Diversity and Plant Growth‐Promoting Potential of Duckweed‐Associated Bacteria on Wolffia globosa Biomass Production and Nutritional Quality

TL;DR

This study identifies bacteria that significantly boost the growth and nutritional quality of Wolffia, a superfood, suggesting their use as natural biofertilizers.

Contribution

The study discovers and evaluates duckweed-associated bacteria that enhance Wolffia growth and quality, identifying specific strains like Pseudomonas toyotomiensis.

Findings

108 bacterial isolates from duckweed showed high diversity, with Pseudomonadota being the dominant phylum.

Six PGPB strains increased Wolffia globosa biomass by 54.67%–77.75%, with Pseudomonas toyotomiensis W5–11 showing the highest improvement.

Selected PGPB produced IAA, siderophores, and solubilized phosphate, linking IAA production to protein accumulation.

Abstract

Wolffia (Lemnoideae) is recognised as a nutritional superfood with increasing interest in commercial cultivation. Its growth and biomass quality are influenced by abiotic factors and duckweed‐associated bacteria (DAB) that support nutrient cycling, stress tolerance and metabolism. This study assessed DAB diversity and their effects on Wolffia growth to select effective plant growth‐promoting bacteria (PGPB) and elucidated beneficial plant‐microbe interactions. A total of 108 isolates, representing 66 species from 41 genera across four phyla, were obtained from duckweeds collected in six provinces of Thailand. The culturable DAB community showed high taxonomic diversity, dominated by the phylum Pseudomonadota, particularly Alphaproteobacteria. The isolates, along with the known PGPB of
Spirodela polyrhiza
, were evaluated for growth promotion in axenic
Wolffia globosa
using co‐cultivation. Six PGPB enhanced growth by 54.67%–77.75% without reported pathogenicity. Several PGPB originally isolated from
S. polyrhiza
also showed growth‐promoting effects on Wolffia. Large‐scale co‐cultivation demonstrated that
Pseudomonas toyotomiensis
W5–11 most effectively increased dry weight and chlorophyll content (3.18‐ and 2.75‐fold, respectively). All selected isolates produced indole‐3‐acetic acid (IAA), siderophores and solubilised phosphate. Correlation analysis revealed that IAA production was associated with protein accumulation. These findings suggested that PGPB represent promising biofertilisers that enhance Wolffia productivity and nutritional quality through various mechanisms.

Co‐cultivation screening of 116 duckweed‐associated bacteria revealed isolates that enhanced
Wolffia globosa
biomass up to 77.75%, with
Pseudomonas toyotomiensis
W5–11 increasing dry weight 3.18‐fold and chlorophyll content 2.75‐fold, demonstrating their potential as biofertilisers for sustainable duckweed production.

Figures

Sampling locations in Thailand.

Frequency of occurrence of bacteria in duckweed samples.

EPG of PGPB on W. globosa DKU‐25W after co‐cultivation in the Hoagland medium for 7 days (a) and growth comparison with selected PGPB and axenic control (b). Black bars indicate high EPG values > 50%. Error bars represent standard error (n = 3).

Phylogenetic placement of six selected PGPB and 48 closely related taxa based on the sequence analysis of the 16S rRNA gene using the Neighbour‐Joining method based on the maximum composite likelihood model. The complete deletion method was used to remove gaps and missing data, the final dataset contained 1262 positions. Name in bold type are the isolates obtained from this study. Numbers on branches indicate percentages of bootstrap sampling (> 50%), derived from 1000 samples. Bars, 0.02 substitutions per nucleotide position.

EPG of PGPB after co‐cultivation under various bacterial density conditions in Hoagland medium for 7 days. Error bars indicate standard error (n = 3). Asterisks (*) mark the cell density selected for large‐scale co‐cultivation. Different letters (a‐e) indicate significant differences between treatments (p < 0.05, one‐way ANOVA with Duncan's multiple range test).

Effects of PGPB on dry weight (a), chlorophyll content (b), protein content (c) and starch content (d) after large‐scale co‐cultivation in Hoagland medium for 7 days. Error bars indicate standard deviation (n = 3). Different letters (a–c) indicate significant differences between treatments (p < 0.05, one‐way ANOVA with Duncan's multiple range test).

Growth of W. globosa DKU‐25W in large‐scale co‐cultivation with selected PGPB compared to axenic duckweed (control) in Hoagland medium for 7 days, demonstrating visible differences in biomass production and chlorophyll content between treatments.

Correlogram matrix of Pearson's correlation coefficients between bacterial PGP traits and Wolffia growth parameters. Asterisks (*) indicate statistically significant correlations (p < 0.05, Pearson correlation test, n = 6).

Tables

TABLE 1. Undescribed species found in this study.

Isolate	Accession no.	Closest species	Similarity (%)
Phylum Bacteriodota
30–45	LC868795	Chryseobacterium mollis DSM 18016T (jgi.1107922)	98.26
M10	LC868796	Pedobacter kyungheensis KACC 16221T (JSYN01000008)	96.68
Phylum Pseudomonadota; Class Alphaproteobacteria
Rhizobiales
25–25	LC868791	Allorhizobium pseudoryzae J3‐A127T (DQ454123)	98.64
30–1	LC868794	Devosia oryziradicis G19T (MT992792)	98.29
25–26	LC868792	Pleomorphomonas oryzae DSM 16300T (AUHB01000035)	96.79
25–39	LC868793	Pleomorphomonas oryzae DSM 16300T (AUHB01000035)	96.79
Sphingomonadales
M18	LC868797	Novosphingobium hassiacum W‐51T (AJ416411)	98.52
Phylum Pseudomonadota; Class Betaproteobacteria
Burkholderiales
25–17	LC868790	Formosimonas limnophila AHQ‐12T (HG421033)	97.78
25–11	LC868789	Roseateles asaccharophilus KIN192T (AY136099)	98.62
Rhodocyclales
25–7	LC868788	Uliginosibacterium aquaticum IMCC34675T (MK138654)	98.39

TABLE 2. Diversity of DAB in duckweed samples.

Location	Site 13	Site 17	Site 25	Site 30	Site 33	Site M
No. of isolate in phylum
Actinomycetota	2	0	3	4	2	3
Bacillota	5	3	2	2	2	5
Bacteroidota	0	0	0	1	1	2
Pseudomonadota
Alphaproteobacteria	11	2	14	5	5	3
Betaproteobacteria	0	1	4	3	1	0
Gammaproteobacteria	0	1	5	4	7	5
No. of isolate	18	7	28	19	18	18
Species richness	14	6	21	15	10	14
Shannon–Wiener diversity index	2.55	1.75	2.94	2.65	2.17	2.55

TABLE 3. PGP traits and related properties.

Bacteria	W12–8 a	W7–16 a	W5–11	W5–13 a s	30–37	25–48
PGP factor
IAA production (μg/mL)	13.19 ± 1.41	4.87 ± 2.54	109.03 ± 11.77	4.45 ± 2.20	59.27 ± 3.27	15.00 ± 0.30
Siderophore production	1.60	1.21	1.21	1.12	1.61	1.35
P solubilisation	1.23	1.23	2.72	1.40	1.00	3.11
K solubilisation	4.17	0.00	6.40	0.00	1.04	5.39
Nitrogen utilisation
N‐free	−	−	−	−	−	−
Casamino acid	+	+	+	+	+	+
NH4Cl	+	−	+	−	−	+
NaNO3	+	−	+	−	−	+
Nitrate reduction	+	+	+	+	+	+
Denitrification	−	−	−	−	−	−
Plant colonising factors
Motility	−	−	+	−	+	−
EPS production	+	++	+	+	+	+

TABLE 4. PGP effects of PGPB after large‐scale co‐cultivation with Wolffia globosa DKU25‐W in Hoagland medium for 7 days.

Isolate	Dry weight (mg)	Chlorophyll content (μg/g)	Protein content (g/100 g)	Starch content (g/100 g)
Control	19.4 ± 1.7c	197.76 ± 9.32c	8.91 ± 0.93ab	3.49 ± 0.80b
W12–8	38.4 ± 1.7b	293.40 ± 1.20c	9.39 ± 0.12a	2.88 ± 0.70bc
W7–16	42.7 ± 2.3b	432.17 ± 2.77ab	6.82 ± 0.28c	2.18 ± 0.03c
W5–11	61.7 ± 0.9a	542.96 ± 3.66a	10.91 ± 0.37a	2.69 ± 0.01bc
W5–13	36.6 ± 11.4b	445.58 ± 2.14a	6.21 ± 0.21c	2.59 ± 0.12bc
30–37	17.3 ± 1.1c	315.43 ± 18.37bc	9.50 ± 1.84a	4.98 ± 0.38a
25–48	34.2 ± 4.3b	275.98 ± 5.06c	7.33 ± 0.51bc	3.02 ± 0.15bc

Full text

Introduction

1

Duckweeds (Lemnoideae) are small flowering plants that have been extensively studied as a valuable and environmentally friendly biomass due to their ability to accumulate high amounts of starch and protein under varying culture conditions (Ishizawa et al. 2021; Sree and Appenroth 2022; Ziegler et al. 2015). Wolffia, a genus of rootless duckweed with high nutritional value, boasts protein and carbohydrate contents up to 50.89% and 31.33% of dry weight, respectively (Hu et al. 2022). It also exhibits high levels of essential fatty acids, particularly linoleic acid and α‐linolenic acid (26.6% and 38.57% of total fatty acid, respectively) (Arbex et al. 2015; Calder 2015). In addition, Wolffia contains abundant phenolic compounds, flavonoids and antioxidants (Hu et al. 2022; Yang et al. 2017; Yaskolka Meir et al. 2021). Moreover, vitamin B12 production by duckweed endophyte bacteria has also been reported (Acosta et al. 2024), further enhancing its nutritional value.

Duckweed is an aquatic plant that maintains a close association with bacteria. Duckweed–microbe interaction studies are crucial for understanding the diversity, functional roles and evolutionary impacts. Such investigations elucidate microbiota assembly processes and reveal mechanisms by which duckweed actively controls recruitment and maintenance of microbial partners within the holobiont—an integrated ecological unit formed by the host plant and its associated microbial community (Simon et al. 2019). Duckweed has co‐evolved with its microbial community in mutualistic and commensal relationships, selectively recruiting beneficial bacteria through the release of exudates, metabolites and other chemical signals. Within the community, bacteria cooperate and contribute to duckweed metabolism and physiology through various activities, such as nitrogen metabolism and the degradation of toxic compounds (Acosta et al. 2020; Bunyoo et al. 2022; Ishizawa et al. 2017a; Khairina et al. 2021; Lu et al. 2014; Popržen et al. 2023; Toyama et al. 2022; Yamaga et al. 2010; Yamakawa et al. 2018). While the composition of these bacterial communities is influenced by environmental factors, the duckweed core microbiome, including families such as Beijerinckiaceae, Caulobacteraceae, Comamonadaceae, Methylophilaceae, Rhizobiaceae and Sphingomonadaceae, remains conserved and maintains functional stability across habitats (Acosta et al. 2020; Bunyoo et al. 2022; Inoue et al. 2022; Ishizawa, Kuroda, et al. 2020; Saimee et al. 2025).

Plant growth‐promoting bacteria (PGPB) benefit plants through various mechanisms including nitrogen fixation, phosphate solubilisation, siderophore production and phytohormone synthesis (Bulgarelli et al. 2013; Parnell et al. 2016). Plant growth promotion may occur through individual high‐impact bacterial strains, or synergistic interactions among bacterial communities where different species contribute complementary functions that collectively benefit the plant. Several duckweed‐associated bacteria (DAB) have been reported as PGPB. Acinetobacter calcoaceticus P23, isolated from Lemna aequinoctialis , represents the first reported PGPB associated with Lemna spp. (Ishizawa, Ogata, et al. 2020; Iwashita et al. 2020; Khairina et al. 2021; Phạm et al. 2025; Shuvro et al. 2023; Yamaga et al. 2010; Yamakawa et al. 2018). P23 exhibits phosphate‐solubilising activity (Khairina et al. 2021), phenol degradation capabilities (Yamaga et al. 2010) and the capacity to enhance chlorophyll content in Lemna minor and Lactuca sativa (Suzuki et al. 2014). Chryseobacterium strains promote Lemna gibba growth under nitrogen‐limiting conditions by utilising organic nitrogen sources, such as amino acids, thereby reducing competition for inorganic nitrogen between duckweed and bacteria (Khairina et al. 2021). Furthermore, Ensifer sp. SP4 promotes Spirodela polyrhiza growth through metabolic interactions. The duckweed assimilates amino acids, notably glutamic acid, from the bacteria, stimulating the synthesis of amino acids, chlorophyll and the enzyme ribulose‐1,5‐bisphosphate carboxylase/oxygenase (RuBisCO), thus enhancing photosynthesis and the Calvin cycle (Toyama et al. 2022). In our previous studies, DAB were isolated from S. polyrhiza enriched in a sterilised, partially treated wastewater effluent from a poultry farm. Several bacterial species were detected across multiple locations, suggesting their role as core DAB capable of promoting duckweed growth, including Acinetobacter soli , Acidovorax kalamii, Brevundimonas vesicularis , Pseudomonas toyotomiensis and Shinella curvata. Among these, Sh. curvata W12–8 exhibited the highest growth promotion in Hoagland medium with effect on plant growth (EPG) value of 208.72% based on frond proliferation, followed by Paracoccus marcusii W7–16 (171.31%), Novosphingobium subterraneum W5–13 (156.96%) and A. kalamii W7–18 (156.96%). Notably, Pa. marcusii W7–16 showed the greatest enhancement of duckweed growth under wastewater conditions, significantly increasing both the dry biomass and root length of S. polyrhiza (Boonmak et al. 2024).

Moreover, Wolffia species have a long history of consumption in Southeast Asia, particularly in Thailand, Laos, Myanmar and Cambodia. The main species commonly consumed are Wolffia arrhiza and Wolffia globosa (Appenroth et al. 2017). In recent years, Wolffia consumption has been increasing, leading to growth in commercial cultivation. However, Wolffia cultivation technology remains limited. The Wolffia sold in local markets is often harvested from natural ponds, which may carry risks of contamination with pathogenic microorganisms, helminth ova, toxins and various heavy metals (Boonyapookana et al. 2002; Grag and Chandra 1994; Markou et al. 2018; Mitrovic et al. 2005). To address these food safety concerns, closed‐system cultivation of Wolffia for consumption has been developing. Effective cultivation strategies must balance between biomass production which focuses on maximising yield through growth rate and biomass accumulation, and nutritional quality which targets the optimisation of protein content, essential amino acids profiles and bioactive compounds that determine Wolffia's value as a superfood. Wolffia growth and biomass quality are influenced by both abiotic factors, such as nutrients, light and temperature and biotic factors, including associated bacteria. Application of PGPB not only improves Wolffia cultivation process, but also reduces chemical fertiliser usage. However, Wolffia presents unique challenges as a plant model due to its tiny, rootless morphology, making it difficult to obtain and maintain axenic cultures. Consequently, the studies of PGPB associated with rootless duckweeds remain insufficient. This study aimed to isolate and investigate the effects of DAB on Wolffia growth promotion and biomass quality under axenic conditions to improve Wolffia cultivation and enhance its biomass quality as a valuable superfood.

Materials and Methods

2

DAB and Bacterial Isolation

2.1

A total of six duckweed samples were collected from six provinces in Thailand. Although more than one kind of duckweed was found in the same location, all samples comprised Wolffia mixed with other duckweed species (Figure 1 and Table S1). The duckweed samples were brought to the laboratory in an ice box within 1–2 days. To remove loosely attached bacteria from the plant surface, 0.5 g of the samples were washed twice with 50 mL of sterilised 0.85% saline solution and shaken on a rotary shaker at 150 rpm for 30 min. The duckweeds were crushed and suspended in 1 mL of 0.85% saline solution for dilution plating. After dilution, 0.1 mL of each dilution was spread in triplicate on Reasoner's 2A (R2A) agar (Himedia, India), and incubated at 20°C for 7–14 days. Although R2A agar is commonly used for the isolation of environmental heterotrophic bacteria, some culturable bacteria associated with duckweed may still not be recovered under these cultivation conditions. Distinct colonies were selected and purified on R2A agar. The isolates were stored in tryptic soy broth (TSB; Himedia, India) containing 20% glycerol at −20°C. Furthermore, eight PGPB for S. polyrhiza were also included in this study, namely Acidovorax kalamii W7–18, Agrobacterium pusense W9–7, Novosphingobium subterraneum W5–13, Paracoccus marcusii W7–16, Phenylobacterium haematophilum W12–15, Pseudomonas toyotomiensis W5–11, Ps. toyotomiensis W7–9 and Shinella curvata W12–8. All isolates were positive for IAA production, siderophore production and phosphate solubilisation. Furthermore, N. subterraneum W5–13 and Pa. marcusii W7–16 were able to produce carotenoids, including astaxanthin, zeaxanthin, ζ‐carotene and β‐zeacarotene, which are effective as strong antioxidants to alleviate plant stress (Boonmak et al. 2024).

Sampling locations in Thailand.

Bacterial Identification

2.2

Genomic DNA was extracted using PureLink Genomic DNA Mini kit (Invitrogen, USA) according to the manufacturer's protocol. The 16S rRNA gene was amplified using the primers 27F (5′‐AGAGTTTGATCMTGGCTCAG‐3′) and 1492R (5′‐TACGGYTACCTTGTTACGACTT‐3′). Polymerase chain reaction (PCR) was performed with a Bio‐Rad T100 Thermal Cycler (Bio‐Rad Laboratories, USA). The PCR products were sequenced using Barcode‐Tagged Sequencing (BTSeqTM) Service (U2Bio, Korea). The nucleotide sequences were compared to sequences of type strains in the EzBioCloud database (Yoon et al. 2017) for taxonomic identification. The sequences of closely related taxa were aligned by ClustalW (Higgins 1994). Phylogenetic trees were constructed from evolutionary distance data using the Neighbour‐Joining method (Saitou and Nei 1987) based on the Maximum Composite Likelihood model (Nei and Kumar 2000) of MEGA12 software (Kumar et al. 2024). The complete deletion method was used to remove gaps and missing data. Confidences for the phylogenetic tree were estimated from bootstrap analysis (1000 replicates) (Felsenstein 1985). The 16S rRNA gene sequences were deposited in the DNA Data Bank of Japan (DDBJ) database.

Preparation of Axenic Wolffia

2.3

Wolffia globosa DKU‐25W was obtained from site 25 which was a natural pond in Pathum Thani province, Thailand (Figure 1). One or two fronds within the same colony of the duckweed were transferred to a sterilised tube containing 1 mL of 0.3% sodium hypochlorite. The tube was agitated for 2–3 min until only the meristem retained its colour. The duckweed was transferred to 1 mL of 2% sodium thiosulfate, gently shaken for 30 s and washed by 1 mL of distilled water twice. Individual fronds were aseptically transferred to 5 mL of sterilised modified Hoagland medium (Toyama et al. 2018) supplemented with 1% sucrose and 100 μg/mL cefotaxime. The cultures were incubated at 28°C under a light intensity of 6000 lx with 16:8 h light:dark for 7 days. For maintenance, the axenic duckweeds were aseptically transferred to fresh 0.5× Schenk & Hildebrandt (SH) medium (Schenk and Hildebrandt 1972) without antibiotics every 10 days as recommended by the Rutgers Duckweed Stock Cooperative (RDSC, http://www.ruduckweed.org). Microbial contamination was checked by homogenising Wolffia fronds and plating the homogenate on R2A agar (Himedia, India). While no growth indicated axenic status, some bacteria may require alternative growth media and thus remain undetected. The clone showing no culturable microbial growth was propagated and maintained as a collection stock for further experiments.

Screening of PGPB by Co‐Cultivation Using Suspension Method

2.4

The plant growth‐promoting (PGP) ability of bacteria was evaluated by co‐cultivating them with duckweed using the suspension method modified from Phạm et al. (2025). Axenic duckweed, confirmed to be free of culturable microorganisms on R2A agar, was proliferated in 100 mL of sterilised 0.5× SH medium supplemented with 1% sucrose at 28°C under 6000 lx with 16:8 h light:dark for mass cultivation. The duckweed was transferred to the fresh medium at 7‐day intervals until a suitable amount was reached. Then, the duckweed was transferred to Hoagland medium for 5 days before testing. Bacterial culture was grown in 15 mL of 0.5× TSB (Himedia, India) at 28°C, 150 rpm for 48 h. The bacteria were collected by centrifugation at 8000 rpm for 5 min and washed twice by 10 mL of Hoagland medium. The optical cell density was measured at 600 nm using a spectrophotometer (Shimadzu, Japan). The bacterial culture was suspended in six‐well plates containing 10 mL of Hoagland medium to obtain an OD_600_ of 0.3. One loop of duckweed was placed in a well, in triplicate for each bacterial isolate. The axenic duckweed was used as a negative control. The co‐culturing plates were incubated under the previous conditions for 7 days, and measured surface coverage area of Days 0 and 7 using Ilastik programme (Romano et al. 2022). The relative growth rate (RGR) was calculated as a previous study using the equation (Ishizawa et al. 2017b):

$$[eqn]$$

where N 0 and N 7 are size in pixel‐based surface area measurements of fronds obtained from the software before and after 7 days of co‐cultivation, respectively. The effect on plant growth (EPG) was calculated as (Ishizawa et al. 2017b):

$$[eqn]$$

where G(T) is the RGR of duckweed in the presence of bacteria, and G(C) is that in the axenic duckweed. The standard error (SE) of the EPG was calculated from the following formula:

$$[eqn]$$

where SE(G(T)) and SE(G(C)) are the standard error of G(T) and G(C), respectively.

Co‐Cultivation of Wolffia With Selected PGPB Using Attachment Method

2.5

The selected PGPB from the screening test were further tested by attachment method (Phạm et al. 2025). Bacterial isolates were cultured in 15 mL of TSB at 28°C, 150 rpm for 48 h. Cells were harvested by centrifugation at 8000 rpm for 5 min and suspended in Hoagland medium. Fifty millilitres of the bacterial suspensions were adjusted to an OD_600_ of 0.3 and 1.0, then 0.5 g fresh weight of axenic Wolffia was added to each suspension. The co‐cultures were incubated at 28°C under light intensity of 6000 lx with 16:8 h light:dark for 24 h. One loop of bacterial‐attached duckweed was then transferred into six‐well plates containing 10 mL of Hoagland medium in triplicate. Axenic Wolffia was used as a negative control. The co‐cultures were incubated under the previous conditions for 7 days and analysed using Ilastik programme (Romano et al. 2022) to calculate the RGR and EPG values. For counting Wolffia‐attached bacteria on Day 1 of co‐cultivation, the duckweed was crushed and plated by dilution on TSA. Bacterial numbers were then compared with those in the original bacterial suspension.

Optimisation of Bacterial Cell Density for Wolffia Growth Promotion

2.6

From the previous experiment, the co‐cultivation test using the suspension method was selected to determine the optimal bacterial concentration for Wolffia growth promotion. The selected PGPB were co‐cultivated with Wolffia at OD_600_ values of 0.1, 0.3, 0.5, 1.0, 2.0 and 3.0, using the previously described suspension method in triplicate. The surface coverage area of Wolffia was analysed using the Ilastik programme (Romano et al. 2022) to calculate the RGR and EPG values.

Large‐Scale Co‐Cultivation and Their Effect on Duckweed

2.7

The axenic Wolffia was cultivated in the Hoagland medium for 5 days before testing. The PGPB were grown in 50 mL of TSB at 28°C, 150 rpm for 48 h. The bacterial cells were collected by centrifugation at 8000 rpm for 5 min and adjusted to an appropriate OD_600_ value with Hoagland medium. Subsequently, 0.5 g fresh weight of Wolffia was added to 300 mL of the bacterial suspension. The experiment was performed in triplicate and used axenic duckweed as the control. The co‐cultivation test was incubated under the previous conditions. After 7 days, the Wolffia biomass was used to determine the bacterial number on duckweed using dilution plating on R2A agar, and to analyse dry weight, chlorophyll, total starch and total protein content. Duckweed biomass was analysed by oven‐drying at 60°C for 24–36 h until a constant weight was achieved. Chlorophyll content was analysed by homogenising 0.01 g fresh weight of Wolffia with 1 mL of methanol in the dark for 1 h. The supernatant was then used for absorbance measurements at 650 and 665 nm, following the previously reported method by Ishizawa et al. (2021). Total starch and total protein content were analysed from 0.01 g dry sample of Wolffia using Total Starch Assay Kit (Megazyme, USA) and Pierce BCA Protein Assay Kits (Thermo Scientific, USA), respectively.

Determination of PGP Traits

2.8

PGP traits of the selected PGPB were evaluated including IAA production, siderophore production, as well as their ability to solubilise phosphate and potassium. For IAA production assays, 50 μL of active bacteria culture (OD_600_ of 0.2) were inoculated into 5 mL of 0.5× TSB supplemented with 0.05% glucose and 0.05% l‐tryptophan, incubated at 28°C, 150 rpm for 48 h. Supernatants were collected after centrifugation at 8500 rpm for 5 min. IAA production was analysed using HPLC (Nexera LC‐40 series, Shimadzu, Japan) equipped with a Cosmosil SC18‐MS‐II column (Nacalai Tesque, Japan). The mobile phase consisted of ethanol, acetic acid and water (60:1.5:40, v/v/v). Elution was performed at 40°C with a flow rate of 0.3 mL/min. Detection was performed at 280 nm using a UV detector, and the peaks were identified by comparison with a standard IAA solution (Sigma, USA) (Boonmak et al. 2024). Siderophore production, phosphate solubilisation and potassium solubilisation were assessed on chrome azurol S (CAS) agar (Schwyn and Neilands 1987), Pikovskaya agar (Pikovskaya 1948) and Aleksandrov agar (Aleksandrov 1958), respectively. All tests were performed in triplicate. Efficiency of activity was determined by dividing the diameter of the positive zone observed on the tested agar by the diameter of the colony.

Furthermore, nitrogen cycle‐related metabolism and bacterial activities were further evaluated. The nitrogen utilisation was tested as described previously by Khairina et al. (2021). Nitrate reduction and denitrification were assessed in 8 mL of R2A broth supplemented with 1 g/L KNO_3_ with an inverted Durham tube (Boonmak et al. 2024). Motility was tested in 0.5× TSA with 0.3% agar, prepared in deep tubes. Production of extracellular polymeric substance (EPS) was determined using crystal violet assay (Leeprasert et al. 2022). All experiments were performed in triplicate.

Statistical Analysis

2.9

Statistical analyses were performed using SPSS Statistics software and R version 4.4.3 (RC Team 2024) via RStudio server on posit.cloud. One‐way analysis of variance (ANOVA), followed by Duncan's multiple range test as a post hoc test, and independent samples t‐tests were used to evaluate statistical significance. Pearson's and Spearman's rank correlation coefficient, and principal component analysis (PCA) were used to analyse the relationships between bacterial PGP traits and Wolffia growth and biomass quality. Correlation matrices were visualised as correlograms in RStudio using ‘psych’ and ‘Hmisc’ packages (Harrell Jr and Dupont 2025; Revelle 2024). PCA was performed using the ‘prcomp’ function in RStudio. Statistical significance was set at p < 0.05.

Result

3

Bacterial Isolation and Identification

3.1

A total of 108 DAB isolates were obtained from R2A agar after incubation at 20°C for 7–10 days. The bacterial count in duckweed samples ranged from 1.60 × 10^6^ to 2.79 × 10^7^ CFU/g. The highest counts were observed in sample from Klaeng District, Rayong Province (Site 30), followed by Lam Luk Ka District, Pathum Thani Province (Site 25), both of which are natural ponds. In contrast, the lowest count was detected in sample from Pran Buri District, Prachuap Khiri Khan Province (Site 13), which is a garbage run‐off site in an urban area (Table S1). Hence, pollution may have reduced the DAB abundance. In total, 66 species across 41 genera were identified. Most isolates belonged to the phylum Pseudomonadota (66%), particularly the class Alphaproteobacteria (Figure 2). Among these, 57 species were known species. Ten isolates showed < 98.7% similarity to all valid type strains, indicating they likely represent nine novel species (Table 1). These candidates were affiliated with the genera Allorhizobium, Chryseobacterium, Devosia, Formosimonas, Novosphingobium, Pedobacter, Pleomorphomonas, Roseateles and Uliginosibacterium, based on phylogenetic placement (Figures S1–S3). The Shannon–Wiener diversity index of DAB ranged from 1.75 to 2.94 (Table 2). The highest diversity was found at site 25 (Figure 2), a natural pond environment where four duckweed genera (Landoltia, Lemna, Spirodela and Wolffia) coexisted. This contrasts with other sampling sites, where only one or two duckweed genera were present. The most common genus was Rhizobium, with at least one species across samples, including R. glycinendophyticum, R. ipomoeae, R. rhizoryzae, R. rosettiformans and R. wuzhouense.

Frequency of occurrence of bacteria in duckweed samples.

Effects of DAB on Wolffia Growth

3.2

Screening for PGPB was performed by co‐cultivating the axenic W. globosa DKU‐25W with the DAB using the suspension method in the Hoagland medium for 7 days. EPG values of the bacterial isolates on duckweed were presented in Table S2. Among these, 104 isolates enhanced duckweed growth by 2.7%–77.83%, with 12 isolates showing high EPG value > 50% (Figure 3a). The highest EPG value was found in Leclercia adecarboxylata M9 (77.83% ± 2.28%), followed by Enterobacter cloacae M2 (77.80% ± 2.59%), Shinella curvata W12–8 (77.75% ± 3.23%), Paracoccus marcusii W7–16 (71.74% ± 2.41%) and Pseudomonas toyotomiensis W5–11 (71.29% ± 1.53%), respectively. In contrast, 12 isolates exhibited inhibitory effects on duckweed growth ranging from −2.87% to −42.69%, with the strongest inhibition observed in Sporosarcina contaminans 30–25, which exhibited an EPG value of −42.69% ± 5.40%. Bacterial isolates selected for further experiments were required to have EPG values > 50% and no known reports of pathogenicity to humans or plants. Therefore, L. adecarboxylata M9 and E. cloacae M2, which are known for their pathogenicity, were excluded from this study (Davin‐Regli et al. 2019; Elbehiry et al. 2024; Zayet et al. 2021). Based on the criteria, six PGPB were selected for further experiments including Sh. curvata W12–8, Pa. marcusii W7–16, Ps. toyotomiensis W5–11, Novosphingobium subterraneum W5–13 (with EPG value of 59.89% ± 5.65%), Rhizobium rosettiformans 30–37 (55.67% ± 2.61%) and Asticcacaulis excentricus 25–48 (54.67% ± 1.47%) (Figure 3b). All selected PGPB belonged to the phylum Pseudomonadota. Five isolates were classified in the class Alphaproteobacteria; only Pseudomonas toyotomiensis belonged to the class Gammaproteobacteria (Figure 4).

EPG of PGPB on W. globosa DKU‐25W after co‐cultivation in the Hoagland medium for 7 days (a) and growth comparison with selected PGPB and axenic control (b). Black bars indicate high EPG values > 50%. Error bars represent standard error (n = 3).

Phylogenetic placement of six selected PGPB and 48 closely related taxa based on the sequence analysis of the 16S rRNA gene using the Neighbour‐Joining method based on the maximum composite likelihood model. The complete deletion method was used to remove gaps and missing data, the final dataset contained 1262 positions. Name in bold type are the isolates obtained from this study. Numbers on branches indicate percentages of bootstrap sampling (> 50%), derived from 1000 samples. Bars, 0.02 substitutions per nucleotide position.

Optimisation of Duckweed Co‐Cultivation With DAB

3.3

Before large‐scale cultivation, the co‐cultivation method and bacterial concentration were evaluated. Co‐cultivation of W. globosa with three effective PGPB (W12–8, W7–16 and W5–11) was performed using the attachment method at OD_600_ of 0.3 and 1.0. Bacterial count revealed that the initial number of duckweed‐attached bacteria at Day 0 was comparable to the bacterial density in the cell suspension, confirming successful PGPB colonisation on duckweed (Table S3). However, the initial bacterial concentration had minimal effect on plant growth promotion (0.68%–9.84% EPG), with no statistically significant differences observed by t‐test (p > 0.05) in the attached growth co‐cultivation (Figure S4). Furthermore, the plant growth promotion achieved by these PGPB using the attachment method was significantly lower compared to the suspension growth method (71.29%–77.75% EPG), as confirmed by one‐way ANOVA (p < 0.05). These findings indicate that the attachment method is unsuitable for biomass enhancement in Wolffia. Hence, suspension culture was selected for large‐scale co‐cultivation.

Bacterial concentration for co‐cultivation using the suspension method was optimised by evaluating six selected PGPB at varying initial densities (Figure 5). Increasing initial bacterial density significantly enhanced duckweed growth up to an optimal threshold (p < 0.05, one‐way ANOVA), beyond which no further significant improvements were observed. The optimal cell density differed among isolates. Pa. marcusii W7–16 and

N. subterraneum

W5–13 showed the highest EPG at OD 600 of 1.0 (117.26*% and* 85.90*%, respectively*). Interestingly, Pa. marcusii W7–16 exhibited a slight decline in EPG at higher concentrations, whereas the EPG of W5–13 decreased considerably to 14.82%–22.38%. Sh. curvata W12–8, Ps. toyotomiensis W5–11 and

A. excentricus 25*–48 showed the highest EPG values at OD 600 of 3.0, with their EPG increasing as initial cell densities increased. However, no significant differences in EPG values were observed for W12–8 at OD 600 of 2.0–3.0 or W5–11 at OD 600 of 1.0–*3.0, and these data are summarised in Table S4. Based on these results, the lowest concentrations that achieved maximum growth were selected for large‐scale co‐cultivation. Therefore, OD 600 of 2.0 and 1.0 were used for Sh. curvata W12–8, and Ps. toyotomiensis W5–11, respectively. A. excentricus 25–48 showed the highest EPG at OD 600 of 3.0, with potential for further improvement at higher densities. Interestingly, R. rosettiformans 30–37 demonstrated high EPG values at both low (OD_600_ = 0.1) and high (OD_600_ = 2.0) densities. These values were significantly higher than at intermediate densities, indicating a non‐linear dose–response relationship that requires further study.

EPG of PGPB after co‐cultivation under various bacterial density conditions in Hoagland medium for 7 days. Error bars indicate standard error (n = 3). Asterisks () mark the cell density selected for large‐scale co‐cultivation. Different letters (a‐e) indicate significant differences between treatments (p < 0.05, one‐way ANOVA with Duncan's multiple range test).*

Evaluation of PGPB for PGP Traits

3.4

Typical PGP traits (IAA and siderophore production, and solubilisation of phosphate and potassium), nitrogen‐related metabolisms and plant colonisation factors of six PGPB were summarised in Table 3. These traits contribute to nutrient availability for plants and may play a role in enhancing plant tolerance to abiotic stress. All isolates could produce IAA with concentrations ranging from 4.45 to 109.03 μg/mL. They also exhibited phosphate solubilisation and siderophore production with efficiency of activity of 1.00–3.11 and 1.12–1.61, respectively. Potassium solubilisation activity (1.04–6.40) was observed in all isolates except W7–16 and W5–13. Bacteria capable of utilising a wide range of nitrogen sources are generally better adapted to complex environments and more competitive. However, in nitrogen‐limited conditions, they may outcompete duckweed for nitrogen sources. All isolates utilised amino acids as organic nitrogen sources, but W7–16, W5–13 and 30–37 could not utilise NH_4_ and NO_3_. All PGPB were positive on nitrate reduction, but not denitrification and nitrogen fixation.

Plant surface colonisation ability, which is a crucial factor for PGPB stability and efficacy, was assessed through bacterial motility and EPS production. Only W5–11 and 30–37 exhibited motility, suggesting chemotactic ability to achieve effective colonisation. All isolates produced EPS, with W7–16 showing the highest level, which may contribute to its superior colonisation and PGP activity.

Large‐Scale Co‐Cultivation and Plant Growth‐Promoting Effects

3.5

Large‐scale co‐cultivation of W. globosa with six PGPB was performed using optimal bacterial concentrations. After 7 days, duckweed biomass was analysed for dry weight, chlorophyll, starch and protein content (Figure 6 and Table 4). Duckweed cultured with Ps. toyotomiensis W5–11 showed a biomass increase, with fronds forming layers and greener colour (Figure 7). The increased frond production led to the highest dry weight (3.18‐fold greater than the control) and highest chlorophyll content (2.75‐fold). The other PGPB also significantly enhanced dry weight (1.76–2.20‐fold) (p < 0.05, one‐way ANOVA), except R. rosettiformans 30–37. For chlorophyll content, the other isolates significantly increased production (1.60–2.25‐fold) (p < 0.05, one‐way ANOVA), except for Sh. curvata W12–8 and A. excentricus 25–48. Protein enhancement is one of the key goals for improving duckweed quality. Axenic Wolffia exhibited a total protein content of 8.9% of dry weight, whereas Wolffia co‐cultivated with PGPB showed total protein content ranging from 6.21% to 10.91% of dry weight. The highest content was observed in Ps. toyotomiensis W5–11 (10.91%), followed by R. rosettiformans 30–37 (9.50%) and Sh. curvata W12–8 (9.39%). In contrast, Pa. marcusii W7–16 and N. subterraneum W5–13 decreased protein levels. Among PGPB, only R. rosettiformans 30–37 significantly increased total starch content (4.98%) (p < 0.05, one‐way ANOVA).

Effects of PGPB on dry weight (a), chlorophyll content (b), protein content (c) and starch content (d) after large‐scale co‐cultivation in Hoagland medium for 7 days. Error bars indicate standard deviation (n = 3). Different letters (a–c) indicate significant differences between treatments (p < 0.05, one‐way ANOVA with Duncan's multiple range test).

TABLE 4: PGP effects of PGPB after large‐scale co‐cultivation with Wolffia globosa DKU25‐W in Hoagland medium for 7 days.

Growth of W. globosa DKU‐25W in large‐scale co‐cultivation with selected PGPB compared to axenic duckweed (control) in Hoagland medium for 7 days, demonstrating visible differences in biomass production and chlorophyll content between treatments.

While PGP mechanisms are diverse and often strain‐specific, making them difficult to fully elucidate, we explored potential correlations between common PGP traits and duckweed growth parameters. Despite the limited sample size (six isolates), this analysis aimed to identify possible relationships that could guide future research. Plant growth parameters were analysed for correlations with bacterial PGP activities using Pearson's correlation coefficient to determine the direction and strength of these relationships (Figure 8). IAA production strongly positively correlated with all growth parameters, particularly protein production (r = 0.85). Protein accumulation was associated with bacterial ability to produce IAA, solubilise potassium, followed by phosphate solubilisation. Dry weight showed positive correlations with potassium solubilisation, phosphate solubilisation and IAA production, but a slight negative correlation with siderophore production. Starch accumulation was positively correlated with IAA production and negatively with phosphate solubilisation, suggesting stress‐induced starch synthesis under low phosphate availability conditions. Chlorophyll content showed a strong negative correlation with siderophore production (r = −0.59). Overall, siderophore production exhibited negative correlations with nearly all plant growth parameters, except for starch accumulation, which was only marginally affected. To assess the reliability of these correlations, additional analyses were conducted using Spearman's rank correlation and PCA as shown in Figures S5 and S6, respectively. Spearman's rank correlation supported the trends observed in Pearson's correlation analysis. Bacterial IAA production and potassium solubilisation were associated with duckweed protein content, whereas siderophore production showed an inverse relationship with dry weight and a positive association with starch content. PCA provided a multivariate overview of these patterns, with the first two principal components explaining 69.2% of the total variability (PC1: 42.7%, PC2: 26.5%). In the PCA biplot, IAA production and nutrient‐solubilising traits clustered with protein content along PC1, while siderophore production was separated from dry weight and chlorophyll content and aligned with starch content along PC2.

Correlogram matrix of Pearson's correlation coefficients between bacterial PGP traits and Wolffia growth parameters. Asterisks () indicate statistically significant correlations (p < 0.05, Pearson correlation test, n = 6).*

Discussion

4

This study evaluated the PGP potential of DAB on W. globosa , focusing on biomass productivity and quality. A total of 108 bacterial isolates were obtained from duckweeds collected from six provinces across Thailand. These isolates were taxonomically classified into 66 species across 41 genera. The bacterial community exhibited high diversity, spanning four phyla: Actinomycetota, Bacillota, Bacteroidota and Pseudomonadota, with the majority belonging to the class Alphaproteobacteria, consistent with previous reports on natural duckweed microbiomes (Acosta et al. 2020; Bunyoo et al. 2022; Matsuzawa et al. 2010). Water quality and duckweed species both influence the composition of DAB communities. Bacterial abundance was higher in duckweed collected from natural ponds than in plants from urban garbage run‐off sites. This pattern suggested that pollutants near waste deposits created conditions that were unfavourable for the growth of DAB. Site 25 showed the highest bacterial diversity and contained four coexisting duckweed genera. This observation supports the idea that greater duckweed species richness can promote microbial diversity. Despite environmental variability affecting community composition, DAB maintain stable core taxa through selective recruitment mechanisms. Duckweeds selectively recruit bacterial communities through the release of exudates, metabolites and chemical signals that establish mutualistic relationships within the holobiont. This selective recruitment shapes unique microbiomes with conserved core taxa that remain stable across diverse environments. Although morphological differences between rootless and rooted duckweeds influence microbial colonisation patterns (Saimee et al. 2025), core bacterial families, including Beijerinckiaceae, Caulobacteraceae, Comamonadaceae, Methylophilaceae, Rhizobiaceae and Sphingomonadaceae, were consistently maintained across samples (Acosta et al. 2020; Inoue et al. 2022; Ishizawa, Kuroda, et al. 2020). Among these core taxa, four families were observed in this study, including Caulobacteraceae, Comamonadaceae, Rhizobiaceae and Sphingomonadaceae. The most abundant family was Rhizobiaceae (25 isolates, 20.5% of all new isolates), which was found in all samples, followed by Comamonadaceae (six isolates, 4.9%) and Sphingomonadaceae (six isolates, 4.9%), found in four locations. These core bacteria may function as both PGPB and keystone species that facilitate inter‐microbial communication and coordinate community networks, thereby supporting a stable and functional duckweed microbiome (Saimee et al. 2025).

All new isolates, along with eight PGPB for S. polyrhiza , were evaluated for their PGP effects on W. globosa . Most isolates promoted plant growth, while only 12 isolates showed inhibitory effects. Among PGPB, 12 isolates demonstrated high PGP potential with EPG values > 50%. However, some of these isolates belonged to taxa with reported pathogenicity. For example, Leclercia adecarboxylata is a rare opportunistic pathogen in immunocompromised individuals (Zayet et al. 2021); Enterobacter cloacae is a multidrug‐resistant nosocomial pathogen (Davin‐Regli and Pagès 2015) and Agrobacterium pusense is known to cause crown gall disease in plants (Basavand et al. 2021). These isolates are therefore unsuitable for environmental applications. Six PGPB demonstrated high efficacy and safety for duckweed, including Sh. curvata W12–8, Pa. marcusii W7–16, Ps. toyotomiensis W5–11, N. subterraneum W5–13, R. rosettiformans 30–37 and A. excentricus 25–48 (EPG: 54.67%–77.75%; Figure 3b). W. globosa co‐cultivated with these PGPB in the Hoagland medium showed higher RGRs than axenic duckweed in nutrient‐rich SH medium (Figure S7). This result indicated that PGPB could promote duckweed growth in the Hoagland medium comparable to SH medium, suggesting their potential as biofertiliser candidates to replace chemical fertilisers. Although the PGP efficiency on W. globosa growth was lower than on S. polyrhiza with EPG values 2.7%–77.83% and 7.07%–208.72%, respectively (Table S5), the four top‐performing PGPB were the same for both species (Boonmak et al. 2024). Few studies have reported cross‐species effects of DAB. For instance, Acinetobacter calcoaceticus P23, isolated from L. aequinoctialis , most effectively promoted the growth of its original host, followed by other Lemna spp. and S. polyrhiza (Toyama et al. 2017), possibly due to phylogenetic proximity. In contrast, the effectiveness of S. polyrhiza ‐derived PGPB on W. globosa likely stems from conserved biochemical traits, such as compatible root exudates, strong surface adhesion and phytohormone‐mediated responses that support chlorophyll synthesis and biomass accumulation (Hu et al. 2018; Toyama et al. 2022; Zhalnina et al. 2018). These findings underscore that functional host–microbe compatibility depends primarily on specific biochemical mechanisms including exudate chemistry, adhesion properties and hormone signalling pathways, rather than mere phylogenetic relatedness, suggesting that PGPB applications may be effective across related plant species that share key physiological traits.

Fundamental PGP traits were evaluated in the selected PGPB strains. All strains showed positive activities in IAA production, siderophore production and phosphate solubilisation (Table 3). The overall impact of these traits on duckweed growth remained unclear. Many studies on duckweed‐associated PGPB have reported varied results. For instance, Ac. calcoaceticus P23 possesses phosphate‐solubilising activity, Acidobacteria sp. F‐183 produces IAA, while Acidobacteria sp. TBR‐22 exhibited no positive PGP traits (Yoneda et al. 2021). Similarly, Chryseobacterium spp. promoted L. gibba growth under nitrogen‐limited conditions through IAA and siderophore production (Khairina et al. 2021), and Ensifer sp. SP4, a PGPB of S. polyrhiza , was only capable of siderophore production (Toyama et al. 2022). Notably, Ps. toyotomiensis W5–11 exhibited particularly high IAA production in tryptophan‐supplemented media (109.03 ± 11.77 μg/mL). Although exogenous IAA had minimal impact on L. minor growth (Utami et al. 2018), most endophytic bacteria associated with Landoltia, Lemna and Wolffia tested positive in the Salkowski assay, suggesting that internal IAA may still play an important physiological role in these hosts (Gilbert et al. 2022). Moreover, Pseudomonas putida A3–104/5, capable of both producing and degrading IAA, showed oxidative stress resistance when exposed to hydrogen peroxide, implying that IAA may serve a protective role in bacterial stress responses (Popržen et al. 2023). Regarding nitrogen metabolism, none of the six PGPB exhibited nitrogen fixation or denitrification activity, although all isolates were positive for nitrate reduction. Notably, R. rosettiformans has been reported to carry nifH gene, a key enzyme of nitrogen fixation, though its expression may be host‐specific and primarily associated with legumes (Burbano et al. 2011; Kaur et al. 2011). The nitrogen utilisation profiles of the isolates were also examined, as competition for inorganic nitrogen compounds can occur between bacteria and duckweed under nitrogen‐limited conditions (Khairina et al. 2021). Pa. marcusii W7–16, N. subterraneum W5–13 and R. rosettiformans 30–37 could not utilise NH_4_ ^+^ and NO_3_ ^−^, indicating their potential applicability as PGPB under nitrogen‐limited ecosystems. In addition to the metabolic activities described above, nutrient recycling from bacterial biomass may also contribute to growth promotion in co‐cultivation systems. Ishizawa et al. (2023) demonstrated that Pelomonas saccharophila MRB3 underwent spontaneous cell lysis, releasing NH_4_ ^+^ and PO_4_ ^3−^ that were subsequently utilised by duckweed, with a considerable fraction of bacterial nitrogen converted to plant‐available forms through this process. Such nutrient release from bacterial cells could potentially occur in suspension co‐cultivation systems, where bacterial cells remain in direct contact with duckweed fronds throughout the cultivation period. However, determining the contribution of this mechanism to growth promotion would require monitoring of both bacterial population and nutrient dynamics over time, which should be investigated in future studies.

The attachment method was less effective in growth promotion, regardless of the initial bacterial concentrations. This reduced efficacy may be attributed to the limited surface area available for bacterial attachment on W. globosa , restricting the total number of associated cells. In the attached growth system, initial bacterial density primarily influences velocity of colonisation but not the final population (Ishizawa et al. 2019), and bacterial attachment tends to support plant growth temporarily, depending on strain‐specific behaviour (Ishizawa, Ogata, et al. 2020). The absence of roots and root exudates in W. globosa likely creates different microenvironments compared to rooted plants, possibly affecting bacterial colonisation stability and plant‐microbe interactions, as some strains preferentially attach to roots rather than fronds (Iwashita et al. 2020; Yoneda et al. 2021). To investigate the colonisation dynamics and interactions between PGPB and Wolffia, temporal analysis of bacterial density on the frond surface using microscopy should be performed in future studies. Therefore, the suspension method was selected for evaluating PGPB effects on Wolffia biomass quality. The optimal initial cell concentration varied among the six PGPB (Table S6). These concentrations likely represented optimal concentrations that promote plant growth without inducing adverse plant defence responses. Growth promotion generally increased with bacterial inoculum density until reaching an optimal threshold. Beyond this point, excessive bacterial populations may limit duckweed growth through physiological saturation or environmental constraints such as light, water or nutrient availability, leading to a significantly reduced growth‐promoting efficacy. This may trigger plant immune responses or phytotoxic effects (Madhaiyan et al. 2007; Pillay and Nowak 1997), leading to suboptimal growth and intensify nutrient competition both between bacteria and plant, as well as among bacterial cells (Lopes et al. 2021). However, for some isolates, plant growth benefits stabilised at the optimal concentration, with higher bacterial densities showing no additional positive or negative effects on growth.

In this study, the protein content of axenic W. globosa grown under nutrient‐limited Hoagland conditions was relatively low (8.91*%*), compared with farm‐produced Wolffia in Thailand (23.93%–31.50%) (Boonarsa et al. 2024; Dhamaratana et al. 2025; On‐Nom et al. 2023). Under nutrient‐rich conditions, increased frond proliferation and chlorophyll biosynthesis often lead to increased protein accumulation and reduced starch storage (Cui et al. 2011; Reid 2004; Toyama et al. 2022; Zhao et al. 2015). In this study, duckweed co‐cultivated with PGPB exhibited low protein contents (6.82%–10.91%). This likely reflects the limited nutrients in Hoagland medium, which are less enriched than alternative media or farm‐applied fertilisers (Appenroth et al. 2018). Under such conditions, duckweed may have partially relied on bacterial‐mediated nutrient supply (Toyama et al. 2017; Zenir et al. 2023), which was insufficient for high protein accumulation. Notably, protein levels in most treatments were comparable to the axenic control, except for Pa. marcusii W7–16 and N. subterraneum W5–13, which significantly increased dry weight but reduced protein content. This reduction may indicate that these strains either suppressed protein biosynthesis or diverted resources towards other metabolic processes in W. globosa. Starch accumulation in duckweeds is commonly induced under stress conditions, such as nutrient deficiencies (especially nitrogen or phosphate) leading to starch contents as high as 28%–38% (Fujita et al. 2016; Li et al. 2021). Salinity, heavy metal toxicity and other environmental factors also trigger starch accumulation (de Morais et al. 2019; Sree and Appenroth 2022). In this study, all PGPB induced relatively low starch accumulation (2.18%–4.98%). Only R. rosettiformans 30–37 increased starch accumulation compared to other treatments, possibly due to declining bacterial attachment or nutrient competition that induced mild stress in duckweed. The stability of PGPB colonisation on the frond surface is a critical limitation that can influence the overall potential of PGP (Ishizawa, Ogata, et al. 2020).

Additionally, we analysed the relationships between typically bacterial PGP traits and Wolffia growth and biomass quality. Chlorophyll production is an important factor in plant growth due to its essential role in photosynthesis (Ishizawa et al. 2021; Liu et al. 2021; Toyama et al. 2022). Photosynthesis provides carbon resources that support frond proliferation and protein/starch synthesis, potentially affecting dry weight. Plants allocate photosynthetic carbohydrates differently depending on environmental conditions (Hartmann et al. 2020). The correlations between chlorophyll content and other variables showed that increased chlorophyll was positively associated with dry weight, weakly positively correlated with protein content, and negatively correlated with starch content. These trends are consistent with the allocation of carbohydrates towards biomass production. During co‐cultivation test between Wolffia and PGPB, all treatments used Hoagland medium without additional tryptophan. Therefore, bacteria could only retrieve tryptophan from its biosynthesis pathways or secretion from Wolffia. In case of Pseudomonas toyotomiensis W5–11 which produced highest IAA in this study, trpA and trpB which encode tryptophan synthase subunit alpha (QSL94784.1) and tryptophan synthase subunit beta (QSL94785.1) have been reported in whole genome sequence of P. toyotomiensis strain SM2 (CP070505) in GenBank database. Similarly, these two genes also found in the whole genome sequence of Rhizobium rosettiformans strain MAE2‐X (CP032405). This suggests that both P. toyotomiensis W5–11 and R. rosettiformans 30–37 might synthesise tryptophan and use it for IAA production. Even though the number of tested PGPB was limited, our results suggested that bacterial IAA production was associated with Wolffia protein content. This observation aligned with previous findings in L. minor , which indicated that endogenously produced IAA may be more effective than exogenous application (Utami et al. 2018). Unlike in terrestrial plants, exogenous IAA could enhance chlorophyll content and photosynthetic rate in tomato (Liu et al. 2024) and Syringa villosa (Jin et al. 2023). Based on mechanistic understanding from model plants such as Arabidopsis thaliana , IAA produced by PGPB could contribute to protein accumulation through auxin‐responsive signalling pathways (Paque and Weijers 2016; Woodward and Bartel 2005). Bacterially derived IAA has been proposed to act through the conserved ARF/Aux–IAA signalling pathway, which regulates genes involved in nitrogen assimilation (Guilfoyle and Hagen 2007; Weijers et al. 2005). This regulation may involve increased activity of nitrate reductase, glutamine synthetase and glutamate synthase, which convert inorganic nitrogen into amino acids required for protein synthesis (Mashiguchi et al. 2011; Zhao et al. 2001). IAA signalling may also promote the transcription and translation of genes encoding ribosomal proteins and translational machinery, thereby increasing cellular capacity for protein synthesis (Guilfoyle and Hagen 2012). However, direct evidence linking auxin signalling to nitrogen‐assimilating enzyme activity in Wolffia remains limited and warrants further investigation. Elevated photosynthesis has been reported to increase the RGR and metabolite profiles of S. polyrhiza , especially its amino acid profile. Ensifer sp. SP4, despite lacking detectable IAA production activity, promoted S. polyrhiza protein content by 2.4‐fold through direct transfer of organic nitrogen compounds, particularly glutamic acid, possibly by triggering increases in photosynthetic and metabolomic activities (Toyama et al. 2022). These findings suggested that PGPB operate through multiple mechanisms, including both phytohormone production and direct metabolite transfer, which can affect plant physiology and metabolomic profiles. This is consistent with our observation of a positive correlation between bacterial IAA production and Wolffia protein content. Despite plants requiring essential macronutrients such as nitrogen, phosphorus and potassium, phosphate solubilisation barely influenced Wolffia growth in this study, possibly due to the universal phosphate‐solubilising ability of the selected PGPB isolates, which reduced the observable differences among treatments. Reports on potassium‐solubilising bacteria (KSB) associated with duckweed are scarce. However, KSB are recognised as important biofertilisers in agriculture (Ahmad et al. 2016; Bakhshandeh et al. 2017; Xiao et al. 2017). In this study, K‐solubilising isolates significantly enhanced dry weight and protein content, highlighting the potential importance of KSB for duckweed growth. Conversely, siderophore production was negatively associated with chlorophyll synthesis in Wolffia, which coincided with reduced plant biomass and protein content. The reduction in chlorophyll content likely decreased photosynthetic efficiency, subsequently triggering stress responses in the plant. Hence, starch accumulation, which is a typical stress response in duckweed, was observed to be positively associated with siderophore production, as shown by Spearman correlation analysis and PCA (Figures S5 and S6). This finding contrasts with previous knowledge in which siderophores are generally considered beneficial for plant growth by facilitating iron acquisition. Such stress may result from incompatibility between bacterial siderophore types and the plant requirements, altered uptake of certain heavy metals or overproduction of siderophores that disrupt iron homeostasis (Singh et al. 2022). The inhibitory effects of siderophores on plant growth have been reported by Eigharlou et al. (2024), who demonstrated that siderophores from Amycolatopsis lurida inhibited both root and shoot growth of ryegrass and redroot, with enhanced suppression observed when siderophores functioned synergistically with other secondary metabolites (Eigharlou et al. 2024). Furthermore, the study revealed differential sensitivity between plant types, with ryegrass (monocotyledonous) exhibiting greater susceptibility to siderophore‐induced toxicity compared to redroot (dicotyledonous). It is noteworthy that Wolffia belongs to the monocotyledonous group. This taxonomic‐based sensitivity difference may relate to the negative correlation between siderophore production and Wolffia growth observed in our study. Nevertheless, the specific effects of siderophore production on Wolffia require further experimental validation. These findings revealed associations between specific bacterial PGP traits and Wolffia growth and biomass quality, suggesting the relevance of targeted PGPB selection based on these traits. These correlations may serve as foundational indicators for the selection of effective PGPB. Further investigation of the molecular and ecological mechanisms involving these interactions may unlock new strategies for enhancing productivity in sustainable aquatic plant cultivation systems.

Conclusion

5

This study emphasises the PGP potential of DAB, especially taxa associated with Wolffia and Spirodela, on W. globosa . Ps. toyotomiensis showed the highest potential by enhancing dry biomass and chlorophyll content. Bacterial traits, particularly IAA production and potassium solubilisation, influenced Wolffia growth promotion. Overall, these complex duckweed–bacteria interactions suggest that duckweed PGPB hold promise as bioinoculants for improving Wolffia cultivation.

Author Contributions

S.K. performed the data curation, investigation, formal analysis, writing – original draft and editing. M.M. provided project administration, resources, and editing – original draft. C.B. provided conceptualisation, funding acquisition, project administration, resources, supervision, formal analysis, investigation, methodology, writing – original draft, and editing. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by Kasetsart University through the Graduate School Fellowship Programme.

Ethics Statement

The authors have nothing to report.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Sampling locations, on‐site water qualities and bacterial counts. Table S2: Effects on plant growth (EPG) of newly DAB isolates on Wolffia globosa DKU25‐W after co‐cultivation in the modified Hoagland medium for 7 days. Table S3: Total count of bacteria in cell suspension and on Wolffia globosa DKU25‐W after attachment in six‐well plate for 24 h. Table S4: EPG values of the potential PGPB at various cell concentrations on W. globosa DKU25‐W after co‐cultivation in the modified Hoagland medium for 7 days. Table S5: EPG values of the potential PGPB from previous study on W. globosa DKU25‐W after co‐cultivation in the modified Hoagland medium for 7 days. Table S6: Total count of bacteria on Wolffia globosa DKU25‐W before and after large‐scale co‐cultivation for 7 days. Figure S1: Phylogenetic placement of the isolate 30–45, M10 and closely related taxa in the Phylum Bacteriodota based on the sequences analysis of the 16S rRNA gene using the Neighbour‐Joining method based on the maximum composite likelihood model. The complete deletion method was used to remove gaps and missing data, the final dataset contained 1282 positions. Name in bold type are the isolates obtained from this study. Numbers on branches indicate percentages of bootstrap sampling (> 50%), derived from 1000 samples. Bars, 0.05 substitutions per nucleotide position. Figure S2: Phylogenetic placement of the isolate obtained from this study and closely related taxa in the Phylum Pseudomonadota, Class Alphaproteobacteria, based on the sequences analysis of the 16S rRNA gene using the Neighbour‐Joining method based on the maximum composite likelihood model. The complete deletion method was used to remove gaps and missing data, the final dataset contained 1273 positions. Name in bold type are the isolates obtained from this study. Numbers on branches indicate percentages of bootstrap sampling (> 50%), derived from 1000 samples. Bars, 0.02 substitutions per nucleotide position. Figure S3: Phylogenetic placement of the isolate obtained from this study and closely related taxa in the Phylum Pseudomonadota, Class Betaproteobacteria, based on the sequences analysis of the 16S rRNA gene using the Neighbour‐Joining method based on the maximum composite likelihood model. The complete deletion method was used to remove gaps and missing data, the final dataset contained 1347 positions. Name in bold type are the isolates obtained from this study. Numbers on branches indicate percentages of bootstrap sampling (> 50%), derived from 1000 samples. Bars, 0.02 substitutions per nucleotide position. Figure S4: EPG of three PGPB after co‐cultivation at bacterial concentrations of OD600 = 0.3 and 1.0 using an attachment method in Hoagland medium for 7 days. Error bar is standard error, SE. Figure S5: Correlogram displaying Spearman's correlation coefficients among various and negative correlations, respectively. Circle size represents correlation strength. Numbers inside show the exact correlation values. Asterisks indicate statistically significant correlations (p < 0.05). Figure S6: Principal component analysis showing relationships between bacterial PGP traits (red arrows) and Wolffia biomass quality parameters (blue arrows). The biplot was constructed using the ‘prcomp’ function in R. Arrow directions and length indicated the strength and direction of correlations between variables. Figure S7: Relative growth rate of axenic W. globosa DKU‐25W after co‐cultivation with PGPB in the Hoagland medium compared with axenic W. globosa DKU‐25W grown in SH medium. Error bar is standard error, SE.