Author for correspondence: D. R. Parker Tel: +909–787–5126 Fax: +909 7873993 Email:firstname.lastname@example.org
• Ecotypic variation in selenium (Se) hyperaccumulation in plants is reported here among populations of Stanleya pinnata (Brassicaceae), which has a broad biogeographical range in the western USA.
• In a glasshouse study, Se and sulfur accumulation were examined in 16 populations of S. pinnata. Plants grown from seed (collected from sites representing the species range) were subjected to five treatments differing in selenate (SeO42−) and sulfate (SO42−) concentrations.
• The populations differed in shoot Se concentration by 1.4- to 3.6-fold, depending on the treatment, and these concentrations were positively correlated with the indigenous soil Se levels at the collection sites. Shoot S concentrations varied by less than two-fold, and did not correlate with the shoot Se levels. All populations accumulated SeO42− preferentially over SO42−. By contrast, Brassica juncea seedlings grown in a similar solution series consistently accumulated SO42− preferentially over SeO42−. Biomass production differed up to three-fold between populations.
• S. pinnata is a primary Se accumulator, but populations exhibit significant ecotypic differences in Se accumulation. Environmental concerns about Se are common, and the broad adaptation of S. pinnata makes it an attractive candidate for phytoremediation.
In 1831 at Ft. Randal, South Dakota, USA, it was reported that livestock grazing in certain areas would contract ‘alkali’ disease. In the 1920s, high selenium (Se) in forage was found to be the causative agent of alkali disease, thus identifying Se as an ecotoxicological hazard, and spurring numerous investigations into the geobotany of Se. This led to the identification of unique plants, mostly endemic to western USA, capable of (hyper)accumulating Se to thousands (primary accumulators) or hundreds (secondary accumulators) of mg kg−1 (Rosenfeld & Beath, 1964; Mayland et al., 1989).
More recently, disposal of saline irrigation wastewater in hydrologically closed sinks in semiarid western USA has concentrated salts, Se, and other trace elements to levels sufficient to harm wildlife (Presser et al., 1994). A striking example of the ecotoxicology of Se occurred in the mid 1980s at Kesterson Reservoir (Merced Co., CA, USA), where high Se in agricultural wastewater was linked to deformities and deaths of waterfowl embryos (Ohlendorf et al., 1986). Due to the threat it poses to wildlife, methods for removing Se from both contaminated soils and sediments are being sought.
Phytoremediation has been proposed as one solution to this problem wherein plants remove Se, predominantly as selenate (SeO42−), from the soil and accumulate it in their shoots (Banuelos et al., 1990, 1997a; Parker et al., 1991). Seleniferous plant material could be harvested and landfilled, used as a supplement to low-Se animal diets, or incorporated back into the soil to promote microbial Se volatilization (Parker et al., 1991). Plant-enhanced volatilization of methyl-selenide compounds from leaves and/or the rhizosphere has also been proposed to augment removal in harvested shoots (Terry et al., 1992; Zayed & Terry, 1994). The search for superior germplasm is ongoing (Banuelos et al., 1997c), and the ideal candidate plant for Se phytoremediation should be tolerant of the wide range of aerial and edaphic conditions found in western USA, including heat, drought, salinity, and high boron (Parker et al., 1991). Such plants should also have a rapid growth rate and large biomass production, coupled with the ability to accumulate high concentrations of Se in shoot tissue even when the soil is high in soluble SO42−, as is often the case (Bell et al., 1992).
To date, the only primary accumulators evaluated specifically for use in phytoremediation have been species of Astragalus, which have proven difficult to grow, and tend to produce small biomass (Parker et al., 1991; Bell et al., 1992; Duckart et al., 1992; Retana et al., 1993). Other studies have focused on Brassica species (mostly B. juncea and B.napus), which seem to be secondary Se accumulators, probably because they are avid accumulators of S (Banuelos et al., 1997a). Stanleya pinnata, a member of the Brassicaceae, is a previously unstudied candidate that is attractive because it is a putative primary Se accumulator, with a reported maximum of 2380 mg kg−1 Se (Inhat, 1989), it is widespread and broadly adapted in western USA, and it has a perennial habit and potentially large biomass. With respect to Se metabolism, S. pinnata has been shown to be similar to the Se-accumulating Astragalus species, producing the characteristic Se-methylselenocysteine and selenocystathionine (Shrift, 1969). Additionally, S. pinnata produces a seleno-wax unknown in any other species (McColloch et al., 1963).
Bell et al. (1992) described the unique ability of primary Se accumulators to accumulate SeO42−preferentially over SO42−. Sulphur and Se share similar chemistries so that SeO42− and SO42− enter plants via the same carrier and compete strongly for plant uptake and assimilation (Legget & Epstein, 1956). Bell et al. (1992) proposed a method for classifying plants as primary, secondary, or nonSe accumulators based on a discrimination coefficient (DC) derived from the ratio of [plant Se/S] to [nutrient solution Se/S]. Bell et al. (1992) found that the nonaccumulator Medicago sativa (alfalfa) had a DC of <1, indicating discrimination against SeO42−. The primary accumulator Astragalus bisulcatus had a mean DC of 5.43, which indicates significant preferential accumulation of SeO42− over SO42−. The Brassicaceae are well-known accumulators of S (Mengel & Kirkby, 1997), so high shoot Se concentrations reported for S. pinnata may thus be a function of avid S uptake coupled with indiscriminant Se uptake. Alternatively, S. pinnata might be a true primary accumulator that exhibits preferential uptake of SeO42− over SO42−.
Plant populations are classified as ecotypes when they have evolved heritable, phenotypic differences that directly correlate to environmental factors (Turresson, 1922). Baker (1987) and Macnair (1993) have reviewed plant adaptation to high-metal soils, and report that ecotypic variation with regard to tolerance and accumulation of metals is a common phenomenon. Recently, differential cadmium accumulation in populations of Thlaspi caerulescens has been investigated in the context of identifying superior germplasm for phytoremediation (Lombi et al., 2000). To date, there have been no reports of true ecotypic variation in Se accumulation. Due to its extensive range in western USA, S. pinnata offers an opportunity to investigate ecotypic differentiation with respect to soil Se.
The first objective of this research was to determine whether populations of S. pinnata differ ecotypically with respect to Se accumulation, and, thus, to identify populations with greater promise for phytoremediation of Se-laden soils. Secondly, we sought to clarify whether S. pinnata behaves more as a primary or secondary accumulator with respect to preferential uptake of SeO42− vs SO42−. For comparison, we also examined Se accumulating tendencies in a single genotype of B. juncea, a species that has been extensively studied for phytoremediation of Se (Banuelos et al., 1997b,c; de Souza et al., 1998; Pilon-Smits et al. 1999) and other trace metals (Kumar et al., 1995; Salt et al., 1995).
Materials and methods
Site sampling and seed collection
Populations of Stanleyapinnata (Pursh) Britton were found at 25 sites throughout western USA. Green leaf material, soil samples from 0 to 15 cm and 15–30 cm depths, and seeds were collected from each site. From these 25 populations, 15 that represented the widest range in geographical distribution (Fig. 1) and soil and leaf Se concentrations (Table 1) were chosen for use in the glasshouse experiment. One additional population of S. pinnata was obtained from a commercial seed company (Plants of the South-west, Santa Fe, NM, USA), and is designated throughout as Plants of the Southwest (PSW). This population (accession no. P2920) was harvested in New Mexico, but more precise information about location, edaphic, and environmental factors was not available.
Table 1. Topsoil and subsoil Se, field-collected leaf Se, elevation, and geographical coordinates for each field-collected Stanleya pinnata population
Soil Se (mg kg−1)
Seed germination and seedling transplant
Germination towels on trays were dampened with nutrient solution (see below) containing 100 µM Gibberellin A3-acetate (Sigma). Seeds were placed onto the towels and covered with one layer of filter paper. Each tray was placed in a transparent plastic bag, sealed, placed into a growth chamber (16 h day, 8 h night), and kept moist with deionized water. Seeds germinated after 5–50 d. Seedlings with 3–5 cm of shoot and 1–4 cm of root were planted into Sunshine Potting Mix no. 4 (Western Farm Services, Riverside, CA, USA), two seedlings per container (7.5 cm deep × 5 cm wide × 3 cm wide), grown in a glasshouse for 10–55 d, and watered with tap water. When seedlings reached c. 8 cm in height, their growth was slowed by limited root space. Thus, at the time of transplant into sand culture, time from imbibition varied from 20 to 60 d, but seedling size did not vary.
Plants were transplanted into 7.5-l pots each filled with 10 kg of deionized-water-rinsed silica sand. The root ball of the plant was cleaned of potting mix by rinsing in tap water and manually stripping mix from the roots. Due to the large number of plants, transplanting spanned 6 d. Initially, each pot contained eight plants, but plant density was later thinned to five plants per pot. A reservoir containing 120 l of nutrient solution supported four pots. The basal nutrient solution contained (in mM) 1 NH4NO3, 1 CaCl2, 0.25 KCl, 0.1 MgSO4, and (in µM) 10 NaH2PO4, 10 Fe EDTA, 1 MnCl2, 1 ZnCl2, 0.1 CuCl2, 0.1 NaMoO4, and 3 H3BO3 (Parker et al., 1991). Nutrient solution was automatically siphoned to the pots 5 times d−1, with excess solution draining back into the reservoir (Parker et al., 1991). Water lost to evapotranspiration was replaced using deionized water. Solution P levels were checked periodically and replenished as necessary. Residual carbonates in the sand buffered the nutrient solution at a pH of 7.55 throughout the experiment. Average night and day temperature was 18.3°C and 25.6°C, respectively, with a high of 38.3°C and a low of 13.3°C. PPFR (400–700 nm) was measured on 3 d at solar noon, and averaged 1035 µmol m−2 s−1.
Experimental design and treatments
After 15–20 d of plant establishment in the sand culture, treatments were imposed. The day before treatment initiation, the pots were thinned to five plants per pot. The culled plants were counted, dried, and weighed to determine biomass at the time of treatment imposition (which ranged from just 2–3.5% of the final harvest weights). There were five treatments with varying concentrations of Se and S as Na2SeO4 and Na2SO4, respectively. Selenate concentrations ranged from 2 to 80 µM, while sulphate concentrations ranged from 0.5 to 15.5 mM (Table 2). Treatments 1–5 are identical to treatments 3, 5, 8, 10, and 11, respectively, in Bell et al. (1992). Each of the 16 populations was subjected to the five treatments in a randomized block design, replicated four times. There was no treatment 2 for population MT1 because there were not enough seedlings for all five treatments. The 16 populations were randomly grouped into four groups of four, such that each grouping always contained the same four populations sharing a common reservoir. After transplanting, the plants grew for a total of 54–60 d before harvest. Shoots were then cut at sand level and frozen in plastic bags. Solution samples from each tank were collected immediately after treatment imposition and on the day of harvest.
Table 2. Initial Se and S concentrations in the nutrient solutions, molar ratios of Se : S for each treatment, percent depletions of Se and S from the nutrient solutions, and the depletion that can be accounted for by the total shoot Se content
Molar ratio Se : S
Se depletion due to shoot Se
Analysis of plants, soils, and solutions
Plants were freeze-dried, weighed, and ground to pass a 0.43 mm mesh. Soil samples were air dried and pulverized in an Agnstrom mill. Plant material was microwave-digested by first combining 250 mg plant material, 1 ml deionized H2O, 2 ml 30% (v/v) H2O2, and 2 ml concentrated HNO3 (Sah & Miller, 1992). Soil digests consisted of 300 mg pulverized soil, 4.5 ml concentrated HNO3, 1.5 ml HCl, and 1 ml deionized H2O (Milward & Kluckner, 1989). Digestions were carried out in 110-ml Teflon-lined vessels, and heated for 20 min at 570 W in a microwave oven. For plant Se analysis, appropriate subsamples were diluted in 6 M HCl, heated for 20 min in a 90°C water bath (Zhang et al., 1999), and immediately analysed by hydride vapour-generated atomic absorption spectroscopy (HVG-AAS). For soil Se analysis, subsamples and 0.2 ml of 0.2 M K2S2O4 (Zhang et al., 1999) were diluted to 50 ml with 6 M HCl, heated at 92–95°C for 1 h, and analysed by HVG-AAS. For S analysis of the plant shoots, suitable dilutions were made with 250 µg l−1 Yttrium as an internal standard, and were analysed using inductively coupled plasma optical emission spectroscopy (ICP-OES). Nutrient-solution samples taken at harvest were similarly analysed for Se by HVG-AAS and for S by ICP-OES. The data were analysed using standard ANOVA procedures (SAS Institute, 1985) to assess the significance of the treatment effects.
Brassica juncea experiment
In a separate experiment, seeds of Brassica juncea (L.) Czern. (PI 426308) were obtained from USDA North Central Regional Plant Introduction Station, Ames, IA, USA. Plants were grown hydroponically in a growth chamber using treatments 3, 5, and 8–11 exactly as described in Bell et al. (1992). Plants grew for 18 d after transfer of seedlings into 7.5-l pots. Solutions were completely replaced four times to maintain solution [SeO42−] and [SO42−] within ±10% of target values. Plants were harvested and analysed for Se and S as described in the previous section.
Results and Discussion
During collection, S. pinnata was found prospering in heat, drought, and intense sunlight. The species seems to thrive in erosive and/or disturbed areas such as badlands and roadcuts, and was found growing on wide variety of soil types, ranging from loose sand to packed gravel/shale to dense clays. The geographical range of the populations was wide (Fig. 1), and elevations ranged from 660 to 2270 m. Individual populations tended to be small (usually no more than a few hectares) and disjunct, factors known to contribute to local adaptation (Baker et al., 1994), and were interspersed with other plant species.
Soil Se varied from <0.1–9.3 mg kg−1 (Table 1), and about half of the 15 used in this study were seleniferous (Se ≥ 1.5 mg kg−1) (Elrashidi et al., 1989). Leaf Se ranged from <1–1200 mg kg−1 (Table 1). Leaf Se and topsoil Se from all 25 sites showed a significant positive correlation (leaf Se = −30 + 90 × soil Se; r = 0.80, P < 0.001), and a similar relationship existed for the 15 populations used in the glasshouse study (not shown).
Averaged across the five Se : S treatments, there were significant differences in biomass production between the 16 populations when grown in the glasshouse (Fig. 2). Population CO4 produced the highest yield (31 g per pot), while the biomass of WY2 (10 g per pot) was significantly smaller than that of any other population; the overall mean biomass production was c. 20 g per pot. The adaptive value of this size variation under field conditions is unknown, but it does suggest genetic differences in growth potential between the populations. Variation in leaf morphology was also apparent, although not readily quantifiable, and S. pinnata thus shows the polymorphism often noted in hyperaccumulating taxa (Baker et al., 1994).
For all populations except PSW, UT1, and UT6, there were no significant differences in biomass production between the five treatments (data not shown). With PSW and UT1, biomass production under treatment 3 was c. 10% less than average, while a similarly low value was obtained with UT6 under treatment 1 (data not shown). Although statistically significant, these differences are clearly minor and, overall, plant growth was neither stimulated nor inhibited by the Se and S concentrations used in this experiment.
Shoot Se concentrations
Fig. 3 depicts the shoot Se concentrations for each treatment and population, with the populations ranked according to the treatment 4-values. This ranking is useful because it emphasizes the similarity in trends between treatments 3–5, and, to a lesser degree, treatment 2 (Fig. 3). With treatment 1, few significant differences were found between any of the populations, though CA1, CA2, NV1, and PSW were slightly lower than average (Fig. 3a). Shoot Se concentration ranged from 320 to 460 mg kg−1, with error bars overlapping for most treatments (Fig. 3a). All populations were equally able to concentrate Se at the high Se/S ratio (0.012) of this treatment, possibly due to less competition from sulphate. The uniform Se concentrations may also be due to the approx. 95% depletion of solution Se in this treatment by the time of harvest (Table 2 and below).
Treatment 2 had both the lowest solution Se (2 µM) and the lowest solution Se/S ratio (0.0008), and thus the greatest competitive inhibition of SeO42− uptake by SO42−. As a result, plants in this treatment took up very little Se (shoot Se = 30–90 mg kg−1) compared to the other treatments (Fig. 3).
Within a population, treatment 4 always had the highest Se concentration (Fig. 3b), probably because this treatment had the highest solution Se (80 µM). Because factors governing plant Se uptake are complex, treatments 3 and 5 produced very similar shoot Se levels (Fig. 3b). While treatment 3 has only one-third the solution Se of treatment 5, the Se/S ratio of treatment 3 is twice that of treatment 5 (Table 2). Thus, in treatments 3 and 5, the initial solution Se and the Se/S ratios seem to offset in their effect on plant Se accumulation.
Treatments 3–5 produced nearly identical trends in shoot Se concentration across the populations (Fig. 3b), even among those with lower accumulation. A Spearman’s analysis of rank correlation (Steel & Torrie, 1980) of population Se concentration within each treatment revealed that population rankings for treatments 3–5 all corresponded highly (rs ≥ 0.95), and that treatments 2–5 all corresponded reasonably well (rs ≥ 0.78). This consistency in rankings strengthens the notion that the populations are responding differentially to the available selenate as mitigated by the competing sulphate ion. Unsurprisingly, the ranking of populations within treatment 1 did not correspond well with the other treatments’ rankings (rs ≤ 0.27; Fig. 3).
Analysis of the nutrient solutions at harvest revealed that Se depletion ranged from 95% in treatment 1 to 36% in treatment 5 (Table 2). Shoot accumulation of Se accounted for only 59% or less of the total depletion (Table 2). The larger, remaining fraction of solution Se depletion can be ascribed to Se accumulation in the roots and/or to Se volatilization (Terry et al., 1992), neither of which were measured in this experiment. However, the characteristic Se-odour described by Rosenfeld & Beath (1964) was apparent during the glasshouse experiment, suggesting that substantial volatilization of Se occurred. Moreover, if one assumes a typical root : shoot Se ratio of ≤0.5 coupled with the observation that Se accumulators such as A. bisulcatus often have somewhat greater shoot than root Se concentrations (Rosenfeld & Beath, 1964; Bell et al., 1992), it seems very unlikely that the ‘missing’ Se in Table 2 is predominantly root Se. These results suggest that S. pinnata may be capable of volatilizing large quantities of Se, a subject of ongoing study in our lab.
Within a given Se treatment, depletion of solution Se in the four population groupings did not differ (data not shown), which indicates that all the plants in a given treatment were exposed to the same Se level. Thus, differences in Se uptake between populations within treatments were due to actual rates of Se uptake, and not to any differences in solution Se concentration between the population groupings.
Shoot S concentrations for each population are presented in Fig. 4, and are presented in the same order as Fig. 3 (i.e. ranked by the treatment 4 shoot Se concentration). Treatments 2–5 were averaged because, within a population, the variation between them was inconsequential (note small error bars in Fig. 4). Treatment 1 had the lowest solution S, and the corresponding plants had shoot S concentrations lower than those in treatments 2–5, with the exception of PSW (Fig. 4). The reason for the variation in PSW is unknown. Clearly, S accumulation did not follow the same trend as Se accumulation (compare Figs 3, 4), and there is no correlation between shoot Se and S concentrations. Within a given treatment, populations varied in S concentration by less than twofold (Fig. 4) which is significantly less than the variation in shoot Se, where populations differed by as much as 3.6-fold (Fig. 3b, treatment 5). Depletion of nutrient solution S was ≤22% in all treatments except treatment 1, and was substantially less than the corresponding depletions of Se (Table 2).
Treatments 2 and 3 had identical solution S initially (Table 2), and most S. pinnata genotypes showed no difference in shoot S concentration between these treatments (data not shown), despite a 10-fold difference in solution Se (Table 2). Thus, Se does not appear to stimulate S accumulation in the shoots of S. pinnata. This finding contrasts with results using a number of other species (Bell et al., 1992, and references therein), and with B. juncea (compare the second and third treatments in Table 3).
Table 3. SeO42− and SO42− treatments, shoot Se and S, and discrimination coefficient (DCi) values for the solution culture experiment with Brassica juncea
Treatment SeO42− (M)
Shoot Se (mg kg−1)
Shoot S (g kg−1)
Values are means, SE (in parentheses); n = 3.
The S concentrations for S. pinnata ranged from 12.3 mg g−1 (CA1, treatment 1) to 31.6 mg g−1 (MT1, treatment 5) (Fig. 4). In the companion experiment, B. juncea accumulated S to between 13.1 and 23.8 mg g−1 (Table 3). The Brassicaceae often accumulate in excess of 1 mg g−1 (Mengel & Kirkby, 1987), so the high S concentrations in S. pinnata and B. juncea are not surprising. The tendency for Se-accumulating plants also to take up S is further evident in A. bisulcatus, which can take up S to at least 17 mg kg−1 (Bell et al., 1992).
Primary accumulation: Se concentration and discrimination coefficients
The WY2 population was able to concentrate Se to 1130 mg kg−1 in treatment 4. When grown at the same selenate and sulphate levels, Astragalus bisulcatus had a shoot Se concentration of 1160 mg kg−1 (Bell et al., 1992). Thus, at least one S. pinnata population was found to concentrate Se to the level of a well-known primary accumulator. In the other treatments, the WY2 shoot Se concentrations ranged from 47% to 80% of those previously found in A. bisulcatus. It should be noted that Bell et al. (1992) used frequent solution changes to maintain Se concentrations within 5–10% of the target values. In contrast, the solution Se concentrations here were sufficiently depleted that the shoot Se concentrations of S. pinnata were probably lower than if the target solution [Se] had been more assiduously maintained. In the companion experiment with well-maintained solution [Se] and [S], B. juncea shoot Se ranged from 16 to 320 mg kg−1 (Table 3), thus attaining Se concentrations that ranged from 7% to 60% (average: 22%) of those found previously in A. bisulcatus (Bell et al., 1992).
A discrimination coefficient (DC = [Se/S]plant/[Se/S]solution) indicates the ‘preference’ of a plant for Se or S during uptake and subsequent metabolism (Bell et al., 1992). Treatment 1 was omitted from the calculation of DC for S. pinnata because of excessive solution depletion of both Se and S (Table 2). The DC values based on initial solution Se/S (DCi) indicate that, as a species, S. pinnata accumulates Se preferentially over S. The average DCi for most populations was about 2, with only PSW and MT1 bordering on 1 (Table 4). The CO4 and ND1 populations had mean DCi values > 2.5, and WY2 had a mean value of 4.39 (Table 4) which is similar in magnitude to the mean DC of 5.43 (whole-plant basis) reported by Bell et al. (1992) for A. bisulcatus. The values in Table 4, thus, offer further evidence that S. pinnata can be classified as a primary Se accumulator as proposed by Bell et al. (1992).
Table 4. DCi (SE) and average DCi based on initial solution [Se/S] for each Stanleya pinnata population for treatments 2–5, and average Se content per pot for each population (based on shoot Se and biomass)
It is important to note that the relative depletions in solution Se vs S concentrations (Table 2) would, if anything, have caused an underestimation of the ‘true’ DC values. The actual [Se/S]solution values of all the nutrient solutions decreased over the course of the experiment. Thus, the DCi values (Table 4) calculated based on the initial solution composition (Table 2), are lower than the ‘true’ DC values which would instead reflect the time-averaged values of [Se/S]solution. Assiduous maintenance of [Se/S]solution was not a goal of this large experiment, nor feasible, but S. pinnata would undoubtedly yield higher DC values if exposed to more constant [Se/S]solution values.
To date, S. pinnata is the only Brassicaceae that exhibits the metabolic traits of a primary Se accumulator (Shrift, 1969). In the study by Bell et al. (1992) reanalysis of the literature, Brassica species had mean DC values of ≤1.7. In our solution-culture experiment, B. juncea and had very consistent DC values of approx. 0.6 (Table 3), which are quite similar to those found previously in Medicago sativa (Bell et al., 1992). Thus, B. juncea preferentially accumulated S over Se and, together with the shoot Se concentrations discussed above, this result is consonant with B.juncea’s classification as a secondary Se accumulator. Even with well-maintained SeO42− and SO42− solution concentrations, shoot Se in B juncea reached only 13–77% (average = 37%) of the levels in the corresponding treatments with S. pinnata, population WY2.
Ecotypic variation in Stanleya pinnata
This is the first report of ecotypic variation in Se accumulation. Under the controlled conditions of sand culture, the S. pinnata populations grown in treatments 2–5 displayed consistent patterns of Se accumulation (Fig. 3), as well as obvious differences in their tendency to accumulate Se. Moreover, the glasshouse shoot Se concentrations for treatments 2–5 were positively correlated with the soil concentration of Se at the collection site for each population. Treatment 5 produced the strongest such correlation which is shown in Fig. 5. The four highest-accumulating populations of S. pinnata were collected from the four soils highest in Se. These populations, especially CO4 and WY2, seem to have evolved a superior ability to accumulate Se which distinguishes them from other ecotypes of S. pinnata.
Because S. pinnata occurs on both seleniferous and nonseleniferous soils (17 of the 25 original sites had soil Se <1.5 mg kg−1 in our study), this species is analogous to the ‘pseudometallophytes’ described by Baker & Proctor (1990). Its tendency to accumulate Se, however, seems to be largely constitutive in that this trait is expressed to a relatively high degree in all populations. The ‘superior’ Se-accumulating ecotypes may have evolved as a direct result of the differential phytotoxicity between seleniferous and nonseleniferous soils. However, adaptation to Se could be driven by other factors, as has been proposed for nickel accumulators. Plants with high shoot Ni concentrations have the ability to deter both herbivores and pathogens (Boyd et al., 1994; Boyd & Moar, 1999). High levels of shoot Se might have similar effects (Vickerman & Trumble, 1999), thus conferring a competitive advantage. Further studies are underway to understand better the evolutionary ecology of this unique species.
Phytoremediation potential of Stanleya pinnata
Each population’s biomass production (Fig. 2) and Se concentration (Fig. 3) were multiplied to calculate the mass of Se accumulated by the shoots in each treatment. Table 4 shows the results for treatment 5, which had the largest (3.6-fold) variation in Se accumulation between populations. The overall average for treatment 5 was 8.5 mg Se per pot, with a range from 4.5 (UT5) to 26 (CO4) mg Se per pot (Table 4). Although WY2 had the highest concentration of shoot Se (Fig. 3b), it produced the lowest biomass of any population (Fig. 2), and its Se content was, thus, about average (Table 4). Thus, the CO4 population is presently the most promising source of material for developing an S.pinnata genotype useful for phytoremediation purposes. The commercially available PSW accession had below-average shoot Se concentrations (Fig. 3), DC values (Table 4), and mean Se content (Table 4). This emphasizes the importance of investigating a number of naturally occurring populations when searching for superior phytoremediation candidates.
To date, S. pinnata and Astragalus species are the only primary Se accumulators that have been studied as phytoremediation candidates. Greater biomass potential (the largest field specimen found was c. 1.5 m high × 2 m wide), broader adaptation, and relative ease of cultivation are the main advantages of S. pinnata over Astragalus species. The advantages of S. pinnata over previously studied Brassica species, are preferential uptake of Se over S such that greater shoot concentrations of Se are attained, and indications that its metabolism is similar to A. bisulcatus, and that copious volatilization of Se from leaves may occur in S. pinnata.
Recently, Pilon-Smits et al. (1999) genetically modified a line of B. juncea so that it overexpresses ATP-sulphurylase and, probably as a result of an increased ability to accumulate S, is consequently able to take up and volatilize more SeO42− than wild types. As opposed to primary accumulation, this amplification of secondary Se accumulation presents an alternative means to achieve removal of Se from soil. However, this approach does nothing to mitigate the competition between S and Se, both for uptake and for metabolism, including volatilization of methyl selenides.
Stanleya pinnata is native to western and mid-western USA (Rollins, 1939), and may also be native to specific locales where Se contamination is presently occurring. Thus, S. pinnata’s broad adaptation to the climates and soils of western USA make this plant an ideal candidate for Se phytoremediation. The use of native plants for phytoremediation may also be a desirable management goal, in order to keep non-native plants out of already disturbed areas.
Further investigation is needed to assess S. pinnata’s germination rate, establishment, growth, and Se uptake under field conditions. Also, this species needs to be evaluated for tolerance to B and salinity (Parker et al., 1991). The rate at which this plant volatilizes Se is of special interest in light of the Se depletion from the nutrient solutions unaccounted for by tissue Se concentrations (Table 2). Finally, a comparative study of Se metabolism in S. pinnata, A. bisulcatus, and Brassica species would provide insights into the mechanisms, genetics, and evolution of Se accumulation.
We thank Merrick Myers, Judith F. Pedler, Tham Nguyen, David N. Thomason, Laska Whiteaker, and Yiqiang Zhang for their invaluable assistance in the glasshouse and laboratory phases of this study.