SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 8 June 2007
Published in Soil Sci Soc Am J 71:1204-1214 (2007)
DOI: 10.2136/sssaj2006.0014
© 2007 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Zinn, Y. L.
Right arrow Articles by Resck, D. V. S.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Zinn, Y. L.
Right arrow Articles by Resck, D. V. S.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Zinn, Y. L.
Right arrow Articles by Resck, D. V. S.
Related Collections
Right arrow Soil Organic Matter
Right arrow Carbon Sequestration
Right arrow Tropical Soil Management

FOREST, RANGE & WILDLAND SOILS

Edaphic Controls on Soil Organic Carbon Retention in the Brazilian Cerrado: Texture and Mineralogy

Yuri L. Zinna,*, Rattan Lala, Jerry M. Bighama and Dimas V. S. Resckb

a School of Natural Resources, Ohio State Univ., 2021 Coffey Rd., Columbus, OH 43210-1085
b Embrapa Cerrados Agric. Research Center, P.O. Box 08223, 73310-970 Planaltina-DF, Brazil

* Corresponding author (zinn.14{at}osu.edu).


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Soil organic carbon (SOC) retention is a function of climate, vegetation, drainage, and management interactions, but also of intrinsic soil properties such as texture, mineralogy, and structure. To assess these edaphic controls, three soils of the Brazilian savanna (Cerrado) under similar climate, vegetation, and slope but of contrasting texture were sampled to 1-m depth and characterized for textural, chemical, and mineralogical properties, and SOC concentration (in bulk samples and clay, silt, and sand fractions). The basic assumption was that SOC particle size determines its retention mechanism: colloidal forms are retained by sorption, while particulate organic matter (>20 µm) can occur outside or inside aggregates. It was hypothesized that SOC retention is controlled simultaneously by soil texture, mineralogy, and depth. The three soils are clayey, loamy, and sandy Haplustox, all kaolinitic with minor contents of Fe and Al oxides, vermiculite, and illite. The SOC concentrations in particle size fractions were inversely related to the content of the respective fraction in soil (SOC dilution effect), thus SOC partition is better assessed by determination of SOC pools in each particle size on a total soil mass basis rather than on a size-fraction concentration basis. The positive linear relations between SOC and clay + silt concentrations in bulk soil were explained mostly by greater clay-sized SOC pools, which could be modeled as a function of clay content (related to specific surface area) and depth. Quantitative clay mineralogy showed that bulk SOC and clay-sized SOC pools were well correlated with Fe oxides in topsoil and amorphous Al oxides in subsoil, but this mineralogical control is secondary to the textural control, since it depends on clay content.

Abbreviations: BET, Brunauer–Emmet–Teller • CBD, citrate–bicarbonate–dithionite • POM, particulate organic matter • SOC, soil organic carbon • SSA, specific surface area


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Quantitative and qualitative data on SOC can be important inputs for land use planning and environmental modeling. Traditionally, SOC retention is thought to depend mainly on factors such as climate, topography, vegetation, and management (e.g., Jenny, 1941), and edaphic controls are often considered secondary. It is well known that anoxia related to poor soil drainage retards decomposition and results in high SOC accumulation (Tan et al., 2004), but this effect is usually related to topography. In aerobic soils, the edaphic factor most likely controlling SOC retention is texture, as reviewed by Zinn (2005). Direct relations between concentrations of SOC and clay (or clay + silt) were demonstrated for kaolinitic soils of the tropics (Feller and Beare, 1997) and different temperate soils in Iowa (Konen et al., 2003), eastern Canada (Carter et al., 2003), and Argentina (Galantini et al., 2004). Such correlations may not occur across broad geographic regions (Rühlmann, 1999; Müller and Höper, 2004), however, because the effect of diverse climate, topography, and land use can preponderate over edaphic controls. For instance, SOC and clay contents can be negatively correlated in regions where coarse-textured Spodosols and Histosols accumulate the highest C stocks (Silver et al., 2000; Vejre et al., 2003).

Despite the considerable evidence on the textural control, the exact reasons why texture affects SOC retention are poorly understood. Early proposed mechanisms included sorption and inactivation of SOC-decomposing enzymes in clayey soils, thus SOC accumulation (Bremer, 1965), and lower biomass input in sandy soils (Munn et al., 1978). Feller et al. (1996) concluded that correlations between SOC and clay concentrations in the tropics could be caused by stronger biomass inputs or aggregation in finer textured soils. Oades (1988) stated that clay content affects many soil properties and thus its relation with SOC is difficult to be proven causative, but he proposed that finer textured soils have more cation bridges (Ca2+ in neutral, Fe3+ and Al3+ in acidic soils) to bind organic molecules to clay particles. Christensen (1992) supported this hypothesis and concluded that the principal effect of higher clay (or clay + silt) contents is to stabilize derivates from the microbial decomposition of organic debris, which are readily lost in sandy soils. Probably all or many of these mechanisms occur simultaneously, but the sorption of colloidal or soluble SOC compounds to clay surfaces is certainly a major factor. Interactions between clays and microbial biomass may also affect SOC retention and dynamics (Hassink and Whitmore, 1997; Müller and Höper, 2004), and soil texture is often critical for SOC dynamics in croplands. For example, SOC loss with cultivation under tropical or subtropical climate is usually greater in coarse than in finer textured soils (Oades, 1988; Feller et al., 1991; Lepsch et al., 1994; Silva et al. (1994); Zinn et al., 2005a). In the Brazilian Amazon, Telles et al. (2003) observed that SOC in clayey Oxisols had a much longer turnover time (700–400 yr) than in coarser textured Ultisols and Spodosols (300–100 yr) for the 0- to 40-cm depth.

If SOC is mostly stabilized by sorption to clays, then clay activity (resulting from its surface area and charges) necessarily affects SOC retention. This mineralogical control is even less understood, however: although in vitro SOC sorption is often stronger or faster by smectites than other phyllosilicates (Varadachari et al., 1995; Dontsova and Bigham, 2005), field data indicate that soils rich in high-activity clays do not retain more SOC than those dominated by low-activity clay, including Oxisols (Dalal and Mayer, 1986; Moraes et al., 1995; Feller and Beare, 1997; Hassink, 1997; Wattel-Koekkoek et al., 2001; Gonzalez, 2002; Krull and Skjemstad, 2003; Wattel-Koekkoek and Buurman, 2004). Perhaps in consequence, SOC stocks in kaolinitic soil profiles in the tropics are similar to those in temperate, less-weathered soils (Sanchez and Logan, 1992) despite faster decomposition rates associated with higher temperatures. The considerable SOC retention in kaolinitic soils is partly due to significant contents of Fe and Al oxides, hydroxides, and oxyhydroxides (oxides for simplicity) in the clay fraction, as demonstrated in studies using synthetic oxides (e.g., Parfitt et al., 1977; Kaiser and Zech, 1997; Varadachari et al., 1997) and oxide-bearing soils (e.g., Jardine et al., 1989; Kahle et al., 2004). In the tropics, high SOC concentrations in soils or fractions rich in Fe oxides were demonstrated using magnetic susceptibility (Hughes, 1982; Shang and Tiessen, 1998, 2000) and selective extractions. For instance, the upland, kaolinitic soils studied by Wattel-Koekkoek and Buurman (2004) contained much higher Fe oxides (and SOC) than lowland, poorly drained Vertisols. Additionally, Fe and Al oxides may also enhance SOC stabilization by improving soil structure and aggregate stability (Oades and Waters, 1991; Feller et al., 1996).

The literature on edaphic controls on SOC retention mostly refers to surface layers, where pedoturbation and spatial variability are most intense. Mathematical descriptions of SOC profiles have been attempted since the 1960s and can be important predictive tools (Minasny et al., 2006), but there is a generalized dearth of attribute correlations and models for subsoil layers, posing additional complexity to fully understanding SOC retention mechanisms. In his pioneer work with Brazilian Oxisols, Bennema (1974) modeled SOC profiles and briefly discussed the effect of texture. More recently, Zinn et al. (2005b) studied three soils in the Cerrados of central Brazil and reported linear correlations between bulk SOC and clay + silt concentrations for seven layers to a depth of 1 m, with R2s ranging from 0.53 to 0.91. Based on the predictable, depthwise decline of the intercept and slope of those linear relations, SOC concentration could be modeled as a function of texture and depth, demonstrating that the textural control is exerted in surface and subsurface layers. The underlying mechanisms for this textural control were not studied in that work, however, and are the focus of this study.

The basic assumption in most studies is that SOC particle size is an indicator of its retention mechanism: colloidal and soluble forms are retained mostly by sorption on clays; particulate organic matter (POM; larger than sand size) may exist outside or inside aggregates, and silt-sized SOC commonly has intermediate properties. The rationale for this study is that the relative importance of these mechanisms is determined by soil texture, mineralogy, structure, and depth, which then control SOC retention. Therefore, the objective of this study was to describe and quantify the textural and mineralogical controls on SOC retention in the top 1 m of Cerrado soils. It was hypothesized that SOC retention, partition through particle size classes, and retention mechanisms are controlled by texture and depth, and that clay mineralogy affects SOC retention through sorption of colloidal and soluble forms.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The study sites comprised three soils of contrasting texture (sandy, loamy, and clayey), located in the Cerrado region of the State of Minas Gerais, about 50 km north of João Pinheiro (sandy and loamy soils) and 50 km north of Paracatu (clayey soil; see map in Zinn et al., 2005b). A ceteris paribus comparability of soil- and SOC-forming factors required in this case similar climate, topography, and vegetation, while soil texture varied with the parent material. The climate is the Aw of the Köppen classification (tropical humid, megathermic), with mean annual temperature of 22.5°C and mean annual precipitation of 1440 mm (Zinn, 2005). All soils are located on wide interfluves (<5% slope) of slightly dissected plateaus, at altitudes of approximately 570 m above sea level. For each textural group, samplings were conducted in paired plots of native vegetation and Eucalyptus camaldulensis Dehnh. plantations, part of a study on impacts of afforestation on SOC stocks and pools. The 1.25-ha native vegetation stands comprise a savanna woodland with lower (grasses and herbs) and higher strata (trees <15 m), as described by Oliveira-Filho and Ratter (2002), containing a litter layer of about 5 Mg ha–1 (Zinn, 2005) and similar aboveground biomass in all plots. The 25-ha eucalyptus stands were planted immediately after clearance of native vegetation, harvested at age 7 yr, then allowed to coppice until age 14 yr (at the time of sampling in August 2003). The experimental design was a factorial with three replications, in which the main factor was soil texture at three levels (sandy, loamy, clayey) and the secondary factor was land use at two levels (Cerrado and eucalyptus). Since changes in soil properties due to land use are not discussed here, however, a conventional ANOVA is not adequate to interpret quantitative differences due to soil type, which constitutes the main scope of this work. Thus, only mathematical regressions are used here for statistical analysis of data and variability is presented as standard errors of means, both conducted with a standard electronic spreadsheet and the software JMP IN 5.1 (SAS Institute, Cary, NC). In each soil type, true replication was achieved by sampling three independent and neighboring pairs of Cerrado/eucalyptus stands (see sampling map in Zinn, 2005). From each stand, five subsamples within a circle of 2-m radius were collected with a Dutch auger and bulked to form one composite soil sample. Soil samples were obtained for 0- to 5-, 5- to 10-, 10- to 20-, 20- to 30-, 30- to 40-, 50- to 60-, and 90- to 100-cm depths.

Soil samples were air dried, passed through a 2-mm sieve, and characterized according to the Brazilian standard procedures (Centro Nacional de Pesquisa de Solos, 1997). Briefly, soil texture was determined by the pipette method after dispersion by 3 h shaking in 0.1 M NaOH; pH was determined in deionized water and 1.0 M KCl; exchangeable Al+3, Ca+2, and Mg+2 were extracted with 1.0 M KCl and determined by atomic absorption in a 2380 AA spectrophotometer (PerkinElmer, Norwalk, CT); available K+ was extracted with 0.025 M H2SO4 + 0.05 M HCl and quantified by flame photometry; Al3+ + H+ were extracted with 1.0 M Ca acetate at pH 7.0 and titrated with 0.1 M NaOH.

Primary particle size separates (sand, silt, and clay) were obtained for three selected depths (0–5, 30–40, and 90–100 cm). For the 0- to 5-cm layer, soil samples from native Cerrado and eucalyptus plots were used separately; for the two other layers, samples from Cerrado and eucalyptus were mixed in a 1:1 ratio, because of the limited amount of soil material available and no significant differences in bulk SOC concentration due to land use for all depths (Zinn, 2005). Soil samples (enough to produce approximately 10 g of clay) were dispersed in 0.1 M NaOH and not by sonication, commonly used for SOC fractionations (Christensen, 1992). Chemical dispersion was chosen because: (i) it is the standard textural analysis procedure in Brazil, so results reported here are directly referable to the Brazilian literature; (ii) no chemical characterization of SOC other than C/N ratios was envisaged; (iii) total dispersion of Oxisols may not be achieved with ultrasound (Oorts et al., 2005); and (iv) a method for use in simple laboratories without ultrasound probes was desirable. The sand fraction was collected by wet sieving, whereas clay (<2 µm) was automatically separated from silt (2–20 µm) by sequential decantation and siphoning (Rutledge et al., 1967). The sand and silt separates were oven dried, whereas the suspended clay was flocculated with 0.5 M MgCl2, sequentially centrifuged and washed, quickly frozen in liquid N2, and freeze-dried. Total C and N in particle-size separates and bulk soil were determined by the dry combustion method in a Variomax CNHOS analyzer (Hanau, Germany) using 1.0 g of finely ground samples. An undesirable effect of dispersion with NaOH is that it is the standard solvent for humic and fulvic acids, and thus variable amounts of SOC can remain soluble after clay flocculation. In fact, C recovery from the three fractions varied between 70 and 90% for the 0- to 5-cm depth, and 90 to 100% for the 90- to 100-cm depth. Thus, when estimating SOC particle-size pools, the unrecovered SOC in each sample was accounted for in the clay size, which by convention adsorbs colloidal and soluble forms, although it may be expected that a minor part of the silt-sized pool may also contribute to the unrecovered portion.

Clay mineralogy was determined qualitatively by x-ray diffraction (CuK{alpha}, Ni filtered) of oriented slides with a Philips XRG-3100 generator (35 kV, 20 mA) and goniometer (Mahwah, NJ). Quantitative analyses involved different techniques. Thermogravimetry was used to determine gibbsite (dehydration weight loss of 31.2% in the 220–290°C range) and kaolinite (14% weight loss, 400–590°C) with a Seiko TG/DTA 200 (Torrance, CA), according to Karathanasis and Harris (1994). Total Fe oxides were extracted with citrate–bicarbonate–dithionite (CBD, Mehra and Jackson, 1960), whereas amorphous Fe and Al oxides were extracted with acid NH4 oxalate (Schwertmann, 1964). Iron and Al concentrations in both extracts were determined by atomic absorption in a 2380 AA spectrophotometer (PerkinElmer, Norwalk, CT). Finally, clay was dissolved in HF + aqua regia according to Sawhney and Stilwell (1994), total K in solution was determined by flame emission, and illite estimated assuming a K2O content of 10% w/w. The x-ray, thermogravimetry, and total K analyses were conducted on clay treated with hot H2O2 for SOC removal, whereas the selective extractions used intact clays. The Brunauer–Emmet–Teller (BET) method of N2 sorption–desorption (Brunauer et al., 1938) was used to determine the specific surface area (SSA) of soil <2 mm and the clay and silt fractions (without H2O2 removal of SOC) with a Micromeritics Flowsorb 2300 (Norcross, GA).


   RESULTS AND DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Soil Characterization
All sites consisted of highly weathered, kaolinitic soils characterized by low cation exchange capacity and moderate to strongly acid reaction (Table 1). The pH in KCl was always lower than pH in H2O (data not shown), i.e., none of the soils has an acric character. Regardless of the textural group, all soils were classified as Haplustox (Zinn, 2005), except for Replicate 1 of the sandy soils, classified as a Quartzipsamment because of clay contents consistently <150 g kg–1, the lower limit used by Soil Taxonomy for Oxisols. In samples from Replicates 2 and 3, clay contents were above that limit and are named "sandy Haplustox," but in fact they represent a Quartzipsamment–loamy Haplustox transition rather than a typical Oxisol. For a better illustration of the importance of these small textural differences in properties of the sandy soils, Tables 1, 3, and 4 show data of the Quartzipsamment separately.


View this table:
[in this window]
[in a new window]

  		Table 1. Textural, chemical and mineralogical characterization of soils (<2 mm) under native Cerrado (mean of three replicates, except for sandy soils), standard errors shown in parentheses.

 
Soil Texture and Soil Organic Carbon Size Partition
Table 2 shows the results of particle-size fractionation and C/N analyses of the different size fractions. Dispersion with NaOH was complete in all soils, as shown by clear quartz sand separates and x-ray diffraction patterns of silt samples showing quartz with only traces of kaolinite and Fe and Al oxides (Zinn, 2005). Despite the use of the same dispersion method, particle-size distribution differs somewhat from that shown in Table 1. These discrepancies are due to the pipette method used for textural characterization (weighing clay or clay + silt suspended in 20 mL from a 1-L suspension with 20 g of dispersed soil), contrasting with the repeated siphoning of large volumes of suspensions of 30 to 100 g of soil samples. This difference is especially important for the silt fractions in sandy soils, below the detection limit of the pipette method. The results in Table 2 for the 0- to 5-cm depth of the loamy and clayey Haplustox show means for Cerrado and eucalyptus samples, because there were no significant differences in SOC concentration and size partition between the two land uses. In the sandy soils, however, SOC size partition (but not bulk SOC concentration) changed under eucalyptus, thus only the data from native Cerrado soils were used (Zinn, 2005).


View this table:
[in this window]
[in a new window]

  		Table 2. Particle size distribution, soil organic C (SOC) concentration, and C/N ratio of particle size fractions. Mean of three replicates, standard errors shown in parentheses.

 
Soil texture strongly affected SOC concentrations in bulk soils (Table 1, see also Zinn et al., 2005b, Table 2) and also in particle size fractions (Table 2). The partition of SOC throughout particle sizes is important as indicators of its "quality": SOC in coarse fractions is usually less decomposed and has high C/N ratios, whereas an important part of SOC sorbed to clays is highly altered and has lower C/N ratios (Christensen, 1992). Table 2 shows that, at most depths but especially at 0 to 5 cm, the SOC concentration in any size fraction tended to be inversely related to the content of that fraction in the soil. Similar results were reported by Christensen (1992) and Schulten and Leinweber (2000), who reviewed the literature showing SOC enrichment factors for clay and silt (but not sand) fractions as exponentially and inversely related to each fraction's content. For surface soils under grassland in North America, Amelung et al. (1998) detected similar trends for the sand and clay fractions (but not silt), ascribing it to a "simple dilution effect," an expression that will also be used here. This effect is described as follows: (i) total SOC, which exists in a size continuum, is distributed during particle size fractionation throughout the clay, silt, and sand size fractions; (ii) assuming that any soil contains SOC in all size classes and in relatively constant proportions (valid for the soils studied), a particle size fraction is more or less SOC enriched if its content in soil is lower or higher, respectively. The SOC dilution effect is graphically depicted in Fig. 1 , and can be described by simple exponential or logarithmic functions. The fact that the SOC dilution effect is pronounced in the 0- to 5-cm depth, moderate for 30 to 40 cm, but virtually nonexistent for 90 to 100 cm suggests that the total amount of SOC (Table 1) is critical to the intensity of SOC dilution.


Figure 1
View larger version (24K):
[in this window]
[in a new window]

  		Fig. 1. Graphical view of the soil organic C (SOC) dilution effect for the sand, silt, and clay fractions (in the regression equations, y and x correspond to values on y and x axes).

 
Because the SOC dilution effect was originally reported for temperate soils and now in a tropical savanna, it probably is of widespread occurrence. The method for soil dispersion and fractionation may have little importance: in the present study, SOC concentrations in particle size fractions of the two coarser soils were similar to those obtained for the same soils sampled in 1997, but dispersed by sonication (Zinn et al., 2002). The dilution effect helps in explaining why silt fractions may show higher SOC concentrations than the clay, despite its much lower SSA and surface charges. Table 2 shows that the sandy and loamy soils have a very low (<2%) silt content, which shows consequently much higher SOC concentrations than the clay fractions. In the clayey Haplustox, with a considerable silt content, the clay fraction had the highest SOC concentration. The SOC dilution effect is critical in modeling the textural control on SOC size partition. The mathematical functions shown in Fig. 1 can be adjusted to different depths, but a more useful model would predict the SOC pool in each size fraction and depth. The SOC size pools for each soil type shown in Fig. 2 are the product of SOC concentration in each size fraction vs. the concentration of that fraction in soil, corrected for SOC solubilized by the NaOH. Considering all soils and depths, the overall trends for the relative (percentage, Fig. 2a) pools are: (i) the clay fraction is the main SOC pool, even in sandy topsoils, in accord with Christensen (1992) and Feller et al. (1991); (ii) the sand-sized SOC pool is greater in coarser soils (in accord with Feller et al., 1991; Kay, 1998; Bird et al., 2003), and also in surface layers because of abundant fresh residues; and (iii) the silt-sized SOC pool is nearly constant (7–12% of total SOC) throughout textures and profiles, as reported by Feller et al. (1991) on uncultivated topsoils (0–20 cm) in tropical Africa, except for the relatively more important silt-sized pool (>20% of total C). The sand-sized SOC pool decreases with depth, where C inputs are lower and the relative importance of clay SOC increases, in accord with Bird et al. (2003).


Figure 2
View larger version (29K):
[in this window]
[in a new window]

  		Fig. 2. Soil organic C (SOC) size pools (mean of three replicates) for three selected depths in (a) relative and (b) absolute units, representing total SOC partitioning throughout particle size fractions. Bars represent standard error.

 
The SOC size pools in absolute units (grams per kilogram of soil, Fig. 2b) indicate somewhat different trends, and explain further the relation of bulk SOC concentration with clay and silt contents. It is notable that the sand-sized SOC pool (i.e., POM-C) varies little with soil texture, being approximately 5.5 g kg–1 for 0 to 5 cm, 1.0 g kg–1 for 30 to 40 cm, and 0.5 g kg–1 for the 90- to 100-cm depth. These data suggest that, throughout the textural range studied, POM-C attains similar equilibrium levels, probably determined by the balanced input and decomposition of coarse residues, which in its turn depends mostly on soil depth and the climate–vegetation–slope combination. Conversely, the clay-sized SOC pool is proportional to clay content (see also Fig. 3a ), and the silt- sized SOC pool shows an intermediate trend. Thus, the direct relations between SOC and clay + silt concentrations are explained by greater SOC pools in the clay size, and secondarily in the silt size.


Figure 3
View larger version (26K):
[in this window]
[in a new window]

  		Fig. 3. (a) Clay-sized soil organic C (SOC) pool as a function of clay content throughout the profiles; (b) plot of intercepts and slopes [from linear relations in (a)] vs. depth; and (c) plot of estimated vs. measured clay-sized SOC pool (dotted lines mark the 95% confidence interval).

 
The successful modeling of bulk SOC concentrations based on texture and depth (Zinn et al., 2005b) suggested the possibility of similarly modeling the SOC size partition in soil profiles. First, the absolute SOC pools in Fig. 2b were plotted as a function of respective fraction contents in soil. No relations were found for sand (not shown), but linear functions explained well clay-sized SOC pools for the three depths (Fig. 3a). The intercepts and slopes of the linear functions in Fig. 3a were plotted against depth, and were well explained by power and logarithm functions, respectively (Fig. 3b). These functions were then used to parameterize Eq. [1]:

Formula 1[1]
where d is depth in centimeters, clay content is in grams per kilogram, and the SOC pool is in grams of clay-sized SOC per kilogram of soil. Figure 3c shows the plot and the 95% confidence interval of values measured vs. estimated by Eq. [1], resulting in the following fit:

Formula 2[2]
where NS is nonsignificant. Although Eq. [1] explained the data well, the intercept and slope parameters were modeled based on only three depths (n = 3 for Fig. 3b), less than the seven depths used by Zinn et al. (2005b), so this must be considered an exploratory model. Considering that the sand-sized SOC pool is constant for the range of depth and texture, the silt-sized SOC pool can be estimated by difference with total (bulk) SOC concentration.

Soil Texture, Specific Surface Area, and Soil Organic Carbon Retention
The retention of SOC is probably limited by the surface area available for sorption (Kaiser and Guggenberger, 2003). Table 3 shows that BET-N2 SSA of bulk soils (<2 mm) varies little among replicates of the same soil and depth, and is proportional to clay content and depth. Figure 4a shows the SSA of bulk soil as a function of clay content. The same slopes and similar intercepts for the three depths indicate that SSA is mostly determined by clay content with little or no effect of depth, i.e., higher SSA values in subsoil are due to slightly higher clay contents (Table 1). On the other hand, SOC concentrations were also correlated with SSA, but with a pronounced effect of depth (Fig. 4b). A model to predict SOC based on texture, SSA, and depth would be of little use since SSA determination is difficult and data are scarce in the literature. On the other hand, prediction of SSA based on texture, SOC, and depth can be intrinsically useful and also give some insight into the relation between SSA and SOC. A multivariate regression of SSA as a function of texture (clay, silt, and fine and coarse sand), SOC concentration, and depth showed that SSA can be modeled based solely on clay, silt, and SOC concentrations. Figure 4c shows the fit of Eq. [3]:

Formula 3[3]
where clay, silt, and SOC concentrations are expressed in grams per kilogram of soil.


View this table:
[in this window]
[in a new window]

  		Table 3. Specific surface area of bulk soil and clay and silt fractions (soil organic C not removed by peroxide). Mean of three replicates, except for sandy soils and silt fractions (n = 1), standard errors shown in parentheses.

 

Figure 4
View larger version (23K):
[in this window]
[in a new window]

  		Fig. 4. a) Linear relations between Brunauer–Emmet–Teller N2 specific surface area (SSA) and clay content (n = 9); (b) linear relations between soil organic C (SOC) concentrations and SSA n = 9); and (c) fit of the model of SSA as a function of clay, silt, and SOC contents (n = 27, dotted lines mark the 95% confidence interval).

 
Although SSA tended to increase in subsoil layers, depth was not a significant component of Eq. [3], contrasting with Eq. [1] and the SOC-prediction model by Zinn et al. (2005b), both using soil depth as an independent variable. According to Eq. [3], the effect of SOC is to decrease SSA, which can be explained by organic matter coating clay surfaces that would otherwise be exposed to N2 (see also Christensen, 1992), which is corroborated also in the SSA of clay fractions (Table 3). Additionally, the SSA estimator for silt in Eq. [3] is similar to that for clay, although the latter showed much higher measured SSA values (Table 3). Silt SSA was proportional to its SOC concentration, which in its turn is determined by depth and the SOC dilution effect, as follows:

Formula 4[4]
where SOC is expressed as grams of SOC per kilogram of silt. Equation [4] indicates that any aggregation of silt grains promoted by SOC does not result in occlusion of mineral surfaces, as is the case for bulk soil and the clay fraction. Thus, the SSA estimators for clay and silt fractions in Eq. [3] are restricted to that specific regression, and may be applicable only to soils of comparable texture and mineralogy, which however comprise a large area of the humid tropics.

The apparently contrasting effect of depth on bulk SOC concentration and SSA indicates that for a textural range of soils, SOC retention for a specific depth is proportional to the <20-µm fraction, because a finer texture results in higher SSA and thus SOC sorption. The fact that SOC concentrations decrease toward the subsoil, while SSA increases slightly, indicates not only the occlusion of clay surfaces by SOC, but also that the input of organic matter to a specific depth is the main factor affecting SOC retention, followed then by texture. Thus, across a textural range, SOC concentrations are directly related to SSA, but in a single soil profile, SSA is inversely related to SOC concentration.

Soil Texture and Carbon/Nitrogen Ratios
The C/N ratio is an indication of SOC quality and decomposition status. For bulk soils, C/N ratios decline with depth and tend to increase with clay content, except for the 90- to 100-cm depth (Table 1). For particle-size separates (Table 2), C/N ratios are proportional to particle size, in accord with Christensen (1992). For the clay fraction, C/N ratios vary little (10–12) with texture, tending to decrease slightly in the subsoil. For the sand fractions, C/N ratios were >20, indicating as expected a lower degree of alteration. The high C/N ratios in deeper layers in the two coarser textured soils, reaching 75 for the loamy Haplustox at 90 to 100 cm, probably reflect the presence of charcoal resulting from ancient fire events (Teixeira et al., 2002), visible in the sand separates. Finally, C/N ratios of silt fractions are intermediate between the clay and sand in topsoil, approaching those of clay in subsoil layers, which suggests that silt-sized SOC differs from clay-sized SOC, at least in surface layers. Silt-sized SOC is less decomposed and probably not retained by sorption, as also indicated by the SSA function (Eq. [4]). Based on SOC vs. clay + silt relations, Hassink (1997) and Six et al. (2002) suggested the existence of a "silt + clay associated or protected SOC" pool. The present data suggest instead, however, a heterogeneous composition in which any SOC protection is mostly due to sorption by clays, while silt-sized SOC is probably a mass of organic microdebris or colloidal clots loosely associated with the mineral component.

Clay Mineralogy and Soil Organic Carbon Retention
The quantitative mineralogy of selected soil depths is shown in Table 4. The sum of mineral fractions does not attain the total of 1000 g kg–1 due to the difficulty in quantifying hydroxy-interlayered vermiculite (HIV), residual water, and other amorphous components. Also, quantification of goethite by thermogravimetry was not feasible due to interference of HIV; hence the crystalline Fe oxides represent goethite + hematite. In all cases, however, the mineral composition was determined to approximately 80%, and the results are consistent with the x-ray diffraction patterns. Clay is almost totally kaolinitic in the Quartzipsamment, with small amounts of HIV, gibbsite, and goethite. The importance of the accessory minerals, including hematite and illite, increases gradually with increase in clay content until reaching significant amounts of Fe and Al oxides and illite in the clayey Haplustox. Amorphous Fe oxides are low (<8%) in relation to total Fe oxides; this is not the case for amorphous Al oxides, which are approximately 20% of the gibbsite content. There is little deviation among replicates and depths for each soil type, except for amorphous Fe and Al oxides.


View this table:
[in this window]
[in a new window]

  		Table 4. Quantitative mineralogy of clay fractions from selected depths of the studied soils. Numbers in parentheses are standard errors. Crystalline Fe2O3 (goethite + hematite) is the difference between cirate–bicarbonate–dithionite and oxalate extracts.

 
The BET-N2 SSA values for the clay fraction in the clayey Haplustox (Table 3) are similar to those determined by Feller et al. (1992) for another site in the Cerrado. Higher clay SSA in lower depths (except for the loamy Haplustox) may be explained by preferential illuviation of finer clay particles or the concurrent decrease in SOC concentration, as explained before. The latter hypothesis is supported by the negative correlations between SOC and SSA in clay fractions (Table 5), especially for the SOC-rich 0- to 5-cm depth. For all depths, SSA is best correlated (negatively) with kaolinite contents, probably because of larger crystallite size of kaolinite in comparison to accessory minerals. Amorphous Al oxides and crystalline Fe oxides are also well correlated (positively) with SSA, as reported earlier by Feller et al. (1992); however, clay SSA could not be modeled based on quantitative clay mineralogy by multivariate or other regression functions.


View this table:
[in this window]
[in a new window]

  		Table 5. Correlations between oxalate (ox) and citrate-bicarbonate–dithionite (cbd) extractable mineral contents and specific surface area (SSA), soil organic C (SOC), total N, and C/N ratios in clay fractions at selected depths.

 
Both SOC and total N concentrations are negatively correlated with Fe and Al oxide concentrations in the clay fractions (Table 5). This unexpected trend is in fact caused by the SOC dilution effect discussed above, which means that correlations between SOC concentrations and mineralogy are flawed if calculated for the isolated clay fraction. Alternatively, when concentrations of each clay mineral in whole soils (calculated from data in Table 2) were correlated with bulk SOC concentration, strong correlations were obtained with all minerals shown in Table 4, a consequence of the strong correlations between SOC and clay + silt (Zinn et al., 2005b). Thus, only correlations better than those with clay + silt can be considered indicative of a mineralogical control, which occurred for Fe oxides and amorphous Al oxides (Fig. 5 ). There is a clear effect of depth on the correlations with SOC concentration: crystalline Fe oxides were closely related (R2 = 0.91) for the 0- to 5-, but not for the 90- to 100-cm depth (R2 = 0.66). Conversely, amorphous Al oxides were better correlated with SOC in the 90- to 100- (R2 = 0.90) than 0- to 5-cm depth (R2 = 0.75). As a comparison, the coefficients for SOC vs. clay + silt (Zinn et al., 2005b) were R2 = 0.77 for the 0- to 5-cm depth, and R2 = 0.85 for the 90- to 100-cm depth. For the 30- to 40-cm depth, SOC correlations with crystalline Fe and amorphous Al oxides were similar. Kleber et al. (2005) reported that SOC concentrations in 12 acid subsoils (mostly under temperate climate) were best explained by amorphous Fe oxides, but especially by combined Fe and Al amorphous oxides, which was partially corroborated by desorption with pyrophosphate. It is also possible that, in subsurface layers of Oxisols, amorphous Al oxides are in equilibrium with "exchangeable" Al3+ forms, that may be complexed by organic compounds and reduce their mineralization rate (Sollins et al., 1996; Schwesig et al., 2003). This stabilization by amorphous Al oxides may also occur in surface layers, but this effect is obscured by the sorption of much higher amounts of colloidal SOC by Fe oxides. Correlations with amorphous Fe oxides were lower than with crystalline Fe oxides, perhaps because of the relatively low contents of the former. When only the clay-sized SOC pool is considered, all correlations were slightly improved (not shown, see Zinn, 2005).


Figure 5
View larger version (25K):
[in this window]
[in a new window]

  		Fig. 5. Linear relations (n = 9) between concentrations of total soil organic C (SOC) and clay-sized crystalline Fe oxides (cryst. Fe2O3) and amorphous Fe and Al oxides (Fe2O3- and Al2O3-oxal.) at three selected depths.

 
Detection of the SOC dilution effect is critical also for studies on mineralogy–SOC relations. Wattel-Koekkoek et al. (2001) reported similar SOC concentrations in clay fractions of six kaolinitic and six smectitic tropical soils and concluded that SOC in soil clays is independent of mineralogy. This conclusion is partially undermined, however, because in that work all smectitic soils were very clayey and all but one kaolinitic soil had <30% clay, so the SOC dilution effect was strong: g SOC kg–1 clay = 40.97exp(–0.0013clay) (R2 = 0.62), where clay is in grams per kilogram of soil. Studying long-cultivated, thick Ap horizons of seven illitic soils within a narrow range of texture (all soils >70% silt) and Fe oxide contents (22.5–26.3 g CBD-extractable Fe kg–1 clay), Kahle et al. (2002) reported a negative correlation between SOC and CBD-extractable Fe concentrations in clay fractions that could not be explained. But when their data for whole soils are correlated, SOC was correlated significantly but weakly with clay (R2 = 0.50) and CBD-extractable Fe (R2 = 0.28) contents. When data by Wiseman and Puttmann (2005) for three German forest soils are recalculated (excluding Gley and arable soils), for the 10- to 40-cm depth, bulk SOC is better correlated with oxalate-extractable Fe (R2 = 0.85, n = 6) than with oxalate-extractable Al (R2 = 0.65, n = 6). For the 40- to 110-cm depth, correlations with oxalate-extractable Al (R2 = 0.88, n = 9) were better than with oxalate-extractable Fe (R2 = 0.65, n = 9). Thus, reinterpretation of the latter two works after accounting for the SOC dilution effect produced correlations that support the conclusions drawn here, including the differential role of Fe and Al oxides in stabilizing SOC in surface and subsoil layers.

The preferential sorption of SOC by Fe and Al oxides rather than by phyllosilicates has been demonstrated by many independent workers for other soil orders (Jardine et al., 1989; Kalbitz et al., 2000; Kaiser and Guggenberger, 2003; Kahle et al., 2004). Nevertheless, it is evident that the literature on mineralogical controls on SOC retention is in need of a comprehensive review that accounts for variability due to: (i) environmental factors—climate, soil order, texture, organic inputs, etc.; (ii) in situ and in vitro experimental conditions; (iii) sorption mechanisms—e.g., cation bridges and coordination—associated with specific mineral and organic types; and (iv) eventual saturation of the sorption capacity of common SOC compounds by clay minerals, according to their crystal sizes and aggregation in the soil fabric.


   CONCLUSIONS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
The strong correlation between SOC and clay + silt concentrations in bulk soils can be mostly explained by enhanced sorption of colloidal-soluble SOC by a larger surface area resulting from higher clay contents. The role of silt in this textural control is probably indirect, i.e., silt particles contribute to the soil plasma and thus to an overall finer texture that favors SOC stabilization. Therefore, higher clay + silt contents result in higher silt- and clay-sized SOC pools and thus higher bulk SOC concentrations. The role of the clay fraction in SOC retention is not homogeneous, since SOC concentrations are better correlated with contents of Fe oxides and amorphous Al oxides than with other clay minerals such as kaolinite. These data can also help to explain why soils dominated by high-activity clays are similar to those with low-activity clays in regard to SOC retention.


   ACKNOWLEDGMENTS
 
This work is part of the doctoral dissertation of Y.L. Zinn, sponsored by the Capes Foundation (Ministry of Education, Brazil) and by a Presidential Fellowship from the Graduate School of the Ohio State University (OSU). We thank Embrapa Cerrados (in special Mrs. Jesuíno S. Caldas and Mr. Wantuir C. Vieira) and V&M Florestal Co. (in the person of Dr. Hélder B. Andrade) for their support with the sampling operations. We also thank Mr. F.S. Jones and Y. Raut (Soil Science Dep., OSU) for their important help in the laboratory procedures.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Current address: The Capes Foundation, Ministry of Education, Cx. Postal 365, Brasília DF 70375, Brazil

All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher.

Received for publication January 11, 2006.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Soil Sci.Home page
Y. L. Zinn, R. Lal, J. M. Bigham, and D. V. S. Resck
Edaphic Controls on Soil Organic Carbon Retention in the Brazilian Cerrado: Soil Structure
Soil Sci. Soc. Am. J., June 8, 2007; 71(4): 1215 - 1224.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Zinn, Y. L.
Right arrow Articles by Resck, D. V. S.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Zinn, Y. L.
Right arrow Articles by Resck, D. V. S.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Zinn, Y. L.
Right arrow Articles by Resck, D. V. S.
Related Collections
Right arrow Soil Organic Matter
Right arrow Carbon Sequestration
Right arrow Tropical Soil Management

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome