The variability of soil carbon and other selected soil properties as indicators of sustainable land management under tomato production at Mooketsi ZZ2 farm in Limpopo, South Africa

Abstract

The soil is the most important component in sustainable land management that varies spatially due to the combined effect of biological, physical, and chemical processes that occur over time. Although there has been extensive research on some of the soil characteristics and their effects on different crop yields, the interactive effect of various management practices and soil properties on carbon as well as the main factor or factors controlling soil C under short-term continuous tomato production are not yet fully understood. The objectives of the study were (i) to determine the spatial variation of soil carbon (C) and other selected soil properties within the tomato production field and (ii) to investigate the inter-relationship between soil C and the selected soil properties within the tomato production field at Mooketsi ZZ2 Farm. To achieve these objectives, a detailed soil survey was conducted, whereby a systematic soil sampling strategy was carried out where one sample was collected every 40 m using an auger at depth of 0-15 cm on a 23-hectare farm. The total number of samples collected were 132. A handheld Global Positioning System (GPS) was used to record the geographical coordinates, the latitude, and the longitude of where each sample was taken, this was used to create spatial variability maps. A handheld cone penetrometer was used to determine the penetrative resistance of the soil before soil samples were collected. The collected soil samples were analysed for physical and chemical soil properties such as particle size distribution, aggregate stability, soil organic carbon (SOC), soil pH, electrical conductivity (EC), and soil extractable phosphorus. The soil colour was also determined for the collected soil samples. The coefficient of variation (CV) showed that high variation exists in SOC with a CV of 38.72%, clay content with CV of 43.48%, silt content with CV of 50.70% and EC with CV of 59.60%. Semivariograms which are important for spatial analysis showed variation of soil properties within Mooketsi ZZ2 farm. Spatial dependency, which is the nugget/sill ratio, showed that extractable P had a weak spatial dependence with 1.00 nugget/sill ratio. Soil pH (KCl), EC, MWD, clay, silt and sand had moderate spatial dependence with the following nugget/sill ratios: 0.60; 0.44; 0.38; 0.48; 0.41 and 0.41 respectively. The SOC and PR both had 0 nugget/sill ratio which is a strong spatial dependence. The correlation results showed that SOC had weak correlation with the silt, sand and clay content having correlation coefficients of 0.30, -0.27 and 0.2, respectively. This means that texture does not influence the spatial variation of SOC across the tomato field. Mean weight diameter (MWD) was positively correlated with sand (r= 0.50) and negatively correlated with silt content (r = -0.46) and clay content (r= -0.51) showing that there was weak aggregation in the Glenrosa soil. The electrical conductivity had relatively weaker positive correlation with both clay content (r= 0.33) and silt content (r= 0.29) and it was negatively correlated with sand content (r= -0.32). The positive relationship between clay content and silt content with EC might be because; finer particles have more negatively charged sites that can hold onto the cations. The negative relationship between EC and sand content might be because; sandy soils tend to have low organic matter levels, which is important in binding soil particles. The low correlations between soil properties might be because, the Glenrosa soil has low clay content which means less surface area to hold cations and soil particles is available. This leads to poor soil structure and poor nutrient holding capacity of the soil. Overall, the results revealed that there was wide spatial variation within the soil properties of the study area. The RMSE values showed that kriging is reliable to characterize pH, MWD, SOC, P, PR, clay, silt, and sand with moderate to good accuracy, but it is less reliable when it comes to EC. From the inter-relationship results, it can be concluded that there is no soil property that has strong influence on SOC for the case considered. This indicates that none of these properties could serve as a proxy for predicting soil C or as parameters that can assist in soil C management options. The observed spatial variation could have an implication in the optimization of tomato yield in the study area. This bids for the adoption of site-specific soil nutrient management in the area in order to optimize tomato production because over and under fertilisation would be costly for the farm.