The study was performed in the Gulf of Tunis (central Mediterranean Sea, 37°N 10°30'E). The brown meagre is caught by commercial fishing boats using artisanal fishing gears (mostly gillnets and trammel nets) and trawl, at depths ranging from 20 to 100 m, according to information provided by experienced fishermen who know the fishing area well. A total of 276 individuals were randomly collected on a monthly basis from October 2008 to September 2011. For each fish, the total length (TL) was measured to the nearest millimetre and the total weight (TW) to the nearest gram. The sex (male, female or undetermined) was determined by macroscopic observation of the gonads. Sagittal otoliths were removed, cleaned and stored dry in labelled envelopes for later treatment and examination.

The sex-ratio (males:females) for the entire sample was estimated. A chi-square test was used to detect differences in the sex-ratio of the sampled fish. The Mann-Whitney U test was applied to test the differences between sexes according to the TL (cm) and TW (g) (Zar 1999).

The relation between length and weight was described by the power function TW=aTL^b. The regression parameters a, b and the coefficient of determination (r²) were estimated for the whole population and for each sex by least square linear regression, after log transformation of both variables. According to Ricker (1973), the slope of the regression was corrected to follow a geometric mean regression. A Student t-test was used to appreciate the hypothesis of isometric relationship, whereas analyses of covariance (ANCOVA) were used to detect any significant differences in the linear relationships between sexes (Zar 1999).

Before estimating the age, two preliminary steps were applied to choose the calcified structure used and to validate the periodicity of observed growth increments. Comparing the results between the otoliths and the scales, in old fishes age is often underestimated or impossible to estimate when studied by scalimetry. Therefore, sagittal otoliths are the most common and widely used calcified structure. They appear to be highly suitable for age estimation as they are metabolically inert, so any material deposited remains unaltered and cannot be resorbed (Campana and Neilson 1985). Furthermore, some otolith preparation techniques that enhance the visibility of annuli are available, and their applicability depends on the specific characteristics of fish otoliths (Arneri et al. 1998).

Otoliths of S. umbra, like those of most Sciaenids species, are too thick to allow direct reading of the growth marks, so thin cross-sections through the core (nucleus) of the otoliths were used. The preparation method used to section into thin slices was described in Mahé et al. (2009). Otoliths were embedded in translucent polyester resin and two thin cross-sections (0.45 mm thick) were made using a high saw. These sections were then viewed under transmitted light against a white background, using a binocular microscope linked to a video camera connected to a computer. Paraffin oil was used to improve the visualization of growth increments. Each sample was then analysed using TNPC software (numeric treatment of calcified pieces, www.tnpc.fr).

The association of one opaque zone and one translucent zone was considered an annulus. The path of the growth marks was counted from the nucleus towards the tip of the inner face next to the sulcus (Fig. 1A). Measurements were taken from the otolith core to each opaque zone (R_n) and to the proximal margin of the otolith (R_o) to the nearest 0.01 mm.

The periodicity of opaque zone formation was validated by the marginal increment analysis (Jearld 1983). The marginal increment ratio (MIR, 0.01 mm) of fish with more than one opaque zone was calculated by the equation MIR=R₀-R_n/ R_n-R_n-1, where R_n and R_n-1 are the radii of the ultimate and penultimate annuli, respectively. The Kruskal-Wallis test was used to test the homogeneity of MIR among months followed by a Student-Newman-Keuls (SNK) non-parametric test after null hypothesis rejection. Correlation between MIR and temperature was tested using the Spearman rank correlation coefficient (r_s) (Zar 1999). The mean sea water temperature data were supplied by the National Institute of Meteorology of Tunisia (www.meteo.tn). According to the study on the reproductive characteristics of S. umbra in the Gulf of Tunis (Chakroun-Marzouk and Ktari 2003), the peak of the reproductive season is between May and June, so 1 May was considered the birth date of the fish.

To assess ageing precision, the growth increment was counted independently by two readers without prior knowledge of fish length or sex. The variability of the within-reader age estimates was calculated with the absolute percent error (APE) (Beamish and Fournier 1981) with the equation

where R is the number of times each fish is aged, Xij the i(th) ageing of the j(th) fish and X the mean age calculated for the j(th) fish. Only interpretations with an agreement between two readings were used in the successive elaborations.

The von Bertalanffy growth function is the most commonly used to describe fish growth. It was fitted to the observed length-at-age data as follow:

where TL_t is the total length (cm) at age t, TL_∞ is the asymptotic total length (cm), k is the growth curvature (years^–1), t is the age (years) and t₀ is the theoretical age at which length is zero. These parameters were estimated using Fishparm, a computer program based on a nonlinear model with an algorithm for least-square estimation (Saila et al. 1988).

Separate analyses were done for males and females. The growth curves were compared between sexes using the Hotteling test (Zar 1999). A comparison with other areas was performed producing a synoptic table and an auximetric grid of Pauly et al. (1996), in order to judge the coherence among present and other data.

Catch curves were constructed by plotting the log of the number of fishes caught against their age class. Only age classes fully recruited to the sampling gear (i.e. the age classes in the descending limb of the catch curve) were used to estimate total mortality (Z). Z was estimated by the slope of the linear regression through the descending right limb of the catch curve (Pauly 1983). The instantaneous rate of natural mortality (M) was estimated by the updated empirical method of Hoenig (1983), M=4.899t_max^–0.916, where t_max is the maximum age (Then et al. 2015). Then, the instantaneous rate of fishing mortality (F) was obtained from the difference between Z and M. The exploitation ratio E was estimated by dividing F by Z (Gulland 1971).

A total of two hundreds and 76 specimens composed of 121 males, 117 females and 38 specimens with undetermined sex were sampled. The sex-ratio of 1: 0.97 was not significantly different from the1: 1 ratio (χ²=0.7, n=238).

Males ranged in size from 17.6 to 43.3 cm and females from 19.1 to 49.2 cm. The mean TL of males (TL=28.98±0.52 cm) was not significantly different from that of females (TL=30.73±0.62 cm) (Mann-Whitney U test, n=238, p=0.06). Fish of undetermined sex were ranged between 15.3 and 26.3 cm (TL=20.04±0.50 cm) (Fig. 2). The total weight of fish was between 67 and 1063 g (TW=335.12±19.28 g) for males and between 84 and 1565 g (TW=424.01±27.73 g) for females. The mean TW was significantly different between the two sexes (Mann-Whitney U test, n=238, p=0.04). Total weight of fish with undetermined sex varied between 43 and 239 g (TW=99.83±8.29 g).

The parameters of the length-weight linear regressions are provided for each sex and all individuals in Table 1. Our data suggested that S. umbra showed positive allometric growth. The sex effect was not significant on the length-weight relationships (ANCOVA, n=238, p>0.05).

Growth increments were counted on otolith cross-sections (Fig. 1A). Among the total sections examined (n=276), 252 (91.3%) yielded useful age estimates and 24 were discarded as they showed disagreement between the two readings. The value of APE (2.79%) demonstrated the very good reproducibility between readings, the reliability and the high precision of sectioned otolith for ageing the species.

The marginal increment analysis showed considerable individual variation (Fig. 3) but with a consistent pattern of higher values from August to February and lower values from March to June, the months corresponding to the mark formation period (Kruskall-Wallis test, p=0.009; SNK, p=0.001). The monthly trend in marginal increment analysis showed that only one mark is deposited per year. Significant correlation was found between mean marginal increment values and sea surface temperature (r_s=0.12, p<0.05).

The relationships between TL and otolith radius (Ro) were calculated for males and females. Fish length and the radius of the calcified structure was closely correlated (p=0.00, Fig. 4). The analysis of covariance showed a significant difference in the length-radius relationship between the two sexes (ANCOVA, p<0.05).

The observed age of fish ranged from 1 to 31 years (Fig. 1B). Females reached a higher maximum age (31 years) than males (22 years). No significant differences in mean ages between males (3.74±0.39 years) and females (4.27±0.43 years) were observed (Mann-Whitney U test, n=226, p=0.34). Most of the fish, accounting for 69% of the total sample, were between 1 and 3 years old (Table 2). Fish older than 9 years old were poorly represented in the sample. More than 50% of the maximum length was achieved by the first year.

The growth curves of age-at-observed length data were well described through von Bertalanffy fits (Fig. 5). Significant differences were found between the von Bertalanffy growth curves of both sexes (Hotelling’s T² test=10.07, p<0.05).

The von Bertalanffy relationships between age and length, derived from the assumed annual periodicity of the growth increments, were described by the growth parameters: TL_∞=43.83 cm, k=0.145 year^–1, t₀=–4.88 years for males and TL_∞=50.10 cm, k=0.105 year^–1, t₀=–5.71 years for females.

Total mortality, corresponding to the slope of the descending limb of the catch curves, was Z=0.51 year^–1 for males and Z=0.41 year^–1 for females (Fig. 6). Values of natural mortality were M=0.29 year^-1 for males and M=0.22 year^–1 for females. Calculation of fishing mortality gave F=0.22 year^–1 for males and F=0.19 year^–1 for females. With the values of M and F known, the exploitation ratios were computed as E=0.43 for males and E=0.46 for females.

This study showed no significant difference between the length-weight relationships of males and females of S. umbra. This result corroborates the study realized in the northwestern Adriatic Sea (La Mesa et al. 2008). The length-weight relationship parameters obtained from different areas are given in Table 1. The slope values (b) of the length-weight relationships for both sexes (b=3.09 for males, b=3.18 for females) and for the entire sample (b=3.15) showed a positive allometric growth. These results are different from those found by Chakroun-Marzouk and Ktari (2003) in the same area and by Dulčić and Kraljević (1996) in the east Adriatic Sea, where the growth was isometric. Our results corroborated those obtained in the western Mediterranean Sea, around the Balearic Islands and the eastern coast of the Iberian Peninsula (Morey et al. 2003, Grau et al. 2009). In the northern Aegean Sea (Karakulak et al. 2006) and the northwestern Adriatic Sea (La Mesa et al. 2008), a positive allometric growth was also reported. The fluctuating pattern of the length-weight relationship can be assigned to factors such as food availability and feeding rates (Bagenal and Tesch 1978), the temporal effect related to the spawning season (Mahé et al. 2017), composition in size of the individuals sampled and the period of sampling (Moutopoulos and Stergiou 2002). Table 1 shows differences in the size interval of captured fishes, samples showing a positive allometry being composed of larger individuals than those with isometric growth. For this reason, Petrakis and Stergiou (1995) suggested that the use of the length-weight relationship and its comparison must be limited to the interval of size used to evaluate linear regression parameters.

The large size and thickness of the otoliths in Sciaenids species do not allow a direct reading on the surface of the whole otolith under reflected or transmitted light. The dense calcium carbonate deposition reduces light transmission, making it almost impossible to distinguish hyaline and opaque zones (Arneri et al. 1998). Consequently, different otolith preparation techniques (burnt, polished and sectioned) have been tried. Cross-sections have proved to be a very effective method for determining the age of Mediterranean brown meagre (Arneri et al. 1998). The significant correlation observed between the TL and the radius of the otolith cross-section indicated that this structure is useful for estimating age and for reconstructing the past growth history of S. umbra.

The low values of APE found in this study, like those determined by La Mesa et al. (2008) in the Adriatic Sea, indicate a high precision between readers and suggest the good precision of the otolith cross-section method for ageing this species.

Marginal increment analysis confirmed the deposition periodicity of growth increments, showing an annual pattern. It also proved that the opaque zone was deposited between March and June. This period is identical to that observed for the formation of the annulus on the scales of the brown meagre in the same area (Chakroun-Marzouk and Ktari 2003). In the Adriatic Sea, La Mesa et al. (2008) noted the thinnest marginal increment value in July. The formation of alternating translucent and opaque zones might be related to a complex control by environmental and endogenous factors (Beckman and Wilson 1995). Indeed, physiological changes are produced mainly by cyclic variations of temperature, photoperiodism and food supply (Pannella 1980). The temperature mean showed the same trend as the marginal increment analysis, with a small time lag (1 or 2 months). This pattern is linked to temporal fluctuations in metabolism and water temperature, where seasonal temperature variations exceed 3°C to 4°C (Høie et al. 2008, Neat et al. 2008). The patterns of opaque and translucent zone formation have been linked to a seasonal change in otolith daily increments widths (Beckman and Wilson 1995) themselves in relation to variations in the rhythm of the metabolism of calcium (aragonite) and organic material (otolin) (Morales-Nin 1987).

The examination of otolith cross-sections showed that the maximum age observed in the Gulf of Tunis was 31 years for a female of 49.2 cm. A very similar result of the same age was found for a female of 49.7 cm, poached in the marine reserve of Scandola, protected since 2014 (Morat et al. 2017). It appears that the brown meagre of the Gulf of Tunis has the longest life span reported until now in a natural environment. Indeed, in Maltese waters, the Sea of Marmara, the northwestern Adriatic and the southeastern Black Sea, the oldest individual was 26 years old (Table 3). In a previous study of the growth of S. umbra in the Gulf of Tunis, Chauvet (1991) used burnt broken whole otoliths and determined a maximum age of 21 years. Using scales, the maximum readable age was only 9 years (Chakroun-Marzouk and Ktari 2003). Further differences in observed growth pattern may therefore be due, in part, to the calcified structure used or its preparation method.

In the growth parameters of the von Bertalanffy model obtained, the theoretical maximal length value is close to the observed largest size. S. umbra grows quickly during the first two years, attaining approximately 55% of its maximum length, but the annual growth rate drops rapidly thereafter. A similar result was found previously in the Gulf of Tunis, where males and females attained 50% of their maximum length during the first two years of life (Chakroun-Marzouk and Ktari 2003). In the Adriatic Sea, La Mesa et al. (2008) also reported a rapid increase in size during the first two to three years of life. Juveniles used all the energy available for somatic growth. High negative values of t₀ were reported in Maltese waters (Ragonese et al. 2004) and in the Adriatic Sea (La Mesa et al. 2008), probably reflecting the fast growth rate of juveniles.

The comparison of the growth parameter estimates of S. umbra from different geographical areas (Fig. 7, Table 3) shows substantial discrepancies. In the Gulf of Tunis, females were found to grow larger than males. The sexual dimorphism was reported too in the same area (Chauvet 1991, Chakroun-Marzouk and Ktari 2003), in Maltese waters (Ragonese et al. 2004) and in the Adriatic Sea (La Mesa et al. 2008). The estimates of the asymptotic length TL_∞ were lower than those found previously in the same area for both sexes (Chauvet 1991, Chakroun-Marzouk and Ktari 2003) but slightly higher in Maltese waters (Ragonese et al. 2004) (Fig. 7). In all the growth studies on S. umbra, the values of the Brody growth coefficient k confirm a slow attainment of maximal size, which is rather characteristic of long life cycle species. The high growth performance of S. umbra reported by Artüz (2006) in the Sea of Marmara was the result of the high Brody growth coefficient recorded in this locality, coupled to the high asymptotic length. This result seems to reflect, the growth of recruits and juveniles, although the length range of the studied sample is not reported.

These variations in growth performance between different regions are probably the result of the following factors:

(1) Variations in environmental conditions (such as temperature and food availability) among sampled areas.
(2) Different size distributions (probably caused by different types of sampling gear). In fact, growth model estimates are greatly affected by the lack of very young or old individuals and the unequal abundance of size classes (Mouine-Oueslati et al. 2015).
Distribution difference between sexes when there was sexual dimorphism as for this species.
(4) Mortality, which may affect the variability of the maximum length and has a relatively large and consistent impact on the size of structure of communities (Blanchard et al. 2005).
(5) A combination of the above factors.

Gulland (1971) suggested that a fish stock is optimally exploited when fishing mortality is equal to natural mortality. More recently, Pauly (1987) proposed a lower threshold value, E=0.4. Basing his suggestion on research by Beddington and Cooke (1983), Pauly (1987) proposed a lower optimum fishing mortality, F_opt=0.4M. (F>F_opt=0.12 for males and F>F_opt=0.09 for females). Estimates for females are considered more relevant for population modelling, because aspects of female population dynamics, including age-specific and total life-time fecundity and mortality, determine the potential increase in population numbers (Bishop et al. 2006). In the Gulf of Tunis, during the study period, the stock of S. umbra was slightly overexploited (E>0.4 and F>F_opt). On the Tunisian coasts, females reached ages of 31 years, which is the maximum age recorded in the Mediterranean Sea. The studied life history parameters of the population of the brown meagre of the Gulf of Tunis suggest that the species is long-lived, is slow growing and slow-maturing, and has a low natural mortality. Because of all these previous characteristic, S. umbra is a species vulnerable to inconsiderate fishing pressure. In the Gulf of Tunis, the brown meagre seems preserved from excessive fishing and particularly from spearfishing. Recent recreative fishing activity had a negative impact on its stocks on the northwestern coasts of the Mediterranean Sea, but it this is a very limited activity on the Tunisian coasts.

The results obtained from the current study provide an actualization of practical biologically related parameters for stock assessment and management of the brown meagre of the Gulf of Tunis. Regular follow-up of life history traits of this species are necessary for its optimal management on the Tunisian coasts.