Page 1
SOIL, 6, 53–88, 2020
https://doi.org/10.5194/soil-6-53-2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.
SOIL
Variations in soil chemical and physical properties
explain basin-wide Amazon forest soil carbon
concentrations
Carlos Alberto Quesada
1
, Claudia Paz
1,2
, Erick Oblitas Mendoza
1
, Oliver Lawrence Phillips
3
,
Gustavo Saiz
4,5
, and Jon Lloyd
4,6,7
1
Instituto Nacional de Pesquisas da Amazônia, Manaus, Cx. Postal 2223 – CEP 69080-971, Brazil
2
Universidade Estadual Paulista, Departamento de Ecologia, CEP 15506-900, Rio Claro, São Paulo, Brazil
3
School of Geography, University of Leeds, LS2 9JT, UK
4
Department of Life Sciences, Imperial College London, Silwood Park Campus, Buckhurst Road, Ascot,
Berkshire SL5 7PY, UK
5
Department of Environmental Chemistry, Faculty of Sciences, Universidad Católica de la Santísima
Concepción, Concepción, Chile
6
School of Tropical and Marine Sciences and Centre for Terrestrial Environmental and Sustainability Sciences,
James Cook University, Cairns, 4870, Queensland, Australia
7
Universidade de São Paulo, Faculdade de Filosofia Ciências e Letras de Ribeirão Preto, Av Bandeirantes,
3900, CEP 14040-901, Bairro Monte Alegre , Ribeirão Preto, São Paulo, Brazil
Correspondence:
Carlos Alberto Quesada (quesada.beto@gmail.com)
Received: 26 April 2019 – Discussion started: 11 June 2019
Revised: 8 November 2019 – Accepted: 2 December 2019 – Published: 11 February 2020
Abstract.
We investigate the edaphic, mineralogical and climatic controls of soil organic carbon (SOC) concen-
tration utilising data from 147 primary forest soils (0–30 cm depth) sampled in eight different countries across
the Amazon Basin. Sampled across 14 different World Reference Base soil groups, our data suggest that sta-
bilisation mechanism varies with pedogenetic level. Specifically, although SOC concentrations in Ferralsols and
Acrisols were best explained by simple variations in clay content – this presumably being due to their relatively
uniform kaolinitic mineralogy – this was not the case for less weathered soils such as Alisols, Cambisols and
Plinthosols for which interactions between Al species, soil pH and litter quality are argued to be much more
important. Although for more strongly weathered soils the majority of SOC is located within the aggregate frac-
tion, for the less weathered soils most of the SOC is located within the silt and clay fractions. It thus seems that
for highly weathered soils SOC storage is mostly influenced by surface area variations arising from clay content,
with physical protection inside aggregates rendering an additional level of protection against decomposition. On
the other hand, most of the SOC in less weathered soils is associated with the precipitation of aluminium–carbon
complexes within the fine soil fraction, with this mechanism enhanced by the presence of high levels of aromatic,
carboxyl-rich organic matter compounds. Also examined as part of this study were a relatively small number of
arenic soils (viz. Arenosols and Podzols) for which there was a small but significant influence of clay and silt
content variations on SOM storage, with fractionation studies showing that particulate organic matter may ac-
count for up to 0.60 of arenic soil SOC. In contrast to what were in all cases strong influences of soil and/or
litter quality properties, after accounting for these effects neither wood productivity, above-ground biomass nor
precipitation/temperature variations were found to exert any significant influence on SOC stocks. These results
have important implications for our understanding of how Amazon forest soils are likely to respond to ongoing
and future climate changes.
Published by Copernicus Publications on behalf of the European Geosciences Union.
Page 2
54
C. A. Quesada et al.: Variations in soil chemical and physical properties
1
Introduction
The soil organic carbon (SOC) pool is a function of the
amount and quality of organic material entering the soil and
its subsequent rate of mineralisation, which can be controlled
by the various stabilisation processes that protect SOC from
decomposition (Bruun et al., 2010). For example, organic
carbon may be stabilised in mineral soils through interactions
with oxides and clay minerals (Kahle et al., 2004; Kaiser and
Guggenberger, 2003; Mikutta et al., 2007; Saidy et al., 2012;
Saiz et al., 2012; Wiseman and Püttmann, 2006), with SOC
physically entrapped in soil aggregates (Baldock and Skjem-
stad, 2000) and/or stabilised by intermolecular interactions
between SOC and the surface of clays and Fe and Al hydrox-
ides (Oades, 2018). Thus, chemical adsorption on mineral
specific surface area (SSA) has an important role in C stabil-
isation (Kahle et al., 2003; Saggar et al., 1996, 1999; Saidy
et al., 2012).
Specific surface area is itself dependent on clay miner-
alogy, with low-activity clays (LACs) being 1
:
1 alumino-
silicates such as kaolinite (hereafter simply referred to as
1
:
1 clays) with low SSA and low cation exchange capacity
(
I
E
). This contrasts with high-activity clays (HACs) which
are 2
:
1 alumino-silicates such as smectites and illites (here-
after simply referred to as 2
:
1 clays) with a much larger
I
E
and SSA (Basile-Doelsch et al., 2005; Lützow et al., 2006).
Hydrous Fe and Al oxides also provide reactive surface ar-
eas for organic matter binding, with the content of Fe and Al
oxides in soils often reported as strongly correlated to C con-
tent (Eusterhues et al., 2005; Kleber et al., 2005; Saidy et al.,
2012; Wiseman and Püttmann, 2006). Hydrous iron and Al
oxides nevertheless show different surface properties to those
of clays. Specifically, whilst surface charges of clays are pre-
dominantly negative in the tropics (Sanchez, 1976), hydrous
oxides generally have positive charges and associated anion
exchange capacities, which can further substantially vary in
extent in different oxide types and levels of crystallinity (Cor-
nell and Schwertmann, 2003). Thus, the SSA of clay and ox-
ide mixtures, their chemical nature, and the types of charge
predominant in organic matter may all play an important role
in the C stabilisation process (Saidy et al., 2012).
For acidic soils, SOC stabilisation by Fe and Al oxides is
likely to be dominated by ligand exchange (a pH-dependent
process) involving carboxyl groups of SOC and simple OH
groups on the surface of the oxides (Kaiser and Guggen-
berger, 2003; Lützow et al., 2006; Wagai and Mayer, 2007): a
similar sorption mechanism to that occurring on the edges of
1
:
1 clay minerals such as kaolinite (Oades, 2018). Iron and
Al oxides can also increase the stabilisation of SOC through
interactions with clay minerals via a promotion of the forma-
tion of aggregates which then serve to preserve SOC (Kita-
gawa, 1983; Wagai and Mayer, 2007), also forming bridges
between kaolinite and charges in organic matter which are
mainly conferred by cationic amino (R-NH
2
) and sulfhydryl
(R-SH) groups (Wiseman and Püttmann, 2006). Other fac-
tors such as the pH of soil and the organic matter loading
present in the system also influence C stabilisation by min-
eral surfaces (Saidy et al., 2012).
Hydrous oxides themselves also vary in their capacity to
stabilise C, with amorphous Fe and Al oxides having com-
paratively higher capacity to stabilise C than more crystalline
oxides (Kleber et al., 2005; Mikutta et al., 2005). For ex-
ample, on a mass basis, the C sorption capacity of ferrihy-
drite is 2.5 times higher than that of goethite (Kaiser et al.,
2007), while amorphous Al oxides have a greater sorption
capacity than ferrihydrite (Kaiser and Zech, 2000). Despite
these complexities, because many heavily weathered soils
consist primarily of kaolinite (Sanchez, 1976) it is common
to find strong relationships between [SOC] and soil clay frac-
tion when only soils dominated by 1
:
1 clays are considered
(Burke et al., 1989; Dick et al., 2005; Feller and Beare, 1997;
Telles et al., 2003).
A second process that may also protect organic matter
against microbial decay and which should be much more rel-
evant to 2
:
1 clays soils is the co-precipitation of dissolved
organic matter (DOM) with Fe and Al (Baldock and Skjem-
stad, 2000; Boudot et al., 1989; Nierop et al., 2002; Scheel
et al., 2007). DOM can be precipitated in the presence of Al,
Fe and their hydroxides, with an efficiency of up to 90 % of
all DOM present in the solution of some acidic forest soils
(Nierop et al., 2002). The extent to which DOM precipitates
is largely influenced by soil pH, with higher pH values lead-
ing to an increase in precipitation (Nierop et al., 2002). This
is because pH affects both the solubility of DOM (which
decreases at low pH) and the speciation of Al. At higher
pH levels (> 4.2) the formation of hydroxide species such as
Al(OH)
3
and tridecameric Al (Al
13
) controls the solubility of
Al, but with Al
+
3
predominating at lower pH. Moreover, the
chemical nature of the carbon inputs into a soil may also po-
tentially influence the nature and extent of any DOM precip-
itation reactions, with high molecular weight derived from
lignin and tannins (e.g. aromatic compounds) with a large
number of functional groups likely to be preferentially pre-
cipitated from DOM (Scheel et al., 2007, 2008).
The retention of such precipitated DOM in the soil can
contribute substantially to total soil C pools and is consid-
ered one of the most important processes of SOC stabilisa-
tion (Kalbitz and Kaiser, 2008). Indeed, mineralisation rates
of such metal–DOM precipitates have been reported to be
28 times lower than that of original DOM, with the resis-
tance of precipitates against microbial decay increasing with
aromatic C content and large C : N ratios. This then results in
a relatively stable pool that accumulates in the soil (Scheel
et al., 2007). Exchangeable Al concentrations are often very
high for Amazon Basin forest soils (Quesada et al., 2011),
and with Al–OM co-precipitations particularly important in
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 3
C. A. Quesada et al.: Variations in soil chemical and physical properties
55
such developing soils (Kleber et al., 2015), stabilisation of
DOM by precipitation with Al is likely to be of consider-
able importance, especially in the western area of the Ama-
zon Basin where actively evolving soils dominate (Quesada
et al., 2010).
Given the range of potential mechanisms discussed above,
no single soil property should be considered likely to have
overriding control of SOC concentrations for Amazon Basin
forest soils. And indeed, although there is a current per-
ception that clay content alone exerts strong influence over
SOC concentration of Amazon forest soils (Dick et al., 2005;
Telles et al., 2003), all of this work has been done with
highly weathered soils and with SOC from soil characterised
by 2
:
1 mineralogical assemblages not showing any sort of
simple clay content dependency (Quesada and Lloyd, 2016).
This suggests that for such soils – as has already been shown
to be the case for other regions of the world with similar pe-
dogenetic levels (Bruun et al., 2010; Percival et al., 2000) –
variations in clay quality, oxide content and metal–DOM in-
teractions are likely to be just as, if not more, important in
influencing the extent of SOC stabilisation.
With the forest soils of the Amazon Basin varying sub-
stantially in their chemical and physical properties (Quesada
et al., 2010, 2011), it is important to consider how the dif-
ferent soils of the basin may differ in the mechanisms by
which they stabilise and store SOC. Specifically, we hypoth-
esised that soil groups with contrasting pedogenetic develop-
ment should differ in their predominant mechanism of SOC
stabilisation, and that soils which share more similar chemi-
cal and mineralogical characteristics should also share simi-
lar mechanisms of SOC stabilisation. Specifically, we ratio-
nalised that strongly weathered soils dominated by 1
:
1 clays
should have their C pools influenced primarily by clay con-
tent. On the other hand, given that Al is the main product
of weathering in the less weathered soils of western Ama-
zonia (Quesada et al., 2011), and with clay contents already
shown to not explain their SOC densities well (Quesada and
Lloyd, 2016), we hypothesised that interactions between Al
and organic matter were likely to be the main stabilisation
mechanism for such soils.
Finally, soil organic matter (SOM) is a complex mixture
of carbon compounds and different soil minerals. SOM con-
sists of various functional pools, which are stabilised by dif-
ferent mechanisms, each associated with a given turnover
rate. Aiming to simplify this complexity, several soil organic
matter partitioning methods have been developed to sepa-
rate SOM in different operationally defined pools or frac-
tions with contrasting chemical and physical characteristics
(Denef et al, 2010). Such fractionation methods provide ad-
ditional support for understanding soil carbon stabilisation
mechanisms and also provide useful constraints for models
of soil carbon dynamics (Trumbore and Zheng, 1996; Zim-
mermann et al., 2007).
Here we explore the climatic, edaphic and mineralogical
conditioning of soil carbon pools across the diverse forest
Figure 1.
Geographic distribution of 147 study sites across the
Amazon Basin, according to the different soil groups. Each point is
a 1 ha forest inventory permanent plot. Geographical locations have
been manipulated in the map to allow visualisation of site clusters
at this scale.
soils of the Amazon Basin, focusing on three major ques-
tions:
1. What are the major edaphic and climatic factors ex-
plaining observed variations in soil organic C across the
basin?
2. Are the likely contrasting stabilisation mechanism pat-
terns hypothesised to operate also consistently related to
different SOC physicochemical fraction distributions?
3. How should the contrasting SOC retention mechanisms
identified above influence our understanding of the
likely responses of the Amazon Basin forests to future
changes in climate?
2
Materials and methods
2.1
Study sites and sampling
Soils of 147 1 ha primary forest plots, representing 14 soil or-
ders, were sampled across the Amazon Basin as part of this
study (Table 1). These include forests in Brazil, Venezuela,
Guyana, French Guyana, Ecuador, Colombia, Peru and Bo-
livia (Fig. 1).
Details of soil sampling protocol, laboratory analysis and
soil classification can be found in Quesada et al. (2010,
2011), which described a subset of the soils detailed here.
For each site at least five soil cores were taken across the
1 ha plot to the depth of 2.0 m, with an additional 2.0 m soil
pit also sampled in each plot. Within each soil core, sam-
ples were collected over the following standardised depths:
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 4
56
C. A. Quesada et al.: Variations in soil chemical and physical properties
T
ab
le
1.
Climate
and
site
details
and
summary
of
soil
ph
ysical
and
chemical
characteristics
(0.0–0.3
m).
Abbre
viations
used:
T
A:
mean
annual
temperature;
P
A:
mean
annual
pre-
cipitation;
EV
:
ele
v
ation;
6
B
:
sum
of
bases;
IE:
ef
fecti
v
e
cat
ion
exchange
capacity;
6
B
(
R
):
total
reserv
e
bases;
Ch:
Chlorite;
Gi:
gibbsit
e;
Go:
goethite;
He:
Haematite;
Il:
Illite;
Ka:
kaolinite;
Mi:
Mica;
Mu:
Musco
vite;
Or/K:
orthoclase/K-feldspar;
Pl:
Plagioglase;
Sm:
Smectite;
Albite:
Al;
Microcline:
Mc;
ND
–
not
determined.
Soils
from
the
are
nic
group
(Arenosols/Podzols)
follo
wed
by
F
indicate
seasonally
flooded
white
sands.
F
or
the
mineralogies,
blank
columns
indicat
e
that
measurements
were
not
made;
∗
:
identification
uncertain;
0:
none
identified.
Sites
ha
v
e
been
numbered
and
ordered
according
to
their
upper
-layer
(0.0–0.3
m)
soil
C
content
as
gi
v
en
in
T
able
A1
(Appendix).
Soil
Classification
Locat
ion
T
A
P
A
E
V
pH
P
article
fraction
6
B
I
E
6
B(R)
Mineralogy
(
◦
C)
(mm)
(m)
Sand
Clay
Silt
(mmol
c
kg
−
1
)
1
◦
2
◦
1
Plinthosols
Braz
il,
Acre
25.1
1705.1
260
4.57
0.61
0.16
0.22
7.1
13.2
189.3
Ka
Mu,
Go,
He
2
Gle
ysols
Per
u,
north
26.3
2751.5
126
4.26
0.53
0.21
0.27
4.0
36.2
40.6
Mi
Ka
3
Cambisols
Per
u,
south
25.2
2457.0
358
4.53
0.23
0.36
0.41
11.0
15.4
206.5
Ka
Or/K,
Mu,
Ch
4
Podzols
F
Brazil,
Roraima
27.9
1836.0
46
4.91
0.78
0.05
0.17
1.1
2.8
20.1
5
Plinthosols
Braz
il,
Acre
25.0
1689.5
259
4.45
0.62
0.17
0.22
7.4
14.1
215.0
Mu
Ka,
Gi,
He
6
Ferralsols
V
enez
uela
28.0
2382.0
70
4.68
0.79
0.16
0.06
1.1
5.8
20.6
7
Alisols
Per
u,
south
25.4
2457.6
216
4.21
0.40
0.22
0.38
7.5
23.0
463.6
8
Podzols
Bra
zil,
Amazonas
27.1
2289.2
92
4.10
0.96
0.02
0.02
3.1
20.1
3.1
Pl
He,
Ch
9
Alisols
Per
u,
south
25.3
2536.5
216
4.41
0.18
0.29
0.53
5.7
29.6
362.1
Il-Sm
Mi,
Ka,
Al,
Go,
Gi
10
Re
gosol
Bra
zil,
Mato
Grosso
25.6
2353.1
280
5.34
0.77
0.12
0.11
20.2
22.3
109.0
Ka
Gi,
He,
Or/K
11
Acrisols
Bra
zil,
P
ará
26.8
2191.6
55
3.74
0.84
0.11
0.06
0.2
1.3
44.7
12
Acrisols
Braz
il,
Acre
26.0
1919.8
194
4.13
0.62
0.23
0.15
6.2
9.1
85.1
13
Ferralsols
V
enez
uela
28.1
2337.0
58
4.16
0.85
0.14
0.02
1.3
7.6
21.7
14
Ferralsols
Braz
il,
Mato
Grosso
25.5
1613.1
352
4.20
0.78
0.16
0.06
1.5
5.8
38.2
Ka
Gi,
Go,
He
15
Luvisols
Per
u,
south
25.2
2457.0
358
6.12
0.29
0.08
0.63
32.9
36.3
326.3
Mu
Ka,
Pl,
Or/K,
He,
Gi
16
Gle
ysols
Bra
zil,
Roraima
27.2
1839.0
60
4.40
0.73
0.17
0.10
4.2
8.3
41.1
17
Arenosols
F
Peru,
north
26.3
2751.5
127
4.14
0.94
0.03
0.04
1.7
4.0
13.0
Il-Sm
Ka
18
Ferralsols
Braz
il,
P
ará
26.7
2211.9
35
4.09
0.73
0.22
0.04
2.4
10.6
63.7
19
Plinthosols
Braz
il,
Acre
25.9
1907.0
203
4.23
0.19
0.18
0.62
10.2
29.2
145.9
Il-Sm
Ka
20
Alisols
Per
u,
south
25.4
2457.6
216
4.32
0.20
0.40
0.40
7.0
35.6
578.0
Il-Sm
Mi,
Ka,
Al
21
Gle
ysols
Per
u,
south
25.4
2457.6
217
4.05
0.17
0.39
0.44
3.4
41.4
486.0
Mi
Ka,
Il-Sm,
Al
22
Arenosols
Guyana
26.4
2813.3
125
4.73
0.96
0.02
0.02
2.5
3.4
8.0
23
Plinthosols
Braz
il,
Amazonas
26.4
2593.7
71
3.98
0.26
0.20
0.54
1.2
10.1
44.5
24
Ferralsols
Braz
il,
P
ará
26.7
2211.9
44
4.02
0.80
0.14
0.06
2.0
6.4
52.2
25
Plinthosols
Braz
il,
Mato
Grosso
25.3
1509.7
281
4.65
0.66
0.24
0.10
7.4
12.7
51.3
Ka
Gi,
He
26
Ferralsols
Braz
il,
Mato
Grosso
25.0
1854.4
326
4.19
0.86
0.10
0.04
1.2
12.1
9.5
Ka
Gi,
Mi
27
Acrisols
Bol
ivia
23.3
1142.6
447
5.88
0.75
0.10
0.14
17.8
18.2
230.1
Ka
Gi,
He,
Or/K,
Pl
28
Cambisols
Bol
ivia
24.8
813.4
310
6.06
0.48
0.18
0.35
51.3
51.6
679.7
Ka
Gi,
Go,
He,
Mu
29
Ferralsols
Boli
via
23.9
1451.2
299
4.63
0.74
0.20
0.06
1.6
12.5
48.8
Ka
Gi,
Go,
He,
Pl
30
Arenosols
Peru,
nor
th
26.3
2751.5
126
4.07
0.82
0.02
0.16
4.2
4.9
4.1
Mu
∗
Ch
31
Acrisols
Guyana
26.4
2813.3
124
4.24
0.81
0.15
0.05
3.4
10.0
17.6
32
Fluvisols
Peru,
south
25.1
2399.4
381
5.08
0.02
0.48
0.50
64.9
65.7
435.1
33
Plinthosols
Braz
il,
Acre
25.9
1946.3
205
5.19
0.18
0.20
0.63
31.0
41.8
546.4
34
Plinthosols
Braz
il,
Amazonas
26.3
2553.3
70
4.01
0.22
0.19
0.59
4.2
14.4
62.7
35
Plinthosols
Braz
il,
Amazonas
26.3
2553.3
70
3.94
0.14
0.13
0.73
4.6
16.0
44.0
36
Acrisols
Braz
il,
P
ará
26.8
2178.1
38
3.96
0.24
0.46
0.30
2.6
15.6
18.7
Ka
Sm
37
Arenosols
Guyana
26.7
2282.1
97
4.79
0.97
0.01
0.02
4.0
4.5
6.9
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 5
C. A. Quesada et al.: Variations in soil chemical and physical properties
57
T
ab
le
1.
Continued.
Soil
Classification
Locat
ion
T
A
P
A
E
V
pH
P
article
fraction
6
B
I
E
6
B(R)
Mineralogy
(
◦
C)
(mm)
(m)
Sand
Clay
Silt
(mmol
c
kg
−
1
)
1
◦
2
◦
38
Acrisols
Braz
il,
Mato
Grosso
25.6
2353.1
274
4.65
0.79
0.11
0.10
15.7
20.4
66.8
Ka
Gi
39
Alisols
Peru,
south
25.3
2536.5
216
5.06
0.02
0.46
0.52
49.9
56.7
978.3
Ka
Mu,
Or/K,
Ch,
He
40
Ferralsols
Boli
via
24.2
1456.7
198
4.70
0.58
0.36
0.06
13.2
20.1
36.5
Ka
Gi,
Sm
41
Ferralsols
Braz
il,
P
ará
26.8
2191.6
43
4.23
0.52
0.39
0.09
2.7
13.6
77.7
42
Plinthosols
Colom
bia
25.8
2804.1
106
4.50
0.21
0.42
0.37
10.0
62.1
327.0
Il-Sm
Ka,
Mi
43
Acrisols
Guyana
25.7
2932.2
124
4.44
0.82
0.13
0.05
2.8
10.9
31.0
44
Ferralsols
Guyana
26.6
2633.8
108
4.25
0.79
0.18
0.03
2.7
13.0
21.8
45
Ferralsols
Guyana
26.6
2633.8
106
4.03
0.76
0.20
0.04
2.9
11.2
22.0
46
Acrisols
Braz
il,
P
ará
26.8
2178.1
40
4.00
0.64
0.25
0.11
3.0
18.1
10.8
Ka
Gi
47
Lixisols
V
enez
uela
25.3
1364.4
291
5.43
0.71
0.10
0.19
17.8
17.9
45.2
Ka
Sm,
Mi
48
Cambisols
Peru,
nor
th
26.3
2805.5
97
5.15
0.10
0.28
0.62
50.7
57.2
496.4
Il-Sm
Ka,
Mi,
Al,
Mc
49
Plinthosols
V
enez
uela
25.8
2810.2
98
4.13
0.38
0.31
0.31
2.8
44.4
233.4
Il-Sm
Ka,
Mi,
Gi
50
Podzols
F
Brazil,
Amazonas
27.1
2289.2
100
4.73
0.89
0.09
0.02
1.3
3.1
1.6
51
Gle
ysols
V
enez
uela
28.0
2499.0
89
4.61
0.83
0.14
0.03
1.9
8.8
20.4
52
Cambisols
Braz
il,
Acre
25.7
1803.7
278
5.56
0.39
0.25
0.35
73.7
73.7
564.9
Ka
Pl,
Or/K,
Mu,
He
53
Alisols
Bol
ivia
25.0
3076.8
229
4.24
0.43
0.25
0.32
4.8
18.0
304.4
54
Plinthosols
Colom
bia
25.8
2804.1
107
4.29
0.19
0.43
0.38
10.2
62.60
385.1
Il-Sm
55
Cambisols
Peru,
south
25.4
2457.6
219
4.22
0.47
0.29
0.24
2.2
33.80
185.1
Il-Sm
Ka,
Mi,
Gi
56
Cambisols
Ecuador
24.9
3172.3
261
4.95
0.47
0.30
0.23
77.9
84.80
928.5
Il-Sm
Ka
57
Ferralsols
Boli
via
23.9
1451.2
300
4.39
0.73
0.21
0.06
1.7
15.2
50.1
Ka
Gi,
He,
Mu
58
Alisols
Bra
zil,
Rondônia
27.2
2208.0
78
3.81
0.20
0.34
0.46
2.0
30.0
78.6
59
Plinthosols
Braz
il,
Acre
25.9
1907.0
205
5.07
0.16
0.25
0.59
50.0
56.2
345.3
Il-Sm
Mi,
Ka
60
Cambisols
Peru,
south
25.6
2095.9
203
5.60
0.15
0.25
0.60
85.5
86.9
1047.9
61
Ferralsols
Braz
il,
Amazonas
26.9
2409.0
114
4.29
0.25
0.62
0.13
2.6
16.6
45.0
62
Ferralsols
Guyana
26.6
2633.8
101
4.37
0.82
0.15
0.03
3.6
12.2
19.4
63
Leptosols
Frenc
h
Guyana
25.0
3329.2
140
4.34
0.60
0.32
0.08
4.5
24.0
74.0
64
Cambisols
Peru,
south
25.4
2457.6
218
3.91
0.40
0.44
0.17
2.2
44.7
272.8
Il-Sm
Mi,
Ka,
Al,
Mc
65
Alisols
Col
ombia
25.8
2777.6
120
4.13
0.58
0.20
0.22
2.4
26.0
80.1
Ka
Il-Sm,
Mi,
Gi
66
Alisols
Bra
zil,
Rondônia
27.2
2208.0
83
3.82
0.27
0.26
0.48
1.2
29.3
75.0
67
Arenosols
Guyana
26.8
2158.5
102
4.53
0.90
0.01
0.09
3.0
7.8
28.3
68
Ferralsols
Frenc
h
Guyana
24.9
3329.2
140
4.40
0.52
0.38
0.10
4.6
13.2
72.6
Ka
Gi,
Go
69
Alisols
Peru,
nor
th
26.3
2805.5
97
5.20
0.32
0.27
0.40
68.8
92.1
464.1
Il-Sm
Ka,
Sm,
Gi
70
Plinthosols
Peru,
nort
h
26.3
2814.8
113
4.55
0.38
0.47
0.14
7.9
37.7
275.4
Ka
Il-Sm
71
Ferralsols
Braz
il,
Mato
Grosso
25.3
1509.7
281
4.20
0.47
0.45
0.08
4.8
22.0
103.0
Ka
Gi,
He,
Go,
Or/K
72
Ferralsols
Braz
il,
P
ará
26.9
2197.2
42
4.03
0.46
0.48
0.06
2.7
17.0
71.1
73
Cambisols
Ecuador
24.9
3172.3
266
4.63
0.36
0.29
0.35
89.8
124.7
835.0
Il-Sm
Ka,
Mi
74
Plinthosols
Boli
via
25.0
3076.8
229
4.07
0.30
0.23
0.47
4.2
9.7
261.9
75
Arenosols
Guyana
26.8
2289.6
98
4.86
0.97
0.03
0.00
4.6
6.0
6.3
76
Acrisols
Guyana
26.8
2289.6
98
4.20
0.59
0.36
0.05
4.1
19.1
27.7
77
Ferralsols
Braz
il,
P
ará
25.4
1883.1
145
3.78
0.23
0.66
0.10
4.0
21.4
10.7
Ka
Sm
78
Plinthosols
V
enez
uela
25.8
2810.2
98
3.97
0.24
0.36
0.40
2.9
53.4
296.3
Il-Sm
Ka,
Mi,
Gi
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 6
58
C. A. Quesada et al.: Variations in soil chemical and physical properties
T
ab
le
1.
Continued.
Soil
Classification
Locat
ion
T
A
P
A
E
V
pH
P
article
fraction
6
B
I
E
6
B(R)
Mineralogy
(
◦
C)
(mm)
(m)
Sand
Clay
Silt
(mmol
c
kg
−
1
)
1
◦
2
◦
79
Fluvisols
Peru,
south
25.0
3192.2
274
4.51
0.02
0.50
0.47
47.5
58.1
952.4
80
Ferralsols
Braz
il,
P
ará
26.7
2211.9
42
3.79
0.33
0.54
0.14
4.1
22.8
61.6
Ka
Go,
Gi
81
Cambisols
Peru,
south
25.5
2079.3
203
5.93
0.05
0.31
0.64
78.9
80.7
1253.3
82
Acrisols
Guyana
26.8
2387.0
90
4.07
0.60
0.34
0.06
3.3
18.0
28.5
83
Alisols
Peru,
nor
th
26.3
2777.8
126
4.47
0.78
0.10
0.13
1.3
24.6
114.7
Il-Sm
Mi,
Ka
84
Plinthosols
V
enez
uela
27.9
2510.0
114
4.44
0.63
0.30
0.07
2.6
10.0
21.1
85
Alisols
Bra
zil,
Rondônia
27.7
923.5
83
3.64
0.48
0.34
0.18
1.6
23.8
40.4
86
Plinthosols
Peru,
nort
h
26.3
2751.5
127
4.46
0.33
0.34
0.33
10.8
51.3
94.0
Il-Sm
Ka,
Gi
87
Alisols
Ecua
dor
25.3
3008.9
237
4.61
0.43
0.34
0.23
33.8
51.0
441.8
Ka
Il-Sm,
Mi
88
Acrisols
Peru,
nor
th
26.3
2814.8
113
4.42
0.32
0.46
0.22
5.4
37.1
224.9
Ka
Sm,
Gi
89
Alisols
Bra
zil,
Rondônia
26.2
2205.4
78
3.63
0.47
0.18
0.35
1.5
23.1
32.6
90
Gle
ysols
Peru,
nor
th
26.7
2645.1
140
4.31
0.62
0.16
0.22
4.1
17.8
172.0
Ka
He
91
Gle
ysols
Ecua
dor
25.3
3008.9
235
4.39
0.03
0.57
0.40
71.0
78.3
832.3
Il-Sm
Ka,
Sm
92
Cambisols
Peru,
south
25.5
2079.3
203
6.07
0.05
0.42
0.52
68.2
68.2
1225.3
Mica
Il,
Ka
93
Acrisols
V
enez
uela
26.2
3425.0
109
4.79
0.88
0.08
0.03
3.0
12.2
6.5
Ka
Gi,
Mi
94
Acrisols
Bol
ivia
24.1
1270.3
268
6.30
0.49
0.31
0.20
21.3
22.2
209.0
Ka
He,
Pl
95
Ferralsols
Braz
il,
Amazonas
27.1
2289.2
100
4.34
0.08
0.85
0.05
1.9
21.4
96
Alisols
Peru,
nor
th
26.3
2814.8
114
3.99
0.38
0.18
0.45
3.8
29.9
185.2
97
Cambisols
Boli
via
24.3
1066.0
373
5.23
0.55
0.18
0.26
60.7
61.9
283.4
98
Ferralsols
Braz
il,
Amazonas
27.0
2444.4
111
4.17
0.30
0.59
0.11
2.9
12.4
34.9
99
Cambisols
Braz
il,
Roraima
27.0
1855.0
153
4.25
0.43
0.36
0.22
9.5
16.1
120.0
100
Acrisols
V
enez
uela
26.2
3425.0
99
5.03
0.89
0.05
0.06
1.8
11.3
1.9
Ka
Sm,
Mi
101
Alisols
Ecua
dor
23.8
3710.7
431
4.49
0.40
0.33
0.27
11.1
26.6
333.9
Ka
Mi,
Gi
102
Alisols
Per
u,
north
26.3
2814.8
113
4.03
0.39
0.46
0.15
3.0
30.7
85.4
103
Alisols
Bra
zil,
Rondônia
26.2
2205.4
87
3.84
0.60
0.17
0.24
1.8
19.7
32.1
104
Luvisols
Bra
zil,
Acre
25.7
1883.8
228
4.26
0.14
0.55
0.31
25.8
45.9
461.1
105
Alisols
Ecuador
23.8
3710.7
432
4.77
0.41
0.29
0.30
20.4
30.5
330.2
Ka
Mi
106
Ferralsols
Braz
il,
Amazonas
27.1
2245.7
95
4.24
0.16
0.68
0.16
2.4
28.1
4.7
Ka
Sm,
Gi
107
Acrisols
Peru,
nor
th
26.8
2630.0
122
3.98
0.44
0.22
0.34
2.6
20.2
68.8
Ka
Go.
Gi
108
Gle
ysols
Col
ombia
25.8
2799.9
120
4.34
0.32
0.34
0.34
5.5
35.1
150.0
Ka
Mu,
Gi,
He,
Go
109
Gle
ysols
Bra
zil,
Roraima
27.3
1840.0
62
4.51
0.78
0.15
0.06
2.6
7.3
22.3
110
Plinthosols
Braz
il,
Amazonas
26.4
2593.7
71
4.00
0.36
0.18
0.45
1.9
14.9
47.3
111
Ferralsols
Braz
il,
Amazonas
27.1
2245.7
93
3.98
0.09
0.78
0.13
3.5
21.6
42.1
112
Ferralsols
Braz
il,
Amapá
26.8
2377.1
80
4.05
0.04
0.81
0.15
5.3
21.5
25.5
Ka
Gi
113
Cambisols
Peru,
south
25.5
2079.3
203
5.96
0.08
0.31
0.61
80.1
81.2
1304.2
114
Cambisols
Braz
il,
Acre
25.8
1652.5
236
5.92
0.25
0.37
0.38
80.3
80.4
845.6
Ka
Mu,
Pl,
Or/K,
He
115
Alisols
Bra
zil,
Roraima
27.3
1841.0
126
4.08
0.33
0.44
0.23
3.8
16.4
73.9
116
Fluvisols
Peru,
south
25.2
2477.1
356
6.72
0.01
0.45
0.54
85.3
85.9
1688.1
Ka
Mu,
Ch,
Pl,
Go
117
Ferralsols
Braz
il,
Amazonas
27.1
2193.2
110
4.27
0.13
0.46
0.41
1.9
17.5
24.3
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 7
C. A. Quesada et al.: Variations in soil chemical and physical properties
59
T
ab
le
1.
Continued.
Soil
Classification
Locat
ion
T
A
P
A
E
V
pH
P
article
fraction
6
B
I
E
6
B(R)
Mineralogy
(
◦
C)
(mm)
(m)
Sand
Clay
Silt
(mmol
c
kg
−
1
)
1
◦
2
◦
118
Acrisols
Frenc
h
Guyana
24.9
3329.2
140
4.16
0.33
0.57
0.10
5.3
21.3
31.2
119
Ferralsols
Braz
il,
P
ará
26.9
2175.8
43
4.13
0.23
0.68
0.09
2.7
8.5
70.5
120
Ferralsols
Braz
il,
P
ará
26.7
2211.6
45
4.27
0.14
0.79
0.08
3.1
17.7
68.3
121
Ferralsols
Braz
il,
Amazonas
27.1
2193.2
112
4.14
0.10
0.69
0.20
1.0
11.3
30.2
122
Alisols
Ecuador
23.8
3710.7
431
4.37
0.42
0.31
0.28
9.0
32.4
288.2
Ka
Mi,
Il-Sm
123
Podzols
F
Colombia
25.8
2799.9
120
4.27
0.75
0.01
0.25
6.4
7.1
3.3
Mu
Ch
124
Cambisols
Bol
ivia
24.3
1066.0
373
6.84
0.58
0.19
0.23
75.6
75.9
566.7
125
Fluvisols
Peru,
south
25.2
2457.0
356
6.41
0.48
0.52
0.00
84.5
85.2
1688.7
Mu
Ka,
Ch,
Or/K,
Pl
126
Ferralsols
Braz
il,
P
ará
25.1
2015.9
197
3.84
0.03
0.89
0.08
6.4
29.7
16.6
Ka
0
127
Ferralsols
Braz
il,
Mato
Grosso
25.1
1665.8
373
4.10
0.46
0.49
0.05
2.3
19.5
28.3
Ka
Gi,
Go,
He,
Mu
128
Acrisols
Frenc
h
Guyana
24.9
3329.2
140
4.76
0.12
0.68
0.20
10.9
15.5
87.9
Ka
Gi,
Go
129
Lixisols
Fre
nch
Guyana
24.9
3329.2
140
4.85
0.18
0.65
0.17
13.6
16.9
65.5
130
Ferralsols
Braz
il,
Amazonas
27.1
2289.2
106
3.94
0.20
0.68
0.12
3.7
21.9
5.0
Ka
0
131
Ferralsols
Braz
il,
Amazonas
27.1
2193.2
105
3.56
0.08
0.54
0.38
2.4
11.6
30.0
132
Lixisols
Frenc
h
Guyana
24.9
3329.2
140
4.74
0.17
0.62
0.21
16.2
17.4
64.2
133
Acrisols
Braz
il,
Amazonas
26.9
2457.9
119
4.29
0.08
0.81
0.11
2.7
15.4
43.3
134
Ferralsols
Braz
il,
Amazonas
27.1
2245.7
93
4.08
0.10
0.80
0.10
4.9
19.8
8.6
135
Fluvisols
Ecuador
23.8
3710.7
394
5.09
0.35
0.32
0.34
81.7
81.9
1181.6
Il-Sm
Ka,
Sm
136
Plinthosols
Braz
il,
Roraima
27.2
1841.0
59
4.43
0.43
0.38
0.19
4.0
14.2
72.1
137
Cambisols
Boli
via
24.2
1383.6
248
5.67
0.63
0.18
0.19
76.5
76.6
755.3
Il-Sm
Ka,
Mi
138
Gle
ysols
Bra
zil,
Roraima
27.2
1840.0
64
4.61
0.60
0.26
0.14
2.9
8.6
24.7
139
Podzols
F
Peru,
north
26.3
2777.8
124
4.07
0.60
0.08
0.32
6.8
8.5
10.5
Pl
Ch,
Ka,
He
140
Cambisols
Bra
zil,
Roraima
27.1
1846.0
85
4.02
0.33
0.43
0.25
3.1
15.3
64.0
141
Leptosols
V
enez
uela
26.3
2820.7
366
5.26
0.53
0.16
0.30
23.5
27.7
142
Umbrisols
Boli
via
24.2
1456.7
195
4.74
0.29
0.36
0.35
6.6
25.5
259.7
143
Umbrisols
Boli
via
24.2
1456.7
195
4.90
0.29
0.36
0.35
8.7
20.8
179.5
Ka
Il,
Mi
144
Cambisols
Bol
ivia
24.2
1383.6
248
6.17
0.84
0.07
0.09
50.7
52.2
715.1
Mi
Ka
145
Podzols
F
V
enezuela
26.2
3425.0
99
4.88
18.2
18.6
146
Gle
ysols
Peru,
nor
th
26.3
2801.3
114
4.03
6.20
62.3
147
Podzols
F
Peru,
north
26.7
2646.5
127
4.25
0.69
0.06
0.25
3.9
7.1
20.0
Mu
∗
Ch
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 8
60
C. A. Quesada et al.: Variations in soil chemical and physical properties
0–0.05, 0.05–0.10, 0.10–0.20, 0.20–0.30, 0.30–0.50, 0.50–
1.00, 1.00–1.50 and 1.50–2.00 m. They were collected using
an undisturbed soil sampler (Eijkelkamp Agrisearch Equip-
ment BV, Giesbeek, the Netherlands) and/or collected from
the pit walls at the same depths. All samples were air dried as
soon as possible with roots, detritus, small rocks and particles
over 2 mm then removed in the laboratory. Samples, sieved
at 2 mm, were used in the laboratory for analysis. Through-
out this paper only results for surface soils (0–0.30 m) are
reported, which is the layer that hold the bulk of soil C in
tropical forest systems (Batjes and Dijkshoorn, 1999).
2.2
Soil classification
Soils were classified according to their Reference Soil Group
(RSG), which represents the great order level in the World
Reference Base for Soil Resources (IUSS (International
Union of Soil Science) Working Group WRB, 2014). Our
classification were based on the requisite field and laboratory
observations taken following the standard approach from
WRB Guidelines for Soil Descriptions (Jahn et al., 2006).
2.3
Laboratory analysis
Soil samples were analysed at different institutions depend-
ing on sampling location: Max-Planck Institute für Biogeo-
chemie (MPI), Jena, Germany; Instituto Venezuelano de In-
vestigaciones Cientificas (IVIC), Caracas, Venezuela; or In-
stituto Nacional de Pesquisas da Amazonia (INPA), Manaus,
Brazil. All laboratories were linked through inter-calibration
exercises and strictly adhered to the same methodologies and
sample standards. For the Venezuelan soils, only cation ex-
change capacity was measured at IVIC, with all remaining
analysis being determined at MPI and INPA. Soil total re-
serve bases were analysed at INPA and Leeds laboratories
(University of Leeds, School of Geography). For samples
collected after 2008 (i.e. not included in Quesada et al. 2010)
all analyses were performed at INPA.
Chemical analysis
Soil pH was determined in H
2
O as 1
:
2
.
5. Exchangeable
cations were determined at soil pH using the silver thiourea
method (Ag-TU; Pleysier and Juo, 1980), with the analysis
of filtered extracts then done by atomic absorption spectrom-
etry (AAS) at INPA and IVIC or by ICP-OES at MPI. Each
sample run was checked and standardised with extracts from
the Montana SRM 2710 soil standard reference (National In-
stitute of Standards of Technology, Gaithersburg, MD, USA).
Effective cation exchange capacity (
I
E
) was calculated as the
sum of [Ca]
E
+
[Mg]
E
+
[K]
E
+
[Na]
E
+
[Al]
E
, where
[
X
]
E
represents the exchangeable concentration of each element in
mmol
c
kg
−
1
soil. Total phosphorus was determined by acid
digestion at 360
◦
C using concentrated sulfuric acid followed
by H
2
O
2
as described in Tiessen and Moir (1993). In the
same acid digestion extract, total concentration for Ca, Mg,
K and Na was determined and the weathering index Total
Reserve Bases,
6
RB
, calculated. This index is based on to-
tal cation concentration in the soil and is considered to give
a chemical estimation of weatherable minerals (Delvaux et
al., 1989; Quesada et al., 2010), with
6
RB
equal to [Ca]
T
+
[Mg]
T
+
[K]
T
+
[Na]
T
, where
[
X
]
T
represents the total con-
centration of each element in mmol
c
kg
−
1
soil.
Leaf litter lignin estimates were available for 72 of the
147 sites, having been obtained using the acid detergent fi-
bre method (Van Soest, 1963) as part of the studies of Que-
sada (2008) and Paz (2011).
2.4
Determination of soil organic C and its fractions
Concentrations of total SOC and N were determined in an
automated elemental analyser (Nelson and Sommers, 1996;
Pella, 1990). All samples were free of carbonates as con-
firmed by their acidic nature (Table 1). The partitioning of
SOC in its different fractions was also performed for a subset
of sites (
n
=
30) previously selected by Paz (2011) to account
for the large variation in weathering, climate and chemical
properties of soils occurring across forest sites in the Ama-
zon Basin. The fractionation was done in compound samples
by depth (0–5, 5–10, 10–20 and 20–30 cm) to better repre-
sent the soil conditions in the 1 ha sampling plot. The frac-
tionation scheme followed Zimmermann et al. (2007), which
yields five different fractions of labile C associated with the
clay and silt (
C
+
S
), resistant C associated with clay and
silt (
R
C
+
S
), C associated with sand and stable aggregates
(
S
+
A
), particulate organic matter (POM), and the dissolved
organic C (DOC) component. Samples were dispersed us-
ing a calibrated ultrasonic probe operating with an output
energy of 22 J mL
−
1
. They were subsequently wet sieved to
separate < 63 µm particles (
C
+
S
) from > 63 µm soil particles
(POM and
S
+
A
). The entire < 63 µm solution was then cen-
trifuged for 4 min at 1200 rpm. The
C
+
S
obtained after cen-
trifugation was oven dried at 40
◦
C for 48 h and subsequently
weighed. The
R
SOC
was obtained by incubating 1 g of
C
+
S
with 150 mL of sodium hypochlorite 6 % (adjusted to pH 8).
After this reaction, the remaining material was washed with
distilled water and oven dried at 40
◦
C for 48 h. The labile
C
+
S
fraction was determined as the difference of total C as-
sociated with clay and silt and the
R
C
+
S
. The DOC sample
was obtained by vacuum filtering a 50 mL aliquot of the total
water volume used in the wet sieving (after centrifugation)
through a membrane filter of 0.45 µm and C was determined
using a TOC analyser.
S
+
A
and POM were separated fol-
lowing the procedures described in Wurster et al. (2010) and
Saiz et al. (2015). In short, 25 mL of sodium polytungstate
solution (1.8 g cm
−
3
, Sometu - Europe™, Berlin, Germany)
was added to the > 63 µm dried samples placed in 50 mL cen-
trifuge tubes. Samples were then centrifuged for 15 min at
1800 rpm and left to rest overnight. After this time, samples
were left in the freezer for approximately 3 h, after which
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 9
C. A. Quesada et al.: Variations in soil chemical and physical properties
61
POM and
S
+
A
was separated by washing the frozen su-
pernatant with distilled water. Both fractions were washed
with distilled water to remove any residue of polytungstate
solution and then dried at 40
◦
C for 48 h. All fractions were
analysed in the same way as SOC.
Given that some tropical soils have aggregates that are
very strong and resistant to disruption by sonication, the
> 63 µm fraction often contains clay aggregates and therefore
S
+
A
represents the entire coarse fraction. The recovery of
C
after fractionation averaged 97.7 %.
2.5
Selective mineral dissolution
Soil samples were extracted for Fe and Al using estab-
lished standard techniques as described in detail in Van
Reeuwijk (2002). In short, replicate samples were shaken
for 16 h using dithionite–citrate and Na–pyrophosphate so-
lutions. The extraction with ammonium oxalate–oxalic-acid
solution at pH 3 was performed in the dark by shaking for
4 h. All extracts were determined for Fe and Al concentra-
tions in AAS. These methods provide useful quantitative es-
timates of soil oxide composition (Parfitt and Childs, 1988).
The dithionite–citrate solution dissolves all iron oxides, such
as goethite, gibbsite, ferrihydrite, halloysite and allophane,
but with hematite and goethite only partially dissolved. Al-
though this mineral dissolution method has a broad capacity
to estimate Fe and Al in such minerals, it does not differ-
entiate its various crystalline forms or between short-range
(amorphous) minerals and crystalline structures. The ammo-
nium oxalate – oxalic acid solution on the other hand, specifi-
cally dissolves short-range order minerals such as allophane,
imogolite, ferrihydrite, Al–humus complexes, lepidocrocite,
Al–vermiculite and Al hydroxy interlayer minerals. There-
fore, the difference between the two methods is often used
to estimate the amount of crystalline minerals in the soil,
viz. (Fe
d
-Fe
o
), while negative values indicate the predom-
inance of short-range minerals. Further interpretation of se-
lective dissolution data according to Parfitt and Childs (1988)
is shown in Table 2.
2.6
Soil physical properties
Soil particle size distribution was determined using the
pipette method (Gee and Bauder, 1986) is are reported here
as a fraction (ranging from 0 to 1). Soil bulk densities were
determined using samples collected inside the soil pits at the
same depths of other samples using standard container rings
of known volume (Eijkelkamp Agrisearch Equipment BV,
Giesbeek, the Netherlands). These were subsequently oven
dried at 105
◦
C until constant weight.
2.7
Mineralogy
Bulk soil mineralogical characterisation (less than 2 mm)
was attained through X-ray diffractometry (XRD) using
a PW1050 unit (Philips Analytical, the Netherlands) attached
to an X-ray generator DG2 (Hiltonbrooks Ltd, Crewe, UK).
XRD analyses require sample particle size to be very fine in
order to obtain adequate statistical representation of the com-
ponents and their various diffracting crystal planes, as well
as to avoid diffraction-related artefacts (Bish and Reynolds,
1989). Therefore, samples were ground with a mortar
and pestle using acetone to avoid sample degradation from
heat. Powdered samples were then mounted in holders by a
back-filling method with the aid of a micro-rugose surface to
minimise preferred orientation of the phases present. Sam-
ples were continuously scanned from 3 to 70
◦
(2
θ
) Ni-
filtered CuK
α
radiation (
λ
=
1
.
54185
A
◦
) working at 40 kV
and 40 mA. The scanning parameters were 0.020
◦
step size
and 1.0 s step time. Interpretation and semi-quantitative anal-
ysis of the scans were achieved using the Rietveld refinement
method built in to the Siroquant software (SIROQUANT;
Sietronics Pty Ltd, Canberra, Australia). All samples were
analysed at the Facility for Earth and Environmental Analy-
sis at the University of St. Andrews, Scotland, UK.
2.8
Climatic and terrain elevation data
Mean annual temperature (
T
A
) and precipitation (
P
A
) data
come from BioClim (https://www.worldclim.org, last ac-
cess: 13 July 2018) and site elevation (
E
V
) estimates ob-
tained from the Shuttle Radar Topography Mission database
(SRTM).
2.9
Statistical analysis
All analyses were carried out using the R statistical platform
(R Development Core Team, 2016). In the exploratory data
phase, the non-parametric Kendall
τ
was used to quantify
the strength of bivariate associations with the aid of the cor-
relation function available within the agricolae package (de
Mendiburu, 2013). Multivariate ordinary least squares (OLS)
regression was then performed relating SOC to other soil
properties with candidate variables chosen with reference to
the Kendall rank correlations matrices, after which there was
an exhaustive exploration of regression models taking into
account the a priori hypothesis outlined in the Introduction.
As a check to ensure that we had not overlooked any of the
measured variables as important potential determinants of
[C] regression models, we also then checked for the mini-
mum Akaike information criterion (AIC) regression models
using the dredge function available within MuMIn (Barto´n,
2013) and used variance inflation factor (VIF) to account
for possible collinearity in AIC selected models. Principal
coordinates of soil mineralogical compositions were under-
taken using the princomp function after first transforming the
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 10
62
C. A. Quesada et al.: Variations in soil chemical and physical properties
Table 2.
A guide for interpretation of selective dissolution data following Parfait and Childs (1988).
Form
Description
Fe
d
Dissolves almost all iron oxides not differentiating between crystalline and short-range oxides. Provides estimates of
total amount of iron oxides in the soil.
Fe
o
Estimates short range minerals such as ferrihydrite and possibly other amorphous minerals. Do not extract crystalline
oxides.
Fe
p
Extracts a variety of Fe forms and thus does not specifically relate to any particular form of Fe in soil. Should not be
used to estimate Fe-humus complexes.
Al
d
Probably arises from Al substitution in both crystalline and amorphous oxides. Free Al and interlayer Al. Similar to
Fe
d
it provides wide estimates of Al oxides in the soil.
Al
o
Estimates Al in short-range minerals, such as allophane and imogolite. May also represent Al substitution in ferrihydrite
and the presence of Al hydroxy interlayer minerals. Do not extract crystalline Al hydroxides.
Al
p
Correspond to Al-humus complexes in most soils such as those occurring in Podzols and Andosols.
Fe
d
-Fe
o
Provides estimation of crystalline oxides only. Excludes the content of ferrihydrite and other short-range oxides which
are extracted by Fe
o
.
data using the acomp function available within the compo-
sitions package (van den Boogaart and Tolosana-Delgado,
2008). Kruskal–Wallis multiple comparison tests (Siegel and
Castellan Jr., 1998) were undertaken using the kruskalmc
command available within the pgirmess package (Giraudoux,
2013).
3
Results
3.1
Clustering of soils types
The distribution of the sampled sites across the Amazon
Basin is shown in Figure 1, with the soils sampled divided a
priori into three “clusters” according to their World Resource
Base RSG classification (IUSS, 2014): (1) the typically more
strongly weathered Acrisol and Ferralsol soil types domi-
nated by low-activity clays (LACs); (2) other less weath-
ered soils types (the Alisol, Cambisol Fluvisol, Gleysol, Lep-
tosol, Lixisol, Luvisol, Plinthosol, Regosol and Umbrisol soil
groups) typically dominated by high-activity clays (HACs);
(3) exceptionally sandy soils (Arenosols and Podzols) which
we here refer to as “arenic” soil types. From Fig. 1 the major-
ity of the LAC soils sampled come from the eastern area of
the basin, with the majority of the HAC soils found closer to
the Andes Cordillera. Arenic soils are less abundant than ei-
ther LAC or HAC soils and were sampled in both the eastern
and western portions of the basin.
The contrasting chemistry of the three soil groups is shown
in Fig. 2, where soil effective cation exchange capacity,
I
E
,
is plotted as a function of soil clay fraction,
8
clay
(0 to 0.3 m
depth), with different symbols for each RSG and with the
contrasting
I
E
vs.
8
clay
domains indicated by different back-
ground colours. This shows a minimal overlap between the
arenic soil types and LAC and HAC soil types, with some of
the former having relatively high
I
E
despite their very low
clay content. There is some overlap between the LAC and
HAC soil clusters at intermediate
I
E
and/or
8
clay
, though
with it also being clear that none of the sampled LAC soils
were characterised by a high
I
E
and that none of the HAC
soils had a very high or very low clay content.
3.2
Mineralogical analysis
Distinctions between the LAC and HAC clusters are further
illustrated in Fig. 3, where for a subset of the main dataset,
mineralogical analysis of the bulk soil had been undertaken
using X-ray diffraction (XRD) spectroscopy and for which
the results of a principal components analysis (PCA) ordi-
nation are shown in Fig. 3a. Here it can be seen that the
first PCA axis (PCA1) serves to primarily differentiate the
soils according to their clay activity with the 1
:
1 clay min-
erals gibbsite, goethite and kaolinite, all with large negative
weightings on the PCA1 axis and with the 2
:
1 potassium
feldspar, plagioclase, smectite–illite and chlorite minerals all
with positive weightings. Accordingly (although mineralogy
is not used in the RSG classification system), almost all sites
within our RSG-based LAC cluster are located with negative
scores along the PCA1 axis and with almost all HAC soils
with positive values. All four arenic soils analysed had high
PCA scores.
The contrast between the three soil groups is further shown
in Fig. 3b where, shown as a compositional plot, the contrast-
ing relationships between the 1
:
1 and 2
:
1 minerals are con-
sidered along with variations in quartz content. This diagram
emphasises the almost total lack of 2
:
1 minerals found with
the LAC soil cluster, with these soils essentially being a mix-
ture of 1
:
1 minerals (primarily kaolinite: see Table 1) and
quartz in varying proportions. On the other hand, the HAC
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 11
C. A. Quesada et al.: Variations in soil chemical and physical properties
63
Figure 2.
Contrasting chemical characteristics of the three soil
groups, evidenced by the relationship between top-soil clay frac-
tion and effective cation exchange capacity (0–30 cm). Triangles
with yellow background represent the arenic soil group, consist-
ing of Arenosols (green) and Podzols (black). Filled circles with
pink background represent the low-activity clay soils (LACs) which
consists of Ferralsols (yellow) and Acrisols (red). Soils with high-
activity clay (HAC) are shown as open squares with light blue
background. They are the Alisol (black), Cambisol (pink), Fluvi-
sol (grey), Gleysol (green), Leptosol (brown), Lixisol (red), Luvi-
sol (purple), Plinthosol (blue), Regosol (cyan) and Umbrisol (light
green) soil groups.
soils are all characterised by a high quartz content and with
less than 20 % 1
:
1 minerals present. Also of note, two Cam-
bisols, one Regosol and one Gleysol had 2
:
1 minerals con-
stituting less than 1 % in their fine-earth fraction. Not unex-
pectedly, having a quartz content of > 97 %, all four arenic
soils are found clustered in the bottom right-hand corner of
the compositional triangle.
3.3
Univariate and bivariate comparisons
Using data averaged over the upper 0.3 m of the sampled
soil profiles, Fig. 4 shows as boxplots the contrasts between
our three a priori soil groups in terms of the following: car-
bon density [C]; total reserve bases
6
RB
; effective cation ex-
change capacity
I
E
; fractional sand, silt and clay contents
(
8
sand
,
8
silt
and
8
clay
); and concentrations of dithionite and
oxalate extractable aluminium and iron, viz. [Al]
d
, [Al]
o
,
[Fe]
d
and [Fe]
o
(original data available in Table 1 and Ap-
pendix Table A1). This shows that, although there was no sig-
nificant difference between the three clusters in [C] (Fig. 4a;
Kruskal–Wallis test;
p>
0
.
05), there were significant differ-
ences in the underlying chemistry at
p<
0
.
05 not only be-
tween the arenic soil cluster and both the LAC and HAC
clusters for
6
RB
(Fig. 4b),
I
E
, (Fig. 4c), [Al]
d
(Fig. 4d),
[Al]
o
(Fig. 4e), [Fe]
d
(Fig. 4f) and [Fe]
o
(Fig. 4g) but also
with HAC soils with higher
6
RB
,
I
E
, [Fe]
d
and [Fe]
o
than
the soils in the LAC cluster (
p<
0
.
05). For pH, the situation
was more complicated, with the HAC soils having higher val-
ues than the LAC soils (
p<
0
.
05) but showing no difference
between the arenic soils and either the LAC or HAC soils.
Despite there being many differences in soil properties sig-
nificant at
p<
0
.
05 or better (non-parametric Kruskal–Wallis
test), considerable overlap existed in LAC and HAC for all
seven soil chemical properties presented in Fig. 4.
In terms of soil texture, as would reasonably be expected,
8
sand
was significantly higher at
p<
0
.
05 for the arenic ver-
sus LAC and/or HAC clusters (Fig. 4i). As would be ex-
pected, we also observed significantly lower
8
clay
for the
arenic soils (
p>
0
.
05, Fig. 4j). On the other hand, there was
no difference between
8
silt
for the arenic vs. LAC soils, both
of which, in turn, had a significantly lower
8
silt
than the soils
of the HAC cluster (
p<
0
.
05; Fig. 4k). As is also evident from
Fig. 2, there was much more variation in
8
clay
for the LAC
soils compared to the HAC soils, with
8
clay
ranging from
0.05 to 0.89 and 0.07 to 0.57 for LAC and HAC respectively.
Using Kendall’s
τ
as a non-parametric measure of associ-
ation, correlations between a wide range of soil and climate
properties potentially involved in differences in soil carbon
storage are shown in Table 3. This takes the form of four one-
sided correlation matrices, viz. one half-triangle for each of
the arenic, LAC and HAC clusters as well as for the (com-
bined) dataset as a whole. Here, with
n>
30 for the LAC
and HAC clusters we have indicated in bold all cases where
τ >
0
.
30 for these two groupings (as well as the combined
dataset) with this associating roughly with the probability of
Type-II error being less than 0.05. For the arenic soil cluster
with
n
=
13 the equivalent value is
τ >
0
.
52 and where one or
more of the four groupings has
p>
0
.
05, this has been indi-
cated for all four matrices using italicised characters to help
cross-referencing across the four diagonal matrices.
Table 3 shows that, whilst there are many correlations
which are significant at
p
=
0
.
05 or better, in only a few
cases were there significant correlations found for the same
bivariate combinations in two or more of the three soil clus-
ters and/or when the three clusters are considered together.
For example, although there is clear association between soil
texture and soil carbon density for the LAC soils (
τ
= −
0
.
56
and
τ
=
0
.
54 for
8
sand
and
8
clay
respectively), this is not
the case for the HAC soils (
τ
=
0
.
06 and
τ
=
0
.
19), and the
association is also much less clear for the arenic grouping
(
τ
= −
0
.
17 and
τ
= −
0
.
24). Consequently, when all three
soil clusters are considered together we find
τ
of only
−
0.21
and 0.31 for
8
sand
and
8
clay
. That is to say, when all soils
are considered together there is much weaker association be-
tween soil carbon concentration and soil texture than when
LAC soils are considered on their own. This is also the case
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 12
64
C. A. Quesada et al.: Variations in soil chemical and physical properties
T
ab
le
3
.
K
endall’
s
τ
correlations
between
a
wide
range
of
soil
and
climate
properties
potentially
in
v
olv
ed
in
dif
ferences
in
soil
carbon
storage.
F
our
one-sided
correlation
matrices
are
sho
wn,
one
for
each
of
the
arenic,
LA
C
and
HA
C
clusters
as
well
as
for
the
(combined)
dataset
as
a
whole.
Here,
with
n>
30
for
the
LA
C
and
HA
C
clusters
we
ha
v
e
indicated
in
bold
all
cases
where
τ
>
0.30
for
these
tw
o
groupings
(as
well
as
the
combined
dataset)
with
this
associating
roughly
wit
h
the
probability
of
T
ype-II
error
being
less
than
0.05.
F
or
the
arenic
soil
cluster
with
n
=
13
the
equi
v
alent
v
alue
is
τ
>
0
.
52,
and
we
ha
v
e
indicated
all
cases
where
one
or
more
of
the
four
groupings
has
p
>
0
.
05,
using
italicised
characters
to
help
cross-referencing
across
the
four
diagonal
matrices.
[C]
CN
T
a
P
a
E
v
D
b
sand
clay
silt
pH
I
E
[
P
]
t
TRB
[Fe]
d
[Fe]
o
[Fe]
do
[Al]
d
[Al]
o
[Al]
do
Lo
w-acti
vity
clay
soils
All
[C]
0.08
0.19
0.13
−
0.05
−
0.47
−
0.56
0.54
0.28
−
0.12
0.37
0.08
−
0.09
0.27
0.27
0.23
0.37
0.30
0.14
soils
CN
0.14
0.02
−
0.07
−
0.01
−
0.07
0.04
−
0.02
−
0.16
−
0.16
0.01
−
0.09
−
0.44
−
0.05
0.10
−
0.03
0.10
0.11
−
0.05
combined
T
a
0.01
0.33
0.16
−
0.52
−
0.21
−
0.27
0.27
0.10
−
0.28
−
0.06
−
0.29
−
0.21
−
0.14
0.03
−
0.18
−
0.24
−
0.16
−
0.04
P
a
0.06
−
0.08
−
0.10
−
0.10
−
0.01
0.05
−
0.01
−
0.08
0.12
−
0.07
−
0.17
−
0.16
−
0.19
−
0.08
−
0.06
−
0.15
−
0.18
0.01
E
v
0.00
−
0.32
−
0.61
0.00
−
0.07
0.09
−
0.11
0.00
0.30
0.08
0.12
0.12
0.09
0.13
0.04
0.16
0.14
−
0.02
D
b
−
0.33
−
0.10
−
0.07
−
0.08
0.08
0.48
−
0.48
−
0.26
0.09
−
0.32
0.00
0.05
−
0.20
−
0.31
−
0.11
−
0.20
−
0.18
−
0.06
Sand
−
0.21
0.34
0.07
−
0.05
−
0.04
0.14
−
0.87
−
0.46
0.22
−
0.41
−
0.16
−
0.07
−
0.37
−
0.32
−
0.21
−
0.33
−
0.28
−
0.10
Clay
0.31
−
0.16
−
0.03
0.03
−
0.02
−
0.23
−
0.59
0.33
−
0.22
0.38
0.16
0.04
0.36
0.27
0.22
0.31
0.27
0.10
Silt
0.05
−
0.43
−
0.17
0.05
0.16
0.02
−
0.51
0.10
−
0.10
0.30
0.16
0.21
0.25
0.35
0.08
0.28
0.26
0.03
pH
0.01
−
0.21
−
0.31
−
0.02
0.36
0.20
0.05
−
0.13
0.06
−
0.06
0.07
0.17
−
0.16
−
0.04
−
0.14
0.03
0.04
−
0.10
I
E
0.13
−
0.47
−
0.31
0.06
0.30
−
0.03
−
0.40
0.25
0.42
0.19
0.32
0.04
0.28
0.32
0.13
0.43
0.45
−
0.09
[
P
]
t
0.17
−
0.48
−
0.35
−
0.03
0.28
0.05
−
0.36
0.26
0.38
0.24
0.49
0.30
0.41
−
0.09
0.37
0.36
0.17
0.20
TRB
0.02
−
0.64
−
0.39
0.02
0.36
0.09
−
0.34
0.17
0.47
0.31
0.60
0.60
0.31
−
0.11
0.31
0.09
−
0.01
0.28
[Fe]
d
0.18
−
0.32
−
0.19
−
0.02
0.16
−
0.04
−
0.33
0.34
0.24
0.05
0.34
0.51
0.38
0.16
0.75
0.44
0.19
0.32
[Fe]
o
0.13
−
0.41
−
0.26
0.02
0.24
−
0.10
−
0.36
0.27
0.35
0.17
0.52
0.37
0.46
0.39
−
0.09
0.40
0.53
−
0.36
[Fe]
do
0.12
−
0.15
−
0.04
−
0.03
0.05
0.02
−
0.20
0.27
0.06
−
0.08
0.10
0.31
0.15
0.66
0.05
0.25
−
0.03
0.49
[Al]
d
0.28
−
0.15
−
0.25
0.03
0.12
−
0.24
−
0.22
0.35
0.10
0.00
0.28
0.32
0.21
0.49
0.43
0.32
0.66
0.02
[Al]
o
0.19
−
0.19
−
0.29
0.04
0.20
−
0.20
−
0.19
0.30
0.10
0.08
0.37
0.23
0.26
0.25
0.56
0.01
0.62
−
0.32
[Al]
do
0.05
0.08
0.18
−
0.04
−
0.14
0.03
−
0.04
0.05
−
0.03
−
0.19
−
0.17
0.03
−
0.09
0.18
−
0.31
0.43
−
0.03
−
0.41
Arenic
soils
[C]
CN
T
a
P
a
E
v
D
b
sand
clay
silt
pH
I
E
[
P
]
t
TRB
[Fe]
d
[Fe]
o
[Fe]
do
[Al]
d
[Al]
o
[Al]
do
High-
[C]
0.30
−
0.03
0.02
−
0.01
−
0.29
0.06
0.19
−
0.20
0.08
0.08
0.21
0.02
0.14
0.05
0.08
0.29
0.17
−
0.01
acti
vity
CN
0.15
0.33
−
0.15
−
0.32
−
0.17
0.39
−
0.21
−
0.37
−
0.20
−
0.40
−
0.35
−
0.56
−
0.13
−
0.31
0.00
0.03
−
0.12
0.15
clay
T
a
−
0.32
−
0.03
−
0.13
−
0.61
−
0.05
0.08
−
0.11
−
0.04
−
0.33
−
0.28
−
0.32
−
0.41
−
0.02
−
0.21
0.06
−
0.20
−
0.27
0.23
soils
P
a
0.31
0.16
−
0.70
0.02
−
0.19
−
0.11
0.18
0.03
−
0.12
0.08
−
0.03
0.06
0.05
0.06
0.03
0.18
0.18
−
0.08
E
v
0.26
−
0.33
−
0.40
0.31
0.19
−
0.01
0.04
0.01
0.40
0.29
0.22
0.44
0.02
0.15
−
0.01
0.05
0.18
−
0.18
D
b
−
0.17
0.33
0.00
0.09
−
0.45
−
0.07
−
0.10
0.16
0.21
0.04
0.07
0.14
0.03
−
0.08
0.10
−
0.33
−
0.30
0.15
sand
−
0.24
0.09
0.19
−
0.12
−
0.24
0.22
−
0.43
−
0.63
−
0.01
−
0.28
−
0.30
−
0.31
−
0.18
−
0.19
−
0.06
0.02
0.03
−
0.03
clay
0.00
−
0.03
0.19
−
0.03
0.12
−
0.56
−
0.33
0.05
−
0.04
0.21
0.32
0.19
0.26
0.18
0.13
0.20
0.18
−
0.06
silt
0.23
−
0.11
−
0.32
0.19
0.35
−
0.22
−
0.87
0.20
0.01
0.25
0.13
0.30
0.03
0.11
−
0.02
−
0.15
−
0.12
0.05
pH
−
0.03
0.55
0.26
−
0.16
−
0.47
0.34
0.37
−
0.06
−
0.42
0.32
0.32
0.43
0.08
0.22
−
0.05
−
0.02
0.09
−
0.17
I
E
0.43
−
0.04
−
0.23
0.20
−
0.01
0.06
−
0.11
−
0.14
0.18
−
0.25
0.35
0.65
0.13
0.41
−
0.09
0.06
0.26
−
0.24
[
P
]
t
0.24
−
0.45
−
0.22
0.15
0.48
−
0.34
−
0.45
0.15
0.53
−
0.65
0.41
0.51
0.36
0.25
0.21
0.19
0.12
0.00
TRB
0.00
−
0.03
0.03
−
0.18
0.18
−
0.30
−
0.15
0.09
0.29
0.09
−
0.08
0.27
0.16
0.41
−
0.06
0.03
0.22
−
0.24
[Fe]
d
0.26
−
0.16
0.08
−
0.34
0.10
−
0.11
0.20
−
0.14
−
0.22
0.11
−
0.01
−
0.08
0.02
0.20
0.62
0.39
0.10
0.19
[Fe]
o
−
0.06
−
0.32
−
0.04
−
0.04
0.01
0.16
−
0.17
0.05
0.03
−
0.12
−
0.05
−
0.02
−
0.26
0.20
−
0.18
0.33
0.58
−
0.46
[Fe]
do
0.13
0.00
0.40
−
0.39
−
0.10
−
0.23
0.24
0.06
−
0.26
0.18
−
0.04
−
0.06
0.24
0.36
−
0.45
0.20
−
0.16
0.49
[Al]
d
0.15
0.08
0.29
−
0.13
−
0.13
0.05
−
0.12
0.36
−
0.02
0.29
0.04
0.12
0.00
0.21
0.14
0.21
0.58
−
0.12
[Al]
o
0.27
−
0.03
0.06
0.05
0.15
−
0.04
0.13
0.20
−
0.18
0.12
−
0.02
0.16
0.16
0.38
−
0.02
0.45
0.48
−
0.54
[Al]
do
−
0.18
−
0.06
0.13
−
0.20
−
0.06
−
0.22
−
0.16
−
0.09
0.22
−
0.22
−
0.02
0.02
0.02
−
0.29
0.14
−
0.30
−
0.21
−
0.73
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 13
C. A. Quesada et al.: Variations in soil chemical and physical properties
65
Figure 3.
Contrasting mineralogical characteristics of the different soils in this study.
(a)
Principal components analysis (PCA) ordination
on semi-quantitative X-ray diffraction (XRD) spectroscopy data.
(b)
Compositional plot showing contrasting relationships between the 1
:
1
and 2
:
1 minerals considered along with variations in quartz content.
for the relationship between [C] and soil bulk density,
D
b
,
for which we find
τ
= −
0
.
47 for LAC soils but markedly
lower values for the HAC and arenic soils (
τ
= −
0
.
29 and
τ
= −
0
.
17 respectively), as well as for the combined dataset
(
τ
= −
0
.
33).
In a similar vein, although a high cation exchange capacity
(
I
E
) is clearly associated with a high [C] for LAC soils (
τ
=
0
.
37) and perhaps the arenic soils as well (
τ
=
0
.
43), for the
HAC soils we find a
τ
of only
−
0.08 for the [C] vs.
I
E
. Not
surprisingly then, for the dataset as a whole
τ
=
0
.
13 for the
I
E
vs. [C] correlation.
On the other hand (simple physically based bivariate cor-
relations such as
T
a
vs.
E
v
aside) there are cases where
the strength of the bivariate associations seems to be con-
sistent across all three soil groups. For example, taking the
relationship between total phosphorus, [P]
t
, and mean an-
nual air temperature,
T
a
, shows
τ
= −
0
.
29,
τ
= −
0
.
32 and
τ
= −
0
.
22 for the LAC, HAC and arenic soils respectively
and with the combined dataset yielding
τ
= −
0
.
35. A second
example is the relationship between dithionite extractable
aluminium [Al]
d
and
8
clay
for which we find
τ
=
0
.
31 for
LAC soils,
τ
=
0
.
20 for HAC soils and
τ
=
0.36 for arenic
soils, with
τ
=
0
.
35 for the dataset as a whole. Although we
found many correlations between the variation oxalate and
dithionite extraction metrics for Fe and Al, it was only [Al]
d
that, on its own, showed any marked association with [C],
with this being for the LAC soils (
τ
=
0
.
37). However, we
do also note that
τ
=
0
.
29 for the HAC soils and
τ
=
0
.
28
for the dataset as a whole.
Also of note are the many cases where there are reasonably
high
τ
values found for both the LAC and HAC soils, but not
for the arenic ones: for example in the correlations between
total reserve bases,
6
B
, and organic matter CN (carbon–
nitrogen) ratio for which we observe
τ
= −
0
.
44 for LAC
soils and
τ
= −
0
.
56 for HAC soils, but with a value of only
τ
= −
0
.
03 for the soils in the arenic cluster.
3.4
Carbon and soil texture associations
With a high
τ
observed for several [C] vs. soil texture re-
lationships (Sect. 3.3), the correlations between soil carbon
content and
8
clay
are shown in Fig. 5, with a separate panel
used for each of the three soil clusters and with each panel
having different ranges for both the
x
and
y
ordinates. For
the LAC soils (Fig. 5a) a strong linear relationship exists
(
r
2
=
0
.
58), with there being little apparent difference be-
tween the Ferralsol and Acrisol RSGs. But when the LAC
OLS regression line is repeated again within the arenic soil
group [C] vs.
8
clay
graph of Fig. 5b (for which we also note
that the variability in
8
clay
is only 1
/
10 of that for Fig. 5a,
with [C] being 4 times larger) it is clear not only that does
soil clay content exert little or any control over [C] for these
sandy soils, but also that many of the Podzols have [C] well
in excess of even the highest clay content LAC soils. With the
LAC OLS regression line again repeated for the HAC soils in
Fig. 5c it is similarly clear that many of the HAC soils have
[C] appreciably higher than is expected on the basis of the
highly significant LAC [C] vs.
8
clay
relationship, but with
no detectable [C] vs.
8
clay
correlation when considered on
their own (
r
2
=
0
.
01).
The underlying OLS regressions of Fig. 5 are outlined in
more detail in Table 4. Here, as well as providing a [C] vs.
8
clay
OLS regression summary for the combined dataset as
a whole, we also examined the effects of including
8
silt
in
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 14
66
C. A. Quesada et al.: Variations in soil chemical and physical properties
Table 4.
Summary of OLS regression coefficients for soil organic carbon and texture associations.
b
SE
β
t
p
Lower
Upper
LAC soils:
r
2
=
0
.
57,
p<
0
.
001, AIC
=
292.1
intercept
9.56
1.03
–
9.31
0.000
7.50
11.62
Clay fraction
17.91
2.15
0.762
8.32
0.000
13.60
22.24
LAC soils:
r
2
=
0
.
61,
p<
0
.
001, AIC
=
288.6
intercept
8.50
1.08
–
7.84
0.000
6.32
10.68
clay fraction
16.58
2.13
0.716
7.75
0.000
12.24
20.89
silt fraction
14.39
6.19
0.212
2.32
0.024
1.94
26.83
LAC soils:
r
2
=
0
.
61,
p<
0
.
001, AIC
=
286.7
intercept
8.44
1.06
–
7.96
0.000
6.32
10.57
(clay
+
silt) fractions
16.23
1.79
0.789
9.07
0.000
12.63
19.82
HAC soils:
r
2
=
0
.
00,
p<
0
.
335, AIC
=
628.2
intercept
16.16
3.21
–
5.04
0.000
9.78
22.54
clay fraction
9.58
9.87
0.088
0.97
0.335
−
10.07
29.22
HAC soils:
r
2
=
0
.
05,
p<
0
.
006, AIC
=
625.3
intercept
21.67
4.02
–
5.41
0.000
13.70
29.69
clay fraction
9.26
9.64
0.088
0.96
0.340
−
9.94
28.44
silt fraction
−
16.29
7.40
−
0.196
−
2.21
0.037
−
31.03
−
1.55
HAC soils:
r
2
=
0
.
05,
p<
0
.
259, AIC
=
627.8
intercept
23.36
4.03
–
5.81
0.000
15.35
31.37
(clay
+
silt) fractions
−
6.87
6.04
−
0.103
−
1.14
0.259
−
18.90
5.16
Arenic soils:
r
2
=
0
.
07,
p<
0
.
206, AIC
=
119.92
intercept
8.35
14.55
–
0.574
0.579
−
24.07
40.77
clay fraction
431.39
319.17
0.352
1.352
0.206
−
279.75
1142.55
Arenic soils:
r
2
=
0
.
23,
p<
0
.
119, AIC
=
118.26
intercept
−
0.38
14.04
–
−
0.03
0.979
−
32.13
31.38
clay fraction
143.77
80.24
0.254
1.79
0.107
−
37.75
325.30
silt fraction
228.66
310.22
0.254
0.74
0.480
−
473.18
930.39
Arenic soils:
r
2
=
0
.
31,
p<
0
.
035, AIC
=
116.34
intercept
1.09
12.08
–
0.09
0.930
−
25.84
28.01
(clay
+
silt) fractions
154.67
63.43
0.225
2.44
0.035
13.26
296.07
All soils:
r
2
=
0
.
01,
p<
0
.
13, AIC
=
1154.3
intercept
16.14
1.96
–
8.220
0.000
12.25
20.15
clay fraction
7.98
5.23
0.106
1.524
0.130
−
2.37
18.32
All soils:
r
2
=
0
.
00,
p<
0
.
32, AIC
=
1156.3
intercept
15.96
2.43
–
6.58
0.000
11.18
20.79
clay fraction
7.98
5.25
0.106
1.52
0.131
−
2.41
18.36
silt fraction
0.68
6.01
0.007
0.10
0.917
−
11.25
12.51
All soils:
r
2
=
0
.
01,
p<
0
.
232, AIC
=
1155.2
intercept
16.01
2.43
–
6.59
0.000
11.20
20.80
(clay
+
silt) fractions
4.80
3.96
0.084
1.21
0.228
−
3.03
12.63
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 15
C. A. Quesada et al.: Variations in soil chemical and physical properties
67
Figure 4.
Contrasts between the three soil clusters for selected variables. Statistical differences are given through the non-parametric
Kruskal–Wallis test.
(a)
SOC concentration,
(b)
total reserve bases,
(c)
effective cation exchange capacity,
(d)
soil pH,
(e)
dithionite–citrate
extractable Al,
(f)
Al
o
oxalate extractable Al,
(g)
Fe
d
dithionite–citrate extractable Fe,
(h)
oxalate extractable Fe,
(i)
sand fraction,
(j)
clay
fraction and
(k)
silt fraction.
the [C] vs.
8
clay
regression models: this being either as an
additional term or as part of a single (
8
silt
+
8
clay
) predictor
– the latter, of course, also being equal to [1
−
8
sand
]. Com-
paring the equations for LAC, this analysis shows that the
addition of the
8
silt
term to the [C] vs.
8
clay
regression in-
creases the
r
2
from 0.57 (Table 4a) to 0.61 (Model b) with
a change in Akaike’s information criterion (
1
AIC) of
−
3.9
and with the coefficients for both terms having very similar
slopes, viz. 16
.
6
±
2
.
1 g C kg
−
1
clay and 14
.
4
±
6
.
2 g C kg
−
1
silt. For these LAC soils, taking silt and clay together as the
one soil texture metric (Table 4c) yields a similar
r
2
with an
intermediate slope of 16
.
2
±
1
.
8 g C kg
−
1
(clay
+
silt).
Despite the strong relationships found for the LAC soils
for both
8
clay
and
8
silt
, no correlation was evident for the
HAC soils and, of the three models tested, none had a
r
2
greater than 0.05 (Table 4d–f). For the arenic soils, the ad-
dition of
8
silt
term to a simple [C] vs.
8
clay
model led to
a
1
AIC of only
−
1.7 (compare equations of Table 4g and
h). Nevertheless, where a summation term (
8
clay
+
8
silt
)
was tested as a single predictor variable this resulted in a
marked improvement over and above the [C] vs.
8
clay
re-
lationship with a
1
AIC of
−
3.6 and
r
2
of 0.31 (Table 4i). It
is worth noting that Table 4i shows that the fitted slope for the
arenic soils was 155
±
63 g C kg
−
1
(clay
+
silt), a value nearly
10 times that found for the LAC soils (Table 4c). When all
three soils groupings were considered together there was no
significant relationship between [C] and
8
clay
, this being the
case for either with
8
clay
considered on its own, or for
8
clay
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 16
68
C. A. Quesada et al.: Variations in soil chemical and physical properties
Figure 5.
Associations between soil organic C and clay fraction for the three soil groups.
(a)
Low-activity clay (LAC),
(b)
arenic soils and
(c)
soils containing high-activity clays (HACs). Only LAC shows a significant regression. Non-significant regressions in arenic and HAC
soils are shown as dotted lines.
considered in conjunction with
8
silt
, and with all three mod-
els tested having
r
2
≤
0
.
01 and
p>
0
.
13 (Table 4j–l).
3.5
Soil carbon chemical and mineralogical associations
As already noted in Sect. 3.1, of the many strong correla-
tions between the aluminium and iron oxide measured and
soil carbon concentration, one of the strongest and the most
consistent across the three soil groups was the [C] vs. Al
d
relationship, and this relationship is shown for all three soil
groupings in Fig. 6 (log–log scale) with the appropriate re-
gression coefficients shown in Table 5 (models m to o). Rea-
sonably strong relationships were found between [C] and Al
d
for both the LAC (Fig. 6;
r
2
=
0
.
27,
p<
0
.
0001) and HAC
soils (Fig. 6c;
r
2
=
0
.
23,
p<
0
.
0001), but not for the arenic
grouping (Fig. 6b;
r
2
=
0
.
09
p>
0
.
17). Here direct compari-
son with the soil texture models of Table 4 according to the
AIC values is confounded by slightly different datasets for
the HAC soils (due to Al
d
only having been determined for
77 of the 83 HAC soils), with the relationships examined here
being log–log as opposed to linear. Nevertheless, the very
different
r
2
between the two model types (with
r
2
=
0
.
27
much lower for the [C] vs. Al
d
relationship than for any of
the [C] vs. soil texture models for the LAC soils, for which
r
2
>
0
.
57, and with this being the other way around for the
HAC soils –
r
2
=
0
.
23 for the [C] vs. Al
d
relationship but
none of the soil texture models have
r
2
>
0
.
05) suggests that
for the HAC soils Al
d
is a much better predictor of [C] than
soil texture. In addition, simple soil texture metrics were the
better predictors for the LAC soils.
With any role of [Al]
d
in the modulation of [C] also likely
to be dependent on soil pH (see Introduction) we then probed
potential interactions of [Al]
d
and pH, at the same time eval-
uating the potential role of other measured mineralogical fac-
tors. This was done by testing a range of multivariate models
and selecting on the basis of AIC, the net result of which is
shown in Table 6 (model
q
). This model, which also involves
both pH and [Fe]
o
, has a
1
AIC of
−
17.7 as compared to the
univariate [Al]
d
model of Table 5n, suggesting a drastic im-
provement through the addition of the two additional terms.
Nevertheless, using data for 41 of the 77 HAC sites for which
we had leaf litter lignin content (
3
) measurements, there was
a clear relationship between the model residuals of Eq. (6q)
in Table 6 (Fig. 7a), with this relationship also being evi-
dent (though to a lesser extent) when a simpler model involv-
ing just [Al]
d
and pH was applied (
r
2
=
0
.
25, AIC
=
85.1;
Fig. 7b). In both cases residuals increase with increasing
3
,
meaning that at high
3
the models tend to underestimate [C]
and vice versa at low
3
.
With this lignin effect being consistent with any pH de-
pendent [Al]
d
precipitation reaction mechanism as originally
postulated, we thus probed a possible role of
3
as a factor in-
teracting with both pH and Al
d
using the more limited dataset
of 41 HAC sites for which the requisite data were available.
Model comparisons are shown in Table 7. Starting first with a
simple model of [C] as a function of [Al]
d
, [Fe]
o
and pH (Ta-
ble 7t, which is the same model as Table 6q but in this case
with the reduced “leaf lignin only” dataset), the addition of a
3
term clearly results in a marked improvement in the model
fit (Table 7u;
r
2
=
0
.
46,
1
AIC
= −
3.50). Moreover, for this
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 17
C. A. Quesada et al.: Variations in soil chemical and physical properties
69
Figure 6.
The association between soil organic C and dithionite extractable Al (Al
d
) for the studied soils. The regression line for LAC soils
(a)
is repeated as a dotted line in
(b)
(arenic) and
(c)
(HAC) for comparison.
Table 5.
Summary of OLS regression coefficients for soil organic carbon and dithionite extractable Al.
b
SE
β
t
p
Lower
Upper
LAC soils:
r
2
=
0
.
27,
p<
0
.
0001, AIC
=
30.26
intercept
2.36
0.100
–
23.69
0.000
2.16
2.57
[Al]
d
0.372
0.084
4.39
0.000
0.201
0.542
HAC soils:
r
2
=
0
.
23,
p<
0
.
0001, AIC
=
95.83
intercept
2.50
0.08
–
31.25
0.000
2.34
2.66
log [Al]
d
0.300
0.060
5.00
0.000
0.180
0.419
Arenic soils:
r
2
=
0
.
09,
p
=
0
.
17, AIC
=
37.05
intercept
3.42
0.433
–
7.96
0.000
2.47
4.38
[Al]
d
0.343
0.236
0.17
0.174
−
0.176
0.863
All soils:
r
2
=
0
.
08,
p<
0
.
0004, AIC
=
200.18
intercept
2.69
0.052
52.13
0.000
2.59
2.79
[Al]
d
0.141
0.039
3.65
0.000
0.06
0.217
reduced dataset at least, the [Fe]
o
term then becomes redun-
dant (Table 7v;
r
2
=
0
.
47,
1
AIC
= −
2.0).
The goodness of fit of Eq. (7v) is shown in Fig. 8 where
the fitted soil carbon densities,
[ ˆ
C
]
, are plotted as a function
of the actual values (log–log scale). This shows that Eq. (7v)
provides a reasonable and unbiased fit across a wide range of
[C] for HAC soils, though with two locations (viz. POR-02,
a Plinthosol in the west of the basin, and RIO-12, a Lixisol
on the basin’s northern periphery) being substantially over-
estimated by the model.
Probing the effect of litter quality on soil C storage further,
we examined the relationship of
3
with both leaf litter and
soil C : N ratios (denoted
8
L
CN
and
8
S
CN
respectively), this ex-
ercise being undertaken with a view to see if we could find
statistically significant relationships between
3
and one or
both of
8
L
CN
and
8
S
CN
. This was so as to allow incorporation
of litter quality surrogate measures into an analysis using the
full HAC soil dataset. As is shown in Fig. 9, there were in-
deed significant log–log relationships between
3
and both
8
L
CN
and
8
S
CN
for both HAC soils (but not for LAC soils and
not between
8
L
CN
and
8
S
CN
for HAC soils), with the HAC
3
vs.
8
S
CN
giving a better fit (
r
2
=
0
.
32,
p<
0
.
0001; Fig. 9b).
Considering that the correlation between [C] and the C : N
ratio in HAC soils is very low (
τ
=
0
.
1, Table 3), we then
took
8
S
CN
as our best available surrogate for litter quality and
tested the effect of adding this variable to the original HAC
model as given in Table 6q, finding that this term provided for
a substantial reduction in AIC when added to a model already
including pH, [Al]
d
and [Fe]
o
. Further, upon the inclusion
of the
8
S
CN
term the negative [Fe]
o
term became, as for the
lignin models of Table 7, redundant (Table 6s).
The goodness of fit of the equation of Table 6s is shown in
Fig. 10, where the fitted soil carbon densities
[ ˆ
C
]
are plotted
as a function of the actual values (log–log scale). This shows
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 18
70
C. A. Quesada et al.: Variations in soil chemical and physical properties
Figure 7.
The effect of litter lignin content, a surrogate for the abundance of aromatic C compounds, on the residuals of model regressions
in Table 6q
(a)
and a simplified additional model with only pH and Al
d
included
(b)
.
Table 6.
Summary of OLS regression coefficients for soil organic carbon in HAC soils.
b
SE
β
t
p
Lower
Upper
VIF
HAC soils: log[C] (mg g
−
1
),
r
2
=
0
.
32,
p<
0
.
001, AIC
=
78.09
intercept
1.490
0.313
–
4.77
0.000
0.867
2.113
pH
0.241
0.066
0.359
3.66
0.000
0.109
0.372
1.18
log [Al]
d
(mg g
−
1
)
0.403
0.071
0.673
5.66
0.000
0.261
0.544
1.62
log [Fe]
o
(mg g
−
1
)
−
0.156
0.055
−
0.347
−
2.84
0.006
−
0.266
−
0.047
1.72
HAC soils: log[C] (mg g
−
1
),
r
2
=
0
.
55,
p<
0
.
001, AIC
=
46.42
intercept
−
1.387
0.522
–
−
2.56
0.010
−
2.429
−
0.344
pH
0.262
0.054
0.399
4.91
0.000
0.155
0.368
1.18
log [Al]
d
(mg g
−
1
)
0.314
0.059
0.524
5.30
0.000
0.195
0.432
1.71
log [Fe]
o
(mg g
−
1
)
−
0.010
0.050
−
0.018
−
0.20
0.844
−
0.110
0.090
2.19
Soil C : N ratio
1.132
0.181
0.567
6.29
0.000
0.777
1.500
1.36
HAC soils: log[C] (mg g
−
1
),
r
2
=
0
.
56,
p<
0
.
001, AIC
=
44.46
intercept
−
1.417
0.496
–
−
2.85
0.006
−
2.406
−
0.426
pH
0.259
0.050
0.395
5.12
0.000
0.158
0.359
1.08
log [Al]
d
(mg g
−
1
)
0.307
0.045
0.513
6.78
0.000
0.216
0.396
1.01
Soil C : N ratio
1.155
0.160
−
0.573
−
7.24
0.000
0.837
1.474
1.07
that Eq. (6s) provides a reasonable and unbiased fit across a
wide range of [C] for HAC soils, though with the same two
locations that were overestimated by the lignin model (Fig. 9)
being similarly overestimated.
3.6
Alternative models
Although we have used AIC to assist with model selection
in Sect. 3.3, 3.4 and 3.5, candidate models had been in all
cases guided by the background knowledge and hypothesis
as outlined in Sect. 1. It is therefore worth pointing out that if
one takes a simple information-criterion-guided model selec-
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 19
C. A. Quesada et al.: Variations in soil chemical and physical properties
71
Table 7.
Summary of coefficients from OLS regression models for HAC soils. Interactions of soil organic carbon, soil pH, leaf litter lignin
content (
∧
) and dithionite extractable Al.
b
SE
β
t
p
Lower
Upper
VIF
HAC soils: log[C] (mg g
−
1
),
r
2
=
0
.
38,
p<
0
.
001, AIC
=
42.37
intercept
0.887
0.482
–
1.84
0.073
−
0.090
1.864
pH
0.286
0.091
0.395
3.13
0.003
0.101
0.471
1.09
log [Al]
d
(mg g
−
1
)
0.469
0.107
0.673
4.37
0.000
0.251
0.687
1.58
log [Fe]
o
(mg g
−
1
)
−
0.055
0.087
−
0.092
−
0.63
0.532
−
0.233
0.122
1.47
HAC soils: log[C] (mg g
−
1
),
r
2
=
0
.
46,
p<
0
.
001, AIC
=
38.77
intercept
−
0.488
2.556
–
−
1.91
0.064
−
10.07
0.300
pH
0.318
0.087
0.449
3.62
0.000
0.140
0.496
1.12
log [Al]
d
(mg g
−
1
)
0.415
0.104
0.584
3.97
0.000
0.203
0.626
1.70
log [Fe]
o
(mg g
−
1
)
0.019
0.089
0.006
0.22
0.830
−
0.161
0.200
1.70
log [
∧
] (mg g
−
1
)
0.942
0.410
0.341
2.29
0.027
0.109
1.774
1.20
HAC soils: log[C] (mg g
−
1
),
r
2
=
0
.
47,
p<
0
.
001, AIC
=
36.83
intercept
−
4.676
2.340
–
−
2.00
0.054
−
9.417
0.065
pH
0.319
0.086
0.452
3.70
0.000
0.143
0.494
1.12
log [Al]
d
(mg g
−
1
)
0.428
0.083
0.618
5.18
0.000
0.261
0.595
1.07
log [
∧
] (mg g
−
1
)
0.909
0.377
0.323
2.41
0.021
−
0.145
1.674
1.04
Figure 8.
Fitted vs. observed SOC densities for regression
model 7v (Table 7).
tion approach then it is possible to find models with a lower
AIC than those presented in Tables 4 and 6. For example,
for LAC soils there is a model involving all of
8
sand
,
8
clay
,
[Al]
d
, [Al]
o
[Fe]
d
, [Fe]
do
and
8
S
CN
which provides a signif-
icantly better fit than Eq. (b) of Table 4 (
1
AIC of
−
19.9).
But for this model many of the terms had VIF > 10 and after
removal of these terms then the simpler [C]
=
8
sand
+
8
clay
equation is only 0.2 AIC units higher.
Likewise, if one applies a “blind” information criterion
selection criterion to the HAC soils then it is possible to
find a log–log model significantly better than that of Ta-
ble 6s, which retains the [Al]
d
term but with log
6
RB
substituting pH and, moreover, with an additional
8
clay
term included (
r
2
=
0
.
65,
p<
0
.
0001,
1
AIC
= −
20.5). Fur-
ther, modifying this “blindly selected” equation, by rein-
serting our previously rationalised pH term in preference
to log
6
RB
term (thus effectively adding a
8
clay
term to
the Equation of Table 6s), results in a markedly inferior
fit (
1
AIC
= +
10.3). Nevertheless, the resulting equation,
viz. [C]
=
pH
+
log[Al]
d
+
log(
8
S
CN
)
+
8
clay
(
r
2
=
0
.
63), is
still a marked improvement on the equation of Table 7v
(
1
AIC
= −
10.2).
For the smaller arenic soils dataset (
n
=
10) the lowest
AIC linear model is as in Table 4h (i.e. with, combined to-
gether, clay and silt only;
r
2
=
0
.
31,
p
=
0
.
035). However
we do note that there does exist a virtually uninterpretable
log–log model found through the AIC minimisation proce-
dure which involves all of pH (negative coefficient),
8
sand
[Al]
d
, [Fe]
d
and
8
S
CN
(positive coefficients) with an impres-
sive sounding
r
2
=
0
.
85 (but due to the low degrees of free-
dom for which
p
is only < 0.039).
3.7
Checking for model biases
In order to check if there were any systematic biases in the
final models used (the models as presented in Table 4b for
LAC soils, Table 4i for arenic soils and Table 6s for HAC
soils), standardised model residuals were examined in rela-
tion to the soil variables
8
sand
,
8
clay
,
8
silt
, [Al]
d
, [Al]
o
[Fe]
d
,
pH and CN ratio along with mean annual temperature
T
A
and mean annual precipitation
P
A
climate variables and two
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 20
72
C. A. Quesada et al.: Variations in soil chemical and physical properties
Figure 9.
The relationship of leaf litter lignin content with both leaf litter and soil C : N ratios.
(a)
Soil C : N ration as a function of leaf litter
C : N ratio,
(b)
soil C : N ratio as a function of leaf litter lignin concentration and
(c)
leaf litter C : N ratio as a function of leaf litter lignin.
Figure 10.
Fitted vs. observed SOC densities for regression
model 6s (Table 6).
vegetation-associated characteristics available for over 100
of the study sites, viz. the above-ground wood productivity
and above-ground biomass. These data are essentially as in
Quesada et al. (2012) but in an updated and expanded form
(Brienen et al., 2015). These relationships are shown in the
Appendix Fig. A1, which shows that there was little if any
evidence of systematic model bias, with the strongest asso-
ciation found for the standardised residuals being with
P
A
(
τ
=
0
.
09,
p
=
0
.
18).
3.8
SOC fractions and mineralogy
Further adding to our analysis, Table 8 shows results for
soil carbon fractions for a subset of our study sites (
n
=
30).
The [C] range in this reduced dataset is similar to the main
dataset, with LAC soils ranging from 8.8 to 25.3 mg g
−
1
,
with the arenic group ranging from 4.2 to 108.6 mg g
−
1
and with the HAC soils ranging from 5.5 to 24.8 mg g
−
1
. It
also shows very similar relationships between the relevant
edaphic parameters and [C] as found for the larger dataset
and described in Sect. 3.2. Comparing the Kendall
τ
from
Table 8 with results from Table 3, we find very similar cor-
relations for both LAC and for all groups combined, but
with [C] in the reduced dataset having stronger correlations
with clay content and Al
d
in LAC soils (
τ
=
0
.
64,
p<
0
.
01
and
τ
=
0
.
61,
p<
0
.
01, respectively). The main difference
between datasets occurs in HAC soils, where the reduced
dataset used for fractionations shows stronger correlations
between [C] and both clay content and
I
E
(
τ
=
0
.
49,
p<
0
.
02
and
τ
=
0
.
72,
p<
0
.
001, respectively) than is the case in the
larger dataset (Table 3).
Soil C fractionations revealed fundamental differences be-
tween the three soil groups as shown in detail in Fig. 11. LAC
soils (Fig. 11a) had an average fraction of 0.49 (or 49 %) C
in clay-rich aggregates (sand and aggregates fraction,
S
+
A
),
with this increasing with [C] up to 0.74. This increase in
S
+
A
fraction in high [C] soils seems to occur at the expense
of the labile clay and silt fraction (
C
+
S
), which represents
a fraction of 0.20 of soil carbon on average, but only 0.09 in
the higher [C] soils. The proportion of C in POM and DOC
fractions varied little across the range of soil [C], while the
resistant carbon associated with clay and silt (
R
C
+
S
) aver-
aged 0
.
2
±
0
.
07 and showed no clear pattern.
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 21
C. A. Quesada et al.: Variations in soil chemical and physical properties
73
T
ab
le
8.
K
endall’
s
τ
correlations
for
soil
or
g
anic
carbon
fractions
and
a
range
of
soil
and
mineralogical
properties.
F
our
one-sided
correlation
m
atrices
are
sho
wn
for
each
of
the
arenic,
LA
C
and
HA
C
clusters
as
well
as
for
the
(combined)
dataset
as
a
whole.
Data
sho
wn
here
are
a
subset
of
our
entire
dataset
(
n>
30).
W
e
ha
v
e
indicated
in
bold
all
cases
where
the
probability
of
T
ype-II
error
is
less
than
0.05.
F
or
the
entire
dataset,
a
τ
>
0
.
22
is
associated
with
a
probability
of
p
<
0
.
05.
F
or
the
HA
C
soil
cluster
with
n
=
13
the
equi
v
alent
v
alue
is
τ
>
0
.
36,
and
for
LA
C
with
n
=
12
the
v
alue
is
τ
>
0
.
39.
F
or
the
arenic
soil
cluster
,
with
only
n
=
5,
the
associated
probability
of
p
<
0
.
05
requires
a
τ
>
0
.
80.
W
e
ha
v
e
indicated
all
cases
where
one
or
more
of
the
four
groupings
has
p
>
0
.
05
–
using
italicised
characters
to
help
cross-referencing
across
the
four
diagonal
matrices.
Lo
w-acti
vity
clay
soils
[C]
CN
[
C
]
cs
[
C
]
r
[
C
]
sa
[
C
]
pom
[
C
]
doc
Sand
Clay
Silt
pH
I
E
TRB
[Fe]
d
[Fe]
o
[Fe]
do
[Al]
d
[Al]
o
q
1
q
2
All
[C]
−
0.03
0.24
0.33
0.55
0.55
0.48
−
0.67
0.64
0.30
−
0.03
0.33
0.12
0.53
−
0.09
0.55
0.61
−
0.03
−
0.48
−
0.52
soils
CN
0.20
0.30
0.39
−
0.18
0.00
0.30
−
0.24
0.21
−
0.06
−
0.45
0.21
−
0.61
−
0.14
0.09
−
0.24
−
0.06
0.15
0.12
0.03
combined
[
C
]
cs
0.43
−
0.09
0.79
−
0.21
0.21
0.45
−
0.27
0.24
0.45
−
0.36
0.36
−
0.27
0.11
0.12
0.15
0.33
0.00
0.03
−
0.24
[
C
]
r
0.49
0.07
0.56
−
0.12
0.42
0.55
−
0.36
0.33
0.30
−
0.33
0.39
−
0.36
0.08
0.03
0.12
0.30
0.03
0.00
−
0.27
[
C
]
sa
0.42
0.42
−
0.10
−
0.01
0.15
0.27
−
0.52
0.55
0.03
0.18
0.30
0.15
0.44
−
0.06
0.39
0.27
0.06
−
0.39
−
0.48
[
C
]
pom
0.44
0.20
0.42
0.31
0.10
0.21
−
0.33
0.30
0.27
−
0.06
0.18
0.03
0.29
−
0.24
0.39
0.33
−
0.24
−
0.21
−
0.18
[
C
]
doc
0.68
0.33
0.32
0.46
0.48
0.17
−
0.52
0.61
0.03
−
0.24
0.48
−
0.27
0.26
0.00
0.27
0.39
0.06
−
0.39
−
0.55
sand
−
0.31
0.18
−
0.27
−
0.32
−
0.05
−
0.03
−
0.29
−
0.91
−
0.21
0.18
−
0.48
0.21
−
0.44
−
0.06
−
0.33
−
0.52
−
0.12
0.27
0.55
clay
0.34
−
0.06
0.11
0.32
0.23
−
0.14
0.48
−
0.66
0.12
−
0.21
0.52
−
0.18
0.41
0.09
0.30
0.48
0.15
−
0.30
−
0.64
silt
0.01
−
0.27
0.43
0.15
−
0.31
0.15
−
0.12
−
0.32
−
0.02
0.06
0.06
0.15
0.29
0.24
0.15
0.33
-0.06
0.15
−
0.24
pH
−
0.07
−
0.53
0.17
−
0.01
−
0.32
0.01
−
0.21
−
0.06
−
0.06
0.33
−
0.33
0.30
−
0.23
−
0.03
−
0.18
−
0.18
0.03
0.00
−
0.09
I
E
0.20
−
0.34
0.37
0.31
−
0.09
0.13
0.11
−
0.41
0.35
0.16
0.32
−
0.36
0.11
−
0.03
0.18
0.12
0.09
−
0.18
−
0.33
TRB
0.02
−
0.66
0.35
0.17
−
0.35
−
0.02
−
0.15
−
0.25
0.15
0.40
0.58
0.42
0.29
−
0.06
0.33
0.09
−
0.30
−
0.21
0.12
[Fe]
d
0.29
−
0.25
0.16
0.38
0.01
−
0.04
0.27
−
0.38
0.49
0.03
0.13
0.34
0.34
0.08
0.69
0.72
−
0.14
−
0.32
−
0.35
[Fe]
o
0.03
−
0.32
0.24
0.32
−
0.23
−
0.04
0.02
−
0.32
0.34
0.16
0.22
0.40
0.42
0.47
−
0.24
0.24
0.58
0.24
−
0.15
[Fe]
do
0.21
−
0.19
−
0.01
0.20
0.15
−
0.20
0.26
−
0.24
0.37
−
0.15
−
0.01
0.17
0.15
0.58
0.04
0.45
−
0.30
−
0.52
−
0.24
[Al]
d
0.21
−
0.01
−
0.02
0.24
0.19
−
0.18
0.32
−
0.36
0.55
−
0.09
−
0.15
0.15
0.05
0.64
0.34
0.43
0.12
−
0.33
−
0.61
[Al]
o
0.09
−
0.02
−
0.01
0.22
0.10
−
0.20
0.22
−
0.32
0.47
−
0.03
−
0.06
0.22
0.07
0.32
0.53
0.11
0.63
0.12
−
0.27
q
1
−
0.16
−
0.22
0.28
−
0.03
−
0.44
0.20
−
0.35
0.10
−
0.31
0.34
0.39
0.04
0.29
−
0.20
0.04
−
0.32
−
0.46
−
0.35
0.30
q
2
−
0.35
−
0.16
−
0.09
−
0.30
−
0.22
−
0.13
−
0.35
0.26
−
0.30
0.09
0.12
0.03
0.15
−
0.26
−
0.03
−
0.19
−
0.28
−
0.08
0.10
High-acti
vity
clay
soils
[C]
CN
[
C
]
cs
[
C
]
r
[
C
]
sa
[
C
]
pom
[
C
]
doc
sand
clay
silt
pH
I
E
TRB
[Fe]
d
[Fe]
o
[Fe]
do
[Al]
d
[Al]
o
q
1
q
2
[C]
0.10
0.74
0.62
0.43
0.72
0.73
−
0.26
0.49
0.10
0.36
0.72
0.49
0.38
0.28
−
0.05
0.18
0.18
0.23
−
0.26
CN
0.40
−
0.05
0.28
0.37
0.18
0.25
0.44
−
0.21
−
0.33
−
0.28
−
0.08
−
0.31
0.05
0.10
−
0.03
0.15
0.21
−
0.41
0.18
[C]
cs
1.00
0.40
0.56
0.22
0.67
0.60
−
0.36
0.59
0.21
0.31
0.62
0.64
0.38
0.38
−
0.05
0.23
0.23
0.49
−
0.31
[C]
r
0.80
0.60
0.80
0.19
0.44
0.76
−
0.13
0.41
−
0.03
0.18
0.44
0.36
0.62
0.41
0.13
0.41
0.31
0.10
−
0.49
[C]
sa
0.60
0.40
0.60
0.40
0.45
0.32
0.17
−
0.04
−
0.12
0.19
0.25
−
0.09
0.04
−
0.04
0.01
−
0.06
−
0.09
−
0.09
0.14
[C]
pom
1.00
0.40
1.00
0.80
0.60
0.54
−
0.23
0.36
0.08
0.18
0.49
0.31
0.36
0.51
−
0.28
0.31
0.41
0.15
−
0.13
[C]
doc
1.00
0.40
1.00
0.80
0.60
1.00
−
0.09
0.41
−
0.07
0.30
0.60
0.38
0.49
0.36
0.09
0.17
0.20
0.12
−
0.30
sand
−
0.60
−
0.40
−
0.60
−
0.40
−
0.60
−
0.60
−
0.60
−
0.56
−
0.69
−
0.33
−
0.33
−
0.51
−
0.10
−
0.26
0.18
0.00
−
0.15
−
0.41
0.33
clay
0.20
0.00
0.20
0.00
0.60
0.20
0.20
−
0.60
0.26
0.31
0.41
0.64
0.28
0.44
−
0.15
0.13
0.28
0.49
−
0.46
silt
0.60
0.40
0.60
0.40
0.60
0.60
0.60
−
1.00
0.60
0.33
0.23
0.36
−
0.05
0.05
−
0.08
−
0.10
0.10
0.21
−
0.13
pH
0.32
0.53
0.32
0.53
−
0.11
0.32
0.32
0.11
−
0.53
−
0.11
0.59
0.56
0.26
0.00
0.18
−
0.05
−
0.10
0.46
−
0.23
I
E
0.11
0.11
0.11
−
0.11
0.53
0.11
0.11
−
0.11
0.11
0.11
0.00
0.56
0.26
0.21
−
0.08
0.00
0.05
0.36
−
0.13
TRB
0.60
0.00
0.60
0.40
0.20
0.60
0.60
−
0.60
0.60
0.60
−
0.11
−
0.32
0.28
0.23
0.00
0.13
0.13
0.64
−
0.36
[Fe]
d
0.00
−
0.20
0.00
0.20
−
0.40
0.00
0.00
0.00
0.00
0.00
0.32
−
0.53
0.40
0.44
0.21
0.64
0.28
0.13
−
0.62
[Fe]
o
−
0.60
−
0.40
−
0.60
−
0.40
−
1.00
−
0.60
−
0.60
0.60
−
0.60
−
0.60
0.11
−
0.53
−
0.20
0.40
−
0.36
0.49
0.74
0.08
−
0.36
[Fe]
do
0.60
0.00
0.60
0.40
0.60
0.60
0.60
−
0.20
0.60
0.20
−
0.11
0.11
0.60
0.00
−
0.60
0.05
−
0.41
0.05
−
0.23
[Al]
d
0.00
−
0.60
0.00
−
0.20
0.00
0.00
0.00
0.00
0.40
0.00
−
0.11
−
0.11
0.40
0.60
0.00
0.40
0.49
0.03
−
0.36
[Al]
o
0.60
0.00
0.60
0.40
0.60
0.60
0.60
−
0.20
0.60
0.20
−
0.11
0.11
0.60
0.00
−
0.60
1.00
0.40
−
0.03
−
0.15
q
1
−
0.20
0.00
−
0.20
0.00
−
0.60
−
0.20
−
0.20
0.20
−
0.60
−
0.20
0.53
−
0.53
−
0.20
0.40
0.60
−
0.60
0.00
−
0.60
−
0.21
q
2
0.00
−
0.20
0.00
−
0.20
0.40
0.00
0.00
0.00
0.40
0.00
−
0.74
0.32
0.00
−
0.60
−
0.40
0.40
−
0.20
0.40
−
0.80
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 22
74
C. A. Quesada et al.: Variations in soil chemical and physical properties
Figure 11.
Proportion of total soil carbon in the different pools for
the three soil groups varying as a function of their SOC content.
(a)
LAC,
(b)
arenic and
(c)
HAC soils. Dissolved organic carbon
(DOC), particulate organic matter (POM), sand and aggregates (
S
+
A
), silt and clay (
S
+
C
), and resistant SOC associated with silt and
clay fractions.
On the other hand, the arenic group have most of their car-
bon associated with POM and
S
+
A
fractions (average pro-
portion of 0.47 and 0.25, respectively) (Fig. 11b, Table 8),
with the fraction of POM reaching 0.70 in soils with higher
overall [C]. Seasonally wet sands (denoted with
F
follow-
ing the soil type in Table 1) had the highest POM fractions,
averaging 0.6 of total [C], but despite the differences in [C]
related to soil drainage, POM and
S
+
A
fraction were still
the main stores of SOC in well drained sands (0.33 and 0.3
of total [C], respectively).
HAC soils consistently had most of their [C] associated
with the clay and silt fraction (0.43) and the resistant carbon
(0.28) associated with clay and silt (
R
C
+
S
). On average 0.72
of [C] was found in these two fine-earth fractions (Fig. 11c).
The
S
+
A
fractions only had on average 0.13 of HAC soils
[C], while POM and DOC had 0.13 and 0.01 respectively.
In general, the HAC fractions varied little in proportion to
increasing [C].
Soil C fractions in the three groups also differed in the
way they relate to other edaphic properties such as texture,
the abundance of Fe and Al oxides, and bulk soil mineralogy
(Table 8). In LAC, soil carbon associated with both
C
+
S
and
R
C
+
S
fractions did not show any significant correlation with
Fe and Al oxides, nor with clay content, but
C
+
S
was corre-
lated with soil silt content (Kendall
τ
=
0
.
45,
p<
0
.
025). On
the other hand, the
S
+
A
fraction, the main pool of SOC, was
significantly correlated to clay content (
τ
=
0
.
55,
p<
0
.
01).
S
+
A
was also negatively correlated with our PCA axis 1,
which indicates a positive relationship with the abundance of
1
:
1 clay minerals (see Sect. 3.2) as axis 1 (
q
1
Table 8) rep-
resents to a large degree the abundance of kaolinite, Goethite
and Gibbsite (Kendall
τ
= −
0
.
39,
p<
0
.
05).
S
+
A
was also
negatively correlated to sand content (Kendall
τ
= −
0.52,
p<
0
.
01),
S
+
A
was also significantly correlated to Fe ox-
ides (Kendall
τ
=
0
.
44,
p<
0
.
03 and 0.39
p<
0
.
05 for Fe
d
and Fe
d-o
; Table 8). The DOC fraction was significantly cor-
related to clay (Kendall
τ
=
0
.
61
p<
0
.
01),
I
E
(Kendall
τ
=
0
.
48,
p<
0
.
02) and Al
d
(Kendall
τ
=
0
.
39
p<
0
.
05). DOC
was also correlated to
q
1
(Kendall
τ
= −
0
.
39,
p<
0
.
05). The
POM fraction was significantly correlated to Fe
d
−
o
(Kendall
τ
=
0
.
39,
p<
0
.
05).
The small number of arenic soils in this analysis (
n
=
5)
makes correlations unreliable and difficult to interpret. At
n
=
5, a Kendall
τ
=
0
.
8 does not differentiate critical values
at
p
=
0
.
1 and 0.05, and significance can only be attained for
Kendall
τ
=
1. Therefore, correlations in Table 8 should be
taken just as a guidance for the direction of the relationship
and are not considered further here.
HAC fractions showed totally different correlations to
edaphic properties when compared to LAC soils. For exam-
ple, the
C
+
S
fraction was significantly correlated to clay
content (
τ
=
0
.
59,
p<
0
.
01),
I
E
(
τ
=
0
.
62,
p<
0
.
01) and with
the weathering index TRB (
τ
=
0
.
64,
p<
0
.
01).
C
+
S
also
showed a positive correlation with PCA axis 1, indicating a
positive correlation with the abundance of 2
:
1 clays (
τ
=
0
.
49,
p<
0
.
02).
R
C
+
S
in HAC soils also showed an effect
of both Fe
d
and Al
d
(Kendall
τ
=
0
.
62,
p<
0
.
01 and 0.41,
p<
0
.
04, respectively) and
I
E
(Kendall
τ
=
0
.
44,
p<
0
.
03).
In striking difference to LAC,
S
+
A
in HAC soils was
an insignificant storage for SOC and showed no significant
correlation to the concentration of any oxides, clay content
or any other of the measured parameters. DOC on the other
hand behaved in a more similar manner to LAC soils, also
showing significant associations with
I
E
(
τ
=
0
.
60
p<
0
.
01)
and clay content (
τ
=
0
.
41,
p<
0
.
04) and an iron oxide ef-
fect (Fe
d
:
τ
=
0
.
49,
p<
0
.
02). POM on the other hand was
correlated to Fe
o
(
τ
=
0
.
51,
p<
0
.
02) and Al
o
a (
τ
=
0
.
41,
p<
0
.
05) and
I
E
(
τ
=
0
.
49,
p<
0
.
02, respectively).
3.9
Carbon stocks versus carbon concentrations
Although the analysis here has focused on soil carbon con-
centrations, for carbon inventory purposes the actual carbon
stock (i.e. carbon per unit ground area:
C
S
) is usually of more
interest, and with the two being related according to
C
s
=
0
Z
d
[
C
]
z
·
ρ
z
·
d
z,
(1)
where [C]
z
and
ρ
z
represent the carbon concentrations and
bulk density of the soil at depth
z
below the soil sur-
face respectively and
d
is the maximum sampling depth.
Thus with the actual calculations done layer by layer (viz.
0 to 0.05 m, 0.05 to 0.10 m, 0.10 to 0.20 m and 0.20 to
0.30 m) Fig. 12a–c show the relationship between [C] and
ρ
for the three soil groups with regressions shown that
were significant at
p<
0
.
05 or better. This shows a rea-
sonably strong relationship for the LAC soils across the 0
to 0.3 m depth (Fig. 12a, log(
ρ
)
=
0.881–0.298
×
log[C]:
r
2
=
0
.
43,
p<
0
.
001) and with a similar though somewhat
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 23
C. A. Quesada et al.: Variations in soil chemical and physical properties
75
less convincing relationship being observed for the HAC
soils (Fig. 12b, log(
ρ
)
=
0.678–0.219
×
log[C]:
r
2
=
0
.
25,
p<
0
.
001) but no readily discernable relationship evident for
the arenic soils (Fig. 12c, log(
ρ
)
=
0.697–0.233
×
log[C]:
r
2
=
0
.
20,
p<
0
.
08).
These negative [C] vs.
ρ
associations across all three soil
groupings necessitate that
C
S
is a saturating function of [C]
as is shown in Fig. 12d–f, with the slopes of the log–log scal-
ing relationships being 0
.
62
±
0
.
05 for LAC soils (Fig. 12d),
0
.
71
±
0
.
05 for the HAC soils (Fig. 12e), 0
.
23
±
0
.
15 for the
arenic soils (Fig. 12f) and 0
.
59
±
0
.
04 for the dataset as a
whole. This means, for example, that – on average – an in-
crease in [C] of 50 % will result in an increase in
C
S
of only
(1.5
0
.
59
−
1) or just 27 %.
This negative covariance between [C] vs.
ρ
also means
that within a given soil group variation in
C
S
is typically
much less than for [C]. For example, as is shown in Ta-
ble 9, the 12 RSGs examined show a lower coefficient of
variation for
C
S
than is the case for [C], with this difference
being especially marked for Cambisols (0.63 for [C] vs. 0.39
for
C
S
). Also shown in Table 9 are the mean
C
S
values for
the 12 RSGs we have examined as compared to the values
given by Batjes (1996), for which we note that in the ma-
jority of cases our estimates are surprisingly close, with one
exception being the Alisols for which our estimate of around
46 t C ha
−
1
is only 53 % that of the Batjes (1996) estimate of
ca. 86 t C ha
−
1
to 0.3 m depth. Our Leptosols and Podzol
C
S
estimates are also much higher than those of Batjes (1996).
4
Discussion
According to our analysis, the three soil groups studied here
are characterised by different soil C stabilisation mecha-
nisms. Specifically, highly weathered soils, dominated by
low-activity clays such as Ferralsols and Acrisols (our LAC
group) have SOC densities that are strongly dependent on
their clay and silt contents. However, such simple relation-
ships with fine-earth fractions could not explain SOC vari-
ations for the less weathered soils. For the HAC group-
ing, SOC stabilisation is predominantly related to interac-
tions with Al, and the formation of Al/organic matter co-
precipitates. For our arenic soils group, it appears that most
of the SOC present is in loose particulate organic matter form
and is therefore not stabilised by mineral interactions, though
with a surprisingly strong effect of their small clay and silt
content variations.
Such differences in the stabilisation mechanisms can arise
from the different soils examined being at contrasting pedo-
genetic development stages and/or differences in parent ma-
terial. Highly weathered soils such our LAC group have been
under constant tropical weathering rates for timescales that
range from 100 million to 2 billion years (Hoorn et al., 2010;
Quesada et al., 2011), with some of the central and eastern
Amazon Basin soils having suffered several cycles of weath-
ering (Herrera et al., 1978; Irion, 1978; Quesada and Lloyd,
2016). This extreme weathering of LAC soils has resulted in
a deep uniformity in their mineralogy, which is dominated
by kaolinite (Sombroek, 1984), and in the depletion of rock-
derived elements. It has also resulted in the development of
favourable soil physical properties such as free drainage, low
bulk densities and the formation of very deep soil horizons
(Quesada et al., 2010).
Nevertheless, it also needs to be remembered that the
Amazon Basin has a complex mosaic of soils, with ca. 40 %
having young and intermediate pedogenetic development
levels (Quesada et al., 2011; Richter and Babbar, 1991;
Sanchez, 1976). Most of these less weathered soils occur in
the west of the basin and were influenced by the uprising of
the Andean Cordillera (Hoorn et al., 2010), thus having much
younger geological ages. Much of the soil formation process
in this region only came into effect after the Pliocene, with
most of the substrate in that region having less than 2 million
years (Hoorn and Wesselingh, 2011; Quesada et al., 2011;
Quesada and Lloyd, 2016). Soils in that region have a di-
verse mineralogy, with a high abundance of 2
:
1 clays and
sometimes also some rock-derived easily weatherable miner-
als (Irion, 1978; Quesada et al., 2010, 2011; Sombroek, 1966,
this study). One important characteristic of many HAC soils
is the very high amount of Al that is released through the
weathering of 2
:
1 clays (Marques et al., 2002). High active
clays are unstable in environments depleted of silica, alkaline
and alkaline earth cations, thus releasing soluble aluminium
from the octahedral internal layers of the 2
:
1 clay minerals,
with such Al release also increasing with depth (Quesada et
al. 2011).
The arenic soil group on the other hand is strongly influ-
enced by its parent material. It comprises the Arenosol and
Podzol reference groups, with the latter also being predom-
inantly sandy in Amazonia (Do Nascimento et al., 2004).
Both soil types are thought to have evolved from the weath-
ering of aeolian and riverine sediments of siliceous rocks,
or in some cases, being locally weathered and deposited in
colluvial zones through selective erosion (Buol et al., 2011;
Driessen et al., 2000). As quartz usually makes up more than
90 % of their mineral fraction, their surface exchange capac-
ity is very small, resulting in very low nutrient levels as a
consequence of a high degree of leaching (Buol et al., 2011;
Quesada et al., 2010, 2011). The very low nutrient content
of these soils, often associated with high groundwater levels,
results in the formation of thick root mats in the soil sur-
face (Herrera et al., 1978) which then strongly influences the
amount and vertical distribution of their SOC stocks.
Therefore, our HAC, LAC and arenic soils groups consist
of very different soils, with contrasting geological formation
and chemical and physical properties. Not surprisingly, such
wide variations also resulted in different mechanisms of SOC
stabilisation.
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 24
76
C. A. Quesada et al.: Variations in soil chemical and physical properties
Figure 12.
Variations in bulk density
(a)
LAC,
(b)
HAC and
(c)
arenic soils, and top-soil SOC stocks, and SOC stocks of
(d)
LAC,
(e)
HAC
and
(f)
arenic soils as a function of SOC content. Significant regression lines (see text for details) for each soil group are plotted together for
comparison.
Table 9.
Mean soil organic carbon stocks (0–30 cm) for 12 RSGs examined in this study. Stocks from Batjes (1996) are also given for
comparison. CV stands for coefficient of variation.
RSG
n
Soil carbon concentration
Soil carbon stock
SOTER-LAC estimated
soil carbon stock
Mean (mg g
−
1
)
CV
Mean (t ha
−
1
)
CV
Mean (t ha
−
1
)
CV
Acrisol
18
16.3
0.35
49.5
0.27
44.0
0.50
Alisol
20
16.6
0.28
45.6
0.27
85.7
0.42
Arenosol
6
12.3
0.23
29.6
0.31
20.7
0.50
Cambisol
19
21.3
0.63
58.9
0.39
55.9
0.61
Ferralsol
34
17.1
0.35
47.3
0.26
50.5
0.48
Fluvisol
5
21.0
0.33
54.6
0.33
34.2
0.52
Gleysol
10
24.5
1.03
70.1
0.84
67.4
0.62
Leptosol
2
32.0
0.75
115.2
0.63
51.5
0.63
Lixosol
3
21.9
0.36
65.4
0.17
38.5
0.45
Luvisol
2
15.3
0.57
43.8
0.46
46.7
0.51
Plinthosol
18
14.2
0.40
41.1
0.44
34.0
0.48
Podzol
7
48.3
0.92
98.9
1.32
54.9
0.54
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 25
C. A. Quesada et al.: Variations in soil chemical and physical properties
77
4.1
Mechanisms of SOC stabilisation
4.1.1
SOC stabilisation in low-activity clay soils
Since soil C content might reasonably be expected to de-
pend, at least in part, on specific surface area (SSA) because
a higher density of exchange sites per unit volume should
result in more soil carbon stabilisation through mineral–
organic-matter associations (Saidy et al., 2012), the uniform
mineralogy of 1
:
1 soils means that, as shown in Fig. 5 and
elsewhere (Burke et al., 1989; Dick et al., 2005; Feller and
Beare, 1997; Telles et al., 2003), for LAC soil organic C
scales linearly with clay content since the variation in clay
content is the main source of variation in SSA.
The observed variation in clay fractions across LAC soils
studied here was large, from 0.05 to 0.89. This reflects dif-
ferences in parent material, with Acrisols tending to have
sandier top soils (West et al., 1997). Central and East Ama-
zonia are known for having very clay-rich soils, often having
clay content well above 60 % (Chauvel et al., 1987; Som-
broek, 1966), with such clays originating from ancient fluvio-
lacustrine sediments deposited on the Barreiras and Alter do
Chão geological formations locally known as Belterra clays
(Sioli, 1984; Sombroek, 1966, 2000). Other regions where
Ferralsols dominate, such as the southern fringe of the basin
(Quesada et al., 2010), often have much sandier soils.
The uniformity in the clay vs. C relationships shown by
our best OLS models indicates an overruling effect of clay
content and with some effect from silt (Table 4). The supe-
rior predictive power of sand content (
−
[clay
+
silt]), com-
pared to clay, as a main determinant of SOC in highly
weathered tropical soils has already been shown by Saiz et
al. (2012), with these authors concluding that sand content
shows less confounding effects than that of clay in these sys-
tems. The association of clay with aluminium and iron ox-
ides in highly weathered tropical soils may promote the for-
mation of sesquioxides. Saiz et al. (2012) have shown that
these particles confer the soil a coarse-like texture, which
exerts a strong influence on soil bulk density and water re-
tention properties. Furthermore, results from Fig. 3a, c also
suggest a wide variation of Fe oxides to occur on LAC soils,
with Fig. 6 and Tables 3 and 5 indicating that the abundance
of Al
d
is also correlated with SOC. This could be related to
increments in SSA resulting from the greater abundance of
such minerals (Eusterhues et al., 2005; Kleber et al., 2005;
Wiseman and Püttmann, 2006; Saidy et al., 2012) in which an
increment in the number of exchange sites may provide addi-
tional stabilisation of carbon via direct complexation (Parfitt
et al., 1997; Schwertmann et al., 2005) and with direct inter-
actions between SOC, Fe and Al oxides, with clay particles
(Wiseman and Püttmann, 2006) also being important. How-
ever, Fe and Al hydroxides may also indirectly protect car-
bon from decomposition through their role in the formation
of stable aggregates which make carbon physically inaccessi-
ble to decomposers (Kitagawa, 1983; Six et al., 2004; Wagai
and Mayer, 2007). This may be of importance for LAC soils
since stable clay aggregates were found to store most of the
SOC (Sect. 3.5).
Using the Zimmerman et al. (2007) soil carbon fraction-
ations to gain further insights into the stabilisation mecha-
nisms that underlie soil organic matter dynamics (Denef et
al., 2010), Fig. 11a shows that the sand and aggregate (
S
+
A
)
fraction is responsible for holding most of the SOC in LAC
soils. This fraction is essentially formed by a mixture of clay,
silt, oxides and organic matter, and within this fraction ag-
gregation may promote increased SOC protection as it influ-
ences the accessibility of substrate to microorganisms, thus
limiting the extent that the diffusion of reactants and prod-
ucts from extracellular synthesis (i.e. soil enzymes) can reach
the organic matter (Sollins et al., 1996). For example, pore
spaces inside aggregates can be too small to allow access
of bacteria (Van Veen and Kuikman, 1990) and efficient en-
zyme diffusion (Sollins et al., 1996). This then retains SOC
in inaccessible micropores inside aggregates (Baldock and
Skjemstad, 2000), which ultimately protects SOC from de-
cay, explaining the positive correlation often found between
the level of soil aggregation and SOC concentration (Six et
al., 2004; Tisdall and Oades, 1982).
Soil aggregation level is also affected by other chemi-
cal, microbial, plant, animal and physical processes, many
of which seem to be favoured by the tropical climate and
thriving biological activity of the tropical moist forest envi-
ronment. For instance, microbial activity releases polysac-
charides that act as binding agents in soil aggregates (Lynch
and Bragg, 1985; Oades, 1993) and fungal hyphae are known
to bind solid particles together (Sollins et al., 1996). Plant
roots also influence soil aggregation by releasing exudates
that can directly flocculate colloids and bind or stabilise ag-
gregates (Glinski and Lipiec, 1990). Root exudates may also
foster microbial activity which can lead to aggregate forma-
tion and stabilisation. Plant roots and associated hyphae can
also enmesh soil particles by acting as a “sticky string bag”
(Oades, 1993) which binds soil particles. Further, the pres-
sure exerted by roots and soil fauna on soil promotes aggre-
gation (Oades, 1993; Sollins et al., 1996). Soil fauna (includ-
ing earthworms, termites, collembola, beetles, isopods and
millipeds) form fecal pellets and excrete binding agents that
form aggregates (Oades, 1993; Sollins et al., 1996). Never-
theless, the presence of Fe and Al oxides in these soils may
also favour the formation of soil aggregates (Kitagawa, 1983;
Wagai and Mayer, 2007) since they act as binding agents
with clays in a process thought to be associated with the large
abundance of aggregates in Ferralsols and Acrisols (Paul et
al., 2008; Sanchez, 1976; Sollins et al., 1996).
Soil C stabilisation on the surface of Amazonian Ferralsols
and Acrisols (1
:
1 clays) is thus interpreted here as the sum-
mation of the effect of variations in kaolinite clay content
(varying SSA) and the additional physical protection given
by the extensive level of aggregation common to these soils.
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 26
78
C. A. Quesada et al.: Variations in soil chemical and physical properties
4.1.2
Processes of C stabilisation in arenic soils
Since quartz is devoid of significant surface area and ex-
change sites, the retention of SOC in sand-rich soils is diffi-
cult to predict on the basis of soil physiochemical properties
as there is no, or very little, mineral–organic-matter interac-
tion. Thus, the bulk SOC variation in our arenic soil group
most likely reflects varying edapho-environmental condi-
tions such as groundwater levels and/or moisture regimes,
vertical root distribution and/or litter quality. However, small
changes in clay and silt content were still found to have large
effects on soil [C] (Table 4), with this OLS regression giv-
ing a slope 10 times greater than that of LAC soils. This
is similar to what Hartemink and Huting (2008) found for
150 Arenosols in southern Africa, where soil carbon con-
tent varied from about 0.5 to 12 g kg
−
1
alongside a change in
clay fraction ranging from effectively zero to just 0.12. Sim-
ilar findings (i.e. 0.8 to 14.5 g kg
−
1
) were also obtained on
heavily coarse-textured soils sampled along a 1000 km mois-
ture gradient spanning from southern Botswana into southern
Zambia (Bird et al., 2004).
In addition, groundwater fluctuations and the often ex-
tremely low nutrient availability of these soils often result
in the formation of root mats, covering the top 10 to 50 cm of
the soil surface with an impressive mixture of roots and or-
ganic matter in different stages of decomposition (Herrera et
al., 1978). Such soil mats may reasonably be expected to ex-
ert a strong influence on soil SOC concentrations, since they
concentrate the inputs of organic matter into a single layer
close to the surface. Moreover, because many of these soils
are seasonally waterlogged (Quesada et al., 2011) the associ-
ated anaerobic conditions should also inhibit decomposition.
It is therefore not a surprise that we observed some of the
highest [C] in these soils.
Our fractionation results again provided additional infor-
mation for the understanding of SOC retention, with the bulk
of the SOC in arenic soils found as free particulate organic
matter, and with this proportion increasing as [C] increases
(Fig. 11b). This was particularly the case for seasonally wet
sands (up to 60 % of SOC), but with POM also being a sig-
nificant fraction of the total SOC even in the drier sands
(
∼
30 %). The implication here is that chemical recalcitrance
of organic matter may also have a role in these soils: favour-
ing the maintenance of residual, slowly decaying organic par-
ticles.
High chemical recalcitrance may be common in such ex-
tremely dystrophic arenic soils, with total P levels often as
low as 10 mg kg
−
1
. Total P in LAC soils is ca. 10 times
greater than in arenic soils, and generally 20–50 times greater
in HAC soils (Quesada et al., 2010). Such a low level of nu-
trient content often results in high levels of plant investment
in secondary defense compounds against herbivory (Coley et
al., 1985; Fine et al., 2004), and such chemical recalcitrance
may affect the decomposition process and thus slightly in-
crease residence time of uncomplexed C in the soil. This may
affect POM levels particularly, considering that the most re-
calcitrant part will have a slower turnover, or be left unde-
composed following microbial attack. This is given support
by the observations made by Luizão and Schubart (1987),
who found that leaf litter decomposition in Amazonian white
sands takes twice as long than for Ferralsols and Acrisols
during the dry season, and nearly 7 times longer in the wet
season when decomposition is more dynamic in the non-
white sand soils. Organic acids from residual decomposition
from these soils are known to colour the rivers of the region –
the Rio Negro, with its head waters within a vast white sand
forest region (Quesada et al., 2011), gets its name by virtue
of its high humic and fulvic acid content (Fittkau, 1971).
4.1.3
SOC stabilisation in high-activity clay soils
Our results suggest that Al–organic-matter (Al–OM) interac-
tion, or co-precipitation is a fundamental mechanism of SOC
stabilisation for the less weathered HAC forest soils of the
Amazon Basin, with the OLS models presented here involv-
ing complex interactions between Al species (Al
d
), soil pH
and the abundance of aromatic, carboxyl-rich organic mat-
ter. The complexity of the models and their high ability to
explain SOC densities suggest that this mechanism is funda-
mental to an understanding of HAC soil C storage.
To our knowledge this is the first time that Al–OM inter-
actions have been suggested as a key factor explaining SOC
densities for Amazon forest soils. Nevertheless, with DOC
being ubiquitously present in such a highly dynamic sys-
tem, and with exchangeable Al often abundant as has already
been shown to be the case in western Amazon soils (Que-
sada et al., 2010, 2011; Marques et al., 2002; this study),
it is intuitive that Al–OM interactions should encompass a
continuum from low-polymeric metal–organic complexes to
highly crystalline phases with surface attached organic mat-
ter (Kleber et al., 2015). Thus Al–OM interactions forming
co-precipitates are likely to be a widespread mechanism that
has previously been overlooked because most of the studies
in the Amazon Basin have to date only focused on highly
weathered soils such as Ferralsols and Acrisols (i.e. Telles et
al., 2003). Nevertheless, with less weathered soils occupying
circa 40 % of the Amazon Basin (Quesada et al., 2011), it is
important to further investigate the role of Al–OM interac-
tions, in particular with regard to their influence over SOC
mean residence times (MRTs), since they are likely to be dif-
ferent from what is known for Ferralsols. For example, MRT
of SOC in Amazon Ferralsols is about 10 years (Trumbore
and Barbosa De Camargo, 2009) as determined by
14
C stud-
ies, but to our knowledge, no
14
C information is available
for western Amazon soils, nor is such information is avail-
able for MRT of Al–OM co-precipitates. As organic poly-
electrolytes reorganise on mineral surfaces over time, they
form additional polar covalent bonds, and this ageing process
can then lead to a decreased desorbability of OM (Kleber et
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 27
C. A. Quesada et al.: Variations in soil chemical and physical properties
79
al., 2015). This means that MRT of Al–OM co-precipitates
could well extend to decades or even centuries.
In that respect, it is clear that organic matter becoming
co-precipitated with Al results in it becoming more resis-
tant to microbial decay (Kalbitz and Kaiser, 2008; Nierop et
al., 2002). At Al–OM concentrations typical of forest soils,
up to 80 % of DOC can co-precipitate (Nierop et al., 2002;
Scheel et al., 2007), with mineralisation rates of Al–OM co-
precipitates formed from DOM much lower than the com-
pounds from which the original, unprecipitated DOM origi-
nates (Boudot et al., 1989; Scheel et al., 2007). For instance,
using incubations, Scheel et al. (2007) found that the min-
eralisation extent of Al–OM precipitates ranged from 0.5 %
to 7.7 % while the DOM that originated the precipitates had
much higher rates (5 % to 49 %). Kalbitz and Kaiser (2008)
found that up to 50 % of total SOC in their study site was
stabilised from DOM following Al–OM interaction, with the
authors suggesting that Al co-precipitation has a stronger ca-
pacity to reduce mineralisation than sorption in phyllosili-
cates.
The formation of Al–OM co-precipitates is influenced by
several factors and interacting processes with, according to
the extensive review from Kleber et al. (2015), the most im-
portant factors being the prevalent metal-to-carbon ratios in
the soil solution (M
/
C), the presence of aromatic organic
compounds, the pH value of soil solution and the metal
species present (in which Fe also may have a role). Increas-
ing M
/
C ratios increase the probability of reaction with OM
while the solution pH controls the solubility and speciation
of metals (Al, Fe). With an increasing pH, the efficiency of
the process increases, causing larger amounts of precipitates
(Scheel et al., 2007). Also, co-precipitation occurs preferen-
tially with aromatic, carboxyl-rich organic structures such as
those derived from lignin and tannin decomposition due to
their higher affinity for Al complexation sites (Scheel et al.,
2007, 2008; Kleber et al., 2015), interactions which were also
made clear through the importance of litter lignin content and
soil C : N ratio in our OLS results. With regard to metal spe-
ciation, our OLS models selected for dithionite extractable
Al (Al
d
) which, having a broad capacity to extract Al bear-
ing minerals, we interpret as a continuum of likely different
forms such as free Al (Al
+
3
), Al from Al-interlayer miner-
als, Al–OM complexes, and both crystalline and amorphous
Al hydroxides (particularly at higher pH values).
In interpreting the use of soil C : N ratios as a surrogate
for litter quality it needs to be borne in mind, however, that
because log(C : N)
=
log[C]
−
log[N], this means that em-
bedded in equation 6s is what is known as a “whole-part”
correlation (Chayes, 1971) Formula and randomisation tech-
niques exist to estimate the extent to which slopes and corre-
lation coefficients may be biased by the presence of the same
terms on both sides for OLS regression equations if their (co-
)variances and/or correlations are known (Bartko and Pet-
tigrew, 1968; Lloyd et al., 2013). But unfortunately, due to
complex interactions between the fitted terms in such a situ-
ation, these cannot be readily applied in a multivariate con-
text (Lloyd et al., 2013). Nevertheless we can say that, even
though the observed (bivariate) correlation between log[C]
and log(CN) for the HAC soils in our dataset is relatively
low (
r
2
=
0
.
23) it is almost certain that the relatively steep
log–log slope of 1.16 for the log(CN) effect within Table 6
is inflated. Thus, caution would need to be exercised in ap-
plying this equation in any sort of predictive framework. Fur-
ther insights into carbon stabilisation mechanisms may again
be found from the fractionations study. Specifically, Fig. 11c
suggests that for HAC the Al–OM precipitates are held to-
gether within
C
+
S
fractions, this despite there being no sim-
ple correlations with clay fraction in the extended dataset.
Although this could perhaps be attributed to the use of only
a subset of sites used in the fractionation analysis, where the
reduced dataset shows stronger associations between [C] and
clay content, we suggest that such colloidal sized Al–OM
precipitates should be stored alongside the fine-earth frac-
tion. Remarkably 75 % of SOC that occurs is associated with
C
+
S
(and its resistant fraction) in these soils, with this frac-
tion being reasonably consistent across a range of soil [C].
4.2
Possible influences of confounding factors
As noted in the Introduction, our approach to modelling the
[C] storage potential has been primarily hypothesis based,
but also as noted in Sect. 3.6, there were some models that
– on the basis of their AIC – are statistically superior to
those presented as best models here. For example, in mod-
elling the [C] storage of HAC soils solely on the basis of soil
mineralogical properties, then a model also including both
Fe
o
and Al
o
seemed to be the best (equation of Table 6q).
Nevertheless, following our rationalisation that plant organic
matter quality inputs should also be important, once the sur-
rogate soil CN ratio metric was added to the model, then the
difficult-to-explain apparent negative Fe
o
effect became re-
dundant (equations of Tables 6r and 6s). Likewise in Sect. 3.6
we also noted that total reserve bases seemed to be a better
predictor than pH in a model of soil C stocks with [Al]
d
and
CN ratio as covariates. We chose pH for our final model on
the basis of its known effect on the SOC precipitation process
and with the apparent Total Reserve Bases effect rationalised
as a simple consequence of its high correlation with pH in
HAC soils (
τ
=
0
.
52,
p<
0
.
0001; Table 3).
Also, not included in our final models were the effects
of either mean annual temperature or precipitation, which,
as well as showing poor associations with SOM storage for
all three of our soil groups when considered individually as
well as when all soils were pooled together as a whole, also
showed no significant association with model residuals (Ap-
pendix Fig. 1). Nor – as is suggested by the lack of any sys-
tematic bias of model residuals with above-ground wood pro-
ductivity – was there any suggestion of variations in carbon
inputs influencing forest C stocks. This suggests that, across
the temperature and precipitation range of our dataset, litter
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 28
80
C. A. Quesada et al.: Variations in soil chemical and physical properties
input quality and soil mineral stabilisation mechanisms are
the primary determinants of the SOM storage variations: a
result which is consistent with microbial decomposition rates
acclimating to both temperature (Bradford et al., 2008) and
precipitation (Deng et al., 2012).
Our findings do not negate the possibility that future cli-
mate changes will have a significant impact on soil car-
bon stocks in the Amazon Basin. For example, Cotrufo et
al. (2013) have postulated that although interactions of or-
ganic materials within the soil mineral matrix are the ulti-
mate controllers of SOM stabilisation over long timescales,
it is the microbially mediated delivery of organic products
to this matrix that provides the critical link between plant
litter inputs and what products are available for stabilisa-
tion. In this respect a consideration of depths substantially
greater than the upper 0.3 m examined here must also be
critical for the accurate determination of any future changes
in climate stocks, as below 0.3 m Amazon Basin forest soil
C are generally quite low, and there is likely existing reac-
tive mineral surfaces yet to be saturated with SOM (Que-
sada, 2008; Quesada et al., 2010). Moreover, any future in-
puts into these lower layers, for example as might be medi-
ated though increased litter inputs as a consequence of CO
2
fertilisation (Lloyd and Farquhar, 2008), are likely to be mi-
crobially derived (Schrumpf et al., 2013). Quite likely the
extent of any such additional stabilisation of SOM at these
lower depths will differ between HAC, LAC and arenic soils
in accordance with the different stabilisation mechanisms as
suggested throughout this paper. But in the absence of more
detailed information and, indeed, precise confirmations as to
the apparent different mechanisms involved in SOM storage
as suggested here, then whether or not it is really the case
that Amazon forest soil C stocks are currently increasing in
response to higher litter inputs with soil developmental stage
also influencing that response must remain a matter of simple
conjecture.
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 29
C. A. Quesada et al.: Variations in soil chemical and physical properties
81
Appendix A
Figure A1.
Standardised regression model residuals plotted against selected climatic, edaphic and vegetation variables.
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 30
82
C. A. Quesada et al.: Variations in soil chemical and physical properties
Table A1.
Soil carbon and associated measures of the study soils (0.0–0.3 m). [C]: C concentration; CN: carbon–nitrogen ratio;
ρ
: bulk
density;
R
C
: total soil C; Fe
d
: dithionite extractable iron; Fe
o
: oxalate extractable iron; Al
d
: dithionite extractable aluminium; Al
o
: oxalate
extractable aluminium; Al
o
: pyrophosphate extractable aluminium.
Soil
[C]
CN
ρ
R
C
Fe
d
Fe
o
Fe
d
–Fe
o
Al
d
Al
o
Al
p
(mg g
−
1
)
(kg dm
−
3
)
(Mg ha
−
1
)
g kg
−
1
1
5.03
9.04
1.05
14.26
19.61
2.54
17.08
4.71
2.46
0.60
2
6.78
11.07
1.15
119.23
2.01
0.00
2.01
1.79
0.77
3
6.79
7.64
1.43
24.50
3.15
0.80
2.35
0.78
0.60
0.54
4
6.80
22.67
1.30
21.90
0.24
0.19
0.05
0.28
0.03
0.27
5
7.51
10.52
1.17
25.11
13.90
11.57
2.33
2.69
4.42
0.57
6
7.60
12.67
1.11
23.80
4.65
1.13
3.52
0.87
1.00
1.68
7
7.73
7.00
1.27
25.96
11.25
3.44
7.81
1.55
0.93
1.09
8
7.93
15.94
1.34
42.57
0.25
0.22
0.04
0.17
0.18
0.12
9
8.20
7.20
1.26
27.02
11.24
5.43
5.81
2.20
3.16
0.98
10
8.31
9.97
1.21
48.02
4.50
2.38
2.12
0.69
0.82
1.36
11
8.31
13.07
1.40
35.01
8.60
1.61
6.99
1.40
0.53
1.54
12
8.49
7.80
1.29
29.95
17.04
0.66
16.38
2.01
0.80
0.83
13
9.00
12.86
1.17
30.05
5.60
1.11
4.49
0.86
0.71
1.12
14
9.03
14.53
1.03
27.81
20.46
1.02
19.44
1.65
0.69
2.27
15
9.07
8.04
1.38
29.66
6.24
0.62
5.62
0.88
0.77
0.14
16
9.10
13.00
1.33
35.07
14.60
1.43
13.17
1.33
0.44
0.82
17
9.12
14.19
0.89
21.69
0.37
0.34
0.03
0.03
18
9.41
11.88
1.38
37.06
7.65
0.35
7.30
1.76
0.72
2.14
19
9.43
9.34
1.32
32.36
14.40
4.73
9.67
2.23
2.83
1.72
20
9.77
6.71
1.08
28.50
15.43
6.16
9.27
3.05
3.97
1.35
21
10.12
7.29
1.18
30.78
11.04
7.04
4.00
2.52
4.01
1.46
22
10.14
22.79
1.34
35.31
0.30
0.07
0.23
0.12
0.33
0.06
23
10.52
12.57
21.66
1.60
0.79
0.81
0.49
0.75
0.80
24
10.52
12.35
1.46
43.57
7.35
0.54
6.81
1.77
0.86
2.17
25
10.61
13.65
1.02
32.88
0.57
0.55
0.02
2.78
1.68
2.94
26
10.71
14.23
1.31
39.13
3.18
1.37
1.81
2.49
6.60
1.16
27
10.75
9.56
1.34
45.40
28
10.76
9.56
1.27
48.91
29
10.85
12.99
1.15
33.01
9.59
6.28
3.31
6.47
13.08
2.11
30
11.26
13.47
0.95
16.36
0.68
0.56
0.12
0.24
0.11
0.03
31
11.28
13.22
1.29
38.63
6.03
0.68
5.35
1.69
0.87
1.69
32
11.50
7.52
1.40
38.91
4.77
1.17
3.60
0.79
0.89
0.21
33
11.60
7.16
1.37
33.97
28.62
3.00
25.62
2.54
1.14
0.86
34
11.61
9.58
1.06
55.28
10.14
5.03
5.11
1.65
0.94
1.16
35
11.66
11.32
25.81
0.23
0.15
0.08
0.31
0.59
0.62
36
11.68
16.10
1.43
46.93
7.40
6.43
0.97
2.81
7.30
1.19
37
11.77
22.47
1.23
34.37
1.26
0.09
1.17
0.11
0.31
0.03
38
11.88
10.08
1.31
43.66
6.12
1.83
4.29
2.75
7.64
2.03
39
11.99
7.05
1.25
37.36
14.87
5.20
9.68
1.24
6.37
0.86
40
12.09
11.68
1.01
33.12
11.54
6.37
5.17
7.71
15.97
0.02
41
12.17
11.46
1.05
39.92
10.62
0.68
9.94
2.51
1.53
1.73
42
12.33
8.18
0.85
28.95
20.53
8.55
11.98
4.13
5.67
1.92
43
12.60
12.90
1.23
39.71
4.26
0.96
3.30
1.04
0.92
1.84
44
12.65
14.28
1.24
41.86
6.66
0.69
5.97
2.04
0.95
1.96
45
12.69
11.69
1.35
45.64
6.24
0.96
5.28
1.83
1.17
2.20
46
12.85
16.87
0.99
36.07
6.76
3.14
3.62
3.89
9.89
1.14
47
12.88
11.43
1.47
54.28
76.11
50.27
25.84
10.00
14.20
1.03
48
13.03
8.34
0.97
33.32
35.32
10.88
24.44
4.30
4.31
1.52
49
13.08
9.52
1.07
38.04
11.44
10.08
1.36
4.31
7.37
1.47
50
13.35
17.63
34.87
1.20
0.88
0.32
1.30
3.16
3.32
51
13.40
14.89
1.26
42.55
0.22
0.20
0.02
0.46
0.82
0.53
52
13.54
9.90
1.25
31.89
7.72
6.12
1.60
1.63
3.48
0.58
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 31
C. A. Quesada et al.: Variations in soil chemical and physical properties
83
Table A1.
Continued.
Soil
[C]
CN
ρ
R
C
Fe
d
Fe
o
Fe
d
–Fe
o
Al
d
Al
o
Al
p
(mg g
−
1
)
(kg dm
−
3
)
(Mg ha
−
1
)
g kg
−
1
53
13.65
8.58
1.29
45.24
20.01
1.84
18.17
3.69
1.66
2.60
54
13.73
8.55
0.85
31.36
20.71
15.97
4.74
5.62
8.85
1.84
55
14.18
10.38
1.23
46.31
13.81
8.50
5.31
6.26
8.99
2.02
56
14.23
8.03
1.14
41.99
15.87
8.30
7.57
3.83
6.90
1.22
57
14.25
13.84
1.15
43.24
7.47
3.02
4.45
4.18
14.53
3.00
58
14.40
11.08
0.92
44.70
32.60
2.53
30.07
4.76
1.54
2.63
59
14.41
9.86
1.36
50.54
26.80
12.77
14.03
2.82
3.72
3.62
60
14.46
6.80
1.32
40.82
14.49
9.93
4.56
1.34
1.12
0.41
61
14.87
11.62
0.76
31.39
5.19
2.74
2.45
1.48
1.20
0.92
62
14.87
14.66
1.12
46.06
5.25
0.57
4.68
1.72
1.04
2.99
63
14.93
12.63
1.40
63.47
11.82
1.41
10.41
3.38
2.10
2.57
64
15.11
9.37
1.09
43.71
13.93
10.64
3.29
3.23
8.43
1.87
65
15.11
15.77
1.14
43.09
3.08
1.94
1.14
2.35
3.91
1.71
66
15.40
12.83
0.92
40.55
28.85
2.06
26.79
4.32
1.48
3.21
67
15.44
16.08
1.10
41.04
4.20
0.15
4.05
0.36
0.48
0.76
68
15.65
12.18
1.15
49.26
10.23
2.88
7.35
3.13
3.64
2.95
69
15.68
8.91
1.15
40.69
17.57
13.13
4.44
3.50
6.65
1.45
70
15.89
9.35
0.91
37.79
32.32
19.93
12.38
7.68
12.92
0.67
71
15.92
14.96
0.90
43.15
44.70
2.36
42.34
4.96
3.16
4.09
72
15.97
11.81
1.36
57.74
12.00
0.90
11.10
2.62
1.91
2.43
73
16.01
7.96
1.28
52.90
17.77
8.16
9.61
4.55
7.43
2.35
74
16.06
9.16
1.00
38.89
14.73
2.17
12.56
2.85
1.69
1.92
75
16.16
31.81
0.99
28.95
0.72
0.08
0.64
0.21
0.26
0.10
76
16.25
13.15
1.07
46.40
10.50
1.17
9.33
2.75
1.44
2.37
77
16.40
13.67
0.98
44.21
18.34
5.36
12.98
5.33
11.12
0.96
78
16.40
9.79
1.07
45.01
16.24
11.59
4.65
4.68
7.32
1.80
79
16.79
6.98
1.08
41.36
22.14
5.90
16.24
2.95
2.86
1.55
80
16.79
13.15
1.13
51.93
15.72
1.20
14.52
3.47
1.70
2.42
81
16.85
6.78
1.41
52.47
16.55
11.13
5.42
1.50
0.86
0.50
82
17.02
15.00
0.97
43.39
3.50
1.10
2.40
1.98
2.33
2.01
83
17.11
12.70
1.15
66.72
7.73
7.42
0.31
2.90
5.58
1.74
84
17.20
14.33
1.07
46.18
21.45
1.45
20.00
2.12
1.11
2.51
85
17.32
11.65
1.02
41.95
86
17.35
10.77
0.89
43.74
7.23
5.37
1.85
3.11
4.57
1.25
87
17.40
9.20
1.01
44.51
22.17
7.42
14.74
5.49
8.64
2.01
88
17.84
10.62
0.87
41.30
22.57
10.48
12.08
5.37
9.32
2.28
89
17.93
11.96
0.92
70.74
7.07
1.92
5.15
2.18
1.24
2.04
90
18.02
10.14
1.06
54.78
9.63
3.94
5.69
1.54
1.85
1.23
91
18.16
7.49
0.90
38.83
18.45
13.74
4.71
6.13
12.86
1.81
92
18.35
7.58
1.37
55.53
23.89
21.99
1.89
4.25
8.34
0.61
93
18.40
17.36
1.22
64.33
2.11
2.11
3.88
2.25
94
18.48
10.80
1.29
69.52
95
18.84
16.82
0.92
48.09
9.15
2.24
6.91
2.13
1.60
1.87
96
18.97
10.83
0.71
40.47
15.87
1.73
14.14
3.62
2.06
12.22
97
19.80
11.65
1.27
67.10
15.55
1.58
13.97
2.54
1.88
1.26
98
20.05
12.23
0.92
44.10
6.89
2.83
4.06
2.04
1.52
1.22
99
20.10
11.82
1.23
71.13
22.00
2.41
19.59
1.83
0.84
1.23
100
20.49
18.68
1.14
63.59
101
20.87
10.06
0.98
51.94
12.72
6.46
6.26
5.62
9.49
2.56
102
21.01
10.72
0.96
52.89
14.70
2.13
12.57
3.60
2.07
3.68
103
21.40
12.49
0.92
40.49
12.63
1.41
11.22
3.50
1.24
2.65
104
21.46
8.82
1.27
57.95
37.53
5.34
32.19
4.70
3.25
3.72
105
21.53
9.82
0.96
51.38
16.61
14.91
1.70
6.88
13.92
1.65
106
21.68
13.35
0.98
60.36
6.95
2.65
4.30
3.39
7.61
1.61
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 32
84
C. A. Quesada et al.: Variations in soil chemical and physical properties
Table A1.
Continued.
Soil
[C]
CN
ρ
R
C
Fe
d
Fe
o
Fe
d
–Fe
o
Al
d
Al
o
Al
p
(mg g
−
1
)
(kg dm
−
3
)
(Mg ha
−
1
)
g kg
−
1
107
21.76
11.69
0.96
53.12
14.82
1.65
13.17
3.89
2.05
4.19
108
21.85
13.71
0.82
38.35
16.61
15.48
1.13
4.51
10.66
1.79
109
21.90
16.85
0.97
46.77
1.20
0.90
0.30
0.44
0.80
0.76
110
21.99
13.83
48.94
16.75
3.54
13.21
3.07
1.36
2.30
111
22.70
11.65
0.89
52.62
7.70
2.98
4.72
2.45
1.98
1.49
112
22.73
13.15
0.99
63.55
19.64
10.34
9.30
9.47
37.03
1.85
113
22.77
6.82
1.60
80.81
17.42
11.91
5.51
1.48
0.88
0.40
114
22.83
10.88
1.27
69.23
10.57
8.53
2.04
1.86
4.45
0.68
115
23.00
15.33
0.93
58.49
11.41
2.31
9.10
2.83
1.77
1.22
116
23.09
9.07
1.15
78.66
23.52
7.08
16.44
1.66
1.45
0.24
117
23.20
13.47
0.91
56.77
9.02
2.59
6.43
2.68
1.83
1.52
118
23.21
12.93
0.98
60.79
26.40
2.12
24.28
5.73
1.75
4.32
119
23.34
12.53
0.94
55.58
13.50
1.04
12.46
3.90
2.77
3.26
120
23.53
11.93
1.09
57.84
19.62
0.87
18.75
4.97
2.55
3.10
121
23.65
12.24
0.97
59.31
9.75
2.89
6.86
3.02
1.89
2.17
122
24.03
10.83
1.05
60.18
12.33
4.92
7.41
5.15
7.97
2.84
123
24.30
22.03
1.34
3.12
0.60
0.41
0.19
0.09
0.02
0.03
124
24.30
11.05
1.27
74.80
16.70
1.36
15.34
0.95
1.01
0.59
125
24.76
9.49
1.05
68.86
21.66
6.28
15.38
1.77
1.44
0.66
126
25.39
15.15
0.77
52.49
14.82
1.09
13.73
3.28
2.26
1.60
127
25.48
16.20
0.86
67.13
21.55
2.85
18.70
3.88
2.25
4.09
128
25.57
11.35
1.13
77.51
36.21
1.60
34.61
7.66
3.27
2.61
129
25.82
10.79
0.94
64.92
58.14
2.19
55.95
9.61
2.77
2.50
130
25.87
17.21
1.02
70.11
8.44
3.55
4.89
4.71
11.92
1.26
131
26.57
12.57
0.89
61.29
9.71
3.12
6.59
3.02
1.97
2.37
132
26.86
9.86
1.03
76.89
53.64
2.19
51.45
9.60
1.99
1.69
133
27.00
11.82
0.82
55.83
8.72
3.58
5.14
3.32
2.49
2.57
134
27.09
11.56
0.93
64.68
7.71
2.64
5.07
2.51
1.82
1.85
135
28.80
9.05
0.79
45.16
10.39
9.63
0.76
3.83
8.40
2.47
136
30.80
15.40
1.18
95.49
67.20
2.03
65.17
5.88
1.24
1.72
137
30.82
10.75
0.88
81.36
21.34
12.14
9.20
7.99
31.37
2.36
138
32.80
14.26
0.88
74.97
3.70
1.41
2.29
2.49
4.70
2.79
139
41.81
20.72
0.63
53.70
0.24
0.08
0.16
0.10
0.21
0.01
140
46.70
16.10
1.36
115.48
21.40
3.17
18.23
3.74
3.14
2.53
141
49.08
13.36
1.32
166.86
20.10
2.87
17.23
10.49
2.55
1.81
142
60.47
11.31
14.50
20.27
11.69
143
61.44
11.77
0.60
97.92
9.02
8.34
0.67
19.53
43.52
10.26
144
63.43
12.51
0.43
87.72
11.14
5.25
5.89
7.36
22.54
8.61
145
89.26
25.82
1.58
363.55
0.36
0.34
0.02
1.34
2.75
1.18
146
93.06
12.50
0.89
219.25
147
119.82
20.79
0.34
42.19
0.90
0.10
0.80
0.45
0.54
0.27
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 33
C. A. Quesada et al.: Variations in soil chemical and physical properties
85
Data availability.
All data used in the figures and analysis here
are made available in the tables, with the exception of the above-
ground wood productivity and above-ground biomass data (Ap-
pendix Fig. A1), which are available in Brienen et al. (2015).
Author contributions.
CAQ and EOM conducted field sampling.
CAQ, CP and GS performed laboratory analysis. OLP and JL sup-
ported the study and the maintenance of the RAINFOR network.
All authors reviewed and commented on the manuscript. CAQ and
JL performed data analysis and wrote the manuscript.
Competing interests.
The authors declare that they have no con-
flict of interest.
Acknowledgements.
This paper is a product of the RAINFOR
network. It integrates the effort of several researchers, technicians
and field assistants across Amazonia. We thank the following indi-
viduals in particular: Michael Schwarz, Gabriel Batista de Oliveira
Borges, Claudia Czimczk, Jens Schmerler, Alexandre J. B. Santos,
Gareth Lloyd, Timothy R. Baker, Sandra Patiño, Yadvinder Malhi,
Jonas O. Moraes Filho, Orlando F. C. Junior, José Edivaldo Chaves
and Raimundo Nonato de Araújo Filho.
Financial support.
Support for RAINFOR has come princi-
pally from the Natural Environment Research Council (grant
nos. NE/D005590/1 (“TROBIT”), NE/B503384/1, NE/D01025X/1,
NE/I02982X/1, NE/F005806/1 and NE/I028122/1), the Gordon and
Betty Moore Foundation (grant no. 1656), the European Research
Council (advanced grant no. 291585, “T-FORCES”) and the EU
Seventh Framework Programme (grant nos. GEOCARBON283080,
AMAZALERT-282664). Jon Lloyd was supported by a Royal So-
ciety of London Research Merit Award and a São Paulo Excellence
Chair.
Review statement.
This paper was edited by Elizabeth Bach and
reviewed by two anonymous referees.
References
Baldock, J. A. and Skjemstad, J. O.: Role of the soil matrix and
minerals in protecting natural organic materials against biologi-
cal attack, Org. Geochem., 31, 697–710, 2000.
Bartko, J. J. and Pettigrew, K. D.: The Teacher’s Corner: A Note on
the Correlation of Parts with Wholes, Am. Stat., 22, p. 41, 1968.
Barto´n, K.: MuMIn: Multi-model inference, R package version 1.9.
13, The Comprehensive R Archive Network (CRAN), Vienna,
Austria, 1–76, 2013.
Basile-Doelsch, I., Amundson, R., Stone, W. E. E., Masiello, C. A.,
Bottero, J. Y., Colin, F., Masin, F., Borschneck, D., and Meu-
nier, J. D.: Mineralogical control of organic carbon dynamics in
a volcanic ash soil on La Réunion, Eur. J. Soil Sci., 56, 689–703,
2005.
Batjes, N. H.: Total carbon and nitrogen in the soils of the world,
Eur. J. Soil Sci., 47, 151–163, 1996.
Batjes, N. H. and Dijkshoorn, J. A.: Carbon and nitrogen stocks in
the soils of the Amazon Region, Geoderma, 89, 273–286, 1999.
Bird, M. I., Veenendaal, E. M., and Lloyd, J. J.: Soil carbon in-
ventories and
δ
13
C along a moisture gradient in Botswana,
Glob. Change Biol., 10, 342–349, https://doi.org/10.1046/j.1365-
2486.2003.00695.x, 2004.
Bish, D. L. and Reynolds, R. C.: Sample preparation for X-ray
diffraction, Rev. Mineral. Geochem., 20, 73–99, 1989.
Boudot, J. P., Hadj, A. B., Steiman, R., and Seigle-Murandi, F.:
Biodegradation of synthetic organo-metallic complexes of iron
and aluminium with selected metal to carbon ratios, Soil Biol.
Biochem., 21, 961–966, 1989.
Bradford, M. A., Davies, C. A., Frey, S. D., Maddox, T. R., Melillo,
J. M., Mohan, J. E., Reynolds, J. F., Treseder, K. K., and Wallen-
stein, M. D.: Thermal adaptation of soil microbial respiration to
elevated temperature, Ecol. Lett., 11, 1316–1327, 2008.
Brienen, R. J. W., Phillips, O. L., Feldpausch, T. R., Gloor,
E., Baker, T. R., Lloyd, J., Lopez-Gonzalez, G., Monteagudo-
Mendoza, A., Malhi, Y., Lewis, S. L., Vasquez Martinez, R.,
Alexiades, M., Alvarez Davila, E., Alvarez-Loayza, P., Andrade,
A., Aragao, L. E. O. C., Araujo-Murakami, A., Arets, E. J. M.
M., Arroyo, L., Aymard C, G. A., Banki, O. S., Baraloto, C., Bar-
roso, J., Bonal, D., Boot, R. G. A., Camargo, J. L. C., Castilho, C.
V., Chama, V., Chao, K. J., Chave, J., Comiskey, J. A., Cornejo
Valverde, F., da Costa, L., de Oliveira, E. A., Di Fiore, A., Er-
win, T. L., Fauset, S., Forsthofer, M., Galbraith, D. R., Grahame,
E. S., Groot, N., Herault, B., Higuchi, N., Honorio Coronado, E.
N., Keeling, H., Killeen, T. J., Laurance, W. F., Laurance, S., Li-
cona, J., Magnussen, W. E., Marimon, B. S., Marimon-Junior, B.
H., Mendoza, C., Neill, D. A., Nogueira, E. M., Nunez, P., Pal-
lqui Camacho, N. C., Parada, A., Pardo-Molina, G., Peacock, J.,
Pena-Claros, M., Pickavance, G. C., Pitman, N. C. A., Poorter,
L., Prieto, A., Quesada, C. A., Ramirez, F., Ramirez-Angulo, H.,
Restrepo, Z., Roopsind, A., Rudas, A., Salomao, R. P., Schwarz,
M., Silva, N., Silva-Espejo, J. E., Silveira, M., Stropp, J., Tal-
bot, J., ter Steege, H., Teran-Aguilar, J., Terborgh, J., Thomas-
Caesar, R., Toledo, M., Torello-Raventos, M., Umetsu, R. K., van
der Heijden, G. M. F., van der Hout, P., Guimaraes Vieira, I. C.,
Vieira, S. A., Vilanova, E., Vos, V. A., and Zagt, R. J.: Long-term
decline of the Amazon carbon sink, Nature, 519, 344–348, 2015.
Bruun, T. B., Elberling, B., and Christensen, B. T.: Lability of soil
organic carbon in tropical soils with different clay minerals, Soil
Biol. Biochem., 42, 888–895, 2010.
Buol, S. W., Southard, R. J., Graham, R. C., and McDaniel, P.
A.: Soil genesis and classification, John Wiley & Sons, 543 pp.,
2011.
Burke, I. C., Yonker, C. M., Parton, W. J., Cole, C. V., Schimel, D.
S., and Flach, K.: Texture, climate, and cultivation effects on soil
organic matter content in US grassland soils, Soil Sci. Soc. Am.
J., 53, 800–805, 1989.
Chauvel,
A.,
Lucas,
Y.,
and
Boulet,
R.:
On
the
gene-
sis of the soil mantle of the region of Manaus, Cen-
tral Amazonia, Brazil, Cell. Mol. Life Sci., 43, 234–241,
https://doi.org/10.1007/bf01945546, 1987.
Chayes, F.: Ratio correlation: a manual for students of petrology
and geochemistry, University of Chicago Press, 99 pp., 1971.
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 34
86
C. A. Quesada et al.: Variations in soil chemical and physical properties
Coley, P. D., Bryant, J. P., and Chapin, F. S. I.: Resource avail-
ability and plant antiherbivore defense, Science, 230, 895–899,
https://doi.org/10.1126/science.230.4728.895, 1985.
Cornell, R. M. and Schwertmann, U.: The iron oxides: structure,
properties, reactions, occurrences and uses, John Wiley & Sons,
137 pp., 2003.
Cotrufo, M. F., Wallenstein, M. D., Boot, C. M., Denef, K., and
Paul, E.: The M icrobial E fficiency-M atrix S tabilization
(MEMS) framework integrates plant litter decomposition with
soil organic matter stabilization: do labile plant inputs form sta-
ble soil organic matter?, Glob. Change Biol., 19, 988–995, 2013.
Delvaux, B., Herbillon, A. J., and Vielvoye, L.: Characterization
of a weathering sequence of soils derived from volcanic ash
in Cameroon: Taxonomic, mineralogical and agronomic impli-
cations, Geoderma, 45, 375–388, https://doi.org/10.1016/0016-
7061(89)90017-7, 1989.
de Mendiburu, F.: Statistical procedures for agricultural re-
search. Package “Agricolae”, version 1.4-4, Comprehensive
R Archive Network, Institute for Statistics and Mathemat-
ics, Vienna, Austria, available at: http://cran.r-project.org/web/
packages/agricolae/agricolae.pdf (last access: 15 May 2019),
2013.
Denef, K., Plante, A. F., and Six, J.: Characterization of soil or-
ganic matter, in: Soil carbon dynamics: An integrated methodol-
ogy, Cambridge University Press, 91–126, 2010.
Deng, Q., Hui, D., Zhang, D., Zhou, G., Liu, J., Liu, S., Chu, G., and
Li, J.: Effects of precipitation increase on soil respiration: a three-
year field experiment in subtropical forests in China, PLoS One,
7, e41493, https://doi.org/10.1371/journal.pone.0041493, 2012.
Dick, D. P., Gonçalves, C. N., Dalmolin, R. S., Knicker, H., Klamt,
E., Kögel-Knabner, I., Simões, M. L., and Martin-Neto, L.: Char-
acteristics of soil organic matter of different Brazilian Ferralsols
under native vegetation as a function of soil depth, Geoderma,
124, 319–333, 2005.
Do Nascimento, N. R., Bueno, G. T., Fritsch, E., Herbillon, A.
J., Allard, T., Melfi, A. J., Astolfo, R., Boucher, H., and Li,
Y.: Podzolization as a deferralitization process: a study of an
Acrisol–Podzol sequence derived from Palaeozoic sandstones in
the northern upper Amazon Basin, Eur. J. Soil Sci., 55, 523–538,
2004.
Driessen, P., Deckers, J., Spaargaren, O., and Nachtergaele, F.: Lec-
ture notes on the major soils of the world, Food and Agriculture
Organization (FAO), 334 pp., 2000.
Eusterhues, K., Rumpel, C., and Kögel-Knabner, I.: Stabilization
of soil organic matter isolated via oxidative degradation, Org.
Geochem., 36, 1567–1575, 2005.
Feller, C. and Beare, M. H.: Physical control of soil organic matter
dynamics in the tropics, Geoderma, 79, 69–116, 1997.
Fine, P. V., Mesones, I., and Coley, P. D.: Herbivores promote habi-
tat specialization by trees in Amazonian forests, Science, 305,
663–665, 2004.
Fittkau, E. J.: Esboco de uma divisao ecologica da regiao Amazon-
ica, edited by: Idrobo, I. M., 363–372, 1971.
Gee, G. W. and Bauder, J. W.: Particle-size analysis, in: Methods
in Soil Analysis, Part 1, Physical and Mineralogical Methods,
edited by: Klute, A., American Society of Agronomy and Soil
Science Society of America, Madison, Wisconsin, USA, 383–
409, 1986.
Giraudoux, P.: pgirmess: data analysis in ecology, R package ver-
sion 1.5. 8, R Foundation for Statistical Computing Vienna, Aus-
tria, 64 pp., 2013.
Glinski, J. and Lipiec, J.: Soil physical conditions and plant roots.
CRC press. Boca Raton, FL, 250 pp., 1990.
Hartemink, A. E. and Huting, J.: Land cover, extent, and properties
of Arenosols in Southern Africa, Arid Land Res. Manag., 22,
134–147, 2008.
Herrera, R., Jordan, C. F., Klinge, H., and Medina, E.: Amazon
ecosystems: Their structure and functioning with particular em-
phasis on nutrients, Interciencia, 3, 223–232, 1978.
Hoorn, C. and Wesselingh, F.: Amazonia: landscape and species
evolution: a look into the past, John Wiley & Sons, 447 pp., 2011.
Hoorn, C., Wesselingh, F. P., Ter Steege, H., Bermudez, M. A.,
Mora, A., Sevink, J., Sanmartín, I., Sanchez-Meseguer, A., An-
derson, C. L., and Figueiredo, J. P.: Amazonia through time: An-
dean uplift, climate change, landscape evolution, and biodiver-
sity, Science, 330, 927–931, 2010.
Irion, G.: Soil infertility in the Amazonian rain forest, Naturwis-
senschaften, 65, 515–519, https://doi.org/10.1007/BF00439791,
1978.
IUSS: (International Union of Soil Science) Working Group WRB:
70 World Reference Base for Soil Resources 2014, International
Soil Classification System For Naming Soils And Creating Leg-
ends For Soil Maps, Rome, 2014.
Jahn, R., Blume, H.-P., Asio, V. B., Spaargaren, O., and Schad, P.:
Guidelines for soil description, FAO, Rome, 97 pp., 2006.
Kahle, M., Kleber, M., Torn, M. S., and Jahn, R.: Carbon storage in
coarse and fine clay fractions of illitic soils, Soil Sci. Soc. Am.
J., 67, 1732–1739, 2003.
Kahle, M., Kleber, M., and Jahn, R.: Retention of dissolved organic
matter by phyllosilicate and soil clay fractions in relation to min-
eral properties, Org. Geochem., 35, 269–276, 2004.
Kaiser,
K.
and
Guggenberger,
G.:
Mineral
surfaces
and
soil
organic
matter,
Eur.
J.
Soil
Sci.,
54,
219–236,
https://doi.org/10.1046/j.1365-2389.2003.00544.x, 2003.
Kaiser, K. and Zech, W.: Sorption of dissolved organic nitrogen by
acid subsoil horizons and individual mineral phases, Eur. J. Soil
Sci., 51, 403–411, 2000.
Kaiser, K., Mikutta, R., and Guggenberger, G.: Increased stability
of organic matter sorbed to ferrihydrite and goethite on aging,
Soil Sci. Soc. Am. J., 71, 711–719, 2007.
Kalbitz, K. and Kaiser, K.: Contribution of dissolved organic matter
to carbon storage in forest mineral soils, J. Plant Nutr. Soil Sci.,
171, 52–60, 2008.
Kitagawa, Y.: Goethite and hematite in some solls from the amazon
region, Soil Sci. Plant Nutr., 29, 209–217, 1983.
Kleber, M., Mikutta, R., Torn, M. S., and Jahn, R.: Poorly crystalline
mineral phases protect organic matter in acid subsoil horizons,
Eur. J. Soil Sci., 56, 717–725, 2005.
Kleber, M., Eusterhues, K., Keiluweit, M., Mikutta, C., Mikutta, R.,
and Nico, P. S.: Mineral–organic associations: formation, prop-
erties, and relevance in soil environments, in Advances in agron-
omy, Elsevier, 130, 1–140, 2015.
Lloyd, J. and Farquhar, G. D.: Effects of rising temperatures and
[CO
2
] on the physiology of tropical forest trees, Philos. T. R.
Soc. B, 363, 1811–1817, https://doi.org/10.1098/rstb.2007.0032,
2008.
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Page 35
C. A. Quesada et al.: Variations in soil chemical and physical properties
87
Lloyd, J., Bloomfield, K., Domingues, T. F., and Farquhar, G. D.:
Photosynthetically relevant foliar traits correlating better on a
mass vs. an area basis: of ecophysiological relevance or just a
case of mathematical imperatives and statistical quicksand?, New
Phytol., 199, 311–321, 2013.
Luizão, F. J. and Schubart, H. O. R.: Litter production and decom-
position in a terra-firme forest of Central Amazonia, Experientia,
43, 259–265, 1987.
Lynch, J. M. and Bragg, E.: Microorganisms and soil aggregate sta-
bility, in: Advances in soil science, Springer, 133–171, 1985.
Marques, J. J., Teixeira, W. G., Schulze, D. G., and Curi, N.: Min-
eralogy of soils with unusually high exchangeable Al from the
western Amazon Region, Clay Miner., 37, 651–661, 2002.
Mikutta, R., Kleber, M., and Jahn, R.: Poorly crystalline miner-
als protect organic carbon in clay subfractions from acid subsoil
horizons, Geoderma, 128, 106–115, 2005.
Mikutta, R., Mikutta, C., Kalbitz, K., Scheel, T., Kaiser, K., and
Jahn, R.: Biodegradation of forest floor organic matter bound
to minerals via different binding mechanisms, Geochim. Cos-
mochim. Ac., 71, 2569–2590, 2007.
Nelson, D. W. and Sommers, L. E.: Total carbon and total nitrogen,
in: Methods of Soil Analysis: Part 3 – Chemical Methods, edited
by: Sparks, D. L., American Society of Agronomy/Soil Science
Society of America, Madison, WI, 961–1010, 1996.
Nierop, K. G., Jansen, B., and Verstraten, J. M.: Dissolved organic
matter, aluminium and iron interactions: precipitation induced by
metal/carbon ratio, pH and competition, Sci. Total Environ., 300,
201–211, 2002.
Oades, J. M.: The role of biology in the formation, stabilization
and degradation of soil structure, in: Soil Structure/Soil Biota
Interrelationships, Elsevier, 377–400, 1993.
Oades,
J.
M.:
An
Introduction
to
Organic
Matter
in
Mineral
Soils,
in:
Minerals
in
Soil
Environments,
edited
by:
Dixon,
J.
B.
and
Weed,
S.
B.,
89–159,
https://doi.org/10.2136/sssabookser1.2ed.c3, 2018.
Parfitt, R. L. and Childs, C. W.: Estimation of forms of Fe and Al-a
review, and analysis of contrasting soils by dissolution and Moss-
bauer methods, Soil Res., 26, 121–144, 1988.
Parfitt, R. L., Theng, B. K. G., Whitton, J. S., and Shepherd, T. G.:
Effects of clay minerals and land use on organic matter pools,
Geoderma, 75, 1–12, 1997.
Paul, S., Flessa, H., Veldkamp, E., and López-Ulloa, M.: Stabiliza-
tion of recent soil carbon in the humid tropics following land use
changes: evidence from aggregate fractionation and stable iso-
tope analyses, Biogeochemistry, 87, 247–263, 2008.
Paz, C. P.: Distribuição das frações do carbono orgânico nos so-
los de florestas maduras na bacia Amazônica: o papel das pro-
priedades do solo, da qualidade da liteira e do clima, MSc thesis,
Instituto Nacional de Pesquisas da Amazôˆnia, Brazil, 62 pp.,
2011.
Pella, E.: Elemental organic analysis, Part 2, State of the art, Am.
Lab., 22, 28–32, 1990.
Percival, H. J., Parfitt, R. L., and Scott, N. A.: Factors controlling
soil carbon levels in New Zealand grasslands is clay content im-
portant?, Soil Sci. Soc. Am. J., 64, 1623–1630, 2000.
Pleysier, J. L. and Juo, A. S. R.: A single-extraction method us-
ing silver-thiourea for measuring exchangeable cations and effec-
tive CEC in soils with variable charges, Soil Sci., 129, 205–211,
1980.
Quesada, C. A. and Lloyd, J.: Soil–Vegetation Interactions in Ama-
zonia, in: Interactions Between Biosphere, Atmosphere and Hu-
man Land Use in the Amazon Basin, edited by: Nagy, L., Fors-
berg, B. R., and Artaxo, P., Springer Berlin Heidelberg, Berlin,
Heidelberg, 267–299, 2016.
Quesada, C. A., Lloyd, J., Schwarz, M., Patiño, S., Baker, T. R., Cz-
imczik, C., Fyllas, N. M., Martinelli, L., Nardoto, G. B., Schmer-
ler, J., Santos, A. J. B., Hodnett, M. G., Herrera, R., Luizão, F.
J., Arneth, A., Lloyd, G., Dezzeo, N., Hilke, I., Kuhlmann, I.,
Raessler, M., Brand, W. A., Geilmann, H., Moraes Filho, J. O.,
Carvalho, F. P., Araujo Filho, R. N., Chaves, J. E., Cruz Junior, O.
F., Pimentel, T. P., and Paiva, R.: Variations in chemical and phys-
ical properties of Amazon forest soils in relation to their gene-
sis, Biogeosciences, 7, 1515–1541, https://doi.org/10.5194/bg-7-
1515-2010, 2010.
Quesada, C. A., Lloyd, J., Anderson, L. O., Fyllas, N. M., Schwarz,
M., and Czimczik, C. I.: Soils of Amazonia with particular ref-
erence to the RAINFOR sites, Biogeosciences, 8, 1415–1440,
https://doi.org/10.5194/bg-8-1415-2011, 2011.
Quesada, C. A., Phillips, O. L., Schwarz, M., Czimczik, C. I., Baker,
T. R., Patiño, S., Fyllas, N. M., Hodnett, M. G., Herrera, R.,
Almeida, S., Alvarez Dávila, E., Arneth, A., Arroyo, L., Chao,
K. J., Dezzeo, N., Erwin, T., di Fiore, A., Higuchi, N., Hono-
rio Coronado, E., Jimenez, E. M., Killeen, T. J., Lezama, A. T.,
Lloyd, G., López-González, G., Luizão, F. J., Malhi, Y., Mon-
teagudo, A., Neill, D. A., Núñez Vargas, P., Paiva, R., Peacock,
J., Peñuela, M. C., Peña Cruz, A., Pitman, N., Priante Filho, N.,
Prieto, A., Ramírez, H., Rudas, A., Salomão, R., Santos, A. J.
B., Schmerler, J., Silva, N., Silveira, M., Vásquez, R., Vieira,
I., Terborgh, J., and Lloyd, J.: Basin-wide variations in Amazon
forest structure and function are mediated by both soils and cli-
mate, Biogeosciences, 9, 2203–2246, https://doi.org/10.5194/bg-
9-2203-2012, 2012.
Quesada, C. A. N.: Soil vegetation interactions across Amazonia,
University of Leeds (School of Geography), 236 pp., 2008.
R Development Core Team: R: A Language and Environment for
Statistical Computing, edited by: Austria, R., Foundation for Sta-
tistical Computing Vienna Austria, 0(01/19), ISBN 3-900051-
07-0, 2012.
Richter, D. D. and Babbar, L. I.: Soil diversity in the tropics, Else-
vier, Adv. Ecol. Res., 21, 315–389, 1991.
Saggar, S., Parshotam, A., Sparling, G. P., Feltham, C. W., and
Hart, P. B. S.:
14
C-labelled ryegrass turnover and residence
times in soils varying in clay content and mineralogy, Soil Biol.
Biochem., 28, 1677–1686, 1996.
Saggar, S., Parshotam, A., Hedley, C., and Salt, G.:
14
C-labelled
glucose turnover in New Zealand soils, Soil Biol. Biochem., 31,
2025–2037, 1999.
Saidy, A. R., Smernik, R. J., Baldock, J. A., Kaiser, K., Sander-
man, J., and Macdonald, L. M.: Effects of clay mineralogy and
hydrous iron oxides on labile organic carbon stabilisation, Geo-
derma, 173, 104–110, 2012.
Saiz, G., Bird, M. I., Domingues, T. F., Schrodt, F., Schwarz, M.,
Feldpausch, T. R., Veenendaal, E. M., Djagbletey, G., Hien, F.,
Compaore, H., Diallo, A., and Lloyd, J.: Variation in soil car-
bon stocks and their determinants across a precipitation gra-
dient in West Africa, Glob. Change Biol., 18, 1670–1683,
https://doi.org/10.1111/j.1365-2486.2012.02657.x, 2012.
www.soil-journal.net/6/53/2020/
SOIL, 6, 53–88, 2020
Page 36
88
C. A. Quesada et al.: Variations in soil chemical and physical properties
Saiz, G., Bird, M., Wurster, C., Quesada, C. A., Ascough, P.,
Domingues, T., Schrodt, F., Schwarz, M., Feldpausch, T. R.,
Veenendaal, E., Djagbletey, G., Jacobsen, G., Hien, F., Com-
paore, H., Diallo, A., and Lloyd, J.: The influence of C
3
and C
4
vegetation on soil organic matter dynamics in contrasting semi-
natural tropical ecosystems, Biogeosciences, 12, 5041–5059,
https://doi.org/10.5194/bg-12-5041-2015, 2015.
Sanchez, P. A.: Properties and Management of Soils in the Tropics,
Wiley, New York, 618 pp., 1976.
Scheel, T., Dörfler, C., and Kalbitz, K.: Precipitation of dissolved or-
ganic matter by aluminum stabilizes carbon in acidic forest soils,
Soil Sci. Soc. Am. J., 71, 64–74, 2007.
Scheel, T., Haumaier, L., Ellerbrock, R. H., Rühlmann, J., and Kalb-
itz, K.: Properties of organic matter precipitated from acidic for-
est soil solutions, Org. Geochem., 39, 1439–1453, 2008.
Schrumpf, M., Kaiser, K., Guggenberger, G., Persson, T., Kögel-
Knabner, I., and Schulze, E.-D.: Storage and stability of or-
ganic carbon in soils as related to depth, occlusion within ag-
gregates, and attachment to minerals, Biogeosciences, 10, 1675–
1691, https://doi.org/10.5194/bg-10-1675-2013, 2013.
Schwertmann, U., Wagner, F., and Knicker, H.: Ferrihydrite–humic
associations, Soil Sci. Soc. Am. J., 69, 1009–1015, 2005.
Siegel, S. and Castellan Jr., N.: Nonparametric statistics for the
behavioural sciences, 2nd Edn., McGraw-Hill, Boston, 399 pp.,
1998.
Sioli, H.: The Amazon and its main affluents: hydrography, mor-
phology of the river courses, and river types, in: The Amazon,
Springer, 127–165, 1984.
Six, J., Bossuyt, H., Degryze, S., and Denef, K.: A history of re-
search on the link between (micro) aggregates, soil biota, and
soil organic matter dynamics, Soil Till. Res., 79, 7–31, 2004.
Sollins, P., Homann, P., and Caldwell, B. A.: Stabilization and
destabilization of soil organic matter: mechanisms and controls,
Geoderma, 74, 65–105, 1996.
Sombroek, W. G.: A Reconnaissance of the Soils of the Brazilian
Amazon Region, Centre for Agricultural Publications and Docu-
mentation, Wageningen, 292 pp., 1966.
Sombroek, W. G.: Soils of the Amazon region, in The Amazon,
Springer, 521–535, 1984.
Sombroek, W. G.: Amazon landforms and soils in relation to bio-
logical diversity, Acta Amazon., 30, 81–100, 2000.
Telles, E. de C. C., de Camargo, P. B., Martinelli, L. A., Trumbore,
S. E., da Costa, E. S., Santos, J., Higuchi, N., and Oliveira Jr., R.
C.: Influence of soil texture on carbon dynamics and storage po-
tential in tropical forest soils of Amazonia, Global Biogeochem.
Cy., 17, 1–12, 2003.
Tiessen, H. and Moir, J. O.: Total and Organic Carbon, in: Soil Sam-
pling and Methods of Analysis, edited by: Carter, M. R., 187–
199, Lewis Publishers, Boca Raton, FL, 1993.
Tisdall, J. M. and Oades, J.: Organic matter and water-stable aggre-
gates in soils, J. Soil Sci., 33, 141–163, 1982.
Trumbore, S. and Barbosa De Camargo, P.: Soil carbon dynamics,
Amazonia and Global Change, 186, 451–462, 2009.
Trumbore, S. E. and Zheng, S.: Comparison of fractionation meth-
ods for soil organic matter
14
C analysis, Radiocarbon, 38, 219–
229, 1996.
van den Boogaart, K. G. and Tolosana-Delgado, R.: “Composi-
tions”: a unified R package to analyze compositional data, Com-
put. Geosci., 34, 320–338, 2008.
van Lützow, M., Kögel-Knabner, I., Ekschmitt, K., Matzner, E.,
Guggenberger, G., Marschner, B., and Flessa, H.: Stabilization
of organic matter in temperate soils: mechanisms and their rele-
vance under different soil conditions – a review, Eur. J. Soil Sci.,
57, 426–445, 2006.
Van Reeuwijk, L. P.: Procedures for soil analysis, 6th Edn., Interna-
tional Soil Reference Information Centre, ISRIC, Wageningen,
The Netherlands, 2002.
Van Soest, P. J.: Use of detergents in the analysis of fibrous feeds,
2. A rapid method for the determination of fiber and lignin, J.
Assoc. Off. Agr. Chem., 46, 829–835, 1963.
Van Veen, J. A. and Kuikman, P. J.: Soil structural aspects of decom-
position of organic matter by micro-organisms, Biogeochemistry,
11, 213–233, 1990.
Wagai, R. and Mayer, L. M.: Sorptive stabilization of organic matter
in soils by hydrous iron oxides, Geochim. Cosmochim. Ac., 71,
25–35, 2007.
West, L. T., Beinroth, F. H., Sumner, M. E., and Kang, B. T.: Ul-
tisols: Characteristics and impacts on society, in: Advances in
Agronomy, Vol. 63, Elsevier, 179–236, 1997.
Wiseman, C. L. S. and Püttmann, W.: Interactions between mineral
phases in the preservation of soil organic matter, Geoderma, 134,
109–118, 2006.
Wurster, C. M., Saiz, G., Calder, A., and Bird, M. I.: Recovery of
organic matter from mineral-rich sediment and soils for stable
isotope analyses using static dense media, Rapid Commun. Mass
Sp., 24, 165–168, 2010.
Zimmermann, M., Leifeld, J., Schmidt, M. W. I., Smith, P., and
Fuhrer, J.: Measured soil organic matter fractions can be related
to pools in the RothC model, Eur. J. Soil Sci., 58, 658–667, 2007.
SOIL, 6, 53–88, 2020
www.soil-journal.net/6/53/2020/
Document Outline