The accelerating loss of biodiversity and ecosystem functions has become one of the most pressing global environmental challenges, as habitat degradation, land-use intensification, climate change, and the decline of keystone species such as large herbivores have fundamentally reshaped ecosystems across biomes. The United Nations decade on ecosystem restoration 2021-2030 calls for action to “prevent, halt, and reverse” these trends. Among the most promising, yet contested, restoration strategies is rewilding, which seeks to restore natural ecosystem processes following human disturbance and to promote self-regulating, dynamic ecosystems by reducing human control. Rewilding emphasizes the recovery of trophic interactions through the re-establishment of locally extinct large native herbivores and predators, or their ecological proxies when extinct. This approach builds on the recognition that virtually all terrestrial ecosystems evolved under the persistent influence of large mammals, and that human-induced megafauna extinctions over the last ~50,000 years have profoundly changed ecosystem dynamics.
Despite growing interest in rewilding, robust empirical evidence for its long-term ecological effects remains scarce. Most studies are conceptual, short-term, or spatially limited, and continuous multi-decadal datasets capturing the full trajectory of rewilding are virtually absent. Consequently, as almost all empirical evidence stems from a period of accelerating anthropogenic climate warming, the ecological effects of rewilding and climate change are difficult to disentangle. Yet, ecosystem responses to rewilding prior to modern warming may have differed fundamentally from those observed under ongoing climate change.
To address this gap, we leverage a unique, century-scale, multi-trophic dataset from the Swiss National Park, the world’s oldest large-scale scientific rewilding site (170 km2, established 1914). Upon its foundation, all former human land-use, such as alpine farming, hunting, forestry and mining, was halted, and the area was placed under strict protection. Red deer (Cervus elaphus), locally extinct at park establishment, naturally recolonized in 1918 and rapidly increased in abundance until stabilizing at high population levels by the late 1990s, just as local summer temperatures began rising. This distinct chronology enables us to disentangle sequential phases of land-use cessation, herbivore recovery, and subsequent climate warming, an analytical separation rarely possible in existing ecological time series.
We compiled 1,281 vegetation surveys (1917-2025) from 129 permanently marked plots with species-level cover estimates, encompassing 337 vascular plant species and a mean re-survey interval of nine years. These high-resolution data are complemented by annual red deer counts since 1915 and continuous local climate records since 1917. Sequential phases of rewilding and climate warming were identified using breakpoint analyses of deer abundance and summer temperature time series. Plant community trajectories were visualized with a cumulative-sum approach to capture non-linear dynamics, and generalized mixed-effects models, along with standardized effect sizes, quantified the impacts of herbivory and warming on plant diversity, temporal turnover, and spatial heterogeneity across grasslands and forests.
By spanning pre-warming and warming periods within a strictly protected, long-term rewilding site, our study provides rare, century-scale empirical evidence of how large herbivore recovery shapes vegetation dynamics and plant biodiversity. Looking ahead, the recent return of wolves adds an additional layer of trophic complexity, offering an exciting opportunity to track how cascading predator-herbivore-plant interactions continue to shape ecosystem dynamics in the coming decades.