Recently, global and some regional observations of soil carbon stocks and t
urnover times have implied that warming may not deplete soil carbon as much
as predicted by ecosystem models. The proposed explanation is that microbi
al respiration of carbon in 'old' mineral pools is accelerated less by warm
ing than ecosystem models currently assume. Data on the sensitivity of soil
respiration to temperature are currently conflicting. An alternative or ad
ditional explanation is that warming increases the rate of physico-chemical
processes which transfer organic carbon to 'protected', more stable, soil
carbon pools. These processes include adsorption reactions, some of which a
re known to have positive activation energies. Theoretically, physico-chemi
cal reactions may be expected to respond more to warming than enzyme-mediat
ed microbial reactions. A simple analytical model and a complex multi-pool
soil carbon model are presented, which separate transfers between pools due
to physico-chemical reactions from those associated with microbial respira
tion. In the short-term, warming depletes soil carbon. But in the long-term
, carbon losses by accelerated microbial respiration are offset by increase
s in carbon input to the soil (net production) and any acceleration of soil
physico-chemical 'stabilization' reactions. In the models, if net producti
on rates are increased in response to notional warming by a factor of 1.3,
and microbial respiration (in all pools) by 1.5, then soil carbon at equili
brium remains unchanged if physico-chemical reactions are accelerated by a
factor of about 2.2 (50% more than microbial reactions). Equilibrium soil c
arbon increases if physico-chemical reactions are over 50% more sensitive t
o warming than soil respiration. (C) 2001 Annals of Botany Company.