fishmethods (version 1.13-1)

growth_LEP: A flexible maximum likelihood approach for fitting growth curves to tag-recapture data

Description

Estimation of von Bertanffy growth parameters from tag-recapture data (Laslett et al. 2002)

Usage

growth_LEP(l1=NULL,l2=NULL,dt=NULL,measurer = NULL,
           gmodel=1,use_parameter_boundaries=T,graphs=T,
           K_start_bounds=list(K1=NULL,lower_K1=0,upper_K1=Inf,
                 K2=NULL,lower_K2=0,upper_K2=Inf), 
           mu_Linf_start_bounds=list(mu_Linf=NULL,lower_mu_Linf=0,upper_mu_Linf=Inf,
                 sigma_mu_Linf=NULL, lower_sigma_mu_Linf=0,upper_sigma_mu_Linf=Inf),
           A_start_bounds=list(mean_age=NULL,lower_mean_age=0,upper_mean_age=Inf,
                 sigma_age=NULL,lower_sigma_age=0,upper_sigma_age=Inf),
           resid_error_start_bounds=list(sigma_resid=NULL,lower_sigma_resid=0,
                  upper_sigma_resid=Inf),
            measurer_error_start_bounds=list(use_measurer=F,sigma_measure=NULL,
                  lower_sigma_measure=0,upper_sigma_measure=Inf),
           vb_log_k_parms=list(alpha=NULL,lower_alpha=0,upper_alpha=Inf,fix_beta=T,
                 beta=NULL,lower_beta=0,upper_beta=Inf),
           nlminb.control=list(eval.max=10000,iter.max=10000,trace=10),
           tmb.control=list(maxit=10000,trace=FALSE))

Value

List containing the parameter_estimates, AIC, random effects A, the original predicted values, the original model residuals, results of the adjustment of A (Af, predicted and residuals for l1 and l2 used for plotting (see Laslett et al., 2004)), and convergence statistics (from nlminb; convergence=0 is successful convergence).

Arguments

l1

vector of release lengths of tagged fish.

l2

vector of recapture lengths.

dt

vector of time increment between tagging and recapture.

measurer

vector of integers specifying the recapturer type for each row: scientist = 0; fisherperson=1). Not required. Default = NULL.

gmodel

model to fit. 1 = standard von Bertalannfy growth model; 2 = VB log k model of Laslett et al. (2002). Default=1

use_parameter_boundaries

use parameter boundary values (T/F). Applies to all parameters estimated in the model. Default=T.

graphs

plot the observed values of l1 and l2 and the fitted growth curve model versus Af(the corrected measures of A; Lasett et al. 2004). Residuals plots (observed versus fitted) are also provided.

K_start_bounds

list of starting values (K1 and K2), lower(lower_K1 and lower_K2) and upper (upper_K1 and upper_K2) parameter boundaries for K1 and K2. If gmodel = 1, only K1 values are used.

mu_Linf_start_bounds

list of starting, lower and upper boundary values for estimated parameters mu_Linf and sigma_mu_Linf.

A_start_bounds

list of starting, lower and upper boundaries values for estimated mean_age, and starting, lower and upper boundaries values for sigma_age, both used to define the log-normal random effects distribution of A.

resid_error_start_bounds

list of starting, lower and upper boundary values for the estimated residual (measurement) error parameter sigma_resid

measurer_error_start_bounds

list of starting, lower and upper boundary values for the estimated measurer error parameter sigma_measure. Specify use_measurer=T to estimate the parameter. Default is F.

vb_log_k_parms

If gmodel=2, a list of starting, lower and upper values for estimated parameters alpha and beta. To fix beta to a constant value, specify fix_beta=T and enter a fixed value in beta

nlminb.control

controls for the nlminb function. See function nlminb for more information.

tmb.control

controls for the TMB function. See package TMB for more information.

Author

Gary A. Nelson, Massachusetts Division of Marine Fisheries

Details

The von Bertalanffy growth model or the VB log k model of Laslett et al. (2002) is fitted to tag release-capture lengths and times-at-a-large data following Laslett et al. (2002). The distribution of A is assumed log-normal. In addition, adjustments are made to A (age) following Laslett et al. (2004) to correct bias which permits simple graphical checking of the fitted growth curve model. If argument graph = TRUE, plots of l1 and l2 observed versus predicted, and residuals are created for checking model fit. Refer to Laslett et al. (2002) for more details.

References

Lasett, G. M., J. P. Eveson and T. Polacheck. 2002. A flexible maximum likelihood approach for fitting growth curves to tag-recapture data. Canadian Journal of Fisheries and Aquatic Sciences 59: 976-986.

Lasett, G. M., J. P. Eveson and T. Polacheck. 2004. Estimating the age at capture in capture-recapture studies of fish growth. Australian and New Zealand Journal of Statistics 46: 59-66.

See Also

growhamp grotag grotagplus

Examples

 if (FALSE) {
	data(lepdata)
  growth_LEP(l1=lepdata$l1,l2=lepdata$l2,dt=lepdata$dt,measurer=NULL,
               gmodel=1,use_parameter_boundaries=T,graphs=T,
               K_start_bounds=list(K1=0.2,lower_K1=0,upper_K1=Inf,K2=0.12,lower_K2=0,
                  upper_K2=Inf), 
               mu_Linf_start_bounds=list(mu_Linf=189.624,lower_mu_Linf=0,upper_mu_Linf=Inf,
                  sigma_mu_Linf=11.032,lower_sigma_mu_Linf=0, upper_sigma_mu_Linf=Inf),
               A_start_bounds=list(mean_age=1.76,lower_mean_age=0,upper_mean_age=Inf,
                   sigma_age=0.165,lower_sigma_age=0,upper_sigma_age=Inf),
               resid_error_start_bounds=list(sigma_resid=3.547,lower_sigma_resid=0,
                    upper_sigma_resid=Inf),
               measurer_error_start_bounds=list(use_measurer=F,sigma_measure=3.547,
                     lower_sigma_measure=0,upper_sigma_measure=Inf),
               vb_log_k_parms=list(alpha=2.955,lower_alpha=0,upper_alpha=Inf,fix_beta=T,
                      beta=30,lower_beta=0,upper_beta=30),
                      nlminb.control=list(eval.max=10000,iter.max=10000,trace=10),
                      tmb.control=list(maxit=10000,trace=FALSE))
 }