Biogeographical survey of soil microbiomes across sub-Saharan Africa : structure, drivers, and predicted climate-driven changes
Loading...
Date
Authors
Cowan, Don A.
Lebre, Pedro Humberto
Amon, C.E.R.
Becker, R.W.
Boga, H.I.
Boulange, A.
Chiyaka, T.L.
Coetzee, T.
De Jager, Pieter Christiaan
Dikinya, O.
Journal Title
Journal ISSN
Volume Title
Publisher
BMC
Abstract
BACKGROUND : Top-soil microbiomes make a vital contribution to the Earth’s ecology and harbor an extraordinarily
high biodiversity. They are also key players in many ecosystem services, particularly in arid regions of the globe such
as the African continent. While several recent studies have documented patterns in global soil microbial ecology,
these are largely biased towards widely studied regions and rely on models to interpolate the microbial diversity of
other regions where there is low data coverage. This is the case for sub-Saharan Africa, where the number of regional
microbial studies is very low in comparison to other continents.
RESULTS : The aim of this study was to conduct an extensive biogeographical survey of sub-Saharan Africa’s top-soil
microbiomes, with a specific focus on investigating the environmental drivers of microbial ecology across the region.
In this study, we sampled 810 sample sites across 9 sub-Saharan African countries and used taxonomic barcoding to
profile the microbial ecology of these regions. Our results showed that the sub-Saharan nations included in the study
harbor qualitatively distinguishable soil microbiomes. In addition, using soil chemistry and climatic data extracted
from the same sites, we demonstrated that the top-soil microbiome is shaped by a broad range of environmental
factors, most notably pH, precipitation, and temperature. Through the use of structural equation modeling, we also
developed a model to predict how soil microbial biodiversity in sub-Saharan Africa might be affected by future
climate change scenarios. This model predicted that the soil microbial biodiversity of countries such as Kenya will be
negatively affected by increased temperatures and decreased precipitation, while the fungal biodiversity of Benin will
benefit from the increase in annual precipitation.
CONCLUSION : This study represents the most extensive biogeographical survey of sub-Saharan top-soil microbiomes
to date. Importantly, this study has allowed us to identify countries in sub-Saharan Africa that might be particularly
vulnerable to losses in soil microbial ecology and productivity due to climate change. Considering the reliance of many economies in the region on rain-fed agriculture, this study provides crucial information to support conservation
efforts in the countries that will be most heavily impacted by climate change.
Description
The sequencing data analyzed in this study has been deposited in the SRA
NCBI submission portal (BioProject ID PRJNA807934). The R scripts used for
the analysis of the sequencing data can be found in the GitHub page https://
github. com/ Pedro HLebre/ AfSM_ scripts.
Additional file 1. Figure S1. Distribution of samples across the 9 African countries according to their land cover (LC) classification. Land cover codes used were the following: LC_1 - Rainfed croplands; LC_2 - Mosaic Cropland (50-70%) / Vegetation (grassland, shrubland, forest) (20-50%); LC_3 - Mosaic Vegetation (grassland, shrubland, forest) (50-70%) / Cropland (20-50%); LC_4 - Closed to open (>15%) broadleaved evergreen and/or semi-deciduous forest (>5m); LC_5 - Closed (>40%) broadleaved deciduous forest (>5m); LC_6 - Open (15-40%) broadleaved deciduous forest (>5m); LC_10 - Mosaic Forest/Shrubland (50-70%) / Grassland (20- 50%); LC_11 - Mosaic Grassland (50-70%) / Forest/Shrubland (20-50%); LC_12 - Closed to open (>15%) shrubland (<5m); LC_13 - Closed to open (>15%) grassland; LC_14 - Sparse (>15%) vegetation (woody vegetation, shrubs, grassland); LC_17 - Closed to open (>15%) vegetation (grassland, shrubland, woody vegetation) on regularly flooded or waterlogged soil; LC_18 - Artificial surfaces and associated areas (urban areas >50%); LC_19 - Bare areas.
Additional file 2. Figure S2. Significant (p-value < 0.01) variation of soil chemistry and climatic variables across African countries. Significance was calculated using the Kruskal-Wallis test for non-parametric data distributions, while pair-wise comparison was calculated using the pairwise Wilcox test. Significant results are indicated using the following nomenclature: * - p-value < 0.05; ** - p-value < 0.01; *** - p-value < 0.001.
Additional file 3.Figure S3. Average relative abundance of the top bacterial (A) and fungal (B) taxa across the sampled sub-Saharan Africa countries.
Additional file 4.Figure S4. Relative abundance of dominant (201 bacterial, 43 Fungal and 7 archaeal) phylotypes across soil samples.
Additional file 5. Figure S5. Relationship between the relative abundance of dominant phylotypes across soil samples and their main environmental predictors, as determined by semipartial correlation analysis. Phylotypes were grouped into environmental categories based on the correlation between phylotype and its major environmental predictor: positive correlation with pH – high pH; negative correlation with pH – low pH; positive correlation with phosphate – high Phosphate; negative correlation with phosphate – low Phosphate; negative correlation with Sodium – low Sodium.
Additional file 6. Figure S6. A-priori ecological model tested using SEM. MAP and MAP are represented as exogenous variables (black rectangles), soil chemistry and vegetation index are represented as endogenous variables (blue rectangles), while the Shannon diversity and abundance of PGPT are represented as response variables (green rectangles). The color and direction of the arrows represent the nature and direction of the causal relationships between variables: red – negative relationship; black – positive relationship.
Additional file 7. Figure S7. MIROC6 model predictions for mean annual temperature (oC) (A) and mean annual precipitation (mm) (B) under too different GH emission scenarios (SSP126 and SSP585), predicted for 2040-2060 and 2080-2100 temporal windows. The predicted datasets are grouped according to country, as indicated by the vertical dashed lines.
Additional file 8. Figure S8-A. Predicted prokaryotic Shannon biodiversity index values (expressed as natural log scale) in soils of the 9 sub- Saharan Africa countries used in this study, for 2040-2060 and 2080-2100 under two distinct GH emission scenarios (SSP126 and SSP585), and comparison with current predicted Shannon biodiversity as estimated by SEM. Pairwise significance values of differences in biodiversity means between the different years and scenarios are represented by the brackets with the following nomenclature: * - p-value < 0.05; ** - p-value < 0.01; *** - p-value < 0.001.
Additional file 9. Figure S8-B. Predicted abundance values of PGPB (expressed as natural log scale) in soils of the 9 sub-Saharan Africa countries used in this study, for 2040-2060 and 2080-2100 under two distinct GH emission scenarios (SSP126 and SSP585), and comparison with current predicted Shannon biodiversity as estimated by SEM. Pairwise significance values of differences in biodiversity means between the different years and scenarios are represented by the brackets with the following nomenclature: * - p-value < 0.05; ** - p-value < 0.01; *** - p-value < 0.001.
Additional file 10. Figure S8-C. Predicted fungal Shannon biodiversity values (expressed as natural log scale) in soils of the 9 sub-Saharan Africa countries used in this study, for 2040-2060 and 2080-2100 under two distinct GH emission scenarios (SSP126 and SSP585), and comparison with current predicted Shannon biodiversity as estimated by SEM. Pairwise significance values of differences in biodiversity means between the different years and scenarios are represented by the brackets with the following nomenclature: * - p-value < 0.05; ** - p-value < 0.01; *** - p-value < 0.001.
Additional file 11. Figure S8-D. Predicted abundance values of PGPF (expressed as natural log scale) in soils of the 9 sub-Saharan Africa countries used in this study, for 2040-2060 and 2080-2100 under two distinct GH emission scenarios (SSP126 and SSP585), and comparison with current predicted Shannon biodiversity as estimated by SEM. Pairwise significance values of differences in biodiversity means between the different years and scenarios are represented by the brackets with the following nomenclature: * - p-value < 0.05; ** - p-value < 0.01; *** - p-value < 0.001.
Additional file 12. Table S1. Metadata for all the sites used in the study, which include the latitude and longitude GPS coordinates, physicochemical properties of the sample soils, macroclimatic variables for each site, and soil texture and land cover classifications based on the macroclimatic variables.
Additional file 13. Table S2. Taxonomy of prokaryotic taxa in the dominant fraction of the microbial community, at the Class taxrank.
Additional file 14. Table S3. Metadata of the dominant phylotypes, including taxonomy, functional predictions (based on FAPROTAX and manual curation), and ecological groups based on the main environmental predictor.
Additional file 15. Table S4. Table with the semi-partial correlation analysis results, in which the correlation values (r) and associated p-values of the variable with the highest correlative value are displayed for each dominant phylotype that was significantly (p-value < 0.05) correlated with environmental factors.
Additional file 16. Table S5. Taxonomy of the taxa considered as plant-growth-promoting
Additional file 17. Table S6. Net estimates and corresponding significance values for the environmental variables associated with soil health in the SEM model.
Additional file 18. Table S7. Number of samples allocated for each country, and number of samples collected.
Additional file 19. Table S8. Variable codes, meaning and units for the environmental variables used in this study.
Additional file 1. Figure S1. Distribution of samples across the 9 African countries according to their land cover (LC) classification. Land cover codes used were the following: LC_1 - Rainfed croplands; LC_2 - Mosaic Cropland (50-70%) / Vegetation (grassland, shrubland, forest) (20-50%); LC_3 - Mosaic Vegetation (grassland, shrubland, forest) (50-70%) / Cropland (20-50%); LC_4 - Closed to open (>15%) broadleaved evergreen and/or semi-deciduous forest (>5m); LC_5 - Closed (>40%) broadleaved deciduous forest (>5m); LC_6 - Open (15-40%) broadleaved deciduous forest (>5m); LC_10 - Mosaic Forest/Shrubland (50-70%) / Grassland (20- 50%); LC_11 - Mosaic Grassland (50-70%) / Forest/Shrubland (20-50%); LC_12 - Closed to open (>15%) shrubland (<5m); LC_13 - Closed to open (>15%) grassland; LC_14 - Sparse (>15%) vegetation (woody vegetation, shrubs, grassland); LC_17 - Closed to open (>15%) vegetation (grassland, shrubland, woody vegetation) on regularly flooded or waterlogged soil; LC_18 - Artificial surfaces and associated areas (urban areas >50%); LC_19 - Bare areas.
Additional file 2. Figure S2. Significant (p-value < 0.01) variation of soil chemistry and climatic variables across African countries. Significance was calculated using the Kruskal-Wallis test for non-parametric data distributions, while pair-wise comparison was calculated using the pairwise Wilcox test. Significant results are indicated using the following nomenclature: * - p-value < 0.05; ** - p-value < 0.01; *** - p-value < 0.001.
Additional file 3.Figure S3. Average relative abundance of the top bacterial (A) and fungal (B) taxa across the sampled sub-Saharan Africa countries.
Additional file 4.Figure S4. Relative abundance of dominant (201 bacterial, 43 Fungal and 7 archaeal) phylotypes across soil samples.
Additional file 5. Figure S5. Relationship between the relative abundance of dominant phylotypes across soil samples and their main environmental predictors, as determined by semipartial correlation analysis. Phylotypes were grouped into environmental categories based on the correlation between phylotype and its major environmental predictor: positive correlation with pH – high pH; negative correlation with pH – low pH; positive correlation with phosphate – high Phosphate; negative correlation with phosphate – low Phosphate; negative correlation with Sodium – low Sodium.
Additional file 6. Figure S6. A-priori ecological model tested using SEM. MAP and MAP are represented as exogenous variables (black rectangles), soil chemistry and vegetation index are represented as endogenous variables (blue rectangles), while the Shannon diversity and abundance of PGPT are represented as response variables (green rectangles). The color and direction of the arrows represent the nature and direction of the causal relationships between variables: red – negative relationship; black – positive relationship.
Additional file 7. Figure S7. MIROC6 model predictions for mean annual temperature (oC) (A) and mean annual precipitation (mm) (B) under too different GH emission scenarios (SSP126 and SSP585), predicted for 2040-2060 and 2080-2100 temporal windows. The predicted datasets are grouped according to country, as indicated by the vertical dashed lines.
Additional file 8. Figure S8-A. Predicted prokaryotic Shannon biodiversity index values (expressed as natural log scale) in soils of the 9 sub- Saharan Africa countries used in this study, for 2040-2060 and 2080-2100 under two distinct GH emission scenarios (SSP126 and SSP585), and comparison with current predicted Shannon biodiversity as estimated by SEM. Pairwise significance values of differences in biodiversity means between the different years and scenarios are represented by the brackets with the following nomenclature: * - p-value < 0.05; ** - p-value < 0.01; *** - p-value < 0.001.
Additional file 9. Figure S8-B. Predicted abundance values of PGPB (expressed as natural log scale) in soils of the 9 sub-Saharan Africa countries used in this study, for 2040-2060 and 2080-2100 under two distinct GH emission scenarios (SSP126 and SSP585), and comparison with current predicted Shannon biodiversity as estimated by SEM. Pairwise significance values of differences in biodiversity means between the different years and scenarios are represented by the brackets with the following nomenclature: * - p-value < 0.05; ** - p-value < 0.01; *** - p-value < 0.001.
Additional file 10. Figure S8-C. Predicted fungal Shannon biodiversity values (expressed as natural log scale) in soils of the 9 sub-Saharan Africa countries used in this study, for 2040-2060 and 2080-2100 under two distinct GH emission scenarios (SSP126 and SSP585), and comparison with current predicted Shannon biodiversity as estimated by SEM. Pairwise significance values of differences in biodiversity means between the different years and scenarios are represented by the brackets with the following nomenclature: * - p-value < 0.05; ** - p-value < 0.01; *** - p-value < 0.001.
Additional file 11. Figure S8-D. Predicted abundance values of PGPF (expressed as natural log scale) in soils of the 9 sub-Saharan Africa countries used in this study, for 2040-2060 and 2080-2100 under two distinct GH emission scenarios (SSP126 and SSP585), and comparison with current predicted Shannon biodiversity as estimated by SEM. Pairwise significance values of differences in biodiversity means between the different years and scenarios are represented by the brackets with the following nomenclature: * - p-value < 0.05; ** - p-value < 0.01; *** - p-value < 0.001.
Additional file 12. Table S1. Metadata for all the sites used in the study, which include the latitude and longitude GPS coordinates, physicochemical properties of the sample soils, macroclimatic variables for each site, and soil texture and land cover classifications based on the macroclimatic variables.
Additional file 13. Table S2. Taxonomy of prokaryotic taxa in the dominant fraction of the microbial community, at the Class taxrank.
Additional file 14. Table S3. Metadata of the dominant phylotypes, including taxonomy, functional predictions (based on FAPROTAX and manual curation), and ecological groups based on the main environmental predictor.
Additional file 15. Table S4. Table with the semi-partial correlation analysis results, in which the correlation values (r) and associated p-values of the variable with the highest correlative value are displayed for each dominant phylotype that was significantly (p-value < 0.05) correlated with environmental factors.
Additional file 16. Table S5. Taxonomy of the taxa considered as plant-growth-promoting
Additional file 17. Table S6. Net estimates and corresponding significance values for the environmental variables associated with soil health in the SEM model.
Additional file 18. Table S7. Number of samples allocated for each country, and number of samples collected.
Additional file 19. Table S8. Variable codes, meaning and units for the environmental variables used in this study.
Keywords
Soil microbiome, Microbial biodiversity, Climate change, Ecosystem predictions, Sub-Saharan Africa (SSA), SDG-13: Climate action
Sustainable Development Goals
Citation
Cowan, D.A., Lecre, P.H., Amon, C.E.R. et al. 2022, 'Biogeographical survey of soil microbiomes across sub-Saharan Africa: structure, drivers, and predicted climate-driven changes', Microbiome, vol. 10, art. 131, pp. 1-22. DOI : 10.1186/s40168-022-01297-w