The Seduction by Mechanism: A Reply to Tilman

Many ecologists (Clark et al. 1973; Pielou 1981; Hall 1988; Peters 1988) have criticized ecological models whose state variables are unmeasurable or whose predictions are untestable under natural conditions. Tilman's (1988) model of plant strategies is one of the best mechanistic models in plant ecology because it largely avoids those pitfalls and yet applies to common and widespread patterns of vegetation structure and dynamics found in the field. The popularity of the model suggests that it will have a large impact on future research in plant community ecology. It is because the best models deserve the most concentrated scrutiny that we designed the experiments described in Shipley and Peters (1990a); we believe that scientific ideas, including our own, can only improve after their weaknesses have been identified. We are happy to see that Tilman's (1990) response to, and reanalysis of, our work was done in the same constructive spirit. We would like to take this opportunity to comment further on some aspects of Tilman's (1988) model of plant strategies that we left unclear in our original paper and to respond to the criticisms in Tilman (1990). According to Tilman (1990), one of the main conclusions in Shipley and Peters (1990a) is incorrect because we had misinterpreted his theory. In responding to this, it is important o emphasize which aspects of the model we wanted to test since there are at least two components. First, there is a hypothesis at the level of whole-plant physiology, relating the interactions among relative growth rate, biomass partitioning, and physiological parameters involving carbon fixation and nutrient uptake. This hypothesis is developed in detail in the appendix of Tilman (1988) and is incorporated into the full version of the simulation model Allocate but was not of primary concern to us. Second, there is a simplified version of Allocate, explored in the text (chaps. 4-8), that incorporates a modification of the physiological hypothesis. This modified version is used to explain patterns of biomass allocation, life histories, and population dynamics along gradients of productivity, disturbance, and time (i.e., succession). The predictions concerning vegetation structure and dynamics along gradients of fertility and disturbance produced by the simplified version of Allocate are logically dependent on the assumed negative correlation between maximum relative growth rate and biomass partitioning to nonphotosynthetic tissues, as stated in Shipley and Peters (1990a). This negative correlation was introduced when Tilman (1988) assumed that "species-to-species differences in morphology are much greater than species-to-species differences in nutrientand light-saturated rates of photosynthesis per unit biomass or respiration rates per unit biomass" (Tilman 1988, p. 60; Tilman 1991). Thus, the physiological parameters of Allocate