We analyze the nonlinear stage of modulation instability in passively mode-locked fiber lasers leading to chaotic or noise-like emission. We present the phase-transition diagram among different regimes of chaotic emission in terms of the key cavity parameters: amplitude or phase turbulence, and spatio-temporal intermittency. In addition to short-pulse and cw emission regimes, ring fiber lasers used for passive mode locking via nonlinear polarization rotation exhibit the noise-like pulse emission mode of operation [1][2][3][4][5]. The resulting intense pulses have a broadband spectrum and low coherence length, which is of interest for metrology applications such as optical coherence tomography. Noise-like or turbulent laser emission may occur with both anomalous or normal cavity dispersion. Switching among various chaotic regimes involving different pulse timescales is obtained by adjusting the intracavity polarization controllers, which must be set away from the condition of minimal polarizer transmissivity for low power signals, so that weak mode-locking results. When operating in the average anomalous dispersion regime, coexistence and interaction of isolated soliton pulses with a chaotic background may lead to complex temporal dynamics involving, e.g., soliton rain and their condensation in liquid, glass, or crystal states [6,7]. In this work we point out that the dynamics of turbulent light emission from fiber lasers with passive mode locking may be analyzed by means of a universal model, namely the complex Ginzburg-Landau equation (CGLE) [8]. This approach permits us to directly relate the key cavity parameters to phase transitions among different regimes of chaotic behavior. Previous studies have been mostly devoted to the stable short-pulse generation regime, where nonlinear gain describes a fast saturable absorber [9][10][11]. In that case, for mode-locked soliton stability, a quintic nonlinear gain saturation term is added to the CGLE [11]. Here we analyze the modulation instability (MI) of cw emission and its associated nonlinear development and consider situations where the intracavity polarization analyzer acts as fast nonlinear gain saturation element. In this case the cubic CGLE is an appropriate minimal model.Light propagation in a fiber laser may be described in the mean field approximation by the CGLE [12]:where t 0 is a continuous slow temporal variable that replaces the round-trip number, τ 0 is a retarded time, E is the complex field envelope defined so that jEj 2 measures optical power, t r is the round-trip time for a cavity of length L, g 0 > 0 is the average distributed power gain in the cavity at the center frequency ω 0 , and T is the amplitude transmission coefficient of the output coupler. Moreover, we approximated the cavity bandwidth limited gain as Gω g 0 1 − 4∕B 2 ω − ω 0 2 , so that β 0 4g 0 ∕B 2 , β 2 is the average group velocity dispersion (GVD), γ 0 ω 0 n 2 ∕cA eff is the nonlinear coefficient of the fiber, n 2 is the nonlinear index, and A eff the effective mode area. Gain b...