News & Comment

that most parameters can be estimated, and that the fundamental mechanisms involved can be understood. It thus represents a powerful alternative to the complex forest simulators that often have not yielded to easy interpretationlO. If the forest architecture hypothesis is correct, and if the model is indeed sufficient to capture the major features of forest dynamics, then two important pra dictions can be made. First, with good estimates of the parameters, simulations of the model should give a stable coexistence. Second, important quantitative features of the forest, such as age and size structure and their dynamics within gaps, should be reproduced by the model. Having these two predictions borne out amounts to a simultaneous test of the hypothesis and the particular model representation of it. If the second prediction were not borne out, there would be reason to be worried that the model is inadequate to capture the variation in forest structure on which the mechanism of coexistence depends. That does not seem to be the case here, and so we can regard the first prediction as a strong test of the forest architecture hypothesis. Based on point estimates of the parameters of the model, the hypothesis appears to account for coexistence of only the subcanopy and the understorey species. The reason for this failure is that the trade-off function specified by circles in Fig. 1 gives too high a recruitment advantage to the subcanopy and understorey species and, as a consequence, the canopy species is excluded. However, the second trade-off function, which lessens this advantage, does lead to stable coexistence of all three species. Kohyama says that the second trade-off function lies within the range of the accuracy of the data, although he does not give a 95% confidence region to support this claim. If he had, we would be able to say that the data are consistent with the forest architecture hypothesis of coexistence of all three species. However, we have seen that the data are also consistent with insufficiency of the hypothesis for coexistence of all three species. Only improvements in estimates of the parameters with more data can settle the issue.