Appendix 5

DETAILS OF THE WORK CARRIED OUT DURING THE CONSULTANCY

1. AIYURA WATER SUPPLY DATA

2. FISH DISTRIBUTION AND DATA ANALYSIS

All available data concerning fish supplies to farmers' ponds was looked at. This covered the period December 1983 to July 1985, when supplies ceased. Ponds were considered private unless clearly institutional or group-owned.

There was no data concerning growth in private farmers' ponds. There is one report of 8 fish, total weight 5.5 kg, being harvested at Bundaira Corrective Institution from a pond of unknown size (possibly 300 m²) one year after stocking at 1.5 g each. There is another account of a 0.3 ha secondary sewage treatment pond in which 75 fingerlings of 5 cm length all survived to reach average weights of 0.6 kg six months later, a yield equivalent to 300 kg/ha/year.

In fact the dimensions given for the ponds are not very reliable as they are based on owners' estimates. Quite a number were said to be 2, 3 or even 5 m deep. The actual existence of some of the ponds is doubtful. Mr. P. Sagom was unable to find some in nearby villages for which fish had already been supplied: He suggested that villagers wanted rather to stock their local rivers

3. BROODSTOCK GROWTH RATES

Female brood fish were separated from males (except during spawning attempts) and held under good conditions. This means they were either kept at very low density (about 1 fish/100 m²) and fed broiler feed via a demand feeder about 2–3 percent bodyweight daily or at moderate densities (1 fish/5–10 m²) and fed via a demand feeder 3–6 percent bodyweight daily or more. This may have been more feed than necessary, the intention being to ripen the broodstock as fast as possible and learn something about their maximum growth rates. At less than about 500 kg/ha no feeding should be needed, especially if the pond is manured regularly. At 1 000 kg/ha feeding at 2–3 percent bodyweight should be plenty. Excessive feeding risks causing a lot of fat to be deposited, which may actually interfere with spawning, particularly if starchy food is used.

Four females were lost during this period, two via a partial dike collapse and two disappearing from a small pond possibly due to predators, or being stolen. Despite this there was considerable biomass increase, and in some cases certain fish could be identified giving individual growth rates. Results are shown below.

21–22.11.85			20.1.86			5.2.86
description	wt	(kg)	description	wt	(kg)	description	wt(kg)
gold	1.1	→	gold	1.7	→	gold	1.85
gold	1.0	→	gold,black specks	1.6	→	gold, black specks	1.75
gold	1.0		brown	1.35	→	brown	1.56
long gold	0.8		gold	1.25	→	gold	1.38
brown, small eyes	0.7		brown	1.2	→	brown	1.26
brown	0.7		gold	1.2	→	gold	1.25
gold	0.7		red-gold	1.14		gold	1.21
gold with lump	0.7		yellow, small eyes	1.13	→	yellow, small eyes	1.20
brown	0.64		gold	1.1		long red-gold	1.19
brown	0.6		gold with lump	1.0		gold	1.15
yellow, small eyes	0.55		gold	1.0		brown	1.09
silver	0.52		brown	0.92		gold	1.08
brown	0.52		gold	0.92		long red-gold	1.08
gold	0.5		brown	0.86		gold	1.06
long gold	0.5		gold	0.81		brown	1.00
gold	0.5		gold	0.8		long red-gold	0.98
silver-grey	0.5		long dirty gold	0.74		yellow	0.86
gold	0.44		long gold	0.71		red-gold, specks	0.84
long gold	0.4		gold	0.7		long gold	0.78
brown	0.4
gold	0.4		total	20.1		total	22.57
gold	0.4		average (19 fish)	1.06		average (19 fish)	1.19
gold	0.38
long dirty gold	0.35
total	14.3
average (24 fish)	0.6		plus			plus
			1 fish unweighed			1 fish unweighed
			new recruit, yellow	0.2	→	yellow	0.27

4. SPAWNING TRIALS AT AIYURA

From the outset difficulty was expected in spawning the carp at Aiyura. There were only limited numbers of rather small females which had been held unfed in one pond together with males and fingerlings.

On 21–22.11.85 all 24 females were checked, and it was felt that one or two might be ready or nearly ready for breeding. They were placed in a separate 2 000 m² pond (pond 2) and feeding started using broiler crumble in a perforated-bag demand feeder which they learned to use within a few days. Feeding was more-or-less ad libitum, over 6 percent of bodyweight per day.

On 5.12.85 all were checked again and the four most promising were placed in the 60 m² spawning pond in the corner of pond 3. This had been recently repaired and filled. Next day six males were put into the spawning pond, with added pine branches. No eggs were seen the following day, but the day after that a few hundred eggs were seen to have been produced (probably on the first night) but already to have been covered with fungus. The following night there was a partial collapse of the area of the dike containing the outlet and overflow pipe, with the loss of two females and three males. The remaining fish were replaced in their ripening ponds.

On 17.12.85 all females were again checked. Thirteen unripe fish were put into pond 3 (also about 2 000 m²). Eight somewhat-promising fish went into the spawning pond and the most promising fish (which had a distinct asymmetry or lump on one side but also appeared ripe) went into the newly-dug 50 m² holding pond A (see map 2) in which were placed three males. By this time a small amount of Arenga pinnata fibre had been obtained. The fibre was made into kakabans (Fig. 5) by being clamped between strips of wood or bamboo to make 180 × 40 cm egg catchers. Casuarina branches, long-needled pine branches and two kakabans were arranged as egg substrate. No spawning occurred on the following 3 nights. On 20.12.85 the pond was emptied. The females, the 3 males and the spawning substrates were all transferred to the spawning pond in the hope of provoking breeding of the eight somewhat-promising fish. Nothing happened during the following three nights. On 23.12.85 the 3 males and the female were taken to the hatchery. The males were placed in a running water tank of 600 1 at about 19°C. The female was put into a 2 800 1 outdoor static water tank covered with clear plastic which was at about 22–26°C. On 27.12.85 the female was examined and found to have spawned while alone 2–3 days previously.

The fish in pond 3 and the spawning pond continued to be fed (at a somwehat lower rate) using demand feeders. On 2.1.86 the spawning pond was lowered. Two fish were missing. One 600 g female was transferred to the hatchery and placed in a 600 1 tank standing inside a larger covered tank as a water bath to stabilise the temperature. The following day the female received about 0.2 mg dried carp pituitary at 2.45 p.m. and about 2 mg at 10.30 p.m., when males also received half doses. Overnight temperatures were about 24°C. At 10 a.m. the next day a few eggs could be obtained, at 11.10 she was stripped. The eggs came out easily enough but not well separated and with quite a lot of blood and other tissue. They were fertilized with sperm from 2 males, washed using Woynarovich and Woynarovich's modified method described by Rothbard (1981) and incubation attempted in 3 containers. These consisted of 2 sheet-metal conical incubators, one connected directly to the main water supply and one connected via an improvised constant-head tank. There was also a conical net incubator connected directly to the water supply and immersed in a tank. Only the last was capable of keeping the eggs in motion. There was no hatching.

On 14.1.86 the following arrangements were made. A hapa was placed in holding pond A with 1 female and 2 males. Three females and 5 males were placed directly in A. Another hapa was placed in pond 1 with 3 males and 2 females (one of only 200.g which had been found by one of the labourers while sorting fish from pond 4 to have easily-expressed eggs). A hapa with 1 female and 2 males was placed in the spawning pond, and free in the spawning pond were the remaining 14 females plus 3 males. Substrates consisted of the kakabans plus casuarina in pond 1, casuarina and pine in the remaining sites. Spawning occurred only in the hapa in pond 1, and was estimated by eye as having produced 10 000–15 000 eggs, of which most were good. Hatching started on 17.1.86 and substrates were removed the next day. On 20.1.86 the larvae were stocked in pond 3, but amounted to only 2 500 in number.

Guessing that the superior clarity of the water in pond 1 may have been an important factor, the female and 2 males from the hapa in holding pond A were transferred to another hapa in pond 1 on 15.1.86. All the free fish were removed from holding pond A which was then refilled and left for several days to see whether the water clarity would improve (it didn't). The best of these females were taken to the 2 800 1 aerated static water covered tank outside the hatchery with 2 males and pine substrate. Other females went into the spawning pond (where there were still 3 males). There was no spawning at any site. On 17.1.86 females in the tank by the hatchery, the hapa in the spawning pond and the hapa in pond 1 were all injected in the morning (8 – 9 a.m.) with approximate 0.3 mg/kg preparatory doses of carp pituitary. Between 3 – 4.30 p.m. females received second doses of about 3 – 4 mg/kg and one male of each site received around 1.5 mg/kg. Casuarina was replaced by kunai grass (Imperata cylindrica) but pine remained. Next morning the fish in the two hapas had failed to spawn. On checking at 10.30 a.m. the female in the hapa in the spawning pond dribbled eggs. She could perhaps have been stripped there and then on to wet substrate and milt washed over the eggs. Even better might have been to try the use of diluted urine as a substitute for weak solution (Appendix 3, section 3.2.3). Unfortunately these ideas only came later. In the event the female and one male companion were put into the hapa in pond 1 with the fish which had also failed to spawn after injection, to give them one more night together. They again failed to spawn.

Numbers of eggs were seen on both substrates in the hatchery tank, estimated as 70 percent good, with a total of 2 – 5 000. Incubation took place in the same tank with aeration and low flow of water (about 4 1/min). Due to the small egg numbers seen it was decided to prepare only the 60 m² holding pond A for fry rearing and to transfer all females to pond 2. Hatching started on 21.1.86; the next day the substrates were removed and a little hard-boiled eggyolk fed in the evening. On 23.1.86 the fry were siphoned out of the tank into a cloth lined fish trough, transferred to plastic bags and taken down to the ponds. The total number turned out to be 27 000. There was no choice but to stock the bulk of them with the 2 500 fry in pond 3, which were 3 days older.

On 4.2.86 the females were transferred from pond 2 to hapas in the spawning pond and examined. Two females (of about 1.2 kg and 0.85 kg respectively) were put into a hapa in pond 1, in which were put 4 kakabans (a large quantity of Arenga pinnata fibre having been obtained) and 5 small males. Both females spawned that night. Egg numbers were estiimated by eye as about 10 000 of which about three quarters appeared to be good and fertilized (i.e., they were clear and yellowish the following day). One female (1 150 g) and 3 males were put into a covered, aerated, stagnant water tank of about 2 800 1 standing outside the hatchery, also with 4 kakabans. They failed to spawn on the following two nights. Therefore on 6.2.86 the female received about 0.3 mg carp hypophysis at 7.15 a.m. and about 4 mg at 2.45 p.m., when two of the males were also injected with about 1.2 mg hypophysis each. There was no spawning in the following two nights so the fish were returned to the ponds.

In the hapa in pond 1, hatching started on 7.2.86. Kakabans were removed on 9.2.86 and a suspension of one whole egg (prepared by method A described by Rothbard, 1982) was given. This was repeated the following morning. (when hard-boiled egg-yolk is dispersed into water through cloth discrete particles can be seen in the water, and larvae clearly eat them. However, the whole-egg suspension method produces particles which are extremely tiny, and fish were not seen to have ingested them. Thus despite comments in ADCP (1983) hard-boiled egg-yolk appears to be better.) That afternoon larvae were collected from the hapa and distributed into ponds 2, B, C and D. A total of about 12 800 larvae were produced from this spawning.

Tentative conclusions therefore are as follows. Fecundity seems to be quite low, but may improve. A convenient and predictable routine system still has to be worked out. Brood females increased in average size from 600 g to about 1 200 g in 2½ months. If they continue to be held at low density separate from males, spawning should become much easier in the future. It may still be necessary to resort to injections to induce spawning if nothing happens after one or two nights under good conditions, but stripping and artificial fertilization are not envisaged as being carried out routinely. It seems that the water turbidity in any ponds with fish or with muddy pond floors inhibits spawning. Possible solutions might be one of the following. Keep a grassy-floored pond just for spawning (Dubisch style). This would have to be left empty for the rest of the time, since it would otherwise become muddy. Spawning could then be carried out inside it, or in a hapa placed in it. Arrange concrete or plastic tanks for spawning by the hatchery (outdoors, static and covered with clear plastic for warmth). This has the convenience of mains electricity but the considerable inconvenience of distance from the ponds. Bring down or build tanks by the ponds, perhaps with a small accumulator-powered aquarium aerator or two. This last is recommended; as well-designed tanks would have a variety of uses. One design is shown in Figs 2 and 3 and described in section 4.8. Kunai grass and pine seemed better than casuarina as a substrate, but kakabans are more convenient than either. The site now has about a dozen kakabans. The fibre should last for years, though the wood or bamboo may need periodic replacement.

Some of the larger fingerlings in pond 4 should be considered as candidate broodstock. If the best of these were stocked in one large fertilized pond at low density (not more than 1 fish per 20 m²) and well fed, the site should have plenty of broodstock within a few months. Males and females (and fish of as yet indeterminate sex) could be included in this group, with the likely bonus of fry production by uncontrolled breeding. Pond 1 would be well sutied for this use.

5. FRY REARING

Fry were reared in two ponds (pond 3 and holding pond A) for about 3 weeks. At the end of this period, those in pond 3 were almost ready for harvesting and transfer to a fingerling pond, averaging about 0.5 g each. Those in pond A grew more slowly due to the higher density of stocking. Four ponds (ponds 2, B, C, D) were stocked with larvae just before the consultant left.

Pond 3 was drained on 13.1.86 in order to catch and check the female broodstock. Some work was done on repairing the banks and filling started 2 days later. The following day, 200 kg of cow manure collected from nearby fields were mixed with water and distributed in the pond. Seven kilograms of NPK fertilizer (12, 12, 17) were placed on a small try immersed near the water inlet. On 17.1.86 another 13 kg NPK were added. By this time the water depth averaged about 50 cm. The next day 1 litre of Dipterex was poured into the pond water at various places around the edge. On 20.1.86, 2 500 larvae from the natural spawning in the hapa in pond 1 were stocked in pond 3. Three days later, about another 21 300 larvae from the induced spawning in the hatchery tank were distributed (total stocking density 11.9 larvae/m²). Small numbers of rotifers were first seen in the plankton that day. Quite a number of back-swimmers (notonectids) were also seen. Feeding was started using 0.5 kg ground broiler feed distributed as a slurry in the morning and afternoon around the pond edges. To save time spent in grinding and sifting feed, this was altered next day to a mixture of half chicken feed and half wheat flour amounting to a total of 1 kg/day. On the morning of 25.1.86, about 2.25 1 of kerosine were spread over the water surface, mainly around the edges, by mixing the kerosine with sand and scattering the mixture. On 27.1.86 the feeding rate was increased to 1.5 kg daily of ground and sifted broiler feed plus what flour (half and half). About 20 kg of 2-month-old compost (cow manure plus grass) was also mixed with water and dispersed. On 28.1 kerosine (about 1.75 1) was again spread in the calm of the morning. On 31.1.86 pond 3 received an additional 30 kg compost, 30 kg cow manure, 2.5 kg NPK and 1 kg urea. The morning feed was increased to 1 kg (but the afternoon feed inadvertently forgotten). Thereafter the pond received 2 kg of feed per day. Wheat flour was not used from 3.2.86, the whole quantity being made up of ground and sifted broiler feed. From 4.2.86 this was sometimes spread as a dust on the water surface if the weather was calm, rather than as a slurry. On 7.2.86 an additional 50 kg cow manure was given. By 11.2.86, the average fry weight was about 0.5 g. Fry were plentiful and active but were not harvested due to lack of a fingerling pond available for further growing. The size difference between the two age-groups was still perceptible.

The 70 m² holding pond A was refilled after a brief draining on 15.1.86. On 19.1.86 it received a double dose of cow manure (14 kg) and fertilizer (700 g NPK and 300 g urea). On 22.1.86 it received 50 ml Dipterex and the following day about 5 400 larvae (77/m²) from the induced spawning in the hatchery were released. On 24.1.86 feeding of one dispersed egg-yolk in the morning and one in the evening commenced, and the next day a small quantity of fine chicken feed plus wheat flour slurry was also given morning and evening. Around 0.25 1 of kerosine was spread in the morning. The feeding rate was increased to 0.25 kg mixed powders plus one egg yolk, given morning and evening, on 27.1.86. A second kerosine treatment (0.25– 0.5 1) was given the next morning, and rotifers were first seen in the plankton the same day at about 2/ml. On 31.1.86, the morning feed was increased to 0.5 kg and egg feeding ceased, but the afternoon feed was forgotten. Thereafter, the daily ration was 1 kg (0.5 kg in the morning, 0.5 kg in the afternoon). On 3.2.86 flour was left out and the feed from then on consisted only of ground and sifted broiler pellets. From the following day, this was generally given as a dust on the surface rather than as a slurry. The plankton on 6.2.86 was full of what were either rotifers or very large, slow-moving ciliates. The next day an additional 10 kg of compost were dispersed. Fry were plentiful and active, but on 11.2.86 still very much smaller than those in pond 3, estimated at about 0.1–0.2 g average weight.

On 3.2.86, the small holding ponds B (21 m²), C (49 m²) and D (45 m²) were fertilized. They had recently been emptied but had refilled by seepage, etc. Considering “standard” fertilization as consisting of 100 g/m² cow manure plus 10 g/m² of NPK (12, 12, 17), B received a single dose (210 g NPK and 2.1 kg cow manure), C received a triple dose (1 470 g NPK and 14.7 kg manure) and D a six-times normal treatment (2 700 g NPK and 27 kg manure). On 9.2.86 B received about 10 ml Dipterex and the other two ponds around 30 ml each. All were stocked the following afternoon with larvae from the second natural spawning in a hapa in pond 1, at a rate of 35–38 larvae/m² (i.e., B received some 740, C 1 850 and D 1 660).

Pond 2 was emptied on 4.2.86 to enable female broodstock to be checked, and refilling started the same day. On 7.2.86, it received 300 kg cow manure and 1.5 kg urea. On 9.2.86 it was treated with 1 000 ml Dipterex and the following afternoon some 8 500 larvae (4.3/m²) from the natural spawning in the hapa in pond 1 were distributed.

6. CONSTRUCTION AND USE OF SMALL EXPERIMENTAL PONDS

Three ponds of 10 m² each were dug in a vacant area above pond 4 in January 1986. Digging all three took perhaps six man-days, and outlet and overflow systems about another three, using bamboo pipes. Some mistakes were made. Outlet pipes were not low enough and had to be replaced. Grass was not removed before soil was piled up to form banks. This grassy layer leaks somewhat and limits the depth for static water conditions to 50 cm. In one pond this has been made good by cutting out a V-shaped strip of earth to one spade's depth around the pond at this height and refilling it with good clay. This may be worth doing in the other two ponds when next they are drained. Meanwhile, however, all three ponds are capable of maintaining flowing or static water conditions with a depth of about 50 cm and with complete and easy drainage. A fourth similar pond was constructed in February, and there is room for more. Private farmers' ponds in PNG average less than 100 m² each. Therefore preliminary trials on management of village-style ponds may be carried out in these 10 m² units. However promising results should be re-tested in ponds of 50–200 m².

A small experiment was started, looking at the growth of carp in stagnant water without feeding under moderately high manuring rates. This experiment was designed to give as much data in as short a time as possible and to be somewhat protected against predation or other fish losses.

Each pond was stocked with 10 fish covering a range of sizes and of different colours. Therefore, if all fish are weighed regularly, it is hoped that each will be identifiable. Therefore, each fish should produce a measurable growth rate figure at each weighing; ten figures per pond per growth period instead of just one average. Although all ponds were stocked at the same numerical rate of one fish/m², the biomass in each pond was very different. They were supposed to represent conditions in one particular pond at different stages in the growing process. Thus it is hoped that some information representing what might be a year's growth can be obtained within a few months. At the beginning all ponds are receiving the same rate of manuring. However, if this is seen to be too low for the ponds with the higher biomass of fish (as judged by the growth rates) the operator can increase the rate of manure application or harvest some of the big fish and replace them by small ones, as an intelligent farmer might also decide to do. The main thing of course is to weigh the fish in all the ponds regularly (every 2–4 weeks) and to keep good records of all measurements, harvests, inputs and changes.

The starting rate of manure application is 1 kg of moist cow manure/pond/day, on a 5-day week basis. It is therefore equivalent to about 100–150 kg dry matter/ha/week. This is low by intensive manuring standards discussed in Appendix 4, section 5.3. It was chosen to fall somewhere between what was thought might give maximum production and what it was felt that PNG villagers might be persuaded to try.

Data from the first “growth” period of this experiment is shown below.

In pond El four fish disappeared (and the smallest died during the catching and weighing process). The row “weight of fish which survived” shows the total biomass of those fish which were weighed on both occasions. Into pond 3 a number of small fish had entered by jumping out of a hapa which had been standing in that pond when the experiment was first set up. The excess fish were removed on weighing but unfortunately among them was one of the fish originally weighed. Therefore only 9 fish contributed to the biomass loss figure. However, these difficulties illustrate the value of using identifiable fish of different sizes in small growth trials. Had this experiment been based only on average weight increase or decrease such losses would have prevented any information being obtained. In this case, useful information was obtained.

Fish lost weight during the period, and weight loss increased as the stocking biomass increased. Possible explanations are insufficient manuring (in new ponds with little topsoil) and/or excessive turbidity due to the clay nature of the soil. Unfortunately the ponds were harvested by draining, so any fertility which may have developed in the water was lost. It is suggested that the experiment be continued for one or two periods more of about a month each. The manuring rate should be doubled after weighing if little or no growth has occurred. If growth continues poor despite very high manuring rates, the next thing to try might be a comparison of growth rates using different manures; poulty waste, compost and coffee pulp, perhaps at somewhat lower fish biomass levels.

Future experimental ponds should be built with more sloping banks. This will tend to make them slightly more productive and perhaps a little warmer; it will also result in lower erosion. There is room for at least 6 ponds of 50 m² each near to the holding ponds C and D, all of which can be drained through the pipe passing beneath pond 4 which has just been completed. Care should be taken to remove the topsoil layer, grass and roots from the area where soil is piled up to make banks, as otherwise this grassy layer will leak.

7. USING A LAMP TO CATCH INSECTS FOR FISH FEED

Filewood (1967) states: “Insects are probably the cheapest and most available protein source in the highlands; they can be collected by a night-light hung over the pond or used as an insect-trap”. Similarly Glucksman (1971) says: “Cheap local sources of protein include maggots, ants and ant eggs and insects captured at night with a light”. This has been embodied in advice to farmers. The government's extension booklet “Grow good carp book 2B, Management of pond with supplementary feeding” (1970) shows a man with a simple kerosine-burning wick lamp with the caption, “Insects are good food for carp. Hang a lamp over the pond at night. Insects will fly into the light and fall into the pond”.

A small experiment was carried out to see how effective this advice might be. A kerosene lamp (Dietz D-lite No. 90) with an adjustable wick inside a glass envelope was placed with its base just at water level inside a glass aquarium (69 cm × 38 cm × 29 cm) filled with water to a depth of about 10 cm. This was arranged on an open area of lawn, unobstructed in all directions except for one small bush at one side. There were no electric lights left on in the near vicinity. The lamp was lit at dusk and left burning overnight. In the morning the insects which had fallen into the water were collected using an aquarium handnet and weighed (without drying) on a digital balance. The volume of kerosene burned was measured. This was carried out on three separate nights. One of these nights was wet and had noticeably abundant insect life (as judged by riding a motorcycle during the evening). The results were as follows:

Fuel used over 3 night 1 050 ml, to catch 0.74 g wet weight of insects.

On the basis of this very preliminary trial, the fuel cost of insects as a fish food works out at slightly over 500 Kina/kg (i.e., more than one thousand times the price of manufactured livestock feeds).

8. PELLETS AND FLAKES

Pellets may be a convenient and economical way of presenting feeds under certain circumstances, e.g., when feeding broodstock held at high densities, when using bait rod demand feeders (see Fig. 1) or when feeding fish in tanks. As a simple way to make small quantities of pellets, broiler grower was powdered by passing through a small hand-powered mincing machine (hole diameter about 2 mm). Wheat flour or dried powdered sweet potato were mixed with the powdered feed at two levels to test their effectiveness as binders. Measured quantities of water were added to the dry powders and well mixed in. Then the resulting pastes were pelleted through the same mincing machine. Pellets were spread out until dry to the touch and then tested for cohesion by soaking in cups of water with occasional swirling. Visual estimates were made of the percentage of the material in each cup which retained its pelleted from after different periods of immersion. The results are tabulated below (quantities in grams).

Clearly wheat flour slightly improved the quality of the pellets while sweet potato slightly lowered it. However, pellets which consisted just of broiler feed, without any binder, seemed very satisfactory. As wheat flour is dearer than the chicken feed and has a lower protein level, its use increases the cost and lowers the value of the feed, and is therefore not recommended.

Flakes are useful for feeding fish in aquaria and for feeding small fish in tanks. As they float on the surface of the water they provide a way of making food available for a long time without the use of a feeder and they encourage the early feeding of those small fish which do not like to feed off the floor. Flakes were produced by grinding broiler starter in the handpowered mincing machine and seiving the resulting powder through a fine cloth (pore size about 1/4 mm). The fine powder was then mixed with one quarter its weight of wheat flour. Water was added to form a wet paste. This was spread by hand as thinly as possible onto polythene sheeting and then left to dry (initially indoors, later in a small solar drier). The resulting flakes were rather variable in thickness but had excellent cohesion, lasting for many hours in water. It would be worth experimenting with differing amounts of water in the paste, and with lower levels of wheat flour.

Residue from the seiving appeared to consist largely of maize particles. It was kept and used for feeding fish being held at low densities (below about 500 kg/ha), which benefit most from additions of energy to their diet but receive adequate protein from natural food.

9. COLLECTION OF MULLETS

Mullets were collected from the Labu Batu area opposite the wharf at Lae, on 29 – 30 January 1986. Considerable help was received from Marco Sappu who works at Unitech in the Department of Chemical Technology. He made contact with the Kidd family at Gumao village who provided fishing assistance, a place to stay, canoes, their beach seine and a fish cage.

Fish which were caught in the sea were placed in a net cage standing in front of the village. They were placed in a hapa standing in a natural freshwater pond if they came from one of several such ponds or from the Markham river mouth. They were carried in large plastic bags to the wharf where the DPI vehicle waited with an oxygen cylinder. There the bags were filled with oxygen. In both cases the fish were transported in water of the same salinity as they had been held in.

About 300 fish reach Aiyura alive, of which about half appeared to be in reasonable condition. A little malachite green was put into the water in their bags, and they were hung inside the water of pond 1 for the temperatures to equilibrate. Pond water was added gradually to the water in the bags over about 30 minutes until the temperature difference had been reduced from about 4°C to about 1°C and the fish were then allowed to swim out into the pond. They were counted as they did so. By this time the salinity of those bags containing salt water had also dropped (from a taste-estimated 10–15 ppt to perhaps one third of this level). Fish sizes ranged from about 1.6 cm (0.05 g) to over 12 cm (15 g) each. No identification as to species was attempted due to the absence of identification keys, but samples were preserved in 4 percent formalin, for possible future study.

Mullets, although resistant to an extremely wide temperature and salinity range (if the change is not too sudden) very easily lose their scales and become susceptible to infection if handled roughly. A few fry were caught in a smaller 5 m x 1.20 m seine made of fine nylon cloth (pore size about 1 mm × 1 mm). This seemed less likely to damage them than the large beach seine. A factor which made the journey difficult was the uncertain nature of the boat connexion between the village and the Lae wharf. Various delays led to the journey back through the Markham valley being made in the heat of the day. Though shaded, the bags were not well insulated so that water temperatures rose to about 30°C. An unknown number of fish died at Labu Butu (due largely to inexperienced handling) and about 35 percent of those packed were dead on arrival.

If there is interest in mullets in future, the following might help improve their survival. In whatever net they are caught they should be allowed to remain in the water, carefully concentrated in the centre and then lifted out, preferably using an aquarium handnet. A seine of fine netting of 5–10 m length and 1–1½ m depth made of a cloth of pore size 1–2 mm might be very useful, especially about a month or two earlier in the year when smaller fry should be about. Push nets, either square or triangular, operated by one person going around with his own bucket (and a torch if at night) might work out as fast as using one seine net between six people, and a great deal gentler to the fish. (This is how Mugil cephalus of 1.5–2 cm length are caught in Israel.) With good handling, more than 80 percent should survive the journey.

It might be worth putting malachite at a low rate (0.1–0.2 ppm) into the water before the bags are filled with oxygen. It might also be worth packing the bags in an insulated box into which ice could be put from time to time to keep the temperature fairly constant. The position of the sea cage in front of the village turned out to have a very variable salinity with changing tides. Further up one of the creeks might have been better, especially if a place could be found where tidal currents are not so strong. Cages or seines for larger fingerlings would be better made out of knotless nylon which causes less fish damage.

It should be possible on a well-planned trip to catch and bring home a few thousand fish at a cost of a few toia each. There must be some thought given in advance as to where they are to be stocked. Care must be taken to minimize temperature shocks, particularly if the fish are also undergoing a big salinity change. Mullets generally do well in manured polyculture ponds where they live largely on detritus (partly-decayed material), algae and zooplankton. They are not known as predators and do not breed in freshwater.

10. TILAPIA BREEDING

A group of small tilapia was found in a tank at the hatchery. They had apparently been caught in the Ramu river at Yonki several months previously. On 27.12.85 they were sexed as far as possible and a group of 9 males, 14 females and 4 unidentified (average weight of whole group 19 g) was placed in a round tank. This was of 285 cm diameter. The floor had been covered with about 200 kg of sand, and it had been filled with water to a depth of about 60 cm. It was outdoors, without flow but covered with clear plastic sheeting, so that temperatures were generally in the range of 24–28°C. Five carp of about 10 g each were also put into the tank, which was fed (once or twice on most days) with small amounts of flaked food made from broiler grower and wheat flour (Appendix 5, section 8).

At the same time an indoor aquarium was set up with an airlift aerator set into a layer of sand on the floor. One male and two females (unweighed) were introduced. It was placed in the sunniest corner of the hatchery and covered with clear plastic sheeting. However temperatures remained low, 18– 24°C.

Within about 10 days a disease developed in the aquarium (ichthyopthyriasis perhaps) and the male and 1 female had died. The second female was bathed in 50 ppt salt solution, left for 10 hours in 10 ppt salt and the aquarium finally refilled so that it remained slightly brackish (about 2 ppt). Small amounts of malachite green were also added occasionally. The female survived and another male was introduced later but no breeding took place.

On 20.1.86 the outdoor tank was drained and the mouths of all females checked for fry or eggs. Four were found to have bred and their mouth contents were washed out. 190 free-swimming fry were obtained and an uncounted number (about 200) of newly-hatched or just-hatching eggs. The free-swimming fish were placed in a 600 l outdoor tank, covered with clear plastic and partly filled with water from the tilapia breeding tank. They survived and grew, receiving small quantities of egg-yolk and finely-powdered flakes. The nonswimming larvae were put into various floating baskets and an upwelling airlift system but none reached a free-swimming stage and all had died by the end of the month.

If this work is to be continued it is suggested that the period between checking the females be increased to about a month or six weeks.