Geoderma 310 (2018) 238–248

Contents lists available at ScienceDirect

Geoderma
journal homepage: www.elsevier.com/locate/geoderma

Effect of conversion from sugarcane preharvest burning to residues green- MARK

trashing on SOC stocks and soil fertility status: Results from different soil
conditions in Brazil
E.F. Lucaa,⁎, V. Chaplotb,c, M. Mutemac, C. Fellerd, M.L. Ferreirae, C.C. Cerrif,†, H.T.Z. Coutog
a
Instituto Florestal de São Paulo (IF), Rua Profa. Ana C. M. Ferraz, 78 CEP 13412-018 Piracicaba, SP, Brazil
b
Institut de Recherche pour le Développement (IRD), Laboratoire d'Océanographie et du Climat: Expérimentations et approches numériques, UMR 7159, 4 place Jussieu,
75252 Paris Cedex 05, France
c
SAEES, Centre for Water Resources Research, University of KwaZulu-Natal, PB X01, Scottsville 3209, Pietermaritzburg, Republic of South Africa
d
Institut de Recherche pour le Développement (IRD), UMR Eco & Sols, SupAgro, bât. 12, 2 place Viala, 34060 Montpellier Cedex 2, France
e
Universidade Nove de Julho (UNINOVE), Av. Luiz Ignácio Anhaia Mello, 1363. Vila Prudente, CEP 03155-000 São Paulo, Brazil
f
Centro de Energia Nuclear na Agricultura (CENA/USP), Av. Centenário s/n, CEP 13400-000 Piracicaba, SP, Brazil
g
Escola Superior de Agricultura Luiz de Queiroz (ESALQ/USP), Av. Pádua Dias s/n, CEP 13400-000 Piracicaba, SP, Brazil

A R T I C L E I N F O A B S T R A C T

Editor: Bernd Marschner Land use and land management have a recognized impact on the cycle of elements such as carbon and nitrogen
Keywords: and are thus expected to play an important role in the mitigation against human-induced environmental changes
C sequestration such as global warming. Brazil is the world's largest producer of sugarcane and has in recent years, experienced a
Green-trashing shift from traditional cane preharvest burning (B) to mechanized green-trashing (GT), whose consequences on
Land degradation soil organic carbon and nutrients are still uncertain. This study performed in the São-Paulo state, Southeast
Soil aggregation Brazil, investigated the impact of shifting from B to GT on soil organic carbon and nitrogen stocks, and chemical
Soil fertility elements at three 4-year and one 12-year old trial sites. Soil samples were collected from four sites representing
different soil types (clayey to sandy-clay Hapludox; Hapudult and sandy Quartzpsamment). Sampling was up to
0.4 m (0–0.05; 0.05–0.1; 0.1–0.2; 0.2–0.4 m) depths at six randomly selected pits from six plots (of 0.2 ha size
each) per treatment arranged in randomized block design. The samples were analysed for carbon and nitrogen
content and other soil properties such as texture, bulk density, aggregate stability, pH and chemical con-
centrations. Overall, conversion from B to GT increased fresh organic matter retained on soil surface by
12.8–13.0 Mg ha− 1 y− 1, and soil organic carbon and nitrogen stocks in the top 0.4 m soil layer by
1.06 Mg C ha− 1 y− 1 and 0.11 Mg N ha− 1 y− 1, respectively. The highest carbon sequestration rate (108‰ y− 1)
occurred in the most carbon depleted soils, which were sandy soils with lowest proportion of macroaggregates
and chemical elements such as calcium, magnesium and potassium. All the chemical elements also showed a
general increase in concentration, especially within the topmost soil layer, because increased soil organic matter
had greater ability to tighten them together. Considering the average sequestration rate of 1.06 ton ha− 1 y− 1,
the resulting soil carbon sequestration rate for Brazil where 8.8 million hectares are under green-trashing of
sugarcane residues would be 9.3 million ton C y− 1. However, intermittent tillage every 4–5 years for replanting
poses limitations to carbon sequestration because a 12 year experiment which was sampled 1 year after re-
planting did not show any carbon sequestration, suggesting that all carbon stored between planting and re-
planting was lost at tillage. There is need for further research to quantify the dynamics of carbon stocks during
and immediately after tillage and to find sugarcane replanting techniques which minimize soil disturbance and
associated carbon losses.

Abbreviations: B, manual harvest of sugarcane (Saccharum officinarum L.) with prior residue burning; ECEC, effective cation exchange capacity; GT, mechanical harvest with residue
mulching of sugarcane (Saccharum officinarum L.); Hx, Typic Hapludox; Ht, Typic Hapudult; H++Al3 +, potential acidity; Qt, Quartzpsamment; SB, sum of exchangeable bases; SOCS, soil
organic carbon stocks (Mg ha− 1); SONS, soil organic nitrogen stocks (Mg ha− 1); V, base saturation (%); ρb, soil bulk density (Mg m− 3)
⁎
Corresponding author.
E-mail addresses: efluca@if.sp.gov.br (E.F. Luca), vincent.chaplot@ird.fr (V. Chaplot), macdexo@gmail.com (M. Mutema), christian.feller@ird.fr (C. Feller),
mauecologia@yahoo.com.br (M.L. Ferreira), htzcouto@usp.br (H.T.Z. Couto).
†
In memoriam.

http://dx.doi.org/10.1016/j.geoderma.2017.09.020
Received 18 November 2016; Received in revised form 4 September 2017; Accepted 11 September 2017
0016-7061/ © 2017 Published by Elsevier B.V.
E.F. Luca et al. Geoderma 310 (2018) 238–248

1. Introduction properties such as soil bulk density (ρb), pH, available phosphorus (P),
exchangeable cations (ECEC), and aggregate stability. Evaluating the
Soils store about 1500 Pg C in the top 1 m depth, which is three changes in SOC and SON stocks in conjunction with soil aggregation
times the amount of atmosphere carbon (C) estimated at 830 Pg C. The might yield important information on the controls of soil C and N se-
C pool in this soil reservoir has been depleted by land use changes and questration because soil structure and, especially, soil aggregation plays
land miss-uses; therefore, soil C sequestration via improved manage- an important role in soil carbon stabilization and destabilization me-
ment is increasingly seen as a credible solution to sequestrate atmo- chanisms (Chaplot and Cooper, 2015). The study was carried out at four
spheric C for mitigating against climate change (Paustian et al., 2016). farms showing different soil types but similar climates in the São Paulo
Yet, discrepancies still exist on the impact of crop residue management State, Southeast Brazil.
on soil C stocks and the main soil and environmental controls.
Brazil, with 8.8 million ha of sugarcane in 2014, is the largest su-
garcane producer of the world (CONAB, 2013). This crop is assumed to 2. Materials and methods
play an important role in the global C budget. Cerri et al. (2007) esti-
mated that sugarcane production in Brazil to produce 156 million m3 2.1. Site location and experimental design
hydrated alcohol and 71 million m3 anhydrous alcohol raised total at-
mospheric C by 0.172 Pg C during the 1975–2000 period. Brazilian Comparisons between two sugarcane trash management systems
farmers produce on average 4.5 metric tons ha− 1 year− 1 of dry matter involving manual harvest with prior residue burning (denoted B) and
sugarcane residues, which are mostly burnt. Assuming that 2.35 Mg of mechanical harvest with residue mulching referred as green-trashing
the residues contain 1 Mg C (Macedo, 1998), then it suggests Brazil is (GT) were carried out on sugarcane (Saccharum officinarum L.) planta-
contributing large amounts of C to the atmosphere each year from the tions at four sites in São Paulo state, Southeast Brazil (Fig. 1). One site,
sugar cane industry (Cerri et al., 2007). Elsewhere, a study by Dominy denoted Hx4, was a 4-year experiment located at São Martinho farm
et al. (2002) under sub-tropical humid conditions in South Africa de- (21° 22′ S, 48° 03′ W) in the Pradópolis area. Soil at this site was
monstrated that continuous sugarcane production, a common practice classified as clayey typic Hapludox (SSS, 1992), or Rhodic Ferralsol – FRr
worldwide, resulted in drastic soil C losses. These authors pointed to a (FAO, 1988). Two sites were located at Santa Luiza farm (21° 36′ S, 48°
decline of soil C content from 40 to 50 g C kg− 1 under natural grass- 22′ W) in the Matão area. One was a 4-year trial, while the other one
lands to 17–33 g C kg− 1 after 30 years of continuous sugarcane pro- was a 12-year experiment. The sites were denoted Ht4 and Hx12, re-
duction. A shift in management is, therefore, expected to help store spectively. The Ht4 site soil was classified as typic Hapudult (SSS, 1992),
back part of the lost C. or Haplic Acrisol – Ach (FAO, 1988); while the Hx12 soil was typic
The traditional practice of sugarcane harvesting consists of pre- Hapludox (SSS, 1992), or Haplic Ferralsol – FRh (FAO, 1988). The last
harvest burning of both dead and green leaves to facilitate manual site, denoted Qt4, was a 4-year experiment located at Da Pedra farm
harvesting of the stalks. These leaves constitute a large C pool which (21° 14′ S, 47° 36′ W) in the Serrana area. The soil there was classified
could be returned to the soil instead of being emitted to the atmosphere. as sandy Quartzpsamment (SSS, 1992), or Ferralic Arenosol – ARo (FAO,
Mechanized harvesting of the sugarcane stalks provides an opportunity 1988).
to abandon the practice of trash burning. Brazil began mechanized Soil textural characteristics for the top 0–040 m horizons are pre-
harvesting in the early 1950s (Ripoli and Ripoli, 2009). Currently, 84% sented in Table 1. Globally, soil texture was similar in B and GT
or 4.1 million ha of sugarcane produced in São Paulo state is green- treatments at a given site. Soils at Hx4 had greatest clay content
trashed (São Paulo, 2003) yet the consequences of green-trashing on averaging 687 and 660 g kg− 1 for the B and GT, respectively. Qt4 had
atmospheric C sequestration in soils are not determined. There are also lowest average clay content soils at 90 and 100 g kg− 1 for B and GT,
very few world studies on the impact of land use change in general respectively. Soil texture showed little variation with depth at all sites,
(Batjes, 1996) and green-trashing in particular on soil C stock replen- except for Qt4 where clay content increased dramatically from
ishment such as by Blair (2000) and Noble et al. (2003) in Australia, 60 g kg− 1 in the 0–0.05 m horizon to 100 g kg− 1 in the 0.05–0.10 m
Dominy et al. (2002) and Graham et al. (2002) in South Africa, and layer. The four sites were characterised by a tropical climate tropical
Ball-Coelho et al. (1993), Orlando Filho et al. (1998), Souza et al. with mean annual precipitation varying from 1470 to 1550 mm. Most
(2005), Galdos et al. (2009) and Pinheiro et al. (2010) in Brazil. of the precipitation comes between October and March each year. The
Pinheiro et al. (2010) showed that 14 years of green-trashing in São long-term mean annual temperature for São Paulo state is 23 °C (Pereira
Paulo State increased soil organic C stocks within the 0.1 m layer by et al., 2002).
0.93 tons C ha− 1 yr− 1. At the São Martinho farm, also in the São Paulo All the four sites were under conventional sugarcane cultivation for
State, Galdos et al. (2009) observed enhanced soil organic C stocks of at least 40 years before the start of the experiments. The conventional
0.15 tons C ha− 1 y− 1 in the 0–0.2 m layer after 8 years of green- sugarcane planting procedure involved heavy disking two to three
trashing. Noble et al. (2003), using a nine year experiment in the humid times followed by sub-soiling and light disking of the whole area.
wet tropics of north Queensland, showed that green-trashing increased Thereafter, 0.20 to 0.30 m deep furrows were opened for fertilizer ap-
soil C stocks by as much as 0.44 Mg C ha− 1 y− 1. While most published plication and planting of sugarcane pieces. The sugarcane pieces were
studies show that abandonment of preharvest residue burning (B) in planted at 1.4 m inter-row and 0.5 m in-row. Fertilizer application was
favour of green-trashing (GT) built up soil C, other studies, such as Ball- done once a year at a rate of 25 kg N ha− 1, 125 kg P2O5 ha− 1 and
Coelho et al. (1993), pointed to insignificant gains of soil C. There is 125 kg K2O kg ha− 1 at the time of planting, and 85 kg N ha− 1,
thus a need to better understand the impact of sugarcane trash man- 50 kg P2O5 ha− 1 and 100 kg K2O kg ha− 1 thereafter. Sugarcane har-
agement on soil C and other nutrient stocks such as soil nitrogen level. vesting was done each year during the driest months of winter.
Such an understanding will contribute to international initiatives Conventional harvesting was performed manually and was preceded by
seeking to develop land management practices fostering C sequestration burning of the trash (B). Mechanical harvesting (GT) was performed by
to soils. Amongst these initiatives is the ‘4 per mille Soils for Food Se- machine and the trash was left on the soil surface. The trash manage-
curity and Climate’ launched at the COP21 with as its goal to increase ment techniques (B and GT) were replicated 3 times at Hx4, Ht4 and
global soil organic C stocks by 4 per 1000 (or 0.4%) per year (Lal, 2015; Qt4, and only 2 times at Hx12. Each replicate plot size was 70 m long
Minasny et al., 2017; Chabbi et al., 2017; Dignac et al., 2017). and 29.4 m wide (Fig. 2) Soil and trash sampling was performed within
This study aimed at investigating the impact of abandoning pre- a 50 × 12.6 m2 space in each plot. The sugarcane plants were left to
harvest trash burning (B) in favour of green-trashing (GT) on soil or- grow for 4 to 5 years before replanting.
ganic C (SOCS) and nitrogen stocks (SONs) and on a series of other soil

239
E.F. Luca et al. Geoderma 310 (2018) 238–248

Fig. 1. Locations of municipalities in which the farms housing experimental sites (Pradópolis-São Martinho farm; Matão-Santa Luiza farm and Serrana-Da Pedra farm) were located in
São-Paulo State, Brazil.

Table 1
Soil clay, silt and sand content (g kg− 1) for different soil layers (0–0.05, 0.05–0.10, 0.10–0.20 and 0.20–0.40 m) under different trash management techniques (B: burning, and GT: green-
trashing) at the different experiment sites (Hx4 = 4-year trial, Ht4 = 4-year trial, Qt4 = 4-year trial and Hx12 = 12-year trial).

Site Soil type Soil Layer (m) B GT

SSS (1992) FAO (1988) clay silt sand clay silt sand

g kg− 1

Hx4 Typic Hapludox Rhodic Ferralsol (FRr) 0–0.05 710 130 160 660 120 220
0.05–0.1 680 140 180 660 140 200
0.1–0.2 670 150 180 650 160 190
0.2–0.4 nd nd nd 670 140 190
Mean 687 140 173 660 140 200
Ht4 Typic Hapudult Haplic Acrisol (Ach) 0–0.05 180 60 760 200 40 760
0.05–0.1 180 60 760 180 20 800
0.1–0.2 200 20 780 200 60 740
0.2–0.4 240 60 700 220 40 740
Mean 200 50 750 200 40 760
Qt4 Quartzpsamment Ferralic Arenosol 0–0.05 60 40 900 100 40 860
(ARo) 0.05–0.1 100 40 860 100 20 880
0.1–0.2 100 40 860 80 60 860
0.2–0.4 100 60 840 120 40 840
Mean 90 45 865 100 40 860
Hx12 Typic Hapludox Haplic Ferralsol (FRh) 0–0.05 320 80 600 320 60 620
0.05–0.1 340 60 600 340 60 600
0.1–0.2 320 80 600 360 40 600
0.2–0.4 360 60 580 340 60 600
Mean 335 70 595 340 55 605

240
E.F. Luca et al. Geoderma 310 (2018) 238–248

Table 2
The average amount of trash input and remaining at two of the study sites in São Paulo
state, Brazil.

Site Input Remaining

Mg DM ha− 1 yr− 1 Mg DM ha− 1 yr− 1 C N

−1
Mg ha

Hx4 13.9 a 4.5 a 1.61 a 0.0223 a

Qt4 12.8 a 3.6 b 1.36 a 0.0209 a

DM = dry matter biomass.

C and N are the total carbon and nitrogen, respectively.
Different letters in the same column indicates significant difference at p < 0.05.

2.4. Soil aggregate stability

A laboratory method based on the wet-sieving technique (Kemper

and Rosenau, 1986) where 4 g of top (0–0.05 m) bulk soils were placed
in a 0.2 mm sieve and rapidly immersed into deionised water for
30 min, then wet-sieved using a vertical oscillation of 33 cycles min− 1
Fig. 2. Representation of plot and locations of sampling positions for soils and litter; and 13 mm amplitude for 6 min, was used. The soil fraction over
horizontal lines represent sugarcane planting lines.
0.2 mm, which included macro-aggregates (MA) and coarse sands
(CSd), was dried at 105 °C for 16 h before weighting. This fraction was
subsequently wet-sieved and dispersed in a NaOH solution
2.2. Soil and trash sample collection and preparation (0.05 mol L− 1), dried and weighted, in order to determine the CSd. The
proportion of stable macro-aggregates (> 0.2 mm) was calculated by
Fig. 2 illustrates the sampling strategies adopted for both soils and the difference between the > 0.2 mm fraction and CSd. Aliquots of
sugarcane trash. Six soil pits were excavated per replicate plot after su- the < 0.02 mm fraction were used to determine the micro-aggregate
garcane harvesting in the fourth year and soil samples collected from 0 to fractions by the pipette method (Gee and Bauder, 1986). The meso-
0.05, 0.05–0.1, 0.1–0.2 and 0.2–0.4 m layers. Samples from corresponding aggregate (0.02–0.2 mm) content was calculated by the difference be-
soil depths at each plot were put together to make composite samples. The tween the total weight (4 g) and the other two fractions (Barthès et al.,
bulk soil samples (n = 88), of about 1 kg each, were then air-dried and 2008).
gently crushed to pass a 2 mm sieve. These samples were later analysed for
SOCC, SONC and other soil chemical properties. Another part of the bulk
soils corresponding to the top 0–0.05 m layers (n = 22) were also air dried 2.5. Statistical analysis
separately and analysed for aggregate stability. In addition, two un-
disturbed samples per soil depth of each pit (n = 176) were collected by All data were first checked for Normality (Shapiro-Wilk test) and
using a metallic ring (0.05 m height × 0.05 m diameter) for soil bulk variance stability (Box-Cox Test). The data were transformed in order to
density (ρb) determination. Litter was sampled from GT plots at Hx4 and correct them and then analysed again for Normality and Variance
Qt4 (n = 30) after harvesting in the fourth year, using 0.5-m long by Homogeneity before a complete randomized (one-way) analysis of
0.7 m wide quadrates. The litter samples were dried at 60 °C until constant variance was performed (GLM procedure was used). When differences
weight. Litter samples from the other two sites were not considered in the were significant, a multiple comparison procedure was performed using
analysis because they were heavily damaged by pests. t-test for comparison of average values at 5% probability level (SAS, 9.3
version, SAS Institute, 2012). The t-tests on SOCS and SONS were per-
2.3. Soil chemical analyses formed on data normalized by the mean of all samples. Data on bulk
density were used in the calculation procedures of SOCs and SONs. One-
Total carbon and nitrogen in the bulk soils and in the litter were on-one relationships between SOCC, SONC, SOCS, SONS, ρb, pH, content
determined using dry combustion with a CN 2000 LECO auto analyser of P, K, Ca, Mg, ECEC, and soil aggregate stability on one hand and soil
(LECO Corporation, St. Joseph, MI). Soil pH was determined in a texture and duration since abandonment of preharvest burning on the
0.01 mol L− 1 CaCl2 suspension (1:2.5 soil/solution, v/v). Total soil other hand were evaluated using Spearman correlations, while the
acidity (exchangeable H++Al3 +) was assessed by a SMP neutral multiple associations were explored by performing principal compo-
(pH = 7) buffer procedure, while available phosphorus (P) and ex- nent analysis (PCA) using Statistica software (StatSoft, 1996). The PCA
changeable potassium (K), calcium (Ca) and magnesium (Mg) were technique extracts important information from a multivariate dataset
extracted by resin in a 1:2.5 soil/solution v/v. P was assayed by col- and displays the relationships amongst the study variables.
orimetric method, K by flame photometry, and Ca and Mg by atomic
absorption spectrometry. All these procedures were standard methods 3. Results
as described by Raij et al. (2001).
As soil pH was lower than 5.6 in all cases, we assumed that in- 3.1. Change of soil organic carbon and nitrogen with shift from B to GT
organic carbon was absent in the soils and that organic carbon equalled
to total carbon. The total soil organic carbon and organic nitrogen The average amount of dry matter litter input onto the GT plots
stocks (SOCS, SONS) of each soil layer were estimated from ρb, SOCC varied from 12.8 Mg ha− 1 year− 1 at Qt4 to 13.9 Mg ha− 1 year− 1 at
and SONC following the same procedure described by Poeplau et al. Hx4 (Table 2). Shifting from B to GT resulted in overall increase of soil
(2017). The fraction of rock fragments in soils at all the study was not organic carbon (SOCC) and nitrogen (SONC) content for the soil profile
significant. The stocks were subsequently expressed by equivalent soil (Fig. 3). However, changes were only significant in the top layers with
mass (e.g. Ellert and Bettany, 1995) in order to account for changes in SOCC increasing from 11.9 ± 0.4 to 14.5 ± 0.3 g kg− 1 (21%) in the
soil volume consecutive to shifts in soil management. 0–0.05 m layer, while SONC increased from 0.73 ± 0.03 to

241
E.F. Luca et al. Geoderma 310 (2018) 238–248

0.88 ± 0.02 and 0.69 ± 0.02 to 0.83 ± 0.04 g kg− 1 for the 0–0.5 In terms of stocks, the whole profile (0–0.4 m) increase following
and 0.05–0.1 m soil layer, respectively. The greatest benefits, corre- the shift from B to GT was from 59.9 ± 1.3 to 64.1 ± 1.3 Mg C ha− 1
sponding to increases of 43 and 61% for SOCC and SONC respectively, for SOCS, from 3.7 ± 0.1 to 4.2 ± 0.1 Mg N ha− 1 for SONS (Table 2).
accrued at Qt4 which was characterised by lowest soil clay content. The These changes were significant at p < 0.05. The surface (0–0.05 m)
gain in topsoil SOCC showed a general decrease with increasing soil clay layer also experienced the greatest gains in the stocks (22% for both
content. However, Hx4 recorded greater increase than Ht4. The gain in SOCs and SONs). Across sites, only Qt4 and Hx12 gained SOCs and
topsoil SOCC at Hx4 was also greater than at Hx12, which was a longer- SONs significantly in the top layers. However, Qt4 also gained SONS
term experiment but with much lower soil clay content. Benefits to significantly in the lower layers. On average, SOCS and SONS increased
topsoil SONC also decreased with soil clay content. However, Hx12 with soil clay content in the order Hx4 > Hx12 > Ht4 > Qt4, while
gained more topsoil SONC than Ht4, which was on lower clay content gains in these stocks following the shift from B to GT was in reverse
soil. The high overall gain in SONC in the 0.05–0.1 m layer was driven order Qt4 > Ht4 > Hx12 > Hx4.
by the uncharacteristically high positive gain at Ht4 from 0.45 ± 0.15
to 0.71 ± 0.35 g kg− 1, which corresponded to 58% over 4 years.
While Hx4 and Ht4 showed positive SOCC and SONC gains in all their 3.2. Change of soil nutrients and other properties with shift from B to GT
horizons, Qt4 and Hx12 showed depletions in some horizons. The re-
sultant SOCc/SONc ratios did not vary much with soil depth under both Sites Ht4 and Qt4 exhibited relatively higher soil bulk densities (ρb)
treatments at all sites, and they varied from 17.7 to 19.8 and from 16.2 in comparison to Hx4 and Hx12 (Table 3), reflecting their lower soil
to 17.5 for B and GT, respectively. Qt4 exhibited greatest decreases of clay content (Table 1). Overall, abandonment of B for GT resulted in
SOCC/SONC ratios in corresponding soil layers following the shift from significantly greater ρb in the top 0.2 m horizon. However, not all sites
B to GT. recorded significant changes of the ρb following the shift. The longer
experiment (Hx12) experienced the most significant change of ρb (at

Fig. 3. Comparisons of soil total carbon (Ct, g kg− 1)

and nitrogen (Nt, g kg− 1) in different soil depths at the
experimental sites Ht4 (a and b), Ht4 (c and d), Qt (e
and f) and Hx12 (g and h) for the two sugarcane trash
management B: burning and GT: green-trashing. The
figures (a), (c), (e) and (g) are for Ct (g kg− 1), while
(b), (d), (f) and (h) are for Nt (g kg− 1).

242
E.F. Luca et al. Geoderma 310 (2018) 238–248

Table 3
Mean and standard deviation (between brackets) of soil total carbon (SOCS), nitrogen (SONS) stocks for the different soil layers at the four experimental sites.

Site and Soil layers (m) SOCS SONS

Mg C ha− 1 Mg N ha− 1

B GT % change B GT % change

†
Mean (n = 24)
0–0.05 8.02 (0.3) 9.82(0.2) 22a 0.49(0.02) 0.60(0.02) 22a
0.05–0.1 8.06(0.5) 8.34(0.4) 3 0.47(0.03) 0.57(0.04) 21a
0.1–0.2 16.76(1.0) 17.3(0.8) 3 1.08(0.06) 1.15(0.08) 6
0.2–0.4 27.04(1.6) 28.5(1.3) 5a 1.68(0.11) 1.82(0.10) 8a
Sum 59.88(1.3) 64.1(1.3) 7a,b 3.72(0.09) 4.15(0.11) 12a,b

Hx4 (n = 6)
0–0.05 12.2(2.5) 15.3(2.3) 25 0.83(0.15) 0.94(0.13) 13
0.05–0.1 25.2(3.4) 28.7(3.0) 14 1.71(0.24) 1.91(0.23) 12
0.1–0.2 51.0(5.4) 56.0(5.6) 10 3.59(0.51) 3.87(0.48) 8
0.2–0.4 91.4(8.4) 103(12) 13 6.32(0.82) 7.02(0.91) 11
Sum 179.8 203 13b 12.45 13.74 10b

Ht4 (n = 6)
0–0.05 6.9(1.1) 7.6(1.3) 10 0.46(0.18) 0.53(0.13) 15
0.05–0.1 13.2(3.2) 14.2(2.4) 8 0.83(0.16) 1.12(0.32) 35
0.1–0.2 25.6(4.4) 28.6(4.8) 12 1.73(0.26) 2.19(0.55) 27
0.2–0.4 45.8(8.3) 51.9(6.3) 13 3.30(0.61) 4.10(0.59) 24
Sum 91.5 102.3 12b 6.32 7.94 27b

Qt4 (n = 6)
0–0.05 5.4(1.2) 7.7(0.5) 43a 0.25(0.09) 0.40(0.05) 60a
0.05.-0.1 11.3(2.4) 13.6(0.8) 20a 0.46(0.11) 0.68(0.07) 48a
0.1–0.2 23.0(5.1) 25.8(3.6) 12 0.96(0.22) 1.30(0.29) 35a
0.2–0.4 40.1(6.9) 42.4(4.5) 6 1.60(0.24) 1.99(0.30) 24a
Sum 79.8 89.5 12b 1.47 1.84 25b

Hx12 (n = 4)
0–0.05 10.8(0.8) 12.5(0.1) 16a 0.61(0.05) 0.74(0.04) 21a
0.05–0.1 20.8(1.7) 23.0(1.3) 11 1.18(0.06) 1.36(0.09) 15a
0.1–0.2 43.8(4.1) 43.9(1.7) 0 2.56(0.24) 2.61(0.15) 2
0.2–0.4 82.3(7.8) 77.1(3.7) −6 4.75(0.42) 4.41(0.23) −7
Sum 157.7 156.5 −1b 9.10 9.12 0b

Mean† = mean of normalized data using the average for Hx4 as a reference.
n = number of plots.
See Table 1 for properties of soil at each site.
a
Significant difference between burn and green-trashing treatments at P ≤ 0.05. The dry matter biomass values for the two sites were also significantly different.
b
Change (%) for sum of the 4 soil layers.

12% for the 2 layers). The shift from B to GT was also associated with did not show any significant change. Ht4 experienced the greatest
significant increase of soil aggregation across sites, except Ht4 (Fig. 4). losses in available P. It was also observed that Hx12 recorded increasing
Macroaggregation increased significantly by 17, 24 and 39% at Hx4, available P with soil depth. It was also observed that Hx12 was the only
Qt4 and Hx12 respectively. However, meso and micro-aggregation de- site with decreasing K+ following the shift from B to GT. Ht4 and Qt4
creased significantly at each of these three sites. It was also interesting recorded big gains of K+ in the subsoil layers. Hx12 gained more Ca2 +
to note that all site soils, except the sandy Qt4, were dominated by the in all its soil layers than shorter-term trials. The shorter-term experi-
macroaggregate class following the shift. Ht4 was peculiar in that none ments showed much greater accumulation of Mg2 + in the surface than
of the aggregate classes (i.e. micro, meso and macro aggregates) subsoil layers.
changed significantly over 4 years. Overall, sum of exchangeable bases (SB) and base saturation (V)
The change from B to GT was accompanied by a decrease in average increased significantly following the shift from B to GT, and they tended
available P in all soil layers with greatest loss occurring in the topsoil to increase with soil depth (Table 4). Potential acidity (H+ + Al3 +)
(Table 3). However, Ca2 + and Mg2 + increased significantly, while K+ changed significantly in all layers, but with an increase in the surface

Fig. 4. Aggregate size distribution for different combinations of soil type

and sugarcane trash management (B: burning vs GT: green-trashing)
technique at the four experimental sites. MI = microaggregates,
ME = mesoaggregates, and MA = macroaggregates. Different letters at
a site indicate significant difference between B and GT treatments at
P ≤ 0.05.

243
 E.F. Luca et al.

Table 4
Mean and standard deviation (between brackets) of available phosphorus (P) and exchangeable potassium (K+), calcium (Ca2 +) and magnesium (Mg2 +) for different combinations of soil type and sugarcane trash management (B: burning vs GT:
green-trashing) techniques at the four experimental sites.

Site and Soil layers (m) ρb P K+ Ca2 + Mg2 +

Mg m− 3 mg kg− 1 mmolc kg− 1 mmolc kg− 1 mmolc kg− 1

B GT % change B GT % change B GT % change B GT % change B GT % change

Mean (n-24)
0–0.05 1.42(0.01) 1.52(0.01) 7a 19.6(0.08) 16.2(0.09) −17a 1.41(0.06) 1.32(0.06) −6 17.6(0.09) 20.0(0.09) 14a 6.35(0.29) 10.41(0.39) 64a
0.05–0.1 1.42(0.01) 1.52(0.01) 7a 18.8(0.14) 16.3(0.09) −13a 1.35(0.06) 1.28(0.06) −5 16.4(0.08) 21.1(0.14) 29a 5.83(0.26) 7.85(0.45) 35a
0.1–0.2 1.52(0.02) 1.60(0.01) 5a 17.0(0.07) 14.2(0.08) −16a 1.29(0.07) 1.23(0.06) −5 17.7(0.10) 19.3(0.11) 9a 6.10(0.36) 7.00(0.31) 15a
0.2–0.4 1.49(0.02) 1.51(0.01) 1 25.0(0.16) 23.4(0.17) −6a 0.82(0.06) 0.87(0.04) 6 11.5(0.06) 15.5(0.10) 35a 3.87(0.26) 5.70(0.33) 47a

Hx4 (n = 6)
0–0.05 26.9(6.0) 23.8(6.9) −12 1.26(0.33) 1.29(0.26) 2 19.3(7.8) 18.8(5.1) −3 6.39(2.67) 8.72(1.63) 36
0.05–0.1 1.30(0.05) 1.31(0.09) 1 24.6(8.9) 27.5(3.0) 12 1.16(0.41) 1.18(0.33) 2 17.1(6.2) 17.9(6.2) 5 5.66(2.81) 7.01(2.19) 24
0.1–0.2 1.30(0.05) 1.31(0.09) 1 24.4(5.9) 20.1(6.6) −18 0.95(0.37) 0.96(0.17) 1 22.8(6.9) 20.6(8.6) −10 8.25(3.36) 7.30(2.00) − 12
0.2–0.4 1.37(0.11) 1.37(0.11) 0 35.4(8.7) 31.4(11.6) −11 0.75(0.31) 0.74(0.13) −1 15.6(2.7) 19.0(6.4) 22 4.96(1.47) 6.85(2.61) 38
Mean 1.24(0.08) 1.26(0.10) 2 27.8(7.4) 25.7(7.0) −8 1.03(0.35) 1.04(0.22) 1 18.7(6.6) 19.1(6.6) 2 6.32(2.6) 7.47(2.1) 18

Ht4 (n = 6)

244
0–0.05 21.0(6.0) 12.0(6.2) −43 1.05(0.46) 1.35(0.27) 29 22.5(5.1) 20.5(4.0) −9 8.54(1.36) 14.23(1.77) 67a
0.05–0.1 1.52(0.03) 1.66(0.06) 9a 25.2(8.9) 10.7(4.2) −58 1.19(0.41) 1.63(0.37) 37 19.1(4.8) 24.2(5.9) 27 7.93(0.91) 9.83(1.39) 24
0.1–0.2 1.52(0.03) 1.66(0.06) 9a 16.0(5.9) 7.9(2.1) −51 1.10(0.32) 1.71(0.54) 55 20.7(6.1) 20.8(3.5) 0.5 7.28(1.45) 8.97(1.11) 23
0.2–0.4 1.65(0.06) 1.73(0.05) 5a 18.1(8.7) 12.7(4.4) −30 0.42(0.09) 1.16(0.40) 176a 11.2(2.6) 15.1(6.0) 35 4.07(1.73) 5.67(1.68) 39
Mean 1.63(0.07) 1.67(0.06) 2 20.0(7.4) 10.8(4.2) −46 0.94(0.32) 1.46(0.39) 55 18.4(4.65) 20.2(4.85) 10 6.96(1.36) 9.68(1.49) 39

Qt4 (n = 6)
0–0.05 15.1(6.0) 14.1(3.2) −7 0.42(0.07) 0.48(0.11) 14 13.0(2.5) 13.5(3.0) 4 3.10(0.50) 6.83(1.71) 120a
0.05–0.1 1.49(0.05) 1.60(0.02) 7a 12.8(8.9) 12.6(4.6) −2 0.20(0.10) 0.31(0.09) 55 13.4(3.4) 15.1(6.4) 13 2.90(0.74) 3.55(1.18) 22
0.1–0.2 1.49(0.05) 1.60(0.02) 7a 14.5(5.9) 14.2(5.1) −2 0.14(0.05) 0.38(0.08) 171a 12.6(3.0) 12.3(3.7) −2 2.56(0.63) 2.75(0.74) 7
0.2–0.4 1.56(0.08) 1.68(0.04) 8a 26.9(8.7) 23.7(10.3) −12 0.16(0.10) 0.26(0.08) 63 8.6(2.6) 7.6(2.4) −12 2.13(0.61) 1.92(0.88) − 10
Mean 1.60(0.09) 1.66(0.09) 4 17.3(7.4) 16.1(5.8) −7 0.23(0.08) 0.36(0.09) 57 11.9(2.87) 12.1(3.9) 2 2.67(0.62) 3.76(1.13) 41

Hx12 (n = 4)
0–0.05 15.4(6.0) 15.0(1.6) −3 2.89(0.31) 2.17(0.59) −25 15.7(3.0) 27.0(5.3) 72a 7.37(1.18) 11.87(2.43) 61a
0.05–0.1 1.36(0.10) 1.52(0.03) 12a 12.9(8.9) 14.6(2.3) 13 2.85(0.22) 1.99(0.30) −30a 15.9(0.5) 26.9(9.0) 69a 6.84(0.60) 11.02(4.03) 61
0.1–0.2 1.36(0.10) 1.52(0.03) 12a 12.9(5.9) 14.8(1.9) 15 2.94(0.52) 1.86(0.34) −37a 14.9(3.6) 23.6(5.2) 58 6.31(1.60) 8.96(2.16) 42
0.2–0.4 1.50(0.11) 1.60(0.04) 7 19.5(8.7) 25.7(6.4) 32 1.93(0.73) 1.33(0.20) −31 10.7(4.0) 20.3(4.2) 90a 4.32(1.28) 8.35(1.36) 93a
Mean 1.47(0.07) 1.46(0.04) −1 15.2(7.4) 17.5(3.0) 15 2.65(0.44) 1.84(0.36) −31 14.3(2.8) 24.5(6.1) 71a 6.21(1.16) 10.05(2.94) 62

n = number of plots.
See Table 1 for properties of soil at each site.
a
Significant difference at P ≤ 0.05 between burn and green-trashing treatments.
Geoderma 310 (2018) 238–248
E.F. Luca et al. Geoderma 310 (2018) 238–248

the 0–0.4 m soil layer by 7% over four years (Table 2), which corre-
sponded to 1.75% year− 1, under different soil conditions tested in
tropical Brazil. The 22% increase (an increase of 5.5% year− 1) of
surface (0–0.05 m) soil layer SOCS was greatest and only significant
change. Across sites, the greatest gain in surface soil SOCS and SONS
occurred at a site with lowest topsoil clay content which agreed with
results from other studies (e.g. Thorburn et al., 2012; Pinheiro et al.,
2010; Orlando Filho et al., 1998). Pinheiro et al. (2010) observed a 50%
increase of SOCS on sandy Brazilian soils 13 years following a shift from
burning to green-trashing. Orlando Filho et al. (1998) reported lower
SOCS gain with an average of 6.5% on clayey soils in Brazil. Elsewhere,
Thorburn et al. (2012) reported increased topsoil (0–0.2 m) SOCS only
after 5 years of green-trashing on soils of different textures in Australia.
The fact that our study sites recorded significant increases of surface
SOCS in fewer years than other studies was probably a consequence of
high productivity of the study sugarcane fields. The fields produced 12
to 13 metric tons year− 1 of green-trash (Table 2), which were com-
pletely broken down and decomposed by soil fauna thus potentially
contributing to the high SOCS. These amounts of biomass were com-
parable to observations by other recent studies in Brazil; for example,
Carvalho et al. (2013) who reported trash retention rates of average
11.3 and 11.5 tons ha− 1 year− 1, respectively. The greater increase of
SOCS and SONS in more degraded soils with lower clay content can be
explained by the fact these soils had lost more C due to aggregate break
down (Chaplot and Cooper, 2015).
Fig. 5. Principal component analysis showing relationships between multiple controlling While the overall results from our study show a general increase of
factors (D: experiment duration, BD: soil bulk density, pH, Bio: mulch biomass, MA: soil
SOCs, one site (Hx12) showed decrease of SOCS and no change of SONS
macroaggregates, SOCC and SONC for soil organic carbonand nitrogen content, Clay, Silt
and Sand for soil clay, silt and sand content, Ca: calcium, Mg: magnesium, Al: aluminium,
for the whole soil profile (Table 2). This result agreed with long-term
K: potassium, SB: sum of exchangeable bases, V: base saturation and ECEC: effective study results from other land uses and environments. For example,
cation echangeable capacity) as active variables and rates of change in soil carbon (% Dimassi et al. (2014) reported on rapid SOC accumulation in surface
SOCS) and nitrogen stocks (%SONS) for the 0–0.05 m soil layer were used as supple- (0–0.1 m) layers in the first 4 years of green-trashing with a plateau
mentery variables. being reached in 28 years and then followed by a decrease thereafter,
but the subsoil (0.1–0.3 m) layer was reported to experience continuous
(0–0.05 m) layer and decrease in the subsurface layers. The gains in decline throughout. Blair (2000) also reported decreasing SONS fol-
effective cation exchange capacity (ECEC) tended to decrease with soil lowing a similar shift management to green-trashing, but Ball-Coelho
depth. Surface soil pH did not change significantly, but subsoil pH et al. (1993) reported no significant change of SOCS. While SOCS de-
changed significantly following the shift from B to GT with the change creased by average 7% at Hx12 in 12 years of trashing for 12 years,
increasing with soil depth. The variations of these soil chemical prop- Blair (2000) reported a decrease of 13% in four years.
erties varied with site and soil depth. However, ECEC exhibited a trend There was no clear explanation proffered for the decrease of SOCS in
similar to SOCs and SONs in that it decreased with soil clay content in short and long-term green-trashing experiments, but one possible
the order Hx4 > Hx12 > Ht4 > Qt4. reason is that tillage for replanting of sugarcane done every 4–5 years at
Hx12 and other farms in Brazil contribute to significant losses of SOCS.
Tillage breaks down soil aggregates thereby exposing the physically
3.3. Factors controlling the changes in SOCS
protected SOC to preferential erosion and accelerated decomposition
(Chaplot and Cooper, 2015). Tillage might induce “priming” by in-
The principal component analysis (PCA) in Fig. 5 shows the multiple
corporating fresh organic material deep into the soil profile, which
relationships between rate of change in soil SOCS and SONS on one
accelerates SOC decomposition by microbial activity (Dimassi et al.,
hand and the controlling factors at the other. The two major principal
2014). The fact that SOC accumulation in green-trash treatments takes
components accounted for 88.9% of the dataset variability, with PC1
place only in the surface layer makes it vulnerable to decomposition
accounting for 61.4% while PC2 accounted for 27.5%. PC1 was closely
and release of CO2 to the atmosphere. Tillage operations every
associated with factors such as SOCC, SONC, ECEC, soil texture (Clay,
4–5 years further jeopardise the long-term capacity of green-trashing to
Silt and Sand content) and bulk density (BD), while PC2 was more
sequester atmospheric C to the soil. Most of the atmospheric C se-
closely associated with experiment duration (D). The PCA results show
questered during the 4–5 year period is immediately released back into
no direct link between rate of changes in SOCS and SONS, and soil
the atmosphere at tillage stage; therefore, green-trashing might not
texture; but SOCC and SONC tended to increase with soil clay and silt
sequester significant amounts of soil C in the long-term. Our study re-
content, ECEC. There was also tendency for greater sequestration of
sults showed that SONS behaved in similar trend to SOCS, but there was
organic carbon and nitrogen in soils characterised by low biomass ac-
no overall soil profile SONS change at Hx12. However, it is important to
cumulation on soil surface, macroaggregation, chemical elements and
note that other past studies on the same sites e.g. Pinheiro et al. (2010)
short experiment duration. These observations were confirmed by re-
and Galdos et al. (2009) observed lower C sequestration rates after 14
sults from Spearman rank correlation analyses (Table 5). (See Table 6.)
and 8 years, respectively, than our results.

4. Discussion 4.2. Controls of the changes in SOCs following green-trashing

4.1. Impact of green-trashing on soil carbon stocks Generally, degraded soils with low C and nutrient stocks experience
the greatest change in SOCS following abandonment of mismanagement
On average, green-trashing of sugarcane residues increased SOCS in and putting in place rehabilitation initiatives on that land (Guo and

245
 E.F. Luca et al.

Table 5
Mean and standard deviation (between brackets) of sum of bases (SB), potential acidity (H++Al3 +), effective cation exchange capacity (ECEC), base saturation (V) and pH for different combinations of soil type and sugarcane trash management (B:
burning vs GT: green-trashing) techniques at the four experimental sites.

Site and Soil layers (m) SB H++Al3 + ECEC V pH (CaCl2)

mmolC kg− 1 mmolC kg− 1 %

B GT % change B GT % change B GT % change B GT % change B GT % change

Mean (n = 24)
0–05 25.4(0.12) 31.7(0.13) 25a 32.8(0.08) 33.1(0.08) 1a 58.1(0.90) 64.8(1.14) 12a 45.3(1.51) 50.4(1.43) 11a 4.98(0.05) 4.99(0.04) 0.2
0.05–0.1 23.5(0.10) 30.2(0.13) 29a 34.9(0.11) 33.0(0.09) − 5a 58.4(1.47) 63.1(1.47) 8a 42.4(1.57) 49.0(1.92) 16a 4.89(0.04) 5.03(0.05) 3a
0.1–0.2 25.1(0.13) 27.6(0.14) 10a 34.6(0.13) 30.2(0.07) − 13a 59.7(0.12) 57.7(1.12) − 3a 43.2(1.78) 48.4(1.08) 12a 4.87(0.06) 5.09(0.04) 5a
0.2–0.4 16.2(0.08) 22.1(0.3) 36a 37.1(0.12) 31.9(0.08) − 14a 53.3(0.10) 54.0(0.96) 1 30.8(1.55) 40.1(1.88) 30a 4.56(0.05) 4.82(0.06) 6a

Hx4 (n = 6)
0–05 26.9(10.6) 28.8(6.4) 7 55.9(8.4) 60.5(6.7) 8 82.8(3.5) 89.3(7.7) 8 32.2(11.8) 32.2(5.9) 0 4.82(0.51) 4.65(0.15) −4
0.05–0.1 23.9(8.7) 26.1(8.1) 9 57.7(9.0) 61.8(9.2) 7 81.6(3.5) 87.9(2.7) 8a 29.0(12.5) 29.8(9.4) 3 4.65(0.26) 4.63(0.21) − 0.4
0.1–0.2 32.0(9.8) 28.9(11.0) −10 53.2(10.5) 51.9(5.1) −2 85.2(7.4) 80.7(7.8) −5 36.5(8.0) 35.1(9.5) −4 4.73(0.35) 4.80(0.22) 1
0.2–0.4 21.3(3.9) 26.6(9.1) 25 53.4(9.0) 51.3(4.3) −4 74.7(1.5) 77.9(7.5) 4 28.5(5.1) 33.6(9.3) 18 4.63(0.29) 4.77(0.20) 3
Mean 26.0 27.6 55.0 56.4 81.3 83.9 31.9 32.7 4.71 4.71

Ht4 (n = 6)

246
0–05 32.1(6.0) 36.1(5.8) 12 19.0(1.6) 19.5(2.2) 3 19.0(1.6) 19.5(2.2) 9 62.5(3.7) 64.7(5.8) 4 5.35(0.12) 5.42(0.25) 1
0.05–0.1 28.2(5.5) 35.9(7.6) 27 19.7(1.9) 18.9(2.6) −4 19.7(1.9) 18.9(2.6) 14 58.6(4.6) 64.5(7.3) 10 5.30(0.13) 5.47(0.30) 3
0.1–0.2 29.1(6.5) 31.5(4.5) 8 19.0(2.0) 17.7(1.6) −7 19.0(2.0) 17.7(1.6) 2 59.9(5.8) 63.9(3.8) 7 5.26(0.19) 5.52(0.16) 5
0.2–0.4 15.7(3.3) 21.9(7.6) 39 27.4(3.3) 22.5(4.0) − 18 27.4(3.3) 22.5(4.0) 3 34.8(7.3) 48.0(11.1) 38 4.48(0.35) 4.97(0.45) 11
Mean 26.3 31.3 21.3 19.6 21.3 19.6 54.0 60.3 5.10 5.35

Qt4 (n = 6)
0–05 16.5(3.0) 20.8(4.7) 26 21.7(4.8) 25.6(5.3) 18 38.2(3.6) 46.4(3.8) 21a 43.5(9.4) 44.9(9.9) 3 4.96(0.25) 4.79(0.25) −3
0.05–0.1 16.5(4.0) 19.0(7.4) 15 23.6(7.5) 23.1(5.2) −2 40.1(6.1) 42.1(6.2) 5 41.9(12.4) 44.3(13.2) 6 4.92(0.36) 4.97(0.41) 1
0.1–0.2 15.3(3.6) 15.4(4.3) 1 25.1(8.6) 23.2(4.3) −8 40.3(5.5) 38.6(3.3) −4 39.3(13.3) 39.8(0.4) 1 4.90(0.38) 4.97(0.24) 1
0.2–0.4 10.9(3.2) 9.8(3.2) −10 24.0(6.0) 24.1(3.0) 0.4 34.9(4.9) 33.9(1.9) −3 31.9(10.1) 28.8(8.7) − 10 4.66(0.27) 4.67(0.22) 0.2
Mean 14.8 16.3 23.6 24.0 38.4 40.3 39.1 39.5 4.86 4.85 0

Hx12 (n = 4)
0–05 26.0(4.2) 41.1(7.9) 58a 34.5(1.5) 27.1(1.8) − 21a 60.5(3.3) 68.1(6.2) 13 42.8(4.7) 59.8(6.3) 40a 4.78(0.07) 5.08(0.09) 6a
0.05–0.1 25.6(0.9) 39.9(3.2) 56a 38.5(2.3) 28.0(1.6) − 27a 64.1(2.8) 67.9(3.0) 6 39.9(1.2) 57.7(7.6) 45a 4.68(0.06) 5.04(0.12) 8a
0.1–0.2 24.1(5.6) 34.5(7.5) 43 41.0(5.2) 27.9(2.5) − 32a 65.1(3.5) 62.4(6.1) −4 36.9(7.8) 54.8(7.3) 49a 4.60(0.14) 5.06(0.14) 10a
0.2–0.4 17.0(5.4) 30.0(5.6) 76a 43.5(5.1) 29.8(3.9) − 31a 60.5(6.1) 59.7(3.7) −1 27.8(7.9) 50.0(7.8) 80a 4.46(0.13) 4.86(0.24) 9a
Mean 23.2 36.3 39.4 28.2 62.5 64.5 36.9 55.6 4.63 5.01

n = number of plots.
See Table 1 for properties of soil at each site.
a
Significant difference at P ≤ 0.05 between burn and green-trashing treatments.
Geoderma 310 (2018) 238–248
E.F. Luca et al. Geoderma 310 (2018) 238–248

Table 6 in the content of exchangeable bases in the green mulching treatment,

Spearman rank correlations between experiment duration (D), biomass accumulated which was concomitant to a decrease in K content. In contrast, Orlando
on soil surface (Bio), soil texture (Clay, Silt and Sand content), bulk density (BD),
Filho et al. (1998), also in Brazil, found a decrease in soil available P
macroaggregates (MA), organic carbon and nitrogen content (SOCc and SONc, re-
spectively), chemical properties (pH, Ca: calcium, Mg: magnesium, Al: aluminium, K: content and exchangeable Ca and Mg, but an increase in available K,
potassium, SB: sum of exchangeable bases, V: base saturation and ECEC: effective results that they explained by nutrient leaching and nutrient plant as-
cation echangeable capacity) and changes in soil organic carbon and nitrogen stocks similation (Orlando Filho et al., 1998; Ball-Coelho et al., 1993). Overall,
per year (%SOCS and % SONS, respectively). green-trashing increased SOCS, SONS and chemical elements, especially
%SOCS %SONS
in the topmost layer, because increased soil organic matter formed had
greater ability to tighten the elements together.
D −0.77 a
− 0.77a
Clay −0.80a − 0.80a 5. Conclusions
Silt −0.80a − 0.80a
Sand 0.80a 0.80a
SOCc −0.80a − 0.80a This study aiming to assess the impact of changing sugarcane trash
SONc −0.80a − 0.80a management from preharvest burn to green-trashing on carbon se-
BD 0.60a 0.60a questration into soils showed that this shift can increase soil carbon
Bio −0.32 − 0.32
stocks by an average 17.5‰ per year, with most of the carbon se-
P 0.00 0.00
K −0.80a − 0.80a questration occurring in the superficial 0–0.05 m soil layer. The highest
Ca −0.80a − 0.80a sequestration rate (108‰ per year) occurred in the most carbon de-
Mg −0.40 − 0.40 pleted soils which happened to be sandy soils. Considering an average
SB −0.80a − 0.80a sequestration rate of 1.06 ton ha− 1 y− 1, the resulting soil carbon se-
Al −0.60a − 0.60a
questration rate for Brazilian sugar cane soils alone would be 9.3 mil-
ECEC −0.80a − 0.80a
V 0.00 0.00 lion ton C y− 1. Green-trashing appears to be an efficient technique for
pH 0.60a 0.60a atmospheric C sequestration to soils to meet the objective of the ‘4 per
MA −0.40 − 0.40 mille Soils for Food Security and Climate’ which was launched at the
a
COP21. However, intermittent tillage may pose limitations to carbon
Significant at p < 0.05.
sequestration using green-trashing because the longer experiment
which was sampled one year after tillage for replanting did not show
Gifford, 2002). The present study confirmed this by showing negative
any carbon sequestration, suggesting all the carbon stored over four
correlations between soil organic carbon and nitrogen content and rate
years between planting and replanting was lost at tillage. This calls for
of change in SOCS (Fig. 5). This did not only apply to C and nutrient
further research to quantify the dynamics of carbon stocks during and
contents, but also to all the studied chemical elements. The higher rate
immediately after tillage and to seek for sugarcane replanting techni-
of C sequestration in degraded soils reflects a sink due to past land
ques that minimize soil disturbance and associated carbon losses.
mismanagement. The high rate of SOCS change in degraded soils can be
explained by high amount of available sites for insertion of the C and
Acknowledgements
other elements, but the sequestration capacity is likely to be finite and
will only correspond to the refilling of the depleted C (Mackey et al.,
We thank FAPESP (process 12.648-3) (Research Support Foundation
2013; Six et al., 2002).
of the State of São Paulo) for supporting this research and providing a
The highest sequestration rate also corresponded to sandy soils
post-graduate grant to the first author, as well as the CTC (Sugarcane
(Fig. 5) because they are more prone to degradation due to their low
Technology Center, former Copersucar), São Martinho, Santa Luiza and
aggregate stability and high erosion rate, which dramatically affect
Da Pedra for providing the experimental areas.
SOCS by biological decomposition of the organic C. The sandy soils also
have high potential for preferential detachment and transport of or-
References
ganic C by water (Müller-Nedebock et al., 2016). Low proportion of
macroaggregates under the studied sandy soils (Fig. 5) confirmed their
Ball-Coelho, B., Tiessen, H., Stewart, J.W.B., Salcedo, I.B., Sampaio, E.V.S.B., 1993.
higher initial degradation level and thus highest C sequestration rate Residue management effects on sugarcane yield and soil properties in northeastern
following abandonment of preharvest burning of sugarcane trash in Brazil. Agron. J. 85, 1004–1008.
favour of green-trashing. Barthès, B.G., Kouakoua, E., Larré-Larrouy, M.C., Razafimbelo, T.M., de Luca, E.F.,
Azontonde, A., CSVJ, Neves, Freitas, P.L., Feller, C., 2008. Texture and sesquioxide
Sandier soils also experienced greater increase of soil bulk density effects on water-stable aggregates and organic matter in some tropical soils.
following the shift from preharvest burning to green-trashing, which Geoderma 143, 14–25.
was in agreement with results from other studies elsewhere (e.g. Batjes, N.H., 1996. Total carbon and nitrogen in the soils of the world. Eur. J. Soil Sci. 47,
151–163.
Osunbitan et al., 2005). Soil compaction, which results in greater bulk Blair, N., 2000. Impact of cultivation and sugarcane green-trash management on carbon
densities, may help in sequestering C because of reduced gaseous ex- fractions and aggregate stability for a chromic luvisol in Queensland, Australia. Soil
change between soils and the atmosphere. Tillage Res. 55, 183–191.
Bruce, J.P., Frome, M., Haites, E., Janzen, H., Lal, R., 1999. Carbon sequestration in soils.
The increased proportion of macroaggregates following green- J. Soil Water Conserv. 54, 382–389.
trashing at all the studied sites also inform on the C sequestration Carvalho, J.L.N., Otto, R., Franco, H.C.J., Trivelin, P.C.O., 2013. Input of sugarcane post-
mechanism in soils. Soil aggregation by biological and/or physical harvest residues into the soil. Sci. Agric. 70 (5), 336–344.
Cerri, C.E.P., Sparove, G., Bernoux, M., Easterling, W.E., Melillo, J.M., Cerri, C.C., 2007.
mechanisms, which enhance carbon and nitrogen protection (Feller and
Tropical agriculture and global warming: impacts and mitigation options. Sci. Agric.
Beare, 1997), most probably had a role in enhancing C storage under 64 (1), 83–99.
green-trashing. Elsewhere in Brazil, Souza et al. (2005) reached similar Chabbi, A., Lehmann, J., Ciais, P., Loescher, H.W., Cotrufo, M.F., Don, A., 2017. Aligning
agriculture and climate policy. Nat. Clim. Chang. 7 (5), 307–309. http://dx.doi.org/
conclusions at Hapludox sites where green cane trash left on the soil
10.1038/nclimate3286.
surface resulted in increased mean diameter of soil aggregates by 62%. Chaplot, V., Cooper, M., 2015. Soil aggregate stability to predict organic carbon outputs
Blair (2000) reported a similar increase in Australia. All these results from soils. Geoderma 243-244, 205–213.
point to carbon sequestration benefits of enhanced soil aggregation. CONAB (Companhia Nacional de Abastecimento), 2013. Acompanhamento da safra
brasileira: cana-de-açúcar: safra 2013/2014 – segundo levantamento, agosto/2013.
Generally, there is a positive effect of sugarcane green-trashing on Brasilia. Accessed at < http://www.conab.gov.br//> on 3 Nov 2014.
soil chemical fertility, especially in medium to long term (Bruce et al., Dignac, M.F., Derrien, D., Barré, P., Barot, S., Cécillon, L., Chenu, C., Chevallier, T.,
1999; Tivet et al., 2013). The present study pointed to a sharp increase Freschet, G.T., Garnier, P., Guenet, B., Hedde, M., Klumpp, K., Lashermes, G., Maron,

247
E.F. Luca et al. Geoderma 310 (2018) 238–248

P.A., Nunan, N., Roumet, C., Basile-Doelsch, I., 2017. Increasing soil carbon storage: from soils by sheet erosion and main controls. Earth Surf. Process. Landf. http://dx.
mechanisms, effects of agricultural practices and proxies: a review. Agron. Sustain. doi.org/10.1002/esp.3916.
Dev. 37, 14. http://dx.doi.org/10.1007/s13593-017-0421-2. Noble, A.D., Moody, P., Berthelsen, S., 2003. Influence of changed management of su-
Dimassi, B., Mary, B., Wylleman, R., Labreuche, J., Couture, D., Piraux, F., Cohan, J.-P., garcane on some soil chemical properties in the humid wet tropics of north
2014. Long-term effect of contrasted tillage and crop management on soil carbon Queensland. Aust. J. Soil Res. 41, 1133–1144.
dynamics during 41 years. Agric. Ecosyst. Environ. 188, 134–146. Orlando Filho, J., Rosseto, R., Muraoka, T., Zotelli, H.B., 1998. Efeitos do sistema de
Dominy, C.S., Haynes, R.J., Van Antwerpen, R., 2002. Loss of soil organic matter and despalha (cana crua x cana queimada) sobre algumas propriedades do solo. STAB-
related soil properties under long-term sugarcane production on two contrasting Açúcar, Álcool e Subprod. 16 (6), 30–34.
soils. Biol. Fertil. Soils 36, 350–356. Osunbitan, J.A., Oyedele, D.J., Adekalu, K.O., 2005. Tillage effects on bulk density, hy-
Ellert, B.H., Bettany, J.R., 1995. Calculation of organic matter and nutrients stored in soils draulic conductivity and strength of a loamy sand soil in southwestern Nigeria. Soil
under contrasting management regimes. Can. J. Soil Sci. 75, 529–538. Tillage Res. 82, 57–64.
FAO (Food and Agricultural Organisation of the United Nations), 1988. FAO/Unesco soil Paustian, K., Lehmann, J., Ogle, S., Reay, D., Robertson, G.P., Smith, P., 2016. Climate-
map of the world, revised legend, with corrections and updates. In: World Soil smart soils. Nature 49. http://dx.doi.org/10.1038/nature17174.
Resources Report. 60 FAO, Rome Reprinted with updates as Technical Paper 20, Pereira, A.R., Angelocci, L.R., Sentelhas, P.C., 2002. Agrometeorologia: fundamentos e
ISRIC, Wageningen, 1997 (1988). aplicações práticas. In: Guaíba: Agropecuária, 478p.
Feller, C., Beare, M.H., 1997. Physical control of soil organic matter dynamics in tropical Pinheiro, E.F.M., Lima, E., Ceddia, M.B., Urquiaga, S., Alves, B.J.R., Boddey, R.M., 2010.
land-use systems. Geoderma 79, 49–67. Impact of preharvest burning versus trash conservation on soil carbon and nitrogen
Galdos, M.V., Cerri, C.C., Cerri, C.E.P., 2009. Soil carbon stocks under burned and un- stocks on a sugarcane plantation in the Brazilian Atlantic forest region. Plant Soil
burned sugarcane in Brazil. Geoderma 153, 347–352. 333, 71–80.
Gee, G.W., Bauder, J.W., 1986. Particle size analysis. In: Klute, A. (Ed.), Methods of Soil Poeplau, C., Vos, C., Don, A., 2017. Soil organic carbon stocks are systematically over-
Analysis, Part 1. Agronomy. 9 Am. Soc. Agron. Inc., Madison, Wis. estimated by misuse of the parameters bulk density and rock fragment content. Soil 3,
Graham, M.H., Haynes, R.J., Meyer, J.H., 2002. Soil organic matter content and quality: 61–66.
effects of fertilizer applications, burning and trash retention on a long-term sugarcane Raij, B.V., Andrade, J.C., Cantarella, H., Quaggio, J.A., 2001. Análise química para
experiment in South Africa. Soil Biol. Biochem. 34, 93–102. avaliação da fertilidade de solos tropicais. Inst. Agron, Campinas.
Guo, L.B., Gifford, R.M., 2002. Soil carbon stocks and land use change: a meta-analysis. Ripoli, T.C.C., Ripoli, M.L.C., 2009. Biomassa de cana-de-açúcar: colheita, energia e
Glob. Chang. Biol. 8, 345–360. ambiente. 2. Ed., Piracicaba. 333p.
Kemper, W.D., Rosenau, R.C., 1986. Aggregate stability and size distribution. In: Klute, A. São Paulo (State), 2003. Decreto n. 47.700, 11/03/2003. Regulamenta a Lei 11.241, 19/
(Ed.), Methods of Soil Analysis, second ed., Part 1, Physical and Mineralogical 09/2002, que dispõe sobre a eliminação gradativa da queima da palha da cana-de-
Methods. American Society of Agronomy, Madison, pp. 425–442. açúcar e dá providências correlatas. Diário Oficial do Estado, São Paulo, 12 mar.
Lal, R., 2015. Cover cropping and the “4 per thousand” proposal. J. Soil Water Conserv. SAS Institute, 2012. SAS/STAT User's Guide, Version 9.3, Ninth Edition. v. 1/2. Cary, NC.
70 (6). http://dx.doi.org/10.2489/jswc.70.6.141A. Six, J., Conant, R.T., Paul, E.A., Paustian, K., 2002. Stabilization mechanisms of soil or-
Macedo, I.C., 1998. Greenhouse gas emissions and energy balances in bioethanol pro- ganic matter: implications for C-saturation of soils. Plant Soil 241, 155–176.
duction and utilization in Brazil. Biomass Bioenergy 14, 77–81. Souza, Z.M., Prado, R.M., Paixão, A.C.S., Cesarin, L.G., 2005. Sistemas de colheita e
Mackey, B., Prentice, I.C., Steffen, W., House, J.I., Lindenmayer, D., Keith, H., Berry, S., manejo da palhada de cana-de-açúcar. Pesq. Agrop. Brasileira 40, 271–278.
2013. Nature climate change. In: Untangling the Confusion Around Land Carbon SSS (Soil Survey Staff), 1992. Keys to Soil Taxonomy, 5th Ed. Pocahontas Press,
Science and Climate Change Mitigation Policy, http://dx.doi.org/10.1038/ Blacksburg.
nclimate1804. StatSoft Inc, 1996. StatiStica for Windows (Computer Program Manual). StatSoft, Inc.,
Minasny, B., Malone, B.P., McBratney, A.B., Angers, D.A., Arrouays, D., Chambers, A., Tulsa.
Chaplot, V., Chen, Z.S., Chengg, K., Das, B.S., Fielda, D.J., Gimona, A., Hedley, C.B., Thorburn, P.J., Meier, E.A., Collins, K., Robertson, F.A., 2012. Changes in soil carbon
Hong, S.Y., Mandal, B., Marchant, B.P., Martin, M., McConkey, B.G., Mulder, V.L., sequestration, fractionation and soil fertility in response to sugarcane residue re-
O'Rourke, S., Richer-de-Forges, A.C., Odeh, I., Padarian, J., Paustian, K., Pan, G., tention are site-specific. Soil Tillage Res. 120, 99–111.
Poggio, L., Savin, I., Stolbovoy, V., Stockmann, U., Sulaeman, Y., Tsui, C.C., Vågen, Tivet, F., Sá, J.C.M., Lal, R., Briedis, C., Borszowskei, P.R., Santos, J.B., Farias, A., Eurich,
T.G., vanWesemael, B., Winowiecki, L., 2017. Soil carbon 4 per mille. Geoderma 292, G., Hartman, D.C., Nadolny Jr., M., Bouzinac, S.A., Seguy, L., 2013. Aggregate C
59–86. depletion by plowing and its restoration by diverse biomass-C inputs under no-till in
Müller-Nedebock, D., Chivenge, P., Chaplot, V., 2016. Selective organic carbon losses sub-tropical and tropical regions of Brazil. Soil Tillage Res. 126, 203–218.

248