Attractor distributions in oscillatory networks in Table 1.

The same as Table 1 with all data measured within the network structures of periodic oscillations in Table 1. Since the existence of oscillation is a necessary condition for chaos, much more effective searching can be made among all the periodic GRNs. Obviously, we do observe considerably richer chaotic behaviors than that of Table 1. The amount of chaotic samples is also counted out of every tests, and all these chaotic samples will be used in the following for studying the mechanism of chaos.