MARAM_IWS_DEC14_PENG_B12.pdf (945.42 kB)
Download file

Updated Analyses of the Results from the Island Closure Feasibility Study for the Dassen/Robben and St Croix/Bird Island Pairs given Revised Data and Responses to Matters Raised in Documents MARAM/IWS/DEC14/Peng/A3

Download (945.42 kB)
posted on 25.03.2022, 13:17 by Doug ButterworthDoug Butterworth, Naseera MoosaNaseera Moosa, Liam Furman, William Robinson, Susan HollowaySusan Holloway
Repetition of earlier analyses of the data from the island closure feasibility study, given some corrected and extended data series, makes little change to results for Robben and Dassen Islands, except that the estimated power of foraging–related response variables to achieve statistically significant results decreases appreciably. However for St Croix Island, the inference of a negative impact of fishing is strengthened. The rationales offered in MARAM/IWS/DEC14/Peng/A3 for using closure instead of catch as a covariate in the analysis, and for restricting data to the years from 2008 onwards, are questioned. Comparisons conducted by applying the MARAM/IWS/DEC14/Peng/B4 approach, which uses annual means of response variables, indicate that the use of closure rather than catch as the covariate generally results in poorer precision and fewer statistically significant estimates of the fishing effect parameter λ. Furthermore when catch is used as the covariate, appreciably better precision for estimates of λ is generally achieved by including all years in the analyses, rather than by restricting them to the period from 2008 onwards. Importantly comparative estimates of λ from the Peng/B4 approach are shown to achieve better precision generally than those from the more complex and data-intensive Peng/A3 approach, thus negating the assertions in Peng/A1 and Peng/A2 that estimates from the former are compromised by their dependence on response variable means alone. The failure of Peng/A3 to report the variance estimates needed for input to the power analysis required for the feasibility study is noted. Furthermore Peng/A3 offers no specification of the simulation studies necessary to carry out a power analysis for the estimators which it proposes, so that it has failed to address this key first step in this overall closure study process. Peng/A3 has provided some strongish evidence that closures may benefit penguins, but for the Eastern Cape colonies only. However it has failed to address the primary aim of the feasibility study itself to ascertain for how long an experimental closures programme would need to continue for reliable determination of the impact of fishing in the near vicinity of island colonies on penguin reproductive success. Use of the Peng/B4 approach indicates that this period is appreciably lengthened if data for analyses are to be restricted to the years from 2008 onwards only, and particularly so if closure replaces catch as a covariate.



Department of Mathematics and Applied Mathematics, University of Cape Town