Disentangling temporal vegetation dynamics through trait-based trajectory analysis

Long-term vegetation resurveys provide key opportunities to investigate community assembly processes. Yet, the metrics most commonly used to quantify temporal change capture only limited dimensions of community dynamics. For example, variation in species richness overlooks shifts in species composition, whereas beta diversity quantifies the magnitude of change but provides little information about its direction. Trajectory analysis overcomes these limitations by integrating both magnitude and direction of compositional change, thereby offering a powerful framework for analysing community dynamics through time. However, species-level trajectories may be strongly influenced by pseudo-turnover, where replacement among functionally similar species artificially inflates apparent change and obscures meaningful long-term trends. To address this issue, we compared trajectories derived from species-level composition with those obtained after grouping species into functional categories based on key morphological and functional traits representing dominant ecological strategies. To assess the robustness of this approach, we further evaluated the sensitivity of the results using alternative grouping schemes. We applied this framework to a long-term vegetation resurvey dataset comprising 80 permanent plots distributed along an elevational gradient in the Central Apennines (Italy), surveyed four times between 2005 and 2025. Species-based trajectories were generally long and showed weak directionality, indicating substantial compositional reshuffling across surveys. In contrast, ordinations based on functional groups produced shorter and more directional trajectories, revealing clearer and more coherent temporal patterns among plots located at similar elevations. Functional aggregation therefore, reduced noise associated with pseudo-turnover and improved the interpretability of long-term vegetation dynamics. Overall, trait-based trajectory analysis may represent a robust methodological framework for detecting directional changes in vegetation time series and advancing our understanding of community assembly processes.