Drought conditioning of rhizosphere microbiome influences maize water use traits

Beneficial plant–microbe interactions can improve plant performance under drought; however, we know less about how drought-induced shifts in microbial communities affect plant traits. We cultivated Zea mays in fritted clay with soil microbiomes originating from contrasting environments (agriculture or forest) under two irrigation treatments (well-watered or water limited). Using this design, we investigated whether water conditioning was carried forward through the microbiome to affect a subsequent plant cohort that was subjected to either a well-watered or water limited treatment. Regardless of the microbiome-origin, plants inoculated with a microbiome from a water limited legacy had traits that allowed them to avoid stress but conserve water. They produced longer roots to explore soil, generated greater soil dissolved organic carbon, potentially stimulating the microbiome, and slower soil water content loss during drought. A well-watered legacy resulted in plants that delayed permanent stomatal closure and higher photosynthetic nitrogen use efficiency. In plants with a forest-originated microbiome, a well-watered legacy and water treatment also resulted in higher rates of photosynthesis and stomatal conductance. These results demonstrate that soil microbiomes can be developed to influence plant drought performance, impacting crop resilience, using short-term microbial conditioning.


Introduction
The frequency and duration of drought events is increasing across agricultural land area (Masson-Delmotte et al. 2021), driving growing concerns about global food supplies and bioenergy security.The effects of climate change are compounding and increasing the negative impacts of drought on worldwide crop yield (Lesk et al. 2016).These declines in crop yield not only threaten food security but cascade into severe economic costs (Boyer et al. 2013).Drought often leads to lower crop productivity by decreasing stomatal conductance, eventually limiting CO 2 uptake and photosynthesis (Farquhar and Sharkey 1982).Lower CO 2 uptake, or productivity, leads to declines in plant growth (Xu et al. 2009;Hussain et al. 2019), resulting in lowered crop yield (Ciais et al. 2005).Maintaining plant productivity and growth under limited water availability is key to maintaining crop yields as world population and food demand continues to rise (Ciais et al. 2005;Passioura 2007;Boyer et al. 2013).
There is growing interest in utilizing microbes to improve plant and ecosystem functions for specific functions, such as improved productivity (Backer et al. 2018;Cavicchioli et al. 2019;Compant et al. 2019;de Vries et al. 2020).Soil and phyto-microbiomes create close associations with their host plant and can play large roles in ecosystem functioning and plant productivity (van der Heijden et al. 2008), providing a promising solution to help alleviate plant stress.Microbial associations can increase a plant's ability to deal with drought by, for example, directly altering root growth (Fan et al. 2011;Marasco et al. 2013;Henkes et al. 2018) and water and nutrient scavenging through the extension of the root system (Orfanoudakis et al. 2010).These responses improve the plant's ability to maintain higher rates of stomatal conductance and continued CO 2 uptake during moderate drought (Ulrich et al. 2019).Microbes can also directly influence the rhizosphere soil water available to the plant by excreting exopolysaccharides that create a gel-like substance that increases soil adhesion to roots and maintains soil structure (Sandhya et al. 2009).Beyond drought tolerance, many studies have shown improved plant structure and form through microbial inoculations (Timmusk et al. 2014;Dhawi et al. 2015).Trees and crops inoculated with beneficial microbes can have higher growth (Timmusk et al. 2014;Rolli et al. 2015) and produce higher yields (Chen et al. 2021).Most of these applications focus on individual plant growth promoting bacteria or mycorrhizal fungi; however, interactions between microbes can alter their effects on plants (Lau and Lennon 2011), and developing complex microbial communities may provide a more robust method of improving plant traits (Busby et al. 2017;Compant et al. 2019).Inoculations of microbial communities can provide advantages over a single microbial strain as they may have a competitive advantage and longer persistence in natural systems.
Plants can alter the soil chemistry and microbial communities creating plant-soil feedbacks, where the feedbacks have legacies that affect the growth, function, and interactions of future plant communities (Bever 2003; Van der Putten et al. 2013).Legacies of plant community-induced shifts in microbiomes can influence future generations of microbial-plant interactions (Heinen et al. 2020), and abiotic factors, such as extreme weather, warming, and nitrogen enrichment can shift the microbial community in a way that alters future plantplant interactions (Meisner et al. 2013).These microbial legacies can be beneficial for plant drought avoidance and tolerance, and understanding these interactions could help identify robust microbiomes that can support plant growth under limited water availability.Some plant soil-feedback studies have investigated drought effects on plant growth and competition, where drought legacies on microbiomes can give some plant species a competitive advantage over others (Kaisermann et al. 2017).Further, bacteria isolated from a droughted environment can enhance plant drought tolerance traits (Marulanda et al. 2009), and natural microbial consortia sourced from water stressed conditions can allow maize to avoid drought stress by increasing root length compared to microbiomes from nonstressed plants (Zhang et al. 2022).Additionally, one study found that a microbiome intentionally developed under drought can improve the fitness of future plant generations experiencing drought conditions (Lau and Lennon 2012), suggesting a beneficial legacy.While Vol.: (0123456789) these studies provide evidence that drought-induced rapid shifts in microbial communities can influence plant competition and fitness, we are still lacking information on how these stress-induced shifts in microbial communities can affect plant growth and function.
In this study we investigated microbiome legacy effects on Zea mays (maize) growth and function using two different source microbiomes, one originated from a forest (M Forest ) and one from an agricultural field (M Ag ), under well-watered (W W ) and limited water (W L ) conditions.We used a plant-soil feedback approach to condition the microbiomes under an initial microbial selection generation (G 0 ) of maize growth with these two watering conditions.We then propagated these microbiomes into a second, experimental, generation (G 1 ) of plants and tested for a legacy effect under the same and altered watering treatments (Fig. 1).This design allowed us to assess how a short-term drought inflicted on microbiomes originating from diverse sources affects plant growth and function.We hypothesized that the watering treatment would have a strong effect on plant growth; however, we also expected a growth-promoting legacy from microbiomes that received a W L treatment during the microbial conditioning phase.The legacy would manifest so that microbiomes that received a W L conditioning treatment would promote plant growth and function (photosynthesis and stomatal conductance) compared to microbiomes that received a W W conditioning treatment.We further hypothesized that the legacy of the W L conditioning would be stronger in plants inoculated with the microbiome originating from agricultural soil, as this microbiome was adapted to maize under low irrigation conditions.
Fig. 1 Representation of experimental design.Generation 0 (G 0 ) describes the water conditioning phase where the microbiomes were either well-watered (W w ; large water drop) or water limited (W L ; small water drop).The parent microbiomes in the conditioning phase originated from either agricultural (M Ag ; yellow) or forest (M Forest ; green) soil.Soil was collected from each conditioning phase pot and the microbiome was transferred to a new pot and soil for Generation 1 (G 1 ).Half of the pots containing W W -conditioned microbiome were selected at random to receive W W watering treatment again at G 1 , while the remaining pots containing W W -conditioned microbiome were instead given W L watering treatment at G 1 .Likewise, half of the pots containing W L -conditioned microbiome received W W watering treatment and half received W L watering treatment at G 1 .This obtains all eight combinations of microbial conditioning treatment, watering treatment, and microbiome parent source (shown in the bottom row of the figures) specified by the experimental design.Image courtesy of LANL.Don Montoya

Experimental design
We investigated the effects of microbial conditioning, microbiome parent, and watering treatments on plant growth and function.We were interested in studying two levels of microbial conditioning treatment (W L and W W ); two levels of watering treatment (W L and W W ); and two microbiome parent sources (M Ag and M Forest ).The experiment was performed as a completely randomized 2 × 2 × 2 full factorial design, consisting of all possible combinations of these levels, giving eight combinations in total.The experiment was carried out in seven replicates per condition for a total of 56 plants.

Experimental set-up
The experiment was conducted in a greenhouse with air temperature averaging 21-22 °C during both day and nighttime hours.Fifty-six Zea mays plants (B73 line: USDA National Plant Germplasm System) were grown from seed under 14 h/10 h day/ night cycles.To equalize the daylight hours over the season, natural light was supplemented with growth lights (ROI-E720, Grower's Choice, Ontario, CA) set to 50% of maximum natural light intensity (PPFD 2000 µmol m −2 s −1 ) to produce PPFD 1000 µmol m −2 s −1 both in the morning and evening.All plants were grown in 9.8L pots in 6L 20-30 mesh size artificial, fritted clay (GreensGradeTM, PROFILE(R) Products LLC, Buffalo Grove, IL, USA) to control soil chemistry and sterility prior to inoculation.We did not autoclave the clay due to leaching issues; however, the fritted clay manufacturing process produced an extremely low microbial load.Fritted clay was rinsed multiple times prior to planting to reduce the potential for solute toxicity, especially manganese (Adams et al. 2014).
Following the experimental design, half of the 56 plants were selected at random to be inoculated with microbiome source M Ag and half were inoculated with source M Forest .The agricultural microbiome (M Ag ) was collected from a maize agricultural field from Akron, CO (40.1605°N, 103.2144°W) that has historically not been irrigated.The forest microbiome (M Forest ) was collected from a mid-elevation mixed conifer forest in Los Alamos, NM (35.8800°N, 106.3031°W).The two source microbiomes started with different community compositions that remained different throughout the water conditioning phase of the experiment (Moore et al., 2023).Plant inoculation consisted of inoculating both the fritted clay and the seed.The fritted clay was inoculated using a 1:20 soil-water slurry.Z. mays seeds were soaked in a 10% bleach solution for 10 min, rinsed three times with sterile water, and soaked in the natural soil-water slurry for 20 min.Two seeds were then planted per pot and 500 mL of the water slurry was used to water each pot.
Generation 0 (G 0 ) was used as the conditioning phase.Half of the plants inoculated with M Ag and half of the plants inoculated with the M Forest were chosen at random to be grown under well-watered (W W ) conditions at G 0 ; all remaining plants were grown under water limited (W L ) conditions at G 0 .Plants were maintained at 64% and 45% volumetric water content, respectively.To allow for sufficient germination, all plants were maintained at 64% volumetric water content for two weeks after planting.Moisture contents were selected based on fritted clay measured saturation and the moisture characteristic curve (Steinberg et al. 2005).The 45% moisture condition was selected because it was high enough to allow plant growth, but low enough to impose a water limited condition.Soil moisture in each pot was monitored using a soil moisture probe (TEROS 10, METER Group, Inc, WA, USA), and each pot was watered to the target volumetric water content three times per week using filtered tap water enhanced with FloraGro, FloraMicro, and FloraBloom (General Hydroponics, Santa Rosa, CA, USA) fertilizers with doses of 200, 200, and 100 μL L −1 of water, respectively.When both seeds germinated, the smaller seedling was pruned from the pot.When neither planted seed germinated, both seeds were carefully dug up with a sterilized spatula and removed, and the pot re-planted.Plants were grown until the emergence of the 11 th leaf.Because the G 0 watering treatment affected plant growth rate, W W plants were harvested at 8 weeks and W L plants at 10 weeks.10 cm soil cores were then collected to sample the soil and rhizosphere microbiome of each pot and stored to inoculate a new seed.Microbial communities remained distinct after Gen 0 and differences between Gen 0 and Gen 1 show that microbial communities continued to shift throughout the experimental Gen 1 (Fig. S1a).
After the completion of G 0 , we used serial propagation of the microbiome (i.e., the microbiome from each pot in G 0 was used to inoculate one pot in G 1 ) to transfer the conditioned microbiome to a microbial selection generation with a fresh seed from the same seed batch as used in G 0 (G 1 ).G 1 seeds were inoculated with G 0 microbiome using the same method described above.The G 0 soil cores were collected surrounding the intact plants.Therefore, soil cores included fine roots, which were also incorporated in the inoculum slurry.All seeds were planted in fresh, rinsed fritted clay so that the conditioned microbiome, and not differences in soil chemistry, provided the legacy factor.While, we do acknowledge that our soil slurry likely had nutrient carry-over from G 0 , this would be very minimal due to high dilution at planting.Seedlings were allowed to germinate and grow under W W treatment for two weeks, after which half were given the W W watering treatment and half were given the W L watering treatment (Fig. 1), as specified by the experimental design.After emergence of the 11 th leaf, plant height, stem diameter, and maximum photosynthesis rates were measured as described below.After measurement, the plants were subjected to a terminal drought phase where irrigation was discontinued.During this phase, stomatal conductance (g s ) and soil volumetric water content (VWC) were measured daily during the hours of 08:00-10:00 using a porometer (SC-1; Decagon Devices Inc., WA, USA).Once the newest fully-expanded leaf reached g s = 0, we proceeded with the post terminal drought measurements described below.

Plant and soil traits pre-terminal drought
We measured plant growth response characteristics non-destructively starting one day before initiating the terminal drought.Plant height (cm) was measured from the base of the stem to the tip of the longest leaf using a ruler.Stem diameter (mm) was measured in two orthogonal directions approximately 2 cm from the base of the plant using calipers.Stem diameter was calculated as the averaged of two orthogonal measurements per plant.
We evaluated plant function from light saturated photosynthesis (A sat ) and g s measured on the second fully expanded leaf using an open-system infrared gas exchange system (LI6400XT, LI-COR Inc, Lincoln, NE, USA).Light was controlled at saturating light (measured from light response curves; data not shown) set to 2,000 μmol m −2 s −1 .Reference CO 2 concentrations and flow rate were controlled at 400 ppm and 400 μmol m −2 s −1 , respectively.Relative humidity inside the chamber was controlled at 20%, similar to the ambient humidity inside the greenhouse.Intrinsic water use efficiency (WUE i ) was calculated as A sat /g s .
Leaf traits, such as leaf carbon and nitrogen content, are often used as proxies to describe plant function and were measured on the same leaf as plant gas exchange, from 2-3 cm excised from the apex of the leaf.Leaf area (cm 2 ) was measured on the apex cutting of the leaf using a scanner (Canon Image Runner C5560i) and analyzed using ImageJ v1.53 image analysis software.The leaf was placed in a 60 °C drying oven for 48 h for dry mass (g) and then ground using a ball mill (12621-1664, VWR International LLC, PA, USA) to analyze for % nitrogen (N) and % carbon using an elemental analyzer (Eurovector, Pavia, Italy).Nitrogen per unit area (N area ; g m −2 ) was calculated as nitrogen per unit mass (N mass ; mg g −1 ) multiplied by leaf mass/ leaf area.Photosynthetic nitrogen use efficiency (PNUE) was calculated as A sat divided by N area .
Plants can directly encourage microbial growth during drought by altering their root exudation chemistry (Karst et al. 2017;Preece et al. 2018;de Vries et al. 2019); therefore, we also investigated soil pore water chemistry.We analyzed the concentration of dissolved organic carbon (DOC; mg L −1 ) and nitrate (NO 3 − ; mg L −1 ) in the pore water, which was collected using Rhizon filter pore water collectors (Rhizosphere Research Products, Wageningen, Netherlands).We placed two collectors per pot, one at the surface (1-11 cm) and one in the deeper profile (6-17 cm).To get a representative sample covering the entire pot depth, surface and deep profile samples were combined to collect at least 20 mL pore water.Samples were stored at -20 °C until thawed and analyzed for DOC using OI Analytical Aurora 1030 TOC analyzer.NO 3 − and NH 4 + were analyzed using a Dionex ICS-2100 (Thermo Fisher Scientific Inc., USA) and Varian Cary 100Bio UV-VIS Spectrometer, respectively.10 cm soil cores and biocrust samples were then collected for microbiome analysis (Supplemental Methods S1).

Post terminal drought traits
During the terminal drought, VWC declined.The rate of decline was obtained by fitting a linear decay function: where t indicates time passed since beginning the terminal drought in days, a is the starting VWC before beginning the terminal drought and b is the decay factor.Parameters a and b were estimated for each pot separately using non-linear least squares to determine the best fit.The W W water treatment plants had higher starting VWC; therefore, we normalized each VWC and terminal drought day count measurement to a proportion of the highest VWC and total drought days for each plant.VWC at permanent stomatal closure was measured as the VWC on the morning when the plant reached zero g s .
After the first fully expanded leaf reached g s = 0, plants were harvested and measured for root biomass.To harvest roots, stems were cut 2 cm from the base of the plant.Root systems were carefully pulled from the pot and gently shaken to release soil.Root bulbs were washed in 500 mL sterilized filtered water.Remaining soil in the pot was hand searched for remaining fine roots.Roots were immediately frozen in -20 °C for root morphology analysis (total root length (cm); root length of < 1 mm diameter, 1-2 mm diameter, and > 2 mm diameter (cm), and root projected area (cm 2 )).For root morphology analysis, roots were placed in a refrigerator to thaw for no more than 72 h.Root systems were imaged (4800 Perfection, Epson America Inc., Epson, CA, USA.) at 600 dpi and analyzed using WinRhizo pro software (Regent Instruments Inc., QC, Canada).Samples were dried at 60 °C for at least 48 h.Specific root length (SRL, m g −1 ) of the entire root system was calculated as root length per dry root mass.

Statistical analyses
We employed the analysis of variance (ANOVA) to investigate the simultaneous effects of microbiome source, conditioning treatment, and watering treatment on each above-and belowground trait of interest, using a significance threshold of = 0.05 .We (1) VWC = a(1 − b) t , carried out separate statistical analyses for each trait.All analyses were conducted using R statistical software (R Core Team 2020).When no statistically significant interaction effects were identified, we estimated main effects.For statistically significant main effects, we estimated and reported the difference in factor level means (W W -W L (G 0 ); W W -W L (G 1 ); M Forest -M Ag ) and constructed a 95% confidence interval for the true difference.When statistically significant interaction effects were identified, we carried out multiple comparisons of all treatment means using the Tukey procedure to examine factor effects jointly, as it is not appropriate to combine factor levels to investigate main effects when interaction effects are present.When assumptions of a linear model were not met, we performed a Box-Cox transformation of the response Y, given by where is a parameter that is determined from the data (Box and Cox 1964;Sakia 1992).
Differences in microbial community composition were statistically compared with a PERMANOVA analysis using the adonis function in the R package vegan (Oksanen et al. 2022; see Supplemental Methods S1).

Conditioning microbial legacy effect
A single generation of water-limited microbial conditioning enhanced root and stem growth in the subsequent generation regardless of the microbiome parent and water treatment.Microbiome conditioning did not affect plant height (Table 1; Fig. 2a), but plants grown under a microbiome with W L conditioning had thicker stems than plants with a W W conditioned microbiome (11.95 ± 0.40 mm vs 10.95 ± 0.45 mm, respectively, Fig. 2b).Interestingly, dry root mass was unaffected (Fig. 2c); however, there was a microbiome conditioning effect on root surface area and length of mid and coarse diameter roots.Plants showed marginal evidence for longer fine roots (p = 0.069), and had longer mid (440.93 ± 31.86 cm vs 330.25 ± 27.08 cm) and coarse (1341.88 ± 90.10 cm vs 1045.32 ± 101.12 cm) roots when grown with microbes having a W L than W W conditioning (Table 1; Fig. 2d-f microbes produced plants with greater root surface area than W W conditioned microbiomes (1407.88 ± 96.17 cm 2 vs 1132.45 ± 102.54 cm 2 ; Fig. 2g).Plants grown with W L conditioned microbes had marginally longer SRL (47.47 ± 2.62 m g −1 vs 40.74 ± 2.88 m g −1 ) than those with W W conditioned microbes (Table 1; Fig. 2h).
For A sat , the conditioning legacy effect depended on both the watering treatment and the parent microbiome, characterized by a statistically significant three-factor interaction involving conditioning treatment, watering treatment, and microbiome treatment (p = 0.006, Table 2).Tukey multiple comparisons showed that M Forest plants treated with W w at G 1 obtained statistically significantly higher mean A sat (17.71 ± 0.85 μmol m −2 s −1 ; Fig. 3a) when given the W w conditioning treatment than when conditioned with the W L treatment (12.06 ± 0.67 μmol m −2 s −1 ; p = 0.011).Additionally, M Forest plants treated with W w at G 1 also obtained statistically significantly higher mean A sat than W L M Ag plants treated with the W L conditioning treatment (11.36 ± 1.28 μmol m −2 s −1 ; mean ± se; p = 0.003).
There was a similar three-factor interaction for g s (p = 0.031).Under the W L conditioning, varying the levels of microbiome or watering treatment had no discernable effect on mean g s (Fig. 3b; left panel).Under the W W conditioning, increasing the watering treatment from W L to W W resulted in an increase in mean g s for M Forest but not for M Ag .Tukey multiple comparisons showed that M Forest plants treated with W W at G 1 obtained statistically significantly higher mean g s (0.17 ± 0.03 mol m −2 s −1 ; Fig. 3b) when given the W w conditioning treatment than plants given three of the four treatment combination involving W L conditioning: (1) W L conditioning with M Ag and W L watering treatment (0.09 ± 0.03 mol m −2 s −1 ; p = 0.001); (2) W L conditioning with M Forest and W W watering treatment (0.11 ± 0.03 mol m −2 s −1 ; p = 0.029); (3) W L conditioning with M Forest and W L watering treatment (0.11 ± 0.03 mol m −2 s −1 ; p = 0.025).Additionally, W w -conditioned M Forest plants treated with W W at G 1 obtained statistically significantly higher mean g s than W w -conditioned M Forest plants treated with W L at G 1 (0.11 ± 0.03 mol m −2 s −1 ; p = 0.025).
There were no significant pairwise differences among any other treatment combinations for A sat or g s responses.There was no conditioning effect on WUEi (Fig. 4a; data shown as differences between treatment means) or leaf chemistry (Table S1; Fig. S2), but W W conditioning increased PNUE (31.14 ± 14.21 μmol CO 2 g −1 N s −1 ) over W L conditioning (25.31 ± 10.06 μmol CO 2 g −1 N s −1 ; Table 2; Fig. 4b).
W W conditioned microbiomes also allowed plants to delay permanent stomatal closure longer during the terminal drought, even when the rate of VWC decline was elevated for W W conditioning.Plants grown with W L conditioned microbiomes reached Factor level mean differences are not shown for characteristics with hypothesis test results for main effects with p > 0.05 (Table 2) Vol.: (0123456789) permanent stomatal closure almost two days earlier than plants grown with W W conditioned microbiome (13.29 ± 0.66 days vs 15.07 ± 0.71 days, respectively; Table 2, Fig. 4c).This occurred despite an increase in the decay factor describing VWC decline over time in plants with W W conditioned microbiomes (0.60 ± 0.02% vs 0.54 ± 0.02%; Table 2; Fig. 5) relative to plants with W L conditioned microbiomes.VWC at stomatal closure did not respond to microbial water conditioning.Microbial W L conditioning increased pore DOC over W W conditioning (11.11 ± 0.145 vs 8.25 ± 0.98 mg L −1 ; Table 1; Fig. 4d).However, pore water NO 3 − and NH 4 + did not respond to microbial water conditioning (Table 2; Fig. 4e).Soil microbial community composition varied and showed distinct clustering by conditioning treatment (Fig. S1b; p = 0.001).

Watering treatment effect
The G 1 watering treatment affected some plant traits without interacting with the G 0 conditioning treatment, showing that our watering treatment imposed stress while still allowing plants to continue functioning.Aboveground growth reduced and root length increased under W L watering treatment, even when root biomass decreased.Plants grown under the W W treatment were taller (115.89 ± 2.32 vs 103.42 ± 2.28 cm; Table 1, Fig. 2a) with increased stem diameter (12.97 ± 0.35 vs 9.94 ± 0.29 mm; Fig. 2b) compared to those grown under W L .Plants from W W had greater root biomass than W L treatment plants (3.66 ± 0.19 vs 2.94 ± 0.12 g; Table 1; Fig. 2c).However, when separated into root diameter classes, there was only an overall watering treatment effect for mid diameter root length with W W treatment producing plants with longer mid root length than W L treatment (445.99 ± 33.85 vs 325.19 ± 23.62 cm; Fig. 2e).SRL of W W plants was shorter than for W L plants (40.11 ± 2.48 vs 48.11 ± 2.94 m g −1 ; Fig. 2h).
Plants grown under W L watering treatment experienced stomatal closure ~ 2.5 days more quickly (12.93 ± 0.54 days vs 15.43 ± 0.77 days; Fig. 4c) than plants grown under a W W treatment.Similarly, the VWC decay factor was higher in plants given the W W treatment (0.60 ± 0.01% vs 0.54 ± 0.02%VWC day −1 ) than in plants given W L .VWC at stomatal closure was lower for the W L watering treatment (21.62 ± 0.35%VWC) than W W (23.06 ± 0.34%VWC; Table 2).W L watering treatment pots also had higher DOC (14.98 ± 1.28 mg L −1 vs 5.38 ± 0.26 mg L −1 ; Fig. 4d) than W W treatment pots.Pore water NO 3 − was almost three times higher in W L compared to W W pots (33.21 ± 3.64 mg L −1 vs 11.36 ± 2.21 mg L −1 ; Fig. 4e).NH 4 + did not respond to water treatment (Table 1).Soil microbial community composition varied and showed distinct clustering by watering treatment (Fig. S1b; p = 0.001).
M Forest demonstrated a higher abundance of nitrogen-fixing bacteria during the conditioning phase (G 0 ).This trend continued and was more pronounced in G 1 , where M Forest showed a higher abundance (hits) of nitrogen-fixing bacteria than M Ag (Fig. S2).

Discussion
Our experiment showed that just a single generation of water treatment conditioning generated a microbial community that could carry forward its water treatment legacy to influence the structure and function of a new maize plant.Root and belowground traits were strongly influenced by a W L legacy with root the differences in mean response between plants treated with well-watered (W W ) and water-limited (W L ) conditioning (light blue); W W and W L watering treatment (dark blue); and microbiomes originating from forest (M Forest ) and agricultural (M Ag ) soil (brown).Factor level mean differences are not shown for characteristics with hypothesis test results for main effects with p > 0.05 (Table 1) Vol:.( 1234567890) surface area, specific root length, lengths of mid-to coarse-roots, and soil pore water dissolved organic carbon increasing in plants inoculated with a W L conditioned microbiome.A W W conditioned microbiome also positively affected some plant functional traits.Plants with a W W conditioning legacy kept stomata open longer before reaching their point of permanent stomatal closure, had higher photosynthetic nitrogen use efficiency, and had higher rates of photosynthesis and stomatal conductance under M Forest and W W treatment.These results provide important evidence of how quickly microbiomes can shift to affect the function of future plant growth.Our findings provide a promising experimental design approach toward the development of microbiomes to secure agriculture under future climate stressors.
Water conditioning legacy on plant structure Many studies have shown that microbiomes influence plant belowground growth, where plant growth promoting bacteria increases root dry or fresh biomass (Marulanda et al. 2009;Marasco et al. 2013;Schwartz et al. 2013) by, e.g., altering plant signal cascades to increase lateral root growth or promoting production of root hairs that increase plant surface area for water uptake (Creus et al. 2005;Molina-Favero et al. 2008).Our watering conditioning induced a legacy effect, such that depending on the trait, the W L conditioning either benefitted or disadvantaged plant structure in relation to drought independent from the original microbiome source.Plants with a microbiome originating from W L conditioning had greater root length due to shift in the morphology of their root system such that they produced longer roots per biomass investment (e.g., longer SRL) than plants growing with a microbiome that originated from W W conditioning.Greater root length (Fig. 2, S5; Table 1), theoretically assists plants growing under limited water availability in foraging for water (Comas et al. 2013;Marasco et al. 2013).Similarly, plants that originated from the W L conditioning had thicker stems, even though height did not increase.Plantassociated microbiomes, particularly plant growth promoting bacteria, can either increase (Bashan et al. 2012;Molina-Romero et al. 2017) or decrease stem diameter (Molina-Romero et al. 2017).Greater stem thickness could lead to higher rates of water transport or, overall, higher biomass.The longer root length resulting from W L conditioning could have increased plant water uptake; therefore, thicker stems, if associated with larger conduits and stem cross sections, could have developed to better facilitate water movement through the plant (Abd El-Rahim et al. 1998).Although we did not specifically measure hydraulic conductance in this study, greater water transport through the stem often leads to greater primary productivity (Brodribb and Feild 2000;Mencuccini 2003;Liu et al. 2020).However, during times of water scarcity, greater water transport can cause more rapid water uptake from the soil and eventually lead to more rapid water deficits (Leakey et al. 2006).In our experiment, W L conditioning still had a soil water retention advantage over W W conditioning in that W L -conditioned plants lost VWC more slowly during the terminal drought (Table 2; Fig. 5); therefore, this may not apply to our system.Overall, these results suggest that microbiomes quickly respond to drought in a way that affects plant growth traits toward water uptake and that shift is conserved in the next plant cohort.Water conditioning legacy on plant functioning While W L legacy resulted in beneficial plant structural shifts, whether W W or W L legacy was beneficial differed between plant functional traits.The W L conditioning resulted in greater DOC (Fig. 2d), showing the important control that the microbial legacy had on DOC.Pre-experimental growth substrate (fritted clay + fertilized water) pore water DOC was quite low (~ 5 mg l −1 ); however, it is likely that the fertilized water caused some accumulation of DOC over time in our pots.Even so, we saw a conditioning treatment effect on DOC regardless of the water treatment; therefore, the conditioning treatment effects are due to alterations of the microbiome during the conditioning phase.DOC treatment differences could be an indicator of either increased microbial metabolism of soil organic carbon, cells sloughing off during growth, or direct exudation roots (Calvo et al. 2019).Root exudates play a direct role in signaling and providing a carbon source for the microbial community that can help alleviate drought stress (Kiers et al. 2011).Although severe droughts may limit root exudation (Preece and Peñuelas 2016;Karst et al. 2017), greater carbon exudation or DOC is commonly found in systems exposed to moderate drought compared to a control treatment (Sanaullah et al. 2012;Preece et al. 2018).The amount of DOC has been shown to be controlled by microbial taxa (Sanaullah et al. 2012;Albright et al. 2020). 14C labeling studies have shown that drought-induced increases in DOC do not necessarily lead to increases in recently exuded C in microbial biomass (Sanaullah et al. 2012), suggesting DOC is directly coming from the plant instead of increased microbial biomass.The legacy of W L conditioning promoted higher DOC in the soil, suggesting that a drought adapted microbiome stimulates C exudation or root sloughing from the plant that, in principle, could promote microbiome activity that may also benefit the plant under drought stress.
A W W conditioning treatment benefitted M Forest , which produced higher rates of photosynthesis and stomatal conductance under well-watered conditions.However, due to higher rates of photosynthesis and stomatal conductance in the forest soil when microbiomes experienced W W during both the conditioning and water treatment phase, W L during the conditioning or water treatment phases seemed to prohibited gas exchange rate increases (Fig. 3).It is possible that the M Ag treatment did not experience an observable W W conditioning treatment legacy because this microbiome was already adapted to maize drought conditions.Less responsive stomata are more commonly found in more drought tolerant species (Skelton et al. 2015;Henry et al. 2019) and, while often associated with lower stomatal conductance, lower stomatal sensitivity and less efficient xylem can help plants maintain hydraulic safety (Gleason et al. 2017).Further, we hypothesize that plants grown under M Forest combined with W W water treatment and W W conditioning may have experienced upregulation of plant gas exchange compared to the W L treatment because M Forest was enriched with potential nitrogen fixing bacteria (Figs.S3, S4).N fixation could positively influence photosynthetic capacity and CO 2 uptake via impacts of increased N on photochemistry (Henneron et al. 2020).We note that further experiments are needed to definitely attribute function to these microbial taxa.These are taxa that could explain treatment differences, and nitrogen fixation is one potential mechanism, but we acknowledge that these taxa and other taxa in our systems could be driving different functions and, thus, the different plant responses observed between our treatments.Even so, rhizospheric nitrogen fixing bacteria play a key role in plant nutrient availability as they are able to directly reduce atmospheric nitrogen, creating higher nitrogen availability to plants while also producing phytohormones and solubilizing phosphorus (Swarnalakshmi et al. 2020).This could lead to higher nitrogen availability for the photosynthetic apparatus (Cleveland et al. 1999;Reed et al. 2011) and, therefore, increased rates of photosynthesis.
Photosynthetic nitrogen use efficiency (PNUE) also experienced a watering treatment legacy effect; however, this was not dependent on parent microbiome (Fig. 4b).W W conditioning led to plants with higher PNUE than plants with W L legacy, independent of the parent microbiome.This is aligned with the Least-cost Hypothesis that states that plants optimize their traits to efficiently use and acquire the more limiting resource (either water or nitrogen) (Field et al. 1983;Wright et al. 2003).In our study, W L microbiome conditioning tended to produce lower rates of photosynthesis, (significant interaction effect, Table 2); however, N area did not differ between plants with different microbiomes (Table 2).This suggests that the plants experiencing the W W treatments were Vol:.( 1234567890) better able to utilize similar concentrations of leaf nitrogen.Water and nitrogen are the factors that most limit global crop growth (Mueller et al. 2012), and importantly, these results show that the microbiome plays a role in this relationship, where W W conditioning of the microbiome allowed the plant to maximize plant nitrogen utilization per the rate of photosynthesis.Time to stomatal closure strongly controlled by microbial community Contrary to our hypothesis, corn plants that were inoculated with a W W conditioned microbiome were able to last longer during the terminal drought before reaching the point of permanent stomatal closure (Fig. 4c).This suggests the microbiomes retained a feature from W W conditioning that allowed them to prolong plant activity even under drought.The microbial community can have direct and indirect effects on how long plants keep stomata open during drought and have specifically been shown to delay wilting in rice (Redman et al. 2011).Microbes can exude exopolysaccharides that directly retain water and nutrients within the rhizosphere (De Caire et al. 1997;Alami et al. 2000;Mehboob et al. 2009).In addition, microbes can control plant hormonal signals and can reduce abscisic acid induced stomatal closure (Perrig et al. 2007;Kang et al. 2014).In our system, we also found that pots with W W conditioning lost soil water more quickly than pots with a W L legacy, yet these plants also took longer to close their stomata (Fig. 5).This extended time to stomatal closure under faster soil drying might be accomplished by altering plant hormonal signals or by allowing the plants to better access the more limited water available in these pots.

Implications and conclusions
We found that just a single generation of plant growth under different watering conditions induced rapid shifts in microbial function to influence a subsequent cohort of plant structure and function under varying watering treatments.These results present novel microbiome selection methods to improve plant performance relative to other studies that have used mycorrhizal fungi (Bowles et al. 2016), plant growth promoting bacteria (Rolli et al. 2015), and arid region microbiomes (Marulanda et al. 2009) in that they show that, regardless of the microbiome origin, we can very quickly modify the entire microbiome to produce plant traits of interest using droughting techniques.Similarly, we know from previous studies that drought alters the legacy of microbial communities (de Vries et al. 2019); however, our work further shows that shifting the microbial community alone can influence plant structure and function.
We hypothesized a priori that, of our two microbiomes, the microbiome originating from nonirrigated agricultural soil would show a stronger microbial legacy response to drought than the forest originated microbiome.Surprisingly, our results do not support this hypothesis.Except for A sat and g s grown under M Forest and W W treatment (Fig. 3), there were no clear differences between the conditioning legacies of the two microbiomes even though had different starting compositions (Tables 1 and 2; Moore et al. 2023).What was clear, however, was that the M Forest promoted greater plant growth than M Ag 5), a trend that occurred regardless of watering treatment and further supports results discussed in Moore et al. (2023).The shorter plants and lower root mass seen in M Ag plants might suggests that, similar to our W L legacy effects, plants exposed to M Ag would use less water and might be able to withstand drought for longer periods of time independently of the microbiome conditioning treatment.However, the decrease in aboveground growth was not accompanied by increased root length or area that would improve the plant's ability to access sources of water (Fig. 2; Comas et al. 2013).There is a trade-off associated with drought tolerance and plant size where a smaller plant might have a greater ability to survive but the smaller, drought adaptive traits often lead to lower productivity.Our results suggest that M Ag , when compared to M Forest , might actually be limited in its ability to shift in response to drought in a way that is beneficial for the crops growing under water limitation.Further, depressed growth in M Ag plants could have been influenced by the yield drag phenomenon, where years of agricultural plant growth in the same field decreases yield through an altered microbiome (Johnson et al. 1992;Benitez et al. 2021).While the difference in growth suggests important functional differences between plants grown in the presence of different microbiomes, it is important to note that height alone might not indicate improved plant yield, especially under drought where light interception is key (Comas et al. 2019).The impacts of microbiomes from a historically droughted environment on plant drought tolerance traits are currently under investigated.However, the differences we found between the two original parent microbiomes are opposite to a recent study by Zhang et al. (2022) who showed that a microbiome sourced from a droughted grass conferred higher maize drought tolerance than a microbiome sourced from irrigated agricultural conditions.Our results, however, agreed with Zhang et al. (2022) in that the microbiome able to best improve plant traits came from a natural soil rather than controlled, agricultural conditions.These results have important implications for agricultural applications trying to improve plant growth or yield through modification of the microbiome, where it is important to perform rigorous investigations of the community being amended.Further, sourcing that a microbiome from agricultural soil that has already gone through generations of selection under plant species and nutrient amendments might not produce the most effective microbial community to improve plant performance.
Our study shows that it is possible to rapidly induce a shift in complex microbial communities to alter plant performance.Examples showing that microbial communities can be pressured to modify composition in a way that alters plant function, rather than plant growth alone, are rare; however, our results are aligned with a study where a microbiome from a three-generation drought experiment improved aspects of plant fitness, such as fruiting and flowering time (Lau and Lennon 2012).In addition, Panke- Buisse et al., (2015) also adjusted plant flowering time through microbiome directed evolution.These results suggest that, while directly trying to evolve plants in a timely manner in response to global change may be an unlikely or impractical solution for improving plant structure and function in future environments (Lau et al. 2007;Lau and Terhorst 2020), achieving improved plant performance by imposing short-term treatments on microbial communities may be a promising avenue to ameliorate plant drought stress.

Fig. 2
Fig. 2 Plant growth trait response differences between the two factor level means for each of the three experimental factors: microbiome conditioning (G 0 ), watering treatment (G 1 ), and microbiome parent.Effects on plant height (a), stem diameter (b), dry root mass (c), fine root length (d), mid-diameter root length (e), coarse root length (f), root surface area (g), and specific root length (h).Bars in figures show the differences in

Fig. 3
Fig. 3 Three-factor interaction plots for response of interest (a) saturated photosynthesis (A sat ) and (b) stomatal conductance (g s ) (b).The left panels contain watering treatment-bymicrobiome two-factor interaction plots when the third factor, conditioning, is W L .The right panels contain watering treatment-by-microbiome two-factor interaction plots when the

Fig. 4
Fig. 4 Plant function response differences between the twofactor level means for each of the three experimental factors: microbiome conditioning (G 0 ), watering treatment (G 1 ), and microbiome parent.Effects on intrinsic water use efficiency (WUE i ; a), photosynthetic nitrogen use efficiency (PNUE; b), time to stomatal closure (c), pore water dissolved organic carbon (DOC; d), and nitrate (NO 3 − ; e).Bars in figures show

Fig. 5
Fig. 5 Volumetric water content (VWC) decline during the terminal drought.Normalized to highest content per pot, the rate that VWC declined until the plant reached the wilting point where stomatal conductance = 0 for plants that received the well-water conditioning (W W ; black) or water limited conditioning (W L ; gray) microbiomes.Symbols represent the current watering treatment: W L (circle) or W W (triangle)

Table 1
Plant morphology ANOVA p-value results for the modeling including microbiome origin (Mbiome), watering treatment (Tmt), microbiome water conditioning (Conditioning), and all possible interactions between these factors In all cases, numerator and denominator degrees of freedom were 1,48 respectively Bold values indicate significant response of α Variables that were transformed are noted by superscripts indicating the degree of Box-Cox transformation used, based on lambda (l): Transformed Variable = (x^l-1)/l *Vol:.(1234567890)

Table 2
Plant function ANOVA p-value results for the modeling including microbiome origin (Mbiome), watering treatment (Tmt), microbiome water conditioning (Conditioning), and all possible interactions between these factorsNumerator and denominator degrees of freedom for light saturated photosynthesis (A