Increased fitness and plasticity of an invasive species in its introduced range: a study using Senecio pterophorus

Authors


Correspondence author. E-mail: lidiacano@ub.edu

Summary

  • 1When a plant species is introduced into a new range, it may differentiate genetically from the original populations in the home range. This genetic differentiation may influence the extent to which the invasion of the new range is successful. We tested this hypothesis by examining Senecio pterophorus, a South African shrub that was introduced into NE Spain about 40 years ago. We predicted that in the introduced range invasive populations would perform better and show greater plasticity than native populations.
  • 2Individuals of S. pterophorus from four Spanish (invasive) and four South African (native) populations were grown in Catalonia, Spain, in a common garden in which disturbance and water availability were manipulated. Fitness traits and several ecophysiological parameters were measured.
  • 3The invasive populations of S. pterophorus survived better throughout the summer drought in a disturbed (unvegetated) environment than native South African populations. This success may be attributable to the lower specific leaf area (SLA) and better water content regulation of the invasive populations in this treatment.
  • 4Invasive populations displayed up to three times higher relative growth rate than native populations under conditions of disturbance and non-limiting water availability.
  • 5The reproductive performance of the invasive populations was higher in all treatments except under the most stressful conditions (i.e. in non-watered undisturbed plots), where no plant from either population flowered.
  • 6The results for leaf parameters and chlorophyll fluorescence measurements suggested that the greater fitness of the invasive populations could be attributed to more favourable ecophysiological responses.
  • 7Synthesis. Spanish invasive populations of S. pterophorus performed better in the presence of high levels of disturbance, and displayed higher plasticity of fitness traits in response to resource availability than native South African populations. Our results suggest that genetic differentiation from source populations associated with founding may play a role in invasion success.

Introduction

As a consequence of the colonization of a new geographical range, introduced organisms are subject to genetic bottlenecks, random genetic drift and increased levels of inbreeding. These genetic changes associated with founding may reduce allelic diversity and promote the fixation of alleles, leading to genetic differentiation of the introduced populations (Ellstrand & Elam 1993). Moreover, introduced organisms in a geographically separate new range face new biotic and abiotic selection pressures. Their ability to adapt to the new conditions will contribute to the success of the invasion, enabling a small subset of introduced species to become invaders as a result of rapid evolutionary changes in the new range (Ellstrand & Schierenbeck 2000; Sakai et al. 2001; Lambrinos 2004). Many studies have demonstrated that plants undergo a genetic shift towards larger size in their introduced range (Crawley 1987; Blossey & Nötzold 1995; Siemann & Rogers 2001; Leger & Rice 2003; Erfmeier & Bruelheide 2005; Stastny et al. 2005, but see Willis et al. 2000 and Thébaud & Simberloff 2001), which has sometimes been attributed to a change in the allocation of resources to anti-herbivore defences (EICA hypothesis; Blossey & Nötzold 1995). However, genetic shifts can affect other traits which are relevant to the new environment (Bossdorf et al. 2005), including reproductive, morphological (e.g. DeWalt et al. 2004), ecophysiological (e.g. Niinemets et al. 2003) and chemical (e.g. Joshi & Vrieling 2005) traits.

Adaptive plasticity of morphological or physiological traits may contribute to the fitness homeostasis of invaders in variable environments (Williams et al. 1995; Rejmanek 2000). As adaptive phenotypic plasticity may play an important role in the fitness of plants, it is a potential target for selection (Schlichting & Pigliucci 1995). Therefore, when assessing evolutionary changes in the introduced range of an invasive species, plasticity can be considered as a trait in itself – a property of a genotype –, which may differ with respect to its native range (Richards et al. 2006). However, studies that compare plasticity between native and invasive populations are scarce (Leger & Rice 2003; Lavergne & Molofsky 2007) and few authors have attempted to compare plasticity for morphological or ecophysiological traits (but see Kaufman & Smouse 2001 and DeWalt et al. 2004).

The environmental characteristics of the area invaded may influence which morphological and ecophysiological characteristics of the invader differ between its introduced and native ranges. In a Mediterranean climate, plants undergo periods of stress with water deficits and high irradiance from late spring to early autumn. This can produce functional–structural changes in the leaves, such as a decrease in leaf water content and stomata conductance, leading to photosynthesis decline (Chaves et al. 2003). Plants that optimize their water status and photoprotective mechanisms during periodic stress episodes will, therefore, gain an advantage over plants that do not.

In this study, we searched for recent genetic differentiation in quantitative traits of introduced populations of Senecio pterophorus, a South African perennial shrub, which is invading NE Spain. Specifically, we examined: (i) the fitness traits; (ii) one leaf trait whose adaptive plasticity is likely to improve fitness (specific leaf area, SLA); and (iii) ecophysiological responses that contribute to fitness in different environments. Senecio pterophorus is well-suited to testing for evolutionary changes in its new range, because its distribution in continental Europe is restricted to NE Spain (Chamorro et al. 2006) and the environmental growth conditions of the different populations are well known and relatively similar. Moreover, its rapid expansion in recent years may reflect that it has overcome the lag phase after its introduction about 40 years ago. We carried out a common garden experiment with plants from invasive (Spanish) and native (South African) populations. We predicted that invasive populations would be fitter and better adapted in ecophysiological terms to environmental conditions in the introduced range than native populations. Given that Spanish populations of S. pterophorus mainly colonize disturbed areas, we also predicted that Spanish invasive populations would be fitter in disturbed environments than native South African populations. We therefore manipulated disturbance and water availability, which in turn allowed us to test for differences between origins in plastic responses to heterogeneous environments. Although plasticity is more directly measured by comparing phenotypes of the same genotype in different environments, estimates of plasticity at the population level allow a more realistic evaluation of the role of plasticity in natural populations in a comparative framework. We examined: (i) whether invasive populations perform better in the introduced range than native populations; and (ii) whether invasive populations show greater plasticity in response to different treatments than native populations.

Methods

study species

Senecio pterophorus D.C. (Asteraceae) is a dwarf shrub native to the Eastern Cape Province, South Africa, where it forms scattered populations in forest margins, grasslands and the fynbos. It was introduced to South Australia in ship ballast about 1930 and spread to south-east South Australia (Scott & Delfosse 1992), where it tolerates a wide range of soil types and occupies areas with 500–1500 mm annual rainfall (Walsh 1999). In South Australia it causes heavy productivity losses in agricultural areas and it is a strong competitor that excludes native species in natural communities (Parsons & Cuthbertson 1992). In Europe, it has only been recorded in the British Isles (Stace 1997) and NE Spain (Pino et al. 2000). However, it occurs only very sporadic in the British Isles and does not form stable populations (Stephen Harris, Oxford University Herbaria, personal communication).

It was accidentally introduced into NE Spain 30–40 years ago, probably via sheep wool or cotton, and from 1995 onwards it has been widely reported in a large number of localities in the province of Barcelona (Pino et al. 2000; Chamorro et al. 2006). Although it forms dense and persistent populations in river beds and human disturbed areas, S. pterophorus is currently spreading rapidly into less disturbed communities such as grasslands and shrublands. The invasion potential in Mediterranean plant communities in Spain depends on the degree of competitiveness of the resident vegetation and on fluctuations in water availability (Caño et al. 2007). Forests seem to be less invasible than shrubland. Also, invasion success is higher when seedlings establish during autumn than when they establish during spring (Caño et al. 2007). Many ecological and life-history traits have been shown to contribute to the invasion success of S. pterophorus in Spain, such as higher relative growth rate, biomass production and reproductive output than native congeners (Garcia-Serrano et al. 2004, 2005, 2007; Sans et al. 2004).

plant material

Seeds of S. pterophorus were collected from eight populations in 2003 and 2004: four South African populations and four Spanish populations (see Table 1). The South African populations were sampled widely across their native range, in regions with differing climatic conditions. The Transkei region (Sidwadwenii population), in the north of the Eastern Cape Province, has a subtropical climate with a pronounced dry winter and most of the annual rainfall concentrated in summer. Towards the south of the Eastern Cape Province (Bushman Valley population), rainfall is more unpredictable but the climate is generally characterized by summer rainfall with occasional hot dry periods in summer and extended dry periods in winter. Populations from the Cape Region in the Western Cape Province (Rhodes Memorial and Vlakkensberg populations) receive most of the rain in winter. The Spanish populations were collected in their current introduced range, which is characterized by a Mediterranean climate with summer drought and a relatively rainy autumn (Table 1).

Table 1.  Sites of origin of populations used in the common garden experiments
Source populationHabitat typeCoordinatesTemperature, (°C, Max./Min.)Rainfall (mm)Bioclimatic diagnosis
YearWinterSpringSummerAutumn
  1. Max., annual average of the monthly average of the maximum temperatures; Min., annual average of the monthly average of the minimum temperatures. Climatic data and bioclimatic diagnosis follows Rivas-Martínez (2000).

Spain
 Garden site 41°23′N 2°07′E20.1/12.7 597130140107221Thermomediterranean dry
 MatadeperaTemporary flooded dry basin41°36′N 2°01′E20.8/9.6 594115183109187Mesomediterranean dry
 SabadellRuderal vegetation-invaded river bed41°34′N 2°05′E20.8/9.6 594115183109187Mesomediterranean dry
 RipolletPathway border in the river margin41°30′N 2°08′E20.8/9.6 594115183109187Mesomediterranean dry
 BesòsRuderal vegetation-invaded river bed41°26′N 2°13′E20.8/10.9 540108109102221Mesomediterranean dry
South Africa
 Rhodes MemorialGrass-invaded coastal Fynbos33°57′S 18°27′E  22/11 515252 84 49150Thermomediterranean subhumid
 VlakkenburgGrass and acacia-invaded coastal Fynbos34°01′S 18°23′E21.6/11.11008492185 88243Thermomediterranean subhumid
 Bushman ValleyS. pterophorus– invaded dried-up dam33°32′S 26°23′E26.1/13.4 610 67176201166Infratemperate dry
 SidwadweniiRuderal vegetation-invaded abandoned field31°28′S 28°51′E  24/11 650 44183264159Subtropical subhumid

Seeds were air-dried and stored at room temperature in paper bags until sowing. One hundred randomly selected seeds from each population were weighed to assess whether differences in plant biomass could be attributed to seed size, and to account at least partly for maternal effects. Seeds were sown on February 2005 in bedding cells with a 1 : 1 mixture of sterilized sand and peat, and placed in a greenhouse bench under optimal conditions (20 °C and natural photoperiod, after Afán (2000)). Seedlings, each with four to six true leaves, were transplanted into the experimental field in April 2005. The height and maximum crown diameters of all seedlings were measured prior to planting. These variables showed a significant allometric relationship with the above-ground biomass (r2 = 0.645, F1,61 = 115.60, P < 0.0001), and were used to estimate seedling size. Seedlings were sprayed with water until establishment to minimize transplant shock and seedling mortality.

experimental design

The study was carried out between April 2005 and January 2006 at the Experimental Field Services of the University of Barcelona, Spain, which is an outdoor field previously used for other garden experiments (see Table 1). The climate is Mediterranean, with rainfall averaging 600 mm year−1 and with mean monthly temperatures ranging from 29 °C in July to 10 °C in January. The experiment used a complete, two-way factorial, randomized block design. The layout consisted of four blocks (9 × 5 m), 1 m apart. Each block was sub-divided into four equal plots (2 × 4 m) 1 m apart, which were randomly assigned to one of the combinations of two factors: disturbance and watering. In the ‘disturbed treatment’ (hereafter D) all the above-ground biomass was periodically removed. In the ‘undisturbed treatment’ (hereafter ND) we did not remove the resident vegetation (dominated by turfs of Brachypodium phoenicoides, Dactylis glomerata, Lolium rigidum, Erodium malacoides, Chenopodium album, Amaranthus retroflexus and Convolvulus arvensis), which provided a 100% cover across the entire ND subplots. In the plots receiving the ‘water treatment’ (hereafter W) water was periodically added to the soil, while no water was added to the ‘no-water treatment’ (hereafter NW), which received ambient rainfall. W plots were irrigated at regular intervals two or three times a week. We applied a total of 260 L of water per plot throughout the experiment. Plots were watered with a hose with a fine sprinkler to allow the water to soak into the soil and to minimize lateral movement of water. In each 2 × 4 m subplot, 32 seedlings (four individuals × four populations × two origins) were transplanted in April 2005 in a randomized design, 0.5 m apart to avoid intraspecific competition.

data collection

Fitness traits

Survival and flowering (percentage of original transplants showing at least one flower head) were monitored from April 2005 every 2 weeks until the end of the experiment, that is, January 2006. The number of seed heads was counted in all reproductive individuals every second month and at harvest; total seed head production per plant was calculated as the total accumulated number of seed heads per plant. In January 2006, we uprooted each individual and calculated the final total biomass by weighing plants after drying at 60 °C for 48 h.

Leaf parameters

Two leaf parameters related to plant performance and fitness were evaluated using a subset of the plants: SLA and the relative water content of leaves (RWC). Decreases in SLA and mean leaf area are often associated with drought exposure and high irradiance (Gratani & Bombelli 1999; Wright et al. 2001). Leaf parameters were sampled twice during the dry season: in the second week of June and at the end of July. On each occasion we collected 8–16 fully expanded leaves per plant from five to eight randomly selected plants per population and treatment combination. Plants used for leaf collection were randomly selected in June and again in July, in order to avoid systematic effects of leaf excision on plant growth. Leaves were excised pre-dawn and immediately weighed (fresh weight, Wf). Then we hydrated leaves to saturation for 24 h at 4 °C in Petri dishes with distilled water. Leaves were weighed again (saturated weight, Ws). We scanned the hydrated leaves to measure the leaf area by Image J software (National Institute of Health, Bethesda, MA). Finally, we measured their dry weight after drying them at 60 °C to constant weight (dry weight, Wd). We estimated the SLA and the pre-dawn RWC of leaves as follows:

image
image

Chlorophyll fluorescence parameters

At the beginning of July (mid-summer), leaf chlorophyll fluorescence was measured in a subset of the plants. Parameters of chlorophyll fluorescence are indicators of the level of damage caused by high irradiance and/or low water availability. Chlorophyll fluorescence was determined in two fully expanded leaves per plant, in five plants per population and treatment combination with a portable, pulse-modulated fluorometer (Mini-Pam Photosynthesis Yield Analyzer, Walz, Effeltrich, Germany), which includes a leaf-clip holder (2030-B; Waltz) and a micro-quantum sensor to monitor photosynthetic active radiation (PAR). Steady-state fluorescence (F) under the prevailing ambient light and maximal fluorescence inline image during a saturating pulse (> 5000 µmol photons m−2 s−1, 800 ms duration) were determined at midday on a sunny day. The same leaves were dark-adapted by wrapping in aluminium foil. After at least 20 min, maximum fluorescence (Fm) and minimal fluorescence (Fo) were measured and used to obtain the intrinsic efficiency of open photosystem II centres during illumination, inline image, which was calculated after Oxborough & Baker (1997). Non-photochemical quenching NPQ (thermal energy dissipation) was calculated using the Stern–Volmer equation:

image

NPQ is related to rapid responses of metabolic mechanisms involving processes of non-radiative dissipation of excess energy as heat. At high irradiances, when photosynthesis is saturated acting as electron sink, NPQ is expected to increase and inline image is expected to decrease. At not-saturating irradiations, plants displaying lower NPQ or higher inline image are expected to have higher photosynthetic rates.

statistical analysis

Survival curves within treatment were calculated using the product-limit method and compared with the log-rank test (Peto & Pike 1973). As we only compared two origins, here the log-rank test is equivalent to the Mantel–Haenszel test. SEs were calculated using the method of Greenwood (Altman 1991).

Initial seed and seedling sizes were analysed by a mixed model, with the origin (South African vs. Spanish) as a fixed effect and populations (nested within origin) as a random effect. Mean growth, percent plants flowering and mean seed head production at the end of the experiment were analysed by a mixed model, with block effect, disturbance, water treatment and origin as fixed effects and population (nested within origin) as a random effect. To avoid pseudo-replication, in each analysis we computed the mean growth and the mean seed head number per population for each treatment subplot within a block. The effects of origin, population, disturbance and water addition on leaf and chlorophyll fluorescence parameters were analysed by a mixed model with origin, disturbance and water addition as fixed effects, and population (nested within origin) as a random effect. PAR was used as a covariate if its effect on the model was significant. Proportional data were arcsine-square–root transformed and non-proportional data were log-transformed to achieve normality and homoscedasticity of residuals (Sokal & Rohlf 1981). All analyses were performed by fitting general linear models, using the GLM procedure of sas (SAS Institute Inc. 1999). We used type-III sums of squares for the calculation of F statistics. Post-hoc multiple comparisons of marginal means were performed with the LSMEANS statement in the GLM procedure.

Significant effects of plant origin were considered indications of genetically based differences in trait or ecophysiological response between the two origins (South African vs. Spanish); significant effects of population reflected genetic variability for a given trait or response within the species; significant effects of treatments indicated plasticity or responsiveness of the corresponding parameter; and interactions between origin and treatment showed genetic differences in plasticity or responsiveness due to plant origin (Schlichting 1986). To interpret the direction of the response of the Spanish invasive and South African native populations to disturbance and water factors, we presented the norms of reaction of those ecophysiological parameters that showed a significant effect of the disturbance × origin and/or water × origin interaction. The average values per population of these parameters were regressed against the average fitness (survival), which is taken as an indication of the extent to which they contribute to fitness. Since the relationship between a parameter and fitness depends on the context of the environment (Dudley 1996), we performed separate linear regressions in D, ND, W and NW treatments.

Results

differences between invasive and native populations in fitness traits

Survival

The highest mortality of the two origins was recorded in ND–NW treatment (Fig. 1). Most deaths occurred between June and September, with dramatically higher mortality in NW plots than in W plots during this period (Fig. 1). These months coincided with the drought period (Meteorological Service of Catalonia, Government of Catalonia). Survival of invasive and native populations was affected in different ways by disturbance. Mortality of native populations in D treatment was significantly higher than that of Spanish invasive populations, whereas no significant differences in survival were detected in ND treatment between native and invasive populations (Fig. 1).

Figure 1.

Seasonal course of plant survival of Spanish invasive populations and South African native populations of Senecio pterophorus growing in disturbed (D) and undisturbed (ND) plots, and in watered (W) and unwatered (NW) plots. Points are based on all the populations of the same origin, as there were no significant differences between populations within origin (log-rank test, P > 0.05). Bars represent SEs calculated by the method of Greenwood (Altman 1991). Significant differences between origins at each treatment are indicated by * when P < 0.05, according to the log-rank test.

Final biomass

The final biomass of plants was more affected by spatial heterogeneity (block effect) than by genetic differences (origin and population effect, Table 2). Moreover, growth showed a plastic response to environmental conditions, as indicated by the significant effect of disturbance treatment and disturbance × water interaction. Furthermore, the significant effect of the water × origin and disturbance × water × origin interaction indicated differences in plasticity between the two origins (Table 2). Spanish invasive populations displayed significantly higher final biomass than South African native populations in D–W plots (Fig. 2a). In this treatment, the increase of growth was concentrated from August to October (data not shown), when the average increase in Spanish populations was 235.04 g compared to 71.46 g (over three times lower) in South African populations.

Table 2. anova of the effect of continental origin, population (pop.), disturbance (disturb.) and water supply (water) on Senecio pterophorus total biomass, percent plants flowering, seed head production at the end of the experiment and several ecophysiological parameters (see text for details) measured in June and July. (Only parameters showing significant effect of the origin and/or the origin × treatment interaction are shown)
 Biomass (g)Flowering (%)Seed headsSLA June (cm2 g−1)RWC June (%)SLA July (m2 g−1)RWC July (%)inline imageNPQ
Source of variationd.f.Fd.f.FFFFd.f.FFFF
  • 0.05 < P < 0.1,

  • *

    P < 0.05,

  • **

    P < 0.01;

  • ***

    P < 0.0001. Arrows indicate superiority in performance of invasive populations.

Origin10.86113.83**15.74**  1.79  5.54*1  0.20 1.44 0.03 0.82
Population (origin)61.186 2.40* 0.90  0.99  7.86***6  1.48 6.89*** 1.54 2.86*
Block36.18**3 1.25 1.09 – –
Disturbance19.38**111.76** 9.50**437.04***  1.111361.15*** 5.12* 0.2211.74**
Water10.14155.96***31.30***  5.85*274.56***1  5.42*97.95***45.31***13.54**
Disturb. × Water18.52**1 1.97  3.37  0.041  0.68 4.11* 1.58 3.22
Disturb. × Origin11.171 6.22* 8.06**  0.32 54.15***1  7.08** 0.8511.27** 1.21
Water × Origin15.05*113.99**13.19**  0.07 21.79***1  0.20 2.26* 3.29 3.39
Disturb. × Water × Origin15.09*1 0.22  0.02  0.021  9.65** 4.66* 4.40* 1.56
Disturb. × Pop. (origin)60.396 0.42 0.18  1.58  3.80**6  2.33* 3.37** 1.31 0.81
Water × Pop. (origin)60.826 1.57 0.74  0.31  3.02*6  1.17 0.62 1.59 2.48*
Disturb. × Water × Pop. (origin)20.976 0.85  2.06  2.21*5  2.31* 8.13*** 2.42* 0.89
Error  6    6    
Figure 2.

(a) Total biomass, (b) mean percent plants flowering and (c) mean seed head production per flowering plant of Spanish invasive and South African native populations of Senecio pterophorus measured at the end of the experiment in relation to the disturbance (D) and no disturbance (ND) treatments, and treatments with additional water (W) and without (NW). Values are means ± SE. Significant differences between origins at each treatment are indicated by * when P < 0.05 and + when 0.1 > P > 0.05, based on post-hoc multiple comparisons of marginal means in anova.

We did not find significant differences between origins in seed size (F1,6 = 0.12, P = 0.742), nor in seedling size (F1,6 = 0.004 , P = 0.982).

Flowering

Spanish invasive populations had greater reproductive fitness than South African native populations in all treatments, except in the ND–NW treatment, in which no plant flowered (Fig. 2, Table 2). Moreover, in the D–NW treatment, no South African plant flowered and two of the four South African populations did not flower in any treatment (population effect, Table 2). Water addition significantly enhanced flowering and seed head production of both origins. However, the effect of this treatment differed between origins, as South African plants did not flower in NW plots (Fig. 2) (water × origin, Table 2). Moreover, disturbance favoured flowering and seed-head production in Spanish plants, whereas it had no significant effect on fitness of South African plants (disturbance × origin, Table 2) (Fig. 2).

Overall, Spanish populations displayed higher fitness than South African populations in treatments with disturbance, and similar or slightly higher fitness than South African populations in treatments without disturbance.

differences between invasive and native populations in plasticity of sla

Overall disturbance significantly decreased the SLA (Table 2, Fig. 3a). However, the plastic response of SLA to disturbance differed between Spanish invasive and South African native populations as measured in July (disturbance × origin interaction, Table 2). Note that the norms of reaction crossed (Fig. 3a) showing that the response of one population to disturbance was opposite to that of the other. There was a significant effect of disturbance × population on SLA, further indicating genetic variation for plasticity (Table 2).

Figure 3.

Norms of reaction to (a) disturbance treatment (D, disturbance; ND, no disturbance) and (b) water treatment (W, water addition; NW, no water addition) of leaf and chlorophyll fluorescence parameters in Spanish invasive and South African native populations of Senecio pterophorus. Only parameters showing significant effect of the origin × treatment interaction are shown. SLA, specific leaf area; RWC, relative water content; inline image, efficiency of the reactive centres of photosystem II; NPQ, non-photochemical quenching of photosystem II. Values are means ± SE based on means for individual populations. Significant differences between origins at each treatment are indicated by * when P < 0.05 and + when 0.1 > P > 0.05.

Under disturbance conditions, invasive populations had lower SLA than native populations in July (Fig. 3a). In contrast, native populations displayed lower SLA than invasive ones in ND plots. SLA in July was negatively correlated with survival in the disturbance treatment but not the ND treatment (Table 3).

Table 3.  Linear regressions relating the within-treatment parameter means to within-treatment survival of Senecio pterophorus. (Only values from July sampling were used to calculate regressions for RWC and SLA)
 DNDWNW
βtβtβtβt
  1. D, disturbance; ND, no disturbance; W, water addition; NW, no water addition. See text for details on parameters. Standardized regression coefficients (β) are reported. Boldface indicates significant coefficients according to t-test, *P < 0.05, **P < 0.001 ***P < 0.0001.

RWC (%)0.8032.03**0.8967.26***0.5192.28*0.6953.48*
SLA (cm2 g−1)–0.524–2.31*–0.289–1.090.2230.86–0.635–2.96*
inline image0.5792.66*0.8135.04**0.3841.560.4031.59
NPQ–0.529–2.33*–0.827–5.31**–0.550–2.46*–0.174–0.64

differences between invasive and native populations in ecophysiological responses

The ecophysiological response to disturbance and water supply differed between Spanish invasive and South African native populations (disturbance × origin and water × origin interaction, Table 2).

Overall water restrictions decreased the RWC both in June and July (Table 2, Fig. 3b) but invasive populations maintained higher RWC than native ones in both no-watered and disturbed plots in June (Fig. 3b). The effect of the interaction of both treatments on RWC also differed between invasive and native populations (disturbance × water × origin, Table 2). RWC was significantly higher in invasive populations than in native ones in June in D–W plots (mean ± SE, 84.23% ± 0.80% vs. 77.84% ± 0.90%, P = 0.014) and D–NW plots (75.93% ± 0.80% vs. 64.54% ± 0.90%, P = 0.010), a difference that was maintained in July only in D–W plots (83.67% ± 1.22% vs. 74.45% ± 1.29%, P = 0.008). Increased RWC correlated positively with survival in all treatments (Table 3).

Under disturbance conditions, invasive populations had higher efficiency of the open photosystem II centres, inline image, than native populations (disturbance × origin interaction, Table 2, Fig. 3a). In contrast, native populations displayed higher inline image than invasive ones in ND plots. inline image correlated with survival in both treatments (Table 3). Moreover, in W plots, invasive plants also had significantly higher inline image and lower thermal energy dissipation, NPQ, than native populations (Fig. 3b), and these responses correlated with survival in this treatment (Table 3).

The effect of the interaction of both treatments on NPQ differed between invasive and native populations (disturbance × water × origin, Table 2). Invasive populations displayed lower mean NPQ in the D–W treatment than native populations (2.06 ± 0.19 vs. 2.71 ± 0.21, P = 0.070).

Discussion

The evolution of an introduced species in its new range can contribute to its invasiveness (Sakai et al. 2001; Lambrinos 2004). We found that invasive populations of S. pterophorus outperformed native populations when grown in common garden conditions, except under the most stressful conditions (i.e. in an undisturbed, non-watered environment), which revealed a genetic differentiation between the two origins.

differences between invasive and native populations

Spanish invasive populations suffered less mortality than South African native populations throughout summer drought under disturbance conditions. Moreover, in disturbed and watered plots, invasive populations showed up to three times greater biomass than native populations. No plant of either origin flowered in undisturbed and unwatered plots, but invasive populations had greater reproductive fitness in all the other treatments.

In the disturbed environment, Spanish populations reduced their SLA and maintained their pre-dawn RWC more than South African populations. Although disturbance is considered to be a favourable factor by reducing competition with established vegetation, in the Mediterranean climate, plants may be exposed to high irradiance and transpiration rates in the absence of vegetation canopy. A decrease in SLA has been associated with greater irradiance and water stress by several authors (Gratani & Bombelli 1999; Wright et al. 2001). This is also suggested by our finding that SLA decreased from June to July (data not shown). Moreover, this response is adaptive in a disturbed environment as it correlates with survival in this treatment. The lower SLA displayed by the invasive populations of S. pterophorus in disturbed plots is also indicative of increased sclerophylly in Spanish populations. On the other hand, higher pre-dawn RWC in this treatment in invasive populations is indicative of their improved water content regulation. Note that Spanish invasive populations maintained higher RWC under water restrictions (NW treatment), as well. In addition, greater efficiency of open photosystem II centres inline image, in the invasive populations growing in disturbed plots reflects less damage of photosystem II in these populations than in native ones. Therefore, adaptive plasticity of SLA and improved water content regulation and photoprotective mechanisms can contribute to the greater survival of invasive populations in a disturbed environment.

Moreover, lower thermal energy dissipation (NPQ) of the invasive populations in disturbed watered plots may be associated with the higher biomass of Spanish plants in this treatment, since lower levels of thermal energy dissipation under non-limiting water conditions (W treatment) is likely to be due to a higher photochemical sink for electrons (i.e. higher photosynthetic rate; Fleck et al. 1998).

the role of phenotypic plasticity

Plasticity of fitness traits can play an important role in invasion success (Baker 1974). Different reaction norms of fitness traits can contribute to invasiveness in variable environments. These include a flat or invariable reaction norm of fitness traits, which reflects constant fitness across a broad range of environments (the ‘jack-of-all trades’ scenario, i.e. a general purpose genotype), increased plasticity of fitness traits in response to high resource availability (‘master-of-some’) or a combination of the two abilities together (‘jack-and-master’) (Richards et al. 2006). In our study, the growth response of the invasive populations of S. pterophorus resembled the ‘master-of-some’ type. However, invasive populations of S. pterophorus also displayed higher seed head production in less favourable conditions, which resembled a ‘jack-of-all-trades’ or a certain ‘jack-and-master’ behaviour. Increased plasticity of fitness traits in S. pterophorus may allow greater invasiveness in variable environments through a greater ability to capitalize on high resource availability. Indeed, although S. pterophorus primarily colonizes disturbed habitats, it is currently spreading into more competitive natural communities (Chamorro et al. 2006), taking advantage of small disturbances or periods of greater water availability (Caño et al. 2007).

Bradshaw (1965) stated that phenotypic plasticity at one level of organization may contribute to the maintenance of homeostasis at another level. Our results suggest that adaptive plasticity of SLA in Spanish invasive populations of S. pterophorus plays an important role in improving fitness in disturbed conditions when water is likely to limit survival.

Although evolutionary changes in plasticity may contribute to the invasiveness of introduced species, few studies have compared plasticity among native and invasive populations. Leger & Rice (2003) found that the invasive populations of Eschscholzia californica displayed greater plasticity in biomass than native populations under reduced competition, which fits with our results for S. pterophorus. Kaufman & Smouse (2001) also found greater plasticity in growth in introduced populations of Melaleuca quinquenervia in response to variation in pH. A recent study also demonstrates that invasive genotypes of Phalaris arundinacea are more plastic across a gradient of moisture than native genotypes (Lavergne & Molofsky 2007). Therefore, the greater plasticity of the introduced populations vs. native ones seems generally to be due to a greater ability to increase fitness in response to high resource availability.

mechanisms involved in genetic differentiation

The phenotypic differences between Spanish and South African populations may reflect various processes. Since the seeds used were collected directly in the field, maternal environmental effects could have affected our results. Seed size is a commonly measured parental effect that affects plant size, survival and flowering (Roach & Wulff 1987). However, we did not find significant differences in seed or seedling sizes between Spanish and South African populations. Moreover, although maternal effects are expressed early in the life cycle, they are expected to decrease as plants age (Roach & Wulff 1987); in our study, the differences in survival and growth rate were only detected after several months’ growth. It seems more likely that the differences found are genetically based.

The better performance of the invasive Spanish populations does not necessarily imply post-introduction evolutionary changes. Rather, it may be related to which genotypes were introduced from the native range and which were able to persist under novel conditions, which in turn may be due to stochastic and/or selection processes. In the colonization of a new range, founder effects may indeed interact with selection (Eckert et al. 1996). Genetic bottlenecks can cause genetic drift and increase levels of inbreeding, reducing genetic diversity (Ellstrand & Elam 1993). However, such a scenario may favour purging of deleterious alleles via selection, which could explain in part why invasive populations performed better than native ones. In particular, our results suggest that introduced populations may evolve greater invasiveness through selection in the presence of high levels of disturbance. Also, multiple introductions from the native range may counteract the effect of drift by increasing genetic diversity and selection diversification. However, whether multiple introductions have occurred remains unknown.

The better performance of invasive populations may also have been achieved through selection acting on mean values and/or on plasticities of morphological or physiological traits (Via et al. 1995; Pigliucci & Schlichting 1996). For instance, we found significant linear regressions between SLA and fitness, which is usually indicative of potential for directional selection on a trait (Lande & Arnold 1983). However, since the mechanisms of selection were not the object of our study, further studies should clarify the effects of selection on ecophysiological traits (directional, stabilising or disruptive) and discriminate between direct and indirect selection resulting from the high level of correlation between traits (Lande & Arnold 1983). More importantly, since rapid evolutionary changes emerge as a potential explanation for the success of invasions, studies are needed to specifically investigate the role of the different genetic mechanisms involved in the genetic differentiation between the native and the introduced ranges.

Acknowledgements

We would like to thank Heidi Hawks, Tony Dold, Gurutze Calvo-Ugarteburu and Olga Cruz for their help in seed collection. We are grateful to Margarita Roldán and Montserrat Bassa for their technical assistance and to two anonymous referees for valuable comments on previous versions of this manuscript. This research was partially funded by the Science and Technology Department of the Spanish Government (project REN2001-2837), with a fellowship for the first author, and by the European Research Group (GDRE 122) ‘Mediterranean and mountain ecosystems in a changing world’.

Ancillary