Search

Soybean [Glycine max (L.) Merr.] yield in claypan soils varies systematically with soil properties and landscape position. This is likely caused by soil interactions with soybean roots. Field observations of soybean root distribution are needed to reveal its effect on yield variability. This study examined profile distributions of soybean root length density (RLD) and average root diameter (ARD) as a function of landscape position, depth to claypan (DTC), apparent soil electrical conductivity ([EC.sub.a]), clay-maximum translated depth ([D.sub.t]), and other soil properties. A landscape of claypan soils was sampled postharvest at two sites near Centralia, MO, in 2001. Roots were washed from soil cores in 15-cm layers (15-120 cm) and measured with image analysis. Root length density and ARD were significantly related to landscape position, DTC, [D.sub.t], and [EC.sub.a]. Predictions of RLD and ARD were best from 15 to 60 cm, the depths with the greatest influence from claypan soil morphology. Soil profile distributions of base cations, P, and pH matched root density profiles. Soybean roots were inhibited in E horizons above the claypan and stimulated 20 to 40 cm below it. Soybean roots below the claypan had about 20 to 30% smaller diameter. We conclude that DTC and rapid estimators of claypan morphology, such as [EC.sub.a], can be used to predict soybean root distribution in claypan soil landscapes.

Nuclear magnetic resonance (NMR) relaxation time (T2) distributions are well known to be linked to the pore size distribution and show promise as a method of estimating soil texture. As traditional laboratory methods used for soil texture estimates in soil science are generally time consuming, in this study, we explore an alternative approach based on NMR T2 distributions to estimate the soil texture of water‐saturated soil samples collected from three field sites. Using two T2 cut‐off times, T2a and T2b, the T2 distribution of a soil was partitioned into three regions, short, intermediate and long relaxation times, each of which represents the fraction of clay, silt and sand, respectively. Two approaches for determining the cut‐off times were used: the first used T2 cut‐off times determined from the data from all sites and the second used site‐specific T2 cut‐off times. The NMR estimates of soil texture were compared to measurements of soil texture made using the sieve–pipette method and laser diffraction particle size analysis (LDPSA). The results show that there is no universal cut‐off time for estimating the clay, silt and sand fractions based on the NMR T2 distributions. The accuracy of NMR measurements to estimate the soil texture depends on the magnetic susceptibility of the measured material. For soils with low magnetic susceptibility (<2 × 10−4 SI) using site‐specific cut‐off times, the NMR‐derived soil texture (root mean squared error [RMSE] = 9.43%) more closely matches the soil texture measured from the sieve–pipette method than the soil texture determined using LDPSA (RMSE = 11.88%). However, the NMR estimate of soil texture breaks down for soils with high magnetic susceptibility (>4 × 10−4 SI). These results suggest that the NMR method can provide reasonable estimates of the soil texture for soils with low magnetic susceptibility. Highlights: No universal cut‐off time exists for estimating the clay, silt and sand fraction based on the NMR T2 distributions.NMR can provide better estimates of clay, silt and sand fractions for soils with low magnetic susceptibility than LDPSA.Site‐specific cut‐off times are determined through comparison between NMR data and sieve–pipette measurements.NMR yields reasonable estimates of the soil texture for soils with low magnetic susceptibility. [ABSTRACT FROM AUTHOR]

Determination of in-season N requirement for corn (Zea mays L.) is challenging due to interactions of genotype, environment, and management. Machine learning (ML), with its predictive power to tackle complex systems, may solve this barrier in the development of locally based N recommendations. The objective of this study was to explore application of ML methodologies to predict economic optimum nitrogen rate (EONR) for corn using data from 47 experiments across the US Corn Belt. Two features, a water table adjusted available water capacity (AWCwt) and a ratio of in-season rainfall to AWCwt (RAWCwt), were created to capture the impact of soil hydrology on N dynamics. Four ML models--linear regression (LR), ridge regression (RR), least absolute shrinkage and selection operator (LASSO) regression, and gradient boost regression trees (GBRT)--were assessed and validated using "leave-one-location-out" (LOLO) and "leave-one-year-out" (LOYO) approaches. Generally, RR outperformed other models in predicting both at planting and split EONR times. Among the 47 tested sites, for 33 sites the predicted split EONR using RR fell within the 95% confidence interval, suggesting the chance of using the RR model to make an acceptable prediction of split EONR is ~70%. When RR was used to test split EONR prediction with input weather features surrogated with 10 yr of historical weather data, the model demonstrated robustness (MAE, 33.6 kg ha-1; R² = 0.46). Incorporating mechanistically derived hydrological features significantly enhanced the ability of the ML procedures to model EONR. Improvement in estimating in-season soil hydrological status seems essential for success in modeling N demand. [ABSTRACT FROM AUTHOR]