Spatial Prediction of Soil Antibiotics Based on High-Accuracy Surface Modeling

Abstract

The spatial prediction of soil antibiotic is more difficult than other normal soil properties due to the diverse sources of soil antibiotics. Few studies have attempted to predict soil antibiotic residues in intensive vegetable cultivation areas. High-accuracy surface modeling (HASM) is regarded as an important new technique in the pedometrics and digital soil mapping fields. A total of 100 surface soil samples were collected from the north-central part of the Shandong Province of China. The antibiotic concentrations, including ciprofloxacin (CF), enrofloxacin (EF), norfloxacin (NF), and fluoroquinolones (FQs), were analyzed using high-performance liquid chromatography–tandem mass spectrometry. We employed splines to compare its performance with that of HASM method. The errors of HASM for NF, CF, EF, and FQ were less compared to splines. HASM has less mean absolute error (MAE) and root mean square error (RMSE) than splines. The RMSEs of splines for FQ, CF, EF, and NF were 3.02, 2.34, 3.46, and 2.64 times lager than those of HASM, respectively. Therefore, HASM can be considered as an alternative and accurate method for interpolating soil antibiotics. It can also make the map more consistent with the true spatial distributions.