1 Bacterial community diversity in the rhizosphere of 2 hyperaccumulator species from Island (Malaysia) 3 4 Séverine Lopez,1 Antony van der Ent,1,2 Sukaibin Sumail,3 John B. Sugau,3 Matsain Mohd 5 Buang,4 Zarina Amin,5 Guillaume Echevarria,1,2 Jean Louis Morel1 and Emile Benizri1 6 7 1Université de Lorraine, INRAE, Laboratoire Sols et Environnement, 54000, Nancy, France. 8 9 2Centre for Mined Land Rehabilitation, Sustainable Minerals Institute, The University of 10 Queensland, St Lucia, 4072, QLD, Australia. 11 12 3Sabah Parks, Kota Kinabalu, Sabah, Malaysia. 13 14 4Forest Research Centre, Sabah Forestry Department, Sandakan, Sabah, Malaysia. 15 16 5Biotechnology Research Institute, Universiti Malaysia Sabah, Kota Kinabalu, Sabah, 17 Malaysia. 18 19

20 SUMMARY 21 The Island of Borneo is a major biodiversity hotspot, and in the Malaysian state of Sabah, 22 ultramafic soils are extensive and home to more than 31 endemic nickel hyperaccumulator 23 . The aim of this study was to characterize the structure and the diversity of the 24 rhizosphere bacterial communities of several of these nickel hyperaccumulator plants and 25 factors that affect these bacterial communities in Sabah. The most abundant phyla were 26 Proteobacteria, Acidobacteria and Actinobacteria. At family level, Burkholderiaceae and 27 Xanthobacteraceae (Proteobacteria phylum) were the most abundant families in the 28 hyperaccumulator rhizospheres. Redundancy analysis based on soil chemical analyses and 29 relative abundances of the major bacterial phyla showed that abiotic factors of the studied sites 30 drove the bacterial diversity. For all R. aff. bengalensis rhizosphere soil samples, irrespective 31 of studied site, the bacterial diversity was similar. Moreover, the Saprospiraceae family showed 32 a high representativeness in the R. aff. bengalensis rhizosphere soils and was linked with the 33 nickel availability in soils. The ability of R. aff. bengalensis to concentrate nickel in its 34 rhizosphere appears to be the major factor driving the rhizobacterial community diversity unlike 35 for other hyperaccumulator species. 36 37

38 INTRODUCTION 39 Ultramafic soils cover 3% of the terrestrial surface (Guillot and Hattori, 2013; Echevarria, 40 2018). The largest areas are found in Cuba, , Philippines, Indonesia, Turkey and 41 Malaysia (van der Ent, 2015). Due to the nature of the ultramafic bedrock, they naturally contain 42 high metal concentrations, particularly nickel (Ni), (Co), chromium (Cr) and manganese 43 (Mn; Proctor and Woodell, 1975). They are also characterized by high concentrations of iron 44 (Fe) and magnesium (Mg), with a low calcium/magnesium ratio (Harrison and Rajakaruna, 45 2011). In addition, they have low levels of macronutrients such as nitrogen (N), phosphorus (P) 46 and potassium (K; Shallari et al., 1998; Boyd and Jaffré, 2009) and a low water holding capacity 47 (Kay et al., 2011). In the state of Sabah in Malaysia (Northern tip of Borneo Island), ultramafic 48 soils are extensive and cover an area of approximatively 3500 km2 (Proctor et al., 1988). This 49 region has a great floristic richness with about 8000 species, including more than 50 5000 species inventoried in the Kinabalu Park (Beaman, 2005, van der Ent et al., 2015a). The 51 park was established in 1964 as the first fully protected area in Sabah, and has an area of 754 52 km2 around Mount Kinabalu, the highest mountain on the island of Borneo (4095 m a.s.l.). 53 54 Because of their extreme properties, ultramafic soils support a particular flora with high rates 55 of endemism which have evolved both morphological and physiological adaptations and allow 56 it to survive in these environments. Within this flora, hyperaccumulator plants have the ability 57 to concentrate metals in their aerial parts, whatever the concentration found in the soil (Baker, 58 1981). There are about 700 taxa of plants that are known to hyperaccumulate one or more 59 metals, and around 90% of these accumulate Ni (van der Ent et al., 2013; Pollard et al., 2014; 60 Galey et al., 2017; Reeves et al., 2017; van der Ent et al., 2017; Kidd et al., 2018; Nkrumah et 61 al., 2019). In the case of Ni hyperaccumulator plants, the threshold of Ni accumulation is at 62 least 1000 mg Ni kg-1 of dry weight in aerial parts (0.1% of plant biomass; Brooks et al., 1977). 63 Nickel hyperaccumulator plants occurring in Sabah mainly belong to the botanical families of 64 , Rubiaceae, Salicaceae and (Reeves, 2006). The known Ni 65 hyperaccumulators of this region are Rinorea aff. bengalensis and R. aff. javanica (Violaceae; 66 Brooks et al., 1977; Brooks and Wither, 1977), Phyllanthus balgooyi and Phyllanthus 67 rufuschaneyi (Phyllanthaceae; Baker et al., 1992; Hoffmann et al., 2003; MesjaszPrzybylowicz 68 et al., 2016; Bouman et al., 2018), Dichapetalum gelonioides (Dichapetalaceae; Baker et al., 69 1992; Nkrumah et al., 2018b), Psychotria sarmentosa (Rubiaceae; Reeves, 2003) and Shorea 70 tenuiramulosa (Dipterocarpaceae; Proctor et al., 1989). Since prospecting campaigns conducted 71 in Kinabalu Park, new nickel hyperaccumulator species have been discovered such as Actephila 72 alanbakeri (Phyllanthaceae; van der Ent et al., 2015a) in Kinabalu Park or Antidesma montis- 73 silam (Phyllanthaceae; Nkrumah et al., 2018c) at Mount Silam. 74 75 Due to the high concentrations of certain elements (Ni, Mg), ultramafic soils harbour specific 76 microbial communities (Schipper and Lee, 2004; Rajkumar et al., 2009; Branco, 2010; Lopez 77 et al., 2019a, 2019b). It has been shown that the presence of these elements can reduce microbial 78 biomass (Wang et al., 2008), affect microbial enzymatic activities (Kuperman and Carreiro, 79 1997; Kandeler et al., 2000) and consequently can alter the mineralization of soil organic matter 80 (Epelde et al., 2012). Based on highthroughput Illumina sequencing approaches, recent studies 81 have revealed that Proteobacteria, Actinobacteria and Acidobacteria are the major phyla in the

82 rhizosphere of Ni hyperaccumulator plants growing on ultramafic soils (Venter et al., 2018; 83 Visioli et al., 2019; Lopez et al., 2019a, 2019b). These phyla are commonly encountered as the 84 predominant taxa in agricultural, polluted or ultramafic soils (Janssen, 2006; Rastogi et al., 85 2010). Unlike other environments, Ni hyperaccumulator plant rhizospheres typically have a 86 high representation of the Chloroflexi phylum (Lopez et al., 2017). 87 88 The Island of Borneo is known to be one of the major hotspots of biodiversity for plants and 89 vertebrates (Myers et al., 2000), however there is a paucity of data available on the soil bacterial 90 diversity present in this region (Tripathi et al., 2012; Lee-Cruz et al., 2013). There is currently 91 no known study on the rhizobacterial diversity of Ni hyperaccumulator plants from Borneo 92 Island. The rhizobacterial communities can, on the one hand, provide mineral elements 93 necessary for the plant and, on the other hand, improve their resistance to stress such as those 94 induced by metals (Lebeau et al., 2008). Moreover, the bacteria can also affect mobility and the 95 availability of metals present in the soil, favouring or not their transfer from the rhizosphere to 96 the plant and therefore phytoextraction (Kidd et al., 2009; Sessitsch et al., 2013). The aims of 97 this study were to characterize the structure and the diversity of the rhizosphere bacterial 98 communities of Ni hyperaccumulator plants, and to highlight the physicochemical factors that 99 may influence the composition of these bacterial communities. 100 101 RESULTS 102 103 Study area and sample collection 104 During fieldwork, 52 Ni hyperaccumulator plants and their associated rhizosphere soil were 105 sampled from 10 different ultramafic sites (Table 1 and Fig. 1) around Mount Kinabalu Park. 106 The samples were collected associated with the following species: Flacourtia kinabaluensis 107 Sleumer (Salicaceae), Mischocarpus sundaicus Blume (Sapindaceae), Phyllanthus balgooyi 108 Petra Hoffm. & A.J.M. Baker, P. rufuschaneyi Welzen, R.W. Bouman & Ent (Phyllanthaceae), 109 Psychotria sarmentosa Blume (Rubiaceae), Rinorea aff. bengalensis (Wall.) Kuntze, R. aff. 110 javanica Kuntze, Rinorea sp. (Violaceae) and Walsura pinnata Hassk. (Meliaceae). The highest 111 plant diversity was found at sites 06 and 10, with six plant species harvested. Site 01, with the 112 highest elevation (1410 m a.s. l.) has only one species: P. balgooyi. Flacourtia kinabaluensis 113 and M. sundaicus showed a better geographical repartition (found in 6 of the 10 sites), followed 114 by P. balgooyi and P. sarmentosa (5 sites), R. aff. javanica (4 sites), R. aff. bengalensis, the 115 undetermined Rinorea sp. and W. pinnata (3 sites) and finally P. rufuschaneyi (2 sites). 116 117 Chemical analyses of rhizosphere soil samples from each site 118 Abiotic soil parameters were measured as seen in Tables 2–4. The mean pH ranged from 6.0 119 (Sites 01 and 05) to 7.4 (Site 07) with an average pH of 6.6 (Table 2). The C/N ratio of the 120 rhizosphere soils showed significant differences (P < 0.05) depending on the site. Sites 01 and 121 05 had statistically the highest C/N ratio (21 and 20 respectively) in comparison to sites 02, 03, 122 06, 07 and 09 (lower than 14). Due to its rhizosphere origin, the soil had an organic C content 123 corresponding to an average of 96% of the total C. The elements that were most extracted by 124 DTPA in the rhizosphere soils were Ni (193 μg g-1), Fe (140 μg g-1) and Mn (102 μg g-1) as 125 seen in Table 3. There were no significant differences for DTPA-extractable Ni and Zn among

126 the different rhizosphere soil samples. Although Cr was present in soil samples (total Cr), this 127 element was poorly available in the rhizosphere soils (<1 μg g-1). While site 09 had the highest 128 Co DTPA-extractable (29 μg g-1), Site 01 had the highest Cu (14 μg g-1) and Pb (5.8 μg g-1) 129 DTPA-extractable. Mean total CEC was 27 cmol+ kg-1 with the highest values for both sites 05 130 (43 cmol+ kg-1) and 08 (40 cmol+ kg-1), compared to sites 09 (12 cmol+ kg-1) and 10 (17 cmol+ 131 kg-1; Table 4). The values of the exchangeable elements were relatively low, except for Mg 132 (average of 15 cmol+ kg-1) and Ca (average of 6.9 cmol+ kg-1). Site 05 showed the highest 133 concentrations of exchangeable elements, while sites 09 and 10 had the lowest. There were no 134 significant differences for exchangeable Ca, Mn and Ni between the different sites. 135 136 Ni accumulation in plant species 137 All plant species can be considered as Ni hyperaccumulator plants in relation to the 138 concentrations of this metal in their leaves even if high variability in concentrations for the 139 same plant species were reported (Table 5). This was the case for M. sundaicus, which can 140 accumulate in its leaves from 600 to 3600 μg of Ni g-1. Similarly, for P. sarmentosa, the 141 accumulation of Ni varied from 42 to 20,800 μg of Ni g-1. The highest average Ni levels in the 142 leaves were found respectively for P. rufuschaneyi (19,000 μg of Ni g-1), R. aff. bengalensis 143 (16,100 μg of Ni g-1) and R. aff. javanica (15,200 μg of Ni g-1). Bioconcentration factor (BCF) 144 and translocation factor (TF) of nickel were estimated for the different Ni hyperaccumulators 145 sampled (Table 5). Phyllanthus rufuschaneyi showed the best BCF (factor of 95) which was 146 statistically higher than that of M. sundaicus (factor of 16). However, M. sundaicus had one of 147 the highest TF (factor of 9.6), following by F. kinabaluensis (factor of 9.3) and W. pinnata 148 (factor of 11). 149 150 Bacterial community diversity from Ni hyperaccumulator rhizospheres 151 Upon bioinformatic analysis of the 52 rhizosphere DNA samples, 2,189,071 sequences were 152 obtained and grouped under 2475 different Operational Taxonomic Units (OTUs) belonging to 153 23 phyla. The α-diversity indices (observed OTUs, Chao1 and Shannon) for the rhizosphere 154 soil samples did not have significant differences between the different Ni hyperaccumulator 155 species (data not shown). However, differences can be highlighted when considering the α- 156 diversity of each site (Table 6). For the 2475 observed OTUs, the range was from 853 (Site 01) 157 to 1748 (Site 06), and the Shannon index varied from 7.48 (Site 07) to 9.40 (Site 06). The Chao1 158 index, which represents OTUs by considering rare bacterial species (singletons and 159 doubletons), is an extrapolation of the total number of OTUs in the sample that could have been 160 found with a perfect depth of sequencing. At least, 83% of the diversity of the Ni 161 hyperaccumulator rhizosphere samples were represented (observed OTUs/ Chao1 × 100). The 162 relative abundances of each phylum were estimated (Fig. 2). Furthermore, 1.2% of the 163 sequences could not be taxonomically affiliated. Among the 10 most abundant phyla, 164 Proteobacteria (46%) and Acidobacteria (21%) were most represented at all sites regardless of 165 the plant species. This was followed by Actinobacteria (6.3%), Rokubacteria (5.5%) and 166 Bacteroidetes (4.3%). Firmicutes were significantly more abundant in site 07 (17%) than in 167 other sites (1.8%). Conversely, this site showed the lowest representativeness of Acidobacteria 168 (12% for an average of 22% for the other sites), Chloroflexi (2.5% against 3.3%) and 169 Planctomycetes (1.9% against 4.0%). No significant difference was observed between the

170 different sites for the phyla of Actinobacteria, Bacteroidetes and Latescibacteria. When grouped 171 in function of the plant species (data not shown), significant differences were found for the 172 Bacteroidetes phylum with a higher relative abundance in M. sundaicus rhizosphere (7.8%) 173 compared to P. rufuschaneyi (3.1%) and P. sarmentosa (2.5%). In addition, the Rokubacteria 174 and Gemmatimonadetes phyla presented higher relative abundances in the rhizospheres of R. 175 aff. bengalensis (8.6 for Rokubacteria and 3.6% for Gemmatimonadetes) compared to those of 176 P. balgooyi (3.6% and 1.2%) and P sarmentosa (3.3% and 1.0%). 177 178 At the family level, a network analysis was carried out on the 66 bacterial families having a 179 relative abundance greater than 1% in Ni hyperaccumulator rhizospheres (Fig. 3 and Supporting 180 Information Table S1). Among the 66 families, 45% belonged to the Proteobacteria phylum, 181 17% to Firmicutes and 9% to the Bacteroidetes. In total, nine families were common to the nine 182 different Ni hyperaccumulator plants. Among these common families, five belonged to the 183 Proteobacteria phylum (Burkholderiaceae, Desulfarculaceae, Haliangiaceae, 184 Nitrosomonadaceae and Xanthobacteraceae), two were from the Acidobacteria phylum 185 (Blastocatellaceae and Pyrinomonadaceae) and one was from the Actinobacteria 186 (Micromonosporaceae) and the Gemmatimonadetes (Gemmatimonadaceae) phyla. 187 Burkholderiaceae (relative abundance of 5.4% on average) and Xanthobacteraceae (4.8%), 188 from the Proteobacteria phyla, were the most abundant families in the different soils. 189 Phyllanthus balgooyi samples presented a particularity with six families which were only found 190 at more than 1% of representativeness in the rhizosphere of this plant species, while M. 191 sundaicus had only two own families, F. kinabaluensis, Rinorea sp. and R. aff. bengalensis 192 possessed one family and P. rufuschaneyi, P. sarmentosa, R. aff. javanica and W. pinnata had 193 not specific family present in their rhizosphere soils. 194 195 Relation between bacterial community diversity and soil abiotic parameters 196 To establish any links between the bacterial community diversity and physicochemical 197 parameters of the different sites, a Redundancy Analysis (RDA) was performed between the 198 relative abundance of the 10 major phyla and the soil physicochemical parameters (Fig. 4). The 199 envfit function of the Vegan package was used to keep only the abiotic variables significantly 200 correlated (P < 0.10) with the relative abundances of the bacterial phyla. The considered abiotic 201 parameters were the percentages of total nitrogen, total carbon and organic C, concentrations 202 of Cu, Mn, Ni, Co, Pb, Fe and Zn extractable with DTPA, exchangeable Ni, K and Mg, the total 203 CEC and the pH. The main plan F1–F2 explained 89% of the total variability. Axis F1, which 204 explained 62% of the total variability, clearly discriminated R. aff. bengalensis rhizosphere 205 samples (Rinb) among all samples. Indeed, excepted for the sample Rinb-02b, all rhizosphere 206 soil samples of R. aff. bengalensis (Rinb) had both negative abscissa and ordinate. Their 207 discriminations were based on basic pH, high Ni exchangeable and DTPA-extractable amounts 208 and conversely low concentrations of Cu, Co and Pb DTPA-extractable. In addition, these 209 samples showed lower relative abundances of Proteobacteria and Firmicutes. No clear 210 discrimination could be established for the rhizosphere soil samples from the other Rinorea 211 species (R. aff. javanica: Rinj and Rinorea sp.: Rin). Axis 2 explained 27% of the total 212 variability and globally discriminated rhizosphere samples from P. rufuschaneyi (Phyr; 213 excepted Phyr-02a, Phyr-02b and Phyr-02c), which had positive ordinates and were negatively

214 correlated with C and N percentages, total CEC and exchangeable Mg and K concentrations. 215 Furthermore, these soil samples were characterized by high abundances of Proteobacteria and 216 Acidobacteria. Finally, the rhizosphere samples from P. balgooyi and those from other 217 hyperaccumulator plant species were distributed over the whole F1–F2 plan, and it was not 218 possible to correlate them with the studied rhizosphere soil physicochemical parameters. 219 Spearman correlations were calculated between the relative abundance of the 10 most abundant 220 phyla and the rhizosphere soil physicochemical parameters (Fig. 5). Only significant 221 correlations (P < 0.1) were kept. DTPAextractable Ni was positively correlated with Firmicutes 222 phyla (P = 0.005), Gemmatimonadetes (P = 0.014), Latescibacteria (P = 0.006), Planctomycetes 223 (P = 0.002), Rokubacteria (P = 0.002) but negatively correlated with Proteobacteria (P = 0.015). 224 Irrespective of phyla, DTPAextractable Ni correlations were opposite to altitude correlations. 225 Gemmatimonadetes and Rokubacteria phyla, with the highest relative abundances in R. aff. 226 bengalensis rhizospheres, showed the same correlations with soil physicochemical parameters 227 and were positively correlated with pH (P < 0.001 and P = 0.004) and exchangeable Mg (P = 228 0.072 and P = 0.072), but were negatively correlated with DTPA-extractable Co (P < 0.001 and 229 P = 0.007), Cu (P < 0.001 and P = 0.004) and elevation (P = 0.003 and P = 0.002). Organic C 230 and total C were positively correlated with Bacteroidetes (P = 0.042 and P = 0.088), Firmicutes 231 (P = 0.001 and P < 0.002) and Proteobacteria (P = 0.026 and P = 0.024) while other phyla were 232 negAtively correlated with these soil parameters. 233 234 To establish the influence of abiotic factors on family abundances, Spearman correlations were 235 calculated between the relative abundance of bacterial families and the rhizosphere soil 236 physicochemical parameters (Fig. 6). Only significant correlations (P < 0.1) were maintained. 237 In order to aid interpretation, the analysis was restricted to nine families common for the nine 238 Ni hyperaccumulator plant rhizospheres, and the 11 families found only in a plant rhizosphere 239 at a relative abundance upper than 1%. Thereby, 20 bacterial families were linked with 19 240 abiotic parameters. Among the common families for the nine plant species, 241 Gemmatimonadaceae (Gemmatimonadetes phylum) and Haliangiaceae (δ-Proteobacteria) 242 families had opposite responses, with positive correlations for Gemmatinomanadaceae to pH 243 (P < 0.001) and exchangeable Mg (P = 0.058) conversely to Haliangiaceae (P = 0.002 and P < 244 0.001), and negative correlations to Co-DTPA (P = 0.001), Cu-DTPA (P < 0.001) and elevation 245 (P = 0.003) while these correlations were positive for Haliangiaceae (P = 0.007 for CoDTPA P 246 = 0.100 for Cu-DTPA and P = 0.036 for elevation). In the case of the families with more than 247 1% of representativeness in one rhizosphere plant species, Saprospiraceae found in R. aff. 248 bengalensis rhizosphere was the family the most positively linked with Ni-DTPA (P < 0.001). 249 The six families from P. balgooyi rhizosphere soil had the same trends and all were negatively 250 correlated with pH values (P < 0.001 to P = 0.015) and positively with DTPA-extractable Co 251 (P < 0.001 to P = 0.058) and elevation (P < 0.001 to P = 0.010). 252 253 Using the Vegan package on R software, the graphical representation NMDS (non-metric 254 multidimensional scaling) allowed for the comparison of bacterial community diversity (at 255 OTU level) from the different plant species rhizospheres with the Ni DTPA-extractable 256 concentrations found in the soil samples (Fig. 7). With this representation, we saw that R. aff. 257 bengalensis samples (Rinb) had a similar bacterial diversity because samples were close on the

258 NMDS plan. Moreover, these samples presented a Ni DTPA-extractable concentration higher 259 than 220 mg kg-1. As for the R. aff. bengalensis samples, rhizosphere soils from R. aff. javanica 260 (Rinj) had a close bacterial community diversity with the four samples clustered together. Based 261 on these results, no clear group can be highlighted for the other samples. 262 263 DISCUSSION 264 Ultramafic soils are multiple stressed environments due to the high concentration of potentially 265 toxic elements (Ni, Co, Mn and Mg), and low nutrient status (N, P, K and Ca; Proctor, 2003; 266 Galey et al., 2017). The totality of these abiotic factors constitutes the ‘serpentine syndrome’ 267 (Kazakou et al., 2008). The rhizosphere soils collected in this study had total Ni, Mg, Mn, Fe 268 concentrations and Mg/Ca ratios comparable to those expected for ultramafic soils worldwide 269 (Reeves, 1999; Freitas et al., 2004; Ghaderian et al., 2007; Kazakou et al., 2010). However, 270 variations in the concentrations of trace elements and the Mg/Ca ratio were found between the 271 different studied sites. This can be attributed to alteration and leaching processes, combined 272 with biological activity, which may differ depending on the location of the sampling sites 273 (Alexander, 2004). Indeed, it is known that weathering plays an important role in soil processes, 274 especially in tropical climates. Malaysia, like Indonesia, Brazil, Cuba, Australia and New 275 Caledonia, has large ultramafic massifs whose alteration generates the formation of a large 276 lateritic cover (i.e., Ferralsols) which can reach several meters in depth (Nahon, 2003). 277 278 Around Kinabalu Park in Sabah, many Ni hyperaccumulator plant species have been identified 279 and most belong to the Plant Kingdom Order of the (families of Dichapetalaceae, 280 Phyllanthaceae, Salicaceae, Violaceae; van der Ent et al., 2015a). The Malpighiales is one of 281 the largest Orders of flowering plants, containing approximately 16,000 species in 42 families 282 globally, accounting for approximately 7.8% of (Nkrumah et al., 2018a). Based on 283 our sampling, the plants with the highest Ni concentrations in their leaves were P. rufuschaneyi 284 (33,600 μg g-1), R. aff. bengalensis (29,600 μg g-1) and R. aff. javanica (20,800 μg g-1). 285 Moreover, P. rufuschaneyi had the higher BCF measured (ratio of 95). In previous studies, the 286 concentrations found in the leaves of these plants ranged from 100 to 10,700 μg g-1 for P. 287 rufuschaneyi (van der Ent et al., 2017), from 1 to 17,500 μg g-1 for R. aff. bengalensis (Brooks 288 and Wither, 1977; van der Ent et al., 2017) and from 2 to 6000 μg g-1 for R. aff. javanica (Brooks 289 et al., 1977; van der Ent et al., 2013, 2020). In contrast to the results of this study, lower leaf 290 concentrations were measured by these authors in both Rinorea species. This genus is known 291 to be a facultative Ni hyperaccumulator plants, meaning that it can be found on both ultramafic 292 and non-ultramafic soils. However, the ability of hyperaccumulation only occurs in specimens 293 growing on ultramafic soils (Brooks and Wither, 1977; Pollard et al., 2014) and Sabah has an 294 estimated 8000 plant species, of which, over half are known to occur on ultramafic soils (van 295 der Ent et al., 2015b). This makes this region a global hotspot for Ni hyperaccumulator plants 296 species. As the sampling in this study focused only on ultramafic soils, this could explain the 297 higher Ni concentrations found in the leaves of the two Rinorea species. Even if P. rufuschaneyi 298 had the higher Ni concentration in leaves and the higher BCF, the highest TF were found for 299 species with lower Ni in leaves, as W. pinnata (maximum of 3100 μg Ni g-1 in leaves with a TF 300 of 11), M. sundaicus (maximum of 2200 μg Ni g-1 in leaves with a TF of 9.6) and F. 301 kinabaluensis (maximum of 8100 μg Ni g-1 in leaves with a TF of 9.3). These hyperaccumulator

302 plants, with the lowest BCF, appeared to hold lower Ni concentrations in their roots than other 303 studied hyperaccumulator plants, but have therefore evolved better mechanisms for transferring 304 Ni from the roots to aerial parts. This must allow plants to maintain their hyperaccumulation 305 characteristic and therefore their competition with other plants. 306 307 Regardless of the site prospected and the plant species collected, the most abundant bacterial 308 phyla were Proteobacteria (46%), Acidobacteria (21%) and Actinobacteria (6.3%). These 309 results concur not only with studies that use cultivation-based approaches (AbouShanab et al., 310 2010; A,lvarez-López et al., 2016) but also with studies based on molecular techniques using 311 highthroughput sequencing techniques. These studies underline that these three phyla were 312 predominant in different environments with contaminated and non-contaminated soils (Roesch 313 et al., 2007; Buée et al., 2009b; Chu et al., 2010; Zhao et al., 2014; Bordez et al., 2016; Saad et 314 al., 2017) but also in the rhizosphere of hyperaccumulators found in temperate climate such as 315 Albania or Greece (Lopez et al., 2017, 2019a). Many studies suggest that Proteobacteria and 316 Actinobacteria are the most common phyla in the rhizosphere of many plant species (Green and 317 Bohannan, 2006; Singh et al., 2007; Kaiser et al., 2016) and bacteria belonging to these two 318 phyla have been defined as copiotrophic (Fierer et al., 2007; Kopecky et al., 2011; Lienhard et 319 al., 2014). Moreover, it is currently accepted that copiotrophic bacteria are known to prefer 320 carbon-rich environments, such as rhizospheres (Semenov et al., 1999; Buée et al., 2009a; Yang 321 et al., 2017). Consequently, the positive correlations between Proteobacteria and both total C 322 and organic C contents underline the adaptability of this phylum to the rhizosphere, which is 323 known to content a wide variety of substances such as rhizodeposits. Acidobacteria phylum is 324 known to be one of the most abundant bacterial phyla in terrestrial ecosystems (Barns et al., 325 1999), playing a crucial role in the C cycle due to the ability of these bacteria to degrade 326 complex polysaccharides such as cellulose and lignin (Ward et al., 2009). These compounds 327 are commonly found in the rhizosphere environments. Indeed, the release of polysaccharides 328 as rhizodeposits has been frequently observed on the root surface of many plants (Oades, 1978) 329 and more particularly at the 330 331 root tip, where it can form a droplet in the presence of water (Samtsevich, 1965). Moreover, 332 senescence of epidermal and cortical root cells provides sources of carbon compounds for 333 bacteria (Werner, 2000). So, our results confirm previous studies focusing on the diversity of 334 bacterial communities from non-cultivated hyperaccumulator plants, which underlined high 335 relative abundances of Proteobacteria, Actinobacteria and Acidobacteria (Venter et al., 2018; 336 Lopez et al., 2019a, 2019b; Visioli et al., 2019). Following these bacterial phyla, the fourth 337 most abundant phylum was Rokubacteria (5.54%). This phylum occurs in many ecosystems, 338 including soils, plant rhizospheres, volcanic mud, frequently with a relative abundance less than 339 1% (Becraft et al., 2017) with the exception of a study conducted in Argentina focusing on the 340 soybean rhizosphere (Figuerola et al., 2015). In our study, this phylum, as well as 341 Gemmatimonadetes, was more abundant in the rhizosphere of R. aff. bengalensis (8.58% and 342 3.60% respectively for Rokubacteria and Gemmatimonadetes). However, this study showed 343 that the highest DTPA-extractable Ni concentrations were measured in the R. aff. bengalensis 344 rhizosphere (252 mg kg-1), compared to the other collected rhizosphere samples (193 mg kg- 345 1). In addition, the relative abundances of Rokubacteria and Gemmatimonadetes were

346 positively correlated with DTPA-extractable Ni. These observations were also found at family 347 level, with Saprospiraceae with a representativeness beyond 1% in R. aff. bengalensis 348 rhizosphere, and a positive correlation of this family with DTPA-extractable Ni. The presence 349 of this family was previously reported in Italian ultramafic soils (Visioli et al., 2019) which 350 confirms the adaptation of these bacteria to high Ni concentrations and therefore their high 351 representativeness in biome highly concentrated in Ni such as R. aff. bengalensis rhizosphere 352 soils. In the same way, all the other families linked positively with Ni-DTPA extractable 353 concentrations (Blastocatellaceae, Desulfarculaceae, Gemmatimonadaceae and 354 Pyrinomonadaceae) have been promoted in R. aff. bengalensis rhizosphere. Blastocatellaceae 355 had a relative abundance of 1.06% in R. aff. bengalensis rhizosphere soils against 0.93% for 356 the other plant species, Desulfarculaceae 1.79% vs. 1.67%, Gemmatimonadaceae 3.46% vs. 357 1.98% and Pyrinomonadaceae 3.91% vs. 2.09%. At OTU level, these observations were 358 confirmed by the NMDS analysis, which showed that bacterial communities present in all R. 359 aff. bengalensis rhizospheres were similar and linked with high Ni-DTPA extractable 360 concentrations, whatever the sampled site. As previously highlighted in a study on Ni 361 hyperaccumulator species from Halmahera Island (Indonesia; Lopez et al., 2019b), it appears 362 that R. aff. bengalensis, characterized by both high amounts of bioavailable Ni in its rhizosphere 363 and a particular rhizosphere bacterial community, may be able to create, through its 364 rhizodeposition, a particular biotic and abiotic environment near its roots. Rhizodeposits are 365 known to be essential parameters structuring the bacterial community in the rhizosphere 366 (Walker et al., 2003) and forming a basis for chemotaxis to mediate attraction and repulsion 367 among particular microbial species (Kumar et al., 2009). R. aff. bengalensis seems to drive its 368 own rhizosphere bacterial diversity, whatever the sites studied and their differences in terms of 369 physicochemical composition. The same conclusions may be reached for the other Rinorea 370 species: R. aff. javanica. All specimens from this hyperaccumulator had a close bacterial 371 community diversity as shown in the NMDS analysis. As observed in Halmahera (Lopez et al., 372 2019b), this feature can be specific to Rinorea genus as opposed to other regional nickel 373 hyperaccumulators. 374 375 In this study, except for these two species of Rinorea, for which Ni-DTPA extractable 376 concentrations were the most important driver of the bacterial diversity, for the other plant 377 species, all the rhizosphere abiotic parameters were the determinants of rhizosphere 378 microbiome composition. The results, particularly for the site 07, are examples that highlight 379 the importance of soil physicochemical parameters on the rhizosphere bacterial community 380 diversity. In the case of this site, the relative abundance of Firmicutes is significantly higher 381 (17.24%) compared to those observed in the other sites (1.84%). In contrast, the rhizosphere 382 soil samples from this site showed the lowest relative abundances of Acidobacteria (11.9% for 383 an average of about 21.7% in the other sites), Chloroflexi (2.47% vs. 3.32%) and 384 Planctomycetes (1.93% vs. 4.01%). However, the soil pH in this site had the higher values (7.4). 385 This can explain the low representativeness of Acidobacteria, whose relative abundance is 386 known to be higher in soils with more acidic pH (Naether et al., 2012). Conversely, Firmicutes 387 is the most abundant because this phylum is favoured by basic pH (Lauber et al., 2009). 388 Furthermore, it is generally in soil characterized by basic pH, where high Shannon index values 389 were found (Fierer and Jackson, 2006). However, in the Site 07, this index was significantly the

390 weakest with an average of 7.34 while the average is 8.90 for the other sites. So, we can 391 hypothesize that Firmicutes, favoured by particular soil physicochemical conditions, is 392 dominant because of a decrease in the bacterial diversity within the rhizosphere soil samples 393 collected in this site. Nevertheless, these results underlined the substantial role of soil 394 physicochemical parameters on the structure and the diversity of rhizosphere bacterial 395 communities, as many other authors have shown (Latour et al., 1996; Marschner et al., 2004; 396 Buée et al., 2009a). 397 398 CONCLUSION 399 A total of nine Ni hyperaccumulator plant species were collected in 10 sampling sites in 400 Malaysia. Of these, P. rufuschaneyi, R. aff. bengalensis and R. aff. javanica showed the highest 401 concentrations of Ni in their leaves. In addition, P. sarmentosa was found to accumulate Al to 402 the detriment of Ni when Al is present in high concentration in the rhizosphere soil. For all 403 rhizosphere soils collected, the most abundant phyla were Proteobacteria, Acidobacteria and 404 Actinobacteria. At the family level, Burkholderiaceae and Xanthobacteraceae from the 405 Proteobacteria phylum were the most abundant families in the different soils. Based on the 406 study area, the soil physicochemical properties of the prospected sites were the major factors 407 that impacted the microbial community diversity. Indeed, whatever the plant species found at 408 the same site, the rhizosphere bacterial diversity was similar. Conversely, Rinorea aff. 409 bengalensis had the ability to select a particular rhizosphere microbiome able to resist to high 410 Ni levels in its rhizosphere and characterized by high relative abundances of Rokubacteria and 411 Gemmatimonadetes. Regardless of the site considered, all rhizosphere samples from R. aff. 412 bengalensis presented a close bacterial diversity, a similar result which was already observed 413 in the neighbouring region of the island of Halmahera (Indonesia). 414 415 It might be interesting to confirm if it was the species R. aff. bengalensis, with particular 416 exudates, or the Ni concentration found in the rhizosphere soils, which were able to select a 417 specific microbiome. For this purpose, the rhizosphere bacterial diversity of R. aff. bengalensis 418 specimens needs to be compared, in controlled conditions, with the bacterial diversity of other 419 Ni hyperaccumulator plant species found in Borneo Island under different Ni amendment 420 conditions. Due to that, we may conclude if a high soil Ni concentration would be the major 421 factor structuring the bacterial communities of Rinorea, or if specific rhizodeposits produced 422 by this plant species could be the more important driver of bacterial communities in the 423 rhizosphere of this Ni hyperaccumulator plant. 424 425 EXPERIMENTAL PROCEDURES 426 427 Study area and sampling 428 The study area is located in the northeastern part of the Borneo Island, in the Malaysian state 429 of Sabah. During the fieldwork in July 2017, hyperaccumulator plants were sampled with their 430 associated rhizosphere soil. About 100 g of rhizosphere soil (i.e. soil adhering closely to the 431 roots of hyperaccumulator plants, at a 10–20 cm depth) were collected for each 432 hyperaccumulator plant. For detailed information and properties of ultramafic soils in the study 433 locations, please refer to previous studies from van der Ent et al. (2016a, 2016b, 2018).

434 435 Soil chemical analyses 436 At harvest, fresh rhizosphere soils collected from each site were sieved (2 mm) and the soil 437 moisture content was determined by heating subsamples to 105oC until constant weight. The 438 phytoavailable elements (i.e. chemically extractable) were extracted with a DTPATEA solution 439 (0.005 M diethylene triamine pentaacetic acid, 0.01 M calcium chloride dihydrate, 0.1 M 440 triethanolamine, pH 7.3) according to Lindsay and Norvell (1978), and the concentrations in 441 the extracted solutions were measured via ICP-AES (Inductively Coupled Plasma-Atomic 442 Emission Spectrometer; ICPAES, Liberty II, Varian). Soil samples (500 mg subsamples) were 443 acid-digested using freshly prepared Aqua Regia (6 ml 37% hydrochloric acid and 2 ml 70% 444 nitric acid per sample) for a 2-h period and diluted with distilled water to 50 ml before ICP- 445 AES analysis. Exchangeable elements associated with the soil cation exchange capacity (i.e. 446 CEC) were extracted in 0.0166 M Co(NH3)6Cl3 at a soil solution ratio of 1:20 (2.5 g:50 ml) 447 and 1 h shaking time according to international ISO standard 23470 (ISO 23470:2007). Soil pH 448 was measured using a pH meter in a soil-water suspension (soil to water ratio = 1:5). Total C 449 and N and organic C were quantified by combustion at 900 oC with a CHNS analyser (vario 450 MICRO cube, Elementar Analysensysteme GmbH). 451 452 Plant analyses 453 Plant parts (leaves, twigs and roots) were washed with distilled water and dry at 70oC for 48 h. 454 Subsamples (0.5 g) of dry and ground plant tissue were acid-digested at 95oC in 3 ml 455 concentrated nitric acid (65%) and 1 ml hydrogen peroxide (30%) for 2 h at 95oC. The digest 456 was diluted to 40 ml with distilled water before analysis with ICP-AES. Concentrations of the 457 following elements Ni, Co, Cr, Cu, Zn, Mn, Fe, Mg, Ca, Na, K, S and P were measured. Trace 458 metals translocation in these plants from shoot to root was measured using TF: TF = Cl/Cr, 459 where Cl and Cr are metal concentrations (μg g-1) in the leaves and roots respectively. 460 Furthermore, trace metals bioconcentration factor (BCF) in these plants was determined by 461 calculating the ratio of metal concentration in the shoot to that of the soil: BCF = Cl/Cs, where 462 Cl and Cs are total metal concentrations in leaves of the plant and in soil respectively. 463 464 Microbial diversity by high-throughput 16S rRNA amplicon sequencing 465 At harvest, 2 g of fresh rhizosphere soils were frozen and stored at -80oC as soon as possible 466 after fieldwork. Genomic DNA extractions from 0.5 g of fresh soil samples were performed 467 using the FastDNA™ SPIN kit for Soil (MP Biomedicals™, France) in accordance to 468 manufacturer’s protocol. Concentrations of DNA solutions were measured with a 469 spectrophotometer (SmartSpec Plus spectrophotometer, BIO-RAD) and sent to ADNid 470 (Montferrier-sur-Lez, France) for high-throughput 16S rRNA amplicon sequencing. Briefly, 471 barcoded amplicon sequencing was performed using the modified primers SD-Bact-0909-a-S- 472 18 (50-ACTCAAAKGAATWGACGG-30) and S-*-Univ-*-1392-a-A-15 (50- 473 ACGGGCGGTGTGTRC30) targeting a fragment of 484 bp of the 16S rRNA gene V6-V8 474 region (Klindworth et al., 2013). The primer modification consisted of the incorporation of the 475 Nextera XT® transposase sequence (Illumina Inc., San Diego, CA) in the 50 end of the forward 476 and reverse primers, and additional four random nucleotides in the forward primer to increase 477 the nucleotide diversity (Bartram et al., 2011). Amplicons were generated, purified and further

478 quantified. The concentration of the amplicons was re-adjusted to 1.66 ng μl-1 and 1 μl of each 479 library was used as a template in a second PCR where the Nextera XT® barcodes and the 480 Illumina adapters necessary for hybridization to the flow cell were added with the Nextera XT 481 Index kit. The resulting amplicons were purified and pooled in equimolar concentrations. 482 Libraries were mixed with Illumina-generated PhiX control libraries (5%) and a 2 × 300 bp 483 sequencing was performed with the MiSeq Reagent Kit V3-600 cycles. 484 485 Bioinformatics and statistical analysis 486 The obtained sequence reads were de-multiplexed, quality trimmed (sequence of low quality 487 (limit = 0.05) were removed, sequences with no ambiguous nucleotides were allowed and with 488 a minimum length of 400 nucleotides were kept) and for the ones with at least two reads were 489 assigned to OTUs at 97% similarity. Taxonomic affiliation was done with the SILVA database 490 with a confidence threshold of 80% (Silva.nr_v132, https://www.arbsilva.de/). This Targeted 491 Locus Study project has been deposited at DDBJ/ENA/GenBank under the accession 492 KBWT00000000. The version described in this article is the first version, KBWT01000000. In 493 order to be able to compare the data obtained from each rhizosphere sample, the high- 494 throughput sequencing results were normalized to the sample with the lowest total counts 495 (sample ‘Phyb-08b’ with 21,144 reads). Alpha (Shannon and Chao1 indexes) and Beta (NMDS 496 analysis with unweighted Unifrac method) diversities were studied using the QIIME software 497 (Quantitative Insights Into Microbial Ecology, version 1.8.0; Caporaso et al., 2010). Statistical 498 analyses were carried out on XLSTAT software (XLSTAT version Ecology 18.07) and R 499 software (version 3.5.3). To assess the significance of the differences between the different 500 samples, we have checked the normality of the data set with the Shapiro Wilk test and the 501 variance homogeneity with the Bartlett test. After validation of the conditions, a one-way 502 ANOVA was performed, followed by Tukey’s HSD test for paired comparisons in parametric 503 tests. 504 505 ACKNOWLEDGEMENTS 506 Sabah Biodiversity Centre provided S. Lopez with Access and Export Licences to permit this 507 research in Sabah (Malaysia). The French National Research Agency (ANR) through the 508 National ‘Investissements d’avenir’ program (ANR-10-LABX-21, LABEX RESSOURCES21) 509 is acknowledged for funding the PhD scholarship of S. Lopez. The ANR is also acknowledged 510 through the ANR14-CE04-0005 Project ‘Agromine’ for funding support. A. van der Ent was 511 the recipient of a Discovery Early Career Researcher Award (DE160100429) from the 512 Australian Research Council. 513 514

515 REFERENCES 516 Abou-Shanab, R.A.I., van Berkum, P., Angle, J.S., Delorme, T.A., Chaney, R.L., Ghozlan, 517 H.A., et al. (2010) Characterization of Ni-resistant bacteria in the rhizosphere of the 518 hyperaccumulator Alyssum murale by 16S rRNA gene sequence analysis. World J Microbiol 519 Biotechnol 26: 101–108. 520 Alexander, E.B. (2004) Varieties of ultramafic soil formation, plant cover and productivity. In 521 Ultramafic rocks: Their soils, vegetation and fauna. Proceedings of the 4th international 522 conference on serpentine ecology. St. Albans, Herts, UK: Science Reviews, pp. 9–17. 523 A,lvarez-López, V., Prieto-Fernández, A., Janssen, J., Herzig, R., Vangronsveld, J., and Kidd, 524 P.S. (2016) Inoculation methods using Rhodococcus erythropolis strain P30 affects bacterial 525 assisted phytoextraction capacity of Nicotiana tabacum. Int J 18: 406–415. 526 Baker, A.J.M. (1981) Accumulators and excluders -strategies in the response of plants to heavy 527 metals. J Plant Nutr 3: 643–654. 528 Baker, A.J.M., Proctor, J., van Balgooy, M.M.J., and Reeves, R.D. (1992) Hyperaccumulation 529 of nickel by the flora of the ultramafics of Palawan, Republic of the Philippines. In The 530 Vegetation of Ultramafic (Serpentine) Soils, Baker, A.J.M., Proctor, J., and Reeves, R.D. (eds). 531 Andover: Intercept, pp. 291–304. 532 Barns, S.M., Takala, S.L., and Kuske, C.R. (1999) Wide distribution and diversity of members 533 of the bacterial kingdom Acidobacterium in the environment. Appl Environ Microbiol 65: 534 1731–1737. 535 Bartram, A.K., Lynch, M.D.J., Stearns, J.C., MorenoHagelsieb, G., and Neufeld, J.D. (2011) 536 Generation of multimillion-sequence 16S rRNA gene libraries from complex microbial 537 communities by assembling paired-end Illumina reads. Appl Environ Microbiol 77: 3846– 538 3852. 539 Beaman, J.H. (2005) Mount Kinabalu: hotspot of plant diversity in Borneo. Biologiske Skrifter 540 55: 103–127. 541 Becraft, E.D., Woyke, T., Jarett, J., Ivanova, N., GodoyVitorino, F., Poulton, N., et al. (2017) 542 Rokubacteria: genomic giants among the uncultured bacterial phyla. Front Microbiol 8: 2264. 543 Bordez, L., Jourand, P., Ducousso, M., Carriconde, F., Cavaloc, Y., Santini, S., et al. (2016) 544 Distribution patterns of microbial communities in ultramafic landscape: a metagenetic approach 545 highlights the strong relationships between diversity and environmental traits. Mol Ecol 25: 546 2258–2272. 547 Bouman, R., van Welzen, P., Sumail, S., Echevarria, G., Erskine, P.D., and van der Ent, A. 548 (2018) Phyllanthus rufuschaneyi: A new nickel hyperaccumulator from Sabah (Borneo Island) 549 with potential for tropical agromining. Bot Stud 59: 1–12. https://doi.org/10.1186/s40529- 550 0180225-y. 551 Boyd, R.S., and Jaffré, T. (2009) Elemental concentrations of eleven new Caledonian plant 552 species from serpentine soils: elemental correlations and leaf-age effects. Northeast Nat 16: 93– 553 110. 554 Branco, S. (2010) Serpentine soils promote ectomycorrhizal fungal diversity. Mol Ecol 19: 555 5566–5576. 556 Brooks, R.R., and Wither, E.D. (1977) Nickel accumulation by Rinorea bengalensis (Wall.) 557 O.K. J Geochem Explor 7: 295–300.

558 Brooks, R.R., Lee, J., Reeves, R.D., and Jaffré, T. (1977) Detection of nickeliferous rocks by 559 analysis of herbarium specimen of indicator plants. J Geochem Explor 7: 49–57. Buée, M., De 560 Boer, W., Martin, F., van Overbeek, L., and Jurkevitch, E. (2009a) The rhizosphere zoo: an 561 overview of plant-associated communities of microorganisms, including phages, bacteria, 562 archaea, and fungi, and of some of their structuring factors. Plant and Soil 321:189–212. 563 Buée, M., Reich, M., Murat, C., Morin, E., Nilsson, R.H., Uroz, S., and Martin, F. (2009b) 454 564 pyrosequencing analyses of forest soils reveal an unexpectedly high fungal diversity. New 565 Phytol 184: 449–456. 566 Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., 567 Fierer, N., et al. (2010) QIIME allows analysis of high-throughput community sequencing data. 568 Nat Methods 7: 335–336. https://doi.org/ 10.1038/nmeth.f.303. 569 Chu, H., Fierer, N., Lauber, C.L., Caporaso, J.G., Knight, R., and Grogan, P. (2010) Soil 570 bacterial diversity in the Arctic is not fundamentally different from that found in other biomes: 571 bacterial biogeography in arctic soils. Environ Microbiol 12: 2998–3006. 572 Echevarria, G. (2018) Genesis and behaviour of ultramafic soils and consequences for nickel 573 biogeochemistry. In Agromining: Farming for Metals: Extracting Unconventional Resources 574 Using Plants, Mineral Resource Reviews, van der Ent, A., Echevarria, G., Baker, A.J.M., and 575 Morel, J.L. (eds). Cham: Springer International Publishing, pp. 135–156. 576 Epelde, L., Martín-Sánchez, I., González-Oreja, J.A., Anza, M., Gómez-Sagasti, M.T., and 577 Garbisu, C. (2012) Impact of sources of environmental degradation on microbial community 578 dynamics in non-polluted and metalpolluted soils. Sci Total Environ 433: 264–272. 579 Fierer, N., and Jackson, R.B. (2006) The diversity and biogeography of soil bacterial 580 communities. Proc Natl Acad Sci USA 103: 626–631. 581 Fierer, N., Bradford, M.A., and Jackson, R.B. (2007) Toward an ecological classification of soil 582 bacteria. Ecology 88: 1354–1364. 583 Figuerola, E.L.M., Guerrero, L.D., Türkowsky, D., Wall, L.G., and Erijman, L. (2015) Crop 584 monoculture rather than agriculture reduces the spatial turnover of soil bacterial communities 585 at a regional scale: reduced turnover of soil bacterial communities. Environ Microbiol 17: 678– 586 688. 587 Freitas, H., Prasad, M.N.V., and Pratas, J. (2004) Analysis of serpentinophytes from north–east 588 of Portugal for trace metal accumulation–relevance to the management of mine environment. 589 Chemosphere 54: 1625–1642. 590 Galey, M.L., van der Ent, A., Iqbal, M.C.M., and Rajakaruna, N. (2017) Ultramafic geoecology 591 of south and Southeast Asia. Bot Stud 58: 18. 592 Ghaderian, S.M., Mohtadi, A., Rahiminejad, M.R., and Baker, A.J.M. (2007) Nickel and other 593 metal uptake and accumulation by species of Alyssum (Brassicaceae) from the ultramafics of 594 Iran. Environ Pollut 145: 293–298. 595 Green, J., and Bohannan, B.J.M. (2006) Spatial scaling of microbial biodiversity. Trends Ecol 596 Evol 21: 501–507. 597 Guillot, S., and Hattori, K. (2013) Serpentinites: essential roles in geodynamics, arc volcanism, 598 sustainable development, and the origin of life. Elements 9: 95–98. 599 Harrison, S., and Rajakaruna, N. (2011) Serpentine: The Evolution and Ecology of a Model 600 System. Berkeley, CA: University of California Press.

601 Hoffmann, P., Baker, A.J.M., Madulid, D.A., and Proctor, J. (2003) Phyllanthus Balgooyi 602 (Euphorbiaceae S.L.), a new nickel-hyperaccumulating species from Palawan and Sabah. 603 Blumea Biodivers Evol Biogeogr Plants 48: 193–199. 604 Janssen, P.H. (2006) Identifying the dominant soil bacterial taxa in libraries of 16S rRNA and 605 16S rRNA genes. Appl Environ Microbiol 72: 1719–1728. 606 Kaiser, K., Wemheuer, B., Korolkow, V., Wemheuer, F., Nacke, H., Schöning, I., et al. (2016) 607 Driving forces of soil bacterial community structure, diversity, and function in temperate 608 grasslands and forests. Sci Rep 6: 33696. 609 Kandeler, E., Tscherko, D., Bruce, K.D., Stemmer, M., Hobbs, P.J., Bardgett, R.D., and 610 Amelung, W. (2000) Structure and function of the soil microbial community in microhabitats 611 of a heavy metal polluted soil. Biol Fertil Soils 32: 390–400. 612 Kay, K.M., Ward, K.L., Watt, L.R., and Schemske, D.W. (2011) Plant speciation. In 613 Serpentine: The Evolution and Ecology of a Model System, Harrison, S., and Rajakaruna, N. 614 (eds). Berkeley, CA: University of California Press, pp. 71–96. 615 Kazakou, E., Dimitrakopoulos, P.G., Baker, A.J.M., Reeves, R.D., and Troumbis, A.Y. (2008) 616 Hypotheses, mechanisms and trade-offs of tolerance and adaptation to serpentine soils: from 617 species to ecosystem level. Biol Rev 83: 495–508. 618 Kazakou, E., Adamidis, G.C., Baker, A.J.M., Reeves, R.D., Godino, M., and Dimitrakopoulos, 619 P.G. (2010) Species adaptation in serpentine soils in Lesbos Island (Greece): metal 620 hyperaccumulation and tolerance. Plant and Soil 332: 369–385. 621 Kidd, P., Barceló, J., Bernal, M.P., Navari-Izzo, F., Poschenrieder, C., Shilev, S., et al. (2009) 622 Trace element behaviour at the root–soil interface: implications in phytoremediation. Environ 623 Exp Bot 67: 243–259. 624 Kidd, P.S., Bani, A., Benizri, E., Gonnelli, C., Hazotte, C., Kisser, J., et al. (2018) Developing 625 sustainable agromining systems in agricultural ultramafic soils for nickel recovery. Front 626 Environ Sci 6: 44. 627 Klindworth, A., Pruesse, E., Schweer, T., Peplies, J., Quast, C., Horn, M., and Glockner, F.O. 628 (2013) Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next- 629 generation sequencing-based diversity studies. Nucleic Acids Res 41: e1. 630 Kopecky, J., Kyselkova, M., Omelka, M., Cermak, L., Novotna, J., Grundmann, G.L., et al. 631 (2011) Actinobacterial community dominated by a distinct clade in acidic soil of a waterlogged 632 deciduous forest. FEMS Microbiol Ecol 78: 386–394. 633 Kumar, K.V., Srivastava, S., Singh, N., and Behl, H.M. (2009) Role of metal resistant plant 634 growth promoting bacteria in ameliorating fly ash to the growth of Brassica juncea. J Hazard 635 Mater 170: 51–57. 636 Kuperman, R.G., and Carreiro, M.M. (1997) Soil heavy metal concentrations, microbial 637 biomass and enzyme activities in a contaminated grassland ecosystem. Soil Biol Biochem 29: 638 179–190. 639 Latour, X., Corberand, T., Laguerre, G., Allard, F., and Lemanceau, P. (1996) The composition 640 of fluorescent pseudomonad populations associated with roots is influenced by plant and soil 641 type. Appl Environ Microbiol 62: 2449–2456. 642 Lauber, C.L., Hamady, M., Knight, R., and Fierer, N. (2009) Pyrosequencing-based assessment 643 of soil pH as a predictor of soil bacterial community structure at the continental scale. Appl 644 Environ Microbiol 75: 5111–5120.

645 Lebeau, T., Braud, A., and Jézéquel, K. (2008) Performance of bioaugmentation-assisted 646 phytoextraction applied to metal contaminated soils: a review. Environ Pollut 153: 497–522. 647 Lee-Cruz, L., Edwards, D.P., Tripathi, B.M., and Adams, J. 648 M. (2013) Impact of logging and forest conversion to oil palm plantations on soil bacterial 649 communities in Borneo. Appl Environ Microbiol 79: 7290–7297. 650 Lienhard, P., Terrat, S., Prévost-Bouré, N.C., Nowak, V., Régnier, T., Sayphoummie, S., et al. 651 (2014) Pyrosequencing evidences the impact of cropping on soil bacterial and fungal diversity 652 in Laos tropical grassland. Agron Sustain Dev 34: 525–533. 653 Lindsay, W.L., and Norvell, W.A. (1978) Development of a DTPA soil test for , iron, 654 manganese, and copper. Soil Sci Soc Am J 42: 421–428. 655 Lopez, S., Piutti, S., Vallance, J., Morel, J.-L., Echevarria, G., and Benizri, E. (2017) Nickel 656 drives bacterial community diversity in the rhizosphere of the hyperaccumulator Alyssum 657 murale. Soil Biol Biochem 114: 121–130. 658 Lopez, S., Goux, X., Echevarria, G., Calusinska, M., Morel, J.L., and Benizri, E. (2019a) 659 Community diversity and potential functions of rhizosphere-associated bacteria of nickel 660 hyperaccumulators found in Albania. Sci Total Environ 654: 237–249. 661 Lopez, S., Goux, X., van der Ent, A., Erskine, P.D., Echevarria, G., Calusinska, M., et al. 662 (2019b) Bacterial community diversity in the rhizosphere of nickel hyperaccumulator species 663 of Halmahera Island (Indonesia). Appl Soil Ecol 133: 70–80. 664 Marschner, P., Crowley, D., and Yang, C.H. (2004) Development of specific rhizosphere 665 bacterial communities in relation to plant species, nutrition and soil type. Plant and Soil 261: 666 199–208. 667 Mesjasz-Przybylowicz, J., Przybylowicz, W., Barnabas, A., and van der Ent, A. (2016) Extreme 668 nickel hyperaccumulation in the vascular tracts of the tree Phyllanthus balgooyi from Borneo. 669 New Phytol 209: 1513–1526. 670 Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., and Kent, J. (2000) 671 Biodiversity hotspots for conservation priorities. Nature 403: 853–858. 672 Naether, A., Foesel, B.U., Naegele, V., Wüst, P.K., Weinert, J., Bonkowski, M., et al. (2012) 673 Environmental factors affect acidobacterial communities below the subgroup level in grassland 674 and forest soils. Appl Environ Microbiol 78: 7398–7406. 675 Nahon, D. (2003) Altérations dans la zone tropicale. Signification à travers les mécanismes 676 anciens et/ou encore actuels. C R Geosci 335: 1109–1119. 677 Nkrumah, P.N., Chaney, R.L., and Morel, J.L. (2018a) Agronomy of ‘metal crops’ used in 678 Agromining. In Agromining: Farming for Metals, van der Ent, A., Echevarria, G., Baker, 679 A.J.M., and Morel, J.L. (eds). Cham: Springer International Publishing, pp. 19–38. 680 Nkrumah, P.N., Echevarria, G., Erskine, P.D., and van der Ent, A. (2018b) Contrasting nickel 681 and zinc hyperaccumulation in subspecies of Dichapetalum gelonioides from Southeast Asia. 682 Sci Rep 8: 1–15. 683 Nkrumah, P.N., Echevarria, G., Erskine, P.D., and van der Ent, A. (2018c) Nickel 684 hyperaccumulation in Antidesma montis-silam: from herbarium discovery to collection in the 685 native habitat. Ecol Res 33: 675–685. 686 Nkrumah, P.N., Echevarria, G., Erskine, P.D., Chaney, R.L., Sumail, S., and van der Ent, A. 687 (2019) Growth effects in tropical nickel-agromining ‘metal crops’ in response to nutrient 688 dosing. J Plant Nutr Soil Sci 182: 715–728.

689 Oades, J.M. (1978) Mucilages at the root surface. J Soil Sci 690 29: 1–16. 691 Pollard, A.J., Reeves, R.D., and Baker, A.J.M. (2014) Facultative hyperaccumulation of heavy 692 metals and metalloids. Plant Sci 217–218: 8–17. 693 Proctor, J. (2003) Vegetation and soil and plant chemistry on ultramafic rocks in the tropical 694 Far East. Perspect Plant Ecol Evol Syst 6: 105–124. 695 Proctor, J., and Woodell, S.R.J. (1975) The ecology of serpentine soils. In Advances in 696 Ecological Research, MacFadyen, A. (ed). London: Academic Press, pp. 255–366. 697 Proctor, J., Lee, Y.F., Langley, A.M., Munro, W.R.C., and Nelson, T. (1988) Ecological studies 698 on Gunung Silam, a small ultrabasic mountain in Sabah, Malaysia. 699 I. Environment, forest structure and floristics. J Ecol 76: 320–340. 700 Proctor, J., Phillipps, C., Duff, G.K., Heaney, A., and Robertson, F.M. (1989) Ecological 701 studies on Gunung Silam, a small ultrabasic mountain in Sabah, Malaysia II some forest 702 processes. J Ecol 77: 317–331. 703 Rajkumar, M., Vara Prasad, M.N., Freitas, H., and Ae, N. (2009) Biotechnological applications 704 of serpentine soil bacteria for phytoremediation of trace metals. Crit Rev Biotechnol 29: 120– 705 130. 706 Rastogi, G., Osman, S., Kukkadapu, R., Engelhard, M., Vaishampayan, P.A., Andersen, G.L., 707 and Sani, R.K. (2010) Microbial and mineralogical characterizations of soils collected from the 708 deep biosphere of the former Homestake gold mine, South Dakota. Microb Ecol 60: 539–550. 709 Reeves, R.D. (1999) Nickel hyperaccumulation in the serpentine flora of Cuba. Ann Bot 83: 710 29–38. 711 Reeves, R.D. (2003) Tropical hyperaccumulators of metals and their potential for 712 phytoextraction. Plant and Soil 249: 57–65. 713 Reeves, R.D. (2006) Hyperaccumulation of trace elements by plants. In Phytoremediation of 714 Metal-Contaminated Soils, Morel, J.L., Echevarria, G., and Goncharova, N. (eds). Dordrecht: 715 Kluwer Academic Publishers, pp. 25–52. 716 Reeves, R.D., Baker, A.J.M., Jaffré, T., Erskine, P.D., Echevarria, G., and van der Ent, A. 717 (2017) A global database for plants that hyperaccumulate metal and metalloid trace elements. 718 New Phytol 218: 407–411. 719 Roesch, L.F., Fulthorpe, R.R., Riva, A., Casella, G., Hadwin, A.K., Kent, A.D., et al. (2007) 720 Pyrosequencing enumerates and contrasts soil microbial diversity. ISME J 1: 283–290. 721 Saad, R.F., Kobaissi, A., Machinet, G., Villemin, G., Echevarria, G., and Benizri, E. (2017) 722 Crop rotation associating a legume and the nickel hyperaccumulator Alyssum murale improves 723 the structure and biofunctioning of an ultramafic soil. Ecol Res 33: 799–810. 724 Samtsevich, S.A. (1965) Active excretions of plant roots and their significance. Soviet Plant 725 Physiol 12: 731–740. 726 Schipper, L.A., and Lee, W.G. (2004) Microbial biomass, respiration and diversity in ultramafic 727 soils of west dome, New Zealand. Plant and Soil 262: 151–158. 728 Semenov, A.M., van Bruggen, A.H.C., and Zelenev, V.V. (1999) Moving waves of bacterial 729 populations and total organic carbon along roots of wheat. Microb Ecol 37: 116–128. 730 Sessitsch, A., Kuffner, M., Kidd, P., Vangronsveld, J., Wenzel, W.W., Fallmann, K., and 731 Puschenreiter, M. (2013) The role of plant-associated bacteria in the mobilization and 732 phytoextraction of trace elements in contaminated soils. Soil Biol Biochem 60: 182–194.

733 Shallari, S., Schwartz, C., Hasko, A., and Morel, J.L. (1998) Heavy metals in soils and plants 734 of serpentine and industrial sites of Albania. Sci Total Environ 209: 133–142. 735 Singh, B.K., Munro, S., Potts, J.M., and Millard, P. (2007) Influence of grass species and soil 736 type on rhizosphere microbial community structure in grassland soils. Appl Soil Ecol 36: 147– 737 155. 738 Tripathi, B.M., Kim, M., Singh, D., Lee-Cruz, L., Lai-Hoe, A., Ainuddin, A.N., et al. (2012) 739 Tropical soil bacterial communities in Malaysia: pH dominates in the equatorial tropics too. 740 Microb Ecol 64: 474–484. 741 van der Ent, A. (2015) Heavy metal farming. Aust Sci 36: 26–27. 742 van der Ent, A., Baker, A.J.M., van Balgooy, M.M.J., and Tjoa, A. (2013) Ultramafic nickel 743 laterites in Indonesia (Sulawesi, Halmahera): mining, nickel hyperaccumulators and 744 opportunities for phytomining. J Geochem Explor 128: 72–79. van der Ent, A., Erskine, P., and 745 Sumail, S. (2015a) Ecology of nickel hyperaccumulator plants from ultramafic soils in 746 Sabah (Malaysia). Chem 25: 243–259. 747 van der Ent, A., van Balgooy, M., and van Welzen, P. (2015b) Actephila alanbakeri 748 (Phyllanthaceae): a new nickel hyperaccumulating plant species from localised ultramafic 749 outcrops in Sabah (Malaysia). Bot Stud 57: 1–8. 750 van der Ent, A., Echevarria, G., and Tibbett, M. (2016a) Delimiting soil chemistry thresholds 751 for nickel hyperaccumulator plants in Sabah (Malaysia). Chem 26: 67–82. 752 van der Ent, A., Erskine, P., Mulligan, D., Repin, R., and Karim, R. (2016b) Vegetation on 753 ultramafic edaphic ‘islands’ in Kinabalu Park (Sabah, Malaysia) in relation to soil chemistry 754 and elevation. Plant and Soil 403: 77–101. 755 van der Ent, A., Callahan, D.L., Noller, B.N., MesjaszPrzybylowicz, J., Przybylowicz, W.J., 756 Barnabas, A., and Hugh, H.H. (2017) Nickel biopathways in tropical nickel hyperaccumulating 757 trees from Sabah (Malaysia). Sci Rep 7: 41861. 758 van der Ent, A., Cardace, D., Tibbett, M., and Echevarria, G. (2018) Ecological implications of 759 pedogenesis and geochemistry of ultramafic soils in Kinabalu Park (Malaysia). Catena 160: 760 154–169. 761 van der Ent, A., de Jonge, M.D., Mak, R., MesjaszPrzybylowicz, J., Przybylowicz, W.J., 762 Barnabas, A.D., and Hugh, H.H. (2020) X-ray fluorescence elemental mapping of roots, stems 763 and leaves of the nickel hyperaccumulators Rinorea cf. bengalensis and Rinorea cf. javanica 764 (Violaceae) from Sabah (Malaysia), Borneo. Plant Soil. in Press. 765 https://doi.org/10.1007/s11104-019-04386-2. 766 Venter, A., Siebert, S., Rajakaruna, N., Barnard, S., Levanets, A., Ismail, A., et al. (2018) 767 Biological crusts of serpentine and non-serpentine soils from the Barberton Greenstone Belt of 768 South Africa. Ecol Res 33: 629–640. 769 Visioli, G., Sanangelantoni, A.M., Conti, F.D., Bonati, B., Gardi, C., and Menta, C. (2019) 770 Above and belowground biodiversity in adjacent and distinct serpentine soils. Appl Soil Ecol 771 133: 98–103. 772 Walker, T.S., Bais, H.P., Grotewold, E., and Vivanco, J.M. (2003) Root exudation and 773 rhizosphere biology. Plant Physiol 132: 44–51. 774 Wang, Y., Li, Q., Shi, J., Lin, Q., Chen, X., Wu, W., and Chen, Y. (2008) Assessment of 775 microbial activity and bacterial community composition in the rhizosphere of a copper 776 accumulator and a non-accumulator. Soil Biol Biochem 40: 1167–1177.

777 Ward, N.L., Challacombe, J.F., Janssen, P.H., Henrissat, B., Coutinho, P.M., Wu, M., et al. 778 (2009) Three genomes from the phylum Acidobacteria provide insight into the lifestyles of 779 these microorganisms in soils. Appl Environ Microbiol 780 75: 2046–2056. 781 Werner, D. (2000) Organic signals between plants and microorganisms. In The rhizosphere. 782 Boca Raton, FL: CRC Press, pp. 213–238. 783 Yang, Y., Wang, N., Guo, X., Zhang, Y., and Ye, B. (2017) Comparative analysis of bacterial 784 community structure in the rhizosphere of maize by high-throughput pyrosequencing. PLoS 785 One 12: e0178425. 786 Zhao, J., Zhang, R., Xue, C., Xun, W., Sun, L., Xu, Y., and Shen, Q. (2014) Pyrosequencing 787 reveals contrasting soil bacterial diversity and community structure of two main winter wheat 788 cropping systems in China. Microb Ecol 67: 443–453. 789 790

791

792

site 04 Borneo Island site 06 site 05 site 03

site 08 site 09 site 02 Site 10

site 07

site 01 793

794 Fig. 1 Map showing the localities from which metal accumulator plants and rhizosphere soil

795 samples were collected

100% c abc ab ab abc bc a bc ab bc d bcd cd cd bcd ab a abc abc 90% a

80%

bc c abc 70% abc ab ab a abc bc abc

60%

50% ab a a bc ab b c c ab a bcde a a a bc a ab b ab bc ab ab a bc 40% bc abcd abc cde a abc de c Relative abundance ab c a bc b a bcb a ab cde a a a bcde c ab e a ab ab a a ab bc bc bc a 30% a a ab a a ab a ab a a a a ab a a 20% a a a a a ab a a a 10% ab ab ab ab b ab

0% Site 01 Site 02 Site 03 Site 04 Site 05 Site 06 Site 07 Site 08 Site 09 Site 10 Acidobacteria Actinobacteria Bacteroidetes Chloroflexi Firmicutes Gemmatimonadetes Latescibacteria Nitrospirae Planctomycetes Proteobacteria Rokubacteria Others (<1%) 796

797 Fig. 2 Relative abundance of bacterial phyla identified in the rhizosphere soil samples (%) for

798 each site. “Others (<1%)” refers to Armatimonadetes, BRC1, Cyanobacteria, Dependentiae,

799 Elusimicrobia, Entotheonellaeota, GAL15, Patescibacteria, Spirochaetes, Verrucomicrobia,

800 WPS-2 and Zixibacteria. Bars marked with the same letter for each phylum do not differ

801 significantly according to TukeyHSD test at p < 0.05

Rin Fla Rinj

Psy

Rinb

Phyb

Mis Phyr Wal

Legend: Acidobacteria Chloroflexi Nitrospiraea Actinobacteria Firmicutes Planctomycetes Bacteroidetes Gemmatimonadetes Proteobacteria

Relative abundances: 1% 6% 802

803 Fig. 3 Bacterial network analysis of the most abundant families (>1%) for the Ni

804 hyperaccumulator rhizosphere soils. Octagons represent plant species and circles represent the

805 different families. The size of a circle indicates the mean of the family relative abundance (1 to

806 6%). The line color indicates a representativeness of a family greater than 1% in a plant

807 rhizosphere soil. Fla: F. kinabaluensis, Mis: M. sundaicus, Phyb: P. balgooyi, Phyr: P.

808 rufuschaneyi, Psy: P. sarmentosa, Rinb: R. aff. bengalensis, Rinj: R. aff. javanica, Rin: Rinorea

809 sp., Wal: W. pinnata

810

811 Fig. 4 Redundancy Analysis (RDA) performed between the soil physicochemical

812 characteristics and the relative abundancy of the 10 major bacterial phyla. Fla: F. kinabaluensis,

813 Mis: M. sundaicus, Phyb: P. balgooyi, Phyr: P. rufuschaneyi, Psy: P. sarmentosa, Rinb: R. aff.

814 bengalensis, Rinj: R. aff. javanica, Rin: Rinorea sp., Wal: W. pinnata. Number 01 to 10: the

815 site number. a, b and c: the replicate at each site. Abbreviations: % of soil organic carbon

816 (%Corg), % of soil total nitrogen (%N), % of total carbon (%C), soil pH (pH), soil Cation

817 Exchange Capacity (xx-CEC), extractable element (xx-DTPA)

818

819 DTPA CEC CEC DTPA DTPA DTPA - DTPA DTPA CEC - - CEC DTPA DTPA ------CEC - - - N C Corg - pH % % % Co Cr Cu Fe Mn Ni Pb Zn CEC Ca K Mg Mn Ni Elevation Acidobacteria Actinobacteria Bacteroidetes Chloroflexi Firmicutes Gemmatimonadetes Latescibacteria Planctomycetes Proteobacteria Rokubacteria

-1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1 820

821

822

823 Fig. 5 Correlation matrix between soil physicochemical parameters and the relative abundance

824 of the ten major bacterial phyla. Based on the Spearman correlation coefficient r, red colors

825 correspond to a negative correlation and blue colors to a positive correlation, the color gradient

826 was linked with the r coefficient value, the blanks correspond to a non-significant coefficient

827 (p > 0.1). Abbreviations: % N: % total nitrogen, % C: % total carbon, % Corg: % organic C,

828 xx-DTPA: DTPA-extractable elements, XX-CEC: exchangeable elements

829

830

831 DTPA CEC CEC DTPA DTPA DTPA - DTPA DTPA CEC - - CEC DTPA DTPA ------CEC - - - N C Corg - pH % % % Co Cr Cu Fe Mn Ni Pb Zn CEC Ca K Mg Mn Ni Elevation Blastocatellaceae Burkholderiaceae Desulfarculaceae Gemmatimonadaceae families Haliangiaceae Micromonosporaceae Nitrosomonadaceae Pyrinomonadaceae Common Xanthomonadaceae Fla Clostridiales vadinBB60 Mis A4b Pseudonocardiaceae A21b Chitinimonadaceae Phyb Hyphomonadaceae Koribacteraceae Ktedonobacteraceae Parvularculaceae Rin TRA3-20 Rinb Saprospiraceae

-1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1 832

833

834 Fig. 6 Correlation matrix between soil physicochemical parameters and the relative abundance

835 of the major bacterial families common from all plant species or found in only one plant species

836 rhizosphere. Based on the Spearman correlation coefficient r, red colors correspond to a

837 negative correlation and blue colors to a positive correlation, the color gradient was linked with

838 the r coefficient value, the blanks correspond to a non-significant coefficient (p > 0.1).

839 Abbreviations: Fla: F. kinabaluensis, Mis: M. sundaicus, Phyb: P. balgooyi; Rin: Rinorea sp.,

840 Rinb: R. aff. bengalensis, % N: % total nitrogen, % C: % total carbon, % Corg: % organic C,

841 xx-DTPA: DTPA-extractable elements, XX-CEC: exchangeable elements

Phyr-10a Rinb-02b

Psy-09 Phyr-02c Phyr-10c Phyr-10b Phyr-02a Phyb-10 Rin-10 Rinb-02c

0.2 Wal-10 Phyr-02b Phyb-09 Mis-09 Rin-09 Wal-06 Rinb-06b Fla-10 Fla-09 Rinj-03Rinb-03a Rinb-03c Phyb-08a Psy-06 Rinb-03bRinb-02a Psy-10 Rinb-06c Rinb-06a Rin-08 Mis-06 220 Mis-03 Mis-08 Rinj-06 Rinj-04 Mis-07 Fla-06 Phyb-08b Fla-04 Phyb-01a Phyb-01b Rinj-07 NMDS2 0.2 Phyb-05c

- Psy-08 Phyb-05a Phyb-01c Wal-07 Phyb-05b Fla-08 stress: 0.102 Psy-05 Fla-07 r2: 0.363

0.6 Mis-04 - p < 0.001

-1.5 -1.0 -0.5 0.0 0.5 NMDS1 842

843 Fig. 7 Non-metric multidimensional scaling (NMDS) related with Ni DTPA-extractable

844 concentration (green lines). Fla: F. kinabaluensis, Mis: M. sundaicus, Phyb: P. balgooyi, Phyr:

845 P. rufuschaneyi, Psy: P. sarmentosa, Rinb: R. aff. bengalensis, Rinj: R. aff. javanica, Rin:

846 Rinorea sp., Wal: W. pinnata. Number 01 to 10: the site number. a, b and c: the replicate at each

847 site

848 849

850 Table 1 Sampled sites, species and codes of the hyperaccumulator plants. The letters a to c

851 correspond to the different plant replicates collected at a same site

Site Site name Elevation Hyperaccumulator plants and codes number (m) 01 Bukit Kulung 1410 P. balgooyi (Phyb-01a, Phyb-01b, Phyb-01c) 02 Pahu 340 P. rufuschaneyi (Phyr-02a, Phyr-02b, Phyr-02c), R. aff. bengalensis (Rinb-02a, Rinb-02b, Rinb-02c) 03 Monggis 420 M. sundaicus (Mis-03), R. aff. javanica (Rinj-03), R. aff. bengalensis (Rinb-03a, Rinb-03b, Rinb-03c) 04 Wuluh River 780 F. kinabaluensis (Fla-04), M. sundaicus (Mis-04), R. aff. javanica (Rinj-04) 05 Wuluh River Km 905 P. balgooyi (Phyb-05a, Phyb-05b, Phyb-05c), 7 P. sarmentosa (Psy-05) 06 Wuluh River Km 680 F. kinabaluensis (Fla-06), M. sundaicus (Mis-06), 3 P. sarmentosa (Psy-06), R. aff. javanica (Rinj-06), R. aff. bengalensis (Rinb-06a, Rinb-06b, Rinb-06c), W. pinnata (Wal-06) 07 Lobou 385 F. kinabaluensis (Fla-07), M. sundaicus (Mis-07), R. aff. javanica (Rinj-07), W. pinnata (Wal-07) 08 Panataran River 470 P. balgooyi (Phyb-08a, Phyb-08b), Rinorea sp. (Rin-08), M. sundaicus (Mis-08), P. sarmentosa (Psy-08), F. kinabaluensis (Fla-08) 09 Bukit Mongilan 750 F. kinabaluensis (Fla-09), M. sundaicus (Mis-09), P. balgooyi (Phyb-09), P. sarmentosa (Psy-09), Rinorea sp. (Rin-09) 10 Bukit Lompouyu 700 F. kinabaluensis (Fla-10), P. balgooyi (Phyb-10), P. rufuschaneyi (Phyr-10a, Phyr-10b, Phyr-10c), P. sarmentosa (Psy-10), Rinorea sp. (Rin-10), W. pinnata (Wal-10) 852

853 Table 2 pH, total N, C/N ratio and organic C of rhizosphere soils. Mean and range are provided,

854 n is the number of soil samples per site. Data followed by different letters are significantly

855 different at p < 0.05 (TukeyHSD test)

Site n pH N (%) C/N C organic (%) 01 3 6.0 [5.8 – 6.5] de 0.37 [0.19 – 0.53] 21 [15 – 28] a 6.3 [2.3 – 11] ab 02 6 6.8 [6.5 – 7.1] abc 0.50 [0.43 – 0.67] 13 [12 – 16] b 6.4 [5.1 – 11] ab 03 5 7.0 [6.8 – 7.3] ab 0.49 [0.42 – 0.66] 12 [11 – 13] b 5.7 [5.0 – 7.6] b 04 3 6.9 [6.4 – 7.2] abc 0.53 [0.30 – 0.78] 16 [14 – 17] ab 8.1 [5.0 – 12] ab 05 4 6.0 [5.7 – 6.3] e 0.90 [0.63 – 1.11] 20 [17 – 22] a 18 [11 – 23] a 06 8 6.4 [6.0 – 6.9] cde 0.60 [0.29 – 1.03] 13 [10 – 15] b 7.8 [2.9 – 16] ab 07 4 7.4 [6.7 – 7.7] a 0.49 [0.23 – 0.73] 14 [11 – 15] b 6.7 [2.5 – 10] ab 08 6 6.4 [6.2 – 6.8] bcde 0.81 [0.24 – 1.92] 15 [12 – 21] ab 13 [2.9 – 34] ab 09 5 6.1 [5.7 – 6.4] de 0.53 [0.42 – 0.66] 14 [12 – 17] b 7.3 [4.9 – 10] ab 10 8 6.7 [6.3 – 6.9] bcd 0.37 [0.24 – 0.54] 16 [13 – 24] ab 5.8 [3.9 – 12] b 856

857 Table 3 DTPA-extractable elements of rhizosphere soils. Mean and range are provided, n is the number of samples. Data followed by different

858 letters are significantly different at p < 0.05 (TukeyHSD test)

Site n Mn (mg kg−1) Fe (mg kg−1) Co (mg kg−1) Ni (mg kg−1) Cu (mg kg−1) Zn (mg kg−1) Pb (mg kg−1) 01 3 68 [61 – 76] bcd 152 [43 – 247] b 8.7 [4.2 – 11] b 110 [78 – 145] 14 [8.2 – 22] a 6.1 [4.0 – 9.4] 5.8 [3.0 – 18] a 02 6 140 [98 – 172] abc 67 [52 – 83] b 4.5 [3.1 – 6.3] b 295 [243 – 436] 1.6 [1.3 – 1.8] b 3.3 [2.5 – 4.0] 0.14 [0.07 – 0.27] b 03 5 68 [32 – 108] cd 96 [42 – 151] b 2.9 [1.0 – 5.1] b 207 [165 – 291] 0.6 [0.4 – 0.8] b 2.9 [1.9 – 4.7] 0.27 [0.13 – 0.46] b 04 3 40 [12 – 72] d 92 [48 – 161] b 2.6 [0.6 – 5.7] b 232 [149 – 318] 0.7 [0.4 – 1.3] b 2.3 [1.3 – 3.3] 0.18 [0.13 – 0.23] b 05 4 97 [23 – 173] bcd 571 [282 – 900] a 12 [7.7 – 20] b 211 [126 – 334] 1.5 [1.1 – 1.9] b 4.0 [2.3 – 6.5] 0.20 [0.08 – 0.34] b 06 8 101 [45 – 145] bcd 122 [52 – 211] b 5.0 [3.0 – 8.1] b 228 [157 – 389] 1.1 [0.4 – 1.9] b 3.8 [1.3 – 10] 0.33 [0.12 – 0.50] b 07 4 37 [11 – 88] d 67 [51 – 102] b 1.9 [0.3 – 5.9] b 148 [63 – 354] 1.0 [0.7 – 1.6] b 2.1 [0.76 – 3.9] 0.18 [0.09 – 0.23] b 08 6 45 [30 – 55] d 261 [68 – 757] b 2.7 [1.4 – 3.6] b 141 [67 – 252] 2.0 [1.2 – 2.6] b 7.0 [0.84 – 22] 1.2 [0.39 – 3.1] b 09 5 163 [82 – 216] ab 112 [44 – 312] b 29 [6.9 – 46] a 196 [72 – 513] 1.5 [1.1 – 1.7] b 3.4 [1.7 – 6.0] 0.09 [0.05 – 0.17] b 10 8 186 [109 – 257] a 37 [23 – 53] b 7.9 [3.9 – 14] b 163 [63 – 239] 1.6 [1.1 – 3.1] b 2.1 [1.4 – 2.9] 0.18 [0.10 – 0.26] b 859

860 Table 4 Exchangeable elements of rhizosphere soils. Mean and range are provided, n is the number of samples. Data followed by different letters

861 are significantly different at p < 0.05 (TukeyHSD test)

Site n CEC (cmol+ kg−1) Ca (cmol+ kg−1) K (cmol+ kg−1) Mg (cmol+ kg−1) Mn (cmol+ kg−1) Ni (cmol+ kg−1) 01 3 17 [8.0 – 31] ab 5.7 [0.49 – 14] 0.15 [0.08 – 0.22] b 7.9 [5.5 – 12] bcd 0.40 [0.16 – 0.58] 0.19 [0.12 – 0.27] 02 6 27 [22 – 37] ab 11 [5.5 – 23] 0.28 [0.18 – 0.45] ab 12 [10 – 16] bcd 0.18 [0.11 – 0.28] 0.20 [0.09 – 0.50] 03 5 25 [22 – 30] ab 3.6 [1.9 – 6.4] 0.25 [0.21 – 0.30] ab 18 [15 – 20] bcd 0.11 [0.03 – 0.19] 0.07 [0.02 – 0.10] 04 3 33 [18 – 49] ab 11 [1.1 – 28] 0.26 [0.07 – 0.44] ab 18 [13 – 23] abcd 0.13 [0.02 – 0.33] 0.08 [0.03 – 0.17] 05 4 43 [36 – 53] a 1.0 [0.59 – 1.7] 0.60 [0.45 – 0.73] a 35 [27 – 47] a 0.33 [0.18 – 0.50] 0.16 [0.06 – 0.30] 06 8 30 [18 – 42] ab 5.8 [0.62 – 17] 0.37 [0.13 – 0.60] ab 19 [14 – 29] b 0.31 [0.09 – 0.80] 0.16 [0.05 – 0.25] 07 4 30 [24 – 43] ab 5.0 [2.1 – 7.7] 0.25 [0.17 – 0.33] ab 21 [15 – 30] abc 0.06 [0.01 – 0.22] 0.05 [0.01 – 0.16] 08 6 40 [18 – 76] a 14 [3.2 – 28] 0.35 [0.09 – 0.86] ab 20 [7.8 – 40] b 0.07 [0.01 – 0.11] 0.05 [0.02 – 0.09] 09 5 12 [6.9 – 24] b 2.7 [1.0 – 5.9] 0.15 [0.07 – 0.28] b 5.2 [2.3 – 13] cd 0.33 [0.02 – 1.29] 0.21 [0.08 – 0.45] 10 8 17 [8.7 – 33] b 8.2 [5.0 – 23] 0.15 [0.09 – 0.25] b 5.0 [1.0 – 7.9] d 0.06 [0.01 – 0.11] 0.13 [0.03 – 0.25] 862

863 Table 5 Nickel concentrations in soils ([Ni]soils), leaves ([Ni]leaves) and roots ([Ni]roots) and bioconcentration (BCF) and translocation (TF) factors

864 from Ni hyperaccumulator plant species. Mean and range deviation are provided, n is the number of samples. Data followed by different letters are

865 significantly different at p < 0.05 (TukeyHSD test)

Plant n [Ni]soils [Ni]leaves [Ni]roots BCF TF (mg g−1) (mg g−1) (mg g−1) F. kinabaluensis 6 1.9 [0.6–3.4] ab 5.2 [2.2–8.1] c 0.7 [0.2–1.3] cd 27 [8.8–44] ab 9.3 [3.3–16] a M. sundaicus 6 2.0 [0.4–3.5] ab 2.2 [0.6–3.6] c 0.2 [0.08–0.4] d 16 [7.4–34] b 9.6 [8.2–15] a P. balgooyi 10 1.8 [1.1–3.4] b 5.0 [1.8–11] c 2.8 [0.6–6.3] abc 44 [15–146] ab 2.0 [0.8–3.3] b P. rufuschaneyi 6 3.4 [3.1–3.7] a 19 [6.9–34] a 4.9 [2.8–6.4] a 95 [16–195] a 3.7 [2.2–6.4] b P. sarmentosa 5 1.8 [0.5–2.8] b 12 [0.04–21] abc 3.8 [0.2–6.8] ab 62 [0.7–98] ab 2.8 [0.2–5.2] b Rinorea sp. 3 2.3 [0.8–3.0] ab 7.5 [3.9–14] abc 1.9 [1.1–2.6] abcd 79 [32–137] ab 3.7 [2.1–5.4] b R. aff. bengalensis 9 2.7 [1.9–3.7] ab 16 [6.1–30] ab 5.1 [2.5–7.6] a 63 [21–111] ab 3.1 [1.7–4.5] b R. aff. javanica 4 2.4 [1.2–3.8] ab 15 [10–21] abc 3.1 [1.4–5.1] abcd 66 [33–101] ab 5.5 [3.5–7.8] ab W. pinnata 3 2.5 [2.2–2.8] ab 3.1 [1.8–4.1] bc 0.3 [0.1–0.5] bcd 18 [16–21] ab 11 [8.4–17] a 866

Table 6 α-diversity of rhizosphere soils. Mean and range are provided, n is the number of samples. Data followed by different letters are significantly different at p < 0.05 (TukeyHSD test)

Sites n Observed OTUs Chao1 Shannon Site 01 3 853 [711 - 935] d 930 [772 - 1014] c 8.25 [8.05 - 8.41] ab Site 02 6 1574 [1330 - 1702] abc 1709 [1546 - 1807] a 8.82 [7.11 - 9.35] a Site 03 5 1682 [1501 - 1824] ab 1806 [1630 - 1935] a 9.34 [8.91 - 9.66] a Site 04 3 1433 [1139 - 1661] abcd 1555 [1231 - 1824] ab 9.05 [8.57 - 9.34] ab Site 05 4 1398 [1138 - 1639] abcd 1562 [1320 - 1792] ab 8.77 [7.74 - 9.39] ab Site 06 8 1748 [1488 - 1861] a 1867 [1635 - 1992] a 9.40 [8.96 - 9.71] a Site 07 4 1233 [1038 - 1602] bcd 1427 [1211 - 1816] abc 7.48 [6.39 - 8.57] b Site 08 6 1538 [1320 - 1723] abc 1693 [1545 - 1808] a 8.65 [6.79 - 9.51] ab Site 09 5 1140 [604 - 1663] cd 1222 [643 - 1774] bc 8.44 [7.78 - 9.31] ab Site 10 8 1524 [984 - 1684] abc 1637 [1042 - 1853] ab 9.00 [8.47 - 9.38] a

33