Assessing the Impact of Environmental and Management Variables on Mountain Meadow Yield and Feed Quality Using a Random Forest Model.

Seasonal climate variability and agronomic management profoundly influence both the productivity and nutritive value of temperate hay meadows. We analyzed five years of data (2019, 2020, 2022-2024) from 15 meadows in the central Spanish Pyrenees to quantify how environmental variables (January-June minimum temperatures, rainfall), management variables (fertilization rates (N, P, K), livestock load, cutting date), and vegetation (plant biodiversity (Shannon index)) drive total biomass yield (kg ha), protein content (%), and Relative Feed Value (RFV). Using Random Forest regression with rigorous cross-validation, our yield model achieved an R of 0.802 (RMSE = 983.8 kg ha), the protein model an R of 0.786 (RMSE = 1.71%), and the RFV model an R of 0.718 (RMSE = 13.86). Variable importance analyses revealed that March rainfall was the dominant predictor of yield (importance = 0.430), reflecting the critical role of early-spring moisture in tiller establishment and canopy development. In contrast, cutting date exerted the greatest influence on protein (importance = 0.366) and RFV (importance = 0.344), underscoring the sensitivity of forage quality to harvest timing. Lower minimum temperatures-particularly in March and May-and moderate livestock densities (up to 1 LU) were also positively associated with enhanced protein and RFV, whereas higher biodiversity (Shannon ≥ 3) produced modest gains in feed quality without substantial yield penalties. These findings suggest that adaptive management-prioritizing soil moisture conservation in early spring, timely harvesting, balanced grazing intensity, and maintenance of plant diversity-can optimize both the quantity and quality of hay meadow biomass under variable climatic conditions.