Detection of Calcium Concentration Variation of Wheat Genotypes Grown on Sodic Soils Using Red Edge Spectral Variables.

Poorly structured sodic soils constrain crop growth due to nutrient deficiencies and toxicities. Calcium (Ca), magnesium (Mg) and potassium (K) deficiencies, in particular, can be a problem for crops grown in sodic soils. Concentrations of these nutrients in plant tissue could potentially provide a signal to identify crop genotypes tolerant of sodic conditions. This research was undertaken on 18 different wheat genotypes grown on a sodic Vertosol (Exchangeable Sodium Percent (ESP)=2.2 in surface and 20 in subsoil) at Billa Billa, Southern Queensland, Australia in 2018. Results demonstrate the possibility that the red edge (RE) spectral band (680-750 nm) may be used to detect variations in plant Ca concentration among wheat genotypes. Leaf reflectance data in the range 400-900 nm were collected by using a handheld spectroradiometer Field Spec® (Analytical Spectral Devices Inc., Colorado, USA) at the pre-anthesis stage. At the same stage, plant Ca, Mg, K concentration (%) was measured by destructive sampling of young mature leaves (YML) at plot level followed by laboratory analysis. The plant Ca concentration was significantly positively correlated with Mg (R²=0.58, RMSE=0.015, and p < 0.001) and K (R²=0.32, RMSE=0.28, and p=0.017). Normalized Difference Red Edge Index (NDRE) was derived from the RE and near infrared (NIR) wavelengths. To determine Ca, Mg, K concentration among the wheat genotypes, Red edge inflection points (REIP) of 18 genotypes were computed by using a four-point linear interpolation method. REIP showed a significant shift towards longer wavelengths as an indicator for healthy genotypes. The first derivative of reflectance spectra showed significant variations between genotypes, which demonstrates crop structural variations, indicating deviations in Ca, Mg, and K level. REIP was derived by following the first derivative of reflectance and was observed from 724-725 nm, which divides the entire RE area under 680-750 nm into two portions. The area under spectral curve from 724-750 nm was less than the area under the curve from 680-724 nm. This indicates that there is a shift in reflectance towards longer wavelengths for structurally healthier genotypes with higher chlorophyll content. Examination of the red edge symmetry (RES) to determine plant chlorophyll content demonstrated that RES is a strong indicator of plant chlorophyll content (R²=0.66, p < 0.0001). Leaf chlorophyll index (LCI) was derived from RE spectra for 18 wheat genotypes and also had a significant correlation with Ca concentration (R²=0.85, p < 0.0001). A regression model followed by Principal Component Analysis (PCA) was performed between Red edge variables (NDRE and REIP) and plant Ca, Mg, and K concentrations. This analysis indicated that REIP as a red edge variable is a useful indicator of plant nutrient concentrations. For example, correlation with REIP for Ca (R²=0.82, RMSE=0.01), Mg (R²=0.52, RMSE=0.02), and K (R²=0.40, RMSE=0.28) were closer than those for NDRE, Ca (R²=0.79, RMSE=0.01), Mg (R²=0.48, RMSE=0.02), and K (R²=0.29, RMSE=0.31). The results were validated using 25% of the field measured samples after 75% were used in model calibration. In stepwise regression analysis, only Ca concentration was significantly correlated with both NDRE and REIP (p < 0.001). The results revealed that REIP, NDRE, and RES can provide significant spectral cues to understand crop nutrient variations, especially Ca under sodic soil constraints. This result suggests that remote sensing has potential to select for genotypes tolerant to soil constraints.