IOTC 4 WPTmT5 3 Rev_ Stock assessments of albacore (Thunnus alalunga) in the Indian Ocean by Age-Structured Production Model (ASPM) Tom Nishida /, Takayuki Matsumoto /, Kazuharu Iwasak / and Toshihide Kitakado 3/ / National Research Institute of Far Seas Fisheries, Fisheries Research Agency, Shimizu, Shizuoka, Japan / Environmental Simulation Laboratory (ESL), Kawagoe, Saitama, Japan 3/ Tokyo University of Marine Science and Technology, 5-7, Konan 4, Minato, Tokyo, 8-8477, Japan July, 4 Abstract Indian Ocean Albacore stock assessment was attempted by ASPM. Because of large uncertainties in extremely large number of drift gillnet CAA (catch-at-age) matrix (98-99) (max million fish) caused by the fundamental problem (no size data), we could not obtain the plausible and realistic results. To overcome this type of situation, we plan to develop additional option to the current ASPM software that can handle original size or CAS (catch-at-size) data, so that ASPM can conduct stock assessment when no or not enough size data situation (NB: when no size data, that option can use substituted size data from other areas and conduct assessments). Submitted to the IOTC 5th WPTmT (temperate tuna) working group meeting, July 8-3, 4, Busan, Korea Page of 9
IOTC 4 WPTmT5 3 Rev_. Introduction We attempted the stock assessment on albacore (Thunnus alalunga) (ALB) in the Indian Ocean based on AD Model Builder implemented Age-Structured Production Model (ASPM) (ver. 5) (Nishida et al) (4) using the data for 63 years from 95-. It is important to have a few stock assessments from simple (e.g. ASPIC), medium (e.g. ASPM, ASAP) to integrated models (e.g. SS3), so that we can compare under different structure of the dynamic models and confirm results. This is also important from another aspect, i.e., we will have more Line of Evidence in the Weight of Evidence approach if we have similar results in a few stock assessments. This means that we have more certainty (confident) in the stock status even there are large uncertainties in the data and models.. Input information To implement ASPM, we used ALB annual nominal catch by gear, standardized CPUE (STD_CPUE), CAA (catch-at-age) by gear and biological information. Below are descriptions of the data used in the ASPM runs.. stock structure In the Pacific and the Atlantic Ocean, two (north and south) stocks hypothesis has been used and stock assessments have been conducted for each stock. As for the Indian Ocean, it has a very small northern part, thus a single stock hypothesis has been applied, although there are some knowledge on intermingled areas with Pacific and Atlantic stock in its eastern and western end respectively. Nevertheless, we assume a single stock hypothesis for the 4 stock assessment as in the past.. Fleet We used 5 types of fleet (gears), i.e., tuna longline (Japan): LL(J), tuna longline (Taiwan,China) LL(T) and drift gillnet in high seas (GILL) by Taiwan,China, purse seine (PS) and others(oth), which is defined in the data sets produced by the IOTC Secretariat. OTH includes small scale surface fisheries such as troll, pole and lines, lines, gillnet (off shore) and other minor fisheries..3 Nominal catch by gear We used the nominal catch data by gear (fleet) from the IOTC Secretariat. Fig. shows the trends of catch by fleet type (in weight and number). Page of 9
IOTC 4 WPTmT5 3 Rev_ 5 ALB catch (weight) by gear ALB catch by gear (number) 4 3 95 954 958 96 966 97 974 978 98 986 99 994 998 6 tons Other Purse seine Driftnet LL(T) LL(J) million fish 8 6 4 OTH PS GILL LL(T) LL(J) 95 953 956 959 96 965 968 97 974 977 98 983 986 989 99 995 998 4 7 Fig. Trend of albacore tuna catch in the Indian Ocean by gear type in weight (left) and in number (right). (Source: IOTC Secretariat, 4) However, catch in number in GILL (98-99) is very high comparing to the one used in stock assessment (Fig. ). According to Miguel Herrera (IOTC data manager), this gap is caused by the following reason: There are no size samples available from GILL of Taiwan,China, thus substitutions need to apply to compute numbers. Then large discrepancy between and 4 is due to change in the substitution schemes, i.e., Secretariat used average ALB weight PS data (about 4 kg), while 4, the one in OTH (surface fisheries which are more in agreement with the sizes that driftnet fisheries catch in southern waters)(*) (3Kg). That is why number in 4 is 8 times higher than in (see below). (a) ALB Catch (no)() (GILL: ave PS) (b) IO ALB catch (no) (4) (GILL: ave wt OTH) million fish 8 6 4 OTH PS GILL LL(T) LL(J) million fish 8 6 4 95 953 956 959 96 965 968 97 974 977 98 983 986 989 99 995 998 4 7 OTH PS GILL LL(T) LL(J) 95 953 956 959 96 965 968 97 974 977 98 983 986 989 99 995 998 4 7 Fig. Indian Ocean Albacore catch by gear in number Number of GILL is estimated by IOTC Secretariat using (left) Average weight (4kg) of ALB caught by PS fisheries () (right) Average weight of ALB (3Kg) caught by OTH fisheries (4).4 CAA (GILL) CAA by gear is provided by Secretariat. However age compositions of GILL CAA (98-99) are constant in this period. As they vary by year, we assume that age composition (selectivity) of GILL are similar to OTH [see (*) above] then we estimated GILL CAA (Box ). Page 3 of 9
IOTC 4 WPTmT5 3 Rev_ Box Estimation of GLL CAA using annual age composition (selectivity patterns) of OTH () TWN GILL catch in tons 3 [B] TWN GILL catch (tons) (98-99) % 98 983 984 985 986 987 988 989 99 99 99 () Age comp of OTH gear [A] Age compostion of OTH 8% 6% 4% % a4 a3 a a a % 98 983 984 985 986 987 988 989 99 99 99 (3) GILL CAAw is computed by () x () assuming selectivity of OTH similar to GILL 3 [C]=[A]x[B] Catch by weight (CAAw) (tons) TWN GILL a4 a3 a a a 98 983 984 985 986 987 988 989 99 99 99 Then using average weights of ALB by age shown Table (below), which are estimated growth equation by Wells et al (3) and length-weight relationship by Penney (994) (see page ). Age (middle of year).5.5.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5.5.5.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5 weight (kg).8 3.7 6. 9.. 5. 8..8 3.4 5.7 7.8 9.7 3.4 3.8 34. 35. 36. 37. 37.7 38.3 (4) Estimate CAA of GILL by (3)/ average weight by age [D] = [C]/ave wt by age Catch by number (CAA) TWN GILL (million fish) 8 6 4 a4 a3 a a a 98 983 984 985 986 987 988 989 99 99 99 Page 4 of 9
IOTC 4 WPTmT5 3 Rev_.3 Plus group age The IOTC Secretariat provide CAA (age -+) by fleet. According to IOTC-4-WPTmT-6, plus group age are different among RFMOs (Oceans) (Fig. 5). We need to decide scientifically valid plus group. + group age used in the recent assessments by Ocean (RFMO) S Pacific (WCPFC) N Pacific (ISC) N Atlantic (ICCAT) Indian (IOTC) 5 5 5 Age Fig. 5 Plus group used in recent stock assessments in different tuna RFMOs The IOTC Secretariat provide CAA (age -+) by fleet and we explore optimum plus group using this CAA. Based on personal communications with three professors, Butterworth (Cape Town University), Hiramatsu (Tokyo University) and Shono (Kagoshima University), they suggest three rough clues to decide the optimum plus age group: (i) (ii) (iii) There will be biases in the stock assessment results if the population in plus group is more than % or less than % of the total population. If catch is included in the plus group in any year, it will be difficult to conduct assessments. If the age determination is difficult starting from some age (by otolith reading for example), that age and older ages should be pooled as the plus group. Then we investigated these three criteria to select plus group age. Regarding criterion (i), Fig 6 shows compositions of the plus group in the total catch, which suggested Age 5 or younger ages, satisfied (ii) % criteria. Regarding (ii), we investigated (zero) catch in CAA then we found years 95-95, there are catch in Age 5+ or younger plus age groups. This we will use the data from 6 years (95-). Compositions (%) of the plus group in the total cacth (no.) 5 4 3 a+ a9+ a8+ a7+ a6+ a5+ a4+ a3+ a+ Fig. 6 Compositions of the plus group in the total catch Page 5 of 9
IOTC 4 WPTmT5 3 Rev_ Table number of catch in plus group ages (+ to +). (95-95 include catches) a+ a9+ a8+ a7+ a6+ a5+ a4+ a3+ a+ 95 95 95 6 6 34 5 5 68 3 953 5 74 87 9 4 39 339 7 954 8 4 93 367 44 697 848 546 378 955 856 989 6 47 437 699 59 45 6598 956 885 8 588 86 56 459 446 67 9837 957 9 5 757 5 68 44 3747 5733 834 958 73 93 333 377 455 796 668 983 384 959 97 547 353 443 5354 888 7547 49 4495 96 734 7 38 3944 4737 77 6796 994 3855 96 495 886 53 398 4588 687 639 8856 86 96 67 75 789 358 46 6665 64 9597 3648 963 43 334 437 5544 649 9975 8964 385 579 964 634 45 755 3465 459 6743 6844 6 396 965 636 799 3553 437 54 759 6863 944 649 966 39 583 3397 43 566 797 6858 9877 683 967 984 59 7 69 367 487 577 776 38 968 4359 693 859 74 9 954 864 4966 758 969 437 54 667 835 9673 55 36 9 997 97 33 6354 68 65 3487 5 4376 687 646 97 66 859 348 6439 8839 7853 3673 375 38584 Regarding the criterion (i), we selected the growth curve by Well et al (3) that cover age up to 5 (Fig. 7) (for details, see page ), which suggested that age -5 are valid and CAA in other age need to be pooled. Then we checked (ii) and (iii) and age 5+ satisfied these two conditions. As a conclusion, we decide to use Age 5+ (plus group). Fig. 7 Page 6 of 9
IOTC 4 WPTmT5 3 Rev_.4 CPUE Table Eight standardized CPUE (STD_CPUE) in 6 (sub) areas (Figs 4 and 5) STD_CPUE ALL (Fig. 5) North (SS3) Area (IOTC core area) South Area a (TWN core) Area b (TWN core) Area 3 (JPN core) Japan () () Taiwan (3) (4) (5) (6) (7) (8) Korea (not available as of July ) N North area (SS3) S South area 3 (JPN core) 4 S South area (IOTC core) E 4 E 6 E 8 E E E 4 E N S South area b(twn) 4 S South area a(twn) E 4 E 6 E 8 E E E 4 E Fig 4 Seven core (sub) areas defined by Japan, Taiwan and IOTC Page 7 of 9
975 977 979 98 983 985 987 989 99 993 995 997 999 3 5 7 9 scaled (ave=) IOTC 4 WPTmT5 3 Rev_ Fig 5 Whole area for STD_CPUE (Taiwan LL) Fig. 6 shows eight STD_CPUE available in WPTmT5. Then, we compared relations between total catch vs. 8 STD_CPUE (Fig. 7). STD_CPUE (Japan) had the positive correlation while 6 STD_CPUE (Taiwan) negative. Among 6 TWN STD_CPUE, STD_CPUE in the whole area has the highest negative correlation. Hence we used it for ASPM. As ASPM use whole area, this STD_CPUE in the whole area is consistent to this approach. 3.5 Comparison of All STD_CPUE 3.5.5.5 J_S J_N T_ALL T_N T_S_IOTC T_Sa T_Sb T_S_JPN Fig. 6 Comparisons of 8 STD_CPUE series Page 8 of 9
STD_CPUE (scaled) IOTC 4 WPTmT5 3 Rev_ STD_CPUE (scaled) STD_CPUE Japan ( series) 4 Relation between Catch vs STD_CPUE Japan_South Relation between Catch vs STD_CPUE J_North STD_CPUE (sacled) 3 y = E-5x +.689 R² =.486 STD_CPUE (Scaled).5.5 y = E-5x +.684 R² =.69 3 4 5 catch (tons) 3 4 5 catch (tons) Taiwan (6 series) Relation between Catch vs STD_CPUE TWN ALL Relation between Catch vs STD_CPUE TWN IOTC.5.5.5 y = -E-5x +.6999 R² =.6 STD_CPUE (scaled).5.5.5 y = -E-5x +.5335 R² =.9 3 4 5 catch (tons) 3 4 5 catch (tons) STD_CPUE (scaled).5.5 Relation between Catch vs STD_CPUE TWN_North y = -E-5x +.58 R² =.996 STD_CPUR (scaled).5.5.5 Relation between Catch vs STD_CPUE TWN_South (JPN) y = -4E-5x +.64 R² =.366 3 4 5 catch (tons) 3 4 5 catch (tons) Relation between Catch vs STD_CPUE TWN a Relation between Catch vs STD_CPUE TWN S b.5 y = -E-5x +.4593 R² =.846.5 y = -E-5x +.65 R² =.55.5.5.5.5 3 4 5 cacth (tons) 3 4 5 catch (tons) Fig. 7 Relations between total ALB catch vs. 8 STD_CPUE series Page 9 of 9
M IOTC 4 WPTmT5 3 Rev_.5 Biological information In the ASPM, three types of age-specific biological inputs are needed, i.e., natural mortality-at-age (M), weights-at-age (beginning and mid-year) and proportion maturity-at-age. Based on the review of these parameters by Nishida et al (4) (IOTC-4-WPTmT5-6), we follow suggestions made by that paper. () Natural mortality vector (M) (Box ) Box Natural mortality (M) used in ASPM Age Base case M (Age )=.4 M (age 5+)=.7 (Lee and Liu, 99) M(age -4): propotions of abobe two Ms.4.364.383 3.94 4.566 5 or older.7.5 Proposed M by age and sensitivty for 4 assessment in the Indian Ocean.4.3. sensitivity (N Pacifc and N Atlantic) Indian (combined). 3 4 5 6 7 8 9 3 4 5 Age Page of 9
IOTC 4 WPTmT5 3 Rev_ () Beginning- and mid-year weights-at-age Beginning- and mid-year weights-at-age are computed as explained in Box 3 Box 3 Computation process of beginning- and mid-year weights-at-age as follow: (a) using the growth equation by Wells et al (3), size-at-age was calculated, (b) using the length-weight relationship, W = (.378-5 )*L 3.973 by Penney (994) (S Atlantic), weightat-age was calculated as shown in Table below. Growth curve by Wells et al (3) (Otolith)(N. Pacific) LW relation (S. Atlantic) 4 6 5 4 fork length (cm) 8 6 weight (kg) 3 4 3 4 5 6 7 8 9 3 4 5 Age 4 6 8 4 fork length (cm) Age (beginning of year) weight (kg) Age (middle of year) weight (kg)..5.8.7.5 3.7 4.9.5 6. 3 7.6 3.5 9. 4.5 4.5. 5 3.6 5.5 5. 6 6.6 6.5 8. 7 9.4 7.5.8 8. 8.5 3.4 9 4.6 9.5 5.7 6.8.5 7.8 8.8.5 9.7 3.6.5 3.4 3 3. 3.5 3.8 4 33.5 4.5 34. 5 34.7 5.5 35. 6 35.7 6.5 36. 7 36.6 7.5 37. 8 37.4 8.5 37.7 9 38. 9.5 38.3 38.6.5 38.9 Page of 9
IOTC 4 WPTmT5 3 Rev_ (3) Maturity-at-age We assume that the fecundity is proportional to maturity. We use maturity at-age based on biological data in the South Pacific Ocean by Farley et al () (Table ) and the estimation method by Hoyle (8). Table Maturity-at-age based on Farley () and Hoyle (8) Age 3 4 5 6 7 8 9 + Maturity at-age.9.47.75.88.94.97.99.99. Maturity-at-Age (S Pacific, ).8.6.4. 3 4 5 6 7 8 9 3 4 5 Fig. 8 Maturity-at-age (S Pacific) based on Hoyle (8) and Farley () Page of 9
3. ASPM runs (base case and sensitivity runs) IOTC 4 WPTmT5 3 Rev_ We use the base case as below. Catch and CAA : 95- Taiwan STD_CPUE (global) (98-) Hybrid age specific M Wells (Growth) and Penny (LW) CAA and Wt-at-age Farley s Maturity-at-age Steepness=.7 CV (CPUE)=. Sigma (SR)=.7 B=B95 But we could not get the conversions, then we explore further to search optimum parameters around this base case scenario. Then we found the most plausible option (Table 3). Then we run sensitivities and result are shown in Box 4. As a result, Base case produce the most plausible results which are depicted in Figs. 9- and the conclusion of the result is described in Box 5. M hybrid (.-.4) Table 3 Most plausible ASPM run around the base case scenario h (steepness) Sigma (SR) CPUE CV Weighting (CAA) SSB ( tons) Total likelihood R SSBmsy MSY ( tons) SSB/SSBmsy.67... 58-53.77.47 7 4.66.53 F/Fmsy Box 4 Results of 7 sensitivity runs Revised base case : the best scenario scenario base case sensitivity () sensitivity () sensitivity (3) sensitivity (4) sensitivity (5) sensitivity (6) sensitivity (7) plus group 5+ M hybrid (.-.4)..3.4 hybrid (.-.4)..3.4 h Weighting Sigma (SR) CPUE CV (steepness) (CAA) SSB ( tons) Total likelihood R SSBmsy MSY ( tons) SSB/SSBmsy.67... 58-53.77.47 7 4.66 not converged.67... 574-53.688.46 73 4.68 + not converged not plausible results F/Fmsy.53.5 3 Page 3 of 9
IOTC 4 WPTmT5 3 Rev_ Fig 9 Results of base case ASPM run () Fig Results of base case ASPM run () Page 4 of 9
IOTC 4 WPTmT5 3 Rev_ Fig Results of base case ASPM run (3) (projection) BOX 5 Page 5 of 9
IOTC 4 WPTmT5 3 Rev_ 4. Discussion (Box 6-8) Box 6 Box 7 Page 6 of 9
IOTC 4 WPTmT5 3 Rev_ Box 8 5 Future works (Box 9-) Box 9 Page 7 of 9
IOTC 4 WPTmT5 3 Rev_ Box 6. Summary (Box ) Box Page 8 of 9
IOTC 4 WPTmT5 3 Rev_ Acknowledgements We sincerely thank to Miguel Herrera, Data manager (IOTC) for providing the nominal catch and Catch-At-Age (CAA) data of albacore tuna in the Indian Ocean. We also appreciate Rebecca Rademeyer (University of Cape Town, South Africa) helped ASPM runs. References Beverton, R. J. H., and S. Holt. 957. On the dynamics of exploited fish populations. Reprinted in 993 by Chapman and Hall, London. 553 pp. ICCAT. 997. Report for biennial period 996-97. Part I (996), Vol.. Int. Int. Comm. Cons. Atl. Tunas. 4pp. Lee, L. K., Chang, F.C., Chen, C. Y, Wang, W. J. and Yeh, S. Y. () Standardized CPUE of Indian albacore (Thunnus alalunga) based on Taiwanese longline catch and effort statistics dating from 98 to. IOTC--WPTmT4 Matsumoto, T. () Review of Japanese longline fishery and its albacore catch in the Indian Ocean. IOTC WPTmT4??. 3pp. Matsumoto, T., Kitakado, T. and Okamoto, H. () Standardization of albacore CPUE by Japanese longline fishery in the Indian Ocean. IOTC WPTmT4. 6pp. Nishida, T. and Matsumoto, T. () Stock and risk assessments of albacore tuna (Thunnus alalunga) in the Indian Ocean by A Stock-Production Model Incorporating Covariates (ASPIC). IOTC--WPTmT3-7. 5pp. Nishida, T. and Wang, S-P. () Stock assessment of swordfish (Xiphias gladius) in the Indian Ocean by ASPIC (98-8). IOTC--WPB- Restrepo, V. 997. A stochastic implementation of an Age-structured Production model (ICCAT/ SCRS/97/59), 3pp. with Appendix (unlisted references will be provide upon request) Page 9 of 9