All four human malaria species are found in Papua New Guinea (PNG), with a predominace of Plasmodium falciparum and Plasmodium vivax 1. The outcomes of insecticide-treated bednet (ITN) interventions for malaria control in PNG tend to suggest a differential protective effect against P. falciparum and P. vivax. Millen 2 working in the coastal areas of Madang Province showed a significant reduction in the prevalence of P. falciparum among children between one to four years of age, after treated bednets were introduced, but there was no effect on P. vivax prevalence. Graves et al 3 who introduced ITNs in different villages in the same province reported a significant reduction in the incidence of P. falciparum in children 0–4 years old, 4–10 weeks after the ITNs were introduced; however, they also found no effect on the incidence of P. vivax. In the Highlands Region, the P. falciparum:P. vivax ratio in all age-groups changed from 3.8:1, before ITNs were introduced, to 1:1 six months afterwards 4. Other phenomena such as P. vivax relapses may also lead to a change in the prevalence of P. vivax in humans independent of ITN interventions. Relapses are a feature of vivax malaria that may occur 8–10 weeks after a previous attack (short-term relapses) or about 30–40 weeks later (long term relapses). The form and frequency of relapses depend on the infecting strain and occur due to the activation of hypnozoites in liver cells 5.

Little is known about the impact of ITNs on the relative abundance of mosquitoes infected with either P. falciparum or P. vivax. One study on the biting pattern of Plasmodium-infected Anopheles punctulatus mosquitoes, the main vector in PNG, suggests a differential feeding behaviour of mosquitoes infected with P. vivax and P. falciparum, with P. vivax-infected mosquitoes having a tendency to bite earlier than P. falciparum-infected mosquitoes 6. If indeed P. vivax-positive mosquitoes have a greater tendency to bite earlier than P. falciparum-positive ones, people may be more exposed to P. vivax than P. falciparum before going to sleep under the protection of bednets. This could result in changes in the proportion of people infected with the two species following a bednet intervention. This paper describes the biting cycle of P. vivax- and P. falciparum-infected mosquitoes and the impact of an ITN intervention on the relative abundance of mosquitoes infected with either parasite species.

The impact of ITNs on the parasite ratio in mosquitoes was investigated on Lihir Island in New Ireland Province. The island, about 240 km², is bounded by 70 km of coastline with savannah grassland in the north-east and swampy vegetation with thick mangrove shores in the south-west. The average annual rainfall is about 4,000 mm with no distinct dry and wet seasons. The inhabitants live in coastal villages, usually adjacent to a sandy beach or coral reef. Most of the people of Lihir Island do not normally go to bed before 22:00. In September 1993, the PNG government started the sale of permethrin-treated bednets in all 25 villages on the island, at a subsidized price of US$ 3.00 per net. By October 1993, over 50% of the island population reported that they were sleeping under treated bednets. Pre-treatment blood slides obtained from 471 randomly selected residents (all age-groups) of five indicator villages in September 1993 revealed prevalence rates of 31% for P. falciparum and 6.3% for P. vivax, hence a parasite formula (P. falciparum:P. vivax) of 5:1. Follow-up blood slide surveys were not carried out by the Health Department team, but entomological monitoring of the bednet intervention was performed by the Papua New Guinea Institute of Medical Research. Mosquito sampling was conducted on a fly-in fly-out basis as the entomology laboratories were located about 700 km from Lihir island. Sampling could, therefore, only be carried out during four months; one month before and three months after intervention. However, mosquitoes were processed individually to determine infection with malaria parasites.

Mosquitoes were collected using the all-night (18:00 - 06:00) landing catch method and CDC light-traps placed in bedrooms between 18:00 and 06:00. Human biting mosquitoes were caught by two male adults seated outside houses and sporozoite rates were determined for P. falciparum and P. vivax as described previously 67. Differences in sporozoite rates were evaluated using the Chi-square test. To investigate the extent of variation in monthly sporozoite rates in the An. punctulatus complex in the absence of bednets, sporozoite rates were analysed for P. vivax and P. falciparum for 12 consecutive months (January – December 1997), in the highly malarious village of Yauatong in the East Sepik Province. This village, where less than 3% of the population slept under bednets, has already been described by Bockarie et al 6.

In August 1993, one month before ITNs were introduced on Lihir Island, mosquitoes were collected in the villages of Wurtol and Talis, for four consecutive nights using the all-night landing catch and CDC light-traps methods. Post-intervention collections were carried out using the same methods in October, November and December 1994. All of the 2,209 Anopheles mosquitoes collected were morphologically identified as members of the An. punctulatus group including Anopheles. farauti s.l. (95.5%), An. punctulatus (3.8%) and Anopheles koliensis (0.6%). A total of 943 mosquitoes were processed from the landing catch and 26 (2.8%) were sporozoite positive including 14 (1.5%) with P. falciparum and 12 (1.3%) with P. vivax. Mosquitoes with mixed infections were not included in the analysis for this paper. The distribution of sporozoite positive mosquitoes in the three four-hour periods (18:00-22:00, 22:00-02:00 & 02:00-06:00) showed that a higher proportion of infected mosquitoes biting during the pre-bedtime period (18:00-22:00) were infected with P. vivax compared to those infected mosquitoes biting later at night (Figure 1). Four P. vivax and one P. falciparum positive mosquitoes were caught during the pre-bedtime period and the sporozoite rates for this period were 1.6% and 0.4% respectively compared to 1.2% and 1.8% for the rest of the night. Only 7% of the P. falciparum-infected mosquitoes were caught during the pre-bedtime period compared to 33% P. vivax-infected mosquitoes. However, statistical significance could not be obtained for the difference in the species-specific sporozoite rates for the pre-bedtime period probably because of the low sporozoite rates normally observed for wild caught malaria vectors. Sporozoite rates of the An. punctulatus group of mosquitoes seldom exceed 3% 7. Bockarie and co-workers 6, who processed 4,168 An. punctulatus mosquitoes from landing catches in East Sepik Province, also could not establish a statistically significant difference between the sporozoite rates for P. vivax and P. falciparumduring the pre-bed time period despite observing two times more P. vivax- than P. falciparum-infected mosquitoes during this period. According to Bockarie et al 6, the tendency for younger females to bite earlier than older ones may explain the early biting habit of P. vivax-infected mosquitoes which on average are younger than P. falciparum-infected mosquitoes because of the shorter duration of sporogony for P. vivax (seven days) compared to P. falciparum (nine days) 5. They also showed that An. punctulatus mosquitoes infected with L1 larvae of Wuchereria bancrofti were biting earlier than those infected with L2 and L3 larvae.

A total of 1,302 mosquitoes including light-trap catches were processed to assess the impact of ITNs on parasite ratio. The 394 mosquitoes collected before ITNs were introduced included eight (2.0%) P. falciparum-positive and four (1.0%) P. vivax-positive specimens giving a parasite ratio of 2:1. The sporozoite rate determined from 908 mosquitoes caught after ITNs were introduced showed a statistically significant (P = 0.027) decrease for P. falciparum (0.7%) and a slight increase for P. vivax (1.3%) resulting in a post intervention parasite ratio of 1:2 (Figure 2).

However, other phenomena such as relapse may also lead to a change in the prevalence of this species in humans independent of bednet interventions. Relapse is a future of vivax malaria. It represents a reseeding of the bloodstream by dormant parasites ('hypnozoites') contained in the liver. P. falciparum does not produce hypnozoites. Relapses may occur 8–10 weeks after a previous attack (short-term relapses) or about 30–40 weeks later (long term relapses). To investigate the extent of variation in sporozoite rates in the An. punctulatus complex in the absence of bednets, monthly sporozoite rates were analysed for P. falciparum and P. vivax for 12 months (January – December 1997) in the highly malarious village of Yauatong in the East Sepik Province. A total of 5,002 mosquitoes were processed from this village, with monthly sample sizes ranging between 118 and 1426 females (Table 1). The annual sporozoite rates for P. falciparum and P. vivax were 3.1% and 0.6% respectively. Malaria infected mosquitoes were observed in all monthly samples with the exception of September. For the 11 months for which infected mosquitoes were observed, the monthly sporozoite rate for P. falciparum was higher than P. vivax. Monthly P. falciparum:P. vivax formula varied from 8:1 to 1.2:1. In the absence of bednets, P. falciparum remained the dominant species in monthly mosquito collections suggesting that other factors other than relapse may be responsible for the relative increase in the proportion mosquitoes carrying P. vivax following the introduction of ITNs in a community. In a malaria endemic area in India, where P. vivax relapse rate ranged from 23% to 44%, the vectors mosquitoes (Anopheles culicifacies and Anopheles stephensi) remained uninfected during the months (nontransmission season) when relapse rate was highest 10. It is, therefore, unlikely that the increase in the P. vivax infection rate of An. farauti observed during this study was predominantly due to relapse.

These findings suggest that people sleeping under treated bednets in the Lihir villages may be more exposed to P. vivax- than P. falciparum-infected mosquitoes before going to sleep under the protection of bednets. This difference in the biting behaviour of mosquitoes infected with different malaria parasites may partly explain the change in the P. falciparum:P. vivax formula after the introduction of bednets. A similar differential feeding behaviour of mosquitoes infected with the two species have been reported in the East Sepik Province of Papua New Guinea 6. Other studies in Papua New Guinea have shown that the use of untreated bednets to protect against mosquito bites can also lead to an increase in the proportion of P. vivax infections in mosquitoes 8 and humans 48. Recent studies in Madang Province showed that P. vivax is becoming more prevalent relative to P. falciparum with the increasing use of ITNs in the Province. On Bagabag island, where over 60% of the inhabitants slept under bednets, there was no significant difference (P = 0.129) in the prevalence of P. falciparum (19.5%) and P. vivax (17.1%) parasites 9. On the other hand, the prevalence of P. vivax (9%) was significantly lower (P < 0.0001) than P. falciparum (23%) in Liksul village where less than 15% of the population were sleeping under bednets [11].

MJB designed the study and drafted the manuscript. HD supervised the field work and performed the laboratory analysis.