The study was conducted in three areas of the Babahoyo River and the fish farm center located in the Province Los Rios (Ecuador). The area has a tropical climate with an average temperature of 25°C, an annual rainfall of 2400 mm and a relative humidity of 82%. The salinity of water, both in the river and the fish farm, does not exceed 0.1%, the pH was between 7.0 and 7.29, the range of temperature is 19.7ºC and 24.7ºC cultured fish, while the dissolved oxygen in the river and fish farm is between 6.8 and 8.9 mg/L, respectively. The conductivity values are about 145 mS/cm.

One hundred matured fish samples (following the rules described by Konings, 1989 ) of C. festae comprising of 50 individuals from natural origin (wild population) and 50 from a cultured environment (private fish farms, cultured stock) were collected at dawn over the month of May 2016 with the help of standard fishing gears like cast and hand nets. Since male and female could not be differentiated morphologically, sexing of the sampled fish was not carried out. Specimen collection was performed weekly by purchasing representative samples of the two selected populations from local fishermen (wild fish) or fish farm (cultivated fish). Wild fishes were caught from three different locations within their natural geographic distributions in Babahoyo River (Provincia de los Rios, Ecuador). Cultured fishes were collected from fish farm. Just after catching, the fish specimens were kept in a glass flow through aquaria with continuous air and filled with 200 L of dechlorinated tap water, transported alive and housed in two masonry tanks (capacity of 500 L) (dissolved oxygen = 6.20 ± 0.0 mg/L, temperature = 20.5 ± 0.2 °C and pH = 5.6 ± 0.1). The fish rested for 48 h before the experiment, with fasting time of 24 h before stunning. On the day of the experiment, the water in the tank was reduced by half; the fish were quickly caught with a net and transferred to a plastic box (100 L) and kept indoor. For stunning, the fish were placed at the same time in a mixture of 40 L of ice and 40 L of water (0.8 °C) until the apparent stunning (20 min) was over. After confirmation of death, the fish were identified, and live weight and pH was performed. Then, the collected fish were rapidly bagged, packed in an undistorted condition and stored at 0 ± 2 ºC in chilly bins containing ice, and carefully brought to a laboratory in Facultad de Ciencias Pecuarias of Escuela de Zootecnia of the Universidad Técnica Estatal de Quevedo (Ecuador) until further analysis.

The study was carried out according to Ecuadorian national recommendations for the management of fish, taking into considerations the rules on animal welfare.

Muscle pH was determined after death (pH ₀ ), at 2 h (pH ₂ ) and 12 h (pH ₁₂ ) postmortem by inserting a pH electrode (portable meat pHmetre, HI99163, Hanna Instruments Ltd, UK) into the Flesh Quality Cut; dorsal part of the fillet posterior to the head. The instrument was frequently calibrated using pH 4.01 and pH 7.00 buffers, and the electrode was also cleaned to obtain consistent results. Muscle pH was measured in duplicate and expressed as mean value from both.

At the laboratory, the fishes were kept in boxes with ice in cold stores at 2 ± 1 °C, and flake ice was added to the boxes as required. Laboratory analyses started 24 h after death when rigor mortis had passed in most of the chilled fishes. Fish were dissected with a scalpel and scissors, and fins, scales, head, entrails, bones and flesh were removed and weighed. Head, guts, bones and flesh yield were calculated according to methodology propose by Rutten et al. (2004) . Biometric traits as standard length (cm), body weight (g), gutted body weight (g) and weight without head, viscera and skin (g), were utilised used in order to estimate slaughter yield and dress-out as expressed in the following equations:

After 45 min filleting, surface colour measurements on right fillets were recorded at three positions using a portable colorimeter (Lutron RGB-1002 Chroma Meter) equipped with light source C and a 2º observed angle, calibrated to a white standard. The L*, a*, b* system colour profile was measured in all harvested fish. The colour variables calculate were L*, a* and b* where L* describes lightness (+L* = white, -L* = black), a* red-green chromaticity (+a* = red, –a* = green) and b* yellow-blue chromaticity (+b* = yellow, -b* = blue) as recommended by CIE (1976) . For each fillet, three measurements (along the length of the fillet) were done on the interior part of fillet, and values were combined to one mean value per fish for each of the three colour variables measured.

Water holding capacity (WHC) was determined using the method described by Grau & Hamm (1953) and it was measured in two ways: drip loss and cooking loss. To determine drip loss, two cubes of 10 mm × 10 mm × 20 mm were cut of fresh muscle.

The cubes were suspended on a pin inside a sample bottle (200 mL) taking care that the meat did not touch the sides of the bottle and stored for 24 h at 2 ± 1°C. The amount of drip measured between 24 h and 48 h postmortem , as the difference between the sample mass before and after, was expressed as a percentage of the starting mass:

To evaluate cooking loss, the samples (∼ 30 g) were trimmed of external fat, weighed prior to cooking, placed in a polyethylene bag and immersed in a water bath (JP Selecta, Barcelona, Spain) at 80°C until the internal temperature of sample reached 70°C. The temperature was repeatedly monitored by a Type Kflexible high-temperature thermocouple (Hanna Instruments, USA) inserted into the geometric centre of each piece. Once the samples were cooled at room temperature (approximately 15ºC) for 40 min, they were re-weighed (after gently blotting on filter paper). Cooking loss percentage was calculated as follows:

Muscle samples were homogenized using a 20000 rpm grinder. Wet, crude protein, total fat and ash percent of fish raw meat were determined according to AOAC (2000) . The crude protein content was measured by the block digestion method (UNE 55-020), ashing was done at 550 °C for 24 h (ISO R-936), and the moisture content was determined by drying at 102 °C for 24 h (ISO R-1442). Fat percentage was measured according to the Soxhlet method (ISO R-1443) using a Foss Tecator AB Soxtec 2050. Analyses were determined in duplicate, according to the mean value of two determinations and expressed in mg per 100 g of raw meat.

Skinned and deboned muscle from individual fish was blended into homogeneous flesh and total lipid was extracted with chloroform/methanol (2:1 v/v) containing 0.01% of butylated hydroxytoluene (BHT) as antioxidant ( Folch et al. , 1957 ). The organic solvent was evaporated under a stream of nitrogen and the lipid content was determined gravimetrically. Aliquots of the lipids extracted were converted to fatty acid methyl esters (FAME) according to Chistie (1993) . FAME were separated and identified on GC Perkin Elmer Clarus 500 gas chromatograph with a flame ionization detector (FID) equipped with a TR-FAME capillary column (30 m × 0.25 mm i.d., 0.25 µm film thickness, Shinwa Inc.), using helium as a carrier gas at a flow rate of 0.5 mL/min. The injection and detector were maintained at 250 and 260 °C, respectively. The oven temperature was programmed at 100 °C, followed by an increase of 2 °C/min to 220 °C, with a final hold time of 20 min. Individual fatty acids were identified by comparing their retention times with those of a standard fatty acid mix Sulpeco 37 (Sigma Chemical Co. Ltd., Poole, UK). Nonadecanoic acid methyl ester (19:0 ME) was used as an internal standard. Individual fatty acids were expressed as a percentage of total fatty acids identified and mg/g muscle raw tissue of fish, and grouped as follows: saturated fatty acid (SFA), monounsaturated (MUFA), polyunsaturated fatty acid (PUFA), n-6 and n-3. The PUFA/SFA, DHA/EPA, ∑n-6/∑n-3, atherogenicity (IA) and thrombogenicity (IT) indices were also calculated. IA indicates the relationship between the sum of the main saturated fatty acids and that of the main classes of unsaturated, the former being considered pro-atherogenic (favoring the adhesion of lipids to cells of the immunological and circulatory system), and the latter anti atherogenic (inhibiting the aggregation of plaque and diminishing the levels of esterified fatty acid, cholesterol, and phospholipids, thereby preventing the appearance of micro and macro coronary diseases). Finally, IT shows the tendency to form clots in the blood vessels.

IA and IT indices were calculated by using the Ulbricht & Southgate (1991) equations as follows:

Approximately 1 g of fish raw meat was subjected to the wet mineralisation by Kjeldahl method using a mixture of nitric and sulphuric acid (2:1, w/w) according to Alasalvar et al. (2011) . Mineral contents were determined by plasma absorption spectrometer using a 200-DV (Perkin-Elmer, Waltham, USA). The following elements were measured: potassium (K), calcium (Ca), magnesium (Mg), manganese (Mn), phosphorus (P), iron (Fe), zinc (Zn) and copper (Cu). Analyses were determined in duplicate, according to the mean value of two determinations and expressed in mg per 100 g of fish raw meat.

A total of 100 fish flesh samples were analysed for different parameters. Normal distribution was checked for all data with Kolmogorov-Smirnoff test and homogeneity of the variances with the Levene test. After verification of normal distribution, the effect of the production system (wild and cultured) on carcass and fillet characteristics, fatty acid composition and nutritional value was evaluated using one-way ANOVA with the production systems as the fixed effect. Statistical treatment of the data was done by calculating means and standard error of mean. Differences were considered statistically significant at p <0.05. The statistical data were obtained using SPSS software, version 15.0 (IBM, Chicago, IL, USA).

Body yield of C. festae under two production systems is shown in Table 1. The coefficient of variation (data not show) ranged from 12.66% for flesh to 39.97% for guts, with mean values of 14.83% and 15.60% for head and skin + bones, respectively. One way ANOVA showed differences ( p <0.05) among head, skin + bones and flesh percentages of cultured and wild fish samples. The percentage of head was significantly ( p <0.05) higher in wild fish, while the percentages of skin + bones and flesh were higher in cultured fish. However, no significant differences were found among guts percentage, slaughter yield and dress-out. The edible portion represents 29.2% of fish weight, higher in cultured compared to wild.

The flesh quality characteristics of C. festae are shown in Table 1. The coefficient of variation did not exceed 3% at pH values, it was about 10% in L*, 21% by cooking loss, 35% for drip loss and exceeded 45% in a* and b* (data not show). The pH postmortem was the main factor which influences the quality of meat. In the first one hour postmortem , the pH dropped from 7.15 to 6.72, 7.13 to 6.74 and 7.17 to 6.69 in the total population, cultured and wild fish, respectively. This fall took place mainly in the first 2 h postmortem (from 7.15 to 6.73, 7.13 to 6.72 and 7.17 to 6.74, respectively). Chromatic variables ( L *, a *, b *), drip loss and cooking loss ranged from 41.8 to 71.8, from 1.35 to 9.91, from -2.44 to 11.34, from 1.09 to 4.81 and from 19.44 to 47.27, respectively (data not shown). L *, a *, b * values indicate a pale meat with high L * value (have a tendency to white), low a * value and low b * value. None of the variables showed significant differences ( p >0.05) among populations.

The results of proximate analysis of muscle tissue of C. festae samples are shown in Table 1. The cultured fish had higher ( p <0.05) crude protein, total fat and ash percentages compared to wild specimens. There was no significant difference of wet in muscles between cultured fish with wild fish.

The fatty acid profiles of cultured and wild C. festae are listed in Table 2. In the present study, more abundant saturated fatty acids were palmitic (27.91%), stearic (9.0%) and myristic (7.12%) fatty acids. Monounsaturated oleic acid was the most abundant (19.74%) fatty acid in fish muscle samples, and linoleic acid was the most abundant (8.19%) polyunsaturated fatty acid. Muscle tissue of C. festae included 55.7% saturated, 21.2% monounsaturated and 23.2% polyunsaturated fatty acids.

Besides, significant differences ( p <0.05) were found in the content of most of the fatty acids analyzed between cultured and wild fish except for caprylic, lauric, palmitic, SFA, DHA and arachidonic contents which were similar ( p >0.05) between systems.

Table 3 shows the trace mineral composition of the meat of cultured and wild C. festae . P, K and Ca were predominant elements among eight minerals analysed and constituted 95.3% and 95.4% of total trace minerals content in cultured and wild C. festae , respectively. P, Ca, Mg were higher ( p <0.05) in cultured fish compared with wild fish. However, the content of K, Fe and Zn were lower in cultured fish. No significant differences ( p >0.05) were found in Cu and Mn content.

The skeletal muscle (fillet or flesh) is the major part of the edible portion of fish and flesh yield depends on the species, sex and size, structural anatomy of the fish and farming conditions. In the present study, flesh yield was much lower than those presented for tilapia ( Oreochromis sp.) (33%) and channel catfish ( Ictalurus punctatus ) (>38%) and similar to the 32% reported by Neto et al. (2012) in Pacu ( Colossoma bidens ) and Tambaqui ( Colossoma macropomum ) fish. Based on production systems, the flesh yield obtained was higher in cultured fish than wild fish (Table 1) and similar than those presented by Sulieman & James (2011) for Nile Tilapia ( Oreochromis niloticus ) who reported values higher in farmed fish than in wild fish (37.1% and 32.2 % respectively). However, the results were lower than those presented by Rørå et al. (2001) who indicated values between 40% and 70% in cultured system. The values for flesh yield and dress-out did not agree with Intarak et al. (2015) who indicated that dress-out and flesh yield did not change with slaughter weight.

The differences in yield of different part of body could have been caused by the slaughter weight and to food. In addition, factors inherent to cichlids such as the anatomical body shape, head size and final weight influenced in the yield body in these species ( Rojas-Runjaic et al. , 2011 ).

The values for pH postmortem , chromatic variables (L*, a*, b*), drip loss and cooking loss were not influenced by production system (Table 1). The pH postmortem is the most important factor which influences the texture of meat; minor changes in pH impact dramatically the connective tissue properties ( Huss, 1995 ). The pH could be considered as an appropriate index of quality control of fish meat. In this study, the drop of pH within 12 h after death was in accordance with Robb et al. (2000) and Roth et al. (2009) who reported that muscle pH displayed a rapid decline in muscle pH during the first 12 h postmortem , then between 12 and 24 h decline was less pronounced and from 24 h pH did not show variations, indicating that the fish were reaching its end pH.

The values obtained for drip losses and cooking losses were higher than those found by Wangtueai & Vichasilp (2015) in Nile tilapia, who reported cooking and drip losses values fluctuating between 17.78% - 26.37% and 0.7 - 3.95%, respectively. Intarak et al. (2015) , in Panga fish ( Pangasius bocourti Sauvage), reported values of 4.88% to 2.88% for drip loss, with significant decreases with increasing live weight.

L*, a* and b* value indicate a pale meat and the results of our study were greater that Lima et al. (2015) in Nile tilapia who reported value of L* 47.53 and 52.77 respectively. As expected, a* and b* values registered in these studies (25.9 - 27.6 and 25.5 - 27.2, respectively) were much higher than those recorded in our study.

The proximate composition of fish is affected by a diversity of factors such as: size, temperature, salinity, production system and feeding among other ( Gonzalez-Artola, 2004 ). The normal variations between the constituents in fish are: 66 - 81% for wet, 16 - 21% for protein, 0.2 - 25% for fat and 1.2 - 1.5% for ash ( Chandrashekar & Deosthale, 1993 ).

The results obtained in this study were similar that those obtained for Nile tilapia ( Michelato et al. , 2016 ). In a study in Turkey of cultured Oncorhynchus mykiss ( Mashaii et al. , 2012 ), the protein (17.4 - 17.9%) and ash (1.38 - 1.59%) contents were similar than our results, while the moisture (75.2 - 76.9%) and total fat (2.48 - 4.88%) contents were lower and higher, respectively. In general, the wet contents in the muscle fish of this study were within the range reported by Jabeen & Chaudhry (2011) and Campagnoli de Oliveira et al. (2010) . Fat content was affected by production systems and this variation in the total fat content was consistent with the conclusions of Gonzalez-Artola (2004) and Jabeen & Chaudhry (2011) . Fish is often classified on the basis of their fat content into lean fish (fat<2%), low fat fish (fat = 2 - 4%), medium fat fish (fat = 4 - 8%) and high fat fish (fat>8%) ( Gonzalez-Artola, 2004 ). Based on this classification, C. festae is considered a lean fish (Table 1).

In this study, crude protein of 17.33% wet weight was similar than protein levels for common carp ( Cyprinus carpio ) (16% wet weight; FAO, 2008 ) but much lower within the range of 18.64 - 22.7% and 18.4 - 20.8% reported for the Cichlidae family by Perea et al. (2008) and Sulieman & James (2011) and respectively. According to Hernandez & Aguilera (2012) ash content (1.36% wet weight) observed in this study was located within the ranges reported and was affected by production systems.

Fish suffers changes in body composition in response to diet and environmental conditions, and the differences found in fatty acid composition between production systems could be attribute to differences between condition in captivity and in the wild.

More than 50% of the SFAs identified, palmitic acid (C16:0) was the most abundant, ranging from 26.8% to 28.9%. Other studies also reported this fatty acid as the most abundant ( Aggelousis & Lazos, 1991 ). Caproic (C6:0), capric (C10:0), myristic (C14:0), pentadecanoic (C15:0), stearic (C18:0) and arachidic (C20:0) acids were significantly different between systems but no significant differences were found for total SFAs. Oleic acid was the principal MUFA in both systems. Difference was observed between cultured and wild fish. Alasalvar et al. (2011) also reported oleic acid as the most abundant of the MUFAs.

Regarding to PUFA content, linoleic acid (C18:2n - 6) was the dominating fatty acids according to results found by Jabeen & Chaudhry (2011) for Cyprinus carpio , Labeo rohita and Oreochromis mossambicus in Indus River (Pakistan). Linoleic acid values were different ( p <0.05) among systems, being higher in cultured system.

In n-3 family fatty acids, EPA, DHA and linolenic acid were the most important as Romero et al. (2000) and Mashaii et al. (2012) indicated. According to Leaf & Weber (1988) , DHA and EPA are key components for a healthy diet in humans. Although in this study DHA and EPA values were low, it is important to highlight that, contrary to expectations, wild fish had higher EPA content (22.89 mg/g). Finally, suitable choice of dietary lipid in cultured fish will allow improve the fatty acids profile, especially in n-3 PUFAs. In n-6 family fatty acid, linoleic acid was the most important and arachidonic acid (C20:4 n-6) content was higher in cultured fish ( p <0.05) and this fatty acid is precursor for prostaglandin and thromboxane biosynthesis aiding the blood clotting process during wound healing ( Jabeen & Chaudhry, 2011 ).

In our study, the SFA content was higher than unsaturated fatty acids. Higher amounts of SFA in the present study might be assumed as a disadvantage of these fish. The n-3/n-6 ratio in this study (0.86) was similar than those found by Hoseini et al. (2013) in farmed Big head carp ( Hypophthalmichthys nobilis ) and Grass carp ( Ctenopharyngodon idella ). An increase in the human dietary of n-3/n-6 fatty acid ratio is essential in the diet and nutritionists believe that this ratio should be 0.1 - 0.2 and consider higher ratios (>0.2) more beneficial to human health ( FAO/WHO, 1994 ). Simopoulos (2008) suggested that the n-3/n-6 ratio should be kept between 1:1 and 1:4. Amount of n-3/n-6 in cultured fish was 0.72 and in wild fish 0.99; differences were observed between two production systems ( p <0.05). Those results could be due to wild fish probably ingests higher rate of natural foods containing more EPA and DHA.

IA and IT indices are indicators of flesh lipids quality, and were calculated to determine the potential health impact on human consumers. In our study, the mean values of IA and IT indices were 1.39 and 0.86, respectively; higher than those values recorded by Hoseini et al. (2013) .

According to Alasalvar et al. (2011) the concentration of trace minerals in fish is influenced by numerous factors such as seasonal and biological differences, food source and environment. In this study, P, K and Ca were predominant elements among eight minerals analysed and constituted 95.3% and 95.4% of total trace minerals content in cultured and wild C. festae , respectively. The results obtained were similar than those obtained by Ravichandran et al. (2012) ; P, Ca, Mg, were higher in cultured than wild fish, however K, Fe and Zn were lower in cultured fish. P values were similar than those found by Perea et al. (2008) for Nile tilapia with value fluctuating between 191mg/100 g and 285mg/100 g. Adeniyi et al. (2012) found K values higher in wild fish than farmed fish, in line with the results obtained in this study (farmed fish: 107.8mg/100 g, wild fish: 94.28 mg/100 g). Moreover, Ca and Mg values were higher than those reported by Mogobe et al. (2015) in different freshwater species ( Marcusenius altisambesi, Schilbe intermedius, Brycinus lateralis, Oreochromis andersonii, Barbus poechii ). In the present study, the Fe values obtained were in consonance with FAO (2001) , who indicated adequate range from 0.23 mg/100 g to 2.1 mg/100 g. Mazumder et al. (2008) defined the decreasing order of magnitude (Zn>Fe>Mn>Cu) which is evident in most of the fishes, while in our work, the Cu content exceeded that of Mn.

Hence, rearing system significantly influences most of the analyzed characteristics of carcass and flesh of C. festae . Overall, cultured C. festae is desirable for its greater flesh yield and appears to be the best diet for its higher content of protein, fat, MUFA, PUFA and PUFA/SFA ratio, and lower SFA content and IA and IT indices. Finally, from a nutritional point of view, C. festae flesh presents very desirable characteristics for human consumption.