution. Species that reproduce primarily by seed rather than clonal growth will have a higher frequency of genetic recombination, as will species with relatively short generation times such as herbaceous perennials. As such, r-selected species may be good candidates for studying post-establishment evolution, though we do not suggest that the role of K-selected species should be ignored. In testing hypotheses related to herbivore damage, it is appropriate to consider the relative effects of specialist and generalist herbivores on a given study species (e.g., Huang et al. 2010). The ability of a species to self-fertilize may also affect the likelihood of adaptive evolution. Species that cannot self-fertilize will have higher outcrossing rates and may accumulate beneficial adaptations more rapidly than self-fertilized species. However, species that self-fertilize may exhibit stronger founder effects that could accelerate divergence from ancestral populations. Byers and Waller (1999) conducted an in-depth review of the evolutionary advantages and disadvantages of inbreeding, showing that while inbreeding populations tend to rapidly accumulate mutations, inbreeding may increase the likelihood of purging disadvantageous alleles from a population. Species that are known to hybridize with congeners may not be well suited for evolutionary studies, since the introduction of genetic material from other species may introduce novel traits that could mask the influence of selection (Roman 2006).

Consider introduction history
Documenting the introduction and historic spread of a given species can reveal the types and durations of selection pressures that the species has undergone. The possibility of multiple introductions should also be considered since repeated introductions can increase genetic variation and/or result in novel genetic admixtures not found in the native range (Maron et al. 2004). Where detailed records of species introduction are unavailable, genetic analyses can potentially determine relatedness of introduced populations. Hufbauer and Sforza (2008), for example, used chloroplast DNA to infer the history of introduction from Eurasia to North America for two species of Centaurea.

Incorporate competition in manipulations
Because individuals in nature rarely grow in isolation, tests in which individuals are exposed to actual competition are likely to be more ecologically relevant. Incorporation of multiple competitive scenarios in common garden experiments will enhance our understanding of traits that are affected by competition. For example, growing individuals along a gradient of competitive stress (measured as the number of individuals per pot) would provide more information on the ability of a given species to obtain and utilize resources. The ratio of the number of individuals in a pot to soil volume can also be kept constant to avoid confounding competition with density (Gurevitch et al. 1990). 1. Evaluate the variability in abiotic and biotic conditions across ranges 2. Choose study species with appropriate life-history traits 3. Include introduction history 4. Incorporate competition in manipulations 5. Measure multiple traits and avoid the term "competitive ability" 6. Use standard metrics where possible

Measure multiple traits and avoid the term "competitive ability"
Previous studies related to post-establishment evolution have used the term "competitive ability" as synonymous with vegetative growth, despite discussing growth in the context of energetic tradeoffs with other traits that may be affected by resource competition, including reproductive effort and herbivore tolerance (Blossey and Nötzold 1995;Bossdorf et al. 2004). A strict definition of "competitive ability" is lacking. Grime defined "competition" as the tendency of neighboring plants to use the same resources (Grime 1973), and thus "competitive ability" should be the ability to secure those resources. These resources may then be utilized for a variety of traits related to survival and fecundity, including (but not limited to) growth, reproduction, and herbivore tolerance, but the use of any one of these traits as being synonymous with "competitive ability" would be misleading. Future research could quantify variation in a broader suite of traits and refer more generally to survival and fecundity rather than using the loosely defined term "competitive ability." 6. Use standard metrics where possible.
The variety of metrics used to measure growth, reproduction, and herbivore defense has made it difficult to integrate data across studies. Utilizing standard metrics or converting units to those of standard metrics could facilitate such comparisons. Vilà and Gimeno (2006), for example, used the number of stems of Oxalis pes-caprae as a growth metric, but converted their data to units of biomass using an allometric equation. Total biomass is non-dimensional and removes bias for vertical rather than lateral growth.
In studies measuring reproductive effort, a metric that demonstrates the energetic investment in reproductive biomass relative to total biomass is useful in accounting for the role of plant size in determining the amount of fruit produced. The reproductive mass ratio (ReMR=[fruit mass][total biomass] -1 ) has been used to describe the production of reproductive structures relative to total biomass production (Abrahamson and Gadgil 1973;Bastlová and Kvet 2002;DeWalt et al. 2004).
Compensatory growth response is a good candidate as a standard measure for quantifying the effect of herbivory. While many studies have used quantified defense chemical concentrations, not all plant species utilize the same chemical pathways. As a metric of response to herbivory, compensatory growth is relevant across all species that utilize different chemical pathways and therefore facilitates comparisons across studies. Agrawal et al. (2005) used plant biomass to measure the response to herbivory for native and introduced species of Brassicaceae and found that exposure to herbivory resulted in a significant decrease in biomass production for native plants and no difference in biomass production in introduced plants.

Conclusion
Research on post-establishment evolution offers the potential for a better understanding of how nonnative plant populations interact with and adapt to their host environments. These data are relevant not only to invasion ecology, but also to studies of successional ecology and range expansions, as these fields also deal with new species interactions and novel habitat conditions (Davis et al. 2001;Simmons and Thomas 2004;Mitchell et al. 2006). The focus of previous work in this field on testing the EICA hypothesis has demonstrated that while EICA may describe evolution in some study systems, it does not fully explain the evolutionary response of nonnative plant populations to their invaded habitats. A broader examination of potential selection pressures is required to more fully understand post-establishment evolution. Careful consideration of experimental design parameters should be given to future research in this field. The lessons learned from tests of the EICA hypothesis can be used to guide future research that tests the EICA or ERCA hypotheses, or any other selection pressure that may drive post-establishment evolution.
Understanding how nonnative populations change over time is fundamental to their effective management, particularly with respect to weed risk assessments that attempt to predict the ways in which a given species might interact with a given habitat. Data on the evolutionary response of invading plant species can be used to incorporate evolutionary potential into such predictions, filling a knowledge gap that will allow researchers to predict not only the immediate impact of species invasions, but also how rapid evolutionary changes might over time alter the type or magnitude of those impacts (Whitney and Gabler 2008). Future studies that evaluate additional selection pressures will increase our understanding of how habitat characteristics drive adaptations. This information could be used to create regionally specific weed risk assessments that would prioritize management efforts and more effectively respond to current and potential threats (Daehler et al. 2004;Whitney and Gabler 2008). This enhanced understanding of how certain habitat characteristics drive rapid evolution will become increasingly relevant as trade, development, and global change alter habitat characteristics around the world.

acknowledgements
We thank Elizabeth Farnsworth, Kimberly Lellis-Dibble, and two anonymous reviewers for comments and suggestions on early drafts of this manuscript.

references
Abhilasha D, Joshi J (2009) Enhanced fitness due to higher fecundity, increased defence against a specialist and tolerance towards a generalist herbivore in an invasive annual plant. Journal of Plant abstract Urban reserves, like other protected areas, aim to preserve species richness but conservation efforts in these protected areas are complicated by high proportions of alien species. We examined which environmental factors determine alien species presence in 48 city reserves of Prague, Czech Republic. We distinguished between archaeophytes, i.e. alien species introduced since the beginning of Neolithic agriculture up to 1500 A. D., and neophytes, i.e. modern invaders introduced after that date, with the former group separately analysed for endangered archaeophytes (listed as C1 and C2 categories on national red list). Archaeophytes responded positively to the presence of arable land that was in place at the time of the reserve establishment, and to a low altitudinal range. In addition to soil properties, neophytes responded to recent human activities with the current proportion of built-up area in reserves serving as a proxy. Endangered archaeophytes, with the same affinity for past arable land as other archaeophytes, were also supported by the presence of current shrubland in the reserve. This suggests that for endangered archaeophytes it may have been difficult to adapt to changing agricultural practices, and shrublands might act as a refugium for them. Forty-six of the 155 neophytes recorded in the reserves are classified as invasive. The reserves thus harbour 67% of the 69 invasive neophytes recorded in the country, and particularly worrisome is that many of the most invasive species are shrubs and trees, a life form that is known to account for widespread invasions with high impacts. Our results thus strongly suggest that in Prague nature reserves there is a high potential for future invasions.

Introduction
Urbanization is the most dramatic form of natural habitat destruction making cities a rather hostile environment for natural wildlife. Conserving the native biodiversity in urbanized areas is therefore particularly challenging because remnants of natural habitats in urban areas are restricted to small and isolated patches. These often harbour fragmented populations of native plants and animals that face risks associated with small population sizes and pressures from heavily altered urban environments (Raupp et al. 2010. Maintaining biological biodiversity in urban landscapes has recently become a conservation priority and protecting natural remnants within cities is increasingly viewed as important (Celesti-Grapow and Blasi 2003, Turner et al. 2004, Palmer et al. 2008, Toth et al. 2009, Vermonden et al. 2009). Many cities are located in naturally species-rich areas (Kühn et al. 2004) and their heterogeneous environments (Zerbe et al. 2003) have the potential to support high numbers of species. However, overall species diversity in entire cities has been intensively studied (Klotz 1987, 1988, 1990, Pyšek 1989, Knapp et al. 2008, 2010 and it has been repeatedly documented that urban environments and the associated life styles promote introductions of alien species (Pyšek 1998, Celesti-Grapow et al. 2001, 2006, Chocholoušková and Pyšek 2003, LaSorte et al. 2007, Ricotta et al. 2009). Plant invasions are strongly dependent on propagule pressure and pathways associated with human activities that are extremely pronounced in urban areas, and also create favourable conditions for the establishment of arriving species via disturbances , Pyšek et al. 2010c, Essl et al. 2011. These factors contribute to plant species richness in cities that is higher than surrounding landscapes (Haeupler 1974), and overall alien species contribute to the remarkably high species richness of European cities (Pyšek 1993(Pyšek , 1998.
However, as in other protected areas, the conservation focus in nature reserves in urban areas is on the diversity of native species. Urban areas are where these "two diversities" come into the sharp conflict that results from the mismatch between human efforts to protect natural biodiversity and their activities that create ideal environments for alien species invasions. This matrix of urban development and nature reserves is therefore an appropriate testing ground to explore resistance patterns of natural vegetation against penetration by alien plants. It has been shown for other environments that nature reserves and protected areas possess some resistance against invasions (Pyšek et al. 2003, Foxcroft et al. 2011. This study analyses patterns of species richness of alien vascular plants in the city of Prague, Czech Republic. Based on the same data set as in Jarošík et al. (2011) where the factors shaping the richness of native, including endangered, butterflies and plants were analysed, this paper focuses on alien plants. It aims to identify the factors that contribute to the levels of reserve invasions (in the sense of Hierro et al. 2005, Chytrý et al. 2008a) and that determine the numbers of archaeophytes and neophytes currently observed in nature reserves in Prague.

Study area
The city of Prague, Czech Republic, contains 88 nature reserves in an area of 496 km 2 (Kubíková et al. 2005). Complete plant species lists were available for 48 reserves, and resulted from a systematic surveys of flora in the late 1980s/early 1990s (Špryňar and Marek 2001; Appendix 1). The reserves analysed are evenly distributed across the metropolitan area, both in its central and peripheral areas between 14°13'51 and 14°43'80"N and 49°56'33" and 50°12'23"E, and were established between 1953 and 1990. The vegetation protected in the reserves ranges from plant communities of rocky outcrops and unused quarries to thermophilous grasslands and shrublands, and seminatural woodlands (see Jarošík et al. 2011 for details). Nineteen of the reserves (mainly rocks and unused quarries) were originally established to protect geological or palaeontological sites and 15 to conserve rare plants (Kubíková et al. 2005). At present all of them represent sites protected from human influence other than tourism.

Plant data
For each reserve, the total number of vascular plant species was recorded; only naturally occurring species were considered; planted shrubs and trees were excluded. The total number of plant species in all reserves was 1309 (about a half of the Czech flora; Kubát et al. 2002), mean species number per reserve was 291, ranging from 117 to 683.
Species were classified into native and alien, with archaeophytes and neophytes distinguished among the latter group (e.g. Pyšek et al. 2002b, 2004c, Preston et al. 2004). Archaeophytes are plant species introduced to the Czech Republic since the beginning of Neolithic agriculture up to 1500 A. D., mostly from the Mediterranean region and Western Asia, while neophytes arrived after that date and represent a much more variable group in terms of the areas of origin (Pyšek et al. 2002b. The effect of the time of arrival is still detectable at present and both groups markedly differ in their ecology and habitat affinities in Central Europe (Pyšek et al. 2004b, 2008a. In total there were 175 archaeophytes and 155 neophytes in the reserves analysed (mean per reserve 14.0 and 20.3, range 4-63 and 6-104, respectively).
Since some archaeophytes appear on national red lists despite their alien origin, (Holub and Procházka 2000, Cheffings et al. 2005, Zajac et al. 2009), we recorded the number of species that are considered endangered in the Czech Republic for each reserve following the red-list classification in Kubát et al. (2002). This group (further referred to as "endangered") included species classified in categories C1 (critically threatened taxa, corresponding to the IUCN category "critically endangered") and C2 (strongly threatened taxa, corresponding to the IUCN category "endangered"; see Holub and Procházka 2000 for details). In total there were 15 endangered species in the reserves analysed (mean = 0.6, range 0-8).
The numbers of neophytes, archaeophytes and endangered archaeophytes recorded in reserves were used to calculate their proportions among total numbers of species, used in statistical analyses.

Explanatory variables
Explanatory variables describing factors that were assumed to affect the patterns of alien species occurrence in the reserves reflect geography (no. 1-7 below), habitat characteristics (8-9), substrate (10-12) and urbanization (13-16), and included: 1. reserve area (ha); 2. degree of reserve isolation (categorical: isolated, > 1 000 m from the closest reserve; clustered, < 300 m from the closest reserve; neighbouring, adjoining other reserves); 3. reserve age, expressed as years since the establishment; 4. aspect (north to north-east; plain; south-east and west; south and south-west; valley with all aspects present); 5. mid altitude, i.e., the mid value between minimum and maximum altitude; 6. altitudinal range, i.e. the difference between maximum and minimum altitude; 7. presence or absence of railway; 8. past habitat, reflecting the proportional representation of the following habitat types at the time of reserve establishment, with each type treated as an independent variable: forest, arable land, pasture, grassland, orchards, shrubland (including rocky outcrops) and built-up area; 9. present habitat, referring to the current state, using the same classification; 10. soil type (categorical variable with following levels: alluvial; acid; calcareous; neutral; acid and alluvial; acid and neutral; acid and calcareous; acid, neutral and calcareous); 11. presence or absence of bare rock; 12. presence or absence of a quarry; 13. minimum distance to natural habitat; 14. minimum distance to built-up area; 15. built-up perimeter, i.e. length of perimeter formed by built-up area; 16. natural perimeter, i.e. length of perimeter formed by other than built-up area.
The variables are the same as in Jarošík et al. (2011) where further details and means and ranges for continuous variables are given.

Statistical analysis
The response variables were proportional representations of species numbers of archaeophytes (mean 11.8%, range 2.4-21.5%), neophytes (mean 6.0%, range 2.0-17.4%) and endangered archaeophytes (mean 0.2%, range 0-1.3%) within all wild-growing species of vascular plants in each reserve. To prevent these proportions from speciespoor reserves having undue influence, the proportions were weighted by the total numbers of species in each reserve (e.g. Pyšek et al. 2010a). The response variables were then analysed as a function of the environmental characteristics of the 48 nature reserves. These analyses were made by boosted trees (Friedman 2002) in a commercial statistical software TreeNet® v.1 (Friedman 1999(Friedman , 2001. This data mining technique enables to make predictions and identify the most important predictors by screening a large number of candidate variables, without requiring any assumptions about the form of the relationships between predictors and the response variable, and without a priori formulated hypotheses (Hochachka et al. 2007). The method is more flexible than traditional statistical analyses also because it enables to reveal structures in the dataset that are other than linear, and to solve complex interactions. Importantly, the technique is nonparametric and thus not affected by spatial autocorrelations and by collinearity of the predictor variables (e.g. Jarošík 2011).
Using trees, the data are successively split along coordinate axes of the predictors, represented by the environmental characteristics, so that at any node, the split is selected that maximally distinguishes the response variable, represented by the proportional representation of the species, in the left and the right branches (Breiman et al. 1984, De'ath andFabricius 2000). This is done using binary recursive partitioning, with a best split made based on Gini impurity measure (e.g. Steinberg andColla 1995, Cutler et al. 2007). In boosted trees (Friedman 1999(Friedman , 2001, five hundred six-node classification trees were sequentially built from residual-like measures from previous trees. At each iteration, a tree was built from a default (50%) random subsample of the data set, producing a default incremental improvement in a model (0.01 learning rate at each iteration). The calculations were made with Huber-M regression loss criterion having breakdown 0.9, and minimum number of training observations in terminal nodes equal to three.
Five-fold cross-validation was used to obtain estimate of regression accuracy for each tree, and the best tree, having the smallest cross-validated mean absolute error, was chosen for interpretation. The quality of the best tree was expressed as R 2 value (Friedman 1999(Friedman , 2001. Predictors of the best tree were ranked based on improvements of all splits associated with a given variable across all trees in the model, with the raw importance scores rescaled so that the most important predictor always got a score of 100. The resulting relative importance scores thus provided a relative measure of each predictor's contribution to the model's predictive power. Partial dependence plots (Friedman 2001, Hastie et al. 2001, Cutler et al. 2007) were used to graphically characterize relationships between the individual predictor variables and predicted probabilities of species presence.
The absolute numbers of archaeophytes, neophytes and endangered archaeophytes closely correlated with their proportional representation in the species pool of each reserve (Spearman's rank: archaeophytes r s = 0.76; neophytes r s = 0.70; endangered archaeophytes r s = 0.99). However, it cannot be a priori excluded that the alien species respond very differently than native species to the predictors; if so, it may not be appropriate to weigh the proportions of alien species by the total number of wild growing species in each reserve, as it could change some of the conclusions presented.
To verify that the results on proportions are generic, all analyses were repeated using numbers of alien species as the response variable. Comparing to previous analyses on proportions, there were no changes in conclusions, and thus only the results on proportions are presented.

results
The most important factors affecting the proportion of archaeophytes among all species in a reserve were mainly the presence of arable land before the reserve was established, but also its altitudinal characteristics ( Figure 1A): archaeophytes were more abundant when the reserve had a low altitudinal range (Figure 2). The proportion of neophytes consistently increased with the proportion of present built-up area and depended on soil type: neophytes were more represented on alluvial and neutral to calcareous soils than in reserves with acidic soils (Figure 1B and 3).
As with all archaeophytes, the proportion of endangered archaeophytes among all species in a reserve positively depended on the past presence of arable land ( Figure 1C and 4A), but there was also an important effect of shrubland. The proportion of endangered archaeophytes abruptly declined in reserves with less than 30% of currently present shrubland ( Figure 4B).

Factors that determine the level of invasion of urban nature reserves
Prague nature reserves are important sanctuaries for native plants because they harbour approximately half of the native flora in the Czech Republic (Špryňar and Marek 2001, Kubát et al. 2002. Alien species in the reserves studied, on the other hand, constitute a much lower proportion of the total alien species richness in the country, but the figure differs with respect to the time of arrival: while the 175 archaeophytes represent 53% of all archaeophytes registered in the Czech Republic, the 155 neophytes found in reserves are only 15% of the 1046 neophytes (Pyšek et al. 2002b). On average, 17.8% of reserve floras are formed by alien species; this is much higher proportion than found in a larger set of 302 reserves in the whole of the Czech Republic where alien species make up on average 6.1% of the reserve flora (Pyšek et al. 2002a(Pyšek et al. , 2004a. Our study shows that the numbers of alien species in urban nature reserves can be predicted by relatively few factors. We used a number of variables that reflected site geography, land-use history and connectivity, and propagule pressure, but only five of them were needed to explain from 54 to 71% of the overall variability. That habitats were the most important factor for archaeophytes corresponds well to the recent results of studies on regional determinants of plant invasions in the Czech Republic and Europe that show habitat identity to play a decisive role, more important than propagule pressure and climate (Chytrý et al. 2008a, b, Pyšek et al. 2010b. For neophytes, the strongest effect of the proportion of built-up area reflects that it is a surrogate for propagule pressure by human activities. The effect of soil type corresponds to the well-known avoidance of acidic soils by neophytes and their affinity to resource-rich habitats , 2008b, Blumenthal et al. 2009). This is consistent with the finding of Chytrý et al. (2008a) that habitat is disproportionally a more important determinant of the level of invasion by archaeophytes than neophytes, with propagule pressure also playing an important role in the latter group.
The results also reflect that the two groups differ in long-term dynamics . Archaeophytes, prehistoric invaders to Central Europe, respond to the presence of arable land that was in place at the time of the reserve establishment; this is in accordance with the results of a previous study from Prague nature reserves that found native plants generally responding to the past factors, unlike butterflies that were more affected by current landscape settings . From the nature conservation perspective this implies that future levels of reserve invasion by archaeophytes were "imprinted" at the time of their establishment. The effect of low altitudinal range that also significantly affected the numbers of archaeophytes may also be related to habitat structure in a reserve since hilly sites were traditionally considered less suitable for agriculture in a lowland region.
Neophyte introductions, however, continue at an accelerating rate in Europe , Hulme et al. 2009), and a substantial proportion of them were introduced when most reserves were already established. Neophytes, therefore, respond to more recent ongoing human activities with the current proportion of built-up area in our study serving as a proxy for this activity. Still, the historical signal, similar to that for archaeophytes, is present and manifested by the effect of soil properties.

Endangered archaeophytes: better not lose them
Though it is is questionable whether species of alien origin should be a part of red lists, these species are perceived by botanists as elements of local nature, especially when they are rare, and many of them are typical of traditional cultural landscapes in Europe and considered to be species of cultural and historical importance (Cheffings et al. 2005). From the management perspective, this attitude is justified because rare archaeophytes are unlikely to exert any impact. Moreover, it is generally difficult to draw the line between the archaeophyte and native status, resulting sometimes in uncertain labelling (Pyšek et al. 2002b. In some cases the decision to assign a species the archaeophyte rather than native status is based on the fact that native range or habitat of the given species is unknown, and criteria applied are mostly indirect. A strong argument for protecting rare archaeophytes follows from the fact that many archaeophytes have native world ranges which are not known or are highly uncertain, and some archaeophytes are regarded as alien throughout their known global range (so-called anecophytes; Zohary 1962). As pointed by Cheffings et al. (2005), if such species were excluded from conservation efforts on account of their non-native status, it would lead to them being ignored almost everywhere and they would effectively fall through the conservation net. Endangered archaeophytes, in addition to the same affinity for past arable land as other archaeophytes, are supported by the presence of current shrubland in a reserve. This seems to indicate that there are species among this group for which it might have been difficult in the past to adapt to changing agricultural practices and that new technologies might have negatively impacted their population dynamics (Kropáč 2006). For such species, shrublands serve as refugia. This is further supported by the fact that it is current, not past shrublands that play a role. This finding points out that from the viewpoint of conservation of endangered archaeophytes, nature reserves could be more important than previously thought.

From endangered to endangering: rare archaeophytes and invasive neophytes
Our results suppport previously raised concerns about studies that analyse patterns of regional plant invasions and lump all aliens regardless of the time of immigration. It has been repeatedly shown that archaeophytes and neophytes are ecologically distinct groups that differ in habitat affinities, historical dynamics, pollination patterns and response to climate (Pyšek et al. 2004b. This is even more pronounced in nature reserves where the categories of alien species stand on opposite sides from the conservation point of view, ranging from threatened species included on red lists to potentially dangerous invaders that are targeted by management efforts. Obviously, species numbers used to infer general patterns that can lead to universal management recommendations ) are one side of the coin, but to realize the immediate threat at the local scale, species identities need to be taken into account.
On the other hand, 46 of the 155 neophytes recorded in reserves (Appendix 2) are classified as invasive (Pyšek et al. 2002b). Also, the above figure about the percentage of all neophytes recorded in the set of reserves studied among their total number in the Czech Republic, 15%, changes dramatically if expressed for a subgroup of invasive neophytes: Prague reserves harbour 67% of the 69 invasive neophytes recorded in the country (Pyšek et al. 2002b). Therefore, the potential threat to Prague nature reserves by alien plants may be in fact greater than inferred from species numbers. The list from reserves includes the majority of noxious invaders of the Czech flora, species that often exert a high impact on vegetation and species diversity of invaded communities, e.g. Heracleum mantegazzianum, Reynoutria japonica, Helianthus tuberosus, Lupinus polyphyllus, Solidago gigantea, S. canadensis ). Invasive species in Prague reserves include diverse life forms and it should be seen as warning that there are many species of shrubs and trees, a life form that is known to account for many widespread invaders, often with a high impact Pyšek 2006, Pyšek et al. 2009): Robinia pseudoacacia (recorded in 38 reserves), Symphoricarpos albus (24), Mahonia aquifolium (21), Quercus rubra (18), Syringa vulgaris (17), Lycium barbarum (10), Populus ×canadensis (4), Ailanthus altissima (3), Rhus hirta (2) and Amorpha fruticosa (1). Considering that none of the reserves was free of alien species, that the most invaded reserves harboured up to as much as 17.4% of neophytes and 32.3% of all aliens, and which species are most represented, our results strongly suggest that in Prague nature reserves there is a warning potential for future invasions unless appropriate control measures are imposed by nature conservation authorities. To make targeted practical recommendations specific to particular reserves with distinct environmental and vegetation settings, studies on current status of individual invasive species and their dynamics over time are needed, ideally initiated by state/municipal administration.

acknowledgements
We thank two anonymous reviewers for their comments, Laura Meyerson for editing our English, and Jan Pergl, Ivan Ostrý and Zuzana Sixtová for technical assistance. The study was funded by grants no. 206/09/0563 from the Czech Science Foundation, and long-term research plans no. AV0Z60050516 (from the Academy of Sciences of the Czech Republic), and no. MSM0021620828, and project no. LC06073 (both from the Ministry of Education, Youth and Sports of the Czech Republic). Petr Pyšek acknowledges support from by the Praemium Academiae award (AV0Z60050516) from the Academy of Sciences of the Czech Republic. List of invasive neophytes recorded in Prague nature reserves. Species are ranked according to the decreasing number of reserves in which they were recorded (n = 48). Plant names according to Kubát et al. (2002).

Introduction
Due to globalization, more and more species are being transported across the globe and introduced to regions where they did not occur before. Such species have taken step 1 of the invasion process (Kolar and Lodge 2001;Strayer 2005, 2006). Species that have also established one or more self-sustaining populations in the wild in their exotic range (step 2) are called exotic established species here. Those species that have additionally spread substantially from their point of introduction (step 3) are called exotic invasive species here. Although the term "invasive" is sometimes reserved for species with clear negative impacts, we are not restricting our definition of invasive species in this way. Certainly, however, many exotic invasive species do have devastating effects, e.g. on biodiversity or ecosystem services (Pimentel et al. 2005;Kettunen et al. 2009). A central question of invasion biology has been which species with which characteristics are the ones that establish themselves and become invasive. What separates such species from those that have not established populations in exotic ranges? Most studies focusing on this question investigated plants and vertebrates (reviewed in Kolar and Lodge 2001;Jeschke and Strayer 2006;, whereas only a few studies have looked at invertebrates (Mondor et al. 2007;Statzner et al. 2008;Šefrová and Laštůvka 2009). Here, we investigate this question for ants (Hymenoptera: Formicidae).
Most countries and regions of the world are now populated by numerous exotic ant species. Hawaii, for instance, has no native ants but 51 exotic established ant species (Starr et al. 2008). McGlynn (1999a) listed 147 ant species that have established themselves beyond their native range. Despite the existence of many invasive ants, most studies have focused on a few invasive ant species, e.g. the Argentine ant Linepithema humile, the big-headed ant Pheidole megacephala, or the red imported fire ant Solenopsis invicta (Holway et al. 2002). These species are also included in the Global Invasive Species Database's (2011) list of 100 of the World's Worst Invasive Alien Species. As most studies on invasive ant species have focused on a few species, quantitative studies that compare many species have been largely lacking (but see McGlynn 1999b; Lester 2005; King and Porter 2007).
Despite this lack of formal quantitative analyses, exotic established and invasive ants are often assumed to have the following characteristics (Passera 1994;McGlynn 1999b;Holway et al. 2002;Tsutsui and Suarez 2003): their colonies have (1) more reproducing queens (polygyny) and (2) more workers than the colonies of native species; (3) they form new nests more frequently via budding than native species; and their workers are (4) more frequently monomorphic, (5) smaller, and (6) more frequently sterile than the workers of native or unsuccessfully introduced species. There are only few previous studies that quantitatively tested any of these six assumptions by comparing many ant species. With respect to assumption 5 that exotic ant species have smaller workers than native species, McGlynn (1999b) compared worker head width -a standard measure of body size in ants (Wilson 1980;Hölldobler and Wilson 1990;Kaspari 1993) -between 78 exotic and 233 native ants, finding that the workers of exotic ants are smaller than their native relatives. Lester (2005) similarly found for 66 species introduced to New Zealand that smaller species are better able to establish themselves than larger species. By contrast, King and Porter (2007) found no obvious difference in body size between 94 native and 13 exotic ant species in Florida. Their results also do not support assumption 2 that exotic ant species form larger colonies than native species. In conclusion, quantitative tests of the six mentioned assumptions have been largely lacking, and the few tests that do exist had mixed results.
Using a dataset with more than 400 species, we quantitatively tested the six assumptions about the characteristics of exotic ants by comparing traits of exotic and native ant species in North America. These comparisons were done twice: once between exotic established ants and native ants, and once between exotic invasive ants and native ants. We expected that differences will be more pronounced for the latter comparison, as exotic invasive ants are the subset of exotic established ants that have successfully completed the full invasion process. If our analysis identifies traits that are related to the success of ant species in new environments, these traits should be more pronounced in the subset of invasive species.

Geographic Focus
We focused on regions that are particularly well investigated with respect to exotic ants: 14 states of the U.S. (Alabama, Arizona, Arkansas, California, Florida, Georgia, Illinois, Louisiana, Mississippi, Missouri, North Carolina, Ohio, South Carolina, and Tennessee) and one state of Mexico (Baja California). This study area has definite and naturally given borders (e.g. coastlines) and is part of the Nearctic and Neotropic bioregion, with mean annual temperatures from 10° C in northern Illinois and northern California, to 25° C in southern Baja California and southern Florida (Geodata 2011). The area includes not only a wide range of temperatures but also a diversity of biomes, thus covering a broad range of conditions.

Species List
Our species list is based on the database AntWeb (2009) which is hosted by the California Academy of Sciences and lists native as well as exotic ant species. We added further exotic species from literature sources (given in Appendix 1) and based on personal communications with other researchers (see Acknowledgements), giving a total of 416 ant species. The category Exo (exotic established ant species; n = 93 species) includes all species that were reported as exotic, introduced, alien, non-indigenous, or non-native in at least one of the above mentioned 15 states. Since the discovery and determination of an ant colony requires a certain amount of persistence of that colony, it seems likely that such exotic species have established at least one colony in the relevant state(s) and have thus completed the first two steps of the invasion process (see above). The category Inv (exotic invasive ant species; n = 42) is a sub-category of Exo and includes those species of this category that were given as exotic, introduced, alien, non-indigenous, or non-native in at least two of the above mentioned 15 states. Due to their occurrence in at least two states, it seems likely that they have spread and have thus completed all three steps of the invasion process, even if this cannot be fully revealed. Ants can spread either by human transport over short to large distances (jump dispersal) or naturally over short distances by themselves. Given that 86% of the species in this category were reported as present in directly neighboring states, many of them have probably spread naturally. Independently of whether or not the species in this category really completed the full invasion process, most of them have been more successful in their exotic range than the other species in the category Exo which were only reported to be present in one state. We thus expect stronger differences between ants of this category Inv and native ants (Nat; n = 323 species) than between all exotic (category Exo) and native ants. Our complete species list is given in Appendix 2.

Traits
In a literature search until July 2009, we collected data on traits of the 416 ant species in our species list. The sources included scientific papers, books, websites, and personal communications with researchers (Appendix 1 and Acknowledgements). Synonyms and antiquated names of species were noticed. Tools for searching were Google, Google Scholar, Google Books, and the ISI Web of Science. The complete dataset with references for all data points is provided as Appendix 3. It includes 2536 data points from over 300 different sources; data on worker head width of 414 ant species were measured ourselves. The nine specific traits we analyzed are as follows: A) Gyny -the degree of gyny, i.e. the number of reproductive queens (corresponding to assumption 1 mentioned in the Introduction; n = 226 species). We differentiated between obligate monogyny (only one functional queen; n = 103), obligate polygyny (two or more functional queens; n = 67), and facultative monogyny/polygyny (n = 56). B) Colony size -the mean colony size, defined as the average number of workers in a colony (corresponding to assumption 2; n = 227). For exotic species, it is the average number of workers in colonies in both the native and exotic range. Since this trait only relates to the workers of each species, we excluded the two parasitic species Pogonomyrmex colei (Snelling) and Anergatus atratulus (Schenk) that do not have a worker caste. C) Founding -how new nests are founded (n = 190), either by the queen alone (independent; n= 144), with the help of accompanying workers (dependent; n = 35), or a mix of these strategies (n = 11). The category "independent" includes claustral, semiclaustral, and pleometrotic founding strategies. The category "dependent" is applicable to species that found new nests via budding, splitting, sociotomy, or fission. This trait corresponds to assumption 3 in the Introduction, but to an extended version of this assumption, as only budding as a characteristic of exotic ants has been previously assumed and has thus been mentioned in the Introduction. We consequently extended assumption 3 to assumption 3a: Exotic ants form new nests more frequently in a dependent way than native ants. With respect to forming new nests, we additionally included information on social parasitism in our dataset: D) Parasitism (n = 225) -we differentiated between facultatively or obligately parasitic species (n = 35; e.g. optional slaveholders, dulotic ants, or inquilines) and non-parasitic species (n = 190). We assumed that parasites, which depend on their host species being present in the exotic range, are found less frequently among exotic than among native species (assumption 3b).
The remaining traits only concern the workers of each species. As for colony size, we again excluded the two species Pogonomyrmex colei and Anergatus atratulus that do not have a worker caste. E) Morphs -the morphology of the workers (n = 386) with the following categories: monomorphic (n = 265), dimorphic (n = 44), or polymorphic (n = 77) worker caste. This trait corresponds to assumption 4 in the Introduction. For testing assumption 5 on worker body size, we used data on head width and total body length. For simplicity, no differentiation was made between monomorphic, dimorphic, and polymorphic ant species here. F) HW (head width; n = 414) -since literature values were only available for less than half of the species in our dataset (HW1; n = 178), we measured head width for all species ourselves, using the software ImageJ (2009). These measurements (HW2; n = 414) were carried out with digital photographs, showing the frontal view of workers. Following Hölldobler and Wilson (1990), we measured maximum head width without the workers' eyes. We measured five individuals of each species (except where this was not possible due to lacking photos) and then calculated average head width. Photos were acquired from authoritative websites (AntWeb 2009;Discover Life 2009;Mississippi Entomological Museum 2009). As the data from the literature (HW1) were highly correlated with the data we measured ourselves (HW2) (r = 0.954; analysis performed for species where we had both HW1 and HW2; n = 178), we merged these two variables to HW, using the mean of HW1 and HW2 for species where both data were available. G) TL (total body length) -an alternative measure of body size in ants is total body length (n = 313). We collected these data from the literature, thereby not discriminating between different methods to measure body length, as such information was often not provided.
H) Reproduction -the reproductive ability of workers (n = 179), discriminating workers that are sterile (n = 90; without ovarioles) from those that are potentially fertile (n = 89; able to produce males, trophic eggs, or are thelytoke, i.e. produce females). This trait corresponds to assumption 6. I) Stinger (n = 388) -we discriminated workers with a functional stinger (n = 232) from those without a stinger or a rudimental or non-functional stinger (n = 156). This trait does not belong to an assumption mentioned in the Introduction. In fact, no clear assumption with respect to the frequency of a functional stinger in exotic as compared to native ant species can be found in the literature, possibly because two intuitively reasonable lines of thought lead to opposite expectations. On the one hand, a functional stinger represents a weapon that might be beneficial to survive in an exotic environment, hence one could expect that workers of exotic ants are more frequently equipped with a stinger than those of native ants. On the other hand, the stinger is a phylogenetically primary trait that has been secondarily lost in many ant species of derived clades (Hölldobler and Wilson 1990), questioning the adaptive advantage of having a stinger under at least some environmental conditions. Analyzing our data on stinger presence may help assess which of these two conflicting lines of thought is more applicable to exotic ants in North America.
For all metric literature data (colony size, HW1, and TL), we adopted means reported in the literature for a given species. If no mean but only an interval was reported (minimal and maximal limits for di-or polymorphic species; colony sizes for different colony ages), we calculated the mean by averaging the minimal and maximal value of each interval. If data for a given trait and species were available from more than one source, we calculated the mean by averaging across sources.

Analyses
Comparisons were done between native (Nat) and exotic established (Exo) ants, and between native and exotic invasive (Inv) ants. We applied univariate analyses, multivariate analyses of the raw data, and multivariate analyses of phylogenetically corrected data. For the univariate analyses, we performed two different two-sample tests for each of our nine traits. In these tests, one sample consisted of Nat species; the other sample consisted of Exo species for the first test and of Inv species for the second test. In the multivariate analyses, the independent variables were the traits, and the dependent variable was the species category: Nat/Exo for the first comparison and Nat/Inv for the second comparison. In our analyses, we followed the approach taken by Strayer (2006, 2008) and Jeschke and Kokko (2008) to combine the strengths of univariate and multivariate analyses by performing both types of analysis and interpreting them jointly. Univariate analyses have the advantage that all species can be considered for which data on a certain trait are available. In multivariate analyses, however, only those species can be considered for which data on all traits are available. As our dataset includes empty cells, multivariate analyses will have a smaller sample size than univariate analyses. On the other hand, the species being lost in the multivariate analyses will be those that are not as well investigated as species for which data on all traits are available. The data on the latter species are thus probably more reliable, and some data of species additionally included in univariate analyses might not be reliable. An additional drawback of univariate analyses is that their results have to be interpreted with caution, as spurious correlations cannot be detected, and the relative importance of different variables for explaining observed variations cannot be inferred. The latter can only be achieved with multivariate analyses. As Strayer (2006, 2008) and Jeschke and Kokko (2008), we performed two types of multivariate analyses: one with the raw data and one with phylogenetically corrected data. Details of these three com-plementary analyses -(A) univariate analyses of the raw data, (B) multivariate analyses of the raw data, and (C) multivariate analyses of phylogenetically independent contrasts -follow in the next paragraphs. Their complexity increases from A to C, while their sample size decreases. When interpreting the results, we consider all three analyses jointly. For the reasons given above, however, we put most weight on the multivariate analyses of phylogenetically independent contrasts. If not stated otherwise, statistical analyses were performed with PASW Statistics version 17.0.
A) For the univariate analyses, we ran two-tailed binomial tests for the binary variables parasitic, reproduction, and stinger; U-tests were done for the ordinal variables gyny, founding, and morphs; and t-tests for unequal variances were run for the metric variables HW, TL, and colony size. B) As mentioned above, multivariate analyses can only consider those species for which data on all included variables are available. As our dataset includes empty cells, reducing the number of variables in the multivariate analysis increases the number of species in the analysis and thus the sample size. This is one reason why it is necessary to consider a priori knowledge and thoroughly think about which variables should be included in a multivariate analysis and which should be excluded; further reasons are given in Burnham and Anderson (2002). We excluded the variable TL (total body length), as it was available for fewer species than HW (head width), the other variable quantifying body size. We also excluded the variable parasitism from the multivariate analyses, as it was uninformative in the reduced dataset of the multivariate analyses: all species in the reduced dataset were non-parasitic. The remaining seven traits were included as independent variables in the multivariate analyses: gyny, colony size, founding, HW, morphs, reproduction, and stinger. The species list decreased to n = 70 for the comparison of native and exotic established ants, and to n = 60 for the comparison of native and exotic invasive ants. For all combinations of traits, but excluding interactions, we calculated multiple logistic regressions (due to our binary dependent variables: Nat, Exo, and Inv). For each of the two comparisons, we thus calculated 2 7 -1 = 127 regression models. We evaluated the models by means of AIC c values (Akaike's information criterion corrected for small sample size, Burnham and Anderson 2002). C) As our data quantify traits of species that are phylogenetically related to each other, they are not independent of each other. To correct for this phylogenetic dependence, we calculated phylogenetically independent contrasts (Felsenstein 1985;Garland et al. 1992Garland et al. , 1999Garland et al. , 2005Pagel 1992), using Mesquite version 2.71 (Maddison and Maddison 2009) and the PDAP module version 1.14 (Midford et al. 2008). Branch lengths were set according to Pagel's (1992) method. The phylogeny was taken from the literature and is freely available from the authors upon request (sources are provided in Appendix 1). A drawback is that phylogenetic relationships among ant species are not fully resolved, which is one reason why the results of raw-data multivariate analyses should be considered as well. Another reason is that similarities and differences between both types of analyses are informative (Garland et al. 1999). Independent contrasts are not binary, so we ran linear regressions in this case that were forced through the origin, which is necessary when analyzing independent contrasts (Garland et al. 1992). The same traits and reduced datasets were used as for multivariate analyses of the raw data.

results
The above mentioned existing assumptions on differences between exotic and native ant species are supported by the univariate analyses. As expected, ant species that are exotic in North America (Exo and Inv) are more often polygynous than native (Nat) species (assumption 1; Fig. 1a; for both comparisons, Nat vs. Exo and Nat vs. Inv, p < 0.001) and form new nests more frequently in a dependent way (assumption 3a; Fig. 1b; both p < 0.001). Social parasitism is infrequent among all ant species but especially rare among exotic established (3%) and invasive (0%) species (native species: 20%; assumption 3b; Fig. 1c; both p < 0.001). Workers of exotic ants are mostly monomorphic, whereas those of native species tend to be more polymorphic (assumption 4; Fig. 1d; p < 0.001 and p < 0.05, respectively). Furthermore, head widths of exotic established (mean = 0.725 mm; SE = 0.037) and invasive species (mean = 0.699 mm; SE = 0.050) are significantly smaller than those of native species (mean = 1.051 mm; SE = 0.032) (assumption 5; Fig. 2a; both p < 0.001). As expected, the differences for total body length (TL) are similar to those of head width: workers of exotic established (mean = 2.99 mm; SE = 0.17) and invasive species (mean = 2.81 mm; SE = 0.24) have significantly shorter bodies than those of native species (mean = 4.22 mm; SE = 0.13) (both p < 0.001). Sterility is more frequent among workers of exotic established (78%) and invasive (93%) species than among natives (40%) (assumption 6; Fig. 1e; both p < 0.001). Finally, colonies of exotic established (mean ≈ 123037; SE ≈ 63591) and invasive species (mean ≈ 136777; SE ≈ 87659) are larger than those of native species (mean ≈ 6265; SE ≈ 1688) (assumption 2; Fig. 2b), but differences are not significant here. Please note that colony size is highly variable, however, ranging from five individuals to millions of workers for the species analyzed here. Thus, all assumed trait differences (assumptions 1-6) are shown by the data and are significant in the univariate analyses except for colony size which is highly variable. Still, the trend shown by the data for colony size is in the assumed direction. Also as expected, differences between exotic and native species are usually more pronounced if only exotic invasive species are compared with the native species. With respect to the presence of a functional stinger, workers of exotic established (72%) and invasive (76%) ants are significantly more frequently equipped with a functional stinger than those of native species (56%) (Fig. 1f; both p < 0.01).
Only multivariate analyses can reveal the relative importance of traits to differentiate between exotic and native ant species. In the four types of multivariate analyses that we performed, the only variable that is included in all models with substantial empirical support (Akaike weight ω i ≥ 0.05, Table 1) is colony size, suggesting that a large colony size is the most important characteristic of exotic ants as compared to native ants in North America. The mode of colony founding and the reproductive ability of workers also appear to be particularly important variables to differentiate between exotic and native ant species, as they are included in many models with substantial empirical support. The remaining variables seem less important.  Stinger absent Stinger present Figure 1. Results of univariate analyses, categorical traits. Differences among native (Nat), exotic established (Exo), and exotic invasive (Inv) ants in North America. Illustrated are means ± SE. Asterisks indicate significant differences between native and exotic established species, and between native and exotic invasive species: ***p < 0.001, **p < 0.01, *p < 0.05.

Discussion
We examined traits of exotic and native ant species in North America in order to test previously postulated but insufficiently tested assumptions on the characteristics of exotic ant species. Our results support all investigated assumptions. Thus indeed, colonies of exotic ant species have more reproducing queens (polygyny; assumption 1) and more workers (assumption 2) than colonies of native species; they form new nests more frequently in a dependent way than native species (assumption 3a); parasites table 1. Results of multivariate analyses. Shown are multivariate regression models with substantial empirical support, i.e. with Akaike weights ω i ≥ 0.05, calculated on the basis of Δ i AIC c values which are shown as well. Models with low empirical support, i.e. with Akaike weights ω i < 0.05, are not shown. Each regression model compared native ants with either exotic established or exotic invasive ants, as indicated. Model inputs were either raw data or phylogenetically independent contrasts, also as indicated. Each model included the given variables' main effects. are found less frequently among exotic than among native species (assumption 3b); and the workers of exotic species are more frequently monomorphic (assumption 4), smaller (assumption 5), and more frequently sterile than the workers of native species (assumption 6). As expected, differences between exotic invasive and native species are more pronounced and in the same direction than differences between exotic established and native species. Of the six assumptions, only assumptions 2 and 5 were, to our knowledge, previously tested (see Introduction). Assumption 2, which says that exotic ant species tend to form larger colonies than native species, has been previously tested by King and Porter (2007) who had mixed results depending on the inclusion or exclusion of the red imported fire ant Solenopsis invicta in their analysis. Our results are based on a much larger dataset and support assumption 2.

Model (variables included
Our results also support assumption 5, which says that the workers of exotic ant species are smaller than those of native species, and are in line with previous tests of this Figure 2. Results of univariate analyses, metric traits. Differences among native (Nat), exotic established (Exo), and exotic invasive (Inv) ants in North America. Asterisks indicate significant differences between native and exotic established species, and between native and exotic invasive species: ***p < 0.001 a Nat: median = 0.923 (range: 0.227 to 3.750), mean = 1.051 (SE = 0.032); Exo: median = 0.633 (range: 0.322 to 1.828), mean = 0.725 (SE = 0.037); Inv: median = 0.609 (range: 0.323 to 1.818), mean = 0.699 (SE = 0.050) b Nat: median = 296 (range: 5 to 200000), mean = 6265 (SE = 1688); Exo: median = 300 (range: 20 to 3000000), mean = 123037 (SE = 63591); Inv: median = 550 (range: 25 to 3000000), mean = 136777 (SE = 87659); differences between Nat and Exo (p = 0.071) and Nat and Inv (p= 0.146) were not significant here. Please note that the y-axis is log 10 -scaled here. assumption by McGlynn (1999b) and Lester (2005), whereas King and Porter (2007) did not find an obvious difference in body size between exotic and native ant species. As mentioned above (Methods), we did not differentiate between monomorphic, dimorphic, and polymorphic ant species when measuring body size but simply used the average of five individuals. This approach was also used by King and Porter (2007), thus it cannot explain different results between their and our study. McGlynn (1999b) only used monomorphic species to avoid this problem. Using this approach for our dataset, thus restricting the analysis to monomorphic species, does not change our finding that workers of exotic ant species are typically smaller than workers of native ant species: Nat (n = 189): median = 0.781 (range: 0.227 to 2.380), mean = 0.829 (SE = 0.030); Exo (n = 76): median = 0.599 (range: 0.322 to 1.828), mean = 0.671 (SE = 0.039); Inv (n = 34): median = 0.564 (range: 0.323 to 1.818), mean = 0.653 (SE = 0.056). Both differences, between native and exotic species (p < 0.01) and between native and invasive species (p < 0.01), also remain significant for the restricted dataset. Another approach to correct body size for polymorphism was applied by Lester (2005) who measured the smallest and largest available ant for each species. Neither approach -taking the average of measured individuals, restricting the analysis to monophoric species, or measuring the smallest and larges available ant -is perfect, and this point demands further attention in future studies. Given currently available evidence, however, it seems that the workers of exotic ant species really are often smaller than the workers of native species.
Our results show that exotic ants have a suite of characteristics that separate them from the native ant fauna of North America ( Table 2). The most important of these characteristics is, according to our multivariate analyses, colony size, followed by the founding strategy and reproductive ability of workers. Indeed, the univarate tests for differences in colony size were not significant, whereas the multivariate analyses identified this variable as the most important characteristic of exotic ants in North America. The non-significant results of the univariate analyses are less surprising when considering the huge variation in this trait, ranging from five individuals to several millions. Still, the trend shown by the univariate analyses was consistent with the multivariate analyses that exotic ant species have larger colonies than native species. Also, more weight should be put on the multivariate analyses, for the reasons given above in the Methods section. Within exotic species, there was a trend that exotic invasive species have larger colonies than exotic established species. It is reasonable that larger colonies have an advantage over smaller ones, especially when they are competing or even fighting with each other. A challenge with colony size is data availability: as the size of a colony depends on its stage (founding stage, ergonomic stage, or reproductive stage; Hölldobler and Wilson 1990), it would be desirable to compare colony size by correcting for colony stage. This was not possible for the current study due to lack of data. Another question is whether colony size of exotic species should be measured only in the native range, only in the exotic range, or in both ranges. We decided to take the average colony size in both ranges, as this measure best represents a species' overall average colony size. Other approaches are possible as well, but any approach currently suffers from the shortage of data on colony size. Once more data are available, it should be tested if colony size remains the most important characteristic of exotic ants.
In addition to colony size, the sterility of workers is an important characteristic of exotic ant species in North America, which is in line with the literature (Passera 1994;Holway et al. 2002). Species with large colonies consisting of sterile workers have a high probability to spread and thus to become invasive. Sterile workers do not invest their resources and those of their colony in their own reproduction, hence reproductive rivalries with the queen(s) or other workers are avoided (e.g. worker policing). Besides a large colony size and worker sterility, our results also show that exotic ants form new nests more frequently in a dependent way than native species, either via budding, splitting, sociotomy, or fission. Such a nest-founding strategy reduces the risk of the queen to die of starvation or predation. The queen can dedicate her resources to reproduction while the accompanying workers take care of foraging, defense, brood care, and other tasks.
Besides testing previously postulated assumptions, we also investigated if workers of exotic ant species differ from those of native ant species in the presence of a functional stinger. Our analyses revealed such a difference, showing that workers of exotic ant species are more frequently equipped with a functional stinger than workers of native ant species. Within exotic species, a functional stinger is more frequent in exotic invasive than exotic established species. These findings support the line of thought mentioned above that a functional stinger is a weapon that helps to survive in an exotic environment. Our dataset also showed a significant relationship of stinger presence with worker body size: workers of species with a functional stinger are smaller (mean head width in mm = 0.788, SE = 0.026, n = 232) than those of species without a functional stinger (mean = 1.259, SE = 0.052, n = 156) (p < 0.001, two-tailed t-test for unequal variances). Thus, species with large workers appear to often have lost their stinger in the course of evolution, probably because their large size is sufficient to be competitive. In species with small workers, on the other hand, the stinger may at least partly compensate for the size disadvantage.
A weakness of our dataset is that it does not include information on ant species that were introduced to North America but did not establish there. Even although sub-samples of ant species introduced to North America exist (Suarez et al. 2005), it is impossible to know all ant species that were introduced to North America and all that were not. We can, however, speculate to which extend observed differences between ex-table 2. Summary of the combined results. Listed is the suite of traits that characterize exotic as compared to native ants, based on this study's results.

Characteristics of exotic ants
Large colony size New nests founded by queen with workers † rather than queen alone Not socially parasitic on other ants Sterile workers Monomorphic workers More than one queen per colony (polygyny) Small body size Equipped with a functional stinger otic and native ant species are influenced by an introduction bias (Blackburn and Duncan 2001;Cassey et al. 2004;Jeschke and Strayer 2006;Blackburn and Jeschke 2009;van Kleunen et al. 2010). An introduction bias with respect to body size seems likely: Introductions of ants and other invertebrates are typically unintentional , and since it is easier for small organisms than for large ones to remain undetected by the human eye, small ant species are probably more frequently introduced to North America than large species (see Mondor et al. (2007) and Šefrová and Laštůvka (2009) for similar arguments on aphids and lepidopterans, respectively). It is thus possible that observed differences between exotic and native ant species in body size are at least partly attributable to an introduction bias. There might be introduction biases with respect to other investigated traits as well, but they are less obvious and possibly less pronounced.
Another weakness of this study is that it can only provide correlative patterns rather than causative findings. For example, our finding that exotic ants have larger colonies than native ants does not necessarily imply that they successfully established because they form larger colonies. Despite limitations of this study, it may nonetheless contribute to a better understanding of exotic ant species. Together with a few other studies, this study also sheds light on differences between native and exotic invertebrates. Combining these differences with those found for vertebrates and plants will substantially improve our understanding of the general characteristics of exotic as compared to native species.

Impact scores of invasive plants are biased by disregard of environmental co-variation and non-linearity abstract
Prioritisation of high-impact species is becoming increasingly important for management of introduced species ('neobiota') because of their growing number of which, however, only a small fraction has substantial impacts. Impact scores for prioritising species may be affected by the type of effect model used. Recent studies have shown that environmental co-variation and non-linearity may be significant for effect models of biological invasions. Here, we test for differences in impact scores between simple and complex effect models of three invasive plant species (Heracleum mantegazzianum, Lupinus polyphyllus, Rosa rugosa).
We investigated the effects of cover percentages of the invasive plants on species richness of invaded communities using both simple linear effect models ('basic models') and more complex linear or nonlinear models including environmental co-factors ('full models'). Then, we calculated impact scores for each invasive species as the average reduction of species richness predicted by basic and full effect models.
All three non-native species had negative effects on species richness, but the full effect models also indicated significant influence of habitat types. Heracleum mantegazzianum had uniform linear effects in all habitats, while effects of L. polyphyllus interacted strongly with habitat type, and R. rugosa showed a marked non-linear relationship. Impact scores were overestimated by basic effect models for H. mantegazzianum and R. rugosa due to disregard of habitat effects and non-linearity, respectively. In contrast, impact of L. polyphyllus was underestimated by the basic model that did not account for the strong interaction of invader cover and habitat type.
We conclude that simple linear models will often yield inaccurate impact scores of non-native species. Hence, effect models should consider environmental co-variation and, if necessary, non-linearity of the effects of biological invasions on native ecosystems.

Introduction
Impacts differ in quality and quantity among introduced species ('neobiota'). Many non-indigenous species can be considered neutral, but some have severe implications for native biodiversity, ecosystems, human health and economy (Gaertner et al. 2009, Goodenough 2010. Globally, roughly 1% of the introduced species have become invasive with substantial impacts (i.e. invaders sensu Davis and Thompson 2000), but this proportion varies among regions and is likely to increase with longer residence times (Williamson 1996, Richardson andPysek 2006). The human and financial resources available for management of biological invasions do not allow controlling all invasive species. Hence, it is increasingly necessary to prioritise management of neobiota according to rankings of their impacts (Parker et al. 1999, Byers et al. 2002, Thiele et al. 2010a. Different approaches have been taken to study the impact of invasive species. Regarding species richness of invaded communities, for instance, removal or seedaddition experiments have been used (Meffin et al. 2010), as well as comparisons of invaded and uninvaded sites (Adams and Engelhardt 2009, Flory and Clay 2010, Maurel et al. 2010, multi-year studies of expanding margins of invasive stands (Brewer 2008), comparisons of different stages of invasion (Prévosto et al. 2006), and correlational studies of gradients of invader abundance (Isermann et al. 2007, Gooden et al. 2009). All these approaches have their pros and cons concerning inference about invader impacts (cf. Adams andEngelhardt 2009, Meffin et al. 2010).
For impact assessment, according to the framework provided by Parker et al. (1999), and expanded by Thiele et al. (2010a), it is advantageous to relate invader effects to direct measures of invader abundance or equivalent measures, because the density of stands of a specific invasive species and the consequent impacts may vary among invaded sites, forming continuous gradients. Further, the relationship of effect and invader abundance may be non-linear which may affect impact estimates and also management decisions (Yokomizo et al. 2009).
Relationships of invader abundance and ecosystem traits may be modified by environmental factors (Houlahan andFindlay 2004, Surrette andBrewer 2008), so that one actually has to deal with a triangular relationship. A negative correlation may partly be the result of environmental conditions facilitating the invasion, while constraining the ecosystem trait. For instance, disturbance events that destroy most biomass of a given plant community may reduce species numbers, but could increase the abundance of invasive plant species. So, a negative relationship of invader abundance and species richness among sites with different disturbance regimes may be caused by environmental factors, instead of being a genuine effect of the invader. Moreover, effects of an invasive species on a particular ecosystem trait often vary among habitat types. For example, invasion of a tree species may decrease the cover of native herbaceous plants in a formerly treeless habitat, whereas it even could facilitate native species in forests, as has been shown for Cinchona spp. (Fischer et al. 2009, Jäger et al. 2009). Different or contrasting effects can be modelled as interactions of invader abundance and habitat type. Although it is important to include environmental variables into effect models of invasive species, the potential strength of the influence of environmental variation on impact scores has rarely been tested.
The aim of the present study was to assess to which extent impact scores of invasive species are affected by environmental variation, and by choice of a linear vs. non-linear effect model. We used empirical data of three invasive plant species for model building and calculation of impact scores. Richness of vascular plant species of invaded communities was chosen as response variable.

Study species and field data
We used three plant species, Heracleum mantegazzianum, Lupinus polyphyllus and Rosa rugosa, that are invasive and widespread in Europe; all three are able to form dominant stands with >90% cover (Thiele et al. 2010b).
Heracleum mantegazzianum (Apiaceae) is a monocarpic tall herb native to the Western Greater Caucasus. It invades mesic grasslands, waste ground, riverbanks, roadsides and forest edges where it forms tall-herb stands (Tiley et al. 1996. Lupinus polyphyllus (Fabaceae) is a nitrogen-fixing perennial tall herb from Pacific North America that spreads by seeds and rhizomes. In Central Europe, it mainly invades unimproved and mesic grasslands in mountain areas and transforms them into tall-herb stands, especially when grassland management is abandoned (Otte et al. 2002, Otte andMaul 2005).
Rosa rugosa (Rosaceae) is a shrub originating from the Pacific coasts of East Asia. In Europe, it mainly invades coastal habitats, but it can also become invasive on inland habitats, such as Calluna heath (Bruun 2005, Kollmann et al. 2007, Isermann 2008, Thiele et al. 2009).
We estimated cover percentages of the invasive study species and recorded all vascular plant species on plots of 16, 25 or 100 m² (see Table 1) that were established in different communities invaded by the study species. In H. mantegazzianum all plots were situated inside of invaded areas, while in L. polyphyllus and R. rugosa, sampling also included areas adjacent to invaded stands, but still inside of the same plant community. Cover percentages of the invaders ranged between zero and (almost) 100% percent. Details on the datasets are given in Table 1.
The plots were assigned to habitat types based on plant communities of Central Europe (see Ellenberg 2009) or, in few cases, based on physiognomy, land use and disturbance regime where assignment to plant communities by floristic composition was not feasible. Definitions of habitat types are given in the Supplement (Table A1).

Effect modelling
First, we calculated 'basic' effect models that only contained invader cover as predictor of species richness. For all three invaders, species counts within the plots clearly deviated from normal distributions. Thus, we used Generalized Linear Models (GLM) with (quasi-)Poisson distribution and log-link which fitted the data well.
Next, we looked for the best models of species richness considering invader cover, habitat type, squared invader cover and interactions of (squared) invader cover with habitat type, as potential predictor variables. We tested the significance of predictor variables with likelihood ratio tests, starting with the basic model ('invader cover') and adding one additional predictor variable at a time, in the aforementioned sequence. Only variables that significantly improved the model fit were included in the final effect model ('full model'). All effect models were calculated with R 2.10.1 (R Development Core Team 2009).
The models of L. polyphyllus included plot size as an additional predictor variable to account for the possibility that 100-m² plots contained more species than 25-m² plots because of species-area relationships. For R. rugosa, we calculated Generalized Linear Mixed Models (GLMM) that included a random site effect to account for nestedness of the plots. The GLMM were calculated with 'lmer' of the 'lme4' package in R.

Calculation of impact scores
We calculated impact scores by averaging predicted species loss over all sample sites using the basic and full models, thus yielding two impact scores per invasive species. Impact scores were calculated as the difference between predicted species richness at zero invader cover and predicted richness given the actual cover of the invader, averaged over all plots for each invasive species (see Equ. 1).
Equ. 1: where ŷ = predicted species richness, x i = invader cover of site i, and n = number of sample plots per invasive species.
For calculation of the predicted species richness, we used estimates of all predictor variables in the model. If habitat type was a significant predictor, we used estimates of all habitat categories in the calculations, even if the estimates of some categories did not differ significantly from zero according to the GLM t test or z test.
We calculated percentile (2.5-97.5%) confidence intervals for impact scores with bootstrapping using 10,000 resamples with replacement of same size as the original sample.

results
We found different effect models for the three invasive species (Table 2, Figure 1). Models of H. mantegazzianum and R. rugosa contained main effects of habitat types, indicating that habitat types differed in species richness. Among habitat types invaded by L. polyphyllus, i.e. ruderal and managed grasslands, species richness differed not per se, but the relationship of invader cover and species richness interacted with habitat type (Figure 1c).
The three invaders had negative effects on vascular plant species richness of invaded communities, but the shapes of the relationships differed. Heracleum mantegazzianum had uniform linear effects in all habitat types, while the effects of L. polyphyllus were much stronger in ruderal than in managed grasslands. Rosa rugosa showed a nonlinear relationship, so that species richness only was reduced at high cover percentages (Figure 1e). The details of the basic and full models (estimates etc.) are given in the Supplement (Tables A2-4).
Impact scores differed markedly between basic and full models (Figure 1). For the basic and full model of H. mantegazzianum, confidence intervals of impacts scores did not overlap, indicating that part of the apparent effect on species richness predicted by the basic model was, indeed, due to general differences in species richness between invaded habitat types. For L. polyphyllus and R. rugosa, impact scores calculated with full models were just outside the confidence intervals of the basic models, suggesting significant differences between model variants, but the confidence intervals overlapped by roughly 50%. Impact scores of L. polyphyllus tended to be higher when taking the stronger effect in ruderal grasslands into account, whereas impact scores of R. rugosa tended to be lower when considering non-linear effects and general differences in species richness between invaded habitat types.

discussion
The case studies presented here show that effect models of invasive species may take various forms. Concerning impacts on biodiversity, models should consider environmental variation, because general biodiversity trends along ecological gradients may lead us to believe that impacts are higher than they actually are. However, underestimation of impacts may happen as well.
For instance, H. mantegazzianum invades managed grasslands, ruderal (abandoned) grasslands and tall-herb stands -these habitat types resemble, and quite often are, a successional sere (Thiele and Otte 2006). Along this sere, species richness generally decreases, while average cover of H. mantegazzianum increases. Only part of the decline in species richness is attributable to H. mantegazzianum that shades out lower growing grassland plants (Thiele et al. 2010b), while another part is due to generally lower species numbers in tall-herb stands compared to the grasslands that they replace during succession (Neuhäusl and Neuhäuslova-Novotna 1985). In other words, species numbers of tall-herb stands are low even if H. mantegazzianum covers only a minor proportion of the area. In contrast, impact of L. polyphyllus was underestimated using a simple model that included no potential interactions of invader cover and habitat type. In fact, reduction of species richness was much stronger in ruderal than managed grasslands. Hence, inclusion of the interaction term increased the impact score, although ruderal grasslands represented only 25% of the plots sampled.
Several recent studies have found that inclusion of environmental variables into effect models changed predictions of invader effects on biodiversity: apparent negative effects of Lythrum salicaria L. and Rhamnus frangula L. on richness of rare native species disappeared (Houlahan and Findlay 2004), negative correlation of Lonicera japonica Thunb. with species richness became less strong (Surrette and Brewer 2008), but on the other side, effect estimates of (native) Hippophaë rhamnoides increased, i.e. became Habitat type was marginally significant in sequential likelihood ratio tests when added after L. polyphyllus cover and before squared L. polyphyllus cover and the interaction terms (p = 0.033), but the main effect of habitat type did not improve the 'full' model that contained Lupinus cover and the interaction of cover and habitat type (p = 0.833). Figure 1. Effect models and impact scores. Left panels show the 'full' effect models of Heracleum mantegazzianum (A), Lupinus polyphyllus (C) and Rosa rugosa (E) that included habitat type as a co-factor and, if significant, non-linear terms of cover percentages of the invasive species. Right panels (B, D, F) show the corresponding impact scores calculated with 'full' models and with 'basic' models that only included a linear term of invader cover. more negative, by up to 50% when including environmental co-variables (Isermann et al. 2007). Further, Spartina anglica C.E. Hubb. reduces native species richness in lowsalinity marshes, while facilitating richness in mudflats (Hacker and Dethier 2006), and differences in impact on species richness among sites (Mediterranean islands) have been found for Ailanthus altissima (Mill.) Swingle and Oxalis pes-caprae L. (Vilá et al. 2006). Hence, it is the rule rather than an exception that disregarding environmental co-variation leads to biased estimates of invader effects on species richness.
Also non-linear relationships between invader abundance and biodiversity can substantially influence impact assessments. In the case of R. rugosa, the impact score was lower when introducing a quadratic term into the effect model, while inclusion of habitat type did not change the score much (basic model, 2.8; habitat model, 2.9; quadratic model, 2.1). A non-linear increase of effect with invader abundance will commonly lead to higher impact scores compared to a linear model (Thiele et al. 2010a), but the model of R. rugosa included both a negative quadratic and a positive linear term, so that in fact the non-linear model predicted a lesser effect. Non-linear effects on species richness were also found for the shrubs Lantana camara L. (Gooden et al. 2009) and Hippophaë rhamnoides (Isermann et al. 2007). These results suggest that non-linearity of abundance-effect relationships of invasive species should be considered in future impact studies.
Further advances of impact assessment and a deeper understanding of invaderdiversity relationships could be expected from studies that take traits of the resident species into account, e.g. by modelling richness of functional plant groups or of species of high-conservation value. For the invaders studied here, it has been shown that different plant types -size classes, life forms, sociological groups -are affected differently (Thiele et al. 2010b). Regarding theory of biological invasions it would be interesting to investigate whether or not impact models could be generalised within groups of invasive species (e.g. life forms or strategy types). Another question is how to represent environmental variation in effect models. Here, we used habitat type as a co-factor which appears practical and useful, but might not be the optimal solution. The suitability of environmental properties and scaling of the variables (metric, categorical) will depend on both the invasive species and the ecosystem trait under study. Finally, multiple regression models, as applied here, account for main effects and interactions of invader and environment, but do not consider possible correlations among them. Thus, effect estimates may still be affected by spurious correlations. Further improvements could be achieved by more advanced models using, e.g., corrected path coefficients instead of regression coefficients.
Besides scientific approaches, impact assessment is grounded in normative definitions of impact or ecological damage. Not all negative relationships of invasive species with ecosystem traits necessarily have to be considered as damage, but definitions might focus on legal conservation resources or set thresholds discerning minor adverse effects from 'significant' damage (Bartz et al. 2010). In a normative sense, changes in species numbers may not directly quantify the ensuing damage (Ingo Kowarik, pers. comm.). Merging scientific and normative approaches into practical implementations of impact assessment of invasive species remains a challenge for applied invasion biology and environmental planning.

Conclusions
Simple linear effect models may often yield inaccurate impact scores of invasive species. Hence, it appears to be advisable to consider somewhat more complex models that include environmental co-variables and, if applicable, non-linear effect terms. Models should preferentially be based on samples representing the full environmental gradient of invaded habitats and the full range of equilibrium abundances or cover percentages of the invasive species.

acknowledgements
The study on Heracleum mantegazzianum by JT and AO was funded by the European Union within the project 'Giant Alien' (EVK2- CT-2001-00128  Guo (2011) points to problems arising from different approaches to estimating the proportions of floras that are native or alien, specifically those across and within various regions. This results in inconsistency of numbers reported from internal administrative units by underestimating the numbers of species that are alien to the region and overestimating native species richness. Resulting species numbers and proportions for smaller units within large countries, or whole continents, can be seriously biased if only species alien to the larger unit as a whole are considered alien, while all other species are considered native.

alien species databases for large regions need to be built bottom-up
This issue has been dealt with in detail in some invasion biology literature from other regions of the world, a fact that is not acknowledged in Guo (2011). Moreover, even for the US it is not true that previously published plant richness data for each state were always based on species either native or exotic to the entire country, as stated by Guo (2011). The issue has been thoroughly considered in some US state floras, such as California, where species with their origin in other parts of the US were consistently counted as aliens (Rejmánek et al. 1991;Rejmánek and Randall 1994). The analysis by these authors was based on detailed data from the Jepson Manual (Hickman 1993) which represents an excellent regional dataset collated through intensive historical research. However, this is not only true for California; across the US the nativity concept has been consistently applied at the regional level. Rejmánek and Randall (1994, their Table 2) give the numbers of native and naturalized alien plant species for 37 regions in North America and Hawaii, both at the state level or smaller. They derived the lists from regional or state botanical floras and carefully re-checked each species in terms of their status and origin. For North America there is good coverage of the territory generally; Palmer (2005) reports data from 1870 floras, ranging in size from local checklists to states, provinces and countries. A survey conducted to determine what data currently exists for alien species in the US identified 319 datasets. These datasets ranged from county to national and global scales, with most of them related to plants (Crall et al. 2006). Therefore the data from which native vs alien status can be correctly assessed does exist, it just needs to be searched for in appropriate sources. The above indicates that the problem may be more political/management related, rather than scientific, i.e. rather related to large databases covering the whole country and built for pratical purposes, such as USDA and NRCS, referred to by Guo (2011). Despite biases in the literature on biological invasions, resulting from varying research intensity and regional differences in the level of knowledge , for many regions with a botanical tradition the data can be found in basic botanical works. Floras and checklists of individual regions are generally more precise, for obvious reasons, than if regional information is derived from continental-scale summary floral works or checklists,where the level of inaccuracy, especially when it comes to origin of species and treatment of aliens, can be quite high (Pyšek 2003). The difference between a database of alien species built from the bottom-up and country data downscaled from a continental-wide account, can be illustrated by a comparison of European situation as reported by  and Weber (1997). Generally, the bottom-up approach of building alien species databases is scientifically more rigorous and provides more exact information than that resulting from application of nationwide or continental checklists to regional situations.

Illustrating the magnitude of bias on european data
European data (based on the DAISIE project; DAISIE 2009) provide additional and more systematic illustration of the problem pointed by Guo (2011), as well as of its magnitude. The treatment of European alien flora was based on compiling regional checklists and accurately distinguishing between alien species with their native range entirely outside the continental boundaries (termed "alien to" Europe) and alien species of European origin, including those that are native in a part of Europe but alien to another. Both subgroups together formed the "alien in" Europe category, comprising species alien across all regions Pyšek et al. 2009b). This approach, previously used also by Weber (1997) in his analysis of the continental flora based on Flora Europaea, represents a rare case in studies on biological invasions when administrative boundaries are an advantage because the history and tradition of botan-ical research relates to individual nations. Therefore, country-based research is ideally suited to reliably distinguish native and alien species at this scale.
On the continent as a whole there are 3749 naturalized (sensu Richardson et al. 2000;Blackburn et al. 2011) aliens in Europe, of which 1780 are alien to Europe. Therefore 47% of alien species are of extra-European origin . To estimate total alien species richness and its contribution to the total plant diversity in Europe, only species with extra-European origin are relevant. The 1780 naturalized aliens in this group add to 10,928 native species (as given in Winter et al. 2009), resulting in 14% of plant species in Europe being alien.
At the scale of individual countries, for example considering incorrectly only those aliens with their region of origin entirely outside of Europe would move 522 of the neophytes (alien species introduced since the beginning of the 16th century) in the Czech Republic (49.9% of the total number of neophytes) into the "native" category. This would reduce the number of neophytes recorded from 1046 to 524, and their proportional contribution to the country's entire flora from 25.3% to 12.7% (based on data in Pyšek et al. 2002). The difference between figures obtained by applying the two approaches is of similar magnitude for other countries in Europe (www.europealiens.org).

reconciling ecological and biogeographical approaches
The problem is essentially one of scale. There is no specific scale at which invasions should be studied, and the patterns and processes found by researchers differ according to the grain of the study (e.g., Hamilton et al. 2005;Pyšek and Hulme 2005). As pointed by Guo (2011), the concept of nativity/origin status depends on the size of the unit examined. It needs to be emphasized that with an increase in scale, the approach changes from an ecological one (with focus on colonization and evaluation of nativity at the level of a community or habitat), to a biogeographical one (addressing the issue of origin at the level of a region, country or continent). This reflects the differences in perceiving new species through ecological, biogeographical and anthropocentric approaches, and also in the definitions used (e.g. colonizers, invaders and weeds; Rejmánek 2000). As the scale of study decreases, the boundary between the concepts of "native" and "alien" becomes more blurred; biogeographical delimitation works with arbitrarily defined regions to which species are traditionally assigned as native or alien, and the definition of an alien is bound to introduction and dispersal resulting from human activities (Richardson et al. 2000;Blackburn et al. 2011). However, at the ecological scale, the application of the principle "the species would not be present had not been for the activities of humans" (Pyšek et al. 2004) becomes more vague. When classifying species as alien, authors generally do not consider the effect of landuse changes and global environmental change on the spread of native species through landscapes transformed by human activities (Chytrý et al. 2012). Such species would not be present in some of their locales without the influence of humans either, but are considered native to the region as a whole, mostly due to lack of information needed to make such distinctions. This can be illustrated for the British Isles; the New Atlas of the British and Irish Flora (Preston et al. 2002) is probably the only available source that distinguishes between native and alien occurrences of species that are native to the British Isles as a whole, thereby consistently applying the principles normally used at the biogeographical scale to an ecological grain of study. Using such information reveals species that are good dispersers in their native geographical range and are not confined to their native habitats, exhibiting thus potential to spread (Fig. 1).
our results are only as good as our data As pointed by Guo (2011), not distinguishing between the two groups according to origin (or ignoring the "internal" introductions) can have serious consequences for management since these species impact on ecosystems in the same way as "true exotics". Some continental-scale analyses of invasive species impacts indeed restrict the assessment to invasive species originating from overseas (e.g., Nentwig et al. 2009;Kumschick et al. 2011; but see Vilà et al. 2011) and data to separate the effects of both origin-based groups are not available. There is however one piece of evidence from the United States. Paini et al. (2010) showed that immediate present-day threat from known invasive insect pests is greater from within the United States than from the outside and advocate for invasive species policy paying increased awareness of state-level post-border biosecurity.
In terms of science, lumping both groups of origin together may obscure ecological analyses since the groups differ in many respects; the data gathered for Europe again allow for a more detailed insight. Besides the fact that biased species richness complicates comparisons between regions for which the data was collected by one approach or the other (Guo 2011), there are differences in historical dynamics and rates of introduction. Aliens of European origin for example tended to start spreading earlier in Europe than those from overseas . Similarly, comparison of the alien flora of California from different time periods shows that species of within-continental origin increased their percentage contribution over the period of 50 years from about 8 to 15% (Rejmánek and Randall 1994). European data further show that both groups differ in terms of distribution (almost all among the 15 most widely distributed species in terms of the number of region have their native ranges outside Europe), taxonomy (only a few large genera which have successfully invaded are predominantly non-European), hybridization rate (European natives hybridize more frequently) or habitat affiliation (aliens of European origin tend to occur in a wider range of habitats, most likely profitting from a better habitat match and longer residence times) Pyšek et al. 2009b).
Moreover, comparisons seeking for determinants of regional invasibility based on floras of whole states or groups of states can be misleading. This is because high environmental heterogeneity across a large area may hide the patterns that could be seen if comparing alien species richness within smaller, more homogeneous areas that are more suitable for revealing causal connections between environmental factors, anthropic disturbance, history, and numbers of alien species (Rejmánek and Randall 1994).
In invasion biology, it is becoming increasingly important to synthesize existing data on alien species abundances and distributions (Crall et al. 2006;Pyšek et al. 2009aPyšek et al. , 2010Stohlgren et al. 2011). The important issue raised by Guo (2011) points to that in invasion biology, a field currently enjoying intensive research but relying heavily, for theory, on data collected in different parts of the world by varying means and approaches, it is essential that assumptions of each study are clearly defined and stated. Our results are only as good as our data, and comparability of results drawn from different regions is a necessary step towards building a general framework and synthesis of the field.

acknowledgments
The author was supported by long-term institutional research plans no. AV0Z60050516 (Academy of Sciences of the Czech Republic), no. MSM0021620828 (Ministry of Education of the Czech Republic), and by the Praemium Academiae award from the Academy of Sciences of the Czech Republic.   Figure 1. Number of 10 ×10 km grid cells in which species native to the British Isles, that also occur in localities where their occurrence is considered alien, were mapped in 1987-1999. Species above and below the unity line are recorded more frequently as "aliens" and "natives", respectively. Based on data from Preston et al. (2002), using species with at least 100 "alien" occurrences and excluding hybrids. For species in the top left part of the plot, more than 90% of records refer to alien occurrences, those with the greatest disproportions towards the "alien" occurrences are labelled. Note the log scale.