ecological modelling 219 ( 2 0 0 8 ) 342–360 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/ecolmodel Modeling forest leaf-litter decomposition and N mineralization in litterbags, placed across Canada: A 5-model comparison C.F. Zhang a , F.-R. Meng a , J.S. Bhatti b , J.A. Trofymow c , Paul A. Arp a,* a Faculty of Forestry and Environmental Management, Fredericton, New Brunswick, Canada E3B 6C2 b Canadian Forest Service, Northern Forestry Centre, 5320 122 Street, Edmonton, Alberta, Canada T6H 3S5 c Canadian Forest Service, Pacific Forestry Centre, 506 Burnside Road West, Victoria, B.C., Canada V8Z 1M5 article info Article history: Published on line 26 August 2008 Keywords: Organic matter decomposition Nitrogen mineralization Climate Forest litter type Modeling abstract The performances of five carbon (C) and nitrogen (N) cycling models (FLDM, CENTURY, SOMM, DOCDOM and CANDY) were compared for their ability to quantify mass and N remaining in 10 g leaf-litterbags across the wide range of litter types and sites of the Canadian Intersite Decomposition Experiment (CIDET), over a 6-year period. These models differ in their structure and assumptions, number of compartments, predictor variables and coefficients. Calibrated model outputs displayed similar but not identical trends for mass and N remaining, but differed substantially in mass and N contents per model-defined compartments. The quality of fit between model calculations and data varied as follows: FLDM > CENTURY > DOCMOD > CANDY > SOMM for mass remaining (0.73 < r 2 < 0.92), and FLDM > DOCMOD > CENTURY > SOMM > CANDY for the changing N con- centrations inside the bags (0.40 < r 2 < 0.80). FLDM calculations were the most consistent by CIDET site (21), litter type (10), and years of litterbag retrieval (1993–1998). Best-fitted models were used to project mass remaining and N concentrations inside the bags over the next 50 years, using mean July and January air temperatures, and annual precipitation and initial lit- ter composition as independent predictor variables. Projected model outputs converged for mass remaining, but diverged for the N concentrations, i.e., (1 ± 0.5) g and (2 ± 1)% at year 50, respectively. © 2008 Elsevier B.V. All rights reserved. 1. Introduction Predicting the rate at which forest litter decomposes is an important aspect of assessing past, current and future carbon (C) and nitrogen (N) responses of forests under changing cli- mate conditions. For this, there is a need to test the reliability and accuracy of existing models, and also to continue with the development of new models (Knorr et al., 2005; Powlson, 2005; Zhang et al., 2007). Recent studies have therefore focused on ∗ Corresponding author. E-mail address: arp2@eunb.ca (P.A. Arp). producing data on actual rates of forest litter decay for a wide range of litter, site and climate conditions. Among these stud- ies are: the Long-Term Intersite Decomposition Experiment (LIDET) in the United States (LIDET, 1995), the Decomposition Study in Europe (DECO: Jansson and Reurslag, 1992), and the Canadian Intersite Decomposition Experiment (Trofymow and CIDET Working Group, 1998). Factors that influence the rate of organic matter degradation are well known, namely litter quality (Heal et al., 1997), litter temperature and moisture, 0304-3800/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.ecolmodel.2008.07.014