Soil storage temperature and air-drying did not significantly change bacterial taxa in the short-term

Soil sampling

Soil samples for this study were collected in June 2023 from Camden farm, Sydney, NSW, Australia. The site is managed under a wheat cultivation system. Within a 1 km radius, soil samples representing three soil textures—sandy, silty loam, and clay loam—were collected (Schad, 2023). The sample collection was conducted using a sterilised hand trowel, disinfected with 80% alcohol before each use. About 1 kg of topsoil (0–10 cm depth) was collected at each site, subsequently stored in transparent, sealed bags, and transported to the laboratory, 1 h’s drive away. Bulk soil was collected, and visible plant tissues and roots were removed. Upon laboratory arrival, samples were homogenised and subset into four subsamples for the different storage treatments: frozen at −20 °C, refrigerated at 4 °C, room temperature, and air-drying. For the room temperature samples, the samples were sealed with Parafilm to maintain the moisture. DNA was extracted on days 1, 3, 7, 14, and 28, except for samples frozen at −20 °C, which were extracted only on days 1 and 28.

The fundamental physical and chemical soil properties were quantified following our previous protocol (Du, Minasny & Rabbi, 2024). Total carbon and total nitrogen were determined via dry combustion using a LECO CN analyser (LECO Corporation, St. Joseph, MI, USA). Electrical conductivity was assessed using a 1:5 soil-to-deionized water ratio. Soil pH was measured with 25 ml of 0.01 M CaCl₂ solution. Additionally, soil particle size distribution was analysed using the hydrometer method. As summarised in Table 1, silty loam exhibited the highest TC (2.47%), exceeding sandy (1.47%) and clay loam (1.22%) soils. Clay loam showed the greatest EC and pH (266 µS cm⁻¹ and 6.26, respectively).

Soil DNA sequencing and analysis

Soil DNA was extracted using Qiagen DNeasy PowerSoil Pro Kit (Qiagen, Hilden, Germany) following the manufacturer’s standard procedures. The purified DNA was assessed for yield and quality using the Quantifluor dsDNA system. Three replicate DNA samples were extracted per soil sample, and these replicates were pooled prior to sequencing. Metabarcoding of bacterial communities was conducted using primers 341F and 805R (Takahashi et al., 2014). A two-step PCR protocol was used to generate dual-indexed amplicons adapted from the Illumina protocol for 16S Metagenomic Sequencing Library Preparation. The concentration of PCR amplicons was then measured by fluorimetry using the Quantifluor dsDNA system. Amplicons were then pooled at equimolar concentrations, purified using SPRI beads normalised to a concentration of 10 nM and sequenced by the IMB Sequencing Facility at the University of Queensland on an Illumina MiSeq (2 × 300 bp).

Bioinformatics

The bioinformatics workflow was carried out using the LatchBio workflow. Cutadapt was used to remove primer sequences for pair-end reads (Martin, 2011). All subsequent processing and analysis were carried out in R version 4.0 (R Core Team, 2020). Amplicon sequence variants (ASVs) and taxonomic inference were conducted through DADA2 package (Callahan et al., 2016a). During quality filtering in DADA2, forward and reverse reads were truncated to lengths of 270 and 240 bp, respectively. Following truncation, reads exceeding maximum expected error thresholds of 2 (forward) and 3 (reverse) were also discarded. The ‘Pseudo Pooling’ method was used for the main ‘dada’ function. Chimeras were identified and removed with the ‘consensus’ method of the “removeBimeraDenovo” function. The naïve Bayesian Classifier was used to assign taxonomy to genus level with the reference to Silva 138 database (Quast et al., 2012) using the “assignTaxonomy” function of DADA2.

We filtered out ASVs detected with fewer than 15 reads across all samples and found in fewer than two samples, minimising potential biases (Du et al., 2025). Then, the ASVs were rarefied to the minimum reads (33,000 reads) to standardise the bacterial community. The distinction of taxa as abundant or rare was determined by their relative abundance. Specifically, ASVs exhibiting a relative abundance exceeding 0.1% were classified as abundant taxa, whereas those below 0.01% were identified as rare taxa (Zheng et al., 2021).

Statistical analysis

Using the ‘vegan’ package (Oksanen et al., 2013), we analysed various alpha diversity indices to discern the impacts of storage methods and duration on bacterial diversity. A linear mixed-effects model with random effects including ‘soil types’ was applied to evaluate the effect of storage methods and time on microbial alpha diversity by the ‘lme4’ package (Bates et al., 2015). We created the phylogenetic tree by ‘phyloseq’ package (Callahan et al., 2016b) and calculated the weighted UniFrac distance for the principal coordinate analysis (PCoA) to show bacterial community differences between storage methods and time by the ‘microeco’ package (Liu et al., 2021). Additionally, a permutational multivariate analysis of variance (PERMANOVA) was conducted to quantify the variances in bacterial distribution due to storage methods and time. All data visualisations were performed using R Studio (version 4.3.2).

Alpha diversity

Through 16S rRNA sequencing, 6,005 ASVs (33,000 sequences per sample after rarefication) were generated in the soil bacterial community analysis. Across the 1-month storage experiment, we did not find a significant difference among the storage methods over time in terms of four alpha diversity metrics for total, abundant, and rare taxa (Table 2). For the total taxa, the mean Shannon index was slightly higher under 4 °C storage compared to the other conditions (Fig. 1A). However, this trend was not consistent across all soil types. Specifically, the increase was observed in silty loam and sandy soil but was absent in the clay loam (Fig. S1). For the abundant and rare species, the Shannon index fluctuated slightly but remained consistent throughout the storage period (Figs. 1B and 1C).

Beta diversity

Based on Weighted UniFrac distances, the four storage treatments did not result in a significant difference in the beta diversity of the total taxa (Fig. 2A). The bacterial community composition also didn’t show significant differences between storage methods and time for the total (Fig. 2B), abundant (Fig. 2C) and rare taxa (Fig. 2D). In the principal coordinate analysis (PCoA), the first two axes explained a substantial portion of the variance for both the total (67.9% and 24.76%) and abundant communities (64.2% and 26.31%). In contrast, the variance explained was considerably lower for the rare taxa (30.52% and 7.42%). Notably, the groups of three soil types were clearly clustered, across total, abundant, and rare taxa, indicating that the four preservation methods effectively maintained the general beta diversity of each soil type.

Community composition

Despite initial differences between the three soil types, their microbial community compositions at the phylum level remained stable over time across all storage methods. This stability was observed for the total (Fig. S2), abundant (Fig. S3), and rare taxa (Fig. S4). Across all communities, Actinobacteria, Proteobacteria, and Acidobacteria were the three most abundant phyla. We also noted a minor decrease in the relative abundance of Acidobacteria among the abundant taxa in the air-dried silty loam and the sandy soil stored at room temperature with field moisture.