Two feeding trials were carried out to determine the effect of feeding Azolla microphylla as partial replacement of the protein in boiled soya bean in diets based on sugar cane juice for meat ducks. The experimental design for the trials involved five treatments, four replications (pens) and 10 birds per pen. Diets, containing five levels of crude protein, which were formulated from boiled soya beans and Azolla microphylla and with sugar cane juice ad libitum as the energy source, were fed to 200 growing ducks (Cherry Valley hybrids). The rations were fed from one month of age until the birds reached 60 to 70 days of age, when two ducks (one male and one female) per pen were slaughtered for carcass evaluation.

In the first trial, actual intakes of Azolla were lower than expected, and supplied 15.2, 17.9, 24.4, and 30.3% of the total protein in diets Az15, Az30, Az45, and Az60, respectively. Daily gains (g) were 28.6, 30.4, 28.3, 24.1, and 21.4 for the control, Az15, Az30, Az45, and Az60 diets, respectively and decreased with level of azolla when this exceeded 15% of the dietary protein (P<0.001). Feed conversion ratios deteriorated with increasing consumption of Azolla (P<0.001).

In the second trial, actual intakes of Azolla were also lower than expected, and supplied 14.7, 17.5, 19.2, and 22.6% of the total protein in diets Az15, Az25, Az35, and Az45, respectively. Gains (g/day) were 32.6, 31.5, 30.2, 28.2, and 24.1 for the control, Az15, Az25, Az35, and Az45 diets, with decrease in gain with azolla level beyond15% of the protein (P<0.001). Feed conversion ratios increased with increasing consumption of Azolla (P<0.001). All the carcass traits were similar for all treatments in both trials.

Aquatic plant species, because of their growth habit, appear not to accumulate secondary plant compounds and therefore offer a greater potential than tree leaves as a source of protein for monogastric animals. Of these species the water fern Azolla, which grows in association with the blue-green alga Anabaena azollae, is perhaps the most promising from the point of view of ease of cultivation, productivity and nutritive value (Lumpkin and Plucknett 1982; Van Hove and López 1983).

The concept of using aquatic plants for different purposes is now receiving special attention. Azolla is one of the plant resources with high biomass and protein production. Little information is available on maximum levels of inclusion in the diets of ducks and for this reason two trials were carried out in which different amounts of whole boiled soy ad libitum. In the first trial intakes of boiled soya bean were adjusted weekly according to mean pen liveweight, while in the second trial the daily a bean were given, and Azolla and sugar cane juice supplied allowance of boiled soybeans was fixed throughout the trial.

Azolla is an aquatic fern genus, with six species distributed widely throughout warm-temperate and tropical zones. Its most remarkable feature is its symbiotic relationship with the nitrogen-fixing blue-green alga Anabaena azollae within its leaf cavities (Reynaud and Franche 1987). Azolla has been used for centuries in Asia as green manure, as a fertilizer for rice fields and supplement in diets for pigs, poultry and fish (Moore 1969; Cassani 1981; Lumpkin 1984). Some strains of Azolla can fix as much as 2-3 kg of nitrogen/ha/day. Azolla contains around 27% crude protein and 10% carbohydrates on a dry weight basis (Pullin and Almazan 1983).

Azolla is widely distributed and is found in ponds, ditches and channels containing stagnant water with a temperature range of 15- 350 Celcius (Singh and Subudhy 1978). It grows in aquatic habitats and absorbs nutrients mainly from water. In shallow water the plant roots attach to the soil and absorb nutrients from the soil.

Poultry, particularly ducks and chickens, can be raised on a diet including fresh Azolla. It has long been recognized as a feed for wildfowl in the USA and for domesticated ducks in China. It is reported to be fed to domestic fowl in Vietnam (Vo Minh Kha and Tran Quang Thuyet 1970).

In India, Subudhi and Singh (1978) concluded that fresh Azolla could replace about 20% of commercial feed in the diet of young chickens. They estimated that to replace this much commercial feed would require about 9 kg of fresh Azolla each day for 100 chickens and that this amount could be produced in a shallow pond 60 m² in area.

The experimental design, housing and management were identical in both trials. The experiments were conducted between March and June 1993, in the dry season. Two hundred 4 week old ducklings with initial liveweights of between 700 and 750 g were used in each experiment. The birds were identified and then individually weighed initially and at slaughter, and then by group once per week throughout the experiment. They were allocated to five treatments with four replications per treatment, and 10 birds per group. The initial total body weight of the ducks in each treatment was approximately the same. The experimental period was preceded by a preliminary period of one week.

The ducks were housed in a shed divided into pens constructed from bamboo, with thatched roofs and rice straw covered packed earth floors, with an average density of five ducks/m². During the daytime the birds also had access to outside sand yards (one duck/m²), and for approximately six hours per day to a fresh water swimming area (five ducks/per m²).

Feeders and drinkers were round plastic basins (40 cm diameter, 10 cm deep). Natural light only was used in the day time, and electric bulbs (4 w/m² of floor area) at night from hatching until the end of the experiment. The mean diurnal temperature during the experiment was around 30°C.

The ducklings were fed a commercial type starter diet between 15 and 32 days of age, and then were allocated at random to one of the experimental diets. The birds were group-fed different amounts of soya beans adjusted according to liveweight. The experimental diets were all based on reconstituted sugar cane juice offered ad libitum as the major energy source, with varying amounts of Azolla and whole boiled soya beans as the protein source. Total crude protein requirements were estimated from NRC recommendations (NRC 1984) and pilot trials were carried out to determine feed intakes for increasing liveweights. Daily boiled soya bean allowances were calculated so that Azolla offered ad libitum would supply approximately 0 (control), 15, 30, 45 or 60% of the daily crude protein intake (Table 1). A vitamin-mineral premix (0.5% of the diet) and common salt (0.25% of the diet) were mixed with the whole boiled soybeans.

Table 1: Composition of the diets
	N x 6.25 from soya	Fresh Azolla	Cane juice
Control	100	-	ad lib.
Az15	85	ad lib.	ad lib.
Az30	70	ad lib.	ad lib.
Az45	55	ad lib.	ad lib.
Az60	40	ad lib.	ad lib.

The dry soya beans were weighed out daily and boiled each morning for half an hour in order to destroy the antinutritional factors in the grain. After boiling the grain was put into a bamboo basket to drain and cool before feeding.

Boiled soya beans were offered twice a day, in the morning and afternoon, in amounts according to the treatment and according to liveweight. Daily amounts of whole boiled soya beans started at 12.5 g crude protein/day at 1.0 kg body weight, increasing weekly to a maximum of 25 g crude protein/day at 2.0 kg. The weight of soya beans offered was adjusted every week according to the mean pen live weight for that week.

Azolla was harvested each afternoon during the experimental period and was put into a bamboo basket for one hour before feeding in order to drain the excess water from the plant. The fresh Azolla was offered ad libitum three, four or five times per day, at a rate of 1 kg fresh weight per pen at each feeding. The times increased with the age of the birds to minimize losses. Refusals were collected and weighed the following morning.

Due to technical problems concerned with the purchase and storage of sugar cane, reconstituted raw sugar solutions were used. Raw sugar was dissolved in cold water in a proportion of 2.2 kg raw sugar to 5 kg water, approximately, and then was mixed and measured as Bohmer degrees, to give a value of 11 Bohmer, equal to 20°Brix (total dissolved sugars). This was offered to the ducks in plastic basins covered with a grille to ensure that there was minimum spillage of the juice. The reconstituted sugar cane juice was offered ad libitum three, four or five times per day, with 800 ml/pen in each feeding, the times increasing with the age of the ducks to ensure that there was minimum loss. Juice remaining in the basins each morning was collected and weighed to calculate amounts consumed.

Sample preparation and analyses of the dietary ingredients for dry matter (DM), crude protein (CP), crude fibre (CF), ether extract (EE), ash, calcium, phosphorus and carotene were carried out at Cantho University using standard procedures.

At the end of the experiment, the ducks were weighed individually and two ducks per pen (one male and one female) were selected at random for carcass analysis. The birds were slaughtered at 72 days of age, eviscerated and plucked, and the carcass weight recorded.

The data were analysed by analysis of variance using the GLM (General Linear Model) procedure of MINITAB version 8.2 program statistical software (Minitab Inc., 3081 Enterprise drive, State College, PA, 1680-3008, 1991, USA). The balanced design of the trial permitted the use of GLM with initial live weight as a covariate. A similar method was used to analyse the carcass traits.

An economic analysis was carried out to compare the cost and profitability of the dietary treatments.

The chemical composition data of the whole boiled soya beans, Azolla and sugar cane juice are shown in Table 2. The Azolla had a mean dry matter content of 5.6%, crude protein 26.7%, crude fibre 11%, ether extract 4.6%, Ca 0.8% and total P 0.4% (DM basis).

Table 2: Chemical composition of the whole boiled soya beans, Azolla and sugar cane juice
Analysed composition			Sugar cane juice
	Soya Beans	Azolla
Dry matter	91.8	5.6	20.0
As % of DM
N x 6.25	38.8	26.7	1
Crude fiber	8.4	11.2	--
Ether extract	20.0	4.6	--
Ash	5.1	15.1	--
Ca	0.4	0.8	--
P (total)	0.5	0.4	--
Carotene (mg/kg)	--	326	--

The raw sugar used in the experiment was collected from a small sugar refining factory in the area, and is the syrup resulting from the evaporation of sugar cane juice. The sample used in this experiment had the following chemical composition: (% as received basis): ?Brix 70; dry matter 70%; crude protein 1 %; ash 4.9%; sucrose 58.1%; reducing sugars (as glucose) 8.5%.

Total dry matter intake was increased by Azolla when offer level of soya bean protein was only slightly less than on the control diet (diet Az15). With increased restriction of soya bean, dry matter intake decreased linearly (r= 0.81) according to protein consumed as soya bean (Table 3).

Table 3: Effect of replacing boiled soya beans with Azolla in sugar cane juice based diets on feed intake and performance of growing ducks (Trial 1).
Item	Ctl	Az15	Az30	Az45	Az60	SE/Prob
No birds	40	40	40	40	40
Liveweight, g
Initial	717	735	698	741	697
Final	1861	1940	1828	1706	1564
Daily gain	28.6	30.4	28.3	24.1	21.4	±0.63/0.01
Feed intake, g/d
Soya bean	55.7	52.8	42.7	34.2	25.4
Azolla	--	233	234	279	301	±4.48/0.01
Cane juice	400	442	422	433	426	±8.89/0.05
Total DM	131	149	136	134	127	±1.9/0.001
Total CP	20.6	23.0	19.5	17.2	14.5	±0.15/0.01
CP from Azolla:
g/day	--	3.5	3.5	4.2	4.4	±0.06/0.01
% of total CP	--	15.2	17.9	24.4	30.3
Feed conversion,
kg DM/kg gain	4.59	4.86	4.77	5.57	5.95	±0.18/0.01

The decrease in total dry matter intake appeared to be caused by the inability of the ducks to eat more azolla when soya bean was restricted beyond 73% of the control level. Intakes of Azolla were lower than expected, increasing from a mean of 233 g/day on the Az15 diet to only 301 g/day on the Az60 diet. The proportion of the crude protein intake supplied by Azolla was only 15.2, 17.9, 24.4, and 30.3% of the total, for diets Az15, Az30, Az45 and Az60, respectively. But more significant was the decrease in total protein intake which fell from 23 g/day on diet Az15 to only 14.5 g/day on diet Az60. As there is no evidence of palatability problems with Azolla it is possible that the bulky nature of the plant was the reason for the lower than expected intakes. Mean intakes of cane juice were 400, 442, 421, 432, and 426 ml/day for the control, Az15, Az30, Az45 and Az60 treatments, respectively.

Growth rate was linearly related to protein intake both total and that derived from the soya beans; however, the correlation was stronger with total protein (r=0.98) compared with soya bean protein (r=0.93). When soya bean protein was restricted to up to 73% of that in the control diet the ducks compensated by eating azolla to achieve higher (treatment Az15) or the same total intake of protein (Az30). However, when soya bean intakes were restricted to less than 60% of the control level, the ducks were not able to eat enough azolla to compensate for the deficiency in protein and growth rate fell. Thus it can be concluded that azolla can effectively replace up to 17% of the normal level of soya bean protein with no reduction in grrowth, and that low levels of Azolla in addition to the normal level of soya bean actually stimulate growth (Treatment Az15). In contrast with the results for growth, feed conversion ratios deteriorated with offer level of Azolla and were related directly and negatively with soya bean protein level (r= -0.94).

Carcass parameters (Table 4) were not affected by dietary treatment (P>0.05).

Table 4: Effects of replacing whole soya bean with Azolla in diets on carcass parameters of ducks (Trial 1).
Item	Control	Az15	Az30	Az45	Az60	SE/prob
Liveweight, g	1911	1915	1845	1737	1608
Carcass
Weight, g	1384	1268	1324	1253	1117	±70/0.65
Yield, % LW	72	67	72	72	69	±2.7/0.79
Heart, g	11.6	13.6	12.3	13.7	13.5	±1.4/0.317
Liver, g	44.6	48.2	45.4	48.6	48.3	±3.2/0.99
Gizzard, g	51.9	51.3	50.8	44.7	44.7	±3.5/0.44
Breast, g	278	276	271	264	251	±18.1/0.71
Breast
muscle, g	172	168	156	151	131	±15.2/0.66
Breast fat+
skin, g	51.2	58.0	58.5	58.0	62.3	±6.8/0.56
Breast
bone, g	49.5	51.7	56.2	58.3	47.7	±4.5/0.24
Carcass
measurements, cm
Length	22.7	22.3	22.6	23.4	22.2	±0.6/0.232
Small int. length	205	191	214	204	210	±8.86/0.99
Large int. length	12.0	11.4	12.9	12.0	12.6	±0.77/0.39

Optimum economic results (Table 5) were obtained with an offer level of protein of 23 g/day made up of 73% from boiled soya beans and 17% from Azolla.

Table 5: Economic analysis of replacing soyabeans with Azolla, Vietnamese Dong (1US$ = 10.500 VND) (Trial 1).
Item	Control	Az15	Az30	Az45	Az60
Daily intakes, g
Soya bean	55.7	52.8	42.7	34.2	25.4
Fresh Azolla	--	233	234	279	301
Cane juice	400	442	422	433	426
Premix, salt	0.20	0.20	0.16	0.12	0.09
Costs of the diet/day:
Boiled soybean	184	174	141	113	83.8
Fresh Azolla		23.3	23.4	27.9	30.1
Cane juice	72.4	80.5	76.3	78.3	77.1
Premix, salt	3.8	3.6	2.9	2.3	1.7
Total cost, VND
Per day	260	2826	244	221	193
Trial period	10400	11264	9740	8852	7708
Cost of feed/ kg
gain	9091	9345	8619	9173	8890
Profit/bird	-104.9	-417	433	-167	95

In the first trial the protein supply was adjusted weekly and increased with the age and liveweight of the ducks. In this trial the daily amounts of boiled soya beans offered were the same throughout the experiment and the proportions of crude protein contributed by Azolla were correspondingly lower than in trial 1. This is the system used by small scale farmers in Vietnam.

The experimental design, housing, feeding and management procedures were identical to those followed in trial 1.

The experiment was carried out between April and June, 1993. A total of 200 F1 Cherry Valley hybrids with initial average live weight between 900 and 970 g were used. The ducklings were fed a commercial type starter diet between 1 and 28 days of age, and then were allocated at random to one of the experimental diets. The initial total body weight of the ducks in each treatment was approximately the same. The birds were identified and individually weighed at the beginning (28 days) and also at slaughter. Group weights were taken once per week throughout the experiment. The experimental period was preceded by a preliminary period of one week. The ducks were group fed with different amounts of soya bean and had free access to Azolla. All the ducks were offered reconstituted sugar cane juice ad libitum as the main source of energy.

Total crude protein requirements were estimated from NRC recommendations (NRC 1984) and pilot trials to determine feed intakes for increasing liveweights. Daily boiled soya bean allowances were restricted to provide 100, 85, 75, 65 and 55% of the recommended levels of protein. A vitamin-mineral premix (0.5% of the diet) and common salt (0.25% of the diet) were mixed with the whole boiled soya bean only for the control group. Details of the feeding regime are in Table 6.

Table 6: Proportions of ingredients (dry matter basis) in the diets of fattening ducks (28-60 days) (Trial 2)
	Protein from
	Soya	Azolla	Cane juice
Control	100	--	ad lib.
Az15	85	ad lib.	ad lib.
Az25	75	ad lib.	ad lib.
Az35	65	ad lib.	a d lib.
Az45	55	ad lib.	ad lib.

The chemical composition of the whole boiled soya beans was the same as in Trial 1 (Table 1). The raw syrup used in this experiment had the following chemical composition (% as received basis): ?Brix 72.3; dry matter 72%; crude protein 1 %; ash 4.5%; sucrose 56.2%; reducing sugars (as glucose) 13%.

As in Trial 1, intakes of Azolla were lower than expected, increasing from 257 g/day on the Az15 diet to only 284 g/day on the Az45 diet. Total protein intake thus decreased from 25.5 and 25.9 on control and Az15 diets to 18.6 g/day on the Az45 diet. As in Trial 1, growth rate was better correlated with total protein intake (r=0.97) than with intake of soya bean protein (r=0.94) indicating a benefit from the Azolla supplementation at the low level of substitution (Table 7).

Table 7: Effect of replacing boiled soybeans with Azolla in sugar cane juice based diets on feed intake and performance of growing ducks (Trial 2).
Item	Control	Az15	Az25	Az35	Az45	SE/prob
Liveweight, g
Initial	966	910	924	943	919
Final	2046	1951	1918	1873	1710
Daily gain	32.6	31.5	30.2	28.2	24.1	±1.1/0.001
Feed intake, g/day
Soya bean	70.0	60.0	52.5	45.5	38.5
Azolla	--	257	274	268	284	±4.45/0.01
Cane juice	279	342	347	340	338	±3.35/0.01
Total DM	120	137	133	125	119	±0.72/0.01
Total CP	25.5	25.9	23.4	20.8	18.6	±0.06/0.01
CP as Azolla,%	--	14.7	17.5	19.2	22.6
Feed conversion,
kg DM/kg gain	3.7	4.3	4.4	4.4	4.9	±0.16/0.01

A supplementary source of vitamins and minerals was not included in the diets of the groups that received Azolla and yet there were no obvious health problems or deficiency symptoms indicating that the Azolla was supplying these essential nutrients. This apparent capacity of Azolla to supply vitamins and minerals is an important advantage in rural areas where premixes may not be available or are expensive. However, it should be mentioned that the ducks were given commercial starter diets, which contained vitamins and minerals, up to one month of age, and some of these would have been stored in the body.

There were no effects of treatment on carcass characteristics (Table 8).

Table 8: Effects of replacing whole soybeans with Azolla in the diets of fattening ducks on carcass parameters (Trial 2).
Item	Control	Az15	Az25	Az35	Az45	SE/prob
Liveweight, g	1901	1786	1768	1858	1641
Carcass Weight, g	1339	1287	1262	1326	1179	±73.7/0.54
Yield, % LW	70	72	71	71	72	±1.6/0.33
Heart, g	11.4	12.8	10.8	12.5	9.9	±1.0/0.15
Liver, g	43.6	40.2	37.4	40.9	37.6	±3.3/0.39
Gizzard, g	55.4	53.7	53.2	53.5	48.7	±4.9/0.65
Breast weight, g	273	251	254	251	232	±20.8/0.06
Breast muscle, g	170	143	148	149	137	±15.6/0.23
Breast fat+ skin, g	48.3	55.3	57.1	49.5	39.1	±8.16/0.26
Breast bone, g	41.7	46.6	45.3	48.5	51.0	±5.56/0.20
Carcass
measurements, cm
Length	22.6	22.6	23.0	24.4	22.5	±0.72/0.75
Small int. length	204	196	202	217	207	±7.66/0.56
Large int length	12.9	13.2	12.6	13.5	11.3	±0.71/0.70

The deeper yellow color of the legs, beaks and skin of the ducks fed with Azolla compared to the control group was noted and substantiated the observations made in Trial 1.

The economic analysis is presented in Table 9. Daily feed costs were less with the lower levels of soya bean but the reduction in rate of gain on these diets cancelled out this advantage.

Table 9: Economic analysis (VND) of replacing soyabeans with Azolla in the diet of fatteningducks (1US$ = 10,500 VND)
Item	Control	Az15	Az25	Az35	Az45
Daily intakes, g
Soya bean	70.0	60.0	52.5	45.5	38.5
Fresh Azolla	-	257	2749	268	284
Cane juice	279	342	347	340	338
Premix, salt	0.26	-	-	-	-
Cost of the diet
(VND/day):
Soya bean	231	198	173	150	127
Fresh Azolla		25.7	27.4	26.7	28.3
Cane juice	50.5	61.9	62.7	61.6	61.2
Premix, salt	4.8	-	-	-	-
Total cost (VND)
Per day	286	286	263	239	217
Trial period	9448	9425	8693	7871	7149
Cost feed/kg gain	8755	9057	8743	8459	9023
Profit/bird	265	-58.8	256	503	-18.0

A low digestibility of the protein in Azolla could be one explanation of the poorer growth rates when a high level of substitution of soya beans was attempted. However, in a digestibility trial with broiler chicken, Alcantara and Querubin (1985) concluded that nutrient digestibilities of crude protein, crude fat, and crude fibre were not affected by the level of Azolla in the ration, and that broilers can readily digest the crude fibre in Azolla, but not that in rice bran, so that digestibility may not have been a limiting factor.

The other important characteristic of a feed protein for ducks is its balance of essential amino acids. These were not analyzed in either of the trials but all indications from the literature are that most strains of Azolla strains have a well balanced array of essential amino acids and constitute high quality protein if cultivated in good conditions (Van Hove 1989).

There are few references to feeding trials that have been carried out with Azolla for poultry. In an experiment in India, White Leghorn females were fed a commercial poultry feed and fresh Azolla at levels of 5, 12.5, or 16% on a dry matter basis. The birds receiving the diet with 5% Azolla grew faster than the control group and those given the diet with 12.5% Azolla grew only slightly slower, although at 16% inclusion growth rates were significantly reduced (Singh and Subudhi 1978).

A series of feeding trials was conducted by Escobin (1987) to determine the effects of feeding fresh Azolla as partial replacement to traditional paddy rice-snail-shrimp (PSS) based rations for laying Mallard and growing Muscovy ducks. In terms of production efficiency, the results showed no significant differences between the dietary treatments containing 0, 20, 30 and 40% Azolla replacing corresponding levels of PSS, both for Mallard (egg production) and Muscovy (meat production) ducks.

From the results of these trials it can be concluded that fresh Azolla can partially replace whole soya beans up to a level of about 20% of the total crude protein in diets of fattening ducks based on sugar cane juice, without any problems and with no adverse effects on growth rate, or health. Cost of feed per kg gain was lowest, and net profit per bird highest for this treatment. However, at levels of replacement above this, rates of gain and feed conversion efficiency were significantly poorer.

Grateful acknowledgement is made to the Swedish International Development Authority (SIDA) for the financial and material support of this project. Facilities for the experiments were provided by the University of Cantho through the cooperation of Mr Bui Xuan Men and help was received from Ms Kim Dong and Mr Sanh in field work and from Ms Xuan Dung in the laboratory.

(Received 1 June 1995)