A systematic assessment of the occurrence of trace element hyperaccumulation in the flora of New Caledonia

Vidiro Gei1, Sandrine Isnard2,3, Peter D. Erskine1, Guillaume Echevarria1,4, Bruno Fogliani5, Tanguy Jaffré2,3, Antony van der Ent1,4*

1Centre for Mined Land Rehabilitation, Sustainable Minerals Institute, The University of Queensland, St Lucia, QLD 4072, Australia

2botAnique et Modelisation de l’Architecture des Plantes et des végétation (AMAP), Université Montpellier, IRD, CIRAD, CNRS, INRA, Montpellier, France

3botAnique et Modelisation de l’Architecture des Plantes et des végétation (AMAP), IRD, Herbier de Nouvelle-Calédonie, Nouméa, New Caledonia

4Laboratoire Sols et Environnement, Université de Lorraine – INRAE, F54000 Nancy, France

5Équipe ARBOREAL (AgricultuRe BiOdiveRsité Et vALorisation), Institut Agronomique néo-Calédonien (IAC), 98890 Païta, New Caledonia

*Corresponding author. E-mail: [email protected]

ABSTRACT New Caledonia is a global biodiversity hotspot known for its metal hyperaccumulator plants. X-ray fluorescence technology (XRF) has enabled non-destructive and quantitative determination of elemental concentrations in herbarium specimens from the ultramafic flora of the island. Specimens belonging to six major hyperaccumulator families (Cunoniaceae, Phyllanthaceae, Salicaceae, Sapotaceae, Oncothecaceae and Violaceae) and one to four specimens per species of the remaining ultramafic taxa in the herbarium were measured. XRF scanning included a total of c. 11 200 specimens from 35 orders, 96 families, 281 genera and 1484 species (1620 taxa) and covered 88.5% of the ultramafic flora. The study revealed the existence of 99 nickel hyperaccumulator taxa (65 known previously), 74 manganese hyperaccumulator taxa (11 known previously), eight cobalt hyperaccumulator taxa (two known previously) and four zinc hyperaccumulator taxa (none known previously). These results offer new insights into the phylogenetic diversity of hyperaccumulators in New Caledonia. The greatest diversity of nickel hyperaccumulators occur in a few major clades (Malphigiales and Oxalidales) and families (Phyllanthaceae, Salicaceae, Cunoniaceae). In contrast, manganese hyperaccumulation is phylogenetically scattered in the New Caledonian flora.

KEYWORDS: cobalt – Cunoniaceae – manganese – nickel – Oncothecaceae – Phyllanthaceae – Salicaceae – Sapotaceae – ultramafic – Violaceae – zinc.

INTRODUCTION

Ultramafic soils and hyperaccumulator plants Ultramafic soils derive from mantle rock consisting largely of magnesium-iron-silicate minerals. These soils have relatively high concentrations of the trace elements nickel (Ni), cobalt (Co) and chromium (Cr), but at the same time have cation imbalances as a result of high magnesium (Mg) and low calcium (Ca) (Proctor, 2003; Echevarria, 2018). Ultramafic outcrops contribute disproportionately to regional plant diversity throughout the globe and especially in biodiversity hotspots including California, Cuba, Borneo and New Caledonia (Jaffré, 1992; Borhidi, 1992; Harrison et al., 2006; van der Ent, Erskine & Sumail, 2015a; van der Ent et al., 2015b, d, 2016; Isnard et al., 2016; Galey et al., 2017). The chemical and physical adversity of ultramafic soils is associated with major transformations and adaptations of some plants, such as accumulation of trace elements at extremely high concentrations in their leaves known as ‘hyperaccumulation’ (Brooks et al., 1977; van der Ent et al., 2012; Reeves, van der Ent & Baker, 2018b). Although ultramafic vegetation is often highly species-rich, plants that hyperaccumulate trace elements are comparatively rare. Current estimates suggest that hyperaccumulation occurs in < 0.2% of angiosperms (Baker & Brooks, 1989; Reeves, 2003) and in 1–2% of the ultramafic flora (van der Ent et al., 2015c). Worldwide there are at least 523 Ni, 53 copper (Cu), 42 Co, 20 zinc (Zn) and 42 manganese (Mn) hyperaccumulators known to date (Baker & Brooks, 1989; Reeves, 1992; 2003; van der Ent et al., 2012; Reeves et al., 2018a; Reeves, van der Ent & Baker, 2018b). Most known hyperaccumulator plants hyperaccumulate Ni; however, the disparity in the number of known Ni hyperaccumulator taxa in comparison to the numbers of hyperaccumulators that accumulate other elements is due to the availability of an initial field-testing method using dimethylglyoxime (DMG)-treated paper (Gambi, 1967; Gei et al., 2018) and because of the presence of ultramafic soils worldwide which are enriched in Ni (Brooks, 1987; Echevarria, 2018).

A small portion of plants growing on ultramafic soil hyperaccumulate trace elements; however, most plants are metal excluders, i.e., they maintain a low and relatively constant concentration of trace elements in their tissues when compared to concentrations in the soil (Baker, 1981). Hyperaccumulation involves high metal concentration in plant tissues especially in the leaves (van der Ent et al., 2012), implying extremely high physiological tolerance to specific elements. As such, plant species growing on ultramafic soils show a gradient in physiological responses than can lead to the accumulation of variable concentrations of metal in their tissues. Threshold values for hyperaccumulation have been established as an order of magnitude higher than the “normal” concentration in plants (Jaffré & Schmid, 1974; Brooks et al., 1977). In a more recent review, van der Ent et al. (2012) used the criteria of: (i) two to three orders of magnitude higher than plants on normal soils, and (ii) one order of magnitude higher than in typical plants on metalliferous soils. This concept was applied to all elements, not just Ni. For the main metals that have been studied in plants, the hyperaccumulator thresholds are: 300 µg g-1 for Co, Cu and Cr, 1000 µg g-1 for Ni, 3000 µg g-1 for Zn and 10 mg g-1 for Mn (Reeves, 2003; Reeves et al., 2018a; van der Ent et al., 2012) (Table 1). Plants that accumulate >10 000 µg g-1 of Ni have been further referred to as ‘hypernickelophores’ (Jaffré & Schmid, 1974). The phylogenetic distribution of metal hyperaccumulation indicates repeated evolution of the trait during plant evolution (Krämer, 2010; Jaffré et al., 2013; Cappa & Pilon-Smits, 2014). Such repeated independent evolution is suggestive of an adaptive benefit of hyperaccumulation, although selective advantages are probably diverse and depend on the metal, species physiology, phylogeny and ecology (Boyd, 2004; Reeves et al., 2018a). In New Caledonia, hyperaccumulation is particularly common in a few clades, which are over- represented in the native flora (Pillon et al., 2010; Jaffré et al., 2013). As such, the disharmony of the New Caledonian flora, as observed in many island floras, has been suggested to be strongly driven by preadaptation to ultramafic soils (Pillon et al., 2010). As hyperaccumulation is a complex trait, a better understanding of the

evolution of this phenomenon requires a broad knowledge on the phylogenetic diversity of hyperaccumulator plants. the Ultramafic flora of New Caledonia and hyperaccumulator plants New Caledonia is an archipelago located in the south- western Pacific and is a renowned global biodiversity hotspot (Myers et al., 2000; Mittermeier et al., 2004). The total land area of the island is approximately 19 000 km2 and it harbours an exceptional flora of c. 3300 vascular species with c. 75% endemism (Morat et al., 2012; Pillon, Barrabé & Buerki, 2017). In New Caledonia, ultramafic outcrops cover about one-third of the main island (Grande Terre), including the islands of Belep and des Pins (totalling c. 5600 km2) (Pelletier, 2006). New Caledonia is recognized as a global hotspot of Ni hyperaccumulator plants with 65 Ni and 11 Mn hyperaccumulator plant species currently recorded (Jaffré et al., 2013; Losfeld et al., 2015a). Approximately 70% of the currently known Ni hyperaccumulators in New Caledonia were discovered between 1974 and 1980. In 1976, Pycnandra acuminata (Pierre ex Baill.) Swenson & Munzinger [formerly Sebertia acuminata (Pierre ex Baill.) Engl.] was discovered to have the highest ever recorded concentration of Ni in a living organism (257.4 mg g-1 or 25.74 percent by weight (wt%) in the latex) (Jaffré et al., 1976, 2018). Considerable interest was generated in New Caledonia to discover more hyperaccumulators, most of which were discovered in 1979 in Cunoniaceae, Phyllanthaceae and Salicaceae (Jaffré, Brooks & Trow, 1979a; Jaffré et al., 1979b; Kersten et al., 1979) and in 1980 from Argophyllaceae, Celastraceae, Cunoniaceae, Phyllanthaceae and Violaceae (Jaffré, 1980). The remainder of the recorded Ni hyperaccumulators were discovered in 2007 and 2013 (Amir et al., 2007; Jaffré et al., 2013). Apart from the Ni hyperaccumulators, New Caledonia also hosts 11 Mn hyperaccumulators; the first discovered in 1977 was Denhamia fournieri (Pancher & Sebert) M.P.Simmons subsp. fournieri [formerly Maytenus bureaviana (Loes.) Loes.] in Celastraceae and Alyxia poyaensis (Boiteau) D.J.Middleton (formerly Alyxia rubricaulis (Baill.) Guillaumin subsp. poyaensis Boiteau) in Apocynaceae (Jaffré, 1977). These were followed by discoveries of three Mn hyperaccumulators in the Proteaceae (Jaffré, 1979), Garcinia amplexicaulis Vieill. ex Pierre in Clusiaceae, Gossia clusioides (Brongn. & Gris) N.Snow var. ploumensis (Däniker) N.Snow comb. ined. in Myrtaceae (Jaffré, 1980) and more recently Polyscias pancheri (Baill.) Harms in Araliaceae, Gossia diversifolia (Brongn. & Gris) N.Snow in Myrtaceae and Grevillea meisneri Montrouz. in Proteaceae (Losfeld et al., 2015a).

Herbarium discoveries of hyperaccumulator plants Global herbaria are the greatest depositories of genetic, taxonomic and biogeographical information on the plant kingdom (Camerini, 1987; Soberón, Llorente & Benitez, 1996; Wen et al., 2015). Many known hyperaccumulator taxa were discovered through investigations of pieces of dried leaf samples taken from herbarium specimens (Reeves, 2003). Most of the known New Caledonian Ni and Mn hyperaccumulator taxa were primarily discovered through destructive sampling of herbarium specimens (e.g., Brooks et al., 1977; Jaffré, Brooks & Trow, 1979a; Kersten et al., 1979) and a few field-collected and dried samples (e.g., Jaffré et al., 1979b). These discoveries involved testing possible Ni hyperaccumulators with DMG-treated paper before further analysis with atomic absorption spectrophotometry (AAS), which is limited to a single element at a time. In most cases, the only element measured was Ni (e.g., Jaffré, Brooks & Trow, 1979a; Jaffré et al., 1979b), although two or three elements (Mn, Co and Ni) were sometimes analysed, as in Phyllanthus L., Homalium Jacq. and Hybanthus Jacq. (Kersten et al., 1979; Jaffré 1980). These analytical techniques have therefore hindered the discovery of hyperaccumulator plant taxa for elements other than Ni. Later studies (e.g., Fernando et al., 2008; Boyd & Jaffré, 2009; Jaffré et al., 2013; Losfeld et al., 2015a) employed inductively coupled plasma-atomic emission spectroscopy (ICP-AES) for multiple elements; however, these studies were narrower in scope. The destructive sampling of herbarium specimens was both time- and resource-consuming and is no longer allowed on a large scale by most herbaria. The advent of portable X-ray fluorescence

technology makes it possible to non-destructively measure the foliar element composition of thousands of samples in a short period of time (Gei et al., 2018; van der Ent et al., 2019). As a global hyperaccumulator hotspot, New Caledonia offers a unique opportunity to assess the occurrence of hyperaccumulation in a flora at both a regional and phylogenetic scale. As many Ni hyperaccumulators are already known, based on classical analytical techniques, the XRF method was tested on six selected families known globally to contain Ni hyperaccumulator plants, followed by a systematic testing of all the species (one to four specimens per taxon) that are known to occur on ultramafic soils. This first complete assessment of hyperaccumulation in a regional flora enables a confident elucidation of the phylogenetic distribution and the diversity of metal hyperaccumulation.

MATERIAL AND METHODS

XRF scanning and sample selection The XRF analysis was undertaken at the Herbarium of New Caledonia (NOU) managed by the Institut de Recherche pour le Développement (IRD). This regional herbarium contains c. 90 000 specimens of vascular plants, covering >95% of the flora. The taxonomy is based on the Checklist (“Florical”) of the indigenous flora of New Caledonia (Morat et al., 2012; Munzinger et al., 2019 [continuously updated]). Only specimens identified at the species or infraspecies level were kept in the dataset. We included infraspecific levels, when available, as close relatives may exhibit different physiology. When infraspecific taxa were available, we nevertheless kept specimens with identification at the species level. This study only considered dicotyledonous plants. Following this method, 11 204 specimens were measured with XRF based on the following selection: all specimens from families known to host hyperaccumulator species in more than one region in the world or because they host many and strong (‘hypernickelophore’) hyperaccumulator species (Cunoniaceae, Oncothecaceae, Phyllanthaceae, Salicaceae, Sapotaceae and Violaceae) were selected. All these specimens were analysed regardless of their ecology and soil preference (if any). This systematic screening was aimed to uncover new hyperaccumulating taxa and included 7257 specimens; all taxa (one to four specimens per taxon) that are reported from ultramafic soils in New Caledonia (Isnard et al., 2016) were also selected, which represented another 3950 specimens. The total sampling included 1620 taxa from the native dicotyledonous flora of New Caledonia. Among the c. 1550 dicotyledonous species known to occur on ultramafic soils (obligate or facultative) in New Caledonia (Isnard et al., 2016), 1372 species (1498 taxa) belonging to 269 genera and 95 families were scanned (difference due to missing specimens: on loan, lacking in the collection or no specimens occurring on ultramafic, for ultramafic facultative taxa), which represented 88.5 % of the total ultramafic flora (92% of the ultramafic genera and 97% of the families).

Each measurement was taken from a specimen attached to standard herbarium cardboard sheet placed on top of a 10 × 10 cm2, 99.99% pure titanium plate to provide a uniform background and to block transmitted X- rays. Full details on the handheld XRF instrument and calibration are provided in Supporting Information (Data S1). The limits of detection (LODs) for Co, Mn, Ni and Zn were estimated by visual inspection of the log- transformed regression models of XRF data against corresponding ICP-AES measurements (Supporting Information, Data S1; Table 1). Two concentration classes for XRF-corrected values were considered: ‘low- range’ and ‘high-range’ hyperaccumulators, the latter have at least two to five times the classical value used to identify hyperaccumulators confidently (i.e., Ni > 5000 µg g-1, Mn > 20 mg g-1, Co > 1000 µg g-1, Zn 1 mg g-1) (Table 1). There were numerous ‘low-range’ hyperaccumulator records for Ni, Mn, Co and Zn; however, to avoid potential ‘false-positives’ the current work focuses principally on ‘high-range’ data. The foliar concentrations refer to the corrected XRF values as detailed in Supporting Information (Data S1) and Table 1.

RESULTS

Focus on six major hyperaccumulator families Six families known to contain a high diversity of hyperaccumulators or at least few strong hyperaccumulators (of Ni for example) were studied in detail. The measurements included all genera in each family present and 92–100% of known species. All families contained mostly Ni and Mn high-range hyperaccumulators, concentrated in a few genera, and a few specimens had high-range Co and Zn foliar concentrations (Fig. 1).

Cunoniaceae Seven genera in this family occur in New Caledonia, including three endemic genera (Codia J.R.Forst. & G.Forst., Hooglandia McPherson & Lowry, Pancheria Brongn. & Gris) (Fig. 1a). Of the 2519 specimens measured, 283 had high-range concentrations (Fig. 1a). No specimens of Cunoniaceae were Zn hyperaccumulators. Only Codia, Geissois Labill. and Pancheria contained Ni hyperaccumulators. Cunonia L., Pancheria and Spiraeanthemum A.Gray contained Mn hyperaccumulators. Pancheria was the only genus that simultaneously hyperaccumulated Ni and Mn. Only Spiraeanthemum and Pancheria had high- range Co specimens (Fig. 1a; Table 4). No specimens of Hooglandia and Weinmannia L. were recorded with high-range values for any element. Species already known as hyperaccumulators of Ni in Geissois [e.g. Geissois bradfordii H.C.Hopkins, Geissois lanceolata (Guillaumin) H.C.Hopkins, Geissois pruinosa Brongn. & Gris] and Pancheria (e.g., Pancheria engleriana Schltr.) were clearly identified as high-range species (Fig. 2a; Supporting Information, Table S2.1). XRF measurements revealed few new hyperaccumulator species in these genera (Geissois trifoliolata Guillaumin, Pancheria calophylla Brongn. & Gris, Pancheria xaragurensis H.C.Hopkins & Pillon) and two species of Codia, which had high-range values (Codia albicans Vieill. ex Pamp., Codia triverticillata H.C.Hopkins & Pillon). Several hyperaccumulators of Ni that had low-range values were also recorded (Fig. 2a; Supporting Information, Table S2.1). Geissois lanceolata and Geissois pruinosa were the only Cunoniaceae with high XRF-concentrations for almost all measured specimens (Table 2). No Mn hyperaccumulators were previously known from this family; however, XRF measurements revealed 15 high- range hyperaccumulators (eight in Pancheria) and 25 low-range hyperaccumulators (Fig. 2b; Supporting Information, Table S2.2), which makes this family an important group in Mn hyperaccumulation. Two Pancheria spp. (P. multijuga Guillaumin and P. reticulata Guillaumin) belong to the top 15 Mn hyperaccumulators, with all measured specimens having high-range values (Table 3).

Phyllanthaceae All the samples from the five genera that are present in New Caledonia were measured; however, only Phyllanthus contained hyperaccumulators (Fig. 1b). Of the 1401 specimens of Phyllanthus measured, 122 specimens had high-range concentrations (Fig. 1b). Phyllanthus combined Ni, Mn, Zn and Co hyperaccumulation, although Zn, Co and Mn hyperaccumulating specimens were few (< 10) (Fig. 1b). The genus contained an exceptional diversity of Ni hyperaccumulators, since c. 40% of measured taxa (75/177) were low-range or high- range hyperaccumulators (Fig. 3a; Supporting Information, Table S2.1). Among these 75 taxa, 29 had specimens with high-range values, most species being already known as Ni hyperaccumulators (Supporting Information, Table S2.1). Sixteen species or varieties constituted new records (e.g., Phyllanthus aeneus Baill. var. longistylis M.Schmid, Phyllanthus fractiflexus M.Schmid, Phyllanthus tixieri M.Schmid) (Supporting Information, Tables S2.1, S2.6) and a few others were low-range Ni hyperaccumulators. In comparison, fewer species were found to be Mn hyperaccumulators; however, all were new records (Fig. 3b; Supporting Information, Table S2.2). Five taxa were recorded within high-range Mn but were sometimes represented by a single specimen calling for further validation. Five taxa of Phyllanthus belonged to the top 15 Ni and one species to top 15 Mn hyperaccumulators (Tables 2-3). All these species had almost all their

specimens measured in the high-range, and 50% of their specimens had >23 mg g-1 Ni (Supporting Information, Table S2.1). Phyllanthus serpentinus S.Moore had the second highest recorded maximum Ni concentration in this study, and all specimens of this species had concentrations > 22 mg g-1 (Table 2). Several species of Phyllanthaceae were found among the few high-range Co and Zn hyperaccumulators (Tables 4-5).

Sapotaceae Thirty-two specimens, out of 2130 specimens measured, had high-range XRF foliar concentrations (Fig. 1c). Of the six genera occurring in New Caledonia, only the endemic genus Pycnandra Benth. had high- range hyperaccumulator specimens for Ni and Mn (Fig. 1c; Tables 3–4; Supporting Information, Tables S2.1, S2.2). Apart from this genus, single specimens of Pichonia Pierre (Ni) and Pleioluma (Baill.) Baehni (Mn) were found to hyperaccumulate (Fig. 1c). Sapotaceae include the famous “blue sap” species Pycnandra acuminata, well known for its strong Ni hyperaccumulation. All the specimens of this species were measured with high- range Ni concentration. Pycnandra caeruleilatex Swenson & Munzinger is another known Ni hyperaccumulator species, more recently described. Two more species were revealed as high-range Ni hyperaccumulators, including the recently described and rare Pycnandra kouakouensis Swenson & Munzinger (Swenson & Munzinger, 2016) and the commoner Pycnandra sessilifolia (Pancher & Sebert) Swenson & Munzinger. With Pycnandra acuminata, Pycnandra kouakouensis was one of the strongest Ni hyperaccumulators with samples consistently having concentrations >19 mg g-1 and > 40 mg g-1, respectively. Specimens of Pichonia daenikeri (Aubrév.) Swenson, Bartish & Munzinger indicated another genus in Sapotaceae as a hyperaccumulator, with one high-range specimen and 50% of the specimens with > 1600 µg g-1. Additionally, Pycnandra, Pichonia and Planchonella Pierre contained several low-range Ni hyperaccumulator species (Fig. 4a). Three genera representing 15 species were found to be low-range Mn hyperaccumulators in Sapotaceae (Fig. 4b; Supporting Information, Table S2.2). High-range Mn hyperaccumulators were represented by only two taxa [Pleioluma sebertii (Pancher) Swenson & Munzinger and Pycnandra decandra (Montrouz.) Vink subsp. coriacea (Baill.) Swenson & Munzinger and a few specimens (one and two, respectively); however, several specimens of both taxa have been recorded with Mn concentrations >10 000 µg g-1 (Supporting Information, Table S2.2). Sapotaceae also included several low- range Zn hyperaccumulators detected in this study (Supporting Information, Table S2.4).

Salicaceae We recorded 66 specimens (out of 834 specimens) with high-range foliar concentrations. Among the four genera present in New Caledonia, only the two genera Homalium and Xylosma contain high- range Ni hyperaccumulators; however, no Mn hyperaccumulators were detected (Fig. 1d). High-range Mn hyperaccumulators were only represented by few specimens in Casearia. No specimens of Lasiochlamys had high-range XRF foliar concentration for any elements (Fig. 1d). Homalium kanaliense (Vieill.) Briq. var. kanaliense was the only Co high-range hyperaccumulator and Casearia silvana Schlechter was the only high- range Zn hyperaccumulator in the family (Fig. 1d; Tables 4-5). Salicaceae have been extensively studied in the past and all Xylosma and Homalium species with high- range Ni hyperaccumulation threshold were already known from the literature (Supporting Information, Table S2.6). Several new low-range hyperaccumulators were however detected in all genera (Fig. 4a; Supporting Information, Table S2.1). Some, such as Casearia silvana, were previously known as marginal hyperaccumulators [8–1490 µg g-1 for 18 specimens in Jaffré et al. (1979b)]. No Mn hyperaccumulators were previously known from this family and this study revealed two Casearia species with high-range Mn values (Fig. 4b; Supporting Information, Table S2.2). Several low-range Mn hyperaccumulators were detected in Homalium and Lasiochlamys. In Salicaceae there were no taxa in the top 15 Ni and Mn hyperaccumulators; however, several species were hypernickelophores (Supporting Information, Table S2.1).

Violaceae One hundred and nine specimens, out of 294 specimens measured, had high- range foliar concentrations. Only the genus Hybanthus contained high-range hyperaccumulators for Ni, Mn, Co and Zn (one specimen for Mn and Zn in Agatea A.Gray) (Fig. 1e; Tables 2, 4-5, Supporting Information, Tables S2.1-S2.4). In Violaceae, Hybanthus austrocaledonicus and Hybanthus caledonicus are well known Ni hyperaccumulators (Brooks, Lee & Jaffré, 1974). Hybanthus austrocaledonicus contained the highest number of specimens with high-range values recorded for both Ni and Mn (Fig. 4a-b), whereas both Hybanthus species were listed in the top 15 Ni hyperaccumulators (Table 2). Hybanthus austrocaledonicus was the strongest Ni hyperaccumulator in this study, and 50% of the specimens had concentrations > 20 mg g-1 (Table 2). Hybanthus austrocaledonicus was also among the top 15 hyperaccumulators for Zn and Co (Tables 4-5). In the genus Agatea, three species were detected as marginal (low-range) Ni hyperaccumulators, two of which were already known as Ni hyperaccumulators (Fig. 4a; Supporting Information, Tables S2.1, S2.6). Agatea rufotomentosa was also a new Mn hyperaccumulator (Fig. 4b; Supporting Information, Table S2.2).

Oncothecaceae Oncothecaceae are a small endemic family with only one genus and two species. Of the two species occurring on ultramafic soils, only Oncotheca balansae was recorded as a Ni hyperaccumulator, with many specimens observed to have high-range values. Of the 45 specimens measured, 25 specimens had high- range foliar concentrations of Ni (Fig. 4a). No other hyperaccumulators were detected in this family assessment of hyperaccumulation in herbarium specimens from other families XRF screening revealed several Ni, Co, Mn and Zn hyperaccumulators (Table 6; Fig. 5a-b). Some Ni and Mn hyperaccumulators were known from previous studies (Jaffré & Schmid, 1974; Jaffré, 1977, 1980; Jaffré, Brooks & Trow, 1979a; Jaffré et al., 1979b; Kersten, 1979; Fernando et al., 2008; Losfeld et al., 2015, etc.) (Supporting Information, Tables S2.2 and S2.7), and new Co and Zn hyperaccumulators were revealed (Supporting Information, Tables S2.3, S2.4).

Hyperaccumulators of Ni Fifty-three Ni low-and/or high-range hyperaccumulators were recorded across 21 families, with 40 taxa falling in the low-range values (Ni < 5000 µg g-1) (Fig. 5a; Supporting Information, Table S2.1). High-range foliar concentrations (≥5000 µg g-1) were found in 13 taxa (Fig. 5a), with nine new records (Supporting Information, Table S2.1). These new records occurred in Argophyllaceae (two species), Celastraceae (one taxon), Ebenaceae (one species), Euphorbiaceae (three taxa), Proteaceae (one species) and Rubiaceae (one species). Overall, the diversity of hyperaccumulators within families was low and, among the four specimens measured per taxon, only a few exhibited high-range values (Fig. 5a). However, some species reached high values for all specimens measured and fell within the top 15 Ni species (Table 2), e.g., the intensively studied Psychotria gabriellae (Baill.) Guillaumin (Rubiaceae) and, unexpectedly, Cleidion velutinum McPherson (Euphorbiaceae) (> 60 mg g-1) and Gynochthodes collina (Schltr.) Razafim. & B.Bremer (Rubiaceae) (> 59 mg g-1) (Supporting Information, Table S2.1). In Rubiaceae, the 17 genera occurring on ultramafic substrates were included for a total of 104 species (107 taxa) over 135 known species from ultramafic substrates; however, only four species (included a potentially new species) were recorded, three within high-range hyperaccumulators. Psychotria comptonii S.Moore was a newly recorded low-range Ni hyperaccumulator (Fig. 5a).

Hyperaccumulators of Mn One hundred and sixteen taxa belonging to 28 families were recorded with low- and/or high-range Mn concentrations with 72 taxa falling in the low- range values (<20 mg g-1) (Fig. 5b). High-range Mn concentrations were recorded in 44 taxa, with 18 taxa reaching concentrations ≥50 mg g-1, of which 37 are

new records (Supporting Information, Tables S2.2 and S2.7). Taxa recorded with high-range concentrations were particularly diverse in Myrtaceae (nine taxa), Proteaceae (seven species) and Apocynaceae (six elements (Ni, Mn and Co). Some combinations appeared more prevalent, including Co-Ni (seven species) and Mn-Ni (four species). A combination with Zn was rare in the families screened and was found only in a single specimen of Hybanthus austrocaledonicus (Ni-Zn). Simultaneous hyperaccumulation of two elements represented the most common syndrome, and none accumulated four elements (Supporting Information, Table S2.5).

Major families of hyperaccumulators in the New Caledonian flora Across the ultramafic flora of New Caledonia, Phyllanthaceae contained the greatest number of Ni (high- range) hyperaccumulators (35% of total taxa). Cunoniaceae and Salicaceae also contained many Ni hyperaccumulator taxa (19% and 18%, respectively); the remaining families each contained < 10% of the total Ni hyperaccumulator taxa (Fig. 6a). The study revealed two new families with Ni hyperaccumulating taxa (Fig. 6a). The pattern is different for Mn hyperaccumulation; Cunoniaceae, a new hyperaccumulator family, contained the majority of hyperaccumulators (21%), followed by Myrtaceae (13%), Proteaceae (10%), Apocynaceae (9%) and Phyllanthaceae (7%). The remaining families each contained <4% of the total number of Mn hyperaccumulator taxa (Fig. 6b). Most of these families were new discoveries for Mn hyperaccumulation. Few Co (six taxa) and Zn (four taxa) hyperaccumulators were recorded. Simultaneous hyperaccumulators were detected in Cunoniaceae, Phyllanthaceae, Proteaceae, Salicaceae, Violaceae, Rubiaceae and Celastraceae; however, none occurred in Myrtaceae (Supporting Information, Table S2.5).

Phylogenetic patterns of hyperaccumulation in the New Caledonian flora Hyperaccumulation (high-range values) is phylogenetically scattered; however, the greatest diversity is concentrated in a few orders as c. 62% of the hyperaccumulators detected occurred in Oxalidales and Malpighiales (rosids), followed by Gentianales (c. 7.0%), Proteales (c. 6.5%) and Ericales (c. 6.0%) (Fig. 7). Only Malpighiales and Proteales contain hyperaccumulators for all elements. A single species of Celastrales, Denhamia fournieri (two varieties), was recorded as a hyperaccumulator of Ni and Mn. Rosids (70%) and asterids (22%) contained 92% of all hyperaccumulators. The distribution of Mn hyperaccumulators was more scattered, compared to Ni, as few Mn hyperaccumulators occurred in the magnoliids, Myrtales or Apiales (Fig. 7). When more marginal hyperaccumulators are considered (low-range values), Oxalidales and Malpighiales still represent 60% of all hyperaccumulators. Ericales contained many low-range hyperaccumulators and held c. 10% of all hyperaccumulators.

DISCUSSION

Herbarium XRF screening reliability The regression corrected XRF readings may over- estimate Ni and Mn concentrations, as shown by the discrepancies between the published foliar elemental concentrations and the present studies for most of the taxa. For example, the 23 specimens of Phyllanthus serpentinus had a Ni concentration ranging from 22 400 to 102 000 µg g-1 in this study, whereas Jaffré (1980) and Jaffré et al. (2013) recorded 3750–31 791 µg g-1 from seven samples (Supporting Information, Table S2.6). Another example is Pycnandra acuminata for which all XRF derived values ranged between 19 800–89 700 µg g-1, whereas other studies reported lower values, between 11 625–25 760 µg g-1 (Lee et al., 1978; Sagner et al., 1998; Schaumlöffel et al., 2003; Perrier et al., 2004; Callahan et al., 2008). Reported values for Ni had not exceeded 63 000 µg g-1 (Jaffré & Schmid, 1974) or 55 000 µg g-1 for Mn (Jaffré, 1979) in the New Caledonian flora sampled across a wide phylogenetic and geographical range. The over- estimation of the XRF values may be explained by sample heterogeneity

to which XRF is more sensitive than bulk methods (i.e., ICP-AES after acid digestion of a leaf fragment). Biological material is inherently heterogeneous in its elemental distribution, and the structure and thickness cannot be controlled. Calibration is a critical issue to reduce the extent of matrix effect on the accuracy of the calibration model (Arantes de Carvalho et al., 2018). We used a matrix- matched standard based on analysing a subset of samples by a validated reference method (ICP-AES); however, XRF measures a small spot (6 mm) and the X-ray fluorescence signal is therefore affected by sample morphology and elemental localization. For example, in Pycnandra acuminata Ni is highly concentrated in the dense laticifer network, where Ni can exceed 25 wt% (i.e., 250 000 µg g-1) locally, which may lead to artefacts in the XRF measurement. Nevertheless, by screening all the specimens available in families known to hold several hyperaccumulator taxa (Cunoniaceae, Phyllanthaceae, Salicaceae) or a few strong hyperaccumulators of Ni (Sapotaceae, Violaceae, Oncothecaceae) (e.g., Jaffré, Brooks & Trow, 1979a; Jaffré et al., 1979b, 2013; Kersten et al., 1979; Jaffré, 1980; Losfeld et al., 2015a), we show that the method is reliable to detect hyperaccumulators.

Other limitations of the method include the use of field preservation methods during collection of herbarium specimens (e.g., soaking of specimens in ethanol for temporary preservation during transport), which probably results in leaching of (certain) elements, and such specimens should be avoided when possible (Gei et al., 2018). Even though this is commonly done in many tropical regions, the relatively short geographical distances in New Caledonia meant that this is not routinely done in the NOU Herbarium. Existing superficial contamination with soil particles cannot be excluded from herbarium specimens; however, it can be effectively gauged from concomitant high Fe and Cr concentrations, and suspect specimens should be treated with caution. This is rarely an issue with hyperaccumulators, however, as in hyperaccumulators the foliar Ni, Co, Mn, or Zn concentrations are generally substantially higher than in the soil in which the plant grew (van der Ent et al., 2018).

New hyperaccumulator records Using high-range values only, to avoid XRF-calibration bias (“false positives”), we found that most previously known species were detected even when only a single specimen was available in the herbarium. The detection of a new record, when based on a single herbarium, will require more specimens to be scanned or reference methods to confirm results. When comparing data with literature, the discrepancies between the published foliar elemental concentrations and XRF measured appear much less important in the low-range measurements (e.g., Jaffré et al., 2013; Supporting Information, Tables S2.6 and S2.7). As such the XRF method is more sensitive to high foliar concentration, for the reason provided earlier. The XRF method may also allow for the detection of strong accumulators and marginal hyperaccumulators, i.e., species in which the range of metal accumulation remains below or unfrequently above the hyperaccumulation threshold. Accordingly, several species already known as marginal hyperaccumulators fell within the low-range or below minimal threshold values [e.g., Casearia silvana Schltr., Agatea longipedicellata (Baker f.) Guillaumin & Thorne, Homalium austrocaledonicum Seem.] (Jaffré et al., 2013; Supporting Information, Tables S2.6 and S2.7). For many other species, for which a single specimen with values around the minimum threshold was found, hyperaccumulation might be facultative in nature sensu Pollard, Reeves & Baker (2014). For instance, no Capparaceae taxa were recorded as Ni hyperaccumulators in our study; however, previous ICP-AES analysis of Capparis artensis Montrouz. revealed it to be a marginal Ni hyperaccumulator from three samples with a range of 321–1249 µg g-1 (Jaffré et al., 2013).

The high-range records found in this study, from screening c. 11 200 leaf samples from dried herbarium specimens, can confidently be considered as hyperaccumulators. In total we discovered 34 Ni and 63 Mn hyperaccumulators, covering two and ten new families, respectively, and we also identified eight for Co and

four for Zn for which no records were available in New Caledonia (Tables 4-5; Supporting Information, Table S2.6 and S2.7). The new families that contained high-range Ni hyperaccumulators include only one species (Kermadecia pronyensis in Proteaceae and Diospyros calciphila F.White in Ebenaceae), which require further validation with ICP-AES. A new count of 99 and 74 hyperaccumulator taxa for Ni and Mn, respectively, has been compiled for New Caledonia with species belonging to 19 and 32 genera, 12 and 22 families, respectively. In addition to this hyperaccumulator list, many low-range hyperaccumulators (including facultative and marginal taxa) were recorded: 150 Ni (1000–5000 µg g-1), 122 Mn (10–20 mg g-1), 34 Co (≥ 1000 µg g-1), and ten Zn (≥ 10 000 µg g-1). This mass screening confirms that New Caledonia is a global hotspot for hyperaccumulator plants, as we can confidently estimate that Ni hyperaccumulation occurs in 3.8% (high-range) or c. 10% (including low-range) of the dicotyledonous flora [2419 dicotyledonous species (Munzinger et al., 2016)]. This is well above the 0.16% of dicots estimated globally for Ni (Borhidi, 2001). These results also indicate that Mn hyperaccumulation, which has been understudied in New Caledonia, is an important phenomenon. Overall, Mn hyperaccumulation occurs in 2.8% (high- range) or c. 7.9% (including low-range) of dicots. In addition to the high incidence of hyperaccumulators, many species have extraordinary levels of accumulation, 57 taxa (including Gynochthodes sp. nov.) being hypernickelophores (Ni > 10 000 µg g-1), of which 16 were previously known (Jaffré et al., 2013). There were no prior records of Zn hyperaccumulators in New Caledonia; however, globally there are 15 taxa, of which nine are in Brassicaceae (Baker & Brooks, 1989). Mass screening has revealed only four high-range Zn hyperaccumulators in New Caledonia. All of these hyperaccumulators belong to Malpighiales (Phyllanthaceae, Violaceae and Salicaceae) (Table 5, Supporting Information, Table S2.4).

Worldwide, 26 taxa from 12 families have been recorded as Co hyperaccumulators; however, this could be an over-representation of numbers of species due to soil contamination (Krämer, 2010; Lange et al., 2017). Previously two Co hyperaccumulators were known from New Caledonia (Homalium kanaliense, Phyllanthus serpentinus) (Jaffré, 1980). Here we record eight high-range Co hyperaccumulators, representing Malphigiales (Phyllanthaceae, Salicaceae and Violaceae), Oxalidales (Cunoniaceae), Gentianales (Rubiaceae) and Proteales (Proteaceae). XRF screening led to the discovery of hyperaccumulators with more than one trace metal (Supporting Information, Table S2.5). Most instances of simultaneous hyperaccumulation occurred in Malpighiales. The majority of simultaneous hyperaccumulators were bi-hyperaccumulators of either Co + Ni or Mn + Ni. Unlike the Ni hyperaccumulators in Malaysia (van der Ent & Mulligan, 2015), there was no significant correlation between Co and Ni (data not shown).

Phylogenetic diversity of hyperaccumulators This study reveals that the high diversity of hyperaccumulators are phylogenetically clustered in a few major clades, especially Oxalidales and Malpighiales, followed by Proteales, Gentianales and Ericales. The phylogenetic clustering of Ni hyperaccumulation, as already observed in the Celastrales, Oxalidales and Malpighiales (COM) clade in New Caledonia (Pillon et al., 2010; Jaffré et al., 2013), is confirmed for Oxalidales and Malpighiales in this study. However, in Celastrales a single species, Denhamia fournieri (two varieties), was recorded as a hyperaccumulator (Ni, Mn and marginally Co). We showed that c. 62% of the hyperaccumulators (79% for Ni and 40% for Mn) belong to Malpighiales and Oxalidales in New Caledonia. Hyperaccumulation of Mn is phylogenetically more dispersed with Proteales, Myrtales (Gossia), Gentianales (Apocynaceae and Rubiaceae) and Apiales (Apiaceae, Pittosporaceae and Araliaceae) each representing c. 10% of hyperaccumulators. In the main orders, there is also a strongly unbalanced distribution of hyperaccumulators in a few major families, especially for Ni. Most of the Ni hyperaccumulators (c. 72%), are concentrated in a small number of families (Phyllanthaceae, Cunoniaceae and Salicaceae) already well known for containing hyperaccumulators in New Caledonia. Phyllanthus (Malpighiales: Phyllanthaceae) is a well-

known Ni hyperaccumulator from other tropical regions, including Cuba and Malaysia (Reeves, 2003; van der Ent, Erskine & Sumail, 2015a). In New Caledonia, the genus has 27 taxa (+ 44 in low-range) recorded as Ni hyperaccumulators [17 already known (Jaffré et al., 2013)], of which 80 species are known to occur on ultramafic substrates. Salicaceae (also Malpighiales) includes many Ni hyperaccumulators and hypernickelophores, in three main genera (Casearia, Homalium and Xylosma) with several species already having been identified (Jaffré et al., 1979b, 2013). Oxalidales are exclusively represented by Cunoniaceae, containing both Ni and Mn hyperaccumulators. Cunoniaceae are well-known for containing Ni hyperaccumulator taxa [11 species, see Jaffré, Brooks & Trow (1979a); Jaffré (1980); Jaffré et al. (2013)], and data presented here found between 15 (high-range) and 47 (including low-range) Ni-hyperaccumulator taxa; of these, several are hypernickelophores. Cunoniaceae were not previously known as Mn hyperaccumulators, and only a few species were recognized to accumulate Mn (Losfeld et al., 2015a); however, they hold the greatest diversity of Mn hyperaccumulators known globally, as 15 species (+ 13 in low-range) were detected, all of them new records.

Sapotaceae (Ericales) are another important Ni hyperaccumulator family, in which the genus Pycnandra is placed (APG IV, 2016). Pycnandra is endemic to New Caledonia, and recent discoveries of new species make it the largest endemic genus with 59 species and 62 taxa (Swenson & Munzinger, 2016). At the time of screening, 44 Pycnandra taxa were available. Three species of Pycnandra (Pycnandra acuminata, Pycnandra caeruleilatex and Pycnandra kouakouensis) have characteristic blue-green latex (Swenson & Munzinger 2010a, b, 2016) and were detected as hypernickelophores. Note that Pycnandra acuminata and Pycnandra kouakouensis belong to Pycnandra subgenus Sebertia (Pierre ex Engl.) Swenson & Munzinger, whereas Pycnandra caeruleilatex belongs to Pycnandra subgenus Trouettia (Pierre ex Baill.) Swenson & Munzinger, suggesting at least two independent origins of Ni hyperaccumulation in the genus. Our study revealed another high-range hyperaccumulator, Pycnandra sessilifolia (subgenus Trouettia), and five low-range hyperaccumulators.

On a global scale, Brassicaceae have one of the highest incidences of Ni hyperaccumulation, notably in Odontarrhena C.A.Mey. [synonym Alyssum section Odontarrhena (C.A.Mey.) Hooker] and Noccaea Moench (Reeves et al., 2018a). This family, represented by only six species (four genera) in New Caledonia, does not occur on ultramafic substrates. The only hyperaccumulator in Brassicales in New Caledonia, Capparis artensis, is a facultative hyperaccumulator. In Asterales, another major Ni hyperaccumulator clade worldwide (essentially in Asteraceae), only one hyperaccumulator (tentatively identified as Argophyllum ellipticum Labill.) was known from New Caledonia (Jaffré et al., 2013). Our study shows that Asterales actually contain several high-range and marginal Ni hyperaccumulators, essentially in Argophyllum J.R.Forst. & G.Forst. Manganese hyperaccumulators were detected in Alseuosmiaceae (Platyspermation crassifolium Guillaumin) and Phellinaceae (Phelline confertifolia Baill., P. lucida Vieill. ex Baill.). These three families (Argophyllaceae, Alseuosmiaceae and Phellinaceae), forming a clade with an Australasian distribution (Kårehed et al., 1999), are ultramafic specialists, all endemic, with low infrageneric diversity. This phylogenetic distribution and the similar Australasian distribution of these lineages suggest that hyperaccumulation could have been more widespread millions of years ago in this region. Western Australia and Queensland, despite extensive ultramafic outcrops, have remarkably few hyperaccumulators, while Tasmania and New Zealand have none, despite wide areas with ultramafic soils (van der Ent et al., 2015b). Why hyperaccumulation is so widespread in New Caledonia compared to Australia or New Zealand remains an open question. Several factors might account for the high numbers of hyperaccumulators in New Caledonia, such as the diversity in ultramafic soils (cambisol, colluvial ferralsols, eroded ferralsols) in which metals (Co, Ni, Mn) are often highly available to plants (Isnard et al., 2016). The geological history of New Caledonia and

the late Eocene subduction/collision, resulting in the emplacement of a large ophiolitic nappe in the Latest Eocene (38–34 Mya), implies a long time of exposure of large areas of ultramafic soils that could have triggered the evolution of hyperaccumulation. In Cuba, the highest diversity of hyperaccumulator plant species occurs on the oldest ultramafic outcrops (Reeves, 1999). Metal hyperaccumulators are almost exclusively woody plants in New Caledonia and, as observed globally (Borhidi, 2001), climbers are rare among hyperaccumulators with a few high-range Ni (only Gynochthodes collina) or Mn (four and one species of Alyxia and Agatea, respectively) records. When considering the flora of New Caledonia, it appears that much of the diversity of hyperaccumulators occurs in major orders, Oxalidales and Malpighiales, which account for c. 24% of the dicotyledonous flora, followed by Gentianales (c. 17%), Ericales (c. 14%) and Apiales (c. 7%) (Morat et al., 2012). These clades were recognized as over-represented in the flora of New Caledonia possibly as a result of a physiological preadaptation to ultramafic soils (Jaffré et al., 1987; Pillon et al., 2010). In New Caledonia the long-time frame of exposure to ultramafic soils could have offered time for the evolution of hyperaccumulation in these clades (Isnard et al., 2016). However, the fact that some over-represented lineages in Sapindales exhibit a high diversity of species on ultramafic soils (148 species over 194 in total), but only a few marginal hyperaccumulators, suggests that physiological adaptations for tolerating edaphic stresses imposed by ultramafic soils and hyperaccumulation do not coincide. We confirm that several families that are mostly ultramafic obligates (Isnard et al., 2016) do not contain any hyperaccumulators of any element (Ni, Mn, Co, Zn) in New Caledonia (e.g., Dilleniaceae and Myodocarpaceae), perhaps because these species are specialized for nutrient-poor soils. A few Ericaceae do contain hyperaccumulating taxa, and these include Dracophyllum verticillatum Labill. (Mn) and other low-range hyperaccumulator species; however, their diversity is low.

CONCLUSIONS This study demonstrates the way in which massive herbarium XRF scanning can be used to obtain elemental concentrations from existing plant material collections held in herbaria globally. This rapid and non- destructive method gains access to this untapped resource of information and is anticipated to be transformative in the discovery of hyperaccumulator plants at regional scales (van der Ent et al., 2019). The approach also enables a new way to elucidate evolution, phylogenetic diversity and ecology of hyperaccumulation in tropical floras by systematically assessing all the taxa from a given region. The discovery of hyperaccumulator plants is also the starting point for detailed ecophysiological investigations using microanalytical techniques such as synchrotron X-ray fluorescence microscopy and proton induced X- ray emission spectroscopy (van der Ent et al., 2017). The physics of X-ray fluorescence are well-understood, and standardized measurement conditions mean that the accuracy of obtained concentration data may be further optimized in the future using the raw data (i.e., by refitting and processing of the XRF spectrum collected from each specimen). Ultimately, timely identification of hyperaccumulators will assist in the development of management strategies for conservation of these rare taxa (Whiting et al., 2004; Erskine, van der Ent & Fletcher, 2012).

ACKNOWLEDGEMENTS V.G. was the recipient of an Australia Awards PhD Scholarship from the Australian Government. The French National Research Agency through the Agence Nationale de la Recherche (ANR-10-LABX-21, LABEX RESSOURCES21, ANR-14-CE04-0005 Project “Agromine”) and through the ANR-14-CE04-0005 Project “Agromine” is acknowledged for funding support. A.v.d.E. was the recipient of a Discovery Early Career Researcher Award (DE160100429) from the Australian Research Council. We thank Irene Nigote for help with the XRF measurements, and Jacqueline Fambart for technical support in the herbarium (NOU). We also thank Philippe Birnbaum for matching all barcode entries in the XRF to the herbarium plant database.

REFERENCES Amir H, Perrier N, Rigault F, Jaffré T. 2007. Relationships between Ni hyperaccumulation and mycorrhizal status of different endemic plant species from New Caledonian ultramafic soils. Plant and Soil 293: 23–35. APG IV. 2016. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG IV. Botanical Journal of the Linnean Society 181: 1–20. Arantes de Carvalho GG, Bueno Guerra MB, Adame A, Nomura CS, Oliveira PV, Pereira de Carvalho HW, Santos D, Nunes LC, Krug FJ. 2018. Recent advances in LIBS and XRF for the analysis of plants. Journal of Analytical Atomic Spectrometry 33: 919–944. Baker AJM. 1981. Accumulators and excluders–strategies in the response of plants to heavy metals. Journal of Plant Nutrition 3: 643–654. Baker AJM, Brooks RR. 1989. Terrestrial higher plants which hyperaccumulate metallic elements – a review of their distribution, ecology, and phytochemistry. Biorecovery 1: 81–126. Borhidi A. 1992. The serpentine flora and vegetation of Cuba. In: Baker AJM, Proctor J, Reeves RD, eds. The vegetation of ultramafic (serpentine) soils. Andover: Intercept Ltd., 83–96. Borhidi A. 2001. Phylogenetic trends in Ni-accumulating plants. South African Journal of Science 97: 544– 547. Boyd RS. 2004. Ecology of metal hyperaccumulation. New Phytologist 162: 563–567. Boyd RS, Jaffré T. 2009. Elemental concentrations of eleven New Caledonian plant species from serpentine soils: elemental correlations and leaf-age effects. Northeastern Naturalist 16: 93–110. Brooks RR. 1987. The serpentine factor. In: Brooks RR, ed. Serpentine and its vegetation: a multidisciplinary approach. Portland: Dioscorides Press, 454 pp. Brooks RR, Lee J, Jaffré T. 1974. Some New Zealand and New Caledonian plant accumulators of nickel. Journal of Applied Ecology 62: 493–499. Brooks RR, Lee J, Reeves RD, Jaffré T. 1977. Detection of nickeliferous rocks by analysis of herbarium specimens of indicator plants. Journal of Geochemical Exploration 7: 49–57. Callahan DL, Roessner U, Dumontet V, Perrier N, Wedd AG, O’Hair RAJ, Baker AJM, Kolev SD. 2008. LC– MS and GC–MS metabolite profiling of nickel(II) complexes in the latex of the nickel-hyperaccumulating tree Sebertia acuminata and identification of methylated aldaric acid as a new nickel(II) ligand. Phytochemistry 69: 240–251. Camerini JR. 1987. Darwin, Wallace, and maps. Ph.D. Thesis, University of Wisconsin. Madison, Wisconsin: Madison ProQuest Dissertations Publishing. Cappa JJ, Pilon-Smits EAH. 2014. Evolutionary aspects of elemental hyperaccumulation. Planta 239: 267– 275. Echevarria G. 2018. Genesis and behaviour of ultramafic soils and consequences for nickel biogeochemistry. In: Van der Ent A, Echevarria G, Baker AJM, Morel JL, eds. Agromining: extracting unconventional resources from plants. Cham: Springer International Publishing, 135–156. Erskine PD, van der Ent A, Fletcher A. 2012. Sustaining metal-loving plants in mining regions. Science 337:

1172–1173. Fernando DR, Woodrow I, Jaffré T, Dumontet V, Marshall AT, Baker AJM. 2008. Foliar manganese accumulation by Maytenus founieri (Celastraceae) in its native New Caledonian habitats: populational variation and localization by x‐ray microanalysis. New Phytologist 177: 178–185. Galey MC, van der Ent A, Iqbal MCM, Rajakaruna N. 2017. Ultramafic geoecology of South and Southeast Asia. Botanical Studies 58: 18. Gambi OV. 1967. Primi dati sulla localizzazione istologica del nichel in Alyssum bertolonii Desv. Giornale Botanico Italiano 101: 59–60. Gei V, Erskine PD, Harris HH, Echevarria G, Mesjasz- Przybylowicz J, Barnabas AD, Przybylowicz WJ, Kopittke PM, van der Ent A. 2018. Tools for the discovery of hyperaccumulator plant species and understanding their ecophysiology. In: van der Ent A, Echevarria G, Baker AJM, Morel JL, eds. Agromining: farming for metals: extracting unconventional resources using plants. Cham: Springer International Publishing, 189–204. Harrison S, Safford HD, Grace JB, Viers JH, Davies K. 2006. Regional and local species richness in an insular environment: serpentine plants in California. Ecological Monographs 76: 41–56. Isnard S, L’Huillier L, Rigault F, Jaffré T. 2016. How did the ultramafic soils shape the flora of the New Caledonian hotspot? Plant and Soil 403: 53–76. Jaffré T, Schmid M. 1974. Accumulation du nickel par une Rubiacée de Nouvelle Calédonie, Psychotria douarrei (G.Beauvisage) Däniker. Comptes Rendus de l’Académie des Sciences Série D - Sciences Naturelles 278: 1727–1730. Jaffré T, Brooks RR, Lee J, Reeves RD. 1976. Sebertia acuminata: a hyperaccumulator of nickel from New Caledonia. Science 193: 579–580. Jaffré T. 1977. Accumulation du manganèse par des espèces associées aux terrains ultrabasiques de Nouvelle- Calédonie. Comptes Rendus de l’Académie des Sciences Série D - Sciences Naturelles 284: 1573–1575. Jaffré T. 1979. Accumulation du manganèse par les Protéacées de Nouvelle Calédonie. Comptes Rendus de l’Académie des Sciences Série D - Sciences Naturelles 289: 425–428. Jaffré T, Brooks RR, Trow JM. 1979a. Hyperaccumulation of nickel by Geissois species. Plant and Soil 51: 157–161. Jaffré T, Kersten W, Brooks RR, Reeves RD. 1979b. Nickel uptake by Flacourtiaceae of New Caledonia. Proceedings of the Royal Society B: Biological Sciences 205: 385–394. Jaffré T. 1980. Étude écologique du peuplement végétal des sols dérivés de roches ultrabasiques en Nouvelle Calédonie. Paris: ORSTOM, 273 pp. Jaffré T, Morat PH, Veillon JM, MacKee HS. 1987. Changements dans la végétation de la Nouvelle-Calédonie au cours du Tertiaire: la végétation et la flore des roches ultrabasiques. Adansonia 4: 365–391. Jaffré T. 1992. Floristic and ecological diversity of the vegetation on ultramafic rocks in New Caledonia. In: Baker AJM, Proctor J, Reeves RD. The vegetation of ultramafic (serpentine) soils: Proceedings of the first international conference on serpentine ecology. Andover: Intercept Ltd., 101–107. Jaffré T, Pillon Y, Thomine S, Merlot S. 2013. The metal hyperaccumulators from New Caledonia can broaden our understanding of nickel accumulation in plants. Frontiers in Plant Science 4: 279. Jaffré T, Reeves RD, Baker AJM, van der Ent A. 2018. The discovery of nickel hyperaccumulation in the New Caledonian tree Pycnandra acuminata: 40 years on. New Phytologist 218: 397–400. Karehed J, Lundberg J, Bremer B, Bremer K. 1999. Evolution of the Australasian families Alseuosmiaceae, Argophyllaceae, and Phellinaceae. Systematic Botany 24: 660–682. Kersten W, Brooks RR, Reeves RD, Jaffré T. 1979. Nickel uptake by New Caledonian species of Phyllanthus. Taxon 28: 529–534.

Krämer U. 2010. Metal hyperaccumulation in plants. Annual Review of Plant Biology 61: 517–534. Lange B, van der Ent A, Baker AJM, Echevarria G, Mahy G, Malaisse F, Meerts P, Pourret O, Verbruggen N, Faucon MP. 2017. Copper and cobalt hyperaccumulation in plants: a critical assessment of the current state of knowledge. New Phytologist 213: 537–555. Lee J, Reeves RD, Brooks RR, Jaffre T. 1978. The relation between nickel and citric acid in some nickel- accumulating plants. Phytochemistry 17: 1033–1035. Losfeld G, L’Huillier L, Fogliani B, McCoy SG, Grison C, Jaffré T. 2015a. Leaf-age and soil-plant relationships: key factors for reporting trace-elements hyperaccumulation by plants and design applications. Environmental Science and Pollution Research 22: 5620–5632. Losfeld G, Mathieu R, L’Huillier L, Fogliani B, Jaffré T, Grison C. 2015b. Phytoextraction from mine spoils: insights from New Caledonia. Environmental Science and Pollution Research 22: 5608–5619. Mittermeier RA , Robles Gil P , Hoffmann M , Pilgrim J, Brooks T, Mittermeier CG, Lamoreux J, da Fonseca GAB. 2004. Hotspots revisited: Earth’s biologically richest and most endangered terrestrial ecoregions. Mexico City: CEMEX. Morat P, Jaffré T, Tronchet F, Munzinger J, Pillon Y, Veillon JM, Chalopin M, Birnbaum P, Rigault F, Dagostini G, Tinel J, Lowry PP. 2012. The taxonomic reference base FLORICAL and characteristics of the native vascular flora of New Caledonia. Adansonia 34: 179–221. Munzinger J, Morat P, Jaffré T, Gâteblé G, Pillon Y, Rouhan G, Bruy D, Veillon JM, Chalopin M. 2019. [continuously updated]. FLORICAL: checklist of the vascular indigenous flora of New Caledonia. Available at: http://publish.plantnet-project.org/project/florical. Myers N , Mittermeier RA , Mittermeier CG , da Fonseca GAB, Kent J. 2000. Biodiversity hotspots for conservation priorities. Nature 403: 853–858. Pelletier B. 2006. Geology of the New Caledonia region and its implications for the study of the New Caledonian biodiversity. In: Payri C, Richer de Forges B, eds. Compendium of marine species from New Caledonia. Nouméa: Centre IRD de Nouméa, 17–30. Perrier N, Colin F, Jaffré T, Ambrosi J-P, Rose J, Bottero J-Y. 2004. Nickel speciation in Sebertia acuminata, a plant growing on a lateritic soil of New Caledonia. Comptes Rendus Geoscience 336: 567–577. Pillon Y, Barrabé L, Buerki S. 2017. How many genera of vascular plants are endemic to New Caledonia? A critical review based on phylogenetic evidence. Botanical Journal of the Linnean Society 183: 177–188. Pillon Y, Munzinger J, Amir H, Lebrun M. 2010. Ultramafic soils and species sorting in the flora of New Caledonia. Journal of Ecology 98: 1108–1116. Pollard AJ, Reeves RD, Baker AJM. 2014. Facultative hyperaccumulation of heavy metals and metalloids. Plant Science 217–218: 8–17. Proctor J. 2003. Vegetation and soil and plant chemistry on ultramafic rocks in the tropical Far East. Perspectives in Plant Ecology, Evolution and Systematics 6: 105–124. Reeves RD. 1992. Hyperaccumulation of nickel by serpentine plants. In: Proctor J, Reeves RD, Baker AJM, eds. The vegetation of ultramafic (serpentine) soils. Andover: Intercept Ltd., 253–277. Reeves RD. 1999. Nickel hyperaccumulation in the serpentine flora of Cuba. Annals of Botany 83: 29–38. Reeves RD. 2003. Tropical hyperaccumulators of metals and their potential for phytoextraction. Plant and Soil 249: 57–65. Reeves RD , Baker AJM , Jaffré T , Erskine PD, Echevarria G, van der Ent A. 2018a. A global database for hyperaccumulator plants of metal and metalloid trace elements. New Phytologist 218: 407–411. Reeves RD, van der Ent A, Baker AJM. 2018b. Global distribution and ecology of hyperaccumulator plants. In: van der Ent A, Echevarria G, Baker AJM, Morel JL, eds. Agromining: farming for metals: extracting unconventional resources using plants. Cham: Springer International Publishing, 75–92.

Sagner S, Kneer R, Wanner G, Cosson J, Deus-Neumann B, Zenk M. 1998. Hyperaccumulation, complexation and distribution of nickel in Sebertia acuminata. Phytochemistry 47: 339–347. Schaumlöffel D, Ouerdane L, Bouyssiere B, Łobiński R. 2003. Speciation analysis of nickel in the latex of a hyperaccumulating tree Sebertia acuminata by HPLC and CZE with ICP MS and electrospray MS-MS detection. Journal of Analytical Atomic Spectrometry 18: 120–127. Soberón J, Llorente J, Benitez H. 1996. An international view of national biological surveys. Annals of the Missouri Botanical Garden 83: 562–573. Swenson U, Munzinger J. 2010a. Taxonomic revision of Pycnandra subgenus Trouettia (Sapotaceae), with six new species from New Caledonia. Australian Systematic Botany 23: 333–370. Swenson U, Munzinger J. 2010b. Revision of Pycnandra subgenus Sebertia (Sapotaceae) and a generic key to the family in New Caledonia. Adansonia 32: 239–249. Swenson U, Munzinger J. 2016. Five new species and a systematic synopsis of Pycnandra (Sapotaceae), the largest endemic genus in New Caledonia. Australian Systematic Botany 29: 1–40. van der Ent A, Baker, AJM, Reeves RD, Pollard AJ, Schat H, 2012. Hyperaccumulators of metal and metalloid trace elements: facts and fiction. Plant and Soil 362: 319–334. van der Ent A, Mulligan DM. 2015. Multi-element concentrations in plants parts and fluids of Malaysian nickel hyperaccumulator plants and some economic and ecological considerations. Journal of Chemical Ecology 41: 396–408. van der Ent A, Erskine PD, Sumail S. 2015a. Ecology of nickel hyperaccumulator plants from ultramafic soils in Sabah (Malaysia). Chemoecology 25: 243–259. van der Ent A, Jaffré T, L’Huillier L, Gibson N, Reeves RD. 2015b. The flora of ultramafic soils in the Australia–Pacific region: state of knowledge and research priorities. Australian Journal of Botany 63: 173– 190. van der Ent A, Rajakaruna N, Boyd R, Echevarria G, Repin R, Williams D. 2015c. Global research on ultramafic (serpentine) ecosystems (8th International Conference on Serpentine Ecology in Sabah, Malaysia). Australian Journal of Botany 63: 1–16. van der Ent A, Wong KM, Sugau J, Repin R. 2015d. Plant diversity of ultramafic outcrops in Sabah (Malaysia). Australian Journal of Botany 63: 204–215. van der Ent A, Erskine PD, Mulligan DR, Repin R, Karim R. 2016. Vegetation on ultramafic edaphic islands in Kinabalu Park (Sabah, Malaysia) in relation to soil chemistry and altitude. Plant and Soil 403: 77–101. van der Ent A, Callahan DL, Noller BN, Mesjasz- Przybyłowicz J, Przybyłowicz WJ, Barnabas A, Harris HH. 2017. Nickel biopathways in tropical nickel hyperaccumulating trees from Sabah (Malaysia). Scientific Reports 7: 41861. van der Ent A, Mulligan DR, Repin R, Erskine PD. 2018. Foliar elemental profiles in the ultramafic flora of Kinabalu Park (Sabah, Malaysia). Ecological Research 33: 659–674. van der Ent A, Echevarria G, Pollard AJ, Erskine, PD. 2019. X-ray fluorescence ionomics of herbarium collections. Scientific Reports 9: 4746. Wen J, Ickert-Bond SM, Appelhans MS. Dorr LJ, Funk VA. 2015. Collections-based systematics: opportunities and outlook for 2050. Journal of Systematics and Evolution 53: 477–488. Whiting SN, Reeves RD, Richards D, Johnson MS, Cooke JA, Malaisse F, Paton A, Smith JAC, Angle JS, Chaney RL , Ginocchio R , Jaffré T , Johns R , McIntyre T, Purvis OW, Salt DE, Schat H, Zhao FJ, Baker AJM. 2004. Research priorities for conservation of metallophyte biodiversity and their potential for restoration and site remediation. Restoration Ecology 12: 106–116.

Figure 1. Number of specimens with high-range XRF leaf foliar concentrations for each element. Total: the total number of specimens. (a) Cunoniaceae, (b) Phyllanthaceae, (c) Sapotaceae, (d) Salicaceae, (e) Violaceae and Oncothecaceae.

Figure 2. Detail of specimen numbers by species for Cunoniaceae for Ni (a) and Mn (b) records. Darker colours: high-range; lighter colours: low-range.

Figure 3. Detail of specimen numbers by species for Phyllanthaceae for Ni (a) and Mn (b) records. Darker colours: high- range; lighter colours: low-range.

Figure 4. Detail of specimen numbers by species for Oncothecaceae, Salicaceae, Sapotaceae and Violaceae for Ni (a) and Mn (b) records. Darker colours: high-range; lighter colours: low-range.

Figure 5. Detection of Ni hyperaccumulator plant across herbarium specimens (one to four specimens per species). Detail of specimen numbers by species for Ni (a) and Mn (b) records. Darker colours: high-range; lighter colours: low-range.

Figure 6. Proportion of high-range Ni (a) and Mn (b) hyperaccumulator families for the full dataset (all herbarium specimens measured). In (b), families representing 1% (top right quarter) include Stemonuraceae, Alseuosmiaceae, Winteraceae, Ericaceae, Lauraceae, Clusiaceae, Euphorbiaceae. * denotes a new family record.

Figure 7. Phylogenetic distribution of hyperaccumulators of New Caledonia. The horizontal axis shows the number of taxa for each element (Ni, Mn, Co, Zn). a, only high-range hyperaccumulators represented. b, combination of low-range and high-range hyperaccumulators.

Table 1. Threshold values of hyperaccumulation for different trace metal, their relative initial power-fit regressions, inverse power-fit regressions, and correlation coefficients as per identified ≥LODs at 0.95 confidence intervals. * threshold value used in global database (Reeves et al., 2018a). High-range values are use in this work as confident threshold for detection of hyperaccumulators with the XRF method.

“Low range” “High-range” XRF hyperaccumulat hyperaccumulator Equation from calibration Element LOD or (µg g-1) curve r2 (µg g-1) (µg g-1)* y = XRF; x = ICP-AES

Cobalt ≤ 140 ≥ 300 ≥ 1000 y = 0.844

Manganese ≤ 336 ≥ 10 000 ≥ 20 000 y = 0.7282

Nickel ≤ 190 ≥ 1000 ≥ 5000 y = 0.865

Zinc ≤ 305 ≥ 3000 ≥ 10 000 y = 0.9603

Table 2. Top hyperaccumulator taxa for Ni showing the total number of specimens that were analysed, how many specimens were in the element detection threshold (in bracket number of low-range specimens) and in the high-range, with the Ni foliar concentration range and median.

Number of Ni analyses Number of specimens -1 Taxa [Ni] µg g-1 Ni range [median] (µg g ) >LOD- Total ≥5000 <5000 CUNONIACEAE Geissois lanceolata 25 0 25 5550‒51 300 [19 900] Geissois pruinosa 36 3 30 1300‒42 500 [14 750] EUPHORBIACEAE Cleidion velutinum 4 0 4 23 000‒61 400 [25 700] PHYLLANTHACEAE Phyllanthus serpentinus 23 0 23 22 400‒102 000 [51 300] Phyllanthus luciliae 5 0 5 21 100‒69 100 [44 800] Phyllanthus favieri var. favieri 6 1 (1) 5 1800‒59 400 [25 700] Phyllanthus parangoyensis 5 0 5 9200‒51 600 [23 300] Phyllanthus montrouzieri var. poyaensis 2 0 2 9230‒42 800 [26 000] RUBIACEAE Psychotria gabriellae 5 0 5 22 700‒79 900 [34 500] Gynochthodes collina 3 0 3 18 350‒59 900 [41 000] SAPOTACEAE Pycnandra acuminata 18 0 18 19 800‒89 700 [35 200] Pycnandra kouakouensis 3 0 3 42 700‒80 500 [47 900] VIOLACEAE Hybanthus austrocaledonicus 102 18 84 680‒122 000 [20 000] Hybanthus caledonicus 22 8 14 270‒54 800 [10 700]

Table 3. Top hyperaccumulator taxa for Mn showing the total number of specimens that were analysed, how many specimens were in the element detection threshold (in bracket number of low-range specimens) and in the high-, with the Mn foliar concentration range and median.

Number of Mn analyses Number of specimens Mn range [median] (µg g-1) Taxa -1 [Mn] µg g >LOD- Total ≥20 000 <20 000 ALSEUOMIACEAE Platyspermation crassifolium 4 0 4 20 400‒85 400

APOCYNACEAE

Alyxia poyaensis 4 0 4 27 500‒69 900 [41 800]

CELASTRACEAE Denhamia fournieri subsp. drakeana 4 0 4 37 800‒77 200 [43 600]

Denhamia fournieri subsp. fournieri 4 0 4 25 500‒100 000 [41 600]

CUNONIACEAE

Pancheria multijuga 14 0 14 23 700‒238 000 [80 400]

Pancheria reticulata 19 0 19 25 000‒ 224 700 [93 700] ERICACEAE

Dracophyllum verticillatum 4 0 4 20 100‒80 500 [36 400]

MYRTACEAE

Gossia alaternoides 4 0 4 64 600‒81 800 [69 900]

Gossia kaalaensis 4 1 (0) 3 6900‒103 000 [55 000]

Gossia pancheri 4 0 4 42 900‒97 500 [90 300]

PHYLLANTHACEAE

Phyllanthus fractiflexus 3 0 3 20 000‒82 100 [22 600]

PROTEACEAE

Kermadecia pronyensis 4 0 4 22 800‒78 600 [39 600]

Virotia angustifolia 4 2 (2) 2 11 100‒74 400 [42 400]

Virotia neurophylla 4 0 4 103 000‒364 000 [192 000]

RUBIACEAE

Psychotria pseudomicrodaphne 4 2 (2) 2 14 900‒158 000 [36 900]

Table 4. All high-range hyperaccumulator taxa for Co showing the total number of specimens that were analysed, how many specimens were in the element detection threshold (in bracket number of low-range specimens) and in the high- range, with the Co foliar concentration range and median.

Number of Co analyses

Number specimens -1 Taxa [Co] µg g-1 Co range [median] µg g >LOD-< Total ≥ 1000 µg g-1 1000 µg g-1 CUNONIACEAE Pancheria engleriana 69 18 (5) 1 160‒1250 [300]

Spiraeanthemum meridionale 57 35 (22) 5 150‒1740 [400]

PHYLLANTHACEAE

Phyllanthus favieri var. favieri 6 4 (2) 1 140‒2130 [580] Phyllanthus fractiflexus 3 2 (1) 1 190‒1450 [920]

SALICACEAE

Homalium kanaliense var. boulindae 8 0 1 1640

PROTEACEAE Kermadecia pronyensis 4 1 (1) 3 920‒3090 [2410]

RUBIACEAE

Gynochthodes sp. nov (Dagostini 87) 1 0 1 1930

VIOLACEAE

Hybanthus austrocaledonicus 109 6 2 140‒1210 [200]

Table 5. All high-range hyperaccumulator taxa for Zn showing the total number of specimens that were analysed, how many specimens were in the element detection threshold (in bracket number of low-range specimens) and in the high-range, with the Zn foliar concentration range and median.

Number of Zn analyses Number of specimens Zn range [median] Taxa -1 [Zn] µg g µg g-1 >LOD- Total ≥ 10 000 <10 000 PHYLLANTHACEAE Phyllanthus veillonii 4 0 (3) 1 10 300

SALICACEAE

Casearia silvana 110 8 (0) 1 1810‒10 200 [2210]

VIOLACEAE Agatea longipedicellata 35 1 1 3200‒12 300 [7730]

Hybanthus austrocaledonicus 109 30 (2) 1 1550‒10 300 [2120]

Table 6. Summary of New Caledonian genera in the Cunoniaceae, Oncothecaceae, Phyllanthaceae, Salicaceae, Sapotaceae, and Violaceae families; with all species samples of corrected XRF values. Range, means, and medians of cobalt, manganese, nickel and zinc foliar concentrations in μg g-1.

Number of -1 mean/median foliar concentration in µg n taxa Range of foliar concentration in µg g -1 hyperaccumulators (high Order Family Genera g /sample ranges species) Co Mn Ni Zn Co Mn Ni Zn Co Mn Ni Zn Manilkara 1/26 ‒ ‒ ‒ ‒ ‒ ‒‒ ‒ ‒ ‒ ‒ ‒ Mimusops 1/34 ‒ 530‒1180 ‒ ‒ ‒ 720/650 ‒ ‒ ‒ ‒ ‒ ‒ Pichonia 7/178 ‒ 430‒660 260‒24 500 3840 ‒ 530/520 1750/640 3840/3840 ‒ ‒ 1 ‒ Ericales Sapotaceae Planchonella 37/717 ‒ 430‒20 000 250‒2810 1540‒2100 ‒ 2420/1050 580/430 1710/1590 ‒ ‒ ‒ ‒ Pleioluma 17/328 ‒ 440‒23 700 250‒440 ‒ ‒ 4050/2770 300/260 ‒ 1 ‒ ‒ Pycnandra 59/847 150‒160 430‒39 200 270‒89 700 1550‒7590 155/155 2870/1240 11 3200/2670 1 1 4 ‒ Icacinales Oncothecaceae Oncotheca 2/80 170 440 7750‒21 700 ‒ 170/170 440/440 6670/6580 ‒ ‒ ‒ 1 ‒ Antidesma 1/11 ‒ 650‒9640 ‒ ‒ ‒ 2890/2570 ‒ ‒ ‒ ‒ ‒ ‒ Breynia 1/15 ‒ 2000‒3840 ‒ ‒ ‒ 2900/2900 ‒ ‒ ‒ ‒ ‒ ‒ Phyllanthaceae Cleistanthus 1/53[4] ‒ 430‒7300 ‒ ‒ ‒ 1520/830 ‒ ‒ ‒ ‒ ‒ ‒ Bischofia 1/5 ‒ ‒ ‒ ‒ ‒ ‒ ‒ ‒ ‒ ‒ ‒ ‒ Phyllanthus 177/1270 140‒2130 430‒82 100 240‒102 000 1570‒10150 338/190 3480/1540 5430/1090 2620/1910 2 5 29 1

Malphigiales Casearia 6/190 350 430‒42 800 260‒3420 1810‒10 200 350/350 3210/1570 740/430 3000/2180 0 2 0 1 Homalium 22/274 140‒1640 430‒19 900 260‒35 300 1530‒2810 240/140 3040/1470 3500/880 1940/1910 1 6 ‒ ‒ Salicaceae Lasiochlamys 10/120 ‒ 440‒18 800 260‒7210 1770‒2080 100/100 3660/1980 930/570 1920/1920 ‒ ‒ ‒ ‒ Xylosma 21/250 ‒ 450‒2800 250‒24 500 1510‒2260 ‒ 1630/740 2690/1500 1710/1550 ‒ ‒ 7 ‒ Agatea 6/127 170 430‒21 200 260‒3610 1540‒12 300 170/170 2600/1280 950/600 3590/1960 ‒ 1 ‒ 1 Violaceae 22 100/13 Hybanthus 5/168 140‒1210 430‒63 600 270‒122 000 1510‒10 300 240/170 4560/1330 2450/2110 1 1 4 1 700 Codia 15/425 140‒960 430‒17 800 250‒7870 1750 220/170 1820/1270 1030/700 1750/1750 ‒ ‒ 2 ‒ Cunonia 25/618 140‒800 430‒22 100 250‒4120 1730‒2180 210/150 2817/1460 670/550 1900/1800 ‒ 2 ‒ ‒ 11 Oxalidales Cunoniaceae Geissois 15/317 140‒460 430‒20 000 250‒51 300 1610‒2620 160/140 2280/1100 2046/1907 ‒ 1 9 ‒ 200/7300 Hooglandia 1/2 ‒ 1440‒2560 ‒ ‒ ‒ 2000/2000 ‒ ‒ ‒ ‒ ‒ ‒ Pancheria 30/827 140‒1250 430‒238 000 240‒32 900 1550‒2130 220/150 13 2900/750 1778/1700 1 3 8 ‒

30

Spiraeanthemum 7/210 140‒1740 440‒44 900 250‒4370 ‒ 360/220 7880/4440 980/670 ‒ 1 2 ‒ ‒ Weinmannia 5/120 ‒ 440‒9820 ‒ ‒ ‒ 1420/1020 ‒ ‒ ‒ ‒ ‒ ‒

31