It is estimated that by 2050 the world population will exceed 9 billion people and that crop productivity should increase by 70% to meet food demands (FAO et al., 2021). However, both the number and severity of drought events are predicted to increase in near future, especially from 2019–2034, impacting crop productivity (Spinoni et al., 2020; Waseem et al., 2021). This is critical for plants with high water demand, such as tomato, which require between 400 and 600 mm of irrigation water for seasonal production (Ronga et al., 2019a; Ronga et al., 2019b). Therefore, it is imperative to develop sustainable strategies to optimize plant productivity and food quality under water scarcity.

Today, it is widely accepted that microorganisms have important roles in determining host fitness during periods of water deficit stress (Naylor et al., 2017; Xu and Coleman-Derr, 2019). Many bacteria from the endosphere and rhizosphere microbiome have been described and highlighted as providing a benefit to their host by several well-documented mechanisms and others that have not yet been fully clarified. A main bacterial mechanism is the production of the enzyme 1-aminocyclopropane-1-carboxylate (ACC) deaminase (ACCD), which catalyzes the hydrolysis of the immediate precursor of ethylene, the ACC, to ammonia and α-ketobutyrate (Tiwari et al., 2018). Many plants growth-promoting bacteria (PGPB), can also produce phytohormones such as indole acetic acid (IAA), which regulates various development processes in the plant, such as root initiation, stem elongation, apical dominance, root size and distribution, resulting in greater water and nutrient absorption from the soil (Orozco-Mosqueda et al., 2023). Traditionally, PGPB have been isolated, characterized, identified, bioaugmented and re-inoculated into soil or plant seeds, with the finality of optimizing plant fitness under a stressing condition, such as nutrient limitation or water stress (Barra et al., 2017; Barra et al., 2019; Rahnama et al., 2023). However, despite the promising findings of bacterial bioinoculants in laboratory and growth chamber conditions, their performance at the field level has frequently been suboptimal, resulting in limited commercial adoption on a large scale (O’Callaghan et al., 2022). Several hypotheses have been formulated to explain these dissimilarities and inconsistencies among controlled and field conditions, such as low inoculum survival due to competition with native microorganisms, starvation, limited distribution of the inoculant in the soil profile, and variations in soil properties, among other (Kaminsky et al., 2019). Furthermore, the utilization of inoculums based solely on individual bacteria may be constrained due to the intricate synergistic interactions and complex interconnected networks that exist among various components of the microbiota (Backer et al., 2018; Qiu et al., 2019). Auspiciously, some studies have showed that inoculation with bacterial consortia is a more effective approach than inoculation with a single strain, since bacteria seem to function synergistically and can along compete for certain ecological niches (Berendsen et al., 2018a; Woo and Pepe, 2018). For these reasons, the use in situ of all (or most) microbiota, their synergy and the complex networks, avoiding previous cultivation, could represent a promising alternative for breaking down the barriers of pre-existing microbial niches and, thus, more efficiently incorporate bioinoculants with beneficial properties for crops.

Plant-microbe relationships are often shaped by alterations in radical exudation of primary and secondary plant metabolites, such as organic acids, tryptophan and strigolactone, which can be substrates for microbial growth, elicit chemotactic responses, facilitate root colonization, or inhibit the growth of some specific microbial taxa (Rudrappa et al., 2008; Hassan and Mathesius, 2012; Liu et al., 2017; Massalha et al., 2017; Menezes-Blackburn et al., 2021). The process by which the plant modulates root exudation of metabolites depends on various intrinsic plant factors as well as external factors such as nutrient availability, management practices, abiotic stresses, soil conditions and the surrounding organisms (Compant et al., 2019; Rolfe et al., 2019; Menezes-Blackburn et al., 2021). In turn, plant-associated microorganisms can induce specific systemic changes in root exudation, and therefore also modulate the plant-associated microbiome structure (Korenblum et al., 2020).

In recent years, Host-Mediated Microbiome Engineering (HMME) has emerged as a novel and eco-friendly in situ microbiome engineering strategy based on the plant’s intrinsic ability to recruit and maintain a beneficial microbiome (Lakshmanan et al., 2012; Jochum et al., 2019; Mueller et al., 2019). By employing a host multigenerational approach and visualizing changes in the host phenotype, this strategy enables the indirect selection of a specific microbiome, wherein the plant recruits a diverse array of microorganisms. These microorganisms, encompassing beneficial, neutral, or potentially negative types, collectively contribute to the overall well-being and favorability of the plant (Swenson et al., 2000; Raaijmakers and Mazzola, 2016; Mathur and Roy, 2021). Although HMME has only recently been encompassed as a subject of study, this natural plant defense mechanism was evidenced in early reports of ‘suppressive soils’(Cook and Rovira, 1976; Jayaraman et al., 2021). Natural suppression of soil- borne pathogens were induced by crop monoculture (Wildermuth, 1982; Sturz et al., 1999; Cook, 2003; Sagova-Mareckova et al., 2015; Sahu et al., 2018), wherein plants ‘recruited’ microorganisms involved in disease protection after a pathogen outbreak, a phenomena called ‘The Cry for Help’ (Bakker et al., 2018; Berendsen et al., 2018b; Li et al., 2019; Rolfe et al., 2019).

We recently reported that HMME over multiple generations can be achieved by managing three co-occurring factors; i) a favorable microbiota as a donor of microorganisms, ii) a stressor factor to induce plant response and shape microbial community assemblages, iii) a host model as modulator (Durán et al., 2021). Extreme environments are an interesting habitat likely to contain microbiota with the potential to enhance plant fitness (Jeon et al., 2009; Mishra et al., 2011; Gonc et al., 2015; Peixoto et al., 2016) due to strong selective pressure. These environments may lead to the evolution of novel mechanisms for stress tolerance (Convey and Smith, 2006; Kasana and Gulati, 2011; Liu et al., 2019). In this way, Antarctica is a unique ‘laboratory’ to study life in extreme conditions, where the combination of an extensive glacial layer, intense katabatic winds, and extremely low precipitation rates makes them the oldest, coldest, and driest deserts on Earth (Rycroft, 1987). So far, several studies show that the Antarctic microorganisms could confer plants protection against different abiotic stresses (Gallardo-Cerda et al., 2018; Yarzábal et al., 2018; Garcia et al., 2021; Styczynski et al., 2022).

In this study, we investigated the effect of Antarctic microbiota donation and plant microbiome selection by HMME on the improvement of water deficit stress tolerance in tomato plants. We focused our study on addressing three major questions: (1) Is it possible to improve the tolerance of tomato plants to water deficit through Antarctic microbiota transference using HMME? (2) Can extreme microbiota survive in agricultural soil? (3) If so, which taxonomic groups co-evolve with the host-plant during the multigenerational water deficit selection? By answering these questions, we hope to contribute new approaches to enhance sustainable agriculture through the development of a new generation of in situ bioinoculants.

The potential for microbiome soil transplant to improve tomato plant tolerance against water deficit stress was assayed through the HMME after Antarctic microbiome soil donation. The Antarctic soil microbiomes were chosen as sources of desirable microorganisms due to their inherent capability to withstand extreme arid conditions characterized by limited liquid water availability (Morales-Quintana et al., 2021; Ball et al., 2022).

A noticeable enhancement in the tolerance of tomato seedlings to water deficit stress was observed in all soil mixtures throughout successive generations of HMME. This was evidenced by a significant improvement in survival time, plant stress index decrease, biomass accumulation, and a reduction in leaf proline content. This enhanced plant response was primarily evidenced in R+F, which had the highest plant biomass under water deficit in G10. In general, at the beginning of the study, higher levels of proline were found in leaves of tomato seedlings subjected to water deficit as compared to the well-irrigated control. The proline levels gradually decreased in the leaves of plants grown in each soil mixture over successive generations, with a more pronounced decline becoming apparent from G5 onwards. Concurrently, an evident improvement in tolerance to water deficit stress began to manifest. The intracellular accumulation of osmolytes, such as proline, represents a crucial physiological plant response aimed at mitigating cell membrane damage and preserving cellular integrity (Claussen, 2005; Montesinos-Pereira et al., 2014). Elevated proline levels have been linked to the response to water stress, which can be also induced by inoculating the plant with PGPB. (Wang et al., 2012; Palika et al., 2013; Shintu and Jayaram, 2015). For example, Bacillus spp. have been shown to be efficient PGPB that also have the capability to reduce proline levels in plants (Wang et al., 2012; Palika et al., 2013; Shintu and Jayaram, 2015; Mehboob et al., 2022). The decreased levels of proline, together with the improved performance of the tomato seedlings under water deficit, which were even comparable to the well-irrigated controls, indicated that the plant ceased to be dramatically stressed towards the last generations of the study. While undergoing generations, the rhizosphere microbiota associated with tomato plants was concurrently restructured and microbial alpha/beta diversity indices declined across generations. Thus, microbial communities became phylogenetically more similar and taxa abundances were more uniform within treatments, but communities became more dissimilar from one another as generations progressed. The most dominant phyla found in the rhizosphere soil of tomato seedlings in all soils over all generations were Proteobacteria, Actinobacteria, Acidobacteria and Firmicutes. These phyla consistently emerge as dominant taxa in the rhizosphere soils of diverse plant species (Jochum et al., 2019; Mueller et al., 2019; Jiang et al., 2022), including those inhabiting Antarctic soils (Zhang et al., 2020) and Andisols (Lagos et al., 2014; Acuña et al., 2020). Possible reasons for their prevalence lies in these bacterial taxa possessing specific life strategies and characteristics that enable them to endure various environmental stresses. These include fast growth rates (Ling et al., 2017), metabolic versatility (Lladó and Baldrian, 2017) and phenotypic plasticity (Perraud et al., 2020). For example, the spore-forming capacity of Firmicutes and Actinobacteria allows them to enter a dormant state during periods of stress by water deficit, while the less suitable bacterial lineages decreases in abundance (Naylor et al., 2017). Although the dominant taxa persisted, water deficit and multigenerational selection significantly influenced on the rhizobacterial microbiota composition, which could be linked with changes in soil moisture and the release of primary and secondary metabolites by tomato plants (Bakker et al., 2018; Berendsen et al., 2018b; Li et al., 2019; Rolfe et al., 2019). These findings align with the hologenome theory of evolution, which postulates that eukaryotic organisms establish close relationship with microbiota that could be inherited and exert significant effects on host fitness (Carrier and Reitzel, 2018). The host-associated microbiota can undergo dynamic changes, influenced by modifications in the environment, host factors, and microbial genomes (Carrier and Reitzel, 2018). We unveiled a notable variation in the structure of the bacterial microbiota across each evaluated generation, highlighting the dynamic nature of these host-microorganism interactions. As expected, the rhizospheric microbiota underwent an adaptation process in intermediate generations (G2 and G5), where several taxonomic taxa as Rhodococcus and Paenibacillus were suppressed, while Sphingomonas was enhanced. Overall, when assessing the multigenerational-selected soil microbiome (after G10) under both control and water deficit conditions, the bacterial genera Sphingomonas and Bacillus, along with the archaea Candidatus Nitrosocosmicus, emerged as the most prevalent taxa. Notably, under water deficit stress conditions, Bacillus and Candidatus Nitrosocosmicus exhibited higher abundance compared to other taxa, reflecting the impact of the HMME approach on shaping the microbial composition. Bacillus spp, are ubiquitous inhabitant of soils, with a wide distribution in natural environments and are frequently associated with diverse plant species and niches. Numerous publications have evidenced the importance of species from this genus in alleviating water deficit stress in plants via phytohormone synthesis, volatile compounds production, increasing nutrient availability and the activity of the enzyme ACCD, which reduces ethylene-induced stunting in plants (Barra et al., 2016; Durán et al., 2016; Gagné-Bourque et al., 2016; Barra et al., 2017; Tiwari et al., 2018; Moreno-Galván et al., 2020; Astorga-Eló et al., 2021). The increase in abundance of Candidatus Nitrosocosmicus (Archaea, Thaumarchaeota phylum) was evidenced in all soils after G10, but especially in tolerant plants to water deficit from R+F soils. Members of this phylum are relatively rare across many environments but seem to be particularly abundant in desert soils (Fierer et al., 2012; Marusenko et al., 2013). This archaea has been reported to be involved in biogeochemical cycles, particularly in carbon and in nitrogen cycles, as an ammonium oxidizing archaea (AOA), important for the control of soil nitrification (Lehtovirta-Morley et al., 2016; Wang et al., 2019). A previous work described that this archaea is tolerant to prolonged water deficit events (Hammerl et al., 2019), therefore water deficit would benefit its proliferation with respect to other less tolerant microorganisms. Due to the remarkable increase in abundance of Bacillus spp. and Candidatus Nitrosocosmicus under water deficit and their consistent dominance suggests their potential involvement in key functional processes within the soil ecosystem, particularly under water deficit stress conditions. In addition, it is well-documented that chloroform treatment can significantly reduce the microbial population size by at least 70% (Ingham and Horton, 1987). Therefore, to further validate the role of the microbiome in water stress tolerance, fumigation of soils was carried out after G10, after which new seedlings were grown in the fumigated soils. After fumigation, plants once again became susceptible to water deficit stress, strongly indicating the significant role of the microbiota in conferring acquired tolerance to water deficit in plants. It is also important to highlight that given the inherent soil buffering capacity, the transplantation of donor soils had minimal effects on key soil receptor attributes, including macronutrient levels, soil organic matter (SOM) content, and pH values (Nanzyo, 2002; Ugolini and Dahlgren, 2002). Therefore, we propose that these prokaryotic taxa could play a significant role in enhancing the tolerance of tomato seedlings to water deficit, especially in soils engineered with soils from Fildes Bay. Although there is a noticeable trend of enrichment with these taxa in the engineered soils associated with enhanced water stress tolerance, further research is needed to provide stronger evidence and confirm these findings. Additional studies are required to investigate the molecular and physiological basis of their interactions with plants, as well as their potential roles in nutrient acquisition and stress mitigation pathways. Moreover, it is essential to unravel the specific mechanisms and interactions by which these taxa could contribute to plant-microbe associations and nutrient cycling processes. By comprehending the functional contributions of Bacillus and Candidatus Nitrosocosmicus, we can advance the fine-tuning and optimization of the microbiome engineering approach for the development of sustainable and efficient biofertilization strategies.

The beneficial role of plant-associated microorganisms isolated from the rhizosphere and utilized as plant bioinoculants under conditions of water deficit stress has been extensively demonstrated, providing promising results under controlled experimental conditions (Asaf et al., 2017; Ximénez-Embún et al., 2018; Zhang et al., 2018; Kim et al., 2022). However, studies investigating the impact of environmental stressors, such as water deficit stress, on the adaptation, composition and functionality of plant-associated microbiota over time are relatively recent and limited in number. In this context, the FAPROTAX analysis is a valuable and accurate tool utilized for rapid functional screening and grouping of data derived from the 16S rRNA gene in terrestrial ecosystems (Sansupa et al., 2021). As expected, chemoheterotrophy and aerobic chemoheterotrophy were the predominant functions in all samples and treatments, possibly due to the environment in which the plant developed. However, the significant increase in the relative abundance of AOA, particularly Candidatus Nitrosocosmicus, particularly in the R+F treatment under water deficit conditions suggests a potential link between enhanced water deficit tolerance and increased oxidation of ammonium (NH₄ ⁺) or ammonia (NH₃) to nitrate (NO₃^-) in the soil. This, in turn, may lead to an augmented nitrogen absorption capacity by the plant. Nevertheless, studies of the functionality of the plant-associated microbiota, including transcriptomic and proteomic analysis, throughout and after the HMME approach are still necessary to corroborate this hypothesis.

Our findings demonstrate the potential to enhance the water deficit tolerance of tomato plants by transferring the Antarctic microbiome through 20% w/w soil inoculation. This enhancement could be achieved through the recruitment, maintenance, and restructuring of microbiomes via multigenerational plant selection. Interestingly, our results show that specific taxa from Antarctic soils were capable of persisting in the tomato rhizosphere throughout the entire 96-week experiment. For instance, 45 taxonomic groups (ASVs observed) from Fildes Bay remained consistent during the multigenerational selection process under water deficit conditions. This suggests that some taxonomic groups co-evolved with the host plant during the multigenerational water deficit selection.

When examining the relative abundance data across the course of multigenerational selection, it becomes evident that Bacillus spp. and Candidatus Nitrosocosmicus could potentially serve as pivotal contributors to enhancing water deficit tolerance in tomato seedlings, particularly in soils from Bahía Fildes. However, further trials are needed to validate this hypothesis. It is worth noting that understanding the underlying mechanisms governing microorganism-plant interactions is vital for optimizing crop productivity in a rapidly changing world. For instance, exploring the organic exudates utilized during host signaling in the ‘cry for help’ process, studying interactions within the plant endosphere, and unraveling the assembly mechanisms of beneficial microbiota can pave the way for the development of more resilient bioinoculants suitable for agroecosystems.

The original contributions presented in the study are included in the article/ Supplementary files , further inquiries can be directed to the corresponding author/s.

RR, PB and PD wrote the main manuscript text. GL analyzed the metagenomic data. PB, VC, PD, and LH critically revised the manuscript and approved the final version. All authors contributed to the article and approved the submitted version.