modelGr.R

In evanchildress/perform:

model{
    #performance parameters
    maxAdd~dnorm(5,0.01)T(0,50)
    ctMax<-tOpt+maxAdd
    tOpt~dnorm(11,0.01)T(0,50)
    sigma~dunif(0,10)

    #derivative of the von Bert is linear, intercept and slope(with length) of hourly growth rate
#     beta1Scaled~dnorm(0,10)T(0,10)
#     beta1<-beta1Scaled/100
    beta1~dnorm(0,100)T(0,)
    beta2~dnorm(0,1000)T(,0)

#     beta2Scaled~dnorm(0,1)T(-1,0)
#     beta2<-beta2Scaled/1000

    #variation on growth rate at tOpt
    #epsScaled<-1.5
#     epsScaled~dunif(0,100)
#     eps<-epsScaled/1000

    eps~dunif(0,0.1)
    #eps<-0.00000000001
#     tauEpsScaled<-1/pow(epsScaled,2)
#     tauEps<-1/pow(eps,2)


    for(t in 1:nTimes){
      perf[t]<-ifelse(temp[t]>tOpt,1-(((temp[t])-tOpt)/(tOpt-ctMax))^2,
                    exp(-((temp[t]-tOpt)/(2*sigma))^2))
    }
    for(i in 1:nObs){

      # errScaled[i]~dnorm(0,tauEpsScaled)#variation on growth at tOpt
      # err[i]<-errScaled[i]/10000
      grMaxExpected[i]<-beta1+beta2*startLength[i]
      # grMax[i]<-grMaxExpected[i]+err[i] #von bert plus noise

      p[i]<-sum(perf[startTime[i]:endTime[i]]) #summed performance over growth period

      gr[i]~dnorm(grMaxExpected[i]*p[i],1/pow(eps*p[i],2))
      grExp[i]<-grMaxExpected[i]*p[i]
    }

  }