INTRODUCTION
In the Mediterranean Sea, the Gobiidae family has great specific and ecological diversity (Kovačić, Reference Kovačić2005) which explains the interest in their use as models in research on life-history evolution (Miller, Reference Miller, Potts and Wootton1984). However, having no market value, these fish are not targeted by commercial fishing gear, which makes their collection and regular monitoring difficult to carry out. Therefore, data on the reproduction of more than half of the species of the Gobiidae family from the Mediterranean Sea and north-east Atlantic are missing (Kovačić, Reference Kovačić2001).
Among the Gobiidae, the rock goby, Gobius paganellus (Linnaeus, 1758), is mainly an intertidal species (Miller, Reference Miller, Whitehead, Bauchot, Hureau, Nielsen and Tortonese1986). However, it is common in subtidal area to a depth of approximately 5 m (Dunne, Reference Dunne1978). In the latter area, the rock goby is one of the most abundant benthic species (Patzner et al., Reference Patznerr, Santosp, Re and Nashr1992). Thus, G. paganellus lives in shallow waters where it occurs on sheltered rocky shores and on muddy and sandy bottoms with much weed cover preferably in pools, under stones and clumps of algae (Miller, Reference Miller, Whitehead, Bauchot, Hureau, Nielsen and Tortonese1986; McDowall, Reference McDowall1997). It is a relatively large goby found along the Mediterranean, north-western Atlantic from western Scotland to Senegal, Black Sea, Indian Ocean and Red Sea (Miller, Reference Miller, Whitehead, Bauchot, Hureau, Nielsen and Tortonese1986, Reference Miller, Quero, Hureau, Karrer, Post and Saldanha1990; Engin & Seyhan, Reference Engin and Seyhan2009). In Tunisia, it is found in some north-east lagoons (Bizerta, Ghar El Melh, Ichkeul), in the Tunis lagoon (Menif, Reference Menif2000) and in the Gulf of Gabès (Ben Othman, Reference Ben Othman1973; Hajji et al., Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012). The reproductive biology of this species has been studied in some marine areas of the Atlantic (Miller, Reference Miller1961; Dunne, Reference Dunne1978; Faria & Almada, Reference Faria and Almada1995; Azevedo & Simas, Reference Azevedo and Simas2000), Black Sea (Engin & Seyhan, Reference Engin and Seyhan2009) and Mediterranean Sea (Miller, Reference Miller, Whitehead, Bauchot, Hureau, Nielsen and Tortonese1986; Hajji et al., Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012). Available data showed that the breeding season exhibited considerable geographical variations (Miller, Reference Miller1961, Reference Miller, Whitehead, Bauchot, Hureau, Nielsen and Tortonese1986; Dunne, Reference Dunne1978; Azevedo & Simas, Reference Azevedo and Simas2000; Engin & Seyhan, Reference Engin and Seyhan2009; Hajji et al., Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012). In addition, there is no information on the reproduction of this goby in Mediterranean coastal lagoon environments that constitute one of the most productive humid zones in the Mediterranean basin (Diawara et al., Reference Diawara, Tlig-Zouari, Rabaoui and Ben-Hassine2008) and have, according to most fishery biologists, a particular importance in Mediterranean fisheries production (Quignard et al., Reference Quignard, Mazoyer, Viane, Man Wai and Benharrat1983). Moreover, in comparison with the open sea, these typical brackish stretches of water are characterized, in both space and time, by a strong variability in their physicochemical parameters (Pérès, Reference Pérès1967; Skinner & Zalewski, Reference Skinner and Zalewski1995; Marzano et al., Reference Marzano, Liaci, Fianchini, Gravina, Mercurio and Corriero2003) that is considered as physiological obstacles to the species reproduction and is often advanced to explain the absence or weakness of lagoonal reproductive activity (Albaret, Reference Albaret, Durand, Dufour, Guiral and Zabi1994).
Among these lagoons, Bizerta lagoon is a south-Mediterranean brackish stretch of water, located on the north-east coast of Tunisia. Over the decade 1895–1905, this lagoon, was one of the richest fishing lagoons in the world (Salvator, Reference Salvator1900). Thus, statistics of fish production in this lagoon have shown 30 fish species of commercial interest (Zaouali, Reference Zaouali1979). In addition, from 1906, with the introduction of shellfish, it had an aquacultural industry. To this richness can be added the sedentary benthic fish, such as Gobiidae, that play an important role in the food chain by allowing energy transfers between the epifauna, on the one hand, and the fish-eating fish and birds, on the other hand (Costa, Reference Costa1988). These resident species constitute the prey of numerous migratory fish of commercial interest (Costa, Reference Costa1988). Furthermore, among the Gobiidae, G. paganellus, which appears amongst the most abundant Gobiidae species in Bizerta lagoon (Louiz, Reference Louiz2010), have a regulating role with respect to lagoon invertebrate communities that constitute an important component of its diet (Azevedo & Simas, Reference Azevedo and Simas2000). However, like most coastal lagoons which are, naturally, fragile environments, Bizerta lagoon constitutes the receptacle for the discharges of anthropogenic activities. The disturbances generated by these dischargs pose a threat to biodiversity and the balance of the ecosystem. These environmental disturbances require the management of space and species, mainly in areas where fish, such as Gobiidae, are born and die. This management requires knowledge of the biology of these species, including that of reproduction, to lead to the development of effective conservation measures. Indeed, reproductive parameters of fish species are part of basic criteria for assessing the status of stocks (Jakobsen et al., Reference Jakobsen, Fogarty, Megrey and Moksness2009) and estimating their productive potential. Consequently, knowledge of the rock goby can contribute to the diagnosis of the functional state of the environment and, therefore, to the implementation of a rational approach of its protection. Furthermore, it will allow, by comparison with previous data on the marine environment and species reproduction, appreciation of the impact of lagoonal life on the reproductive parameters of G. paganellus. All these reasons prompted us to take an interest in studying the reproduction of G. paganellus in Bizerta lagoon. Thus, in the present research, we describe various aspects of the demographic and reproductive biology of this species by analysing sex-ratio, size at sexual maturity, gonad annual cycle and the impact of some environmental factors on the sexual cycle. In addition the hepatosomatic index and the condition index were also analysed to obtain information on the general physiological status of the animals. A histological study of male and female gonads was also performed to identify the main phases of their evolution.
MATERIALS AND METHODS
Site of collection
Goby sampling sites were located in Bizerta lagoon (Figure 1). This lagoon extends for about 150 km2 and its depth varies from 0.5 to 12 m with an average depth of 7 m (Ghrabi et al., Reference Ghrabi and Yoshida2002). Bizerta lagoon is connected to the Mediterranean Sea by a straight canal of 7 km; it is also connected to Ichkeul lagoon by the narrow channel of Oued Tinja. The exchanges of water with the Mediterranean Sea and Ichkeul lagoon determine the average salinity of the lagoon, which varies between 32.5‰ and 38.5‰. The average water temperature ranged from 13°C to 32.5°C, respectively, during the winter and summer of 2006. Superficial deposits of the lagoon are composed of three facies: a sandy facies, a silty to silty clay facies and a clay facies (Soussi et al., Reference Soussi, Levy and Zaouali1983). Bizerta lagoon bottom is carpeted by seagrass meadows of Cymodocea nodosa and the algae Caulerpa prolifera (Frisoni et al., 1886).
Fig. 1. Location of the sampling sites in the Bizerta lagoon (north-east Tunisia).
The discharges of anthropogenic activities, following the demographic growth, led to destabilization of this ecosystem (Louiz et al., Reference Louiz, Menif, Ben-Attia and Ben-Hassine2007, Reference Louiz, Kinani, Gouze, Ben-Attia, Menif, Bouchonnet, Porcher, Ben-Hassine and Aït-Aïssa2008, Reference Louiz, Ben-Attia and Ben-Hassine2009) as evidenced by the decrease in the amounts of fish caught. Thus, it is important to have better understanding of the influence of the environmental conditions of this lagoon on the reproduction of Gobius paganellus.
Sampling
A total of 1489 gobies belonging to G. paganellus were collected, from January 2005 to December 2006, at a transect of 0.5–4 m depth in five sites (Figure 1). Indeed, in order to cover the maximum of the stock, the sampling stations are located at shallow depths where the species is common, even abundant (Dunne, Reference Dunne1978; Patzner et al., Reference Patznerr, Santosp, Re and Nashr1992). Moreover these sites are, like the rest of the lagoon, covered with vegetation and are representative of the sedimentary facies of the brackish stretch of water. Fish were caught by means of a beam trawl like the one used by Magnhagen (Reference Magnhagen1990) and by Menif (Reference Menif2000) to capture Gobius niger, especially as the capture of such species on the bottom covered by seagrass meadows requires the use of trawling which is the most appropriate mode of capture (Harmelin-Vivien & Francour, Reference Harmelin-Vivien and Francour1992). This gear, with 8 mm mesh size, is frequently used by Tunisian fishermen, who call it Gargamellou, to capture benthic fish in shallow waters such as in Bizerta lagoon. Samples were collected in the third week of each month with the intention of fishing 60 individuals monthly (4–132 animals per month, average 55, standard deviation 40.1). Specimens for biological studies were frozen until the day of accomplishing the measurement. Only live gobies were processed for histology immediately after being caught. Environmental parameters such as temperature, salinity and dissolved oxygen were recorded.
Collected data and methods of study
Each specimen was measured for total length (TL), standard length (SL) (±0.01 mm) using a digital caliper. A monthly sub-sample of specimens for each 5 mm length-class was selected. Each of these was sexed by external examination of the urogenital papilla (Miller, Reference Miller, Potts and Wootton1984; Bouchereau & Marques, Reference Bouchereau and Marques1998) and by macroscopic observation of the gonads. Total body wet weight (TW), eviscerated body wet weight (EW) (±0.01 g), liver weight (LW) and gonad weight (GW) (±0.001 g), were also measured. Gonad observation and weighing were not generally possible for smaller specimens (<5 cm TL).
The changes in sex proportion were analysed according to months and size (5 mm length intervals). Gender proportion was calculated using the formula:
$${\rm proportion\,of\,males} \lpar \% {\rm M} \rpar = {\rm M} \times 100/\lpar {\rm M} + {\rm F}\rpar$$
where M is the number of males, F is the number of females, and M + F is the total number of males and females.
The gonadosomatic index (GSI) was calculated as
$${\rm GSI} = \left({{\rm GW}/{\rm EW}} \right)*{\rm 100}$$
Accumulation and depletion of reserves in the liver and the muscles of the G. paganellus in Bizerta lagoon were studied through the analysis of hepatosomatic index (HSI) and the condition index (K). These indices were calculated as follows:
$${\rm HSI} = \left({{\rm LW}/{\rm EW}} \right)\times {\rm 100}\rpar$$
$${\rm K} = \left({{\rm EW}/{\rm TL}^3 } \right)\times {\rm 100}\rpar$$
Gonads were macroscopically staged according to the gonad development classification for gobies based on a five point scale of maturity (Miller, Reference Miller1961).
-
• Stage I: immature, beginning of gametogenesis; gonads not developed and sex only distinguishable by the shape of the genital papilla.
-
• Stage II: developing and recovering spent.
-
• Stage III: maturation; gonads developed and made of a compact tissue, white streamed line in males and yellow granular in females.
-
• Stage IV: ripe, reproductive stage; inflated and ripe gonads, for females and for males.
-
• Stage V: spent; ovary and testes empty, loose, red. Few remains of eggs in resorption process.
The size at sexual maturity and the mean size at which 50% of fish were sexually mature (L50) were determined in fish sampled during the breeding season to avoid classifying individuals in sexual rest as immature (Atse et al., Reference Atse, Konan and Kouassi2009). The proportions of mature (female, male) were estimated for reproductive active fish (Stages III–V) per 5 cm size-class, for pooled data during two years by fitting the sigmoid, logistic curve (Chen & Paloheimo, Reference Chen and Paloheimo1994). The equation used is as follows:
$$Px = {\rm 1}/{\rm 1} + {\rm e}^{ - {\rm r}\lpar \left({{\rm TL } - {\rm L5}0} \right)}$$
where r is a fitted constant and P x is the proportion of the mature fish at length TL.
Histology
Both male and female gonads were dissected and quickly fixed by immersion in alcoholic Bouin's solution, dehydrated in graded ethanol solution and embedded in paraffin. Embedded tissues were cut at 7 µm sections for ovaries and at 5 µm for testes. The paraffin was removed through a toluene series. Tissues were rehydrated and stained with haematoxylin and eosin as described in Martoja & Martoja (Reference Martoja and Martoja-Pierson1967). Histological analysis was conducted in ten males and in nine females. 20 histological sections were prepared from each gonad. Sections were photographed by means of an Olympus (CX41) microscope provided with a numerical camera (Olympus SP-51OUZ). The gonad tissue was examined in terms of cell types, based on the descriptions of Wallace & Selman (Reference Wallace and Selman1981).
Statistical analysis
The fit of the distributions to data sets was assessed by a dithered Kolmogorov–Smirnov goodness-of-fit test. The homogeneity of variances was checked by Levene test. The comparison of means was carried out using univariate analysis of variance (ANOVA) followed by the post-ANOVA test of Gabriel. If the normality assumption is not verified, statistical significance was tested by using Kruskal–Wallis one-way ANOVA on ranks. Differences in the percentage were analysed by the Chi-square test. Pearson's linear regression was used for correlation analyses. Statistical analyses were conducted using GraphPad InStat and GraphPad Prism v. 3.0a for MacIntosh (GraphPad Software, San Diego, CA, USA). The comparison of the TL50 values for males and females was carried out using a specific toxicological data processing program REGTOX (v. EV7.0.5) (Vindimian et al., Reference Vindimian, Garric, Flammarion, Thybaud and Babut1999) based on models of linear and non-linear regression. Goodness of fit was assessed by the sum of squares F-test and standard error of the TL50 value. All statistical tests were two-tailed, and a P value of 0.05 or less was considered statistically significant.
RESULTS
Characteristics of the overall examined samples
A total of 1486 individuals were examined, 660 were males, 826 females and three small individuals (38 mm TL) were not sexed because the urogenital papilla was not individualized. Sexual dimorphism shown by the morphology of the urogenital papilla does not presuppose that individuals are adults. In order to get a monthly larger sample set we combined the two year samples. The size (TL) range of fish was from 48 to 122 mm in males and from 39 to 125.3 mm in females (Table 1) (Figure 2).
Fig. 2. Length–frequency distribution of males and females of Gobius paganellus collected in Bizerta lagoon.
Table 1. Biometric data of Gobius paganellus collected in Bizerta lagoon during the years 2005 and 2006. N, number; TL, total length; SL, standard length; EW, eviscerated weight.
Sex-ratio
The total number of female samples was greater than that of males. The estimated sex-ratio (males:females) for all samples was 0.79:1, which was significantly different from unity (χ2 = 18.54, P < 0.001). Monthly sex-ratio fluctuated from 0.3 to 1.06 (mean, 0.78, standard error, 0.26). The proportion of females was higher than that of males in January, February, June, July and in November (Table 2). It was noted that sex-ratio is in favour of females for size-class between 65 and 85 mm (P < 0.05. The males were significantly more numerous than females over the range from 115 to 130 mm (P < 0.05) (Table 3).
Table 2. Monthly variation of the sex proportion (%) of Gobius paganellus in the Gulf of Gabès. N, number; significance: *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Table 3. Evolution of sex proportion according to the size (total length, TL) in Gobius paganellus. N, effective; *, denotes significant difference (*, P < 0.05; **, P < 0.01; ***, P < 0.001).
Gonad annual cycle
GONADOSOMATIC INDEX (GSI)
The monthly variation in the female GSI (both years combined) showed a stage of pre-maturation that spreads from October (0.6) to November (2.9), followed by a phase of fast maturation marked by a considerable increase of the GSI from November to December (7.8). However, there is no statistical difference observed between GSI values from December (7.8), January (9.3) and February (8.3) (Gabriel post-ANOVA test). The spawning phase is characterized by a decrease of the GSI which occurs from January (9.3) to March (3.9) (Figure 3A). Finally the sexual rest extended from April (0.4) to September (0.24), and is marked by a relatively constant GSI.
Fig. 3. Monthly evolution of (A) the gonadosomatic index (GSI); (B) the hepatosomatic index (HSI); (C) the condition index (K) for males (M) and females (F) of Gobius paganellus from Bizerta lagoon (mean ± SEM) over the period January 2005–December 2006.
The males showed different evolution patterns of GSI to females, and the maximum GSI mean of females (10) was higher than that of males (0.6) (Figure 3A). The phase of fast maturation of gametes was located between October and November. The highest values were observed in February (0.62) (Gabriel post-ANOVA test). The spawning period occurred in February–March and the sexual inactivity phase spread from April to August.
HEPATOSOMATIC INDEX (HSI)
The HSI of both sexes showed a clear seasonal pattern inversely related to the GSI (Figure 3B). The HSI gradually increased from April to August and reached its maximum at 6.82 and 5.25 for males and females, respectively. In December, the beginning of spawning of this species, this index showed a decrease and reached its minimum in March (2.1 for males and 3 for females).
CONDITION INDEX (K)
Temporal trends of K yielded a weak apparent seasonal variation in males (Figure 3C). In females, a significant decrease appeared in spawning phase, from January (2) to March (1.8) (Gabriel post-ANOVA test). The condition index gradually increased from March and reached its maximum in May (2.1). A second fall was recorded during the hot season and reached a minimum in August (2%).
Impact of some environmental factors on sexual cycle
Periodic correlation modelling of females GSI and environmental parameters showed a periodic variation of photoperiod (R 2 = 0.99) (Figure 4A), temperature (R 2 = 0.82) (Figure 4B) and salinity (R 2 = 0.25) (Figure 4C). We demonstrated, as well, a negative regression between GSI and temperature (P < 0.0001), on the one hand and between the GSI and photoperiod (P < 0.0001), on the other hand. Furthermore the evolution of the GSI of this species was positively correlated with salinity, conductivity and dissolved oxygen in water (P < 0.001) (Table 4).
Fig. 4. Periodic regression modelling of GSI (solid line) in females of Gobius paganellus and environmental parameters (dashed): (A) photoperiod; (B) temperature; (C) salinity.
Table 4. Relation between environmental parameters and males and females gonadosomatic index of Gobius paganellus from Bizerta lagoon. T, temperature; Phot, photoperiod; Sal, salinity; Cond, conductivity; Oxy, dissolved oxygen. r, Pearson correlation coefficient; *denotes significant correlation with Pearson's test (*, P < 0.05; **, P < 0.01; ***, P < 0.001).
Gonad structure
Histological sections of testes from specimens captured in October showed that lobules were subdivided into cysts where germ cell maturation takes place. Cysts contain mainly spermatocytes and spermatids and a glandular mass (GM) homologue of Leydig cells (Figure 5A). This glandular tissue appears also in small clusters between seminiferous tubules (Figure 5B). Seminiferous tubules are enclosed by myoid cells forming a thin interlobular tissue by Sertoli cells and blood vessels. In a mature individual examined in February, different stages of spermatogenic were observed. In fact, the development of male germ cells takes place centripetally in cysts (Figure 5C, D). Spermatogonia, located at the periphery of the cysts, will evolve into spermatocytes and then spermatids and sperm in the lumen of the cyst. During this phase the glandular mass represents about a third of total mass of the organ. In late breeding season (March) seminiferous tubules began to be emptied of sperm cells and to be reabsorbed to give way for the glandular mass (Figure 5E, F). During the sexual rest (May), histological sections of testes showed two parts: one part seminiferous and the other glandular (Figure 5I, J). The glandular part is the closest to mesorchium. This phase of relative inactivity extends from April to September.
Fig. 5. Histological sections of Gobius paganellus testis: (A, B) Stage II: testis in the beginning of maturation (October, TL = 80 mm); (C, D) Stage III: mature testis (February, TL = 79 mm); (E, F) Stage III: testis in the end of reproduction period (March, TL = 95 mm); (I, J) Stage V: spent (May, TL = 112 mm). ST, seminiferous tubes; MG, glandular mass composed by lending cells; SP, sperm; white arrow, spermatocytes; black arrow, Sertoli cell; Ps, seminiferous portion.
Four stages of ovarian development were also described. Females collected in October showed early gonad development containing primary oocytes and cortical alveolar follicles (Figure 6A). In December gonads are in more advanced maturing phase at which there is a follicle formation and a development of cortical alveolar oocytes in early vitellogenesis. Ovaries contain previtellogenic oocytes in various stages of development (Figure 6B). In January, at maximum gonadal growth, histological examination of ovaries revealed structure with two synchronous populations of oocytes: one predominant population relatively synchronized of mature oocytes (vitellogenic oocytes) and a second lower population previtellogenic oocytes dominant and early vitellogenesis (Figure 6C). During periods of post-spawning in April, ovaries contain remains of primary oocytes (PO) which have not completed their maturity and atretic follicles (AF). The ovarian wall becomes thicker (Figure 6D).
Fig. 6. Ovarian structures in the females of Gobius paganellus: (A) Stage II: developing ovary (October, TL = 75 mm); (B) Stage III: ripping ovary (December, TL = 84 mm); (C) Stage IV: mature ovary (January, TL = 75 mm); (D) Stage V: ovary in post-spawning (April, TL = 103 mm). OP, primary oocyte; FAC, cortical alveolar follicle; OV, vitellogenic oocyte in maturation phase (early vitellogenic oocyte); OVm, mature vitellogenic oocyte (hydrated oocyte); FA, atresic follicle.
In ripe ovaries (Stage IV), hydrated oocytes of G. paganellus from Bizerta lagoon ranged from 544 µm to approximately 1022 µm in diameter (799.5 ± 114.5 µm) (Figure 6). Results of oocyte size–frequency distributions, constructed for five ripe fish, showed two populations of oocytes in the ovary at one time (Figure 7), a fairly synchrounous population of hydrated oocytes and a more heterogeneous population of smaller oocytes (Figure 6C).
Fig. 7. Oocyte size–frequency for Gobius paganellus ripe ovary.
Size at sexual maturity
During our study, mature gonads were found only in males with total length equal to or greater than 70 mm (SL = 55 mm) (Figure 8A) and the smallest mature female was 58 mm total length (SL = 45 mm) (Figure 8B). Distribution of GSI in adult females during the breeding period, is entirely independent of size (Kruskal–Wallis test, P < 0.05). For small size-classes below 45 mm, females have less than 3% GSI and thus seem immature (Figure 8B).
Fig. 8. Length–GSI relationship for (A) males and (B) females. Broken line marks presumptive size at first sexual maturation.
The proportion of individuals per total length group that had reached maturity, plotted against size, formed a logistic curve. TL50% values were found to be 78.3 mm (95% confidence interval 77–84 mm) for males and 79 mm (95% confidence interval 75–85 mm) for females (Figure 9; Table 5). There is no statistical differences observed between the TL50 values of males and females (P = 0.92, F test).
Fig. 9. Length at sexual maturity (TL50) of Gobius paganellus from Bizerta lagoon.
Table 5. The variation of the proportions of mature individuals (%) according to the size.
DISCUSSION
Reproduction biology of Gobius paganellus has so far never been studied in Mediterranean lagoons. In this study, the total length of the specimens recorded in Bizerta lagoon ranged between 48 and 122 mm in males and between 39 and 125.3 mm in females. The minimum size collected in this lagoon was greater than that of the Atlantic population (20 mm) (Azevedo & Simas, Reference Azevedo and Simas2000 ) and smaller than that of the Black Sea (50 mm) (Engin & Seyhan, Reference Engin and Seyhan2009) and the southern Tunisian coast, Gulf of Gabès population (90 mm) (Hajji et al., Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012). This discrepancy could be explained by the nature and therefore the selectivity of fishing gears, especially in the case of the Gulf of Gabès. Indeed, the fish examined by Hajji et al. (Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012) were collected from the traditional fishery, which usually uses trammel nets with a mesh diameter higher than that of the beam trawl that we used to catch gobies within the Bizerta lagoon. However, the maximum size of both sex specimens was smaller than that of the Atlantic (130 mm) (Azevedo & Simas, Reference Azevedo and Simas2000), Black Sea (140 mm) (Engin & Seyhan, Reference Engin and Seyhan2009) and southern Tunisian coast, Gulf of Gabès population (143 mm) (Hajji et al., Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012). This difference could be explained by differences in ecological conditions of the prospected regions and by the rigour of the lagoon ones. Indeed, harsher environmental lagoonal conditions can have an impact on the fish life expectancy. In contrast, the distribution pattern of habitat in Bizerta lagoon, as types of substrate (soft) and vegetation (seagrass meadows of Cymodocea nodosa and the algae Caulerpa prolifera), is homogeneous (Frisoni et al., Reference Frisoni, Guelorget, Pertuisot and Fresi1986). In addition, the depths of the lagoon are not, in general, important (7 m average). In this case, a different bathymetric distribution, depending on the size of G. paganellus, seems unlikely in this environment.
The analysis of the sex-ratio showed significant difference between the proportion of males and those of females of the entire sample, according to the months and size. The population of G. paganellus from Bizerta lagoon is characterized by a significant predominance of females that corresponds to a sex-ratio different from the hypothetical ratio 1:1. This result is different from that reported by Hajji et al., (Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012) for the Gulf of Gabès population where there is a significant predominance of males. In the Black Sea, the sex-ratio of G. paganellus population was not significantly different from unity (Engin & Seyhan, Reference Engin and Seyhan2009). According to Miller (Reference Miller, Potts and Wootton1984), the sex-ratio is generally balanced in gobies. However, a predominance of one sex over the other was mentioned by other authors (Kvarnemo, Reference Kvarnemo1994; Kvarnemo et al., Reference Kvarnemo, Forsgren and Magnhagen1995; Azevedo & Simas, Reference Azevedo and Simas2000). Engin & Seyhan (Reference Engin and Seyhan2009) have captured fish by freediving, using spear gun and hand net. This sampling method allowed them to obtain a homogeneous sample rate. The monthly changes in the numerical proportions of sexes, revealed a predominance of females during the spawning period. But after spawning and during the period of sexual rest, males become more numerous, with the exception of June and July. This result is different from those found by Hajji et al. (Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012) in the Gulf of Gabès, where males are favoured in October, November and December. During the breeding season gobies of both sexes exhibit different behaviours (ethological). Males defend the nest while females are very active and move around a lot, which makes them more vulnerable to the fishing gear used. Thus, the benthic trawl cannot easily catch the breeding males, which, according to Miller (Reference Miller, Potts and Wootton1984), guard the eggs in nests. Therefore, males predominate before and after spawning. The decrease in the proportion of males during June and July, corresponding to the post-spawning period, can be explained by an eventual mortality. Indeed, during the breeding season males provide a major effort accompanied by a large energy consumption for maintenance and guarding nests of eggs. The observed difference in the monthly sex-ratio variations between Bizerta lagoon and the Gulf of Gabès is probably due to the selectivity of the artisanal fishing gears used by Hajji et al. (Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012) whose wide mesh holds back the large-sized gobies where males dominate.
In Bizerta lagoon, a predominance of males in large fish (TL >12 cm) was noted. The predominance of males in the largest size-class has already been observed by Hajji et al. (Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012) in the Gulf of Gabès (TL >12 cm) and by Azevedo & Simas (Reference Azevedo and Simas2000) in São Miguel (TL > 10 cm). Several hypotheses can be put forward to explain the imbalance in the sex-ratio towards males and the relationship sex-ratio/size observed. Thus, this imbalance can be attributed to differential growth between sexes (Bruslé & Quignard, Reference Bruslé and Quignard2004), a different life expectancy or behavioural events, feeding and spawning grounds, that may introduce a bias in sampling.
The study of the reproductive cycle of G. paganellus in Bizerta lagoon revealed that it is a single fish egg with a marked seasonal character. The gonad ripening period extends from October to December, while the spawning period occurs from December–January to March (winter–early spring) that is characterized by short days and low temperatures (present study). Indeed, there is no statistical difference observed between GSI values from December, January and February (Gabriel past-ANOVA test). The sexual inactivity phase extends from April to September. Our findings show that the increase in GSI is parallel to the increase in salinity; however, the maximum laying does not correspond to maximum salinity. In addition, Pearson correlation coefficients showed that temperature and photoperiod play a more important role than salinity and dissolved oxygen in the maturation of gonads in G. paganellus. In fact, the close interaction between temperature and photoperiod is known for its natural role in the maturation and spawning fish (De Vlaming, Reference De-Vlaming1972, Reference De-Vlaming1975). The results of anatomical gonads showed that gonad maturation of G. paganellus in Bizerta lagoon starts in October. The spawning period, which takes place from December to March, seems earlier than that observed in the marine environment of the Gulf of Gabès (Tunisian south coast). In fact, for G. paganellus from the Gulf of Gabès (Hajji et al., Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012), the spawning period occurs from January to March. This difference seems to be related to the lagoonal conditions, especially those of temperature, that are harsher than those of the open sea. Furthermore, the spawning of this species in Bizerta lagoon is earlier than those described in other works (Figure 10) and ends earlier than in these sites. Then, reproduction of this species occurs between January and May in Ponta Delgada, Azores (Azevedo & Simas, Reference Azevedo and Simas2000), between January and June in Italian Mediterranean waters (Naples) (Miller, Reference Miller, Whitehead, Bauchot, Hureau, Nielsen and Tortonese1986) and between December and June in Portugal, Parede (Faria & Almada, Reference Faria and Almada1995). According to Miller (Reference Miller1961), Dunne (Reference Dunne1978) and Engin & Seyhan (Reference Engin and Seyhan2009), the reproduction of this species in the Irish Sea (Isle of Man), Ireland (the shores of Carna) and the Black Sea (Varna) is strictly in spring and it unfolds, respectively, from April to June, March to June and March to May (Figure 10).
Fig. 10. Representation of the spawning season of Gobius paganellus according to the latitude of various localities. Irish Sea, Isle of Man (Miller, Reference Miller1961), the shores of Carna Ireland (Dunne, Reference Dunne1978); Black Sea, Varna (Engin & Seyhan, Reference Engin and Seyhan2009); Italy, Naples (Miller, Reference Miller, Whitehead, Bauchot, Hureau, Nielsen and Tortonese1986); Portugal, Parede (Faria & Almada, Reference Faria and Almada1995); Ponta Delgada, Azores (Azevedo & Simas, Reference Azevedo and Simas2000); Gulf of Gabès, Tunisian south coast (Hajji et al., Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012) and Bizerta lagoon (present study).
These discrepancies between spawning periods observed in various localities could be explained by ecological conditions, including photoperiod and temperature that decrease from the equator to the North Pole. In fact, in the southern part of the distribution area of this species there have been less harsh climatic conditions than in northern areas, allowing the extension of the reproduction period and its shifting to the winter months. The set of observations showed a difference of the breeding season between the Mediterranean Sea and the Atlantic, between the two Mediterranean basins, Western and Oriental, and between lagoon and sea environments. Hydroclimatic factors, including temperature and salinity, could be the cause of these variations. Indeed, ecological factors can fluctuate between regions but also between sea and lagoons, according to the currents and depths (Augier, Reference Augier1973; Astraldi et al., Reference Astraldi, Balopoulos, Candela, Font, Gacic, Gasparini, Manca, Theocharis and Tintore1999; Drira et al., Reference Drira, Hamza, Belhassen, Ayadi, Bouaïn and Aleya2008, Reference Drira, Bel Hassen, Hamza, Rebai, Bouain, AyadI and Aleya2009), that have an effect on the reproductive biology of fish (Ben Jrad et al., Reference Ben Jrad, Fehri-Bedoui, Ben-Slama and Ben-Hassine2010). It is interesting to report the observations of De-Vlaming (Reference De-Vlaming1972) and Moiseyeva & Rudenko (Reference Moiseyeva and Rudenko1979) which showed that compared with photoperiod, temperature has a greater importance on gonadal development and reproduction of various Gobiidae species. In addition, a genetic characterization of G. paganellus populations could contribute to a better understanding of the observed discrepancies between spawning periods.
Histological study of G. paganellus gonads allowed a description of their structure during all reproduction phases. In fact in males we confirmed the centripetal maturation of sexual cells and the presence of a glandular mass. According to Colombo & Brurighel (Reference Colombo and Burighel1974), the glandular mass is the homologue of Leydig cells in the testis of Gobius genus. It is situated along the mesorchium and it is distinct from the seminiferous area. This cell type is also found in small groups within the interlobular space in G. paganellus (Stanley et al., Reference Stanley, Chieffi and Botte1965).
Oocyte maturation is consistent with published description of the cystovarian type of teleost oocyte maturation (Wallace & Selman, Reference Wallace and Selman1981). Ovaries architecture in this species shows a group-synchronous structure with a single clutch of hydrated oocytes and a more heterogeneous population of smaller oocytes which will be spawned in future breeding seasons (Murua & Saborido-Rey, Reference Murua and Saborido-Rey2003). This architecture is generally characteristic of iteroparous species having only one annual egg production with a relatively short spawning season and where the yolk accumulation mostly depends on body reserves (Murua & Saborido-Rey, Reference Murua and Saborido-Rey2003).
In order to better understand energetic transfers in G. paganellus from Bizerta lagoon, we analysed the evolution of HSI and K based on the sexual cycle. Indeed, the relationship between seasonal variations in HSI and other parameters such as the GSI can help to assess the physiological condition of the animal and duration of the gonads maturation (Kume et al., Reference Kume, Furumitsu, Tanaka and Yamaguchi2009). The results obtained showed an increase of energy storage in the liver for two sexes, in the post-reproductive period, from May to July. Then a notable decrease in liver mass has been observed at the beginning or during the breeding season (November). The HSI of females is slightly higher than that of males. Hajji et al. (Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012) reported, in both sexes, a gradual increase of HSI from August to November. These differences, in the mobilization and use of hepatic reserves, between a lagoon and a marine environment on Tunisian coasts, seem to fit a diversity of reproductive strategies that could characterize the different populations, marine and lagoonal, of G. paganellus. Furthermore, the condition factor (K) showed a rather apparent seasonal cycle in the two sexes; however females had the lowest levels by contrast with males. The amount of somatic lipids can depend largely on the different levels of reproductive investment shown by the breeding individuals (Malavasi et al., Reference Malavasi, Fiorin, Franco and Torricelli2004). The low variation of (K) observed in this study, in both sexes, suggests that this coefficient is not affected by maturation of genital products.
In the present study, well-developed gonads were found only in specimens with total length equal to or greater than 58 mm (females) or 70 mm (males). Lengths at sexual maturity, expressed as L50, were found to be 78.3 mm for males and 79 mm for females. These results confirm those mentioned in Azores (Atlantic) for the same species (Azevedo & Simas, Reference Azevedo and Simas2000). According to these authors, the size at first maturity, expressed as total length in G. paganellus, is 70 mm in males and 60 mm in females, that corresponds to individuals aged about one year (Azevedo & Simas, Reference Azevedo and Simas2000). This contrasts with the findings of Miller (Reference Miller1961) in the Isle of Man (Irish Sea) and Dunne (Reference Dunne1978) in Carna (Ireland), which indicate a size at first maturation from 2 to 3 years old. It is the same for G. paganellus from the Gulf of Gabès (Hajji et al., Reference Hajji, Ouannes-Ghorbel, Ghorbel and JarbouI2012) which attains sexual maturity at a size of 103.7 mm for females and 114.4 mm for males. However, a smaller size, at first sexual maturity (L50m = 52 mm; L50f = 55 mm), was observed in the Black Sea (Engin & Seyhan, Reference Engin and Seyhan2009). This difference could be explained by ecological conditions of lagoons and the Black Sea, which are different from the open sea. Therefore, to be mature at a young age may be a strategy of adaptation of gobies to harsh lagoonal conditions and to their short life expectancy, especially in the anthropized environments, as in Bizerta lagoon. More studies should be undertaken to obtain more information on life traits of this fish.
ACKNOWLEDGEMENTS
The authors are grateful to the anonymous referees whose suggestions and comments improved the submitted manuscript.
