Synergistic effects of arbuscular mycorrhizal fungi and nitrogen on tea root architecture using machine learning predictions

• A reliable root phenotype prediction model based on machine learning was established. • The interaction mechanism between AMF and nitrogen was quantitatively evaluated. • This study deepens the understanding of the ecological characteristics of tea roots. • The integration of experimental data with machine learning offers an innovative analytical framework for investigating biological systems. Root system architecture (RSA) governs plant nutrient acquisition and environmental adaptability, yet the synergistic effects of arbuscular mycorrhizal fungi (AMF) and multi-gradient nitrogen on tea ( Camellia sinensis cv. ‘Fuding Dabaicha’) RSA remain underexplored, hindering the optimization of nutrient uptake efficiency and root adaptability for sustainable cultivation. Previous studies on tea have focused on phosphorus or single nitrogen levels, leaving gaps in understanding AMF-nitrogen interactions and the application of machine learning (ML) for predictive root phenotyping. Here, we conducted a sand culture experiment to investigate AMF inoculation effects on ‘Fuding Dabaicha’ seedlings under four nitrogen levels (0, 0.1, 1, 10 mM (NH 4 ) 2 SO 4 ). Root parameters were quantified using WinRHIZO software, and four ML algorithms, multilayer perceptron, random forest, XGBoost, and Tabular Probabilistic Formal Network, were applied for multi-output regression of total root length, projected area, surface area, average diameter, and volume. AMF colonization increased from 15.62 % at 0 mM to 30.39 % at 1 mM, slowing to 32.50 % at 10 mM; extraradical hyphal length followed similar trends (2.45–6.12 m/g). AMF enhanced root length (up to 43.75 %), projected area (up to 48.74 %), surface area (up to 48.73 %), and volume (up to 71.31 %) at 1 mM, with no effect on diameter. ANOVA and MANOVA confirmed additive effects, with 1 mM + AMF yielding peak values ( P < 0.05). Random forest and XGBoost achieved high predictive accuracy (R 2 > 0.76), except for diameter (R 2 ≈ 0.08). These findings demonstrate that AMF and moderate nitrogen optimize tea RSA for nutrient foraging, while high nitrogen limits benefits. This ML framework supports precision agriculture by reducing fertilizer inputs and advances root phenotyping for tea and other woody crops.