Modeling the Colonization of Hawaii by Hoary Bats (Lasiurus cinereus)

Abstract

The Hawaiian archipelago, the most isolated cluster of islands on Earth, has been colonized successfully twice by bats. The putative “lava tube bat” of Hawaii is extinct, whereas the Hawaiian Hoary Bat, Lasiurus cinereus semotus, survives as an endangered species. We conducted a three-stage analysis to identify conditions under which hoary bats originally colonized Hawaii. We used FLIGHT to determine if stores of fat would provide the energy necessary to fly from the Farallon Islands (California) to Hawaii, a distance of 3,665 km. The Farallons are a known stopover and the closest landfall to Hawaii for hoary bats during migrations within North America. Our modeling variables included physiological, morphological, and behavioral data characterizing North American Hoary Bat populations. The second step of our modeling process investigated the potential limiting factor of water during flight. The third step in our modeling examines the role that prevailing trade winds may have played in colonization flights. Of our 36 modeling scenarios, 17 (47 %) require tailwind assistance within the range of observed wind speeds, and 7 of these scenarios required <10 m s⁻¹ tailwinds as regularly expected due to easterly trade winds. Therefore the climatic conditions needed for bats to colonize Hawaii may not occur infrequently either in contemporary times or since the end of the Pleistocene. Hawaii’s hoary bats have undergone divergence from mainland populations resulting in smaller body size and unique pelage color.

1 Introduction

The Hawaiian archipelago is the most isolated cluster of islands on planet Earth. Among terrestrial animals, the only groups to reach the archipelago and flourish with significant radiations are birds, land snails, spiders, and a variety of insects that include flies, beetles, katydids, and moths (Zimmerman 1970). That’s it. No rodents, amphibians, or reptiles other than marine sea turtles have ever established on any Hawaiian island without anthropogenic association. There are only two living endemic mammals in Hawaii, one is the Hawaiian Monk Seal, Monachus schauinslandi, and the other is the animal the Hawaiians call Ōpe‘ape’a (oh-pay-ah-pay-ah), the Hawaiian Hoary Bat. This bat currently is recognized as a subspecies of the hoary bat, Lasiurus cinereus semotus, although historically it has been given full species status and reduced to a subspecies several times (see Tomich 1986). Subfossils of L. c. semotus are known from at least 5,500 YBP (Olson and James 1982), and it is believed that its ancestors colonized the Hawaiian Islands in the early Holocene or the late Pleistocene. Another bat, now extinct, successfully colonized Hawaii even earlier, flourished in the Hawaiian Islands since at least 130,000 YBP (Olson and James 1982), and probably survived until at least 1,760–1,460 YBP. Despite ample skeletal remains (Olson and James 1982), this bat has yet to be described officially, but appears to be related to lasiurine bats. This now extinct bat, the putative “lava tube bat,” was slightly smaller in cranial and forearm measurements than the extant Hawaiian Hoary Bat (N. Simmons, personal communication). It has been suggested that the ancestral stock may have been Lasiurus borealis. Thus, two species of bats, probably both lasiurines with origins in North America, have successfully colonized Hawaii.

Lasiurine bats are the only bats found in the Americas that seem to have a proclivity for successful colonization of distant oceanic islands. Both L. cinereus and L. blossevillii (as the subspecies brachyotis) occur on the Galapagos Islands, 925km W of Ecuador (McCracken et al. 1997). Lasiurus cinereus, L. seminolus, and Lasionycteris noctivagans regularly occur on Bermuda (Hall 1981; Van Gelder and Wingate 1961; Yates et al. 1976). The distance from North Carolina to Bermuda overwater is an impressive 1,064 km. Even more impressive Lasiurus cinereus occasionally appears on Iceland, and there is a single record of this bat from the Orkney Islands north of Scotland (Nowak 1999). From Newfoundland, the northeastern limit of distribution for the hoary bat in North America to Iceland is 2,428 km. Despite the extreme distance that would need to be covered in a single uninterrupted flight to Iceland, hoary bats have apparently made such flights on multiple occasions. However, the longest overwater flights with subsequent successful colonization for any bat (or dispersal by a land mammal) are from mainland North America to the Hawaiian archipelago. The shortest distance from San Francisco to Hawaii is 3,665 km, with this landfall at the eastern tip of Maui Island. Of course even this flight would be considerably longer if immigrating bats did not maintain the minimum straight line flight distance or if they made landfall on a different island.

We have timed Hawaiian Hoary Bats flying directly between two points while foraging and calculated their flight speed as 11 m s⁻¹. At that speed, a bat would take 3.86 days to reach Hawaii. To a flying animal’s advantage, the prevailing trade winds in the North Pacific usually would assist flight along the projected course from California to Hawaii. In present times, trade winds between North America and Hawaii typically are about 7 m s⁻¹ and blow from the NE in this region about 90 % of the time (Fig. 10.1). Furthermore, it is not unusual for periods of higher wind speed to occur. Consequently it is plausible that a bat departing from the West Coast of North America, following the trade winds, would achieve a moderate tailwind assistance directed towards the Hawaiian archipelago. In a storm, it is easily possible that winds could reach speeds of 25 m s⁻¹ or more. We can imagine ancestral hoary bats flying near San Francisco or even more likely the Farallon Islands (see below) getting swept on a southwesterly direction by a wind storm. Such a scenario is easily imaginable given the autumn aggregation of hoary bats of both sexes on Southeast Farallon Island, 48 km offshore from San Francisco (Cryan and Brown 2007). As the winds abate, these bats are far at sea and are committed to continue flying in the direction of least resistance with the normal trade winds. The main volcanic islands of the Hawaiian archipelago stretch 620 km in an alignment nearly perpendicular to the approach of a bat that would be flying from California with the trade winds. With rain forest and a rich biomass of nocturnal flying insects, Hawaii would therefore present an obvious and welcome sight to an exhausted hoary bat.

Fig. 10.1

Prevailing wind currents in the northeastern Pacific Ocean with five NOAA weather stations used for historical wind speed values indicated by red triangles. The shortest distance from North America to Hawaii is indicated by the black line. Note that prevailing winds off coastal northern California lead directly to the Hawaiian archipelago

To go beyond simple speculation of a colonization flight from the Farallon Islands to Maui, it is necessary to examine in more detail key physiological constraints for a bat of this size and the particular wing morphology and flight mechanics of the hoary bat. Could a hoary bat fly that long and that far from mainland North America to Hawaii without running out of two vital resources: energy and water? In this chapter we input the best available data into a series of models and calculations from a viewpoint of energy and water storage and expenditure, flight dynamics, and meteorological conditions to suggest how hoary bats may have managed the successful colonization of Hawaii some 10,000 years ago.

We conducted a three-stage analysis to determine the conditions under which hoary bats may have originally colonized Hawaii. The first, and most obvious, question is whether hoary bats could store enough fat to provide the energy necessary to fly from California to Hawaii. To answer this and subsequent questions discussed below, we used the software program FLIGHT [v 1.2, and the companion guide, Modeling the Flying Bird (Pennycuick, 2008)]. Though primarily designed for migratory birds, the principles involved work equally well for any flying vertebrate (birds, bats, pterosaurs) and the software has been used successfully for studies of bats (e.g., Grodzinski et al. 2009; McGuire et al. 2012). Users input basic morphometric and body composition values and the program can then be used to model flight, including long-distance migration, from aerodynamic principles. The program simulates a migration in 6-min interval, calculating the fuel requirement and hence mass change in each interval, as well as several other factors such as wingbeat frequency, flight speed, and distance traveled. At each 6-min interval, the remaining fuel stores and body mass are recalculated, repeating until the fat store is completely exhausted.

The second step of our modeling process investigated the potential for water to become a limiting factor in the flight range of hoary bats. If nutrient stores are completely exhausted, the animal will no longer be able to power flight and will fall into the ocean without successfully colonizing Hawaii. Similarly, if the animal becomes excessively dehydrated, unable to maintain water balance, it will not be possible to continue flying. The output of the FLIGHT simulations considers only the former scenario. Therefore, we used these simulations as a base model and developed our own calculations to determine if water balance may become a limiting factor. We considered various scenarios of water loss to determine under which conditions water may be a more limiting resource than fat.

The final component of our calculations is determining what role the prevailing trade winds may have played in the original colonization flights. Our energy and water balance models assume the animal is flying in still air. Such an assumption is clearly not valid. As described above, the prevailing trade winds blow in such a manner that a bat departing from California would likely receive some degree of tailwind assistance (and guidance by following the path of least resistance) on a flight to Hawaii. For each model, we calculated the minimum tailwind speed required (or headwind that could be tolerated) for a bat to fly 3,665 km to Hawaii before either fat or water was depleted. We know that bats have successfully colonized Hawaii at least twice and thus our objective was not to determine if such flights are possible, but rather to provide a context to the likelihood of such flights and the role of environmental conditions. Could a bat only reach Hawaii if blown by gale force storm winds, or could such a flight be reasonably expected to occur under regularly expected environmental conditions? If the predicted environmental conditions are not exceptional, there may be implications for repeated colonization events and a continued influx of genetic material to the population.

2 Methods and Assumptions

As with all models, assumptions must be made; however, there is a solid body of data for many of the variables that contribute to a robust modeling process. The following outlines the data we have used as well as the explicit assumptions and approximations we have made to arrive at all the inputs into our model. Additional assumptions are made implicitly by accepting the default configurations of the FLIGHT software. Our first assumption was that the hoary bats that first colonized Hawaii were morphologically similar to modern North American Hoary Bats (L. c. cinereus). Therefore, all morphological and physiological parameters are determined from North American Hoary Bats. The values of the various parameters input into our models and calculations are given in Table 10.1.

Table 10.1 Values used for FLIGHT models and water loss calculations

Perhaps the most crucial information input into FLIGHT are the morphometric dimensions of the subject, specifically wing span and wing area, from which aspect ratio is calculated. Aspect ratio determines aerodynamic and energetic efficiency in flight (Norberg and Rayner 1987) and thus is a crucial measure when modeling an extreme flight. We used values specific to North American Hoary Bats (L. c. cinereus), obtained from Norberg and Rayner (1987).

The remaining critical values for our simulations pertain to body composition. Flight range is largely determined by the amount of fat as a proportion of total body mass (how much fuel is in the tank?). The FLIGHT model calculations assume that fat is the primary fuel source with ~5 % of the total energy coming from protein sources, consistent with observations of migratory birds (Jenni and Jenni-Eiermann 1998; McWilliams et al. 2004). While the proportion of fat determines the energy available for flight, the fat-free mass is more important to consider for water balance. Fat is stored nearly anhydrously while lean tissue contains 70–80 % water by mass. Therefore, fat-free body mass was another important component of our model.

In the course of other research, one of us (LPM) has gathered information about the body composition of North American Hoary Bats. Hoary bats were collected in late summer and during spring migration. To avoid potentially confounding effects of reproductive physiology, we limit ourselves here to only male bats. For details on collection methods, see McGuire (2012). Total body mass was recorded from each bat before the pectoralis muscles were carefully dissected and weighed. The carcass was dried to a constant mass at 70 °C to determine total body water and then extracted with petroleum ether for 6 h in a Soxhlet apparatus to determine total fat mass (correcting for tissue subsamples). We determined the minimum, mean, and maximum fat fraction (fat mass/total body mass), fat-free mass (total body mass–fat mass), and flight muscle fraction (pectoralis mass/total body mass) among all bats sampled. The range of fat fraction in L. c. cinereus was relatively small, so we also considered the greatest fat fraction [measured by quantitative magnetic resonance (McGuire and Guglielmo 2010)] observed in a recent study of fall migrating Silver-Haired Bats (Lasionycteris noctivagans; McGuire et al. 2012) which was substantially greater than any values observed in our sample of hoary bats. Such a large fat fraction likely represents the extreme range of what may be expected to occur in a migrating hoary bat. We found that varying flight muscle fraction had negligible effects on the outcome of the model and conservatively used the minimum observed flight muscle fraction, thus reducing the number of variable parameters in the analysis.

FLIGHT allows the user to input starting and cruising altitudes. A bat departing from the California coast would begin flying at approximately 0 m asl, and we estimated a cruise altitude of 500 m asl. In practice, for long-distance flights (i.e., the 3,665 km required to reach Hawaii), the flight altitude does not dramatically affect the results of the simulation. Assuming a cruising altitude of 500, 1,000, or even 2,000 m merely changes flight range approximately 1 % of the total flight range. Therefore, although estimates of cruising altitude in migrating bats are lacking, the actual altitude chosen has little impact on the conclusions of our models.

A bat flying over the Pacific Ocean would have no exogenous source of fresh water. Assuming that endogenous metabolic water production is the only water input once a flight has begun, we calculated respiratory water loss based on different water vapor density deficits. Total evaporative water loss is the sum of respiratory water loss and cutaneous water loss (assumed to be 10 % of respiratory water loss; Carmi et al. 1992). Thus, when an animal cannot drink free water, water balance is maintained as evaporative water loss is replaced by metabolic water production.

Sophisticated and detailed calculations of total evaporative water loss require far more detailed information than is currently available for flying bats. One method of calculating respiratory water loss would be to know to what temperature the bats are able to cool their expended breath (then assume saturation) and the temperature and relative humidity of the ambient air. With each breath, bats will inspire a lung volume of ambient air and exhale a lung volume of saturated air. Based on the temperature of those gases, one could determine the water vapor densities in g m⁻³ and the difference is the respiratory water loss. To avoid assumptions regarding ambient temperatures and humidities and the ability to cool expended breath, we ran calculations at three different water vapor density deficits which cover a range of ecologically relevant scenarios. In a study of evaporative water loss in flying Swainson’s Thrushes (Catharus ustulatus), Gerson and Guglielmo (2011) ran water vapor density deficits of 3.2 and 13.2 g m⁻³, arguing that such conditions are relevant to the flight of migrating Swainson’s Thrushes. Songbirds migrating across the Sahara desert experience deficits of approximately 23 g m⁻³ (Schmaljohann et al. 2008, 2009). Based on this range, we evaluated water balance at deficits of 3, 10, and 20 g m⁻³. The 20 g m⁻³ scenario is likely far more extreme than would be experienced over the ocean (more comparable to flying over a desert), but provides context to the two other scenarios which are more realistic.

We calculated total initial body water based on the mean of the proportion of total body water in the hoary bats described above (McGuire 2012). We assumed the critical water balance threshold (below which dehydration precludes survival/continued flight) was 30 % loss of the total body water pool (Carmi et al. 1992). Hoary bats have been reported to survive water loss equivalent to 28 % total body mass with no ill effects (Shump and Shump 1982), so 30 % water loss should be a realistic limit. Shump and Shump (1982) also reported that hoary bats have lower rates of evaporative water loss than several other bats in the family Vespertilionidae, perhaps due to their dense pelage and to the presence of pelage on the wing and tail membranes (an intriguing observation which, if water loss is a limiting factor to long-distance flight, may partially explain why lasiurines appear more likely to colonize remote landmasses than other vespertilionids). Finally, we took tidal volume from Canals et al. (2005), which reports lung volume for South American L. c. villosissimus, and we therefore assume that lung volume is similar for L. c. cinereus.

In each 6-min interval of the FLIGHT model, the number of breaths can be determined given the wingbeat frequency (output by the model) and the respiratory frequency (breaths × wingbeat⁻¹). We assumed the respiratory frequency to be 1 (as reported in Suthers et al. 1972). Therefore, given the number of breaths and the respiratory water loss per breath (plus 10 % to account for cutaneous water loss), we calculated the total water loss in each interval. To calculate the total metabolic water production, we determined the water that would be produced based on the catabolism of fat and protein. Fat catabolism produces 0.029 g H₂O kJ⁻¹ and protein produces 0.155 g H₂O kJ⁻¹ (Jenni and Jenni-Eiermann 1998). The amount of each fuel burned (in kJ) is reported by the software for each interval. Based on these values, we calculated metabolic water production (and hence, net water loss) and the remaining total body water at each interval. It was then possible to determine if/when water balance became a limiting factor.

Ultimately our models will be affected importantly by (1) initial proportion fat (how much energy is available to burn), (2) initial fat-free mass (how much body water is available), and (3) water vapor density deficit (rate of evaporative water loss). There are 4 estimates of fat proportion and 3 estimates of initial fat-free mass, hence 12 base models (fat and fat-free combinations) that use the FLIGHT simulations to predict flight range assuming no winds and no water limitation. After running the 12 base models and the 3 water loss conditions for each (36 conditions total), we evaluated which conditions were energy or water limited. If the simulation exhausted fat reserves before reaching Hawaii (3,665 km), or if total body water pool dropped more than 30 %, we increased airspeed by adding a tailwind, optimized such that the bat reached Hawaii exactly as fat/water stores were exhausted.

3 A Model for Colonization of Hawaii Hoary Bats from North America

3.1 Is Fat Limiting?

From the base FLIGHT models, we examined how long and how far a hoary bat could fly nonstop until fat stores were completely exhausted (temporarily ignoring the effects of water balance and wind). The flight range is determined strictly by fat fraction. Potential flight distance is largely independent of the size of the bat, although the flight speed and hence duration will vary (Pennycuick 2008). Assuming neutral winds, energy is limiting for most conditions, but at the highest fat proportions in our models, bats could reach Hawaii before fat stores were exhausted (Fig. 10.2). The model predicts that a bat with a fat fraction of approximately 0.24 could just reach Hawaii without tailwind assistance (assuming water is not limiting, below). In our models we only considered the body composition of males, but female hoary bats regularly carried larger fat stores than males (McGuire 2012). During spring migration, one-third of the females examined by McGuire (2012) had fat fractions >0.17, suggesting the upper limits of this curve represent realistic body composition expectations.

Fig. 10.2

Total possible flight distance based on energy stores without considering wind or water limitations. The solid horizontal line indicates the distance from southern California to Hawaii. At the highest fat fractions (>0.24), hoary bats could reach Hawaii without any tailwind assistance (assuming water is not limiting). Only the highest fat-free mass scenario is plotted here for simplicity. At the highest fat fractions, lower fat-free masses would be able to fly a few tens of kilometers farther

Even at fairly modest fat proportions, the model predicts the bats could fly for long durations (Fig. 10.3). Maximum flight duration ranged from 27.2 to 126.3 h (5.26 days!) depending on the initial proportion fat. Flight duration was somewhat variable depending on the size of the hypothetical bat considered. There is little difference in flight duration when fat stores are slight; at a fat fraction of 0.076, flight duration ranged from 27.2 to 30.9 h. However, as the proportion of fat increases, the differences in flight duration corresponding to variation in fat-free mass are more exaggerated. At the highest fat proportion, maximum flight duration was between 110.0 and 126.3 h. Holding fat fraction and wing morphometry constant, but varying body mass, both drag and wing loading will be affected, resulting in variable flight duration predictions. These models assume fat stores are completely exhausted (i.e., absolutely zero fat remaining at termination) which is probably not realistic so perhaps these are slight overestimates.

Fig. 10.3

Predicted maximum flight durations for hoary bats with varying body composition. The three lines indicate the low (solid), mid (dashed), and high (dotted) estimates of fat-free mass included in the models (see Table 10.1 for values)

Our model also predicts airspeed from aerodynamic principles. In the model, true airspeed varies within a flight based on rules regarding maximum power speed in the initial climb, maintaining constant muscle work in the early period of level flight and ultimately maintaining the maximum range speed as fat stores and hence body mass declines. Predicted airspeed varied from 5.9 to 12.5 m s⁻¹, and mean airspeed among flights ranged from 6.1 to 10.8 m s⁻¹. Such estimates are reasonable compared to direct observations of hoary bats. Foraging Hawaiian Hoary Bats averaged 11 m s⁻¹ (range 5.9–15.2 m s⁻¹) (Bellwood and Fullard 1984; Jacobs 1996; FJB unpublished data). North American Hoary Bats average 7.7 m s⁻¹ in foraging flight, and some individuals fly as fast as 12 m s⁻¹ (De la Cueva Salcedo et al. 1995). These authors did not control for the effects of wind speed, which may increase or decrease the apparent flight speed depending on direction, and thus are not directly comparable to the flight speeds estimated by our models. Furthermore, optimal flight speed theory (Hedenstrom and Alerstam 1995) predicts different speed when foraging and migrating. Consistent with this theory, commuting bats do indeed fly faster than foraging bats (Grodzinski et al. 2009). Thus, while previous hoary bat flight speed observations are not directly comparable, they indicate that the flight speeds predicted by the model are reasonable and may in fact represent underestimates.

3.2 Is Water Limiting?

Equally important to depletion of energy stores in considering the likelihood of a bat to successfully continue prolonged flight over an ocean barrier is depletion of body water. Might a hoary bat run out of adequate stores of body water before it ran out of energy? Which would be the more limiting during a flight from San Francisco to Hawaii?

We considered whether water balance was more limiting than energy balance (i.e., would water run out before fat, regardless of whether the bat could reach Hawaii or not?). Energy balance was more limiting than water balance at 3 g m⁻³ deficit. Conversely, water was always more limiting than energy at 20 g m⁻³. At 10 g m⁻³ the limiting resource was related to the combination of fat-free mass and fat fraction. At lower fat proportions (0.076, 0.115), water was more limiting than energy when fat-free mass (largely the size of initial water stores) was low (17 and 20 g, respectively). At higher fat proportions, water was always more limiting than energy.

Accounting for scenarios where water was more limiting than energy, we reevaluated the potential flight range (Fig. 10.4). Comparing Fig. 10.4 with Fig. 10.2 clearly illustrates the potential of water balance to limit flight range. Only bats with the highest fat fraction and lowest water vapor density deficit had sufficient energy and water reserves to reach Hawaii in neutral winds. At high respiratory water deficits, water has a much greater impact on defining flight range than fat. With a 20 g m⁻³ deficit, flight range was limited to approximately 500 km, well short of Hawaii.

Fig. 10.4

Flight range accounting for energy and water limitations assuming neutral winds. Solid, dashed, and dotted lines represent low, mid, and high fat-free mass values. The symbols indicate the respiratory water loss scenarios: low water loss (circles), mid (triangles), and high water loss (squares). See Table 10.1 for values of fat-free mass and water loss rates. Higher water loss scenarios limit flight range, but higher initial fat-free mass provides a larger starting water pool which increases the potential flight range before water becomes limiting. Water balance was never limiting for the low water loss scenario and therefore the three lines are overlaid and equivalent to Fig. 10.2. The horizontal black line at 3,665 km represents the minimum distance from California to Hawaii

3.3 Tailwind Assistance

The next step in the modeling process is the role of tailwind assistance. For each of our 36 model scenarios, we know the theoretical limit to flight range (Fig. 10.4) and the minimum required distance (3,665 km to Hawaii). We determined the minimum required tailwind assistance for the bats to reach Hawaii. Or to put it another way, what tailwind speed would be required to compensate for given energy or water limitations? For practical purposes we assumed a constant wind speed for the duration of the flight and that the wind and flight direction were parallel, both leading a wayward bat on the shortest distance to Hawaiian refuge. If one or the other is off-angle, the vectors will change and for the same wind/airspeed, the ground speed would be lower. Given the prevailing trade winds in the region (Fig. 10.1), such an assumption is reasonable.

To find the minimum required tailwind, we took a fixed wind speed and added it to the airspeed in each 6-min interval of the model output. Adding wind speed and airspeed gives ground speed and thus a new distance estimate can be made for each time interval. We solved for the minimum wind speed necessary to reach 3,665 km in the total flight time possible for a given scenario. In other words, our models assume the bats reach Hawaii exactly as fat and water reserves are exhausted. We determined the range of observed wind speeds in this region of the Pacific Ocean from five weather stations, two island stations, and three environmental buoys [Fig. 10.1; data obtained from National Oceanic and Atmospheric Administration (NOAA) Climate Services http://www.climate.gov/#dataServices]. The five stations had variable temporal coverage, with the most extensive records obtained from Hilo International Airport (1943–present). We constrained the records to only include the months of September–October to coincide with autumn migration and the arrival of hoary bats in California as this is the most likely period when a bat may be swept out to sea and forced to fly to Hawaii. Weather records from these stations include mean and maximum sustained wind speeds recorded at daily intervals. To be conservative, we considered the mean wind speeds (mean 4.51 m s⁻¹, max 25.8 m s⁻¹), rather than maximum sustained wind speed (mean 7.0 m s⁻¹, max 36.0 m s⁻¹). We obtained wind direction information from the Comprehensive Ocean–atmosphere Dataset (COADS, Woodruff et al. 1987) which we obtained through the National Virtual Ocean Data System (NVODS). As for wind speed, we considered the long-term average wind direction for the months of September and October. Required tailwind assistance is highly variable depending on the modeling scenario. Bats with the highest fat proportions and lowest rates of respiratory water loss don’t require wind assistance (Figs. 10.5 and 10.6), rather could reach Hawaii even in the face of a slight headwind! At the opposite end of the range, bats facing a 20 g m⁻³ water deficit would require tailwinds of at least 45 m s⁻¹ (162 km h⁻¹), corresponding to a category 2 (or higher) cyclone. Therefore, if bats face 20 g m⁻³ water deficit, colonizing Hawaii represents a truly rare and remarkable occurrence. At a more moderate 10 g m⁻³ deficit, required tailwinds ranged from 6 to 32 m s⁻¹ (22–115 km h⁻¹), not far off of our originally speculated 25 m s⁻¹ storm winds in some cases, and perhaps even much less.

Fig. 10.5

Required tailwind assistance to reach Hawaii before water or energy reserves are exhausted. Symbols and line styles indicate the various modeling scenarios as in Fig. 10.4. The horizontal gray lines indicate the wind speed thresholds for cyclone categories 1–5. At the highest water loss estimates, our models predict that bats could only have reached Hawaii under extreme cyclone winds (category 2 or greater)

Fig. 10.6

Distribution of required tailwind speeds, replotted from Fig. 10.5 to provide alternate perspective. The horizontal boxplot indicates the historical range of sustained wind speeds (data from 5 NOAA weather stations indicated in Fig. 10.1). Note that observed wind speed (from NOAA) is always positive, independent of direction, whereas model scenarios include negative wind speeds to indicate possible headwinds. It is clear that many of our modeling scenarios, accounting for both energy and water limitations, are plausible given the typical wind speeds in the regions

In Fig. 10.6, we have overlaid a histogram of the required tailwind speeds for each of our modeling scenarios and a boxplot of the historically observed range of wind speeds. With the exception of the high water loss scenario, many of our models require tailwinds that could be reasonably expected in the region. Of our 36 modeling scenarios, 17 (nearly half) require tailwind assistance within the range of observed wind speeds. Considering maximum sustained wind speeds, 24 (two-thirds) of our models fall within the range of observed values. Furthermore, seven scenarios required <10 m s⁻¹ tailwinds as may be regularly expected due to the easterly trade winds. Therefore, the climatic conditions needed for bats to colonize Hawaii may not be exceptionally infrequent in occurrence either in contemporary times or at least since the end of the Pleistocene, ~11,000 YBP. During the Pleistocene, prevailing winds would have fluctuated greatly between glacial and interglacial times.

4 Modeling Conclusions

The modeling exercises we conducted for this chapter have examined the physiological, behavioral, and morphological mechanisms and adaptations that we judge were present in the ancestral population in North America from which hoary bats colonized Hawaii. Most of these attributes probably remain relatively unchanged in North America, while the Hawaiian Hoary Bat populations have undergone some noted changes either due to genetic drift (such as founder effects) or natural selection in a novel environment.

Our model provides insights into specific questions crucial to a 3,665 km one-way flight by the ancestral bats that traveled to Hawaii. Can hoary bats carry enough fat to sustain them on a flight to Hawaii? Our model suggests they could. Would dehydration be an issue on that flight? It is apparent that certain dehydration scenarios (Figs. 10.4 and 10.5) have a greater impact on colonization likelihood than energy balance. The extreme variation in the water balance models arises from greater uncertainty in the respiratory water loss processes of flying bats. We have reliable, empirically measured data regarding body composition and can model energy expenditure based on established aerodynamic principles. How bats regulate and compensate for respiratory water losses is poorly understood. How much wind assistance would the bats need to reach Hawaii? Depending on the scenario, some hoary bats could have arrived in Hawaii with light wind assistance or even with slight headwinds; however, prevailing trade winds are likely to have assisted flights with wind speeds in the range of 7–25 m s⁻¹ common throughout much of the year.

While our model predicts that hoary bats can store adequate fat to power continuous flight up to 5 day duration, such flights are well beyond the scope of flight behavior known in any other bat species. Even large pteropodid bats have not been successful in colonizing water gaps greater than 1,000 km (Bonaccorso and McNab 1997) across the Pacific Ocean. However, comparable flights do occur regularly in bird migration. Barnacle Geese (Branta leucopsis) fly 14 h nonstop during migration (Butler et al. 1998). Great Knots (Calidris tenuirostris) fly 4 days nonstop from Australia to China (Pennycuick and Battley 2003). Pacific Golden Plovers (Pluvialis fulva) fly nonstop from Alaska to Oahu (Hawaii) taking an average of 4 days over the 4,900 km passage (Johnson et al. 2012). The champions of continuous flight among avian migrants are Bar-Tailed Godwits (Limosa lapponica) which make nonstop flights from Alaska to New Zealand that regularly take >7 days of continuous flight (Battley et al. 2012). Impressive transoceanic flights are also known from smaller songbirds. The 25 g Northern Wheatear (Oenanthe oenanthe) is similar in body size to North American Hoary Bats and is suspected of making a nonstop migratory flight lasting 4 days from Baffin Island in the Canadian Arctic, across the Atlantic Ocean to the British Isles (Bairlein et al. 2012). Even tiny hummingbirds are known to migrate across the Gulf of Mexico (Lasiewski 1962). Therefore, while such extreme endurance flights are not known to regularly occur among bats, such flights are certainly possible.

5 Post-colonization Ecology and Evolution of Hoary Bats in Hawaii

5.1 Hawaiian Founder Population

Neither the size of the original founder population nor the number of subsequent colonizing events for hoary bats arriving in Hawaii is known; however, estimation of each can possibly be inferred from future examination of molecular genetics in modern populations of Hawaiian Hoary Bats. Theoretically, a population can be founded by a single pregnant female, but the loss of diversity would be severe, and inbreeding depression in such a small founding population would make extinction much more likely (Amy Russell, personal communication). Looking at the founding of Triaenops rufus (Hipposideridae) populations in Madagascar from a source population on the African continent, Russell et al. (2008) found their simulation data to be consistent with a founding population size as small as ~10–25 individuals which subsequently led to speciation events in Madagascar in this genus.

Generally from late August to the end of October each year, hoary bats (both in North America and in Hawaii) swarm in social groups during evening flights, during which time copulations occur (Cryan and Brown 2007, Christopher Todd, personal communication). A single colonization event, if originating during a period of L. cinereus autumnal swarming along the North American coast or on offshore islands such as the Farallons, possibly could have resulted in the arrival of individuals in Hawaii sufficient for a founder population of similar size to the model demonstrated for Trianeops (Russell et al. 2008). Subsequent arrivals of individuals or small groups in separate migration events would have enriched the genetic diversity of the population, particularly early in the founding process.

5.2 Morphological Divergence of Modern Hawaiian Hoary Bats

Hawaiian Hoary Bats have undergone significant character displacement from their mainland counterparts in a number of physical traits. Phenotypic divergence since arrival in Hawaii has included appreciable reductions in body size compared to the presumed ancestral stock in North America. These traits include smaller body mass, skull size, and forearm length (Jacobs 1996; Tomich 1986; FJB personal observation). Jacobs (1996) found that adult female semotus were on average 45 % smaller in body mass and 8.4 % smaller in forearm length than cinereus. Although he made no conclusion regarding males, our own recent data (FJB, personal observation) indicates similar proportionate reductions in male body size between these two subspecies.

The smaller body mass of Hawaiian compared to North American Hoary Bats also has resulted in a lesser wing loading (Jacobs 1996) in the Hawaiian subspecies which in turn makes possible a more maneuverable and acrobatic flight. Because of the dominance of fast flying moths in the nocturnal insect fauna of Hawaii (Belwood and Fullard 1984; Bonaccorso, personal communication), the ability to chase and intercept the evasive flight of moths is of critical importance to hoary bats in Hawaii. In all habitats from sea level to 1,600 m, moths are the most abundant taxa in Hawaiian Hoary Bat diets (Bonaccorso et al. 2013).

Given that hoary bats have been resident in Hawaii for at least 5,500 years, one may wonder if there is significant movement between island populations within Hawaii or if there is little gene flow between islands. Comparing skins and skulls of semotus from Kauai, Maui, and Hawaii Islands, there are striking differences in pelage coloration between islands (FJB, personal observation). Bats from Maui have much more distinct fur tipping to render the typical “hoary” appearance similar to mainland cinereus (Fig. 10.7). Despite the small water gap (46 km) between the closest portions of coastline separating Maui and Hawaii, the bats from the island of Hawaii in general possess a bright red-brown under coat with much less white tipping on the fur. The lack of white tipping on adult male semotus is particularly noticeable on the crown of the head (Fig. 10.7). In fact bats of Hawaii Island were more similar in appearance to those from the most distant major island, Kauai. The similarity of Maui bats to mainland cinereus in body size and pelage color, and their distinct difference from bats on Hawaii Island in these characteristics, suggests the possibility of multiple colonization events for separate island groups in Hawaii, with relatively little genetic exchange among islands post-colonization. Future molecular genetics studies sampling these different islands may resolve such questions.

Fig. 10.7

Comparison of pelage coloration in adult Lasiurus cinereus semotus from Hawaii Island (female upper and lower left, male lower right) and L. c. cinereus from North America (upper right); L. c. semotus adult males from the island of Maui more closely resemble L. c. cinereus (Photos of L. c. semotus by Jack Jeffrey and L. c. cinereus by Adam Miles)

5.3 Present Hawaiian Distribution and Habitat Use

Hawaiian Hoary Bats presently occur on Kauai, Molokai, Oahu, Maui, Lanai, and Hawaii (Tomich 1986). They probably occur on Kahoolawe; however, no effort at ascertaining this is shown in the published literature. Furthermore, Hawaiian Hoary Bats are recorded from sea level to virtually the summit of Mauna Loa Volcano, which is 4,169 m above sea level. We have observed this bat flying over almost every conceivable habitat found in Hawaii including over saltwater embayments, rivers, forests (including both virgin native forest and forest dominated by nonnative trees), grasslands and pastures, eucalyptus plantations, fruit orchards, within deep gulches, as well as over suburban and urban landscapes. Occasionally, Hawaiian Hoary Bats fly and forage over almost barren lava flows despite the apparent scarcity of insect prey that might be expected. Conducting echolocation surveys on both the Big Island of Hawaii and Kauai, foraging activity is common on both windward and leeward sides of these islands. The windward coasts and mountain slopes of the major Hawaiian Islands are incredibly wet and are (or were) covered by rain forests where rainfall in many locations exceeds 4,000 mm year⁻¹, whereas the leeward slopes and coasts of these same islands are much drier. Thus, Hawaiian Hoary Bats truly are extreme habitat generalists.

The subfossil record substantiates that hoary bats arrived in Hawaii at least 5,500 YBP and perhaps as much as 130,000 YBP (Olson and James 1982). Molecular data Morales and Bickham 1995), as well as zoogeographic considerations, and wind patterns suggest that hoary bats arrived in Hawaii from an origin somewhere along the coastline of North America between northern California and southern Alaska. DNA analyses may confirm the region of origin by comparing Hawaiian and North American populations. Our modeling scenarios suggest that the conditions necessary for hoary bats to successfully colonize Hawaii may be reasonably expected and thus provide suggestive evidence that hoary bats may have reached Hawaii more than once. Molecular evidence from DNA may also shed insights into whether more than one colonization event in Hawaii occurred and the degree of continuing exchange among islands. Since their arrival in Hawaii, hoary bats have diverged from the ancestral stock. Both physical and behavioral characteristics distinguish this bat from their North American counterparts. Furthermore, present-day L. c. semotus from the neighboring islands of Maui and Hawaii have striking differences in cranial measurements and pelage coloration. Could these two islands have been colonized independently with low subsequent gene flow between them? Further investigation is required to fully answer the question of how bats became the only terrestrial mammals to colonize the Hawaiian archipelago, one of the most isolated insular landmasses on Earth.