Terceiro and Ross: Estimation of age from length data for Pomatomus saltatrix 



535 



oped by pooling data collected by the fisheries agen- 

 cies of several Atlantic coast states during 1982 to 

 1986 (Crecco et al. 1 ; Terceiro 2 ). 



There were concerns about the potential for biased 

 results in using the ASMFC pooled key, because of the 

 influence of interannual variation in growth, recruit- 

 ment, or mortality rate, given the broad temporal and 

 geographic scale over which the age-length data were 

 collected (Westrheim and Ricker, 1978). As a result, no 

 consensus was reached on the utility of the ASMFC 

 pooled key, catch-at-age data, or subsequent analyses 

 presented in Crecco et al. 1 and Terceiro 2 . 



Following implementation of the bluefish Fisheries 

 Management Plan in 1990, stock assessment work for 

 bluefish was renewed by ASMFC. A yield-per-recruit 

 analysis was developed to provide biological reference 

 points for the new assessment. Parameters of the von 

 Bertalanffy growth function L, = L, nl {l-e~ A '"~'o']), de- 

 rived from weighted mean calculated lengths at annu- 

 lus formation presented in NOAA ( 1989), were used as 

 an expedient alternative to the ASMFC pooled key to 

 age MRFSS-sample bluefish length-frequency data by 

 cohort slicing (i.e., solving for t in the von Bertalanffy 

 growth equation, given L, nf , L t , K, and t ). Length at 

 age, weight at age, and partial recruitment vectors for 

 the yield-per-recruit analysis were developed from this 

 version of the MRFSS length-age data. 



Published estimates of mean calculated bluefish 

 length at annulus formation exhibit considerable varia- 

 tion (Barger, 1990; Chiarella and Conover, 1990; Hamer, 

 1959; Lassiter, 1962; NOAA, 1989; Richards, 1976; 

 Wilk, 1977) (Table 1). The variation is likely due to 

 1) variation in growth over time, 2 1 sampling and avail- 

 ability bias, 3) varying influence of Lee's phenomenon 

 (Jones, 1958), 4) use of different conventions in fixing 

 the assumed birthdate of bluefish, and 5) differential 

 proportions of spring and spawned fish in the length 

 frequency samples collected by different investigators 

 (Wilk, 1977; Kendall and Walford, 1979; Chiarella and 

 Conover, 1990). Variation in mean lengths at age and 

 subsequently derived growth parameters, likely inher- 

 ent when data from several disparate sources are con- 

 sidered, could lead to biased results from cohort slic- 

 ing. The Eleventh NEFSC SAW 1 recommended testing 



'Crecco, V..M. Terceiro, and C. Moore. 1987. A stock assessment of 

 Atlantic coast bluefish, Pomatomus saltatrix. Special report pre- 

 pared for the Atlantic States Marine Fisheries Commission, Wash- 

 ington, D.C. 



-Terceiro. M. 1987. Status of Atlantic coast bluefish— 1987. U.S. 

 Dep. Commer. NOAA, Natl. Mar. Fish. Serv., Northeast Fish. Cent., 

 Woods Hole lab. Ref. Doc. 87-10, 54 p. 



'NEFSC. 1990. Report of the Eleventh NEFSC Stock Assessment 

 Workshop, Fall 1990; Woods Hole, MA. U.S. Dep. Commer., NOAA, 

 Natl. Mar. Fish. Serv., Northeast Fish. Sci. Cent. Ref. Doc. 90-09, 

 121 p. 



alternative methods, such as mixture of distributions 

 methods, and data sources in lieu of a time series of 

 age-length keys for the estimation of bluefish ages from 

 length-frequency data. 



This work compares results from the application of 



1) the cohort slicing method using growth parameters, 



2) a pooled age-length key applied in the standard 

 manner, 3) the iterated age-length key method (IALK) 

 as presented by Kimura and Chikuni ( 1987) and Hoenig 

 and Heisey (1987), and 4) MULTIFAN, a mixture of 

 distributions method (Fournier et al., 1990), to blue- 

 fish length-age data collected by the North Carolina 

 Division of Marine Fisheries (NCDMF) from 1986 to 

 1989. Our objective was to determine how these meth- 

 ods interpret the known-age test data, and suggest 

 whether any are an acceptable alternative to the re- 

 source-intensive method of developing fisheries spe- 

 cific age-length keys. 



Methods 



Data sources 



North Carolina bluefish length-age test data Bluefish 

 are landed in the commercial long haul seine (April- 

 October), pound net (May-October), and winter otter 

 trawl (October-May) fisheries in the estuarine and 

 coastal waters of North Carolina. The NCDMF has 

 collected biological sample data from these commer- 

 cial fisheries for bluefish since 1982. This time series 

 is the largest and most consistent body of bluefish 

 length-age data available. NCDMF bluefish length-age 

 data collected from spring 1986 to winter 1989 in = 

 1469, 13-84 cm, ages to 9, ages determined from 

 scale annuli), aggregated by calendar year, were used 

 as a test data set for comparing the performance of 

 alternative methods for estimating bluefish ages from 

 length data (Table 2, Figure 1). 



Investigators have not agreed on a birthdate con- 

 vention for Atlantic coast bluefish. NCDMF fisheries 

 biologists employ an unweighted average date of 1 

 June, between peak spawning of the spring ( 1 April ) 

 and summer ( 1 August) aggregations. This date roughly 

 corresponds with the time of annulus formation dur- 

 ing late May to mid-June. 



Bluefish recruit to North Carolina fisheries during 

 the summer fishing season. Bluefish with fork lengths 

 ranging from 15 to 32 cm that are captured before 1 

 June and possess scales that do not yet exhibit the 

 first annulus are typically classified as age 0. Inspec- 

 tion of modes in length-frequency data and the timing 

 of spawning suggests that the largest of these bluefish 

 originate from the spring cohort of the previous calen- 

 dar year, while the smallest are fish from the summer 



