Re-analysis of the island closure experiment results to implement the suggestions of the December 2020 International Panel
The suggestions of the December 2020 International Panel for further analysis of the results from the island closure experiment, in particular to use the same data within a common framework to facilitate comparisons, and to include month as a covariate, are implemented for all response variables exc...
| Main Authors: | , |
|---|---|
| Format: | Report |
| Language: | unknown |
| Published: |
Faculty of Science
2021
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/11427/33665 https://open.uct.ac.za/bitstream/11427/33665/1/FISHERIES_2021_JUN_SWG-PEL_35%20Re-analysis%20of%20the%20ICE%20results.pdf |
| Summary: | The suggestions of the December 2020 International Panel for further analysis of the results from the island closure experiment, in particular to use the same data within a common framework to facilitate comparisons, and to include month as a covariate, are implemented for all response variables except chick survival. Broadly, the results indicate that aggregated and disaggregated approaches give the same results for estimates and variances of the island closure effect (see Figure 2). This is especially a consequence of implementing the disaggregated approach with a nesting structure as advocated by the Panel, which is shown to be statistically justified. Inclusion of the month co-variate does impact results to some extent, the more so for foraging data for the west coast islands (see Figure 3). The sensitivities investigated generally make little difference to results; these checks include extensions to incorporate data prior to 2008. As to be expected, REML based estimates of CIs are somewhat wider than those based on MLE. |
|---|