International Council forthe Exploration of the Sea

ICES C.M. 19951P: 14Theme Session on Causes ofObserved Variation in Fish Growth

Abstract

•

Factors influencing weight at age of cod off eastern Newfoundland (NAFO Divisions 2J+3KL)

Peter A. Shelton and George R. LillyScience Branch, Department of Fisheries and OceansNorthwest Atlantic Fisheries Centre, P.O. Box 5667

St John's, NewfoundlandCanada AIC 5Xl

{eI ~c.Jv.J "tlo.Lc.u} }I /(c,'Jl.lfckn 1J;tI<f)!v> tf/

Growth increments and condition factor in the northern cod stock (~AFO Divs. 2J, 3K and3L) measured from research vesse1 trawl sampies show considerable variabllity over the period1978 to 1994. This variability in the data is examined through a variety 9f plots. Scatter plots ofgrowth and condition data against temperature, cod biomass and capelin'abundance in codstomachs are used to look for possible corre1ations. A near-decadel scale cycle in the area of thecold intermediate layer (CIL) is correlated with the cycle in cod growth. The generally decliningtrend in temperature is correlated with the declining trend in growth. Correlations between theabundance of capelin in cod stomachs and cod growth or condition are not easy to interpret andprobably require more attention to temporal and spatial patterns of abundance of predator and preyto resolve. Correlations with cod biomass are weak and, with the exception of Div. 3L, arepositive, Le. do not provide evidence of density dependent growth. It is important to take changesin growth into account in future management of this cod stock and predictive models based onvariables such as the area ofthe.CIL would be very useful.

Introduction

Growth of Atlantic cod (Gadus morlllla) varies considerably among stocks, with much ofthe differcnce attributable to variability in ambicnt temperature (Brander 1995). There is alsopersistent geographie variability within some stocks, such as those around Iceland (J6nsson 1965)and off Labrador and eastern Newfoundland (Fleming 1960; May et al. 1965). Within the latterregion there has historically been a cline in growth, from very slow growth off central Labrador torelatively rapid growth on the southwestern Grand Bank.

Annual variability in growth has been reported for many eod stocks, with importantimplieations for fisheries management. Studies of faetors responsible for this variability haveemphasized temperature, stock size and the abundance of prey. In the stock of eod occupyingNorthwest Atlantie Fisheries Organisation (NAFO) Divisions 2J, 3K and 3L (eommonly ealled thenorthem eod stock), ehanges in length-at-age of eod were found to be affeeted by both the biomassofthe eod stock (WeHs 1984, 1986; MilIar and Myers 1990) and temperature (MilIar and Myers1990). There has also been cancern that ead growth might be affected by changcs in abundanec ofcapelin, the most important prey of eod on the Northcast Newfoundland Shclf (Lilly 1987).Declincs in eapelin abundance have becn eorrelated with reduetions in growlh rate of eod in watersaround Iccland (Steinarsson and Stcfansson 1991) and with a reduction in bath growth rate and

•

2

somatie eondition of eod in the Barents Sea (Mehl and Sunnanill991; Jprgensen 1992).However, neither Akenhead et al. (1982) nor Millar et al. (1990) eould find a statistieallysignificant relationship between cod growth and capelin abundance off Newfoundland. The lackof statistical signifieance might reflect weak linkage between cod and its prey, but might also refleetan inability to detect a link because of uncertainty in the timeseries of capelin abundance (Akenheadet al. 1982; Shelton et al. 1991).

In 1990 the ICES Multispecies Assessment \Vorking Group undertook an analysis ofgrowth in cod from four Arcticlboreal systems (Barents Sea, Iceland, Greenland andNewfoundland, Anon 1991). Length at age and length increments were examined for stockdensity, capelin abundance and temperature effects. A significant residual year effect was found inmost of the data sets after accounting for these three factors. At a subsequent meeting of the\Vorking Group (Anon 1992) cod stornach content data for the North Sea, Baltic Sea, Barents Sea,Iceland, Newfoundland and Northeast USA were compiled into a common format and analysed insome detail. At this same meeting further, but preliminary analysis of the cod growth data basewas carried out taking into account the information in the cod stornach content data base. It wasconc1uded (Anon 1992) that in the eomparison among ecosystems "the relationship betweengrowth and feeding remains confounded by the inter-dependencies of growth, temperature­dependent stornach evacuation and differences in environmental temperature".

Brander (1995) extended the analysis of the temperature effect to 17 cod stocks andconc1uded that 92% of the variance in the logarithm of mean weight at age for ages 2 to 4 fishamong stocks can be explained by a ANCOVA model with age (dass) and temperature(eontinuous) effects. The weight at age data used in Brander's study eame mainly from thesampling of the commercial eatch. For the stock off Labrador and the east eoast of Newfoundland,and possibly for other stocks, the weight at age in the commercial eatch is inferred from a eonstantlength-weight relationship. Although the comparisons among cod stocks by Anon (1991, 1992)and Brander (1995) has much merit, we believe that more detailed intra-stock analyses are requiredif the combined and probably interacting effects of temperature, food, population density andfishing mortality on cod growth are going to be disentangled.

In this paper we return to the problem of explaining cod growth variability off southern •Labrador and the cast eoast of Newfoundland in further detail. In particular we attempt to (i)account for the length-stratified approach to the subsampling of the research trawl catch, (ii)consider thermal measures additional to those made at Station 27, (iii) give greater consideration toweight at age data, (iv) evaluate condition factor data, and (iv) give further consideration to thepossible effects of capelin abundance in cod stomachs on cod growth. The approach in this studyremains primarily descriptive and exploratory. Formal hypothesis testing and statistical inferenceleading to predictive models ofuse in stock assessments have been attempted in the past (e.g.Miliar and Myers 1990) and are an objective for future studies by ourselves. Strong signals inocean temperature data (e.g. Colbourne 1995), eod population size (see Bishop et al. 1995), theabundance of capelin in cod stomachs (Lilly 1993, 1994, 1995), and possibly the abundance ofcapelin in the ocean (Carscadden 1994) should facilitate statistical inference, however, beforeproceeding further with such analyses, wc believc that a solid base of data description andexploration is required. Although the present paper docs not complcte this task far thc southcrnLabrador and cast coast of Newfoundland region, wc hope it is a further step in this direction.

3

Data sources

Temperature measurements

Temperature profiles from surface to bottom (175m) are routincly taken throughout the yearat Station 27 just off St 10hn's (Fig. 1) at the commencement and termination of nearly everyresearch cruise (Colbourne et al. 1994, Colbourne 1995). For this analysis all data for the period1978 to 1994 have been used to compute annual average water column temperature. In addition,the bottom temperature anomaly from the longterm mean (1961-90), following MilIar and Myers(1990), has been considered.

• Oceanographic transects across the Hamilton Bank (Seal Island) in Div. 21, off BonavistaBay across the Div. 3K1 Div. 3L boundary, and in Div. 3L shorewards from the Flemish Cap alongthe 470 N latitude have been surveyed regularly in summer since the 1950s (Colbourne et al. 1994,Colbourne 1995). Annual estimates of the area of the cold intermediate layer (CIL), defined as~O°C, have been caIculated as an index of thermal conditions within the northern cod habitat(Coibourne 1995) and are used in this stu~y.

Estimates of cod abundance

•

Two sources of information on cod abundance were considered - research vessel (RV)estimates for each division and estimates from sequential population analysis. RV surveys havebeen conducted by Canada during the fall in Divs. 21, 3K and 3L since 1977, 1978 and 1981respectively (excluding 1984 in Div. 3L). All surveys in Divs. 21 and 3K were conducted with the74 m stern trawler R.~ 'Gadus Atlantica'. Surveys in Divs. 3L were conducted with the 51 mside trawler R.~ 'A. T. Cameron' and the sister 50 m stern trawlers R.~ 'Wilfred Templernan'and R.~ 'Alfred Needler'. The 'Gadus Atlantica', 'Wilfred Templernan' and 'Alfred Needler'deployed an Engcl-145 trawl, whereas the 'A. T. Cameron' deployed a Yankee 41-5 trawl. In allinstances, a 29 mm meshliner was inserted in the codend. Tows were made at 3.5 knots for 30min at each fishing station, and catches from the few tows of duration other than 30 min wereappropriatcly adjusted. No adjustments were made for possible between-vessel differences incatching efficieney. Details regarding areas and 10eations of strata and changes in survey patternare provided by Bishop et al. (1994), Lilly and Davis (1993) and Bishop (1994). The mostnotable change in survey eoverage was the addition of depths between 100 and 200 m innorthwestern Division 3K (St. Anthony Shelf and Grey Islands Shelf) in 1984 and subsequentyears. Fishing in all divisions and years was eonducted on a 24-h basis. Set alloeation withineach stratum is random, with the number of sets determined primarily by the area of the stratum.Boundaries to strata correspond to water depth eontours. The RV estimates of trawlable biomassby division obtained from the program STRAP (Smith and Somerton 1981) are provided in Bishopet al. (1995).

Aversion of s~quentialpopulation analysis, ADAPT, has been routinely applied in northerneod assessments to eommercial eatch at age data and tuned using the RV estimates of abundaneefrom STRAP. ADAPT estimates ofbiomass in Divs. 2J3KL eombined are provided in Bishop etaI. (1993).

4

Length and weight at age sampIes from fall RV surveys

The number of cod sampled for which length and body weight measurement are availableand for which age was determined directly from the otolith (rather than an age-length key) variesacross years and among ages (Tables 1-3). During the early years of the surveys, otoliths foraging were obtained from a sampIe of 25 cod per 3-cm length-group per division. An additionalsampIe of 5 cod per 3-cm length-group was frozen at sea and thawed in the laboratory, whereotoliths were extracted and body weight (both whole and gutted, head-on) was recorded. All fishwere measured (fork length, cm) at sea. Additional sampIes of frozen fish were collected in Div.21 in 1984 and 1985. The number of frozen fish increased in 1988 and subsequent years, whenDivs. 21 and 3K were both subdivided into 2 areas, and a sampIe of 5 cod per 3-cm length-groupwas obtained from each area. The number of cod for which weights were available increaseddramatically when weighing at sea was initiated (in 1989 in Divs. 21 and 3K and in 1990 in Div. •3L, Table 2). Length-at-age was determined for a total of 30,535 cod (Table 1) over the studyperiod. Body weights were obtained for 13,822 of these cod (Table 2).

The cod for which length and weight at age were determined represent a subsampIe of thetrawl catch at each station. The subsampIe is length stratified - where possible 25 fish are collectedfrom each 3 cm length interval in each of the three divisions over the course of the survey. From1989 onwards in Divs. 2J and 3K and from 1990 onwards in Div. 3L body weights have beenrecorded at sea prior to freezing and these weights have replaced the thawed weight previouslyentered into the data base. Although these two weights are expected to differ systematically, acorrection has not yet been applied (although the data are avialable to do so) and in the presentstudy the differe~ce has bcen ignored.

In addition to the subsampIe, the length frequency of the entire catch for each set, or arandom portion of the catch if it is too large to process, is also determined. The sampie lengthfrcquency is transformed into a population length frequency in each division using the programSTRAP (Smith and Somerton 1981). In this study population mean length and weight at age bydivision were obtained by weighting the individual measurements in the bio10gical sampIe by theratio of the sampIe number per 3 cm length dass to the STRAP estimate of the number in thepopulation in each 3 cm length dass. •

The average number of length and weight sampIes by age for the period 1978 to 1994 aresummarised in Table 3. It is evident that after age 10 the number of length sampIes is considerab1yreduced «20 per year on average). The number of sampIes for weight is 10w «20 per year) afterabout age 8. For this reason the present analysis was restricted to ages 10 and 1ess for length atage and ages 8 and less for weight at age. Although the number of sampIes are also low for agesless than 3, there is relatively little variance in length and weight at age among sampIes within ayear, so ages 1 and 2 have been induded in the analysis.

Cod condition measurements from fall RV sampIes of length and weight at age

Condition was cxprcssed as Fulton's condition factor:K = 100*(\V/L3)

where \V is body weight (g) and L is fork length (ern). 80th whole body (round) weight andguUed (head on) weight were used. Round weight is of greater significance when considering theimpact of growth on stock assessments and the TAC. Gutted weight,with liver and gonad weightmeasured separately, are of greater significance when considering aspects of bioenergetics and are

•

•

5

eurrently subjects of further study not reported here. Beeause growth in eod is not isometrie,annual variability in eondition was eompared within speeifie eod length-groups rather than withinage-groups.

Cod stornach contents from fall RV surveys

Stomaehs were eolleeted from up to 3 randomly seleeted eod per lO-em length-group perstation in 1980-1982 and 3 per 9-em length group in 1983-1994. Stomaehs were not eolleetedfrom fish which showed signs of regurgitation, such as food in the mouth or a flaecid stomaeh.Stomaehs were individually tagged and excised. In 1978-1993, the stomaehs were fixed andpreserved in 4% formaldehyde solution in semvater prior to examination of their eontents in thelaboratory. In 1994, the stomaehs were frozen prior to examination.

Examination involved separation of food items into taxonomie eategories. Fish anddeeapod erustaeea were identified to species, but most other groups were assigned to higher ordertaxa. Hems in eaeh taxon were plaeed brieflyon absorbent paper to remove exeess liquid, and theneounted and weighed to the nearest 0.1 g.

The quantity of eapelin in the stomaehs of the eod from a specified sampie was expressedas a mean partial fullness index (Fahrig et aI. 1993):

where Wej is the weight (g) of eapelin in fish j, Lj is the length (ern) of fish j, and n is the numberof fish in the sampie. This index is based on the assumption that stomaeh capacity is apowerfunetion of length, and is analogous to Fulton's eondition faetor. Mean total fullness index wasealeulated as

I n WTFI=-L~·104

n'_ 1 L 3J- j

where \Vlj is the total weight of prey in fish j.

For simplieity, the present analysis was restrieted to eod within the 36-71 em length range,and all eod within this range were pooled. Cod smaller than about 30-35 cm cannot feed on thelargest capelin and eod larger than about 70 em tend to feed to an inereasing extent on groundfishand erabs (Lilly 1991).

Capelin abundanee estimates

Estimates of capelin biomass from hydroaeoustie surveys condueted by Canada and Russiaduring the fall in Divs. 2J and 3K are available intermittently from 1974 onwards (Miller 1994).Considerable uneertainty exists regarding some of the estimates, partieularly those in recent years(see for example Carseadden 1994).

6

Data description

Temperature

\Vater temperatures on the shelf off Labrador and Newfoundland have been below thelongterm mean for most of the period since 1980, with two unusually cold periods in the mid­1980s and early 1990s (CoIbourne et al. 1994). These recent two cold periods separated by awarmer period ean be c1early seen in the annual average water eolumn temperature und bottomtemperature anomaly plotted in Fig. 2. Underlying this near-deeadel seale variability is longerperiod variability apparent as a downward trend in temperature over the study period with eurrentvalues weil below the relatively warm 1950s and 1960s (CoIbourne 1995). The near-deeadel sealevariability is c1early illustrated in the unnual variability in the area of the CIL calculated along thethree transeets within the study area Fig. 3). As weil, the upward trend in the area of the CIL, •eorresponding to the eooling trend in Fig. 2, ean be seen.

Capelin

Estimates of eapelin biomass from hydroaeoustie surveys eondueted by Canada and Russiaduring the fall in Divs. 21 and 3K deelined to very low levels in the early 1990s with the Canadiunseries showing a very abrupt drop between 1989 and 1990. Low levels of eapelin in the acoustiefall survey in Divs. 21 and 3K, as weH as in the spring survey in Div. 3L have persisted insubsequent years. The 1983 und 1986 eapelin year c1asses are thought to havc been particularlystrong (Carseadden 1994).

Peaks in eapelin PFI in eod stomaehs (Fig. 5) in Div. 21 oecur in the mid 1980s and in thelate 1980s, eorresponding to the high values in the Canadian hydroacoustie survey. Subsequentlythe eapelin PFI dropped to a very low level und their was a eorresponding drop in TA. In Div. 3Keapelin PFI increased over the 1980s and was fairly eonstant over the early 1990s, dropping to alow level only in the most reeent year for which data are available (1994). The eapelin PFI in Div.3L increased in 1990 and has remained at a high level up to und including the most reeent survey.It is c1ear that there is little compensation in the eontribution of other food in the diet of eod and that •the trends in TFI are therefore driven mostly by the eapelin contribution (Lilly 1991).

It is important to eonsider changing geographie patterns of eod and eapelin abundanee inthe interpretation of these data (Lilly 1994). Despite the apparent low abundanee of eapelin in the1990s, eod in Divs. 3K und 3L had a relatively high eapelin PFI, at least partly beeause the eapelinchanged their distribution and occupied the restricted area whcre bulk of thc remaining eodpopulation was eneountered during the fall RV survey. The low eapelin PFI in Div. 21corresponds to a very low abundanee of eod in this division after 1989.

Cod abundanee

The time series of eod abundance in fall RV surveys and from the ADAPT estimates areillustrated in Figs. 6 and 7. The RV estimates fluetuated with no apparent trend up to 1988. Highestimates in 1986 in Divs.21 and 3K relative to estimates before und after are not easily interpretcd.Thc progressive dccline in Div. 2J, Div. 3K und Div. 3L in the Jute 1980s und 1990s to the eurrentlow biomass is c1early seen. The ADAPT time series is smoother and suggests that the biomassrose to a peak in 1984 following extension of jurisdiction and then declined monotonically to 1992,

•

7

after which sequential population analysis has not been applied. These perspectives on the changesin cod stock size, although not independent because of the tuning exercise, are somewhat differentand both have been considered in the data exploration exercise that follows.

Cohort length at age

The cohort length at age for Divs. 2J, 3K and 3L are plotted in Fig.8 for ages ~ 10. Notethat because the point estimates are from the fall of each year and growth is slow over the winter,most of the increment in length or weight is occurring in the following year. Cohorts are takenback to age 0 at which age length is assumed to be 0 to facilitate the identification of cohorts in theplots. Lines roughly parallel to the x-axis identify the ages whereas lines roughly parallel to the y­axis indicate cohorts. Ridges corresponding to fast growing cohorts and valleys corresponding toslow growing cohorts can be clearly seen in the plot fer Div. 2J and to a lesser extent in the plot forDiv. 3K. The Div. 3L "surface" is relatively flat.

The ridges in the cohort growth correspond to fast growing 1976-78 cohorts and fastgrowing cohorts from the early to mid 1980s. The elevated length increment does not occur in allages in these cohorts but by age 4 the effect of the cumulative incremcnt is clear and remainsdistinct to age 6, after which the general slowing down in growth and increasing mcasurementeITor tend to obscure the pattern. The valleys correspond to slow growing cohorts from the mid1970s,late 1970s-early 1980s and late 1980s. The most marked declines in growth in thesecohorts (lowest points to the valleys) occurred in the mid 1980s and early 1990s, primarily withinages 2 to 5, particularly in Div. 2J but also in Div. 3K. Year-to-year variability is much lesspronounced in the Div. 3L data.

In addition to these cohort and year effects in the growth, it is evident that there is anoverall decreasing trend in length at age in Divs. 2J and 3K illustrated by the downward slope ofthe iso-age lines with time. A similar trend is not apparent in the Div. 3L data.

• Cohort weight at age

The cohort weight at age plots (ages ~ 8) for the 3 divisions (Fig. 9) show similar patternof ridges and valleys to those described above for the cohort length at age plots. The overalldownward trend in weight at age in Divs. 2J and 3K is also quite marked, illustrated by thedecrease in area of the rectangles bounded by age and cohort lines in the right portion of the plots.There is some indication of an increase in weight at age for the 1989+ cohorts in the older ages andmost recent years.

Year effects in length and weight increments

In an attempt to see the year effects in length and weight increments more clearly, thegeneral approach taken in the 1992 Multispecies Assessment \Verking Group (Anon 1992) wasadopted. A multiplicative model accounting for age and year effects were fitted to the logarithm ofthe annual increments for each division separately,

Ln(X 1··) = u· + ß· + EY I J ,

8

where Xij is the length or weight in a particular division at age i in year j, <Xi is the age effeet, ßj is

the year effeet and e is normally distributed error. The vector ßj for both length and weight is

plotted for each division in Fig. 10. The year effects for both length and weight inerements inDivs. 2J and 3K have a similar pattcrn and are compatible with the downward trend and near­decadel seale variability described in the cohort plots. As explained above, the effeet ascribed toyear t oeeurs mainly in year t+1, requiring that these year effeet estimates be lagged by one year inthe subsequent data exploration exercises described below.

Condition

Annual variability in eondition basd on round weight was similar among eod length­groups, but differed among divisions (Fig. 11). In Div. 2J, average condition based on roundweight inereased slowly but irrcgularly in the 1980s before deelining abruptly in the early 1990s.There is some indication of an inerease in the most reeent two years. In Div. 3K, conditionfluetuated without trend until the mid-1980s, when it inereased for about 4 years before droppingbaek to the earlier level. In Div. 3L, there was a small inerease in the 1990s.

Condition based on gutted weight (Fig. 12) was less variable than condition based onround weight. In Div. 2J, there was no trend until a rapid decline in the early 1990s, followed bysome improvement in reeent years. In Div. 3K, condition based on gutted weight did not rise asprominently in the mid-1980s as did eondition based on round weight. In Div. 3L, eonditionbased on gutted weight was variable but without trend. Similar patterns in annual variability ineondition based on gutted wcight are apparent when the data are presented by age-group (Bishopand Baird 1994; Taggart et al. 1994; Bishop et al. 1995).

•

Data exploration

A data exploration exercise was carried out to eompare the patterns described above for cod •growth and the "environmental" variables. The primary tool adopted for doing this was seatter plotanalysis and the construction of 95% bivariate normal density ellipses. Correlation in the variablesis scen by the eollapsing of the ellipses along the diagonal. If the ellipses is fairly round and notoricntated along the diagonal then the variables are not eorrelated. In this exercise correlations arenot taken to indieate eause and effeet, but to suggest possible relationships that could be examinedfurther.

Correlations with eapelin

The PA for capelin in eod stomaehs is plotted against cod Iength inerements, weightinerements and eondition faetor in Fig. 13. Correlations with length and weight inerements aregenerally weak. The Div. 2J PFI for capelin show little correlation with cod length and weightinerements but is positively correlated with condition factor in Divs. 2J and 3K and negativelycorrelated with condition in Div. 3L. Div. 3K capelin PA is negatively corrclated with Iength andweight increments in all divisions but in some cases the correlation is very weak. There is noeorrclation with condition factor. Capelin PFI for Div. 3L is weakly negatively correlated withIcngth and weight increments. However there is a relatively tight cIlipses indieating negative

•

•

9

correlation in the scatter plot with Div. 21 and 3K condition factors and a positive correlation withDiv. 3L condition.

The generaIly negative correlations, albeit weak, between capelin PA and cod length andweight increments are unexpected. Increased consumption of capelin, and hence generaIlyincreased total consumption, might rather be expected to translate into increased growth increment.The positive correlation between capelin PFI in Div. 21 and cod condition in Divs. 21 and 3K isreassuring. The differing correlations between capelin PA and condition between Divs. 21/3K andDiv. 3L might suggest that good feeding conditions in the north occur at the expense of fish in thesouth, and vice versa. It is of interest that the intermediate area, Div. 3L shows virtuaIly nocorrelation with any of the growth or condition data.

Correlations with cod biomass

The RV survey biomass estimates for the three divisions and the ADAPT estimate of 3+biomass are plotted against the growth and condition data in Fig. 14. \Vith few exceptions, thecorrelations are weakly positive. Correlations with Div. 3L condition are negative. The tightestpositive correlations are between weight increment estimates in Div. 21 and RV estimates in Div. 21as weIl as ADAPT estimates. Correlations with population size are generaIly anticipated to benegative because of density dependent effects, however in this study period the correlations areconsistent with a growth contribution to the observed biomass changes.

Correlations with Station 27 temperature data

Average annual water column temperature and annual bottom temperature anomaly areplotted against the growth and condition data in Fig. 15. Correlations vary in strength but aregeneraIly positive with growth and condition data for Divs. 21 and 3K but negative with the Div.3L data. Correlations are weakest with respect to Divs. 21 and 3K length increments. \Vith theexception of Div. 3L, these correlations are consistent with warm water conditions causingenhanced growth.

Correlations with area of the CIL

Annual estimates of the area of the CIL for the three seetion as weIl as the average area forall three sections combined are plotted against growth and condition data in Fig. 16. Correlationsare generally negative, with the exception of Div. 3L , and in some cases are quite strong. Forexample, weight increments in Divs. 21 and 3K are considerably smaIler when the area of the CILis large. Div. 21 length increment data are also negatively correlated with area of the CIL but therelationship with Div. 3K length increment data is not dear.

Further exploration oLthe correlation with area of the CIL

To examine the correlation in growth increments with the average area of the CIL from thethree seetions more closely, the totallength and weight increment in each cohort from the beginningof age 2 to the end of age 5 was plotted against the average area of the CIL experienced by thecohort over that period (Fig. 17). For the CIL data this is equivalent to a 3 year running mean andbrings out the near-decadel scale cycle very clearly. Evaluating the growth increment over the three

10

years of the cohort's life when most of the growth is occurring and for which most of the data arcavailable brings out the near-decadel scale of variability in growth more clearly than can be seen byplotting year effects in the increments. This is probably at least in part because the ycar to yearerrors in the sampIe estimates canceI. \Varren (1993) applied the KaIman Filter technique to thecod growth data for Divs. 21, 3K and 3L for a longer time period. This technique appears to bepurticularly useful in picking out both the downward trend and the near-decadel eyele in growth.For the time periods that overlap the pattern is very similar in the study by \Varren (1993) and inthe present study.

In the case of both length and weight increments, but most cIearly with respect to weight,the increments decline when the arca of the CIL increases and increase when the area of thc CILdeclines. The overall declining trend in growth, particularly strong with respect to weight, is alsoapparent, corresponding to the increasing trend in CIL, most easily seen in the plots in Fig.3.

Diseussion and Conclusions•

Cod growth has varied considerably over the time period considered in this study. Thereappears to be at least two components to this variabiIity: (i) a downward trend over the entircperiod and (ii) an approximately decadel eycle in growth. The combined effect, as illustrated inFig. 17 is that the weight increment by cohorts between the ages of 2 and 5 reached a minimum inDivs. 21 and 3K in the 1986 and 1987 cohorts. Tbe decline in weight increment is quitcsubstantial and could perhaps be referred to as a "collapse in growth". The 1976 and 1977 cohortsbenefited from being at both the high end of the trend and the high point in thc approximatelydecadel eycIe, and increased by nearly 1.6 kgs over ages 2 to 6. In comparison, the 1986 and1987 cohorts were at the low end of the trend and the low point in the cycle and increased by onlyabout 400 g in Div. 21 and 700 g in Div. 3K. Thus thc weight increment over these ages decreasedby more than 1 kg for cod in Div. 21. The timeseries for Div. 3L is less complete but shows muchless evidence of a similar trend or eycle.

Isolating the causes for the trend and eycle in cod growth in Divs. 21 and 3K would have"important implications for management of the northem cod stock from both thc TAC-fishing •mortality aspeet and from a spawner stock per recruit point of view. Several variables cxamined inthis study show weak to strong corrc1ation with these changes, but causation is still a matter ofspeculation. Relationships with temperature, particularly the area of the CIL, may have the mostpromise with respect to cxplaining thc near-decadel eycle in cod growth. The downward trend incod growth is correlated with the overall downward trend in ocean temperaturc over the period ofthc study. Howcver, other factors, such as thc generally incrcasing fishing mortality over thestudy period mayaIso be important. Fishing mortality could progressively wced out the fastgrowing genotypes from thc population leaving only genotypes with a low growth capacity.

The effect of capelin on cod growth, whilc anticipated to occur, is not easy to isolate in thccurrent data. It is thought that the 1983 and 1986 year cIasses werc strong (Carscadden 1994,leading to biomass peaks about two years Iater in the hydroacoustic surveys (Fig. 4). Cod growthwas in fact low following the 1983 year class, however two years after the 1986 eapelin yearclasseod growth rate was high and/or inereasing. Capelin in eod stomachs in Div. 21 inereased in thelate 1980s when cod growth rate and condition werc high and then decreased sharply to very lowlevels after 1990, eoineident with the abrupt decline in the weight inerement und the eonditionfaetor, although the overall eorrelation of eapelin PFI with weight is weak. The general increasc incapelin in eod stomachs in Divs. 3K and 3L over the time period is not compatible with changes in

•

•

ll

growth and condition in Div. 3K, but is consistent with the trend of improving condition in Div.3L. At this stage the capelin effect on cod growth and condition should be interpreted asequivocal. It seems almost certain that resolution of the temporal and spatial patterns of abundanceof cod and capelin are an essential prerequisite to the interpretation of capelin effects on cod (seeLilly 1994 for more insight into this). .

Density dependent effects on cod growth rate have been suggested in the past, particularlywhen the pattern in the residuals are examined after removing the predominant temperature effect(e.g. MilIar and Myers 1990). In this data visualization and exploration exercise there is noevidence of density dependent growth in northern cod in Divs 21 and 3K, although there arecorrelations compatible with density dependent growth in Div. 3L. In Divs. 21 and 3K cod growthand condition are in fact generally positively correlated with biomass estimates supporting theinterpretation of an accelerated growth component to the increase in biomass following extensionof jurisdiction in 1977.

Difference in the sign of the correlations between Divs. 2113K and Div. 3L with respect tothe variables considered in this study suggest that fish on the Labrador and north eastNewfoundland shelves are responding differently to those on the northern Grand Bank. This isparticularly evident with respect to the increasing condition factor in Div. 3L in the late 1980s andearly 1990s at a time when condition factor was dropping precipitously in Divs. 21 and 3K.Reasons for the spatial differences over the range of northern cod stock need to be examined inmore detail.

The cod stock in NAFO Divs. 2J3KL are presently under a fishery moratorium followingthe very large decline in biomass in the late 1980s and early 1990s. This decline coincided withrapidly increasing fishing mortality which must be considered to be the major cause. Howeverwhat amounts to a collapse in growth in cohorts entering the fishery at this time amplified thefishing mortality effect considerably and reduced the biomass and yield per recruit. While it isimportant to resolve the various contributors to the decline in the northern cod stock, the isolationof predictive relationships between cod growth and, for example, temperature and capelinconsumption, could have considerable value in adjusting future fisheries management for changesin cod growth rate.

Acknowledgements

\Ve acknowledge the considerable efforts ofDFO personnel who collected the sampIes onwhich this analysis is based. Our recently retired colleague Claude Bishop and his predecessors inthe Gadoids Section in particular have done much to ensure that a comprehensive biological database is available for northern cod. Our colleague, Eugene Colbourne, is thanked for providingStation 27 and CIL data from the oceanographic data base and for advice on interpretation.

References

Akcnhcad, S. A., J. Carscadden, H. Lcar, G. R. Lilly, and R. Wclls. 1982. Cod-capcIin interactions off northeastNewfoundland and Labrador, p. 141-148. In M. C. Mercer (cd.) Multispecies approaches to fishericsmanagement advicc. Can. Spec. Publ. Fish. Aquat. Sei. 59.

Anon. 1991. Report ofthe Multispecics Assessment Working Group, Woods Hole, 4-13 Decembcr, 1990. ICESC.M. 19911ASSESS:7, 246p.

12

Anon. 1992. Report of the Multispeeies Assessment Working Group, Copenhagen, 16-25 June 1992. ICES C.M.19921ASSESS:16, 152p.

Bishop, C.A., E.F. Murphy, M.B. Davis, J.W. Baird and G.A Rose. 1993. An assessment of the eod stock inNAFO Divisions 2j+3KL. NAFO SCR Doe., No. 86, Serial No. N2271 , 51p.

Bishop, C. A 1994. Revisions and additions to stratifieation sehemes used during research vessel surveys inNAFO Subareas 2 and 3. NAFO SCR Doe. 94143, Serial No. N2413. 23 p.

Bishop, C. A., J. Anderson, E. Dalley, M. B. Davis, E. F. Murphy, G. A. Rose, D. E. Stansbury, C. Taggart, andG. Winters. 1994. An assessment of the eod stock in NAFO Divisions 2J+3KL. NAFO SCR Doe.94/40, Serial No. N241O. 50 p.

Bishop, C. A, and J. W. Baird. 1994. Spatial and temporal variability in eondition faetors of Divisions 2J and3KL eod (Gadus morlzua). NAFO Sei. Coun. Studies 21: 105-1I3.

Bishop, C. A, D. E. Stansbury, and E. F. Murphy. 1995. An update of the stock status of Div. 2J3KL eod. DFOAtlantie Fisheries Res. Doe. 95134. 38 p.

Brander, K. M. 1995. The effeet oftemperature on growth of Atlantic eod (Gadus morlzua L.). ICES J. mar. Sei.52: 1-10.

Carseadden, J. 1994. Editor. Capelin in SA2 + Div. 3KL. DFO Atlantie Fisheries Research Doeument 94118, 164p.

Colbourne, E., S. Narayanan, and S. Prinsenberg. 1994. Climatie ehanges and evironmental eonditions in the .Northwest Atlantic, 1970-1993. ICES mar. Sei. Symp., 198:311-322.

Colbourne, E. 1995.' Oeeanographie eonditions and climate change in the Newfoundland region during 1994. DFOAtl. Res. Doe. 9513, 36p.

•

Fahrig, L., G. R. Lilly, and D. S. Miller. 1993. Predator stomaehs as sampling tools for prey distribution: Atlantic •eod (Gadus morlzua) and eapelin (Mallotus villosus). Can. J. Fish. Aquat. Sei. 50: 1541-1547.

Fleming, AM. 1960. Age, growth and sexual maturity of eod (Gadus morhua L.) in the Newfoundland area, 1947­1950. J. Fish. Res. Board Can. 17: 775-809.

J6nsson, J. 1965. Temperature and growth of eod in Icelandie waters. ICNAF Special Publieation 6: 537-539.

Jprgensen, T. 1992. Long-term ehanges in growth of North-cast Aretic eod (Gadus morlzua) and someenvironmental influenees. ICES J. mar. Sei. 49: 263-277.

Lilly, G. R. 1987. Interactions betwcen Atlantie eod (Gadus morhua) and eapelin (Mallotus villosus) off Labradorand castern Newfoundland: a review. Can. Teeh. Rep. Fish. Aquat. Sei. 1567: 37 p.

Lilly, G. R. 1991. Interannual variability in predation by eod (Gadus morhua) on eapelin (Mal/olus villosus) andother prey off southern Labrador and northeastern Newfoundland. ICES mar. Sei. Symp. 193: 133-146.

Lilly, G.R. 1994. Predation by Atlantie eod on eapelin on the southern Labrador and Northeast Newfoundlandshelves during aperiod of changing spatial distributions. ICES mar Sei. Symp. 198:600-611.

•

•

13

LiIly, G.R. 1995. Did the feeding level of the cod off southern Labrador and eastern Newfoundland decline in the1990s? DFO Atlantic Fisheries Research Document 95174, 25p.

LiIly, G. R., and D. J. Davis. 1993. Changes in the distribution of capelin in Divisions 21, 3K and 3L in theautumns of recent years, as inferred from bottom-trawl by-catches and cod stornach examinations. NAFOSCR Doc. 93/54, Serial No. N2237. 14 p.

May, A. W., A. T. Pinhorn, R. WeHs, and A. M. Fleming. 1965. Cod growth and temperature in theNewfoundland area. ICNAF Special Publication 6: 545-555.

Mehl, S., and K. Sunnanä. 1991. Changes in growth ofNortheast Aretie cod in relation to food eonsumption in1984-1988. ICES mar. Sei. Symp. 193: 109-112.

MilIar, R. B., L. Fahrig, and P. A. Shelton. 1990. Effeet of eapelin biomass on eod growth. ICES C.M.1990/0:25. 10 p.

MilIar, R. B., and R. A. Myers. 1990. ModeHing environmentaHy induced change in size at age for AtianticCanada eod stocks. ICES C.M. 1990/0:24. 13 p.

MilIer, D.S. 1994. Results from an acoustic survey for eapelin (Mallotus villosus) in NAFO Divisions 2J3KL inthe autumn of 1993, p.91-98. In I. Carseadden [ed.] Capelin in SA2 + Div. 3KL. DFO Atiantic Fisheries

Research Document 94/18.

Sheiton, P. A., L. Fahrig, and R. B. MilIar. 1991. Uneertainity associated with cod-capelin interactions: how muchis too much? NAFO Sei. Coun. Studies. 16: 13-19.

Smith, S. I., and G. D. Somerton. 1981. STRAP: A user-oriented computer analysis system for groundfishresearch trawl survey data. Can. Tech. Rep. Fish. Aquat. Sci. 1030: 66 p.

Steinarsson, B. iE., and O. Stefansson. 1991. An attempt to explain cod growth variability. ICES C.M.1991/0:42.20 p.

Taggart, C. T., I. Anderson, C. Bishop, E. Colbourne, I. Hutchings, G. LiIly, J. Morgan, E. Murphy, R. Myers, G.Rose and P. Sheiton. 1994. Overview cf cod stocks, biology, and environment in the northwest Atianticregion ofNewfoundland, with emphasis on northern eod. ICES mar. Sci. Symp. 198: 140-157.

Warren, W.O. 1993. Some applieations of the Kaiman Filter in Fisheries Research. ICES C.M. 19931D:57, 19p.

WeHs, R. 1984. Orowth of cod in Divisions 21, 3K and 3L, 1971-1983. NAFO SCR Doc. 84/90, Serial No.N881. 6 p.

WeHs, R. 1986. Declines in the average length at age of eod in Divisions 21 and 3K during 1977-85. NAFO SCRDoe. 86/83, Serial No. N1205. 2 p.

14

Table 1. Number of fish sampled from RV surveys by NAFO division for which both length andage have been determined.

YEAR NAFO

FREQUENCYI2j 13k 131 TOTAL---------+--------+--------+--------+

78 1 453 I 457 1 0 1 910---------+--------+--------+--------+

79 I 460 1 510 I 0 1 970---------+--------+--------+--------+

80 I_ 600 1 706 I 0 1 1306---------+--------+--------+--------+

81 1 650 1 687 I 614 1 1951---------+--------+--------+--------+

82 1 844 1 747 I 882 I 2473---------+--------+--------+--------+

83 I 725 I- 800 I 797 I 2322---------+--------+--------+--------+

84 I 1055 1 844 I 0 I 1899---------+--------+--------+--------+

85 1 987 1 659 I 801 I 2447---------+--------+--------+--------+

86 1 569 1 670 I 699 I 1938---------+--------+--------+--------+

87 1 819 1 681 I 743 I 2243---------+--------+--------+--------+

88 1 852 1 1000 I 742 I 2594---------+--------+--------+--------+

89 1 891 1 1055 I 713 I 2659---------+--------+--------+--------+

90 1 853 1 971 I 706 1 2530---------+--------+--------+--------+

91 1 546 1 764 I 578 1 1888---------+--------+--------+--------+

92 I 263 I 540 1 494 I 1297---------+--------+--------+--------+

93 I 95 I 355 I 378 I 828---------+--------+--------+--------+

94 I 62 I 92 1 126 I 280---------+--------+--------+--------+TOTAL 10724 11538 8273 30535

•

•

•

•

15

Table 2. Number of fish sampled from RV surveys by NAFO division for which both weight andage have been determined.

YEAR NAFO

FREQUENCY\2j 13k 131 I TOTAL---------+--------+--------+--------+

78 I 132 I 120 I 0 I 252---------+--------+--------+--------+

79 I 113 I 119 I 0 I 232---------+--------+--------+--------+

80 I 140 I 156 I 0 1 296---------+--------+--------+--------+

81 I 145 I 141 I 138 I 424---------+--------+--------+--------+

82 I 135 I 160 I 51 I 346---------+--------+--------+--------+

83 1 173 I 156 I 152 I 481---------+--------+--------+--------+

84 I 532 I 167 I 0 I 699---------+--------+--------+--------+

85 I 506 I 143 I 147 I 796---------+--------+--------+--------+

86 I 119 I 130 I 142 I 391---------+--------+--------+--------+

87 I 104 I 13~ I 161 I 397---------+--------+--------+--------+

88 I 200 I 249 I 156 I 605---------+--------+--------+--------+

89 I 890 I 1055 I 142 I 2087---------+--------+--------+--------+

90' I 852 I 970 I 706 I 2528---------+--------+--------+--------+

91 I 546 I 764 I 576 I 1886---------+--------+--------+--------+

92 I 263 I 538 I 494 I 1295---------+--------+--------+--------+

93 I 95 I 355 I 377 I 827---------+--------+--------+--------+

94 I 62 I 92 I 126 I 280---------+--------+--------+--------+TOTAL 5007 5447 3368 13822

16

Table 3. Range and average annual number of fish' sampled for Iength and weight at age from RVsurveys by age, alI three divisions combined.

Lenoth al aoe samoies Weiohl al aoa samoiesDivision Aoe Years Mean N iMin N 'MaxN Years IMeanN MinN IMaxN2J 1 12 14.24 0 73 12 3.47 0 18

2 17 68.12 7 193 17 33.41 7 1921 3 171 88.47 15! 202 17 44.47 7 202

4 17 83.88 13 190 17 45.94 9 1875 17 84.82 81 265 17 42.06 4 1766 17 70.881 2\ 195 17 29.47 2 1217 17 61.35 11 133 17 25.12 1 1138 141 52.651 01 125 141 22.76 0 1259 14 41.29 0 113 14 16.06 0 7210 14 29.41 0 74 14 12.18 0 3711 14 15.12 0 36 14 7.29 0 2612 14 9.82 0 23 14 5.24 0 1113 '14 4.59 0 11 14 2.88 0 814 10 3.12 01 18 10 1.94 0 615 10! 1.181 01 5 10 1.00 0 416 9 0.71 01 3 9 0.47 0 217 6 0.59 01 3 6 0.47 0 218 4 0.35 01 3 4 0.18 0 119 1 0.06 01 1 1 0.06 0 120 1 0.06 01 1 1 0.00 0 021 1 0.06 0 1 1 0.06 0 122 1 0.061 0 1 1 0.00 0 0

3K I 0 2 0.291 0 4 2 0.24 0 4I 1 161 23.941 0 72 16 10.94 0 72I 2 17 87.82 101 211 17 45.18 5 2111 3 17 98.24 381 175 17 48.41 8 168

4 17 84.531 21 151 17 43.941 6 1515 17 80.18 11 164 17 37.06 1 1636 17 70.651 2 130 17 31.41 2 937 17 58.59 4 121 17 24.35 4 1098 17 55.71 1 121 17 22.88 0 1219 15 42.94 0 100 15 17.88 0 7910 141 30.591 01 77 14 14.06 0 5011 141 18.881 01 51 14 8.29 0 3012 14 11.181 01 30 14 5.47 0 1313 13 5.291 01 14 13 3.18 0 9

I 14 13' 4.531 0 14 13 3.00! 0 8I 15 131 2.291 0 13 13 1.47 0 9

16 101 1.531 0 5 10 1.24 0 417 5 0.591 0 3 5 0.531 0 318 • 5 0.29 0 1 5 0.24 0 119 2 0.12 01 1 2 0.12 0 120 3 0.18 01 1 3 0.18 0 121 3 0.24 01 2 3 0.24 0 222 1 0.12 01 2 1 0.12 0 2

I 23 11 0.061 01 1 1 0.061 0 13L I 0 51 2.711 01 21 5 0.071 0 1

I 1 111 12.141 01 46 11 2.361 0 13I 2 131 71.431 01 203 13 24.711 0 79

3 131 92.141 01 154 13 38.431 0 1174 131 81.211 0 148 13 38.57 0 1165 131 67.50 0 120 131 29.00 0 916 131 61.71 0 99 13 27.43 0 987 131 58.29 0 147 13 17.71 0 648 131 47.43 01 104 13 14.71 0 759 111 37.57 01 77 11 15.29 0 6610 121 19.57 01 53 12 8.50 0 3111 11 14.29 01 44 11 7.36 0 1912 11 10.21 01 28 11 5.57 0 16

13 10 6.361 01 21 10 4.14 0 121 14 9 3.861 01 13 9 2.57 0 10

15 91 2.291 01 6 9 2.14 0 6I 16 91 1.001 01 3 9 0.93 0 3I 17 61 0.501 01 2 6 0.43 0 2I 18 31 0.211 01 1 3 0.21 0 1I 20 21 0.291 01 3 2 0.291 0 3I 21 11 0.211 01 3 1 0.211 0 3I 22 11 0.071 01 1 1 0.071 0 1I 24 11 0.071 01 1 1 0.001 01 0: I I I I ! I I

Years = number of years with sampies. N = number of samoies Der vear. I I

I I I I I 1 I

I I

•

•

::: .: : rtEMIS CAP,:;: ~ ;,_:;::?:;

1------,r-.......--~.7';-'-t!': :,:,," :.••.•• ,J._. ••••••_•••••__.

oIl)

~-1-------1---_----:L...-+- -L.._+-__-"'_-.60 55 50 45 40

LONGITUDE

54'

52'

50'52'SB'SB'60'

50'

.,.....;, oall'-'3L' Co. 1''3' .._.0 ..

,..'"

4B'

,~,.,........f "'1 ..1, ~"\

~ ""Q~

I,4B'

54' 52' 50' 48' 48'60' SB'

-'-.1Fig. 1. Maps showing NAFO division boundaries, the loeation of Station 27 and the position ofthe three seetions. The Seal Island section erosses the Hamilton Bank and is referred to in the textas the Hamilton Bank seetion.

18

-D- AVETEMP

-+-- BOTANOM

1

1.5 ~

CIQ)

"C- 0.5Q)

1.......-+....

+t{~- +ca o 0....Q)

IQ.EQ)

-0.5 II- 0

I I I I-1 I I I I I I I I

co 0> 0 ..- C\J C') '<t LO co !' co 0> 0 ..- C\I C') '<t

l'-- l'-- co co co co co co co co co co 0> 0> 0> 0> 0)

Year

-(J

Fig. 2. Time series of Station 27 average annual water column temperature and the bottom (175m) temperature anomaly from the 1961 - 1990 mean.

50 -

45 1- 40 1E

35 1::.::I

Q) 30 I....ca

25 1-/+ .~

C'"l/) I +--r- 20VcaQ) 15...<

10 I5 I0

co 0> 0l'-- !' co

\C16+ +

+/'+

I I I I I I I I I I I

..- C\I C') '<t LO co !' co 0> 0 ..- C\I C') '<tco co co co co co co co co 0> 0> 0> 0> 0>

Year

--0- BONCIL

-+- FLEMCIL •

--D- HAMCIL

Fig. 3. Annual area of the cold intermediate layer (CIL) on the Hamilton Bank, Bonavista andFIemish Cap sections.

4.-------------------__--,3

c:J Capelin_ Otherprey 2J

94

3K

-3L

90 92888684828078

3r-

o78 80 82 84 86 88 90 92 94

4,........!.~---:~-.:::::..-~-.....:::.:::.....-......::;=---=-=----==--=-:.....,

2

>< 3Q)"0C

~ 2Q)

.f::iu..

•

e-e Canada

0··0 Russia

<\.....\

C!......

\o +--r---r-..-~O'~..~9_r__r__,__r_,_.,.__r_,.__...,:_'tq=:+_;

g 1500o'-'

~ ro n 00 ~ M M M 00 ~ ~

Year

2000 -r------~-----------__,

cncn(lj

E 1000o1).

.~Q).

0..: 500(lj()

­...

Fig. 4. Canadian and Russian fall capelin hydroacoustic estimates.

2r-

1 r-

o 1.-.1--1 ...L..-....I1_

78 80

r- -

82

-

LL84

_r-

86

Year'

88

r-, ..

r-

90 92

-

II~94

-\0,

Fig. 5. Partial fullness indices (PFI) for capelin in cod stomaehs in the fall RV survey, togetherwith the PFI for other prey, by division. .

20

450000 -

400000 -

350000I-UI

s::::300000 -;-0-- 250000 ~UI i

UI200000 ~C'll

E0 150000 :)m ,

100000i

,

50000I

0CX)

t--

""'+""'=-+0) 0 T""" C\I ('I) ~ 1.0 (0 "'" (Xl 0> 0 T""" C\I ('I) ~

"'" CX) CX) (Xl (Xl (Xl (Xl (Xl (Xl (Xl (Xl 0> 0) 0> 0> 0)

Year

--D-- RV2J

--0- RV3K

-+-- RV3L

•

Fig. 6. Annual biomass estimates by division from the fall RV trawl survey.

1200000 -

- 1000000UIs::::0- 800000 --UI IUIC'll 600000E I0 1m 400000+

M200000 -

0 I I

('I) -.:t 1.0 <0 t-- (Xl

"'" t-- "'" "'" t-- "'"

I I I I

0) 0 T""" N ('I) -.:t 1.0 (0 t-- CX) 0> 0 T""" C\I"'" (Xl (Xl (Xl CX) (Xl ro (Xl ro CX) ro 0> 0> 0)

Year

•

Fig. 7. ADAPT estimates of the 3+ beginning of year biomass in NAFO Divs. 2J3KL combined.

Div 2J cohort length at age90

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94

Year

Div 3K cohort ·length at age90

s~ 50

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94

Year

Div 3L cohort length at age90

80

70

60.........S~ 50

30

20

10

Year

Fig. 8. Cohort lengths. at age for each division for ag~s 10 and less. Lines orientated roughlyparallel to the the x-aXlS denote ages. Length at age 0 IS taken to be 0 so that the Iines orientatedroughly parallel to the y-axis denoting cohorts intercept the x-axis to give the year in which thecohort arose.

Div 2J cohort weight at age Div 3K cohort weight at age Div 3L cohort weight at age

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 9·o~~~z~z~z~~~nnwM~~M~M~MMOO~~~W

1000 I

3000

4000

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94

.....~

~

~ 0.0 0.0--' '-"" '-""

.... +.J +.J~

~ 2000 ~ 200050 2000~ ..... .....1)

~ ~;;.;;.

Year Year Year

Fig. 9. Cohort weights at age for each division for ages 8 and less.

NN

•

0.40.2

0-u -0.2Q)-- -0.4Q)

Q) -0.6.~- -0.8caQ)

-1a:-1.2-1.4-1.6

co 0) 0 T""" C\I C') ~ 10 CD f'- co 0) 0f'- f'- CX) CX) CX) ro CX) ro CX) CX) ro co 0)

Year

0.8 f0.6,

- 0.4 I0

O.~ tQ)--Q)

-0.2 TQ) I.~ -0.4 +- -0.6 tcaQ)

a: -0.8 -r

-1 I-1.2 J

I I I I I I I I I-1.4 .co 0) 0 T""" C\I C') oo::t 10 CD f'- CX) O'l 0 T""" C\I C') ~

f'- f'- CX) CX) ro CX) CX) CX) CX) ro CX) CX) O'l O'l O'l O'l O'l

Year

23

-D-- 2J Une

--0-- 3K Une

--+-- 3L Une

---0- 2J Wine

--0-- 3K Wine

--+-- 3L Wine

Fig. 10. Relative year effects in cohort length and weight increments in each division, aIl agescombined.

94929088

-0- 36-44cm-6-- 45-53 cm

- 54·62cm- ....- 63·71 cm

2J

3K

78 80 82 84 86 88 90 92 94

78 80 82 84 86

3L

-------~----/~~

0.85 --r------------ --,

0.75

0.70

0.80

0.75

0.70

o. 0.65 4---r---"T""---r--r-,r--r-T-.,.-J...,.-...,...-.--r--'r---r-1~r--r-io~

0.80

x_ 0.85 --r--------- ----,Ci).c-g> 0.80Q)

...J-.. 0.75-.cCl

'Q5 0.703:-g 0.65 -+--,--.--.--.-----r-r---.-.---,-,,---r-r-r--r-'-"T-".---t:::::::s~ 0.85 -,-------------------,

94929088

-0- 36-44cm-6-- 45-53 cm- 54-62cm- .... - 63-71 cm

78 80 82 84 86

.6

-------~------... lt•

1.0

0.8

0.9

0.8

-.cCl

'Q) 0.8~

"C -1--r-,--r-""1c----r----.--.----r---.....---.--.--.--.--,~__"T--IC 0.7:Jo~ 1.1 --r-----------------,

1.0

0.9

3L

1.1 .....-----------------,2J

\\\..

o O.7 +-~,---,-----r----r~=r=::;:::::r==;:..-.-_r_.--l_r--r-l~ 78 80 82 84 86 88 90 92 94;:1.1--r------------------,

Ci).c-g> 1.0Q)

--J_ 0.9

78 80 82 84 86 88 90 92 94

Year78 80 82 84 86 88 90 92 94

Year

Fig. 11. Average condition (round weight) of cod by division in ?.emlength groups from sampies taken in the fall RV surveys.. A eondltlon faetorof 0.9, which is sometimes assumed to be generally appheable to eod (eg.Brander 1995), is shown for referenee.

. Fig. 12. Average eondition (gutted weight, head on) of eod byjivision in 9 cm length groups from sampIes taken in the fall RVmrveys. A eondition faetor ofO.77, whieh is the overall averagereported by Taggart et al. (1994), is shown for referenee.

•

25

(Scatter Plot Matrix )Ip2JC I I. " ,:,,~ /": \ ,{ 1/' , ' .I .'\. V' • . \ V:",

I(~I~OO:>2.0 · . .1.0 ·., · .., . '\. ;, :.', i :"/· "-.. '. ''; \ .. :0.0 .. ·. , , '. .' - .. '

\... · .

3.0Ip3KC I / . """- ~ · "- /. "\. . . , .:\ 0\ ., ... .. ....~ - . ., , .

2.0 · . . .'. . · . . · . i\.. · . · .. '. ~ '.~. K":' K· · · · .

~ ../1.0 .~.· ":/ · .../

,. "~ ',' .' ~K~ './0.0 .' /

1.5 : \ !I Ip3LC I \ ·. \ '. '" '\ . .' : \

I~>.':\1(.)- ·. .. J.. · " ·. ·1.0 . . ..

!~.'. .

\, . I~.0.5 · " \ :' · 1"'- '. '.' ~ ... ·.

~'. . 1\ .. J .. ' \ ... .... ., " . .0.0 v,. ..........If:~ ......... 2JLINC eV .......V·"eV·,v· v· 2. .

I~· . ·. .' .:.> ., '\~

· . . .-0.5 ..... ,'. J ... I,' :.. • I • _',t' 'I .... .. ..' . · .', ," : .'. ..-1.0

.-/ I'- . ["-.... ~/ f'>... '..-/~-

F2 3KLINC :- --~ ~-

0.2 ,: :--....-. --@V • k, '. · ..~i" •• · .' & · .....( .. ' ..' ':' ~ · .' · . . "

~I • ~ .:. "

~ '. .:.... ~." ;.-0.2 1--:' • .-- '. . r--..:. •• · ..-0.8 -. r-.-.",,- -.- .-0.4 "\ '" \ .' .\ 3LLINC V I l/.' · '1/ !/.' .,11 .

.'. · ·. .' I

I· · '. ·. . : .':'/ · ..0.0

I\·.~·. . ·. · .' ,, .

/\ ·./'.

~ " ·· .) ':/ ..)-0.4 / \ J \. · ·· · ·'" 11;:·' 2JWINC (:

'1/. I/. .. 11..·. . .,' .\.. \ '. " -' . ·0.0 ·. , .'" 1\'.')' · .

. :~).,' . ·-0.5 · . " · . . . · ·. ·. · .

·1.0 . ' .. :.) .. / .... ) : ·· ./1'\ ,1"-.. · .. · .. ·. ,;

. .· . . .

1.0

I~~ v- k 3KWINC~ l;>----V ......

0.5", ") ..- .. ' .~16 ·u ' .. · '. · · · '.' '" .1K····.'- • _,I0.0 ·... .

I~'-'s • y .' .y . .. .'.~ • I · ., .

'./-0.5 I--. '---", I'-- i'-. .- I'- '..-/ r---.... • ./

0.4 V: .......... /(0' ~ : ...............1(:·': v. ·v / ."\ 3LWINC V .--"'" L/ ............v· '\0.2 . . .. ·'. ~ ) · · ..0.0 - 1\ . ·. · ... '.. · :

-0.2 . . :. .' .1\ ": J · · . · . ·· · y' · :. J .. J-0.4 ·. / ·. '". . . · ·

'/. ,/

~( : \ / .' \ / V · / .. \ · · \ 2JCOND ?;K

O:>1.05 '. ·. · . . . · . . . ·'.}.>./ .. .

~ . ~")~ . .: · · . · . . ! · '.

0.95 I\.\ . ·. . .' - ·0.85 , \"., / · '/ .' ;/ \ ... J . /1.05 V'.. li '. .. ~ ( .:. \

· . V" V .' I .' ' ... \

93KCOND .. \· . . · .1.00 .. · ') · ..

0.95 1\" k" \. ::) . :.,:-; ..../ '. " ;\.. :. ~.,: J ~ ... J0.90 :.j/ .' . '.. / · ... .. · . , .1.00

~.">I/' ."""-v· \~v· .~

Ir \l / .. \ '.~v. ~ 3LCOND

0.95 .- • ",1 • • si - . .1\\

I ". •~:.:'. ~.. ~ ..0.90 · . 1(· ." / "./ · , · .

f',........ • · ./"-...: .. : ./. . , ,

0.0 -0.5 0.5 -0.4 0.00.30.85 1.00 0.90 1.00 0.90 1.000.0 1.0 2.0 0.0 1.5 3.00.0 1.5 -1.0 0.0 -0.8 0.0 0.&0.4 0.0 0.4 -1.0

, ,\

Fig, 13. Scatter plot matrix of the partial fullness index of capelin in each division (P21C, P3KC,P3LC) against the relative annuallength increments for all ages combined (21 LINC, 3K LINC,3K LINC), relative annual weight increments for all ages combined (21 WINC, 3K WINC, 3LWINC) and round weight condition factor for cod in the 54 - 62 cm length dass (21 COND, 3KCOND, 3L COND),

26

Scatt.r Plot Matrix )

::1[§J <.~~.: M .<.-/-. r.,.:,. V; .:VS.i/7~· ~10::.. l(:. 1\)o 0 "/,01/ ..;/". '.' '.'-'. o' .... /. ,y' .... /I~ .... -I .. , I~\\

;::1/.. IRV3K I V .... 1/.... .~\ . ".\ V.' . .~ ,.1: .' "I;: : l( . I~' '.\1= .:".) .. :'J "~J . "·1' .\ ~':I' .... .'.' Ir ~':J .. . "':J250000 i/.. V .. . IRV3L I V 0 ..\ / /. ;'\ .. "'Ir v ." ..~.150000 ",.' / :.. .' ........, 1 '. '.' . ..50000 ~"/ • :' / .. • y; '.. .. /" .:' " .,"/ . \ .1=1//~:.: .'V:'.:': ~3KLB I.. ·::,V.·I/}/.::)V: ·1 /~.I~· ~400000 : / .' J. j ./. ./ .. / . / / I( . >J . \ :. ,

;:~r0~P:"j;. !r 3KLINC ~··.:::010S>0~'-:~~.:0.·0.8~ '.~ I'-~" (j .. [,;(, -----~ ._ I~0.4..: 1/ . " (.. 3L UNC ~ -: i/. • '1. / .... /. ... 11 •":)

.::: . .. )' j': .. J...j :. ;.. ' . .) :' '" / .'::J if .:.j ....: ;/. ..

.~:~ 1(·:' JV:'~ '.' V. '.:.. I/:~ , ·/.. 1/·" ~WINC f:: J '·1 f.:·..:1/'· .' ... :.\·1.0 '. / • '. /. "/. / •••• • ':.j . .y ...; .. ' 1(, ' ../ 1'\ .' ).

H'<:J0--:~k .(t .h~ ..-::,./:3K·WINC ~IC:~:'~·'I\).'0:5~/;:'./~"'... / (1 ('7'" .../ ···/I~···~~·: ,.'(.)"0.4 .' '\. '''' v '\'"\1/'/ ( 3LWINC 1(' '."17 I'.)0.2 \.. ..

~.~ .... ) . . ." .:. .... '.. "\ . '.' . ....•0:4 •••• • J • \" ~ • , • 'I\. ' ..

1.05 r: ::':..)V. . v. .. v:. . I.: ( :-:. V. '"V:- ..... ':,'1 . , .; ~COND I/;r., .:..)~::0YYU'" .1~··0vD·IQ~ 0'0.1.05 • \ ' .' v '. v ". 1(;" 3KCOND ')1.00 • • • 1/':' ,0.95 _ 'J' =. . ... ;/ ..) o' ••:": / '.' '/ .'.~. ...." 1\'" •0.90 : .': • : • '. /', , • .,?:, .... '.,/'. ·,,--/,·.I\.· r---..' / ,'\ '" R\~IQI~ 3LCOND~:~~ ~:.~~~ ~'l' ~.. v:-- ..... ~I,,('" I .:. ."0•• : ".090""'. \ ••••••••,.. '. • ••

. I "'-..... .' ./ 1'-.. .: ~ '. • . ' • .' ..... f'.-.:. :.-/ •'. I": .lDIOOOO rollHllllO 3OOOoom<xxi 2oo()(JC)JO(lOOI000000 ·1.0 0.0 -0 8 0.0 0.&0.4 0.0 0.4 ·1.0 0.0 ·0.5 0.5 -0.4 0.0 0.3085 1.00 0.90 1.00 0.90 1.00

Fig. 14. Scatter plot matrix ofbiomass estimates by division from the fall surveys (RV2J, RV3K,RV3L) and the ADAPT estimate of 3+ biomass for all divisions combined (2J3KL B) againstgrowth and condition data (same abbreviations as in Fig. 13).

•

\ .

27

(Scatter Plot Matrix )1.5

!AVET I.

D.

~-0.5 ~~ --: :.. ,~ V · '."",CJ) (!)~ @ I · , . "· .', · -,' L:--Y ~

~ . . · . ... .., ..--- '1'-0.5 -0.3

() IBOTAN 11(8 .n V 10~V V \):0.1 V .' . . ..

-0.1 .. ~'. /'. '..' · . ·. . . · . · .

-0.3 ~ . · ~ .. ' -/ .: / · . .. ':/ -./ '.' :/-0.5 . , . . " .. .' . :. " ;. .. ..,0.0 "B': 2JLrnC ()

V ..-........V·"() V·~/'.......V "'~-0.5 0' . . .. .' · . '. · . . .~."'" ~:'.. ' · _.~ ~ " . -, .:.;;; ·\ ~: " .. ' . • .·Z,'· :.'. ..

-1.0I'-... ~ r-- :...-----r--..:./I~

3KLINC -;---~. ~0.2 D (j) V · ......V, '. · -~@ '. .~'.:. '.' 69 .......... , . -..

':' ~· .... , .

~~-0.2 .... '. : " ~

· .I'- --0.8 -.

0.4 /. I:':) / " \ . '\ 3LLINC 1/ I '. l/. . 'V '. V·' .,11 .''. · . .:

.'.:/ , . · .0.0 .. · . ·. .' ·. / . ..)-0.4 I \ .... / \. • I · '/ \ :/ ··.1/.)Ir) /,,/ 1/·' 2JWrnC v. v: .. 1/·: . .: .\.. ' 11 :. . .0.0 -. . I'

-0.5 '. '. . ·. · . ·. · .-1.0 '(.. :. ../ .... ) : · f\ I..' .... .. .' /

.· ./ .1.0

10 k 3Kwrnc~ - ......... '-u' V V t::" . .0.5 ..- ~(J)' . .., .& '.,

0.0 :' ','. .., ." '.' .y · '. · ',- .. ' .'

~.'~... .' .' · ~\ .'.

-0.5.. 1'- . ('..... ~"- --- I'- i'-

-----0.4 /. \.

~:)1(:·': 1/. ·v . I 3Lwrnc V .'-.V "1/' '\0.2 . · . . .0.0 I ::'.1 ·. . · ..· . · · '.. · : . .

-0.2 I\: .... ) . · · . ~ · . ';-0.4 \0 .J . . I . y' · ·. · · -/ /.) I : \ .\ 1/ V J .. \ · \ 2JCOND ()1.05 . ..\'. · ., . . · , · '. \ ....: .' . . .... .;.: / . · J.0.95 .. . , " I ••• . .' · · ..

0.85 .. / ,. ., . J ' .. / ,/ .' ;/1\ '.' /1.05 .'

I~.),{ . \/ ..

/~ ··V 11 .' · ... \ / .. 3KCOND ".\1.00 . . r .:. · · ·.0.95 .

·. :: J. . ./.' .' . .. . :/ ·. /:I\"~' '.,: /. · . ~'.'0.90 .. . '\ ··.1 .. · . '. j. . .

1.00 .\

~~k~· V'. ~Ir :J / ..'\ ':~

v. "'" 3LCOND

0.95

:~"l· . . . .' . . .

1\\.'. · '.

0.90, . t •

~:.. ~..· · : /........ 1'---.... ·-0.5 0.5 UD.5 -0.1 0.3 -1.0 0.0 -0.8 0.00.&0.4 0.0 0.4 -1.0 0.0 -0.5 0.5 -0.4 0.00.30.85 1.00 0.90 1.00 0.90 1.00

Fig. 15. Scatter plot matrix of average annuaI water column temperature (AVET) and bottomtemperature anomaly (BOTAN) from Station 27 measurements against growth and condition data(same abbreviations as in Fig. 13).

1.000.950.90

•

28

Scatter Plot Matrix

45~ HAMCIL3525

~+----:::~~~~~--:;;,f-=----::;;i-~=:-i---:;;=:::-k=---+:-;---+-;;::.;:--b:=:::::---f;:::=::::--i:=::::::::--l-o::;:::::==:i I

40

:n

2:l 0;::~t=-~~::::"-h--.--~:::::::~~+-::~i-:;:::'~~+;:~~:::::+;::::::==+==~---110.0

-0.5·1.0

0.2-0.2

-O.8+---r-~+...".....-=--+-r-"""""~-,-~:"'-H--:-"""",""+---r::-""-+r-~~=-I--:"'--:::;::>--t--:T.'\--t~""",,---:-1b--;-,,,;+-"?"::---+-::7.""-""<j I0.4

0.0

-0.4 h-~-';'4-~--':+..-~-4~~..:J1~""":"-;,1-.l,~+-r--~~~4-.L;""-r~-=----'l:-t"'--~""';"'R"':-oC.-t--=--"'.-l10.0

-0.5·1.01.0J-...L.....:....j.~---..::.l.I-----':........:+-....:L-...:l.+-..:...L-+-L-:.L.-f-=--...L..--+---~~=!-=---'--+----:....4-.f.--.....;~-p---.,;..--i1

0.50.0

-0.50.4J-~-+----,~--.J---.,....-+------,,--+.~-:-I---:-~..J..-.....--;'-+----:---Ir-="=~f---,~-h-~...,...--:-"----I10.20.0

-0.2-O.4:1-~....:.,.:~~....,.....:L+--.;;~~W~,..:L./>.--~~J--.-\,..:..,/-+;,..---~:..",..-~....1.:.;...t.r--1---.:::---e~~+-=~~+--..,;:--L.j11.050.95

0.85J--+-~,.+---'-:~-l..~~';":'-h.L:';"'4~":"'-4f---JL.:.:M+-7-~L.,f--=7...--:4-LT!"\+--;~....L+-:T-:-tr~~:;:,;",t-~",~11.051.000.950.90

W-:""':"'~:"'-"'-A-~~H--,-<~~4-.J,.....!H"'7'""-'-.L......;+--~4-++~:::""-H~-+-""","",-+::~~1

15 25 35 452:1 3J 40 ro 25 3J 35 2:1 3J 40 -1.0 0.0 ·0.8 0.0 0.5-0.4 0.0 0.4 ·1.0 0.0 -0.5 0.5 -0.4 0.0 0.3 0.85 1.00 0.90 1.00 0.90 1.00

Fig. 16. Scatter plot matrix of average annual area of the cold intermediate layer on the HamiltonBank (HAMCIL), Bonavista (BONCIL) and Flemish Cap (FLEMCIL) as weIl as the annualaverage of all three (AVECIL) against growth and condition data (same abbreviations as in Fig.13).

• J

•

~ .

40 I~ 35c - 1~ E

~ ~ 30 I- G) 25 IE E.:.:: G)

~] 20 1<> :: 11---+---'-+--+1-+-1--+-1---'11--+--+1-+-1--+1-jl-+-I--+1-+-1--+-I-I(---j

C") ~ ~ CD r--. CX) m 0 T"" C\I c") ~ LO CD I"- CX) Cl 0I"- I"- r--. r--. r--. I"- I"- CX) CX) CX) CX) CX) CX) CX) CX) CX) CX) Cl

Cohor1

29

--0- 2JSLlNC

--0- 3KSLINC

--+-- 3LSLINC

• RUNCIL

1600

140

- ,~Cl 35- 1400 + /. "".- Ic IG)1200 +/X 30 --D-- 2JSWINC

E I -~ + /+"",G) "'" +-+'-u + 25 --0- 3KSWINCc 1000

I'C 3LSWINCc 800 t20

ca-E 600

I15 • RUNCIL

.:.::tr

400 1011l- I..J

U 200 5

I0

I 0LO CD r--. CX) Cl o T"" C\I c") ~ LO CD r--. CX) Cl

r--. I"- r--. l"- r--. CX) CX) CX) CX) CX) CX) CX) CX) CX) CX)

Cohor1

Fig. 17. The total increment in length and weight in each cohort from the beginning of age 2 to theend of age 5 plotted together with the average area of the eIL over that period (all three sectionscombined).