Divergence of desiccation-related traits in two Drosophila species of the takahashii subgroup from the western Himalayas

Water conservation is crucial to the ecological success of diverse insect taxa as well as terrestrial arthropods (Edney, 1977; Hadley, 1994). The ability to maintain water balance is associated with species distribution patterns (Willmer et al., 2000). Several studies have shown substantial variation in desiccation resistance of Drosophila species living in different habitats, i.e. xeric versus mesic species (Gibbs and Matzkin, 2001; Gibbs et al., 2003). A major conclusion of these studies is that insects can adapt to variation in water availability. Insects can increase their desiccation resistance through three different avenues of water balance: (1) higher bulk water, (2) reduced rate of water loss and (3) greater dehydration tolerance (Hadley, 1994; Gibbs et al., 1997). In addition to physiological adaptations, behavioral responses such as clustering in bed bugs (Climax lectularius) represent a group effect for water conservation (Benoit et al., 2007a). Insects with higher initial body water content can survive longer under arid conditions, e.g. laboratory-selected desiccation-resistant strains of D. melanogaster have shown a 300% increase in hemolymph water content compared with control strains (Folk et al., 2001). If laboratory selection responses result in increases in the bulk water content, we may expect similar changes in wild-caught flies of diverse Drosophila species. However, comparative studies of diverse Drosophila species from subtropical habitats have generally not considered this aspect of water balance (Parkash and Munjal, 1999). Thus, it would be interesting to compare the water budget of Drosophila species differing in their desiccation resistance levels.

Most arthropods can tolerate ∼30–50% loss of body water, but some taxa that are adapted to drier habitats have exhibited higher dehydration tolerance, i.e. ∼40–60% body water loss, before succumbing to death (Hadley, 1994; Willmer et al., 2000; Benoit et al., 2005). In contrast, comparative studies have found little consistent evidence that desert insects are able to tolerate lower water levels (dehydration tolerance) or store more water (Hadley, 1994; Gibbs and Matzkin, 2001; Gibbs et al., 2003). For example, enhanced desiccation resistance in cactophilic Drosophila species from North American deserts is associated with reduced rates of water loss despite lack of changes in dehydration tolerance (Gibbs and Matzkin, 2001). However, few studies have tested the generality of such findings in diverse Drosophila species from other continents. Further, changes in epicuticular lipid mass and/or body melanisation have been shown to be negatively correlated with cuticular water loss (Parkash et al., 2008a; Parkash et al., 2008b). For drosophilids, it is not clear whether changes in rate of water loss mainly affect the ability to cope with drier habitats or whether other avenues of water balance have also evolved under contrasting climatic conditions.

There is evidence that acquisition of carbohydrates as energy reserves alleviates the consequences of desiccation stress in laboratory-selected desiccation-resistant strains of D. melanogaster (Graves et al., 1992; Gibbs et al., 1997; Chippindale et al., 1998; Djawdan et al., 1998). In contrast, large-sized insects (locusts and tse-tse flies) and a new set of laboratory-selected desiccation-resistant strains of D. melanogaster have shown storage of higher levels of lipid content under dehydration stress (Loveridge and Bursell, 1975; Telonis-Scott et al., 2006). Further, a single study has shown that utilization of energy metabolites (carbohydrates and lipids) is a function of different durations of desiccation stress for three mesic and two desert Drosophila species (Marron et al., 2003). However, storage and utilization of energy metabolites for conferring desiccation resistance in wild populations of diverse Drosophila species of the subgenus Sophophora are largely unknown.

Several studies have shown beneficial effects of cold stress acclimation in diverse insect taxa, but acclimation to desiccation stress has received less attention (Bale, 2002; Hoffmann et al., 2003). Enhanced drought tolerance has been reported in a soil-dwelling springtail by pre-acclimation to a mild drought stress (Sjursen et al., 2001). Drought acclimation has also been found to be beneficial to the arctic collembolan Onychiurus arcticus (Holmstrup and Somme, 1998), and to Folsomia candida (Holmstrup et al., 2002), Belgica antarctica (Benoit et al., 2007b) and Cryptophgus antarcticus (Elnitsky et al., 2008). In contrast, a single study has shown increased desiccation resistance in females of four Drosophila species due to dehydration acclimation (Hoffmann, 1991). In that study, desiccation acclimation ability was also associated with species distribution patterns; D. birchii, a rainforest Drosophila species, lacked an acclimation response whereas species with widespread distribution showed acclimation effects. However, adaptive changes due to acclimation in diverse Drosophila species remain unknown.

In the present work, we have focused on Drosophila nepalensis Okada 1955 and Drosophila takahashii Sturtevant 1927, two species that are co-distributed along an altitudinal gradient in the western Himalayas but that differ in their abundance level. Drosophila nepalensis was first described from collections made from Nepal by Okada (Okada, 1955). Subsequently, Parshad and Paika (Parshad and Paika, 1964) reported association between relative abundance of D. nepalensis and lower temperature (T_ave=18.20°C) as well as humidity (∼43–45% relative humidity) at Manali (2050 m). Thus we expect that these two species will differ in their desiccation-related traits; therefore, we compared D. nepalensis and D. takahashii for evolved physiological mechanisms that may affect their adaptations to drier climatic conditions. We tested three different routes of water balance that may result in interspecific differences of desiccation resistance. We examined the role of cuticular components (cuticular lipid mass and body melanisation) for interspecific differences in desiccation resistance. We also analysed the relationship between desiccation resistance and dehydration tolerance under different humidity conditions. We assessed whether higher desiccation potential is associated with greater storage of energy metabolites. Finally, we examined the desiccation acclimation potential of the two species.

Sympatric populations of D. nepalensis and D. takahashii (N=150–200 flies from each site) were collected in a single trip during autumn in October 2008 from five altitudinal localities of the western Himalayas (Fig. 1). Wild-caught individuals of a midland locality (Solan, 1440 m; 30.55°N) were used to initiate 20 isofemale lines (for all analyses, 20 lines with 10 replicates each were used unless otherwise indicated). All cultures were maintained at low density (60–70 eggs per vial; eggs were 40×100 mm in size) on cornmeal-yeast-agar medium at 21°C and 65±1% relative humidity in a temperature- and humidity-controlled incubator for five generations before experimental analysis. All assays were performed on 8-day-old flies because the trait values did not vary as a function of age between 6 and 21 days (Gibbs and Matzkin, 2001; Parkash et al., 2008a). Climatic data for thermal variables and relative humidity were obtained from the Indian Meteorological Department, Government of India, New Delhi (Indian Meteorological Department, 2010). Percent abundance was estimated as the number of individuals of a particular Drosophila species divided by the total number of individuals of all the different Drosophila species in the samples collected from a given locality.

For both species, body melanisation of individual female flies was visually scored with an Olympus stereo-zoom microscope SZ-61 (www.olympus.com) from the dorsal and lateral views of the female abdomen, giving values ranging from 0 (no melanisation) to 10 (complete melanisation) for each of the six abdominal segments (second to seventh). Further, the relative size of each abdominal segment was calculated in proportion to the largest (fourth) abdominal segment, which was assigned a value of 1.0. Because the abdominal segments differ in size, these relative sizes (i.e. 0.86, 0.94, 1.0, 0.88, 0.67 and 0.38 for the second to seventh segments, respectively) were multiplied with segment-wise melanisation scores. Data on percent melanisation were calculated as the sum of the observed weighted melanisation scores of abdominal segments per fly divided by the sum of the relative size of each abdominal segment multiplied by 10 per fly, times 100 (Parkash et al., 2008a).

Cuticular lipid mass was estimated individually on 8-day-old flies of each species following the Gibbs method (Gibbs et al., 2003). Flies were dried overnight at 60°C to obtain measurements of dry mass. Each dried fly was kept in HPLC-grade hexane in a 2 ml Eppendorf tube (www.tarson.com) for 1 h; thereafter it was removed from the solvent and was again dried at room temperature and finally reweighed on a Sartorius microbalance (model CPA26P, 0.001 mg precision; www.sartorius.com). Cuticular lipid mass per cm² was calculated as the difference in mass following solute extraction divided by the surface area (cm²). The surface area was calculated by following Edney's formula: 12M^0.67, where M is wet mass (Edney, 1977).

Desiccation resistance was measured as the time to lethal dehydration (LT₁₀₀) under dry air. Ten batches of 10 female individuals were isolated in individual dry plastic vials (40×100 mm) with 2 g of silica gel at the bottom and were covered with a foam disc. The vials were then placed in a desiccator chamber (Secador electronic desiccator cabinet; www.tarsons.in) that maintains 0–6% relative humidity. The number of immobile flies was counted after every 1 h interval, and LT₁₀₀ values in dry air were recorded. We pooled data on isofemale lines for survival curve analysis.

To estimate total body water content and dehydration tolerance (%), 10 flies of each isofemale line were used. First, individual flies were weighed on Sartorius microbalance (model CPA26P, 0.001 mg precision) and then reweighed after drying at 60°C overnight. Total body water content was estimated as the difference between masses before and after drying at 60°C. Further, after mild anesthesia (1 min) with solvent ether, flies were weighed on a Sartorius microbalance both before and after desiccation stress until death. Dehydration tolerance was estimated as the percentage of total body water lost until death due to desiccation, and was calculated by the formula: (wet body mass – body mass at death)/(wet body mass – dry body mass) × 100 (Gibbs et al., 1997).

For calculation of the rate of water loss in D. nepalensis and D. takahashii, we followed method of Wharton (Wharton, 1985), modified by Benoit et al. (Benoit et al., 2005) and Yoder et al. (Yoder et al., 2009). Total body water content (m) was calculated as the difference between wet or fresh (f) and dry mass (d), i.e. m=f–d. Individual flies were weighed and placed at 0a_v (a_v=percent relative humidity/100) for a specified time at 1 h intervals (1 to 8 h) and reweighed. The rate of water loss was derived from the slope of regression line on a plot of ln(m_t/m₀) against time according to Wharton's exponential equation (Wharton, 1985) m_t=m₀e^–k_t, where m_t is the water lost at time t, and m₀ is the initial water content. Rate (k_t) is the slope of the regression line and expressed as% per hour.

To measure acclimation pre-treatment duration, 10 female individuals of each replicate were subjected to desiccation stress at ∼0–5% relative humidity. The initial body water content in replicate groups was recorded. The time period in which flies lost ∼15–17% body water was defined as the pre-treatment duration. Further, for the recovery period, individuals were placed on non-nutritive agar and tested at hourly intervals for increase in body water until the lost body mass was regained. Such flies were subjected to desiccation stress until death to test the increased desiccation resistance due to acclimation. Increased desiccation survival (h) was calculated after subtracting the desiccation resistance (h) of non-acclimated (control) from acclimated individuals. Control and treatment experiments were run simultaneously under identical experimental conditions.

We tested the changes in desiccation resistance, dehydration tolerance and rate of water loss due to variation in humidity level (0, 22, 33 and 45% relative humidity) in D. nepalensis and D. takahashii. Different relative humidity conditions were maintained by using saturated solutions of calcium sulphate for 0a_v (Toolson, 1978), and potassium acetate, magnesium chloride and potassium carbonate for 22, 33 and 45% relative humdity, respectively, as suggested by Winston and Bates (Winston and Bates, 1960). The relative humidity was checked regularly with a dial hygrometer (model 4186, Control Company, www.control3.com).

Individual flies were carefully pinned to a microdissection dish at its anterior and posterior ends with microdissection pins, and a narrow incision was made through the cuticle with a third pin while visually observing through a stereo-zoom microscope. The leaking extractable hemolymph was absorbed with an absorbent tissue moistened with an isotonic saline solution (Folk et al., 2001). Extractable hemolymph content was estimated as reduction in mass following hemolymph blotting (Cohen et al., 1986; Hadley, 1994). Further, tissue water was estimated after subtracting exsanguinated mass before and after drying. From same data, we also calculated hemolymph water content by subtracting tissue water from total body water content.

Individual virgin adult female flies were dried in 2 ml Eppendorf tubes (www.tarsons.in) at 60°C for 48 h and then weighed on a Sartorius microbalance (model CPA26P, 0.001 mg precision). Thereafter, 1.5 ml di-ethyl ether was added in each eppendorf tube and kept for 24 h under continuous shaking (200 r.p.m.) at 37°C. Finally, the solvent was removed and flies were again dried at 60°C for 24 h and reweighed. Lipid content was calculated per individual fly by subtracting the lipid free dry mass from initial dry mass per fly (Hoffmann et al., 2001).

We tested the assay sensitivity by measuring cuticular lipids as well as total body lipids in the replicate samples. For removal of cuticular lipids, we treated flies with hexane (without shaking). However, for total body lipids, we first removed cuticular lipids followed by 24 h treatment with di-ethyl ether with continuous shaking at 200 r.p.m. We did not find a difference in the estimates of either cuticular lipids or total body lipids under our assay conditions.

For trehalose and glycogen content estimation, 10 flies of each isofemale line were homogenized in a homogenizer (Labsonic M, www.sartorius.com) with 300 μl Na₂Co₃ and incubated at 95°C for 2 h to denature proteins. An aqueous solution of 150 μl acetic acid (1 mol l^–1) and 600 μl sodium acetate (0.2 mol l^–1) was mixed with the homogenate. The homogenate was subsequently centrifuged (Fresco 21, ThermoFisher Scientific, Pittsburgh, PA, USA) at 12,000 r.p.m. (9660 g) for 10 min. This homogenate was used for independent estimations of trehalose and glycogen as given below.

For trehalose estimation, aliquots (200 μl) were placed in two different tubes; one was taken as a blank whereas the other was digested with trehalase at 37°C using the Megazyme trehalose assay kit (K-Treh 10/10, www.megazyme.com). In this assay, released d-glucose was phosphorylated by hexokinase and ATP to glucose-6-phosphate and ADP, which was further coupled with glucose-6-phosphate dehydrogenase and resulted in the reduction of nicotinamide adenine dinucleotide (NAD). The absorbance by NADH was measured at 340 nm (UV-2450-VIS, Shimadzu Scientific Instruments, Columbia, MD, USA). The pre-existing glucose level in the sample was determined in a control reaction lacking trehalase and subtracted from total glucose concentration.

For estimation of glycogen, a 50 μl aliquot was incubated with 500 μl Aspergillus niger glucoamylase solution (8.7 U ml^–1 in 200 mmol l^–1 of acetate buffer) for 2 h at 40°C with constant agitation and the suspension was centrifuged at 4000 r.p.m. (1073g) for 5 min. It mainly hydrolyzed alpha-(1,4) and alpha-(1,6) glycosyl linkages and was suited for breakdown of glycogen. Glucose concentration was determined with 20 μl of supernatant from the suspension and added to 170 μl of a mixture of G6-DPH (0.9 U ml^–1), ATP (1.6 mmol l^–1) and NADP (1.25 mmol l^–1) in triethanolamine hydrochloride buffer (380 mmol l^–1 TEA–HCl and 5.5 mmol l^–1 of MgSO4) and 10 μl of hexokinase solution (32.5 U ml^–1 in 3.2 mol l^–1 ammonium sulphate buffer), and absorbance was measured at 340 nm.

Protein levels were determined using the bicinchoninic acid (BCA) method as followed by Marron and coworkers (Marron et al., 2003). For the protein assay, 10 virgin female flies per isofemale line were homogenized in 3 ml distilled water and centrifuged at 10,000 r.p.m. for 5 min. Further, a 50 μl aliquot was taken from the supernatant and treated with 2 ml BCA reagent (Sigma-Aldrich, Bangalore, India) and incubated at 25°C for 12 h. Absorbance was recorded at 562 nm and protein concentration was determined by comparison with a standard curve.

We measured each metabolite (carbohydrates, lipids or proteins) in 10 replicate sets of 20 isofemale lines before and after its utilization under desiccation stress until death in D. nepalensis and D. takahashii. Flies were subjected to different durations of desiccation stress (at 5 h intervals). In addition, total energy budget was calculated using standard conversion factors following Schmidt-Nielsen (Schmidt-Nielsen, 1990).

For each trait, population means (20 isofemale lines, 10 replicates each) are presented ±s.e.m. Results of analysis of co-variance (ANCOVA; with body mass as a covariate) were used to compare different ecophysiological traits as well as energy metabolites in D. nepalensis and D. takahashii. We used nested ANOVA to assess significant changes in the levels of energy metabolites (carbohydrates, lipids or protein) due to their utilization during desiccation stress until death. Total energy budget in D. nepalensis and D. takahashii due to differential storage of energy metabolites was calculated using standard conversion factors (Schmidt-Nielsen, 1990; Marron et al., 2003). Pearson correlations were based on 10 isofemale lines (10 replicates each). For correlations between desiccation resistance and dehydration tolerance and rate of water loss at different humidity levels (0, 22, 33 and 45% relative humdity), we standardized data for each trait by dividing each individual value by the overall mean. Further, increases in desiccation resistance and dehydration tolerance in desiccation-acclimated flies compared with non-acclimated (control) flies were analyzed using ANOVA. Results were deemed significant at P<0.05. Statistica (release 5.0, Statsoft Inc., Tulsa, OK, USA) was used for calculations as well as figures.

Data on percent abundance of wild-caught flies of D. nepalensis and D. takahashii from five altitudinal localities (512–1951 m) as a function of changes in relative humidity of sites of origin of populations are shown in Fig. 1A,B. Drosophila nepalensis is more abundant (30–40%) in highland localities but occurs less frequently in low to mid altitudinal localities. In contrast, D. takahashii is more abundant in lowland localities. The highland localities are moderately colder and drier (T_ave=15.2°C, 42.5% relative humidity) whereas the lowland localities are warmer with higher relative humidity (T_ave=27.6°C, 62.8% relative humidity). Thus, significant reductions in T_ave (∼2°C per 200 m) and relative humidity (∼3.9% per 200 m) along an elevational gradient may act as selection agents, affecting species abundance.

Fig. 2 illustrates a comparison of desiccation-related traits in D. nepalensis and D. takahashii. For both species grown at 21°C, there is lack of differences in cuticular trait values (body melanisation and cuticular lipid mass; Fig. 2A). In contrast, we found significant differences in the desiccation resistance of these two species (Fig. 3B) but a modest difference in their rate of water loss, i.e. 0.3% h^–1 (1.5% h^–1 in D. nepalensis versus 1.8% h^–1 in D. takahashii; Fig. 3C). However, the level of dehydration tolerance is much higher in D. nepalensis (82.32%) than D. takahashii (50%; Fig. 3D).

Data on trait variability in isofemale lines were subjected to ANCOVA and the results are shown in Table 1. Between-species differences for cuticular traits (body melanisation: F=0.84, P=0.51; cuticular lipids: F=0.52, P=0.63) and two energy metabolites (body lipids: F=3.92, P=0.28; proteins: F=3.87, P=0.31) were not significant. In contrast, we found significant F-values for interspecific differences for desiccation resistance (F=9445.14, P<0.001), dehydration tolerance (F=6937.35, P<0.001), hemolymph content (F=9850.04, P<0.001) and carbohydrate content (F=13921.10, P<0.001; Table 1).

Data on comparison of basic measures of total water content, hemolymph content and tissue water are given in Table 2. For each trait, means (±s.e.m.) of 20 isofemale lines per species are shown along with differences between these two species. Wet and dry mass and total body water content were ∼10–14% higher in D. nepalensis compared with D. takahashii (Table 2, Fig. 3A). However, total hemolymph content as well as hemolymph water content were significantly higher (∼2.4-fold) in D. nepalensis than in D. takahashii (Fig. 3B). Interestingly, D. nepalensis can sustain a greater loss of body water (0.875±0.009 mg fly^–1) before succumbing to death than D. takahashii (0.472±0.004 mg fly^–1). Thus, there is much higher dehydration tolerance (82.32%) in D. nepalensis than D. takahashii (50.78%; Table 2). Furthermore, higher levels of hemolymph and dehydration tolerance in D. nepalensis showed a significant positive correlation (r=0.92±0.06, P<0.001; Fig. 3C). A similar positive correlation was evident between lower levels of hemolymph and dehydration tolerance in D. takahashii (r=0.76±0.09, P<0.001; Fig. 3C).

Correlations (based on 10 isofemale lines, 10 replicates each) between desiccation resistance and total hemolymph content, tissue water content or carbohydrate content for D. nepalensis and D. takahashii are shown in Fig. 4. For each trait, mean values were significantly higher in D. nepalensis than in D. takahashii. For both species, desiccation resistance was positively correlated with hemolymph content (D. nepalensis: r=0.90±0.08, P<0.001; D. takahashii: r=0.75±0.16, P<0.001; Fig. 4A) as well as carbohydrate content (D. nepalensis: r=0.95±0.05, P<0.001; D. takahashii: r=0.79±0.11, P<0.001; Fig. 4B) but not with tissue water content (D. nepalensis: r=0.20±0.34, P=0.43; D. takahashii: r=0.23±0.39, P=0.36; Fig. 4C). Thus, both species showed similar patterns of trait correlations.

We observed higher storage levels of trehalose (91.23±1.52 μg fly^–1) and glycogen (46.01±1.03 μg fly^–1) in D. nepalensis than in D. takahashii (trehalose: 42.05±0.98 μg fly^–1; glycogen: 38.20±0.68 μg fly^–1; Fig. 5). In contrast, the storage levels of body lipids (D. nepalensis: 60.81±1.26 μg fly^–1; D. takahashii: 59.99±1.05 μg fly^–1) and proteins (D. nepalensis: 52.80±1.09 μg fly^–1; D. takahashii: 51.680±0.98 μg fly^–1) were similar in both species. Based on standard conversion factors (Schmidt-Neilsen, 1990; Marron et al., 2003), we compared the energy budget due to each metabolite (carbohydrates or lipids or proteins) in D. nepalensis and D. takahashii (Table 3). The energy budget of D. nepalensis was approximately 70% higher than that of D. takahashii because of the greater carbohydrate storage level. Further, we observed ∼1.22-fold differences in total energy budget of these two related Drosophila species, consistent with higher storage of carbohydrates in D. nepalensis.

A comparison of utilization patterns of trehalose and glycogen as a function of different durations of desiccation stress for both species is shown in Fig. 5. We observed similar slope values for trehalose utilization in both species (D. nepalensis: b=–1.80±0.003; D. takahashii: b=–1.76±0.004; the negative sign indicates a decrease in metabolite level with time due to utilization; Fig. 5A) but much lower utilization of glycogen (D. nepalensis: b=–0.56±0.002, P<0.001; D. takahashii: b=–0.55±0.002, P<0.001; Fig. 5B).

Further, data on metabolite utilization in 20 isofemale lines of each species were subjected to nested ANOVA (Table 4). Results showed significant F-values due to utilization of carbohydrates (D. nepalensis: F=9687.36, P<0.001; D. takahashii: F=2979.14, P<0.001; Table 3) but lack of changes in lipids (D. nepalensis: F=0.10, P=0.74; D. takahashii: F=3.66, P=0.48; Table 3) and proteins (D. nepalensis: F=2.75, P=0.51; D. takahashii: F=1.10, P=0.68; Table 3) levels during desiccation stress until death. Thus, there was no utilization of body lipids and proteins under desiccation stress in either species.

We measured possible effects of four different humidity conditions (0, 22, 33 and 45% relative humdity) on desiccation resistance, dehydration tolerance and rate of water loss (Fig. 6). We standardized the data on trait values for species comparison and to avoid scaling effects. We found simultaneous changes in desiccation resistance as well as rate of water loss under variable humidity conditions, but there were no corresponding changes in dehydration tolerance. In contrast, desiccation resistance showed a significant negative correlation with rate of water loss in both the species. Thus, we found that changes in desiccation resistance were not correlated with dehydration tolerance (Fig. 6A), but were negatively correlated with rate of water loss (Fig. 6B) under different humidity conditions in both species.

Fig. 7 illustrates the comparison of desiccation resistance and dehydration tolerance (mean ± s.e.m.; 20 isofemale lines for each species) in non-acclimated (control) and acclimated female flies of D. nepalensis and D. takahashii. For both physiological traits, there was a significant increase in the trait value (∼6 h in desiccation resistance and ∼ 5% in dehydration tolerance) in acclimated D. nepalensis, but such responses were not observed in D. takahashii. Further, results of ANOVA explaining the trait variability are shown in Table 5. We observed significant F-values for increases in desiccation resistance as well as dehydration tolerance (desiccation resistance: F=3039.10, P<0.001; dehydration tolerance: F=2316.56, P<0.001) due to acclimation in D. nepalensis, but no such acclimation effects were observed in D. takahashii (P>0.39).

In the present study, we found significant differences in desiccation-related traits (desiccation resistance, dehydration tolerance, water budget and energy metabolites) between D. nepalensis and D. takahashii, which differ in their abundance under field conditions. Female individuals of both the species grown at 21°C did not differ in their cuticular traits (body melanisation and cuticular lipids mass). However, we observed a modest difference in rate of water loss (1.8% h^–1 in D. takahashii versus 1.5% h^–1 in D. nepalensis), which cannot account for the twofold differences in desiccation resistance between these two species. In D. nepalensis, greater accumulation of hemolymph as well as carbohydrate content is significantly correlated with higher desiccation resistance. Thus, abundance of D. nepalensis under drier habitats can be explained by increased bulk water, higher dehydration tolerance and desiccation acclimation potential. In contrast, lower dehydration tolerance as well as desiccation resistance are consistent with the lower adaptive potential of D. takahashii for drier climatic conditions.

Changes in epicuticular lipid mass and/or body melanisation are associated with reduction in cuticular water loss in drosophilids (Parkash et al., 2008a; Parkash et al., 2008b). However, in the present study, female individuals of D. nepalensis and D. takahashii grown at 21°C did not show differences in body melanisation or cuticular lipid mass, despite twofold differences in their desiccation resistance levels. These observations suggest that other physiological mechanisms of water balance may be responsible for such species-specific differences.

In insects, resistance to desiccation involves accumulation of bulk water in hemolymph (Folk et al., 2001; Folk and Bradley, 2005). However, such changes were not evident in xeric versus mesic Drosophila species from North America, i.e. desert Drosophila species did not show accumulation of extra bulk water when compared with mesic species (Gibbs and Matzkin, 2001). However, the physiological basis of accumulation of bulk water has not shown consistent patterns in laboratory-selected and wild populations of Drosophila species (Hoffmann and Parsons, 1989; Folk et al., 2001; Folk and Bradley, 2005). Further, associations between hemolymph and increased levels of carbohydrates have also been reported in laboratory-selected desiccation-resistant lines (Folk et al., 2001). In contrast, in the case of xeric versus mesic Drosophila species, bulk water as well as storage levels of carbohydrates and other energy metabolites did not vary (Marron et al., 2003). In the present study, D. nepalensis and D. takahashii showed significant differences in bulk water and hemolymph content, as well as storage levels of carbohydrates. Thus, we may argue that species-specific differences in desiccation resistance are associated with changes in bulk water and carbohydrate storage level. Our results on the differences in the physiological basis of water-balance-related traits in D. nepalensis and D. takahashii are in agreement with the water bound hypothesis (Gibbs et al., 1997). However, further studies are needed to support such arguments in diverse insect taxa as well as terrestrial arthropods.

Insects can enhance their desiccation resistance by increasing their total body water content, reducing the rate of body water loss and tolerating a larger proportion of overall water loss from the body (Hadley, 1994; Gibbs et al., 1997). In laboratory-selected desiccation-resistant lines, Gibbs and coworkers assessed the relative contribution of increased water content and reduced rate of water loss for evolved differences in desiccation resistance between D and C strains of D. melanogaster (Gibbs et al., 1997). In their study, laboratory-selected desiccation-resistant strains showed an increase in desiccation resistance by 10 h because of a reduced rate of water loss, but an increase by 9 h on the basis of increased bulk water (Gibbs et al., 1997). However, no previous study has examined the relative contributions of different water-balance-related traits in wild-caught populations of Drosophila using the calculations suggested by Gibbs and coworkers (Gibbs et al., 1997). In the present study, we partitioned different measures of the water budget that support greater desiccation resistance in D. nepalensis as compared with D. takahashii. Because D. nepalensis stores more bulk water than D. takahashii, it has ∼4.93 h longer desiccation resistance (bulk water difference between species/water loss of D. nepalensis, i.e. 0.133/0.027=4.93 h). We compared water loss (WL) in these two species (WL_{D. nepalensis}=0.027 mg h^–1 and WL_{D. takahashii}=0.030 mg h^–1; ratio of water loss=0.027/0.030=0.90) which can account for ∼2.38 h of increased desiccation resistance in D. nepalensis than D. takahashii (desiccation resistance of D. takahashii/reduced WL=21.40/0.90=23.78 h; 23.78–21.40=2.38 h). Further, D. nepalensis lost more water content (0.875 mg fly^–1) before succumbing to death (dehydration tolerance) compared with D. takahashii (0.472 mg fly^–1), and this difference in dehydration tolerance can contribute ∼10.47 h of increased desiccation potential for D. nepalensis (between species difference in total water lost during desiccation/water loss per hour in D. nepalensis=0.875–0.472/0.027=15.40 h; 15.40–4.93=10.47 h). In fact, we found that D. nepalensis exhibited ∼18 h more desiccation survival than D. takahashii in our laboratory assays. The present calculations have suggested ∼27% (4.93 h) and 14% (2.38 h) increased desiccation resistance in D. nepalensis due to differences in bulk water content and water loss, respectively. However, greater dehydration tolerance of D. nepalensis contributed a major proportion (∼58%, ∼10.47 h) of the difference in desiccation survival between these two Drosophila species. Thus, higher desiccation resistance of D. nepalensis is associated with higher dehydration tolerance in comparison with other routes of water conservation.

Numerous studies have implicated the changes in the storage levels of energy metabolites (carbohydrates) in laboratory-selected desiccation-resistant (D) strains of Drosophila (Graves et al., 1992; Chippindale et al., 1998; Djawdan et al., 1998). In contrast, a new set of laboratory-selected desiccation-resistant strains showed higher accumulation of lipids in desiccation-resistant lines compared with controls (Telonis-Scott et al., 2006). Further, a comparative study has shown higher desiccation resistance in xeric than mesic Drosophila species, despite a lack of differences in carbohydrate storage levels (Marron et al., 2003). However, the higher desiccation potential of xeric species was related to a reduced rate of metabolite utilization compared with mesic species (Marron et al., 2003). In the present study, our interspecific comparisons have shown a positive correlation between carbohydrate storage levels and species-specific desiccation potential. The two sympatric Drosophila species differ in their total energy budget (1.22-fold) because of higher carbohydrate storage in D. nepalensis, which is consistent with the contrasting levels of desiccation potential in the two species. In contrast to the results of a previous study on laboratory-selected desiccation-resistant lines (Telonis-Scott et al., 2006), we did not find that higher storage of body lipids in D. nepalensis compared with D. takahashii conferred desiccation resistance. Interestingly, these two related Drosophila species did not vary in their rate of utilization of metabolites under desiccation stress.

In the present study, we found significant differences in dehydration tolerance between D. nepalensis and D. takahashii (82% versus 50%). Our results for D. nepalensis differ from dehydration tolerance levels observed in field populations of D. melanogaster (Parkash et al., 2008a) and laboratory-selected desiccation-resistant lines of D. melanogaster (Gibbs et al., 1997). Further, Drosophila species collected from contrasting habitats, i.e. xeric versus mesic, also did not differ in their dehydration tolerance (Gibbs and Matzkin, 2001). In contrast, there is evidence of higher dehydration tolerance in other invertebrate species, e.g. 89.2% in the semi-aquatic beetle Peltodytes muticus (Arlian and Staiger, 1979) and 75% in Belgica antarctica (Benoit et al., 2007b). However, mechanisms promoting dehydration tolerance are less clear in Drosophila, although several studies have suggested that both desiccation and low temperature stimulate the production of trehalose and glycerol (Holmstrup et al., 2001; Yoder et al., 2006; Holmstrup et al., 2010). Accumulation of trehalose has been correlated with dehydration tolerance in Anthonomus pomorum (Kostál and Simek, 1996) and in Polypedilum vanderplanki (Watanabe et al., 2002; Watanabe, 2006). In the present work, trehalose levels differed significantly between D. nepalensis and D. takahashii and were correlated with desiccation resistance. Thus, the higher trehalose level in D. nepalensis may confer greater dehydration tolerance, but this needs further investigation.

Most studies have investigated interspecific and intraspecific differences in desiccation resistance at 5% relative humidity (Gibbs et al., 1997; Gibbs and Matzkin, 2001; Parkash et al., 2008a), whereas less attention has been paid to the effects of a range of lower humidity conditions (Benoit et al., 2007b; Eckstrand and Richardson, 1981). In the present study, we found that rate of water loss increased with decreasing relative humidity conditions, but there were no corresponding changes in dehydration tolerance (Fig. 6A). Based on standardized data, we found a lack of correlation between desiccation resistance and dehydration tolerance. This observation suggests possible independence of factors affecting these desiccation-related traits. However, as expected, we found a negative correlation between desiccation resistance and rate of water loss (Fig. 6B) when measured at different humidity conditions.

Several studies have shown that dehydration pre-treatment followed by rehydration can equip insects or terrestrial arthropods to tolerate a subsequential exposure to dehydration (Holmstrup and Somme, 1998; Sjursen et al., 2001; Benoit et al., 2007b). However, a single study has analyzed acclimation to desiccation stress in four Drosophila species from Australia (Hoffmann, 1991). This study showed a lack of acclimatory response in tropical rainforest D. birchii but improved desiccation resistance due to acclimation in D. serrata, D. simulans and D. melanogaster. Further, D. birchii and D. serrata are sibling species of the montium species subgroup and have shown divergence in acclimation response to dehydration stress, and such observations are in agreement with divergence in their desiccation resistance (Hoffmann, 1991). Subsequently, another study on a single population of D. melanogaster provided indirect evidence that acclimation to desiccation resistance may be affected by reduction in cuticular water loss rates, but not through changes in respiratory water loss (Bazinet et al., 2010). Our results on D. nepalensis and D. takahashii, which belong to the takahashii species subgroup, showed an increase in desiccation resistance as well as dehydration tolerance in D. nepalensis after acclimation, but there was a lack of effects in D. takahashii, and such changes are consistent with their abundance level under field conditions.

In the present study, we found contrasting levels of desiccation resistance (twofold differences) between two related Drosophila species that occur sympatrically but differ in their abundance level. Contrary to general expectations, we did not find corresponding differences in rate of water loss (0.3% h^–1 difference between species), which is consistent with a lack of interspecific differences in cuticular traits (body melanisation and cuticular lipid mass). However, D. nepalensis is characterized by higher hemolymph and carbohydrate content compared with D. takahashii, and both of these traits are correlated with species-specific differences in desiccation resistance. A major difference between these two Drosophila species relates to dehydration tolerance (82% in D. nepalensis versus 50% in D. takahashii). To our knowledge, such a high level of dehydration tolerance has not been observed in any other Drosophila species so far, although similar levels do occur in some terrestrial arthropods such as Belgica antarctica (Benoit et al., 2007b). Thus, D. nepalensis and D. takahashii differ mainly in two modes of water conservation, i.e. bulk water as well as dehydration tolerance, despite modest differences in rate of water loss. Interestingly, these two Drosophila species differ in their acclimation potential to dehydration stress. Drosophila nepalensis responds to acclimation and has shown improved desiccation resistance as well as dehydration tolerance, whereas there is a lack of acclimation response in D. takahashii. Our results suggest that under field conditions, even closely related Drosophila species may evolve quite different physiological mechanisms that may impact their abundance level.

We are indebted to the reviewers for several helpful comments that improved the manuscript.