Cropping system is a key determinant of soil health after accounting for environmental and edaphic variability

• Propensity score weighting revealed actual impacts of cropping system on soil health. • Grazed pastures had higher organic matter and carbon indicators than row crops. • Soil pH and PSR were elevated in row crop systems. • Cropping system consistently affected carbon-based indicators. • The relative importance of environmental and edaphic variables was indicator-specific. Soil health assessment is critical for understanding and promoting sustainable soil management practices. Soil health assessment methods incorporate a variety of inherent environmental and edaphic properties, including climate variables and texture, as well as cropping system. However, these inherent properties often vary systematically among cropping systems and potentially confound the effects of cropping system on soil health. Using a new soil health dataset in Florida, we adopted propensity score weighting to model each sample’s probability of belonging to a particular cropping system based on soil drainage class, clay content, mean annual temperature (MAT), and mean annual precipitation (MAP). By balancing these factors across systems, this method enabled unbiased estimation of cropping system effects. The effects of cropping system on soil health indicators remained statistically meaningful after propensity score weighting using the same significance threshold ( α = 0.05). Soil organic matter and carbon-related indicators were consistently higher in grazed pastures than in row crop systems, indicating that differences in disturbance and management intensity across cropping systems regulated soil health. Similarly, soil pH and phosphorus saturation ratio (PSR) were higher in row crop systems than in grazed pastures, which likely reflected the differences in chemical inputs among these systems. In contrast, our results showed that before propensity score weighting, Mehlich-3 extractable nutrients were higher in row crops than in hayfields and grazed pastures; however, these effects became non-significant after weighting, suggesting that they were partially driven by differences in inherent properties among cropping systems. Variance partition analysis further confirmed cropping system as a consistent driver of the carbon-based indicators, while the impacts of inherent properties were indicator-specific. Overall, our results demonstrate the importance of considering variability in inherent properties when evaluating cropping system impacts on soil health. Propensity score weighting offers a robust approach for addressing complex interactions among soil health drivers.