The X-ray fluorescence screening of multiple elements in herbarium specimens from the Neotropical region reveals new records of metal accumulation in plants Célestine Belloeil, Pierre Jouannais, Charles Malfaisan, Rolando Reyes Fernández, Severine Lopez, Dulce Montserrat Navarrete Gutierrez, Swann Maeder-Pras, Paola Villanueva, Romane Tisserand, Melina Gallopin, et al. To cite this version: Célestine Belloeil, Pierre Jouannais, Charles Malfaisan, Rolando Reyes Fernández, Severine Lopez, et al.. The X-ray fluorescence screening of multiple elements in herbarium specimens from the Neotropical region reveals new records of metal accumulation in plants. Metallomics, Royal Society of Chemistry, 2021, 13 (8), ฀10.1093/mtomcs/mfab045฀. ฀hal-03298402฀ HAL Id: hal-03298402 https://hal.archives-ouvertes.fr/hal-03298402 Submitted on 23 Jul 2021 HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. 1 The X-ray fluorescence screening of multiple elements in herbarium 2 specimens from the Neotropical region reveals new records of metal 3 accumulation in plants 4 5 Célestine Belloeil1, Pierre Jouannais1, Charles Malfaisan1,2, Rolando Reyes Fernández3, 6 Severine Lopez4, Dulce Montserrat Navarrete Gutierrez5,6, Swann Maeder-Pras1, Paola 7 Villanueva1, Romane Tisserand5, Melina Gallopin1, Dubiel Alfonso-Gonzalez7, Ilsa M. Fuentes 8 Marrero8, Serge Muller2, Vanessa Invernon2, Yohan Pillon9, Guillaume Echevarria5,10, Rosalina 9 Berazaín Iturralde7, Sylvain Merlot1* 10 11 1 12 91198, Gif-sur-Yvette, France 13 2 14 naturelle, CNRS, Sorbonne Université, EPHE, Université des Antilles, Paris, France 15 3 16 Biotecnología Vegetal, Mayabeque, Cuba, CP: 32700 17 4 ECONICK, 1 rue Granville, 54000 Nancy, France 18 5 Université de Lorraine, INRAE, Laboratoire Sols et Environnement (LSE), 54000 Nancy, 19 France 20 6 Universidad Autónoma de Chapingo, Texcoco de Mora, State of México, México. 21 7 Jardín Botánico Nacional, Universidad de La Habana, La Habana, Cuba. CP: 19230. 22 8 Instituto de Ecología y Sistemática, Ministerio de Ciencia, Tecnología y Medio Ambiente, La 23 Habana, Cuba, C.P : 11900 24 9 25 Institut Agro, Univ. Montpellier, Montpellier, France Université Paris-Saclay, CEA, CNRS, Institute for Integrative Biology of the Cell (I2BC), Institut de Systématique, Evolution, Biodiversité (ISYEB), Muséum national d’Histoire Universidad Agraria de La Habana (UNAH), Facultad de Agronomía, Laboratorio Laboratoire des Symbioses Tropicales et Méditerranéennes (LSTM), IRD, INRAE, CIRAD, 1 1 10 2 Australia Centre for Mined Land Rehabilitation, SMI, University of Queensland, QLD 4072 St. Lucia, 3 4 *For correspondence. E-mail sylvain.merlot@i2bc.paris-saclay.fr 5 Running head: Metal Accumulation in the Neotropical Flora 6 7 KEYWORDS 8 9 X-Ray Fluorescence, Metal, Hyperaccumulator, Cuba, Ionomics, Agromining 10 2 1 ABSTRACT 2 Plants have developed a diversity of strategies to take up and store essential metals in order to 3 colonize various types of soils including mineralized soils. Yet, our knowledge of the capacity 4 of plant species to accumulate metals is still fragmentary across the plant kingdom. In this study, 5 we have used the X-Ray Fluorescence technology to analyze metal content in a wide diversity 6 of species of the Neotropical flora that was not extensively investigated so far. In total, we 7 screened more than 11 000 specimens representing about 5000 species from herbaria in Paris 8 and Cuba. Our study provides a large overview of the accumulation of metals such as 9 manganese, zinc and nickel in the Neotropical flora. We report 30 new nickel 10 hyperaccumulating species from Cuba, including the first records in the families Connaraceae, 11 Melastomataceae, Polygonaceae, Santalaceae and Urticaceae. We also identified the first 12 species from this region of the world that can be considered as manganese hyperaccumulators 13 in the genera Lomatia (Proteaceae), Calycogonium (Melastomataceae), Ilex (Aquifoliaceae), 14 Morella (Myricaceae) and Pimenta (Myrtaceae). Finally, we report the first zinc 15 hyperaccumulator, Rinorea multivenosa (Violaceae), from the Amazonas region. The 16 identification of species able to accumulate high amounts of metals will become instrumental 17 to support the development of phytotechnologies in order to limit the impact of soil metal 18 pollution in this region of the world. 19 3 1 Introduction 2 The ability of plants to colonize most of the terrestrial ecosystems reside in their capacity to 3 take up and store nutrients and metals from the soil. Metals such as iron (Fe), copper (Cu), 4 manganese (Mn), zinc (Zn), and nickel (Ni) are cofactors of numerous enzymes playing 5 essential biological functions including respiration and photosynthesis. However, when present 6 in excess these metals play adverse effects, triggering oxidative and genotoxic stresses.1–3 7 Therefore, every plant has to regulate the homeostasis of metals according to the availability of 8 metals and their needs during life cycle. The regulation of metal homeostasis is particularly 9 challenging for plant species growing on soils depleted in available metals or on metalliferous 10 soils enriched in metals such as ultramafic (e.g. serpentine) and calamine soils.4,5 A limited 11 number of species have acquired the capacity to tolerate and accumulate high concentration of 12 metals in their above-ground tissues. These species are called metal hyperaccumulators when 13 the concentration of metal in above-ground tissues collected on plants growing in their natural 14 environment reaches a threshold presently fixed at 1000 ppm (or 1000 ug/g dry weight) for Ni, 15 3000 ppm for Zn and 10 000 ppm for Mn, corresponding to 2-3 orders of magnitude above the 16 levels recorded in plant leaves growing on common soils.6,7 Today, about 700 metal 17 hyperaccumulator species have been identified, with the large majority (c. 530 species) 18 hyperaccumulating Ni.8 Metal hyperaccumulators are widely distributed across the tree of life 19 in more than 70 families but are found more frequently in some specific eudicotyledon families, 20 ie Phyllanthaceae, Brassicaceae, Asteraceae, Cunoniaceae, Euphorbiaceae and Salicaceae. The 21 capacity of plants to accumulate metals is a rare trait even in metalliferous flora suggesting an 22 important evolutionary cost. However, metal accumulation is also proposed to provide selective 23 advantage through the protection from insects and pathogens, allelopathy or adaptation to 24 drought.9,10 Species that are able to accumulate metals thus represent good models to identify 25 the mechanisms involved in the regulation of metal homeostasis in plants and to study the 4 1 evolution of this complex adaptive trait.11–13 Hyperaccumulator species are also foreseen as 2 crops to extract and recycle metals from contaminated or naturally metal-rich soils.14,15 3 The number of plant species known to accumulate metals has considerably increased in the past 4 few years thanks to the use of handheld X-ray fluorescence (XRF) analyzer to screen herbarium 5 collections from Malaysia, Papua New Guinea and New Caledonia. These regions contain very 6 diverse floras developing on ultramafic soil.16–18 Handheld XRF analyzers are particularly well 7 suited to screen large herbarium collections because they allow to measure non-destructively 8 virtually all elements from magnesium to uranium, very rapidly and with a good sensitivity on 9 dry leaf material.19,20 However, the XRF spectra analysis software used in these analyzers are 10 usually not developed for the quantification of elements from carbon-based matrix. Therefore, 11 the precise quantification of metals from dry leaves still requires a calibration and a 12 confirmation of XRF data using quantitative methods including inductively coupled plasma 13 (ICP) or microwave plasma (MP)-atomic emission spectrometers (AES). 14 Despite the fact that the Neotropical region is the richest region for plant species on Earth,21 15 handheld XRF was not used yet to perform large scale screens of herbarium specimens from 16 this area. The dimethylglyoxime assay was previously used to screen herbaria from Cuba and 17 Brazil, both countries containing large ultramafic outcrops, and allowed the identification of 18 130 and 20 Ni hyperaccumulators respectively.22–25 In addition, Ni hyperaccumulators of the 19 Psychotria genus (Rubiaceae) were identified in several countries of the Neotropical region.26,27 20 However, to our knowledge, no Zn or Mn hyperaccumulators had yet been reported from the 21 Neotropical region. 22 In this work, we used handheld XRF to screen large collections of plant specimens in the 23 herbarium of the French National Museum of Natural History in Paris (P) and in herbaria in 24 Cuba (HAJB and HAC) to identify plant species able to accumulate metals from the Neotropical 25 region. This analysis provides a large picture of metal accumulation in plant species from this 5 1 region. In particular, we have identified new accumulating species for Ni, Mn and Zn. Some of 2 these species belong to plant families or genera that were not known so far to contain metal 3 accumulating species. We believe that these original data will be instrumental to broaden our 4 knowledge of the mechanisms involved in metal accumulation in plants and to develop 5 sustainable phytotechnologies to extract metals from soil in the neotropical region. 6 6 1 Materials and methods 2 Handheld XRF screening of herbarium specimens 3 We used a Niton XL3T 950 GOLDD+ handheld XRF analyzer (Thermo Fisher Scientific, 4 Waltham, Massachusetts, USA) to screen herbarium specimens essentially as previously 5 described 6 Saint Louis, Missouri, USA) and the X-ray fluorescent spectra were recorded in the soil mode 7 for 30 sec. using the main filter. The analysis of the spectra by the proprietary XRF 8 quantification software (ver. HH-XRF_8.4G) provided an estimated concentration in part per 9 million (ppm) for the following elements: As, Au, Co, Cu, Fe, Hg, Mn, Mo, Pb, Rb, Sr, Th, U, 10 W, Zn, Zr. The handheld XRF analyzer was associated with a barcode reader (Motorola 11 CS3070) to record the specimen identification number when available. 12 Strategy of herbarium screening 13 We screened the Herbarium of the French National Museum of Natural History in Paris (P) 14 containing c. 6 million specimens of vascular plants organized according to the APGIII 15 classification.28 We restricted the screen to the America sector (AME) as defined in P, 16 excluding specimens from USA and Canada. The screen of P was mostly targeted at species 17 belonging to clades known to contain metal hyperaccumulator species including families of the 18 COM clade (eg Cunoniaceae, Euphorbiaceae, Phyllanthaceae, Violaceae, Salicaceae), 19 Proteaceae and Rubiaceae (entire genus Psychotria). When possible, we screened 3 to 5 20 specimens collected from distinct geographical areas per species. The metadata linked to each 21 specimen were recovered from the P herbarium database Sonnerat using the unique barcode 22 number, or manually collected from the label of the specimen with the help of the citizen 23 science program “Les herbonautes” (http://lesherbonautes.mnhn.fr/missions/13907009). 18 . Each specimen was placed on a 2 mm titanium foil (#269489, Sigma-Aldrich, 7 1 In addition, species of the Cuban flora were screened at the Johannes Bisse Herbarium (HAJB) 2 of the Jardín Botánico Nacional in Havana. In this case, we screened every dicotyledonous 3 species from the reference collection that contains one specimen of all species present in Cuba. 4 For species and genera with noticeable concentration of metals, we screened additional 5 specimens at HAJB and at the Onaney Muñíz Gutiérrez Herbarium (HAC) of the Institute of 6 Ecology and Systematics in Havana. For the Cuban flora, we used the most recent nomenclature 7 of the species available.29 Following preliminary results, additional specimens of Rinorea 8 multivenosa and R. longistipulata were kindly provided as a loan by the Missouri Botanical 9 Garden Herbarium (MO). 10 11 Quantitative elemental analysis and calibration of the XRF data 12 Dry leaf samples (c. 50 mg) were weighted and mineralized in polypropylene tubes (#352070, 13 Falcon brand-Corning, Tewksbury, USA) at 120°C in a heating block for 4 h in 2 mL of 65 % 14 HNO3 (#30709-M, Sigma-Aldrich, Saint Louis, USA) and additionally for 4 h after the addition 15 1 mL of 30% H2O2 (#7722-84-1, Sigma-Aldrich, Saint Louis, Missouri, USA). Mineralized 16 samples were analyzed on an Agilent 4200 MP-AES (Agilent Technologies, Santa Clara, USA). 17 The concentration of metals in the samples were calculated using metal standard calibration 18 curves. For the calibration of the XRF analyses, we gathered 124 samples of dry leaves from 19 plant species accumulating various amounts of Ni and Mn originating from herbarium 20 specimens and samples collected in the field. The samples were analyzed in parallel by XRF 21 and MP-AES as described above. A linear regression modelling the relationship between the 22 elemental concentration measured by XRF and MP-AES was obtained for Ni ([Ni] MP-AES= 23 0.6762 [Ni] XRF; r2=0.94) and Mn ([Mn] MP-AES= 0.2793 [Mn] XRF; r2=0.95) (Fig. S1). 24 8 1 Data representation and statistical analyses 2 The cartographies were generated using the QGIS software (version 2.18.28, QGIS Geographic 3 Information System, Open Source Geospatial Foundation Project, http://qgis.osgeo.org) with 4 the GRASS v. 7.6.0 extension. Publicly available base maps originate from Natural Earth 5 (Admin 0 - Countries, https://www.naturalearthdata.com). 6 For metal concentration analysis, unless indicated, we used the elemental concentration 7 provided by the handheld XRF analyzer. Metal concentrations below to the limit of detection 8 (LOD) were replaced by the lowest concentration measured in this study: Zn (6 ppm), Cu (8 9 ppm), Ni (29 ppm) and Mn (39 ppm). 10 Statistical analyses and data representation were performed using R v.3.6.330 and R studio 11 v.1.3.109. The visualization of data was performed with R packages ggplot2 v.3.3.231 and dplyr 12 v.1.0.2.32 The analysis of the relationship between metal concentration and plant families was 13 analyzed using a linear mixed effect model (LMM) with the R packages lme4 v.1.1.23, 14 lmerTest v.3.1.2 and car v3.0-10.33–35 Metal concentration data were Log10 transformed. The 15 models were constructed using the plant family as a fixed effect and both genus and species as 16 random effects. p-values were obtained using the Satterthwaite's approximation for degrees of 17 freedom. The R script used for this analysis is available upon request to the authors. 18 9 1 Results and discussion 2 Extent of the X-Ray Fluorescent screen of the Neotropical flora 3 Combining the X-Ray Fluorescent (XRF) screen of the herbaria in Paris (P) and in Cuba (HAJB 4 and HAC), we analyzed a total of 11 321 herbarium specimens originating from 39 countries 5 and territories of the Neotropical region (Fig. 1, Table S1). We emphasized the investigation of 6 the Cuban flora (4282 specimens) because Cuba is known to contain a large diversity of Ni 7 accumulating species. However, this flora was not investigated so far to identify species able 8 to accumulate other metals than Ni.22,23 The important representation of specimens from Brazil 9 (2155 specimens) and French Guiana (1181 specimens) reflects the richness of the P herbarium 10 for these flora.28 The screened specimens correspond to 5053 plant species, mostly 11 dicotyledonous, distributed in 933 genus and representing 161 families (Fig. S2). The screen of 12 the P herbarium was targeted to species from families and genera known to contain metal 13 hyperaccumulators. As a consequence, the most represented families in our study are Rubiaceae 14 (2325 specimens), Phyllanthaceae (832 specimens), Salicaceae (777 specimens), Celastraceae 15 (741 specimens), Chrysobalanaceae (644 specimens) and Violaceae (539 specimens). 16 Overall, our screen covers approximately 5% of the diversity of the Neotropical flora that is 17 estimated to contain from 90 000 to 100 000 seed plant species.21 18 Distribution of metal concentration in herbarium specimens 19 The concentration of several elements was measured in parallel by handheld XRF in leaves of 20 the 11 321 herbarium specimens (Table S1). The distributions of the concentrations of the 21 essential metals Cu, Mn, Zn and Ni measured in these specimens are presented in Fig. 2. The 22 concentration of Fe measured in herbarium specimens was not considered in our study because 10 1 a fraction of specimens is contaminated with Fe-rich soil particles, randomly affecting the 2 quantification of the concentration of this element. 3 The distribution of Cu concentrations measured in the herbarium specimens is relatively 4 restricted around the median value of 79 ppm (mean=91 ppm), with only few outliers or 5 samples below the limit of detection (LOD) of 8 ppm. In contrast, the concentrations of Mn and 6 Zn measured in the specimens showed a wider distribution around the median values of 218 7 ppm (mean=650 ppm) for Mn and 50 ppm (mean=93 ppm) for Zn. Only a limited number of 8 specimens displayed a concentration below the LOD for these metals that was estimated at 39 9 ppm for Mn and 6 ppm for Zn. For both metals, we also observed high outliers with metal 10 concentrations up to 48 811 ppm for Mn and 24 267 ppm for Zn, which likely indicates the 11 presence of hyperaccumulator specimens. These results indicated that these two essential 12 metals are available to plants in most of the soils where the specimens have been collected and 13 that plant species can accommodate with wide concentrations of Mn and Zn. The Ni 14 concentrations measured in herbarium specimens showed a very different distribution. A large 15 fraction (85%) of the specimens contains Ni below the LOD of 29 ppm. This result indicates 16 that the sensitivity of handheld XRF is not sufficient to detect Ni in the majority of specimens 17 originating from standard soils. We were however able to detect and measure Ni concentration 18 in 1742 specimens, likely originating from Ni enriched soils such as the ultramafic outcrops 19 from Cuba. In these specimens, Ni concentrations showed a right skewed distribution with a 20 median value of 124 ppm (mean=4306 ppm) and ranging from 29 ppm to 71 901 ppm Ni. This 21 very large distribution of Ni concentrations can be explained by the presence in this dataset of 22 Ni concentrations measured from either excluder or hyperaccumulator species.36 23 11 1 Analysis of manganese accumulation at the family level 2 Because the concentration of Mn can be measured in most of the specimens and displayed a 3 wide distribution, it was possible to test for differences in Mn accumulation at the plant family 4 level. The analysis of the distribution of Mn concentrations in 10 families selected from visual 5 inspection of the data and containing more than 30 specimens revealed large variations of the 6 mean Mn concentration from 443 ppm in Rubiaceae to 4030 ppm in Aquifoliaceae (Fig. 3; 7 Table 1). Using a linear mixed effect model (Fig. 3, Table S2), we did not reveal significant 8 difference (p>0.01) in the accumulation of Mn between the Proteaceae family, known for its 9 propensity to accumulate high concentrations of Mn,37 and the Ochnaceae, Clusiaceae, 10 Melastomataceae, Theaceae and Aquifoliaceae families. In contrast, we observed significant 11 differences (p<0.01) in the mean concentration of Mn between Proteaceae and the Violaceae, 12 Salicaceae, Phyllanthaceae and Rubiaceae families. 13 Our data thus reveal that several plant families tend to accumulate higher concentration of Mn 14 suggesting that the species from these families have developed specific nutrition strategies 15 leading to an increased uptake, tolerance and accumulation of Mn.38 The high capacity of 16 Proteaceae species to take up and accumulate Mn is proposed to be the consequence of their 17 phosphorus nutrition strategy based on the exudation of large amounts of organic acid in the 18 rhizosphere. The acidification of the rhizosphere leads to the solubilization of phosphate and 19 other elements including Mn.37 It was also showed that Proteaceae species are able to 20 accumulate high amount of Aluminum (Al) and that there is a trade-off between Al and Mn 21 accumulation in these species.39,40 These observations suggest that the uptake and accumulation 22 of both Mn and Al in Proteaceae depends on the acidification of the rhizosphere. Interestingly, 23 species of the Melastomataceae and Theaceae families are known to accumulate high amounts 24 of Al.41,42 Here, we observed that species of the Melastomataceae and Theaceae families also 25 have a tendency to accumulate high amounts of Mn. It will be therefore interesting to test if 12 1 species of these families as well as species of the Aquifoliaceae and Ochnaceae families use a 2 nutrition strategy based on the important acidification of the rhizosphere, leading to the 3 solubilization and accumulation of Mn or Al depending on the availability of these metals in 4 the soil. 5 6 Identification of species accumulating manganese 7 The XRF screen of the herbaria revealed 69 specimens containing above 10 000 ppm Mn in 8 leaves (Table S1). These specimens belong to 30 species from 18 genera (14 families) including 9 Roupala and Lomatia (Proteaceae), Calycogonium and Miconia (Melastomataceae), Morella 10 (Myricaceae), Pimenta (Myrtaceae), and Ilex (Aquifoliaceae) (Table 2; Fig. 4). The analysis of 11 several specimens from these species revealed a large variability of Mn content suggesting a 12 high intraspecific genetic variability and/or a strong influence of edaphic conditions on Mn 13 accumulation. To obtain a better quantification of Mn accumulation in these specimens, we 14 calibrated the XRF measurements with MP-AES on a set of plant samples accumulating various 15 amounts of metals. This calibration indicated that the XRF analysis overestimates the 16 concentration of Mn in dry leaves material by a factor of 3.58 in our experimental set-up (Fig. 17 S1). According to this calibration, only three species, Lomatia dentata, Pimenta oligantha and 18 Morella shaferi, are predicted to accumulate more than 10 000 ppm Mn currently used as a 19 threshold to define Mn hyperaccumulators (Table 2). The direct measurement of Mn 20 concentration in herbarium specimens by MP-AES confirmed that P. oligantha and M. shaferi 21 are able to accumulate more than 10 000 ppm Mn in leaves and further revealed that Ilex 22 obcordata and Calycogonium bissei can also be considered as Mn hyperaccumulators. The 23 confirmation that the other species accumulating high concentration of Mn can be considered 24 as hyperaccumulators will require further sampling and quantitative elemental analyses. 13 1 The species accumulating above 10 000 ppm Mn in their leaves described in this study 2 represent, to our knowledge, the first Mn hyperaccumulator species described in the 3 Neotropical flora including Cuba. It was recently shown that several Proteaceae species 4 belonging to the Roupala, Embothrium and Lomatia genera growing in Chile accumulate high 5 concentrations of Mn but so far none of the analyzed sample had ever been reported to reach 6 the hyperaccumulation threshold.39 Here, we have identified a sample of Lomatia dentata 7 collected in the Valdivia region in Chile that is predicted to accumulate more than 10 000 ppm 8 Mn. This result also suggests that additional sampling of Proteaceae species from South 9 America may reveal additional Mn hyperaccumulators. More interestingly, we identified new 10 Mn hyperaccumulators such as Ilex obcordata (Aquifoliaceae), Morella shaferi (Myricaceae) 11 and Calycogonium bissei (Melastomataceae) belonging to families that were not known to 12 contain Mn hyperaccumulators. Recently, Ilex paraguariensis, also known as “yerba mate”, 13 was shown to be able to accumulate more than 10 000 ppm Mn when grown ex-situ on an Mn 14 enriched soil.43 Together with the identification of I. obcordata, these results suggest that Mn 15 hyperaccumulation could be revealed in other Ilex species. More generally, our results support 16 the idea that, in contrast to Ni hyperaccumulation, the hyperaccumulation of Mn does not define 17 a specific biological adaptation strategy but rather describe the capacity of some species to 18 accumulate and tolerate very high concentration of Mn when this metal is highly available in 19 soils.36 The frequency distribution of Mn concentrations at the level of the Neotropical flora 20 appears to be unimodal (Fig. 2). Furthermore, the species in which we have recorded Mn 21 concentrations above 10 000 ppm belong to plant families that have a propensity to accumulate 22 elevated concentration of Mn (Fig. 3). Finally, even in these Mn hyperaccumulator species we 23 observed a very dispersed distribution of Mn concentration between specimens (Fig. 4). These 24 results suggest that the threshold of 10 000 ppm Mn used to qualify Mn hyperaccumulators 25 defines a limit in the upper tail of the distribution of Mn concentrations measured in plants 14 1 rather than separates two distinctive modes (ie non-hyperaccumulator and hyperaccumulator) 2 in this distribution as previously described for Ni.36 3 4 Analysis of the accumulation of zinc 5 We identified 43 specimens containing more than 1000 ppm Zn in leaves as measured by XRF 6 (Table S1). However, we suspected that several specimens from Cuban herbaria collections 7 were contaminated by a Zn-rich glue used for the mounting of the specimens. Such 8 contamination with a glue was previously observed in specimens from the herbarium of Sabah. 9 18 Nevertheless, we identified in the P herbarium four specimens of Rinorea multivenosa 10 (Violaceae) collected in Brazil, containing from 9280 to 17 830 ppm Zn (Fig. 5). The 11 concentration of Zn measured in R. multivenosa specimens is significantly higher than the Zn 12 concentration recorded in the other species of the Rinorea genus or more generally in the 13 Violaceae family. To confirm the high accumulation of Zn in leaves of R. multivenosa, we 14 analyzed six additional specimens collected in Brazil and Colombia obtained from the Missouri 15 Botanical Garden herbarium (MO). The analysis of the samples from MO confirmed the 16 capacity of R. multivenosa to accumulate high concentration of Zn (8517 ± 4936 ppm, n=10) 17 in leaves (Table 3). As a comparison, we measured low concentrations of Zn (56 ± 26 ppm, 18 n=6) in R. paniculata. Interestingly, the analysis of 3 specimens of R. longistipulata from MO 19 revealed that this species, closely related to R. multivenosa and originating from the same 20 geographical region,44 is also able to accumulate noticeable amount of Zn (1055 ± 314 ppm, 21 n=3). 22 The highest concentrations of Zn measured by XRF in leaves of R. multivenosa are very likely 23 to exceed the 3000 ppm threshold value currently used to define Zn hyperaccumulators. R. 24 multivenosa, a 2-3 m tall tree growing in the Amazonia region,44 would thus correspond to the 25 first Zn hyperaccumulator species identified in the Neotropical region. The labels of the R. 15 1 multivenosa specimens that have been analyzed do not provide precise information of the nature 2 of the soil on which they have been collected. However, the accumulation of high Zn in leaves 3 of one cultivated specimen collected in 1874 in the Quinta da Boa Vista park (P02141358), that 4 later became the zoological garden of Rio de Janeiro, suggest that R. multivenosa is able to 5 accumulate Zn from soils that are not particularly enriched in Zn. This hypothesis still needs to 6 be supported by quantitative measurement of Zn content in freshly-collected R. multivenosa 7 leaves and corresponding soil samples. Interestingly, the closely related species R. 8 longistipulata is also able to accumulate noticeable amounts of Zn, suggesting that the Zn 9 accumulation trait was present in the common ancestor of these 2 species. Previously, 10 specimens of Rinorea longiracemosa from Sabah (Malaysia, Borneo Island) were shown to 11 accumulate high concentrations of Zn in leaves.8 The Rinorea genus is also known to host 12 several noticeable Ni hyperaccumulators from the south east Asian region, eg R. bengalensis 13 45–47 14 its polyphyly observed in molecular phylogenetic studies.48 These results thus indicate that both 15 Zn and Ni hyperaccumulation appeared repeatedly in the Rinorea genus as this was previously 16 documented for the Noccaea (Brassicaceae) and Dichapetalum (Dichapetalaceae) 17 genera.8,47,49,50 It should be noted that the limits of the genus Rinorea require re-appraisal as suggested by 18 19 Analysis of the accumulation of nickel 20 We screened the P herbarium for species belonging to families and genera known to contain Ni 21 hyperaccumulators including Phyllanthus (Phyllanthaceae), Psychotria (Rubiaceae) and 22 Violaceae. As expected, our XRF screen identified specimens with foliar Ni concentration 23 largely exceeding 1000 ppm in the Psychotria species P. grandis, P. costivenia and P. 24 papantlensis that are known Ni hyperaccumulators from the Neotropical region.22,27 We also 25 confirmed the capacity of Phyllanthus nummularoides from the Dominican Republic to 16 1 hyperaccumulate Ni.51 More interestingly, we measured high concentration of Ni in leaves of 2 Blepharidium guatemalense (≡ B. mexicanum, Rubiaceae) and in four Orthion (Violaceae) 3 species O. caudatum, O. malpighiifolium, O. oblanceolatum and O. subsessile, as well as in the 4 closely related monospecific genus Mayanea (M. caudata) from Guatemala, that were recently 5 identified as new Ni hyperaccumulators from Guatemala and Mexico.52,53 6 The systematic screen of the HAJB herbarium and the screen of the HAC herbarium in Cuba 7 confirmed the accumulation of Ni in most of the hyperaccumulator species previously identified 8 in the Cuban flora (Table S3).22,23 Our work provides an updated nomenclature of the Cuban 9 species and references to herbarium specimens. We further observed the accumulation of Ni 10 above 1000 ppm in new species belonging to genera that had previously been investigated in 11 the Cuban flora, including Buxus braimbridgeorum, Mosiera acunae, Sapium parvifolium, 12 Euphorbia podocarpifolia and Phyllanthus excisus (Table 4). In contrast, we did not confirm 13 the accumulation of high Ni concentration in Anastraphia crassifolia (≡Gochnatia crassifolia, 14 Asteraceae), 15 erythrospermum, Euphorbiaceae), Ouratea nitida (Ochnaceae) and Psychotria osseana 16 (Rubiaceae). In few cases (eg Leucocroton genus), specimens corresponding to species 17 previously identified as Ni hyperaccumulators were not available at the time of our study in 18 HAJB, precluding the confirmation of previous observations. The differences observed with 19 previous studies could be explained by changes in the identification of species or because we 20 analyzed different specimens collected on non-ultramafic soils. More interestingly, we 21 identified species able to accumulate high concentrations of Ni in families and genera that were 22 not known to contain Ni accumulators (Fig. 6, Table 4). We measured above 1000 ppm Ni by 23 XRF in specimens from the Aristolochiaceae, Urticaceae, Connaraceae, Melastomataceae, 24 Santalaceae and Polygonaceae families. In particular, 3 species of the Pilea genus (Urticaceae), 25 P. fruticulosa, P. mayarensis and P. microphylla accumulates more than 10 000 ppm Ni in Garcinia ruscifolia (Clusiaceae), Bonania erythrosperma (≡Sapium 17 1 leaves. We also measured above 1000 ppm Ni in specimens from new species from families 2 and genera previously known to contain Ni hyperaccumulators as in Pimenta and Calyptrantes 3 (Myrtaceae), Daphnopsis (Thymelaeaceae), Aristolochia (Aristolochiaceae)54 and Allophylus 4 (Sapindaceae) genera. The identification of 2 specimens (Ekman10003, Alain7754) of Pimenta 5 oligantha accumulating more than 1000 ppm Ni is intriguing because several specimens of this 6 species accumulate high amounts of Mn but not Ni (Table 2). 7 The calibration of XRF data with MP-AES quantification indicated that our XRF analysis 8 overestimates the Ni concentration by a factor of 1.48 (Fig. S1). To confirm the quantification 9 of Ni, we directly measured metal concentration by MP-AES on some herbarium specimens 10 corresponding to new Ni accumulating species (Table 4). These results confirmed that Pilea 11 fruticulosa is able to accumulate up to 25 700 ppm Ni indicating that this species, and very 12 likely two other Pilea species, P. mayarensis and P. microphylla, are novel Ni 13 hyperaccumulators from the Urticaceae family. This analysis also confirmed that 14 Crossopetalum rhacoma (Celastraceae), Rourea glabra (Connaraceae), Casearia crassinervis 15 (Salicaceae), Allophylus reticulatus (Sapindaceae), Daphnopsis angustifolia (Thymelaeaceae), 16 Dendrophthora tetrastachya (Santalaceae) and Coccoloba oligantha (Polygonaceae), are able 17 to accumulate more than 1000 ppm Ni in leaves and can therefore be considered as new Ni 18 hyperaccumulators from Cuba. However, for other species including Aristolochia lindeniana 19 (Aristolochiaceae), Miconia costata (Melastomataceae) and species of the Calyptranthes genus 20 (Myrtaceae), Ni quantification on additional leaf samples will be necessary to confirm their Ni 21 hyperaccumulator status. 22 Our analysis revealed new Ni hyperaccumulator species in the Neotropical flora, thus 23 expanding our knowledge of the diversity of Ni hyperaccumulators in this region of the world. 24 However, except for the Cuban flora, our screen was targeted to selected plant families and 25 genera known to contain Ni hyperaccumulator species. Other plant groups known to contain Ni 18 1 hyperaccumulators in the Tropics were not specifically targeted.24,25 We therefore think that 2 further efforts to screen tropical floras will reveal additional Ni hyperaccumulator species. We 3 show here that the systematic screen of the Cuban flora allowed us to identify new Ni 4 hyperaccumulators, such as Pilea fruticulosa (Urticaceae), that belong to families and genera 5 that were not known to contain hyperaccumulators before. These new findings will likely 6 motivate the targeted investigation of related species in future screens to identify new Ni 7 hyperaccumulator species in flora of the pantropical region and in other regions of the world. 8 9 New combinations of functional traits 10 Our screening has revealed several combinations of functional traits that had never been 11 reported before in angiosperms, including metal (hyper)accumulation associated with particular 12 root symbioses. Morella shaferi is a new Mn hyperaccumulator from Cuba (Table 2) and, as a 13 member of Myricaceae, a likely actinorhizal plant forming root nodule with nitrogen fixing 14 Frankia.55 Ni hyperaccumulation has been revealed in the Polygonaceae species Coccoloba 15 baracoensis and possibly in C. oligantha (Table 4), which both belong to a genus that is 16 considered as ectomycorrhizal.56 The only similar case known so far was Shorea tenuiramulosa 17 from Borneo, accumulating up to 1787 ppm Ni and belonging to Dipterocarpaceae, a family 18 widely considered as entirely ectomycorrhizal.57–59 Although the symbiotic status of Morella 19 and Coccoloba remains to be validated, the combination of hyperaccumulation with root 20 symbiosis makes those plants ideal candidates as nurse plants for ecological restoration. 21 In addition, we report novel cases of metal accumulation in parasitic plants. We 22 observed high Ni content in Dendrophthora tetrastachya (Santalaceae), an aerial ligneous 23 parasite, which unidentified host was likely a nickel accumulator (Table 4). We also observed 24 high Mn content in Schoepfia cubensis (Table 2), that belongs to a woody genus considered as 25 hemi-parasitic.60 Therefore, these two members of the order Santalales, which partially or 19 1 entirely depend on other plants for their mineral nutrition, can withstand high metal 2 concentrations in their leaf tissue. Previous reports of parasitism on nickel hyperaccumulating 3 plants include the leafless vine Cuscuta (Convolvulaceae) on herbaceous Streptanthus 4 polygaloides (Brassicaceae) in California, and the non-photosynthetic herbaceous root-parasite 5 Phelipanche (Orobanchaceae) on another herbaceous Brassicaceae, Odontarrhena (Alyssum) 6 in Albania and Lesbos Island.61–63 In these cases, the Ni content of the parasite was much lower 7 than that of the hosts, but sometimes exceeding 1000 ppm. Recently, a Ni concentration of 1350 8 ppm was reported in Amyema scandens (Loranthaceae), a root semi-parasitic woody vine from 9 New Caledonia.17 Thus, even if metal hyperaccumulation is proposed to provide elemental 10 allelopathy and a protection against parasitic plants,10 our results further illustrate that several 11 unrelated lineages of plants with different parasitic strategies evolved metal tolerance and 12 accumulation mechanisms to circumvented the toxicity of their hyperaccumulator host. 13 14 Conclusions 15 In this work, we have used XRF to analyze the elemental composition of more than 11 000 16 herbarium specimens representing more than 5000 plant species and 161 families from the 17 Neotropical region. The quantification of metals in leaves of a wide diversity of plant species 18 can help to identify the phylogenetic effects on plant ionomes, reflecting both nutrition and ion 19 homeostasis strategies.64–68 The screen of the Neotropical flora with handheld XRF led us to 20 the identification of the first hyperaccumulators of Zn and Mn in this region of the world. In 21 addition, the systematic screen of the Cuban flora allowed us to identify new Ni 22 hyperaccumulators belonging to families and genera in which Ni hyperaccumulation was not 23 described before. Our results suggest that further efforts to screen the flora of yet 24 underinvestigated regions such as tropical Africa and Madagascar will undoubtedly yield new 25 metal hyperaccumulator species. Our results also indicate that the systematic screen of flora is 20 1 essential to identify hyperaccumulator species in new plant groups. However, because of the 2 scarcity of digitalized information about the nature of the soil linked to the majority of 3 herbarium specimens, systematic screens are difficult to develop in large herbaria and are 4 probably better adapted to herbaria dedicated to specific geographic regions containing 5 metalliferous outcrops.24 6 Our study also reinforces the position of Cuba as a major hot spot for the diversity of metal 7 hyperaccumulator species. While previous studies reported 130 Ni hyperaccumulators on this 8 island,22,23 we identified 30 new species predicted to hyperaccumulate Ni (Table S3) and the 9 first 4 species hyperaccumulating Mn (Table 2). New Caledonia, which flora contains from 70 10 to 100 Ni hyperaccumulators and from 10 to 70 Mn hyperaccumulators, is another very rich 11 territory for metal hyperaccumulators.8,17 Even if the recent XRF analyses of these flora are 12 difficult to compare because of differences in calibration and because results still need to be 13 confirmed by quantitative analyses,17 Cuba and New Caledonia have a comparable number and 14 diversity of Ni hyperaccumulators, but Cuba has a more limited number of Mn 15 hyperaccumulators (Table S4). The Cuban and New Caledonian metallophyte flora have 16 interesting similarities, but also differences associated with their distinct biogeographical 17 situation.69 On both islands, Ni hyperaccumulation is observed in species of the Celastraceae 18 and Euphorbiaceae families, and Mn hyperaccumulation in Myrtaceae. In addition, Ni 19 hyperaccumulation has been reported in endemic species of the genera Casearia (Salicaceae), 20 Phyllanthus (Phyllanthaceae), and Psychotria (Rubiaceae), that is likely the outcome of 21 convergence. Other important families for Ni hyperaccumulation in Cuba are Asteraceae, 22 Buxaceae, and Ochnaceae, that are poorly represented or absent in New Caledonia.70 On the 23 other hand, in New Caledonia Ni and Mn accumulation are common in the locally diverse 24 families Cunoniaceae and Proteaceae, that are virtually absent in Cuba.29,71 21 1 A better knowledge of the metal hyperaccumulator species at the physiological and molecular 2 levels will become instrumental for the development of phytotechnologies to limit the impact 3 of metal pollution in different region of the world.72 For example the Mn hyperaccumulator and 4 putative nitrogen fixer Morella shaferi, represents a good candidate species for the restoration 5 of degraded soils in Cuba and Mn accumulated in its biomass could be recycled to produce Mn 6 for green chemistry.73 The identification metal hyperaccumulators, as well as the identification 7 of related non-accumulator species in a large diversity of plant families is a prerequisite to 8 identify the genes involved in metal accumulation in plants using comparative studies 74–76. 9 22 1 ACKNOWLEDGEMENTS 2 The authors would like to thank all the members of the Herbaria P, HAJB and HAC for their 3 precious assistance. We also thank the “Les Herbonautes” community for their help with the 4 “Fan de metal” mission (http://lesherbonautes.mnhn.fr/missions/13907009), the Missouri 5 Botanical Garden for the loan of Rinorea specimens, José Luis Gómez Hechavarría (Jardín 6 Botánico, Holguín, Cuba) for the identification of species in Cuban flora and Antony van der 7 Ent for advices on XRF herbarium screening. The authors are grateful to the Groupement 8 d’Intérêt Scientifique sur les Friches Industrielles (GISFI) from the Université de Lorraine for 9 providing key support with the use of one of the portable X-ray Fluorescence apparatus. 10 This work was funded by the CNRS MITI X-Life program (X-TreM project to SM). This work 11 has also benefited from student fellowships from the Saclay Plant Sciences-SPS (ANR-17- 12 EUR-0007) to SMP and from the SCAC of the French Embassy in Cuba to RR and DAG. 13 23 1 Figure legends 2 3 Fig. 1. Extent of the XRF analysis in the Neotropical region. The size of the circles is 4 proportional to the number of screened specimens (indicated within the circle). Countries with 5 the highest number of screened specimens are highlighted. 6 7 Fig. 2. Distribution of metal concentrations measured by XRF in the specimens of the 8 Neotropical region. The distributions of Cu, Mn, Zn and Ni concentrations measured in leaves 9 in all specimens (n=11 321) by XRF are represented using violin plots with a Log10 scale. To 10 better visualize the distribution of Ni, a violin plot containing only the specimens with a 11 concentration above the LOD is represented. In the box plots, the median is represented as a 12 bold line with both the first (Q1) and the third quartile (Q3). Whiskers extend from the 13 minimum to the maximum [± 1.5 (Q3-Q1)]. The outliers, located outside the whiskers, are 14 represented by brown dots. 15 16 Fig. 3. Distribution of Mn accumulations in 10 selected families. The distributions of Mn 17 concentrations are represented using violin plots with a Log10 scale. Families are organized 18 according to the APG IV classification. n represents the number of specimens per family. 19 Significant differences between the mean concentration of Mn in Proteaceae and other families 20 are shown (p<0.01, linear mixed effect model, Table S2). 21 22 Fig. 4. Distribution of Mn accumulations in various genera. The distributions of Mn 23 concentrations are represented using violin plots with a Log10 scale. Black dots represent 24 individual XRF measurements. Because of the high number of specimens in Psychotria and 25 Phyllanthus, only the outliers are represented (brown dots). Genera are organized according to 24 1 the APG IV classification of the corresponding families. n represents the number of specimens 2 per genus. 3 4 Fig. 5. Zn accumulation in the Violaceae family. The distributions of Zn concentration in the 5 Violaceae family, the Rinorea genus, and in the R. longistipulata, R. multivenosa and R. 6 paniculata species are represented using box plots. Dots represent individual XRF 7 measurements. n represents the number of individual XRF scans. 8 9 Fig. 6. Distribution of Ni accumulations in genera from the Cuban flora containing 10 accumulating species. The distributions of Ni accumulations are represented using box plots 11 with a Log10 scale. Dots represent individual XRF measurements. n represents the number of 12 specimens. Genera are classified according to the APG IV classification of the corresponding 13 families. 14 25 1 References 2 1 R. Hänsch and R. R. Mendel, Curr. Opin. Plant Biol., 2009, 12, 259–266. 3 2 I. V Seregin and A. D. Kozhevnikova, Russ. J. Plant Physiol., 2006, 53, 257–277. 4 3 E. Andresen, E. Peiter and H. Küpper, J. Exp. Bot., 2018, 69, 909–954. 5 4 E. Kazakou, P. G. Dimitrakopoulos, A. J. M. Baker, R. D. Reeves and A. Y. Troumbis, 6 7 Biol. Rev., 2008, 83, 495–508. 5 M. Wójcik, C. Gonnelli, F. Selvi, S. Dresler, A. Rostański and J. Vangronsveld, in 8 Phytoremediation, eds. A. Cuypers and J. B. T.-A. in B. R. Vangronsveld, Academic 9 Press, 2017, vol. 83, pp. 1–42. 10 6 U. Krämer, Annu Rev Plant Biol, 2010, 61, 517–534. 11 7 A. van der Ent, A. J. M. Baker, R. D. Reeves, A. J. Pollard and H. Schat, Plant Soil, 12 13 2013, 362, 319–334. 8 14 R. D. Reeves, A. J. M. Baker, T. Jaffré, P. D. Erskine, G. Echevarria and A. van der Ent, New Phytol., 2018, 218, 407–411. 15 9 A. C. Hörger, H. N. Fones and G. M. Preston, Front. Plant Sci., 2013, 4, 395. 16 10 A. Manara, E. Fasani, A. Furini and G. DalCorso, Plant. Cell Environ., 2020, 43, 17 2969–2986. 18 11 J. J. Cappa and E. A. H. Pilon-Smits, Planta, 2014, 239, 267–275. 19 12 U. Krämer, Curr. Opin. Plant Biol., 2018, 42, 66–75. 20 13 M. Hanikenne and C. Nouet, Curr. Opin. Plant Biol., 2011, 14, 252–259. 21 14 A. van der Ent, A. J. M. Baker, R. D. Reeves, R. L. Chaney, C. W. N. Anderson, J. A. 22 Meech, P. D. Erskine, M.-O. Simonnot, J. Vaughan, J. L. Morel, G. Echevarria, B. 23 Fogliani, Q. Rongliang and D. R. Mulligan, Environ. Sci. Technol., 2015, 49, 4773– 24 4780. 25 15 C. M. Grison, A. Velati, V. Escande and C. Grison, Environ. Sci. Pollut. Res., 2015, 26 1 2 22, 5686–5698. 16 3 4 Sarramegna, V. Gei, P. D. Erskine and A. van der Ent, Chemoecology, 2020, 30, 1–13. 17 5 6 V. Gei, S. Isnard, P. D. Erskine, G. Echevarria, B. Fogliani, T. Jaffré and A. van der Ent, Bot. J. Linn. Soc., 2020, 194, 1–22. 18 7 8 C. Do, F. Abubakari, A. C. Remigio, G. K. Brown, L. W. Casey, V. Burtet- A. van der Ent, A. Ocenar, R. Tisserand, J. B. Sugau, G. Echevarria and P. D. Erskine, J. Geochemical Explor., 2019, 202, 49–58. 19 I. Purwadi, V. Gei, G. Echevarria, P. D. Erskine, J. Mesjasz-Przybyłowicz, W. J. 9 Przybyłowicz and A. van der Ent, eds. A. van der Ent, A. J. M. Baker, G. Echevarria, 10 M.-O. Simonnot and J. L. Morel, Springer International Publishing, Cham, 2021, pp. 11 183–195. 12 20 T. Jaffré, Y. Pillon, S. Thomine and S. Merlot, Front. Plant Sci., 2013, 4, 279. 13 21 A. Antonelli and I. Sanmartín, Taxon, 2011, 60, 403–414. 14 22 R. D. Reeves, A. J. M. Baker, A. Borhidi and R. Berazaín, Ann. Bot., 1999, 83, 29–38. 15 23 R. D. Reeves, A. J. M. Baker, A. Borhidi and R. Berazaín, New Phytol., 1996, 133, 16 17 217–224. 24 18 R. D. Reeves, A. J. M. Baker, T. Becquer, G. Echevarria and Z. J. G. Miranda, Plant Soil, 2007, 293, 107–119. 19 25 R. I. Berazaín and T. S. Filgueiras, Rev. del Jardín Botánico Nac., 2002, 23, 67–74. 20 26 R. L. McAlister, D. A. Kolterman and A. J. Pollard, Aust. J. Bot., 2015, 63, 85. 21 27 G. L. McCartha, C. M. Taylor, A. Ent, G. Echevarria, D. M. Navarrete Gutiérrez and 22 23 A. J. Pollard, Am. J. Bot., 2019, 106, 1377–1385. 28 G. Le Bras, M. Pignal, M. L. Jeanson, S. Muller, C. Aupic, B. Carré, G. Flament, M. 24 Gaudeul, C. Gonçalves, V. R. Invernón, F. Jabbour, E. Lerat, P. P. Lowry, B. Offroy, 25 E. P. Pimparé, O. Poncy, G. Rouhan and T. Haevermans, Sci. Data, 2017, 4, 170016. 27 1 29 W. Greuter and R. Rankin, Plantas vasculares de Cuba: inventario preliminar. 2 Vascular plants of Cuba: a preliminary checklist, Botanischer Garten & Botanisches 3 Museum Berlin & Jardín Botánico Nacional, Universidad de La Habana, Universität 4 Berlin Königin-Luise-Str. 6-8, D-14195 Berlin, Germany, 2017. 5 30 R Core Team, R Found. Stat. Comput., 2019. 6 31 H. Wickham, ggplot2, Springer International Publishing, Cham, 2016. 7 32 H. Wickham, R. François, L. Henry and K. Mûller, 2018. 8 33 D. Bates, M. Mächler, B. Bolker and S. Walker, J. Stat. Softw., 9 10 DOI:10.18637/jss.v067.i01. 34 11 A. Kuznetsova, P. B. Brockhoff and R. H. B. Christensen, J. Stat. Softw., DOI:10.18637/jss.v082.i13. 12 35 J. Fox and S. Weisberg, CAR - An R Companion to Applied Regression, 2019. 13 36 A. van der Ent, G. Echevarria, P. N. Nkrumah and P. D. Erskine, Ann. Bot., 2020, 126, 14 15 1017–1027. 37 16 H. Lambers, P. E. Hayes, E. Laliberté, R. S. Oliveira and B. L. Turner, Trends Plant Sci., 2015, 20, 83–90. 17 38 P. J. White and K. Neugebauer, Plant Soil, 2021, 461, 63–68. 18 39 M. Delgado, S. Valle, P. J. Barra, M. Reyes-Díaz and A. Zúñiga-Feest, Plant Soil, 19 2019, 444, 475–487. 20 40 T. Jaffré, C. R. Acad. Sc. Paris, 1979, 289, 425–428. 21 41 S. Jansen, T. Watanabe and E. Smets, Ann. Bot., 2002, 90, 53–64. 22 42 S. Jansen, T. Watanabe, P. Caris, K. Geuten, F. Lens, N. Pyck and E. Smets, Plant 23 24 25 Biol., 2004, 6, 498–505. 43 E. Magri, E. K. Gugelmin, F. A. P. Grabarski, J. Z. Barbosa, A. C. Auler, I. Wendling, S. A. Prior, A. T. Valduga and A. C. V. Motta, Ecotoxicol. Environ. Saf., 2020, 203, 28 1 111010. 2 44 W. H. A. Hekking, University of Amsterdam, 1988. 3 45 R. R. Brooks and E. D. Wither, J. Geochemical Explor., 1977, 7, 295–300. 4 46 R. R. Brooks, E. D. Wither and B. Zepernick, Plant Soil, 1977, 47, 707–712. 5 47 E. Fernando, M. Quimado and A. Doronila, PhytoKeys, 2014, 37, 1–13. 6 48 G. A. Wahlert, T. Marcussen, J. de Paula-Souza, M. Feng and H. E. Ballard, Syst. Bot., 7 2014, 39, 239–252. 8 49 M. A. Koch and D. A. German, Front. Plant Sci., 2013, 4, 267. 9 50 P. N. Nkrumah, G. Echevarria, P. D. Erskine and A. van der Ent, Sci. Rep., 2018, 8, 10 9659. 11 51 R. D. Reeves, Plant Soil, 2003, 249, 57–65. 12 52 R. D. Reeves, A. van der Ent, G. Echevarria, S. Isnard and A. J. M. Baker, eds. A. van 13 der Ent, A. J. M. Baker, G. Echevarria, M.-O. Simonnot and J. L. Morel, Springer 14 International Publishing, Cham, 2021, pp. 133–154. 15 53 D. M. Navarrete Gutiérrez, A. J. Pollard, A. van der Ent, M. Cathelineau, M.-N. Pons, 16 J. A. Cuevas Sánchez and G. Echevarria, Chemoecology, DOI:10.1007/s00049-021- 17 00338-4. 18 54 19 S. Lopez, A. van der Ent, P. D. Erskine, G. Echevarria, J. L. Morel, G. Lee, E. Permana and E. Benizri, Plant Soil, 2019, 436, 543–563. 20 55 D. A. Wilcox and D. A. Cowan, Plant Soil, 2016, 406, 375–388. 21 56 J. Alvarez-Manjarrez, R. Garibay-Orijel and M. E. Smith, Mycorrhiza, 2018, 28, 103– 22 23 115. 57 24 25 J. Proctor, C. Phillipps, G. K. Duff, A. Heaney and F. M. Robertson, J. Ecol., 1989, 77, 317. 58 A. van der Ent and D. Mulligan, J. Chem. Ecol., 2015, 41, 396–408. 29 1 59 S. S. Lee, Ambio, 1990, 19, 383–385. 2 60 J. Kuijt and B. Hansen, in Flowering Plants. Eudicots, eds. J. Kuijt and B. Hansen 3 (deceased), Springer International Publishing, Cham, 2015, pp. 167–168. 4 61 R. S. Boyd, S. N. Martens and M. A. Davis, Madroño, 1999, 46, 92–99. 5 62 A. Bani, D. Pavlova, E. Benizri, S. Shallari, L. Miho, M. Meco, E. Shahu, R. Reeves 6 7 and G. Echevarria, Ecol. Res., 2018, 33, 549–559. 63 8 9 816. 64 10 11 65 66 67 Y. Pillon, D. Petit, C. Gady, M. Soubrand, E. Joussein and G. Saladin, Plant Soil, 2019, 434, 481–489. 68 18 19 Y. Pillon, H. C. F. Hopkins, F. Rigault, T. Jaffré and E. A. Stacy, New Phytol., 2014, 202, 521–530. 16 17 K. Neugebauer, M. R. Broadley, H. A. El-Serehy, T. S. George, J. W. McNicol, M. F. Moraes and P. J. White, Physiol. Plant., 2018, 163, 306–322. 14 15 M. R. Broadley, N. J. Willey, J. C. Wilkins, A. J. M. Baker, A. Mead and P. J. White, New Phytol., 2001, 152, 9–27. 12 13 P. G. Dimitrakopoulos, M. Aloupi, G. Tetradis and G. C. Adamidis, Plants, 2021, 10, K. Neugebauer, H. A. El‐Serehy, T. S. George, J. W. McNicol, M. F. Moraes, M. C. M. Sorreano and P. J. White, Physiol. Plant., 2020, 168, 790–802. 69 20 Y. Pillon, D. A. González, H. Randriambanona, P. P. Lowry, T. Jaffré and S. Merlot, J. Biogeogr., 2019, 46, 2457–2465. 21 70 Y. Pillon, J. Munzinger, H. Amir and M. Lebrun, J. Ecol., 2010, 98, 1108–1116. 22 71 P. Morat, T. Jaffré, F. Tronchet, J. Munzinger, Y. Pillon, J.-M. Veillon, M. Chalopin, 23 P. Birnbaum, F. Rigault, G. Dagostini, J. Tinel and P. P. Lowry, Adansonia, 2012, 34, 24 179–221. 25 72 A. T. Schettini, M. G. P. Leite, M. C. T. B. Messias, A. Gauthier, H. Li and A. R. 30 1 2 Kozovits, Flora, 2018, 238, 175–182. 73 3 4 C. Bihanic, K. Richards, T. K. Olszewski and C. Grison, ChemCatChem, 2020, 12, 1529–1545. 74 V. S. García de la Torre, C. Majorel‐Loulergue, G. J. Rigaill, D. Alfonso‐González, L. 5 Soubigou‐Taconnat, Y. Pillon, L. Barreau, S. Thomine, B. Fogliani, V. Burtet‐ 6 Sarramegna and S. Merlot, New Phytol., 2021, 229, 994–1006. 7 75 8 9 S. K. Meier, N. Adams, M. Wolf, K. Balkwill, A. M. Muasya, C. A. Gehring, J. M. Bishop and R. A. Ingle, Plant J., 2018, 95, 1023–1038. 76 P. Halimaa, Y.-F. Lin, V. H. Ahonen, D. Blande, S. Clemens, A. Gyenesei, E. Häikiö, 10 S. O. Kärenlampi, A. Laiho, M. G. M. Aarts, J.-P. Pursiheimo, H. Schat, H. Schmidt, 11 M. H. Tuomainen and A. I. Tervahauta, Environ. Sci. Technol., 2014, 48, 3344–3353. 12 13 31 1 Supplementary data legend 2 3 Fig. S1. MP-AES calibration of the XRF data. The concentration of Mn (A) and Ni (B) was 4 measured by XRF and MP-AES in a collection of 124 dry leaves samples from species 5 accumulating various amounts of metals. Only samples with a metal concentration above the 6 LOD for XRF were used for the calibration. The linear regression modelling the relationship 7 between XRF and MP-AES measurements is represented on the graphs. The coefficient of 8 determination (r2) is used to measure the goodness-of-fit of the model. 9 10 Fig. S2. Extent of the XRF analysis in the Neotropical region. The number of families (A), 11 genera (B) and species (C) analyzed per country is represented by the size of colored circles. 12 The countries represented by the most important number of specimens are highlighted. 13 32 MEX CUB n = 310 n = 4282 COL n = 482 GUF n = 1181 PER n = 397 CHL n = 187 Fig. 1 BRA n = 2155 Fig. 2 Metal concentration (Log10 ppm) 5 4 3 2 1 Cu Mn all Zn Ni Ni >LOD *** *** olia cea n= e 68 *** Aqu if *** Rub ia n = ceae 232 5 5 anth ace n = ae 832 Mel asto mat ace n = ae 216 The acea n= e 34 4 Phy ll 3 Prot eace n = ae 247 Och nace a n= e 76 Clus iace n = ae 124 Viol ace n = ae 539 Sali cace n = ae 777 2 1 Fig. 3 Mn concentration (Log10 ppm) Fig. 4 Mn concentration (Log10 ppm) Caly c ogo niu n= m 33 Mi c on n = ia 67 n= Il e x 68 Phy llan th n = us 584 Psyc ho n = tria 200 5 5 Pim en n = ta 27 4 Mor ell n= a 17 3 Lom ati n= a 29 2 1 Rou p n = ala 122 Fig. 5 Zn concentration (Log10 ppm) 5 4 3 ngis tipu lata n= 6 R. m ultiv enos n= a 20 R. p anic ulat a n= 6 2 1 R. lo Viol acea e n= 539 Rino n = rea 179 Fig. 6 Ni concentration (Log10 ppm) tran the n= s 75 Pim en n = ta 27 Allo phyl u n= s 14 Dap h nop s n = is 16 Coc colo b n= a 54 Psyc ho n = tria 200 5 Lun din n = ia 12 5 Caly p 4 llan th n = us 584 3 Phy 2 1 Pile n= a 56 B Mn MP-AES (Log10 ug.g-1) n = 73 y = 0.2793x r2 = 0.9469 Mn XRF (Log10 ppm) Fig. S1 Ni MP-AES (Log10 ug.g-1) A n = 45 y = 0.6762x r2 = 0.9354 Ni XRF (Log10 ppm) A MEX CUB n = 19 n = 149 B MEX n = 55 GUF n = 85 GUF n = 26 COL n = 26 PER n = 28 CHL n=8 COL n = 73 PER n = 78 BRA n = 55 C MEX CUB n = 165 n = 2624 GUF n = 418 COL n = 286 PER n = 272 CHL n = 68 Fig. S2 CUB n = 754 BRA n = 1028 CHL n = 15 BRA n = 170