Science of the Total Environment 712 (2020) 136405

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Can soil phosphorus availability in tropical forest systems be increased by

nitrogen-fixing leguminous trees?
Seldon Aleixo a,b, Antonio Carlos Gama-Rodrigues a,⁎, Emanuela Forestieri Gama-Rodrigues a,
Eduardo Francia Carneiro Campello c, Erika Caitano Silva a, Jan Schripsema b
a
Laboratório de Solos, Universidade Estadual do Norte Fluminense Darcy Ribeiro, Campos dos Goytacazes, RJ, Brazil
b
Grupo Metabolômica, Laboratório de Ciências Químicas, Universidade Estadual do Norte Fluminense Darcy Ribeiro, Campos dos Goytacazes, RJ, Brazil
c
Laboratório de Leguminosas Florestais, Embrapa Agrobiologia, Seropédica, RJ, Brazil

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• The leguminous trees increased all soil

P-fractions.
• 31P NMR spectroscopy identified and
quantified a large amount of Po and Pi
compounds.
• Leguminous trees and N fertilization
promote Po and Pi accumulation in top-
soil (0–10 cm).

a r t i c l e i n f o a b s t r a c t

Article history: Understanding the role of N-fixing leguminous trees for phosphorus (P) cycling in highly weathered tropical soils
Received 8 November 2019 is relevant for the conservation of natural forests as well as the sustainable management of agroforests and forest
Received in revised form 19 December 2019 plantations with low P input in the Brazilian Atlantic Forest region. We hypothesized that N-fixing leguminous
Accepted 27 December 2019
trees can increase the availability of soil P by exploiting different P sources without causing a depletion of soil or-
Available online 03 January 2020
ganic P due to efficient biogeochemical cycling, but empirical evidence remains scarce. For this purpose, 31P nu-
Editor: Elena Paoletti clear magnetic resonance spectroscopy (31P NMR) was used for quantifying soil P forms and the Hedley
sequential extraction to determine soil P fractions. The studied sites were forestry systems with leguminous
Keywords: trees: mixed forest plantations with different proportions of fast-growing N-fixing leguminous trees; pure plan-
Atlantic Forest tations, and agroforestry systems with leguminous trees. The results show that all N-fixing leguminous trees and
N-fixing leguminous trees N mineral fertilization positively affected the concentrations of available soil P in relation to the control treat-
Soil organic phosphorus ments. There were increases of all P fractions through cycling in all forest sites. 31P NMR spectra clearly identified
Sequential extraction method and quantified that a large amount of phosphomonoesters followed by phosphodiesters in the form of DNA, as
Solution 31P NMR spectroscopy
well as high reserves of Pi species (ortho-P and pyrophosphate) in the first eleven years of growth at pure plan-
Oxisols
tations, mixed plantations or agroforests. The relations between both ortho-P and DNA with the resin-Pi, NaHCO3-
Ultisols
Pi and NaOH-Pi fractions suggest that both analysis methods provide complementary information about the soil P
transformations. Thus, the paper highlights the importance of the use of different N-fixing leguminous tree

⁎ Corresponding author at: UENF/CCTA/Laboratório de Solos, Av. Alberto Lamego 2000, Campos dos Goytacazes, RJ CEP 28013-602, Brazil.
E-mail address: tonygama@uenf.br (A.C. Gama-Rodrigues).

https://doi.org/10.1016/j.scitotenv.2019.136405
0048-9697/© 2020 Published by Elsevier B.V.
2 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405
species under different environmental conditions, production systems and management practices for recovering
heavily degraded areas, which may be a suitable strategy through efficient management of P in highly weathered
tropical soils in the Brazilian Atlantic Forest biome.
1. Introduction
Throughout the world over the years rainforests are being converted
into pasture and monoculture farming. In Brazil, the Atlantic Forest
biome used to cover an area between 100 and 120 million ha (Mha) be-
tween the states of Ceará and Rio Grande do Sul. Today only about 7% of
this biome is still covered with the original forest in fragments with an
area over 100 ha (SOS Mata Atlântica, 2018), as large areas were
deforested to provide construction materials, firewood and provide
land to sugarcane and coffee plantations, and later pastures with low
levels of productivity. This resulted in a high level of soil degradation
(top soil erosion, reduction of organic matter and phosphorus loss).
These lands do not currently serve as reserves of biological diversity,
nor do they generate economic activity to support the rural population
or generate jobs. Thus, tree planting constitutes a possible economic al-
ternative for the recovery of these degraded lands, especially in those
areas with low natural regeneration potential of the forest. This would
reduce the exploitation pressure on the remaining natural forest frag-
ments. In view of this reality, the use of leguminous trees associated
with N-fixing bacteria and arbuscular mycorrhizal fungi (AMF) is a tech-
nique that has shown viability for soil recovery and sustainability in sev-
eral regions in the State of Rio de Janeiro, as witnessed from the positive
influence on several attributes of soil quality (Gama-Rodrigues et al.,
2008; Chaer et al., 2011). Among these, increases in the organic carbon
(C) and nitrogen (N) levels of soil and microbial biomass (Gama-
Rodrigues et al., 2008), increase in the mineralization rates of C and N
of the soil (Nunes et al., 2016) and in the abundance and richness of
the soil fauna (Manhães et al., 2013; Bianchi et al., 2017). But the effects
of such leguminous tree plantings on soil phosphorus (P) form and
availability remain relatively unexplored. Since P is derived from min-
eral weathering and organic matter turnover (as opposed to N derived
from fixation), and often has low availability in tropical settings, under-
standing the effects of legumes on soil P in replanted forests is impor-
tant for elucidating the trajectory of forest restoration.
While N and P are thought to limit primary production across much
of the terrestrial surface (Lambers et al., 2008; Vitousek et al., 2010;
Fisher et al., 2012), P is thought to be more limiting on highly weathered
or ancient tropical forest soils (Gama-Rodrigues et al., 2014), particu-
larly where N-fixing leguminous trees are abundant, despite general N
enrichment conditions of such tropical ecosystems (Houlton et al.,
2008; Nasto et al., 2014). Thus in the lowland forests on stable tropical
cratons, such as those in the Brazilian Amazon (Quesada et al., 2010)
and some of the Brazilian Atlantic Forest (Viana et al., 2018), P may be
more limiting. In these environmental conditions with a high abun-
dance of leguminous trees, the current hypothesis is that the addition
of N via biological fixation may allow greater root phosphatase activity
involved in the cleavage of the soil organic P (Po) thus increasing the
local supply of inorganic P (Pi) for plants, since N fixation entails a
high P cost (Houlton et al., 2008; Nasto et al., 2014, 2017; Png et al.,
2017). Although the magnitude of this interaction reflects differences
between plant species (Batterman et al., 2018; Soper et al., 2019), this
type of N-P interaction would allow the persistence of N-binding le-
gumes in tropical ecosystems rich in N and soil organic matter (SOM),
but substantially limited by P. Also, association with arbuscular mycor-
rhizal fungi (AMF) enhances the capacity of N-fixing leguminous trees
to uptake soil P from Po pool mineralized via enzymes (Nasto et al.,
2014). Thus, increasing P availability affects the amount of N in the eco-
system by affecting N-fixation via feedbacks, while increased activity of
the phosphatase enzymes affects the cycling rate, but not the total P
© 2020 Published by Elsevier B.V.
concentrations within the ecosystem (Wang et al., 2007; Vitousek
et al., 2010; Marklein and Houlton, 2012). However, it is still unclear
how the acceleration of P cycling would affect Po stocks in the soil sur-
face layer because the continuous increase in P availability via increased
phosphatase enzyme activity would lead to depletion of the Po stock in
the forest soils (Fan et al., 2018), limiting the ecosystem growth. In
order to avoid this, the action of the leguminous trees should guarantee
the replacement of Po reserves simultaneously with the activity of the
phosphatase enzyme. Thus, the N-P interaction complex still presents
knowledge gaps that need to be addressed in order to improve the
use of the various soil P forms of varying lability in tropical forest sys-
tems. In this sense, we hypothesized that N-fixing leguminous trees in
association with arbuscular mycorrhizal fungi can increase the availabil-
ity of soil P by exploiting different P sources without causing depletion
of soil Po due to efficient biogeochemical cycling of managed tree plan-
tations in highly weathered soils within the Atlantic Forest biogeo-
graphical domain. The present paper aims to answer the following
questions:
(1) Does the presence of N-fixing leguminous trees affect P fractions
and P forms in the soil differently compared to non-legumes?
(2) Would there be an increase in the labile Pi reserve with a de-
crease in labile Po with the other fractions of P and forms of P
not being affected?
The answer to these questions will improve the understanding of
how the soil P availability is influenced by the use of N-fixing legumi-
nous trees. Thus, the objective of the present paper was to evaluate
the amount and distribution of the different P fractions and forms in
agroforests and forest systems with N-fixing leguminous trees.
For this purpose, solution 31P nuclear magnetic resonance spectros-
copy (NMR) was used to determine the amounts and chemical nature of
P compounds. Complementarily, we used the Hedley sequential extrac-
tion method to characterize different inorganic and organic fractions of
P of distinct biological lability. The integrated use of these two method-
ologies of soil P analysis would, therefore, allow advances in the knowl-
edge about P transformation, useful for establishing management
strategies that aim to increase the availability of P for plants in tropical
forest/agroforestry systems.
2. Materials and methods
2.1. Description of sites
We performed this study with three forest sites located in the State
of Rio de Janeiro, Brazil, all within the Atlantic Forest biome region. A
brief description of the three sites with the respective treatments of
N-fixing leguminous trees is given below. More extensive information
is given in Tables 1 and S1.
Site 1, “Santo Antônio da Aliança Farm” located in the municipality of
Valença (22°22′20″ S, 43°47′23″ W), at an altitude of 650 m, in strongly
undulating relief, sloping around 40 cm m−1. Dominant vegetation was
classified as “Semideciduous Seasonal Forest” (IBGE, 2012). Study was
carried out with an unfertilized pasture and revegetated with mixed
forest plantations with different proportions of fast-growing N-fixing
leguminous trees [0% (T1 - control), 25% (T2), 50% (T3) and 75% (T4)].
Site 2, “Carrapeta Farm” located in the municipality of Conceição de
Macabú (22°05′04″ S, 41°52′07″ W), at an altitude of 160 m, in strongly
 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405 3

Table 1
Description of the three forest sites in the Atlantic Forest region of Brazil.

Sites Soil order Forest Treatments Management Experimental design Selected inoculation
systems
Strains of Arbuscular
N2-fixing mycorrhizal
bacteria fungi

1 Red-yellow Mixed forest T1 0% leguminous Seven tree species of the Fabaceae family were planted: Acacia BR 3465, BR Gigaspora
latosola plantations trees enrichment auriculiformis A. Cunn. ex Benth., Acacia mangium Willd., Pseudosamanea 3609, BR margarita,
(oxisolb) (control) guachapele (Kunth e Harms.), Dalbergia nigra Vellozo, Enterolobium 3407, BR Glomus
T2 25% leguminous contortisiliquum (Vell.) Morong, Piptadenia gonoacantha (Mart.) and 3446. clarum
trees enrichment Mimosa caesalpiniifolia Benth. Six tree species were planted at adding
T3 50% leguminous economic value to the plantation: Peltophorum dubium (Spreng.) Taub.,
trees enrichment Handroanthus heptaphylla (Vell.) Mattos, Handroanthus chrysotrichus
T4 75% leguminous (Mart. ex. DC.) Mattos, Khaya sp., Ceiba speciosa A. St.-Hil. and Colubrina
trees enrichment glandulosa Perk. All tree species were 9 years old. Spacing of 2 m × 2 m
following the level curves of the soil, in a total of 200 plants used per plot
of 35 m × 25 m, in a randomized complete block design with four
treatments (T1, T2, T3 and T4) and three replicates. The region's climate
is type Cwa (Köppen classification system), and is characterized by an
annual rainfall of 1466 mm and a mean annual temperature of 20.4 °C
2 Red-yellow Pure T5 Acacia Acacia auriculiformis and Mimosa caesalpiniifolia planted at a spacing of BR 3465, BR Gigaspora
argisol leguminous auriculiformis 3 m × 2 m following the level curves of the soil, in a total of 250 plants 3609. margarita,
(ultisol) tree T6 Mimosa used per plot of 75 m × 20 m. Experimental plots of each leguminous BR 3407, BR Glomus
plantations caesalpiniifolia tree species were arranged adjacent to each other at the same elevation. 3446. clarum
Thus, each leguminous tree species was considered to be a fixed effect
treatment. The two leguminous tree species were 11 years old. The
region's climate is type Aw (Köppen classification system), and is
characterized by an annual rainfall of 1170 mm and mean annual
temperature of 23 °C
T7 Secondary forest Forest fragment of the Atlantic Forest in secondary ecological succession – –
(control)
T8 Pasture (control) Degraded pasture without grazing with predominance of Imperata – –
brasiliensis Trin., Melinis minutiflora P. Beauv and Paspalum maritimum
Trin. Both covers approximately 40 years old
3 Red-yellow Agroforestry T9 Acacia Euterpe oleraceae and Musa sp. were planted on the same row BR 3465, BR Gigaspora
argisol systems angustissima interspersed 1.5 m from each other, each one at spacing of 3 m × 3 m 3609. margarita,
(ultisol) T10 Gliricidia sepium following the level curves of the soil, in a total of 16 plants of each species Glomus
per plot of 9 m × 9 m, in a randomized complete block design with four clarum
T11 Pueraria replicates. A. angustissima and G. sepium were planted between the rows BR 2613 –
phaseoloides of E. oleraceae and Musa sp. at spacing of 3 m × 1 m. Kaya sp. was
planted in the center of the leguminous trees hedgerows, corresponding
to one individual per plot. P. phaseoloides was also planted between the
rows of E. oleraceae and Musa sp. at spacing of 0.6 m × 0.2 m. The
region's climate is type Aw (Köppen classification system), and is
characterized by an annual rainfall of 1354 mm and mean annual
temperature of 23.5 °C
T12 Low input of Spontaneous vegetation in the rows of Euterpe oleraceae and Musa sp. – –
nitrogen was predominantly Panicum maximum. 45 g of urea (50 kg/ha N) was
fertilizer applied only once on the vegetation
T13 Without Spontaneous vegetation in the rows of Euterpe oleraceae and Musa sp. – –
nitrogen was predominantly Panicum maximum
fertilizer
(control)

Additional information of the fertilization design of all evaluated sites are displayed at the Table S1 in Supplementary files.
a
Brazilian Soil Classification System (Embrapa, 2013).
b
USDA Soil Taxonomy (Soil Survey Staff, 2014).

undulating relief, sloping around 35 cm m−1. Dominant vegetation was
classified as “Tropical Ombrophilous Dense Forest” (IBGE, 2012). The
study was carried out with an unfertilized pasture and revegetated
with pure tree plantations of the leguminous trees Acacia auriculiformis
(T5) and Mimosa caesalpiniifolia (T6) (Table S1). A forest fragment of the
Atlantic Forest in secondary ecological succession (T7) without pres-
ence of leguminous trees, and a degraded pasture (T8) were used as
controls and were located adjacent to the pure plantations (Gama-
Rodrigues et al., 2008).
Site 3, “Experimental Field of Embrapa Agrobiology” located in the
municipality of Seropédica (22°44′29″ S, 43°42′19″ W), at an altitude
of 33 m, in slightly undulating relief, sloping around 3 cm m−1. Domi-
nant vegetation was classified as “Tropical Ombrophilous Dense Forest”
(IBGE, 2012). Study was carried out with an unfertilized pasture and
revegetated with agroforestry systems composed of Euterpe oleraceae,
Musa sp. and Kaya sp. in arrangement of hedgerows with the legumi-
nous trees Acacia angustissima (T9) and Gliricidia sepium (T10), and
with the forage legume Pueraria phaseoloides (T11). From the same
site, spontaneous vegetation with low nitrogen fertilizer input (T12)
and spontaneous vegetation without nitrogen fertilizer (T13 - control)
were also evaluated.
2.2. Soil physical-chemical characterization
At each forest site, a composite sample of soil collected between the
plantation lines, formed of 10 single soil samples taken at 0–10 cm
depth, was taken from each plot in all treatments. At site 2, pasture
and secondary forest treatments, soil samples were taken in a plot of
the same size as the legumes. In the pasture, no sampling was taken
on the ruts formed by cattle walking along the contours. Soil samples
were air dried in the shade and passed through a 2-mm sieve. Total car-
bon (C) and nitrogen (total-N) were determined by using dry combus-
tion in an automated elemental analyzer CHNS/O Series II 2400 (Perkin
Elmer's Inc., Shelton, CT). The following measurements were performed
4 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405

according to the methods described in Embrapa (2011): granulometry by colorimetry for P reactive to ammonium molybdate in SPECORD 210
was performed by pipette method after pretreatment with 35% H2O2 PLUS (Analytik Jena, Jena, DE). A blank (no soil) and a sample of a stan-
and 1.0 M HCl to remove organic matter and carbonates; pH was mea- dard laboratory soil (from a primary Tropical Ombrophilous Dense For-
sured with a glass electrode in a 1:2.5 water stirred suspension and P est at southern part of the state of Bahia, Brazil) were included with each
was extracted with Mehlich-1. Total P was determined by the method procedure.
proposed by Hedley et al. (1982): a H2SO4 + H2O2 digestion of 5.0 g According to the description of Aleixo et al. (2017), the soil P frac-
of macerated soil. The total P concentration was determined by colorim- tions were characterized using its degree of availability to plants: Labile
etry for inorganic P (Pi) reactive to ammonium molybdate at 880 nm by P (resin-Pi + NaHCO3-Pi + NaHCO3-Po) represents readily available
the method of Murphy and Riley (1962) in the SPECORD 210 PLUS forms and adsorbed with low energy to mineral soil surfaces. Moder-
(Analytik Jena, Jena, DE). The results of physical and chemical analyses ately labile P (NaOH-Pi + NaOH-Po) represents forms adsorbed with
of soils are displayed in Table 2. The soil properties before the experi- higher energy to surfaces of Fe and Al oxides. Occluded P (sonic-
ments started were the same as the control treatments of each site: Pi + sonic-Po) represents the P adsorbed at the internal surfaces of
Treatment 1 at site 1; treatment 7 at site 2, and treatment 13 at site 3. soil aggregates. HCl-Pi fraction represents the P bound to Ca element
(Ca-P, apatite) with very restricted availability. Residue-P corresponded
2.3. Fractionation of soil P to the chemically stable Po and relatively insoluble Pi forms, which pre-
vious steps could not extract.
P fractions in soil samples were extracted sequentially using the
Hedley method (Hedley et al., 1982) that separates soil P into five inor- 2.4. NaOH–Na2EDTA extraction
ganic fractions, three organic fractions, and one residue fraction. The ex-
traction procedure was performed as follows: 0.5 g of air-dried soil was Soil P was extracted by shaking 4.0 g of soil with 32 mL of a solution
placed in 15 mL tubes. P fractions of soil were extracted sequentially containing 0.25 mol L−1 NaOH + 0.05 mol L−1 of disodium ethylenedi-
using anion exchange resin (AER) that measured 1 × 5 cm (ANION aminetetraacetate dihydrate (Na2EDTA·2H2O) (99.0–101.0%; Sigma-
204UZRA), 0.5 mol L−1 NaHCO3 at pH 8.5, 0.1 mol L−1 NaOH, Aldrich Co., St. Louis, USA) for 10 h at 25 °C. After previous tests, we
0.1 mol L−1 NaOH plus ultrasonic treatment performed for 2 min, and choose an extraction time of 10 h to limit the degradation of P com-
1.0 mol L−1 HCl. Residue-P was obtained by H2SO4 + H2O2 in a block di- pounds and thus improve the recovery of total P in the sample in accor-
gester at 200 °C. Each extraction step was performed for 16 h at 120 rpm dance with Aleixo et al. (2019). After this step, each extract was
(21 °C) and followed by centrifugation at 5000 rpm for 20 min, the ex- centrifuged at 5000 rpm for 15 min and the supernatant filtered
tract was reserved and 5 mL NaCl 0.5 mol L−1 was added to the soil that through 0.45 μm membrane filters (75–80% nitrate and acetate;
remained in the tubes from the previous steps. This step was followed Millipore Inc., Massachusetts, USA), frozen with liquid nitrogen, lyoph-
by another centrifugation at 5000 rpm for 5 min; the supernatant was ilized (~52 h) and stored. Triplicate lyophilized NaOH–Na2EDTA ex-
placed in the same containers from the previous extract. Total P (Pt) in tracts for each soil were blended for 31P NMR spectroscopy.
each extract of 0.5 mol L−1 NaHCO3, 0.1 mol L−1 NaOH, and
0.1 mol L−1 NaOH ultrasonic was determined using one aliquot of the 2.5. Elemental contents in NaOH–Na2EDTA extracts and 31P NMR measure-
extract from the initial extraction steps that was autoclaved at 121 °C ment parameters
and 103 kPa for 3 h with 1.0 mL of 24 mol L−1 H2SO4 and 10 mL of
(NH4)2S2O8 added. After cooling, the volume was completed to 20 mL An aliquot (3.0 mL) of each NaOH–Na2EDTA extract of soil was
with deionized water. Inorganic P (Pi) in resin, HCl, residue-P and Pt ex- autoclaved at 121 °C and 103 kPa for 3 h with 1.0 mL of 24 mol L−1
tracts was determined at 880 nm by the method of Murphy and Riley H2SO4 and 10 mL of (NH4)2S2O8 added. After cooling, the volume was
(1962). The Pi in NaHCO3 and NaOH extracts was determined at adjusted to 20 mL with deionized water for molybdate reactive P anal-
700 nm by the method of Dick and Tabatabai (1977). Organic P (Po) ysis by colorimetry at 880 nm by the method of Murphy and Riley
was calculated as the difference between the Pt and Pi concentrations (1962) in SPECORD 210 PLUS (Analytik Jena, Jena, DE). These soil P mea-
(Po = Pt − Pi) in each extract. All extracts from the steps were analyzed surements were described as PNaOH–EDTA total. To obtain the concentra-
tions of paramagnetic ions (Fe and Mn), an aliquot (3.0 mL) of each
NaOH–Na2EDTA extract of soil was taken for analysis by inductively
Table 2
Physical and chemical attributes of soils (0–10 cm) under different forest sites. coupled plasma atomic emission spectrometry (ICP AES) after
H2SO4 + H2O2 digestion following the modified method proposed by
Sites Treatments Clay pH (H2O) Mehlich-P Total soil P SOC Total N Doolette et al. (2009, 2011). This enabled the calculation of the ratio of
% mg kg−1 mg kg−1 g kg−1 P relative to Fe and Mn [P/(Fe + Mn)] in the solution for 31P NMR spec-
1 T1 30.5 5.4 3.7 490 23.2 2.0 troscopy, in accordance with McDowell et al. (2006). Briefly, the aver-
T2 30.5 5.5 4.3 460 24.5 2.0 age concentrations of Fe and Mn were 1.0 ± 0.003 mg L−1 and
T3 30.5 5.4 4.7 490 24.1 2.4 0.07 ± 0.01 mg L−1 (n = 9), respectively (Table S2). The average rela-
T4 30.5 5.2 6.7 530 21.8 3.1
tive ratios of PNaOH-EDTA total to Fe and Mn in the solution 31P NMR spec-
2 T5 18.5 4.8 2.8 520 32.1 0.8
T6 29.1 4.5 2.4 545 31.9 1.5 troscopy was 0.35 ± 0.07, n = 9. This means that short recovery delay
T7 38.6 4.2 2.8 595 31.2 1.3 times (D1) were sufficient to obtain accurate and quantitative results
T8 28.5 4.7 2.0 525 29.4 1.5 (Fig. S1, Table S3) (McDowell et al., 2006).
3 T9 31.8 4.4 1.2 495 15.1 1.3
T10 33.0 4.3 2.3 525 12.1 1.2
2.6. Solution 31P NMR spectroscopy
T11 32.0 4.3 3.1 450 11.1 1.0
T12 32.5 4.4 3.8 675 13.2 1.2
31
T13 36.2 4.4 1.1 445 12.2 1.1 P NMR spectroscopy was carried out after re-solubilizing 200 mg
Total soil P: H2SO4 + H2O2 digestion. SOC: soil organic carbon. Site 1 – mixed forest plan- of each lyophilized extract (equivalent to ~1.0 g of soil) with 0.9 mL so-
tation: (T1) 0% leguminous trees, (T2) 25% leguminous trees, (T3) 50% leguminous trees lution containing 0.25 mol L−1 NaOH + 0.05 mol L−1 Na2EDTA plus
and (T4) 75% leguminous trees cover; Site 2 – pure plantations of leguminous trees: 0.1 mL of deuterium oxide (99.9 at.% D; Sigma-Aldrich Co., St. Louis,
(T5) Acacia auriculiformis, (T6) Mimosa caesalpiniifolia, (T7) secondary forest and (T8) pas- USA), and 0.1 mL of a solution of 2.5 μg mL−1 of methylenediphosphonic
ture; Site 3 – agroforestry systems: (T9) Acácia angustíssima, (T10) Gliricidia sepium, (T11)
Pueraria phaseoloides, (T12) spontaneous vegetation with low input of nitrogen fertilizer
acid (MDPA) (≥99.0%; Sigma-Aldrich Co., St. Louis, USA), as an internal
and (T13) spontaneous vegetation without nitrogen fertilizer. See detailed description of standard. Each extract was centrifuged at 14,000 rpm for 5 min and
forest sites in the Materials and methods section and Table S1. then transferred to a 5 mm NMR tube. Solution 31P NMR spectra were
 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405
acquired with broadband 1H decoupling operating in the radio fre-
quency band of 202.446 MHz on a Avance III HD 500 MHz (Bruker
Inc., DE), using the standard Bruker pulse program zg30 (6.0 μs), 0.4 s
data acquisition time, 0.5 s recovery delay time (D1) and operating at
25 °C (total repetition rate 0.9 s). Approximately 50,000 scans were ac-
quired for each sample (total measurement time 13.3 h). Longer relax-
ation delays were suggested in most studies (Cade-Menun and Liu,
2014), but too long relaxation delays lead to a decreased signal to
noise ratio. Therefore, the relaxation delays in our measurements
were intentionally short, and 30-degree pulses were used to obtain
the highest signal to noise ratio possible in a short time (Fig. S1;
Table S3), as was shown in the paper by Aleixo et al. (2019).
The free induction decay (FIDs) data was processed with line-
broadening exponential function of 5 Hz, before Fourier-transform, to
improve accuracy in the 31P spectra (processed data). Fully automatic
baseline correction was applied and the phase was adjusted manually.
Chemical shifts of signals (δ) were determined in parts per million
(ppm) relative to an external standard of H3PO4 (≥85.0%; Merck KGaA,
Darmstadt, DE). It should be noted here that some signals are affected
by eventual variations in chemical shift, due to small temperature vari-
ations. Accordingly, spectra were calibrated to the ortho-P signal (stan-
dardized at δ 6.0 ppm). No further alignment procedures were used on
these data.
Signal areas were obtained by integration and the concentration of
each P compound was calculated from the integration values of the
MDPA (mean ± standard error, δ 17.57 ± 0.02 ppm, n = 9). Signals
were assigned to P compounds based on literature reports of model
compounds spiked in NaOH–Na2EDTA soil extracts and were
interpreted as suggested by Cade-Menun (2015) and Doolette et al.
(2009, 2011). All spectral processing and peaks areas were calculated
by integration using Delta NMR Software (Jeol USA Inc., Peabody,
USA). The relative proportion of each P compound detected in 31Po
31
RMN total (diester-P + monoester-P) and Pi NMR total (ortho-P + pyro-
phosphate) was calculated and then compared to the PNaOH-EDTA total,
as well as the proportion of the 31Pi NMR total and 31Po NMR total in the
total soil P composition obtained by H2SO4 + H2O2 digestion (Turner
and Engelbrecht, 2011; Aleixo et al., 2019).
In the present paper, the soil organic P forms are generally grouped
as orthophosphate monoesters (monoester-P) (R1OPO2− 3 , where R1 are
carbon moieties) and orthophosphate diesters (diester-P) (R1OR2OPO− 2 ,
where R1 and R2 are carbon moieties), while the inorganic P forms are
grouped in orthophosphate (PO− 4−
4 ) and pyrophosphate (P2O7 ). Or-
ganic/inorganic polyphosphates (linear chains of orthophosphate with
high-energy phosphate-phosphate bonds) and phosphonates (contain-
ing a group with a direct carbon-P bond) are not easily detected by 31P
NMR (Cade-Menun and Liu, 2014). Phosphate monoesters are often
considered to be of limited availability to plants (Turner and
Engelbrecht, 2011), because of their high stability in the soil (Condron
et al., 2005). In contrast, phosphate diesters constitute the largest por-
tion of labile Po due to their fast mineralization (Bowman and Cole,
1978; Morrissey et al., 2015).
2.7. Statistical analysis
The data were subjected to the Kolmogorov-Smirnov and Lilliefors
normality test, homoscedasticity and normal distribution residuals,
which evaluate the normal distribution of analyzed variables. Only 31P
NMR data were transformed by √X when not normally distributed (as
determined by Kurtosis for values significantly different from zero;
p b 0.05). For each evaluated forest site, the analysis of variance
(ANOVA) was carried out and the Tukey test (p = 0.05) was used to
compare the concentrations of P fractions obtained by the Hedley's se-
quential extraction method of leguminous trees treatments in relation
to the control of each site. The relative contribution of P fractions to
total extracted P was calculated for each treatment at all the sites. The
relative increase of P fractions and P forms of leguminous trees
5
treatments was quantified in relation to the control of each site. The re-
lationships between the P fractions and the P form and of them with the
physical and chemical attributes of soils were measured using the
Pearson's correlation and only the significant ones were presented in
the text. The R software (www.r-project.org) were used for all statistical
analyses.
3. Results
3.1. Distribution of the P fractions
At site 1 (mixed forest plantations), there was a significant in-
crease in the contents of all soil P fractions at the highest planting
density of leguminous trees - 75% (T4) (Fig. 1, Table 3). The highest
relative increments were in the HCl-P i (+49.3%), NaOH-P o
(+45.1%), resin-Pi (+42.7%) and NaHCO3-Pi (+35.2%) fractions. In
the other P fractions, the relative increase varied between 16.7%
(residue-P) and 23.7% (sonic-P o ). The relative increase of total P o
was 37.5%, of the total Pi was 22.1%, and of the extracted P t was
22.9% (Fig. 1). There was a relative increase of 27.3% of the labile P
fraction (resin-Pi + NaHCO3-Pi + NaHCO3-Po), 34.8% of the moder-
ately labile P fraction (NaOH-Pi + NaOH-Po), 21.2% of the occluded
P fraction (sonic-Pi + sonic-Po), and 17.3% of the stable P fraction
(HCl-Pi + residue-P) (Fig. 1). The Po corresponded to 64.8%, 67.4%
and 30%, respectively, of the labile P, moderately labile P and oc-
cluded P fractions (Table 3).
The relative distribution of each P fraction in relation to extracted
Pt was similar in all treatments (Fig. 1). Thus, of the extracted Pt, on
average, the resin-Pi fraction corresponded to 0.5%, the NaHCO3-Pi
fraction 1.2%, the NaHCO3 -Po fraction 3.3%, the NaOH-Pi fraction
9%, the NaOH-Po fraction 16.5%, the sonic-P i fraction 11.4%, the
sonic-Po fraction 4.7%, the HCl-Pi fraction 1.1% and the residue-P frac-
tion 52.5%. On average, the total Po (NaHCO3-Po + NaOH-Po + sonic-
Po) represented 24.5% of the extracted Pt, and the total Pi (NaHCO3-
Pi + NaOH-Pi + sonic-Pi + HCl-Pi) represented 23.1%. Thus, the geo-
chemical P (total Pi + residue-P) constituted around 75.6% of the ex-
tracted Pt.
At site 2 (pure leguminous tree plantations), there was wide var-
iation in the concentrations for all P fractions across the forest covers
(Fig. 2, Table 3). In leguminous tree plantations (T5 and T6), there
were significant increases for all fractions of labile P (resin-P i ,
NaHCO 3-Pi and NaHCO3 -P o) in relation to pasture (T8 - control).
There were also significant increases in the fractions of NaOH-P i
and HCl-Pi. Also all P fractions from the secondary forest (T7) pre-
sented higher values, statistically similar to the leguminous tree
plantations, except for the NaOH-Pi fraction in relation to the Mimosa
caesalpiniifolia plantation (T6). Thus, the two leguminous tree spe-
cies presented higher concentrations of total Pi in relation to pasture,
but similar to the secondary forest (Fig. 2). However, the Acacia
auriculiformis plantation (T5) had the lowest total Po concentration,
while Mimosa caesalpiniifolia (T6) and pasture (T8) were similar to
each other, and all of these coverages were lower than the secondary
forest (T7) (Fig. 2).
The proportion of Po was 66.2% of the labile P fraction in the plan-
tation of Acacia auriculiformis, while in the other vegetation cover it
was, on average, 76.4%. In the plantations with the two leguminous
trees, the Po corresponded, on average, to 55.5% of the moderately
labile P fraction, while in the pasture and the secondary forest it
was, on average, 68.9% (Table 3). The Po corresponded to 22.1%,
36.9%, 44.1% and 46.9% of the occluded P fraction, respectively in Aca-
cia auriculiformis, Mimosa caesalpiniifolia, pasture and secondary for-
est (Fig. 2).
At all treatments the resin-Pi fraction of the extracted Pt ranged
from 0.2 to 0.5%, the NaHCO3 -P i fraction ranged from 0.5 to 1.3%,
the NaHCO3-Po fraction ranged from 2.0 to 4.3%, the NaOH-Pi fraction
ranged from 7.4% to 9.9%, the NaOH-Po fraction ranged from 12.1 to
6 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405

Fig. 1. Relative distributions of the phosphorus fractions in soils under different land uses. Site 1 – mixed forest plantations: (T1 - control) 0% leguminous trees, (T2) 25% leguminous trees,
(T3) 50% leguminous trees, (T4) 75% leguminous trees cover. See detailed description of forest sites in the Materials and methods section and Table S1. Labile P: resin-Pi + NaHCO3-
Pi + NaHCO3-Po. Moderately labile P: NaOH-Pi + NaOH-Po. Occluded P: sonic-Pi + sonic-Po. Stable P: residue-P + HCl-Pi. Total Po: sum of the NaHCO3-Po, NaOH-Po and sonic-Po. Total
Pi: sum of the resin-Pi, NaHCO3-Pi, NaOH-Pi, sonic-Pi and HCl-Pi.

16.9%, the sonic-Pi fraction ranged from 3.9 to 5.4%, the sonic-Po frac- the total Pi represented 12.3 to 18.2%. Thus, the geochemical P con-
tion ranged from 1.5% to 4.2%, the residue-P fraction ranged from stituted around 78.4% of the extracted Pt.
61.3 to 65.7%, while the HCl-Pi fraction varied around 0.1% At site 3 (agroforestry systems), the treatment with low input of ni-
(Table 3). Total Po represented 17.2 to 25% of the extracted Pt, and trogen fertilizer (T12) promoted a significant increase of all soil P

Table 3
Phosphorus fractions (mg kg−1) of soils under different forest sites.

Sites Treatments Labile P Moderately labile P Occluded P Ca-P Stable P

Resin-Pi NaHCO3-Pi NaHCO3-Po NaOH-Pi NaOH-Po Sonic-Pi Sonic-Po HCl-Pi Residue-P ∑Pi ∑Po Extracted Pt

1 T1 3.1 c 7.8 c 22.7 b 62.3 c 105.5 c 77.8 d 33.3 b 6.9 b 364.6 d 158.0 d 161.4 b 161.4 b
T2 3.0 c 7.8 c 23.3 b 64.3 bc 107.8 c 82.9 c 33.3 b 7.3 b 382.7 c 165.3 c 164.4 b 164.4 b
T3 3.6 b 9.0 b 24.5 b 68.8 b 128.2 b 87.8 b 33.3 b 8.2 b 404.1 b 177.4 b 186.0 b 186.0 b
T4 4.5 a 10.6 a 27.8 a 74.2 a 153.1 a 93.3 a 41.1 a 10.3 a 425.3 a 192.9 a 222.0 a 222.0 a
Means 3.6 8.8 24.6 67.4 123.7 85.5 35.3 8.2 394.2 173.4 183.5 751.1
CV (%) 16.4 13.6 12.1 8.2 17.4 8.9 19.1 21.0 6.1 8.5 13.9 8.3
2 T5 3.8 a 9.5 a 25.9 b 78.7 ab 88.1 c 39.4 bc 11.2 c 1.0 b 468.9 c 132.3 a 125.2 c 125.2 c
T6 3.6 a 7.3 b 37.1 a 85.5 a 118.5 b 43.9 ab 25.7 b 1.2 ab 538.6 b 141.5 a 181.3 b 181.3 b
T7 3.1 a 9.6 a 41.2 a 71.7b c 159.5 a 45.6 a 40.2 a 1.4 a 589.8 a 131.3 a 240.9 a 240.9 a
T8 1.6 b 3.9 c 16.7 c 62.9 c 138.3 ab 32.0 c 25.3 b 0.6 c 538.6 b 100.9 b 180.2 b 180.2 b
Means 3.0 7.6 30.2 74.7 126.1 40.2 25.6 1.0 534.0 126.5 181.9 842.3
CV (%) 28.3 30.8 31.7 11.2 20.9 13.2 40.1 28.9 8.1 12.1 22.5 10.0
3 T9 6.7 b 24.1 b 25.1 a 108.7 c 118.7 bc 33.5 c 28.0 b 2.2 c 346.9 c 175.1 c 171.8 b 171.8 b
T10 8.8 a 40.2 a 20.9 a 155.9 b 137.2 ab 42.3 b 36.3 ab 5.3 b 404.5 b 252.5 b 194.5 ab 194.5 ab
T11 6.4 b 26.4 b 26.8 a 108.6 c 123.8 bc 33.8 c 29.8 b 2.3 c 357.6 c 177.5 c 180.4 b 180.4 b
T12 9.0 a 41.3 a 21.2 a 169.8 a 161.3 a 46.7 a 43.4 a 6.3 a 446.1 a 273.0 a 225.9 a 225.9 a
T13 5.7 b 23.4 b 22.8 a 107.6 c 108.3 c 32.8 c 26.6 b 2.1 c 366.1 c 171.6 c 157.7 b 157.7 b
Means 7.3 31.1 23.4 130.1 129.9 37.8 32.8 3.7 384.3 210.0 186.1 780.3
CV (%) 18.4 25.5 9.7 20.8 14.1 14.9 19.1 49.3 9.5 20.8 12.5 13.0

Within a column, values followed by the same letter are not significantly different among treatments of each forest site by Tukey test (p = 0.05). ∑Pi: sum of the resin-Pi, NaHCO3-Pi,
NaOH-Pi, sonic-Pi and HCl-Pi. ∑Po: sum of the NaHCO3-Po, NaOH-Po and sonic-Po. Extracted Pt: sum of the ∑Pi and ∑Po. Site 1 – mixed forest plantations: (T1) 0% leguminous trees,
(T2) 25% leguminous trees, (T3) 50% leguminous trees and (T4) 75% leguminous trees cover; Site 2 – pure plantations of leguminous trees: (T5) Acacia auriculiformis, (T6) Mimosa
caesalpiniifolia, (T7) secondary forest and (T8) pasture; Site 3 – agroforestry systems: (T9) Acácia angustíssima, (T10) Gliricidia sepium, (T11) Pueraria phaseoloides, (T12) spontaneous veg-
etation with low input of nitrogen fertilizer and (T13) spontaneous vegetation without nitrogen fertilizer. See detailed description of forest sites in the Materials and methods section and
Table S1.
 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405 7

Fig. 2. Relative distributions of the phosphorus fractions in soils under different land uses. Site 2 – pure leguminous trees plantations: (T5) Acacia auriculiformis, (T6) Mimosa caesalpiniifolia,
(T7 - control) secondary forest, (T8 - control) pasture. See detailed description of forest sites in the Materials and methods section and Table S1. Labile P: resin-Pi + NaHCO3-Pi + NaHCO3-
Po. Moderately labile P: NaOH-Pi + NaOH-Po. Occluded P: sonic-Pi + sonic-Po. Stable P: residue-P + HCl-Pi. Total Po: sum of the NaHCO3-Po, NaOH-Po and sonic-Po. Total Pi: sum of the
resin-Pi, NaHCO3-Pi, NaOH-Pi, sonic-Pi and HCl-Pi.

fractions when compared to treatment with spontaneous vegetation
without nitrogen fertilizer (T13 – control) (Fig. 3, Table 3). In addition,
T12 was only slightly superior to the treatment with Gliricidia sepium
(T10). Treatments with Acacia angustissima (T9) and Pueraria
phaseoloides (T11) always presented lower values of P fractions when
compared to treatment with nitrogen fertilization (T12). T10 treatment
was similar to T9 and T11 only for the NaOH-Po fraction, and there was
no statistical significant difference between T10 and other treatments
for the sonic-Po fraction. On average, the relative proportion of Po was
30% of the labile P fraction in T10 and T12, whereas the other treatments
it was 45% (Fig. 3). On average, the Po corresponded to 47.8% of the mod-
erately labile P fraction in T10 and T12, while in the other treatments it
was 52%. The Po, on average, corresponded to 46.6% of the occluded P
fraction in all treatments (Fig. 3).
In all treatments of site 3, the resin-Pi fraction of the extracted Pt,
ranged from 0.8 to 1%, the NaHCO3-Pi fraction ranged from 3.4 to 4.7%,
the NaHCO3-Po fraction ranged from 2.2 to 3.7%, the NaOH-Pi fraction
ranged from 15.2 to 18.3%, the NaOH-Po fraction ranged from 15.6 to
17.3%, the sonic-Pi fraction ranged from 4.7 to 5%, the HCl-Pi fraction
ranged from 0.3 to 0.7%, and the residue-P fraction from 47.2 to 52.7%.
Total Po represented 22.7 to 25.2% of the extracted Pt, and the total Pi
represented 24.7 to 29.7%. Thus, the geochemical P constituted around
76.1% of the extracted Pt (Table 3).
Organic C was negatively correlated with the fractions of resin-Pi,
NaHCO3-Pi, NaOH-Pi and moderately labile P, and positively correlated
with the fractions of residue-P and geochemical P (Table 4). Total-N
was correlated with the fractions of sonic-Pi and occluded P. Clay was
correlated with the fractions of NaOH-Po, sonic-Po and extracted Pt.
The pH value was correlated with the fractions of sonic-Pi and occluded
P (Table 4).
3.2. P composition determined by solution 31P NMR spectroscopy
3.2.1. Peak identification
Representative solution 31P NMR spectra of selected treatments
of each forest site are shown in Figs. 4, 5 and 6. The spectra displayed
were individually reduced to the spectral region of δ 9.0 to
−6.0 ppm. Inorganic P was detected as ortho-P at δ 6.0 ppm (n =
9) and pyrophosphate at δ −4.89 ± 0.02 ppm (n = 9). Monoester-
P signals detected were divided into three general groups, based on
their locations: Mono1, Mono2 and Mono3. The Mono2 group as
downfield of ortho-P signal between δ 5.9 and 4.0 ppm (n = 9). Spe-
cific peaks in this region were detected at δ 5.89 ± 0.07; 5.69 ± 0.02;
5.49 ± 0.03; 5.38 ± 0.01; 5.22 ± 0.01; 5.2 ± 0.02; 4.9 ± 0.02; 4.71 ±
0.03; 4.51 ± 0.02 ppm (n = 9); δ 4.9 ± 0.02 ppm (n = 5). At Mono1
group a specific peak in this region were detected at δ 6.81 ±
0.09 ppm (n = 6). At Mono3 group were not detected peaks, gener-
ally distributed from δ 3.9 to 2.0 ppm. In diester-P region only the de-
oxyribonucleic acid (DNA) was specifically detected and identified at
δ −0.98 ± 0.05 ppm (n = 9). Long chain organic and/or inorganic
polyphosphates were not detected, generally distributed from δ
−4.0 to −20.9 ppm, and phosphonates generally distributed from
δ 38.4 to 9.9 ppm. According to Turner et al. (2012) the orthophos-
phate monoesters regions were dominated by the inositol
hexakisphosphate (IHP) stereoisomers, which are classified as
myo-IHP containing six phosphates grouped in four signals, with a
specific arrangement of 1:2:2:1 (respectively C2, C1 and C3, C4 and
C6, and C5 phosphates), scyllo-IHP presents six phosphates on the
inositol ring grouped at one signal, D-chiro-IHP contains an assign-
ment to the equatorial phosphate groups in the 2-equatorial/4-
axial (D-chiro-IHP 2e/4a) conformer, and neo-IHP presents six
8 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405

Fig. 3. Relative distributions of the phosphorus fractions in soils under different land uses. Site 3 – agroforestry systems: (T9) Acácia angustíssima, (T10) Gliricidia sepium, (T11) Pueraria
phaseoloides, (T12) spontaneous vegetation with low input of nitrogen fertilizer and (T13 - control) spontaneous vegetation without nitrogen fertilizer. See detailed description of forest
sites in the Materials and methods section and Table S1. Labile P: resin-Pi + NaHCO3-Pi + NaHCO3-Po. Moderately labile P: NaOH-Pi + NaOH-Po. Occluded P: sonic-Pi + sonic-Po. Stable P:
residue-P + HCl-Pi. Total Po: sum of the NaHCO3-Po, NaOH-Po and sonic-Po. Total Pi: sum of the resin-Pi, NaHCO3-Pi, NaOH-Pi, sonic-Pi and HCl-Pi.

phosphates at 4-equatorial/2-axial conformer (neo-IHP 4e/2a). 3.2.2. Chemical composition of soil P

Sugar phosphates (glucose-1-phosphate and -6-phosphate) can be
found in this spectral region (McLaren et al., 2015). In addition, α-
and β-glycerophosphate peaks, initially detected in Mono2 group
in some studies, are actually phospholipid degradation products dur-
ing NaOH–Na2EDTA extraction-method, while the detected peaks of
mononucleotides are formed predominantly from ribonucleic acid
(RNA) degradation (He et al., 2011; Young et al., 2013; Schneider
et al., 2016).
The total P extracted in NaOH–Na2EDTA and determined by solution
31
P NMR spectroscopy (31PNMR total) ranged from 200.4 to 469 mg kg−1
(341 ± 31.7 mg kg−1), and represented 38 to 83% (mean 63 ± 5.2%) of
the total soil P (Table 5). Values determined in extracts by colorimetry
for reactive P to ammonium molybdate (P NaOH–EDTA total) were similar
to those determined by 31PNMR total, ranging from 211.4 to 558.mg kg−1
(mean 373 ± 37.1 mg kg−1) and represented 40 to 85% (mean 69 ±
5.7%) of the total soil P (Table 5). On average, the 31PNMR total

Table 4
Correlation coefficients between phosphorus and physical and chemical attributes of soils (0–10 cm) under different forest sites (n = 9).
31
Phosphorus fractions P NMR spectroscopy Soil attributes

ortho-P Pyrophosphate Monoester-P DNA C Total N Clay pH

Resin-Pi 0.87⁎⁎ – – 0.84⁎⁎ −0.84⁎⁎ – – –

NaHCO3-Pi 0.76⁎ – – 0.76⁎ −0.88⁎⁎ – – –
NaHCO3-Po – – – – – – – –
NaOH-Pi 0.81⁎⁎ – – 0.81⁎⁎ −0.80⁎⁎ – – –
NaOH-Po – – – – – – 0.76⁎ –
Sonic-Pi – – – – – 0.75⁎ – 0.75⁎
Sonic-Po – – – – – – 0.75 –
HCl-Pi – – – – – – – –
Residue-P – – – – 0.72⁎ – – –
Extracted Pt – – – – – – 0.75⁎ –
Labile P 0.81⁎⁎ – – – – – – –
Moderately labile P 0.84⁎⁎ – – 0.77⁎ −0.72⁎ – – –
Occluded P – – – – – 0.72⁎ – 0.71⁎
Geochemical P – – – – 0.75⁎ – – –

Labile P: sum of the resin-Pi, NaHCO3-Pi and NaHCO3-Po. Moderately labile P: sum of the NaOH-Pi and NaOH-Po. Occluded P: sum of the Sonic-Pi and Sonic-Po. Geochemical-P: sum of the
Total Pi fractions and Residue-P. Extracted Pt: sum of the resin-Pi, NaHCO3-Pi, NaHCO3-Po, NaOH-Pi, NaOH-Po, Sonic-Pi, Sonic-Po, HCl-Pi and Residue-P.
⁎ Significant at p b 0.05.
⁎⁎ Significant at p b 0.01.
 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405 9

Fig. 4. 31P NMR spectra of soil in NaOH–Na2EDTA extracts under different land uses. Spectra are scaled to the orthophosphate peaks (δ 6.0 ppm). Details in the pyrophosphate, Monoester-P
(Mono1, Mono2 and Mono3) and DNA regions are shown. Spectra were plotted with 5 Hz line broadening for the main spectra. Site 1 – mixed forest plantations: (1) 0% leguminous trees
and (4) 75% leguminous trees. See detailed description of forest sites in the Materials and methods section and Table S1.

concentrations corresponded to 91% of the PNaOH–EDTA total. A linear re-
gression between 31PNMR total and PNaOH–EDTA total, with the intercept
forced through the origin was described by the equation:

At site 2 (pure leguminous tree plantations), the two species of legu-
31
√ ðPNaOH–EDTA totalÞ ¼ 1:04621  0:008  √ PNMR total ; r 2
¼ 0:980; pb0:0001; n ¼ 9:
The 31Pi RMN total values (sum of ortho-P and pyrophosphate) ranged
from 143.2 to 275.3 mg kg−1 and represented 27 to 47% of the total soil
P, while 31Po NMR total values (sum of monoester-P and diester-P)
ranged from 57.2 to 200.4 mg kg−1 and represented 10 to 39% of the
total soil P (Table 5).
The ortho-P values were in the range of 124.4 to 237.6 mg kg−1,
representing, on average, 86% of the 31Pi NMR total, while pyrophosphate
concentrations ranged from 18.8 to 38.5 mg kg−1. Most of the 31Po NMR
total occurred as a monoester-P (mean of 94%) and ranged from 52.6 to
191.9 mg kg−1, while diester-P concentrations (DNA) ranged from 4.1
to 17 mg kg−1 (Table 6).
At site 1 (mixed forest plantations), the highest planting density of
leguminous trees (T4) promoted a high increases in ortho-P (+22.4%),
monoester-P (+106%) and DNA (+60%) in relation to control treat-
ment (T1, without leguminous trees), but there was practically no
change in the pyrophosphate concentration (Table 6).
minous trees (T5 and T6) promoted high increases in the concentrations
of all P species in relation to secondary forest (T7) and pasture (T8)
(Table 6). Notably in relation to pasture, in the Acacia auriculiformis
(T5) and Mimosa caesalpiniifolia (T6) plantations the increments were,
respectively, +55,4% and +53% for ortho-P, +105% and +57% for pyro-
phosphate, +265% and +227% for monoester-P, and +85% and +55%
for DNA.
At site 3 (agroforestry systems), Gliricidia sepium in hedgerow (T10)
and spontaneous vegetation with nitrogen fertilization (T12) promoted
a high increase in concentrations of all P forms in relation to spontane-
ous vegetation without nitrogen fertilizer treatment (T13 - control). In
T10, the increase was +23% for ortho-P, +67% for pyrophosphate,
+112% for monoester-P and +160% for DNA, while increases of +40%
for ortho-P, +74% for pyrophosphate, +163% for monoester-P and
+227% for DNA occurred in T12 (Table 6). Thus, the overall effect of le-
guminous trees on soil P across all three sites showed that the most

Fig. 5. 31P NMR spectra of soil in NaOH-Na2EDTA extracts under different land uses. Spectra are scaled to the orthophosphate peaks (δ 6.0 ppm). Details in the pyrophosphate, Monoester-P
(Mono1, Mono2 and Mono3) and DNA regions are shown. Spectra were plotted with 5 Hz line broadening for the main spectra. Site 2 – pure plantations of leguminous trees: (5): Acacia
auriculiformis, (6) Mimosa caesalpiniifolia, (7) secondary forest and (8) pasture. See detailed description of forest sites in the Materials and methods section and Table S1.
10 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405

Fig. 6. 31P NMR spectra of soil in NaOH-Na2EDTA extracts under different land uses. Spectra are scaled to the orthophosphate peaks (δ 6.0 ppm). Details in the pyrophosphate, Monoester-P
(Mono1, Mono2 and Mono3) and DNA regions are shown. Site 3 – agroforestry systems: (T10) Gliricidia sepium, (T12) spontaneous vegetation with low input of nitrogen fertilizer and
(T13) spontaneous vegetation without nitrogen fertilizer. See detailed description of forest sites in the Materials and methods section and Table S1.

positively affected P form was monoester-P, followed by DNA, pyro- 4. Discussion

phosphate and orthophosphate (Fig. 7).
ortho-P was positively correlated with pyrophosphate, monoester-P
and DNA. Pyrophosphate was positively correlated with monoester-P
and DNA. Monoester-P was correlated with DNA (Table 7). As for corre-
lations between P compounds and soil P fractions, ortho-P was posi-
tively correlated with the fractions of resin-Pi, NaHCO3-Pi and NaOH-
Pi, and with the fractions of labile P (resin-Pi + NaHCO3-Pi + NaHCO3-
Po) and moderately labile P (NaOH-Pi + NaOH-Po) (Table 4). DNA was
positively correlated with the fractions of resin-Pi, NaHCO3-Pi and
NaOH-Pi, and with the fraction of moderately labile P. There were no
significant correlations between 31Pi NMR total and 31Po NMR total, nor be-
tween the soil attributes and these two P groups.
Table 5
Phosphorus fractions in NaOH–Na2EDTA extracts of soil under different forest sites, deter-
mined by solution 31P NMR spectroscopy.
4.1. Leguminous trees and soil P fractions
The total P concentrations extracted from the soils were low in all
sites, with the average proportion of residual-P N 50% and HCl-Pi ≤ 1%,
thus revealing the characteristic of highly weathered tropical soils
(Gama-Rodrigues et al., 2014; Viana et al., 2018). In this soil environ-
ment, both leguminous trees and N-mineral fertilization positively af-
fect the concentrations of all P fractions under different climatic
conditions, production system and management practices of each forest
site. Regarding the control (non- leguminous trees) of each forest site,
the largest relative increase of total Po was observed at site 1 (mixed for-
est plantations), while total Pi had the largest increases at sites 2 (pure
leguminous tree plantations) and 3 (agroforestry systems). In the labile
P (resin-Pi + NaHCO3-Pi + NaHCO3-Po), the largest increases occurred

Sites Treatments NaOH-Na2EDTA extraction (mg kg−1) Table 6

Phosphorus compounds determined by solution 31P NMR spectroscopy in NaOH–
31 31 31
Pi NMR total Po NMR total P NMR total P NaOH–EDTA total Na2EDTA extracts of soils under different forest sites.
1 T1 183.8 (38) 83.9 (17) 267.7 (55) 293.0 (60)
Sites Treatments 31
Pi NMR total (mg kg−1) 31
Po NMR total (mg kg−1)
T4 216.6 (41) 170.2 (32) 386.8 (73) 423.4 (80)
2 T5 230.9 (44) 200.4 (39) 431.3 (83) 442.0 (85) ortho-P Pyrophosphate Monoester-P Diester-P
T6 219.9 (40) 179.3 (33) 399.2 (73) 434.4 (80)
DNA M/D
T7 172.9 (29) 78.1 (13) 251.0 (42) 272.2 (46)
T8 143.2 (27) 57.2 (11) 200.4 (37) 211.4 (40) 1 T1 151.3 32.5 77.7 6.2 12.6
3 T10 244.5 (47) 156.4 (30) 400.9 (76) 438.9 (84) T4 185.2 31.4 160.3 9.9 16.2
T12 275.2 (41) 193.8 (29) 469.0 (70) 558.1 (83) 2 T5 192.4 38.5 191.9 8.5 22.7
T13 191.7 (43) 72.5 (16) 264.2 (59) 289.1 (65) T6 190.3 29.6 172.1 7.2 23.7
T7 153.4 19.5 74.0 4.1 17.9
Values in parentheses are the proportion (%) of the total soil P (H2SO4 + H2O2 digestion).
31 T8 124.4 18.8 52.6 4.6 11.5
Pi NMR total: sum of ortho-P and pyrophosphate. 31Po NMR total: sum of monoester-P and
3 T10 208.4 36.1 142.9 13.5 10.6
diester-P. 31PNMR total: sum of 31Pi NMR total and 31Po NMR total. PNaOH–EDTA total: total P-ex-
T12 237.6 37.6 176.8 17.0 10.4
tractable in NaOH-Na2EDTA and analyzed by molybdate reactive colorimetry. Site 1 –
T13 170.1 21.6 67.3 5.2 12.9
mixed forest plantations: (T1) 0% leguminous trees, (T2) 25% leguminous trees, (T3)
50% leguminous trees and (T4) 75% leguminous trees cover; Site 2 – pure plantations of Site 1 – mixed forest plantations: (T1) 0% leguminous trees and (T4) 75% leguminous
leguminous trees: (T5) Acacia auriculiformis, (T6) Mimosa caesalpiniifolia, (T7) secondary trees; Site 2 – pure plantations of leguminous trees: (T5) Acacia auriculiformis, (T6) Mi-
forest and (T8) pasture; Site 3 – agroforestry systems: (T9) Acácia angustíssima, (T10) mosa caesalpiniifolia, (T7) secondary forest and (T8) pasture; Site 3 – agroforestry systems:
Gliricidia sepium, (T11) Pueraria phaseoloides, (T12) spontaneous vegetation with low (T10) Gliricidia sepium, (T12) spontaneous vegetation with low input of nitrogen fertilizer
input of nitrogen fertilizer and (T13) spontaneous vegetation without nitrogen fertilizer. and (T13) spontaneous vegetation without nitrogen fertilizer. See detailed description of
See detailed description of forest sites in the Materials and methods section and Table S1. forest sites in the Materials and methods section and Table S1.
 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405
a Legume
200 = +29.6%
b
150
Concentration (mg kg-1)
100
50
0
ortho-P
Fig. 7. Phosphorus compounds determined by solution 31P NMR spectroscopy in NaOH–Na2EDTA extracts of soils under different forest sites. Legume: mean of (T4) 75% leguminous trees,
(T5) Acacia auriculiformis, (T6) Mimosa caesalpiniifolia and (T10) Gliricidia sepium. Control: mean of (T1) 0% leguminous trees, (T7) secondary forest and (T8) pasture and (T13)
spontaneous vegetation without nitrogen fertilizer. Values followed by the same letter are not significantly different among treatments by “t-test” (p = 0.05). See detailed description
of forest sites in the Materials and methods section and Table S1.
at site 2, followed by sites 3 and 1 (Figs. 1, 2 and 3). In this fraction, Pi
forms were more responsive than Po.
An increase of over 50% in total soil N only occurred at site 1, sug-
gesting a strong influence of the treatment with 75% density of N-
fixing leguminous trees (T4) (Table 2). The maintenance of total N con-
centrations at sites 2 and 3 does not indicate that there were no changes
in the various forms of N, especially the increase in the labile organic N
−
(potentially mineralizable) and mineral N (NH+ 4 + NO3 ), which con-
tribute to the increase of N availability in the soil. Thus, there would
be conditions for the different plant and microorganism species to in-
crease the production of extracellular phosphatase enzymes, which
are rich in N (~15%; Treseder and Vitousek, 2001) present in the rhizo-
sphere and bulk soil. There would be an increase in Po mineralization
under these soil environmental conditions and consequently an in-
crease in the availability of P, as indicated by the increase in the concen-
trations of the labile fractions of resin-Pi and NaHCO3-Pi (Fig. 1, Table 3).
Thus, the positive correlations of N with the resin-Pi, NaHCO3-Pi,
NaHCO3-Po (all with r = 0.99, p b 0.01, n = 4) and NaOH-Pi (r = 0.96,
p b 0.05, n = 4) fractions at site 1 evidenced that there would be close
association between the dynamics of N and P in forest systems with
N-fixing leguminous trees.
The pure plantation of Acacia auriculiformis (T5) and Mimosa
caesalpinifolia (T6) and the Gliricidia sepium in hedgerows (T10) at
sites 2 and 3, respectively, showed a distinction of the ability of the legu-
minous trees to positively and significantly alter the P fractions in the
soil, whereas the hedgerows of Acacia angustissima (T9) and the
Pueraria phaseoloides (T11) forage legume did not provide significant al-
terations of these fractions at site 2 (Table 3). In turn, the mixed forest
plantation with different leguminous trees showed the potential of
this production system to positively affect P fractions as a function of
Table 7
Correlation coefficients between soil P compounds (0–10 cm) under different forest sites
(n = 9).
ortho-P Pyrophosphate
Pyrophosphate 0.78⁎ – –
Monoester-P 0.85⁎⁎ 0.83⁎⁎
DNA 0.90⁎⁎⁎ 0.81⁎⁎
⁎ Significant at p b 0.05.
⁎⁎ Significant at p b 0.01.
⁎⁎⁎ Significant at p b 0.001.
11
Control
a
= +145.7%
a = +46.8%
b = +94.5%
a
b
Pyrophosphate Monoester-P DNA
the planting density of these species (Site 1) (Fig. 1), but without
distinguishing the effect of each species on the P dynamics in the soil.
At site 3, the N mineral fertilization (T12) promoted the greatest in-
creases (+72.8%) in the fraction of labile Pi (Resin-Pi + NaHCO3-Pi) in
relation to spontaneous vegetation (T13 - control) (Fig. 3). This increase
in P availability may be due to a higher production of extracellular phos-
phatases in response to the addition of N (Olander and Vitousek, 2000;
Tresedear and Vitousek, 2001; Marklein and Houlton, 2012). However,
phosphatase activity in plant roots tends to respond more strongly to
N fertilization than soil microbial biomass phosphatase (Marklein and
Houlton, 2012), showing a clear distinction in the magnitude of action
between plants and microorganisms on the availability of P in tropical
soils with a high weathering degree. Olander and Vitousek (2000) re-
ported that N fertilization increased the levels of extractable P (Bray-1
P), but reduced the adsorption strength of forest soils in Hawaii. On
the other hand, elevated N fertilization plus atmospheric N deposition
significantly decreased soil labile P fractions (NaHCO3-Pi and NaHCO3-
Po) in the Acacia auriculiformis plantation, but they did not change in
the Eucalyptus urophylla plantation on acrisol in subtropical China
(Chen et al., 2018).
Although there were no significant changes in soil carbon concentra-
tions, which varied from medium to high at all evaluated sites (Table 2),
significant positive correlations between the C and the labile fractions of
Pi (resin-Pi and NaHCO3-Pi) suggest that P lability would be affected by
soil organic matter. The N input modifies the soil organic matter quality
(Malhi et al., 2006). In this sense, as the evaluated forest systems
showed a high production and conservation of soil organic matter
(Costa et al., 2014), it is probable that there has been an increase in
the production of reactive organic acids that would compete with phos-
phate for the specific adsorption sites, and also act as organic ligands
(Sato and Comerford, 2006), and thus displacing a part of the P adsorbed
by the mineral phase to the soil solution. Concurrent with the role of or-
ganic acids in P availability in the soils of the present paper, the contin-
uous transfer of P to the soil via litterfall by leguminous trees and other
Monoester-P species that compose the evaluated forest systems (Costa et al., 2014)
may also contribute to increase the bioavailability of P. The addition of
– P-rich residues significantly increased the resin-Pi, NaHCO3-Pi and
0.75⁎ NaOH-Pi fraction concentrations in highly weathered tropical soils
under various agroforestry systems (Nziguheba et al., 1998; Lehmann
et al., 2001). In this context, there should be no change in the total P
amount of the soil, but only its recycling rate within the assessed forest
12 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405
systems. We expected that there would be the maintenance or a de-
crease of the HCl-Pi and residue-P (insoluble P) fractions and the subse-
quent transformation of P to the fractions of medium and high lability of
P. Thus, it is plausible that the input of N provided by the use of legumi-
nous trees or N fertilization favored the root system of the trees to ab-
sorb P from the subsoil and to enrich the topsoil through the
contribution of P-rich residues. Helfenstein et al. (2018), using sequen-
tial extraction together with isotopic analyses, showed that P in the HCl
fraction has been biologically recycled from B horizon and redistributed
in different forms (organic, adsorbed, and re-precipitated) in pasture-
covered soil in Hawaii.
As all the leguminous trees in this study were inoculated with
arbuscular mycorrhizal fungi (AMF), and assuming the persistence
and effectiveness of this symbiosis and the ability of the hyphae to pro-
duce extracellular phosphatases as well as plants (Turner, 2008), which
allows AMF to also use Po as the only source of P, there would then be a
greater absorption capacity of Pi from the soil solution by the symbiotic
association plant-AMF. In future studies the dominant mechanisms of P
acquisition (plant or AMF derived phosphatase) in response to the addi-
tion of N (biological or mineral) should be investigated. In the case of
AMF, a high fraction of the net C fixed by the plants is required for fungal
growth, although phosphatase production represents a substantial en-
ergy cost for plants in terms of N. For this reason, Treseder and
Vitousek (2001) suggested that the alternation between these two
mechanisms might be more efficient in tropical ecosystems developed
on strongly-weathered soils. As a result, and considering the double in-
oculation (diazotrophic bacteria + AMF) of the leguminous trees used
in the present paper, it can be hypothesized that a double symbiosis
(N-fixation + P-AMF) would predominate in the initial phase of plant
growth, in which the AMF would be responsible for expanding the
root system in order to increase the soil P acquisition, and with this to
enable maximizing N-fixation; while in the mature phase of develop-
ment the production of extracellular phosphatases would predominate
due to the greater N availability, thus reducing the cost of fixed C allo-
cated in the symbiosis with N-fixers. In this context, tropical tree species
with different mycorrhizal associations can produce specific phospha-
tase enzymes to explore different Po compounds (Turner, 2008;
Steidinger et al., 2015), thus reducing competition for soil P in agrofor-
estry systems and mixed forest plantations on tropical strongly weath-
ered soils.
In the present paper, the increase in the Pi fraction concentrations
was not at the cost of the Po fractions (NaHCO3-Po, NaOH-Po and
sonic-Po), in which there were significant increases or statistically did
not differ from the reference coverage, except of the NaOH-Po and
sonic-Po fractions of the soil in the treatment with Acacia auriculiformis
(T5) at site 2 (Table 3). In other words, the increase or maintenance of
the soil Po fractions concentrations suggests that the amount of labile
Pi would exceed the demand of the plants and microorganisms in the
late development phase of the evaluated forest systems; which would
result in negative feedback on the production of phosphatase and main-
tenance of soil P levels near the limit level to reduce the production of
this enzyme. Thus, the increase and/or maintenance of the Po concentra-
tions in the soils of the treatments of the present paper, especially of la-
bile P, indicate that the transfer rate of Po to the soil via litterfall should
be higher than its mineralization. In this context, the amount of the la-
bile Pi fraction at site 2 in the treatments with Acacia auriculiformis
(T5) (13 kg/ha) and Mimosa caesalpinifolia (T6) (11 kg/ha) was 3–5
times higher than the annual amount of P required by these species
based on the amount of P (2.6 to 3.7 kg/ha/year) transferred to the
soil via litterfall production (Costa et al., 2014), while in the secondary
forest (T7) the amount of labile Pi was 6.5 times greater than the annual
amount of P required (Table 3). This is corroborated by Vincent et al.
(2010) to report that the removal of litter in tropical forest soils in
Panama reduced the Po concentration in the surface layer of these
soils. The results found in the present paper showed that the balance be-
tween accumulation and mineralization of Po in the soil varied between
the forest sites. On the other hand, the results shown do not corroborate
the hypothesis proposed by Olander and Vitousek (2000) based on the
model of McGill and Cole (1981), in which the addition of N via fertili-
zation or biological N-fixation would exhaust the Po pool over time by
increasing the mineralization and the availability of P in the soil, since
these mechanisms are strongly coupled. Similarly, Fan et al. (2018) re-
ported a reduction in the NaOH-Po fraction concentration, but not for
the resin-Pi and NaHCO3-Po fractions in a latosol under nitrogen fertili-
zation in a subtropical forest ecosystem in China. In turn, Koutika et al.
(2016) reported decreased concentrations of the resin-Pi and NaHCO3-
Po fractions in sandy soils under mixed plantations of Acacia mangium
and Eucalyptus urophylla × grandis hybrid in the Republic of Congo.
4.2. Leguminous trees and soil P compounds
The use of the NaOH–Na2EDTA extraction and solution 31P NMR
spectroscopy for determining and identifying the P compounds in the
soil corroborated the results obtained with the sequential fractionation
of P in the soils of all evaluated forest sites. The correlations between
ortho-P and DNA with the resin-Pi, NaHCO3-Pi and NaOH-Pi fractions
suggest that both methods used acted in the same soil P pool
(Table 4). The positive effect of leguminous trees and N fertilization
was distinct among P forms, in which Po compounds were more respon-
sive than Pi compounds in all forest sites (Fig. 7). Among the Po com-
pounds, monoester-P showed the highest average rate of relative
increase. These results showed the potential of leguminous trees to in-
crease the reserves of soil P compounds (ortho-P, pyrophosphate,
monoester-P, and DNA) in the first eleven years of growth at pure plan-
tations, mixed plantations or hedgerow (Table 6, Fig. 7). This potential
was well evidenced by the increase of P reserves in the pure Acacia
auriculiformis (T5) and Mimosa caesalpinifolia (T6) plantations in rela-
tion both to the secondary forest (T7) and the degraded pasture (T8)
in site 2 (Table 6). The concentration values of ortho-P, pyrophosphate
and monoester-P in all evaluated forest sites were much higher than
those reported by Turner and Engelbrecht (2011) in oxisols and ultisols
under forest ecosystems in Panama, except for the DNA in which the
concentrations were similar.
The accumulation of monoester-P and DNA in evaluated forest soils
depends on a set of abiotic stabilization processes. Adsorption in the
mineral soil is the main process that affects the selective accumulation
of some compounds (Celi and Barberis, 2005), mainly in soils at
pH b 5.0, which protects them from enzymatic attack (Condron et al.,
2005). The complexation and precipitation with polyvalent cations of
these compounds of Po and the incorporation within the structures of
the soil organic matter are other conservation processes of the Po in ter-
restrial ecosystems (Celi and Barberis, 2005; McLaren et al., 2015).
Overall, the results displayed by the strong correlation between the
ortho-P and pyrophosphate with DNA evidence this would be the main
source of P for the readily-available Pi pool (Table 7). The high values of
the monoester-P/diester-P ratio (Table 6), as an index of the lability de-
gree of Po compounds in the soil, and the close correlation between DNA
and Pi fractions of different lability levels (resin, NaHCO3 and NaOH ex-
tracts) reinforce this hypothesis (Table 4). The pool of diester-P consti-
tutes labile Po (DNA) forms due to their rapid turnover (Bowman and
Cole, 1978; Morrissey et al., 2015). Thus, the enrichment of the organic
pool would be predominantly monoester-P, evidenced by its higher rel-
ative increasing rates (Fig. 7), as previously discussed. The close interre-
lationships between P compounds suggest that the transformation of
the monoesters-P and diester-P to pyrophosphate may be due to high
fungal activity of the soil (Koukol et al., 2006, 2008). In turn, since pyro-
phosphate has biological origin, its transformation to ortho-P requires
hydrolysis by the phosphatase enzymes, since this compound can be
considered functionally similar to Po (Turner and Engelbrecht, 2011;
Reitzel and Turner, 2014) at times of soil P limitation.
Our results evidenced that the magnitude of the increments of the P
compounds varied with the environmental conditions (climate and
 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405
soil), age system, and leguminous trees of each forest site. Thus, further
research is needed to highlight the theoretical plausibility of the com-
plex interconnection between all P species determined by solution 31P
NMR spectroscopy, considering the specificities of leguminous tree spe-
cies at each forest site evaluated. For this, the construction of a structural
model of the soil P cycle with inputs (P-litterfall) and outputs (plant up-
take) would make it possible to measure the magnitude of P transfor-
mations caused by leguminous trees present in a given forest system.
5. Conclusion
The integrated use of solution 31P NMR spectroscopy and sequential
P extraction improved our understanding of inorganic (Pi) and organic
phosphorus (Po) transformations in forest systems with N-fixing legu-
minous trees in tropical soils. The P fractionation method displayed
the potential of leguminous trees, inoculated with diazotrophic bacteria
and arbuscular mycorrhizal fungi, to increase P concentrations in the
soil through cycling and accumulation of Pi and Po fractions at different
degree of availability to plants (labile P, moderately labile P and oc-
cluded P). In turn, the use of 31P NMR spectroscopy clearly identified
and quantified that a large amount of the Po species found in the soil
consisted of phosphomonoesters followed by phosphodiesters in the
form of DNA, as well as high reserves of Pi species (ortho-P and pyro-
phosphate) in the first eleven years of growth at pure plantations,
mixed plantations or agroforests. This study highlighted the importance
of the use of different N-fixing leguminous trees with the purpose of re-
covering heavily degraded soils, considering the environmental condi-
tions (climate and soil), production systems and management
practices. This may be a suitable strategy for the efficient management
of P cycling in highly weathered tropical soils in the Brazilian Atlantic
Forest biome.
Declaration of competing interest
The authors declare no conflict of interest.
Acknowledgments
We thank the Fundação Carlos Chagas Filho de Amparo à Pesquisa
do Estado do Rio de Janeiro (FAPERJ) for providing a PhD scholarship
to the first author – (Grant E-26/100.611/2014), the Conselho Nacional
de Desenvolvimento Científico e Tecnológico (CNPq) – (Grant 2013/
475222-0, Projeto Universal), and the Coordenação de Aperfeiçoamento
de Pessoal de Nível Superior (CAPES) – (Finance Code 001,
88882.14549/2019-01) for it financial support. We would also like to
thank the Embrapa Agrobiologia for it technical support. The authors
tank the anonymous reviewers for their comments and suggestions on
the manuscript. The authors declare no conflict of interest. Any use of
trade, firm, or product names is for descriptive purposes only.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.scitotenv.2019.136405.
References
Aleixo, S., Gama-Rodrigues, A.C., Costa, M.G., Sales, M.V.S., Gama-Rodrigues, E.F., Marques,
J.R.B., 2017. P transformations in cacao agroforests soils in the Atlantic Forest region
of Bahia, Brazil. Agrofor. Syst. 91 (3), 423–437. https://doi.org/10.1007/s10457-016-
9939-6.
Aleixo, S., Gama-Rodrigues, A.C., Gama-Rodrigues, E.F., Schripsema, J., 2019. Organic phos-
phorus of soils under cacao agroforests in the Atlantic coast of Brazil. Geoderma Re-
gional 17, e00220. https://doi.org/10.1016/j.geodrs.2019.e00220.
Batterman, S.A., Hall, J.S., Turner, B.L., Hedin, L.O., Walter, J.K.L., Sheldon, P., van Breugel,
M., 2018. Phosphatase activity and nitrogen fixation reflect species differences, not
nutrient trading or nutrient balance, across tropical rainforest trees. Ecol. Lett. 21
(10), 1486–1495. https://doi.org/10.1111/ele.13129.
13
Bianchi, M.O., Scoriza, R.N., Resende, A.S., Campello, E.F.C., Correia, M.E.F., Silva, E.M.R.,
2017. Macrofauna edáfica como indicadora em revegetação com leguminosas
arbóreas. Floresta e Ambiente 24, e00085714. https://doi.org/10.1590/2179-
8087.085714.
Bowman, R.A., Cole, C.V., 1978. Transformations of organic phosphorus substrates in soils
as evaluated by NaHCO3 extraction. Soil Sci. 125, 49–54. https://doi.org/10.1097/
00010694-197801000-00008.
Cade-Menun, B., 2015. Improved peak identification in 31P-NMR spectra of environmental
samples with a standardized method and peak library. Geoderma 257–258, 102–114.
https://doi.org/10.1016/j.geoderma.2014.12.016.
Cade-Menun, B., Liu, C.W., 2014. Solution 31P nuclear magnetic resonance spectroscopy of
soils from 2005 to 2013: a review of sample preparation and experimental parame-
ters. Soil Sci. Soc. Am. J. 78, 19–37. https://doi.org/10.2136/sssaj2013.05.0187dgs.
Celi, L., Barberis, E., 2005. Abiotic stabilization of organic phosphorus in the environment. In:
Turner, B.L., Frossard, E., Baldwin, D.S. (Eds.), Organic Phosphorus in the Environment.
CABI, Wallingford, pp. 113–132. https://doi.org/10.1079/9780851998220.0000.
Chaer, G.M., Resende, A.S., Campello, A.F.C., Faria, S.M., Boddey, R.M., 2011. Nitrogen-
fixing legume tree species for the reclamation of severely degraded lands in Brazil.
Tree Physiol. 31 (2), 139–149. https://doi.org/10.1093/treephys/tpq116.
Chen, H., Chen, M., Li, D., Mao, Q., Zhang, W., Mo, J., 2018. Responses of soil phosphorus
availability to nitrogen addition in a legume and a non-legume plantation. Geoderma
322, 12–18. https://doi.org/10.1016/j.geoderma.2018.02.017.
Condron, L.M., Turner, B.L., Cade-Menun, B., 2005. Chemistry and dynamics of soil organic
phosphorus. In: Sharpley, A.N., Sims, J.T. (Eds.), Phosphorus: Agriculture and the En-
vironment. American Society of Agronomy, Madison WI, pp. 87–121. https://doi.org/
10.2134/agronmonogr46.c4.
Costa, M.G., Gama-Rodrigues, A.C., Zaia, F.C., Gama-Rodrigues, E.F., 2014. Leguminosas
arbóreas para recuperação de áreas degradadas com pastagem em Conceição de
Macabu, Rio de Janeiro, Brasil. Scientia Forestalis. 42 (101), 101–112. http://www.
ipef.br/publicacoes/scientia/nr101/cap10.pdf.
Dick, W.A., Tabatabai, M.A., 1977. Determination of orthophosphate in aqueous solutions
containing labile organic and inorganic phosphorus compounds. J. Environ. Qual. 6,
82–85. https://doi.org/10.2134/jeq1977.00472425000600010018x.
Doolette, A.L., Smernik, R.J., Dougherty, W.J., 2009. Spiking improved solution
phosphorus-31 nuclear magnetic resonance identification of soil phosphorus com-
pounds. Soil Sci. Soc. Am. J. 73, 919–927. https://doi.org/10.2136/sssaj2008.0192.
Doolette, A.L., Smernik, R.J., Dougherty, W.J., 2011. A quantitative assessment of phospho-
rus forms in some Australian soils. Soil Research 49, 152–165. https://doi.org/
10.1071/SR10092.
Embrapa - Empresa Brasileira de Pesquisa Agropecuária, 2011. Manual de análise química
de solos, plantas e fertilizantes. Embrapa solos, Brasília.
Embrapa - Empresa Brasileira de Pesquisa Agropecuária, 2013. Sistema brasileiro de
classificação de solos. Embrapa solos, Brasília.
Fan, Y., Lin, F., Yang, L., 2018. Decreased soil organic P fraction associated with
ectomycorrhizal fungal activity to meet increased P demand under N application in
a subtropical forest ecosystem. Biol. Fertil. Soils 54, 149–161. https://doi.org/
10.1007/s00374-017-1251-8.
Fisher, J.B., Badgley, G., Blyth, E., 2012. Global nutrient limitation in terrestrial vegetation.
Glob. Biogeochem. Cycles 26. https://doi.org/10.1029/2011GB004252.
Gama-Rodrigues, E.F., Barros, N.F., Viana, A.P., Santos, G.A., 2008. Alterações na biomassa e
na atividade microbiana da serapilheira e do solo, em decorrência da substituição de
cobertura florestal nativa por plantações de eucalipto, em diferentes sítios da Região
Sudeste do Brasil. Revista Brasileira de Ciência do Solo 32 (4), 1489–1499. https://doi.
org/10.1590/S0100-06832008000400013.
Gama-Rodrigues, A.C., Sales, M.V.S., Silva, P.S.D., Comerford, N.B., Cropper, W.P., Gama-
Rodrigues, E.F., 2014. An exploratory analysis of phosphorus transformations in trop-
ical soils using structural equation modeling. Biogeochemistry 118, 453–469. https://
doi.org/10.1007/s10533-013-9946-x.
He, Z.Q., Olk, D.C., Cade-Menun, B.J., 2011. Forms and lability of phosphorus in humic acid
fractions of Hord silt loam soil. Soil Sci. Soc. Am. J. 75, 1712–1722. https://doi.org/
10.2136/sssaj2010.0355.
Hedley, M.J., Stewart, J.W.B., Chauhan, B.S., 1982. Changes in inorganic and organic soil
phosphorus fractions induced by cultivation practices and by laboratory incubations.
Soil Sci. Soc. Am. J. 46, 970–976. https://doi.org/10.2136/
sssaj1982.03615995004600050017x.
Helfenstein, J., Tamburini, F., von Sperber, C., Massey, M.S., Pistocchi, C., Chadwick, O.A.,
Vitousek, P.M., Kretzschmar, R., Frossard, E., 2018. Combining spectroscopic and iso-
topic techniques gives a dynamic view of phosphorus cycling in soil. Nat. Commun.
https://doi.org/10.1038/s41467-018-05731-2.
Houlton, B.Z., Wang, Y.P., Vitousek, P., Field, C., 2008. A unifying framework for dinitrogen
fixation in the terrestrial biosphere. Nature 454, 327–330. https://doi.org/10.1038/
nature07028.
IBGE - Instituto Brasileiro de Geografia e Estatística, 2012. Manual técnico da vegetação
brasileira. Instituto Brasileiro de Geografia e Estatística, Rio de Janeiro, Brazil.
Koukol, O., Novák, F., Hrabal, R., Vosátka, M., 2006. Saprrotrophic fungi transform organic
phosphorus in spruce needle litter. Soil Biol. Biochem. 38, 3372–3379. https://doi.org/
10.1016/j.soilbio.2006.05.007.
Koukol, O., Novák, F., Hrabal, R., 2008. Composition of the organic phosphorus fraction in
basidiocarps of saprotrophic and mycorrhizal fungi. Soil Biol. Biochem. 40,
2464–2467. https://doi.org/10.1016/j.soilbio.2008.04.021.
Koutika, L.S., Mareschal, L., Epron, D., 2016. Soil P availability under eucalypt and acacia on
Ferralic Arenosols, republic of the Congo. Geoderma Regional 7 (2), 153–158. https://
doi.org/10.1016/j.geodrs.2016.03.001.
Lambers, H., Raven, J.A., Shaver, G.R., Smith, S.E., 2008. Plant nutrient-acquisition strate-
gies change with soil age. Trends in Ecology and Evolution 23, 95–103. https://doi.
org/10.1016/j.tree.2007.10.008.
14 S. Aleixo et al. / Science of the Total Environment 712 (2020) 136405
Lehmann, J., Günther, D., Socorro, Da Mota, M., Pereira De Almeida, M., Zech, W., Kaiser, K.,
2001. Inorganic and organic soil phosphorus and sulfur pools in an Amazonian
multistrata agroforestry system. Agrofor. Syst. 53 (2), 113–124. https://doi.org/
10.1023/A:1013364201542.
Malhi, S.S., Lemke, R., Wang, Z.H., Chhabra, B.S., 2006. Tillage, nitrogen and crop residue
effects on crop yield, nutrient uptake, soil quality, and greenhouse gas emissions.
Soil Tillage Res. 90 (1–2), 171–183. https://doi.org/10.1016/j.still.2005.09.001.
Manhães, C.M.C., Gama-Rodrigues, E.F., Moço, M.K.S., Gama-Rodrigues, A.C., 2013. Meso-
and macrofauna in the soil and litter of leguminous trees in a degraded pasture in
Brazil. Agrofor. Syst. 87 (5), 993–1004. https://doi.org/10.1007/s10457-013-9614-0.
Marklein, A.R., Houlton, B.Z., 2012. Nitrogen inputs accelerate phosphorus cycling rates
across a wide variety of terrestrial ecosystems. New Phytol. 193, 696–704. https://
doi.org/10.1111/j.1469-8137.2011.03967.x.
McDowell, R.W., Stewart, I., Cade-Menun, B.J., 2006. An examination of spin-lattice relax-
ation times for analysis of soil and manure extracts by liquid state phosphorus-31 nu-
clear magnetic resonance spectroscopy. J. Environ. Qual. 35, 293–302. https://doi.org/
10.2134/jeq2005.0285.
McGill, W.B., Cole, C.V., 1981. Comparative aspects of cycling of organic C, N, S and P
through soil organic matter. Geoderma 26, 267–268. https://doi.org/10.1016/0016-
7061(81)90024-0.
McLaren, T.I., Smernik, R.J., McLaughlin, M.J., McBeath, T.M., Kirby, J.K., Simpson, R.J.,
Guppy, C.N., Doolette, A.L., Richardson, A.E., 2015. Complex forms of soil organic
phosphorus–a major component of soil phosphorus. Environ. Sci. Technol. 49,
13238–13245. https://doi.org/10.1021/acs.est.5b02948.
Morissey, E.M., Mchugh, T.A., Preteska, L., Hayer, M., Dijkstra, P., Hungate, B.A., Schwartz,
E., 2015. Dynamics of extracellular DNA decomposition and bacterial community
composition in soil. Soil Bio. Biochem. 86, 42–49. https://doi.org/10.1016/j.
soilbio.2015.03.020.
Murphy, J., Riley, J.P., 1962. A modified single solution method for the determination of
phosphate in natural waters. Analytical Chemistry Acta 27, 31–36. https://doi.org/
10.1016/S0003-2670(00)88444-5.
Nasto, M.K., Alvarez-Clare, S., Lekberg, Y., Sullivan, B.W., Townsend, A.R., Cleveland, C.C.,
2014. Interactions among nitrogen fixation and soil phosphorus acquisition strategies
in lowland tropical rain forests. Ecol. Lett. 17, 1282–1289. https://doi.org/10.1111/
ele.12335.
Nasto, M.K., Osborne, B.B., Lekberg, Y., Asner, G.P., Balzotti, C.S., Porder, S., Taylor, P.G.,
Townsend, A.R., Cleveland, C.C., 2017. Nutrient acquisition, soil phosphorus
partitioning and competition among trees in a lowland tropical rain forest. New
Phytol. 214, 1506–1517. https://doi.org/10.1111/nph.14494.
Nunes, D.A.D., Gama-Rodrigues, E.F., Barreto, P.A.B., Gama-Rodrigues, A.C., Monroe,
P.H.M., 2016. Carbon and nitrogen mineralization in soil of leguminous trees in a de-
graded pasture in northern Rio de Janeiro, Brazil. J. For. Res. 27, 91–99. https://doi.
org/10.1007/s11676-015-0164-3.
Nziguheba, G., Palm, C.A., Buresh, R.J., Smithson, P.C., 1998. Soil phosphorus fractions and
adsorption as affected by organic and inorganic sources. ant and Soil 198, 159–168.
https://doi.org/10.1023/A:1004389704235.
Olander, L.P., Vitousek, P.M., 2000. Regulation of soil phosphatase and chitinase activity by
N and P availability. Biogeochemistry 49 (2), 175–191. https://doi.org/10.1023/A:
1006316117817.
Png, G.K., Turner, B.L., Albornoz, F.E., Hayes, P.E., Lambers, H., Laliberté, E., 2017. Greater
root phosphatase activity in nitrogen-fixing rhizobial but not actinorhizal plants
with declining phosphorus availability. J. Ecol. 105 (5), 1246–1255. https://doi.org/
10.1111/1365-2745.12758.
Quesada, C.A., Lloyd, J., Schwarz, M., Patiño, S., Baker, T.R., Czimczik, C., Fyllas, N.M.,
Martinelli, L., Nardoto, G.B., Schmerler, 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., Paiva, R., 2010. Variations in chemical
and physical properties of Amazon forest soils in relation to their genesis. Biogeosci-
ences 7, 1515–1541. https://doi.org/10.5194/bg-7-1515-2010.
Reitzel, K., Turner, B.L., 2014. Quantification of pyrophosphate in soil solution by
pyrophosphatase hydrolysis. Soil Biol. Biochem. 74, 95–97. https://doi.org/10.1016/
j.soilbio.2014.03.001.
Sato, S., Comerford, N.B., 2006. Organic anions and phosphorus desorption and bioavail-
ability in a humid brazilian ultisol. Soil Sci. 171, 695–705. https://doi.org/10.1097/
01.ss.0000228043.10765.79.
Schneider, K.D., Cade-Menun, B.J., Lynch, D.H., Voroney, R.P., 2016. Soil phosphorus form
organic and conventional forage fields. Soil Sci. Soc. Am. J. 80, 328–340. https://doi.
org/10.2136/sssaj2015.09.0340.
Soil Survey Staff, 2014. Soil survey field and laboratory methods manual. In: Burt, R. (Ed.),
Soil Survey Investigations Report n51 Version 1.0. U.S. Department of Agriculture,
Natural Resources Conservation Service.
Soper, F., Nasto, M., Osborne, B.B., Cleveland, C.C., 2019. Nitrogen fixation and foliar nitro-
gen do not predict phosphorus acquisition strategies in tropical trees. Journal Ecology
107, 118–126. https://doi.org/10.1111/1365-2745.13044.
SOS Mata Atlântica, 2018. Atlas dos remanescentes florestais da Mata Atlântica
(2016–2017). Relatório Técnico, São Paulo, Brazil https://www.sosma.org.br/link/
Atlas_Mata_Atlantica_2016-2017_relatorio_tecnico_2018_final.pdf, Accessed date:
16 August 2018.
Steidinger, B.S., Turner, B.L., Corrales, A., Dalling, J.W., 2015. Variability in potential to ex-
ploit different soil organic phosphorus compounds among tropical montane tree spe-
cies. Funct. Ecol. 29, 121–130. https://doi.org/10.1111/1365-2435.12325.
Treseder, K.K., Vitousek, P.M., 2001. Effects of soil nutrient availability on investment in
acquisition of N and P in Hawaiian Rain Forests. Ecology 82, 946–954. https://doi.
org/10.1890/0012-9658(2001)082[0946:EOSNAO]2.0.CO;2.
Turner, B.L., 2008. Resource partitioning for soil phosphorus: a hypothesis. J. Ecol. 96,
698–702. https://doi.org/10.1111/j.1365-2745.2008.01384.x.
Turner, B.J., Engelbrecht, B.M.J., 2011. Soil organic phosphorus in lowland tropical rain for-
ests. Biogeochemistry 103, 297–315. https://doi.org/10.1007/s10533-010-9466-x.
Turner, B.L., Cheesman, A.W., Godage, H.Y., Riley, A.M., Potter, B.V.L., 2012. Determination
of neo- and D-chiro-inositol hexakisphosphate in soils by solution 31P NMR spectros-
copy. Environ. Sci. Technol. 46 (9), 4994–5002. https://doi.org/10.1021/es204446z.
Viana, T.O., Gama-Rodrigues, A.C., Gama-Rodrigues, E.F., Aleixo, S., Moreira, R.V.S., Sales,
M.V.S., Marques, J.R.B., 2018. Phosphorus transformations in alfisols and ultisols
under different land uses in the Atlantic forest region of Brazil. Geoderma Regional
14, e00184. https://doi.org/10.1016/j.geodrs.2018.e00184.
Vincent, A.G., Turner, B.L., Tanner, E.V.J., 2010. Soil organic phosphorus dynamics follow-
ing perturbation of litter cycling in a tropical moist forest. Eur. J. Soil Sci. 61, 48–57.
https://doi.org/10.1111/j.1365-2389.2009.01200.x.
Vitousek, P., Porder, S., Houlton, B.Z., Chadwick, O.A., 2010. Terrestrial phosphorus limita-
tion: mechanisms, implications, and nitrogen-phosphorus interactions. Ecol. Appl. 20,
5–15. https://doi.org/10.1890/08-0127.1.
Wang, Y.P., Houlton, B.Z., Field, C.B., 2007. A model of biogeochemical cycles of carbon, ni-
trogen and phosphorus including symbiotic nitrogen fixation and phosphatase pro-
duction. Glob. Biogeochem. Cycles 21. https://doi.org/10.1029/2006GB002797.
Young, E.O., Ross, D.S., Cade-Menun, B.J., Liu, C.W., 2013. Phosphorus speciation in riparian
soils: a phosphorus-31 nuclear magnetic resonance spectroscopy and enzyme hydro-
lysis study. Soil Sci. Soc. Am. J. 77 (5), 1636–1647. https://doi.org/10.2136/
sssaj2012.0313.