A comparison of phenotypic plasticity in the native dandelion Taraxacum ceratophorum and its invasive congener T. officinale

Authors

  • Marcus T. Brock,

    Corresponding author
    1. Division of Biological Sciences, University of Missouri, Columbia, Missouri 65211, USA;
    2. Present address: Department of Plant Biology, University of Minnesota
      Author for correspondence: Marcus T. Brock Tel: +1 612 624 3124 Fax: +1 612 625 1738Email: brockmt@umn.edu
    Search for more papers by this author
  • Cynthia Weinig,

    1. Department of Plant Biology, University of Minnesota, St Paul, Minnesota 55108, USA;
    Search for more papers by this author
  • Candace Galen

    1. Division of Biological Sciences, University of Missouri, Columbia, Missouri 65211, USA;
    Search for more papers by this author

Author for correspondence: Marcus T. Brock Tel: +1 612 624 3124 Fax: +1 612 625 1738Email: brockmt@umn.edu

Summary

  • We compared plastic responses to variation in the light environment for sympatric populations of native and exotic dandelion species, Taraxacum ceratophorum and Taraxacum officinale.
  • Plasticity in leaf size, inflorescence height, reproductive phenology and dispersal-related traits were measured under experimentally altered light quality (red : far-red light ratio, R : FR) and light intensity (photosynthetically active radiation, PAR). To test whether differences in means and reaction norms of dispersal-related traits between species affected colonization potential, we created seed-dispersal models based on seed-fall rate and release height.
  • Differences in plasticity between species were not systematic, but varied in direction and magnitude among traits. Taraxacum officinale produced larger leaves that exhibited greater plasticity in size under variable light intensity than T. ceratophorum. Plasticity in scape length at flowering occurred in relation to R : FR ratio in both species, but tended to be greater in T. ceratophorum. Seed-bearing scapes of T. officinale were taller and more canalized in height across light regimes than scapes of T. ceratophorum. Seeds of T. officinale were smaller than seeds of T. ceratophorum.
  • Models predict greater dispersal in T. officinale within open and vegetated habitats. In contrast to the idea that plasticity promotes invasiveness, results suggest that the lack of plasticity in dispersal-related traits enhances the colonization potential of T. officinale.

Introduction

Plastic ‘all-purpose genotypes’, especially in sedentary organisms, may better tolerate environmental variation encountered during a lifetime, or between generations, through seed dispersal. For this reason, phenotypic plasticity is fundamental to ecological breadth (Sultan, 1995). Species (or genotypes) that are more plastic are hypothesized to grow and reproduce under a wider range of environments than more canalized genotypes. Recognizing this idea, Baker (1965, 1974) viewed plasticity as a major trait characterizing agricultural weeds. He postulated that a ‘jack-of-all-trades’ genotype capable of maintaining fitness should be more tolerant of norms of environmental variation (disturbance) found in agricultural fields. This view of agricultural weeds has also influenced attempts to characterize invasive species. Although previous attempts at classification have been largely unsuccessful (Mack, 1996; Williamson, 1999; but see Rejmanek & Richardson, 1996), predictive models based on biological characteristics may help determine which species are likely to become invasive (Rejmanek, 2000). Local adaptation in populations of invasive species provides a less satisfactory explanation of broad ecological range than plasticity (Sultan, 1995; Parker et al., 2003), because nonindigenous species often go through bottlenecks during introduction that reduce genetic variation (Novak et al., 1991; Sakai et al., 2001; Tsutsui & Case, 2001). A high degree of plasticity may enable genetically depauparate, exotic populations to maintain fitness across environments and facilitate colonization of unoccupied habitats (Novak et al., 1991; Williams et al., 1995; Rejmanek, 2000; Parker et al., 2003). Using native and exotic dandelions (Taraxacum), we compare the plasticity of a suite of traits across a range of light environments in the field. We then incorporate experimental results into simple models of seed dispersal to assess the contribution of trait plasticity to colonization potential in Taraxacum.

Plants can respond to variation in red : far-red (R : FR) light ratio (foliar shade) through the shade-avoidance syndrome: a set of plastic responses that includes increased internode length, reduced branching, reduced investment in root tissue, and faster time to reproduction under foliar shade (Smith, 1982; Smith & Whitelam, 1997). Shade avoidance is mediated by the phytochrome family of photoreceptors and allows plants to anticipate increasingly crowded conditions (Smith, 1982; Ballaréet al., 1997). Shade avoidance is triggered by a reduction in the ratio of red to far-red wavelengths (R : FR) resulting from the selective absorption of red wavelengths of light by neighboring and/or overtopping leaves (Ballaréet al., 1987, 1990). Studies of transgenic lines lacking functional phytochrome genes (Schmitt et al., 1995) as well as experimental manipulations of light quality (Dudley & Schmitt, 1996; Schmitt & Dudley, 1996) have demonstrated that internode elongation increases relative fitness by enhancing light interception under crowded conditions. Along with vegetative stems, the leaves and reproductive stems (e.g. peduncles and scapes) of various species respond to foliar shade (Young & Schmitt, 1995; van Hinsberg & van Tienderen, 1997; de Kovel & de Jong, 1999). For wind-pollinated species, modification of flower height via scape elongation could increase fitness by reducing obstruction from vegetative organs during pollen dispersal and receipt (Young & Schmitt, 1995). In animal-pollinated species, too, flower height can influence pollinator visitation rate, interspecific pollen transfer and resulting mating patterns (Levin & Kerster, 1973; Waddington, 1979). Greater seed release height should enhance dispersal of wind-dispersed seeds (Pasquill & Smith, 1983; Greene & Johnson, 1989), increasing their colonization potential (Nathan et al., 2002).

In this study we conduct a field experiment to quantify plastic responses to altered light quality (R : FR ratio) and quantity (intensity of photosynthetically active radiation, PAR) in natural populations of the native alpine dandelion Taraxacum ceratophorum and its exotic congener Taraxacum officinale. We focus on the shade-avoidance response because it is relevant to success in the heterogeneous light environment separating subalpine and true alpine vegetation, and because the genetic and developmental pathways regulating shade-avoidance responses are well characterized (Smith, 1995, 2000). Consequently, comparing plastic responses to the R : FR light environment between closely related invasive and native species may ultimately enhance our understanding of the genetic basis of traits underlying invasiveness. In this experiment, we specifically address whether native and exotic Taraxacum species differ in: (1) the capacity to alter leaf size, producing sun vs shade leaves under alternative light environments; (2) average inflorescence height and its plastic response to light quality; (3) the degree to which maturation (flowering and fruiting phenology) is accelerated under foliar shade; and (4) the capacity to match dispersal-related traits, including scape length, achene mass and pappus size, to habitat openness as indexed by light environment. We then apply a simple ballistic model to data on dispersal-related traits to address how interspecific differences in trait means and plasticity affect colonization potential for T. ceratophorum and T. officinale.

Materials and Methods

Study system

Taraxacum officinale (L.) Weber, the common dandelion, is a pervasive weed across North America (Holm et al., 1997). After introduction during European settlement (Josselyn, 1672; Solbrig, 1971; Mack, 2003), range expansion of T. officinale has resulted in the formation of sympatric populations with Taraxacum ceratophorum (Ledeb.) DC., a native species of alpine tundra in the Rocky Mountains (Brock, 2004). In areas of sympatry T. officinale commonly grows along disturbed roads and trails, but can also be found interspersed with T. ceratophorum from krummholz (treeline) vegetation into higher alpine meadows. The krummholz creates a diversity of light environments because of its patchwork of open alpine meadows interspersed with stands of bushy willows (Salix glauca) and spruce (Picea engelmannii). For herbaceous plants, photon flux density and R : FR decrease with proximity to the willow overstory and, in meadow, patches of high plant density. Plasticity to light quality and/or irradiance should enhance the ability of T. officinale to intercept more light and potentially increase the seed-release height. Both shade-avoidance traits would facilitate invasion into and through the heterogeneous light environment of the krummholz, as T. officinale plants invade open alpine tundra. Furthermore, T. officinale plants have lower water-use efficiency than those of T. ceratophorum (Brock, 2003), suggesting that mesic habitats or microsites with partial shading (both associated with stands of willows; Totland & Esaete, 2002) offer abiotic conditions conducive to invasion by T. officinale.

On Pennsylvania Mountain (Park County, CO, USA, 39°15′ N, 106°07′ W) T. ceratophorum is a diploid (2n = 16) obligate outcrosser (Brock, 2004), while T. officinale is triploid (3x = 24) and asexual (agamospermous). From mid-July to early August the flowering phenologies of these congeners overlap extensively (Brock, 2004). Physiological studies have shown that, on average, T. ceratophorum plants have less above-ground surface area than those of T. officinale, but produce thicker leaves that assimilate more carbon per unit water lost than T. officinale plants (Brock, 2003). From an acaulescent rosette of leaves both species produce a head-type inflorescence at the end of a leafless scape. These two congeners are taxonomically distinguished by bracts (phyllaries) subtending the mature inflorescence − T. officinale produces noncorniculate, reflexed phyllaries while those of T. ceratophorum plants are horn-like and clasp tightly around the inflorescence (Cronquist et al., 1994). A broad variety of insects move indiscriminately between congeneric inflorescences, transferring equivalent amounts of pollen from T. officinale to T. ceratophorum and between T. ceratophorum inflorescences (Brock, 2003). Deposition of T. officinale pollen on T. ceratophorum stigmas produces moderate seed crops that consist of hybrid and selfed progeny because of a breakdown in self-incompatibility in the presence of interspecific pollen (mentor effect; Brock, 2004). After flowering, bracts subtending the inflorescence (phyllaries) close protectively around developing fruits. Once the seeds mature, the bracts reflex to expose mature achenes (single-seeded fruit hereafter referred to as seeds), each topped by a pappus that aids in wind dispersal.

Experimental design

The experiment was conducted in June 2002 within sympatric populations of T. ceratophorum and T. officinale distributed along an elevation gradient (3555–3575 m) at the upper edge of the krummholz belt on Pennsylvania Mountain. A split-plot design was established within the transition zone between T. officinale (lower) and T. ceratophorum (upper) habitats, where plants of the two species commingled. The area was subdivided into 15 blocks containing T. ceratophorum and another 15 containing T. officinale. Plants of both species were individually examined for phyllary characteristics at flowering in an attempt to exclude possible hybrid plants within the experimental populations. One exotic plant exhibited phyllary characteristics intermediate to T. ceratophorum and T. officinale, and was replaced with the nearest T. officinale plant that had not initiated scape elongation. Within each block, three plants with immature inflorescences were randomly assigned to receive control, foliar shade or neutral shade treatments. On 17 June, panels made of clear acrylic (42.0 × 42.0 cm, GE Polymershapes, Huntersville, NC, USA) were placed over controls. Identical panels were painted with a 3.5% solution of Hostaperm Violet RL pigment (Lee, 1985; Weinig, 2000) in a clear gloss polyeurathane (Minwax Co., Upper Saddle River, NJ, USA) and placed over individuals in the foliar shade treatment to simulate light passing through a leaf canopy. We measured the R : FR ratio (655–665/725–735 nm; Smith, 1994) transmitted through painted panels using a spectroradiometer (LI-1800, Li-Cor, Lincoln, NE, USA) and standardized pigment application with a gravity-fed paint gun to simulate a natural canopy R : FR ratio of 0.45 (Smith, 1982). The pigment solution reduced PAR by 70%. Plants in the neutral shade treatment were covered by knitted shade cloth (70% Black, Dewitt Sudden Shade, Sikeston, MO, USA) placed over 12 gauge vinyl panels to control for the reduction in PAR without altering light quality (R : FR). Panels were suspended at a height of 12.7 cm directly over Taraxacum plants, exposing each rosette to assigned light treatments.

Each individual was monitored daily for the onset of flowering. At flowering, the plant was enclosed in a box by attaching four side panels of identical light transmittance (each 40.6 mm × 63.5 cm) to the original (top) panel. Enclosures ensured that both the plant and its elongating scape were contained in the appropriate light environment. Boxes were raised 37 mm off the ground to facilitate air movement and access to water runoff during rainstorms. Nonetheless, control, neutral shade and foliar shade boxes varied slightly but significantly with respect to both soil moisture content (w/w) (6.1 ± 0.7%[±95 confidence limits (CL)]; 8.3 ± 1.2%; and 8.7 ± 1.2%, respectively, F2,72 = 6.51, P = 0.0025) and excess temperature (degrees deviation from ambient, −1.3 ± 2, −5.3 ± 2 and −4.5 ± 2°C, respectively, F2,14 = 5.17, P = 0.0208). Although control treatments differed significantly from both shading treatments for soil moisture content and temperature (all Tukey's P ≤ 0.05), foliar shade and neutral density treatments were not significantly different from one another (both Tukey's P ≥ 0.8). Plastic responses to changes in light quality, tested by comparisons between neutral density and foliar shade, are not confounded by either temperature or moisture gradients.

Scape length at the onset of flowering was measured with calipers as the distance along the stalk between the inflorescence and the ground. For plants producing multiple inflorescences, scapes of all inflorescences were measured and the values averaged to calculate mean scape length per plant. To ensure fruit maturation in T. ceratophorum, flowers were pollinated daily by ‘brushing’ stigmas with a donor inflorescence randomly selected from conspecific plants at least 15 m away. If total rainfall over a 2 d period was less than 1 mm, plants were given 500 ml water using aqua cones (Gardner's Supply Co., Burlington, VT, USA) to reduced runoff and soil surface evaporation. Even with supplemental watering, the severe drought caused inflorescence abortion and plant mortality, reducing sample sizes in control, foliar shade and neutral shade treatments for both T. ceratophorum (n = 10, 15, 11, respectively) and T. officinale (n = 15, 15, 13, respectively).

After flowering, inflorescences were monitored daily until seed maturation (bracts reflexing to reveal mature seeds). At that time scape length was recorded again. Sample size at the time of seed maturation was further reduced in control, foliar shade and neutral shade treatments for T. ceratophorum (n = 7, 14, 8, respectively) and T. officinale (n = 14, 14, 13, respectively). At the end of the 50 d experiment, the length of the longest leaf was recorded (as was total leaf number per plant) to determine if the light environment caused plastic adjustment in leaf size (or number).

Mature infructescences were harvested and stored in paper cups to prevent damage to the pappus. Two randomly selected seeds from each plant were weighed using a semimicrobalance (Sartorius R160P, Sartorius Corp., Edgewood, NY, USA), and for each the radius of the pappus ‘disk’ was measured with calipers. Plant means for each seed trait were calculated by averaging the pair of measurements. Hemipteran seed predation further reduced sample size in control, foliar shade and neutral shade treatments for T. ceratophorum (n = 6, 10, 6) and T. officinale (n = 9, 12, 10).

Data analysis

anova (here and elsewhere, Proc MIXED, ML option, SAS version 6.12; SAS Institute, Inc., Cary, NC, USA) was conducted to test the effects of treatment, species, and the random spatial blocking factor (nested within species) on scape length, using a split-plot design (Steel et al., 1997). As sample sizes differed between flowering and seed maturation, separate anovas were performed on data collected at each stage. Scape length at flowering and seed maturation was log-transformed to meet assumptions of normality. Here and elsewhere, a posteriori Tukey's pairwise comparisons were used to test the significance of differences among treatments for each species. To evaluate causes of variation in days to flowering and days from flowering to seed maturation, we used a model identical to that described above. Three-way anovas with species and treatment as fixed effects, and block (nested within species) as a random factor, were also conducted to test for significant sources of variation in the total number of leaves, length of longest leaf, seed mass and pappus radius.

Species- and environment-specific models of dispersal potential

Because seed height and mass are thought to be key determinants of dispersal in dandelions, we used a simple ballistic modeling framework to compare the average dispersal distance of T. ceratophorum and T. officinale seeds in open and crowded habitats. The ballistic model (Pasquill & Smith, 1983):

x = Hu/F

predicts the average horizontal dispersal distance (x) in metres from a maternal plant based on the release height of the seed (H); the horizontal wind speed (u) averaged from H to the ground; and the fall rate of the seed (F). This model predicts local dispersal distance as a function of horizontal wind speed, but may underestimate rare long-distance seed dispersal via updraft (Tackenberg et al., 2003). For seed-release height we used the mean scape length for each species under control and foliar shade treatments. This procedure assumes that scape length at fruit set is an accurate surrogate for seed head height (i.e. that stem angle relative to the ground is 90°), that minimum wind speed required for seed abscission is equivalent for the two species, and that seed paths are unobstructed (Ford, 1985; Greene & Johnson, 1989). To estimate species-specific fall rates, we randomly selected 15 infructescences per species (one infructescence per plant) from maternal lines pooled over all three treatments. Pappus radius and seed mass did not vary significantly among treatments (see Fig. 2). Two seeds per infructescence were weighed and dropped from a height of 2 m under still conditions in the laboratory (University of Missouri-Columbia, MO, USA). The time (s) for seeds to reach the ground was recorded and averaged for the pair. Variation in seed-fall rate between species was analyzed using a one-way anova (Proc GLM, SAS version 6.12). The same anova model was re-run on the residuals produced from a regression of fall rate (dependent variable) on seed mass (independent variable) (Proc REG, SAS version 6.12), to determine if species differences in fall rate mainly reflected variation in seed mass.

Figure 2.

Reaction norms for (a) days to flowering and (b) fruit set in Taraxacum ceratophorum and Taraxacum officinale ( ± 95% CL) when exposed to control (square), neutral shade (triangle) and foliar shade (circle) treatments.

The average dispersal distance for each species was estimated over a range of wind speeds (0–7 m s−1), recorded at ≈ 10 cm above ground in open tundra meadows on Pennsylvania Mountain (C.G., unpublished data). To construct confidence limits we conservatively calculated minimum seed-dispersal distance by combining the lower 95% CL value of scape length (H) with the upper limit for seed-fall rate (F). To determine how F and H independently contribute to species differences in seed dispersal under each canopy condition, we ran the model twice more for T. officinale, replacing each dispersal parameter with that of T. ceratophorum. As wind speeds fall by ≈ 66% in stands of Salix glauca (A. Dona & C.G., unpublished data), we reduced wind speed under vegetative conditions to the range 0–2.5 m s−1.

Results

Leaf length varied significantly between species (F1,28 = 26.56, P = 0.0001; for anovas see Supplementary material, Table S1), among treatments (F2,40 = 11.37, P = 0.0001), and with the species × treatment interaction (F2,40 = 3.66, P = 0.0346). Leaf length in T. ceratophorum did not vary with treatment (all Tukey's tests, P ≥ 0.74; Fig. 1). By contrast, plants of T. officinale produced significantly longer leaves under both shading regimes than in the control treatment (both Tukey's tests, P ≤ 0.03; Fig. 1) and had leaves of similar length under foliar shade and neutral shade (Tukey's test, P = 0.287). This pattern indicates that the common dandelion produces larger ‘shade’ leaves in response to reduced light intensity, but not in relation to modifications in light quality. Species, treatment, and species × treatment interaction, did not contribute significantly to variation in leaf number (all factors, P ≥ 0.2).

Figure 1.

Reaction norms for Taraxacum officinale and Taraxacum ceratophorum for (a) leaf length, and (b) flowering and (c) fruiting scape lengths, in response to control (square), neutral shade (triangle) and foliar shade (circle). Points show means and brackets the associated 95% confidence intervals.

At flowering, scape length did not differ between species (F1,28 = 1.96, P = 0.1725), but showed significant plasticity among treatments (F2,45 = 21.58, P = 0.0001; Fig. 1). Flowering scapes developing under control and neutral shade did not differ significantly in length ( = 5.4 ± 0.6 cm, ±95% CL) and ( = 6.2 ± 0.8 cm, respectively; Tukey's test, P = 0.1451). However, plants in both these treatments had significantly shorter scapes than plants in foliar shade ( = 8.3 ± 0.9 cm, both Tukey's tests, P = 0.0003). A marginally significant species × treatment interaction (F2,45 = 2.85, P = 0.0681) indicates a tendency for species to differ in reaction norms of scape length at flowering (Fig. 1). Scape length tends to be more plastic in T. ceratophorum than in T. officinale, because flower-bearing scapes of native plants tend to be shorter than those of exotics under control conditions (Tukey's P = 0.1184, Fig. 1).

The phenology of flowering and seed maturation differed between species and exhibited significant plasticity among treatments (for flowering, F1,28 = 36.14, P = 0.0001; F2,45 = 5.03, P = 0.0107, respectively; for seed maturation, F1,28 = 110.69, P = 0.0001; F2,36 = 12.84, P = 0.0001, respectively). However, the species × treatment interaction was not significant at either stage (F2,45 = 1.96, P = 0.1524 and F2,36 = 2.36, P = 0.7362, respectively). Taraxacum officinale plants flowered and matured seeds more quickly than T. ceratophorum plants (both Tukey's tests, P = 0.0001; Fig. 2). For both species, plants exposed to neutral shade flowered significantly later than control plants (Tukey's test, P = 0.0087; Fig. 2). Plants in foliar shade had intermediate flowering schedules that did not differ significantly from the flowering schedules of either control or neutral shade groups (Tukey's test, P = 0.4866 and P = 0.0900, respectively). Plants in control and foliar shade treatments did not differ in timing of seed maturation (Tukey's test, P = 0.0997), but matured seeds earlier than plants under neutral shade (both Tukey's test, P = 0.0055; Fig. 2).

After flowering, scapes of T. officinale elongated more than scapes of T. ceratophorum ( = 14.7 ± 1.6 cm and  = 10.7 ± 1.4 cm, respectively, F1,28 = 17.02, P = 0.0003). A marginally significant species × treatment interaction (F2,36 = 3.06, P = 0.059) indicates that plants of T. ceratophorum and T. officinale differed in reaction norms for the length of seed-bearing scapes in relation to the light environment (Fig. 1). In T. ceratophorum, seed-bearing scapes exhibited significant plasticity in length among treatments. Individuals in control and neutral shade treatments produced scapes of equivalent length (Tukey's test, P = 0.50) that were significantly shorter than scapes produced under foliar shade (both Tukey's tests, P = 0.004; Fig. 1). By contrast, seed-bearing scapes of T. officinale did not vary significantly in length among light treatments (all Tukey's tests, P ≥ 0.09; Fig. 1).

Seed mass varied significantly between species (F1,25 = 50.91, P = 0.0001), but not among treatments or with the species × treatment interaction (F2,22 = 1.30, P = 0.2932; F2,22 = 0.20, P = 0.8212, respectively). Seeds of T. officinale were smaller than those of T. ceratophorum (0.29 ± 0.037 vs 0.52 ± 0.055 mg, respectively, Fig. 3). Species, treatment, and the species × treatment interaction did not contribute significantly to variation in pappus radius (F1,25 = 0.12, P = 0.730; F2,22 = 1.74, P = 0.1994; F2,22 = 1.29, P = 0.2963, respectively; Fig. 3).

Figure 3.

Reaction norms for (a) pappus radius and (b) seed mass in Taraxacum ceratophorum and Taraxacum officinale ( ± 95% CL) plants in response to control (square), neutral shade (triangle) and foliar shade (circle) treatments.

Seed-dispersal potential

Mean seed-fall rate differed significantly between T. ceratophorum ( = 0.50 ± 0.06 m s−1) and T. officinale ( = 0.25 ± 0.06 m s−1; F1,28 = 36.54, P = 0.0001). The more rapid descent of seeds in T. ceratophorum mainly reflects the species’ difference in seed mass. Regression of fall rate on seed mass dramatically reduced the amount of residual variation explained by species (F1,28 = 2.89, P = 0.10). The ballistic model of seed dispersal predicts that in the open, plants of T. ceratophorum should disperse seeds only 36.3% the distance of T. officinale (Fig. 4). Substituting mean H and F of T. ceratophorum into the model for seed dispersal in T. officinale reduces dispersal distance by 27.5 and 49.0%, respectively (Fig. 4a). This analysis suggests that both the taller seed-bearing scapes and smaller seeds of T. officinale contribute to its superior seed-dispersal capacity in open habitats. The model predicts similar trends in vegetated habitats, with seeds of T. ceratophorum predicted to disperse 44.4% the distance of T. officinale (Fig. 4b). However under foliar shade where congeners are more similar in scape length, interspecific variation in seed size largely accounts for the superior dispersal potential of T. officinale. Substituting H and F of T. ceratophorum for T. officinale reduces seed dispersal by 12.3 and 49.2%, respectively (Fig. 4).

Figure 4.

Predicted mean seed-dispersal distances (±95% CL; see Materials and Methods for calculation) based on modification of a ballistic model of seed dispersal to incorporate species-specific parameters for seed release height (H) and seed-fall rate (F). The two plots illustrate the comparative superiority of seed dispersal in Taraxacum officinale in (a) open and (b) vegetated habitats. Intermediate lines result from substituting F or H of T. officinale with values from T. ceratophorum to examine how each trait contributes to the difference in dispersal capacity between species.

Discussion

Taraxacum ceratophorum and T. officinale exhibit phenotypic plasticity in relation to light quantity (PAR) and quality (R : FR). Differences in plasticity between congeners varied among traits; neither species demonstrated consistently greater plasticity. Vegetative plasticity in response to the light environment was observed only in T. officinale, which produced significantly longer leaves under reduced light intensity. Conversely, only T. ceratophorum displayed a significant plastic response in the length of the seed-bearing scape. In both Taraxacum species, foliar shade induced equivalent elongation of flowering scapes and acceleration of flowering and seed set. These results suggest that differences between invasive and native species in plasticity are not systematic, but instead vary among traits. Models of seed dispersal suggest that canalization for a longer scape and smaller seeds should confer a seed-dispersal advantage for T. officinale relative to native T. ceratophorum. Long-distance dispersal, accelerated seed maturation and a higher degree of plasticity in leaf size may enable the invasive T. officinale to colonize and persist under a wider range of habitats than its native congener.

Vegetative plasticity

Plants of T. officinale produced significantly longer leaves on average than plants of T. ceratophorum. Additionally, plasticity for leaf length in relation to light environment was detected only in plants of T. officinale. In response to the reduction in light intensity under neutral shade and foliar shade treatments, leaves of T. officinale were significantly elongated relative to controls. Increasing leaf area in response to resource limitation (reduced PAR) increases ‘light foraging’ in plants and is selectively advantageous when shade varies predictably over the growing season or within the canopy of an individual plant (shade leaves; Ryser & Eek, 2000; Steinger et al., 2003). Reductions in PAR and R : FR ratio have been shown to modify leaf development in T. officinale, resulting in longer, more rounded leaves with greater surface area (Sanchez, 1967, 1971; Sanchez & Cogliatti, 1975). Vegetative plasticity in Taraxacum has also been shown to vary with breeding system. In its ancestral range, where both sexual and asexual populations of T. officinale occur, leaf size of apomicts is more plastic under experimental shade (de Kovel & de Jong, 1999). Although breeding system covaries with ploidy in T. officinale, similar results have been found in hexaploid apomictic and sexual plants of Antennaria parlinii (Michaels & Bazzaz, 1989). Colonizing apomictic plants exhibit greater plasticity in vegetative growth than sexual plants across light (PAR) and nutrient gradients (Michaels & Bazzaz, 1989). These results support the idea that plasticity in vegetative traits may compensate for loss of genetically based variation in apomictic and/or colonizing lineages. Greater leaf plasticity in North American T. officinale probably contributes to its ecological breadth by providing an advantage in harvesting light under shaded or crowded conditions. Conversely, leaves of alpine T. ceratophorum have lower specific leaf area and use water more efficiently than those of T. officinale, increasing their functional value under the predominantly open and dry conditions above the timber line (Brock, 2003).

Plasticity in reproductive traits

Under foliar shade, plants of both T. ceratophorum and T. officinale produced significantly longer flowering scapes than under control or neutral shade treatments. At this developmental stage, scape length was marginally more plastic in plants of T. ceratophorum, because they tended to produce shorter inflorescences than plants of T. officinale under open conditions. Greater inflorescence height in dense, crowded environments may increase access to inflorescences for flying insects, enhancing pollination success (Pyke, 1981; Lortie & Aarssen, 1999). Although pollen accrual should not alter fecundity in asexual T. officinale, intraspecific pollen movement is required for seed set in the obligate outcrosser T. ceratophorum (Brock, 2004), providing a possible selective advantage for increased flower height under foliar shade. Insect visitors often discriminate between species based on inflorescence height (Levin & Kerster, 1973; Waddington, 1979; Peakall & Handel, 1993; Gumbert & Kunze, 1999); a further consequence of the lack of a species difference between native and exotic congeners at this stage may be elevated interspecific visitation and heterospecific pollen transfer. Hand-pollination of T. ceratophorum with pollen from T. officinale results in a low level of hybridization (Brock, 2004). Similar inflorescence heights may facilitate asymmetrical gene flow between this pair of dandelion species, especially under shaded conditions along the edge of the willow canopy.

Plasticity in reproductive phenology

Plants of T. officinale reached both the flowering and fruit stages significantly earlier than plants of T. ceratophorum. The timing of both reproductive stages was advanced under foliar shade in comparison with the neutral shade treatment, although the congeners did not differ in the degree of plastic response. Results show that, independently of the reduction in PAR, foliar shade accelerates reproductive maturity, a common shade-avoidance response (Davis & Simmons, 1994; Halliday et al., 1994; Devlin et al., 1996). By enabling plants to anticipate competitors, accelerated rates of flowering and fruiting may facilitate reproduction in advance of canopy closure (Ballaréet al., 1997). In comparison with T. ceratophorum plants, flowers and fruits developed earlier in T. officinale plants. Faster rates of reproduction are common life-history traits in weedy plants (Baker, 1965; Higgins et al., 1996; Marco et al., 2002), and should favor higher fecundity before increased crowdedness (Botto & Smith, 2002; Callahan & Pigliucci, 2002). Fitness benefits and costs associated with developmental acceleration are unclear in this and other systems (Callahan & Pigliucci, 2002; Schmitt et al., 2003), but early seed maturation is associated with seed size and germination disadvantages in Plantago lanceolata (Lacey et al., 2003). We also found evidence of a trade-off in timing of seed maturation vs size, with larger seeds of T. ceratophorum maturing significantly later than the smaller seeds of T. officinale. Seed size increases germination success in Taraxacum, suggesting a fitness consequence of advanced maturation (Tweney & Mogie, 1999). The potential to produce hybrid generations will allow us to examine the genetic basis of correlations between reproductive timing, seed size, germination rate and dispersal potential in future work on this system.

Plasticity in dispersal-related traits

The length of seed-bearing scapes differed significantly in plasticity between T. ceratophorum and T. officinale. Decreased R : FR ratio induced a significant increase in scape length in T. ceratophorum, but not in T. officinale. The longer and less plastic scape of T. officinale is consistent with a history of selection for dispersal potential, while short stature is a common plant adaptation in high alpine environments (Clausen et al., 1940; Rochow, 1970). All else being equal, seeds released from greater heights should move further in wind-dispersed species (see ballistic model), but at the cost of potential lodging in windy habitats (Emery et al., 1994; Schmitt et al., 1995).

Traits that enhance seed dispersal and establishment are often associated with ruderal, colonizing or invasive species. For example, apomictic A. parlinii plants in disturbed and early successional habitats have lower seed mass and a pappus morphology that increases drag coefficient compared with sexual conspecifics (O’Connell & Eckert, 2001). Although pappus size did not vary between Taraxacum species (or with light environment), seeds of T. officinale are significantly smaller than seeds of T. ceratophorum. In gravity- and wind-dispersed seeds, mass is inversely related to dispersal distance.

Seed-dispersal models tailored to each Taraxacum species predict that, under sympatry in open habitats, seeds of T. officinale should disperse an average 2.7 times further than seeds of T. ceratophorum. Both greater release height and smaller seeds work in concert to confer a dispersal advantage for T. officinale. The model predicts that seeds of T. officinale should have a 2.2-fold dispersal advantage under closed canopy conditions. However, this prediction requires that seed paths are unobstructed (Ford, 1985; Greene & Johnson, 1989). In vegetated habitats, dispersal spectra of T. officinale and T. ceratophorum may actually converge, given obstruction by vegetative organs, reduced wind speed, and the capacity of plastic scape elongation in native plants partially to compensate for interspecific variation in seed size (Fig. 4). Calibration of seed-dispersal models to habitat structure will require spatially explicit modeling approaches that incorporate habitat heterogeneity.

The greater seed-dispersal potential of T. officinale may incur several costs. As seed size partly accounts for the difference in seed-fall rate between species, seeds of T. officinale may travel further than seeds of T. ceratophorum, but suffer reduced germination or establishment success on arrival (Mogie et al., 1990; Tweney & Mogie, 1999; Gravuer et al., 2003). Furthermore, canalization for increased infructescence height could increase the risk of scape lodging in open, windy alpine habitats (Emery et al., 1994; Schmitt et al., 1995) or encourage seed predation (Ehrlen et al., 2002).

Plasticity and invasiveness

Although the process of species invasion may select for general-purpose genotypes (Williams et al., 1995; Parker et al., 2003), our results suggest that invasive species are not more phenotypically plastic than closely related native species for all traits. Because we have not compared ancestral and exotic populations, we cannot directly address how the process of invasion selected on plasticity in the common dandelion. Nonetheless, results of this study provide insights into whether plasticity explains invasiveness. In Taraxacum, interspecific variation in plasticity is stage- and trait-specific, indicating that attempts to characterize invasive species as more or less plastic will probably meet with little success. Furthermore, genotype-specific plastic responses in phenotypic traits can buffer components of fitness across a range of environments. For example, Sultan & Bazzaz (1993) showed that different genotypes maintained fruit weight over a range of light limitation through different sets of plastic responses. Attempts to classify invasive species by the ‘strength’ of plasticity in specific traits could be confounded by failing to measure all traits contributing to invasiveness. Instead, studies that forego the quantification of plasticity, and focus on direct proxies of fitness across a range of environments, may prove more useful in attempts to identify invasive species (Rejmanek, 2000; Sultan, 2001).

Acknowledgements

We thank two anonymous reviewers for their helpful comments on the manuscript; A. Dona, L. Dudley, and M. Carter for assistance in the field; the University of Colorado at Colorado Springs for providing access to the field site on Pennsylvania Mountain; and M. Ellersieck for statistical advice. R. Kaczorowski, J. Schmitt and lab, and the members of the Paint Shop at the University of Missouri provided help with production of the acrylic treatment panels. This research was supported by a TWA Scholarship 2002–03 to M.B. and by NSF grant DEB-0087412 to C.G.

Supplementary material

The following material is available as supplementary material at http://www.blackwellpublishing.com/products/journals/suppmat/NPH/NPH1300/NPH1300sm.htm

Table 1anovas (from SAS Proc MIXED) for effect of species (Taraxacum ceratophorum and Taraxacum officinale), treatment (control, neutral shade, foliar shade) and their interaction on scape length, developmental time (days to flowering and fruiting), vegetative traits (longest leaf and number of leaves), and achene geometry (pappus radius and seed weight). The block term nested within species (not shown) was designated as a random factor.

Ancillary