Losses of NO and N2O emissions from Venezuelan and other worldwide tropical N-fertilized soils


  • Sorena Marquina,

    1. Laboratorio de Química Atmosférica, Centro de Química, Instituto Venezolano de Investigaciones Científicas, Caracas, Venezuela
    Search for more papers by this author
  • Loreto Donoso,

    1. Laboratorio de Química Atmosférica, Centro de Química, Instituto Venezolano de Investigaciones Científicas, Caracas, Venezuela
    Search for more papers by this author
  • Tibisay Pérez,

    Corresponding author
    • Laboratorio de Química Atmosférica, Centro de Química, Instituto Venezolano de Investigaciones Científicas, Caracas, Venezuela
    Search for more papers by this author
  • Jenie Gil,

    1. Laboratorio de Química Atmosférica, Centro de Química, Instituto Venezolano de Investigaciones Científicas, Caracas, Venezuela
    Search for more papers by this author
  • Eugenio Sanhueza

    1. Laboratorio de Química Atmosférica, Centro de Química, Instituto Venezolano de Investigaciones Científicas, Caracas, Venezuela
    Search for more papers by this author

Corresponding author: T. Pérez, Laboratorio de Química Atmosférica, Centro de Química, Instituto Venezolano de Investigaciones Científicas (IVIC), Aptdo. 20632, Caracas 1020-A, Venezuela. (tperez@ivic.gob.ve)


[1] N fertilization significantly increases N2O and NO soil fluxes to the atmosphere. In spite of the expansion of agricultural activities in tropical managed soils from the developing world, there is little information about the loss of applied nitrogen (LAN) as NO and N2O from these areas. In this work, we determined LAN-N2O and LAN-NO from different crops during the growing season at a sandy soil experimental field and two active farms with loamy and clay soils, respectively. Tillage (T) and no-tillage (NT) farming were separately evaluated. All of the evaluated areas were located in the Venezuela savanna region. A large range of LAN-N2O (0.30–6.1%) and LAN-NO (0.26–2.1%) were recorded, with overall average values of 1.9% and 0.9%, respectively. LAN values were mainly affected by soil texture and rainfall pattern, which affected soil moisture and water-filled pore space. Also, soil management (T and NT) and the chemical composition of the N fertilizer played important roles. The overall average of LAN-N2O is about two times higher than the IPCC default value of 1%; therefore, our results suggest that a higher factor should be considered for cropping systems in tropical savanna regions.

1 Introduction

[2] Soils are the largest source of nitrous oxide (N2O) and a significant source of nitric oxide (NO) [Denman et al., 2007]. N2O from natural soils (6.6 Tg N yr−1) and that from agricultural soils (2.8 Tg N yr−1) together account for 53% of the total sources [Intergovernmental Panel on Climate Change, 2007], with tropical soils responsible for roughly two thirds of the natural total soil sources. NO global soil emission estimates are also significant, comprising between 1.6 and 8 Tg N yr−1, or 17.1% of the total NO sources [Denman et al., 2007], and like N2O, global potential emission of NO magnitude rank in decreasing order from agricultural fields > savannas > forests > other natural systems, with tropical ecosystems being the largest potential emitters of NO [Yan et al., 2005]. These gases are of global relevance: nitrous oxide is an important greenhouse gas representing 6% of the total greenhouse gas radiative forcing [Denman et al., 2007; World Meteorological Organization (WMO), 2011], and it is also the main source of stratospheric NO, which is one of the most important catalysts for stratospheric ozone depletion [Crutzen, 1979]. Nitric oxide participates in the oxidation of tropospheric hydrocarbons that lead to tropospheric ozone formation, which is a greenhouse gas and also affects the oxidative capacity of the atmosphere [Crutzen, 1979].

[3] Globally, the atmospheric N2O concentration average during 2010 was 323.2 ppb, up to 20% above the preindustrial level [WMO, 2011]. The increase of N2O in the troposphere of ~ 0.25% yr−1 [e.g., Denman et al., 2007; WMO, 2011] has been attributed to large increases in the application of inorganic fertilizer during the last century [Kroeze et al., 1999]. Despite the global importance of N2O, the number of studies of direct N2O soil fluxes from tropical areas is limited. This has led to larger uncertainties in the global natural and agricultural N2O budget [Bouwman et al., 1995; Stehfest and Bouwman, 2006] and NO sources estimates [Stehfest and Bouwman, 2006; Yan et al., 2005]. Recent inverse modeling approaches [Hirsh et al., 2006; Huang et al., 2008] suggested that N2O land sources in the tropics (Equator to 30°N) are larger by a factor of 20% to 64% than prior estimates.

[4] For countries with insufficient ground information of N2O emissions, the current Intergovernmental Panel on Climate Change (IPCC) guidelines for national greenhouse gases inventory recommend the “Tier 1” method for calculating direct N2O emissions for N addition from mineral fertilizers, organic amendments, and crop residues [Klein et al., 2006]. This methodology sets a default global emission factor for N2O of 1%, with a large uncertainty, between 0.3% and 3%. Due to the scarce studies in the tropical region, most of the information used to establish the default value is from the temperate region. Although the IPCC guidelines have not proposed a global emission factor value for NO, a study review of 99 measurements from agricultural soils found that 0.7% of the N applied is lost as NO [Bouwman et al., 2002; Bouwman and Boumans, 2002]. Considering that N-fertilized soils in the tropics is rapidly growing [Bouwman et al., 2011; Sutton and Bleeker, 2013; van Vuuren et al., 2011], additional studies to determine N losses as N2O and NO from this region are needed.

[5] In this work, we measured gas fluxes of NO and N2O from two active farms with clay and loamy soils (2005–2006 and 2007, respectively). We also measured N2O fluxes at a sandy soil experimental field from 1991 to 1994. From these sandy soil experimental fields, we used previously reported NO fluxes [Pérez et al., 2007] to compare with the data measured in this work to evaluate the effect of soil moisture, soil texture, and land management (tillage, fertilization, and crop rotation) on N fertilizer losses as NO and N2O. Also, we made a compilation of N2O and NO loss data from our data set and other tropical agroecosystems in order to estimate possible LAN-N2O and LAN-NO values for this latitude.

2 Field Measurements

2.1 Sampling Sites

[6] Study was conducted in the Venezuelan savanna region. Two well-defined periods are characteristic of the region, a dry season between December and April and a rainy season between May and November, with annual rainfall ranging between 700 and 2238 mm. Air temperature ranged between 23 and 33°C (minimum and maximum mean annual values, respectively) [Instituto Nacional de Meteorología e Hidrología, 2009].

[7] Measurements were made at three different sites: a set of experimental controlled plots with different crop types located in the Estación Biológica de Los Llanos, Calabozo, Guárico State (Guárico with sandy soils; GS) (8°53′N, 67°19′W) during 1991, 1992, and 1994; and corn fields under local farmers' land management—one located at Fundo Tierra Nueva farm in Guárico State (Guárico with clayey soils; GC) (9°23′33″N, 66°38′30″W) during 2005 and 2006 and the other one at Nardini farm located in Portuguesa State (Portuguesa with loamy soils; PL) (9°6′45″N, 69°2′12″W) during 2007. Soil properties are listed in Table 1.

Table 1. Soil Properties (0 to 5 cm of Depth) of Study Sites
(1991–1994)b, c(2005–2006)d(2007)e
  1. a

    Mean ± standard deviation (n).

  2. b

    T-91, T-92, and T-94.

  3. c

    Data from Cárdenas et al. [1993] and Pérez et al. [2007].

  4. d

    NT-05, T-06, and NT-06.

  5. e

    T-07 and NT-07.

Sand (%)712330
Clay (%)134814
Silt (%)162956
Textural class (common name)Sandy loam (sandy)Clay (clayey)Silt loam (loamy)
Bulk densitya (g cm−3)1.4 ± 0.1 (19)1.4 ± 0.2 (9)1.5 ± 0.2 (6)
pHa5 ± 1 (28)6 ± 1 (472)8 ± 1 (268)
Tsoila (°C)29 ± 8 (570)28 ± 3 (1176)29 ± 4 (400)

[8] At the experimental plots site (GS), 1 ha of savanna land was divided into four (250 m2 each), and Zea mays L. (corn), Brachiaria decumbens grass (pasture), and Andropogon sorghum (sorghum) were planted, while the last one was left as a control area. Fertilizer application rates for each area were done according to local farm practices (Table 2). On each site, two metal frames were inserted at 10 cm of depth at least 3 days prior to the measurements. Gas fluxes were measured during the year of land conversion (1991), as well as 2 (1992) and 4 (1994) years after that. Other sampling details are described elsewhere [Pérez et al., 2007; Sanhueza et al., 1994].

Table 2. Loss of the Applied Nitrogen (LAN) as NO and N2O From Venezuelan Sampling Sitesa
 Crop (Management-Year) and Fertilizer (kg N ha−1)bSampling Period (Rainfall in mm)WFPS (%)LAN-N2O (%)LAN-N2O (%) Corrected by Unfertilized SoilcLAN-NO (%)LAN-NO (%) Corrected by Unfertilized Soilc
  1. a

    Data are average ± standard deviation, and numbers of measurements are reported in parentheses. N/A, not available.

  2. b

    Land management: T, tillage; NT, no-tillage. Fertilizer: A, ammonium; AN, ammonium nitrate; U, urea.

  3. c

    The native unfertilized fields did not undergo any soil management (e.g., plowing, seeding).

  4. d

    Data and calculations are from Pérez et al. 2007.

Guárico-sandyCorn (T-91) A(56) + U(70)29 May–18 Jul (458)22 ± 16 (296)d0.30 ± 0.040.28 ± 0.051.7 ± 1.2d1.5 ± 1.0d
Pasture (T-91) A(173.6)29 May–18 Jul (458)20 ± 16 (276)d0.11 ± 0.010.09 ± 0.011.2 ± 0.9d1.0 ± 0.7d
Corn (T-92) A(58) + U(70)31 May–13 Aug (858)36 ± 12 (149)d2.6 ± 0.12.5 ± 0.10.79 ± 1.83d0.8 ± 1.8d
Sorghum (T-92) A(60) + U(70)31 May–13 Aug (858)42 ± 13 (142)d0.91 ± 0.040.84 ± 0.040.64 ± 0.60d0.6 ± 0.6d
Corn (T-94) A(54)19 Jul–23 Sep (766)33 ± 16 (103)d0.64 ± 0.020.55 ± 0.020.99 ± 1.30d0.9 ± 1.3d
Sorghum (T-94) A(60)19 Jul–23 Sep (766)34 ± 16 (106)d0.46 ± 0.010.39 ± 0.010.64 ± 0.67d0.6 ± 0.6d
Guárico-clayeyCorn (NT-05) A(54) + AN(46)31 May–8 Jul (N/A)97 ± 6 (110)3.8 ± 0.2N/A0.6 ± 0.1N/A
Corn (T-06) A (65) + AN(58)31 May–29 Jul (279)39 ± 2 (244)3.4 ± 0.42.8 ± 0.42.1 ± 0.21.48 ± 0.18
Corn (NT-06) A(56) + AN(58)31 May–29 Jul (279)58 ± 3 (232)6.1 ± 0.75.5 ± 0.70.72 ± 0.050.05 ± 0.01
Portuguesa-loamyCorn (T-07) A(30) + U(46) + U(46)14 May–11 Jul (371)36 ± 5 (7278)0.8 ± 0.1N/A0.34 ± 0.04N/A
Corn (NT-07) A(30) + U(46) + U(46)14 May–11 Jul (371)56 ± 6 (7268)2.1 ± 0.2N/A0.26 ± 0.02N/A

[9] At the cornfield sites (GC and PL sites), measurements were made in adjacent fields (~ 1 km2 each) under tillage (T) and no-tillage (NT) management. These cornfields had the same land management for at least the previous 7 years, and they were planted and fertilized with the typical regional farmers' land management. Also, during 2006, measurements were made in an adjacent unfertilized field (fallow area was not planted or fertilized). In each cornfield sampling area (approximately 150 m2), eight PVC rings (10 cm height × 26 cm diameter) were placed randomly and inserted 5 cm on the ground after fertilization and seeding, and they were left throughout the sampling. Fertilization rates were similar in the five sampling areas (100 to 120 kg N ha−1), distributed in two applications for Guárico-clayey and three applications for Portuguesa-loamy (Table 2).

2.2 Flux Measurements

2.2.1 N2O Soil Emissions

[10] At the GS site, a stainless steel-glass chamber was used, similar to the one described by Conrad et al. [1983]. This chamber was inserted into the metal frame, which was insulated with distilled water. N2O concentrations in the chamber were measured directly using a Shimadzu gas chromatograph (model GC-8AIE) equipped with an electron capture detector and an automatic injection system. Gas samples were supplied to the injection valves by a pump and recirculated back to the chamber at a flow rate of 250 mL min−1. During the measurement of each chamber (60 min), four gas samples every 10 min and two calibration gases (at the beginning and at the end) were injected. N2O standard calibration gas (303.4 ± 0.9 ppbv) was prepared and calibrated by Max Planck Institute. N2O fluxes were measured four times per day during the first 3 weeks after fertilization. After that period, measurements were made every other week until the end of the growing season. More details about the flux measurement procedure are described elsewhere [Sanhueza et al., 1990].

[11] At the GC and PL sites, fluxes were measured daily during the sampling period using a closed chamber technique (PVC base and chamber) [Vitousek et al., 1989]. We collected 20 mL of air samples at 1, 10, 20, and 30 min after chamber closure by means of silicone or glass syringes provided with stopcocks. N2O was analyzed using a Shimadzu GC-8A gas chromatograph equipped with an electron capture detector at 375°C, a main separating column packed with Porapak Q (2 m), and N2 UHP as carrier gas. We used standard calibration gas with two N2O concentrations (320 ppbv and 800 ppbv) both manufactured by Scott-Marrin, Inc.

[12] N2O mixing ratio values as a function of time in closed chambers were used to calculate the slope with the best lineal regression. In cases with nonlineal behaviors, these were not used for calculations.

[13] The N2O fluxes determined by means of our chamber measurement methodologies have a high level of confidence according to the parameters suggested by Rochette and Eriksen-Hamel [2008]. The score of these factors (chamber design, seal and soil surface, air sample handling and storage, and determination of dC/dt) for our methodologies were between 1.5 and 2.5, which are qualified as “good” and “very good” [Rochette and Eriksen-Hamel, 2008].

2.2.2 NO Soil Emissions

[14] At the cornfields (GC and PL sites), NO soil emissions were also measured daily during the entire period using the dynamic chamber technique (PVC base and chamber) [Keller and Reiners, 1994]. We analyzed the NO mixing ratios in the chambers by means of a chemiluminescence NOx Analyzer (model LMA-3D, manufactured by Scintrex, Inc.) based on the reaction between NO2 and luminol solution. This portable equipment allowed in situ measurements (response time 0.2 s). Also, we installed a data logger model DAS 1245 (Datastick Systems, Inc., Santa Clara, CA) with a handheld computer (Palm OS 3.5, Zire 72 model) from which we monitored the trace gas concentrations during measurements and data storage for subsequent data transfer. NO emissions were calculated with the NO mixing ratio values as a function of time during chamber enclosure using the best lineal regression of the data. For the experimental plots of GS site, we used data reported in Pérez et al. [2007] to calculate LAN-NO from these soils.

2.3 Soil Texture and Bulk Density

[15] Bulk density was determined by means of the core method [Blake and Hartge, 1994]. For the GS site (1991–1994), triplicates were taken between 0 and 10 cm and between 0 and 100 cm of depth for the GC and PL sites (2005–2006 and 2007, respectively). Soil texture was determined by the hydrometer method [Gee and Bauder, 1994] at the same depths as those for bulk density. These soil analyses for the GC and PL sites were made by personnel of the Instituto Nacional de Investigaciones Agrícolas (INIA) de Venezuela.

2.4 Soil Water Content

[16] For the GS and GC sites (1991–1994 and 2005–2006, respectively), we determined soil water content gravimetrically. For that, we took soil samples between 0 and 2 cm of depth near each base or frame. We weighed duplicate 10 g of those and then oven dried them at 105°C for 48 h. Subsequently, we stored the samples in desiccators until their weight became constant. For the PL site (2007), we measured soil water content volumetrically using soil moisture sensors ECH2O Decagon Devices, model EC-5. These sensors were located at 2 cm of depth inside each PVC base (eight for each plot), and they measured the variation on soil water content every 10 min during the sampling time [Saxton et al., 1986].

[17] Water filled pore space was calculated according to Saxton et al. [1986] from the upper 0–2 cm layer using equation (1), where θg and θv are the average values of the water content of each plot during the sampling period gravimetrically and volumetrically calculated, respectively, BD is the bulk density mean value, and PD is the particle density assumed as 2.61 g/dsg (dry soil gram) [Davidson and Schime, 1995].

display math(1)

2.5 LAN Calculations

[18] Loss of applied nitrogen (LAN) as NO and N2O is a unit to calculate the direct nitrogen fraction from fertilizer application that is emitted as these nitrogen gases during a specific period of time (equation (2). This value is expressed as a percentage of the N applied.

display math(2)

[19] Emission from fertilized soil in equation (2) corresponds to the average of all gas fluxes measured during the sampling period in each site, expressed as kg N ha−1 (see Table 2 for sampling period). For example, the emission value used in equation (2) for the PL site is the average of the gas fluxes measured in eight chambers daily during 58 days.

[20] When we have measurements of NO and N2O emissions from a fallow area or a native unfertilized field without any soil management (e.g., plowing or seeding), we also calculated a LAN relative to unfertilized soil (LAN* in equation (3)). This LAN corrected by unfertilized soil is similar to an emission factor value (EF) because it represents the percentage increases of superficial emissions from a cultivated and fertilized soil in comparison to the emission from an unfertilized soil. However, in our LAN* calculation, the unfertilized control plot has not all other conditions equal to those of the fertilized plot (e.g., plowing), which is proposed for EF calculations [Bouwman and Boumans, 2002].

display math(3)

[21] For equation (3), both emissions from fertilized and unfertilized soil were calculated as the average of all gas fluxes measured during the sampling period.

2.6 Statistics

[22] We used a Kolmogorov-Smirnov test to verify that the data had normal distribution. We confirmed correlations using SPSS Statistic 17 by means of Pearson and Spearman coefficients for data with normal or no distribution, respectively. We calculated a mean LAN value for each site (GS, GC, and PL) and then made a comparison of two measured numbers [Taylor, 1982]. We used one-way ANOVA to compare LAN values for different textures from tropical data.

3 Results and Discussion

[23] Values of the loss of applied nitrogen (LAN) as NO and N2O from direct soil emission obtained at the Guárico-sandy (GS), Guárico-clayey (GC), and Portuguesa-loamy (PL) N-fertilized fields are shown in Table 2. LANs calculated directly in relation to the amount of N fertilizer applied and those corrected by unfertilized soils are given. Rainfall during sampling periods and water-filled pore space (WFPS) mean values are also given.

[24] In general, for loamy and clayey soils, LAN-N2O values were larger than LAN-NO, while for sandy soils, we obtained the opposite (except for 1992; Table 2). This result is mainly affected by soil texture and soil moisture relationship, which promotes anaerobic conditions that increase N2O production and emission. Larger N2O emissions were observed under controlled experiments at higher WFPS values (45 to 90%) due to enhanced denitrification [Hernandez-Ramirez et al., 2009]. The texture effect on soil moisture is evident from the WFPS values obtained in each site despite the difference in precipitation during the sampling period, i.e., more precipitation on the GS site than on the GC and PL sites (Table 2). However, these LAN values were also affected by land management.

3.1 Effect of Soil Texture and Moisture

[25] LAN-N2O values found in this study decrease in the following order: Guárico-clayey > Portuguesa-loamy ≈ Guárico-sandy (Table 2). We obtained a Spearman correlation coefficient of 0.661 (p = 0.014, N = 13) between LAN-N2O values and soil clay content. Additionally, we obtained a positive lineal correlation between the Log LAN-N2O versus water-filled pore space (r = 0.707, p = 0.015, N = 11) (Figure 1, grey values and linear fit). These results suggest that LAN-N2O values are affected by soil texture and soil moisture. For instance, our site Guárico-clayey (GC), which had the largest LAN-N2O values, has Vertisol soil. This soil type has expansible clays that shrink and crack when dry and increase in volume when wet [Sabburg et al., 1997]. This expansible property of Vertisol clay soils induces bulk density changes with water content, which at the same time influence soil oxygen diffusion. Therefore, Vertisol soils saturate with lower water content, promoting more anaerobic sites, which should increase N2O production and emission in comparison to soils with less clay content [Silver et al., 2000]. Our results are consistent with other tropical studies. For example, for Puerto Rico grasslands, the LAN-N2O from clayey-Vertisol was about three times higher than the values recorded from clayey-Ultisol and clayey-Oxisol soils [Mosier and Delgado, 1997]; and also, significantly more N2O lost from a fine-texture soil than from a coarse-texture soil was observed in N-fertilized sugarcane fields in Australia [Weier et al., 1996].

Figure 1.

Relationship between the logarithm of LAN-N2O and soil water-field pore space. Letters and lines represent the individual data points and linear regression: grey (this work data set) and black (data from Table 4; B: Brazil, CR: Costa Rica, Hw: Hawaii, J: Java, Mu: Maui, Mg: Madagascar, My: Malaysia, PR: Puerto Rico, and V: Venezuela).

[26] The lowest LAN-N2O recorded in this study was produced during the 1991 growing season at the Guárico-sandy fields (corn and pasture T-91; Table 2). This period was associated with the occurrence of an El Niño episode, which started in the April-May-June, 3 month running mean of the NOAA Oceanic Niño Index (http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.shtml). El Niño produces a dry climate in the Venezuelan savanna region, and the scarce rainfall that occurred during the 1991 growing season produced lower soil moisture and, in consequence, lower LAN-N2O observed in this study (Table 2). This result is in agreement with a recent study of rainfall manipulation in a Puerto Rican tropical forest, where the authors concluded that a drier climate could cause substantial reductions in global emissions of nitrous oxide from soil [Cleveland and Sullivan, 2012].

[27] It is important to point out that at Portuguesa-loamy fields (T-07 and NT-07 in Table 2), lower soil moisture periods were produced immediately after fertilization, which would in part explain the relatively much lower LAN-N2O values, in comparison with the ones observed in Guárico-clayey (NT-05, T-06, and NT-06). It is likely that timing of fertilization and rainfall is an important factor that controls LANs. Liu et al. [2011] found that the time interval between top dressing and irrigation or rainfall induces different patterns of N2O emissions.

[28] In the case of LAN-NO values, we did not find statistical differences between the GS, GC, and PL sites probably due to a high variability (Table 2). However, our measurements show that after relative dry periods and rewetting due to soft rainfalls, significant emission pulses of NO were observed, which is in agreement with findings in natural tropical savanna soils [Davidson, 1992; Harris et al., 1996; Johansson et al., 1988; Johansson and Sanhueza, 1988]. At the GC fields, the NO pulses produced in the fertilized soils (T-06 and NT-06) were four times larger than the pulse from the unfertilized control field, while only small changes in the N2O fluxes were observed during these periods. Higher NO pulses could be attributable to the soft rewetting of the uppermost part of the soil that promotes optimal soil moisture conditions to emit NO. We also did not observe flux pulses of NO and N2O when strong rainfall events occurred, even long after fertilization, again confirming that both nutrient availability after fertilization and optimum soil moisture are not achieved immediately for microorganisms to produce these trace gases emissions [Liu et al., 2011].

3.2 Land Management and Fertilization

[29] The measurements made at Guárico-clayey (GC) during 2006 and Portuguesa-loamy (PL) in 2007 allow us to make a direct comparison between tillage and no-tillage soil management. In each campaign, T and NT fields received the same type and similar amount of N fertilizer (Table 2), and also, they were under the same climatological conditions (e.g., solar irradiation and rain pattern).

[30] At both sites, higher LAN-NO from tilled soil were observed (T-06 and T-07 in Table 2). This is likely attributable to a larger soil-atmosphere gas exchange in plowed soils and also to a high oxygen availability to decompose organic matter and produce NO by means of aerobic microbial reactions (such as nitrification and nitrifier denitrification in the soil anaerobic microsites), which leads to larger NO production [Davidson et al., 2000; Firestone and Davidson, 1989; Wrage et al., 2001]. In comparison to our LAN-NO values, around two times higher LAN-N2O values were found in both no-tillage management fields (NT-06 and NT-07 in Table 2). The larger LAN-N2O values from no-tillage than tillage management could be attributable to the enhancement of anaerobic conditions in the no-tillage soil, which favors N2O production by means of denitrification or nitrate ammonification microbial reactions [Baggs, 2011; Firestone and Davidson, 1989]. Similar results, with lower NO and higher N2O emissions in NT soils, were reported from cornfields in Colorado [Liu et al., 2006]. This higher LAN-N2O could offset the store of C in no-tillage soils. However, other studies did not find N2O soil emission differences between T and NT management [Chapius-Lardy et al., 2009; Jantalia et al., 2008]. Other studies found differences in N2O emissions from no-tillage or reduced-tillage according to soil aeration, time of management implantation, N fertilizer placement, and climate conditions [Rochette, 2008; van Kessel et al., 2013]. Rochette [2008] found an increase in N2O emissions in poorly aerated soil, but no differences were found in soils with good and medium aeration. Meanwhile, van Kessel et al. [2013] found a significant increase of 57% under dry climatic conditions when no-till/reduced-till was implemented 10 years or less but found a significant decrease (27%) when no-till/reduced-till was implemented 10 years or more. Additionally, for humid climates, the authors recommend N fertilizer placement at more than 5 cm of depth in order to decrease N2O emissions when implementing no-till or reduced-till.

[31] An additional aspect for the analysis is that, historically, Portuguesa-loamy crop soils have been more intensively managed (with almost four rotations per year) in comparison to Guárico-clayey crop soils (one to two rotations per year). According to Stevenson and Cole [1999], intensive management of soils may reduce the microbial activity. This assumption was confirmed by measurements of dehydrogenase enzyme activity taken during the sampling period (F. Herrera, 2007; personal communication). At the Guárico-clayey soils, this enzyme activity was 170 ± 11 µg TPF dsg−1 (microgram of triphenilformazan per dry soil gram), whereas in Portuguesa, soils were 108 ± 5 µg TPF dsg−1. This, in part, could explain the higher LAN-NO and LAN-N2O values at the GC fields, compared to those of the PL fields. However, it is important to consider that different fertilization regimes were used at these fields, especially the application of ammonium nitrate at the GC site.

[32] At the GC sites, the first fertilization was made with NH4Cl and the second one with NH4NO3. Now, in 2006, during the immediate period after first and second fertilizations (about 10 days), the soil moisture was quite similar, allowing a comparison of the effect of the two different fertilizers (Table 3). Clearly, LAN-NO and LAN-N2O from T and NT fields are higher when NH4NO3 was applied—around 10 times in the case of N2O and 2–3 times for NO. This is in agreement with fertilization experiments in tropical natural savanna soils [Donoso et al., 1993; Sanhueza et al., 1990], which found that the addition of nitrate produces a large enhancement in NO and N2O emissions whereas ammonium has a smaller effect and that the larger increase was produced with NH4NO3. Veldkamp et al. [1998] found that the fertilizer types did not show significant differences in LAN-N2O and LAN-NO from pastures in Costa Rica; however, in their fertilization experiment, they did not apply NH4NO3.

Table 3. LAN-NO and LAN-N2O From Guárico-Clayey Fields During 2006 Campaign 10 Days After First (NH4Cl) and Second Fertilization (NH4NO3) Events
GasManagementaLAN (%)b
  1. a

    NT, no-tillage; T, tillage.

  2. b

    Calculated from emissions produced during 10 days after fertilization.

NONT0.12 ± 0.010.46 ± 0.023.7
T0.38 ± 0.040.71 ± 0.081.9
N2ONT0.56 ± 0.046.47 ± 0.4111.5
T0.15 ± 0.021.40 ± 0.079.2

[33] The analysis of the result of land management (tillage and crop rotation) and fertilization indicates that both may affect LANs; special attention should be paid to the quite high values obtained from NT when soils are fertilized with NH4NO3.

3.3 LAN From Tropical N-Fertilized Agricultural Fields

[34] In the literature, LANs for NO and N2O from tropical N-fertilized soils are very scarce. The available information is reviewed in Table 4. Most of the studies were performed at biological or agricultural experimental stations.

Table 4. Loss of Applied Nitrogen (LAN) as N2O and NO From Tropical Agricultural Soils
ReferenceLocationTexture (Soil Type)CropManagementaTotal N Application Rate (kg N ha−1)Chemical N Application Rate (kg N ha−1)Organic N Application Rate (kg N ha−1)Crop Residue (kg N ha−1)Type of Chemical and Organic NdWFPS (%)LAN-N2O (%)LAN-NO (%)
  1. a

    NT, no-tillage; T, tillage; C, continuously; I, intermittent.

  2. b

    LAN values corrected by unfertilized soils. These values are equivalent to emission factors (EF) according to IPCC.

  3. c

    Farmers rotated between 20 and 45 heads of cattle daily among individual pastures.

  4. d

    A, ammonium; N, nitrate; AN, ammonium nitrate; U, urea; U+, urea with phosphorous and potassium; ChM, chicken manure; ZM, zebu manure.

Weitz et al. [2001]Costa RicaClayMaize-TaroNT737300AN700.44; −0.60b 
Weitz et al. [2001]Costa RicaLoamyMaize-TaroNT19519500AN861.08; 1.07b 
Weitz et al. [2001]Costa RicaClayPapaya-BalsaNT26026000AN770.29; 0.18b 
Weitz et al. [2001]Costa RicaLoamyPapaya-BalsaNT32532500AN902.16; 2.87b 
Khalil et al. [2002]MalaysiaLoamyMaize/GroundnutT3221800142A30–800.59 
Khalil et al. [2002]MalaysiaLoamyMaize/GroundnutT18018000A30–800.78 
Khalil et al. [2002]MalaysiaLoamyMaize/GroundnutT30090168142A, ChM30–800.99 

Chapius-Lardy et al. [2009]

MadagascarClayMaize-soybeanNT57.133 + 4.6172.5A + U, ZM60–750.46 

Chapius-Lardy et al. [2009]

MadagascarClayMaize-soybeanT54.633 + 4.6170A + U, ZM60–750.47 
Suratno et al. [1998]JavaClayRice floodedC868600U930.36 [0.25–0.52] 
Suratno et al. [1998]JavaClayRice floodedI868600U930.47 [0.32–0.60] 
Weier et al. [1996]Australia SugarcaneT16016000N 0.13 
Weier et al. [1996]Australia SugarcaneNT16016000N 15.4 

Mosier and Delgado [1997]

Puerto RicoClay (Oxisol)Grassland 757500A55–880.8b 

Mosier and Delgado [1997]

Puerto RicoClay (Vertisol)Grassland 757500A50–903.3b 

Mosier and Delgado [1997]

Puerto RicoClay (Ultisol)Grassland 757500A40–800.8b 

Matson et al. [1996]

MauiClaySugarcane 848400U450.0130.014

Matson et al. [1996]

MauiClaySugarcane 454500U570.120.027

Matson et al. [1996]

MauiClaySugarcane 353500U410.490.004

Matson et al. [1996]

MauiSilty claySugarcane 222200U400.0580.038

Matson et al. [1996]

MauiSilt loamSugarcane 202000U 0.0300.016

Matson et al. [1996]

MauiSilt loamSugarcane 393900U400.240.034

Matson et al. [1996]

MauiClaySugarcane 343400U400.340.012

Matson et al. [1996]

HawaiiSilt loamSugarcane 959500U580.412.07

Matson et al. [1996]

HawaiiSilt loamSugarcane 12412400U661.001 

Matson et al. [1996]

HawaiiSilt loamSugarcane 949400U680.3460.76

Veldkamp et al. [1998]

Costa RicaLoamyPasturecNT30030000AN72.96.8b1.3b

Veldkamp et al. [1998]

Costa RicaLoamyPasturecNT75.275.200A 11.4b2.52b

Veldkamp et al. [1998]

Costa RicaLoamyPasturecNT75.275.200U+ 11.6b2.22b

Veldkamp et al. [1998]

Costa RicaLoamyPasturecNT75.275.20 N 12.5b1.53b

Veldkamp et al. [1998]

Costa RicaLoamyPasturecNT75.275.200U 14.7b2.05b

Veldkamp et al. [1998]

Costa RicaLoamyPasturecNT75.218.8 × 400A, U+, N, U882.9b0.57b

Veldkamp et al. [1998]

Costa RicaLoamyPasturecNT75.218.8 × 400A, U+, N, U7613.6b2.98b

Veldkamp et al. [1998]

Costa RicaLoamyPasturecNT75.218.8 × 400A, U+, N, U879.5b1.5b

Veldkamp et al. [1998]

Costa RicaLoamyPasturecNT75.218.8 × 400A, U+, N, U8024.2b3.28b
Veldkamp and Keller [1997]Costa RicaLoamy(Andisol)Banana36036000AN50–802.95.7
Veldkamp and Keller [1997]Costa RicaClay (Inceptisol)Banana 36036000AN50–801.35.1

Jantalia et al. [2008]

BrazilClay (Ferrasol)Maize-wheatNT162450117 20–400.41 

Jantalia et al. [2008]

BrazilClay (Ferrasol)Maize-wheatT14145096 20–400.7 

Jantalia et al. [2008]

BrazilClay (Ferrasol)Sorghum-wheatNT2531050148 20–400.24 

Jantalia et al. [2008]

BrazilClay (Ferrasol)Sorghum-wheatT2531050148 20–400.29 
Watanabe et al. [2000]Thailand(Calciustalfs)Maize 62.562.500A10 (soil moist)0.22–0.44b 
Watanabe et al. [2000]Thailand(Haplustoxs)Maize 62.562.500U + A10 (soil moist)0.19–0.38b 
Watanabe et al. [2000]Thailand(Tropaqualfs)Maize 757500U10 (soil moist)0.12–0.24b 
Watanabe et al. [2000]Thailand(Ocharaqualfs)Maize 46.946.900A + N10 (soil moist)0.08–0.15b 

3.3.1 LAN-N2O

[35] As shown in Table 4, a large range of LAN-N2O values (0.01–24.2%) have been reported for tropical agricultural fields, which is wider than the range of 0.3–3% given in the IPCC Guidelines for National Greenhouse Gas Inventories [Klein et al., 2006]. The tropical variation is probably caused by a combination of soil management, environmental conditions, and N fertilizer application rates [Kim et al., 2013]. Some of the low values are related with low total N fertilizer inputs [e.g., Chapius-Lardy et al., 2009], a complex fertilizer management in which fertilizer is applied below the soil surface and is carefully timed to crop requirement [Matson et al., 1996], or low WFPS (< 40%) during the cropping season [Jantalia et al., 2008]. The higher LAN-N2O values were produced with high N application rates on very fertile soils that exceeded the N demand [Veldkamp et al., 1998]. Now the range of LAN-N2O values of 0.3%–6.1% obtained in this study from N-fertilized fields at the Venezuela savanna region (Table 2) is within the tropical reported values (Table 4), and the overall average of 1.9% is in the upper part of the IPCC range. As discussed earlier, in our study, relatively low soil moistures produce low LAN-N2O, whereas our NT management and NH4NO3 fertilization promote higher LAN values. Therefore, due to the fact that NT management is rapidly increasing in the tropics, its contribution to the global N2O agricultural source is also likely to increase.

[36] Due to the fact that there is a large range of LAN-N2O values (Table 4) and that soil moisture is a key factor affecting N2O soil emissions, we plotted the logarithm of LAN-N2O as a function of water-filled pore space using data from Table 4 and this study (Figure 1). We found a statistically significant positive correlation for both data sets: grey for this study (r = 0.707, p = 0.015, N = 11) and black for the data from Table 4, including our results (r = 0.457, p = 0.002, N = 45). We attribute the lower correlation (~ 46%) between LAN-N2O and WFPS for all the tropical data available in the literature not only to differences in the soil moisture measurement techniques used among the studies (such as depth of sampling and gravimetric or volumetric measurements) but also to differences in the fluxes estimated from chamber measurements [Parkin et al., 2012; Rochette and Eriksen-Hamel, 2008]. However, there is a significant positive relationship between LAN-N2O and soil moisture from tropical soils. Further systematization of soil water content measurement techniques and the chamber methodology in future studies would contribute to diminishing the high variability. Also, future studies need to take into consideration the dependency of direct N2O emission on N input, which could be linear or nonlinear [Kim et al., 2013].

[37] Considering that extension of tropical cultivated soils, from savanna or forest conversion, is rapidly occurring, more field studies are needed to establish appropriate LAN factors for the tropical agriculture. For that, using available data (Table 4 and our results), we made a box plot for LAN-N2O as a function of soil texture and calculated the simple average of each one (Figure 2). We found that LAN-N2O values from loamy soils are significantly larger (p < 0.05) than those from clay and sandy soils, which are not statistically different between them (Figure 2). Probably, this absence of difference between tropical LAN values from clayey soils and sandy soils could be attributable to factors that affect N2O production in each type of soil; i.e., aeration of sandy soils and high soil moisture conditions of clayey soils (N2O is reduced to N2) both lead to lower N2O emission [Davidson et al., 2000].

Figure 2.

Box plot of LAN-N2O values from tropical N-fertilized fields according to soil texture (data from Table 4 and this study). Mean ± standard deviation and median in parentheses for each soil are given. Different superscript letters indicate significant difference at p < 0.05. Study number for each soil are as follows: 22, clayey; 22, loamy; and 6, sandy. Asterisks indicate extreme outlier value, and filled circle indicates outlier value.

[38] Additionally, we used the simple average calculated above and calculated a weighted LAN-N2O value for tropical soils (3.0 ± 3.7%) based on the number of studies with different texture found in the literature (44% for clayey, 44% for loamy, and 12% for sandy) (Figure 2). High variability of this result could be attributable to variability in the length of the period of measurement, the frequency of measurement within that period, and high N fertilizer application rates combined with optimal soil moisture conditions of each study. However, this texture-weighted LAN-N2O value for tropical soils is three times higher than the IPCC default value of 1% [Klein et al., 2006], suggesting that a higher factor should be considered for cropping systems in tropical savanna regions, which are generally established on fertile loamy to clayey soils. This is in agreement with the findings of Crutzen et al. [2008], which indicate that N2O global emission from N fertilization is most likely underestimated. This Crutzen assertion has been confirmed by recent inverse modeling approaches that suggest a larger N2O land source in the tropics by a factor of 20 to 64% (Equator to 30°N) than prior estimates [Hirsh et al., 2006; Huang et al., 2008].

3.3.2 LAN-NO

[39] The overall average of LAN-NO of 0.9% (0.26–2.1%) obtained in this study from agricultural fields in the savanna region of Venezuela is considerably higher than the fertilize-induced emission factor of 0.5% proposed by Veldkamp and Keller [1997], which is based on 12 observations in temperate areas. Also, our value exceeds the global mean estimates of 0.7% derived by Bouwman et al. [2002]. On the other hand, with the exception of sugar cane in Maui (0.004–0.038%), the LAN-NO values reported for tropical sites are similar or higher (0.57–5.7%) than the one obtained in this work (Table 4).

[40] Similar to N2O, using Table 4 data and our results, we determined LAN-NO simple average values as a function of soil texture (Figure 3). In this case, we did not find statistical differences between LAN-NO values from clayey, loamy, and sandy soils (Figure 3). However, using these simple averages, we calculated a weighted LAN-NO value for tropical soils (1.3 ± 1.3%) based on the number of studies with different texture found in the literature (29% for clayey, 52% for loamy, and 19% for sandy) (Figure 3). Additionally, using the same data set, we found a positive correlation between LAN-NO from tropical soils and N fertilizer application rate (r = 0.677; p < 0.01; N = 31), suggesting that this factor could be used to estimate N fertilizer losses as NO.

Figure 3.

Box plot of LAN-NO values from tropical N-fertilized fields according to soil texture (data from Table 4 and this study). Mean ± standard deviation and median in parentheses for each soil are given. Different superscript letters indicate significant difference at p < 0.05. Study numbers for each soil are as follows: 9, clayey; 15, loamy; and 6, sandy. Asterisk indicates extreme outlier value, and filled circles indicate outlier value.

[41] The overall LAN-NO average from tropical N-fertilized fields according to soil texture of 1.3 ± 1.3% also has high variability, which could be attributable to the same factors proposed for LAN-N2O as a function of soil texture (factors that affect gas production, length and frequency of measurements, and combination of fertilizer application rates and soil moisture). However, this value is almost two times larger than the value estimated by Bouwman et al., 2002. Therefore, it is likely that the loss of applied N as NO in the tropic is higher than the one from temperate agricultural fields, and special attention should be paid to the quite high values obtained from tillage agricultural soils. This could be relevant to future tropospheric ozone production, in regions with low anthropogenic NO emissions.

[42] In contrast to LAN-N2O, we did not find a significant correlation between LAN-NO and soil moisture, probably because the differences in the measurement of this variable are more evident for this gas. We propose that it should be considered that the large emission pulses of NO are produced when small changes on top surface soil moisture occurs (e.g., rewetting due to soft rainfalls).

[43] Finally, our LAN-N2O and LAN-NO values estimated by soil textural classes (Figures 2 and 3) can be used for scaling up direct N2O and NO emissions from agricultural tropical areas when estimating national and regional N2O and NO inventories using Geographic Information Systems (GIS) methodology and soil texture maps.


[44] We determined the first LAN values from direct NO and N2O emissions for Venezuelan agroecosystems under the typical Venezuelan farmers agricultural management (tillage and no-tillage). Although we found a wide range of LAN-NO and LAN-N2O values (0.26–2.1% and 0.1–6.1%, respectively) from our cultivated soils, these values will be used to better constrain the Venezuelan national greenhouse gas inventory.

[45] It is important to point out that our N2O/NO soil-atmosphere fluxes were mainly affected by a combination of soil texture and rainfall pattern, which determined soil moistures and WFPS. Also, we found that the chemical composition and application rate of the N fertilizer could affect these N losses. Tillage or no-tillage land management and crop rotation also play important roles. Our LAN-NO values from tillage-managed fields were higher than the ones from no-tillage, whereas for N2O, higher values were recorded in no-tillage fields. Total LAN values (NO + N2O) (between 1.1 and 6.8% of the applied fertilizer) imply that mitigation strategies are necessary to reduce the emissions of NO and N2O from both types of Venezuelan agricultural practices. Nevertheless, no-tillage agricultural practices require particular attention because although they decrease soil erosion, our results show that they enhance N2O emissions as compared to tillage agriculture. Therefore, some mitigation strategies for these particular soils should be considered in the future.

[46] A large range of LAN-N2O and LAN-NO values were recorded from Venezuelan and other worldwide tropical cultivated and fertilized soils. This high variability could be associated with differences in fluxes estimated from chamber measurements and also to differences in field measurements, such as length and frequency of measurements and combination of fertilizer application rates and soil moisture.

[47] Overall average of 0.9% was derived for LAN-NO, and this is higher than previous global mean estimates. In the case of LAN-N2O, the average value of 1.9% is two times higher than the IPCC Tier 1 recommended value of 1%. Therefore, in order to avoid an underestimation of global N2O emissions, our results suggest that a higher LAN factor should be considered for cropping systems in tropical savanna regions.


[48] This work was supported by the U.S. Environmental Protection Agency Cooperative Agreement CR8211758-01-0 (1991–1994), the National Science Foundation grant NSF-0312004, and the Venezuelan National Science Foundation (Fondo Nacional de Ciencia, Tecnología e Innovación, FONACIT) grant G-205000435 (2005–2007). We thank Evelyn Cabrera and Manuel de Jesús Mujica (Instituto Nacional de Investigaciones Agrícolas, INIA) for assistance in soil physical analysis, and Adriana Giuliante, Rafael Rasse, Oscar Corona, and Alcides Rojas (Atmospheric Chemistry Lab, IVIC) for assistance in sampling fields and soil and gas analysis.