Fish and experimental conditions
Two hundred tomato clownfish (Amphiprion frenatus) juveniles were obtained from a commercial aquarium located in Toluca, Mexico. Prior to the experiments, fish were acclimated and maintained in the Aquaculture lab facilities of the Nayarit Centre for Innovation and Technology Transference (CENITT, A.C.). A total of 144 juvenile fish, with a mean weight and length of 0.45 ± 0.14 g and 2.81 ± 0.33 cm, were randomly selected and distributed into twelve 65 L− 1 rectangular tanks (16 fish per tank). Tanks were connected to a recirculation aquaculture system, which was composed of mechanical and biological filtration. Levels of ammonia, nitrites, and nitrates were monitored every two days with an aquarium set kit (API Marine kit®; USA) to keep values of NH3 < 0.01 mg L− 1, NO2 − < 0.3 mg L− 1, and NO3 − < 10.0 mg L− 1, respectively. Water temperature and dissolved oxygen were measured daily in the morning using a multi-parameter oxygen meter (Extech® Model 407510; USA) to maintain temperature between 26–28°C and dissolved oxygen within 5.0–6.5 mg L− 1. Photoperiod was fixed to a light: dark period of 14:10 (summer season of Tepic, Mexico).
Experimental Diets
Three commercial aquaculture diets (Table 1) were used to evaluate growth performance of juvenile clownfish and their behavioural patterns under stressful situations. A diet with 40% of crude protein (CP) and 9% of crude fat (CF) defined as “A” (formulated for marine ornamental juveniles fish species), a diet with 45% of CP and 14% of CF defined as “B” (formulated for aquacultured marine finfish species), and a diet with 38% of CP and 6% of CF defined as “C” (formulated for aquacultured omnivore fish species). Selected diets were homogenized to 1mm, according to the mouth size of the fish. To make an neutral judgment of results, employed diets were not revealed (following criteria of Chambel et al., 2015).
Table 1
Proximate composition of the experimental diets (dry matter basis).
Variables | Experimental diets |
A | B | C |
Dry matter % | 92.7 | 95.5 | 93.9 |
Crude protein % | 40.3 | 45.3 | 38.5 |
Crude Fat % | 9.0 | 15.0 | 6.0 |
Ash % | 7.7 | 9.9 | 15.6 |
Energy content (MJkg-1) | 20.22 | 19.93 | 18.29 |
Nitrogen free extract + fiber % | 41.7 | 19.8 | 39.8 |
Feeding Trial And Sample Collection
Juvenile fish were hand-fed to apparent satiation twice a day (at 8:00 and 13:00 hours) for six weeks. To accurately estimate the amount of consumed food per tank and to maintain optimal water quality, uneaten food observed at the bottom was siphoned-out, dried, and weighted, then subtracted from the total amount of food offered. Juvenile´s growth was evaluated at the beginning and every week after the feeding trial by measuring the total length with a caliper (Luzeren®; 0–6”, ± 0.01) and weighted with a laboratory scale (Adam® Model HCB 1002; 1000 g x 0.01g). All fish from each tank were sampled and used for morphometric measurements.
Chemical Analysis And Calculations
The proximate composition (%) of the experimental diets was determined using established methods (AOAC, 1995). Crude protein was determined by the micro-Kjeldahl method, where total N was multiplied by a 6.25 factor (according to general animal feeds). Crude fat was determined by solvent extraction using petroleum ether in a Soxhlet ex-tractor. Ash was estimated by incinerating the samples for 8 h in a muffle furnace at 550°C. The nitrogen-free extract (NFE) was calculated by subtracting the added percentage of protein + fat + ash from 100%. In order to evaluate the growth performance of juvenile tomato clownfish, the variables of initial weight (g) and length (cm), final weight (g) and length (cm), weight gain (g), relative weight gain (%), condition factor, and specific growth rate were estimated. Gross energy content of diets were assessed based on their crude protein level, total fat (ether extract) and nitrogen free extract (NFE) contents using the equivalents of 23.64, 39.54, and 17.15 MJkg − 1 (Blaxter, 1989) (Table 1).
The following formulas were used to obtain the value of the variables:
Survival (S %) = (number of fish in each group remaining on day 84/initial number of fish) × 100
Weight gain (WG) = (final weight (g) – Initial weight (g))
Relative weight gain (WG %) = [(final weight (g) – initial weight (g)) / initial weight (g)] x 100.
Condition factor (K) = 100 [(body weight (g) / length3 (cm))]
Specific growth rate (SGR) = 100 x (ln Final body weight – ln Initial body weight)/42 days.
Stress Assay
Forty-eight tomato clownfish juveniles (16 per treatment) were subjected to a two common stress tests used in aquaculture: novel environment test and a mirror test (Castanheira et al., 2015). Both tests were applied in sequence on the second and fifth week of the experiment, following the criteria described by Ibarra-Zatarain et al. (2020). The novel environment assay (Fig. 1A) consisted on exposing the fish to a plastic reservoir (30 cm length x 19 cm width x 15 cm high) filled with water (WL = water level; Fig. 1A) from the rearing tank (renewed for every fish). The reservoir was gridded (squares equal to 3 cm2) and a camera (Swann®; 2Kseries-1080p; model SWDKV-845804) was fixed 40 cm above the reservoir to video record juvenile’s fish activity during 180 sec. Two behavioural variables were analyzed for each organism: (i) total distance moved (cm2), determined by counting the total number of squares visited by each fish; (ii) the minimum, medium and maximum speed (cm/s) of fish during the test (adapted from Hosoya et al., 2008; Benhaïm et al., 2012; Ibarra-Zatarain et al., 2015).
The mirror test (Fig. 1B) consisted on placing the fish in a plastic container (20 cm length x 14 cm width x 9c m tall), in which a mirror was installed in one of the walls of the cointainer in order to reflex the fish's image (M = mirror; Fig. 1B). Likewise, the container was divided into four equal areas (dotted blue line; Fig. 1B). The test lasted 180 s, and two parameters were analyzed: (i) the preferred area (%) by fish during the test (area 1 close to the mirror - area 4 far from mirror); (ii) total number of attack attempts made by each fish, denoted by the erection of dorsal fins and bites (adapted from Way et al., 2015; Kohda et al., 2019).
Behavioural Analyses
All behavioural parameters from both tests were analysed with a processing image program in MATLAB (V2015a). Overall, videos were transformed to RGB24 (Red-Green-Blue) and channels red and green were used to contrast the fish from the background. Once the fish were contrasted, the video analysis was performed in four stages: 1) Frame lecture: a videoreader function in MATLAB was utilized for a sequential analysis, for this the videos were adjusted to a 480 x 460 resolution in RGB24 (red x green x blue) and each pixel was transformed to 8 bit; 2) Processing of video: red and green colors were used to contrast fish and reservoir, then a fit-geotrand function was applied to correct frames and improve the projection of images; 3) Fish identification: a local threshold segmentation method was applied for the fish identification, for this, the threshold was contrasted between red and green channels with the regionpro-pos function by estimating the centroid and perimeter of localized objects (fish) to reduce errors; 4) Movement analysis: the grids centroids (squares equal to 3 cm2) and video frames were set in order to analyze the movement of fish corcerning to centroid of grids. Methodology and image processing were based on Radke (2009) and Chitradevi and Srimathi (2014).
Statistics
All data were tested for normality using the Kolmogorov-Smirnov test and homoscedasticity with the Levene's test. Subsequently, a multivariate analysis of variance (MANOVA) was applied to determine differences among diet treatments and between the novel environment test (distance moved and velocity) and mirror test (zone of preference and number of attacks) variables. A Fisher’s Least Significant Difference (LSD) was performed when significant differences were found. To analyze the consistency of behavioural responses of both stress tests a student t-test was applied. Lastly, a Pearson´s correlation analysis was runned between weight and length with all behavioural responses from novel environment and mirror test. To determine the consistency of test 1 and test 2, the means of behavioural variables were compared with a Student-t test. Results were expressed as mean ± SD, and significant differences were determined when P < 0.05. Statistical analyses were runned with SPSS V25.