Will global environmental changes drive smooth and predictable vegetation changes, or force abrupt and catastrophic shifts? If vegetation has a single linear relationship with environmental drivers, then responses to environmental changes should be predictable and reversible. However, in some systems, intrinsic vegetation feedbacks maintain multiple stable states (MSS) of vegetation in hysteresis across similar environments. Because of this hysteresis, catastrophic and nearly irreversible state shifts occur if extrinsic drivers push systems past a threshold. Biome-scale MSS are present globally, but catastrophic state transitions have not been documented at these scales. We use statistical estimates of both composition and tree density to test whether the Upper Midwestern US had biome-scale MSS in the 1800's before EuroAmerican settlement. Using modern Forest Inventory Analysis data, we evaluate whether ongoing fire suppression, climate, and land-use changes drove a catastrophic state shift resulting in a novel vegetation-environment relationship. Specifically, we fit a Bayesian mixture model to determine if tree density of both the past and modern landscapes is bimodal relative to three environmental covariates: climatic moisture balance, soil moisture, and the first Principal Component of environmental space. We also explore whether joint relationships between species and the environment have changed between the two time periods.
Results/Conclusions
Pre-European settlement tree density was significantly bimodal across a wide range of geographic and environmental space, providing evidence that over 99% of the past landscape existed in savanna and forest MSS. However, modern vegetation is both unimodal and intermediate to the past savanna and forest states, indicating that ongoing fire suppression and logging have shifted the system to a single, novel state. Thus, the modern vegetation-environment relationship does a poor job predicting either historical species composition or tree density, and vice versa. If modern land management continues to maintain forests in a unimodal, but noisy relationship with the environment, vegetation should have a smooth response to moderate future environmental changes. However, climate projections predict a much drier future compared to either historical or modern observations, making it unclear how long fire suppression efforts can continue to be successful at maintaining forested systems in the region. Here, we show that complex interactions between anthropogenic activities, climate, and disturbances can forceboth biome-level shifts in MSS and create novel vegetation-environment relationships. The historical shift of biome-scale MSS to a novel intermediate state documented here should serve as a useful model for the consequences of anthropogenic activities elsewhere in the world.