Introduction

From climate change to the global mixing of species, research continues to document human impacts on the environment. Palumbi (2001) refers to humans as the “world’s greatest evolutionary force,” because we impose selection pressures (from increased atmospheric CO2 to prodigious use of antibiotics) affecting nearly all taxa. In addition to our direct effect on evolutionary trends, humans also introduce organisms to novel environments, often vast distances from their native range. Many of these new environments present strong abiotic selection pressures to naïve genotypes (Caño et al. 2008; Vellend et al. 2007)—inhibiting germination or preventing establishment of arriving disseminules, resulting in very few successful introductions (Williamson and Fitter 1996). Propagules that do survive and reproduce are often genetically bottlenecked and highly susceptible to genetic drift—at least in the early stages of invasion or until subsequent, genetically variant, propagules arrive (Lockwood et al. 2005).

Alternatively, novel habitats may lack strong biotic selection pressures, due to the absence of the pathogens and specialist herbivores found in the native range (Colautti et al. 2004; Mitchell and Power 2003). This ‘release’ from natural enemies in the introduced habitat may select for phenotypes allocating relatively fewer resources toward now superfluous defense structures and chemistry (Blossey and Notzold 1995). The combination of reduced herbivory and disease and exposure to a novel environment can favor (select for) phenotypes that germinate earlier, produce more aboveground biomass, grow faster, and are more fecund (Blair and Wolfe 2004; Blossey and Notzold 1995; Brown and Eckert 2005; Grosholz and Ruiz 2003; Siemann and Rogers 2003). The net effect can be a rapid evolutionary shift to a phenotype unique to the introduced range with greater competitive ability compared with individuals in the native range (Blair and Wolfe 2004; Hairston et al. 2005; Lee 2002; Vellend et al. 2007). Therefore, it is of value to understand phenotypic shifts between geographic regions.

Few studies investigating the presence of an invasive phenotype have assessed the competitive abilities of native and introduced populations (Bossdorf et al. 2005 and references therein). Therefore, it is difficult to assess whether phenotypic differences in the introduced populations actually confer enhanced competitive ability. The few studies that have examined competitive ability have unfortunately only included a ‘with’ and ‘without’ competition comparison (Blumenthal and Hufbauer 2007; Blair and Wolfe 2004; Bossdorf et al. 2004; Vilà et al. 2003, except McKenney et al. 2007), which precludes elucidation of neighbor identity effects (Bossdorf et al. 2005). Including both inter- and intra-specific competition treatments as well as a competition-free control would more fully reveal competitive mechanisms and resource allocation patterns (Bazzaz et al. 1987).

One well-established method to determine phenotypic variation for ecological traits of interest is to grow native and introduced populations of the target species in a common garden (Hendry and Kinnison 1999). With adequate representation from both ranges, phenotypic divergence can be estimated efficiently, and if divergence in these characters is observed, then it presents evidence for genetic differentiation (Bossdorf et al. 2005). Most common garden studies have focused on aboveground biomass or seed production as fitness metrics (Bossdorf et al. 2005 and references therein). Aboveground biomass and seed production may be adequate metrics of fitness for annual plant species, but many invaders also reproduce vegetatively (Pyšek 1997), with clonal expansion largely responsible for local dispersal and competitor displacement (Barney et al. 2005). In addition, belowground competition for resources (e.g., water, nutrients) can often be more intense than competition for resources aboveground (i.e., light) due to similarities in belowground allometry (Casper and Jackson 1997). Root interactions between introduced plants and their native competitors remain largely unstudied, despite the importance of belowground interactions in influencing competitive outcomes (Wilson 1988). Therefore, quantification of aboveground and belowground dynamics between introduced and native populations, especially in response to a common competitor in the introduced range, would provide valuable insight regarding the mechanisms contributing to successful invasions.

The objectives of this study were to: (1) assess the variation among native and introduced populations of Artemisia vulgaris for specific phenotypic traits related to competitive ability; (2) compare phenotypic responses between native and introduced populations under various competitive environments; and (3) determine if an endemic ‘invasive phenotype’ exists in the introduced range that is distinct from the ‘native phenotype.’ We hypothesized that introduced populations would germinate earlier, grow taller, and would produce more above and belowground biomass when compared with native populations. Additionally, we postulated that introduced populations would be more competitive (i.e., inter-specific competition would be more asymmetric than intra-specific competition) than native populations.

Methods

Plant material and experimental design

The invasive perennial A. vulgaris L. (mugwort) was used as the model system to address the above objectives. A. vulgaris has a long history of human uses in its native Eurasian range (Barney and DiTommaso 2003), and has likely been introduced in many different locations and times in North America where it is non-native (Barney 2006). A. vulgaris is a vigorous competitor along roadsides and old fields, often displacing native vegetation along a distinct invasion front (T.H. Whitlow, personal observation). Viable seeds are produced annually, but vegetative reproduction and propagation is the principal mechanism for colonizing neighboring habitats (Barney et al. 2005). We selected the North American native perennial herb Solidago canadensis L. (Solidago hereafter), as the introduced-range competitor for this study, as both species are members of the Asteraceae, have similar life history traits and environmental tolerances, and occupy similar habitats (e.g., old fields).

A. vulgaris germplasm for this study was collected by the authors, collaborators, or purchased from sources with verifiable collection locations (Table 1), and stored dry at 4°C until use. In April 2006, seeds of 15 European (EU) and 12 North American (NA) populations of A. vulgaris (Table 1) and locally collected Solidago were sown in soil-less media. Flats were maintained in a greenhouse at 26/23°C day/night temperatures with natural lighting and watered as needed. The day of emergence was recorded for each seedling. Five weeks after seeding, individual seedlings were transplanted in 15.3 cm top-diameter pots lined with weed fabric and filled with white sand (pH 6.5), and top-dressed with 6 g slow-release fertilizer Osmocote 14-14-14 (N–P-K). Individual plants were tagged so as to be able to track them through harvest. At the time of transplanting, the height (cm) of each shoot was recorded.

Table 1 Locations of European- and North American-collected A. vulgaris populations and number of replications per treatment used in this common garden study. Single A. vulgaris (A), two A. vulgaris (AA), and one A. vulgaris and one Solidago (AS) per pot treatments were used in addition to 20 replicates each of single (S) and double Solidago (SS)
Full size table

The competitive dynamics of A. vulgaris were examined using plants derived from populations from both the native (15 accessions, EU) and the introduced (12 accessions, NA) ranges. The following three treatments were imposed on all 27 accessions: single A. vulgaris (A), two A. vulgaris individuals of the same population (AA), one A. vulgaris and one Solidago (AS). Solidago treatments consisted of one (S) or two (SS) Solidago individuals, which were used to calculate competition indices for Solidago relative to A. vulgaris. The AA and AS treatments were used to determine neighbor identity effects (sensu Bossdorf et al. 2004). Pots were arranged in a completely randomized design in a 15 × 60 m2 turfgrass field (Festuca spp.). Pots were placed into pre-drilled holes spaced 1 m apart in the field so the tops of the pots were level with the soil surface. Turfgrass was mown to a height of 2.5 cm as needed to prevent interaction with treatments. Overhead irrigation of ~2.5 cm was applied when needed—usually one to two times per week.

Data collection and statistical analyses

In order to avoid genetic mixing between resident A. vulgaris populations and those used in this study, target plants were destructively harvested before flowering. Eight weeks after transplanting (13 weeks after germination), the experiment was terminated, which coincided with inflorescence initiation. The following variables were recorded for each plant in each pot: height of tallest ramet and total number of ramets. Aboveground tissue for each plant was cut at soil level and dried at 70°C until constant mass was achieved, and then biomass determined by weighing. Pots were placed at 4°C until roots could be washed (<1 week). Belowground tissue was harvested by washing soil off roots/rhizomes with water. Roots of different plants were separated easily by pulling them apart manually. Belowground tissue was dried at 70°C until constant mass was achieved, and biomass determined by weighing. Root-to-shoot biomass ratios (R/S) were calculated for each individual.

In order to determine whether the measured-dependent variables differed between populations from the native range (EU) and the introduced range (NA), a mixed-model ANOVA was used. Population origin (EU and NA), competition treatment (A, AA, AS), and the interaction between origin and treatment were considered fixed effects, while population nested within origin was considered a random effect. Pre-planned contrasts were performed between origins within competition treatments with a corrected error of α = 0.05/3(contrasts) = 0.017 for each dependent variable. All analyses were performed using the JMP v5.1 statistical software package (Cary, NC). All means presented are least square means ± 1 SE.

Competition indices

We calculated plant competition indices for each A. vulgaris origin (EU and NA) among the various competition treatments. These indices allow comparisons of the competitive strength between native and introduced A. vulgaris, as well as their relative effects on Solidago. The relative competitive performance index (Cpi) quantifies the proportional decrease in plant fitness due to competition (Keddy et al. 1998), and was calculated as:

$$ {\text{Cpi}} = \left[ {{{\left( {P_{\text{single}} - P_{\text{comp}} } \right)} \mathord{\left/ {\vphantom {{\left( {P_{\text{single}} - P_{\text{comp}} } \right)} {P_{\text{single}} }}} \right. \kern-\nulldelimiterspace} {P_{\text{single}} }}} \right] \times 100, $$

where P single is A. vulgaris performance when grown alone (A), and P comp is A. vulgaris performance when grown in (intra- or inter-specific) competition (AA or AS). If Cpi = 0, then A. vulgaris performance is unaffected by the presence of a neighbor (i.e., no competition). If Cpi > 0, then A. vulgaris performance is greater without competition, and if Cpi < 0 then A. vulgaris performance is greater with competition. To compare A. vulgaris fitness when grown in intra-specific competition (AA) to A. vulgaris grown in inter-specific competition with Solidago (AS) (i.e., does neighbor identity matter?), we calculated the relative competition index (RCI) (Jolliffe et al. 1984) as:

$$ {\text{RCI}} = \left[ {{{\left( {P_{\text{mono}} - P_{\text{mix}} } \right)} \mathord{\left/ {\vphantom {{\left( {P_{\text{mono}} - P_{\text{mix}} } \right)} {P_{\text{mono}} }}} \right. \kern-\nulldelimiterspace} {P_{\text{mono}} }}} \right] \times 100, $$

where P mono is A. vulgaris performance when grown in monoculture (AA), and P mix is A. vulgaris performance when grown in mixture (AS). If RCI = 0, then A. vulgaris performance does not differ based on neighbor identity (i.e., AA = AS). If RCI > 0, then A. vulgaris performance is greater when grown with an A. vulgaris neighbor, and if RCI < 0 then A. vulgaris performance is greater when grown in the presence of a Solidago neighbor. Cpi and RCI were calculated for each variable (height, total biomass, etc.) as the mean of individual plants within each population, followed by averaging populations across origin. Cpi and RCI values were also calculated for Solidago. Pairwise contrasts were made between origins for all competition indices except S–SS.

Results

Life history and allometric variables

Seedlings of introduced North American A. vulgaris populations emerged earlier than European populations (~4 days: F 1,25 = 18.5, P < 0.0001). However, by the end of the study, native (EU) A. vulgaris was 27% taller than introduced (NA) A. vulgaris across all treatments, although they had 61% fewer ramets per clone, and 18% less total biomass (Table 2). Averaged across origins, A. vulgaris individuals in the AA treatment had 76, 52, and 65% less aboveground, belowground, and total biomass, respectively, than A. vulgaris individuals grown alone (A), while individuals in the AS treatment had 31, 28, and 30% less aboveground, belowground, and total biomass, respectively, than A. vulgaris individuals alone (A).

Table 2 Mixed-model F-statistic values and probabilities for independent variable effects on six measured dependent variables for A. vulgaris
Full size table

A significant origin by competition treatment interaction was found for height, number of ramets, aboveground and total biomass, and R/S ratio (Table 2). Introduced A. vulgaris was shorter, but yielded more ramets than native individuals in all competitive environments (Fig. 1a, b). Aboveground biomass was greater in introduced A. vulgaris without a neighbor or with a Solidago neighbor (Fig. 1c). Belowground and total biomass were always greater in introduced than native populations, regardless of competitive environment (Fig. 1d, e). R/S ratios were higher in introduced populations in intra-specific competition and without a neighbor (Fig. 1f).

Fig. 1
figure 1

Height (a), number of ramets (b), aboveground (c), belowground (d), and total biomass (e), and R/S ratio (f) for native (EU—solid) and introduced (NA—open) populations across the competition treatments: A. vulgaris grown alone (A), two A. vulgaris plants (AA), and A. vulgaris-Solidago mixture (AS). For competition treatment comparisons between native and introduced populations, asterisks indicate significant differences at P < 0.017 based on pairwise contrasts

Full size image

Competition indices

Within a population origin, the relative Cpi was typically greater for intra-specific competition (A–AA) than for inter-specific competition (A–AS) (Table 3). Although we could not compare statistically, Solidago was more highly suppressed by an A. vulgaris neighbor, native, or introduced, than another Solidago (S–AS > S–SS). The RCI was larger (more negative) for introduced populations than for native populations in ramet number and aboveground biomass (Table 3). Solidago individuals maintained higher fitness (RCI nearer zero) with native A. vulgaris neighbors with respect to belowground biomass and R/S ratio (Table 3).

Table 3 Relative Cpi and RCI as percentages (±SE) for European (EU) and North American (NA) A. vulgaris populations and Solidago (S). Pairwise contrasts were made between origins within a treatment comparison. Origins followed by a different letter within a treatment comparison are significantly different at P < 0.05
Full size table

Discussion

Averaged across all competition treatments, introduced A. vulgaris seedlings emerged earlier, and produced more ramets and belowground and total biomass than native populations. However, introduced A. vulgaris plants were, on average, much shorter than native A. vulgaris plants, despite producing more aboveground biomass. This is a consequence of individuals from the introduced range investing in a greater number of relatively short ramets with an enhanced belowground root/rhizome network, while native populations invested in fewer, relatively tall ramets. In the absence of competition, introduced A. vulgaris individuals produced substantially more belowground and total biomass, more ramets, and had a higher R/S ratio than native populations. In the presence of a competitor, introduced A. vulgaris performed better than native populations for most ecological traits measured. Our results partially support the hypothesis of more competitive populations in the introduced range via the evolution of an ‘invasive phenotype’ (sensu Blossey and Notzold 1995).

We found only five studies examining an invasive phenotype in the published literature that included an inter-specific competition treatment (Blumenthal and Hufbauer 2007; McKenney et al. 2007; Blair and Wolfe 2004; Leger and Rice 2003; Vilà et al. 2003), and two studies that incorporated an intra-specific competition treatment (Bossdorf et al. 2004; Leger and Rice 2003). Vilà et al. (2003) found that aboveground biomass of the North American invader St. John’s wort (Hypericum perforatum L.) was reduced 90% when grown in inter-specific competition with Lolium multiflorum Lam. Another North American invader, Silene latifolia Poiret, produced 50% fewer leaves when grown in competition with a grass mixture (Festuca rubra L., Festuca arundinacea Schreb., Lolium perenne L., Cynondon dactylon L. (Pers.)) than when grown alone (Blair and Wolfe 2004). Similarly, in Chile, the invasive California poppy (Eschscholzia californica Cham.) produced less aboveground biomass and fewer flowers when grown in inter- and intra-specific competition than when grown alone (Leger and Rice 2003). Aboveground biomass and silique production of the invasive biennial herb, garlic mustard (Alliaria petiolata (M. Bieb.) Cavara and Grande) were reduced 67 and 99%, respectively, when grown in intra-specific competition, while height was unaffected (Bossdorf et al. 2004). Blumenthal and Hufbauer (2007) found that across 14 species, individuals under no competition outperformed those under low and high competition. Similarly, in our study, S. canadensis competition reduced A. vulgaris aboveground, belowground, and total biomass, and ramet number compared with A. vulgaris grown alone (Fig. 1).

Clearly, including a competition treatment in such biogeographical comparisons is critical to assess the evolution of increased competitive ability. However, only the occurrence of a significant origin (native versus introduced) by competition treatment interaction would indicate that populations from the two continents responded differently to the presence of a competitor. If introduced populations maintain more competitive phenotypes (higher fitness) relative to native populations when a competitor is present, this suggests that introduced and native populations differ in competitive ability (Blair and Wolfe 2004). Of the six studies listed above, only Leger and Rice (2003) and Blumenthal and Hufbauer (2007) reported a significant origin by competition interaction. Introduced E. californica populations produced more aboveground biomass and flowers than native populations (Leger and Rice 2003) and Blumenthal and Hufbauer (2007) reported 14 non-native species produced more aboveground biomass than their native conspecifics. Regardless of phenotypic differences, the lack of statistical origin by competitive environment interaction in A. petiolata, S. latifolia, and H. perforatum suggests no shift in competitive ability since the introduction despite phenotypic shifts (Blair and Wolfe 2004; Bossdorf et al. 2004; Vilà et al. 2003). McKenney et al. (2007) found no phenotypic differences between introduced and native populations of the perennial forb Lepidium draba whether grown with a strong or weak competitor. In our study, a significant origin by competitive environment interaction was found for height, total number of ramets, aboveground and total biomass, and R/S ratio with introduced A. vulgaris outperforming native populations in inter-specific competition with Solidago.

A comparison between phenotypes grown alone and with a competitor does not, however, adequately address whether an invasive phenotype has evolved, because we would expect individual plants to be smaller when two plants are competing for the same resources within a limited soil volume relative to a single plant (Tilman 1982). The establishment of a replacement series that includes multiple proportions of each species, or an additive design using a range of densities, would be most appropriate to assess competitive ability (Freckleton and Watkinson 2000; Jolliffe 2000). However, experimental limitations where large numbers of populations from different origins are used often preclude using these elaborate designs. Instead, in addition to an inter-specific competition treatment, we also included an intra-specific (same population) competition treatment to address the question of how neighbor identity affects the introduced and native A. vulgaris phenotypes (sensu McKenney et al. 2007). Several studies have demonstrated that neighbor identity can alter phenotypes (e.g., Reader et al. 1994), but we are most interested in the relative effects of neighbor identity. We calculated the RCI to assess neighbor identity effects, with a negative value (AA < AS) indicating that A. vulgaris is a superior competitor against Solidago than with itself. Additionally, a more positive RCI for Solidago (SS-AS) represents greater suppression by an A. vulgaris neighbor than a Solidago neighbor. Therefore, if these competition indices are greater for introduced than native A. vulgaris, we can conclude a shift toward increased competitive ability, and enhanced suppression of a common introduced range competitor, which would support the origin by competitive environment interaction found previously.

The RCI for A. vulgaris, with the exception of the R/S ratio, was negative for both native and introduced populations, indicating A. vulgaris maintains higher fitness in the presence of Solidago than another A. vulgaris. As predicted, introduced A. vulgaris RCI was greater than native A. vulgaris for the competitively important traits of ramet number and aboveground biomass, suggesting introduced populations fair much better against Solidago relative to themselves than do native A. vulgaris populations. Additionally, Solidago belowground biomass was more highly suppressed when paired with an introduced A. vulgaris individual than a native A. vulgaris individual. Our findings support the conclusion that introduced A. vulgaris individuals that invest more heavily in belowground biomass were more competitive than native A. vulgaris individuals that invest more heavily in aboveground biomass against a common introduced-range competitor of similar life history.

In addition to a shift in phenotype between geographic ranges, A. vulgaris also displays plasticity in nearly all ecological traits measured. In the parlance of Richards et al. (2006), A. vulgaris appears to be a Jack-and-master for the traits measured under the conditions imposed. A species exhibiting Jack-and-master plasticity performs best under favorable conditions (i.e., no competition in our study) and maintains lower, yet still high performance under less favorable conditions (i.e., presence of competitor). In our study, A. vulgaris always did best with no competition, while performance was lower, but varied little between inter- and intra-specific competitions. Additionally, the introduced populations outperformed native populations for all ecological traits measured (except height), which Richard et al. (2006) predict for fitness plasticity of an invasive species.

Our study period was intentionally short to elucidate competitive mechanisms and hierarchies that manifest early and are maintained (often non-linearly) throughout the life of the population, with species and/or individuals establishing early becoming dominant (Barney et al. 2005; Weiner 1990). Previous studies have demonstrated that A. vulgaris populations that establish rapidly early become more abundant in successive years (Barney et al. 2005). Shifts in phenotype and competitive hierarchies in our study occurred in the crucial stages of population establishment of a perennial species—seedling to flowering. Additionally, we were concerned with terminating the experiment before target plants flowered to preclude the creation of novel genotypes and potential escapes of this difficult to control perennial. Results should also be viewed within the context of the treatments imposed. Neighbor identity and density would likely change the magnitude of the ecological traits measured.

To the best of our knowledge, this is the first study to quantify both aboveground and belowground performances under both inter- and intra-specific competitions of an invasive species. Our results suggest that since initial introduction to North America in the early nineteenth century (Barney 2006), A. vulgaris has shifted from being largely an aboveground competitor in the native range (i.e., producing relatively few tall ramets) for being a dominant belowground competitor in the introduced range, as highlighted by the production of numerous relatively short ramets with more extensive rhizome networks. This demonstrates a shift in competitive architecture from aboveground to belowground dominance, and provides partial support for the evolution of a more competitive phenotype. Additionally, A. vulgaris possesses plasticity for ecological traits to maintain high performance in response to the presence of competitors. We did not assess the predicted decrease in defensive compounds or increased herbivory susceptibility (no variation was observed in what very little herbivory we noted) in introduced populations (Blossey and Notzold 1995). However, the more competitive invasive phenotype documented here may increase the probability of establishment, and ultimately dominance, in native ecosystems comprising species of similar life history.