Vaughan and Prager: Decline in abundance of Pagrus pagnis off the southeastern United States 



365 



Table 6 



Static yield per recruit (\TR) and spawning potential ratio (SPR) of red porgy off southeastern United States. Estimates are based 

 on mean age-specific fishing mortahty rates from calibrated VPA on primary or alternative catch matrices (see text), are made 

 under several assumptions about natural mortality rate, M. and treatment of missing values in hook-and-line abundance index, 

 and use selectivity for most recent time period, ( 1992-961. 



Spawning potential ratio 



Analysis 



YPR(gl 



Total 



Female 



Eggs 



Male 



Percentage 



of males 



in relation 



to no fishing' 



Primary (MARMAP) catch matrix 

 (A/=0.28) 



1972-78 



1982-86 



1992-96 



(A/=0.28, zeroes in index treated as missing) 

 1972-78 

 1982-86 

 1992-96 



(M=0.20) 

 1972-78 

 1982-86 

 1992-96 



(A/=0..35) 

 1972-78 

 1982-86 

 1992-96 



Alternate catch matrix 

 (M=0.28) 



1972-78 



1982-86 



1992-96 



Percent relative reduction in numbers of mature males between fished and unfished conditions; i.e. proportion of males under fished conditions 

 is .v'7 of proportion of males under unfished conditions. 



presented both sets of results, which are different, but not 

 remarkably so. 



Several issues arise from choosing as "primary" the 

 catch matrix developed from fishery-independent age data 

 and as "alternate" the matrix from fishery-dependent da- 

 ta. The choice was by necessity somewhat subjective, but 

 was based on the long-term ( 1979-94). continuous nature 

 of the fishery-independent data on age at size and its larg- 

 er sample sizes, in contrast to the intermittent age sam- 

 pling in the fishery, which required frequent interpolation 

 of age-length keys to construct the alternate catch matrix. 

 The different selectivity of the fishery-independent gear 

 should not significantly bias the primary catch matrix, but 

 fishery-independent samples at larger sizes were some- 

 times very small, making it necessary to pool data across 

 longer periods for the largest sizes. Parallel analyses were 

 conducted with the two catch matrices to determine sensi- 

 tivity of major results to the choice of matrices as primary 

 and alternate. 



Size selectivity of the gear can have much greater effect 

 in fitting growth models. Fishery-dependent sampling, by se- 

 lecting larger fish, may overestimate mean size at younger 

 ages, but estimate L^ more accurately, whereas fishery- 

 independent sampling, by selecting smaller fish, may un- 

 derestimate L^. In a simulation study, Goodyear (1995) 

 noted: ". . . samples drawn from size-selective gears or fish- 

 eries yield biased estimates of mean length at age. . . . 

 Even slight changes in sampling protocol can result in mis- 

 leading temporal shifts of estimates of size at age." This as- 

 pect of the differing selectivity between gears seems impor- 

 tant to us, and it can bias estimation of growth models. As a 

 result, we cannot say whether estimated temporal differences 

 in size at older ages result from changes in growth patterns 

 in response to long-term overexploitation (Harris and McGov- 

 em, 1997) or from a change in sampling-gear selecti\'ity in the 

 fishery-independent sampling gear (Potts et al., 1998). 



Despite questions about size at age over time, patterns 

 of population benchmarks and stock status estimated from 



