Postflexion larvae and juvenile daily growth patterns of the Alborán Sea sardine ( Sardina pilchardus Walb . ) : influence of wind

1 Instituto Español de Oceanografía, Centro Oceanográfico de Baleares, Palma de Mallorca, Apdo. 291, 07015, Baleares, Spain. E-mail: francisco.alemany@ba.ieo.es 2 Instituto Español de Oceanografía, Centro Oceanográfico de Málaga, Puerto Pesquero s/n, Apdo. 285, 29640, Fuengirola, Málaga, Spain. 3 Instituto Español de Oceanografía, Centro Oceanográfico de Gijón, Avda. Principe de Asturias 70 bis, 33213 Gijon , Asturias, Spain.


INTRODUCTION
It is well-known that population size, especially in the case of small pelagic species, can exhibit large interannual fluctuations (Kawasaki, 1991;Wyatt and Larrañeta, 1988).These fluctuations are mainly attributable to variations in recruitment success, which in turn are mainly due to very high and extremely variable mortality (Peterson and Wroblesky, 1984;Bradford, 1992) during the planktonic stages (Hjort, 1914(Hjort, , 1926)), which represent a critical period (May, 1974).Several authors have proposed hypotheses to explain the mechanisms controlling larval survival (Cushing, 1975(Cushing, , 1990;;Lasker 1978), whose validity has been supported by field studies (Beaugrand et al., 2003;Platt et al., 2003;Fortier and Gagne, 1990;Parrish et al., 1981;Lasker, 1985;Peterman and Bradford, 1987).In most cases it has been recognised that variance in recruitment can be explained through relatively simple atmosphere-ocean interaction models.
Following the SARP approach (Bakun et al., 1991), daily growth studies of surviving juveniles were used to identify the periods when juveniles were born, and hence to determine the environmental conditions favourable for larval survival.This methodology was first applied in the CALCOFI area (Methot, 1983), and later in the SE Atlantic (Painting et al., 1998;Van der Lingen, 2002), SW Atlantic (Anonymous, 1989) and NE Atlantic (López-Jamar et al., 1995), in the latter case to analyse the recruitment process of European sardine (Sardina pilchardus Walbaum 1792).The present study was undertaken within the framework of a research project one of whose main objectives was to define the Optimal Environmental Windows (OEWs) (Cury and Roy, 1989;Guisande et al., 2001) for Alborán Sea European sardine populations, which have showed major fluctuations in the last few decades (Abad et al., 1998).
The sardine in the western Mediterranean has a protracted spawning season, from autumn to spring (Alemany, 1997;Olivar et al., 2003).In the Alborán Sea the presence of sardine eggs has been reported from September to May (Rodríguez, 1990).Along the narrow northern shelf of the Alborán Sea, several bays constitute important nursery grounds for small pelagic fishes, among which the Málaga bay is the most important (García et al., 1988).Sardine is the most abundant small pelagic fish species in the Alborán Sea (Abad et al., 1998) and, like anchovy, is one of the traditional target species of the Mediterranean Spanish purse seine fleet (Alemany and Álvarez, 1993).
The main objective of this work was to determine the birthdates of juvenile sardines of the Alborán Sea for the 2004 spawning season by means of daily growth analysis in order to define optimal environmental conditions of surviving sub-cohorts.We define juveniles as pre-reproductive individuals that have all the fin ray elements and scales and are morphologically similar to adults.In addition, a comparison was made between the daily growth patterns of the surviving juveniles from different periods of the spawning season, and between growth patterns of surviving juveniles and postflexion larvae born in the same period.
As the preliminary results showed that the daily growth rates of juveniles belonging to different subcohorts appeared to be synchronised, mainly from April onwards, suggesting that they were strongly influenced by some environmental factor, we explored the relationships between daily growth rates in juveniles and an environmental index based on wind direction and speed.

The study area
The Alborán Sea is a hydrographically complex region.It straddles the transition between the Mediterranean Sea and the Atlantic Ocean, and is mainly conditioned by the water masses flowing through the Straits of Gibraltar.However, the inflow of Atlantic waters only affects the slope  en postflexión nacidas en el mismo mes, dando apoyo a la hipótesis de "mayor es mejor" en relación con la supervivencia larvaria.La influencia de las condiciones ambientales sobre el crecimiento se estudió relacionando la evolución temporal de un índice de crecimiento basado en la variabilidad de las anchuras de los incrementos diarios con un sencillo descriptor ambiental, un "índice de viento" calculado teniendo en cuenta tanto la dirección como la intensidad de los mismos en el área.Este análisis indica una relación positiva tanto de la supervivencia larvaria como de la evolución de las tasas de crecimiento diarias en larvas y juveniles con periodos de viento de componente norte y oeste, que producen afloramiento en la plataforma continental y condiciones de mar calma en las bahías donde tiene lugar la cría.Al contrario, períodos de viento de levante y sur, que interrumpen el afloramiento e inducen condiciones de temporal en las zonas de las bahías, producen un descenso de las tasas de crecimiento de los juveniles y una menor supervivencia en las larvas.

Postflexion larvae and juvenile sampling
Sardine larvae were sampled off the Alborán Sea coast during the acoustic-ichthyoplanktonic survey carried out on board the R/V "Cornide de Saavedra" in November-December 2003.Selection of optimal sites for sampling postflexion larval stages of sardine was based on zooplankton samples from night oblique standard bongo 60 tows, carried out along a set of transects quasi-parallel to the coastline off the northern Alborán Sea, which were briefly examined on board for the presence of postflexion sardine larvae.All selected postflexion sardine larvae sites were located in the bays of Málaga and Almería (Fig. 1).Surface night hauls were carried out between 14 and 18 December 2003 using a quadrangular bongo net with black tinted 1 mm mesh.In order to prevent larval damage, 10 minute hauls were carried out at two knots, and sardine postflexion larvae were immediately sorted and frozen in liquid N 2 for storage.In the laboratory, sardine postflexion larvae were thawed at ambient temperature and then measured to standard length using an image analysis program (NIH Image).A representative sample of postflexion larvae from each sample site was selected: 64 from Almería, ranging between 16.04 and 22.92 mm, and 55 from Málaga, ranging between 16.14 and 22.70 mm.The otoliths were then extracted using fine needles, dropped onto a slide and cleansed with a drop of distilled water.After drying, the sagitta were fixed with nail lacquer.
Juvenile sardines, belonging to age class 0 were sampled from the commercial purse seine fleet landings at the Alborán Sea ports of Málaga and Caleta de Vélez (Fig. 1).But, as the fishery does not target juveniles, they were only occasionally found in commercial landings (Table 1).Seven samplings were carried out between May and September 2004, with a total of 311 individuals sampled.Standard length was measured to the nearest mm and both sagittal otoliths were removed.After cleaning and drying, the otoliths were mounted separately on glass slides with the sulcus acusticus on the glass side and embedded in Eukitt.

Reading otoliths
The daily nature of microincrements in sardine larvae otoliths was validated by Ré (1984), and the location of hatch ring was defined by Alemany and Álvarez (1994).Taking into account these references, the larval sagitta-one per individual-were read by one reader using the OTO program for McIntosh (Andersen and Moksness, 1988), as described in García et al. (2003).Daily microincrements were counted and their width measured for every individual.All the final sagitta measurements for the numerical analysis (counts, widths, radius) were done at 1000 x magnification.
Juvenile otoliths were analysed with the first MS-DOS version of the OTO program, using the methodology proposed by Methot (1981), as described in Álvarez and Alemany (1997).For each individual, a preliminary examination of each pair of otoliths was carried out in order to select those showing the clearest rings and the most complete readable transects (without breaks), from the focus to the posterior margin of the otolith.Increment widths were measured from live images using a video-coordinated digitiser interfaced with a microcomputer.In each case, the optimal focus was chosen.Several readings at different magnifications (160 x, 630 x and 1000 x) were made along a transect located at ±5º of the longest radius, from the focus to the posterior margin of the otolith.Groups of homogeneous daily increments were sequentially counted and measured.Otoliths were progressively polished on one side between readings to reveal increments, using 30.9 and 0.3 µm lapping film.All otoliths were read independently by two readers.Data from different readings were finally combined to calculate the age of each individual.Whenever discrepancies in age estimations exceeded four days, a joint reading session was carried out and discussed, and subsequently the otolith was again independently examined by the two readers.When discrepancies remained-only 7% of cases-the data were rejected.Interpolation procedures were not applied, since in all the otoliths the whole reading transect, from the hatching ring (Alemany and Álvarez, 1994) to the border, was analysed in at least one of the subsessions.All otoliths in which some part of the transect was unreadable, or subdaily increments were not clearly distinguishable from true daily microincrements, as well as all otoliths that got broken during the polishing procedure before complete readings, were also rejected.All 40 pairs of otoliths from the August sample were rejected because, for unknown reasons, they appeared to be degraded and did not show a clear enough internal structure.Therefore, out of 311 pair of otoliths available, only the data from 156 individuals were finally considered.
As a first step to evaluate the reliability of juvenile otolith readings, length at age data were analysed using the Fishparm software package to obtain Von Bertalanffy growth parameters.Furthermore, to detect sub-cohorts and to explore general growth patterns and compare them with those obtained from otolith readings, a modal progression analysis of the standard length distributions was carried out using Battacharya's method, again using the Fishparm software.

Otolith data post-processing
To calculate the birthdate of each individual, the age in days was directly subtracted from the sampling date, since all the counts started in the recognisable hatching day ring (Alemany and Álvarez, 1994).Rough data of otolith readings from postflexion larvae were processed using the OTO program (Andersen and Moksness, 1988) to obtain the width of each daily microincrement per specimen.Daily increment widths were later averaged to estimate the mean otolith growth.Rough data of juvenile otoliths were processed using Matlab software to calculate the width of each daily growth microincrement,  since the OTO software version used did not provide these data directly.Then, five temporal series of these microincrement widths were calculated by averaging, for each calendar day, daily increment width data corresponding to such calendar day from individuals born within the same month.It must be highlighted that the individual series of microincrement widths were similar to the averaged values, with small standard deviations that remained more or less constant throughout the whole time periods.
After that, considering the close relationship between otolith and somatic growth (see Results section) as an indicator of daily growth performance, a growth index was calculated for each of the five averaged series, by subtracting from the width of each microincrement the width of the previous one.Finally, a global series was obtained by averaging the data for each calendar day from the five series considered.The mean growth in length of each of the five groups of juveniles was back-calculated from otolith microincrement width data.This was done by applying the equations obtained from linear regressions between otolith and somatic growth, calculated from both larva and juvenile otolith radius vs. standard length pairs of data (see Figs. 5A  and 5B).Although using simple linear relationships is not the most accurate method for back-calculation (Francis, 1990), we considered the results reliable enough for our purposes.This was because regression values were obtained from individuals sampled within a short period and back-calculation was only applied to a limited range of lengths of the same individuals.Regression coefficients obtained from postflexion larvae analysis were applied for the larval period up to 26 mm in length, whereas for the juvenile phase, from 36 mm onwards, regression values estimated from juveniles were used.The growth in length during metamorphosis was not back-calculated, because due to the important morphological changes that take place during this phase, the available linear relationships are not valid to obtain reliable estimations.

Wind index
To explore the possible relationships between daily growth rates and environmental factors, a wind index based on the wind induced upwelling index proposed by Bakun and Parrish (1990) was developed.Basic wind data coming from a "Seawatch" meteorological buoy located off Algeciras, which can be considered representative of the whole study area, were obtained from the "Puertos del Estado" web site (www.puertos.es).Hourly records of wind stress and direction from November 7 onwards were averaged to calculate a mean daily wind intensity and direction vectors.Northern and western winds produce upwelling in the Alborán Sea area (Bakun and Agostini, 2001;Sarhan et al., 2000) and calm sea surface conditions in the inshore areas of the northern Alborán Sea, where the nursery areas of sardine are located (Ramírez et al., 2004).Thus, each wind vector was decomposed and added up, considering N-S and W-E components as positive and S-N and E-W ones as negative, and the result were multiplied by the wind intensity, using the resulting value in m/s as a "wind index".The formulae used were: -1 st quadrant: (V * ⎜cos a⎟) -(V * ⎜sen a⎟) -2 nd quadrant: (V * ⎜cos a⎟) + (V * ⎜sen a⎟) -3 rd quadrant: (V * ⎜sen a⎟) -(V * ⎜cos a⎟) -4 th quadrant:-(V * ⎜sen a⎟) -(V * ⎜cos a⎟) where V=wind velocity in m/s and a=wind direction.
Finally, the daily data series was smoothed by means of running average over 3.

Regression between wind and growth indices
To analyse the relationship between sardine growth and wind, linear regression analysis between temporal series of growth and wind indices were applied.Two regressions analyses were performed, the first considering the whole series of data and the second only the period comprised between November and April, coinciding with the first growth phase of juveniles characterised by a progressive increasing in daily increment width.

Birthdate distribution and standard length modal progression in juveniles
Postflexion larvae sampled in Málaga bay were born between 7 and 27 November, with 61% of birthdates concentrated between 15 and 23 and peaking on 20 November (Fig. 2a).Postflexion larvae from Almería bay were born in general a little earlier, between 1 and 22 November, with 64% of birthdates between 9 and 17 November, with the mode on 20 November (Fig. 2b).
The birthdate distribution, estimated from daily microincrement counts in otoliths from juvenile sardine (Fig. 2b), showed three main modal groups, with birthdates centred in mid-November, late December and early February, respectively.The best represented sub-cohort, including almost 2/3 of individuals, was that formed by specimens born in November.Modal progression analysis in standard length distributions also revealed at least three welldefined sub-cohorts (Fig. 3).Fitting the modal standard lengths of the respective sub-cohorts to linear functions, daily growth rates were approximately 0.4 mm day -1 and the mean birthdates were between mid-November and early March.

General growth pattern
Length at age data from postflexion larvae otolith readings are shown in Figure 4a.The best fit (R 2 =0.8044) was obtained with a logarithmic function.According to these results, the mean daily growth rate between 20 and 40 days (postflexion lar-vae) was 0.32 mm day -1 .Regarding juveniles, despite the limited range of ages considered, between 116 and 294 days, sardine standard length at age data fitted well to a von Bertalanffy growth model (Fig. 4b), giving the following growth parameters: L ∞ =240 mm; K=0.93 and t 0 =0.001.This implies a mean growth in this range of ages of 0.4 mm day -1 .

Somatic and otolith growth relationship
Otolith and somatic growth showed a close relationship in both postflexion larvae and juveniles.Linear regressions between otolith radius and standard lengths for larvae and juveniles (Figs.5a and b) presented R values higher than 0.9, with p <0.001.

Daily increment width evolution: postflexion larvae/juveniles and among-subcohort comparisons
Postflexion larvae from both Málaga and Almería bays showed an initial period of continuous increase in increment widths, from around 1 µm in the first days of life, during the yolk sac reabsorption, to >2 µm in 18-day-old larvae (Fig. 6).From day 18 to 28, microincrement widths remained stabilised around the same values.Maximum values, between 2.5 and 3 µm, were reached by day 34.Finally, they follow a decreasing trend from day 34 onwards, reaching values <2.5 µm in the last daily microincrements of post larvae more than 40 days old.
The evolution of average daily increment widths in the five groups of juveniles considered (Fig. 7) revealed two main phases.In the first phase microincrement widths increased almost continuously, from 1 to 5-7 µm, whereas in the second phase microincrement widths fluctuated, showing a general decreasing trend.The length of the first phase was variable, and progressively shorter in the individuals born later.For example, it was about 5.5 months in specimens born in October and only 3.5 months in those born in February.Within this phase, in all groups, an initial inflexion in otolith growth rates was observed when increment widths reached about 2.5 µm.At this moment the larvae were around 20 days old and their standard length was about 20 mm.Later, also in all the groups, a period of stabilisation in the increment widths was observed, around 3.5 µm in the October-December groups and between 4 and 5 µm in the January-February groups.This stabilisation in daily increment widths started when individuals were about 50 days old.This period of stabilisation was progressively shorter in the five considered groups.Besides these inflexion points that occur in all groups irrespective of the date, other periods of relative increase or decrease in microincrement widths seem to have been synchronised among groups.As for comparisons among increment widths from larval or juvenile otoliths (Fig. 6), the latter always showed wider increments.The differences became evident from the 18 th day of life, when in postflexion larvae the daily microincrement widths remained stable, whereas in the surviving juveniles the daily microincrements continued being progressively wider.This difference was greater from day 34 onwards, when daily microincrements in juveniles continued increasing their width, whereas the last microincrements of the analysed postlarvae showed a decreasing trend.

Back-calculated lengths
Evolution of mean back-calculated standard lengths of the five groups of juveniles studied is shown in Figure 8.As expected, growth in length revealed two different phases.According to these back-calculated growths, specimens born later grew faster than those born at the beginning of the spawning season.These differences became more evident in the juvenile stage.For example, individuals born in February took less than five months to reach a standard length of 80 mm, whereas those born in October needed more than six months to attain the same length.

Influence of environmental factors on juvenile sardines growth
A similar evolution pattern was observed for the environmental (wind) and biological (growth (Fig. 9).The match seems especially strong from late autumn to early spring, whereas from May onwards the relationship is less clear.It must be pointed out that, despite the visual correspondence in the temporal evolution of the two variables, regression analysis including the whole series of data showed a weak but significant relationship (R= 0.12; P=0.036).Considering only the period from November to April, when visually the concordance between the two series is higher, the R value was 0.24, with P=0.0017.

DISCUSSION
The length at age data obtained for postflexion larvae match those presented in previous studies on sardine postflexion larvae growth in the Alborán Sea (Ramírez et al., 2001;Ramírez et al., 2004).For juvenile growth, our mean length at age values are very similar to those reported by Álvarez and Morales-Nin (1992) and Álvarez (2002) for autumn and winter spawned juvenile sardines in the Balearic Sea, and also to those found by Álvarez and Alemany (1997) for winter spawned juvenile sardines from Atlantic Iberian stock.Furthermore, despite the narrow range of length-at-age data considered, the parameters of the fitted von Bertalanffy growth equation are coherent, with a L ∞ value (240 mm) close to the maximum-recorded lengths in the Alborán populations and t 0 close to 0. This growth pattern was similar to those obtained previously from annual ring readings in the area (Alemany and Álvarez, 1993).The growth rates estimated from modal progression analysis, around 0.4 mm day -1 , also coincided with the growth rates estimated from daily increment readings during the juvenile stage.The coherence of results suggests that the microincrement count was reliable, reflecting the true age of analysed individuals with a low deviation.However, some bias cannot be discarded, mainly in the oldest specimens sampled in September.In this species, the microincrements are usually easily readable during the first growth phase, from the nucleus to the moment in which increments reach their maximum width.At this stage, in some cases it is rather difficult to distinguish true daily rings from subdaily increments.In older juveniles, the increments laid down during late summer are also difficult to count, since they are much more irregular and less defined than those formed in spring.
Estimated postflexion larvae and juvenile birthdates were within the spawning period of the species in the area.Curiously, the birthdates of the best-represented sub-cohort of juveniles (November) matched the birthdates of the postflexion larvae sampled in December.This confirms that survival rates were high in the larvae born in November.However, since the juvenile sampling scheme was not strictly random, and could not be extrapolated to total catch, this does not necessarily mean that this sub-cohort is the one which most contributed to 2004 recruitment.The other two sub-cohorts detected, those born in December and February, could have been less abundant in the first juvenile samples because in the first samplings they would be not fully recruited to the fishing areas.Furthermore, other sub-cohorts born in spring could have finally been fully incorporated into the fishery once the sampling programme finished, as suggested by the presence of some small isolated specimens around 80 mm in the August sampling.Therefore, these later sub-cohorts would be not properly represented in our sampling scheme.
The comparison between the growth rates during the first weeks of life of postflexion larvae and that of surviving juveniles born in the same month showed that the postflexion larvae that survived to juveniles were those that had a faster growth during larval stages and were therefore bigger at a specific age.This supports both the "the bigger is the better" and "the faster is the better" hypothesis (Miller et al., 1988;Folkvord and Hunter, 1986).
The general pattern of daily microincrement width development in juveniles involved an increasing trend until maximum values over 5 µm were reached in late spring, followed by a later decreasing phase in which values fluctuated.This is the same pattern found previously in other daily growth studies on sardine juveniles (Álvarez and Butler, 1992;Álvarez and Morales-Nin, 1992;Alemany, 1993;Álvarez and Alemany, 1997;Álvarez, 2002), and even in other winter spawned pelagics off the Iberian peninsula, as Scomber scombrus (Cotano and Álvarez, 2003;Villamor et al., 2004).These patterns are probably influenced by endogenous factors, such as the decrease in potential growth with age; but they are also modulated by environmental factors, such as temperature and food abundance (Catalán et al., 2006;Ramírez et al., 2004).Since in all the aforementioned cases the start of the fluctuating and decreasing second growth phase was synchronic among autumn, winter and early spring spawned juveniles, and coincided with the beginning of stratification processes which produce a decrease of production in surface waters, we hypothesise that it is triggered by the end of the planktonic bloom.From satellite-derived phytoplankton pigments in the Alborán Sea, García-Gorriz and Carr (1999) stated that the algal bloom period lasts from November to March and the non-bloom period from May to September.Transitional periods occur in April-May, when thermal stratification starts, and in October-November, coinciding with maximum wind variability and loss of stratification within the basin.These figures are coherent with our findings, since the phase of increasing daily growth rates would coincide with the high production period and the decreasing trend would match with the non-bloom period.Mid-and late winter spawned specimens would grow relatively faster than those born earlier because their larval and early juvenile phase, when their growth potential is highest, match the maximum planktonic production.This cycle of annual production could also influence the survival of larval sardines.According to our results, survivors' birthdates match the algal bloom period, in agreement with Cushing's (1975Cushing's ( , 1990) "match-mismatch" theory.However, shorter temporal scale phenomena would modulate these global figures, both in juvenile growth rates and in larval survival.Some of the inflexions observed in daily microincrement width evolution during the first growth phase would be attributable to endogenous factors, as would be expected for the metamorphosis process and their inherent behavioural changes, which would lead to a temporal stabilisation in the increasing trend of microincrement widths.Other changes in otolith microincrement width must be driven by environmental factors.The results of this study suggest that wind stress plays a crucial role in the recruitment process of the Alborán Sea sardine, by both determining larval survival and modulating juvenile growth.The three first periods of northern and western wind predominance coincided exactly with the estimated birthdates of the three detected recruit sub-cohorts.It could be hypothesised that both the increased food abundance, resulting from coastal upwelling events, and calm sea weather conditions in the inshore bays (nursery areas) induced by northern and western winds enhance larval survival.Thus, these periods of northern and western winds would be those in which "Lasker events" (Lasker, 1978;Peterman and Bradford, 1987;Mendelssohn and Mendo, 1987;Cury and Roy, 1989) occur, favouring larval survival.On the other hand, stormy conditions in coastal areas induced by Levantine winds modify spatial and temporal availability of food to fish larvae and thus negatively influence their feeding and growth (Maillet and Checkley, 1991), and hence their survival.A parallel study on larval condition based on the same postflexion larvae analysed in this study (García et al., 2006) suggests that the growth strategy of sardine larvae in the Alborán Sea is adapted to pulses of low food availability, which would agree with the fluctuating daily growth rates observed.
According to our results, wind stress and direction also directly influence daily growth rates during the juvenile period, mainly during late autumn-early spring, when water masses are not stratified.Taking into account the immediacy of the biological response (variation in growth rates) to the considered wind index-both increasing growth rates during NW winds and decreasing growth rates during SE episodes-the effect of wind must not only be related to enhancement of production through upwelling processes, but also to a more direct impact on juvenile feeding rates or metabolism, as occurs in larval stages.Of course, as the low correlation values between the two indices considered suggest, environmental factors other than wind determine larval and juvenile growth.With regard to larval growth, temperature and food availability have been defined as the main factors affecting larval growth and condition (Buckley, 1984;Heath, 1992).Their importance for sardine larvae development has been demonstrated in other Mediterranean populations (Ramírez et al., 2004;Catalán et al., 2006).Several studies have shown the direct impact of temperature on the otolith accretion rate (Mosegaard et al., 1987;Fitzhugh and Nixon, 1997).However, during periods in which these factors remain more stable, as is the case for temperature during winter, when the water column is mixed, the wind must become a more important factor modulating larval and juvenile sardine growth in the Alborán Sea.This study has therefore detected a relationship between wind stress and sardine juvenile growth and sardine larval growth and survival, but further studies will be required to understand the mechanisms driving this relationship.
FIG. 2. -a) Birthdate distribution of sardine postflexion larvae sampled in December 2003 in the bays of Málaga and Almería; b) Birthdate distribution of sardine juveniles from Alborán Sea population sampled throughout 2004.N: number of individuals.
FIG. 5. Linear regressions between a) otolith radius and fish length in sardine postflexion larvae sampled in December 2003 in Málaga and Almería bays, and b) otolith radius and fish length in sardine juveniles sampled from Alborán Sea population throughout 2004.
FIG. 8. -Mean back-calculated growth of 2004 sardine juveniles from Alborán Sea population born in the different spawning season months.The regression values presented in Figure 5a were used for larval period and for juveniles those showed in Figure 5b.