Sampling and extraction methods for soil inorganic N determination to calibrate the EM38 in irrigated fields
MetadataShow full item record
Precise management of N variability in crop fields are required to increase yields and ensure sustainable and economic crop production, whilst not having a negative impact on the environment. A popular type of sensor for characterizing soil variability is the EM38-MK2 that measures apparent electrical conductivity (ECa), operating on the principle of electromagnetic induction (EMI). After analysis, inorganic N results can be calibrated to ECa measurements. It has been established that NH4+-N and NO3ˉ-N can be predicted from ECa. This study presented three main research aims to: i) compare single and composite samples for the determination of NH4ˉ-N, NO3ˉ-N, and total inorganic N (TIN), ii) determine if the saturated paste extract (SATe) could replace the standard 2.0 M KCl extraction for determination of NH4ˉ-N and NO3ˉ-N, and iii) determine what combination of single or composite ECa measurements and inorganic N at different sampling depths would produce the most statistically significant inorganic N prediction model. EMI surveys were conducted on four study sites under centre pivots, located on commercial irrigation farms in the districts of Douglas, Luckhoff, Hofmeyr and Empangeni. Using ECa data with the “Electrical Conductivity Sampling Assessment and Prediction” (ESAP) software and it‟s featured “Response Surface Sampling Design” (RSSD) sampling methodology, soil sampling points were identified based on the degree of ECa variability. Before sample collection, additional ECa readings were taken at each sampling point, one in the centre and one on each corner of a 1 m2 area. Afterwards soil samples were collected in the same manner in 300 mm depth increments up to 1500 mm. Samples collected in the centre were considered single, while those from the corners were composited. Concentrations of NH4+-N and NO3ˉ-N in KCl and SATe soil extracts were simultaneously determined colorimetrically. For the first aim, inorganic N concentrations in KCl extracts was loge transformed and pooled to compare sampling methods irrespective of study site, sampling point, and depth. For the second aim, data of inorganic N concentrations determined in KCl and SATe extracts were transformed and pooled for comparison, irrespective of site, sampling point, sampling method, and depth. The third aim was divided into three parts, determining agreement between single and composite ECa measurements, determining what inorganic N values to use, i.e. what sampling method and extract, and finally model calibration. Statistical analysis focused on assessing agreement using the Bland-Altman method for assessing agreement on a 95% confidence interval and multiple-linear regression calibration models were developed in Microsoft Excel. Results revealed poor agreement between single and composite samples for NO3ˉ-N and TIN. A composite sample taken in a 1 m2 area was more suitable when investigating NO3ˉ-N or TIN. Good agreement was found for NH4ˉ-N and a single sample proved sufficient. Agreement between the two extracts was poor for both NH4+-N and NO3ˉ-N and it was concluded that SATe cannot replace KCl for inorganic N determination. Agreement between single and composite ECa measurements was good and one ECa measurement was sufficient per 1 m2 sampling location. Based on the conclusions from the previous research questions, inorganic N results used for model development were those from composite samples extracted using KCl and an average between the single and composite ECa measurements was used. Values of inorganic N, ECa and elevation were loge transformed. Results showed that the majority of the calibration models were statistically insignificant except for one sampling depth (900 to 1200 mm) at Douglas (ECa 0 to 750 mm; R2=0.54 and ECa 0 to 1500 mm; R2=0.57). It was concluded that for the sites investigated, inorganic N was not the dominant soil property influencing ECa.