Spatial Prediction of Soil Microbial Enzyme Activities Using On-The-Go Vis–NIR Spectroscopy

Despite being key indicators of soil fertility and quality, microbial enzyme activities are rarely assessed spatially due to the high number of samples required and labour-intensive assays. This study aimed to evaluate the potential of on-the-go visible–near infrared (vis–NIR) spectroscopy to predict and map the spatial distribution of β-1,4-glucosidase (BG), acid phosphatase (ACP), alkaline phosphatase (ALP), and arylsulphatase (ARS) activities. An on-the-go (tractor-pulled sensor platform) vis–NIR spectroscopy sensor, coupled with partial least squares regression (PLSR), was used to estimate microbial enzyme activities in two fields, namely, Cayenne and Mortier, Belgium. Spatial maps were developed for both measured and predicted enzyme activities and analyzed using Local Moran's I and Bivariate Moran's I spatial statistics. Results showed strong correlations between total organic carbon (TOC), pH, and enzyme activities for samples from these two sites. A notable negative correlation existed between soil pH and TOC. PLSR model prediction accuracy varied by enzyme, with the highest for ARS (R² = 0.70, ratio of performance to interquartile distance [RPIQ] = 2.79), followed by BG (R² = 0.69, RPIQ = 2.93), ACP (R² = 0.64, RPIQ = 2.95), and ALP (R² = 0.60, RPIQ = 1.81). Spatial analysis demonstrated a strong agreement between measured and predicted enzyme activity maps, except for ARS in the Mortier field. Bivariate Moran's Iindicated positive spatial correlations between observed and predicted enzyme activities. In the Cayenne field, the highest Bivariate Moran's I was 0.64 for ACP, while in the Mortier field, the highest value was 0.58 for ALP. Overall, PLSR models performed better in the Cayenne field; hence, spatial predictions of enzyme activities were generally reliable, except for ARS in the Mortier field. These findings demonstrate that on-the-go line vis–NIR spectroscopy can provide reliable, high-resolution maps of soil microbial activities, offering a practical tool for guiding precision fertilizer recommendations.