Abstract
The exceptionally high growth rate and high flavonoid content make the giant duckweed Spirodela polyrhiza (L.) Schleid. (Arales: Lemnaceae Martinov) an ideal organism for food production and metabolic engineering. To facilitate this, identification of the genetic basis underlying growth and metabolic traits is essential.
Here, we analysed growth and content of 42 metabolites in 137 S. polyrhiza genotypes and characterized the genetics underpinning these traits using a genome‐wide association (GWA) approach.
We found that biomass positively correlated with the content of many free amino acids, including L‐glutamine, L‐tryptophan, and L‐serine, but negatively correlated with specialized metabolites, such as flavonoids. GWA analysis showed that several candidate genes involved in processes such as photosynthesis, protein degradation, and organ development were jointly associated with multiple metabolic traits.
The results suggest the above genes are suitable targets for simultaneous optimization of duckweed growth and metabolite levels. This study provides insights into the metabolic diversity of S. polyrhiza and its underlying genetic architecture, paving the way for industrial applications of this plant via targeted breeding or genetic engineering.
Keywords: duckweed metabolism, mGWAS, Spirodela polyrhiza, targeted metabolomics, trait variation
A genome‐wide association study in duckweed revealed candidate genes associated with plant growth, photosynthesis, starch metabolism, and protein turnover.

INTRODUCTION
With a biomass duplication rate of less than three days, Spirodela polyrhiza (L.) Schleid. is one of the fastest‐growing angiosperms (Ziegler et al., 2015). Due to its high levels of flavonoids and amino acids, S. polyrhiza is one of the best organisms for producing food resources and pharmaceuticals (Zhao et al., 2015; Acosta et al., 2021; Baek et al., 2021; Böttner et al., 2021; Smith et al., 2024). To fully realize its industrial potential, specific improvements in its growth and metabolite production are needed, which require detailed understanding of the genetic mechanisms controlling these traits. Although previous forward genetic studies on metabolic traits in species like Zea mays L. (Chen et al., 2016), Oryza sativa L. (Chen et al., 2014; Zhang et al., 2019; Cu et al., 2021) and Arabidopsis thaliana (L.) Heynh. (Angelovici et al., 2013; 2017) identified many promising candidate genes, the genetic principles controlling growth and metabolism in S. polyrhiza largely remain unknown, hampering its further biotechnological optimization.
For most plants, to survive natural stress factors, such as herbivory, they must carefully allocate resources to growth or defence pathways. On the molecular level, free amino acids play a pivotal role in balancing resource distribution between biomass production and synthesis of defence metabolites. Biosynthesis of amino acids can account for 50% of all carbon compound syntheses and consumes 32% of total fixed carbon dioxide (Smith et al., 1961), making them a major carbon sink for biomass production in plants (Noctor & Foyer, 1998). The central metabolic function of free amino acids is highlighted by their association with key metabolic pathways, such as glycolysis, tricarboxylic acid (TCA) cycle, pentose phosphate pathway, urea cycle and photorespiration (Fig. 1) (Noctor & Foyer, 1998). Besides their function in biomass gain, amino acids are precursors of many specialized metabolites like flavonoids and phenolic acids (Fig. 1), which are involved in plant defence. Biosynthesis of specialized metabolites often negatively impacts plant biomass, primarily through consumption of nutrient resources (Züst et al., 2011) and autotoxicity (Dick et al., 2012; Li et al., 2021). Consequently, optimization for high crop yields often leads to high susceptibilities to pests due to a lack of defence metabolites, a phenomenon known as the growth–defence trade‐off (Huot et al., 2014; Züst & Agrawal, 2017). On a regulatory level, resource allocation from growth to defence pathways is controlled by phytohormones such as jasmonic acid (JA), abscisic acid (ABA), salicylic acid (SA) and indole‐3‐acetic acid (IAA) (Aftab et al., 2010; Huot et al., 2014; Živanović et al., 2020; Hui et al., 2023). Therefore, understanding the genetic mechanisms controlling levels of phytohormones, free amino acids and related metabolites is key for optimizing the biomass yield and specialized metabolite content in plants.
Fig. 1.

Overview of the biosynthesis of metabolites analysed in this study. Metabolites are connected to their biochemical precursors with arrows. All quantified compounds were sorted into three categories: amino acid metabolism, specialized metabolism, and phytohormones, as shown in boxes coloured red, blue and green, respectively. Direct degradation products and biosynthetic precursors of amino acids such as tyramine and shikimic acid are grouped together with all quantified amino acids in the group amino acid metabolism. Amino acids were subcategorized into five families based on their last common precursor (serine, pyruvate, aspartate and glutamate) or on common chemical properties (aromatic amino acids). All specialized metabolites quantified in this study are phenylpropanoids. Phenolic acids are also precursors of flavonoids. In contrast to amino acids and specialized metabolites, compounds with a major function in signalling were characterized as phytohormones. Chemically derived metabolites with the same precursors were further grouped in biosynthetic families, illustrated by frames of different colours. 3‐PGA = 3‐phosphoglycerate; ABA = abscisic acid; CoA = coenzyme A; cZ = cis‐zeatin; cZR = cis‐zeatin riboside; DAHP = 3‐deoxyarabinoheptulonate‐7‐phosphate; E4P = erythrose‐4‐phosphate; IAA = indole‐3‐acetic acid; iP = isopentenyladenine; iPR = isopentenyladenine riboside; JA = jasmonic acid; JA‐Ile = jasmonic acid‐isoleucine conjugate; PEP = phosphoenolpyruvate; R5P = ribose‐5‐phosphate; SA = salicylic acid; TCA = tricarboxylic acid; tZ = trans‐zeatin; tZR = trans‐zeatin riboside.
Here, we focus on understanding the genetic control regulating levels of free amino acids and their related specialized metabolites in giant duckweed. We quantified content of 42 metabolites (Fig. 1) in 137 genotypes of S. polyrhiza and performed a genome‐wide association study (GWAS) on the content of these metabolites. We specifically aim to address the following questions: (i) to what extent do growth and metabolism vary in S. polyrhiza; (ii) which metabolites are associated with the intra‐specific variations of biomass and growth in S. polyrhiza; and (iii) which genes control these metabolic traits?
MATERIAL AND METHODS
Data collection and sample preparation
In a previous study we estimated the growth of 138 S. polyrhiza genotypes under herbicide treatment and under control conditions (Höfer et al., 2024b). We quantified growth as relative growth rate of frond area (RGR of FA) and frond number (RGR of FN) and measured biomass as fresh weight (FW) and dry weight (DW) at the end of a 7‐day growth period. For each genotype, growth and biomass were estimated in three technical replicates. The quantified growth and biomass parameters are publicly available (Höfer et al., 2024a). A table documenting geographic origin of all genotypes analysed in this study can be found at https://datadryad.org/stash/downloads/file_stream/2894056.
All analyses included in this study are based on plant material grown under control conditions (Höfer et al., 2024b). After harvesting, we freeze‐dried plant material and stored it at room temperature (RT) until metabolite extraction. To have sufficient plant material for metabolite analysis, triplicates were pooled for each genotype. The RGRs were estimated (Höfer et al., 2024b) based on a published method (Ziegler et al., 2015), applying the formula: , where Nt is the frond number or frond area at the endpoint of the assay, N 0 is the frond number or frond area at the start of cultivation and t is the duration of the bioassay.
Metabolite extraction
To investigate the genotype variation in metabolite content via liquid chromatography ‐ mass spectrometry (LC‐MS), we extracted 42 metabolites from pooled triplicates of 137 genotypes (Table S1, Fig. 1). The number of analysed genotypes was reduced from 138 to 137 due to mislabelling of one genotype, which was excluded from all analyses. The protocols used for LC–MS analysis of metabolite levels are described in Table S1–S5. The entire extraction and quantification procedure is based on a previously published method (Schäfer et al., 2016).
The samples were ground with two steel balls at 30 Hz for 30 s in a TissueLyser (Qiagen, Venlo, Netherlands). Next, 10 mg aliquots of the homogenized samples were placed in 96‐well BioTubes (Axygen, New York, NY, USA) and 800 μL extraction buffer (methanol:water:formic acid, 15:4:1 (v/v/v)) containing [13C6]‐IAA (1 ng), [2H5]‐JA (10 ng), [2H6]‐ABA (10 ng), [2H4]‐SA (10 ng), [2H5]‐tZ (0.5 ng), [2H6]‐iP (0.2 ng) and [2H6]‐iPR (0.2 ng) were added to each tube. Tubes were sealed with a pre‐cooled mat and homogenized in a TissueLyser at 25 Hz for 60 s. Samples were incubated at −20 °C overnight then the homogenization step was repeated at 25 Hz for 60 s. We then centrifuged the samples at 2,250 g for 20 min at 4 °C and transferred 600 μL of the supernatant in new BioTubes. The pellet was kept for re‐extraction. We diluted an aliquot of 1 μL supernatant in 99 μL algae amino acid standard mix −13C, 15N (0.5 ng·μL−1) (Sigma‐Aldrich, St. Louis, MO, USA) and transferred the dilutions into 96‐well PCR plates (Kisker, Steinfurt, Germany). The PCR plates were then stored at −20 °C till analysis of all amino acids and several secondary metabolites (Methods S1 and S2, Table S1). The pellet was reextracted with 600 μL extraction buffer without standards, homogenized at 25 Hz for 60 s before incubation at −20 °C for 30 min. After another centrifugation step at 2,250 g for 20 min at 4 °C, we transfered 600 μL of supernatant from the reextracted samples to the remaining supernatant from the first part of the extraction. We centrifuged the combined supernatants at 2,250 g for 20 min at 4 °C. Before loading samples, we conditioned the HR‐X columns (Macherey‐Nagel, Düren, Germany) with 600 μL MeOH, followed by 600 μL extraction buffer and discarded the flow through. We collected the flow‐through of our samples in Nunc 96‐well Deep Well Plates (Thermo Fisher Scientific, Waltham, MA, USA). Next, we added another 200 μL extraction buffer and collected the flow through in the same well plate. We then evaporated the MeOH at 45 °C. Next, we added 850 μL of 1 N formic acid to each sample, mixed the samples at 20 Hz for 30 s using a TissueLyser and centrifuged them at 2,250 g for 20 min at 4 °C.
Before starting the next purification step, we first conditioned HR‐XC columns with 600 μL MeOH and then with 600 μL 1 N formic acid and discarded the flow‐through. We loaded samples on the conditioned HR‐XC columns and subsequently washed the columns with 1.2 mL 1 N formic acid, discarded the flow through and eluted the sample with 1 mL 0.2 N formic acid in 80% (v/v) MeOH. We collected the eluates in BioTubes. After sample homogenization by inversion, we transferred 50 μL supernatant into a 96‐well PCR plate for analysis of phytohormones (Method S3, Table S1). The sealed plates were stored at −20 °C until analysis.
For analysis of IAA and phenolic acids (Method S4), the remaining eluate was evaporated under a constant nitrogen stream at 45 °C and samples were reconstituted to 50 μL with 0.2 N HCOOH in 80% (v/v) MeOH. We covered the plates before mixing at 20 Hz for 30 s. After centrifugation at 2,250 g for 20 min, samples were transferred to a 96‐well PCR plate and stored at −20 °C till analysis. For extraction of cytokinins (Method S5), we washed the HR‐X column twice with 1 mL 0.35 N NH4OH and discarded the flow through. Cytokinins were finally eluted with 1 mL 0.35 N NH4OH in 60% (v/v) methanol. Samples were then dried under constant nitrogen flow before being reconstituted to 50 μL with 0.1% (v/v) acetic acid. The samples were then homogenized at 20 Hz for 30 s, before incubation in an ultrasonic bath for 5 min. Samples were then centrifuged at 2,250 g for 20 min at 4 °C, the supernatant was transferred to a 96‐well PCR plate and stored at −20 °C until analysis (Method S5).
LC–MS analysis of metabolite contents
We quantified 42 metabolites via LC–MS. For separation and quantification, we used a Nexera X3 UHPLC system (Shimadzu, Kyoto, Japan) coupled to an LCMS‐8060 system (Shimadzu, Kyoto, Japan). For separation of all metabolites, we used a Zorbax RRHD Eclipse XDB‐C18 column (50 × 3 mm, 1.8 μm) (Agilent, Santa Clara, CA, USA) with a 1290 Infinity II inline filter (0.3 μm) (Agilent, Santa Clara, CA, USA). For all measurements, the autosampler was pre‐cooled to 5 °C during the entire measurement procedure. As mobile phase, we used 0.05% (v/v) formic acid and 0.1% (v/v) acetonitrile as solvent A and 100% methanol as solvent B, that were applied in gradient mode at a constant flow rate of 0.5 ml·min−1 in all methods. The column oven temperature was always 42 °C. For analyses, we used an electrospray ionization source with the following parameters: nebulizing gas flow: 3 l·min−1, heating gas flow: 10 l·min−1, drying gas flow: 10 l·min−1, interface temperature: 300 °C, DL temperature: 250 °C, heat block temperature: 400 °C and CID gas flow: 270 kPa. For metabolites measured in positive and negative ionization modes, we applied interface voltages of 4000 V and −3000V, respectively (Table S2–S6). A detailed description of individual metabolite analysis can be found in Methods S1–S5.
Correlation and principal components analysis (PCA)
Next, we used regression and PCA to identify correlation patterns between metabolic parameters and to identify influences of population structure on metabolite levels and growth. We established multiple correlations for all metabolite and growth data using the ggcorrplot R‐package (v 0.1.4.1) (Kassambara, 2023). We used Pearson correlation coefficient to measure strength of linear correlations. All PCAs were conducted with the R package pcaMethods (v 1.88.0) (Stacklies et al., 2007). We calculated 95% confidence intervals based on standard deviations using the vegan package (v 2.6‐2) (Oksanen et al., 2022) for all PCAs. All three‐dimensional scatter plots were created using the R package scatterplot3d (v 0.3.42) (Ligges & Mächler, 2003). To identify metabolite categories, we sorted all 42 metabolites into nine groups based on their chemical characteristics and biochemical precursors: flavonoids, phenolic acids, aromatic amino acids, glutamate family, jasmonates, serine family, pyruvate family, aspartate family, and cytokinins (Fig. 1). A PCA on metabolite concentrations was subsequently done for all 137 genotypes.
Since our lab has identified four genetic populations for S. polyrhiza (Xu et al., 2019; Wang et al., 2024), we next investigated to what extent differences in metabolite concentrations and growth are explained by population structures. Assignment of genotypes to different genetic populations was done previously. Due to the high proportion of clonal genotypes in our accession, 137 genotypes were grouped into 97 clonal families for PCA on population structure of growth parameters and metabolite levels, respectively. For analysis, each clonal family was represented by the genotype with the highest sequencing coverage, as described previously (Höfer et al., 2024b).
Genome‐wide association study
To explore the genetic basis underlying metabolic traits, we conducted GWAS on growth data and free metabolite content of all representative genotypes. We used single nucleotide polymorphisms (SNPs) and structural variations (SVs) (>50 bp) as genetic markers for all GWAS analyses. To allow their use for GWAS, SVs were re‐coded according to a previously published method (Lemay & Malle, 2022). To correct for missing data, we performed an imputation of both SNP and SV datasets using beagle 5.4 v 22Jul22.46e (Browning et al., 2018).
We conducted GWAS on SNP and SV data using the vcf2gwas platform (v 0.8.7) (Vogt et al., 2021). Markers were pruned using the Plink software integrated into vcf2gwas, with phased r 2 thresholds of 0.33 and 0.15 for SVs and SNPs, respectively. For filtering low abundance alleles, we applied a minor allele frequency (MAF) threshold of 5%, leaving 10,057 SNPs and 1,182 SVs for analysis. We corrected for population structure through PCA from the input genotype files. Here, we estimated population structure based on four PCs accounting for the four genetic populations of S. polyrhiza. All GWAS were conducted using a univariate linear mixed model (Zhou & Stephens, 2012). The genotype data, including annotation, were published previously (Wang et al., 2024) and can be found at: https://github.com/Xu‐lab‐Evolution/Great_duckweed_popg (accessed 22 March 2024).
RT‐qPCR quantification of candidate genes
Our GWAS detected a deletion in SpUBP7 that was associated with increased content of L‐glutamine and L‐serine. To validate the effect of the deletion on gene function, we studied expression of SpUBP7 (SpGA2022_056000) in two different allelic backgrounds.
For this, we cultivated genotypes SP012 (low content of L‐serine and L‐glutamine, no deletion in SpUBP7) and SP187 (high content of L‐serine and L‐glutamine, homozygous for deletion in SpUBP7) in N‐medium (Appenroth, 2015) at 26 °C, 135 μmol photons m−2·s−1 16/8 h light/dark photoperiod. For each genotype, five replicates were made, each consisting of ten fronds growing as colonies. The fronds were cultivated in plastic beakers (product 560/250, transparent, round, 250 mL, Plastikbecher.de, Giengen, Germany), covered with perforated lids to allow gas exchange. Each beaker was filled with 150 mL N‐medium.
After five days of cultivation, we separately harvested root and frond tissue of each replicate for RNA extraction. We extracted RNA from <20 mg fresh weight using the InnuPREP RNA mini kit (Analytik, Jena, Germany). For each sample, RNA concentration was measured using a Nanodrop and integrity was checked via gel electrophoresis on a 1% agarose gel. We used 600 ng RNA for each cDNA synthesis following instructions of the RevertAid First Strand cDNA synthesis kit (Takara, Shiga, Japan). For all cDNA syntheses, we used oligo‐dT primers. qPCRs were carried out using a RotorGene Q system (Qiagen, Venlo, Netherlands) applying a time program with an initial denaturation of 98 °C for 3 min, followed by 40 cycles with 98 °C for 3 s and 60 °C for 20 s. Before qPCR, we estimated primer efficiency for SpUBP7 using serial dilutions of pooled cDNA templates from root and frond tissue samples. Primer specificity was checked by evaluating qPCR products on a 2% agarose gel and with melting curve analysis. We conducted all qPCR reactions according to instructions of the KAPPA SYBR FAST kit (Roche, Basel, Switzerland). All cDNA samples were diluted at 1:100 in nuclease‐free water before analysis. We used ALPHA‐ELONGATION FACTOR 1 SpaEF (SpGA2022_005771) and GLYCERINALDEHYDE‐3‐PHOSPHATE DEHYDROGENASE SpGAPDH (SpGA2022_054082) as references for visualization of candidate gene expression according to the delta–delta Ct method with multiple reference genes, as published previously (Hellemans et al., 2007). The primers used can be found in Table S7. Primers for the reference genes SpaEF and SpGAPDH were used in previous publications (Wang et al., 2024; Höfer et al., 2024b).
Software and statistics
We conducted all statistical analyses with R v 4.2.0. Standard errors of means were calculated using the plotrix R package (Lemon, 2006). For all LC–MS measurements, we recorded and quantified analyte and standard peaks with LabSolutions software v 5.97. We estimated the broad sense heritability (H2.c) from our comprehensive dataset, which includes repeatedly analysed genotypes with the lme4 R package (Bates et al., 2015) following a previously published method (Cullis et al., 2006).
The significance of correlations was evaluated using the F‐test. We used Bonferroni corrected P < 0.05 to determine significant genetic markers associated with our traits. The P‐values for GWAS were calculated using the Wald test. For SpUBP7, we compared the SV effect of heterozygous and WT samples using a two‐tailed Student's t‐test. We used Student's t‐test for comparison of gene expression between two genotypes. We conducted the Levene test to check for homogeneity of variances using the car R package (Fox & Weisberg, 2019).
RESULTS
Intraspecific variation of growth rate and metabolic traits
We quantified 42 metabolites, including 20 free amino acids and their derivatives, 11 specialized metabolites, and 11 phytohormones (Fig. 1) among 137 S. polyrhiza genotypes. L‐Asparagine and L‐glutamine were the most abundant free amino acid in S. polyrhiza, with average genotype tissue concentrations of 114.4 and 57.0 μmol·mg−1 DW, respectively (Table S8). ABA was the least abundant metabolite, with an average concentration of 37.0 pmol·mg−1 DW (Table S9).
Among all quantified metabolic features, the content of flavonoids and phenolic acids had highest intraspecific variation (Table S10, Figures [Link], [Link], [Link], [Link]). Chlorogenic acid and luteolin‐7‐O‐glycoside showed 214.9‐ and 183.8‐fold differences, respectively, among genotypes (Table S10, Figure S3). Growth rate, quantified either as RGR of FN or FA, had <2.4‐fold differences (Figure S1).
On average, growth and biomass had higher broad‐sense heritability than metabolite levels (Table 1, Tables S8–S10). All growth and biomass parameters had overall high broad‐sense heritability values between 0.6 and 0.9, with fresh weight being most heritable (Table 1). A total of 25 metabolites had high heritability values (H2.c > 0.3), while the remaining 16 metabolites had low heritability (H2.c < 0.3) (Tables S8–S10). Together, these findings suggest that growth rate and metabolic traits vary within S. polyrhiza, and most of these traits are heritable.
Table 1.
Phenotype variation and heritability of biomass and growth rate in S. polyrhiza.
| Fitness trait | Mean | SE | Fold change | H2.c |
|---|---|---|---|---|
| Fresh weight | 227.7 mg | 7.3 mg | 10.4 | 0.865 |
| Dry weight | 16.67 mg | 0.5 mg | 10.4 | 0.602 |
| RGR of FA | 0.323 d−1 | 0.004 d−1 | 2.4 | 0.841 |
| RGR of FN | 0.259 d−1 | 0.003 d−1 | 2.2 | 0.802 |
H2.c = broad sense heritability according to Cullis et al. (2006); RGR of FA = relative growth rate of frond area; RGR of FN = relative growth rate of frond number; SE = standard error.
Metabolic traits correlate with biomass and growth
Plant metabolic traits are often associated with growth and biomass (Angelovici et al., 2013; Chen et al., 2016; Živanović et al., 2020; Cu et al., 2021). In S. polyrhiza, metabolites showed stronger correlations with biomass than growth rate. In total, 19% and 29% of quantified metabolites correlated significantly with growth rate (RGRs of FN and FA) and biomass (FW and DW) (Fig. 2A, Figures S5 and S6), respectively. Among 20 quantified free amino acids, 11 had a significant positive correlation with biomass (DW), including L‐glutamine, L‐tyrosine, L‐serine and L‐threonine (Fig. 2A, Figure S6), whereas five were positively correlated with growth rate (RGR of FN). For phytohormones, only IAA and JA‐Ile were positively correlated with biomass. Interestingly, most of the quantified specialized metabolites, while positively correlated with cytokinins, were negatively correlated with biomass (Fig. 2A). Among all metabolites, cyanidine‐3‐C‐glycoside had the strongest negative correlation with both growth rate and biomass (FW) (Fig. 2A, Figures S5 and S7).
Fig. 2.

Correlation matrix (A) and principal coordinates analyses (PCA) on growth parameters (B) and metabolite content (C). (A) Tissue concentrations of free amino acids showed strong positive correlation patterns. In most cases, content of free amino acids was positively correlated with flavonoid content and biomass. Metabolite concentrations and growth rates were measured for 137 genotypes. All metabolites were categorized according to Fig. 1. The strength of the correlation was estimated using Pearson correlation coefficient and is reflected by the intensity of red colour for positive and blue for negative correlations. Asterisks highlight significant correlations (F‐test, P < 0.05). Simultaneous correlations of dry and fresh weight are highlighted with red margins. (B and C) Population‐based differences in growth (B) and metabolite levels (C). All PCAs were conducted on 97 representative genotypes assigned to four previously defined genetic populations (Xu et al., 2019). The ellipses show 95% confidence intervals calculated based on standard deviation. The colours of ellipses correspond to the genetic populations. ABA = abscisic acid; cZ = cis‐zeatin; cZR = cis‐zeatin riboside; DW = dry weight; FW = fresh weight; IAA = indole‐3‐acetic acid; iP = isopentenyladenine; iPR = isopentenyladenine riboside; JA = jasmonic acid; JA‐Ile = jasmonic acid‐isoleucine conjugate; RGR of FA = relative growth rate of frond area; RGR of FN = relative growth rate of frond number; SA = salicylic acid; tZ = trans‐zeatin; tZR = trans‐zeatin riboside.
For all quantified metabolites, there was a strong correlation within each metabolite group (Fig. S8). Most free amino acids were positively correlated with each other and with flavonoids. Often, phenolic acids were positively correlated with flavonoids (Fig. 2A), reflecting their biosynthetic relationships (Fig. 1).
Because S. polyrhiza has a strong population structure (Xu et al., 2019; Wang et al., 2024), which can confound phenotypic variations and reduce the power of GWAS, we further investigated whether metabolism differed among populations, using a PCA but no population‐specific differences were found at metabolic level (Fig. 2B, C), indicating GWAS can be used to study most metabolic traits in S. polyrhiza.
Genome‐wide association studies on metabolite traits
We used a GWA approach to identify the genetic basis underlying different metabolic traits. Here, we focused on 10,057 unlinked SNPs and 1,182 SVs, of which 75 SNPs and 15 SVs were significantly associated with the 42 metabolites (Tables S11–S13). Among them, 13 SNPs and three SVs were associated with several metabolites, indicating their pleiotropic role in metabolism. Surprisingly, we did not find any SNPs and only one SV, located in an intergenic region (>30 Kb from next open‐reading frame), were associated with growth rate and biomass (Tables S12 and S13).
Among metabolic traits, several loci were associated with content of L‐glutamine, L‐tyrosine, L‐tryptophan, L‐serine, chlorogenic acid and IAA. Most notably, several SNPs located within the LIGHT HARVESTING COMPLEX OF PHOTOSYSTEM II 5 (SpLHCB5), a structural component of photosystem II (de Bianchi et al., 2008), were associated with content of L‐glutamine, L‐valine, L‐tryptophan and L‐tyrosine (Fig. 3A–C, Table S13). Further, an SNP upstream of a PLANT U‐BOX CONTAINING PROTEIN 4 (SpPUB4), which functions as a regulator of cell division in meristematic regions (Kinoshita et al., 2015), was associated with L‐tyrosine content. Three SNPs associated with the content of L‐tyrosine, chlorogenic acid and IAA were linked to SpUGT89B1, SpCYP71AU50 and SpMYBC1, three genes functioning in secondary metabolite biosynthesis (Caputi et al., 2012; Yamaguchi et al., 2014; Ke et al., 2021) (Fig. 3B,D,E). An SNP within SpDPE2, which is functionally involved in starch metabolism, is associated with IAA content (Li et al., 2022) (Fig. 3E). For SVs, the content of L‐glutamine and L‐serine were associated with a 94‐bp intronic deletion in UBIQUITIN‐SPECIFIC PROTEASE 7 SpUBP7 (Fig. 4A–C). Homologues of SpUBP7 have been shown to stabilize ubiquitin upon proteasome binding and regulate proteasomal activity (Leggett et al., 2002; Wu et al., 2019). Presence of the intronic deletion was associated with increased content of L‐glutamine and L‐serine (Fig. 4D,E) and increased expression of SpUBP7 in roots (Fig. 4F,G). These findings suggest that biomass and metabolite content are coordinated through photosynthesis, starch metabolism, cell division, secondary metabolite biosynthesis, and protein degradation.
Fig. 3.

Manhattan plots of SNP‐based GWAS for free amino acids. Each plot shows the GWAS result on concentrations of L‐glutamine (A), L‐tyrosine (B), L‐tryptophan (C), chlorogenic acid (D) and indole‐3‐acetic acid (IAA) (E) with their corresponding QQ plots on the right side. Bonferroni corrected P = 0.05 (Wald‐test) at 4.97 10−6 was used as significance threshold and is shown as a red line. All significant markers are labelled with the names of gene candidates.
Fig. 4.

The presence of the intronic deletion in SpUBP7 is associated with increased content of L‐glutamine and L‐serine. SV‐based GWAS on L‐glutamine (A) and L‐serine (B). The red line represents Bonferroni corrected P = 0.05 (Wald‐test) at 4.23 10−5. (C) The associated 94 bp deletion (red bar) is located within the 7th intron of SpUBP7. Presence of the deletion is associated with increased L‐glutamine (D) and L‐serine (E) content. (F) SpUBP7 shows genotype‐dependent expression differences at root level. SP187, a genotype that is homozygous for the deletion, shows enhanced SpUBP7 expression compared to genotype SP012 that lacks the deletion. (G) No significant differential expression of SpUBP7 between SP187 and SP012 was found in fronds.
DISCUSSION
Using targeted metabolomics on 137 S. polyrhiza genotypes, we identified strong positive correlations between chemically related metabolites. Additionally, candidate genes involved in photosynthesis (SpLHCB5), protein degradation (SpUBP7) and organ development (SpLOB, SpAGL62) were repeatedly associated with different metabolite levels, suggesting strong genetic co‐regulation. Since levels of free amino acids and specialized metabolites correlated with plant biomass in many cases, some of these candidate genes are potentially of use for joint optimization of biomass and nutrient level.
Free amino acids have large pool sizes and turnover rates in duckweed and serve as major carbon and nitrogen sources for starch and storage protein production (Evans et al., 2018). Previous work found that supplementation of growth medium with amino acids such as L‐glutamine was associated with increased protein content and biomass in duckweed (Shi et al., 2023). In agreement with these observations, associations between free amino acid levels and biomass were mostly positive. Compared to free amino acids, phenylpropanoids and phytohormones have higher functional diversity. In S. polyrhiza, anthocyanins alleviate Cr(VI) stress (Oláh et al., 2009), whereas luteolins and apigenins are involved in stress responses to copper and UV‐light, respectively (Böttner et al., 2021). The content of these three flavonoid classes is inversely correlated with biomass in S. polyrhiza (Oláh et al., 2009; Böttner et al., 2021), which can be explained by increased allocation of carbon resources to flavonoid biosynthesis. In comparison, our study revealed more diverse association patterns of flavonoid levels with biomass. While the content of sinapic acid, luteolins and cyanidin‐3O‐glycoside was negatively associated with biomass, content of apigenin‐8‐C‐glycoside and chlorogenic acid showed positive correlations with biomass. In other plant species, changes in profiles of glycosylated flavonoids regulate plant growth by influencing auxin transport (Ringli et al., 2008), possibly explaining different association patterns of glycosylated luteolins and apigenins to plant biomass in our study. Content of chlorogenic acid was associated with increased leaf growth and content of antioxidants and proteins, suggesting a regulatory function in plant growth (Zhang et al., 2024). The overall positive association of cytokinin levels with most phenylpropanoids was previously explained by their function in coordinating resource allocation to defence metabolism during plant development (Brütting et al., 2017). For phytohormones, levels of JA‐Ile were positively correlated with biomass. In duckweeds, jasmonates show a bimodal dose–response pattern, with small concentrations stimulating growth and turion germination, and high concentrations delaying growth (Appenroth et al., 1991; Piotrowska et al., 2010). Since metabolites were measured in unstressed plant material, JA content likely never exceeded constitutive levels, which would be required for significant growth inhibition. Together, these correlation patterns suggest that S. polyrhiza can be simultaneously optimized for higher biomass and increased content of certain phenylpropanoids.
On a genetic level, we identified for most specialized metabolite and phythohormone levels genetic associations, whereas for many free amino acids, no significant associations were found. For growth rate and biomass, we did not find significant associations for any markers located in genic regions, likely due to their highly polygenic features. Previous GWAS on complex traits found that success in identifying genetic associations greatly depends on sample size and phenotype variations (Chen et al., 2014; Schizophrenia Working Group of the Psychatric Genomics Consortium, 2014; Duncan et al., 2019). In line with previous reports (Chen et al., 2016; Pham et al., 2023), secondary metabolite levels had the highest average phenotypic variation across all metabolic traits, whereas variation in growth parameters was lowest. For specialized metabolites, high variation in tissue concentration was explained by low trait complexity (Chen et al., 2014; Haghi et al., 2022; Pham et al., 2023). While primary metabolites and growth are regulated by many small effect loci, specialized metabolites are often controlled by a few major‐effect loci (Chen et al., 2014, 2016). Since small effect loci are often diffusely distributed across the genome, GWAS on primary metabolite levels and growth often require large sample sizes to detect them (Duncan et al., 2019). Compared to other association studies on plant metabolic traits (Angelovici et al., 2013; Chen et al., 2014, 2016; Cu et al., 2021), our GWAS suffered from a low sample size, likely explaining the lack of genetic associations for growth traits and several metabolites.
On a physiological level, differences in growth were correlated with photosynthesis rate in duckweed (Sree et al., 2015). Hence, metabolism of storage molecules plays a major role in plant growth. Interestingly, many biomass‐correlated metabolites were associated with genes influencing photosynthetic efficiency (SpLHCB5), protein turnover (SpUBP7), starch metabolism (SpDPE2) and cell‐cycle control (SpPUB4). These genes provide suitable candidates for future reverse genetic studies, which are required for further elucidation of their exact function in duckweed metabolism. Taken together, this study provides insights into the molecular basis of growth regulation and metabolite homeostasis in S. polyrhiza. Through identification of gene candidates associated with metabolic traits, this study contributes to laying the foundations for further optimization strategies in S. polyrhiza.
AUTHOR CONTRIBUTIONS
MH analysed the data. MH, MS, SW, and YW performed the experiments. YW and SX provided resources and technical infrastructure. SX and MH conceived and supervised the project. MH wrote the manuscript. All authors contributed to the final version of the manuscript.
CONFLICT OF INTEREST
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
FUNDING INFORMATION
This research was funded by the “Deutsche Forschungsgemeinschaft”, grant number 427577435 and 435681637 to SX.
Supporting information
Figure S1. Kernel density distribution of growth and biomass parameters measured for 137 genotypes of S. polyrhiza: Fresh weight (A), RGR of frond number (B), RGR of frond area (C) and dry weight (D) for 137 genotypes of S. polyrhiza.
Figure S2. Kernel density distribution of free amino acid concentrations measured for 137 genotypes of S. polyrhiza: L‐Alanine (A), L‐Arginine (B), L‐Asparagine (C), L‐Aspartic acid (D), L‐Glutamic acid (E), L‐Glutamine (F), L‐Isoleucine (G), L‐Leucine (H), L‐Phenylalanine (I), L‐Proline (J), L‐Serine (K), L‐Threonine (L), L‐Tryptophan (M), L‐Valine (N), Glycine (O), L‐Histidine (P), L‐Methionine (Q), L‐Tyrosine (R), Shikimic acid (S) and Tyramine (T).
Figure S3. Kernel density distribution of secondary metabolite concentrations measured for 137 genotypes of S. polyrhiza: Chlorogenic acid (A), Coumaric acid (B), Caffeic acid (C), Sinapic acid (D), Apigenin (E), Luteolin (F), Cyanidin‐3‐O‐glycoside (G), Luteolin‐7‐O‐glycoside (H), Luteolin‐8‐C‐glycoside (I), Apigenin‐8‐C‐glycoside (J) and Apigenin‐7‐O‐glycoside (K).
Figure S4. Kernel density distribution of phytohormone concentrations measured for 137 genotypes of S. polyrhiza: Abscisic acid (A), Salicylic acid (B), Jasmonic acid (C), Jasmonic acid‐Isoleucine conjugate (D), N6–Isopentenyladenine (E), N6–Isopentenyladenine riboside (F), trans‐Zeatin (G), trans‐Zeatin riboside (H), cis‐Zeatin (I), cis‐Zeatin riboside (J), Indole‐3‐acetic acid (K).
Figure S5. Scatterplots of all significant correlations (F‐test, P ≤ 0.05) of fresh weight with individual free metabolite contents. Fresh weight significantly correlated with contents of L‐Glutamine (A), L‐Tyrosine (B), L‐Tryptophane (C), L‐Threonine (D), L‐Isoleucine (E), L‐Serine (F), Glycine (G), Chlorogenic acid (H), Sinapic acid (I), Luteolin (J), Cyanidin‐3‐O‐glycoside (K), Luteolin‐8‐C‐glycoside (L), Luteolin‐7‐O‐glycoside (M), Apigenin‐8‐C‐glycoside (N), Jasmonic acid–Isoleucine (O) and Indole‐3‐acetic acid (P).
Figure S6. Scatterplots showing all identified significant correlations (F‐test, P ≤ 0.05) of individual free metabolite contents with dry weight. Dry weight was significantly correlated with levels of L‐Glutamine (A), L‐Tyrosine (B), L‐Tryptophane (C), L‐Threonine (D), L‐Isoleucine (E), L‐Serine (F), Glycine (G), Chlorogenic acid (H), Sinapic acid (I), Apigenin‐8‐C‐glycoside (J), Jasmonic acid–Isoleucine (K), Indole‐3‐acetic acid (L).
Figure S7. Correlation of growth with contents of Cyanidin‐3‐O‐glycoside. Both metabolites showed strong negative correlation patterns with RGR of frond number (Pearson, ρ = −0.57) (A) and RGR of frond area (Pearson, ρ = −0.52) (B).
Figure S8. Principal coordinate analysis (PCA) on levels of 42 metabolites, that were categorized into nine groups according to their biosynthetic origin. Contents of all free metabolites except that of salicylic acid (SA) and Jasmonates (JA and JA‐Ile) formed a cluster.
Data S1. Supplemental Methods.
Data S2. Supplementary Tables.
ACKNOWLEDGEMENT
We thank Marie Serwaty‐Sárazová for maintaining the giant duckweed collection.
Editor: T. Hildebrandt
DATA AVAILABILITY STATEMENT
All raw data and scripts used for analysis can be found at: https://doi.org/10.17632/xwsxpfcysd.1
References
- Acosta K., Appenroth K.J., Borisjuk L., Edelman M., Heinig U., Jansen M.A.K., Oyama T., Pasaribu B., Schubert I., Sorrels S., Sree K.S., Xu S., Michael T.P., Lam E. (2021) Return of the Lemnaceae: Duckweed as a model plant system in the genomics and postgenomics era. The Plant Cell, 33, 3207–3234. 10.1093/plcell/koab189 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Aftab T., Masroor M., Khan A., Idrees M., Naeem M. (2010) Salicylic acid acts as potent enhancer of growth, photosynthesis and artemisinin production in Artemisia annua L. Journal of Crop Science and Biotechnology, 13, 183–188. 10.1007/s12892-010-0040-3 [DOI] [Google Scholar]
- Angelovici R., Batushansky A., Deason N., Gonzalez‐Jorge S., Gore M.A., Fait A., DellaPenna D. (2017) Network‐guided GWAS improves identification of genes affecting free amino acids. Plant Physiology, 173, 872–886. 10.1104/pp.16.01287 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Angelovici R., Lipka A.E., Deason N., Gonzalez‐Jorge S., Lin H., Cepela J., Buell R., Gore M.A., Dellapenna D. (2013) Genome‐wide analysis of branched‐chain amino acid levels in Arabidopsis seeds. The Plant Cell, 25, 4827–4843. 10.1105/tpc.113.119370 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Appenroth K.J. (2015) Media for in vitro‐cultivation of duckweed. Duckweed Forum, 3, 180–186. [Google Scholar]
- Appenroth K.J., Dathe W., Hertel W., Augsten H. (1991) Photophysiology of turion germination in Spirodela polyrhiza (L.) SCHLEIDEN. VII. Action of Jasmonic acid. Journal of Plant Physiology, 138, 345–349. 10.1016/S0176-1617(11)80298-1 [DOI] [PubMed] [Google Scholar]
- Baek G., Saeed M., Choi H.‐K. (2021) Duckweeds: Their utilization, metabolites and cultivation. Applied Biological Chemistry, 64, 73. 10.1186/s13765-021-00644-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bates D., Mächler M., Bolker B., Walker S. (2015) Fitting linear mixed‐effect models using lme4. Journal of Statistical Software, 67, 1–48. 10.18637/jss.v067.i01 [DOI] [Google Scholar]
- Böttner L., Grabe V., Gablenz S., Böhme N., Appenroth K.J., Gershenzon J., Huber M. (2021) Differential localization of flavonoid glucosides in an aquatic plant implicates different functions under abiotic stress. Plant, Cell & Environment, 44, 900–914. 10.1111/pce.13974 [DOI] [PubMed] [Google Scholar]
- Browning B.L., Zhou Y., Browning S.R. (2018) A one‐penny imputed genome from next‐generation reference panels. The American Journal of Human Genetics, 103, 338–348. 10.1016/j.ajhg.2018.07.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Brütting C., Schäfer M., Vanková R., Gase K., Baldwin I.T., Meldau S. (2017) Changes in cytokinins are sufficient to alter developmental patterns of defense metabolites in Nicotiana attenuata . The Plant Journal, 89, 15–30. 10.1111/tpj.13316 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Caputi L., Malnoy M., Goremykin V., Nikiforova S., Martens S. (2012) A genome‐wide phylogenetic reconstruction of family 1 UDP‐glycosyltransferases revealed the expansion of the family during the adaptation of plants to life on land. The Plant Journal, 69, 1030–1042. 10.1111/j.1365-313X.2011.04853.x [DOI] [PubMed] [Google Scholar]
- Chen W., Gao Y., Xie W., Gong L., Lu K., Wang W., Li Y., Liu X., Zhang H., Dong H., Zhang W., Zhang L., Yu S., Wang G., Lian X., Luo J. (2014) Genome‐wide association analyses provide genetic and biochemical insights into natural variation in rice metabolism. Nature Genetics, 46, 714–721. 10.1038/ng.3007 [DOI] [PubMed] [Google Scholar]
- Chen W., Wang W., Peng M., Gong L., Gao Y., Wan J., Wang S., Shi L., Zhou B., Li Z., Peng X., Yang C., Qu L., Liu X., Luo J. (2016) Comparative and parallel genome‐wide association studies for metabolic and agronomic traits in cereals. Nature Communications, 7, 12767. 10.1038/ncomms12767 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cu S.T., Warnock N.I., Pasuquin J., Dingkuhn M., Stangoulis J. (2021) A high‐resolution genome‐wide association study of the grain ionome and agronomic traits in rice Oryza sativa subsp. indica. Scientific Reports, 11, 19230. 10.1038/s41598-021-98573-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cullis B.R., Smith A.B., Coombes N.E. (2006) On the design of early generation variety trials with correlated data. Journal of Agricultural, Biological and Environmental Statistics, 11, 381–393. 10.1198/108571106X154443 [DOI] [Google Scholar]
- de Bianchi S., Dall'Osto L., Tognon G., Morosinotto T., Bassi R. (2008) Minor antenna proteins CP24 and CP26 affect the interactions between photosystem II subunits and the electron transport rate in grana membranes of Arabidopsis. The Plant Cell, 20, 1012–1028. 10.1105/tpc.107.055749 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dick R., Rattei T., Haslbeck M., Schwab W., Gierl A., Frey M. (2012) Comparative analysis of benzoxazinoid biosynthesis in monocots and dicots: Independent recruitment of stabilization and activation functions. The Plant Cell, 24, 915–928. 10.1105/tpc.112.096461 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Duncan L.E., Ostacher M., Ballon J. (2019) How genome‐wide association studies (GWAS) made traditional candidate gene studies obsolete. Neuropsychopharmacology, 44, 1518–1523. 10.1038/s41386-019-0389-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Evans E.M., Freund D.M., Sondervan V.M., Cohen J.D., Hegeman A.D. (2018) Metabolic patterns in Spirodela polyrhiza revealed by (15)N stable isotope labeling of amino acids in photoautotrophic, heterotrophic, and mixotrophic growth conditions. Frontiers in Chemistry, 6, 191. 10.3389/fchem.2018.00191 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fox J., Weisberg S. (2019) An R companion to applied regression, 3rd edition. SAGE, Thousand Oaks, CA, USA. [Google Scholar]
- Haghi R., Ahmadikhah A., Fazeli A., Shariati V. (2022) Candidate genes for anthocyanin pigmentation in rice stem revealed by GWAS and whole‐genome resequencing. The Plant Genome, 15, e20224. 10.1002/tpg2.20224 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hellemans J., Mortier G., De Paepe A., Speleman F., Vandesompele J. (2007) qBase relative quantification framework and software for management and automated analysis of real‐time quantitative PCR data. Genome Biology, 8, R19. 10.1186/gb-2007-8-2-r19 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Höfer M., Schäfer M., Wang Y., Wink S., Xu S. (2024a) Data from: Genetic mechanism of non‐targeted‐site resistance to diquat in Spirodela polyrhiza [Dataset]. Dryad. 10.5061/dryad.2fqz612ww [DOI] [PMC free article] [PubMed] [Google Scholar]
- Höfer M., Schäfer M., Wang Y., Wink S., Xu S. (2024b) Genetic mechanism of non‐targeted‐site resistance to diquat in Spirodela polyrhiza . Plants, 13, 845. 10.3390/plants13060845 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hui T., Zhang Y., Jia R., Hu Y., Wang W., Wang Y., Wang Y., Zhu Y., Yang L., Xiang B. (2023) Metabolomic analysis reveals responses of Spirodela polyrhiza L. to salt stress. Journal of Plant Interactions, 18, 2210163. 10.1080/17429145.2023.2210163 [DOI] [Google Scholar]
- Huot B., Yao J., Montgomery B.L., He S.Y. (2014) Growth‐defense tradeoffs in plants: A balancing act to optimize fitness. Molecular Plant, 7, 1267–1287. 10.1093/mp/ssu049 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kassambara A (2023) Ggcorrplot: Visualization of a correlation matrix using ‘ggplot2’. R Foundation for Statistical Computing, Vienna, Austria. [Google Scholar]
- Ke Y., Abbas F., Zhou Y., Yu R., Fan Y. (2021) Auxin‐responsive R2R3‐MYB transcription factors HcMYB1 and HcMYB2 activate volatile biosynthesis in Hedychium coronarium flowers. Frontiers in Plant Science, 12, 710826. 10.3389/fpls.2021.710826 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kinoshita A., ten Hove C.A., Tabata R., Yamada M., Shimizu N., Ishida T., Yamaguchi K., Shigenobu S., Takebayashi Y., Iuchi S., Kobayashi M., Kurata T., Wada T., Seo M., Hasebe M., Blilou I., Fukuda H., Scheres B., Heidstra R., Kamiya Y., Sawa S. (2015) A plant U‐box protein, PUB4, regulates asymmetric cell division and cell proliferation in the root meristem. Development, 142, 444–453. 10.1242/dev.113167. [DOI] [PubMed] [Google Scholar]
- Leggett D.S., Hanna J., Borodovsky A., Crosas B., Schmidt M., Baker R.T., Walz T., Ploegh H., Finley D. (2002) Multiple associated proteins regulate proteasome structure and function. Molecular Cell, 10, 495–507. 10.1016/s1097-2765(02)00638-x [DOI] [PubMed] [Google Scholar]
- Lemay M.‐A., Malle S. (2022) A practical guide to using structural variants for genome‐wide association studies. In: Torkamaneh D., Belzile F. (Eds), Genome‐wide association studies. Springer, New York, NY, USA, pp 161–172. 10.1007/978-1-0716-2237-7_10 [DOI] [PubMed] [Google Scholar]
- Lemon J. (2006) Plotrix: A package in the red light distric of R. R‐News, 6, 8–12. [Google Scholar]
- Li J., Halitschke R., Li D., Paetz C., Su H., Heiling S., Xu S., Baldwin I.T. (2021) Controlled hydroxylations of diterpenoids allow for plant chemical defense without autotoxicity. Science, 371, 255–260. 10.1126/science.abe4713 [DOI] [PubMed] [Google Scholar]
- Li X., Apriyanto A., Castellanos J.F., Compart J., Muntaha S.N., Fettke J. (2022) Dpe2/phs1 revealed unique starch metabolism with three distinct phases characterized by different starch granule numbers per chloroplast, allowing insights into the control mechanism of granule number regulation by gene co‐regulation and metabolic profiling. Frontiers in Plant Science, 13, 1039534. 10.3389/fpls.2022.1039534 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ligges U., Mächler M. (2003) Scatterplot3d–An R package for visualizing multivariate data. Journal of Statistical Software, 8, 1–20. 10.18637/jss.v008.i11 [DOI] [Google Scholar]
- Noctor G., Foyer C.H. (1998) A re‐evaluation of the ATP:NADPH budget during C3 photosynthesis: A contribution from nitrate assimilation and its associated respiratory activity? Journal of Experimental Botany, 49, 1895–1908. 10.1093/jxb/49.329.1895 [DOI] [Google Scholar]
- Oksanen J., Simpson G.L., Blanchet G.F., Kindt R., Legendre P., Minchin P.R., O'Hara R.B., Solymos P., Stevens M.H.H., Szoecs E., Wagner H., Barbour M., Bedward M., Bolker B., Borcard D., Carvalho G., Chirico M., De Caceres M., Durand S., Evangelista H.B.A.F., Rich F.M., Furneaux B., Hannigan G., Hill M.O., Lahti L., McGlinn D., Ouellette M.‐H., Ribeiro Cunha E., Smith T., Stier A., Ter Braak C.J.F., Weedon J. (2022) Vegan: Community ecology package. R Foundation for Statistical Computing, Vienna, Austria. [Google Scholar]
- Oláh V., Combi Z., Szöllösi E., Kanalas P., Mészáros I. (2009) Anthocyanins: Possible antioxidants against Cr (VI) induced oxidative stress in Spirodela polyrrhiza . Cereal Research Communications, 37, 533–536. 10.1556/CRC.37.2009.Suppl.4 [DOI] [Google Scholar]
- Pham C.H., Do T.D., Nguyen H.T.L., Hoang N.T., Tran T.D., Vu M.T.T., Doi H.H., Bui T.‐G.T., Henry R.J. (2023) Genome‐wide association mapping of genes for anthocyanin and flavonoid contents in Vietnamese landraces of black rice. Euphytica, 220, 11. 10.1007/s10681-023-03268-0 [DOI] [Google Scholar]
- Piotrowska A., Bajguz A., Czerpak R., Kot K. (2010) Changes in the growth, chemical composition, and antioxidant activity in the aquatic plant Wolffia arrhiza (L.) Wimm. (Lemnaceae) exposed to Jasmonic acid. Journal of Plant Growth Regulation, 29, 53–62. 10.1007/s00344-009-9113-8 [DOI] [Google Scholar]
- Ringli C., Bigler L., Kuhn B.M., Leiber R.‐M., Diet A., Santelia D., Frey B., Pollmann S., Klein M. (2008) The modified flavonol glycosylation profile in the Arabidopsis rol1 mutants results in alterations in plant growth and cell shape formation. The Plant Cell, 20, 1470–1481. 10.1105/tpc.107.053249 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schäfer M., Brütting C., Baldwin I.T., Kallenbach M. (2016) High‐throughput quantification of more than 100 primary‐ and secondary‐metabolites, and phytohormones by a single solid‐phase extraction based sample preparation with analysis by UHPLC–HESI–MS/MS. Plant Methods, 12, 30. 10.1186/s13007-016-0130-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schizophrenia Working Group of the Psychatric Genomics Consortium (2014) Biological insights from 108 schizophrenia‐associated genetic loci. Nature, 511, 421–427. 10.1038/nature13595 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Shi H., Ernst E., Heinzel N., McCorkle S., Rolletschek H., Borisjuk L., Ortleb S., Martienssen R., Shanklin J., Schwender J. (2023) Mechanisms of metabolic adaptation in the duckweed Lemna gibba: An integrated metabolic, transcriptomic and flux analysis. BMC Plant Biology, 23, 458. 10.1186/s12870-023-04480-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Smith D.C., Bassham J.A., Kirk M. (1961) Dynamics of the photosynthesis of carbon compounds II. Amino acid synthesis. Biochimica et Biophysica Acta, 48, 299–313. 10.1016/0006-3002(61)90478-4 [DOI] [PubMed] [Google Scholar]
- Smith K.E., Schäfer M., Lim M., Robles‐Zazueta C.A., Cowan L., Fisk I.D., Xu S., Murchie E.H. (2024) Aroma and metabolite profiling in duckweeds: Exploring species and ecotypic variation to enable wider adoption as a food crop. Journal of Agriculture and Food Research, 18, 101263. 10.1016/j.jafr.2024.101263 [DOI] [Google Scholar]
- Sree K.S., Sudakaran S., Appenroth K.‐J. (2015) How fast can angiosperms grow? Species and clonal diversity of growth rates in the genus Wolffia (Lemnaceae). Acta Physiologiae Plantarum, 37, 204. 10.1007/s11738-015-1951-3 [DOI] [Google Scholar]
- Stacklies W., Redestig H., Scholz M., Walther D., Selbig J. (2007) pcaMethods–A bioconductor package providing PCA methods for incomplete data. Bioinformatics, 23, 1164. 10.1093/bioinformatics/btm069 [DOI] [PubMed] [Google Scholar]
- Vogt F., Shirsekar G., Weigel D. (2021) vcf2gwas‐python API for comprehensive GWAS analysis using GEMMA. Bioinformatics, 38, 839–840. 10.1093/bioinformatics/btab710 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang Y., Duchen P., Chávez A., Sree K.S., Appenroth K.J., Zhao H., Höfer M., Huber M., Xu S. (2024) Population genomics and epigenomics of Spirodela polyrhiza provide insights into the evolution of facultative asexuality. Communications Biology, 7, 581. 10.1038/s42003-024-06266-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wu R., Zheng W., Tan J., Sammer R., Du L., Lu C. (2019) Protein partners of plant ubiquitin‐specific proteases (UBPs). Plant Physiology and Biochemistry, 145, 227–236. 10.1016/j.plaphy.2019.08.032 [DOI] [PubMed] [Google Scholar]
- Xu S., Stapley J., Gablenz S., Boyer J., Appenroth K.J., Sree K.S., Gershenzon J., Widmer A., Huber M. (2019) Low genetic variation is associated with low mutation rate in the giant duckweed. Nature Communications, 10, 1243. 10.1038/s41467-019-09235-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yamaguchi T., Yamamoto K., Asano Y. (2014) Identification and characterization of CYP79D16 and CYP71AN24 catalyzing the first and second steps in l‐phenylalanine‐derived cyanogenic glycoside biosynthesis in the Japanese apricot, Prunus mume Sieb. et Zucc. Plant Molecular Biology, 86, 215–223. 10.1007/s11103-014-0225-6 [DOI] [PubMed] [Google Scholar]
- Zhang M., Xiao Q., Li Y., Tian Y., Zheng J., Zhang J. (2024) Exploration of exogenous chlorogenic acid as a potential plant stimulant: Enhancing physiochemical properties in Lonicera japonica . Physiology and Molecular Biology of Plants, 30, 453–466. 10.1007/s12298-024-01435-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhang P., Zhong K., Zhong Z., Tong H. (2019) Genome‐wide association study of important agronomic traits within a core collection of rice (Oryza sativa L.). BMC Plant Biology, 19, 259. 10.1186/s12870-019-1842-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhao Y., Fang Y., Jin Y., Huang J., Bao S., Fu T., He Z., Wang F., Wang M., Zhao H. (2015) Pilot‐scale comparison of four duckweed strains from different genera for potential application in nutrient recovery from wastewater and valuable biomass production. Plant Biology, 17, 82–90. 10.1111/plb.12204 [DOI] [PubMed] [Google Scholar]
- Zhou X., Stephens M. (2012) Genome‐wide efficient mixed‐model analysis for association studies. Nature Genetics, 44, 821–824. 10.1038/ng.2310 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ziegler P., Adelmann K., Zimmer S., Schmidt C., Appenroth K.J. (2015) Relative in vitro growth rates of duckweeds (Lemnaceae)–The most rapidly growing higher plants. Plant Biology, 17, 33–41. 10.1111/plb.12184 [DOI] [PubMed] [Google Scholar]
- Živanović B., Milić Komić S., Tosti T., Vidović M., Prokić L., Veljović Jovanović S. (2020) Leaf soluble sugars and free amino acids as important components of abscisic acid‐mediated drought response in tomato. Plants, 9, 9. 10.3390/plants9091147 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Züst T., Agrawal A.A. (2017) Trade‐offs between plant growth and defense against insect herbivory: An emerging mechanistic synthesis. Annual Review of Plant Biology, 68, 513–534. 10.1146/annurev-arplant-042916-040856 [DOI] [PubMed] [Google Scholar]
- Züst T., Joseph B., Shimizu K.K., Kliebenstein D.J., Turnbull L.A. (2011) Using knockout mutants to reveal the growth costs of defensive traits. Proceedings of the Royal Society B: Biological Sciences, 278, 2598–2603. 10.1098/rspb.2010.2475 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Kernel density distribution of growth and biomass parameters measured for 137 genotypes of S. polyrhiza: Fresh weight (A), RGR of frond number (B), RGR of frond area (C) and dry weight (D) for 137 genotypes of S. polyrhiza.
Figure S2. Kernel density distribution of free amino acid concentrations measured for 137 genotypes of S. polyrhiza: L‐Alanine (A), L‐Arginine (B), L‐Asparagine (C), L‐Aspartic acid (D), L‐Glutamic acid (E), L‐Glutamine (F), L‐Isoleucine (G), L‐Leucine (H), L‐Phenylalanine (I), L‐Proline (J), L‐Serine (K), L‐Threonine (L), L‐Tryptophan (M), L‐Valine (N), Glycine (O), L‐Histidine (P), L‐Methionine (Q), L‐Tyrosine (R), Shikimic acid (S) and Tyramine (T).
Figure S3. Kernel density distribution of secondary metabolite concentrations measured for 137 genotypes of S. polyrhiza: Chlorogenic acid (A), Coumaric acid (B), Caffeic acid (C), Sinapic acid (D), Apigenin (E), Luteolin (F), Cyanidin‐3‐O‐glycoside (G), Luteolin‐7‐O‐glycoside (H), Luteolin‐8‐C‐glycoside (I), Apigenin‐8‐C‐glycoside (J) and Apigenin‐7‐O‐glycoside (K).
Figure S4. Kernel density distribution of phytohormone concentrations measured for 137 genotypes of S. polyrhiza: Abscisic acid (A), Salicylic acid (B), Jasmonic acid (C), Jasmonic acid‐Isoleucine conjugate (D), N6–Isopentenyladenine (E), N6–Isopentenyladenine riboside (F), trans‐Zeatin (G), trans‐Zeatin riboside (H), cis‐Zeatin (I), cis‐Zeatin riboside (J), Indole‐3‐acetic acid (K).
Figure S5. Scatterplots of all significant correlations (F‐test, P ≤ 0.05) of fresh weight with individual free metabolite contents. Fresh weight significantly correlated with contents of L‐Glutamine (A), L‐Tyrosine (B), L‐Tryptophane (C), L‐Threonine (D), L‐Isoleucine (E), L‐Serine (F), Glycine (G), Chlorogenic acid (H), Sinapic acid (I), Luteolin (J), Cyanidin‐3‐O‐glycoside (K), Luteolin‐8‐C‐glycoside (L), Luteolin‐7‐O‐glycoside (M), Apigenin‐8‐C‐glycoside (N), Jasmonic acid–Isoleucine (O) and Indole‐3‐acetic acid (P).
Figure S6. Scatterplots showing all identified significant correlations (F‐test, P ≤ 0.05) of individual free metabolite contents with dry weight. Dry weight was significantly correlated with levels of L‐Glutamine (A), L‐Tyrosine (B), L‐Tryptophane (C), L‐Threonine (D), L‐Isoleucine (E), L‐Serine (F), Glycine (G), Chlorogenic acid (H), Sinapic acid (I), Apigenin‐8‐C‐glycoside (J), Jasmonic acid–Isoleucine (K), Indole‐3‐acetic acid (L).
Figure S7. Correlation of growth with contents of Cyanidin‐3‐O‐glycoside. Both metabolites showed strong negative correlation patterns with RGR of frond number (Pearson, ρ = −0.57) (A) and RGR of frond area (Pearson, ρ = −0.52) (B).
Figure S8. Principal coordinate analysis (PCA) on levels of 42 metabolites, that were categorized into nine groups according to their biosynthetic origin. Contents of all free metabolites except that of salicylic acid (SA) and Jasmonates (JA and JA‐Ile) formed a cluster.
Data S1. Supplemental Methods.
Data S2. Supplementary Tables.
Data Availability Statement
All raw data and scripts used for analysis can be found at: https://doi.org/10.17632/xwsxpfcysd.1
