A COMPARISON OF FOREST GAP MODELS - MODEL STRUCTURE AND BEHAVIOR

Forest gap models share a common structure for simulating tree populat ion dynamics, and many models contain the same or quite similar ecolog ical factors. However, a wide variety of formulations are being used t o implement this general structure. The comparison of models incorpora ting different formulations is important for model validation, for ass essing the reliability of model projections obtained under scenarios o f climatic change, and for the development of models with a wide range of applicability. This paper reviews qualitative and quantitative com parisons of the structure and behaviour of forest gap models. As examp les of qualitative model comparisons, the different formulations used for the height-diameter relationship, for the maximum growth equation, and for the effects of temperature and drought on tree growth are rev iewed. The variety of formulations currently in use has the potential to influence simulation results considerably, but we conclude that lit tle is known on the sensitivity of the models in this respect. The qua ntitative model comparisons performed so far allow us to draw the foll owing conclusions: (1) Gap models are quite sensitive to the formulati on of climate-dependent processes under current climate, and this sens itivity is even more pronounced under a changed climate. (2) Adaptatio ns of forest gap models to specific regions have required detailed sub -models of species life history, thus complicating model comparison. ( 3) Some of the complex models developed for region-specific applicatio ns can be simplified without hampering the realism with which they sim ulate species composition. (4) Attempts to apply the models without mo dification beyond the area for which they were developed have produced controversial results. It is concluded that the sensitivity of forest gap models to the exact process formulations should be examined caref ully, and that more systematic comparisons of model behaviour at a ran ge of test sites would be desirable. Such studies could improve our un derstanding of forest dynamics considerably, and they would help to fo cus future research activities with gap models.