|Highly structured ﬁssion-fusion societies in an aerial-hawking, carnivorous bat
ANA G . P OPA -LI S S EANU * , F A B I O B O N T A D I N A †, O L G A M O R A * & C A R L O S I B A‘ N˜ EZ *
*Estacio´ n Biolo´ gica de Don˜ ana (CSIC), Sevilla
yDivision of Conservation Biology, Zoological Institute, University of Bern
(Received 20 December 2006; initial acceptance 21 January 2007;
ﬁnal acceptance 28 May 2007; published online 29 October 2007; MS. number: 9222R)
In some group-living animals, societies are far from being static but are instead dynamic entities encom- passing multiple scales of organization. We found that maternity colonies of giant noctule bats, Nyctalus lasiopterus, form ﬁssionefusion societies, where group composition in single tree roosts changes on a daily basis but social cohesion in the larger group is preserved. The population inside a small city park was com- prised of three distinct but cryptic social groups coexisting in close proximity. Each social group used a dis- tinct roosting area, but some overlap existed in the boundaries between them. Social groups were stable at least in the mid term because adult females were loyal to roosting areas and young females returned to their natal social groups in successive years. Our results suggest that distinct social groups with separate roosting areas may have existed for at least 14 years. The ﬁndings described support the hypothesis that roost-switching behaviour in forest bats permits the maintenance of social bonds between colony mem- bers and enhances knowledge about a colony’s roosting resources. Fissionefusion societies in forest bats might have evolved as a mechanism to cope with changing conditions in the environment by restructur- ing subgroups or adjusting subgroup size, to maximize the amount of information that can be transferred between colony members, or as a consequence of territory inheritance by philopatric female offspring. Other factors such as resource competition or kin selection could limit the size and composition of ﬁssionefusion societies and promote strong social structuring within populations.
Keywords: ﬁssionefusion; giant noctule bat; Nyctalus lasiopterus; radiotracking; roosting behaviour; roost-switching;
The tendency of conspeciﬁcs to aggregate is widespread in both plants and animals. Essential resources might be patchily distributed in space and time, forcing individuals to come together. Other selective pressures can favour group living, including predator avoidance (Hamilton
1971), increased foraging efﬁciency (Beauchamp 1999) and cooperative breeding (Emlen 1984). Conversely, liv- ing in groups can impose ﬁtness costs, leading to direct competition for resources between group members (West-Eberhard 1979), facilitating the spread of parasites and diseases (Davies et al. 1991; Van Vuren 1996),
Correspondence and present address: A. G. Popa-Lisseanu, Estacio´ n Biolo´ gica de Don˜ ana (CSIC), Avda. Mar´ıa Luisa s/n, Pabello´ n del Peru´ , 41013 Sevilla, Spain (email: firstname.lastname@example.org). F. Bontadina is at the Zoological Institute, Division of Conservation Biology, Univer- sity of Bern, Baltzerstrasse 6, CH-3012 Bern, Switzerland.
requiring common decisions necessary for group coordi- nation which can generate conﬂict of interest (Conradt & Roper 2000, 2005), and occasionally producing altruistic behaviours, which beneﬁt apparently only the recipient but not the donor (West et al. 2006).
Social animals must continuously balance the trade-off between the costs and beneﬁts of group living (Alexander
1974). Sociality can thus be a dynamic process (Couzin
2006) in which groups might assemble or split in response to a variety of intrinsic factors (such as age or reproductive status) and extrinsic factors (such as food availability or landscape complexity). Some examples of animals having this ﬂexible, ‘ﬁssionefusion’ social behaviour are lions, Panthera leo (Packer et al. 1990), primates such as chim- panzees, Pan troglodytes, or spider monkeys, Ateles sp. (Symington 1990), dolphins, Tursiops sp. (Lusseau et al.
2006), elephants, Loxodonta africana (Wittemyer et al.
2005; Archie et al. 2006), red deer, Cervus elaphus
(Albon et al. 1992), spotted hyenas, Crocuta crocuta (Hole- kamp et al. 1997) and orange-fronted parakeets, Aratinga canicularis (Cortopassi & Bradbury 2006). The pattern of temporal associations within these ﬁssionefusion socie- ties is not random, but appears to be tied to individual preferences, in some cases mediated by kinship, resulting in complex social structures (Wittemyer et al. 2005; Archie et al. 2006; Lusseau et al. 2006). Although it is accepted that ﬁssionefusion behaviours allow animals to adapt to changing conditions in their environment by adjusting group size, the ultimate forces shaping the evolution of this type of social organization are still poorly understood (Chapman et al. 1995; Lehmann & Boesch 2004).
Temperate bats offer a good model system to assess the adaptability of group living at several scales, because of their complex life history related to the seasonality of food resources, roost requirements and energetic constraints imposed by ﬂight. Most temperate bat species have
‘seasonally variant’ social interactions (Bradbury 1977), with sexually segregated units during the breeding season (females forming relatively large maternity colonies and males roosting solitarily or in small groups) and many dif- ferent grouping patterns during mating and hibernation, including solitary and colonial roosting. It has recently been suggested that maternity colonies of tree-dwelling bats form ﬁssionefusion societies (Kerth & Ko¨ nig 1999; O’Donnell 2000; Willis & Brigham 2004), where group members are spread among multiple roosts on a given day with the composition of subgroups varying from day to day. Single bats, or sometimes whole groups, switch roosts regularly (Lewis 1996; Kerth & Ko¨nig 1999; O’Donnell & Sedgeley 1999; Willis & Brigham 2004; Russo et al.
2005). Several hypotheses have been proposed to explain roost switching including avoidance of predators, antipar- asite strategy (i.e. roosts are left vacant to interrupt ecto- parasite life cycles), minimization of distance to foraging areas, ephemerality of roost trees, and speciﬁc thermoreg- ulatory requirements in relation to variable microclimatic conditions (reviewed in Lewis 1995; Lewis 1996; Kerth & Ko¨ nig 1999; Kunz & Lumsden 2003). In the latter case, we expect that individuals with different thermoregula- tory requirements, for example lactating versus pregnant females, might differ in their roost-switching patterns (Willis & Brigham 2004). Recent studies best support two alternative hypotheses: (1) roost switching could be a way of maintaining social bonds between bats belonging to a colony which is spread over large areas of forest (O’Donnell 2000; Willis & Brigham 2004; O’Donnell & Sedgeley 2006); or (2) roost switching could serve to en- hance and share knowledge about a large pool of roosts (Kerth & Reckardt 2003; Russo et al. 2005; O’Donnell & Sedgeley 2006). Even if forest bats change roosts often, they nevertheless appear to be loyal to roosting areas (Brigham et al. 1997; O’Donnell & Sedgeley 1999; Cryan et al. 2001) and even to speciﬁc trees, over the mid, and possibly the long term (Willis et al. 2003).
The giant noctule, Nyctalus lasiopterus, is the largest and one of the rarest European vespertilionid bats (body mass ¼ 50 g; forearm ¼ 65 mm; wing span ¼ 450 mm). It has a Circum-Mediterranean distribution (Ib‘an˜ ez et al.
2004), possibly related to its dietary specialization: it is
the sole predator known to catch nocturnally migrating songbirds which concentrate in Mediterranean regions in spring and autumn (Bruderer & Liechti 1999), while itself on the wing (Ib‘an˜ ez et al. 2001, 2003; Popa-Lisseanu et al. 2007). During summer, it hunts insects in the open like other aerial-hawking bats. Individuals roost sexually segregated, in trees (Ib‘an˜ ez et al. 2004): most adult males appear to be solitary throughout the year whereas adult fe- males and their young aggregate in breeding colonies dur- ing spring and summer, joining males in the mating season in autumn. In some localities, only one of the two, either female and young breeding colonies or all- year male populations (females arriving only in autumn from unknown areas) have been found, suggesting that sexes might also show local and/or altitudinal segregation apart from roost segregation, with breeding colonies located in the lower or warmer areas (C. Ib‘an˜ ez, A. Guille´n, P. Agirre-Mendi, J. Juste & A. Popa-Lisseanu, unpublished data; cf. Barclay 1991). No data on hibernation exist.
We studied social structure and roost use by individuals in a giant noctule breeding population, located in a small urban park in southwestern Spain. The south of the Iberian Peninsula, which is a main conﬂuence of bird migratory routes, is the most intensely deforested region in the Mediterranean basin (Arribas et al. 2003). Few nat- ural roosts are available for forest-dwelling bats, and some historic urban parks constitute ‘roosting islands’ for giant noctules in an otherwise treeless agricultural or urbanized landscape. We report patterns of roost use by giant noctule bats from an urban park across several years, with the following aims: (1) assess whether maternity colonies of giant noctule conform to the ﬁssionefusion society model, as has been proposed for smaller tree-dwelling bat species. (2) Deﬁne population structure and the limits of ‘colony’ or ‘social group’, considered ambiguous con- cepts for forest bats (e.g. Lewis 1996). More speciﬁcally, we question whether each tree contained one social group, whether all bats in the park belonged to a single social group scattered in many different tree roosts, or whether a few social groups, with members scattered in several tree roosts, coexisted within the park. (3) Test whether single trees are used over multiple years and if bats are loyal to roosting areas over time. (4) Test whether roost-switching patterns, in particular frequency of roost switching, differ between individuals or between different reproductive periods. (5) Evaluate the hypotheses pro- posed to explain roost-switching behaviour in forest bats.
The study was conducted in Mar´ıa Luisa Park, situated in Seville, Andalusia, Spain (37o 240 N, 5o 590 W, altitude
10 m asl). This 23-ha park was established in 1850 and has a dense subtropical vegetation, mostly exotic species in- cluding large specimens of Platanus sp., Gleditsia triacanthos and Sophora japonica, and tall palm trees, for example Washingtonia ﬁlifera. A breeding population of c. 500 giant noctules use the cavities and hollows of these mature trees
and roost under the dry leaves of Washingtonia (Ib‘an˜ ez et al. 2004).
The area has a typical Mediterranean climate, with hot dry summers and precipitation occurring mostly in autumn and winter. Average annual rainfall is w550 mm, mean an- nual temperature 18.6o C and there are almost 3000 hours of sunshine per year. The land surrounding the city of Seville is mostly devoted to agriculture, with a few fragmented natu- ral vegetation (mostly shrubs) patches.
The giant noctule population is comprised primarily of females and their young who are born in late Mayeearly June and start ﬂying in July. Adult females lactate until early August (Ib‘an˜ ez et al. 2004). Most bats abandon the roosting area from August to November, and females begin aggregating at roosts in March.
Capture, Monitoring and Radiotracking
Bats were netted when emerging from or returning to several trees with accessible roosting cavities in 1999e
2006. Netting was conducted regularly (but never more frequently than once per month to minimize distur- bance). Exit counts were also performed at some of these roosts. Tree cavities were occasionally inspected with a small infrared video camera after adults had emerged, to determine the timing of parturition.
Bats were individually marked with 5.2-mm aluminium alloy rings (Porzana, Ltd, Icklesham, U.K.), and beginning in 2003, also with subcutaneously implanted transpon- ders (ID100, Trovan, EID Ibe´ rica, Spain). Transponders (2.2 x 11.5 mm) were inserted between the shoulder blades using a Trovan robust applicator with a 12-gauge needle (EID Ibe´ rica, Spain). Transponders have been suc- cessfully used to mark smaller bats, for example Myotis bechsteinii (8e14 g) and Eptesicus fuscus (15e20 g), with no apparent adverse effects (Kerth & Ko¨ nig 1996, 1999; Kerth & Reckardt 2003; Wimsatt et al. 2005). Bats were classiﬁed based on age, sex and reproductive status. Lactat- ing females had enlarged nipples surrounded by hairless skin. Juveniles had cartilaginous plates in the metacarpale phalangeal joints (Anthony 1988). During 2003, we caught
11 bats at several tree roosts and equipped them with col- lared radiotransmitters (Pip Ag392, Biotrack, Dorset, U.K.). Owing to transmitter failure, no data were collected for one individual (Table 1). In April 2004, we attached radio- transmitters to 15 different individuals captured at three dif- ferent sites inside the park (Table 1). All tagged bats were adult females with average or above body mass. To afﬁx transmitters, the ends of a Teﬂon collar were glued together around the neck of the animal. The collar was also attached to the back of the neck with surgical cement (Skin-Bond, Smith and Nephew United, Largo, FL, U.S.A.), after clipping the fur to prevent the transmitter from rotating around the
Table 1. Roost-switching behaviour of radio tagged giant noctules
No. of days tracked
No. of trees
used FR prelact. FR lact. FR total Social group
92.1 July 1992 13 3 d d 4.33 I
92.2 July 1992 16 7 d d 1.86 I
1 AprileJune 2003 47 16 d d d III
2 AprileJune 2003 46 11 d d d III
3 AprileMay 2003 17 6 d d d III
4 AprileMay 2003 13 7 d d d III
5 July 2003 16 6 d d d I
6 July 2003 15 5 d d d I
7 July 2003 9 3 d d d I
8 OcteNov 2003 16 3 d d d II
9 OcteNov 2003 14 4 d d d I
10 OcteNov 2003 11 2 d d d II
11 AprileJuly 2004 68 14 1.54 3.75 2.09 II
12 AprileJuly 2004 68 8 3 6 3.88 II
13 AprileMay 2004 12 3 d d d I
14 AprileMay 2004 26 7 2.08 d 2.08 I
15 AprileJune 2004 70 8 4.45 4.67 4.53 I
16 AprileJuly 2004 72 19 1.86 3.11 2.23 I
17 AprileJune 2004 41 11 2.35 d 2.35 I
18 AprileMay 2004 29 4 2.25 d 2.25 II
19 AprileMay 2004 27 3 2.33 d 2.33 II
20 AprileMay 2004 23 8 1.83 d 1.83 II
21 AprileJune 2004 66 8 2.47 2.33 2.41 III
22 AprileMay 2004 34 8 2.75 d 2.75 III
23 AprileJune 2004 55 15 2.85 1.78 2.41 III
24 AprileJune 2004 60 7 3.00 7.00 3.69 III
25 AprileMay 2004 25 4 d d d III
27 (no. of bats) 909 (total days)
73þ4* (no. of trees)
Frequency of roost switching (FR) ¼ days/roost before moving to another roost for the prelactation (prelact.) and lactation (lact.) periods. Social group ¼ the cluster-deﬁned group that each bat was assigned to, according to site of capture and cluster analysis reported in the text. Data on FR for bats tracked in 1992 were not considered for the calculation of mean ± SD total FR.
*Trees used only in 1992.
neck. The thin teﬂon collar was designed to fall off after 1e6 months. The total mass of transmitter, collar and glue was w1.85 g, representing less than 5% of body mass (Aldridge
& Brigham 1988). Speciﬁcally, radiotransmitter mass repre- sented 3.0% and 4.4% of the mass of the largest (61.8 g)
Table 2. Number of giant noctules detected by transponder readers
1 and 2 and assigned to their individual social groups, January
No. of bats belonging to each social group
and smallest (42 g) tagged bats, respectively. Capture and marking of bats were approved by the Environmental Coun- cil of the Junta de Andaluc´ıa.
Social group I
Social group II
Social group III
In 2003, we located radiotagged bats using handheld telemetry receivers (FT-250 RII, Yaesu Musen Co., Ltd, Sapporo, Japan; Falcon V, Wildlife Materials International, Inc., IL, U.S.A.) and three-element Yagi antennae (AF Antronics, Inc., IL, U.S.A.), approximately four times a week. In 2004, the bats were tracked to tree roosts every day, until the signal was lost. Transmitter life was approx- imately 2 months.
No. of bats marked
No. of bats registered at reader 1
No. of bats registered at reader 2
81 61 114
61 (2904) 1 (1) 0 (0)
0 (0) 48 (1919) 6 (8)
Emergence by marked bats was continuously monitored using two automatic transponder readers (LID 650, Tro- van, EID Ibe´ rica, Spain) installed at two roost trees located
265 m apart, from January 2004 to August 2006. While entering or leaving the tree cavity, the bats ﬂew through a circular antenna around the entrance so that their indi-
Numbers in parenthesis indicate bats visits*days (the sum of the
number of days that bats of each social group were registered at
(Matusita 1955; Krebs 1989) as follows:
vidual code, the date and time were recorded.
¼ pir r¼1
Effect of Transponder Marking and
Radiotagging on Bats
Transponder injection caused no bleeding and the small hole created by the needle healed within a few days. None of the recaptured transponder-marked bats (N ¼ 58) showed evidence of scabs or scarring, and transponders re- mained positioned between the shoulder blades. We did not detect any adverse effect of radiotagging on the bats. We recaptured and removed tags from only three individ- uals. However, two transmitters with detached collars were found on the ground near roosts (1 and 2 months after tagging, respectively). Another two bats, no longer carrying transmitters, were recaptured 1 and 2 years, re- spectively, after tagging. Their body mass had not declined and they showed no outward sign of having carried a transmitter. In addition, 10 out of 13 radiotagged bats (76.9%) belonging to the two social groups monitored with transponder readers (excluding bats for which radio- transmitters were recovered) were detected in successive years, a similar proportion to transponder-marked bats of these two social groups detected at the readers (76.8%; Table 2), suggesting that radiotagging did not affect survival.
The existence of an organized versus random structure in the giant noctule population of the Mar´ıa Luisa Park was assessed using two approaches: (1) the similarity in the use of different tree roosts by individual bats; (2) the degree of association between pairs of bats.
In the ﬁrst approach, a cluster analysis was conducted based on the similarity in the use of roost trees. The similarity, or overlap, was calculated using the FreemaneTukey statistic
where FTij is the overlap, or similarity, in the use of available roost trees by individuals i and j, and pir is the proportion of days, from the total number of radiotracking days, that bat i was found in tree r (idem for bat j ). This allowed us to iden- tify groups of bats showing similar roost use. A dissimilarity index was then entered for the calculations, deﬁned as 1- overlap.
We created hierarchical groupings using four different clustering methods (UPGMA or unweighted pair-group average, Ward’s weighted method, SLINK or single linkage method, and CLINK or complete linkage method), as agreement between the outcome of different clustering algorithms is usually a sign of a pronounced structure in the data.
In the second approach, we relied again on hierarchical clustering (UPGMA, Ward’s method, SLINK and CLINK), now based on a matrix of associations between all possible pairs of bats. The association index for a given pair was calculated by dividing the number of days that two bats roosted together in a particular roost by the number of days both bats were radiotracked. In this analysis we only used data for the 15 bats tracked in 2004, as they were monitored simultaneously but had been captured at three different sites within the park (ﬁve bats at each site). To assess the integrity of the groups, we compared the outcome of the four clustering methods.
The trees used by bats were assigned as belonging to one (or, rarely, to several) of the three groups identiﬁed by cluster analysis, according to which group used them. For all recaptured bats, we noted whether the tree where each individual was recaptured belonged to the same group as the tree where it had been captured previously. Given that bats switch frequently from one tree to another, in the absence of a population substructure all bats would have the same probability of recapture in all three subsamples of trees.
Likewise, if there was no population substructuring and bats switched randomly between roosts throughout the park, the two automatic reading devices installed at two different trees should with equal probability detect bats captured at any site within the park. We assigned each bat to the group where it was ﬁrst captured. We evaluated using a c2 test whether bats belonging to each of the three
groups were detected with equal probability, that is if they roosted at random, in the two trees equipped with tran- sponder readers. The number of individuals from each group that were detected by each device was compared with the expected frequency calculated from the number of bats marked with transponders in each group assuming that they all had the same probability to be detected at each reader. Additionally, we recorded the number of days that bats from each group were detected at each reader, to distinguish frequent from rare events (i.e. if a bat visited a ‘foreign’ group regularly or only occasionally).