Published May, 2009

A New Spectrophotometric Method for

Rapid Semiquantitative Determination
of Soil Organic Carbon
Omar R. Harvey* Most methods to determine soil organic C (SOC) content are constrained by the time
Water Management & Hydrological Sciences or equipment required, thereby limiting their use in determining the spatial variability
Texas A&M Univ. of SOC across large areas. In this study, a new spectrophotometric method for the rapid
3408 TAMU determination of SOC was developed. The method is based on Beer’s law, A = εlc, where
College Station, TX 77843-3408 A is absorbance, ε is absorptivity, l is the path length, and c is the concentration of the
absorbing species, and uses soil extract absorbance at 300 nm (A300nm) as a proxy for SOC.
Bruce E. Herbert For organic C extractions, 1 mol L−1 HCl followed by a 0.25 mol L−1 NaOH–0.1 mol L−1
Dep. of Geology and Geophysics Na4P2O7⋅10H2O solution was used (1:250 soil/extractant ratio). Evaluation of the method
Texas A&M Univ. using 146 soil samples from 11 Major Land Resource Areas (MLRAs) across Texas indicated
3115 TAMU that SOC was linearly related to A300nm. For soils containing up to 50 g kg−1 SOC, the
College Station, TX 77843-3115 relationship could be described by the equation A300nm = 5.0 SOC (r2 = 0.89). The standard
error and relative prediction deviation associated with the SOC values determined using the
J. Pat Harris newly proposed method were 2.9 g kg−1 SOC and 3.2, respectively. Calibration sample size,
Texas Transportation Institute
MLRA, soil texture, and inorganic C content had no significant effect on SOC extractability
3135 TAMU
or method performance. In addition to being rapid, the method was accurate, stable, easy
College Station, TX 77843-3135
to execute, amendable to field use, and shows potential for use across large geographic areas
comprised of soils from different parent materials and land uses.
Eric A. Stiffler Abbreviations: EDTA, ethylenediamine tetraacetic acid; MLRA, major land resource area; SEp, standard
Dep. of Geology and Geophysics error of prediction; SOC, soil organic carbon; UV, ultraviolet.
Texas A&M Univ.
3115 TAMU
College Station, TX 77840-3115

Julie-Ann Crenwelge
Dep. of Agricultural Leadership, Education &
Communication
Texas A&M Univ.
NUTRIENT MANAGEMENT & SOIL & PLANT ANALYSIS

2116 TAMU
College Station, TX 77843-2116

P hysical, chemical, and biological processes occurring in soils

are greatly influenced by SOC content. Estimates of SOC
are often required for a wide variety of agricultural, environ-
and Zimmer, 2007), making them impractical for applications
requiring rapid analysis of a large number of samples. For such
applications and applications where spatial variability in SOC
mental, and engineering applications. Both quantitative and is more critical than measurement accuracy, semiquantitative
semiquantitative methods are available for SOC determina- methods provide an attractive alternative to quantitative meth-
tion (Schumacher, 2002). Although quantitative methods for ods. In addition to being faster and cheaper, semiquantitative
SOC are more accurate than semiquantitative methods, they methods for SOC determination often have the added advan-
are also typically more expensive and time consuming (Abella tage of being easily adapted for in-field use.
A number of semiquantitative methods are available for
Soil Sci. Soc. Am. J. 73:822-830
doi:10.2136/sssaj2008.0268 laboratory and in-field estimation of SOC. These include soil
Received 15 Aug. 2008. color measurements (Fernandez et al., 1988), loss-on-ignition
*Corresponding author ([email protected]). (Konen et al., 2002), remote sensing (Chen et al., 2005),
© Soil Science Society of America and nondestructive or destructive spectroscopic techniques
677 S. Segoe Rd. Madison WI 53711 USA (McCarty et al., 2002; Bowman et al., 1991). Selecting an ap-
All rights reserved. No part of this periodical may be reproduced or
propriate semiquantitative SOC method can be challenging
transmitted in any form or by any means, electronic or mechanical,
including photocopying, recording, or any information storage and
because each method has different advantages and disadvan-
retrieval system, without permission in writing from the publisher. tages. Semiquantitative methods vary in measurement accu-
Permission for printing and for reprinting the material contained racy, degree of site specificity, applicability for large-scale use,
herein has been obtained by the publisher. and environmental friendliness.

822 SSSAJ: Volume 73: Number 3 • May–June 2009

 For example, reported correlations between SOC and soil easily adaptable to field use, is less affected by soil properties
color for studies using quantitative soil color measurements and shows the potential for use across large geographic areas
to estimate SOC vary from 0.01 to 0.94 (Wills et al., 2007; comprised of soils from different parent materials and land
Konen et al., 2003; Lindbo et al., 1998; Ibarra-F. et al., 1995; uses. Procedures for a new ultraviolet (UV)–visible light spec-
Schulze et al., 1993; Fernandez et al., 1988), indicating that trophotometric method for indirectly estimating SOC were
the accuracy of soil color as a proxy for SOC varies spatially developed. In developing the method, we assumed that the
and is largely site dependent. Variability in the results from base extractable UV-active soil components/total SOC ratio
soil color–SOC studies have also been observed at different was relatively constant and therefore, for efficient extractions,
spatial scales, leading some researchers to conclude that soil soil extract absorbance could be used as a proxy for total SOC.
color–SOC relationships are more suited for small-scale rather Additionally we hypothesized that, for extractions of similar
than large-scale applications (Schulze et al., 1993; Fernandez efficiency, absorbance of the UV-active components in soil ex-
et al., 1988). tracts adheres to Beer’s law:
Studies evaluating loss-on-ignition (LOI) for estimating
SOC also suggest strong soil property, landscape, and land use A = ε lc [1]
influences on method performance (Konen et al., 2002; Abella
and Zimmer, 2007). For example, LOI is often only consid- where A is absorbance, ε is the proportionality term or absorp-
ered to be suitable for use with coarse-textured or noncalcare- tivity of the analyte, l is the path length through the sample,
ous soils. In addition to oxidation of SOC, loss in structural and c is the concentration of the analyte. We also hypothesized
water from clays and other inorganic constituents as well as that a single absorptivity can be found for all soils, irrespective
decomposition of carbonates and hydrated salts also contribute of other soil properties. We evaluated the feasibility of using
to the weight loss observed with LOI (Schulte and Hopkins, the proposed method for large-scale SOC determination.
1996). Loss-on-ignition is therefore likely to result in an over-
estimation of SOC. Abella and Zimmer (2007) also found MATERIALS AND METHODS
that the relationship between LOI and SOC varied across 10 Soils and Humic Material
ecosystem types. The addition of soil texture, pH, or CaCO3 A total of 146 (82 calcareous, 62 noncalcareous) soil samples
did not significantly improve the relationship, suggesting that from 11 MLRAs across Texas was used (Fig. 1). Soils as well as data on
factors such as land use and management were major factors their classification and physicochemical properties were obtained from
influencing the results. Texas A&M University Soil Characterization Laboratory. Data on
Wet oxidation methods for SOC determination, such as physicochemical soil properties included: soil pH (1:1 water extracts);
the widely used Walkley–Black method (Walkley, 1947), are sand, silt, and clay contents (Kilmer and Alexander, 1949); calcite
generally suitable for the analysis of a small number of samples. and dolomite content (Dreimanis, 1962); cation exchange capacity
For a large number of samples, the method becomes cumber- (U.S. Salinity Laboratory Staff, 1969); base saturation (Soil Survey
some and time consuming (Bowman, 1998). Of even greater Staff, 1996); and total SOC (Nelson and Sommers, 1982). Samples
concern is the fact that wet oxidation methods use and gener- were previously air dried and ground to <2 mm when received and
ate hazardous materials such as Cr6+ and concentrated acid. were used without further processing. For verification purposes, SOC
Nonoxidative colorimetric methods such as the basic ethylene- in triplicate samples of 25 randomly selected soils were determined
diamine tetraacetic acid (EDTA) method are typically more by difference between total C content (determined by elemental
environmentally friendly but are also prone to soil property analysis) and total inorganic C content (Dreimanis, 1962). Total C
effects (Bowman et al., 1991). For example, heating (85°C) contents for verification samples were measured by elemental analysis
and incubation times of up to 4 h are required for efficient using a Vario EL III CNS elemental analyzer (Elementar Americas,
extraction of SOC from soils with high clay and organic matter Mt. Laurel, NJ). Values were then compared with those obtained
contents. Additionally, Fe3+, a major constituent of most soils, from the Soil Characterization Laboratory database, which uses dry
is a serious interferent since it forms a colored complex with combustion for total C content (Nelson and Sommers, 1982) and
EDTA. The soil extract color formed due to Fe3+–EDTA com- the method of Dreimanis (1962) for inorganic C. Regression analysis
plexation could potentially be misinterpreted as color formed indicated that soil organic C determined for verification samples was
due to SOC, thereby resulting in overestimation of SOC. not significantly different (α = 0.05) from that determined by the Soil
Developing semiquantitative methods for SOC estima- Characterization Laboratory, hence SOC contents from the database
tion that are independent of soil properties, soil origin, or land were used in all analyses. Sampling and analytical techniques used
use are important for several reasons. In addition to providing by the Texas A&M Soil Characterization Laboratory are available at
a general method for SOC estimation, such methods would soildata.tamu.edu/methods.pdf (verified 20 Jan. 2009).
allow (i) better standardization of semiquantitative SOC meth- A commercial humic acid (Sigma-Aldrich, St. Louis, MO) was
ods, (ii) better estimation and modeling of the spatiotemporal used as reference material for spectral comparison of soil extracts
variability in SOC across small as well as large geographic ar- to humic or aromatic components. With the exception of sieving
eas, (iii) cross-study comparisons, and (iv) the development of through a 63-μm sieve, to remove large particulate matter, the refer-
more consistent databases of SOC pools under different land ence material was used without any further pretreatment.
use and management practices (Lal et al., 2001; Kern, 1994).
The purpose of this investigation was to develop and
evaluate an alternative semiquantitative method for estimating
SOC that, in addition to being quick, easily executable, and

SSSAJ: Volume 73: Number 3 • May–June 2009 823

 Method Development For method evaluation, absorbance at the wavelength selected
Initial experiments were geared toward identifying a suitable for analysis and SOC determined by dry combustion (SOCDC) data
extractant, soil/solution extraction ratio, and appropriate UV wave- for 73 (50% of the 146 soil samples) randomly selected samples were
length for SOC quantification. These experiments were done using used as calibration samples to estimate the absorptivity (ε) in Beer’s
triplicate samples of 27 soils representing a range of inorganic charac- law. The calculated ε was then used to determine SOC in the remain-
teristics (13 calcareous, 14 noncalcareous) and SOC contents (0–25 ing 73 evaluation samples based on their absorbance at the selected
g kg−1). The influence of soil/extractant ratios (1:20, 1:60, 1:150, wavelength. Method performance was evaluated by comparing the
1:200, 1:250, and 1:450) and three soil extractants on the extraction parameters (r2, slope, and intercept) of the SOCDC vs. spectrophoto-
of organic C were evaluated. The extractants tested were 0.25 mol L−1 metrically estimated SOC regression line with those of the ideal 1:1
NaOH, 0.25 mol L−1 NaOH + 0.05 mol L−1 EDTA (after Bowman line with an intercept and slope of 0 and 1, respectively. Additionally,
et al., 1991), and 0.1 mol L−1 HCl followed by 0.25 mol L−1 NaOH standard chemometric statistics such as the standard error of predic-
+ 0.1 mol L−1 Na4P2O7⋅10H2O (pyrophosphate) (after Schnitzer, tion (SEp) and the relative prediction deviation (RPD) associated with
1982). The UV–visible light (Vis) spectra (190–800 nm) for soil the spectrophotometrically determined SOC were calculated accord-
extracts from 12 evaluation samples were compared with each other ing to Islam et al. (2003). The SEp for the evaluation samples was
and with the spectra of the reference humic material. All chemicals calculated as
used were reagent grade and extractants were made in nanopure wa-
( )⎤⎥
1/2
ter (Diamond UV ultra-pure water system, Barnstead International, ⎡ ∑ SOCspect − SOC DC
Dubuque, IA). The UV-Vis spectra and UV absorbance data were SE p = ⎢ [2]
⎢ n −1 ⎥
collected on a Hitachi dual-beam spectrophotometer (U-3010) with ⎣ ⎦
a path length of 1 cm. The appropriate wavelength for SOC quanti-
fication was taken to be the wavelength around the absorbance max- where SOCspect is the SOC determined by the proposed method and
ima that resulted in the best correlation between absorbance and the n is the number of samples in the validation data set. The relative pre-
known SOC for the evaluation samples. diction deviation for the evaluation samples was calculated as

Method Evaluation RPD = SD DC SE p [3]

Before evaluating the method’s performance, an analysis of cova-
riance was conducted to assess potential contributions of sample ori- where SDDC is the standard deviation of the SOC determined by dry
gin and soil properties to A300nm, The analysis was conducted using combustion and SEp is from Eq. [2].
all 146 soil samples, with MLRA representing the sample origin. The Since the number of samples required for establishing a robust
soil properties evaluated were SOC, soil texture, and whether the soil calibration will influence the overall analysis time and cost, we were
was calcareous or noncalcareous. These soil properties were selected also interested in evaluating the effect of calibration sample size on
because they were known to influence organic C extractability. The the performance of the proposed method. In doing so, we adopted a
analysis of covariance was executed in SPSS (Version 14.0, SPSS Inc., more stochastic approach in evaluating method performance. Rather
Chicago), considering both single and interactive effects of factors. than a single randomization (as described above), multiple random-
izations consisting of different ratios of calibration/validation samples
were considered. Calibration/validation splits considered
were 1:9, 1:3, 1:2, 1:1, 2:1, 3:1, and 9:1. A total of 63
independent randomizations were performed for each
calibration/validation split, resulting in the generation of
63 unique calibration and validation sets for each split.
Method performance was subsequently evaluated by
comparing the distributions of calibration and validation
regression parameters and chemometric statistics (Islam
et al., 2003).

RESULTS AND DISCUSSION

Soil Organic Carbon Extraction
and Analysis
In addition to being from several distinctly
different land resource areas (Fig. 1), the soils used
covered a wide range of moisture and temperature
regimes as well as taxonomic families (Table 1), tex-
tural classes (Fig. 2), and chemical properties (Table
2). Such diversity is critical to method development
since both the physical and chemical properties of
soils are likely to influence organic C extractability
and subsequently UV absorbance–concentration
relationships (Stevenson, 1994). Although com-
Fig. 1. Major land resource areas from which samples were obtained (NRCS, 2007). mercial and soil humic material are considered to

824 SSSAJ: Volume 73: Number 3 • May–June 2009

Table 1. Taxonomic classification of soils used from each Major Land Resource Area (MLRA).
MLRA† Soil series Taxonomic class‡
1 Blackland Prairies Austin fine-silty, carbonatic, thermic Udorthentic Haplustolls
Heiden fine, smectitic, thermic Udic Haplusterts
Dimebox fine, smectitic, thermic Udic Haplusterts
Branyon fine, smectitic, thermic Udic Haplusterts
Houston Black fine, smectitic, thermic Udic Haplusterts
2 Claypan Area Robco loamy, siliceous, active, thermic Aquic Arenic Paleustalfs
Bonneville sandy-skeletal, mixed, mesic Humic Dystroxerepts
Tabor fine, smectitic, thermic Oxyaquic Vertic Paleustalfs
Lufkin fine, smectitic, thermic Oxyaquic Vertic Paleustalfs
Mabank fine, smectitic, thermic Oxyaquic Vertic Paleustalfs
Zulch fine, smectitic, thermic Udertic Paleustalfs
Annona fine, smectitic, thermic Vertic Paleudalfs
Burleson fine, smectitic, thermic Udic Haplusterts
Falba fine, smectitic, thermic Aquic Paleustalfs
Shiro fine, mixed, active, thermic Udic Paleustalfs
3 Coastal Marsh Mustang siliceous, hyperthermic Typic Psammaquents
4 Coastal Prairies Snakecreek fine-loamy, siliceous, semiactive, hyperthermic Aquic Glossudalfs
Orcadia fine, smectitic, hyperthermic Oxyaquic Glossudalfs
Hockley fine-loamy, siliceous, semiactive, hyperthermic Plinthic Paleudalfs
5 East Texas Timberlands Spurger fine, smectitic, thermic Albaquultic Hapludalfs
Raylake fine, smectitic, thermic Chromic Dystruderts
Mooreville fine-loamy, siliceous, active, thermic Fluvaquentic Dystrudepts
Gallime fine-loamy, siliceous, semiactive, thermic Glossic Paleudalfs
Nona fine-silty, siliceous, active, thermic Natric Vermaqualfs
Evadale fine-silty, siliceous, active, thermic Typic Glossaqualfs
Fetzer fine-loamy, siliceous, active, thermic Aquic Paleudalfs
6 Edwards Plateau Eckrant clayey-skeletal, smectitic, thermic Lithic Haplustolls
Noelke loamy-skeletal, mixed, superactive, thermic Lithic Petrocalcic Calciustolls
Cho loamy, carbonatic, thermic, shallow Petrocalcic Calciustolls
Ozona loamy, mixed, superactive, thermic, shallow Petrocalcic Calciustolls
Rumple clayey-skeletal, mixed, active, thermic Typic Argiustolls
Pedernales fine, mixed, superactive, thermic Typic Paleustalfs
Krum fine, smectitic, thermic Udertic Haplustolls
Venus fine-loamy, mixed, superactive, thermic Udic Calciustolls
7 Grand Prairies Denton fine-silty, carbonatic, thermic Udic Calciustolls
Lewisville fine-silty, mixed, active, thermic Udic Calciustolls
Topsey fine-loamy, carbonatic, thermic Udic Calciustolls
Brackett loamy, carbonatic, thermic, shallow Typic Haplustepts
8 High Plains Lamesa fine-loamy, mixed, superactive, thermic Aeric Endoaqualfs
Lenorah fine-loamy, mixed, superactive, calcareous, thermic Aeric Halaquepts
Hindman coarse-loamy, mixed, superactive, thermic Torrifluventic Haplustepts
Cedarlake fine-loamy, mixed, superactive, calcareous, thermic Typic Halaquepts
Seagraves fine-loamy, mixed, superactive, thermic Typic Haplustalfs
9 North Central Prairies Abilene fine, mixed, superactive, thermic Pachic Argiustolls
Bastrop fine-loamy, mixed, active, thermic Udic Paleustalfs
10 Rolling Plains Vernon fine, mixed, active, thermic Typic Haplustepts
Shep fine-loamy, mixed, superactive, thermic Calcic Haplustepts
11 Trans-Pecos Tornillo fine-loamy, mixed, superactive, hyperthermic Ustifluventic Haplocambids
Martillo fine, mixed, superactive, thermic Sodic Ustic Haplocambids
Straddlebug fine-loamy, mixed, superactive, thermic Sodic Ustic Haplocambids
Chilicotal loamy-skeletal, mixed, superactive, thermic Ustic Haplocalcids
Reagan fine-silty, mixed, superactive, thermic Ustic Haplocalcids
Potrillo coarse-loamy, mixed, superactive, thermic Ustic Haplocalcids
Pandale fine-loamy, mixed, superactive, thermic Ustic Haplocalcids
Nillo fine-silty, mixed, superactive, calcareous, thermic Ustic Torrifluvents
† Numbers correspond to Fig. 1.
‡ Subgroup level classification according to Soil Survey Division Staff (1993).

SSSAJ: Volume 73: Number 3 • May–June 2009 825

 Fig. 3. Ultraviolet–visible light spectra of soil extracts and reference
humic acid. Spectra were obtained using 1 mol L−1 HCl followed by
0.25 mol L−1 NaOH + 0.1 mol L−1 Na4P2O7⋅10H20 as an extractant
at a soil/extractant ratio of 1:250.
Fig. 2. Particle size distribution of soils used in the method
calibration and validation, n = 146. In evaluating the effect of the soil/extractant ratio on the
relationship between A300nm and SOC, we found that for a
be significantly different (Malcolm and MacCarthy, 1986), we given extractant and soil, a soil/extractant ratio of 1:250 pro-
found that the UV-Vis spectra for the soil extracts and reference duced the best linearity and correlation across the evaluation
humic acid in our study were qualitatively similar (Fig. 3). This range (0–25 g kg−1). Higher soil/extractant ratios (1:20, 1:60,
was consistent with recent work by Sierra et al. (2005), which 1:150, and 1:200) became nonlinear around 10 to 15 g kg−1
showed that irrespective of source, all organic matter exhib- SOC, while the lower soil/extractant ratio (1:450) resulted in
ited common fluorescence behavior in the UV region that was poor resolution at SOC below around 5 g kg−1 SOC. A soil/
attributable to humic components. Additionally, it is fairly extractant ratio of 1:250 was therefore selected for all subse-
well agreed that the absorption maxima (occurring around quent extractions. Further discussions comparing extractant
260–300 nm) observed in either material is attributable to the efficiency are therefore based on a 1:250 soil/extractant ratio.
presence of aromatic components (Traina et al., 1990; Novak Results from initial experiments showed that while NaOH
et al., 1992; Stevenson, 1994). Although the final pH of the provided quantitative extractions for SOC determination in
HCl + NaOH–pyrophosphate extracts (11.6 ± 0.3) were lower noncalcareous soils, for calcareous soils NaOH alone was not
than the NaOH (12.7 ± 0.02) and NaOH–EDTA (12.6 ± an efficient extractant (Fig. 4). This observation was consistent
0.01) extracts, we obtained a fairly narrow range of absorption with observations made by Bowman et al. (1991) for KOH-
maxima (280–300 nm) irrespective of the extractant used or only extracts. Stevenson (1994) suggested that both clay and
the soil tested. This suggested no apparent effect of extractant CaCO3 in soils can physically protect organic matter from ex-
on the extracted UV-Vis active SOC fraction and was consis- traction. While physical protection of SOC from extraction
tent with all extractants targeting the aromatic SOC fractions was probable in our study, we believe that the dominant ex-
(Traina et al., 1990; Novak et al., 1992). Normalizing the soil changeable cation(s) present in the system may also influence
extract spectra to that of the reference humic acid suggested organic C extractability. In comparing the solubility of a refer-
that the spectra were most similar around 290 to 320 nm (data ence humic material in NaOH (0.05 mol L−1) and saturated
not shown). We chose absorbance at 300 nm (A300nm) for our Ca(OH)2 (?0.025 mol L−1 or 0.05 mol L−1 OH−), we found
analyses; however, other wavelengths within the 290- to 320- that the absorbance of the NaOH extract was at least an order
nm range are expected to produce similar results. of magnitude higher than that of the Ca(OH)2 extract. Since
the equivalent OH− concentration and the pH of both so-
Table 2. Summary of selected soil chemical properties for soil samples lutions were similar (pH > 12), we believe that the primary
used, n = 146.
reason for the difference in absorbance observed was differ-
Base ences in the interaction of Ca2+ and Na+ with the humic
Statistic pH† CEC‡ SOC§ CaCO3¶ CaMg(CO3)2¶
saturation acid. It is plausible that Ca2+ forms an organo–Ca complex
cmolc kg−1 ———————— g kg−1 ——————– with the humic acid that is more condensed and less prone
Min. 3.6 1.4 0.0 1.0 0.1 40 to dissolution than an organo–Na complex, which is likely
Max. 9.0 61.6 46.0 797 152 1000 to be more dispersed. Other studies have also reported dif-
Median 7.6 17.0 4.6 130 12.5 100 ferences in organic matter behavior in the presence of mon-
Mean (SD)6.8 (1.5) 21.0 (15.6) 8.0 (9.0)217 (228)15.3 (18.6) 810 (290) ovalent vs. polyvalent cations. Majzik and Tombácz (2007)
† Determined in 1:1 soil/water mixtures. found that shear tolerance of humic acid–montmorillonite
‡ Cation exchange capacity. suspensions were enhanced with increased Ca2+loading due
§ Dry-combustion soil organic C. to increased flocculation. Stevenson (1994) also suggested
¶ For calcareous samples only, n = 82.

826 SSSAJ: Volume 73: Number 3 • May–June 2009

that, in many soils, polyvalent cations may maintain soil organic
matter in a flocculated and insoluble condition.
Bowman et al. (1991) found that a mixture of KOH and
EDTA was more efficient at the extraction of SOC from cal-
careous soils than NaHCO3 or KOH only. The results from
our initial experiments were consistent with this finding.
Evaluations using the NaOH–EDTA extractant mixture and
the 13 calcareous soils (shown in Fig. 4) showed an increase
in both A300nm and its correlation with SOC (r2 = 0.82) com-
pared with NaOH. Despite this apparent improvement in
correlation for calcareous soils, we found that when the non-
calcareous evaluation samples (n = 14) were included in the re-
gression, the A300nm–SOC relationship for the NaOH–EDTA
extracts was only linear up to 10 g kg−1 SOC, with an intercept
that was 2.5 times that of the A300nm–SOC relationship of
the HCl + NaOH–pyrophosphate extracts. These observations
suggested that the NaOH–EDTA extractant was inefficient in
soils with high SOC and was likely to overestimate SOC, par-
ticularly for soils containing less SOC. Nonlinearity at higher
SOC is possibly the reason why Bowman et al. (1991) recom-
mended that, when using KOH–EDTA as an extractant for
high-organic-matter, high-clay soils, heating at 85°C for 2 h
was needed for efficient SOC extraction. Bowman et al. (1991)
also noted that the formation of colored EDTA–Fe3+complexes
were likely to be a major interferent in estimating SOC using
Fig. 4. Correlation between extract absorbance at 300 nm (A300nm)
the KOH–EDTA extractant. Both temperature and interferent
and soil organic C (SOC) determined by dry combustion for selected
effects were likely to have affected our results for the NaOH– calcareous (n = 13) and noncalcareous (n = 14) soils using 0.25 mol L−1
EDTA extracts since all of our experiments were conducted at NaOH solution as an extractant and a soil/extractant ratio of 1:250.
room temperature and Fe3+ is a common constituent of oxi-
dized soils. In fact, it is possible that the higher intercept ob- source, extracted SOC fractions exhibited similar absorption
served in the NaOH–EDTA extracts compared with the HCl behavior around 300 nm (Sierra et al., 2005).
+ NaOH–pyrophosphate (which forms colorless complexes) Although the ability of the pyrophosphate anion to
extracts, was due to the presence of colored EDTA–Fe3+ com- form insoluble or colorless soluble complexes with met-
plexes, resulting in false positive detection or overestimation of als (Alexandrova, 1960) was important in enhancing SOC
SOC in the NaOH–EDTA extracts. extractability and removing interference effects, we believe
The best extraction results were obtained using the HCl that the initial addition of excess HCl was also advantageous.
+ NaOH–pyrophosphate extractant at a soil/extractant ratio Besides being important for carbonate dissolution, the added
of 1:250. In addition to having intercepts close to zero, the HCl probably played an important role in disrupting soil ag-
A300nm–SOC relationships were linear across the full SOC gregation and clay–organic interactions, thereby resulting in
evaluation range (0–25 g kg−1 SOC) with statistically similar increased dissolution of organic C (Stevenson, 1994).
slopes and intercepts (α = 0.05) for both noncalcareous (r2 =
0.89) and calcareous soils (r2 = 0.91). Therefore, a
Table 3. Analysis of covariance of the effect of soil organic C (SOC), Major Land
common equation could be used to describe the Resource Area (MLRA), soil texture, and whether soils were calcareous or noncal-
A300nm–SOC relationship for both soil types, sug- careous on soil extract absorbance at 300 nm. Extractions were carried out using
gesting that the HCl + NaOH–pyrophosphate 0.1 mol L−1 HCl followed by 0.25 mol L−1 NaOH + 0.1 mol L−1 Na4P2O7⋅10H20
extractant was equally efficient at extracting UV- as the soil extractant at a soil/extractant ratio of 1:250, n = 146.
active SOC from both classes of soils. This was con- Source of variation df Sum of squares Mean square F value P>F
sistent with the analysis of covariance results using SOC† 1 0.527 0.527 25.326 < 0.001
all 146 soil samples, which showed no significant MLRA 10 0.115 0.012 0.553 0.848
influence on the A300nm–SOC relationship wheth- Calcareous‡ 1 0.016 0.016 0.787 0.377
er soils were calcareous or not (Table 3). Table 3 Texture§ 13 0.143 0.011 0.529 0.901
also shows no significant influence of MLRA or soil MLRA × SOC 10 0.171 0.017 0.822 0.608
textural class on A300nm, thereby suggesting that Calcareous × SOC 1 0.000 0.000 0.007 0.935
neither SOC source nor clay content had any sig- Texture × SOC 13 0.302 0.023 1.116 0.356
nificant impact on the extraction efficiency of UV- Error 96 1.997 0.021
active SOC when HCl + NaOH–pyrophosphate Total 145 30.032
was used as the extractant. That MLRA did not sig- † Determined by dry combustion.
nificantly influence the A300nm–SOC relationship ‡ Whether soil was calcareous or not.
could be attributed to the fact that, irrespective of § Based on USDA textural class delineations.

SSSAJ: Volume 73: Number 3 • May–June 2009 827

 1. Weigh 0.1 g of air-dried soil into a 50-mL polypropylene
centrifuge tube with 5 mL of 1 mol L−1 HCl.
2. Shake by hand for 15 to 20 s at 1-min intervals for 5
min or on a wrist-action shaker for 5 min.
3. Add 20 mL of NaOH–pyrophosphate extractant
solution and shake by hand for 15 to 20 s at 1-min
intervals for an additional 5 min.
4. Filter approximately 10 mL of extract through 0.45-
μm polycarbonate syringe filters.
5. Measure the absorbance of the filtrate, relative to a
reagent blank, at 300 nm.
6. Use a calibration curve (developed using Beer’s law and
soils of known SOC content) to determine the SOC
content of the sample based on its measured absorbance.

Method Performance
The correlation between A300nm (obtained using the pro-
posed method) and SOC for the calibration samples (n = 73)
is shown in Fig. 5. In contrast to what was observed for the
NaOH-only extracts (Fig. 4), Fig. 5 indicates that for soil ex-
tracts obtained using the proposed method, a common regres-
sion line could be used to describe the relationship between
A300nm and SOC in both calcareous and noncalcareous soils,
thereby eliminating the need for developing separate calibra-
tions for calcareous and noncalcareous samples. Figure 5 high-
Fig. 5. Relationship between soil extract absorbance at 300 nm (A300nm)
and soil organic C (SOC) for calibration samples (n = 73) using the
lights two important features of the proposed method: (i) the
proposed method. Dashed lines show the 95% confidence interval. A300nm–SOC relationship of the HCl + NaOH–pyrophosphate
extracts followed Beer’s law; and (ii) the relationship was linear
Based on all the initial experimental data, the following pro- beyond the maximum SOC content of the evaluation samples
cedure for rapid determination of SOC in soils is proposed: (25 g kg−1 SOC).
Based on the regression equation in Fig. 5, Beer’s law for
calibration samples could be written as

A300nm = 5.0 SOC [4]

where εl = 5.0 and SOC is the known SOC content (in g kg−1)
for each of the 73 calibration samples. For a standard cuvette
with a path length of 1 cm, εl and ε would be equal. Equation
[4] can be rearranged to give

SOC = 20 A300nm [5]

Equation [5] is more intuitive for using A300nm to estimate

SOC. When Eq. [5] was used to estimate SOC in the 73 vali-
dation samples (not used in the calibration), estimated SOC
values were in close agreement with SOC determined by dry
combustion (Fig. 6). No significant difference (α = 0.05) was
found between the regression line in Fig. 6 and the ideal 1:1
line. The standard error and relative prediction deviation as-
sociated with spectrophotometric SOC (compared with dry-
combustion SOC) were 2.9 g kg−1 SOC and 3.2, respectively,
indicating that, in addition to being fairly accurate, the pro-
posed method is also robust, with the standard deviation of
dry-combustion-measured SOC being more than three times
the SEp (Islam et al., 2003).
Further evidence of method stability was observed by com-
Fig. 6. Comparison of soil organic C (SOC) for validation samples (n = 73)
as measured by dry combustion and determined spectrophotometrically
paring the effect of calibration sample size on calibration and
using the proposed method and the relationship shown in Eq. [3]. validation regression parameters. For a calibration sample size
Dashed lines show the 95% confidence interval. ranging from 10% (1:9 calibration/validation split) to 90% (9:1

828 SSSAJ: Volume 73: Number 3 • May–June 2009

calibration/validation split) of the 146 samples Table 4. Variation in method calibration and validation regression parameters when
used in the study, ε and r2 for calibration equa- the data set (n = 146) was split randomly into calibration and validation sets of dif-
ferent sizes. Values are the means (with the standard deviation in parentheses) based
tions were similar (Table 4). Table 4 also shows on 63 independent randomizations for each calibration/validation (C/V) split.
that, irrespective of the calibration sample size,
the accuracy of the proposed method in de- Calibration† Validation‡
termining SOC was similar, indicating that by C/V ε§ r2 Slope Intercept r2 SEp¶ RPD¶
using a single calibration equation developed 1:9 5.2 (0.5) 0.87 (0.09) 0.97 (0.10) 0.2 (0.2) 0.89 (0.01) 2.9 (0.3) 3.0 (0.3)
using quantitatively determined SOC and soil 1:3 5.1 (0.3) 0.88 (0.06) 0.99 (0.07) 0.2 (0.2) 0.89 (0.02) 2.9 (0.2) 2.9 (0.3)
extract absorbance from as few as 15 samples 1:2 5.2 (0.2) 0.89 (0.04) 0.96 (0.07) 0.2 (0.1) 0.89 (0.02) 2.9 (0.3) 3.0 (0.3)
(?10% of the 146 samples used in the study), 1:1 5.1 (0.2) 0.89 (0.03) 0.98 (0.07) 0.2 (0.3) 0.89 (0.03) 3.0 (0.3) 2.9 (0.4)
the SOC content of a large number of soils 2:1 5.1 (0.1) 0.89 (0.02) 0.97 (0.09) 0.3 (0.3) 0.89 (0.04) 2.9 (0.4) 3.0 (0.6)
could be determined. As with any other ana- 3:1 5.1 (0.1) 0.89 (0.02) 0.97 (0.09) 0.3 (0.4) 0.87 (0.06) 3.0 (0.4) 2.9 (0.6)
lytical technique, it is recommended that SOC 9:1 5.1 (0.1) 0.89 (0.01) 0.96 (0.16) 0.3 (0.9) 0.87 (0.12) 2.6 (0.8) 3.4 (1.4)
contents in the calibration and prediction data † Regression of soil extract absorbance at 300 nm against dry-combustion soil organic C (SOC).
sets cover a similar range. The results shown in ‡ Comparison of SOC determined by dry combustion to that determined
Table 4 indicate that the use of Eq. [3] with soil spectrophotometrically for samples not included in the calibration set.
extracts obtained using the proposed method § Absorptivity, for SOC measured in g kg−1 with a path length of 1 cm.
was suitable for estimating SOC content in ¶ Chemometric parameters: SEp, standard error of prediction; RPD, relative prediction deviation.
the soils used in this study. Since the method
loss-on-ignition in northern Arizona forest soils. Soil Sci. Soc. Am. J.
was not evaluated with samples containing 71:545–550.
−1
>50 g kg SOC, we would recommend developing a separate Alexandrova, L.N. 1960. The use of sodium pyrophosphate for isolating free
calibration for soils with higher SOC. humic substances and their organic-mineral compounds from the soil.
Sov. Soil Sci. (Engl. Transl.) 2:190–197.
CONCLUSIONS Bowman, R.A. 1998. A reevaluation of the chromic acid colorimetric procedure
for soil organic carbon. Commun. Soil Sci. Plant Anal. 29:501–508.
The ease of execution, rapid analysis, and fairly good accu- Bowman, R.A., W.D. Gunze, and D.J. Savory. 1991. Spectroscopic method
racy of the proposed method makes it an attractive alternative for estimating soil organic matter. Soil Sci. Soc. Am. J. 55:563–566.
for both field and laboratory use. Although not accurate enough Chen, F., D.E. Kissel, L.T. West, D. Rickman, J.C. Luvall, and W. Adkins.
to be considered quantitative, the proposed method would be 2005. Mapping surface soil organic carbon for crop fields with remote
suitable for quick analysis of a large number of samples where sensing. J. Soil Water Conserv. 61:51–57.
Dreimanis, A. 1962. Quantitative gasometric determination of calcite and
accuracy to within ±10% was not required. With good-quality dolomite by using Chittick apparatus. J. Sediment. Petrol. 32:520–529.
field spectrophotometers now readily available, the proposed Fernandez, R.A., D.G. Schulze, D.L. Coffin, and G.E. Van Scoyoc. 1988.
method could be easily adapted for field use. Modification for Color, organic carbon, and pesticide adsorption relationships in a soil
field use could include substituting weighing of soil samples landscape. Soil Sci. Soc. Am. J. 52:1023–1026.
with using a scoop, and making corrections for soil moisture Ibarra-F., F.A., J.R. Cox, M.H. Martin-R., T.A. Crowl, D.F. Post, R.W. Miller, and
G.A. Rasmussen. 1995. Relationship between buffelgrass survival, organic
and temperature effects, if necessary. Additionally since this carbon, and soil color in Mexico. Soil Sci. Soc. Am. J. 59:1120–1125.
study was conducted on air-dried soils and under room-tem- Islam, K., B. Singh, and A. McBratney. 2003. Simultaneous estimation of
perature conditions (?24°C), the method may need to be several soil properties by ultra-violet, visible, and near infrared reflectance
modified for wet soils or extreme temperature conditions. spectroscopy. Aust. J. Soil Res. 41:1101–1114.
The ratio of soluble aromatic/total SOC present in the Kern, J.S. 1994. Spatial pattern of soil organic carbon in the contiguous
United States. Soil Sci. Soc. Am. J. 58:439–455.
soil is likely to be the major limitation to applicability of the Kilmer, V.J., and L.T. Alexander. 1949. Methods of making mechanical
proposed method. The method assumes that this ratio is more analysis of soils. Soil Sci. 68:15–24.
or less constant across a large spatial extent and therefore can Konen, M.E., C.L. Burras, and J.A. Sandor. 2003. Organic carbon, texture,
be used as a proxy for total SOC. While this may be true for and quantitative color measurement relationships for cultivated soils in
most soils, since nonaromatics are usually rapidly biodegraded north central Iowa. Soil Sci. Soc. Am. J. 67:1823–1830.
Konen, M.E., P.M. Jacobs, C.L. Burras, B.J. Talaga, and J.A. Mason. 2002.
(Schnitzer, 1982), for soils comprising a significant nonaro- Equations for predicting soil organic carbon using loss-on-ignition for
matic or insoluble aromatic fraction, the method will probably north central U.S. soils. Soil Sci. Soc. Am. J. 66:1878–1881.
underestimate SOC. Lal, R., R.F. Follett, and J.M. Kimble. 2001. Research and development
priorities. p. 433–438. In R.F. Follett et al. (ed.) The potential of U.S.
grazing lands to sequester carbon and mitigate the Greenhouse Effect.
ACKNOWLEDGEMENTS Lewis Publ., Boca Raton, FL.
This work was funded by the Texas Department of Transportation Lindbo, D.L., M.C. Rabenhorst, and F.E. Rhoton. 1998. Quantifying hydric
(TXDOT). We thank Ms. Darlene Goehl of TXDOT for her soil indicators with soil color. p. 95–106. In M.C. Rabenhorst et al.
continued support and discussions throughout the project. We (ed.) Quantifying soil hydromorphology. SSSA Spec. Publ. 54. SSSA,
also thank Dr. C.T. Hallmark, Ms. D. Prochaska, and the staff of Madison, WI.
the TAMU Soil Characterization Laboratory for assistance with soil Majzik, A., and E. Tombácz. 2007. Interactions between humic acid
samples. Additionally, we would like to thank Dr. F. Hons and Dr. K. and montmorillonite in the presence of calcium ions: II. Colloidal
McInnes for presubmission manuscript review and suggestions. interactions: Charge state, dispersing and/or aggregation of particles in
suspension. Org. Geochem. 38:1330–1340.
Malcolm, R.L., and P. MacCarthy. 1986. Limitations in the use of commercial
REFERENCES humic acids in water and soil research. Environ. Sci. Technol. 20:904–911.
Abella, S.R., and B.W. Zimmer. 2007. Estimating organic carbon from McCarty, G.W., J.B. Reeves, V.B. Reeves, R.F. Follett, and J.M. Kimble. 2002.

SSSAJ: Volume 73: Number 3 • May–June 2009 829

 Mid-infrared and near-infrared diffuse reflectance spectroscopy for soil (TOC) in soils and sediments. Available at www.epa.gov/nerlesd1/cmb/
carbon measurement. Soil Sci. Soc. Am. J. 66:640–646. research/papers/bs116.pdf (accessed 12 Aug. 2008; verified 19 Jan.
Nelson, D.W., and L.E. Sommers. 1982. Total carbon, organic carbon, and 2009). USEPA Exposure Res. Lab., Las Vegas, NV.
organic matter. p. 539–580. In A.L. Page et al. (ed.) Methods of soil Sierra, M.M.D., M. Giovanela, E. Parlanti, and E.J. Soriano-Sierra. 2005.
analysis. Part 2. Chemical and microbiological properties. 2nd ed. Agron. Fluorescence fingerprint of fulvic and humic acids from varied origins
Monogr. 9. ASA and SSSA, Madison, WI. as viewed by single-scan and excitation/emission matrix techniques.
Novak, J.M., G.L. Mills, and P.M. Bertsch. 1992. Estimating the percent Chemosphere 58:715–733.
aromatic carbon in soil and aquatic humic substances using ultraviolet Soil Survey Division Staff. 1993. Soil survey manual. Agric. Handbk. 18. U.S.
absorbance spectrometry. J. Environ. Qual. 21:144–147. Gov. Print. Office, Washington, DC.
NRCS. 2007. Geospatial data gateway. Available at datagateway.nrcs.usda. Soil Survey Staff. 1996. Soil survey laboratory methods manual. Soil Surv.
gov/NextPage.aspx?Progress=1&AValue=1&QuickCounty=*&QuickSt Invest. Rep. 42. U.S. Gov. Print. Office, Washington, DC.
ate=Texas&ExtentMinX=-125.5&ExtentMinY=16.35&ExtentMaxX=- Stevenson, F.J. 1994. Humus chemistry: Genesis, composition, reactions. 2nd
65.0&ExtentMaxY=59.0&HitTab=2 (accessed 25 Dec. 2007, verified ed. John Wiley & Sons, New York.
26 Jan. 2009). NRCS, Washington, DC. Traina, S.J., J. Novak, and N.E. Smeck. 1990. An ultraviolet absorbance
Schnitzer, M. 1982. Organic matter characterization. p. 581–594. In A.L. Page method of estimating the percent aromatic carbon content of humic
et al. (ed.) Methods of soil analysis. Part 2. Chemical and microbiological acids. J. Environ. Qual. 19:151–153.
properties. 2nd ed. Agron. Monogr. 9. ASA and SSSA, Madison, WI. U.S. Salinity Laboratory Staff. 1969. Diagnosis and improvement of saline
Schulte, E.E., and B.G. Hopkins. 1996. Estimation of organic matter by weight and alkali soils. Rev.ed. Agric. Handbk. 60. U.S. Gov. Print. Office,
loss-on-ignition. p. 21–31. In F.R. Magdoff et al. (ed.) Soil organic matter: Washington, DC.
Analysis and interpretation. SSSA Spec. Publ. 46. SSSA, Madison, WI. Walkley, A. 1947. A critical examination of a rapid method for determining
Schulze, D.G., J.L. Nagel, G.E. Van Scoyoc, T.L. Henderson, M.F. organic carbon in soils:Effect of variations in digestion conditions and of
Baumgardner, and D.E. Stott. 1993. Significance of organic matter in inorganic soil constituents. Soil Sci. 37:251–257.
determining soil colors. p. 71–90. In J.M. Bigham and E.J. Ciolkosz Wills, S.A., C.L. Burras, and J.A. Sandor. 2007. Prediction of soil organic
(ed.) Soil color. SSSA Spec. Publ. 31. SSSA, Madison, WI. carbon content using field and laboratory measurements of soil color.
Schumacher, B.A. 2002. Methods for determination of total organic carbon Soil Sci. Soc. Am. J. 71:380–388.

830 SSSAJ: Volume 73: Number 3 • May–June 2009