The evolution of the duckweed ionome mirrors losses in structural complexity

Abstract Background and Aims The duckweeds (Lemnaceae) consist of 36 species exhibiting impressive phenotypic variation, including the progressive evolutionary loss of a fundamental plant organ, the root. Loss of roots and reduction of vascular tissues in recently derived taxa occur in concert with genome expansions of ≤14-fold. Given the paired loss of roots and reduction in structural complexity in derived taxa, we focus on the evolution of the ionome (whole-plant elemental contents) in the context of these fundamental changes in body plan. We expect that progressive vestigiality and eventual loss of roots might have both adaptive and maladaptive consequences that are hitherto unknown. Methods We quantified the ionomes of 34 accessions in 21 species across all duckweed genera, spanning 70 Myr in this rapidly cycling plant (doubling times are as rapid as ~24 h). We related both micro- and macroevolutionary ionome contrasts to body plan remodelling and showed nimble microevolutionary shifts in elemental accumulation and exclusion in novel accessions. Key Results We observed a robust directional trend in calcium and magnesium levels, decreasing from the ancestral representative Spirodela genus towards the derived rootless Wolffia, with the latter also accumulating cadmium. We also identified abundant within-species variation and hyperaccumulators of specific elements, with this extensive variation at the fine (as opposed to broad) scale. Conclusions These data underscore the impact of root loss and reveal the very fine scale of microevolutionary variation in hyperaccumulation and exclusion of a wide range of elements. Broadly, they might point to trade-offs not well recognized in ionomes.


INTRODUCTION
The duckweeds (Lemnaceae) consist of 36 species exhibiting broad variation, including, in recently derived species, the progressive evolutionary loss of a fundamental plant organ, the root.This progressive loss of roots is accompanied by an overall reduction in vascular tissues in derived taxa.Given the paired loss of roots and reduction in structural complexity, we focus here on the evolution of the ionome and place it in the context of these fundamental changes in body plan.
In contrast to vascular land plants, duckweeds have miniscule bodies in direct contact with water and limited to non-existent root systems.This results in small distances for ion translocation (Zhang et al., 2009).However, the relative differences in translocation distance can be large: frond sizes of Spirodela are >1 cm, but in Wolffia only <1 mm.Duckweed roots are considered adventitious, lacking lateral roots and root hairs (An et al., 2019).Root-forming species have flexibility in their root systems, which can develop or elongate in stressful situations or drop off (Landolt, 1986).Root functions in anchorage, aggregation to form duckweed mats and aiding dispersal by attachment have all been proposed (Cross, 2017;Ware et al., 2023).In the highly derived Wolffioideae, the shrinking of body size and complete root loss have evolved to maximize growth rate, improve mobility and enhance adaptability to changing environments (Wang et al., 2010;Michael et al., 2020;Yang et al., 2021).We expect that duckweeds, representing this unique example of progressive root reduction through to complete loss, will illustrate a gradient of phenotypic changes resulting in altered internal macronutrient and trace element compositions (Ware et al., 2023).
At the fine scale, duckweed habitats differ in their availability of elements; thus, adaptation of accessions to their environments can occur through different elemental storage and exclusion strategies (Mkandawire and Dudel, 2007;Zhang et al., 2009;Van Dam et al., 2010;Lahive et al., 2011).The tolerance of duckweed to elemental extremes is an important trait driving adaptive (and sometimes strongly invasive) strategies in the wild (Wang, 1991;Naumann et al., 2007;Ekperusi et al., 2019).To date, however, the tolerance of only a handful of duckweed accessions to external elemental concentrations has been assessed, with reports focusing on growth vigour vis-à-vis single elements in Lemna and Landoltia species.Studies quantifying elemental composition are rare, with the broadest study looking at only a single genus, Wolffia, with 11 species being assessed (Appenroth et al., 2018).We collected existing reports of duckweed elemental variation; however, serious confounding factors plague interpretation of different studies, owing to discordant methods and quantification (Table 1).
Here, we bridge this gap, reporting whole-plant ionome compositions in 34 duckweed accessions spanning 21 species and representing the worldwide range of all five duckweed genera (Fig. 2; Supplementary Data Table S1).We place these data into an evolutionary context, focusing on 11 key macro-, microand trace elements, contrasting microevolutionary variation (accession-level, within-species variation) with macroevolutionary trends (between genera).These results reveal extensive ionomic variation at both the within-species and between-genus levels, with particularly clear trends for differences in Ca and Mg accumulation, in addition to possible excess Cd accumulation in the rootless Wolffia/Wolffiella.We discern a broad evolutionary trajectory towards very low levels of essential Ca and Mg, in addition to increased Cd accumulation, in the recently derived rootless species.This suggests a potentially deleterious consequence associated with the root loss and body-wide reduction in vasculature.

Plant growth and care
Duckweed accessions were grown in axenic conditions from single isolates or from five to ten individuals, depending on the size of duckweeds, in 100 mL of nutrient medium (N medium) in individual sealed sterile glass conical flasks.Duckweeds were sourced from the Landolt Collection (now housed in Milan).The N medium was described by Appenroth et al. (1996) [KH 2 PO 4 , 0.15 mm; Ca(NO 3 ) 2 , 1 mm; KNO 3 , 8 mm; MgSO 4 , 1 mm; H 3 BO 3 , 5 µm; MnCl 2 , 13 µm; Na 2 MoO 4 , 0.4 µm; and FeEDTA, 25 µm].Concentrations of elements in the supplied N medium, including the presence of other trace elements, were measured by inductively coupled plasma mass spectrometry (ICP-MS) and are presented in the Supplementary Data (Dataset S1).Weekly media changes were performed, with rinses in Milli-Q (Millipore) water to regulate nutrient composition availability.Plants were grown at 100 µmol m −2 s −1 under broad-spectrum (white) LED lights at 22 °C/18 °C with a 16 h day/night cycle.Four-week-old duckweed cultures were washed on plastic sieves using a three-step protocol for 2 min each of Milli-Q (Millipore)  water, CaCl 2 and Na-EDTA and harvested into individual samples from flasks of individual populations.These were harvested for ICP-MS analysis on day 1, 3 and 5 after media change, n = 6 per time point.Four-week-old cultures are clonally reproduced and therefore suitable replicates, given the very low generational variation and low mutation rates shown in duckweed mutation accumulation experiments (Xu et al., 2019).

Imaging and microscopy
All samples were cleared, then stained with Fluorescent Brightener 28 (calcofluor) following the protocol described by Kurihara et al. (2015) and imaged on a Leica TCS SP5 confocal microscope.In short, plants were cleared, based on the ClearSee procedure described by Kurihara et al. (2015), with slight modification.Given that fluorescent markers were not being used, plants were fixed overnight in ethanol and acetic acid (3:1 v/v) rather than paraformaldehyde, because this reduced the toxicity and requirement for vacuum infiltration, which can be damaging to the air spaces.Plants were then rinsed three times with reverse osmosis water and left for 30 min, after which the reverse osmosis water was replaced with ClearSee solution (10 % xylitol, 15 % sodium deoxycholate and 25 % urea; Kurihara et al., 2015) and left to clear for 2 weeks.Before imaging, plants were stained for 1 h with calcofluor in ClearSee (100 μg mL −1 ), then washed in ClearSee for 1 h.Imaging was carried out using a confocal laser scanning microscope (Leica SP5), using a 405 nm diode laser at 12 % and hybrid detector with a range of 440-450 nm, gain of 25 % and pinhole of 0.5 Airy units.  ,10,11,12,13,1 27.3

Quantification of elemental tissue concentrations
For ICP-MS, we used a method adapted from the study by Danku et al. (2013).Briefly, 5-20 mg (fresh weight) was harvested per sample, placed in Pyrex test tubes and dried at 88 °C for 24 h.The dry weight was recorded, then 1 mL concentrated trace metal grade nitric acid Primar Plus (Fisher Chemicals) spiked with an internal standard was added to the samples, which were digested further in DigiPREP MS dry block heaters (SCP Science; QMX Laboratories) for 4 h at 115 °C.Before the digestion, 20 µg L −1 of indium (In) was added to the nitric acid as an internal standard for assessing errors in dilution, variations in sample introduction and plasma stability in the ICP-MS instrument.Then 0.5 mL of hydrogen peroxide (Primar, for trace metal analysis, Fisher Chemicals) was added to the samples and they were digested for additional 1.5 h at 115 °C.After digestion, samples and blanks were diluted to 10 mL with Milli-Q (Millipore).Direct water and elemental analysis was performed using an ICP-MS, PerkinElmer NexION 2000, with 22 elements monitored (Li, B, Na, Mg, P, S, K, Ca, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Rb, Sr, Mo, Cd and Pb) in the collision mode (He).To correct for variation between and within ICP-MS analysis runs, liquid reference material was prepared using pooled digested samples and run after every nine samples in all ICP-MS sample sets.The calibration standards were prepared from single element standard solutions (Inorganic Ventures; Essex Scientific Laboratory Supplies Ltd, Essex, UK).Sample concentrations were calculated using an external calibration method within the instrument software.Further data processing, including calculation of final elements concentrations (in milligrams per kilogram), was performed in Microsoft Excel.Log 10 -transformations, z-score calculations and graphical representation were performed using R (v.3.0.2 'Frisbee Sailing'; R Development Core Team, 2023; see http://www.R-project.org), and RStudio v.1.0.136 (RStudio Team, 2020) was used for all statistical analyses.To calculate relationships between elements, the corrplot package (McKenna et al., 2016) was used in R with Pearson correlations on log 10 -transformed data.

Broad scale evolution of the ionome
We focus on ionomes from day 5 after media change (Fig. 3), which is representative of other time points (none of the 11 elements upon which we focus was significantly different across days by ANOVA).The full raw dataset is given in the Supplementary Data (Dataset S2); elements we considered for further analysis are shown in the Supplementary Data (Fig. S1).Concentrations were consistent for all elements for all accessions between time points except for a handful of elements in certain accessions depicted in the Supplementary Data (Fig. S2).These exceptions show a small minority of accessions decreasing in K, Ca, Fe and Cd and others still increasing (e.g.Ca, Cu and Fe).For accumulators showing the latter pattern, such as Spirodela intermedia 9227, the maximum concentration capacity of Ca on day 1 after media changes was still not reached, despite high nutrient provision throughout a 4-week experimental period, and the accession could still prolong uptake.
In the overall dataset of 34 accessions, the broadest contrast observed was between the Lemnoideae and Wolffioideae (rooted and rootless, respectively) for Ca, Mg and Cd accumulation (Fig. 3A).All ancestral representatives of (rooted) Lemnoideae (Spirodela, Landoltia and Lemna) consistently exhibited two to three times higher Ca content relative to the derived rootless Wolffioideae (P ≤ 0.01; log 10 , ANOVA with Tukey's post-hoc test).Likewise, on average, Mg accumulation was 1.8 times higher in the rooted species relative to the rootless Wolffia and Wolffiella.Ca and Mg showed a positive correlation (Table 2; Supplementary Data Figs S3 and S4).We observed further variation for Mg in the Lemna genus, where there emerged a gradient of Mg accumulation across Lemna sections (Figs 1 and 3A, D).The highest Mg levels were in the Uninerves section (Figs 3A  and 4), which includes the invasive Lemna minuta and Lemna yungensis (now Lemna valdiviana), as described by Tippery et al. (2015) and Bog et al. (2020), both alien within Europe (Kirjakov and Velichkova, 2016;Ceschin et al., 2018).This association of Mg accumulation with increased root vasculature (and with reduced frond vasculature in Lemna) stood in strong contrast to the uniformly very low Mg in rootless Wolffioideae.Cadmium concentrations varied significantly between rooted and non-rooted duckweeds (Fig. 3A; P < 0.05; log 10 , ANOVA with Tukey's post-hoc test) in a manner inverse to Ca and Mg.The unrooted Wolffioideae species (especially Wolffiella) showed the highest Cd concentrations.Only the submerged Lemna trisulca exhibited Cd comparably high to the Wolffioideae (Fig. 3).
Rootless species exhibiting variation in at least two elements included Wolffiella lingulata, Wolffiella hyalina and Wolffia brasiliensis (Fig. 3E).In contrast, the species in our panel from the multi-rooted, more ancestral duckweed representatives, Spirodela and Landoltia, showed the greatest ionomic consistency across all accessions (Fig. 3B).Spirodela species had the highest tissue content of Ca in our panel, but other elements were not as variable between accessions.

Fine-scale ionome variation and identification of extreme accumulators in Lemna
We observed the greatest within-genus ionome variation in the Lemna genus (n = 20 accessions, six biological replicates of each; Fig. 3C, D).Lemna also harboured several extreme accumulators, each standing as outliers for the accumulation of three or more elements.Lemna trisulca 7192 has a submerged growth pattern and accumulated the greatest number of elements in amount and number from the panel, showing very high tissue concentrations of four essential elements (P, Ca, Zn and Fe), in addition to Cd, and low K levels (Fig. 3D).Lemna yungensis 9210 accumulated high S and Mn and also exhibited low K (Fig. 3C).The K levels trended negatively against the enhanced accumulation of other macroand microelements in both Le.trisulca and Le.yungensis and across our panel as a whole (Table 2; Supplementary Data Fig.S4).

Fine-scale ionome variation between Lemna species
We noted variation at the level of several accession pairs, most obviously between Le. yungensis accessions (Fig. 3C).
Notably, Le. yungensis 9208 greatly accumulated Mg, and Le.yungensis 9210 exhibited extreme accumulation of S and Mn, but low K.When comparing Le. yungensis with Lemna valdiviana clones, none of the accessions showed large differences in ionomes between ten elements, with consistent levels of B and S (Fig. 5A).Comparing Lemna minor with Lemna turionifera and their interspecific hybrid Lemna japonica, Le. japonica accessions had lower Mo and a slight increase in Na and K in specific Le.japonica clones (Fig. 5B); however, neither of these ionome changes was significant in comparison to the whole duckweed panel.clone-level variation in some elements, but none varied significantly from the overall population by as much as one SD (Fig. 5C).

DISCUSSION
The broad variation we observed in duckweed ionomes at levels of genera, species and sister accessions is presumably attributable, in large part, to both morphological differences and adaptation to micro-environments.The most robust differences were at the genus level for Ca, Mg and Cd.The accumulation difference for Ca is perhaps explained, in part, by a storage mechanism as calcium oxalate (CaOx) within frond crystal ultrastructures in rooted genera, in the fronds of Spirodela and Lemna (Landolt and Kandeler, 1987) and in the root of Le. minor (Franceschi, 1989;Mazen et al., 2003).In Le. turionifera, Ca influxes through roots and is stored in both fronds and roots, and in exceptional cases it can also be effluxed out of roots (Ren et al., 2022).In contrast, Wolffioideae species have soluble Ca in cell sap and accordingly also cannot store excess Ca in the roots (Landolt and Kandeler, 1987;Appenroth et al., 2017); thus Ca and Mg might be lower in Wolffiodeae because they lack roots as a storage organ.Given that Ca was kept sufficiently available in our experiment through media refreshes, and rooted duckweeds use their roots as an additional storage compartment (Ren et al., 2022), this might result in overall higher accumulation when compared with their rootless counterparts.
Given the broad contrasts in Ca between genera, it is interesting to consider these results alongside the importance of roots for elemental uptake and segregation of individual elements between the frond and root in duckweed species.The excision of roots makes only a modest change to the frond ionome, showing that roots are vestigial and overall not required for nutrient uptake in replete media conditions (Ware et al., 2023).This supports the notion that duckweed roots might be adventitious (Landolt, 1986;An et al., 2019).Although, surprisingly, removal of roots increased elemental composition in some cases (Ware et al., 2023), the picture is more complicated, in that rootless species do not naturally exhibit elevated Mg or Ca in our data, indicating evolutionary adjustment of ion homeostasis upon root loss.The Wolffia genome harbours a derived complement of Ca export and cell wall-thickening genes, possibly minimizing potential for apoplastic transport, which, coupled with inability for storage as CaOx, results in less specialized mechanisms to manoeuvre and store Ca content overall (Michael et al., 2020).In contrast, clones of Le. aequinoctialis, Le. minuta and Le.minor exhibit marked Ca accumulation (storage) to alleviate Mg toxicity from a contaminated mine and in high Mg:Ca ratio media or wastewater (Van Dam et al., 2010;Paolacci et al., 2016;Walsh et al., 2020).This suggests specific adaptation of Ca storage and transport mechanisms to particular ionomic challenges.
The Mg gradient across Lemna species is not necessarily correlated with strict overall inferred ancestral and derived forms (Wang et al., 2011;Tippery et al., 2015) and root vascular complexity is not sufficiently varied between rooted duckweeds to account for this (Ware et al., 2023).Instead, higher specific Mg uptake in the Uninerves section of Lemna might be associated with their reduced frond vascular complexity (Figs 3A and 4).With typical frond nerves numbering ≤16 in in Spirodela and between three and seven in other Lemna species (Les et al., 2002), only one nerve is present in Le. yungensis and Le.minuta, with Le. yungensis (now Le. valdiviana) having the longer nerve of the two (Landolt, 1980;Crawford et al., 1996).It is thought that this simplified vascular system might contribute to their invasive status (Kirjakov and Velichkova, 2016;Kadono and Iida, 2022).Reduced vascular complexity and ionomic differences could also offer enhanced potential for adaptation to varied environments, showing higher Mg tolerance (Paolacci et al., 2016) and possibly, therefore, survival in hard water.
Although some variation in mineral content among Wolffia species has been reported by Appenroth et al. (2018), Wolffiella  have received little attention and can be under-reported owing to clones having restricted biogeography and not being readily available (Landolt, 1986;Kimball et al., 2003).Therefore, multi-elemental compositions of rooted and rootless duckweeds have not been compared directly before.In this respect, we see relative accumulation of Cd, especially in Wolffiella compared with the rooted species.This is somewhat surprising, because it might be expected that Cd accumulation would be detrimental to minuscule plants with no root segregation away from photosynthetically active tissue.We note, however, that Wolffia species also exhibit tolerance to As and have been considered as candidates for phytoremediation, accumulating more than Lemnoideae (Zhang et al., 2009).Additionally, there is good evidence that Wolffia has moderate tolerance to Cd and increased accumulation capacity even in extreme concentrations (>200 µm).In fact, a handful of Wolffia species show Cd uptake in as little as 30 min from solution via apoplastic transport, which increases linearly with Cd concentration (Boonyapookana et al., 2002;Xie et al., 2013).We therefore speculate that loss of roots could have reduced control of heavy metal uptake whilst, at the same time, root loss removes a potential mechanism of uptake and a storage compartment available to rooted species (Verma and Suthar, 2015;Ma et al., 2023;Zheng et al., 2023).Wolffioideae perhaps evolved higher tolerance mechanisms to Cd toxicity, such as compartmentalization to vacuoles and complexation via conjugates (Schreinemakers, 1986).Although Cd was not supplied in a dedicated quantity in N medium preparation, we quantified the presence of Cd by ICP-MS in the media used (Supplementary Data Dataset S1) and suggest that this comes from chemical impurities, as indicated by Appenroth et al. (2018).We infer that Wolffioideae species might have a potential for heavy metal accumulation at higher dosages than those given here, perhaps also in the wild through adaptation to contaminated habitats (Zhang et al., 2009).
Our results showed that the genus with the greatest diversity of specific accumulators was Lemna.The Lemna accessions with most extreme ionomes, Le. trisulca 7192 and Le.yungensis 9208, also harbour the most divergent root architecture, in comparison to other species of Lemna.Lemna trisulca is characterized by a submerged growth habit but smaller cortical cells, giving a thin, reduced root compared with other Lemna species, and Le.yungensis 9208 often displays an additional layer of cortical cells and irregularly large extracellular airspaces in the root cortex (Ware et al., 2023).
Thus, these differential root vasculature components, coupled with minimal frond vasculature, might play a role in producing the contrasting elemental profiles observed.Both Le. trisulca and Le.yungensis accumulated >1000 mg kg −1 dry weight for several elements and can therefore be considered hyperaccumulators (Zayed et al., 1998;Zhang et al., 2009).For this reason, these two species might have potential to be used in combination to alleviate multi-elemental toxicity in watercourses.Lemna trisulca accumulated greater Zn and Cd than floating species, possibly because of increased absorption through submerged fronds.Although Le. trisulca had the greatest variation overall and maximal micronutrient levels, the associated high Cd accumulation might be problematic for any applications in nutrition.It is also unclear whether this trait is common in other Le.trisulca accessions owing to limited availability of clones in stock centres; however, this species has previously been noted for its Cd accumulation potential (Kara and Kara, 2005).
A greater appreciation for duckweed variation in the micronutrients Ca, Mg, Fe and Zn is clear from our study, with particular accessions acting as hyperaccumulators for multiple nutritionally relevant elements.This is not the case for trace elements, such as Na and Cu (and especially Mn and the heavy metal Cd), for which the variation in tissue concentration was less dramatic than seen in other reports (Table 1).This is probably attributable to the combined effect of low presence of these elements in our supplied media or that comparisons across literature are confounded by variables disallowing truly quantitative comparisons between studies.This is particularly evident for Cd, which we supplied in only trace amounts (Supplementary Data Dataset S1), whereas external Cd concentrations vary 500-fold between studies.
Synthetic biology, including the tailoring of ionomic profiles in duckweeds, is an important goal of the duckweed research community (Lam and Michael, 2022).Interestingly, the Spirodela genome sizes are the smallest and the ionomes the least variable among all duckweeds here (Wang et al., 2011;An et al., 2018); additionally, the amenability of Spirodela to genetic transformation (Yang et al., 2018a, b) makes it a strong candidate as a minimal scaffold for synthetic biology.We also suggest that because their ionomic profiles are so variable, the species harbouring larger genomes will be particularly valuable to mine natural variation to inform transgenic approaches in the smaller, highly tractable Spirodela genome.For the fine-scale variation between Lemna species of interest, the vast ionome differences between Le. yungensis 9208 and 9210 can be ascribed best to local adaptation.Given that these accessions are closely related and were both originally isolated from the same region in Bolivia, one might expect more similar ionome profiles, but instead our data show that duckweeds exhibit strongly contrasting local variation in elemental uptake.Interestingly, this region of Bolivia is reported to be atypically harsh for duckweed, growing on sheer rock faces with waterfall spray with low nutrient availability (Landolt, 1998).It will be valuable to characterize Le. yungensis species further, in order to determine the genetic basis for their adaptation to specialized habitats.Given that Le. yungensis and Le.valdiviana showed no other significant internal differences between ten elements, this supports their unification as one species owing to lack of genetic differentiation (Bog et al., 2020).Lemna minuta is an invasive species in introduced regions with ecological significance (Ceschin et al., 2018), as an opportunist species in replete N and P with additional higher Mg tolerance (Njambuya et al., 2011;Paolacci et al., 2016;Ceschin et al., 2020) one would expect drastic differences in the ionome in comparison to Le. minor.Despite this, there were no clear pattern differentiating two Le.minuta from two Le.minor clones grown in nutrient-rich medium (N medium; Appenroth et al., 1996; measured here in Supplementary Data Dataset S1).Elemental differences seem to be at the clonal level, and opportunism therefore probably depends on unique situations in the wild.Recent data classified Le. japonica as a hybrid between Le. minor and Le.turionifera (Braglia et al., 2021;Volkova et al., 2023).Hybrid Lemna japonica clones had slightly reduced Mo compared with their parents, and one clone had significantly higher Na.It could be that hybridization might result in ionome differences important for altered adaptation to varied environments, as found in other plant species (Arnold et al., 2016;Wong et al., 2022).Taken together, between these groups of Lemna species, subtle interspecies differences for elements were clear.The physiological differences between species and their clones in light of genetic differences deserve future attention in duckweed.

Conclusions
Here, we detailed broad-and fine-scale diversity for the accumulation of physiologically and nutritionally important elements across all five duckweed genera.This variation is associated with dramatic morphological reductions in fundamental plant organs and genome expansions.Thus, disentangling the concurrent effects of dramatic genome size expansions, organ reduction and ecological adaptations will be a great challenge.However, at the more microevolutionary scale, within-species, accession-level variation points to clear promise in mapping alleles responsible for this observed variation.
One might speculate that the observed ionomic changes might be a maladaptive spandrel associated with root loss in derived taxa, but it is hard at this point to identify what the exact trade-off might be; this is for dedicated mechanistic and ecological work on the rootless taxa.Beyond highlighting these enigmatic correlates of root loss and the consequences of organ loss and vestigiality, this work serves to establish phenotypic variation across the ionome at both the fine and broad scale.This serves as a basis for future genomic characterization of causal alleles, in addition to rational development of targeted duckweed lines for both important nutritional and phytoremediation goals.
SUPPLEMENTARY DATA Supplementary data are available at Annals of Botany online and consist of the following.
Figure S1: raw elemental composition of duckweed whole plants between days 1, 3 and 5 following media change by ICP-MS.Figure S2: outlier accessions with dynamic elemental concentrations over sampling days 1, 3 and 5 after media change.Figure S3: principal component analysis for 11 plant macro-and micronutrients and heavy metals.

Fig. 3 .
Fig.3.The evolution of the duckweed ionome across genera, species and accessions.(A) Relative levels of elemental accumulation across rootless and rooted subgroups, respectively.The heat map is coloured by z-scores for the four most differentially accumulated elements.Significant differences were determind by ANOVA with Tukey's post-hoc test set at **P < 0.01 and *P < 0.05 between Wolffioideae and Lemnoideae.The z-scores (number of standard deviations away from the mean) were generated for each element using log 10 -transformation of values (in milligrams per kilogram) on day 5.The x-axis is arranged with basal forms on the left and derived forms on the right.Separating lines indicate genus and subgroup boundaries.We. = Wolffiella (2), Wo. = Wolffia (5), Le. = Lemna (20), La. = Landoltia (2) and Sp.= Spirodela (5).Within Lemna, sections Biformes, Alatae, Uninerves and Lemna are marked from left to right.(B-E) Radar plots showing differences in ionome profiles between individual accessions: (B) Spirodela and Landoltia; (C) Lemna sections Biformes, Alatae and Uninerves; (D) Lemna section Lemna; and (E) Wolffiella and Wolffia species.Species are ordered in the panels according toTippery et al. (2015), from the most ancestral representative at the top left to the most derived at the bottom right.Numbers after species represent clone numbers.Asterisks represent a significant increase or decrease of ±2 relative to all normalized element concentrations for all species based on the mean and SD.The complete dataset of 17 elements and three time points is given in the Supplementary Data (Dataset S2).

Fig. 4 .
Fig.4.Increased Mg content mirrors the reduction of frond vasculature within Lemna.The four sections of Lemna represent the highest Mg content in the species with most reduced vasculature for section Uninerves, with transitional sections Biformes and Alatae and the most developed frond vasculature in section Lemna, with reduced Mg.The Mg content is plotted from day 5 averaged values for each accession within each section: Uninerves, n = 6; Biformes, n = 2; Alatae, n = 2; and Lemna, n = 10.Sections are ordered and described according toLandolt (1986) andTippery et al. (2015).Violin plots represent the spread of data for each group, with the middle line plotting the mean.

Fig. 5 .
Fig. 5. Elements high in N medium show limited differences in internal ionomes between pairs of Lemna species.(A) Lemna yungensis (now merged with Lemna valdiviana) and Le.valdiviana accessions.(B) Lemna minor, Lemna turionifera and their interspecific hybrid species, Lemna japonica.(C) Accessions of cosmopolitan Lemna minor and invasive European alien Lemna minuta.Heat maps for z-scores from day 5 are presented for each accession.Ten elements were selected based on those intentionally added and present in the highest concentrations in N medium.The z-scores ± 2SD represent a significant increase or decrease relative to all normalized elements.
Figure S4: intensity and direction of correlations between eight elements in 34 duckweed accessions.Table S1: accessions studied in this work, with Landolt codes and locations.Dataset S1: summary elements present in N medium, as measured by ICP-MS.Dataset S2: all ionomics data (in milligrams per kilogram) for 22 elements for 34 accessions on days 1, 3 and 5 post media change quantified by ICP-MS.FUNDING K.E.S. is supported by a Biotechnological and Biological Sciences Research Council (BBSRC) PhD scholarship (BB/ M008770/1).L.Y. was supported by the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme [grant number ERC-StG 679056 HOTSPOT].This work was also supported by the University of Nottingham's Future Food Beacon of Excellence.

Table 1 .
Elemental tissue concentration of duckweeds gathered from the literature.Elements are ordered by type (macro, micro, trace elements and heavy metals) reported from the literature and included in our experiment.

Table 2 .
Mg and Ca were correlated strongly and positively with various elements, whereas K was negatively correlated.Element pairs were significantly correlated across 34 duckweeds at three time points.The R values correspond to positive or negative Pearson correlations derived from log 10 -transformed data for eight elements.Data are given to two decimal places.