Leveraging mixed-effects models to interpret longitudinal soil carbon dynamics

Agriculture significantly contributes to greenhouse gases emissions, while soils play a key role in climate change mitigation by storing atmospheric carbon as soil organic carbon (SOC). However there is still a gap in modelling studies describing SOC dynamics in tropical and subtropical environments. Linear mixed-effects models provide a parsimonious framework to analyze SOC stocks and decomposition processes, yet their performance in this context remain unclear. This study investigates SOC dynamics and persistence in three long-term conservation agriculture soils from Brazil’s diverse climatic zones: subtropical (Ponta Grossa), transitional (Londrina), and tropical Cerrado (Lucas do Rio Verde), considering three soil layers (0–20, 20–40, and 40–100 cm). Through a 30-month incubation experiment, we measured CO₂ emissions and SOC stocks. We then applied mixed models to evaluate temporal changes in SOC dynamics, accounting for fixed effects of site, soil layer and incubation time, as well as random effects associated with repeated measurements within experimental units. A strong three-way interaction between between site, soil layer, and incubation time (p<0.0001) indicated that SOC dynamics rates are distinctively driven by the interplay of environment conditions and soil depth. Ponta Grossa showed the highest SOC stocks (59 Mg ha^-1). A significant decrease was observed in the 20–40 cm layer (−19.67 Mg ha⁻¹) relative to 0–20 cm, whereas 40–100 cm exhibited the highest adjusted SOC stocks values. The model successfully captured the high temporal dependency of the data (=0.99), providing insights into the long-term dynamics and decomposability of SOC under controlled conditions.