Evolutionary origins of abnormally large shoot sodium accumulation in nonsaline environments within the Caryophyllales

Summary The prevalence of sodium (Na)‐‘hyperaccumulator’ species, which exhibit abnormally large shoot sodium concentrations ([Na]shoot) when grown in nonsaline environments, was investigated among angiosperms in general and within the Caryophyllales order in particular. Shoot Na concentrations were determined in 334 angiosperm species, representing 35 orders, grown hydroponically in a nonsaline solution. Many Caryophyllales species exhibited abnormally large [Na]shoot when grown hydroponically in a nonsaline solution. The bimodal distribution of the log‐normal [Na]shoot of species within the Caryophyllales suggested at least two distinct [Na]shoot phenotypes within this order. Mapping the trait of Na‐hyperaccumulation onto the phylogenetic relationships between Caryophyllales families, and between subfamilies within the Amaranthaceae, suggested that the trait evolved several times within this order: in an ancestor of the Aizoaceae, but not the Phytolaccaceae or Nyctaginaceae, in ancestors of several lineages formerly classified as Chenopodiaceae, but not in the Amaranthaceae sensu stricto, and in ancestors of species within the Cactaceae, Portulacaceae, Plumbaginaceae, Tamaricaceae and Polygonaceae. In conclusion, a disproportionate number of Caryophyllales species behave as Na‐hyperaccumulators, and multiple evolutionary origins of this trait can be identified within this order.


Introduction
Sodium (Na) is not considered to be an essential element for plants (White & Brown, 2010), although it is required (in micronutrient quantities) for the C 4 photosynthetic pathway (Cheeseman, 2015) and some halophytes (euhalophytes) grow better when supplied with Na (Greenway & Munns, 1980;Albert, 1982;Flowers & Colmer, 2008;Munns & Tester, 2008;Rozema & Schat, 2013). In addition, in some environments, for example where there is low K + phytoavailability, plant growth can benefit from a source of Na, as Na + can replace K + as a cationic osmoticum in the vacuole (White, 2013). The accumulation of excessive Na concentrations in plant tissues is, however, detrimental to plant growth as Na + interferes with metabolism in the cytoplasm, mitochondria and plastids (Flowers et al., 2015).
It is estimated that > 6% of the world's land, and 5-15% of the world's agricultural land, is adversely affected by Na concentration through either salinity or sodicity (Munns & Tester, 2008). Saline soils are generally dominated by NaCl, although there are often significant concentrations of Ca 2+ , Mg 2+ , SO 2À 4 and CO 2À 3 . They are defined as having < 15% of their exchangeable cations as Na + and soil solutions with electrical conductivity (EC e ) > 2 dS m À1 in a saturated paste extract, which equates to an NaCl concentration of 20 mM and pH < 8.5. Sodic (alkali) soils are generally dominated by Na 2 CO 3 , and are defined as having > 15% of their exchangeable cations as Na + and soil solutions with EC e > 2 dS m À1 and pH > 8.5 in a saturated paste extract. Saline-sodic soils have > 15% of their exchangeable cations as Na + and soil solutions with EC e > 2 dS m À1 and pH < 8.5 in a saturated paste extract.
Halophytes can also be grouped into 'ionotypes', which are defined as characteristic ionomic features of plant species that are conserved in diverse environments (Albert & Popp, 1977;Gorham et al., 1980;Albert et al., 2000;Flowers & Colmer, 2008;White et al., 2012). Commelinid monocots (e.g. Poaceae, Cyperaceae, Juncaceae) are classed as 'Na-excluders' and generally exhibit lower shoot Na concentrations ([Na] shoot ) than other angiosperms growing in the same environment and Na/K quotients less than unity, whereas many eudicots are characterized by comparatively large [Na] shoot and tissue Na/K quotients greater than unity (Albert & Popp, 1977;Gorham et al., 1980;Albert, 1982;Flowers & Colmer, 2008;Yang et al., 2012). Several families in the Caryophyllales (Amaranthaceae (Chenopodioideae), Caryophyllaceae, Tamaricaceae) exhibit exceptionally large [Na] shoot and tissue Na/K quotients when grown in saline environments (Albert & Popp, 1977;Gorham et al., 1980;Albert, 1982;Flowers & Colmer, 2008;Yang et al., 2012;Zhang et al., 2012). It has also been observed that some Caryophyllales species have exceptionally large [Na] shoot , even when grown in nonsaline environments (Collander, 1941;Patel et al., 1980;Glenn & O'Leary, 1984;Broadley et al., 2004). For example, in a phylogenetically balanced study of the ionomes of 117 angiosperm species belonging to 25 orders grown hydroponically in a nonsaline solution containing 0.1 mM Na, it was noted that [Na] shoot varied significantly among eudicot orders (P < 0.05) and that three of the seven Caryophyllales species studied had conspicuously large [Na] shoot (Broadley et al., 2004). It has been suggested that Na might have a special role in the biology of euhalophyte Caryophyllales, whose maximal growth requires Na accumulation (Flowers & Colmer, 2008), and that the characteristic ionome of the Caryophyllales might reflect their unusual ecology (White et al., 2015). Although Caryophyllales species can inhabit a variety of biomes worldwide, they comprise a significant proportion of the flora of many deserts (Fahn & Cutler, 1992), coastal regions (Kadereit et al., 2012) and soils with unbalanced mineral composition for plant nutrition, such as gypseous (Moore et al., 2014) and ultramafic/serpentine (White & Pongrac, 2016) soils.
The Caryophyllales order comprises over 11 000 species currently partitioned into c. 700 genera and 38 families (The Plant List, 2013;APGIV, 2016). About 5% of species in the Caryophyllales are halophytes and the order contains 35-40% of all known halophytic angiosperm species (Flowers et al., 2010(Flowers et al., , 2016Saslis-Lagoudakis et al., 2014). Of the most populous families in the Caryophyllales (> 50 species), the halophytic character is particularly prevalent in the Amaranthaceae (17.3% species), Frankeniaceae (16.7% species) and Tamaricaceae (31.1% species). In contrast with observations on other angiosperm orders, the halophytic character appears to be rarely lost in Caryophyllales lineages, such as the Chenopodioideae and Tamaricaceae, once it has evolved (Bromham, 2015). It has been suggested that the halophytic character might evolve from ancestors with a general complement of stress-tolerance traits that enable lineages to adapt to a wide range of environmental challenges (Kadereit et al., 2012;Saslis-Lagoudakis et al., 2014;Bromham, 2015). It is therefore noteworthy that the Caryophyllales order contains many succulent species (Kadereit et al., 2012;Rozema & Schat, 2013), many species that possess salt glands, which are specialized multicellular structures that excrete salt onto the leaf surface, or bladder cells, which are modified trichomes that accumulate salt and then burst (Thomson et al., 1988;Fahn & Cutler, 1992;Salama et al., 1999;Flowers et al., 2010;LoPresti, 2014), many species exhibiting C 4 and Crassulacean acid metabolism (CAM) photosynthetic pathways (Silvera et al., 2010;Sage et al., 2011;Kadereit et al., 2012), many species that hyperaccumulate potentially toxic elements (White & Pongrac, 2016) and many species adapted to arid (Ehleringer et al., 1997) or alkaline (Yang et al., 2012) environments. The C 4 photosynthetic pathway has evolved many times within the Caryophyllales (Sage et al., 2011) and Kadereit et al. (2012) observed that the rate of gain of the C 4 photosynthetic character was greater in salt-tolerant Chenopodioideae lineages, which they attributed to shared adaptations between C 4 photosynthesis and salt tolerance as part of a wider drought tolerance syndrome. A similar dependence of the evolution of C 4 photosynthesis with succulence and coastal habitat was also observed (Kadereit et al., 2012). CAM has also evolved many times within the Caryophyllales and is associated with succulence and other traits enabling water use efficiency in arid or saline environments (Edwards & Ogburn, 2012).
The present study investigated the prevalence of 'Nahyperaccumulator' species, which exhibit abnormally large [Na] shoot (> 4 mg Na g À1 dry matter (DM)) when grown in nonsaline conditions (< 20 mM Na + in the rhizosphere solution), among the angiosperms in general and the Caryophyllales in particular. The prevalence of this phenomenon among angiosperms is currently unknown and this study provides an original insight to its occurrence and evolutionary origins within the Caryophyllales order. It was observed that only the Caryophyllales species Atriplex hortensis and Beta vulgaris of the 12 halophytic species studied, representing 10 angiosperm orders, behaved as Na-hyperaccumulators when grown in compost. Similarly, when 334 angiosperm species representing 35 angiosperm orders were grown hydroponically in a nonsaline solution containing 0.1 mM Na, a disproportionate number of Caryophyllales species exhibited abnormally large [Na] shoot . The bimodal distribution of the log-normal [Na] shoot of species within the Caryophyllales suggested at least two distinct [Na] shoot phenotypes within this order. Mapping the trait of Na-hyperaccumulation in nonsaline environments onto the phylogenetic relationships between Caryophyllales families (Crawley & Hilu, 2012;Hern andez-Ledesma et al., 2015;Yang et al., 2015), and between subfamilies within the Amaranthaceae, suggested that the trait had evolved several times within this order: in an ancestor of the Aizoaceae, but not the Phytolaccaceae or Nyctaginaceae, in ancestors of several lineages formerly classified as Chenopodiaceae, but not in the Amaranthaceae sensu stricto, and possibly in ancestors of species within the Cactaceae, Portulacaceae, Plumbaginaceae, Tamaricaceae and Polygonaceae. It is possible that the ability to hyperaccumulate Na + might benefit plants by providing an alternative osmoticum to K + , especially in environments with low K availability (White, 2013). Thus, Na-hyperaccumulation might have served Caryophyllales during their evolution in overcoming the selection pressures associated with the colonization of arid or saline environments, which require succulence and water conservation (Fahn & Cutler, 1992;Nobel, 2003;Flowers & Colmer, 2008;Kadereit et al., 2012).

Responses of halophytic species from different angiosperm orders to salinity
Responses to salinity were studied in 12 halophytic species, from 10 angiosperm orders, catalogued in the eHALOPH Halophytes Database (Flowers et al., 2016) Species were chosen on the basis of their availability from suppliers and their ability to grow in the glasshouse. Seeds of all species were obtained from Chiltern Seeds (Wallingford, UK), except C. cunninghamiana, H. tilliaceous, K. virginica and P. maritima, which were obtained from Rareexoticseeds (Montreal, Canada), Kenni Koala's Aussie Seed Store (Australia), Floridawildflowers (Crescent City, FL, USA) and Scotia Seeds (Brechin, UK), respectively. Seeds were germinated in the dark at between 10°C and 25°C, according to species requirements, on the surface of filter paper moistened with deionized water. Once a radicle was observed, individual seedlings were transplanted to rockwool plugs (2.5 9 2.5 9 4 cm 3 ; Grodan, Hedehusene, Denmark) held in plastic trays in a glasshouse compartment at The James Hutton Institute, Dundee (UK; latitude 56°27 0 26″N, longitude 3°4 0 17″W), in which the experiment was subsequently performed, and irrigated with tap water containing 0.14 mM Na. The glasshouse compartment maintained a maximum of 25°C by day and a minimum of 15°C at night using automatic venting and supplementary heating.
Established seedlings were transferred to pots containing 1 l Levington Professional compost (ICL, Ipswich, UK) before the experiment. Two sets of plants, with up to 12 replicate plants per species in each set, were exposed to either nonsaline or saline irrigation. Plants were irrigated with 100 ml solution wk À1 . Plants receiving the nonsaline treatment were irrigated with tap water containing 0.14 mM Na. The experiment was initiated by increasing the NaCl concentration in the irrigation water of the saline treatment to 50 mM for the first week, then 150 mM for the second week and, finally, 300 mM for the third week. Plants were harvested on 12 December 2014, 3 wk after the first addition of NaCl to the saline irrigation water. The fresh weight (FW) of whole shoots was determined immediately, and then samples were dried in an oven at 70°C to a constant weight and their DM was determined. Dried samples were milled to a powder using a ball mill (C + N Laboratory Mill; Christy and Norris Ltd, Chelmsford, UK), digested using HNO 3 in sealed tubes in a microwave oven (MARS Xpress, CEM Corporation, Matthews, NC, USA), cleared using hydrogen peroxide (H 2 O 2 ) and analysed for Na concentration using inductively coupled plasma-mass spectrometry (ICP-MS; ELAN DRCe, PerkinElmer, Waltham, MA, USA), as described by White et al. (2012).

Phylogenetic effects on shoot sodium concentrations in plants grown hydroponically in a nonsaline solution
Phylogenetic effects on shoot Na concentrations in angiosperm species were assessed by combining data from six glasshouse experiments in which plants were grown hydroponically using a Nutrient Film Technique (NFT), essentially as described by Broadley et al. (2003). The final dataset comprised 334 species from 35 orders (Supporting Information Table S1). In all experiments, seeds were germinated in the dark on the surface of filter paper moistened with deionized water at temperatures between 4°C and 25°C, depending on their requirements. Once a radicle was observed, individual seedlings were transplanted to rockwool plugs (2.5 9 2.5 9 4 cm 3 ; Grodan) held in plastic trays and irrigated with tap water. Plastic trays were either placed in a weaning room at 25°C or in the glasshouse compartment in which the experiments were subsequently performed. Once seedlings were established, the rockwool plugs containing the plants were transferred to the NFT system. Whenever possible, two rockwool plugs constituted each replicate and up to six replicates were obtained for each plant species. For experiments at both Warwick-HRI, Wellesbourne (UK; latitude 52°12 0 18″N, longitude 1°36 0 00″W) and The James Hutton Institute, the glasshouse maintained a maximum of 20°C by day and a minimum of 15°C at night using automatic venting and supplementary heating. The recirculating nutrient solution contained 2 mM Ca(NO 3 ) 2 , 2 mM NH 4 NO 3 , 0.75 mM MgSO 4 , 0.5 mM KOH, 0.25 mM KH 2 PO 4 , 0.1 mM FeNaEDTA, 30 lM H 3 BO 3 , 25 lM CaCl 2 , 10 lM MnSO 4 , 3 lM CuSO 4 , 1 lM ZnSO 4 and 0.5 lM Na 2 MoO 4 . This was adjusted daily to pH 6, with H 2 SO 4 , and solutions were replaced completely once or twice each week. Seedlings were harvested during the exponential growth phase, 18-73 d after transfer to the hydroponic system, depending on the plant growth rate. Whenever possible, shoots were separated into leaves and stems. The FW of whole shoots or leaves was determined immediately and then samples were dried in an oven New Phytologist (2017)

Research
New Phytologist at 70-80°C to a constant weight and their DM was determined. Dried samples were milled to a powder using a ball mill, acid digested and their Na concentrations were determined either by inductively coupled plasma-emission spectrometry (JY24; Jobin-Yvon, Longjumeau, France), as described by Broadley et al. (2003;Exps 1-4), or by ICP-MS, as described by White et al. (2012;Exps 5, 6).
Exp 1, described by Broadley et al. (2004), was undertaken in a glasshouse compartment at Warwick-HRI between July and October 2001 to survey calcium (Ca), potassium (K), magnesium (Mg), Na, organic-N and phosphorus (P) concentrations in leaves of a phylogenetically balanced set of 117 angiosperm species belonging to 25 orders. Exps 2(A-C) were undertaken sequentially in a glasshouse compartment at Warwick-HRI between May and November 2003 to survey Ca concentrations in leaves of Magnoliid and monocot orders, with replication at the taxonomic level of the family. Six species representing three Magnoliid orders, 54 species representing eight monocot orders and nine other angiosperm species were grown in this experiment. Exp 3, described by White et al. (2007), was undertaken in a glasshouse compartment at Warwick-HRI between July and August 2004 to survey selenium (Se) concentrations in leaves of 35 angiosperm species chosen to represent the range of ecological strategies for Se accumulation reported in angiosperms. Exp 4, described by White et al. (2015), was undertaken in a glasshouse compartment at Warwick-HRI between June and August 2004 to survey leaf concentrations of Ca and Mg in as many Caryophyllales families as possible, with replication at the taxonomic level of the genus. Forty-six Caryophyllales species were studied, representing eight families and 29 genera, together with 33 other angiosperm species. Exp 5 was undertaken in a glasshouse compartment at The James Hutton Institute between July and October 2011 to survey leaf Ca and Mg concentrations in a range of serpentine and nonserpentine plant species. These included 28 Caryophyllales species and 35 other angiosperm species. Exp 6 was undertaken in a glasshouse compartment at The James Hutton Institute between July and November 2015 to survey leaf Ca and Mg concentrations in a range of Arecaceae species, with replication at the taxonomic level of the genus. Twenty-three Arecaceae species were studied, representing six genera, together with 11 other angiosperm species. Each experiment had several species in common with other experiments, allowing cross comparisons (Table S1). In total, 53 species, representing 22 families and 15 orders, were grown in more than one experiment.

Statistics
Data are expressed as the mean and SE or SD of n observations. Statistical differences between treatments were assessed for each species by Student's t-test. Estimates of variation in [Na] shoot were assigned between and within orders (n = 35), families (n = 79) and species (n = 334) using analyses of variance (ANOVA). All statistical analyses were performed using R 3.3.0 (R Core Team, 2016) employing a linear model of: [Na] shoot~O rder + Family + Species.

Results
Twelve halophytic angiosperm species were grown in compost in pots that were irrigated with either nonsaline or saline solution. The shoot FW of most of these species did not differ significantly between plants that were irrigated with nonsaline and saline solutions (Table 1). However, the shoot FWs of A. officinalis (P = 0.0193) and K. virginica (P = 0.0430) were lower in plants irrigated with saline solution than in those irrigated with nonsaline solution, whereas the shoot FWs of A. hortensis (P = 0.0090) were greater in plants irrigated with saline solution than in those irrigated with nonsaline solution. Previous studies have also suggested that halophytic Atriplex species grow best under slightly saline conditions (Black, 1960;Wallace et al., 1973;Storey & Wyn Jones, 1979;Albert, 1982;Glenn & O'Leary, 1984;Redondo-G omez et al., 2007;Glenn et al., 2012;Norman et al., 2013).
The distribution of [Na] shoot among the angiosperm species studied did not fit a simple normal distribution (Fig. 1a) and the log-normal distribution of [Na] shoot appeared to comprise the sum of at least three individual log-normal distributions (Fig. 2a). The [Na] shoot of Caryophyllales species differed by several orders of magnitude, from 0.05 mg g À1 DM in Lewisia cotyledon (Montiaceae) to 40.02 mg g À1 DM in D. bellidiformis (Aizoaceae).
The distribution of [Na] shoot among the Caryophyllales appeared to comprise a normal distribution (mean = 0.393, SD = 0.185 mg g À1 DM, n = 42 species) plus up to 19 species with abnormally large [Na] shoot (Fig. 1b). The low probabilities of these species being part of the normal distribution suggested that there were at least two distinct [Na] shoot phenotypes among Caryophyllales species. The species with [Na] shoot at the limit for inclusion in the normal distribution were Plumbago auriculata (P = 0.0153, rank no. 41), Gomphrena serrata (P = 0.0106, rank no. 42), Rumex hydrolapathum (P = 0.0003, rank no. 43) and Limonium sinuatum (P = 0.0001, rank no. 44).
The distribution of log-normal [Na] shoot of Caryophyllales species appeared to comprise the sum of two discrete log-normal distributions (Fig. 2b). The first log-normal distribution (mean = À0.3717, SD = 0.3299, n = 49 species) contained 49 species and the second log-normal distribution (mean = 1.246, SD = 0.2756, n = 12 species) contained 12 species (Fig. 2b). As these two log-normal distributions differed significantly (P < 0.0001), these data suggest that there are at least two distinct [Na] shoot phenotypes among Caryophyllales species. Considering the species with log [Na] shoot at the extremes of these two distributions, the log [Na] shoot of Psylliostachys suworowi had a greater probability of being in the first rather than the second log-normal distribution (P = 0.0076 vs P = 0.0017), whereas the log [Na] shoot of Spergula arvensis had a greater probability of being in the second rather than the first log-normal distribution (P = 0.0210 vs Table 1 Shoot fresh weight (FW), dry matter (DM) and sodium concentration ([Na] shoot ) of 12 halophytic angiosperm species grown in pots irrigated with either 100 ml nonsaline (0.14 mM Na) or saline (50-300 mM Na) solution wk À1 The experiment was initiated by increasing the NaCl concentration in the irrigation water of the saline treatment to 50 mM for the first week, then 150 mM for the second week and, finally, 300 mM for the third week. Plants were harvested 3 wk after the first addition of NaCl to the saline irrigation water. Data are expressed as mean AE SE of n observations.

Research
New Phytologist P = 0.0007). The trait of abnormally large [Na] shoot when plants are grown in nonsaline solutions will henceforth be termed 'Nahyperaccumulation', and the discrete set of 12 Caryophyllales species with large log [Na] shoot are considered to be 'Nahyperaccumulators'.
The evolutionary origin of Na-hyperaccumulation was sought by comparing the number of Na-hyperaccumulator species and the mean [Na] shoot in different families of the Caryophyllales (Fig. 3). The 12 Caryophyllales species exhibiting Nahyperaccumulation were distributed across five of the 10 Caryophyllales families represented in this study. However, the trait was most prevalent in the Aizoaceae. Six of the seven Aizoaceae species studied exhibited Na-hyperaccumulation. These six species were among the seven Caryophyllales species with the largest [Na] shoot (Table S1). Consequently, the Aizoaceae had the largest mean [Na] shoot (24.47 AE 5.07 mg g À1 DM, n = 7 species) of all the Caryophyllales families. The only Cactaceae species studied, Echinofossulocactus sp., also had one of the largest [Na] shoot measured (17.60 mg g À1 DM, n = 1 experiment). In addition, two of the 12 Amaranthaceae species studied (A. hortensis, B. vulgaris), two of the 20 Caryophyllaceae species studied (Silene armeria, S. arvensis) and one of the two Portulacaceae species studied (Portulaca grandiflora) could also be considered as Na-hyperaccumulators (Table S1). However, as there were proportionally fewer Na-hyperaccumulator species in these families, and the Na-hyperaccumulator species in these families generally had smaller [Na] shoot than the Aizoaceae Nahyperaccumulator species, their mean [Na] shoot was less than the mean [Na] shoot of the Aizoaceae (Fig. 3). No Nahyperaccumulator species were observed in the Phytolaccaceae, Nyctaginaceae, Montiaceae, Polygonaceae or Plumbaginaceae. Based on the phylogenetic relationships between Caryophyllales families proposed recently (Crawley & Hilu, 2012;Hern andez-Ledesma et al., 2015;Yang et al., 2015) and the data from the experiments reported here (Table S1), it appears that the trait of Na-hyperaccumulation might have evolved several times within the Caryophyllales (Fig. 3). It is likely that the trait evolved in an ancestor of the Aizoaceae, but not the Phytolaccaceae or

Discussion
Angiosperm species can be classified into 'Na-excluders', 'Naresponders' and 'Na-accumulators' based on the relationship between their [Na] shoot value and the salinity of the irrigation solution (cf Baker, 1981). This terminology, originally proposed to categorize the responses of plant species to toxic elements ('heavy metals') in the environment, also appears to be valid for Na accumulation, as the accumulation of excessive Na + can be toxic to plants and plant species respond to Na + in their environment either by excluding this cation or accumulating it safely in their tissues. Only two of the 12 halophytic species studied in detail in this article could be classified as Na-accumulators (Table 1). These were the Caryophyllales species A. hortensis and B. vulgaris, which both had exceptionally large [Na] shoot when irrigated with either nonsaline or saline solution. A similar response of [Na] shoot to increasing salinity in the root environment has been observed previously for other Caryophyllales species, including members of the Atriplex, Salicornia and Suaeda genera (Albert, 1982;Glenn & O'Leary, 1984). However, not all Caryophyllales species exhibit this trait, and the response of [Na] shoot to increasing salinity in the root environment of, for example, the miohalophytes Rumex dentatus and Limonium perezii is typical of Na-excluders, whereas the response of [Na] shoot to increasing salinity in the root environment of, for example, Sarcobatus vermiculatus is reminiscent of Na-responders (Glenn & O'Leary, 1984). The prevalence of Na-accumulator species, which exhibit abnormally large [Na] shoot when grown under nonsaline conditions, was assessed by combining data from six glasshouse experiments in which 334 angiosperm species representing 35 angiosperm orders had been grown hydroponically in a nonsaline solution containing 0.1 mM Na for 18-73 d (Table S1). It was observed that a relatively large number of Caryophyllales species exhibited abnormally large [Na] shoot (> 10 mg g À1 DM) when grown in nonsaline solution (Table S1; Fig. 1). The distribution of the log-normal [Na] shoot of Caryophyllales species appeared to comprise two discrete log-normal distributions containing 49 and 12 species, respectively (Fig. 2), suggesting that there are at least two distinct [Na] shoot phenotypes among Caryophyllales species. The [Na] shoot distinguishing between these two distributions was c. 4 mg Na g À1 DM.
The ability of plants to accumulate Na when growing in nonsaline environments is not considered to be an evolutionary advantage (Cheeseman, 2015). Indeed, it has been suggested that grazing by herbivores has selected for glycophyte species that maintain [Na] shoot below c. 1-2 mg g À1 DM (Cheeseman, 2015). Nevertheless, it is possible that the ability to accommodate large [Na] shoot might be an enabling trait allowing species to adapt to a variety of abiotic environmental challenges. It might confer the ability for osmotic adjustment in environments with low K phytoavailability or contribute to tolerance to arid or saline environments (Flowers & Colmer, 2008;Kadereit et al., 2012;White, 2013). However, it can be observed that the trait of Na-hyperaccumulation within the Caryophyllales is not directly correlated with the expression of either C 4 photosynthesis or CAM, tissue succulence, halophytism in general or the euhalophytic trait in particular (Table S2).
The trait of Na-hyperaccumulation in nonsaline environments is less ubiquitously exhibited by Amaranthaceae species (Table S2, and references therein). However, it is exhibited by many species formerly classified as Chenopodiaceae. It is exhibited by the Betoideae, B. vulgaris and Hablitzia tamnoides, by some Camphorosmoideae (e.g. Bassia hyssopifolia and Maireana brevifolia), by many Chenopoidioideae including most, but not all, Atriplex and Chenopodium species, by Corispermum hyssopifolium and Corispermum pallasii ssp. membranaceum, by all the Salicornioideae studied, including several Salicornia and Tecticornia species, by many Salsoloideae, and all Suaeda species (Table S2). By contrast, the trait is not exhibited by any Amaranthaceae sensu stricto (Amaranthoideae, Gomphrenoideae), with the exception of Ptilotus polystachyus. Although many Amaranthaceae species possess bladder cells or salt glands (Thomson et al., 1988;Fahn & Cutler, 1992;Flowers et al., 2010;LoPresti, 2014), there does not appear to be a direct correlation between the presence of salt glands and the ability to hyperaccumulate Na in nonsaline environments (Table S2).
In the experiments reported here, no Na-hyperaccumulator species were observed in the Plumbaginaceae or Polygonaceae (Table S1; Fig. 3). Nevertheless, several species in these families have been reported to accumulate large [Na] shoot when grown in nonsaline environments (Table S2, and references therein). In addition, all six species of Tamaricaceae studied to date appear to accumulate large [Na] shoot when grown in nonsaline environments (Patel et al., 1980;Ding et al., 2010;Li et al., 2010;Gorai & Neffati, 2011;Sghaier et al., 2015;Sharif & Khan, 2016). It is, perhaps, noteworthy that many species in the Plumbaginaceae and Tamaricaceae possess salt glands, whereas members of the Polygonaceae do not (Thomson et al., 1988;Fahn & Cutler, 1992;Salama et al., 1999;Flowers et al., 2010). Again, there does not appear to be a direct correlation between the occurrence of salt glands and the ability of a species to hyperaccumulate Na in nonsaline environments (Table S2).
In conclusion, phylogenetic relationships between Caryophyllales families suggest that the trait of Na-hyperaccumulation in nonsaline environments has evolved several times within this order (Fig. 3). The data presented here suggest that the trait evolved in an ancestor of the Aizoaceae, but not the Phytolaccaceae or Nyctaginaceae. It is also likely that the trait also evolved in an ancestor of species formerly classified as Chenopodiaceae (subfamilies Betoideae, Chenopodioideae, Camphorosmoideae, Salsoloideae, Salicornioideae, Suaedoideae), but not the Amaranthaceae sensu stricto (subfamilies Amaranthoideae, Gomphrenoideae). In addition, it is possible that the trait evolved in ancestors of the Sarcobataceae, Portulacaceae, Cactaceae, Tamaricaceae, Plumbaginaceae and Polygonaceae, but further studies are required to explore these hypotheses. Future studies should focus on the elucidation of the evolutionary origin of Nahyperaccumulation in nonsaline environments among species formerly classified as Chenopodiaceae, among species in the Cactaceae and Portulacaceae, which are currently under-represented in published studies, and among species in the Plumbaginaceae, Tamaricaceae and Polygonaceae, to determine the extent of the trait in these families.

Supporting Information
Additional Supporting Information may be found online in the Supporting Information tab for this article: Table S1 Shoot sodium concentrations in 334 species from 35 angiosperm orders grown hydroponically in a nonsaline solution containing 0.1 mM Na + in at least one of six glasshouse experiments Table S2 Occurrence of sodium (Na)-hyperaccumulator species, having shoot Na concentrations > 4 mg g À1 dry matter when grown in nonsaline environments, within the Caryophyllales order, together with their halophytic and photosynthetic characteristics Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.
New Phytologist is an electronic (online-only) journal owned by the New Phytologist Trust, a not-for-profit organization dedicated to the promotion of plant science, facilitating projects from symposia to free access for our Tansley reviews. Regular papers, Letters, Research reviews, Rapid reports and both Modelling/Theory and Methods papers are encouraged. We are committed to rapid processing, from online submission through to publication 'as ready' via Early View -our average time to decision is <28 days. There are no page or colour charges and a PDF version will be provided for each article.
The journal is available online at Wiley Online Library. Visit www.newphytologist.com to search the articles and register for table of contents email alerts.
If you have any questions, do get in touch with Central Office (np-centraloffice@lancaster.ac.uk)