Fish stand out among vertebrates as the group with the most remarkable diversity of reproductive strategies (Helfman et al., 2009), resulting mainly from the heterogeneity of secondary sexual characters, oocyte morphology, parental care, and the variety of courtship behaviors (Vazzoler, 1996). Their reproductive tactics tend to vary according to the intensities of intra and interspecific interactions or even in response to environmental variations, which may be natural or resulting from anthropic activities (Wootton, 1992; Souto et al., 2017).

Reproductive strategies and tactics are among the most important aspects of fish autecology and are directly related to the maintenance of their populations in each environment (Nikolsky, 1969; Braga, 2006; Monaco et al., 2014). In this sense, knowing these strategies and tactics contributes to the development of management plans for these biological populations, directing actions aimed at minimizing possible damage resulting from threats such as predatory fishing (Chapman et al., 1998; Ribeiro et al., 2008) and anthropic disturbances that affect the dynamics of aquatic ecosystems (Souto et al., 2017).

Conducting this type of study also fills a significant knowledge gap about the life history of tropical fish species (Raunkiaeran Gap) (Hortal et al., 2015). This is even more relevant for the Amazon basin, which has a high environmental heterogeneity (Sioli, 1984) and is home to the most extraordinary diversity of freshwater fish in the world, with a large part of this diversity present in creek ecosystems (Oberdorff et al., 2019).

In the Amazon basin, hydrological variations are considered the main predictor of natural seasonal variations in the environmental conditions of aquatic ecosystems (Espírito-Santo et al., 2013). In large rivers, this directly contributes to flooding pulses, which are predictable variations directly related to the reproductive activities of Amazonian fish (Junk et al., 1989). However, in upland streams, this seasonal variation of environmental conditions is much more unpredictable and is more strongly associated with the rainfall regime along the catchment (Kramer, 1978; Tomasella et al., 2008). This local rainfall is considered the primary regulator of the environmental conditions of the streams and, consequently, the main predictor of the reproductive tactics of the fish living in this environment (Kramer, 1978).

Hyphessobrycon heterorhabdus (Ulrey, 1894) is a small characid that is widely distributed in the lower Amazon region (Faria, 2020), standing out among the most abundant species in streams of the Eastern Amazon (Benone et al., 2017; Ferreira et al., 2018; Prudente et al., 2018; Santos et al., 2019). This species inhabits small tributaries and forms schools of five to 30 individuals that occupy backwater areas close to the banks, where they feed on items available both in the water column and associated with the substrate of these banks (Brejão et al., 2013).

Like other Amazonian characids, H. heterorhabdus is regionally known as “piaba” and economically distinct for being on the list of ornamental species popularly known as flag tetra (IBAMA, 2012). However, information about their reproductive ecology is still lacking.

In this sense, the present study aimed (i) characterize the reproductive biology of H. heterorhabdus considering the trilateral continuum model of Winemiller, Rose (1992), based on sex ratio, gonadal maturation stages, reproductive activity, growth pattern, allometric condition factor (K), spawning type, fecundity, and length where 50% and 100% of population to reach sexual maturity; and (ii) understand how this reproductive activity responds to temporal variation in environmental conditions in the upland streams of the Eastern Amazon. We hypothesized that the species would have reproductive traits corresponding to an opportunistic strategy, such as a prolonged period of reproductive activity that is positively related to hydrological variations such as precipitation and variation in depth and width of the channel, with a smaller influence of physicochemical characteristics of the water.

Study area. This study occurred in a catchment of 1,240 ha on the left margin of the Guamá River, in the municipality of Capitão Poço, state of Pará, Eastern Brazilian Amazon (Fig. 1). The local vegetation is classified as equatorial sub-perennial forest (IBGE, 1992). However, the catchment landscape is currently dominated by farmland and cattle pasture (Pacheco, Bastos, 2001), with small remnants of secondary (Silva et al., 1999) and riparian forests, which are considered areas of permanent preservation under article 30 of the Brazilian Forest Code, federal law number 12,651/12. The region has a humid tropical climate, subtype Af in the Köppen classification adapted by Peel et al. (2007). The mean annual temperature is 26.9 ºC, which varies only slightly over the year. The mean annual rainfall is 2,370 mm, with a rainy season between January and May, and a dry season between August and November (Pacheco, Bastos, 2001; INMET, 2021).

We sampled three low-order streams (1st to 2nd order sensu Strahler, 1957) with similar physical habitat characteristics. In each stream, we defined a 50 m stretch divided by six cross sections, resulting in five 10 m longitudinal sections where we measured the local environmental condition and sampled the fish species. The streams were assessed every two months between March 2019 and January 2020.

Data sampling. Before the fish collection, we measured four physicochemical characteristics of the water in each stream, such as Dissolved Oxygen (%), Electrical Conductivity (μS/cm-1), pH, and Temperature (ºC), using a Horiba U-50 multiparameter device. In each cross-section established at each stream, we measured the wetted width (WW – m), the thalweg depth (TD – cm) and the percentage of structured substrate (SS – %). We determined the WW (transversal distance between stream-flooded margins) and the TD using a ruled pole. The SS was obtained by summing the percentages of leaf litter, woody fragments and roots visually estimated at five equidistant points within each cross-section.

We recorded the flow speed (FS – m/s) of the stream at three equidistant points along the channel by measuring the time taken by a floating object to move a known distance. This information was used to calculate the stream discharge (D – m3/s) based on the equation, Q=a*vm , where Q is the discharge, Vm the mean flow speed, and A the mean transect area. The mean transect area was calculated by = ∑ A n` , where A is the area of the transect, which is given by the sum of

, where, Zn is the the measured depth of each segment, and w is the width of each segment. Accumulated monthly rainfall data were provided by the meteorological station of the Instituto Nacional de Meteorologia (INMET 2021 – Station A248), located in the municipality of Capitão Poço (Fig. S1). We used the average monthly values of the environmental variables obtained for each stream for the statistical analysis.

The specimens of Hyphessobryconheterorhabdus were collected from each 10 m longitudinal section for 12 min by three people using rectangular sieves (80 cm x 60 cm) with a 2 mm mesh. The individuals captured were euthanized with an overdose of Eugenol (6 ml / 3 L of water) and fixed in a 4% formalin solution in the field and preserved in 70% ethanol solution under vouchers numbers GEA.ICT 12101–12106 in the ichthyological collection of Grupo de Ecologia Aquática (GEA), Universidade Federal do Pará (UFPA), Belém, State of Pará, Brazil.

In the laboratory, ten individuals were randomly selected from each stream every two months, resulting in 30 individuals per month and 180 individuals in total. The selected specimens were measured for their total weight (Wt) using an analytical balance with a precision of 0.0001g, and for their standard length (Ls) in millimeters using a caliper with 0.01 mm precision. The specimens were eviscerated through a longitudinal incision in the ventral region to remove the gonads, which were also weighed (Wg) with a precision of 0.0001 g and conditioned in 70% alcohol. Afterward, the specimens were checked for their eviscerated weight (Wevi) in grams.

The gonads of the 180 individuals were assessed macroscopically to a previous definition of sex and gonadal maturation stage, following Vazzoler (1996), and subsequently, subjected to routine histological analysis, following Prophet et al. (1995). Considering that mature females are easily identifiable macroscopically because of the size of the gonad and the large number of oocytes, only three gonads from females at this stage were subjected to histological analysis to corroborate the previously identification of the stage. The microscopic definition of the sex and gonadal maturation stages was based on the presence and frequency of different types of cells of oogenic and spermatogenic lineages based on the classification proposed by Núñez, Duponchelle (2009).

Males were classified into four stages:1) immature, characterized by the presence of undifferentiated germ cells and spermatogonia surrounded by a large amount of connective tissue; 2) maturing, characterized by the presence of spermatogonia located along the seminiferous tubules, spermatocytes, spermatids, and a small amount of spermatozoids; 3) mature, characterized by the presence of a large number of spermatozoids in the lumen of the tubule; and 4) spent, characterized by the presence of practically empty seminiferous tubules and a few residual spermatozoids.

The females were classified into five gonadal stages:1) immature, characterized by the predominance of previtellogenic oocytes (stage I), which present a basophilic homogenous ooplasm, large central nuclei with central or sub-central nucleoli, and a high nucleoplasmic ratio; 2) maturing: characterized by the presence of previtellogenic oocytes (stage I), with a predominance of oocyte in early vitellogenesis (stage II). The latter is distinguished from stage I oocytes by the presence of cortical alveoli. Oocytes in advanced vitellogenesis (stage III) can also be observed at this stage of gonadal maturation, which is characterized by the presence of the chorion being clearly visible, folicular cells and the theca generally well developed; the nucleus or germinal vesicle is still visible and located in a central position; 3) mature: characterized by the predominance of stage IV oocytes, which are characterized by cytoplasm filled with large yolk globules. Previtelogenic (stage I) and early vitellogenic (stage II) oocytes were also observed at a low frequency during this stage of gonadal maturation; 4) spawned: characterized by the presence of previtellogenic oocytes (Stage I) and new batches of vitellogenic oocytes (stage II and III), which makes the ovary partially filled. Post-ovulatory follicles and some atretic oocytes were also found in this gonadal maturation stage; 5) resting: It presents characteristics similar to those of an immature female, with a predominance of previtellogenic oocytes (stage I), distinguishing it from this gonadal stage by the presence of a thicker ovarian wall. Some atretic follicles were also observed in this gonadal stage.

Reproductive traits. The proportion of males and females in the population was evaluated for the whole study period and each month. Differences in this sex ratio were tested using the Chi-square test (χ2) of adherence, considering as the null hypothesis that the sex ratio of this population does not differ from 1:1, as proposed by Vazzoler (1996).

The period of reproductive activity of the population of Hyphessobrycon heterorhabdus was evaluated through the Gonadosomatic Index (GSI) and the variation in the frequency of maturation stages over the sampled period. The GSI was obtained through the equation, GSI=Wg/Wt*100 , where Wg is the gonad weight and Wt is the individual's total weight. The GSI values ​​were verified for the statistical assumptions of normality and homoscedasticity and tested for their variation between the studied months using the non-parametric Kruskal Wallis test followed by a Wilcoxon multiple comparison test. This test was performed separately for males and females and disregarded immature specimens.

The growth pattern of Hyphessobrycon heterorhabdus was evaluated through the weight-length relationships of the individuals and followed the model proposed by Järvi (1920) (see Froese, 2006), Wt=a* L s b , where Wt is the individual total weight; L s the individual standard length; a is the proportionality coefficient, and b the allometric coefficient. To verify possible differences in the growth pattern between the sexes, the residuals of this relationship were evaluated according to the statistical assumptions of normality and homoscedasticity and tested between males and females using a Mann-Whitney U test. In case of significant differences, the weight-length relationship was evaluated separately for males and females, and the residues of this relationship were tested using a linear regression model to verify a possible non-random pattern of variation. In the case of non-random patterns, the weight-length relationship was adjusted through a polyphasic model proposed by Bervian et al. (2006).

The polyphasic model consists of using parameters a and b of the Huxley equation (1924), y+f a * x f b , where f a = a 1 + a 2 + a 1 1 + e RSCa* Ls−LSC and f b = b 1 + b 2 + b 1 1 + e RSCb* Ls−LSC . The parameters a1 and b1 represent the coefficients of proportionality and allometry determined in the first growth phase, and a2 and b2 are the same coefficients determined in the second growth phase. RSC (a/b) is the rate of change of the coefficients between the first and the second growth phase, and LSC is the standard length when the change in the growth pattern occurs.

The allometric condition factor (K) of each specimen, which reflects the balance between energy acquired through feeding and spent on activities such as reproduction and ecological interactions (Braga, 1986), was estimated through the equation K+Wevi/L s b , where Wevi is the eviscerated body weight obtained from the total mass of the specimen subtracting the weight of the gonad (Wg) and stomach (We); Ls is the standard length and b the allometric coefficient obtained through the weight-length relationship. K values ​​were obtained separately for males and females, evaluated according to the statistical assumptions of normality and homoscedasticity and submitted to a Kruskal-Wallis test to verify its variation between the months sampled, followed by a Wilcoxon multiple comparisons test. In the case of polyphasic growth, the individuals had their allometric condition factor (K) calculated based on the allometric coefficient of their respective lengths.

The mean length where 50% (L50) and 100% (L100) of population to reach sexual maturity was obtained separately for males and females through the logistic equations P = A * 1 + e r * L s − ( L 50 − 1 and P = A * 1 + e r * L s − L 100 − 1 respectively, where P is the proportion of adult individuals, A is the proportionality coefficient, r is the phase change rate parameter (from juveniles to adults), Ls is the standard length in mm, and L₅₀ and L100, are the mean length at sexual maturation of 50% and 100% of the population respectively. A length class with an amplitude of 2 mm for this analysis was defined. Both equations were adjusted using the Solver routine of the Microsoft Office Excel 2016® software.

The spawning type and fecundity of Hyphessobrycon heterorhabdus were defined by considering ten randomly selected mature gonads. The spawning type was defined based on the frequency distribution of oocyte diameters, macroscopically measured, and corroborated by the assessment of the frequency of occurrence of different phases of oocyte development in the histological sections. The oocytes were manually dissociated and subsequently photographed using a Motic stereomicroscope, model BA310E (zoom 40x), equipped with a camera. The diameter of the oocytes was measured using ImageJ® software with a spherical object of known diameter as a reference. The fecundity of H. heterorhabdus was estimated considering the number of vitellogenic oocytes, defined as oocytes in stage IV of maturation (Núñez, Duponchelle, 2009). Macroscopically, vitellogenic oocytes were distinguished from the other oocytes because they were larger, fuller, and yellowish. In addition, only oocytes with a diameter greater than the smallest microscopically identified vitellogenic oocyte (stage IV) were used for fecundity estimation. Oocyte count was also performed under a stereomicroscope, 40x zoom.

Effects of environment variables. To verify the predictive capacity of the physicochemical variables of water and the physical habitat in the reproductive activity of H. heterorhabdus, the variables was submitted to a Multiple Linear Regression Model (MLR), with GSI as the response variable. The selection of the variables with the highest predictor capacity of the GSI variation was validated through the Akaike selection criterion (AIC) (Zuur et al., 2009). The best model, according to AIC criteria, was used to interpret the relationships between reproductive activity and environmental variables. The analysis was run separately for males and females, considering a significance level of 5%, with R 4.1.1 software, using the MuMIn (Bartoń, 2020), Car (Fox, Weisberg, 2019), and Vegan (Oksanen et al., 2020) packages.

Considering the whole sample period, the streams presented an average value of 60.31% of dissolved oxygen, an average Electrical Conductivity of 22.60 μS/cm-1, pH of 5.60 and temperature of 26.28 °C. Concerning the channel morphology, the streams presented an average wetted width of 2.68 m, thalweg depth of 38.05 cm, and flow speed of 0.12 m/s. The average percentage of structured substrate was 67.62%. The variation of these environmental characteristics in the different sampling months can be seen in the Fig. S2.

Of the 180 analyzed specimens, 111 were male and 69 females. Males presented a mean standard length (Ls) of 20.65 mm (varying from 11.63 mm to 26.88 mm), while the females presented a mean standard length of 23.01 mm (varying from 12.68 mm to 32.57 mm). The population showed a difference in the sexual proportion between males and females for the sampled period (χ2 = 8.88; df = 2; p < 0.05), with 1.6 males for each female. In the assessment by month, only January showed a difference in the sexual proportion (χ2 = 4.8; df = 2; p < 0.05), with 2.3 males for each female (Fig. 2).

Gonadal maturation activity, based on Gonadosomatic Index (GSI) values, differed for males (H(5;75) = 25.913; p < 0.05) and females (H(5;50) = 21.65; p < 0.05) of H. heterorhabdus between the sampled months. The specimens recorded the highest gonadal maturation activity between March and May 2019, with a decrease in gonadal maturation activity from July 2019 on and reaching their lowest average values ​​between September and November 2019. A resumption of gonadal maturation was observed in January 2020. According to the multiple comparison test, for males, the months of the period with the highest intensity of rainfall (March and May) differed from the two months with the lowest intensity of rainfall (September and November). Therefore, differences were also evidenced between January and March (Fig. 3A). For females, the values ​​of GSI of the period with the highest intensity of rainfall (March and May) also differed from the two months with the lowest intensity of rainfall (September and November). Nevertheless, January also differed from September (Fig. 3B).

Variations in the frequency of the stages of gonadal maturation evidenced, for both sexes, a predominance of mature individuals in March 2019 (males = 63.15%; females = 63.63%), a higher frequency of immature between May (males = 35%; females = 30%) and July (males = 37.5%; females = 21.40%) and a higher frequency of maturing individuals in September (males = 43.7%; females = 71.40%). Spent males were recorded throughout the entire period studied, with a lower frequency in March 2019 (5.20%) and a higher frequency in January 2020 (38.00%) (Fig. 3C). Females of H. heterorhabdus also showed a higher predominance of gonads in the spawned stage in January (77.7%). Finally, resting females were observed only in September (14.28%) (Fig. 3D).

Hyphessobrycon heterorhabdus showed differences in growth patterns between males and females (W = 4723; df = 108; p < 0.05). Males showed positive allometric growth in a single phase represented by the model Wt=0.0000126*L s 3.12 (R² = 0.95) (Fig. S3), while females showed negative allometric growth in two phases (Fig. S3), with the first phase represented by model Wt=0.0001350*L s 2.29 (R² = 0.93), up to 20.71 mm, and the second phase represented by the model Wt=0.0000817*L s 2.59 (R² = 0.93).

The allometric condition factor (K) varied between the months sampled both for males (H (5; 75) = 14, p < 0.05) and females (H (5; 50) = 12.482; p < 0. 05). In both sexes, an increase in this parameter was observed between July and September, followed by a decrease in November, which was more marked in females, and a new rise in January. According to the multiple comparison test, for males, it was evidenced that the months with the highest rainfall (March and May) differed from the month with the lowest rainfall (September) and the month in which rainfall stars to increase (January). However, differences were also observed between March and July (Fig. 4A). For females, the K values ​​of the period with higher rainfall (March–May) also differed from the month with lower rainfall (September). However, January also differed from March (Fig. 4B).

The lengths where 50% (L50) and 100% (L100) of the population reached sexual maturity for males of H. heterorhabdus were 18.0 mm and 22.0 mm, respectively (Fig. 5A). For females, L50 was 19.7 mm, which was 1.7 mm higher than that of males, with a L100 of 27.0 mm, which was 5 mm higher than that of males (Fig. 5B).

The total number of oocytes at different developmental stages in mature females ranged from 176 to 655 (324 ±161; n = 10), with a diameter between 0.120 mm and 0.650 mm. A bimodal distribution of oocyte diameter frequency was detected with at least two batches of oocytes at different stages of development (Fig. 6A). The presence of oocytes at different developmental stages of maturation can also be visualized through macroscopic analysis of the mature ovaries of H. heterorhabdus (Fig. 6B). Histological sections revealed oocytes at all stages of maturation, indicating a continuous release of oocytes after reaching sexual maturity, reinforcing the idea of batch spawning in this species. The mean fecundity was 197 vitellogenic oocytes (179 ± 110) with a range between 108 to 420 oocytes, whose diameter varied from 0.242 mm to 0.650 mm.

Considering the global Multiple Linear Regression Model (MLR) (Tab. S4), followed by the Akaike Information Criterion (AIC), the variables with the greatest predictive capacity of variation in the GSI of males were water temperature, electrical conductivity, and stream discharge (Tab. S5). According to the selected model, these variables explained 10.74% of the variation in the GSI (p < 0.05) (Tab. 1). For females, the MLR highlighted the percentage of structured substrate and dissolved oxygen as the variables with the best predictive capacity for GSI variation (Tab. S6). According to the selected model, these variables explained 29.47% of the variation in the GSI of females (p < 0.05) (Tab. 1).

The flag tetra Hyphessobrycon heterorhabdus showed reproductive activity throughout the studied period, with a pronounced activity of gonadal maturation between March and May 2019. The GSI values ​​of females were positively associated with the percentage of structured substrate and mean values ​​of dissolved oxygen. For males, the GSI values were positively associated with water temperature, electrical conductivity and stream discharge. The response to environmental variations was more evident for females than males.

The sex ratio between males and females provides basic information about the reproductive dynamics of a population (Vazzoler, 1996; Wootton, 1999; Vicentini, Araújo, 2003). Although a 1:1 male to female ratio is expected in a natural environment, this can vary throughout the species’ life cycle due to factors such as mortality, growth rate, and behavior (Raposo, Gurgel, 2001; Vicentini, Araújo, 2003; Fagundes et al., 2020). In the present study, the higher proportion of males in periods of higher reproductive activity indicates the possibility of oocytes from a single female to be fertilized by spermatozoa from more than one male, thus increasing the genetic variability of this population, as suggested by Wootton et al. (1978).

Regarding gonadal maturation, Hyphessobrycon heterorhabdus showed reproductive activity throughout the sampling cycle, with greater intensity between March and May. According to Azevedo (2010), 20% of small characids have a long reproductive period that can last up to six months, which is strongly associated with an opportunistic strategy (Winemiller, 1989; Winemiller, Rose, 1992). Fish with this strategy are also characterized by having reduced body size, early sexual maturation size, multiple spawning, and low fecundity (Winemiller, 1989) and are common in stream ecosystems (Kramer, 1978). Long reproductive periods also contribute to the increase in population density (Winemiller et al., 2008), corroborating the fact that this species is among the most abundant in streams of the Eastern Amazon, even in areas that are affected by anthropogenic activities (Ferreira et al., 2018; Prudente et al., 2018; Santos et al., 2019).

About L50, males showed a smaller size at first sexual maturation than females. This pattern is typical in Neotropical characids, such as Astyanax scabripinnis (Jenyns, 1842) (Veregue, Orsi, 2003) and A. lacustris (Lütken, 1875) (Súarez et al., 2017), and even for other small fish, such as Characidium lauroi Travassos, 1949 (Braga, 2006) and Aspidoras fuscoguttatus Nijssen & Isbrücker, 1976 (Araujo, Garutti, 2003). This smaller size may be related to the fact that females demand higher energy for gonadal maturation (Pereira-Filho et al., 2011) since the energy investment for oocyte maturation seems to be greater than for sperm maturation (Vazzoler, 1996; Wootton, 1999). Therefore, females would require a larger body size (Bromley, 2003). Individuals with larger body sizes also have a greater capacity to absorb oxygen, which is fundamental to biochemical processes involved in gonadal maturation (Pauly, Cheung, 2017).

The growth pattern of H. heterorhabdus also appears to reflect the energy demand of reproductive activity and the significant female reproduction investment. Although females maintain a negative allometric growth throughout their development, the increase in the value of the allometric coefficient, after 20.71 mm, is an important indicator decrease in investment in growth when analyzed proportionally to weight. Considering that this change occurs at a length where more than 50% of the females are sexually mature, it is likely that this change is related to a greater investment in reproductive activities, such as gonadal maturation.

The bimodal distribution of oocyte diameters defines the spawning type of H. heterorhabdus as group-synchronous, characterized by two populations of oocytes recognized in the ovary throughout the reproductive season (Wallace, Selman, 1981; Lubzens et al., 2010). This pattern was evidenced in congeneric species, Hyphessobrycon santae (Eigenmann, 1907) (Sales et al., 2015) and Hyphessobrycon anisitsi (= Psalidodon anisitsi (Eigenmann, 1907)) (Gonçalves et al., 2013) in reservoirs of the state of Minas Gerais and upper Paraná River basin, respectively, and in Hyphessobrycon eques (Steindachner, 1882) (Santana et al., 2019) in floodplains of the Brazilian Pantanal and the Paraguay and Amonguijá rivers. On the other hand, the spawning type of H. heterorhabdus differs from the congeneric species Hyphessobrycon bifasciatus Ellis, 1911 and H. eques in a lake of the upper Paraná River basin (Gonçalves et al., 2013). Seasonal fluctuation of environmental parameters is considered the primary selective pressure of this biological strategy (Winemiller, 2005). In seasonal environments, fish species tend to present a total spawning, which generally occurs during the high-water periods, when better environmental conditions benefit the offspring’s survival (Gonçalves et al., 2013). In environments with less seasonal variation, batch spawning would minimize intraspecific competition and enhances the use of the microhabitats, thus contributing to the viability of the offspring (Barbieri, 1992). According to Winemiller (1989), batch spawning is frequently observed in fish with an opportunistic reproductive strategy.

The average fecundity of H. heterorhabdus was lower (324 ± 161), considering congeneric species that inhabit reduced seasonal variation environments in the Brazilian Southeast, such as H. bifasciatus, which presents an average of 482 (± 221) mature oocytes, and H. eques, with an average of 486 (± 159) mature oocytes, as well as those inhabiting environments with greater seasonal variation in the same basin, such as H. anisitsi with 931 (± 521) mature oocytes (Gonçalves et al., 2013). In contrast, H. heterorhabdus showed a higher fecundity than H. eques (average = 191.9 mature/vitellogenic; min = 67; max = 624 oocytes) sampled in the floodplains of the Brazilian Pantanal and the Paraguay and Amonguijá rivers (Santana et al., 2018). As suggested by Gonçalves et al. (2013), environmental conditions can be important predictors of fish fecundity. According to these authors, marked seasonal variations in environmental conditions restrict periods of favorable conditions for reproduction, resulting in lower fecundity in places with more stable environmental conditions. However, according to Wootton (1992) fish fecundity can also be affected by body size, where in species with a larger body size, females have a larger coelomic cavity, being able to house a greater number of mature oocytes.

The batch spawning of H. heterorhabdus reinforces the idea of low seasonal fluctuation of the environmental condition of the Amazonian streams concerning Brazilian subtropical streams, such as savanna and Atlantic Forest streams (Veregue, Orsi, 2003; Gonçalves et al., 2013) and large amazon rivers with a marked flood pulse regime (Junk et al., 1989). However, this reproductive tactic can also be related to the habitat use by H. heterorhabdus, which inhabits the water column in backwater areas close to the banks and occupies microhabitats in the stream during high and low rainfall. The batch spawning also matches with the low fecundity recorded in H. heterorhabdus, which does not require a high production of oocytes to be released in a short time when environmental conditions are more favorable.

The predictive effect of environmental variables in the reproductive activity of H. heterorhabdus was more evident for females (29.47% of explanation; p < 0.05) than for males (10.74%; p < 0.05), which may be associated with the fact that oocyte production demands more energy (Pereira-Filho et al., 2011) than the production and maturation of spermatozoa (Vazzoler, 1996; Wootton, 1999) (Tab. 1). The elevated GSI values observed in male individuals can be partially attributed to the augmentation of mean water temperature, thereby reinforcing the notion of its significance in the reproductive biology of fish (Larsen et al., 2011). High water temperatures allow an increase in the metabolic rates of fish, affecting their reproductive performance (Vagner et al., 2019). The relationship between fish reproductive activity and temperature is frequently reported for fish from temperate regions (Migaud et al., 2010), while for tropical stream fish, mainly from the Amazon, reproductive activity is often associated with the local rainfall regime (Carvalho et al., 2021; López-Rodríguez et al., 2021). However, experimental evidence has demonstrated that increased water temperature leads to a reduction in the duration of spermatogenesis and an increase in cell proliferation in Neotropical characids, resulting in heightened GSI values in these fish (Quirino et al.,2021).

The increase in the reproductive activity of females, expressed by the increase in the GSI values, coincided with the increase in the average values ​​of dissolved oxygen, corroborating the idea that this physicochemical parameter of the water is an essential factor for the reproduction of tropical fish (Pauly, Cheung, 2017). The positive relationship between the GSI and the percentage of structured substrate may indicate the importance of transporting organic matter from adjacent regions to the stream bed. The cumulative organic matter allows for the formation of favorable environments for spawning, shelter, and feeding of juvenile fish (Kramer, 1978; Casatti, 2005), which are essential to supply the energy invested in reproduction (Castillo-Rivera et al., 1994). This process would be essential since H. heterorhabdus has a diet predominantly composed of items of allochthonous origin (Benone et al., 2020). However, studies focused on the relationship between fish feeding and reproduction are scarce (Ballesteros et al., 2009) and non-existent for stream ecosystems.

In this sense, we believe that females may have a greater ability to adjust their physiological conditions to the temporal variation of the stream’s environmental conditions. On the other hand, we think that gonadal maturation in males would be less affected by the measured environmental variables because the time for gonadal maturation is shorter and has a lower energetic cost than in females. Another possibility is that the gonadal maturation of males is also influenced by behavioral or even hormonal stimuli resulting from female gonadal maturation. However, behavioral and physiological studies involving these issues are still insufficient (Nakatsuru, Kramer, 1982; Bhat et al., 2021).

The results obtained in the present study demonstrate that H. heterorhabdus presents a reproductive activity synchronized with environmental conditions related to the rainfall regime. However, a portion of the reproductive activity responds to other factors not yet clarified in this study, which may be better evaluated in experimental studies focused on triggers of gonadal maturation and spawning in this species. Even so, reduced body size and multiple spawning in an extended reproductive activity period tend to fit the species in an opportunistic strategy according to the classification proposed by Winemiller (1989) and Winemiller, Rose (1992). The presence of H. heterorhabdus in the list of species authorized by the Brazilian government to be exploited for ornamental or aquarium purposes (Interministerial Regulatory Instruction nº 01/2012 of extinct Ministry of Fisheries and Aquaculture of Brazil) makes the information herein compiled of paramount importance for elaborating strategies aimed at conservation and sustainable exploitation practices, such as period and minimal size of capture.