A note on the sufficiency of low resolution ecosystem models