Moderate forest disturbance as a stringent test for gap and big-leaf models
- 1Pacific Northwest National Laboratory, Joint Global Change Research Institute at the University of Maryland–College Park, 5825 University Research Court, Suite 3500, College Park, Maryland, MA 20740, USA
- 2Department of Geographical Sciences, 1150 LeFrak, University of Maryland, College Park, Maryland, MA 20742, USA
- 3Climate Sciences Department, Lawrence Berkeley National Laboratory, 1 Cyclotron Rd., MS 74-0171, Berkeley, CA 94720, USA
- 4Pacific Northwest National Laboratory, 902 Battelle Boulevard, Richland, WA 99352, USA
- 5Department of Civil, Environmental and Geodetic Engineering, The Ohio State University, 470 Hitchcock Hall, 2070 Neil Avenue, Columbus, Ohio, OH 43210, USA
- 6Virginia Commonwealth University, Department of Biology, P.O. Box 842012, 1000 West Cary Street, Richmond, VA 23284-2012, USA
Abstract. Disturbance-induced tree mortality is a key factor regulating the carbon balance of a forest, but tree mortality and its subsequent effects are poorly represented processes in terrestrial ecosystem models. It is thus unclear whether models can robustly simulate moderate (non-catastrophic) disturbances, which tend to increase biological and structural complexity and are increasingly common in aging US forests. We tested whether three forest ecosystem models – Biome-BGC (BioGeochemical Cycles), a classic big-leaf model, and the ZELIG and ED (Ecosystem Demography) gap-oriented models – could reproduce the resilience to moderate disturbance observed in an experimentally manipulated forest (the Forest Accelerated Succession Experiment in northern Michigan, USA, in which 38% of canopy dominants were stem girdled and compared to control plots). Each model was parameterized, spun up, and disturbed following similar protocols and run for 5 years post-disturbance. The models replicated observed declines in aboveground biomass well. Biome-BGC captured the timing and rebound of observed leaf area index (LAI), while ZELIG and ED correctly estimated the magnitude of LAI decline. None of the models fully captured the observed post-disturbance C fluxes, in particular gross primary production or net primary production (NPP). Biome-BGC NPP was correctly resilient but for the wrong reasons, and could not match the absolute observational values. ZELIG and ED, in contrast, exhibited large, unobserved drops in NPP and net ecosystem production. The biological mechanisms proposed to explain the observed rapid resilience of the C cycle are typically not incorporated by these or other models. It is thus an open question whether most ecosystem models will simulate correctly the gradual and less extensive tree mortality characteristic of moderate disturbances.