Introduction

Understanding the life cycle of coastal marine fishes is essential for effective management of exploited stocks. This can, however, be remarkably difficult to achieve because many species have complex life histories (Forrester and Swearer 2002; Gillanders et al. 2003; Figueira 2009; Fromentin et al. 2009) often including migrations between offshore marine environments and inshore coastal habitats (Joyeux and Ward 1998; Day et al. 2012). For many species, early life stages occupy inshore nurseries and feeding grounds for several months to years, until they recruit to adult stocks (Beck et al. 2001; Able 2005; Vasconcelos et al. 2007). The condition and growth of juveniles in these areas are expected to influence their recruitment success to the adult population (van der Veer et al. 1990; Beverton and Iles 1992; Ciotti et al. 2014). Therefore, it is particularly important to understand the role, in overall metapopulation function, of the various coastal habitats colonized by the juveniles and to identify nursery grounds among them.

In this regard, temperate coastal lagoons have, so far, received little attention (Kjerfve 1994). Like estuaries, these transitional ecosystems may offer favourable developmental conditions with high primary productivity, lower salinity than seawater and some protection from predation (Claireaux and Lagardère 1999; Gillanders et al. 2003; Dahlgren et al. 2006; Stierhoff et al. 2009; Vasconcelos et al. 2010, 2011). Lagoons can, however, comprise a diverse mosaic of environments with different physico-chemical characteristics (Pauly and Yáñez-Arancibia 1994), and hence differ markedly in their suitability as habitats for early life stages of fishes. The Gulf of Lions, in the Northwestern Mediterranean, is a case in point (Mouillot et al. 2005), with a series of contiguous lagoons that are occupied seasonally by juveniles of numerous highly prized fish species (Quignard et al. 1984), among which the gilthead seabream Sparus aurata L. 1758 is an emblematic example (Audouin 1962; Lasserre 1976).

S. aurata is a euryhaline and eurythermal species found in both the open sea and brackish transitional waters along the coasts of the eastern Atlantic, the Mediterranean and the Black Sea (source FAO.org). In the Gulf of Lions, reproduction occurs in winter at sea and pelagic larvae drift for up to 3 months, their appearance inshore in spring coinciding with metamorphosis to the juvenile stage (Audouin 1962; Bodinier et al. 2010). The juveniles colonise sheltered areas, in particular the coastal lagoons, which they occupy massively over the summer, migrating out to sea in the autumn, when temperatures of the lagoons and open sea become similar (Audouin 1962; Lasserre 1976). Despite their apparent importance as juvenile habitats, the exact role of the various coastal lagoons in local S. aurata metapopulation function is unclear. In particular, the marked abiotic and biotic differences among them could be significant for their relative quality as nurseries. Indeed, there is evidence that a majority of the local adult population originates from relatively small, shallow, brackish lagoons (Mercier et al. 2012).

We focused on four lagoons of the Gulf of Lions with different environmental conditions and freshwater inputs, which harbour large populations of S. aurata juveniles each year. We investigated the hypothesis that shallow brackish lagoons would produce larger juveniles in better condition. To this end, we measured and compared various traits of body condition (Fraser 1989; Norton et al. 2001; Kerambrun et al. 2011) and otolith growth rates (Panfili et al. 2002) of juveniles captured at the end of their first summer of residence. These were then related to lagoon conditions, particularly temperature, salinity, phytoplankton biomass and dissolved oxygen. Terrestrial organic matter (OM) inputs to coastal ecosystems modify the composition, abundance and biomass of benthic prey (Drake et al. 2002; Salen-Picard et al. 2002; Nicolas et al. 2007; Kostecki et al. 2010), which subsequently influence growth and condition of juvenile fishes (Houde 1997). We therefore also investigated the hypothesis that the better habitat quality of brackish water lagoons would be explained, at least in part, by food web enrichment from terrestrial inputs in OM. We therefore analysed the δ13C and δ15N stable isotope signatures in muscle of the juveniles from the four lagoons and related these to signatures of major sources of OM in benthic food webs.

Materials and Methods

Study Area

We focused on four lagoons of the Languedoc-Roussillon region (Mauguio, Thau, Bages and Salses-Leucate; Fig. 1), which all harbour large numbers of S. aurata juveniles each year (Audouin 1962; Lasserre 1976) although they differ in their morphological and physico-chemical characteristics (Table 1). Thau and Salses-Leucate are 2 to 4 m deep on average and exhibit water characteristics close to those of seawater, with low phytoplankton biomass and eutrophication. By contrast, Mauguio is shallow (0.8 m deep on average), brackish and eutrophic, with high phytoplankton biomass. Bages possesses intermediate depth and salinity values, with low phytoplankton biomass and eutrophication.

Fig. 1
figure 1

Map of the study area showing the four studied lagoons (Mauguio, Thau, Bages and Salses-Leucate) in bold

Full size image
Table 1 Morphological and physico-chemical characteristics of the Mauguio, Thau, Bages and Salses-Leucate lagoons for the year studied
Full size table

Fish Sampling

Using fyke nets, 140 juveniles were collected during the autumn of 2011, at the end of their first summer of residence in the four lagoons. To cover the full period of juvenile migration from each lagoon, the fish from Mauguio (N = 44) were collected between September and November, while those from Salses-Leucate (N = 30), Thau (N = 42) and Bages (N = 24) were captured only in October, November and December, respectively. On the day of capture, all fish were transported on ice to the laboratory, weighed to the nearest 0.1 g (total mass = Mt) and measured to the nearest mm (total length = Lt). Two portions of epaxial white muscle were collected and stored at −80 °C for subsequent analysis of lipid and stable isotope composition. Finally, the head of each fish was removed and stored at −20 °C for subsequent extraction of sagittal otoliths.

Condition Indices

Body condition was evaluated for all individuals by the Fulton’s condition factor (K; Bagenal and Tesch 1978), commonly used as an indicator of general well-being and calculated as follows:

$$ K=\frac{100{M}_t}{L_t^3} $$

where M t is the total mass (mg) and L t the total length (mm).

Condition was further investigated by determining total lipid content and concentrations of lipid classes in the muscle of 15 individuals per lagoon. Approximately 100 mg of muscle was ground on dry ice with a mixer mill MM400 (Retsch). Lipids were extracted with a Tenbroek Homogeniser in dichloromethane/methanol (2:1, v/v) (Folch et al. 1957), then spotted onto S-III chromarods (Iatron Laboratories) and separated into triacylglycerols (TAG), free fatty acids, free sterols (ST), sterol- and wax-esters, ketones, acetone mobile polar lipids and phospholipids (Parrish 1987). Chromarods were scanned by a flame ionization detection (FID) system (Iatroscan Mark-VI, Iatron Laboratories), and chromatograms were analyzed using integration software (Peak Simple version 3.2, SRI). Concentrations of lipid classes were estimated individually in μg mg−1 of wet muscle sample mass (w.m.), before calculating total lipid content as the sum of all lipid classes (in μg mg−1 w.m.). The TAG/ST ratio, which compares the concentrations of TAG, an indicator of reserve lipids, and ST, an indicator of structural lipids, was then used as an index of the immediate nutritional and energetic status of the juveniles (Håkanson 1993; Harding and Fraser 1999; Norton et al. 2001).

Otolithometry

Taking the same individuals used for lipid analyses, sagittal otoliths were extracted, washed and prepared for daily ring counts as described by Tomás and Panfili (2000). They were embedded in polyester resin and transverse sections of approximately 1-mm width, including the central nucleus, were made with a precision saw (Buehler, Isomet 1000). The sections were polished to approximately 10-μm depth with the nucleus exposed on the surface. They were photographed with a digital camera system (Olympus ProgRes C5) and ProRes Capture pro 2.5 software linked to a microscope (Olympus BX41, ×400). All pictures were imported into Perfect Image software.

Examination of the otolith sections revealed multiple concentric rings from the nucleus to the edge, clearly visible on all otoliths, in particular along the ventral axis, which was the maximum growth axis of the otolith in S. aurata. In accordance with previous literature (Morales-Nin et al. 1995; Panfili et al. 2002), the daily deposition of these increments was validated using five alizarine-marked S. aurata from an experimental study (Darnaude pers. comm.). Daily otolith rings were counted twice along the ventral axis by two independent readers, from the first visible increment corresponding to mouth opening, 3 days after hatch (Bodinier et al. 2010) to the edge of otolith, namely the capture date. Otoliths were read again if the two estimated ages in days differed by more than 5 %, to achieve a consensual age estimate between the two readers (Campana and Jones 1992). The two separate readings of otolith daily increments were compared by paired t tests after checking normality and homoscedasticity, and when there were non-significant differences, the mean reading was used.

Teleost otoliths can exhibit clear concentric marks (checks) that correspond to major life stage changes, such as metamorphosis from larva to post-larva, or major environmental changes (Campana 1992; Joh et al. 2011; Rey et al. 2012). In S. aurata, two successive checks were observed on the otoliths: (i) a band of five to 15 discontinuous and narrower increments at about 96 days after hatching, corresponding to metamorphosis (Bodinier et al. 2010), and (ii) a later single check mark followed by a clear change in the growth axis (appearance of an accessory primordium) on the ventral side. Because this latter check generally indicates a change of habitat in juvenile fish (Daverat et al. 2011; Feutry et al. 2012), it has provided a useful marker to delineate timing of migration from highly saline offshore areas to low-salinity estuaries (Hsu et al. 2009; Chang and Iizuka 2012). After validation of the presence of such a check on the edge of the otoliths of five of the smallest juveniles (Lt = 30–39 mm) captured in April–May in the four lagoons (two in Mauguio and one in the other three lagoons), we used it as the mark for the beginning of lagoon residence. This allowed estimation of the age at lagoon entrance (in days) and back-calculation of the date of lagoon entrance for each individual. Distances (in μm) were also measured along the maximum otolith growth axis between the nucleus and the check (i.e. larval marine life) and between the check and otolith edge (i.e. juvenile lagoon residence) to calculate absolute open-sea and lagoon daily growth rates (AGR, Panfili et al. 2002):

$$ AGR=\frac{S_f-{S}_i}{T_f-{T}_i} $$

where, for each period of life, Sf and Si are the final and initial distances to the nucleus (in μm) measured along the maximum growth axis and Tf and Ti the corresponding final and initial ages estimated (in day).

Stable Isotope Analysis

Carbon (δ13C) and nitrogen (δ15N) stable isotope analysis constitutes a powerful tool for the description of the structure of coastal food webs and OM transfers through them (Carlier et al. 2008; Layman et al. 2012). Provided that distinct isotopic signatures can be identified for the primary producers, the method allows assessment of the main OM source(s) that sustain animal feeding (Kostecki et al. 2012). White muscle gives the most reliable results for predator isotopic signatures, with the lowest inter-individual variability (Pinnegar and Polunin 1999). Therefore, stable isotopic signatures in carbon (δ13C) and nitrogen (δ15N) were determined for this tissue in 34 juveniles, at least 8 per lagoon among the 15 individuals used for lipid and otolith analyses. Each muscle sample was freeze-dried separately and ground to a fine powder in a mortar and pestle. As total lipid content was low (i.e. <5 % dry mass), individual δ15N and δ13C values were determined on samples without lipid extraction (Bodin et al. 2009). Thus, 0.5 mg was weighed in a tin capsule and analysed with a continuous flow isotope-ratio mass spectrometer (Delta V Advantage, Thermo Scientific) coupled to an elemental analyzer (Flash EA1112 Thermo Scientific). Results were expressed as parts per thousand (‰) differences from international standard reference materials, i.e. the Vienna Pee Dee Belemnite (VPDB) for C and atmospheric N2 for N:

$$ \delta X\left({\mbox{\fontencoding{U}\fontfamily{wasy}\selectfont\char104}} \right)=\left[\frac{R_{sample}}{R_{s \tan dard}}-1\right]\times {10}^3 $$

where X is the 13C or 15N, and R the corresponding ratio 13C/12C or 15N/14N.

Precision was assessed by repeated analysis of an internal laboratory standard (acetanilide) and was around 0.15 ‰ for both C and N.

Statistical Analyses

All statistical analyses in this work were carried out using R software (version 2.12.0, R Development Core Team 2010 ) taking α < 0.05 as the limit for statistical significance.

Inter-lagoon differences in fish biometry, condition, age, otolith growth rates and muscle isotopic signature were assessed after testing for data normality (Shapiro-Wilk test) and homoscedasticity (Bartlett test). When these conditions were met, one-way ANOVA were applied, followed by Tukey post hoc tests. Otherwise, data were compared by non-parametric Kruskal-Wallis tests followed by post hoc multiple comparison tests with Bonferroni correction.

To avoid size bias when comparing fish isotopic signature among lagoons, effect of fish size on muscle isotopic signature was evaluated separately for δ13C and δ15N, by Pearson correlation tests. Inter-habitat differences in the δ13C and δ15N of S. aurata muscle were interpreted in relation with isotopic signatures reported for the major potential OM sources at the base of the benthic food web in each lagoon, i.e. local macrophytobenthos (1) and microphytobenthos (2) and the marine (3) and terrestrial (4) particulate organic matters (POM) that end up mixed in the lagoon water column and sediment (Table 2). For each OM source, from 6 to 24 multiple-year estimates of δ13C and δ15N, signatures were compiled from the recent local literature, all specific to the study area and the four lagoons (Vizzini et al. 2005; Carlier et al. 2008; Harmelin-Vivien et al. 2008; Carlier et al. 2009; Pernet et al. 2012; Escalas et al. 2015). When possible, specific average values and standard deviations were calculated for each source and lagoon. Otherwise (i.e. for local marine and terrestrial POMs, microphytobenthos and the macroalgae of Thau and Bages), we used average signatures and standard deviations derived from the data available for all lagoons. Because, at this spatial scale, the variability in OM signatures among lagoons was always smaller than that observed among OM sources in all lagoons, at least for one of the two isotopes, we are confident that this did not bias identification of food source contributions to S. aurata growth. As normality and homoscedasticity were observed, differences in isotopic signatures among all potential major OM sources were tested for each lagoon by one-way ANOVAs, followed by multiple comparison Tukey post hoc tests (Table 2). The contributions to muscle composition of the four different OM sources were then estimated with mixing model analyses based on Bayesian methods (Parnell et al. 2010), using the R software package ‘stable isotope analysis in R’ (SIAR). A trophic level (TL) of 3.5 was considered for S. aurata (Froese and Pauly 2010). Thus, trophic enrichment factors (TEF) between OM sources (TL = 1) and S. aurata juveniles (TL = 3.5) in the model were calculated for δ13C and δ15N as follows:

Table 2 Carbon and nitrogen isotopic signatures (mean ± SD) retained for each organic matter (OM) source in the four lagoons studied
Full size table
$$ TE{F}_{\mod el}=2.5\times TE{F}_{marine\ organism} $$

where TEFmarine organism represents the average TEF value expected for marine organisms.

For this, TEFmarine organism for both δ13C and δ15N were obtained by averaging the TEF values listed for marine invertebrates and fish species in Caut et al. (2009), resulting in final values (±SD) of 1.26 ‰ (±0.67) (n = 35) and 2.26 ‰ (±0.75) (n = 42), for δ13C and δ15N, respectively. SIAR analyses using the TEFmodel derived from these two TEFmarine organism were run separately for each lagoon to estimate the respective contributions (medians and 50th Bayesian credibility intervals over 500,000 iterations) of the four OM sources to the final composition of muscle tissue.

Results

Size and Body Condition

Total length (Lt) and mass (Mt) of juveniles varied significantly among the four lagoons (Table 3). Individuals from Mauguio and Bages were significantly larger and heavier than those from Thau and Salses-Leucate (Kruskal-Wallis tests, H = 53.52, p value = 1.42 10−11 and H = 60.39, p value = 4.86 10−13, respectively). Fulton’s condition factor K was also significantly different among lagoons, the highest values being observed in Mauguio (Kruskal-Wallis test, H = 52.2, p value = 2.65 10−11).

Table 3 Age and condition (mean ± SD) of juveniles of Mauguio, Thau, Bages and Salses-Leucate, analysed in this study
Full size table

Muscle total lipid content differed significantly among lagoons (Kruskal-Wallis test, H = 17.00, p value = 7.08 10−4, Table 3). However, although the highest mean value was in juveniles from Mauguio, post hoc multiple comparisons did not reveal significant differences among lagoons. The TAG/ST ratio was, however, significantly higher (Kruskal-Wallis test, H = 25.97, p value = 9.68 10−6) in Mauguio and Bages than that in Salses-Leucate, with Thau having an intermediate value (Table 3).

Age and Otolith Growth Rates

Ages at the time of sampling were not statistically different (ANOVA, F = 2.54, p value = 0.07) and were approximately 10 months for all fish (Table 3). Similarly, no significant difference was found among lagoons in the duration of either the offshore larval life (ANOVA, F = 0.66, p value = 0.58) or the inshore juvenile life (ANOVA, F = 1.22, p value = 0.31); all individuals had spent approximately 4 months (mean ± SD; 122.3 ± 9.6 days) at sea before entering the four lagoons and, then, approximately 6 months (182.4 ± 11.6 day) feeding and growing in these ecosystems before their capture in the autumn, before they could migrate back at sea. Back-calculation of hatching dates revealed that reproduction extended at least from December to February, and colonization of lagoons occurred at least from April to June (Fig. 2). The data indicated that the juveniles had entered the lagoons at different times (ANOVA, F = 36.46, p value = 4.14 10−13), mostly during April in Mauguio, from late April to early May in Salses-Leucate, during the second half of May in Thau and from June in Bages (Fig. 2).

Fig. 2
figure 2

Lagoon entrance dates (boxplot) for the S. aurata juveniles captured in the four lagoons during autumn (n = 15 per lagoon). The box for each lagoon contains 50 % of the data, with the thick horizontal line indicating the median; box whiskers (dotted lines) represent the first and fourth quartiles of the total range and circles represent extreme values. Letters indicate significant differences between lagoons (p < 0.05), when present

Full size image

Growth during the marine larval phase did not differ among lagoons (Kruskal-Wallis, Fig. 3a), with a mean overall AGR of 2.8 ± 0.4 μm day−1. Growth during the lagoon phase was, however, significantly higher in Mauguio (9.6 ± 0.7 μm day−1) and Bages (9.3 ± 0.7 μm day−1) than that in Thau (8.5 ± 1.0 μm day−1) and Salses-Leucate (8.5 ± 0.7 μm day−1) (ANOVA, F = 10.05, p value = 2.22 10−5, Fig. 3b).

Fig. 3
figure 3

Boxplot of the a open-sea growth rate and b lagoon growth rate of the 60 juveniles of S. aurata from the four lagoons (n = 15 per lagoon). Circles represent extreme values; letters (a, b) indicate significant differences (p < 0.05) between lagoons

Full size image

Organic Matter Sources Exploited for Growth

For the size range studied (153–205 mm), C and N isotopic ratios in muscle were not affected by fish size (Pearson correlation test: δ13C, t = −2.049, df = 29, p value = 0.05952, R 2 = 0.126; δ15N, t = 1.699, df = 29, p value = 0.09985, R 2 = 0.090). Muscle isotopic signatures varied according to the lagoons for both carbon (ANOVA: F = 28.03, p value = 1.90 10−8) and nitrogen (ANOVA: F = 31.56, p value = 5.63 10−9). Fish from Mauguio had significantly lower δ13C signatures (−19.86 ± 1.53 ‰) than those from Thau (−14.92 ± 0.52 ‰), Bages (−14.24 ± 0.91 ‰) and Salses-Leucate (−12.43 ± 0.48 ‰). On the other hand, muscle δ15N signatures were significantly higher in Mauguio (16.65 ± 0.74 ‰) and lower in Salses-Leucate (8.93 ± 0.22 ‰) compared to Thau (12.06 ± 0.72 ‰) and Bages (12.20 ± 0.65 ‰). This revealed three distinct trophic groups: one in Mauguio with high δ15N and depleted δ13C, one in Salses-Leucate with low δ15N and high δ13C and one in Bages and Thau with intermediate δ15N and high δ13C.

The mixing models revealed that this reflected significant differences among lagoons in the origin of the OM at the base of the sea bream food web, with Mauguio being particularly different from the other three (Table 4). In Mauguio, terrestrial POM and lagoon macroalgae represented 68 to 98 % of the OM exploited, with the former representing 33 to 49 %. Marine POM accounted for less than 9 % of OM exploited in Mauguio. In the three other lagoons, the two main sources of OM (47–93 %) were marine POM and microphytobenthos, with the former representing >20 % in all three lagoons. Terrestrial POM represented about 25 % in Thau and Bages but less than 5 % in Salses-Leucate.

Table 4 Mixing-model estimates of organic matter source contributions to S. aurata somatic growth in Mauguio, Thau, Bages and Salses-Leucate lagoons
Full size table

Discussion

To our knowledge, the current data are the first to compare size and condition with accurate estimates of the age, lifetime growth rates and OM sources of juvenile fish among lagoon nursery grounds. Our results demonstrate that differences in the size and condition of S. aurata juveniles among lagoons at the end of the colonization period were due to differences in growth rate during the lagoon phase. This, in turn, was associated with different food webs and sources of OM. This therefore indicates that shallow brackish lagoons afford better growth rates and produce juveniles in better condition compared to deeper more saline lagoons.

Linking Size and Condition to age and Growth Rates

Previous studies on growth of S. aurata juveniles in the Gulf of Lions reached a similar conclusion when comparing Mauguio and Thau (Audouin 1962; Quignard et al. 1983), but not when comparing Thau and Bages, juveniles from Bages being smaller than Thau (Audouin 1962). In these earlier studies, however, the age of the fish was not known exactly. In the current study, we demonstrated that, after spending approximately 6 months in the lagoons, individuals from Mauguio and Bages were significantly bigger and heavier than those from Thau and Salses-Leucate. Values for the condition factor K in Thau, Bages and Salses-Leucate (1.4) were not significantly different to each other and similar to values reported for wild individuals in Spain and Greece (Laiz-Carrión et al. 2005). However, juveniles from Mauguio were in better ‘body condition’ with a K of 1.6 that is more similar to reports for individuals in aquaculture, which are typically characterised by higher K than wild fish (Arechavala-Lopez et al. 2011).

Direct biochemical condition indices are the most accurate means of measuring the energetic status of marine organisms (Vogt et al. 2002). In fish, lipids constitute the main energy source. The content and composition of tissue lipids can influence growth, reproduction, behaviour/movements, thermal acclimatisation and immune responses (Evans 1994). While total lipid content provides a good estimation of condition, separation and quantification of lipid classes reveal information about levels of metabolically available lipids (Sargent et al. 2002). In this study, the larger size of fish from Mauguio and Bages was generally correlated with a greater accumulation of storage lipid (TAG) and a higher TAG/ST ratio, which indicates larger reserves for use as energetic substrates or as structural components of new tissues.

The otolithometry analysis provided much novel information about the reasons for inter-lagoon differences. As in previous studies (Campana 1992; Joh et al. 2011; Rey et al. 2012), otolith reading revealed the timing of successive life stages in individual fish, indicating the durations of open-sea (larval) and lagoon (juvenile) phases and determining daily growth rates (Panfili et al. 2002). Otolith growth rate is directly related to somatic growth rate in fishes (Campana 1992); so, it is very interesting that the differences in body size and condition of the S. aurata juveniles among lagoons were not due to different ages or duration of larval phases, but to differences in growth rate in the lagoon phase. The date at which individuals entered the lagoons could also play a role in defining their final condition, if they encounter different intervals in the seasonal patterns of water temperature and productivity. The fact that larger fish were found in Mauguio and Bages, however, is not consistent with this theory because they entered first and last, respectively.

Environmental Influences on Condition

The condition and growth rate of juvenile fish depend on food availability (Fraser 1989). In this study, this likely contributed, at least in part, to the differences among lagoons. Indeed, the highest TAG/ST ratios and growth rates were observed in Mauguio and Bages, where the percentage of saturation in dissolved oxygen, often used as a proxy for productivity (Taylor et al. 1992; Bearzi et al. 2008), is maximal (Table 1). This boosts secondary production and probably increased prey availability for S. aurata, as the diet of this opportunistic predator is particularly flexible at the juvenile stage (Kraljevic and Dulčić 1997; Parra and Yufera 2000; Mariani et al. 2002; Tancioni et al. 2003; Chaoui et al. 2005). Digestibility and nutritional quality of the food might also play a role, since they both modulate growth and condition of S. aurata in aquaculture (Morais et al. 2006). However, the origin of OM in coastal food webs can affect the abundance and nutritional quality of the benthos (Kidd et al. 2001; Salen-Picard et al. 2002; Darnaude 2005; Vizzini et al. 2005; Gilliers et al. 2006); so, variation in food web carbon sources, revealed by the δ13C and δ15N isotope analyses, may have contributed significantly to the differences in growth and condition.

Small inaccuracies in the isotopic signatures of the OM sources used to parameterize the SIAR analyses cannot be fully excluded, mainly for Thau and Bages where site-specific signatures could not be found for all sources. We used, however, mean values from similar neighbouring lagoons influenced by the same watersheds (IFREMER 2011). Moreover, variations in OM source signatures among lagoons at our spatial scale were smaller than those observed between OM sources in all lagoons, with at least one of the two isotopes always allowing full separation of all sources. Therefore, we are reasonably confident that any bias in estimating contributions of the OM sources would not obscure the marked variation among lagoons, especially since the variance around the means was incorporated into the SIAR model. Another potential source of error is the kinetics of isotopic signature turnover in fish muscle. In juvenile fishes, growth is the main factor that causes turnover of muscle isotope signatures (Hesslein et al. 1993; MacAvoy et al. 2001; Maruyama et al. 2001; Sakano et al. 2005; Suzuki et al. 2005; Logan et al. 2006; Guelinckx et al. 2007); it takes anything from 2 to 7 months for juvenile fish muscle to reflect the isotopic signature of their food (Maruyama et al. 2001; Sakano et al. 2005; Guelinckx et al. 2007). The juveniles were captured in autumn as they migrated towards the sea to overwinter offshore. The otolith data indicate that they had been in the lagoons for at least 5 months; so, the isotopic signature of their muscle should reflect their lagoon food web.

In Mauguio, where juveniles grew the fastest and accumulated the most lipid reserves, food webs were highly dependent on terrestrial POM. That was not the case for the three other lagoons, where the major OM sources were lagoon microphytobenthos and marine POM. These results are in general agreement with the existing literature. Terrestrial POM uptake by coastal marine communities is generally weak (Deegan and Garritt 1997; Kwak and Zedler 1997; Paterson and Whitfield 1997; Vizzini and Mazzola 2003). In transitional systems under limited freshwater influence, in situ primary production is generally the major source of OM for growth of young fish (Boesch and Turner 1984; Melville and Connolly 2003; Kostecki et al. 2012; Le Pape et al. 2013), which was what we observed in Salses-Leucate and, to a lesser extent, in Thau and Bages. Many studies have, however, found significant incorporation of terrestrial OM into juvenile fish food webs in estuarine areas with major terrestrial inputs (Darnaude et al. 2004; Leakey et al. 2008; Pasquaud et al. 2008; Vinagre et al. 2008; Kostecki et al. 2010). This is the case in the Mauguio lagoon where, despite a high phytoplankton biomass and density (Henard and Vaulot 1979), the trophic importance of the pelagic pathway for benthic communities is limited and food webs are known to be primarily based on macroalgae and sedimentary OM (Vizzini et al. 2005). Because this latter, in turn, consists largely of terrestrial POM, our results corroborate previous observations and also indicate that, when present, OM of terrestrial origin in benthos is transferred up to seabream juveniles. Terrestrial OM use by marine fishes has already been demonstrated in several coastal and estuarine ecosystems (Darnaude 2005; Martinho et al. 2009; Kostecki et al. 2010; Le Pape et al. 2013), where freshwater inputs into nursery habitats can control both the availability of suitable prey for fish juveniles (Houde 1997; Darnaude et al. 2004) and their size (Le Pape et al. 2003). Studies in extensive aquaculture in brackish ponds have shown that supplementary feeding of S. aurata juveniles with high-quality food can increase their growth by 10 to 20 % (El-Ghobashy et al. 1993; Sadek et al. 2004). Therefore, taken together, it seems reasonable to speculate that the freshwater inputs into Mauguio, which are rich in terrestrial POM, contributed significantly to the higher growth and condition of resident seabream juveniles, through effects on prey availability and nutritional quality.

Abiotic conditions such as water temperature, salinity, dissolved oxygen and perhaps pollution can also influence growth and condition of fishes (Tandler et al. 1995a; Claireaux and Lagardère 1999; Boeuf and Payan 2001; Gilliers et al. 2006). In ectotherms, warmer temperatures generally promote growth by accelerating all metabolic processes (Claireaux and Lagardère 1999), including digestion and assimilation (McCue 2006). Temperature has a very marked influence on growth of S. aurata (Tandler et al. 1989), optimal rearing temperatures for the species’ post-larvae being between 20 and 28 °C (Requena et al. 1997). However, as the four lagoons provide temperatures in this range and have overlapping values, this abiotic factor does not seem to be responsible for the differences in fish growth and condition. On the contrary, differences in salinity or dissolved oxygen levels among lagoons might have partially modulated growth in S. aurata juveniles. Indeed, although S. aurata larvae grow well over a wide range of salinities (Tandler et al. 1995), including all those found during S. aurata presence in the four lagoons investigated, the optimal salinity for their growth is around 28 (Klaoudatos and Conides 1996; Conides et al. 1997; Laiz-Carrión et al. 2005). As shown in Table 1, salinities close to 28 are most often found in Bages, the waters in Mauguio typically being less saline and more saline in Thau and Salses-Leucate. This might have contributed to the differences in growth between Bages and Thau, but other environmental factors might be involved. For example, juvenile fishes are strongly affected by hypoxia, and their growth rate in coastal nurseries can be correlated with levels of dissolved oxygen (Stierhoff et al. 2009). Among the four lagoons, Thau is the worst in terms of water oxygenation (Table 1), with a mean dissolved oxygen of 6.2 mg L−1, due to frequent values between 2 and 5 mg L−1 during the summer (IFREMER 2012). Such low oxygen concentrations can have negative effects on fish growth and survival (Diaz and Rosenberg 2008; Stierhoff et al. 2009) and may therefore have contributed to the lower growth rates in this lagoon. Thus, the highest growth and the larger lipid reserves of juveniles in Bages are likely due to a combination of optimal salinities and oxygen availability.

Altogether, the data indicate that Mauguio was the best habitat for juvenile seabream in the particular year studied. This may not, however, always be the case. Among the four lagoons, Mauguio is the most susceptible to dystrophic crises that cause localized hypoxia (IFREMER 2010). No hypoxic events (i.e. oxygen concentration below 2 mg L−1) were recorded in any of the lagoons during the summer of this study (2011), but, because hypoxia can cause extensive mortality and inhibit growth rates of fishes (Diaz and Rosenberg 2008; Stierhoff et al. 2009), such events can be expected to have negative effects on fish growth and survival. Similarly, diverse pollutions could be associated with the freshwater inputs and thus alter the nursery function of lagoons. The 15N enrichment found in Mauguio sediments and primary producers has long been considered an indicator of wastewater pollution (e.g. McClelland and Valiela 1998). The Mauguio lagoon is also polluted by pesticides and persistent organic pollutants (POPs), mainly associated with the freshwater inputs (Souchu et al. 2010; Bec et al. 2011; Brehmer et al. 2011). For demersal fish, contamination of nursery areas can cause a decline in growth and population density (Gilliers et al. 2006). Previous studies have demonstrated a negative impact of contaminants on the development and recruitment of various aquatic organisms in Mauguio (Brehmer et al. 2011). The exceptional growth rates and body condition reached by S. aurata juveniles in Mauguio could therefore well be pondered by negative effects of freshwater inputs in terrestrial POM on fish physiological status and survival. Particular attention should therefore be paid to reduce pollution and control eutrophication in Mauguio to preserve its potential as nursery site for local S. aurata populations.

Implications for S. aurata Life Cycle and Metapopulation Function in the Gulf of Lions

Growth and body condition are primary criteria for classifying nursery habitats (Gilliers et al. 2006; Vasconcelos et al. 2009), indicating that Mauguio and, to a lesser extent, Bages might be key nursery sites for S. aurata in the Gulf of Lions. For many juvenile fishes, the first winter is a critical period (Miranda and Hubbard 1994; Fullerton et al. 2000; Jolley et al. 2013), with severe conditions in the open-sea such as low temperatures, limited food resources and an elevated risk of predation. According to the ‘bigger is better hypothesis’ (Miller et al. 1988), the largest juveniles with the best body condition may have increased chances of survival over winter. They may be less susceptible to predation due to gape limitations, and better able to resist starvation and tolerate physiological extremes, allowing them to retain condition and performance for a longer period (Sogard 1997; D’Alessandro et al. 2013) and so survive until the next spring.

The finding that shallow brackish lagoons produce larger and heavier juveniles than deeper ones with higher salinity is interesting in light of the recent findings on lifetime migrations of S. aurata adults in the Gulf of Lions (Mercier et al. 2012), which suggested that most of the S. aurata adults fished offshore originated from a shallow, brackish lagoon nursery. Juveniles from lagoons like Mauguio, in better condition after their first summer, may therefore recruit proportionally more to the adult population. Such differences in the contributions of lagoons to adult stocks, if confirmed, clearly are of profound consequence for local population dynamics and stock management strategies.

Conclusions and Perspectives

This study provided clear evidence that lagoons differ in their quality as juvenile fish habitats, which may result in large differences in the contribution of nursery areas to adult stocks (Mercier et al. 2012). Our results indicate that ongoing and future modifications of river inputs to the coastal zone in the NW Mediterranean, under the combined influence of climate change and anthropogenic freshwater use (Dolbeth et al. 2008), could profoundly alter the nursery function of transitional and coastal ecosystems. If enrichment of food webs in lagoons is altered, the consequences for fisheries resources could be dramatic. It is essential to have effective sustainable management plans for these fragile ecosystems, to ensure their water quality, and so preserve their carrying capacity as nurseries of highly prized coastal fish species.