CONTENTS Abstract Introduction The Fishery Stock Abundance and Biomass Indices Maturity Mortality Estimation of Fishing Mortality Rates and Stock Size Biological Reference Points Short-Term Projections Conclusions Acknowledgments Literature Cited
Northeast Fisheries Science Center Reference Document 03-14
S. E. Wigley, J. K.T. Brodziak and L. Col
A Report of the 37th Northeast Regional Stock Assessment Workshop: Assessment of the Gulf of Maine and Georges Bank witch flounder stock for 2003
National Marine Fisheries Serv., Woods Hole Lab., 166 Water St., Woods Hole, MA 02543
Web version posted September 19, 2003Citation: Wigley, S.E.; Brodziak, J.K.T.; Col, L. 2003. Assessment of the Gulf of Maine and Georges Bank witch flounder stock for 2003. Northeast Fish. Sci. Cent. Ref. Doc. 03-14; 186 p.
Information Quality Act Compliance: In accordance with section 515 of Public Law 106-554, the Northeast Fisheries Science Center completed both technical and policy reviews for this report. These predissemination reviews are on file at the NEFSC Editorial Office.ABSTRACT
The 2003 analytical assessment for witch flounder in USA waters, covering the period 1982- 2002, estimates 2002 fishing mortality and spawning stock biomass for stock status, and provides short-term projections of median landings, discards and spawning stock biomass for various fishing mortality scenarios.
USA commercial landings increased during the 1960's from 1,200 mt to about 3,000 mt, then fluctuated between 2,000 and 3,000 mt until 1983 through 1985 when landings abruptly increased to about 6,000 mt. Landings subsequently declined to 1,500 mt in 1990. Since the early 1990s, landings have fluctuated between 2,000 and 3,000 mt. In 2002, USA commercial landings totaled 3,186 mt, a 5% increase over 2001; and 117% higher than in 1990, the lowest value since 1964. Research survey indices of abundance and biomass remained fairly stable from 1963 until the late 1970s; autumn indices declined during the early and mid 1980s, reaching record low levels in late 1980s and early 1990s. Abundance sharply increased in 1993 and has continued to increased to near record high levels in 2002; however, the age composition still remains truncated.
The VPA indicates that fishing mortality (ages 8-9, unweighted) increased from 0.26 in 1982 to 0.67 in 1985, declined to 0.22 in 1992, increased to 1.13 in 1996, then declined to 0.41 in 2002. Spawning stock biomass declined from 16,897 mt in 1982 to about 3,800 mt in 1996. With recent increases in recruitment and declines in fishing mortality, SSB has increased to 18,296 mt in 2002 . Since 1982 recruitment of age 3 has ranged from approximately 3 million fish (1984 year class) to 67.6 million fish (1997 year class). Over the 1982-2002 period, average recruitment of age 3 fish (the 1979 - 2000 year classes) was 19.6 million (the geometric mean equaled 14.4 million fish). The 1995-1999 year classes appear to be above average, and the 1997 year class is the largest in the VPA time series. In addition to the VPA, an alternative model, a statistical catch at age model (SCAA) was also conducted. The SCAA generally confirmed trends in the VPA results.
The biological reference points were updated by applying the approach used to estimate MSY proxies for witch flounder. Fmsy is approximated as F40% (0.23), the SSBmsy proxy is 25,248 mt, the product of 40%MSP (1.2882 kg spawning biomass) and average long-term recruitment (19.6 million). The MSY proxy is 4,375 mt, the product of yield per recruit at F40% ( 0.2232 kg) and average recruitment.
Based on the ADAPT VPA, the witch flounder stock was not overfished, but overfishing was occurring in 2002. Fully recruited fishing mortality in 2002 was 0.41, nearly double Fmsy (0.23), and spawning stock biomass was estimated to be 18,296 mt in 2002, 72% of SSBmsy (25,248 mt). Recent year classes appear to be above average. Although the spawning stock biomass has increased, the age structure still remains truncated. Fishing mortality should be reduced to Fmsy or below to allow the age structure to rebuild.
The witch flounder (Glyptocephalus cynoglossus, L.) or grey sole is a deep water boreal flatfish occurring on both sides of the North Atlantic. In the Northwest Atlantic, witch flounder are distributed from Labrador to Georges Bank and in continental slope waters southward to Cape Hatteras, North Carolina. In U.S. waters, the species is commercially abundant in the Gulf of Maine-Georges Bank region [defined as Northeast Fisheries Science Center (NEFSC) Statistical Reporting Areas (SA) 511-515, 521-522, 525-526, and 561-562; Figure 1], and, in the absence of any stock structure information, is assumed to comprise a single stock unit. Prized as a table fish, witch flounder receives a high ex-vessel price relative to other flounders and represents an important by-catch component in the New England mixed species groundfish fishery. Annual landings during the period 1960-1982 averaged 3,000 metric tons (mt), ranging from 1,000 to 6,000 mt (Lange and Lux 1978, Burnett and Clark 1983). More recently, landings declined from a peak of 6,660 mt in 1984 to a low of 1,490 mt in 1990. Landings for 2002 were 3,186 mt.
Previous witch flounder stock assessments were conducted by Burnett and Clark (1983), Wigley and Mayo (1994) and Wigley et al. (1999). An assessment update was conducted for this stock in 2002 and reviewed at Groundfish Assessment Review Meeting (GARM; NEFSC 2002). The GARM assessment indicated average fishing mortality (ages 7-9, unweighted) increased from 0.21 in 1982 to 0.59 in 1985, declined to 0.24 in 1990, increased to 0.96 in 1996, then declined to 0.45 in 2001. Spawning stock biomass declined from 18,000 tons in 1982 to about 4,000 tons in 1995 and then increased to 11,368 mt in 2001. Since 1982, recruitment at age 3 has ranged from approximately 3 million fish (1984 year class) to 84 million fish (1997 year class) with a mean of 22 million fish. In 2001, the SSB was slightly above ½ SSBmsy (9,950 mt), the minimum stock size threshold, and fishing mortality (F= 0.45) was three times higher than Fmsy, the maximum fishing mortality threshold; thus, witch flounder was not overfished but overfishing was occurring in 2001.
This assessment of witch flounder in the Gulf of Maine - Georges Bank region and southward (USA portions of NAFO Subareas 4, 5 and 6), presents a benchmark analytical assessment for the stock for the 1982-2002 period, estimates 2002 fishing mortality and spawning stock biomass for stock status, and provides short-term projections of median landings, discards and spawning stock biomass for various fishing mortality scenarios. This assessment provides estimates of discards from the shrimp fishery and large-mesh otter trawl fishery based upon analyses of sea sampling, commercial and research vessel survey data through 2002.
Witch flounder is managed under the New England Fishery Management Council's Multi-species Fisheries Management plan since 1987. A brief summary of groundfish management regulations affecting witch flounder is presented in Table 1. Significant changes in regulations include increased minimum size in 1983 and 1987; increases in mesh size in 1982, 1983, 1994, 1999; effort reductions in 1996 and 2002; and implementation of closed areas in 1994 and 1998 (Figure 2). The western Gulf of Maine area closure, Cashes Ledge area closure and the seasonal rolling closures overlap the witch flounder distribution (Figure 3 and Figure 4). Management regulations for the northern shrimp fishery also impact witch flounder (Table 2); significant changes in the shrimp fishery include a monthly 10% by-catch limit which restricted the possession of groundfish to 10% by weight of shrimp in the mid-1980's to early 1990s; and the implementation of the Nordmore grate to exclude groundfish in 1992.
There is no recreational fishery for witch flounder.
USA commercial landings in 2002 totaled 3,186 mt, a 5% increase over 2001 (Table 3); and 117% higher than in 1990, the lowest value since 1964 (Figure 5). Canadian landings from the stock have been negligible (32 mt in 2001; Table 3). Landings from the Grand Banks (NAFO Divisions 3LNO) during 1985 to 1990 are not included in this assessment. Canadian landings from the western Scotian Shelf (NAFO Division 4X) are not considered due to the fact that, until recently, witch flounder were reported as 'other flounders' by Canada, and cannot be separated from other flounder species. Furthermore, samples from the western Scotian Shelf indicate slower growth of witch flounder than in the Gulf of Maine, suggesting a different phenotypic stock.
The western Gulf of Maine (SA 513 and 514) and the central basin (SA 515) provide nearly a third of the USA witch flounder landings (Table 4); landings from Georges Bank are confined to the deeper waters north of the South Channel (SA 521, 522; Table 4). Otter trawl catches account for about 98% of witch flounder landings, with sink gillnets comprising the remainder (Table 5). Catches are generally highest during March-July (Table 6) when witch flounder form dense pre-spawning aggregations (Burnett et al. 1992). The majority of witch flounder are landed in Maine ports, primarily Portland, with lesser amounts landed in New Bedford and Gloucester, MA (Table 7).
Although culling and grading practices vary by port, witch flounder have historically been landed as either 'small' or 'large'; however, three market categories ('peewee', 'medium', and 'jumbo') were added in some ports beginning in 1982 (Table 8, Figure 6). Since the early 1990s, the proportions of witch flounder landings from the peewee and small market categories have steadily increased. In 2002, witch flounder less than 45 cm ('peewee' and 'small' market categories) constituted 87% of total landings (Table 8, Figure 6). The current regulated minimum landing size for witch flounder is 36 cm (14 inches).
Length frequency and age sampling data for witch flounder landings from the Gulf of Maine-Georges Bank region are summarized by quarter and market category in Table 9 (because some ports do not cull into 'peewee' or 'jumbo' categories, NEFSC sampling protocols incorporate these categories into the 'small' and 'large' categories, respectively). Until 1982, sampling was minimal and sporadic. During 1982-1988, an average of 48 length frequency samples (approximately 100 fish per sample) was obtained annually over all market categories, representing 1 sample per 102 mt landed. In 1990, sampling requirements were adjusted to 1 sample per 50 mt to obtain more samples from the 'large' market category. However, samples for the 'large' market category have been difficult to obtain due to the sharp decrease in the landings of larger fish in recent years (Table 8 and Table 9). Sampling intensity during 2001-2002 averaged 39 samples annually, representing 1 sample per 80 mt landed; nonetheless, even with this increased sampling intensity, inadequate numbers of samples were obtained for some market categories and quarter combinations. In 2002, of the 35 samples collected, 15 were small samples (43%), 10 were medium (29%) and 10 were large (29%). Compared with the 2002 market category landings distribution by weight (small 87%; medium: 10%; large: 3%), sampling in 2002 adequately approximated the market category distribution of landings on an annual basis. As in previous years, it was necessary to pool some quarters for some market categories. A summary of pooling procedures by year, market category and quarter is presented in Table 10.
Commercial Landings at Age
Commercial age data for the years 1982 to 2002 were available for this assessment. Quarterly age-length keys (ALKs) were applied to corresponding commercial landings length frequency data by market category. Resulting estimates of annual age compositions (age 0 to 14+) are presented in Table 11 and Figure 7. No discernible changes in growth are evident during the 1982-2002 period; although landings mean weights and mean lengths at ages 6 to 8 declined in 1996-2002, this may be an artifact of poor sampling in recent years.
The Fisheries Observer Program (FOP), which began in 1989, has generated various levels of coverage for different fisheries. Prior to the FOP, NEFSC conducted sea sampling on an ad-hoc basis. The northern shrimp fishery, the small-mesh otter trawl fishery, and the large-mesh otter trawl fishery are three fisheries in which discarding of witch flounder occurs. In this assessment, discard estimates have been estimated for the shrimp fishery and the large-mesh otter trawl fishery.
Northern shrimp fishery
Since the 'shrimp season' spans a calendar year, in this report, the year in which most of the fishing occurred will be used to identify the entire season. For example, 1990 will refer to the shrimp season from December 1, 1989 to May 31, 1990. An overview of the procedures used to estimate witch flounder discarded in the shrimp fishery from 1982 to 1998 is presented in Figure 8. These estimation procedures were used in the 1994 assessment (Wigley and Mayo 1994), reviewed by the SAW 18 (NEFSC 1994), and extended through 1997 using the same methodology. The ratio of witch flounder discarded (kg) to days fished was calculated using FOP data for individual shrimp seasons, 1989-1997, by fishing zone. Since depth is an important factor influencing discards (Wigley MS 1994), discard ratios were calculated for each of three fishing zones (zone 1 = 0-3 miles from shore, zone 2 = 3 - 12 miles, and zone 3 = greater than 12 miles) in each season. For the most part, fishing zones are analogous to depth zones. Statistical testing of zonal discard rates indicated differences between fishing zones in most years. The zone-specific discard rates were weighted by the days fished in each zone to calculate a weighted mean discard rate for each season (Table 12). To estimate witch flounder discard rates prior to the FOP, (i.e., 1982-1988), a simple linear regression was employed using 1989-1992 (years in which the Nordmore grate was not required) weighted mean discard rates and annual indices of witch flounder abundance. The NEFSC autumn bottom trawl survey index of age 3 fish was found to be the best predictor of annual discard rates (r2 = 0.97, p = 0.0127; Figure 9; Wigley MS 1994).
With no 1998-2002 FOP sampling in the northern shrimp fishery, an alternative method of survey filtering was explored to estimate witch flounder discard rates; however, due to insufficient length frequency data at small sizes, this method did not prove fruitful. As used for the years prior to the FOP, a simple linear regression using 1993-1997 (years in which the Nordmore grate was required) annual shrimp season discard rates and annual survey indices of autumn age 3 fish was employed (r2 = 0.87, p = 0.0206). This five-point regression may not be as robust as the r2 suggests, as four of the points are clustered (Figure 9).
To obtain total weight of witch flounder discarded during a shrimp season, season discard rates (kg per day fished) were multiplied by the total number of days fished by the commercial fleet in each season (Table 13). Estimated discard weight was then translated into discarded numbers at age by applying witch flounder sea-sampled discard length-frequencies (Figure 10) expanded up to the total discard weight and then applying NEFSC spring bottom trawl survey ALKs. Detailed information on this method is given in Wigley (MS 1994). For 1995-2002, days fished were estimated from the Vessel Trip Reports (VTR) using a stratification level of year, ton class, port group, month, and fishing zone. To derive the number of trips by fishing zone, the proportion of VTR trips by fishing zone was applied to the number of trips in the weighout database. Days fished per trip in each fishing zone were derived from the VTR data. Days fished per trip were then multiplied by the estimated number of trips for each fishing zone to derive estimated days fished by fishing zone, and then summed over year and fishing zone.
For the 1982-1997 time period, discard estimates of numbers at age and weight were derived on a shrimp season basis due to the limited number of length frequency samples in December. To adjust the shrimp fishery discard-at-age from a shrimp season basis to calendar year, the ratio of December days fished to the entire shrimp season days fished was used to apportion of the weight and numbers discarded into December and January-May categories. The December discards-at-age were shifted back one age, and then re-combined with the January-May matrix of the corresponding calendar year. The December discard weight was combined with January-May of the same calendar year. Mean lengths and mean weights at age in the re-combined catch at age were weighted by the numbers at age from each category.
Without 1998-2002 FOP sampling, discard length-frequency data were unavailable to partition the 1998-2002 estimated discard weight into numbers at length; thus discarded numbers at age were derived by apportioning discard weight by the average age composition (calendar year) of discards in 1993-1997 and then dividing by the average 1993-1997 discard mean weights at age (Figure 11). The average 1993-1997 mean weights at age from the FOP were consistent with trends in mean weights from the NEFSC survey during the 1998-2002 time period (Figure 11).
Witch flounder discards in the northern shrimp fishery ranged from a low 0.8 mt in 2002 to a high of 34 mt in 1988 and 1995 (Table 13). Similarly, number of witch flounder discarded ranged from 40,000 fish discarded in 2002 to 1.8 million fish in 1994 (Table 13). Estimates of age compositions of discarded witch flounder in the shrimp fishery are presented in Table 14 and Figure 12. Discarded witch flounder from the shrimp fishery range from age 0 to 6, with ages 1 to 3 most commonly discarded (Table 14 and Figure 12).
Large-mesh otter trawl fishery
Discard estimation from the large-mesh otter trawl fishery is confounded by the lack of FOP coverage prior to 1989, sparse coverage in the beginning of the program, and the recent implementation of year-around and seasonal area closures. As a result, three estimation scenarios were examined (Table 15): 1) utilizing a survey filter method (Table 16); 2) utilizing the at-sea observer data (Table 17); and 3) utilizing the Vessel Trip Report data (Table 18). The estimated discards (in weight and numbers) are presented in Table 19. Each method is described below.
The method used in previous witch flounder assessments to estimate large-mesh otter trawl discards was based upon a method developed by Mayo et al. (1992) which utilizes survey and commercial catch at length data, commercial gear retention ogives, and information on culling practices. An overview of the method is presented in Figure 13. Research vessel length frequency data were filtered through commercial gear retention ogives corresponding to the predominant mesh size employed in the large-mesh fishery (130, 140, and 152 mm) and then through a culling practice ogive. Due to the sparse gear retention studies for witch flounder, mesh selection ogives were taken from Walsh et al. (1992) for American plaice. Given the high value and low abundance of this species, the culling practice of commercial fishermen was assumed to be nearly knife edged at the minimum landing size. A semi-annual ratio estimator of survey filtered 'kept' index to semi-annual numbers landed was used to expand the estimated 'discard' survey index to obtain numbers of fish discarded at length. The method used in this analysis differs from the method described by Mayo et al (1992) which employs an expansion factor derived from a linear regression from the ratios of kept to landed at length. A spreadsheet illustrating the method used is presented in Table 16 for 1993 using the spring survey and commercial landings from quarters 1 and 2. Semi-annual numbers of discard fish at length were apportioned to age using the corresponding season NEFSC ALK. Estimated numbers of discarded witch flounder in the large-mesh otter trawl fishery are presented by season in Table 19. Results indicate that in recent years, numbers discarded at sea comprised as much as 54% of the witch flounder landed. The general pattern of discarding appears to be consistent with that expected given strong recruitment during 1979-1981 and the mid-1990s.
Given the distribution of juvenile witch flounder in the western Gulf of Maine and the recent implementation of year-around area closures and seasonal rolling area closures in the western Gulf of Maine, there was some concern regarding the application of the survey filter method to estimate discards in recent years. Since the commercial fishery does not have year-around access to the population estimated by the NEFSC survey, it may be inappropriate to use the survey filter method to estimate discards. For the 1989-2002 period, discard weight to kept weight ratios (D/K ratio) were calculated from FOP data on a semi-annual basis (Table 17). Total discard weight was derived by multiplying the D/K ratio by the commercial landings. The number of sea sampled trips varied from no trips in the second half of 1992 to 83 trips in the second half of 2002. The D/K ratios ranged between 0.02 and 0.50. Given the limited number of trips, tows and available discard length frequencies, discards at age were derived only for the 1995-2002 time period (Table 19).
The Vessel Trip Report data were explored for information on discarding of witch flounder. Reporting of discard information in the logbooks is known to be incomplete. To eliminate problems associated with incomplete reporting, a subset of the VTR data was used. The VTR subset included only logbooks which reported discards of any species (Delong et al. 1997), assuming that operators who report discards of any species would reliably report witch flounder discards. This subset was used to estimate discard ratios (discard weight/kept weight) semi-annually for large-mesh otter trawl gear from 1994 to 2002. Limitations of this analysis are: 1) the dealer data used to expand discard rates to total discard weight do not contain information on mesh size, precluding partitioning of otter trawl fisheries into small and large mesh trips; 2) there is no area information on dealer data to isolate trips from the Gulf of Maine-Georges Bank region. From this analysis, results suggest that discard rates range between 4% and 9% (Table 18). These estimates should be reviewed cautiously as not all fishermen report discards. Discarded numbers at age were estimated by expanding the FOP length frequencies and applying the survey age/length keys (Table 19).
For estimates of total catch at age, discards from the large-mesh otter trawl fishery were derived using the survey filter method from 1982-1994 and using the FOP method for 1995-2002 (Table 20 and Figure 14). Total discards at age (from the shrimp fishery and large-mesh otter trawl fishery combined) are presented in Figure 15.
Total Catch at Age
Total catch at age compositions (including commercial landings, discards from the northern shrimp fishery and the large-mesh otter trawl fishery) are presented in Table 21 and Figure 16. The age composition data reveal strong 1979-1981 year classes (Table 21). The 1989 and 1993 year classes also appear to have been strong; however, these cohorts were heavily discarded in both the shrimp and large-mesh otter trawl fisheries (Table 14 and Table 20; Figure 12 and Figure 14). The poor 1984 year class is also evident as well as the truncated age-structure since the early 1990s.
Since witch flounder landings are highest during March-July, the average weights-at-age in the catch approximate mid-year weights. Mean weights at age at the beginning of the year (January 1; Table 22) were derived from mid-year weights using procedures described by Rivard (1980).
STOCK ABUNDANCE AND BIOMASS INDICES
Commercial catch rates (landings per unit effort, LPUE, expressed as landings in mt per day fished) were derived for vessel tonnage classes 2-4 [Class 2 consists of vessels 5 to 50 gross registered tons (GRT); Class 3, 51 to 150 GRT; and Class 4, 151 to 500 GRT]. These vessel classes account for greater than 95% of annual witch flounder otter trawl landings. LPUE indices for the Georges Bank-Gulf of Maine region were computed for: 1) all trips landing witch flounder, and 2) trips in which 40% or more of the total landings comprised witch flounder (Table 23). These '40% trips' may represent effort that is 'directed' towards witch flounder, a species historically taken as by-catch.
For all trips landing witch flounder, increases in LPUE occurred in 1977-1978 for tonnage classes 2 and 3 and in 1982 for tonnage class 4, and remained high during the early 1980s; however, LPUE indices declined steadily for all tonnage classes from 1986 to 1990. Since the early 1990s, LPUE indices have steadily increased and are among the highest in the time series (Table 23, Figure 17a). Indices for 40% trips peaked in the early 1980s , then declined to a low in 1994, and have increased slightly in recent years (Table 23, Figure 17a). Effort (days fished) associated with all trips and 40% trips increased during the late 1970s and early 1980s, peaked during 1985-1988, and have generally declined since (Figure 17b). While there is some evidence of increased directed effort in the early and mid 1980s [a period in which both witch flounder and American plaice were abundant and a small directed fishery emerged (Burnett and Clark 1983)], it is likely that LPUE indices derived for all trips landing witch flounder provide the best measure of relative abundance. In 1994 the NEFSC commercial data collection system changed from a voluntary to a mandatory system in which fishermen self-report fishing effort. Investigation is still on-going to determine if the time series of LPUE data can be extended (considered one series) or whether the post 1993 LPUE derived under the mandatory system constitutes a separate time series. Effort (days fished) for 1994 to 2002 may be underestimated in this report since effort is based upon preliminary VTR data, which do not represent 100% of the trips.
Research Vessel Survey Indices
The NEFSC has conducted annual research vessel stratified random bottom trawl surveys during autumn since 1963 and during spring since 1968. Details on survey sampling design and the use of survey data in stock assessments are given in Azarovitz (1981) and Clark (1981), respectively.
In September 2002, an offset in the trawl wraps was detected which may have effected the NEFSC bottom trawl surveys conducted from winter 2000 to the spring 2002. Extensive analyses of existing data sets and experimental studies were conducted to evaluate the offset issue (NEFSC 2002). These analyses were reviewed by a panel of experts and they concluded that no adjustments to the survey time series were justified (Groundfish Science Peer Review, 2003).
The Commonwealth of Massachusetts Division of Marine Fisheries (DMF) began an inshore trawl survey in 1978 which complements the NEFSC survey in coastal Massachusetts waters in that depths less than 27 meters (the lower depth limit sampled by the NEFSC offshore survey) are sampled (for details of this survey, see Howe et al. 1981). Additionally, the Northern Shrimp Technical Committee of the Atlantic States Marine Fisheries Commission (ASFMC) has conducted an annual northern shrimp survey during August in the Gulf of Maine since 1983, with catch data for witch flounder available from 1984 on (for details of the shrimp survey, see Northern Shrimp Technical Committee MS 1984). All three surveys provide useful information relative to trends in abundance, distribution, and recruitment of witch flounder in the Gulf of Maine-Georges Bank region. Strata utilized in the derivation of indices of relative abundance and biomass for witch flounder are as follows: NEFSC, offshore strata 22-30, 36-40 (Figure 18); Massachusetts DMF, regions 4 and 5 (Figure 19); and northern shrimp, strata 1, 3, 6, and 8 (Figure 20).
Witch flounder are generally distributed throughout the Gulf of Maine, along the Northern Edge and southern flank of Georges Bank, and southward along the continental shelf as far south as Cape Hatteras, NC (Figures 3a and 3b). Juvenile witch flounder (<25 cm) are distributed along the western Gulf of Maine, with a few in the canyon areas in the Mid-Atlantic region (Figures 4a and 4b). Concentrations of witch flounder along the western portion of the Gulf of Maine are observed in the ASMFC shrimp survey. Although this survey has limited spatial coverage (Figure 21), most of the juvenile range is covered.
In response to a research recommendation from SARC 29, analyses were conducted to examine if the use of additional strata in the NEFSC bottom trawl survey might be appropriate. Burnett and Clark (1983) used NEFSC survey strata set 22,24,26-30, 33-40 in the first witch flounder assessment; however, Burnett (MS 1987) suggested that fish from strata 33, 34 and 35 exhibited different growth rates indicating these fish may be from a different stock inhabiting the western Scotian shelf. Based on this information, Wigley and Mayo (1994) revised the witch flounder survey strata set excluding 33, 34, and 35, and included strata 23 and 25. Following a method developed by Cadrin (2003), witch flounder catches for the entire autumn bottom trawl survey time series were examined by individual stratum. The stratified mean number per tow in each stratum was summed over the time period, and the percentage contribution of each stratum was calculated as well as the percentage of annual stratum sampling which produced no catch (Table 24). Results indicate that the current strata set (22-30, 36-40) accounts for approximately 93% of the survey catch and that only minor differences exist between the strata sets used in previous assessments. This analysis also indicated that stratum 6 contributed to the overall witch flounder catch. The stratified mean weight (kg) per tow was calculated for three strata sets: set 1 (22-30, 36-40); set 2 (22, 24, 26-30, 36-40); and set 3 (6, 22, 24, 26-30, 36-40). The trends of these biomass indices (and their variance) are indistinguishable (Figures 22a and 22b). The inclusion of stratum 6 is not justified due to its geographical discontinuity with the core strata. Since no additional strata were identified as contributing to the total catch, or improved the precision of the estimates of mean weight per tow, the strata set 22-30, 36-40 will continue to be used.
Research vessel survey indices of abundance, biomass, and mean length for NEFSC surveys, Mass. DMF surveys, and ASMFC shrimp surveys are presented in Tables 25-27 and Figure 23, Figure 24, Figure 25, Figure 26, Figure 27, and Figure 28, respectively. Length frequency data from these surveys are presented in Figure 29, Figure 30, Figure 31, and Figure 32. A summary of available age data from NEFSC surveys is given in Table 28; survey age samples collected during 1976 to 1979 have not been aged. Too few age samples are collected during DMF surveys to reliably characterize the age composition of witch flounder in the inshore areas, and no age samples are collected on ASMFC surveys. Age-specific relative abundance indices from NEFSC spring and autumn surveys 1980-2002, and preliminary spring 2003 are presented in Table 29, Figure 33 and Figure 34. Mean length and mean weights at age from the NEFSC spring and autumn surveys area given in Tables 30 and Table 31 and Figure 35, Figure 36, and Figure 37.
While NEFSC spring survey indices tend to be more variable due to the pre-spawning aggregations of witch flounder, spring and autumn indices generally display similar trends. Abundance and biomass remained fairly stable from 1963 until the late 1970s (Table 25, Figure 23 and Figure 24); autumn indices declined during the early and mid 1980s, reaching record low levels in 1987. Abundance sharply increased in 1993, due to a large age 0 index (Table 29, Figure 34) and has continued to increased to near record high levels in 2002. During the same time, mean length declined (Figure 27 and Figure 28). The age structure has been truncated since the late 1980's (Figure 33 and Figure 34).
Length frequency data from the ASMFC shrimp survey suggest that incoming year classes can be identified prior to their appearance in the NEFSC surveys. Thus, the ASMFC survey appears to be more useful in providing a pre-recruit index than in characterizing the population as a whole (Table 27, Figure 26 and Figure 32). The ASMFC survey data indicate improved recruitment in recent years (see length frequency modes at 12 cm, corresponding to age 1 fish, during 1991-1994, 1997, and 1999 (Figure 32). Significant numbers of small fish were also observed in the NEFSC autumn survey during the same years (Figure 30).
Mean lengths at age from NEFSC spring and autumn surveys are presented in Table 30 and for ages 4 to 8 in Figures 35a and 35b. Mean lengths at age for ages 5 to 7 appear to have increased approximately 3-5 cm from 1980 to the late 1980's, and then declined (Figures 35a and 35b); however, Von Bertalanffy growth analyses detected no significant changes in resulting growth parameters over the time period.
NEFSC spring and autumn survey mean weights at age are given in Table 31 and Figure 36 and Figure 37. Survey mean weights are variable, however, similar declines in mean weights for ages 6-9 were observed during the mid-1990s to 2002 in both the commercial landings and spring and autumn surveys.
Witch flounder maturity observations have been collected on the NEFSC research bottom trawl surveys since 1977. The NEFSC spring surveys were used for maturity analyses as these surveys occur closest to and prior to spawning (Halliday 1987). In the previous witch flounder assessment, probit analyses (SAS 1985) of maturity at age data revealed that there have been six maturity stanzas over the assessment period (GARM NEFSC 2002). The proportion at which 50% of the fish are mature at age (A50) was significantly different for the time periods 1980-1982, 1983-1984, 1985-1990, 1991-1993, 1994-1999, and 2000-2002. Due to small sample sizes, it was necessary to pool individual years, however, individual years were examined, and then pooled into time blocks. Trends in female A50 and L50 were similar, progressively decreasing from 1980-1982 to 1985-1990, then increasing in 1991-1993, then declining in 1994-1999 and increasing in 2000-2002 similar to 1983-1984 levels. The maturity stanzas used revealed sharp changes in proportion mature, uncharacteristic of the assumed gradual biological process. The maturity stanzas also revealed, in a few instances, biologically infeasible outcomes, i.e. over the life span of a cohort, the proportion mature at age would decrease. Given these issues, a method which has been applied to Georges Bank cod (L. O'Brien, NEFSC, pers. comm.) was employed to minimize the abrupt changes yet still capture the changing trends in maturity over time. This method used logistic regression and a five-year moving time block to estimate annual maturity ogives. For example, the proportion mature in 1982 was estimated using NEFSC spring maturity data from 1980, 1981, 1982, 1983 and 1984. Likewise, the 1983 maturity ogive used maturity data from 1981 to 1985. Annual maturity ogives were derived for 1982 to 2001 using 1980 - 2003 data. The annual 2002 maturity ogive was assumed to be equal to the 2001 ogive (Table 32, Figure 38). In addition to the annual maturity ogives, a single ogive using maturity data from the entire time series was also calculated (Table 32). It was concluded that the moving time block method was appropriate for use in the VPA.
Stratified mean weight per tow of mature (spawning stock) witch flounder was calculated for spring NEFSC research vessel surveys (Table 33, Figure 39) using the six maturity stanzas. This analysis will be updated to incorporate the moving time-block maturity estimates in the next assessment update. The spawning stock biomass indices closely track total biomass indices except in most recent years, indicating a larger proportion of immature fish in the population.
Burnett (MS 1987) estimated instantaneous natural mortality (M) to be 0.16 from a regression of survey-derived instantaneous total mortality (Z) estimates on commercial fishing effort. Halliday (1973) used a value of M = 0.15 for females and M = 0.2 for males in an assessment of Scotian Shelf witch flounder. In the present study, virtual population analyses, yield per recruit and spawning stock biomass per recruit analyses were performed assuming M = 0.15.
Estimates of instantaneous total mortality (Z) were computed from NEFSC spring and autumn research vessel bottom trawl survey catch per tow at age data by combining cohorts over the following time periods: 1982-1985, 1986-1989,1990-1993, 1994-1997 and 1997-2001. Given the variability in age at full recruitment to the sampling gear observed during the survey time series (Table 34), estimates were derived for each time period and each season by taking the natural logarithm of the ratio of pooled age 7+ to pooled 8+. For example, the estimate of Z for 1982-1985 was computed as:
Spring: ln (sum age 7+ for 1982-1985 / sum age 8+ 1983-1986) Autumn: ln (sum age 6+ for 1981-1984 / sum age 7+ 1982-1985).
To evaluate Z over identical year classes within each of the survey series, different age groups were used in the spring and autumn.
Total mortality estimates from the two survey series exhibited similar trends, although autumn estimates were generally lower than those in the spring (Table 34 and Figure 40a). With no objective basis to select one survey series over another, total mortality was calculated by taking the geometric mean of the spring and autumn estimates during each time period. Total mortality ranged between 0.34 and 0.71 over the time series (Table 34). Additionally, annual estimates of total mortality were calculated, and smoothed with a three year moving average (Figure 40b).
ESTIMATION OF FISHING MORTALITY RATES AND STOCK SIZE
Virtual Population Analysis and Calibration
The ADAPT calibration method (Parrack 1986, Gavaris 1988, Conser and Powers 1990) was applied to estimate abundance at age in 2003 using catch-at-age estimates (i.e., landings plus discards from the shrimp and large-mesh otter trawl fishery; Table 21). Estimates of stock sizes, their associated statistics, and F in the terminal year are summarized in the Table 35.
New VPA software is now available in the NOAA Fisheries Toolbox. To bridge the transition between the software used in the last assessment update (FACT 1.5) and the current software, NFTv2.0.11, the accepted 2002 VPA (NEFSC 2002) formulation and input data was re-run using the NFTv2.0.11 software. The summary statistics of the two VPAs (RUN 61-f) reveal only slight changes in stock size estimates and fishing mortality (Table 35), and these minor changes are attributed to the use of the exact catch equation and other improvements in precision.
An initial formulation (RUN 100) based upon the 2002 VPA was performed to estimate 2003 stock sizes for ages 4 to 10 (Table 35) using a catch-at-age matrix including ages 3-11+ and NEFSC spring and autumn abundance indices for ages 3 to 11+ as tuning indices. All indices were given equal weighting. Autumn survey indices were lagged forward one year and one age to calibrate with beginning year population sizes of the subsequent year. A flat-top partial recruitment (PR) pattern was assumed, with full fishing mortality on ages 7 and older. The F on ages 10 and 11+ in the terminal year was estimated as the average of F on ages 7 through 9. The F on ages 10 and 11+ in all years prior to the terminal year was derived from weighted estimates of Z for ages 7 through 9. Instantaneous rate of natural mortality (M) was assumed to be 0.15. Spawning stock biomass (SSB) was calculated at time of spawning (March) and mean weight at age calculated by the Rivard method (Table 22).
The results of the initial run indicated that coefficients of variation (CV) for estimated ages ranged between 29% and 44% and the CVs for survey catchability coefficients (q) were consistent, ranging from 11% to 27%.
Two alternative formulations included: 1) using a total catch at age in which large-mesh otter trawl discards were estimated using the survey filter method for 1982-1994 and Fisheries Observer Program data for 1995 to 2002 [RUN 200]; and 2) estimating age 3 stock size using survey tuning indices [RUN 201]. Results from these alternative formulations provided estimates of stock size, F and spawning stock biomass consistent with the base run [RUN 100]. RUN 201 stock size for age 3 was poorly estimated (CV = 63%). Based on these runs, the partial recruitment pattern indicated that age 7 was not fully recruited. An alternative formulation (RUN 300) was conducted using a partial recruitment vector where the fully recruited age was increased from 7 to 8. Assuming full recruitment at age 8, the F on ages 10 and 11+ in the terminal year was estimated as the average of F on ages 8 and 9. The F on ages 10 and 11+ in all years prior to the terminal year was derived from weighted estimates of Z for ages 8 through 9. This partial recruitment pattern is consistent with recent mesh regulation changes.
The final formulation (RUN 301-f) included a 3 to 11+ catch at age with large-mesh otter trawl discards estimated using both the survey filter method and FOP data; an updated partial recruitment vector reflecting current management regulations was derived from the 1999-2002 F pattern taken from a penultimate calibration run; annual maturity ogives estimated by the five year moving time block with the 2002 maturity vector assumed to be equal to 2001. Ages 1 and 2 were deleted from the catch at age, this allowed recruitment in 2003 to be estimated using the the geometric mean; there is no difference between 1-11+ vs 3-11+ on VPA results for fishing mortality and spawning stock biomass. Based on the final formulation, two sensitivity analyses were conducted to evaluate the selection of tuning indices. The VPA was tuned with only NEFSC spring survey indices and then tuned with only NEFSC autumn survey indices (Table 35). Estimates of F and SSB from analyses using a single tuning series bounded the F and SSB estimated using both spring and autumn tuning indices. Using only the spring tuning series (RUN 301-f-spr), F was slightly higher (F= 0.43) and SSB is slightly lower (15,798 mt) then the final run (RUN 301-f). Conversely, using only the autumn tuning indices (RUN 301-f-aut), F is slightly lower (F = 0.39) and SSB is slightly higher (21,569 mt; Table 35) then the final run (RUN 301-f) .
VPA Estimates of Fishing Mortality, Spawning Stock Biomass and Recruitment
A full listing of the final ADAPT VPA calibration output and diagnostics is presented in the Appendix and the results, including estimates of F, stock size and spawning stock biomass at age are given in Table 36. The mean residual for the VPA calibration was 0.791 and the CV on age 3-10 stock sizes ranged from 31% to 64% while the CVs on the estimates of survey catchabilities were between 13% and 26% (Appendix). The normalized survey indices and standardized residuals are presented in Figure 41 and Figure 42.
The VPA indicates that fishing mortality (ages 8-9, unweighted) increased from 0.26 in 1982 to 0.67 in 1985, declined to 0.22 in 1992, increased to 1.13 in 1996, then declined to 0.41 in 2002 (Table 37 and Figure 43). Spawning stock biomass declined from 16,897 mt in 1982 to about 3,800 mt in 1996. With recent increases in recruitment and declines in fishing mortality, SSB has increased to 18,296 mt in 2002 (Table 37 and Figure 44). Since 1982 recruitment of age 3 has ranged from approximately 3 million fish (1984 year class) to 67.6 million fish (1997 year class; Table 37 and Figure 44). Over the 1982-2002 period, average recruitment of age 3 fish (the 1979 - 2000 year classes) was 19.6 million (the geometric mean equaled 14.4 million fish). The 1993-2000 year classes appear to be above average, and the 1997 year class is the largest in the VPA time series (Table 37 and Figure 44).
The relationship between spawning stock biomass and recruitment (age 3) is presented in Figure 45. The negative stock-recruitment relationship observed in previous assessments continues with the addition of the 2000 year class.
Precision of F and SSB
The uncertainty associated with the estimates of stock size and fishing mortality from the final VPA was evaluated using a bootstrap procedure (Efron 1982). One thousand bootstrap iterations were performed to derive standard errors, coefficients of variation (CVs) and bias estimates for the stock size estimates at the start of 2003, the catchability estimates (q) of the abundance indices used in calibrating the VPA, and the 2002 fully recruited fishing mortality rate (age 8+). Frequency distributions of the 2002 mean fishing mortality and spawning stock biomass bootstrap estimates were generated and cumulative probability curves produced (Figure 46 and Figure 47).
Bootstrap results suggest that the CVs of the 2003 abundance estimates ranged between 32% to 84%, 24% for 2002 F8-9 and 14% for 2002 spawning stock biomass. There is an 80% probability that the 2002 F (0.41) lies between 0.31 and 0.56 (Figure 46), and the 2002 SSB (18,296 mt) lies between 15,603 mt and 22,969 mt (Figure 47).
A retrospective analysis was conducted on the final VPA (Run 301-f) from 2002 to 1992 by sequentially removing the terminal year of the data to evaluate internal consistency of the current ADAPT formulation with respect to terminal estimates of F, SSB, and recruits at age 3 for the seven years prior to the current assessment. Results indicate that average F was underestimated (Figure 48a) and spawning stock biomass was consistently overestimated (Figure 48b). The retrospective analysis indicated that the number of age 3 recruits were generally overestimated, and the 1995-1997 year classes were considerably overestimated (Figure 48c).
Statistical Catch-at-age model
A statistical catch-at-age analysis was conducted for the witch flounder stock. An age-structured forward-projection model (a.k.a., age-structured production model) was fit to fishery and survey data during 1937-2002. This model provided an alternative long-term perspective on resource dynamics in comparison to VPA-based analyses that were limited to the period 1982-2002. Age-structured population dynamics of witch flounder were described using forward-projection methods for statistical catch-at-age analyses (Fournier and Archibald 1982, Methot 1990, Ianelli and Fournier 1998, Quinn and Deriso 1999). Models were fit to data with the AD Model builder software for nonlinear optimization (Otter Research 2001).
Six alternative statistical catch-at-age models were developed and fit. Brodziak and Wigley (2003 ms) contains a complete description of the basic model and input data. Common features of the six models were:
- Natural mortality was M=0.15 for all age classes.
- Catch scenario 2 was used (same as used in VPA).
- Fishery selectivity was estimated for historic (1937-1993) and current (1994-present) time periods.
- NEFSC spring and fall survey biomass and numbers at age data were used.
- Emphasis values for likelihood components were: Recruitment 1=10, Fishery age composition 2=1, NEFSC Fall survey age composition 3=1, NEFSC Fall survey biomass index 4=100, NEFSC Spring survey age composition 5=1, NEFSC Spring survey biomass index 6=100, Catch biomass 7=100, Fishing mortality 8=1, Fishing mortality penalty 9=1
The primary differences among the six alternative models were:
- Dome-shaped selectivity possible for fishery, spring, and fall surveys; time frame is 1937-2002.
- Flat-topped selectivity for fishery, spring, and fall surveys; time frame is 1937-2002.
- Dome-shaped selectivity possible for fishery, spring, and fall surveys; time frame is 1963-2002.
- Flat-topped selectivity for fishery, spring, and fall surveys; time frame is 1963-2002.
- Flat-topped selectivity for fishery and spring survey; Dome-shaped selectivity possible for fall survey; time frame is 1963-2002.
- Flat-topped selectivity for fishery and spring survey; Dome-shaped selectivity possible for fall survey; time frame is 1937-2002.
Models 1, 2, and 6 were considered to be the primary models, while models 3, 4, and 5 provided sensitivity analyses to the choice of time frame. The Northern Demersal Working Group (WG) reviewed the model diagnostics. In general, the selectivity patterns of models that allowed dome-shaped fishery or survey selectivity appeared to be too sharply domed to be biologically plausible. In contrast, models with the assumption of flat-topped selectivity provided a poorer fit to the data, as measured by the root-mean squared errors for the NEFSC fall and spring survey biomass index and the catch biomass fits to the data. The WG chose to reduce the emphasis on the NEFSC fall and spring survey biomass index and the catch biomass likelihood components to 10 down from 100. This choice alleviated the problem of implausible selectivity patterns in the fishery and the survey. As a result, the WG concluded that Model 1 with reduced emphasis values was the best alternative of the statistical catch-at-age analyses (SCAA). Model results are reported to confirm the basic trends of VPA-based results and show the likely effect of extending the assessment time horizon back to 1937.
Model results showed that current fishery selectivity at age was estimated to be lower at ages 1-6 than historic selectivity (Figure 49). This was consistent with increases in fishery mesh size and changes in discarding practices (e.g., shrimp fishery) that occurred around 1994. The resulting catch biomass predictions generally matched observed catch biomass (Figure 50) with some moderate deviations in the early 1980s.
Model results showed that NEFSC fall survey selectivity was dome-shaped with a peak at age-5 (Figure 51). The NEFSC spring survey selectivity was flat-topped with full selection occurring at roughly age-7. The resulting predicted NEFSC fall and spring survey indices generally matched the trends in observed indices (Figure 52 and Figure 53). Both surveys indicate a longterm decline in biomass from the 1970s through the early 1990s. Biomass increases in the late-1990s differed moderately between the fall and spring surveys.
There was general agreement between VPA and SCAA results during 1982-2002. Spawning biomass estimates were very similar during 1989-1999 (Figure 54, SCAA estimate of 10.5 kt in 2002). The VPA indicates a smaller decrease in spawning biomass during 1982-1988 and a greater increase during 2000-2002. Fishing mortality estimates were also similar (Figure 55, SCAA estimate of 0.48 in 2002). Both VPA and SCAA estimates increased to roughly the mid-1990s and then declined. Recruitment estimates also exhibited similar patterns (Figure 56, SCAA estimate of 14.1 million age-1 fish in 2002), although the VPA indicated larger increases in recruitments during the late-1990s. Despite differences in model configuration and estimation approach, the SCAA generally confirmed point estimates and trends in the VPA results.
BIOLOGICAL REFERENCE POINTS
Yield-per-recruit (Y/R) and spawning stock biomass per recruit (SSB/R) analyses were performed using the Thompson and Bell (1934) method for witch flounder ages 3 to 20. Input vectors for partial recruitment, maturation at age and mean weights at age were all updated since the last assessment. Mean weights at age used in the Y/R analyses were computed as an arithmetic average of catch mean weights at age (Table 21) over the 1999-2002 period. Mean weights at age for use in the SSB/R analyses were derived by applying the length-weight relationship for witch flounder to predicted lengths at age from von Bertalanffy growth curve analyses of NEFSC survey data from 1980-2002. The maturation ogive from the entire time series (1980-2003) was also used (Table 32). Given the changes in regulated mesh size in 1999, the exploitation pattern used in the yield and SSB per recruit analyses and short-term projections was computed from the 1999-2002 VPA results. Geometric mean F at age was computed for the 1999-2002 period and divided by the geometric mean of the fully recruited annual Fs to derive the partial recruitment vector. The final exploitation pattern was smoothed, applying full exploitation on ages 8 and older, viz.
Age 3 Age 4 Age 5 Age 6 Age 7 Ages 8+ 0.0036 0.0229 0.0703 0.1931 0.5282 1.000
The biological reference points were updated by applying the approach used to estimate MSY proxies for witch flounder (NEFSC 2002). Fmsy is approximated as F40% (0.23), the SSBmsy proxy is 25, 248 mt, the product of 40%MSP (1.2882 kg spawning biomass) and average long-term recruitment (19.6 million). The MSY proxy is 4,375 mt, the product of yield per recruit at F40% ( 0.2232 kg) and average recruitment.
In 2002, spawning stock biomass was slightly greater than ½ SSBmsy (12,624 mt), the minimum stock size threshold, and fishing mortality in 2002 was nearly double Fmsy, the maximum fishing mortality threshold; therefore, witch flounder was not overfished but overfishing was occurring in 2002 (Figure 58).
To evaluate the effects of simultaneous changes in the three input vectors described above (i.e. partial recruitment, maturation and mean weights) on F40%, Y/R, SSB/R and the SSBmsy proxy, a decomposition analysis (P. Rago, NEFSC, pers. comm.) was conducted. This analysis is analogous to decomposing a sum of squares in an analysis of variance (decomposing the total resulting difference into its components).
For F40%, Y/R and SSB/R:
Total effect = effect of vector 1 + effect of vector 2 + effect of vector 3 + interaction terms.
Total effect = SSB/R effect + Recruit effect + interaction term.
The effect is the difference between the former YPR estimate and the current YPR estimate, for F40%, Y/R, SSB/R and for SSBmsy.
To accomplish this, the former YPR analysis (Run 0) was re-run using ages 3-20 to coincide with the ages used in the current YPR analysis (Run 1). Then, YPR analyses were conducted where each former vector was replaced with a current vector (Runs 2 through 7), until all vectors were replaced with current vectors (Run 8). The resultant F, Y/R and SSB/R at 40%MSP from each run (Runs 0 to 8) are reported in Table 39. The total effect of changing all three vectors at once equals Run 1 - Run 8.
Results of the decomposition analysis (Table 39) indicate that changes in F40% were effected most by new partial recruitment vector. Changes in Y/R resulted from the interaction of all three new input vectors while changes in the SSB/R resulted from the interaction between the mean weights and maturity vectors. Changes in SSBmsy were effected most by changes in new mean age 3 recruitment.
SHORT-TERM PROJECTIONS FOR 2004 AND 2005
Short-term stochastic projections were performed to estimate landings, discards and SSB during 2003-2005 under various F scenarios using bootstrapped VPA calibrated stock sizes in 2002. The partial recruitment, maturity ogive, and mean weights at age were the same as described in the yield and SSB per recruit section (Table 40). Recruitment (age 3) in 2003-2005 was derived by re-sampling the cumulative density function based on the empirical observations during 1982-2001 (1979-2000 year classes). Fishing mortality was apportioned among landings and discards based on the proportion observed landed at age during 1999-2002. The proportion of F and M which occurs before spawning equals 0.1667 (March 1); M was assumed to be 0.15. Spawning stock biomass in 2002 was estimated to be 18,296 mt. The F scenarios are: status quo F2003 = 0.41, Fmsy = 0.230, 75% of Fmsy = 0.17 and landings2003 = landings2002 (F= 0.199). Fishing at the status quo F (0.41) or at the target (Fmsy = 0.23) in 2003 - 2005 is expected to allow biomass to increase above SSBmsy and initiate rebuilding of the age structure (Table 40). Comparison of the current age structure and the age structure under MSY conditions are given in Figure 59.
Based on the ADAPT VPA, the witch flounder stock was not overfished, but overfishing was occurring in 2002. Fully recruited fishing mortality in 2002 was 0.41, nearly double Fmsy (0.23), and spawning stock biomass was estimated to be 18,296 mt in 2002, 72% of SSBmsy (25,248 mt) . Recent year classes appear to be above average. Although the spawning stock biomass has increased, the age structure still remains truncated. Fishing mortality should be reduced to Fmsy or below to allow the age structure to rebuild.
We would like to recognize all those who diligently collected the biological samples from the commercial fisheries and from research vessel surveys. We wish to express our appreciation to Betty Holmes for her technical assistance. We thank Jay Burnett for the age determinations. We thank all members of the Stock Assessment Workshop's Northern Demersal Working Group and the SARC panel for their review and helpful comments.
Azarovitz, T.R. 1981. A brief historical review of the Woods Hole Laboratory trawl survey time series. In: Doubleday, W.G. and D. Rivard (eds.), Bottom Trawl Surveys, p. 62-67. Can. Spec. Publ. Fish. Aquat. Sci. 58.
Brodziak, J., and S. Wigley. 2003 manuscript. Statistical catch-at-age analysis of the witch flounder stock, 1937-2002. Northern Demersal Working Group Meeting May 12-16 2003, Working Paper W3, NEFSC, Woods Hole, MA 02543.
Burnett, J.M. MS 1987. The population biology of the witch flounder, Glyptocephalus cynoglossus (L.), in the Gulf of Maine-Georges Bank region. M.Sc. Thesis, Department of Wildlife and Fisheries Biology, University of Massachusetts, Amherst, MA, 116 p.
Burnett, J. and S.H. Clark. 1983. Status of witch flounder in the Gulf of Maine - 1983. NMFS/NEFC, Woods Hole Laboratory Ref. Doc. No. 83-36, 31 p.
Burnett, J., M.R. Ross, and S.H. Clark. 1992. Several biological aspects of the witch flounder (Glyptocephalus cynoglossus (L.)) in the Gulf of Maine-Georges Bank region. J. Northw. Atl. Fish. Sci. 12: 15-25.
Cadrin, S.X. 2003. Stock assessment of yellowtail flounder in the southern New England - Mid-Atlantic area (SAW 36). NEFSC CRD 03-03.
Cadrin, S. X. 1999. A precautionary approach to fishery control rules based on Surplus production modeling. NOAA Tech. Memo. NMFS-F/SPO-40. 17-22 pp.
Clark, S.H. 1981. Use of trawl survey data in assessments. In: Doubleday, W.G. and D. Rivard (eds.), Bottom Trawl Surveys, p. 82-92, Can. Spec. Publ. Fish. Aquat. Sci. 58.
Conser, R.J. and J.E. Powers. 1990. Extensions of the ADAPT VPA tuning method designed to facilitate assessment work on tuna and swordfish stocks. Int. Comm. Conserv. Atlantic Tunas, Coll. Vol. Sci. Pap., 32:461-467.
Delong, A., K. Sosebee, and S. Cadrin. 1997. Evaluation of vessel logbook data for discard and CPUE estimation. NEFSC Laboratory Reference Document 97-xx.
Efron, B. 1982. The jackknife, the bootstrap and other resampling plans. Phila. Soc. for Ind. and Appl. Math. 38: 92 p.
Fournier, D. A., and C. P. Archibald. 1982. A general theory for analyzing catch at age data. Can. J. Fish. Aquat. Sci. 39:1195-1207.
Gavaris, S. 1988. An adaptive framework for the estimation of population size. CAFSAC Res. Doc. 88/29, 12 p.
Halliday, R.G. 1973. The flatfish fisheries of the Scotian Shelf. Int. Comm. Northw. Atl. Fish., Res. Doc. 73/103, Ser. No. 3064 (mimeo), 46 p.
Halliday, R.G. 1987. Size and age at sexual maturity of Atlantic argentine, Argentina silus: a critique. Environ. Biol. Fishes 19(2):165-261.
Howe, A.B., F.J. Germano, J.L. Buckley, D. Jimenez, and B.T. Estrella. 1981. Fishery resource assessment, coastal Massachusetts. Completion Report, Massachusetts Division of Marine Fisheries, Commercial Fisheries Review Div. Project 3-287-R-3.
Ianelli, J. N., and D. A. Fournier. 1998. Alternative age-structured analyses of NRC simulated stock assessment data. NOAA Tech. Memo. NMFS-F/SPO-30. pp. 81-96.
Lange, A.M.T. and F.E. Lux. 1978. Review of the other flounder stocks (winter flounder, American plaice, witch flounder, and windowpane flounder) off the northeast United States. NMFS, NEFC, Woods Hole Lab. Ref. Doc. No. 78-44, 53 pp.
Mayo, R.K., L. O'Brien, and N. Buxton. 1992. Discard estimates of American plaice, Hippoglossoides platessoides, in the Gulf of Maine northern shrimp fishery and the Gulf of Maine-Georges Bank large-mesh otter trawl fishery. Appendix to NEFSC Ref. Doc. 92-07, Res. Doc. SAW 14/3, 40 p.
Methot, R. D. 1990. Synthesis model: an adaptive framework for analysis of diverse stock assessment data. Int. North Pac. Fish. Comm. Bull. 50:259-277.
NEFSC [Northeast Fisheries Science Center]. 2002. A Report of the Groundfish Assessment Review Meeting (GARM), October 2002. NEFSC Ref. Doc. 02-16, 511p.
NEFSC [Northeast Fisheries Science Center]. 2002. Final Report of the Working Group on Re-Evaluation of biological Reference Points for New England Groundfish. NEFSC Ref. Doc. 02-04.
NEFSC [Northeast Fisheries Science Center]. 1994. Report of the 18th Northeast Regional Stock Assessment Workshop (18th SAW), Stock Assessment Review Committee (SARC) consensus summary of assessments. NEFSC Ref. Doc. 94-22, 281 p.
Northern Shrimp Technical Committee. MS 1984. Results of the 1983 northern shrimp survey in the western Gulf of Maine, August 1983. Unpublished Report, Woods Hole, MA, 16 p.
O'Brien, L. and R.W. Brown. 1996. Assessment of the Georges Bank Haddock stock for 1994. NEFSC Ref. Doc. 95-13, 108 p.
Otter Research. 2001. An introduction to AD Model Builder Version 6.0.2 for use in nonlinear modeling and statistics. Otter Research, Ltd. Sidney, B.C. V8L 3S3, Canada.
Parrack, M.L. 1986. A method of analyzing catches and abundance indices from a fishery. Int. Comm. Conserv. Atlantic Tunas, Coll. Vol. Sci. Pap., 24:209-211.
Quinn, T. P., II, and R. B. Deriso.1999. Quantitative fish dynamics. Oxford University Press, New York, 542 pp.
Rivard, D. 1980. APL programs for stock assessment. Can. Tech. Rep. Fish. Aquat. Sci. 953.
SAS. 1985. SAS User's Guide: Statistics, version 5. SAS Institute, Cary, NC. 956 p.
Thompson, W.F. and F.W. Bell. 1934. Biological statistics of the Pacific halibut fishery. 2. Effect of changes in intensity upon total yield and yield per unit of gear. Rep. Int. Pac. Halibut Comm. 8: 49 p.
Walsh, S.J., R.B. Millar, C.G. Cooper, and W.M. Hickey. 1992. Codend selection in American plaice: diamond versus square mesh. Fish. Res., 13:235-254.
Wigley, S.E. MS 1994. Estimation of foregone yield associated with the discarding of witch flounder (Glyptocephalus cynoglossus) in the Gulf of Maine northern shrimp (Pandalus borealis) fishery, 1982-1992. M.S. thesis. North Carolina State University, Raleigh, NC. 91 p.
Wigley, S.E. and R.K. Mayo. 1994. Assessment of the Gulf of Maine-Georges Bank witch flounder stock for 1994. Northeast Fisheries Science Center Reference Document 94-17, 86 p.
Wigley, S.E. , J.K.T. Brodziak and S.X. Cadrin. 1999. Assessment of the witch flounder stock in Subareas 5 and 6 for 1999. Northeast Fisheries Science Center Reference Document 99-16, 153 p.