Phenotypic flexibility in migrating bats: seasonal variation in body composition, organ sizes and fatty acid profiles

Many species of bats migrate long distances between breeding and wintering areas [>1000 km (Fleming and Eby, 2003)], but the physiological adaptations and trade-offs involved have been poorly studied (McGuire and Guglielmo, 2009). Some species of bats can fly continuously for long durations [e.g. >6 h (Barclay, 1989)] and travel great distances [e.g. >50 km (Best and Geluso, 2003)] just for daily foraging, and consequently the degree to which migration poses additional flight-related physiological challenges to bats is unknown. A night of migratory flight may be no more demanding than a long night of foraging. With the exception of aerial foraging insectivores such as swallows, the endurance flights required for migratory movements of most birds far exceed the typical amount of time spent flying at any other time of the year (e.g. Portugal et al., 2011). In birds, the high energy demands of migration are often reflected in dramatic changes in physiology, and lead to seasonal changes in numerous characters at the molecular, biochemical, tissue, organ and whole-animal levels (e.g. Guglielmo and Williams, 2003; McWilliams et al., 2004; Guglielmo, 2010). Such reversible changes are referred to as phenotypic flexibility (Piersma and Drent, 2003; Piersma and van Gils, 2011).

We examined migration-related phenotypic flexibility in migrating and non-migrating bats. We hypothesized that bats and birds, despite being distantly phylogenetically related, would face similar physiological challenges during migration (McGuire and Guglielmo, 2009). Mammalian exercise physiology has been studied extensively (e.g. Roberts et al., 1996; Weber et al., 1996a; Weber et al., 1996b; Weibel et al., 1996; McClelland, 2004), but the high metabolic rates required to sustain flight [∼15 times basal metabolic rate (Speakman and Thomas, 2003)] and selection for high energy density substrates (fat, not carbohydrates) (Jenni and Jenni-Eiermann, 1998) make it unclear whether the ‘mammalian exercise model’ (McWilliams et al., 2004) is appropriate for migrating bats. Migratory birds may serve as a better group for comparison. Bats and birds are both vertebrate endotherms that have evolved the capacity for true powered flight (Rayner, 1988; Maina, 2000), and thus may have converged on similar solutions to some of the physiological challenges of migration. If bat migration is indeed more physiologically demanding than normal foraging flight, we predicted that we would observe phenotypic changes similar to those previously shown in migratory birds. Specifically, we examined seasonal variation in body composition (fat and lean stores), organ sizes (exercise and digestive organs) and the fatty acid (FA) composition of adipose tissue and flight muscle membranes.

Several studies have assumed that fat is the primary fuel for migratory flight in bats (O'Shea, 1976; Fleming and Eby, 2003; McGuire et al., 2012), but there are few detailed studies of fuel metabolism in bat migration. Bats can fuel flight with recently ingested nutrients (Welch et al., 2008; Voigt et al., 2010a) and use a mixed fuel strategy, relying on both recently ingested nutrients and endogenous fat stores (Voigt et al., 2012). When exogenous nutrients are not available, fasted bats can fuel short periods of flight, and possibly endurance migratory flights as well (Voigt et al., 2012), exclusively with stored fat (Welch et al., 2008; Voigt et al., 2010a). Seasonal fat deposition is common in many species of bats that migrate to and from hibernacula (e.g. Krulin and Sealander, 1972; Ewing et al., 1970; Kunz et al., 1998), but it is difficult to distinguish between fat deposited to fuel migration and that deposited to survive months of hibernation. Studies of non-hibernating migratory bat species are rare. Brazilian free-tailed bats (Tadarida brasiliensis) do not hibernate and deposit fat stores for spring and autumn migration (O'Shea, 1976). Fat deposition consistent with the timing of migration has also been reported in several West African species (O'Shea and Vaughan, 1980) and lesser long-nosed bats (Leptonycteris yerbabuenae) (Ceballos et al., 1997). In the current study, we predicted that fat stores would be larger in bats during spring migration than in summer when they are not migrating.

Migrating bats and birds face two seemingly opposed activities, flight and refuelling, each of which may require increased capacities of different physiological systems (although the difference may be less pronounced in aerial compared wtih terrestrial foraging species). The contrasts between high-intensity exercise during flight and hyperphagia during refuelling have led numerous studies of birds to examine changes in ‘exercise organs’ or ‘flight machinery’ (e.g. flight muscles, heart, lungs) and ‘digestive’ or ‘nutritional organs’ (stomach/gizzard, intestines, kidneys, liver) (Hume and Biebach, 1996; Piersma, 1998; Piersma and Gill, 1998; Piersma et al., 1999; Battley et al., 2001; Guglielmo and Williams, 2003; Landys-Ciannelli et al., 2003; Bauchinger et al., 2005). Increases in flight machinery may increase exercise performance on long migratory flights (Pennycuick, 1998; Guglielmo and Williams, 2003), or may provide extra protein to be catabolized in flight to maintain blood glucose, replenish metabolite pools for the Krebs cycle, or provide metabolic water (Jenni and Jenni-Eiermann, 1998; Gerson and Guglielmo, 2011). Thus, we predicted that migrating bats would have enlarged exercise organs.

Digestive organs can be catabolized in flight, but are needed to maximize fuelling rate and minimize time spent at stopover [‘migration takes guts’ (McWilliams and Karasov, 2005)]. In some cases, digestive organs appear to represent surplus mass that can be eliminated for a more energy efficient flight [‘guts don't fly’ (Piersma and Gill, 1998)]. Consequently, changes in the size of digestive organs can indicate migratory strategy. Birds that migrate long distances without stopping to refuel (typically crossing ecological barriers such as oceans or deserts) reduce the size of digestive organs prior to or during flight, and rebuild these organs upon arrival at stopover sites (Hume and Biebach, 1996; Piersma, 1998; Piersma and Gill, 1998; Piersma et al., 1999; Battley et al., 2000; Battley et al., 2001; Bauchinger et al., 2005). Alternatively, species that stop frequently along the migratory route may maintain enlarged digestive organs to minimize refuelling time at each stopover [e.g. western sandpipers, Calidris mauri (Guglielmo and Williams, 2003)]. Most bats remain nocturnal during migration and therefore must complete all foraging and migratory flight within the hours of darkness, unlike nocturnally migrating songbirds, which forage by day and fly at night. We predicted that the additional time pressure faced by migratory bats would select for enlarged digestive organs to facilitate rapid refuelling.

Migrants may alter the FA composition of adipose stores and cell membranes to facilitate migration (McWilliams et al., 2004; Price, 2010). FA chain length and degree of unsaturation affect the potential energy derived from fat stores (more ATP from longer FAs with fewer double bonds) (Price, 2010). However, these same factors affect the rate of mobilization: short FAs with more double bonds are preferentially mobilized (Price et al., 2008) and preferentially oxidized (Leyton et al., 1987; Price et al., 2011). Consequently, adipose FA profiles present a trade-off between rapid mobilization and energy density. Adipose FA composition is largely determined by diet (Price, 2010), and although bats generally do not demonstrate the same degree of dietary plasticity as birds (an insectivorous bat will not suddenly switch to a diet of seeds or berries), there is some limited evidence that bats select insect prey based on FA content (Schalk and Brigham, 1995). If migrating bats are able to alter the FA composition of adipose stores either through dietary selection or by preferentially retaining particular FAs, we predict a shift towards shorter chain length and more unsaturated FAs, resulting in more rapid mobilization.

Variation in the FA profiles of cell membrane phospholipids (PLs) has been suggested to affect whole-animal exercise performance. A pervasive theme in both mammalian and avian literature is the importance of essential dietary polyunsaturated FAs (PUFAs; n-3 and n-6), though the pattern is not consistent. Various studies have found increased exercise performance (or increased proportion in migrants compared with non-migrants) with high n-6 PUFA (Pierce et al., 2005; Ayre and Hulbert, 1997; Ruf et al., 2006), high n-3 PUFA (Maillet and Weber, 2007), a high ratio of n-6:n-3 (Klaiman et al., 2009) or a low ratio of n-6:n-3 (Guglielmo et al., 2002). Alternatively, muscle PL composition may not affect exercise performance at all; rather, differences may simply arise due to differences in adipose FAs (Price and Guglielmo, 2009). Therefore, the role of muscle PL composition in whole-animal exercise performance remains unclear.

Muscle PL composition has also been implicated in torpor and hibernation studies. Munro and Thomas (Munro and Thomas, 2004) described improved torpor performance (e.g. torpor depth, torpor bout duration, metabolic rate) with high membrane PUFA content. Based on the Munro and Thomas (Munro and Thomas, 2004) framework, migrating insectivorous bats should seek to maximize PUFAs. However, another review suggested that a high n-6:n-3 ratio is important for maintaining membrane function at low body temperatures (Ruf and Arnold, 2008). Male hoary bats are more likely to use torpor during spring migration than females (Cryan and Wolf, 2003), and therefore we predicted that sexes would differ in muscle PL composition during migration. Specifically, we predicted that males would have either greater total PUFAs or greater n-6:n-3 if the Munro and Thomas (Munro and Thomas, 2004) or Ruf and Arnold (Ruf and Arnold, 2008) hypotheses were supported, respectively.

The hoary bat [Lasiurus cinereus (Beauvois 1796)] is the most widely distributed bat species in North America (Shump and Shump, 1982), and is believed to migrate longer distances than other migratory species (Cryan et al., 2004). Hoary bats exhibit sexual size dimorphism, where females are ∼3% larger than males (Williams and Findley, 1979). Hoary bats are solitary and roost in exposed foliage (Willis and Brigham, 2005; Carter and Menzel, 2007; Cryan and Veilleux, 2007). In summer the sexes are largely segregated, with males more commonly found in mountainous regions in the western part of North America, whereas females are widespread throughout the eastern part of the continent (Cryan, 2003). Mating occurs during autumn migration (Cryan, 2008) and females delay pregnancy until spring, migrating north while pregnant (Cryan and Wolf, 2003). The winter distribution is not well understood, though it is thought that most individuals overwinter in southern California and Mexico (Cryan, 2003).

Migrating L. cinereus were captured from 5 to 17 May 2009 by setting mist nets across creeks in Bernalillo County, New Mexico, USA (35°12′N, 106°18′W), or around water pools in the Manzano Mountains, Cibola National Forest, New Mexico, USA [34°59′N, 106°21′W; see Cryan and Wolf (Cryan and Wolf, 2003) for a description of the region]. Non-migrating bats were captured between 20 July and 1 August 2008 and 2009 in mist nets set across creeks in Cypress Hills Interprovincial Park, Saskatchewan, Canada [49°34′N, 109°53′W; see Willis and Brigham (Willis and Brigham, 2005) for description]. For the non-migrants, we determined age (sub-adult or adult) by the degree of ossification of the metacarpal-phalanges joint (Anthony, 1988). By autumn the metacarpal-phalanges joint is fully ossified, and thus it was not possible to determine the age of bats during spring migration. Lactating females were identified by manually expressing milk from the mammary glands. Sub-adults and lactating females were released immediately upon capture. We collected 15 female and 15 male migrants (New Mexico), and eight female and seven male non-migrants (Saskatchewan). All non-migrating females were post-lactating and all females collected during spring migration were pregnant. We euthanized the bats immediately following capture by cervical dislocation under isoflurane anaesthesia. We recorded body mass (±0.1 g) and forearm length (±0.05 mm) and immediately excised samples of pectoral muscle, liver and adipose tissue, which we transferred to pre-weighed individual vials (for liver and muscle; 2 ml Cryotube, Cryo.S, Grenier Bio-One) or 600 μl O-ring-sealed screw cap microcentrifuge tubes (for adipose; Fisherbrand, Thermo Fisher Scientific, Pittsburgh, PA, USA) and froze in a liquid nitrogen cooled dry-shipper (Taylor-Wharton CX-100, Theodore, AL, USA). The remainder of the carcass was frozen in a sealed plastic bag at –20°C. Samples were transported back to the laboratory either in liquid nitrogen cooled cryoshippers or packed in dry ice. Pectoralis, liver and adipose tissue samples were stored at –80°C; the remaining tissues were stored at –20°C.

All animal collection and experimental protocols were approved by the University of Western Ontario Animal Use Sub-committee (protocol no. 2008-003-04) and conducted under permits from the New Mexico Department of Game and Fish (permit no. 3424), the United States Department of Agriculture-Forest Service (permit no. SND502), Saskatchewan Ministry of Environment (permit nos 08FW080 and 09FW045) and the Saskatchewan Ministry of Tourism, Parks, Culture and Sport (permit nos SP-CHPP-02-08 and SP-CHPP-01-09). Samples from New Mexico were imported to Canada under the approval of the Canadian Food Inspection Agency (permit no. A-2009-01022-3).

Carcasses were thawed overnight at 4°C prior to dissection. We removed the remaining pectoral muscle, intestines (large and small combined), stomach, remaining liver tissue, kidneys, heart and lungs. Intestines and stomach were opened to remove all contents, rinsed in 0.9% NaCl and blotted dry. We recorded the wet mass of each organ or tissue (±0.0001 g), correcting for subsamples taken in the field. We dried organs to a constant mass at 70°C, then placed them in pre-weighed filter paper envelopes (Whatman #1) and extracted the lipid fraction with petroleum ether (boiling point 30–60°C) for 6 h in a Soxhlet apparatus. Similarly, we dried the remainder of the carcass at 70°C, and homogenized it with a heavy-duty blender (model CB15l Waring Commercial, Torrington, CT, USA). We divided the homogenate into two to three pre-weighed filter paper envelopes (Whatman #1) for Soxhlet extraction with petroleum ether.

Total lipids were extracted by adding the sample (75–120 mg muscle or 6–12 mg adipose) to 8 ml chloroform:methanol (1:1 v/v) containing butylated hydroxytoluene (25 mg l^–1), homogenizing for 2×10 s (Polytron PT 10-35, Kinematica, Bohemia, NY, USA), adding 4 ml chloroform and homogenizing for an additional 1×10 s. The homogenizer was rinsed with an additional 6 ml chloroform methanol (2:1 v/v) to ensure complete transfer of the sample to the sample tube. The sample was then centrifuged for 15 min at 2056 g, and gravity filtered (Whatman #1) into a new tube. The previous sample tube was rinsed with 12 ml chloroform:methanol (2:1 v/v), which was also filtered into the new tube. To separate aqueous solutes, we added 7.5 ml 0.25% KCl, incubated in a 70°C water bath for 10 min, and discarded the aqueous layer. The remaining organic phase was transferred to a pear flask (25 ml) and evaporated under vacuum (Rotovapor, Buchi, Switzerland) at 60°C. Dried samples were either suspended in 1–2 ml chloroform:methanol (1:1 v/v) under nitrogen for overnight storage at –20°C, or immediately resuspended in 100 μl chloroform for separation of the different lipid fractions. PL, neutral lipid (NL; primarily triglycerides) and non-esterified FA (NEFA) fractions were separated with Supelclean solid phase extraction tubes (LC-NH₂, 100 mg; Supelco, Sigma-Aldrich Canada, Oakville, ON, Canada). The columns were conditioned with 2 ml hexane prior to addition of the samples. After each elution the samples were centrifuged for 1 min at 1370 g. NLs were eluted with 1.8 ml chloroform:isopropanol (2:1 v/v). NEFAs were eluted with 1.6 ml isopropyl ether:acetic acid (98:2 v/v). PLs were eluted with 3 ml methanol. We added heptadecanoic acid (17:0; 3 mg ml^–1 in hexane) as an internal standard to the NL and PL fractions.

For transesterification, the NL and PL fractions were dried at 70°C under a stream of N₂, resuspended in 2 ml of 1 mol l^–1 acetyl chloride in methanol and incubated at 90°C for 2 h. The samples were then dried under N₂, resuspended in 1 ml methanol and dried under N₂ again to remove any residual HCl and H₂O. Finally, the samples [now fatty acid methyl esters (FAMEs)] were resuspended in dichloromethane for analysis by gas chromatography (Agilent Technologies 6890N, Hewlett Packard, Palo Alto, CA, USA). We used a J&W Scientific high-resolution gas chromatography column (DB-23, Agilent Technologies), a flame ionization detector and He as a carrier gas [as described in Klaiman et al. (Klaiman et al., 2009)]. The temperature programme was 2 min at 80°C, increase 5°C min^–1 for 20 min, hold 180°C for 3 min, increase 1.5°C min^–1 for 13.3 min, hold 200°C for 0 min, increase at 10°C min^–1 for 4 min, and hold 240°C for 4 min. Fatty acids were identified by comparing relative retention time (retention time/retention time of internal standard) to known standards (Supelco C8–C24 FAME mix, Supelco 37 component FAME mix, and Supelco PUFA no. 3 from menhaden oil). For analysis, we did not consider short chain fatty acids (less than 16 carbons) or any fatty acids that comprised less than 0.5% of the total FA content.

All analyses were completed with the software R (version 2.9.2) (R Development Core Team, 2009). In all cases where we considered body mass, we used empty mass (capture mass – stomach content mass). We first used a linear model to test for sex and migration effects on forearm length to confirm that there was no systematic size difference between the migrating and non-migrating bats. We then tested for sex and migration effects on body mass, dry lean mass and fat mass. For all linear models, we started with a full model including main effects and all two-way interactions. We then removed non-significant terms and re-evaluated the model until only significant terms remained. All masses were log_e-transformed prior to analysis.

For statistical analysis of organ sizes, we considered wet intestine mass, dry lung, kidney and stomach masses, and dry lean (extracted) liver, heart and pectoralis masses as in Guglielmo and Williams (Guglielmo and Williams, 2003). When comparing organ sizes we originally considered forearm length as a measure of body size but found that organ sizes were more strongly correlated with body mass. Thus, we calculated corrected body mass as log_e(body mass – organ mass) as a measure of body size accounting for part whole correlation, substituting total wet mass, total dry mass or total dry lean mass as appropriate. For each organ, we used general linear models to test for the effects of sex and migration controlling for corrected body mass (including all two- and three-way interactions). We sequentially removed non-significant terms and re-evaluated the model until only significant terms remained. Whether considering models correcting for forearm length or body mass, the results were qualitatively the same.

We also used principal components analysis (PCA) to conduct a multivariate analysis of migration-related changes in muscle and organ sizes. We entered all organ masses and total fat mass into a PCA and used MANOVA to test for effects of sex, migration and body mass on the retained PC axes.

We compared the fatty acid profiles of adipose NLs and muscle PLs as described in Klaiman et al. (Klaiman et al., 2009). We arcsin square root transformed the proportions of each FA and conducted two-way ANOVA to test for effects of sex, migration and sex×migration interaction. Furthermore we calculated the double bond index as DBI=Σ[(proportion FA_i)(number of double bonds per FA_i)]. We also calculated the n-6:n-3 ratio, and the total proportions of MUFA, PUFA and saturated FA.

Body composition is summarized by sex and migratory status in Table 1. Female hoary bats were 3.4% larger than males (forearm length: F_1,42=30.91, P<0.0001) but there was no effect of migratory status on body size (F_1,42=0.081, P=0.78). Body mass was greater in females (F_1,42=88.55, P<0.0001) and migrating bats weighed ∼15% less than non-migrants (F_1,42=23.91, P<0.0001). Dry lean mass was lower in migrants (F_1,42=5.73, P=0.021) and males (F_1,42=32.87, P<0.0001). The effect of migratory status on fat storage was sex dependent (sex × migration: F_1,41=6.24, P=0.017; Fig. 1). Absolute fat mass decreased during migration for males (F_1,20=4.92, P=0.038), but when controlling for fat-free mass (accounting for overall decrease in body mass) there was no difference in fat content of migrating and non-migrating males (F_1,19=1.31, P=0.27). For females, there was no difference in the absolute fat mass between migrating and non-migrating bats (F_1,21=2.19, P=0.15), but when controlling for fat-free mass the relative fat content increased during migration (F_1,20=5.53, P=0.029), accounting for as much as 24% of body mass in one individual. In summer, there was no difference in the relative fat content of males and females (F_1,12=0.0034, P=0.95).

Organ masses are summarized by sex and migration in Table 1. Upon visual inspection of the untransformed organ masses, we noted obvious outliers for intestine and pectoralis mass (Fig. 2). There were three female non-migrants with much larger intestines than any other bats. These were the first three females we captured in the summer. These same individuals and one other non-migrant bat (a male) had much smaller pectoralis muscles than would have been expected for their size. After applying the log_e transformation to intestine mass, the outliers were no longer problematic. Migrating bats had smaller intestines than non-migrating bats (F_1,42=14.97, P<0.0001). However, the pectoralis mass outliers were still apparent even after data transformation and thus were excluded from the analysis. Excluding outliers, there was no effect of migratory status on pectoralis mass (F_1,38=1.09, P=0.30). However, it appears that the pectoralis muscles in four of the non-migrants were much smaller than expected for bats of their size.

No other outliers were apparent in the data set. Migrating bats had larger lungs than non-migrants (F_1,42=8.49, P=0.0057) but there was no difference in heart size (F_1,42=0.13, P=0.72). There was no difference between migrants and non-migrants in the size of stomach (F_1,42=0.15, P=0.70). Similarly, liver size was not affected by migratory status (F_1,41=0.36, P=0.55), though females had larger livers than males (F_1,41=33.01, P<0.0001). Kidney size was affected by both sex (females had larger kidneys; F_1,41=9.03, P=0.0045) and migratory status (F_1,41=5.33, P=0.026), with smaller kidneys in migrating bats.

Excluding the three outliers among the non-migrant females, the results of the PCA largely reflected the conclusions of the individual organ analyses. The first two principle components (PC1 and PC2) were retained in the analysis, accounting for 43.1 and 13.4% of the total variance, respectively (Table 2). All loadings in PC1 were positive and approximately equal, suggesting that this PC reflects body size. This is further confirmed by the strong correlation between PC1 and body mass (Pearson's r=0.87). Lungs, heart and pectoralis loaded positively on PC2, while intestines, stomach, kidneys and fat had negative loadings (Table 2), suggesting that this PC represented an axis of exercise machinery and digestive machinery.

Including both PCs, MANOVA indicated significant effects of migratory status (Wilks' λ=0.764, F_2,37=5.73, P=0.0068), sex (Wilks' λ=0.199, F_2,37=74.64, P<0.0001) and body mass (Wilks' λ=0.408, F_2,37=26.84, P<0.0001). PC1 was related to body mass (F_1,38=48.76, P<0.0001) and sex (F_1,38=137.92, P<0.0001), but not migratory status (F_1,38=0.098, P=0.76), further confirming this PC as an indication of body size (the inclusion of sex reflects sexual size dimorphism). PC2 was related to migratory status (F_1,38=10.21, P=0.0028), but not body mass (F_1,38=0.24, P=0.63) or sex (F_1,38=0.27, P=0.61). Therefore, migrating bats were shifted towards increased exercise machinery and reduced digestive machinery.

We found migration-related differences in the proportions of all adipose NL FAs except 18:0 (Fig. 3). Although some sex × migration interactions indicated that the proportions of certain FAs changed for one sex but not the other, the general pattern was a decrease in 16:0 and increases in 18:2n-6 and 18:3n-3 during migration. This pattern is further supported by decreased saturated FAs (F_1,42=35.53, P<0.0001) and increased PUFAs. A sex × migration interaction indicated a greater PUFA increase for males (F_1,20=15.12, P=0.00091) than females (F_1,21=8.89, P=0.0071). There was no overall change in MUFAs (F_1,42=0.44, P=0.51). DBI was greater in migrants (F_1,42=45.57, P<0.0001) and in males (F_1,42=5.91, P=0.019). There was no difference in the n-6:n-3 ratio of migrating and non-migrating bats (F_1,42<0.0001, P=0.99).

There were differences in FA composition between migrants and non-migrants for five of the nine FAs comprising muscle PLs (Fig. 4). Sex × migration interactions indicated sex-specific responses for several FAs, notably 18:2n-6 and 22:6n-3. Males had more 18:2n-6 and less 22:6n-3 during migration, while the pattern was reversed for females. This led to a significant change in the n-6:n-3 ratio, which was higher during migration for males (migrants: 0.89±0.06; non-migrants: 0.56±0.02; F_1,20=12.4, P=0.0022) and lower during migration for females (migrants: 0.59±0.03; non-migrants: 0.74±0.05; F_1,21=8.98, P=0.0069). DBI was similarly affected, decreasing during migration for males (migrants: 2.25±0.04; non-migrants: 2.48±0.05; F_1,20=14.08, P=0.0013) and increasing marginally for females (migrants: 2.43±0.03; non-migrants: 2.31±0.08; F_1,21=3.15, P=0.091). The composition of MUFAs, PUFAs and total saturated FAs was not affected by migratory status (all P>0.05).

We found migration-related changes in all aspects of our study, suggesting that for these bats, as in birds, migration presents distinct physiological challenges compared with other periods of the annual cycle (e.g. Gwinner, 1990; Dingle, 1996; Jenni and Jenni-Eiermann, 1998; Guglielmo, 2010). Sex was important for every aspect of physiology we examined, a pattern not typically observed in migratory birds. Migrating females carried relatively larger fat stores during migration, but males did not. In summer, some females stood out as outliers with larger intestines and smaller pectoralis muscles. Separate male and female effects were evident in adipose FA composition, and males and females changed muscle PL composition in opposite directions. Below we discuss these findings in the context of previous studies of migratory birds, and consider bat life-history factors that may contribute to the observed sex differences.

We predicted that migrating bats would weigh more than non-migrating bats due to the increased mass of fat stores and hypertrophy of muscles and digestive organs. Contrary to our expectations, body mass was reduced in migrating bats. Particularly surprising was the overall reduction of lean mass in migrants (see discussion of digestive organs below). A reduction in body mass reduces wing loading and consequently lowers the energetic cost of flight (Bowlin and Wikelski, 2008; Voigt et al., 2010b). The wing loading effect is in addition to the efficient flight that hoary bats achieve due to a high aspect ratio. Of 81 bats measured in the family Vespertilionidae, hoary bats had the fourth highest aspect ratio (Norberg and Rayner, 1987). High aspect ratio and seasonally lowered wing loading suggest that the energetic costs of migration have played an important role in the evolution of this species (although wing morphology adaptations for an open-space aerial hawking foraging strategy have also surely contributed).

Increases in the relative size of fat stores for migrating females, but not males, may reflect an increased energetic cost of migration for females. Female hoary bats generally migrate greater distances than males, and therefore females may require larger fat stores. Alternatively, the discrepancy may arise as a result of differences in thermoregulatory strategy during migration. Migrating bats could use daily torpor during migration and spare energy stores to fuel migratory flight (McGuire et al., 2012). This strategy could greatly reduce the overall energetic costs of migration; migratory birds incur twice the energy cost during stopover compared with actual migratory flight, largely because of thermoregulatory costs during overnight roosting (Wikelski et al., 2003; but see Carpenter and Hixon, 1988; Wojciechowski and Pinshow, 2009). However, Cryan and Wolf (Cryan and Wolf, 2003) demonstrated (at the same sites we collected our migrants) that during spring migration female hoary bats defend normothermic body temperature, while males readily use torpor when ambient temperature is decreased. Females may not lower body temperature because of the potential detrimental effects on the developing fetuses. If females do not use daily torpor, they would either need larger fat stores to support higher metabolic rates during diurnal roosting periods (compared with torpid males), or they would need to frequently replenish fuel stores that are depleted in both migratory flight and defence of body temperature. We argue (see below) that the former scenario is more consistent with our observations.

The general pattern of reduced digestive organs is similar to the pattern observed in numerous studies of migratory birds (e.g. Battley et al., 2000; Bauchinger et al., 2005), consistent with the ‘guts don't fly’ hypothesis [minimize gut mass to minimize the cost of transport (Piersma and Gill, 1998)]. Individually, intestines and kidneys were smaller in migrating bats (mean intestine mass 43% smaller, mean kidney mass 6–9% smaller; Table 1), and combined digestive organs (except liver) loaded together on the second PC axis, indicating smaller digestive organs in migrants. Most bird species for which reduced guts have been observed migrate vast distances over habitat that does not permit foraging. In the case of bar-tailed godwits (Limosa lapponica baueri), non-stop migratory flights may cover >11,000 km and last 9 days as birds fly from Alaska to New Zealand (Gill et al., 2009; Battley et al., 2012). Garden warblers (Sylvia borin) migrate >2500 km across the Sahara Desert without feeding (Bauchinger et al., 2005). However, digestive capacity (and digestive tract organ sizes) is typically linked to demand (McWilliams and Karasov, 2005). For a hoary bat migrating from southern California to Canada, there is ample suitable foraging habitat along the route, and therefore it is somewhat surprising that the changes in digestive organs are not similar to the increases observed in western sandpipers (C. mauri), which stop to refuel frequently along their migratory route (Guglielmo and Williams, 2003). Instead, it would appear that the bats in our study were more likely to deposit fuel (larger fat stores for the females that could not save energy through the use of daily torpor) prior to migration and minimize time spent foraging along the migratory route. We did not observe sex effects in the reduction of digestive organs, suggesting that both sexes reduce foraging effort similarly. To compensate for reduced foraging during migration, females that do not use torpor could deposit larger fat stores prior to migration, which is indeed what we observed. In the autumn when females are not pregnant, there should be no sex effects in either foraging effort or the size of fat stores. Both of these predictions are consistent with observations of fall migrating silver-haired bats (Lasionycteris noctivagans) where few bats foraged and there was no difference in body composition with regards to sex (McGuire et al., 2012). However, only two of the migrants in our study had empty stomachs. Similar ratios of migrating bats with and without stomach contents were observed in a study of hoary and silver-haired bats during autumn migration in Alberta, Canada (Reimer et al., 2010). Therefore, foraging does not cease completely during migration. Bats may forage briefly each night rather than alternating extended periods of refuelling and migratory flight (Voigt et al., 2012).

Increased exercise machinery has been frequently observed in migratory birds, but the individual organ changes differ from the observations of the bats in our study. Birds may increase the size of their heart and flight muscle (Marsh, 1984; Piersma, 1998; Piersma et al., 1999; Portugal et al., 2009), but changes in lung mass are rarely recorded. In the hoary bat PCA, all exercise organs (heart, lungs, pectoralis muscle) loaded together, indicating larger exercise organs in migrating bats. However, the evidence for changes in flight muscle mass is weak (not significant in individual organ comparison, smaller factor loading in PCA), suggesting that flight muscle size varies with body mass, as has been observed in many migratory birds (e.g. Marsh, 1984; Dietz et al., 1999; Lindström et al., 2000). When comparing individual organs, migrating bats had larger lungs (mean lung mass 21% larger; Table 1) but there was no difference in heart or flight muscle size. Lung mass change appears to be a novel component of bat migration physiology that may result from the less rigid structure of the mammalian lung, which enables phenotypic flexibility that is not possible given the design of the avian lung (Maina, 2000). Larger lungs in migrants, independent of body mass, may be associated with increased capacity for aerobic exercise. An intriguing alternative is that increased lung mass may be associated with exposure to low oxygen concentration, as observed in rodents (Burri and Weibel, 1971). Deer mice at high altitudes have been shown to increase lung mass in response to the lower oxygen concentration (Hammond et al., 2001). If bats spend more time flying at high altitudes during migration, and over a long enough period, lung mass may be increased to compensate for lower oxygen concentration. This suggestion is highly speculative and requires further investigation.

The data we collected from some of the non-migrating female hoary bats suggest that all organ size changes we observed may be conservative estimates of seasonal migration-related phenotypic flexibility. Female bats arrive in Cypress Hills in late May or early June (Willis et al., 2006). Parturition occurs shortly after arrival (mid-June), and the young become volant ∼5 weeks later (Shump and Shump, 1982). The earliest date we captured post-lactating females (in either year) was 20 July; the latest date we captured lactating females was 22 July. Therefore, some of the earliest females we captured (outliers in Fig. 2) may only have ceased lactating days earlier. The outliers shown in Fig. 2 suggest that there may be a rapid and dramatic change in body composition at weaning. In these early season females, intestines were approximately two times larger and pectoralis muscles were only ∼60% of the size that would be expected for their body mass. If these organ size changes are typical, then our late summer female samples may be better considered as pre-migratory samples. Comparison of migrating bats and lactating females may present a more extreme picture than the potentially conservative size changes we documented.

Adipose FA composition was shifted towards increased PUFAs and decreased saturated FAs. Such changes would slightly reduce the net ATP production per gram of triglycerides (Price, 2010), but would greatly increase the potential mobilization of FA stores. In a study of ruffs (Philomachus pugnax), the relative mobilization of 18:2n-6 and 18:3n-3 (two FAs that increased in hoary bats; Fig. 3) was ∼50% greater than 16:0 (decreased during migration in hoary bats; Fig. 3) (Price et al., 2008). A similar pattern (but less pronounced) has been observed in studies on rats and humans (Raclot and Groscolas, 1993; Raclot, 2003). High-intensity migratory flight demands a continuous supply of energy substrates. Consequently, the trade-off between potential energy and mobilization rate may favour those FAs that permit a high rate of sustained energy substrate delivery to the flight muscles. The reduction in energy density may be compensated by deposition of larger fat stores, or refuelling more frequently.

Interpretation of migration-related changes in the FA composition of muscle PL is notoriously difficult (Price, 2010). Many studies have reported effects of different muscle PL profiles, but these effects may in fact simply reflect differences in adipose composition. This is not likely the case in our study. The most notable differences in muscle PL FA were the changes in the n-6:n-3 ratio, increasing for males and decreasing for females. The only adipose FA where males and females changed in opposite directions was 18:1n-9, which was not affected by sex or migration in muscle PL. Sex-biased differences in muscle PL but not adipose NL suggest that any differences in muscle PL are not simply consequences of adipose NL composition.

It is difficult to interpret how muscle PL changes might affect exercise performance given that males and females changed in different manners. To interpret exercise performance effects we would need to understand sex differences in the energetic cost or energetic strategies associated with migratory flight. However, sex-based variation in the muscle PL n-6:n-3 ratio is consistent with the Ruf and Arnold (Ruf and Arnold, 2008) hypothesis that an increased n-6:n-3 ratio is associated with torpor use. During spring migration, male hoary bats readily use torpor while females rarely lower their body temperature (Cryan and Wolf, 2003), and accordingly we observed an increased n-6:n-3 ratio in males and a decreased ratio in females. The prediction that migrating males should maximize total PUFAs (Munro and Thomas, 2004) is not supported by our data, as we did not observe any differences in muscle PL total PUFA content.

Although bats and birds represent two phylogenetically distant lineages, they have converged on a number of similar physiological strategies associated with migratory behaviour. We observed patterns consistent with fat being used as the primary fuel for migration, as has been shown in previous studies of migratory birds. Bats are able to deposit large fat stores (as observed in the migrating females), and the FA profile of adipose stores indicates that mobilization of FAs to maintain high delivery rates in flight is more important than the energy density of the fat stores. Furthermore, changes in digestive organ sizes indicate that bats may favour a time-minimizing migration strategy, minimizing time spent foraging en route. Like birds, enlarged exercise organs of bats reflect the increased aerobic exercise demands during migratory periods. However, the increase in lung size that we observed has not been previously documented in birds and highlights the fact that the migration physiology of bats is in many ways distinct from that of birds.

The coincidence of spring migration and pregnancy is perhaps the most important contrast between the migration of birds and bats. In birds, migration and breeding are temporally isolated, and thus physiological consequences of migration are relatively easily isolated compared with the situation in bats. Clearly, reproductive physiology is an important consideration when interpreting bat migration-related phenotypic flexibility. Pregnancy and lactation may affect nearly all aspects of migration physiology that we considered. We collected bats during spring migration and summer non-migratory periods because these were the only times of year we could reliably capture hoary bats. Comparison of autumn migrating and wintering bats will help to isolate migration effects from reproductive effects. We were unaware of reliable sites for capturing autumn migrants or overwintering bats. Future work should focus on finding such sites or studying species where such sites are known.

 
• DBI

  double bond index

 
• FA

  fatty acid

 
• FAME

  fatty acid methyl ester

 
• MUFA

  monounsaturated fatty acid

 
• NEFA

  non-esterified fatty acid

 
• NL

  neutral lipid

 
• PC

  principal component

 
• PCA

  principal components analysis

 
• PL

  phospholipid

 
• PUFA

  polyunsaturated fatty acid

Thanks to Edwin Price and Alex Gerson for helpful discussions and developing laboratory analysis methods. We also thank Mark Brigham, Craig Willis, Robert Barclay, Blair Wolf, Paul Cryan and Ernie Valdez for discussion and assistance in determining field sites and co-coordinating collecting trips. Lauren Hooton, Amanda Adams, Quentin Hays, Julia Kilgour, Erin Gillam and Joe Poissant provided invaluable assistance in the field. Christian Voigt and two anonymous reviewers provided valuable feedback that improved the manuscript.