Introduction

Serpentine soils derived from ultramafic rocks are often enriched in Ni, Cr and Co and deficient in macronutrients, such as N, P, K and Ca. They also typically lack organic matter and present a low water-holding capacity. These characteristics convert these areas in stressful environments for plant growth (Brady et al. 2005; Brooks 1987). Nonetheless, some plant species known as hyperaccumulators have developed physiological mechanisms enabling them to tolerate and even thrive under these conditions, and to accumulate extreme concentrations of trace metals in their harvestable organs without showing signs of toxicity. Hyperaccumulator plants are recognized for their potential application in Ni phytomining processes. Nickel hyperaccumulation has been documented in approximately 450 species, and in temperate latitudes these mainly belong to the Brassicaceae family and particularly the Odontarrhena (syn. Alyssum) genus (Pollard et al. 2014).

Nickel phytomining cultivates hyperaccumulating plants which are able to bioaccumulate this metal from Ni-enriched soils in their harvestable plant parts, representing a biological technique for the recovery of this valuable metal (Nkrumah et al. 2016). Ni bio-ores (i.e. plant ashes) can be generated with up to 25% Ni, which is significantly higher than that present in current lateritic ores (< 1.5%) or normal Ni-ores (approx. 3%) where conventional mining processes are considered economically unviable (Chaney et al. 1998). Ni hyperaccumulators within the genus Odontarrhena are candidate species for phytomining in serpentine areas with a Mediterranean-type climate (Chaney et al. 2014). However, metal removal efficiency can be limited by a slow plant growth rate and low biomass. Several agronomic management strategies have been proposed to optimize the phytomining process, which focus on maximizing the yields of hyperaccumulator plants or their metal uptake and accumulation. Examples include soil pH modification, application of fertilizers, use of plant growth regulators, or bioaugmentation with plant-associated bacteria (Alvarez-Lopez et al. 2016a, b; Bani et al. 2015; Cabello-Conejo et al. 2014a; Chaney et al. 2014; Chaney and Baklanov 2017; Kidd et al. 2009, 2017).

Plant growth-promoting rhizobacteria (PGPR) enhance plant biomass production and nutrient availability in various ways, such as through the production of phytohormones (auxins, cytokinins and gibberellins) or Fe(III)-specific chelating siderophores, or by solubilizing inorganic P or fixing atmospheric N2 (Fravel 2005). They can also reduce plant stress through the release of 1-aminocyclopropane-1-carboxylic acid (ACC) deaminase which reduces the high levels of ethylene by consuming its immediate precursor, the ACC (Abou-Shanab et al. 2006; Glick et al. 1998). Other well-known mechanisms of PGPR include increasing water uptake, alteration of root morphology, production of antibiotics and the induction of plant defense mechanisms (Kidd et al. 2017). PGPR can also influence soil trace metal mobility and phytoavailability (Sessitsch et al. 2013).

Hyperaccumulator plants harbor distinct bacterial communities in the rhizosphere with a high genetic diversity, as well as a high level of resistance to metals (such as Ni) and capacity to promote plant growth (Benizri and Kidd 2018; Kidd et al. 2017; Thijs et al. 2017). Many studies have isolated Ni-tolerant PGPR strains associated with Ni-hyperaccumulators and successfully applied these as inoculants in metal-(hyper)accumulating plants to promote Ni removal from the soil (reviewed by Benizri and Kidd 2018; Kidd et al. 2017). Most of these studies evaluated improvements in plant biomass and/or shoot Ni uptake and accumulation compared to non-inoculated plants. However, the effects of applying these inoculants on plant physiological status have rarely been studied. Also, in many cases the specificity of these plant-bacterial associations is unclear (Benizri and Kidd 2018) and the application of these bioinoculants could themselves also cause plant stress. Moreover, their possible combination with other beneficial agronomic practices, such as the use of fertilization or soil amendments, has not to date been addressed within the field of phytomining.

Abiotic stresses, including trace metal toxicity, result in cellular damage and oxidative stress in plants due to the formation of reactive oxygen species (ROS) (Cuypers et al. 2010; Foyer et al. 1997; Hall 2002; Sharma and Dubey 2005). Plants have developed various strategies for trace metal detoxification that involve both enzymatic and non-enzymatic components. The antioxidant defense system protects plant cells from oxidative damage by regulating the level of cellular ROS (Mittler 2002; Fabiano et al. 2015). Hyperaccumulators are capable of storing high concentrations of metals in their aerial parts without showing toxicity symptoms (usually two or to three orders of magnitude higher than in plant leaves growing on uncontaminated soils, van der Ent et al. 2013). Hyperaccumulators can alleviate ROS damage caused by metal toxicity more efficiently than non-hyperaccumulating plants (Chiang et al. 2006). Increased production of glutathione in Noccaea goesingensis (syn. Thlaspi goesingense) and other Ni hyperaccumulators within the Noccaea genus is thought to provide protection against oxidative damage induced by high Ni concentrations (Freeman et al. 2004). Boominathan and Doran (2002) reported higher activities of the antioxidative enzymes, superoxide dismutase (SOD) and catalase (CAT), in Odontarrhena bertolonii compared to the non-hyperaccumulator Nicotiana tabacum, which they attributed to more developed mechanisms for alleviating oxidative damage in the Ni-hyperaccumulator.

The main aim of this study was to investigate the effects of beneficial rhizobacterial strains, previously isolated from Odontarrhena serpyllifolia, on three Ni-hyperaccumulating Odontarrhena spp. growing in serpentine soil, and the effects of these bioinoculants on plant physiological status (antioxidant defense system). A second objective was to study the interaction between fertilization regimes and inoculation on the growth, shoot biomass production and Ni phytoextraction capacity of the three Odontarrhena species: Odontarrhena serpyllifolia (Desf.) Jord. & Fourr. (syn. Alyssum serpyllifolium subsp. lusitanicum T.R. Dudley & P.C.Silva), Odontarrhena inflata (Nyár.) D.A.German (syn. Alyssum inflatum Nyár.) and Odontarrhena bracteata (Boiss. & Buhse) Španiel, Al-Shehbaz, D.A.German & Marhold (syn. Alyssum bracteatum Boiss. & Buhse). The latter two species have not previously been studied for their Ni phytoextraction potential.

Materials and methods

Experimental set-up

Soil (upper 0–20 cm) was collected from non-vegetated areas in the Melide ultramafic complex in NW Spain (where Odontarrhena serpyllifolia is a native species). Soils were air-dried, sieved through an 8-mm stainless steel mesh and mixed with washed quartz sand (10% w/w) to improve aeration and drainage. Half of the soil was amended with cow manure (organic amendment) at a ratio of 2% w/w (computed on a DW basis) and the other half was fertilized with inorganic NPK. P and K were added in the form of KH2PO4 at an application rate of 80 and 100 kg ha−1, respectively, and N in the form of NH4NO3 equivalent to 80 kg ha−1. After amendment soils were adjusted to 70% water holding capacity and left for 2 weeks to attain equilibrium before planting.

Three Ni-hyperaccumulating plant species were selected for the experiment: Odontarrhena serpyllifolia which is endemic to the Iberian Peninsula, and Odontarrhena inflata and Odontarrhena bracteata which are both native to serpentine soils of western Iran. Seeds were randomly collected at each site from 10 to 15 individual plants growing in close proximity and mixed together. Seeds were germinated in plastic flats on a 2:1 perlite:quartz mixture (2:1 v/v) in a growth chamber under controlled conditions (temperature 22–25 °C, photosynthetic photon flux density (PPFD) of 190 μmol m−2 s−1, under a 16/8 h light/dark cycle). Seeds were watered with distilled water until germination and thereafter with a serpentine-like nutrient solution which consisted of 2 mM MgSO4, 0.8 mM Ca(NO3)2, 2.5 mM KNO3, 0.1 mM K2HPO4, 75 μM KCl, 20 μM FeEDDHA chelate, 10 μM H3BO3, 2 μM MnCl2, 1 μM ZnSO4, 0.5 μM CuSO4, 0.2 μM Na2MoO4 and 10 μM NiSO4 in deionized water (based on Chaney et al. 2009). One-month-old seedlings of each Odontarrhena species were transplanted into pots containing 1 kg of either NPK- or manure-amended soil (one plant per pot). After a 14 day adaptation period, seedlings were inoculated with one of 5 rhizobacterial strains. Non-inoculated pots were also prepared as a control treatment.

Five bacterial isolates with plant growth promoting traits were selected for this study. Four of these strains were previously isolated from the rhizosphere soil of Odontarrhena serpyllifolia growing in three ultramafic outcrops of the Iberian Peninsula: Melide (Paenarthrobacter nitroguajacolicus strain LA44 [formerly Arthrobacter nitroguajacolicus; Busse 2016]), Trás-os-Montes (Paenarthrobacter nicotinovorans strain SA40 [formerly Arthrobacter nicotinovorans; Busse 2016] and Stenotrophomonas sp. strain MA98), and Sierra Bermeja (Pseudoarthrobacter oxydans strain SBA82 [formerly Arthrobacter oxydans; Busse 2016]) (Alvarez-Lopez et al. 2016a). Strain LA44 was shown to be an efficient mobilizer of Ni from ultramafic rocks under in vitro conditions, and also shows intense IAA-production and is highly Ni-resistant (Alvarez-Lopez et al. 2016a; Becerra-Castro et al. 2013). Strains SBA82, SA40 and MA98 show moderate Ni tolerance and were characterized for the ability to solubilize inorganic P (SBA82), or as IAA- (SA40 and SBA82) or siderophore- (all three strains) producers (Álvarez-López et al. 2016a). In previous studies, the strains LA44, SBA82 and SA40 significantly increased plant growth and, in some cases, shoot Ni concentrations of Odontarrhena serpyllifolia growing in serpentine soil (Becerra-Castro et al. 2013; Cabello-Conejo et al. 2014a). Finally, the Rhodococcus fascians strain ALM1 was found to colonize the rhizosphere of Odontarrhena serpyllifolia when growing in substrates enriched with ultramafic rock, and is classified as an IAA- and ACC- producer (unpublished results; deposited in the EMBL database under accession number MF692766).

Fresh cultures of bacterial strains were grown in 869 liquid medium (1.0 g tryptone, 0.5 g yeast extract, 0.5 g NaCl, 0.1 g glucose, 0.035 g CaCl2.2H2O in 1 L deionized water; Mergeay et al. 1985) for 48 h, then harvested by centrifugation (6000 rpm, 10 min), washed and re-suspended in 10 mM MgSO4 to an optical density of 1.0 at 580 nm (about 108 cells per mL). Each plant pot was inoculated with 20 mL of each bacterial suspension around the base of the plant. The same amount of sterile 10 mM MgSO4 was added to non-inoculated plants. After one month growth the inoculation procedure was repeated as above. Six replicate pots were established for each plant species, fertilization regime (NPK or manure) or inoculation treatment. Plants were watered with distilled water and kept under controlled conditions (temperature 22–25 °C, PPFD of 190 μmol m−2 s−1, under a 16/8 h light/dark cycle) for a total of three months. Before harvesting, various aliquots of approximately 0.1 g of fresh weight (FW) leaves were collected, frozen in liquid N2 and stored at −80 °C, for evaluating the physiological status of the plants.

Soil analyses, plant biomass and nutritive status

The air-dried, <2 mm fraction of soils were characterized for the following physico-chemical properties before planting and after harvesting. Soil pH was measured in H2O using a 1:2.5 soil:solution ratio. Total C and N were analyzed by combustion with a CHN analyzer (Model CHN-1000, LECO Corp., St Joseph, MI). Available P was extracted with Olsen’s reagent (0.5 M NaHCO3 adjusted to pH 8.2, 1:20 w/v) and determined colorimetrically using the molybdenum blue method (Murphy and Riley 1962). Pseudo-total metal concentrations were determined using 0.2 g of soil digested in a 3:1 mixture of concentrated HNO3:HCl in Teflon PFA vessels in a microwave accelerated reaction system (MarsXpress; CEM Corp., USA). Total concentrations of metals in digests were analyzed by inductively coupled plasma optical emission spectrometry (ICP-OES, model Vista-PRO, Varian). A certified soil standard (ISE 979 Rendzina Soil) was used to ensure quality of metal quantification. Exchangeable cations (Ca, Mg, Al, Na and K) were extracted with 1 M NH4Cl (1:20 w/v, 16 h equilibration) and determined by ICP-OES. Ni availability was assessed after extraction with DTPA-TEA (0.005 M DTPA with 0.01 M CaCl2 and 0.1 M triethanolamine (TEA) at pH 7.3, 1:10 w/v, 2 h shaking) and with 10 mM Sr(NO3)2 (1:4 w/v, 2 h shaking). Filtrates of both extractions were analyzed by ICP-OES as above (Everhart et al. 2006; Lindsay and Norvell 1978).

At harvest, plants were separated in shoots and roots, washed with pressurized tap water and then by distilled water, and dried for 48 h at 45 °C to determine dry weight (DW) biomass. Dried shoot biomass (approximately 0.1 g) was digested in a 2:1 HNO3:HCl mixture on a hot plate at 160 °C, and the concentration of micro- and macro-nutrients were determined by ICP-OES. Data were expressed in mg kg−1 DW plant material. Phytoextracted Ni was calculated as the product of the shoot Ni concentration and shoot DW yield.

Plant physiological status

All the physiological parameters were determined for all three plant species grown in manure-amended soils. However, the biomass of O. inflata and O. bracteata grown in the NPK-treated soil was only sufficient to determine photosynthetic pigments.

MDA and H2O2

Malondialdehyde (MDA) was estimated by the thiobarbituric acid (TBA) test, according to Heath and Packer (1968). MDA concentration was calculated from the extinction coefficient 155 mM−1 cm−1 and expressed as nmol MDA g−1 fresh weight (FW) shoot tissue. Hydrogen peroxide (H2O2) levels were evaluated by methods described by Velikova et al. (2000). The content of H2O2 was determined using the extinction coefficient 0.28 μM−1 cm−1 and expressed as μmol H2O2 g−1 FW.

Chlorophylls and carotenoids

The concentration of total chlorophyll (Chltot), chlorophyll a (Chla) and b (Chlb), and carotenoids were determined spectrophotometrically according to Arnon (1949) using 80% acetone (v/v).

Enzymes assays

Frozen leaf tissues (0.1 g) were homogenized in 0.1 M potassium phosphate buffer (pH 7.8) containing 1 mM EDTA and 2% (w/v) polyvinylpyrrolidone (PVP) in an ice bath. The homogenate was centrifuged for 20 min at 15,000 g at 4 °C. The activities of different enzymes were determined in the supernatant and the protein concentration was determined according to Bradford (1976) using bovine serum albumin (BSA) as a standard. Superoxide dismutase (SOD) activity was assayed using a modified NBT (Nitro Blue Tetrazolium) method (Giannopolitis and Ries 1977). The reaction mixture contained 60 μL leaf extract, 50 mM phosphate buffer (pH 7.8), 75 μM NBT, 2 μM riboflavin, 13 mM L-methionine, and 0.1 mM EDTA. One unit of SOD activity was defined as the enzyme concentration required for 50% inhibition of the reduction of NBT at 560 nm. Enzyme activity was expressed as U mg−1 protein min−1. Catalase (CAT) activity was determined according to Aebi (1984). Assay mixtures contained 50 μL leaf extract, 50 mM potassium phosphate buffer (pH 7.0) and 10 mM H2O2. Catalase activity was calculated using the absorbance coefficient of H2O2 (0.039 mM−1 cm−1). Enzyme activity was expressed as U mg−1protein min−1. One unit of CAT gives a decomposition rate of 1 μmol H2O2 per minute at 240 nm at 25 °C. Ascorbate peroxidase (APX) activity was determined according to Nakano and Asada (1981). The assay mixture contained 50 μL leaf extract, 50 mM potassium phosphate buffer (pH 7.0), 0.5 mM ascorbate, 0.5 mM H2O2 and 0.2 mM EDTA. APX activity was determined from the decrease in absorbance at 290 nm due to oxidation of ascorbate in the reaction. The extinction coefficient (ε) of 2.8 mM−1 cm−1 for reduced ascorbate was used in calculating the enzyme activity which was expressed as U mg−1 protein min−1. Glutathione reductase (GR) activity was assayed according to Smith et al. (1988). The assay mixture contained 40 μL leaf extract, 50 mM potassium phosphate buffer (pH 7.8), 1 mM GSSG, 0.75 mM DTNB, 0.1 mM NADPH and 0.5 mM EDTA. The increase in absorbance at 412 nm was measured when DTNB was reduced to TNB by GSH in the reaction. The extinction coefficient of TNB (14.15 M−1 cm−1) was used to calculate the activity of GR which was expressed as U mg−1 protein min−1.

Phenylalanine ammonia lyase (PAL) and tyrosine ammonia-lyase (TAL) enzyme assays

Leaf samples (0.1 g) were homogenized in 50 mM Tris-HCl buffer (pH 8) containing 15 mM 2-mercaptoethanol and centrifuged at 7000 g for 15 min in 4 °C. The supernatant was collected and centrifuged at 13,000 g for 20 min at 4 °C. The total protein concentration was determined using the Bradford method (Bradford 1976). The method of Beaudoin-Eagan and Thorpe (1985) was used to determine PAL and TAL activities. The enzyme activity was expressed in μmol (cinnamic or coumaric acid) mg protein−1 min−1 at 290 and 333 nm respectively, where 1 U is defined as 1 μmol (cinnamic or coumaric acid) mg protein−1 min−1.

Total phenolic content

Leaf material (0.02 g) was homogenized in 95% (v/v) methanol and total phenolic content was determined by the Folin-Ciocalteu method (Singleton and Rossi 1965). The phenolic contents were estimated using a standard curve of gallic acid and expressed as μg g−1 FW.

Statistical analyses

All statistical analyses were carried out using the SPSS software (version 22). The effects of fertilization, bacterial strain and plant species were analyzed using two- and three-way ANOVA and individual means were compared using the Duncan’s test with p < 0.05 as a significance threshold.

Results

Effects of fertilization and bacterial inoculants on soil properties

The physicochemical properties of the untreated soil and the two fertilized soils (NPK or manure) before planting are presented in Table 1. All soil samples presented a slightly acidic pH (5.8–5.9). Total C content ranged from 4.0 to 4.7%, and was significantly lower in the fertilized soils, while total N increased after addition of both NPK and manure. Cation exchange capacity (CEC) ranged from 6.2 to 8.4 cmolc kg−1, and was significantly higher after amendment with manure but lower after NPK addition. Manure addition significantly increased exchangeable Ca and K. As is characteristic of serpentine soils, the Ca/Mg ratio was <1 in all soils, but was significantly higher in the manure–amended soil compared to either the untreated or NPK-fertilized soil (p < 0.05). Manure addition also led to a slight but significant increase in water retention capacity and decrease in soil density. Dissolved organic carbon and soil respiration were also significantly enhanced in manure-amended soil. Total Ni and Cr concentrations varied from 3.9 to 4.6 g kg−1 and from 1.9 to 2.5 g kg−1, respectively, and tended to be lowest after incorporating manure. Total concentrations of Co, Cu (only in manure) and Mn decreased significantly after NPK fertilization and manure amendment, while P and K concentrations increased significantly after manure addition (p < 0.05). Both DTPA- and Sr(NO3)2-extractable concentrations of Cr were below the detection limit. DTPA-extractable concentrations of Ni, Mn and Co were consistently higher than Sr(NO3)2-extractable concentrations. Inorganic fertilization and manure addition induced a significant decrease in the DTPA-extractable concentrations of these metals compared to untreated soil. Sr(NO3)2-extractable concentrations of Ni and Co were only significantly reduced after amending the soil with manure (p < 0.05).

Table 1 Physico-chemical properties of the untreated and treated serpentine soil before planting of Odontarrhena species
Full size table

Soil pH, Ni and Co availability were analyzed after plant growth. Soil pH in NPK-fertilized plots was not modified by plant growth or bacterial inoculation. A slight but significant reduction in soil pH was observed for O. bracteata grown in manure-amended soils and inoculated with strains LA44 and SBA82. Plant growth increased DTPA-extractable concentrations of Ni, but bacterial inoculants did not induce strong changes in metal availability. Nonetheless, a few general effects of inoculants were observed (Table 2). In NPK-fertilized pots the DTPA-extractable Ni concentrations were significantly higher after inoculating with the strains SA40, MA98 and ALM1 (p < 0.05). This was the case for all three plant species (albeit not significant for O. serpyllifolia) and the highest increase was observed for O. bracteata after inoculation with strain MA98: DTPA-extractable Ni increased from 87.9 to 102.4 mg kg−1. This effect was not reflected in Sr(NO3)2-extractable Ni concentrations in this treatment. In manure-amended pots, both DTPA- and Sr(NO3)2-extractable Ni concentrations were actually lower after growth of O. serpyllifolia and inoculation with strains MA98 and ALM1 (decreasing from 88.5 mg kg−1 to 80.2 and 81.3 mg kg−1, and from 5.0 mg kg−1 to 4.3 and 4.7 mg kg−1, respectively). In contrast, Sr(NO3)2-extractable Ni concentrations increased in both O. inflata and O. bracteata pots inoculated with strain LA44. The same was observed for O. bracteata inoculated with SBA82. An effect of inoculation on Co availability in manure-amended soils was only observed for O. bracteata. DTPA-extractable Co increased by 1.4-fold in ALM1-inoculated soils after growth of O. bracteata compared to non-inoculated soils. In NPK-fertilized soils with the same species, available Co (DTPA-extractable) increased by 1.2 to 1.3-fold after inoculation with SA40, MA98 and ALM1 strains. In NPK-fertilized pots, a consistent increase in DTPA-extractable Co concentrations was observed after inoculation of both O. serpyllifolia and O. bracteata with the strains SA40, MA98 and ALM1 compared to non-inoculated soils (Table 2). A slight but significant increase in the Sr(NO3)2-extractable concentration of this element was also observed in O. serpyllifolia.

Table 2 Concentrations of Sr(NO3)2-extractable Ni and Co (mg kg−1) and DTPA-extractable Ni and Co (mg kg−1) at harvest. Plants were inoculated with rhizobacterial strains LA44, SBA82, SA40, MA98, ALM1 and non-inoculated (NI)
Full size table

Effects of fertilization and bacterial inoculants on plant biomass production

Figure 1 shows the shoot and root dry weights of O. inflata, O. serpyllifolia and O. bracteata. The highest shoot DW yields in NPK-fertilized soil were obtained in O. serpyllifolia, which were approximately 20-fold higher than either O. inflata or O. bracteata (considering non-inoculated plants). In manure-amended soils, O. serpyllifolia and O. inflata showed the best growth, with shoot DW yield of about 2.5-fold higher than that of O. bracteata (non-inoculated plants). Plants always grew better in the manure-amended soil than after inorganic fertilization, with highly significant differences between the two amendment types in all three species. The amendment*species interaction factor were significant for all growth parameters (p < 0.01) (Table 3). This stimulation in biomass production was most pronounced in the Iranian Odontarrhena species. Shoot DW yields of O. inflata and O. bracteata were enhanced by up to 47- and 16-fold when grown in manure-amended soils, while the corresponding increase in yield for O. serpyllifolia was only 1.9-fold. Similarly, root DW yields were highest for O. serpyllifolia, and in the beneficial effect of manure addition was most pronounced in O. inflata and O. serpyllifolia (root DW increased by 16- to 23-fold in non-inoculated plants).

Fig. 1
figure 1

Effect of 5 different rhizobacterial inoculants on the shoot dry weight yield (SDW) and root dry weight yield (RDW), of O. inflata (a and d), O. serpyllifolia (b and e) and O. bracteata (c and f). Significant differences between treatments for each plant are indicated with different letters above bars (p < 0.05)

Full size image
Table 3 Significance of the effects of the main factors amendments (Am), species and strains, their first order interactions, and their second order interaction on shoot dry weight yield (SDW), Root dry weight yield (RDW), shoot Ni concentration, shoot Ni removal, chlorophyll a (Chl a), chlorophyll b (Chl b), total chlorophyll (Total Chl), carotenoids (Car), superoxide dismutase (SOD), catalase (CAT), ascorbate peroxidase (APX), and glutathione reductase (GR) activities, malondialdehyde (MDA), hydrogen peroxide (H2O2), phenolic compounds (phenolic com.), phenylalanine ammonia-lyase (PAL) and tyrosine ammonia-lyase (TAL) activities
Full size table

The effects of the bacterial inoculants on the growth of the Ni-hyperaccumulators were variable and depended on the strain and type of fertilization applied. None of the bacterial inoculants negatively affected plant growth in either soil type (although shoot DW tended to be lower in LA44-inoculated O. inflata grown in manure-amended soil). The most pronounced result was a general beneficial effect on the growth of O. serpyllifolia from which these strains were originally isolated. In manure-amended soils four out of the five strains evaluated significantly stimulated shoot and root DW yields of this species (albeit not always statistically significant). Strains LA44, SBA82 (not significantly), MA98 and ALM1 increased mean shoot DW from 1.2 g plant−1 to 1.9, 1.6, 1.9 and 1.9 g plant−1, respectively (p < 0.05). However, this growth-promoting effect was not observed in NPK-fertilized soils. For both O. inflata and O. bracteata fewer inoculants were found to stimulate the growth of these two species, but in this case beneficial effects were observed in both soil types. The effect of bacterial strains on root biomass generally followed the same pattern as that observed in shoot biomass. In manure-amended pots, O. inflata inoculated with Pseudoarthrobacter oxydans strain SBA82 showed the highest biomass production but this was the only inoculant with a stimulatory effect in this species: shoot and root DW yield was increased by 1.6- to 2.9-fold and by 2.1- to 2.6-fold, respectively, compared to other inoculated or non-inoculated plants. Strains SA40 and MA98 tended to increase root DW yield in this species. In NPK-fertilized pots, two strains were beneficial for the growth of O. inflata: root and shoot DW yields were increased after inoculation with strain SBA82 and Paenarthrobacter nicotinovorans strain SA40. The PGP effect was more pronounced on root yields than shoots (Fig. 1). Strain SBA82 also enhanced shoot DW yields of O. bracteata when grown in NPK-fertilized soils. Two additional strains had a positive effect on this latter species: Paenarthrobacter nitroguajacolicus strain LA44 stimulated plant yields in both soil types (albeit not always significant) and MA98 increased root and shoot yields by 1.6- and 1.9-fold in manure-amended soil, respectively.

Globally, the effects of bacterial strains on shoot DW yield was insignificant (p ˃ 0.05), but a significant effect of inoculation was found on root DW yield (p < 0.01). On the other hand, plant responses to inoculants were significantly different between species (p < 0.01 for species* strain interaction) (Table 3). The amendment*strain interaction factor was insignificant, but the amendment*species*strain second order interaction was significant for both shoot and root DW.

Effects of fertilization and bacterial inoculants on shoot Ni concentration and phytoextraction efficiency

Shoot Ni concentration was generally lower in O. serpyllifolia when this species was grown in manure-amended soils compared to NPK-fertilized soils, while the opposite was observed for both O. inflata and O. bracteata. At the species level, shoot Ni concentrations showed considerable variation but were of a similar magnitude in O. serpyllifolia and O. inflata (4 to 9 g kg−1), but up to 10-fold lower in O. bracteata (0.2 to 1.2 g kg−1). On the other hand, in all three species the shoot Ni yields were significantly higher when grown in manure-amended compared to NPK-fertilized soils, and this difference was particularly pronounced in O. inflata and O. bracteata. In accordance with the shoot Ni concentrations, yields were consistently higher for O. serpyllifolia and O. inflata. The effect of soil amendment on shoot Ni yields was highly significant for all three hyperaccumulators (amendment*species interaction factor, p < 0.001; Table 3).

The effects of bacterial inoculation on shoot Ni accumulation and yields varied according to the fertilization regime and the inoculant applied. The global effect of inoculation on shoot Ni concentration and yield was insignificant, but the plant species*strain interaction factor was significant (p < 0.05 for shoot Ni concentration and p < 0.001 for phytoextracted Ni) (Table 3). Moreover, the amendment*species*strain second order interaction was significant at a level of p < 0.001 in the two-way ANOVA (Table 3). As observed for plant growth, shoot Ni concentrations and yields of O. serpyllifolia were unaffected by bacterial inoculants in NPK-fertilized soils. On the other hand, a generalized increase in shoot Ni concentration was observed in inoculated plants of this species when grown in manure-amended soils. The increase in shoot Ni concentration was significant for strains LA44, MA98 and ALM1. In accordance, shoot Ni yields were also significantly enhanced in O. serpyllifolia grown in manure-amended soil and inoculated with these three strains (and tended to increase with SBA82) (Fig. 2). Shoot Ni yields obtained for non-inoculated plants were 5.9 mg plant−1 and increased to 13.7 mg plant−1 in MA98-inoculated plants, representing an increase of 2.3-fold in Ni removal. Similarly, there was no significant effect of bacterial inoculation on shoot Ni concentrations of O. inflata in soils with inorganic fertilization (although concentrations tended to be higher in SBA82-inoculated plants). Moreover, inoculating this species in the manure-amended soil led to a significant reduction in shoot Ni concentrations but this was not reflected in Ni yields (Fig. 2). Ni yields of O. inflata were significantly enhanced in both NPK-fertilized and manure-amended soils after inoculation with Pseudoarthrobacter oxydans strain SBA82. This effect, a result of the enhanced shoot biomass, was most pronounced in plants grown in manure-amended soils. Shoot Ni concentrations were not enhanced by bacterial inoculants in O. bracteata, and in fact, they were significantly reduced in manure-amended soils with strain LA44. One isolate, Rhodococcus fascians strain ALM1, significantly increased Ni yields of O. bracteata and this was only observed when grown in manure-amended soil due to an enhanced biomass production.

Fig. 2
figure 2

Effects of rhizobacteria and amendments on shoot Ni concentration and shoot Ni removal of three Ni hyperaccumulator species O. inflata (a and d), O. serpyllifolia (b and e) and O. bracteata (c and f). Significant differences between treatments for each plant are indicated with different letters above bars (p < 0.05)

Full size image

Effect of bacterial inoculants on plant nutritive and physiological status

Nutrient concentrations

There were no generalized effects of bacterial inoculation on shoot macro- or micro-nutrients in plants grown in either NPK-fertilized or manure-amended soils. However, some strains significantly influenced the plant nutritional status (Table 4). In NPK-fertilized soils, some inoculants were found to increase the shoot Fe concentration of all three species. This was significant for SA40, MA98 and ALM1 in O. serpyllifolia (increasing by up to 1.6-fold), for MA98 and ALM1 in O. inflata (up to 3.7-fold), and for SBA82 and MA98 in O. bracteata (up to 3.7-fold). On the other hand, shoot Fe concentration in plants grown in manure-amended soils decreased significantly after inoculation, and this was the case for all three Odontarrhena species (only significant for SBA82 and SA40 in O. serpyllifolia). The mean Fe concentration was 0.39, 0.69 and 0.22 g kg−1, respectively, in non-inoculated plants of O. inflata, O. serpyllifolia and O. bracteata and was reduced to 0.16 and 0.12 in O. inflata and O. bracteata inoculated with strain ALM1, and to 0.44 g kg−1 in O. serpyllifolia inoculated with SBA82. Shoot Ca concentrations were significantly reduced in O. inflata after inoculation in NPK-fertilized soils. In the same species, inoculation led to a significant reduction in shoot P concentration. In contrast, inoculating O. serpyllifolia in manure-amended soils generally led to an increase in shoot K concentration, which was significant for all strains except SA40. Finally, some inoculants significantly increased shoot Co concentration. In manure-amended soil, strain MA98 increased shoot Co concentration from 26.1 to 46.6 mg kg−1 in O. serpyllifolia and strain ALM1 increased Co accumulation in O. bracteata (from 16.1 to 26.1 mg kg−1). In NPK-fertilized soil the highest Co concentrations were found in SBA82-inoculated O. serpyllifolia and in MA98-inoculated O. bracteata (Table 4).

Table 4 Shoot macro- and micro-nutrient concentrations in the three Odontarrhena species grown in NPK-fertilized and manure-amended soils (mean concentration ± SE)
Full size table

Photosynthetic pigments

Photosynthetic pigments (total chlorophyll, chlorophyll a and b, and carotenoids) were of a similar magnitude in all three Odontarrhena species when grown in both soils (NPK- and manure-amended), but the highest levels were generally found in O. serpyllifolia (data not shown). In non-inoculated plants, Chla ranged from 0.71 to 2.08 mg g−1 FW shoot tissue, Chlb from 0.50 to 0.67 mg g−1 FW shoot tissue, Chltot from 0.97 to 2.88 mg g−1 FW shoot tissue and carotenoids from 0.41 to 1.20 mg g−1 FW shoot tissue.

There were few or even no significant effects of inoculation on photosynthetic pigments determined in O. serpyllifolia when grown in either soil. A significant increase in the concentration of Chlb and carotenoids was observed in O. serpyllifolia inoculated with strain SA40 and grown in NPK-fertilized soil (1.3-fold increase). In contrast, all inoculants enhanced chlorophyll (a, b and total) and carotenoid concentrations in O. inflata grown in NPK-fertilized soil: Chla increased by 1.6- to 2.5-fold, Chlb by 1.5- to 2.7-fold, Chltot by 1.6- to 2.6-fold and carotenoids by 1.4- to 2.5-fold. Three strains (SBA82, SA40 and MA98) also increased chlorophyll and carotenoid concentrations in this species when grown in manure-amended soil. In the case of SBA82, this coincided with the significant increase in shoot DW yield. Finally, all pigments were significantly increased in O. bracteata after inoculation with strain MA98 (by up to 1.6-fold), and this occurred in both soils (NPK- or manure-amended). In manure-amended soil it again coincided with the increase observed in shoot biomass.

Antioxidant enzymes

Activities of the antioxidant enzymes SOD and GR were of a similar magnitude in the leaves of all three plant species, while APX and CAT activity were significantly higher in O. inflata than either O. serpyllifolia or O. bracteata. Enzyme activities for all three plant species grown in manure-amended soils are presented in Fig. 3. It was not always possible to determine APX activities in all treatments for O. serpyllifolia and the results of this enzyme are not presented. SOD activity significantly decreased in O. inflata inoculated with strains SBA82, SA40, MA98 and ALM1 strains to 80, 72, 75 and 79% of that determined in non-inoculated plants. Both CAT and APX activities decreased after inoculation with strains SBA82 (which stimulated growth) and SA40, this was significant in the former where activities were 55 and 47% of non-inoculated plants, respectively. However, no effect of bacterial inoculation was found on GR activity in this species (Fig. 3). Inoculation of O. serpyllifolia with strain SBA82 reduced SOD activity to 68% of non-inoculated plants (Fig. 3), but there were no significant effects on GR and CAT activities in this hyperaccumulator. The CAT activity significantly decreased in O. bracteata when inoculated with LA44, SBA82, SA40, MA98 and ALM1 strains by 35–62%. However, no significant effects of microbial inoculants were observed on APX and SOD activities in this hyperaccumulator (Fig. 3). The species*strain interaction was significant for SOD, CAT and APX, but not for GR (Table 3).

Fig. 3
figure 3

Influence of rhizobacteria on the activities of (a) SOD, (b) CAT and (c) GR activities in leaves of the three Ni hyperaccumulator species O. inflata, O. serpyllifolia and O. bracteata grown in manure-amended soil. Significant differences between treatments for each plant are indicated with different letters above bars (p < 0.05)

Full size image

Malondialdehyde (MDA) and hydrogen peroxide (H2O2)

Lipid peroxidation indicators (MDA and H2O2) were generally higher in leaves of O. bracteata than O. inflata and O. serpyllifolia growing in manure-amended soil (Fig. 4). Bacterial inoculation influenced these parameters in some cases and the plant species*strain interaction factor was significant (p < 0.01; Table 3). MDA accumulation was significantly reduced in SBA82-inoculated plants of O. inflata compared to non-inoculated plants. The same pattern was observed for H2O2 concentration. Similarly, MDA accumulation was reduced in O. serpyllifolia after inoculation with LA44, SA40, MA98 and ALM1 (albeit not significant). In the case of O. bracteata, a significant reduction in MDA and H2O2 concentrations was observed after inoculation with the MA98 strain (Fig. 4).

Fig. 4
figure 4

Effects of rhizobacteria on (a) MDA and (b) H2O2 concentration in leaves of the three Ni hyperaccumulator species, O. inflata, O. serpyllifolia and O. bracteata grown in manure-amended soil. Significant differences between treatments for each plant are indicated with different letters above bars (p < 0.05)

Full size image

Phenylalanine ammonia lyase (PAL) and tyrosine ammonia-lyase (TAL) activity, and total phenolic compounds

Activities of PAL and TAL in leaves were generally of a similar magnitude amongst the three plant species when grown in manure-amended soil (Fig. 5), while phenolic compounds followed the order O. bracteata > O. inflata > O. serpyllifolia (Fig. 6). Bacterial inoculation had a significant effect on these parameters, but this depended on the plant species. The plant species*strain interaction factor was significant for all three parameters (Table 3). Inoculation of O. inflata with strains LA44 and SBA82 induced a significant increase in PAL (2.1- and 1.6-fold, respectively) and TAL (1.6- and 1.3-fold) compared with the non-inoculated plants, but did not influence the concentration of phenolic compounds. A similar increase in the activity of both enzymes was found in O. bracteata after inoculation with SA40, and in this species an increase was also observed in phenolic compounds after inoculation with ALM1. In contrast, PAL activity was significantly lower in O. serpyllifolia inoculated with the SBA82 strain, but there was no effect on either TAL activity or phenolic compounds (Figs. 5 and 6).

Fig. 5
figure 5

Influence of rhizobacteria on (a) phenylalanine ammonia-lyase (PAL) and (b) tyrosine ammonia-lyase (TAL) activities in leaves of the three Ni hyperaccumulator species, O. inflata, O. serpyllifolia and O. bracteata grown in manure-amended soil. Significant differences between treatments for each plant are indicated with different letters above bars (p < 0.05)

Full size image
Fig. 6
figure 6

Effects of rhizobacteria on total phenolic compound concentration in the leaves of the three Ni hyperaccumulator species, O. inflata, O. serpyllifolia and O. bracteata grown in manure-amended soil. Significant differences between treatments for each plant are indicated with different letters above bars (p < 0.05)

Full size image

Discussion

As expected the serpentine soil used in this study was deficient in nutrients (N, P and K) and presented a low Ca/Mg ratio which are limiting factors for plant growth. The addition of both inorganic fertilizers and manure improved N and P content. Amending with manure additionally enhanced exchangeable K and improved the Ca/Mg ratio compared to NPK-fertilized soil. Application of inorganic N, P and K to serpentine soils has been shown by various authors to stimulate shoot biomass production and Ni yield of Ni-hyperaccumulators (Chaney et al. 1998). An early study by Robinson et al. (1997) found Odontarrhena bertolonii (syn. Alyssum bertolonii) increased its biomass by 3-fold compared to control plants after the addition of N, P, K as NH4NO3, NaH2PO4 and KCl (each at 10 g m−2). Shoot biomass production and Ni yield of Odontarrhena muralis (syn. Alyssum murale) was enhanced in an Oregon ultramafic soil in the USA using P and N chemical fertilizers (Li et al. 2003). Bani et al. (2015) carried out field experiments using O. muralis in ultramafic Vertisols in Albania. The application of NPK fertilizers (applied as NH4NO3, Ca(H2PO4)2 and K2SO4 at 100 kg ha−1 of P and K, 65 kg ha−1 of Ca and a split 100 kg ha−1 N application) significantly increased the shoot yield of O. muralis without reducing Ni concentration. It is worth noting that in this study, O. serpyllifolia which is native to the site where the serpentine soil was collected showed significantly better growth on the NPK-fertilized soil than the Iranian hyperaccumulators, which may reflect its adaptation to this soil. In contrast, only a few studies have considered incorporating organic amendments into Ni phytomining systems and the potential improvements in organic matter content and fertility of serpentine soils (Alvarez-Lopez et al. 2016b; Nkrumah et al. 2018). Broadhurst and Chaney (2016) found O. muralis did not survive in a serpentine soil (Brockman variant) amended with mature dairy manure compost (added at 10% w/w on DW basis). Contrasting results were found by Alvarez-Lopez et al. (2016b) who evaluated the effect of adding composted sewage sludge to the same serpentine soil as that used in this study and found beneficial effects on soil properties and plant growth. Compost addition improved CEC, Ca/Mg ratio and organic matter content of the soils, and an addition rate of 5% w/w (calculated on DW basis) was optimal for biomass production of O. serpyllifolia. Moreover, these authors also showed that the growth of this Ni-hyperaccumulator was significantly better in compost-amended soil than in NPK-fertilized soils (100:100:125 kg NPK ha−1). However, they highlighted the possible risks of excess concentrations of potentially phytotoxic elements (PPE), particularly Cu, when incorporating these types of organic wastes into phytomining systems. This study confirmed the results of Álvarez-López et al. (2016b) in that soil fertility was significantly improved and the Ni-hyperaccumulator O. serpyllifolia grew better in the ultramafic soil after addition of organic matter rather than inorganic fertilization. Here the positive effect was obtained using cow manure and we also showed that the benefits to biomass production was common to all three Ni-hyperaccumulators, suggesting that the addition of organic soil amendments could be an interesting option for phytomining systems. In addition to an improvement in soil nutrition the addition of manure in this study led to an improvement in physical properties (soil density, water retention capacity) which is likely to have been even more pronounced after plant growth and root proliferation. Moreover, the increase in DOC and respiration after manure addition reflect a higher soil biological activity. Finally, the manure-amended soils used in this study did not present significantly higher concentrations of PPE (DTPA- or Sr(NO3)2-extractable) compared to the NPK-fertilized soils, thus avoiding potential risks of phytotoxicity which may occur with other types of organic wastes. The benefits of using organic amendments are likely due to the provision of plant nutrients but also to improvements in soil structure, porosity and water holding capacity. Plants grown in manure-amended soils had a higher shoot concentration of P, K and Ca, and lower shoot Mg concentration. This was consistently seen amongst all three hyperaccumulators. Furthermore, the Ca:Mg shoot concentration quotient was significantly higher in all three species when grown in manure-amended soil compared to NPK-fertilized soil, and increased from 2.2 to 3.6, 0.8 to 1.4, and 0.4 to 0.9 in non-inoculated O. serpyllifolia, O. inflata and O. bracteata, respectively. The Ca:Mg quotient has been shown to be an important factor for growth and Ni uptake of hyperaccumulators (Ghasemi and Ghaderian 2009). Shoot Ni concentration in O. inflata and O. bracteata was higher in plants grown in manure-amended than NPK-fertilized soil, and this coincided with the increase in the Ca:Mg quotient. However, this effect was not observed in O. serpyllifolia. In NPK-fertilized soil, sufficient biomass for determining all the physiological parameters was only obtained for O. serpyllifolia. For this species, lipid peroxidation (MDA content was 38% lower), total phenolic content and CAT activity was also found to be lower in plants grown in the manure-amended soil than in NPK-fertilized soils.

The shoot biomass of O. serpyllifolia and O. inflata in the manure-amended soil reached up to around 2 g DW plant−1. This is similar to the shoot DW yields obtained by Álvarez-López et al. (2016b) for O. serpyllifolia and O. bertolonii, and also those obtained by Cabello-Conejo et al. (2014b) for O. muralis, growing in the same soil (with either compost amendment or NPK fertilization, respectively). The Ni-hyperaccumulator O. serpyllifolia has been suggested to present a limited application potential in phytomining systems compared to other Odontarrhena sp., such as O. muralis or O. corsica, due to their low productivity (Cabello-Conejo et al. 2013). These results however suggest that proper agronomic management could increase the biomass productivity of these species and hence their application in phytomining.

Soil Ni and Co availability decreased after incorporating both manure and NPK fertilizer and after a 2-week equilibrium period, i.e. before planting. Several authors have also observed a significant decrease in soil metal extractability after application of amendments to soil, as a result of complexation or precipitation processes (Vangronsveld et al. 2009). Álvarez-López et al. (2016b) observed a significant reduction in the Ni availability (assessed via NH4Cl soil extractions) of the same serpentine soil when amended with composted sewage sludge which they attributed to the formation of organo-metallic insoluble complexes. However, in this study, Ni and Co availability (assessed via DTPA and Sr(NO3)2 soil extractions) was actually higher after plant growth than before planting and also higher than the original untreated soil. At harvest, no significant differences were observed between soils in DTPA-extractable metal concentrations, while Sr(NO3)2-extractable concentrations of Ni were lower in manure-amended soil compared to NPK-fertilized soil. The latter could reflect the more prolific plant growth and a higher Ni uptake by plants grown in the manure-amended soil. Shoot Ni concentrations were higher in O. inflata and O. bracteata grown in manure-amended than NPK-fertilized soil, while the opposite was observed in O. serpyllifolia. In accordance, shoot Ni yields in non-inoculated plants were significantly higher in O. inflata and O. bracteata grown in manure-amended soil but did not differ significantly in the case of O. serpyllifolia which showed better growth in NPK-fertilized soil. Shoot Ni concentrations and Ni yields were consistently higher for O. inflata and O. serpyllifolia, while O. bracteata showed a lower capacity for Ni accumulation and hence phytomining application. Overall, the results indicate a potential use of organic amendments to maximize biomass production and Ni yield of some hyperaccumulating Odontarrhena spp. within phytomining systems but this will require field-scale evaluations.

The rhizobacterial strains tested here were all originally isolated from O. serpyllifolia and identified as having plant growth-promoting traits. In two previous studies, the strains LA44 and SA40 (both belonging to the Paenarthrobacter genus) and Pseudoarthrobacter oxydans strain SBA82, were found to promote the growth of this hyperaccumulator when grown in two different serpentine soils (Becerra-Castro et al. 2013; Cabello-Conejo et al. 2014a). Cabello-Conejo et al. (2014a) used the same serpentine soil as that used here. Strain SA40 also stimulated O. serpyllifolia growth in a Cd- and Ni-contaminated non-serpentine agricultural soil (Cabello-Conejo et al. 2014a). A beneficial effect of these inoculants was therefore expected on the growth of O. serpyllifolia in this study. Surprisingly, no effect was observed in the NPK-fertilized soil and SA40 had no beneficial effect in either soil. This is in contrast to the results of the previous studies and could be due to the fact that neither of the previous studies applied fertilizers to the soils. In nutrient-deficient serpentine soils the PGP traits of the bacterial inoculants are more likely to be activated, whereas this effect may be lost when fertilizing the soil with readily available forms of NPK. Despite this, a strong and significant bacterial-induced stimulation in the growth of this hyperaccumulator was observed in the manure-amended soil after inoculation with strains LA44 and SBA82, as well as Stenotrophomonas sp. strain MA98 and Rhodococcus fascians strain ALM1. These results confirm the potential use of the strains LA44 and SBA82 to promote the phytomining capacity of this species, and also show a potential for strains MA98 and ALM1 which have not previously been tested in inoculation studies. Moreover, one of the aims of this study was to assess whether or not these plant-associated bacteria have a beneficial effect on other Ni-hyperaccumulating Odontarrhena spp. The response of plants to the bacterial inoculants differed between the three Odontarrhena spp. and depended on the soil in which they were grown, as reflected through the significant plant species*strain and amendment*strain interaction factors. However, beneficial effects of inoculation were found for at least some strains and in some cases the PGP effect was common to all three hyperaccumulators, indicating that these inoculants could have a generalized use in phytomining systems based on the cultivation of Odontarrhena spp. This was the case for the siderophore-producing Pseudoarthrobacter oxydans strain SBA82 which enhanced shoot DW yield of O. inflata in both soils (by 1.6-fold in NPK-fertilized and 6.4-fold in manure-amended soil), of O. bracteata in NPK-fertilized soil and of O. serpyllifolia in manure-amended soil. The IAA-producer Paenarthrobacter sp. LA44 and siderophore-producer Stenotrophomonas sp. MA98 promoted growth in two of the three hyperaccumulators, O. serpyllifolia and O. bracteata. The PGP effect of strain MA98 is particularly interesting since this strain was found in association with the Ni-hyperaccumulator O. serpyllifolia in three different ultramafic outcrops across the Iberian Peninsula but was not present in the culturable bacterial community associated with two non-accumulators, Dactylis glomerata and Santolina semidentata (Alvarez-Lopez et al. 2016a).

Further research is necessary to fully understand the mechanisms involved in this bacterial-induced growth promotion and why the same isolate with specific PGP traits does not induce the same plant response under distinct soil conditions. Alvarez-López et al. (2017) studied the effects of combining soil amendment (using composted sewage sludge) with bacterial inoculation on the growth and Cd/Zn accumulation capacity of Salix caprea and Nicotiana tabacum in contaminated mine tailings. Bacterial inoculants stimulated tobacco biomass when this was grown in the compost-amended soil, but the same effect was not observed in unamended soil. These authors attributed this to the requirement of the bacterial strains for a minimal amount of organic matter for their successful establishment in the soil and protection against metal toxicity. The same reasoning could explain why the most pronounced effects of the bacterial inoculants in this study were found in the manure-amended soil.

Several authors have identified rhizobacterial isolates associated with Ni-hyperaccumulators with capacity to enhance soil Ni bioavailability (Abou-Shanab et al. 2006; Becerra-Castro et al. 2013; Ma et al. 2009). However, bacterial inoculation generally had no effect on soil Ni availability in manure-amended soils, and only a few isolated effects were observed in NPK-fertilized soils. In the latter, three bacterial strains (SA40, MA98 and ALM1) induced an increase in DTPA-extractable Ni concentrations, but this was not reflected in plant Ni accumulation. Becerra-Castro et al. (2013) showed that strains LA44 and SBA82 (also used here) were able to mobilize Ni from ultramafic rock in in vitro studies, however, neither of these strains influenced extractable soil Ni concentrations in this study. Nonetheless, three strains (LA44, MA98 and ALM1) significantly increased shoot Ni concentrations, but surprisingly this was only observed in the original host of these isolates, O. serpyllifolia. The combined increase in shoot Ni concentration and shoot DW yields led to significant increases in phytoextracted Ni. On the other hand, despite a reduction in shoot Ni concentration (only in manure-amended soil) in O. inflata after inoculation with strain SBA82 the stimulation in plant growth led to a significant increase in phytoextracted Ni. Similarly, the Ni phytoextraction capacity of O. bracteata was significantly improved after inoculation with strain MA98 due to the stimulation in plant growth. These results confirm those of other authors showing enhanced Ni removal by hyperaccumulators using rhizobacterial inoculants (Abou-Shanab et al. 2003) and non-hyperaccumulators (Ma et al. 2009).

Metal toxicity to plants has been assessed in many situations and plant types ((hyper)accumulators and non-accumulators) using stress indicators, such as antioxidant enzyme activities, lipid peroxidation or phenolic metabolites (Gratão et al. 2005). The antioxidant defense system includes enzymatic components (SOD, CAT, APX and GR) and non-enzymatic components (such as carotenes, anthocyanin and tocopherols), which interact with stress metabolites and each other (Gratão et al. 2005; Verma and Dubey 2003). Malondialdehyde (MDA) and H2O2 are indicators of lipid peroxidation and oxidative damage in plants (Dewir et al. 2006). To the best of our knowledge this is the first study to evaluate plant-microbe interactions on the physiological status of Ni-hyperaccumulators in natural metal-enriched serpentine soils. The inoculation process did not apparently invoke plant stress (no increase in any of the stress parameters assessed was observed). O. inflata showed particularly high activities of catalase (CAT) but this species has previously been shown to present high activities of this enzyme (Ghasemi et al. 2009). In some cases, inoculated plants showing growth promotion also tended to present lower activities of antioxidative enzymes (particularly SOD and CAT), alongside lower concentrations of MDA and H2O2, indicating a protective effect of these inoculants on the plants. Superoxide dismutase (SOD) activity was significantly reduced in SBA82-inoculated O. inflata and O. serpyllifolia in manure-amended soil, and catalase (CAT) and ascorbate peroxidase (APX) were lower in the same treatment in O. inflata. In parallel, MDA and H2O2 concentrations decreased in O. inflata, which coincided with the highest shoot DW yields and phytoextracted Ni and indicates a lower level of stress in these plants and improved Ni removal capacity. Similar reductions were found in CAT activity of O. bracteata and GR activity of O. serpyllifolia after inoculation with MA98 in manure-amended soils, where the highest shoot DW yields and Ni removal were obtained. Moreover, in O. bracteata a parallel reduction in MDA and H2O2 was observed. Antioxidant enzymes activities were correlated with the stress markers MDA and H2O2 in O. inflata and O. bracteata. Other studies have reported improved growth and reductions in the production of MDA and H2O2 in plants exposed to As or Pb after inoculation with PGP bacterial strains (Singh et al. 2016; Janmohammadi et al. 2013). No clear effects of bacterial inoculation were observed on PAL or TAL activities in leaves (enzymes in the phenylpropanoid pathway) or the content of phenolic compounds.