Large pelagic fisheries and climate variability: A comparative analysis of the spatio-temporal patterns in the tropical Indian Indian, Atlantic and Pacific Oceans Ana Corbineau, Tristan Rouyer, Bernard Cazelles, Alain Fonteneau, Jean-Marc Fromentin, Frédéric Ménard 4th Fishery GIS Symposium Rio de Janeiro, 25 – 29 August 2008 Brazil a Longline fishery in the Indian, Atlantic and Pacific Oceans • Industrial fisheries since the 50’s in the Indian, Atlantic and Pacific Oceans ¾ Catch variability in the short and long terms 7000 Catch (number x 1000) 6000 5000 4000 Atlantic 3000 2000 Indian 1000 Pacific 0 1957 1966 1975 1984 1993 2002 Catch for the principal large pelagic species exploited in the tropical Oceans (Japanese longliners, from 1957 to 2004, database: IOTC, ICCAT, IATTC/WCPFC) • Catch variability ¾ Populations dynamics Æ migration … ¾ Fisheries activity Æ overexploitation, targetting… ¾ Climate forcing Æ regime shift… Goals Characterizing and comparing the patterns of variability of catch of large pelagic fishes (LPF) in the tropical Indian, Atlantic and Pacific oceans 9 TUNA yellowfin (YFT) bigeye (BET) 9 SWORDFISH (SWO) 9 BLUE MARLIN (BUM) Highlight climate effects on LPF using the Southern Oscillation Index Species composition of longline Japanese catches Average ve ge fishing s g mapp of o Japanese J p ese Longline o g e ((1990-2002) 990 00 ) 50-20 -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 -2500 40 40 30 30 20 20 10 10 0 0 -10 -10 -20 -20 -30 -30 -40 -40 -50-220 -330 -440 -550 -660 -770 -880 -990 -1100 -1110 -1120 -1130 -1140 -1150 -1160 -1170 180 170 160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0 YFT LL Japan 1990-2002 SBTBFT BET SWO20000 YFT -110 -22-5 00 ALB Courtesy by Alain Fonteneau (2007) BET SWO ALB ¾ Yellowfin/ Bigeye ¾ Albacore Equatorial areas Gyres areas Biogeochemical Provinces (Longhurst, 1998) Biogeochemical provinces Æ based on water properties properties, primary production and nutrient fluxes Atlantic Indian Pacific NPTGW MONSW coastal gyres equatorial MONSE NPTGE ¾ 18 Provinces ¾ 4 Species NPTGW 50 0 72 2 catch time series (6 time series)) X 100 BET 150 5 SWO 4 3 2 1 0 2000 MONSE 2000 NPTGE 1992 0 1992 1984 1976 1968 800 1984 400 1960 YFT Catch (x1000) 1200 Catch (x100 00) 2000 1992 1984 1976 BET 1976 0 2000 8 1992 12 1984 BUM 1976 MONSW 1968 0 1968 1960 300 1968 4 1960 16 Ca atch (x1000) 100 x1000) Catch (x 2000 1992 200 1960 2000 1992 1984 1976 1968 1960 catch (x x1000) 1984 1976 1968 1960 Catch (x1000) Fisheries time series in the Biogeographical provinces 30 BUM 20 10 0 66 time series Methods – Wavelet Analysis Highlight periodic components in a time series (time-frequency decomposition) Fourier Spectrum The wavelet spectrum, contrary to Fourier analysis show frequencies evolution in time. For each time series (66) we calculated the wavelet spectrum in order to highlight periods where the interannual variability was significant. Methods – Classification (clustering) Based on the wavelet spectra of 2 temporal series: WPS WPS S P Ms x N Mp x N Covariance Matrix RSP = S x P t Single Value Decomposition RSP = U Γ Vt Distances (P and S) Distance Matrix Cluster CARB NSE MON MON NSE CARB AFR EA MONS SW WAR RM SPSG ARCH GW NPTG NPTG GW WAR RM NA ATR ARCH TGE NPT NPTG GW NPTG GW NPT TGE NPT TGE AU USE AU USE AU USE PEQ QD NA ATR ISSG WTRA WTRA ATL SA AU USE WTRA CARB ETRA NEC PN PN NEC PN NEC SPSG SPSG ETRA ETRA AFR EA ARCH ARCH RM WAR PEQ QD PEQ QD PEQ QD ARA AB ARA AB ARA AB CARB ISSG ISSG AFR EA SA ATL MON NSE MON NSE MONS SW MONS SW MONS SW EA AFR WTRA RM WAR PN NEC SPSG ATL SA NPT TGE YFT YFT BUM BUM YFT YFT BUM BUM SWO BUM YFT SWO BET BUM SWO BET SWO BUM YFT YFT BET SWO SWO SWO YFT BET SWO SWO BUM BUM SWO BUM YFT BET BUM BET SWO BET SWO SWO YFT BET BET YFT BUM BET BET SWO BUM BET BET SWO BET BET BET SWO BET SWO BUM BUM YFT YFT SWO YFT YFT BET AT IO IO AT IO IO PA PA PA PA PA PA AT PA PA PA PA PA PA PA PA PA PA AT IO AT AT AT PA AT AT AT PA PA PA PA PA AT AT IO PA PA PA PA PA PA IO IO IO AT IO IO IO AT IO IO IO IO IO IO AT PA PA PA AT PA Results I – Indian, Atlantic and Pacific Oceans Classification Classification - 66 catch Based time on theseries 66 WPS- YF FT BE ET SW WO BU UM Equatorial IO Tropical IO Gyre A AT Equatorial A AT CAR RB WAR RM ARC CH AUS SE NPTG GW NPTG GE PNE EC PEQ QD SPS SG Intrra provincce Inte er Provincce Intra sp pp Inter sp pp PA P IO AO A Inte er Ocea an Results I – Indian, Atlantic and Pacific Oceans Factors Modalities Especies P i Provinces Summary Structuration by province in each ocean The “species” species is the factor that less influence the structuration of the time series Relationships to Climate Variability ¾ Climatic Indice: SOI – Southern Oscillation Index (Trenberth, 1997) Difference between sea level pressure anomalies at Tahiti (17°S – 149°W) and Darwin (12°S – 130°E) Southern Oscillation Index Annual SOI from 1952 to 2004 Cross-spectrum SOI vs Catch time series 2004 2000 1996 1992 1988 1984 1980 1976 1972 1968 1964 1960 1956 1952 2.5 2 5 2 1.5 1 0.5 0 -0.5 -1 -1.5 -2 -2.5 2.5 Wavelet Cross – spectrum Detect D t t relationships l ti hi b between t 2 ttemporall series i cross-spectrum t CLUSTER CARB B CARB B NPTGW W CARB B ISSG G NPTGE E WARM M ARAB B ARAB B ARAB B NPTGW W SPSG G SPSG G WARM M NPTGE E PEQD D WTRA A SATL L NPTGE E WARM M PNEC C MONSW W NPTGE E MONSW W MONSW W MONSW W MONSE E NPTGW W MONSE E EAFR R MONSE E EAFR R CARB B ARCH H AUSE E WTRA A EAFR R NATR R NATR R SATL L ISSG G EAFR R ISSG G ETRA A SPSG G PEQD D SATL L MONSE E ETRA A ARCH H PEQD D ETRA A PEQD D ARCH H ARCH H WARM M AUSE E PNEC C PNEC C SPSG G PNEC C NPTGW W WTRA A WTRA A AUSE E AUSE E YFT BUM BUM SWO YFT YFT BUM BET BUM SWO YFT YFT BUM SWO SWO SWO YFT YFT BET YFT SWO YFT BUM BET SWO BUM SWO BET BET BUM BUM YFT BET YFT BET SWO SWO BET SWO BET BET BET SWO BET SWO YFT SWO YFT BUM BUM BUM SWO BET BET SWO BET SWO YFT BET BET BUM SWO BET BUM YFT BUM AT AT PA AT IO PA PA IO IO IO PA PA PA PA PA PA AT AT PA PA PA IO PA IO IO IO IO PA IO IO IO IO AT PA PA AT IO AT AT AT IO IO IO AT PA PA AT IO AT PA PA AT PA PA PA PA PA PA PA PA PA PA AT AT PA PA Results II – Relationships to Climate Variability Classification Based on the 66 WCS YF FT BE ET SW WO BU UM Equatorial IO Tropical IO Gyre A AT Equatorial A AT CAR RB WAR RM ARC CH AUS SE NPTG GW NPTG GE PNE EC PEQ QD SPS SG Intrra provincce Inte er Provincce Intra sp pp Inter sp pp PA P IO AO A Inte er Ocea an Results II – Relationships to Climate Variability Factors Modalities Especies P i Provinces Conclusions The biogeographic provinces is the main factor that structurate the variability of catch time series in each ocean The biogeographic provinces defined by Longhurst, caracterized by water properties, influence the top predators distribution (that represent the higher trophic level). Large scale climate variability (SOI) influences the patterns of variation of large pelagic fishes, but the geographic locations modulated its impact. Perspectives Does the limit of those provinces change during the time? The SIG could underline these changes and link the spatial distribution of populations to environmental fluctuations fluctuations. Use more accurate fishing data (i.e.1°x1° grid) Thanks for your attention Thanks to: CNRS financial support Frédéric Ménard (PhD supervisor) IRD – UR Thetis Bernard Cazelles, Jean Marc Fromentin, Tristan Rouyer and Alain Fonteneau CLUSTER - MCA Spectre S Spectre P Ms x N Mp x N Cross Covariance Cross-Covariance RSP = S x P t Matrice de co covariance ariance SVD RSP = U Γ Vt U Γ V Les colonnes de U (Ms x Ms)contiennent les vecteurs singuliers de S Matrice rectangulaire de Ms x Mp avec des valeurs sur la diagonale: valeurs singulières Les lignes de V (Mp x Mp)contiennent les vecteurs singuliers de P A = Ut x S B = Vt x P (A=K x N) N=temps (B=K x N) N=temps K nbre K= b sing.vectors i t K nbre K= b sing.vectors i t A= leading pattern (chaque ligne) B= leading pattern (chaque ligne) Matrice de Distance Comparaison entre les leading pattern et singular vector