Vegetation/Ecosystem Modeling and Analysis Project:Comparing biogeography and biogeochemistry models in a continental‐scale study of terrestrial ecosystem responses to climate change and CO2 doubling

Abstract

We compare the simulations of three biogeography models (BIOME2, Dynamic Global Phytogeography Model (DOLY), and Mapped Atmosphere‐Plant Soil System (MAPSS)) and three biogeochemistry models (BIOME‐BGC (B io G eochemistry C ycles), CENTURY, and Terrestrial Ecosystem Model (TEM)) for the conterminous United States under contemporary conditions of atmospheric CO 2 and climate. We also compare the simulations of these models under doubled CO 2 and a range of climate scenarios. For contemporary conditions, the biogeography models successfully simulate the geographic distribution of major vegetation types and have similar estimates of area for forests (42 to 46% of the conterminous United States), grasslands (17 to 27%), savannas (15 to 25%), and shrublands (14 to 18%). The biogeochemistry models estimate similar continental‐scale net primary production (NPP; 3125 to 3772 × 10 12 gC yr−1) and total carbon storage (108 to 118 × 10 15 gC) for contemporary conditions. Among the scenarios of doubled CO 2 and associated equilibrium climates produced by the three general circulation models (Oregon State University (OSU), Geophysical Fluid Dynamics Laboratory (GFDL), and United Kingdom Meteorological Office (UKMO)), all three biogeography models show both gains and losses of total forest area depending on the scenario (between 38 and 53% of conterminous United States area). The only consistent gains in forest area with all three models (BIOME2, DOLY, and MAPSS) were under the GFDL scenario due to large increases in precipitation. MAPSS lost forest area under UKMO, DOLY under OSU, and BIOME2 under both UKMO and OSU. The variability in forest area estimates occurs because the hydrologic cycles of the biogeography models have different sensitivities to increases in temperature and CO 2. However, in general, the biogeography models produced broadly similar results when incorporating both climate change and elevated CO 2 concentrations. For these scenarios, the NPP estimated by the biogeochemistry models increases between 2% (BIOME‐BGC with UKMO climate) and 35% (TEM with UKMO climate). Changes in total carbon storage range from losses of 33% (BIOME‐BGC with UKMO climate) to gains of 16% (TEM with OSU climate). The CENTURY responses of NPP and carbon storage are positive and intermediate to the responses of BIOME‐BGC and TEM. The variability in carbon cycle responses occurs because the hydrologic and nitrogen cycles of the biogeochemistry models have different sensitivities to increases in temperature and CO 2. When the biogeochemistry models are run with the vegetation distributions of the biogeography models, NPP ranges from no response (BIOME‐BGC with all three biogeography model vegetations for UKMO climate) to increases of 40% (TEM with MAPSS vegetation for OSU climate). The total carbon storage response ranges from a decrease of 39% (BIOME‐BGC with MAPSS vegetation for UKMO climate) to an increase of 32% (TEM with MAPSS vegetation for OSU and GFDL climates). The UKMO responses of BIOME‐BGC with MAPSS vegetation are primarily caused by decreases in forested area and temperature‐induced water stress. The OSU and GFDL responses of TEM with MAPSS vegetations are primarily caused by forest expansion and temperature‐enhanced nitrogen cycling.