Half of the metabolites detected in root tissue extracts were significantly different in pairwise comparisons of experimental treatments, with 28% having highest abundance in phosphate‐sufficient‐, 30% in phosphate‐deficient‐, and 25% in soil‐ extract‐grown roots . The significantly different metabolites could be grouped into four main clusters . Cluster I consists of three metabolites significantly different between all experimental treatments. Cluster II is composed of metabolites abundant in phosphate‐sufficient roots, including nucleosides, organic acids, amino acids, and, notably, all phosphorous compounds present in this dataset. The higher abundance of phosphorous compounds in phosphate‐sufficient roots compared with phosphate‐deficient‐ or soil‐extract‐ grown roots is in line with the phosphate quantification of plant tissues , in which the highest free phosphate was detected in phosphate‐sufficient plants, as would be expected. Cluster III includes metabolites abundant in phosphate‐deficient roots. All these metabolites are nitrogenous compounds, likely due to the N–phosphate imbalance of phosphate‐deficient plants. Cluster IV contains metabolites distinct for soil‐extract‐grown roots, and is split in two subclusters: IVa includes metabolites with low abundance in soil extract roots, which are mostly nitrogenous compounds, whereas IVb includes metabolites with high abundance in soil extract roots, which are mostly organic acids.Overall, 137 metabolites were identified in root exudates . Only phenylacetaldehyde was significantly different between exudates of phosphate‐sufficient and ‐deficient plants,vertical plant growing which explains why these conditions are not separated in a principal component analysis .

Plants grown in soil extracts had a distinct exudate composition, with 27 and 25 distinct compounds vs phosphate‐sufficient and phosphate‐deficient root exudates, respectively. Most of these distinct metabolites were most abundant in soil extract controls , showed medium abundance in soil extract exudates, and had low abundance in the other conditions . Metabolite comparisons between soil extract with and without plants revealed that half of the metabolites detected were altered in abundance, causing a distinct grouping in a principal component analysis . Fifty percent of these metabolites were depleted in the presence of plants . Although individual metabolite levels varied somewhat across laboratories, this finding was consistent across participating laboratories . Distinct metabolites included organic acids, carbohydrates, amino acids, and nucleosides, and these compounds contained various groups, such as phosphate, N, or sulfur . Furthermore, citric acid exhibited an interesting but statistically insignificant trend of higher abundance in soil extract exudates vs controls .Metabolites that were detected in root tissue and root exudates showed distinct patterns: 42% of these metabolites were significantly different in roots and 43% in exudates, depending on environments. Only 23% of the compounds were significantly different in both datasets, which indicates that root exudates are metabolically distinct from root tissue . We similarly found that only 50% of the metabolites depleted from soil extract were significantly different in root tissues, with 29% of high abundance in soil extract roots , 25% of low abundance, and 46% are not detected . This study investigated the reproducibility of morphological and metabolic responses of the model grass B. distachyon grown in EcoFABs in phosphate‐sufficient and phosphate‐deficient mineral medium and in chemically complex but sterile soil extract. We purposely chose phosphate starvation as an experimental system, as the morphological and metabolic responses of plants are well described and should be reproducible in a system such as the EcoFAB.

The soil extract medium was added to represent a more natural environment, but it was sterilized to exclude the effects of microbial metabolism on exudation and to lower variability of the system. We found that B. distachyon FW, phosphate content, and metabolic profiles were distinct for our experimental conditions and that these responses were reproducible across the four participating laboratories. The traits investigated included tissue FW and phosphate content, total root length, and metabolic profiles of roots and exudates. These results compare favorably to a related study comparing three Arabidopsis thaliana genotypes grown in soil in pots by 10 laboratories where, similar to this study, materials were distributed from one laboratory, growth conditions were monitored at each laboratory, and one laboratory analyzed leaf morphology and metabolomic and transcriptomic profiles. Although one trait was similar between a core group of four laboratories, all traits significantly varied across laboratories. The authors attributed the variance to the strong influence of small environmental changes in their soil pot system . Our EcoFAB setup comprised a more uniform and controlled growth environment than pots filled with soil, which is likely one cause of the higher reproducibility observed here. Another equalizing factor mighthave been the use of sterilized soil extract in this study, which did not take into consideration the complex physical and mineral properties of soil, or the effects of microorganisms. It could be that integrating these factors in future EcoFAB studies might increase the variability of the system. It will be important to investigate the reproducibility as well as the morphological and metabolic responses of plants to microbial communities and soil mineralogy, as natural soils were identified as main contributors shaping root morphology, plant C exudation, plant– microbe interactions, and rhizosphere extension . Overall, we conclude that the reproducibility of plant traits in soil extract EcoFABs is a promising first step towards developing plant growth systems generating reproducible data that are relevant to field environments.

Root metabolic profiles were clearly distinct between experimental treatments. Phosphate‐ sufficient roots were abundant in nucleosides, amino acids, organic acids, and phosphorous compounds, whereas phosphate‐deficient roots accumulated nitrogenous compounds, and soil‐extract‐grown roots were deficient in nitrogenous compounds, but accumulated carbohydrates . It will be interesting to investigate whether shoot metabolic profiles are similarly distinct between experimental treatments in a future study. The metabolites detected in B. distachyon root exudates in this study were comparable to metabolites detected in exudates of other grasses, such as wheat , maize , rice , Avena barbata , and dicots such as Arabidopsis . Similarly, the B. distachyon exudation profile varied with developmental stage, as reported for other plants . The largest exudate metabolic differences in this study were observed between plants grown in soil extract and soil extract controls without plants. Surprisingly, we did not find many statistical differences in exudates of plants grown in phosphate‐sufficient vs ‐deficient conditions. For many plants, an increase in organic acid exudation in low phosphate conditions was reported , which was not found in our dataset. This might be due to several reasons. First, plants were grown without phosphate for the entire growth period and might have ceased differential exudation when sampled after 3 wk. Second, the small EcoFAB volume likely allows for re‐uptake of exuded metabolites,growing strawberries vertically masking differential exudation of compounds. Third, the exudation response of B. distachyon to phosphate starvation might not be as pronounced as in other species and be below the detection limit in our assay. Future experiments focusing on the timing and magnitude of B. distachyon exudation changes in response to phosphate starvation would be able to address these points. The clear differences observed for FW, tissue phosphate content, and root metabolic profile indicate that the plants indeed were starved for phosphate in our experimental setup.The main differences in exudate metabolic profiles in this study were due to a depletion of metabolites from soil extract by plants . With our experimental setup, we are unable to determine whether metabolites are depleted due to uptake by plant roots or due to, for example, chemical reactions caused by an altered pH around plant roots. Experiments with isotopically labeled compounds spiked into soil extract could address the fate of metabolites of interest in future experiments. In addition to depletion of metabolites, a trend for increased citric acid levels in soil‐extract‐ grown plants was observed. This might constitute a starvation response, given that exudation of organic acids is a characteristic of phosphate‐limited plants . The fact that half of the soil extract metabolites are depleted by plants is surprising, as it suggests that plants not only are producers, but also consumers of a significant amount of compounds.

Among them is pterin, which is a folate precursor. Folate is an essential part of human diet, and thus studying uptake of pterin by plants to elevate folate levels might be an interesting bio-fortification strategy . Xanthine is part of the purine degradation pathway in plants and can act as a sole N source for A.thaliana growth . Similarly, there could be direct utilization of thymine and thymidine for synthesis of nucleic acids and of N‐acetyl‐L‐glutamic acid for synthesis of amino acids. In addition, plants deplete complex organoheterocyclic compounds such as the ascorbic acid precursor gulonolactone , as well as simple carbohydrates such as sucrose. Uptake of these compounds by roots would indicate that plants grow partially heterotrophically in specific environments, importing simple and complex biomass precursors. There is only a small amount of literature regarding uptake of metabolites by roots: amino acids and sugars were reported to be imported by roots in mineral medium assays where compounds were spiked in , whereas organic acids are likely not imported at significant amounts . There is evidence that plants are capable of importing C from environments , but overall, the scope of how much and which metabolites are taken up by plants from natural environments is currently unknown. In another experimental system comprising cyanobacteria and associated heterotrophs, it was found that the primary producer depleted 26% of biological soil crust metabolites, whereas soil heterotrophs only depleted 13% of metabolites . This might suggest that photoautotroph organisms in general not only release, but also deplete, a significant amount of compounds from the environment. Plants might compete with microbes for nutrient soil organic compounds in certain environmental conditions. Besides nutritional functions, compounds could act as signals, as exemplified by a recent study that found the depletion of plant‐derived phenolic acids to be associated with rhizosphere microbes . Many of the plant‐depleted metabolites contained N, phosphate, or S groups , which suggests that plants not only use inorganic forms, but also more complex compounds as nutrients. Consistent with this hypothesis, compounds containing the N, phosphorus , and S groups are low in soil‐extract‐grown roots, likely indicating a fast turnover rate. It was suggested that amino acid uptake might account for 30–90% of imported N, depending on the environmental conditions , but overall, data on how much elements are taken up as inorganic vs organic compounds is missing. By contrast to N‐, P‐, and S‐containing compounds, carbohydrate‐type compounds were of high abundance in soil‐extract‐grown roots, likely due to a low external demand for carbohydrates by plant tissues . Interestingly, plants depleted metabolites from soil extract in a selective manner, suggesting that the plant controls depletion of metabolites to a certain degree. Similarly, the difference between root and exudate metabolic profiles indicates that plants control exudation to some degree. Selectivity in import and export processes could be achieved by the presence of transport proteins that were described for a number of metabolites , and investigation of transport processes is a promising direction for future studies. We conclude that plants not only significantly alter their environment by export, but also by depletion of metabolites.In this study, plants were grown in basal salt medium widely used in standard laboratory settings, and in soil extract medium that includes water‐soluble metabolites but that excludes additional factors defining soils, such as presence of other metabolically active organisms or solid soil particles. We observed an increased root : shoot ratio in plants grown in soil extract, which might point to nutrient limitations , consistent with the low phosphate content of soil extract and of soil‐extract‐grown plants . Interestingly, altered root : shoot ratios were recently also detected for wheat genotypes grown in different soils , suggesting that different soils might affect root : shoot ratio and possibly also metabolic profiles in different ways. The most prominent phenotypic difference observed for soil‐extract‐grown plants was the four‐fold increase in root hair length compared with other plants . Root hair elongation can be caused by altered nutrient levels , and depends on the growth condition used . Further, the response to phosphate is concentration dependent , which might be the cause for the different root hair phenotype observed in phosphate‐ deficient medium vs phosphate‐limited soil extract.