Growth of microorganisms in a Martian regolith simulant at reduced water activity

TL;DR

This study shows that some microorganisms can grow in a simulated Martian soil at low water availability, suggesting potential for life on Mars.

Contribution

The research demonstrates microbial growth in a Martian regolith simulant at water activity levels lower than previously thought possible.

Findings

Microbial DNA increased at water activities as low as 0.34, suggesting possible replication.

No detectable microbial growth occurred at a water activity of 0.12 over 60 days.

Statistical analysis confirmed significant differences in DNA yields between low water activity levels.

Abstract

Water activity (aw) quantifies the free water available for microbial growth. At the cellular level, liquid water is paramount for replication and proliferation. Research on Earth-like life suggests that microbial replication is limited by an aw threshold of ≥ 0.585, below which replication ceases. On Mars, liquid water is typically unstable, but gaseous water exchanges between the atmosphere and the upper regolith are substantial. A variety of salts widespread across the Martian surface are capable of hydration and deliquescence, including sulfates that can undergo hydration–dehydration changes when exposed to different levels of aw. This study investigates microbial growth at different aw levels in a commercially available Mojave Mars Simulant 2 (MMS-2), a fine-grade basaltic soil modified with 2-4 wt% calcium sulfate and oxides (Fe₂O₃, SiO₂, MgO, CaO) to mimic Martian regolith composition. Growth was monitored by quantifying extracted deoxyribonucleic acid (DNA) from samples incubated at aw 1, 0.75, 0.65, 0.34 and 0.12 under Earth-like conditions (30 °C, ≈1 bar). At aw = 1, DNA mass from MMS-2 and Bacillus subtilis-spiked MMS-2 peaked on day 15 and day 3, respectively. At lower aw levels (0.75±0.02 and 0.65±0.02), DNA mass reached its peak after 20 and 30 days, respectively. Samples incubated at aw = 0.34±0.02 exhibited reduced DNA yields, with a maximum on day 30, whereas no detectable increase in DNA occurred at aw = 0.12 ± 0.02 over 60 days. Statistical comparisons with the aw = 0.12 control were significant (e.g., aw = 0.34 ± 0.02, cleanroom vs. aw = 0.12 ± 0.02 at day 30: p = 0.0098; Benjamini–Hochberg (BH) False Discovery Rate across day 20, 30, 45: q = 0.0153). These findings suggest that atmospheric water can be adsorbed into regolith grains and salts, supporting microbial persistence and DNA accumulation consistent with possible replication at reduced water activity under Earth ambient conditions.

The online version contains supplementary material available at 10.1038/s41598-026-35595-2.

Figures

Full text

Introduction

Two primary physical parameters that define habitable environments on Earth are temperature and water activity (a_w_)^1,2^. Water activity (a_w_) quantifies the availability of free water for microbial growth. At equilibrium, a_w_ can be derived from the atmospheric relative humidity (RH) relative to liquid water, as a_w_=RH/100^3^. In the microbial realm, most organisms cannot replicate below approximately 0.90 water activity^4^, whereas some extremophilic archaea and bacteria can germinate at approximately 0.60 a_w_^5^. Theoretical survival limits of approximately 0.61 a_w_ have been proposed for bacteria, archaea, and eukaryotes, indicating that similar physicochemical constraints and water-activity thresholds govern all three domains of life^1^. Thus, water’s thermodynamic availability, expressed as a_w_, constrains habitability^6,7^. The lowest documented water activity limit for life on Earth is a_w_= 0.5851 supporting the germination of xerophilic, osmophilic and halophilic fungi, Aspergillus penicillioides in high-glycerol (liquid, but viscous) media^8^. Interestingly, recent studies have shown that in hyper-arid areas of Earth, such as certain sites of the Atacama Desert (e.g., Yungay and Maria-Elena South), where the mean atmospheric RH is 21% and 17% (i.e., a_w_ = 0.21 and 0.17), yet maximum RH values reach 86.8% and 54.7% (a_w_ = 0.868 and 0.547) viable microbes are still present^9^. Another study that investigated the microbial communities inhabiting the arid Namib Desert, identified 89 bacterial strains that represented 20 distinct genera, all exhibiting adaptations to high salinity, low water availability, and elevated temperatures^10^.These findings underscore the importance of investigating indigenous desert microbiomes to explore lower water activity tolerances. For example, a recent study assessed the active microbial community in hyper-arid subsurface habitats of the Atacama Desert by sequencing extracted intracellular deoxyribonucleic acid (DNA)^11^. Expanding the knowledge of the limits of life is essential for exploring the ecosystems of the Earth, understanding the habitability of extraterrestrial environments, and defining the planetary protection protocols that must be implemented in space missions. The COSPAR Planetary Protection panel uses water activity to define Mars’ ‘Special Regions’^12^ and to evaluate Venus’ cloud habitability^13^. For space exploration, a working assumption is that cell replication requires a_w_ > 0.50 and T > 245 K (−28 °C)^12^.

Present-day Mars is a hyper arid environment. The surface is covered by regolith, which consists of an unconsolidated mixture of mineral grains, salts, and dust that is in contact with the atmosphere and can capture water via absorption or adsorption from the near-surface atmosphere. Liquid water is unstable on Mars under most present-day conditions; however hydrated salts or liquid brines can remain transiently stable on a diurnal timescale. Recent studies have highlighted the role of sulfate salts (magnesium, calcium, and iron) and other salts in Mars’ surface diurnal and seasonal water cycles^14^. On Mars, there is a substantial interchange of water (in gas form) between the atmosphere and the uppermost centimetres of the surface^15^. The recent observations of NASA’s Perseverance and Curiosity rovers on Mars suggest that the water activity at the ground can frequently go above the limit given by the COSPAR Planetary Protection panel for terrestrial-like cell reproduction of 0.5 a_w_, but this happens only at night when the temperature at the surface drops below 190 K (− 83 °C)^15,16^. Therefore, the present-day Mars surface conditions differ significantly from the known, tolerated limits for cell replication on Earth. Experiments with cryptoendolithic microbial communities, in their natural environment on Earth, have shown that they can adapt to water availability by adsorbing water vapor through physical mechanisms, independent of biological activity, within porous sandstone rocks^17^. Additional research has examined the survivability of organisms in rock simulants by spiking suitable organisms^18–21^. A recent study with spiked Martian regolith analogues demonstrated for the first time that a closed deliquescence system of 98% RH can rehydrate the soil enough to reactivate methanogenic archaea^22^. However, no laboratory studies have explored soil microbial growth at low a_w_ values when water is supplied solely via the atmosphere. The purpose of this research is to investigate if indigenous microorganisms in a Mojave derived Martian regolith simulant can tolerate extremely reduced water activity levels when incubated under Earth-like hyper arid conditions. The Mojave Mars Simulant 2 (MMS-2) soil is a commercially available, super-fine- grade soil produced by crushing basaltic rocks to the size of Martian regolith (to produce MMS-1 regolith)^23^ and adding between 2 and 4 wt% gypsum (i.e. calcium sulfate dihydrate), iron(III) oxide, silicon dioxide, magnesium oxide and calcium oxide^24,25^ to mimic the composition of Martian regolith. The Mojave soil harbours a natural microbiome, some of which has been identified^26,27^. In our previous work with MMS-2, we found a few natural soil inhabitants, and other airborne microbes that could reproduce at a_w_ = 1 suggesting the potential of this simulant to investigate microbial growth when water is supplied solely as vapor through the atmosphere^28^.

The central hypothesis of our study is that microorganisms inhabiting rock substrates of desertic areas are adapted to extreme aridity and may replicate or persist at lower water activity levels than typically culturable organisms. To test this, we developed a tailored experimental setup for monitoring microbial growth within opaque mineral matrices, where conventional methods are not feasible. By tracking DNA mass over time, we generate growth curves in these challenging substrates, enabling direct quantification of DNA accumulation consistent with possible replication under reduced a_w_ conditions.

Materials and methods

To explore the limits of life on Earth and the possibility of life on Mars, it is essential to advance the state-of-the-art life detection techniques that allow the investigation of cell replication in its natural environmental niche. Most growth experiments are implemented in liquid media, where the cells can be investigated through traditional optical methods (like microscopy or optical density measurements). Other experiments use culture-based techniques, but this method is inappropriate as only < 2% of bacteria are culturable^29^. Optical density (OD) and colony-forming units (CFU) are routinely used to monitor the growth phases of microorganisms. Therefore, the standard growth curve methods limit the study to monitor only those organisms that can be cultivated in broth or agar media. In our experimental setup, OD or CFU are not adequate here because we evaluate growth in an opaque, endolithic matrix (natural MMS-2), and many residents are not readily cultivable on agar. We therefore implemented a culture independent approach: DNA mass quantification. Previous studies assessed growth with culture independent methods, often qPCR on extracted DNA^30–32^.

In our experimental setup, the water reservoir was separated (Fig. 1), and water vapour was delivered via the headspace of each sealed plate to hydrate the regolith. This compact setup allows multiple independent experiments and replicates in parallel. Each Petri dish was double sealed with parafilm and a sterile polyethylene sample bag to maintain constant a_w_ and minimise contamination. Each plate was only opened once for the analysis. We first established the protocol with a controlled case: The regolith plates were kept in an incubator at 30 °C, the optimal growth temperature for Bacillus subtilis (spiked MMS-2 sets) and typical of warm desert ground temperatures^33^. As a positive control, MMS-2 was spiked with B. subtilis chosen for ease of cultivation and spore tolerance to desiccation while vegetative cells replicate only down to ~0.941 a_w_^34^. To quantify DNA mass, we used commercially available fluorescent dyes that bind to double-stranded DNA and Qubit 4.0 fluorometer (detection limit 5 pg µL^-1^). This experimental setup can be broadly used to monitor microbial growth from various desert soils under reduced a_w_ that mimic atmospherically supplied water.

Fig. 1Experimental set-up of the MMS-2 incubation. (a) One side of a two compartment Petri dish was filled with 1 g of heat-treated MMS-2; the other side contained 10 mL of Milli-Q water or a saturated salt solution. (b) Plates were sealed with Parafilm and placed in a sterile bag before incubation at 30 °C in a forced convection oven at ambient pressure. For each condition and sampling day, three replicate plates were opened once and then discarded after DNA extraction.

Demonstration of growth curves based on DNA mass measurements with Bacillus subtilis in liquid media

A control experiment was implemented in liquid media to demonstrate the validity of DNA mass measurements and monitor cell replication over time. We compared the quantifications obtained using a Qubit 4.0 fluorometer with standard measurements such as optical density (OD) and colony-forming units (CFU) count. Bacillus subtilis subsp. subtilis (NCIMB 3610, Scotland) purchased from NCIMB, Aberdeen, was used for this purpose, where the cells were revived and cultured in nutrient broth (701222, Merck, UK). For solid media, 1.5% (w/v) agar (05040, Merck, UK) was used. Fresh cultures were inoculated with overnight culture samples and grown at 30 °C with continuous shaking at 150 rpm. A 1 mL aliquot of the suspension was used to inoculate 100 mL of nutrient broth in triplicate 250 mL conical flasks and was incubated in an orbital shaking incubator at 30 °C, 150 rpm.

Comparison of three methods: (a) The OD of batch cultures was measured at 600 nm in 1 mL aliquots (and, if necessary, diluted in nutrient broth) in 1.5 mL spectrophotometer cuvettes hourly for the first 3 h, then every 30 min thereafter. The specific growth rate was determined by calculating the slope of a semi-logarithmic plot of OD_600_ against time during exponential growth. (b) The CFU count method was also applied at each measurement point where 1 mL of culture was serially diluted either to 1:100, 1:1000 or 1:1000000, depending on the growth stage, from which 100 µL was plated in nutrient agar and incubated overnight at 30 °C. The colony forming units (CFU) were counted, and CFU mL^-1^ was calculated using the formula: (number of colonies*dilution factor)/volume of culture plated (mL ). (c) A separate batch culture was prepared for DNA extractions. Every 1 h or 30 min, 1 mL of culture was removed and aliquoted in a sterile Eppendorf tube (Eppendorf, UK). The tubes were centrifuged at 1800 g for 10 min, the pellet was washed and resuspended in nuclease-free water (NEB, UK). DNA extractions from Bacillus subtilis cultures were performed using Zymo’s Quick-DNA Miniprep Plus kit (D4068, Zymo Research, UK) according to the manufacturer’s instructions. DNA was eluted in 50-100 µL of nuclease-free water and quantified using Qubit 1X dsDNA High Sensitivity (HS) assay kit for Qubit 4.0 fluorometer (ThermoFisher Scientific, UK).

Heat-treatment of MMS-2

To minimise the water activity of the regolith to a minimum (target RH < 10%) and start all replicates from a common baseline with non-detectable DNA, the regolith used for the experiment was exposed to heat. Mojave Mars Simulant 2 (MMS-2) (commercial Martian regolith analogue; The Martian Garden, Austin, TX, USA) was heated at 150 °C for 3.5 h in a hot-air oven in a stainless steel Petri dish with lid. The MMS-2 was placed in a closed steel petri dish and heated in the hot-air oven. For the MgSO₄ amended variant, 5% (w/w) MgSO₄ (746452, Merck, UK) was pre-mixed with the soil in the steel dish before heat treatment. pH was measured (potentiometry) in Milli-Q water at a 1:1 soil: water (w/v) ratio using a pH probe (Atlas Scientific, USA) and benchtop pH meter (Hanna Instruments, HI 2210, UK). Measured in triplicate, the average pH was 10.03 (untreated) and 9.87 (post-heat).

Regolith incubation experimental setup

Heat-treated MMS-2 was weighed, and 1 g was transferred into one side of a two-compartment sterile Petri dish (G635161, Greiner Bio-One, UK); see Fig. 1. The other compartment was filled with 10 mL of Milli-Q water or a saturated salt solution, serving as a water source and humidity buffer to control the water activity. To maintain a stable relative humidity (i.e. water activity), the Petri dishes were sealed with Parafilm, and each of them was kept inside a sterile polyethylene sample bag (129-9816, VWR, UK) as shown in Fig. 1b. For the positive control at a_w_ = 1, Bacillus subtilis cells harvested at OD ~1 (~10^9^ cells mL^-1^) were spiked in MMS-2. Because vegetative B. subtilis does not replicate below ≈0.94 a_w_, we also ran a negative control at a_w_ = 0.80. Before spiking, the cells were centrifuged at 1800 g for 10 min. The pellet was washed twice in nuclease-free water (NEB, UK) and resuspended to 10 µL; the suspension was added to 1 g of the regolith in an Eppendorf tube. The tube was gently vortexed and aseptically transferred to the Petri dish. For the experiments conducted in a standard BSL-2 laboratory, the transfer of soil, salt solutions or water were performed in a laminar flow hood to minimize contamination. For two of the lower water activity cases (a_w_ = 0.34 and 0.12), all experimental preparation and extractions were carried out in an ISO class 5 cleanroom (Supplementary Figure S1)^35^. The sealed plates were incubated at 30 °C. Three replicates were prepared per sampling day for each condition. To monitor the growth of microorganisms in MMS-2 in the presence of additional salts (i.e. along with the standard 2-4% (w/w) CaSO_4_), we pre-treated a set of MMS-2 mixed with 5% (w/w) MgSO_4_ at 150 °C. The salt was aseptically transferred into a sterile falcon tube containing weighed MMS-2 and gently vortexed until a homogenous mixture was obtained. From this mixture, 1 g was filled in one compartment while the other half contained 10 mL of Milli-Q water. Different incubation periods were chosen depending on the growth rate and experiments, and DNA was extracted at distinct moments of the growth phase.

Water activity control and monitoring

In each sealed experimental plate, the regolith reaches equilibrium with the overlying gas above shortly after sealing. The water activity can then be derived from the headspace relative humidity (RH), with respect to liquid water, as a_w_ = RH/100. Saturated salt solutions are routinely employed in biological, chemical, physical, and industrial applications to establish and maintain defined RH levels within sealed systems^36–38^. In such environments, these solutions function as passive humidity buffers, dynamically equilibrating with the surrounding air by absorbing or releasing water vapor to stabilise the RH at a salt-specific, constant value. Saturated salt solutions were prepared at 30 °C, and the equilibrium headspace relative humidity (RH) above the solution was measured. The salt solutions used to control the water activity in the closed system were sodium chloride (2361009, Fisher Scientific, UK), magnesium nitrate hexahydrate (63084, Merck, UK), magnesium chloride (208337, Merck, UK) and lithium chloride (L9650, Merck, UK). Table 1 summarises the minimum amount of salt to be added to 50 g of sterile water to obtain a saturated solution and the resulting stable headspace relative humidity at 30 °C. To measure the constant humidity and temperature, 10 mL of the saturated solution was aseptically poured into a petri dish and sealed using parafilm and a sterile bag (Fig. 1b). The relative humidity (RH) and temperature for each salt solution were measured for up to 60 days every 4 h, using the Maxim Integrated iButtons (DS1923-F5, Digi-Key, UK) data loggers and the software OneViewWire (64bit, version 0.3.19.47). In one experimental case, MMS-2 soil was supplemented with 5% MgSO₄ (w/w), with pure water serving as the source. In this experiment, the water-absorbing properties of the salt led to an equilibrium relative humidity (RH) of 96% (a_w_ = 0.96) (Fig. 2). This setup also caused the soil to absorb a significant amount of water, which was subsequently quantified. In all other experiments, the soil consisted of pure MMS-2. The iButtons were taped on the inner part of the lid using double-sided tape so that the sensor membrane faced the solution and regolith. To ensure that iButtons and tape did not introduce contamination or bias into the soil cultivation experiments, only one of the three replicates in each set contained an iButton. At the end of each experimental period, DNA extraction was performed on all three replicates, and the results were found to be comparable. Additionally, two separate experiments were conducted solely to monitor relative humidity (RH) and temperature (T) using iButtons; however, DNA extraction was not performed in these cases. RH and T measurements were taken in triplicate, the average RH is reported to calculate the a_w_. The standard deviation for all cases was ±2% RH (±0.02 a_w_). At the end of each incubation period, the plates with regolith were weighed to record the water absorption. In the experiments with pure MMS-2, only the tests with a_w_ = 1 showed a measurable water-induced weight increase. This water uptake can be attributed to the hydration of calcium sulfate, from anhydrite to gypsum.

Table 1. Saturated salt solution concentration required for 50 g of H_2_O at 30 °C.SaltsSaturation salt mass (per 50 g of H_2_O)Equilibrium relative humidity (30 °C)NaCl18 g75%Mg(NO_3_)2.6H_2_O40 g65%MgCl_2_28 g34%LiCl42.5 g12%

Fig. 2Headspace relative humidity (RH) at 30 °C in sealed two compartment plates. RH (%) of MMS-2 plates over 60 days in a closed setup (iButton resolution 0.04% RH). Pure water and saturated salts imposed stable RH at 30 °C: water 100%, NaCl 75%, Mg(NO₃)₂·6H₂O 65%, MgCl₂ 34%, LiCl 12%. In MMS-2 amended with 5% (w/w) MgSO₄ with Milli-Q water as the source, the headspace stabilised at ~96% RH (a_w_ ≈ 0.96) and the soil absorbed substantially more water. Values are means of three replicates; SD was ±2% RH for all cases.

We conducted an additional control experiment to confirm that the iButtons’ readings of headspace humidity are in equilibrium with the relative humidity (RH) within the MMS-2 regolith. In this configuration, one iButton sensor was placed within the plate, facing the salt solution, to measure the headspace RH above the liquid brine pool, while a second iButton sensor was placed in direct contact with the soil to measure the RH of the regolith grains (Supplementary Figure S2-a). Relative Humidity (RH) measurements were taken in triplicate, with a standard deviation of ±2%. For pure soil samples exposed to saturated saline solutions of MgCl₂ (headspace average RH 34%) and LiCl (headspace average RH 12%), the measured RH values in direct contact with the soil were 34% and 11%, respectively (Supplementary Figure S2-d).

MMS-2 DNA extractions

DNA was extracted during distinct stages of the regolith cultivation experiment, and the amount of DNA mass was measured using a Qubit 4.0 fluorometer. All MMS-2 extractions were obtained using DNeasy powersoil pro kit (47014, Qiagen, UK). Three replicates were analysed per sampling day. Approximately 500 mg (±20 mg) sample of the Martian simulant from the incubated petri dishes was aseptically weighed and directly transferred into the PowerBead tubes. The tubes were then filled with 800 µL of the C1 lysis buffer and were briefly vortexed and incubated at 65 °C for 10 min. The Powerbead tubes were secured to a bead beater (BeadBug™ Microtube homogeniser) and homogenised at 2500 RPM for two cycles, 30 s each, with a 30 s break between each cycle. The tubes were centrifuged at 15,000 g for 2 min, and 400–500 µL of the supernatant was carefully transferred to a centrifuge tube using sterile 100–200 µL filter tips for maximum recovery. Hereafter, the extraction procedure followed the manufacturer’s instructions, and the final DNA was eluted in 40 µL of nuclease-free water (NEB, UK). The DNA was quantified using a Qubit 1X dsDNA High Sensitivity (HS) assay kit (Q33230, ThermoFisher Scientific, UK) for Qubit 4.0 fluorometer (ThermoFisher Scientific, UK), which has a lower limit of DNA detection of 5 pg µL^-1^. The protocol was validated previously in our study^28^. From the eluted solution, 5 µL of extracted DNA was used as input for each sample type consistently across each replicate. We report total DNA mass extracted from 500 ± 20 mg of regolith as our primary measurement, rather than mass–normalised DNA yield as accurately removing excess beads and clumps to achieve exactly 0.50 g is impractical in the matrix. Normalised total DNA yield (ng g⁻¹) and its propagated error are provided in the Supplementary tables (S1–S7b). If \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:y=\frac{D}{m},\:$$\end{document} with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:m=0.50\pm\:0.02$$\end{document} g and s.e.(D) from triplicates, then \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{s}.\mathrm{e}.\left(y\right)\approx\:\sqrt{(\mathrm{s}.\mathrm{e}.(D)/m{)}^{2}+(D\text{}{\Delta\:}m/{m}^{2}{)}^{2}}$$\end{document} .

Statistical analysis

For each water activity regime, we compared the peakday DNA mass (n = 3) to the a_w_ = 0.12 negative control (n = 3) using two-tailed Welch’s t-tests (unequal variances)^39^. Because all control replicates were zero, this is equivalent to a one sample t-test of the treatment mean against 0 with df = 2; both formulations yield identical \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:t\:$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:p$$\end{document} . To assess within condition temporal change, we applied the Kruskal–Wallis test across all sampling days (Day 0–60) and report the p-values (Supplementary tables S1-S7b) The Kruskal-Wallis test is a non-parametric method that compares the distributions of multiple groups making it well suited for our dataset, which includes small sample sizes, non-normal distributions, and many zero values^40^. For the a_w_= 0.34 (cleanroom) vs. a_w_ = 0.12 (cleanroom) contrasts at Days 20, 30, 45, we applied the Benjamini–Hochberg False Discovery Rate procedure (α = 0.05). This method adjusts p-values to control the expected proportion of false positives among the results identified as significant when performing multiple comparisons^41^. The three raw pvalues: p(20), p(30), p(45), were sorted in ascending order p(1) ≤ p(2) ≤ p(3). For each ordered p(i), we computed q(i) = min \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left(m/i)\cdot\:p(i),q(i+1\right)$$\end{document} , with m = 3 tests and q(3) = p(3). The reported q-values were then mapped back to the original days (20, 30, 45) and are listed in Table 2. Unless noted, tests are two-tailed; we report means ± s.e. in figures/tables. Raw replicate values and ng g⁻¹ with propagated error (from 500 ± 20 mg) are provided in supplementary tables S1–S8; the day wise cleanroom comparisons appear in Table 2. Days where both groups were identically zero were marked n/a (test not applicable).

Fluorescence microscopy

Fluorescence microscopy was used to visualise the cells in the regolith under different incubation conditions. Samples of 500 mg aseptically transferred to a 2 mL Eppendorf tube, to which 600 µL of nuclease-free water (NEB, UK) was added and vortexed. The tubes were centrifuged at 15,000 g for 2 min, and the supernatant was transferred to a fresh tube. To this, Hoechst 33342 staining solution (HB9888, Hello Bio, UK) was added to make the final concentration of the dye to 5 µM^42^ and incubated in the dark for 15–20 min at room temperature. The sample tubes were centrifuged at 1800 g for 10 min. Retaining ~100 µL of the pellet, the rest was discarded, and fresh 100 µL of nuclease-free water was added to the tube. The total 200 µL of stained sample was transferred to a flat black 24 well plate (82426, ibidi, Germany) and imaged using Zeiss LSM880 Confocal Microscope (University of Aberdeen, UK) in both bright field and 488 nm excitation (Alexa Fluor 488 channel) for the fluorescence.

Sequencing

The microorganisms in the soils were investigated phylogenetically. For this, the library for nanopore sequencing of all the samples was prepared using Ligation Sequencing Kit V14 (SQK-LSK114, Oxford Nanopore Technologies), which is compatible with their new chemistry R10.4.1 flongle (FLO-FLG114) flow cells. The gDNA for the nanopore library was prepared using nuclease-free water and was diluted in 25 μL for library preparation. The preparation steps were followed according to the manufacturer’s instructions, except the volume for all the steps was halved as the samples were sequenced using a Flongle Flow Cell (R10.4.1) with a minimum of 50 or above active nanopores. As a negative control, a library with just nuclease-free water was prepared strictly following the protocol with all steps included. ONT’s Flongle Flow Cells require 125 μL priming mix and 30 μL gDNA library. It is important to note that the operator’s pipetting technique will impact the accuracy while handling such low concentrations and may affect the result. At every library preparation stage, the specified reagents were transferred in an Eppendorf tube before adding the sample containing gDNA to avoid losses due to pipetting. The bases were called for 20–24 h using MinKNOW GUI software (ONT, V5.3.6) with a Qscore of 5. The passed Fastq files were analysed using EPI2ME Desktop Agent software (ONT, V3.5.7) with What’s In My Pot (WIMP) workflow that can produce a real-time taxonomic classification. WIMP is a workflow that classifies each sequence in the MinION™ FASTQ file uploaded by the Desktop Agent™^43^. The Centrifuge classification engine assigned each sequence to a taxon based on the NCBI RefSeq database. The library preparation protocol and sequence analysis were previously validated in our work^28^.

Results

Demonstration of growth curves based on DNA mass measurements with Bacillus subtilis

This preliminary experiment aims to demonstrate the reliability of DNA extraction kits and Qubit fluorometer in monitoring cell growth. Using Bacillus subtilis as a positive control organism, we obtained growth curves (in liquid broth at 30 °C, 150 rpm) quantifying the OD_600_, CFU mL^-1^ and total extracted DNA. All data followed the typical bacterial growth phases (Fig. 3), reaching the stationary phase after 6.5 h. The exponential phase started 2.5 h after the initial inoculum and incubation. To calculate growth rates (r), log2(OD) vs. time was plotted (Supplementary Figure S3), and the exponential slope (r) was calculated using the equation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:r = \frac{{log2\:(OD_{2} /OD_{1} )}}{{{\mathrm{T}}2 - {\mathrm{T}}1}}$$\end{document} . The doubling time (d) of cells was obtained using \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:d=\frac{ln2}{\mathrm{l}\mathrm{n}(1+r)}$$\end{document} , and the OD, the CFU and total extracted DNA curves resulted in a doubling time of 42.5 min, 28.1 min and 41.81 min, respectively. This confirms that DNA extraction and Qubit DNA mass quantification are viable methods to monitor cell replication and colony growth.

Fig. 3Bacillus subtilis growth curve in nutrient-rich media cultivated at 30 °C, 150 rpm. (Left) Extracted DNA and optical density (OD_600_) growth curves in liquid media. (Right) Extracted DNA and colony forming units (CFU mL^-1^), in agar plates. All three curves had a consistent log phase, beginning at the 2nd hour and reaching stationary phase 6 h later. Each data point represents an average of 3 replicates, and the standard error bars are indicated.

Positive control: Bacillus subtilis growth curve in spiked-MMS-2 soil at aw = 1 and aw = 0.8

A control case with B. subtilis spiked in MMS-2 was set up to monitor the growth of microorganisms in the regolith. Since the lowest value of water activity tolerated by Bacillus subtilis to reproduce is 0.941^34^, this test was implemented only at an RH value of 100% (a_w_ = 1) and, as a negative control, at a_w_ = 0.8. The results were compared with those obtained for the MMS-2 soil at a_w_ = 1 without spiking. Day 0 values were obtained by extracting DNA from spiked and non-spiked MMS-2 samples, resulting in a total of 0.91 ng and 0 ng (too low to be detected by ubit 4.0), respectively (Supplementary tables S1, S4). From Fig. 4, we can see the growth of Bacillus subtilis in MMS-2 reaching exponential growth within 24 h and starting to decline after 3 days, where the total extracted DNA dropped from 8 ng to 2 ng. On the 15^th^ day, only a total of 0.4 ng of DNA was obtained. Similarly, a stationary phase was observed in non-spiked MMS-2 after 7 days, where the total extracted DNA was about 5-6 ng starting from no detectable DNA on day 0, with a short lag phase of 1 day. Although the spiked soils contained a natural MMS-2 microbiome, the growth was predominantly dominated by Bacillus subtilis, as confirmed by our sequencing analysis (Supplementary figure S5). The growth of the natural microbiome in MMS-2 is slower than that of Bacillus subtilis in MMS-2 as the maximum growth was observed on day 15. The relative humidity throughout the set-up was constantly between 99% and 101% (Fig. 2), and the temperature was 30 °C. The negative control, conducted at lower water activity level, with a_w_ = 0.8, confirms the absence of B. subtilis growth in MMS-2 beyond 24 h (Supplementary table S2). Kruskal–Wallis tests indicated significant temporal change in DNA mass within each group (Supplementary tables S1–S2).

Fig. 4DNA growth curve of Bacillus subtilis** in spiked-MMS-2 vs. Natural microbiome in MMS-2**. Curves show total extracted DNA from 500 ± 20 mg of regolith. Spiked MMS-2 began at 0.91 ng and peaked on Day 3, while non-spiked MMS-2 began below detection and peaked on Day 15. DNA from B. subtilis spiked MMS-2 declined to near baseline by Day 15. Spiked MMS-2 at a_w_ = 0.80 showed no growth beyond Day 1. Curves are means of three replicates; shaded areas show standard error. Mass normalised yields (ng g⁻¹) are reported in Supplementary tables S1–S2.

Microbial growth in non-spiked MMS-2 soil at reduced water activity conditions

We next investigated growth in MMS-2 (containing only 2-4% CaSO₄ as included in the commercial formulation) under different a_w_ conditions. On day 0, before set-up, we obtained no detectable DNA for all heat-treated non-spiked MMS-2 samples. As seen before in Fig. 4, after a short lag phase of 1 day, for a_w_=1 (RH=100%), we observed the exponential growth phase on day 3. Now, for 0.75 a_w_, 0.65 a_w_, and 0.34 a_w_, there is a lag phase of 7 days, 15 days and 15 days, respectively (Fig. 5). The maximum growth recorded as per the DNA quantification at 1 a_w_, 0.75 a_w_, 0.65 a_w,_ and 0.34 a_w_ was reached around 15 days, 20 days, and 30 days, respectively, after which the measurements showed a decline. To investigate in depth any potential growth at the lowest water activity range, two extra incubation experiments were conducted in an ISO class 5 cleanroom (Supplementary Figure S1): one at 0.34 a_w_ and one negative control with 0.12 a_w_ (Supplementary Figure S2-b). The maximum total extracted DNA obtained in the laboratory and cleanroom experiments were 6.42 ng (a_w_=1), 5.53 ng (0.75 a_w_), 5.40 ng (0.65 a_w_), was 2.56 ng (0.34 a_w_) and 1.68 ng (clean room, 0.34 a_w_). The experiments incubated at 0.12 a_w_ did not show any DNA over 60 days (Supplementary tables S4 - S8). For all experiments with an a_w_ of 0.65 or higher, detectable DNA remained after 60 days of incubation. In contrast, samples incubated at 0.34 a_w_ showed no detectable DNA on day 60, both in the laboratory and cleanroom analyses. Statistical comparisons were performed using two tailed Welch t-tests (n = 3 vs. n = 3) comparing each condition at its peak day to the a_w_ = 0.12 negative control. Significant differences were observed for each water activity case where the calculated *p-*values were 0.0073, 0.0174, 0.0265, and 0.0460 for a_w_ = 1, 0.75, 0.65 and 0.34 respectively (Fig. 5). For the tests in the clean room, with a_w_= 0.34, the comparison yielded a p-value of 0.0098 indicating the difference is statistically significant with a 99% confidence level. Kruskal–Wallis tests confirmed significant temporal change in DNA mass within each a_w_ group (Supplementary tables S4 - S8), except for a_w_ = 0.12, where no growth was observed. To identify the earliest day at which differences emerge at a_w = 0.34 (cleanroom), we performed day wise Welch contrasts versus the day matched a_w = 0.12 (cleanroom) control and applied Benjamini–Hochberg FDR across days 20, 30, and 45. There was no significant difference through day  20 (p = 0.4226; q = 0.4226), but DNA became detectable by day  30 (p = 0.0098; q = 0.0153) and day 45 (p = 0.0102; q = 0.0153) (Table 2).

Fig. 5DNA growth curve of the natural microbiome in MMS-2 at reduced water activities, from 500 ± 20 mg of regolith. The lag phase significantly increases with a decrease in water activity where at a_w_ = 1, maximum growth was seen on day 15, while at a_w_ = 0.34, maximum growth was obtained on day 30. Quantitatively, the DNA mass yield was reduced by ~3 times at a_w_ = 0.34. There was no detectable DNA for all samples at 0.12 a_w_. The cleanroom data includes 0.34 and 0.12 a_w_. Growth curves are an average of three replicates for each data point, and the shaded area represents the calculated standard error. The* p*-values were obtained using two-tailed t-tests comparing peak DNA concentrations of each sample type to the negative control (n = 3 biological replicates). The total DNA yield (ng g^-1^) is presented in table S4-S7b (supplementary tables).

Table 2. Statistical tests comparing replicate values for each day at 0.34 w with the corresponding day of the negative control at 0.12_w_.Daysa_w_ = 0.34 (cleanroom)a_w_ = 0.12 (cleanroom)p-value (two-tailored)BH_q_R1(ng)R2(ng)R3(ng)R1(ng)R2(ng)R3(ng)0000000n/an/a15000000n/an/a202.864000000.42260.4226301.981.41.660000.00980.0153451.1360.8480.8480000.01020.015360000000n/an/a*p-*values are two-tailed Welch’s t-tests comparing each day at a_w_ = 0.34 (cleanroom) to the corresponding day at a_w_ = 0.12 cleanroom control (n = 3 vs. 3). Because the control values are identically zero, these tests are equivalent to one-sample t-tests vs 0 with df = 2, yielding the same t and p. “n/a” indicates the test is not applicable because both groups were all zeros (zero variance). Benjamini–Hochberg false discovery rate (BH-FDR) correction was applied across the three day-wise tests (Days 20, 30, 45); adjusted q-values are shown.

Microbial growth in non-spiked MMS-2 analogue soil in the presence of 5% MgSO4 (w/w) at aw = 0.96

To compare the effect of adding other additional sulfates on soil’s water retention capabilities and microbiome growth, 5% (w/w) MgSO_4_ was mixed with MMS-2, preheated, and incubated. Depending on the environmental conditions, this salt has the potential to absorb water and even deliquesce and melt into a liquid brine. Figure 6a shows the amount of water captured by the soil at the end of each incubation period. For 1 g of MMS-2 containing 2–4% CaSO₄, included in the commercial formulation a maximum of approximately 0.2 g of water was absorbed. In contrast, supplementing the soil with 5% MgSO₄ increased water absorption to around 0.5 g, equivalent to nearly 50% of the soil’s weight. The absorption of water on the regolith is very different depending on the type of salts and the RH of the atmosphere. This is illustrated in Fig. 6b. We observed a drop in humidity from saturation to 90% RH in the first few days and stabilised to 96% RH after 10 days, the equilibrium RH for this salt concentration (Fig. 2).

Fig. 6MMS-2 absorption of water with and without 5%**** MgSO4 at 100% RH, from 500 ± 20 mg of regolith. (a) Bar plot indicating the weight of water absorbed at different incubation periods by 1 g of MMS-2 (containing 2–4% CaSO₄ as included in the commercial formulation) with 0 or 5% w/w of MgSO_4_. At the end of 60 days of incubation, ~0.5 g of water was absorbed by the MMS-2 with 5% MgSO_4_, half the weight of the initial soil volume. (b) Images of the petri dishes showing the absorption of water. The pure regolith absorbed measurable amounts of water only at 100% relative humidity (RH). At lower water activity levels, the soil remained dry, with no significant changes in weight observed (c) The lag phase in the presence of 5% MgSO_4_ is about 30 days, and the maximum extracted DNA was half the amount (~3.2 ng) obtained from MMS-2 without any salt (~6 ng). Growth curves are an average of three replicates for each data point, and the shaded area represents the calculated standard error. The total DNA yield (ng g^-1^) is presented in table S3 (supplementary table).

No growth was observed in the three replicates for almost 45 days. Figure 6c shows that the maximum DNA was obtained after almost 40 days of incubation and was only ~3.2 ng, lower than the 5-6 ng of maximum DNA obtained in our previous experiments (Supplementary table S3). One of the replicates had no detectable DNA throughout the incubation period. We have sequenced the DNA extracted from one of the replicates that showed DNA growth (Supplementary figure S5). Statistical analysis using the Kruskal–Wallis test showed significant differences (Supplementary table S3). A summary of maximum extracted DNA under different water activities incubated at 30 °C is presented in Table 3, while the individual replicates data is given out in supplementary information (table S1 -S8).

Table 3. Summary of maximum extracted DNA under different water activities incubated at 30 °C.SampleSalt/water activity (a_w_)Total incubation time (days)Average peak DNA mass (ng)Peak dayB subtilis spiked MMS-2Milli-Q water: 1 aw158.0933B subtilis spiked MMS-2NaCl: 0.8 aw151.0501Non-spiked MMS-2Milli-Q water: 1 aw706.42615Non-spiked MMS-2 + 5% MgSO_4 (w/w)Milli-Q water: 0.96 aw703.2045Non-spiked MMS-2NaCl: 0.75 aw605.53320Non-spiked MMS-2Mg(NO_3)2.6H_2_O: 0.65 aw605.40630Non-spiked MMS-2MgCl_2: 0.34 aw602.10130Non-spiked MMS-2 (Cleanroom)MgCl_2: 0.34 a_w601.68030Non-spiked MMS-2 (Cleanroom)LiCl: 0.12 aw_600–

Visualisation of the cells at reduced water activity levels

To confirm the presence of active cells in MMS-2, the cells were harvested from the regolith during their log phase and dyed with Hoechst 33342, which binds to extracellular and intracellular DNA by penetrating the cell membranes, see Fig. 7. Each representative image captured an area of 1.7 × 10⁻⁴ cm². Post heat treatment, we observed very few intact cells (0 to 2 per area) and some e-DNA within the soil particles, as seen in the ‘Day 0’ image. At water activity levels 1 and 0.75, the cells were harvested from day 10 and day 15, respectively. At 1 a_w_, at least 10 to 15 cells were in each area, and at 0.75 a_w_, we observed 8-12 cells. For water activity 0.65, the cells from the regolith were harvested on day 25; we observed fewer cells (2-8 per area) that varied significantly in size. It is essential to note the presence of a few intact cells at 0.34 a_w_ (1-3 per area), though much smaller than other water activity levels. We also found 1-3 oval and rod-shaped cells per area in the cells extracted from MMS-2 with 5% MgSO_4_, harvested from day 45 (Supplementary Figure S4). Cells appeared to be weaker and smaller. A total of two replicates and two images per each field was taken, and a representative image is shown.

Fig. 7Fluorescence microscopic imaging of the soil samples at different water activities. Day 0 is the soil taken after heat treatment, and the cells at 1 a_w_, 0.75 a_w_, 0.65 a_w_, and 0.34 a_w_ were harvested on day 10, day 15, day 25 and day 30, respectively. The soil supernatant containing the cells was incubated with 5 µM of Hoechst 33,342 dye for 15 min at RT. The images were taken using Zeiss LSM880 Confocal Microscopy at 100x with Brightfield and 488 nm excitation (Alexa Fluor 488 channel) and represent two replicates. The images were edited using the Zeiss Zen 3.10 software, and a scale bar of 10 μm is indicated.

Discussion

Our study monitored indigenous microbial growth in MMS-2 regolith under Earth ambient conditions at reduced water activities (a_w_) of 0.75 and 0.65, revealing a progressive increase in DNA concentrations over time. Notably, at a_w_ = 0.34, we observed a substantial increase in DNA levels over 30 days, consistent with DNA accumulation and possible replication. In contrast, the negative control (a_w_ = 0.12) showed no detectable DNA throughout the cultivation period. Statistical analysis confirmed a significant peak day increase in DNA at a_w_ = 0.34 (cleanroom) (p = 0.0098; BHFDR q ≈ 0.0153), indicating that the observed DNA accumulation at this water activity is unlikely to be a chance effect. These findings represent a meaningful experimental advance in astrobiology, offering insights into the limits of terrestrial life and the potential for microbial survival in Mars-analogue environments. DNA mass extracted from 500 mg of regolith (±20 mg) was quantified using Qubit. The total DNA yield (ng g^-1^) and DNA yield error due to ±20 mg of MMS-2 is calculated and presented in table S1-S7b (supplementary tables). The assay kit used for Qubit measurements consists of a fluorescent dye that specifically binds to double-stranded DNA. Fluorescent DNA dyes have been traditionally used in molecular biology as they can detect even picograms of double-stranded DNA^44,45^. The Qubit 1X dsDNA HS assay has a detection limit of 5 pg µL^-1^ and a coefficient of variation of approximately 5–10% for homogeneous liquid samples at concentrations ≤0.1 ng µL^-1^^46^. However, these metrics strictly apply to uniform liquid matrices. In our experiments, the intrinsic heterogeneity of mineral grains, the patchy spatial distribution of their associated microbiomes, and the variable affinities of DNA and extraction reagents for mineral surfaces introduce additional variability beyond that of homogeneous systems. We therefore treat the standard deviation and propagated standard error from triplicate measurements as the most appropriate representation of experimental uncertainty.

Our liquid culture control demonstrated that DNA mass measurements reproduce classical growth dynamics obtained from optical density (OD_600_) and CFU mL⁻¹ (Fig. 3). Doubling times derived from DNA, OD_600_, and CFU curves were in close agreement, confirming that DNA mass can serve as a robust proxy for microbial replication, even in opaque matrices that are inaccessible to OD or colony counting. The control experiment with MMS-2 inoculated with Bacillus subtilis demonstrated growth without an extended lag phase at a_w_ = 1 (Fig. 4). In contrast, while the MMS-2 microbiome was able to grow at low a_w_, our experiments with B. subtilis showed no growth at a_w_ = 0.8, as expected (Fig. 4). Although the spiked MMS-2 contained both its natural microbiome and B. subtilis, sequencing results (Supplementary Figure S5) indicate that B. subtilis largely dominated during the growth phase. This dominance is likely due to its faster doubling time compared to the native microbiome during the 10-day incubation period. These observations suggest that pure cultures of well characterised laboratory strains may not be ideal probes of the ultimate low a_w_ limits relevant to Mars. In contrast, natural field microbiomes, already adapted to environmental extremes, may provide more realistic insight into terrestrial extremophiles and potential Martian habitability.

In the Martian soil simulant (MMS-2) experiments, the initial DNA content was not detectable when the regolith was exposed to a heat treatment (Day 0, Figs. 4 and 5, and 6). Then, starting from an undetectable DNA mass in all the experimental cases and replicates, we have monitored the gradual increase in the DNA mass during the cultivation phase. While the initial application of a heat-treatment likely favoured microorganisms capable of surviving heat shock or forming spores, this selection aligns with the experimental goal of focusing on organisms resilient under extreme conditions, including desiccation. Our results suggest DNA accumulation consistent with possible cell replication within a Martian analogue regolith (MMS-2, containing 2–4 % CaSO₄) at a_w_ = 0.34 ± 0.02 and T = 30 °C (RH ≈ 34 ± 2 %). This water activity is notably lower than the previously reported value for glycerol-rich liquid/viscous media of a_w_= 0.585109. Previous experiments with microorganisms under reduced a_w_ have investigated the effect of sugars or salts to reduce the water activity and the impact of this on growth^47–52^. A more recent study of decrease in the assimilation of amino acids at low a_w_ has estimated a putative lowest limit for anabolic active life at a_w_ = 0.540^53^. We emphasise, however, that our measurements are based on DNA mass in a solid regolith under Earth ambient temperature and pressure, not on direct growth in stable liquids; thus we do not propose to redefine canonical global limits, but rather to show that rock associated microbiomes can exhibit DNA based signatures at lower a_w_ than previously demonstrated under these specific conditions.

The different cell densities observed in the fluorescence microscope images (Fig. 7) illustrate the impact of water activity on cell reproduction. We further validated our results by conducting the same experimental setup in an ISO class 5 cleanroom environment, along with a lower water activity negative control (0.12 a_w_). For 0.34 ± 0.02 a_w_, we observed maximum growth after 30 days, whereas no growth was detected throughout the 60 days of incubation at 0.12 ± 0.02 a_w_. The soil RH for these two cases was also monitored over 60 days, with values of approximately 34% and 11%, respectively (Supplementary Figure S2-b). The observed trends of increasing lag phase and decreasing peak DNA with lower water activity (Figs. 5 and 6) provide strong evidence of true microbial growth dynamics. The extended lag reflects the time required for microbes to adapt to stresses, while the subsequent rise in DNA indicates exponential replication, and the later decline corresponds to cell death or dormancy. Together, these patterns demonstrate that water availability not only controls the rate and extent of growth but also reveals the physiological limits of microbial proliferation in MMS-2 soil. Across treatments, we observed a systematic pattern: decreasing a_w_ produced longer lag phases, lower peak DNA, and eventual return to baseline, which together are characteristic of genuine growth - decline dynamics rather than noise.

Although non-replicative mechanisms such as anhydrobiosis, dormancy, or gradual lysis of stressed cells could theoretically contribute to DNA release^54,55^, our experimental results suggest these are unlikely to explain the observed dynamics. In all experiments conducted at reduced water activity (a_w_ ≤ 0.65), DNA concentrations initially fell below the Qubit detection threshold for up to 20 days, even though any DNA from dormant or lysed cells would have been extracted by the same protocol used. Following this initial phase, we consistently observed a lag period whose duration increased as water activity decreased, followed by a smooth, reproducible exponential rise in DNA concentration and a subsequent decline. Notably, at a_w_ = 0.12, DNA remained undetectable throughout the entire cultivation period. If DNA release from dead or dormant cells were significant, it should have been detected under any of these conditions. The reproducible, coordinated DNA trajectories across replicates, experiments, and a_w_ levels are therefore inconsistent with purely stochastic, non-replicative release and instead point to active microbial replication as the main driver. Nevertheless, future work should incorporate ATP assays, RNA-based methods, or isotopic tracers to directly probe metabolic activity and viability at low a_w_. The pH of MMS soils is measured for MMS-1 type grains and has been reported to be 8.86^56^ and our measured pH value for non-treated and pre-heated (150 °C, 3.5 h) MMS-2 was 10.03 and 9.87 respectively. Since DNA can denaturalise within minutes in an alkaline environment^57^, we hypothesise that the alkalinity of the MMS-2 soil may be partially responsible for its degradation during the DNA mass decline phases. Other rock matrices could be more effective in preserving DNA after cell death.

While pure liquid water is unstable on Mars under its present-day conditions, hydrated salts, thin films of water or liquid brines can be transiently stable^14^. Additionally, a diurnal atmospheric-surface water exchange of 0.5–10 g water per m^2^ is observed^16^. At night, when the temperature drops below 190 K, the surface water activity can exceed 0.5, which is the lowest limit for cell reproduction. During the day, when the temperature rises above the cell replication limit of 245 K, water activity falls below 0.02. Thus, water on Mars may be adsorbed onto mineral surfaces and taken up by salts through hydration and deliquescence, and some of this water may persist into the warmer daytime portion of the cycle^15^. In MMS2 supplemented with 5% MgSO₄, we observed strong water uptake (up to ~0.5 g H₂O per 1 g soil, Fig. 6a), forming a visibly wet soil. Headspace RH decreased from saturation and stabilised at ~96 % (Fig. 2), consistent with formation of hydrated MgSO₄ ^54^. Despite this enhanced water availability, no DNA was detected for ~45 days (Fig. 6c), and the eventual peak DNA mass (~3.2 ng) was notably lower than in MMS-2 at 100 % RH. This reduction in both peak DNA and cell size/abundance (Fig. 7, Supplementary Figure S4) highlights how salt composition and a_w_ together shape microbial dynamics. Other hygroscopic salts, including chlorides and perchlorates, can deliquesce at high RH, forming brines with very low a_w_. Implementing hydration–dehydration cycles under Mars like low pressure, CO₂dominated atmospheres, and subfreezing temperatures will be crucial to assess microbial survival and metabolic activity under more realistic conditions.

Future habitability studies with Martian simulants should explore the combined effects of different salt assemblages, grain sizes, and mineralogies. Our procedure cultivates soil microbes within their natural solid matrix, without adding exogenous carbon sources, so the native community relies on regolith organic matter and mineral redox processes. Other regoliths and rocks may differ in organic content, microbiome composition, and redox chemistry, leading to different growth curves and extraction efficiencies. For example, simulants containing higher concentrations of sulfates or perchlorates could affect both water retention and DNA recovery. Consequently, comparisons of habitability across matrices with distinct chemical compositions must be made cautiously, especially when extrapolating from DNA yields alone.

From a planetary protection perspective, these experiments raise the possibility that extremophilic or extremotolerant terrestrial organisms could persist and occasionally replicate under the cold, arid conditions characteristic of Martian rocks, provided sufficient humidity is available. The current COSPAR operational definition of “Special Regions” areas where terrestrial microorganisms might replicate derives largely from terrestrial experiments at ambient pressure and temperature, with a_w_ treated as the primary limiting parameter^12^. Our data, suggesting DNA based evidence for possible replication at a_w_ ≈ 0.34 under Earth ambient P/T, indicate that rock associated desert microbiomes may be more resilient to low a_w_ than previously assumed. We do not claim that global limits for life should be revised solely on this basis, but our results highlight that rock–microbe systems can mobilise and retain enough moisture to cross replication thresholds in niches that would traditionally be considered too dry.

The experiments reported here were conducted at Earth ambient pressures (~1 bar), a temperature of 30 °C, and a standard terrestrial atmosphere (pN₂/pO₂ ≈ 78/21 %). In contrast, the Martian surface experiences pressures of 6–10 mbar, a CO₂ dominated atmosphere with only trace N₂ and O₂, and far lower temperatures. Future work will extend this approach to Mars relevant pressure, temperature, and gas composition, including diurnal thermal and RH cycling, to evaluate the combined effects of these variables on microbial persistence and replication under low a_w_ conditions. Because our experiments were performed under oxygenated conditions, obligate anaerobes would be disadvantaged or excluded; future studies cultivating rocks and soils in CO₂ rich, low O₂ environments, especially using natural samples known (or expected) to contain anaerobes, may further constrain water activity limits for these taxa.

Conclusion

Our results show that DNA-based evidence supports microbial survival and DNA accumulation consistent with possible growth within the MMS-2 Martian regolith simulant under reduced water activity, indicating that terrestrial microorganisms can persist under Mars-analogue humidity conditions. Specifically, indigenous cells in MMS-2 were able to survive and exhibit DNA accumulation at a_w_ as low as 0.34, when incubated under Earth-like conditions within their native rock matrix. These findings extend the known physicochemical limits of life in solid substrates and provide new insight into the potential habitability of hyper-arid extraterrestrial environments. This experimental framework developed here, cultivating regolith microbiomes in-situ and tracking DNA mass, can be combined with complementary techniques (like RNA-analysis, isotopic labelling and ATP quantification) and applied to a variety of Martian-regolith analogues, with different salt compositions. Future studies should explicitly test microbial growth under Mars like pressure, CO₂ rich atmospheres, and cyclic temperature/RH regimes, yielding a more comprehensive picture of Martian habitability. Importantly, our results also inform Planetary Protection strategies by highlighting the resilience of terrestrial microbes under near-Martian humidity, reinforcing the need to carefully assess contamination risks and sterilisation requirements for future missions.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1

Supplementary Material 2

Supplementary Material 3

Supplementary Material 4

Supplementary Material 5

Supplementary Material 6