SCIENTIFIC COMMITTEE SIXTH REGULAR SESSION August Nuku alofa, Tonga

SCIENTIFIC COMMITTEE SIXTH REGULAR SESSION 10 19 August 2010 Nuku alofa, Tonga Evaluaton of longlne mtgaton to reduce catches of North Pacfc strped marln n the Hawa-based tuna fshery 1 WCPFC SC6 2010/EB WP 03 Keth Bgelow 2 and Bruno Mourato 2,3,4 1 PIFSC Workng Paper WP-10-005. Issued 30 July 2010. 2 NOAA Fsheres Pacfc Islands Fsheres Scence Center, Honolulu, Hawa, USA 3 Unversdade Federal Rural de Pernambuco, Departmento de Pesca e Aqucultura, Laboratóro de Oceanografa Pesquera, Recfe-PE, Brazl 4 Unversty of Hawa Jont Insttute for Marne and Atmospherc Research, Honolulu, Hawa, USA

1 Introducton Strped marln n the North Pacfc are prmarly harvested n longlne fsheres targetng speces such as tunas (Thunnus spp.) and swordfsh (Xphas gladus). Annual catches have declcned from ~17,000 mt n the early 1960s to ~3,000 mt n 2006 (Fgure 1). Strped marln are prmarly harvested by longlne fsheres from Japan n the northwest Pacfc and the USA n the central Pacfc wth smaller catches from Korea and Chnese-Tape longlne fleets and are also targeted n coastal fsheres off Japan and Chnese- Tape and support valuable recreatonal fsheres off Australa, New Zealand, Mexco and the Unted States. The Internatonal Scentfc Commttee for Tuna and Tuna-lke speces n the North Pacfc (ISC) completed a strped marln assessment n 2007 consderng a stock n the entre North Pacfc (Pner et al. 2007). The assessment was conducted n SS2 wth a structure of 29 dfferent fsheres, defned by regon, country and gear. Nne fsheres, all of them longlne fsheres from the western or central Pacfc, provded reasonable measures of abundance. Two assessment scenaros were based on alternatve assumptons about the steepness of the stock-recrutment relatonshp and under these two scenaros the spawnng bomass of the North Pacfc strped marln stock was estmated to have been fshed down to between 6% and 16% of ts 1952 abundance. Ths low range of spawner abundance suggested that the stock was n a depleted condton and the ISC offered the followng conservaton advce (ISC 2007): Whle further gudance from the management authorty s necessary, ncludng gudance on reference ponts and the desrable degree of reducton, the fshng mortalty rate of strped marln (whch can be converted nto effort or catch n management) should be reduced from the current level (2003 or before), takng nto consderaton varous factors assocated wth ths speces and ts fshery. Untl approprate measures n ths regard are taken, the fshng mortalty rate should not be ncreased. Brodzak and Pner (2010) presented probable status of N. Pacfc strped marln by conductng two assessment scenaros to account for dfferent hypotheses about the steepness of the stock-recrutment dynamcs. Values of relatve fshng mortalty rates durng 2001 2003 were 367% of F MSY under scenaro 1 and 190% of F MSY under scenaro 2. Correspondng estmates of strped marln bomass were below S MSY and ranged from 29% of S MSY under scenaro 1 to 44% of S MSY under scenaro 2. In relaton to MSYbased reference ponts, strped marln was experencng overfshng and the stock was consdered depleted under each steepness scenaro. Gven the hgh estmated fshng mortalty, the objectves of ths study were to conduct analyses of potental longlne catch reductons of N. Pacfc strped marln whle mantanng target bgeye tuna catches. Longlne mtgaton was based on modfcaton of longlne gear and spatally closed areas. The analyss was conducted on the Hawa-based longlne fshery whch comprses ~10% of the total N. Pacfc catch of strped marln snce 2000 and s well suted to analyses of longlne mtgaton because detaled operatonal and catch data have been gathered by the Pacfc Islands Regonal Observer Program (PIROP) snce 1994. The Hawa-based pelagc longlne fshery s composed of two sectors whch target bgeye tuna (Thunnus obesus) wth deep gear and swordfsh (Xphus gladus) wth shallow gear. The medan depth of the deepest hook on 266 deep sets was 248 m, whereas that on 333 shallow sets was 60 m (Bgelow et al. 2006). The study consdered the deep set fshery due to larger strped marln catches and potentally greater mtgaton optons as deep gear fshes at a greater range of depth and habtat than 1

shallow gear targetng swordfsh. Aspects of gear mtgaton consdered n the study were the effcacy of removng shallow hooks adjacent to longlne floats and converson of termnal gear from Japanese style tuna hooks to 18/0 crcle hooks. A spatal and temporal analyss was conducted to nvestgate the exstence of strped marln catch rate (CPUE) hot spots. An evaluaton of establshng tuna longlne fshery closures was conducted wth the trade-off between strped marln catch reductons and loss of target bgeye catch. 2 Methods 2.1 Observer data Levels of observer coverage ncreased from 4.7% of longlne sets n 1995, nearly trpled n 2000 and averaged 24.4% durng 2001 2006 (Walsh et al. 2009). The PIROP observers record operatonal detals of the longlne set and retreval, speces specfc estmates of fsh catch on a partcular hook poston, condton upon longlne retreval (.e. lve or dead), subsequent dsposton (.e. retaned or dscarded), and lengths. They also montor nteractons wth protected or endangered speces (PIRO 2010). Deep sets were characterzed as a longlne set wth at least 15 hooks between floats and shallow sets used less than 15 hooks between floats. Catch data were gathered by PIROP observers on 33,049 deep longlne sets north of the equator on 2,640 commercal fshng trps from March 1994 through December 2009. 2.2 Hook-at-capture Observers recorded the partcular hook poston, numbered sequentally between two successve floats, for each anmal caught. These hook-at-capture data are avalable even when depths are not recorded, and dstrbutons of hook-at-capture were generated from the 33,049 observed sets that montored 68,165,026 hooks. A total of 1,534,262 anmals representng 125 fsh, elasmobranch, turtle, mammal and brd speces or speces groups were caught. Hook-at-capture dstrbutons by speces and correspondng catch rate by hook number were only consdered n the analyss f at least 5,000 ndvduals were caught. Some speces were rarely encountered and ths crteron reduced the analyss to 1,463,000 ndvduals of 19 speces (Table 1). A proporton (~14%) of ndvduals was reported as caught but wthout a hook-atcapture number. In these cases, the avalable catch rate by hook number dstrbuton for a speces was expanded (rased) to represent all captures for the purposes of modelng the effects of elmnatng partcular hook numbers. 2.3 Longlne catch scenaros wth removng hooks adjacent to floats and redstrbuton of hooks For modelng the removal of hooks near floats, sequental hook postons were renumbered from shallowest (nearest the float) to deepest (half way between floats) for each confguraton of hooks between floats (the number of hooks between floats can vary). The catch and effort were then summed by hook number. Sx scenaros based on catch rate by hook number were developed to portray the change n catches expected to occur by removng hooks adjacent to longlne floats. Three scenaros consdered the removal of hook #1, hooks #1 2 and hooks #1 3 adjacent to the float. Removal of hooks actually occurs on both sdes of each float, thus the removal of hooks #1 3 consttutes a total of sx hooks removed. An addtonal three scenaros consdered the same sequence of hook removal though the removed hooks were redeployed n the deeper depth pattern by addng to the total length of the set (wth 2

addtonal manlne and floats). For example, f hooks #1 3 were removed, then they were deployed at poston #4 and deeper n an extenson to the length of the modeled set. Each scenaro consdered a hypothetcal longlne set wth 2,800 hooks deployed and 28 hooks between floats. The sx scenaros were further stratfed by catches kept (landed) and released (bycatch) to quantfy changes expected by hook removal. Proportons of kept and released ndvduals by speces were based on 33,049 observed sets (Table 1) and the scenaros assumed that proportons dd not change through tme. 2.4 Catchablty of Japanese tuna and crcle hooks Sxteen Hawa-based tuna longlne vessels were contracted to alternate large non-rnged stanless steel crcle hooks ( C hooks sze 18/0 made n Korea) wth the vessel s conventonal hooks (composed entrely of Japanese style tuna hooks sze 3.6 sun (hereafter referred as a tuna hooks) on all longlne deployments. Vessels mxed two ndvdual hook types (e.g. C, tuna, C, tuna) throughout the longlne set n a 1:1 rato, and there was no change n operatonal characterstcs n order to mnmze sources of varaton. Vessels chose where they fshed and were allowed to retan and sell ther catch. At the begnnng of the feld trals, all vessels were mandated to mx the alternate hook types throughout the entre longlne set and to mantan a 1:1 rato of hook types throughout the trals. Branchlne snaps wth 10-cm cable tes allowed for easy dentfcaton of hook type and correspondng fsh catch. Vessels conducted trps between July 2005 and February 2006. Every trp was accompaned by a Natonal Oceanc and Atmospherc Admnstraton (NOAA) certfed Pacfc Islands Regon observer who collected nformaton on all catch by speces, hook type, sequental hook number between two floats, lfe status, catch dsposton (retaned, dscarded), and length measurements of some landed speces. The observer conducted a daly tally of the numbers of each type of hook deployed, and evaluated the vessel s ablty to follow expermental protocols. A total of 1,182 longlne sets were analyzed. The most numerous 18 ndvdual speces were chosen for analyss representng a mnmum catch rate of 0.14 fsh per longlne set. Generalzed Lnear Mxed Models (GLMM) were appled to explctly model the underlyng processes n the catch rate (CPUE, number per 1,000 hooks) data and to estmate catchablty of hook types. GLMMs were employed to accommodate the herarchy n whch longlne sets wthn a trp are expected to be more smlar than sets across trps. For each speces, the GLMM predcts mean catch (µ ) per set usng three categorcal and two contnuous varables wth a log lnk: log( ) N H T V 2 3 2 3 1Lat 2Lat 3Lat 4Lon 5Lon 6Lon log( E ) where N s the mean local abundance; H, hook type effect; T, tme (year:quarter) effect; V, vessel effect; Lat and Lon are thrd order (cubc) effects of lattude and longtude and offset E s the number of hooks deployed durng longlne operaton. Catch rates are correlated wthn a trp, thus trp was assgned as the groupng varable n GLMMs wth no estmated random effects. GLMMs were appled usng the glmm.admb module n R (verson 2.7.2 for Lnux) and consdered both posson and negatve bnomal response dstrbutons. Model selecton was conducted by AIC. 3

2.5 Strped marln catch rate (CPUE) hot spots and effects of spatally closed areas Logbooks estmates of stophord catches n the Hawa longlne fshery are based due to msdentfcaton, especally blue marln whch was estmated to be nflated by 29.4% whle catches of other marln speces are negatvely based (Walsh et al. 2005). A scentfc database of corrected stophord catches from 1994 to 2003 was used to nvestgate spatal CPUE hot spots and evaluated tuna longlne fshery closures. In order to predct the areas wth hgh CPUE we appled generalzed addtve models (GAM) usng penalzed regresson splnes, estmated by penalzed teratve least squares (Wood, 2006). Two models were ft to the data: - Reduced model log( ) s Lon, Lat log( E ) - Full model log( ) Y s Lon, Lat s( M ) s( V ) s( HBF ) log( E ) where µ s the number of strped marln caught per set; s (Lon, Lat) represents the effect of the spatal locaton as an sotropc bvarate functon (Wood and Augustn, 2002); Y, year effect as a factor; M, month effect; V, vessel effect; HBF, the number of hooks deployed between two successve floats and offset E s the number of hooks deployed durng longlne operaton. Intally the models were ft for the entre perod (1994 2003) and also on a quarterly bass to assess seasonal varaton. Moreover, n order to delneate and assess the persstence of the CPUE hot spot areas we also ft models for each year separately. GAMs were ftted usng the mgcv package (Wood 2006) n R 2.10.1 (R Development Core Team, 2009) wth the optmum degrees of smoothng beng estmated by the generalzed cross-valdaton (GCV) crteron. The selecton of predctors and the decson on entry or excluson was based on AIC and the total devance explaned. Two error dstrbutons (posson and negatve bnomal) were consdered and the decson of error dstrbuton was based on the lowest values of AIC and BIC, total devance explaned and adjusted-r 2. The latter was defned as the proporton of varance explaned, where orgnal varance and resdual varance were both estmated usng unbased estmators (Wood, 2006). The spatal predctons were generated from the contrbuton of the smoothed lattude-longtude term and the predctons occurred unformly across the locatons of observed fshng and not exclusvely over a regular spatal grd. Furthermore, spatal predctons wth overall mean of +1 standard devaton were also llustrated to assess f hgh CPUE perssted spatally across years followng the methodology n Watson et al. (2008). Impact of fshery closures was estmated on a spatal grd from 180º to 128ºW and the equator to 40ºN (Fgure 2). All possble contguous squares and rectangles were consdered whch resulted n 79,380 closure optons. The percentage reducton n strped marln and bgeye catches was estmated for each closed area. There was no temporal structure for the closures and no redstrbuton of fshng effort was consdered. 4

3 Results 3.1 Observer data and hook-at-capture Attrbutes of gear confguraton on observed deep sets were smlar to prevously values estmated for the longlnes montored durng 1996 1999 (Table 1 n Bgelow et al. 2006). Hook-at-capture results are ntally llustrated for two speces strped marln and bgeye tuna (Fgure 3). The hooks are numbered sequentally from one float to the subsequent float and ndcate that strped marln are caught on shallow hooks whle bgeye are on deeper hooks. For each hook between float confguraton, the hooks were renumbered from shallow to deep and catch and effort were summed by hook number. The resultng CPUE by hook number s llustrated for 19 speces n Fgure 4. The shallowest hooks, adjacent to the longlne floats, have substantally hgher CPUE for wahoo, four stophord speces, oceanc whte-tp shark and mahmah. The deepest hooks have hgher CPUE for bgeye tuna and bgeye thresher shark. 3.2 Longlne catch scenaros wth removng hooks adjacent to floats and redstrbuton of hooks Table 2 summarzes the percentage of catch by speces on hook#1 and hooks#1 2. There are dfferences by quarter for most speces. Strped marln had the lowest percentage n reductons durng quarter 1 (12% on hook#1 and 23% on hooks#1 2) and hghest percentage durng quarter 4 (25% on hook#1 and 43% on hooks#1 2). Table 3 consders catch scenaros resultng from removal of hooks adjacent to floats and the redstrbuton of hooks to deeper depths. Catch reductons n Tables 2 3 dffer slghtly because estmates n Table 2 were from the entre hook-at-capture CPUE dstrbuton (Fgure 4) whereas Table 3 consdered one partcular gear type (28 HBF). The largest catch reductons n terms of percentage occurred for strped marln, spearfsh and dolphnfsh. Target bgeye catches declned 1.5%, 4% and 7.8% by removng hook #1, hooks #1 2 and hooks #1 3 adjacent to the float. Total catches of all landed fsh declned at hgher percentages (8%, 16% and 23%) manly due to lower CPUE of mahmah. Total bycatch declned at the lower percentages (5%, 12% and 19%) due to lower CPUE of blue shark and lancetfsh. The redstrbuton of hook #1, hooks #1 2 and hooks #1 3 ncreased target bgeye catches by 6%, 12% and 17%. Albacore, bgeye thresher shark and sckle pomfret also had smlar catch ncreases as bgeye tuna wth removng and redstrbutng hooks. Total bycatch was largely unaffected by removng and redstrbutng hooks. 3.3 Catchablty of Japanese tuna and crcle hooks Tuna hooks caught 4,630 bgeye and 642 strped marln whereas 18/0 crcle hooks caught 4,722 bgeye and 370 strped marln. Detals for the remanng 16 speces are not dscussed as results are ncluded n another manuscrpt currently under revew. The GLMM hook type coeffcents are equvalent to catchablty and nterpreted as the magntude of the CPUE dfferences between hook types consderng the ncluson of other sgnfcant explanatory varables. The crcle hook coeffcents for bgeye tuna and strped marln were 1.011 (95% CI 0.949 1.073) and 0.574 (95% CI 0.431 0.718), respectvely. Bgeye CPUE was not sgnfcantly dfferent between tuna and crcle hooks. A coeffcent of 0.574 ndcates that strped marln CPUE was ~42.6% reduced on crcle hooks compared to tuna hooks. 5

3.4 Strped marln catch rate (CPUE) hot spots and effects of spatally closed areas Model results of the GAM analyss are provded n Table 4. A negatve bnomal error dstrbuton was consstently preferred over a posson dstrbuton. The reduced models wth only the smoothed lattudelongtude term explaned ~13% of the devance, whereas the full model wth effects of month, vessel and gear (HBF) explaned ~24% of the devance. The spatal predcton aggregated over the entre tme-seres (1994 2003) ndcated an area of hgh CPUE at 25ºN and 165ºW (Fgure 5). The hgh CPUE occurred n the same area n quarters 1 through 3, but CPUE was lower and more dspersed n quarter 4 (Fgure 6). However, when ndvdual years were consdered the spatal dstrbuton of hgh CPUE was hghly varable and not persstent, suggestng a lack of CPUE hot-spots (Fgure 7). The closure analyss dd not dentfy areas of potentally hgh strped marln reductons wth mnmal reductons of target bgeye catch (Fgure 8), rather there s a co-occurrence n catch of both speces. The lack of an optmal spatal closure s reflected n the hgh nterannual spatal varaton of CPUE (Fgure 7). 4 Dscusson Analyses of the effects on removng shallow hooks and changng from tuna to crcle hooks both demonstrated moderate strped marln CPUE reductons wth mnmal or no reductons on target bgeye CPUE. Strped marln CPUE hot-spots exst n the Baja Calforna n the eastern Pacfc, near New Zealand and n the northwest Pacfc; however there were no hot-spots dentfed that were spatally persstent n the area fshed by the Hawa-based tuna fshery. Modfyng longlne gear to fsh deeper has been demonstrated to provde conservaton benefts to stophords by comparng CPUE on shallow and deep gear (Suzuk et al. 1977) and by expermental longlne trals (Boggs 1992, Beverly et al. 2009). An estmated 16 and 30% reducton n strped marln catch could occur f hook#1 and hooks#1 2 were not fshed, respectvely; and longlne scenaros wth a specfc gear confguraton of 28 HBFs ndcated reductons of 18 and 34%. Three of the longlne scenaros changed the gear confguraton by redstrbutng hooks to deeper depths. Whle these scenaros ndcate ncreased CPUE and catches of target bgeye, there are operatonal dffcultes as more manlne wll need to be deployed, thus ncreasng both the settng and retreval tmes. Consderng a longlne wth 2,800 hooks deployed and 28 hook postons between floats, f two hooks are removed adjacent to the float then 14% of the effort (4 of every 28 hooks) would have to be redstrbuted. Longlne retreval s typcally completed by 2 am and the set commences agan at 6:30 am thus there s lttle tme for extendng retrevals wth the current observed operatonal patterns. Usng large (18/0) crcle hooks had a larger effect on CPUE (42% reducton) than removng shallow hooks. Reduced catchablty occurred for most speces on large crcle hooks and we contend that these reductons are a functon of 18/0 crcle hook morphology. Although there are a myrad of types, szes, and shapes of hooks used wthn longlne fsheres around the world, the mnmum wdth appears to be the prmary metrc nfluencng catchablty rather than gape or straght total length. The 18/0 crcle hook had a mnmum wdth (4.9 cm) that was 57% wder than the Japanese tuna (3.1 cm). The larger mnmum wdth relates to a smaller probablty of ngeston and probably accounts for the reduced catchablty of non-bgeye speces. 6

We assume that the there would be a multplcatve effect of a ~70% reducton n strped marln CPUE by both removng shallow hooks and mplementng large crcle hooks. Whle there are demonstrated conservaton benefts to a varety of speces n not fshng shallow hooks or mplementng large crcle hooks, there s economc concern n the Hawa-based tuna fshery of lost revenue due to lower catch rates of stophords, dolphnfsh and some pelagc sharks that are often retaned and marketed. 5 Acknowledgments We thank captans and crews of Hawa-based longlne vessels and observers. We thank Wllam Walsh for constructng a scentfc database of corrected stophord catches. 6 References Beverly, S., Curran, D., Musyl, M., and Molony, B. 2009. Effects of elmnatng shallow hooks from tuna longlne sets on target and non-target speces n the Hawa-based pelagc tuna fshery. Fsh. Res. 96:281 288. Bgelow, K., Musyl, M.K., Posson, F., and Kleber, P. 2006. Pelagc longlne gear depth and shoalng. Fsh. Res. 77: 173 183. Boggs, C.H., 1992. Depth, capture tme, and hooked longevty of longlne-caught pelagc fsh: tmng btes of fsh wth chps. Fsh. Bull. 90, 642 658. Brodzak, J., and Pner, K. 2010. Model averagng and probable status of North Pacfc strped marln, Tetrapturus audax. CJFAS 67:793 805. Internatonal Scentfc Commttee for Tuna and Tuna-lke Speces n the North Pacfc Ocean [ISC]. 2007. Report of the Seventh Meetng of the Internatonal Scentfc Commttee for Tuna and Tuna-lke Speces n the North Pacfc Ocean, 25-30 July 2007, Busan, Korea. 53 p. Internatonal Scentfc Commttee for Tuna and Tuna-lke Speces n the North Pacfc Ocean [ISC]. 2010. Report of the Bllfsh Workng Group Workshop, 12-13 July 2010, Vctora, Brtsh Columba, Canada. 23 p. Pacfc Islands Regonal Offce. 2010. Hawa Longlne Observer Program Observer Feld Manual. Natonal Oceanc and Atmospherc Admnstraton, Pacfc Islands Regon, Honolulu, Hawa. Pner, K., Conser, R., DNardo, G., and Brodzak, J. 2007. Stock synthess 2 senstvty runs for strped marln assessment WG 2007. Internatonal Scentfc Commttee for Tuna and Tuna- Lke Speces n the North Pacfc/MARWG-SWORWG761 1/Workng Paper 2. R Development Core Team. 2009. R: A language and envronment for statstcal computng. R Foundaton for Statstcal Computng, Venna, Austra. ISBN 3-900051-07-0, URL http://www.r-project.org. 7

Suzuk, Z., Warashna, Y., and Kshda, M. 1977. The comparson of catches by regular and deep tuna longlne gears n the western and central equatoral Pacfc. Bull. Far Seas Fsh. Res. Lab. 15:51 89. Walsh, W.A., Ito, R. Y., Kawamoto, K. E., and McCracken, M. 2005. Analyss of logbook accuracy for blue marln (Makara ngrcans) n the Hawa-based longlne fshery wth a generalzed addtve model and commercal sales data. Fsh. Res. 75: 175 192. Walsh, W.A., Bgelow, K. A., and Sender, K. 2009. Decreases n Shark Catches and Mortalty n the Hawa-Based Longlne Fshery as Documented by Fshery Observers. Marne and Coastal Fsheres: Dynamcs, Management, and Ecosystem Scence 1:270 282. Watson, J., Essngton, T., Lennert-Cody, C., and Hall, M. 2008. Trade-Offs n the Desgn of Fshery Closures: Management of Slky Shark Bycatch n the Eastern Pacfc Ocean Tuna Fshery. Cons. Bol. 23(3):626 635. Wood, S. N. 2006. Generalzed addtve models: an ntroducton wth R. Chapman & Hall. London. Wood, S.N. and Augustn, N.H. 2002. GAMs wth ntegrated model selecton usng penalzed regresson splnes and applcatons to envronmental modelng. Ecol. Model. 157: 157-177. 8

Table 1. Catch of 19 most common speces observed on 33,049 sets n the Hawa-based tuna longlne fshery wth vald hook-at-capture and percentage retaned. Speces Catch Vald hook-atcapture Percentage retaned Bgeye tuna Thunnus obesus 286,621 275,351 94.54 Yellowfn tuna Thunnus albacares 72,184 69,636 90.72 Wahoo Acanthocybum solandr 32,149 21,272 94.78 Albacore Thunnus alalunga 47,425 45,947 97.51 Skpjack tuna Katsuwonus pelams 61,026 59,803 82.84 Strped marln Kajka audax 34,365 33,611 94.40 Spearfsh Tetrapturus angustrostrs 28,612 28,012 92.21 Swordfsh Xphas gladus 12,367 11,854 58.45 Blue marln Makara ngrcans 9,358 8,952 96.28 Blue shark Pronace glauca 146,307 142,834 0.33 Bgeye thresher Alopas superclosus 11,646 11,129 8.63 Oceanc whte-tp shark Carcharhnus longmanus 5,303 5,124 4.22 Pelagc stngray Dasyats volacea 11,526 7,151 3.63 Dolphnfsh Coryphaena hppurus 137,599 98,238 92.31 Opah Lamprs guttatus 28,238 19,802 97.57 Sckle pomfret Taractchthys stendachner 94,586 69,358 96.05 Lancetfsh Alepsaurus ferox 314,747 246,916 0.02 Escolar Lepdocybum flavobrunneum 45,908 36,491 88.66 Snake mackerel Gempylus serpens 83,033 62,900 1.84 9

Table 2. Percentage of ndvduals taken on the frst and frst/second hooks closest to the longlne float observed on 33,049 sets n the Hawabased tuna longlne fshery. Speces All quarters 1 st quarter 2 nd quarter 3 rd quarter 4 th quarter Hook 1 Hooks 1-2 Hook 1 Hooks 1-2 Hook 1 Hooks 1-2 Hook 1 Hooks 1-2 Hook 1 Hooks 1-2 Bgeye tuna 1.13 2.89 1.12 2.97 1.09 2.77 1.43 3.44 1.01 2.68 Yellowfn tuna 3.69 8.78 2.93 7.32 4.71 10.57 3.69 8.56 3.80 9.20 Wahoo 12.31 25.07 9.75 20.40 12.23 25.62 14.11 27.87 12.11 23.98 Albacore 1.28 3.77 1.56 4.20 1.23 3.83 1.98 5.81 0.98 2.90 Skpjack tuna 8.23 18.54 5.76 13.50 9.74 21.71 13.10 28.44 8.98 20.14 Strped marln 16.71 30.64 12.15 23.30 17.78 31.81 25.20 43.22 17.28 32.18 Spearfsh 22.07 38.28 19.14 34.51 24.80 42.63 29.69 49.21 21.57 37.14 Swordfsh 6.58 14.57 5.70 10.95 3.93 9.64 8.10 17.04 8.04 18.60 Blue marln 14.10 25.52 11.87 21.46 13.93 25.62 15.95 27.37 14.14 26.51 Blue shark 4.73 10.69 4.95 10.79 4.72 10.58 5.57 12.25 4.16 9.80 Bgeye thresher 1.14 3.16 1.23 3.13 1.04 2.87 1.50 4.63 1.19 3.31 Oceanc whtetp shark 15.14 26.82 14.22 24.95 15.75 26.93 15.98 36 15.73 27.73 Pelagc stngray 3.89 9.99 2.55 7.24 3.77 9.15 5.36 12.59 3.22 9.10 Dolphnfsh 22.82 38.42 26.74 43.56 25.51 42.94 19.98 34.35 21.28 35.91 Opah 1.02 2.86 0.58 1.74 1.08 3.45 0.92 2.48 0.92 2.54 Sckle pomfret 0.56 1.67 0.45 1.56 0.51 1.44 0.50 1.51 0.62 1.86 Lancetfsh 3.30 8.48 2.88 7.25 2.88 7.67 3.97 9.71 2.41 6.83 Escolar 7.29 15.99 6.75 15.38 7.08 15.78 7.72 15.95 7.21 15.94 Snake mackerel 5.67 12.99 4.86 11.25 6.28 13.93 7.96 17.33 5.01 11.96 10

Table 3. Nomnal CPUE (number per 1000 hooks) and percentage landed for 19 speces observed on 33,049 longlne sets n the Hawa-based fshery. Sx longlne scenaros ndcate hypothetcal catches by removng varous hooks wth and wthout redstrbuton for a longlne set wth 28 hooks between floats and 2,800 hooks. Removng hook #1 adjacent to float Speces Nomnal Percentage Percentage Removng no hooks (status quo) Catch percent Common name code cpue landed bycatch Catch Landed Bycatch Catch of status quo Landed Bycatch Bgeye tuna 916 4.15 94.54 5.46 11.63 11.00 0.64 11.47 98.56 10.84 0.63 Yellowfn tuna 1 1.06 90.72 9.28 2.96 2.68 7 2.81 95.22 2.55 6 Wahoo 57 0.47 94.78 5.22 1.33 1.26 0.07 1.15 86.97 1.09 0.06 Albacore 5 0.69 97.51 2.49 1.92 1.88 0.05 1.89 98.42 1.85 0.05 Skpjack tuna 2 0.90 82.84 17.16 2.52 2.08 0.43 2.26 89.97 1.88 0.39 Strped marln 92 0.51 94.40 5.60 1.42 1.34 0.08 1.15 81.54 1.09 0.06 Spearfsh 94 0.42 92.21 7.79 1.18 1.09 0.09 0.89 75.71 0.82 0.07 Swordfsh 91 0.18 58.45 41.55 0.51 0.30 1 0.47 91.49 7 0.19 Blue marln 93 0.14 96.28 3.72 0.39 0.37 0.01 0.33 84.45 0.31 0.01 Blue shark 167 2.15 0.33 99.67 6.01 0.02 5.99 5.68 94.57 0.02 5.66 Bgeye thresher 147 0.17 8.63 91.37 0.47 0.04 0.43 0.47 98.77 0.04 0.43 Oceanc whte-tp shark 138 0.08 4.22 95.78 2 0.01 1 0.18 82.72 0.01 0.17 Pelagc stngray 193 0.17 3.63 96.37 0.47 0.02 0.45 0.45 95.72 0.02 0.43 Dolphnfsh 914 2.03 92.31 7.69 5.68 5.24 0.44 4.26 75.04 3.93 0.33 Opah 467 0.41 97.57 2.43 1.15 1.13 0.03 1.14 99.13 1.12 0.03 Sckle pomfret 908 1.35 96.05 3.95 3.79 3.64 0.15 3.78 99.67 3.63 0.15 Lancetfsh 909 4.64 0.02 99.98 13.00 0.00 13.00 12.57 96.66 0.00 12.56 Escolar 13 0.68 88.66 11.34 1.89 1.68 1 1.74 91.82 1.54 0 Snake mackerel 295 1.22 1.84 98.16 3.43 0.06 3.36 3.22 93.86 0.06 3.16 Total catch 59.96 33.83 26.13 55.91 31.07 24.84 (numbers of fsh) 11

Table 3 contnued. Nomnal CPUE (number per 1000 hooks) and percentage landed for 19 speces observed on 33,049 longlne sets n the Hawabased fshery. Sx longlne scenaros ndcate hypothetcal catches by removng varous hooks wth and wthout redstrbuton for a longlne set wth 28 hooks between floats and 2,800 hooks. Removng hooks #1-2 adjacent to float Removng hooks #1-3 adjacent to float Catch percent Catch percent Common name Catch of status quo Landed Bycatch Catch of status quo Landed Bycatch Bgeye tuna 11.17 96.04 10.56 0.61 10.73 92.20 10.14 0.59 Yellowfn tuna 2.61 88.43 2.37 4 2.36 79.89 2.14 2 Wahoo 0.97 73.35 0.92 0.05 0.81 60.83 0.76 0.04 Albacore 1.82 94.87 1.78 0.05 1.72 89.36 1.68 0.04 Skpjack tuna 1.94 77.20 1.61 0.33 1.62 64.40 1.34 8 Strped marln 0.94 66.11 0.88 0.05 0.75 53.21 0.71 0.04 Spearfsh 0.68 57.84 0.63 0.05 0.52 43.98 0.48 0.04 Swordfsh 0.41 81.00 4 0.17 0.36 70.77 1 0.15 Blue marln 8 71.86 7 0.01 3 60.19 2 0.01 Blue shark 5.26 87.58 0.02 5.24 4.77 79.33 0.02 4.75 Bgeye thresher 0.45 96.22 0.04 0.42 0.44 92.43 0.04 0.40 Oceanc whte-tp shark 0.15 69.40 0.01 0.14 0.13 58.06 0.01 0.12 Pelagc stngray 0.42 88.72 0.02 0.40 0.38 80.58 0.01 0.37 Dolphnfsh 3.29 57.95 3.04 5 2.60 45.80 2.40 0 Opah 1.12 97.17 1.09 0.03 1.08 93.80 1.06 0.03 Sckle pomfret 3.73 98.50 3.59 0.15 3.65 96.31 3.51 0.14 Lancetfsh 11.85 91.12 0.00 11.85 10.91 83.89 0.00 10.91 Escolar 1.55 81.91 1.38 0.18 1.36 71.55 1.20 0.15 Snake mackerel 2.94 85.75 0.05 2.88 2.63 76.68 0.05 2.58 Total catch 51.60 28.49 23.11 47.03 25.98 21.05 (numbers of fsh) 12

Table 3 contnued. Nomnal CPUE (number per 1000 hooks) and percentage landed for 19 speces observed on 33,049 longlne sets n the Hawabased fshery. Sx longlne scenaros ndcate hypothetcal catches by removng varous hooks wth and wthout redstrbuton for a longlne set wth 28 hooks between floats and 2,800 hooks. Re-dstrbutng shallow hooks Removng hook #1 adjacent to float Removng hooks #1-2 adjacent to float Removng hooks #1-3 adjacent to float Catch percent Catch percent Catch percent Common name Catch of status quo Landed Bycatch Catch of status quo Landed Bycatch Catch of status quo Landed Bycatch Bgeye tuna 12.35 106.14 11.67 0.67 13.04 112.05 12.32 0.71 13.65 117.35 12.91 0.75 Yellowfn tuna 3.03 102.55 2.75 8 3.05 103.17 2.77 8 3.01 101.67 2.73 8 Wahoo 1.24 93.66 1.18 0.06 1.13 85.57 1.07 0.06 1.03 77.42 0.97 0.05 Albacore 2.04 105.99 1.99 0.05 2.13 110.68 2.08 0.05 2.19 113.73 2.13 0.05 Skpjack tuna 2.44 96.89 2.02 0.42 2.27 90.06 1.88 0.39 2.06 81.97 1.71 0.35 Strped marln 1.24 87.81 1.17 0.07 1.09 77.13 1.03 0.06 0.96 67.72 0.91 0.05 Spearfsh 0.96 81.54 0.89 0.07 0.79 67.49 0.73 0.06 0.66 55.98 0.61 0.05 Swordfsh 0.50 98.53 9 1 0.48 94.50 8 0 0.46 90.07 7 0.19 Blue marln 0.35 90.94 0.34 0.01 0.32 83.84 0.31 0.01 0.30 76.60 8 0.01 Blue shark 6.12 101.85 0.02 6.10 6.14 102.18 0.02 6.12 6.07 100.97 0.02 6.05 Bgeye thresher 0.50 106.37 0.04 0.46 0.53 112.26 0.05 0.48 0.56 117.64 0.05 0.51 Oceanc whte-tp shark 0.19 89.08 0.01 0.19 0.18 80.96 0.01 0.17 0.16 73.90 0.01 0.15 Pelagc stngray 0.49 103.09 0.02 0.47 0.49 103.51 0.02 0.47 0.48 102.56 0.02 0.47 Dolphnfsh 4.59 80.81 4.24 0.35 3.84 67.61 3.54 0.30 3.31 58.28 3.06 5 Opah 1.23 106.76 1.20 0.03 1.31 113.36 1.28 0.03 1.38 119.39 1.34 0.03 Sckle pomfret 4.07 107.33 3.91 0.16 4.36 114.92 4.18 0.17 4.65 122.58 4.46 0.18 Lancetfsh 13.53 104.09 0.00 13.53 13.82 106.31 0.00 13.82 13.88 106.77 0.00 13.88 Escolar 1.87 98.88 1.66 1 1.81 95.57 1.60 1 1.72 91.07 1.53 0 Snake mackerel 3.46 101.07 0.06 3.40 3.43 100.04 0.06 3.37 3.34 97.59 0.06 3.28 Total catch 61 33.46 26.75 60 33.24 26.96 59.86 33.06 26.80 (numbers of fsh) 13

Table 4. Model selecton results for strped marln CPUE for reduced and full Generalzed Addtve Models. Reduced Model Full Model GCV % Dev. AIC BIC R-sq (adjust) GCV % Dev. AIC BIC R-sq (adjust) Posson (1994-2003) 2.56 12.90 298344.80 298615.10 0.08 2.11 28.10 264957.30 265546.60 2 Neg. Bnomal (1994-2003) 12.20 243544.70 243815.10 0.08 26.10 229964.30 230556.50 1 Neg. Bnomal (1 quarter, 1994-2003) 0.08 66974.18 67198.49 12.40 20.70 65183.98 65609.00 0.16 Neg. Bnomal (2 quarter, 1994-2003) 12.70 54603.46 54821.31 0.07 0.18 52742.03 53114.34 21.70 Neg. Bnomal (3 quarter, 1994-2003) 13.20 32209.07 32432.81 0.04 26.10 29802.98 30213.69 0.14 Neg. Bnomal (4 quarter, 1994-2003) 8.88 83735.05 83959.78 0.06 26.20 78954.18 79397.45 4 Neg. Bnomal (1994) 9.69 11405.00 11556.45 0.08 18.40 11083.81 11310.02 0.18 Neg. Bnomal (1995) 11.50 26179.00 26371.47 0.14 26.80 25061.10 25411.29 7 Neg. Bnomal (1996) 12.70 22695.00 22885.77 0.09 22.90 22023.39 22357.96 7 Neg. Bnomal (1997) 7.70 20967.00 21159.25 0.07 24.70 19778.15 20080.49 1 Neg. Bnomal (1998) 29.70 21444.00 21637.28 8 39.40 20523.13 20786.58 0.38 Neg. Bnomal (1999) 12.60 26162.00 26348.46 0.11 20.90 25416.37 25783.76 3 Neg. Bnomal (2000) 20.90 15761.66 15952.04 0.17 32.50 14826.31 15131.25 0.30 Neg. Bnomal (2001) 14.00 27338.24 27539.24 0.09 24.80 26238.87 26605.34 0.18 Neg. Bnomal (2002) 12.20 22442.22 22648.29 0.08 25.60 20984.31 21335.21 2 Neg. Bnomal (2003) 18.00 38407.32 38615.91 0.16 27.20 37201.19 37594.69 7 14

Fgure 1. Tme-seres of annual strped marln catches by gear n the North Pacfc, Source: ISC 2010. Note: Longlne strped marln catch s negatvely based due to speces msreportng n the US fleet. 15

Fgure 2. Spatal grds consdered n an analyss of fshery closures. 16

Fgure 3. Frequency of hook-at-capture (horzontal axs) for two speces n four commonly used longlne gear confguratons. (HBF: hooks between floats) observed n the Hawa-based tuna fshery. Hooks are numbered sequentally from each float to the adjacent float. 17

Fgure 4. Comparson of catch rates by hook number for 19 speces caught by the Hawa-based tuna fshery (n=33,049 sets). Hook #1 s adjacent to the longlne float whle hook #19 s assumed to be the deepest hook. 18

Fgure 4 contnued. Comparson of catch rates by hook number for 19 speces caught by the Hawabased tuna fshery (n=33,049 sets). Hook #1 s adjacent to the longlne float whle hook #19 s assumed to be the deepest hook. 19

Fgure 4 contnued. Comparson of catch rates by hook number for 19 speces caught by the Hawabased tuna fshery (n=33,049 sets). Hook #1 s adjacent to the longlne float whle hook #19 s assumed to be the deepest hook. Fgure 5. Spatal predctons of CPUE (number per 1,000 hooks) for strped marln n the Hawa-based tuna longlne fshery based on reduced and full GAM models (1994 2003). Reduced model (1994-2003) Full model (1994-2003) Lattude ( S) 0 10 20 30 40 1.2 2 1.8 1.6 1.4 0.8 Lattude ( S) 0 10 20 30 40.8 2 0.4 0.6 1 1.2 2.2 1.4 0.8 0.4 0.6 1.2-170 -165-160 -155-150 -145-140 -170-165 -160-155 -150-145 -140 Longtude ( W) Longtude ( W) 20

Fgure 6. Quarterly spatal predctons of CPUE (number per 1,000 hooks) for strped marln n the Hawa-based tuna longlne fshery based on reduced and full GAM models (1994 2003). Reduced model (1 quarter, 1994-2003) Full model (1 quarter, 1994-2003) Lattude ( S) 0 10 20 30 40 1.8 1.4 0.8 1.2 1.6 0.8 0.4 1.2 0.6 Lattude ( S) 0 10 20 30 40 1.4 0.8 0.6 0.6 0.4-170 -165-160 -155-150 -145-140 Longtude ( W) Reduced model (2 quarter, 1994-2003) -170-165 -160-155 -150-145 -140 Longtude ( W) Full model (2 quarter, 1994-2003) Lattude ( S) 0 10 20 30 40 1.6 2 1.4 1.8 1.2 0.6 Lattude ( S) 0 10 20 30 40 0.8 0.5 0.1 0.1 0.3 0.5 0. -170-165 -160-155 -150-145 -140 Longtude ( W) -170-165 -160-155 -150-145 -140 Longtude ( W) 21

Fgure 6 contnued. Quarterly spatal predctons of CPUE (number per 1,000 hooks) for strped marln n the Hawa-based tuna longlne fshery based on a reduced and full GAM models (1994 2003). Reduced model (3 quarter, 1994-2003) Full model (3 quarter, 1994-2003) Lattude ( S) 0 10 20 30 40 2.8 2 3.2 1.2 1.6 1.8 0.6 0.8 1 0.6 0.4 0.6 0.8 0.4 0.4 Lattude ( S) 0 10 20 30 40 0.4 0.4 0.3 0.1 0.3 0.4 0.1-170 -165-160 -155-150 -145-140 Longtude ( W) -170-165 -160-155 -150-145 -140 Longtude ( W) Reduced model (4 quarter, 1994-2003) Full model (4 quarter, 1994-2003) Lattude ( S) 0 10 20 30 40 2 1.2 1.6 1.4 1.8 1.6 0.8 Lattude ( S) 0 10 20 30 40 0.16 0.1 0.06 0.08 0.04 0.12 0.1 0.06 0.08 0.1 0.04 0.02 0.02-170 -165-160 -155-150 -145-140 Longtude ( W) -170-165 -160-155 -150-145 -140 Longtude ( W) 22

Fgure 7. Spatal predctons of CPUE for strped marln n the Hawa-based tuna longlne fshery based on reduced and full GAM models for ndvdual years from 1994 to 2003. Contour lnes are ndvdual years wth areas of hgh CPUE (mean + 1 standard devaton). Reduced models Full models Lattude ( S) 0 10 20 30 40 Lattude ( S) 0 10 20 30 40-170 -165-160 -155-150 -145-140 -170-165 -160-155 -150-145 -140 Longtude ( W) Longtude ( W) Fgure 8. Trade-offs between the percentage reducton n strped marln and bgeye tuna catch for spatal closures n the Hawa-based tuna longlne fshery. Each pont represents a dfferent closure for the perod 1994 2003. 23