Introduction

In serpentine environments, plants have developed different strategies to avoid Ni-toxicity and can be classified into three different groups depending on their type of strategy: excluders, indicators, and hyperaccumulators (Baker 1981; Tomsett and Thurman 1988; Verkleij 1990). Ni-hyperaccumulating plants are characterized by their ability to accumulate more than 1000 μg g−1 (dry weight) of Ni in their leaves.

For a long time, Ni-hyperaccumulating plants have been considered as non-mycorrhizal species (Leyval et al. 1997; Pawlowska et al. 2000; Coles et al. 2001). Indeed, in the past, studies of Ni-hyperaccumulators have concerned mostly non-tropical plants, especially Brassicaceae which are known as non-mycorrhizal (Smith and Read 1997; Wang and Qiu 2006). Recently, Turnau and Mesjasz-Przybylowicz (2003) have shown Berkheya coddii and three other Ni-hyperaccumulating Asteraceae from South Africa to be consistently colonized by arbuscular mycorrhizal fungi (AMF).

In New Caledonia, one of the floristically richest serpentine environments in the world, over 50 species of Ni-hyperaccumultating plants have been reported (Jaffré 1980). Their mycorrhizal status remains almost unknown: the only reported study concerned AMF colonisation of Phyllanthus favieri (Euphorbiaceae) a strong Ni-hyperaccumulator (Perrier et al. 2006). Some of these species are relatively abundant (Jaffré and Veillon 1991) and could be used in restoration programs.

New Caledonian ultramafic soils are characterized by P, N, K and Ca deficiencies (Brooks 1987). Despite their high metal contents, AMF are abundant in these soils and seem to play an important role in the adaptation of plants to this environment (Amir et al. 1997; Perrier et al. 2006). In other respects, it becomes more and more clear that, in a metal rich substrate, AMF can reduce the concentration of Ni in shoots and thereby increase metal-tolerance of plants (Heggo and Angle 1990; Hilderbrandt et al. 1999; Joner and Leyval 2001; Vivas et al. 2005).

When one consider the aforementioned elements, the relationships existing between Ni-hyperaccumulation by certain plants and their mycorrhizal status in ultramafic soils can be questioned. It seems clear that analysis of these relationships could help us to understand plant adaptation to toxic metalliferous soils. Several fundamental questions on these points include: are New Caledonian Ni-hyperaccumulators generally mycorrhizal? Is mycorrhizal colonisation of strong hyperaccumulators as consistent as that of weak accumulators? Does the mycorrhizal status of each Ni-hyperaccumulating species vary depending on Ni concentration in leaves or soil? If there is an inhibition of mycorrhizal colonisation, how can it be explained? If AMF are present in Ni-hyperaccumulator roots, are they more tolerant to Ni than other AMF isolates from ultramafic soils?

To address these questions, nine Ni-accumulators from New Caledonia were studied and the relationships between Ni-hyperaccumulation and different mycorrhizal parameters investigated.

Materials and methods

Study sites and studied plant species

Nine plant species with different levels of Ni-accumulation were chosen. They were selected because they are easy to collect or because they are strong hyperaccumulators. They were sampled in 3 different sites, located in Rivière Bleue, Mont Koghis and the Koniambo Massif. The average annual temperature of these sites ranges between 18°C and 21°C, with a maximum of 30°C and a minimum of 10°C.

The Rivière Bleue site is a tropical humid forest in the south of the main island (60 km east of Noumea), with an altitude of 160 m and an annual precipitation of 3,180 mm. Samples were collected at points located within 3–4 km of the «grand Kaori». The main soil studied is an alluvial lateritic soil (oxisol), relatively deep, lying on peridotite.

The Koghis site is a tropical humid forest located 20 km north of Noumea. The altitude of the site is 400 m, and average annual precipitation 1,750 mm. The sampling points are about 300 m from the restaurant «Auberge du mont Koghis». The soil is a brown hypermagnesian clayey soil formed on serpentinised peridotite.

The third site, in the Koniambo Massif (in the north of the main island), is at an altitude of 710 m and has annual precipitation of 1,800 mm. The sampling site is located in a tropical humid forest dominated by Nothofagus balanse. The soil is a colluvial lateritic soil (oxisol) lying on unweathered peridotite.

Some chemical characteristics of soil samples of the three sites are shown in Table 1. The plant species sampled in each of the three sites and their ability to accumulate Ni are presented in Table 2.

Table 1 Some chemical characteristics of representative samples of the studied soils
Full size table
Table 2 List of the studied plant species and their ability to accumulate Ni
Full size table

AMF isolates used

Five Glomus sp. isolates, all from New Caledonian soils, were used for studies on spore germination in relation to Ni-hyperaccumulation. FSCtAc was isolated from a dry forest non-ultramafic soil collected at Pointe Maa (just north of Noumea). T2R3 and SFONL were isolated from lateritic ultramafic soils collected respectively in Koniambo massif and Ouenarou (near Rivière Bleue). «Psychot1» and «Sebert1» were isolated from roots of Psychotria douarrei and Sebertia acuminata at Rivière Bleue. The two latter isolates were used to check if their Ni-tolerance is higher than three isolates collected under non-hyperaccumulating plants. FSCtAc, (non-ultramafic isolate) was used in all spore germination experiments because of its relatively low Ni-tolerance.

General approach

During the first stage, 8–10 individuals of each of the 9 plant species were sampled and characterised for the following parameters: total Ni content in leaves, Ni-DTPA in soil under plant, type of mycorrhizae, mycorrhizal frequency and mycorrhizal density.

During the second stage, one moderate Ni-hyperaccumulator (Geissois pruinosa) and 3 strong hyperaccumulators (Phyllanthus favieri, Psychotria douarrei and Sebertia acuminata) were studied in more detail. In this case, 18–40 individuals (depending on the difficulty to find and collect the roots) were sampled for each species. The same parameters as above were measured; but we also documented the arbuscule frequency and the number of viable spores in the soil under each plant. A statistical analysis of the correlations between these different parameters was then carried out.

During the third stage, AMF spore germination in contact with sterile sand containing different concentrations of Ni, or in contact with soils or root crush of Ni-hyperaccumulating plants was assessed to check if there was inhibition of spore germination in these conditions.

Root sampling and assessment of mycorrhizal colonisation

Depending on the objective (see above) 8–40 haphazardly chosen individuals of each plant species were selected per site. The area explored for each site and each species was about 1 ha and the distance between the individuals was 10–30 m. The fine roots were sampled and placed in plastic bags. To avoid confusion between the roots of different species, the plants were dug up if they were small and the roots were followed from the main root to the finer roots, which were sampled; then the plant was replanted. For bigger plants, the roots were traced carefully, from the main root to the finer roots. The roots were kept refrigerated until treatment. In the laboratory the roots were prepared following Koske and Gemma (1989) procedure. After being thoroughly rinsed with water, they were immersed in a 10% KOH bath at 90°C for 90 min, a few drops of H2O2 were added after 1 h to lighten the colour. They were then rinsed again and stained by immersion in a trypan blue bath for 15 min at 90°C. For each sample, 30 haphazardly selected stained segments (approximately 1 cm long) were observed under the microscope. A root was considered arbuscular mycorrhizal when arbuscules, vesicles or hyphal coils were present. Mycorrhizal frequency percentage (F) mycorrhizal density percentage (M) and arbuscule density percentage (A) were determined according to the Trouvelot et al. (1986) method.

Soil and leaf sampling, and nickel analyses

Soils under plants were sampled in each of the 3 sites. The surface under each plant was scraped to eliminate the litter, and the soil was taken from a 15 cm deep top layer. In a few cases, soil was also taken 10 m away from the plant to avoid the influence of Ni-rich litter. After a thorough mix, all samples were sieved through a 2 mm mesh before treatment. For each sampled plant, a few healthy representative leaves were collected on three different branches using an automatic CHNOS analyser (Thermoquest Finnigan). All chemical analyses were performed at the chemistry laboratory of IRD Noumea. The leaves were washed thoroughly with water and then dried under neon light. Total C and N were measured.

The total Ni content in leaves (expressed in μg g−1 of dry leaf) and the diethylene triamine pentacetic acid (DTPA) extractable Ni in soils (expressed in μg g−1 of dry soil) were performed following the Lindsay and Norvell (1978) method, by measurement with an ICP-OES (Inductively Couple Plasma—Optical Emission Spectroscopy).

AMF spore counts

The number of viable AMF spores present in each soil sample was determined after wet-sieving and centrifugation. The spores were extracted from three 50 g aliquots of dry soil. The aliquots were sieved through 1 mm and 38 μm sieves. The fraction collected between the two sieves was centrifuged for 5 min at 1,000 g. The pellets were then resuspended in 50% sucrose and centrifuged for 1 min at 1,000 g. AMF spores were then recovered from the supernatant, which was poured through a 38 μm sieve, thoroughly rinsed with water and retransfered to filter paper for counting. Only spores filled with intact cytoplasm and which were not black were recorded.

Bioassay of Ni-tolerance of AMF spore germination

Aliquots of 30 g of fine sand autoclaved 1 h at 120°C were placed into 5 cm diameter Petri dishes. A cellulose filter membrane (0.45 μm, 47 mm diameter) was applied on the surface of the sand of each plate. Eighty AMF spores obtained after wet sieving and centrifugation (see above), were placed on the membrane. Distilled water or Ni solution (NiSO4) was then added to the sand until saturation caused the membrane to stick to the sand by water capillarity. Four Ni concentrations were used: 0, 10, 20, 30 μg g−1. Three replicates of each treatment were prepared. The Petri dishes were then sealed with parafilm and incubated 30 days at 25°C. The percentage of germinated spores was determined at 45× magnification. The same technique was used to test the influence of soils on spore germination but the sand was replaced by the tested soil. In another experiment, roots of a Ni-hyperaccumulating plant were crushed roughly with a hand grinder; then 1 g of root crush was spread on the surface of a fine layer of sand just under the cellulose membrane.

Data analysis

Two types of statistical analyses were performed using the computer program Kyplot v2.0 beta 13. To study the relationships between Ni-accumulation and mycorrhizal status, total correlations between Ni-accumulation parameters and mycorrhizal parameters were analysed. ANOVA (Newman-Keuls test) was used to compare values of different treatments, especially for AMF spore germination data.

Results

Mycorrhizal status of the 9 studied plants in relation to Ni in soil and leaves

The nine studied Ni-hyperaccumulating plants were characterised by very different Ni contents in leaves (Fig. 1a). The lower values (for Pancheria alateroides and Cloezia artensis) were under the minimal level which define Ni-hyperaccumulation, whereas the higher values (Psychotria douarrei and Sebertia acuminata) exceeded 2.2%. The DTPA-extractable Ni concentrations in soil under plants (Fig. 1b) were lower than 50 μg g−1 for the 2 weak accumulators, and varied from 84 μg g−1 to 326 μg g−1 for the moderate hyperaccumulators. Among the four strong hyperaccumulators, only Phyllanthus favieri had a relatively moderate level of DTPA-Ni in soil under plants (132 μg g−1); the values measured for the others were above 750 μg g−1.

Fig. 1
figure 1

Relationships between leaf Ni content, DTPA-Ni in soil under plants and mycorrhizal intensity (M) of the 9 studied species (means of 8–10 samples for each value); G = Geissois. The coefficient r indicates the total linear correlation between the three parameters; r (a, b): correlation between Leaf Ni content and DTPA-Ni; r (a–c): correlation between Leaf Ni content and M; r (b, c): correlation between DTPA-Ni and M; ** = significant at P < 0.01; NS = non significant

Full size image

A highly significant correlation (r = 0.823) between Ni content in leaves and DTPA-Ni in soil under plant was recorded. The roots of the 9 plant species were all more or less colonised by AMF (Fig. 1c). No other types of fungal symbiots were observed. Mycelium and vesicles were present. Arbuscules and coils were generally present in the same root fragments. Values of root colonisation by AMF were high for the weak and moderate accumulators (>40%), and very low (<12%) for the strongest hyperaccumulators. The negative correlation between M and Ni content in leaves was highly significant (r = 0.886), whereas the correlation between M and DTPA-Ni in soil under plant was not significant but close to the significance level (P = 0.06).

Correlations between mycorrhizal parameters and Ni-hyperaccumulation for Geissois pruinosa

The values of Ni content in leaves of G. pruinosa were very variable depending on the Ni concentration in soil (Table 3). Indeed, a few soil samples did not contain detectable extractable Ni, whereas others contained more than 600 μg g−1 Ni. Root colonisation by AMF was also quite variable: M varied from 8.5% to 95%. Arbuscules were generally present but not frequent. An average number of 57 AMF spores per 100 g of soil under plant was recorded.

Table 3 Variation of Ni-hyperaccumulation in Geissois pruinosa (25 samples), DTPA extractable Ni in soil under plant and mycorrhizal parameters of the plant
Full size table

The correlation between M an leaf Ni content was negative and highly significant, whereas the correlation of M with the DTPA-Ni in soil under plant was close to the significance level. The correlations obtained with the other mycorrhizal parameters (F, A and spore numbers) were not significant.

Correlations between mycorrhizal parameters and Ni-hyperaccumulation for Phyllanthus favieri

The values of leaf Ni content of P. favieri varied from 7,800 μg g−1 to 42,200 μg g−1 and those of DTPA-Ni in soil under plant from 25 μg g−1 to 225 μg g−1 (Table 4). Three root samples (13% of the total number of samples) did not show any AMF colonisation, but most of them were highly colonised.

Table 4 Variation of Ni-hyperaccumulation in Phyllanthus favieri (21 samples), DTPA extractable Ni in soil under plant and mycorrhizal parameters of the plant
Full size table

Significant negative correlations were recorded between M and leaf Ni content (r = −0.580), M and DTPA-Ni in soil under plant (r = −0.464), F and Ni content in leaves (r = −0.662).

Correlations between mycorrhizal parameters and Ni-hyperaccumulation for Psychotria douarrei

The minimal and maximal values of Ni content in leaves were respectively 5,300 and 35,530 μg g−1, with a mean of 17,800 μg g−1 (Table 5). Nine plant samples (20% of the total number of samples) did not show any AMF colonisation in their roots and all had more than 20,000 μg g−1 Ni in their leaves. The other plant samples were more or less colonised by AMF and showed arbuscules in the root cortex.

Table 5 Variation of Ni-hyperaccumulation in Psychotria douarrei (43 samples), DTPA extractable Ni in soil under plant and mycorrhizal parameters of the plant
Full size table

The high number of samples studied for this easy to collect plant allowed a more robust statistical analysis so that more significant correlations were obtained: all the mycorrhizal parameters were negatively and significantly correlated to leaf Ni content and to DTPA-Ni in soil under plant.

Correlations between mycorrhizal parameters and Ni-hyperaccumulation for Sebertia acuminata

Ni content in leaves of S. acuminata varied from 10,490 μg g−1 to 41,280 μg g−1 and DTPA-Ni in soil under plant from 572 μg g−1 to 883 μg g−1 (Table 6). Nine samples (50% of the total number of samples) did not show any AMF colonisation; among them 7 had more than 25,000 μg g−1 Ni in their leaves. The other 9 samples were generally very weakly colonised: only one with M > 12%. Among these 9 last samples 7 had less than 16,000 μg g−1 Ni in their leaves.

Table 6 Variation of Ni-hyperaccumulation in Sebertia acuminata (18 samples), DTPA extractable Ni in soil under plant and mycorrhizal parameters of the plant
Full size table

F was negatively and significantly correlated with leaf Ni content (Table 6). The correlation between M and leaf Ni content was very close to the significance level (P = 0.055).

Influence of soils collected under hyperaccumulators on AMF spore germination

Figure 2 shows the effects of soils on AMF spore germination. About 60% of the spores germinated on sand with distilled water. A solution of 10 μg g−1 Ni reduced the germination to 26%. A non-ultramafic soil, with 1 μg g−1 Ni did not affect on spore germination. The value of germination was 30.5% when spores were in contact with an ultramafic soil having 34 μg g−1 DTPA-Ni. Soils collected under Geissois pruinosa and Psychotria douarrei, characterised by a relatively high level of DTPA-Ni (respectively 110 and 278 μg g−1), induced a strong inhibition (Fig. 2). The inhibition induced by the soils was greater for soil with greater extractable Ni.

Fig. 2
figure 2

Influence of soils collected under Geissois pruinosa and Psychotria douarrei on AMF spore germination (isolate FSCtAc) in comparison with other soils and with a solution of 10 μg g-1 Ni. Ni-DTPA concentrations of the soils are shown on the top of the histograms. Different letters indicate significant differences at P ≤ 0.05 (ANOVA, Newman-Keuls test)

Full size image

Influence of root crushes of Ni-hyperaccumulators on AMF spore germination

Root crushes of Geissois pruinosa and Psychotria douarrei inhibited AMF spore germination (Fig. 3). The inhibition was more important for P. douarrei and was related to the concentration of total Ni in roots; among the 2 root samples of this plant the sample from Rivière Bleue, with 7,733 μg g−1 total-Ni, reduced the germination to only 4%.

Fig. 3
figure 3

Influence of 3 samples of root crush (Psychotria douarrei from mont Koghis, or Rivière Bleue (RB) and Geissois pruinosa from mont Koghis) on AMF spore germination (strain FSCtAc) in comparison with a solution of 10 μg g-1 Ni and a control of H2O. Total Ni concentrations of the root crushes are shown at the top of the histograms. Different letters indicate significant differences at P ≤ 0.05 (ANOVA, Newman-Keuls test)

Full size image

Concentrations of total and available Ni in roots of Ni-hyperaccumulators

The nickel absorbed by the roots was not totally in inactivated and non-mobile forms after the roots were crushed (Table 7). For G. pruinosa, about 25% of the Ni was extractable with KCl. The root crush of P. douarrei released 80% of the Ni in mobile form, whereas for S. acuminata the entire Ni concentration was extractable with KCl.

Table 7 Concentrations of total Ni and available Ni (extractable with KCl) in crushed roots of 3 Ni-hyperaccumulating plants (means of four values with standard errors)
Full size table

Tolerance to Ni of 2 AMF isolates from roots of hyperaccumulators in comparison with other isolates

The effect of three Ni concentrations on AMF spore germination varied according to each isolate tested (Fig. 4). Isolates FSCtAc and T2R3 were less tolerant: germination percentage was about five times lower in the presence of Ni in comparison with the control without Ni. SFONL was moderately tolerant, whereas Psychot1 and Sebert1 (isolated respectively from P. douarrei and S. acuminata) were highly tolerant to Ni. Indeed, for these latter two, the germination percentage was not affected by Ni at the highest concentration used.

Fig. 4
figure 4

Influence of different Ni-concentrations on AMF spore germination of two AMF isolates collected under hyperaccumulating plants (Psychot1 and Sebert1), in comparison with three other isolates. Ni−/Ni+ indicates, for each isolate, the ratio between the control without Ni and the average of the three Ni-treatment values, and show the tolerance of the isolates. For each isolate, different letters indicate significant differences at P ≤ 0.05 (ANOVA, Newman-Keuls test)

Full size image

Discussion

The ability of Ni-accumulating plants to absorb and store nickel is highly variable (Brooks 1987; Reeves 1992). The nine plant species studied here are representative of this diversity. A highly significant positive correlation between leaf Ni content and DTPA-Ni in soil under plants was highlighted for all the series of the studied samples. Boyd et al. (1999) analysed the relationships between leaf Ni content and DTPA-Ni in soil under Psychotria douarrei and did not find a significant correlation, but this could be due to the lower number of samples they studied. It is known that metal hyperaccumulators increase the toxicity of the soil underneath them, through their Ni-rich litter (Schlegel et al. 1991; Boyd and Jaffré 2001), and roles of this plant strategy (for example chemical defense against pathogens and competition with other plants) have been hypothetised (Boyd et al. 1994; Boyd and Jaffré 2001; Jhee et al. 2005).

The nine endemic Ni-accumulators studied were all more or less mycorrhizal, but only AM were observed. Apart from Phyllanthus favieri, which was previously reported as mycorrhizal (Perrier et al. 2006), the mycorrhizal status of all these species was documented for the first time. The first report of mycorrhizal colonisation of Ni-hyperaccumulating plants was by Turnau and Mesjasz-Pzybylowicz (2003) who showed AMF colonisation of Berkheya coddii and three other Ni-accumulating Asteraceae from South Africa. Another recent study concerned mycorrhizal colonisation of the Zn-hyperaccumulator Thlaspi praecox in metal polluted soils (Vogel-Mikus et al. 2005). In the light of these results, it is clear that metal-hyperaccumulating plants are not non-mycorrhizal, as was generally assumed in the past (Leyval et al. 1997; Pawlowska et al. 2000; Coles et al. 2001). However, our study clearly shows the strongest hyperaccumulating plants are less colonised by AMF than weak or moderate accumulators. Thus, Sebertia acuminata, commonly called «blue sap tree» because it contains the highest known level of Ni in its latex (Jaffré et al. 1976), was poorly colonised by AMF: 50% of the studied samples were not colonised, the other samples showed generally less than 12% mycorrhizal density (M). Another strong hyperaccumulator, Psychotria douarrei, showed AMF colonisation varied largely according to the levels of Ni concentrations in plant and soil. Therefore, what factors determine the inhibition of AMF colonisation of the strong hyperaccumulators?

The correlation analysis on variables studied in natural conditions requires a relatively high number of samples to reduce the heterogeneity due to the numerous factors, which influence the variables. In this study, negative correlations between nickel concentrations in soil and leaves and mycorrhizal parameters were recorded in all cases, but with various significance levels. The correlations were particularly clear for P. douarrei because of the high number of samples studied (43). These results mean that high levels of nickel in leaves and/or in soil are related to low levels of mycorrhizal colonisation.

Correlation does not necessarily indicate a causal relationship, but allows us, as a first step, to propose the following hypothesis: inhibition of root colonisation by mycorrhizal fungi may occur for high levels of Ni-hyperaccumulation and/or high levels of Ni in soil under plants due to Ni-hyperaccumulation. If this hypothesis is correct, soils taken from under plants, or roots of Ni-hyperaccumulators should inhibit germination or growth of AMF. In fact, soils collected under G. pruinosa or under P. douarrei induced a strong inhibition of AMF spore germination (Fig. 2). An ultramafic soil containing a lower level of DTPA-Ni caused a significantly lower inhibition, whereas a non-ultramafic soil did not affect spore germination. Only 10% of the spores germinated in contact with the soil collected under P. douarrei and containing 278 μg g−1 DTPA-Ni, which was 6 times lower than the control (sand). Inhibition of AMF spore germination by heavy metals in polluted soils has already been reported by Leyval et al. (1995) who also noticed that mycorrhizal parameters tended to be negatively correlated with total heavy metal concentrations (especially Cd, Zn and Pd) in soil. The concentrations of Ni in New Caledonian ultramafic soils are highly variable. DTPA extractable concentrations generally vary from 1 μg g−1 to 400 μg g−1 (Brooks 1987; Amir and Pineau 2003), but can reach 1,500 μg g−1 in the 15 cm upper soil under strong Ni-hyperaccumulating plants (as shown in Table 5). However, this concentration generally decreases in the deeper soil (results not shown). We noted that the inhibition induced by the 10 μg g−1 Ni solution in sand was similar to that obtained with a soil containing 34 μg g−1 DTPA-Ni. Indeed, DTPA-Ni soil does not measure the available and active fraction of Ni, but the releasable fraction, so that only a part of this fraction is active at any time.

It was not possible to test the direct effect of roots on AMF spore germination under field conditions, but crushed roots had significant negative effects on the germination. The roots with a high content of total-Ni induced a greater inhibition: only 4% of the spores in contact with root crush of P. douarrei collected from Rivière Bleue germinated. The roots of S. acuminata could not be tested in this experiment because of the difficulty to obtain a sufficient quantity of fine roots of this tree (roots are very deep in the soil) but, as shown in Table 7, 80–100% of the root Ni of the two strongest hyperaccumulators is in available form (extractable with KCl). Inside hyperaccumulator organs, Ni is generally complexed with organic acids, amino-acids (histidine) or nicotianamine (Krämer et al. 1996; Sagner et al. 1998; Ouerdane et al. 2006). These latter authors have found all these forms not only in the shoots but also in the roots and xylem of Thlaspi caerulescens. Organic acid complexes such as citrate and malate were found in Psychotria douarrei (Kersten et al. 1980) and in Sebertia acuminata (Lee et al. 1977; Sagner et al. 1998; Perrier et al. 2004). It must be noted that the root crush in the present experiment could also have released Ni stored in the cell vacuoles (Perrier et al. 2004), which is normally inactive in non-crushed roots. However, in natural conditions, the metal stored in the vacuoles is continuously released with the exfoliation of the dead cells, so that Ni concentration in the rhizoplane and in the rhizosphere is higher than in the bulk soil and can prevent the initiation of the root colonisation by AMF. Indeed, Bathia et al. (2004) reported that a high proportion of nickel is stored inside epidermal cells. In addition, these authors outlined that inside tissues Ni is present in a highly soluble and mobile form. Thus, the results of the two germination tests confirm that AMF development may be inhibited in soil under Ni-hyperaccumulators and in contact with their roots. It is interesting to note that in the roots of the moderate hyperaccumulator G. pruinosa only 20% of the total Ni was extractable with KCl, whereas the two strong hyperaccumulators contained more than 80% of KCl extractable Ni (Table 7). This could mean that the storage of Ni in the root in unavailable forms is reduced when the level of nickel is very high. This needs to be further investigated, and more precisely, it would be beneficial to identify the chemical forms of Ni inside the roots and in the rhizosphere.

Despite the clearly toxic levels of Ni in the soil and in the root surface of Ni-hyperaccumulators, at least some AMF colonisation was observed. This can be explained by the high Ni-tolerance level of the AMF associated with these plants. Indeed, the two isolates from roots of P. douarrei and S. acuminata are clearly more tolerant to Ni than the three other isolates tested. These two isolates are indifferent to concentrations of Ni up to 30 μg g−1 and are able to colonize at least parts of roots of Ni-hyperaccumulators when Ni content do not exceed about 20,000 μg g−1 in leaves. Tolerant AMF isolates to Cd and Zn have already been reported (Gildon and Tinker 1981; Weissenhorn et al. 1994), but Ni-tolerance of AMF has not been reported before.

Current research is dedicated to the identification by DNA sequencing of hyperaccumulator-related AMF isolates and to their better characterisation, including the study of their effects on plants. The relation between the hyperaccumulation phenomenon and mycorrhizal status also requires further investigation, in particular by comparison of pairs of plant species of the same genus (a Ni-hyperaccumulator and a non-accumulator), for example Geissois pruinosa and G. montana, or Psychotria douarrei and P. baillonii.