International Workshop on Climate and Oceanic Fisheries Rarotonga, Cook Islands 3-5 October 2011 Modelling the impact of climate change on Pacific tuna stocks and fisheries Patrick Lehodey & Inna Senina CLS, Space Oceanography Division, Marine Ecosystem Department, France John Hampton, Simon Nicols, Peter Williams Oceanic Fisheries Programme, SPC, New Caledonia
Tuna and fisheries in the tropical Pacific O. Skipjack Katsuwonus pelamis Thunnus albacares Page 2 Yellowfin Tropical sp. Bigeye Temperate sp. (Bluefin) Albacore Thunnus alalunga Thunnus obesus
Distribution of tuna catch -20 60-30 -40-50 -60-70 -80-90 -100-110 - 120-130 -140-150 -160-170 180 170 160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0-10 -20 60 50 50 40 40 30 20 WCPFC IATTC ICCAT 30 20 10 10 0 0-10 -20 IOTC -10-20 -30-30 -40-40 -50 yellowfin skipjack bigeye albacore bluefin -50-60 -20-30 -40-50 -60-70 -80-90 -100-110 - 120-130 -140-150 -160-170 180 170 160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0-10 -60-20 YFT BFT Total prises 90-97 8000 SKJ BET ALB Catch WCPFC in WCPO Tuna catch in the WCPC
ECOSYSTEM MODEL 3-D Models Ocean Biogeochem. SEAPODYM Prey model Lehodey et al 2010 Predator s population dynamics model Lehodey et al 2008 Senina et al 2008 Ocean Physics 3-D models Primary Production from satellites 4
ECOSYSTEM MODEL 3-D Models Ocean Biogeochem. SEAPODYM Prey model Lehodey et al 2010 Predator s population dynamics model Lehodey et al 2008 Senina et al 2008 Ocean Physics 3-D models -20 60 50 Primary Production from -30-40 -50-60 -70-80 -90 satellites - 100-110 - 120-130 -140-150 -160-170 180 170 160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0-10 -20 60 50 40 40 30 30 20 20 10 10 0 0-10 -10-20 -20-30 -30-40 -50-60 -20 Blue = skipjack; Red = bigeye (kindly from A. Fonteneau) -30-40 -50-60 -70-80 -90-100 -110-120 -130-140 - 150-160 - 170 180 170 160 150 140 130 120 110 100 90 80 70 FISHERIES 60 50 40 30 20 10 0-10 -40-50 -60-20 5
Modeling : Integrated approach SEAPODYM: Spatial Ecosystem And Populations Dynamics Model Modeling the interaction of oceanic variables and fishing impact with tuna biology and population dynamics (Lehodey et al 2008; Senina et al 2008; Lehodey et al 2010) Page 6 Age-structured Population Growth mortality by cohort Movement toward feeding grounds Mortality Feeding Habitat = Food abundance x accessibility (T,DO) IF MATURE Seasonal switch Spawning success Recruitment Movement toward spawning grounds Spawning Habitat = Food & T for larvae Absence of adults preys 15 parameters
Modeling : Integrated approach Currents Food ( necton) Page 7 Dissolved O2 Age-structured Population Growth mortality by cohort Movement toward feeding grounds Mortality Feeding Habitat = Food abundance x accessibility (T,DO) IF MATURE Seasonal switch Temperature Spawning success Recruitment Movement toward spawning grounds ph Spawning Habitat = Food & T for larvae Absence of adults preys Primary Production
Modeling : Integrated approach 1- Predict observed variability 2 Project Climate Change impact Page 8 Calibration Predicted catch Age-structured Population Growth mortality by cohort Movement toward feeding grounds Mortality Feeding Habitat = Food abundance x accessibility (T,DO) IF MATURE Seasonal switch Fisheries Observed effort/catch Spawning success Recruitment Movement toward spawning grounds Spawning Habitat = Food & T for larvae Absence of adults preys
Predicting observed variability (skipjack) 1st optimization experiments using outputs from: 1: Ocean-NPZD models (ESSIC) forced by NCEP reanalysis 2: Ocean-NPZD models (OPA- PISCES) forced by NCEP reanalysis Monthly catch data by fishery: purse-seine seine (x3; x6) long-line line (x0; x1) pole-and and-line (x3; x3) Quarterly length frequencies data for each fishery by 5x5, 5x10 or 10x20 degree squares 1/ ESSIC + data WCPO (ref. Senina et al. 2008) WCPO EPO 2/ OPA-PISCES + data WCPO & EPO
Predicting observed variability (skipjack) Small to medium size fish caught by surface gears Loop animation 1993-2000 Anecdotical (but informative) longline catch of large fish Predicted biomass and observed catch (% circles) Fishing events used in optimization : catch-effort : 174,221 size frequencies : 1,571 Predicted biomass and observed catch (% circles) In each cell of the grid (2 ) at each time step (month), the model predicts: i) the abundance of fish by size (age) ii) The catch by size for each fishery
Predicting observed variability (skipjack) Validation methods Spatial goodness-of-fit for catch and CPUE Exporting the model. Ex Indian Ocean R-squared goodness of fit Predicted vs. observed catch & LF Sub-tropical pole-and-line (solid lines predicted catch) 1997, II quarter Tropical purse-seine (free schools) 1997, IV quarter
Predicting observed variability (bigeye) 1st optimization experiments using outputs from: 1: Ocean-NPZD models (ESSIC) forced by NCEP reanalysis 2: Ocean-NPZD models (OPA- PISCES) forced by NCEP reanalysis EL NINO Larvae density 1 st semester 1998 LA NINA Larvae density 1 st semester 1999 Page 12 Total Pacific stock biomass SEAPODYM estimate Young and surface catch 1 st semester 1998 Young and surface catch 1 st semester 1999 SPC MULTIFAN-CL estimate for WCPFC Adult and longline catch 1 st semester 1998 Adult and longline catch 1 st semester 1999 Fishing events used in optimization : catch-effort : 362,424 size frequencies : 1,492
Projecting Climate Change impact Skipjack Bigeye Forcing: IPSL-CM4 & PISCES 2000 2000 Larvae density 2050 2050 2000 2000 Adults biomass 2050 2050 Changes in predicted distribution of larvae and adult biomass (g m -2 ) in the Pacific Ocean under IPCC A2 scenario, with historical fishing effort until 2000, then a projection based on average fishing effort for the period 1980-2000.
But conclusions from this 1 st study i) Even for the historical period, the ocean predicted by the climate model is quite different from the ocean forced by an atmospheric reanalysis (ie closer to observations) ii)we cannot simply transport the parameterisation obtained with historical reanalyses. We need to optimize the tuna parameters to this «new environment» using fishing data in the historical period iii) The climate model produced a bias in temperature in the midlatitude 1 T s 2 s 5 a Unit SKJ BET Parameters estimated by the ESSIC OPA- IPSL ESSIC NCEP IPSL model PISCES Habitats spawning o C 30.5 29.8 29.5 26.2 26.6 26.2 Optimum of the temperature function spawning o C 3.5* 2.05 3.5 0.82 2.19 0.9 Std. Err. of the temperature function 3 Larvae foodpredator trade-off coefficient - 3.67 2] 0.001* 0.63 2e-5] 0.34 Optimum of the 4 T a adult temperature function at maximum age o C 26* 23.4 25* 13 [5.00 [8 temperature function C 1.62 3.5] 1.1 2.16 3.99] 5] Std. Err. of the adult at maximum age 6 Ô Oxygen threshold ml l -1 3.86 1.5 5.46 0.46 0.74 1.02 Movements Maximum 8 V M sustainable speed B.L. s -1 1.3 1.13 0.95 0.32 0.97 0.1 * fixed; [ ] reaching min or max value boundary
Correction of temperature fields Page 15 1st layer temperature Jan 2001 5 C! WOA climatology IPSL - CM4 IPSL - CM4 corrected T c : Corrected temperature (IPSL corrected) T cl-obs : Temperature climatology from observation (WOA) T cl-mod :Temperature climatology from biased model (average IPSL 1900-2000) T anom : Temperature anomaly (difference between IPSL temperature and T cl-mod ) latitudinal temperature transects along 159 E in January, for the WOA climatology (dotted blue) and the original (dotted black) and corrected (red) IPSL-CM4 temperature outputs, averaged in the epipelagic layer defined for SEAPODYM
Projecting Climate Change impact (II) 1 T s 2 s 4 T a 5 a Unit SKJ Parameters estimated by the ESSIC OPA- IPSL IPSLc model PISCES Habitats spawning o C 30.5 29.8 29.5 30* Optimum of the temperature function spawning o C 3.5* 2.05 3.5 3 Std. Err. of the temperature function 3 Larvae food-predator - trade-off coefficient 3.67 2] 0.001* 0.498 temperature function C 26* 23.4 25* 20.1 Optimum of the adult at maximum age temperature function C 1.62 3.5] 1.1 4.9] Std. Err. of the adult at maximum age 6 Ô Oxygen threshold ml l -1 3.86 1.5 5.46 2.07 Movements Maximum 8 V M sustainable speed B.L. s -1 1.3 1.13 0.95 0.72 Note that the new optimization experiment also beneficiated of technical improvements in the likelihood approach New results are more realistic over the historical period though there are still several parameters that cannot be correctly estimated
2050 2000 Projecting Climate Change impact (II) SKIPJACK LARVAE 1st Exp with IPSL-CM4 2 nd Exp after T correction 2099 Threshold point: Mean SST in 10 N-10 S > Optimal spawning temperature + 1 Std Er.
Projecting Climate Change impact (II) Recent experimentation provided lethal temperature limits for yellowfin tuna J. Wexler, D. Margulies, V. Scholey (2011). Temperature and dissolved oxygen requirements for survival of yellowfin tuna, Thunnus albacares, larvae. Journal of Experimental Marine Biology and Ecology 404: 63 72
2000 Projecting Climate Change impact (II) SKIPJACK TOTAL BIOMASS (Both simulations used average 1990-2000 fishing effort to project fishing impact) 1st Exp with IPSL-CM4 2 nd Exp after T correction 1 2 2050 1 2 actual fishing effort average 1990-2000 fishing effort 2099 Under this fishing effort scenario, the stock biomass is predicted to be mainly driven by larval recruitment
Predicting observed variability (Albacore) July January September March November May Seasonal migration of adults
Young biomass Adult biomass Predicting observed variability (Albacore) Parameters optimization achieved with south Pacific albacore fishing data and extended to all Pacific domain: the north albacore stock is well predicted (with increasing biomass as observed). 1983 1993 2003
Optimization with IPSL-CM4 historical period SEAPODYM Multifan-CL (WCPFC assessment) 12 fisheries
Projecting Climate Change impact 1 T s 2 s Unit Alb Parameters estimated by the model NCEP IPSLc Habitats spawning o C 28.9 25.8 Optimum of the temperature function spawning o C 2.09 0.81 Std. Err. of the temperature function 3 Larvae food-predator - trade-off coefficient 5* 1.3 4 T a temperature function C 17.47 10.8 Optimum of the adult at maximum age 5 a temperature function C 2.23 4] Std. Err. of the adult at maximum age 6 Ô Oxygen threshold ml l -1 3.93 4.6 Movements Maximum 8 V M sustainable speed B.L. s -1 1.65* 0.59 Species Fork length (cm) Lower lethal O 2 levels (ml l -1 ) Skipjack 50 1.87 75 2.16 Yellowfin 50 1.14 75 1.77 Bigeye 50 0.40 75 0.50 Albacore 50 1.23 75 1.03 Particularly high O2 threshold value compared to observation Model bias in O2 as for T?
Projecting Climate Change impact Albacore Larvae density Total biomass 2050 2099 2000 Habitat of adults is shrinking in the tropical region. The south-east Pacific becomes more favorable
Projecting Climate Change impact 2000 Albacore Total biomass However increasing pco2 could lead to changes of C/N ratio (Oschlies et al. 2008). In that case DOC would decrease in the south east Pacific. 2099 2050
Projecting Climate Change impact Sensitivity to O2: Projection using Oxygen Climatology (blue lines) Larvae density Total biomass 2099 2050 2000
Conclusions (1/3) Understanding the impact of climate change on tuna populations is linked to our capacity to explain, model and predict the effect of natural variability for the historical period for which we have data. With realistic environmental forcing our model seems relatively robust. Parameter optimization approach provides a good framework to measure the progress and evaluate the outputs. But even with a limited number of parameters (15), it is not easy We need more data (tagging, larvae, etc ) Detail of skipjack biomass prediction (February 2002) with optimization using outputs from OPA- PISCES 2 forced by atmospheric reanalysis NCEP, and observed catch (circles) Same detail with optimization experiment using ocean reanalysis SODA 1 and Primary production derived from satellite
Conclusions (2/3) Ocean prediction from Climate models are not fully comparable for historical period to results achieved from ocean reanalysis. So a model parameterized from ocean observation or ocean reanalyses needs to be adjusted to the climate model environment. Climate model ensemble simulations could help to solve the problem of bias but ideally we would need climate model simulation with realistic (forced by data reanalyses) historical variability (ENSO, PDO, NAO) More work with biogeochemical models and experimentation is needed to provide reliable projections of dissolved O2 More detailed fishing data exist that would be helpful for optimization, but they are very difficult to access Last skipjack configuration based on ocean reanalysis SODA 1 and satellite PP
Conclusions (3/3) New simulations with temperature corrected forcing confirm the extension of skipjack population to subtropical and eastern Pacific Ocean, but with lower biomass and a decreasing trend after the 2070 s, driven by large extension of unfavourable equatorial spawning grounds. Application to albacore has been achieved and seems robust. However at the difference of skipjack, simulations for this species indicated high sensitivity to O2, for which the biogeochemical models are still unclear. Fishing impact will remain a key driver of tuna stocks, and conservation measures can take 10 to >20 years according to life span of the species to be fully effective. We need to act now to rebuild heavily exploited stocks (e.g. Bigeye). SST above 33-34 C will be likely a threshold for spawning of tropical tunas. Unless genetic plasticity allows for some adaptation? Healthy stock would increase the diversity of genes present in the population Several mechanisms needs to be fully evaluated (e.g., spawning migration) investigated in details (food competition) or added (bioenergetics).
This is only the beginning of the story... Thank you for your attention and for the invitation!