Non‐random species loss weakens functional diversity and species asynchrony, destabilizing grassland communities

Abstract Environmental stressors such as grazing and climate variability often drive non‐random plant species loss. However, their effects on the temporal stability of plant communities remain unclear. We conducted a manipulative experiment in a semi‐arid grassland using four ecologically motivated species‐loss scenarios defined by the order of species removal along local abundance ranks: removal starting from the most abundant species, removal starting from the least abundant species, simultaneous removal from both ends of the abundance rank, and loss of random species. We evaluated the consequences of these scenarios for community stability at the local community scale by (1) testing for overall differences in community stability, (2) identifying pathways linking species loss to community stability and (3) quantifying the relative contributions of key stabilizing pathways and their components. Community stability showed no apparent decline under any species loss scenario, suggesting the short‐term robustness of grassland communities to species loss. However, the pathways maintaining this apparent stability differed markedly among scenarios. The loss of dominant species, or of both dominant and rare species, reduced stability by decreasing functional diversity and/or compensatory dynamics. Rare species loss alone had minimal effects, whereas random loss unexpectedly increased community stability by enhancing species‐level stability. The consequences of species loss for community stability therefore depended on which species were removed, with losses of dominant species causing the most pronounced reductions in stability. Synthesis . Our findings demonstrate that grassland community stability is shaped more by the order of loss of species than by the total number of species lost and how these losses reorganize stabilizing processes. To maintain grassland stability under global change, conservation efforts should consider species' functional roles and extinction drivers, because random species loss scenarios may underestimate the ecological consequences of realistic, non‐random loss pathways.