Abstract :
[en] Temperate permanent grasslands play a central role in European livestock systems, providing forage and regulating nutrient and water cycles. However, under climate change, these systems are increasingly exposed to rising temperatures, altered precipitation regimes, and intensifying climatic extremes. In regions such as Wallonia in southern Belgium, the vulnerability of grasslands is determined by the interaction between plant functional strategies, soil constraints, management regimes, and seasonal shifts in biomass production, not just by mean warming trends. Understanding these coupled processes is essential for anticipating forage reliability and supporting adaptive management.
Despite extensive research on grassland modeling, several gaps persist. First, while process-based models are well established for highly productive temperate grasses, fewer studies have explicitly evaluated contrasting functional strategies within a unified modeling framework. This limits systematic comparison across heterogeneous pedoclimatic contexts. Second, soil water and nitrogen processes are generally represented, but their interaction with dominant plant functional types under variable soil textures and water-holding capacities is insufficiently examined. Third, grazing is often represented as a single, aggregated biomass removal flux, which ignores the spatial specificity and behavior-driven nature of defoliation. Consequently, few modeling frameworks jointly represent functional contrasts, soil-mediated constraints, climatic variability, and grazing-induced heterogeneity within a coherent, management-oriented perspective.
This thesis aims to develop and formalize a process-based modeling framework to analyze the functioning of temperate permanent grasslands under climate change across contrasting pedoclimatic and management conditions in Wallonia. The research follows a progressive trajectory, combining grassland modelling, climate impact assessment, grazing experimentation, and agent-based modelling. First, a process-based grassland growth model was developed and evaluated under cutting management. This model explicitly represents soil water and nitrogen balances, distinguishing contrasting plant functional types that reflect acquisitive and conservative growth strategies. Evaluating the model across heterogeneous pedoclimatic contexts demonstrated its ability to reproduce seasonal growth trajectories and interannual variability patterns, supporting its structural robustness within defined environmental boundaries. Next, the modeling framework was applied to analyze productivity-stability trade-offs under multiple climate scenarios. The results show a systematic seasonal redistribution of biomass production, characterized by enhanced spring growth and recurrent summer declines, which are driven by intensified drought and heat extremes. Conservative functional strategies tend to sustain higher productivity under stress, whereas acquisitive strategies display greater interannual stability. These contrasting responses generate an emergent productivity-stability trade-off, the magnitude of which depends strongly on soil water-holding capacity, altitude, and exposure to extreme events. Thus, vulnerability emerges primarily through the erosion of temporal buffering capacity rather than uniform declines in mean annual yield.
To complement the modeling analyses, an experimental grazing study was conducted to investigate how grazing intensity shapes sward structural heterogeneity, regrowth dynamics, and animal performance. The results suggest that moderate grazing intensity does not necessarily homogenize pasture structure and can increase live weight gain per hectare without compromising individual performance under non-severe water stress. Structural heterogeneity arises from plant-animal feedback, including selective grazing, localized depletion, avoidance behavior, and compensatory regrowth. Based on these findings, a spatially explicit, agent-based grazing model was developed. By coupling bite-scale intake constraints with movement decisions in heterogeneous swards, the model reproduces realistic post-grazing sward height gradients and intake trajectories without prescribing spatial defoliation patterns. Thus, grazing-induced heterogeneity emerges mechanistically from explicit plant-animal-space interactions rather than from imposed utilization rules.
In a separate modeling component, the grass growth model, applied under cutting management to future climate trajectories, reveals significant regional differentiation within Wallonia. Upland areas, such as the High Ardennes, show comparatively buffered stability responses. In contrast, clay-limestone and sandy lowlands (e.g., the Jurassic and Hennuyere Campine regions) exhibit amplified summer vulnerability. Increases in consecutive dry days and extreme temperatures play a dominant role in destabilizing seasonal production. These results suggest that grassland vulnerability is more influenced by pedoclimatic configuration than administrative boundaries.
Beyond empirical findings, this thesis clarifies the epistemic position of process-based modeling in grassland research. Rather than universal predictive accuracy, validation is framed as contextual robustness-structural consistency across functional types and pedoclimatic gradients. The framework enables strong mechanistic and comparative inferences regarding relative contrasts, seasonal redistribution, and emergent processes. However, absolute long-term quantitative projections and community compositional shifts depend on structural assumptions and scenario choices.
This work demonstrates that integrating plant functional strategies and soil processes into a unified, process-based modeling framework enhances our understanding of the vulnerability of temperate grasslands to climate change. By explicitly representing the interactions between functional traits, soil-mediated constraints, and climatic forcing, the framework allows for a mechanistic exploration of the trade-offs between productivity and stability across different pedoclimatic contexts. In parallel, developing a spatially explicit, grazing module extends the framework to include a more detailed representation of plant-animal interactions at the event scale, laying the methodological groundwork for future multi-cycle integrations. Transforming climate uncertainty into a structured comparison of vulnerability pathways supports risk-oriented analyses that emphasize stability, seasonal flexibility, and context-specific adaptation rather than maximizing yield alone. Thus, the framework provides a coherent analytical basis for comparing management configurations and identifying those that may remain comparatively robust across plausible future climatic trajectories.
