EPS, Soil Microbes & Land Cover in Tropical Soils

Telechargé par kidindakagro1
Soil Biology and Biochemistry 187 (2023) 109221
Available online 22 October 2023
0038-0717/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Extracellular polymeric substances are closely related to land cover,
microbial communities, and enzyme activity in tropical soils
Laurent K. Kidinda
a
,
b
,
*
, Doreen Babin
c
, Sebastian Doetterl
d
, Karsten Kalbitz
a
,
Basile B. Mujinya
b
, Cordula Vogel
a
,
**
a
Chair of Soil Resources and Land Use, Institute of Soil Science and Site Ecology, TU Dresden, Dresden, Germany
b
Biogeochemistry and Ecology of Tropical Soils and Ecosystems, University of Lubumbashi, Lubumbashi, Democratic Republic of the Congo
c
Julius Kühn Institute (JKI) - Federal Research Centre for Cultivated Plants, Institute for Epidemiology and Pathogen Diagnostics, Braunschweig, Germany
d
Soil Resources Group, Department of Environmental Systems Science, ETH Zurich, Zurich, Switzerland
ARTICLE INFO
Original content: Extracellular polymeric
substances are closely related to land cover,
microbial communities, and enzyme activity in
tropical soils (Original data)
Keywords:
High-throughput amplicon sequencing
Carbon cycling
Nutrient cycling
Soil geochemistry
Bacterial 16S rRNA gene
Fungal internal transcribed spacer
ABSTRACT
Extracellular polymeric substances (EPS) form the main matrix of microbial biolms and play a crucial role in
maintaining microbial life. However, factors inuencing EPS concentration and production in soil are poorly
understood. Here we show that EPS are closely related to microbial communities and nutrient acquisition in
tropical forest and cropland soils with varying iron-aluminum-manganese concentrations and total reserve in
base cations. We found under homogenized moisture and temperature conditions that EPS concentration and
production efciency (i.e., EPS per unit of microbial biomass) depend more on land cover than on geochemical
soil properties. EPS concentration and production efciency were higher in cropland than in forest soil and were
related to the higher relative abundance of microbial sequences identied as Paenibacillaceae, Ramlibacter,
Chaetosphaeria, Burkholderiaceae, and Xanthobacteraceae, pointing to potential EPS producers. In contrast, lower
EPS concentration in forest soil was related to the higher relative abundance of microbial sequences associated
with e.g., Gemmatimonas and Massilia, suggesting potential EPS degradation. We also found that EPS production
efciency was positively related to microbial investment in nutrient acquisition, implying that EPS production
likely follows the same principles as extracellular enzyme activity. That is, EPS production may increase when
resources are scarce to facilitate nutrient acquisition, and decrease when resources are abundant. Overall, mi-
crobial community composition and resource demand seem to control EPS degradation and accumulation in
tropical soils, which could inuence microbially-driven carbon and nutrient cycling.
1. Introduction
Extracellular polymeric substances (EPS), mainly composed of pro-
teins and polysaccharides, form the main matrix of microbial biolms
(Flemming et al., 2016) and can account for more than 90% of the dry
mass of a biolm, with microbial cell biomass constituting less than 10%
(Flemming and Wingender, 2010). EPS play a variety of roles in main-
taining microbial life (Flemming et al., 2016), including the stabilization
of extracellular enzymes for prolonged catalytic activity under varying
environmental conditions (De Beeck et al., 2021). Additionally, EPS
facilitate microbial cells interaction with their environment, alleviate
environmental stress, trap nutrients, and serve as a readily available
nutrient source (de Brouwer et al., 2002; Flemming et al., 2016). Thus,
depending on the physiological needs of microbes, EPS can act as both a
nutrient sink and source (De Beeck et al., 2021).
Our knowledge of the factors affecting EPS concentration and pro-
duction comes largely from wastewater, activated sludge, aquatic en-
vironments, and single microbial cultures (Wang et al., 2014;
Gonz´
alez-García et al., 2015; Premnath et al., 2021; Rath et al., 2022),
but rarely from soil (e.g., Redmile-Gordon et al., 2015; Zethof et al.,
2020; Olagoke et al., 2022). Most soil EPS studies so far have focused on
soil aggregation (e.g., Zethof et al., 2020; Bettermann et al., 2021) and
inoculation of EPS-producing microbial strains (Sandhya et al., 2009;
Sandhya and Ali, 2014; Costa et al., 2018). However, the factors inu-
encing EPS have been largely ignored. In general, EPS concentration and
production can be inuenced by microbial community composition,
* Corresponding author. Chair of Soil Resources and Land Use, Institute of Soil Science and Site Ecology, TU Dresden, Dresden, Germany.
** Corresponding author.
E-mail addresses: [email protected] (L.K. Kidinda), [email protected] (C. Vogel).
Contents lists available at ScienceDirect
Soil Biology and Biochemistry
journal homepage: www.elsevier.com/locate/soilbio
https://doi.org/10.1016/j.soilbio.2023.109221
Received 26 April 2023; Received in revised form 11 September 2023; Accepted 20 October 2023
Soil Biology and Biochemistry 187 (2023) 109221
2
environmental stresses (e.g., salinity, metal pollution, drought, and
temperature), and nutrient availability (Siddharth et al., 2021). EPS
concentration and production in the absence of environmental stresses
may largely depend on the composition and abundance of microbial
communities, and their interactions with resource availability. Thereby,
microbes can greatly vary in their ability to produce EPS (Vuko et al.,
2020). For instance, some strains of Pseudomonas, Bacillus, and Paeni-
bacillus are known to produce high amounts of EPS in vitro (Siddharth
et al., 2021). However, in a complex environment like soil, where
diverse microbial communities interact, it is often unclear which specic
microbial taxa are associated with an increase or decrease in EPS con-
centration and production. While not all microbes produce EPS (Flem-
ming and Wingender, 2010), non-producing ones may still benet from
EPS, such as by decomposing them during starvation (Smith and
Schuster, 2019). Hence, the balance between EPS-producing and
non-producing microbial taxa in soil can determine EPS concentration,
depending on environmental resource availability.
EPS biosynthesis is an energy-intensive process for microbes (Costa
et al., 2018), expected to provide benets comparable to energy in-
vestment (Flemming et al., 2016). It is unclear how EPS production ef-
ciency, dened as EPS concentration per unit of microbial biomass
(Redmile-Gordon et al., 2015), relates to microbial investment in
nutrient acquisition, particularly in deeply weathered soils with limited
nutrients. Nutrient imbalance is thought to stimulate EPS production
efciency, especially when carbon (C) is in excess and nitrogen (N) is
limited (Redmile-Gordon et al., 2015). However, in tropical forest soils
where N is rarely limiting for microbes (Jing et al., 2020; Wang et al.,
2020), it is unclear how EPS production efciency relates to C avail-
ability. In addition, microbes in the tropics have to cope with often
deeply weathered soils that are poor in rock-derived nutrients but rich in
iron (Fe) Manganese (Mn), and aluminum (Al) oxides (Chadwick and
Asner, 2016; Doetterl et al., 2021). In response to limited nutrients,
microbes produce more extracellular enzymes (Sinsabaugh et al., 2009;
Kidinda et al., 2022), which may affect investment in EPS production.
Generally, in acidic tropical soils, both enzymes and EPS can be sorbed
to mineral surfaces and complexed with Fe and Al (Mikutta et al., 2011;
Olagoke et al., 2020). Moreover, microbes may increase EPS production
as a protective mechanism under high acidity (Fang and Zhong, 2002).
Therefore, it is not clear how EPS concentration and production ef-
ciency may vary in geochemically diverse soils where they may be
inuenced by organo-mineral interaction and sorption effects. For
tropical soils developed from different parent materials, organo-mineral
interactions and concentrations of rock-derived nutrients may vary
signicantly (Doetterl et al., 2021), potentially inuencing EPS con-
centration and production efciency differently.
Rapid land cover change in many tropical regions due to socio-
economic pressures (Tyukavina et al., 2018; Doetterl et al., 2021) has
led to signicant changes in soil biogeochemical properties on which
microbes depend (Kidinda et al., 2023). Conversion of stable forests with
high biomass productivity to croplands with limited nutrient inputs and
biomass availability, and soils more prone to erosion and degradation,
can result in changes in qualitative and quantitative C inputs and
nutrient availability. Consequently, microbial properties such as
biomass and nutrient acquisition are more dependent on organic matter
(OM) in forests, while they are more dependent on geochemical soil
properties in croplands (Kidinda et al., 2023). Furthermore, conversion
of forest to cropland may lead to a decrease in fungal abundance relative
to bacteria (Kidinda et al., 2023) and a change in microbial community
composition (Zhang et al., 2022). However, it is unclear how microbial
community composition differs between tropical forest and cropland in
geochemically contrasting soils and whether this difference can affect
EPS concentration and production efciency.
To understand how soil microbial communities and investment in
nutrient acquisition relate to EPS concentration and production ef-
ciency in geochemically distinct forest and cropland soils, we seek to
answer the following questions.
(i) How do patterns of EPS concentration and production efciency
differ between forest and cropland soils with varying geochem-
ical properties? We hypothesize that soils with highest FeAlMn
and lowest rock-derived nutrient concentrations will have lowest
EPS concentrations because of limited resources and the potential
for these metals to affect microbial activity. However, EPS pro-
duction efciency will increase in these soils as a potential mi-
crobial response to promote resource acquisition and mitigate the
stress caused by these metals. In geochemically similar soils, EPS
concentration will be highest in forest, because high C inputs can
stimulate microbial biomass. Conversely, EPS production ef-
ciency will be highest in cropland soil, as disturbances can
stimulate EPS production per unit of microbial biomass.
(ii) How do microbial communities differ between forest and crop-
land soils with varying geochemical properties? How does the
enrichment or decline of certain microbial taxonomic groups
relates to EPS concentration and production efciency? We hy-
pothesize that cropland soil will have different taxonomic com-
positions compared to forest soil due to differences in OM inputs
and disturbances that alter physicochemical soil properties.
Furthermore, soils with highest FeAlMn concentrations will
have different taxonomic compositions compared to those with
highest rock-derived nutrient concentrations, as microbes may
have adapted to the presence of these metals and their inuence
on resource availability. Specic fungal and bacterial taxa (po-
tential EPS-producers) will be positively associated with EPS
concentration and production efciency, while other taxa (po-
tential non-EPS-producers) will be negatively associated.
(iii) How does microbial investment in nutrient acquisition relate to
EPS concentration and production efciency? We hypothesize
that microbial investment in nutrient acquisition in the form of
extracellular enzymes will be positively related to EPS production
efciency, as EPS potentially play a crucial role ensuring pro-
longed catalytic enzyme activity. However, if microbes have to
invest more in nutrient acquisition, their biomass is likely to be
low resulting in low EPS concentration despite a high EPS pro-
duction efciency.
To answer these questions, we collected tropical montane forest and
cropland soils developed from geochemically distinct parent materials
(mac, mixed sedimentary rocks, and felsic) in tropical Africa. On these
samples, we analyzed microbial properties under homogenized moisture
and temperature conditions to better assess the effects of microbial
communities and nutrient acquisition on EPS concentration and pro-
duction efciency in soils that differ in total reserve in base cations
(TRB) and metal (Fe, Al, and Mn) concentration.
2. Materials and methods
2.1. Study sites
Study sites of the original soil sampling were located in montane
forest and cropland in the border region between the eastern Democratic
Republic of the Congo, western Rwanda, and southwestern Uganda
along the Albertine Rift. The climate is tropical humid with monsoonal
dynamics (K¨
oppen Af - Am). Mean annual temperatures range from 15.3
to 19.2 C and mean annual precipitation ranges from 1697 to 1924 mm.
The topography is hilly with smaller plateaus and ridges, steep slopes
and various valley shapes. Forest is classied as lower montane forest to
lower montane cloud forest, which vary in species diversity and
composition (Bruijnzeel and Hamilton, 2000), but generally have
similar functional traits and stand age (Doetterl et al., 2021). Cropland is
characterized by subsistence cultivation of cassava (Manihot esculenta)
with no usage of chemical fertilizers. Nevertheless, occasional applica-
tions of organic household waste and agricultural residues as soil
amendments have been reported by the farmers. Tillage is done
L.K. Kidinda et al.
Soil Biology and Biochemistry 187 (2023) 109221
3
manually with traditional hoes to prepare the land. Under forest, surface
erosion is marginal, while cropland has lost 1043 cm of surface soil
over the past 66 years (Wilken et al., 2021). Soils in Democratic Republic
the Congo are predominantly Nitisols developed from mac magmatic
rocks, typically mac alkali basalts rich in base cations and metals. Soils
in Uganda are predominantly Lixisols and Andosols developed from
felsic magmatic and metamorphic rocks, typically gneiss and granite,
where volcanic ash deposition have been mixed in during the early and
mid-Holocene (Biggs et al., 2021). Soils in Rwanda are predominantly
Ferralsols and Acrisols developed from a mixture of sedimentary rocks
with different geochemical properties, consisting of alternating layers of
quartz-rich sandstone, siltstone, and dark clay schists containing fossil
organic C (Schlüter and Trauth, 2006; Doetterl et al., 2021).
2.2. Soil sampling
Soil samples were collected between March and June 2018, as part of
a larger research project detailed by Doetterl et al. (2021). We sampled
soils developed from three parent materials (mac, mixed sedimentary
rocks, and felsic) in both forest and cropland. For each parent material,
four topographic positions (plateau/ridge, upslope, midslope, and val-
ley/foot slope) were selected. In forest, we established three 40 ×40 m
plots at each topographic position, totaling 36 plots. In cropland, we
employed a stratied random sampling design with one 3 ×3 m plot for
each topographic position, totaling 100 plots. In each plot, two and four
(for cropland and forest) 1-m soil cores were collected. Samples were
taken in 10-cm increments before making depth-explicit composite
samples for each plot. For the current study, we considered three forest
and three to four cropland plots per topographic position in each
geochemical region and analyzed them at three soil depths (topsoil
010 cm, shallow subsoil 3040 cm, deeper subsoil 6070 cm), resulting
in 225 samples. This selection aimed to cover a wide range of variation
in geochemical soil properties, providing a better basis for studying the
effects of microbial properties on EPS concentration and production
efciency. It is important to note that differences in sampling design
between forest and cropland led to larger distances between cropland
plots compared to forest plots. To determine if this introduced uncer-
tainty into the analysis, we previously quantied the distance between
the empirical distribution of microbial properties in forest and cropland
soils using the two-sample Kolmogorov-Smirnov test. Results showed no
sampling bias toward signicantly lower or higher values of microbial
properties (Kidinda et al., 2023).
2.3. Incubation experiment
A laboratory incubation experiment using the 225 samples under
standardized temperature (20 C) and moisture conditions (60% water
holding capacity) was conducted to allow the stabilization of microbial
communities and activity. A detailed description of this incubation
experiment can be found in Bukombe et al. (2021) and Kidinda et al.
(2022). Note that the purpose of this experiment was not to mimic
real-world conditions, but to better assess the effects of microbial
communities and their investment in nutrient acquisition on EPS con-
centration and production efciency for which the standardization of
moisture and temperature conditions is a prerequisite. Briey, 50 g
sample aliquot (12-mm sieved and air-dried) were added to a 100 mL
beaker and soil moisture was adjusted to 60% of the maximum soil water
holding capacity. During incubation, CO
2
samples were taken when the
CO
2
concentration had reached 10003000 ppm. The end of incubation
was determined when the standard deviation of means of respiration
rate between three consecutive measurement time points was smaller
than the standard deviation between three replicates of the same mea-
surement time point. This indicated that CO
2
production did not in-
crease further within the measurement error. Following this rationale,
the incubation experiment ended after 120 days for forest and 67 days
for cropland soils. Soils were then sieved with a 2-mm sieve before
microbial and chemical analyses were performed.
2.4. Soil microbial analyses
2.4.1. Microbial biomass C, enzyme activity, and nutrient acquisition
Our measurements of microbial biomass C (MBC) and extracellular
enzyme activity, and assessment of microbial investment in nutrient
acquisition, are described in Kidinda et al. (2022, 2023), where they
were originally published to address other specic research questions. In
the current study, we use these data to investigate their relationship with
EPS concentration and production efciency. Briey, MBC was
measured using the chloroform fumigation-extraction method (Vance
et al., 1987). Here, MBC was calculated as the difference in K
2
SO
4
extractable organic C between fumigated and non-fumigated sample
aliquots, with 0.45 as a factor for microbial biomass extraction ef-
ciency (Beck et al., 1997). Potential extracellular enzyme activity was
measured uorometrically in the soil suspension following German et al.
(2011). We measured ve enzymes relevant to the acquisition of C
(cellobiohydrolase (CB) and β-glucosidase (BG)), N (N-acetylglucosa-
minidase (NAG) and leucine-aminopeptidase (LAP)), and P (acid phos-
phatase (AP)). Briey, a soil slurry was prepared by sonicating 1 g of
2-mm sieved soil in 50 mL of 50
μ
M sodium acetate trihydrate buffer at
an energy of 60 J mL
1
and current power (W) of 34 J s
1
(Marx et al.,
2001). Prior to enzyme analysis, an additional 50 mL of sodium acetate
was added to the homogenized suspension and stirred with a magnetic
stirrer. To measure the activity of CB, BG, NAG, AP, and LAP, we used
the following substrates, respectively: β-D-cellobioside (200
μ
M),
β-D-glucopyranoside (200
μ
M), N-acetyl-β-D-glucosaminide (200
μ
M),
phosphate (400
μ
M), and L-Leucine-7-amido-4-methylcoumarin (100
μ
M). All assays included standards, controls for soil and substrates and
the sample in four replicates. Microplates were incubated at 30 C for 1 h
and measured uorometrically at an excitation wavelength of 360 nm
and an emission wavelength of 450 nm using a microplate reader
(Synergy HTX Multi-Mode Reader, Bio-Tek Instruments, Inc., USA).
Based on potential enzyme activity, two indicators of microbial in-
vestment in nutrient acquisition were calculated (Kidinda et al., 2023).
The rst is vector characteristics (vector angle and length), calculated as
proportional ratios of measured potential extracellular enzyme activity
(Moorhead et al., 2013). Vector length increases with microbial in-
vestment in C relative to N and P. Vector angle <45indicates a pre-
dominant investment in N acquisition and angle >45indicates a
predominant investment in P acquisition (Cui et al., 2019; Moorhead
et al., 2013). The second indicator comprises the specic C- (CE
MBC
), N-
(NE
MBC
), and P- (PE
MBC
) acquiring enzyme activity, calculated by
summing all measured enzymes targeting the same nutrient and
normalizing to MBC (Medeiros et al., 2015). MBC was further normal-
ized to SOC (MBC
SOC
) to reect C availability to microbes (Insam and
Domsch, 1988; Mendoza et al., 2020).
2.4.2. Extracellular polymeric substances
Extracellular polymeric substances (EPS) were extracted using cation
exchange resin (Redmile-Gordon et al., 2014). Briey, 2.5 g of 2-mm
sieved soil was placed in a centrifuge tube, and 25 mL of 0.01 M
CaCl
2
was added to extract soluble microbial products. Then, depending
on SOC content, cation exchange resin (Sigma-Aldrich/DOWEX, Saint
Louis, USA, PN 91,973) followed by 25 mL of phosphate-buffered saline
was added to the centrifuge tube to extract EPS. The supernatants con-
taining EPS were transferred to a new centrifuge tube and frozen at
20 C. Prior to measurement, EPS extracts were ltered using syringe
lters (CHROMAFIL ® PET-45/25, polyester). EPS-polysaccharides
were quantied by the sulfuric acid-phenol method using D
(+)-glucose (Roth, PN X997) as a calibration standard (Dubois et al.,
1956). Briey, 1 mL of EPS extract was added to two test tubes (with and
without 25
μ
L of 80% phenol). Then, 2.5 mL of 9598% sulfuric acid
were added to each test tube and placed in a water bath at 90 C for 10
min. Absorbance was measured at 480 nm wavelengths using a
L.K. Kidinda et al.
Soil Biology and Biochemistry 187 (2023) 109221
4
microplate reader (Synergy HTX Multi-Mode Reader, Bio-Tek In-
struments, Inc., USA). EPS-proteins were quantied using bovine serum
albumin as calibration standard following Lowrys method (Lowry et al.,
1951) modied for microplate format by Redmile-Gordon et al. (2013).
Briey, Lowry reagent ‘Awas prepared by sequentially combining 3.5 g
CuSO
4
5H
2
O 100 mL
1
H
2
O, 7 g NaK tartrate 100 mL
1
H
2
O, and 70 g
Na
2
CO
3
L
1
0.35 N NaOH in a 1:1:100 ratio (v:v:v). Lowry reagent ‘B
was prepared in the same manner as reagent A, except that CuSO
4
5H
2
O
was replaced by deionized water. Then, 100
μ
L of reagent A or B was
added to 100
μ
L of EPS extracts or standard solutions in a 96-well
microplate in four replicates. The microplate was incubated in the
dark at room temperature for 10 min. Subsequently, 100
μ
L folin-phenol
(2 N diluted 10-fold in H
2
O) was added, and the microplate was incu-
bated in the dark at room temperature for 30 min. The absorbance was
measured at 750 nm emission wavelengths using the same device as for
EPS-polysaccharide. Specic EPS-protein (EPS-protein
MBC
) and
EPS-polysaccharide (EPS-polysaccharide
MBC
) concentrations were
calculated by normalizing EPS per soil unit to the unit of MBC and used
to indicate EPS production efciency (Redmile-Gordon et al., 2015).
2.4.3. DNA extraction
To characterize microbial community in the investigated soils, total
community DNA extraction from a 0.5 g sample aliquot was conducted
using a FastDNA SPIN Kit for soil (MP Biomedicals, Santa Ana, CA,
United States) following manufacturer recommendations. Sixty samples
consisting of ve replicates of the two land cover types (forest and
cropland) and three parent materials (mac, mixed sedimentary rocks,
and felsic) were selected from our incubated soils. We restricted the
extraction to two soil depths (010, 3040 cm) since we expected that
low variation in microbial biomass and enzyme activity between the
3040 cm and 6070 cm subsoil samples (Kidinda et al., 2023) will
result into comparable microbial communities. DNA quality was
assessed by agarose (0.8%) gel electrophoresis, followed by ethidium
bromide (0.005%) staining and UV-light photography (Intas Gel Jet
Imager, 2004; Intas, G¨
ottingen, Germany).
2.4.4. Illumina sequencing and sequence processing
Library construction and sequencing of the 16S rRNA gene V3V4
region or ITS2 fragment was carried out by Novogene (Cambridge, UK)
on an Illumina platform (PE250) using the primers Uni341F (5
-
CCTAYGGGRBGCASCAG-3
) and Uni806R (5
- GGACTACNNGGG-
TATCTAAT -3
) targeting Bacteria (Sundberg et al., 2013) or gITS7 (5
-
GTGARTCATCGARTCTTTG-3; Ihrmark et al., 2012) and ITS4 (5
-
TCCTCCGCTTATTGATATGC -3; White et al., 1990) to characterize the
fungal community. Read pairs for which none of the two primer se-
quences could be detected were sorted out using Cutadapt (Martin,
2011). Primer-trimmed sequence reads were corrected for errors, and
merged before amplicon sequence variants (ASVs) were identied using
DADA2 version 1.24.0 in R (Callahan et al., 2016) with the following
parameters: maxN =0, truncQ =2, minLen =50 (all only for ITS),
trimRight =c(5,5) (only for 16S), maxEE =c(2,2), matchIDs =TRUE,
maxMismatch =0, and otherwise default parameters. For the 16S rRNA
gene dataset, only sequences with a length of 404429 bp were
considered. For both datasets, chimeric sequences were identied
within the DADA2 pipeline and removed. Taxonomic afliations to each
16S-ASV or ITS-ASV were assigned in a Galaxy workow (Cock et al.,
2013) with an Expect Value cut-off of 0.001 and a percent identity
cut-off of 80% against the SILVA 138.1 SSU Ref NR99 (Quast et al.,
2013) or the UNITE database version 8.3 (Nilsson et al., 2019) for
Bacteria or Fungi, respectively. Sequences that belonged to cyano-
bacteria/chloroplasts and mitochondria were removed from the dataset.
In addition, microbial ASVs with less than ve reads in the entire dataset
were excluded to account for PCR and sequencing artifacts. The number
of retained ASVs was 13,945 for fungal ITS and 23,040 for bacterial 16S
rRNA gene. The average number of quality-ltered sequences per sam-
ple was 157,989 for fungal ITS and 80,051 for 16S rRNA gene.
2.5. Geochemical soil properties
For a detailed description of the methods used to analyze physico-
chemical properties of the investigated soils, see Doetterl et al. (2021)
and Kidinda et al. (2022, 2023). In short, soil organic C (SOC), total N,
total Al, Fe, and Mn, and bioavailable phosphorus (P) concentrations,
pH (KCl), as well as the total reserve in base cations (calcium, sodium,
potassium, magnesium) were determined on air-dried aliquots prior to
incubation (Doetterl et al., 2021). Total dissolved N (TDN) was analyzed
at the end of incubation in 0.01 M CaCl
2
extracts (Rennert et al., 2007).
The chemical index of alteration (CIA), calculated as described in
Kidinda et al. (2023), was used to reect changes in geochemical
properties among soil samples. In our study, the CIA index reects the
geochemical status of soils and correlates positively with FeAlMn
concentrations and negatively with TRB.
2.6. Data analysis and statistics
2.6.1. K-means clustering of geochemical soil properties
To classify samples into geochemically distinct soils based on
geochemical variables (i.e., nutrients and metals) known to affect soil
microbes, we performed K-means clustering. The TRB as a measure of
rock-derived nutrient concentrations was chosen as the rst set of var-
iables because it is one of the most important determinants of microbial
activity (Bukombe et al., 2021; Kidinda et al., 2022). It may thus affect
EPS concentration and production efciency, especially in deeply
weathered, tropical soils, depleted in base cations. In addition, total
FeAlMn concentrations were chosen as a second set of variables that
may affect resource availability to microbes through organo-mineral C
stabilization and P immobilization (Mikutta et al., 2011), which in turn
may affect EPS concentration and production efciency. Prior to cluster
analysis, we converted the TRB and the sum of total FeAlMn con-
centrations into Z-scores and performed clustering using the R packages
statsand ‘factoextra. The analysis resulted in three clusters, referred to
as geochemical clusters in this study. The rst cluster (High TRB/Low
FeAlMn) grouped soil samples derived from felsic parent material that
have high TRB and low FeAlMn concentrations (Fig. 1a). Additionally,
this cluster is characterized by high pH (KCl) and high bioavailable P
concentration (Table 1). The second cluster (Low TRB/High FeAlMn)
is dominated by soils derived from mac parent material (79%) and has
low TRB but high FeAlMn concentrations. This cluster is also char-
acterized by high CIA, low bioavailable P and low pH (KCl). The third
cluster (Low TRB/Low FeAlMn) grouped soil samples derived pre-
dominantly from mixed sedimentary rocks (56%) and felsic parent
material (36%). This cluster is characterized by low TRB, low FeAlMn
and bioavailable P concentrations, and low pH (KCl). The number of
samples differed among geochemical clusters and land cover types and
was highest for Low TRB/Low FeAlMn and lowest for High TRB/Low
FeAlMn (Fig. 1b). This aspect was further considered when perform-
ing mean comparisons among the clusters.
2.6.2. Variance analysis of nutrient acquisition and EPS
We performed a two-way Analysis of Variance (ANOVA) to compare
means among geochemical clusters and land cover types at each soil
depth. Because of the unequal number of samples within groups, we
used type III Sums of Squares (SS) ANOVA. Type III SS calculates un-
weighted means by summing the means of each level of our independent
variables and dividing by the total number of levels to deal with an
unequal number of samples (Graefe et al., 2022). Note that due to
marginal effects of topography on microbial properties (Kidinda et al.,
2022, 2023), topographic positions were combined within each land
cover type. We then conducted a one-way ANOVA to compare means
among soil depths in each land cover type and geochemical cluster. Prior
to running ANOVA, the assumptions of normal distribution of residuals
and homogeneity of variances were tested using Shapiro-Wilks and
Levenes tests (Webster and Lark, 2019). We performed Order-Norm
L.K. Kidinda et al.
Soil Biology and Biochemistry 187 (2023) 109221
5
transformation because all variables deviated from the ANOVA as-
sumptions (Peterson and Cavanaugh, 2019). Factors that signicantly
affected EPS concentration and production efciency or nutrient
acquisition were further considered for pairwise mean comparison using
lsmeansR package and Tukey correction (Midway et al., 2020).
2.6.3. Microbial communities
2.6.3.1. Unconstrained ordination. To explore similarities and dissimi-
larities among microbial communities, Non-Metric Multidimensional
Scaling (NMDS) was applied to log-transformed relative abundance of
microbial ASVs. NMDS was performed based on Bray-Curtis dissimi-
larity matrices using the function metaMDSof ‘veganR package with
three dimensions and 100 permutations considering a stress level <0.1
and a non-metric t R
2
>0.95 (Oksanen et al., 2019).
2.6.3.2. Permutational multivariate analysis of variance. To test whether
geochemical clusters, land cover and soil depth had signicant effects on
the observed microbial community assemblages and thus harbored
different microbial communities, we performed a three-way Permuta-
tional Multivariate Analysis of Variance (PERMANOVA). PERMANOVA
was performed on log-transformed relative abundance of microbial
ASVs based on Bray-Curtis dissimilarity matrices and 999 permutations
using ‘vegdist and ‘adonis2 functions. Factors that signicantly
affected microbial communities were further considered for pairwise
PERMANOVA post hoc tests using ‘pairwise.adonis2 function and
Bonferroni correction. These analyses were performed using ‘veganR
package (Oksanen et al., 2019).
2.6.3.3. Constrained ordination. To visualize variations in bacterial and
fungal community composition associated with geochemical clusters,
land cover, soil depth, and chemical properties, we performed a con-
strained ordination analysis. To decide whether microbial community
data were homogeneous or heterogeneous and therefore more suitable
for linear or unimodal ordination methods, we performed Detrended
Correspondence Analysis and examined the length of the rst ordination
axis (Lepˇ
s and ˇ
Smilauer, 2003). For both bacterial and fungal ASV
datasets, the length of the rst ordination axis was >4, indicating a
unimodal data structure suitable for Canonical Correspondence Analysis
(CCA). To t soil chemical variables onto CCA, we used the envt
function. Envt allowed us to determine the relative contribution of
environmental variables to the separation of microbial communities
along ordination axes. Here, we included resource-related variables (e.
g., SOC, total N, TDN, bioavailable P), pH, and the geochemical status of
Fig. 1. Geochemical clusters discriminating soil samples based on their total reserve in base cations (TRB) and total FeAlMn concentrations (a). Panels b-d" show
the number of samples in each geochemical cluster.
Table 1
Means ±standard deviation of soil chemical properties in each geochemical cluster. Soil depths (010 cm, 3040 cm, 6070 cm) are combined for each land cover.
TRB: total reserve in base cations. Total FeAlMn: sum of total iron, aluminium, and manganese concentrations. CIA: Chemical index of alteration. TDN: Total
dissolved nitrogen. SOC: soil organic carbon.
Parameters Unit High TRB/Low FeAlMn Low TRB/High FeAlMn Low TRB/Low FeAlMn
Cropland Forest Cropland Forest Cropland Forest
TRB cmol
c
kg
1
30.0 ±8.6 24.6 ±2.7 6.0 ±4.4 11.4 ±6.8 7.0 ±7.1 9.3 ±4.9
Total FeAlMn % 8.2 ±2.8 4.8 ±2.4 18.2 ±3.3 18.4 ±3.2 7.7 ±1.9 5.8 ±3.5
CIA 88.5 ±4.6 91.0 ±4.2 98.7 ±1.0 98.3 ±0.7 95.2 ±3.1 92.8 ±5.7
TDN mg kg
1
69.2 ±47.9 319.0 ±191.3 102.9 ±137.4 230.6 ±322.0 63.9 ±51.1 93.3 ±119.2
SOC % 2.3 ±1.0 3.4 ±1.5 2.4 ±1.3 4.0 ±2.7 2.4 ±1.1 2.0 ±1.8
Total N % 0.2 ±0.1 0.3 ±0.2 0.2 ±0.1 0.4 ±0.2 0.2 ±0.1 0.2 ±0.1
Soil C:N 10.7 ±1.4 10.2 ±1.4 11.7 ±1.7 11.9 ±2.9 12.1 ±1.8 19.1 ±26.0
Soil C:P 9.4 ±6.7 50.0 ±52.7 34.5 ±42.9 27.8 ±23.7 31.7 ±21.5 44.3 ±82.6
Soil N:P 0.8 ±0.5 4.9 ±5.0 2.7 ±3.1 2.4 ±1.7 2.6 ±1.8 3.3 ±5.5
pH (KCl) 5.1 ±0.5 5.6 ±0.4 4.3 ±0.3 3.7 ±0.5 4.1 ±0.3 4.2 ±0.9
Bioavailable P mg kg
1
175.1 ±64.1 59.0 ±63.2 4.7 ±4.4 17.9 ±15.7 24.7 ±60.6 13.4 ±22.2
L.K. Kidinda et al.
1 / 16 100%
La catégorie de ce document est-elle correcte?
Merci pour votre participation!

Faire une suggestion

Avez-vous trouvé des erreurs dans l'interface ou les textes ? Ou savez-vous comment améliorer l'interface utilisateur de StudyLib ? N'hésitez pas à envoyer vos suggestions. C'est très important pour nous!