Age, growth, mortality and yield per recruit of the dourado Salminus brasiliensis, Corumbá Reservoir, Goiás State, Brazil

In order to evaluate the fate of the migratory species dourado Salminus brasiliensis in the first years of impoundment in Corumbá Reservoir we estimated age, growth, mortality and yield per recruit. Samplings were carried out monthly in Corumbá Reservoir and its main tributaries (Goiás State) from March 1998 to February 1999 using gillnets. After one year of impoundment, age was estimated from scales and the maximum number of rings was six for males and five for females. Rings are formed annually in May and June. The asymptotic length and growth rate for males and females were 37.1 cm and 0.77 and 56.6 cm and 0.52, respectively. A dominance of juveniles was verified in the reservoir and its tributaries. The instantaneous total mortality rate (Z) was 1.59 and the annual total mortality rate (A) was 79.6%. The highest yield per recruit (1200g) and the highest average weight (1900g) were obtained in simulations with low values of fishing (F) and natural (M) mortalities. Based on the above information we describe the Corumbá Reservoir impoundment has influenced the growth of the dourado. As regards this study, we recommend that the fishing effort not be applied during the trophic upsurge period and that the monitoring of the dourado assemblage continue. Fishery programs management for this species should be carried out with subsequent monitoring involving efficient communication, realistic practices and involvement of fisher organizations.


Introduction
The dourado from upper rio Paraná was formerly identified as Salminus maxillosus, but this is actually a junior synonym of S. brasiliensis, the valid name for that species.According to Lima et al. (2003) it is distributed in the Paraná, Paraguay and Uruguay river basins, Laguna dos Patos drainage basin and the upper Chaparé and Mamoré river basin in Bolivia.It is the largest characin of the Paraná basin and could reach more than one meter in length (Godoy, 1975;Britski et al., 1999).
The diet of this piscivore includes a great variety of prey, which reflects its relative abundance in the environment.In its adult phase on the upper Paraná River floodplain it fed on mainly corimba Prochilodus lineatus, piau-bosteiro Schizodon altoparanae and lambari Moenkhausia intermedia (Almeida et al., 1997).It is the most valuable in the artisanal and sport fishing of the upper Paraná River basin (Agostinho et al., 2003).Given its popularity for all types of fishing, Agostinho et al. (2005) consider it a good candidate for an umbrella species in conservations strategies.Salminus brasiliensis has been impacted by the construction of dams because it can migrate up to 1000 km to spawn (Petrere Jr., 1985) and needs floodplains and marginal lagoons for the larvae to complete development and the juveniles to find food and shelter (Agostinho et al., 2004).
This study examines growth parameters from an assemblage of S. brasiliensis after the first year of the impoundment of Corumbá Reservoir.Considering this fact, the objective of this paper is a) to determine periodicity and time of age ring formation, b) to estimate age and growth parameters (length) for each sex, c) to determine age structure for the different environments sampled in the reservoir area (the reservoir itself, upstream, downstream and the tributaries), d) to estimate the mortality rate and e) to evaluate yield using yield per recruitment models.Based on the aims above we expect to describe how the Corumbá Reservoir impoundment has influenced the growth of the dourado.

Study area
The Corumbá River (500 km long with a 34,000 km 2 drainage area) is the second most important tributary of the Paranaíba River, which, along with the Grande River, form the Paraná River (Paiva, 1982).Corumbá Reservoir is located in the upper third of the Corumbá River and was closed in September 1996.It does not have a fish ladder.Its surface has an area of 65 km 2 and an average depth of 23 m.Average hydraulic retention time is 30 days.Annual precipitation is above 1,500 mm, with maximum values in January and minimum values in June.The dry period is from April to September, with a minimum annual temperature of 10 o C in July and a maximum of 31.5 o C in October.
Samplings were carried out from March 1998 to February 1999 in 12 sites distributed in the Corumbá River and Corumbá Reservoir and their main tributaries.Five sites were located in the reservoir (Pedra Lisa -LISA, Jacuba -JACU, Pirapitinga -CPIR, Ponte -PONT and Peixe -COPE), two upstream from the reservoir (Porto das Moitas -MOIT and Areia -AREI) and one downstream (JUSA).Four sites were located in the tributaries (two in the Pirapitinga River -PIRA and SAPE; two in the Peixe River -PFOZ and LINI) (Fig. 1 and Table 1).
Date and sampling site, standard length (Ls) in centimeters, total weight (Wt) in grams and sex of each individual were recorded.Five individuals were sampled from each length class by sex.About 10 scales from the region immediately posterior to the insertion of the left pectoral fin were taken from each individual and stored in an envelope.Scales were cleaned and sterilized and placed between glass slides, according to the methodology of Vazzoler (1982).The reader counted three times and measured age rings using a stereomicroscope.In the case of different results, the data were not considered.Measurements were taken in mm between the center and the edge of the scale and between the center and each ring using an ocular micrometer.

Data analysis
The coefficient of variation (CV) of the standard length per growth ring was used to determine age precision for males and females (Witherel & Burnett, 1993).
The period of age ring formation was determined using analysis of the marginal increment of the scales per month, considering 2 to 6 year olds (Lai et al., 1996;Campana, 2001;Isely & Grabowski, 2007).
Growth in length was modeled for the sexes separately using the von Bertalanffy equation (Beverton & Holt, 1957).In order to adjust the von Bertalanffy model, a non-linear procedure was carried out (King, 1995).Seed values were obtained using the Ford-Walford protocol (Walford, 1946).The von Bertalanffy growth equation is: ] Where: L s = average standard length (cm) of the individuals with age t; L = maximum standard length (cm) that, on average, the individuals can reach, also called asymptotic length; k = refers to the growth rate (year -1 ); t = age of the individuals; t 0 = parameter related to the average standard length of the individuals at hatching.
The value of t 0 was zero, given that this parameter does not have biological relevance (Moreau, 1987).
The age structure was established using the frequency distribution of the ages per environment (reservoir, upstream and tributaries).A total of 254 individuals were considered for this analysis.The individuals were deposited in a museum collection and were not part the analyses.
Instantaneous total mortality (Z) was obtained converting the length of each individual into age using the formula t = ln e (1-L/L ) k. Later, a linear regression was applied, with age (x axis) and Ln e of the abundance per age class (y axis).The slope of this relationship corresponds to Z (Van Den Avyle, 1993).Annual mortality (A), i.e. the number of fish that die during a year, was obtained using the formula A = 1 -e -Z (Ricker, 1975).The model proposed by Beverton & Holt (1957) and modified by Jones (Ricker, 1975) was used to simulate the yield per recruit.This model is based on the premise that the stock biomass varies with the rates of growth and mortality.Thus, yield is obtained from variations in the available biomass or dynamic pool (Van Den Avyle, 1993).The model is expressed by the following equation: Where: F = instantaneous fishing mortality; N 1 = number of individuals recruited to age 1 (in years); r = time (in years) between age 1 and when the fish becomes susceptible to the fishing gear; W = asymptotic average weight; k = growth coefficient; We arbitrarily set N 1 = 1000, but then adjusted these values multiplying by the representation (i.e.proportion) of the species at age 1 relative to the other species (Miranda et al., 2000).
The main information necessary to calculate yield per recruit and average weight were the parameters (a and b) of the weight-length relationship (0.0118 and 3.1616, respectively), von Bertalanffy growth equation parameters, asymptotic length (L = 47.2cm) and weight (W = 2331.0g),standard length of first maturation (L 50 = 32.0cm) and age of first sexual maturation (tL 50 = 1.82 years), obtained solving the von Bertalanffy equation for L 50.
Simulations of yield per recruit were conducted varying instantaneous natural mortality (M), fixed arbitrarily from 0.1 (low) to 0.6 (high), and instantaneous fishing mortality (F), ranging from 0.1 to 0.6.Instantaneous total mortality (Z) was computed as the sum of M and F (Ricker, 1975).The yield simulations were carried out using the package SAS (SAS Institute Inc., 2000).

Results
A total of 195 (100 males and 95 females) Salminus brasiliensis individuals out of 300 caught was analyzed.In the downstream site only one individual was sampled and in the Piratininga River the dourado was not caught during this sampling period.Standard length of caught individuals varied from 8.5 cm to 41.2 cm for males and 8.5 cm to 63.5 cm for females.Maximum number of rings observed was 6 for males and 5 for females (Table 2).
Formation of the age rings for S. brasiliensis occurred in March through August since the lowest marginal increment averages were found in this period (Fig. 2).It is evident from these data that formation of age rings follows an annual periodicity and indicates that the scales of the dourado are ad- Salminus brasiliensis presented values of L " equal to 37.1 cm for males and 56.6 cm for females and growth rates k equal to 0.77 and 0.52 for males and females, respectively (Table 3).Females presented larger sizes and higher growth rates than males.
A dominance of juveniles, particularly females less than one year of age and one-year-old males, was verified in Corumbá Reservoir (Fig. 3).In addition, adults up to 4 years of age were also caught.There was a dominance of juveniles upstream and in the tributaries; however, they were in lower quantities when compared to the reservoir.
Causes of mortality of the fish stock in Corumbá Reservoir are only natural, since fishing is prohibited.The total instantaneous mortality rate Z was 1.59, which corresponds to an annual total mortality rate A of 79.6%, with Ln (N) = -1.5864(age) + 6.03 (R 2 = 0.96); N=4.
The best yield per recruit and average weight were obtained in the simulations with low values of fishing mortality F and natural mortality M (Figs. 4 a and b).
which reached a maximum value of 22.3% for males and 18.5% for females.These values show good precision in the reading of the growth rings.The high degree (82%) of agreement between the readers also suggests that scales can reliably estimate age.This was also verified for S. brasiliensis studied in the Paraná River near the municipality of Porto Rico (Paraná State).For that study, the coefficient of variation was not above 20% either (Feitosa et al., 2004).
There is some controversy in the literature about which factors could create growth marks in tropical fish.Various authors consider multiple causes (Brett, 1979;Ambrósio & Hayashi, 1997;Jepsen et al., 1999;Ambrósio et al., 2003).Formation of rings in S. brasiliensis coincides with the period of low precipitation, low temperatures, short daylight and high water transparency in the region.The seasonal growth linked to mark formation indicates that the former two variables are the most important but, as emphasized by Fontoura & Agostinho (1996), they are not the only deter- Table 3. Parameters estimated for the von Bertalanffy model using non-linear procedure (L = asymptotic length (cm), k = growth rate (year -1 ), F = Fisher Statistic, P = level of significance, SST = sum of the corrected total squares) for S. brasiliensis males and females in the area of influence of Corumbá Reservoir.

Discussion
Calcified structures present periodic growth and are useful in determining age in fish.Scales were chosen in this study because they are easy to collect and prepare, and have given satisfactory results for the Neotropical region (Castagnolli, 1971;Sverlij & Espinach-Ros, 1986;Barbieri et al., 2001;Feitosa et al., 2004).Variability in age determination may be due to difficulty in distinguishing between false and true rings.This is associated with errors in the criteria used to determine age and due to the subjectivity existing in every age estimate (Casselman, 1983;Kimura & Lyons, 1991;Campana, 2001;Fernandes et al., 2002).Another difficulty was the sample.Few large-sized individuals were caught.Thus, the precise age (for purposes of validating the formation periodicity of the ring) was evaluated using the coefficient of variation, mining factors.
The studies carried out on this species in the Mogi-Guaçu River (São Paulo State) show that the formation of age rings occurs in December and January.This is related to the high precipitation, air and water temperatures and photoperiod, in addition to coinciding with the spawning period (Barbieri et al., 2000).However, Leporinus acutidens studies carried in middle Paraná River stretches report growth mark formation during autumn/winter before the damming and during spring after it (Araya et al., 2008).
Studies developed in the Mogi-Guaçu River for S. brasiliensis reveal asymptotic lengths of 71.0 cm for males and 87.4 cm for females (Barbieri et al., 2001).In the Paraná River, asymptotic length was found to be 76.7 cm for males and 81.1 cm for females (Feitosa et al., 2004).This study agrees in general that females reach greater asymptotic length than males.The value of k estimated for S. brasiliensis in Corumbá Reservoir was high and indicates fast growth.High growth rates of the dourado in the first year have been determined in Corumbá Reservoir.This is due to the fact that the reservoir is now (trophic upsurge period) and, therefore, allows rapid colonization and high reproductive and recruitment success (Kimmel and Groeger, 1986;Agostinho et al., 1999).This strategy seems to enable piscivores to feed on other fish and this resource had the highest availability in Corumbá Reservoir (Luz- Agostinho et al., 2006).
Like other characins, S. brasiliensis eggs drift to the lower parts of the tributaries while undergoing embryonic development and are washed into floodplains and marginal lagoons where the larvae complete development and juveniles find food and shelter (Agostinho et al., 2004).Our study shows the high frequency of juveniles in Corumbá Reservoir, just after its impoundment, and indicates that this species exploits its abundant but temporary resources (food and shelter) to complete its development.A similar result is reported for Leporinus species in Yacyretá (Araya et al., 2005(Araya et al., , 2008)).The high mortality of the dourado in the first and second year could be because open areas with low turbidity can increase predation on eggs and larvae.Studies carried out in the Tocantins River after the impoundment of Lajeado Reservoir found a reduction in the density of fish eggs and larvae (Agostinho et al., 2007) and demonstrated that the changes in the transport-sedimentation process led to an increase in water transparency (especially in the reservoir), which influenced predation on eggs and larvae.According to Luz- Agostinho et al. (2006), Galeocharax knerii, Serrasalmus spilopleura (actually S. maculatus), Cichla monoculus (actually C. kelberi), Cichlasoma paranense, Pseudoplatystoma corruscans, Rhamdia quelen and Hoplias malabaricus from Corumbá Reservoir mainly eat fish (forager species and the younger of other species).
The values found for yield per recruitment and per weight show that this species has the economic potential to lead a fishery in Corumbá Reservoir as long as the fishing effort is maintained at low levels.In spite of this economic potential, as the reservoir ages this stage of high production (trophic upsurge period) will tend towards trophic depletion and migratory and large-bodied fish which reached peak abundance shortly after impoundment will decrease, and assemblages shifted towards a dominance of sedentary, medium-sized species, as mentioned by Agostinho et al. (1999).Consequently, we recommend that the fishing effort not be applied during the trophic upsurge period and that the monitoring of the dourado assemblage continue.Fishery programs management for this species should be carried out with subsequent monitoring involving efficient communication, realistic practices and involvement of fisher organizations.

Fig. 1 .
Fig. 1.Location of the sampling sites in the area of Corumbá Reservoir.
probability of the beta distribution; Γ = complete gamma function (fully defined); X 1 = e -kr ; X 2 = e -kl ; l = time (in years) between age 1 and when the fish are no longer susceptible to the fishery; P = Z/k; Z = instantaneous total mortality; Q = b + 1; b = angular coefficient of the weight-length relationship; R = N 1 e -Mr ; M = instantaneous natural mortality.

Fig. 4 .
Fig. 4. (a) Simulation of the yield per recruit (g) in different scenarios of fishing mortality (year -1 ).Each curve represents a natural mortality rate (year -1 ) (0.1 to 0.6 with intervals of 0.1) for S. brasiliensis in the area of influence of Corumbá Reservoir.(b) Simulation of the average weight (g) in different scenarios of fishing mortality (year -1 ).Each curve represents a natural mortality rate (year -1 ) (0.1 to 0.6 with intervals of 0.1) for S. brasiliensis in the area of influence of Corumbá Reservoir.

Table 1 .
Morphometric and physiographic characteristics of the sampling sites.

Table 2 .
Sample size (N), average standard length (Ls, cm) and coefficient of variation (CV) of the standard length per growth ring in Salminus brasiliensis scales from the area of influence of Corumbá Reservoir.
Fig. 2. Monthly average (mean ± standard error; SE) of the marginal increment in scales of the dourado Salminus brasiliensis in the area of influence of Corumbá Reservoir.equate to estimate age.