Content area
Full text
The architecture (here, the size distribution combined with the spatial pattern of individuals) of natural forest at demographic equilibrium can be used to infer the demographic processes that drive the forest dynamics. In particular, a constant growth rate and a constant mortality rate for all trees would generate an exponential distribution of their size, whereas the metabolic scaling theory predicts a power distribution. In an undisturbed tropical rainforest in French Guiana, the diameter distribution was significantly steeper than the best-fit exponential distribution and significantly flatter than the best-fit power distribution. A simple individual-based model of forest dynamics with asymmetric competition between trees, where the strength of competition was regulated by a single parameter, was able to predict the observed distribution. Competition drove the forest into a self-organized state with stronger inequalities of size among trees, a lower mean competition index, and a spatial pattern of trees that deviated from complete spatial randomness.
Keywords. demographic modeling, forest dynamics, French Guiana, scaling theory, space-dependent competition
(ProQuest: ... denotes formulae omitted.)
Models of forest dynamics have proven to be useful tools for the sustainable management of forest stands (Vanclay 1994). With the current issues on global change, predictions at larger scales than the stand and a better understanding of forest functioning have become a necessity. Starting from gap models and space-independent models in the 1970s, development in forest dynamics modeling has moved toward greater details on processes at the individual level, such as the three-dimensional modeling of light interception by the canopy (Chave 1999), or has integrated processes at a lower scale than the individual and the annual time step, like photosynthesis at the leaf level on a daily basis (Friend et al. 1997). Mimicking biological processes by increasing the level of details in forest simulators increases the generality of the model. However, a drawback of increased details is to replace a complex reality by a numerical representation that is also complex, thus preventing from understanding the behavior of the model and scaling it up to address global questions (Van Nes and Scheffer 2005). More concerning is that the increased level of detail in models does not necessarily improve the quality of predictions (Lischke et al. 1998, Pfister and Stevens 2003, Tietjen and Huth 2006), thus bringing...