ELSEVIER

Aquaculture 148 (1997) 85- 104

Survival and growth of juvenile fluted giant clams, Tridacna squamosa, in large-scale grow-out trials in the Solomon Islands I Timothy P. Foyle, Johann D. Bell *, Mark Gervis, Idris Lane International Center

for Living Aquatic Resources Management, Coastal Aquaculture Centre, P.O. Box 438, Honiara, Solomon Islands

Accepted 6 July 1996

Abstract Two large-scale experiments were conducted with juvenile Tridacna squamosa at coastal village sites in the Solomon Islands to provide robust estimates of survival and growth during grow-out. Juveniles were reared to _ 24 mm shell length (IO- 11 months) in land-based tanks and then transferred to subtidal sites near villages for grow-out for 8 months. During the first experiment, which started in 1993, survival varied greatly among sites, ranging from 7% to 83%. Exposure (‘fetch area’) and geographic location had a significant influence on survival, indicating that adverse weather conditions affected mortality rates. Inserts made of 5-mm galvanized mesh (‘settlement rings’) fitted to the inside perimeter of grow-out cages significantly reduced loss of clams from cages after transfer to villages. Growth varied from 2.3 to 8.6 mm month-’ and was significantly, positively correlated with the number of clams surviving in the cages. During the second experiment, which commenced in 1994, survival was greater and less variable, ranging from 42% to 83%. Average growth also improved, ranging from 4.0 to 7.2 mm month- ‘. Growth was positively correlated with water flow, and negatively correlated with Secchi disc visibility and the number of predatory ranellid gastropods found in the grow-out cages. A weak negative relationship between growth and survival occurred during the second experiment. Modifications to aquaculture protocols in 1994 were evaluated in a series of concurrent experiments conducted at the village sites. Grooves in the bases of cages, support of clams using pieces of coral, and retention of ‘settlement rings’ for 4 instead of 2 months had no effect on survival and growth of the clams. However, survival of juveniles preconditioned in the ocean was significantly higher than those reared exclusively in tanks.

* Corresponding author. Tel.: + 677-29255; ’ ICLARM Contribution No. 1228. 0044-8486/97/$17.00 PII SOO44-8486(96)0

Copyright 1402-O

fax: + 677-29130;

e-mail: [email protected]

0 1997 Elsevier Science B.V. All rights reserved

86

T.P. Foyle et al./Aquaculture

148 (1997) 85-104

On the basis of these two experiments, the mean time required to grow 7’. squamosa from 24 to 50 mm shell length, a size suitable for sale to the aquarium market, was only 5 to 7 months. Our

results reveal that village-based fanning can be economically feasible in the Solomon Islands, provided reliable hatchery facilities are available. Keywords: Tridacna squamosa; Giant clams; Survival; Growth;

Juveniles; Farming trials in villages

1. Introduction Giant clams (Family Tridacnidae) have been identified as important candidates for aquaculture in the tropical Indo-Pacific because of their traditional role as a food item, the high value of the adductor muscle, relatively rapid growth rates, and the necessity to re-establish over-exploited populations (Mum0 and Heslinga, 1983; Crawford et al., 1987). In the past 15 years, much information has been published on the biology and ecology of giant clam species (for reviews, see Munro, 1993; Lucas, 1994) and considerable progress has been made in their culture. Techniques have now been developed for culture of all the stages of the life cycle (Braley, 1992; Calumpong, 1992). A promising market for cultured clams originally appeared to be the lucrative Taiwanese trade in adductor muscle (Heslinga et al., 1990), which favoured cultivation of the larger species such as Tridncna gigas. Hambrey and Gervis (1993) have recently cast doubt on the economic feasibility of growing clams at village sites in less developed countries for the protracted periods, at least 7 years, required to supply the market for adductor muscle. However, a substantial demand for small tridacnids has developed in the aquarium trade (Heslinga et al., 1988, 1990). The aquarium market promises to be beneficial to growers in developing countries because they can sell small cultured clams after a short grow-out period (Bell et al., 1996) and has spurred interest in cultivating the smaller species, such as T. crocea, T. maxima, and T. squamosa. The focus of giant clam research at the Coastal Aquaculture Centre (CAC) of the International Center for Living Aquatic Resources Management (ICLARM) has been the introduction of culture techniques for coastal villagers. Initially, low cost methods were developed for T. gigas and small-scale trials at village sites were conducted between 1989 and 1992 (Bell et al., 1996). Since 1993, six species of giant clams have been propagated at the CAC and a series of large-scale, replicated, grow-out experiments at village sites has commenced. The trials were designed to accurately measure survival and growth of the six species across a wide range of locations. With this information, the feasibility of low-cost aquaculture for villagers can be assessed. This paper presents the results of two major grow-out experiments for juvenile (- 24 mm) T. squamosa conducted during consecutive years at village sites throughout much of the Solomon Islands. These experiments, principally designed to assess variability in the performance of this species at a large number of sites, revealed important differences in survival and growth between years and among sites. We attempted to identify the factors regulating this variation in survival and growth through multiple regression analysis. During the second experiment, we also evaluated aquaculture protocols potentially useful for reducing this variability.

T.P. Foyle et al./Ayuaculture

148 (1997) 85-104

87

We found that growth was significantly correlated with water flow, the amount of particulate matter in the water, and presence of predatory ranellid snails. We also discovered that preconditioning the clams in the ocean prior to distribution to villages, and using small mesh inserts in the cages, substantially reduced loss of clams after transfer. The robust estimates of survival and growth indicate that T. squamosa can be grown from 24 to 50 mm, an acceptable size for sale to the aquarium trade, in 5 to 7 months at attractive rates of return for villagers.

2. Materials

and methods

The general method was similar for the two main grow-out experiments. These experiments started in October 1993, and October 1994, and ran for a period of 8 months (Table 1). 2.1. Village sites Sites selected for the experiments spanned 500 km along the Solomon Islands chain (Fig. 1). Sites located in the New Georgia Islands were considered western sites while eastern sites were those situated in the Russell Islands, Florida Islands, and on Guadalcanal. Western sites were serviced by the CAC field station located in the New

Table I Summaries

of the main features of the two village-based

grow-out

experiments

using T. squamoscr

Feature

Experiment 1993

1994

No. of cohorts used Cohort type Dates of spawning Date of harvest Dates of distribution No. of original sites No. of sites used in analysis Mean shell length at transfer Distribution of clams in cages No. of clams per cage Time ‘settlement rings’ left in cages Cage design Total no. of cages at a site No. cages used for site comparisons No. cages for other experiments Duration of experiment Shell length measurements initially 5 months 8 months

2 Multiple parents December 7,9, 1992 October 12, 1993 October l3- 15, 1993 I6 I4 23.7 mm + 7.5 s.d., CV = 32% Scattered in cage 250 I week or I month Smooth bases 4 4

I Multiple parents November 2, 1993 October 18, 1994 October 19, 1994 IO IO 23.9* 3.8 s.d., CV = 16% Placed upright in grooves 150 2 months Grooved bases, principally 6 3 3 8 months

8 months 100 clams per site 20- 100 clams per site Up to 200 clams per site

50 clams per cage

30 clams per cage 50 clams per cage

T.P. Foyle et al./Aquaculture

88

148 (1997) 8.5-104

I

157’E

159’E

161°E

Fig. 1. Village sites used in the experiments. Letter designations (a, b) refer to sites where farmers shifted the location of cages between the first and second experiment. l , Coastal Aquaculture Center (CAC) and hatchery; * , poor husbandry sites, deleted from the analysis; t, sites damaged by storm; tt, CAC field station.

Georgia Islands while eastern sites were serviced by the CAC main office and hatchery on Guadalcanal (Fig. 1). Most sites were situated at villages. One was located at the CAC field station (Fig. 1) where clams were grown under simulated village conditions. Sites were selected based on interviews with prospective farmers, diversity of habitat, and suitability for rearing giant clams. Sites were of sufficient water depth, close to villages, protected from oceanic wave action, and away from sources of freshwater and sediment input, as recommended by Calumpong (1992) and Govan (1993). Prior to transfer of the clams to the sites, the farmers were trained in husbandry at workshops conducted by CAC staff. They were taught to clean cages regularly to prevent fouling by algae and to remove the predators of giant clams (Govan, 1995) two to three times a week. 2.2. Production

and distribution

of seed clams

Juveniles were grown in land-based nursery tanks at the CAC on Guadalcanal (Fig. 1) using standard spawning and rearing methods (Braley, 1992). After 10 to 11 months, clams were harvested, sorted by size to reduce the coefficient of variation, counted, then randomly assigned to sites. Juvenile clams were packed in plastic bags with water and air, placed in insulated containers, and transported by boat or aircraft to the grow-out sites. Clams which did not reach sites on the first day were transferred temporarily to perforated polyethylene containers and submerged in lagoonal water. The clams remained in transit for a maximum of 3 days.

T.P. Foyle et ul./Aquacubure

148 (1997) 85-104

89

c&nent base of cage Fig. 2. Diagram of the placement of a 5-mm mesh insert (“settlement ring”) into a clam cage. Front sides and top (19-mm mesh) of the cage have been cut away.

2.3. Experimental design and data collection The two grow-out experiments were designed to provide robust estimates of mean survival and growth for juvenile T. squamosa and to test statistically whether survival and growth of this species differed among sites. Replicate grow-out cages were therefore placed at the sites (Table 1). The 0.5-m’ cages, measuring 77 X 65 X 13 cm, consisted of a reinforced concrete base with sides and top made from 19-mm galvanized welded mesh. A 5-mm galvanized mesh insert (the ‘settlement ring’) was tied to the inside perimeter of each cage (Fig. 2) to reduce the loss of clams while they attached to the base of the cage by their byssal threads. Two trestles, made from 12-mm-diameter steel reinforcing bar, were also constructed at each grow-out location and positioned to be at least 1 m deep at low tide to reduce turbulence from waves and prevent damage from motorized canoes. Upon distribution of the clams, the cages were labelled and, if warranted, moved to quiet water for 5 to 7 days to allow clams to attach before being transferred to the trestles. CAC technical staff visited each site monthly to count the clams remaining in the cages, discuss problems with the farmers, and re-emphasize proper husbandry. Shell length measurements were taken at distribution, and 5 and 8 months after transfer to the villages (Table 1). 2.4. Differences between the 1993 and 1994 experiments Several modifications were made to the experimental design of the second experiment, based on experience gained during the first one. In addition, environmental monitoring was incorporated into the 1994 experiment and aquaculture protocols were examined in a series of concurrent experiments (described in subsequent sections). For the 1993 experiment, 16 sites were selected. This was reduced to ten sites in 1994 (Table 1) based on the performance of sites and farmers during the first year. In some cases, sites remained the same but farmers shifted cages to nearby locations (Fig. 1). For the 1993 experiment, two cohorts were mixed to provide sufficient numbers of juvenile T. squamosa and 250 clams were assigned to each cage. Crowding was noted at some of the best sites by the end of the first experiment and so 150 clams were assigned to each cage from one cohort for the 1994 experiment (Table 1).

90

T.P. Foyle et al./Aquaculture

148 (1997) 85104

At distribution, the coefficient of variation (CV) of initial shell lengths was 32% for the 1993 experiment (Table 1). Improved grading reduced the CV to 16% for the 1994 study. Mean shell lengths at transfer ranged from 22.5 to 25.7 mm for sites used in the 1993 experiment and 22.6 to 24.6 mm for sites used in 1994. The first experiment revealed that T. squamosa did not attach strongly to the bases of the cages. Hence, cages with smooth cement bases, used initially, were mainly replaced in the 1994 experiment with cages having grooved cement bases. The grooves were made by impressing 12 mm reinforcing bar into the wet cement. Farmers were instructed to remove the small mesh inserts (‘settlement rings’) from the cages after 1 week for the 1993 experiment. Farmers left the rings in place in some cases; these rings were removed by technical staff after 4 weeks. For the 1994 experiment, ‘settlement rings’ were removed by field staff, mostly after 2 months (Table 1). 2.5. Physical and biological measurements Because strong site differences were noted during 1993, monitoring of physical and biological features of the sites was incorporated in the 1994 experiment. Water flow at the sites was examined on up to four occasions between January and April, 1995, by deploying calcium sulphate (‘plaster of Paris’) clods which erode in proportion to the movement of water. They were made following the procedure of Doty (1971) and sanded to 15 ( + 0.2) g weight. Four clods were attached with silicone to the cement base of a small 17- 19 cm X 17-19 cm X 7.5 cm cage, similar in construction to the clam cages used in the experiment. One cage of clods was placed at each site, resting on the cover of a cage of clams. To determine dissolution of clod material under conditions of no water movement, three or four control clod cages were deployed at three sites, sealed in a 40-80-l polyethylene (‘Nally’) bin, resting in shallow water. All clod cages were retrieved after around 24 h. The ‘diffusion factor’, DF, a dimensionless index of water movement (Doty, 1971), was calculated by dividing the weight of calcium sulphate dissolved from each clod during 24 h by the mean weight lost from the control clods during the same period of time. Govan (1995) documented high mortality of cultured giant clams due to predatory gastropods of the genus Cymatium (Ranellidae). Hence, the numbers of ranellids found

in each cage were recorded during the monthly site visits. Temperature, salinity, and horizontal Secchi disc readings were taken monthly at the sites between March and June, 1995. Salinity was measured to the nearest 0.5%0 using an Atago S-10 refractometer. The depths of water at which the grow-out cages were placed were measured in May, 1995 and standardized to mean low water level using published tide levels for the nearest port. Calumpong and Solis-Duran (1993) and Lucas et al. (1989) found that storm surge and wave turbulence reduce the survival and growth rate of giant clams. Although the village sites were not affected by ocean swell, they varied with respect to exposure to wind-driven waves. Exposure to wave turbulence generated by wind was estimated for both the 1993 and 1994 experiments by calculating the ‘fetch area’ around a site. The area of ‘open’ water was drawn on a chart, photocopied, cut out, and weighed. Fetch

T.P. Foyle et al./Ayuaculture

148 (1997) 85-104

91

area was calculated by dividing this weight by the weight of a square of paper of known area. Open water was defined as any water between the site and the nearest land or, if no land, to the closest reef that broke the oceanic swell. 2.6. Concurrent

experiments

During the 1994 experiment, four other experiments were conducted at village sites to examine whether modifications to the aquaculture protocols improved growth or survival of juvenile T. squamosa during grow-out. In addition to the three standard ‘grooved’ cages used for comparing the ten sites, another three cages were deployed at each site and manipulated in the series of experiments, described below. The two sets of cages were interspersed on the trestles. 1. Tank-reared vs ocean-reared clams. This experiment was conducted at two village sites and compared survival and growth of three cages of seed clams reared exclusively in tanks with three cages of seed clams preconditioned in a floating ocean nursery (see Calumpong, 1992). Clams came from the same cohort. The seed clams reared in the ocean nursery were transferred to cages with ‘settlement rings’ suspended from floats at a depth of 1 m offshore from the CAC hatchery. They were held in this manner for 12 weeks prior to distribution. At transfer, ocean-reared juveniles were substantially larger (28.4 mm + 6.0 sd.) than their tank-reared counterparts (23.9 mm f 3.8 s.d.1. 2. Grooved vs smooth cages. The effectiveness of grooves in the bases of the cages in assisting T. squamosu to attach was evaluated at four sites by comparing survival and growth of tank-reared clams in three cages with grooved cement bases to those in three cages with smooth cement bases. 3. No-coral vs coral. This experiment, conducted at four sites, examined whether clams placed upright in the grooves had better survival when supported by pieces of dead Acropora coral. The coral ‘fingers’ were placed in rows along each side of the grooves to support the clams as they attached by their byssal threads. At each site, three grooved cages with coral supports were compared with three grooved cages without coral. The coral pieces were removed after 1 month. 4. ‘Settlement ring’ retention: 2 months vs 4 months. The protocol for the ‘no-coral vs. coral’ experiment was modified in December, 1994 because there was no significant difference in survival between cages with and without coral pieces after 1 month. Consequently, ‘settlement rings’ were left on half of the six cages for a longer period (4 months) than the other half (2 months) to determine whether this affected survival or growth of the clams. 2.7. Analysis of data For the 1993 experiment, two farmers neglected their clams; these sites were removed from the analyses (Fig. 1). In addition, cages at two of the easternmost sites (Sites 13, 14, Fig. 1) were badly damaged by storm surge on January 28, 1994. Data collected at these two sites after the storm passed were also excluded from all analyses.

T.P. Foyle et al./Aquaculture

92

148 (1997) 85-104

Table 2 List of variables used in the multiple regression equations for assessing effects on survival and growth during the 1993 and 1994 experiments. For geographic location, sites in the New Georgia group were categorized as western sites, whereas those in the Russell Islands, Florida Islands, and Guadalcanal were considered eastern sites (see Fig. 1). For water flow, temperature, salinity, and Secchi disc readings, mean values of all the measurements were used in the regressions Independent

variable

Time Fetch area Settlement ring time Geographic location Survival Cage depth Ranellids present Water flow Temperature Salinity Secchi disc reading a Removed

Units

No. days after transfer km* 1 week or 4 weeks ‘Dummy variable’ 1 = east, 2 = west % m Total no. found at a site ‘Diffusion factor’, unitless “C %c m

from analysis due to high correlation

1993 experiment

1994 experiment

Survival

Survival

Growth

::

J

Growth

;

J

:: J

; :: J

; ::a

Y J

with other variables.

No such problems were encountered in the 1994 experiment; data from all ten sites were used. All tests were conducted using Statistica software (StatSoft, Inc., Tulsa, OK, USA). Differences between sites and between treatments of the concurrent experiments were examined by ANOVA. Survival per cage and mean growth per cage at the end of the experiments were used in these tests. For site comparisons in the 1994 experiment, only data from the three ‘standard’ grooved cages at a site were included in the ANOVA tests. All data had homogeneity of variances. The importance of physical and biological variables in explaining differences in survival and growth among sites was evaluated using multiple regression. The magnitudes of the beta values (standardized partial regression coefficients) and partial correlations of the independent variables calculated during the regression analysis revealed the relative contribution of each independent variable in the explanation of the variance of the dependent variable (Winer et al., 1991). Beta values were therefore used to rank independent variables in relative order of importance. The number of independent variables used differed among the two experiments and for the dependent variables of growth and survival within experiments (Table 2). In all cases, data points with residual values falling outside +3 standard deviations were removed as outliers. For the 1994 experiment, only data from the three ‘standard’ grooved cages at a site were used in these regressions. Survival data were made linear by calculating the ‘hazard rate’ (Kennedy and Gehan, 1971). The hazard rate is the number of individuals dying per time unit in the interval divided by the mean number of survivors alive at the midpoint of the interval. This approach has been adopted in biomedical studies on survival because loss of individuals can be very high at first (such as after an operation) and then stabilize with time. Due to

T.P. Foyle et ul./Aquuculture

148 (1997) 85-104

93

high initial loss of clams in the 1993 experiment, the ‘hazard rate’ was a more effective linearization technique than log transformations of the data. The hazard rate is defined as follows:

Di Hf=

(Ni-D,/2).T

where H, = the hazard rate; Di = the number of individuals dying in the time interval; Ni = the number of individuals alive at the beginning of the time interval; T = the number of days in the interval. For the first experiment, mean growth per site was compared with mean percentage survival using simple regression on data collected 5 months after transfer to the villages. For the 1994 experiment, individual growth measurements taken at the end of the experiment (8 months) were used to provide enough data points for the multiple regression equation. Individual growth was approximated as follows: G = (SL, - &)/T where G = growth (mm month-‘); SL, = final shell length (mm> of an individual; & = mean initial shell length (mm> per cage; T = time in months between measurements.

3. Results 3.1. Experiment Survival was highly variable among sites, ranging from 6.7% to 82.6% at the termination of the first experiment. It was generally poorer at the eastern sites compared with those in the west (Table 3). At the completion of the experiment, mean survival at sites where ‘settlement rings’ were retained for 4 weeks was 49.0% (k23.0 s.d.1 whereas it was 15.7% (+9..5 s.d.) at sites where ‘settlement rings’ were retained for 1 week. At all sites, loss of clams was highest during the first 3 months. After this time, survival stabilized (Fig. 3). Mean survival after 8 months was 32.4%. The results of the multiple regression analysis are presented in Table 4. All four variables examined had a significant impact on survival. Based on the magnitudes of the beta values and partial correlations, geographic location and time were the most important variables explaining the ‘hazard rates’ of survival, followed by fetch area and ‘settlement ring’ time. The weakness of the ‘settlement ring’ variable was partially due to its moderate correlation with both geographic location and fetch area (Table 4). This is because farmers at western sites left the rings in place more often than their counterparts at eastern sites (Table 3). Rings were also removed more commonly from exposed sites after 1 week but left in place for 4 weeks at more sheltered sites. Mean shell growth at the sites varied from 2.3 to 8.6 mm month-’ and followed similar trends to the survival data (Table 3) since sites with the highest survival also had

T.P. Foyle et ul./Aquaculture

94

148 (1997) 8.5-104

Table 3 Information on the time ‘settlement rings’ remained in the cages, mean survival, and shell growth at the 14 sites analysed from the 1993 experiment. Growth values are based on shell length measurements taken at 5 months (March). These measurements were used because crowding appeared to inhibit growth at the best sites by the end of the experiments (see Fig. 5(a)>. For site designations, refer to Fig. 1 ‘Settlement ring’ retention time

Site

% Mean survival at end of experiment (k s.d.)

(weeks) Western sites 1 2a 4aa 5 6a 7 8a 10 Mean, west

Eastern sites 11 12 13 14 15 16a Mean, east

Mean shell growth (mm month ’ )

82.6 ( f 4.4) 16.3 ( f 7.6) 6.7 ( f 3.7) 42.4(+_5.8) 33.8 (* 1.1) 40.6 (+_ 5.3) 57.4 ( f 8.4) 59.6 ( f 10.8) 42.4 (+ 23.9)

4

8.6 3.9 3.6 4.4 4.9 6.3 4.9 5.1 5.2

11.5(f4.0) 12.8 ( i 4.4) 10.6 (+ 2.5) 13.9(+1.6) _” _b

1

Overall

2.9 3.4 2.5 2.3

1 Ott

I

b

_

12.2 (C 3.2)

2.8

32.4 ( f 24.2)

4.4

a CAC field station. b Storm damaged sites; data excluded after passage of the storm (January

0

b

_

I

Dee

Fig. 3. Survival curves over a period of 8 months experiment ( n ).

I

I

Feb

I

28, 1994).

I

APr

for Tridacna squamosa

I

I

Jun

during

the 1993 (Cl) and 1994

T.P. Foyle et al./Aquaculture

148 (1997) 85-104

95

Table 4 Results of multiple regression analysis of the ‘hazard rates’ (see text) for survival of clams during the 1993 experiment (a) and the correlation matrix of variables (b) (a) Variable

B x 10-s

Beta

Partial correlation

Intercept Geographic location Time Fetch area Settlement ring time

2012.4 - 563.9 - 3.6 4.6 - 53.5

- 0.46 - 0.43 0.28 -0.13

- 0.40 -0.52 0.26 -0.13

t value (d.f. = 381) 23.9 - 8.6 - 11.9 5.4 - 2.5

Time

Fetch area

Settlement ring time

Geographic

- 0.39 0.39 -0.14

0.42 - 0.46

P-level ***

i +* **s *. . *

(b) Fetch area Settlement ring time Geographic location Hazard rate

0.04 0.05 0.06 - 0.46

location

- 0.44

B, partial regression coefficient. Beta, standardized partial regression coefficient. Independent variables ranked in order of importance based on the magnitudes overall regression, R* = OS?;* rc4, 38,j = 95.5. , Significant at P < 0.05; , significant at P < 0.001.

of the beta values.

For the

the best growth rates. Regression revealed that the two variables were highly correlated (Fig. 4). The time required to grow clams from 24 to 50 mm shell length, a size suitable for sale to the aquarium market, was calculated to be 6.8 months. This is a conservative

I

0

20

I

40 % Survival

I

I

60

80

Fig. 4. Regression line and 95% confidence interval of mean growth versus mean survival for the 1993 experiment (n = 12). Data collected at 5 months. The strong positive correlation between the two variables indicates that sites with the best survival also had the highest growth rates during this experiment.

T.P.

96

Foyle

et al./Aquaculture

148

y = 24.01

(1997)

8.5-104

r2 = 0.82

+ 3.83x

y=24.31

+5.18x

lo0

, 0

, 2

. ;

, 4

;

, 6

.

, 8

.

( 10

lime (months)

Fig. 5. Plots of mean shell length versus time for the (a) 1993 experiment (n= 36) and (b) 1994 experiment (n = 30). Solid line represents regression of entire data set. Dashed line is best growth (Site I), which is based on data collected at 5 months for the 1993 experiment and at 5 and 8 months for the 1994 experiment. Dotted lines represent calculated time to grow from 24 mm to 50 mm shell length.

estimate months). site (Site collected

because crowding appeared to inhibit growth at the best sites by June (8 Excluding data for June reduces the time interval to 6.3 months. For the best l), calculated time to 50 mm was 3.3 months, based on the shell length data at 5 months (Fig. 5).

3.2. 1994 experiment 3.2.1. Overall site differences Both survival and growth were less variable during the 1994 experiment than in 1993. Average survival (66.6%) was twice as great in the 1994 study (Table 5). The severe mortality seen in the initial 3 months of the 1993 experiment did not eventuate in 1994 (Fig. 3). Mean growth improved slightly in the 1994 experiment to 5.3 mm month-’ (Table 5), which reduced the average time to reach 50 mm shell length to under 5 months (Fig. 5). At eastern sites in particular, survival and growth were substantially higher for the second experiment. Site differences remained important during the 1994 experiment; both survival

T.P. Foyle et al./Aquaculture

148 (1997) 85-104

97

Table 5 Information on water flow, abundance of predators, water clarity, mean survival, and shell growth for the ten sites used in the 1994 experiment. For site designations, refer to Fig. I. Growth values are based on shell length measurements taken at the end of the experiment. For water flow measurements, DF = ‘diffusion factor’ (Doty, 1971) Site

Mean water (DF)

Total no. ranellids found

Mean Secchi disc reading

5.6 6.4 3.7 4.3 3.3 4.3 2.2

1 27 5 22 39 3 6

Eastern sites 11 14 16b Mean, east

2.7 3.2 4.3

Mean, overall

4.0

flow

% Mean survival at end of experiment ( f s.d., n = 3)

Mean shell growth (mm month ’ )

8 19 20 10 8 13 7

76.2 62.7 75.1 76.7 80.7 64.7 42.2 68.3

7.1 5.7 3.6 5.7 4.6 5.5 6.0 5.5

32 4 15

16 13 19

52.7 (k 83.1 (k 51.6(zt 62.4 (+

10.6) 5.7) 15.1) 18.3)

4.2 4.9 4.2 4.4

15

13

66.6 (* 15.5)

5.3

(ml

Western sites 1 3 4b a 6b 8b 9 10 Mean, west

(+ 2.3) ( f 5.3) (i3.1) (5 5.5) (k 9.5) ( + 14.4) ( i 18.0) (+ 14.3)

a CAC field station.

(one-way ANOVA; F&,, 20j = 7.0, P < 0.001) and growth (one-way ANOVA; Fo, 2oj = 43.0, P < 0.001) differed significantly among sites. Trends in growth and survival among sites were consistent for the two experiments (cf. Tables 3 and 5). Multiple regression results for survival and growth during the 1994 experiment are presented in Table 6. Except for time, no variables significantly explained the ‘hazard rates’ of survival. This may have been caused by the low mortality between time intervals and therefore low hazard values. For growth, all the variables tested had significant effects except salinity. Based on the beta values and partial correlations, the three most important variables explaining growth were Secchi disc visibility, water flow, and the number of predatory ranellids found in the cages. Unlike the 1993 experiment, percentage survival was only a weak predictor of growth and the relationship between the two variables was negative. The variables used in the multiple regression were correlated to various extents (Table 7). For instance, fetch area was moderately correlated to water flow (0.541, revealing that water movement often increased at more exposed sites. Similarly, fetch area was moderately correlated to Secchi disc visibility (0.59) suggesting that sheltered sites were generally more turbid. Temperature was not included in the regressions due to high correlations with geographic location (0.75) and other variables.

T.P. Foyle et al./Aquaculture

98

148 (1997) 85-104

Table 6 Results of multiple regression analyses of the ‘hazard rates’ (see text) for survival and growth (mm monthof clams during the 1994 experiment Variable

B

Beta

Partial correlation

Survival (dlf: = 215) Intercept Secchi disc reading Time Cage depth Fetch area Geographic location Ranellids present Water flow Salinity

475.1 x10-s 12.0x 10-s -0.5x 10-r -50.8x 1O-5 -0.2x 10-s - 15.3 x 10-s 1.0x 10-s 6.7x IO-’ - 11.7x 10-s

0.27 -0.18 -0.11 - 0.07 - 0.03 0.06 0.04 - 0.03

0.13 -0.18 - 0.08 - 0.03 - 0.02 0.04 0.02 -0.01

- 0.68 0.47 -0.41 0.29 - 0.27 - 0.24 -0.14 - 0.07

- 0.37 0.31 - 0.28 0.15 -0.14 - 0.20 -0.15 - 0.03

Growth (d$ = 1480) Intercept Secchi disc reading Water flow Ranellids present Fetch area Geographic location Cage depth Survival Salinitv

17.12 - 0.22 0.58 - 0.05 0.01 - 0.89 -0.81 -0.01 - 0.23

t value

’)

P-level

0.22 1.9 -2.8 - 1.1 -0.4 -0.2 0.6 0.3 -0.2

n.s. n.s. ** n.s. n.s. ns. n.s. n.s. n.s.

2.8 -15.3 12.4 -11.1 6.0 -5.5 -7.9 -5.7 - 1.2

** *** *** *** *** *** ** * *** ns.

B, partial regression coefficient. Beta, standardized partial regression coefficient. Independent variables ranked in order of importance based on the magnitudes of the beta values. For the overall survival regression, R* = 0.10; FC8,2,5j = 2.9. For the growth regression, RZ = 0.46; FC8.,480j = 157.9. * * , Significant at P < 0.01; * * * , significant at P < 0.001; n.s., not significant.

Table 7 Correlation

matrix for variables used in the multiple regression

Variable

Time

Fetch area

Geographic location

Cage denth

Fetch area Geographic location Cage depth Ranellids present Water flow Temperature Salinity Secchi disc reading Hazard rate Survival Growth

-0.01 0.01

0.35

-0.01 -0.01

0.19 0.21

- 0.25 - 0.07

- 0.50

-0.02 0.01 -0.03 - 0.02

0.54 -0.13 0.37 0.59

0.33 0.75 - 0.29 - 0.39

-0.19

0.09 0.09 -0.12

-0.12 0.18 0.33

analyses

Ranellids present

Water flow

- 0.08 -0.57 0.29 0.48

- 0.04 0.29 -0.28 0.01

-0.10 0.61 0.29

-0.03 - 0.03 - 0.29

0.12 - 0.02 - 0.22

0.07 0.21 0.30

of the 1994 experiment

Temperature

-0.66 -0.69 -0.10 0.11 0.23

Salinity

Secchi reading

Survival

0.62 0.11 0.32 - 0.03

0.20 -0.12 -0.42

0.01

T.P. Foyle et al./Aquuculture

Table 8 ANOVA results of percent survival and mean growth per cage (mm monthconducted during the 1994 experiment Effect

Survival

d.f.

F

Error

*** I I * **

F

P value

0.9 6.6 0.5

ns. ns.

8

Treutment: growed Site

versus smooth cages 3 19.1

I

Treatment Site X treat Error

3 16

Treaunent Site X treat Error

.I*

0.11 1.55

Treatment: no coral versus cord Site 3

Treatment: Site

’ ) for the concurrent experiments

Growth P value

Treutment: tunk-reared uersus ocean-reared Site 1 33.2 Treaunent 1 57.3 Site X treat 1 12.6

99

148 (1997) 85-104

1 3 16

13.4 0.0 0.3

n.s. n.s.

l

** ns. ns.

‘settlement ring’ 2 months versus 4 months 3 13.4 ***

Treatment Site X treat Error

* , Significant

1 3 16 at P < 0.05;

3.2.2. Concurrent

0.0 0.3

ns. n.s.

199.7 1.9 2.7

28.3 0.0 1.8

28.3 2.3 1.o

111

ns. n.s.

*** ns. ns.

**

I

ns. n.s.

’ ’ , significant at P < 0.01; * * * , significant at P < 0.001; n.s., not significant.

experiments

In most cases, there were significant differences in growth and survival between sites for the concurrent experiments (Table 8). The patterns of these differences were similar to the overall analysis of site differences described in the previous section. There was a significant site X treatment interaction for survival in the ‘tank-reared versus ocean-reared’ sub-experiment. This was because survival of juvenile clams preconditioned in the ocean was consistently high at the two sites used for this experiment (83 and 92%) whereas survival of tank-reared individuals was variable at the two locations (42 and 77%). However, tank-reared individuals grew at a significantly faster rate once placed at the grow-out locations (Table 8). After 8 months, shell lengths were similar between the tank-reared (71.5 mm + 2.4 s.d.) and ocean-reared (72.1 mm + 2.8 s.d.1 animals because tank-reared juveniles were smaller when placed in the field. No significant effects were found for treatments in the other sub-experiments (Fig. 6, Table 8). Survival and growth were similar for juveniles placed in grooved or smooth cages, supported with coral or left free-standing, or grown in cages where ‘settlement rings’ were removed after 2

100

T.P. Foyle et al./Aquaculture

al 6

148 (1997) 85-104

100 80

TT

TO

GS

NC C

52 s4

Fig. 6. Comparisons of (a) mean survival and (b) mean growth frc ,m the four concurrent experiments conducted during the 1994 trials. Data pooled across sites. Key: T, tank-reared clams; 0, ocean-reared clams; G, grooved cages; S, smooth cages; NC, no coral pieces in cages; C, coral piecesin cages; S2, settlement rings removed after 2 months; S4, settlement rings removed after 4 months.

months or 4 months. Overall, clams appeared to be slightly less clumped in grooved cages compared with smooth cages.

4. Discussion Our study shows that great differences in the performance of giant clams grown at village sites can occur from year to year. Mean survival (67%) during the second experiment, for instance, was twice that during the previous year (32%). The statistical importance of geographic location and fetch area in explaining survival in the 1993 experiment indicates that adverse weather conditions affected mortality during that year: strong winds and heavy precipitation were more common during the first experiment than the second. Predation by ranellid snails would also have had an impact since high numbers of ranellids were observed at some eastern sites early in the 1993 experiment. The fact that the cohorts used for the two experiments were derived from different parents, and that aquaculture protocols were improved in 1994, may also have accounted for the difference in performance between years. ‘Settlement rings’, for instance, remained in place for at least 2 months during the 1994 experiment compared with 1 week or 1 month during the 1993 experiment. The importance of settlement rings for

T.P. Foyle et al./Ayuaculture

148 (1997) 85-104

101

reducing mortality of small juvenile clams has been confirmed in a subsequent experiment conducted at the CAC. After 2 weeks, the number of T. maxima juveniles remaining in cages with ‘settlement rings’ was 85% compared with 20% in the cages without rings (J. Bell, unpublished data, 1994). In addition to annual variability, our research revealed great variability in growth and survival among sites. These differences were generally consistent for the two experiments. Sites with high survival and growth in the first experiment usually yielded good results in the second one. Similarly, sites which performed poorly in the first experiment generally proved less satisfactory in the second. This outcome was unexpected since all the sites were surveyed by trained technical staff at the CAC and appeared suitable for farming giant clams based on the criteria of Govan (1993). Strong differences in survival between comparable sites have not been reported widely in the literature, probably because most other studies have been done at limited numbers of locations. When multiple sites have been selected, it has often been to test widely different environmental conditions. Gomez and Belda (1988) found that T. derusa juveniles displayed poor survival and growth when reared in silty water, but that growth and survival improved remarkably when transferred to relatively clear water. Lucas et al. (1989) found that survival and growth of juvenile T. gigas were poorer at a more oceanic locality than at a sheltered bay. They attributed this to disturbance from turbulence. In the 1993 experiment, ‘fetch area’ was an important variable explaining survival. Since most sites were well protected from the oceanic swell, this suggests that even moderate levels of turbulence may deleteriously affect survival. As T. squumosu attaches weakly to the cement bases of the cages, such turbulence may continually dislodge clams. During the 1993 experiment, the highest survival occurred at the most sheltered site (Site 11, situated in a narrow strait with high water flow but well protected from waves generated by wind. With moderate weather, however, more exposed sites may be suitable for clam rearing. This appears to have been the case in the 1994 experiment. Proper site selection is therefore important for ensuring relatively high yields for farmers. Increase in shell length was correlated strongly with survival in the 1993 experiment. Locations that yielded the highest survival were also the best for growth. The correlation between growth and survival was weak and negative in the 1994 experiment, implying that crowding began to affect growth by the end of the second experiment, even though the number of clams placed in the cages was reduced from 250 to 150 for the 1994 study. Growth was significantly correlated to seven variables in the 1994 experiment, the most important being Secchi disc visibility, water flow, and number of ranellid snails found in the cages. Growth may have improved as Secchi disc visibility decreased because the particulate matter and phytoplankton in the water, which reduced visibility, provided more food for juveniles; Klumpp et al. (1992) calculated that filter feeding contributed 65% of the nutritional requirements of small juvenile T. gigus. While turbidity would reduce the amount of light available to the symbiotic zooxanthellae, Lucas et al. (1989) found that 50% shade only reduced growth of juvenile T. gigus marginally compared with full sunlight.

T.P. Foyie et al./Aquaculture

102

148 (1997) 85-104

High water flow may promote growth by bringing particulate matter, phytoplankton, or nutrients to the clams. The presence of ranellid gastropods may reduce growth if clams expend energy in defence of the predators. Ranellid numbers may also be an indirect indicator of the amount of husbandry performed by the farmers; poor husbandry may account for reduced growth. There are, however, two reasons to be cautious about the significance of these correlations. Firstly, measurements made at monthly intervals may not reflect fully the true environmental or biological conditions at the sites. Secondly, the variables were correlated with each other to various extents so these effects will require further validation. Changes to aquaculture protocols introduced in 1994 and tested in the concurrent experiments, were generally ineffective in improving survival and growth. The exception was preconditioning clams in the ocean, which increased their robustness once transferred to the village sites. Presumably, their larger size (28 mm) at transfer, compared with exclusively tank-reared animals (24 mm>, was partly responsible for their significantly greater survival. Although preconditioned clams grew at a slower rate, this would not impose a penalty on farmers. Being larger to begin with, preconditioned juveniles attained the same size as tank-reared clams after 8 months. The advantage of preconditioning clams now needs to be weighed against the costs of providing ocean grow-out facilities and the potential for greater mortality during the preconditioning period. Mean growth of juveniles in these experiments (4.4 and 5.3 mm month-‘) was generally higher than reported in other studies (Table 9) and compares favourably with growth rates of juvenile T. gigas in the Solomon Islands (4.1 mm month-’ + 1.4 s.d.; Bell et al., 1996). Based on these experiments, T. squamosa clams can, on average, be grown to a size suitable for the aquarium market in only 5 to 7 months. Economic viability is a more difficult concept to address since it depends on the initial cost of the seed clams, which in turn varies with the scale of production (cf. Hambrey and Gervis, 1993). The CAC currently sells lo-month-old (- 24 mm> seed T. squamosa to farmers for US$O.33 each, a price believed to be economically viable for a small hatchery (M. Gervis, unpublished analysis, 1994). For T. squamosa which average 50 mm shell length, producers presently receive a mean of US$l.86 per clam from distributors to the aquarium trade. The ‘break-even’ level for farmers, assuming no additional costs, is

Table 9 Rates of growth of juvenile Tridacna squamosa reported Growth (mm month3.3 4.0-4.5 2.3 2.3-7.0 2.9 2.3-8.6

in the literature

Size (mm)

Location

Source

28-67 29-76 13-49 4-50 20-35 24-80

Palau

Beckvar(l981) Munro and Heslinga ( 1983) Gomez and Belda (1988) Crawford et al. (1988) Solis (1989) This study

’)

Philippines Great Barrier Reef Philippines Solomon Islands

T.P. Foyle er al./Aquaculrure

148 (1997) 85-104

103

therefore 18% survival during the grow-out period. Compared with local wages in the Solomon Islands, the value of small live tridacnids destined for the aquarium market is high. Bell et al. (1996) calculated that farmers growing T. gigas to 70 mm could make substantial income with 54% survival during the grow-out phase. Village based farming of tridacnid clams therefore appears to be economically feasible, provided sites are satisfactory and a stable source of hatchery-reared juveniles is available. Our experiments provide detailed information on survival and growth of T. squumosa juveniles during ocean culture and point to some of the factors affecting grow-out. The best sites appear to be well sheltered with strong flow delivering sufficient suspended particulate matter and phytoplankton to the clams. Similar large-scale grow-out trials are now necessary for the other species of giant clams valuable to the aquarium trade. Further research also needs to be carried out on: (1) the physical and biological factors accounting for variability in survival and growth, (2) the importance of size at stocking on survival, and (3) the effects of genetic differences between cohorts on survival and growth Understanding these factors is the key to maximizing survival and growth and improving economic returns for farmers.

Acknowledgements

We thank S. Soule, H. Tafea, and staff at the ICLARM Coastal Aquaculture Centre for technical assistance, and G. Courtois de Vicose for culturing the ocean-reared clams. The manuscript benefited from the comments of J. Lucas, J. Maclean, J. Munro, and M. Prein. The experiments would not have been possible without the cooperation and participation of the villagers.

References Beckvar, N., 1981. Cultivation, spawning, and growth of the giant clams Tridacm gigus, T. derasu, and 7. squamosa in Palau, Caroline Islands. Aquaculture, 24: 21-30. Bell, J.D., Lane, I., Gervis, M., Soule, S., and Tafea, H., 1996. Village farming of the giant clam, Triducnu gigas, for the aquarium market: initial trials in the Solomon Islands. Aquaculture Res., in press. Braley, R.D. (Editor), 1992. The Giant Clam: Hatchery and Nursery Manual. ACIAR Monograph No. 15, Australian Centre for International Agricultural Research, Canberra, 144 pp. Calumpong, H.P. (Editor), 1992. The Giant Clam: An Ocean Culture Manual. ACIAR Monograph No. 16, Australian Centre for International Agricultural Research, Canberra, 68 pp. Calumpong, H.P. and Solis-Duran, E., 1993. Constraints in restocking Philippine reefs with giant clams. In: W.K. Fitt (Editor), The Biology and Mariculture of Giant Clams: A Workshop Held in Conjunction with the 7th International Coral Reef Symposium, 21-26 June 1992, Guam, USA. ACIAR Monograph No. 47, Australian Centre for International Agricultural Research, Canberra, pp. 94-98. Crawford, C.M., Lucas, J.S. and Munro, J.L., 1987. The mariculture of giant clams. Interdisciplinary Sci. Rev., 12: 333-340. Crawford, CM.. Braley, R.D. and Nash, W.J., 1988. Interspecific growth rates of cultured giant clams on the Great Barrier Reef. In: J.W. Copland and J.S. Lucas (Editors), Giant Clams in Asia and the Pacific. ACIAR Monograph No. 9, Australian Centre for International Agricultural Research, Canberra, pp. 193-196.

104

T.P. Foyle et al./Aquaculture

148 (1997) 85-104

Doty, M.S., 1971. Measurement of water movement in reference to benthic algal growth. Bot. Mar., 14: 32-35. Gomez, E.D. and Belda, C.A., 1988. Growth of giant clams in Bolinao, Philippines. In: J.W. Copland and J.S. Lucas (Editors), Giant Clams in Asia and the Pacific. AClAR Monograph No. 9, Australian Centre for International Agricultural Research, Canberra, pp. 17% 182. Govan, H., 1993. Participatory research in giant clam farming. Naga. ICLARM Q., 16(l): 8-10. Govan, H., 1995. Cymatium muricinum and other ranellid gastropods: major predators of cultured tridacnid clams. ICLARM Tech. Rep. 49, 136 pp. Hambrey, J. and Gervis, M., 1993. The economic potential of village based farming of giant clams (Triducna gigas) in Solomon Islands. In: W.K. Fitt (Editor), The Biology and Mariculture of Giant Clams: A Workshop Held in Conjunction with the 7th International Coral Reef Symposium, 21-26 June 1992, Guam, USA. ACIAR Monograph No. 47, Australian Centre for International Agricultural Research, Canberra, pp. 138-146. Heslinga, G.A., Watson, T.C. and Isamu, T., 1988. Giant clam research and development in Palau. In: J.W. Copland and J.S. Lucas (Editors), Giant Clams in Asia and the Pacific. ACIAR Monograph No. 9, Australian Centre for International Agricultural Research, Canberra, pp. 49-50. Heslinga, G.A., Watson, T.C. and Isamu, T., 1990. Giant Clam Farming. Pacific Fisheries Development Foundation (NMFS/NOAA), Honolulu, Hawaii, 179 pp. Kennedy, A.D. and Gehan, E.A., 1971. Computerized simple regression methods for survival time studies. Comput. Programs Biomed., 1: 235-244. Klumpp, D.V., Bayne, B.L. and Hawking, A.J., 1992. Nutrition of the giant clam Tridacm gigas. I. Contribution of filter feeding and photosynthates to respiration and growth. J. Exp. Mar. Biol. Ecol., 155: 105-122. Lucas, J.S., 1994. The biology, exploitation, and mariculture of giant clams (Tridacnidae). Rev. Fish. Sci., 2: 181-223. Lucas, J.S., Nash, W.J., Crawford, CM. and Braley, R.D., 1989. Environmental influences on growth and survival during the ocean-nursery rearing of giant clams, Triducnu gigas CL.). Aquaculture, 80: 45-61. Munro, J.L., 1993. Giant clams. In: A. Wright and L. Hill (Editors). Nearshore Marine Resources of the South Pacific. Forum Fisheries Agency, Honiara, Institute of Pacific Studies, Suva, pp. 43 I-449. Munro, J.L. and Heslinga, G.A., 1983. Prospects for the commercial cultivation of giant clams (Bivalvia: Tridacnidae). Proc. Gulf Caribb. Fish. Inst., 35: 122-134. Solis, E.P., 1989. Growth of laboratory reared tridacnid clam juveniles under natural and laboratory conditions. In: E.C. Zaragoza, D.L. de Guzman and E.P. Gonzales (Editors), Culture of Giant Clams (Bivalvia: Tridacnidae). Proceedings of the Symposium Workshop on the Culture of Giant Clams (Bivalvia: Tridacnidae), 15- 17 March 1988, Silliman University, Dumaguete City, pp. 46-56. Winer, B.J., Brown, D.R. and Michels, K.M., 1991. Statistical Principles in Experimental Design. 3rd edn., McGraw-Hill Series in Psychology, McGraw-Hill, Inc., New York, 1057 pp.