Comparison of different methods for the determination of total organic carbon and humic substances in Brazilian soils

Submitted on 1 1/09/2014 and appr oved on 07/28/2015. 1 Universidade Federal de Viçosa, Departamento de Solos, Viçosa, Minas Gerais, Brasil. raphael@ufv .br 2 Uberlândia, Minas Gerais, Brasil. ildeujr1107@hotmail.com 3 CNPq, Brasília, Distrito Federal, Brasil. eribeiro@cnpq.br 4 Universidade Federal do Espírito Santo, Departamento de Produção Vegetal, Alegre, Espírito Santo, Brasil. eduardo.mendonca@ufes.br *Corresponding author: raphael@ufv .br Comparison of different methods for the determination of total organic carbon and humic substances in Brazilian soils


INTRODUCTION
Nowadays there is a great interest in carbon cycling in the context of climate change.The dynamics of this element has been modified as a result of human activities, primarily due to fossil fuel use and secondarily to changes in land use (IPCC, 2013).Changes in soil organic matter contents are directly linked to the modifications of landuse, affecting this important carbon natural reservoir in the Earth.Hutchinson et al. (2007) discussed some management practices that contribute to the mitigation of CO 2 emissions by trapping C in soils, indicating challenges and opportunities for policymakers, farmers, and soil scientists.
It is well known the importance of soil organic matter for soil quality and plant growth.Thus, several methods have been proposed for its measurement (Walkley & Black, 1934;Mebius, 1960) and a number of papers discussed the most common ones.The establishment of the relationship among different methods of organic carbon determination is important for soil carbon inventories focusing "Land Use Change and Forestry" evaluations for the Kyoto Protocol and the fitting of data previously obtained by different analytical methodologies (Lettens et al., 2007).
Regarding the soil organic carbon analysis, the dichromate-oxidation method (Walkley-Black) is preferred for routine purposes or when the soils have high carbonate contents.In this procedure, the heat needed for oxidation can be obtained by the energy released from H 2 SO 4 or by an external source such as a hot plate or a digestion block.The temperature of 120 °C obtained from the reaction of concentrated H 2 SO 4 with water and dichromate may be enough to oxidize the active organic carbon forms, but not the recalcitrant forms (Allison, 1960).Usually, 77% of the total organic carbon is determined by this method.Therefore, a correction factor of 1.30 is recommended to convert the amount of organic carbon measured into total organic carbon (Nelson & Sommers, 1982;Gillman et al., 1986).
However, other correction factors are available to compensate this incomplete oxidation.For example, Mikhailova et al. (2003) indicated correction factors from 1.09 to 2.27 for Russian topsoils.
The loss-on-ignition method is an easy procedure for soil carbon analysis, requiring equipment available in most laboratories such as a muffle furnace, a drying oven, and a scale (Konen et al., 2002).This method causes partial oxidization of the organic matter using moderate to high temperatures, and the weight lost during the procedure is related to soil organic content.On the other hand, Grewal et al. (1991) demonstrated that this procedure may cause structural water loss, especially from oxy-hydroxides, as well as structural changes in some inorganic compounds.
The dry combustion method measures the CO 2 generated by the complete decomposition of organic material at temperatures at or above 900 °C.This method is currently considered as the most reliable, with recoveries close to 100% (Lettens et al., 2007), and it is used as reference method in several studies.However, these elemental analyzers are expensive, which is the main limiting factor for their use.
The aim of this work was to compare the most commonly used methods to measure total organic carbon in soils and humic substances of Brazilian soil samples, as well as to revise the recovery factor used for the dichromate-oxidation method.Some recently studies have made similar comparisons, but using a fewer number of soil classes and none of them evaluated humic substances.
Humic substances were fractionated according to the alkaline or acid solubility (Schnitzer, 1982).Soil fulvic (FA) and humic (HA) acids were solubilized after extraction with Na OH 0.1 mol L -1 , while the humin fraction remained in the residual portion.Then, the alkaline supernatant was acidified (pH < 2), promoting HA precipitation while FA remained soluble.The carbon content in each fraction was measured by the dichromateoxidation method (Walkley-Black), with (WBH) and without (WB) heating at 130 °C, for 30 min (Nelson & Sommers, 1982).Individual contents of total organic carbon (TOC) in fulvic acids, humic acids and humim (expressed in g C kg -1 soil) were used to calculate the TOC in humic substances with (THSH) and without (THS) heating in the analysis.
Soil samples were crushed to pass through a 500 µm sieve.Organic C was analyzed by dichromateoxidation method, with (WBH) and without (WB) heating at 130 °C, for 30 min (Nelson & Sommers, 1982), and by the loss-on-ignition (LOI) method of dry weight after 5 h at 450 °C (Davies, 1974) and 6 h at 600 °C (Goldin, 1987).The LOI results were divided by the factor 1.724 to change from organic matter to organic C. Soil samples were further grounded and passed through a 140 µm sieve to measure total C by the dry-combustion method in a CHN-LECO 600.
Results obtained with this method were considered as reference and used to measure the efficiency of the other methods.
Percent recovery (%R) of organic C (OC) was obtained by each evaluated method and compared with the dry-combustion (DC) method, according the following equation: The recovery factor (RF) for each method evaluated was obtained using the average of percent recovery ( ) considering all samples evaluated by the equation: All results were expressed on an oven-dry basis.All analyzes were done in triplicate.Data were submitted to regression and correlation analysis using the SAEG, a statistical package developed by the Federal University of Viçosa (SAEG, 2007).

RESULTS AND DISCUSSION
The amount of organic C obtained showed differences among the studied methods (Table 2) and, as expected, were highly correlated with each order, which is in agreement with the literature (Miyazawa et al., 2000;Brunetto et al., 2006;Jankauskas et al., 2006;Pereira et al., 2006;Bianchi et al., 2008).The highest correlation coefficient with the reference method (DC) was obtained by the dichromate-oxidation (Walkley-Black) with external heating (r = 0.98, n = 17, p < 0,001), which indicates the high concordance between the methods.When heating was not used, the correlation was lower (r = 0.92, n = 17, p < 0.001).
The recovered mean (M) of all soils (Table 2) for the dichromate-oxidation method without heating was lower than some reported values.Considering the average of percent recovery obtained for all samples (n = 17), the efficiency of this measurement was low, 71% for all soils and 81% considering only the Oxisols.These results indicate that the amount of OC in Brazilian soils is mostly underestimated by routine laboratories, which largely use the dichromate oxidation without heating.If this methodology is used, the recovery factor should be 1.41, or 1.23 for Oxisols.These values are intermediate to 1.33 suggested by Walkley & Black (1934), without using external heating.When external heating was used, the mean recovery was 102%, which is very close to the reference method (DC).Similar accurate estimation using WB with external heating was observed by Wang et al. (2012) for Chinese calcareous soils.On the other hand, the LOI method showed lower concordance with the DC method, as also expressed by the correlation coefficients [r = 0.85 (LOI450) and r = 0.90 (LOI600), n = 17, p < 0.001].Also, the estimated organic C amount (M) was almost two times higher than the reference method (DC), when the average of individual percent recovery (n = 17) was considered.This result should be related to loss of structural water from oxy-hydroxides and some mineral compounds due to the high temperature, as suggested by Grewal et al. (1991); Dias & Lima (2004); Escosteguy et al. (2007), and Segnini et al. (2008).
Effects of external heat on humic substances estimation were similar to those of soil samples.Using an additional heating source improved recovery power of the method, and the results were more similar to the DC method.These results will be discussed below.
The recovery factor (RF, Table 2) considering the mean value of percent recovery obtained for all samples evaluated was 1.41, 0.98, 0.59 and 0.53 for dichromateoxidation with and without external heating, and LOI at 450 o C and 600 o C, respectively.The factor close to 1 obtained for Walkley-Black method with external heating indicates high capacity of this procedure for oxidizing all carbon organic presented in the soil sample.
The relationship between total organic carbon contents of the reference method (DC) and those of the other methods may be better described by regression equations (Table 3).The lowest R 2 coefficients of the methods dichromate-oxidation without external heating and losson-ignition indicate no linear association between these methods and the DC method, and that the use of a factor may be not a good choice to the final C content in the soil sample.In addition, the intercepts are significantly different from zero (p < 0.01) for all equations obtained, which means that more accuracy can be obtained using not a single factor, but considering also the intercept.Some studies forced the intercept to zero to find a single factor, but this was avoided in this study.
When regressions were carried out for the Oxisols, different equations and R 2 coefficients were obtained, although intercepts were still different from zero.This points out the difficult in using a single factor for all types of soils.According to this statement, La Manna et al. (2007) verified effects of vegetation type on soil organic C contents, and Lettens et al. (2007) noted that RF can be influenced by land use, soil texture and depth.Maybe for this reason, Konen et al. (2002) suggested the inexistence of a universal factor applied to different soils, and De Vos et al. (2007) recommended a specific RF for each laboratory and soil type in order to standardize the results of soil C organic estimation.These results suggest that the factor 1.33 for soil organic carbon, as recommended by Walkley & Black (1934) and currently in use in Brazilian laboratories, should be revised and other criteria need to be considered for the correction of partial oxidizing reaction observed when external heating is not used.A possibility to increase accuracy of this procedure in routine analysis is to use the dichromate-oxidation method with external heating at 130 °C, for 30 min.
The large difference between the amount of carbon in the total humic substances measured by the dichromate oxidation without external heating (WB) and the reference method (DC) (Table 2) may be due to the low capacity of the former to oxidize the organic carbon in the humin fraction (Table 4), which is the humic fraction most resistant to decomposition and protected by mineral fraction.This result means that the humin is the main organic matter fraction underestimated by the Walkley-Black method without external heating.When external heating is used (WBH), the efficiency is nearly 100% (Table 2) and much more humim is recovery (Table 4).Therefore, when it is not possible to use the reference method, the amount of carbon in humin fraction should be measured by dichromate-oxidation method with external heating.The humic and fulvic acid fractions are in the extracted solution, so the mean difference between this method with and without heating is about 11%, which is much lower than 42% for the humin fraction.

Table 1 :
Soil classes of studied samples and their particle size composition

Table 2 :
Organic carbon content (g C kg -1 soil) from Brazilian soils and their humic substances according to different methods Organic carbon in soil humic substances measured by dichromate-oxidation with (THSH) and without (THS) external heating.3Numbers in the parenthesis refer to the percent recovery in relation to DC method (reference) in each sample.For mean value (M), numbers in the parenthesis are the average of all percent recovery (n=17). 4Mean values. 5Recovery factor considering average of all percent recovery (n=17)

Table 4 :
Organic carbon in the soil humic fractions extracted with the dichromate-oxidation method (Walkley-Black) with (WBH) and without (WB) external heating

Table 3 :
Relationship between total organic carbon (y) from the reference method (dry combustion) and organic carbon values (x) measured by different methods evaluated in soil and humic substances Methods used for soil organic carbon: dichromate-oxidation (Walkley-Black) with (WBH) and without (WB) external heating, loss-on-ignition at 450 o C (LOI450) and 600 o C (LOI600). 2 Total organic carbon in soil humic substances measured by dichromate-oxidation (Walkley-Black) with (THSH) and without (THS) external heating.** Significant (p < 0.01).All values expressed as g C kg -1 soil.