• The relationship between zinc (Zn) tolerance and hyperaccumulation in Thlaspi caerulescens was investigated from F1 and F2 crosses within and among metallicolous and nonmetallicolous Mediterranean populations.
• F1 offspring were grown on increasingly Zn-enriched soils to test Zn enrichment effects, and many families of F2 offspring were grown on a Zn-rich soil.
• Tolerance of F1 offspring depended on stress intensity. Tolerance of interecotype crosses was intermediate between that of the intraecotype crosses. No difference in Zn accumulation appeared among the F1 offspring from the three crosses involving metallicolous parents. Otherwise, none of these offspring exceeded the Zn hyperaccumulation threshold (10 000 mg kg−1), unlike the nonmetallicolous ones. The latter also showed the highest mortality. In some F2 families from interecotype crosses, hyperaccumulation values exceeded 15 000 mg kg−1 in nontolerant offspring, whereas tolerant offspring displayed lower values (c. 10 000 mg kg−1). There was no difference between tolerant and nontolerant offspring when they showed low hyperaccumulation.
• Therefore, the relationship between tolerance and hyperaccumulation in F1 and F2 crosses depended on the hyperaccumulation level of plants.
Heavy metals occur both as natural trace elements, for example nickel (Ni) in serpentine soils derived from ultramafic rocks (Baker et al., 1992), or as residues from human activities, such as in sewage sludge or on mine spoils enriched with zinc (Zn), lead (Pb) or cadmium (Cd) (Ernst, 1990). That some species from local floras are able to survive, grow and reproduce in areas highly polluted with toxic heavy metals by evolving tolerant ecotypes has been known for several decades (Antonovics et al., 1971). Some tolerant plant species are also able to accumulate very high heavy metal concentrations in their shoots without suffering from metal toxicity (Baker, 1981). Such species are called hyperaccumulators and c. 400 hyperaccumulating species have been reported (Brooks et al., 1997). For zinc, a concentration threshold of 10 000 mg kg−1 dry weight is currently admitted for a species to be considered a hyperaccumulator (Baker & Brooks, 1989). In Europe, Zn hyperaccumulators occur almost exclusively in the Brassicaceae; two well-studied examples are Arabidopsis halleri and several species of Thlaspi and Alyssum (Reeves & Brooks, 1983). These species can occur on serpentine soils, and on calamine soils enriched in Zn, Pb and Cd, often due to mining and industrial activities.
Such metal-tolerant and hyperaccumulating species are of considerable interest for the ecological remediation of moderately polluted soils by means of phytoextraction, a method that consists of successive harvests of the above-ground plant biomass where heavy metals are highly concentrated, leading with time to the removal of the metal from soils (Chaney et al., 1997). Improving metal uptake and metal tolerance in a phytoextraction perspective would be easier if these two traits were positively correlated. However, previous studies on this topic have produced contrasting results. Ingrouille & Smirnoff (1986) found independent phenotypic relationships between Zn tolerance and accumulation in British populations of Thlaspi caerulescens. Krämer et al. (1997) showed that nickel hyperaccumulation in Thlaspi goesingense was associated with nickel tolerance. Meerts & Van Isacker (1997) and Escarréet al. (2000) showed that Zn tolerance was lower for nonmetallicolous populations of T. caerulescens compared with metallicolous populations, but that nonmetallicolous populations showed higher foliar Zn concentrations. Assunção et al. (2003a) found that the relationship between Zn, Cd and Ni accumulation and tolerance in T. caerulescens was metal and ecotype-specific. Roosens et al. (2003) found a negative relationship between Cd tolerance and hyperaccumulation in metallicolous populations of T. caerulescens, and a negative correlation between shoot Zn and Cd concentrations and shoot to root biomass ratios. However, to understand the genetic relationship between tolerance and hyperaccumulation, it is necessary to study the cosegregation of these traits in F2 or F3 offspring from crosses between species or populations differing in their tolerance and hyperaccumulation levels. Macnair et al. (1999) crossed the Zn tolerant and hyperaccumulating species Arabidopsis halleri with the nontolerant and nonaccumulating species Arabidopsis lyrata ssp. petraea. They found an independent segregation of Zn tolerance and hyperaccumulation, and demonstrated that these traits were under independent genetic control. Likewise, for T. caerulescens, Assunção et al. (2003b) detected independent segregation patterns of Zn tolerance and hyperaccumulation using the F2 and F3 offspring from a single intraspecific cross between a plant from a metallicolous population and a plant from a nonmetallicolous population. These contrasting results concerning the relationship between metal tolerance and hyperaccumulation could be due to species, population and metal specificity of tolerance and hyperaccumulation (Schat et al., 1999).
In previous experiments (Frérot et al., 2003), Zn hyperaccumulation differences appeared in F1 offspring from controlled crosses involving Mediterranean metallicolous and nonmetallicolous populations of T. caerulescens. Zn hyperaccumulation showed genetic inheritance, and low Zn hyperaccumulation (threshold of 10 000 mg kg−1 Zn in a contaminated soil, see also Escarréet al., 2000) of metallicolous plants was dominant to high Zn hyperaccumulation of nonmetallicolous plants. The purpose of the present work is to investigate the relationship between Zn tolerance and hyperaccumulation by using F1 and F2 offspring. Its originality concerns the use of several metallicolous and nonmetallicolous populations of T. caerulescens in numerous controlled crosses between and within these populations. In addition, F1 offspring were grown in increasingly Zn-enriched soils to test whether Zn tolerance and hyperaccumulation relationship depended on level of Zn enrichment.
Materials and Methods
Thlaspi caerulescens J. & C. Presl (Brassicaceae) is an annual, biennial or perennial herb, occurring in southern, western and central Europe, eastwards to Poland and Yugoslavia, and in parts of north Europe (Tutin et al., 1993). It has been studied largely for its Zn tolerance and hyperaccumulation, although it also shows Ni and Cd tolerance and hyperaccumulation (Meerts & Van Isacker, 1997; Escarréet al., 2000; Assunção et al., 2003a).
In the Languedoc-Roussillon (south of France), T. caerulescens occurs on soils strongly enriched with Zn, Pb and Cd (metallicolous ecotype), in abandoned mining areas where Zn and Pb were extracted in the past. Mine sites are located in the upper valley of the Vis river, adjacent to the Causse du Larzac (south of the Cévennes). T. caerulescens is also found on calcareous normal soils (nonmetallicolous ecotype), on the Causse du Larzac and its surroundings (the Séranne Mountain). Seeds were collected from five metallicolous (hereafter M) populations. ‘Moyen-Age’ (MG) and ‘Les Avinières’ (AV) are mining sites abandoned several hundreds of years ago (Demange, 1973; Bailly Maitre, 1990). At ‘Les Malines’ (MAL), ‘Saint-Bresson’ (SB) and ‘Pommiers’ (POM), mining activity ceased one or two decades ago (Rolley, 2002). Seeds of five nonmetallicolous (hereafter NM) populations were collected: ‘Les Infruts’ (INF), ‘Cirque de Navacelles’ (NAV), ‘Saint-Michel’ (SM) and ‘Charnier’ (CH) from the Causse du Larzac, and ‘Séranne’ (SER) from the Séranne Mountain.
Seedlings from the metallicolous M populations were grown on a contaminated mixture made of 90% of normal soil (ammonium-EDTA extractable metal concentrations: 50 mg kg−1 Zn and 1 mg kg−1 Cd) from the experimental station of the Centre d’Ecologie Fonctionnelle et Evolutive (CEFE/CNRS Montpellier, France) and 10% of contaminated soil from the AV site (ammonium-EDTA extractable metal concentrations: 33 800 mg kg−1 Zn and 200 mg kg−1 Cd). Seedlings from the nonmetallicolous NM populations were grown on normal soil from the CEFE station. These plants were first grown in an unlit glasshouse, then placed outside in order to induce flowering. Metallicolous and nonmetallicolous individuals were manually crossed in all maternal × paternal combinations (M × M, M × NM, NM × M and NM × NM) by rubbing open anthers of male parents on the stigmas of female parents. In order to avoid self-fertilisation, flowers were emasculated prior to anther dehiscence. On each plant, some flowers were also self-pollinated.
Several individuals in each M and NM population were used per cross type, so that the pooled results could be referred to ecotypes (M and NM, respectively, for M × M and NM × NM intraecotype crosses) or interecotypes (M × NM and NM × M crosses) offspring. It was not possible to obtain exactly the same number of F1 offspring per cross type, because successful crosses produced only a few seeds, some crosses failed, and some adults died during the experiment. Therefore, for all tests, the F1 individuals were pooled according to their cross type. Moreover, in a previous experiment (J. Escarré, unpublished), there was no difference in Zn hyperaccumulation between populations within ecotype (metallicolous or nonmetallicolous) cultivated in a substrate with 20 000 mg kg−1 Zn (see treatment 4 below). The F2 offspring were obtained by manual self-pollination of several F1 offspring from the four cross types (M × M, M × NM, NM × M and NM × NM) cultivated in the conditions described below.
F1 and F2 offspring cultivation
Seedlings from F1 offspring were transplanted into 1-l pots in each of four soil treatments (Table 1). Treatments 1, 2 and 3 consisted of a mixture of 50% of normal soil from the CEFE station, 25% sand and 25% compost (horse manure, 30% organic matter, pH = 7). The control treatment 1 was free of added metals. Treatments 2 and 3 were Zn-enriched with zinc sulphate (respectively, 1000 and 3000 mg kg−1 Zn). Zinc sulphate was added as powder and manually homogenized with the soil in each pot. Treatment 4 was obtained by mixing 50% of normal soil from the CEFE station with 50% of soil from the AV polluted site.
Table 1. Soil treatments used to estimate tolerance and hyperaccumulation levels in different crosses among metallicolous and nonmetallicolous populations of Thlaspi caerulescens
The normal soil came from the experimental station of the CEFE (CNRS Montpellier, France). ‘AV soil’ was collected on the former mine site ‘Les Avinières’.
About 100 F1 offspring were used for each treatment. The M × M crosses showed less mortality than the other crosses during seedling culture, so that their initial number at transplantation on the four soil treatments was higher. Pots were placed in a random block design. F1 offspring were grown in an unlit glasshouse and homogenously watered with deionised water for three months. Five to six families per F2 cross type were selected, and up to 15 plants per family were grown for five months under the same conditions as F1 offspring in treatment 4. Some leaves per plant were then collected for Zn analysis and the F2 offspring were allowed to flower.
Treatment 4 was used because in previous experiments (Escarréet al., 2000; Frérot et al., 2003), this mix of normal and polluted soil revealed striking differences in Zn hyperaccumulation between metallicolous and nonmetallicolous populations. It was thus selected for the cultivation of the F2 individuals, and also because its moderate toxicity compared with treatment 3 (see the results for the F1 offspring) allowed the survival of most F2 individuals as seed source for further experiments on genetic determinism of Zn hyperaccumulation ability. In addition, in accordance with the current phytoextraction concern, the plant performances have also to be tested in soil conditions close to those observed in field conditions.
Determination of Zn tolerance and hyperaccumulation
Zn tolerance on F1 and F2 offspring was evaluated by survival rates (SR), that is the ratio of plant number alive at the end of the experiment to initial plant number.
As necrosis and chlorosis of leaves (Assunção et al., 2003a,b) and yield reduction are visual symptoms of both Zn deficiency and toxicity (Brennan et al., 1993; Chaney, 1993), we also estimated Zn tolerance of F1 offspring using two other criteria. First, we quantified the vitality rate (VR) as the proportion of tolerant plant number to living plant number, in which nontolerant plants had more than half their leaves suffering from chlorosis and necrosis, while tolerant plants had more than half of their leaves unaffected. This index was closely linked to plant survival because most plants presenting more than half their leaves with necrosis and chlorosis died after some months of growth and did not flower. Second, we defined a tolerance index (TI) as the ratio of above-ground biomass produced on contaminated soil (treatments 2, 3 or 4) to above-ground biomass on uncontaminated soil (treatment 1), following Meerts & Van Isacker (1997) and Escarréet al. (2000). We thus obtained three series of mean TI for each cross type. To calculate each mean TI, we divided the mean biomass of the individuals from a given cross in the contaminated treatment by the mean biomass of the individuals from the same cross in the noncontaminated treatment 1.
Zn concentrations in mature leaves were measured with the zincon method developed for Arabidopsis halleri (Macnair & Smirnoff, 1999). This method is based on UV-visible spectrophotometry using zincon as coloured Zn-chelating agent. Mature leaves were selected because they showed a higher Zn concentration than young leaves, and thus gave a better estimation of the Zn hyperaccumulation ability of the plant (Perronnet et al., 2003). Zn concentration in F1 offspring was estimated from the mean of two measurements, and was measured as late as possible (3 months of growth), that is just after the estimation of survival rates, owing to their rapid decrease in treatment 3 (see Results). Zn concentration in F2 offspring was measured after 5 months of growth, before the less tolerant individuals began to die, allowing the comparison of Zn hyperaccumulation between tolerant (i.e. living) and nontolerant (i.e. dead) plants.
For all analyses, the F1 offspring were pooled into cross types. The F2 offspring were pooled into cross types for survival data and Zn hyperaccumulation data analyses. For Zn hyperaccumulation, data for tolerant and nontolerant F2 individuals were also analysed at the family level.
As survival and vitality rates are binomial responses, we analysed the differences between them by a generalized linear model using a logit link function, with the GENMOD procedure of SAS (SAS, 2001). Two main factors were considered for the F1 and F2 offspring: treatment and cross type. Using the GLM procedure of SAS (SAS, 2001), the differences between tolerance indexes for the F1 individuals were analysed by a two main factors analysis of variance (treatment and cross type) after a log-transformation of the data to improve the normality. The differences in Zn hyperaccumulation among F1 cross types were analysed by a three main factors analysis of variance (block, treatment, cross type). The comparisons between Zn hyperaccumulation levels among F2 individuals pooled by cross types were detected by a single factor analysis of variance. Among the five families having more than two individuals per type (tolerant and nontolerant), the differences between tolerant and nontolerant F2 individuals within family were tested by a two factors analysis of variance (family and tolerance level). Type III sums of square were always used because data sets were unbalanced. Comparisons of means were performed with least-squares means tests (SAS, 2001).
Survival and vitality rates (SR and VR) of F1 offspring showed significant differences among treatments, crosses and their interaction (Table 2). For this reason, we analysed the differences among crosses for each treatment.
Table 2. Comparisons of survival rates (SR) and vitality rates (VR) of Thlaspi caerulescens by a least-means square test, for the four F1 crosses in the four different zinc (Zn)-rich treatments, and for the F2 crosses in treatment 4 (Table 1)
In the four treatments, there was no significant difference in SR and VR of F1 offspring from the reciprocal M × NM and NM × M crosses (Fig. 1a,b). In treatment 1, the differences between the four types of cross reach the significant threshold for VR (d.f. = 3, χ2 = 7.95, P = 0.047) and tended to be significant for SR (d.f. = 3, χ2 = 6.81, P = 0.079). The offspring from the M × M crosses were significantly less tolerant to a low Zn concentration than those from the NM × NM crosses for the two indexes. In treatment 2, no overall significant difference among the four cross types appeared for either SR or VR (SR: d.f. = 3, χ2 = 1.67, P = 0.64; VR: d.f. = 3, χ2 = 5.90, P = 0.12), even though a significant difference was obtained between the offspring from NM × M crosses and those from NM × NM crosses for VR (Fig. 1b). In treatment 3, the offspring from M × M crosses displayed significantly higher tolerance parameters than those recorded for other crosses (SR: d.f. = 3, χ2 = 29.43, P < 0.0001; VR: d.f. = 3, χ2 = 35.46, P < 0.0001), among which no significant differences were detected (Fig. 1a,b). In treatment 4, the same trends were found as in the treatment 3 for VR ( d.f. = 3, χ2s = 45.81, P < 0.0001), but not for SR ( d.f. = 3, χ2 = 6.28, P = 0.099). The offspring from M × M crosses showed significantly higher SR and VR than the offspring from NM × NM crosses, and had a higher VR than the offspring from NM × M crosses (Fig. 1a,b).
The TI for the three contaminated treatments are represented in Fig. 2. In treatment 2, the TI were not significantly different among crosses (d.f. = 3, 28, F = 0.31, P = 0.82). In treatment 3, biomass indices were significantly different (d.f. = 3, 26, F = 9.49, P = 0.0002). The offspring from the interecotype crosses had an intermediate TI, both similar to the M × M offspring, whereas the NM × NM cross was significantly different from the M × NM and M × M cross types. In treatment 4, the overall differences were significant (d.f. = 3, 26, F = 5.00, P = 0.0072), with the same statistical pattern as in treatment 3.
The SR for F2 offspring in treatment 4 (Fig. 3) showed significant differences between cross types (Table 2), in particular, the offspring from M × M crosses were significantly more tolerant than the offspring from NM × NM crosses, in accordance with the results for the F1 offspring in the same treatment.
Zn hyperaccumulation in the F1 offspring showed significant differences among treatments and crosses, and their interaction (Table 3).
Table 3. anova for zinc (Zn) hyperaccumulation of Thlaspi caerulescens (a) for F1 crosses grown in four treatments and (b) for F2 crosses grown in treatment 4 (Table 1)
In treatment 1, all four cross types had low Zn concentration in their shoots (Fig. 4). In treatments 2 and 4, the offspring from the NM × NM crosses contained a significantly higher Zn concentration, and in treatment 3 a significantly lower Zn concentration, than the other cross types, which showed the same accumulation levels in the three treatments. Only the NM × NM crosses in treatment 4 had a Zn concentration higher than the hyperaccumulation threshold for Zn (10 000 mg kg−1).
The differences in Zn hyperaccumulation values between F2 offspring in treatment 4 were significant (Table 3). The offspring from M × NM and NM × M crosses had a similar Zn hyperaccumulation level, between the lowest Zn concentration in M × M crosses and the highest Zn concentration in NM × NM crosses (Fig. 5). These results were in accordance with the results for F1 offspring for the same treatment.
Nontolerant F2 individuals hyperaccumulated significantly higher Zn concentrations than tolerant F2 individuals in the interecotype crosses (Fig. 6). Intra-ecotype crosses showed no significant differences in Zn hyperaccumulation. For this reason, only the families from the interecotype crosses were represented in Fig. 7. Many of these families showed very unbalanced sample sizes between tolerant and nontolerant individuals (because no more than two individuals died), so that they were not taken into account for statistical comparisons of hyperaccumulation. Five families were then retained. They reached the threshold of significance for differences between them (d.f. = 4, 62, F = 2.46, P = 0.054), and exhibited significant differences for the interaction between family and tolerance level (d.f. = 4, 62, F = 4.07, P = 0.005). Three showed a significantly higher Zn concentration in nontolerant individuals (MAL × INF, P = 0.0074; MG × SER, P = 0.0002; SER × AV, P = 0.0001), and two showed no difference between tolerant and nontolerant individuals (MAL × CH, P = 0.46 and NAV × AV, P = 0.22).
Expression of tolerance
The results on F1 and F2 offspring showed complex patterns of tolerance due to variations in relation to stress intensity and to the way in which tolerance was assessed.
In treatment 1, the F1 offspring from M × M crosses suffered from Zn deficiency compared with the NM × NM crosses, as revealed by a low survival rate and a low vitality rate. Assunção et al. (2001) suggested that the depleted growth of metallicolous populations of T. caerulescens in a Zn-poor medium could be explained by higher Zn sequestration efficiency in metallicolous populations compared with nonmetallicolous populations, leading to a deficiency in Zn necessary for several metabolic functions. On the contrary, there was no overall difference for survival, vitality or growth among crosses in treatment 2.
In the two treatments with a high level of Zn (3 and 4), the F1 offspring from NM × NM crosses showed significant lower survival, vitality and growth than the F1 offspring from M × M crosses. This confirms previous work by Meerts & Van Isacker (1997), Escarréet al. (2000) and Assunção et al. (2003a), which have shown that tolerance differences between metallicolous and nonmetallicolous populations appeared on highly Zn-concentrated treatments. Treatment 3 was almost as toxic for interecotype crosses as for NM individuals for survival and vitality. In treatment 4, the F1 offspring from interecotype crosses were significantly more tolerant than the NM individuals for VR and as tolerant as M individuals for SR and TI. This indicated that treatment 3 was more toxic than treatment 4, although the concentration of extractable Zn was greater in the latter treatment (Table 1). This could be due to the greater solubility of Zn sulphate (96.5 g 100 ml−1 in cold water) in treatment 3 compared with Zn availability in the AV contaminated soil (Meerts et al., 2003).
Relationship between Zn tolerance and hyperaccumulation
Comparisons of the responses of F1 and F2 offspring from M × M and NM × NM crosses to the different treatments illustrate several important points concerning the relationship between Zn tolerance and hyperaccumulation. In treatment 4, we obtained a negative relationship between tolerance and Zn hyperaccumulation. F1 or F2 offspring from NM × NM crosses were less tolerant but more hyperaccumulating than F1 or F2 offspring from M × M crosses. On the contrary, the results in treatment 3 were in favour of a positive relationship as the NM ecotype, less tolerant than the M ecotype, was the least Zn hyperaccumulating. The results from treatment 3 could be explained by its extreme toxicity, which may have caused rapid elimination of the least tolerant individuals, so that the Zn concentration had been measured on the few surviving most tolerant NM plants, which were also the least Zn hyperaccumulating ones. Previous studies have indeed shown that nonmetallicolous populations of T. caerulescens have an enhanced ability to accumulate Zn with regard to metallicolous populations (Meerts & Van Isacker, 1997; Escarréet al., 2000; Assunção et al., 2001; Meerts et al., 2003). Therefore, in nonmetallicolous individuals grown on contaminated soil, excessive amounts of Zn could be present in the cytoplasm and disturb metabolic processes, as a result of a reduced ability to sequestrate Zn in leaf cell vacuoles (Assunção et al., 2001). This helps explain why hyperaccumulation and tolerance may be negatively correlated in the offspring from M × M and NM × NM crosses.
However, when taking into account the F1 offspring from interecotype crosses, Zn tolerance and hyperaccumulation were not correlated. Plants from M × NM and NM × M crosses indeed showed the same hyperaccumulation level as the plants from M × M crosses in the four treatments, whereas their tolerance levels were significantly different in treatments 3 and 4, implying a more complex kind of relationship between tolerance and hyperaccumulation than a simple inverse relationship. Nevertheless, in these three crosses, there were no plants that exceeded the Zn hyperaccumulation threshold (10 000 mg kg−1). Only the F1 offspring from NM × NM crosses in treatment 4 were ‘true’ hyperaccumulators. The results relating to F1 offspring also demonstrated that low Zn hyperaccumulation ability is dominant over high Zn hyperaccumulation ability, in accordance with Frérot et al. (2003) (except for treatment 3 due to heavy toxicity). The intermediate tolerance level of F1 offspring from interecotype crosses suggested a partial dominance component in the genetic make-up of Zn tolerance ability.
With F2 offspring, and in agreement with the study of Assunção et al. (2003b), we obtained an inverse relationship between Zn tolerance and hyperaccumulation with offspring from several interecotype crosses: the high Zn hyperaccumulation ability occurred mainly in the less tolerant individuals, and the low Zn hyperaccumulation ability occurred mainly in the tolerant individuals. Otherwise, the results relating to both the F1 and F2 crosses suggest that the relationship between tolerance and hyperaccumulation depends on the hyperaccumulation level of plants. When Zn concentration in the aerial parts was around or below 10 000 mg kg−1, the survival rates were very high in F1 and F2 offspring, and there was no difference between tolerant and nontolerant plants in F2 offspring. Conversely, most nontolerant F2 offspring had Zn hyperaccumulation values exceeding 15 000 mg kg−1. Thus, among the five families whose sample sizes between tolerant and nontolerant individuals allowed comparisons, three F2 families clearly showed a negative relationship. In the two other families, the inverse relationship did not appear but the Zn accumulation levels for both nontolerant and tolerant individuals were quite low compared with the levels for nontolerant individuals in the other three families.
F2 data from interecotype crosses between metallicolous and nonmetallicolous populations of the hyperaccumulator Thlaspi caerulescens suggest that Zn hyperaccumulation and Zn tolerance are ruled by counteracting processes, resulting in that high hyperaccumulation ability and low tolerance associated with nonmetallicolous populations. Conversely, high tolerance and restriction of metal uptake in aerial parts below the hyperaccumulation threshold could have been coselected in metallicolous populations. The relationship between tolerance and hyperaccumulation in F1 and F2 crosses also depended on the hyperaccumulation level of plants: below the Zn hyperaccumulation threshold, there was no relationship between tolerance and hyperaccumulation. Therefore, the independency of the two characters is still debatable. In addition, plants both Zn tolerant and Zn hyperaccumulating (and conversely) were obtained in our segregates, so that the genetic make-up remains to be elucidated.
We thank Dr Pierre Saumitou-Laprade and Prof. Pierre Meerts for their helpful suggestions on the experiments. We also thank Dr John Thompson for his relevant commentaries on the manuscript, and Prof. Mark R. Macnair for his indications during the development of the zincon method for Thlaspi caerulescens. This research was supported by the ‘Environnement, Vie et Sociétés’ and by ‘Biodiversité’ CNRS programs, by travel grants of the Direction des Relations Internationales of the CNRS (France) to J.E and of the Fonds National de la Recherche Scientifique (Belgium) to C.L., in addition to a grant to H. Frérot from the CNRS-Région Languedoc Roussillon.