Tiered maize and wheat nutrient removal coefficients estimated from available data

Summary

This meta-analytical study quantified variation in nutrient removal coefficients for maize and wheat by synthesising field experiment and on-farm data into a tiered estimation framework. Statistical and machine learning models achieved moderate predictive accuracy (R² 0.32–0.45) on replicated field trials but performed poorly on on-farm data (R² 0.08–0.36), highlighting substantial data limitations. The authors recommend using regional (Tier 2) coefficient estimates for national and global nutrient budget assessments until on-farm data quality and coverage improve.

UK applicability

The findings are relevant to UK arable farming as maize and wheat are significant UK crops; however, the study is global in scope and may require adaptation of regional Tier 2 coefficients specific to UK conditions. The poor predictive performance on on-farm data suggests that UK farm-level nutrient balance estimates would benefit from locally calibrated coefficients rather than reliance on global models.

Key measures

Harvest index (HI); nitrogen (N), phosphorus (P) and potassium (K) concentrations in crop grain and residues; model prediction accuracy (R²); Tier 1 (global mean), Tier 2 (regional), and Tier 3 (statistical/machine learning) coefficient estimates

Outcomes reported

The study assessed variation in nutrient removal coefficients (harvest index, nitrogen/phosphorus/potassium concentrations in grain and residues) for maize and wheat using three tiered estimation approaches. Mixed-effects and random forest models were developed to predict these coefficients globally, with accuracy varying substantially between replicated field experiment data and on-farm data.

Topic tags