ANALYSIS OF NUTRIENT-CYCLING DYNAMICS, FOR PREDICTING SUSTAINABILITY AND CO2-RESPONSE OF NUTRIENT-LIMITED FOREST ECOSYSTEMS

The dynamics of nutrient cycling within a forest system involve proces ses operating on many different timescales, ranging from seconds to th ousands of years. In their response to environmental change, trees can respond initially in one direction and then later reverse the respons e due to a process acting on a longer time scale. Thus, theoretical mo dels are required for the interpretation of short-term experiments whe re CO2 concentration or nutrient availability are modified. Comins and McMurtrie (1993) delineate the G'DAY plant-soil model, which attempts to incorporate nutrient cycling responses on various time scales in a model of moderate complexity, using the Century model of soil nutrien t dynamics coupled with a simple model of tree growth. The model shows a complex response to doubling CO2 under nitrogen-limited conditions; large or small short-term responses can be independently combined wit h either large or small long-term responses, depending on certain deta ils of plant and soil dynamics. Comins and McMurtrie describe the long -term behaviour of the system by calculating the equilibrium of the mo del, excluding the passive soil organic matter pool. This analytical r esult makes clear which parameters of the model determine the long-ter m dynamics. Subsequently it has been found that a more general descrip tion of long-term dynamics requires understanding both the equilibrium pool sizes and the equilibration time of the system. This paper deriv es an analytical expression for the equilibration time, again permitti ng the important parameters of the model to be identified. It also des cribes an intermediate level of approximation, using differential equa tions for slow soil organic matter carbon and nitrogen pools. This is useful for generalisations of the model, and for describing the long-t erm dynamics in a gradually changing environment, as opposed to the dy namics following a step change. It is found that most parametrisations of G'DAY which have low nitrogen loss rates also have relatively long equilibration times. The equilibration time is approximately predicte d by the slow pool decay time divided by the 'slow nitrogen loss fract ion' (defined as the fraction of nitrogen released from the slow soil organic matter pool which leaves the system without ever re-entering t he slow pool). In a slowly changing environment, the response of G'DAY can be approximately predicted using equilibrium analysis and a time lag, where the appropriate time lag is equal to the predicted equilibr ation time. (C) 1997 Elsevier Science B.V.