CHAPTER 11 PACIFIC OCEAN PERCH by Paul D. Spencer and James N. Ianelli Executive Summary The last full assessment for Pacific ocean perch (POP) was presented to the Plan Team in 2006. The following changes were made to POP assessment relative to the November 2006 SAFE: Summary of Changes in Assessment Inputs Changes in the Input Data (1) The harvest time series were revised and updated through August 30, 2008. (2) The 2006 AI survey age composition was included in the assessment. (3) The 2006 and 2007 size compositions from the Aleutian Islands fishery were included in the assessment. (4) The historical Aleutian Islands survey data were updated based on the estimates provided by the AFSC/RACE Division. Changes in the Assessment Methodology There were no changes in the assessment methodology. Summary of Results A summary of the 2008 assessment recommended ABC’s relative to the 2007 recommendations is shown below. BSAI Pacific ocean perch are not overfished or approaching an overfished condition. Assessment Year 2007 2008 Projections Year 2008 2009 2009 2010 M 0.062 0.062 0.060 0.060 Tier 3a 3a 3a 3a B (mt) 331,158 331,158 307,507 307,507 100% B (mt) 132,463 132,463 123,003 123,003 40% B (mt) 115,905 115,905 107,627 107,627 35% SSB (mt) 152,580 150,397 133,264 131,374 Total Biomass (mt) 452,941 448,782 401,725 398,804 Max F (=F ) 0.059 0.059 0.057 0.057 abc 40% F (F ) 0.070 0.070 0.068 0.068 ofl 35% Max ABC (mt, yield at F ) 21,656 21,349 18,817 18,630 40% Recommended ABC 21,656 21,349 18,817 18,630 OFL (mt, yield at F ) 25,727 25,363 22,331 22,107 35% EBS Eastern AI Central AI Western AI Total Area apportionment 20.3% 22.4% 22.6% 34.7% 100% ABC (2007) 4,160 4,970 5.050 7,720 21,900 TAC (2007) 2,160 4,970 5,050 7,720 19,900 Catch (2007) 870 5,097 4,660 7,824 18,451 ABC (2008) 4,200 4,900 4,990 7,610 21,700 TAC (2008) 4,200 4,900 4,990 7,610 21,700 ABC (2009) 3,822 4,207 4,261 6,528 18,817 ABC (2010) 3,784 4,165 4,219 6,463 18,630 Responses to the comments of the Statistical and Scientific Committee The SSC December 2007 minutes included the following comments concerning all stock assessments: The SSC notes that the approach for calculating ABC and other biological reference points is not fully described in the SAFE's. It would be desirable to have a general description in the introduction of the SAFE. In each SAFE chapter, specific details could be provided, if the calculation is done differently. For example, the range of years that is used to calculate average recruitment for converting SPR to B 40 should be given. We continue to assume that the equilibrium level of recruitment is equal to the average of age 3 recruits from 1980-2008 (year classes between 1980 and 2005) for Pacific ocean perch as detailed in the Amendment 56 Reference Points section of the Projections and Harvest Alternatives of this stock assessment. The SSC December 2007 minutes included the following comments concerning all rockfish: For all of the rockfish assessments, the SSC recognizes the efforts of the stock assessment authors to respond fully to the 2006 CIE review comments. The SSC requests that the draft response to the CIE review be finalized and made available. The response to the 2006 CIE rockfish review is available online at the following web address: ftp://ftp.afsc.noaa.gov/afsc/public/rockfish/RWG%20response%20to%20CIE%20review.pdf The following comments were made in the December, 2006 meeting of the SSC: Explore model sensitivity to natural mortality estimates in relation to the degree of change allowed for time varying selectivity. Explore alternative priors for natural mortality and evaluate model sensitivity to these changes. Evaluate/compare external estimates of natural mortality to model estimates. To evaluate the model sensitivity to natural mortality estimates in relation to the degree of change allowed for time-varying fishery selectivity, a series of model runs was conducted in which the coefficient of variation (CV) of the prior distribution for the time-varying selectivity parameters was increased, thus relaxing the penalty on the temporal variation in the selectivity parameters. CV values of 0.01, 0.05, 0.25, 0.5, 0.75, and 1.0 were evaluated. As the CV increased a better fit was obtained for the age and length composition data, thus decreasing to total likelihood. Estimates of natural mortality and survey catchability are shown below. The estimates of M were essentially constant at 0.06 due to the prior distribution used for this parameter, and were thus unaffected by the penalties on the time-varying fishery selectivity. In addition, the estimates of M are not expected to be greatly affected by time-varying selectivity because the selectivity for the older ages in the asymptotic selectivity curve does not show much temporal variability, and it is the abundance at these ages that offer the most insight regarding mortality. Interestingly, the estimates of survey catchability increased from about 1.3 to 2.0, thus decreasing the size of the population. The reason for this may be that the decrease in the scale of the population may be necessary to match the age and size composition data while allowing the different fishery selectivity patterns. 0.1 2.5 0.09 0.08 2 ) 0.07 q) M ( ( y y 0.06 1.5 it alit abil ort 0.05 ch M at C ral 0.04 M 1 y u e t q v Na 0.03 ur S 0.02 0.5 0.01 0 0 0 0.2 0.4 0.6 0.8 1 1.2 CV of prior distribution of selectivity parameters To evaluate alternative priors for natural mortality, a series of model runs was conducted in which the CV for the prior distribution for M was increased. CV values of 0.01, 0.05, 0.1, 0.25, 0.5, 0.75, and 1.0 were evaluated. As the CV was increased, the estimate of M increased from 0.05 (the mean of the prior distribution) to 0.13, which was also the estimate when M was estimated freely with any prior distribution. The higher estimates of M correspond to larger population sizes, which require a smaller survey catchability coefficient to fit the survey biomass data (shown below). Model estimates of M > 0.09 are not consistent with model –independent estimates of M (shown below), which range from approximately 0.04-0.09, and thus indicate the utility of using a prior distribution to constrain this parameter. 0.20 2.00 0.18 1.80 0.16 1.60 ) 0.14 1.40 q) M ( ( y ality 0.12 1.20 abilit ort 0.10 1.00 ch M at C ral 0.08 M 0.80 y u e t q v Na 0.06 0.60 ur S 0.04 0.40 0.02 0.20 0.00 0.00 0 0.2 0.4 0.6 0.8 1 1.2 CV of prior distribution natural mortality The expected value for the prior distribution of natural mortality (M) were based upon catch curve estimates for British Columbia POP by Archibald et al. (1981). A review of other empirical methods for estimating M was presented in the Rockfish Working Group response to the CIE review, including empirical relationships between natural mortality and longevity and growth. These techniques are applied to BSAI POP data below to obtain empirical estimates of M A catch curve analysis was applied to the age composition data from the 2000, 2002, 2004, and 2006 AI surveys. The analysis was applied to ages 8-23 to obtain approximately fully selected fish which incurred mortality during the years 1977-1998. An estimate of total mortality (Z) of 0.1041 was obtained from which an estimate of fishing mortality during 1977 to 1998 can be subtracted to obtain M. An estimate of fishing mortality during the period can be obtained from the assessment, which is 0.049. Model-independent estimates of fishing mortality do not exist, but as a worst case scenario we may assume the fishing accounts for ¾ of the total mortality, which would put the estimate of M at ~ 0.078. Given the conservative management of the fishery, it is likely that fishing contributes a smaller proportion of the total mortality, and that M is lower than 0.078. Hoenig (1983) related M to maximum age (t ) for wide variety of species (mollusks, fish, and m cetaceans) and obtained the following regression equation (cid:3) ln(M) = 1.44− 0.82*ln(t ) m Hoenig (1983) also introduced a simplified approach that relies on the relation between M and the proportion of the stock (P) expected to survival to the t under exponential mortality. m − ln(P) M = t m Because the value of P is not known, use of values between 0.01 and 0.05 has been suggested (Quinn and Deriso 1999). Finally, Alverson and Carney (1975) identified a relationship between M, t and von m Bertalanffy growth curve K parameter. 3K (cid:3) M = exp(t*K)− 1 The parameter t* is the “critical” age when an unfished cohort reaches maximum biomass, and was approximated by Alverson and Carney (based on regression analyses) as 0.38 t . m For the empirical relationships above the estimate of M may be sensitive to any outliers in the estimated of tm. A cumulative distribution on the aged POP from all AI trawl surveys was created, and the maximum age as well as the ages corresponding to the 99% and 95% percentiles were used; these values were 104, 59, and 51 years, respectively. The table below summarizes the resulting estimates of natural mortality. T m Method 104 69 51 Hoenig regression 0.0441 0.0660 0.0888 Hoenig simplfied method (with P = 0.01) 0.0443 0.0667 0.0903 Alverson and Carney 0.0005 0.0051 0.0175 INTRODUCTION Pacific ocean perch (Sebastes alutus) inhabit the outer continental shelf and upper slope regions of the North Pacific Ocean and Bering Sea. Pacific ocean perch, and four other associated species of rockfish (northern rockfish, S. polyspinis; rougheye rockfish, S. aleutianus; shortraker rockfish, S. borealis; and sharpchin rockfish, S. zacentrus) were managed as a complex in the two distinct areas from 1979 to 1990. Known as the POP complex, these five species were managed as a single entity with a single TAC (total allowable catch). In 1991, the North Pacific Fishery Management Council separated POP from the other red rockfish in order to provide protection from possible overfishing. Of the five species in the former POP complex, S. alutus has historically been the most abundant rockfish in this region and has contributed most to the commercial rockfish catch. Since 2001, POP in the Bering Sea-Aleutian Islands area have been assessed and managed as a single stock. The rationale for this change is based upon the paucity of data in the EBS upon which to base an age-structured assessment, and the limited amount of data available in 2001 to suggest that the EBS POP represent a discrete stock (Spencer and Ianelli 2001). Information on Stock Structure A variety of types of research can be used to infer stock structure of POP, including age and length compositions, growth patterns and other life-history information, and genetic studies. Spatial differences in age or length compositions can be used to infer differences in recruitment patterns that may correspond to population structure. In Queen Charlotte Sound, British Columbia, Gunderson (1972) found substantial differences in the mean lengths of POP in fishery hauls taken at similar depths which were related to differences in growth rates and concluded that POP likely form aggregations with distinct biological characteristics. In a subsequent study, Gunderson (1977) found differences in size and age composition between Moresby Gully and two other gullies in Queen Charlotte Sound. Westrheim (1970, 1973) recognized “British Columbia” and “Gulf of Alaska” POP stocks off the western coast of Canada based upon spatial differences in length frequencies, age frequencies, and growth patterns observed from a trawl survey. In a study that has influenced management off Alaska, Chikuni (1975) recognized distinct POP stocks in four areas – eastern Pacific (British Columbia), Gulf of Alaska, Aleutian Islands, and Bering Sea. However, Chikuni (1975) states that the eastern Bering Sea (EBS) stock likely receives larvae from both the Gulf of Alaska (GOA) and Aleutian Islands (AI) stock, and the AI stock likely receives larvae from the GOA stock. An alternative approach to evaluating stock structure involves examination of rockfish life- history stages directly. Stock differentiation occurs from separation at key life-history stages. Because many rockfish species are not thought to exhibit large-scale movements as adults, movement to new areas and boundaries of discrete stocks may depend largely upon the pelagic larval and juvenile life-history stages. Simulation modeling of ocean currents in the Alaska region suggest that larval dispersal may occur over very broad areas, and may be dependent on month of parturition (Stockhausen and Herman 2007). In 2002, an analysis of archived Sebastes larvae was undertaken by Dr. Art Kendall; using data collected in 1990 off southeast Alaska (650 larvae) and the AFSC ichthyoplankton database (16,895 Sebastes larvae, collected on 58 cruises from 1972 to 1999). The southeast Alaska larvae all showed the same morph, and were too small to have characteristics that would allow species identification. A preliminary examination of the AFSC ichthyoplankton database indicates that most larvae were collected in the spring, the larvae were widespread in the areas sampled, and most are small (5-7 mm). The larvae were organized into three size classes for analysis: <7.9 mm, 8.0-13.9 mm, and >14.0 mm. A subset of the abundant small larvae was examined, as were all larvae in the medium and large groups. Species identification based on morphological characteristics is difficult because of overlapping characteristics among species, as few rockfish species in the north Pacific have published descriptions of the complete larval developmental series. However, all of the larvae examined could be assigned to four morphs identified by Kendall (1991), where each morph is associated with one or more species. Most of the small larvae examined belong to a single morph, which contains the species S. alutus (POP), S. polyspinus (northern rockfish), and S. ciliatus (dusky rockfish). Some larvae belonged to a second morph which has been identified as S. borealis (shortraker rockfish) in the Bering Sea. Rockfish identification can be aided by studies that combine genetic and morphometric techniques and information has been developed to identify individual species based on allozymes (Seeb and Kendall 1991) and mitochondrial DNA (Gharrett et al. 2001, Rocha-Olivares 1998). The Ocean Carrying Capacity (OCC) field program, conducted by the Auke Bay laboratory, uses surface trawls to collect juvenile salmon and incidentally collects juvenile rockfish. These juvenile rockfish are large enough (approximately 25 mm and larger) to allow extraction of a tissue sample for genetic analysis without impeding morphometric studies. In 2002, species identifications were made for an initial sample of 55 juveniles with both morphometric and genetic techniques. The two techniques showed initial agreement on 39 of the 55 specimens, and the genetic results motivated re-evaluation of some of the morphological species identifications. Forty of the specimens were identified as POP, and showed considerably more morphological variation for this species than previously documented. Because stocks are, by definition, reproductively isolated population units, it is expected that different stocks would show differences in genetic material due to random drift or natural selection. Thus, analysis of genetic material from North Pacific rockfish is currently an active area of research. Seeb and Gunderson (1988) used protein electrophoresis to infer genetic differences based upon differences in allozymes from POP collected from Washington to the Aleutian Islands. Discrete genetic stock groups were not observed, but instead gradual genetic variation occurred that was consistent with the isolation by distance model. The study included several samples in Queen Charlotte Sound where Gunderson (1972, 1977) found differences in size compositions and growth characteristics. Seeb and Gunderson (1988) concluded that the gene flow with Queen Charlotte Sound is sufficient to prevent genetic differentiation, but adult migrations were insufficient to prevent localized differences in length and age compositions. More recent studies of POP using microsatellite DNA revealed population structure at small spatial scales, consistent with the work of Gunderson (1972, 1977). These findings suggest that adult POP do not migrate far from their natal grounds and larvae are entrained by currents in localized retention areas (Withler et al. 2001). Interpretations of stock structure are influenced by the technique used to assess genetic analysis differentiation, as illustrated by the differing conclusions produced from the POP allozyme work of Seeb and Gunderson (1988) and the microsatellite work of Withler et al. (2001). Note that these two techniques assess components of the genome that diverge on very different time scales and that, in this case, microsatellites are much more sensitive to genetic isolation. Protein electrophoresis examines DNA variation only indirectly via allozyme frequencies, and does not recognize situations where differences in DNA may result in identical allozymes (Park and Moran 1994). In addition, many microsatellite loci may be selectively neutral or near-neutral, whereas allozymes are central metabolic pathway enzymes and do not have quite the latitude to produce viable mutations. The mutation rate of microsatellite alleles can be orders of magnitude higher than allozyme locus mutation rates. Most current studies on rockfish genetic population structure involve direct examination of either mitochondrial DNA (mtDNA) or microsatellite DNA. Dr. Anthony Gharrett of the Juneau Center of Fisheries and Ocean Sciences has examined the mtDNA and microsatellite variation for POP samples collected in the GOA and BSAI. The POP mtDNA analysis was performed on 124 fish collected from six regions ranging from southeast Alaska to the Bering Sea slope and central Aleutian Islands. No population structure was observed, as most fish (102) were characterized by a common haplotype. Preliminary results from an analysis of 10 microsatellite loci from the six regions resulted in 7 loci with significant heterogeneity in the distribution of allele frequencies. Additionally, the sample in each region was statistically distinct from those in adjacent regions, suggesting population structure on a relatively fine spatial scale consistent with the results on Gunderson (1972, 1977) and Wither et al. (2001). Ongoing genetic research with POP is focusing on increasing the sample sizes and collection sites for the microsatellite analysis in order to further refine our perception of stock structure. FISHERY POP were highly sought by Japanese and Soviet fisheries and supported a major trawl fishery throughout the 1960s. Catches in the eastern Bering Sea peaked at 47,000 (metric tons, t) in 1961; the peak catch in the Aleutian Islands region occurred in 1965 at 109,100 t. Apparently, these stocks were not productive enough to support such large removals. Catches continued to decline throughout the 1960s and 1970s, reaching their lowest levels in the mid 1980s. With the gradual phase-out of the foreign fishery in the 200-mile U.S. Exclusive Economic Zone (EEZ), a small joint-venture fishery developed but was soon replaced by a domestic fishery by 1990. In 1990 the domestic fishery recorded the highest POP removals since 1977. The history of S. alutus landings since implementation of the Magnuson Fishery Conservation and Management Act (MFCMA) is shown in Table 1. The domestic POP fisheries has been managed with separate ABCs for the BS and AI areas. The ABCs, TACs, and catches from 1988 to 2008 are shown in Table 2. Estimates of retained and discarded POP from the fishery have been available since 1990 (Table 3). The eastern Bering Sea region generally shows a higher discard rate than in the Aleutian Islands region. For the period from 1990 to 2007, the POP discard rate in the eastern Bering Sea averaged 25%, and the 2007 discard rate was 42%. In contrast, the discard rate from 1990 to 2007 in the Aleutian Islands averaged 15%, with a 2007 discard rate of 14%. The removals from trawl and hydoracoustic surveys are shown in Table 4. Historically, POP have been assessed with separate selectivity curves for the foreign and domestic fisheries (Ianelli and Ito 1992), although examination of the distribution of observer catch reveals interannual changes in the depth and areas in which POP are observed to be caught. For example, in the late 1970s and since 1990 POP are predominately taken in depths between 200 m and 300 m, although during the low catch periods of the mid-1980s a large portion of POP were observed to be captured at depths greater than 500 m (Table 5). The area of capture has changed as well; during the late 1970s POP were predominately captured in the western Aleutians, whereas from the early 1980s to the mid-1990s POP were captured predominately in the eastern Aleutians. Establishment of area-specific TACs in the mid-1990s redistributed the POP catch such that about 50% of the current catch is now taken in the western Aleutians (Table 6). Note that the extent to which the patterns of observed catch can be used as a proxy for patterns in total catch is dependent upon the degree to which the observer sampling represents the true fishery. In particular, the proportions of total POP caught that were actually sampled by observers were very low in the foreign fishery, due to low sampling ratio prior to 1984 (Megrey and Wespestad 1990). DATA Fishery Data Catch per unit effort (CPUE) data from Japanese trawl fisheries indicate that POP stock abundance has declined to very low levels in the Aleutian Islands region (Ito 1986). By 1977, CPUE values had dropped by more than 90-95% from those of the early 1960s. Japanese CPUE data after 1977, however, is probably not a good index of stock abundance because most of the fishing effort has been directed to species other than POP. Standardizing and partitioning total groundfish effort into effort directed solely toward POP is extremely difficult. Increased quota restrictions, effort shifts to different target species, and rapid improvements in fishing technology undoubtedly affect our estimates of effective fishing effort. Consequently, we included CPUE data primarily to evaluate its consistency with other sources of information. We used nominal CPUE data for class 8 trawlers in the eastern Bering Sea and Aleutian Islands regions from 1968-1979. During this time period these vessels were known to target on POP (Ito 1982). Length measurements and otoliths read from the EBS and AI management areas were combined to create fishery age/size composition matrices (Table 7). Years that were not selected for age or length composition were rejected due to low samples sizes of fish measured (<300; years 1973-1976, 1985- 1986), and/or otoliths read (<150; years 1984, 1987, 1989). In 1982, the method for aging otoliths at the Alaska Fisheries Science Center changed from surface reading to the break and burn method (Betty Goetz, Alaska Fisheries Science Center, pers. comm.), as the latter method is considered more accurate for older fish (Tagart 1984). The time at which the otoliths collected from 1977 to 1982 were read is not known for many vessels and cruises. However, the information available suggests that otoliths from 1977 to 1980 were read prior to 1981, whereas otoliths from 1981 and 1982 were read after 1982. Survey Data The Aleutian Islands survey biomass estimates were used as an index of abundance for the BSAI POP stock. Since 2000 the survey has occurred biennially, although the 2008 survey was canceled due to a lack of funding. Note that there is wide variability among survey estimates from the portion of the southern Bering Sea portion of the survey (from 165 o W to 170 o W), as the post-1991 coefficients of variation (CVs) range from 0.41 to 0.64 (Table 8). The biomass estimates in this region increased from 1,501 t in 1991 to 18,217 t in 1994, and have since ranged between 12,099 t (1997) and 74,208 t (2004); the 2006 estimate is 23,701 t. The estimated biomass of Pacific ocean perch in the Aleutian Islands management area region (170o W to 170o E) appears to be less variable, with CVs ranging from 0.13 to 0.24. The biomass estimates from the AI trawl survey increased from a low of 82,378 t in 1980 to 625,273 t in 1997. Since 1997, the trawl survey estimates declined to 511,770 t in 2000 and 468,585 t in 2002 before increasing to 576,799 t in 2004 and 667,341 t in 2006. Age composition data exists for each Aleutian Islands survey, and the length measurements and otoliths read are shown in Table 9. Historically, the Aleutian Island surveys have indicated higher abundances in the Western (543) and Central (542) Aleutian Islands, and this pattern was repeated in the 2006 survey (Figure 1). In particular, areas near Amchitka and Kiska Islands, Tahoma Bank-Buldir Island, and Attu Island and Stalemate Bank showed high CPUE in 2006 survey tows. In the central Aleutians, large tows were observed in near Adak Island and the Delarof Islands. The biennial EBS slope survey was initiated in 2002. The most recent slope survey prior to 2002, excluding some preliminary tows in 2000 intended for evaluating survey gear, was in 1991, and previous slope survey results have not been used in the BSAI model due to high CVs, relatively small population sizes compared to the AI biomass estimates, and lack of recent surveys. The 2008 EBS slope survey was completed, but the 2006 survey was canceled due to lack of funding. The survey biomass estimates of POP from the 2002, 2004, and 2008 surveys were 76,665 t, 112,273 t, and 111,302 t, respectively, with CVs of 0.53, 0.38, and 0.40. The slope survey results are not used in this assessment, and the feasibility of incorporating this time series will be evaluated in future years.