development takes place varies from fresh and clear to murky and heavily charged with organic debris (Laird, 1946a). Tests of the water from several breeding places at Palmalmal showed hydrogen-ion concentrations ranging from pH 5.5 to pH 7.0. The species was only encountered in places heavily shaded by vegetation, and its breeding places were generally located just within the edge of the jungle which bordered the station.

Simple laboratory experiments were performed to investigate the importance of M. inornatus as a larvicidal agent. A single fourth instar Megarhinus larva was placed in each of eight 250 cc. beakers. These containers each held 5 cm. (approximately 160 cc.) of water from a natural breeding place of the predacious mosquito. Known numbers of both larval and pupal stages of three other mosquito species which showed habitat preferences like those of M. inornatus were added to each of these beakers.

Twenty late instar larvae of Aëdes albolineatus (Theobald, 1904) were isolated in containers 1a and 1b. A similar number of fully-developed larvae of Aëdes scutellaris (Walker, 1859), and of Armigeres lacuum Edwards, 1922, were added to containers 2a, 2b, and 3a, 3b respectively. Twenty pupae of a mixture of all three species were placed in containers 4a and 4b. The beakers were then covered with mosquito-netting to prevent any hatched imagines from escaping and so confusing the results of the investigation.

The surviving larvae and pupae were counted at 4 p.m. on each of the succeeding five days, and the numbers eaten by the Megarhinus larvae in each twenty-four hour period determined. Any larval or pupal remains present were removed, and the number of developmental stages in each container restored to twenty.

The results of the investigation on M. inornatus are set out in Table 1. Information under each experiment number in the table refers collectively to containers (a) and (b) of that number.

It will be seen that the predator destroyed fewer larvae of Armigeres lacuum than of either species of Aëdes. This is probably simply due to the fact that larvae of Armigeres are almost twice as large as those of the two species of Aëdes used in the experiments. The fact that very few pupae were destroyed by the predacious larvae was due to two main considerations. In the first place, the compact and heavily-chitinized condition of the bodies of pupae as compared with those of larvae make it difficult for Megarhinus larvae to seize them. Secondly, pupae can elude their pursuers by evasive darting movements, while larvae, with their long and lightly-armoured bodies and more predictable movements, seldom escape capture.

When not feeding, a larva of M. inornatus usually lies relaxed and almost motionless at the bottom of its container, only moving its head from time to time as it follows the progress of other larvae in its vicinity. Such a resting larva becomes tense just before making a sudden strike at a potential victim. Captured larvae are usually grasped by the posterior part of the abdomen near the base of the siphon, pupae in the neighbourhood of the paddles.

When a large and powerful larva such as that of Armigeres lacuum is seized, there is a brief initial struggle during which the victim drags its captor about the container in frenzied efforts to tear itself free. Such efforts, however, were never observed to meet with any measure of success. The effective nature of the predacious culicine's hold on a captured larva was strikingly demonstrated when an Armigeres larva was grasped with forceps and lifted from the water ten seconds after a Megarhinus larva had seized it. So firmly were the mouth-parts of the predator embedded in the tissues of its victim that the former, although swung gently back and forth in the air for thirty seconds, dangled below the Armigeres without relinquishing its grip. The insects were then replaced in their container, where the M. inornatus larva proceeded to consume its prey as though nothing untoward had happened.

After the initial struggle the Megarhinus larva rises to the surface for air, keeping its prey fully submerged meanwhile. As the victim's resistance grows weaker, its siphonal valves can be seen spasmodically opening and closing under water. M. inornatus larvae frequently commence feeding before their victims' struggles have ceased. Relatively small larvae such as those of Aëdes were observed to be swallowed bodily within two or three minutes, and the predator required about five minutes to deal with the abdomen and prothorax of a fully-grown larva of Armigeres lacuum. Siphons of large larvae are often rejected early in the swallowing process. The heavily-chitinized head-capsules of such larvae are rejected later, a Megarhinus larva sometimes remaining quiescent for as long as an hour with the head of a victim protruding from its mouth.

Besides feeding on the developmental stages of mosquitoes, including smaller members of their own species, these formidable larvae devour other small aquatic animals. They were observed to eat both larval and pupal stages of Chironomidae, small nymphs of dragonflies (Anisoptera), and even small tadpoles, in the laboratory at Palmalmal. Paine (1934) records that larvae of the Javanese Megarhinus splendens Wiedemann devour young tadpoles, and prey on the larvae of Tipulidae and Chironomidae as well as those of mosquitoes. Paine also observes that M. splendens does not find larvae with very hairy bodies distasteful. The same may be said of M. inornatus, for this species was seen to feed on fourth instar larvae of Tripteroides quasiornata (Taylor, 1915), a form with a dense covering of stellate hairs on the abdomen. For that matter, Aëdes albolineatus larvae, which figured in the laboratory experiments already discussed, bear abdominal hairs like those of Tripteroides.

Single coconut husks were never found to be occupied by more than three fourth instar Megarhinus larvae at Palmalmal. On all but one occasion developmental stages of other mosquito species were

absent from husks containing such larvae (Laird, 1946a). Second and third instar larvae of the predator were observed attacking fully-grown larvae of Aëdes scutellaris and Culex papuensis (Taylor, 1914) in the field.

It is thus obvious that under natural conditions larvae of M. inornatus exercise an appreciable measure of control over the breeding of mosquitoes with similar habitat preferences. Furthermore, the adult female of this species does not suck blood. Therefore, M. inornatus is to be considered a beneficial insect.

Attempts have been made to introduce M. inornatus into other countries, as an agent in the biological control of disease-transmitting and pest mosquitoes. In 1929 consignments of this insect were sent from New Britain and liberated in heavily-shaded parts of Hawaii (Swezey, 1930). It was hoped that the predacious larvae would check the breeding of certain day-flying pest mosquitoes. However, after breeding for several generations in the neighbourhood of Honolulu, the M. inornatus colony died out within a year of its introduction (Swezey, 1931). An attempt to establish this mosquito in Fiji was made during 1933 (Paine, 1934). This also proved unsuccessful, although Megarhinus splendens appears to have become established in Fiji after its introduction from Java some years ago (Lever, 1943).

Despite the failure of previous attempts to utilize M. inornatus as an agent in the control of noxious mosquitoes, it is considered that further efforts in this direction would be well worth while. Closer attention should be paid to the habitat preferences of this insect in any future introduction attempts. It was mentioned earlier that M. inornatus and mosquitoes belonging to the scutellaris group of the genus Aëdes inhabit similar breeding places in nature. Members of the latter group play an important part in the transmission of filariasis, particularly in those eastern Polynesian islands where this disease is highly endemic. Great numbers of these mosquitoes develop in household containers or in husks and tins in open places where the shade-loving M. inornatus would not be expected to breed. Such breeding is easily brought under control by artificial means. However, this has the effect of forcing Aëdes scutellaris and related species to retreat to the shelter of the jungle, where their breeding places are much harder to locate. As such jungle breeding places are suited to the development of M. inornatus, this predator might well be of value in a campaign against the filaria-transmitting mosquitoes.

All the other aquatic predators studied inhabit long-established ground pools. The hydrogen-ion content of the water was found to range from pH 4.5 to pH 7 5, but did not appear to have any direct bearing on the composition of the macrofauna. The vegetation of such pools generally consists of masses of filamentous green algae (Spirogyra sp.), and various emergent and marginal plants Echinochloa colona O. Kuntze, and Eleusine indica Gaerter, are typical of the grasses bordering these pools. Clumps of Pennisetum macrostachyum (Brongn.) and of Centotheca latifolia (Linnaeus) are often present. The marginal flora commonly includes Ageratum conyzoides Linnaeus (Compositae), Sida sp.? (Malvaceae), and Clitoria sp.? (Papilionaceae). At the edge of jungle clearings straggling masses of Cucumis sp.? (Cucurbitaceae) overhang pools. Fresh-water swamps have much

the same macroflora as the smaller permanent pools, but generally contain a greater amount of emergent vegetation. A species of Equisetum (E. variegatum Schleicher?) makes up the bulk of the swamp flora at Palmalmal.

These and other plants afford mosquito larvae shade and a degree of shelter (Laird, 1946b). Pools with a comprehensive macroflora are seldom without developmental stages of mosquitoes unless they are heavily populated with various kinds of animals which prey on these insects.

Anopheles farauti Laveran, 1902, and Culex pullus Theobald, 1905, are typical of the mosquito fauna of permanent pools in the Jacquinot Bay area. Third and fourth instar larvae of these two species were used in the feeding experiments with various predacious inhabitants of ground pools, discussed in the following pages. It was not deemed necessary to supply pupae to the predators in pure culture, as the habits and size of the two kinds of pupae concerned are very similar. The aim of these experiments was to find not only whether the predators concerned destroy mosquito larvae and pupae, but whether they exhibit any preference between anopheline and culicine larvae.

The following technique was employed in all the laboratory feeding experiments with aquatic predators which inhabit ground pools in nature. Eight enamelled metal basins 12 cm. in diameter were filled to a depth of 8 cm. with boiled and filtered water from a natural pool, the hydrogen-ion content of which averaged pH 6.5. These basins were placed on a shelf within the laboratory, so that they were shaded from the direct rays of the sun by the overhanging roof. The temperature of the water which they held did not differ markedly from that of shaded natural pools.

At the commencement of each experiment a single specimen of the predator concerned was placed in each basin. Twenty^* late instar larvae of A. farauti were placed in each of containers 1a and 1b, and the same number of C. pullus larvae in containers 2a and 2b. Ten^† larvae of each of these species were placed in containers 3a and 3b; and twenty pupae of a mixture of both species were added to containers 4a and 4b. The basins were then covered with mosquito-netting, to prevent the escape of hatched imagines.

Daily counts of the number of the developmental stages of both genera eaten by the predators were made in the manner already described for the experiments with M. inornatus (p. 456), and were continued for five days in each case. The information under each experiment number in Tables 2–9 again refers collectively to containers (a) and (b) of the number concerned.

Culex (Lutzia) halifaxi Theobald, 1903. (Text-figs. 2a, 2b).

The habitat preferences of this predacious mosquito resemble those of A. farauti and C. pullus (Laird, 1946a). The results of feeding experiments carried out with fourth instar C. halifaxi larvae are given in Table 2.

Text-fig. 2.—Culex (Lutzia) halifaxi, fourth in star laiva. a, Head. b, Terminal abdominal segments.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

It will be seen that there was little significant difference between the mean numbers of anopheline and culicine larvae destroyed in experiments 1 and 2, where larvae of one species only were available. In experiment 3, where larvae of both Anopheles and Culex were available, four times as many of the former as of the latter were consumed by the predacious larvae. As larvae of C. halifaxi rest at the surface of the water with their bodies very nearly parallel to the surface film, it would seem that their apparent preference for anopheline larvae as food is merely due to the fact that these surface-dwelling larvae are more convenient to capture than culicines. The mean number of pupae consumed by a single C. halifaxi larva in twenty-four hours was just under one. This predator was observed to meet with the same difficulty as did larvae of M. inornatus in capturing pupae.

C. halifaxi larvae capture their prey with a sudden lunge like that made by M. inornatus, and the feeding process closely resembles that of the latter insect. Unlike M. inornatus, C. halifaxi was rarely seen to swallow the more heavily chitinized parts of its victim. Late instar larvae of C. halifaxi show no reluctance to attack unwary members of their own kind. Haddow (1942) records that Culex (Lutzia) tigripes Grandpré and Charmoy displays similar cannibalistic tendencies.

Whenever C. halifaxi appeared in a pool already populated by other Culicidae, a decrease in the numbers of the other species present soon became apparent (Laird, 1946a). Haddow (1942) records