Multi-scale approaches to understand the future of Earth's forests under climate change

Summary:

• Focus: Impact of climate change on Earth’s forests

• Climate change is a massive experiment with the Earth’s ecosystems

  • Climate models have a lot of uncertainty about its impacts on forest carbon storage

  • Uncertainty about emissions and response by forests

• Key questions:

  • Improving predictions of how individual trees respond

  • Scaling individual response predictions to entire globe

  • Using these models to inform decision making

• Ecophysiology of climate responses

  • Plant physiology

    • Surface energy fluxes

    • Hydrology

    • Carbon cycle

  • Focus: water loss through transpiration and carbon uptake via photosynthesis; both done via stomata on plant leafs

  • Plant hydraulic transport: using atmospheric pressure to pull water up from roots to leafs

    • Droughts make it hard to pull water, end up pulling air through channels

    • Embolism (air bubbles) can block water transport, kills trees

  • Can we link water transport to carbon uptake?

    • Using optimality theory

    • What’s the optimal behavior of leaf pores to optimize its carbon uptake and avoid dying from embolism?

    • Constrained optimization: optimal use of limited water

    • Incorporate competition among different tree species

  • Using this to build an individual tree level model

    • Tree dynamic properties

    • Boundary conditions: Wind, air temp, dryness, etc.

    • Model effectively models tree mortality due to physiological stress in experimental conditions

  • How well can we predict regional mortality patterns in the wild?

    • Compared model to FIA dataset of real tree observations in US

      • Challenge: FIA separates human and fire death from “other”, which is a combination of insects and droughts

    • Model explains 52% of variance in the dataset

    • Model over-predicts mortality and results vary significantly across species

• Modeling risks to forests as nature-based climate solutions

  • Impacts of forests

    • Albedo of forests (darker forests warm climate relative to bright snow)

    • Additional carbon storage in the forests

    • Shifting of activities: move logging from one forest to another

    • Risks to durability: how likely is a forest to burn and lose carbon

  • Climate/ecosystem risks: fire, drought, biotic agents, wind, ecological dynamics, etc.

  • Human-system risks: financial failure, management failure, policy changes, poor governance, economics, etc.

  • Goal: forecast these risks and understand their impacts on forests

    • Do different methods result in different predictions about the future?

  • US Disturbance modeling project

    • Landsat MTBS dataset

    • Historical and future drought/climate stress and insect risks modeled using FIA data

    • Downscalked 6 CMPI6 Earth System Models to 4km (SSPs: 245, 370, 585)

    • Forecast tree mortality from fire, drought and insects

    • Cross-validated accuracy: 

      • Fire R2 = .62

      • Drought R2 = .18

      • Insect R2 = .31

      • Lots of covariation across risks

    • Projecting climate risks to forests across different US regions

  • US Forest carbon modeling

    • Model:

      • Mechanistic vegetation model: CMIP6 carbon in vegetation (6 models)

      • Growth-mortality: Demographic approach: stand growth curves and stochastic & historical land future projections

      • Empirical Niche: Random Forest models of total vegetation carbon in 24 major forest groups

    • Applied to population of real forest carbon projects

      • Earth system models project 45% gains in forest carbon (note: missing most fire and drought/insect mortality drivers)

      • Growth-mortality model: 4% gain (losses in the West and gains in the East)

      • Empirical niche models predict 4% loss

    • Looking at forest carbon project we see that the losses in forests are expected to be larger than the buffer pools used by the project: significant risk of carbon storage failure

  • Global forest risk assessment

    • High risk in US West, Canada, Siberia, Amazon

    • However, there is a lot of inconsistency among risk datasets