Water quality and plankton densities in

E-mail: danielle.johnston@utas.edu.au Abstract Water quality and plankton densities were moni-tored in shrimp ponds at 12 mixed shrimp-mangrove forestry farms in Ca Mau province, south-e

Water quality and plankton densities in mixed

shrimp-mangrove forestry farming systems in Vietnam

D Johnston1, M Lourey2, D Van Tien3, T T Luu3& T T Xuan3

7250, Australia

Correspondence: Danielle Johnston, School of Aquaculture, Tasmanian Aquaculture and Fisheries Institute, University of Tasmania, Locked Bag 1±370, Launceston, Tasmania 7250, Australia E-mail: danielle.johnston@utas.edu.au

Abstract

Water quality and plankton densities were

moni-tored in shrimp ponds at 12 mixed

shrimp-mangrove forestry farms in Ca Mau province,

south-ern Vietnam, to detail basic water chemistry and

assess whether conditions are suitable for shrimp

culture In general, water quality was not

optimal for shrimp culture In particular, ponds

were shallow (mean + 1SE, 50.5 + 2.8 cm), acidic

(pH , 6.5), had high suspended solids (0.3 +

0.03 g l 1), low chlorophyll a/phytoplankton

con-centrations (0.2 + 0.05 mg l 1 and 8600 + 800

cells l 1 respectively) and low dissolved oxygen

(DO) levels (3.7 + 0.15 mg l 1) Eight out of the 12

farms sampled had potentially acid sulphate soils

(pH , 4.2) Salinity, DOand pH were highly

vari-able over short time-periods (hours); DOin

particu-lar was reduced to potentially lethal levels

(1±2 mg l 1) Seasonal variations in water

chemis-try and plankton communities (i.e salinity, DO,

phosphate, temperature, phytoplankton and

zoo-plankton densities) appear to be driven by

differ-ences in rainfall patterns The presence or absence

of mangroves on internal pond levees (`mixed'

versus `separate' farms) and the source of pond

water (rivers versus canals) were of lesser

import-ance in determining water quality patterns and

plankton biomass Zooplankton and macrobenthos

densities were sufficient to support the current

(low) stocking densities of shrimp However, natural

food sources are not adequate to support increases

in production by stocking hatchery reared post

larvae Increasing productivity by fertilization and/

or supplemental feeding has the potential for ad-verse water quality and would require improve-ments to water management practices Some practical strategies for improving water quality and plankton densities are outlined

Keywords: shrimp, water quality, Vietnam, integrated farming, mangroves, extensive shrimp culture, shrimp aquaculture

Introduction Shrimp aquaculture in Vietnam has undergone rapid expansion over the past two decades, particu-larly in the Mekong Delta (Lovatelli 1997; Phuong & Hai 1998) Despite this expansion, shrimp yields per unit area are in decline (de Graaf & Xuan 1998; Johnston, Trong, Tuan & Xuan 2000a, b) Poor shrimp yields in the Mekong Delta and other coun-tries with similar farming systems have been attrib-uted to several factors, including low quality and quantity of shrimp seed, poor pond management and infrastructure, overexploitation of wildstock and whitespot disease outbreaks (Sinh 1994; Binh, Phillips & Demaine 1997; Primavera 1998; de Graaf & Xuan 1998; Johnston, Clough, Xuan & Phillips 1999; Johnston et al 2000a, b) However, the extent to which poor water quality has contrib-uted remains largely unstudied, particularly in remote regions such as the Mekong Delta in Viet-nam Good water quality in shrimp ponds is essen-tial for survival and adequate growth (Boyd 1990;

Burford 1997) In the Mekong Delta, low primary

production, rapid rates of water column respiration,

and low rates of benthic decomposition have

already been suggested as possible factors limiting

shrimp production (Alongi, Dixon Johnston, Tien &

Xuan 1999a; Alongi, Tirendi, Trott & Xuan

1999b) On other South-east Asian shrimp farms,

disease problems have been attributed to poor water

quality (Phillips 1998) This study aims to address

the lack of basic information on water quality in

extensive shrimp ponds in Vietnam and comment

on the potential for deleterious effects on shrimp

aquaculture

Shrimp ponds in Vietnam are primarily extensive

shrimp±rice and shrimp-mangrove integrated

systems, although there has been an increase in

the number of improved extensive and

semi-inten-sive ponds (Binh & Lin 1995; Binh et al 1997) The

extensive ponds in the Mekong Delta rely on tidal

flushing for water exchange and post-larval

recruit-ment, so farmers have little control over the water

quality in their ponds Fortunately, extensive shrimp

farms such as those in southern Vietnam have low

stocking densities, little or no fertilization and no

supplementary feeding, so do not generate

signifi-cant amounts of organic effluent However,

man-grove deforestation, due to the uncontrolled

increase in the number of aquaculture ponds and

increasing population pressure, has emerged as a

threat to water quality in the region (de Graaf &

Xuan 1998; Johnston et al 1999; 2000b) The

effects of deforestation include acidic run-off and

discharges from ponds constructed on acid sulphate

soils (Phillips 1998), increased coastal erosion,

sal-inity intrusion and loss of shrimp nursery grounds

(Hong 1993; Macintosh 1996) Population pressure

and reliance by local communities on the

water-ways for transport and market locations may have

important impacts on water quality on a regional

basis

Data on shrimp pond water quality in the

Mekong Delta is limited to investigations of water

column (Alongi et al 1999a) and benthic

(Alongi et al 1999b) processes in just two

shrimp-mangrove ponds We introduce data from 12

ponds on 12 shrimp-mangrove forestry farms and

cover a range of environments, farm types and

both wet and dry seasons We present the first

information on phytoplankton, zooplankton and

macrobenthos densities, which are particularly

im-portant as they form the basis of the natural food

webs in extensive ponds and, in some cases, may

limit shrimp productivity The specific aims of this study were:

1 To describe water quality and plankton densities

in shrimp ponds from mixed shrimp-mangrove forestry farming systems in the Mekong Delta

2 To establish important trends with season (wet, dry), farm type (`mixed', `separate') and pond water source (rivers, canals)

3 To identify situations where water quality may

be deleterious to shrimp production

4 Make recommendations to improve pond water quality and plankton densities

Materials and methods Sample collection

The study was conducted in 12 shrimp ponds ranging in size from 0.5 to 6 ha at 12 (there is traditionally one pond per farm) mixed shrimp-mangrove forestry farms in the Ca Mau province of the Mekong Delta of Vietnam (Fig 1) (see Johnston

et al 1999) These are integrated extensive farming systems where ponds are effectively ditches dug either separate to or through mangroves Each pond consists of a series of long (250±800 m), narrow (3±4 m) interconnected channels separated

by internal levees and surrounded by a dyke Ponds are connected to external waterways via a single sluice gate through which water is exchanged Exchange during grow out is generally minimal although water levels can be maintained during tides of sufficient height and losses due to leakage are common Recruitment and harvesting of wild shrimp (primarily Metapenaeus spp.) occur on con-secutive flood and ebb tides of the spring tide period Recruitment is followed by 10±12 days of grow out (Johnston et al 1999) There is little or

no supplementary feeding, aeration, liming or fertil-izer treatment

Water samples or in situ measurements were col-lected from 20 cm below the surface at two stations within each pond, one 5±6 m from the sluice gate and one in the middle to back of the pond Two farm types were sampled: `separate' farms have separate shrimp pond and mangrove areas so the internal levees within each pond are devoid of mangroves;

on `mixed' farms the internal levees have man-groves planted at high densities Farms of each type were further categorized based on their location and source of pond water, i.e from a large river (rivers) or a small canal (canals) Sampling

was carried out in the morning between 08.00 and

10.00 hours, twice during the dry season (April and

June 1996) and twice during the wet season

(August and October 1996)

Vertical and diurnal profiles for pH, salinity,

tem-perature and DOwere recorded from ponds and

their adjacent river/canals (source waters) Both

profiles were measured in situ using Hydrolabß

Datasonde 3 dataloggers calibrated to factory

speci-fications Vertical profiles were recorded at

approxi-mately 11.00 hours in the morning during October

1996 (wet season) Data was recorded every 5 s as

the datalogger was lowered through the water

column Diurnal profiles in the ponds were recorded

for 7-day periods using dataloggers deployed 20 cm

from the bottom and approximately 6 m from the

sluice gate Profiles representative of the general

patterns were selected and are presented here

Water quality analyses Temperature and DOwere measured using an Orion oxygen meter; redox potential with Orion electrodes and salinity with a refractometer Replicate water samples from each station were filtered through a 0.45-mm filter and analysed for dissolved ammo-nium (NH4‡), nitrite (NO2-N), phosphate (PO4-P) and iron (Fe) using Pharmacia Biochrom (Palintest) test kits designed for a Novaspec II spectrophotom-eter The ammonia and nitrite tests were rated for saline water with incorporation of a conditioning agent to prevent the precipitation of salts Rudimen-tary facilities precluded the use of standard methods for total ammonia, NO2-N, PO4-P and total Fe ana-lyses (Grasshoff, Ehrhadt & Kremling 1983) Repli-cate 100-mL samples were filtered onto preweighed GFC filter papers, dried at 60C for 6±8 h and

China Hanoi

Laos

Thailand

Vietrlam Cambodia

Ca Mau

Ho Chi Minh

Kilometres

W N

E

S

Ca Mau

District Tran Van Thoi

District Dam Doi District Cai Nuoc

Song Granh Hao

Kinh Sau Dong

Kinh Doi Cuong

Kinh Dam Cung

Song Bay Hap

Bien Tay

Song Cua Lon

Kien Vang

Rach Duong Keo

District Ngoc Hien

South China Sea

Hoc Nang Bien Bong

Song Bo De Song Dam Chim Song Cua

Lo n

K

in h a iN

h a

Song D

am D oi

Song Dam Chim

ai N

Kinh Ong Don

TG3

184

Figure 1 Map of Ca Mau province in southern Vietnam, the location of this study The farms were located in State Fisheries Forestry Enterprises Tam Giang 3 (TG3) and 184.

reweighed for determination of total suspended

solids Particulate matter from 100 mL of each

sample was filtered onto a GFC filter, the chlorophyll

was extracted in 90% acetone and quantified by

spectrophotometry (Parsons, Maita & Lalli 1984;

Stirling 1985) Pond sediment from each sampling

station was collected and dried Dry soil pH and

redox potential were determined on slurries of the

dried sediment mixed with deionized water

Plankton and macrobenthos density

Water samples (1000 mL) were collected at each

station and fixed in 4% seawater±formalin for

phytoplankton density determination The samples

were allowed to settle for 24±48 h in the laboratory

and excess water was removed to a final volume of

100 mL Phytoplankton density (cells l 1) was

deter-mined on replicate 0.1-mL subsamples using a

Palmer±Maloney plankton counter Water (60 L)

was collected from 20 cm depth using a 15-L bucket

and filtered through a 30-mm plankton net The

zooplankton collected were fixed in 4% formalin in

seawater and the solution made up to 30 mL The

number of zooplankton in two 1-mL replicate

sub-samples were counted and the density of

zooplank-ton (no m 3) was extrapolated The contents of two

benthic grab samples (area 0.025 m2) were pooled

and fixed in 4% seawater±formalin for

macro-benthos density determination The organisms

were removed from the sample and the total wet

weight biomass (g m 2) and density (no m 2) were

determined

Statistics

Univariate anovas were used to explore seasonal

differences in the parameters measured between

the farm locations (river versus canal) and farming

type (`mixed' versus `separate') Given that the

design was not fully balanced, it was not possible

to do a single analysis involving all factors of interest

simultaneously, i.e season, farm type and location

Therefore, separate analyses were conducted to

ex-plore the interaction between season and farm type

effects and the interaction between season and

loca-tion In the season, location analysis, farms were

nested within location; therefore, the final anova

design was a three-factor mixed model However,

in the season, farm type analysis there was an

un-balanced number of farms, therefore, farm was not

included in the analysis, resulting in a two-way orthogonal design For each parameter measured, the assumptions of anova were checked using re-sidual plots; where the assumptions had been vio-lated, a square-root transformation was used In those analyses that had significant factors, Tukeys HSD post hoc test was used to determine the nature

of the differences

Additionally a manova (multivariate analysis of variance) was used to explore these structured data because more than one parameter was measured at each farm (pond) In this analysis, differences among levels in a factor (season, farm type, location) could be explored in multivariate space allowing differences to be found that would not be seen in univariate space Following the manova, significant differences were explored using a Canonical Dis-criminant Analysis (CDA) Each group was plotted

in the reduced multivariate space, in which the new axes (CD1 and CD2) explained a proportion of the total variability in the data Superimposed on this plot was the association between the new axes [which display the differences among the groups (farms)] and the parameters that were measured This is displayed as a vector diagram in which the direction and length of the vector is a measure of the association between the parameter and the axes This allowed differences among the groups (farms)

to be interpreted with respect to the water quality parameters measured

Results The shrimp ponds studied here were typically shallow, averaging just 50.5 + 2.8 cm (range 10±140 cm) On average, salinity was higher in the dry season (mean + 1SE 27.4 + 0.7) than the wet season (16.7 + 0.7) (F1, 89ˆ 121; P , 0.001) and higher at `mixed' farms (22.9 + 0.8) than `sep-arate' farms (20.3 + 1.5) (F1, 89ˆ 6.3; P ˆ 0.014) (Fig 2) The reduction in salinity during the wet season was more pronounced in ponds that source their water from rivers than in those that source water from canals (F1, 69ˆ 5.6; P ˆ 0.039) (Fig 3) Temperature was approximately 1C higher in ponds that source water from canals (28.5 + 0.3C) than from rivers (27.4 + 0.3C) (F1, 69ˆ 7.2; P ˆ 0.022) (Fig 3) There were no significant seasonal or farm type trends in tempera-ture Although variable (0.5±9.6 mg l 1), DOcon-centrations were generally low (3.7 + 0.15 mg l 1)

10

5

4

3

2

0

0.1

3)

4

3

2

1

0

0.2

0.3

0.4

0.5

0.2

0.4

0.6

0.8

1)

20

25

30

Dry

Dry

Dry

Dry

Dry

Wet

27 26

28

30

Dry

0.1 0

0.2

0.4

Dry

0.15 0.1

0.2

0.25 0.3

Dry

0.5 0

1

1.5 2

Dry

20

0

40 60

Dry

Figure 2 Mean + SE of a range of water quality parameters in ponds from two types of shrimp-mangrove farm (`mixed' and `separate') in the wet and dry seasons.

10

5

4

3

2

0

0.1

4

3

2

1

0

0.2

0.3

0.4

0.5

0.2

0.4

0.6

0.8

20

25

30

Dry

Dry

Dry

Dry

Dry

Wet

27 26

28

30

Dry

0.1 0

0.2

0.4

Dry

0.15 0.1

0.2

0.25 0.3

Dry

0.5 0

1

1.5 2

Dry

20

0

40 60

Dry

Figure 3 Mean + SE of a range of water quality parameters in ponds of shrimp-mangrove farms that obtain their water from rivers and canals in the wet and dry season.

There were no significant temporal or seasonal

trends in either ammonia and nitrite concentrations

(Figs 2 and 3), with mean concentrations of

0.13 + 0.02 mg l 1 and 0.01 + 0.002 mg l 1

re-spectively Phosphate concentration in the wet

season (0.41 + 0.03 mg l 1) was double that of the

dry season (0.21 + 0.02 mg l 1) (F1, 45ˆ 21.87;

P ˆ 0.001), but the wet season increase was larger

at `separate' (levees bare of mangroves) farms

than `mixed' (levees with mangroves) farms

(F1, 65ˆ 4.5, P ˆ 0.038) (Fig 2) Ponds were turbid

(suspended solid loads of 0.3 + 0.03 g l 1; range

0.03±1.54 g l 1) in both seasons and regardless

of farm type and water source Chlorophyll a

concentrations were generally low averaging

0.2 + 0.05 mg l±l and ranged from 0 to 0.5 mg l±l

Phytoplankton densities encountered during this

study were highly variable ranging from 1000 to

36 500 cells l 1 Similarly, zooplankton densities ranged from 1900 to 119 000 no m 3 Phytoplank-ton and zooplankPhytoplank-ton densities were around twofold higher in the dry season (11000 + 1300 cells l 1 and 33 600 + 4000 no m 3respectively) than the wet season (7000 + 1000 cells l 1 and 16 400 +

1700 no m 3 respectively) (F1, 89ˆ 12.1; P ˆ 0.001 for phytoplankton and F1, 89ˆ 15.3;

P , 0.001 for zooplankton) (Fig 3) Zooplankton densities were also 1.6 times higher in ponds at

`mixed' farms (28 300 + 1300 no m 3) than `sep-arate' farms (18 200 + 3000 no m 3) (F1, 89ˆ 6.14; P ˆ 0.015) (Fig 2) In contrast, macrobenthos biomass was threefold higher in ponds at `separate' farms (26 + 7 gm 2) than `mixed' farms (10 +

2 gm 2) (F1, 89ˆ 7.13; P ˆ 0.009) (Fig 2) Pond sediments were not highly reducing with a mean

eH of 7 mV Soil surrounding the ponds was acidic

at eight out of the 12 farms sampled (pH , 4.2), indicating that the majority of farms had acid sul-phate soils

The CDA explained 87% of the variation among the groups (farms) on the first two axes (Fig 4) The greatest difference among the groups was along the first axis (CD axis 1), which explained 72% of the variation This difference was largely due to the difference between the wet versus the dry season groups, with the two wet season groups situated at one end of CD axis 1 and the two dry season groups

at the other end The vector diagram of parameters (Fig 4) suggests that high salinity and zooplankton density and to a lesser extent high temperature and high DOoccurred during the dry season (these four parameters have vectors that are pointing positively along CD axis 1) The second axis (CD axis 2) ex-plained 15% of the variation among groups and separated the `separate' from the `mixed' farms Macrobenthos biomass is higher in the `separate' farms than the `mixed' farms and to a lesser extent

NH4, zooplankton density and phytoplankton dens-ity were also higher in `separate' farms The `mixed' farms during the wet season also seemed to be char-acterized by higher PO4and suspended solids con-centrations

Vertical profiles of DO, pH, salinity and tempera-ture within two ponds and their adjacent river/ canals are presented (Figs 5 and 6) These stations represent a `separate' pond (pond 22) and a `mixed' pond (pond 23) Both profiles were measured in the same season to allow for comparisons The general patterns were typical of others measured in the area and in other seasons The salinity in pond 22 was

CD Axis 2 (15 %)

CD Axis 1 (72 %) 3

−3

Sepwet

Sepdry Mixwet

SS Zoop Temp Salinity

Po4 NH4 Do 2 Phyto Macroden Macro biomass

Mixdry

Figure 4 Results of the canonical discriminant analysis

are displayed on the first two axes (CD axis 1 and 2) The

mean and 95% confidence limits for each group is

dis-played in the reduced multivariate space In the top

right-hand corner of the graph is a vector diagram for the

parameters measured The direction and length of the

vector for each parameter is an indication of the

associ-ation between the parameter and the CD axes and can be

used to interpret the differences among the groups (farms).

The vector for ammonia lies along the macrobenthos

oxygen concentration; Macroden, macrobenthos density;

Macro Biomass, macrobenthos biomass; MixDry, `mixed'

farms dry season; MixWet, `mixed' farms wet season NH4,

ammonia concentration; Phyto, phytoplankton; PO4

phos-phate concentration; SepDry, `separate' farms dry season;

SepWet, `separate' farms wet season; SS, suspended solids;

Temp, temperature; Zoop, zooplankton density.

similar to that below '3 m in the adjacent river and

increased with depth in both the river and the pond

(Fig 5) The salinity in pond 23 was higher than the

adjacent river (Fig 6) but in this case the water

column in both pond and canal appeared to be

well mixed In the river adjacent pond 22,

tempera-ture increased with depth in a similar pattern to

salinity In the pond however, thermal stratification

was evident with a sharp decrease in temperature

with depth In pond 23 a thermal gradient was

absent There were vertical gradients of DOin all

source waters (canals/rivers) and ponds DOlevels

in the canals/rivers decreased with depth from around 5±5.5 mg l 1 at the surface to around

4 mg l 1at 2 m (Figs 5 and 6) At the deeper station DOwas constant from 2 to 10 m Dissolved oxygen levels in deeper waters in pond 23 were depleted to a greater extent than pond 22 Both ponds and the canals/rivers were acidic with pH of around 4.5±6 OpH in pond 22 was similar to its adjacent river waters, whereas pH in pond 23 was higher than its adjacent canal waters

12

Pond 22 River 10

8

6

4

2

0

DO (mg L−1)

pH

12

10

8

6

4

2

0

Figure 5 Vertical profiles of DO, pH, salinity and temperature in the pond at farm 22 (a `separate' farm) and in the adjacent river from which water is obtained Profiles were recorded at approximately 11.00 hours during October 1996 (wet season) Data are individual recordings taken every 5 s as the datalogger was lowered through the water column.

Salinity, pH, temperature and DOin ponds varied

considerably over a time-frame of hours The

diur-nal profile in Figure 7 is typical of patterns displayed

in both seasons, in both `mixed' and `separate' ponds

and regardless of water source Water depth in pond

22 was maintained at a reasonably constant level

throughout the 7-day period when measurements

were taken However, dramatic, short-term

reduc-tions and increases in pond depth occurred with the

tide (about every 6 h; see diagonal arrows in Fig 7),

followed by a slow decrease in depth in the period between these larger events The rapid reductions in the depth coincided with increases in DO(see diag-onal arrows in Fig 7), whereas rapid increases in pond depth led to smaller increases in DOlevels Between these rapid events there were two distinct patterns; DOand pH were maintained or increased during the day or were drawn down during dark periods DOconcentrations were lowest (1±2 mg l 1) shortly before sunrise (vertical arrows in Fig 7)

3

2

1

0

DO (mg L−1)

pH

12

8

4

0

Pond 23 River

Figure 6 Vertical profiles of DO, pH, salinity and temperature in the pond at farm 23 (a `mixed' farm) and in the adjacent river from which water is obtained Profiles were recorded at approximately 11.00 hours during October 1996 (wet season) and in close proximity to the sluice gate in the pond Data are individual recordings taken every 5 s as the datalogger was lowered through the water column.

Water quality

Most of the seasonal patterns of water quality and

food chain dynamics observed here were driven by

differences in rainfall patterns between the wet and

dry season Low pond salinity and vertical

stratifica-tion (Fig 5) during the wet season were attributed

to run-off from monsoon rainfall events Lower

sal-inity at `separate' than `mixed' farms was most likely

due to higher run-off from bare internal levees on

the `separate' farms During the wet season, salinity

in ponds that obtain water from rivers was lower

than in ponds that obtain water from canals Water

exchange is achieved on the rising tide, so pond

waters fill with the surface water of the adjacent

river or canal If water exchanges are made while

the layer of freshwater dominates the surface

(Fig 5), then a considerable reduction in pond

sal-inity may occur This suggests that the dilution of

river waters by run-off is greater than in canals and

that these differences are passed on to pond waters

This is consistent with the idea that freshwater

fluxes would be greater in rivers than canals

be-cause rivers have larger catchments

Pond temperature was higher at farms that

obtain their water from canals rather than rivers,

due to greater heating of the smaller water mass in

the shallow canals compared with the larger rivers

(combined with reduced tidal flushing) Shading may moderate temperature changes, as thermal stratification in ponds with no mangroves was greater than in ponds where mangroves lined the levee banks A thermally stratified water column results in poor circulation and possibly stagnation from benthic heterotrophic processes (Alongi et al 1999a) Poor pond design (one sluice gate, long narrow shallow channels) and lack of mechanical aeration contribute to stratification Thermal strati-fication was evident in pond 22 and may maintain the vertical dissolved oxygen gradient (Fig 5) How-ever, the presence of a similar DOgradient in pond

23 in the absence of a vertical temperature (or salinity) gradient (Fig 6) suggests that stratification

is not necessary for high surface-water DOconcen-trations to develop

Ammonia and nitrite concentrations in ponds were well below toxic levels recorded for shrimp (, 1.3 mg l 1 NO2-N and , 7.7 mg l 1 NH4-N for Metapenaeus macleayi Haswell and , 4.1 mgl 1

NH4-N for Peneaus monodon Fabricius (Allan Maguire & Hopkins 1990) and are consistent with levels recorded previously in integrated shrimp-mangrove ponds and canals in Ca Mau province (Hong 1996; Binh et al 1997) The highest levels

of nitrogenous nutrients encountered in this study were most likely due to point sources of pollution in the source waterways rather than shrimp excretion,

1

0.8

0.6

0.4

0.2

0

8.5

8

7.5

7

6.5

6

1800 2100

300 600 900 1200 1500 1800 2100

300 600 900 1200 1500 1800 2100

300 600 900 1200 1500 1800 2100

300 600 900 1200 1500 1800 2100

300 600 900 1200 1500 1800 2100

300 600 900

1800 2100

300 600 900

1200 1500 1800 2100

300 600 900

1200 1500 1800 2100

300 600 900

1200 1500 1800 2100

300 600 900

1200 1500 1800 2100

300 600 900

1200 1500 1800 2100

300 600 900

14 12 10 8 6 4 2 0

1 )

31 30 29 28 27 26 25

Depth DO

pH Salinity

Time Figure 7 Diurnal cycles of depth, DO, salinity and pH over seven days within a grow-out cycle in the pond at farm 22 in March 1997 (dry season) Curves are means integrated by 3-h intervals of datalogger readings taken every 20 min Readings were taken 20 cm above the pond bottom, approximately 6 m from the sluice gate Vertical arrows indicate lowest levels of DOduring the 7-day grow-out period and diagonal arrows indicate rapid fluctuations in pond depth and DO.