borderline significantp = 0.05
The optimal growth increment model for females were improved by allowing for heterogenous variance between MPAs and control areas (ΔAIC = −3.14), with the variance being higher in the MPAs (18.5) than in control areas (13.6). For males, protection had a borderline significant effect on both skipped moulting ( figure 2 ; LRT: status × TL: L = 3.75, d.f. = 1, p = 0.05) and growth increment ( figure 3 ; LRT: status × TL: L = 4.07, d.f. = 1, p = 0.04).