UPDATED STANDARDIZED CPUE FOR ALBACORE CAUGHT BY JAPANESE LONGLINE FISHERY IN THE ATLANTIC OCEAN,

SCRS/00/160 UPDATED STANDARDIZED CPUE FOR ALBACORE CAUGHT BY JAPANESE LONGLINE FISHERY IN THE ATLANTIC OCEAN, 1975-1999 Koji Uosaki 1 SUMMARY CPUEs of Atlantic albacore (Thunnus alalunga) caught by the Japanese longline fishery from 1975 to1999 were standardized using two kinds of model with different error structures (log-normal and distribution) under the hypotheses of separate north and south stocks. The standardized CPUEs decreased for 1975-1999 in the north Atlantic, whereas the CPUEs fluctuated for 1975-87, and showed no clear trend in the south Atlantic. The standardized CPUEs between these two models were fairly similar in both the north and the south Atlantic. RÉSUMÉ La CPUE du germon atlantique (Thunnus alalunga) capturé par les palangriers japonais de 1975 à 1999 a été standardisée au moyen de deux types de modèles comportant une structure différente de l erreur (distribution log-normale et de ) selon l hypothèse de l existence de stocks distincts nord et sud. La CPUE standardisée a diminué en 1975-1999 dans l Atlantique nord, alors qu elle a fluctué en 1975-1987 sans montrer de tendance claire ldans l Atlantique sud. Les deux modèles donnaient une CPUE standardisée assez similaire dans l Atlantique nord et sud. RESUMEN Las CPUEs del atún blanco atlántico (Thunnus alalunga) capturado por la pesquería de palangre japonesa entre 1975 y 1999, fueron estandarizadas por medio de dos modelos con diferentes estructuras de error (lognormal y ) con la hipótesis de dos stock separados, al norte y al sur. Las CPUEs estandarizadas descendieron de 1975 a 1999 en el Atlántico norte, mientras que en el Atlántico sur las CPUEs fluctuaron entre 1975 y 1988 sin una tendencia clara. Las CPUEs estandarizadas de estos dos modelos eran bastante similares tanto en el norte como en el sur del Atlántico. KEYWORDS Atlantic, albacore, abundance index, standardization 1. INTRODUCTION For Atlantic albacore caught by Japanese longline fishery, the CPUE standardization using the General Linear Model (GLM) with the assumption that the error structure belongs to log-normal had been carried out (Uozumi 1994a). Uosaki (1996, 1997) carried out the standardization assuming error structure with to overcome a problem of zero CPUE treatment. In that study, standardized CPUEs were calculated separately for the 1959-69, 1969-75 and 1975-92 periods according to Uozumi (1994b), which classified three periods of "Target", "Transition" and "By-catch" in the history of the 1 National Research Institute of Far Seas Fisheries, 7-1, Orido 5-chome, Shimizu-shi, 424-8633 Japan

Japanese longline fishery on the Atlantic Ocean based on their fishery strategy. In the present study the standardized CPUE for the north and south stocks in the Atlantic was updated using those two models up to 1999 as the By-catch period. 2. MATERIALS AND METHODS 2.1. Data The data used in this study were obtained from the Japanese longline fishery statistics based on the logbooks and compiled at the National Research Institute of Far Seas Fisheries. The data was made by breaking the Task II catch-and-effort data down with gear configuration information i.e., number of hooks per basket, for 1975-1999. The data in 1999 was still preliminary. CPUE was defined as the number of fish caught per 1,000 hooks. Observations with less than 3,000 hooks were excluded from this analysis. In addition, observations which belong to the EEZ were also excluded to avoid inconsistency of data coverage during a long historic period. 2.2. Standardization In order to standardize CPUE of albacore, two different models were used, i.e., model with the assumption of log-normal error structure, and model with the assumption of error structure, using GLM and GENMOD procedures of SAS software package (Ver. 6.12), respectively. Year, season, subarea and gear configuration were incorporated as main effects. Quarter was used for fishing season. Nine subareas for each stock were defined based on the spatial distribution pattern of nominal CPUE of albacore (Fig. 1). The gear configuration was categorized to four levels (3-7, 8-11, 12-15 and 16-20 hooks between float). The model used was: ln(cpue+constant)=µ+y+q+a+g+interactions+e where µ : intercept Y : effect of year Q : effect of quarter A : effect of subarea G : effect of gear e : error term Interactions : any combinations of two-way interaction between main factors In order to include observations with fishing effort but no catch of albacore, a value which adds to CPUE (constant) were used. The constants 1.0 and 0.1 were selected for north and south stock, respectively, as approximate 10% of mean CPUE of albacore. The 10% of mean CPUE as the constant was recommended in Bluefin Species Group (ICCAT, 1997). The model is: E(C)=H exp(µ+y+q+a+g+interactions) where E(C): expectation of catch in number which belong to distribution. H: number of hooks used. In the analysis of GENMOD procedure, overdispersion was observed, therefore DSCALE option which estimates dispersion parameter was included. All statistics were adjusted according to the dispersion parameter appropriately. Model selection was according to the result of Uosaki (1997) based on Akaike s Information Criterion (AIC). The models were as follows: model north stock : ln(cpue+0.1)= µ+y+q+a+g+q*a+q*g south stock : ln(cpue+0.1)= µ+y+q+a+g+y*q+q*a+q*g+a*g model north stock : E(C)=H exp(µ+y+q+a+g+y*q+q*a+q*g) south stock : E(C)=H exp(µ+y+q+a+g+y*q+q*a+q*g+a*g)

As done at the ICCAT albacore species group meeting in 1998 (Madrid), standardized (predicted) CPUE for the model was calculated as follows; predicted CPUE=exp(LSM+1/2(MSE))-constant where LSM: least square mean of response variable for the effect of year MSE: mean square error of the model 3. RESULT AND DISCUSSION The ANOVA for the model using GLM procedure, and goodness of fit for the model using GENMOD procedure are shown in Tables 1 and 2. For the model, F value shows that the models are highly significant in the ANOVA (Table 1). The R-squares were 0.53 and 0.34 for north and south stocks, respectively. Whereas, for the model, scaled Pearson χ 2 in Table 2 show the model are highly significant as well. Although, the effect of gear in the south stock (χ 2 =0.37) were not significant, interactions which include effect of gear (Q*G and A*G) were significant. The scaled CPUEs which were adjusted to 1.0 in 1975, for 1975-1999 are shown in Fig 1, and those for 1959-1999 are shown in Table 3 for the north, Table 4 for the south stock. For the north stock there is no substantial difference in the CPUE trend between models for 1975-99. The CPUEs declined from 1.0 to around 0.2 for the period. For the south stock, there is no substantial difference in the CPUE trend between the models for 1975-99, as well as the north stock. The CPUEs fluctuated and showed no clear trend for 1975-88. And then, the CPUEs were, however, stable after 1989. LITERATURE CITED ICCAT. 1997. Report of the bluefin tuna methodology session. ICCAT Col. Vol. Sci. Pap. XLVI (1); pp.187-212. UOSAKI, K. 1996. Updated standardized CPUE for albacore caught by Japanese longline fishery in the Atlantic Ocean. ICCAT Col. Vol. Sci. Pap. XLIII; pp.311-317. UOSAKI K. 1997. Updated standardized CPUE for albacore caught by Japanese longliner in the Atlantic Ocean, 1959-1995. ICCAT Col. Vol. Sci. Pap. XLVI (3); pp.95-100. UOZUMI, Y. 1994a. The historical trend of standardized CPUE for albacore caught by Japanese longline fishery in the Atlantic Ocean. ICCAT Col. Vol. Sci. Pap. XLIII; pp.261-267. UOZUMI, Y. 1994b. A historical review of Japanese longline fishery and albacore catch in the Atlantic Ocean. ICCAT Col. Vol. Sci. Pap. XLIII; pp.163-170. UOZUMI, Y. 1997. Recent Status of the Japanese longline fishery in the Atlantic Ocean laying stress on albacore catch. ICCAT Col. Vol. Sci. Pap. XLVI (3); pp.101-103.

Table 1. ANOVA for the model (lognormal error structure assumption) in the north and south Atlantic. North Atlantic South Atlantic Source DF SS Mean Sq. F Value Pr > F Source DF SS Mean Sq. F Value Pr > F Model 71 13669.33 192.53 161.18 0.0001 Model 167 9710.27 58.15 48.00 0.0001 Error 11307 13506.17 1.19 Error 15959 19332.35 1.21 Corr. Tot. 11378 27175.50 Corr. Tot. 16126 29042.63 R-square= 0.503 C.V.= -103.52 R-square= 0.334 C.V.= -80.88 Source DF Type III SS Mean Sq. F Value Pr > F Source DF Type III SS Mean Sq. F Value Pr > F Y 24 1034.06 43.09 36.07 0.0001 Y 24 501.08 20.88 17.24 0.0001 Q 3 46.70 15.57 13.03 0.0001 Q 3 86.03 28.68 23.67 0.0001 A 8 5429.00 678.62 568.13 0.0001 A 8 2883.08 360.38 297.50 0.0001 G 3 14.33 4.78 4.00 0.0074 G 3 21.80 7.27 6.00 0.0004 Q*A 24 406.90 16.95 14.19 0.0001 Y*Q 72 531.97 7.39 6.10 0.0001 Q*G 9 72.03 8.00 6.70 0.0001 Q*A 24 1191.12 49.63 40.97 0.0001 A*G 9 115.77 12.86 10.62 0.0001 Q*G 24 478.19 19.92 16.45 0.0001 Table 2. Goodness of fit for the model ( error structure assumption) in the north and south Atlantic. North Atlantic South Atlantic Source NDF DDF F Pr>F ChiSquare Pr>Chi Source NDF DDF F Pr>F ChiSquare Pr>Chi Y 24 11272 67.83 0.0001 1628.01 0.0001 Y 24 15975 32.56 0.0001 781.35 0.0001 Q 3 11272 129.10 0.0001 387.31 0.0001 Q 3 15975 21.40 0.0001 64.20 0.0001 A 8 11272 838.57 0.0001 6708.57 0.0001 A 8 15975 193.57 0.0001 1548.55 0.0001 G 3 11272 6.48 0.0002 19.43 0.0002 G 3 15975 1.41 0.2380 4.23 0.2380 Y*Q 72 11272 16.09 0.0001 1158.32 0.0001 Y*Q 72 15975 20.56 0.0001 1480.33 0.0001 Q*A 24 11272 33.00 0.0001 791.95 0.0001 Q*A 24 15975 67.74 0.0001 1625.86 0.0001 Q*G 9 11272 7.48 0.0001 67.32 0.0001 Q*G 9 15975 8.16 0.0001 73.43 0.0001 A*G 24 15975 24.91 0.0001 597.89 0.0001

Table 3. Standardized CPUEs and scaled CPUEs with the confidence intervals for the model and those for the model in the north Atlantic. The value of scaled CPUE was adjusted to 1.0 in 1975. CPUEs for 1959-74 is revised from Uosaki (1996), by adding 1/2 mean square error to least square mean at obtaining index. Year Target Transition Baycatch (scaled) Target Transition Baycatch (scaled) Mean Mean Mean Mean Lower Upper Mean Mean Mean Mean 59 29,086 5,976 5,002 7,133 2,274 7,243 60 20,266 4,164 3,157 5,470 2,104 6,700 61 23,537 4,836 3,135 7,395 1,755 5,589 62 30,304 6,226 4,651 8,310 2,410 7,677 63 17,983 3,695 3,007 4,528 1,236 3,938 64 20,412 4,194 3,845 4,574 1,232 3,923 65 13,036 2,678 2,507 2,861 0,791 2,518 66 13,119 2,695 2,469 2,941 0,924 2,943 67 14,364 2,951 2,677 3,252 1,067 3,397 68 16,232 3,335 2,984 3,725 0,989 3,150 69 14,024 12,386 2,881 2,545 3,259 1,000 3,185 3,185 70 12,651 2,943 2,651 3,264 2,954 2,954 71 7,251 1,687 1,535 1,851 1,806 1,806 72 4,676 1,088 0,948 1,244 1,135 1,135 73 5,316 1,237 1,059 1,439 1,176 1,175 74 4,554 1,059 0,912 1,227 1,420 1,420 75 4,299 2,835 1,000 0,840 1,190 1,000 1,501 1,000 76 2,630 0,928 0,770 1,116 1,925 1,036 77 1,613 0,569 0,458 0,705 1,011 0,693 78 1,320 0,466 0,385 0,562 0,632 0,386 79 1,550 0,547 0,453 0,658 0,434 0,644 80 1,163 0,410 0,345 0,487 1,234 0,507 81 1,590 0,561 0,482 0,652 2,051 0,524 82 1,224 0,432 0,371 0,501 2,206 0,429 83 1,185 0,418 0,348 0,500 1,273 0,436 84 1,106 0,390 0,327 0,464 1,303 0,407 85 1,189 0,419 0,359 0,489 2,021 0,343 86 0,804 0,284 0,238 0,338 2,275 0,195 87 0,659 0,232 0,193 0,278 0,931 0,177 88 0,928 0,327 0,276 0,388 0,688 0,272 89 0,933 0,329 0,284 0,381 0,986 0,248 90 0,731 0,258 0,219 0,303 0,860 0,160 91 0,762 0,269 0,228 0,316 1,007 0,178 92 0,646 0,228 0,192 0,269 1,049 0,149 93 0,712 0,251 0,212 0,296 0,889 0,223 94 0,726 0,256 0,217 0,301 0,896 0,154 95 0,616 0,217 0,185 0,255 0,652 0,127 96 0,645 0,228 0,195 0,264 0,835 0,191 97 0,607 0,214 0,183 0,249 1,101 0,174 98 0,526 0,185 0,159 0,216 0,975 0,170 99 0,609 0,215 0,182 0,254 1,000 0,217

Table 4. Standardized CPUEs and scaled CPUEs with the confidence intervals for the model and those for the model in the south Atlantic. The value of scaled CPUE was adjusted to 1.0 in 1975. CPUEs for 1959-74 is revised from Uosaki (1996), by adding 1/2 mean square error to least square mean at obtaining index. Year Target Transition Baycatch (rescaled) Target Transition Baycatch (rescaled) Mean Mean Mean Mean Lower Upper Mean Mean Mean Mean 59 45,626 15,987 13,357 19,130 7,231 51,404 60 36,419 12,761 10,739 15,158 6,192 44,015 61 27,832 9,752 8,554 11,117 4,325 30,745 62 23,544 8,250 7,290 9,334 3,267 23,222 63 21,071 7,383 6,535 8,339 2,755 19,586 64 22,339 7,827 6,977 8,780 3,008 21,383 65 15,338 5,374 4,860 5,943 2,436 17,319 66 16,143 5,656 5,069 6,310 2,390 16,992 67 17,137 6,005 5,272 6,837 2,285 16,244 68 14,767 5,174 4,543 5,890 2,274 16,166 69 7,825 10,249 2,742 2,384 3,148 1,000 7,109 7,109 70 6,554 1,753 1,520 2,017 4,808 4,808 71 7,774 2,080 1,800 2,397 6,208 6,208 72 6,252 1,672 1,422 1,960 3,377 3,377 73 4,044 1,082 0,897 1,296 1,674 1,674 74 4,902 1,312 1,013 1,679 1,749 1,749 75 3,738 0,721 1,000 0,807 1,233 1,000 1,501 1,000 76 0,767 1,065 0,815 1,380 1,925 1,282 77 0,674 0,935 0,749 1,161 1,011 0,674 78 0,803 1,114 0,902 1,371 0,632 0,421 79 0,536 0,744 0,618 0,891 0,434 0,289 80 0,669 0,929 0,812 1,059 1,234 0,822 81 1,017 1,412 1,236 1,611 2,051 1,366 82 0,889 1,233 1,090 1,393 2,206 1,470 83 0,670 0,931 0,770 1,119 1,273 0,848 84 0,673 0,934 0,817 1,066 1,303 0,868 85 1,031 1,431 1,276 1,604 2,021 1,346 86 0,884 1,226 1,077 1,395 2,275 1,515 87 0,606 0,841 0,722 0,976 0,931 0,620 88 0,453 0,629 0,558 0,706 0,688 0,458 89 0,575 0,799 0,718 0,886 0,986 0,657 90 0,559 0,776 0,700 0,858 0,860 0,573 91 0,592 0,822 0,741 0,911 1,007 0,670 92 0,509 0,706 0,629 0,791 1,049 0,699 93 0,499 0,692 0,621 0,770 0,889 0,592 94 0,573 0,795 0,720 0,877 0,896 0,597 95 0,421 0,584 0,524 0,649 0,652 0,434 96 0,450 0,625 0,561 0,695 0,835 0,556 97 0,488 0,678 0,604 0,759 1,101 0,734 98 0,516 0,716 0,632 0,809 0,975 0,650 99 0,509 0,707 0,616 0,809 1,000 0,666

Fig.1. Subarea used in standardization of CPUE. Scaled CPUE 1,200 1,000 0,800 0,600 0,400 North stock m odel P oisson m odel 0,200 Fig. 2. Scaled standardized CPUEs of the north and south Atlantic using two different models. The value of CPUE was adjusted to 1.0 in 1975. Scaled CPUE 0,000 1,800 1,600 1,400 1,200 1,000 0,800 0,600 75 80 85 90 95 South stock m odel model 0,400 0,200 0,000 75 80 85 90 95