Background/Question/Methods Individual trait variation has important ecological implications for population and community properties. In particular, individual variation of metabolic rate links directly with the energy use estimation and scaling patterns in community ecology. Here, we examine the mean-variance relationship of individual metabolic rate by testing Taylor's law (a power-function relationship between mean and variance) for individual rescaled metabolic rate across tree communities in tropical forests. We use a constraint-based model called maximum entropy theory of ecology (METE) to estimate and predict the parameters of Taylor's law. Results/Conclusions Our results show that, when assuming a universal metabolic scaling between metabolic rate and aboveground biomass, METE generates the form of Taylor's law but fails to predict its slope. When setting the metabolic scaling exponent as a community-level parameter in the METE model, the estimated and predicted slope of Taylor's law agree with each other. Our parameterized METE model reveals the positive effect of number of individuals on the metabolic scaling exponent. These results suggest that fluctuation scaling of individual metabolic rate can be explained solely by the macroecological constraints of communities, without relying on the physiological or genetic characters of individual organisms.