MASTHING: a process-based model of mast seeding in European beech

Simone Bregaglio , Sofia Bajocco, Carlotta Ferrara, Michał Bogdziewicz, Andrew Hacket-Pain, Francesco Chianucci

Masting, the synchronised interannual variation in seed production, shapes forest regeneration and many ecosystem processes, yet process-based models remain underdeveloped. Here we introduce MASTHING (MASting THeory modellING), an individual-tree model coupling phenology, carbon gain, resource storage, temperature cues, and environmental vetoes on reproduction. We parameterised MASTHING for European beech (Fagus sylvatica L.) using 43 years of data from 97 trees across 11 sites in England, testing three variants: i) resource budget only, and extensions with ii) additive and iii) interactive temperature cue effects. The results reveal that the critical variable is not whether weather cues are incorporated, but how they are coupled to internal resource dynamics: the additive formulation (RB+WC) reproduced the observed interannual variability but degraded temporal performance relative to RB alone, whereas the interactive formulation (RB×WC) in which cue sensitivity is gated by resource status improved overall performance and classification skill (R2 = 0.82). RB×WC provided the lowest forecast error at country calibration level (RMSE = 0.27), demonstrating structural robustness when parameterisation cannot be tailored to local conditions, while simpler formulations remained competitive for point-forecast accuracy at finer calibration scales. The model reproduced the long-term decline in masting intensity under climate warming (R2 = 0.81). For four years of data excluded from model calibration, the model correctly identified the 2024 low-production event across multiple sites, though the 2025 crop was systematically overestimated, pointing to insufficient carry-over depletion after consecutive poor seed years as the main structural limitation. MASTHING provides a platform for testing masting hypotheses, evaluating climate change impacts, and supporting operational seed forecasting.