The influence of a large reservoir on the reproductive activity of the white piranha , Serrasalmus brandtii ( Lütken , 1875 ) in Southeast Brazil

In order to assess the influence of the Três Marias dam on the reproduction of the white piranha, Serrasalmus brandtii, 1569 specimens captured from June 2011 to May 2012 in three sections of the São Francisco River basin were analyzed: section 1 = Três Marias reservoir (TMR); section 2 = São Francisco River (SFR) segment immediately downstream of TMR; and section 3 = SFR segment at the confluence with the Abaeté River. Total length (TL), body weight (BW), gonadosomatic index (GSI), and gonadal maturation stages were determined for each specimen, while vitellogenic follicle diameter, and absolute (AF) and relative (RF) fecundities were also determined for females. Reproductive activity, including spawned females and spermed males, was recorded throughout the year in the three river segments. Higher frequencies of females and males at the maturing/mature stage were found in the November/December and January/February, which coincided with the period of higher temperature, rainfall, and dissolved oxygen concentration. Mean GSI and vitellogenic follicle diameter were higher in sections 1 and 3 than in section 2 (P <0.05), while AF and RF were higher in section 1 (P <0.05) than in sections 2 and 3. The results of the present study showed that S. brandtii had lower reproductive performance in section 2, probably due to thermal disturbances caused by the dam, thus confirming the influence of dams on the reproduction of fish downstream from hydroelectric plant reservoirs.


Introduction
Most of the world's major rivers have some kind of damming that causes negative impacts to entire fish communities (Olden & Naiman 2010).Although important for economic development, dams cause serious and irreversible impacts to the natural hydrological regime of rivers including altering the downstream flow (Agostinho et al. 2010, Nunes et al. 2015).Such impacts have been detected downstream of several hydroelectric plants in different parts of the world (Clakson & Childs 2000, Todd et al. 2005, Olden & Naiman 2010).In the São Francisco River in the Southeast Region of Brazil, negative impacts of the Três Marias dam on the reproductive activity of the ichthyofauna have been detected in migratory species, such as Prochilodus argenteus (Arantes et al. 2010), Leporinus reinhardti (Weber et al. 2013) and Brycon orthotaenia (Nunes et al. 2015), and foraging species, such as Astyanax fasciatus and A. bimaculatus (Normando et al. 2014), but there have been no studies evaluating impacts on typically sedentary species or species pre-adapted to lentic environments such as Serrasalmus brandtii.
Among Brazilian carnivorous fish, piranhas and pirambebas exhibit a higher degree of agility in catching prey than other species due to their voracity, especially when they inhabit lentic waters (Braga 1981).The white piranha, S. brandtii, is native to Brazil and endemic to the São Francisco River basin.It belongs to the Serrassalmidae family (Mirande 2010), which includes piranhas and pirambebas, which are voracious carnivorous freshwater fish that are restricted to South America and occur in Brazilian reservoirs (Jégu 2003).Males of the species perform uni-parental care and the species is pre-adapted to lentic environments (Braga 1981, Agostinho 2003).
Considering that studies related to the reproduction of S. brandtii have only been conducted in lentic environments (Honorato-Sampaio et al. 2015), the objective of the present study was to comparatively analyze the reproduction of S. brandtii in three sections of the São Francisco River basin -the Três Marias reservoir, a lentic environment, and two lotic sections of the São Francisco River -and to evaluate the impact of the dam on the reproduction of the species.

Study area
A total of 1569 specimens of S. brandtii were captured in three sections of the São Francisco River (SFR) basin: section 1 = Três Marias reservoir (TMR) (18º23'27" S, 45º13'12" W); section 2 = SFR in the first 34 km downstream of the TMR (18º07'59" S, 45º14'01" W); and section 3 = 34 to 54 km downstream of the TMR, after the confluence of the SFR with the Abaeté River (18º00'49" S, 45°10'51" W) (Figure 1 and Table 1).Fish were sampled bimonthly from June 2011 to May 2012 using gill nets with meshes ranging from 3.0 to 7.0 cm between opposite knots.The fish, if still alive, were killed by cross-sectioning the cervical cord, following the ethical principles for animal handling established by the Brazilian College of Animal Experimentation -COBEA (www.cobea.org.br).

Sampling, biological indices, and biometry
All fish were dissected and measured for total length (TL), body weight (BW), and gonad weight (GW).These biometric data were used to calculate the gonadosomatic index (GSI = GWx100/BW) and the Fulton condition factor (K = BWx100/TL 3 ) for each of the three river sections.

Histology, gonadal maturation stage and spawning type
For histological analysis, fragments from the middle region of the ovaries and testes were fixed in Bouin's fluid for 24 hr, embedded in paraffin, sectioned at 5 μm thickness, and stained with haematoxylin-eosin (HE).Stages of gonadal maturation, spawning type, and the frequency distribution were established based on the macro-and microscopic characteristics of the gonads and on variation in the gonadosomatic index (Honorato-Sampaio et al. 2009, Weber et al. 2013, Normando et al. 2014, Nunes et al. 2015, Brandao et al. 2017).
Size at first gonadal maturation was determined as the smallest total length of females and males with gonads in the maturing/mature stage from each section (Boncompagni-Júnior et al. 2013, Brandão et al. 2017).

Histometry, fecundity and sex ratio
The diameter of 50 vitellogenic follicles, with little shrinkage and intact spherical shape, at the maturating/mature stage of development was measured from histological slides using an Olympus BX 50 light microscope with Olympus CellSens Standard 1.9 software.
Table 1.Water temperature, dissolved oxygen concentration and pH for the three studied sections of the São Francisco River Basin (SFR) from June 2011 to May 2012: section 1 = Três Marias dam; section 2 = SFR immediately downstream of the Três Marias dam; and section 3 = confluence between SFR and Abaeté River.In order to determine fecundity, sub-samples of mature ovaries (n = 10 from for each section) were collected.Samples from the middle region of the ovaries were fixed in a modified Gilson solution (100 ml of 60% ethanol, 880 ml of distilled water, 15 ml of 80% nitric acid, 18 ml of glacial acetic acid, and 20 g of mercuric chloride).Dissociated vitellogenic follicles were separated and counted under a stereoscopic microscope.The number obtained in the sub-sample was extrapolated for the total weight of the ovaries through the simple rule of three.Absolute fecundity (AF) was calculated using the equation: AF = OVA × GW, where OVA is the number of follicles per gram of ovary.Relative fecundity (RF) was calculated using the equations TL (AF/ TL) and GW (AF/GW).
The sex ratio in the three sections of the basin was determined as the ratio of the absolute frequency of females to that of males.The chi-square test (x 2 ; p <0.05) was applied to detect possible differences in the proportions between the sexes.

Statistical analysis
The variables of the three sections were tested for normality (Shapiro-Wilk) and subjected to an analysis of variance to test for significant differences in mean values of TL, BW, GSI, K, DF, AF, AF/TL and AF/GW.The parametric Duncan test was used to compare mean values.When the assumption of normality was not met, even after appropriate transformations, the data were subjected to the nonparametric Kruskal-Wallis test.A significance level of P < 0.05 was employed for all tests.
Females and males had statistically higher values for TL, BW, and GSI in the section immediately downstream from the Três Marias reservoir (section 1), than in sections 2 and 3 (Table 2).
Fish in reproductive activity, including spawned females and spermed males, were recorded throughout the year in all three sections.The peak for spawned females and spermed males in the three sections occurred in the bimonthly periods of November/December and January/ February.A long reproductive period (Figure 4) and the occurrence of spawning females with follicles in all stages of development, including post-ovulatory folicles, confirm that S. Brandtii is a partial spawner.
The size at first gonadal maturation for each section was: section 1 = females 14.2 cm and males 12.5 cm; section 2 = females 12.1 cm and males 11.8 cm; and section 3 = females 12.8 cm and males 11.6 cm.Absolute fecundity (AF) and relative fecundity (RF), calculated using TL and GW, were greater in section 1 (P < 0.05) than in sections 2 and 3 (Table 2).

Discussion
The results of the present study showed that S. brandtii reproduces in all three of the analyzed sections of the São Francisco River basin, although with lower performance in section 2 immediately downstream of the Três Marias dam.The physical and chemical conditions of the water in this section are known to be unfavourable for the reproduction of several fish species (Sato et al. 2005).The lower reproductive performance of S. brandtii in this section may be related to the low dissolved oxygen and temperature of the water due to thermal stratification of the reservoir during the summer and to water of the hypolimnion entering the turbines, as well as the anthropic impact of flow control (Santos et al. 2012).Impacts caused by the release of cooler water with low dissolved oxygen have been detected downstream of several other dams in different locations throughout the world (Clarkson & Childs 2000, Donaldson et al. 2008).
Fish from sections 2 and 3 had significantly lower lengths and weights than those from section 1, which may also be due to the negative environmental impacts (Nikolsky 1963) immediately downstream of the dam, as was also observed for Schizodon knerii (Brandão et al. 2016).Larger size at first gonadal maturation was observed for females in section 1, which may be related to differences in environmental conditions and food availability (Nikolsky 1963, Pawson et al. 2000).It could also be related to the production of sex hormones, such as 17-β oestradiol, which is responsible for both the somatic growth and gonadal development (Blázquez et al. 1998, Arantes et al. 2010).
The morphological characteristics of the ovaries and testes of S. brandtii we found to be similar to those of other species of the family Serrassalmidae (Honorato-Samapio et al. 2009).As has been reported for other piranhas, the mature ovaries of S. brandtii exhibited asynchronous development with follicles in different growth stages (i.e.perinucleolar, cortical alveolar and vitellogenic follicles; Honorato-Samapio et al. 2009, Marcon et al. 2017).Three gonadal maturation stages for females and males were established in the present study, which are similar Table 2. Biological variables for females and males of S.brandtii captured in three sections of the São Francisco River basin (SFR) from June 2011 to May 2012.Section 1 = Três Marias dam; section 2 = SFR immediately downstream of the Três Marias dam; and section 3 = confluence of the SFR with the Abaeté River.N = number of fish caught; TL = total length; BW = body weight; GSI = gonadosomatic index at maturing/mature stage; K = Fulton condition factor; DF = diameter of the vitellogenic follicle; AF = absolute fecundity; relative fecundity RF = AF/TL and AF/gonadal weight (GW).to those that have been reported for other species of Serrassalmidae (Honorato-Samapio et al. 2009) and for fish of other groups (Weber et al., 2013, Normando et al. 2014, Nunes et al. 2015, Brandao et al. 2017).

Females
The GSI can reflect environmental quality since it is directly linked to energy reserves available for breeding (Le Cren, 1951, Yoda & Yoneda 2009, Thomé et al. 2012).The GSI for three sections followed gonad development, with higher values in section 1, thus confirming that S. brandtii preferentially favors lentic environments (Braga 1981).The higher GSI values for females caught above the Três Marias dam may be a consequence of the greater number of vitellogenic follicles produced (fecundity) and/or the larger vitellogenic follicles (diameter) for the females of this section.
The highest frequency of spawned females and spent males for the three sections occurred in the bimonthly periods of November/ December and January/February, which coincide with high temperatures and rainfall, and longer photoperiods.Indeed, high values for these environmental factors are known to favor the reproductive activity of fish in the Neotropical Region (Lowe McConnel 1987).
The Fulton condition factor (K) provides important information on the physiological state of fish, assuming that individuals with higher K values are in better health (Nikolsky 1963, Vazzoler 1996, Froese 2006).In the present study, the highest K values for females and males were recorded in section 1 where water temperature, and oxygen quality were more favorable.
Histometric analysis of maturing/mature ovaries showed four classes of vitellogenic follicle diameter, indicating asynchronous folliculogenesis with distinct populations of vitellogenic follicles that will be spawned in batches.These findings are consistent with that observed for other teleosts that have fractioned spawning and batch fecundity (Melo et al. 2011, Armstrong & Witthames 2012).Asynchronous folliculogenesis is characteristic of species with long reproductive periods, multiple or fractioned spawning or batch fecundity.As are histological characteristics of spawned females, frequency of maturation stages and variation in the gonadosomatic index (Nikolsky 1963, Núñez & Duponchelle 2009;Lubzens et al. 2010), as is the case for S. brandtii of the present study.
The present study found batch fecundity and statistically higher fecundity relative to total length and gonadal weight in section 1, followed by section 3, showing that, as with GSI and vitellogenic follicle diameter, fecundity may be influenced by more favorable reproductive conditions for S. brandtii, particularly in section 1.  2012).In addition to the known effects of the hypothalamic-pituitary axis on reproduction, Jensen et al. (2004), Chakrabarty et al. (2012), andMarcon et al. (2017) found differences in the concentrations of vitellogenin, estradiol and aromatase, and in vitellogenic follicle diameter, in impacted environments, indicating impairment of follicular development.The present results showed that fish in section 2 may be experiencing endocrine alteration in the advanced secondary follicles; however, further investigation into the pathways leading to this change are needed.
Sex ratio may be related to differences between the sexes, selectivity of the sampling apparatus, food availability, and/or population stratification (Carvalho et al. 2009).Furthermore, male parental care of eggs and larvae, a behavior peculiar to piranhas and pirambebas (Braga 1981, Agostinho 2003), make them less readily captured by gill nets.On the other hand, a 1:1 sex ratio has been reported for the piranha Pygocentrus piraya in the Três Marias Reservoir (Cruz et al., 1996, Ferreira et al., 1996).In general, the present study sampled slightly more females than males in the three studied sections of the São Francisco River Basin.
The results of the present study showed that S. brandtii, a sedentary species, reproduces in all three of the studied sections of the São Francisco River basin, although with lower reproductive performance in section 2, which is immediately downstream of the Três Marias dam and where fish had lower values for TL, BW, GSI, K, FD, AF, and RF.These findings are probably related to the thermal disturbances caused by the release of cooler water from the hypolimnion of the reservoir since tropical fish decrease their feeding activity under low temperatures, which negatively affects their health (Lowe McConnel 1987).Indeed fish are able to find favorable conditions for reproduction by moving away from dams, such as in section 3, thus confirming the negative impact of dams on fish reproduction downstream of reservoirs of hydroelectric power plants.

Figure 1 .
Figure 1.Study area in the São Francisco River (SFR) basin, showing the locations of the studied sections: section 1 = Três Marias Reservoir; section 2 = section of the SFR above the confluence with the Abaeté River, extending from the dam 34 km downstream to the confluence with the Abaeté River; and section 3 = section below the confluence with the Abaete River, extending from 34 to 54 km downstream from the dam.
Data expressed as mean ± standard deviation (SD); different letters in the same row indicate statistically significant differences between sections (p < 0.05).

Figure 3 .
Figure 3. Histological sections of testes of S. brandtii in different gonadal maturation stages and stained with HE. a M1 = resting stage: containing only spermatogonia (SPG).b small amount of spermatozoa (SPZ) in the lumen of the seminiferous tubules in a moment of rapid development, which is rare to find in testis tissue prior to M2. c M2 = maturing/mature stage: with seminiferous tubules filled with spermatozoa (SPZ).d M3 = spent stage: empty lumen of the seminiferous tubules or with a small amount of residual spermatozoa.Bars: a = 20 µm; b and c = 80 µm; d = 40 µm.
Physical and chemical conditions of water are known to be the main factors influencing reproductive potential, as evidenced by vitellogenic follicle number and diameter (Yoda & Yoneda 2009, Armstrong & Witthames