PS 18-247 - Developing multi-scaled, probabilistic forecasts of woodland dynamics in the Great Basin

Abstract: Climate change is expected to cause major changes in the distribution of species worldwide. We can better anticipate these changes by quantifying and predicting shifting species abundance at their range limits. Pinyon-juniper woodlands are a dominant ecosystem across the western US, covering 170,000 km². Through prehistoric and historic times, the distribution of pinyon-juniper woodlands has been incredibly dynamic, with several range-wide expansion and contraction events. The ability to predict future shifts in pinyon-juniper abundance and extent will play a key role in anticipating shifting carbon source-sink dynamics, wildlife habitat, and forage production across the western US. Forecasts that include underlying demographic rates of woodland species are likely to be more robust to novel climate conditions because they provide a mechanistic basis for the persistence and viability of woodland populations. Traditional approaches to demographic modeling require time intensive field data collection. However, satellite and historical aerial imagery can serve as an alternative source of vegetation data with high spatial and temporal resolution and extent, making them extremely useful for monitoring long-term changes in vegetation cover and structure. These long-term observations of vegetation can be used to develop and train forecast models that can then be validated and updated when new data become available, creating the opportunity for testable predictions of future climate response. The objective of our project is to leverage historical aerial imagery and satellite-derived vegetation products to quantify range-wide shifts in woodland species abundance. Using spatiotemporal demographic approaches, we can: 1) forecast range expansion and/or contraction given future climate projections, and 2) build models to assess demographic performance in response to temporal and spatial variation in climate. Preliminary analyses show great promise for using remotely sensed data products to understand and forecast changes in the distribution and abundance of pinyon-juniper woodland species. For example, our preliminary hierarchical spatiotemporal matrix model accurately predicts woodland expansion in low cover areas along the range margins but performs less well in high cover areas. This may indicate that separating out demographic processes like growth vs. survival is necessary to improve forecasts in high cover areas. Future work will focus on estimating demographic rates from historical aerial imagery, to develop structured population models of stand dynamics.