117 The number of islets surveyed each season varied from 11 in 2017-17 to 25 in 2022-23, and the total number of field surveys from 5 in 2017-18 and 2018-19 to 12 in 2022-23. Overall, a nesting activity was recorded on 27 out of the 28 surveyed islets and the mean total nesting success was estimated at 42.4 ± 10.3% (the nesting success per islet and per season is available in Table 16 SI Appendix). The seasonal predicted number of nests over the surveyed islets, as generated by Bourgogne et al. (2024c), is available in Table 16 SI Appendix. These data were used to produce a graphic representation of the spatiotemporal variability of the nesting activity at the scale of the GLS (Figure 23). An important heterogeneity in both the total number of nests and their distribution is highlighted in space and time at the scale of the GLS area (Chi-squared test, X-squared = 493.95, df = 135, p-value < 0.001 ***). A significant variability in the inter-season total number of nests is reported (Kruskal-Wallis test, chi-squared = 6, df = 6, p-value =0.423*), with a maximum of 532 nests recorded during 2017-18 and a minimum of 303 nests in 2020-21. The mean number of nests was also significantly different among islets (Kruskal-Wallis test, chisquared = 91.045, df = 27, p-value < 0.001***), with a maximum of 56.8 ± 14.8 nests on Kie (islet #26 on the study area map) and minimum of 2.5 ± 2.3 nests on Tere (islet #14 on the map). The predicted number of nests is stable over time for some islets, whereas it can significantly fluctuate from one season to another for others. For example, the explained percentages of the total predicted number of nests over the GLS are consistently high on Atire (6.9 ± 1.1%, islet #1 on the map) or on Gi (6.8 ± 2.4%, islet #16 on the map), and consistently low on Ugo (0.7 ± 0.7%, islet #20 on the map) or on Noe (0.8 ± 0.6%, islet #10 on the map). They can also substantially vary from one season to another, for example on Ndo (4.4 ± 3.9%, islet #23 on the map) or on Redika (3.1 ± 2.4%, islet #2 on the map). Over the full study period, the initial 28 surveyed islets have been divided into four sets, each of them composed of 7 islets (one fourth of the total number of surveyed islets) sorted by the descending order of importance in terms of predicted number of nests, and referred to as 1st to 4th set of islets (Table 16 SI Appendix). The cumulative percentage of the total predicted number of nests (based on the ‘nest_islet_median’ variable) and the cumulative percentage of the total length of available nesting beach linear (based on the ‘periv_fav’ variable) values have been attributed to each set of islets (Figure 24). Overall, no significant difference in the cumulative percentage of the available nesting beach linear values is observed between the four sets of islets (Chi-square test, X-squared = 2.276, df = 3, p-value = 0.517), yet the cumulative percentage of the total number of nests values appear to be significantly different (Chi-squared test, X-squared = 22.73, df = 3, p-value < 0.001 ***). The 1st set of islets,
RkJQdWJsaXNoZXIy MjE1NDI=