Light in Tropical Forest Models: What Detail Matters?

Abstract:

Representations of light in models of tropical forests are typically unconstrained by field data and rife with assumptions, and for good reason: forest light environments are highly variable, difficult and onerous to predict, and the value of improved prediction is unclear. Still, the question remains: how detailed must our models be to be accurate enough, yet simple enough to be able to scale them from plots to landscapes? Here we use field data to constrain 1-D, 2-D, and 3-D light models and integrate them with simple forest models to predict net primary production (NPP) across an Andes-to-Amazon elevation transect in Peru. Field data consist of novel vertical light profile measurements coupled with airborne LiDAR (light detection and ranging) data from the Carnegie Airborne Observatory. Preliminary results indicate that while 1-D models may be “good-enough” and highly-scalable where forest structure is relatively homogenous, more complex models become important as forest structure becomes more heterogeneous. We discuss the implications our results hold for prediction of NPP under a changing climate, and suggest paths forward for useful proxies of light availability in forests to improve and scale up forest models.