© 2016. Published by The Company of Biologists Ltd | Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665
RESEARCH ARTICLE Desiccation tolerance in Anopheles coluzzii: the effects of spiracle size and cuticular hydrocarbons Arthur C. Arcaz1, Diana L. Huestis1,*, Adama Dao2, Alpha S. Yaro2, Moussa Diallo2, John Andersen1, Gary J. Blomquist3 and Tovi Lehmann1,‡
ABSTRACT contributing greatly to their role in disease transmission. These The African malaria mosquitoes Anopheles gambiae and Anopheles sibling species withstand dry conditions during short dry spells in – coluzzii range over forests and arid areas, where they withstand dry the rainy season (RS) as well as the 3 8 month-long dry season spells and months-long dry seasons, suggesting variation in their (DS). Their presence in both arid and humid environments suggests desiccation tolerance. We subjected a laboratory colony (G3) and variability in their desiccation tolerance (DT), as was previously wild Sahelian mosquitoes during the rainy and dry seasons to hypothesized (Coluzzi, 1982; Coluzzi et al., 1979; Toure et al., desiccation assays. The thoracic spiracles and amount and 1994, 1998). Recent evidence suggests that during the DS, composition of cuticular hydrocarbons (CHCs) of individual A. coluzzii persists in the Sahel via a form of dormancy mosquitoes were measured to determine the effects of these traits (aestivation) whereas A. gambiae engages in long-distance on desiccation tolerance. The relative humidity of the assay, body migration to re-colonize Sahelian villages in the RS (Dao et al., water available, rate of water loss and water content at death 2014; Huestis and Lehmann, 2014; Lehmann et al., 2010). Eco- accounted for 88% of the variation in desiccation tolerance. Spiracle physiological studies revealed surprisingly modest variation in DT size did not affect the rate of water loss or desiccation tolerance of the between species and carriers of chromosomal inversions that were colony mosquitoes, as was the case for the total CHCs. However, six associated with aridity clines geographically and seasonally (Fouet – CHCs accounted for 71% of the variation in desiccation tolerance and et al., 2012; Gray et al., 2009; Lee et al., 2009; Lyons et al., 2014) three accounted for 72% of the variation in the rate of water loss. Wild findings that are difficult to reconcile with the distinct environments A. coluzzii exhibited elevated desiccation tolerance during the dry they inhabit. The mechanisms underlying DT in anopheline season. During that time, relative thorax and spiracle sizes were mosquitoes are poorly understood. Such knowledge might help smaller than during the rainy season. A smaller spiracle size uncover subpopulations with elevated or reduced DT associated appeared to increase A. coluzzii’s desiccation tolerance, but was with aestivation or other ecophysiological states. not statistically significant. Seasonal changes in CHC composition Adult anophelines have two pairs of spiracles in the thorax and were detected in Sahelian A. coluzzii. Stepwise regression models seven considerably smaller pairs in the abdomen (Snodgrass, 1959). suggested the effect of particular CHCs on desiccation tolerance. In We focused on the large metathoracic and mesothoracic spiracles, conclusion, the combination of particular CHCs along with the total which vary in length (Nagpal et al., 2003; Wagoner et al., 2014), yet amount of CHCs is a primary mechanism conferring desiccation this variation has not been examined in terms of its effect on DT. tolerance in A. coluzzii, while variation in spiracle size might be a Mosquitoes close their spiracles presumably to optimize the balance secondary mechanism. between gas exchange and water loss (Gray and Bradley, 2006) based on external and internal factors, such as relative humidity KEY WORDS: Mosquito, Aestivation, Anopheles gambiae, (RH), hydration and starvation status (Chown and Nicolson, 2004; Desiccation resistance, Water balance, Dry season Krafsur, 1971). In most insects, a muscle contraction closes the spiracles, which open because of the elasticity of the thoracic cuticle INTRODUCTION (Miller, 1960). During flight, all four thoracic spiracles are open, The importance of Anopheles gambiae and Anopheles coluzzii presumably to maximize gas exchange (Herreid and Fourtner, (previously the S and M forms of A. gambiae, respectively) in the 1981). In Drosophila, at least one spiracle is partially open even in transmission of malaria is widely recognized (WHO, 2014), yet extremely dry environments to allow the organism to breathe (Gibbs despite extensive research, major gaps remain in our understanding et al., 2003). of fundamental aspects of their biology. The ranges of these vectors The primary role of the insect epicuticle is to minimize cover steamy forests, dry savannas and semi-arid areas in Africa, evaporation, although it also protects against radiation, infection and injury (Wigglesworth, 1933; Hadley, 1994). The waxy
1Laboratory of Malaria and Vector Research, NIAID, NIH, Rockville, MD 20852, epicuticle is composed mainly of hydrocarbons and other lipids USA. 2Malaria Research and Training Center (MRTC)/Faculty of Medicine, (Hadley, 1994; Haverty et al., 1991; Howard and Blomquist, 2005; Pharmacy and Odonto-stomatology, Bamako, BP 1805, Mali. 3Department of Page et al., 1990). An increased amount of total cuticular Biochemistry and Molecular Biology, University of Nevada, Reno, NV 89557, USA. *Present address: Office of the US Global AIDS Coordinator and Health Diplomacy, hydrocarbons (CHCs) as well as changes in CHC composition US Department of State, Washington DC, USA. were linked to elevated DT in insects (Bazinet et al., 2010; Benoit
‡ and Denlinger, 2007; Chown and Nicolson, 2004; Chung and Author for correspondence ([email protected]) Carroll, 2015; Denlinger and Armbruster, 2014; Gibbs et al., 1997; This is an Open Access article distributed under the terms of the Creative Commons Attribution Gibbs and Rajpurohit, 2010; Howard and Blomquist, 2005). The License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed. rate of water loss (RWL) of insects increases rapidly above a ‘transition’ temperature – an observation that led to the phase-
Received 30 November 2015; Accepted 9 March 2016 transition model that describes cuticle permeability as a function of Journal of Experimental Biology
1675 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665
respectively (Huestis and Lehmann, 2014). The humidity in each Abbreviations treatment was maintained using supersaturated salt solutions of BWA body water available, defined as wet mass−end mass (mg) potassium fluoride (KF) and potassium chloride (KCl; Sigma, St BWCD body water content at death, defined as (end mass−dry Louis, MO, USA), respectively. The salt solutions had a sludge-like mass)/end mass (%) consistency when poured into Petri dishes to maximize the surface CHC cuticular hydrocarbon area of each solution. Five Petri dishes were placed in separate DS dry season DT desiccation tolerance, expressed as survival time (h) plastic, air-tight containers (50×40×20.3 cm). Humidity and RH relative humidity (%) temperature were monitored in the containers using HOBO data- RS rainy season loggers (Onset Computer Corporation, Bourne, MA, USA). In the RWL rate of water loss, defined as BWA/DT, i.e. the loss of body dry and the humid treatment chambers, mean (±s.e.m.) RH and −1 water available over desiccation time (mg h ) temperature were 22.0±0.11% RH (min.=20.1%, max.=43.1%; WL wing length measured from the alular notch to the junction of N=978) and 27.6±0.005°C (min.=26.5°C, max.=27.9°C; N=978) the radius 3 vein and the outer margin (mm) and 80.7±0.019% RH (min.=77.6%, max.=82.4%; N=984) and 27.7±0.005°C (min.=26.6°C, max.=28.2°C; N=978), respectively. Each treatment was replicated in two identical air-tight containers. temperature-dependent melting of its surface lipids (Wigglesworth, One hundred, 3 day old A. gambiae G3 females were 1945). Although not all evidence is consistent with the model’s individually weighed using a microbalance (AD 6000, predictions, most biophysical and eco-physiological studies showed PerkinElmer, Waltham, MA, USA) to the nearest 1 µg, after light that populations and species from warmer and drier environments anesthesia in diethyl ether (Mallinckrodt Chemicals, St Louis, MO, tend to exhibit more lipids that presumably melt at higher USA) for 6 s. This initial mass is referred to as ‘wet mass’. temperatures (Gibbs and Pomonis, 1995; Gibbs and Rajpurohit, Mosquitoes were then placed individually into a 50 ml centrifuge 2010). In A. gambiae s.l., 48 CHCs were identified, ranging in chain tube covered with a net, and then randomly assigned to one of the length from 17 to 47 carbons (Caputo et al., 2007, 2005). Variation two treatments, described above. Individuals were placed in in CHC composition in anopheline mosquitoes has been linked to containers after the RH reached within 5% of the expected value. species (Caputo et al., 2007; Milligan et al., 1993), age (Caputo The treatment chambers were set inside a controlled insectary et al., 2005; Hugo et al., 2006) and mating status (Polerstock et al., chamber maintained at 28°C with a 13 h:11 h light:dark 2002). The possibility that CHC composition plays a dual role in photoperiod, without access to food or water. mate recognition and hence in speciation, while at the same time After 1 h, and thereafter every 2 h, mosquitoes were checked for affecting tolerance to arid conditions (Chung and Carroll, 2015), is mortality, defined by their inability to stand upright when the tubes of special interest given the contentious speciation processes in this were agitated. Upon verification of death, mosquitoes were species complex (Coetzee et al., 2013; Coluzzi et al., 1985; della immediately weighed to obtain their ‘end mass’, transferred to a Torre et al., 2002; Lehmann and Diabate, 2008; Simard et al., 2009). closed 1.5 ml tube with desiccant (Sigma) and heated to 65°C in a However, as in the case of spiracle size, the effect of variation in desiccating oven (HB-1D, Techne, Staffordshire, UK). After 4 days CHC composition on DT, although hypothesized, has not been in the desiccating oven, the samples were re-weighed to obtain their empirically evaluated in mosquitoes to the best of our knowledge. ‘dry mass’. Water loss in insects occurs by evaporation across the cuticle, by The mosquitoes in each treatment were ranked based on survival transpiration through spiracles and via excrement (Gray and duration under the desiccating assay. The top and bottom 20%, as Bradley, 2005; Hadley, 1994; Toolson and Hadley, 1987). Here, well as 10 randomly selected mosquitoes from the center of the we investigated the mechanisms underlying DT in laboratory and distribution (for a total of 50 mosquitoes), were dissected by field A. coluzzii mosquitoes, focusing on the roles of the size of the removing the wings and legs in order to obtain morphological spiracles and of the CHCs. We tested the hypotheses that compared measurements and subsequently CHC measurements. with mosquitoes collected during the RS, those collected in the DS exhibit (i) greater DT, (ii) a smaller relative spiracle size and (iii) a Field DT assay higher total amount of CHCs and changes in their composition. Study sites and mosquito collection Further, we hypothesized that the changes in CHC composition will Indoor resting mosquitoes were collected (using mouth aspirators) include (iv) a reduction in the relative quantity of alkenes, methyl- between 08:00 h and 10:00 h in February and August 2011 from the branched and short-chain alkanes that would result in elevating the village of Thierola, Mali (13.6583N, 7.2155W), located in the melting temperature of CHCs and thus decrease permeability of the Sahel. The last rains fall in October and no surface water can be cuticle to water, in accordance with the phase transitional model found over a 30 km radius from December to the end of May (Dao (Gibbs and Rajpurohit, 2010). Finally, we hypothesized that (v) et al., 2014; Lehmann et al., 2010). The climate in the area has been similar mechanisms in Sahelian A. coluzzii also operate in a related described previously (Huestis and Lehmann, 2014). Collected laboratory colony. Although the order of these predictions follows mosquitoes were identified visually as A. gambiae s.l. and classified our conceptual scheme, the laboratory experiment offers greater by sex and gonotrophic state. They were kept in the shade, covered control over key parameters and it is treated earlier in the Results with a wet towel until subjected to the desiccation assay around section. 11:00 h the same day. Mosquitoes were placed individually into modified 15 ml tubes MATERIALS AND METHODS that were cut 7 cm from the top, and the open side was covered Laboratory DT experiment by a net (the top was kept on). These inserts were placed in 50 ml The A. gambiae G3 colony is maintained in an insectary (27°C, tubes with 15 g of Drierite™, a calcium sulfate desiccant 80% RH, 12 h:12 h light:dark photoperiod) as described previously (W. A. Hammond Drierite Company, Ltd, Xenia, OH, USA), (Artis et al., 2014). The RH of the two treatment chambers was set at with the netted opening pointing up (to avoid contact between the
20% and 80%, simulating the conditions in the Sahelian DS and RS, mosquito and the desiccant). Parafilm was wrapped around the top Journal of Experimental Biology
1676 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665 of the 50 ml tube as an additional sealant. Samples were checked for Data analysis mortality every 2 h (between 00:00 h and 06:00 h, inspection was Mosquito DT was estimated as midpoint survival based on the last done every 3 h). Upon death (defined as above), mosquitoes were time the mosquito was observed alive and the time it was found individually placed into a 1.5 ml vial with silica gel and kept in dead. Mosquitoes that died in the first 2 h were excluded from the direct sunlight for 12 h to expedite desiccation. To identify the DT analyses because of the concern that their death was due to species and molecular form, PCR and restriction enzyme assays handling accidents. were conducted on two legs (Fanello et al., 2002). The specimens For the laboratory experiment, body water available (BWA) was were shipped to the NIH (MD, USA) for additional measurements defined as wet mass−end mass (mg), whereas body water content at (below). death (BWCD) was defined as (end mass−dry mass)/end mass (%). The RWL was defined as BWA/DT (mg h−1) and expressed the loss Morphological measurements of body water available over the desiccation time. Because the Body size was estimated by wing length (WL). Thirteen wing microbalance for measuring the mass of individual mosquitoes was landmarks were used to measure WL of A. gambiae G3 mosquitoes not available in the field, only DT was measured in the field (Huestis et al., 2011). The same distance was measured on experiment. The effect size of independent variables in multivariate A. coluzzii using a dissecting microscope equipped with ANOVA was estimated using ‘semipartial’ ω2 statistics, i.e. after all micrometer ruler under ×40 magnification to the nearest 0.1 mm other effects were partialed out of the effect in question (without (Yaro et al., 2006). After the wings and legs had been removed, adjustment on the dependent variable). This measures the adjusted mosquitoes were placed on a slide with their lateral side facing up. effect as a proportion of the dependent variation remaining after All morphological landmarks were taken using a 3D digital partialing all other effects. This adjustment, based on the covariance microscope (Starlite 200, Quality Vision International, Rochester, with all other factors in the model may result in a lower value than NY, USA) using a routine that was created in Measure-X software the corresponding r2 values in a univariate regression model. (RAM Optical, Rochester, NY, USA). Both left and right lateral Prior to binning of GC-trace peaks, adjustments were made to sides were measured for each mosquito and averaged (unless peak time and peak area based on the time and area of the damage on one side prevented landmark identification, in which pentadecane standard. Peaks with standardized area below 1% of case only one side was used). Thorax size was estimated based on the pentadecane and singleton peaks were excluded. The amount of three landmarks on the thorax (Fig. S1). A combination of eight and total CHCs of each specimen was then estimated as the sum of the 13 points around the contour of the metathoracic spiracles and (adjusted) areas of all CHCs. The CHC composition was expressed mesothoracic spiracles, respectively, were used to assess the size of as the proportion of each (adjusted) peak area of the total (defined each spiracle (Fig. S2). The length, perimeter and area were above). The resulting compositional data were transformed to calculated using functions in Measure-X. minimize interdependence among peaks by the log contrast transformation (Aitchison, 1986), using n-C29 (time=22.7 min) as CHC measurement the common denominator for each mosquito, i.e. transformed- CHCs were extracted and analyzed by gas chromatography (GC). CHn=log(proportion CHn/proportion n-C29). The denominator Each sample was submerged in 12 μl of solvent (10 ppm peak must be present in all specimens, but its identity has no effect pentadecane standard in hexane) for 10 min. A 2 μl sample of the on the results (Aitchison, 1986). extraction solution was injected into an Agilent 6850 Series II gas The absence of a particular CHC peak from certain specimens chromatograph (Agilent Technologies, Santa Clara, CA, USA) complicated the analysis of compositional data because of the equipped with a HD-1 column (30 m, 0.32 mm, 0.25 μm) heated uncertainty of whether the absence reflected a detection problem or from 75 to 150°C at 25°C min−1 and then from 150 to 310°C at a true absence (Aitchison, 1986). We analyzed both the fraction of 8°C min−1 and then held at 310°C for an additional 10 min. The specimens that exhibited each CHC as well as the relative quantity injector was set at 300°C. Helium was used as the carrier gas of each CHC after imputation of a small value (−3.5) when the log (constant flow of 1 ml min−1) and the injection mode was set on contrast transform value was not defined due to zero. This value was pulsed splitless [25 psi (∼172 kPa) pulse pressure]. A 20 Hz data selected because the transformed value of the lowest detected CHC collection rate was used. EZChrom SI data system (Agilent peak was −3.45. Imputation was not performed for CHCs that were Technologies, Santa Clara, CA, USA) was used to calculate the observed in <4 specimens in both seasons (or treatments in area under the peaks (standard option). the laboratory experiment). To address multiple tests, we used Output of the gas chromatograph was visually inspected and the binomial test (which estimates the probability of obtaining the occasional corrections of peak area were manually made using split observed number of significant tests at the 0.05 level given the total peak and alignment of the base of each peak. Exceptionally broad number of tests) as a global test to detect departure from the null peaks with height to area ratio below 0.12 (5.9% of the total peaks) hypothesis, such as across multiple CHCs, and the sequential were excluded. Bins were constructed based on the clusters of the Bonferroni test (Holm, 1979) to detect whether a particular peaks’ retention times and the gaps between clusters. Bins were significant test on the individual test level is significant given the identified by matching their retention time with that of standard total tests performed. Statistical analyses were performed using SAS alkane mix (C21-C40, Sigma-Aldrich 04071), which was diluted 1:9 9.3 (SAS, 2011). with our pentadecane standard solution) and run under the same GC program. Peaks were putatively identified as specific CHCs based RESULTS on alignment with these standards and with peaks previously Laboratory experiment identified by GC-MS using a pool of 30 G3 females of mixed DT and body water content gonotrophic state and ranging in age from 2 to 15 days (and Mosquitoes subjected to the desiccation assay survived longer in the separately, a pool of 30 G3 males, both mated and virgin). Other high humidity treatment: 36.2 versus 16.3 h (P<0.0001, ANOVA, peaks were putatively categorized as oxygenated lipids (mostly fatty F1,95=47.8). Considerable variation in survival (DT) within acids) and excluded from the analysis. treatment was indicated by the spread of the values (Fig. 1A). On Journal of Experimental Biology
1677 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665
A ABC Dry Dry ns Wet 0.010.02 0.03 0.04 0.05 0.06 0.2 0.4 0.6 0.8 1.0 0.50.55 0.6 0.65 0.7 DEF Wet 60
0 102030405060 Dry Survival (h) Wet 50 B Wet mass End mass Dry mass
40 40 N=39 30 x N N=39 =0.77 =39 x x =1.26 (h) DT 20 =0.29
10 20
0 C Per cent 60
0 0.01 0.03 0.05 0.2 0.4 0.6 0.8 1.0 0.5 0.6 0.7 40 N=44 RWL (mg h–1)BWA (mg) BWCD (%) x=0.78 N=44 x=1.19 Fig. 2. Variability in the components of water balance and their N=44 20 interrelationships. Variation in the rate of water loss (RWL; A), body water x=0.25 available (BWA; B) and body water content at death (BWCD; C), and their – relationship to DT (D F) in the G3 colony by treatment (Ndry=39, Nwet=44). 0 0.20.4 0.6 0.8 1.01.2 1.4 1.6 1.8 ’ Mass (mg) was larger (P<0.004, F1,81>8.9, Brown and Forsythe s test; Fig. 2C), suggesting that the capacity to withstand a greater water Fig. 1. Variation in desiccation tolerance and changes in body mass loss is enhanced under dry conditions. However, this might reflect values in 3 day old G3 colony mosquitoes under dry and humid the higher dry mass of mosquitoes subjected to the dry treatment treatments. (A) Variation between treatments in desiccation tolerance (DT), (above). The relationships of these components of mosquito water measured as survival time (Ndry=39 and Nwet=44). The mean and median – survival time are denoted by a diamond and a line (dividing the box), balance and DT were evaluated graphically (Fig. 2D F) and their respectively. (B) Changes in body mass distribution before the assay (wet specific effects evaluated in a multivariate regression model. After mass) and at the end of the assay (end mass), and dry mass under the dry removal of insignificant interaction terms, the model accounted for treatment (20% relative humidity, RH). Arrows (black) point to the mean mass 88% of the variation in DT. The component-specific effect sizes, (x, shown above the arrow with sample size) of each group. (C) Same as in B after partialing out correlation with other factors (see Materials and for the humid treatment (80% RH). methods), revealed that the amount of water available for a mosquito explains most of the variation in DT (25%) followed by RWL (15%) average, mosquitoes lost 36% of their initial wet mass during the and treatment (5%; Table 1). Note that in this formulation, the assay, corresponding to a reduction of ∼13% in body water content ambient RH effect is mostly subsumed in RWL. BWCD accounted (from 78% to 65%). There was no significant difference in initial for only 1% and its confidence interval included zero. As expected, (wet) mass and in the end mass between treatments (P>0.062, increasing RWL or BWCD decreased DT, whereas larger amounts F1,81>3.56, ANOVA; Fig. 1B,C); however, the dry mass of the of initial body water, increased DT (Table 1). mosquitoes subjected to the dry treatment was higher than that of the mosquitoes subjected to the humid treatment (P<0.001, F1,81>40.8, Body size and spiracle size effects on DT ANOVA; Fig. 1B,C), indicating that a small part of the mass lost Contrary to expectation based on the ratio of surface areatovolume, the during the assays was likely nutritional reserves rather than water. effects of body size, typically measured by WL and thorax perimeter −1 The RWL increased at lower RH (0.032 versus 0.012 mg h , on DT and on RWL were not significant (P>0.21, F1,51<1.57; Fig. 3). P<0.0001, F1,81>221.4, Welch’s ANOVA; Fig. 2A), and so did its The length, perimeter and area of both the mesothoracic and variance (P<0.0003, F1,81>15.2, Brown and Forsythe’s test; metathoracic spiracles were positively correlated with WL and Fig. 2A), with few mosquitoes under dry conditions exhibiting thorax perimeter (0.11
P<0.0001, F1,81>69.4, Welch’s ANOVA; Fig. 2C), and its variance less sensitive to the degree to which it is opened. As the spiracle may Journal of Experimental Biology
1678 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665
Table 1. The effect of rate of water loss (RWL), body water available (BWA) and body water content at death (BWCD) on desiccation tolerance of laboratory G3 mosquitoes (r2=0.88) a Source FN,D P Estimate Effect size (90% CI)
Model 116.75,76 0.0001 ND ND − − Intercept 19.81,77 0.55 10.9 ND − – Treatment (dry/wet) 36.81,77 0.0001 29.1 0.056 (0.002 0.15) − – RWL 67.21,77 0.0001 2001 0.102 (0.021 0.21) – RWL×treatment (dry/wet) 27.71,77 0.0001 1552 0.042 (0.000 0.13) – BWA 170.01,77 0.0001 54.5 0.258 (0.13 0.38) – BWCD 8.11,77 0.0056 70.2 0.012 (0.000 0.079) aEstimate of effect size based on the semipartial ω2, measuring the proportion of the dependent variable specifically accounted for by each effect after partialing out the correlations between the other independent effects on each other (see Materials and methods). N and D subscripts represent the numerator and denominator d.f. of the F-test. ND, not determined. have settled in a random state upon death, the degree of spiracle amounts of total CHCs (as well as body size-adjusted total CHCs) opening might add considerable variation to its area and perimeter. did not increase DT or decrease RWL (P>0.35 ANCOVA, not shown). Therefore, we used spiracle length in subsequent analyses. Considerable variation was observed among individual females. Although a high proportion of mosquitoes exhibited n-C29, methyl- CHCs branched C33 (MebrC33) and MebrC25 (91%, 88% and 86%; A total of 30 CHCs were putatively identified in 64 females that were Fig. 6), no CHC was detected in all of the G3 mosquitoes and only subjected to the laboratory desiccation assay. Surprisingly, the total 13 of 30 CHCs were detected in 50% or higher proportion. The amount of CHCs in the mosquitoes subjected to the dry treatment was methyl-alkane 2MebrC27 was detected in 5 of 26 mosquitoes higher than in those subjectedtothehumidtreatment(P<0.006, subjected to the dry treatment but in 0 of 38 from the humid Welch’s ANOVA) and a larger variance was also evident in that group treatment (P<0.008, Fisher exact test; P>0.05 at the multi-test level; (P<0.019, Brown and Forsythe’s test; Fig. 5). In part, this difference Fig. 6). The alkanes n-C21, n-C23 and n-C25 were found in a higher was produced by five mosquitoes (20%, Ndry=26) that had considerably proportion of the mosquitoes subjected to the dry treatment larger amounts of CHCs (Fig. 5), but even excluding these individuals, (P<0.01, Fisher exact test; Fig. 6), whereas C27:1 exhibited the the total amount of CHCs was elevated in that group, suggesting that reverse trend. The relative quantity of the CHCs, expressed as the log under harsh desiccation conditions, mosquitoes can respond by contrast-transformed values of CHCs (Fig. 6), showed differences elevating the total amount of CHCs (Fig. 5). However, higher between treatments in nine of 25 CHCs (overall P<0.0001, binomial
60 60 50 50 40 40 30
30 (h) DT DT (h) DT 20 20 10 10 0.06 0.06 Dry Dry Wet Wet 0.05 0.05 ) ) –1
–1 0.04 0.04 0.03 0.03 RWL (mg h RWL RWL (mg h RWL 0.02 0.02 0.01 0.01 0.08 0.10 0.12 0.005 0.01 0.015 0.02 2.7 2.8 2.9 3.0 3.1 2.7 2.8 2.9 3.0 3.1 Metathoracic Mesothoracic WL (mm) Thorax perimeter (mm) Spiracle size (mm)
Fig. 3. Effects of body size on DT and RWL under the two treatments. Body Fig. 4. Effects of spiracle size on DT and RWL under the two treatments. size was measured as mean wing length (WL) and thorax perimeter. Spiracle size was measured as the length of the mesothoracic and Regression lines have been added to convey trends although the relationships metathoracic spiracles. Regression lines have been added to convey trends are not statistically significant (see Results). For both WL and thorax size, although the relationships are not statistically significant (see Results). Ndry=26 sample sizes were Ndry=26 and Nwet=28. and Nwet=28. Journal of Experimental Biology
1679 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665
Table 2. The effect of cuticular hydrocarbon (CHC) composition on desiccation tolerance (DT) and RWL of the G3 colony based on stepwise regression models DT (r2=0.71) RWL (r2=0.73)
a a Source FN,D P Estimate Effect size (90% CI) Source FN,D P Estimate Effect size (90% CI)
Model 167,45 0.0001 ND ND Model 32.24,48 0.0001 ND ND − – – Treatment 621,45 0.0001 24.0 0.39 (0.19 0.54) Treatment 17.2 0.0001 0.010 0.26 (0.07 0.26) (dry/wet) (dry/wet) − – – n-C24 4.31,45 0.0440 5.64 0.03 (0.0 0.16) n-C21 6.51,47 0.0143 0.0038 0.04 (0.0 0.18) – – n-C25 8.71,45 0.0049 6.39 0.06 (0.0 0.21) 3MebrC29 15.01,47 0.0003 0.0031 0.09 (0.0 0.25) − – – n-C32 211,45 0.0001 6.38 0.13 (0.01 0.3) MebrC29 3.31,47 0.075 0.0023 0.02 (0.0 0.14) − – MebrC29 5.21,45 0.0270 5.97 0.03 (0.0 0.17) – n-C27 7.91,45 0.0075 5.17 0.05 (0.0 0.20) − – MebrC31 28.51,45 0.0001 10.7 0.17 (0.03 0.36) aEstimate of effect size based on the semipartial ω2 is described in Table 1 (see Materials and methods). test). Three (MebrC35, 3MebrC29 and n-C23) of the nine CHCs available for morphometric and CHC analyses. They represented 23 were significant in the multi-test level (Fig. 6). and 11 gravid females collected during the DS and the RS, To evaluate the effects of total CHCs as well as CHC composition respectively. In agreement with predictions (Introduction), the mean on DT, we used a stepwise regression model (Materials and DT of females collected in the DS (15.7) was higher than that of methods). A set of six CHCs and treatment accounted for 71% of the females collected in the RS (11.1, P<0.0003, N=81, one-way variation in DT (treatment alone accounted for 38%), and all were Wilcoxon test; Fig. 7A). In the subset of mosquitoes used in the statistically significant (P<0.05; Table 2). The effects of n-C25 and morphometric and CHC analyses, however, the difference was only n-C27 on DT were positive, whereas those of n-C24, n-C32, marginally significant (P<0.093, N=34, one-way Wilcoxon test; MebrC29 and MebrC31 were negative. Of these, n-C25, n-C27 and Fig. 7A), probably reflecting the smaller sample size. MebrC29 were elevated in the dry treatment (Fig. 6). Considering RWL, 73% of its variation was accounted for by the effect of the Body size and spiracle size effects on DT assay RH (treatment), 3MebrC29, MebrC29 and n-C21. In a model Body size measured by WL in the (early) DS was larger than that containing the relative abundance of the selected CHCs, increasing in the RS (3.18 versus 2.95 mm, P<0.001, F1,76=24.9; Fig. 7B), total CHC amount decreased RWL (Table 2). Variable-selection consistent with earlier findings (Huestis et al., 2012). The effect of regression models may produce misleading results based on WL on DT was significant overall (Fig. 7B); however, it was spurious correlations unless independent data are available for confounded by the effect of season because when season was validation tests. Notably, MebrC29 is the only CHC that showed a included in the model, the body size effect was insignificant (P>0.1, negative effect on DT and a positive effect on RWL. ANCOVA), yet season remained significant (P<0.029, in the presence of body size). The interaction between season and WL was Field experiment not significant (P>0.4, ANCOVA). A total of 81 gravid A. coluzzii mosquitoes were subjected to the Spiracle size (length) of both the metathoracic and mesothoracic desiccation assay in the field but only 34 intact specimens were spiracles was correlated with thorax size (r=0.44 and 0.39, respectively, P<0.014) more than with WL (0.24 and 0.13, respectively, P>0.12). Indeed, thorax size was modestly correlated with WL (r=0.62, P<0.001, N=38). Metathoracic spiracle size and Dry 5 mesothoracic spiracle size were positively correlated with each other 5 Wet 4 (r=0.51, P<0.0007). 3 Seasonal allometry was found between WL and thorax size 4 2 (Fig. 8A), showing that for the same WL, expected thorax size
Total CHCs Total 1 in the DS (least square means =2.856 mm) was smaller than that in 0 the RS (least square means=2.989 mm, P<0.013, ANCOVA; 3 5101520Fig. 8A). This difference (0.133 mm) amounts to 70% of the No. of CHCs detected standard deviation of thorax size (0.18 mm; Table S1). The
Total CHCs Total 2 hypothesis of relatively smaller spiracle size during the DS was tested using least square means in ANCOVA (one-side test). The interactions between season and both measures of body size 1 P<0.005 were insignificant and, therefore, were removed. The hypothesis was confirmed for both spiracles (P<0.05), except for the 0 metathoracic spiracle with respect to thorax size (P<0.084; Dry Wet Fig. 8B), indicating that during the DS, spiracle size is smaller than expected from body size even relative to the already smaller Fig. 5. Differences in total cuticular hydrocarbons (CHCs) between thorax size during the DS. treatments. The larger variance of the dry treatment (box plot and inset) was The effect of spiracle size on DT was evaluated in univariate and detected by Brown and Forsythe’s test of homogeneity of variance (P<0.019). The difference between means was evaluated by Welch’s ANOVA. Inset multivariate ANCOVA. Although a larger spiracle appeared to shows the relationship between total CHCs and the number of CHC peaks decrease DT, according to the hypothesis (Fig. 9), these effects were
(note the 5 extreme values). Ndry=26 and Nwet=38. non-significant (including in multivariate analyses, not shown). Journal of Experimental Biology
1680 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665
Fig. 6. Variation in CHC composition 1.0 A of G3 mosquitoes between Wet treatments. (A) The proportion of Dry mosquitoes exhibiting each CHC as a measure of abundance. (B) The relative 0.8 quantity of each CHC using log contrast- transformed values. Lines represent 95% confidence interval (CI) of the mean 0.6 (single test level). Significant differences in frequency between treatments are indicated by asterisks below bars. Black and red asterisks indicate significance at 0.4 the individual and multi-test level, ‡ respectively ( P<0.1, *P<0.05, **P<0.01 and ***P<0.001).
0.2 Proportion of mosquitoes exhibiting each CHC 0 C21 C22 C23 C24C25:1C25 MeC253MeC25C26 C27:1C27 MeC27 2MeC27MeC28C29:1C29 MeC293MeC29C30 MeC30C31:1C31 MeC31C32 MeC32MeC332MeC33MeC352MeC352MeC39
*** ** ‡ ** * ‡ ** 2.0 B 1.5
1.0
0.5
0
–0.5
–1.0
–1.5
–2.0
Relative quantity of each CHC –2.5
–3.0
–3.5 C21 C22 C23 C24C25:1C25 MeC253MeC25C26 C27:1C27 MeC27 2MeC27MeC28C29:1MeC293MeC29C30 MeC30C31:1C31 MeCC32 MeC32MeC332MeC33MeC352MeC352MeC39 31
‡ ** * ** * * * ** ‡ * ** CHCs
CHCs test; Fig. 10A), whereas 2MebrC37 and n-C28 were detected in During the DS, total CHC appeared elevated and its distribution higher proportions of DS mosquitoes (P<0.035, and P<0.054, wider than that in the RS, yet these differences were not significant respectively, Fisher exact test), suggesting these CHCs were more (P>0.15, Welch’sANOVA,F1,33=1.69). Total CHCs did not concentrated during the DS. Considering the log contrast-transformed vary significantly in relation to body size by season or across values (Fig. 10B), significant seasonal differences in relative quantity seasons (P>0.2). Finally, a higher amount of total CHCs appeared were detected in nine of 21 CHCs, indicating a seasonal shift in to increase DT in the DS but this effect was non-significant composition (P<0.0001, binomial test). Further, six of the nine CHCs (P>0.3, ANCOVA, F1,30=0.8), after removal of non-significant were also significant in the multi-test level (Fig. 10B). interactions, as was also the case during the RS (P>0.3 ANCOVA, To evaluate the effects of total CHCs as well as CHC composition F1,30=0.81). on DT, we used a stepwise regression model (including season). A set The composition of CHCs was overall similar across seasons of five CHCs and total CHCs accounted for 66% of the variation in (Fig. 10A). Only C31:1 was season specific; it was present in all the DT and all were statistically significant (P<0.05; Table 3). The effects RS specimens (N=10) but in none of the DS specimens (N=23, of the total CHCs, n-C28, n-C33, and MebrC33 were positive, P<0.001, Fisher exact test; Fig. 10A). Similarly, the alkene C27:1 was whereas those of C29:1 and n-C31 were negative. Of these, MebrC33 present in a smaller fraction of DS mosquitoes (P<0.006, Fisher exact and C29:1 were significantly different between seasons (Fig. 10), Journal of Experimental Biology
1681 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665
Fig. 7. Seasonal difference in DT in wild gravid Anopheles AB coluzzii and variation in body size in relation to DT. (A) DT in the rainy season (RS) and dry season (DS). Black and blue 30 distributions represent, respectively, the full number and a subset DT=9.87ϫWL–16 of specimens used in the desiccation assay and in the N=78, R2=10%, P<0.003 morphometric and CHC analyses. The seasonal difference is highly significant over the total data (P<0.0003, N=81, Wilcoxon test one-way) and significant in the subset (P<0.047, N=34, Wilcoxon test one-way). (B) Seasonal variation in body size (measured as WL) is summarized in the horizontal box whisker 20 (top; see Results). Red and blue dots represent DT values of gravid mosquitoes from the DS and RS, respectively. The 16.6 regression line shows the relationship of DT and WL regardless of DT (h) DT 15.7 season (see Results).
11.7 11.1 10
0 DS RS 2.6 2.8 3.0 3.2 3.4 WL (mm) although MebrC33 was reduced during DS, contrary to its positive mosquitoes exhibited higher DT, smaller spiracle size relative to effect on DT, while C29:1 was reduced during DS, in accordance body size and specific seasonal changes in their CHC composition, with its negative effect on DT. Likewise, n-C28 was more common in some of which correlated with increased DT. These findings, and the DS, in accordance with its positive effect on DT. especially the seasonal allometries in thorax and spiracle size of A. coluzzii, are consistent with local persistence throughout the DS, DISCUSSION referred to as aestivation (Adamou et al., 2011; Dao et al., 2014; Using laboratory and field experiments, we have evaluated the Lehmann et al., 2010). However, there was insufficient evidence effects of spiracle size, CHC composition and total CHC amount on that smaller spiracle size increased DT, possibly because of limited desiccation tolerance of the African malaria mosquito. Many studies sample size. There was partial congruence between the results of the have speculated on the importance of tolerance to aridity for the field and the laboratory experiments in respect to the roles of these persistence of members of the A. gambiae complex in dry mechanisms on DT. environments (Bayoh et al., 2001; Coluzzi et al., 1985, 1979; Huestis and Lehmann, 2014; Lehmann and Diabate, 2008) and Effects of body and spiracle size on DT about the effects of the spiracles and CHCs in minimizing water loss Consistent with our results, a seasonal increase in WL during the (Benoit and Denlinger, 2007, 2010; Fouet et al., 2012; Gray and early DS was reported based on specimens collected a year earlier Bradley, 2005; Gray et al., 2009; Hidalgo et al., 2014; Huestis et al., (2010) in Thierola (Huestis et al., 2012). As predicted based on 2012; Mamai et al., 2014; Nagpal et al., 2003; Wagoner et al., surface-to-volume ratio (Gray and Bradley, 2005), larger WL 2014). Our results have produced support for at least three of our increased DT in a univariate regression (Fig. 7); however, this effect five hypotheses (see Introduction): during the DS, A. coluzzii did not persist within season and was confounded with season. In
ABFig. 8. Seasonal allometry in thorax–wing size and spiracle size in relation to both wing and thorax size. (A) Rainy Variation in thorax size in relation to WL by season (red and WL 3.2 * blue denote DS and RS, respectively). The effect of season Dry (P<0.012) and WL (P<0.001) in a covariance analysis (interaction term was not significant) are shown. Horizontal Rainy Thorax lines denote mean thorax size of each season adjusted to WL ‡ values (least square means). (B) Mean spiracle size adjusted 3.0 Rainy Metathoracic Dry to WL and thorax size and their 95% CI computed in a covariance analysis. The significance of one-way tests of the * Rainy Dry hypothesis of smaller relative spiracle size during the DS is WL ‡ shown in blue (*P<0.05 and 0.05 Rainy Mesothoracic Thorax Dry * 2.6 2.8 3.0 3.2 3.4 0.08 0.09 0.10 0.11 0.12 WL (mm) Adjusted spiracle size (ls mean; mm) Journal of Experimental Biology 1682 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665 the laboratory experiment, no effect of body size (WL and thorax Table 3. The effect of CHC composition on DT of A. coluzzii in Mali perimeter) on DT or RWL was detected (Fig. 3). Possibly, the G3 based on stepwise regression models (r2=0.66) a mosquitoes used were less variable in size than wild mosquitoes as Source FN,D P Estimate Effect size (90% CI) suggested by their coefficient of variation (Table S1). Model 6.8 0.0004 ND ND Relative to body size (WL), the thorax of A. coluzzii was smaller in 6,27 Intercept 18.31,21 0.0003 53.1 ND the DS than in the RS (Fig. 8). Moreover, during the DS, the thoracic – Total CHCs 15.51,21 0.0008 7.9 0.25 (0.023 0.48) – spiracles relative to WL and relative to thorax perimeter were smaller n-C28 5.61,21 0.0273 6.4 0.09 (0.00 0.32) − – than they were during the RS (Fig. 8). Smaller spiracle size (relative to C29:1 5.91,21 0.0243 5.6 0.095 (0.00 0.32) − – body size) was concurrent with elevated DT during the DS (although n-C31 10.11,21 0.0046 6.8 0.16 (0.00 0.40) – the correlation was not statistically significant), in agreement with it n-C33 20.31,21 0.0002 7.9 0.33 (0.06 0.54) MebrC33 16.1 0.0006 12.2 0.26 (0.027–0.48) being a mechanism to minimize transpiration. It may also be 1,21 a 2 explained by the hypothetically lower demands for gas exchange due Estimate of effect size based on the semipartial ω is described in Table 1 (see to restricted movement of aestivating mosquitoes (Huestis et al., Materials and methods). 2012). The relative reduction in thorax size is consistent with this The ‘spiracular index’ (ratio of the length of the mesothoracic explanation and further suggests that during the DS, mosquitoes spiracle to the length of the thorax) was compared among African allocate fewer resources to locomotion, yet the larger wings suggest anophelines with respect to the aridity of their habitat that mosquitoes carry heavier loads (blood, sugar meals or water) (Vinogradskaia and Detinova, 1978). The members of A. gambiae during their short foraging excursions. Such allometries highlight s.l. were grouped based on ‘intermediate index’ as moderately that, as in other diapause developmental programs, the insect xero-resistant although large differences were found between responds to certain token stimuli before metamorphosis (Denlinger, mosquitoes from savanna versus forest habitats. Seasonal 2002; Denlinger and Armbruster, 2014; Huestis and Lehmann, variation in spiracle size was suggested in A. stephensi from arid 2014), consistent with aestivation. Could the shift in morphology be a regions in India, showing a smaller spiracle during the DS (Nagpal result of differential survival during the DS instead? Accordingly, the et al., 2003). Wagoner and colleagues (2014) have examined the distribution of the spiracular ratios during the RS would subsume that variation in spiracle size between G3 mosquitoes raised from eggs of the DS. However, contrary to this prediction, over 75% of spiracle: under short and long photoperiod regimes. They found that, wing ratio values during the DS were below the lowest 95% contrary to their hypothesis, mosquitoes exposed to short confidence interval of the corresponding RS distributions (Fig. 8). photoperiod, corresponding to the DS, exhibited larger WL and Further study into the seasonal and spatial variation in this larger relative mesothoracic spiracle length than mosquitoes morphometric signal will help determine its potential as a marker exposed to long photoperiod. Direct evaluation failed to detect an of aestivation as well as its function. effect of spiracle size on DT in A. coluzzii and in the G3 colony of A. gambiae (Fig. 4). Altogether, we conclude that variation in spiracle size in A. coluzzii may be a secondary mechanism underlying its DT. Nonetheless, transpiration through the spiracles r=–0.11, P>0.5, N=34 r P N might be regulated behaviorally by closing the spiracles (Gray and 30 =–0.05, >0.7, =34 Bradley, 2006; Hadley, 1994; Krafsur, 1971), as shown in Movies 1 and 2 morphologically by the setae (cuticular ‘hairs’) that surround the opening (Mamai et al., 2016). 20 Effects of the total amount and composition of CHCs on DT During the DS, the total amount of CHCs of A. coluzzii appeared 10 higher than that during the RS, especially in a subset of the DS mosquitoes, although the difference was not significant (Fig. S3). Moreover, higher total CHCs alone did not increase DT in A. coluzzii 0 and in G3 mosquitoes. Likewise, total CHCs alone did not reduce DT (h) DT r=–0.07, P>0.7, N=23 r=-0.05, P>0.8, N=23 RWL of G3 mosquitoes. Conversely, in combination with particular 30 r=–0.07, P>0.8, N=11 r=+0.04, P>0.9, N=11 CHCs, the total amount of CHCs elevated DT in A. coluzzii, suggesting total CHC amount plays a modest role as a mechanism conferring elevated DT on members of the A. gambiae complex. 20 Adaptation to desiccation stress during two independent selection experiments on Drosophila showed stability in the total amount of CHCs, but the relative abundance of longer chain CHCs increased, 10 reducing RWL (Gibbs et al., 1997; Kwan and Rundle, 2010). Consistent with our results, G3 mosquitoes exposed to short photoperiod and low RH, simulating the DS, exhibited greater total 0 CHCs than mosquitoes exposed to long photoperiod and high RH, 0.04 0.06 0.08 0.10 0.12 0.050 0.075 0.100 0.125 0.150 and also revealed differences in CHC composition between these Mesothoracic Metathoracic groups (Wagoner et al., 2014). Seasonal changes in the CHC Spiracle size (mm) composition were detected in A. coluzzii (Fig. 10A), with the notable absence of C31:1 from all DS specimens as opposed to its Fig. 9. Effects of mesothoracic and metathoracic spiracle size (length) on DT. Top, overall results; bottom, results by season. Red and blue denote DS presence in all RS specimens. The alkene C27:1 was detected in a and RS, respectively. Correlation coefficients and their significance are shown smaller fraction of DS mosquitoes, whereas 2MebrC37 and n-C28 for the top panels. were detected in higher proportions of DS mosquitoes. Similarly, Journal of Experimental Biology 1683 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665 A Fig. 10. Seasonal variation in CHC composition of A. coluzzii. (A) The 1.0 proportion of mosquitoes exhibiting each CHC as a measure of its abundance. (B) The relative quantity of each CHC using log contrast- transformed values. Significant 0.8 differences in frequency between seasons are indicated by asterisks below bars. Black and red asterisks indicate significance at the individual 0.6 and multi-test level, respectively ‡ ( P<0.1, *P<0.05, **P<0.01 and ***P<0.001). Ndry=23 and Nrainy=12. 0.4 0.2 Proportion of mosquitoes exhibiting each CHC 0 C23 C24 C25:1C25 M C26 C27:1C27 MeC27 2MeC27C28 MeC28C29:1MeC293MeC29C30 MeC30C31:1C31MeC312MeC31C32 MeC32C33MeC2MeC33MeC352MeC35C37 2MeC37C38C39MeC392MeC39 eC25 33 B ** ‡ *** * 2.0 Dry season 1.5 Rainy season 1.0 0.5 0 –0.5 –1.0 –1.5 –2.0 –2.5 –3.0 Relative quantity of each CHC –3.5 –4.0 –4.5 –5.0 C23 C24 C25:1C25 M C26 C27:1C27 MeC27 2MeC27C28 MeC28C29:1MeC293MeC29C30 MeC30C31:1C31 MeC312MeC31C32 MeC32C33 MeC2MeC33MeC352MeC35C37 2MeC37C38 C39MeC392MeC39 eC25 33 ‡ ** **** ** *** *** * CHCs * the relative quantity of C27:1, MebrC29, MebrC31, MebrC32 and that melt at higher temperatures. Thus, the elimination of the MebrC33 was reduced in the DS (Fig. 10B). The reductions in the ubiquitous RS alkene C31:1 and the reduction in the relative relative abundance of alkenes and methylated alkanes followed quantity of C27:1 as well as the methyl-branched alkanes from expectations based on studies showing that adding a double bond or A. coluzzii during the DS fit well with this model. In evaluating a methyl group to an alkane (of a given length) reduces its melting direct effects on DT, a more complex pattern arises, possibly temperature by ∼10–50°C, depending on the position where the demonstrating the interactive nature of CHCs in forming the change is made (Gibbs and Pomonis, 1995; Stinziano et al., 2015). permeability of the epicuticle (Gibbs and Rajpurohit, 2010). Thus, According to the phase shift model, desiccation-resistant insects, both ‘expected’ reductions in the relative quantity of the alkene especially in hot and arid environments, possess a layer of CHCs C29:1 and elevated amounts of n-C28 and C33 were observed Journal of Experimental Biology 1684 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665 together with the ‘unexpected’ reduction of C31 and the increase in decreasing MebrC31 increased DT of the G3 mosquitoes (although MebrC33 (Table 3). Despite growing knowledge on lipid-phase it showed no effect on RWL) and was found in a lower amount transitions, how different compounds actually interact to determine during the DS in A. coluzzii (above). The remaining four CHCs that cuticular impermeability to water is poorly understood, and likely affected DT and the two affecting RWL in the laboratory includes interactions among CHCs, other lipids, proteins and experiment, however, showed no effect on A. coluzzii. The probably melanization (Gibbs and Rajpurohit, 2010). differences in CHC composition between the laboratory and the Several CHCs showed consistent patterns, increasing during the field experiments (Table 4) may be attributed to the gonotrophic DS and having a positive effect on DT (n-C28) or decreasing during state (unfed and gravid, respectively), the genetic makeup of the G3 the DS and having a negative effect on DT (C29:1), yet disparities colony (a hybrid between A. coluzzii and A. gambiae from locations were also observed, such as the reduction of MebrC33 during the distant to our field site), and/or the conditions they have been DS and its positive effect on DT. Notably, not all the CHCs that were maintained under. These results highlight the uncertainty in included in the seasonal comparisons were included in the analysis generalizing patterns from the laboratory to the field and the need of the effects of the CHCs on DT (Materials and methods). The for validating studies. Yet, concordance between the field and laboratory experiment reinforced that decreasing MebrC29 laboratory results substantiate the role of CHC composition and enhanced the DT of G3 mosquitoes (Table 2), consistent with its total CHC amount as underlying mechanisms conferring DT in lower amount in A. coluzzii during the DS (above). Similarly, anopheline mosquitoes. Table 4. CHCs detected in the laboratory G3 colony by treatment and in field-collected gravid A. coluzzii (Mali) by season No. mosquitoes exhibiting CHC Difference in relative quantity G3 Mali G3 Mali a b a Time (min) CHC CHC type Ndry Nwet Ndry Nrainy Treatment DT effect Season DT effect 1 15.1 n-C21 Alkane 22 12 00***/** ns/* ND ND 2 16.2 n-C22 Alkane 18 25 00ns/* ns/ns ND ND 3 17.2/17.3 n-C23 Alkane 24 21 2 2 **/** ns/ns ND ND ‡ 4 18.27/18.1 n-C24 Alkane 17 16 2 2 /ns */ns ND ND ‡ 5 18.93 C25:1 Alkene 11 19 7 5 ns/ ns/ns ns/ns ns 6 19.2 n-C25 Alkane 23 18 14 6 **/ns **/ns ns/ns ns ‡ 7 19.3 MeC25 Methyl-branched 24 31 23 10 ns/ns ns/ns ns/ ns 8 19.44 3MeC25 Methyl-branched 5 2 00ns/ns ns/ns ND ND 9 20.1 n-C26 Alkane 2 4 15 3 ns/ns ns/ns ns/ns ns 10 20.8 C27:1 Alkene 17 12 4 7 */ns ns/ns **/** ns ‡ 11 21 n-C27 Alkane 16 32 22 10 /* **/ns ns/ns ns 12 21.35 MeC27 Methyl-branched 5 4 5 1 ns/* ns/ns ns/ns ns 13 21.55 2MeC27 Methyl-branched 5 0 4 3 **/ns ns/ns ns/ns ns ‡ 14 21.85 n-C28 Alkane 0021 6 ND ND /ns * 15 22 MeC28 Methyl-branched 16 19 5 5 ns/* ns/ns ns/ns ns 16 22.5 C29:1 Alkene 10 15 23 10 ns/ns ns/ns ns/** * 17 22.74 n-C29 Alkane 25 33 00ns/ns ns/ns ND ND 18 23 MeC29 Methyl-branched 8 6 23 10 ns/ns */* ns/** ns 19 23.22/23.3 3MeC29 Methyl-branched 6 4 2 2 ns/** ns/*** ND ND 20 23.565 n-C30 Alkane 2 6 13 2 ns/ns ns/ns ns/ns ns 21 23.7 MeC30 Methyl-branched 8 12 1 1 ns/ns ns/ns ND ND ‡ 22 24.0/24.1 C31:1 Alkene 8 8 0 10 ns/ ns/ns ***/ND ND 23 24.4/24.3 n-C31 Alkane 12 11 23 10 ns/ns ns/ns ns/ns ** 24 24.56 MeC31 Methyl-branched 16 25 22 10 ns/ns ***/ns ns/** ns 25 24.6 2MeC31 Methyl-branched 009 1 ND ND ns/ns ns 26 25 n-C32 Alkane 8 8 21 9 ns/ns ***/ns ns/* ns 27 25.4 MeC32 Methyl-branched 6 8 14 7 ns/* ns/ns ns/** ns 28 26.1 n-C33 Alkane 0012 3 ND ND ns/ns *** 29 26.35 MeC33 Methyl-branched 25 31 21 10 ns/ns ns/ns ns/*** *** 30 26.69/26.6 2MeC33 Methyl-branched 13 19 21 7 ns/ns ns/ns ns/* ns 31 29 MeC35 Methyl-branched 6 10 20 8 ns/** ns/ns ns/ns ns 32 29.45/29.05 2MeC35 Methyl-branched 13 11 21 8 ns/ns ns/ns ns/ns ns 33 30.95 n-C37 Alkane 002 3 ND ND ND ND 34 31.54 2MeC37 Methyl-branched 0019 4 ND ND */ns ns 35 32.05 n-C38 Alkane 0013 3 ND ND ns/ns ns 36 33.5 n-C39 Alkane 006 5 ND ND ns/ns ns 37 33.9 MeC39 Methyl-branched 0014 7 ND ND ns/* ns 38 34.1/34.3 2MeC39 Methyl-branched 20 21 16 5 ns/ns ns/ns ns/ns ns No. of CHCs 30 29 33 34 Corresponding differences (P-values) in CHC abundance between treatments (wet/dry) and seasons (rainy/dry), as well as the effect of specific CHCs on desiccation tolerance. Sample sizes were Ndry=25 and Nwet=33 for the dry and wet laboratory treatments, and Ndry=23 and Nrainy=10 for the dry and rainy seasons, respectively. aSignificant difference measured by homogeneity test of contingency tables based on frequency followed by (after the ‘/’) univariate ANOVA on the log contrast transformed values of relative abundance. Black and red asterisks indicate significance in the univariate and multivariate tests. b Significance of the effects of individual CHCs on DT (and RWL, for the G3 colony) evaluated using ANOVA. Color and significance as above. Journal of Experimental Biology 1685 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665 The evidence that variation in CHC composition mediates DT in would a female A. coluzzii survive in the midst of the DS compared A. coluzzii is a prerequisite for serving as a ‘dual-role trait’ that might with her granddaughter during the RS if this was the primary lead to ecological speciation if it also affects mate recognition (Chung mechanism of her aridity tolerance? Notably, there was considerable and Carroll, 2015). Caputo and colleagues (2007) compared CHC overlap in the seasonal distributions of DT despite being different composition of female A. coluzzii, A. gambiae and A. arabiensis from statistically (Fig. 7). A similar magnitude of difference in DT was three villages in Burkina Faso and found differences mostly within recorded between various species and populations of A. gambiae s.l. taxa between villages rather than between taxa, suggesting (Gray and Bradley, 2005; Gray et al., 2009; Lee et al., 2009), which environmental constraints shaping variation in CHC composition. was presumably smaller than anticipated. Even in hibernating Culex Across villages, 2MebrC35 showed a degree of differentiation pipiens, reduction of RWL in diapausing females amounted to between A. arabiensis and A. gambiae, whereas MebrC29 showed a merely 46% of that of non-diapausing females (Benoit and degree of differentiation between A. coluzzii and A. gambiae.All Denlinger, 2007). Based on these considerations, we suspect that, mosquitoes used in their analyses were collected during the RS unless higher DT is manifested only by aestivating mosquitoes (October), precluding seasonal comparison. The qualitative absence while in hidden shelters (which are yet to be found), the of C29:1 from A. coluzzii in the DS shown in our study (above) stands physiologically elevated DT as measured here in A. coluzzii out as a markedly large seasonal effect on CHCs. We found no contributes less to its persistence throughout the 7 month-long DS variation in 2MebrC35, yet MebrC29, which was higher in than its behavior (Huestis and Lehmann, 2014). Unlike hibernating A. gambiae and A. arabiensis (Caputo et al., 2007), was elevated insects that withstand the winter mostly immobilized by near- during the RS (Table 4 and Fig. 10), consistent with A. coluzzii being freezing temperatures, A. coluzzii can move around and replenish its the most arid-tolerant in that region. body water content from blood meals, sugar sources and water in or The elevated total CHCs and change in CHC composition in the near wells or water pots in houses (although these water sources are mosquitoes exposed to dry versus humid treatments suggest a not suitable for reproduction). Their hitherto unknown refugia hardening response as described in several Drosophila species and might provide elevated RH and optimal temperatures (Huestis and differ from findings during acclimation (Stinziano et al., 2015). Lehmann, 2014; Kessler and Guerin, 2008). Despite the growing Within hours after recovery from high desiccation stress, Drosophila understanding of the mechanisms allowing their presence in arid melanogaster females exhibit changes in CHC profile (without conditions, integrating the physiological and behavioral adaptations increasing the total CHC quantity) that increase DT by decreasing of these species to an empirically demonstrated strategy remains to RWL by ∼30% (Bazinet et al., 2010; Kalra et al., 2014). In our assay, be achieved. mosquitoes were not allowed to recover, so the increased total CHCs (Fig. 5) may represent the initiation of the hardening response. Acknowledgements We thank Mr Andre Laughinghouse and Mr Kevin Lee for assisting with mosquito cultures, and Ms Margery Sullivan, Jenna Florio, Olga Muratova and Drs José DT of malaria mosquitoes and persistence in arid Ribeiro, Peter Armbruster and Roy Faiman for discussions on previous drafts of this environments paper. Dr Faiman provided the photograph in Fig. S1. We thank the residents of In the G3 mosquitoes, the ambient RH, RWL, BWA and, to a lesser Thierola and neighboring villages for welcoming us into their homes and treating us with so much warmth and generosity. degree, BWCD accounted for 82% of the variation in DT. Notably, these results suggest variability in mosquito capacity to withstand Competing interests the minimal concentration of water in the body in addition to RWL, The authors declare no competing or financial interests. whereas hydration at the onset of the assay was the only determinant detected by a previous study comparing DT in A. arabiensis and Author contributions T.L. conceived the study, which was designed by D.L.H., A.C.A., A.D., A.S.Y. and A. gambiae (Gray and Bradley, 2005). Our estimate of the effect size T.L. Experiments were carried out by A.C.A., D.L.H., A.D., A.S.Y., M.D., J.A. and of BWCD (including its interaction with RWL) was less than a third G.J.B.; data analysis was performed by T.L. and A.C.A., who also interpreted the of that of RWL (estimated by partitioning R2 among all factors of findings and drafted the article. All authors read and approved the final version of the the model, not shown). The significant effect of wet mass on DT, as manuscript. opposed to that of end mass and dry mass, and its high correlation Funding with RWL suggest that hydration at the onset of the desiccation This study was supported (in part) by the Intramural Research Program of the assay was the primary factor determining the mosquito DT (aside National Institutes of Health, National Institute of Allergy and Infectious Diseases; from the assay RH). As the hydration level of a laboratory mosquito and by a grant from the National Institute of General Medical Sciences reflects merely the time and size of the last sugar meal(s) it took (P20GM103440). Deposited in PMC for immediate release. prior to the assay, rather than an inherent difference in body water Supplementary information content, we conclude that the G3 colony maintains low innate Supplementary information available online at variation in DT, possibly explaining why variation in spiracle size http://jeb.biologists.org/lookup/suppl/doi:10.1242/jeb.135665/-/DC1 had negligible effect on it. During the 7 month-long DS in the Sahel, when RH seldom rises References Adamou, A., Dao, A., Timbine, S., Kassogué, Y., Yaro, A. S., Diallo, M., Traoré, above 25%, survival for small terrestrial insects must pose an S. F., Huestis, D. L. and Lehmann, T. (2011). The contribution of aestivating enormous challenge (Huestis and Lehmann, 2014). Because the mosquitoes to the persistence of Anopheles gambiae in the Sahel. Malar. J. 10,151. field desiccation assay was performed at <10% RH, as opposed to Aitchison, J. (1986). The Statistical Analysis of Compositional Data. London, UK: ∼25% RH of the laboratory assay, we cannot compare the DT of Chapman and Hall. Artis, M. L., Huestis, D. L. and Lehmann, T. (2014). The effects of oviposition-site A. coluzzii and the G3 colony, yet the difference is probably not deprivation on longevity and bloodfeeding rate in Anopheles gambiae. Parasit measured in orders of magnitude (Table S1). The higher DT of Vectors 7, 163. A. coluzzii during the DS may be considered as evidence for its Bayoh, M. N., Thomas, C. J. and Lindsay, S. W. (2001). Mapping distributions of importance during aestivation. However, if this elevated DT is a chromosomal forms of Anopheles gambiae in West Africa using climate data. Med. Vet. Entomol. 15, 267-274. primary component of aestivation, could it be manifested by a mere Bazinet, A. L., Marshall, K. E., MacMillan, H. A., Williams, C. M. and Sinclair, 40% difference (rather than, say, 400%)? How many days longer B. J. (2010). Rapid changes in desiccation resistance in Drosophila melanogaster Journal of Experimental Biology 1686 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665 are facilitated by changes in cuticular permeability. J. Insect Physiol. 56, Holm, S. (1979). A simple sequentially rejective multiple test procedure. 2006-2012. Scand. J. Stat. 6, 65-70. Benoit, J. B. and Denlinger, D. L. (2007). Suppression of water loss during adult Howard, R. W. and Blomquist, G. J. (2005). Ecological, behavioral, and diapause in the northern house mosquito, Culex pipiens. J. Exp. Biol. 210, biochemical aspects of insect hydrocarbons. Annu. Rev. Entomol. 50, 371-393. 217-226. Huestis, D. L. and Lehmann, T. (2014). Ecophysiology of Anopheles gambiae s.l.: Benoit, J. B. and Denlinger, D. L. (2010). Meeting the challenges of on-host and persistence in the Sahel. Infect Genet. Evol. 28, 648-661. off-host water balance in blood-feeding arthropods. J. Insect Physiol. 56, Huestis, D. L., Yaro, A. S., Traore, A. I., Adamou, A., Kassogue, Y., Diallo, M., 1366-1376. Timbine, S., Dao, A. and Lehmann, T. (2011). Variation in metabolic rate of Caputo, B., Dani, F. R., Horne, G. L., Petrarca, V., Turillazzi, S., Coluzzi, M., Anopheles gambiae and A. arabiensis in a Sahelian village. J. Exp. Biol. 214, Priestman, A. A. and della Torre, A. (2005). Identification and composition of 2345-2353. cuticular hydrocarbons of the major Afrotropical malaria vector Anopheles Huestis, D. L., Yaro, A. S., Traore, A. I., Dieter, K. L., Nwagbara, J. I., Bowie, gambiae s.s. (Diptera: Culicidae): analysis of sexual dimorphism and age- A. C., Adamou, A., Kassogue, Y., Diallo, M., Timbine, S. et al. (2012). Seasonal related changes. J. Mass Spectrom. 40, 1595-1604. variation in metabolic rate, flight activity and body size of Anopheles gambiae in Caputo, B., Dani, F. R., Horne, G. L., N’Fale, S., Diabate, A., Turillazzi, S., the Sahel. J. Exp. Biol. 215, 2013-2021. Coluzzi, M., Costantini, C., Priestman, A. A., Petrarca, V. et al. (2007). Hugo, L. E., Kay, B. H., Eaglesham, G. K., Holling, N. and Ryan, P. A. (2006). Comparative analysis of epicuticular lipid profiles of sympatric and allopatric field Investigation of cuticular hydrocarbons for determining the age and survivorship of populations of Anopheles gambiae s.s. molecular forms and An. arabiensis from Australasian mosquitoes. Am. J. Trop. Med. Hyg. 74, 462-474. Burkina Faso (West Africa). Insect Biochem. Mol. Biol. 37, 389-398. Kalra, B., Parkash, R. and Aggarwal, D. D. (2014). Divergent mechanisms for Chown, S. and Nicolson, S. W. (2004). Insect Physiological Ecology: Mechanisms water conservation in Drosophila species. Entomol. Exp. Appl. 151, 43-56. and Patterns. Oxford: Oxford University Press. Kessler, S. and Guerin, P. M. (2008). Responses of Anopheles gambiae, Chung, H. and Carroll, S. B. (2015). Wax, sex and the origin of species: dual roles Anopheles stephensi, Aedes aegypti, and Culex pipiens mosquitoes (Diptera: of insect cuticular hydrocarbons in adaptation and mating. Bioessays 37, 822-830. Culicidae) to cool and humid refugium conditions. J. Vector Ecol. 33, 145-149. Coetzee, M., Hunt, R. H., Wilkerson, R., della Torre, A., Coulibaly, M. B. and Krafsur, E. S. (1971). Behavior of thoracic spiracles of Aedes mosquitoes in Besansky, N. J. (2013). Anopheles coluzzii and Anopheles amharicus, new controlled relative humidities. Ann. Entomol. Soc. Am. 93-97. members of the Anopheles gambiae complex. Zootaxa 3619, 246-274. Kwan, L. and Rundle, H. D. (2010). Adaptation to desiccation fails to generate pre- Coluzzi, M. (1982). Spatial distribution of chromosomal inversions and speciation in and postmating isolation in replicate Drosophila melanogaster laboratory Anopheline mosquitoes. Prog. Clin. Biol. Res. 96, 143-153. populations. Evolution 64, 710-723. Coluzzi, M., Sabatini, A., Petrarca, V. and Di Deco, M. A. (1979). Chromosomal Lee, Y., Meneses, C. R., Fofana, A. and Lanzaro, G. C. (2009). Desiccation differentiation and adaptation to human environments in the Anopheles gambiae resistance among subpopulations of anopheles gambiae s.s. from selinkenyi, complex. Trans. R. Soc. Trop. Med. Hyg. 73, 483-497. Mali. J. Med. Entomol. 46, 316-320. Coluzzi, M., Petrarca, V. and di Deco, M. A. (1985). Chromosomal inversion Lehmann, T. and Diabate, A. (2008). The molecular forms of Anopheles gambiae: intergradation and incipient speciation in Anopheles gambiae. Boll. Zool. 52, A phenotypic perspective. Infect Genet. Evol. 8, 737-746. 45-63. Lehmann, T., Dao, A., Yaro, A. S., Adamou, A., Kassogue, Y., Diallo, M., Sekou, ́ ́ ́ Dao, A., Yaro, A. S., Diallo, M., Timbine, S., Huestis, D. L., Kassogue,Y.,Traore, T. and Coscaron-Arias, C. (2010). Aestivation of the African Malaria Mosquito, ́ A. I., Sanogo, Z. L., Samake, D. and Lehmann, T. (2014). Signatures of aestivation Anopheles gambiae in the Sahel. Am. J. Trop. Med. Hyg. 83, 601-606. and migration in Sahelian malaria mosquito populations. Nature 516, 387-390. Lyons, C. L., Coetzee, M., Terblanche, J. S. and Chown, S. L. (2014). Desiccation della Torre, A., Costantini, C., Besansky, N. J., Caccone, A., Petrarca, V., tolerance as a function of age, sex, humidity and temperature in adults of the Powell, J. R. and Coluzzi, M. (2002). Speciation within Anopheles gambiae–the African malaria vectors Anopheles arabiensis and Anopheles funestus. J. Exp. glass is half full. Science 298, 115-117. Biol. 217, 3823-3833. Denlinger, D. L. (2002). Regulation of diapause. Annu. Rev. Entomol. 47, 93-122. Mamai, W., Mouline, K., Blais, C., Larvor, V., Dabiré,K.R.,Ouedraogo, Denlinger, D. L. and Armbruster, P. A. (2014). Mosquito diapause. Annu. Rev. G. A., Simard, F. and Renault, D. (2014). Metabolomic and ecdysteroid Entomol. 59, 73-93. variations in Anopheles gambiae s.l. mosquitoes exposed to the stressful Fanello, C., Santolamazza, F. and della Torre, A. (2002). Simultaneous conditions of the dry season in Burkina Faso, West Africa. Physiol. Biochem. identification of species and molecular forms of the Anopheles gambiae Zool. 87, 486-497. complex by PCR-RFLP. Med. Vet. Entomol. 16, 461-464. Mamai, W., Mouline, K., Parvy, J.-P., Le Lannic, J., Dabiré, K. R., Ouédraogo, Fouet, C., Gray, E., Besansky, N. J. and Costantini, C. (2012). Adaptation to G. A., Renault, D. and Simard, F. (2016). Morphological changes in the spiracles aridity in the malaria mosquito Anopheles gambiae: chromosomal inversion of Anopheles gambiae s.l (Diptera) as a response to the dry season conditions in polymorphism and body size influence resistance to desiccation. PLoS ONE 7, e34841. Burkina Faso (West Africa). Parasit Vectors 9, 11. Gibbs, A. and Pomonis, J. G. (1995). Physical properties of insect cuticular Miller, P. L. (1960). Respiration in the desert locust: II. The control of the spiracles. hydrocarbons: the effects of chain length, methyl-branching and unsaturation. J. Exp. Biol. 37, 237-263. ́ Comp. Biochem. Physiol. B Biochem. Mol. Biol. 112, 243-249. Milligan, P. J. M., Phillips, A., Broomfield, G., Molyneux, D. H., Toure,Y.T.and Gibbs, A. G. and Rajpurohit, S. (2010). Cuticular lipids and water balance. In Insect Coluzzi, M. (1993). A study of the use of gas chromatography of cuticular Hydrocarbons: Biology, Biochemistry and Chemical Ecology (ed. G. J. Blomquist hydrocarbons for identifying members of the Anopheles gambiae (Diptera: and A. G. Bagneres), pp. 100-120. Cambridge: Cambridge University Press. Culicidae) complex. Bull. Entomol. Res. 83, 613-624. Gibbs, A. G., Chippindale, A. K. and Rose, M. R. (1997). Physiological Nagpal, B. N., Srivastava, A., Kalra, N. L. and Subbarao, S. K. (2003). Spiracular mechanisms of evolved desiccation resistance in Drosophila melanogaster. indices in Anopheles stephensi: a taxonomic tool to identify ecological variants. J. Exp. Biol. 200, 1821-1832. J. Med. Entomol. 40, 747-749. Gibbs, A. G., Fukuzato, F. and Matzkin, L. M. (2003). Evolution of water Page, M., Nelson, L. J., Haverty, M. I. and Blomquist, G. J. (1990). Cuticular conservation mechanisms in Drosophila. J. Exp. Biol. 206, 1183-1192. Hydrocarbons as Chemotaxonomic Characters for Bark Beetles: Dendroctonus Gray, E. M. and Bradley, T. J. (2005). Physiology of desiccation resistance in ponderosae, D. jeffreyi, D. brevicomis, and D. frontalis (Coleoptera: Scolytidae). Anopheles gambiae and Anopheles arabiensis. Am. J. Trop. Med. Hyg. 73, Ann. Entomol. Soc. Am. 83, 892-901. 553-559. Polerstock, A. R., Eigenbrode, S. D. and Klowden, M. J. (2002). Mating alters the Gray, E. M. and Bradley, T. J. (2006). Evidence from mosquitoes suggests that cuticular hydrocarbons of female Anopheles gambiae sensu stricto and Aedes cyclic gas exchange and discontinuous gas exchange are two manifestations of a aegypti (Diptera: Culicidae). J. Med. Entomol. 39, 545-552. single respiratory pattern. J. Exp. Biol. 209, 1603-1611. SAS, I., Inc. (2011). SAS for Windows Version 9.3. Cary, NC: Sas Institute. Gray, E. M., Rocca, K. A., Costantini, C. and Besansky, N. J. (2009). Inversion Simard, F., Ayala, D., Kamdem, G. C., Pombi, M., Etouna, J., Ose, K., Fotsing, 2La is associated with enhanced desiccation resistance in Anopheles gambiae. J.- M., Fontenille, D., Besansky, N. J. and Costantini, C. (2009). Ecological Malar. J. 8, 215. niche partitioning between Anopheles gambiae molecular forms in Cameroon: the Hadley, N. F. (1994). Ventilatory patterns and respiratory transpiration in adult ecological side of speciation. BMC Ecol. 9, 17. terrestrial insects. Physiol. Zool. 67, 175-189. Snodgrass, R. E. (1959). The anatomical life of the mosquito. Smithson. Misc. Haverty, M. I., Thorne, B. L. and Escoubas, P. (1991). Cuticular hydrocarbons: Collect. 139, 87. species and population-level discrimination in termites. USDA For. Serv. Gen. Stinziano, J. R., Sové, R. J., Rundle, H. D. and Sinclair, B. J. (2015). Tech. Rep. PSW 12, 15-23. Rapid desiccation hardening changes the cuticular hydrocarbon profile of Herreid, C. F. and Fourtner, C. R. (1981). Locomotion and Energetics in Drosophila melanogaster. Comp. Biochem. Physiol. A Mol. Integr. Physiol. 180, Arthropods. New York: Plenum Press. 38-42. Hidalgo, K., Mouline, K., Mamai, W., Foucreau, N., Dabiré, K. R., Bouchereau, Toolson, E. C. and Hadley, N. F. (1987). Energy-Dependent facilitation of A., Simard, F. and Renault, D. (2014). Novel insights into the metabolic and transcuticular water flux contributes to evaporative cooling in the sonoran biochemical underpinnings assisting dry-season survival in female malaria desert cicada, Diceroprocta apache (Homoptera: Cicadidae). J. Exp. Biol. mosquitoes of the Anopheles gambiae complex. J. Insect Physiol. 70, 102-116. 131, 439-444. Journal of Experimental Biology 1687 RESEARCH ARTICLE Journal of Experimental Biology (2016) 219, 1675-1688 doi:10.1242/jeb.135665 Toure, Y. T., Petrarca, V., Traoré, S. F., Coulibaly, A., Maïga, H. M., Sankaré, O., dry- and wet-season conditions in female Anopheles gambiae mosquitoes. Sow, M., Di Deco, M. A. and Coluzzi, M. (1994). Ecological genetic studies in the Parasit Vectors 7, 294. chromosomal form Mopti of Anopheles gambiae s.str. in Mali, West Africa. WHO (2014). World Malaria Report 2014, pp. 178. Geneva, Switzerland: World Genetica 94, 213-223. Health Organization. Toure, Y. T., Petrarca, V., Traore, S. F., Coulibaly, A., Maiga, H. M., Sankare, O., Wigglesworth, V. B. (1945). Transpiration through the cuticle of insects. J. Exp. Sow, M., Di Deco, M. A. and Coluzzi, M. (1998). The distribution and inversion Biol. 21, 97-114. polymorphism of chromosomally recognized taxa of the Anopheles gambiae Wigglesworth, V. B. (1933). The physiology of the cuticle and of ecdysis in complex in Mali, West Africa. Parassitologia 40, 477-511. Rhodnius prolixus (Triatomidae, Hemiptera); with special reference to the function Vinogradskaia, O. N. and Detinova, T. S. (1978). [Value of the spiracular index of the oenocytes and of the dermal glands. Q. J. Microsc. Sci. 76, 269-318. as an indicator of the hygrophilic capacity and xeroresistance of Yaro, A. S., Dao, A., Adamou, A., Crawford, J. E., Traoré, S. F., Touré, A. M., Anopheles. 1. Mosquitoes of Africa]. Med. Parazitol. 47, 40-45. Gwadz, R. and Lehmann, T. (2006). Reproductive output of female Anopheles Wagoner, K. M., Lehmann, T., Huestis, D. L., Ehrmann, B. M., Cech, N. B. and gambiae (Diptera: Culicidae): comparison of molecular forms. J. Med. Entomol. Wasserberg, G. (2014). Identification of morphological and chemical markers of 43, 833-839. Journal of Experimental Biology 1688