Description

SebastesSteepness is an R package for generating a prior distribution for recruitment compensation (measured by the steepness parameter of the Beverton-Holt stock-recruit function) for West Coast rockfishes, used in 6 assessment cyles occurring in 2007, 2009, 2011, 2013, 2015, and 2017.

To replicate previous results

SebastesSteepness is intended to improve reproducibility by compiling past data and analytical code. For example to replicate 2017 results using a short MCMC chain please run

# Load 2017 data
data(profile_2017, package="SebastesSteepness")

# Re-run results
Return = SebastesSteepness::calc_steepness_prior( LogLike_zp=-1*profile_2017$negloglike, h_z=profile_2017$steepness_value, StockWeight=profile_2017$stock_weight, RunDir=paste0(getwd(),"/"), NburnIn=1e5, Nsamp=1e5, Nthin=1e1 )

# visualize results
hist( Return$hpred, breaks=seq(0.2,1.0,by=0.04), plot=TRUE, main="Predictive distribution", xlab="Steepness", ylab="")

where the predictive mean for 2017 is r formatC(mean(Return$hpred),digits=3,format="f") and predictive standard deviation is r formatC(sd(Return$hpred),digits=3,format="f"). For a real run, please increase theNburnIn and Nsamp to >1,000,000.

To use for new data

SebastesSteepness can also be used to generate a steepness profile given future data sets. To do so, include the negative log-likelihood, stock weighting, and steepness values (correponding to rows of the negative-loglikelihood table) in similar format to the included data examples. Then use the same code as above.

To generate Type-C priors

Finally, analysts might want a prior that is developed without using data for a given species. To do so, remove the corrresponding column from a data set, e.g., profile_2017$negloglike, and the corresponding entry from profile_2017$stock_weight.

Description of package

If using the software or predictions please cite:

for further information regarding life-history predictions currently available in FishBase.



James-Thorson/SebastesSteepness documentation built on May 7, 2019, 10:19 a.m.