SCIENTIFIC COMMITTEE THIRTEENTH REGULAR SESSION Rarotonga, Cook Islands 9-17 August, 2017 Relatve abundance of yellowfn tuna for the purse sene and handlne fsheres operatng n the Phlppnes Moro Gulf (Regon 12) and Hgh Seas Pocket #1 1 WCPFC-SC13-2017/ SA-IP-07 NOAA Approved (25 September 2017) Keth Bgelow 2, Elane Garvlles 3, and Lala Emperua 4 1 PIFSC Workng Paper WP-17-003. Issued 02 August 2017. https://do.org/10.7289/v5/wp-pifsc-17-003 2 NOAA Fsheres Pacfc Islands Fsheres Scence Center, Honolulu, Hawa, USA 3 Bureau of Fsheres and Aquatc Resources - Natonal Fsheres Research and Development Insttute, BFAR/NFRDI, Phlppnes 4 Bureau of Fsheres and Aquatc Resources Regonal Offce 12, Phlppnes
Abstract Port samplng data were used to estmate effort, catch, CPUE, standardzed CPUE, and speces composton from the purse sene fshery operatng n the southern Phlppnes (Regon 12, SOCCSKSARGEN) and Hgh Seas Pocket #1 and the handlne fshery operatng n Regon 12. A quarterly standardzed CPUE ndex was produced for the purse sene (2005 to 2016) and handlne (2004 to 2016) fshery for use n the 2017 WCPFC yellowfn tuna assessment. Standardzed CPUE was estmated usng Generalzed Lnear Models (GLMs) and removng effects due to vessel and area (fshng ground). The ndex for the 2014 assessment used a GLM that predcted monthly CPUE wth year, month, and vessel effects. The current ndex predcted quarterly CPUE wth a YR:QTR, Area (fshng ground) and Vessel effects. A combned YR:QTR effect was estmated to be consstent wth other fshery CPUE standardzaton methods used n the assessment. There were 12 Area desgnatons n the database; however, Area was relatvely non-nformatve n the model as fshng trps were domnated by a few areas. 1 Introducton Sx tuna speces domnate Phlppne tuna landngs,.e. skpjack tuna (Katsuwonus pelams), yellowfn tuna (Thunnus albacares), bgeye tuna (T. obesus), eastern lttle tuna (Euthynnus affns), frgate tuna (Auxs thazard), and bullet tuna (A. roche). The most common gears used by the commercal sector for catchng these tuna speces are purse senes and rngnets, whle the muncpal fshers use hook-and-lne or handlne. All these gears are operated jontly wth fsh aggregatng devces (FAD), known as payao n the Phlppnes. Skpjack and yellowfn are found throughout the year n all Phlppne waters but are abundant n Moro Gulf, Sulu Sea, and Sulawes Sea off Mndanao Island. Large landngs of these speces occur n General Santos Cty where sx out of eght tuna canneres are located. The objectve of ths study was to use port samplng data to estmate effort, catch, CPUE, standardzed CPUE for yellowfn tuna n the purse sene fshery operatng n the southern Phlppnes (Regon 12, SOCCSKSARGEN) and Hgh Seas Pocket #1 and handlne fshery operatng n Regon 12. 2 Methods Natonal Stock Assessment Program (NSAP) protocols, samplng coverage rates, rasng factors for catch and effort, and qualty control Analyses on fshery performance and relatve abundance were based upon NSAP data collected at the Fshport Complex n General Santos Cty. The Fshport s the major tuna landng ste n Mndanao for handlne, purse sene, and rngnet fsheres. The NSAP samplng was ntated n 1997, though samplng was sparse for several years. Analyses consdered purse sene from 2005 to 2016 and handlne from 2004 to 2016. Wth WPEA- OFMP fundng, samplng of unloaded vessels to total vessels has especally mproved snce 2010. Port samplng data collecton pror to 2013 followed a NSAP protocol where 2
samplng was conducted every thrd day, regardless f the samplng day was on the weekend or a holday. Wth Phlppne purse seners ganng access to Hgh Seas Pocket #1 n 2013, the samplng protocol was altered to montor up to 100% unloadngs from vessel actvty n Hgh Seas Pocket #1 even f landngs occur on a non-samplng day. Samplng occurred where possble on all fshng boats (e.g. handlne, purse sene, rngnet, gllnet) that unloaded ther catch. Data were recorded on NSAP forms whch nclude the followng nformaton based on each fshng trp: A. Year B. Month C. Name of fshng ground D. Regon E. Landng center F. Date of samplng G. Gear H. Vessel name I. No. of fshng days (tme) of the actual fshng operaton J. Total catch by the vessel (no. of boxes/bañeras or weght) K. Sample weght of the catch L. Catch composton weght by speces (scentfc names) M. Name and sgnature of the NSAP samplers/enumerators Collected data are submtted monthly by the Project Leaders or Assstant Projects Leaders to the Natonal Fsheres Research and Development Insttute (NFRDI) offce. Monthly port samplng reports are entered and managed n the NSAP Database System. Two types of data were extracted from the NSAP Database (verson 5.1): 1) samplng of each vessel, hereafter referred to as trp sample, and 2) rased estmates for each month for trps, effort (days), and catch by speces, hereafter referred to as rased monthly estmates. Rased estmates are based on the samplng coverage whch s defned as the coverage of unloaded vessels on days that were sampled (.e. the proporton of sampled vessels unloaded catch to the total unloaded catch for days that were sampled) and the coverage of the samplng days n the month. Annual coverage pror to 2010 was 13% for the handlne fshery and mproved to 32% durng 2010 to 2016. Annual coverage was 6% for the purse sene fshery pror to 2010. Coverage mproved to 11% n the purse sene fshery durng 2010 to 2012 and ncreased agan to 45% durng 2013 to 2016 because more unloadngs were montored. Vessel name entres n the NSAP database were partcularly problematc due to multple spellngs for a unque vessel. Qualty control for purse sene vessels conssted of consoldatng obvous multple spellngs to a sngle vessel assgnment, whch resulted n the 357 purse sene vessels. Statstcal methods to estmate speces relatve abundance Trp sample data were used to estmate fshng effort and catch of ndvdual speces. Statstcal methods were used to estmate relatve abundance or standardzed CPUE by 3
removng effects due to vessel and fshng area. Generalzed Lnear Models (GLMs) were used to estmate relatve abundance. The GLM predcts mean catch (µ) usng three categorcal varables wth a log lnk as follows: log( µ ) = YR : QTR + Area + Vessel + log( Effort ) where YR:QTR s the mean local abundance or year and quarter effect, Area s the area effect, Vessel s the vessel effect (vessel name), and offset Effort s the number of days of the fshng trp. Snce a speces may have nstances of zero catch per quarter, a GLM wth a negatve bnomal dstrbuton was used to accommodate zero observatons. The GLMs were ft n R (R Development Core Team, 2016, verson 3.3.0 for Lnux) wth a MASS lbrary. GLMs were ntally ft wth the YR:QTR effect and then wth sequental addton of other explanatory varables. Model selecton was based on the Akake Informaton Crteron (AIC). Relatve abundance of each speces was calculated from the GLM results usng the predct.glm routne by exponentatng YR:QTR whle constranng other effects (Area and Vessel) to a sngle value. The GLM trends are normalzed to facltate comparson, such that the mean of the entre seres s a value of 1.0. The standardzed CPUE for the Phlppnes purse sene fshery (Bgelow et al. 2014) used n the 2014 assessment (Daves et al. 2014) used a GLM that had separate YR and Month effects: log( µ ) = Year + Month + Area + Vessel + log( Effort ) The YR and Month effects were predcted, and these effects were averaged for each quarter to correspond to the temporal resoluton of the 2014 assessment (Daves et al. 2014). The current use of a combned YR:QTR effect was estmated to be consstent wth other fshery CPUE standardzaton methods used n the 2017 assessment (Tremblay-Boyer et al. 2017). 3 Results and Conclusons Handlne fshery trends effort and catch Yellowfn tuna comprsed ~82.8% of the handlne catch durng 2004 to 2016 (Table 1) and typcally vares between 80 to 90% annually (NFRDI/BFAR 2012). The remander of the catch s composed of blue marln (Makara mazara, ~11.2%), bgeye tuna (~3.4%), albacore (Thunnus alalunga, ~1.3%) and other speces of ~1% (Table 1). Monthly trends n effort, catch, nomnal CPUE, and relatve abundance for the handlne fleet based n General Santos Cty are llustrated n Fgures 1 3. There are no estmates for months when samplng dd not occur; therefore, gaps exst n the effort, catch, nomnal CPUE, and relatve abundance tme-seres. Handlne effort averaged ~10,700 boat days per month and generally ranged from 5,000 to 15,000 days (Fgure 1). Effort durng 2006 to md-2009 was hgher than from md-2009 untl the end of 2013. Handlne effort averaged 18 boat days per trp, although there has been an ncrease over tme due to vessels travelng farther away from port n an attempt to obtan hgher catch rates and/or the use of larger vessels that can reman at sea for longer duratons. Handlne catch of yellowfn tuna averaged ~750 4
mt per month durng 2004 2016 wth low catches n years 2010, 2012, and 2015 (Fgure 2). Handlne speces trends nomnal CPUE and relatve abundance Monthly yellowfn tuna nomnal CPUE for the handlne fleet averaged 88 kgs per boat day and fluctuated from 30 to 170 kgs per boat day (Fgure 3). The CPUE ncreased from 2004 to 2007, declned precptously from 2008 untl the end of 2009, rebounded strongly n 2010, and was relatvely stable from 2011 to 2016. The GLM analyss consdered four models based on effects of: 1) YR:QTR (Fgure 3 black lne), 2) YR:QTR and Vessel (Fgure 3 blue lne), 3) YR:QTR and Area (Fgure 3 red lne), and 4) YR:QTR, Vessel, and Area (Fgure 3 grey lne). Results and dagnostcs ndcated that models based on YR:QTR and Vessel and YR:QTR, Vessel, and Area were statstcally preferred. Relatve trends were smlar for all models, wth YR:QTR and Area havng the most optmstc trend. Inspecton of the Area declaraton ndcated that Moro Gulf was declared for ~79% of the fshng areas from 2004 to 2016; therefore, the area effect s not too nformatve as an explanatory effect n the model. The trend based on YR:QTR and Vessel s consdered the most representatve to llustrate relatve abundance of yellowfn tuna for the handlne. In the comparson between nomnal yellowfn CPUE and relatve abundance, the relatve abundance trend has less varablty and generally follows the trend n nomnal CPUE. Whle the GLMs ncluded a Vessel effect, n realty the relatve abundance trend may be based because the analyss does not adequately quantfy effcency for each handlne vessel. The nomnal ncrease n CPUE for yellowfn tuna (Fgure 3) from 2004 to the end of 2008 may be related to ncreased vessel effcency, such as handlne vessels havng an ncreasng number of pakura or small pump boats whch were ntroduced n 2005. Thus the ncreasng CPUE and relatve abundance may n realty relate to vessels wth more pakura catchng more fsh per boat day. The standardzed CPUE trend from the 2014 and 2017 assessments s llustrated n Fgure 5. The trajectory among trends s smlar, though the 2014 trend has dfferent covarates related to tme (Year and Month; YR:QTR) estmated by the GLMs. Purse sene fshery trends effort, catch and nomnal CPUE Yellowfn tuna comprsed 15.8% of the purse sene catch from 2005 to 2016. The remander of the catch was composed of skpjack tuna (57.9%), mackerel scad (Decapterus macarellus, 9.0%), bullet tuna (Auxs roche, 8.9%), frgate tuna (Auxs thazard, 4.4%), bgeye tuna (1.6%), and other speces representng < 1% of the catch (Table 4). Monthly trends n rased effort and catch and nomnal CPUE for the purse sene fleet based n General Santos Cty are llustrated n Fgures 6 8. Purse sene effort averaged ~ 518 boat days per month (Table 2) and generally ranged from 100 to 1,500 days (Fgure 6). Effort durng 2005 to 2009 was slghtly hgher than effort n 2010 to 2012. There has been an ncrease n purse sene effort from 2013 to 2016 due to 5
re-openng of Hgh Seas Pocket #1 for a lmted number of Phlppne flagged purse sene vessels. Purse sene catch of yellowfn tuna averaged ~ 573 mt per month, and from 2010 to 2012, there was a declne n purse sene catches of yellowfn tuna (Fgure 7). Yellowfn nomnal CPUE n the purse sene fshery averaged 1.18 mt per day and was low durng 2010 and 2011 (Fgure 8). Purse sene fshery trends standardzed CPUE Model results of the GLM analyss are provded n Table 5. The hghest explanatory ablty and lowest AIC were for GLMs wth the ncluson of YR:QTR, Area, and Vessel effects. There were 12 Area desgnatons n the database; however, Area was relatvely nonnformatve n the model as the trps were domnated by three areas. Standardzed CPUE trends for the four models are llustrated n Fgure 9. Trends were consstent among the models from 2005 to 2014, and nomnal CPUE dverged from the other three models n 2014 and was more optmstc. A model based on YR:QTR and Vessel effects was chosen as the model for ncluson n the 2017 yellowfn tuna assessment (Tremblay-Boyer et al. 2017). The model based on YR:QTR, Area, and Vessel had a slghtly hgher explanatory ablty; however, there s an mbalance n the Area covarate as one area (Internatonal Waters) was not declared n the database pror to 2012, but was fshed thereafter. The standardzed CPUE trend from the 2014 and 2017 assessments s llustrated n Fgure 10. The trajectory among trends s smlar, though the 2014 trend has dfferent covarates related to tme (Year and Month; YR:QTR) estmated by the GLMs. 4 References Bgelow, K., E. Garvlles, and N. Barut. 2014. Relatve abundance of skpjack and yellowfn n the Moro Gulf (Phlppne Regon 12). WCPFC- SC10-2014/SA-WP-09, Majuro, Republc of the Marshall Islands, 6 14 August 2014. Daves, N., S. Harley, J. Hampton, and S. McKechne. 2014. Stock assessment of yellowfn tuna n the western and central Pacfc Ocean. WCPFC-SC10-2014/SA-WP-04, Majuro, Republc of the Marshall Islands, 6 14 August 2014. NFRDI and BFAR. 2012. Phlppnes tuna fsheres profle. Bureau of Fsheres and Aquatc Resources Natonal Fsheres Research and Development Insttute Republc of the Phlppnes and Western and Central Pacfc Fsheres Commsson. 84p. Tremblay-Boyer, L, S. McKechne, G. Pllng, and J. Hampton. 2017. Stock assessment of yellowfn tuna n the Western and Central Pacfc Ocean. WCPFC-SC13-2017/SA-WP- XX, Rarotonga, Cook Islands 9-17 August 2017. 6
Table 1. Catch and speces composton (%) estmated by NSAP for the handlne fshery (2004 to 2016) n Regon 12 (SOCCSKSARGEN) based on BFAR NFRDI montorng. Speces Catch Monthly catch mean (mt) Percent (%) (medan) Yellowfn tuna (Thunnus albacares ) 116,806.3 82.79% 768 (687) Blue marln (Makara mazara) 15,835.3 11.22% 108 (84) Bgeye tuna (Thunnus obesus) 4,769.7 3.38% 32( 26) Albacore (Thunnus alalunga ) 1,896.8 1.34% Salfsh (Istophorus platypterus) 696.2 0.49% Black marln (Makara ndca) 656.7 0.47% Swordfsh (Xphas gladus) 333.0 0.24% Moonfsh (Lamprs guttatus) 62.8 0.04% Other 11.9 <0.01% Total 141,091.3 100% Table 2. Mean operatonal and catch characterstcs for handlne (9,727 trps) and purse sene (2,359 trps) fsheres operatng n Regon 12 (SOCCSKSARGEN ) and Hgh Seas Pocket #1 based on BFAR NFRDI montorng. Handlne (2004 2016) Purse sene (2005 2016) Number of trps per month 588 113 Number of days per month 10,735 518 Days per trp 21.9 4.1 Catch (mt) per month 928 3,621 Catch (kgs) per day per vessel 74.7 7,734 Table 3. Results for Generalzed Lnear Models (GLMs) appled to yellowfn tuna n the handlne fshery (2004 to 2016) n Regon 12 (SOCCSKSARGEN ). The percent devance explaned s ((null devance-resdual devance)/null devance). Model selecton was based on the Akake Informaton Crtera (AIC). GLM Model Null devance Resdual devance AIC % devance explaned YR:QTR 11,958 11,293 165,573 5.5 YR:QTR+ Vessel 24,040 10,819 164,972 55.0 YR:QTR+ Area 12,265 11,271 165,313 8.1 YR:QTR+ Area+Vessel 24,327 10,818 164,887 55.5 7
Table 4. Catch and speces composton (%) estmated by NSAP for the purse sene fshery (2005 to 2016) n Regon 12 (SOCCSKSARGEN ) and Hgh Seas Pocket #1 based on BFAR NFRDI montorng. Speces Catch (mt) Percent (%) Skpjack tuna (Katsuwonus pelams) 326,114.4 57.9 Yellowfn tuna (Thunnus albacares) 88,990.9 15.8 Mackerel scad (Decapterus macarellus) 50,744.6 9.0 Bullet tuna (Auxs roche) 49,943.4 8.9 Frgate tuna (Auxs thazard) 24,970.4 4.4 Bgeye tuna (Thunnus obesus) 9,097.8 1.6 Eastern lttle tuna (Euthynnus affns) 5,029.9 0.9 Ranbow runner (Elagats bpnnulata) 4,908.5 0.9 Mahmah (Coryphaena hppurus) 1,183.0 0.2 Other 1,799.9 0.3 Total 562,782.8 100.0 Table 3. Results for Generalzed Lnear Models (GLMs) appled to yellowfn tuna n the purse sene fshery (2005 to 2016) n Regon 12 (SOCCSKSARGEN ) and Hgh Seas Pocket #1. The percent devance explaned s ((null devance-resdual devance)/null devance). Model selecton was based on the Akake Informaton Crtera (AIC). GLM Model Null devance Resdual devance AIC % devance explaned YR:QTR 3,004 2,758 45,748 8.2 YR:QTR+ Vessel 4,296 2,670 45,406 37.8 YR:QTR+ Area 3,434 2,724 45,373 20.6 YR:QTR+ Area+Vessel 4,393 2,660 45,368 39.3 8
Fgure 1. Rased monthly effort n the Phlppne Regon 12 (SOCCSKSARGEN ) handlne fshery based on BFAR NFRDI montorng. Fgure 2. Rased monthly yellowfn tuna catch n the Phlppne Regon 12 (SOCCSKSARGEN ) handlne fshery based on BFAR NFRDI montorng. Fgure 3. Nomnal monthly yellowfn tuna CPUE n the Phlppne Regon 12 (SOCCSKSARGEN ) handlne fshery based on BFAR NFRDI montorng. 9
Fgure 4. Quarterly relatve abundance for yellowfn tuna n the Phlppne Regon 12 (SOCCSKSARGEN ) handlne fshery as determned by Generalzed Lnear Models (GLMs). Each seres s normalzed to a mean value of 1.0. Fgure 5. Comparson of Phlppne relatve abundance ndces used n the 2014 and 2017 yellowfn tuna assessments for the western and central Pacfc Ocean. Indces are for yellowfn tuna n the Phlppne Regon 12 (SOCCSKSARGEN ) handlne fshery as determned by Generalzed Lnear Models (GLMs). Each seres s normalzed to a mean value of 1.0. 10
Fgure 6. Rased monthly effort n the Phlppne Regon 12 (SOCCSKSARGEN ) and Hgh Seas Pocket #1 purse sene fshery based on BFAR NFRDI montorng. 11
Fgure 7. Rased monthly yellowfn tuna catch n the Phlppne Regon 12 (SOCCSKSARGEN ) and Hgh Seas Pocket #1 purse sene fshery based on BFAR NFRDI montorng. Fgure 8. Nomnal monthly yellowfn tuna CPUE n the Phlppne Regon 12 (SOCCSKSARGEN ) and Hgh Seas Pocket #1 purse sene fshery based on BFAR NFRDI montorng. 12
Fgure 9. Quarterly relatve abundance for yellowfn tuna n the Phlppne Regon 12 (SOCCSKSARGEN ) and Hgh Seas Pocket #1 purse sene fshery as determned by Generalzed Lnear Models (GLMs). Each seres s normalzed to a mean value of 1.0. Fgure 10. Comparson of Phlppne relatve abundance ndces used n the 2014 and 2017 yellowfn tuna assessments for the western and central Pacfc Ocean. Indces are for yellowfn tuna n the Phlppne Regon 12 (SOCCSKSARGEN ) and Hgh Seas Pocket #1 purse sene fshery as determned by Generalzed Lnear Models (GLMs). Each seres s normalzed to a mean value of 1.0. 13