Soil biogeography at the habitats of origin of major crops

doi:10.21203/rs.3.rs-3547051/v1

Download PDF

Article

Soil biogeography at the habitats of origin of major crops

https://doi.org/10.21203/rs.3.rs-3547051/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

The rewilding of modern agriculture is challenged by our current uncertainties on the microbiome of major crop wild progenitors (CWPs). Here, we conducted the first global standardised field survey to investigate the edaphoclimatic conditions and soil microbiome of 125 wild populations associated with 10 of the most important CWPs at their centres of origin. The wild populations clustered into four ecoregions, ranging from deserts to tropical seasonal forests and savannas, shaped by two edaphoclimatic dimensions that distinguished areas with high soil sand contents and scarce micronutrients from the more fertile ecoregions characterised by variations in aridity, soil pH and carbon storage potential. We identified a common soil core microbiome with differentiated assemblages across ecoregions, driven by varying environmental preferences among soil biodiversity kingdoms, which reflects potential shifts in their functional profiles. The CWPs created unique microhabitats within ecoregions that strongly influenced the soil community assembly, indicating specific co-evolutionary interactions. These insights into the evolutionary origins of domesticated crops hold the potential to advance microbial-assisted breeding and microbiome rewilding of croplands across the globe.

Earth and environmental sciences/Ecology/Biogeography

Biological sciences/Microbiology/Microbial communities

Biological sciences/Evolution/Coevolution

crop wild progenitors

centres of origin

soil core microbiome

rewilding

plant-microbiome co-evolution

The global food system relies on a small set of crops that trace their origins to specific geographic regions around the world ¹. In these native habitats, the Crop Wild Progenitors (CWPs) and their associated soil biotas have historically co-evolved, establishing interactions that influence plant performance, soil microbiome, and ecosystem functioning ². Hereby, the soil microbiome associated with the CWPs of major crops is an untapped resource to meet global agricultural challenges ^3–6. However, today, we lack the most basic understanding on the soil microbiome at the centres of origin of major crops, especially when considering multiple domestication regions and crops simultaneously ^7,8. The processes of plant domestication and farming have altered host-microbial relationships in modern crops ^9–11, undermining the functionality, resilience, and sustainability of croplands ¹². Reinstating part of the ancient biotas of CWPs that support beneficial interactions in modern crops constitutes the foundations of the microbiome rewilding framework ¹³. For this to happen, surveying soil biodiversity and biogeochemistry at the habitats of origin of major crops is a first step.

Soils are the most biodiverse habitat on Earth ¹⁴, and are the principal inoculum source of microbes and invertebrates for the vicinity of and within root tissues ^15–17. Soil and climatic factors are of paramount importance in determining soil community assembly and modulating the host plant control over the soil microbial recruitment ¹⁸. However, studies addressing the microbiome of CWPs do not focus on native soils under the environmental conditions in which truly wild populations thrive ^9,19. Interestingly, one study revealed that certain taxa at native soils of a CWP were absent from cropland soils, with the latter exhibiting increased diversity and simpler network structure, which suggested a homogenised and less specialised soil microbiome ²⁰. Further, agricultural soils favour copiotrophic lifestyles (e.g., bacteria-dominated microbiomes enriched in Actinobacteria and Firmicutes) and less mutualistic microbes, while indigenous depleted taxa are often linked to nutrient mining strategies and to the degradation of complex biopolymers, thereby contributing to nutrient availability and carbon cycling ^10,18,20,21. These findings hold great potential for soil native microbiomes of CWPs but are limited in extent. A comprehensive biogeographical investigation of soil microbiome patterns across multiple CWPs and microbial kingdoms is absent but crucial to develop a baseline for more applied research. This includes investigating the edaphoclimatic niche of the soil microbiome associated with the CWPs of multiple crops, by determining the soil core microbiome and quantifying their environmental correlates. This approach can serve as a baseline for identifying key members of the soil microbiota of CWPs to be reinstated in modern croplands, and may further improve the performance of microbial inoculants and synthetic communities to changing environmental scenarios ¹².

Here, we investigate the biogeography and edaphoclimatic factors associated with the soil core microbiomes (soil microbes and micro-fauna) at the habitats of origin of major crops. We performed a global standardised field survey at 125 wild populations across the distribution ranges of the recognized CWPs of 10 major crops (maize, rice, wheat, potato, bean, little millet, barley, sunflower, soya and cotton; Fig. 1), which account for 40% of the total area annually harvested worldwide and 63% of the global production of primary crops in the period 2012-2021²². Our survey compiles information on climate, potential ecosystem carbon sinks, soil parameters, and the core phylotypes and co-occurrence networks of soil bacteria, fungi, protists, and invertebrates from six domestication centres (North America, Mesoamerica, Central Andes, Fertile Crescent, Indomalaya region, Yangtze and Yellow River valleys). We addressed the following questions: (1) what is the edaphoclimatic niche of CWPs in their native habitats?, (2) is there a common soil core microbiome embracing different CWPs and ecoregions?, and (3) which edaphoclimatic factors determine the soil biodiversity of CWPs? This multi-kingdom characterization of soil biodiversity at the geographical regions where agriculture originated can serve as a foundation for targeted microbiome rewilding, and can further inform integrated soil conservation strategies, from plant genetic and soil diversities, to their co-evolutionary interactions ^8,23.

Our global standardised field survey provides the first joint assessment of the soil core microbiome − bacterial, fungal, protist and invertebrate communities − and edaphoclimatic components of 10 CWPs, investigated across six of the more ancient and relevant crop domestication centres. The wild populations of our CWPs clustered into four ecoregions that differed widely in their edaphoclimatic attributes, ranging from deserts to tropical seasonal forests and savannas. These ecoregions showed substantial differences in terms of climate, soil texture, soil fertility, soil pH and carbon storage potential. We found that the soil core microbiome common to all CWPs were dominated by the major soil phyla commonly found in studies surveying soil ecosystems worldwide. Despite this commonality, we also found distinct ecological groups of soil core phylotypes that showed variation in their prevalence across ecoregions, driven by varying environmental preferences among kingdoms. These biogeographical patterns of the soil core microbiome point to changes in their life history strategies, showing increased proportional abundances of acidophilic bacteria, as well as fungal and protist parasites in the tropical seasonal forest and savanna ecoregion. Intriguingly, wild populations created specific microhabitats within ecoregions that played a selective role in the assembly of the soil core microbiome. Hence, wild populations within CWP species and ecoregions should be considered a priority for the conservation of the specific co-evolutionary relationships between CWPs and their associated microbiome.

Our biogeographical survey encompassed 125 populations of CWPs at the centres of origin of the 10 major crops in terms of yield and cultivated area (Fig. 1). The wild populations were distributed across all Whittaker biomes except cold and high-precipitation temperate systems (Fig. 2a). The CWPs distributed across four different ecoregions based on climate, primary productivity, and soil properties (Fig. 2c, d; Figure S2): 1) deserts, 2) coastal xeric shrublands, 3) temperate dry forests and shrublands (including dry savanna), and 4) tropical seasonal forests and savannas. The desert ecoregion encompassed the wild populations of barley and wheat from the Fertile Crescent (Israel and Iran ²⁴), and some populations of sunflower from North America (USA ²⁵) and soya from China ²⁶. These populations thrived in arid and semi-arid grasslands with mild temperatures and alkaline soils, where rainfall constraint primary productivity and contributes to higher accumulation of total soil micronutrients compared to other ecoregions (Tables S1 and S2). The tropical seasonal forest and savanna ecoregion comprised the wild progenitors of the little millet from the Indomalaya region (India ²⁷) and rice from the Yangtze and Yellow River valleys (China ²⁸), as well as a few of the populations of wild soya. Populations in this ecoregion are found in warm and high-moisture environments, such as forests and wetlands, with acidic soils due to the high rainfall and accumulation of soil organic carbon and inorganic nitrogen. The temperate dry forests and shrublands ecoregion agglutinated the centres of origin of maize and beans from Mesoamerica (Mexico ^29,30), potato from the Central Andes (Bolivia ³¹), and several wild sunflower populations. The temperate ecoregion was climatically intermediate, presented higher soil concentrations of total phosphorus, but overall lower fertility. Finally, the coastal xeric shrubland ecoregion encompassed the wild cotton populations from the Yucatán Peninsula (Mexico ³²), which experienced the warmest temperatures, the lowest micronutrient concentrations, and the sandiest and saltiest soils. A majority of all CWP populations thrived in drylands with low primary productivity (73% with Aridity Index < 0.65, 53% with Mean NDVI < 0.4, Fig. 2b).

Our characterization of the ecoregions of provenance of CWPs provides a quantitative edaphoclimatic basis to the evolutionary origins of domesticated crops ^24,26–32. Our results show that wild populations are not exclusively circumscribed to marginal environments with unfavourable or harsh conditions, as has often been suggested ³. Indeed, the native edaphoclimatic niche of CWPs relies on two key ordination dimensions that are relevant for agriculture. For instance, certain ecoregions exhibit greater soil sand content and overall low micronutrient concentrations, while others thrive in more fertile soils, yet they vary significantly in terms of aridity, soil pH, and carbon storage potential. These unique edaphoclimatic conditions of ecoregions are likely to exert different selective pressures, resulting in specific soil microbiome assemblages and functions ^13,33. Despite the importance of the different edaphoclimatic origins of CWPs for soil biodiversity and plant-microbiome interactions, no previous research has defined the edaphoclimatic niche across the wild progenitors of major crops ^9,10,19,34. Linking the ecoregions of CWPs with their associated soil microbiome will help to evaluate the domestication and cultivation impacts on crop microbiome composition and guide microbial-assisted agriculture.

The core phylotypes (i.e., zOTUs) of the soil biotic communities of ancestral agriculture accounted for a small portion of the total richness (5–10% of the total number of phylotypes), although these phylotypes dominated in abundance (45–74% of the total number of reads, Fig. 3a). This is in line with previous reports on the global distributions of soil organisms ^35–39. After the assignment of the soil core members that encompassed dominant genera of ecological and agricultural relevance (Figure S4-S7), we identified modules of co-occurrent phylotypes for the four kingdoms (Fig. 3b, Table S3). Although the centres of origin may represent a biogeographically biased sample of soils worldwide, our findings regarding the most prevalent and ubiquitous biotic phyla in CWP soils aligns with global surveys (Fig. 3c). Thus, soil bacteria were dominated by Proteobacteria and Actinobacteriota, fungi by Ascomycota, protists by Cercozoa and Ciliophora, and invertebrates by nematodes, as in soil ecosystems around the world ^35–39. Several reasons could explain this resemblance. First, these phyla are cosmopolitan and include members with diverse dispersal abilities, functionalities, and trophic metabolisms that allow them to reach and adapt to diverse ecological niches ^35–39. Moreover, the CWPs were distributed over a wide variety of edaphoclimatic ecoregions, which rules out the existence of a soil microbiome specialised for a very restricted or extreme set of environmental attributes. Despite the extensive dominance of these phyla in the soil core microbiome of CWPs, their effects on host plant and soil functioning may vary due to differences in environmental preferences at finer taxonomic levels, as explained below ³⁵.

The proportion of nodes in each module of the co-occurrence networks in the four biotic groups exhibited overlap across ecoregions (Fig. 4). All kingdoms had a predominant module in the desert ecoregion (accounting for 55 to 72% of total phylotypes, i.e., module completeness). The remaining ecoregions were dominated by a second major module (34–62%), except for the bacterial group, in which desert and temperate ecoregions share the same dominant module. These biogeographical patterns in soil communities may reflect interactions between soil taxa and plants or overlapping environmental preferences ^38,40. In the desert ecoregion the soil core microbiome presented higher proportions of Actinobacteriota; terrestrial fungi from the Basidiomycota phylum; predatory and decomposer soil protists such as Amoebozoa and Ciliophora, and phototrophic protists such as Ochrophyta; as well as soil invertebrates acting as detritivores and predators such as Annelida, Rotifera, and Tardigrada (Fig. 3c and Fig. 4). Knowing the microbiome patterns of the desert ecoregion, where CWPs thrive, can be essential to address one of the most important challenges in dryland agriculture, namely climate change and land degradation ⁴¹. We should explore the functional properties of these soil microbes and how they modify plant evolutionary responses that ameliorate plant drought stress ¹⁷. The application of the adapted soil microbiome is particularly urgent in low-income world regions where crops grow far from their climate optima and in areas expected to surpass aridity thresholds compromising functional and structural attributes of agroecosystems ^42,43. In contrast, the soil microbiome in tropical and temperate ecoregions exhibited greater abundances of Proteobacteria, Verrucomicrobiota, Acidobacteriota and Gemmatimonadota phyla, and other bacteria with predatory behaviour such as Myxococcota; terrestrial saprophytic fungi belonging to Mortierellomycota phylum; invertebrate parasitic protists such as Labyrinthulomycetes, Apicomplexa and kinetoplastids from the Euglenozoa phylum, along with recognized phototrophs such as Chlorophyta; as well as soil-dwelling invertebrates such as Nematodes and Arthropoda (Fig. 3c and Fig. 4). Our results showing increased abundances of fungal and parasites in soil communities from tropical compared to dryland ecoregions, can be attributed to several factors such as the higher temperature and moisture levels, a wider variety of host species, and the longer transmission seasons ³⁸ (Table S1, Figure S8). Overall, these compositional differences in the soil phylotypes pool across ecoregions could distinctly influence the recruitment and colonisation mechanisms of CWPs for a specific endophytic biota. These variations in soil microbial communities can have far-reaching effects on soil functioning, host phenotype and performance ^13,15,44.

We assessed the factors that determine variations in the abundances of the ecological modules within the soil core microbiome of CWPs (Fig. 5). Soil pH was the main driver of the bacterial modules, along with soil salinity and several micronutrients. This suggests that certain bacterial taxa abundant in the tropical and coastal shrubland ecoregions are selected for tolerance to acidic soils (e.g., Acidobacteriota, and Proteobacteria of the Xanthomonadaceae and Burkholderiaceae families) and salinity (e.g., Gammaproteobacteria of the family Alteromonadaceae and Bacteroidota of the families Flavobacteriaceae and Rhodothermaceae) ^37,45. Fungal modules were mainly influenced by sand content, zinc concentration, primary productivity and aridity. Our results confirm that soil-dwelling fungi are more influenced by vegetation and climate factors, while the impact of edaphic characteristics is less relevant ^35,36. These different environmental preferences among fungal modules may be related with variations in fungal functional groups ^35,46. For instance, there was less diversity of fungal parasites, but higher of saprophytic fungi, in areas with higher aridity and lower productivity (Figure S8). Soil protists are distributed as a function of mean annual temperature and, to a lesser extent, soil pH. This indicates differences in the dominant mode of energy acquisition by soil protist communities ³⁸, with increasing temperatures and lower pH favouring greater proportions of parasites (e.g., Labyrinthulomycetes, Apicomplexa and kinetoplastids) in detriment of other free living heterotrophs (e.g., Amoebozoa and Cercozoa). Soil invertebrates were influenced by a higher number of environmental predictors, likely due to complex interactions among trophic levels ⁴⁰. Our results are consistent with those of Bastida et al. (2020) indicating climate, including its seasonal and daily variability, and vegetation productivity as main drivers of soil invertebrate communities. Taken together, we found that the sets of environmental factors shaping the soil microbiome of CWPs varied among kingdoms, which provides a deeper understanding of the complexity of community-level assembly in native soil environments ^3,44.

Interestingly, the identity of CWPs was the major source of variation among ecological modules of the soil core microbiome (i.e., random effects in the models). This result suggests that interactions between plant hosts and the soil microbiome shape module abundances at the local scale ^47,48, where the active selection of distinct soil microbiomes takes place (Figure S3) ⁴⁹. These selection processes could potentially be triggered by plant phenotypes, or these host phenotypes could be the consequence of plant-independent ecological interactions within the native soil microbiome that do not occur in agricultural systems ^13,33. Hence, distinct wild populations within ecoregions are potential reservoirs of the co-evolutionary relationships between CWPs and soil organisms ^2,8,13.

Our study provides a biogeographical baseline for reinstating beneficial ancestral microbes and biotic interactions within the rewilding framework of modern agriculture ^7,13. This work highlights the conservation value of individual CWP populations, since the ancestral co-evolutionary relationships between CWPs and soil organisms might drive specific plant-soil feedbacks and specific responses to biotic and abiotic stressors. For instance, some authors have demonstrated the role of Bacillus sp. as mitigators of water stress in soybean by increasing abscisic acid, modifying root architecture, and reaching increased photosynthesis and radiation use efficiency in plants ^50–52. We argue that the higher presence of Bacillus sp. in soils of the desert ecoregion would help mitigate water stress of wild soya populations compared to those inhabiting tropical ecoregions, which exhibit lower abundance of Bacillus sp and less arid climate. Our results are meaningful to guide soil microbial inoculation since we decipher the environmental conditions that can promote particular modules of the ancestral core microbiome ⁵³. This knowledge will allow the use of specific microbial inoculants adapted to local conditions ensuring their environmental persistence ^53,54. Likewise, we can use microbiome-environment correlates to infer the potential impacts of different agricultural practices on targeted soil microbial phylotypes. Among the soil biodiversity, bacteria are prone to be influenced by the soil chemical legacy or rapid changes in pH and micronutrients. Protists, fungi and invertebrates might be more susceptible to changes in climate and productivity associated with the expansion of the geographical ranges of crops, with different watering programs, and with shifts in soil texture and the set of cultivated species. The current political and economic context seems adequate to harness the potential of the soil microbiome of ancestral agriculture, as the industry of agricultural microbial products is increasing at a global annual rate of 17% ⁵, and recent research has identified global hotspots for soil biodiversity conservation ²³.

In summary, we found that the populations of CWPs were distributed across widely diverse edaphoclimatic conditions that clustered into four ecoregions. The temperate dry forest and shrubland ecoregion and the coastal xeric shrubland ecoregion had sandier soils and lower micronutrient concentrations. In contrast, the desert and tropical seasonal forest and savanna ecoregions generally feature more fertile soils, although they differed significantly in terms of aridity, soil pH, and carbon storage potential. We described the taxonomic structure of soil core microbes and micro-fauna with Proteobacteria, Actinobacteriota, Ascomycota, Cercozoa, Ciliophora and Nematoda as dominant phyla. Further, we identified ecological modules for the four kingdoms of the soil core community, which exhibited distinct taxonomical profiles and also different abundance patterns across ecoregions. The different assemblages across ecoregions could indicate shifts in the life strategies and functional traits of soil communities. For instance, our analyses indicated that soils from tropical regions harboured greater proportions of acidophilic bacteria, as well as fungal and protist parasites, whereas deserts harboured higher abundances of saprophytic fungi and heterotrophic protists. Most importantly, our results suggest that the individual populations of CWPs host unique and diverse soil microbiomes, and thus potentially different microbial functions. This highlights the wild populations as the conservation unit for the preservation of eco-evolutionary relationships between CWPs and their associated microbiome. Our work lays the groundwork for the rewilding of plant-associated biota in agricultural systems.

To investigate the soil biogeography at the habitats of origin of major crops, we sampled 125 wild populations of 10 species of CWPs that encompassed broad geographic and environmental gradients. For each population, we gathered climate, potential ecosystem carbon sinks (i.e., energy availability to trophic networks), soil fertility and physicochemical data. We measured soil bacteria, fungi, protist, and invertebrate communities through amplicon sequencing. We used principal component and module analyses to characterise the sites of origin of CWPs by distinct edaphoclimatic features (i.e., ecoregions). Then, we identified the core members of ancestral agriculture using relevance (10% more abundant phylotypes) and prevalence (phylotypes present in the populations of at least 4 CWPs) criteria. We calculated co-occurrence networks to identify the main modules of soil biotic communities and assessed their relative abundance patterns across ecoregions. Finally, we used linear mixed effects models to identify the environmental drivers responsible for clustering patterns of the soil core members of ancestral agriculture.

Crop wild progenitors

We selected 10 major crops of global relevance (maize, rice, wheat, potato, bean, little millet, barley, sunflower, soya and cotton), including a woody crop, a tuberous crop, and several herbaceous crops harvested for their seeds. We used the Crop Origins and Phylo Food database to identify crops´ geographic origin and their most likely wild progenitor (https://rubenmilla.github.io/Crop_Origins_Phylo/, ⁵⁵). For each CWP, we selected 10 to 15 truly wild populations in habitats across its native range of distribution, maximising the environmental variation among populations (altitudinal/latitudinal or other edaphoclimatic gradients). In each population, we surveyed a 20 m × 20 m plot. The mean separation distance between native populations within each species was 364 km (ranging from 1 km to 2265 km). In total, we sampled 125 populations of CWPs across disparate regions of the world (Fig. 1).

Soil sampling

In 2020 and 2021, we collected soil samples from each wild population, focusing on soil communities associated with CWPs. Specifically, we targeted the bulk soil compartment, under the area of influence of CWPs, which harbours the highest biodiversity and stands as the foremost source of plant microbes ¹⁷. We selected five plants in the flowering stage separated by at least 5 m. We uprooted the plants along with an entire soil monolith within a 10-cm radius and to a depth of 20-cm. We collected the soil that was not intimately attached to the roots of the plant (i.e., that was not part of the rhizosphere) ⁵⁶. We combined soil samples from each plot to create a composite and homogeneous sample for each population. We immediately stored the soil samples at 4°C and shipped them to the Universidad Rey Juan Carlos (URJC, Madrid, Spain) for processing and analysis. Once in Spain, we froze a subset of each soil sample at -20°C until DNA extraction. We air dried and stored the remaining of each sample for physicochemical analyses. Little millet samples could not leave the country due to legal constraints and were processed at the University of Hyderabad (Telangana, India) and there extracted for soil DNA. Then, soil physicochemical properties were analysed at The International Crops Research institute for the Semi-arid tropics (ICRISAT, Hyderabad, India), following the same procedures applied to the rest of the CWPs.

Soil physicochemical analyses

We analysed electrical conductivity (EC) and pH in deionized water extracts (suspension ratio of 1:5 and 1:2.5 weight:volume, respectively). We estimated soil texture (percentage of sand) using the Bouyoucos hydrometer method. We determined the total soil carbon and nitrogen concentrations by dry combustion using an elemental analyser (C/N Flash EA 112 Series-Leco Truspec). We determined the soil organic carbon (SOC) by dry combustion in 2N HCl pre-treated samples, except for soil samples of little millet that were analysed by the Walkley-Black method. We converted SOC data obtained with the Walkley-Black to dry combustion SOC using an established conversion factor of 1.15 ⁵⁷. We extracted total concentrations of soil micronutrients (Ca, Na, K, B, Mg, Mn, Cu, Fe and Zn) using wet acid digestion and determined them using inductively coupled plasma optical emission spectrometry (ICAP 6500 DUO ICP-OES spectrometer, Thermo-Scientific). We analysed mineral nitrogen and available phosphorus (in the form of NH₄⁺, NO₃⁻ and PO₄⁻) by ionic exchange chromatography in soil water extracts (Metrosep A Supp 5–250/4.0, Herisau).

Amplicon sequencing

We extracted soil DNA from 0.5 g of a sample using the DNeasy PowerSoil Kit (Qiagen) according to the manufacturer’s instructions. We checked the quality and quantity of DNA using a NanoDrop 2000 (Thermo Fisher Scientific, USA) and Qubit Fluorometer (Thermo Fisher Scientific, USA). We used soil genomic DNA for the biotic community profiling using the Illumina MiSeq platform of the Next-Generation Sequencing Facility at the Western Sydney University (Richmond, NSW, Australia). We used the primer sets 799F /1193R and Euk1391f/EukBr to amplify the bacterial 16S and eukaryotic 18S rRNA genes, respectively ^58–60. We used the primer sets FITS7/ITS4 to amplify the nuclear ribosomal internal transcriber spacer (ITS) ⁶¹.

We trimmed primers from paired raw reads and performed an end-trim (base quality < 26) to increase the rate of successful merge. We merged the resultant paired reads using USEARCH ⁶² and FLASH ⁶³, and discarded sequences of low quality (maximum of expected error of 1.0). We denoised the sequences (error-correction) and clustered them into zOTUs (100% sequence similarity) using UNOISE3 ⁶⁴. We annotated representative sequences against the UNITE database (2022.05) using QIIME2 ⁶⁵ for ITS data, and the SILVA database for 16S and 18S rDNA data, respectively. We further checked unassigned zOTUs against the BLAST database to exclude non-fungi and plant sources, respectively. The 18S sequences assigned to embryophytes, bryophytes, metazoans and vertebrates were discarded. Phylotypes (i.e., zOTUs) were rarefied at 12000, 8000, 5000 and 250 reads per sample for bacteria, fungi, protists and invertebrates, respectively. We explored the potential lifestyle of the soil fungal community using the FungalTraits v1.2 database based on the fungal ITS data ⁶⁶.

Environmental data

We gathered field data (slope, elevation, vegetation type and dominant species) at each population, as well as climate and ecosystem productivity data, for the coordinates of each population, from global databases. We obtained climatic variables [mean annual precipitation (MAP), mean annual temperature (MAT), precipitation seasonality (PSEA), temperature seasonality (TSEA) and mean diurnal range temperature (MDR)] from the WorldClim dataset version 2 ⁶⁷, and the UNEP aridity index (AI = precipitation / potential evapotranspiration, where the higher the index the lower the aridity) from the Global Aridity Index and Potential Evapotranspiration Climate Database version 2 ⁶⁸. We used the Normalised Difference Vegetation Index (NDVI) as a proxy for net aboveground primary productivity, based on values from the MODIS satellite imagery MOD13Q1. We followed a standard procedure for NDVI calculation at a global scale ⁶⁷. We acquired 23 NDVI values per year with a pixel size of 250 × 250 m and used them to calculate annual NDVI data for each year in the period from 2000 to 2021. To do so, we averaged the product values between the date of the minimum NDVI (\(n\)) and the date \(n-1\) of the following year at each site. MODIS data are geometrically and atmospherically corrected, and include a reliability index of data quality based on the environmental conditions in which the data were recorded.

Data analyses

We carried out exploratory data analysis to inspect the structure of missing values (4% of the environmental database). We imputed the multivariate missing data using the mice package and the predictive mean matching model with 999 iterations ⁶⁹. Data imputation did not alter the distribution of original data as shown in density plots (Figure S9).

Edaphoclimatic distribution of CWPs. To outline the bioclimatic characterization of the populations of the CWPs surveyed, we used Whittaker’s biomes classification ⁷⁰, the aridity index and NDVI (an indirect measure of vegetation cover and primary productivity). To evaluate the edaphoclimatic differences among the populations surveyed, we ran a Principal Component Analysis (PCA) on 18 centred and scaled local soil and climate variables. The variables with the highest square cosine were used to interpret the PCA axes (FactoMineR package ⁷¹, Figure S1). To describe ecoregions with similar edaphoclimatic conditions, we performed a module analysis based on the first 5 PCA axes, which captured 74% of the total variance. We multiplied the PCA axes by the square root of their eigenvalues to weight the classification outcome according to their importance. We calculated Euclidean distances to overcome the multicollinearity of the edaphoclimatic variables and exploit their orthogonal properties ⁷². We used the K-means algorithm and the Calinski-Harabasz index to select the optimal number of groups (‘cascadeKM function’, vegan package ⁷³, Figure S2). Clustering results were represented using elliptic shapes based on multivariate t-distribution. To test the differences in the dimensions of the edaphoclimatic space by CWPs, we used the Wilcoxon signed-rank test and linear models with the applied Bonferroni-estimated marginal mean comparisons as post-hoc analyses.

Core microbiome assignment. Given our global focus, we identified the soil core microbiome of ancestral agriculture based on both relevance and prevalence criteria ⁷⁴. First, we kept only the 10% most common phylotypes in terms of average relative abundance. Then, within that subset of commonest phylotypes, we selected phylotypes present in at least four CWPs (i.e., 40% of the species surveyed). We conducted these analyses independently for bacteria, fungi, protists, and invertebrates. The resulting abundance tables contain phylotypes that (1) make a substantial contribution to ecosystem functionality ⁷⁵, (2) are prevalent and thereby adapted to contrasting climates and soils, thus likely providing great versatility for their agricultural application, and (3) reduce potential spurious associations of rare taxa, since zero-inflated databases may confound ecological interpretations in downstream co-occurrence network analyses ^76,77.

Structure of the community: co-occurrence network analyses. To explore the structure of the soil core microbiome of ancestral agriculture, we conducted co-occurrence network analyses. We calculated potential co-occurrences between core phylotypes by applying a probabilistic model to presence-absence matrices ^78,79, and obtained a list of pairs of phylotypes that co-occurred more than expected by chance. We built independent networks for bacteria, fungi, protists and invertebrates, where nodes were phylotypes and edges were significant pairwise co-presence between nodes (p-value < 0.05). We determined modularity and clusters (modules of dense connections that are sparsely connected to other areas in the network) using a community finding algorithm via short random walks (‘module_walktrap function’ from igraph package ⁸⁰). The modules can indicate soil microorganisms that potentially interact or share environmental preferences, although other mechanisms responsible for community assemblies such as dispersion, diversification, and drift could also underlie the network structure ²¹. To investigate whether biogeographical patterns exist in the overall network structures, we computed the proportion of nodes of a given module that were present in wild populations of CWP in each ecoregion (module completeness). Given the unequal number of phylotypes across biotic modules (Table S3), we calculated the standardised relative abundance (z-score, \({z}_{ij}\) in Eq. 1) of phylotypes and averaged them per module and wild population. This standardisation allows to obtain a balanced contribution of the phylotypes (i.e., nodes) to the relative abundances of the modules in the networks for downstream analyses.

\({z}_{ij}=\frac{\left({x}_{ij}-{x̄}_{i}\right)}{{\sigma }_{i}}\) Equation 1

Where \({z}_{ij}\) represents the number of reads for the phylotype \(j\) in the wild population \(i\), \({x̄}_{i}\) is the average number of reads for wild population \(i\), and \({\sigma }_{i}\) is the standard deviation of the reads for the wild population \(i\). To test for the differences in the relative abundance of each module across CWPs, we fit linear models and applied Bonferroni-estimated marginal mean comparisons as post-hoc analyses.

Environmental factors determining modules of core members of soil biodiversity. To test if and how the modules in the resulting networks were influenced by environmental factors, we used linear mixed models (LMMs), four groups of predictors (climate: MAT, MDR, PSEA, AI; soil properties: pH, conductivity, sand; soil micronutrient concentrations: NH₄⁺, NO₃⁻, PO₄⁻, Ca, B, K, Mg, Zn, Fe, Mn, Mo; and potential ecosystem carbon sink: mean NDVI, SOC), and the relative abundance (z-score) of the modules as the dependent variable. We retained the above-mentioned subset of variables after checking for multicollinearity patterns in our dataset of environmental predictors by means of the variance inflation factor and using a threshold level of five (regclass package, ⁸¹). To deal with the hierarchical design of the survey, we used the identity of the CWPs as a categorical random intercept in the model. The random factor accounts for local drivers affecting microbial community assembly that were not measured and that may be directly attributed to the identity of the CWP species. We optimised the structure of the fixed components in the LMMs through a backward selection procedure of the predictor variables according to the Akaike Information Criterion (AIC, mass package ⁸²). Model coefficients were estimated using the restricted maximum likelihood criterion (lme4 package, ⁸³) and 95% bootstrap confidence intervals were calculated based on 1000 simulations (jtools package, ⁸⁴). To interpret the outcome of the LMMs, we standardised the predictors and decomposed the marginal coefficient of determination (Rm²) into unique and shared variance by each group of fixed predictors (glmm.hp package, ⁸⁵). We checked LMM assumptions visually and using the Anderson-Darling normality test (nortest package, ⁸⁶). All statistical analyses and visualisations using R version 4.2.4 ⁸⁷.

Acknowledgements

This research was supported by a 2019 Leonardo Grant for Researchers and Cultural Creators, BBVA Foundation, granted to P.G.-P. The BBVA Foundation accepts no responsibility for the opinions, statements and contents included in the project and/or the results thereof, which are entirely the responsibility of the authors. The Spanish Ministry of Science and Innovation also supported this research, with a grant (PID2021-122296NB-100) to R.M. M.J.F.-A. was supported by a ‘María Zambrano’ postdoctoral fellowship (Spanish Ministry of Universities) at the URJC. A.R.P acknowledges the DST, Govt. of India for the award of the JC Bose Fellowship (JCB/2017/000053) and the Ministry of Education, Govt. of India and Institution of Eminence Directorate (IOE), University of Hyderabad (UOH-IOE-RC3-21-065). Special thanks to José María Iriondo Alegría for his advice on methods, concepts and sources of information on crop wild populations.

Contributions

P.G.-P., R.M. and M.D.-B. developed the original idea, designed the research and coordinated all field and laboratory operations, with inputs from B.K.S. Field data were collected by L.A.-R., F.D.A., D.A.-P., M.-K.A., D.C., J.C., G.-L.D., A.H.-L., K.N., S.C.N, B.P. T.S.S.K.P., A.R.P., T.Q.-M., D.S.R., P.V.S.R.N.S, S.S., P.T., Q.Y., Y.Y., E.Z., Y.-G.Z. Data gathering was conducted by M.J.F.-A., M.C., J.G and J.-T.W, with inputs from I.B. Laboratory analyses were carried out by C.G. and J.P. M.J.F.-A., M.C. and J.-T.W analysed the data. The manuscript was written by M.J.F.-A., assisted by M.C., and with contributions from all coauthors.

Additional information

Supplementary information is available for this paper.

Data availability Unpublished raw data in support of these findings will be openly available in a Figshare (https://figshare.com/) or Zenodo (https://zenodo.org/) repository upon acceptance of the manuscript for publication. Data compiled from the literature will not be provided to avoid duplicate deposit of primary data.

Milla, R. Crop Origins and Phylo Food: A database and a phylogenetic tree to stimulate comparative analyses on the origins of food crops. Glob. Ecol. Biogeogr. 29, 606–614 (2020).
Pérez-Jaramillo, J. E., Carrión, V. J., de Hollander, M. & Raaijmakers, J. M. The wild side of plant microbiomes. Microbiome 6, 143 (2018).
Fagorzi, C., Passeri, I., Cangioli, L., Vaccaro, F. & Mengoni, A. When biodiversity preservation meets biotechnology: The challenge of developing synthetic microbiota for resilient sustainable crop production. J. Sustain. Agric. Environ. In press, 1–71 (2023).
Jurburg, S. D. et al. Potential of microbiome-based solutions for agrifood systems. Nat. Food 3, 557–560 (2022).
Singh, B. K., Trivedi, P., Egidi, E., Macdonald, C. A. & Delgado-Baquerizo, M. Crop microbiome and sustainable agriculture. Nat. Rev. Microbiol. 18, 601–602 (2020).
de Vries, F. T., Griffiths, R. I., Knight, C. G., Nicolitch, O. & Williams, A. Harnessing rhizosphere microbiomes for drought-resilient crop production. Science (80-. ). 368, 270–274 (2020).
Pérez-Jaramillo, J. E., Mendes, R. & Raaijmakers, J. M. Impact of plant domestication on rhizosphere microbiome assembly and functions. Plant Mol. Biol. 90, 635–644 (2016).
Chen, Q.-L. et al. Potential of indigenous crop microbiomes for sustainable agriculture. Nat. Food 2, 233–240 (2021).
Brisson, V. L., Schmidt, J. E., Northen, T. R., Vogel, J. P. & Gaudin, A. C. M. Impacts of Maize Domestication and Breeding on Rhizosphere Microbial Community Recruitment from a Nutrient Depleted Agricultural Soil. Sci. Rep. 9, 15611 (2019).
Yue, H. et al. Plant domestication shapes rhizosphere microbiome assembly and metabolic functions. Microbiome 11, 70 (2023).
Coleman‐Derr, D. et al. Plant compartment and biogeography affect microbiome composition in cultivated and native Agave species. New Phytol. 209, 798–811 (2016).
Nerva, L. et al. Breeding toward improved ecological plant–microbiome interactions. Trends Plant Sci. 27, 1134–1143 (2022).
Raaijmakers, J. M. & Kiers, E. T. Rewilding plant microbiomes. Science (80-. ). 378, 599–600 (2022).
Anthony, M. A., Bender, S. F. & van der Heijden, M. G. A. Enumerating soil biodiversity. Proc. Natl. Acad. Sci. 120, 2017 (2023).
Guo, J. et al. Seed-borne, endospheric and rhizospheric core microbiota as predictors of plant functional traits across rice cultivars are dominated by deterministic processes. New Phytol. 230, 2047–2060 (2021).
Vandenkoornhuyse, P., Quaiser, A., Duhamel, M., Le Van, A. & Dufresne, A. The importance of the microbiome of the plant holobiont. New Phytol. 206, 1196–1206 (2015).
Trivedi, P., Leach, J. E., Tringe, S. G., Sa, T. & Singh, B. K. Plant–microbiome interactions: from community assembly to plant health. Nat. Rev. Microbiol. 18, 607–621 (2020).
Escudero-Martinez, C. & Bulgarelli, D. Engineering the Crop Microbiota Through Host Genetics. Annu. Rev. Phytopathol. 61, 257–277 (2023).
Alegria Terrazas, R. et al. A footprint of plant eco-geographic adaptation on the composition of the barley rhizosphere bacterial microbiota. Sci. Rep. 10, 12916 (2020).
Pérez-Jaramillo, J. E. et al. Deciphering rhizosphere microbiome assembly of wild and modern common bean (Phaseolus vulgaris) in native and agricultural soils from Colombia. Microbiome 7, 114 (2019).
Cordovez, V., Dini-Andreote, F., Carrión, V. J. & Raaijmakers, J. M. Ecology and evolution of plant microbiomes. Annu. Rev. Microbiol. 73, 69–88 (2019).
Food and Agriculture Organization of the United Nations (FAO). FAO Global Statistical Yearbook. FAOSTAT Database. https://www.fao.org/faostat/en/#data/QCL.
Guerra, C. A. et al. Global hotspots for soil nature conservation. Nature (2022) doi:10.1038/s41586-022-05292-x.
Brown, T. A., Jones, M. K., Powell, W. & Allaby, R. G. The complex origins of domesticated crops in the Fertile Crescent. Trends Ecol. Evol. 24, 103–109 (2009).
Marek, L. F. Crop Wild Relatives of Sunflower in North America. in North American Crop Wild Relatives, Volume 2 (eds. Greene, S. L., Williams, K. A., Khoury, C. K., Kantar, M. B. & Marek, L. F.) 453–484 (Springer Nature, 2019). doi:10.1007/978-3-319-97121-6_10.
Zhou, Z. et al. Resequencing 302 wild and cultivated accessions identifies genes related to domestication and improvement in soybean. Nat. Biotechnol. 33, 408–414 (2015).
Goron, T. L. & Raizada, M. N. Genetic diversity and genomic resources available for the small millet crops to accelerate a New Green Revolution. Front. Plant Sci. 6, (2015).
Gross, B. L. & Zhao, Z. Archaeological and genetic insights into the origins of domesticated rice. Proc. Natl. Acad. Sci. 111, 6190–6197 (2014).
Matsuoka, Y. et al. A single domestication for maize shown by multilocus microsatellite genotyping. Proc. Natl. Acad. Sci. 99, 6080–6084 (2002).
Kwak, M., Kami, J. A. & Gepts, P. The Putative Mesoamerican Domestication Center of Phaseolus vulgaris Is Located in the Lerma–Santiago Basin of Mexico. Crop Sci. 49, 554–563 (2009).
Spooner, D. M., McLean, K., Ramsay, G., Waugh, R. & Bryan, G. J. A single domestication for potato based on multilocus amplified fragment length polymorphism genotyping. Proc. Natl. Acad. Sci. U. S. A. 102, 14694–14699 (2005).
Vega, M. et al. Multiple domestication events explain the origin of Gossypium hirsutum landraces in Mexico. Ecol. Evol. 13, 1–19 (2023).
Klein, M., Stewart, J. D., Porter, S. S., Weedon, J. T. & Kiers, E. T. Evolution of manipulative microbial behaviors in the rhizosphere. Evol. Appl. 15, 1521–1536 (2022).
Pérez-Jaramillo, J. E. et al. Linking rhizosphere microbiome composition of wild and domesticated Phaseolus vulgaris to genotypic and root phenotypic traits. ISME J. 11, 2244–2257 (2017).
Egidi, E. et al. A few Ascomycota taxa dominate soil fungal communities worldwide. Nat. Commun. 10, 2369 (2019).
Tedersoo, L. et al. Global diversity and geography of soil fungi. Science (80-. ). 346, (2014).
Delgado-Baquerizo, M. et al. A global atlas of the dominant bacteria found in soil. Science 359, 320–325 (2018).
Oliverio, A. M. et al. The global-scale distributions of soil protists and their contributions to belowground systems. Sci. Adv. 6, 1–11 (2020).
Bastida, F. et al. Climatic vulnerabilities and ecological preferences of soil invertebrates across biomes. Mol. Ecol. 29, 752–761 (2020).
Potapov, A., Lindo, Z., Buchkowski, R. & Geisen, S. Multiple dimensions of soil food-web research: History and prospects. Eur. J. Soil Biol. 117, 103494 (2023).
World Atlas of Desertification. (Publication Office of the European Union, 2018). doi:10.2760/9205.
Mahaut, L. et al. Matches and mismatches between the global distribution of major food crops and climate suitability. Proc. R. Soc. B Biol. Sci. 289, (2022).
Berdugo, M. et al. Global ecosystem thresholds driven by aridity. Science 367, 787–790 (2020).
Trivedi, P., Mattupalli, C., Eversole, K. & Leach, J. E. Enabling sustainable agriculture through understanding and enhancement of microbiomes. New Phytol. 230, 2129–2147 (2021).
Rath, K. M., Fierer, N., Murphy, D. V & Rousk, J. Linking bacterial community composition to soil salinity along environmental gradients. ISME J. 836–846 (2019) doi:10.1038/s41396-018-0313-8.
Liu, S. et al. Phylotype diversity within soil fungal functional groups drives ecosystem stability. Nat. Ecol. Evol. (2022) doi:10.1038/s41559-022-01756-5.
Kivlin, S. N., Hawkes, C. V. & Treseder, K. K. Global diversity and distribution of arbuscular mycorrhizal fungi. Soil Biol. Biochem. 43, 2294–2303 (2011).
Miller, E. T., Svanbäck, R. & Bohannan, B. J. M. Microbiomes as Metacommunities: Understanding Host-Associated Microbes through Metacommunity Ecology. Trends Ecol. Evol. 33, 926–935 (2018).
Edwards, J. et al. Structure, variation, and assembly of the root-associated microbiomes of rice. Proc. Natl. Acad. Sci. 112, E911–E920 (2015).
Mondani, F., Khani, K., Honarmand, S. J. & Saeidi, M. Evaluating effects of plant growth-promoting rhizobacteria on the radiation use efficiency and yield of soybean (Glycine max) under water deficit stress condition. Agric. Water Manag. 213, 707–713 (2019).
Prudent, M., Salon, C., Souleimanov, A., Emery, R. J. N. & Smith, D. L. Soybean is less impacted by water stress using Bradyrhizobium japonicum and thuricin-17 from Bacillus thuringiensis. Agron. Sustain. Dev. 35, 749–757 (2015).
Naylor, D. & Coleman-Derr, D. Drought Stress and Root-Associated Bacterial Communities. Front. Plant Sci. 8, 1–16 (2018).
French, E., Kaplan, I., Iyer-Pascuzzi, A., Nakatsu, C. H. & Enders, L. Emerging strategies for precision microbiome management in diverse agroecosystems. Nat. Plants 7, 256–267 (2021).
Hartmann, M. & Six, J. Soil structure and microbiome functions in agroecosystems. Nat. Rev. Earth Environ. 4, 4–18 (2022).
Milla, R. Crop Origins and Phylo Food, GitHub repository. Crop Origins and Phylo Food, GitHub repository (2021).
Ramirez, K. S. et al. Range-expansion effects on the belowground plant microbiome. Nat. Ecol. Evol. 3, 604–611 (2019).
Shamrikova, E. V. et al. Transferability between soil organic matter measurement methods for database harmonization. Geoderma 412, (2022).
Amaral-Zettler, L. A., McCliment, E. A., Ducklow, H. W. & Huse, S. M. A Method for Studying Protistan Diversity Using Massively Parallel Sequencing of V9 Hypervariable Regions of Small-Subunit Ribosomal RNA Genes. PLoS One 4, e6372 (2009).
Stoeck, T. et al. Multiple marker parallel tag environmental DNA sequencing reveals a highly complex eukaryotic community in marine anoxic water. Mol. Ecol. 19, 21–31 (2010).
Herlemann, D. P. R. et al. Transitions in bacterial communities along the 2000 km salinity gradient of the Baltic Sea. ISME J. 5, 1571–9 (2011).
Ihrmark, K. et al. New primers to amplify the fungal ITS2 region - evaluation by 454-sequencing of artificial and natural communities. FEMS Microbiol. Ecol. 82, 666–677 (2012).
Edgar, R. C. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460–2461 (2010).
Magoč, T. & Salzberg, S. L. FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics 27, 2957–2963 (2011).
Edgar, R. C. UNOISE2: improved error-correction for Illumina 16S and ITS amplicon sequencing. bioRxiv 081257 (2016) doi:10.1101/081257v1.
Bolyen, E. et al. Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nat. Biotechnol. 37, 852–857 (2019).
Põlme, S. et al. FungalTraits: a user-friendly traits database of fungi and fungus-like stramenopiles. Fungal Divers. 105, 1–16 (2020).
Fick, S. E. & Hijmans, R. J. WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. Int. J. Climatol. 37, 4302–4315 (2017).
Trabucco, A. & Zomer, R. Global Aridity Index and Potential Evapotranspiration (ET0) Climate Database v2.figshare. Fileset. (2019).
Buuren, S. van & Groothuis-Oudshoorn, K. mice: Multivariate Imputation by Chained Equations in R. J. Stat. Softw. 45, 1–67 (2011).
Stefan, V. & Levin, S. plotbiomes: Plot Whittaker biomes with ggplot2. R package version 0.0.0.9001. at (2022).
Lê, S., Josse, J. & Husson, F. FactoMineR : An R Package for Multivariate Analysis. J. Stat. Softw. 25, (2008).
Testolin, R., Attorre, F. & Jiménez‐Alfaro, B. Global distribution and bioclimatic characterization of alpine biomes. Ecography (Cop.). 43, 779–788 (2020).
Oksanen, J. et al. vegan: Community Ecology Package. at https://cran.r-project.org/package=vegan (2022).
Custer, G. F., Gans, M., van Diepen, L. T. A., Dini-Andreote, F. & Buerkle, C. A. Comparative Analysis of Core Microbiome Assignments: Implications for Ecological Synthesis. mSystems (2023) doi:10.1128/msystems.01066-22.
Zhang, G. et al. Homogeneous selection shapes rare biosphere in rhizosphere of medicinal plant. Ecol. Indic. 129, 107981 (2021).
Delgado-Baquerizo, M. Simplifying the complexity of the soil microbiome to guide the development of next‐generation SynComs. J. Sustain. Agric. Environ. 1, 9–15 (2022).
Guseva, K. et al. From diversity to complexity: Microbial networks in soils. Soil Biol. Biochem. 169, 108604 (2022).
Veech, J. A. A probabilistic model for analysing species co-occurrence. Glob. Ecol. Biogeogr. 22, 252–260 (2013).
de Celis, M. et al. Niche differentiation drives microbial community assembly and succession in full-scale activated sludge bioreactors. npj Biofilms Microbiomes 8, 1–6 (2022).
Csardi, G. & Nepusz, T. The igraph software package for complex network research. InterJournal Complex Sy, (2006).
Adam Petrie. regclass: Tools for an Introductory Class in Regression and Modeling. at https://cran.r-project.org/package=regclass (2020).
Venables, W. N. & Ripley, B. D. Modern applied statistics with S. (Springer-Verlag, 2002).
Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67, (2015).
Long, J. A. jtools: Analysis and Presentation of Social Scientific Data. R package version 2.2.0. at https://cran.r-project.org/package=jtools (2022).
Lai, J., Zou, Y., Zhang, S., Zhang, X. & Mao, L. glmm.hp: an R package for computing individual effect of predictors in generalized linear mixed models. J. Plant Ecol. 15, 1302–1307 (2022).
Juergen Gross and Uwe Ligges. nortest: Tests for Normality. R package version 1.0-4. at https://cran.r-project.org/package=nortest (2015).
R Core Team. R: A language and environment for statistical computing. (R Foundation for Statistical Computing, 2023).
GBIF.org. GBIF Home Page. www.gbif.org (2023).

There is NO Competing Interest.

Supplementaryinformation.pdf
Supplementary information

Download PDF

Version 1

posted

You are reading this latest preprint version

Soil biogeography at the habitats of origin of major crops

Status:

Version 1

Abstract

Figures

Main

Results and discussion

Methods

Crop wild progenitors

Soil sampling

Soil physicochemical analyses

Amplicon sequencing

Environmental data

Data analyses

Declarations

Acknowledgements

Contributions

Additional information

References

Additional Declarations

Supplementary Files

Status:

Version 1