twoindex {MQMF}

twoindex has orange roughy catches with hypothetical cpue

Description

twoindex is a 35 x 4 data.frame of fishery data made up of smoothed real catches but two simulated indices of relative abundance. This data-set is designed to be used to illustrate the implementation of surplus production models when there are more than one time-series of relative abundance indices. This is not currently implemented in the book but is put here for use by readers should they wish to pursue this approach. The indices have been designed to generate a workable answer but also require the use of a penalty on harvest rates to avoid massively inflated harvest rates well above 1. Instead of using simpspm, spm, and negLL1, we need to use simpspmM, spmCE, and negLLM. The cpue series are hypothetical and have been designed to illustrate the use of penalty1 and the use of multiple indices of relative abundance. The real stock assessment uses acoustic survey indices and also uses many years of age composition data inside Stock Synthesis 3, not surprisingly the inclusion of real time-series of indices and of age-composition data leads to very different results.

Format

A data.frame of fishery data

year

the calender year of fishing

catch

the reported catch in tonnes

cpue1

the first index of relative abundance

cpue2

the second index of relative abundance

Subjects

• Surplus production models

• Dynamic model fitting

• -ve log-likelihoods

Source

Catches extracted from Table 4, page 11 of Haddon, M. (2017) Orange Roughy East (Hoplostethus atlanticus) stock assessment using data to 2016 Report to November 2017 SE RAG meeting. CSIRO, Oceans and Atmosphere, Australia. 51p. from https://www.afma.gov.au/fisheries-management/species/orange-roughy Catch data extended to 2019 using AFMA's catchwatch system.

Examples

 data(twoindex)
 fish <- as.matrix(twoindex)
 pars <- log(c(0.04,155000,0.4,0.3))
 bestSP <- nlm(f=negLLM,p=pars,funk=simpspmM,indat=fish,
             schaefer=TRUE,logobs=log(fish[,c("cpue1","cpue2")]),
             steptol=1e-06,harvpen=TRUE)
 namepar <- c("r", "K", "Binit","sigma")
 outfit(bestSP,parnames=namepar)  # best fitting estimates
 # if 'Either ~local min or steptol too small try 'steptol=1e-05'
 # plotprep(width=7,height=5,newdev=FALSE) # for external plot
 answer <- plotspmmod(bestSP$estimate,indat=fish,
                      plotprod=TRUE,maxy=3.4)