Increasing demand for biofuel has intensified land-use change (LUC) for sugarcane (Saccharum officinarum) expansion in Brazil. Assessments of soil quality (SQ) response to this LUC are essential for quantifying and monitoring sustainability of sugarcane production over time. Since there is not a universal methodology for assessing SQ, we conducted a field-study at three sites within the largest sugarcane-producing region of Brazil to develop a SQ index (SQI). The most common LUC scenario (i.e., native vegetation to pasture to sugarcane) was evaluated using six SQI strategies with varying complexities. Thirty eight soil indicators were included in the total dataset. Two minimum datasets were selected: one using principal component analysis (7 indicators) and the other based on expert opinion (5 indicators). Non-linear scoring curves were used to interpret the indicator values. Weighted and non-weighted additive methods were used to combine individual indicator scores into an overall SQI. Long-term conversion from native vegetation to extensive pasture significantly decreased overall SQ. In contrast, conversion from pasture to sugarcane had no significant impact on overall SQ at the regional scale, but site-specific responses were found. In general, sugarcane production improved chemical attributes (i.e., higher macronutrient levels and lower soil acidity); however it has negative effects on physical and biological attributes (i.e., higher soil compaction and structural degradation as well as lower soil organic carbon (SOC), abundance and diversity of macrofauna and microbial activity). Overall, we found that simple, user-friendly strategies were as effective as more complex ones for identifying SQ changes. Therefore, as a protocol for SQ assessments in Brazilian sugarcane areas, we recommend using a small number of indicators (e.g., pH, P, K, Visual Evaluation of Soil Structure -VESS scores and SOC concentration) and proportional weighting to reflect chemical, physical and biological processes within the soil. Our SQ evaluations also suggest that current approaches for expanding Brazilian sugarcane production by converting degraded pasture land to cropland can be a sustainable strategy for meeting increasing biofuel demand. However, management practices that alleviate negative impacts on soil physical and biological indicators must be prioritized within sugarcane producing areas to prevent unintentional SQ degradation over time.
Citation: Cherubin MR, Karlen DL, Cerri CEP, Franco ALC, Tormena CA, Davies CA, et al. (2016) Soil Quality Indexing Strategies for Evaluating Sugarcane Expansion in Brazil. PLoS ONE 11(3): e0150860. doi:10.1371/journal.pone.0150860
Editor: Wenju Liang, Chinese Academy of Sciences, CHINA
Received: December 15, 2015; Accepted: February 20, 2016; Published: March 3, 2016
This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: MRC and ALCF thank the São Paulo Research Foundation (FAPESP) for research grants (Processes # 2013/17581-6; 2014/16612-8 and 2012/22510-8). CEPC, CCC and CAT thank the National Council for Scientific and Technological Development (CNPq) – Brazil for their productivity grants. CAT thanks the CNPq for his postdoc grant (Process 203376/2014-5). CAD thanks the Shell Technology Center Houston for providing him support in form of salaries. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of the authors are articulated in the ‘author contributions’ section.
Competing interests: The authors have declared that no competing interests exist. Although CAD is an employee of Shell Technology Center Houston, this does not alter the authors' adherence to PLOS ONE policies on sharing data and materials.
Increasing global demand for biofuel has accelerated land-use change (LUC) to support bioenergy crops in many countries. In Brazil, the area devoted to sugarcane production increased from 5.8 to 9.0 Mha during the last decade . Even though Brazil is already the world’s largest sugarcane producer, current predictions indicate that an additional 6.4 Mha of sugarcane will be needed to meet the domestic demand for ethanol by 2021 . Sugarcane expansion has primarily occurred on lands previously occupied by extensive pastures [2, 3], most of which are degraded or in the process of being degraded [4, 5]. To obtain long-term energy security, bioenergy systems will need to be agronomically and environmentally sustainable. Intensification of land use through mechanization and agrochemical inputs has direct implications on soil physical, chemical and biological properties and consequently on the quality/health of soils. To prevent unintended consequences, monitoring of soil property changes due to LUC is essential [6, 7]. However, this research topic is still new in Brazil, and we are not aware of any protocol for evaluating soil quality (SQ) changes induced by sugarcane expansion in this region.
Soil quality was defined as the capacity of a specific kind of soil to function, within natural or managed ecosystem boundaries, to sustain plant and animal productivity, maintain or enhance water and air quality, and support human health and habitation . It is a complex functional concept and cannot be measured directly in the field or laboratory; but can be indirectly inferred by soil indicators [8, 9]. Indicators of SQ are those measurable soil properties and processes that have greatest sensitivity to changes in soil function and its ecosystem services [7, 10]. A wide range of soil chemical, physical and biological properties could be measured [7, 11–14], but due to cost it’s not feasible to consider them all, and therefore it is necessary to select a minimum dataset (MDS). Several strategies have been used to define an appropriate MDS including principal component analysis (PCA) [9, 15–20], fuzzy sets [21–22], expert opinion [10, 12] and farmer/local knowledge [23, 24]. According Doran and Parkin  suitable SQ indicators should encompass ecosystem processes, integrate soil properties, be accessible to many users, sensitive to management and climate, and, whenever possible, be components of existing databases. An example for reducing the number of potential SQ indicators was provided by Andrews et al.  through their development of the Soil Management Assessment Framework (SMAF). Starting with an extensive list of 80 or more integrative measurements related to ecosystem processes and functions that reflect SQ, they developed scoring curves only for a small number (i.e., 10) of carefully selected indicators that could reliably detect SQ changes induced by agricultural management practices. In more recent studies, others have shown that small datasets can effectively characterize SQ within different ecosystems. Lima et al.  compared SQ assessment using a total dataset (TDS) of 29 indicators, a MDS of eight indicators based on PCA, and an indigenous set of four indicators based on farmer knowledge to evaluate rice (Oryza sativa L.) production systems in southern Brazil. They concluded that the TDS provided the best assessment of SQ, but the smaller datasets showed the same SQ trends and thus provided meaningful information for land managers. Askari and Holden [18, 19] reduced the number of indicators using PCA from 21 to 3 and from 22 to 7, respectively, and verified that the MDS indicators were suitable to efficiently quantify SQ in grassland and arable fields in Ireland.
After defining a MDS, linear and non-linear techniques, each with their advantages and disadvantages, have been applied to interpret SQ indicators [12, 18, 19, 27]. While linear methods are simple, user-friendly and require little knowledge of the indicator thresholds, non-linear methods can more often assign meaningful scores that better represent the soil functions being represented by the indicators .
Once individual indicators have been scored, it is often convenient, but not essential, to integrate them into an overall SQ index (SQI) that can be used to support decision making and selection of sustainable management practices . Currently, there is no comprehensive, universal SQI that can be used across multiple natural and anthropogenic ecosystems. Many indexing strategies have been developed and tested for specific purposes under particular environmental conditions around the world (e.g., in the U.S.A. [9, 10, 12, 28–30], Brazil , Argentina , Italy , Spain , Ireland [18, 19], South Africa , India  and China [17, 21, 33]). The most user-friendly method to calculate a SQI is to simply add all indicator scores and then divide by the number of indicators [9, 10, 12, 29]. The major concern regarding this method is that when the number of indicators is unbalanced among chemical, physical and biological sectors, the overall SQI misrepresents the sector(s) having fewer indicators. On the other hand, several studies have used methods that assign weights for each indicator. Different criteria that have been used include soil function frameworks [9, 11, 26], principal component loading [9, 12, 20, 27, 33], partial least squares regression coefficients  and correlation with crop yield . Simple and weighted additive SQ indexing strategies provide site-specific responses [9, 12, 18, 19], influenced by existing dataset, soil type, and effects of land use and management practices.
Developing more user-friendly and cost-effective strategies for assessing SQ changes induced by agricultural management practices, especially those associated to bioenergy feedstock production therefore remains a challenge for the scientific community [9, 10, 12, 28]. Our goal was to develop a sensitive and reliable protocol for evaluating SQ impact associated with LUC occurring to increase Brazilian sugarcane production. To do so, we conducted a field-study at three sites where the primary LUC sequence (i.e., native vegetation to pasture to sugarcane) is occurring within of the largest sugarcane-producing region of Brazil. Six SQ indexing strategies with varying complexity were developed and tested. Our hypotheses were that: (i) the LUC sequence would result in SQ degradation; (ii) the SQI approach would be suitable to detect SQ changes due to LUC; and (iii) the simple, more user-friendly strategies would be able to detect SQ changes as effectively as more complex strategies.
Material and Methods
Field sites and experimental design
Land-use change effects on SQ were evaluated at three sites along a 1000-km transect within central-southern Brazil (Fig 1), which is the largest sugarcane-producing region in the world. Soil samples were collected from each LUC phase at: (i) Lat_17S located in southern Goiás state, the largest hotspot of sugarcane expansion in Brazil; (ii) Lat_21S located in western São Paulo state, a transition area between traditional and new sugarcane production cores, and (iii) Lat_23S: located in south-central São Paulo state, which represents the traditional sugarcane production core in Brazil.
Climate patterns at all three sites are similar with rainfall concentrated in the spring and summer (October to April) followed by a dry season in autumn and winter (May to September). The soils are typical for the Brazilian tropical region, well-drained and highly weathered, classified as Oxisols (Lat_17S and Lat_23S) or Alfisols and Ultisols (Lat_21S). A chronosequence was sampled at each site representing the three land uses (native vegetation, pasture and sugarcane) associated with the most common LUC sequence in central-southern Brazil. The synchronic approach (chronosequence) was chosen to represent potential long-term changes occurring in the region due to this LUC. The land use history involves conversion of the native vegetation (Cerrado and Atlantic Forest) to extensive pasturelands in the beginning of 1980’s. These areas are typical Brazilian pasturelands, which are cropped with tropical grasses (Brachiaria and Cynodon genus) and characterized by absence of management, extensive and continuous grazing and low grass productivity and stocking rates (i.e., 1 animal unit ha-1). Sugarcane was subsequently established on a portion of the pasture during the early 1990s at Lat_23S, and more recently in 2009 and 2010 at Lat_17S and Lat_21S. Conversion to sugarcane required intensive, heavy tillage operations (plowing and disking) and the application of lime. Subsequent annual sugarcane management requires fertilization using mineral and/or organic fertilizers, and pesticide applications to control weed, pests and diseases. Traditionally, sugarcane fields were burned before harvest, but at the Lat_23S site, the crop has been mechanically harvested without any burning since 2003 and at the Lat_17S and Lat_21S sites, the crop was never burned. Additional location, climate, soil, land use and management information for each site is available in Cherubin et al. .
Soil sampling and analysis
All soil samples were collected using a consistent experimental design that had four points spaced 50 m apart imposed within each land use. This provided 12 sampling points for each location or 36 sampling points for the three locations. A small trench (30 x 30 x 30 cm) was opened at each sampling point to collect both undisturbed and semi-undisturbed samples from the 0 to 10-, 10 to 20- and 20 to 30-cm layers. This provided 108 samples for physical analyses, and 108 for soil aggregation and macrofauna analyses. An additional 108 disturbed samples were collected for chemical and biological analyses by compositing 12 subsamples taken from each soil layer with a Dutch auger.
Available phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), sulfur (S—sulphate), boron (B), cooper (Cu), manganese (Mn), iron (Fe), zinc (Zn), active acidity (pHCaCl2 0.01mol L-1), potential acidity (H+Al), base saturation (BS) and potential cation exchange capacity (CECpH7) were determined using analytical methods described by Raij et al. . Soil resistance to penetration (SRP) and field-saturated hydraulic conductivity (Kfs) were measured at five and three locations, respectively, within ~5 m of each trench using a digital penetrometer (PenetroLOG®) and the ‘simplified falling-head’ method proposed by Bagarello et al. . Soil structural quality of the 20 x 10 x 25 cm monoliths from each trench was assessed using the Visual Evaluation of Soil Structure (VESS) method [37, 38]. Particle-size was determined using the hydrometer method. Bulk density (BD) was determined using the core method with 100 cm3 cylinders. Soil degree of compactness (SDC) was calculated as SDC = (BD/BDmax) x 100, where BDmax is maximum bulk density, estimated using the pedotransfer function described by Marcolin and Klein . Total porosity (TP) was calculated as TP = 1 - (BD/PD), where, PD is particle density, determined using a gas pycnometer. Soil water content at -6 kPa and -10 kPa water potential was determined using tension tables as described by Ball and Hunter . Soil macroporosity (MaP) was computed as the difference between soil water content at saturation and at -6 kPa. Soil microporosity (MiP) was estimated as the soil water content at -6 kPa. Water-filled pore space (WFPS) was calculated by dividing volumetric moisture at -6 kPa by total porosity as indicated in Wienhold et al. . We also calculated two indexes suggested by Reynolds et al. : i) soil water storage capacity (SWSC) defined as the ratio between water content at field capacity (FC, -10 kPa soil water potential) and TP (SWSC = FC/TP); and ii) soil aeration capacity (SAC) calculated as the ratio between drained pores at soil water potential of -10 kPa (ACt) and TP (SAC = ACt/TP). A structural stability index (SSI) was calculated as suggested by Reynolds et al. : SSI = ((SOC x 1.724) / (silt + clay))*100. Wet macroaggregate stability (AGS) was determined using a vertical oscillator (Yoder, model MA-148) with three sieves (2000, 250, and 53 μm) moving at a speed of 30 oscillations per min for 10 min. Percentage of macroaggregates was calculated by summing aggregate mass for >2000 and 250 μm classes, dividing by the total soil mass, and multiplying by 100. Mean weight diameter (MWD) was calculated as the sum of the proportion of aggregates in each size fraction, with each proportion weighted by the mean diameter of aggregates in that size fraction. Soil organic carbon (SOC) and total nitrogen (TN) were determined by dry combustion on a LECO® CN-2000 elemental analyzer (furnace at 1350°C in pure oxygen). Carbon and nitrogen within microbial biomass (MBC and MBN) were measured by fumigation/extraction as proposed by Vance et al. . Enzymatic activities of β-Glucosidase (BG) and acid phosphatase (AcP) were measured as described by Tabatabai . Immediately after the sampling, soil macrofauna were carefully hand-sorted from each 25 x 25 x 30 cm soil block, according to the standard Tropical Soil Biology and Fertility Institute (TSBF) soil monolith method . Invertebrates were classified into the taxonomic groups: Aranae, Blattodea, Chilopoda, Coleoptera, Dermaptera, Diplopoda, Diptera, Formicidae, others Hymenoptera, Gastropoda, Hemiptera, Isopoda, Isoptera, Oligochaeta, and Scorpiones. Macrofauna density was determined as the number of individuals per surface unit (m2). Ecological indexes were calculated for assessing richness (Margalef’s index) and diversity (Shannon’s index), according to the methods described by Magurran .
Developing the soil quality indexes
Six SQI values were developed using different approaches (Fig 2), although each involved three common steps: selection of SQ indicators as an MDS, transformation of indicator values into unitless 0 to 1 scores using scoring curves, and integration into an overall index [10–13]. The SQIs were compared to identify the most appropriate strategy for assessing SQ changes induced by LUC associated with sugarcane expansion in Brazil. Soil data from the 0 to 10-, 10 to 20- and 20 to 30-cm layers were averaged to create a 0 to 30-cm layer that was then used to calculate an overall SQI that better represented the whole soil profile.
Step 1- Indicator selection.
Three indicator selection approaches were evaluated: (i) the Total Dataset (TDS) which included 38 indicators representing 14 chemical, 14 physical and 10 biological properties and processes; (ii) a MDS-PCA created using PCA on the TDS to reduce data redundancy and identify the most efficient indicators, without depending upon subjective, expert opinion or literature values, and (iii) a five indicator MDS-EO chosen based on expert opinion and literature review. For the MDS-PCA, only seven components with eigenvalues >1 (Kaiser’s criteria) were retained and subjected to varimax rotation to enhance the interpretability of the components (Fig 3). Furthermore, for each component, only the indicators with loading values within 10% of the highest value were retained [12, 17–19, 33]. When more than one indicator was retained, correlation values among them were analyzed (S1 Table). If the indicators were significantly correlated (p<0.01), only the one with the highest loading factor was retained in the MDS to avoid redundancy [9, 12, 17–19]. The MDS-EO was selected taking into account the indicator’s ability to detect soil function changes as well as the ease, practicality and cost-effectiveness for sampling, analysis and interpretation.
Step 2- Indicator interpretation.
All measured indicator values were transformed using non-linear scoring functions. Based on agronomic and environmental soil functions, each indicator was scored using one of the following curves: “more is better” (upper asymptote sigmoid curve), “less in better” (lower asymptote sigmoid curve), and “mid-point optimum” (Gaussian curve), as exemplified in Fig 4. The non-linear Eqs 1 and 2 were used for “more is better” and “less is better” scoring curve shapes, respectively. For “mid-point optimum” curve the Eqs 1 and 2 were jointly used in the increasing and decreasing parts of the curve, respectively. (1) (2) where, Score is the unitless value of the soil indicator which ranging from 0 to 1, a is the maximum score which was equal to 1 in this study, B is the baseline value of the soil indicator where the score equals 0.5, LB is the lower threshold, UB is the upper threshold, x is the measured soil indicator value, and S is the slope of equation set to -2.5.
A) more-is-better function; B) less-is-better function; C) mid-point optimum function.
Threshold and baseline values for each soil indicator were based on literature references and expert’s opinion, as presented in the Table 1. Indicator scoring calculations were performed using a Microsoft Excel® spreadsheet.
Step 3- Indicator integration into an index.
The indicator scores were integrated into indexes through two approaches, simple additive (Eq 3) used to calculate SQI-1, SQI-3 and SQI-5 (Fig 2); and weighted additive (Eq 4) used to calculate SQI-2, SQI-4 and SQI-6 (Fig 2). (3) (4) where, Si is the indicator score, n the number of indicators integrated in the index and Wi the weighted value of the indicators. For the TDS, the indicators were weighted according to a framework developed based on five soil functions (Table 2), as suggest by Karlen and Stott  and later used by Lima et al. . Step by step procedure used for calculate the SQI-2 is shown in the S2 Table. For the MDS-PCA, the indicators were weighted according with proportional variation explained by each principal component (i.e., % variance explained by each component divided by total cumulative variance of all components selected for the MDS). For the MDS-EO the indicators were weighted by chemical, physical and biological sectors, in which each one, regardless of number of indicators, had the same weight (33%) in the final index.
Data were tested for normality using Shapiro-Wilk’s tests (p>0.05). The results indicated that no transformation was required. Principal component analysis (PCA) was performed using PROC FACTOR procedure to select a MDS based on a statistical approach. An analysis of variance (ANOVA) was computed using PROC GLM procedure to test LUC effects on soil indicators and SQI scores. If the ANOVA F statistic was significant (p<0.05), the means were compared using Tukey’s test (p<0.05). Linear correlations among SQI strategies were verified by Pearson’s correlation analysis using PROC CORR procedure. All statistical procedures were completed using the software Statistical Analysis System–SAS v.9.3 (SAS Inc, Cary, USA).
Results and Discussion
Soil quality indicators
Land-use change effects on the 38 soil quality indicators at each site are presented in Table 3. As typically reported for tropical soils, native vegetation sites were characterized by high acidity, low levels of soil organic matter (SOM) and plant-available macronutrients, suitable soil physical conditions, and high activity as well as diversity of edaphic fauna. Long-term conversion from native vegetation to extensive pasture significantly increased soil acidification (i.e., decreased pH and increased H+Al concentrations), depleted SOM (SOC and TN), available macronutrients, B and CTCpH7 and, increased micronutrient (Cu, Fe, Mn and Zn) availability. Poor long-term management, which typically includes continuous grazing without liming and/or applying fertilizer over time [4, 5, 34], is a major factor for SOC and nutrient depletion within Brazilian pastures. Conversion from native vegetation to pasture also degraded soil physical properties. Continuous cattle trampling coupled with SOC depletion, increased soil compaction (i.e., higher BD and SDC) and altered pore size and distribution (i.e., lower MaP and higher MiP). This subsequently reduced soil aeration (SAC), significantly decreased Kfs and available water, and may restrict root growth (i.e., higher SRP and VESS scores). Despite those changes, soil aggregate stability (AGS and MWD) was not affected by pasture establishment. Soil compaction and consequently physical degradation of pasturelands are well documented in the literature (e.g., [48–50]).
Soil biological changes were also observed due to conversion from native vegetation to pasture. Most biological indicators showed site-specific responses, although MBC and MBN tended to be lower within pasture soils, especially at the Lat_17S and Lat_21S sites. Enzyme activities (BG and AcP) showed a decreasing trend under pasture at Lat_17S and Lat_23S, but increased significantly at Lat_21S. Variation in soil acidity, SOC, P availability, microbiological activity and other variables not assessed in this study, may be among the controlling factors affecting enzyme responses at the various sites. Pastures soils generally had a higher density of macrofauna than native vegetation sites, but the increase was dominated by a few taxonomic groups such termites (mainly at Lat_17S), ants, coleopterans and earthworms. Conversely, even though native vegetation samples had lower macrofauna populations, they had a higher richness and diversity of species. Our findings are consistent with others in the literature [51, 52], which generally state that macrofaunal community size in tropical soils tends to increase over time following conversion from native vegetation to pasture.
The LUC from pasture to sugarcane improved soil chemical quality. Liming and annual application of fertilizer (organic and/or mineral) reduced soil acidity and increased macronutrient availability . Short-term sugarcane establishment (<5 years) had no negative impacts on SOC or TN content at Lat_17S and Lat_21S, but as reported by Mello et al.  and Franco et al. , SOC and TN were depleted after more than 20 years (Lat_23S) of sugarcane cultivation. Those decreases presumably are associated with the intensive tillage performed every five years [53, 55] and more than 10 years of pre-harvest burning, which has been shown to deplete SOC over time .
Conversion from pasture to sugarcane also negatively impacted on soil physical indicators, primarily those related to soil structure, such as AGS, MWD, VESS and SSI. Although tillage in preparation for sugarcane replanting (Lat_17S) alleviated soil compaction (i.e., decreased BD and SRP; increased Kfs), our data suggest those positive effects have short-term persistence (i.e., primarily the first year, as reported by Centurion et al. ). Over the entire sugarcane cycle, intensive machinery traffic increases soil compaction again, leading to decreased of aeration, infiltration and water availability as observed at Lat_21S and Lat_23S. Short-term positive tillage effects on soil physical quality are most likely associated with SOC depletion, due to disruption of macroaggregates and exposure of physically and chemically protected C to microbial decomposition  and the subsequent deleterious consequences on soil structure. In addition, several studies have shown that intensive machinery traffic in sugarcane fields has negative impacts on soil physical quality and often decreases sugarcane growth and yield (e.g., [59, 60]). Adverse impacts of current sugarcane management practices on soil physical and structural properties have also markedly increased soil loss and degradation by erosion when compared with native vegetation or pasture . Degraded SQ has thus become a major concern for a sustainable sugarcane production in Brazil .
Overall, LUC from pasture to sugarcane also has negative implications on soil biological indicators. Depletions of soil biota in sugarcane fields can be associated with quantitative and qualitative decreases in SOC. Franco et al.  reported that sugarcane production depletes C input from C3 plants (forest) which is preferable by microorganisms, and that new C from C4 plants (i.e., pasture and sugarcane) was insufficient to offset those losses. Furthermore, intensification of land use and management, that includes considerable mineral fertilizer and pesticides inputs as well as the modification or destruction of native biological habitats by tillage, and soil compaction can led to a reduction or simplification in soil diversity and its ecosystem functions in sugarcane fields .
Soil quality indexing
The three SQ indicator selection approaches (Fig 2) provided different datasets for index calculations. The TDS (38 indicators) provided a wide range of soil indicators and theoretically should have resulted in a more accurate (sensitive) assessment of SQ, due to the very comprehensive evaluation involving chemical, physical and biological soil properties and their interactions. The primary limitations of the TDS approach are the high cost, greater amount of time required for sampling and laboratory analyses, redundancy of indicators, and more complex data interpretation [18, 21, 26]. Using a PCA reduced the TDS to seven principal components (MDS-PCA) that explained approximately 90% of total variance (Fig 3 and Table 4). The seven indicators were: SOC, SAC, pH, Kfs, Mdiver, BG and Mdens. Selecting SQ indicators using PCA has some advantages and disadvantages. According to Andrews et al.  and Mukherjee and Lal , PCA provides a less subjective method of indicator selection, which can help avoid bias and data redundancy. On the other hand, the PCA method requires a large dataset and is less “user friendly,” thus imposing barriers to practical adoption for farm or regional scale SQ assessments. Furthermore, the selected indicators may not be meaningful for farmers and land managers . The expert opinion approach reduced the TDS to five MDS-EO indicators (pH, P, K, VESS and SOC), with the first three being chemical indicators that are widely used to evaluate soil acidity and nutrient availability as well as to guide soil fertility management. As recommended by Doran and Parkin , these indicators are desirable for SQ assessments because they are: easy to sample for, readily available in commercial laboratories at a low cost, and the results can be easily interpreted using pre-defined thresholds. The fourth indicator, VESS score, provides an integrative assessment of soil structural/physical quality through an easily-performed, low-cost, direct on-farm method [37, 38]. VESS integrates soil properties related to size, strength and porosity of aggregates, roots and soil color into a single score, that ranges from 1 (good) to 5 (poor structural quality) [37, 38]. The fifth indicator, SOC, is the most consistent indicator used for SQ assessments  because it influences multiple soil and ecosystem functions . Furthermore, SOC can be analyzed using the same sample collected for chemical indicators and it is routinely analyzed so most farmers have previous records for temporal comparisons. The MDS-EO approach was consistent with Andrews et al.  and Karlen et al. , who recommend that SQ assessments could be made using a minimum of five indicators provided there was at least one each representing soil chemical, physical and biological properties and processes. However, Andrews et al.  did warn that the expert opinion method does truly require expert knowledge of the entire production system and may be subject to disciplinary bias.
Soil quality indicators were individually scored (Eqs 1 and 2) and then, integrated using six strategies (Fig 2). The SQI scores for native vegetation, pasture and sugarcane (0–30 cm depth) at each site are shown in Fig 5. Overall, all six SQI approaches were able to detect SQ changes induced by LUC. Soils from native vegetation sites had significantly greater SQI values, except at Lat_17S, where the soil was more weathered and consequently had very poor chemical quality . In general, LUC from native vegetation to pasture significantly decreased SQ, although the sensitivity among the SQ indexing strategies was slightly different. Conversion from pasture to sugarcane promoted site-specific SQ changes, leading to increases or decreases associated with inherent soil characteristics and historic of land use and management. At Lat_17S, sugarcane cultivation increased SQ, primarily due to soil fertility improvement through lime and fertilizer applications. This was confirmed by SQIs calculated using strategies that gave greater weight to chemical indicators, such as SQI-4 and SQI-5. At Lat_21S, conversion from pasture to sugarcane had essentially no influence on overall SQ, except when SQI-5 was used for the evaluation. In contrast, at Lat_23S, SQI-1, SQI-3 and SQI-4 indicated that long-term sugarcane cultivation significantly decreased SQ likely due to significant SOM depletion . This in turn had negative implications on micro- and macro-faunal activity, cycling and availability of nutrients, and soil structure (Table 3).
§SQI strategies: SQI-1: TDS/non-linear/simple additive, SQI-2: TDS/non-linear/weighted additive, SQI-3: MDS-PCA/non-linear/simple additive, SQI-4: MDS-PCA/non-linear/ weighted additive, SQI-5: MDS-EO /non-linear/simple additive, and SQI-6: MDS-EO /non-linear/weighted additive. *Mean values within each index followed by the same letter do not differ among themselves according to Tukey’s test (p<0.05).
At the regional scale, SQ changes induced by LUC were consistently detected by all six indexing strategies (Fig 6). In general, higher absolute SQI values were observed when using the TDS (SQI-1 and SQI-2) followed by the SQIs from MDS-EO (SQI-5 and SQI-6) and SQIs from MDS-PCA (SQI-3 and SQI-4). An identical sequence was verified by Lima et al. . Native vegetation soils had the highest SQI scores, suggesting they are functioning at 56 to 78% of their potential capacity for the 0–30 cm depth. These results support the hypothesis that natural ecosystems are more balanced, because chemical, physical and biological attributes act collectively, thus enabling soils to perform their functions properly. The SQIs indicated that long-term conversion from native vegetation to extensive pasture decreased SQ indexes by 15 to 23% (Fig 6), resulting in pasture soils that were functioning at between 44 to 66% of their potential capacity. Weighed indexes helped clarify the reasons for overall SQ depletion within pasturelands. The SQI-2 scores (i.e., TDS weighted by soil function framework) indicated that pasture soils had reduced soil functions associated with storage and provision of water, as well as soil aeration (-32%), soil capacity to sustain plant growth (-18%), biological activity (-30%), ability to resist degradation (-22%), and although not statistically significant, the capacity for storage, provision and cycling of nutrients (-6%) when compared to soils under native vegetation (Fig 7). The SQI-4, weighted by PCA loading, showed that SQ depletions in pasturelands were mainly associated with significant decreases of SOC and Kfs (Fig 8A). Using only five selected, but weighted soil indicators (SQI-6) detected that SQ depletion due to conversion from native vegetation to pasture was associated with significant decreases in soil chemical (-23%), physical (-17%) and biological sectors (-22%) (Fig 8B). This was in agreement with results obtained using SQI-2, which was the most complex strategy.
*Mean values within each index followed by the same letter do not differ among themselves according to Tukey’s test (p<0.05).
§Same letter within each soil function indicates that the mean values do not differ among land uses according to Tukey’s test (p<0.05).
Contribution of each principal component (PC) and soil sector in the SQI-4 (A) and SQI-6 (B), respectively under native vegetation (NV), pasture (PA) and sugarcane (SC) in central-southern Brazil. §Same letter within each PC or soil sector indicates that the mean values do not differ among land uses according to Tukey’s test (p<0.05).
Our results effectively described the critical situation associated with most Brazilian pastureland. It is estimated that 70% of those areas are degraded or in the process of being degraded . Recently, a national-scale study verified that the current productivity (i.e., animal unit carrying capacity) of cultivated pasturelands is only 32–34% of their inherent potential . The low productivity of Brazilian pasturelands has multiple causes as reported by Strassburg et al. . Among them are improper pasture management, including seedling failures and bare soil, continuous grazing, absence of liming, maintenance of soil fertility through fertilization, and uncontrolled erosion, which all lead to soil degradation over time [4, 5].
At the regional scale, LUC from pasture to sugarcane showed no significant impact on overall SQ (Fig 6). The SQI scores suggest that sugarcane soils are functioning at 47 to 66% of their capacity. The SQI strategies showed sparse non-significant variations on SQ under sugarcane compared to pasture, ranging from -6% (SQI-3) to +13% (SQI-5). Respectively, SQI-3 and SQI-5 were the indexes that gave the lowest and the highest weight to chemical indicators. Therefore, improving soil fertility attenuated negative implications of sugarcane production on soil physical and biological indicators within overall SQ assessment. This was clearly demonstrated by weighted indexes (SQI-2, SQI-4 and SQI-6). Conversion from pasture to sugarcane had one positive effect on soil functions–that related to nutrient dynamics. In contrast, significant adverse effects were observed in soil functions related to the capacities to sustain biological activity and resist to degradation (Fig 7). SQI-4 showed that under sugarcane only pH scores was improved, while SOC, BG and Mdens scores decreased (Fig 8A). Finally, SQI-6 also was able to indicate that sugarcane production led to significant improvement on soil chemical indicators and decline on physical and biological indicators. Overall, these results indicate that sugarcane expansion over degraded pasturelands seems to be an opportune way to meet increasing domestic and global ethanol demands, avoiding direct competition for land with food crops and natural ecosystems, as reported by Goldemberg et al.  and Strassburg et al. . However, the results clearly indicated the necessity for improved management practices that can mitigate deleterious impacts of sugarcane production on soil physical/structural and biological indicators.
What is the best indexing strategy for assessing sugarcane expansion impacts on soil quality?
All six SQ indexing strategies were able to detect SQ changes induced by LUC, suggesting that any of them could be used for monitoring SQ in sugarcane expansion in Brazil (Fig 6). However, a sensitivity test showed there were slight differences among the strategies (Fig 9). The most complex strategy (SQI-2), which included the 38 indicator TDS and used weighting of the indicator scores provided by the soil function framework had greatest sensitivity to detect SQ changes due to LUC. In contrast, the least sensitive SQI was calculated using TDS without indicator weighting (SQI-1). These results suggest that using a meaningful method (e.g., soil functions) for weighting and integrating indicator scores into an index when a large dataset is available for SQ assessment is best, even though it is more complex than simple additive indexing and does not statistically modify the overall SQ assessment response (Fig 6). There also is no consensus in the literature regarding the benefits of indicator weighting. Andrews et al.  and Askari and Holden  concluded that weighting an additive SQI did not change the relative rankings among treatments, and therefore, this extra step was unnecessary for analyzing vegetable production or other systems. Mukherjee and Lal  also reported similar effectiveness between simple and weighted indexes. They highlighted that appropriate weighting on scores can predict SQ with higher performance which was consistent with our findings. On the other hand, Askari and Holden  showed that a simple additive linear SQI was the most efficient for detecting management practice impacts in arable soils.
Both the MDS-PCA (SQI-3 and SQI-4) and MDS-EO (SQI-5 and SQI-6) strategies were effective for detecting SQ changes using a reduced number of indicators (Fig 9). Similar results were reported by Andrews et al.  and Lima et al. , who both concluded that a reduced number of carefully chosen indicators could adequately provide the information needed for decision-making. From a practical perspective, this means that for any one indexing strategy to become the standard for research, large-scale SQ assessments, or to facilitate discussion and cooperation, it must be rapid, reliable, and economically feasible . SQI-5 and SQI-6 strategies have an important advantage compared to SQI-3 and SQI-4, since the latter two require a large dataset in order to perform a PCA and select fewer indicators. Therefore, simple SQI strategies (SQI-5 and SQI-6), show excellent potential for monitoring SQ changes in sugarcane expansion areas in Brazil. More specifically, between SQI-5 and SQI-6, we suggest opting for SQI-6, which provides balanced weighting for chemical, physical and biological indicators.
The SQ indexing strategies were significantly correlated among themselves (Fig 10), confirming a close relationship between simple and more complex strategies. This was also reported by Mukherjee and Lal . Decreasing correlations between SQI-1 vs SQI-3 (r = 0.93), SQI-1 vs SQI-5 (r = 0.78), SQI-2 vs SQI-4 (r = 0.47), and SQI-2 vs SQI-6 (r = 0.34) were observed. This indicates that as the indexes became simpler, correlations with more complex indexes that used the entire dataset were lower. However, despite lower correlations, simple indexing strategies (i.e., SQI-5 and SQI-6) had the same statistical ability for ranking SQ responses due to LUC (Fig 6) as the more complex strategies.
Simple additive indexes had greater correlations among themselves than weighted indexes (Fig 10), because the simple ones were calculated using only a different number of indicators, while weighted indexes also varied the weighting approaches. Higher correlations were verified between simple additive and weighted additive indexes when fewer indicators were selected (i.e., SQI-1 vs SQI-2, r = 0.66; SQI-3 vs SQI-4, r = 0.89; SQI-5 vs SQI-6, r = 0.92). Fig 9 confirms there was a decreasing trend for sensitivity differences between simple and weighted indexes derived from the same sequence, as the comparisons moved from more complex to simpler strategies.
All six indexing strategies efficiently detected SQ changes due to LUC for sugarcane expansion in Brazilian tropical soils. Both, PCA and EO approaches were useful to reduce the total dataset without any significant interference on SQ ranking among land uses. These results indicate that simple, easily-performed and more user-friendly SQI strategies (e.g., SQI-5 and SQI-6) were as effective and suitable for detecting LUC effects on SQ as more complex SQI strategies (e.g., SQI-1, SQI-2, SQI-3 and SQI-4). Although simple additive and weighted additive SQIs were statistically similar, we recommend using weighted indexes, especially when the number of indicators is unbalanced among chemical, physical and biological components. Therefore, a SQI strategy using a small number of carefully chosen soil indicators, such as pH, P, K, VESS and SOC, and proportional weighting for indicator scores within of each soil sector (chemical, physical and biological) could be adopted as a protocol for SQ assessments in Brazilian sugarcane areas.
Our findings also suggest that long-term LUC from native vegetation to extensive pasture depleted overall SQ, driven by decreases in chemical, physical and biological indicators. In contrast, conversion from pasture to sugarcane had no significant impact on overall SQ, primarily, because chemical improvements offset negative impacts on biological and physical indicators. Therefore, sugarcane expansion into degraded pastureland seems to be a sustainable strategy to meet increasing demands for biofuels. Nevertheless, management practices that alleviate soil physical and biological degradation under sugarcane production must be prioritized to avoid or minimize SQ depletions over time.
S1 Table. Pearson’s correlation coefficients (r) among soil chemical, physical and biological indicators in the land use change areas in central-southern Brazil.
S2 Table. Model of soil functions framework and indicators used to develop the SQI-2.
The authors are grateful to Raízen for all the support received and access provided for this research.
The U.S. Department of Agriculture (USDA) prohibits discrimination in all its programs and activities. USDA is an equal opportunity provider and employer.
Conceived and designed the experiments: MRC CEPC ALCF CCC. Performed the experiments: MRC ALCF. Analyzed the data: MRC. Contributed reagents/materials/analysis tools: MRC CEPC CCC. Wrote the paper: MRC DLK CEPC ALCF CAT CAD CCC.
- 1. Companhia Nacional de Abastecimento—Conab. Acompanhamento da safra brasileira de cana-de-açúcar. v. 2 –Safra 2015/16, n. 2, Segundo Levantamento, 2015. Available: http://conab.gov.br/OlalaCMS/uploads/arquivos/15_08_13_15_58_44_boletim_cana_portugues_-_2o_lev_-_15-16.pdf.
- 2. Goldemberg J, Mello FFC, Cerri CEP, Davies CA, Cerri CC. Meeting the global demand for biofuels in 2021 through sustainable land-use change policy. Energy Policy 2014; 69: 14–18.
- 3. Lapola DM, Schaldacha R, Alcamoa J, Bondeaud A, Kocha J, Koelkinga C, et al. Indirect land-use changes can overcome carbon savings from biofuels in Brazil. Proc Natl Acad Sci USA. 2010; 107: 3388–3393. doi: 10.1073/pnas.0907318107. pmid:20142492
- 4. Dias-Filho MB. Diagnóstico das pastagens no Brasil. Belém, PA: Embrapa Amazônia Oriental; 2014.
- 5. Strassburg BBN, Latawiec AE, Barioni LG, Nobre CA, da Silva VP, Valentim JF, et al.When enough should be enough: Improving the use of current agricultural lands could meet production demands and spare natural habitats in Brazil. Global Environ Chang. 2014; 28: 84–97.
- 6. Fu B, Zhang L, Xu Z, Zhao Y, Wei Y, Skinner D. Ecosystem services in changing land use. J Soils Sediments 2015; 15: 833–843.
- 7. Zornoza R, Acosta JA, Bastida F, Domínguez SG, Toledo DM, Faz A. Identification of sensitive indicators to assess the interrelationship between soil quality, management practices and human health. Soil 2015; 1:173–185.
- 8. Karlen DL, Mausbach MJ, Doran JW, Cline RG, Harris RF, Schuman GE. Soil quality: A concept, definition, and framework for evaluation. Soil Sci Soc Am J. 1997; 61:4–10.
- 9. Mukherjee A, Lal R. Comparison of Soil Quality Index Using Three Methods. PLoS ONE 2014; 9(8): e105981. doi: 10.1371/journal.pone.0105981. pmid:25148036
- 10. Andrews SS, Karlen DL, Cambardella CA. The Soil Management Assessment Framework: A Quantitative Soil Quality Evaluation Method. Soil Sci Soc Am J. 2004; 68:1945–1962.
- 11. Karlen DL, Stott DE, A framework for evaluating physical and chemical indicators of soil quality. In: Doran JW, Coleman DC, Bezdicek DF, Stewart BA, editors. Defining soil quality for a sustainable environment. Madison, WI: SSSA, 1994; pp. 53–72.
- 12. Andrews SS, Karlen DL, Mitchell JP. A comparison of soil quality indexing methods for vegetable production systems in Northern California. Agric Ecosyst Environ. 2002; 90: 25–45.
- 13. Karlen DL, Ditzler CA, Andrews SS. Soil quality: why and how? Geoderma 2003; 114: 145–156.
- 14. Cardoso EJBN, Vasconcellos RLF, Bini D, Miyauchi MYH, Santos CA, Alves PRL, et al. Soil health: looking for suitable indicators. What should be considered to assess the effects of use and management on soil health? Sci Agric. 2013; 70: 274–289.
- 15. Lima ACR, Hoogmoed W, Brussaard L. Soil quality assessment in rice production systems: establishing a minimum data set. J Environ Qual. 2008; 37: 623–630. pmid:18396549
- 16. Armenise E, Redmile-Gordon MA, Stellacci AM, Ciccarese A, Rubino P. Developing a soil quality index to compare soil fitness for agricultural use under different managements in the Mediterranean environment. Soil Till Res. 2013; 130:91–98.
- 17. Chen YD, Wang HY, Zhou JM, Xing L, Zhu BL, Zhao YC, et al. Minimum data set for assessing soil quality in farmland of northeast China. Pedosphere 2013; 23: 564–576.
- 18. Askari MS, Holden NM. Indices for quantitative evaluation of soil quality under grassland management. Geoderma 2014; 230–231: 131–142.
- 19. Askari MS, Holden NM. Quantitative soil quality indexing of temperate arable management systems. Soil Till Res. 2015; 150: 57–67.
- 20. Sánchez-Navarro A, Gil-Vázquez JM, Delgado-Iniesta MJ, Marín-Sanleandro P, Blanco-Bernardeau A, Ortiz-Silla R. Establishing an index and identification of limiting parameters for characterizing soil quality in Mediterranean ecosystems. Catena 2015; 131:35–45.
- 21. Qi YB, Darilek JL, Huang B, Zhao YC, Sun WX, Gu ZQ. Evaluating soil quality indices in an agricultural region of Jiangsu Province, China. Geoderma 2009; 149: 325–334.
- 22. Xia M, Zhao B, Hao X, Zhang J. Soil Quality in Relation to Agricultural Production in the North China Plain. Pedosphere 2015; 25: 592–604.
- 23. Lima ACR, Hoogmoed W, Brussaard L. Farmers’ assessment of soil quality in rice production systems. NJAS-Wagen J Life Sc. 2011; 58: 31–38.
- 24. Tesfahunegn GB, Tamene L, Vlek PLG. Evaluation of soil quality identified by local farmers in Mai-Negus catchment northern Ethiopia. Geoderma 2011; 163: 209–218.
- 25. Doran JW, Parkin TB. Defining and assessing soil quality. In: Doran JW, Coleman DC, Bezdicek DF, Stewart BA, editors. Defining soil quality for a sustainable environment. Madison, WI: SSSA; 1994. pp.1–20.
- 26. Lima ACR, Brussaard L, Totola MR, Hoogmoed W, Goede RGM. A functional evaluation of three indicator sets for assessing soil quality. Appl Soil Ecol. 2013; 64: 194–200.
- 27. Masto RE, Chhonkar PK, Singh D, Patra AK. Alternative soil quality indices for evaluating the effect of intensive cropping, fertilization and manuring for 31 years in the semi-arid soils of India. Environ Monit Assess. 2008; 136: 419–435. pmid:17457684
- 28. de Paul Obade V, Lal R. A standardized soil quality index for diverse field conditions. Sci Total Environ. 2016; 541: 424–434. doi: 10.1016/j.scitotenv.2015.09.096. pmid:26410717
- 29. Karlen DL, Cambardella CA, Kovar JL, Colvin TS. Soil quality response to long-term tillage and crop rotation practices. Soil Till. Res. 2013; 133:54–64.
- 30. Nakajima T, Lal R, Jiang S. Soil quality index of a Crosby silt loam in central Ohio. Soil Till Res. 2015; 146: 323–328.
- 31. Romaniuk R, Giuffré L, Costantini A, Bartoloni N, Nannipieri P. A comparison of indexing methods to evaluate quality of soils: the role of soil microbiological properties. Soil Res. 2011; 49: 733–741.
- 32. Swanepoel PA, du Preez CC, Botha PR, Snyman HA, Habig J. Assessment of tillage effects on soil quality of pastures in South Africa with indexing methods. Soil Res. 2015; 53: 274–285.
- 33. Liu Z, Zhou W, Lv J, He P, Liang G, Jin H. A Simple Evaluation of Soil Quality of Waterlogged Purple Paddy Soils with Different Productivities. PLoS ONE 2015; 10(5): e0127690. doi: 10.1371/journal.pone.0127690. pmid:25997107
- 34. Cherubin MR, Franco ALC, Cerri CEP, Oliveira DMS, Davies CA, Cerri CC. Sugarcane expansion in Brazilian tropical soils—Effects of land-use change on soil chemical attributes. Agr Ecosyst Environ. 2015; 211: 173–184.
- 35. Raij B van, Andrade JC, Cantarella H, Quaggio JA. Análise Química Para Avaliação da Fertilidade de Solos Tropicais. Campinas, SP: Instituto Agronômico, 2001.
- 36. Bagarello V, Iovino M, Elrick D. A simplified falling-head technique for rapid determination of field-saturated hydraulic conductivity. Soil Sci Soc Am J. 2004; 68: 66–73.
- 37. Ball BC, Batey T, Munkholm LJ. Field assessment of soil structural quality–a development of the Peerlkamp test. Soil Use Manage. 2007; 23: 329–337.
- 38. Guimarães RML, Ball BC, Tormena CA. Improvements in the visual evaluation of soil structure. Soil Use Manage. 2011; 27: 395–403.
- 39. Marcolin CD, Klein VA. Determinação da densidade relativa do solo por uma função de pedotransferência para a densidade do solo máxima. Acta Sci Agron. 2011; 33: 349–354.
- 40. Ball BC, Hunter R. The determination of water release characteristics of soil cores at low suctions. Geoderma 1988; 43: 195–212.
- 41. Wienhold BJ, Karlen DL, Andrews SS, Stott DE. Protocol for Soil Management Assessment Framework (SMAF) soil indicator scoring curve development. Renew Agr Food Syst. 2009; 24: 260–266.
- 42. Reynolds WD, Bowman BT, Drury CF, Tan CS, Lu X. Indicators of good soil physical quality: density and storage parameters. Geoderma 2002; 110: 131–146.
- 43. Reynolds WD, Drury CF, Tan CS, Fox CA, Yang XM. Use of indicators and pore volume-function characteristics to quantify soil physical quality. Geoderma 2009; 152: 252–263.
- 44. Vance EC, Brookes PC, Jenkinson DS. An extraction method for measuring microbial biomass C. Soil Biol Biochem. 1987; 19: 703–707.
- 45. Tabatabai MA. Soil enzymes. In: Bottomley PS, Angle JS, Weaver RW. editors. Methods of Soil Analysis: Part 2 –Microbiological and Biochemical Properties. SSSA, 1994; pp. 775–833.
- 46. Anderson JM, Ingram JSI. Tropical Soil Biology and Fertility: A Handbook of Methods, second ed. Wallingford: CABI; 1993.
- 47. Magurran AE. Measuring Biological Diversity. London: Blackwell Publishing; 2004.
- 48. Greenwood KL, McKenzie BM. Grazing effects on soil physical properties and the consequences for pastures: a review. Aust J Exp Agric. 2001; 41: 1231–1250.
- 49. Pietola L, Horn R, Yli-Halla M. Effects of trampling by cattle on the hydraulic and mechanical properties of soil. Soil Till Res. 2005; 82, 99–108.
- 50. Hunke P, Roller R, Zeilhofer P, Schröder B, Mueller EN. Soil changes under different land-uses in the Cerrado of Mato Grosso, Brazil. Geoderma Regional 2015; 4: 31–43.
- 51. Benito NP, Brossard M, Pasini A, Guimaraes MD, Bobillier B. Transformations of soil macroinvertebrate populations after native vegetation conversion to pasture cultivation (Brazilian Cerrado). Eur J Soil Biol. 2004; 40: 147–154.
- 52. Decaens T, Jimenez JJ, Barros E, Chauvel A, Blanchart E, Fragoso C, et al. Soil macrofaunal communities in permanent pastures derived from tropical forest or savanna. Agr Ecosyst Environ. 2004; 103: 301–312.
- 53. Mello FFC, Cerri CEP, Davies CA, Holbrook NM, Paustian K, Maia SMF, et al. Payback time for soil carbon and sugar-cane ethanol. Nat Clim Change 2014; 4: 605–609.
- 54. Franco ALC, Cherubin MR, Pavinato PS, Cerri CEP, Six J, Davies CA, et al. Soil carbon, nitrogen and phosphorus changes under sugarcane expansion in Brazil. Sci. Total Environ. 2015; 515–516, 30–38.
- 55. Silva-Olaya AM, Cerri CEP, La Scala N Jr, Dias CTS, Cerri CC. Carbon dioxide emissions under different soil tillage systems in mechanically harvested sugarcane. Environ Res Lett. 2013; 8: 1–8.
- 56. Cerri CC, Galdos MV, Maia SMF, Bernoux M, Feigl BJ, Powlson D, et al. Effect of sugarcane harvesting systems on soil carbon stocks in Brazil: an examination of existing data. Eur J Soil Sci. 2011; 62: 23–28.
- 57. Centurion JF, Freddi OS, Aratani RG, Metzner AFM, Beutler AN, Andrioli A. Influência do cultivo de cana-de-açúcar e da mineralogia da fração argila nas propriedades físicas de Latossolo Vermelho. R Bras Ci Solo 2007; 31: 199–209.
- 58. Six J, Elliott ET, Paustian K. Soil macroaggregate turnover and microaggregate formation: a mechanism for C sequestration under no-tillage agriculture. Soil Biol Biochem. 2000; 32: 2099–2103.
- 59. Otto R, Silva AP, Franco HCJ, Oliveira ECA, Trivelin PCO. High soil penetration resistance reduces sugarcane root system development. Soil Till Res. 2011; 117: 201–210.
- 60. Souza GS, Souza ZM, Silva RB, Barbosa RS, Araújo FS. Effects of traffic control on the soil physical quality and the cultivation of sugarcane. R Bras Ci Solo 2014; 38: 135–146.
- 61. Weill MAM, Sparovek G. Estudo da erosão na microbacia do Ceveiro (Piracicaba, SP): I—Estimativa das taxas de perda de solo e estudo de sensibilidade dos fatores do modelo EUPS. R Bras Ci Solo 2008; 32: 801–814.
- 62. Martinelli LA, Filoso S. Polluting effects of Brazil’s sugar-ethanol industry. Nature 2007; 445: 364–364.
- 63. Wagg C, Bender SF, Widmer F, van der Heijden MGA. Soil biodiversity and soil community composition determine ecosystem multifunctionality. Proc Natl Acad Sci USA. 2014; 111: 5266–5270. doi: 10.1073/pnas.1320054111. pmid:24639507
- 64. Lal R. Soil carbon sequestration impacts on global climate change and food security. Science 2004; 304: 1623–1627. pmid:15192216
- 65. Karlen DL, Andrews SS, Wienhold BJ, Zobeck TM. Soil quality assessment: Past, present and future. Electronic J Integr Biosci. 2008; 6: 3–14.
- 66. Raij B van, Cantarella H, Quaggio JA, Furlani AMC. Recomendações de adubação e calagem para o Estado de São Paulo. 2° Ed. rev. Campinas, SP: IAC; 1997 (Boletim Técnico, 100).
- 67. Comissão de Química e Fertilidade do Solo (CQFS-RS/SC). Manual de adubação e calagem para os estados do Rio Grande do Sul e Santa Catarina. Porto Alegre, RS: Sociedade Brasileira de Ciência do Solo—Núcleo Regional Sul; 2004.
- 68. Reichert JM, Reinert DJ, Braida JA. Qualidade dos solos e sustentabilidade de sistemas agrícolas. Ci. Ambiente 2003; 27: 29–48.
- 69. Reichert JM, Suzuki LEAS, Reinert DJ, Horn R, Hakansson I. Reference bulk density and critical degree-of-compactness for no-till crop production in subtropical highly weathered soils. Soil Till Res. 2009; 102: 242–254.
- 70. Arshad MA, Lowery B, Grossman B. Physical tests for monitoring soil quality. In: Doran JW, Jones AJ (editors). Methods for assessing soil quality. Madison: SSSA, 1996; p.123–141.
- 71. USDA-NRCS (Natural Resources Conservation Service). Soil quality test kit guide. 2001. Available: http://www.nrcs.usda.gov/Internet/FSE_DOCUMENTS/nrcs142p2_050956.pdf
- 72. Spohn M, Giani L. Impacts of land use change on soil aggregation and aggregate stabilizing compounds as dependent on time. Soil Biol Biochem. 2011; 43: 1081–1088.
- 73. Lopes AAC, Souza DMG, Chaer GM, Reis Junior FB, Goedert WJ, Mendes IC. Interpretation of Microbial Soil Indicators as a Function of Crop Yield and Organic Carbon. Soil Sci Soc Am J. 2013; 77: 461–472.
- 74. Bartz MLC, Pasini A, Brown GG. Earthworms as soil quality indicators in Brazilian no-tillage systems. Appl Soil Ecol. 2013; 69: 39–48.