• Open Access

Infield greenhouse gas emissions from sugarcane soils in Brazil: effects from synthetic and organic fertilizer application and crop trash accumulation

Authors


Correspondence: Janaina Braga do Carmo, tel. + 55 15 32 295 948, fax + 55 15 32 296 001, e-mail: jbcarmo2008@gmail.com

Abstract

Bioethanol from sugarcane is becoming an increasingly important alternative energy source worldwide as it is considered to be both economically and environmentally sustainable. Besides being produced from a tropical perennial grass with high photosynthetic efficiency, sugarcane ethanol is commonly associated with low N fertilizer use because sugarcane from Brazil, the world's largest sugarcane producer, has a low N demand. In recent years, several models have predicted that the use of sugarcane ethanol in replacement to fossil fuel could lead to high greenhouse gas (GHG) emission savings. However, empirical data that can be used to validate model predictions and estimates from indirect methodologies are scarce, especially with regard to emissions associated with different fertilization methods and agricultural management practices commonly used in sugarcane agriculture in Brazil. In this study, we provide in situ data on emissions of three GHG (CO2, N2O, and CH4) from sugarcane soils in Brazil and assess how they vary with fertilization methods and management practices. We measured emissions during the two main phases of the sugarcane crop cycle (plant and ratoon cane), which include different fertilization methods and field conditions. Our results show that N2O and CO2 emissions in plant cane varied significantly depending on the fertilization method and that waste products from ethanol production used as organic fertilizers with mineral fertilizer, as it is the common practice in Brazil, increase emission rates significantly. Cumulatively, the highest emissions were observed for ratoon cane treated with vinasse (liquid waste from ethanol production) especially as the amount of crop trash on the soil surface increased. Emissions of CO2 and N2O were 6.9 kg ha−1 yr−1 and 7.5 kg ha−1 yr−1, respectively, totaling about 3000 kg in CO2 equivalent ha−1 yr−1.

Introduction

The last decade has seen a tremendous shift in the energy system toward the use of biofuels (Food & Agriculture Organization (FAO), 2008) not only because of a growing demand for energy worldwide, but also because of climate change concerns (Fargione et al., 2008). As a consequence, production of sugarcane ethanol has gained global center stage for being considered one of the most economic and sustainable biofuel feedstocks in the world (Goldemberg, 2007).

Sugarcane ethanol is considered to be economic and sustainable primarily because it is produced from a tropical perennial grass with high photosynthetic efficiency that can re-grow up to five times after the first harvest. In addition, sugarcane grown in Brazil, the largest producer in the world, requires less synthetic nitrogen (N) fertilizer than other commonly used biofuel feedstocks such as corn (Heffer & Prud'homme, 2008). Since the use of synthetic N fertilizer in agriculture is one of the main sources of N2O emissions to the atmosphere (Monsier, 1989), lower rates of fertilizer application in Brazilian sugarcane could potentially help reduce emissions of greenhouse gases (GHG) associated with energy production (Goldemberg, 2008).

Recent studies (Boddey et al., 2008; Galdos et al., 2010; Lisboa et al., 2011) have predicted GHG emission savings from the use of sugarcane biofuel produced in Brazil can be substantial. However, because the predictions are based on indirect estimates of GHG emissions in sugarcane agriculture (e.g. Intergovernmental Panel on Climate Change (IPCC), 2007) due to the lack of field data from Brazil, uncertainties are large. Moreover, different management practices that are commonly used in sugarcane cropping systems in Brazil are likely to result in different rates of GHG emissions, which are not only largely unknown, but also not considered in the IPCC estimates.

One of the most common management practices used in sugarcane agriculture in Brazil, which can affect GHG emission rates, is the full recycling of waste products from ethanol production to sugarcane fields in the form of organic fertilizer (Mutton et al., 2010). Organic fertilizer is used in liquid and solid forms as vinasse and filter cake, respectively, both of which are rich in carbon (C) and nitrogen (N) (Macedo et al., 2008), and can affect key biogeochemical processes associated with GHG emissions such as decomposition, respiration, denitrification, and nitrification.

Other important management practices that can potentially affect GHG emissions from sugarcane soils are associated with the harvesting method used. Until recently, sugarcane in Brazil was harvested manually and included the pre-burning of fields to eliminate extraneous leafy straw. As manual harvesting is increasingly substituted by mechanized methods due to a gradual ban on pre-burning practices established by the federal and state governments in Brazil in the early 2000s, the accumulation of sugarcane crop residue (trash) on the soil surface is increasing. However, although changes in GHG emissions associated with the ban of the burning process have been estimated (Galdos et al., 2010), changes associated with crop residue (trash) accumulation are essentially unknown.

In this study, we conducted in situ experiments to assess emission rates of three GHG (CO2, N2O, and CH4) in sugarcane fields in Brazil under different fertilizer application practices typically used in the main stages of the crop, and with increasing crop trash accumulation on the soil surface which results from mechanized harvesting. More specifically, we determined whether or not GHG emissions would vary with 1) the fertilization method used in the first part of the sugarcane cycle (plant cane), and 2) the accumulation of crop trash on the soil surface during the subsequent sugarcane cycle (ratoon cane), with and without the application of liquid waste (vinasse) on the field.

Material and methods

Site description and crop management practices

The study was simultaneously conducted in two different regions to include sites with sugarcane undergoing the two main stages of the crop life cycle (i.e. plant cane and ratoon cane). The two regions, Jaú (22o15′S; 48o34′W) and Piracicaba (22o41′S; 47o33′W), have a long history of sugarcane cultivation, going back at least 100 years, and are located within 80 km from each other in the state of São Paulo, where 60% of the sugarcane is produced in Brazil (Fig. 1).

Figure 1.

Map of Brazil highlighting the state of São Paulo in the Southeast region of the country, and showing the location of the study sites. São Paulo state was responsible for 60% of the country's sugarcane production in 2010–2011.

The specific study sites in each region have had sugarcane planted for 20 years, and have similar topography and climate [Aw (Köppen)] with average annual precipitation around 1390 mm, and an annual average temperature near 21 °C. However, whereas the dominant soil type at the Jaú site is a Haplic Lixisol, the dominant soil at Piracicaba is a Haplic Ferralsol (FAO) (Table 1).

Table 1. Soil characteristics in the experimental sites in the Jaú and Piracicaba regions in the state of São Paulo, Brazil
SiteSoil depth (m)pH CaCl2OM (g dm−3)P (mg dm−3)CEC (mmolc dm−3)V (%)Sand (%)Silt (%)Clay (%)
  1. OM, organic matter; CEC, cation exchange capacity; P, phosphorus content; V, base saturation.

Jaú0–0.24.522224645672211
 0.2–0.44.215103635582022
Piracicaba0–0.24.52216762462929
 0.2–0.44.418226022621028

The sugarcane at the Jaú site was in the first stage of the crop cycle, which is called plant cane. This stage is the first after the planting of sugarcane, therefore, it is usually preceded by soil disturbance associated with tillage. Sugarcane in the plant cane stage is harvested after 16–18 months and the subsequent stages (3–7 ratoon stages) originate from the re-growth of cane stubbles.

Sugarcane at the Piracicaba site had been harvested only once prior to the experiment and thus was in the first ratoon stage. In the ratoon cane stage, the sugarcane re-growths are harvested every 12 months for 3–7 stages, depending on soil type, climate, and cane variety. If the harvest is mechanized (with no pre-burning), crop trash accumulates on the soil surface since only the stalks are harvested and the leaves are left on the field. After the last ratoon cycle, the sugarcane field is commonly re-planted (reformed).

The nutritional demands and conditions for sugarcane undergoing different stages of the lifecycle are different. Therefore, fertilizer application methods are somewhat different for each stage. For instance, in the plant stage of the sugarcane, mineral fertilizer (N, P, and K) is typically applied in planting furrows, whereas in the ratoon stage, mineral fertilizer containing N and K (and less frequently P) is applied onto the soil surface since the crop is already established and the fertilizer cannot be incorporated into the soil. Incorporating fertilizer into the soil is especially difficult if the cane had been previously harvested unburned since a thick mulch of plant material usually accumulates on the soil surface.

Urea is the most commonly used mineral N fertilizer used in Brazilian agriculture. However, because urea is subject to high volatilization losses when surface-applied to soils, urea in Brazilian sugarcane is only used in the plant cane stage. In the ratoon stage, mineral N fertilizer is usually applied on the soil surface in the form of ammonium nitrate to prevent losses (Cantarella et al., 2008).

Filter cake, a solid organic residue of the cane processing in the mill and rich in P, is also typically used in the plant cane stage with mineral fertilizer added in the furrows, but it is not used in ratoon cane. However, in both stages, vinasse, a liquid residue of the ethanol distillation, is broadcast on the soil mainly as a source of K in addition to organic matter and other nutrients.

More than 30 varieties of sugarcane are available to farmers in Brazil. The varieties planted in Jaú and Piracicaba were IAC 95–5000 and CTC 15, respectively, and are among the sugarcane commonly recommended for these particular regions. The general C : N ratio of crop residues is about 100 : 1 (Fortes et al., 2012).

Experimental design and treatments

Experiment with plant cane (Jaú site)

The experiment in the Jaú site was conducted to determine whether or not emissions of GHG in plant cane vary with different fertilization methods commonly used in Brazil during this phase of the sugarcane crop. Therefore, the experiment included a series of treatments combining mineral fertilizer and organic fertilizers (vinasse and filter cake). A control treatment without any fertilizer was also used to determine background emissions.

The experiment started on May 2010, using randomized blocks composed of five treatments and four blocks totaling 20 plots in an area over 1100 m2. Each plot measured 7 × 8 m and contained five 8-m long rows planted with sugarcane, spaced at 1.4 m. Accordingly, the plant cane experiment in the Jaú included the following treatments:

  • T1 – mineral fertilizer containing N, P, and K;
  • T2 – mineral fertilizer containing N and K; filter cake added as source of P;
  • T3 – mineral fertilizer containing N and P; vinasse added as source of K;
  • T4 – mineral fertilizer containing N; filter cake and vinasse added as sources of P and K;
  • C – control, with no fertilizer (organic or mineral).

Mineral fertilizer containing N was applied in the form of urea (450 g N kg−1) in 30-cm deep planting furrows at a rate of 60 kg N ha−1, which is the average amount used for plant cane in the study region (Espironelo et al., 1997; Filoso et al., 2003). Filter cake was applied in furrows at a rate of 30 t ha−1 wet mass, or 19.4 t ha−1 dry mass (33.6 kg per row, wet mass), also based on the amount recommended for the region. Mineral P (triple superphosphate, 410 g P2O5 kg−1) and K (potassium chloride, 600 g K2O kg−1) were added to furrows at rates of 120 and 30 kg ha−1 of K2O and P2O5, respectively (Espironelo et al., 1997).

After mineral fertilizer and filter cake were evenly spread at the bottom of the furrows, 30- to 50-cm seed stem sections of sugarcane were placed inside and covered with soil. Subsequently, vinasse was sprayed over the entire experimental area (i.e., not only in furrows) at an amount equivalent to 100 × 103 L ha−1 (560 L per plot).

To prevent ground water salinization, the amount of vinasse applied in sugarcane plantations in the state of Sao Paulo is regulated according to the concentration and soil exchangeable K (CETESB, 2006). Therefore, because the K concentration of vinasse used in the Piracicaba experiment was relatively high in comparison to that used in Jaú (Table 2), the volume applied was about half of that used in Jaú. Yet, both sites received about 25 kg N ha−1 in the form of vinasse (Table 2).

Table 2. Chemical and biological characteristics of the vinasse applied in the experiments with plant and ratoon cane in the Jaú and Piracicaba sites, respectively, and of the filter cake used in plant cane in the Jaú experiment
ParameterJaú VinassePiracicaba VinasseFiltercake
  1. a

    Biological oxygen demand.

  2. b

    Chemical oxygen demand.

  3. c

    Determined according to COD values.

pH 20 °C4.54.06.1
Moisture (%)----------55
Eletrical conductivity (mS cm−1)6.48.5-----
BODa (gL−1)5.86.2-----
CODb (gL−1)22.622.3-----
Total Nitrogen (mgL−1)2704014.1 (g kg−1)
Organic Carbon (gL−1)8.5c8.4c140 (g kg−1)
Potassium (K mgL−1)204120922582
Phosphorus (P2O5 mgL−1)27,387627

Experiment with first ratoon cane (Piracicaba site)

The experiment at the Piracicaba site was conducted in ratoon cane to determine whether or not the accumulation of crop trash that result from mechanized harvesting and no pre-burn practices changes emissions of GHG. The effects of the use of vinasse in ratoon cane on GHG emissions were also assessed.

The experiment started on November 19, 2010 with sugarcane in the first ratoon stage. The sugarcane at this site was mechanically harvested on October 27, 2010, just a few weeks before the onset of the experiment. Chambers were installed in the field 1 day before fertilizer and vinasse application on November 19, 2010. Gas sample collection started immediately after fertilization.

The experiment was conducted using a randomized block design with a 4 × 2 factorial treatment containing four replicates and a total of 32 plots with five 10-m long rows of sugarcane in 1.4-m intervals. The factors corresponded to the application or not of vinasse in plots with four different levels of crop trash on the soil surface (0, 7, 14 or 21 Mg ha−1, dry material). The amount of trash added simulated amounts observed in sugarcane fields in the region in the absence of pre-harvest burning (Paes & Oliveira, 2005).

Overall, two plots were used as control treatments to account for background emissions, whereas the other plots were treated with mineral fertilizer, following common practices used for ratoon cane in the region.

  • The treatments with ratoon cane (in Piracicaba) included the following:
  • T0 – no trash; mineral fertilizer application plus vinasse (V+) and no vinasse (V−).
  • T7 – 7 Mg of trash; mineral fertilizer application plus vinasse (V+) and no vinasse (V−).
  • T14 – 14 Mg of trash; mineral fertilizer application plus vinasse (V+) and no vinasse (V−).
  • T21 – 21 Mg of trash; mineral fertilizer application plus vinasse (V+) and no vinasse (V−).
  • C – Control, with no fertilizer (organic or mineral).

The trash material used in the experiment was a product of the previous sugarcane harvest at the same site. This material was composed of dry and green sugarcane leaves removed from the area, oven dried at 65 °C to a constant weight, and then transported back to the plots to simulate different levels of trash accumulation. After the trash material was placed onto the soil, the experimental plots were fertilized with 120 kg N ha−1 (ammonium nitrate), and 30 kg P2O5 ha−1 (triple super phosphate), according to the rates recommended for sugarcane in the region (Van Raij et al., 2007). Potassium fertilizer (120 kg K2O ha−1 as potassium chloride) was applied only in plots where no vinasse was used, to prevent K excess. All the mineral fertilizers were surface-applied in a 0.2-m wide band about 0.1 m from the plant row. Vinasse was sprayed over the entire experimental area and on top of the mineral fertilizer at a rate of 56 m3 ha−1 using a motorized pump fit with a flow regulator.

Sample collection and calculation of gas fluxes

In the ratoon cane experiment (Piracicaba site), gas samples were collected on an event basis during a period of 335 days after fertilizer application. Samples were collected using PVC chambers installed in the experimental plots, according to the chamber technique described by Davidson & Schimel (1995) and Allen et al. (2010).

Each plot had three chambers; two in crop rows (on the fertilizer band) and one in a mid row. Overall, 416 samples were collected in 102 chambers during 21 sampling events. The first sampling events occurred immediately after fertilizer application and on the day after, whereas the following sampling events in the first month of the experiment were spaced first by 2 days and later by 3 days. After the first month, sampling became less frequent and occurred about once every 3 weeks. The last four sampling events of the experiment were spaced by 40-day intervals.

At the plant cane experiment in Jaú, samples were also collected in events within a period of 314 days with PVC chambers, but only 54 chambers were used in the experimental plots. Each plot where vinasse was applied had three chambers; two in crop rows and one in a mid row. Where vinasse was not applied, there were only two chambers; one in a crop row and one in a mid row. In total, 216 gas samples were collected during 20 sampling events in the plant cane experiment. Sampling started immediately after fertilizer application and crop trash addition, and was spaced in a similar fashion as in the ratoon cane experiment in Piracicaba, i.e. with more frequent sampling in the first couple of months of the experiment than later on.

The chambers used for gas sampling followed the design described by Varner et al. (2002). They consisted of a round PVC base (30 cm dia., 20 cm height, 0.014 m3), with a 10-cm deep lid containing a small valve to prevent overheating and subsequent increase in the chamber's internal pressure. The chambers were installed by inserting the base into the soil to a depth of 3 cm.

Gas samples were collected using a 60 mL BD plastic syringe (Becton, Dickinson and Co., Franklin Lakes, NJ, USA) and immediately placed in 30 mL glass vials closed with rubber stoppers (Bellco Glass, NJ, USA) , following the procedure described by Sousa Neto et al. (2011). During each sampling event, four samples were collected from each chamber for a period of 30 min. The first sample was collected 1 min after the chamber was closed and the remaining samples after 10, 20, and 30 min.

All samples were analyzed at the São Carlos Federal University, Sorocaba, Brazil, using gas chromatographer Shimadzu GC-2014 (Shimadzu Co., Columbia, MD, USA). The chromatographer was equipped with a packed column, an electron capture detector to analyze N2O, and a flame ionization detector to quantify CO2 and CH4. Prior to detection, CO2 was reduced to CH4 using a methanizer. The method used has been fully described by Keller & Reiners (1994) and Varner et al. (2002).

To calculate emission rates, the gas molar volumes were corrected for ambient air temperature and atmospheric pressure measured at the sampling time. GHG fluxes (f) were calculated according to Jantalia et al. (2008) as:

display math

where, ∆C/∆t is the change in GHG concentration inside of the chamber during the period (∆t) that the chamber is closed; V and A are, respectively, the volume of the chamber and the area of soil covered by the chamber; m is the molecular weight of each GHG (CO2, CH4, and N2O). The emission rate for each experiment was computed using a linear regression based on the curve generated from the gas values measured along the 30 min intervals.

Calculation of cumulative fluxes and emission factors

Cumulative fluxes on a hectare basis were calculated for each treatment using emission values observed in the crop rows and mid rows in accordance to the method suggested by Allen et al. (2010). We estimated that crop rows accounted for 16.8% of the total experimental area and that the space between rows (mid rows) for 83.2%. Annual cumulative estimates of GHG were calculated by linear interpolation between sampling events (Allen et al., 2010).

Emission factors for N2O were calculated by subtracting the cumulative emission rates from plots treated with N fertilizer from those that were not. The calculations followed the methods used by the IPCC (2006) and Allen et al. (2010).

Statistical analyses

We first tested GHG fluxes for normality and subsequently transformed the data using the box-cox transformation method (Statistica,version 10) since they did not follow a normal distribution. ancova was then used on the transformed data to determine whether or not GHG emissions varied among treatments. We considered two different sampling periods: (i) the intensive sampling period that for the Piracicaba experiment went from November 11 to December 17, 2010, and for the Jaú experiment from May 26 to June 22, 2010; (ii) the extensive sampling period that for the Piracicaba went from January 10 to November 1, 2011, and for the Jaú experiment from August 16, 2010 to March 24, 2011. For the Piracicaba data, the test was conducted for the different levels of trash accumulated in the soil (treatments) with or without vinasse application as the categorical factors for the intensive and extensive sampling periods. For the Jaú data the treatments were considered as categorical factors for the intensive and extensive sampling periods. For both experiments, the dependent variables were the GHG fluxes and the co-variables soil temperature and water-filled pore space (WFPS). If the ancova was significant, a post hoc Tukey Honest Test for Unequal Variance was used to test differences among treatments.

Results

Daily gas fluxes during sampling events

Experiment with plant cane (Jaú site) – As expected, daily fluxes of N2O and CO2 in plant cane varied significantly among treatments (ancova F6,38 = 5.62; r2 = 0.39; P = 0.0003 for N2O, and F6,38 = 3.43; r2 = 0.24; P = 0.0082 for CO2), but only in the period following fertilizer application (Fig. 2). In general, fluxes of N2O and CO2 in plant cane treated with organic and mineral fertilizers were higher than in that treated only with mineral fertilizer (T1), and higher than in the control treatment (C) (Fig. 2a and b). However, the results from our Tukey Honest Test for Unequal Variance showed that although daily fluxes in all treatments with organic fertilizers (T2, T3, T4) were significantly different from those in the control treatment (C), only fluxes from T4, which included the application of two types of organic fertilizers (filtercake and vinasse) in combination with mineral fertilizer, were significantly different from those observed in the treatment that included only mineral fertilizer (T1) (P = 0.048 for N2O, P = 0.050 for CO2). Gas fluxes from T1 were not significantly different even from those observed in the control treatment.

Figure 2.

Daily mean fluxes of N2O (a), CO2 (b) and CH4 (c) measured in plant cane during the experiment conducted in the Jaú site between May 2010 and April 2011. The different treatments in the plant cane experiment are represented as: (T1) application of inorganic fertilizer containing N, P, and K; (T2) application of mineral N fertilizer and filter cake as source of K and P; (T3) application of mineral fertilizer containing N and P, plus vinasse as source of K; (T4) application of mineral N fertilizer, filter cake as source of P, and vinasse as source of K. The Control experiment, which excluded any form fertilizer, mineral and organic, is represented as (C).

In contrast to N2O and CO2, daily CH4 fluxes did not differ significantly among treatments (Fig. 2c). In the beginning of the experiment, CH4 fluxes in treatments containing filter cake increased slightly in relation to the control treatment. However, only the treatment with vinasse plus mineral fertilizer (T3) had pronounced changes in daily emissions toward the end of the experiment. Yet, the variance in daily fluxes among sampling dates was larger than the variance among treatments, resulting in no significant differences among treatments.

Experiments with first ratoon cane (Piracicaba site) – Even though methods of fertilization differed for sugarcane in the ratoon and plant cane stages, as described above, the daily GHG fluxes followed similar patterns. Firstly, fluxes of N2O and CO2 varied significantly among treatments (ancova F10,115 = 14.5; r2 = 0.51; P < 0.0001 and ancova F10,115 = 9.51; r2 = 0.41; P < 0.0001, respectively). Moreover, differences in CH4 daily fluxes among treatments were not statistically significantly, with or without vinasse (Figs. 3 and 4).

Figure 3.

Daily mean fluxes of N2O (a, b), CO2 (c, d) and CH4 (e, f) measured in ratoon cane without vinasse application during the experiment in the Piracicaba site between November 2010 and July 2011. The treatments without vinasse application are represented by (V−). Fluxes from crop rows are shown in the left panels (plots a, c, and e), and fluxes from midrows shown in the right panels (plots b, d, and f). The different levels of crop trash added to the soil surface are indicated as: (T0) no crop trash; (T7) 7 Mg crop trash ha−1; (T14) 14 Mg crop trash ha−1; (T21) 21 Mg cane trash ha−1. The control treatment, without application of the any type of fertilizer (organic nor inorganic) is represented by (C).

Figure 4.

Daily mean fluxes of N2O (a, b), CO2 (c, d) and CH4 (e, f) measured in ratoon cane with vinasse application (V+) during the experiment in the Piracicaba site between November 2010 and July 2011. Fluxes from crop rows are shown in the left panels (plots a, c, and e), and fluxes from mid rows shown in the right panels (plots b, d, and f). The different levels of crop trash added to the soil surface are indicated as: (T0) no crop trash; (T7) 7 Mg crop trash ha−1; (T14) 14 Mg crop trash ha−1; (T21) 21 Mg cane trash ha−1. The control treatment, without application of the any type of fertilizer (organic nor inorganic) is represented by (C).

Daily fluxes of N2O and CO2 increased in the first few months of the experiments (Fig. 3 a and c), decreasing exponentially to near background values toward the end. In all experiments, however, N2O and CO2 fluxes increased more in crop rows than in mid rows, and were substantially higher in treatments with vinasse (V+) than without vinasse (V−) (Fig. 4).

Regardless of the trash accumulation level (0, 7, 14, 21) in treatments with vinasse (V+), N2O daily fluxes in treatment plots were significantly higher than in the control plots (C) (P = 0.0001). The same pattern was observed in the mid rows (P < 0.05). However, in the absence of vinasse (V−), N2O fluxes in the treatment plots were significantly higher than in the control plots only when there was 14 Mg ha−1 of trash or more on the soil surface (P = 0.0002 and P = 0.0001).

In contrast to N2O, daily fluxes of CO2 in the presence of vinasse (V+) were significantly higher in treatment plots than in the control plots only when there was at least 7 t ha−1 of crop trash on the soil surface (P = 0.0001). Furthermore, even in the absence of vinasse and in the mid rows, all treatments resulted in significantly higher CO2 fluxes in comparison to control plots.

For CH4, daily fluxes in ratoon cane without vinasse (V−) increased in all treatments after the beginning of the experiment, but quickly decreased to control levels afterward. In the experiments with vinasse (V+) the decrease was more accentuated resulting in negative values. Also, fluxes in treatments with vinasse (V+) were similar in rows and mid rows (Figs. 3e and 4e). However, regardless of the use of vinasse and of the amount of trash on the soil surface, fluxes did not vary significantly among treatments (anova, P > 0.001).

In summary, the use of vinasse (V+) in ratoon cane resulted in higher fluxes than the absence of vinasse (V−), and fluxes were usually higher in crop rows than in mid rows. When vinasse was used, ratoon cane with 14 and 21 Mg of crop trash ha−1 on the soil surface (T14, T21, V+) had the highest N2O flux rates (Fig. 4a), whereas ratoon cane with 21 Mg of trash ha−1 had the highest rates of CO2 fluxes (Fig. 4c). For CH4, the highest fluxes were observed in ratoon cane with 14 Mg of trash ha−1 with vinasse. There was no correlation between N2O and CO2 fluxes.

The main difference among the gases measured was that although N2O and CH4 fluxes in mid rows with vinasse were about half of those in crop rows, fluxes of CO2 in mid rows and crop rows were about the same (Figs. 3c and 4c). Furthermore, fluxes of CO2 were equivalent in the presence or absence of vinasse.

Cumulative emissions

Experiment with plant cane (Jaú site) – By extrapolating the observed daily fluxes data, we estimated that the use of organic fertilizer combined with mineral fertilizer (T2, T3, and T4) resulted in higher cumulative N2O emissions on an annual basis than the use of mineral fertilizer alone. Even when N2O background emissions are taken into account, it is clear that the treatment with mineral fertilizer alone (T1) had lower cumulative emissions on an annual basis than the treatments with organic fertilizers. For instance, cumulative N2O emissions from T1 was only a third of those observed in treatments where vinasse was added (T3 and T4, Fig. 5 a), and only about half of those in the treatment with filter cake added to mineral fertilizer (T2). Similarly, cumulative emissions of CO2 were significantly higher in treatments with vinasse (T3 and T4) than with mineral fertilizer alone (T1).

Figure 5.

Cumulative fluxes of N2O, CO2 and CH4 in plant cane (Jaú) and ratoon cane (Piracicaba). The top three panels (a–c) show fluxes from plant cane; The three mid panels (d–f) show fluxes from ratoon cane without vinasse (V−), and the lower three panels (g, h and i) show fluxes from ratoon cane with vinasse (V+). The different treatments with plant cane are represented as: (T1) application of inorganic fertilizer containing N, P and K; (T2) application of mineral N fertilizer and filter cake as source of K and P; (T3) application of mineral fertilizer containing N and P, plus vinasse as source of K; (T4) application of mineral N fertilizer, filter cake as source of P, and vinasse as source of K; (C) control, where no fertilizer (mineral or organic) was used. The different levels of crop trash added to the soil surface in the experiment with ratoon cane (Piracicaba) are indicated as: (0) no crop trash; (7) 7 Mg crop trash ha−1; (14) 14 Mg crop trash ha−1; (21) 21 Mg crop trash ha−1. The control treatment, without application of the any type of fertilizer (organic nor inorganic), is represented by (C). The roman numerals in each plot indicate statistically significant differences among treatments and control (P < 0.05).

Cumulatively, treatments where vinasse was used (T3 and T4) seemed to be net sinks of CH4, whereas treatments with filter cake added to mineral fertilizer but without vinasse or with mineral fertilizer alone were net sources (Fig. 5c).

Experiment with first ratoon cane (Piracicaba site) – In contrast to the experiments with plant cane, cumulative emissions of N2O generally increased with the application of mineral fertilizer and with the amount of trash covering the soil regardless of the presence of vinasse (Table 4). However, because emissions in the experiment with vinasse increased not only in the crop rows but in mid rows, cumulative emissions values were about twice as high in the experiment with vinasse than without it (Fig. 5 d and g). For instance, in plots with 21 Mg trash ha−1, cumulative N2O emissions were as high as 9000 g ha−1 yr−1 when vinasse was used (V+) in contrast to about 4500 g ha−1 yr−1 when no vinasse was used (V−) (Fig. 5d and g).

For CO2, cumulative emissions were practically the same in the treatments with and without vinasse (Fig. 5e and h), especially when background emissions are taken into account. In both experiments, however, trash accumulation increased CO2 emissions well beyond the values observed in the control treatment. For CH4, there is was no clear pattern among treatments (Fig. 5f and i).

Comparing Cumulative Emissions among Experiments and Treatments

To provide a comparison of emissions between experiments and among all treatments within each experiment, we converted emissions of N2O, CO2, and CH4 to CO2- equivalent and also calculated emission factors for N2O (Tables 3 and 4). Overall, emissions of all gases in CO2-eq were higher in the experiments with plant cane (Jaú site) than with ratoon cane (Piracicaba site).

Table 3. N fertilizer emission factor and emission in CO2 equivalent based on rates of N fertilizer application used in the ratoon cane cycle experiment (Piracicaba site)
Treatments (crop trash level) (Mg ha−1)VinassebAdded N (kg ha−1)aEmission factor (%)cCO2 equivalent (kg CO2 eq ha−1 yr−1)
  1. a

    Emission factor minus background emission observed in the control treatment (i.e., without any N fertilizer). Values were calculated according to the Intergovernmental Panel on Climate Change (IPCC) (2006) and Allen et al. (2010). Data represent means and standard deviation.

  2. b

    Calculated as a function of the amount of N fertilizer applied per hectare.

  3. c

    Calculated according to the Intergovernmental Panel on Climate Change (IPCC), 2007, assuming that N2O GWP = 296 and CH4 GWP = 25.

0With1420.59 ± 0.291289
 Without1200.68 ± 0.41382
7With1421.19 ± 0.841620
 Without1200.96 ± 0.46540
14With1421.89 ± 1.003005
 Without1200.76 ± 0.30427
21With1423.03 ± 1.223060
 Without1202.03 ± 1.151141
Table 4. N fertilizer emission factor and CO2 equivalent based on the N rates applied during the plant cane cycle in Jaú experiment
TreatmentN sourceaAdded N (kg ha−1)bEmission factor (%)cCO2 equivalent (kg CO2 eq ha−1 yr−1)
  1. a

    Calculated as a function of the amount of N fertilizer applied per hectare.

  2. b

    Emission factor minus the N-N2O background emission observed in crop row without N fertilizer. Values were calculated according the Intergovernmental Panel on Climate Change (IPCC) (2006) and Allen et al. (2010). Data represent means and standard deviation.

  3. c

    Calculated according to the Intergovernmental Panel on Climate Change (IPCC), 2007, assuming that N2O GWP = 296 and CH4 GWP = 25.

  4. d

    The total mid row area was 7857 m2.

ControlNone---------------
T1Urea601.11 ± 0.75312
T2Urea + Filter cake1221.10 ± 0.54630
T3Urea + Vinasse872.65 ± 1.131375
T4Urea + Filter cake + Vinasse1491.56 ± 1.011386
MR without vinasse--------------------
dMR with vinasseVinasse212.99 ± 1.1296

In both, plant cane and ratoon cane, the highest values in CO2-eq were linked to the use of vinasse. In plant cane, the use of vinasse resulted in similar values for T3 and T4 despite the fact that T4 included mineral fertilizer (urea) plus filter cake, whereas T3 included only mineral fertilizer (as urea). In ratoon cane, the highest emission values in CO2-eq in the presence of vinasse were associated with the accumulation of at least 14 Mg of crop trash ha−1 on the soil surface. However, when no vinasse was applied but 21 Mg of crop trash ha−1 was added to the soil surface, emissions from ratoon cane were similar to those in plant cane treatments with the use of vinasse plus mineral fertilizer and filter cake (Tables 3 and 4).

In plant cane, the maximum emission value in CO2-eq was about 1380 kg ha−1 yr−1 in T4, whereas in ratoon cane emissions exceeded 3000 CO2-eq. ha−1 yr−1. The minimum emission values were similar in plant and ratoon cane, with 312 CO2-eq in T1 and 382 CO2-eq in T0,V−, respectively (Tables 3 and 4).

In terms of emission factor for N2O, our calculations indicate that the highest value among all experiments and treatments was observed for ratoon cane (Piracicaba) treated with vinasse and the largest amount of crop trash on the soil surface (21 Mg ha−1) (Table 3). The emission factor for mineral fertilizer (ammonium nitrate) applied to ratoon cane without vinasse ranged from 0.68 to 2.03% of the fertilizer applied, and the maximum value was associated with the highest volume of crop trash on the soil surface. When vinasse was applied, however, the emission factor varied from 0.59% to 3.03%, and the highest value was linked to the largest amount of crop trash. In addition, although emission factors were relatively low for all treatments below 21 Mg ha−1 of crop trash in the absence of vinasse (V−), when vinasse was used, factors increased with the amount of trash on the soil.

In plant cane, plots treated with vinasse plus mineral fertilizer had the highest emission factor (2.65% of the N applied as urea fertilizer). Without vinasse the value was 1.1%. Filter cake, which added 62 kg N ha−1, had an emission factor similar to that of urea (Table 3).

Discussion

Agriculture is recognized as a source of considerable GHG emissions on a CO2-equivalent basis (Burney et al., 2010), especially because of its contribution to gases with relatively high global warming potential (GWP) such as N2O and CH4 (Herzog, 2009). Therefore, when the use of sugarcane ethanol became an important alternative for reducing CO2 emissions associated with energy production in the world, it was clear that better information on GHG emissions related to sugarcane agriculture was needed. Sugarcane agriculture includes a series of management and fertilizer practices, and our understanding on how they might affect emissions was lacking, especially for Brazil, the largest producer of sugarcane ethanol in the world. Here, we show, using in situ measurements of emissions from the two main stages of the sugarcane crop (plant cane and ratoon cane stages), that GHG emissions in Brazilian sugarcane can vary significantly with the fertilization method used and also as crop residue (trash) accumulates on the soil surface in sugarcane plantations due to harvesting mechanization.

We determined that in plant cane, emissions of N2O and CO2 increased significantly when organic fertilizers were applied in combination with mineral fertilizer, and that most of the increases occurred shortly after fertilizer application, peaking after about 30 days. Similar patterns were observed in sugarcane fields in Australia (Allen et al., 2010; Denmead et al., 2010) and in corn fields cultivated for biofuel production (Gagnon et al., 2011).

We also observed that emissions of N2O, CO2, and even CH4 increased in ratoon cane as the amount of crop residue on the soil surface increased. The increases were especially significant when vinasse was used, except that changes in CH4 emissions were not significant.

We expected that N2O emissions would increase with the use of vinasse and other organic fertilizers in addition to mineral N fertilizer because more N was added to the soil with the organic material relatively rich in N. However, we did not anticipate the disproportionately high increases that occurred with the use of vinasse in both plant and ratoon cane. When vinasse was applied to plant cane in combination with urea in treatment T3, the amount of N added to the plots increased by only 45% in relation to the N added with urea alone, but the emission factor more than doubled whereas the emission of GHG in CO2 eq increased by about 340% (Table 4). In contrast, when filtercake was applied in combination with urea in treatment T2, the amount of N added increased by 103% but the emission factor remained the same and resulted in an increase in emissions in CO2 eq proportional to the increase in the amount of N used (Table 4). In ratoon cane, the amount of N added with vinasse increased by only 18% but the emission factors increased 1.5–2.5 times in the different treatments with crop trash. Therefore, emissions in CO2 eq increased by up to seven times, especially when 14 Mg of crop residue/ha or more covered the soil surface (Table 3).

Because vinasse is commonly applied throughout the entire surface area of the sugarcane field, the relatively high cumulative emission values that we observed when vinasse was applied were caused, in large part, by enhanced emissions in the mid rows in addition to crop rows. Emissions of CO2 in mid rows were particularly high, indicating that the use of vinasse probably increased soil respiration. Organic fertilizers in liquid form have high water content and high concentrations of dissolved organic carbon (Tejada & Gonzalez, 2005), which not only can fuel nitrification and denitrification processes, and increase N2O production, but also enhance respiration and CO2 emission.

In a study conducted in New Zealand with dairy farm effluent, Barton & Schipper (2001) observed that emissions of N2O and CO2 increased more with the application of mineral fertilizer plus dairy farm effluent than with the application of mineral fertilizer alone, or of mineral fertilizer plus water. They attributed the increase to higher respiration rates followed by depletion of O2 in the soil due to a higher soil water content and lower aeration.

In our experiments, it is possible that the application of vinasse (in liquid form) lowered soil aeration and increased the availability of labile dissolved organic C (DOC) to microorganisms, causing microsite anaerobiosis within soil aggregates due a higher demand of O2. However, despite the fact that large volumes of vinasse are commonly applied in sugarcane fields in the study region, it is unlikely that the practice would lead to anoxic conditions because the soils cultivated with sugarcane are characteristically well drained. On the other hand, if vinasse is applied after an extended period of rainfall, as it was the case in the experiment at the Piracicaba site, soil water saturation and consequent anoxia could occur.

Emissions of N2O in our experiments could also have been favored by the acidic soils that characterize the study region. According to Sánchez-Martín et al. (2008), even when an ample supply of labile C is available, acidic soils limit the completion of the denitrification reaction toward the production of N2, resulting in the accumulation of N2O.

Despite the increases in emissions associated with vinasse, the maximum rates observed for N2O in our study were still considerably lower than those reported by Crutzen et al.(2008) and than those reported for sugarcane fields in Australia where only mineral fertilizer was used, but at higher rates (Denmead et al., 2010). More importantly, any increase in CO2 emissions associated with the use of vinasse might be inconsequential as CO2 is part of the closed sugarcane cropping cycle, meaning that higher emissions do not necessarily result in a net increase in GHG emissions and higher global warming potential.

With regard to crop trash accumulation on the soil surface, it is possible that the increases in emissions of N2O, CO2 that we observed will be canceled out with the end of pre-harvesting burning practices in Brazil, which typically emits large amounts of GHG to the atmosphere (Weier, 1999). Usually more than 70% of the dry mass and much of the N content of leaf tops are lost as GHG gases during harvest pre-burning in sugarcane fields (Mitchell et al., 2000). In addition, N2O, CO2, and CH4 is emitted from the burning of crop trash on the soil surface. Such emissions will no longer exist with the mechanization of harvest in Brazil (Macedo et al., 2008). Also, much of the C in crop residue will be re-mineralized and eventually incorporated into the soil (Galdos et al., 2009).

In short, increases in CO2-eq emissions were associated with the use of vinasse, both in plant cane and in ratoon cane. In plant cane, which includes tillage before planting, vinasse was associated with an increase in N2O emissions in the order of about 1060 kg CO2-eq ha−1 yr−1, and with an increase in CO2 emissions in the order of about 965 kg CO2-eq ha−1 yr−1. Vinasse applied in plant cane contained approximately 850 kg C ha−1, therefore, it is likely that the CO2 emitted resulted from the C content in vinasse and, consequently, it is part of the closed sugarcane cropping cycle. Nevertheless, background emissions in plant cane which were measured in the control treatment (without any type of fertilizer) peaked in the first weeks of the experiments, albeit at a lower rate, indicating that soil tillage before planting (Figueiredo et al., 2010) may have contributed to the increase in emissions as well. However, without a full life cycle analysis, the impact of vinasse application on the energy balance of Brazilian ethanol is unknown (Ometto et al., 2009; D'Agosto et al., 2009; Macedo et al., 2008; Luo et al., 2009; Ramírez, 2011).

Emission Factors

An emission factor is defined by the United Nations Framework Convention on Climate Change as the average emission rate of a given GHG for a given source, relative to units of activity (e.g., fertilizer application). As such, these factors can facilitate comparisons of emissions from various sources and activities.

According to our calculations, emission factors in plant cane and in ratoon cane without vinasse and trash cover up to 14 Mg ha−1 – a usual trash load – were not substantially different from the 1% emission factor estimated for sugarcane ethanol by the Intergovernmental Panel on Climate Change (IPCC) (2007) (Tables 3 and 4). On the other hand, emission factors associated with the use of vinasse, both in plant and ratoon cane, were more than twice the IPCC values.

Relatively high emission factors have been reported in other recent studies with sugarcane as well (Allen et al., 2010; Denmead et al., 2010; Lisboa et al., 2011). However, the controlling factors were likely different. For instance, Denmead et al. (2010) reported an emission factor of 21% for the application of mineral fertilizer in sugarcane growing in Australia, whereas an emission factor of 2.8% was reported for the use of 150 kg ha−1 N banded in a field blanketed with trash. In the first case, the high value reported was associated with a poorly drained soil, which is not typical of sugarcane growing soils in Brazil. In the second case, the high value was associated with a region that has a high rainfall regime (>2100 mm in 292 days), also not typical of the sugarcane growing region in Brazil.

Emission factors reported by Allen et al. (2010) were higher than 1% for fields blanketed with crop trash only when the N fertilizer application rate increased from 100 to 200 kg ha−1. When N fertilizer application was split in two installments, the emission factor was 2.95%. When all of the N fertilizer was added at once, the emission factor increased to 6.7%.

Lisboa et al. (2011) reported an average emission factor of 3.87% for fertilizer application in sugarcane, but this value was based on scientific literature that included several studies with short-period evaluations of N2O emissions.

In our study, emission factors were well above 1% with the use of vinasse only when it was added with other fertilizers to plant cane, or when 14 Mg of crop residue ha−1 or more covered the soils. Therefore, the link between the use of vinasse and higher emission factors is evident. On the other hand, we believe that if vinasse is applied a few weeks before the application of mineral fertilizer, or in a more concentrated form in the crop row (and not in mid rows), N2O emissions could be reduced by avoiding soil conditions that promote denitrification. Therefore, rather than the use of vinasse itself, it is possible that the increase in GHG emissions that we observed in our study was associated with the form and timing of application, which probably could be resolved with better management practices.

Brazil generates an immense volume of vinasse since for each liter of ethanol produced about 10 L of vinasse is generated (Mutton et al., 2010). Therefore, the recycling of vinasse in sugarcane fields has tremendous benefits not only for the environment but for agriculture given that vinasse is also rich in nutrients (Mutton et al., 2010). However, better management practices need to be developed to minimize the negative effects of vinasse application on GHG emissions.

As for sugarcane trash, emission factors were high only in treatments with a rate of accumulation of 14 Mg ha−1 or more. However, such high loads of trash are hardly seen in sugarcane fields in Brazil, where the average is about 7.5 Mg ha−1 (Hassuani et al., 2005). More importantly, large amounts of trash are unlikely to accumulate in sugarcane fields in Brazil because of its growing importance as feedstock for second generation biofuel, and also as it is increasingly used to fuel sugarcane mills (Goldemberg, 2008). Consequently, it is likely that future discussions about accumulation of trash in sugarcane fields in Brazil will focus on limiting removal to preserve soil quality rather than on controlling excess trash accumulation to prevent GHG emissions.

Presently, a very limited number of studies provide any recommendation on the amount of crop trash that should be maintained in agricultural fields in Brazil. However, these studies either focus on prevention of weed proliferation and soil physicochemical protection and restoration (Hassuani et al., 2005) or on minimum crop residue accumulation for soil quality purposes (Gollany et al., 2011). In any case, these studies are crop and site specific and, therefore, are not suitable for sugarcane. Clearly, future studies are needed to provide information for managing trash accumulation in sugarcane fields in Brazil for soil conservation and restoration.

A more pressing issue regarding excess trash accumulation in sugarcane fields in Brazil that needs to be investigated is related to N fertilizer application. According to Vitti et al. (2008), crop trash accumulation may result in higher rates of N fertilizer application as N is retained in decomposing organic material. On the other hand, as harvesting practices change from manual with pre-harvest burning to mechanical without pre-burning and with no- or minimum-tillage systems, soil C and N stocks may be maintained or even increase (Figueiredo & La Scala, 2011), preventing soil impoverishment. However, several models have predicted that a new equilibrium in the soil system will be reached in 20 or more years, and that a reduction in N fertilization rates with trash accumulation on the soil will be possible only in the long term (Thorburn et al., 1999; Robertson & Thorburn, 2007; Van Antwerpen et al., 2008). Regardless of the time frame for a new equilibrium in the soil system, more studies are needed to improve our understanding of trash decomposition rates and C/N ratios in sugarcane agriculture systems, and provide information for better management practices (Trivelin et al., 1995; Cantarella, 1998; Aulakh et al., 2001; Oliveira et al., 2002).

In conclusion, our study provides much needed in situ data on emissions of GHG from sugarcane crop soils in Brazil and a comprehensive characterization of emissions associated with different fertilizer application practices commonly used in the country, and with changes in harvest practices. Furthermore, because the study includes data collected during the entire growing season of the two main stages of the crop (plant and ratoon cane) and data collected particularly intensively in the first 2 weeks after tillage and application of fertilizer, it provides important information about the temporal variation of emissions. However, because many of the conditions that affected GHG emissions in our study are likely inherent to the study region or period (e.g., soil characteristics, climatic conditions, and concentrations of nutrients in fertilizers), the results are not necessarily applicable to other regions where sugarcane is now produced in Brazil or elsewhere. Sugarcane production systems are diverse and understanding how emissions change across regions and under other management scenarios should be a research priority for future studies about GHG emissions from sugarcane systems and ethanol production. Nevertheless, the results of this study should help not only validate available estimates of emissions from Brazilian sugarcane, but also better assess the net C savings capacity of sugarcane ethanol in Brazil.

Acknowledgements

We wish to thank the Sugarcane Technology Center (Centro de Tecnologia Canavieira - CTC) in Piracicaba, SP, and the São Paulo Agro-business Technology Agency (Agência Paulista de Tecnologia dos Agronegócios - APTA) in Jaú, SP, for allowing us to conduct our experiments in their research facilities. We also thank Paulo Queiroz, Gilson Camargo da Silva, Leonardo Martinelli, Luiz Felipe Martins, Eduardo Almeida Anunciação, Fabiana Fracassi, and Aline Daiane Briques for valuable assistance in the field and laboratory. Support for this research was provided to Janaina Braga do Carmo by the São Paulo Research Foundation (FAPESP) through the Young Researcher Program, and to Janaina Braga do Carmo, Solange Filoso, and Luiz Antonio Martinelli through the BIOEN/FAPESP Program (Process Number 08/55989-9). Small Funds were also provided by the Research and Projects Financing Program (FINEP) (Agreement 01.09.0309.00, Process Number 1637/08) as part of the project conducted by Sugarcane Technology Center (CTC). We also thank BIOEN/FAPESP Program (Process Number 08/56147-1) for technical and scientific support.

Ancillary