Estimating survival and exploitation rates from a tagging study on juvenile Atlantic halibut


Timing of releases and recoveries



Download 410.7 Kb.
Page2/5
Date28.03.2018
Size410.7 Kb.
#43749
1   2   3   4   5

3.2 Timing of releases and recoveries


The Brownie et al. (1985) implicitly assumes that releases take place a point in time and that recoveries occur in the year between releases, but makes no assumptions about the timing of recoveries. Hoenig et al. (1998) also assumes that releases occur at a point in time, but allows more flexibility about the timing of the fishing during the year between releases.
Table 4 presents a summary of the number of fish released by month for each year of the program. Most releases (67%) took place in June/July of each year, with all but a few releases falling on the shoulder months. Very few releases took place in 1996 and 2002 and 2003.
Table 5 summarizes the timing of the recoveries in the halibut program. Recaptures take place all year round with more recaptures taking place in June and July.
Both the Brownie et al. (1985) and Hoenig et al (1998) models requires the “recapture” year to follow the tagging point. Consequently the recovery year will be taken as 1 May of each year to 30 April of the subsequent year. As there were no recoveries in January – April 1995, there is no loss of information here. All releases within a year (including outside this interval) will be treated as releases for a given calendar year (i.e. they wll be treated as instantaneous releases) with the tagging “point” to be taken as 1 May of each year.
The resulting matrix of releases and recoveries is presented in Table 6. Only those recaptures that could be matched to releases are used and only those releases and recaptures with known dates are used. Because of the very few releases in 2002, no recoveries were observed; only a single recovery was observed from the 2003 tagging cohort. As halibut are long lived, recoveries can occur many years after release with some recoveries being found up to 11 years after release.

3.3 Model Form


Because of the very few numbers of releases in some years with virtually (or in fact) no recoveries observed, the full model with all parameters varying among years is too rich. Estimates will have very poor precision or may not even be identifiable. Consequently, the starting model to be considered will assume that the instantaneous natural mortality is constant over time (i.e. constant ).
There were virtually no releases from 2002 to 2005 and essentially one year of data for the 2006 release cohort, so only releases from 1995 to 2001 (inclusive) will be used. All recoveries up to 2007 will be used. Because a constant will be fit, it is still possible to estimate exploitation rates after 2001 which are normally non-identifiable in the fully time-dependent model (refer to Hoenig et al. 1998).
Recoveries occur through out the year, so the corresponding formulation by Hoenig et al. (1998) will be employed whose expected counts were presented in Table 4.
Normally, the initial tagging mortality or retention rate () is estimated by holding animals in pens after tagging. In this case, it is my understanding that these pen holding experiments were conducted and little or no tag loss or instantaneous mortality was observed, i.e. it will be assumed that . Double tagging may also be used to estimate long-term tag retention, but no double tags were
The tag-reporting rate () is usually estimated by comparing the rate of return of high value reward tags (where the reporting rate is assumed to be 100%) with the return rate of standard tags. In this study, reward tags were introduced only in 2006 and so very little data is available (Table 7). Based on Table 7, a very rough tag-reporting rate estimate is found as

However, this is based on such little data, that I have no confidence that this is a realistic estimate of the true tag-reporting rate from previous years.


In theory (Hoenig et al., 1998) it is possible to estimate the product -- however, they note (and this was confirmed via a small simulation study) that it is difficult to achieve meaningful levels of precision with an order of magnitude in the number of releases in each year.



Download 410.7 Kb.

Share with your friends:
1   2   3   4   5




The database is protected by copyright ©ininet.org 2024
send message

    Main page