Variation in arsenic accumulation – hyperaccumulation in ferns and their allies

Rapid report


  • Andrew A. Meharg

    Corresponding author
    1. School of Biological Sciences, University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen, AB24 3UU, UK
      Author for correspondence:Andrew A. Meharg Tel: +44 1224 272264 Fax: +44 1224 272703 Email:
    Search for more papers by this author

Author for correspondence:Andrew A. Meharg Tel: +44 1224 272264 Fax: +44 1224 272703 Email:


  • • A range of fern species (45) and their allies, Equisetum (5) and Selaginella (2) species and Psilotum nudum were screened for their ability to hyperaccumulate arsenic, to develop a phylogenetic understanding of this phenomenon. A number of varieties (5) of a known arsenic hyperaccumulator Pteris cretica were additionally included in this study.
  • • This study is the first to report members of the Pteris genus that do not hyperaccumulate arsenic, Pteris straminea and tremula .
  • • A phylogenetic basis for arsenic accumulation in ferns was investigated. Some orders can accumulate more arsenic than others. Although members of the Equisetales and Blechnales did not hyperaccumulate arsenic, they still accumulated relatively high levels in their fronds, approaching 100 mg kg −1 when grown on a soil dosed with 100 mg kg −1 arsenic.
  • • Arsenic hyperaccumulation was identified as a phenomenon at the extreme range of fern arsenic accumulation. Ferns that exhibit arsenic hyperaccumulation arrived relatively late in terms of fern evolution, as this character is not exhibited by primitive ferns or their allies.


Arsenic can be hyperaccumulated by ferns (Ma et al., 2001; Visoottiviseth et al., 2002; Wang et al., 2002; Zhao et al., 2002). Hyperaccumulating ferns identified to date are all located in the order Pteridales, and include a number of Pteris species and Pityrogramma calomelanos. The hyperaccumulating trait appears to be constitutive, regardless of whether the plants were collected from arsenic contaminated sites or not (Zhao et al., 2002), as demonstrated for other elements which are hyperaccumulated by other plants such as cadmium and zinc (Baker & Whiting, 2002). Arsenic hyperaccumulators differ from plant species that hyperaccumulate other metals in that the arsenic hyperaccumulators have a broad geographical range.

Meharg (2002 ) addressed the evolutionary basis of arsenic hyperaccumulation in ferns, outlining that primitive life forms, including early land flora, could have evolved in an arsenic rich environment. If arsenic hyperaccumlation was a trait deployed to enable survival, or some competitive advantage, in such habitats, Pteris species may have simply inherited this trait from their progenitors, or have been confined to arsenic rich habitats too late on in their evolutionary development. Alternatively, hyperaccumulation may have evolved for other reasons, if the mechanisms involved in hyperaccumulation carry out other, as yet unidentified, arsenic unrelated, physiological functions. Pteris species appeared late on in fern evolution. Screening for more primitive ferns and their allies (Psilotales, Equisetales, Selaginellales) for arsenic hyperaccumulation has been limited, with evidence suggesting that more primitive ferns and their allies than the Pteridales do not exhibit hyperaccumulation ( Meharg, 2002 ). These studies have a limited number of species, grown on a range of substrates, with a range of arsenic levels ( Hozhina et al., 2001 ; Kuehnelt et al., 2000 ; Wong et al., 1999 ; Visoottiviseth et al., 2002 ). Drawing inferences on the evolutionary basis of arsenic hyperaccumulation is difficult because of the nature of these studies, although the studies themselves were conducted for objectives not related to establishing a phylogenetic basis for fern arsenic hyperaccumulation. The study reported here screened a range of ferns and their allies for arsenic hyperaccumulation under standard growth conditions to address the phylogenetic basis of arsenic hyperaccumulation.

Materials and Methods

Ferns and their allies (see Table 1 for full species list) were obtained from a range of sources. They were collected from the wild, obtained from the Cruickshank Botanic Gardens (University of Aberdeen) collection, the Department of Plant & Soil Science (University of Aberdeen) collection, local garden nurseries and from fern specialists. All the ferns and their allies were potted into the same seed and potting compost medium (J. Arthur Bower’s, UK), and allowed to acclimate in a temperate fern house (a nonheated glass house with white washed windows) for at least 2 months. They were then transferred into the same potting medium amended with 100 mg kg−1 d. wt arsenic (as Na2HAsO4), and grown for a further 2 months in the fern-house. Two replicate plants were used for each species.

Table 1.  Frond arsenic levels in ferns and their allies. Each value is the average of two replicate plants
SpeciesVariety familyOrderFrond arsenic (mg kg−1)
Adiantumcapillus-venerusPteridaceaPteridales   2.2
AdiantumpedatumPteridaceaPteridales   7.8
AdiantumfragransPteridaceaPteridales  46.5
AneniaphyllitidisSchizaeaceaeSchizaeales  64.3
AspleniumnidusAspleniaceaeAspidiales   3.9
AspleniumosakaAspleniaceaeAspidiales   1.0
AspleniumscolopendriumAspleniaceaeAspidiales  19.6
AspleniumtrichomanesAspleniaceaeAspidiales   0.7
Athyriumfilix-feminaDryoteridaceaeAspidiales  20.6
Blechnum sp. BlechnaceaeBlechnales   4.4
Blechnumpena-marinaBlechnaceaeBlechnales  24.5
BlechnumspicantBlechnaceaeBlechnales  24.0
ConiogrammaintermediaPteridaceaPteridales   1.4
CystoptopterisbulbiferaDryoteridaceaeAspidiales  11.9
CystoptopterisdickeanaDryoteridaceaeAspidiales   2.4
DavalliatrichomanoidesDavalliaceaeAspidiales   2.0
DoodiacaudataBlechnaceaeBlechnales  24.4
DryopterisaffinisDryoteridaceaeAspidiales   4.4
DryopterisdilitataDryoteridaceaeAspidiales   5.5
Dryopterisfilix-masDryoteridaceaeAspidiales   3.2
DryopterisgoldianaDryoteridaceaeAspidiales   3.4
DryopteriserythrosoraDryoteridaceaeAspidiales   2.2
GymnocarpiumdryopterisDryoteridaceaeAspidiales  14.8
HypolepisdistansDennstaedtiaceaeDennstaedtiales   1.2
MatteuciaorientalisDryoteridaceaeAspidiales   8.6
Nephrolepis sp. NephrolepidaceaeAspidiales   1.2
OnocleasensiblisDryoteridaceaeAspidiales  23.5
OsmundaregalisOsmundaceaeOsmundales   7.7
PellaearotundPteridaceaPteridales   2.5
PhegopterisconnectilisThelypteridaceaeDennstaedtiales   5.1
PlatyceriumbifurcatumPolypodiaceaePolypodiales   4.7
PolypodiumaustralePolypodiaceaPolypodiales   3.0
PolypodiuminterjectumPolypodiaceaPolypodiales   3.0
PolypodiumvulgarePolypodiaceaPolypodiales   5.1
PolystichumaculeatumDryoteridaceaeAspidiales   1.2
PolystichumsetiferumDryoteridaceaeAspidiales   1.9
Polystrichumtsu-sumenseDryoteridaceaeAspidiales   8.3
PolystrichumrigensDryoteridaceaeAspidiales   2.2
PterisargyraeaPteridaceaPteridales 361
Pteriscretica chilsiiPteridaceaPteridales1358
Pteriscretica cristaPteridaceaPteridales1506
Pteriscretica mayiiPteridaceaPteridales1239
Pteriscretica parkeriiPteridaceaPteridales2493
Pteriscretica roweriiPteridaceaPteridales1425
PterisstramineaPteridaceaPteridales  78.0
PteristremulaPteridaceaPteridales  16.6
WoodwardiaradicansBlechnaceaeBlechnales  99
Fern allies
EquisetumarvenseEquisetaceaeEquisetales   7.8
EquisetumfluviatileEquisetaceaeEquisetales  72.7
EquisetumhyemaleEquisetaceaeEquisetales  86.2
EquisetumsylvaticumEquisetaceaeEquisetales  65.3
EquisetumtelmateiaEquisetaceaeEquisetales  12.6
PsilotumnudumPsilotaceaePsilotales   0.4
SelaginellacaulesensSelaginellaceaeSelaginellales  26.3
SelaginellakraussianaSelaginellaceaeSelaginellales   9.0

On the harvest date, three of the most recently unfurled fronds were removed with scissors. The fronds were dried at 70°C. The weighed fronds were then digested with 5 ml of Aristar nitric acid at 120°C using a block digester. When digestion was complete, samples were removed and diluted to 15 ml using distilled water. The sample solution was further diluted 10 fold in a solution of 10% HCl, 10% KI and 5% ascorbic acid. All reagents used for arsenic analysis were purchased from BDH (Poole, Dorset, UK), and were of Analar grade or better. The reduced samples were then analysed for arsenic using hydride generation – atomic absorption spectrometry using a Perkin Elmer (Manchester, UK) FIAS 100 flow injection hydride generator interfaced with a Perkin Elmer AAnalyst 300 atomic absorption spectrometer.


Figure 1 presents ferns ranked in order of increasing arsenic levels in their fronds. These values are also reported in Table 1 . Pteris cretica varieties and Pteris longifolia had the highest levels in their fronds, with arsenic levels exceeding 1000 mg kg −1 d. wt. Other Pteris species, notably Pteris tremula and Pteris straminea had comparatively low arsenic levels, 16.6 (range 15.3–18.9) and 78 (range 59.5–96.5) mg kg −1 , respectively. Pteris argyraea had a concentration of 361 (range 320–402) mg kg −1 .

Figure 1.

Arsenic levels in ferns and their allies ranked in order of increasing arsenic concentration in fronds. Ferns are denoted at the order level, except for members of the Pteris genus and Pteris cretica varieties, which are denoted separately. The data presented are the average arsenic concentration for each species tested.

Members of the order Blechnales, and the fern allies the Equisetales had relatively high levels of arsenic, reaching 100 mg kg−1. These include Woodwardia radicans of the Blechnales, which had the highest level of arsenic (99 mg kg−1) for a nonPteridales, and the Equisetales Equisetum fluviatile, E. hyemale and E. sylvaticum (Table 1).

The lowest levels were recorded in the primitive fern ally Psilotum nudum (0.4 mg kg−1), although in the more primitive Selaginellales, arsenic levels were an order of magnitude higher. In the most primitive fern tested, Osmunda regalis (Fig. 2), levels were relatively low at 8 mg kg−1 arsenic.

Figure 2.

Evolutionary tree for ferns derived from Pryer et al. (1995 ). Branch lengths are arbitrary.

For the most represented order, the Aspidiales with 21 species tested, arsenic levels were low, reaching a maximum of 24 mg kg−1 in Onoclea sensiblis. Levels were also relatively low in the Dennstaedtiales and Polypodiales. Differences between orders on log10 transformed arsenic data were analysed statistically and found to be significant at P < 0.001, both when Pteris species were included and excluded from the one-way analysis of variance. The average arsenic levels per order show that the Aspidiales and Polypodiales have lower arsenic levels than the Blechnales and Equisetales (Fig. 3). When averages were calculated for the Pteridales with and without Pteris species, the value without Pteris species is lower than that for the Blechnales and the Equisetales (Fig. 3). The distribution of log10 arsenic levels in fern species is uni-modal, with Pteris species mainly present on the tail at the high end of the distribution (Fig. 4).

Figure 3.

Shoot arsenic concentrations in ferns and their allies averaged by plant order ± standard error of the mean. The average value for each species was used in this calculation. Twenty-one species were tested for Aspidales, five for the Blechnales, five Eqisetales, four Polypodales 15 Pteridales with Pteris species and Pteris cretica varieties, and five Pteridales excluding Pteris species.

Figure 4.

Distribution of arsenic frond concentrations in ferns and their allies, with Pteris species denoted by white bars. Pteris cretica is given one value (the average of the five varieties tested) in this assessment.


The definition of what constitutes a hyperaccumulator is arbitrary, with thresholds above which a plant is considered to hyperaccumulate set to reflect extreme behaviour (Baker & Whiting, 2002). Arsenic levels exceeding 1000 mg kg−1 in fronds of plants grown on soil containing 100 mg kg−1 of this element are remarkable, and would be considered as hyperaccumulation. The Pteris cretica varieties and Pteris longifolia tested here would be considered arsenic hyperaccumulators. These species have previously been recorded as hyperaccumulators by Zhao et al. (2002), along with Pteris vittata, Pteris umbrosa, and the only non-Pteris hyperaccumulating species identified to date, Pityrogramma calomelanos. Pityrogramma calomelanos is also in the Pteridales, but in the family Hemionitidaceae, unlike the Pteris genus, which is in the Pteridacea (Jones, 1987). The genetic relatedness of the Pteris and Pityrogramma has not been established by molecular techniques. Pteris tremula and Pteris straminea, reported here, would not be considered hyperaccumulators, and are the first members of the Pteris genus reported to not hyperaccumulate arsenic. Thus, not all Pteris species hyperaccumulate arsenic. Non-Pteris members of the family Pteridacea, such as Pellaea rotund, Coniogramma intermedia and members of the Adiantum genus, do not hyperaccumulate arsenic.

For ferns and their allies with frond concentrations above 100 mg kg−1, such as Pteris argyraea, definitions concerning arsenic hyperaccumulating become uncertain. However, frond levels above soil, or approaching levels in the soil of 100 mg kg−1 (as for the Equisetales and Blechnales) are high, up to three orders of magnitude higher than the lowest levels observed (in Psilotum nudum).

Some patterns emerge concerning the phylogenetic basis for arsenic hyperaccumulation in ferns. Fern phylogenies are still not complete given the number of fern species (> 11 000) that require molecular screening (Pryer et al., 1995). The Pteridales evolved quite late, and are part of the same evolutionary branch as Aspidiales, Dennstaedtiales, Polypodiales and Blechnales (Fig. 2), which were found not to hyperaccumulate arsenic in this study. Primitive fern allies, the Selaginellales, Psilotales and Equisetales do not hyperaccumulate arsenic. The most primitive fern tested, Osmunda regalis in the Osmundales, also took up relatively low levels. This suggests that hyperaccumulation is not a trait that has been perpetuated through fern evolution, unless the hyperaccumulating progenitors are extinct. This being said, there are distinct differences in how the fern and their allies accumulate arsenic, with Blechnales and Equisetales accumulating up to 100 mg kg−1 arsenic. This suggests an evolutionary basis for this phenomenon.

Meharg (2002 ) outlines a hypothesis concerning the evolutionary basis of arsenic hyperaccumulation. Land plants that evolved in an arsenic rich environment, such as the environment preserved in the Rhiny cherts ( Rice et al., 1995 ), would have required mechanisms for coping with this element. Hyperaccumulation could have been one of these mechanisms, while excluder mechanisms, such as those described in a range of more advanced plant species ( Meharg & Hartley-Whitaker, 2002 ), provide alternative strategies. If hyperaccumulation was a tolerance mechanism, this mechanism may have been lost as plants spread out from hot-springs and other mineralised environments into non-arsenic contaminated environments. Pteris species, for some reason, may have retained these primeval mechanisms, either as evolutionary baggage or because this trait conferred them with some advantage, arsenic related or not. Alternatively, hyperaccumulation evolved in Pteris species at a later stage in response to a particular selection pressure, or perhaps because of confinement to arsenical habitats until late on in their evolutionary development.

There are still many unanswered questions concerning arsenic hyperaccumulation, both at the mechanistic and evolutionary level (Meharg, 2002). Macnair et al. (1999) showed in Arabidopsis halleri that tolerance and hyperaccumulation were coded by different genes. Arsenic hyperaccumulating ferns must be tolerant to grow on soils or media with high arsenic levels (Zhao et al., 2002). Does arsenic hyperaccumulation confer tolerance? If hyperaccumulation has not evolved as a tolerance mechanism, what ecological purpose does it serve? With the identification that there are members of the Pteris genus that do not hyperaccumulate arsenic, and that there is variation within individual species in the extent of arsenic accumulation/hyperaccumulation, such as for Pteris cretica varieties, some of these ecological and mechanistic unknowns can be addressed. Future mechanistic and ecological studies could compare plants which are closely related, yet vary in hyperaccumulating character. Macnair (2002) showed that there could be considerable variation in the zinc hyperaccumulator Arabidopsis halleri in its ability to hyperaccumulate.

The log10 distribution of arsenic levels in fern fronds appears to follow a uni-modal, with the hyperaccumulators occupying the high end of this distribution. A bimodal distribution is observed for arsenic resistances in grasses (Macnair et al., 1992; Meharg et al., 1993). To confirm if the distribution is uni-modal, more fern species would have to be screened for hyperaccumulation character. If the distribution is uni-modal, this may suggest a multigene basis for hyperaccumulation given the plasticity in accumulation character. A simple genetic basis to hyperaccumulation would, most probably have given rise to a bimodal distribution if hyperaccumulation character was sufficiently distinct. Arsenic levels in fern fronds range over four orders of magnitude, suggesting that if hyperaccumulation character were distinct, it would have been observed.

For arsenic hyperaccumulation to occur; firstly, high levels of arsenic must be obtained from the rhizosphere, given that arsenate, the dominant form of arsenic in aerobic soils, has poor mobility like its analogue phosphate. Secondly, once inside the plant, the arsenic must be translocated through the symplast at high concentrations in a manner that does not disrupt cytoplasmic function, given the considerable phytotoxicity of arsenic species (Meharg & Hartley-Whitacker, 2002). Finally, the arsenic is stored at very high concentrations in the fronds. Each of these three phenomena probably requires at least one separate gene, suggesting that hyperaccumulation is under the control of a range of genes.

Given the right selection pressures, other ferns may have evolved arsenic hyperaccumulation. Again, there are statistically different differences in arsenic hyperaccumulation in different fern orders (Fig. 3), suggesting that the selection pressures leading to fern evolution have caused a differential ability to hyperaccumulate arsenic.

Ferns and, in particular, the fern allies Equistum species, are often reported among the flora of mine spoils and mineralized zones, including those rich in arsenic (Wild, 1974; Wong et al., 1999; Kuehnelt et al., 2000; Hozhina et al., 2001; Visoottiviseth et al., 2002). These environments will contain a range of elements generally toxic to plants, such as copper, zinc, silver and lead. The ability to grow in such metalliferous environments may reflect the environment in which progenitors of ferns and their allies evolved, such as around subaerial hot-springs (Rice et al., 1995). This again suggests that the environment in which these plants evolved may still be reflected in their ability to withstand high levels of metals. Why these traits for coping with metaliferous environments have persisted over the time-scale of plant evolution is another matter. Meharg & Cairney, 1999) argue for trees and ericaceous shrubs and their mycorrhizal fungal associates, many of which can constitutively withstand high levels of metals in their rooting environment, regardless of whether the organisms have been collected from metaliferous environments or not, is because of the harsh environments that many of them naturally colonize, such as low pH and waterlogged soils. The redox stresses and high levels of available aluminium, iron and manganese in such habitats may confer resistances to other metals, which have similar modes of phytotoxic action. Thus ferns and allies that have adapted to harsh habitats may have a similar suite of defenses which confer resistances to metals. For example, Equisetum species colonize harsh, nutrient deficient habitats. However, potential general resistance metals cannot explain arsenic hyperaccumulation.

The term hyperaccumulators has been misused for describing plants with high arsenic levels. Visoottiviseth et al. (2002) state that plants adapted to grow on mine spoil soils hyperaccumulate arsenic. The studies they discuss, such as Porter & Peterson (1975) on the arsenic mine spoils of the Tamar valley SW England, concerning sites with percent levels of arsenic in the soil. For a plant to hyperaccumulate a metal, levels in fronds must be higher than in the soils. Studies based on field collection of samples are also open to misinterpretation, particularly on highly contaminated soils, as external contamination of plant tissues cannot be distinguished from internal plant levels (Reeves & Baker, 2000). Contaminated sites are also heterogeneous with respect to soil levels of toxic metals, ranging over orders of magnitude at small spatial scales. Thus identifying if field contaminated plants were growing in a relatively uncontaminated microsite, but have high levels of external shoot contamination through dust particulates is difficult. By screening plants under controlled and standard conditions, as used here, potential artifacts in identifying hyperaccumulators are overcome.


This present study places arsenic hyperaccumulation within the wider context of arsenic accumulation by ferns and their allies. It is the first study to identify Pteris species which do not hyperaccumulate arsenic. It has also identified a phylogenetic basis for arsenic accumulation in ferns, showing that some orders can accumulate more arsenic than others. Arsenic hyperaccumulation was identified as a phenomenon at the extreme range of normal fern. Ferns that exhibit arsenic hyperaccumulation arrived relatively late in terms of fern evolution.