Litter bag technique was used to study the decomposition and nutrient dynamics of marsh litter in the four communities, Carex pseudocuraica(C.pa), Carex lasiocarpa(C.la), Deyeuxia angustifolia(D.aa) and Deyeuxia angustifolia-Shrub(D.aa-Srb), in Sanjiang Plain, northeastern China. In the first year, decomposition could be divided into two periods and mass loss ranged 11.7%—31.4% of the initial mass during summer and autumn, both of which accounted for more than 75% of the annual loss. The decomposition rates ranged from 0.000612 to 0.000945 d-1 depending on flooding depth and duration and were differed significantly among communities. Litter decomposed faster in communities with deeper and perennial flooding than in those with shallow and seasonal flooding. The initial ratios of C∶N and C∶P were also different among the four litter types, but these difference had no effect on the decomposition rates, which might suggest that flooding status rather than litter quality was the main factor influencing the decomposition rates of marsh litter. N concentrations in C.pa and C.la almost continuo
usly increased over time and the final values were 2.8 and 2.4 times higher than the initial ones, respectively. However, nutrient dynamics in D.aa and D.aa-Srb presented another pattern, with a fast decrease during the first month and a following slow increase, and the final values were close to the initial at the end of the experiment. Litter accumulated substantial amounts of N in C.pa and C.la, while net N release from litter was observed both in D.aa and D.aa-Srb. The difference might be caused by the demand of microorganism for nutrition, and then limited by litter C∶N ratios and the availability of nitrogen from soil and marsh water In contrast to N dynamics, P concentrations of all litter types apparently decreased during the first month, then continued to decline in C.pa, remained constant in C.la and D.aa and slightly increased in D.aa-Srb. At the end of the experiment, P concentrations decreased by 56%, -5%, 47% and 24% of the initial values for C.la, C.pa, D.aa and D.aa-Srb, respectively. Net P release was observed in all marsh litter over 480 days of decomposition and the intensity of P release was different among communities which might be regulated by ratios of initial C∶P. The results suggested that in the marsh with N limitation litter tended to accumulate N and release P during the decomposition and the intensity of accumulation or release was closely related to the ratios of initial C∶N and C∶P, respectively.