Not to be cited without prior reference to the authors

ICES CM 2004/CC:31

Interannual variation in life-cycle characteristics of the veined squid (Loligo forbesi)

G.J. Pierce, A.F. Zuur, J.M. Smith, M.B. Santos, N. Bailey & P.R. Boyle

The loliginid squid Loligo forbesi has a flexible life-cycle, involving variable size and age at maturity, presence of summer and winter breeding populations, and extended periods of breeding and recruitment. This paper reviews life history data collected since 1983 from the commercial fishery in Scottish (UK) waters and examines (a) the relationship between size and timing of maturation, (b) evidence for shifts in the relative abundance of the summer and winter breeding populations, and (c) the role of environmental signals in determining the timing of breeding. Evidence from fishery data suggests that, since the 1970s, the summer breeding population has declined while the winter breeding population now dominates and breeds later than was previously the case. Length-weight relationships and size at maturity showed significant inter-annual and seasonal variation during the period 1983-2001 and provide no evidence that there is currently a summer breeding population. Males are shown to decline in relative weight as they mature while females increase in relative weight. There is evidence that timing of breeding and size at maturity are related to environmental variation (winter NAO index).

Key words: life history, time series, environmental factors

G.J. Pierce, J.M. Smith, M.B. Santos, P.R. Boyle: University of Aberdeen, Tillydrone Avenue, Aberdeen AB24 2TZ, UK [tel: +44 1224 272866, fax: +44 1224 272396, e-mail: [email protected]]. A. Zuur: Highland Statistics Ltd, 6 Laverock Road, Newburgh, Aberdeenshire, AB41 6FN, UK [tel: +44 1358 788177, email: [email protected]]. N. Bailey: Fisheries Research Services, The Marine Laboratory, PO Box 101, Victoria Road, Aberdeen, AB11 9DB, UK [tel: +44 1224 876544, e-mail: [email protected]] INTRODUCTION

The veined squid Loligo forbesi has been exploited as a fishery resource around the British Isles for at least 100 years (see Thomas, 1969; Thomas & Davidson, 1972; Howard, 1979; Pierce et al., 1994a). Interest in its biology dates back at least to the 1970s, when Holme (1974) published a short description of the life history of the species in the English Channel. This work already provided some clues as to the complexity of the life-cycle for, although the species was apparently annual, two breeding populations were apparently present, a one breeding in winter and one breeding in summer.

In Scottish waters (Figure 1), two PhDs during the second half of the 1980s (Ngoile, 1987; Lum-Kong, 1989) provided further information on life-cycle biology and reproduction. The basic pattern that emerged was of an annual, winter breeding population. However, two peaks of recruitment were seen, in April and in autumn. Further indications of flexibility and/or complexity included evidence of two sizes at maturity (Boyle & Ngoile 1993a), recruitment throughout the year and the possible existence of morphologically distinct “stocks” (Boyle & Ngoile, 1993b) at Rockall and in coastal waters.

Further studies (e.g. during several EC-funded research projects) in Scottish waters between 1990 and 2001 focused on distribution and abundance (Pierce et al., 1998, 2001; Waluda & Pierce, 1998; Bellido et al., 2001; Bellido, 2002; Pierce & Boyle, 2003; Zuur & Pierce, 2004; Young et al., in press), stock structure (Pierce et al., 1994 b, c; Brierley et al., 1995; Shaw et al., 1999) and ecology (Pierce et al., 1994d; Collins & Pierce, 1996; Pierce & Santos, 1996; Daly et al,, 2001). However, several studies have also examined development and growth (Yau, 1994; Martins, 1997; Hughes, 1998; Pierce et al., 1999; Craig, 2001; Gowland, 2002) as well as reproduction and recruitment (Pierce et al., 1994e; Boyle et al., 1995; Collins et al., 1997, 1999; Shaw & Boyle, 1997; Challier et al., in press).

These studies essentially confirmed the picture established by Ngoile (1987) of an annual life- cycle with a winter breeding peak and two periods of recruitment. However, it is apparent that there are two or three micro-cohorts of Loligo forbesi present in Scottish waters, following different growth trajectories (Collins et al., 1999). It may be argued that some of the apparent variability in life history patterns reflects the persistence of summer and winter breeding populations in various proportions (Zuur & Pierce, 2004): the observed month-to-month pattern of maturation may be interpreted as two periods of recruitment and two periods of maturation. Shaw et al. (1997) confirmed that squid from offshore waters (around Rockall and Faroe) show some genetic differences from the coastal population.

Thus, superimposed on the basic annual life-cycle there may be (in addition to individual variation) variation due to presence of winter and summer breeders, inshore and offshore populations, large and small sizes at maturity,. Unlike relatively long-lived fish species, squid populations are not buffered against environmental change by the presence of several age classes: any factor which affects recruitment success has an immediate impact on population size and, potentially, subsequent reproductive success. There is considerable evidence that distribution and local abundance are affected by environmental variation (see above) and we might logically expect recruitment success, growth, fecundity and breeding success to reflect environmental conditions.

If the life-cycle of Loligo forbesi does indeed display substantial plasticity, we hypothesize that (a) changes in life-cycle characteristics would be evident over the medium- to long-term and that (b) such changes may occur in response to changing environmental conditions. We now have access to life history and fishery data on the species in Scottish waters collected over more than 15 years. In this paper we examine interannual variation in the relative importance of summer and winter breeding, length-weight relationships, size at maturity and the timing of maturity, and ask whether such variation can be related to environmental signals, notably the North Atlantic Oscillation Index – which provides a general indication of the strength of Atlantic inflow into coastal waters of Scotland.

To measure the relative importance of winter and summer breeding, abundance data are needed. Catch-per-unit effort data from commercial trawling may be used as an approximate relative abundance index (see Pierce et al., 1994a; Pierce & Boyle, 2003). It can also be argued that the peak of the fishery is driven by recruitment (see Figure 2, data from Pierce et al., 1994a, e) and is normally 3-4 months ahead of the peak in the proportion of mature animals. Thus by examining interannual trends in the month-to-month pattern of CPUE we may obtain an indication of interannual variation in the importance of winter and summer breeding. A formal analysis of interannual trends in monthly CPUE series was presented in Zuur & Pierce (2004). Here we briefly re-present the same data in terms of the likely interpretation in terms of the timing of breeding.

We go on to examine variation in two basic biological variables, weight at length and maturity at length. Weight at length is expected to follow a standard exponential curve. However, we might also expect variation in relation to the process of maturation. Thus, growing gonads may incur a different energetic cost to growing somatic tissue, mature animals may feed less actively, and gonad growth may involve utilisation of some somatic tissues as an energy source. Seasonal variation might also be expected, relating to changers in food availability, although it may be difficult to disentangle this effect from the effect of maturation. Lastly, we may expect year-to-year differences in growth patterns, related to differences water temperature (see Forsythe, 1993) and/or food supply.

Maturation is commonly assumed to be size-dependent, with the proportion of animals that are mature typically increasing with body length following a logistic curve. However, maturation may also require a seasonal trigger and/or depend on accumulation of sufficient energy reserves. Thus we might expect to see both seasonal and interannual variation. In the case of both weight and maturity at length, differences are expected between males and females. For example previous studies show that males can mature at smaller sizes than females and may mature slightly earlier in the year (Pierce et al., 1994e; Boyle et al., 1995).

If it is possible to tease out interannual differences in weight and maturity at length it might be hypothesised that these could be linked to environmental variation (e.g. larger weight and earlier maturation in warmer years). Another possibility is that there will be a relationship between size and abundance: positive if both parameters represent a response to environmental conditions, negative (at least at high abundance levels) if there is density- dependent competition. The present study aims to test these hypotheses. METHODS

(1) Winter and summer breeding

Commercial landings of squid (kg) and fishing effort (hours) data for UK-registered fishing vessels landing in Scotland during the period 1970-2000 were extracted from the Marine Laboratory FRS database. Since discarding of Loligo is thought to be minimal (Young et al., In Press), landings are assumed to accurately reflect catches. To obtain a reasonably consistent CPUE index we restricted the extraction to “light” and “heavy” single demersal trawls (henceforth referred to as “trawling”) from ICES areas IVa and VIa (henceforth referred to as “coastal waters”) and calculated an overall monthly CPUE index as total landings divided by total effort. Since 1998, Scottish fishing effort data are considered to be unreliable (A. Shanks, FRS Marine Laboratory, pers. comm.). However, as we will demonstrate, although there is evidence that CPUE in 1999 and 2000 was underestimated, the month-to-month pattern is driven by the landings data and examination of the latter confirmed that the underlying seasonal trend remains well-represented.

The overall month-to-month variation in a year was summarised by treating CPUE in each month as a variable, separately for areas IVa and VIa, detrending the data by dividing each value by the annual average (across both areas) and applying PCA.

(2) Length-weight relationships

Length-weight data were derived from monthly market samples collected between 1984 and 2001 under a series of projects. Normally a single box of mixed squid was obtained from the market at Kinlochbervie on the west coast of Scotland. This material was supplemented by market samples from east coast ports and by squid caught during research trawling surveys by FRV Scotia. For almost all squid sampled, dorsal mantle length (DML), body weight, sex and maturity stage (see Boyle & Ngoile, 1993; Pierce et al., 1994) were recorded. Samples were additionally classified by month, year and area (east coast or west coast, divided at 4o W, and Rockall). Monthly data were available for December 1983 to April 1986, July 1989 to September 1995 and March 1996 to December 2001. No weight data were available for squid collected during 1987-88.

The variables used were: body weight, mantle length (both log-transformed), sex, maturity, region, month and year. Data were analysed separately for males and females. Maturity stage was classified following a 5-point scale (Boyle & Ngoile, 1993a; Pierce et al., 1994e). This is a relatively simple scale but was used by many different workers over the course of the date collection period so that consistency of use cannot be 100% assured. Nevertheless, the distinction between immature (stages I – III) and mature (stages IV – V) is quite unambiguous, depending on presence of eggs in the oviducts or spermatophores in the Needham’s sac. On occasion animals were classified as falling between two of the five stages.

Data were available for three regions: Rockall (an offshore bank 200 miles west of Scotland), the West coast of Scotland and the East coast of Scotland. Relatively few data were available for Rockall and these were therefore excluded from the analysis. For pre-1989 samples, the region of origin could not be identified and the analysis was therefore carried out in two stages, firstly for 1989-2001 and secondly for 1982-2001.

Scatter plots of log-transformed weight and length indicated a linear relationship and linear regression was therefore applied, with weight as the dependent variables, length and stage as linear explanatory variables, and year and month as factors. Interaction terms were also included.

The interaction term for month can be thought of as the effect of month on the slope of the relationship with length. Since slope and intercept for the “month” effect tended to be collinear, length data were centred at zero and the analysis repeated. All analysis was carried out using Brodgar software (Highland Statistics Ltd).

(3) Size at maturity

Data on maturity (0 or 1), year and month were extracted for both sexes. As above, for 1989- 2001, Rockall data were not used. The response variable (maturity) is binary. Preliminary application of a binomial GAM indicated that the relationship between maturity and length was non-linear. Hence, GAM is preferred to GLM. Because GAM can cope with non-linear relationships, it was decided to use untransformed length in the models. The Akaike Information Criterion (AIC) was used in conjunction with a forward and backward selection procedure to identify the optimal model, i.e. whether factors or smoothers should be used for year and month, the appropriate number of degrees of freedom, and whether any of the components can be dropped from the model.

One possible interannual difference would be in the timing of maturity. This was investigated by repeating the analysis for animals from December only.

(4) Relationships with abundance and environmental conditions

An annual trawl survey abundance index (squid catch per hour) was derived from survey data (FRS, unpublished data). Values of the winter NAO index were downloaded from the University of east Anglia Climate Research Centre website. Correlations were calculated between these indices and annual coefficients for effects of year on body weight and maturity at length. We also extracted data on the proportion of mature squid in December and June to see whether these figures were correlated with abundance or NAO.

RESULTS

(1) Winter and summer breeding

PCA results (Figure 3) suggest a shift in monthly abundance patterns since 1990. These patterns are illustrated in chronological sequence in Figure 4. In the early 1970s, the fishery generally peaked around October and March-April, with an additional February peak in 1973. In the late 1970s, the spring peak shifted to February-March and the October peak became dominant. The dominance of the October peak extended through the 1980s but with a further strong peak evident in June-July in several years. In the early 1990s the October peak became broader, extending through to January. By the late 1990s the timing of the peak had shifted to December or January and a secondary peak in June and July was again evident.

(2) Length-weight relationships

Females

Regression analysis for females in 1989-2001 demonstrated significant effects of length (t = 231.46, P < 0.0001), maturity stage (t = 11.08, P < 0.0001), region (t = -3.37, P = 0.0008), year, month, and the interactions of length with maturity stage (t = 6.06, P < 0.0001), month and year. Weights at length were generally higher in more mature females, and higher on the east coast (region 1) than the west coast (region 2). There was no significant region-length interaction.

Coefficients for effects of year, month and the year-length and month-length interactions are illustrated in Figure 5. Females are generally heavier at length in January and in May to August than in other months. The interaction between effects of month and length on weight was generally more negative earlier in the year, i.e. the weight-length relationship has a shallower slope in the earlier months of the year. A year effect is also evident, with highest weights being recorded in 1997 and the lowest in 1989. There was also a year-length interaction, with the most pronounced effect relating to 1989 –although this may simply reflect the fact that data collection began mid-1989 so there are no observations for the earlier months of the year.

Using the full 1983-2001 data set (no data for 1987-88), very similar results were obtained. The seasonal pattern is virtually unchanged. Both main and interaction effects for year during 1983-86 are stronger than seen for 1989-2001 (Figure 6), possibly because there were more months with missing data (e.g. only December data were available for 1983).

Males

Regression analysis for males in 1989-2001 demonstrated significant effects of length (t = 291.19, P < 0.0001), maturity stage (t = -24.59, P < 0.0001), region (t = -4.04, P = 0.0001), year, month, and the interactions of length with maturity stage (t = -11.51, P < 0.0001), month and year. Weights at length were generally lower in more mature males, and higher on the east coast (region 1) than the west coast (region 2). There was no significant region-length interaction.

Coefficients for effects of year, month and the year-length and month-length interactions are illustrated in Figure 7. Males are generally heaviest at length in June and lightest in February. The interaction between effects of month and length on weight was generally negative earlier in the year (i.e. the weight-length relationship has a shallower slope in the earliest months of the year) and the highest coefficient was recorded for May. These patterns are similar but not identical to those seen in females. The main year effect was also similar to that seen for females, with highest weights being recorded in 1997 and the lowest in 1989, and the year- length interaction was most pronounced for 1989.

Again, using the full 1983-2001 data set reveals similar patterns, with strong year effects for 1983-1986 (Figure 8).

(4) Size at maturity

Females

The optimal model had the form:

Maturity ~ factor(Year) + factor(Month) + s(Length, df = 11) ; AIC= 5071.0

Thus, Year and Month are fitted as factors, and 11 degrees of freedom are required for the relationship of maturity with length. It as necessary to remove data from the first year (1983) because the corresponding confidence interval was very large (there were not many observations for this year). The smoother for length (DML, cm) was highly significant (χ2 = 397.3, P<0.0001). There was no overdispersion.

The effects of length, month and year on maturity are shown in Figure 9. It may be seen that peak maturity (at length) is in March and there are fewest mature animals (at length) in September.

Males

The optimal model for males was of rather similar structure:

Maturity ~ factor(Year) + factor(Month) + s(Length, df = 10) ; AIC= 8958.9

Again, data for 1983 were removed. The smoother for length (DML, cm) was highly significant (χ2 = 500.9, P<0.0001). There was no overdispersion.

The effects of length, month and year on maturity are shown in Figure 10. It can be seen that there is evidence of two peaks of maturity (as previously reported) in the partial plot for length, the first peak being at around 180 mm mantle length. It may be seen that peak maturity (at length) is in March and there are fewest mature animals (at length) in August, suggesting slightly earlier maturation in males, as previously reported.

(5) Size at maturity in the breeding season

The size-at maturity analysis was repeated using data for December only. As above, data for 1983 were removed since they gave a poor fit to the model.

The best-fits GAMs (see Figure 11), for females and males respectively, had the form:

Mature ~ s(Year, 10) + s(Length, df = 3); AIC= 974.877

Mature ~ s(Year, 10) + s(DML, df = 6); AIC= 1070.537

The effect of length in males was more complex than in females and the partial plots clearly indicate two sizes at maturity in the males.

(6) Interannual variation

There is no evidence that interannual variation in weight at length is related to either winter NAO or squid abundance: all correlations were non-significant.

The simple proportion of males that were mature in the December sample, an index of the timing of maturation was significantly natively correlated with the winter NAO (R = -0.525, P = 0.037) and the equivalent correlation for females was not quite significant (R = - 0.487, P = 0.056). Note that the winter NAO for a given year refers to the winter preceding maturation. Correlations with current conditions (i.e. the subsequent NAO value) were also negative but non-significant. There were non-significant positive correlation between the winter NAO and the proportion mature for both males (R = 0.567, P = 0.055) and females (R=0.583, P=0.100) in the samples in June.

The interannual component of variation in maturity at length (having removed seasonal variation) was related to winter NAO for the following calendar year, i.e. conditions at the end of the current calendar year (R=0.640, P=0.003 for females; R=0.485, P=0.035 for males). These correlations were however not evident if maturity data from December only were used.

DISCUSSION

Although fishery data provide only circumstantial evidence on the life-cycle, the shifting seasonal peaks of the fishery are difficult to reconcile with a fixed annual life-cycle. The peaks in fishery landings suggest the presence of at least two seasonal population components that vary in absolute and relative importance. However, even allowing for this, it is evident that the timing of the life-cycle shows subtle shifts from year to year.

Weight at length

The analysis of weight at length data confirms that there are significant interannual changes. However, perhaps more interesting are the effects of month and maturity stage. In both sexes there appears to be a peak of weight at length in summer, as might be expected if this is the period of highest food availability. In females, there was also a peak in weight at length in January. It is tempting to associate the latter peak with maturity, but in fact the maturity effect has already been removed.

Extending the analysis back to 1983, by incorporating data from Ngoile (1987), there is evidence of wider fluctuations in weight-at-length. Although it may be noted that there was a shift in the seasonal pattern of fishery abundance in the mid-1980s (see Figure 4), and the differences in the length-weight relationship between 1983-86 and 1989 onwards could reflect the presence of different breeding stocks, weight-at-length data cannot be said to provide any strong evidence about the existence of more than one breeding population.

With increasing maturity stage, males become lighter and females get heavier (at length). The latter trend presumably represents the consequence of ovary growth. While males probably invest less energy in gonad growth than females, the negative trend in body weight at length requires further explanation. It is not clear whether somatic tissue is metabolised to fuel gonad growth or if food intake declines in mature males. By analogy with Loligo vulgaris, large males may guard females on the spawning grounds but such behaviour is not expected to occur in advance of spawning. On the other hand, many females carry sperm in the buccal pouch before reaching full maturity (Pierce et al., 1994e; Boyle et al., 1995) so males may incur courtship costs.

Hatfield & Rodhouse (1992) argued that, in the ommastrephid squid Illex argentinus, energy reserves for maturation derived from food rather than stored reserves. The present results for Loligo forbesi are consistent with mobilisation of reserves in males. However, this assumes that animals tend to follow the same growth trajectory whereas it remains possible that animals maturing at larger sizes are also heavier at a given size. In other words, the presence of several microcohorts with different growth patterns could theoretically generate the observed trends. This could theoretically arise from hatchlings of different sizes (see Pecl et al., 2004) and/or from animals experiencing different temperature regimes during their early life, as proposed by Forsythe (1993). Collins et al. (1999) described the existence of at least two microcohorts in female Loligo forbesi and three in males on the west coast of Scotland. Further analysis of somatic and reproductive growth, and feeding rates, using measurements on gonads, mantle tissue and digestive gland may help to resolve this question.

Maturity at length

The analysis of maturity at length effectively investigates variation in size at maturity, while avoiding the need to reduce this to a single parameter. In fact, maturity at length is best represented by a non-parametric smoother rather than a parametric function (such as the traditionally applied logistic curve). The contrasting shapes of these curves for males and females are consistent with the existence of more than one size at maturity in males (see Pierce et al., 2004a) – and this phenomenon was more obvious when dealing with data for December alone. Raya et al. (1999) confirmed that there were size and seasonal components to maturation in L. vulgaris. The present analysis for L. forbesi confirms that there is a distinct seasonal effect on maturity, with the probability of an animal of a given size being mature being much higher in January to March than in August to October. These data clearly suggest the absence of a significant summer breeding component to the population.

Environmental effects

Relationships were found between size at maturity and the winter NAO index, and between the proportion of mature animals in December (the start of the main breeding season) and NAO, thus indicating that both size at maturity and timing of breeding are to some extent under environmental control. Results suggested that size at maturity was positively related to winter NAO, i.e. broadly speaking animals were bigger in warmer years, but that the proportion of mature animals present in December was negatively related to NAO. Sims et al. (2001) showed that the timing of migration in Loligo forbesi in the English Channel could be related to the NAO index, with earlier migration in warmer years.

There was also evidence of a regional effect on weight at length (West coast v East coast). Neither weight nor maturity at length appeared to vary with an index of stock size. Thus, hypothesised effects of intraspecific competition for resources were absent, perhaps because Loligo forbesi rarely reaches very high abundance levels (Young et al., In Press).

ACKNOWLEDGEMENTS

The authors would like to thank everyone who assisted with sampling squid, including: Luise Allcock, Mark Ashmore, Pablo Avila, Antony Bishop, Laurence Challier, Sam Cho, Ischbel Clark, Martin Collins, Sandra Cordes, Heather Daly, Samuel Desormonts, Jamie Dyson, Jacqueline Eggleton, Maxine Grisley, Gordon Hastie, Lee Hastie, Jasmine Haggquist, Suzanne Henderson, Marie-Pierre Holm, Steve Hoskin, Shirley Hughes, Magnus Johnson, Nicola Johnson, John Joy, Eoghan Kelly, Katherine Kelly, Linda Key, Adrian Linnane, Colm Lordan, Amoy Lum-Kong, Shelagh Malham, Clara Martins, Joanne Murphy, Bess Mucklow, Fiona Murray, Mariane Nyegaard, Gill Robertson, Séverine Sachy, Gabi Stowasser, Antoine Vigneau, Jianjun Wang, Cynthia Yau, Iain Young and Xiaohong Zheng.

Data collection during 1990-2001 was financed by the European Commission under projects FAR MA 1.146 (Fishery Potential of Northeast Atlantic Squid Stocks), AIR1-CT92-0573 (Stock Dynamics, Interactions And Recruitment In Northeast Atlantic Squid Fisheries), FAIR CT 1520 (Cephalopod Resource Dynamics: Patterns in Environmental and Genetic Variation), Study Project 96/081 (Data Collection for Assessment of Fished Cephalopod Stocks), Study project 97/107 (Development of software to estimate unreported or misreported catch and effort data and to apply fishery management models) and Study Project 99/063 (Data collection for assessment of cephalopod fisheries).

REFERENCES

Bellido, J.M., 2002. Use of geographic information systems, spatial and environment-based models to study ecology and fishery of the veined squid (Loligo forbesi Steenstrup 1856) in Scottish waters. PhD. University of Aberdeen.

Bellido, J.M., Pierce, G.J. & Wang, J., 2001. Modelling intra-annual variation in abundance of squid Loligo forbesi in Scottish waters using generalised additive models. Fisheries Research 52, 23-39.

Boyle, P.R. & Ngoile, M.A.K. 1993a. Assessment of maturity state and seasonality of reproduction in Loligo forbesi (Cephalopoda: Loliginidae) from Scottish waters. Pages 37-48 in T. Okutani, R. K. O'Dor, and T. Kubodera, editors. Recent Advances in Cephalopod Fisheries Biology. Tokai University Press, Tokyo.

Boyle, P.R. & Ngoile, M.A.K. 1993b. Population variation and growth in Loligo forbesi (Cephalopoda: Loliginidae) from Scottish waters. Pages 49-60 in T. Okutani, R. K. O'Dor, and T. Kubodera, editors. Recent Advances in Cephalopod Fisheries Biology. Tokai University Press, Tokyo.

Boyle, P.R., Pierce, G.J. & Hastie, L.C., 1995. Flexible reproductive strategies in the squid Loligo forbesi. Marine Biology 121, 501-508.

Brierley, A.S., Thorpe, J.P., Pierce, G.J., Clarke, M.R. & Boyle, P.R., 1995. Genetic variation in the neritic squid Loligo forbesi (Myopsida: Loliginidae) in the northeast Atlantic Ocean. Marine Biology 122, 79-86.

Challier L., Pierce G.J. & Robin J.P., In Press. Early growth variation and relationships with recruitment in English Channel and Scottish waters Loligo forbesi. Marine Ecology Progress Series.

Collins, M.A., Boyle, P.R., Pierce, G.J., Key, L.N., Hughes, S.E. & Murphy, J., 1999. Resolution of multiple cohorts in the Loligo forbesi population from the west of Scotland. ICES Journal of Marine Science 56, 500-509.

Collins, M.A. & Pierce, G.J., 1996. Size selectivity in the diet of Loligo forbesi (Cephalopoda: Loliginidae). J. mar. biol. Ass. U.K. 76, 1081-1090.

Collins, M.A., Pierce, G.J. & Boyle, P.R., 1997. Population indices of reproduction and recruitment in Loligo forbesi (Cephalopoda: Loliginidae) in Scottish and Irish waters. Journal of Applied Ecology 34, 778-786.

Craig, S., 2001. Environmental conditions and yolk biochemistry: factors influencing embryonic development in the squid Loligo forbesi (Cephalopoda: Loliginidae) Steenstrup 1856. PhD. University of Aberdeen.

Daly, H.I., Pierce, G.J., Santos, M.B., Royer, J., Cho, S.K., Stowasser, G., Robin, J.P. & Henderson, S.M., 2001. Cephalopod consumption by trawl caught fish in Scottish and English Channel waters. Fisheries Research 52, 51-64.

Forsythe, J.W., 1993. A working hypothesis of how seasonal temperature change may impact the field growth of young cephalopods. In: Okutani, T., O'Dor, R., Kubodera, T. (Eds.), Recent Advances in Cephalopod Fisheries Biology. Tokai University Press, Tokyo, pp. 133- 143.

Gowland, F.C., 2002. Developmental variability in loliginid squid (Loligo forbesi and Sepioteuthis australis). PhD. University of Aberdeen.

Hatfield, E.M.C. & rodhouse, P.G., 1992. production of soma and gonad in maturing female Illex argentinus (Mollusca: Cephalopoda). Journal of the Marine Biological Association of the United Kingdom 72, 281-291.

Holme, N.A., 1974. The biology of Loligo forbesi Steenstrup (Mollusca: Cephalopoda) in the Plymouth area. Journal of the Marine Biological Association of the United Kingdom 54, 481- 503.

Howard, F.G. 1979. Recent trends in the Scottish fishery for Loligo forbesi together with some notes on the biology of the species. CM 1979/K:36, ICES.

Hughes, S.E., 1998. Growth history patterns in squid as assessed by gladius structure. PhD. University of Aberdeen.

Lum-Kong, A. 1989. The reproductive biology of female Loligo forbesi Steenstrup (Cephalopoda: Myopsida). PhD. University of Aberdeen.

Lum-Kong, A., Pierce, G.J. & Yau, C., 1992. Timing of spawning and recruitment in Loligo forbesi Steenstrup (Cephalopoda: Loliginidae) in Scottish waters. Journal of the Marine Biological Association of the United Kingdom 72, 301-311.

Martins, M.C.P.R., 1997. Biology of pre- and post-hatching stages of Loligo vulgaris Lamarck, 1798 and Loligo forbesi Steenstrup, 1856 (Mollusca, Cephalopoda). PhD. University of Aberdeen.

Ngoile, M.A.K. 1987. Fishery biology of the squid Loligo forbesi Steenstrup (Cephalopoda: Loliginidae) in Scottish waters . PhD thesis. University of Aberdeen.

Pecl, G.T., Steer, M.A. & Hodgson, K.E., 2004. The role of hatchling size in generating the intrinsic size-at- age variability of cephalopods: extending the Forsythe Hypothesis. Marine and Freshwater Research 55, 387-394.

Pierce, G.J., Boyle, P.R., Hastie, L.C. & Shanks, A.M., 1994a. Distribution and abundance of the fished population of Loligo forbesi in UK waters: analysis of fishery data. Fisheries Research 21, 193-216.

Pierce, G.J., Bailey, N., Stratoudakis, Y. & Newton, A., 1998. Distribution and abundance of the fished population of Loligo forbesi in Scottish waters: analysis of research cruise data. ICES Journal of Marine Science 55, 14-33.

Pierce, G.J. & Boyle, P.R., 2003. Empirical modelling of interannual trends in abundance of squid (Loligo forbesi) in Scottish waters. Fisheries Research 59, 305-326.

Pierce, G.J., Boyle, P.R., Hastie, L.C. & Key, L., 1994e. The life history of Loligo forbesi (Cephalopoda: Loliginidae) in Scottish waters. Fisheries Research 21, 17-41.

Pierce, G.J., Boyle, P.R., Hastie, L.C. & Santos, M.B., 1994d. Diets of squid Loligo forbesi and Loligo vulgaris in the northeast Atlantic. Fisheries Research 21, 149-164.

Pierce, G.J., Hastie, L.C., Guerra, A., Thorpe, R.S., Howard, F.G. & Boyle, P.R., 1994b. Morphometric variation in Loligo forbesi and Loligo vulgaris : regional, seasonal, sex, maturity and worker differences. Fisheries Research 21, 127-148.

Pierce, G.J. & Santos, M.B. 1996. Trophic interactions of squid Loligo forbesi in Scottish waters. Pages 58-64 in S. P. R. Greenstreet and M. L. Tasker, editors. Aquatic predators and their prey. Fishing News Books.

Pierce, G.J., Thorpe, R.S., Hastie, L.C., Brierley, A.S., Guerra, A., Boyle, P.R., Jamieson, R. & Avila, P., 1994c. Geographic variation in Loligo forbesi in the Northeast Atlantic Ocean: analysis of morphometric data and tests of causal hypotheses. Marine Biology 119, 541-547.

Pierce, G.J., Wang, J., Zheng, X., Bellido, J.M., Boyle, P.R., Denis, V. & Robin, J.-P., 2001. A cephalopod fishery GIS for the Northeast Atlantic: development and application. International Journal of Geographical Information Science 15, 763-784.

Raya, C.P., Balguerías, E., Fernández-Núñez, M.M. & Pierce, G.J., 1999. On reproduction and age of the squid Loligo vulgaris from the Saharan Bank (north-west African coast). Journal of the Marine Biological Association of the United Kingdom 79, 111-120.

Shaw, P.W. & Boyle, P.R., 1997. Multiple paternity within the brood of single females of Loligo forbesi (Cephalopoda: Loliginidae), demonstrated with microsatellite DNA markers. Marine Ecology-Progress Series 160, 279-282.

Shaw, P.W., Pierce, G.J. & Boyle, P.R., 1999. Subtle population structuring within a highly vagile marine invertebrate, the veined squid Loligo forbesi, demonstrated with microsatellite DNA markers. Molecular Ecology 8, 407-417.

Sims D.W., Genner, M.J., Southwa,rd A.J. & Hawkins, S.J., 2001. Timing of squid migration reflects North Atlantic climate variability. Proceedings of the Royal Society of London B 268, 2607-2611.

Thomas, H.J. 1969. Some observations of the exploitation and distribution of squid around Scotland. ICES C.M 1969/K:29, Marine Laboratory, Aberdeen, Scotland.

Thomas, H.J. & Davidson, C. 1972. Further observation on Scottish squid fishery. ICES C.M. K:31, Marine Laboratory, Victoria Road, Aberdeen, Scotland.

Waluda, C.M. & Pierce, G.J., 1998. Temporal and spatial patterns in the distribution of squid Loligo spp. in United Kingdom waters. South African Journal of Marine Science 20, 323-336.

Yau, C. 1994. The ecology and ontogeny of cephalopod juveniles in Scottish waters. Ph.D. University of Aberdeen.

Young, I.A.G., Pierce, G.J., Daly, H.I., Santos, M.B., Key, L.N., Bailey, N., Robin, J.-P., Bishop, A.J., Stowasser, G., Nyegaard, M., Cho, S.K., Rasero, M. & Pereira, J.M.F., In Press. Application of depletion methods to estimate stock size in the squid Loligo forbesi in Scottish waters (UK). Fisheries Research.

Zuur, A.F. & Pierce, G.J., 2004. Common trends in Northeast Atlantic squid time series. Netherlands Journal of Sea Research 52, 57-72.

Figure 1. Map showing the main fishery areas from which squid were sampled. The division into “West coast” and East coast” is at 4o West.

Figure 2. Relationship between maturity patterns (monthly proportion mature for females) and fishery CPUE in Scottish coastal waters, July 1990 - June 1992 (monthly catch per unit effort by trawlers in Scottish coastal waters). Data were taken from Pierce et al., 1994 a, e).

1.0 4 CPUE Female PM 0.8 3

0.6

2

0.4 Proportion mature

1 Coastal fishery CPUE (kg/h) 0.2

0.0 0 JASONDJFMAMJJASONDJFMAMJ Month

Figure 3. PCA plot (axis 2 scores versus axis 1 scores). The original variables were de-trended CPUE data for each month in areas IVa and VIa (N = 24 variables).

5 Trajectory 4 1970s 1980s 3 1990-95 1996-2000 2

1

0

Axis 2 score -1

-2

-3

-4 -5 -4 -3 -2 -1 0 1 2 3 4 5 Axis 1 score

Figure 4. Plots of monthly average trawling CPUE for squid in Scottish coastal waters: (a) 1970- 1974; (b) 1975-1979; (c) 1980-1984; (d) 1985-1989; (e) 1990-1994; (f) 1995-2000. Plots (a) and (f) also show monthly landings (second y axis), demonstrating that landings and CPUE tend to follow similar trends. Vertical lines coincide with January and July in each year.

4.0 150 1.2 CPUE 1970-74 75-79 Landings 1.0

3.0 100

0.8 g/h) g/h) onnes) t k 2.0 50 0.6 ngs ( CPUE (k CPUE ( 0.4 Landi

1.0 0 0.2

0.0 -50 0.0 JAJOJAJOJAJOJAJOJAJO JAJOJAJOJAJOJAJOJAJO Month Month a b

2.5 3.5

80-84 3.0 85-89 2.0

2.5 h)

h) 1.5 2.0 g/ g/ k k

1.5

1.0 CPUE ( CPUE (

1.0

0.5 0.5

0.0 0.0 JAJOJAJOJAJOJAJOJAJO JAJOJAJOJAJOJAJOJAJO Month Month c d

4.0 4.5 150 95-00 Landings 3.5 4.0

3.5 3.0 100

3.0 2.5 90-94 h) h) g/ g/ 2.5 onnes) t k k 2.0 50

2.0 ngs (

CPUE ( 1.5 CPUE ( 1.5 Landi

1.0 0 1.0

0.5 0.5

0.0 0.0 -50 JAJOJAJOJAJOJAJOJAJO JAJOJAJOJAJOJAJOJAJOJAJO Month Month e f

Figure 5. Coefficients (±1 standard error) for nominal variables affecting length-weight relationships in female Loligo forbesi, 1989- 2001: (a) Month (main effect), (b) Month (interaction with length), (c) Year (main effect), (d) Year (interaction with length) a b

0.007 0.02

0.006 0.01

0.005 0.00

0.004 -0.01 ent ent 0.003 ci ci -0.02 0.002

Coeffi Coeffi -0.03 0.001

-0.04 0.000

-0.001 -0.05

-0.002 -0.06 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Month

c d

0.006 0.090

0.004 0.070

0.002 0.050

0.000 0.030 ent ent ci ci

-0.002 0.010 Coeffi Coeffi

-0.004 -0.010

-0.006 -0.030

-0.008 -0.050 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Year Year

Figure 6. Coefficients (±1 standard error) for nominal variables affecting length-weight relationships in female Loligo forbesi, 1983- 2001: (a) Year (main effect), (b) Year (interaction with length) a b

0.030 0.12

0.025 0.10 0.020 0.08 0.015 ent ent 0.010 0.06 ci ci

0.005 0.04 Coeffi Coeffi 0.000 0.02 -0.005 0.00 -0.010

-0.015 -0.02

5 7 2 4 9 1 8 9 0 0 90 97 01 983 991 992 993 1 1984 198 1986 198 1988 1989 19 1991 199 1993 199 1995 1996 19 1998 199 2000 200 1983 1984 1985 1986 1987 198 198 199 1 1 1 1994 1995 1996 1997 1998 1999 200 20 Year Year

Figure 7. Coefficients (±1 standard error) for nominal variables affecting length-weight relationships in male Loligo forbesi, 1989- 2001: (a) Month (main effect), (b) Month (interaction with length), (c) Year (main effect), (d) Year (interaction with length) a b

0.008 0.03

0.02 0.006

0.01 0.004 0.00

0.002 ent ent -0.01 ci ci

0.000 -0.02 Coeffi Coeffi -0.03 -0.002 -0.04

-0.004 -0.05

-0.006 -0.06 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Month c d

0.006 0.08

0.004 0.06

0.002 0.04

0.000 ent ent

ci ci 0.02 -0.002 Coeffi Coeffi 0.00 -0.004

-0.02 -0.006

-0.008 -0.04 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Year Year

Figure 8. Coefficients (±1 standard error) for nominal variables affecting length-weight relationships in male Loligo forbesi, 1983- 2001: (a) Year (main effect), (b) Year (interaction with length) a b

0.03 0.10

0.08 0.02

0.06

0.01 ent ent 0.04 ci ci

0.02 0.00 Coeffi Coeffi

0.00

-0.01 -0.02

-0.02 -0.04

8 9 0 0 8 9 0 0 01 01 991 992 993 991 992 993 1983 1984 1985 1986 1987 198 198 199 1 1 1 1994 1995 1996 1997 1998 1999 200 20 1983 1984 1985 1986 1987 198 198 199 1 1 1 1994 1995 1996 1997 1998 1999 200 20 Year Year

Figure 9. Generalised additive model of maturity in female squid: smoothers and factor coefficients for effects of (a) length (DML, mm), (b) month and (c) year.

0 2 0 1 11) = 0 df DML, s( 0 1 - 0 -2

100 200 300 400

DML (a)

1 2 3 4 5 6 7 8 9 10 11 12 h2 0246 Mont r o f al i t 2 - par 4 - 6 -

Month2 (b)

0 6 84 88 93 94 95 99 01 9 985987 9 98999 991 992 9 9 9 99 998 9 000 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 0 r Year2 o f al i part -2 -4

Year2 (c) Figure 10. Generalised additive model of maturity in male squid: smoothers and factor coefficients for effect of (a) length (DML, mm), (b) month and (c) year. 15 10 5 10) = 0 df L, s(DM -5 10 - -15

100 200 300 400 500 600 DML (a)

0 2 1 2 3 4 5 6 7 8 9 1 11 1 2 h2 ont r M 0 o f al i part -2

Month2 (b)

1 8 1 92 93 96 984 985 987 988 989990 99 9 9 994 995 9 99 999 000 00 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 .5 0 0 . 0 r Year2 o f al i .5 0 - part .0 1 - .5 1 -

Year2 (c)

Figure 11. GAM for maturity of squid in December: partial fits for effects of length and year in females (above) and males (below).

8 84 86 89 90 91 92 93 94 97 9 99 00 01 0 19 19 19 19 19 19 19 19 19 19 19 2 20 3 5 2 1 0 = 3) r Year2 df , o L M al f i D 0 ( s part 5 - -1 -10 -2

100 150 200 250 300 350 Year2 DML 0 . 1 5 .5 0 0 0 . = 7) 0 10) df Year, ( DML, s ( s .5 0 - 5 - .0 1 - -10

100 200 300 400 500 1985 1990 1995 2000 DML Year