Revisiting k: Time-varying stream litter breakdown rates

Abstract

Litter decomposition is usually modeled with the negative exponential model, which assumes constant proportional mass loss. We assessed this assumption and its interpretive consequences using 145 stream litter mass loss time series and process-based simulations. Relatively simple (two to three parameters) models allowing time-varying decay rates produced more accurate predictions and were generally more parsimonious. Decomposition trajectories strongly deviated from constant decay for at least 50% of the time series, with the shape influenced by the degree of decomposition covered by a time series. Finally, simulations and empirical evidence suggested that the degree of decomposition covered can interact with time-varying decay rates and leachability to bias estimates of breakdown rates (k) from negative exponential models, obfuscating comparisons within and across studies. Considering alternative models could accelerate understanding and prediction of litter decomposition dynamics by enabling investigation of time-explicit decomposition dynamics and more precise modeling when warranted.