2017 ESA Annual Meeting (August 6 -- 11)

COS 132-8 - Stabilization of savanna-forest alternative states to closed forests after ~150 years of land use and climate changes

Thursday, August 10, 2017: 10:30 AM
E145, Oregon Convention Center
Kelly Heilman, Department of Biological Sciences, University of Notre Dame, Notre Dame, IN, Jason McLachlan, Biological Sciences, University of Notre Dame, Notre Dame, IN and Carla Staver, Ecology and Evolutionary Biology, Yale University, New Haven, CT
Background/Question/Methods

Globally, the transition from open savanna to closed forest is often abrupt and not fully explained by climate or edaphic factors. Savannas and forests may be alternative stable states driven by fire-vegetation feedbacks, which results in a bimodal distribution in tree density that is not well explained by the environment. This is observed across the tropics, fueling concerns of catastrophic biome transitions with environmental change. Before widespread agro-industrial transformation, the Midwestern US had similar vegetation types, including both savanna and forest. In this study, we asked whether historical vegetation data provided evidence for alternative stable states in the region, and whether these conditions have persisted through ~150 years of regional fire suppression and climatic changes on the modern landscape. We estimated pre-European settlement stem density across five Upper Midwestern states using historical survey data collected in the 19thcentury Public Land Survey (PLS), and compared historic and modern relationships between density, soils, and climate. We hypothesized that PLS tree density distribution was bimodal and not well explained by environmental factors due to intact vegetation-disturbance feedbacks along the savanna-forest boundary before settlement, but fragmentation and fire suppression has eroded savanna-forest bistability, resulting in a modern boundary that is not bimodal.

Results/Conclusions

PLS tree density had a significantly bimodal distribution, with savanna and forest modes. Tree density distributions were not well explained by climate or edaphic factors, with as much as 18% (99,584km2) of the Midwest historically characterized by alternative stable states (Hartigan’s Diptest p <0.05, GAM R2 = 0.659). If these historical environment-vegetation relationships persisted today, 18% of the modern landscape has a climate where both savanna and forest are possible. However, modern tree density is not significantly bimodal, suggesting that fire suppression in the region has eroded the bistability of savanna and forest, such that only forests are stable. Savanna regions in the PLS era have significantly increased in tree density (p < 0.01), providing further support for fire suppression as a mechanism for stabilization. Projecting historic climate-vegetation relationships to future scenarios, 19 - 26% of the Midwest would occur in bistable conditions by 2070, under the CCESM4 model projected climate for IPCC RCP 2.6, 4.5, and 8.5 emissions scenarios. However, under the same future climate scenario and modern climate-vegetation relationships, Midwestern forests are likely to remain stable closed forests. Our results suggest that ~150 years of environmental change have pushed the region into a stable closed forest state.