3.3 Transition
Strong support for a three-way interaction between time, colony and state on transition probabilities was evident from the model selection (model C36, Table A3 in the appendix). At both colonies, breeding individuals were more likely to breed again than to skip reproduction the following year except in 2019/20 at Robben Island (Figure 4a). However, breeding individuals were generally more likely to skip reproduction at Robben Island than at Stony Point. Individuals at Robben Island also showed an increasing prevalence for reproductive skipping from 2014 onwards (Figure 4a). Colony differences were especially evident between 2017/18 and 2019/20, with all estimates at Robben Island >0.3, whereas estimates at Stony Point were all <0.1 (Figure 4a).
Transitions out of the nonbreeding state were more variable and less accurately estimated (Figure 4b). At Robben Island, the probability of nonbreeding birds becoming breeders increased between 2015/16 and 2018/19, followed by a decrease in 2019/20, but estimates at Stony Point showed no clear trend. Overall, nonbreeders were more likely to breed the following year than skip reproduction again (overall probability ± SE = 0.66 ± 0.06 at Robben Island and 0.69 ± 0.06 at Stony Point, based on a constant model). However, individuals skipping reproduction were still generally less likely to breed the next year than breeding individuals (Figure 4).
Finally, transition estimates for prebreeders were also poorly estimated, because of the small number of individuals in this state, especially during the earlier years of the study. However, the estimates suggest a general decrease in the probability of transitioning into a breeding state over time at both colonies, but with prebreeders consistently more likely to breed the following year at Stony Point than at Robben Island (Figure 4c).