An. gambiae used in all experiments were of the Suakoko strain, molecular form M, established by M. Coluzzi in 1987 from material originating in Suakoko, Bong, Liberia. Colony adults were maintained in 86-L cages supplied with honey-soaked sponges, water, and periodic human blood meals. Blood feeding was conducted in accordance with The Ohio State University's Biosafety protocol No. 2005R0020 and Biomedical IRB protocol No. 200440193, FWA No. 00006378. Oviposition cups were placed with caged adults 2 days after each blood meal, and eggs were collected the following day. The laboratory conditions for the colony were 27 ± 1°C, 80 ± 5% RH, and 13:11 (L:D), with 75-minute gradual crepuscular transitions between photophase and scotophase. Laboratory temperatures did not fluctuate as they would under natural field conditions. Instead, two different laboratory temperatures (23 ± 1°C and 27 ± 1°C) were used in the experiments, each of which occurs within the natural temperature range at times when breeding is common in equatorial Africa. These two temperatures allowed for different speeds of energy-reserve consumption and sexual maturation, in case survival rate and maturation rate do not offset one another in a strictly proportional manner. Mosquitoes to be studied at 23°C were transferred to this temperature on the day following pupation.

Larvae for both the colony and experiments were reared as follows: Eggs were disinfected with 0.05% sodium hypochlorite solution and hatched overnight in flat enamel-coated pans of aged tap water. To produce adults of natural extremes of body size, first-instar larvae were placed, 100 each (for large-bodied) or 1000 each (for small-bodied), in 22.8 × 33.0-cm aluminium pans with 450 ml of aged tap water. The larvae were fed a standard diet of pulverized TetraMin^® fish flakes, following a daily schedule that provided 0.5 mg (for large) or 0.01 mg (for small) of food per larva per day. Pupae were collected daily, segregated in separate 38-L cages (internal dimensions ~45 × 29 × 29 cm, h, w, d respectively) according to rearing regime, and supplied with water wicks. Males and females emerging overnight were separated the following morning, by which time no insemination had occurred. All experimental mosquitoes had constant access to plain water throughout the experiments, regardless of enclosure size or sugar-treatment group.

To establish that rearing regimens produced groups of adults with significantly different body sizes, samples of adults were set aside. One wing from each adult was removed, mounted on a slide and measured from the distal end of the alula to the wing tip to the nearest 0.0001 mm using an Auto-Montage digital imaging microscope (Syncroscopy, Frederick, Maryland, USA).

Newly emerged males were divided into three groups of approximately 50 individuals. Each group was placed into an 80-L acrylic cage (internal dimensions ~45 × 51 × 35 cm, h, w, d, respectively) supplied with two water-soaked cotton wicks. The first group received no sugar. The second group was supplied with two additional cotton wicks soaked with 20% sucrose and allowed to sugar feed ad lib. The third group was given a single meal of 20% sucrose before they were placed in the experimental cage. To accomplish this, the group of males was placed in a screen-lidded 500 ml plastic cup, lined with a paper towel soaked with 20% sucrose. After mosquitoes were allowed to feed for 10 minutes, by which time all were engorged and had ceased feeding, they were transferred to the experimental cages. Survival was recorded twice daily. This experiment was repeated in two replicates each, with both large and small males, each size at 23°C and at 27°C. The experiment was terminated on day 6.

After eclosion, males were transferred, 20 each, into two 38-L cages supplied with four water-soaked cotton wicks each. One cage also was supplied with four wicks soaked with 20% sucrose. A 15-watt red lamp provided ambient light during the crepuscular period and onset of scotophase, when the number of males the number of males swarming was recorded at 5-minute intervals through the crepuscular period and first hour of darkness. The number of males with extended antennal fibrillae was also recorded prior to and after the flight period. It was not possible to record this independently during the swarming period. Although the activity of males approximated the distinctive, tightly looping flight described by Charlwood and Jones 24 and interpreted as swarming, the observation cages were too small to determine whether every airbourne male was involved in swarming behaviour. Therefore, for the purposes of this experiment, the number of males in flight (total number of males, less the number at rest), once antennal fibrillae were extended, was considered to be the number swarming. This experiment was replicated four times each, at 23°C and at 27°C, within each size class. It was terminated after 4 nights.

Three separate experiments were conducted to determine the effect of sugar on male insemination success. Three sizes of enclosures were used, to allow either for greater random contact between males and females or for more space for swarming. In all experiments, the female:male ratio was 1:2, to ensure sufficient insemination data for robust statistical analysis in the groups of males expected to perform poorly and die quickly, due to lack of sugar. This consideration outweighed the alternative one, which is to provide males with more opportunities to encounter uninseminated females. In nature, the operational sex ratio is typically skewed strongly toward males. In the first experiment, 60 newly emerged large males were placed into each of two 38-L acrylic cages supplied with four water-soaked cotton wicks each. One cage was also supplied with four wicks soaked with 20% sucrose. Thirty sugar-fed virgin females, 3–5 days old, were placed into each cage at least 4 hours prior to onset of the crepuscular period. The females were allowed to feed from emergence until they were put into experimental cages. To determine cumulative insemination rate over several days, different groups of females in separate cages were left with the same newly emerged males for 1–5 consecutive nights, or until most males were dead, after which females were removed, dissected in saline, the spermatheca removed and placed under a coverslip on a glass slide, and the spermatheca checked visually for the presence of sperm at 100 and 450× magnification. This experiment was replicated four times at each time interval, both at 23°C and at 27°C, and with both large and small males. At 23°C, groups of females were left with males for 1,2,3 or 5 consecutive nights before dissection. At 27°C, the time intervals were 1,2, or 5 consecutive nights.

In the second experiment, the procedure was the same except that 80-L acrylic cages were used, to provide more space for swarming, and the experiment was replicated four times at 27°C with large males only. To determine cumulative insemination rate, different groups of females in separate cages were left with newly emerged males for either 1, 2 or 5 consecutive nights, or until most males were dead, after which females were removed and dissected to check for the presence of sperm in the spermathecae.

In the third experiment, 100 males were released into each of two greenhouse mesocosms, described below, the morning after emergence. One mesocosm was supplied with 10 hanging yellow cellulose sponges (~10 × 6 × 1 cm) covered with honey. Honey was used instead of sucrose so that mosquitoes could locate it easily from a distance, by orienting to its volatiles. In each of three replicates, the two treatments groups were alternated between mesocosms to control for possible local environmental effects. Each morning, 50 5-to-6-day-old females were released from 7-L acrylic cages supplied with cotton wicks soaked with 20% sucrose, where they had acclimated to the mesocosm environment since emergence. Most were recaptured the following morning by backpack aspirator (for host-attracted females) and mouth aspirator (for resting females), prior to the release of the next group of females. Recovery of every released female on each subsequent morning was not possible, so an average of 6 females captured on subsequent mornings may have spent 2 or more nights in the mesocosm. However, this proved to be equally probable in both treatment groups, so systematic errors were not a concern. Captured females were frozen, then dissected to determine daily insemination success. An attempt was made to count resting males each day to determine survivorship. After night 4, all mosquitoes were removed for a final accounting of male survival and female insemination.

The mesocosms used in this experiment were screen enclosures built within two separate rooms located in a greenhouse facility. Each enclosure was a wooden frame, covered with mosquito netting. Both mesocosms had the same dimensions of 2.84 × 3.63 × 2.08 m (w, l, h) (= 21.4 m³). A two-compartment antechamber served as a safe room to prevent mosquito escape. The mesocosms were illuminated primarily by direct and indirect sunlight, although to simulate an equatorial photoperiod (12:12, L:D), a portion of the winter morning was lighted by an incandescent light on a timer. The minimum temperature was maintained by wall radiators heated by steam and regulated by a sheltered, centrally positioned thermostat. Temperatures fluctuated from 23° to 30°C. Humidity cycled between 50% and 90% RH by mist-spraying devices and humidifiers. Walls of concrete cinder blocks, holding beds of wet sand, stood along one wall of each mesocosm to provide resting sites for the mosquitoes and increase moist surface areas that helped to stabilize humidity and provide water for drinking. At dusk males engaged in looping flights in loose aggregations in a distinctive place, indicating that a behaviour akin to natural swarming occurred. However, the swarm did not form over a classic marker, and swarming activity was not quantified.

Survivorship data were analyzed by a Kruskal-Wallis test for multiple treatments. For the purposes of this analysis, the day of death for males with ad-lib sugar was recorded as time of censoring (discontinuation of the experiment). Post-hoc multiple comparisons between treatments and against a control were performed using Statistica 25, applying the method of Siegel and Castellan 26. Other results (body-size difference; sugar effects on fibrillar erection, swarming, and insemination) were analyzed by Student's t test and Chi-square test. Differences among replicates of experiments were found to be trivial, so data were combined prior to analysis.

Wing-length measurements for males reared at densities of 100 per pan ( = 2.87 ± 0.02 mm) and 1000 per pan ( = 2.61 ± 0.02 mm) demonstrated that rearing procedures resulted in two significantly different body-size classes (t test, P < 0.0001). Based on field measurements 2728, these size classes are representative of natural extremes of adult size.

Without sucrose, at each temperature large males lived significantly longer than small males, and males at 23°C lived longer than their same-size counterparts at 27°C. For example, at 23°C large males lived significantly longer ( = 3.72 days) than small males ( = 2.99 days) (P < 0.0001) and longer than both large and small males at 27°C ( = 2.36 days and 1.94 days, respectively) (P < 0.0001) (Table 1, Figure 1). At both low and high temperatures, a single 20% sucrose meal significantly increased the lives of large males by slightly more than 1 day ( = 4.94 days and 3.84 days, respectively) (P < 0.0001). Among small males, a significant extension of life afforded by a single sucrose meal also was slightly more than 1 day at 27°C ( = 3.15 days) and slightly less than 1 day at 23°C ( = 3.77 days) (P < 0.007). In all survival experiments, the majority of males were dead by 3.5 days. Without sugar, the longest survival time was 4.5 days, achieved by a few large males at 23°C (Figure 1).

On the first night (night 1, 24 hours after emergence), a small proportion of both large and small males had erect fibrillae at 27°C at some time during the crepuscular period or early scotophase (Figure 2a,b), whether or not sugar was available, but most fibrillae returned to a recumbent position within 10 minutes during the crepuscular period. Fibrillae of males at 23°C remained recumbent the first night after emergence (Figure 2c,d). By the second night, up to 100% of sugar-fed large males at both temperatures extended their antennal fibrillae (Figure 2b,d), which occurred just prior to the beginning of scotophase and lasted until just after swarming (ca 1.0 hour). On night 2, at 27°C, only 20% of remaining males without sugar had erect fibrillae (Figure 2), which were sustained less than 40 minutes. In contrast, at 23°C 100% of large males both with and without access to sugar had fibrillar erections on night 2. On night 3 at 27°C (Figure 2) and night 4 at 23°C (Figure 2), there were no erect fibrillae among starved males, and none lived through the night.

In general, sugar was necessary for swarming in both large and small males at 27°C, but not at 23°C (Figure 3). Swarming was first observed the second night after emergence. When sugar was available, swarming was significantly more common at 27°C in males of both sizes (X² = 73.01 and 87.13, respectively, P < 0.05 both) and at 23°C in small males (X² = 9.13, P < 0.05) than when sugar was not available. That was not the case for large males at 23°C (X² = 0.54, P > 0.05); those with and without sugar swarmed in similar numbers. By night 3, when sugar was present, a higher proportion of large and small males swarmed at both 23°C (X² = 20.01 and 71.54 respectively, P < 0.05 both) and 27°C (X² = 200 and 151.65 respectively, P < 0.05 both). Without sugar, swarming activity by males of both sizes ceased after the second evening at 27°C, and after the third evening at 23°C (Figure 3). With sugar available, swarming activity was near 100% for both body sizes, starting on night 2 at 27°C and on night 3 at 23°C.

In 38-L cages, when sugar was present, cumulative insemination rates increased over 5 days from 0 to 88% (large males at both temperatures), to 60% (small males at 27°C), and to 73% (small males at 23°C) (Figure 4). In the absence of sugar, insemination increased similarly to those with sugar over the first two nights at 27°C and over the first three nights at 23°C. After two nights of mating opportunity, sugar still had no significant effect in any treatment group, except among small males held at 27°C, in which sugar-deprived males inseminated slightly fewer females (4%) than males with sugar (12%) (X² = 4.60, P < 0.05). By day 3, there were no surviving males at 27°C in sugar-deprived groups. However, at 23°C, both large and small sugar-deprived males were still alive and had inseminated females (32% and 15% respectively). This was still significantly less than insemination rates of those with sugar (58% and 41%, respectively) (X² = 14.17 and 16.58 respectively, both P < 0.05), and no inseminations occurred beyond night 3 among sugar-deprived males.

In 80-L cages at 27°C, the large males with access to sugar produced results similar to those above. Unlike the experiment above, however, sugar-deprived males had very low rates of insemination after the first (1%) and second (3%) nights, and no inseminations after that (Figure 5). Males with sugar had much higher insemination rates after two days (43%) (X² = 44.68, P < 0.05). After five nights, males with sugar had inseminated 82% of the females.

In the mesocosms, results were similar to the 80-L cage experiment above, except that very few inseminations occurred before night 3 (Figure 6). (Unlike the cage experiment, the mesocosm results are based on only single nights of insemination for one cohort of males, so their total performance is based the compilation inseminations of all four nights.) The insemination rate rose substantially on night 3 if honey was available (26.5%) but not if honey was withheld (0%) (P < 0.05). Daily insemination rates remained high after night 3 when honey was available, but most males without honey were dead by night 3. The average cumulative insemination rates were 71% with honey and 1% without honey. Daily counts of resting males suggest that, in each replicate, there was low survivorship the first night, whether or not honey was present ( = 41.3% and = 42.3%, P = 0.51). After the first night, survival remained high in the room with honey while declining in the honey-deprived room. After the 3^rd night, 38% of males remained alive in the honey-supplied room and 3.6% remained alive in the honey-deprived room.