Feeding Trial

An experiment was carried out with a population of juvenile shi drum (U. cirrosa) specimens obtained in May 2018 from a stock of spawners adapted to captivity at the Instituto Español de Oceanografía (Centro Oceanográfico de Murcia, Mazarrón, Spain). Juvenile specimens (7 months old) weighing 18.98 ± 1.20 g (mean ± SD) and measuring 11.80 ± 0.50 cm (mean total length ± SD) were initially fed with a commercial standard diet (Skretting España SA). At the beginning of the study (day 0 of the experiment), 270 specimens were randomly distributed into 2 feeding groups (135 fish per group). Each group was distributed in three 170-L tanks (45 fish per tank) in a flow-through system. Water flow rate in each tank was about 250 L/h. Natural photoperiod conditions were applied throughout the experiment, which varied between 10L:14D and 11L:13D. Light intensity was 250 lux. Rearing temperature was kept between 18 and 20 ºC to ensure food intake by the fish.

Experimental Diets

The control fish received a commercial standard diet (Skretting España SA) (group C) and the experimental fish received a vegetal-based diet (group V). The standard or conventional diet (diet for group C) was formulated and manufactured for gilthead sea bream, but it was used for the shi drum because there is still no commercial feed for this species. The vegetal diet (diet for group V) was formulated and delivered by Biomar (Denmark) and has been used for juvenile sea bream in previous studies (^{Benedito-Palos et al. 2016}, ^{Simó-Mirabet et al. 2018}, ^{Perera et al. 2019}). The diet for group V included partial replacement of fish meal and fish oil with vegetal products at an 84% inclusion level. Pellet size was 2 mm in both the diet for group C and the diet for group V.

The ingredients in the diet for group C were fish meal, fish oil, oilseeds, cereal grains, blood products, chicken flour, soybean and rapeseed oils, calcium, phosphorus, vitamin- mineral mix, and antioxidant and antifungal additives. The ingredients in the diet for group V were fish meal, fish oil, soy protein, corn gluten, wheat gluten, rapeseed cake, wheat flour, rapeseed oil, palm olein, monocalcium phosphate, histidine, yttrium, vitamin-mineral mix, amino-acid and micronutrient mix, and antioxidants. The proximate compositions and percentages of the main ingredients in the diets are shown in Tables 1 and 2. For details on the composition of the diet for group V see also ^{Benedito-Palos et al. (2016)}. Experimental diets were fed ad libitum 3 times a day for 62 d (end of the experiment).

Samplings

Sampling was carried out on day 0 of the experiment (when specimens were 7 months old and had not yet been classified into 2 different dietary groups), on day 27 (when specimens were 8 months old), and on day 62 (when specimens were 9 months old and the experiment came to an end). Fish body length and weight were individually measured on sampling days. The following parameters were also calculated after sampling on day 27 and day 62: feed conversion rate (FCR): total amount of consumed feed/weight gain; specific growth rate (SGR) (% per day): 100 × [(ln final weight - ln initial weight)/days]; condition factor (CF) (g·cm^-3): 100 × (body weight/lenght³ ); and daily intake rate (DIR) (% per day): {supplied feed/[(initial weight + final weight)/2]days)} × 100. For the final sampling, the organosomatic indices (hepatosomatic index, HSI; viscerosomatic index, VSI; and intestine length index, ILI) were also determined from 5 fish per tank (15 fish per group). Each organosomatic index was calculated as follows: HSI (%) = 100 × (liver weight/fish weight); VSI (%) = 100 × (digestive weight/fish weight); and ILI (%) = 100 × (intestinal length/ fish length).

Muscle parameters were measured from 10 fish on day 0, 12 fish per group (4 fish per tank) on day 27, and 12 fish per group (4 fish per tank) on day 62. Percent survival was calculated for both groups at the end of the experiment. On each sampling day, the specimens used for the muscle analysis were slaughtered by an overdose of anesthesia with 60 ppm of clove oil and then shipped to the Faculty of Veterinary Sciences at Murcia (Spain).

Quantitative analysis of muscle growth

After obtaining the body length and weight parameters, cross sections of the specimens were made and whole body slices 5 mm thick were obtained. The entire muscle cross sections from every fish were photographed for measurement by a morphometric analysis system (Sygma-Scan Pro_5). These body slices were subsequently cut into smaller blocks and then snap-frozen in 2-methylbutane over liquid nitrogen. Sections 8 μm thick were later obtained from those frozen blocks with a cryostat (Leica CM 1850) and stained with hematoxylin-eosin for morphometric studies. Muscle growth was quantified by means of the morphometric analysis system. The following parameters were measured: total cross-sectional area of the white muscle, number of white muscle fibers, size (area and minor axis length) of white muscle fibers, and muscle fiber density (number of white fibers/μm² ). Average size was estimated from ~600 fibers (±10 SD) located at the intermediate and apical sectors of the epaxial quadrant of the myotome cross section according to the methodology described by ^{Ayala et al. (2013}, ²⁰¹⁵) for this species.

Statistical analysis

All statistical analyses were performed using SPSS Statistics 24. Mean and standard deviation (SD) values were calculated for each data group. Data distribution for each sampling day was analyzed by the Shapiro-Wilk test for P < 0.05. Data for size of fibers did not show a normal distribution (P < 0.05) and Levene’s test did not show homogeneous variances (P < 0.05) either. Therefore, nonparametric tests were used (Mann-Whitney and Kolmogorov-Smirnov tests) to evaluate the effect of the diet on the size of the fibers (P < 0.05). For most of the other parameters, both the ShapiroWilk test and Levene’s test showed values of P > 0.05 and the analysis of variance was therefore used; however, nonparametric tests were used in cases where P < 0.05.

Body growth and organosomatic indices

Before the start of the experiment, all specimens had been reared under the same culture conditions. At the beginning of the experiment (day 0), the mean weight and mean total body length values (±SD) for these specimens were 18.98 g ± 1.20 and 11.80 cm ± 0.50, respectively. On day 27, fish body lengths were similar in groups C and V (P > 0.05), whereas body weights were significantly lower in V than in C (Table 3). On day 62 of the experiment (fish 9 months old), results were similar to those found on day 27. Therefore, fish body lengths were similar in both groups (P > 0.05). Likewise, FCR and SGR values were similar in both groups (P > 0.05), but DIR values were significantly lower in V than in C (Table 3).

The HSI, VSI, and ILI values were similar in both groups, with no significant differences (P > 0.05) (Table 3). Percent survival was 97.3 (±2) and 99.3 (±0.7) for C and V, respectively (P > 0.05). Occasional deaths did occur during handling independent of the diets.

Muscle growth

The muscle cross section from a 7-month-old specimen, at the beginning of the experiment (day 0), showed the typical morphological mosaic of postlarval and adult specimens, with the small fibers interposed with the big fibers (Fig. 1). On day 27 of the experiment, values for the transverse area of the white muscle were lower in group V than in group C (P < 0.05) (Table 4). Regarding muscle cellularity, muscle hypertrophy was highest in group C (P < 0.05) (Table 4; Fig. 1b, c). In contrast, the number of white fibers and muscle fiber density were higher in V than in C, and the difference was significant for muscle fiber density (P < 0.05). On day 62 of the experiment, the mean value for the transverse area of the white muscle was lower in V than in C, but it was not statistically significant (P > 0.05) (Table 4; Fig. 1d, e). Regarding muscle cellularity, hypertrophy was significantly higher in V than in C; however, hyperplasia and muscle fiber density were higher in C and the difference was significant for muscle fiber density (P < 0.05).

Figure 1 Cross sections of white muscles from 7- (a), 8- (b, c), and 9- (d, e) month-old Umbrina cirrosa specimens. Muscle sections in (b) and (d) correspond to specimens from the control group. Sections in (c) and (e) correspond to specimens from the vegetal group. Hematoxylin and eosin staining. W, white muscle fibers; nW, new white muscle fibers. 

Growth, feed conversion rate, specific growth rate, daily intake rate, and survival

The final body weight of fish was higher in group C, which was parallel to the higher DIR, in comparison with group V. However, SGR and FCR were similar in both experimental groups. FCR for this species is generally 1.28-1.44 during the pre-fattening period (^{Mylonas et al. 2009}). Our FCR values fell within this range for the 9-month-old specimens and were even slightly lower for the 8-month-old specimens (Table 3), indicating both types of feed were efficient. The percentages of protein and fat in both types of feed are suitable for this species according to ^{Segato et al. (2005)}, and this may be reinforced, in part, by the data obtained in the present study. Also, our results seem to indicate that the composition of both feeds (amino acid profile, fatty acids, etc.) provides the essential nutrients for this species.

^{Segato et al. (2005}, ²⁰⁰⁸) fed 2 groups of juvenile shi drum with 2 different diets, the first containing high contents of fish meal (47.5%) and fish oil (16.5%) and the second containing a lesser amount of fish meal (41.0-43.0%) and fish oil (13.0-14.0%) by including vegetable ingredients. These authors recorded similar values of body weight, SGR, FCR, and DIR for fish fed both diets. The protein and fish oil levels used in our experiment were lower than those used by ^{Segato et al. (2005}, ²⁰⁰⁸). The low acceptance of the vegetal diet in our experiment seems to indicate an excessively high level of vegetable ingredients. For the meagre (Argyrosomus regius), a sciaenid of similar characteristics to the shi drum, ^{Ribeiro et al. (2015)} studied the effect of the replacement of 50% fish meal and up to 100% fish oil with plant-based ingredients on the growth of juvenile specimens over a course of 88 d and found that growth and feed efficiency were comparable in all diets. For juvenile gilthead sea bream, ^{Benedito-Palos et al. (2016)} studied the effect of diets with low fish meal and fish oil contents in juveniles over a course of 8 months; the control diet (D1) contained 23% fish meal and the experimental diets (D2, D3) contained only 3% fish meal. These authors found that fish oil in the D1 diet was 15.60%, which was reduced to 6.56% in D2 and 2.5% in D3. The D1 and D2 groups studied by Benedito-Palos et al. (2016) did not show differences in growth or food efficiency during almost the entire experimental period but group D3 showed lower dietary efficiency and less growth during the first 4 weeks of the trial, which was later partially compensated and resulted in a final weight being only 6-7% lower in D3 than in the control. According to ^{Benedito-Palos et al. (2016)}, a very high level of fish meal and fish oil replacement is possible provided that the theoretical requirements for essential nutrients are met; however, the most extreme replacement levels (D3, 3% fish meal and 2.5% fish oil) would require an adaptive period to avoid initial and transient detrimental effects on growth performance. The D3 diet used by ^{Benedito-Palos et al. (2016)} was also tested in our study (the diet for group V), and the lower final weight values for fish in group V may indicate that fish require more time to get fully adapted. However, this issue remains to be studied. Results similar to those found by ^{Benedito-Palos et al. (2016)} were also found for gilthead sea bream juveniles by ^{Simó-Mirabet et al. (2018)} and ^{Perera et al. (2019)}. Mortality in our study was not affected by dietary treatments, which is consistent with the findings of other authors (e.g., ^{Segato et al. 2005}).

Diet influence on organosomatic indices

In our experiment, organosomatic indices were similar in both groups, which indicates good adaptation of the digestive system and liver to the absorption and metabolism of vegetables. Similarly, in the study by ^{Segato et al. (2005)}, HSI did not change significantly among groups fed different diets; however, the vegetal diet produced an increase in VSI, probably due to the larger intestines. The fact that the vegetable percentages in the diets used by ^{Segato et al. (2005)} are different from those employed in our diets may explain the differences in the results between both studies. ^{Benedito-Palos et al. (2016)} also showed that dietary treatments did not have significant effects on organosomatic indices, although they reported the highest HSI value for fish in group D3, for which ILI was also slightly higher with respect to fish from the other groups. ^{Perera et al. (2019)} also discovered that the intestine was enlarged in fish fed a vegetal diet and they assumed this to be an effective adaptation mechanism to preserve maximum growth.

Influence of the diet on muscle growth and muscle cellularity

Growth in fish involves muscle fiber recruitment and hypertrophy. The number of muscle fibers recruited to reach a particular girth varies between families and strains and is influenced by environmental factors including diet, exercise, light, and temperature regimes (^{Johnston 1999}, ^{Ayala et al. 2003}, ^{López-Albors et al. 2003}, ^{Johnston et al. 2011}). In the present study, the transverse area of the white muscle and white fiber hypertrophy were higher in group C than in group V after 27 days. Similar results have been previously found for S. senegalensis by ^{Valente et al. (2016)}, who used diets with increasing percentages of dietary plant ingredients (50%, 75%, and 100%). In the study by ^{Valente et al. (2016)}, fish fed diets with 50% and 75% vegetable proteins did not show differences in body or muscle growth, but fish on 100% vegetal protein showed significantly less body growth (length) and a lower transverse area of the white muscle and its fibrillar size, as well as showing lower fillet firmness; however, the number of fibers (hyperplasia) was similar in all groups. In contrast, our results showed significant differences in hyperplasia and muscle fiber densities, such that both muscle parameters were higher in group V than in group C on day 27. At the end of our experiment, muscle cellularity showed an opposite trend to that observed for day 27. These results suggest that the vegetal diet influenced the relative contribution of hypertrophy and hyperplasia to myotome growth in shi drum and that hypertrophy was the most significantly influenced parameter.

Muscle cellularity influences flesh characteristics, particularly texture, such that muscle fiber density positively correlates with flesh firmness, as has been observed for the shi drum (^{Ayala et al. 2015}) and other fish species (^{Hatae et al. 1984}, ¹⁹⁹⁰; ^{Periago et al. 2005}). Hence, the resulting variation in muscle cellularity in groups V and C can lead to differences in fillet texture at harvest size. Long-term studies would be necessary to verify this hypothesis.