SCTB16 Working Paper

SKJ–5

SEPODYM application to (Katsuwonus pelamis) in the Pacific Ocean: impact of ENSO on recruitment and population

Patrick Lehodey

Oceanic Fisheries Programme Secretariat of the Pacific Community Noumea, New Caledonia

June 2003

1

Introduction

Skipjack tuna are the most important tuna resource in terms of contribution by weight (~1.8 million t year-1) to the total world tuna catch. It is currently the 4th most productive and fished marine species in the World, after Peruvian anchoveta, and Atlantic herring. Most of the skipjack catch comes from the warm waters of the western and central Pacific Ocean (WCPO). Skipjack tuna catches in the WCPO have increased steadily since 1970, more than doubling during the 1980s. The catch has been relatively stable during the 1990s.

To explore the underlying mechanisms by which the environmental variability affects the pelagic ecosystem and tuna populations, a spatial environmental population dynamic model (SEPODYM) has been developed (Bertignac et al., 1998; Lehodey et al., 1998; Lehodey, 2001). SEPODYM is an Advection-Diffusion-Reaction Model (ADRM) at the ocean basin scale, combining a forage (prey) production model with an age structured population model of targeted (tuna predator) species. The model contains environmental and spatial components used to constrain the movement and the recruitment of tuna. All the spatial dynamics are described with an advection-diffusion equation (see previous references). Input data sets for the model are sea surface temperature, oceanic currents, primary production and dissolved oxygen. Preliminary simulations (Lehodey, 2001) were able to reproduce the observed ENSO-related spatio-temporal changes in the distribution of skipjack (Katsuwonus pelamis) population (Lehodey, 1997). However, the run represented a short time-series (1992-95) and was limited to the 20oN - 20oS equatorial region, whereas tuna stocks extend to sub-tropical and temperate oceanic regions. A new input data set simulated by a coupled physical-biogeochemical model (Chai et al., in press; Jiang et al., in press) is used to extend the analysis for skipjack to the Pacific basin over a longer time series (1960-1999). In addition, certain elements of the model parameterization have been improved.

Results of simulations are evaluated by comparing predicted and observed catch by fishery (at the spatial level), predicted and observed length frequencies distribution of the catch, and predicted biomass and recruitment with independent estimates provided by the statistical population dynamics model MULTIFAN-CL (Fournier et al., 1998; Hampton and Fournier, 2001a).

This paper summarises the last results concerning the application of SEPODYM to skipjack in the Pacific Ocean that have been described in a recent article (Lehodey, Chai and Hampton, in press).

Environmental variables

Temperature, currents and primary production are predicted from a coupled ocean-biogeochemical model. The biogeochemical model is driven by a physical model that is a full 3D ocean general circulation model (OGCM). The baseline of the physical model is the Modular Ocean Model (MOM)1. The model covers the entire Pacific (45oS-65oN, 100oE-70oW) except the Southern Ocean section (Li et al., 2001) for the period 1960-1999. The resolution is 2 deg. in longitude, 0.5 deg. in latitude near the equator (10oS to 10oN), and 2 deg. in latitude close to the northern and southern boundaries. There are 40 layers with 10 m resolution within the euphotic zone (120 m) and decreasing resolution with depth below. The surface forcing uses the Comprehensive Oceanic and

1 The Modular Ocean Model was developed by the Geophysical Fluid Dynamics Laboratory (GFDL)/ National Oceanic and Atmospheric Administration (NOAA), United States Department of Commerce

2 Atmospheric Data Set (COADS) monthly wind and heat flux. Details of the physical model and its surface forcing can be found in Li et al (2001). The biogeochemical model is a 10-component improved Nutrient---Detritus (NPZD) ecosystem model designed originally for the equatorial Pacific (Chai et al., 2002; Dugdale et al., 2002). The model includes both nitrate and silicate as major potential nutrients, two sizes of phytoplankton and zooplankton, nonliving detrital particles, as well as total CO2. Currents are averaged over the 0-30 m surface layer, new primary production is integrated over the euphotic zone (0-120 m), and interpolated with the sea surface temperature (SST) on a grid of one-degree square resolution to be used with SEPODYM.

SEPODYM (Spatial Environmental POpulations DYnamics Model)

The structure of the model is summarized on Figure 1. Documentation of most of the model features is provided in Bertignac et al. (1998), Lehodey et al. (1998), and Lehodey (2001). Additional information on recent developments was provided at the last Standing Committee on Tuna and Billfish (Lehodey, 2002) and is detailed in Lehodey et al. (in press). Below are described the main components and structure of the model.

Population structure The tuna population is age-structured to account for growth and gear selectivity. Skipjack population is described with 16 age-classes of 1 quarter, though the time step of computation has a higher time resolution (one month). The last age class is a “+ group” in which are accumulated the oldest individuals. Four components of the population are recorded: juveniles (1st cohort of age 1 quarter), young tuna (from age 2 quarter to age of 1st maturity), adults (all cohorts after age of maturity) and recruits (cohort at age of recruitment). For skipjack, age of first maturity is considered to be 9 months and age of recruitment is one quarter (juvenile group).

Age and Growth Growth parameters are those estimated from MULTIFAN-CL analyses (Hampton, 2002a).

Tuna Forage Given the large spectrum of prey organisms of tuna and the lack of information on their spatio- temporal dynamics, the tuna forage is modeled as a single population (Lehodey et al., 1998; Lehodey, 2001). With the extension of the model to the whole Pacific basin, the parameterization of the forage population has been revised to take into account regions as different as the warm pool (SST>28oC) and the subarctic gyre (Lehodey, 2002). The new parameterization gives a “mean age” or turnover time for F ranging from 4 months in the warmpool to 12-16 months in the subarctic region. The mortality of the forage population is spatially dependent of the density of tuna (Lehodey 2001).

Adult habitat index Ha

The adult habitat index Ha combines the spatial distribution of tuna forage biomass F with an oxygen function (Lehodey 2002b) and a temperature function (1) defined for each species (Lehodey, 2001). Ha is used to constraint the tuna movement and to modulate the natural mortality (see below).

3

-0.4 (SST- 22) θa = 1 / (1 + e ) (1)

Tuna movement Tuna movement is described with an advection-diffusion equation. Tuna larvae are passively transported by surface currents during their first quarter of life (as the forage). Then, young and adult tuna movements are constrained by the adult habitat index Ha. The advection term is proportional to the gradient of the adult habitat and a coefficient of proportionality Χο. An additional advective term due to the current can be selected between 0 (no effect) and 100%. In absence of information for this parameter, a medium value of 50% is used. To consider changes in tuna behaviour according to the quality of habitat, i.e. rapidly leaving a poor habitat region or conversely staying in very favourable habitat, a function is used to increase the diffusion (D) and the advection (Χ) at low values of habitat index. In addition, both D and Χ are proportional to the size of the . A simple linear relationship is used (Eq. 3a and 3b).

Da = D . La . [1 - (Ha / (g2 + Ha) ) ] (3a) Χ,a = Χο . La . [1 - (Ha / (g1 + Ha) )] (3b) with Da and Χ,a the diffusion and advection at age a, La the fish length (in m) at age a, Ha the adult habitat index, and g1 and g2, two coefficients constraining the shape of the curvature towards zero; Χο and D being maximal values of advection and diffusion coefficient respectively. The parameterization is scaled to be in agreement with the range of values estimated from tagging data by Sibert et al. (1999) and with realistic swimming behaviour (advection due to habitat gradient ranges between 0 and ~2 body lengths per second).

Spawning habitat index Hs and Recruitment

A spawning habitat index (Hs) is used to constrain the recruitment to environmental conditions. At each point of the grid, the number of recruits is given by the product of a recruitment scaling value (Rs) and the spawning habitat index. Rs is used to scale the total biomass to independent estimates from the MULTIFAN-CL model. The spawning temperature function (θs) is a Normal function o N(Top, σ) centered on an optimal value (Top= 30 C) with a variance σ = 2. Other environmental effects are investigated. Along with temperature and physical constraints (e.g. advection creating favourable zones of retention for larvae and juveniles), which are already considered in the model, effects of food availability (approximated by primary production P) and predation on larvae (approximated by the biomass of forage organism F). The equation (2) is used to calculate Hs:

(α LnΠ ) Hs = θs e (2) with θs the temperature function, Π the product of all other effects and α a coefficient allowing to scale the amplitude of these effects. Different combinations of effects have been tested: the temperature alone (simulation S1: α = 0; Hs = θs), the food effect (simulation S2: α = 0.5, and Π = P), the predation alone (simulation S3: α = 0.5, and Π = 1/F) and both predation and food effects (simulation S4: α = 0.5, and Π = P/F).

Mortality The total mortality rate (Z) is the sum of natural (M) and fishing mortality (f). The model is age- structured and natural mortality-at-age estimates are those estimated from MULTIFAN-CL.

4 However, these values are expressed in terms of the habitat assuming that the mortality rates are increasing in unfavourable habitat. The habitat index used is the spawning habitat (Hs) for the first age class and the adult habitat (Ha) for the other age classes. The fishing mortality is proportional to the fishing effort, the catchability coefficient of the fishery and the selectivity coefficient for the gear and age (size) considered.

Fisheries The skipjack fisheries are classified into pole-and-line and purse seine fisheries. The former are divided into tropical and Japanese domestic (north of 25oN) pole-and-line fisheries, and the latter are sub-classified by set type categories: unassociated schools, and schools associated with log, or FAD. Each fishery has one constant catchability coefficient and an age-based selectivity function (Lehodey, 2002). The selectivity functions are adjusted to obtain predicted length frequency distributions of catch in agreement with the observed distribution. Fishing effort of each fleet vary by month and in space, with a one degree square resolution except for the Philippine and Indonesia fleets that provide data aggregated by five degree square, and year. The catchability coefficients are scaled to obtain estimated catches at the same average level as observed catches.

Results

Recruitment and biomass Total recruitment and biomass for regions 1 to 6 defined in MULTIFAN-CL application (Fig. 2) are presented in Figure 3 and detail by region in Figure 4 and 5. Most of the recruitment of the skipjack population is predicted to occur in the WCPO, but with a large variability related to ENSO. This is illustrated in Figure 6 where the spatial distributions of skipjack recruitment predicted by SEPODYM are shown for two opposite phases of ENSO. The first simulation (S1) with the spawning habitat index only constrained by SST (α=0) produced relatively low fluctuations in the recruitment and biomass (Fig. 3) in contrast with the variability predicted from the statistical model. These fluctuations are linked to expansion or contraction of the spawning temperature habitat in relation with ENSO events. Additional effects (simulations S2 to S4) in the recruitment increase the fluctuations. Simulation S4 combining all effects provides the best result when comparing the estimates from both MULTIFAN-CL and SEPODYM (Fig 3) and the spatial correlations between predicted and observed catch (Table 1).

During El Niño, the recruitment increases both in the western and central Pacific. While the positive impact of El Niño events in the central region (region 6, Fig. 4) can be linked to the eastward extension of the warm waters of the warm pool, the main factor in the western region (region 5, Fig. 4) would be the primary production since water temperature remains high in this area and primary productivity increases during El Niño. This increase produces higher P/F ratio particularly since the forage development associated to the higher productivity in this region requires a few months. The situation is reversed during La Niña events. This mechanism can be seen as a good illustration of the Cushing’s “match/mismatch” hypothesis (Cushing 1990).

While recruitment and biomass have strong seasonal signals in the northern regions, fluctuations in others regions are dominated by the interannual variability related to ENSO. The impact of ENSO is the strongest in the central region (region 6), where both estimates of MULTIFAN-CL and SEPODYM for recruitment and biomass show relatively good convergence (Fig. 4 and 5). It is also

5 worth noting that the convergence between the two estimates improves when all regions are combined (Fig. 3). This may result from the simple description of the movement in MULTIFAN- CL from one region to another since the movement coefficients are constant over time in the present version. This can make difficult the distinction between recruitment and movement for explaining regional fluctuations in the biomass.

Table 1. Spatial monthly average correlation between predicted and observed skipjack catch (one degree square resolution) by fishery (PL= pole-and-line; PS = purse-seine) for simulation S1 to S4

No Fishery S1 S2 S3 S4 observations PL Japan 39,555 0.379 0.355 0.404 0.374 PL Tropical 111,139 0.713 0.700 0.729 0.728 PS 4,697 0.825 0.817 0.840 0.824 PS Unassociated 72,294 0.544 0.537 0.514 0.536 PS LOG 107,277 0.640 0.624 0.622 0.631 PS FAD 56,233 0.628 0.607 0.626 0.630

Catch by fishery Overall, the estimated length frequency distributions are in close agreement with the observations for pole-and-line, FAD and log fisheries, indicating that selectivity coefficients are reasonably defined (Figure 7 and 8). Total predicted catch by fishery is compared to observed catch in Figure 9. Given the very simple definition of the fisheries and a single constant catchability coefficient by fishery, it is not surprising to observe some discrepancies between the series. However, there is an overall good agreement on the level and fluctuations of catch in time.

Spatial correlations (Fig. 10) are fairly good with r-values ranging in average between 0.5 and 0.8, excepted for the domestic Japan pole-and-line fishery. This latter fishery is seasonal and associated to the warming of water in summer and the dynamic of the Kuroshio and Kuroshio extension. Clearly, the model has too broad a resolution to capture the dynamics of the Kuroshio and its warm- core rings that form excellent skipjack fishing grounds (Sugimoto and Tameishi, 1992).

Discussion Results presented in this paper with the new developments integrated in SEPODYM provide more realistic predictions for skipjack than in the previous versions. The changes in forage and movement parameterization together improve the spatial dynamics of the skipjack population, while the recruitment modelling integrates the main concepts proposed to explain its variability, i.e., the effect of advection, the match/mismatch hypothesis (P/F ratio), or the high mortality associated to starvation during the critical phase of larval and juvenile development (exponential increase of M at very low values of habitat). These new developments allowed the increase in the magnitude of recruitment variation to approach that estimated by the statistical model MULTIFAN-CL, while increasing in the same time the correlation between predicted and observed catch. Additional analyses are still needed to validate and refine this parameterization. For example, further

6 improvements in the spatial dynamics require higher spatial resolution in regions with strong gradients, such as the Kuroshio region and possibly the Coral and Tasman Seas. A better description of these fisheries may be also necessary, as they have been described on the basis of very broad criteria.

Predicted fields of physical–biogeochemical coupled models need to be investigated in details, particularly for the western Pacific, that is a key region for the skipjack recruitment. The capacity of the coupled model to simulate a realistic environment is obviously a major issue for future investigations on the tuna population dynamics. Detailed analyses of the recent period including the strong 1997-98 El Niño followed by the long 1998-2001 La Niña will be helpful as there is a complete coverage of the events by satellites providing SST and chlorophyll concentration. El Niño events appear to be favorable to skipjack recruitment. As the positive (negative) effect of El Niño (La Niña) events on the recruitment is propagated into the stock with a delay of a few months to two-three years, the last La Niña episode of 1998-2001 should lead to a decrease of the skipjack stock biomass in 2002-2003.

References Bertignac, M., Lehodey, P. and Hampton, J. (1998) A spatial population dynamics simulation model of tropical tunas using a habitat index based on environmental parameters. Fish. Oceanogr. 7(3/4):326-335. Chai, F., Dugdale, R.C., Peng, T.-H., Wilkerson, F.P. and Barber, R.T. (2002) One Dimensional Ecosystem Model of the Equatorial Pacific Upwelling System, Part I: Model Development and Silicon and Nitrogen Cycle. Deep Sea Res. II 49:2713-2745. Chai, F., Jiang, M., Barber, R.T., Dugdale, R.C., and Chao, Y. (in press) Interdecadal Variation of the Transition Zone Chlorophyll Front, A Physical-Biological Model Simulation between 1960 and 1990. Journal of Oceanography. Cushing, D. H. (1990) Plankton production and year-class strength in fish populations: an update of the match/mismatch hypothesis. Adv. Mar. Biol. 26:250-293. Dugdale, R.C., Barber, F., Chai, F., Peng, T.H. and Wilkerson, F.P. (2002) One dimensional ecosystem model of the equatorial Pacific upwelling system, Part II: Sensitivity analysis and comparison with JGOFS EqPac Data. Deep Sea Res. II 49 (13-14):2746-2762 Fournier, D.A., Hampton, J. and Sibert, J.R. (1998) MULTIFAN-CL: a length-based, age-structured model for fisheries stock assessment, with application to South Pacific albacore, Thunnus alalunga. Can. J. Fish. Aquat. Sci. 55(9):2105-2116. Hampton, J. (2002a) Stock assessment of skipjack tuna in the western and central Pacific Ocean. 15th SCTB, Hawaii, 22-27th July 2002, Oceanic Fisheries Programme, Secretariat of the Pacific Community, Noumea, New Caledonia. Working Paper SKJ-1:36 pp. http://www.spc.int/OceanFish/Html/SCTB/SCTB15/SKJ-1.pdf Hampton, J. (2002b) Stock assessment of yellowfin tuna in the western and central Pacific Ocean. 15th SCTB, Hawaii, 22-27th July 2002, Oceanic Fisheries Programme, Secretariat of the Pacific Community, Noumea, New Caledonia. Working Paper YFT-1:39 pp. http://www.spc.int/OceanFish/Html/SCTB/SCTB15/YFT-1.pdf Hampton, J. (2002c) Stock assessment of albacore tuna in the south Pacific Ocean. 15th SCTB, Hawaii, 22-27th July 2002, Oceanic Fisheries Programme, Secretariat of the Pacific Community, Noumea, New Caledonia. Working Paper ALB-1:32 pp. http://www.spc.int/OceanFish/Html/SCTB/SCTB15/ALB-1.pdf Hampton, J. and Fournier, D.A. (2001a) A spatially-disaggregated, length-based, age-structured population model of yellowfin tuna (Thunnus albacares) in the western and central Pacific Ocean. Mar. Freshw. Res. 52:93 -963. Hampton, J. and Fournier, D.A. (2001b). Recent enhancements to the MULTIFAN-CL software. 14th SCTB, Noumea, 9-16th August 2001, Oceanic Fisheries Programme, Secretariat of the Pacific Community, Noumea, New Caledonia. Working Paper MWG-1 Hampton, J. and Fournier, D.A. (2002) Recent enhancements to the MULTIFAN-CL Software. 15th SCTB, Hawaii, 22- 27th July 2002, Oceanic Fisheries Programme, Secretariat of the Pacific Community, Noumea, New Caledonia. Working paper MWG-2:5 pp. http://www.spc.int/OceanFish/Html/SCTB/SCTB15/MWG-2.pdf Jiang, M., Chai, F., Barber, R.T., Dugdale, R.C., Wilkerson, F., and Peng, T-H. (in press) A nitrate and silicate budget in the Equatorial Pacific Ocean: A coupled biological-physical model study. Deep Sea Res. II.

7 Lehodey, P., Bertignac, M., Hampton, J., Lewis, A. and Picaut, J. (1997) El Niño Southern Oscillation and tuna in the western Pacific. Nature 389:715-718. Lehodey, P., Andre, J.-M., Bertignac, M., Hampton, J., Stoens, A., Menkes, C., Memery, L. and Grima, N. (1998) Predicting skipjack tuna forage distributions in the equatorial Pacific using a coupled dynamical bio-geochemical model. Fish. Oceanogr. 7(3/4):317-325. Lehodey, P. (2001) The pelagic ecosystem of the tropical Pacific Ocean: Dynamic spatial modelling and biological consequences of ENSO. Prog. Oceanogr. 49:439-468. Lehodey, P. (2002) SEPODYM development and application to skipjack population and fisheries. 15th SCTB, Hawaii, 22-27th July 2002, Oceanic Fisheries Programme, Secretariat of the Pacific Community, Noumea, New Caledonia. Working paper SKJ-5:17 pp. http://www.spc.int/OceanFish/Html/SCTB/SCTB15/SKJ-5.pdf Lehodey P., Chai F., Hampton J. (in press). Modelling climate-related variability of tuna populations from a coupled ocean-biogeochemical-populations dynamics model. Fisheries Oceanography, 12(4/5): --. August 2003 Li, X., Chao, Y., McWilliams, J.C. and Fu, L-L (2001) A Comparison of Two Vertical-Mixing Schemes in a Pacific Ocean General Circulation Model. Journal of Climate 14 (7):1377-1398 Mantua, N. J., Hare, S. R., Zhang, Y., Wallace, J.M. and Francis, R.C. (1997) A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc. 78:1069-1079. Murtugudde, R.G., Signorini, S.R., Christian, J.R., Busalacchi A.J., McClain, C.R. and Picaut, J. (1999) Ocean color variability of the tropical Indo-Pacific Basin observed by SeaWiFS during 1997-1998. J.Geophys.Res. 104:18351- 18366. Sibert, J.R., Hampton, J., Fournier, D.A. and Bills, P.J. (1999) An advection-diffusion-reaction model for the estimation of fish movement parameters from tagging data, with application to skipjack tuna (Katsuwonus pelamis). Can. J. Fish. Aqu. Sci. 56:925-938. Sugimoto, T. and Tameishi H. (1992) Warm-core rings, streamers and their role on the fishing ground formation around Japan. Deep-Sea Res. 39:S183-S201.

8 A Spatial Environmental Population Dynamics Model (SEPODYM)

Fisheries Catchability m: natural mortality fishing effort y lit Tuna selectivity by age ta (space and time) or m m n Larvae-Juvenile g tio in m la sh u Fi p (age q1) po m d re tu c th Tuna (age q2) tru w s o Abundance - gr ge …. Predicted catch by age A (length), space, and time adult tuna Passive transport of larvae movement = Tuna (age qn) + Predicted biomass by diffusion + age, space, and time advection % Prey- gradient Ha and predator Passive transport and energy transfer (E) % fish size coupling

Primary Adult Spawning Currents Temp. Oxygen Forage production Habitat Habitat (u,v) λ (P) (T) (O) (F: Tr, ) (Ha) (Hs)

Figure 1: Schematic view of the structure of SEPODYM

1.0.106 Pole-and-line 120E 130E 140E 150E 160E 170E 180 170W 160W 150W 0.5.106 Purse seine 0.1.106 Other

40N 40N

1 30N 30N 2 3

20N 20N 4

10N 10N

0 0

10S 10S 6 20S 5 20S Skipjack tuna catch (mt) 1972-99

120E 130E 140E 150E 160E 170E 180 170W 160W 150W

Figure 2. Geographical areas defined for the skipjack MULTIFAN-CL application (from Hampton 2002a).

9

Total S1 Total S2 Total S3 Total S4 6.0E+06

4.5E+06 biomass (t)

3.0E+06 J-63 J-66 J-69 J-72 J-75 J-78 J-81 J-84 J-87 J-90 J-93 J-96 J-99

8.5E+06 Total biomass Mult-CL Total S4 7.0E+06

5.5E+06

biomass (t) 4.0E+06

2.5E+06 J-63 J-66 J-69 J-72 J-75 J-78 J-81 J-84 J-87 J-90 J-93 J-96 J-99

Figure 3. Skipjack total biomass in the WCPO (sum of 6 regions shown on Fig. 4) predicted from SEPODYM for the four simulations (S1 to S4) with the different effects tested for the recruitment, and comparison between simulation S4 (thin line) and the MULTIFAN-CL (thick curve) biomass estimate for the same regions.

10 1.5E+04 1 R skj 1 1.5E+04

1.0E+04 1.0E+04

5.0E+03 5.0E+03

0.0E+00 0.0E+00 J-63 J-65 J-67 J-69 J-71 J-73 J-75 J-77 J-79 J-81 J-83 J-85 J-87 J-89 J-91 J-93 J-95 J-97 J-99 J-01

3.0E+04 2 R skj 2 1E+04 8E+03 2.0E+04 6E+03

1.0E+04 4E+03 2E+03 0.0E+00 0E+00 J-63 J-65 J-67 J-69 J-71 J-73 J-75 J-77 J-79 J-81 J-83 J-85 J-87 J-89 J-91 J-93 J-95 J-97 J-99 J-01

5.0E+03 3 R skj 3 4E+03 4.0E+03 3E+03 3.0E+03 2E+03 2.0E+03 1.0E+03 1E+03 0.0E+00 0E+00 J-63 J-65 J-67 J-69 J-71 J-73 J-75 J-77 J-79 J-81 J-83 J-85 J-87 J-89 J-91 J-93 J-95 J-97 J-99 J-01

1.2E+05 4 R skj 4 6E+04 1.0E+05 5E+04 8.0E+04 4E+04 6.0E+04 3E+04 4.0E+04 2E+04 2.0E+04 1E+04 0.0E+00 0E+00 J-63 J-65 J-67 J-69 J-71 J-73 J-75 J-77 J-79 J-81 J-83 J-85 J-87 J-89 J-91 J-93 J-95 J-97 J-99 J-01

1.2E+05 5 R skj 5 4.0E+05 1.0E+05 3.5E+05 8.0E+04 6.0E+04 3.0E+05 4.0E+04 2.5E+05 2.0E+04 0.0E+00 2.0E+05 J-63 J-65 J-67 J-69 J-71 J-73 J-75 J-77 J-79 J-81 J-83 J-85 J-87 J-89 J-91 J-93 J-95 J-97 J-99 J-01

1.0E+05 6 R skj 6 3.0E+05 8.0E+04 2.5E+05 6.0E+04 2.0E+05 4.0E+04 1.5E+05 2.0E+04 1.0E+05 0.0E+00 5.0E+04

J-63 J-65 J-67 J-69 J-71 J-73 J-75 J-77 J-79 J-81 J-83 J-85 J-87 J-89 J-91 J-93 J-95 J-97 J-99 J-01

Figure 4. Skipjack recruitment by region (cf. Fig. 2) estimated by MULTIFAN CL (thick line) (Hampton 2002a) and by SEPODYM (thin line)

11 1 skj B tot. region 1 1.5E+05 1.5E+05

1.0E+05 1.0E+05

5.0E+04 5.0E+04

0.0E+00 0.0E+00 J-70 J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98 J-00

4.0E+05 2 skj B tot. region 2 1.0E+05 3.0E+05 8.0E+04 6.0E+04 2.0E+05 4.0E+04 1.0E+05 2.0E+04 0.0E+00 0.0E+00 J-70 J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98 J-00

6.0E+05 3 skj B tot. region 3 3.0E+05 5.0E+05 2.5E+05 4.0E+05 2.0E+05 3.0E+05 1.5E+05 2.0E+05 1.0E+05 1.0E+05 5.0E+04 0.0E+00 0.0E+00 J-70 J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98 J-00

1.0E+06 4 skj B tot. region 4 1.0E+06 8.0E+05 8.0E+05 6.0E+05 6.0E+05 4.0E+05 4.0E+05 2.0E+05 2.0E+05 0.0E+00 0.0E+00 J-70 J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98 J-00

4.0E+06 5 skj B tot. region 5 4.0E+06 3.5E+06 3.5E+06 3.0E+06 3.0E+06 2.5E+06 2.5E+06 2.0E+06 2.0E+06 1.5E+06 1.5E+06 1.0E+06 1.0E+06 J-70 J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98 J-00

4.0E+06 6 skj B tot. region 6 4.0E+06 3.5E+06 3.5E+06 3.0E+06 3.0E+06 2.5E+06 2.5E+06 2.0E+06 2.0E+06 1.5E+06 1.5E+06 1.0E+06 1.0E+06

J-63 J-65 J-67 J-69 J-71 J-73 J-75 J-77 J-79 J-81 J-83 J-85 J-87 J-89 J-91 J-93 J-95 J-97 J-99 J-01

Figure 5.Skipjack total biomass by region (cf. Fig. 2) estimated by MULTIFAN CL (thick line, left axis) (Hampton 2002) and by SEPODYM (thin line, right axis)

12

Figure 6. Spatial El Niño (average distribution Oct 82-March 83) and La Niña (average distribution Oct 88-March 89) skipjack recruitment. Biomass distribution of 1st skipjack age class (0 to 3 months) in tonnes per degree square.

13 1.0

t Selectivity by gear 0.8 Sel. PLINE 0.6 Sel. PSANI 0.4 Sel. PSUNA Sel. PSLOG 0.2 Sel. PSFAD Selectivity coefficien Selectivity 0.0 0 20406080100 Length (cm)

Figure 7 selectivity function by fishery

0.30 Pred. PL Obs. PL 0.30 0.30 Obs. PSUNA Pred. PSUNA 0.30 0.25 0.25 0.25 0.25 0.20 0.20 0.20 0.20 0.15 0.15 0.15 0.15 0.10 0.10 0.10 0.10 0.05 0.05 0.05 0.05 0.00 0.00 0.00 0.00 0 20406080100 020406080100 Fork length (cm) Fork length (cm)

0.30 Obs. PSLOG Pred. PSLOG 0.30 0.30 Obs. PSFAD Pred. PSFAD 0.30 0.25 0.25 0.25 0.25 0.20 0.20 0.20 0.20 0.15 0.15 0.15 0.15 0.10 0.10 0.10 0.10 0.05 0.05 0.05 0.05 0.00 0.00 0.00 0.00 0 20406080100 020406080100 Fork length (cm) Fork length (cm)

Figure 8 Observed and predicted length frequency distribution by fishery

14 ) 40000 obs c fishery 0 Pole and Line Japan 25000 30000 20000 15000 20000 10000 10000 5000

Skipjack Catch (t Catch Skipjack 0 0 J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98

) 30000 obs c fishery 1 Pole and Line Tropical 30000 25000 25000 20000 20000 15000 15000 10000 10000 5000 5000

Skipjack Catch (t 0 0 J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98

2000 obs c fishery 2 PS animals 2000 1500 1500 1000 1000 500 500 0 0 Skipjack Catch (t) Catch Skipjack J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98

80000 obs c fishery 3 PS Unassociated 80000 60000 60000 40000 40000 20000 20000 0 0 Skipjack Catch (t) Catch Skipjack J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98

60000 obs c fishery 4 PS Log 60000

40000 40000

20000 20000

0 0 Skipjack Catch (t) Catch Skipjack J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98

50000 obs c fishery 5 PS FAD 50000 40000 40000 30000 30000 20000 20000 10000 10000 0 0 Skipjack Catch (t) Catch Skipjack

J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98

Figure 9. Predicted and observed monthly catch of skipjack by fishery

15 1.00 PL Japan ma12 1500 n 0.75 1000 0.50 500 0.25 correlation r correlation 0.00 0 no Observations J-72 J-74 J-76 J-78 J-80 J-82 J-84 J-86 J-88 J-90 J-92 J-94 J-96 J-98

1.00 PL Tropical ma12 1500 0.75 1000 0.50 500 0.25 correlation r correlation

0.00 0 no observations J-71 J-73 J-75 J-77 J-79 J-81 J-83 J-85 J-87 J-89 J-91 J-93 J-95 J-97 J-99

1.00 400 0.75 0.50 200 PS Animal ma12 0.25 correlation r correlation

0.00 0 no observations J-80 J-81 J-82 J-83 J-84 J-85 J-86 J-87 J-88 J-89 J-90 J-91 J-92 J-93 J-94 J-95 J-96 J-97 J-98 J-99

1.00 PS Unas s . ma12 n 1500 0.75 1000 0.50 500 0.25 correlation r correlation

0.00 0 no observations J-80 J-81 J-82 J-83 J-84 J-85 J-86 J-87 J-88 J-89 J-90 J-91 J-92 J-93 J-94 J-95 J-96 J-97 J-98 J-99

PS LOG ma12 n 1.00 1500 0.75 1000 0.50 500 0.25 correlation r correlation

0.00 0 no observations J-71 J-73 J-75 J-77 J-79 J-81 J-83 J-85 J-87 J-89 J-91 J-93 J-95 J-97 J-99

1.00 PS FA D ma12 n 1500 0.75 1000 0.50 500 0.25 correlation r correlation

0.00 0 no observations

J-80 J-81 J-82 J-83 J-84 J-85 J-86 J-87 J-88 J-89 J-90 J-91 J-92 J-93 J-94 J-95 J-96 J-97 J-98 J-99

Figure 10. Spatial one-degree square monthly correlations (crosses = monthly values; thick line = 12-month moving average) between predicted and observed skipjack catch (bars = number of observations) for the fisheries defined in the application

16