Correspondence Luísa Rodrigues, Instituto da Conservação da Natureza e da Biodiversidade, Rua de Santa Marta 55, 1150-294 Lisboa, Portugal. Tel: 00213513507900; Fax: 00213513507987 Email: firstname.lastname@example.org
Regional migrations are important elements of the biology of bats, but remain poorly understood. We obtained a large dataset of recoveries of ringed Miniopterus schreibersii to study the patterns and drivers of migration of a Mediterranean cave-dwelling bat. In spite of the mildness of Mediterranean winters, in average years bats hibernated, and few movements were recorded during this period. After hibernation, females migrated to spring roosts, and again to maternity roosts just before parturition. This late arrival at nurseries could be a strategy to avoid a harmful build-up of parasites. Soon after the juveniles were weaned, the mothers migrated to the roosts where they spent autumn and sometimes also winter. Juveniles remained in the warm nurseries longer, presumably because high roost temperatures speed up growth. The pattern of migration of males was similar to that of females, but they left hibernacula later and remained more mobile during the maternity season. They also arrived at the hibernacula later, possibly because they needed time to build up fat stores after the energetically costly mating season. Maternity colonies spent the yearly cycle in well-defined home ranges (mean=19 030 km2), which overlapped greatly. Bats were furthest from the maternity sites during hibernation, but even then 80% remained within 90 km of them. Each hibernaculum attracted bats from multiple nurseries, from within a mean range of 10 770 km2. We tested two potential drivers for migration – temperature in the roosts and at the foraging areas – but our results supported only the first one. Bats migrated to reach the roosts most thermally suited for each phase of their life cycle, indicating that roost temperature and associated metabolic advantages are key drivers for regional migrations of cave-dwelling bats.
Although most bats are sedentary, migration plays an important role for many species, particularly those living in highly seasonal environments (Fleming & Eby, 2003). Temperate zone bats are known to make migrations at various scales (Strelkov, 1969). Some species make long-distance two-way movements with a strong north-south component (e.g. Pipistrellus nathusii and Lasiurus cinereus), but most make only regional migrations. These movements usually do not follow a clear directional geographic trend, even when they involve two-way seasonal migrations of several hundred kilometres.
The temporal and spatial patterns of migration are important parameters of the biology of bats. Understanding them is also important because many species are threatened (Mickleburgh, Hutson & Racey, 2002) and information on their migratory behaviour is important to plan conservation measures. To conserve a bat species, it may be necessary to protect the network of roosts used throughout the yearly cycle.
Annual cycles of temperate zone bats often involve multiple changes of roost. The most common seasonal movements are hibernation–spring, spring–maternity, maternity–mating and mating–hibernation, but this varies among species and even among populations of the same species. Some migratory species also have sedentary populations, and in some populations only a proportion of individuals migrate (Russell, Medellín & McCracken, 2005).
The long-distance two-way movements that follow a north–south pattern in the temperate zones are probably induced by climate seasonality (Fleming & Eby, 2003). However, the motivation behind regional migrations that do not follow clear geographic trends is less obvious.
It is assumed that temperate zone bats make regional migrations to reach roosts that have microclimatic characteristics particularly suited for each season (McNab, 1982; Baudinette et al., 1994). Some migrations are prompted by the need to move to caves with temperatures suitable for hibernation, but the explanation for movements at other seasons is less clear. Roost temperatures may also be the drivers for these migrations, but there are other plausible hypotheses to explain them. Because some of the regional movements are long enough to place bats in areas with different climates, these migrations could be driven by the advantages of accessing foraging areas with better ambient temperatures. In general, higher night temperatures are more favourable for foraging insectivorous bats, because insect availability, within the same type of habitat, tends to increase with temperature (Jones, Duvergé & Ransome, 1995).
To our knowledge, nobody has compared these hypotheses to explain regional migrations. We did this by testing two predictions: if roost temperatures are the reason why bats migrate, then temperatures at the new roosts should be different from those of the old roosts; if ambient temperatures are driving the migrations, bats should be moving to climatically different regions. We tested these predictions using data on the regional migrations of Miniopterus schreibersii (Kuhl, 1817) in Portugal. In addition, we describe the spatial and temporal patterns of its migrations, based on the results of an 18-year long study and over 8000 recaptures of ringed animals. This was, possibly, the most extensive study of bat migration ever carried out.
It was once considered the most widespread bat in the world (Palmeirim & Rodrigues, 1995), but recent molecular studies have demonstrated that this taxon is a paraphyletic assemblage comprising several species. Following the latest taxonomic studies, we use the name M. schreibersii for Mediterranean, Miniopterus fuliginosus for Asian, Miniopterus oceanensis for Australian (Tian et al., 2004) and Miniopterus natalensis for South African (Miller-Butterworth et al., 2005) populations.
It roosts almost exclusively underground, forming large colonies throughout most of the year. It is the most abundant cave-dwelling species in Portugal, and 12 nurseries and 15 hibernacula (used regularly by more than 100 individuals) are known. The country has been thoroughly surveyed and so it is unlikely that large colonies, particularly maternity colonies, remain unknown.
The migratory behaviour of bats is usually studied using ringed individuals, but a good knowledge of the movements requires large numbers of recaptures. From 1987 to 2005, around 36 000 bats were marked with smoothed metal lipped rings [made initially by Lambournes and later by Porzana (UK)], designed to minimize the risk of damage to the wing. Bats were ringed under licence from the ‘Instituto da Conservação da Natureza’. Following resolutions 4.6 and 5.5 of EUROBATS, we used 4.2 mm rings. Injuries were rare, but 0.007% of the recaptures had developed either tissue irritation under the ring or, in a few cases (0.001%), the ring had moved from the forearm to the upper arm and was partly covered by skin. To check whether this was affecting the foraging ability of the bats, we compared the weight of injured bats with that of unringed animals captured on the same day and at the same site (same sex and age class, using paired t-test) and no significant difference was found (n=20). Likewise, to assess whether rings caused a loss of foraging ability in the first few days after ringing, when the bats are most likely to be stressed, we compared the weight of bats recaptured <30 days after being ringed with that of all other animals captured on the same day and at the same site (same sex and age class, using paired t-test) and no significant difference was also found (n=34). Most important colonies have been monitored annually since 1987 and no declines have been detected. Such a low incidence of problems cannot be assumed for other bat species or ring types.
The ringing programme had two phases. From 1987 to 1992, about 27 000 bats were ringed throughout the whole country, to map migration routes and obtain information to establish a network of protected sites. Since 1993, ringing was limited to a few roosts to obtain long-term data on population dynamics. Analyses about the timing of migration used only data from the first phase to avoid biases. Migratory routes were established from at least two recaptures, thus minimizing the impact of unusual behaviour and of possible errors reading the rings. Although we did not visit roosts in Spain, Spanish colleagues reported a fairly large number of recaptures (n=120).
Generally, bats were captured with harp-traps while emerging from their roosts, but when this was not possible they were captured inside the roosts by hand or hand-net. To minimize disturbance, we avoided making captures in nurseries with flightless young. We estimated the size of compact colonies using the approximate surface area covered by the bats and a reference density of 2000 individuals m−2 (L. Rodrigues, J. M. Palmeirim, pers. obs.).
Because females are strictly philopatric during the maternity season (Palmeirim & Rodrigues, 1995), females found in a maternity colony in this season were assumed to have given birth there. We delimited colony home ranges utilizing the minimum convex polygon, i.e. the smallest convex polygon encompassing all the sites used by its individuals.
We considered two age classes: adults (≥1 year old) and young-of-the-year, using the ossification of the carpal joints and development of nipples and testes (e.g. Baagøe, 1977) and the pattern of moult (Dwyer, 1963). Weight was taken with a Pesola spring balance, rounded to the nearest 0.25 g. The annual cycle was divided into four seasons: hibernation (16th December to 29th February), spring (1st March to 31st May), maternity (1st June to 25th July) and autumn (26th July to 15th December). However, these dates are only approximations, as there are inter-annual variations. Roost temperature was measured with a digital thermometer with a precision of 0.1 °C. Ambient temperatures are the 1961–1990 normals, provided by the ‘Instituto de Meteorologia’.
When do they migrate?
Bats made multiple migrations during the annual cycle, but the rate at which they changed roosts varied during the year. We present (Fig. 1) the average situation of all study populations, because there are timing differences between years and populations. In fact, in one population the majority of bats do not even migrate.
The season when the fewest movements were detected was winter, when bats generally hibernate and seldom change roost. At the end of winter, most females migrated to new sites, and most changed roost at least once during the spring. Some of the spring caves will later harbour nurseries, but at this time females are often not in their own maternity roost. At the end of spring, the great majority migrated to reach their maternity sites, often just before giving birth in early June. After the juveniles were weaned, the females migrated to caves where most spent the autumn and sometimes the winter. The pattern of migrations of males was broadly similar, but differed in that they left the hibernacula later and made more movements during the maternity season.
Seasonal migrations resulted in major population shifts, which we illustrate here with the variation in numbers of individuals using two caves (Fig. 2). The first roost is a typical maternity site (Fig. 2a), used by a large number of animals between April and July, but empty during autumn and winter. The maximum of 4800 individuals was observed in July, when the numbers were swollen by the young-of-the-year. The sex ratio was the highest in the early months, as females tended to arrive earlier, and by the end of the occupation period most of the animals present were young-of-the-year, which were the last to leave. Although the number of males in this cave was high at times, in general they were roosting away from the nursing clusters.
The second roost is a hibernaculum (Fig. 2b), and its occupation was almost inverse of that of the first cave. It harboured many animals during much of the year (up to 2000), but few, mostly males, during most of the spring and summer. All the females roosting here were not breeding. The proportion of females started to decline in early spring, as they abandoned hibernacula earlier.
Where do they go?
Each maternity colony has a well-defined home range (Fig. 3). Outside the maternity season, the five analysed colonies dispersed over a mean area of 19 030 km2 (6360–32 790 km2). We did not estimate an area for the remaining seven colonies because data were insufficient. The mean ranges are similar for males (17 670 km2) and females (16 230 km2). If we consider only the winter range of the maternity colonies, the mean range shrinks to 9880 km2 (900–16 800 km2). There was an extensive overlap in the home ranges of various colonies.
Outside the maternity season, about 80% of the bats remained within 60 km of their nursery (Fig. 4). The average distances between nurseries and all other roosts used by its animals were very similar for both sexes in three out of the four nurseries for which we had good data (n=112, n=175, n=40, P>0.05). In the fourth colony, males travelled significantly further than females (males=58.3 km, females=44.1 km, t-test, n=356, P=0.027). The mean distance travelled by males of all four colonies was 45.2 km and that travelled by females was 49 km. The longest movement recorded between a nursery and other roost was 260 km (one female, between a maternity and a spring roost), but the longest movement recorded was 306 km (one female, between a hibernaculum and a spring roost, in Spain).
The distances travelled from nurseries to hibernacula tended to be longer than the average for all movements; about 80% of the bats hibernated within 90 km of their nursery (Fig. 5). However, a few individuals flew considerably further (three females and one male were found hibernating 237 km away). The average distances between maternity and hibernation roosts were very similar for both sexes in three out of the four maternity roosts for which we had good data (n=96, n=228, n=433, P>0.05). In the fourth colony, males travelled significantly less than females (males=31.2 km, females=44.3 km, t-test, n=211, P=0.033). The mean distance travelled by males of the four colonies was 45.1 km and that travelled by females was 46.8 km.
Each of the 12 hibernacula for which we had sufficient data gathered bats from a mean area of 10 770 km2 (242–35 707 km2). We did not estimate an area for the other hibernacula because data were insufficient. Figure 6 shows two examples of such ‘catchment areas’, as including all of them on one map would make it difficult to read. For the five hibernacula for which we have good data, the mean areas are similar for males (4675 km2) and females (4707 km2).
Why do they migrate?
To investigate whether bats migrate in order to find more suitable roosting or ambient temperatures, we compared the temperatures of the departure roosts and regions, with those where they settled after migration. We did not include roosts used by <100 bats.
The winter to spring migrations allowed bats to roost in warmer sites (3.2 °C warmer, on average) (Fig. 7a). By contrast, night ambient temperatures (i.e. of potential foraging areas) were very similar in the regions used in the two seasons.
Between spring and maternity seasons, bats migrated to even warmer roosts (3.0 °C warmer, on average), but again there was no difference in the night ambient temperatures (Fig. 7b).
The nursery-to-autumn migration placed bats in much cooler roosts (4.5 °C cooler, on average), but did not change their night ambient temperatures (Fig. 7c).
From autumn to winter, bats migrated to even cooler roosts (3.2 °C cooler, on average, although this result must be interpreted cautiously because it is not statistically significant), but located in regions with similar night ambient temperatures (Fig. 7d).
When do they migrate?
Winters in the Mediterranean are short and mild, and the temperature of underground roosts is high when compared with that of hibernacula further north in Europe. In spite of this, in average years, bats spent the full winter hibernating, and few movements were recorded during this period. This situation is similar to that found for other bat species in colder temperate regions (e.g. Tuttle, 1976; LaVal & LaVal, 1980), but contrasts with what we observed in warmer winters, when the traditional mass migration to hibernacula did not take place and the animals remained more scattered, and travelled more frequently. This reveals flexibility in the hibernation and migratory behaviours of M. schreibersii, which is particularly adaptive for a species that lives near the warm geographical limit of hibernation. It is known that above a certain temperature threshold, which varies with the duration of the winter, hibernation becomes energetically difficult because bats use up their fat stores too quickly (Humphries, Thomas & Speakman, 2002).
The tendency for males to leave hibernacula later than females has already been reported for this species in Catalonia (Serra-Cobo et al., 1998), and has also been observed in Myotis grisescens (Tuttle, 1976) and Myotis lucifugus (Davis & Hitchcock, 1965). Tuttle (1976) suggests that this difference may have a significant adaptive value, reducing intraspecific competition for food resources in spring, when energy demands of pregnant females are higher. However, it could also be explained by the need of females to move to warmer roosts early so that their embryos start developing faster. It has been demonstrated for Myotis myotis that roost temperature influences the rate of development of embryos (Zahn, 1999).
The fact that females usually migrate to their maternity roosts just before giving birth is somewhat surprising, as some of these roosts harbour many other bats species in the preceding months. This behaviour was also observed in M. natalensis (Van der Merwe, 1975) and could be an adaptive strategy to delay the build-up of populations of parasites. It has been suggested that changing roosts minimizes parasite build-up in colonial bats (Zahn & Rupp, 2004), and there is evidence that high ectoparasite loads affect the condition of M. schreibersii, especially in nurseries (Lourenço & Palmeirim, 2007).
In some tree-dwelling species, for example Barbastella barbastellus, maternity colonies change location several times during a season (Russo, Cistrone & Jones, 2005), but in M. schreibersii this was one of the periods when they made the least movements, especially females. The heights of the bars for this season in Fig. 1 are not very representative of the situation, because we did not capture animals during nursing to minimize disturbance. Most other species also seem to remain in the same caves throughout this season (e.g. Lewis, 1995). The explanation for this difference between tree- and cave-dwelling species probably lies in the availability of roosts, because caves are less common than tree holes.
Most animals leave nurseries before August, but the young-of-the-year remained there longer than their mothers. There are several possible explanations for this behaviour, also observed in M. oceanensis (Dwyer, 1966), M. myotis (Rodrigues et al., 2003) and M. grisescens (Tuttle, 1976). Abandoning nurseries early should be advantageous for adults because these colonies tend to be heavily parasitized (Lourenço & Palmeirim, 2007). In addition, moving to roosts cooler than those used by nurseries allows bats to enter torpor during daytime, saving metabolic energy (Lausen & Barclay, 2003). For the young-of-the-year, however, it may be preferable to remain in the warm nurseries, in spite of the heavy parasitic pressure, as torpor in cooler roosts could slow down their development (Tuttle, 1975), resulting in a smaller adult size (Ransome, 1998). Adult females could also leave the nurseries as soon as the young-of-the-year become independent in order to move to roosts used by adult males and thus increase their chances of mating, or even just to avoid competition with their offspring.
Our finding that females tend to migrate to their hibernacula earlier than males is consistent with the results of other European studies on this species (Avril, 1997; Serra-Cobo et al., 1998). The same pattern was observed in M. grisescens (Elder & Gunier, 1978), and Tuttle (1976) suggests that it reduces intraspecific competition for food resources when energy demands of mating males are higher. An alternative explanation is that males expend more energy during the mating season, and so they need more time to accumulate fat stores for hibernation. In fact, in M. myotis the weight of females tends to increase steadily through autumn, whereas that of males goes down during the peak of the mating season and only then starts increasing towards winter (Zahn et al., 2007).
Where do they go?
The fact that each maternity colony had a well-defined home-range, used differently throughout the yearly cycle, is probably a consequence of the high degree of philopatry to roosts occupied at certain times of the year (Palmeirim & Rodrigues, 1995). An alternative explanation for having fixed home ranges is limited knowledge of the terrain and roosts; individuals would remain within the region, that is part of the collective knowledge of the colony.
We also observed that these home ranges are large and overlap extensively, with the same area and roosts being used by individuals from several nurseries. This suggests that interference competition or other types of negative interaction between colonies are not important in defining the limits of the home ranges.
The size of the home ranges is so large (up to 32 790 km2) that it is not likely to be determined by resources that are homogeneously distributed across the landscape. This large size is more easily explained by a dependence on patchy resources, such as suitable underground roosts or specific feeding habitats.
Regional migratory movements and extensive colony home-range overlap bring together bats from different colonies outside the maternity season (Fig. 3). This is important in the genetic structuring of populations, as it facilitates gene flow (Palmeirim & Rodrigues, 1995). In fact, it has been demonstrated that in spite of the strict philopatry of females to their natal site, there is a substantial male-mediated gene flow (Ramos Pereira, Salgueiro, Rodrigues, Coelho & Palmeirim, unpubl. data).
The ‘catchment areas’ of the various hibernacula also overlapped greatly, as each of them attracted bats from several nurseries. Individuals do not necessarily hibernate in the suitable roost closest to their nursery. This may occur because after nursing, bats tend to scatter over a large area, and then choose to hibernate in the suitable roost that is closest to where they are in the autumn.
In several species that make long-distance migrations, females tend to migrate further than males (Strelkov, 1969; Russell et al., 2005), but we did not find this in M. schreibersii. Similar migration distances of both sexes were also reported for the same species in France (Avril, 1997) and in M. oceanensis (Dwyer, 1966).
The average distance between maternity and hibernation roosts was <50 km, which is below that reported for the same species in eastern Iberia (120 km; Serra-Cobo et al., 1998) and Serbia (100 km; Paunovic, 1998). This suggests a greater availability of suitable roosts in our study area.
In spite of the very large number of recaptures, the longest movement detected in this study was 306 km, considerably less than the record for this species (833 km in Spain; J. Benzal pers. comm. in Hutterer et al., 2005). This could also be explained by a relative abundance of suitable roosts, but may be partly due to the fact that we concentrated our efforts to recapture ringed bats within Portugal, which is roughly 500 km long by 200 km wide. Had we made a similar search effort throughout Spain, we would probably have found longer movements. In fact, most of the long-distance movements were identified by our Spanish colleagues.
Why do they migrate?
Although widespread among bats, migration is an adaptive behaviour that carries inevitable costs, including those associated with travel: increased energy expenditure and exposure to predation during migration (Fleming & Eby, 2003). In the case of regional migrations, and depending on the distance travelled, these costs may be comparatively small, but there are unavoidable costs associated with adaptation to a new area, such as the time spent optimizing roosting and foraging efficiencies, and an unknown predation scenario. Consequently, bats are only likely to migrate if the advantages of this behaviour more than compensate for its costs, and it is an important ecological issue to understand what these advantages are.
Although almost all the populations we studied migrated extensively, one consisted mostly of resident individuals. Similar exceptions to migratory behaviour were found in M. natalensis (Brown, 1999). This demonstrates that while endogenous mechanisms may be involved in the decision to migrate, migration is triggered by environmental conditions. In addition, we conclude that in the case of most of our study populations, environmental conditions drove the bats to migrate.
We tested two potential environmental drivers for migration: temperature in the roosts and at the foraging areas. Roost temperature was selected because it is generally considered to be the most important climatic parameter of the roost (Tuttle, 1976; Brown & Bernard, 1994), and is the most commonly suggested explanation for the migration of cave-dwelling bats (Van der Merwe, 1973). Temperature at the foraging area was included because it is likely to influence the availability of insects, which has also been proposed as a driver for bat migrations (Richter & Cumming, 2006).
We had predicted that if roost temperatures are the reason why bats migrate, then temperatures at the new roosts should be different from those of the old roosts. Our results validated this prediction, as most migrations resulted in significant changes in the temperature of the roosts used by bats. In addition, these changes were in the direction that we would expect to be metabolically most advantageous for a cave-dwelling bat in a temperate zone. This was particularly obvious in the case of females. The winter-to-spring migration placed them in warmer roosts, which is advantageous because most are pregnant and need high temperatures to allow the embryos to develop. The migration to their maternity quarters brought them to even warmer roosts, and high roost temperatures are known to foster the fast growth of the newly born bats. As nursing was completed, bats migrated to cooler caves, which allowed them to enter torpor during daytime, thus saving metabolic energy. Finally, the migration to the wintering roost brought bats to the coolest caves, and lower temperatures increase the energetic efficiency of hibernation.
These results confirm that roost temperature is a key driver in the migration of M. schreibersii. This conclusion is also corroborated by the fact that the population in which most individuals did not migrate was based at a large cave, which had a variety of microclimates available, suitable for the metabolic needs of various phases of the bats' annual cycle.
The confirmation of the importance of roost temperature does not necessarily imply that the night air temperature in the foraging areas, and consequently the foraging conditions, is also not a relevant driver for migration. However, we did not find any evidence in favour of this hypothesis. Because higher night ambient temperatures favour foraging activity, this hypothesis would be validated had bats migrated to warmer regions during the seasons when foraging is more important, namely spring, summer and autumn. This would have been possible, because within the range of these migrations (using 200 km as a reference) there were variations in low night temperatures of at least 6 °C. However, bats did not take advantage of this possibility as the night temperatures at the newly occupied regions were, on the average, similar to those from which the bats departed. In summary, bats made seasonal migrations to reach the most thermally suited roosts, rather than regions with a more suitable foraging climate.
Our results corroborate the validity of the hypothesis that roost temperature and associated metabolic advantages are key drivers for regional migrations of a cave-dwelling bat. This conclusion should hold for the many other temperate zone bats that have similar yearly life cycles and metabolic requirements. In contrast, our conclusion that night temperatures are not important may be less general, as we worked in a Mediterranean region, where the climate is mild. In colder regions, where low night temperatures are more likely to be limiting to the activity of insect prey, this factor is potentially more important.
We are indebted to our colleagues who generously provided precious help in hundreds of field trips. It is not possible to list all of them, and so we just name those who participated most often: Ana Rainho, Hugo Rebelo, Sofia Lourenço, Maria João Pereira, Tiago Marques, Gabriel Mendes, Francisco Rasteiro, Nuno Vieira and Henrique Vicêncio. Ana, João, Henrique and Inês Moreira, also helped with the data analyses or preparation of the figures. Tony Mitchell-Jones made very useful comments on an earlier version of the paper.