JOURNAL OF THE Vol. 31, No. 2 WORLD AQUACULTURE SOCIETY June, 2000

Growth and Feeding Level in Pre-Weaning Tambaqui Colossoma macropomurn Larvae

ADRIAN SEVILLA AND JORGE GUNTHER' Escuela de Ciencias Bioldgicas, Universidad Nacional, Apartado postal 86-3000, Heredia.

Costa Rica

Abstract.-The relation between growth, feed utili- zation and feed ration was assessed during the first 15 feeding days in Colossoma macropomum larvae fed with Artemia nauplii. Maintenance, optimum and max- imum feeding rations (1.23; 5.48 and 27.2% body weight/d, dry feed, respectively) and feed conversions (0.4 and 0.78 optimum and maximum, respectively) were determined. At maximum feeding the larvae grew 2.6 times faster that at optimum feeding, but the feeding costs increased 1.94 times. 'bo commonly used growth rates (specific growth rate, SGR and co- efficient G ) were compared for the modelling of larval growth. The maximum specific growth rate (SGR, growth rate at maximum feeding) remained relatively constant during the first 15 d of feeding and is thus much more adequate to model the growth of tambaqui larvae, than the coefficient G, which increased more than four times in this period.

The omnivorous serrasalmids of the ge- nus Colossoma are important food fishes in tropical South America. C. macropomum, (tambaqui in Brazil, cachama negra in Ven- ezuela and Colombia), is the most impor- tant fluvial catch in the Amazonas basin (Saint Paul 1986). Culture of tambaqui has been reported from Brazil, Colombia, Peru and Venezuela, with steadily increasing vol- umes (Hernhdez et al. 1992). The tamba- qui has been introduced also in several Cen- tral American countries where the commer- cial production is incipient (Pretto 1989; Van Anrooy et al. 1996). Overall produc- tion in Latin America has attained 7,000 metric tons in 1996 (New 1999).

Positive traits of tambaqui for aquacul- ture are its great size, high growth rate, gre- garious habits, resistance to very low oxy- gen and excellent feed utilization (Saint Paul 1986; Gunther and Boza 1992; Her- nhndez et d. 1992; Gonzfiles and Heredia

'Corresponding author.

1993). Occurrence of intramuscular bones and high susceptibility to skin infection have been reported on the negative side (Bello and Rivas 1992; Van Anrooy et al. 1996). Tambaqui in ponds do reproduce only after hormonal treatment, the fecun- dity is very high, and big females can pro- duce up to several million buoyant eggs. Rearing of the larvae can be achieved either intensively in a hatchery (De Fex 1991) or extensively in prefertilized ponds (Hernh- dez et al. 1992). In the laboratory and be- cause of the small larval size (about 1 mg weight at hatching), feeding with live feed (e.g., Artemia nauplii) is mandatory before the fish can be weaned to artificial feeds at about 100 mg weight (unpublished obser- vations).

Because of the high cost of the Artemia nauplii, it is important to establish the most cost efficient feeding level of the tambaqui larvae in relation to growth rate. The aim of this study is to determine the relation be- tween feeding level and growth rate in pre- weaning tambaqui larvae.

Materials and Methods

Fish

Nine-day-old (posthatch) larvae obtained through induced reproduction were used. The larvae had been starving for 2 d after yolk absorption at experimental start. The initial mean fresh weight and dry weight data were obtained individually from a batch of 25 larvae.

Experimental Facility

An array of 20 small (1 5-L) aquaria was used for the experiment. The aquaria were connected to a recirculation unit with tem-

0 Copyright by the World Aquaculture Society Zoo0

218

GROWTH OF TAMBAQUI LARVAE 219

perature control (28 2 1 C), sedimentator and biological filters. The waterflow through the aquaria was approximately 1 L/ min. In order to avoid accidental feeding with nauplii circulating in the system, the water supply was screened through 50-pm mesh in the aquaria with the four lower feeding levels.

Experimental Design Seven feeding levels, each repeated

twice, were tested. Feeding levels were de- fined using the growth model G (Hogen- doom 1980) and assuming the following values for the growth coefficient: 0 (starv- ing, treatment TOO), 0.03 (T03), 0.07 (T07), 0.15 (T15), 0.25 (T25), 0.35 (T35) and 0.50 (T50). For the initial calculations of the ra- tion dry basis feed conversions of 3, 2.5, 3, 3.5, 4 and 5 were assumed for the six feed- ing levels, respectively. Taking into account the larval mean weight and the number of larvae, a daily ration in mg dry matter feed was allocated for every aquarium. At the begin of the experiment 100 larvae were stocked per aquarium.

Feed Artemia nauplii were hatched every day

and maintained between 5 and 10 C in the refrigerator (1 5 ppt salinity) after measuring the density of the nauplii suspension in the water. Since the dry weight of the nauplii decreased during the day, an average value of 2.35 p,g per nauplius was determined. The feed ration was calculated converting the dry matter ration into numbers of nau- plii and again into water volumes of known density. Larvae were fed three times daily, 0800, 1200 and 1600 h. Light period was 0600 to 1900 h. The water flow through the aquaria was turned off for half an hour at feeding. Feed rests, excreta and dead fish were collected every morning before first feeding through careful siphoning.

Data Retrieval The experiment lasted 15 d. Every three

d, in the early morning of days 4, 7, 10, 13

and 16 after experimental start, 10 larvae were taken out randomly of every aquarium and weighed individually after blotting carefully in filter paper. Dry matter was de- termined in two batches of 5 larvae, dried for 2 h at 105 C. The new values of mean fresh weight and mean dry weight were used to calculate the daily feed ration for the next 3-d period.

Data Analysis The growth rates were calculated as co-

efficient G (Hogendoom 1980) and specific growth rate (SGR, % body weight per d), both for the whole experimental period of 15 d and for every 3-d period.

w p - w.(1/3) G = ' (mg(ll3) per d) (1)

t

SGR = ( e g - 1). 100

(% body weight/d), where (2)

In W, - In Wi t g =

(instantaneous growth rate, mg/d) (3)

where Wi and W, are initial and final fresh weight, respectively and t is the growth pe- riod.

Ration was expressed as specific ration (R, % body weight per d) and the feed con- version wet basis was calculated as R/SGR. The relation between growth rate (G or SGR), feed conversion (FC) and specific ra- tion (R) was approached by the models:

G or SGR = a.(b - exp-c.R) and (4)

R a-(b - exp-c.R)

FC =

(Giinther et al. 1992) (5)

With the model above the following pa- through non linear regression.

rameters were calculated:

Negative growth at starving (R = 0) Maintenance ration (R when SGR = 0) Maximum growth rate (asyntotic value a*b)

220 SEVIUA AND GUNTHER

TABLE 1. Growth and feed utilization parameters of C. macropomum larvae in the 15-d pre-weaning period in dependence of feed ration.

Treat- ment R Wi Wr %DM SGR G FC M

TOO OA 1.23 0.82' 17.19. - 5.09A' -0.0 17A" 1W TO3 3.09AB 1.23 3.34A 15.52A 6.82B 0.028B 0.45A 1 .OA TO7 6.50B 1.23 5.54A 15.6OA 10.5B 0.046B 0.59A 2.OA T1.5 14.6OC 1.23 35.80B 16.39AB 25.09C 0.148C 0.42A 2.5A T25 29.90D 1.23 75.3OC 17.04AB 31.44D 0.210D 0.59A 4.5A T35 51.30E 1.23 85.1OC 16.66AB 32.5D 0.221D 0.95B 2.5A T50 94.97F 1.23 130.60D 16.40AB 36.4E 0.267E 1.m 2.5A

R: ration in 8 body weight per day. Wi and W,: initial and final fresh weight in mg. % DM: % dry matter. SGR: specific growth rate in 8 body weighvd. G: coefficient G in mg03Vd. FC feed conversion wet basis. M: mortality in 5%. Values sharing the same letter in columns are not statistically different: 95% Tukey test.

Values from fish left at day 10. when mortality was complete.

Maximum ration (R at 95% of maximum

Optimum growth rate (SGR when SGR/R

Optimum ration (R at optimum SGR).

growth rate)

is maximum) and

Treatment means were compared by anal- ysis of variance and discrimination of means after W e y , at an error level less than 5%.

40, I50 I I/

~ . " 0 25 SO 75 Ion 0 1 8 I2 16

FItd nion (K body wcighUd.y) mys ofalarm

FIGURE 1. A: Specific growth rate of tambaqui larvae (SGR in 8 body weighud) over the whole 15-d pe- riod in dependence of the feed ration. The curve represents the besf f i r of expression (4 ) using the SGR-values for the whole period. B: Growth of fam- baqui larvae with maximum feeding ration. Contin- uous and broken lines are fined curves after the SGR and the G-models, respectively (expressions I and 2).

Results

Table 1 shows the values of growth and feed utilization of the tambaqui larvae cal- culated for the entire experimental period. Final fresh weight (from 0.82 to 130.6 mg), specific growth rate (from -5.09 to 36.4% body weightld) and coefficient G (from -0.017 to 0.267 mgo.33/d) were significantly dependent on the feed ration which varied between 0 and 94.97% body weight per d. Both growth rates increased from negative values at starving to maximum values at the highest rations (Fig. 1A).

Feed conversion was very low at low ra- tions, increasing to 1.40 at maximum ration. The mortality was low in all feed treatments and did not differ significantly between treatments. In starving larvae (TOO) the mortality was total after 10 d. The dry mat- ter content of the larvae in the feed treat- ments shows an increasing tendency with- out statistic significance.

In the treatment with maximum ration the larvae grew from 1.23 mg to 130.6 mg in 15 d. The growth curve of these larvae (Fig. 1B) could be adjusted much better with the SGR function ( f l = 0.99; SGR = 35.4% body weight/d; linear regression)

GROWTH OF TAMBAQUI LARVAE 221

TABLE 2. Growth and feed utilization in the five ex- perimental 3-d periods.

~~

Period Maximum Maximum Optimum (d) SGR G FC

0-3 34.84 0.12 0.42 3-6 32.43 0.16 0.38 6-9 36.64 0.24 0.45 9-12 34.37 0.30 0.44

12-15 41.73 0.54 0.46

than with the G-function (13 = 0.90; G = 0.25 1; nonlinear estimation).

W = 1.31 .e0.303.t rz = 0.99,

P I 0.0000 (SGR model) (6)

rz = 0.90 (G model) (7)

W = (1.23"'3) + 0.251*t)3

In order to assess differential growth or feed utilization during the five 3-d periods of the experiment, the maximum growth rates SGR and G as well as the optimum feed conversion were determined for each period. Quantitative relations between G, SGR, FC and R were obtained through non- linear regression for each 3-d period and the values of Table 2 were calculated. The re- lation between SGR or G and the feed ra- tion (R) was plotted for each 3-d period and each data set was fitted with the negative exponential model (4) in Figs. 2A, B.

0.60

0.50

0.10

0 30

0 20

0.10

4 . W

0 30 IW 150 2W

Ford mion (% bcdy wciSrrnd.y)

4.10

The maximum growth rate G increased more than fourfold from 0.12 to 0.54 from the first to the fifth 3-d period of the ex- periment, while, measured as SGR, the maximum growth rates remained between 32.4 and 37% body weightld in all first four 3-d periods attaining the highest value (41.7%) in the last experimental period (Ta- ble 2; Figs. 2A, B). The optimum feed con- version was quite similar in all 3-d periods (mean of 0.43).

Since the maximum SGR remains rela- tively unchanged during the whole period of 15 d, the data of all partial periods were pooled together and a unique relation SGR versus R was assessed by nonlinear esti- mation for the whole experimental period.

SGR = 40.38.(0.868 - e(-O.llS.R))

13 = 0.85 (8) The relation WSGR = FC (expression 5 )

is plotted in Fig. 2C against the experimen- tal FC-values for the feeding range between 0 and 30% body weightld.

Finally, using the relation SGR vs. R the values of SGR, R and FC in conditions of starving, maintenance, optimum and maxi- mum growth were calculated (Table 3).

Discussion In spite of their small size, the larvae of

Colossoma macropomum accept readily

50 1

'1 , , 261 0 50 IW 150 2 M

-10

1.50

a 1 '.O0

'B .j 0.50

0.00

I--- .# *

2c 0 1 o i Q ) o

RGURE 2. A: Growth rates of tambaqui larvae (coefficient G) in dependence of feed ration, calculated for every 3-d period of the experiment. Symbolsfrom below to above (jirst to fifth experimental period) are empty squares, full squares, crosses, circles and full triangles. The curves are best fits of expression (4) using the model G for every 3-d period. B: Growth rates of tambaqui larvae as above but expressed as specific growth rate (SGR). Symbols like in Fig. 2A. The curves are bestjts of the expression (4) using the model SGR for every 3-d period. C: Feed conversion values in the feeding range from 0 to 30% body weighud fitted by the expression WSGR obtained in (8).

222 SEVILLA AND GUNTHER

TABLE 3. Specific growth rate (SGR), speci$c ration (R) and feed conversion (FC) in conditions of starv- ing, maintenance, optimum and maximum growth.

Condition R SGR FC

Starving 0 -5.35 - Maintenance 1.23 0 Optimum growth 5.48 13.56 0.40 Maximum growth 27.21 35.03 0.78

-

SGR and R in % body weight per d.

and can be reared with 0-d nauplii of Ar- temia. The mortality in all feed treatments was low and partly due to accidental si- phoning of the larvae during daily cleaning of the aquaria. The larvae with maximum ration grew very well in the experiment, multiplying their weight by a factor of 100 in only 15 d.

Because of its relative constancy during early larval growth, the growth coefficient G has been used in several growth studies, especially in fish larvae (Hogendoorn 1980; Iwama and Tautz 1981; Verreth and Den Bieman 1987; Gunther and Boza 1991, 1992; Cho 1992; Gunther et al. 1992; Ver- reth 1994; Giinther and Ulloa 1995), and was chosen in this experiment for the cal- culation of the daily feed ration. It was therefore a surprise that in this experiment the data of growth vs. time in full-fed lar- vae were much better fitted by the SGR model (Fig. 1B). The reason is that at max- imum feed rations, the G coefficients in- creased strongly during the 15-d experiment (Fig. 2A). The experiment was designed with a constant maximum G of 0.50. This resulted in strong overfeeding in all 3-d growth periods but the last, where a maxi- mum G of 0.54 could only be ascertained by numerical extrapolation. On the con- trary, the maximum SGR values were fairly similar in all 3-d periods (Fig. 2B). It is thus possible to describe the growth of Colos- somu larvae during the 15-d period in de- pendence of feed ration using a unique SGR value (expression 8), so that maintenance, optimum and maximum growth and feed utilization can be estimated for the whole

period (Table 3). The feed utilization of the larvae did not change noticeably during the 15-d period (Table 2). Feed conversion val- ues in dependence of feed ration are thus well fitted by the relation WSGR vs. R (ex- pression 5 ; Fig. 2C), independently of the growth period within the 15-d experiment.

In Table 4 the growth coefficients (G and SGR) in early larval growth of several spe- cies have been compiled and partly recal- culated. All larvae were fed with live Ar- temiu nauplii in excess (with the exceptions of Phoxinus (frozen nauplii) and Clurius (decapsulated cysts)). Temperature was 28 C with the exceptions of Cyprinus (25 C) and Phoxinus (20 C). The theoretical rela- tion between the growth coefficient G and the specific growth rate SGR is given by the equation:

(Hogendoorn 1980), where Wf is the final larval weight of the growth period.

Thus, at constant G, SGR will decrease with increasing weight, and at constant SGR the coefficient G will increase with increasing weight. Five out of the nine stud- ies compiled above show decreasing SGR- values while the G-coefficients stay con- stant. Two studies show simultaneously de- creasing SGR and increasing G and only two cases (including our work) show in- creasing G-values at constant SGR. A mod- el with constant SGR implies a constant rate of cellular divisions which seems im- probable after the very early stages of em- bryonal growth. However, this appears to be the case both in the larvae of Colossoma macropomum and Phoxinus phoxinus, at least during the first 2 vs. 4 wk of larval life. An alternative explanation could be a restricted ration in the first days of life. It has been reported that the small larvae of Colossoma mucropomum are scarcely able to ingest Artemia nauplii (Marthez 1984; De Fex 1991). In Phoxinus phoxinus Kes- temont and Stalmans (1992) mention that they used frozen nauplii because the fish

GROWTH OF TAMBAQUI LARVAE 223

TABLE 4. Growth coefficients G and SCR in early larval growth.

Initial Final Growth weight weight period

Species 0%) (mi$ ( 4 SGR G

Colossoma macropornuma Cichlasoma managuenseb Cichlasoma managuenseC Cichlasoma doviid CIarias gariepinuf CIarias gariepinus' Cyprinus carpi@ Cyprinus carpioh Phoxinus phoxinusl

1.23 3.8 3.72 6.65 2.3

1.1 2 1.86

-

130.15 46.14 74.35 75.2

795 -

67.4

27.9 179

15 10 15 10 28 10 10 10 28

35 34.5-16.9 30.7-12.6 45.5-18.2 53.1-9.2 63.9-23.1

60-24.2 65.3-26.9

10.2

0.12-0.54 0.203 0.177 0.235 0.284 0.3 1

0.260.33 0.351-0.505 0.042-0.103

a This work. Giinther et al. 1992. Giinther and Boza 199 1. Giinther and Ulloa 1995. Hogendoom 1980. ' Verreth and Den Bieman 1987. g Bryant and Matty 1980.

i Kestemont and Stalmans 1992. Hamada et al. 1975 cited in Bryant and Matty 1980.

larvae were not able to catch live nauplii. A defacfo restricted ration would led to a delayed growth in the first days followed by increased growth afterwards, fitting bet- ter to the exponential SGR model. To settle this question, experiments with smaller feed, e.g., rotifers, should be performed. For practical hatchery purposes, however, feeding with Artemia appears to be ade- quate for larval rearing, as is evidenced by the fact, that even at the lowest rations, mortalities were negligible.

The coefficient G increased during the experimental period to a value of 0.54 in larvae of about 130 mg weight, still less than the values of 0.73 and 0.66 observed by Gunther and Boza (1992) in fishes of about 10 vs. 100 g weight fed with for- mulated feed. The values for optimum (0.40) and maximum (0.78) feed conver- sion in larvae fed with Artemia are better than the respective values in Colossoma ju- veniles fed a formulated ration (0.6 and 1 .O; Gunther and Boza 1992), but slighty worse than the values obtained in similar condi- tions in carnivorous cichlid larvae (opti- mum feed conversions of 0.34 (Cichlasoma dovii; Gunther and Ulloa 1995) and 0.36

(Cichlasoma managuense; Gunther et al. 1992)) and maximum wet feed conversions of 0.5 1 (Cichlasoma managuense) and 0.72 (Cichlasoma dovii). As has been stated for juveniles of Colossoma macropomum, the larvae of tambaqui are capable of excellent growth and feed utilization when fed Ar- temia nauplii.

The relations for optimum and maximum growth in the tambaqui larvae allow to de- sign the production strategy of the hatchery between two possible extremes: feeding at maximum level will almost duplicate feed costs (factor 1.94) as compared with opti- mum feeding, but will reduce time-related costs by a factor of 0.39. At maximum feeding, consumption of 0-d Artemia nau- plii will increase from about 140 nauplii/d in a 1.24 mg larva to about 11,500 nauplii/d in a 100 mg larva, for a grand total of about 33,000 nauplii per larva in about 12 d. De- pending on the cost structure in every in- dividual hatchery the most efficient produc- tion strategy can be designed to minimize the combined costs.

Acknowledgments This research was supported by the co-

operative programm UNA-LUW, between

224 SEVILLA AND GUNTHER

the Universidad Nacional, Heredia, Costa Rica and the Department of Fisheries and Aquaculture of the University of Wagenin- gen, Holland.

Literature Cited

Bello, R.A. and W.G. Rivas. 1992. Evaluaci6n y aprovechamiento de la cachama cultivada, como fuente de alimento. FAO. Mtxico, D.E

Bryant, P. L. and A. J. Matty. 1980. Optimisation of Artemiu feeding rate for carp larvae (Cyprinus curpio L.). Aquaculture 21:203-212.

Cho, C. Y. 1992. Feeding systems for rainbow trout and other salmonids with reference to current es- timates of energy and protein requirements. Aqua- culture 100:107-123.

De Fex, R. 1991. Crecimiento y sobrevivencia de lar- vas de cachama (Colossomu mucropomum) con alimento vivo y no vivo. Red Acuicultura (Colom- bia) 5:9-13.

Gondes, J. A. and B. Heredia. 1993. Crecimiento de ejemplares de cachama (Colossomu mucropo- mum) de primera (Fl) y segunda (F2) generaci6n con suministro de alimentos comerciales. Red Acuicultura (Colombia) 7:3-5.

Giinther, J. and J. Boza. 1991. Intensive rearing of juveniles of Cichlusomu munuguense (GUnther 1869) in recirculated systems. UNICIENCIA (Costa Rica) 8:3-10.

Giinther, J. and J. Boa. 1992. Growth performance of Colossomu mucropomum (Cuvier) juveniles at different feed rations. Aquaculture and Fisheries Management 23:81-93.

Giinther, J. and J. ULloa. 1995. Growth and feed uti- lization of Dow cichlid (Cichlusomu dovii) larvae fed Arremiu nauplii. Revista de Biologfa Tropical

Giinther, J., N. Gblvez-Hidalgo, J. Ulloa-Rojas, J. Coppook, and J. Verreth. 1992. The effect of feeding level on growth and survival of jaguar guapote (Cichlusomu munuguense) larvae fed Ar- temiu nauplii. Aquaculture 107:347-358.

Herndndez, A., D. Muiioz, J. A. Ferraz, R. de Fex,

43:277-282.

W. Vdsquez, R. Gonmtles, R. Mornles, F. Al- dntara, T. Ma Luna, C. Kossowski, L. Pkrez, J. A. Morn, P. J. Contreras, F. DCaz, E. M. Fad- ul, and P. Montoya. 1992. Estado actual del cul- tivo de Colossomu y Piuructus en Brasil, Colom- bia, Panam& P e d y Venezuela. Red Acuicultura (Colombia) 6:3-27.

Hogendoon, H. 1980. Controlled propagation of the African catfish, Clurius luzeru (C.&V.). IIl Feed- ing and growth of fry. Aquaculture 21:233-241.

Iwama, G. K. and A. F. Tautz. 1981. A simple growth model for salmonids in hatcheries. Cana- dian Journal of Fisheries and Aquatic Sciences 38:

Kestemont, P. and J. M. Stalmans. 1992. Inicial feeding of European minnow larvae, Phoxinus phoxinus L. 1. Influence of diet and feeding level. Aquaculture 104:327-340.

Martinez, E. 1984. El cultivo de las especies del gb- nero Colossomu en AmCrica Latina. FAO, Santi- ago, Chile.

New, M. B. 1999. Global aquaculture: current trends and challenges for the 21" century. World Aqua- culture 30:8-79.

Pretto Malea, R. 1989. Situaci6n del cultivo de Co- lossomu en PanamB. In A. Hernandez. editor. Cul- tivo de Colossomu. Red Regional de Entidades y Centros de Acuicultura de Amtrica Latina, Bo- got& Colombia.

Saint Paul, U. 1986. Potential for aquaculture of south american freshwater fishes: a review. Aquaculture

Van Anrooy, R., J. Giinther, J. Boza, and N. Gbl- vez. 1996. A preliminary market research about tambaqui (Colossoma mucropomum) in Costa Rica. UNICIENCIA (Costa Rica) 135-1 1.

Verreth, J. 1994. Nutrition and related ontogenetic as- pects in larvae of the African catfish, Clarias gar- iepinus. DSc. thesis. Wageningen Agricultural University, Holland.

Verreth, J. A. J. and H. Den Bieman. 1987. Quan- titative feed requirements of African catfish (Clur- ias guriepinus Burchell) larvae fed with decap- sulated cysts of Arremiu. I. The effect of temper- ature and feeding level. Aquaculture 63:25 1-267.

649-656.

54~205-240.