The nine macroalgae tested in this study were collected along different seasons between 2021 and 2022 at Punta del Hidalgo (28° 33’ 37.32’’ N, 16° 20’ 7.843’’ W) on the north coast of the island of Tenerife (Canary Islands) at low tide in the intertidal zone, in accordance with the Nagoya Protocol, permit reference: ESNC102. These algae were selected for their extensive study in the archipelago in terms of morphology and genetics. Specimen identification was confirmed using a Leica DM 500 Microsystems optical microscope (Wetzlar, Germany), following the classification proposed by Guiry and Guiry in 2023. A voucher specimen of each species was deposited at the University of La Laguna (Tenerife, Spain), see Table 1 .

Each species was gently washed with deionized water, dried at room temperature, and ground into a powder. Seaweed extracts were prepared by adding the dried macroalgal biomass to 150 mL of distilled water (600 mL flask) and sonicating (J.P. Selecta Ultrasounds 3000513, Abrera, Spain) for 30 min at room temperature, after which the mixture was centrifuged to separate the extract from the biomass. This process was repeated three times, then the three extractions were pooled and after removal of solvent by a rotary evaporator (Büchi R-200, Flawil, Switzerland), the extracts of each species were obtained. Table 1 shows the species, the quantity of dry algae and the yield of each extract.

Each extract was analyzed by NMR using D₂O as solvent. We measured ¹H NMR and ¹³C NMR, HSQC (Heteronuclear Single Quantum Coherence), HMBC (Heteronuclear Multiple Bond Correlation) and COSY (Correlated Spectroscopy) spectra using a Bruker Avance II-500 instrument (Bruker, Karlsruhe, Germany) operating at 500 MHz for ¹H NMR and at 125 MHz for ¹³C NMR. The use of standard Bruker software (TOpSpin 2.1, Bruker, Karlsruhe, Germany) aided the provision of two-dimensional NMR spectra.

Tomato (Solanum lycopersicum) var. Robin seedlings were obtained from a local nursery. We sowed tomato seeds in standard 150-cell tomato seedling trays using universal substrate and an automatic sowing machine to ensure germination and growth uniformity. When the seedlings reached the two true leaf stage (two weeks), we started the experiment. The seedling trays were then transferred to a growth chamber with controlled conditions: temperature 24 ± 2°C, photoperiod 16–8 h, humidity 65 ± 2%, and irradiance 300 µmols/m²s. All plants received a half-strength Hoagland solution as water supply (Hoagland and Arnon, 1950). Kelpak™ (BASF, Germany) and Sealgae™ (Biovert, Spain) were purchased from a local supplier.

All water deficit growth tests were carried out over 7 days according to the method described by Jiménez-Arias et al. (Jiménez-Arias et al., 2022). Water stress was induced by watering at 50% field capacity with a half-strength Hoagland solution, compared to control plants well-irrigated at 100% field capacity. All treatments consisted of twenty plants (N=20). Results are the mean of three independent experiments (see schema of the experimental set up in Figure 1 ).

Plant treatment was conducted by dissolving the various algal extracts at 1% w/v in 100 ml distilled water and adding 5 ml directly to the root system, except for the control treatment which received 5 ml of a half-strength Hoagland solution.

Seedling biomass was calculated at the beginning and end of the water deficit period (7 days). We removed the seedlings from the cell trays, carefully washed their roots under water to remove peat, then oven-dried them at 70°C for three days. Relative growth rate (RGR) was estimated according to the formula: RGR = (lnW2 - lnW1)/(t2 - t1), where W1 and W2 are the dry weights of the seedling at times t1 and t2 (the beginning and end of the water deficit, respectively) (Jiménez-Arias et al., 2015). Various indexes were calculated using the weight of the plants at 7 days, such as the stress susceptibility index (SSI) (Ganança et al., 2015), stress tolerance index (TSI) (Farshadfar et al., 2013), relative growth rate (RGR), and plant water use efficiency (WUEp) (Jiménez-Arias et al., 2022).

The fully developed leaves (N = 30) were subjected to gas exchange analyses. Photosynthesis (Pn), intracellular CO₂ (Ci), stomatal conductance (gs) and transpiration rate (E) were measured on the attached leaves using a portable infrared gas analyzer (LCPro, BioScientific Ltd., Hoddesdon, UK). Measurements were at ambient CO₂ concentration, photosynthetic photon flux density (PPFD) of 1000 µmol m^-2 s^-1 (optimized with a light curve) and cuvette airflow of 500 ml min^-1. The values for instantaneous water use efficiency (iWUE) and intrinsic water use efficiency (intWUE) are the ratios between Pn/E and Pn/gs, respectively (Seibt et al., 2008). The ratio between Pn and Ci was also calculated.

We grew pMAPKKK18-LUC Arabidopsis thaliana seedlings (Vaidya et al., 2017) in 24-well plates (25-35 seeds per well) filled with 1 ml of MS medium supplemented with 1% agar for 7 days (d). Subsequently, Arabidopsis seedlings were treated with mock, 25 μM ABA or 0.5% (w/v) G. rugosa extract for 24 h in a solution containing 100 μM luciferin (potassium salt, GoldBio). Seedlings were incubated for 24 h and luminescence was recorded with a LAS-3000 imager (Fujifilm) equipped with a CCD camera using 2 min exposures. We converted eight-bit images to rainbow false color and quantified using Fiji. The experiment was repeated at least twice.

Ten-day-old tomato seedlings (cv. Moneymaker) were treated with mock or 0.5% G. rugosa extract for 6. Total RNA was extracted using a NucleoSpin RNA plant kit. Synthesis of cDNA and quantitative real-time PCR (qRT-PCR) analyses were performed as described by González-Guzmán et al. (2014). cDNAs corresponding to the ABA-responsive genes Sl02g084850 (SlRAB18) and Sl06g067980 (SlLEA) genes were amplified using the same primers they used (González-Guzmán et al., 2014). Expression was normalized using values obtained with Sl06g009970 (SlEF1a).

The G. rugosa extract was dissolved in 80% methanol-1% acetic acid containing internal standards and mixed by shaking for one hour at 4°C. The extract was stored overnight at -20°C, centrifuged and the supernatant dried in a vacuum evaporator. The dried residue was dissolved in 1% acetic acid and passed through an Oasis HLB (reverse phase) column as described in (Seo et al., 2011). To quantify the hormone ABA, the dried eluate was dissolved in 5% acetonitrile-1% acetic acid and the hormone separated using an autosampler and reverse phase UHPLC chromatography (2.6 µm Accucore RP-MS column, 100 mm length x 2.1 mm i.d.; ThermoFisher Scientific) with a 5 to 50% acetonitrile gradient containing 0.05% acetic acid, at 400 µL/min for 21 min. The hormone was analyzed by selected ion monitoring (SIM) with a Q-Exactive mass spectrometer (Orbitrap detector; ThermoFisher Scientific), and its concentration in the extract determined using embedded calibration curves and the Xcalibur 4.0 and TraceFinder 4.1 SP1 programs. The internal standard for quantification was the deuterium-labeled hormone.

After the data normality and homoscedasticity were checked a one-way ANOVA TEST (Duncan´s post hoc, IBM-SPSS24 statistical software package) was used to analyze the differences at p values<0.05 between treatments in all the measures studied.

The specimens included two species from the phylum Rhodophyta (red algae), namely Bonnemaisonia hamifera Hariot and Galaxaura rugosa (J. Ellis & Solander); two species from the phylum Chlorophyta (green algae), namely Dasycladus vermicularis (Scopoli) Krasser and Ulva clathrata (Roth) C.Agardh; and five species from the phylum Ochrophyta (brown algae), namely Cystoseira foeniculacea (Linnaeus) Greville, Cystoseira humilis Schousboe ex Kützing, Lobophora dagamae C.W.Vieira, Colpomenia sinuosa (Mertens ex Roth) Derbès & Solier, and Halopteris scoparia (Linnaeus) Sauvageau (Guiry and Guiry, 2023). Therefore, our study included representative species from all three main macroalgal groups.

The comprehensive metabolic profile analysis of each crude extract was carried out by ¹HNMR and ¹³C NMR, COSY, HSQC and HMBC experiments ( Supplementary Material ). The main metabolites detected in each extract are listed in Table 2 , together with their ¹H and C-13 chemical shifts. The presence of each metabolite was confirmed by comparing its spectroscopic data with those previously described in the literature. NMR spectra analysis of the extract of B. hamifera in D₂O indicated that this extract contains a complex mixture of metabolites, among which floridoside (Ye et al., 2013) appeared as the major metabolite present. The HSQC correlations of signals at δ_C 103.0; 98.5; 96.4; 95.9 and 92.1 ppm, with protons at δ_H 4.44 (d, 7.6); 4.90 (brs); 4.60 (d, 7.4); 4.66 (d, 7.6) and 5.25 (d, 3.8) ppm respectively, indicate the presence of other saccharides in minor amounts. Other metabolites detected were: isethionic acid (Barrow et al., 1993), acetic acid (Sciubba et al., 2020), formic and lactic acid (Ponder and Richards, 1993) and the amino acids alanine (Sciubba et al., 2020), valine (Ryu et al., 2016), threonine (Salvador et al., 2022) and N,N-dimethyltaurine (Blunden et al., 1986).

The 1D and 2D NMR spectra analysis of G. rugosa extract in D₂O showed that this extract also contained a complex mixture of metabolites, including N,N,N-trimethyl taurine (King et al., 1982), N,N-dimethyltaurine, N-monomethyltaurine (McCusker and Klinman, 2009) and taurine (Ye et al., 2013), as well as isethionic acid and floridoside. The NMR data of N-monomethyltaurine and taurine are consistent with those of standard commercial products. Formic acid was also detected.

NMR data for D. vermicularis extract show the presence of α-D-fructofuranose 1,2’:2,3’β-D- dianhydride (DFAIII) as the major compound. DFA III has been isolated as a by-product of inulin, obtained from acid hydrolysis, pyrolysis or an enzymatic reaction of inulin (Uchiyama, 1982; Blize et al., 1994), but has also been described as a natural product isolated from Lycoris radiata (Li et al., 1997). Although inulin has been found in the green alga Ulva lactuca (Aguilera Morales et al., 2018) and inulin-type oligosaccharides have been found in the green alga Acetabularia crenulata (Bourne et al., 1972), to the best of our knowledge this is the first time that this metabolite has been found in a marine organism. The ¹H NMR spectrum of D. vermicularis extract also shows low intensity signals at δ_H 7.96 (s) and 7.40 (s) ppm in the aromatic region, indicating the presence of sulfated coumarins such as dasycladins A and B and signals at δ_H 7.92 (d, 10.1) and 7.36 (d, 10.1) ppm and δ_H 7.32 (d, 8.9) and 7.27 (d, 8.9) ppm indicating the presence of 4-(sulfooxy)benzoic acid and 4-(sulfooxy)phenylacetic acid respectively (Hartmann et al., 2018). Another metabolite detected at low levels was isethionic acid.

The 1D and 2D NMR spectra analysis of U. clathrata extract in D₂O indicated that 3-dimethylsulfoniopropionate (Blunden et al., 1992) is the major compound present in this extract that dissolves in D₂O. Also, 2,3-dihydroxypropanesulfonate (Edmonds and Francesconi, 1983) can be detected in the crude extract of U. clathrata.

The extracts of C. foeniculacea, C. humilis and L. dagamae in D₂O are very similar, mannitol (Yeşil and Akgül, 2022) and citric acid (Pinto et al., 2021) were found, with mannitol being the major constituent of the two in all three extracts. Lactic acid was also detected in the C. foeniculacea extract. While the HSQC spectrum of C. humilis shows a correlation between the signal at δ_H 4.92 (d, 3.9) ppm and the signal at δ_C 98.50, indicating the presence of saccharides, and isethionic acid was again detected. The L. dagamae extract also shows signals at δ_C 102.7: 102.4 and 98.1 ppm, which correlate in the HSQC spectrum with protons at δ_H 4.92: 4.80 and 4.53 ppm respectively, indicating the presence of other saccharides in minor amounts.

The 1D and 2D NMR spectra analysis of C. sinuosa extract in D₂O indicated that 3-hydroxybutyrate (Zweifel et al., 2009) is the major compound present in this extract, which dissolves in D₂O. Another minor metabolite detected by NMR is formic acid.

By contrast, 1D and 2D NMR spectra analysis of H. scoparia extract in D₂O indicated that isethionic acid is the main compound in this extract dissolving in D₂O. Other metabolites detected by NMR are: 2,3-dihydroxypropanesulfonate and phloroglucinol monosulfate (Glombitza and Knöss, 1992).

Betaines and sulphonium compounds are common in polar extracts of marine algae (Blunden et al., 1986; Blunden et al., 1992). In addition to the substances mentioned above, three betaines were detected in five of the nine algae according to the HMBC correlations of the N-methyl groups. The HMBC correlations of the N-Me protons at δ_H 3.26 (s) with both the methyl signal at δ_C 53.6 ppm and the methylene signal at δ_C 64.6 ppm indicate the presence of glycine betaine (Blunden et al., 1992) in B. hamifera extract. The HMBC correlations of the N-Me protons at δ_H 3.21 (s) with the methyl carbons at δ_C 53.5 ppm and the methylene signal at δ_C 78.0 ppm point to another betaine, probably glycine betaine aldehyde (Blunden et al., 1992) in G. rugosa, U. clathrata and H. scoparia extracts. Finally, HMBC correlations of the N-Me protons at δ_H 3.29 (s) with methyl carbons at δ_C 54.3 ppm and with the methylene signal at δ_C 67.0 ppm suggest the presence of another unidentified betaine in B. hamifera, U. clathrata, L. dagamae and H. scoparia extracts.

In order to test the biostimulant activity of the nine algal extracts, tomato seedlings were subjected to water deficit conditions (WD, 50% field capacity) after treating their root systems with the different extracts (see Table 3 ). Seedling dry weight of WD treatment was reduced by 11.2% after 7 days ( Figure 2 ).

Seedlings exposed to WD also showed significant differences in water use efficiency (WUEp), compared to the well-watered (WW) plants, which increased by 66.3% ( Table 3 ). However, the seedling dry weight of B. hamifera, D. vermicularis, C. humilis, and G. rugosa extract treatments was significantly higher compared to WD, comparable to the commercial extract Sealgae™ used as a positive control ( Figure 2 ). Interestingly, treatment with B. hamifera, D. vermicularis, C. humilis, and G. rugosa extracts caused a significant increase in WUE compared to mock-treated WD plants ( Table 3 ). Furthermore, drought stress indexes, RGR and WUE were also significantly higher for the above treatments ( Table 3 ).

After the bioprospection results, G. rugosa extract performs best under WD conditions. For this reason, deeper analyses were carried out under WW or WD conditions ( Figure 3A ). Seedlings treated with the extract show a small but non-significant reduction in plant weight under well-watered conditions, while the protective effect is clearly visible under deficit conditions ( Figure 3A ). It is noteworthy that extracts collected in different seasons can replicate this protective behavior ( Figure 3B ). Growth and tolerance index clearly show the tolerance enhancement by root treatment with G. rugosa ( Table 4 ). Under a low watering regime, treated plants increased in RGR, WUEp, and STI by 31, 53 and 40%, respectively, compared to untreated seedlings, showing an 83% decrease in sensitivity as SSI indicated.

Mock-treated plants exposed to WD continuously decreased stomatal conductance, transpiration, and net photosynthesis ( Figures 4A–C ). Application of the G. rugosa extract led to the recovery of photosynthesis at 5 days after the onset of WD, which also correlated with higher gs and E compared to mock-treated plants subjected to WD ( Figures 4A–C ). Applying the algal extract to well-watered plants also reduced gs, E and Pn, reaching its minimum three days after stress exposure, indicating that the extract’s components likely induce stomatal closure ( Figures 4A–C ). We also calculated the iWUE and int WUE parameters, and the efficiency of CO₂ assimilation according to the Ci ( Figure 5 ). At 3 and 5 days, we observed that plants subjected to WD and treated with the G. rugosa extract showed higher instantaneous and intrinsic WUE than mock-treated plants ( Figures 5A, B ). The algal extract also improved the Pn/Ci ratio in A-WD compared to WD plants ( Figure 5C ).

Seaweed extracts are complex mixtures, so to understand their action mechanisms is challenging. ABA is an important phytohormone for coping with water deficit stress, therefore activation of ABA signaling leads to increased WUE in crops (Mega et al., 2019; Yang et al., 2019; Mao et al., 2022). Recently, an ABA receptor agonist capable of activating ABA signaling in tomato was reported to have biostimulant activity against water deficit stress (Jiménez-Arias et al., 2023). Thus, to further understand the action mechanism of the seaweed extracts reported here, we focused on a sample that showed strong biostimulant potential, i.e., G. rugosa, to establish whether it could activate ABA signaling in either Arabidopsis thaliana (arabidopsis) or tomato. We first measured the ABA concentration in the G. rugosa extract as described in the Materials and Methods section, finding it to be 0.45 ± 0.12 nM. This ABA concentration was used in subsequent mock-treated samples to maintain the background effect due to residual ABA in that extract.

To test the effect on ABA signaling, we used an arabidopsis transgenic line in which the ABA-responsive MAPKKK18 promoter was fused to the LUC reporter (Vaidya et al., 2017). We incubated arabidopsis seedlings treated with mock, ABA or seaweed extract for 24 h in a solution containing 100 μM luciferin ( Figure 6A ). As a result, we observed induction of the LUC reporter by treatment with a 0.5% G. rugosa extract ( Figure 6A ). To test the effect of the extract in wild-type tomato seedlings, we treated tomato seedlings with a 0.5% G. rugosa extract for 6 h and assessed the expression of two ABA-responsive genes, namely Sl02g084850 (SlRAB18) and Sl06g067980 (SlLEA). As a result, we found that G. rugosa extract induced ABA-responsive genes in both arabidopsis and tomato ( Figure 6B ), which may contribute to the protective effect seen in the water deficit experiment.