or pond bottom through species composition on polycultures of large carps and small indigenous species

Aquaculture 286 (2009) 246–253

Contents lists available at ScienceDirect

Aquaculture j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / a q u a - o n l i n e

Effects of intervention in the water column and/or pond bottom through species composition on polycultures of large carps and small indigenous species A. Milstein a,⁎, M.A. Wahab b, A. Kadir b, M.F.H. Sagor b, M.A. Islam b a b

Fish and Aquaculture Research Station Dor, M. P. Hof Ha Carmel, 30820, Israel Dept. of Fisheries Management, Bangladesh Agricultural University, Mymensingh-2202, Bangladesh

a r t i c l e

i n f o

Article history: Received 12 June 2008 Received in revised form 21 September 2008 Accepted 22 September 2008 Keywords: Fish interactions Natural food web Polyculture SIS small indigenous species

a b s t r a c t In Bangladesh, a sustainable semi-intensive pond aquaculture technology including large carp species as cashcrop and small indigenous fish species (SIS) as food for the farmers' families is being optimized. The present paper is on the effects on fish performance and pond ecology of interfering in the water column and/or on the bottom through changes in the polyculture composition. The Control polyculture was the ‘no-effect’ combination consisting of the traditional 33 rohu–33 catla–34 common carp stocking with the addition of 250 SIS and 3 silver carp per 100 m2 of pond, as resulted from a previous experiment. Interferences on the water column were achieved by changing the density of the herbivorous fish (reducing the density of catla to 24/100 m2 and increasing that of silver carp to 12/100 m2), and on the bottom by doing so on the benthophagous fish (replacing 10/100 m2 common carp by the same amount of mrigal). Mola was the SIS included in the polyculture. Interfering in the water column and/or in the pond bottom through the polyculture composition produced complex responses in the pond ecosystem affecting the large carps' performances, while it did not significantly affect the reproduction and the harvested biomass of the small fish mola. Relationships among the different fish species and the environment are described for each polyculture. The four polycultures tested allowed a good production of large carp species as cash-crop, of silver carp as an option to consume or to sell, and of the small species mola as food for the farmers' families. The Control polyculture is appropriate to produce relatively large herbivorous species, mainly silver carp. The polyculture combination in the Water treatment is appropriate to obtain a larger amount of smaller silver carp that can be afforded by the poor people but also smaller rohu and catla, while maintaining the same level of total yield and income with reduced feed conversion ratio (FCR) than in the Control treatment. The polyculture combination in the Bottom treatment allowed a larger fish species diversity and also produced smaller herbivorous fish with still reduced FCR, while maintaining the same level of total yield and income than the Control treatment. The polyculture combination in the Water&Bottom treatment gave the best results: it allowed a larger fish species diversity, is appropriate to obtain a larger amount of small silver carp that can be afforded by the poor people, and gives the highest total yield and income with the lowest FCR. © 2008 Elsevier B.V. All rights reserved.

1. Introduction In Bangladesh, a sustainable semi-intensive pond aquaculture technology including large carp species as cash-crop and small indigenous fish species (SIS) as food for the farmers' families is being optimized. The first steps in the cash-SIS technology optimization were directed to interfere on the pond bottom through the bottom feeding fish (Wahab et al., 2002; Milstein et al., 2002; Wahab et al., 2003; Alim et al., 2004, 2005). The polyculture combinations studied included the filter feeders rohu (Labeo rohita) and catla (Catla catla), the bottom feeders mrigal (Cirrhinus cirrhosus) and common carp (Cyprinus carpio), and the SIS punti (Puntius sophore) and mola (Amblypharyngodon mola). At ⁎ Corresponding author. Tel.: +972 4 6390651x108; fax: +972 4 6390652. E-mail address: [email protected] (A. Milstein). 0044-8486/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.aquaculture.2008.09.036

present the research concentrates on the intervention in the water column through the addition of silver carp (Hypophthalmichthys molitrix). This fish has ecological and socio-economic potential advantages: it is expected to have a strong impact on the pond ecology because it is a very efficient filter-feeder (Milstein et al., 1985a,b; Milstein, 1992), and also on the farmers' family nutrition because it is a cheap fish that the family can afford to eat instead of selling. It is also easily accessible to the poorer section of the population because of its low market price. The first steps of the study of intervention in the water column on the ‘cash-SIS’ polyculture concentrated on the effects of adding silver carp and/or SIS (Kadir et al., 2006, 2007) and of partially substituting herbivorous Indian carps by silver carp (Milstein et al., 2008). This paper presents the effects on fish performance and pond ecology of intervention in the water column and/or on the pond bottom through changes in the polyculture composition.

Table 1 Results of ANCOVA split-plot and Duncan mean multicomparisons of each water quality variable, plankton and benthos PO4–P

TAN

NO2–N NO3–N Chl-a

Phytoplankton

Zooplankton

Benthos

(mg/l)

(mg/l)

(mg/l)

(mg/l)

(mg/l)

(µg/l)

(cells or colonies/l)

(ind/l)

(ind/100 cm2)

⁎⁎⁎ 0.87

⁎⁎⁎ 0.95

⁎ 0.51

⁎⁎ 0.54

ns 0.25

ns 0.34

⁎⁎⁎ 0.60

⁎ 0.46

⁎ 0.50

⁎ 0.48

ns ⁎⁎⁎ ns

ns ⁎⁎⁎ ns

Sign ns ⁎⁎⁎ ns

ns ⁎⁎⁎ ns

ns ⁎⁎⁎ ns

ns ns ns

ns ns ns

ns ⁎⁎⁎ ns

Sign ⁎ ⁎ ns

Mean multicomparisons by treatment (n = 18) Control 29.1 24 _b Water 29.0 28 a_ Bottom 29.0 29 a_ Water&Bottom 29.0 24 _b

5.8 5.8 5.8 5.8

8.0 7.9 8.0 7.8

92 81 80 88

0.81 1.17 1.53 1.15

0.24 0.28 0.31 0.37

0.009 0.009 0.010 0.009

0.02 0.04 0.03 0.04

100 106 95 111

100,440 ab_ 69,670 __c 73,250 _bc 103,780 a__

Mean multicomparisons by month (n = 12) May 31.2 _b____ Jun 29.0 __c___ Jul 26.1 _____f Aug 28.8 ___d__ Sep 31.5 a_____ Oct 27.7 ____e_

5.2 ___d_ 5.8 __c__ 6.7 a____ 5.8 __c__ 5.1 ____e 6.2 _b___

9.0 a___ 8.2 _b__ 7.5 ___d 7.9 __c_ 7.4 ___d 7.5 ___d

159 a___ 80 _b__ 83 _b__ 75 _bc_ 68 __c_ 46 ___d

2.70 a__ 0.65 _bc 1.48 _b_ 1.12 _bc 0.68 _bc 0.36 __c

0.52 a_ 0.49 a_ 0.24 _b 0.23 _b 0.20 _b 0.11 _b

0.005 0.009 0.007 0.011 0.010 0.014

0.02 0.04 0.04 0.03 0.03 0.04

147 a__ 101,620 a_ 40 __c 95,700 a_ 73 _bc 66,830 ab 104 _b_ 58,580 _b 80 _bc 101,580 a_ 176 a__ 96,380 a_

Secchi

DO

(cm)

(mg/l)

ANCOVA models Model significance r2

⁎⁎⁎ 0.90

⁎⁎⁎ 0.72

⁎⁎⁎ 0.99

Variability source Treatment Month Treatment ⁎ month

ns ⁎⁎⁎ ns

Sign ⁎⁎ ⁎⁎⁎ ns

37 a__ 21 __c 25 _bc 24 _bc 27 _b_ 23 __c

%SS 12 77 11

pH

%SS 1 97 2

%SS 28 34 38

Sign ns ⁎⁎ ns

%SS 6 48 46

ns ⁎⁎⁎ ns

6700 5200 7600 5700

18 25 24 18

7625 a__ 6750 ab_ 3125 _bc 2290 __c 6830 ab_ 9625 a__

48 a_ 16 _b 15 _b 13 _b 12 _b 26 _b

A. Milstein et al. / Aquaculture 286 (2009) 246–253

Alkalinity

Temp. (°C)

r2 = coefficient of determination. Sign = significance levels: ⁎ = 0.05, ⁎⁎ = 0.01, ⁎⁎⁎ = 0.001, ns = not significant. %SS = percentage of total sums of squares. Mean multicomparisons: same letters in each column indicate no significant differences at the 0.05 level. a N bN …. (n) = number of observations.

247

248

A. Milstein et al. / Aquaculture 286 (2009) 246–253

2. Methods The experiment was conducted in twelve 140–160 m2 ponds of 1.5 m deep at the Fisheries Field Laboratory, Bangladesh Agricultural University (BAU), Mymensingh, involving 4 treatments, 3 replicates per treatment. The Control polyculture consisted of the traditional stocking of 33 rohu–33 catla–34 common carp per 100 m2, with the addition of 250 SIS and 3 silver carp per 100 m2, which in the previous experiment was found to be a ‘no-effect’ combination in relation to the traditional stocking combination (Milstein et al., 2008). Intervention in the water column (Water treatment) was achieved by changing the density of the herbivorous fish (reducing the density of catla to 24/100 m2 and increasing that of silver carp to 12/100 m2), and on the bottom (Bottom treatment) by doing so on the benthophagous fish (replacing 10/100 m2 common carp by mrigal). In the Water&Bottom treatment both interventions were carried out. The SIS stocked was mola, which has been found to be very rich in nutrients and vitamin A (Thilsted et al., 1997; Roos et al., 2007). Before starting the experiments ponds were drained to eradicate all predatory fishes, embankments and slopes were repaired, and agricultural lime (CaCO3) at 2.5 kg/100 m2 was applied. Ponds were filled up, and fresh cow manure (6.5 kg/100 m2) was applied in the four corners of each pond to promote algal growth. Fish were stocked on 19-May-07 and cultured for 155 days. At stocking the average weight of rohu and of catla was 40 g, mrigal and silver carp 33 g, common carp 12 g, and the SIS mola was 1.5 g. Fingerlings of the large carps were purchased from the local retailer, who collected them from rural nurseries that obtained the fertilized eggs from the nearby Government hatchery. Fingerlings of mola were collected from perennial ponds of the farmers, where farmers keep them together with major carps and where the small fish naturally breed. Inorganic fertilization was performed at 15-day intervals with urea and tri-super-phosphate at 250 g/100 m2 each. Fresh cow manure (6.5 kg/100 m2) was applied in the four corners of each pond, at 15-day intervals. Handmade supplementary feed consisted of rice bran and soaked oil cake (2:1, both weighed dry, before oil cake was soaked in water), given daily at a rate of 3% of the large carps body weight, excluding silver carp. Fish were weighed monthly with a seine net (mesh size of 5 mm) to adjust feeding amounts. Weights of 10 fishes of each species were measured separately to assess health condition and growth. Partial harvesting of mola started two months after stocking, and total fish harvesting was carried out on 20-Oct-07. All mola biomass removed was recorded. At final harvesting all large fish were individually weighed and counted and mola was bulk weighed. Environmental sampling was performed always at around the same hour (9:00 AM). Parameters: at 15-day intervals data were collected onsite on temperature, dissolved oxygen, and pH using digital meters, and water depth was also recorded. At 30-day intervals transparency (Secchi disk) was measured and samples were collected to determine total alkalinity (acid titration following Stirling, 1985), chlorophyll-a (acetone extraction and spectrophotometer reading at 664 and 750 nm wavelength, following Stirling, 1985), and inorganic nutrients (using HACH kits with direct reading spectrophotometer DR/2010). Total ammonia nitrogen (TAN) was determined with the Nessler method (Hach method 8038, range 0 to 2.50 mg/l NH−3–N, estimated detection limit 0.06 mg/l NH−3–N, precision ± 0.015 mg/l NH−3–N). Nitrite nitrogen was determined with the diazotization method (Hach method 8507, range 0 to 0.30 mg/l NO−2–N, precision ± 0.006 mg/l NO−2–N). Nitrate nitrogen was determined with the cadmium reduction method (Hach method 8192, range 0 to 0.40 mg/l NO−3–N, precision ± 0.010 mg/l NO−3–N). Reactive phosphorus was determined with the ascorbic acid method (Hach method 8048, 3− range 0 to 2.50 mg/l PO3− 4 –P, estimated detection limit 0.01 mg/l PO4 –P, 3− precision ± 0.01 mg/l PO4 –P). Plankton samples were collected monthly with a 25 μm mesh size plankton net. Ten liters of water were taken from five different places and depths of the pond, the filtered material was combined, water added to make 50 ml, and preserved in 5% buffered

formalin. Sub-samples of 1 ml were examined using a Sedge Wick-Rafter counting cell under a binocular microscope (Olympus BH2). Identification up to genus level was performed following Bellinger (1992). Bottom samples (zoobenthos) were collected monthly with an Ekman dredge (225 cm2 samples) from 3 sites in each pond, and the organisms present were determined and counted. Analyses were performed at the Water Quality and Pond Dynamics Laboratory of the Faculty of Fisheries, Bangladesh Agricultural University, Mymensingh. Water quality data were analyzed through multivariate and univariate statistical methods. Factor analysis was used to identify ecological processes responsible for the variability of the water quality parameters measured (Milstein, 1993). Split-plot ANOVA using treatment as main factor and month as sub-factor was applied to each water quality parameter and to the factors extracted. Fish data were analyzed through ANOVA to test the effect of the different treatments on the performance of each fish species. Differences between treatment (and month) levels were tested with the Duncan multicomparison test of means, using a significance level of P b 0.05. Survival (percentage) data were normalized using the arcsine of the square root transformation. Feed conversion ratios were transformed

Table 2 ANOVA and Duncan mean multicomparisons of large fish harvesting parameters Variable

Weight

Biomass

Survivala

Growth

Yield

Unit

g

kg/ 100 m2

%

g/day

kg/100 m2/ 155 days

⁎⁎ 0.78

⁎⁎ 0.84

⁎⁎ 0.77

85 99 98 92

1.20 a_ 0.96 _b 0.83 _b 0.91 _b

5.2 a_ 4.9 a_ 4.2 _b 4.3 _b

ns 0.29

⁎ 0.64

⁎ 0.67

93 100 98 94

0.85 a_ 0.56 _b 0.44 _b 0.70 ab

4.1 a_ 2.1 _b 2.3 _b 2.5 _b

ns 0.42

⁎⁎⁎ 0.87

⁎⁎ 0.82

100 91 98 99

6.42 a__ 2.84 __c 4.56 _b_ 3.18 __c

3.0 _b 4.8 a_ 2.2 _b 5.9 a_

ns 0.45

ns 0.35

ns 0.42

79 92 92 83

0.91 0.95 1.20 1.04

3.8 4.6 4.1 3.2

ns 0.20

ns 0.16

ns 0.13

100 98

2.22 2.03

3.4 3.1

Rohu Significance ⁎⁎⁎ ⁎⁎ 0.86 0.80 r2 Mean multicomparisons by treatment Control 226 a_ 6.3 a_ Water 189 _b 6.2 a_ Bottom 169 _b 5.5 _b Water&Bottom 179 _b 5.4 _b Catla Significance ⁎ ⁎ 0.65 0.69 r2 Mean multicomparisons by treatment Control 172 a_ 5.3 a_ Water 129 _b 3.1 _b Bottom 109 _b 3.7 _b Water&Bottom 148 ab 3.4 _b Silver carp Significance ⁎⁎⁎ ⁎⁎ 0.88 0.83 r2 Mean multicomparisons by treatment Control 1033 a__ 3.1 _b Water 472 __c 5.2 a_ Bottom 739 _b_ 2.3 _b Water&Bottom 525 __c 6.2 a_ Common carp Significance ns ns 0.39 0.46 r2 Mean multicomparisons by treatment Control 152 4.1 Water 158 5.0 Bottom 199 4.3 Water&Bottom 173 4.5 Mrigal Significance ns ns 0.16 0.13 r2 Mean multicomparisons by treatment Bottom 379 3.8 Water&Bottom 349 3.5

_b a_ a_ _b

r2 = coefficient of determination. Significance levels: ⁎⁎⁎ = 0.001, ⁎⁎ = 0.01, ⁎ = 0.05, ns = not significant. Mean multicomparisons: different letters in each column indicate significant differences at the 0.05 level. a Statistical tests based on transformed data. Values of means given untransformed.

A. Milstein et al. / Aquaculture 286 (2009) 246–253 Table 3 ANCOVA and Duncan mean multicomparisons of mola, herbivorous (rohu + catla + silver carp) and bottom feeding fish (common carp and mrigal) harvesting biomass (kg/100 m2) and yield (kg/100 m2/155 days) Mola Herbiv. Bottom Total Herbiv. Bottom Total biomass biomass feeders biomass yield feeders yield biomass yield Significance ns ⁎⁎ 0.26 0.80 r2 Mean multicomparisons by treatment Control 0.80 14.7 a_ Water 0.87 14.5 a_ Bottom 1.47 11.4 _b Water&Bottom 1.39 15.0 a_

⁎⁎ 0.83 4.1 _b 5.0 _b 8.1 a_ 6.9 a_

⁎ 0.67 19.6 _b 20.3 _b 21.0 _b 23.4 a_

⁎⁎ 0.82

⁎⁎ 0.80

12.2 a_ 11.8 a_ 8.7 _b 12.6 a_

3.8 _b 4.6 _b 7.5 a_ 6.4 a_

⁎ 0.65 16.8 _b 17.3 _b 17.6 _b 20.3 a_

r2 = coefficient of determination. Significance levels: ⁎⁎ = 0.01, ⁎ = 0.05, ns = not significant. Mean multicomparisons: different letters in each column indicate significant differences at the 0.05.

to ranks (non-parametric technique appropriate for ratios) before performing further analyses. The analyses were run using the SAS statistical package. 3. Results 3.1. Environmental data The first environmental sampling was performed before the seasonal heavy rains started, on the same day that fish were stocked. Thus, grazing effects still did not occur and sunshine allowed phytoplankton development. Afterwards, rains up to 40 mm/day occurred, with stronger rains (up to 150 mm/day) during the first half of June and the second half of July. Table 1 presents the ANOVA results of the water quality, plankton and benthos parameters. Except for the nitrite and nitrate, the models were significant and month was the only or most important source of variability with patterns as indicated in the “Mean multicomparison by month” section of the table. Treatment significantly affected water

249

Table 4 ANCOVA and Duncan mean multicomparisons of feed conversion ratio (FCR) and income obtained by selling the fish of each treatment at the prices indicated in Table 5, in Tk/100 m2 pond (1 U$ = 70 Tk) FCRa

Herbiv. Herbiv. + mola Bottom feeders Total income income income income

Significance ⁎⁎ ⁎ 0.81 0.73 r2 Mean multicomparisons by treatment Control 2.35 a__ 758 a_ Water 2.08 ab_ 660 a_ Bottom 2.04 _b_ 507 _b Water&Bottom 1.84 __c 684 a_

(+) 0.52

⁎⁎⁎ 0.89

⁎ 0.63

867 a_ 777 ab 705 _b 872 a_

163 _b 199 _b 390 a_ 338 a_

1029 _b 976 _b 1096 ab 1209 a_

r2 = coefficient of determination. Significance levels: ⁎⁎⁎ = 0.001, ⁎⁎ = 0.01, ⁎ =0.05, (+)= 0.1, ns = not significant. Mean multicomparisons: different letters in each column indicate significant differences at the 0.05. a Statistical tests based on transformed data. Values of means given untransformed.

transparency (Secchi) and phytoplankton abundance. Water transparency increased when the intervention was either in the water or on the bottom. Phytoplankton counts decreased by 30% when the intervention was in the water alone, had intermediate values not significantly different from either when the intervention was on the bottom alone, and did not differ from the control when both interventions were performed. Phytoplankton communities were composed of Bacillariophyceae (10 genera, mainly Cyclotella, Melosira and Navicula), Chlorophyceae (33 genera, mainly Scenedesmus and Chlorella), Cyanophyceae (14 genera, mainly Gomphosphaeria) and Euglenophyceae (Euglena and Phacus). In the zooplankton Rotifera (7 genera, mainly Brachionus and Polyarthra) and Crustacea (6 genera, mainly nauplia stages of Copepods) dominated. In the benthos the dominant group was Chironomid insect larvae, followed by Mollusca and Oligochaeta. 3.2. Fish performance Table 2 presents the ANOVA results of the large carp analyses. The survival of the Indian major carps (rohu, catla and mrigal) was at least 85%, of common carp 79% and of silver carp 91%. The intervention in the water column through the partial replacement of catla by silver carp and/or the intervention in the pond bottom through the partial replacement of common carp by mrigal, did not significantly affect harvesting weight and biomass, survival, and growth rate and yield of the bottom feeder common carp. Mrigal performed equally well in ponds with and without intervention in the water column. Rohu was affected by the polyculture compositions. The three interventions tested led to 21% and 26% reductions in its harvesting weight and growth rate respectively. Survival was 16% higher when the intervention was either in the water column or in the bottom, but not when both were performed simultaneously. Harvested biomass and yield were respectively 13% and 19% lower when some common carp were substituted by mrigal, either when intervention in the water column was performed or not. Catla was affected by the polyculture compositions. The three interventions tested led to 36 and 48% reduction in harvested biomass and

Table 5 Price of fish of different sizes in the Bangladeshi rural market, Nov-2007, in Tk/kg (1 U$ = 70 Tk)

Fig. 1. Average total yield and income in each treatment including each species in the polyculture. Yield scale in a per ha basis, income in Tk extrapolated to a 1 ha pond (1 U $ = 70 Tk). Similar letters indicate no significant difference in total yield or income at the 0.05 level.

Fish size

100–250 g

250–500 g

500 g–1.0 kg

1–3 kg

Fish species Rohu Catla Silver carp Common carp Mrigal Mola

35–55 30–45 30–35 30–45 35–50 120–150

55–70 45–60 35–50 45–65 50–65 –

70–85 60–80 50–60 65–90 65–80 –

85–250 80–230 55–180 90–190 80–240 –

250

A. Milstein et al. / Aquaculture 286 (2009) 246–253

yield respectively. Harvesting weight and growth rate were respectively 31% and 43% lower when the intervention was either in the water column or in the bottom, but not when both were performed simultaneously. Silver carp was affected by the polyculture compositions. When silver carp was in higher density due to its replacing of some catla, intraspecific competition occurred and silver carp harvesting weight and growth rate were reduced by about 53%, but its harvested biomass and yield increased by 70% whether intervention in the bottom was also carried out or not. When intervention was only on the pond bottom, silver carp harvesting weight and growth rate were reduced by 28%, but its harvested biomass and yield did not increase. When the large carp are grouped by feeding habits (Table 3), intervention on the bottom reduced herbivorous harvested biomass and yield by 22% and 30% respectively, while doubled bottom feeders biomass and yield. The latter effect occurred whether the water column intervention was also performed or not. Together with this, mola, which reproduced in all ponds, presented large harvesting biomass variability among ponds of the same treatment, so that no significant effect due to the interventions in the water column and bottom was detected. The overall result was that the total harvested biomass and total yield were 17–20% higher when both interventions, in the water column and in the bottom, were practiced. Fig. 1a presents the yields of each species in each treatment (average of 3 replicates). Table 4 presents the ANOVA results of feed conversion ratio (FCR) and the income obtained by selling the fish at the prices of the rural markets

in Bangladesh in November 2007 (Table 5). FCR was 13% lower when intervention was only in the bottom, and 24% lower when intervention was both in the bottom and the water column. Intervention in the water column did not affect income, intervention on the bottom alone reduced income from herbivorous fish by 30% and more than doubled income from bottom feeders, and both interventions performed together did not affect income from herbivorous fish and doubled income from bottom feeders. Thus, total income was 17% higher when both interventions were simultaneously performed than in the Control treatment, as shown in Fig. 1b. 4. Discussion 4.1. Factor analysis of the environmental data and its interpretation Factor analysis was performed on the environmental data (Table 6) to identify ecological processes responsible for the variability of the water quality parameters measured. The first factor (F1) is related to the pond bottom, showing a combination of liming effects, decomposition and fish feeding. The positive correlation among alkalinity, pH, TAN, phosphate and Secchi disk transparency reflects the effect of liming on water quality. Liming increases alkalinity and pH, eliminates plankton that precipitates onto the bottom and thereby increases water transparency. The accumulated organic matter on the bottom (a) decomposes, liberating phosphate and TAN (also positive coefficients)

Table 6 Results of factor analysis, split-plot ANOVA and Duncan mean multicomparisons of environmental data Factors

F1

F2

F3

F4

Secchi Dissolved oxygen pH Alkalinity TAN Nitrate Phosphate Chlorophyll-a Benthos Phytoplankton Zooplankton Variance explained (%) Interpretation

0.66 −0.60 0.84 0.88 0.58 −0.24 0.58 0.20 0.59 0.25 0.12 32 Liming, decomposition, feeding on bottom

− 0.24 − 0.24 − 0.06 − 0.15 0.31 0.09 − 0.50 0.15 0.07 0.78 0.54 13 Phytoplankton absorption of phosphate + zooplankton abundance with phytoplankton

−0.35 0.20 0.24 0.09 0.61 0.49 0.05 −0.45 −0.14 0.19 −0.52 12 Nitrifiersphytoplankton competition for TAN + zooplankton abundance with phytoplankton

−0.03 0.40 −0.03 −0.09 −0.04 0.63 0.26 0.61 0.44 0.01 0.06 11 Oxygen related processes

⁎⁎⁎ 0.93 Sign. ⁎⁎ ⁎⁎⁎ ns

⁎⁎ 0.56 Sign. ns ⁎⁎⁎ ns

⁎⁎⁎ 0.60 Sign. ns ⁎⁎⁎ ns

⁎⁎ 0.52 Sign. ⁎ ⁎⁎⁎ ns

ANOVA models Model significance r2 Variance source Treatment Month Treatment ⁎ month

%SS 1 99 0

%SS 13 63 24

%SS 7 67 26

Mean multicomparisons by treatment (n = 18) Control _b Water _b Bottom a_ Water&Bottom _b

a a a a

a a a a

_b a_ _b a_

Mean multicomparisons by month (n = 12) May a__ Jun _b_ Jul __c Aug _b_ Sep _b_ Oct __c

_bc ab_ __c __c ab_ a__

__cd a___ ab__ _bc_ ___d ___d

_b_ __c ab_ _bc __c a__

%SS 9 83 8

Factor coefficients in bold were used for interpretation. r2 = coefficient of determination. Significance levels: ⁎ = 0.05, ⁎⁎ = 0.01, ⁎⁎⁎ = 0.001, ns = not significant. %SS = percentage of total sums of squares. Mean multicomparisons: same letters in each column indicate no significant differences at the 0.05 level. a N b N ….

A. Milstein et al. / Aquaculture 286 (2009) 246–253 Fig. 2. Conceptual representation of the pond ecosystem functioning in each treatment. Size of fish, community circles and arrows represent importance of effects relative to the Control treatment. Each catla, rohu, common carp and mrigal represent 10–12 fish; each silver carp represents 3 fish; mola adults and fry indicate the presence of the SIS in the pond. Meaning of each type of arrow is indicated in the Control graph. F1 to F4 are the four environmental factors. 251

252

A. Milstein et al. / Aquaculture 286 (2009) 246–253

and consuming oxygen (negative coefficient), and (b) provides food for benthic organisms (positive coefficient). This factor decreased with time, as the impact of liming performed before stocking fish decreased and as fish grew up and consumed organic matter and benthic organisms from the pond bottom. Intervention on the pond bottom through partial replacement of common carp by mrigal significantly increased this factor in relation to the control. The reduced stirring effect on the pond bottom of mrigal as compared to that of common carp (Milstein et al, 2002) allowed more accumulation and decomposition of organic matter, which then liberated more phosphate and TAN into the water column and consumed more oxygen. When the intervention was also in the water column, the larger amount of the highly efficient grazer silver carp should have induced phytoplankton reproduction, that then reduced water transparency, absorbed those nutrients and produced more oxygen so that no F1 differences with the Control treatment occurred. The second factor (F2) is a phytoplankton factor showing its relationships in the food chain: phytoplankton uptakes phosphate (negative correlation between phytoplankton and phosphate) and is preyed upon by zooplankton, whose abundance increased with that of phytoplankton (positive correlation between phyto- and zooplankton). This factor had lower values when water sampling was performed after several days of rain and hence low amount of sunshine hours (in May, July and August) that decreased photosynthesis. This factor was not affected by the polyculture composition. The third factor (F3) is a water column factor showing (1) competition for TAN between nitrifying bacteria and phytoplankton, and (2) zooplankton dependence on phytoplankton. Nitrification can only occur when TAN is present (positive correlation between TAN and nitrate), while nutrient absorption promotes phytoplankton development (negative correlation between chlorophyll and nitrogenous nutrients) that promotes zooplankton development (positive correlation between chlorophyll and zooplankton). The low factor value on the first sampling day occurred because nitrification was still not established at the beginning of the culture season, phyto- and zooplankton were rather high since fish grazing still did not start (the environmental sampling was performed several days after pond filling and before fish stocking) and there was enough light for photosynthesis since rains (=covered sky, low sunshine) also did not start. One month later, by the time of the second sampling, nitrification was fully established while fish grazing and rains reduced phyto- and zooplankton, resulting in a strong F3 increase. From then on, phytoplankton steadily increased (F3 decreased) since algae absorb TAN faster than nitrifiers (Hargreaves, 1998), and zooplankton with it. No differences among the different polyculture combinations occurred. The fourth factor (F4) showed positive correlation among nitrate, chlorophyll, benthos and dissolved oxygen, indicating processes producing oxygen and occurring in the presence of oxygen. Photosynthesis increases chlorophyll and oxygen in the water column, nitrification can only occur in aerobic conditions, and benthic organisms require oxygen for respiration. This factor changed with month; with higher values when water temperature was cooler hence oxygen saturation point was higher. Significantly higher F4 values occurred when intervention was in the water column, either if bottom intervention was also performed or not. The stronger grazing pressure due to the larger amount of silver carp induced reproduction of phytoplankton, hence photosynthesis, and the resulting improvement of oxygen conditions favored nitrification and organisms on the pond bottom. The larger amount of silver carp fecal pellets might have also increased food availability on the pond bottom for benthos and benthophagous fish. 4.2. Pond ecosystem functioning in each polyculture tested Interfering in the water column and/or in the pond bottom through the polyculture composition produced complex responses in the pond ecosystem affecting the large carps' performances, while it did not significantly affect the reproduction and the harvested

biomass of the small fish mola. Fig. 2 is a conceptual representation of the pond ecosystem functioning in each polyculture tested, based on the significant interactions observed. The basic trophic relations among the species and the four environmental factors are presented for the Control treatment. All filter feeders feed on plankton and suspended particles (including added feeds) with low electivity capacity (Dewan et al, 1991), but catla prefers zooplankton while rohu, mola and silver carp prefer phytoplankton (Jhingran and Pullin, 1985; Rahman, 1989). Particles originating in all ecosystem components gradually settle onto the bottom, providing organic matter for the decomposing bacteria. Common carp feeds on benthos, and searching for food stirs sediments (Tang, 1970). Stirring of sediments by benthivorous fishes has two effects: (1) it increases diffusion rates across the sediment–water interface (Hohener and Gachter, 1994), and (2) it increases aerobic decomposition by aerating anaerobic sediments (Graneli, 1979; Beristain, 2005). These two effects enhance the TAN and phosphorous flux from the sediments to the water column (Hargreaves, 1998) and consume oxygen, which support our decomposition part of the F1 interpretation. Phosphorous is absorbed by phytoplankton (F2), TAN by nitrifiers and phytoplankton that compete for it (F3) and both nutrients affect phytoplankton, which in turn determines zooplankton abundance. Grazing of phytoplankton by the herbivorous fish induces reproduction of phytoplankton (drawn only for silver carp), hence photosynthesis, and the resulting improvement of oxygen conditions favors nitrification and organisms on the pond bottom (F4). In the Water column intervention catla density was reduced and silver carp density was increased keeping the total fish density, which increased grazing pressure on phytoplankton because silver carp is a faster growing and efficient filter-feeder (Milstein, 1992). The increased grazing pressure improved oxygen regime in the pond by inducing reproduction of phytoplankton (F4), reduced the amount of particles in the water column increasing water transparency (Secchi disk), and led to increased inter- and intraspecific competition among the three herbivorous large fish species (including silver carp itself) that attained smaller size and lower growth rates than in the Control treatment. Catla was more affected by competition with silver carp for food than by catla intraspecific competition, since the expected effect of reduced intraspecific competition due to reduced catla density would be the opposite: increased catla harvesting weight and growth rate and decreased harvesting biomass and yield. The Bottom intervention is characterized by a weaker flow of particles and nutrients from pond sediments into the water column, since mrigal produces a weaker disruption on the pond bottom than common carp when feeding (Milstein et al., 2002). This mrigal weaker action together with a reduction of the amount of common carp resulted in a decrease of particles resuspended into the water column that then had higher transparency (deeper Secchi disk visibility) than in the Control treatment. Those particles remained on the bottom enhancing benthos development and also decomposed liberating nutrients to the water column (increased F1). Since the herbivorous fish composition was the same than in the Control treatment, the predation pressure by the filter feeders was also similar, so that no extra induction of phytoplankton reproduction occurred that could absorb more nutrients and improve oxygen regime (F2, F3 and F4 did not change in relation to the Control). The reduced availability of particles in the water column resulted in a reduction of food availability for the filter feeder large carps that attained lower size and growth rates. This effect on rohu and catla was of a similar magnitude to that produced by silver carp competition in the Water treatment, while for silver carp it had a lower effect than intraspecific competition in the Water treatment. In the Water&Bottom intervention the higher grazing pressure by the increased amount of silver carp induced phytoplankton

A. Milstein et al. / Aquaculture 286 (2009) 246–253

reproduction that improved oxygen regime (F4) and absorbed the nutrients liberated by decomposition, so that F1, F2 and F3 did not change in relation to the Control. The increased grazing pressure compensated phytoplankton increase due to its reproduction, so that phytoplankton biomass and water transparency also did not change. The increased grazing pressure and the reduced particle resuspension by the decreased number of common carp and the presence of mrigal led to increased inter- and intraspecific competition between the large herbivorous species, which attained lower size and growth rates. However, the increased phytoplankton reproduction provided enough food for the reduced amount of catla, which with lower intraspecific competition grew as in the Control treatment. 4.3. Socio-economic considerations The four polycultures tested allowed a good production of large carp species as cash-crop, of silver carp as an option to consume or to sell, and of the small species mola as food for the farmers' families. Mola production was not significantly different among the polycultures tested. Mola price more than doubled their prices in 2005, because of people awareness about the nutrient and vitamin content of this species and at the same time the scarcity of SIS in natural waters. However, SIS production as food for the farmer's family and not as a selling product is still meaningful since the partial harvesting of SIS is very random and the amounts range from 50 g to 350 g, so that farmers mostly opt to consume it during the culture season, and only sell some SIS at total harvest time, when amounts may be over 500 g. Silver carp is cheaper than the other large carps of the same fish size range (Table 5), but its high growth rate results in larger silver carps that are sold at least at the same price than the other large carps of smaller size. This increases total yield and income in relation to polycultures without silver carp (Milstein et al., 2008). Besides, the smaller silver carp are affordable by poor families, which can then increase their protein intake with a large fish. In conclusion, the Control polyculture is appropriate to produce relatively large herbivorous species, mainly silver carp. The polyculture combination in the Water treatment is appropriate to obtain a larger amount of smaller silver carp that can be afforded by the poor people but also smaller rohu and catla, while maintaining the same level of total yield and income with reduced FCR than in the Control treatment. The polyculture combination in the Bottom treatment allowed a larger fish species diversity and also produced smaller herbivorous fish with still reduced FCR, while maintaining the same level of total yield and income than the Control treatment. The polyculture combination in the Water&Bottom treatment gave the best results: it allowed a larger fish species diversity, is appropriate to obtain a larger amount of small silver carp that can be afforded by the poor people, and gives the highest total yield and income with the lowest FCR. Acknowledgements This research was supported under Grant No. TA-MOU-03-C23-022 U.S.–Israel Cooperative Development Research Program, Economic Growth, U.S. Agency for International Development (USAID). The Director of BAURES, Bangladesh Agricultural University (BAU), Mymensingh, is acknowledged for his cooperation during this study. This work could not have been done without the assistance of the staff

253

of the Fisheries Field Laboratory and Water Quality and Pond Dynamics Laboratory, BAU, Mymensingh Bangladesh.

References Alim, M.A., Wahab, M.A., Milstein, A., 2004. Effects of adding different proportions of the small fish punti (Puntius sophore) and mola (Amblypharyngodon mola) to a polyculture of large carp. Aquaculture Research 35, 124–133. Alim, M.A., Wahab, M.A., Milstein, A., 2005. Effects of increasing by 20% the stocking density of large carps of the basic “cash” carp — small fish polyculture of Bangladesh. Aquaculture Research 36, 317–325. Bellinger, E.G., 1992. A Key to Common Algae. The Institute of Water and Environmental Management, London. Beristain, B. T., 2005. Organic matter decomposition in simulated aquaculture ponds. PhD thesis, Fish Culture and Fisheries Group, Department of Animal Science, Wageningen University, The Netherlands. 138 pp. Dewan, D., Wahab, M.A., Beveridge, M.C.M., Rahman, M.H., Sarkar, B.K., 1991. Food selection, electivity and dietary overlap among planktivorous Chinese and Indian major carp fry and fingerlings grown in extensively managed, rain-fed ponds in Bangladesh. Aquaculture Research 22, 277–294. Graneli, W., 1979. The influence of Chironomus plumosus larvae on the oxygen uptake of sediment. Archiv für Hydrobiologie 87, 385–403. Hargreaves, J.A., 1998. Nitrogen biochemistry of aquaculture ponds. Aquaculture 166, 181–212. Hohener, P., Gachter, R., 1994. Nitrogen cycling across the sediment–water interface in the eutrophic, artificially oxygenated lake. Aquatic Science 56, 115–132. Jhingran, V.G., Pullin, R.S.V., 1985. A hatchery manual for the common, Chinese and Indian major carp. ADB & ICLARM Publication. ICLARM Contribution 252. 191 pp. Kadir, A., Kundu, R.S., Milstein, A., Wahab, M.A., 2006. Effects of silver carp and small indigenous species on pond ecology and carp polycultures in Bangladesh. Aquaculture 261, 1065–1076. Kadir, A., Wahab, M.A., Milstein, A., Hossain, M.A., Seraji, M.T.I., 2007. Effects of silver carp and the small indigenous fish mola Amblypharyngodon mola and punti Puntius sophore on fish polyculture production. Aquaculture 273 (4), 520–531. Milstein, A., 1992. Ecological aspects of fish species interactions in polyculture ponds. Hydrobiologia 231, 177–186. Milstein, A., 1993. Factor and canonical correlation analyses: basic concepts, data requirements and recommended procedures. In: Prein, M., Hulata, G., Pauly, D. (Eds.), Multivariate Methods in Aquaculture Research: Case Studies of Tilapias in Experimental and Commercial Systems. ICLARM Studies and Reviews, vol. 20, pp. 24–31. Milstein, A., Hepher, B., Teltsch, B., 1985a. Interactions between fish species and the ecological conditions in mono- and polyculture pond system. I. Phytoplankton. Aquaculture and Fisheries Management 16, 305–317. Milstein, A., Hepher, B., Teltsch, B., 1985b. Interactions between fish species and the ecological conditions in mono- and polyculture pond system II. Zooplankton. Aquaculture and Fisheries Management 16, 319–330. Milstein, A., Wahab, M.A., Rahman, M.M., 2002. Environmental effects of common carp Cyprinus carpio (L.) and mrigal Cirrhinus mrigala (Hamilton) as bottom feeders in major Indian carp polycultures. Aquaculture Research 33, 1103–1117. Milstein, A., Kadir, A., Wahab, M.A., 2008. The effects of partially substituting Indian carps or adding silver carp on polycultures including small indigenous fish species (SIS). Aquaculture 279, 92–98. Rahman, A.K.A., 1989. Freshwater Fishes of Bangladesh. Zoological Society of Bangladesh, Dhaka. 364 pp. Roos, N., Wahab, M.A., Hossain, M.A.R., Thilsted, S.H., 2007. Linking human nutrition and fisheries: incorporating micronutrient-dense, small indigenous fish species in carp polyculture production in Bangladesh. Food and Nutrition Bulletin 28 (2), 280–292. Stirling, H.P., 1985. Chemical and Biological Methods of Water Analysis for Aquaculturists. Institute of Aquaculture, University of Stirling, Scotland, UK. Tang, Y.A., 1970. Evaluation of balance between fishes and available fish foods in multispecies fish culture ponds in Taiwan. Transactions of the American Fisheries Society 99, 708–718. Thilsted, S.H., Roos, N., Hasan, N., 1997. The role of small indigenous fish species in food and nutrition security in Bangladesh. NAGA News letter, July–December (Supplement) 13 pp. Wahab, M.A., Rahman, M.M., Milstein, A., 2002. The effect of common carp Cyprinus carpio (L.) and mrigal Cirrhinus mrigala (Hamilton) as bottom feeders in major Indian carp polycultures. Aquaculture Research 33, 547–557. Wahab, M.A., Alim, M.A., Milstein, A., 2003. Effects of adding the small fish punti (Puntius sophore) and/or mola (Amblypharyngodon mola) to a polyculture of large carp. Aquaculture Research 34 (2), 149–164.