408 



Fishery Bulletin 90(2). 1992 



(Ir 



,)/d, 



where If = TL at recapture, 



Im = TL at tagging, and 

 d = time in days between tagging and 

 recapture. 



A plot of mean daOy growth rate versus TL at tagging 

 for each species suggested asymptotic growth, since 

 growth rate generally declined as size-at-tagging in- 

 creased. Therefore, the von Bertalanffy growth model 

 was chosen as an empirically-based description of 

 growth (Moreau 1987) to which these tagging data 

 were fit. Of the currently available estimating pro- 

 cedures for using tag data to describe growth follow- 

 ing the von Bertalanffy growth equation, Fabens' 

 (1965) method provides the most accurate estimates 

 (Sundberg 1984). Data were analyzed using the Fishery 

 Science Application System (Saila et al. 1988) and 

 Fabens' (1965) iterated least-squares method for 

 estimating K and L^ in the von Bertalanffy growth 

 equation, 



Ir = Im + (L^-U)[l-exp(-Kd)] 



where 1^, Im . and d are defined as above, and 



L^ = the average TL in a population of fish 

 allowed to grow indefinitely following 

 the von Bertalanffy growth function, and 

 K = Brody's growth coefficient (per day). 



Before analysis, data were screened following pro- 

 cedures of Doerzbacher et al. (1988) to eliminate 

 outliers. Fish with growth rates >3 mm/day or <-3 

 mm/day were eliminated from the data set. The mean 

 ± 3 SD for the remaining data were then calculated, and 

 fish with growth rates outside this range were also 

 eliminated from the data set. Sufficient data were 

 avaUable to analyze tagged red drum separately by bay 

 system (except for Sabine Lake and Matagorda Bay). 

 Data for each of the other species were analyzed for 

 all tagging locations combined. 



The measure of effectiveness (P) used by Phares 

 (1980) which is similar to the multiple correlation coef- 

 ficient of linear regression (i?^) was used to determine 

 how well the von Bertalanffy model fit the data: 



P = (SSL-SSE)/SSL, 



where SSL is the sum of squares of (1^ -!„,), and SSE 

 is the residual sum of squares of the model, 



SSE = (1/-1,)2, 



where Ir' is the model's predicted length-at-recapture, 



and n is the number of recaptured tagged fish (after 

 data screening). The value of 1 r' for each tagged fish 

 was calculated following Parrack (1979): 



; = L. 



(Loo-lm)e-K(d)). 



Standard errors of each estimated K and L^ were 

 estimated using 10-fold cross-validation technique (a 

 form of jackknife resampling) described by Verbyla and 

 Litvaitis (1989). For each data set, the original data 

 were randomly partitioned into ten subsamples, nine 

 of which each contained 10% of the data, and one which 

 contained the remainder. The first subsample was ex- 

 cluded from the data set, and K and L^ were reesti- 

 mated. The first subsample was recombined with the 

 data set, and the second subsample was excluded, and 

 so on, until all 10 subsamples had been excluded. The 

 standard error of each parameter of the original data 

 set is approximated by the standard deviation of the 

 mean of the 10 separate estimates made after remov- 

 ing each subsample. 



Results and discussion 



Most of the data reported for recaptured tagged fish 

 during the 1960s were included in the analyzed data 

 set (i.e., few outliers were found). Of 1630 recaptured 

 fish, only 72 (4.4%) fish were excluded from the anal- 

 yses (Table 1). Red drum from the lower Laguna Madre 

 had the greatest proportion of outliers (13 of 69 fish). 

 However, the size range at tagging of the remaining 

 56 fish was comparable to the range of red drum tagged 

 in other bays. These results are similar to those of 

 Doerzbacher et al. (1988) for red drum and black drum, 

 and are supported by Ferguson et al. (1984) who 

 demonstrated that red drum lengths reported by sport- 

 fishermen were accurate. 



Mean daily growth rates of tagged fish during the 

 time between release and recapture were about 0.2 

 mm/day for all species, except red drum which aver- 

 aged about 0.4-0.7 mm/day (Table 1). These means 

 mainly represent the growth of smaller fish within each 

 range because the size data were skewed toward small 

 fish. For example, of 254 recaptured black drum, over 

 250 were <300mmTL at tagging and recapture. 

 However, the estimates of daily growth for black drum, 

 red drum, sheepshead, and spotted seatrout in this 

 study were within the ranges of those reported by 

 Colura et al. (1984), Cornelius (1984), Beckman et al. 

 (1988, 1990, 1991), Doerzbacher et al. (1988), Murphy 

 and Taylor (1989), Matlock (1990), and Green et al. 

 (1990). 



The estimated L^ for black drum, red drum, south- 



oo ' " 



ern flounder, and spotted seatrout tagged in Texas 



