Fish inhabiting artificial reefs have demonstrated reproductive success similar to or better than that of those living in natural reefs, probably due to the greater protection offered by environments of anthropogenic origin (Granneman and Steele 2014). Maricultures are examples of artificial environments that offer protection and foraging resources for fish (Goodbrand et al. 2013).

Mariculture structures alter the environment, resembling a reef environment. This enables the occurrence of species formerly absent at the site (Freitas and Velastin 2010). In addition to cryptic fish species, other organisms such as corals, bryozoans, sponges and echinoderms are attracted to these new consolidated environments. They are mostly suspensivores, thus changing the suspended material and consequently the turbidity (Thrush and Dayton 2002).

Hypleurochilus fissicornis (Quoy and Gaimard, 1824) is a small-sized cryptic marine species that inhabits consolidated environments such as coral and rocky reefs. The species is distributed in the western Atlantic from Vitória (Espírito Santo State, Brazil) to Mar del Plata (Argentina) and in the northeast Atlantic in the Azores (Portugal) (Menezes and Figueiredo 1985, Santos et al. 1997, Floeter et al. 2003). As in other Blenniidae fish, males take care of the nest where females spawn oocytes. The preferential sites for nests of H. fissicornis are rock crevices, empty shells, barnacles and even experimental fouling panels (Menezes and Figueiredo 1985, Omena and Souza 1999).

Small organisms such as blennies have the potential to dominate resources and control the functions of the system in which they live in high abundances, performing a fundamental role in the trophic chain of reef environments (Ackerman and Bellwood 2000). In natural coastal reefs and in artificial ones designed to increase local diversity, H. fissicornis is usually found in low abundances from northern Rio de Janeiro State to the south-central coast of Santa Catarina State (Brotto et al. 2007, Luiz-Jr. et al. 2008, Alves and Pinheiro 2011). On the other hand, in mariculture, the species is found in high abundances (Gerhardinger et al. 2004, Freitas and Velastin 2010), indicating that H. fissicornis probably reproduces more successfully in artificial sheltered environments than in natural coastal habitats.

Aiming to determine the strategy used by H. fissicornis in an artificial environment, the present study intended to (1) characterize the maturity stages of females and males from the development of germ cells (oogenesis and spermatogenesis), (2) identify the reproductive tactics to check for the reproductive success, and (3) investigate the use of cultivation structures in the reproductive process.

Guaratuba Bay (Fig. 1) has an area of 50.19 km² situated on the coastal plain of southern Brazil. The climate of the region is humid mesothermal. Data of air temperature and rainfall, furnished by the technological institute SIMEPAR, were used to characterize the study period. The highest mean monthly temperatures were recorded in January (25.8°C), February (26.1°C) and March (24.1°C), and the lowest in July (15.9°C) and August (16.3°C). The mean annual temperature was 21.0°C. Cumulative rainfall was 2020.6 mm, with the largest volume in June (220.2 mm) and the smallest in August (77.4 mm).

Monthly samplings were conducted between May 2013 and May 2014, in an oyster (Crassostrea brasiliana (Lamarck, 1819)) farm in the Pinheiros River (25°26’S; 49°17’W), near the mouth of the bay. Oysters were placed in mesh tubes, which were attached to a main cable held above the water by floats. These mesh tubes were covered with a cylindrical net (mesh 1 mm) by means of diving in order to prevent the escape of organisms, and brought to the surface following the methodology used by Freitas and Velastin (2010). Shells on which spawn was laid were also collected, as well as water temperature, pH, dissolved oxygen (DO) and salinity data (obtained using a multiparameter probe).

In the laboratory, measurements of total length (TL in cm) and total weight (TW in g) of specimens of H. fissicornis were taken. Gonads were removed, weighed (GW in g) and identified macroscopically for sex and gonadal development (according to Vazzoler 1996 and Brown-Peterson et al. 2011). Some gonads were subjected to routine histological analysis (fixed in ALFAC for 18h and embedded in Paraplast). Slides were stained with hematoxylin-eosin and analysed under a light microscope to characterize the maturity stage of females and males and to confirm the macroscopic identification of gonadal stages. The characterization of the development of female germ cells was based on Lowerre-Barbieri et al. (2013), while the maturity stage for females and males was prepared according to Vazzoler (1996) and Brown-Peterson et al. (2011). The sex ratio (monthly and by length classes) was analysed by a χ² test, with a degree of freedom of 1 and a significance of 0.05 (χ²> 3.84), according to Vazzoler (1996). The length classes (seven classes with a 1-cm range) were established in accordance with Sturges (Sokal and Rolf 1981).

The individual gonadosomatic index (GSI), expressed by the formula GSI=(GW/TW)100, was calculated to analyse gonadal development. To determine the maturation curve, the monthly mean GSI was calculated from the individual GSI of females and males, including juveniles of both sexes. After histological analysis of gonads, the distribution of the monthly percentage frequencies of gonadal development and the frequencies of young and adult individuals were estimated.

The weight-length relationship was calculated by TW=aTL^b, in which a is the linear coefficient and b is the slope (both determined using the least squares method) (Le Cren, 1951). From this, the total and somatic condition factors were calculated, expressed by K=TW/TL^b and K’=BW/TL^b, respectively (BW= body weight of the specimen subtracting the gonad weight).

Length at first maturity (L₅₀) was calculated by $$\mathrm{Fr}=1–\left(e^{–aL{m}^b}\right)$$ (Fr= relative frequency of adult individuals; ℮= base of the Napierian logarithm; a and b are coefficients estimated by the least squares method, transforming the variables involved; and L_m=midpoint of the length classes), according to Fávaro et al. (2003) and Oliveira and Fávaro (2011) in reproductive studies on estuarine fish.

Abiotic variables were summarized by a principal component analysis (PCA). Axes with higher eigenvalues were used to relate with gonadal development stages through Spearman correlation (α=0.05). The broken-stick method was used to select the most important axes. The seasons of the year were defined as autumn (April to June), winter (July to September), spring (October to December) and summer (January to March). These analyses were run using the R 3.1.0 software.

A total of 519 specimens were collected from May 2013 to May 2014 (205 females, 228 males and 86 of unidentified sex, including juvenile and adults with TL ranging from 1.2 to 5.4 cm). No specimens were collected in February and March 2014. TL varied from 2.0 to 7.8 cm (4.78±1.06 cm) and from 2.0 to 8.5 cm (5.64±1.50 cm) for females and males, respectively.

There were no significant differences in the sex ratio between the months, except January, when a larger number of males was found (χ²=4.89; df=1; p<0.05). The analysis of sex ratio by size classes showed differences between the sexes. Intermediate classes (3.2-4.2 and 4.3-5.3 cm) were dominated by females (χ²=5.12 and χ²=8.57; df=1; p<0.05) and the larger classes (6.5-7.5 and 7.6-8.6 cm) by males (χ²=27.55 and χ²=30.11; df=1; p<0.05) (Fig. 2).

Females showed an ovarian maturity scale in agreement with the pattern found in other teleosts, while males of H. fissicornis had a testicular gland associated with the testis, making the determination of male gonadal development more complex.

Unlike most teleosts, H. fissicornis have testes limited to a reduced peripheral part of the testicular gland, so the developing, spawning-capable and partially-spawned stages were difficult to characterize because these stages have all types of sperm cells (differing slightly in number of spermatozoa). This partial release of gametes is an important reproductive tactic because it reduces the probability of larval loss due to environmental fluctuations (Santos 1995).

The extensive reproductive period was also found for Coryphoblennius galerita on the Portuguese coast (Almada et al. 1996). However, Blenniidae species on the coast of Portugal and Spain showed a reproductive peak in summer (Almada et al. 1994, Almada et al. 1996, Carrassón and Baú 2003), whereas H. fissicornis, in southern Brazil, reaches greater reproductive activity in winter, as gonadal maturity is intimately related to the drop in water temperature. This difference in the reproduction period may be used by the species to avoid potential predators, since as youth of piscivorous fishes (as Epinephelidae) were found in the study area only in summer (personal observation). Other piscivorous fish species, such as Centropomus parallelus and Cynoscion sp., have been found in the surroundings of the study area (Gomes and Chaves 2006, Vendel et al. 2010). In addition, the disappearance of H. fissicornis specimens in February and March demonstrates that the bivalve farming is used by the species only as a reproductive site. Therefore, after reproduction, the individuals leave the area, as described for Salaria pavo (Almada et al. 1994). In addition, this behaviour of H. fissicornis also avoids predation in summer.

On the other hand, in colder waters such as those of Argentina, the reproductive period of H. fissicornis is in summer (Delpiani et al. 2012), and this situation may indicate that the ideal temperature range for gonadal development and larval development is found in winter in warm waters and in summer in cold waters. Temperature influences the maturation and growth rate of Parablennius ruber, which are accelerated with increasing temperature (Azevedo and Homem 2002). The relationship of temperature, latitude and reproductive period was also demonstrated by Fávaro et al. (2007) for Atherinella brasiliensis.

With regard to the use of empty shells, several egg development stages were observed in the same shell, indicating that more than one female may have used the same nest in different periods, spawning batches of ovarian follicles under the care of a single male. This strategy is used by Rhabdoblennius nitidus females, which choose their partners by preferentially repeating the choice of another female, in order to reduce the probability of desertion of the male (Matsumoto and Takegaki 2013). Considering that the ability of the male to care for the eggs is directly related to the probability of larval survival, subjecting spawns to several partners is advantageous (Santos 1995). Since the males need to defend their nests, their parental care may explain the longer length observed in males.

The species H. fissicornis was abundant in the bivalve farming studied, and the presence of young individuals for several months is indicative that the species has been reproductively successful at the site. In marine reef environments, the species is present in low abundances (Luiz-Jr. et al. 2008, Hackradt and Félix-Hackardt 2009, Alves and Pinheiro 2011), which may indicate that the farms, more sheltered in the estuary, offer greater protection during the reproductive cycle. However, since the samplings of the works cited above were carried out using visual censuses, cryptic and small-sized species can be underestimated (Ackerman and Bellwood 2000).

The highest frequency of young individuals was recorded in December, coinciding with the end of the reproductive season. The largest number of males in relation to females in January possibly indicates that females leave the breeding site earlier, as only males display parental care for the offspring.

The determination of length at first maturity for H. fissicornis pointed out that individuals with much reduced length enter into the reproductive process. Based on the L₅₀/L_max ratio, the values were 0.27 for the females and 0.23 for the males. This is characterized as an early first maturity, since the value of the L₅₀/L_max ratio is generally between 0.4 and 0.9 for different species (Beverton and Holt 1959).

Our results demonstrate that H. fissicornis has a variety of tactics to protect its offspring (e.g. batch spawning, long reproductive period, parental care, small length at first maturity and low-temperature reproductive period). Thus, the oyster farm is used as a breeding site by H. fissicornis in winter, ensuring the reproductive success of the species.