3 Results
In the studied region, the average soybean yield in the 2015-2017 three-year period was 4.13 Mg ha-1 at 65 plots. The minimal and maximal values of soybean yield were 3.23 Mg ha-1 and 4.97 Mg ha-1, respectively. The box plots of the SOM, FLF, OLF and HF properties contain only the data for the 0.0-0.10-m (L1) layer. The presented data in figure 1 for the other variables are at L1 (0.0-0.10-m), L2 (0.10-0.20-m) and L3 (0.20-0.40-m). The classification of the 65 studied plots revealed that 50, eight, five and two plots formed the textural classes, including clay, sandy clay loam, sandy clay and sandy loam, respectively. Although, the clay exhibited an incremental function with increasing depth, this increment was not sufficient to constitute a textural gradient, and the soils in all the tested areas were Oxisols (Latosols;Fig. 1).
The relationships between the soil physical properties at each layer (L1, L2 and L3) and the soybean yield indicated that seven significant response variables explained the soybean yield. Specifically, the PR, BD, n parameter, I, TP, S-index and macro-porosity presented significant relationships (Table 1). Of these properties, positive and negative relationships were obtained with the S index, macro-porosity, TP and n parameter and with PR, I and BD, respectively (Table 1).
The analysis of the relationships of the SOM,FLF, OLF and HF with the clay, sand, PAW, S and PR properties showed significant values between the clay and sand contents and all forms of SOM studied. Among the properties affected by the soil management (i.e., PAW, S and PR), the significant relationships were only detected between OLF and PR, and between FLF and S index (Table 2).
The RP and BD values obtained at L1 showed a significant negative relationship with the soybean yield. In contrast, PD did not show significant relationship with productivity at L1 the most relevant among the three studied layers, with p=0.32 (Fig. 2).
The α parameter of the Van Genuchten equation also showed non-significant relationship with the soybean yield. However, the n parameter and the inflection point (I) of the SWRC showed positive and negative significant associations with the soybean yield, respectively (Fig. 3).
The total porosity, macro-porosity and S-index properties at L1 exhibited a significant positive relationship with the soybean yield (Fig. 4). The AWC, water content at saturation (θs) and residual water content (θr) showed no significant relationship with the soybean yield in any of the studied layers (Fig. 5). The AWC and θs obtained at L3 had a stronger relationship with productivity, with p=0.53 and p=0.56, respectively, than those obtained for the other layers, whereas the relationship with the θr value obtained for L2 was stronger, with p = 0.60, compared with those found for the other layers.
The clay and sand contents did not show a significant relationship with the soybean yield. However, L1 was found to be the most relevant for clay content, with p=0.16, and L2 was the most important for the sand content, with p=0.22 (Fig. 6). The field capacity that describes the soil water content for the potential matrices obtained at the SWRC inflection point was also not significant at all layers, although a stronger relationship was found at L3, with p=0.52, than in the other layers.
The clay content of the soil exhibited significant positive relationships with the SOM,FLF, OLF, and HF, whereas significant negative relationships of SOM and its fraction (FLF, OLF and HF) were also found with sand contents. The SOM did not present a significant association with the PAW (p=0.46), PR (p=0.19) and S (p=0.20). The FLF was also not significantly related to the PAW (p=0.66), PR (p=0.76) and S index (p=0.14). Whereas, the OLF presented a significant positive relation with PR, but was not significantly related to the PAW (p=0.62) and S (p=0.57). The HF did not show a significant relationship with the PAW (p=57%), PR (p=89%) or S index (p = 71%;Fig. 7).