Northeast Fisheries Science Center Reference Document 02-05
Biological
Characteristics, Population Dynamics,
and Current Status
of Redfish, Sebastes fasciatus Storer,
in the Gulf of Maine
- Georges Bank Region
by R.K. Mayo, J.K.T. Brodziak, M. Thompson, J. Burnett, and S.X.
Cadrin
National Marine Fisheries Serv., Woods Hole Lab., 166 Water St.,
Woods Hole, MA 02543
Print
publication date April 2002;
web version posted April 29, 2002
Citation: Mayo, R.K.; Brodziak, J.K.T.; Thompson, M.; Burnett, J.M.; Cadrin, S.X. 2002. Biological characteristics,
population dynamics, and current status of redfish, Sebastes fasciatus Storer, in the Gulf of Maine - Georges
Bank region. Northeast Fish. Sci. Cent. Ref. Doc. 02-05; 130 p.
Download complete PDF/print version
Abstract
The status of the Gulf of Maine/Georges Bank redfish
(Sebastes fasciatus) stock through 2000 is reviewed, and the
current status of the stock is compared on a relative basis to revised
estimates MSY-based reference points. The 2001 assessment is based
on several sources of information including: the age composition
of USA commercial landings, Northeast Fisheries Science Center (NEFSC)
spring and autumn research vessel survey data, and standardized USA
commercial fishing effort data. This assessment updates the analyses
presented in the 1993 assessment of the Gulf of Maine/Georges Bank
redfish stock as well as that prepared in 2000 by the Northern Demersal
Working Group.
Information on the size and age structure
of the redfish stock is presented including: age composition of the
commercial landings (1969-1985), length composition of inshore and
offshore components of the stock based on NEFSC spring (1968-2000)
and autumn (1963-2000) research vessel surveys, and age composition
of the stock based on NEFSC spring and autumn research vessel surveys
(1975-2000). Several aspects of the biology of the redfish stock are
also presented including patterns in diurnal catchability, length-weight
relationships, analyses of maturity at length, and inshore/offshore
biomass comparisons.
The assessment of current status is based
on several analyses including trends in catch/survey biomass exploitation
ratios; a yield and biomass per recruit analysis; an age-structured
dynamics model which incorporates information on the age composition
of the landings, size and age composition of the population, and
trends in relative abundance derived from commercial CPUE and research
vessel
survey biomass indices; and an age-aggregated biomass dynamics model.
Surplus production estimates were derived from the age-structured
production model, and information on current status of biomass and
fishing mortality
relative to MSY-based reference points is also provided by the biomass
dynamics model.
The fishery on this stock developed during the 1930s.
Landings rose rapidly from less than 100 mt in the early 1930s to over
20,000 mt in 1939, peaking at 56,000 mt in 1942, then declined throughout
the 1940s and 1950s. Redfish have been harvested primarily by domestic
vessels, although distant water fleets took considerable quantities
for a brief period during the early 1970s. The distant water fleet
effort, combined with increased domestic fishing effort, resulted in
a brief increase in total catch to about 20,000 mt during the early
1970s. Landings declined throughout the 1980s and have averaged less
than 500 mt per year during the 1990s.
Exploitation ratios (catch/survey biomass) suggest that
fishing mortality has been very low since the mid-1980s compared to
previous periods. Estimates of fishing mortality derived from the age-structured
dynamics model and the age-aggregated biomass model are similar, both
indicating that current fishing mortality is low relative to past decades
and with respect to Fmsy (<5%). Stock biomass has increased
since the mid-1990s, and is presently estimated to be about 33% of
Bmsy due, in large part, to recruitment of one or more strong year
classes from the early 1990s.
Introduction
Redfish, Sebastes fasciatus Storer, have supported
a substantial domestic fishery in the Gulf of Maine and the Georges
Bank (Great South Channel) regions off the northeast coast of the U.S.
(Northwest Atlantic Fisheries Organization [NAFO] Subarea 5) since
the late 1930s when the development of freezing techniques enabled
a widespread distribution of the frozen product throughout the country.
Landings by domestic vessels rose rapidly, peaking at 56,000 mt in
1942 in Subarea 5, then declined throughout the 1940s and 1950s (Table
1, Figure 1). As landings declined in
local waters, U.S. fishing effort began to expand to the Scotian Shelf
and the Gulf of St. Lawrence (NAFO Subarea 4), and finally to the Grand
Bank of Newfoundland (NAFO Subarea 3). This expansion continued throughout
the 1940s and early 1950s, culminating with a peak U.S catch of 130,000
mt in 1952 (Figure 1). By the mid-1950s,
redfish stocks throughout the Northwest Atlantic were heavily exploited
by U.S and Canadian fleets (Atkinson 1987), and total landings began
to decline in all Subareas.
During the 1960s and early to mid-1970s, catches by distant
water fleets were substantial, at times accounting for 25-30% of the
total Subarea 5 redfish catch (Table 1).
With the declaration of exclusive economic zones by the U.S. and Canada
in 1977, U.S. vessels were prohibited from fishing in all but a small
portion of Subarea 4 off Southwest Nova Scotia. Landings from the Gulf
of Maine subsequently increased temporarily during the late 1970s,
but have been declining throughout the 1980s, and have remained below
1,000 mt per year throughout the 1990s. Recent landings from this stock
are at their lowest level since the directed fishery commenced in 1934.
The status of this stock has been assessed since the
1970s with a variety of techniques including production models (Schaefer
1954, 1957; Pella and Tomlinson 1969; Fox 1975), yield per recruit
(Thompson and Bell 1934; Beverton and Holt 1957) and virtual population
analysis (VPA). A preliminary production model estimate suggested a
long-term potential yield of 20,000 mt from this stock (Mayo 1975)
but this was revised to 14,000 mt when non-equilibrium conditions were
taken into account (Doubleday 1976, Walter 1976), irrespective of the
growth model (exponential or logistic) employed (Mayo 1980). A yield
per recruit analysis performed with M=0.05 and partial recruitment
of 50% at age 6 and full recruitment at age 9, indicated Fmax at 0.13
and F0.1 at 0.06 (Mayo 1993, NEFSC 2001). Virtual population
analysis, which was first performed on this stock using catch at age
data from 1969-1980, indicated that age 9+ fishing mortality rates,
in the range of 0.18 to 0.28 throughout most of the 1970s, were accompanied
by a 62% decline in exploitable biomass (age 5+) between 1969 and 1980
(Mayo et al. 1983). A subsequent analysis which included additional
catch at age data through 1983 indicated that, although F had begun
to decline from a maximum value of 0.28 in 1979 to 0.17 in 1983, exploitable
biomass had been reduced by 75% from the 1969 level by 1984 (NEFC 1986).
The VPA was discontinued after 1986, but further declines in redfish
landings since then suggest that F is now likely to be rather low (at
or below M), rendering the convergence of VPAs somewhat unlikely.
Previous
stock assessments were reviewed at the 2nd and
15th Northeast Regional Stock Assessment Workshops (NEFC
1986, NEFSC 1993) and by the Northern/Southern Demersal Working Group
(NEFSC 2001). The potential for this stock to return to conditions
observed in the 1960s is limited, in part, by the combination of
slow growth and low fecundity of redfish. Even at relatively low
levels
of F, ranging from 0.03 to 0.05, restoration of the 1969 age structure
is not likely to occur except under extremely favorable recruitment
conditions over several decades (Mayo 1987).
COMMERCIAL
FISHERY
Commercial
Catch and Effort
Landings of redfish from Subarea 5 from 1934 through
2000 are given in Table 1 and Figure
1. Landings by domestic vessels rose rapidly from less than 100
mt in the early 1930s to over 20,000 mt in 1939, peaking at 56,000
mt in 1942, then declined throughout the 1940s and 1950s. Redfish have
been harvested primarily by domestic vessels, although distant water
fleets took considerable quantities for a brief period during the early
1970s (Table 1). The distant water fleet
effort, combined with increased domestic fishing effort, resulted in
a brief increase in total catch to about 20,000 mt during the early
1970s. Landings declined throughout the 1980s and have averaged less
than 500 mt per year during the 1990s. Landings in 2000 (319 mt) remain
close to an historic low. Redfish have been harvested almost exclusively
by otter trawlers fishing out of Maine and Massachusetts ports.
Commercial catch per unit effort (CPUE) indices for directed
redfish trips, standardized by vessel tonnage class as described by
Mayo et al. (1979), are listed in Table
1 and illustrated in Figure 2a. The resulting
calculated fishing effort values were derived by dividing total annual
landings by the directed CPUE index. Directed CPUE has declined steadily
from over 10 tons per day fished during the late 1960s to less than
2 tons per day fished since 1984 (Table 1, Figure
2a). This 70-80% decline is consistent with the 60-70% decline
in exploitable biomass estimated by previous VPAs (Mayo et al.
1983; NEFC 1986). Total fishing effort, after peaking during the late
1970s (coincident with the highest estimates of fishing mortality [NEFC
1986]), appeared to stabilize during the mid-1980s before declining
precipitously through 1989.
A depiction of the available effort data is presented
in Figure 2b. Historically, 80-90% of the
total redfish catch and 20-40% of the total number of trips on which
redfish were taken were accounted for in the directed CPUE calculation
(50% redfish trips). These percentages declined sharply between 1979
and 1982, and are now at levels which preclude any definitive interpretation
of the CPUE and effort trends.
Commercial
Length Composition
The available commercial length and age sample data are
summarized in Table 2. Commercial length sampling for redfish has generally
been sufficient to allow quarterly pooling until the 1990s. Sampling
during most years since 1994 has been insufficient to characterize
the length composition of the landings. The apparent improvement in
sampling intensity in recent years is an artifact of the rapid decline
in landings. Even with very low landings, sampling must be maintained
at relatively high levels in order to reflect the age structure of
the population. Age samples have been routinely collected since the
1960s but production ageing ceased after 1985 (Table
2).
Estimates of numbers landed at length were derived from
1969 through 2000 when sample data permitted. In most years prior to
1991, sampling was sufficient to allow pooling of length data on a
quarterly, and in a few cases, semi-annual basis. However, from 1991
to 2000, pooling of samples was required on a semi-annual, and in several
cases, an annual basis. Due to the differences in growth between males
and females, sampling for redfish is conducted separately by sex, and
estimates of numbers landed are also derived separately for males and
females. The overall length composition is then obtained by addition
of the estimates by sex.
Changes in the length composition of the landings between
1969 and 2000 are illustrated in Figure 3.
In 1978, the landings still reflected a fairly broad age structure
in the population of both males and females with the 1971 year class
accounting for the mode between 20 and 30 cm. With the decline in subsequent
recruitment, modes shifted toward larger sizes until fish from the
1978 year class appeared in 1983 and 1984. As landings continued to
decrease throughout the 1980s, modal lengths shifted further until
few fish between 20 and 25 cm could be seen recruiting to the fishery.
Shifts in modal lengths are reflected in annual changes
in mean length of the landings as illustrated in Figure
4. Increases in mean length occur during periods of poor recruitment
(such as 1965-1976) while sharp decreases generally signify the appearance
of a strong year class entering the fishery. The declines which began
in 1976 and 1983 indicate recruitment of the 1971 and 1978 year classes
entering the fishery at age 5. The subsequent overall increasing trend
indicates a gradual ageing of the population as recruitment has declined
over the past 30 years. Mean lengths of the landings have become extremely
variable in recent years as landings have become extremely low and
sampling has deteriorated.
Commercial
Age Composition
Estimates of numbers landed at age were also derived
from the biological sampling data for the period 1969 through 1985.
With the sharp decline in landings evident during the 1980s, ageing
of commercial samples was discontinued after 1985. For the period 1969-1985,
however, estimates of numbers landed at age were derived by applying
quarterly age/length keys, separately by sex, to the estimated numbers
landed at length by sex. The overall age composition was then obtained
by addition of the estimates by sex.
Catch at age and mean weight at age matrices based on
all available commercial length and age data from 1969 through 1985
are given in Table 3, and trends in the age
composition of the landings are illustrated in Figure
5. The sharp discontinuity in the age structure of the population
created by poor recruitment since the 1960s can be inferred from the
age composition of the landings. The most striking feature is the singular
presence of the 1971 year class advancing through the fishery since
1976, followed by the entrance of the 1978 year class during 1983-1985.
By the early 1980s, the fishery had become dependent on a few relatively
strong year classes and recruitment appeared to have collapsed.
RESEARCH
VESSEL SURVEYS
Bottom trawl surveys have been conducted by the Northeast
Fisheries Science Center in the Gulf of Maine - Georges Bank region
since autumn 1963 and spring 1968 (Azarovitz 1981). The NEFSC spring
and autumn bottom trawl survey data were analyzed to evaluate trends
in total abundance and biomass of redfish, diurnal effects on catchability,
differences in density between inshore and offshore regions of the
Gulf of Maine, trends in the size and age composition of the population,
total mortality, relationships between length and weight, and changes
in maturation at length.
Trends
in Total Abundance and Biomass
Abundance (stratified mean number per tow) and biomass
(stratified mean weight per tow) indices have been calculated from
NEFSC spring and autumn surveys based on strata encompassing the Gulf
of Maine and portions of the Great South Channel (strata 24, 26-30,
36-40; Tables 4 and 5; Figures
6a and 6b). Trends in total abundance and biomass are similar in
both spring and autumn surveys. Relative abundance of redfish has declined
sharply in both survey series, from peak levels over of 100 fish per
tow in the late 1960s and early 1970s to generally less than 10 fish
per tow during the mid-1980s through mid-1990s. The decline in biomass
has been of the same order (Figures 6a and 6b).
Both series suggest a slight increase in abundance and biomass between
the mid-1980s and 1990s followed by a sharp increase in autumn 1996
and spring 1997.
Day/Night
Comparisons
Redfish have been observed to exhibit consistent diurnal
patterns in their vertical distribution. Although Kelly and Barker
(1961 ) concluded that there is little evidence of diurnal movement
of planktonic larvae, they also noted a significant decrease in catches
of larval redfish by an Issacs-Kidd midwater trawl during daylight.
This was attributed to possible gear avoidance by larval redfish. Adult
redfish, however, are thought to exhibit very pronounced diurnal movement
patterns. Templeman (1959) noted that, off Newfoundland, redfish catches
from sets made more than one hour before sunrise or after sunset were
negligible compared to those from daytime sets. Catches were also related
to the season, with good catches extending over a longer part of the
day in the brightest months with the longest period of daylight. This
pattern was well known in the commercial redfish fishery as vessels
would often lay to during the night.
In an earlier paper on redfish biology, Steele (1957)
noted the same overall diurnal pattern in redfish catches. In this
study, Steele provided evidence of a 2-3 fold difference in average
catch rates over a 24-hour period. This pattern was correlated , in
part, with the vertical movement of the euphausiid, Meganyctiphanes
norvegica, a major prey item of redfish in the North Atlantic.
Steele (1957) also observed seasonal departures from the general pattern,
and speculated that these differences may be related to the sexual
maturation cycle of males and females. The diurnal response of males
and females differed among seasons.
The presence of a diurnal pattern in redfish activity
in the Gulf of Maine was examined over the period 1992-2000. NEFSC
spring and autumn survey catch data were partitioned into six 4-hour
time blocks as follows: 0001-0400 hr (night2), 0401-0800 hr (dawn),
0801-1200 hr (day1), 1201-1600 hr (day2), 1601-2000 hr (dusk), and
2001-2400 hr (night1). Catch data for valid survey tows within the
total Gulf of Maine strata set as above were selected from the spring,
summer, and autumn surveys. Summer surveys were conducted only in 1992,
1993 and 1994 and the number of tows in the Gulf of Maine which contained
redfish (n=85) was relatively small.
The catch data were analyzed for seasonal and diurnal
effects by ANOVA using PROC GLM (SAS, 1990). Initial analyses indicated
that seasonal effects were not significant; however, based on the observations
of Steele (1957) regarding different seasonal responses by males and
females, further analyses were conducted separately for spring and
autumn data, with summer excluded. In the analyses of diurnal effects,
the last time block (2001-2400 hr) was elected to represent unity and
each of the 5 remaining blocks were related to the last block. The
factors for each time block were re-transformed from log scale to linear
scale.
In the overall analysis, catch rates
from periods 2 (0401-0800 hr), 3 (0801-1200 hr) and 4 (1201-1600 hr)
were significantly different (p < 0.05) from period 6 (2001-2400
hr). These represent dawn and the 2 daytime periods. Catch rates from
the remaining periods (1 and 5), representing dusk (1601-2000 hr) and
night (2001-2400 hr) were not significantly different from period 6.
Analyses of the spring and autumn data revealed possible seasonal differences
(Figure 7). During spring, catch rates from
time periods 2, 3, and 4 were significantly different (p < 0.05)
from those of period 6, but during autumn, none of the time periods
exhibited statistically significant differences in catch rates, although
the general pattern was similar to spring. These differences between
spring and autumn were not due to any pronounced bias in survey station
coverage by time period as the number of stations in both spring and
autumn were almost evenly distributed (Figures
8a and 8b).
In fact, the seasonal differences obtained for the Gulf
of Maine are consistent with the observations of Steele (1957) and
Templeman (1959). When the timing of the NEFSC survey in the Gulf of
Maine is taken into account (spring survey in late April, autumn survey
in late October), it can be seen that this portion of the spring survey
occurs during a period of considerably longer daylight relative to
autumn. There is a 2-month absolute difference in the timing of the
spring and autumn surveys with respect to the corresponding vernal
and autumnal equinoxes. These results are consistent with Templeman's
(1959) observation that good catches occur over a longer part of the
day in the brightest months. The results also seem to corroborate Steele's
(1957) observation that seasonal differences may be related to the
reproductive cycle where females may be more pelagic during the larval
extrusion stage in spring whereas both sexes may occupy bottom during
a greater period of time during the copulation stage in autumn.
Despite the large diurnal differences in catch rates
derived from these analyses, abundance and biomass indices are not
likely to exhibit any substantial bias given the even distribution
of occupied stations over time. It is likely, however, that annual
departures from an even distribution among the six time periods may
impart a degree of inter-annual variability which may partially explain
some of the large year effects exhibited in these data. However, if
the redfish survey indices were to form the basis of an estimate of
absolute biomass, the diurnal differences noted herein must be taken
into account before any estimation is made.
Inshore/Offshore
Comparisons
Indices were also computed for inshore (strata 26, 27,
39, and 40; area: 3,042 square miles) and offshore (strata 24, 28-30,
36-38; area: 17,419 square miles) subsets of the data (Figures
9a and 9b). When two or more strata sets of unequal area are compared
in this manner, the stratified mean catch per tow indices must be considered
to represent the density of fish (index of number or biomass per unit
area) rather than actual abundance or biomass (index of population
size). The inshore Gulf of Maine area from Massachusetts Bay to the
eastern coast of Maine has generally contained higher densities of
redfish compared to the offshore regions, particularly in terms of
numbers (Figure 9a). These fish are generally
smaller than those in the offshore regions, and the index from the
inshore area may be used as a measure of recruitment (Mayo 1980). Trends
in these indices have been consistent with trends in the overall combined
indices (Figures 6a and 6b).
Trends in mean length and weight of redfish from inshore
and offshore strata sets during autumn are illustrated in Figures
10a and 10b. As with commercial mean lengths, sharp declines indicate
the appearance of a relatively strong year class. This is most evident
in the autumn series of inshore data which has provided the most consistent
indicator of recruitment patterns over time. The sharp declines which
occur immediately after 1971, 1978, and 1984 reflect the initial appearance
and subsequent increased influence of these year classes in the inshore
bottom trawl survey indices. The 1991 year class is reflected in the
offshore mean length and weight patterns.
To compare trends in actual abundance and biomass between
regions, the indices must be weighted by the area of each strata set.
This approach provides indices of population size within each strata
set which can be directly compared on the same basis. When viewed in
this manner, it is clear that the greatest fraction of the redfish
population has historically been found in the offshore region of the
Gulf of Maine (Figures 11a and 11b).
Size
Composition
Length composition data from spring, autumn and shrimp
surveys (Figures 12 and 12a)
simultaneously illustrate the changes in relative abundance and size
structure of the population which resulted from the decline in recruitment
over time. The redfish population was composed of a relatively broad
range of sizes in the 1960s resulting from consistent recruitment of
year classes from the 1950s and 1960s. By the mid-1970s, however, abundance
of large fish had declined substantially and only the 1971 year class
remained a dominant feature in the demographics of the population.
The consistency of the survey indices had begun to erode by the beginning
of the 1980s and, throughout this decade, only sporadic indications
of the 1978 and subsequent year classes were evident.
During the 1990s, however, substantial numbers of redfish,
generally between 20 and 25 cm, began to appear, first in spring 1992,
then in autumn 1995 and 1996. These data likely reflect the strength
of one or more year classes from the mid-1980s and early 1990s. In
autumn 1999, a mode at 5 cm could indicate a potentially strong 1999
year class. By 1997, large numbers of redfish up to 30 cm and larger
were appearing consistently. However, the size structure of the population
remains truncated compared to the 1960s and early 1970s. The same pattern
appears in the shrimp survey.
Age
Composition
Age composition estimates are available from NEFSC autumn
surveys from 1975 through 2000 and from NEFSC spring surveys from 1975
through 1990 with some exceptions. The survey otolith collection is
routinely aged to the maximum possible age. For this analysis and the
subsequent analysis of mortality rates, all ages greater than 50 years
were binned at 50+. As the autumn survey has provided the most consistent
set of abundance and biomass indices, priority was given to ageing
of the autumn survey otolith collection. Annual trends are illustrated
in Figure 13. The age composition data clearly
illustrate recruitment patterns and changes in age structure of the
population that are suggested by the length composition data. In 1975
the population still appeared to exhibit a relatively broad age structure.
The 1971 year class is prominently featured in 1975 followed by the
1978 year class in the early 1980s; these two year classes continued
to dominate the demographics of the population through the 1980s.
More recently, the 1985 and 1991 year classes appear
most prominent. As indicated by the length composition estimates, the
age structure of the population during the late 1990s remains truncated
compared to the 1975 and earlier period.
Total
Mortality Estimates
Estimates of instantaneous total mortality were computed
from the age composition data derived from NEFSC autumn surveys from
1975-1996. Annual Z estimates, based on the annual survival rate from
ages 6 and older to ages 7 and older, were highly variable, ranging
between -1.6 to + 1.6. These estimates reflect the high degree of variability
in year class strength evident in the survey abundance indices at age
presented in Figure 13. Therefore, an alternate
approach was attempted.
The 1975-1996 autumn survey age composition data contain
information on cohorts spanning 1925 to as recently as 1995. To minimize
the variability induced by variation in year class strength, separate
catch curves were constructed for each cohort. Since the time span
represented in the age composition data covers the years 1975-1996,
cohorts from years prior to the mid-1970s become truncated at the younger
ages whereas cohorts from years after 1975 become progressively truncated
at the older ages. When combined in a single plot, the mortality on
by various ages spanning the period 1925-1995 is visually represented
(Figure 14). This provides a general indication
of the average mortality sustained by the population over this 70 year
period. It is evident that, in most cases, redfish are incompletely
recruited until ages 5 or 6. However, mortality rates appear to be
relatively consistent for most cohorts after age 6. No attempt was
made at this stage to derive mortality estimates for individual cohorts.
Length-Weight
Analyses
The relationship between length (cm) and weight (kg)
of redfish was examined by season and sex using linear regression (PROC
REG, SAS 1990) of the form:
Ln Weight = a + b* Ln Length.
The analysis is based on 8,567 individual length and
weight measurements collected during NEFSC spring and autumn surveys
since 1992. There are no significant differences (p=0.800) in the length-weight
relationship between spring and autumn. However, differences between
males and females are highly significant (P< 0.01) (Figure
15), with females considerably heavier at a given length.
Maturation
Analyses
Redfish are relatively long-lived, slow growing fish
with an extremely low natural mortality rate compared to most highly
exploited species. Growth studies have indicated maximum ages ranging
from 50-60 years at lengths of 45-50 cm (Mayo et al. 1990).
Perlmutter and Clark (1949) provided early evidence that immature redfish
in the Gulf of Maine exhibited extremely slow growth and that maturation
was delayed until about age 9. Kelly and Wolf (1959) further demonstrated
the extremely slow growth of adult redfish up to age 20. More recently,
Mayo et al. (1981) provided further validation of the slow
growth rates for redfish up to age 7 based on length mode progression
and otolith edge formation. Consequently, an instantaneous natural
mortality rate of 0.05 has been employed in age-structured models,
consistent with the longevity of this species. Moreover, growth and
maturation appear to be linked. The most recent estimates of redfish
maturation suggest a median age of about 5.5 years (Mayo et al.
1990; O'Brien et al. 1993) compared to the 9-10 years indicated
by Perlmutter and Clark (1949).
In this analysis, the relationship between maturation
(Pm) and length is examined within 3 time periods using logistic regression
(PROC LOGISTIC, SAS 1990) of the form:
Pm = e (a + b*Len) / (1 + e (a + b*Len)).
The analysis is based on 3,728 individual maturity stage
observations from 1975 through 2000 within the following periods: 1975-1981,
1982-1991, and 1992-2000. There are 6 maturation stages for male redfish
and 7 stages (including eyed larvae) for females. The development and
present basis for the NEFSC maturity stages are described by Burnett et
al. (1989).
In general, redfish maturation at length remained relatively
constant over the 25 year period analyzed. A slight trend towards decreasing
size at maturity is evident in both the spring and autumn results (Figure
16). Estimates of median length at maturation (L50) for females
varied between 20.3 cm and 22.6 cm. The slightly higher values occurred
in the earliest period. Estimates of L50 for males ranged from 20.2
to 21.3 cm and the higher values also correspond to the 1975-1981 period
(Figure 17).
ASSESSMENT
OF CURRENT STATUS
Yield
and SSB per Recruit
Yield and spawning stock biomass (SSB) per recruit were
calculated according to the methods described by Thompson and Bell
(1934) and Gabriel et al. (1989). Natural mortality was assumed
to be 0.05. Mean weights at age for the yield per recruit calculations
were taken as the 1969-1984 mean of the commercial mean weights at
age (Table 3). Partial recruitment was based
on the fishery selectivity pattern derived from the age-structured
model presented below. This pattern was similar to that employed in
the previously published VPA (Mayo 1993) which was taken from the most
recently published VPA (NEFC 1986) which reflects the recruitment of
the 1971 year class. Growth and maturation data for SSB/R analysis
were taken from the female data presented by Mayo et al. (1990).
Estimates of F0.1 (0.06) and Fmax (0.13)
(Table 6, Figure 18)
are identical to those derived by Mayo (1993); these estimates were
similar to those reported by Mayo (1980) using the Beverton-Holt approach
with the same value of M (0.05) for 89mm mesh (males) and 102 mm mesh
(females). F at 30% of Maximum Spawning Potential was estimated as
0.07, slightly above the estimate of F0.1.
Index
of Exploitation
An index of exploitation (Table
7; Figure 19) was derived for the period
1963-2000, expressed as the ratio of the autumn NEFSC biomass index
(Table 5) to total fishery removals (Table
1). The index fluctuated considerably during the 1960s and 1970s,
generally increased until 1982, then declined sharply during the
1980s. Since 1990, the index of exploitation has remained at an extremely
low level as landings remained low despite the recent increase in
the survey biomass index. However, in contrast to the 1960s and 1970s,
where a substantial portion of the stock persisted in the 30-40 cm
range (Figure 12), during the 1990s, almost
all of the redfish were less than 25 cm, and almost none were larger
than 30 cm. This suggests that, given the present demographics of
the stock, only a small fraction of the biomass would be considered
exploitable. Thus, the exploitation ratio based on the total biomass
index, tends to under-estimate current exploitation relative to the
earlier period in the series.
Age-structured
Dynamics Model
In this section, an age-structured assessment model is
developed for redfish. Age-structured population dynamics of redfish
are modeled in a standard manner using forward-projection methods for
statistical catch-at-age analyses (Fournier and Archibald 1982, Methot
1990, Ianelli and Fournier 1998, Restrepo and Legault 1998). The population
dynamics model, statistical estimation approach, model diagnostics,
and model results are described in sequence below.
Population dynamics model
The age-structured model is based on forward projection
of population numbers at age. This modeling approach is based on the
principle that population numbers through time are determined by recruitment
and total mortality at age through time. The population numbers at
age matrix N=(Ny,a)YxA has dimensions Y by A,
where Y is the number of years in the assessment time horizon and A
is the number of age classes modeled. The oldest age (A) comprises
a plus-group consisting of all fish age-A and older. The time horizon
for redfish is 1934-2000 (Y=67). The number of age classes is 26, representing
ages 1 through 26+.
Recruitment (numbers of age-1 fish) in year y (Ry)
is modeled as a lognormal deviation from average recruitment (µR),
where Vy are iid normal random variables with zero mean
and constant variance.
For all years y from 1935-2000, Ry = Ny1 is
estimated from the recruitment deviation and average recruitment.
Initial population abundance at age in 1934 is based
on recruitment deviations from average recruitment for 1909-1934 and
natural mortality. For all ages a < A, the numbers at age in the
first year (ystart=1) are estimated as lognormal deviations from average
recruitment as reduced by natural mortality
For the plus group, the initial numbers at age is the
sum of numbers at ages 26 and older based on an equilibrium recruitment
deviation for ages 26 and older and natural mortality.
The total instantaneous mortality at age matrix Z=(Zy,a)YxA and
the instantaneous fishing mortality at age matrix F=(Fy,a)YxA both
have dimensions Y by A. Instantaneous natural mortality at age is assumed
to be constant (M) and for all years, y and ages, a
Population numbers at age through time are computed from
the initial population numbers at age, recruitment through time, and
total mortality at age through time. For all ages, a that are younger
than the plus group (a < A), the number at age are sequentially
determined using
For the plus group, numbers at age are the sum of survivors
at age A-1 and plus group survivors
Fishing mortality at age a in year y is modeled as a
separable process, where Sa is selectivity at age a and
Fy is fully-recruited fishing mortality in year y
Fully-recruited fishing mortality in each year is modeled
as a lognormal deviation from average fishing mortality (µF),where
Uy are iid normal random variables with zero mean and constant
variance
Fishery selectivity at age is modeled as being time-invariant
throughout the assessment time horizon. This approach was chosen for
parsimony. In particular, redfish catch-at-age data to estimate fishery
selectivity are limited to 1969-1985, a period when the fishery practices
are believed to have been relatively stable. Fishery selectivity at
age is estimated for ages 1 through 9. For ages older than 9 years,
fishery selectivity is assumed to be equal to the age-9 selectivity
value. This approach was chosen to reflect the asymptotic selectivity
pattern from previous VPA-based assessments of redfish, wherein age
9 was the age of full selectivity. Two constraints are applied to the
estimated selectivity at age coefficients. First, the selectivities
are constrained to average 1 for estimated ages. This forces the scale
of each coefficient to be near unity. Second, a constraint is applied
to ensure that estimated selectivities change smoothly between adjacent
ages. Details of the implementation of both constraints are described
in the section on statistical estimation approach. Last, for each year
the selectivity at age values are scaled so that the maximum selectivity
at age value is unity. This ensures that estimated fully-recruited
fishing mortality rates are directly comparable to biological reference
points such as F0.1.
The fishery catch numbers at age matrix C=(Cy,a)YxA and
the fishery catch biomass at age (yield) matrix Y=(Yy,a)YxA both
have dimensions Y by A. Fishery catch at age in each year is computed
from Baranov's catch equation using population numbers, fishing mortality,
and total mortality at age
Catch biomass at age in each year is the product of catch
numbers at age and mean weight at age, where Wa is the mean
weight at age computed as the average of mean redfish weights at age
from fishery sampling during 1969-1985
Total fishery catch biomass in year y (Yy)
is the sum of yields by age class
The total fishery catch biomass time series is compared
to observed values using a lognormal probability model.
The proportion of fishery catch at age a in year y (Py,a)
is computed from estimated catch numbers
The time series of fishery proportions at age are fitted
to observed fishery values using a multinomial probability model.
Fishery catch-per-unit effort in year y (CPUEy)
is modeled as a catchability coefficient (QCPUE) times exploitable
biomass raised to a power (ßCPUE), where exploitable
biomass is computed at the midpoint of the year
This model for CPUE coincides with the proportionality
model when ßCPUE = 1. The estimated CPUE time series
is fitted to observed values using a lognormal probability model.
The survey biomass index in year y (Iy) for
either the NEFSC autumn or spring survey is modeled as a catchability
coefficient (QSURVEY) times the population biomass that
is vulnerable to the survey, where SSURVEY,a is survey selectivity
at age a and pSURVEY is the fraction of annual total mortality
that occurs prior to the survey
The survey biomass index time series are fitted to observed
values using a lognormal probability model.
Survey selectivity at age is modeled using Thompson's
exponential-logistic model (Thompson 1994), where , ß,
and are parameters and survey
selectivity for redfish is assumed to be time invariant.
This model has the useful property that the maximum selectivity
value is unity. For values of >0
survey selectivity is dome-shaped, while survey selectivity is flat-topped
when =0.
Survey catch proportion at age a in year y (PSURVEY,
y, a) is computed from survey selectivity, the fraction of
mortality occurring prior to the survey, and population numbers at
age
The time series of survey proportions at age are fitted
to observed fishery values using a multinomial probability model.
Statistical
estimation approach
The population dynamics model is fit to observed data
using an iterative maximum likelihood estimation approach. The statistical
model consists of nine likelihood components (Lj) and two
penalty terms (Pk). The model objective function ()
is the weighted sum of the likelihood components and penalties where
each summand is multiplied by an emphasis coefficient (j)
that reflects the relative importance of the data.
Each likelihood component is written as a negative log-likelihood
so that the maximum likelihood estimates of model parameters are obtained
by minimizing the objective function. The Automatic Differentiation
Model Builder software is used to estimate a total of 179 model parameters.
The likelihood components and penalty terms are described below.
1. Recruitment
Recruitment strength is modeled by lognormal deviations
from average recruitment for the period 1909-2000. A total of 92 recruitment
deviation parameters (Vy) and one average recruitment parameter
(µR) are estimated based on the objective function
minimization. The recruitment likelihood component (L1)
is
where
2. Fishery CPUE
Fishery CPUE is modeled by lognormal deviations of predicted
values from observed values, denoted with a superscript "OBS" for all
variables, during 1942-1989, where Wy are iid normal random
variables with zero mean and constant variance
A total of 2 parameters (QCPUE and CPUE)
are estimated based on the objective function minimization. The fishery
CPUE likelihood component (L2) is
3. Fishery age composition
Fishery age composition is modeled as a multinomial distribution
for sampling catch numbers at age. The constant NE ,FISHERY, y denotes
the effective sample size for the multinomial distribution for year
y and is assumed to be constant across time for the years 1969-1985
when redfish catch-at-age data are available. The observed number of
fish at age in the fishery samples is computed as the effective sample
size times the observed proportion at age. The effective sample size
was assumed to be 200 fish in each year during 1969-1985. The negative
log-likelihood of the multinomial sampling model for the fishery ages
(L3) is
The second term in summation over a is a constant that
scales L3 to be zero if observed and predicted proportions
were identical. Nine fishery selectivity coefficients (S1 through
S9) are estimated based on the objective function minimization.
4. Autumn survey age composition
Autumn survey age composition is also modeled as a multinomial
distribution for sampling survey catch numbers at age. The constant
NE ,AUTUMN, y denotes the effective sample size for the
multinomial distribution for year y and is assumed to be constant across
time for the years 1975-2000 when redfish autumn survey catch-at-age
data are available. The observed number of fish at age in the survey
samples is computed as the effective sample size times the observed
proportion at age. The effective sample size was assumed to be 100
fish in each year during each year. The negative log-likelihood of
the multinomial sampling model for the autumn survey ages (L4)
is
As with the fishery age composition, the second term
in the summation over a is a constant that scales L4 to
be zero if observed and predicted proportions were identical. Three
autumn survey selectivity coefficients (AUTUMN, ßAUTUMN, AUTUMN)
are estimated based on the objective function minimization.
5.Autumn survey biomass index
The autumn survey biomass index is modeled by lognormal
deviations of predicted values from observed values during 1963-2000,
where DAUTUMN, y are i.i.d. normal random variables with
zero mean and constant variance
The autumn survey biomass likelihood component (L5)
is
One autumn survey catchability (QAUTUMN) coefficient
is estimated based on the objective function minimization.
6. Spring survey age composition
Spring survey age composition is also modeled as a multinomial
distribution for sampling survey catch numbers at age. The constant
NE ,SPRING y denotes the effective sample size for the multinomial
distribution for year y and is assumed to be constant across time for
the years 1975-1980 and 1984-1990 when redfish spring survey catch-at-age
data are available. The observed number of fish at age in the survey
samples is computed as the effective sample size times the observed
proportion at age. The effective sample size was assumed to be 100
fish in each year during each year.The negative log-likelihood of the
multinomial sampling model for the autumn survey ages (L6)
is
Three spring survey selectivity coefficients (SPRING, ßSPRING, SPRING)
are estimated based on the objective function minimization.
7. Spring survey biomass index
The spring survey biomass index is also modeled by lognormal
deviations of predicted values from observed values during 1968-2000,
where DSPRING, y are i.i.d. normal random variables with
zero mean and constant variance
The spring survey biomass likelihood component (L7)
is
One spring survey catchability (QSPRING) coefficient
is estimated based on the objective function minimization.
8. Catch biomass
Catch biomass is modeled by lognormal deviations of predicted
values from observed values during 1934-1999, where T y are
iid normal random variables with zero mean and constant variance
The catch biomass likelihood component (L8)
is
9. Fishing mortality
Fishing mortality on fully-selected ages is modeled by
lognormal deviations from average fishing mortality for the period
1934-1999. A total of 66 recruitment deviation parameters (Uy)
and one average fishing mortality parameter (µF) are
estimated based on the objective function minimization. The fishing
mortality likelihood component (L9) is
where
10. Fishery selectivity
Two constraints on fishery selectivity are included in
a penalty function. The fishery selectivity penalty function (P1)
is
The first term constrains the fishery selectivity coefficients
to scale to an average of 1. The second term constrains the fishery
selectivity coefficient of age a+1 to be near to the linear prediction
of this value interpolated from age a and age a+2 selectivities over
the range of estimated selectivity coefficients.
11. Fishing mortality penalty
One constraint on fishing mortality is imposed to ensure
that during the early phases of the iterative estimation process that
the observed catch is not generated by an extremely small F on an extremely
large population size. The fishing mortality penalty function (P2)
is
The constraint is weighted with a value of 10 for the
initial estimation phases and is weighted with a value of 0.001 for
the latter and final estimation phases. The value of 0.1 was used because
this is near the maximum computed in previous VPA-based analyses of
the redfish stock. Sensitivity analyses that changed 0.1 to either
0.05 or 0.2 showed virtually no difference in parameter estimates.
Initial values are input for all parameters before the
estimation phases are conducted. A total of seven estimation phases
were used for the iterative minimization of the objective function.
The first phase estimates average recruitment. The second phase estimates
average fishing mortality and fishing mortality deviations. The third
phase estimates recruitment deviations. The fourth phase estimates
fishery and survey selectivity coefficients. The fifth and sixth phases
are placeholders left open for additional parameters, if needed, while
the seventh phase estimates the fishery CPUE catchability and beta
parameters.
The eleven emphasis values used for the baseline analysis
were: 10 (recruitment), 10 (fishery CPUE), 1 (fishery age composition),
1 (autumn survey age composition), 1000 (autumn survey biomass index),
1 (spring survey age composition), 1000 (spring survey biomass index),
1000 (catch biomass), 1 (fishing mortality), 100 (fishery selectivity
penalty), 1 (fishing mortality penalty).
Model
diagnostics
Model diagnostics were the discrepancies between observed
data and predicted values for the catch biomass series (Figure
20), the autumn survey biomass series (Figure
21), the spring survey biomass series (Figure
22), the fishery CPUE series (Figure 23),
fishery age composition series (Figure 24),
autumn survey age composition series (Figure
25), and spring survey age composition series (Figure
26).
Model
results
Key model results of spawning biomass, fishing mortality,
recruitment, and population biomass for the period 1963-2000 are listed
in Table 8 and a full listing of the code
to fit the model, the model output, and the standard deviation of parameters
and other assigned output variables are provided in Appendix
1.
Fishery and survey selectivity estimates at age are shown
in Figure 27. Fishery selectivity was flat-topped
with full selectivity at age 9. While it was assumed that selectivity
for ages 10 and older was equal to age-9 selectivity, this did not
mean that the age-9 fish had to be fully-selected. The autumn survey
selectivity pattern was moderately dome-shaped with full selection
at age 5. In contrast, spring survey selectivity was dome-shaped with
full selection at age 9. The Northern Demersal Working Group (NDWG)
noted that the spring survey selectivity pattern was robust but the
autumn survey selectivity pattern was sensitive to the inclusion of
recent autumn survey age composition data. In particular, autumn survey
selectivity was flat-topped in an initial model run that included the
1996-1998 and 1981-1983 autumn survey age composition data but did
not include the 1999-2000 data.
Recruitment estimates are shown in Figure
28 (see also Table 8). Strong year
classes have been sporadic in recent years with the 1971 and 1992
year classes being very large. Recruitment was higher, on average,
in the 1950s-1960s than in recent years. Overall, the model's ability
to resolve which year class(es) in the early 1990s were strong was
dependent on the recent autumn survey age composition data, in part
due to the lack of commercial fishery age composition data since
1985. The NDWG noted that the earliest recruitment values in the
time series (1934-1962) were not reliable as absolute measures of
recruitment strength by year because these values were sensitive
to assumptions about how to estimate the initial population size
at age in 1934. This sensitivity was a natural consequence of having
little information on annual recruitment variation at the beginning
of the time series. In particular, the extremely large recruitment
estimate in 1942 was sensitive to model assumptions about initial
population size.
Population biomass estimates are shown
in Figure 29 (see also Table
8). Population biomass declined from the 1950s to the late-1980s
and has increased since then. The NDWG noted that the early portion
of the population biomass time series (1934-1951) was less reliable
because there was no relative abundance information during that time
period, i.e., the model was only tuned to catch biomass in the 1930s-1940s.
The NDWG also noted that population biomass estimates in the 1970s-1980s
were very similar to those obtained with an untuned VPA conducted for
SAW 2.
Spawning biomass estimates (at start of the spawning
season) are shown in Figure 30 (see also Table
8). Spawning biomass declined from the 1950s to the late-1980s
and has increased throughout the 1990s. The NDWG noted that the current
population biomass estimate was sensitive to the size of the strong
year class(es) of the early-1990s which could start to appear in fishery
catches.
Fishing mortality estimates are shown in Figure
31 (see also Table 8). Annual estimates
of fishing mortality early in the time series (1934-62) were not
considered to be reliable because they were sensitive to assumptions
about initial population size. Instead, the early estimates of F
provide information on the average fishing mortality that was experienced
by the redfish population as the fishery began. Fishing mortality
increased from 0.05-0.1 in the early 1960s to over 0.20 in the late-1970s
to early-1980s. Since then, fishing mortality has declined and is
currently below 0.01 in 2000.
Stock-recruitment data are shown in Figure
32. Recruitment was below-average throughout 1963-2000, with
the exception of a few strong year classes; for example, the 1971
and 1992 year classes.
Surplus production implied by the age-structured estimates
of exploitable biomass and observed catches is shown in Figure
33. Surplus production was above 10 kt per year during the 1960s
and then declined to very low levels in the 1980s because recruitment
was very low. The recent increase in surplus production is due to strong
recruitment in the early 1990s. The trajectory of surplus production
shows the decline from 1963 to 1990 followed by a sharp increase in
recent years.
Model sensitivity to the assumption that natural mortality
is 0.05 is shown in Figure 34. The likelihood
profile for natural mortality shows that there are values of M from
0.025 to 0.045 that produce a higher value of the total model likelihood
than M=0.05. The biomass time series shows the consequence of higher
or lower values of M on estimated population biomasses.
Model sensitivity to the assumption that each of the
relative abundance indices (autumn and spring survey biomass indices
and CPUE) provides useful information on population trend is shown
in Figure 35. The delete one index sensitivity
analysis shows that the model is robust to the exclusion of one index.
The delete two indices sensitivity analysis shows that the model is
robust to the use of only the autumn or the spring survey series. However,
use of only the CPUE series would produce a substantially different
population biomass trajectory.
Biomass
Dynamics Model
MSY-based
reference points
The current overfishing definition and targets for redfish
are based on an MSY estimate from surplus production analysis (MSY=14,000
mt, Mayo 1980), supplemented with an FMSY proxy from a dynamic
pool model (F20%=0.12), to derive a proxy BMSY (14,000/0.12=60,500
mt, Applegate et al. 1998). As calculated, the current BMSY proxy
is in units of exploitable biomass.
The age-structured model provides some information on
the likely range of MSY based on average recruitment and yield-per-recruit
values. If F0.1=0.06 is assumed to be a suitable proxy for
FMSY, then the average recruitment of 27,954 thousand age-1
recruits would produce an MSY of roughly 4,562 mt. Based on the 95%
confidence interval for the point estimate of average recruitment and
a fixed yield-per-recruit value of 0.1632 at F0.1=0.06,
the 95% confidence interval for MSY would be (4,401 mt - 4,729 mt).
In contrast, if one assumed that FMAX=0.13 was a suitable
proxy for FMSY, the point estimate of MSY would be 5,048
mt with a 95% confidence interval of (4,870 mt - 5,234 mt). Thus, the
age-structured model suggests that MSY may be on the order of 4,400-5,200
mt, a much lower value than that suggested by surplus production analyses.
However, these estimates of recruitment depend considerably on the
average recruitment applied to the yield per recruit estimates. Since
the mid-1960s, recruitment has been extremely low in most years with
the exception of a few very large year classes. Thus, an average value
which captures the observed recruitment pattern is difficult to calculate
for this stock. For similar reasons, these data provide little evidence
of a stock-recruitment relationship. Therefore, an age-disaggregated
approach, in which natural mortality, growth and recruitment are subsumed
into a single parameter, the intrinsic rate of growth (r), may provide
additional insight into the past trajectory of biomass and fishing
mortality for this stock.
A biomass dynamics model (ASPIC, Prager
1994, 1995) was developed to revise the MSY estimate and replace proxies
with direct estimates of MSY reference points that include all available
information on trends in biomass and catch. The analysis includes the
entire time series of catch since the beginning of the fishery (1934-2000),
NEFSC spring and fall survey biomass indices (1968-2000 and 1963-2000,
respectively), and the standardized CPUE series (1952-1990, Figure
36). The three biomass indices are moderately correlated (correlation
ranged from 0.42-0.63: Appendix 2). Initial
attempts to fit ASPIC had problems with convergence and sensitivity
to starting values and random number seeds. In order to reduce the
number of estimated parameters, biomass in 1934 was set equal to K
and therefore removed from estimation (Appendix
2). Initial trials that estimated B1R indicated that biomass in
1934 was near K. The assumption that the stock was at virgin biomass
in 1934 is justified, because there was no fishery prior to 1934 and
incidental catch of redfish in other fisheries was negligible. Furthermore,
life history characteristics of redfish such as long lifespan, slow
growth, slow maturity, and internal fertilization suggest that the
population is "K-selected" and will maintain a relatively stable stock
size near its carrying capacity in the absence of fishing.
Model
results
The model fit the biomass indices well (R2=0.71
for CPUE, 0.59 for the fall survey, and 0.37 for the spring survey; Figures
37, 38, and 39).
Although the observed data represents a large dynamic range (Figure
40), biomass dynamics parameters (r: intrinsic rate of increase
and K: carrying capacity) are largely influenced by a few observations.
For example, r is largely influenced by the large rate of increase
in recent years from strong recruitment, and K is largely determined
by estimates from the early years in the time series, which are not
calibrated with biomass indices (Figure 40).
The estimate of MSY is 20,000 mt (Figure
41) with an 80% confidence limit of 19,000-22,000 mt, which is
similar to a previous estimate from production modeling (Mayo 1975).
The estimate of FMSY (0.09 on total biomass, with an 80%
CI of 0.08-0.10) is consistent with life history and relatively low
productivity of redfish. The estimate of BMSY is 226,000
with an 80% CI of 211,000-244,000 mt. However, estimates of absolute
biomass from ASPIC are commonly misleading, and ratios of biomass
or F to MSY conditions are more reliable (Prager 1994). Comparisons
of biomass estimates from ASPIC, the historical VPA (NEFSC 1986)
and the present age-based dynamics model suggest that ASPIC underestimates
redfish biomass (Figure 42). Therefore,
only relative biomass and F estimates from ASPIC (Figures
43 and 44) should be considered
to be reliable. The estimate of biomass in 2001 is 33% of BMSY with
an 80% CI of 27-40%, and the estimate of F on biomass in 2000 is
estimated as 5% of Fmsy with an 80% CI of 4-7% (Table
9, "REDFISH3" in Table 10).
Sensitivity of ASPIC results to excluding the CPUE series
and estimating biomass in 1934 was assessed with alternative analyses.
Results from sensitivity analyses suggest that estimates are relatively
robust to both decisions (Table 10). Estimates
of MSY, FMSY, and BMSY and B2001/BMSY had
less than 3% difference in estimates among alternative runs, but estimates
of F2000/FMSY had slightly greater sensitivity
(9% difference). However, alternative runs that estimated B1R had problems
converging on a solution. No solution could be found when CPUE was
included and B1R was estimated. Many bootstrap trials could not converge
when B1R was estimated without including CPUE ("REDFISH2"), and results
were sensitive to random number seeds. Including CPUE in the analysis
appears to reduce variance of parameter estimates, and therefore "REDFISH3" was
chosen as the best run.
An additional analysis was performed to assess sensitivity
of model parameter estimates to the recently observed strong recruitment
by truncating the analysis to 1934-1995 ("REDFISHT" in Table
10). Results indicate that the stock is less productive (i.e.,
a 34% decrease in Fmsy) when recent observations are excluded from
the model. Therefore, when the entire time series is included in the
model, there is an explicit assumption that the recently observed high
recruitment is consistent with the long-term reproductive capacity
of the stock.
The capacity of the redfish stock to
rebuild to BMSY was assessed using ten-year stochastic projections
from "REDFISH3" assuming F=0 from 2001 to 2010. Results indicate that
the stock can rebuild to BMSY in 2010 in the absence of
fishing (Figure 45). However, the projection
implicitly assumes the higher productivity indicated by analysis of
the entire time series (i.e., including the recently observed strong
recruitment). As demonstrated in the sensitivity analyses, the estimate
of intrinsic growth rate (r ) is sensitive to recent recruitment observations.
CONCLUSIONS
The biomass of redfish in the Gulf of Maine-Georges Bank
region has increased considerably during the past decade, due primarily
to improved recruitment from several year classes of the early 1990s.
Despite this, total stock biomass is still relatively low and the age
structure remains truncated compared to earlier periods. Biomass in
2000 was approximately 1/3 of the estimated Bmsy. Catches from this
stock have been minimal since the late 1980s and have averaged less
than 500 mt per yr during the 1990s. As such, the current exploitation
rate is extremely low.
Exploitation ratios (catch/NEFSC autumn survey biomass
index) suggest that fishing mortality has been very low since the mid-1980s
compared to previous periods. Fully recruited fishing mortality (ages
9+) in 2000 was less than 0.01, well below any Fmsy reference point.
This is in contrast to the late 1970s and early 1980s when F ranged
between 0.2 and 0.3. These high fishing mortality rates coincided with
a 75% decline in exploitable biomass and a 90% decline in relative
abundance and biomass indices derived from NEFSC bottom trawl surveys
between the early 1970s and the late1980s. The existing proxy for Fmsy
(F20%) is 0.12, a relatively high value considering the
life history of the species. Other more appropriate proxies for Fmsy
are F0.1 (0.06) and F50% (0.04) (Ralston et
al. 1998; Dorn 2002).
ACKNOWLEDGMENTS
We are indebted to members of the Northern Demersal Working
Group who provided a thorough, constructive review of the initial version
of this assessment, and to several members of the Stock Assessment
Review Committee review panel of the 33rd Stock Assessment Workshop
who contributed additional analyses not provided in the initial assessment.
REFERENCES
Applegate, A., S. Cadrin, J. Hoenig, C. Moore, S. Murawski,
and E. Pikitch. 1998. Evaluation of existing overfishing definitions
and recommendations for new overfishing definitions to comply with
the Sustainable Fisheries Act. New England Fishery Management Council
Report.
Atkinson, D.B. 1987. The redfish resources off Canada's
east coast. Proc. Int. Rockfish Symp., Oct. 1986, Anchorage, Alaska.
Alaska Sea Grant Report No. 87-2, p. 15-34.
Azarovitz, T.R. 1981. A brief historical review of the
Woods Hole Laboratory trawl survey time series. In: W.G. Doubleday
and D. Rivard (ed.). 1981. Bottom Trawl Surveys. Can. Spec. Publ. Fish.
Aquat. Sci. 58: 62-67.
Beverton, R.J.H. and S.J. Holt. 1957. On the Dynamics
of Exploited Fish Populations. Fish. Invest. Lond.(2) 19: 533p.
Burnett, J., L. O'Brien, R.K. Mayo, J. Darde and M.
Bohan. 1989. Finfish maturity sampling and classification schemes used
during Northeast Fisheries Center bottom trawl surveys, 1963-1989.
NOAA Tech. Mem NMFS-F/NEC-76, 14 p.
Dorn, M. W. 2002. Advice on west coast rockfish harvest
rate from Bayesian meta-analysis of stock-recruit relationships. N.
Am. J. Fish. Management. 21: 000-000.
Doubleday, W.G. 1976. Environmental fluctuations and
fisheries management. Int. Comm. Northw. Atl. Fish. Sel Papers, No.
1: 141-150.
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.
Fox, W.W. 1975. Fitting the generalized stock production
model by least squares and equilibrium approximation. Fish Bull, U.S.,
73(1): 23-27.
Gabriel, W.L. , M.P. Sissenwine, and W.J. Overholtz.
1989. Spawning stock biomass per recruit analysis: an example for Georges
Bank haddock. North Am. J. Fish. Management 9:383-391.
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.
Kelly, G.F. and R.S. Wolf. 1959. Age and growth of the
redfish, Sebastes marinus, in the Gulf of Maine. Fish. Bull., U.S.,
60: 1-31.
Kelly, G.F. and A. Barker. Vertical distribution of
young redfish in the Gulf of Maine. Int. Comm. Northw. Atl. Fish. Special
Publ. No. 3: 220-233.
Mayo, R.K. 1976. Update of redfish status for 1975 in
ICNAF Subarea 5. Int. Comm. Northw. Atl. Fish. Res. Doc. No. 76/VI/43,
9p.
Mayo, R.K. 1980. Exploitation of redfish, Sebastes marinus
(L.), in the Gulf of Maine-Georges Bank region, with particular reference
to the 1971 year class. J. Northw. Atl. Fish. Sci., Vol. 1: 21-37.
Mayo, R.K. 1987. Recent exploitation patterns and future
stock rebuilding strategies for Acadian redfish, Sebastes fasciatus
Storer, in the Gulf of Maine-Georges Bank region of the Northwest Atlantic.
Proc. Int. Rockfish Symp., Oct. 1986, Anchorage, Alaska. Alaska Sea
Grant Report No. 87-2, p. 335-353.
Mayo, R.K. 1993. Historic and recent trends in the population
dynamics of redfish, Sebastes fasciatus Storer, in the Gulf of Maine-Georges
Bank region. Northeast Fisheries Science Center, Reference Doc. No.
93-03, 24 p.
Mayo, R.K., E. Bevacqua, V.M. Gifford, and M.E. Griffin.
1979. An assessment of the Gulf of Maine redfish, Sebastes marinus
(L.), stock in 1978. Nat. Mar. Fish. Serv., NEFC, Woods Hole Laboratory
Ref. Doc. No. 79-20, 64p.
Mayo, R.K., V.M. Gifford, and A. Jearld Jr. 1981. Age
validation of redfish, Sebastes marinus (L.), from the Gulf of Maine-Georges
Bank region. J. Northw. Atl. Fish. Sci., Vol 2: 13-19.
Mayo, R.K., U.B. Dozier, and S.H. Clark. 1983. An assessment
of the redfish, Sebastes fasciatus, stock in the Gulf of Maine-Georges
Bank region. Nat. Mar. Fish. Serv., NEFC, Woods Hole Laboratory Ref.
Doc. No. 83-22, 55 p.
Mayo, R.K., J. Burnett, T.D. Smith, and C.A. Muchant.
1990. Growth-maturation interactions of Acadian redfish (Sebastes fasciatus
Storer) in the Gulf of Maine-Georges Bank region of the Northwest Atlantic.
J. Cons. int. Explor. Mer, 46: 287-305.
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.
NEFC. 1986. Report of the Second NEFC Stock Assessment
Workshop (Second SAW). National Marine Fisheries Service, NEFC, Woods
Hole Laboratory Ref. Doc. No. 86-09, 114p.
NEFSC. 1993. Report of the Fifteenth Northeast Regional
Stock Assessment Workshop (15th SAW). Stock Assessment Review
Committee (SARC) Consensus Summary of Assessments. National Marine
Fisheries Service, NEFSC Ref. Doc. No. 93-06, 108p.
NEFSC. 2001. Assessment of 19 Northeast Groundfish Stocks.
Report of the Northern and Southern Demersal Working Groups, Northeast
Regional Stock Assessment Workshop. National Marine Fisheries Service,
NEFSC Ref. Doc. No. 01-20, 217p.
O'Brien, L, J. Burnett and R.K. Mayo. 1993. Maturation
of 19 species of finfish off the northeast coast of the United States,
1985-1990. NOAA Tech Report NMFS 113, 66p.
Pella, J.J. and P.K. Tomlinson. 1969. A generalized
stock production model. Bull. Inter.-Amer. trop. Tuna Comm., 13(3):
419-496.
Perlmutter, A. and G.M. Clarke. 1949. Age and growth
of immature rosefish (Sebastes marinus) in the Gulf of Maine and off
Nova Scotia. Fish. Bull., U.S., 51: 207-228.
Prager, M.H. 1994. A suite of extensions to a non-equilibrium
surplus-production model. Fish. Bull. 92: 374-389.
Prager, M.H. 1995. Users's manual for ASPIC: A Stock
Production Model Incorporating Covariates, program version 3.6x. Miami
Laboratory Document MIA-92/93-55, National Marine Fisheries Service.
29p.
Ralston, R, J.R. Bence, W.G. Clark, R.J. Conser, T.
Jagielo and T.J. Quinn II. 2000. Panel Report: West Coast Groundfish
Harvest Rate Policy Workshop. Sponsored by Scientific and Statistical
Committee, PFMC. Alaska Fisheries Science Center, Seattle, WA: March
20-23, 2000.
Restrepo, V. R., and C. M. Legault. 1998. A stochastic
implementation of an age-structured production model. University of
Alaska Sea Grant College Program, Report Number 98-01: 435-450.
SAS 1990. SAS/STAT User's Guide, Version 6, Fourth Edition.
SAS Institute, Inc., Cary, NC.
Schaefer, M.B. 1954. Some aspects of the dynamics of
populations important to the management of the commercial marine fisheries.
Bull. Inter.-Amer. trop. Tuna Comm., 1(2): 25-56.
Schaefer, M.B. 1957. A study of the dynamics of the
fishery for yellowfin tuna in the eastern tropical Pacific Ocean. Bull.
Inter.-Amer. trop. Tuna Comm., 2(6): 245-285.
Steele, D.H. 1957. The Redfish (Sebastes marinus L.)
In the western Gulf of St. Lawrence. J. Fish. Res. Bd. Canada, 14:
899-924.
Templeman, W. 1959. Redfish Distribution in the North
Atlantic. Bull. Fish. Res. Bd. Canada, No. 120: 1-173.
Thompson, G. G. 1994. Confounding of gear selectivity
and the natural mortality rate in cases where the former is a nonmonotone
function of age. Can. J. Fish. Aquat. Sci. 51:2654-2664.
Thompson, W.F. and F.H. 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. Fish. (Pacific Halibut)
Comm. 8: 49p.
Walter, G.G. 1976. Non-equilibrium regulation of fisheries.
Int. Comm. Northw. Atl. Fish., Sel. Papers, No. 1: 129-140.