The hypotheses that were evaluated in this study were that: effluent from biodigesters would be an effective source of nutrients with which to grow duck weed of high protein content; and the protein level in the duck weed would be a function of the amount of effluent added to the pond water.

Six ponds were used each lined with polyethylene film (0.2 mm thickness) having 9.4 m² area and 15 cm water depth. The two treatments were 32 and 4.5 kg effluent/m³ pond water which were estimated to support N concentrations in pond water of 73 and 10 mg N/litre respectively. The effluent was from plastic tube continuous flow biodigesters, charged with pig manure and contained 6.5% of solids and 3.41% of N in the solids. The pond was inoculated with 200 g of duck weed/m² and the total biomass was harvested at 3 day intervals over a period of 9 days. The yield of duck weed was calculated by substracting the amount of seed material in the inoculum.

Biomass yield and protein (N x 6.25) in the duck weed dry matter were linearly related with the N concentration in the pond water and negatively related with root length of the duck weed. Optimum levels of N in pond water were in the range 40-60 mg N/litre. Duck weed of more than 35% protein had a root length less than 10 mm. Duck weed biomass yield at optimum pond N levels was of the order of 100 g fresh biomass/m²/day with 6% dry matter and 30-40% protein (N x 6.25) in the dry matter.

Based on these results, and other observations, the recommended management practices to optimise duck weed yield and protein content are: (i) an initial charge of 33 kg of effluent containing 6.5% solids and 3.4% N in the solids (or comparable amounts to provide 75g N)/m³ of pond) and; (ii) 1.0kg effluent/day/m³ for 27 days, after which the ponds should be emptied and recharged with effluent and "good quality" (root length of less than 10mm) duck weed.

Where rainfall is adequate or there is supplementary irrigation, water plants can be highly productive sources of protein-rich biomass and are ideal complements for fibre-free basal diets such as molasses, sugar cane juice, sugar palm juice and palm oil in pig and poultry feeding systems. The leaves of most water plants are more digestible than the leaves from trees and, generally, they appear to have low concentrations of anti-nutritional factors (Preston 1995).

The water fern Azolla spp is one of the aquatic plants that has had good impact with farmers because of its high grow rate, relatively simple management and easy incorporation in feeding systems for pigs and ducks (Becerra 1991). Observations in Colombia (Rodriguez Lylian and Cuellar Piedad 1994) showed that Azolla can produce an average of 60 g fresh biomass/day all year round although there was a tendency for it to grow better during the rainy season. It supplied 100% of the protein requirements of lactating sows in diets based on sugar cane juice without decrease in their performance.

Another potential resource is duck weed (family lemnaceae). Lemna or duck weed as it is commonly known, is a tiny green plant that grows on pond surfaces. It is easily identified by the presence of only one root per small green frond. This plant grows well in different climates, and is a fast growing, high protein plant that can efficiently absorb nitrogen and phosphorus as well as heavy metals (Logsdon 1989, cited by Becerra et al 1995). A World Bank project in Bangladesh showed that fish yields of an average of 10 tonnes/year/ha were obtained using only duck weed as a supplement to the naturally available fish feed (Skillicorn et al 1993).

Becerra et al (1995) concluded that fresh duck weed a source of protein can only be used in limited amounts to substitute for conventional protein in the diets of fattening ducks. There were no adverse affects on health, but decreases in growth rate and in feed conversion efficiency were considerable when duck weed replaced more than 20% of the soya bean protein. In contrast, Bui Xuan Men et al (1995) showed that fresh duck weed could completely replace roasted soya beans and a vitamin-mineral premix in broken rice based diets for fattening ducks without reduction in growth performance or carcass traits. A major difference between the two experiments was that Becerra et al (1995) worked with duck weed of only 26% protein whereas Bui Xuan Men et al (1995) used duck weed growing in water which had been fertilized with biodigester effluent and contained 38% protein in the dry matter.

The fact that protein yields of duck weed can be as high as 10 tonnes/ha/year (Preston 1995) compared with less than one tonne per year for soya bean protein highlights the potential value of this plant at the farm level. The critical factor appears to be the protein content, which in turn depends on the nutrient status of the medium on which it is grown. This relationship was shown by Stambolie and Leng (1995) when they analysed duck weed (Spirodela spp) grown on sewage water. The protein content rose from 20% to almost 40% in dry matter as the N content of the water was increased from 5 to 40 mg/litre.

The hypotheses to be evaluated in this study were:

* Effluent from biogas digesters would be an effective source of nutrients with which to grow duck weed of high protein content.

* The protein level in the duck weed would be a function of the amount of effluent added to the pond water.

The experiment was done at the "Finca Ecologica" on the University Campus, a small farm established by one of the authors (TRP) to demonstrate integrated farming systems with perennial crops, multi-purpose trees, local breeds of livestock, low-cost plastic biodigesters and duck weed ponds.The area is close to sea level with ranges in temperature from 24 to 38 ?C, and relative humidity in the range 40 to 100%.

Plastic lined ponds were used because of the sandy soil and the need for uniformity among the ponds. They were built using the same sort of polyethylene plastic used for the biogas digesters (0.2 mm thickness). There were 6 ponds of 9.4 m² area and 15 cm water depth. The ponds were located close to the biogas digester to facilitate the use of the effluent. Soil was excavated to about 15 cm and the plastic sheet supported with bamboo poles to gave a total pond depth of 25 cm.

The mixed effluent from 4 experimental biodigesters was used. Each measured 4 m long and 0.84 m in diameter with 1.5 cubic metres liquid volume. They were charged with pig manure and water to give a solids concentration of 2, 4, 6 and 8% for individual digesters. The loading rate was 50 kg/day of the mixtures.

Six ponds were used. The two treatments (each replicated in 3 ponds) were 32 and 4.5 kg effluent/m³ pond water. An inoculum of 200 g of duck weed/m² was added to each pond and the total biomass was harvested at 3 day intervals. The yield of duck weed was calculated by substracting from the total biomass the amount of inoculum. After each harvest the same amount of inoculum was put back again. Samples of water and duck weed were collected at each harvest for determination of nitrogen and dry matter and for measurements of root length.

The yields of duck weed were determined by weighing the total amount of biomass from each pond. Bamboo sticks, the width of the ponds, were used to harvest the duck weed. To measure the length of the roots, samples of duck weed were suspended in plastic containers 8 cm diameter filled with water. Root length was measured by extending 10 individual plants from each sample on millimetric paper and taking the average length.

Dry matter was determined by weighing before and after drying for 48 hr in an oven at 100 ?C and nitrogen by Kjeldahl (AOAC 1988) using a Tekator apparatus. The samples of duck weed were analysed immediately after harvest because when the samples were stored in a refrigerator there were obvious physical changes in their appearance and it was considered this might affect the results.

Samples of water for N determination were taken immediately after harvesting the duck weed. The pond contents were agitated and samples were taken from different points in each pond. These were mixed and the N determination done on the fresh samples using the Kjeltec apparatus.

The effluent added to the ponds contained 6.5±0.65% dry matter (mean and SE) and 3.41±0.13% N in the dry matter. Relationships among the amounts of effluent added to the ponds, the N content of the pond water, the biomass yield, the N x 6.25 content of the duck weed and the root length are shown in Figures 1 to 6. The basic data are in Annex 1.

At the first harvest, fresh duck weed yield was in the range 55 to 65 g/m²/day and did not differ between effluent levels (Figure 1). By the second harvest the yield on the low level of effluent had fallen to 31 g/m²/day while in the ponds with the high effluent level yield had increased to 110 g/m²/day. At the third harvest yield on the low effluent level decreased further to 15 g while at the high level it was maintained at 110 g/m²/day. The interaction between effluent level and harvest number was significant (P=0.01). Trends in N concentrations in the pond water (Figure 2) were similar to those for biomass yield (P=0.13 for the interaction of N concentration with harvest number) except that at the low level of effluent the N concentrations were low and not affected by number of harvests (12.8 to 15.9 mgN/litre) while at the high effluent level, N concentration peaked at the second harvest (53 mgN/litre) and fell subsequently to the value at the first harvest (31-38mgN/litre).

Biomass yield (Y1) and protein (N x 6.25) in the duck weed dry matter (Y2) were linearly related with the N concentration in the pond water (X).

Y1 = 26.0 + 1.41±0.51X; r² = 0.32 (Figure 3)

Y2 = 13.3 + 0.55±0.13X; r² = 0.54 (Figure 4)

Biomass yield (Y3), protein content of the duck weed dry matter (Y4) and N content in pond water (Y5) were negatively related with root length (Z) of the duck weed.

Y3 = 133 - 3.71±0.94Z; r² = 0.49 (Figure 5)

Y4 = 56.3 - 1.49±0.15Z; r² = 0.86 (Figure 6)

Y5 = 56.1 - 1.55±0.36Z; r² = 0.53 (Figure 7)

Conclusions

There was considerable variability in the data but the overall relationships supported the original hypothesis that the protein (N x 6.25) content of duck weed can be manipulated by the addition of biodigester effluent.

On the basis of the analysis of the effluent and an average fresh biomass yield of duck weed of 100 g/m²/day it can be estimated that adding 45 kg fresh effluent (contains 105 g N) to 1.4 1m³ (9.5 m² area) of pond water will raise the N content of the pond water to 50 mg/litre which agrees well fairly with the observed values (Figure 2). An average harvest of 93 g/m²/day of fresh duck weed amounts to an output from the pond of 0.31 g N/day or 2.8 g N in 9 days. Thus after 3 harvests the pond should be recharged with 12 kg of effluent (containing 6.5% solids and 3.41% N in the solids) in order to maintain a constant concentration of 50 mgN/litre. In practice the following management has been applied to the ponds:

(i) An initial charge of 45 kg of effluent

(ii) After 3 harvests (9 days) an additional 24 kg effluent is applied

(iii) After 6 harvests (18 days) an additional 24 kg effluent

(iv) After 9 harvests (27 days) the ponds are emptied, a new charge of 32 kg effluent is introduced and the duck weed seed is replaced using material from another pond which appears to be of "good quality" (defined as having a root length less than 10 mm; see Figure 6).

Theoretically, these amounts of effluent are higher than the absolute requirements. This implies either that not all the N in the effluent is available to the duck weed or there are losses (eg: of un-ionised NH₃) from the pond surface. This is an area requiring further research.

Appreciation is expressed to the Swedish Agency for Research Cooperation with Developing Countries (SAREC) for financing most of this study. The University of Agriculture and Forestry of Ho Chi Minh City made available the site for the Finca Ecologica on the University campus. Special thanks to Nguyen van Lai for advice on analytical procedures and assistance in the laboratory, to Vo Than Hai for help with the day to day management of the ponds and to Bui Xuan An for advice on practical problems of biodigester and pond construction and maintenance.

Skillicorn P, Spira W and Journey W 1993 Duckweed aquaculture - A new aquatic farming system for developing countries.The World Bank: Washington pp76

(Received 20 May 1996)

Level N	Harvest	g/m2/day	Roots	Wat mg N/l	Lemna%DM	Lem%N DM	Lem%Pr
1	1	48.6	8.5	41.7	6.0	6.6	41.6
2	1	1	21.4	16.3	3.6	5.7	35.4
1	1	69.8	9.7	42.9	5.6	7.0	43.5
2	1	54.3	21	7.4	6.8	3.2	20.0
1	1	70.6	10.5	28.0	5.8	6.4	40.2
2	1	56.3	21.3	15.1	6.9	3.3	20.7
1	2	82.9	11.9	63.4	5.5	6.5	40.9
2	2	19.0	33.2	19.2	7.4	2.0	12.7
1	2	124.4	13	49.4	6.4	5.6	34.8
2	2	40.0	28.5	9.3	6.1	3.3	20.4
1	2	122.4	11.4	47.1	6.9	7.1	44.3
2	2	33.2	25.4	19.1	6.1	2.3	14.3
1	3	56.7	12.6	35.8	6.2	6.3	39.2
2	3	18.5	24.8	10.9	7.8	2.2	14.0
1	3	137.0	14.6	38.0	5.9	5.1	32.2
2	3	16.7	25.4	6.8	8.1	2.3	14.2
1	3	137.3	12.4	19.8	6.1	5.9	36.9
2	3	10.4	29.5	20.6	8.7	1.3	8.0