Implementing plant hydraulic architecture within the LPJ Dynamic Global Vegetation Model

Abstract

Aim To implement plant hydraulic architecture within the Lund–Potsdam–Jena Dynamic Global Vegetation Model (LPJ–DGVM), and to test the model against a set of observational data. If the model can reproduce major patterns in vegetation and ecosystem processes, we consider this to be an important linkage between plant physiology and larger‐scale ecosystem dynamics. Location The location is global, geographically distributed. Methods A literature review was carried out to derive model formulations and parameter values for representing the hydraulic characteristics of major global plant functional types (PFTs) in a DGVM. After implementing the corresponding formulations within the LPJ–DGVM, present‐day model output was compared to observational data. Results The model reproduced observed broad‐scale patterns in potential natural vegetation, but it failed to distinguish accurately between different types of grassland and savanna vegetation, possibly related to inadequate model representations of water fluxes in the soil and wildfire effects. Compared to a version of the model using an empirical formulation for calculating plant water supply without considering plant hydraulic architecture, the new formulation improved simulated patterns of vegetation in particular for dry shrublands. Global‐scale simulation results for runoff and actual evapotranspiration (AET) corresponded well to available data. The model also successfully reproduced the magnitude and seasonal cycle of AET for most EUROFLUX forests, while modelled variation in NPP across a large number of sites spanning several biomes showed a strong correlation with estimates from field measurements. Main conclusions The model was generally confirmed by comparison to observational data. The novel model representation of water flow within plants makes it possible to resolve mechanistically the effects of hydraulic differences between plant functional groups on vegetation structure, water cycling, and competition. This may be an advantage when predicting ecosystem responses to nonextant climates, in particular in areas dominated by dry shrubland vegetation.