Study site

The lama-bordo system under study is located in the municipality of Teotongo in the High Mixtec region, Oaxaca, on coordinates 17° 45’ 45” N and 97° 31’ 41” W, at an altitude of 2115 m. The site’s climate is sub-humid temperate with summer rains, with a mean annual temperature of 16.6 °C; the mean annual precipitation is 492 mm and the evaporation is 2799 mm. It is a semiarid zone with restrictions for the production of rainfed crops due to the seasonal variation of rain, which in dry years is insufficient to cover the demands from the crops. The dominant soil group is epileptic Phaeozem, which are shallow, subject to erosion, parental material emergence, and low moisture retention capacity. With the construction of the bordos, the eroded materials accumulate, forming deep colluvial soils in the central part and shallow on the right and left margins.

The system has 11 plots (terraces) with surfaces that vary from 0.22 to 1.32 ha on a total surface of 8.47 ha (Figure 1); they were grouped by age into: young (P1, P2 and P3), middle (P4, P5, P6 and P7) and old (P8, P9, P10 and P11). Those in the low part are the oldest, have higher bordos and show lateral growth from the incorporation of new fractions of land in both margins of the hillsides.

Figure 1 Plots of the lama-bordo system and sampling sites in Teotongo, Oaxaca. 

Soil sampling and laboratory analysis

A preliminary sampling of the soils was carried out, at 0.30 m of depth, to understand the efficiency of the lama-bordo system in improving the soil properties, in relation to lands with traditional tilling and at rest.

For spatial variability, the sampling was carried out at 0.30 m of depth on the right, left and central margins of each plot in the system; the sampling points were located with a Global Positioning System (GPS). The soil samples were placed in plastic bags for their transport and processing in the laboratory; the methods used in the analysis of the properties are shown in Table 1.

Digital elevation model

A total station topographic appraisal was carried out (Sokkia SET 630 R) of the bordos of each terrace; the readings were taken every 10 meters on the high part and low part of each bordo to obtain the incline between plots and to estimate the height of the terrace. The Digital Elevation Model (DEM) of the system was elaborated with the data from this appraisal.

Data analysis

To analyze the spatial behavior of the physical and chemical parameters of the soils, the geographic location of the sampling points was used, the DEM and the specific values for each parameter, and they were interpolated with the Kygring tool in Arc GIS 10.2.2.

The statistical descriptors of each property were obtained in the statistical analysis, and they were grouped by position of sampling and age; the analysis of variance (ANOVA) and means comparison with the Tukey test were carried out with the Minitab 17 software.

Variability of the physical properties of the lama-bordo system

Results show high correlation coefficients between terrace age and silt contents (0.7350, p<0.0001), organic matter (0.5220, p=0.0026), CE (0.41652, p=0.0198) and hydraulic conductivity (0.40964, p=0.0221). The oldest lama-bordo terraces have had more time to retain nutrients and aggregation processes have begun which influence the hydraulic conductivity. The percentage of K and DAP have a negative correlation (0.5320 and 0.5128); the highest values of these properties are found in the young plots. The reduction of the potassium content can be due to a greater accumulation of calcium in the exchange complex.

The position shows high correlation coefficients with the clay content (p=0.0002), FC (p=0.0003), PWP (p=0.0005) and SOC (p=0.0012). The center of the plots continues conserving the water flow line and despite the bordos there is not enough time for the sedimentation of clays; this behavior affects the system’s capacity for moisture retention.

The average texture of the soils in the lama-bordo system is clay-sandy loam, with clay, silt and sand contents of 32.65 %, 18.32 % and 49.03 %, respectively, and variation coefficients (VC) of 29.14 %, 31.27 % and 15.44 %; the soil fraction composition suggests that when sand contents dominate, better conditions for moisture storage in the profile are generated, and water infiltration into the soil is favored. The field capacity (FC) varies between 20 and 41 %, with an average of 31.7 % and VC of 16.75; the permanent wilting point (PWP) varies between 5 and 30 %, with average values of 20.1 and VC of 25.7 %. The hydraulic conductivity has average values of 4.72 mm·h^-1, with a negative correlation with the silt content. The behavior of the hydraulic properties show spatial variability in the capacity of moisture storage in the soil, which can cause a differential growth of the crops and compel to generating management strategies to maintain a more uniform growth. The apparent density has relatively high values, 1.47 Mg m^-3 characteristic of sedimentation materials with scarce influence of soil-formation processes and subject to compacting from the weight of the materials that are continually added (Table 3 and Figure 2).

Figure 2 Spatial variability of the physical properties of the soil in the lama-bordo site. 

The ANOVA showed highly significant differences (α=0.05) for the margins in the contents of clay, sand, VC and PWP (Table 3), suggesting that with the system management and the soil clearance, new materials are incorporated which modify the textural behavior because of the lateral growth of the terraces. The margins change their soil fraction composition and the finer fractions tend to accumulate in the central part; this agrees with what was reported by ^{Sullivan (2000)} for runoff agriculture systems in southwestern United States, where they report that there are changes in the textural patterns of the plots with runoff management. For silts and apparent density there are statistical differences by age, which are variables related to the contribution of materials without weathering that become incorporated into the system’s margins, where coarse fractions predominate; finer fractions of the transported materials tend to accumulate in the central part. For silts and apparent density, there are statistical differences because of the position of the plot (age) (Table 4) which may be explained because it is the easiest fraction to transport by the superficial flow; the hydraulic conductivity reported significant differences and a high coefficient of variation between the right margin and the other positions of the sample. This agrees with what was reported by ^{Kreznor et al. (1989)} and ^{Pachepsky et al. (2001)}, who point out that the geographic formations and the topography are factors highly correlated with the spatial variability of the soil properties.

The highest values of apparent density are related to the incorporation of new materials to the terrace soil; generally it is tuff.

The means test showed significant differences (α=0.05) for the clay, silt, sand contents, apparent density, field capacity, permanent wilting point, and hydraulic conductivity. In the percentage of clay there are significant differences due to position (Table 3); in the center and left margin of the plot there is higher concentration of fine particles than in the right margin, indicating the water flow lines within the plots and the incorporation of coarser materials from the hillsides to the right due to management. The percentage of silt has difference due to age (Table 4); in the plots of the lower part (9, 10 and 11) there is higher percentage in comparison to those located in the high part (plots 1, 2, 3 and 4). The percentage of sand showed difference by margin of the plot; the highest percentage of sand was reported in both margins (right and left) because of the presence of degraded zones and in process of repair with tilling, and a lower proportion was present in the central part of the plot. The apparent density had a difference due to age; the plots from the high part present higher values, associated to the presence of sand materials; the FC and PWP had significant differences in the plot’s margins and there was no difference between the center and left margin. The hydraulic conductivity had significant differences between margins, since the right margin is different from the left and center.

Variability of the chemical properties of the lama-bordo system

The pH presented a low VC (3.65 %) and the rest of the properties (OM; N, P, K and SOC) have high coefficients of variation, which shows the spatial variability of these properties in the lama-bordo system (Table 5 and Figure 3).

Figure 3 Spatial variability of the chemical properties of the soil in the lama-bordo site in Teotongo. 

Figure 3 shows the geospatial variability of the chemical properties of the soil; it is observed that the organic matter contents are higher in the low part of the lama-bordo system, attributable to the age of the terrace construction and to the presence of the soil formation process of aggregation of particles and improvement of the structure, which shows the benignity of the system in the carbon capture that moves with superficial runoffs.

The variance analysis (ANOVA) indicates that the spatial variability of chemical properties of the soil is due to the position of the sampling, to the plot or the position of the plot in the system; significant differences were found in the contents of organic matter, potassium and organic carbon (Table 5).

The system retains in average 72 Mg ha^-1 of soil organic carbon (SOC), which tends to accumulate in a higher proportion in the central part of the terraces with average values of 132 Mg ha^-1 (Table 5). The contents of total nitrogen and phosphorus do not report significant differences despite showing high variation coefficients in their spatial distribution in the system.

The means comparison test shows significant differences in the contents of organic matter due to the position of the plot (age of the lama-bordo); the young plots (9, 10 and 11) report the highest contents of organic matter (92 Mg ha^-1), which is 60o% higher in comparison to the values found in the old plots (1, 2 and 3) (Table 6). As was mentioned previously, soil conservation works such as the lama-bordo system capture the sediments transported by runoffs, foster infiltration, store more water in the terraces, and retain the organic matter transported by the runoff, improving the quality of soils in the sedimentation zones as reported by ^{Mueller et al. (2012)} and ^{Leigh et al. (2013)}.

The potassium contents had significant differences between plots; number 1 is within the high class, compared to number 11, which is within the low class, and total organic carbon showed significant differences in terms of the margin of the plot; the center is different from the right and left margins.

The lama-bordo system captures moisture, retains soils, shapes cultivation areas, captures carbon and reduces risks in the production of crops from the presence of droughts; however, there is abandonment of this production system that translates into the destruction of the bordos and the dragging of material previously retained. The rescue of this Pre-Hispanic technology is a responsibility of the owners of plots with lama-bordo, and the government could support them as a strategy for soil conservation and for production of the grains that the rural population demands from this marginalized zone in the country.