Intensive longitudinal sampling of malaria mosquitoes in the African semi-desert reveals that three morphologically indistinguishable species have distinctive strategies for surviving the dry season. See Letter p.387
The scale-up of interventions against malaria in the past decade has reduced the global death rate of this disease by an impressive 42%. However, more than 600,000 malaria-related deaths still occur each year1 — 90% of them in sub-Saharan Africa — meaning that malaria remains one of the most significant sources of infectious-disease mortality. Africa has long been recognized as a crucible for malaria-control efforts, owing to its particular blend of widespread and dominant mosquito species that transmit malaria. One of the great mysteries of malariology has been how these vector populations survive the dry season, when there is little water in which the mosquitoes can lay their eggs. In this issue Dao et al.2 (page 387) report that they have solved this mystery, but the answer is surprisingly complex, like the vectors themselves.
Three closely related sibling mosquito species belonging to the Anopheles gambiae complex are among the most efficient vectors of malaria3 (there are at least seven species in the complex, collectively referred to as A. gambiae sensu lato (s.l.)). This status is owed to their strong association with humans and their success at exploiting a variety of ecological conditions across tropical Africa, from humid rainforests to the fringes of the Sahara Desert, as long as humans are nearby. However, there is an Achilles heel in the relay of malaria parasites between these vectors and humans — all mosquitoes have an obligate aquatic immature stage, and in the absence of water, they cannot breed. During the long dry season of the African savannas and the Sahel region, the rains cease for months, surface water evaporates, humidity plummets and temperatures soar. As long as there is no permanent surface water from reservoirs or rivers nearby, malaria transmission becomes undetectable and the local vector mosquitoes also disappear, only to return again with the rains.
Understanding malaria-vector ecology during the dry season, when populations have reached their lowest point, has great strategic significance because deploying mosquito control specifically at those times and places can have the greatest impact. There are two main possibilities for what happens to the mosquitoes during the dry season: long-distance migration to and from refugia where water persists; or stasis, in which the vectors enter a state of dormancy (referred to as aestivation or summer diapause4) that allows them to safely ride out the dry season in situ, hidden deep inside (unknown) shelters. Yet finding the disappeared mosquitoes is even harder than it sounds.
In fact, Dao et al. did not solve the mystery directly, by physically locating mosquitoes in hiding places or capturing them in the act of long-distance migration, although such efforts are under way5. Instead, their detective work was indirect, using detailed analyses of mosquito population dynamics over time. Although researchers have adopted conceptually similar approaches in the past, the insights that emerge from Dao and colleagues' data were made possible by a sampling effort that is unprecedented both in its detail, allowing the detection of short-lived phenomena, and in its duration, allowing true seasonal patterns to be distinguished from one-off events.
Based in the Sahelian village of Thierola in Mali, the researchers collected mosquitoes from around 120 houses for 2 weeks of every month for 5 years, yielding about 40,000 A. gambiae s.l. samples. From time-series analysis of the combined data from all three species, the authors inferred a statistically significant repeating seasonal pattern that was unexpectedly complex. They observed the predicted wet-season peak and mid-dry-season trough in vector density, but this was followed by a surprising rise in density in the late dry season, before another low as the dry season ended.
To make biological sense of these data, Dao et al. recognized the importance of splitting A. gambiae s.l. into the three genetically defined units found simultaneously in Thierola: A. gambiae sensu stricto (s.s.), Anopheles coluzzii and Anopheles arabiensis. Mosquitoes from the three groups are very closely related and cannot be physically distinguished at any stage in their development. All three hybridize occasionally in nature, but the first two — only recently named as species6 and not universally recognized as such — diverged evolutionarily much more recently than other species in the complex.
Despite the relative youth and morphological homogeneity of this species complex, the fact that the species radiations were accompanied by, if not promoted by, differential adaptations to environmental heterogeneities7 makes it unlikely that its members would respond uniformly to a common physiological stress. Notwithstanding this expectation, it is striking that, when Dao and colleagues partitioned the data by species, the two closest relatives (A. coluzzii and A. gambiae s.s.) showed the most distinct population dynamics (Fig. 1). The authors also found that the population density of A. gambiae s.s. follows a relatively simple pattern of peak abundance in the wet season and a trough throughout the dry season. By contrast, although the density of A. coluzzii also peaks in the wet season, the onset of population growth precedes that of A. gambiae s.s. by two months and, far from disappearing in the dry season, two peaks in population density are consistently observed, despite the absence of rain.
Dao et al. make the case that these data best fit a model in which A. coluzzii persists locally in a form of diapause and emerges from hiding for two short periods. The cues that provoke this emergence are unknown, but could include abiotic factors, such as increases in humidity or temperature, and biotic factors, such as the need to replenish nutritional reserves — for example, by blood feeding without egg maturation, known as gonotrophic dissociation4. By contrast, it seems that A. gambiae s.s. disappears and, when the rains resume, more slowly recolonizes the area from refugia hundreds of kilometres distant.
Although the population dynamics of A. arabiensis were not statistically different from those of A. gambiae s.s., small numbers of A. arabiensis were collected each dry season, suggesting that at least a fraction of the population remains in place. Whether this implies that the species uses a mixed strategy of diapause and long-distance migration, as the authors propose, or whether there is some other explanation (such as a different type or greater depth of diapause) will require further investigation.
Final proof for these hypotheses will have to come from catching the mosquitoes in the act. Nevertheless, there is now strong evidence that A. coluzzii overcomes the stress of the dry season through local diapause, a strategy that ensures its rapid population expansion at the earliest stages of the rainy season and thereby amplifies disease transmission. The long-distance migration proposed for A. gambiae s.s. will also influence the dynamics of disease transmission and vector control, because both processes determine the ability of vector populations to expand their range and invade distant regions. Unfortunately, we know almost nothing about the environmental cues that prompt these processes, the mechanisms responsible for them or even how generalizable these findings are to mosquito populations elsewhere in tropical Africa. Dao and colleagues' work highlights the urgent need for more field studies to answer these fundamental questions.