Observations of different plant communities over several decades reveal fiv
e types of dynamics: constant, fluctuating, growing, decaying, intermittent
. With a modelling approach based on interacting particle systems (cellular
automata), these distinct types of dynamics can be simulated. In addition,
the proposed model shows a strong tendency towards the formation of patche
s when started with random initial conditions. Further, it gives evidence f
or dominance resulting from the interplay of plant characteristics and patt
erns rather than being attributable to certain plant types alone, so making
questionable the existence of key species and the usefulness of a constant
mapping key over extended periods. The presented model unifies the fundame
ntal dichotomy in vegetation dynamics between determinism (predictability)
and disorder (chance effects) by showing the outcome of both classical theo
ries as special cases.