Native and introduced populations of Solidago gigantea differ in shoot production but not in leaf traits or litter decomposition

Authors


†Author to whom correspondence should be addressed. E-mail: sabine.guesewell@env.ethz.ch

Summary

  • 1Invasive alien plants tend to have a greater specific leaf area and more nutrient-rich tissues than the invaded native vegetation. To test whether these traits differ between introduced and native populations of the same species, we compared 20 European (introduced) and 22 American (native) populations of Solidago gigantea Aiton (Asteraceae) in a common-garden experiment.
  • 2Five plants per population were grown for 2 years in pots and for one summer outdoors in nutrient-rich soil. We recorded shoot number and biomass, leaf production and senescence, flowering, leaf morphology and nutrient concentrations of leaves and litter. In laboratory assays, we compared litter decomposition and nutrient mineralization.
  • 3Shoot growth and leaf traits varied three- to 10-fold among the 42 populations. European plants produced, on average, more shoots than American plants, but did not differ in shoot size, leaf traits or litter decomposition.
  • 4The shoot number and total shoot biomass per plant in the experiment correlated positively with the number of new rhizomes produced by shoots of the same populations at their original field sites.
  • 5We conclude that introduced S. gigantea populations tend to produce more shoots through clonal growth than native populations. This may increase their ability to compete against the established vegetation in dense stands or at nutrient-poor sites.

Introduction

Comparative studies have revealed that invasive alien plants tend to have a greater specific leaf area (SLA) and higher nutrient concentrations in biomass and litter than the native species they displace, or than co-occurring, non-invasive alien plants (Daehler 2003; Ehrenfeld 2003; Lake & Leishman 2004). High SLA and nutrient-rich biomass are characteristic traits of plant species with a ruderal–competitive growth strategy: species that grow quickly, propagate effectively, and tend to dominate in interspecific competition (Grime 2001). It is therefore plausible that such traits contribute to making plant species invasive (Thompson, Hodgson & Rich 1995), especially in nutrient-enriched habitats (Daehler 2003; Lake & Leishman 2004). The SLA and the nutrient concentrations in biomass and litter also play a role for a species’ influence on nutrient cycling, as these traits correlate with the rate of litter decomposition and nutrient mineralization from the litter (Cornelissen & Thompson 1997; Wardle et al. 1998). Accordingly, invasive alien plants often accelerate nutrient cycling, increase nutrient availability and modify soil microbial activity at the invaded sites (Kourtev, Ehrenfeld & Häggblom 2002; Ehrenfeld 2003; Allison & Vitousek 2004).

An important question is why plant invaders have these traits: are they inherent properties of the species, or new properties that have developed in the introduced range? In many cases, contrasting functional traits of plant invaders obviously reflect inherent properties of the species. This is particularly the case when native and alien species represent different growth forms, for example when herbaceous plants invade woody vegetation (Mack & D’Antonio 2003), or when annual grasses invade perennial grasslands (Ogle, Ojima & Reiners 2004). However, there might also be differences between native and introduced populations of the same species, especially if environmental conditions place different selection pressures on functional traits in the two regions. It is commonly observed that introduced populations of a given species are more vigorous and competitive than native populations of the same species (Bossdorf et al. 2005). Functional plant traits such as SLA and nutrient concentrations may also differ between native and introduced populations. It is well known that genotypes of the same species can differ considerably in these traits as adaptive responses to variation in climatic and edaphic factors (Ryser & Aeschlimann 1999; Treseder & Vitousek 2001; Oleksyn et al. 2003). Likewise, more favourable climatic conditions, more fertile soils, or a reduced need for defence against herbivores and pathogens in the introduced range of an invasive plant species might select for genotypes with greater SLA and higher nutrient concentrations (Blossey & Nötzold 1995; Rogers & Siemann 2004). To date, this hypothesis has hardly been tested, as research has focused more on antiherbivore defence than on nutrient-related traits (Willis, Thomas & Lawton 1999; Rogers & Siemann 2004).

Solidago gigantea Aiton (Asteraceae) is a perennial forb native to North America, which has recently invaded a wide range of habitats (roadsides, forest edges, old fields, grasslands, wetlands and riversides) in central Europe (Weber & Jakobs 2005). The species occupies a wide geographical range (Weber 2001), and at least part of the phenotypic variation observed within this range is known to have a genetic basis (Weber & Schmid 1998). A field survey in the native and introduced ranges has shown that the performance of S. gigantea populations (in terms of shoot density, shoot size, reproduction and clonal growth) is on average greater in the introduced range (Jakobs, Weber & Edwards 2004). Common-garden experiments in Europe suggested that differences in average performance between European and American populations have a genetic basis, as they persisted when plants were grown in the same environment (Jakobs 2005). Furthermore, a common-garden experiment in the USA indicated that European plants are more vulnerable to pathogens and herbivores, probably due to reduced defence mechanisms (Meyer, Clare & Weber 2005).

The purpose of this study was to test whether native and introduced populations of S. gigantea also differ in functional plant traits that typically characterize invasive plant species: biomass production, SLA, nutrient concentrations, litter decomposability and nutrient mineralization from the litter. We hypothesized that measures of all these properties would be higher in European (introduced) than in American (native) populations. To test our hypothesis, we compared plant traits between 20 European and 22 American populations in a common-garden experiment.

Methods

plant cultivation

In summer 2000, 45 populations of S. gigantea in the USA and 47 populations in Central Europe were surveyed, including measurements of plant size and sampling of rhizomes (Jakobs et al. 2004). Subsequently, small rhizome segments (with two to three buds) were planted in 0·3-l pots filled with usual potting soil and kept for 2 years outdoors in a 50% shaded growth bed in Zurich (Switzerland) at 550 m a.s.l. During these 2 years, most rhizomes produced roots and one to three small shoots.

In July 2003, 20 European (EU) and 22 American (US) populations were represented by at least five living, healthy plants. Five (randomly selected) plants of each population were transplanted into the loamy soil (pH 6·0) of the experimental garden of the Geobotanical Institute ETH in Zurich at 450 m a.s.l. The experimental area had not been used for several years. Prior to the experiment, it was dug over and weeded twice.

The experimental area was subdivided into five blocks corresponding to an increasing distance from the institute building, which shaded the garden towards the end of the day. In the first block (near the building), plants were shaded approximately 2 hours earlier than in the fifth block, and the soil tended to be moister. Within blocks, plants (one from each population) were planted in rows with 50 cm distance between rows and 25 cm distance within rows, alternating European and American plants. To avoid root disturbance, plants were transferred from the pots directly into the soil. No leaf senescence or other signs of stress were observed after the transplantation.

At 1 and 2 weeks after planting, the experimental area was fertilized with a commercial complete fertilizer (Wuxal, Maag), supplying in total ≈200 mg N and 50 mg P per plant. As the soil of the garden was known to be nutrient-rich, no further fertilizer was applied subsequently.

The experiment lasted from July to October 2003, which was an exceptionally hot, sunny and dry summer in Zurich. Temperatures were above 30 °C on most days in July and August 2003, and mostly above 20 °C in September. Plants were watered daily during this time. In contrast, October was cold and wet, especially towards the end of the month, when plants were harvested.

measurements

To control for possible differences in initial size, the height of each plant was measured 2 weeks after plantation. To assess leaf production and leaf senescence, the first leaf produced after transplantation was tagged with a small ring of thin, plastic-coated wire. As a measure of leaf production rate, the number of new leaves was counted after 10 weeks, on 14 September 2003. From the end of August, plants were checked every 5 days to record the date at which all leaves present before transplantation (leaves older than the labelled one) had senesced.

Measurements on individual leaves were carried out on 1–3 October 2003, when plants were flowering. One healthy, undamaged leaf per plant was sampled on the tallest shoot at approximately two-thirds of plant height. The length, width, area and fresh mass of the leaves were measured immediately (as the weather was cold and rainy we assumed leaf tissues to be water-saturated at sampling time), and dry mass after 24 h drying at 80 °C. To determine nitrogen concentration, the leaves of two and three plants, respectively, were pooled (to obtain enough material), ground and analysed on an elemental analyser (CN-2000, Leco Corporation, St. Joseph, MI, USA).

The final harvest took place on 20 October 2003. Senesced leaves were removed and air-dried at room temperature for decomposition experiments. After recording the number of shoots and inflorescences (= flowering shoots) per plant, all shoots were clipped at the soil surface, dried for 48 h at 80 °C and weighed. Dividing this total above-ground biomass by the number of shoots yielded the mean shoot biomass.

Litter decomposability was tested in laboratory incubations starting on 28 December 2003. The air-dried litter was crushed into pieces ≈5 × 5 mm2, and 200 mg litter from each plant was incubated in plastic Petri dishes (6 cm diameter) containing ≈18 g quartz sand covered with a disc of 0·3-mm polyethylene mesh (Monodur, Verseidag-Techfab, Geldern, Germany). The sand (pH 6) was wetted with a microbial inoculum prepared by mixing ≈500 g topsoil from a ruderal garden site with 5 l water. After 1 h with repeated stirring, the soil was allowed to sediment, the supernatant was passed through a 0·2-mm sieve, and 8 ml of the suspension added to each Petri dish. Petri dishes were placed in cardboard boxes and kept in an incubator at 22 °C and 50% air moisture. To enable gas exchange, the lids were not sealed. The decomposing material was sprinkled with deionized water weekly to replace water lost through evaporation and to maintain the water level at the sand surface.

A further 150-mg subsample of the litter (if enough material was left) was analysed for C and N concentrations on the elemental analyser (CN-2000, Leco). The remaining litter was pooled per population and analysed for Kjeldahl N and P concentrations (1 h digestion at 420 °C with a titane oxide–copper sulphate catalyst; colorimetric analysis on a flow injection analyser from Tecator, Höganäs, SE).

The decomposition experiment was harvested after 4 weeks (27 January 2004), when the most rapidly decomposing litter samples had largely lost their structure. The decomposed litter was recovered from the Petri dishes, dried for 24 h at 80 °C and weighed to determine percentage mass loss. A second decomposition experiment started immediately to test whether the litter had influenced the ‘soil’ (sand) in the Petri dishes so as to affect the decomposition of other plant material. Of the five Petri dishes per S. gigantea population, three received 150 mg cellulose in the form of filter paper (to simulate extremely nutrient-poor litter), and two received 150 mg litter of the grass Molinia caerulea (9·8 mg N g−1, 0·2 mg P g−1) collected in a Molinia-dominated wet meadow (a plant community susceptible to invasion by S. gigantea). This material was removed from the Petri dishes 4 weeks later (22 February 2004), dried for 24 h at 80 °C, and weighed to determine percentage mass loss. Finally, the sand of the Petri dishes that had received cellulose (with no nutrient input since the removal of S. gigantea litter) was extracted with 100 ml water. Concentrations of nitrate, ammonium and phosphate in the water extracts were determined colorimetrically on the flow-injection analyser (Tecator). The release of mineral N and P into the Petri dish by the S. gigantea litter (minus immobilization by microbes growing on the cellulose) was expressed in µg g−1 litter, except for the calculation of correlations with cellulose mass loss, for which the absolute amounts of mineral N and P in the Petri dishes were considered more relevant.

data analysis

To test whether the measured plant variables differed between European (introduced) and American (native) populations, data were analysed with mixed-model anova with origin (EU vs US) as well as experimental blocks as fixed factors, and populations (nested within origin) as random factor. For variables that had been measured at the population level (pooling the five replicates), European and American populations were compared using one-way anova. To meet the assumptions of anova, above-ground biomass (per plant or per shoot) was log-transformed and shoot numbers were square-root transformed; the other variables could be analysed without transformation. Two variables required special tests. As the number of inflorescences was zero for many populations, the proportion of populations of which at least one plant flowered was compared between EU and US populations using a χ2 test. Leaf survival (number of days until tagged leaves had fully senesced) was analysed by comparing the survival curves of European and American plants using a Wilcoxon test.

Associations between variables were described using a matrix of pairwise Pearson correlations: these were calculated from means per population, as some variables were available only at this level. With n = 0·42, the critical value for a significant (P < 0·05) pairwise correlation is | r | = 0·301 (Zar 1996). Because many pairwise associations were considered here in an explorative sense, no significance levels are given for individual correlation coefficients. Qualitatively, correlations with | r |  0·3 were regarded as clearly insignificant, and those with 0·3 < | r |  0·5 as significant but weak. Correlations were also calculated between measures of plant growth in this experiment and at the original field sites of the populations (Jakobs et al. 2004). All analyses were carried out with the jmp statistical software, ver. 5·1 (SAS Institute, Cary, NC, USA).

Results

comparison of european (introduced) and american (native) populations

All traits investigated varied widely among populations (Fig. 1). This variability was similar for European and American populations (Fig. 1; Levene tests for unequal variances were not significant). Based on the means of the five plants per population, there was ninefold variation in above-ground biomass and shoot number (Fig. 1a,b) and sevenfold variation in biomass per shoot (Fig. 1c). Many plants, especially among the US populations, did not flower, but some European populations produced, on average, more than 1·5 inflorescences per plant (Fig. 1d). Of the investigated leaf and litter properties (Fig. 1e–l), the most variable ones were the P concentration of litter (Fig. 1j), and the amount of mineral N released by the litter during the incubation in Petri dishes (Fig. 1l).

Figure 1.

Variability of plant traits investigated among 42 European and American populations of Solidago gigantea. Data are means of five plants per population, grown in an experimental garden from rhizomes collected in the field populations.

European plants produced, on average, a greater above-ground biomass than American ones (Table 1). This was mainly due to a greater number of shoots, whereas mean shoot size did not differ. European plants also produced more inflorescences, and their leaves senesced more quickly (Table 1). None of the other variables differed significantly between European and American populations. This result did not just reflect substantial variability among populations within each range; means were actually nearly identical for the majority of variables, as shown by F values close to 0 (Table 1).

Table 1.  Comparison of shoot, leaf and litter properties of plants from 20 European and 22 American populations of Solidago gigantea grown in a common garden
VariableMean ± SEanova results (F, P)
EuropeUSARangePopulationBlock
  • For each variable, means and SE of the European and American populations were calculated from the means of five plants per population, or from measurements for the five plants pooled. Differences between the two ranges (EU vs USA) were tested with nested anova against variation among populations within ranges (random effect); the block factor accounts for heterogeneity in exposure to sunlight within the experimental garden. For variables that were determined only at the level of populations (five plants pooled), differences between ranges were tested with one-way anova (exceptions noted below). F ratios and significance levels are given (***, P < 0·001; **, P < 0·01; *, P < 0·05; no sign, P = 0·05).

  • Calculated from log-transformed data; positive standard errors are shown.

  • Differences in the proportion of populations of which at least one plant flowered were tested with a χ2 test; the likelihood ratio χ2 statistic is shown. The proportion of individual plants flowering also differed significantly (χ2 = 9·13, P < 0·01).

  • §

    χ2 statistic of a Wilcoxon text comparing the survival curves of European and American plants.

  • Determined after pooling the leaves of two to three plants; the block effect was therefore not tested.

Biomass per plant (g) 18·6 ± 2·3 10·0 ± 1·216·5***3·5*** 2·2
Biomass per shoot (g)  4·7 ± 0·5  4·0 ± 0·4 1·12·5*** 2·7*
Shoot number  4·3 ± 0·3  2·8 ± 0·313·8***2·9*** 1·6
Inflorescences per plant  0·8 ± 0·14  0·4 ± 0·13 3·9*‡
Leaves produced in 10 weeks 34·7 ± 1·6 32·1 ± 1·6 1·03·5*** 2·1
Leaf survival (days) 64·7 ± 1·1 69·1 ± 0·9 7·6**§
Leaf mass (mg)120·2 ± 6·6121·8 ± 6·5 0·03·0*** 2·0
Leaf area (cm2) 24·8 ± 1·2 24·8 ± 1·0 0·03·1*** 4·7**
Specific leaf area (cm2 g−1)212·0 ± 4·1211·9 ± 5·1 0·02·9*** 6·0***
Fresh/dry mass  3·0 ± 0·04  3·1 ± 0·05 2·62·3***11·7***
Leaf N (%)  3·8 ± 0·1  3·7 ± 0·1 0·61·9*
Leaf C (%) 49·0 ± 0·1 48·8 ± 0·2 0·72·5**
Litter N (%)  1·8 ± 0·07  1·7 ± 0·06 0·04·0*** 8·8***
Litter P (mg/g)  3·8 ± 0·3  3·4 ± 0·3 0·8
Litter mass loss (%) 43·5 ± 1·2 44·1 ± 1·6 0·12·7*** 3·8**
N mineralization (µg N g−1 litter) 79·8 ± 62·5 77·8 ± 67·7 0·01·6* 2·7*
P mineralization (µg P g−1 litter) 98·4 ± 12·5104·9 ± 13·6 0·81·3 3·4*
Cellulose mass loss (%)  5·9 ± 1·8  5·4 ± 1·9 0·33·3*** 1·7
Molinia litter mass loss (%) 23·2 ± 1·2 23·6 ± 1·2 0·40·9 1·7

Differences in exposure to sunshine, as represented by the five experimental blocks, had only a marginal influence on the plants’ above-ground biomass (P = 0·06) but significantly affected some of the leaf traits: plants grown near the institute building (partly shaded) had a greater fresh : dry mass ratio, SLA and litter N concentration, and faster litter decomposition than plants fully exposed to sunshine (Table 1). The block effect was similar for European and American populations (no significant block × range interaction) for all traits except litter N concentration. For this trait, European plants responded more than American ones to differences in light conditions (Fig. 2).

Figure 2.

Nitrogen concentration in the litter of plants from European and American populations of Solidago gigantea in relation to their position in the experimental garden; increasing distance from the institute building corresponds to increasing daily duration of exposure to sunshine.

correlations between plant growth, leaf properties and decomposition

The mean above-ground biomass of the 42 S. gigantea populations correlated positively with their mean shoot number (r = 0·64; Fig. 3a), shoot size (r = 0·66), leaf production (r = 0·65; Fig. 3b) and number of inflorescences (r = 0·51). In contrast, pairwise correlations between these measures of plant growth and leaf properties or measures of decomposition were non-significant, with | r | < 0·3. The various leaf traits also hardly correlated with each other, except for SLA and leaf mass (r = −0·60) as well as leaf N and litter N concentrations (r = 0·59). In particular, there was no correlation between the SLA and N concentration of leaves (r = 0·22), nor between the N and P concentrations of the litter (r = −0·05).

Figure 3.

Relationships between mean above-ground biomass of plants from 42 Solidago gigantea European and American populations grown in a common garden and (a) mean shoot number; (b) mean number of leaves produced per shoot during 10 weeks (July–September 2003). Data are means of five plants per population. Closed symbols, European (introduced) populations; open symbols, American (native) populations.

Litter decomposition depended on properties of leaves and litter, although rather weakly: the mass loss of S. gigantea litter mainly correlated with the litter C : N ratio (r = −0·50; Fig. 4a), litter N concentration (r = 0·41), leaf water content (r = 0·38), and SLA (r = 0·31); it also correlated with the amount of mineral N (nitrate + ammonium) released into the Petri dish during the incubation (r = 0·44), but not with the P concentration of the litter nor with the release of phosphate. The mass loss of cellulose incubated in the Petri dishes after removing the S. gigantea litter correlated strongly with the C : N ratio of the S. gigantea litter (r = −0·72; Fig. 4b) and its mass loss (r = 0·63), and more weakly with the amounts of nitrate (r = 0·47) and phosphate (r = 0·35) released into the Petri dishes by the S. gigantea litter. The mass loss of M. caerulea litter varied little, and did not correlate with the mass loss of S. gigantea litter or any of the litter properties.

Figure 4.

Relationships between the mean C : N ratio of litter produced by plants from 42 European and American populations of Solidago gigantea in a common garden and (a) mean mass loss of litter during 4 weeks’ incubation in Petri dishes at 22 °C; (b) mean mass loss of cellulose incubated for 4 weeks in the same Petri dishes after removing S. gigantea litter. Closed symbols, European (introduced) populations; open symbols, American (native) populations.

relationships between plant growth in the common garden and in the field

The number of shoots per plant in this experiment correlated positively with the number of new rhizomes produced per shoot in the same S. gigantea populations at their original field sites, and more weakly with the shoot density, mean shoot height, inflorescence length and rhizome length of these populations (Table 2). Similar correlations were found for the above-ground biomass of plants in this experiment (not shown). The mean biomass per shoot in this experiment correlated weakly with the mean shoot height and inflorescence length at the original field sites, but did not correlate with shoot density, rhizome number or rhizome size in the field (Table 2).

Table 2.  Correlations between mean number of shoots per plant and mean biomass per shoot of the 42 Solidago gigantea populations in the experimental garden and measurements of shoot size, reproduction and vegetative propagation of the same populations at their original field sites (data from Jakobs et al. 2004)
Field sitesCommon garden
Shoot numberShoot mass
  1. The critical value for a significant (P < 0·05) pairwise correlation is | r | = 0·301 (Zar 1996).

Shoot density  (number m−2)0·34−0·00
Shoot height in  late summer (cm)0·35 0·34
Inflorescence length0·43 0·39
New rhizomes per shoot0·58 0·03
Rhizome length0·41 0·05
Rhizome diameter0·18 0·08

Discussion

differences in growth between native and introduced populations

In our experiment, plants from introduced European populations of S. gigantea produced more shoots, more inflorescences and greater above-ground biomass than plants from native American populations. The same difference was found between EU and US plants at their original field sites, and the shoot number and above-ground biomass of plants in our experiment correlated with the shoot density and mean shoot size of the original populations (Jakobs et al. 2004). This correlation is unlikely to reflect a ‘maternal’ effect (due to rhizomes from vigorous field populations containing more reserves), because the rhizome pieces transplanted in 2001 represented at most 5% of the final biomass of plants in our experiment and had been precultivated in pots for 2 years before the experiment. When the small plants grown from these rhizomes (one to two shoots, mostly 20–30 cm high) were transplanted into the experimental garden, shoot height did not differ between EU and US plants (F = 0·54, P = 0·47) and correlated only weakly with the final biomass of the plants (r = 0·36, P = 0·02). Initial size strongly influences the growth of competing plants (Grace, Keough & Guntenspergen 1992), but in this experiment competition between neighbouring plants was likely to be negligible, as the root and rhizome systems did not overlap at the end of the experiment, and the final density of shoots was much lower than in natural populations with a similar shoot size. The difference in shoot production found between EU and US plants therefore probably reflected genetically determined differences, consistent with the results of a common-garden experiment using plants grown from seeds (Jakobs 2005).

The greater above-ground biomass of EU plants in this experiment was almost entirely due to higher shoot number, not to larger shoots. Given the late start and rather short duration of the growth period in the garden (July–October), we believe differences in shoot number reflect differences in the number and activity of rhizome buds formed by the plants during precultivation in pots. In the field survey, Jakobs et al. (2004) found that European populations produced, on average, 2·5 times more rhizomes per shoot than American ones. Assuming that EU plants also produced, on average, more rhizomes per shoot during precultivation, this would be sufficient to explain the higher shoot number (and thus greater above-ground biomass) produced by EU plants during the experiment. In the previous common-garden experiment (Jakobs 2005), which lasted for two growing seasons, European and American plants differed mainly in biomass in the second year, when shoots developed from the rhizomes produced in the first year.

The greater shoot number of EU plants contrasts with results of a similar experiment comparing EU and US plants of the congeneric Solidago canadensis, a species that is almost as invasive as S. gigantea in Europe (Weber 2001): in this case, EU plants produced smaller shoots, smaller inflorescences and fewer vegetative offspring than US plants (van Kleunen & Schmid 2003). The contrasting results obtained for the two Solidago species appear consistent with field observations. In American old fields (Wisconsin), S. gigantea invested more biomass into inflorescences and less into below-ground parts (roots and rhizomes) than S. canadensis (Abrahamson et al. 2005). In Europe, however, S. gigantea tends to produce smaller inflorescences but more rhizomes with more buds than S. canadensis (Weber & Schmid 1998; Weber & Jakobs 2005). Thus an increased allocation to clonal growth, leading to greater shoot densities in field populations, may well characterize EU plants of S. gigantea but not those of S. canadensis. As S. canadensis generally invades drier and more disturbed sites than S. gigantea, this difference would be in line with a more general pattern that clonal growth favours the invasion of moist sites, whereas ruderal traits favour the colonization of drier, more open habitats (Thompson et al. 1995).

no difference in leaf traits and decomposition

As many studies have documented positive correlations between SLA, nutrient concentrations in leaves, net C assimilation rate, and the growth rate of plants (Reich et al. 1999; Shipley et al. 2005), our initial hypothesis that EU plants would grow more quickly than US plants caused us to predict that EU plants should have a greater SLA and higher nutrient concentrations. The absence of difference between EU and US plants in all these traits is consistent with the absence of difference in mean shoot size and leaf production, which roughly reflected the growth rate of individual shoots in this experiment. The exceptional climatic conditions during summer 2003 may have influenced this result, as summer drought strongly affects the leaf-area development of S. gigantea (Weber & Jakobs 2005). We found that the litter of EU plants tended to be more N-rich than that of US plants under the slightly shaded, moister conditions near the building, whereas the trend was opposite in full sunlight (Fig. 2), suggesting the possibility of different leaf-level responses to climatic conditions. On the other hand, recent field studies showed that in Europe, S. gigantea does not differ from native plants in SLA, and even tends to have lower nutrient concentrations when growing at rather nutrient-poor sites (Güsewell, Zuberbühler & Clerc 2005; Chapuis-Lardy et al. 2006). It therefore appears that the invasiveness of European S. gigantea populations is not, or is not consistently, related to the leaf properties often found for invasive alien plants (Daehler 2003; Ehrenfeld 2003; Lake & Leishman 2004).

To test whether native and introduced populations might affect soil processes differently when they colonize a site, we compared the decomposition of S. gigantea, its release of nutrients and the subsequent decomposition of cellulose or M. caerulea litter. The absence of difference between EU and US plants was consistent with the absence of difference in leaf traits, as decomposition depended on the SLA and on the C : N ratio of the litter, as is usually the case in herbaceous plants (Cornelissen & Thompson 1997; Wardle et al. 1998; Schädler et al. 2003). Thus our experiment did not support the hypothesis that EU plants tend to accelerate nutrient cycling more than US plants because of leaves that are more nutrient-rich (Ehrenfeld 2003) or less well defended (Rogers & Siemann 2004).

This negative result is, however, not sufficient to exclude a different impact of EU and US plants on soil processes. Only leaf traits were studied here, while there could also be differences in nutrient cycling from stems, rhizomes or roots. A study in Belgium revealed an increased availability of phosphorus in soils under dense stands of S. gigantea, although P concentrations in plant shoots did not differ between S. gigantea and native vegetation (Chapuis-Lardy et al. 2006). In that case, the greater biomass production of S. gigantea caused a greater input of organic matter to the soil, and this apparently stimulated microbial activity (Chapuis-Lardy et al. 2006). Similar effects have been reported from other species invading herbaceous plant communities, such as Calamagrostis epigeios in nutrient-poor grasslands of Central Europe (Fiala et al. 2003). As our EU plants produced more biomass than US plants without any difference in nutrient concentrations, they must have taken up more N and P from the soil, and would also return more N and P to the soil, promoting increased nutrient mineralization.

large variation among individual populations

For both EU and US plants, the individual populations differed remarkably in leaf traits, litter decomposition and nutrient release (Fig. 1), as well as in their influence on the subsequent decomposition of cellulose, suggesting that the impact of S. gigantea on soil processes might vary substantially among populations. Leaf traits also vary substantially under field conditions: in a preliminary study of 18 field-grown S. gigantea populations in Switzerland, the N and P concentrations of leaves and litter varied threefold among populations (S.G., unpublished data). The large variation among populations found in the present experiment may reflect the wide range of climatic and edaphic conditions under which plants were sampled, and associated differences in selective pressures (Thompson et al. 1995; Weber & Schmid 1998; Jakobs et al. 2004). A practical implication of this variation is that the invasion potential of newly established S. gigantea stands may depend on the origin of the propagules. Decisions regarding the need for control measures (Smith, Londsdale & Fortune 1999) might have to take this variability into account.

Although all measured variables differed significantly among individual populations, there was no correlation between shoot growth and leaf traits. Various reasons may have contributed to the absence of such a relationship. First, SLA and N concentration did not correlate significantly with each other, while a positive correlation is needed to maximize their effect on C assimilation, and thus on growth (Shipley et al. 2005). Second, high SLA and N concentrations are generally associated with a reduced leaf life span (Reich et al. 1999), and tend to make leaves more vulnerable to herbivores and pathogens. In this experiment, the N concentration of leaves and litter correlated negatively with the duration until the tagged leaves had senesced. Third, high SLA and N concentrations can make leaves more vulnerable to herbivory and pathogens (Schädler et al. 2003). We did not observe any sign of herbivory, but most plants were infected by mildew at the end of the summer. A further garden experiment revealed a significant positive relationship between the N concentration of leaves and the severity of powdery mildew infection in S. gigantea (S.G., unpublished data). Even if each of these effects was weak in our experiment, they may have been sufficient in combination to offset any positive effect of high SLA and N concentrations on shoot growth.

a test of the eica hypothesis?

A controversial hypothesis to explain the invasiveness of certain alien plant species is the evolution of increased competitive ability (EICA) hypothesis (Blossey & Nötzold 1995; Hinz & Schwarzlaender 2004; Bossdorf et al. 2005). Recent evidence has both supported (Vilà, Gómez & Maron 2003; Rogers & Siemann 2004; Meyer et al. 2005) and contradicted (Willis, Memmott & Forrester 2000; van Kleunen & Schmid 2003; Bossdorf et al. 2004) this hypothesis. Dietz & Edwards (2006) recommended distinguishing between the colonization of sites with favourable conditions and weak competition, such as anthropogenically disturbed sites (‘primary invasion’), and the invasion of sites with stressful abiotic conditions or strong competition from the native vegetation (‘secondary invasion’). Plant traits associated with fast growth and reproduction, such as high SLA and high nutrient concentrations, should promote the success of invaders mainly during primary invasion, whereas clonal growth and stress tolerance would be most important during secondary invasion. Solidago gigantea is notoriously able to penetrate established vegetation and colonize nutrient-poor, shaded or wet sites (Weber & Jakobs 2005), and therefore represents an example of secondary invasion (Dietz & Edwards 2006). Our finding that the invasive European populations were mainly characterized by the production of more shoots through clonal growth, but not by greater SLA or nutrient concentrations, fits well into the framework suggested by Dietz & Edwards (2006). However, as discussed above, the difference found here (together with the large variation among populations of each continent) might reflect the adaptation to different climatic and edaphic conditions rather than a specific evolutionary process related to enemy release in Europe (Thompson et al. 1995).

Our results suggest that the outcome of experiments testing the EICA hypothesis may depend on their duration. In our study, EU and US plants did not differ in shoot size or leaf traits, but still differed in biomass production, probably because EU plants produced more shoots through clonal growth during the 2-year precultivation. This difference would have been missed in a short-term growth experiment with plants grown from seeds. On the other hand, we speculated that the exceptionally hot weather conditions during our experiment possibly suppressed differences in shoot size or leaf traits, so that the hypothesized differences might have been found in the following, cooler summer. Further traits that were not investigated here, such as more efficient nutrient resorption from senescing leaves, or more effective nutrient storage during the winter, might improve the nutrient status of European plants over a longer period. Thus the ability to produce more shoots through clonal growth under unfavourable conditions (shade, low nutrient supply) appeared to be the main advantage of introduced over native populations of S. gigantea in our garden experiment, but further traits may contribute to making European S. gigantea populations invasive under field conditions.

Acknowledgements

We thank Irene Handke for help with the experiment, Rose Trachsler for nutrient analyses, and Peter J. Edwards as well as two referees for helpful comments on a draft of the manuscript.

Ancillary