Within and between population genetic variation for zinc accumulation in Arabidopsis halleri


  • Mark R. Macnair

    Corresponding author
    1. School of Biological Sciences, University of Exeter, Hatherly Laboratories, Prince of Wales Rd, Exeter EX4 4PS, UK
      Author for correspondence: Mark R. MacnairTel: +1392 263791Fax: +1392 273700 Email: M.R.Macnair@ex.ac.uk
    Search for more papers by this author

Author for correspondence: Mark R. MacnairTel: +1392 263791Fax: +1392 273700 Email: M.R.Macnair@ex.ac.uk


  • • Hyperaccumulator plants in the field show significant variation in the metal concentration in their aerial parts, but little is known of the causes of this variation. This paper investigates the role of soil zinc (Zn) concentration and genetic variation in causing between and within population variation in Zn accumulation in Arabidopsis halleri.
  • • Seed from 17 populations of A. halleri collected in central Europe were grown under standard conditions at three external Zn concentrations and tested for Zn concentration in the leaves.
  • • Between population variation was highest at low external zinc concentrations. At 10 µm Zn some plants had very low leaf Zn concentrations, and were indistinguishable from nonaccumulators. However, at higher Zn concentrations, all plants showed hyperaccumulation. There were no differences in the accumulating abilities of populations from sites with different degrees of contamination.
  • • Heritability of accumulation, determined for individual families from three populations, was quite high (25–50%), indicating that selection for increased accumulating ability should be possible, although selection would be easier at low external Zn concentrations. The Zn concentration of field collected plants was affected partly by plant genotype but not by the total soil Zn around their roots.


Hyperaccumulation is a fascinating phenomenon in which a very small class of species accumulate exceptionally high concentrations of metals in their aerial parts (Brooks, 1998). With only very few exceptions, hyperaccumulation appears to be a species level trait, with all populations and individuals of a species exhibiting the character. There is substantial interest in the phenomenon because of the potential to use hyperaccumulating plants to phytoremediate land (Salt et al., 1998), and some progress is being made in understanding the mechanisms and physiology of the character (Persans & Salt, 2000; Assunção et al., 2001; Clemens, 2001).

In the field, individual plants of a hyperaccumulating species exhibit a very wide variation in phenotype, even within a single population (Bert et al., 2002). For instance, in the zinc hyperaccumulator Arabidopsis halleri, plants sampled from contaminated sites within the Harz mountains, Germany, had concentrations of leaf zinc of 0.6–5% zinc on a d. wt basis (unpublished data). It is important to know what factors lead to this variation. Intuitively, two of the most important determinants could be bioavailable soil zinc concentration and individual genotype. It is expected that soil zinc levels should be related to the plant metal concentration, although the relationship may not be linear, and it is possible that plant metal concentration could be relatively insensitive to soil metal concentration over a wide range of soil metal levels if the curve relating plant metal concentration to soil concentration saturates at quite a low external zinc concentration (Baker, 1981). Clearly genetic variation in accumulating ability could lead to variation in plant metal concentration.

There have been only a few studies of genetic variation in this character, and the majority of work has been conducted on the zinc/nickel hyperaccumulator Thlaspi caerulescens. A number of studies have found that different populations of this species exhibit variation in tolerance and accumulation of zinc, and also in the ability to accumulate zinc in above ground tissues (Ingrouille & Smirnoff, 1986; Pollard & Baker, 1996; Meerts & Van Isacker, 1997; Escarréet al., 2000). Pollard & Baker (1996) studied two populations of T. caerulescens from Britain. They found that there was significant interpopulation variation in zinc accumulation, and one of the populations showed significant genetic variance for this character. Meerts & van Isacker (1997) compared metallicolous and nonmetallicolous populations of T. caerulescens. They found that metallicolous populations were more tolerant of zinc, and accumulated less zinc in aerial parts. They also found substantial variation between families within populations, suggesting heritable variation for this character. Further studies on this species have found that different accessions differ in their relative accumulation of different metals (Lombi et al., 2001; Assunção et al., 2001).

Less work has been conducted on A. halleri.Bert et al. (2000) studied zinc accumulation in two populations of A. halleri. They found that the population from the uncontaminated site accumulated more zinc than the one from a contaminated site. Bert et al. (2002) studied the zinc concentration of field collected plants from a number of contaminated and uncontaminated sites. They found wide variation in zinc concentration, which was not easily reconciled with the contamination status of the sites. They did not, however, test the ability of the plants to accumulate metal, which can only be done under standard conditions. In this paper, I investigate the accumulation of zinc by a number of populations of A. halleri from both contaminated and uncontaminated environments, and test the hypothesis that accumulation is related to soil contamination. In addition, I test whether there is genetic variation within populations for zinc accumulation.

Materials and Methods

Plant material

Populations of A. halleri were sampled on various occasions between 1995 and 1999 for seed and soil. Except as noted below, seed was collected from as many plants as possible, and bulked (wild seed). Plants from some populations were polycrossed (see next section) to produce ‘Exeter grown’ samples. For most populations, between two and six soil samples were collected from the rooting horizon. Populations used and their locations are given in Table 1. In all experiments, seeds were germinated on sand and transferred as soon as plants were big enough to handle (c. 1 wk) into holes in polystyrene rafts which floated on a nutrient solution in 10-l plastic trays (50–70 plants per tray). The nutrient solution was based on one given by Chaney & Bell (1987) and consisted of: MgSO4: 2 mm; Ca(NO3)2: 0.5 mm; KNO3 : 0.5 mm; K2HPO4: 0.1 mm; CuSO4: 0.2 µm; MnCl2: 2 µm; H3BO3: 10 µm; MoO3: 0.1 µm; FeEDDHA: 10 µm. Zn was added at concentrations noted in different experiments as ZnSO4. Trays were grown in a growth room with a 10-h day and temperature 25°C (day) and 15°C (night). Solutions were changed weekly and aerated continuously.

Table 1.  The populations of Arabidopsis halleri used in this paper, with total metal concentrations of soil
PopulationLocationSeed typeDescriptionTotal metal concentrations of soil (mg g−1 d. wt) Expt
  1. Nd: Not determined. W: Wild collected seed. E: Exeter grown seed. SDs of metal concentrations are given in parentheses. The experiment in which the populations were used is also given.

Pec pod SnezkouCzeck rep.ERoadsideNdNdNdNdNd2
Ober schulenbergHarzWMine10 880(254)3660(28.3)2.2(0.28)660(28)2.6(0.28)2
Velkǎ lypǎCzeck rep.WRoadsideNdNdNdNdNd2
BechenhausHarzWCar park3230(523)6190(1683)3.8(0.8)1096(865)14.4(13.8)2
Schwarze GruberHarzWMine17 200(6788)12 660(877)9.8(0.28)660(28.3)19.4(1.4)2
GoseHarzWRiver2480(735)23 020(6930)0.8(1.1)155(64)1.0(0.28)2
GoslarHarzWContaminated pasture23 200(566)5620(452)2.2(0.28)2881(4071)113(7.1)2
OttiliaeschachtHarzWMine24 000(1697)6720(113)9.2(0)400(564)129(7.4)2,3
J-F HalleHarzWMine3660(424)9020(990)11.8(1.4)760(57)14.0(2.3)2


A. halleri is a self-incompatible herb, and polycrosses between plants were made by planting out at least 15 plants in a square of approx 1 m2 in the Exeter experimental garden in the autumn. If more than one polycross was being performed at one time, the different polycrosses were separated by at least 10 m. Though this separation distance may not eliminate gene flow between the plots completely, most of the seed will be set by crossing within the polycross. The plants were allowed to vernalise naturally over winter and flowered in May–July. The plants were protected from pigeons with netting.

Metal analysis

Plant zinc concentration was measured without killing the plants by the methods of Macnair & Smirnoff (1999) in which small leaf or root fragments (10–40 mg) are frozen to break the cells, then extracted with 3% Sulphosalycilic acid and zinc determined using the colorimetric reagent zincon (Sigma-Aldrich, Poole, UK) at pH 9.6. Soil samples were collected from the majority of populations. For soils, samples were air-dried, sieved to remove large stones, then ground in an agate pestle and mortar to a fine powder. The soil was digested with concentrated nitric acid for 12 h at 115°C, after which the temperature was raised to 125°C for 2 h and to 140°C for a further 2 h. The nitric acid was then boiled off at 180°C and the metal redissolved in 1 m HCL. Metal concentrations (Zn, Cu, Cd, Mn, Ag) were determined on a Pye Unicam SP9 AAS using BDH spectrosol standards. The total soil metal concentrations were determined because, though they may not correlate entirely with the bioavailable pool from which the plants would take them up, there is no universally recognised method of determining such a pool.

Experiment 1: The pattern of zinc uptake over time

Ninety plants from Innerste, and 12 plants from Schadenbeck, were established as above in medium containing 10 µm Zn. After 3 wk they were transferred to 250 µm, after a further 4 wk to 500 µm and after another 4 wk to 1000 µm Zn. Leaf zinc concentration was measured on a sample of plants chosen at random weekly after transfer to 250 µm, and root concentrations from one week later (at the first sampling point the roots were too small to sample without killing the plants). Between 20 and 40 plants were sampled on each occasion. At wk 9 (i.e. the week after transfer to 1000 µm), all the surviving plants were sampled.

Experiment 2

This experiment tested whether the zinc accumulating ability of populations differed, whether the populations responded to change in external zinc concentration in the same way, and whether the mean population accumulation phenotype was related to soil zinc status. Fifteen plants from each of the populations listed in Table 1 were grown as above at 10 µm for 4 wk. For two populations, seed germination was poor, and only 3–4 seedlings were grown. Leaf zinc concentration was then determined, and the plants transferred to 250 µm for 5 wk. Leaf zinc was again determined, and the plants transferred to 500 µm for a further 5 wk. All plants were completely randomised across trays in this experiment, and at each sampling two separate leaves were sampled and tested for zinc concentration. The correlation between the two replicate samples was about 0.75 and the mean of the samples was used in all statistical analyses.

Experiment 3

This experiment tests for genetic variation in zinc accumulation. In August 1999, three populations in the Harz mountains (Innerste, Schadenbeck and Ottiliaeschacht) were visited. Twenty plants (only 17 from Innerste) with ripe siliques were identified, and a maternal leaf sample, ripe seed, and a sample of soil from around their roots were collected. Seed was collected from as many siliques as possible (1–5) and pooled. The seed was germinated as above, and five plants from each family (i.e. the seed collected from a single mother) were grown and included in experiment 2.


Experiment 1

Fig. 1 gives the mean leaf and root zinc concentration of the plants sampled each week. It can be seen that the mean concentration rises steadily till about wk 7, when it plateaus out at a mean value of about 90 µmol g−1 f. wt. Note that, assuming that d. wt is about 15% of f. wt, this value corresponds to about 4% zinc (d. wt), which equates roughly with the concentrations found in field collected plants. The change in external zinc concentrations (shown with arrows on Fig. 1) does not have any apparent effect on the pattern of leaf accumulation, but note that root concentration rises on the subsequent sampling point on both occasions. Otherwise, after wk 2, root concentration is always substantially less than shoot concentration, and is relatively unaffected by date or external concentration. At wk 9, leaf zinc concentrations of the 80 plants sampled ranged from 34.7 to 157.4 µmol g−1 f. wt (mean = 81.8 ± 2.6, SD = 23.3).

Figure 1.

Leaf (diamonds) and root (squares) zinc concentration (µmol g f. wt−1) of plants grown over 13 wk in hydroponic solution containing zinc. Before time 0, the plants had been grown at 10 µm Zn; at wk 0 they were transferred to 250 µm, and at times indicated by arrows to 500 µm and 1000 µm. N > 20 in each case. Error bars show SE.

Experiment 2

The overall distribution of plant leaf zinc concentration of all plants grown in experiment 2 at each of the three concentrations is given in Fig. 2. At 10 µm, the variation in zinc concentration is high, with some plants having very low concentrations in their leaves, while others have the higher concentrations expected of these accumulating plants. There is a suggestion in Fig. 2(a) that there may be two overlapping distributions present, but at higher concentrations this is no longer apparent. The variation remains high, but at 250 µm and 500 µm all plants display a hyperaccumulating phenotype. There is a high correlation between the values individual plants show at each of the three concentrations (data not shown), and, since the values are not independent of each other, a principal components analysis was performed. The first principal component (PCA1) was the only variable with an eigenvalue greater than 1, and could be interpreted as a measure of overall accumulation.

Figure 2.

Histograms of leaf zinc concentrations (µmol g f. wt−1) of 530 plants grown in experiment 2. Plants were tested after growth for 4 weeks in 10 µm Zn (a), after a further 5 wk in 250 µm Zn (b) and after 5 wk in 500 µm Zn (c).

The populations vary substantially in mean leaf concentration at each of the three external concentrations and PCA1 (Table 2). There is also some variation in the number of plants showing the very low zinc accumulation phenotype at 10 µm, with particularly Schierke (78%) and Schadenbeck (43%) having higher proportions than the other populations. The variation between populations is greater at lower zinc concentrations than at higher. Thus at 10 µm, the between populations variance component is 47% of the value of the within population variance, but at 500 µm it is only 5.6%. Note that the overall coefficient of variation is not substantially different between these two concentrations (10 µm 37%, 500 µm 26%) so that the difference in the relative magnitudes of the between and within population variance components reflects a genuine shift in the partitioning of variation. After plants have been grown for some time in high external concentrations of zinc, there is less difference between populations, but there remains substantial variation between individuals. Fig. 3 shows the relationship between mean population PCA1 and the total soil zinc concentration of the population. There is no significant trend, and populations displaying both extremes of mean zinc accumulation can be found on both contaminated and uncontaminated sites. There is no relationship between PCA1 and other metal contaminants (data not shown).

Table 2.  Mean zinc concentration (µmol Zn fresh weight−1) of 17 populations at three external zinc concentrations, and the first component of a principal components analysis of all three external concentrations (PCA1). Also given at 10 µM is the number of plants in the sample having a value below 9.5 µmol Zn f. wt∠1, together with the sample size (thus 1/15 means that one plant out of 15 had a low zinc concentration). The result of the one-way ANOVA comparing the means is given, together with the magnitude of the between and within population variance components.
PopulationExternal Zn concentration
10 µM 250 µM500 µMPCA1
Pec pod Snezkou14.7(1/15)42.761.0–0.45
Ober schulenberg13.5(2/13)44.365.8–0.23
Velka lypa15.3(1/15)49.566.1 0.11
Schadenbeck 9.3(6/14)37.168.6–1.11
Schat17.1(1/14)53.970.8 0.66
Lange18.5(0/15)54.772.2 0.92
Bechenhaus19.4(0/15)57.272.4 1.11
Schierke 7.3(11/14)36.073.8–1.24
Schwarze Gruber16.0(0/14)42.374.3 0.09
Gose19.7(0/13)54.176.2 1.08
Goslar16.6(0/15)55.176.5 0.81
J-F Halle16.0(1/13)48.777.4 0.44
Ottiliaeschacht15.6(0/13)51.179.8 0.59
Gerade18.4(0/3)46.184.7 0.75
Silbernaal18.6(0/14)51.690.7 1.22
Between populations 7.4***  5.1***  1.8*6.4*** 
Variance components
Between pop10.1 31.5 19.70.47 
Within pop21.3107.9348.51.20 
Figure 3.

Mean population zinc accumulation (PCA1) of 17 populations of A. halleri from central Europe plotted against the mean total zinc concentration (mg kg−1 d. wt) of the soil of origin of the respective population. Solid symbols: metal concentration measured (Table 1); empty symbols: metal concentration not determined, and an estimated value used (500 µg g−1 d. wt uncontaminated sites, 10 000 µg g−1 d. wt contaminated sites).

Experiment 3

An overall anova of this experiment showed that the three populations differed substantially in mean leaf zinc concentration, and there was significant between family, within populations variation at each of the three external concentrations, and for PCA1 (Table 3). Each population was then analysed separately, testing for between family variance, and the between family (σ2B) and within family (σ2W) variance components estimated. σ2B is an estimate of the additive genetic variance: if all the family members are half-sibs, then 4 σ2B/( σ2B + σ2W) is an estimate of the heritability. If, on the other hand, the family members are all full sibs, then the correct formula is 2 σ2B/(σ2B + σ2W) (Falconer, 1989). Since A. halleri is self-incompatible, families will presumably be a mixture of full and half sibs, and so the correct value will lie between these two extremes. Heritabilities must lie between 1.0 (all variation due to additive genetic effects) and zero (all variation due to environmental effects).Table 4 gives the values of heritability calculated in this way for each population and each external concentration, and PCA1. Most of the heritabilities are significant, and have a value of between about 0.25 and 0.5. The heritability of the PCA is slightly higher (0.37–0.74), due to the greater precision of this overall statistic.

Table 3.  Heirarchical ANOVA of plant leaf zinc concentration grown at three levels of external Zn concentrations
 External Zn concentrations
10 µM250 µM500 µMPCA1
  • *

    P < 0.05,

  • **

    **P < 0.01, **P < 0.001. Between 17 and 20 half-sib families (3-five plants per family) from each of three populations were grown.

Population  26.84**4.25*5.23**8.57***
Family 531.69**1.63**1.77**2.06***
Table 4.  Heritabilities of leaf Zn concentration of three populations of Arabidopsis halleri grown at three external Zn concentrations
PopulationExternal NZ concentrations
10 µM250 µM500 µMPCA1
  1. *P < 0.05, **P < 0.01. Heritability lies between 2 σ2B /2B2W) and 4 σ2B /2B2W) (see text). The significance of the heritability determined from oneway ANOVA is also given.

Ottiliaeschacht0.28–0.56*0.24–0.48 NS0.31–0.62*0.28–0.56*
Innerste0.30–0.60*0.08–0.16 NS0.40–0.80*0.44–0.88**
Schadenbeck0.16–0.32*0.37–0.74**0.12–0.24 NS0.39–0.78*

There is a significant relationship between the mean leaf zinc concentration of the families measured in the laboratory and the field leaf zinc concentration of their mothers (Fig. 4). The partial correlation coefficient (i.e. removing the effect of population) between progeny zinc concentration (PCA1) and maternal zinc concentration is 0.25 (P = 0.03, one-tailed test). There is no relationship, however, between the zinc concentration of either mothers or their progeny and the total soil zinc concentration surrounding the mothers’ roots. This is despite there being significant between site variation in total soil zinc concentration (F2,50 = 6.9, P = 0.002), and considerable variation within sites. However, all sites are heavily contaminated, so there were no maternal plants in soils with low zinc levels. Note that the population with the lowest uptake of zinc (Schadenbeck) is not the population with the lowest mean contamination (Innerste has about half the mean contamination of Schadenbeck and Ottiliaeschacht).

Figure 4.

Mean progeny leaf zinc concentration (PCA1) plotted against maternal field leaf Zn concentration (mg kg−1 d. wt) for plants from three German populations: squares and solid line: Innerste; circles and dashed line: Ottiliaeschacht; triangles and dotted line: Schadenbeck. For each maternal plant, the progeny value is derived from the mean of 3–5 daughter plants; the maternal value was determined from leaf material harvested in the field.


I have shown that there is considerable variation both between and within populations in the ability to accumulate zinc under standard laboratory conditions. In experiment 1, plants accumulated zinc steadily for about 7 wk until they had reached a value typical of field collected plants. The rate of accumulation, and final concentration accumulated, was not particularly affected by the external concentration. The plants showed extensive variation in final zinc concentration. In experiment 2, a range of populations were tested for variation in accumulating ability, and substantial variation was found between populations, particularly at low external zinc concentrations. This variation was unrelated to the contamination status of the soil in which the population grew. In experiment 3, a detailed analysis of three populations showed that between plant variation in zinc accumulation was heritable, with heritabilities between 0.25 and 0.5, and that a small but significant correlation could be found between the field zinc concentration of a maternal plant and the accumulating phenotype of her progeny under standard conditions.

These results indicate that genetic variation in zinc accumulation ability exists in A. halleri, and that it most easily manifested at low external zinc concentrations. At higher concentrations, all plants accumulate substantial amounts of zinc, and the variation between plants with different phenotypes is more likely to be due to chance, local environmental effects, or ontological variation than genes. For instance, differences in root : shoot biomass might lead to differences in the overall rate of accumulation, and, if this character is not genetically controlled, would lead to an environmental effect on leaf zinc concentration. Thus, if a breeder wished to select for increased accumulation ability, it should be possible to obtain a substantial response to selection, but it will be easier to select at low external concentrations, and it is likely that a greater response will be seen in plants grown at low external concentrations. However, the strong correlations between accumulating phenotypes at different external concentrations indicate that plants showing a high concentration at low zinc will show an above average phenotype at high external concentrations as well, and so plants selected at such low concentrations should be more effective accumulators at all concentrations. The results in A. halleri are in contrast to those obtained by Pollard & Baker (1996) who found rather little genetic variation, overall, in Thlaspi caerulescens. However, this species is a self-fertilising species, in contrast to A. halleri, and thus one would expect to find less within populations genetic variance.

There is no relation between accumulating ability and degree of contamination of the site of origin of the seed. Thus the results obtained by Bert et al. (2000), in which they found that a population from an uncontaminated site had a higher intrinsic accumulating ability than one from an uncontaminated site, may simply have been due to the chance selection of the particular populations studied. It is of course possible that the populations sampled from uncontaminated sites were transient, and had been recently colonised from more contaminated areas. However, there is no relation even if only the contaminated sites (most of which supported very large populations of this species) are considered. Note that there is no relation, either, between the concentration of zinc in the maternal plants from highly contaminated sites and the total amount of zinc in the soil around their roots. There is however, a detectable effect of genotype on the zinc concentration of plants collected in the field.

So can we answer the question posed in the introduction: what causes the variation in zinc concentration of specimens of A. halleri in nature? I suggested that either soil metal concentration or plant genotype (or both) could have an effect. At high metal concentrations, it is apparent that external metal concentration has rather little effect on the plant metal concentration, but there is detectable genetic variation both within and between populations. Most of the variation, however, is environmental (i.e. random effects, ontological effects and variations in environmental features uncontrolled and unstudied in these experiments). If plants accumulate metal till they reach a plateau (Fig. 1; Zhao et al., 2000) then a lack of a relationship between external and internal metal concentration might be expected. Genetic variation in concentration at high external concentrations indicates that different genotypes achieve different plateaus.

Thus in contaminated soils, the answer is predominantly environmental effects unrelated to metal contamination, with some evidence of a genetic component. At low metal concentrations, however, the effect of genetic variation is much larger, and it is possible to find individuals that have metal concentrations typical of nonaccumulating species, as well as individuals with much higher concentrations. Thus in these environments, low external concentration can lead to low internal concentration, and a plant with a low zinc concentration could be said to have been affected by both external metal and genotype.


I thank the British Council which funded one of the collecting trips, and Vicky Macnair, who assisted in the collecting trips. Dr Mesicek supplied the seed of the two Czech populations.