Sampling sites

Chile. A sampling area was selected along a longitudinal transect between 33º 30’ and 35º 02’ Latitude S and between 70º 45’ and 71º 48’ Longitude W in the Central Valley of Chile. The soil samples were taken on September 2006 over a total of 13 modal pedons series belonging to Alfisol, Entisol, Inceptisol, Mollisol and Vertisol orders (^{Soil Survey Staff, 1999}). The soils formed under similar vegetation, parent material and climatic conditions (^{Hajek and Di Castri, 1975}). Most soils originated from alluvial sediments subject to seasonal droughts and floods (^{CIREN, 2001}). The selected soils were mostly under maize monoculture with conventional tillage in the farmer’s field for more than 30 years (^{INE, 2007}). The maize crop yield fluctuates between 12 and 15 Mg grain ha^-1 (15% moisture). Depending on the amount of crop residues, they are burned or incorporated by conventional plowing to the Ap horizon (0-0.20 m). Mean annual precipitation in this valley is 676 mm and mean annual temperature is 13.8 °C (^{Santibañez and Uribe, 1993}).

Mexico. ^{Matus et al. (2011)} conducted a study in three representative micro-basins of the Southern region of Mexico in the state of Oaxaca, namely the Cuicateca, Mazateca, and Mixe microbasins, where a sustainable hillside management project was established (^{PMSL, 2004}). These soils are characteristic of steep sites (> 30% slope), Ferrasol (^{FAO-ISRIC-SICS, 1998}), where small farmers conduct agriculture. The sampling area is subjected to slash and burn agriculture following cultivation of maize and beans. Traditional farming (conventional cropping) in this area has low-inputs of nutrients. After 9 and occasionally up to 60 years of cultivation, farmers move into new areas where they clear secondary tree vegetation and establish new crop rotation. The mean annual precipitation in the Cuicateca micro-basin (2180 m) is 700 mm and mean annual temperature 19 °C. These conditions for the Mazateca (1689 m) are 2000 mm and 22 °C and for the Mixe (1422 m), 1500 mm and 22 °C.

Soil sampling

At each sampling site, three soil pits were excavated 100 m apart on a line transect. Soil samples consisted of ~2 kg moist soil from the Ap horizon and they were collected with a steel cylinder (0.05 m × 0.082 m diameter). Bulk density was also determined. In total, we took 39 soil samples in Chile and 9 samples in Mexico. The sampling areas and the scheme used in the present study followed the most common experimental design encompassing a single replicate per soil series when a large-scale experimental area is the only option and the knowledge is restricted to this particular area (^{Webster, 2007}).

All soils were mixed and homogenized and coarse organic matter (leaves, wood, and roots), removed by hand. All soils were air dried and sieved through a 2 mm mesh.

Physical fractionation

We followed the sonication and decantation procedure used by ^{Matus et al. (2011)} to obtain four particle size fractions [coarse sand (250-2000 μm), fine sand (53-250 μm), silt (2-53 μm) and clay (< 2 μm)]. All 48 soil samples, except Palquibudi, Piuchen and Hualañé, were physically fractionated in duplicate or triplicate. All samples were fully dispersed after the light fraction and floatable organic materials were removed (Matus et al., 2011). The fractionation procedure is as follows. About 50 g of moist soils were suspended in half-liter capped plastic bottles containing 10 glass beads (5 mm diameter) and 180 mL water. The suspension was shaken for 16 h (40 cycles min^-1) and sieved through a 250-µm sieve. The material remaining on the sieve contained sand particles and visible fragments of organic debris (plant and animal structures). The material retained on the sieve was washed several times with tap water, then combined with < 250 µm fractions (fine sand, silt and clay) and re-suspended in 100 mL of water, and ultrasonically treated with 910 J mL^-1 (power output 37.3 W). The ultrasonic machine (Fisher Sonic Dismembrator, Model 300) was equipped with a titanium probe with flat tip (18.8 mm diameter and 56 mm length) operating at 20 kHz. The clay and silt fractions were separated after corrected settling time by gravity sedimentation, applying Stoke’s law at room temperature until clear supernatant was obtained (15 decantation cycles). All fractions were oven-dried at 60 ºC and weighed. The physical fractions of silt and clay (< 50 µm) did not show any visual evidence of plant and animal fragments that could have been transferred during the sonication and sedimentation procedure as observed under electron scanning microscope (data not shown). On average, the amount of soil material recovered among all particle size fractions after dispersion and fractionation ranged between 93% and 111%. The slope of the regression analysis (e.g., R² = 0.76, P < 0.01) between the soil mass recovered for coarse sand, fine sand, silt and clay by physical fractionation and using pipette method showed no significant differences.

Carbon analysis

Chilean soils and their physical fractions were analyzed for organic C in duplicates using acid dichromate oxidation by the Walkley and Black method (^{Matus et al., 2009}). The Mexican soil samples for the C concentration in the bulk soil and physical fractions were determined using an automatic C analyzer (TOC, Shimadzu 5000 SSM). The C concentration (g C kg^-1 fraction) in each separate particle size was multiplied by the corresponding dry mass of the particle size to obtain the expression g C kg^-1 soil for each fraction. Several studies have addressed the uncertainties of SOC determination by wet or dry combustion. However, no significant differences have been found (e.g., ^{Arrouays et al., 2001}; ^{Matus et al., 2009}). Therefore, we assumed that the SOC determined by the TOC method in subtropical soils were comparable with those obtained by wet combustion in temperate soils, particularly when the correction factor is used (^{Matus et al., 2009}).

X-ray diffraction

Soil mineralogy of Mexican soils was obtained from ^{PMSL (2004)}, while no records were found for Chilean soils. Identification of clay (< 2 µm) minerals of Colchagua, Macarena, Nilahue, Peteroa and San Vicente soils were conducted by X-ray diffraction (Shimadzu 5A, diffractmeter) as exemplary soils to support early reports of similar genetic horizons sampled by ^{Besoain et al. (1984)} (see Table 1). Briefly, dry soil samples were prepared by removing the soil organic matter by strong oxidation with 30% H₂O₂ at 50 °C. Clay fraction (< 2 μm) was separated following the above indicated procedure and thereafter saturated with Mg (MgCl₂ 1M) to improve the intensity of the reflections at Cu-Kα radiation (40 kV and 20 mA).

Table 1 Average of soil characteristics of soil sampled at 0-0.1 m^†. 

^† ±: standard error of the mean of six samples per soil series, except HLE, PIU and PQB where only three samples were considered. ^‡ AGD = Agua del gato, CLH = Colchagua, CUC = Cunaco, HLE = Hualañé, MAO = Maipo, MPC = Mapocho, MAR = Macarena, NIL = Nilahue, PET = Peteroa, PIU = Piuchen, QHE = Quiahue, PQB = Palquibudi, SVT = San Vicente, CUI =Cuicateca, MAZ = Mazateca, Mix = Mixe. ^§ Latitude and Longitude. ^¶ Chilean soils classified by ^{Soil Survey Staff (1999)} and Mexican soils by (^{FAO-ISRIC-SICS, 1998}). ^# Soil organic carbon. ^?? Coarse sand = 250-2000 µm; Fine sand = 53-250 µm; Silt and clay = < 53 µm. ^‡‡ Clay mineralogy: Ch = Chlorite, K = Kaolinite, S = Smectite, V = Vermiculite, I=Illite, AS = Amorphous silicates (^{Besoain et al., 1984}).

Statistical analysis

A Gaussian distribution of all variables was tested using a skewness test value of 0.5 (^{Webster and Oliver, 2001}). The relationships between the C in the bulk soils and the C associated with fine particle-size fractions versus the silt and clay content of soils were examined by ordinary least square regression. The assumption of linearity and homoscedasticity of the predicted values was assessed. A paired t-test was used to compare the mean values. All regressions, t-test and ANOVAs were computed using JMP 3.2.2. (SAS Institute, Cary, NC, and U.S.A.) for P < 0.05.

Soil characteristics

The sampled temperate soils presented different clay mineralogy (2:1 and 1:1) which were compared with the result presented by ^{Besoain et al. (1984)} (Table 1).

The clay content of the soils ranged from 55±24 to 566±13 g kg^-1 soil in Chilean temperate soils, while these values in Mexican soils ranged from 150±6 to 320±6 g kg^-1 (Table 1). The average of clay mass for 2:1 in Chilean soils was 249±22 g kg^-1 and for 1:1 soils 174±42 g kg^-1, while the mixed clay in Mexican soils averaged 213±53 g kg^-1. In general, the coarse sand fraction was always lower than the fine sand fraction, with the exception of Mixe soil. For Chilean soils, coarse sand ranged 56 ± 11-197±10 g kg^-1 for 2:1 and 46±35-421±16 g kg^-1 for 1:1 soils; while in Mexican soils it ranged between 25±5 and 146±6 g kg^-1.

For Chilean soils, SOC ranged between 7.4±0.6 and 18±0.2 g kg^-1, and 1:1 showed the lowest average C content (9.6±0.7 g kg^-1) compared with 2:1 soil (13.8±0.4 g kg^-1). In general, the agricultural plots of Oaxaca under conventional cropping showed the greatest accumulation of C. The soil C content in cropped soils ranged from24±0.3 g kg^-1 to 120±0.08 g kg^-1. Soil samples taken from the Mixe micro-basin showed the highest C content, while the samples from Cuicateca plots, the lowest (Table 1). On average, SOC in Mexican soils was more than four times higher than that in Chilean temperate soils (Table 2).

Table 2 Average of soil properties by clay type (0-0.1 m). 

^† SOC = Soil organic carbon. ^‡ ± = Standard error of the mean.

Bulk density ranged from 0.61±0.04 to 1.44±0.01 kg m^-3, and the highest values were generally found in temperate soils with lower SOC.

Soil fractions

The C in the clay fraction of Mexican soils was significantly higher (nine times) than the SOC accumulated in the clay fraction of temperate soils. However, C of 2:1 clay was twofold (2 g C kg^-1 soil) than the amount in 1:1 clay soil. Similar results are shown for the silt fraction, although the silt content in both sites were comparable (Table 2).

X-ray diffraction results from chilean soils

X-ray diffraction analyses of clay fraction from five representative soils from Chile were consistent with those of similar modal top soils reported by ^{Besoain et al. (1984)} (Table 1) All soils showed evidence of dominant phases of smectite (2:1), vermiculite (2:1), illite (2.1), chlorite (2:1) and kaolinite (1:1) and some amorphous silicate as reported earlier by ^{Besoain et al. (1984)}. Results indicate that Peteroa and Nilahue soils had very clear reflections at 0.73 nm and 0.36 nm associated with the presence of kaolinite minerals. Both samples maintained the same d-spacing after ethylene-glycol solvation and heating at 490 °C supporting the presence of kaolinite as the dominant clay mineral. The diffraction patterns for San Vicente, Colchagua and Macarena showed the basal reflections associated with smectitic clay. These reflections were moved to a 1.02 nm associated with mica structure when soils were treated and heated. All clay treatments confirmed the presence of smectite as the dominant clay in San Vicente, Colchagua and Macarena soils and minor presence of illite in Colchagua that was not evident in San Vicente and Macarena soils. The clay types for Mexican soils were obtained from ^{PMSL (2004)} and are shown in Table 1.

Relationship of soil organic carbon in the bulk soil, silt and clay fractions

We found a positive and significant (P < 0.05) relationship between SOC and the C in the silt fraction in both, 2:1 and 1:1 top Chilean soils (Figure 1). The slope of the linear regression for the silt fraction ranged from 0.62 to 0.86, but there was no relationship for the clay fraction (Figure 1). The same relationship for subtropical soils from Mexico showed similar trends (Figure 2). There was a significant relationship between SOC and C in the silt fraction. Although the values were concentrated at the two extremes, the slope was 0.70, similar to those values found in temperate soils. The clay fraction of Mexican soils exhibited asymptotic behavior.

Figure 1 Relationship between soil organic C (SOC) and the C in the individual particles of silt and clay of temperate cropped soils with maize in Central Chile. * P < 0.05, ** P < 0.01. 

Figure 2 Relationship between soil organic C (SOC) and the C in the individual particles of silt and clay of subtropical cropped soils with maize in Oaxaca State, Mexico. * P < 0.05, ** P < 0.01. 

Soil fraction and C saturation

Based on the evidence that fine silt and clay fractions can become saturated with SOC, in the present study the approach of ^{Stewart et al. (2008)} was used to decide in which fractions C accumulation is influenced by saturation. A soil fraction exhibiting linear accumulation is interpreted as not being C saturated, while the opposite is true for asymptotic accumulation (^{Diekow et al., 2005}; ^{Stewart et al., 2008}). Although the Mexican soils were located at the extreme of the curve, the clay fraction reached a plateau along the SOC like temperate soils, while the soil C in the silt fraction increased proportionally, and this was interpreted as not influenced by C saturation. Our analyses are based on the fact that there is a hierarchical saturation level from primary to secondary soil structure (soil aggregates) (^{Six et al., 2002}; ^{Kool et al., 2007}). Therefore, we expected clay particles to be saturated before silt particles. In the present study, differences in C in the clay and silt fractions of 1:1 and 2:1 soils suggest that SOC is controlled by organo-mineral interactions because the specific surface area and adsorptive capacity are clearly different among soil minerals (^{von Lützow et al., 2006}; ^{Wagai et al., 2009}). For the subtropical soils examined here, ^{Matus et al. (2011)} found that silt-associated C, unlike clay-associated C, increased linearly with SOC. This is because the silt-sized class includes a partially reactive C fraction (^{Diekow et al., 2005}; ^{Virto et al., 2008}). The silt-sized particle class had a relatively high void volume, where the SOC is located and physically protected (^{Monreal and Kodama, 1997}). Thus, the silt fraction allows a linear C fraction accumulation in similar proportions across the varied soil SOC level.