Newly emerging mathematical models linking population dynamics with these key elements greatly improve historic trophic interaction models and resolve many existing paradoxes. Our study suggests that the “strict homeostasis" hypothesis sounds when the stoichiometric variability of herbivores is less than a threshold (67.6% under the median nutrient availability), but this threshold is independent of algal stoichiometric variability. The median herbivore's stoichiometric variability is about 50%, less than our default threshold, accepting the “strict homeostasis" hypothesis for many herbivores. However, our results show that strict homeostasis is invalid for some herbivores since the threshold depends on herbivore's traits such as turnover rates. An additional observation is that strict homeostasis is more valid when nutrient is more limiting. Results are nearly same in both one-nutrient and two-nutrient models, and robust to perturbation of parameter values and environmental nutrient status. Finally, we show that dynamics are more sensitive to N-associated stoichiometric variability when the environmental N:P ratio shifts from the Redfield ratio to be higher, while dynamics are more sensitive to P-associated stoichiometric variability when the environmental N:P shifts from the Redfield ratio to be lower. *This work is in collaboration with James J. Elser, Robert W. Sterner, Aditya Raghavan.