1413

Effect of ambient oxygen on growth and reproduction in Nile tilapia (Oreochromis niloticus) Jeppe Kolding, Lise Haug, and Sigurd Stefansson

Abstract: Growth, size at maturity, gonadosomatic index (GSI), egg size, and absolute fecundity of Nile tilapia (Oreochromis niloticus) were significantly affected by oxygen levels (1.5 ± 1.0, 2.8 ± 1.4, and 6.0 ± 1.8 mgL–1) in a controlled experiment designed to test the hypothesis (D. Pauly. 1984. J. Cons. Int. Explor. Mer, 41: 280–284) that O2 is the controlling factor for the transition from juvenile to adult in fish, in general, in the context of phenotypic life history plasticity and ‘‘stunting’’ in tilapias. Size at maturity and the estimated asymptotic size decreased with decreasing O2 concentration, as predicted by Pauly’s hypothesis. All fish matured at the same age (18 weeks old), which is in contrast to conventional definitions of stunting. This finding challenges the suggested plasticity in age at first maturity for tilapia. The results also challenge the hypothesis that stunting is a unique recruitment mechanism, as the smaller fish in the group with low oxygen concentration produced smaller and fewer eggs than the larger fish in the group with high oxygen concentration. Re´sume´ : La croissance, la taille a` la maturite´, l’indice gonadosomatique (GSI), la taille des œufs et la fe´condite´ absolue du tilapia du Nil (Oreochromis niloticus) sont significativement affecte´s par les concentrations d’oxyge`ne (1,5 ± 1,0; 2,8 ± 1,4; et 6,0 ± 1,8 mgL–1) dans une expe´rience controˆle´e visant a` ve´rifier l’hypothe`se (D. Pauly. 1984. J. Cons. Int. Explor. Mer, 41: 280–284) selon laquelle l’O2 est le facteur qui controˆle le passage du stade juve´nile a` l’adulte en ge´ne´ral, dans le contexte de la plasticite´ phe´notypique du cycle biologique et du « nanisme » chez les tilapias. La taille a` la maturite´ et la taille estime´e a` l’asymptote sont toutes deux re´duites dans les concentrations plus faibles d’O2, tel que le veut l’hypothe`se de Pauly. Tous les poissons atteignent la maturite´ au meˆme aˆge (18 semaines) contrairement aux de´finitions habituelles du nanisme. Cette observation met en doute la plasticite´ suppose´e de l’aˆge a` la premie`re maturite´ chez le tilapia. Nos re´sultats mettent aussi en doute l’hypothe`se selon laquelle le nanisme est un me´canisme particulier du recrutement, car les poissons plus petits dans le groupe soumis aux faibles concentrations d’oxyge`ne produisent des œufs plus petits et moins abondants que les poissons plus gros dans le groupe soumis aux fortes concentrations d’oxyge`ne. [Traduit par la Re´daction]

Introduction Age and size at first maturity are of major importance in determining the life history of vertebrates. Basic life-history theory predicts that slow growth will result in a higher age at first maturity (Stearns 1992), which is supported by observations of genotypic and phenotypic correlations between these parameters. However, no set threshold would be expected for single parameters (Stearns and Crandall 1984; Stearns 1992); rather, a certain amount of plasticity in life history, including age at first maturity, has been observed under varying environmental conditions, allowing fitness to be maximized in varying environments. The life-history model suggested by Stearns (1992) describes a trade-off between maturing at a fixed size vs. fixed age, with the shape of the reaction norm varying with species and life history, including seasonality of reproduction. It is, however, largely unknown which environmental factors may act as regulators for the observed phenotypic plasticity. The Nile tilapia, Oreochromis niloticus, belongs to the group of cichlids known as ‘‘tilapias’’ (Philippart and Ruwet

1982; McAndrew 2000), which traditionally support major capture fisheries in most African inland waters (LoweMcConnell 1982, 2000; Jul-Larsen et al. 2003). Tilapias tolerate a wide range of temperatures, salinities, and water qualities and show high resistance to parasites and diseases (Pullin and Lowe-McConnell 1982; Beveridge and McAndrew 2000). The relative ease and low cost of rearing tilapias have led to their widespread cultivation and introduction in tropical and subtropical areas (Lowe-McConnell 1958, 2000; Blu¨hdorn and Arthington 1990). Oreochromis niloticus is the second most important cultured freshwater fish worldwide after carp (Pullin 1991; Food and Agriculture Organisation of the United Nations (FAO) 2006). Wild tilapias are frequently observed to mature early and breed prolifically in small water bodies, but not in larger lakes (Fryer and Iles 1972; Lowe-McConnell 1982; Kolding 1993), a phenomenon generally referred to as ‘‘stunting’’ (Pullin 1991). This plasticity in growth and size at maturation is believed to be instrumental for their success in different habitats (Lowe-McConnell 2000). Stunting has been

Received 26 June 2007. Accepted 21 December 2007. Published on the NRC Research Press Web site at cjfas.nrc.ca on 12 June 2008. J20069 J. Kolding,1 L. Haug,2 and S. Stefansson. Department of Biology, University of Bergen, N-5020 Bergen, Norway. 1Corresponding 2Present

author (e-mail: [email protected]). address: Bioforsk Arctic Agriculture and Land Use Division, Box 6232, N-9292 Tromsø, Norway.

Can. J. Fish. Aquat. Sci. 65: 1413–1424 (2008)

doi:10.1139/F08-059

#

2008 NRC Canada

1414

discussed in the context of phenotypic plasticity and lifehistory strategies (Iles 1973; Noakes and Balon 1982; Lorentzen 2000), including environmental factors such as light intensity, temperature (Cridland 1962), and food quality and (or) quantity (Gunasekera et al. 1995; Siddiqui et al. 1997). Other factors such as predation (Iles 1973; De Silva and Amarasinghe 1989; Kolding 1993), growth rate (Huet 1972; Iles 1973), condition of the fish (LoweMcConnell 1958, 1982), and social factors (Jalabert and Zohar 1982; Lowe-McConnell 1982) also have been suggested and partly explored. However, a clear understanding of the causes and the ecological significance of stunting has not been achieved (Pullin and Lowe-McConnell 1982; Brummet 1995; Lorentzen 2000). Iles (1973) suggested that stunting by causing maturation at an earlier age and increasing relative fecundity enables fish to withstand high mortality rates under adverse environmental conditions. As stunting of tilapias is generally observed in small bodies of water, characterized by fluctuating oxygen concentrations, Kolding (1993), based on Pauly’s (1984) hypothesis on the relationship between available oxygen and metabolism, suggested that oxygen level might be one proximate factor determining the ontogenetic switch from growth to reproduction in tilapias. Oxygen is the first limiting physical factor in an aquatic environment (Ross 2000), and fish growth, in general, is oxygen-limited (Pauly 1981, 1984; van Dam and Pauly 1995). Based on the observation that gill area, as a surface, is not growing as fast as the body mass that is to be supplied with oxygen, Pauly (1984) argued that ultimate size in fish is determined by oxygen supply. He suggested a constant, the Qm–Q? ratio (where Qm is the necessary oxygen supply per unit weight at the size of maturation (Wm) and Q? is the necessary supply for maintenance metabolism at maximum size (W?)), as a master regulator determining when an individual will mature. This is summarized here in what may be called a P diagram (Fig. 1), as used by Pauly to develop his argument on the role of oxygen in fish growth (Pauly 1984; van Dam and Pauly 1995). Pauly’s hypothesis would explain the relative constancy of the Lm– L? ratio (length at maturity to ultimate length) generally observed in fish (Beverton and Holt 1959; Mitani 1970; Beverton 1992). Further, for any given population, any cause for elevated maintenance metabolism, such as high temperature, osmotic stress, crowding, or low food density, will thus result in stunting, i.e., a reduced ultimate size and a reduced size at maturity (Fig. 1). The present study tests the general hypothesis of Pauly (1984), i.e., whether low ambient oxygen concentrations can induce stunting in O. niloticus. It also examines Iles’ (1973) suggestion that stunting is a response to stress and therefore associated with increased reproductive effort, as expressed in terms of the gonado-somatic index (GSI), egg size, and fecundity.

Material and methods Fish stock The fish used in the experiment were juveniles from a broodstock of Nile tilapia (Oreochromis niloticus) originating from Lake Manzala in Egypt. The broodstock had been

Can. J. Fish. Aquat. Sci. Vol. 65, 2008 Fig. 1. P diagram, expressing the interrelationships between internal metabolic oxygen supply at maintenance level (Q?), oxygen supply near first maturity (Qm), and the weight of fish. Note that the Qm– Q? ratio is constant for two different habitats: (a) habitat linked with large asymptotic size and large size at maturity, and (b) habitat associated with a small ultimate size and small size at maturity because of higher maintenance metabolism. Note also that the constancy of the Qm–Q? ratio can only be imperfectly mirrored in the Lm–L? ratio, which can be derived from Wm and W? (modified from Pauly 1984).

imported from the University of Stirling, Scotland, by the Norwegian Institute of Water Research (NIVA), Oslo, Norway, in 1983 and brought to the University of Bergen in 1993 where the experiment took place. The juveniles used in this experiment were produced in two batches, each originating from a single female, fertilized by different males, to minimize genetic variability. The two batches were released on 2 and 4 August 1995. From 6 August, all juveniles were mixed and reared under the same conditions. In the first week, the fish were reared in one 150 L aquarium, after which they were transferred to a grey, square, fibreglass tank with a rearing volume of 500 L, equipped with a 1000 Lh–1 bio-filter pump (Eheim 2217; Eheim GmbH and Co. KG, Deizisau, Germany). The juveniles were fed aquarium feed (TetraMin; Tetra GmbH, Melle, Germany) from the day they were released from the mouth of the females until 25 August 1995. From this date until the experiment started on 25 September 1995, they were fed a dry diet (size 1.2 mm pellets; Felleskjøpet Fiskefoˆr AS, Stavanger, Norway) and aquarium feed (cichlid sticks, Tetra GmbH). At the start of the oxygen treatments, the juveniles were 8 weeks old and the initial mean total length (standard deviation, SD) was 4.75 (0.45) cm. Experimental design On 25 September, 781 juveniles were weighed and distributed into six square, grey, fibreglass tanks with open surface and a rearing volume of 200 L (130–131 fish in each tank) in three groups with different oxygen concentrations, with each group consisting of two replicate tanks. The mean oxygen concentrations (SD) in the low, medium, and high oxygen treatments were 18.4% (11.6%), 35.4% (17.7%), and 75.2% (22.2%), corresponding to 1.5 (1.0) mgL–1, 2.8 (1.4) mgL–1, and 6.0 (1.8) mgL–1, respectively. The low and medium O2 concentrations were chosen as tila#

2008 NRC Canada

Kolding et al.

pia are known to live and reproduce and are reared in waters with such low oxygen levels (Welcomme 1969), whereas the higher concentration (6 mgL–1) is above the dose-dependent relationship between O2 and growth (Brett 1979; Ross 2000). All tanks were supplied with dechlorinated, pHadjusted (pH = 7.0) fresh water with a mean temperature of 27.6 8C (1.3 8C) at a flow rate of 1.5–6.0 Lmin–1, which was adjusted according to the increase in biomass. Each of the three oxygen groups consisted of two replicate tanks, a 200 L header tank and one bio-filter pump (Eheim 2217; Eheim GmbH and Co.). Each pump drained the two replicate tanks, removing food and faeces from the water, and pumped the water back into the header tank. To increase water flow, an additional pump (same type) was added on 25 October in the high and medium oxygen groups and on 29 October in the low oxygen group. The oxygen concentration was monitored and regulated by a complete oxygen measurement system (E-module; DAN Technique International A/S, Roskilde, Denmark), which regulated electromagnetic valves (Burkert-Contromatic A/S, N-2026 Skjetten, Norway) controlling the supply of atmospheric air, O2, or N2 from pressure tubs (Hydrogas, Bergen, Norway). Oxygen probes (Oxyguard International A/S, Blokken 59, DK-3460 Birkerød, Denmark) were placed in the header tanks where monitoring and adjustment of the oxygen concentration occurred. Depending on the desired oxygen level and the biomass in each group, air, N2, or O2 diffusers, or a combination of these, were activated. The oxygen concentration in the tanks was further checked manually using Oxyguard Handy MK 111 (Oxyguard International A/S Denmark). Rearing conditions Water temperature in the experimental tanks was kept at 27.6 8C (1.3 8C) by immersed heaters in each header tank and by keeping all tanks in a temperature-controlled room. This temperature is within the range of optimal temperatures for O. niloticus (Wohlfarth and Hulata 1983; Breine et al. 1996). Temperature was lowered approximately 3–4 8C during sampling days, when half of the water in each tank was replaced. The prescribed temperature was re-established during the following night. Light, with intensity ranging from 150 to 270 lx, was provided by fluorescent tubes positioned in the middle of the room, operated by an automatic timer simulating twilight periods. At the start of the experiment, the light regime was kept constant at 9.5 h light – 14.5 h dark. On 2 November, it was changed to 11.5 h light – 12.5 h dark, which was maintained throughout the experiment. Maximum stocking density in the high oxygen group was 20.7 kgm–3, which is well within the range recommended for intensive culture of O. niloticus (Siddiqui et al. 1991). The fish were fed a commercial dry diet (53% protein, 12% fat, 16% carbohydrate, 8% ash; Felleskjøpet Fiskefoˆr, A/S). During week 1 (26 September – 2 October), the fish were fed manually three to eight times daily during daylight hours. Beginning in week 2, food was dispensed every 30 min from automatic feeders during daylight hours. Appetite was checked at the end of each day to ensure that the fish were fed to satiation. The groups were fed rations varying from 1.5% to 9.2% body weightday–1, depending on biomass and growth rate and to minimize negative ef-

1415

fects of excess waste feed. Surplus food was siphoned out daily after the last feeding. The fish were not fed on the fortnightly sampling days and during the last two days at the end of the experiment. Sampling and analyses Length, weight, and maturity Total length (to the nearest 0.1 cm) and wet weight (to the nearest 0.01 g) of all fish were measured every 2 weeks. At the same time, 5–15 fish from each tank were sacrificed for determination of sex, gonadosomatic index (GSI), and state of maturity. Investigations of fecundity and egg size were made from these samples on weeks 16 and 18. Immediately after dissection, the gonads were fixed in Karnovsky fixative (Nylund 1985). Sex was determined based on the morphology, size, and color of the gonads (Babiker and Ibrahim 1979). Gonads that could not be be sexed by macroscopic examination were analysed histologically; they were dehydrated, cleared, infiltrated, embedded in Historesin (Leica historesin embedding kit; Leica 702218500), and sectioned (5 mm) using a Leica microtome (Leica Microsystems GmbH, Wetzlar, Germany). Cross sections were dried, stained with Diff quick (Merz and Dade AG, CH-Dudimger, Switzerland), and examined using a light microscope (12 magnification). Sex was determined based on the histological classification of Latif and Saady (1973a, 1973b). Because low GSI does not necessarily indicate an immature fish, age at 50% maturity was determined from morphological studies of the gonad, rather than from GSI values. Nile tilapia have a very fast reproduction cycle and are able to produce new broods within a few weeks of each other (Peters 1963). However, the readiness to spawn can only be maintained for approximately 1 week, and the ova are subsequently resorbed if appropriate spawning conditions are not available during this period (Peters 1963). The females were classified as mature when oocytes started to be visible, and males when the testis started to become thickened and enlarged (stage 2; Babiker and Ibrahim 1979). Fecundity and egg size Fecundity of tilapias can only be reliably assessed when the eggs of an emergent spawn in ripe fish are separated from those of the latent successive spawns (Peters 1963). Size–frequency distributions of oocytes were therefore made of all ripening or ripe ovaries, defined as tense with numerous pear-shaped oocytes (Latif and Saady 1973b; Babiker and Ibrahim 1979), to separate ripe and ripening females. Subsamples taken from the middle part of each ovary were weighed to the nearest 0.0001 g, and the size (largest diameter) of the oocytes was measured to the nearest 0.04 mm using a binocular microscope (12 magnification). Thereafter, the egg size and fecundity were determined from ripe females by studying the oocytes in addition to the tenseness of the ovary. Only the most developed modal group (G1 oocytes; Kjesbu 1994), separated by a gap from the following oocyte group, was tested for differences in egg size. To avoid possible differences between different sections of the gonad, subsamples were always taken from the middle part of the lobe. #

2008 NRC Canada

1416

Can. J. Fish. Aquat. Sci. Vol. 65, 2008

Calculations Specific growth rate (SGR) was calculated according to the formula of Houde and Schekter (1981): 0 2  1 3 ln W2 Þ  ln ðW1 Þ 5  1A  100 ð1Þ SGR ¼ @exp4 t2  t1

regression, weighted by the total number of individuals for each size class or sampling day. The growth parameters L? and K were determined by fitting the length-at-age data to the von Bertalanffy growth function (VBGF; von Bertalanffy 1957):

where W2 and W1 are body weights at time t2 and t1, respectively, measured in days. Geometric mean weight (GM) was calculated as pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2Þ GM ¼ ðW1  W2 Þ

where K is the ‘‘curvature parameter’’ determining how fast the fish approaches L? (the asymptotic or ultimate length). Fitting was done using the nonlinear iterative numerical search of the minimum sum of squares (SSE = (observed – predicted)2) implemented in FiSAT (Gayanilo et al. 1995). The growth performance index ’ was calculated according to the formula of Munro and Pauly (1983) as

where W1 and W2 are body weights at time t1 and t2, respectively. Condition factor (CF) was calculated as   W  100 ð3Þ CF ¼ L3 where W is body weight (g) and L is total length (cm). Gonadosomatic index (GSI) was calculated as ð4Þ

Wg  100 GSI ¼ Wtot

where Wg is the weight of the gonad and Wtot is the total weight of the fish including gonads. The absolute fecundity (Bagenal 1978) was calculated as ð5Þ

absolute fecundity ¼

Wg N Ws

where Wg is the weight of the gonad from which the subsample Ws was taken, and N is number of ripe eggs (most developed modal group) in Ws. The relative fecundity (Bagenal 1978) was calculated as ð6Þ

relative fecundity ¼

absolute fecundity Wtot

where Wtot is the total weight of the fish including the gonads. The length and age when 50% of the fish were mature, Lm and Agem, respectively, were calculated as ð7Þ

Lm ¼

bL aL

and Agem ¼

bA aA

where a and b are fitted constants of the following logistic model (Gunderson et al. 1980): ð8Þ

1 Pm ¼  100 1 þ expðb  a  xÞ

where Pm is percent mature fish at length or age (x). The effect of oxygen on Lm and Agem was analysed by fitting the fraction of mature females (per 10 mm intervals or by sampling day) to the logistic model by a nonlinear

ð9Þ

ð10Þ

Lt ¼ L1 ð1  eKðtt0 Þ Þ

0 ¼ log10 K þ 2log10 L1

Contrary to L? (dimension: length), and K (dimension: time–1), ’ has the dimension of a growth rate (lengthtime–1) and, hence, can be used to compare different growth performances. Statistical analyses To assess for normality of distribution of the measured variables, a Kolmogorov–Smirnov test was performed (Zar 1984). Weight data were log10-transformed prior to analysis. The homogeneity of variances was tested using the Levene’s F test (Brown and Forsythe 1974). A nested analysis of variance (ANOVA), with replicates nested within oxygen groups, was applied to calculate the effect of oxygen on weight, length, and condition factor. Regressions of SGR against geometric mean weights were analysed using a covariance analysis (Zar 1984). One-way ANOVA was used to test the following: (i) effect of oxygen on GSI in females and males, (ii) effect of oxygen on the growth of each sex, (iii) differences in growth between females and males in each oxygen group, and (iv) differences in length, weight, GSI, egg size, and absolute and relative fecundity of ripe fish in each oxygen group. Significant ANOVAs were further analysed by a Student–Newman–Keuls multiple comparison test to determine differences among experimental groups. A Kruskal–Wallis one-way ANOVA by ranks test (Zar 1984) was used when testing for differences in length of immature and mature females in each oxygen group and differences between replicates in oxygen concentration for each sampling period. Where significant differences were found, a multiple comparison Kolmogorov–Smirnov twosample test was applied. Statistical analyses were performed in Statistica version 5.0 (StatSoft Inc., Tulsa, Oklahoma), and data were considered significant if p < 0.05.

Results Mortality and behaviour No mortality was observed during the experiment in any group, although oxygen availability influenced behaviour in several ways. Fish in the low oxygen group were often observed positioned close to the submerged water inlet performing ‘‘air gulping’’ at the surface, but this behaviour was #

2008 NRC Canada

Kolding et al.

1417

Table 1. Mean length (cm) of female and male Oreochromis niloticus reared under three ambient oxygen concentrations (high, medium, and low). Females Week 2 4 6 8 10 12 14 16 18

High 6.3±1.23a (3) 7.5±0.29a* (5) 9.0±0.27a (5) 9.9±0.82a (12) 11.3±1.12a (15) 12.5±1.15a (14) 12.9±1.47a* (19) 13.6±1.59a* (16) 14.4±1.49a* (19)

Males Medium 6.0±0.53a (3) 7.5±0.35a (6) 8.2±0.54b* (5) 9.1±0.95b* (9) 10.1±1.03b* (11) 11.5±1.35b* (17) 12.3±1.35a* (18) 12.6±1.63a* (18) 13.3±1.44b* (21)

Low 5.7±0.62a (6) 6.5±0.37b (7) 7.7±0.28b* (7) 8.3±0.48c* (13) 9.6±0.74b* (12) 10.1±1.56c* (19) 11.0±1.21b* (14) 11.1±1.28b* (18) 12.0±1.43c* (20)

High 6.7 (1) 8.1±0.12a (5) 9.6±0.9a (5) 10.7±0.99a (7) 11.8±1.21a (5) 13.0±1.95a (15) 15.1±1.84a (11) 15.7±2.45a (14) 17.3±2.81a (9)

Medium 8.1 (1) 7.6±0.86a (4) 9.0±0.76a (5) 10.0±0.57a (11) 11.4±1.03a (9) 12.7±1.60a (13) 14.7±1.47a (12) 15.1±1.59ab (12) 15.8±3.50a (9)

Low 6.3±0.35a (2) 7.0±0.49a (3) 8.4±0.47a (3) 8.9±0.68b (7) 10.5±1.23a (9) 12.0±1.76a (11) 12.6±1.02b (15) 13.7±0.79b (12) 14.5±1.86a (10)

Note: Values are means ± standard deviation (SD), number of observations (n) in parentheses. Values not followed by the same letter on the same date are significantly different (p < 0.05). Values of females followed by a asterisk (*) are significantly different (p < 0.05) from corresponding groups of males.

not observed in the other groups. The fish in the low oxygen group also seemed to have a higher ventilation frequency (although ventilation frequencies were not recorded) and were observed to have a lower food ingestion rate (more uneaten food was found in the tanks despite rations being adjusted to the SGR) compared with the fish in the higher oxygen groups. Growth No consistent differences in growth rate between replicate tanks were found (nested ANOVA); hence, replicates were combined in all cases. Length and weight were reduced in response to the lowered oxygen concentrations (Figs. 2a– 2b), with significant differences in length from week 4 and in weight from week 2. Males grew faster than females in all oxygen groups (p < 0.05; Table 1). Significant differences were found from week 6 in both the low and medium oxygen groups and from week 14 in the high oxygen group. Consistent differences in growth among females were observed as a consequence of oxygen concentration (Table 1). Among males, differences in growth were only found on weeks 14 and 16 between the high and low oxygen groups (Table 1). However, males in the low and medium oxygen groups were relatively smaller than males in the high oxygen group throughout the experiment. Specific growth rate (SGR) decreased in all groups throughout the experiment (Table 2), but less in the low oxygen group than in the high oxygen group. By the end of the experiment, the SGR of the low oxygen group was as high as, or higher than, that of the other two groups. Differences were not evaluated statistically because of the low degrees of freedom (n = 2, replicate tanks). The decreases in SGR were size-dependent and therefore were not related only to oxygen concentration. The size-specific growth rates (SSGR), obtained by plotting the SGR as a function of geometric mean weight (Fig. 3), showed a similar decline in growth rate with increasing weight for all oxygen groups. No significant differences were found between the slopes of the SSGR (test of parallelism, n = 18, p = 0.591; Table 3), suggesting that the fish would approach their ultimate size (L?) at about the same time. An overall significant difference was found between the intercepts of SSGR, but no

Fig. 2. (a) Mean length and (b) mean weight of Oreochromis niloticus under three oxygen concentrations: high, triangles; medium, squares; low, circles. The vertical lines indicating SE may be obscured by symbol; n.s., not significant at p ‡ 0.05.

significant differences were found in the multiple comparison tests. The high oxygen group was found to have the highest value of ’ (Table 4), in accordance with this group reaching the largest size. #

2008 NRC Canada

1418

Can. J. Fish. Aquat. Sci. Vol. 65, 2008 Table 2. Specific growth rate (SGR) of Oreochromis niloticus reared under three ambient oxygen concentrations (high, medium, and low). Week 2 4 6 8 10 12 14 16 18 Average

High 5.18±0.03 4.36±0.02 3.88±0.24 3.04±0.01 3.11±0.18 2.14±0.26 2.22±0.43 1.32±0.02 1.30±0.13 2.95±1.31

Medium 4.80±0.31 4.30±0.02 2.83±0.20 2.95±0.49 3.70±0.08 2.37±0.41 1.52±0.001 1.18±0.06 1.71±0.21 2.81±1.24

Low 3.74±0.20 3.09±0.07 3.83±0.21 2.63±0.06 2.76±0.18 2.15±0.44 1.27±0.31 1.31±0.12 1.81±0.07 2.51±0.93

Note: Values are means ± standard deviation (SD), based on SGR for each replicate (n = 2).

Fig. 3. Regression of specific growth rate versus geometric mean of Oreochromis niloticus under three oxygen concentrations (high, triangles; medium, squares; low, circles).

Condition factor The development in condition factor, with an overall increasing trend, was similar in all groups (Table 5). Significant differences in the condition factor occurred only temporarily between the oxygen groups, but where significant differences were observed, condition factors of the low and medium oxygen groups were always lower than the condition factor in the high oxygen group. Sexual maturity A significant difference in GSI of the females was only found at week 8, when GSI of the high oxygen group was lower than the GSI of the medium and low oxygen groups (Fig. 4a). No significant difference in GSI was found in males (Fig. 4b). An increase in GSI appeared in week 8 (i.e., 16 weeks old) for the females and in week 6 (i.e., 14 weeks old) for the males, which by examination of the gonads, coincided with the commencement of sexual maturation. All sampled females and males were mature from week 16 (i.e., 24 weeks old). During the period of the experiment when both mature and immature specimens

Table 3. Results from the regression of specific growth rate (SGR) versus geometric mean weight of Oreochromis niloticus under three ambient oxygen concentrations (high, medium, and low). The regression equation is given as y =  + x. O2 group High Medium Low

n 18 18 18

 –0.054 –0.058 –0.068

p()