A dynamic rhizosphere interplay between tree roots and soil bacteria under drought stress

  1. Yaara Oppenheimer-Shaanan
  2. Gilad Jakoby
  3. Maya L Starr
  4. Romiel Karliner
  5. Gal Eilon
  6. Maxim Itkin
  7. Sergey Malitsky
  8. Tamir Klein  Is a corresponding author
  1. Department of Plant and Environmental Sciences, Weizmann Institute of Science, Israel

Abstract

Root exudates are thought to play an important role in plant-microbial interactions. In return for nutrition, soil bacteria can increase the bioavailability of soil nutrients. However, root exudates typically decrease in situations such as drought, calling into question the efficacy of solvation and bacteria-dependent mineral uptake in such stress. Here, we tested the hypothesis of exudate-driven microbial priming on Cupressus saplings grown in forest soil in custom-made rhizotron boxes. A 1-month imposed drought and concomitant inoculations with a mix of Bacillus subtilis and Pseudomonas stutzeri, bacteria species isolated from the forest soil, were applied using factorial design. Direct bacteria counts and visualization by confocal microscopy showed that both bacteria associated with Cupressus roots. Interestingly, root exudation rates increased 2.3-fold with bacteria under drought, as well as irrigation. Forty-four metabolites in exudates were significantly different in concentration between irrigated and drought trees, including phenolic acid compounds and quinate. When adding these metabolites as carbon and nitrogen sources to bacterial cultures of both bacterial species, eight of nine metabolites stimulated bacterial growth. Importantly, soil phosphorous bioavailability was maintained only in inoculated trees, mitigating drought-induced decrease in leaf phosphorus and iron. Our observations of increased root exudation rate when drought and inoculation regimes were combined support the idea of root recruitment of beneficial bacteria, especially under water stress.

Editor's evaluation

This paper will be of interest to those interested in plant-microbe interactions under drought. Trees can exchange root exudates for minerals with soil bacteria. In a pot experiment under realistic conditions, the authors indicate that this exchange persists, and may be protective, when plants experience drought stress. The combination of methods used is an important strength: visualizing bacterial colonization of roots, measuring root exudation and mineral uptake, in conjunction with separate assays of bacterial growth responses to chemicals released by trees.

https://doi.org/10.7554/eLife.79679.sa0

eLife digest

The soil surrounding the roots of trees, termed the rhizosphere, is full of bacteria and other communities of microorganisms. Trees secrete organic compounds in to the soil which are thought to influence the behavior of bacteria in the rhizosphere. Specifically, these root secretions, or ‘exudates’, attract and feed soil bacteria, which, in return, release nutrients that benefit the tree.

In 2020, a group of researchers found that some trees in the Mediterranean forest produce more exudates during the long dry season. This suggests that the compounds secreted by roots may help trees to tolerate stress conditions, such as drought. To test this hypothesis, Oppenheimer-Shaanan et al. – including some of the researchers involved in the 2020 study – grew young Cupressus sempervirens conifer trees in drought conditions that starved them of the nutrients phosphorous and iron.

Each tree was planted in a custom-built box which allowed easy access to roots growing in the soil. Two species of bacteria from the forest soil C. sempervirens trees naturally live in were then added to the soil in each box. Microscopy revealed that both species of bacteria, which had been tagged with fluorescent markers, were attracted to the roots of the trees, boosting the bacterial community in the rhizosphere.

Oppenheimer-Shaanan et al. found that the recruitment of the two bacterial species caused the rate at which exudates were secreted from the roots to increase. Compounds in the exudate stimulated the bacteria to grow. Ultimately, levels of phosphorous and iron in the leaves of the starved trees increased when in the presence of these soil bacteria. This suggests that bacteria in the rhizosphere helps trees to survive when they are under stress and have low levels of water.

These findings provide further evidence that plants and bacteria can live together in symbiosis and benefit one another. This could have important implications for forest ecology and potentially how trees are grown in orchards and gardens. For example, specific bacteria and organic compounds in the rhizosphere may be able to improve tree health. However, further work is needed to investigate whether the exudate compounds identified in this study are found more widely in nature.

Introduction

Climate change is characterized by increased temperatures and altered precipitation patterns (Cavin et al., 2013; IPCC, 2014). These environmental changes affect terrestrial ecosystems worldwide, with negative impacts for forest health associated with water limitation (Zhao and Running, 2010; Allen et al., 2010; Choat et al., 2012). Drought adversely affects forest health in many aspects, including seedling recruitment (Pozner et al., 2022), tree productivity (Klein et al., 2014), and mortality of trees (Klein et al., 2019), with increased susceptibility to pathogen or insect attack (Reichstein et al., 2013; McDowell et al., 2022). Moreover, both drought frequency and intensity are projected to increase; particularly in the Mediterranean region and southern Africa (IPCC, 2019). The capacity of trees to survive future droughts is virtually unknown (McDowell et al., 2022).

Trees have evolved mechanisms to cope with drought, including escape, avoidance, and tolerance strategies. Drought has been shown to induce an alteration of carbon allocation from aboveground to belowground organs and increase in the amounts of soluble sugars in the roots (Huang and Fu, 2000; Hasibeder et al., 2015). Root systems mediate water and nutrient uptake, provide physical stabilization, store nutrients and carbohydrates, and provide carbon and nutrients to the soil through the process of fine-root turnover (Brunner and Godbold, 2007; Haichar et al., 2008; Ryan, 2011; Harfouche et al., 2014; Jarzyniak and Jasiński, 2014; Klein et al., 2016). In addition to these roles, trees invest a substantial part of their photosynthesized carbon into root exudates that entice and presumably feed plant-beneficial and root-associated microbiota (Bais et al., 2006; Badri and Vivanco, 2009; Karst et al., 2017; Jakoby et al., 2020). In parallel, the rhizosphere microbes can promote plant growth through various mechanisms such as increasing the availability of nutrients, secreting phytohormones, suppressing pathogens, or having positive effects on the plant metabolism (Pérez-Montaño et al., 2014; Zhou et al., 2016). Microbial utilization and metabolism play a central role in modulating concentration gradients of a variety of compounds right outside root tips, thereby constituting a soil sink (Dakora and Phillips, 2002; Mommer et al., 2016; Martin et al., 2017; Tsunoda and van Dam, 2017). The majority of root exudates typically consists of primary metabolites (20%; sugars, amino acids, and organic acids) and 15% of nitrogen as well as secondary metabolites, complex polymers, such as flavonoids, glucosinolates, auxins, etc. (Vives-Peris et al., 2020). Those plant-derived metabolites were shown to shape microbial communities by allowing bacteria to metabolize them and then establish themselves in the rhizosphere (Venturi and Keel, 2016; Sasse et al., 2017). Although root exudation is ubiquitous among tree species, the amount and composition of root exudates vary. So far, little information is available on how drought influences tree root exudates, their chemical composition, and how root metabolism is connected with shifts in root-associated microbiome composition (Tückmantel et al., 2017; Xu et al., 2018). Recently, it was shown that oak trees (Quercus ilex) shift their exudates from primary to secondary metabolites under drought (Gargallo-Garriga et al., 2018).

The chemical composition of root exudates have a direct effect on the rhizosphere communities. These include plant growth-promoting rhizobacteria (PGPR) genera such as Bacillus, Pseudomonas, Enterobacter, Acinetobacter, Burkholderia, Arthrobacter, and Paenibacillus (Sasse et al., 2017; Zhang et al., 2017). For example, the banana root exudate fumaric acid attracts the Gram-positive Bacillus subtilis N11 and stimulates biofilm formation (Zhang et al., 2014); malic acid exuded by Arabidopsis stimulates binding to roots and biofilm formation on roots by the B. subtilis strain FB17 (Rudrappa et al., 2008). Bacterial growth and antifungal activity of certain species of the Gram-negative Pseudomonas spp. is dependent on organic acids and sugars isolated from tomato root exudates (Kravchenko et al., 2003; Mhlongo et al., 2018). Among PGPR, species of Pseudomonas and Bacillus are the best studied as models for beneficial plant-microbe interaction (Weller et al., 2002; Raaijmakers et al., 2010). Interestingly, the importance of both genera to plants has been verified in multiple metagenomic studies (Loper et al., 2012; Mendes et al., 2013), however, the interplay between these important root bacteria and forest trees remains a mystery. In the forest soil, root exudation is suspected to enhance rhizobacteria, in turn leading to increased decomposition of soil organic matter, that is, increased C mineralization, a process termed microbial priming (Schleppi et al., 2019). Here, we use this term on a wide perspective, even that P release from soil phosphates, for example, is not strictly considered as priming (Dijkstra et al., 2013).

Recently, we have shown that roots of both conifer and broadleaf tree species in an evergreen Mediterranean forest increase their exudation flux during the long dry season (Jakoby et al., 2020). This increase in exudation occurred in spite of the sharp decrease in photosynthesis throughout the dry season, and more so in the coniferous Cupressus sempervirens. Hence, we speculated a specific role for root exudation in tree drought tolerance. Although the interactions between the rhizosphere microbiome and plants play a crucial role in plant growth, it is still unclear how these interactions affect trees under abiotic stresses. Our overall objective was to test whether microbial recruitment by tree root exudates (a form of microbial priming) is beneficial to trees under drought, an abiotic stress that alters tree carbon and nutrient allocation. Here, we link together rhizosphere processes and drought by studying the effect of soil drought on the physical and chemical interaction between the rhizosphere bacteria B. subtilis and P. stutzeri and roots of the conifer species C. sempervirens. We designed a custom-made tree growth facility permitting access to intact roots growing in forest soil, where we measured root exudation rate and composition in response to changes in irrigation and inoculations of these soil bacteria (Figure 1—figure supplement 1). Our major hypothesis regarded the existence of an exudate-induced microbial-tree interaction cycle, starting with tree stress and subsequent exudation, on to enhancement of soil bacteria and their activity, and back to improved tree nutrition.

Results

Soil drought limits tree gas exchange and growth

Soil moisture declined gradually in drought trees down to <10% (V/V) 3 weeks following irrigation cessation, and increased back after re-irrigation (Figure 1—figure supplement 2). In the beginning of the manipulation, which followed the end of the wet season (May 2019), we needed to adjust the irrigation amounts for the irrigated trees, which stabilized within 11 days. Assimilation and stomatal conductance of watered trees fluctuated around 8–12 µmol CO2 m–2 s–1 and 150–300 mol H2O m–2 s–1 throughout the experiment, respectively (Figure 2—figure supplement 1; Figure 2—source data 1). In drought trees, these high rates decreased to zero within 4 weeks, and, upon re-irrigation, increased back to baseline values within 4–5 weeks. Bacterial inoculations had no effect on leaf gas exchange, and hence the interaction irrigation:bacteria was not significant either. During the experiment, sapling biomass increment was 188.6±15.6 and 149.1±14.4 g for irrigated saplings with and without bacterial inoculations, respectively (difference not significant; p=0.753; Figure 2—source data 1). The effect of drought was highly significant (p<0.001), and biomass increment in drought-exposed saplings with and without bacterial inoculations was 22.4±5.7 and 4.5±4.6 g, respectively (p=0.09).

Trees recruit root-associated bacteria during drought

Fluorescently tagged B. subtilis and P. stutzeri, that were modified from native strains isolated from the forest soil, showed attachment and dispersed colonization along Cupressus fine roots, regardless of irrigation, on days 1 and 3 following inoculation (Figure 1—source data 1). Bacteria seemed to localize along crevices in the root epidermis, forming lines along roots, and sometimes co-localized (Figure 1A). The inoculations can be regarded as pulses, allowing us to test the rhizosphere interactions, while declining with time. Upon inoculation, P. stutzeri had higher abundance (~90%) than B. subtilis, but this ratio reversed within a few days or weeks (Figure 1B), with B. subtilis having an advantage over P. stutzeri in drought trees. Interestingly, when irrigation was restored, the drought trees recruited the bacteria in a similar fashion to irrigated trees, with P. stutzeri exhibiting 80% abundance in the rhizosphere at day 3, and then substituted by B. subtilis at days 7 and 15 (Figure 1C). Differences between days in relative abundance were significant (p<0.001; Figure 1—source data 1), except for the difference between days 3 and 15 in soil under drought. Bacterial abundance decreased over time, but, importantly, was always ~10-fold higher in the rhizosphere than in the bulk soil (Figure 1B), further evidencing recruitment by roots. Bacterial abundance was also lower in drought trees than around irrigated (Figure 1B) or re-irrigated trees (Figure 1C) (p<0.001; Figure 1—source data 1). Overall, only in trees under constant irrigation and after two inoculations, bacterial communities of B. subtilis and P. stutzeri seemed to stabilize in the rhizosphere (Figure 1C).

Figure 1 with 2 supplements see all
Tree root recruitment of beneficial bacteria during drought and re-irrigation.

Bacterial dynamics in soil and rhizosphere around roots of irrigated and drought-exposed Cupressus sempervirens saplings. Root colonization (A): epifluorescence and bright-field images of drought tree roots densely colonized by Bacillus subtilis-gfp (green) and Pseudomonas stutzeri-mCherry (red). Orthogonal views of a three-dimensional confocal image were created from a z-stack of x/y-scans on drought trees, 1 and 3 days following inoculation. Dynamics of relative abundance of B. subtilis (green) and P. stutzeri (red) in rhizosphere and soil during drought (B) and during re-irrigation (C). Presented are colony forming units of both species (CFU; expressed as log 10 per g root dry weight) (n=6).

Tree root exudates respond to drought and bacterial inoculation

Under constant irrigation and under drought, bacterial inoculation significantly increased root exudation, by two- to threefold (Figure 2; p<0.001). However, in trees that were exposed to drought and then re-irrigated, this pattern reversed, and inoculated trees had slightly lower exudates than without bacteria. Thus, drought-exposed trees, that were responsive to inoculation, showing an ~50% increase in root exudation, lost this response when re-irrigated (p=0.039 for the interaction irrigation:bacteria). The amount of organic carbon exuded from roots of drought-exposed trees was significantly, ~50% lower than from roots of irrigated trees, and moreover, ~70% lower, in the re-irrigation period (across inoculation treatments; p<0.01; Figure 2—source data 2). Importantly, the same trees that were exposed to drought reduced their exudation when re-irrigated (Figure 2). This observation is further supported by examining correlation coefficients between root exudation rates and the abundance of B. subtilis and P. stutzeri in the rhizosphere. They yielded values above 0.5 for drought trees on day 3 following inoculation (Figure 2C; p=0.046 for P. stutzeri). For B. subtilis, there was a high correlation also with irrigated trees’ exudation on day 7 following inoculation (p=0.093). Correlation coefficients decreased in the re-irrigation period, and were negative in trees that were constantly irrigated. An additional measure of tree carbon allocation into root exudation is the ratio between exudation rate (µg C mg root–1 day–1) and net assimilation rate (µmol C m–2 leaf s–1). This ratio was 0.40–0.45 under drought, decreasing to 0.27–0.35 under re-irrigation (across inoculation treatments). In irrigated saplings, ratios increased from 0.12 to 0.17 without bacteria, to 0.34 and 0.58 in inoculated saplings (in later and earlier phases of the experiment, respectively).

Figure 2 with 1 supplement see all
Tree root exudates increase with bacterial inoculation for both the drought and irrigation treatments (A) and decrease with bacterial inoculation after rewetting the droughted trees (B).

Total organic carbon (TOC) in exudate solutions from roots of irrigated and drought-exposed Cupressus sempervirens saplings, with and without bacterial inoculations. Intact roots were incubated for 48 hr to collect exudates during periods of drought (A) and re-irrigation (B). Boxplots show the log 2 of fold change from baseline exudation rate (at the beginning of the experiment) in µg C mg root–1 day–1. Asterisks indicate significant differences based on two-way ANOVA performed with Tukey’s HSD test (n=6, p<0.05) (see Figure 2—source data 2). Coefficients for the correlations between exudate rate (TOC) and rhizosphere abundance of Bacillus subtilis or Pseudomonas stutzeri (as in Figure 1) for the specific tree groups at 1, 3, and 7 days following inoculation (C).

Figure 2—source data 1

Statistical analysis of gas exchange parameters and sapling biomass.

https://cdn.elifesciences.org/articles/79679/elife-79679-fig2-data1-v2.xlsx
Figure 2—source data 2

Statistical analysis of root exudates total organic carbon.

https://cdn.elifesciences.org/articles/79679/elife-79679-fig2-data2-v2.xlsx

Rhizosphere bacteria induce systemic changes in root exudate composition

Roots of drought-exposed trees displayed substantial changes in exudate composition compared to irrigated trees, and more so when inoculated with rhizosphere bacteria (Figure 3—source data 1). Principal coordinate analysis (PCoA) showed that root exudates blends (samples) from drought trees that were inoculated and those from drought trees that were not inoculated partitioned into distinct clusters, with partial overlap (Figure 3A). When samples from irrigated trees were also considered in the PCoA, they behaved similarly, however there was no single sample common to all three groups (Figure 3A). Metabolomics data identified 149 masses (52.6% of the total mass entities) that were significantly different from either Cupressus root tissues or a root-free control solution incubated in the soil, and were hence regarded as Cupressus exudates (Figure 3—source data 1). About 30 of these metabolites were 2- to 10-fold more abundant in the exudates blends when compared to the controls, and the rest were 15- to 25-fold more abundant (Figure 3B, Figure 3—source data 1). Of the latter number, 44 metabolites were significantly different between irrigated and drought trees (p<0.05; Figure 3—source data 1), and 17 metabolites were significantly different between irrigated and re-irrigated trees (p<0.05) (Figure 3—source data 1). In addition, 13 metabolites were significantly enriched or depleted following bacterial inoculation under drought and one metabolite changed following bacterial inoculation after re-irrigation (Figure 3—source data 1). In two metabolites, the interaction irrigation:bacteria was significant under drought, but not under re-irrigation (Figure 3—source data 1). The metabolites that differed between treatments represented compounds of diverse chemistries, including hydroxycinnamic acid conjugates, hydroxybenzoic acid derivatives, organic acid derivatives, purine derivatives and pyrimidine derivatives, amino acids, and amino acid metabolism derivatives, sugars, and sugar derivatives (Figure 3—source data 1). During drought, many metabolites were depleted compared to their abundance under irrigation, however, upon bacterial inoculation, their exudation was enriched again (Figure 3B). During re-irrigation, many metabolites were enriched compared to their abundance under constant irrigation, and bacterial inoculation did not change the metabolite profile (Figure 3C). Importantly, the exudates in drought conditions consisted mainly of secondary metabolites (70% of total metabolites) associated with tree responses to drought stress (Figure 3—source data 1E). In contrast, the metabolite composition under re-irrigation shifted toward the dominance of primary metabolites (85% of total metabolites), and differences compared to irrigated trees were not significant.

Figure 3 with 1 supplement see all
Rhizosphere bacteria and drought induce systemic metabolic changes in root exudates.

Metabolic profiles of root exudate solutions from roots of irrigated and drought-exposed Cupressus sempervirens saplings, with and without bacterial inoculations. Intact roots were incubated for 48 hr to collect exudates, which were analyzed by mass spectrometry. (DB) drought with bacteria, (D) drought w/o bacteria, (IB) irrigated with bacteria, (I) irrigated w/o bacteria. (A) Principal coordinate analysis of the polar metabolite profiles of exudates from drought-exposed saplings with and without bacterial inoculations (DB and D, respectively; left), and in combination with irrigated saplings (I; right; excluding the group of irrigated trees with bacteria for clarity). (B) Heat map analysis of the fold change of each of 149 identified metabolites (rows) relative to control level (at the beginning of the experiment). Columns represent means for each of the four saplings groups (n=4–6). (C) Same as in (B) for the re-irrigation phase. The nine metabolite names indicate metabolites which were tested in vitro for bacterial growth. For the list of metabolites, see Figure 3—source data 1.

Figure 3—source data 1

Data and statistical analysis of root exudates polar metabolites.

https://cdn.elifesciences.org/articles/79679/elife-79679-fig3-data1-v2.xlsx
Figure 3—source data 2

Statistical analysis of root exudates semi-polar metabolites.

https://cdn.elifesciences.org/articles/79679/elife-79679-fig3-data2-v2.xlsx

In addition to the analysis of polar compounds, an untargeted semi-polar metabolite profiling approach yielded 2285 clustered mass signals which were annotated as semi-polar putative compounds during drought. PCA (principal component analysis) of exuded semi-polar metabolites identified in drought trees with and without bacteria showed clusters that partially overlapped (Figure 3—source data 2 and Figure 3—figure supplement 1A). When samples from irrigated trees were also considered, they behaved similarly. Of the semi-polar compounds that changed significantly under drought, 20 were further identified (Figure 3—figure supplement 1). Finally, three semi-polar metabolites increased significantly in inoculated drought trees compared to drought trees without bacteria: isomaltose, clerodane diterpene, and protocatechuic acid (dihydroxybenzoic acid) (compounds 14, 17, and 20, respectively, in Figure 3—figure supplement 1C).

Metabolites exuded by tree roots promote growth of rhizosphere bacteria

It was of interest to examine for potential utilization of exudates by P. stutzeri and B. subtilis. Among the 44 drought-specific root exudate metabolites identified, nine secondary metabolites that had high abundance relative to the baseline measurement were selected. These metabolites represent diverse chemical classes and different levels across the treatments and stages of the experiments (Figure 3B, Figure 3—source data 1; Table 1). Eight of the metabolites showed significant differences between irrigated and drought trees (typically decreasing with drought, except for 3,4 dihydroxymandelate and 3,5-dimethoxyphenol, which increased; Figure 3B, Figure 3—source data 1; Table 1). A single metabolite, 2,3-dihydroxybenzoate, did not change with drought, however increased with bacterial inoculation under drought. Of the other eight metabolites, six also increased with bacterial inoculation under drought, although not significantly in citrulline and p-coumaric acid. Again, 3,4 dihydroxymandelate and 3,5-dimethoxyphenol differed by decreasing with bacteria, significantly for the latter. Following re-irrigation, none of the metabolites changed significantly between treatments, however citrulline and 3,4 dihydroxymandelate increased under re-irrigation (Figure 3C; Table 1, t-test results not shown).

Table 1
Dynamic changes in selected metabolites across treatments (DB, drought and bacterial inoculation; D, drought; I, irrigation) and along the experiment (drought and recovery following re-irrigation).

t-Test results are for pairwise comparisons between the relative changes of a metabolite between two treatments. Significant values are in boldface. Log 2 fold changes of the metabolites are for pairwise comparisons between two treatments.

Metabolitet-TestLog 2 fold change
Drought DB/DDrought D/IDrought DB/DDrought D/IRe-irrigation DB/DRe-irrigation D/I
2,3-Dihydroxybenzoate0.0130.7590.120.01–0.02–0.07
3,4-Dihydroxymandelate0.6750.010–0.040.190.070.14
3,5-Dimethoxyphenol0.0480.040–0.060.050.010.07
Citrulline0.0650.0050.33–0.44–0.100.19
N-acetylserine0.0430.0290.14–0.13–0.020.06
D-chiro-inositol0.0400.0080.23–0.16–0.050.09
Quinate0.0250.0140.26–0.25–0.08–0.14
Nicotinate0.0390.0110.05–0.030.020.02
p-Coumaric acid0.2270.0010.13–0.21–0.040.06

Bacterial cultures were grown on defined media, or, alternatively, deprived from either glycerol or glutamate as carbon or nitrogen source, respectively, which were substituted by root exudation compounds. Among the nine metabolites, the three phenolic acid compounds (2,3-dihydroxybenzoate; 3,4 dihydroxymandelate [protocatechuic acid], and 3,5-dimethoxyphenol), the two amino acid derivatives (N-acetylserine, L-citrulline), and the shikimate pathway acid (D-Quinic acid) were found to support growth of P. stutzeri (Figure 4). Growth of B. subtilis cultures in the media with 3,4-dihydroxymandelate, with or without 2,3-dihydroxybenzoate, was significantly higher compared to media containing glycerol as a carbon source (2.5- to 3-fold increase; p<0.01). Using 2,3- dihydroxybenzoate alone was less efficient than glycerol for B. subtilis growth, and similar for P. stutzeri (Figure 4A). Next, the metabolites N-acetylserine, L-citrulline, and D-quinic acid yielded similar growth rate compared to glutamate as nitrogen source, and more so for P. stutzeri. In B. subtilis cultures, there was a delay in growth with L-citrulline, and a later decrease. D-quinic acid caused a delay in growth initially but eventually reached higher growth than the control (Figure 4B). p-Coumaric acid was used as sub-optimal carbon source for P. stutzeri, whereas nicotinate and D-chiro-inositol did not serve as carbon or nitrogen source (Figure 4—figure supplement 1). Nevertheless, these two compounds might be still involved in other manners in bacterial life around the roots.

Figure 4 with 1 supplement see all
Specific metabolites in tree root exudates enhance bacterial growth.

Growth curves of Bacillus subtilis (left) and Pseudomonas stutzeri (right) in defined media with specific metabolites identified in root exudate blends of drought-exposed Cupressus sempervirens saplings as exclusive carbon (A) or nitrogen (B) source. Glycerol and monosodium glutamate (MSG) were used as controls for carbon and nitrogen sources, respectively. Root-derived metabolites are 2,3-dihydroxybenzoate (2,3 DHB); 3,4 dihydroxymandelate (3,4 DHM); their 1:1 mixture; 3,5-dimethoxyphenol; N-acetylserine; L-citrulline; and D-quinate. Concentrations are 0.4% by weight for all compounds. OD600, optical density at a wavelength of 600 nm (n=4).

Leaf, root, and soil elements respond to drought and bacterial inoculation

Levels of P, Fe, and Zn were significantly lower in leaves in drought compared to irrigated trees during the middle of the drought period (27.05). This effect was mitigated in inoculated trees under drought for P and Fe, but not for Zn (Figure 5A, Figure 5—source data 1). For Mn, the drought effect was not significant, and the bacterial inoculation rather slightly decreased its leaf content. Key elements were further visualized in leaf and root samples. Elemental distribution images (heat maps) of Cupressus leaves showed Mn concentrations in leaf meristems and Fe spreading across the leaves as particles, with leaves of irrigated trees showing higher K levels (Figure 6A and B). Elemental maps of Cupressus roots under irrigation showed co-localization of Fe, K, and Mn (Figure 6C). In Cupressus roots under drought, Fe and Mn were localized to outer portions of the root, whereas K and Cu were mostly at the inner regions (Figure 6D). Overall, roots of irrigated trees had higher Fe, Mn, and Cl than roots of drought trees, with no change in Zn (Figure 6D).

Leaf and soil elements respond to drought and bacterial inoculation.

Leaf elements (A) and soil phosphorous (B) of irrigated (blue shades) and drought-exposed and re-irrigated (brown shades) Cupressus sempervirens saplings, with and without bacterial inoculations. Values are log 2 of fold change from the baseline leaf elements and soil phosphorous (at the beginning of the experiment) in mg kg–1 concentrations. Asterisks indicate significant differences based on by two-way ANOVA performed with Tukey’s HSD test or Bonferroni test (n=6, p<0.05). See also Figure 5—source data 1.

Composite elemental distribution images of irrigated and drought-exposed leaves (A, B) and fine roots (C, D) of Cupressus sempervirens saplings.

Elemental distribution images (heat maps; a single color per element) were created at 20 µm resolution and scan area of 500 µm with a copper target to detect Fe, Mn, Ca, K, and Cl, and molybdenum or gold target to detect Cu and Zn. The mapping was performed using an X-ray fluorescence microscope applying an ultra-high brightness X-ray source with multiple X-ray targets.

The bioavailability of phosphate was measured by assaying the Olsen-P concentration in the soil. It was found to be responsive to drought, decreasing by twofold compared to irrigated soil (p<0.05; Figure 5B). Importantly, Olsen-P of dry soil increased fourfold when inoculated with soil bacteria (p<0.01). Re-irrigation increased Olsen-P back to its level in wet soil and there was no effect of adding bacteria at this stage. There was a positive correlation between leaf and soil P around drought trees under re-irrigation (r=0.58 and 0.67 for inoculated and un-inoculated trees, respectively). In irrigated trees, r values ranged between 0.03 and 0.23. The correlations between leaf P and photosynthesis were typically <0.50, and increased following the first bacterial inoculation (r=0.70 and 0.65 for inoculated drought and irrigated trees, respectively). In later time points, photosynthesis correlated well with leaf P only in drought trees without inoculation (r=0.80–0.88).

Discussion

Soil drought was imposed on young Cupressus trees growing outside in forest soil, decreasing their soil water and P availability (Figure 1—figure supplement 2, Figure 5B). Trees responded with strong reductions in hydraulic and photosynthetic activity, as well as in leaf nutrients (Figure 2—figure supplement 1, Figure 5A). Despite the lower carbon uptake, trees continued to exude altered metabolite blends (Figure 2, Figure 3). Soil bacteria associated with tree roots and grew preferentially in the rhizosphere (than in the bulk soil; Figure 1), where their presence transiently increased with root exudates (Figure 2, Figure 7). Among ~150 specific metabolites, many decreased in response to drought, but increased back when trees were inoculated with bacteria (Figure 3). Out of nine of these metabolites that were tested in vitro on soil bacteria, six were used by bacteria as either carbon or nitrogen source, sometimes enhancing growth more than the standard source (Figure 4). In turn, drought trees that were inoculated had increased biomass increment, as well as leaf nutrients and slightly higher photosynthetic activity before re-irrigation (Figure 5A, Figure 2—figure supplement 1). Upon re-irrigation, trees gradually recovered, while drastically decreasing their root exudates, both in quantity and quality. While our manipulation changed the native microbial community, it permitted the direct observation of the root-bacteria interplay. This interplay was over a native background, rather than over a sterilized or synthetic media, to simulate the natural conditions as much as possible. Our approach introduces more disturbance than, for example, a microbiome approach, however the latter would not permit direct observation of specific strains and their interactions with tree roots, as reported here.

A dynamic rhizosphere interplay between tree roots and soil bacteria under drought stress.

Leaf CO2 assimilation distributed across the tree tissues (1). Under soil drought, leaf gas exchange and tree growth were inhibited (2). Trees recruited root-associated bacteria through changes in exudation rates and composition (3). The metabolites exuded by roots promoted growth of rhizosphere bacteria (4). In turn, the drought-induced reduction in leaf mineral concentrations (e.g. phosphorous and iron) were mitigated by bacteria (5).

Figure 7—source data 1

Statistical analysis of leaf assimilation, root exudation, and their relationships.

https://cdn.elifesciences.org/articles/79679/elife-79679-fig7-data1-v2.pdf

Costs and benefits for trees and soil bacteria

Our results confirm our hypothesis that desiccated trees will suffer drought stress both belowground (reduced water and nutrient uptake; Figures 5 and 6) and aboveground (closed stomata and reduced carbon and water fluxes; Figure 2—figure supplement 1). The hypothesis that bacterial communities will attach to the tree roots (Figure 1) and remain there over time was also confirmed, albeit bacterial abundance decreased with time after inoculation (Figure 1; see Study limitations below). However, this decrease is expected, considering that inoculations contained bacteria in numbers exceeding the system’s capacity, as previously shown (Bever et al., 2012). Next, root exudation increased in response to bacterial inoculations (Figure 2). The phenomenon of stimulation of exudates by bacteria was demonstrated for exudation of primary metabolites (Canarini et al., 2019). However, the level of exudation was lower under drought, in contrast to our former field observations (Jakoby et al., 2020). It is likely that the imposed drought in our experiment was harsher than the dry season conditions in the field, where roots can explore deeper soil layers. Still, the sharp decrease in exudation rate during re-irrigation highlights the relatively higher exudation rate under the combination of drought and bacterial inoculation. Importantly, when calculating the exudation rate as percent of carbon uptake, the rates of inoculated drought trees were actually threefold higher compared to irrigated trees. Root exudates included carbohydrates and organic acids that fed bacterial communities (Figures 3 and 4). Yet, unexpectedly, most root exudates under drought were secondary, rather than primary metabolites. However, this is in agreement with a similar shift observed in the exudate metabolomes of drought-exposed oak trees (Gargallo-Garriga et al., 2018). Indeed, phenolic acid compounds and amino acid derivatives proved to be superior carbon and nitrogen sources than a sugar and an amino acid.

Finally, desiccated trees supplemented with soil bacteria showed better nutrition in P and Fe (Figure 5); but, unexpectedly, did not recover faster than without bacteria. Thus, photosynthesis recovered faster in drought trees that were not inoculated than those that were (Figure 2—figure supplement 1). Considering that inoculated trees invested significantly more carbon into the rhizosphere than un-inoculated drought trees, it is possible that this carbon cost came at the expense of internal tree reserves, later creating a ‘recovery penalty’ for the trees (Gessler et al., 2020). The carbon cost of root exudation has already been accounted for in relation to P foraging (Dijkstra et al., 2013; Wang and Lambers, 2020), where rhizodeposition may be used for P scavenging rather than for decomposition of soil organic matter. On the other hand, slower recovery might indicate a more controlled acclimation response (Bastida et al., 2019; Gessler et al., 2020). Overall, the rhizosphere interplay between trees and bacteria improved tree nutrition under stress, however the physiological benefit was more subtle than rigorous, mostly in better sustaining photosynthesis under drought, as shown before (Morgan et al., 2005; Boutasknit et al., 2020). Correlations between leaf P and photosynthesis indicated that the improved nutrition was associated with improved physiology, especially following the inoculation under drought, and later in drought trees without inoculation, where individuals with higher leaf P had higher photosynthesis. On the bacterial side, benefits include a growth habitat (Figure 1A), as well as food source (Figure 5), but costs are harder to pinpoint. By congregating in the rhizosphere, and specifically, on the roots, bacteria should face higher competition for resources. High bacterial density was shown to increase competition for iron in soil (Gu et al., 2020).

A dynamic rhizosphere interplay between tree roots and soil bacteria

Four lines of evidence in our study support a mutual interaction between trees and bacteria: (1) bacteria were physically attached to root surfaces; (2) bacteria were always ~10-fold more abundant in the rhizosphere than the bulk soil; (3) root exudation increased following inoculation, and was higher during drought than following it (Figure 1—source data 1, Figure 7—source data 1); and (4) many metabolites were enriched following inoculation in drought trees. The higher root exudation during drought than under re-irrigation is not trivial, considering that wet soil provides a better environment than dry soil for both tree roots and bacteria development (Gao et al., 2021). Our temporal resolution and complex experimental setup do not address the question if exudates attracted the bacteria, or, alternatively, whether the bacteria induced root exudation. It is likely that both processes occurred simultaneously. One must also consider alternative explanations, for example, that exudates act directly on soil elements, rather than solicited by bacteria. During drought, changes in soil water content could become either beneficial (increased Mn and P availability) or harmful (decreased Zn availability) to plant nutrition (Misra, 1999). Tree roots exude a plethora of secondary metabolites into the rhizosphere, which aid in the mobilization and uptake of essential macro-elements as N, P, and microelements like Mg, Mn, Zn, and Fe (Michalet et al., 2013). Several studies reported that beneficial rhizosphere bacteria drive an accumulation of elements (Philippot et al., 2013; Igiehon and Babalola, 2018). Although we showed that multiple exudate metabolites promoted bacterial growth, this does not mean that all of these metabolites act as specific cues for specific bacterial strains in the rhizosphere. However, the intimate attachment of bacteria to roots shown here should ensure the harvest of these metabolites by bacteria.

Plant-microbe interactions at the root level exert strong effects on the whole plant and on nutrient cycling, which are mostly explained by root traits (Teste et al., 2017; Guyonnet et al., 2018). Here, we greatly enhanced the abundance of two bacterial strains, each showing unique dynamics. Previous works found that during the dry season, root-associated communities show elevated abundance for Bacillus, whereas in the rainy season, Proteobacteria increase (Shakya et al., 2013; Taketani et al., 2017). In our experiment, P. stutzeri decreased under drought faster than B. subtilis, but took better advantage of the root exudate metabolites 3,5-dimethoxyphenol, L-citrulline, N-acetylserine, D-quinic acid, and p-coumaric acid (section below). The rhizosphere roles of these two strains might be complementary: while Pseudomonas is abundant under humid conditions (Mendes et al., 2013; Xu et al., 2018), Bacillus dominates plant microbiomes under arid conditions, where Pseudomonas cannot survive.

The chemistry of tree-microbe interactions in the rhizosphere

Interestingly, the phenolic acid compounds that promoted bacterial growth the most (2,3-dihydroxybenzoate; 3,4-dihydroxymandelate; and 3,5-dimethoxyphenol) did not decrease with drought while others did. This observation supports the view of their role in feeding the root bacteria as part of a tree drought resistance strategy. 2,3-Dihydroxybenzoate is a bacterial growth factor (Young et al., 1967). It is the biosynthetic precursor of petrobactin, an unusual 3,4-dihydroxybenzoate catecholate siderophore essential for growth of Bacillus anthracis and Pseudomonas putida (Garner et al., 2004; Neal et al., 2012). Hydroxybenzoic acids also are known as being allelopathic and anti-fungal compounds of plant exudates (Yu and Matsui, 1994; Mandal et al., 2009a). They have been characterized as defensive secondary metabolites that are released by roots of cereals such as wheat and maize to alter root-associated fungal and bacterial communities (Neal et al., 2012; Hu et al., 2018). Benzoic acid can stimulate growth of a Pseudomonas species (Kamilova et al., 2006), and p-hydroxybenzoate can serve as a sole carbon source for the growth of another Pseudomonas species (Harwood et al., 1984). Recently, Ankati and Podile, 2019, showed that colonization of groundnut roots by related Bacillus and Pseudomonas strains resulted in exudation of metabolites such as benzoic and salicylic acids (both identified here) that facilitated root colonization, suppressed fungal growth, and promoted plant growth. Other metabolites identified such as quinate and citrulline were either among the most abundant in exudate blends throughout the experiment, or those that increased in inoculated trees. Quinate is a major compound in Pinus radiata exudates and was shown to stimulate over 700 soil bacterial community taxa (Shi et al., 2011; Zhalnina et al., 2018) and citrulline was identified in soybean exudates under P deficiency (Tawaraya et al., 2014).

Bacterial metabolism plays a key role in root colonization, and carbohydrate metabolism genes of Pseudomonas simiae were linked to its colonization of Arabidopsis roots (Naylor and Coleman-Derr, 2017). Those findings are consistent with our finding of viable carbon and nitrogen sources supporting bacterial growth among root exudates. However, other metabolites were not supportive of growth of both strains. Recently, Ma et al., 2018, demonstrated the effect of nicotine from tobacco root exudates on an antagonistic bacterium for its root colonization, and control of chemotaxis of soil-borne pathogens. p-Coumaric acid has been identified in plant root exudates or residues and in soils around many plant species (Hao et al., 2010). 4-Coumaric acid was identified with anti-fungal activity in tomato roots (Mandal et al., 2009b). p-Coumaric acid increased the relative abundances of microbial taxa with phenol-degrading capability (Zhou et al., 2017). Considering our observations of inhibition of Bacillus growth by p-Coumaric acid, it might be engaged in tree root interaction through generating negative tree-soil microbial interactions. D-chiro-inositol is one of multiple phospho-inositides, essential metabolites, as well as labile messengers that regulate cellular physiology, and are involved in tolerance to abiotic stress (Stevenson et al., 2000).

Among the many exudate compounds elucidated in our analysis, metabolites other than the nine tested here were of interest. These include indole-3-acetic acid (IAA), namely the plant hormone auxin, which is known as active in hydrotropism. Here, IAA had a moderate level which decreased both under drought and re-irrigation. Auxin can be produced in the rhizosphere both by roots and PGPRs (Tsunoda and van Dam, 2017). The amino acid proline, a known plant osmoprotectant under drought, had a moderate level, which decreased under drought, mildly recovering under re-irrigation both with and without inoculation. Another notable metabolite is shikimate, whose pathway includes the aforementioned quinate, and has already been shown to increase root colonization by beneficial Bacillus amyloliquefaciens (Rolfe et al., 2019). Here, shikimate had a high level, mildly decreasing under drought. Gamma aminobutyric acid is another well-known metabolite, which was identified here at a high level, decreasing under irrigation and inoculation. Last, we identified trigonelline, which is known to induce nod gene in Rhizobium bacteria.

Study limitations

We offered a glimpse into the rhizosphere, which was not, however, free of limitations. First, our approach ignored the native rhizosphere microbiome, which probably affected many of the studied parameters. In a complementary fungal study on these young trees, we were unable to identify mycorrhizal partners, however the root pathogen Plectosphaerella cucumerina was identified. Nevertheless, the two bacterial species that we used were typical root colonizers of Cupressus, and the Pseudomonas strain was isolated from the very same forest soil. Second, manipulating the rhizosphere with inoculations can also be criticized as creating disturbance. Moreover, except for the re-inoculated irrigated trees, bacterial communities were not stable. Still, the two bacteria species were released into the forest soil to compete as any other microbial species in the soil system. Third, our sample size was relatively small, with six trees per treatment, however compensated by the Rhizobox design and the intensity of measurements. Fourth, passive exudation and leakage from drought-stressed roots cannot be ruled out in our experiment. However, the observed shift in metabolites and the sensitivity to inoculations support an active exudation. Fifth, whether our results are transferable from saplings to mature trees is questionable. We argue, however, that studying saplings is equally important, as establishment is often the bottleneck for tree survival in the forest (Pozner et al., 2022), and it is the period where microbial partners make the most difference. This is increasingly important in the case of shallow-rooted tree species such as C. sempervirens (Rog et al., 2021). In the forest, mycorrhizal fungi, which were missing here, can have large effects too (Meier et al., 2013). Specifically, C. sempervirens hosts a diversity of active arbuscular mycorrhizal species (Avital et al., 2022). While the latter generally help in tree P nutrition, their activity is mostly reduced under drought. Future studies should test the functions of the three-kingdom interactions among trees, fungi, and bacteria. Finally, our exudate collection technique cannot entirely rule out metabolites from bacterial, rather than plant, origin, nor can our incubations account for loss of exudates to bacterial consumption. Root-free and dissected root controls significantly increase our confidence in the results, but both false positives and false negatives are possible.

Conclusion

In this multidisciplinary study we combined advanced approaches in microbiology (genetic modifications and epifluorescence), plant physiology (leaf gas exchange and exudate collection), and organic chemistry (metabolic profiling) to decipher the dynamic rhizosphere interplay between tree roots and soil bacteria under drought stress. The novelty of our results is by providing a holistic view of the consequence of events in the rhizosphere, from tree water and nutritional shortages, through root exudation of specific metabolites (e.g. phenolic acid compounds) and their benefit to bacterial growth, and finally, to improved tree nutrition (leaf P and Fe). Our next steps aim at further elucidating the mechanisms utilized by both trees and bacteria for their mutual communication and cooperation in the rhizosphere.

Materials and methods

Experimental design and plant material

Request a detailed protocol

The study was performed in custom-made 25 L containers termed ‘Rhizoboxes’ in a net-house in the Weizmann Institute of Science, Rehovot (Israel). These 25 cm long × 25 cm wide × 40 cm tall containers were produced ad hoc from transparent poly(methyl 2-methylpropenoate) and were filled with a mixture (1:1) of washed siliceous sand and forest soil, to preserve the naturally occurring fungi and bacteria in the media, while preventing soil clogging. For every 10 L of mixed soil, 0.5 L Tuff was added to improve aeration. Soil properties were measured before and in the end of the experiment (Table 2). The forest soil was collected from Harel forest, located at the Judean foothills (31° 43′N 34° 57′E, 320 m elevation), 4 km south-west of Beit Shemesh, Israel. Rhizobox containers were covered in black plastic bags, to facilitate root growth, uncovered only during root sampling or exudate measurements. The experiment was conducted during May-July 2019. Weather conditions in Rehovot during May-July 2019 were the following: overall minimum, maximum, and average temperatures were 11.3°C, 34.5°C, and 25°C, respectively, average VPD was 2.00 kPa, and no rain (Beit-Dagan meteorological station, 10 km north of the research institute; Figure 1—figure supplement 1). Global solar radiation was 300–350 W m–2, reduced by ~15% by the net-house.

Table 2
Soil properties from pots where saplings were grown under treatments of drought or irrigated with or without bacteria.
(A) Soil chemistry: pH, electrical conductivity (EC), sodium absorption ratio (SAR), and minerals content (Cl, Na, Ca, Mg, N-NH4, N-NO3, K concentrations)
SoilTreatmentpHEC(ds m–1)Cl(mg kg–1)Na(g L–1)Ca(mg kg–1)Mg (mg kg–1)SAR(mg kg–1)N-NH4 (mg kg–1)N-NO3 (mg kg–1)K of CaCl2 (mg kg–1)
Pre (n=3)Pre (n=3)7.5±0.061.88±0.0787.7±3.054.4±3.6750.7±8.811.8±0.32.8±0.0918.9±3.025.63±3.449.43±3.9
DroughtWith bacteria7.52.51170.0179.41209.917.27.223.83.739.6
DroughtW/o bacteria7.72.64188.980.71222.416.93.219.13.537.3
IrrigatedWith bacteria7.66.31513.654.43358.549.21.337.22.542
IrrigatedW/o bacteria7.56.27544.7191.12716.351.15.141.93.639.6
(B) Physical structure: sand, silt, and clay content, calcite content (CaCO3) and soil porosity (SP).
SoilTreatmentSand %Salt %Clay %CaCO2SP
Pre (n=3)Pre (n=3)9±3.4619.67±2.0871.33±5.513.67±1.1537.33±2.08

Twenty-four 2-year-old C. sempervirens saplings were obtained from the Jewish National Fund, Israel Forest Service (KKL) nursery and transplanted into Rhizoboxes in the Weizmann Institute of Science campus (Rehovot, Israel) in September 2018. Prior to that, seedlings were germinated from seeds collected at Beit She’arim in the Galilee, Israel, and were transferred to plastic ‘quick-pots 585’ (200 mL plugs, 5 cm×5 cm) in the KKL nursery. There, the plants were grown with 2% starter fertilization and irrigated with fertilization until September 2018. The saplings were kept at the net-house, one sapling per Rhizobox, under optimal irrigation of 1.2 L day–1 for a 7-month period, to allow root growth into the mixed forest soil. Saplings were 10.3±0.6 mm in stem diameter and ~50–100 cm in height. Next, saplings were equally divided into four groups and were placed randomly in the net-house. Two groups (12 trees) were exposed to drought conditions, induced by withholding irrigation for 32 days. The volumetric soil water content was measured on each measurement day with moisture sensor (EC-10, Decagon Devices, Pullman, WA) (Figure 1—figure supplement 2). The timing of re-irrigation was determined according to the trees’ physiological conditions, immediately after transpiration and stomatal conductance decreased to zero (Figure 2—figure supplement 1). The other two groups (12 trees) were irrigated continuously with 1.2 L day–1. On the 9th day from the irrigation cessation (May 20, 2019), one drought group and one irrigated group were inoculated with a 1 L bacterial solution (see below). The same groups were inoculated again on the 5th day after re-irrigation (June 16, 2019). Inoculations were made with two native soil bacteria strains over a forest soil background, to simulate natural conditions while permitting quantification and visualization of the bacteria and their interactions with tree roots. Although reducing the complexity of the community, it was shown that pairwise interactions between soil bacteria predict well the community structure (Friedman et al., 2017). All measurements were conducted in the following days: (1) 11 days before irrigation cessation (day –11, 2 May); (2) 24 hr after the first inoculation (day 10, 21 May); (3–5) once a week following irrigation cessation (days 16, 23, and 30, on 27 May, 3 June, and 10 June, respectively); (6) 4 days after re-irrigation (day 36, 16 June); (7) following the second inoculation (day 37, 17 June), and (8) 26 days after re-irrigation (day 58, 8 July). Bacteria counts were made 1, 3, 7, and 15 days (days 10, 13, 16, and 24), following the first inoculation, and similarly following the second inoculation (days 38, 41, 44, and 51). Root exudates were collected at 5 days before drought treatment (day –5, 7 May), 1 day after the first inoculation (day 10, 21 May), and 4 days after re-irrigation (day 36, 16 June; Figure 1—figure supplement 2). Total biomass assessment was determined on six saplings, which were sacrificed before the experiment started and at the end of the experiment, on all saplings. The saplings were separated from the soil and dried at 60°C for 3 days in order to measure biomass.

Soil water content and soil properties

Request a detailed protocol

Soil water content (%, v/v) was measured using a dielectric constant EC-5 soil moisture sensor connected to an Em50 Data logger (Decagon Devices Inc, Pullman, WA) which was programed to record observations at 10 min intervals. The sensors were located in the Rhizobox at 15 cm depths with two repeats for each treatment. Prior to the experiment, soil samples were collected from three different locations of the soil mixture pile, before Rhizoboxes were filled, and sent for soil analysis at Gilat Field Services Laboratory, Israel. Soil structure was 9/20/71% sand/silt/clay, with 3.7% CaCO3 and 37% porosity. Electric conductivity was 1.9 dS m–1; pH was 7.5; sodium absorption ratio 2.1; N-NO3, N-NH4, K, and Olsen-P content was 18.9, 25.6, 49.4, and 11.7 mg kg–1, respectively. Cl, Na, Ca, and Mg content was 224.8, 6.1, 270.1, and 30.4 mg L–1, respectively.

Soil phosphorous extraction and quantification

Request a detailed protocol

Soil samples were collected from each container at three time points during the experiment. The soil was oven dried (60°C for 48 hr) and ground to pass a 2 mm sieve. Each sample was divided into three replicates of 1 g. Bicarbonate (30 mL of 0.5 M solution) was applied and mixed for 1 hr, for extraction according to the Olsen method (Olsen et al., 1954). The extraction was centrifuged at 10,000 g for 10 min, and then an aliquot was transferred to a new tube, and pH was adjusted to 6.0 by H2SO4. After extraction, total phosphate was measured using the Abcam colorimetric phosphate assay kit (ab65622, detection limit of 1 μM), and quantified with a Tecan Infinite M200 plate reader (Tecan, Grödig, Austria).

Live imaging of bacterial root colonization using confocal microscopy

Request a detailed protocol

Bacterial root colonization was visualized and photographed by a confocal super-resolution (CLSM) Nikon A1 + laser system equipped with Plan-Apochromat ×10/NA0.45 (Nikon, Tokyo, Japan). B. subtilis cells expressing GFP were irradiated using a 488 nm laser beam, while P. stutzeri cells expressing mCherry were irradiated using a 555 nm laser beam. For each experiment, both transmitted and reflected light were collected. System control and image processing were carried out using NIS-Elements C software version 4.0 (Nikon).

Pseudomonas strain isolation, characterization, and construction

Request a detailed protocol

Samples of forest soil were collected from Harel forest (5–15 cm depth) in order to isolate native PGPR Pseudomonas strains from tree roots. Soil aliquots of 10 g each were placed in 50 mL PBS (phosphate-buffered saline) and mixed vigorously for 1 min. Subsequently, serial dilutions were plated on LB agar plates and were incubated at 23°C for 24 hr. Pseudomonas strains were isolated and genomic DNA was purified using the Wizard Genomic DNA Purification Kit (Promega, Madison, WI) according to the manufacturer’s instructions. For DNA sequencing, the 16S rRNA genes were amplified using PCR with the forward primer (5-799F AACMGGATTAGATACCCKG) and the reverse primer (6-1192R ACGTCATCCCCACCTTCC′) (Chelius and Triplett, 2001). Sequences were aligned with other sequences downloaded from the GenBank database using NCBI BLAST. P. stutzeri was transformed with plasmids pTns2 and pUC18T-mini-Tn7T-Gm-mCherry, to generate P. stutzeri (attTn7::mCherry) strain with antibiotic resistance as selectable marker, and fluorescence for visual detection (McFarland et al., 2015).

Bacterial strains, inoculum preparation, and bacterial quantification

Request a detailed protocol

B. subtilis and P. stutzeri were used in this study to inoculate the trees. These strains represent two major groups of soil bacteria, that is, the phylum Firmicutes, and the class Gammaproteobacteria, respectively. In addition, both strains offer the advantage of cultivation and genetic manipulation, and both are known to interact with roots of other plants (Weller et al., 2002; Kravchenko et al., 2003; Rudrappa et al., 2008; Zhang et al., 2014; Sasse et al., 2017; Zhang et al., 2017; Mhlongo et al., 2018). B. subtilis (NCIB 3610) containing constitutive GFP (amyE::Phyper-spank-gfp-cm) was a gift from R Losick (Harvard Medical School, Boston, MA) and P. stutzeri was isolated from soil (above). Although the specific B. subtilis strain used here was not recovered from the forest soil, we did identify another B. subtilis strain in our forest soil. In multiple screens of rhizosphere bacteria in our forest site, a total of 54 bacterial strains were identified on plates. Of which, B. subtilis and P. stutzeri were the most consistent, evidencing their important role in our system. Inoculum was prepared by growing the bacterial isolates B. subtilis and P. stutzeri in LB broth at 37°C for 16 hr, followed by centrifugation (4000 rpm for 10 min at 23°C), washing and re-suspension in carbon-free nutrient solution to obtain a cell density of 5 × 108 colony forming units (CFU) mL–1 (OD600 = 1). Equal amounts of bacterial cultures of the two species were mixed and were spread homogeneously in the Rhizobox. To avoid any bias, trees which were not inoculated received the same amount of water as in the inoculation solution at the time of inoculation. To isolate the two bacterial species, soil and roots (10 g of each) were collected into 10 mL PBA and homogenized by 1 min vortex followed by 10 min sonication. The homogenate was serially diluted in sterile PBS and the dilutions were plated on two LB agar medium supplemented with gentamicin (Gm, 30 μg mL–1) to isolate P. stutzeri or chloramphenicol (Cm, 25 μg mL–1) to isolate B. subtilis. Total bacterial counts obtained were expressed as log CFU g–1 of dry weight of soil or root. The relative abundance was estimated by counting each strain of bacteria separately and diving by the total amount of bacteria.

To construct growth curves in response to specific plant metabolites identified in root bacterial exudates, the strains were grown in MSgg medium (Bloom-Ackermann et al., 2016) containing MS salts and 125 µM FeCl3, either with glycerol (0.4% w/v) and glutamate (0.4% w/v) or with root tree metabolite as carbon or nitrogen source (0.4% w/v). Metabolites were identical with those identified in root exudates (Table 1), except for nicotinate, which was tested in the form of nicotine. The cultures were inoculated at an initial OD600 = 0.04 and then incubated at 25°C with shaking at 150 rpm for up to 48 hr in the Tecan plate reader. OD600 were measured continuously over 48 hr.

Leaf gas exchange and water potential

Request a detailed protocol

Leaf physiology was monitored once a week for all the trees in the experiment by leaf gas exchange (i.e. transpiration and net assimilation) with a WALZ GFS-3000 photosynthesis system (Walz, Effeltrich, Germany), equipped with a lamp, set to a light intensity of 1000 µmol m–1 s–1 at ambient air temperature and humidity. CO2 was set to 400 ppm, close to the ambient CO2 level. Gas exchange rates were further calibrated to the actual leaf area by scanning measured leaves to determine projected leaf area. Gas exchange measurements were accompanied by leaf water potential (Ψl) on one to three leaves from each sapling with the Scholander pressure chamber technique (Scholander et al., 1965). Leafy branchlets of 5–6 cm length were cut from all the trees in the experiments and put in a pressure chamber (PMS Instrument, Albany, OR) fed by a nitrogen gas cylinder. Gas pressure within the chamber was gradually increased (~1 MPa min−1) until water emerged from the protruding cut branch surface, and the negative value of the pressure was recorded as water potential (Ψl) in MPa.

Sampling root exudates

Request a detailed protocol

Root exudates were collected from intact fine roots using a non-soil syringe system modified from Phillips et al., 2008. Root tips were sampled from the side windows of the Rhizobox, and were remained attached to the trees during the entire procedure until harvest. We gently washed the intact fine roots using a spray bottle, with autoclaved carbon-free nutrient solution (0.5 mM NH4NO3, 0.1 mM KH2PO4, 0.2 mM K2SO4, 0.2 mM MgSO4, 0.3 mM CaCl2) and fine forceps to remove soil particles and other possible contaminants. The fine roots were placed into a 20 mL sterile plastic syringe and filled with 0.5–1.3 mm acid-washed glass beads and 10 mL autoclaved carbon-free nutrient solution. Then, the syringes were covered with aluminum foil and covered with soil to block sunlight and heat. After 48 hr, the nutrient solution was collected from each syringe system. An additional 10 mL of the double distilled water was flushed through the syringe system to obtain a representative carbon recovery. Two samples and one control (solution subjected to the same process without a root; metabolites that were found in control samples were not determined as exudate metabolites) per Rhizobox were included. Small carbon amounts that were found in the control (root-free) tubes were regarded as contamination and were subtracted from the carbon amounts in the samples. At the end of incubation, the two samples were pooled together and then separated into two tubes, one for exudation rate and the other for metabolomics analysis (below). All solutions were filtered immediately through a 0.22 µm sterile syringe filter (Millex PVDF, Millipore Co., Billerica, MA) and stored in the lab at –80°C until analysis. The solutions were analyzed for dissolved organic carbon on a total organic carbon analyzer (Aurora 1030 W TOC Analyzer coupled with a 1088 rotatory TOC auto sampler; OI Analytical, TX). Root exudation rates were calculated as the total amount of carbon flushed from the pooled root system over the incubation period divided by root dry weight of the investigated root strand, and hereafter referred to as specific exudation rate (μg C mg root–1 day–1). After root exudate collection, roots were cut off the tree and dried by oven (48 hr, 60°C) and then weighed.

Sample preparation for metabolite profiling and metabolite extraction

Request a detailed protocol

Aliquots of root exudates were prepared based on the (lower) amount of organic carbon (aforementioned TOC measurement) in the control samples, in order to normalize for variations in quantity among samples. Aliquots were freeze-dried by a lyophilizer. Extraction and analysis were performed as previously described (Dong et al., 2016), with some modifications: lyophilized exudates and controls were extracted with 1 mL of a pre-cooled (−20°C) homogenous methanol:methyl-tertbutyl-ether (TMBE) 1:3 (v/v) mixture containing the following internal standards: 0.1 μg mL−1 of phosphatidylcholine 34:0 (17:0/17:0) and 0.15 nmol mL–1 of LM6002 (Avanti) standard mix. The tubes were vortexed and then sonicated for 30 min in ice-cold sonication bath (taken for a brief vortex every 10 min). Then, UPLC grade water:methanol (3:1, v/v) solution (0.5 mL) containing internal standards (13C and 15N labeled amino acids standard mix; Sigma) was added to the tubes followed by centrifugation. The upper organic phase was transferred into a 2 mL Eppendorf tube. The polar phase was re-extracted as described above, with 0.5 mL of TMBE. Both organic phases were combined and dried in speedvac and then stored at −80°C until analysis. The lower, polar phase used for polar metabolite analysis was stored at −80°C until analysis.

LC-MS polar metabolites analysis and identification

Request a detailed protocol

Metabolic profiling of polar phase was done as previously described (Zhang et al., 2016) with minor modifications described below. Briefly, analysis was performed using Acquity I class UPLC System combined with mass spectrometer (Thermo Exactive Plus Orbitrap; Waltham, MA). The mass spectrometer was operated under the following parameters: full MS/dd-MS2 mode (1 μscans) at 35,000 resolution from 75 to 1050 m/z, with 3.25 kV spray voltage, 40 sheath gas, 10 auxiliary gas and negative ionization mode. The LC separation was done using the SeQuant Zic-pHilic (150 mm × 2.1 mm) with the SeQuant guard column (20 mm × 2.1 mm) (Merck; Kenilworth, NJ). The mobile phase A was 20 mM ammonium carbonate plus 0.1% ammonia hydroxide in water and the mobile phase B was acetonitrile. The flow rate was kept at 200 μL min–1 and the gradient was as follows: 0–2 min 75% of B, 14 min 25% of B, 18 min 25% of B, 19 min 75% of B, for 4 min. The data processing was done using TraceFinder (Thermo Fisher software), where detected compounds were identified by retention time and fragments were verified using in-house mass spectra library. The results were normalized using the internal standards peak area.

LC-MS semi-polar metabolites analysis and identification

Request a detailed protocol

Metabolic profiling of semi-polar phase was performed using Waters ACQUITY UPLC system coupled to a Vion IMS qToF mass spectrometer (Waters Corp., Milford, MA). The LC separation was performed as previously described (Itkin et al., 2011) with the minor modifications described below. Briefly, the chromatographic separation was performed on an ACQUITY UPLC BEH C18 column (2.1×100 mm, i.d., 1.7 μm) (Waters Corp., Milford, MA). The mobile phase A consisted of 95% water (UPLC grade) and 5% acetonitrile, with 0.1% formic acid; mobile phase B consisted of 100% acetonitrile with 0.1% formic acid. The column was maintained at 35°C and flow rate of the mobile phase was 0.3 mL min–1. Mobile phase A was initially run at 100%, and it was gradually reduced to 72% at 22 min, following a decrease to 0% at 36 min. Then, mobile phase B was run at 100% until 38 min; next, mobile phase A was set to 100% at 38.5 min. Finally, the column was equilibrated at 100% mobile phase A until 40 min. MS parameters were as follows: the source and de-solvation temperatures were maintained at 120°C and 350°C, respectively. The capillary voltage was set to 2 and 1 kV at negative and positive ionization mode, respectively; cone voltage was set at 40 V. Nitrogen was used as de-solvation gas and cone gas at the flow rate of 700 and 50 L h–1, respectively. The mass spectrometer was operated in full scan HDMSE negative or positive resolution mode over a mass range of 50–2000 Da. For the high-energy scan function, a collision energy ramp of 20–80 eV was applied, and for the low-energy scan function 4 eV was applied. Leucine-enkephalin was used as lock-mass reference standard. LC-MS data were analyzed and processed with UNIFI (Version 1.9.4, Waters Corp., Milford, MA). The putative identification of the different semi-polar species was performed by comparison accurate mass, fragmentation pattern, and ion mobility (CCS) values to an in-house semi-polar database, where several compounds were identified vs. standards, when available. Several compounds were identified by their theoretical fragmentation, accurate mass, and CCS.

Leaf elements quantification

Request a detailed protocol

C. sempervirens leaves from each tree were collected at the time points indicated above. They were set to dry in an oven at 60°C for 48 hr and ground to obtain a fine powder. Next, 0.1 g leaf powder was weighed and burned to ash at 550°C for 6 hr. Next, 0.5 mL of fresh nitric acid (HNO3 69%; Merck) was added and incubated for 5 days (Laursen et al., 2009). Subsequently, 19.5 mL of deionized water (Millipore, Milli-Q Biocel Water Purification System, Germany) was added and filtered. The elements quantification in the leaves was performed using inductively coupled plasma mass spectrometry (ICP-MS; Agilent 7500ce; Agilent Technologies, Wokingham, UK) tuned in standard mode. The plasma power was operated at 1450±50 W and the argon carrier and make-up gases were set at 0.83 and 0.17 L min−1, respectively. Sample uptake was maintained at ca. 0.1 mL min−1 by a self-aspirating perfluoroalkoxy micro-flow nebulizer (Agilent Technologies). Elimination of spectral interferences was obtained by the use of an octopole ion guide with the cell gases helium or hydrogen (Laursen et al., 2009). For the series of 235 C. sempervirens samples, three replicates of certified reference material NIST 1575a (Pine Needles) and NIST 1547 (Peach leaves) were included. Only data deviating <±10% from the certified reference values were retained.

Leaf and root element mapping

Request a detailed protocol

Elemental mapping of leaf and root samples was performed using an X-ray fluorescence microscope (AttoMap; Sigray, San Francisco, CA) at the company’s labs. The system uses an ultra-high brightness X-ray source with multiple X-ray targets to provide optimal detection sensitivity; a novel X-ray optics providing large focused X-ray flux, small focus (<10 µm) and large working distance (>25 mm); and a high speed detector. Elemental distribution images (heat maps) were created at 20 µm resolution and scan area of 500 µm with a copper target to detect Fe, Mn, Ca, K, and Cl, and molybdenum or gold target to detect Cu and Zn.

Statistical analysis

Request a detailed protocol

Due to the multidisciplinary nature of the study, statistical analysis was designed separately for the plant physiology and microbial inoculation sections, and for the metabolomics and bacterial growth sections. The two major factors tested in our experiment were drought (treated vs. irrigated saplings) and bacterial inoculation (with and without). All measurements were compared to the baseline samples in the beginning of the experiment, except for the CFU. Effects on root exudation rates were assessed by means of two-way ANOVA. Effects on physiological parameters, tree biomass, soil and leaf elements, and CFU values were assessed by means of a two-way repeated measures ANOVA using the general linear model procedure in using Origin 7 (Origin Lab Corporation, Northampton, MA). ANOVA was followed by Bonferroni tests, and Tukey’s HSD test to determine where the significant difference lay within the dataset. p-Value < 0.05 was taken to be significant. Data presented are the means of at least six replicates (from independent saplings) unless otherwise stated. Thereafter, technical replicates were done while including these controls in the experiment, to ensure reproducibility.

Metabolome values were assessed by means of two-way ANOVA, followed by post hoc Tukey’s test with FDR correction and p-value adjustment for multiple comparisons. Statistically different masses, comparing those from irrigated and drought-exposed saplings, with and without bacterial inoculations, were assigned to metabolites (Figure 3—source data 1 and 2). Then, the chemical annotation of significantly different metabolites was inspected manually and classified according to four levels of confidence of metabolite identification as previously described (Dunn et al., 2013; Sumner et al., 2007; Schymanski et al., 2014). PCoA was performed using MetaboAnalyst V4.0, and the metabolites analysis was performed using R and the interface RStudio. Bacterial growth curves were constructed in Matlab (MathWorks, Natick, MA), and presented as means of four biological replications with standard deviations.

Data availability

All data related to the study are reported in the manuscript and Supplementary Information. Source Data files 1-6 are included in the submission.

References

  1. Report
    1. IPCC
    (2014)
    Summary for Policymakers. In Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
    Cambridge/New York: Cambridge University Press.
  2. Report
    1. IPCC
    (2019)
    IPCC Special Report on Climate Change, Desertification, Lan Degradation, Sustainable Land Management, Food Security, and Greenhouse Gas Fluxes in Terrestrial Ecosystems
    Cambridge, UK & New York, NY, USA: Cambridge University Press.
  3. Book
    1. Olsen SR
    2. Cole CV
    3. Watanabe FS
    4. Dean LA
    (1954)
    Estimation of Available Phosphorus in Soils by Extraction with Sodium Bicarbonate. USDA Circular 939
    Washington D.C: U.S. Government Printing Office.

Decision letter

  1. Meredith C Schuman
    Senior and Reviewing Editor; University of Zurich, Switzerland
  2. Alexander Weinhold
    Reviewer; iDiv, Germany

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "A dynamic rhizosphere interplay between tree roots and soil bacteria under drought" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Alexander Weinhold (Reviewer #1); Richard Phillips (Reviewer #3).

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife in its current form. If the concerns raised by the reviewers can be addressed, the study could be considered for publication in eLife, but the current requests go beyond what we would consider under revisions.

Specifically, important strengths of this study are its attempt to investigate root exudate composition while manipulating native root-microbe interactions and drought stress, in order to draw conclusions about ecological roles of root exudates in relevant scenarios, while furthermore using methods originating in different disciplines for the appropriate manipulation and assessment of experimental factors. The topic is important and many aspects of the experimental design well-chosen, especially considering ecological relevance of the system and the potential power of factorial design, and such a study could be of interest for eLife. Measuring metabolites released from roots and connecting them to their putative functions is extremely difficult, even more so when done with trees and actual soil. Yet all reviewers identified substantial flaws in the analysis and presentation of the results, and independently of these concerns, it is difficult to connect the various results to the authors' stated research questions. Thus, the analysis is inconclusive, and it is generally not clear what we learn out of this study regarding trees under drought stress engaged in interactions with soil bacteria. This is reflected in the divergent focus of the individual reviews, each of which should provide very helpful suggestions for addressing these concerns. In addition to the points raised by the reviewers (below), I would note the following: (1) It is not clear how the bacteria were chosen, except that these are two native soil strains, which is commendable – but are they among the most common, most frequently found associated with the study plant, … ? (2) Metabolite naming is inconsistent, e.g. Figure 7 refers to 3,4-DHB which I think is in fact 3,4-DHM, a supplementary figure refers to nicotine which I think is nicotinate. (3) Heat map colors are confusing, indicating up-regulation (red) and down-regulation (blue) when in fact everything is expressed as a positive fold change, which would be best shown in darker (highest) to lighter (lowest) colors. (4) The first results show that, while roots are recruiting, the experiment is carried out with an unstable microbial community, which makes it more challenging to interpret the results (stability in the experimental microcosms is achieved only for the re-inoculated irrigation treatment). (5) It is misleading to refer e.g. to "drought-specific" root exudate metabolites, as Figure 5 shows that all exudate metabolites are detected across all treatments even if they may have different patterns in each treatment (hard to tell from the color scheme in Figure 5 – difference between slightly darker and slightly lighter yellow hard to assess).

Reviewer #1 (Recommendations for the authors):

Oppenheimer-Shaanan et al. investigated the how inoculation with root bacteria can changes the ability of the Mediterranean cypress to cope with drought stress. Therefore, they compared root exudation (as TOC) of tree saplings under drought stress and under constant irrigation. They show that root exudation increases when trees are inoculated with root bacteria. Inoculated trees also showed higher capabilities to recover from drought stress.

The combination of the different methodologies provides an interesting view, or glimpse as the authors wrote, on the interaction between tree roots and associated bacteria under drought stress. The use of the rhizoboxes, the collection of root exudates and the analysis of the bacteria is very nice experimental set up. It allows to see the different effect of abiotic stress on the different players in the interaction (soil, roots, bacteria, leaves) and to make connections on how they are interacting.

However, I see here some shortcomings in the analysis of the data and the presentation. Especially, the metabolomics data are not easy to understand and could be much better presented. Furthermore, I think that the emphasis that is put on the fold changes of certain metabolites is too high. From this are conclusions drawn that are not well supported by the data and makes it hard to follow later the discussion.

Condensation and maybe a different statistical analysis in some figures in the metabolomics analysis would help to better visualize the changes and to see clearer the effects of bacteria and drought.

I don't feel qualified to review the bacterial stain isolation, characterization and construction.

I am not used so much to the use of those log2 fold changes. That is why I need to do some calculations. If I did those right most of the log2FC in Table 1 are very low there and all smaller than 0.5/-0.5 that means a fold change of less than 1.5 I am wondering if those will have a physiological effect. You than showed the effect of some of the metabolites like in FIGURE 6, where you are testing p- coumaric acid for bacterial growth. In the source data the fold change for this metabolite is either 1.09 or 1.15 only (if I calculate back from your log2 value). I think that is very low and I have my doubts that you can draw some conclusion from that. I acknowledge the effects that you see in bacteria assays, and I don't doubt that. But I just think that the difference in concentrations would not matter. How would this look in a comparison where you for instance test different concentrations of e.g. p-coumaric acid and see if the drought or irrigated concentrations have different effects.

I have also some difficulties with the interpretation and analysis of the metabolomics data. The PCA is a good way of visualizing the differences between treatment groups, but I think a PCoA would be better here. It is getting more and more used in metabolomics research and has to advantage to deal better with non-normal distributed data, which you will most likely have. In combination with the PCoA a PERMANOVA might help you to show effects of the treatment instead of the visual comparison. That would also make either the heatmaps in Figure 5 more understandable. For me, the fold change and their origins are not clear not clear. In source data 4 there is no metabolite 25fold more abundant. So, I am not sure where this data is coming from.

One suggestion would also be to combine the Figure 5 and 6 in one figure, maybe similar to Figure 4. That would make it much easier for the reader to compare the drought and re-irrigation effects.

I like the Figure 1. I am just wondering if this would be better at the end of paper as summary. Also, I can't really see in the figure the differences between the different treatments etc. Also, the (5) is missing in the legend. Wouldn't it be good to show the changes, for instance when you say something is inhibited that the arrow is brocken etc? (Just some artistic suggestion).

In general, I also feel that the source data need to be better referenced in the text (e.g. source data 4 sheet E) Also it needs to be better explained. Source data 3, for instance, needs more explanation on what the two different fold changes columns are. Because if this is used in Figure 4, then I don't know where the data in the boxplot are coming from. When I am looking Figure 4 and see the change of re irrigated TOC. I don't see how this is significant, because the boxplots are within then quantile of the other one. It would help to see the source data.

Line 100ff this sentence comes very unexpected and may need some further explanation.

Line 111: That is introduced a bit to far away. Can you test the microbial priming in your experiment?

Line 124-129 I like the idea of having this results summary at the beginning of the result section.

Line 141: I would put this biomass data in a table or supplemental data. Later in the discussion it was very difficult to find again the data in the text.

Line 152: How was the relative abundance calculated?

Line 125/ Figure 2: The 10fold change would be easier to see with log scale, in Figure 3 you use them.

Line160 – 162: Why is that? How can they be helping to overcome the drought when they are so decreasing in their abundance.

Figure 4 legend refers to abundance in Figure 1 (that is the overview should be Figure 2,3).

Line 202: "…Their exudation was enriched again" I can’t see this in Figure 4 B, this figure is about TOC not single metabolites.

Line 205 How do you know that exudates are mainly secondary metabolites? Where is this shown? How do you classify them? The same in Line 208f. Where does this statement come from? How do you know the percentage and what do you mean of total metabolites (detected, identified, extracted)?

Line 212 What is a feature? That was not explained before. I know it, but do other readers know too?

Line 215ff Where is the data for the chemical richness? It is not in Figure s4B also chemical richness is never explained, it is also not in Source Data 5.

Line 221f there is no figure S7C.

Line 325f those are the wrong figures cited. Again, how do you know that mostly secondary metabolites were exuded. I don't see the data to support this.

Line 336 No, always the inoculated have higher weights. Line 141ff.

Line 354 Debatable with the data and scale shown, if both Figures would be in log scale. Also, the raw data would be nice.

Line 401 where did you identify a salicylic acid

Line 669 the MS parameters for the polar analysis are missing

Reviewer #2 (Recommendations for the authors):

Trees can exchange root exudates for minerals with soil bacteria. In a pot experiment the authors show that this exchange was enhanced when water was withheld, suggesting that trees can recruit beneficial bacteria under drought conditions. The experiment factorially combined the irrigation vs. drought treatment with no inoculation vs. inoculation of pots with two bacterial species. There were strong effects of the bacterial inoculation, but it was not directly tested if they were stronger under drought or irrigation (this could be tested by the interaction term in 2-way anovas). A strength of this study is that the authors measured root exudates, bacterial abundance and concentrations of soil P and minerals during and after the drought. To which extent the results of this short-term experiment with saplings of one particular tree species and two particular bacterial species in reasonably large pots may be extrapolated would have to be tested with further experiments or comparative field studies.

As indicated above, there is one major issue that makes it difficult to judge this manuscript. This is that the main hypotheses concern a potential interaction between the two treatment factors drought and inoculation, yet the authors do not test this interaction statistically. It is not possible to draw conclusions from separate tests of inoculation under irrigation and drought because one cannot draw conclusions about the significance of an interaction from separate tests. Thus, if one test is just significant and the other just not significant, the interaction may be very far from significant.

It should be very easy for the authors to use full general linear models for ALL their measurements (no need to use different analyses for different measurements except for the PCA type of analyses). These can go beyond the mentioned factorial 2-way anova with interaction and also include e.g. TOC as covariate for the analysis of Figure 4, fitting the covariate both as main term and as interaction with the treatment terms and their interaction. One could even add some path-analytic model to test the causal hypotheses implicit in the interpretation of the results.

On issue that did not become clear to me was if the 6 saplings per treatment combination were individually planted in 6 separate boxes (pots) or if perhaps two or three were in one box. My concern stems from looking at Table 2, where pots with four tree species are mentioned, so I wondered why it was not mentioned for the main experiment if each sapling was planted separately into a box. If there were fewer boxes than saplings in the main experiment, then box would have to be added as a random term in the general linear models.

Besides this main concern I find the description of results and the discussion complicated and on the long side. I was also confused that often only three treatments seemed to be compared: irrigated, drought without inoculation and drought with inoculation. For example, in Figures5 and 6 one can see that results should have been available for irrigated without inoculation and irrigated with inoculation, but then these are not always shown (e.g. Figure 5A). When tests are mentioned of irrigation vs. drought it is not even clear if means across the inoculation treatments were used or not.

There are some language issues that the authors should resolve, as well as some minor glitches such as "Schleppi et al. 2020 missing in the reference list.

Reviewer #3 (Recommendations for the authors):

In "A dynamic rhizosphere interplay between tree roots and soil bacteria under drought", the authors present compelling evidence that trees exposed to drought can alter their rhizosphere environment by (1) altering their exudation profiles (amount and composition) and (2) promoting bacteria that can increase the availability of soil nutrients such as phosphorus. The authors used a nice combination of experiments – collecting and analyzing the root exudates from potted saplings of Mediterranean cypress exposed to drought, adding root-associated bacteria with fluorescent markers, and adding exudate compounds to bacterial cultures, and determining the plant physiological responses to bacterial additions and drought. The authors found that while root exudation was decreased by drought, select rhizosphere microbes that were likely promoted by the tree's root exudation profile, buffering the trees from the some of the nutritional consequences of the drought by allowing them to sustain levels of nutrient uptake.

Overall, the study was carried out carefully, and there are several novel aspects. First, while there have been some studies that have looked at how root exudation changes in response to environmental stress or the presence of select microbes, few have combined the two to look at the interactive effects of microbes and stress. Second, few studies have characterized the metabolites released by roots and even fewer have conducted assays to examine how exuded metabolites impact bacterial growth and the consequences of this for plant nutrition. The exudation part of the story alone would have been an excellent contribution to the literature and by making the connections between the root exudates and the bacterial activity, the authors have provided a nice blueprint for how future studies might explore the costs/benefits of plant-microbe interactions.

Nevertheless, there are several issues that if addressed, would strengthen the manuscript. These include:

Improved clarity in writing. There are several places where the writing is unclear and editing would be helpful. In the line-by-line comments, I have included many suggested edits, especially for the Introduction. I didn't provide these throughout the manuscript (in order to focus on other aspects of the manuscript) but I suggest the authors pay careful attention to this issue in all sections of the manuscript.

Revising tables and figures. Overall, several display items could be improved. In my view, Figure 1 does not offer much and fails to capture the factorial nature of the experiment. What about replacing this with a box and arrow diagram that shows the how the presence of microbes likely alters tree drought tolerance (sort of like a path analysis but with words)? For Figure 2, rather than showing the relative abundances (panel B), it would be helpful to see the absolute changes in the two bacterial taxa in C (as opposed to the combined "with bacteria" treatment). If you do decide to include this, you could still keep the figures simple by removing the time series from the figure are interesting, it's not a major part of the story so perhaps that could be moved to SI? For Figure 3, the same comment applies. For Figure 4, it's unclear how a correlation coefficient can be calculated from the data collected. If the exudation rates are generated from root systems that contain both bacterial taxa, how can you have different relationships? For Figure 8b, the treatments were a bit confusing since the color scheme is the same as Figure 4a but the treatments are different.

In Table 2, I assume K in the last column refers to potassium (it's not in the legend). Also, please use consistent units for the soil variables, as some are in mE per L whereas others are mg per kg soil. In the SI, Figure S3 would benefit from having the inoculation and watering treatments added as vertical lines or you could add the sharing and arrows used in S3 to S2 so that the two figures are consistent. Finally, readers are likely to find it easier to make comparisons across figures if the time series data are presented as days of the month (May, 21, June 16, etc.) be as opposed to DD/MM.

There are few points not covered in the manuscripts (or barely covered) that could strengthen the manuscript. First, there's little discussion of the differences between active and passive exudation. Some roots exude compounds merely because a concentration gradient exists between the root apoplast and the soil solution (passive) whereas some exudates are actively exuded. In the case of a drought-stressed root, how much of the exudation is merely of the roots physical integrity? A desiccated rot likely lacks physical integrity to keep exudates from leaking out and it seems like an investigation of how the roots themselves changed with the treatments (especially the drought) is warranted.

Additionally, the manuscript would benefit from more synthesis about what is known about root exudation under drought. There have been many studies of tree root exudation (either solution culture or pulse labeling) that have shown how and why exudation rates change under drought. A brief synthesis of these papers would be very helpful to the readers. Also, I encourage the authors to calculate the percentage of photo assimilate that was exuded (in all four treatments) so that comparisons can be made with other exudation studies.

Finally, while the readers will appreciate the known caveat that no mycorrhizal fungi were detected in the pots, some discussion of how mycorrhiza fungi could affect exudation and bacterial growth would be extremely useful. In forests, arbuscular mycorrhizal fungi (which associate with cypress) play a central role in P nutrition, tree drought tolerance and root-microbial interactions. So, it's worth speculating how some of the patterns and processes described in the study might be altered (even if all you can do is speculate) if the fungi had been present.

Line by line comments:

L1. You might consider using a title that highlights one of the studies main findings: "Altered root exudation during drought stimulates rhizosphere bacteria, with consequences for tree drought tolerance".

L19. "minerals" should be "mineral nutrients" or just "nutrients".

L21. I would be more careful here; exudates can affect nutrient availability directly (e.g.. solubilizing mineral-P) independent of their effects on bacteria.

L22. "Priming" typically refers to the accelerated decomposition of organic matter owing to C (or in some cases N) inputs. In the case of P, phosphates are esther-bonded and so the release of P from soil organic matter is not considered priming. See Figure 2 in Dijkstra et al. 2013 "Rhizosphere priming: a nutrient perspective" for a discussion of this.

L25. Readers may be confused by calling this a factorial design without more details. The two bacterial taxa were not included as individual treatments since they were combined into a "with bacteria" treatment.

L28. State percentage increase of exudation?

L30. Would be clearer to readers if you started a new sentence here. "In a second experiment, we added to bacterial cultures metabolites detected as root exudates, and found that xx percent stimulated bacterial growth".

L34. Could add (recruitment of beneficial bacteria) "especially under water stress".

L39. Maybe just start with "Climate change".

L41. Change to "…ecosystems worldwide, with negative impacts for forest health".

L43-44. Please list references for each impact (recruitment, mortality, etc.)

L47. A closing sentence is needed here.

L49. Replace with "including escape, avoidance and tolerance strategies".

L52. Replace with "Root systems mediate water and nutrient uptake…".

L55. Not sure about the relevance of mentioning fine root turnover here since it's not the focus of the paper.

L49-84. There are lots of redundancies in paragraphs two and three of the Introduction. The two might be better as a single paragraph about what is known about plant exudation responses to drought, and the impacts of these exudate effects on microbes.

L86-89. Is one of these sentences the topic sentence of the paragraph? Please try to make this clear so that the readers know what the paragraph is about from the outset.

L101. As noted above, priming refers to something beyond just root exudates stimulation in that it refers to the accelerated decay of soil organic matter. Some people refer to what you're describing as "microbial activation" which is a hypothesis of how the first step of priming occurs.

L106. "This increase in exudation occurred despite increases in photosynthesis throughout the dry season."

L111 and L118. Replace "priming" with another term please. "exudate-induced microbial growth".

L124-129. This "results abstract" does not see necessary unless the journal specifically requires it.

L140. In the droughted trees in Figure S2, the bacteria treatment seems to be reducing assimilation rates relative to the no bacteria treatment. What might be causing this? This finding seems to be at odds with the other data you present that shows that the presence of bacteria in the drought treatment help trees maintain their nutritional status. Is there some tradeoff that night be occurring which would cause assimilation to go down?

L141. Replace "sapling" with "saplings".

L142-144. Please present a p value for this contrast of a statistical test was conducted.

L146-162. Were statistical tests conducted? If yes, please present the results. If not, please explain why?

L171. It would be helpful if you could use this format – presenting the % difference in exudation – for the other treatments.

L174. How is a correlation coefficient calculated when you have one root system (which are used to generate exudation data) but both types of bacteria? Please explain what variables are being used.

L186. Figure 5B?

L202. Is this meant to be Figure 5B (and not 4B)?

L290. Replace "C source" with "C uptake" or "C assimilation"?

L298. But lower assimilation in the bacteria-drought treatment (as least based on Figure S2).

L311. Replace "sustain" with "remain"?

L365. The was these references are used in this sentence makes it seem like the studies were of tree roots (but they were not).

L486-489. Was any data collected on the changes in PAR over the course of the experiment? Were trees in full sun?

L633. Do you have data on what C contamination existed in the non-root controls? Was this subtracted from the values of the ones with roots (i.e. were the fluxes corrected for background interferences)? Some info in the challenges of getting a C-free root chamber would be helpful to readers who want to repeat these methods.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "A dynamic rhizosphere interplay between tree roots and soil bacteria under drought stress" for further consideration by eLife. Your revised article has been evaluated by Meredith Schuman (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Essential revisions:

(1) eLife has a consultative review procedure. Subsequent to the submission of both reviews, we discussed Figure 2 due to the Editor's concerns. This figure presents data which are very important for interpretation of the authors' proposed mechanism. In the current form of the manuscript the interpretation of these data is misleading.

(1a) These data show that drought-stressed trees release less TOC (exudate) from roots than irrigated trees, and that both drought-stressed and irrigated trees release more TOC when inoculated. In contrast, trees re-irrigated after drought produce similar TOC from roots whether they are inoculated or not (although the figure indicates there is also a significant difference there, which is confusing and perhaps incorrectly indicated). Thus, the inference is rather that inoculated plants produce more TOC from roots, except if they have been subjected to drought and re-irrigation. What the authors imply is that drought-stressed plants produce more exudates when inoculated, which is correct but misleading since irrigated plants do as well. It is rather interesting that re-irrigated plants do not do this. Reviewer 1 commented additionally: "This contrasts with the answer of the authors to my main question: 'Thus, only under drought conditions were trees responsive to inoculation, showing a ~50% increase in root exudation (Figure 2; P = 0.039 for the interaction irrigation:bacteria).' I really cannot see this interaction in Figure 2, and if anything, it seems to go into the wrong direction (i.e. difference with-without bacterial is larger under irrigation)."

(1b) The exudate production has no predictable correlation with abundance of either of the two bacterial strains monitored by the authors. The reason for choosing these strains is now clear in the text, but there is no indication that either responds to differences in root exudates. Thus, although the growth of both strains is affected by (some of) the exudate metabolites which differ (mildly – less than 2x) in abundance between treatments, there is no evidence that these two strains are directly associated with exudate differences. The inoculation affects total TOC from roots, but the effect could be indirect in the ecologically relevant, but complex soil environment of the experiment.

(1c) There are no data to indicate that these bacterial strains are responsible for the nutrient effects observed for inoculated plants undergoing drought, especially as the soil for the experiment was not sterilized. The unsterilized soil is in fact a nice aspect of the design from an ecological perspective, but also makes it more complicated to determine the mechanism by which additional inoculation has its effects.

2) Reviewer 1 emphasized that the writing still needs substantial correction and streamlining.

The full reviews are provided below. Please note that the topics under (1) arose after the reviews were submitted and during the consultation.

Reviewer #1 (Recommendations for the authors):

This manuscript has been improved during revision, including improved statistical analysis. In my view the Discussion would still benefit from a better focus on the main results.

I mentioned in the previous review that parts of the paper were quite long and complicated to read and still find this to be true (see e.g. Discussion lines 413-443; overall, the Discussion has become even longer than it was before). I also mentioned that there were many small language errors and again can still find them in this "revised" version. It is really the authors responsibility to correct the language because they are the "publisher".

https://doi.org/10.7554/eLife.79679.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Comments to the Authors:

Specifically, important strengths of this study are its attempt to investigate root exudate composition while manipulating native root-microbe interactions and drought stress, in order to draw conclusions about ecological roles of root exudates in relevant scenarios, while furthermore using methods originating in different disciplines for the appropriate manipulation and assessment of experimental factors. The topic is important and many aspects of the experimental design well-chosen, especially considering ecological relevance of the system and the potential power of factorial design, and such a study could be of interest for eLife. Measuring metabolites released from roots and connecting them to their putative functions is extremely difficult, even more so when done with trees and actual soil. Yet all reviewers identified substantial flaws in the analysis and presentation of the results, and independently of these concerns, it is difficult to connect the various results to the authors' stated research questions. Thus, the analysis is inconclusive, and it is generally not clear what we learn out of this study regarding trees under drought stress engaged in interactions with soil bacteria. This is reflected in the divergent focus of the individual reviews, each of which should provide very helpful suggestions for addressing these concerns. In addition to the points raised by the reviewers (below), I would note the following:

1) It is not clear how the bacteria were chosen, except that these are two native soil strains, which is commendable – but are they among the most common, most frequently found associated with the study plant, …?

We thank you for this insightful comment. We now add information in the

Methods to clarify our choice: “In multiple screens of rhizosphere bacteria in our forest site, a total of 54 bacterial strains were identified on plates. Of which, B. subtilis and P. stutzeri were the most consistent, evidencing their important role in our system”.

In addition, we modified the text with the first mention of the bacteria in the Results: “Fluorescently tagged B. subtilis and P. stutzeri, that were modified from native strains isolated from the forest soil, showed attachment and dispersed colonization along Cupressus fine roots, regardless of irrigation, on days 1 and 3 following inoculation (Figure 2A, Source Data 2)”.

2) Metabolite naming is inconsistent, e.g. Figure 7 refers to 3,4-DHB which I think is in fact 3,4-DHM, a supplementary figure refers to nicotine which I think is nicotinate.

Thank you so much for catching this error with 3,4-DHM, which we have now corrected. Regarding nicotinate, we added a clarification in the Methods: “Metabolites were identical with those identified in root exudates (Table 1), except for nicotinate, which was tested in the form of nicotine”.

3) Heat map colors are confusing, indicating up-regulation (red) and down-regulation (blue) when in fact everything is expressed as a positive fold change, which would be best shown in darker (highest) to lighter (lowest) colors.

We understand the Editor’s point and hence tried a single-tone heat map, which, to us, was disappointing, since differences among metabolites and among treatments were hardly visible. Therefore, we would like to keep our heat map colors. It is very common to use contrasting colors for different levels of metabolites; the heat map shows the changes of the metabolite levels relative to the beginning of the experiment.

4) The first results show that, while roots are recruiting, the experiment is carried out with an unstable microbial community, which makes it more challenging to interpret the results (stability in the experimental microcosms is achieved only for the re-inoculated irrigation treatment).

We appreciate the Editor’s comment. We are aware that the natural populations of bacteria vary according to the different treatments, but we wanted to focus on the bacteria we added to the soil. Our conclusions from the experiment relate to Bacillus subtilis and Pseudomonas stutzeri addition to the soil, so we also tested the specific metabolites on the growth of these bacteria under laboratory conditions. The study aimed to show the dynamic changes between these specific bacteria and the cypress roots under dry conditions. The dynamics in the rhizosphere and soil during the experiment were affected by both the native microbial community and bacteria inoculation. We did not follow the general microbial community in the rhizosphere and soil in this experiment. Instead, we monitored only the inoculation bacteria by isolating them from the soil by a selective medium and counting them. We point to the dynamics of these two bacteria and not to the whole microbial community. To address this comment, we added text in the “study limitations” section of the Discussion: “Second, manipulating the rhizosphere with inoculations can also be criticized as creating disturbance. Moreover, except for the re-inoculated irrigated trees, bacterial communities were not stable. Still, the two bacteria species were released into the forest soil to compete as any other microbial species in the soil system”. In addition, a sentence was added in the Results to clarify our approach: “The inoculations can be regarded as pulses, allowing us to test the rhizosphere interactions, while declining with time”.

(5) It is misleading to refer e.g. to "drought-specific" root exudate metabolites, as Figure 5 shows that all exudate metabolites are detected across all treatments even if they may have different patterns in each treatment (hard to tell from the color scheme in Figure 5 – difference between slightly darker and slightly lighter yellow hard to assess).

Following the Editor’s comment, we changed the term "drought-specific" to “drought increased” in.

Reviewer #1 (Recommendations for the authors):

Oppenheimer-Shaanan et al. investigated the how inoculation with root bacteria can changes the ability of the Mediterranean cypress to cope with drought stress. Therefore, they compared root exudation (as TOC) of tree saplings under drought stress and under constant irrigation. They show that root exudation increases when trees are inoculated with root bacteria. Inoculated trees also showed higher capabilities to recover from drought stress.

The combination of the different methodologies provides an interesting view, or glimpse as the authors wrote, on the interaction between tree roots and associated bacteria under drought stress. The use of the rhizoboxes , the collection of root exudates and the analysis of the bacteria is very nice experimental set up. It allows to see the different effect of abiotic stress on the different players in the interaction (soil, roots, bacteria, leaves) and to make connections on how they are interacting.

However, I see here some shortcomings in the analysis of the data and the presentation. Especially, the metabolomics data are not easy to understand and could be much better presented. Furthermore, I think that the emphasis that is put on the fold changes of certain metabolites is too high. From this are conclusions drawn that are not well supported by the data and makes it hard to follow later the discussion.

Condensation and maybe a different statistical analysis in some figures in the metabolomics analysis would help to better visualize the changes and to see clearer the effects of bacteria and drought.

We thank reviewer #1for taking the time to thoroughly review our work. We have rigorously revised the manuscript following the valuable suggestions of the reviewer. This includes; Figures have been merged and moved to streamline the narrative; The text was shortened in the Introduction and Results; Instability of the microbial communities was discussed under the study limitations; Information was added concerning the saplings’ biomass; PCA of the polar metabolite profiles of exudates from drought-exposed saplings with and without bacterial inoculations was replaced with PCoA. Please see below detailed responses.

I don't feel qualified to review the bacterial stain isolation, characterization and construction.

I am not used so much to the use of those log2 fold changes. That is why I need to do some calculations. If I did those right most of the log2FC in Table 1 are very low there and all smaller than 0.5/-0.5 that means a fold change of less than 1.5 I am wondering if those will have a physiological effect. You than showed the effect of some of the metabolites like in FIGURE 6, where you are testing p- coumaric acid for bacterial growth. In the source data the fold change for this metabolite is either 1.09 or 1.15 only (if I calculate back from your log2 value). I think that is very low and I have my doubts that you can draw some conclusion from that. I acknowledge the effects that you see in bacteria assays, and I don't doubt that. But I just think that the difference in concentrations would not matter. How would this look in a comparison where you for instance test different concentrations of e.g. p-coumaric acid and see if the drought or irrigated concentrations have different effects.

We thank reviewer #1 for taking the time to thoroughly review our work. First, we note that these are the log 2 fold changes from the pairwise comparison between two treatments, not the changes from the control baseline, which were much higher. A reference to our approach can be found in Warren (2016). Second, fold changes between the treatments might be relatively small, yet significant between treatments (Table 1). Third, small molecules tend to be found at low concentrations in root exudates and may be rapidly consumed by rhizosphere microbes, which affect their recovery and analysis. Here we showed growth curves of the bacteria where very low concentrations of these molecules had physiological effects. Indeed, a small difference in concentration can have a large effect, as observed e.g. in Williams et al. (2011).

Warren, C. R. (2016). Simultaneous efflux and uptake of metabolites by roots of wheat. Plant and Soil, 406(1), 359-374.

Williams, A., Langridge, H., Straathof, A. L., Fox, G., Muhammadali, H., Hollywood, K. A., … and de Vries, F. T. (2021). Comparing root exudate collection techniques: An improved hybrid method. Soil Biology and Biochemistry, 161, 108391.

I have also some difficulties with the interpretation and analysis of the metabolomics data. The PCA is a good way of visualizing the differences between treatment groups, but I think a PCoA would be better here. It is getting more and more used in metabolomics research and has to advantage to deal better with non-normal distributed data, which you will most likely have. In combination with the PCoA a PERMANOVA might help you to show effects of the treatment instead of the visual comparison. That would also make either the heatmaps in Figure 5 more understandable. For me, the fold change and their origins are not clear not clear. In source data 4 there is no metabolite 25fold more abundant. So, I am not sure where this data is coming from.

We thank the reviewer for this suggestion. To address this reviewer’s point, we performed PCoA and PERMANOVA and found similar trends to those observed by the PCA. These changes are reflected in the new Figure 3A and in the Results text: “Principal coordinate analysis (PCoA) showed that root exudates blends (samples) from drought trees that were inoculated and those from drought trees that were not inoculated partitioned into distinct clusters, with partial overlap (Figure 3A). When samples from irrigated trees were also considered in the PCoA, they behaved similarly, however there was no single sample common to all three groups (Figure 3A)”.

One suggestion would also be to combine the Figure 5 and 6 in one figure, maybe similar to Figure 4. That would make it much easier for the reader to compare the drought and re-irrigation effects.

We thank the reviewer for this suggestion, which we adopted.

I like the Figure 1. I am just wondering if this would be better at the end of paper as summary. Also, I can't really see in the figure the differences between the different treatments etc. Also, the (5) is missing in the legend. Wouldn't it be good to show the changes, for instance when you say something is inhibited that the arrow is brocken etc? (Just some artistic suggestion).

Thank you for this suggestion. We have restructured the figure and moved it to the end of the manuscript. We also added the reference to step (5).

In general, I also feel that the source data need to be better referenced in the text (e.g. source data 4 sheet E) Also it needs to be better explained. Source data 3, for instance, needs more explanation on what the two different fold changes columns are. Because if this is used in Figure 4, then I don't know where the data in the boxplot are coming from. When I am looking Figure 4 and see the change of re irrigated TOC. I don't see how this is significant, because the boxplots are within then quantile of the other one. It would help to see the source data.

We have addressed this point by providing further data of TOC in Source Data 3 and by providing a new supplementary file with raw data of TOC and assimilation rates. The analyses of the total amount of carbon were done compared to the zero time of the experiment. At that time, samples were also taken, and all the experimental groups were under similar conditions. There are slight differences between the individual trees in the experiment from the beginning. Therefore, the measurements describe changes between the start point and a subsequent measurements (fold changes) for each individual tree. Only then were statistical analyses done.

The significant differences are compared between two groups and plotted on the graph were examined in statistical tests that are in the Source Data 3. We claim for significant differences between the groups that do not overlap for example between irrigated trees with and without bacteria.

Line 100ff this sentence comes very unexpected and may need some further explanation.

We add explanation: “In the forest soil, root exudation is suspected to enhance rhizobacteria, in turn leading to increased decomposition of soil organic matter, i.e., increased C mineralization, a process termed microbial priming (Schleppi et al., 2019). Here, we use this term on a wide perspective, even that P release from soil phosphates, for example, is not strictly considered as priming (Dijkstra et al. 2013)”.

Line 111: That is introduced a bit too far away. Can you test the microbial priming in your experiment?

We clarified this sentence and toned it down: “Our overall objective was to test whether microbial recruitment by tree root exudates (a form of microbial priming) is beneficial to trees under drought, an abiotic stress that alters tree carbon and nutrient allocation”.

Line 141: I would put this biomass data in a table or supplemental data. Later in the discussion it was very difficult to find again the data in the text.

Biomass data were added in Source Data 1, and the text was modified: “During the experiment, sapling biomass increment was 188.6±15.6 g and 149.1±14.4 g for irrigated saplings with and without bacterial inoculations, respectively (difference not significant; P = 0.753; Source Data 1). The effect of drought was highly significant (P < 0.001), and biomass increment in drought-exposed saplings with and without bacterial inoculations was 22.4±5.7 g and 4.5±4.6 g, respectively (P = 0.09)”.

Line 152: How was the relative abundance calculated?

This information belongs to the Methods: “The relative abundance was estimated by counting each strain of bacteria separately and diving by the total amount of bacteria.”

Line 125/ Figure 2: the 10fold change would be easier to see with log scale, in Figure 3 you use them.

We modified Figure 2.

Line160 – 162: Why is that? How can they be helping to overcome the drought when they are so decreasing in their abundance.

Bacteria in the rhizosphere are even more critical during the drought, even at small amounts. Our study shows this, in agreement with earlier works (Shakya et al., 2013; Taketani et al., 2016, Yang et al., 2021). We show that bacteria adapted to dry conditions (B. subtilis) can improve tree nutrition to higher extent. This claim was also discussed in the Discussion. To address the comment about the unstable bacterial communities, we added text in the “study limitations” section of the Discussion: “Second, manipulating the rhizosphere with inoculations can also be criticized as creating disturbance. Moreover, except for the reinoculated irrigated trees, bacterial communities were not stable. Still, the two bacteria species were released into the forest soil to compete as any other microbial species in the soil system”. In addition, a sentence was added in the Results to clarify our approach: “The inoculations can be regarded as pulses, allowing us to test the rhizosphere interactions, while declining with time”.

Yang, N., Nesme, J., Røder, H.L. et al. Emergent bacterial community properties induce enhanced drought tolerance in Arabidopsis. npj Biofilms Microbiomes 7, 82 (2021).

Figure 4 legend refers to abundance in Figure 1 (that is the overview should be Figure 2,3).

Thank you so much for catching these confusing errors, which we have now corrected.

Line 202: "…Their exudation was enriched again" I can’t see this in Figure 4 B, this figure is about TOC not single metabolites.

Thank you for catching these errors, which we have now corrected. The reference was changed to Figure 3B (the old Figure 6).

Line 205 How do you know that exudates are mainly secondary metabolites? Where is this shown? How do you classify them? The same in Line 208f. Where does this statement come from? How do you know the percentage and what do you mean of total metabolites (detected, identified, extracted)?

Point well taken. We did chemical classification analysis which can found in Source Data 4E, which was now referenced in this section. This is similar with earlier studies like e.g. Gargallo-Garriga et al. (2018). The percentages relate to the identified metabolites with LC-MS that generated a diverse profile of secondary metabolites.

Gargallo-Garriga, A., Preece, C., Sardans, J., Oravec, M., Urban, O., and Peñuelas, J. (2018). Root exudate metabolomes change under drought and show limited capacity for recovery. Scientific reports, 8(1), 1-15.

Line 212 What is a feature? That was not explained before. I know it, but do other readers know too?

Following the reviewer’s comment we changed the text to: “clustered mass signals” to clarify the sentence.

Line 215ff Where is the data for the chemical richness? It is not in Figure s4B also chemical richness is never explained, it is also not in Source Data 5.

The sentence was removed.

Line 221f there is no figure S7C.

Line 325f those are the wrong figures cited. Again, how do you know that mostly secondary metabolites were exuded. I don't see the data to support this.

Thank you so much for catching these errors, which we have now corrected.

Line 336 No, always the inoculated have higher weights. Line 141ff.

Thank you, this has now been corrected: “Finally, desiccated trees supplemented with soil bacteria showed better nutrition in P and Fe (Figure 5), and higher biomass (although not significantly); but, unexpectedly, did not recover faster than without bacteria. Thus, photosynthesis recovered faster in drought trees that were not inoculated than those that were (Figure S3)”.

Line 354 Debatable with the data and scale shown, if both Figures would be in log scale. Also, the raw data would be nice.

Following the reviewer’s comment, we are adding another supplementary file with statistical analysis of root exudation rates and leaf assimilation rates across all trees along the experiment. This analysis is now cited at the respective sections in the Discussion: “root exudation increased following inoculation, and was higher during drought than following it (Source Data 3, 7)”.

Line 401 where did you identify a salicylic acid

We identified 4-AMINOSALICYLIC ACID (See Source Data 4A) as one of the metabolites but not significant to bacteria inoculation.

Line 669 the MS parameters for the polar analysis are missing

Thank you so much for catching these missing. We added the missing details: “Briefly, analysis was performed using Acquity I class UPLC System combined with mass spectrometer (Thermo Exactive Plus Orbitrap; Waltham, MA, USA). The mass spectrometer was operated under the following parameters: Full MS/ dd-MS2 mode (1 μscans) at 35,000 resolution from 75-1050 m/z, with 3.25 kV spray voltage, 40 sheath gas, 10 auxiliary gas and negative ionization mode”.

Reviewer #2 (Recommendations for the authors):

Trees can exchange root exudates for minerals with soil bacteria. In a pot experiment the authors show that this exchange was enhanced when water was withheld, suggesting that trees can recruit beneficial bacteria under drought conditions. The experiment factorially combined the irrigation vs. drought treatment with no inoculation vs. inoculation of pots with two bacterial species. There were strong effects of the bacterial inoculation, but it was not directly tested if they were stronger under drought or irrigation (this could be tested by the interaction term in 2-way anovas). A strength of this study is that the authors measured root exudates, bacterial abundance and concentrations of soil P and minerals during and after the drought. To which extent the results of this short-term experiment with saplings of one particular tree species and two particular bacterial species in reasonably large pots may be extrapolated would have to be tested with further experiments or comparative field studies.

As indicated above, there is one major issue that makes it difficult to judge this manuscript. This is that the main hypotheses concern a potential interaction between the two treatment factors drought and inoculation, yet the authors do not test this interaction statistically. It is not possible to draw conclusions from separate tests of inoculation under irrigation and drought because one cannot draw conclusions about the significance of an interaction from separate tests. Thus, if one test is just significant and the other just not significant, the interaction may be very far from significant.

It should be very easy for the authors to use full general linear models for ALL their measurements (no need to use different analyses for different measurements except for the PCA type of analyses). These can go beyond the mentioned factorial 2-way anova with interaction and also include e.g. TOC as covariate for the analysis of Figure 4, fitting the covariate both as main term and as interaction with the treatment terms and their interaction. One could even add some path-analytic model to test the causal hypotheses implicit in the interpretation of the results.

We thank the reviewer for the enthusiastic feedback. Following up on the reviewer’s suggestion, two-way ANOVA was analyzed to check the different interactions

between the treatments (indicated in the table), which is indicated in the text. We also changed the one-way ANOVA of metabolomes values to two-way ANOVA. We re-analyzed the treatments’ effects on the physiological measurements, bacterial counts, TOC, metabolites (polar and semi polar) and the amounts of leaf elements. Please see all the results in source data files. This issue is now highlighted throughout the Results chapters: “Bacterial inoculations had no effect on leaf gas exchange, and hence the interaction irrigation:bacteria was not significant either”. “Thus, only under drought conditions were trees responsive to inoculation, showing a ~50% increase in root exudation (Figure 2; P = 0.039 for the interaction irrigation:bacteria)”. “In two metabolites, the interaction irrigation:bacteria was significant under drought, but not under re-irrigation (Source Data 4)”. The Methods section was updated too. In addition, following the reviewer’s proposal, a general linear model was tested concerning one of the physiological parameters, namely leaf assimilation, where there was a sharp difference between drought treatment and irrigation, and the exudation rate (total organic carbon). Please see Source Data 7. The differences were similar to those from the analysis done without the linear model. The linear model supported the correlation and indicated the strength of a causal hypothesis, but did not prove the direction of causation. We were unable to use TOC as an independent (predictor) in a linear model. The TOC was dependent (criterion) on drought/irrigation and bacteria inoculation treatments, and these interactions were examined. Finally, following a suggestion by another reviewer, we changed the PCA graphs with PCoA, which resulted with similar results.

On issue that did not become clear to me was if the 6 saplings per treatment combination were individually planted in 6 separate boxes (pots) or if perhaps two or three were in one box. My concern stems from looking at Table 2, where pots with four tree species are mentioned, so I wondered why it was not mentioned for the main experiment if each sapling was planted separately into a box. If there were fewer boxes than saplings in the main experiment, then box would have to be added as a random term in the general linear models.

Thank you, we found your comment helpful and have revised accordingly. First, the erroneous title of Table 2 was corrected: “Soil properties from pots where saplings were grown under treatments of drought or irrigated with or without bacteria”. Second, our experiment was conducted with each sapling being planted in an individual rhizobox. So, we added this information in the Methods text: “The saplings were kept at the net-house, one sapling per rhizobox, under optimal irrigation of 1.2 L day-1 for a seven months period, to allow root growth into mixed the forest soil”.

Besides this main concern I find the description of results and the discussion complicated and on the long side. I was also confused that often only three treatments seemed to be compared: irrigated, drought without inoculation and drought with inoculation. For example, in Figures5 and 6 one can see that results should have been available for irrigated without inoculation and irrigated with inoculation, but then these are not always shown (e.g. Figure 5A). When tests are mentioned of irrigation vs. drought it is not even clear if means across the inoculation treatments were used or not.

We apologize for the length and complexity of our presentation and have tried to improve in multiple places along the text and figures. The opening paragraph of the Results was completely removed. We also condensed and merged paragraphs 2 and 3 of the Introduction into a single paragraph. We were able to better streamline the Results and Discussion by moving Figure 1 to the end; merging Figures 2 and 3; and merging Figures 5 and 6. In the new Figure 3, we compare only the treatments showing significant differences in the total organic carbon.

We add more explanation in to clarify the treatments which were tested. For example, Figure 3: (A) and (B) Metabolic profiles at drought period, (C) at re-Irrigation period. In addition, clarifications were added in text, e.g. in Results: “The amount of organic carbon exuded from roots of drought-exposed trees was significantly, ~50% lower than from roots of irrigated trees, and moreover, ~70% lower, in the re-irrigation period (across inoculation treatments; P < 0.01; Figure 2, Source Data 3, Supplementary Information)”.

There are some language issues that the authors should resolve, as well as some minor glitches such as "Schleppi et al. 2020 missing in the reference list.

English errors were corrected throughout the text. The missing reference was added.

Reviewer #3 (Recommendations for the authors):

In "A dynamic rhizosphere interplay between tree roots and soil bacteria under drought", the authors present compelling evidence that trees exposed to drought can alter their rhizosphere environment by (1) altering their exudation profiles (amount and composition) and (2) promoting bacteria that can increase the availability of soil nutrients such as phosphorus. The authors used a nice combination of experiments – collecting and analyzing the root exudates from potted saplings of Mediterranean cypress exposed to drought, adding root-associated bacteria with fluorescent markers, and adding exudate compounds to bacterial cultures, and determining the plant physiological responses to bacterial additions and drought. The authors found that while root exudation was decreased by drought, select rhizosphere microbes that were likely promoted by the tree's root exudation profile, buffering the trees from the some of the nutritional consequences of the drought by allowing them to sustain levels of nutrient uptake.

Overall, the study was carried out carefully, and there are several novel aspects. First, while there have been some studies that have looked at how root exudation changes in response to environmental stress or the presence of select microbes, few have combined the two to look at the interactive effects of microbes and stress. Second, few studies have characterized the metabolites released by roots and even fewer have conducted assays to examine how exuded metabolites impact bacterial growth and the consequences of this for plant nutrition. The exudation part of the story alone would have been an excellent contribution to the literature and by making the connections between the root exudates and the bacterial activity, the authors have provided a nice blueprint for how future studies might explore the costs/benefits of plant-microbe interactions.

Nevertheless, there are several issues that if addressed, would strengthen the manuscript. These include:

Improved clarity in writing. There are several places where the writing is unclear, and editing would be helpful. In the line-by-line comments, I have included many suggested edits, especially for the Introduction. I didn't provide these throughout the manuscript (in order to focus on other aspects of the manuscript) but I suggest the authors pay careful attention to this issue in all sections of the manuscript.

Revising tables and figures. Overall, several display items could be improved. In my view, Figure 1 does not offer much and fails to capture the factorial nature of the experiment. What about replacing this with a box and arrow diagram that shows the how the presence of microbes likely alters tree drought tolerance (sort of like a path analysis but with words)? For Figure 2, rather than showing the relative abundances (panel B), it would be helpful to see the absolute changes in the two bacterial taxa in C (as opposed to the combined "with bacteria" treatment). If you do decide to include this, you could still keep the figures simple by removing the time series from the figure are interesting, it's not a major part of the story so perhaps that could be moved to SI? For Figure 3, the same comment applies. For Figure 4, it's unclear how a correlation coefficient can be calculated from the data collected. If the exudation rates are generated from root systems that contain both bacterial taxa, how can you have different relationships?

We thank reviewer #3 for his thorough review. We are glad that he found our results interesting and novel, and we thank him for the many insightful comments that contribute to improving our manuscript. Following up on the above suggestions, Figure 1 was moved to the end of the figures (now Figure 7), where it better fits the overview of our experiment. We have also modified it slightly to improve clarity, yet we are hesitant to present a box and arrow diagram, which might look too deterministic, considering that our results are coming from a pot experiment on saplings. Figures 2 and 3 (now Figure 1) were modified as suggested, with absolute changes in the two bacterial taxa. Regarding Figure 4 (now Figure 2), indeed we had a single exudate mean per treatment and time, and two bacterial abundance values, one for each of the two species. While interactions between the populations of the two strains may exist, it is still possible to correlate between each population and the exudate rates. In addition, Figures 5 and 6 were merged into the new Figure 3.

For Figure 8b, the treatments were a bit confusing since the color scheme is the same as Figure 4a but the treatments are different.

We now add more details in the caption for clarity: “Leaf elements (A) and soil phosphorous (B) of irrigated (blue shades) and drought-exposed and re-irrigated (brown shades) Cupressus sempervirens saplings, with and without bacterial inoculations”.

In Table 2, I assume K in the last column refers to potassium (it's not in the legend). Also, please use consistent units for the soil variables, as some are in mE per L whereas others are mg per kg soil.

Thank you for this comment. We modified the Table caption and harmonized the units as suggested, to clarify this point.

In the SI, Figure S3 would benefit from having the inoculation and watering treatments added as vertical lines or you could add the sharing and arrows used in S3 to S2 so that the two figures are consistent.

We added to Figure S3 the same lines as in Figure S2. The caption was updated accordingly.

Finally, readers are likely to find it easier to make comparisons across figures if the time series data are presented as days of the month (May, 21, June 16, etc.) be as opposed to DD/MM.

There are few points not covered in the manuscripts (or barely covered) that could strengthen the manuscript. First, there's little discussion of the differences between active and passive exudation. Some roots exude compounds merely because a concentration gradient exists between the root apoplast and the soil solution (passive) whereas some exudates are actively exuded. In the case of a drought-stressed root, how much of the exudation is merely of the roots physical integrity? A desiccated rot likely lacks physical integrity to keep exudates from leaking out and it seems like an investigation of how the roots themselves changed with the treatments (especially the drought) is warranted.

This is an interesting point that deserves more discussion. Here, we were able to include it in new text that was introduced to the section on study limitations: “passive exudation and leakage from drought-stressed roots cannot be ruled out in our experiment.

However, the observed shift in metabolites and the sensitivity to inoculations support rather an active exudation”.

Additionally, the manuscript would benefit from more synthesis about what is known about root exudation under drought. There have been many studies of tree root exudation (either solution culture or pulse labeling) that have shown how and why exudation rates change under drought. A brief synthesis of these papers would be very helpful to the readers.

Point well taken. Per the reviewer’s comment we improved the focus of our relevant Introduction section, and added a key reference on tree root exudation changes under drought: “The majority of root exudates typically consists of primary metabolites (20%; sugars, amino acids, and organic acids) and 15% of nitrogen as well as secondary metabolites, complex polymers, such as flavonoids, glucosinolates, auxins, etc. (Vives-Peris et al., 2020). Those plant-derived metabolites were shown to shape microbial communities by allowing bacteria to metabolize them and then establish themselves in the rhizosphere (Venturi, 2016; Sasse et al., 2017). Although root exudation is ubiquitous among tree species, the amount and composition of root exudates vary. So far, little information is available on how drought influences tree root exudates, their chemical composition, and how root metabolism is connected with shifts in root-associated microbiome composition (Zhang et al., 2007; Tückmant el al., 2017; Naylor and Coleman-Derr 2018). Recently, it was shown that oak trees (Quercus ilex) under drought shift their exudates from primary to secondary metabolites (Gargallo-Garriga et al. 2018)”.

We also used this information in the Discussion: “Root exudates included carbohydrates and organic acids that fed bacterial communities (Figures 3, 4). Yet, unexpectedly, most root exudates under drought were secondary, rather than primary metabolites. However, this is in agreement with a similar shift observed in the exudate metabolomes of drought-exposed oak trees (Gargallo-Garriga et al. 2018). Indeed, phenolic acid compounds and amino acid derivatives proved to be superior carbon and nitrogen sources than a sugar and amino acid”.

Gargallo-Garriga, A., Preece, C., Sardans, J., Oravec, M., Urban, O., and Peñuelas, J. (2018). Root exudate metabolomes change under drought and show limited capacity for recovery. Scientific reports, 8(1), 1-15.

Also, I encourage the authors to calculate the percentage of photo assimilate that was exuded (in all four treatments) so that comparisons can be made with other exudation studies.

Good point. While true percentages would require a full carbon balance at the whole sapling scale (which was not done here), ratios could be readily calculated. Text was added in the Results section on root exudates: “An additional measure of tree carbon allocation into root exudation is the ratio between exudation rate (µg C mg root-1 day-1) and net assimilation rate (µmol C m-2 leaf s-1). This ratio was 0.40-0.45 under drought, decreasing to 0.27-0.35 under re-irrigation (across inoculation treatments). In irrigated saplings, ratios increased from 0.12-0.17 without bacteria, to 0.34 and 0.58 in inoculated saplings (in later and earlier phases of the experiment, respectively)”.

Finally, while the readers will appreciate the known caveat that no mycorrhizal fungi were detected in the pots, some discussion of how mycorrhiza fungi could affect exudation and bacterial growth would be extremely useful. In forests, arbuscular mycorrhizal fungi (which associate with cypress) play a central role in P nutrition, tree drought tolerance and root-microbial interactions. So it's worth speculating how some of the patterns and processes described in the study might be altered (even if all you can do is speculate) if the fungi had been present.

We thank for the reviewer for this important point. To account for this point, text was added under study limitations: “In the forest, mycorrhizal fungi, which were missing here, can have large effects too (Meier et al. 2013). Specifically, Cupressus sempervirens hosts a diversity of active arbuscular mycorrhizal species (Avital et al. 2022). While the latter generally help in tree P nutrition, their activity is mostly reduced under drought. Future studies should test the functions of the three-kingdom interactions among trees, fungi and bacteria”.

Meier, I. C., Avis, P. G., and Phillips, R. P. (2013). Fungal communities influence root exudation rates in pine seedlings. FEMS microbiology ecology, 83(3), 585-595. Avital S, Rog I, Livne-Luzon S, Cahanovitc R, Klein T (2022) Asymmetric belowground carbon transfer in a diverse tree community. Molecular Ecology.

Rog, I., Tague, C., Jakoby, G., Megidish, S., Yaakobi, A., Wagner, Y., and Klein, T. (2021).

Interspecific soil water partitioning as a driver of increased productivity in a diverse mixed Mediterranean forest. Journal of Geophysical Research: Biogeosciences, 126(9), e2021JG006382.

Line by line comments:

L1. You might consider using a title that highlights one of the studies main findings: "Altered root exudation during drought stimulates rhizosphere bacteria, with consequences for tree drought tolerance".

We thank reviewer for his suggestion. While such a title captures one of the main findings, it misses others, and thus we prefer to retain out title.

L19. "minerals" should be "mineral nutrients" or just "nutrients".

Done.

L21. I would be more careful here; exudates can affect nutrient availability directly (e.g.. solubilizing mineral-P) independent of their effects on bacteria.

Thank you, we modified the sentence “However, root exudates typically decrease in situations such as drought, calling into question the efficacy of solvation and bacteria dependent mineral uptake in such stress”.

L22. "Priming" typically refers to the accelerated decomposition of organic matter owing to C (or in some cases N) inputs. In the case of P, phosphates are esther-bonded and so the release of P from soil organic matter is not considered priming. See Figure 2 in Dijkstra et al. 2013 "Rhizosphere priming: a nutrient perspective" for a discussion of this.

Thank you for your insightful comment. Although the abstract is too condensed to include such details, we do find this point highly relevant, and hence corrected for it in the Introduction: “In the forest soil, root exudation is suspected to enhance rhizobacteria, in turn leading to increased decomposition of soil organic matter, i.e., increased C mineralization, a process termed microbial priming (Schleppi et al., 2019). Here, we use this term on a wide perspective, even that P release from soil phosphates, for example, is not strictly considered as priming (Dijkstra et al. 2013)”. We then come back to this point in the Discussion: “The carbon cost of root exudation has already been accounted in relation to P foraging (Dijkstra et al. 2013, Wang and Lambers 2020), where rhizodeposition may be used for P scavenging rather than for decomposition of oil organic matter. On the other hand, slower recovery might indicate a more controlled acclimation response (Bastida et al. 2019, Gessler et al. 2020)”

Bastida, F., García, C., Fierer, N. et al. Global ecological predictors of the soil priming effect. Nat Commun 10, 3481 (2019).

Dijkstra Feike, Carrillo Yolima, Pendall Elise, Morgan Jack. (2013) Rhizosphere priming: a nutrient perspective. Frontiers in Microbiology 4.

L25. Readers may be confused by calling this a factorial design without more details. The two bacterial taxa were not included as individual treatments since they were combined into a "with bacteria" treatment.

Thank you for the comment, we change the text in line 25 “A 1-month imposed drought and concomitant inoculations with a mix of Bacillus subtilis and Pseudomonas stutzeri, bacteria species isolated from the forest soil, were applied using factorial design”.

L28. State percentage increase of exudation?

Yes, this was now added: “Interestingly, root exudation rates increased 2.3-fold with bacteria under drought”.

L30. Would be clearer to readers if you started a new sentence here. "In a second experiment, we added to bacterial cultures metabolites detected as root exudates, and found that xx percent stimulated bacterial growth".

Thank you, the text was modified as suggested: “Forty four metabolites in exudates were significantly different in concentration between irrigated and drought trees, including phenolic acid compounds and quinate. When adding these metabolites as carbon and nitrogen sources to bacterial cultures of both bacterial species, 8 of 9 metabolites stimulated bacterial growth”.

L34. Could add (recruitment of beneficial bacteria) "especially under water stress".

L39. Maybe just start with "Climate change".

L41. Change to "…ecosystems worldwide, with negative impacts for forest health".

Corrected as suggested.

L43-44. Please list references for each impact (recruitment, mortality, etc.)

References were added: “Drought adversely affects forest health in many aspects, including seedling recruitment (Pozner et al. 2022), tree productivity (Klein et al., 2014), and mortality of trees (Klein et al., 2019), with increased susceptibility to pathogen or insect attack (Reichstein et al., 2013; McDowell et al. 2022)”.

Pozner, E., Bar-On, P., Livne-Luzon, S., Moran, U., Tsamir-Rimon, M., Dener, E., … and Klein, T. (2022). A hidden mechanism of forest loss under climate change: The role of drought in eliminating forest regeneration at the edge of its distribution. Forest Ecology and Management, 506, 119966.

McDowell, N. G., Sapes, G., Pivovaroff, A., Adams, H. D., Allen, C. D., Anderegg, W. R., … and Xu, C. (2022). Mechanisms of woody-plant mortality under rising drought, CO2 and vapour pressure deficit. Nature Reviews Earth and Environment, 1-15.

L47. A closing sentence is needed here.

We added: “The capacity of trees to survive future droughts is virtually unknown (McDowell et al. 2022).”

L49. Replace with "including escape, avoidance and tolerance strategies".

L52. Replace with "Root systems mediate water and nutrient uptake…".

Done.

L55. Not sure about the relevance of mentioning fine root turnover here since it's not the focus of the paper.

Yes, but we wish to provide the full context of our study.

L49-84. There are lots of redundancies in paragraphs two and three of the Introduction. The two might be better as a single paragraph about what is known about plant exudation responses to drought, and the impacts of these exudate effects on microbes.

We have condensed the text and merged these paragraphs into a single one: “Trees have evolved mechanisms to cope with drought, including escape, avoidance and tolerance strategies. Drought has been shown to induce an alteration of carbon allocation from aboveground to below ground organs and increase in the amounts of soluble sugars in the roots (Huang et al., 2000; Hasibeder et al., 2015). Root systems mediate water and nutrient uptake, provide physical stabilization, store nutrients and carbohydrates, and provide carbon and nutrients to the soil through the process of fine-root turnover (Brunner and Godbold, 2007; Haichar et al. 2008; Ryan, 2011; Harfouche 2014; Jarzyniak and Jasiński, 2014; Klein 2016). In addition to these roles, trees invest a substantial part of their photosynthesized carbon into root exudates that entice and presumably feed plant-beneficial and root associated microbiota (Bais et al., 2006; Badri and Vivanco 2009; Karst et al., 2017; Jakoby et al., 2020). In parallel, the rhizosphere microbes can promote plant growth through various mechanisms such as increasing the availability of nutrients, secreting phytohormones, suppressing pathogens, or having positive effects on the plant metabolism (Perez-Montano et al., 2014; Zhou et al., 2015). Microbial utilization and metabolism play a central role in modulating concentration gradients of a variety of compounds right outside root tips, thereby constituting a soil sink (Dakora and Phillips, 2002; Mommer et al., 2016; Martin et al., 2017; Tsonuda and van Dam 2017). The majority of root exudates typically consists of primary metabolites (20%; sugars, amino acids, and organic acids) and 15% of nitrogen as well as secondary metabolites, complex polymers, such as flavonoids, glucosinolates, auxins, etc. (Vives-Peris et al., 2020). Those plant-derived metabolites were shown to shape microbial communities by allowing bacteria to metabolize them and then establish themselves in the rhizosphere (Venturi, 2016; Sasse et al., 2017). Although root exudation is ubiquitous among tree species, the amount and composition of root exudates vary. So far, little information is available on how drought influences tree root exudates, their chemical composition, and how root metabolism is connected with shifts in root-associated microbiome composition (Zhang et al., 2007; Tückmant el al., 2017; Naylor and Coleman-Derr 2018). Recently, it was shown that oak trees (Quercus ilex) under drought shift their exudates from primary to secondary metabolites (Gargallo-Garriga et al. 2018)”.

L86-89. Is one of these sentences the topic sentence of the paragraph? Please try to make this clear so that the readers know what the paragraph is about from the outset.

We admit that the structure was confusing here, and hence reordered these sentences: “The chemical composition of root exudates have a direct effect on the rhizosphere communities. These include plant growth-promoting rhizobacteria (PGPR) genera such as Bacillus, Pseudomonas, Enterobacter, Acinetobacter, Burkholderia, Arthrobacter, and Paenibacillus (Sasse et al., 2017; Zhang et al., 2017). For example, the banana root exudate fumaric acid attracts the Gram-positive Bacillus subtilis N11 and stimulates biofilm formation (Zhang et al., 2014)”.

L101. As noted above, priming refers to something beyond just root exudates stimulation in that it refers to the accelerated decay of soil organic matter. Some people refer to what you're describing as "microbial activation" which is a hypothesis of how the first step of priming occurs.

The text has been modified to reflect these points: “In the forest soil, root exudation is suspected to enhance rhizobacteria, in turn leading to increased decomposition of soil organic matter, i.e., increased C mineralization, a process termed microbial priming (Schleppi et al., 2019). Here, we use this term on a wide perspective, even that P release from soil phosphates, for example, is not strictly considered as priming (Dijkstra et al. 2013)”.

L106 "This increase in exudation occurred despite increases in photosynthesis throughout the dry season."

Following the reviewer’s comment the text was changed: “This increase in exudation occurred in spite of the sharp decrease in photosynthesis throughout the dry season, and more so in the coniferous Cupressus sempervirens”.

L111 and L118. Replace "priming" with another term please. "exudate-induced microbial growth".

Following the reviewer’s comment the text was clarified in both locations: “Our overall objective was to test whether microbial recruitment by tree root exudates (a form of microbial priming) is beneficial to trees under drought, an abiotic stress that alters tree carbon and nutrient allocation” and “Our major hypothesis regarded the existence of an exudate induced microbial-tree interaction cycle, starting with tree stress and subsequent exudation, on to enhancement of soil bacteria and their activity, and back to improved tree nutrition”.

L124-129. This "results abstract" does not see necessary unless the journal specifically requires it.

Per the reviewer’s comment, the section was removed.

L140. In the droughted trees in Figure S2, the bacteria treatment seems to be reducing assimilation rates relative to the no bacteria treatment. What might be causing this? This finding seems to be at odds with the other data you present that shows that the presence of bacteria in the drought treatment help trees maintain their nutritional status. Is there some tradeoff that night be occurring which would cause assimilation to go down?

True. This observation is discussed in the third paragraph in the Discussion, which has now been expanded: “Finally, desiccated trees supplemented with soil bacteria showed better nutrition in P and Fe (Figure 5), and higher biomass (although not significantly); but, unexpectedly, did not recover faster than without bacteria. Thus, photosynthesis recovered faster in drought trees that were not inoculated than those that were (Figure S3). Considering that inoculated trees invested significantly more carbon into the rhizosphere than uninoculated drought trees, it is possible that this carbon cost came at the expense of internal tree reserves, later creating a ‘recovery penalty’ for the trees (Gessler et al. 2020). The carbon cost of root exudation has already been accounted in relation to P foraging (Dijkstra et al. 2013, Wang and Lambers 2020), where rhizodeposition may be used for P scavenging rather than for decomposition of oil organic matter”.

L141. Replace "sapling" with "saplings".

Done.

L142-144. Please present a p value for this contrast of a statistical test was conducted.

Biomass data were added in Source Data 1, and the text was modified: “During the experiment, sapling biomass increment was 188.6±15.6 g and 149.1±14.4 g for irrigated saplings with and without bacterial inoculations, respectively (difference not significant; P = 0.753; Source Data 1). The effect of drought was highly significant (P < 0.001), and biomass increment in drought-exposed saplings with and without bacterial inoculations was 22.4±5.7 g and 4.5±4.6 g, respectively (P = 0.09)”.

L146-162. Were statistical tests conducted? If yes, please present the results. If not, please explain why?

Thank you for the comment. Statistical results were added to the text: “Differences between days in relative abundance were significant (P < 0.001; Source Data 2)”. In addition, in: “Bacterial abundance was also lower in drought trees than around irrigated (Figure 1B) or reirrigated trees (Figure 1C) (P < 0.001; Source Data 2)”.

L171. It would be helpful if you could use this format – presenting the % difference in exudation – for the other treatments.

The text was modified as suggested: “The amount of organic carbon exuded from roots of drought-exposed trees was significantly, ~50% lower than from roots of irrigated trees, and moreover, ~70% lower, in the re-irrigation period (across inoculation treatments; P < 0.01; Figure 2, Source Data 3, Supplementary Information). Under constant irrigation, bacterial inoculation significantly increased root exudation, by 3-fold (P < 0.001). However, in trees that were exposed to drought and then re-irrigated, this pattern reversed, and inoculated trees had slightly lower exudates than without bacteria. Thus, only under drought conditions were trees responsive to inoculation, showing a ~50% increase in root exudation (Figure 2; P = 0.039 for the interaction irrigation:bacteria)”.

L174. How is a correlation coefficient calculated when you have one root system (which are used to generate exudation data) but both types of bacteria? Please explain what variables are being used.

Indeed, we had a single exudate mean per treatment and time, and two bacterial abundance values, one for each of the two species. While interactions between the populations of the two strains may exist, it is still possible to correlate between each population and the exudate rates.

L186. Figure 5B?

L202. Is this meant to be Figure 5B (and not 4B)?

Figure numbers were corrected.

L290. Replace "C source" with "C uptake" or "C assimilation"?

Text was changed: “Despite the lower carbon uptake, trees continued to exude altered metabolite blends”.

L298. But lower assimilation in the bacteria-drought treatment (as least based on Figure S2).

Text was added for clarity: “In turn, drought trees that were inoculated had increased biomass increment, as well as leaf nutrients and slightly higher photosynthetic activity before re-irrigation (Figure 5a, Figure S3, S4)”.

L311. Replace "sustain" with "remain"?

Done.

L365. The was these references are used in this sentence makes it seem like the studies were of tree roots (but they were not).

We thank the reviewer for the important point. The references were replaced with a more relevant citation: “Tree roots exude a plethora of secondary metabolites into the rhizosphere, which aid in the mobilization and uptake of essential macro-elements as N, P, and microelements like Mg, Mn, Zn, and Fe (Michalet et al. 2013)”.

Michalet, Serge, Julien Rohr, Denis Warshan, Clément Bardon, Jean-Christophe Roggy, Anne-Marie Domenach, Sonia Czarnes et al. "Phytochemical analysis of mature tree root exudates in situ and their role in shaping soil microbial communities in relation to tree Nacquisition strategy." Plant physiology and biochemistry 72 (2013): 169-177.

L486-489. Was any data collected on the changes in PAR over the course of the experiment? Were trees in full sun?

Information was added per the reviewer’s question: “Global solar radiation was 300-350 W m-2, reduced by ~15% by the net-house”.

L633. Do you have data on what C contamination existed in the non-root controls? Was this subtracted from the values of the ones with roots (i.e. were the fluxes corrected for background interferences)? Some info in the challenges of getting a C-free root chamber would be helpful to readers who want to repeat these methods.

Point well taken. Information was added in the text: “Two samples and one control (solution subjected to the same process without a root; metabolites that were found in control samples were not determined as exudate metabolites) per Rhizobox were included. Small carbon amounts that were found in the control (root-free) tubes were regarded as contamination and were subtracted from the carbon amounts in the samples”. C contamination was technical and resulted from the use of polypropylene syringes to collect the exudates. We assume that the level of contamination depends on the type of syringe and the manufacturing company, so these should be tested and controlled for on a specific experiment basis.

[Editors’ note: what follows is the authors’ response to the second round of review.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Essential revisions:

1) eLife has a consultative review procedure. Subsequent to the submission of both reviews, we discussed Figure 2 due to the Editor's concerns. This figure presents data which are very important for interpretation of the authors' proposed mechanism. In the current form of the manuscript the interpretation of these data is misleading.

(1a) These data show that drought-stressed trees release less TOC (exudate) from roots than irrigated trees, and that both drought-stressed and irrigated trees release more TOC when inoculated. In contrast, trees re-irrigated after drought produce similar TOC from roots whether they are inoculated or not (although the figure indicates there is also a significant difference there, which is confusing and perhaps incorrectly indicated). Thus, the inference is rather that inoculated plants produce more TOC from roots, except if they have been subjected to drought and re-irrigation. What the authors imply is that drought-stressed plants produce more exudates when inoculated, which is correct but misleading since irrigated plants do as well. It is rather interesting that re-irrigated plants do not do this. Reviewer 1 commented additionally: "This contrasts with the answer of the authors to my main question: 'Thus, only under drought conditions were trees responsive to inoculation, showing a ~50% increase in root exudation (Figure 2; P = 0.039 for the interaction irrigation:bacteria).' I really cannot see this interaction in Figure 2, and if anything it seems to go into the wrong direction (i.e. difference with-without bacterial is larger under irrigation)."

Thank you for this point. We understand that our text was still somewhat confusing, and hence revised it to clarify our arguments and interpretation of the data: “Under constant irrigation and under drought, bacterial inoculation significantly increased root exudation, by 2-3-fold (Figure 2; P < 0.001). However, in trees that were exposed to drought and then re-irrigated, this pattern reversed, and inoculated trees had slightly lower exudates than without bacteria. Thus, drought-exposed trees, that were responsive to inoculation, showing a ~50% increase in root exudation, lost this response when re-irrigated (P = 0.039 for the interaction irrigation:bacteria). The amount of organic carbon exuded from roots of drought-exposed trees was significantly, ~50% lower than from roots of irrigated trees, and moreover, ~70% lower, in the re-irrigation period (across inoculation treatments; P < 0.01; Figure 2, Source Data 3, Supplementary Information). Importantly, the same trees that were exposed to drought, reduced their exudation when re-irrigated (Figure 2)”. We also clarified the significance of our results in the section “A dynamic rhizosphere interplay between tree roots and soil bacteria” in the Discussion: “The higher root exudation during drought than under re-irrigation is not trivial, considering that wet soil provides a better environment than dry soil for both tree roots and bacteria development (Gao et al. 2021)”. Considering that the experiment was held in Israel in May-June, i.e. going into the summer, conditions were very dry and hot (Figure S1, S2), making our observations of higher root exudation in drought vs. re-irrigated trees all the more surprising.

Gao, D., Joseph, J., Werner, R. A., Brunner, I., Zürcher, A., Hug, C., … and Hagedorn, F. (2021). Drought alters the carbon footprint of trees in soils—tracking the spatio‐temporal fate of 13C‐labelled assimilates in the soil of an old‐growth pine forest. Global Change Biology27(11), 2491-2506.

(1b) The exudate production has no predictable correlation with abundance of either of the two bacterial strains monitored by the authors. The reason for choosing these strains is now clear in the text, but there is no indication that either responds to differences in root exudates. Thus, although the growth of both strains is affected by (some of) the exudate metabolites which differ (mildly – less than 2x) in abundance between treatments, there is no evidence that these two strains are directly associated with exudate differences. The inoculation affects total TOC from roots, but the effect could be indirect in the ecologically relevant, but complex soil environment of the experiment.

We agree with the Editor that the data from the tree experiment cannot provide direct evidence of exudate-induced growth enhancement of bacteria. Such evidence comes from our bacterial growth essays, which were done in vitro, rather than in vivo (in terms of the tree involvement; Figure 4). We do see positive correlations between exudation rate and rhizosphere abundance of both strains under drought (on Day 3; Figure 2C), but not under irrigation (except for Bacillus subtilis on Day 7; Figure 2C), nor re-irrigation. Therefore, overall, these correlations were inconsistent over times and treatments. Per the Editor’s comment, we toned down our claim in the opening paragraph of the Discussion: “Soil bacteria associated with tree roots and grew preferentially in the rhizosphere (than in the bulk soil; Figure 1), where their presence transiently increased with root exudates (Figure 2, Figure 7)”. We also agree that the interactions between root exudation rate and bacteria could be indirect in the complex soil environment of the experiment. We acknowledge and expand on that in the Discussion section “A dynamic rhizosphere interplay between tree roots and soil bacteria”: “Our temporal resolution and complex experimental setup do not address the question if exudates attracted the bacteria, or, alternatively, whether the bacteria induced root exudation. It is likely that both processes occurred simultaneously. One must also consider alternative explanations, e.g. that exudates act directly on soil elements, rather than solicited by bacteria. During drought, changes in soil water content could become either beneficial (increased Mn and P availability) or harmful (decreased Zn availability) to plant nutrition (Misra and Tyler 1999). Tree roots exude a plethora of secondary metabolites into the rhizosphere, which aid in the mobilization and uptake of essential macro-elements as N, P, and microelements like Mg, Mn, Zn, and Fe (Michalet et al. 2013). Several studies reported that beneficial rhizosphere bacteria drive an accumulation of elements (Philippot et al.,2013; Igiehon et al., 2018). Although we showed that multiple exudate metabolites promoted bacterial growth, this does not mean that all of these metabolites act as specific cues for specific bacterial strains in the rhizosphere. However, the intimate attachment of bacteria to roots shown here should ensure the harvest of these metabolites by bacteria”. Finally, we refer to this option in the Discussion section on study limitations: “We offered a glimpse into the rhizosphere, which was not, however, free of limitations. First, our approach ignored the native rhizosphere microbiome, which probably affected many of the studied parameters”.

(1c) There are no data to indicate that these bacterial strains are responsible for the nutrient effects observed for inoculated plants undergoing drought, especially as the soil for the experiment was not sterilized. The unsterilized soil is in fact a nice aspect of the design from an ecological perspective, but also makes it more complicated to determine the mechanism by which additional inoculation has its effects.

Point well taken. As for the previous comment, we toned down our claim. This affected text in the Results: “Levels of P, Fe and Zn were significantly lower in leaves in drought compared to irrigated trees during the middle of the drought period (27.05). This effect was mitigated in inoculated trees under drought for P and Fe, but not for Zn (Figure 5A, Source Data 6)”. Please note that the Discussion text is already cautious with this matter, avoiding bald claims of a direct effect of the bacterial strains on leaf elements, e.g. in the first Dicussion paragraph: “In turn, drought trees that were inoculated had increased biomass increment, as well as leaf nutrients and slightly higher photosynthetic activity before re-irrigation (Figure 5a, Figure S3, S4)”, and in the third Discussion paragraph: “Finally, desiccated trees supplemented with soil bacteria showed better nutrition in P and Fe (Figure 5), and higher biomass (although not significantly)”. As detailed above, we acknowledge the role of the existing rhizosphere microbiome in our experiment, including its restrictions on our ability to draw direct evidence about the effects of the chosen strains. However, we do believe that the observed effects on leaf elements can be ascribed to our bacterial inoculations, for the following reasons: (1) When comparing the responses of inoculated drought-exposed trees to those without supplemented bacteria, a significant enhancement emerged. (2) The bacterial inoculations were most probably at higher numbers than the existing bacterial populations (as explained in the text), and hence must have had larger effects than the latter. (3) At least for P nutrition, our soil phosphorous measurements seem to support and complement the beneficial effect of inoculation on drought-exposed trees. (4) Previous studies have shown the beneficial effects of these strains to plants, e.g. via stimulating iron acquisition (Mendes et al. 2013 cited in text) suppressing plant pathogens (Weller et al. 2002, Raaijmakers et al. 2010 cited in text), and specifically fungi (Mendes et al. 2013). Finally, to provide a direct link between these bacterial strains and nutrition benefits to trees, we assume that additional approaches should be applied, e.g. growing plants in hydroponic system, and applying isotopic labeling to track specific elements as they transfer from roots to leaves.

(2) Reviewer 1 emphasized that the writing still needs substantial correction and streamlining.

Following this comment, an English editor has reviewed once again the entire text, correcting the remaining typos and grammar mistakes.

The full reviews are provided below. Please note that the topics under (1) arose after the reviews were submitted and during the consultation.

Reviewer #1 (Recommendations for the authors):

This manuscript has been improved during revision, including improved statistical analysis. In my view the Discussion would still benefit from a better focus on the main results.

I mentioned in the previous review that parts of the paper were quite long and complicated to read and still find this to be true (see e.g. Discussion lines 413-443; overall, the Discussion has become even longer than it was before). I also mentioned that there were many small language errors and again can still find them in this "revised" version. It is really the authors responsibility to correct the language because they are the "publisher".

We thank Reviewer 1 for the valuable inputs. The Discussion is again longer than expected, due to journal’s regulations to avoid Discussion chapters in Supplementary Information. However, it has been condensed as much as possible, while avoiding compromising the wealth of insights arising from such an experiment. Per the reviewer’s comment we were able to mildly shorten the Discussion paragraphs under the section “The chemistry of tree–microbe interactions in the rhizosphere”, and in a few other places in text. Finally, an English editor has reviewed once again the entire text, correcting the remaining typos and grammar mistakes.

https://doi.org/10.7554/eLife.79679.sa2

Article and author information

Author details

  1. Yaara Oppenheimer-Shaanan

    Department of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Formal analysis, Methodology, Writing - original draft
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7005-3074
  2. Gilad Jakoby

    Department of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Formal analysis, Methodology
    Competing interests
    No competing interests declared
  3. Maya L Starr

    Department of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  4. Romiel Karliner

    Department of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  5. Gal Eilon

    Department of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  6. Maxim Itkin

    Department of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Formal analysis, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1348-2814
  7. Sergey Malitsky

    Department of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Formal analysis, Methodology
    Competing interests
    No competing interests declared
  8. Tamir Klein

    Department of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Conceptualization, Supervision, Funding acquisition, Writing - original draft, Writing - review and editing
    For correspondence
    tamir.klein@weizmann.ac.il
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3882-8845

Funding

Edith and Natan Goldenberg Career Development Chair

  • Tamir Klein

Mary and Tom Beck (Canadian Center for Alternative Energy Research)

  • Tamir Klein

Larson Charitable Foundation New Scientist Fund

  • Tamir Klein

Angel mFaivovich Foundation for Ecological Research

  • Tamir Klein

Yotam project

  • Tamir Klein

Dana and Yossie Hollander

  • Tamir Klein

Estate of Emile Mimran

  • Tamir Klein

Estate of Helen Nichunsky

  • Tamir Klein

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank The Weizmann Tree Lab members for support, advice, and helpful discussions throughout. We also thank Guy Shmuel for providing us with a protocol of ICP-MS; Nir Galili for help with earlier experiments; Roee Ben Nissan for assistance with programing of bacteria growth curve; Ron Rotkopf for guidance with statistical analysis; and Sergey Kapishnikov for coordinating the elemental mapping experiments (Weizmann Institute of Science). We thank SH Lau and the engineers of Sigray (San Francisco, CA) for performing the elemental mapping on our samples; Robert Fluhr (Weizman Institute of Science), Daniel Dar (Caltech, CA), and Sophie Obersteiner (Ben Gurion University, Israel), for providing helpful comments on earlier versions of this paper. Funding: The project was funded by The Edith and Nathan Goldenberg Career Development Chair; Mary and Tom Beck-Canadian Center for Alternative Energy Research; Larson Charitable Foundation New Scientist Fund; Angel Faivovich Foundation for Ecological Research; Yotam Project; Dana and Yossie Hollander; Estate of Emile Mimran; and Estate of Helen Nichunsky.

Senior and Reviewing Editor

  1. Meredith C Schuman, University of Zurich, Switzerland

Reviewer

  1. Alexander Weinhold, iDiv, Germany

Publication history

  1. Preprint posted: August 25, 2021 (view preprint)
  2. Received: April 22, 2022
  3. Accepted: July 17, 2022
  4. Accepted Manuscript published: July 20, 2022 (version 1)
  5. Version of Record published: August 17, 2022 (version 2)

Copyright

© 2022, Oppenheimer-Shaanan et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 918
    Page views
  • 356
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Yaara Oppenheimer-Shaanan
  2. Gilad Jakoby
  3. Maya L Starr
  4. Romiel Karliner
  5. Gal Eilon
  6. Maxim Itkin
  7. Sergey Malitsky
  8. Tamir Klein
(2022)
A dynamic rhizosphere interplay between tree roots and soil bacteria under drought stress
eLife 11:e79679.
https://doi.org/10.7554/eLife.79679

Further reading

    1. Ecology
    Ting Tang, Naili Zhang ... Xiaojuan Liu
    Research Article

    Addressing global biodiversity loss requires an expanded focus on multiple dimensions of biodiversity. While most studies have focused on the consequences of plant interspecific diversity, our mechanistic understanding of how genetic diversity within plant species affects plant productivity remains limited. Here, we use a tree species × genetic diversity experiment to disentangle the effects of species diversity and genetic diversity, and how they are related to tree functional diversity and trophic feedbacks. We found that tree species diversity increased tree productivity via increased tree functional diversity, reduced soil fungal diversity and marginally reduced herbivory. The effects of tree genetic diversity on productivity via functional diversity and soil fungal diversity were negative in monocultures but positive in the mixture of the four tree species tested. Given the complexity of interactions between species and genetic diversity, tree functional diversity and trophic feedbacks on productivity, we suggest that both tree species and genetic diversity should be considered in afforestation.

    1. Ecology
    2. Evolutionary Biology
    James P Higham
    Insight

    By spending more time around infants which physically resemble their own, mandrill mothers may increase how frequently their offspring interact with their paternal half siblings.