Content uploaded by Pu Jia
Author content
All content in this area was uploaded by Pu Jia on Sep 29, 2022
Content may be subject to copyright.
Glob Change Biol. 2022;00:1–13. wileyonlinelibrary.com/journal/gcb
|
1© 2022 John Wiley & Sons Ltd.
1 | INTRODUCTIO N
Phosphorus (P), as an essential macronutrient for all life on Earth,
had long been viewed as the second most limiting nutrient element
after nitrogen (N) in cert ain terrestrial ecosystems, especially in
tropical regions with highly weathered soils (Mise et al., 2018; Yang
et al., 2013). Unfortunately, there is an increasing evidence that P
limitation is a widespread phenomenon occurring in various types
of terrestrial ecosystems around the world rather than in some
specific terrestrial ecosystems (Augusto et al., 2017; Elser, 2012;
Hou et al., 2020). Unlike N, P is derived mainly from parent mate-
rial weathering, which is a mechanical process in essence (Chadwick
et al., 1999; Walker & Syers, 19 76; Wardle et al., 2004). Although
total soil P content in terrestrial ecosystems can be considerably
Received: 30 August 20 21
|
Accepted: 13 April 2022
DOI : 10.1111/gcb.16213
RESEARCH ARTICLE
Remarkable effects of microbial factors on soil phosphorus
bioavailability: A country- scale study
Jing- li Lu1 | Pu Jia1 | Shi- wei Feng1 | Yu- tao Wang1 | Jin Zheng1 |
Shu- ning Ou1 | Zhuo- hui Wu1 | Bin Liao2 | Wen- sheng Shu1 | Jie- Liang Liang1 |
Jin- tian Li1
Jing- li Lu and Pu Ji a contribut ed equally to thi s work.
1Institute of Ecologic al Science,
Guangzhou Key Laboratory of Subtropical
Biodiversity and Biomonitoring,
Guangdong Provincial Key Laboratory of
Biotechnology for Plant Development,
School of Life Science s, South China
Normal U niversit y, Guangzhou , People's
Republic of China
2School of Life Science s, Sun Yat- sen
University, Guangzhou, People's Republic
of China
Correspondence
Jie- Liang Liang and Jin- tian Li, S chool
of Life Sciences, South China Normal
University, Guangzhou 510631, People's
Republic of China.
Emails: liang_jieliang@126.com and
lijintian@m.scnu.edu.cn
Funding information
Guangdong Basic and A pplied Basic
Research Foundation, Grant/Award
Number : 2021B1515120039; Key- Are a
Research and Development Program
of Guangd ong Province , Grant/Award
Number : 2019B110207001; National
Natural Science Foun dation of China,
Grant/Award Number: 41622106,
42077117 and 42177009; Natural
Science Foundation of G uangdong
Province of China, Gra nt/Award Numbe r:
2020A1515010937
Abstract
Low soil phosphorus (P) bioavailability causes the widespread occurrence of P- limited
terrestrial ecosystems around the globe. Exploring the factors influencing soil P bio-
availability at large spatial scales is critical for managing these ecosystems. However,
previous studies have mostly focused on abiotic factors. In this study, we explored
the effects of microbial factors on soil P bioavailability of terrestrial ecosystems using
a country- scale sampling effort. Our results showed that soil microbial biomass car-
bon (MBC) and acid phosphatase were important predictors of soil P bioavailability
of agro- and natural ecosystems across China although they appeared less important
than total soil P. The two microbial factors had a positive effect on soil P bioavailability
of both ecosystem types and were able to mediate the effects of several abiotic fac-
tors (e.g., mean annual temperature). Meanwhile, we revealed that soil phytase could
affect soil P bioavailability at the country scale via ways similar to those of soil MBC
and acid phosphatase, a pattern being more pronounced in agroecosystems than in
natural ecosystems. Moreover, we obtained evidence for the positive effects of mi-
crobial genes encoding these enzymes on soil P bioavailability at the country scale
although their effect sizes varied between the two ecosystem types. Taken together,
this study demonstrated the remarkable effects of microbial factors on soil P bioavail-
ability at a large spatial scale, highlighting the importance to consider microbial factors
in managing the widespread P- limited terrestrial ecosystems.
KEY WORDS
agroecosystem, bioavailable soil phosphorus, metagenomics, microbial functional gene, natural
ecosystem, phosphatase, phytase
2
|
LU e t aL.
high, only a small fraction (less than 6% on average) of soil P can be
readily available for plant growth (Cross & Schlesinger, 1995; Morel
et al., 2000; Yang & Post, 20 11). A large proportion of soil P is hardly
available in either organic or inorganic insoluble forms (Rodríguez
& Fraga, 1999). In other words, the low bioavailability of soil P is
a main reason for P limitation in global terrestrial ecosystems (Hou
et al., 2020). Therefore, a better understanding of the mechanisms
that determine soil P bioavailability at large spatial scales is essential
to develop approaches for ameliorating P- limitation in global terres-
trial ecosystems.
Considerable effor ts have been made to identif y abiotic
factors influencing soil P bioavailability at large spatial scales.
For instance, a previous study showed that soil bioavailable
P concentration in 80 grasslands across China decreased with in-
creasing mean annual temperature (MAT; Geng et al., 2017). Hou
et al. (2018) conducted a meta- analysis of bioavailable P in global
natural (seminatural) soils and found that both MAT and mean an-
nual precipitation (MAP) were negatively correlated with bioavail-
able soil P concentration. Additionally, some soil physicochemical
proper ties (such as parent material, sand content, pH, organic
carbon and Al- Fe oxide content) were also reported as import-
ant factors influencing soil P bioavailability at global scale (Achat
et al., 2016; Augusto et al., 2017). However, only a small proportion
of previous studies on the factors influencing soil P bioavailability
at large spatial scales focused on both abiotic and biotic factors
(e.g., Achat et al., 2016; Augusto et al., 2017; Delgado- Baquerizo
et al., 2013; Ringeval et al., 2017). A remarkable example was the
work of Delgado- Baquerizo et al. (2013), which was based on the
data from 224 drylands distributed globally and showed that the
bioavailable soil P concentration was positively correlated with
not only abiotic factors (e.g., labile organic matter fraction), but
also acid phosphatase (ACP). Moreover, the study further revealed
that the ef fects of major abiotic factors on soil P bioavailabilit y
were largely mediated by ACP.
Besides ACP, microbes can also secrete other organic
P- mineralizing enzymes (e.g., alkaline phosphatase [ALP] and phy-
tase) (Alori et al., 2017; Richardson & Simpson, 2011; Rodríguez
et al., 2006). In addition, organic acids (e.g., gluconic acid) pro-
duced by microbes can solubilize inorganic insoluble P largely by
decreasing soil pH (Rodríguez et al., 2006). There is an emerging
evidence that ALP and gluconic acid were among the important
factors affecting soil P bioavailability at small spatial scales (Liang
et al., 2020; Wang et al., 2021). For example, a significant positive
correlation was found between the abundance of phoD (a gene en-
coding ALP) and bioavailable soil P concentration in a 35- year- old
Chinese fir plantation (Wang et al., 2021). However, little is known
about the roles of these microbial enzymes in determining bio-
available soil P concentration at large spatial scales. Note also that
some of these microbial enzymes can be encoded by a couple of
homologous genes. Like ACP (encoded by aphA, olpA or phoN), ALP
and phyt ase are known to be encoded by three (i.e., phoA, phoD
or phoX) and two (i.e., appA or phy) homologous genes, respec-
tively (Alori et al., 2017; Richardson & Simpson, 2011; Rodríguez
et al., 2006). The effectiveness of these microbial functional genes
in determining soil P bioavailability at large spatial scales and their
relative importance in different types of ecosystems are poorly
understood.
In this study, in order to uncover the roles of microbial factors
in determining soil P bioavailability at large spatial scales, we estab-
lished a country- scale dataset by collecting a total of 207 soil sam-
ples from 29 agroecosystems (mainly paddy fields) and 40 natural
ecosystems (including forests, grasslands and Gobi deserts) across
22 provinces in China. These two types of terrestrial ecosystems
were chosen because they are main components of terrestrial eco-
systems and have distinct characteristics. Briefly, agroecosystems
are species- poor systems and sustained largely by human interven-
tions (Bohan et al., 2013; Dodd, 2000; Fuhrer, 2003), while natu-
ral ecosystems are generally open and self- sustainable ones with
higher biodiversity (Liang et al., 2017; Wu et al., 2019). Specifically,
the aims of this study were to explore: (1) besides ACP, whether
and how other microbial factors (e.g., microbial biomass, ALP and
phytase) can affect soil P bioavailability at a large spatial scale;
(2) the relative importance of homologous genes encoding micro-
bial P- mineralizing enzymes in determining soil P bioavailability;
and (3) the extent to which the relative importance of microbial
P- mineralizing enz ymes and their encoding genes may vary be-
tween different types of ecosystems.
2 | MATERIALS AND METHODS
2.1 | A country- scale sampling
We conducted a nationwide field survey in 22 provinces across
mainland China (Figure 1a) from July to August 2018. The study
sites were selected to comprehensively represent the geographic,
climatic and edaphic features of agro- and natural ecosystems na-
tionwide (Tables S1 and S2). At each study site, we collected soil
samples from agro- or/and natural ecosystems. When these two
ecosystem types existed in a given study site simult aneously, the
distance between our sampling sites for them was within 5 km.
In total, 29 agroecosystems (mainly paddy fields) and 40 natural
ecosystems (including 27 forests, 9 grasslands and 4 Gobi deser ts)
were sampled. Gobi is a major type of desert and located mainly
in East Asia (Sternberg, 2015). At each study site, we collected
three soil samples for each ecosystem type at a depth of 0– 20 cm
and each soil sample was collected from three randomly distrib-
uted spots. To minimize the potential impact of vegetation, we col-
lected soil samples about 1 m away from dense vegetation. While
sampling, we recorded the geographic coordinates (i.e., longitude
and latitude) and elevation of each study site using the Global
Positioning System (GPS). Climate factors, including MAT and
MAP for each study site, were derived from the WorldClim Global
Climate Database (http://world clim.org) from 1950 to 2000. Soil
samples were transported back to the laborator y as soon as possi-
ble under refrigeration. Subsequently, we divided each soil sample
|
3
LU et aL.
into three parts. One part was stored at 4°C for measuring soil
microbial biomass carbon (MBC) and nitrogen (MBN), one part was
stored at −20°C for metagenomic sequencing, and another part
was air- dried, ball- milled, sieved and homogenized for physico-
chemical analyses.
2.2 | Soil physicochemical and enzyme activity
measurements
Soil pH was measured in a 1:2.5 (w/v) aqueous solution using a pH
meter (Brand, Germany). Bioavailable soil P was determined accord-
ing to sodium bicarbonate (Olsen) method (Olsen et al., 195 4). After
H2SO4– HClO 4 digestion, total N and total P in soils were determined
using the Kjeldahl and molybdate blue colorimetric method, respec-
tively (Lityński et al., 1976; Sparks & Sparks, 1996). Soil organic C
(SOC) was measured using the dichromate oxidation method (Sparks
& Sparks, 1996). Soil clay content (diameter < 0.0 02 mm) was de-
termined by the method of soil particle- size fractionation (Chen &
Chiu, 2003). MBC and MBN were measured by the chloroform fu-
migation extraction method (Brookes et al., 198 5; Joergensen, 1996;
Wu et al., 199 0). Activities of ACP and ALP were measured by de-
termination of the amount of p- nitrophenol released from soils
after incubation at 37°C for 1 h with the substrate p- nitrophenyl
phosphate in modified UB buffer (pH 6.5 for ACP and pH 11 for
ALP) (Tabatabai, 1994). Using sodium phy tate as the substrate, soil
phytase activit y was assessed as the amount of inorganic P liberated
from sodium phytate solution (Eeckhout & Paepe, 1994).
2.3 | DNA extraction, sequence processing and
metagenomics analysis
Soil DNA was extracted using the FastDNA Spin kit (MP
Biomedicals) following the manufacturer's protocol. DNA qual-
ity was assessed with the NanoDrop 20 00 spec trophotometer
(Thermo Scientific, USA). Silica- based columns were used to purify
DNA samples. After constructing shotgun libraries with an aver-
age insert size of about 300 bp, DNA samples were sequenced on
an Illumina NovaSeq PE250 sequencer (Illumina, USA). Raw reads
quality filtering and assembly were performed as described previ-
ously (Liang et al., 2020).
To get a deeper understanding of the effects of microbial factors
on soil P bioavailability at a large spatial scale, we quantified the rel-
ative abundances of microbial functional genes involved in P cycling,
including those encoding ACP, ALP and phy tase. Based on their cor-
responding KEGG Ontolog y (KO) numbers (Table S3), we retrieved a
representative protein sequence of each gene family, including ACP
encoding genes (aphA, phoN and olpA), ALP encoding genes (phoA,
phoD and phoX ) and phytase encoding genes (appA and phy). We
then used HMMer 3.2 tool (Eddy, 1998) to build a hidden Markov
model (HMM) for each gene subfamily. Detailed information about
the workflow of HMM building and usage was described in our re-
cent work (Li et al., 2021). BBMap v36.x (with parameters build = 1,
minid = 0.97 and k = 14) was used to calculate gene coverage. Our
in- house Perl scripts were used to calculate the relative abundances
of abovementioned microbial functional genes based on the results
of gene coverage.
FIGURE 1 Total and bioavailable soil P concentrations of agro- and natural ecosystems across China. Colors of circles and boxes were
used to indicate ecosystem types. Circles with t wo colors on map of China (a) indicate that we collected soil samples from both agro- and
natural ecosystems simultaneously at the same study site. Some of the circles on the map overlap each other since these sampling sites are
too close to each other. N, the number of study sites; n, the number of samples; *** indicates significant differences between ecosystem
types (p < .001)
!
(
!
(
!
(
!
(
!
(
!
(
!
(
!
(
!
(
!
(
!
(
!
(
Habitat types
Agro- and natural ecosystems both (N = 29)
Natural ecosystems only (N = 11)
0250 500 1,000
KM
±
N, the number of sites 2000
50
100
0
1000
0
Total P (mg kg–1)
Bioavailable P (mg kg–1)
Agroecosystem Natural ecosystem
Agroecosystem Natural ecosystem
(c)
(b)(a)
n, the number of samples
***
n = 87
n = 120
***
n = 87
n = 120
4
|
LU e t aL.
2.4 | Statistical analysis
To detect the significant dif ferences between agro- and natural
ecosystems across China in total soil P, bioavailable soil P and the
relative abundances of P cycling genes (Table S3), we performed the
linear mixed- effects model (LMM) analysis using lmerTest package
(Kuznetsova et al., 2017 ) with study site identit y as a random ef fect.
Furthermore, the effects of dif ferent abiotic and biotic variables on
soil P bioavailability were quantified by the LMMs (Li et al., 2022; Rao
et al., 2021). Before doing that, in order to avoid multicollinearity, we
used a variance inflation factor (VIF) threshold of 3.3 to eliminate
those variables which were strongly correlated (Fanin et al., 2020;
Kock, 2015). In LMMs, study sites were used as random effects to
avoid the potential influences of the three soil samples collected from
each ecosystem. Spearman- rank correlation test was used to explore
the correlations between the 16 variables considered in this study,
including geographic and climatic factors, soil physicochemical prop-
erties and microbial factors. Since there was a strong positive corre-
lation bet ween MBC and MBN (Figure S1), MBN was excluded from
our subsequent analyses. Additionally, we standardized our data by
either log- transformation or Z- score transformation when needed to
improve normality and homogeneity. These analyses were performed
using the R software version 4.1.2 (R Core Team, 2021).
To identify the major factors influencing soil P bioavailabil-
ity, we used Random forest analysis (Breiman, 2001), wherein the
relative importance of study site, MAT, MAP, soil pH, total P, total
N, SOC, clay content, MBC and P- mineralizing enzymes (ACP, ALP
and phyt ase) were ranked. We also used Random forest analysis
to identif y the main predictors of soil P bioavailability among ACP
genes (aphA, phoN and olpA), ALP genes (phoA, phoD and phoX) and
phytase genes (appA and phy). Random forest analysis was done by
randomForest package (Liaw & Wiener, 2002) and rfPermute pack-
age (Archer, 2021) in the R statistical computing environment.
Additionally, we assessed the significance of the models and cross-
validated R2 with 5000 permutations of response variables using A3
package (Fortmann- Roe, 2013) in R 4.1.2 (R Core Team, 2021).
Piecewise structural equation model (SEM) was performed
to quantif y the effects (direct or indirect or both) of explanatory
variables on soil P bioavailability using piecewiseSEM package in R
(Lefcheck, 2016). According to the previously reported potential
causal relationships between explanatory and response variables
(Table S4), we established a priori model for agro- and natural eco-
systems, respectively (Figure S2). To simplify the initial models, we
introduced a composite variable (phosphatase activity) into our
piecewiseSEM model (Schermelleh- Engel & Moosbrugger, 2003),
and then eliminated non- significant pathways until we attained the
final models (Li et al., 2022). Note that composite variables did not
alter the underlying piecewiseSEMs (Grace, 2006; Shipley, 2001). To
avoid the potential influences of the three samples of each ecosys-
tem, we per formed mixed models using lme4 package in R with study
site identity as a random effect. We used Fisher' C test to deter-
mine whether our models adequately represent the observed data
(Oelmann et al., 2021). Moreover, the standardized total effect of
each explanatory variable on soil P bioavailabilit y was assessed using
semEff packages in R (Murphy, 2020).
2.5 | An attempt to upscale our study to the
global scale
Given that a previous study presented the metagenomes of the
global soils (Bahram et al., 2018), from which we can obtain data on
microbial functional genes encoding P- mineralizing enzymes of the
global soils, we have attempted to explore the effects of such genes
on P bioavailability of the global soils. To get a reliable dataset of P bi-
oavailability of the global soils, we first contacted the authors of the
previous study. It was a pity that we were told that our targeted data
were not measured in their study. Second, we not only downloaded a
NetCDF format dataset from the global soil dataset for earth system
modeling (GSDE, http://globa lchan ge.bnu.edu.cn/home), but also
downloaded the global gridded soil phosphorus distribution maps at
0.5- degree resolution (https://daac.ornl.gov/SOILS/ guide s/Global_
Phosp horus_Dist_Map.html). After extracting the data relevant to
bioavailable soil P with ncdf4 package (Pierce, 2021), rgdal package
(Bivand et al., 2021) and raster package (Hijmans, 2021) in R 4.1.2
(R Core Team, 2021), we tested the reliabilit y of the two datasets
using the method of Pineiro et al. (2008). Unfortunately, we found
that the regression lines between predicted and observed values
deviated greatly from the 1:1 line and that the R2 was close to zero
(detailed data not shown). These results indicate the lack of reliabil-
ity. Therefore, no upscaling was done.
3 | RESULTS
3.1 | Soil P bioavailability
We collected a total of 87 and 120 soil samples from 29 agro- and 40
natural ecosystems across China, respectively (Figure 1; Tables S1
and S2). Our study sites covered diverse geographic areas, with lon-
gitude from 99.40 to 129.27°E, latitude from 22.14 to 47.47°N and
elevation from 1 to 2518 m (Table S5). The climates of these areas
also varied greatly, with MAT from 1.20 to 22.9°C and MAP from
124 to 1909 mm (Table S5). Averagely, the total soil P concentra-
tion of agroecosystems was significantly higher than that of natu-
ral ecosystems (1007 vs. 562.1 mg kg−1; p < .001, Figure 1b). The
average concentration of bioavailable soil P of agroecosystems was
45.96 mg kg−1, being much higher than that (8.78 mg kg−1) of natural
ecosystems (p < .001, Figure 1c).
3.2 | Effects of microbial biomass and P-
mineralizing enzymes on soil P bioavailability
The average soil phytase activity of agroecosystems was significantly
(p < .01) higher than that of natural ecosystems, while an opposite
|
5
LU et aL.
pattern was observed for the average soil ACP activit y (Table S5). There
was no significant difference between these two ecosystem t ypes in
soil ALP activity, MBC and MBN (Table S5). Random forest analysis
showed that for both ecosystem types seven out of the 16 factors con-
sidered in this study had a significant effect on soil P bioavailability, re-
spectively (p < .05, Figure 2). Specifically, in agroecosystems, the effects
of total soil P, SOC, ACP, MAT, site, phytase and MBC decreased suc-
cessively (Figure 2a); while, in natural ecosystems, the effect of total soil
P was followed by site, MAP, pH, MBC, MAT and phytase (Figure 2b).
LMM analysis showed that total soil P, MBC and ACP activity positively
affected the soil P bioavailability of both ecosystem types (p < .05,
Figure 3). Additionally, the soil P bioavailability of agroecosystems was
positively affected by phytase ac tivity as well (p < .05, Figure 3a).
Piecewise structural equation models explained 82% and 62% of
the total variance in soil P bioavailabilit y of agro- and natural ecosys-
tems, respectively (Figure 4). In agroecosystems, soil phytase activ-
ity and total soil P had a direct positive effect on soil P bioavailability,
while soil phosphatase activity indirectly affected soil P bioavail-
ability through its direct effect on total soil P (p < .001, Figure 4a).
Notably, the effects of MAT, total soil N and clay content on soil
P bioavailability in agroecosystems were mediated by soil phospha-
tase activity (p < .05, Figure 4a). In natural ecosystems, MBC, total
soil P and pH were the only three factors that could directly affect
soil P bioavailability (p < .05, Figure 4b). The indirect ef fects of MAT,
total soil N, clay content and pH on soil P bioavailability were me-
diated largely by phosphatase activity, while phosphatase activity
could affect soil P bioavailability through its direct effect on total
soil P and MBC (p < .05, Figure 4b). In addition to directly affecting
soil P bioavailability, MBC and soil pH also indirectly affected soil
P bioavailability by directly affecting total soil P (p < .05, Figure 4b).
When direct and indirect effects of individual factors were taken
into account together, five out of the eight factors retained in the
piecewiseSEM for agroecosystems positively affected soil P bio-
availability (Figure 4c), whereas in natural ecosystems all the eight
factors had a positive impact (Figure 4d). Remarkably, all microbial
factors, including MBC, phosphatase and phytase activities, had
positive effects in both ecosystem types (Figure 4c,d). More specif-
ically, the total positive effect of the three microbial factors in agro-
ecosystems was almost equal to that of the t wo soil physicochemical
factors (i.e., total P and N); while, in natural ecosystems the positive
effects of MBC, phosphatase and phytase activities were ranked the
first, fifth and eighth, respectively (Figure 4c,d).
3.3 | Effects of microbial functional genes
encoding P- mineralizing enzymes on soil P
bioavailability
The average relative abundance of microbial functional genes in-
volved in soil P cycling of agroecosystems was significantly (p < .05)
higher than that of natural ecosystems (Figure 5a). A similar pattern
was found for ACP- encoding and ALP- encoding genes (Figure 5b,c).
These results were generally in line with our obser vation that total
and bioavailable soil P was higher in agroecosystems than in natural
ecosystems (Figure 1b,c), although the two ecosystem types exhib-
ited no significant difference in the relative abundance of genes en-
coding phy tase (p > .05, Figure 5d).
In both agro- and natural ecosystems, phoN and phoD were the
most abundant among the homologous genes encoding ACP and
ALP, respectively (p < .05, Table S6). These results were consistent
with thos e of previous studies (Ragot et al., 2015; Tan et al., 2013). As
to the homologous genes encoding phytase, phy was more abundant
FIGURE 2 Main predictors of soil P bioavailability in agro- and natural ecosystems across China. The figure shows the random forest
mean predictor impor tance (the percentage of increase in the mean variance error [MSE]) of abiotic and microbial variables on bioavailable
soil P concentrations for (a) agro- and (b) natural ecosystems. Colors of bars indicate different variable types. The cross- validated R2 and
significance of random forest models are shown. ACP, acid phosphatase; ALP, alkaline phosphatase; MAP, mean annual precipitation; MAT,
mean annual temperature; MBC, microbial biomass carbon; SOC, soil organic carbon. n, the number of samples. Significance levels: **p < .01;
*p < .05; n.s., non- significant (p > .05)
Agroecosystem (
n = 87
)
Natural ecosystem
(
n = 120
)
Total P
ACP
MAT
SOC
P
hytase
MBC
MAP
ALP
Clay
pH
Total P
MAP
pH
MBC
SOC
MAT
Ph
ytase
Total N
ACP
ALP
Site
Site
0
20
40
60
80
Increased in MSE (%)
0
10
20
30
40
50
Increased in MSE (%)
R2 = 0.49, P < 0.001
Climatic variables
Biotic variables
Soil physicochemical variables
**
*
****
** **
*
***
Sampling sites
R2 = 0.44, P < 0.001
Climatic variables
Biotic variables
Soil physicochemical variables
Sampling sites
*
**
(a) (b)
n.s. n.s.
n.s. n.s.
n.s.
n.s.
n.s. n.s.
6
|
LU e t aL.
than appA in both ecosystem types (p < .05, Table S6). Nonetheless,
among the eight functional genes encoding these three enzymes,
five were shown to be significant predictors of soil P bioavailability
of agroecosystems (p < .001, Figure 6a), compared with only two in
natural ecosystems (p < .001, Figure 6b). phoX (encoding ALP) was
the only significant gene predictor shared by both ecosystem types,
although its relative importance varied between ecosystem types.
Note that phoX and phy (encoding phytase) were identified as the
most important gene predictors in agro- and natural ecosystems,
respectively (Figure 6). LMM analysis showed that the relative abun-
dance of phoX and phy had a significant positive effect on the soil
P bioavailability of agro- and natural ecosystems, respectively
(p < .05, Figure 7).
4 | DISCUSSION
As an integral nutrient element for all life on Earth, P in soil can have
a profound ef fect not only on soil C and N cycling, but also on pri-
mary production of terrestrial ecosystems (Elser et al., 20 07; Yao
et al., 2018). However, compared with those of soil C and N (Al- Kaisi
et al., 2005; Bouskill et al., 2020; Kieft et al., 1998), the bioavail-
ability of soil P and its influencing factors at large spatial scales has
received much less attention (Delgado- Baquerizo et al., 2013). To
our knowledge, this study is among the first attempts to explore the
effects of both abiotic and microbial factors (especially functional
genes) on soil P bioavailabilit y in different ecosystem types at a large
spatial scale.
4.1 | Abiotic factors affected soil P bioavailability
As a routine measure to alleviate P limitation in agroecosystems
(Hou et al., 2020), fertilization is known to result in elevated
total P content in agricultural soils (Fuhrer, 2003). In accord-
ance with this common view, we found that on average the con-
centration of total soil P of agroecosystems across China was
higher than that of their nearby natural ecosystems (Figure 1b).
Moreover, our results indicated that total P in soil was the most
important abiotic factor influencing soil P bioavailability of agro-
and natural ecosystems across China (Figures 2– 4). Similarly,
MAT was found to be an important abiotic factor in both eco-
system t ypes (Figures 2– 4). These results supported the previ-
ously known critical roles of abiotic factors in determining soil
P bioavailability at large spatial scales (e.g., Augusto et al., 2017;
Delgado- Baquerizo et al., 2013). Note also that study site iden-
tity was shown as another important predictor of soil P bioavail-
ability of both ecosystem types (Figure 2). This phenomenon
could be considered as a reflection of the composite effect of
abiotic and biotic factors on soil P bioavailability, given that
characteristics of individual study sites include both types of
factors. Nonetheless, one should keep in mind that the interac-
tions bet ween abiotic and biotic factors are very complex and it
is therefore difficult to completely distinguish their effects on
soil P bioavailability. For instance, in agroecosystems, total soil
P concentration was positively correlated with ACP and SOC ,
which, in turn, showed a positive relationship with soil P bio-
availability (Figure S1a).
FIGURE 3 Effects of abiotic and microbial variables on soil P bioavailability in agro- and natural ecosystems across China. Ef fects of
different variables on bioavailable soil P concentrations of (a) agro- and (b) natural ecosystems are assessed by the linear mixed- effects
models with study site identity as a random effec t. Effect sizes of individual predictor variables are normalized coefficients from the
linear mixed- effect s models. The solid circles represent significant effects (p < .05), and the open circles represent non- significant effects
(p > .05). The explained variance (R2) of the model is shown in blue. ACP, acid phosphatase; ALP, alkaline phosphatase; MAT, mean annual
temperature; MBC, microbial biomass carbon. n, the number of samples
Total P
Total N
Phytase
pH
MBC
MAT
Clay
ALP
ACP
R² = 0.79
Normalized effect on soil bioavailable P
-0.250.000.25 0.50 0.75 1.00
R² = 0.70
Normalized effect on soil bioavailable P
-0.20 -0.10 0.00 0.10 0.20 0.30 0.40 0.50
Total P
Total N
Phytase
pH
MBC
MAT
Clay
ALP
ACP
Agroecosystem (
n = 87
)
Natural ecosystem
(
n = 120
)
(a) (b)
|
7
LU et aL.
4.2 | Microbial biomass and P- mineralizing
enzymes affected soil P bioavailability
The potential effect s of microbial biomass on soil nutrient cy-
cling processes had long been recognized (Brookes et al., 19 84;
Jenkinson, 1981). A straightforward reason is that the more the
soil microbial biomass, the higher the microbial enzyme activity. In
agreement with this common view, our result s revealed an impor-
tant role of microbial biomass (as estimated by MBC) in mediating
soil P bioavailability at the country scale (Figures 2– 4). Remarkably,
the effect of MBC on soil P bioavailabilit y was more pronounced in
natural ecosystems than in agroecosystems, although no significant
difference was found between these two ecosystem types in MBC
(Table S5). These results could be attributed partly to a scenario that
human interventions were more intensive in agroecosystems than in
natural ecosystems. For instance, the land- use types of most natu-
ral ecosystems investigated in this study did not change in the past
several decades (Table S2). Moreover, natural ecosystems (including
forests and grasslands) in China generally received fertilizer appli-
cation at a much lower rate than that of agroecosystems (e.g., ap-
proximately 30 kg P ha−1 year−1 for forests vs. 180 k g P ha−1 year−1 for
agroecosystems; Table S1; Peng et al., 2011; Xu et al., 2002). The
more intensive human inter ventions in agroecosystems not only re-
duced the microbial effect on soil P bioavailability, but also resulted
in weaker associations between microbial biomass (estimated by
MBC or MBN) or microbial enzyme (ACP, ALP and phytase) activit y
and SOC or tot al N (Figure S1).
As a P- mineralizing enzyme secreted by microorganisms, ACP
is the only microbial factor that has been reported explicitly to
have a positive effect on soil P bioavailability at a large spatial scale
and can mediate the effects of abiotic factors (Delgado- Baquerizo
et al., 2013). Our results showed that this previously important find-
ing originating from global drylands was likely applicable to other
ecosystem types (especially agroecosystems) in relatively humid re-
gions. For example, we found that ACP activity was an important
predictor of soil P bioavailability of agroecosystems across China
FIGURE 4 Piecewise structural equation models (SEMs) showing ef fects of abiotic and microbial factors on soil P bioavailability in agro-
and natural ecosystems across China. Direct and indirect effects (a, b) and the standardized total effects (c, d) of different factors on soil
P bioavailability of the two ecosystem types are shown. Standardized path coefficients representing the effect sizes of potential causal
variables are indicated by numbers adjacent to arrows. The width of arrows is proportional to the potential causal effect between variables.
The black arrows indicate positive effects, and the red arrows indicate negative effect s. The numbers adjacent to boxes of response
variables denote the explained variance (R2). Phosphat ase activity is the first component of a principal component analysis performed with
ACP and ALP. Colors indicate different types of predictor variables. MAT, mean annual temperature; MBC, microbial biomass carbon. n, the
number of samples. Significance levels: ***p < .001; **p < .01; *p < .05. The priori models are shown in Figure S2
Bioavailable P
Total N pH
Phosphatase
ACP ALP
MBC
Phytase Clay
Total P
MAT
0.29*
0.28**
0.51***
0.22*
0.31** 0.18*
0.26**
-0.24**
-0.22**
0.82***
-0.20*
0.55***
-0.35***
-0.11*
0.02
Bioavailable P
Total N pH
Phosphatase
ACPALP
MBC
Phytase Clay
Total P
MAT
0.89***
0.25**
0.24*
-0.44***
-0.28*
0.38***
0.10
-0.37***
0.30**
-0.30*
0.11
0.14
-0.13
MAT
Clay
pH
Total N
Total P
Phosphatase
Phytase
MBC
-1.0
1.0
0.5
0.0
-0.5
Standardized total effects
from SEM (unitless)
(c)
-0.1
0.4
0.3
0.2
0.1
Standardized total effects
from SEM (unitless)
(d)
0.0
MAT
Clay
pH
Total N
Total P
Phosphatase
Phytase
MBC
(a) (b)
Agroecosystem Natural ecosystem
R2 = 0.82 R2 = 0.62
R2 = 0.65 R2 = 0.79
R2 = 0.57
R2 = 0.55
Fisher’s C = 10.07,
p = 0.86, df = 16
Fisher’s C = 3.82,
p = 0.87, df = 8
R2 = 0.56
R2 = 0.80
R2 = 0.51
R2 = 0.44
Bi
oava
il
a
bl
e
P
T
ota
l
N
pH
Ph
osp
h
atas
e
A
CP
A
L
P
M
B
C
Phy
tas
e
C
lay
T
ota
l
P
MAT
0.89
***
0.25**
0
.24
*
-0.44***
*
-0.28*
0
.38
***
0
.
10
-0.37***
*
0.30**
-0
.
30*
0
.
11
0
.
1
4
-0.13
R
2
=
0
.
82
R
2
=
0
.
65
R
2
=
0
.7
9
R
2
=
0
.5
7
R
2
=
0
.55
F
isher’s C = 10.07
,
p
=
0
.
86,
df
= 16
f
B
i
oa
v
a
il
ab
l
e
P
T
ota
l
N
pH
Ph
osp
h
atas
e
A
C
P
A
L
P
M
B
C
Phy
tas
e
C
la
y
T
ota
l
P
MAT
0.29*
0
.28
**
0.51***
0.22
*
0
.
3
1
**
0.18*
0.18
0
.2
6**
-0.24**
-0.22**
0.82***
-0.20*
0.55***
-0.35***
-
0.11
*
0.02
R
2
=
0
.
62
F
isher’s C = 3.82
,
p
=
0
.
8
7
,
df
= 8
f
R
2
=
0
.5
6
R
2
=
0
.
80
R
2
=
0
.5
1
R
2
=
0
.44
8
|
LU e t aL.
(Figure 2a) and had a significant positive effect on soil P bioavail-
ability (Figures 3a and 4a,c). In contrast, we found that ALP activity
had no significant effect on soil P bioavailability of both agro- and
natural ecosystems across China (Figures 2 and 3). One possible rea-
son for these results was that the effects of ACP and ALP depended
largely on soil pH (Nannipieri et al., 2011). It is generally believed
that ACP and ALP prevail in acidic and alkaline soils, respectively
(Juma & Tabatabai, 1978; Nannipieri et al., 2011). In this study, the
soils of both ecosystem types were slightly acidic (averagely pH < 7,
Table S5) and therefore favored the role of ACP.
A striking finding of this study lied in that phytase activity was
an important predictor of soil P bioavailability of both ecosystem
ecotypes across China (Figure 2) and that phytase activity had a
positive and direct effect on soil P bioavailability of agroecosystems
(Figures 3a and 4a,c). Phytates, including derivatives of inositol pent-
aphosphate and hexaphosphate (Turner et al., 2002), account for ap-
proximately 10%– 50% of the total organic P in soils (Mullen, 2005).
Plants (mostly cereals and legumes) can synthesize phy tates and
accumulate them in seeds for seedlings growth (Lott et al., 2000;
Turner et al., 2002). However, phytates cannot be directly used
as P source by plants unless they are mineralized into bioavailable
inorganic P by microbial phytase (George et al., 2007; Richardson
et al., 2004). We therefore propose that due to crop residue reten-
tion phytates are more likely to accumulate in soils of agroecosys-
tems than in those of natural ecosystems, which thereby requires
phytase to play a more important role in agroecosystems than
in natural ecosystems in release of bioavailable inorganic P from
phytates to meet the P requirement of crops in agroecosystems
(George et al., 2007; Menezes- Blackburn et al., 2013; Richardson
et al., 2004). In accordance with our notion, the average soil phytase
activity of agroecosystems was significantly higher than that of nat-
ural ecosystems (Table S5).
4.3 | Microbial functional genes encoding
P- mineralizing enzymes affected soil P bioavailability
Microbial functional genes encoding P- mineralizing enzymes are
a reflection of the intrinsic abilities of corresponding microor-
ganisms to mineralize organic insoluble P compounds (Rodríguez
FIGURE 5 Relative abundances of microbial functional genes in agro- and natural ecosystems across China. Data on genes involved in
microbial P cycling (a), encoding acid and alkaline phosphat ase (b, c), and phytase (d) are shown, respectively. Colors of boxes were used to
indicate ecosystem types. n, the number of samples. Significant levels: ***p < .001; **p < .01; *p < .05; n.s., non- significant (p > .05)
Agroecosystems Natural ecosystem
0.000
0.001
0.002
0.003
Relative abundanceofgenes
encoding soil acid phosphatase(%)
AgroecosystemsNatural ecosystem
0.00
0.05
0.10
0.15
0.20
Relative abundanceofgenes
involved in soil microbialPcycling(%)
Agroecosystems Natural ecosystem
0.000
0.005
0.010
0.015
0.020
Relative abundanceofgenes
encoding soil alkaline phosphatase(%)
(a) (b)
(c) (d)
Agroecosystems Natural ecosystem
0.0000
0.0005
0.0010
0.0015
Relative abundanceof
genesencodingsoilphytase (%)
n=87
*
n=120 n=87
**
n=120
n=87
***
n=120
n=87
n.s.
n=120
|
9
LU et aL.
et al., 2006), while assaying activities of microbial P- mineralizing
enzymes at specific time points provides only single snapshots of
the expression of their encoding genes in time that is gener ally sus-
ceptible to extrinsic factors (Zaheer et al., 2009). Thus, an impor-
tant step toward a deeper understanding of microbial effects on
soil P bioavailability at large spatial scales is to explore the extent
to which the relative abundances of these functional genes can
affect soil P bioavailability at large spatial scales. Despite it s sig-
nificance, such a topic has not yet been addressed in the literature.
In a wider context, however, there is evidence that incorporating
FIGURE 6 Relative importance of microbial functional genes in predicting soil P bioavailability of agro- and natural ecosystems across
China. The figure shows the random forest mean predictor impor tance (the percentage increase in the mean variance error [MSE]) of the
relative abundances of microbial functional genes on soil P bioavailability of (a) agro- and (b) natural ecosystems. The cross- validated R2
and significance of models are shown. p < .0 01 indicates that models are significant. aphA, olpA and phoN are homologous genes encoding
ACP; phoA, phoD and phoX are homologous genes encoding ALP; appA and phy are homologous genes encoding phytase. n, the number of
samples. Significance levels: **p < .01; *p < .05; n.s., non- significant (p > .05)
Agroecosystem (
n = 87
)
Natural ecosystem
(
n = 120
)
0
10
20
30
40
Increased in MSE (%)
0
10
20
30
Increased in MSE (%)
aphA
olpA
phoX
phoN
phoA
pho
D
phy
appA
phoX
olpA
phy
pho
N
appA
aphA
Site
Site
*
*
**
***
**
*
R2 = 0.20, P < 0.001 R2 = 0.11, P < 0.001
(b)(a)
n.s.
n.s.
n.s. n.s. n.s. n.s.
n.s.
FIGURE 7 Effects of microbial functional genes on soil P bioavailability in agro- and natural ecosystems across China. Effects of dif ferent
functional genes on bioavailable soil P are assessed by the linear mixed- effects models with study site identity as a random effect. Effect
sizes of individual predic tor variables are normalized coefficients from the linear mixed- effects models. The solid circles represent significant
effects (p < .05), and the open circles represent non- significant effect s (p > .05). The explained variance (R2) of the model is shown in blue.
The homologous genes encoding ACP include aphA, olpA and phoN. The homologous genes encoding ALP include phoA, phoD and phoX. The
homologous genes encoding phytase include appA and phy. n, the number of samples
Normalized effect on soil bioavailable P
-0.50 -0.25 0.00 0.25 0.50 0.75
Normalized effect on soil bioavailable P
-0.30-0.15 0.00 0.15 0.30 0.45
phy
phoX
phoN
phoD
phoA
olpA
appA
aphA
phy
phoX
phoN
olpA
appA
aphA
Agroecosystem (
n = 87
)
Natural ecosystem
(
n = 120
)
(a) (b)
R² = 0.55 R² = 0.52
10
|
LU e t aL.
abundance data on microbial functional genes encoding enzymes
responsible for C cycling into a microbially enabled ecosystem
model were able to significantly improve the modeling perfor-
mance of soil microbial respiration and reduce model parametric
uncertainty (Guo et al., 2020).
In this study, ACP- encoding genes (i.e., phoN, olpA and aphA)
were all important predictors of soil P bioavailability of agroecosys-
tems (Figure 6a), whereas none of them was an important predictor
of soil P bioavailability of natural ecosystems (Figure 6b). The pattern
was generally consistent with our obser vation that ACP was an im-
portant predictor of soil P bioavailability of agroecosystems but not
of natural ecosystems (Figure 2). Among these three homologous
genes encoding ACP, phoN was the most impor tant predictor of soil
P bioavailability of agroecosystems (Figure 6a), which was at tributed
at least partly to the fact that its average relative abundance was
higher than those of olpA and aphA (Table S6). Indeed, the numerical
predominance of phoN was reported previously for soils of different
ecosystem types (Neal et al., 2018). An unexpected finding was that
phoX (a gene encoding ALP) rather than ACP- encoding genes had a
positive ef fect on soil P bioavailability of agroecosystems (Figure 6a).
At this stage, we cannot provide an explanation for the seemingly in-
consistent results from different statistical methods.
Although the ALP was found to have a weak effect on soil P bio-
availability of agro- and natural ecosystems across China (Figures 2
and 3), phoX was found to be an important predic tor of soil P bioavail-
ability of both ecosystem types (Figure 6). In addition, the gene had a
significant positive ef fect on soil P bioavailability of agroecosystems
(Figure 7a). This discrepancy was likely explained by the contrasting
effects of the three homologous genes encoding ALP in predicting
soil P bioavailability (Figures 6 and 7), suggesting the need to explore
microbial factors at gene level. A possible reason for the important
role of phoX lied in that PhoX generally has a broad substrate range
(Wu et al., 2007; Zaheer et al., 2009). Despite this, it was somewhat
surprising to find that phoD was not a significant predictor of soil P
bioavailability of both agro- and natural ecosystems across China
(Figure 6), given that the gene has been frequently reported to be a
main factor affecting soil P bioavailability at small spatial scales (Luo
et al., 2009; Ragot et al., 2015; Tan et al., 2013). While our results
could be also attributed partly to the possible functional redundancy
of phoD (Louca et al., 2018), we propose that the role of phoX in deter-
mining soil P bioavailability should not be ignored any longer.
In terms of the homologous genes encoding phytase (i.e., phy
and appA), their relative importance in determining soil P bioavail-
ability was also found to vary considerably between ecosystem
types. More specifically, appA was superior to phy in agroecosys-
tems across China (Figure 6a), while the opposite was true for nat-
ural ecosystems (Figures 6b and 7b). These contrasting patterns
could be attributed partly to a scenario that phyt ases encoded by
these two genes remove phosphate groups from phytates in dif-
ferent C ring positions (Fonseca- Maldonado et al., 2014; Olazaran
et al., 2010).
Irrespective of its relative impor tance, the ALP- encoding
gene phoX was the only functional gene identified as a significant
predictor of soil P bioavailability of both agro- and natural eco-
systems across China (Figure 6). It thus holds promise to develop
gene- informed Earth System Models (ESMs) for soil P bioavailabil-
ity at large spatial scales (Chen & Sinsabaugh, 2021). This is espe-
cially the case if further studies demonstrate that our finding is
applicable at global scale. Nonetheless, once we can predict soil P
bioavailability at large spatial scales in changing environments via
ESMs, we will have the abilit y to provide more sust ainable manage-
ment options for the widespread P- limited terrestrial ecosystems.
5 | CONCLUSIONS
In conclusion, our result s provided the first evidence that phytase
could influence soil P bioavailability at country scale and that
it mediated the effects of several abiotic factors on soil P bio-
availability. Moreover, we showed that many microbial functional
genes encoding P- mineralizing enzymes (including ACP, ALP and
phytase) were significant predictors of soil P bioavailability at a
large spatial scale although their relative importance varied con-
siderably between ecosystem types. Taken together, our find-
ings revealed the remarkable effects of microbial factors on soil
P bioavailability at a country scale, highlighting the importance to
consider microbial factors in managing P- limited terrestrial eco-
systems distributed globally.
ACKNOWLEDGMENTS
We thank Professor AJM Baker (Universities of Melbourne and
Queensland, Australia, and Shef field, UK ) for his help in English lan-
guage editing. This work was supported financially by the National
Natural Science Foundation of China (Nos. 41622106, 42177009
and 42077117), the Key- Area Research and Development Program
of Guangdong Province (No. 2019B110207001), Guangdong Basic
and Applied Basic Research Foundation (No. 2021B1515120039)
and Natural Science Foundation of Guangdong Province of China
(No. 2020A1515010937).
CONFLICT OF INTEREST
The authors declare no conflict of interest.
AUTHORS' CONTRIBUTIONS
JTL and J- LL developed and framed research questions; JLL, JP,
SWF, JZ, SNOu, ZHW and BL conducted the experiments; JLL , PJ
and YT W performed the data analyses; JLL , PJ, J- LL and JTL wrote
the first draft of the manuscript; all authors revised the manuscript.
DATA AVAIL ABILI TY STATEMENT
The supplementary tables including soil physicochemical and microbial
data that support the findings of this study are available in Dryad at
https://doi.org/10.5061/dryad.r2280 gbfr. The metagenomic datasets
have been deposited at EMBL under accession number PRJEB43404
(ERP127366). The code used for statistical analyses is openly available
at https://github.com/1108J ingli/ R- code- for- soil- P- bioav ailab ility.git.
|
11
LU et aL.
ORCID
Jing- li Lu https://orcid.org/0000-0001-8776-2208
Pu Jia https://orcid.org/0000-0002-3451-2798
Shi- wei Feng https://orcid.org/0000-0003-4532-9224
Jie- Liang Liang https://orcid.org/0000-0002-3949-7486
Jin- tian Li https://orcid.org/0000-0002-0848-5730
REFERENCES
Achat, D. L ., Pousse, N., Nicolas, M., Brédoire, F., & Augusto, L. (2016).
Soil properties controlling inorganic phosphorus availability:
General results from a national forest network and a global com-
pilation of the literature. Biogeochemistry, 127(2), 255– 272. h t t ps : //
doi.org/10 .1007/s1053 3- 015- 0178- 0
Al- Kaisi, M. M., Yin, X., & Licht, M. A. (2005). Soil c arbon and nitrogen
changes as influenced by tillage and cropping systems in some Iowa
soils. Agriculture, Ecosystems and Environment, 105(4), 635– 647.
https://doi.org/10.1016/j.agee.2004.08.002
Alori, E. T., Glick, B. R., & Babalola, O. O. (2017). Microbial phosphorus sol-
ubilizat ion and its potent ial for use in susta inable agricult ure. Frontier s
in Microbiology, 8, 971. https://doi.org/10.3389/fmicb.2017.00971
Archer, E. (2021). rfPermute: Estimate permutation p- values for ran-
dom forest import ant metrics. R package version 2.2. ht t p s ://
CRAN.R- proje ct.org/packa ge=rfPer mute
Augusto, L., Achat, D. L ., Jonard, M., Vidal, D., & Ringeval, B. (2017).
Soil parent material— A major driver of plant nutrient limit ations in
terrestrial ecosystems. Global Change Biology, 23(9), 3808– 3824.
https://doi .org /10.1111/gcb.13691
Bahra m, M., Hildeb rand, F., Forslun d, S. K., An derson, J. L ., Soudzil ovskaia,
N. A., Bodegom, P. M., Beng tsson- Palme, J., Anslan, S., Coelho,
L. P., Harend, H., Huerta- Cepas, J., Medema, M. H., Maltz, M. R.,
Mundra , S., Olsson, P. A., Pent, M., Põlme, S., Sunagawa , S., Ryberg,
M., … Bork, P. (2018). Structure and function of the global topsoil
microbiome. Nature, 560(7717), 233– 237. ht tps: //doi.org/10.1038/
s4158 6- 018- 0386- 6
Bivand, R., Keitt, T., & Rowlingson, B. (2021). Rgdal: Bindings for the
‘Geospatial’ data abstraction librar y. R package version 1.5– 28.
https://CRAN.R- proje ct .org/packa ge=rgdal
Bohan, D. A ., Raybould, A., Mulder, C., Woodward, G., Tamaddoni-
Nezhad, A., Bluthgen, N., Pocock, M. J. O., Muggleton, S., Evans, D.
M., Astegiano, J., Massol, F., Loeuille, N., Petit, S., & Macfadyen, S.
(2013). Networking agroecology: Integrating the diversity of agro-
ecosystem interactions. Advances in Ecological Research, 49, 1– 67.
https://doi.org/10.1016/B978- 0- 12- 42000 2- 9.00001 - 9
Bouskill, N. J., Riley, W. J., Zhu, Q., Mekonnen, Z . A., & Grant, R . F. (2020).
Alaskan carbon- climate feedbacks will be weaker than inferred
from short- term experiments. Nature Communications, 11, 5798.
https://doi.org/10.1038/s4146 7- 020- 19574 - 3
Breiman , L. (2001). Random forest. Machine Learning, 45(1), 5– 32.
htt ps://doi.o rg/10.10 23/A:10109 3 340 4324
Brookes, P. C., Landman, A., Pruden, G ., & Jenkinson, D. S. (1985).
Chloroform fumigation and the release of soil nitrogen: A rapid di-
rect ex traction method for measuring microbial biomass nitrogen
in soil. Soil Biology and Biochemistry, 17(6), 837– 842. ht t p s : //d o i.
org /10.1016/003 8- 0717(85)9 0144 - 0
Brookes, P. C., Powlson, D. S., & Jenkinson, D. S. (1984). Phosphorus in
the soil microbial biomass. Soil Biology and Biochemistry, 16(2), 169–
175. ht tps: //doi.org/10.1016/0038- 0717(84)9 0108 - 1
Chadwick, O. A., Derry, L. A., Vitousek, P. M., Huebert, B. J., & Hedin, L.
O. (1999). Changing sources of nutrients during four million years of
ecosystem development. Nature, 397(6719), 491– 497. h t tp s : //d o i.
org /10.103 8/17276
Chen, J., & Sinsabaugh, R. L. (2021). Linking microbial functional gene
abundance and soil extracellular enzyme activity: Implications for
soil carbon dynamics. Global Change Biology, 27(7), 132 2– 1325.
https://doi .org /10.1111/gcb.155 06
Chen, J. S., & Chiu, C. Y. (2003). Characterization of soil organic matter in
different particle- size fractions in humid subalp ine soils by CP/MAS
13C NMR. Geoderma, 117(1– 2), 129– 141. ht tps://doi.o rg/10.1016/
S0016 - 7061(03) 00160 - 5
Cross, A . F., & Schlesinger, W. H. (1995). A literature review and evalua-
tion of the Hedley frac tionation: A pplications to the biogeochemical
cycle of soil phosphorus in natural ecosystems. Geoderma, 64(3– 4),
197– 214 . https://doi.org/10.1016/0016- 7061(94)00023 - 4
Delgado- Baquerizo, M., Maestre, F. T., Gallardo, A., Bowker, M. A .,
Wallenstein, M. D., Quero, J. L., Ochoa, V., Gozalo, B., García-
Gómez, M., Soliveres, S., García- Palacios, P., Berdugo, M.,
Valencia, E., Escolar, C., Arredondo, T., Barraza- Zepeda, C., Bran,
D., Carreira, J. A., Chaieb, M., … Zaady, E. (2013). Decoupling of soil
nutrient cycles as a function of aridity in global drylands. Nature,
502(7473), 672– 676. https://doi.org/10.1038/na tur e12670
Dodd, J. C . (2000). The role of arbuscular mycorrhizal fungi in agro- and
natural ecosystems. Outlook on Agriculture, 29(1), 55– 62. h t tp s : //
doi.org/10.5367/00000 00001 01293059
Eddy, S. R. (1998). Profile hidden Markov models. Bioinformatics, 14 (9),
755– 763. https://doi.org/10.1093/bioin forma tics/14.9.755
Eeckhout, W., & Paepe, M. D. (1994). Total phosphorus, phytate-
phosphorus and phytase activity in plant feedstuf fs. Animal
Feed Science and Technology, 47(1– 2 ), 19– 29. ht t p s : //d oi .
org/10.1016/0377- 8401(94)90156 - 2
Elser, J. J. (2012). Phosphorus: A limiting nutrient for humanity? Current
Opinion in Biotechnology, 23(6), 833– 838. https ://doi.org/10.1016/j.
copbio.2012.03.001
Elser, J. J., Bracken, M. E. S., Cleland, E. E., Gruner, D. S., Harpole, W.
S., Hillebrand, H., Ngai, J. T., Seabloom, E. W., Shurin, J. B., &
Smith, J. E. (2007). Global analysis of nitrogen and phosphorus
limitation of primar y producers in freshwater, marine and terres-
trial ecosystems. Ecology Letters, 10(12), 1135– 1142. ht t p s: //d o i .
org /10.1111/j.1461- 0248 .20 07.01113.x
Fanin, N., B ezaud, S., Sarneel, J. M., Cecchini, S., Nicolas, M., & Augusto,
L. (2020). Relative importance of climate, soil and plant func tional
traits during the early decomposition stage of standardized litter.
Ecosys tems, 23(5), 1004– 1018. https://doi.org/10 .1007/s1002 1-
019- 00452 - z
Fonseca- Maldonado, R., Maller, A., Bonneil, E., Thibault, P., Machado, C.
B., Ward, R . J., & Polizeli, M . L. T. M. (2014). Biochemical properties
of glycosylation and characterization of a histidine acid phospha-
tase (phy tase) expressed in Pichia pastoris. Protein Expression and
Purification, 99(7), 43– 49. https://doi.org/10.1016/j.pep.2014.03.006
Fortmann- Roe, S. (2013). Consistent and clear reporting of results from
diverse modeling techniques: The A3 method. Journal of Statistical
Software, 66(7), 1– 23. https://doi.org/10.18637/ jss.v066.i07
Fuhrer, J. (2003). Agroecosystem responses to combinations of elevated
CO2, ozone, and global climate change. Agriculture, Ecosystems
and Environment, 97( 1– 3), 1– 20. ht tps://doi.o rg/10.1016/S0167
- 8809(03)00125 - 7
Geng, Y., Baumann, F., Song, C., Zhang, M., Shi, Y., Kühn, P., Scholten,
T., & He, J. S. (2017). Increasing temperature reduces the coupling
between available nitrogen and phosphorus in soils of Chinese
grasslands. Scientific Repor ts, 7(1), 43524. https://doi.org/10.1038/
srep4 3524
George, T. S., Quiquampoix, H., Simpson, R. J., & Richardson, A . E.
(2007). Interactions between phytases and soil constituents:
Implications for the hydrolysis of inositol phosphates. In B. L.
Turner, A. E. Richardson, & E. J. Mullaney (Eds.), Inositol phos-
phates: Linking agriculture and the environment (pp. 221– 241). CAB
International.
Grace, J. B. (2006). Structural equation modeling and natural sys tems.
Cambridge University Press.
12
|
LU e t aL.
Guo, X., Gao, Q., Yuan, M., Wang, G., Zhou, X., Feng, J., Shi, Z., Hale, L.,
Wu, L., Zhou, A., Tian, R., Liu, F., Wu, B., Chen, L., Jung, C. G., Niu,
S., Li, D., Xu, X., Jiang, L ., … Zhou, J. (2020). Gene- informed de-
composition model predicts lower soil carb on loss due to persistent
microbial adaptation to warming. Nature Communications, 11, 4897.
https://doi.org/10.1038/s4146 7- 020- 18706 - z
Hijmans, R. J. (2021). Raster: Geographic data analysis and modeling.
R package version 3.5– 9. https://CRAN.R- proje ct.org/packa ge=
raster
Hou, E., Chen, C., Luo, Y., Zhou, G., Kuang, Y., Zhang, Y., Heenan , M., Lu,
X., & Wen, D. (2018). Effects of climate on soil phosphorus cycle
and availability in natural terrestrial ecos ystems. Global Change
Biology, 24(8), 3344– 3356. http s://doi.org /10.1111/gcb.140 93
Hou, E., Luo, Y., Kuang, Y., Chen, C ., Lu, X., Jiang, L., Luo, X., & Wen,
D. (2020). Global meta- analysis shows pervasive phosphorus lim-
itation of aboveground plant production in natural terrestrial eco-
systems. Nature Communications, 11, 637. https://doi.org/10.1038/
s4146 7- 020- 14492 - w
Jenkinson, D. S. (1981). Microbial biomass in soil: M easurement and turn-
ove r. Soil Biochemistry, 5, 415– 471.
Joergensen, R. G. (1996). The fumigation- extraction method to es-
timate soil microbial biomass: Calibration of the kEC value.
Soil Biology and Biochemistry, 28(1), 25– 31. ht t p s : //d oi .
org /10.1016/0038- 0717(95)0 0102 - 6
Juma, N. G., & Tabatabai, M. A. (1978). Distribution of phosphomon-
oesterases in soils. Soil Science, 126(2), 101– 108. ht t ps : //d o i .
org/10.1097/00010 694- 19780 8000- 00006
Kieft , T. L., White, C. S., Loftin, S. R., Aguilar, R., Craig , J. A., & Skaar, D.
A. (1998). Temporal dynamics in soil carbon and nitrogen resources
at a grassland- shrubland ecotone. Ecology, 79(2), 671– 683. ht t p s: //
doi.org/10.1890/0012- 9658(1998)079[0671:TDISC A]2.0.CO;2
Kock, N. (2015). Common method bias in PLS- SEM: A full collinearit y
assessment approach. International Journal of e- Collaboration, 11(4),
1– 10 . https://doi.org/10 .4018/ ijec .20151 0010
Kuznetsova, A., Brockhoff, P. B., & Christensen, R. H. (2017). lmerTest
package: Tests in linear mixed effects models. Jou rnal of Statistical
Software, 82 (13), 1– 26. https://doi.org/10.18637/ jss.v082.i13
Lefchec k, J. S. (2016). piecewis eSEM: Piecewis e structur al equation mo del-
ing in R for ec ology, evolutio n, and system atics. Meth ods in Ecology a nd
Evolution, 7(5), 573– 579. https://doi.org/10.1111/2041- 210X.12512
Li, J. T., Lu, J. L., Wang, H. Y., Fang, Z., Wang, X. J., Feng, S. W., Wang,
Z., Yuan, T., Zhang , S. C., Ou, S. N., Yang, X. D., Wu, Z. H., Du, X.
D., Tang, L. Y., Liao, B., Shu, W. S., Jia, P., & Liang, J. L. (2021). A
comprehensive synthesis unveils the mysteries of phosphate solu-
bilizing microbes. Biological Reviews, 96 (6), 2771– 2793. h t t ps : //d o i .
org /10.1111/brv.12779
Li, S. P., Jia, P., Fan, S. Y., Wu, Y., Liu, X., Meng, Y., Li, Y., Shu, W. S., Li, J.
T., & Jiang, L. (2022). Functional traits explain the consistent resis-
tance of biodiversity to plant invasion under nitrogen enrichment.
Ecology Letters, 25(4), 778– 789. ht tps://doi.org/10 .1111/ele .13951
Liang, C ., Schimel, J. P., & Jastrow, J. D. (2017). The impor tance of anabo-
lism in micr obial contr ol over soil ca rbon stor age. Nature Microbiology,
2(8), 17105. https://doi.org/10.1038/nmicr obiol.2017.105
Liang, J. L ., Liu, J., Jia, P., Yang, T. T., Zeng, Q. W., Zhang, S . C., Liao, B.,
Shu, W. S., & Li, J. T. (2020). Novel phosphate solubilizing bacteria
enhance soil phosphorus cycling following ecological restoration
of land degraded by mining. The ISME Journal, 14 (6), 1600– 1613.
https://doi.org/10.1038/s4139 6- 020- 0632- 4
Liaw, A., & Wiener, M. (2002). Classification and regression by random-
Forest. R News, 2(3), 18– 22.
Lityński, T., Jurkowska, H ., & Gorlach, E. (1976). Analiza chemiczno-
rolnicza. PWN.
Lott, J. N. A., Ockenden, I., Raboy, V., & Batten, G. D. (200 0). Phytic acid and
phosphorus in crop seeds and fruits: A global estimate. Seed Science
Research, 10(1), 11– 33. https://doi.org/10.1017/S0960 25850 0000039
Louca, S., Polz, M. F., Mazel, F., Albright, M., Huber, J. A., O'Connor, M. I.,
Ackermann, M., Hahn, A. S., Srivastava, D. S., Crowe, S. A., Doebeli,
M., & Par frey, L. W. (2018). Function and functional redundancy
in microbial systems. Nature Ecolog y and Evolution, 2(6), 936– 943.
htt ps://doi.o rg/10.10 38/s4155 9- 018- 0519- 1
Luo, H., Benner, R., Long , R. A., & Hu, J. (20 09). Subcellular localization of
marine bacterial alkaline phosphatases. Proceedings of the National
Academy of S ciences of the United State s of America, 106(5 0), 21219–
21223. https://doi.org/10 .1073/pnas.09075 86106
Menezes- Blackburn, D., Jorquera, M. A., Greiner, R., Gianfreda, L., &
Dela, L . M. M. (2013). Phytases and phytase- labile organic phos-
phorus i n manures and so ils. Critical Revi ews in Environme ntal Science
and Technology, 43(9), 916– 954. https://doi.org/10.10 80/106 43
389.2011.627019
Mise, K ., Fujita, K., Kun ito, T., Senoo, K., & Otsu ka, S. (2018). Phosph orus-
mineralizing communities reflect nutrient- rich characteristics in
Japanese arable andisols. Microbes and Environments, 33(3), 282–
28 9. https://doi.org/10.1264/jsme2.ME18043
Morel, C ., Tunney, H., Plénet, D., & Pellerin, S. (2000). Transfer of phos-
phate ions between soil and solution: Perspectives in soil test-
ing. Journal of Environmental Quality, 29(1), 50– 59. h t t p s: //d o i .
org/10.2134/jeq20 00.00472 42500 29000 10007x
Mullen, M. D. (2005). Phosphorus in soil: Biological interac tions. In D.
Hillel (Ed.), Encyclopedia of soils in the environment (pp. 210– 215).
Elsevier Ltd.
Murphy, M. (2020). semEff: Automatic calculation of effects for piece-
wise structural equation models. R package version 0.6.0. ht t p s: //
CRAN.R- proje ct.org/packa ge=semEff
Nannipieri, P., Giagnoni, L ., Landi, L., & Renella , G. (2011). Role of phos-
phatase enzymes in soil. In E. Bünemann, A. Oberson, & E. Frossard
(Eds.), Phosphorus in action (pp. 215– 243). Springer.
Neal, A . L., Blackwell, M., A kkari, E., Guyomar, C., Clark, I., & Hirsch, P.
R. (2018). Phylogenetic distribution, biogeogr aphy and the ef fects
of land management upon bacterial non- specific acid phosphat ase
gene diversity and abundance. Plant and Soil, 427 (1– 2), 175– 189.
htt ps://doi.o rg/10.10 07/s1110 4- 017- 3301- 2
Oelmann, Y., Lange, M., Leimer, S., Roscher, C., Aburto, F., Alt, F., Bange,
N., Berner, D., Boch, S., Boeddinghaus, R. S., Buscot, F., Dassen, S.,
De Deyn, G., Eisenhauer, N., Gleixner, G., Goldmann, K., Hölzel, N.,
Jochum, M., Kandeler, E., … Wilcke, W. (2021). Ab ove- and below-
ground biodiversit y jointly tighten the P cycle in agricultural grass-
lands. Nature Communications, 12, 4431. http s://doi.or g/10.1038/
s4146 7- 021- 24714 - 4
Olazar an, M. G., Blanco, L. R., Trevino, J. G. C., Lopez, J. A. G., & Salvado,
J. M. V. (2010). Expression of a bacillus phytase C gene in Pichia
pastoris and properties of the recombinant enzyme. Applied
and Environmental Microbiology, 76 (16), 5601– 5608. ht t p s : //d o i.
org /10.1128/AEM .00762 - 10
Olsen, S . R., Cole, C . V., Watanabe, F. S., & Dean, L. A. (1954). Estimation
of available phosphorus in soils by extraction with sodium bicarbonate.
U.S. Department of Agriculture.
Peng, Q., Dong, Y., Qi, Y., Xiao, S., He, Y., & Ma, T. (2011). Effects of nitro-
gen fertilization on soil respiration in temperate grassland in Inner
Mongolia , China. Environmental Earth Sciences, 62(6), 1163– 1171.
https://doi.org/10.1007/s1266 5- 010- 0605- 4
Pierce, D. (2021). ncdf4: Interface to unidata netCDF (version 4 or ear-
lier) format data files. R package ver sion 1.19. https://CRAN.R-
proje ct.org/packa ge=ncdf4
Pineiro, G ., Perelman, S., Guerschman, J. P., & Paruelo, J. M. (2008).
How to evaluate models: Observed vs. predicted or predicte d vs.
observed? Ecological Modelling, 216 (3– 4), 316– 322. ht t p s : //d o i.
org/10.1016/j.ecolm odel.2008.05.006
R Core Team. (2021). R: A language and environment for statistical comput-
ing. R Foundation for Statistical Computing. https://www.R- proje
ct.org
|
13
LU et aL.
Ragot, S. A., Kertesz, M. A., & Bünemann, E. K. (2015). phoD alkaline
phosphatase gene diversity in soil. Applied and Environmental
Microbiology, 81(20), 7281– 7289. https://doi.org/10.1128/AEM.
01823 - 15
Rao, C., Coyte, K. Z., Bainter, W., Geha, R . S., Martin, C. R., & Rakoff-
Nahoum, S. (2021). Multi- kingdom ecological driver s of microbiota
assembly in preterm infants. Nature, 591(7851), 633– 638. h t t p s ://
do i.org /10 .10 38/s4158 6 - 021- 03 241 - 8
Richardson, A. E., George, T. S., Hens, M., & Simpson, R. J. (20 04).
Utilisation of soil organic phosphorus by higher plants. In B. L.
Turner, E. Frossard, & D. Baldwin (Eds.), Organic phosphorus in the
environment (pp. 165– 184). CAB International.
Richardson, A. E., & Simpson, R . J. (2011). Soil microorganisms me-
diating phosphorus availability update on microbial phospho-
rus. Plant Physiolog y, 156(3), 989– 996. ht tps://doi.o rg/10.110 4/
pp.111 .1754 48
Ringeval, B., Augusto, L., Monod, H., Van Apeldoorn, D., Bouwman, L.,
Yang, X., Achat, D. L., Chini, L. P., Van Oost, K., Guenet, B., Wang,
R., Decharme, T. N., & Pellerin, S. (2017). Phosphorus in agricultural
soils: Drivers of its distribution at the global scale. Global Change
Biology, 23(8), 3418– 3432. htt ps://doi.or g/10.1111 /gcb.13 618
Rodríguez, H., & Fraga , R. (1999). Phosphate solubilizing bacteria and their
role in plant growth promotion. Biotechnology Advances, 17(4– 5),
319– 33 9. https://doi.org/10.1016/S0734 - 9750(9 9)00014 - 2
Rodríguez, H., Fraga , R., Gonzalez, T., & Bashan, Y. (2006). Genetics of
phosphate solubilization and its potential applications for improv-
ing plant growth- promoting bacteria. Plant and Soil, 287(1– 2), 15–
21. https://doi.org/10.1007/s1110 4- 006- 9056- 9
Schermelleh- Engel, K., & Moosbrugger, H. (2003). Evaluating the fit of struc-
tural equation models: Tests of significance and descriptive goodness-
of- fit measures. Methods of Psychological Research, 8(2 ), 23– 74.
Shipley, B. (2001). Cause and correlation in biology: A user's guide to
path analysis structural equations and causal inference. Cambridge
University Press.
Sparks, D. L., & Sparks, D. L. (1996). Methods of soil analysis. In J. M.
Bigham (Ed.), Chemical methods (pp. 870– 874). Soil Science Society
of America.
Sternberg, T. (2015). Deser t boundaries: The once and future Gobi.
The Geogra phical Journal, 181(1), 61– 72 . http s://doi. org /10.1111/
geoj .12071
Tabatabai, M. A. (1994). Soil enzymes. In R. W. Weaver, J. S. Angle, & P.
S. Bottomley (Eds.), Methods of soil analysis: Microbiological and bio-
chemical properties (pp. 775– 883). Soil Science Society of America.
Tan, H., Barret, M., Mooij, M., Rice, O., Morrissey, J., Dobson, A.,
Griffiths, B., & O'Gara, F. (2013). Long- term phosphorus fertilisa-
tion increased the diversity of the total bacterial community and
the phoD phosphorus mineraliser group in pasture soils. Biology
and Fertility of Soils, 49(6), 661– 672. ht tps://doi.o rg/10.1007/s0037
4- 012- 0755- 5
Turner, B. L., Papházy, M. J., Haygarth, P. M., & McKelvie, I. D. (2002).
Inositol phosphates in the environment. Philosophical Transactions
of the Royal Soc iety of London. S eries B: Biologic al Sciences, 357(1420),
449– 469. https://doi.org/10.1098/rstb.2001.0837
Walker, T. W., & Syers, J. K. (1976). The fate of phosphorus
during pedogenesis. Geoderma, 15(1), 1– 19. h t tp s : //d o i.
org/10.1016/0016- 7061(76)90066 - 5
Wang, C., Xue, L., & Jiao, R. (2021). Soil phosphorus fractions , phos-
phatase activit y, and the abundance of phoC and phoD genes vary
with planting densit y in subtropical Chinese fir plantations. Soil
and Tillage Research, 209(4), 104946 . http s://doi.or g/10.1016/ j.
still.2021.104946
Wardle, D. A ., Walker, L. R., & Bardgett, R . D. (2004). Ecosystem proper-
ties and forest decline in contrasting long- term chronosequences.
Science, 305(5683), 509– 513. https://doi.org/10.1016/j.soilb
io.2021.108144
Wu, H., Xiang, W., Ouyang, S., Forrester, D. I., Zhou, B., Chen, L., Ge, T.,
Lei, P., Chen, L., Zeng, Y., Song, X ., Peñuelas, J., & Peng, C. (2019).
Linkage between tree species richness and soil microbial diver-
sity improves phosphorus bioavailability. Functional Ecology, 33(8),
1549– 1560. htt ps://doi .org/10.1111/1365- 24 35.13 355
Wu, J., Joergensen, R. G., Pommerening, B., Chaussod, R., & Brookes, P.
C. (1990). Measurement of soil microbial biomass C by fumigation-
extraction— An automated procedure. Soil Biology and Biochemistry,
22(8), 1167– 1169. https://doi.org/10.1016/0038- 0717(90)90046 - 3
Wu, J. R., Shien, J. H., Shieh, H. K., Hu, C. C., Gong, S. R., Chen, L. Y., &
Chang, P. C. (20 07). Cloning of the gene and characterization of the
enzymatic properties of the monomeric alkaline phosphatase (PhoX)
from Pasteurella multocida strain X- 73. FEMS Microbiology Letters,
267(1), 113– 120. https://doi.org/10.1111/j.1574- 6968.2006.00542.x
Xu, D., Dell, B., Malajczuk, N., & Gong, M. (2002). Effects of P fertilisa-
tion on productivity and nutrient accumulation in a Eucaly ptus gran-
dis × E. urophylla plantation in southern China. Forest Ecology and
Management, 161(1– 3), 89– 100. http s://doi.org/10.1016/S0378
- 1127(01)00485 - 6
Yang, X., & Pos t, W. M. (2011). Phosphorus transformations as a function
of pedogenesis: A synthesis of soil phosphorus dat a using Hedley
fractionation method. Biogeosciences, 8(10), 2907– 2916. ht t p s ://
doi.org/10.5194/bg- 8- 2907- 2011
Yang, X., Pos t, W. M., Thornton , P. E ., & Jain, A. (2013). Th e distributio n of
soil phosphorus for global biogeochemical modeling. Biogeosciences,
10(4), 2525– 2537. https ://doi.org/10.5194/bg- 10- 2525- 2013
Yao, Q., Li, Z., Song, Y., Wright, S. J., Guo, X., Tringe, S. G., Tfaily, M. M.,
Paša- Tolić, L., Hazen, T. C., Turner, B. L., Mayes, M . A ., & Pan, C .
(2018). Community proteogenomics reveals the sys temic im-
pact of phosphorus availability on microbial functions in tropi-
cal soil. Nature Ecology and Evolution, 2(3), 499– 509. h t t p s: //d o i .
org /10.1038/s 4155 9- 017- 0 463- 5
Zaheer, R., Morton, R ., Proudfoot, M., Yakunin, A., & Finan, T. M. (2009).
Genetic and biochemical properties of an alkaline phosphatase PhoX
family p rotein fo und in many b acte ria. Environmental Microbiology, 11(6),
1572– 1587. https://doi.org/10.1111/j.1462- 2920.2009.01885.x
SUPPORTING INFORMATION
Additional supporting information may be found in the online
version of the article at the publisher ’s website.
How to cite this article: Lu, J.-l., Jia, P., Feng, S.-w., Wang,
Y.-t., Zheng, J., Ou, S.-n., Wu, Z.-h., Liao, B., Shu, W.-s., Liang,
J.-L., Li, J.-t. (2022). Remarkable effect s of microbial factors
on soil phosphorus bioavailability: A country- scale study.
Global Change Biology, 00, 1–13. https: //doi.or g/10.1111/
gcb.16213
