The evolution of sensory divergence in the context of limited gene flow in the bumblebee bat


previous item | next item

Citation
The evolution of sensory divergence in the context of limited gene flow in the bumblebee bat

Material Information

Title:
The evolution of sensory divergence in the context of limited gene flow in the bumblebee bat
Series Title:
Nature Communications
Creator:
Puechmaille, Sebastien J.
Ar Gouilh, Meriadeg
Piyapan
Piyathip
Yokubol, Medhi et al
Publisher:
Nature
Publication Date:
Language:
English

Subjects

Subjects / Keywords:
Evolutionary Biology ( local )
Speciation ( local )
Zoology ( local )
Genre:
serial ( sobekcm )

Notes

Abstract:
The sensory drive theory of speciation predicts that populations of the same species inhabiting different environments can differ in sensory traits, and that this sensory difference can ultimately drive speciation. However, even in the best-known examples of sensory ecology driven speciation, it is uncertain whether the variation in sensory traits is the cause or the consequence of a reduction in levels of gene flow. Here we show strong genetic differentiation, no gene flow and large echolocation differences between the allopatric Myanmar and Thai populations of the world's smallest mammal, Craseonycteris thonglongyai, and suggest that geographic isolation most likely preceded sensory divergence. Within the geographically continuous Thai population, we show that geographic distance has a primary role in limiting gene flow rather than echolocation divergence. In line with sensory-driven speciation models, we suggest that in C. thonglongyai, limited gene flow creates the suitable conditions that favour the evolution of sensory divergence via local adaptation.
Original Version:
Nature Communications, Vol. 2, no. 573 (2011-12-06).

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
This item is licensed with the Creative Commons Attribution Non-Commercial No Derivative License. This license allows others to download this work and share them with others as long as they mention the author and link back to the author, but they can’t change them in any way or use them commercially.
Resource Identifier:
K26-05197 ( USFLDC: LOCAL DOI )
k26.5197 ( USFLDC: LOCAL Handle )

USFLDC Membership

Aggregations:
University of South Florida
Karst Information Portal

Postcard Information

Format:
serial

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

ARTICLE1NATURE COMMUNICATIONS | 2:573 | DOI: 10.1038/ncomms1582 | www.nature.com/naturecommunications© 2011 Macmillan Publishers Limited. All rights reserved. Received 1 Jul 2011 | Accepted 2 Nov 2011 | Published 6 Dec 2011 DOI: 10.1038/ncomms1582 The sensory drive theory of speciation predicts that populations of the same species inhabiting different environments can differ in sensory traits, and that this sensory difference can ultimately drive speciation. However, even in the best-known examples of sensory ecology driven speciation, it is uncertain whether the variation in sensory traits is the cause or the consequence of a reduction in levels of gene ” ow. Here we show strong genetic differentiation, no gene ” ow and large echolocation differences between the allopatric Myanmar and Thai populations of the world  s smallest mammal, Craseonycteris thonglongyai , and suggest that geographic isolation most likely preceded sensory divergence. Within the geographically continuous Thai population, we show that geographic distance has a primary role in limiting gene ” ow rather than echolocation divergence. In line with sensory-driven speciation models, we suggest that in C. thonglongyai , limited gene ” ow creates the suitable conditions that favour the evolution of sensory divergence via local adaptation. 1 School of Biology and Environmental Science & UCD Co nway Institute of Biomolecular and Biomedical Research, University Co llege Dublin , Bel“ eld , Dublin 4, Ireland . 2 Center of Excellence for Vector and Vector-Borne Diseases, Faculty of Science, Mahidol University at Salaya , Nakhon Patho m 73170 , Thailand . 3 Laboratory for Urgent Response to Biological Threats (CIBU), Institut Pasteur , 75724 Paris , France . 4 Faculty of Science, Prince of Songkla University , Hat-Yai 90112 , Thailand . 5 Zoology Department, Yangon University , Yangon , Myanmar . 6 Harrison Institute, Centre for Systematics and Biodiversity Research, Sevenoaks , Kent TN13 3AQ , UK . 7 Hinthada University , Hinthada , Myanmar . 8 Aberdeen Centre for Environmental Sustainability, Institute of Biological and Environmental Sciences, University of Aberdeen , Aberdeen AB24 2TZ , UK . 9 University Rennes 1 / CNRS, UMR 6553 ECOBIO, Station Biologique , Paimpont 35380 , France .  Present address: Max Planck Institute for Ornithology, Sensory Ecology Group, 82319 Seewiesen, Germany . ‚ Deceased. Correspondence and requests for materials should be addressed to S.J.P. (email: s.puechmaille@gmail.com ) or to E.C.T. (email: emma.teeling@ ucd.ie ). The evolution of sensory divergence in the context of limited gene ” ow in the bumblebee bat S é bastien J. Puechmaille 1 ,  , Meriadeg Ar Gouilh 2 , 3 , Piyathip Piyapan 4 , Medhi Yokubol 4 , Khin Mie Mie 5 , Paul J. Bates 6 , Chutamas Satasook 4 , Tin Nwe 5 , ‚ , Si Si Hla Bu 7 , Iain J. Mackie 8 , Eric J. Petit 9 & Emma C. Teeling 1

PAGE 2

ARTICLE 2 NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1582 NATURE COMMUNICATIONS | 2:573 | DOI: 10.1038/ncomms1582 | www.nature.com/naturecommunications© 2011 Macmillan Publishers Limited. All rights reserved. Theoretical models 1,2 and a few empirical studies 3,4 have suggested the role of divergent ecology and sensory perception in speciation. Despite classical studies investigating the role of sensory ecology in speciation, 3,5 it has not yet been possible to conrm whether sensory divergence is initially responsible for limiting gene ow or results as a consequence of limited gene ow 6 . Typically, most classical research on speciation follows a retrospective  spy-glass  approach, whereby the causes and mechanisms of speciation are inferred retrospectively a er the speciation process is nished 7 . Although this approach has undeniably improved our understand-ing of speciation, it o en lacks resolution at the early stages of the process, either because a er speciation is completed, the signal has been confounded by post-speciation changes or simply because of the lack of historical ecological and / or behavioural data when the process was initiated 8 . Templeton 9 suggested that  e closer to the speciation process, the greater the ability to focus upon the genet-ics of speciation  , leading to the view that research is needed at the population / species interface to uncover how speciation occurs 7,8,10 . erefore, studying the population / species interface, using a  mag-nifying glass  approach, can reveal important aspects of speciation and how ecology, behaviour and genetics interact in di erent situa-tions to cause the evolution of barriers to gene ow 7,11 , particularly important in the case of sensory ecology driven speciation. e majority of bats use sound (echolocation) to orient them-selves in their environment, locate their food and communicate 12 . A bat  s echolocation signal and its ability to perceive this signal are functionally linked, as the bat must be physically able to hear the echoes of its outgoing calls 13 . erefore, studying the signal or echo-location call in bats will inform our understanding of their sensory system and perception capabilities 11,13 . Currently the genetic mecha-nisms that control echolocation are being uncovered and studies are starting to explore the role that echolocation has in communication and sexual selection 14 . Recent studies have suggested that changes in echolocation frequency (for example, Rhinolophus ) are associated with assortative mating, reproductive isolation and ultimately spe-ciation, regardless of external barriers to gene ow 5 . erefore, bat species are an excellent mammalian model to study the role of sen-sory drive (that is, echolocation calls) in the speciation process 11,15 . One such species is Craseonycteris thonglongyai , a rare and endangered, charismatic bat species, considered as the World  s smallest mammal and restricted to a 2,000-km 2 region that strad-dles the ai … Myanmar border 16 . Currently, the ai and Myan-mar populations are disjointed, although the long-term presence of suitable intervening karst landscape suggests that this may once have been a continuous distribution ( Fig. 1 ). Despite being morphologically indistinguishable 17 , observed variation in echolo-cation calls between ailand 18 and Myanmar 19 suggested that more than one species might be present. erefore, studying both the allopatric and sympatric populations o ers a unique opportunity to study the factors that in uence population structure, gene ow and sensory trait  s divergence across an entire species range and at di erent evolutionary timeframes. Here we used both a  spyglass  and  magnifying glass  approach to untangle the early drivers of speciation in allopatric and sympatric populations of C. thonglongyai , the sole representative of the mono-typic bat family Craseonycteridae. To illuminate the early stages of the speciation process, we examined the ecological and genetic fac-tors that could have limited gene ow and driven local adaptation in these unique populations. We speci cally focus on the role that echolocation has in this process. We show that currently no gene ow is occurring between the geographically isolated allopatric populations in ailand and Myanmar, and infer that geographic isolation most likely preceded the echolocation divergence. Within the sympatric populations in ailand, we identi ed geographic distance rather than echolocation divergence as the major factor shaping population structure across the entire range. We suggest that echolocation divergence within ailand, most likely resulting from interspeci c competition, evolved in a context of limited gene ow, which itself resulted from the extreme localized dispersal of this species. Results e spyglass approach . To reconstruct the spatio-temporal history of these populations and determine the role that echolocation call variation may have in speciation, we generated and analysed a multi-genic data set (Cytochrome b and partial D-loop, Xand Y-introns and microsatellites; GenBank accession no. GU247601 … 247751, Supplementary Tables S1 and S2 ) and collected acoustic data from throughout the species  range ( Fig. 1 ). Intensive trapping, sampling and acoustic recording were carried out in ailand and Myanmar ( Fig. 1 ). ese data con rmed the marked and signi cant di erence in echolocation calls between ailand and Myanmar ( ailand: N = 4,188, 70.125 … 79.875 kHz and Myanmar: N = 1,472, 76.875 … 83.625 kHz, generalized linear mixed model, P = 0.002; Fig. 2a ; Supplementary Fig. S1 ). Bayesian phylogenetic analyses based on a 1.8-kb mitochondrial fragment of 602 samples and two di erent NUMTs (mitochondrial pseudogenes) showed that the ai and Myanmar populations were reciprocally monophyletic with a posterior probability of 1.00 ( Fig. 2b ). is clear ai … Myanmar split was con rmed at the nuclear level by screening all individuals ( n = 659) for 15 microsatellites ( F st = 0.49, P < 0.001; Fig. 2c … d ) and ve nuclear single-nucleotide polymorphisms (SNPs), which showed almost complete lineage sorting ( Supplementary Fig. S2 ). e overall strong di erentiation of all markers investigated combined with the complete lineage sorting of the mitochondrial fragment ( Fig. 2b ), two nuclear SNPs ( Supplementary Fig. S2 ) and the absence of shared alleles at one microsatellite locus, strongly advocate for a total absence of gene ow between the allopatric ai and Myanmar populations. Relying on the molecular clock, the split between the two populations is dated to approximately 0.4 Mya M14M8M13M1[e]M3THAILANDMYANMARN050KilometresP[g]KN,OL, MJIH[g]G[g]FEDC[e]BA Figure 1 | Map of 21 colonies where C. thonglongyai were sampled and recorded. Grey shaded areas represent limestone areas and blue shaded areas represent inland water bodies (dams). C. thonglongyai distribution in Thailand (green) and Myanmar (dark blue) were taken from Puechmaille et al. 16 A bat is represented next to each site where M. siligorensis was recorded. Myanmar colonies are labelled as M1, M3, M8, M13 and M14. Thailand colonies are labelled A … P. [e] indicates a colony where only echolocation data were recorded, while [g] designates a colony with only genetic data.

PAGE 3

ARTICLE 3 NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1582 NATURE COMMUNICATIONS | 2:573 | DOI: 10.1038/ncomms1582 | www.nature.com/naturecommunications© 2011 Macmillan Publishers Limited. All rights reserved. (95 % highest probability density, 0.268 … 0.545 Mya). Genetic diversity indices of the Myanmar population are signi cantly reduced at the mitochondrial ( ailand, Hs = 0.92; Myanmar, Hs = 0.36; permutation test, P = 0.006) and nuclear ( ailand, Hs = 0.56; Myanmar, Hs = 0.43; permutation test, P = 0.014) levels compared with the ai population. Similarly, nucleotide diversity of the mitochondrial DNA fragment calculated per site in ailand and Myanmar di er strongly with more than 10-fold di erences ( ailand: average = 0.0034, min … max: 0.0028 … 0.0048; Myanmar: average = 0.00027, min … max: 0.00013 … 0.00050), indicating that the Myanmar population probably originated from a small number of individuals, either as a consequence of a strong bottleneck or more likely via a founder e ect. However, to fully elucidate whether echolocation variation drove the isolation of the two populations or evolved a er their split requires knowledge about distribution of species and the echolocation frequency of the population at the time of divergence. As this is not available, we turned to the sympatric ai population to address this question. e magnifying glass approach . As the rst steps of speciation occur at the population / species interface 7 … 9,20 , we investigated the recent spatio-temporal history of the ai population and the relationship between genetic structure and echolocation frequency at the population level. To investigate past changes in distribution, we assessed whether the ai population has been stable, increased in size (demographic expansion) or spat ially expanded (spatial expansion) 21 using mismatch distribution analyses of mitochondrial fragments ( n = 462). A model of stable population size was rejected for all ai colonies (non-overlapping 99 % con dence intervals of 0 and 1 , two mutation parameters that are related to the e ective size of the population before and a er a putative expansion, respectively). While the Southern ai colonies (H, I, J and L, M, N, O, Fig. 1 ) tted both models of demographic and spatial expansion equally well, the Central (E, F and G, Fig. 1 ) and Northern colonies (A, B and D, Fig. 1 ) tted a model of spatial expansion but not a model of pure demographic expansion ( Supplementary Fig. S3 ). Principal component analysis of microsatellite data showed a strong population structure orientated along the Northwest … Southeast direction ( Fig. 3a … d ). is structure was con rmed by a Mantel test showing a very strong correlation between geographic and genetic distances (Mantel test; r = 0.95, R 2 = 0.90, P < 0.001; Fig. 3e ). is strong correlation could theoretically result from di erent evolutionary histories such as secondary contact between two previously isolated populations or spatial expansion from a single population. Although the secondary contact scenario cannot be excluded, mismatch distributions do not support the presence of two isolated populations, but rather suggest that a single population was most likely present in the Southern part of their range and recently expanded northwards. Analyses that simulate the ai population spatially expanding under a wide range of migration rate and population growth rate show that the strong isolation by distance pattern observed ( Fig. 3e ) can result from a re cent spatial expansion when gene ow is restricted to neighbouring populations ( Supplementary Table S3 and Fig. S4 ). e average dispersal distance estimated from the slope of the regression between genetic distance and geographic distance was 2.2 … 5.7 km, indeed demonstrating extreme localized dispersal. e geographically continuous ai population showed an abrupt change in echolocation frequency between colonies having similar echolocation calls (range: 73.35 … 73.96 kHz) for over 100 km in the Northern / Central part of their range and then a sudden increase ( ~ 3 kHz) in < 20 km in the Southern part ( Fig. 4 ). As body size and echolocation frequency are correlated across species 22 , we used forearm length as a proxy of body size and showed that body size di erences are unlikely to explain the observed echolocation di erence (Spearman  s rank correlation, = Š 0.17, P = 0.53). We then used causal modelling on resemblance matrices 23,24 to investigate the combination of factors driving genetic di erentiation between colonies. We rst evaluated the set of diagnostic statistical tests of the seven possible organizational models, which included geographic distance, the presence of barriers (areas without limestone), echolocation distance (measured by the absolute echolocation di erence) and genetic distance (measured by the F st index calculated on 14 microsatellites; Supplementary Information ). Only one organizational model, which included geographic distance, echolocation distance and genetic distance, was fully supported by all statistical expectations ( Supplementary Table S4 ). e six possible models of causal relationships between geographic distance, echolocation distance and genetic distance were then compared with the predictions of the models 23 . Five models were not supported by the data, as one or several of their predictions were not realized ( Supplementary Table S5 ), while one model was fully supported. is latter model supports the primary in uence of geographic distance on genetic distance, which itself in uences echolocation distance. Echolocation and dri . To investigate if this change in echolocation could result from echolocation dri , we assessed the shortand long-term population sizes in the North / Centre and South of Kanchanaburi. We used the actual average colony size 16 as a proxy for short-term population size, and mitochondrial and nuclear genetic diversity found within colonies as a proxy for long-term population size; high genetic diversities are being associated with large 0.06 M 14 M 8 M 13 M 3 A B D E F F G K P H I J L M N O O 0.2 1 1 1 1 Frequency (kHz)Number of pulses 0200400 84 83 82 81 80 79 78 77 76 75 74 73 72 71 70ab cd Figure 2 | Distinction between Thailand and Myanmar populations. ( a ) Distribution of echolocation calls recorded in Thailand (green; n = 4,188) and Myanmar (blue; n = 1,472). ( b ) Bayesian phylogenetic tree based on a 1.8-kb mitochondrial fragment of 602 samples and two different NUMTs (mitochondrial pseudogenes) showing the monophyly of the Thai (green) and Myanmar (blue) C. thonglongyai . The tree is rooted with two distinct NUMTs found in both populations. Bayesian posterior probabilities are shown below branches of interest. ( c ) Graphical representation of 196 Myanmar and 463 Thai individuals according to their probability to belong to each of the two clusters de“ ned by STRUCTURE (black lines separate individuals sampled at different colonies). ( d ) Neighbour-Joining tree showing the relationships between 196 Myanmar (in blue) and 463 Thai (in green) individuals based on 15 microsatellites (Da distance).

PAGE 4

ARTICLE 4 NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1582 NATURE COMMUNICATIONS | 2:573 | DOI: 10.1038/ncomms1582 | www.nature.com/naturecommunications© 2011 Macmillan Publishers Limited. All rights reserved. e ective population sizes and reduced dri , whereas low genetic diversities are being associated with reduced population sizes and increased dri 25 . e actual average colony size (South: 275; North / Centre: 248) did not signi cantly di er between Northern / Central and Southern colonies (Wilcoxon Mann … Whitney rank-sum test, n = 20, P = 0.19). A one-sided permutation test showed that the Northern / Central colonies had signi cantly reduced mitochondrial (North / Centre, Hs = 0.87; South, Hs = 0.94; P = 0.01) and nuclear (North / Centre, Hs = 0.52; South, Hs = 0.57; P = 0.04) gene diversities when compared with the Southern colonies. ese short-term and long-term population sizes suggest that the Northern / Central colonies, which most likely spatially expanded, were equally or more prone to dri than the Southern colonies, and therefore, if dri had a role in echolocation frequency change, the Northern / Central colonies should show similar or higher variation across colonies than their Southern counterparts. Contrary to this expectation, C. thonglongyai in the Northern / Central part of the range shows little echolocation variation over 100 km (colonies A to F, Fig. 4 ). erefore, this observation of a nonrandom pattern of echolocation call variation ( Fig. 4 ) supports the view that variation in echolocation within the ai population may have resulted from adaptation as opposed to dri . Echolocation and genomic selection . ere are marked di erences in echolocation call between the North / Centre and South of Kanchanaburi in ailand ( Fig. 4 ). If natural selection is driving the echolocation divergence within this population, then genes underlying echolocation frequency should show signs of divergent selection on either side of the area of abrupt echolocation change ( Fig. 4 ). is type of divergent selection has been shown in coat colour genes in darkand light-coloured populations of Peromyscus living on dark and light soil, respectively 26 . e 15 microsatellites from 462 Craseonycteris individuals were tested for selection based on the F st outlier approach 27 . One microsatellite locus (CTC1; arrow in Fig. 5a … c ) showed strong signs of divergent selection when comparing colonies on either side of the area of abrupt echolocation change aEigenvalues EigenvaluesEigenvalues EigenvaluesAxis 1–3 Axis 1–5 Axis 1–4 0.20 0.15 0.10 0.05 0.00 0204060 Geographic distance (km) 80100120140Genetic distance (F st )Axis 1–2b c e d Figure 3 | Population structure within Thailand. Principal component analysis (PCA) showing the population structure of individuals orientated along the Northwest … Southeast direction with the “ rst axis (explaining 8.46 % of the variance) versus the ( a ) second axis (5.56 % ), ( b ) third axis (4.97 % ), ( c ) fourth axis (4.34 % ) and ( d ) “ fth axis (4.24 % ). For each graph, the inset represents the eigenvalues of the two axes. Individuals from each of the 14 colo nies are represented in different colours, and the colonies range from A in the Northwest to O in the Southeast. Colony P was excluded f rom the PCA, as only single individual was available from this site. ( e ) Relationship between genetic distance (F st ) versus geographic distance for the 12 colonies with echolocation and genetic data in Thailand ( n = 442 individuals). Signi“ cance was tested using a Mantel test ( r = 0.95, R 2 = 0.90, P < 0.001).

PAGE 5

ARTICLE 5 NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1582 NATURE COMMUNICATIONS | 2:573 | DOI: 10.1038/ncomms1582 | www.nature.com/naturecommunications© 2011 Macmillan Publishers Limited. All rights reserved. (Lositan; P < 0.001, Fig. 5a,d,e ), but not when comparing colonies within each side (Lositan; P > 0.05, Fig. 5b,c ). e identi cation of CTC1 as being under selection was also recovered when testing for selection using a hierarchical framework integrating popula-tion structure ( Supplementary Information ). We then identi ed the genomic region containing the microsatellite locus under selec-tion, which was located on the 5 -side of the recombination signal … binding protein for immunoglobulin kappa J region ( RBP-J gene; Supplementary Table S6 ). e RBP-J gene has been shown to be involved in hair cell formation in the cochlea 28,29 . As bats are using the highest frequencies of all terrestrial mammals, their hearing system and particularly the hair cells in the organ of Corti, where the sound is received and ampli ed, need to be well adapted 22 . e microsatellite itself is most likely not under selection, but may be linked to parts of the genome that are under selection, and the RBP-J gene may be another candidate  echolocation  gene, although further investigations are needed to con rm this. Ecological factors driving divergent echolocation . From an eco-logical perspective, the reasons for this sudden change in echoloca-tion peak frequency are not well known. Under an adaptive hypoth-esis, it would be driven by abiotic or biotic factors 11 . Abiotic factors such as temperature and humidity a ect the attenuation of sound (echolocation call) in the air in a frequency-dependant manner, and therefore in uence the maximum distance of prey detection 30 , a process that could be responsible for di erences in echolocation frequency between populations living in di erent areas and fac-ing dissimilar climatic conditions 11 . Nevertheless, the resulting interplay of temperature and humidity conditions throughout the Craseonycteris range in ailand shows little variation, suggesting that attenuation should be constant throughout the range and there-fore have a limited role in shaping the echolocation call  s frequency ( Fig. 6 ). Furthermore, the observed di erences in echolocation frequency result in high sound attenuation values in the Southern colonies (colonies I to O, Fig. 6 ). If the Southern colonies had a peak frequency similar to Northern / Central colonies ( ~ 74 kHz, colonies A to F), they would bene t from an increased detection distance and the attenuation would be very similar across colonies. Rather, the di erence in echolocation frequency seems to render the South-ern colonies less-optimal in terms of prey detection distance (for example, increased sound attenuation; Fig. 6 ). erefore, it seems unlikely that echolocation frequency has changed to keep attenua-tion constant across the di erent colonies. Response to di erent biotic factors (such as interspeci c com-petition) could explain the pattern of echolocation call variation 31 . A shi in peak frequency (character displacement) may act to avoid using the same frequency bandwidth as another small sympat-ric bat species, Myotis siligorensis . Despite being phylogenetically distant, C. thonglongyai and M. siligorensis produce similar echo-location calls at ~ 70 kHz (ref. 18; Fig. 7 ), are comparable in size, ABCDEFIJLMKNO O F A B D E J L M KI C N Frequency (kHz)Colonies707274767880 Figure 4 | Echolocation call variation within Thailand. Graph showing the variation in frequency of free ” ying C. thonglongyai recorded at different colonies represented on the inset map. Colonies are arranged on the X axis according to their score at the “ rst axis (explaining 98.9 % of the variance) of a principal component analysis on their geographic coordinates (latitude, longitude). For each boxplot, the box represents the 0.25 quantile, median and 0.75 quantile. On either side of the box, the whiskers extend to the minimum and maximum. To showcase possible outliers, whiskers were shortened to a length of one time the box length. The colours in the plot correspond to the colours in the inset map. 277–281277–285281–281281–285281–289281–309285–285285–289289–289300–300300–304300–308300–312304–304304–308304–312308–308308–312ed0.00.20.40.60.81.00.000.050.100.15Southern coloniesExpected heterozygosityFst0.00.20.40.60.81.00.00.20.40.6Northern/Central coloniesExpected heterozygosityFst0.00.20.40.60.81.00.00.10.20.30.40.5All coloniesExpected heterozygosityFstacb Figure 5 | Results of F st outlier approach detecting microsatellites deviating from model of neutral evolution. Solid back lines represent the average expected distribution under a neutral model, and the grey lines indicate the upper and lower 99 % con“ dence interval for the expected. Each dot represents a microsatellite locus (the locus CTC1 is indicated by an arrow), red dots identifying candidates under positive selection, ( a ) for all investigated colonies, ( b ) for colonies North of the area of abrupt echolocation change (colonies A to F), and ( c ) for colonies South of the area of abrupt echolocation change (colonies G to O). Genotype frequencies are represented for Northern / Central (left) and Southern (right) colonies in Thailand, for two loci, ( d ) under selection (CTC1) or ( e ) evolving neutrally (CTC104).

PAGE 6

ARTICLE 6 NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1582 NATURE COMMUNICATIONS | 2:573 | DOI: 10.1038/ncomms1582 | www.nature.com/naturecommunications© 2011 Macmillan Publishers Limited. All rights reserved. forage in the same habitat and exhibit similar hunting behaviour 18 . is hypothesis would be supported if M. siligorensis was more abundant in the south of Kanchanaburi, where C. thonglongyai has a high echolocation call, than in the North / Centre, where C. thonglongyai peak frequency is lower ( Fig. 4 ). To test this character displacement hypothesis, we visually screened 1,775 recordings from the surrounding areas of 14 caves, where we caught C. thonglongyai . Indeed, in six out of eight sites surveyed in the South, M. siligorensis was recorded multiple times (230 / 1,327 recordings), whereas in the North / Centre, no individual was recorded at any of the six sites surveyed (0 / 448 recordings; Fig. 1 ). Discussion e extent of gene ow is of crucial importance during the speciation process; for example, the complete absence of gene ow between two populations of the same species will lead these populations to accumulate genetic di erences through time while keeping a pool of complementary genes within each population. e Bateson … Dobzhansky … Muller model 32 , recently supported by empirical evidence 33 , proposes that given enough time, hybrid sterility and inviability will evolve between these two populations due to negative genetic interactions between two or more loci that have accumulated substitutions, ultimately leading to speciation. is process is most likely ongoing between the allopatric ai and Myanmar populations. e ai and Myanmar populations of C. thonglongyai are genetically isolated from each other and show marked echolocation di erences. Whether these changes in echolocation have had a major role in the separation of the two populations remains difcult to test. e signi cantly reduced diversity of the Myanmar population suggests that either the population was previously more diverse and went through a bottleneck or that the population originated from the ai population via a founder e ect. It is di cult to rule out the bottleneck hypothesis, but if this scenario was correct, it would mean that the population was severely reduced to a point where all individuals shared the same haplotype. Although not impossible, this scenario seems unlikely. e recent origin of the Myanmar population from the ai population via a founder e ect is more likely and would explain the reduced level of diversity, whereby only a few individuals founded the Myanmar population. Although possible, an active dispersal of a few Craseonycteris from ailand to Myanmar (200 km) is unlikely given the very short dispersal distance estimated for Craseonycteris (about 2 … 5 km) and its very localized foraging activity (about 1 km from the colony) 34 . A more plausible scenario would be a  sweepstakes  dispersal event 35 of a few Craseonycteris via storms, cyclones or typhoons. e winter monsoon winds, blowing from East-South-East to WestNorth-West, could easily transport Craseonycteris over the 200 km separating the two populations. Episodes of strong winter monsoon winds correspond to orbital periods and have happened every 100-kyr in the past 600-kyr (refs. 36, 37). is  sweepstakes  event would have been immediately followed by isolation and the absence of gene ow between the two populations. e date of the split was dated between 268,000 and 545,000 years BP, which spans over four periods of high winter monsoon activity 36 . It thus seems that although echolocation di erences match perfectly with genetic di erences, echolocation probably did not have a role in the separation of the ai and Myanmar populations. Interestingly, M. siligorensis was captured and / or recorded at all study sites in Myanmar ( Fig. 1 ), further supporting the possible link between the presence of M. siligorensis and high-frequency echolocation calls. e analyses at the population level within ailand further illuminate the contribution of di erent factors during the speciation continuum 7 . Even in the best known examples of  sympatric  speciation, driven by sensory ecology (for example, the Cichlids in Lake Victoria 3 ), it is uncertain whether the variation in sensory traits is the cause or the consequence of the populations  isolation 6 . e quasi-linear distribution of Craseonycteris along the Kwae river valley 16 combined with limited dispersal abilities have shaped a unique system in which to study the contribution of various factors to population structure and gene ow. Within ailand, isolation by geographic distance seems to be the main factor driving the genetic isolation of populations, a result consistent with recent simulations 38 . If echolocation di erences were the primary driver of the populations  isolation, they should explain a much higher percentage of genetic variance and also consistently match the genetic distance, which was not the case. Our results rather support that echolocation di erences evolved secondarily in a context of limited gene ow. Limited 2.9 3.0 3.1 3.2 ColoniesSound attenuation (dB m–1) ABCDEFI JLMKNO Figure 6 | Sound attenuation for the different colonies. Sound attenuation is higher for Southern colonies using higher frequencies. The attenuation for the observed frequency (red) shows a higher variability compared with calculated attenuation if colonies had a similar frequency (74 kHz, squares; 75.5 kHz, triangles; 77 kHz, diamonds). Colonies are arranged on the X axis according to their score at the “ rst axis (explaining 98.9 % of the variance) of a principal component analysis on their geographic coordinates (latitude, longitude). 50% 0% –50% 80kHz 120kHz 0 200 400 600 800 1,000 30% 0% –30% 50% 0% –50% 80kHz 120kHz 0 200 400 600 800 1,000 50% 0% –50% 5msa f e d b c Figure 7 | Comparison of echolocation calls between C. thonglongyai and M. siligorensis . The calls of C. thonglongyai (top panels) and M. siligorensis (bottom panels) were recorded from free ” ying bats in an open environment. The amplitude panels a and b show the pattern of calls over a period of 1 s for C. thonglongyai (( a ), interval between calls = 62 ms) and M. siligorensis (( b ), interval between calls = 87 ms). Panels c and e show in detail the amplitude for one call per species, whereas panels d and f show the fast Fourier transformation spectrogram of these same calls. Frequency change over time for one call per species can be seen in the spectrograms. For all panels, the X axis represents time (in ms).

PAGE 7

ARTICLE 7 NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1582 NATURE COMMUNICATIONS | 2:573 | DOI: 10.1038/ncomms1582 | www.nature.com/naturecommunications© 2011 Macmillan Publishers Limited. All rights reserved. gene ow can enhance rapid local adaptation 39 and even lead to parapatric speciation in the absence of any environmental variability 38 . We however suggest that in C. thonglongyai , limited gene ow creates the conditions that favour the evolution of echolocation di erences via local adaptation and we predict that reinforcement-like selection might drive the speciation process through acoustic isolation limiting mate choice to similar acoustic types 5,15 ( Supplementary Note 1 ). Further studies investigating mate-choice preference and survival of individuals in relation to echolocation call frequency are needed to test if this predicted reinforcement-like selection would occur. e  echolocation  displacement observed could have two functions, which are not necessarily mutually exclusive ( Supplementary Note 2 for further details); species echolocating at di erent frequencies can in theory detect di erent sizes of insect prey 40 , thereby, partition food resources, especially if the prey size distribution is skewed towards small prey 41 . However, this hypothesis has received little empirical support in the literature, especially when the echolocation frequency di erences between species are less than 10 kHz as in the present case 42 (see ref. 43). Alternatively, this character displacement could enable the two species to avoid jamming each other  s echolocation system by each having a private bandwidth 44 , and therefore reduce interference competition. In echolocating bats, the onset of the emitted call activates a neuronal gating mechanism that establishes a time window during which pulseecho pairs are processed for target distance determination 44 . If an echolocation call of a similar frequency is played to an echolocating bat during this time window, the bat is unable to accurately determine target range 45 . erefore, given the similarity in calls between C. thonglongyai and M. siligorensis , an upward shi in frequency, as seen in the Southern C. thonglongyai population where M. siligorensis is present, could have occurred to avoid acoustic jamming . We note that this idea is consistent with the fact that C. thonglongyai  s calls are also emitted at high frequencies in Myanmar, where M. siligorensis occurs. is work shows that sensory ecology and genetic divergence are associated at the early stage of population di erentiation, and that echolocation divergence most likely occurs in a context of limited gene ow, in response to a competitor bat species. We posit that this echolocation divergence may later act to further drive population di erentiation within ailand. Our work highlights the importance of environmental factors (biotic and abiotic) in contributing to population structure and shows the strength of studying the speciation process using both a  spyglass  and  magnifying glass  approach. Ongoing studies of these unique ai and Myanmar populations of C. thonglongyai will provide an opportunity to study the long-term interaction between gene ow and sensory ecology to fully understand the intricacies of the speciation process in nature. Owing to the central role of echolocation in bat biology and its dual role in foraging and communication, further studies using echolocating bats as models to investigate the interplay of sensory ecology, mate choice and gene ow will undeniably advance our understanding of sensory-driven speciation. Methods Echolocation calls . Free ying bats were recorded at dusk for 1 h outside the cave  s entrance using a D1000X ultrasound detector ( Pettersson Elektronik AB ) and automatically stored in a memory card ( Compaq Flash , SanDisk Ultra II , 1 or 4 GB). A sampling frequency of 384 kHz was used, enabling aliasing-free sounds to be recorded up to 153 kHz. For analysis, calls were processed through a Fast Fourier Transformation (1,024 points, Hanning window , fast Fourier transformation s calculated with 90 % time overlap; so ware BatSound version 3.31 , Pettersson Elektronik AB ). Peak frequency (that is, frequency with most energy) was measured using the function  Pulse Characteristics Analysis  available in BatSound. e  mark positioning  was enabled and measurements were visually checked. Each individual  s recording contained multiple calls. Our echolocation call data set consisted of 5,650 calls from 376 recordings. To compare the peak frequency between ailand and Myanmar, we used a generalized linear mixed model via the  lmer  function implemented in the  lme  package in R 2.12.0 (ref. 46). is model considers the nested structure of individuals within sites and sites within countries. A model with calls nested within individuals was too large to be run; therefore, to remove any potential bias caused by nonindependence between calls from the same recording, we generated 10,000 data sets by randomly picking only one call per recording, when performing the generalized linear mixed model. e generalized linear mixed model was performed 10,000 times resulting in10,000 P -values from which we calculated the median. Sample collection and DNA extraction . C. thonglongyai of both sexes were sampled from 15 colonies in ailand ( n = 468) and 4 colonies in Myanmar ( n = 202; Fig. 1 ). Bats were captured either with a hand-net, while they were roosting during the day, or with mist-nets on emergence at dusk. Bats were placed in cloth bags and immediately released a er processing. Two skin biopsies were taken and stored in 200µ l pure ethanol or dried with silic a-gel until extraction. DNA was extracted from wing biopsy punches using a modi ed salt / chloroform extraction protocol 47 that included an additional chloroform / isoamyl alcohol (24 / 1) step a er adding the saturated NaCl solution. DNA samples were used at a nal concentration of 2 … 5 ng · µ l Š 1 ( Supplementary Methods ). Microsatellite genotyping and analysis . e 15 microsatellite primers used in this study were speci cally developed for C. thonglongyai and were ampli ed and genotyped following Puechmaille et al. 48 (locus CTD114 not used). To check for genotyping consistency, 10 % of samples were ampli ed and genotyped twice. A total of 670 samples were genotyped for 15 microsatellite loci. Samples were collected on di erent dates and therefore it was possible that the same individual had been sampled twice. We thus compared multilocus genotypes using GeneCap 49 to identify and remove duplicates from the data set 50 . Out of 670 samples, 11 samples were duplicates (6 in Myanmar and 5 in ailand), leaving a total of 659 individuals for the analysis (463 and 196 for ailand and Myanmar, respectively). No departures from Hardy … Weinberg and linkage equilibrium were detected at the colony level a er Bonferroni correction using Fstat version 2.9.3.2. For the entire data set, population structure was investigated using the program STRUCTURE version 2.2 (ref. 51). We used a burn-in length of 10,000 and a run length of 100,000 without prior population information. e burn-in and run length were chosen a er an initial test, whereby we looked at the convergence of the values of summary statistics and cons istency between runs. All other parameters were le as by default. Ten independent runs were undertaken for Kvalues ranging from 1 to 10, re ecting the minimum and maximum number of populations suspected. e number of populations was inferred from the K statistic as developed in Evanno et al . 52 Using the 659 individual genotypes, a phylogenetic tree was constructed by the neighbour-joining method with the Da distance. e tree construction was carried out using the program Populations 1.2.30. Within ailand (463 individuals), F st and population di erentiation were calculated and tested using Genepop v 4.0.6 (ref. 53). Mitochondrial DNA and phylogenetic reconstruction . A 1,840-bp mitochondrial DNA fragment encompassing the entire Cytb , tRNA reonine, tRNA Proline and part of the control region was ampli ed by polymerase chain reaction in three overlapping fragments ( Supplementary Fig. S5 and Supplementary Tables S7 and S8 ). Two NUMTs (nuclear pseudogenes) were also ampli ed and used in the phylogenetic analysis. Using the mitochondrial DNA and the two NUMTs, phylogenetic reconstruction was undertaken using the Bayesian inference in BEAST 54 ( Supplementary Methods ). Nuclear introns primers, polymerase chain reaction and genotyping . Five SNPs were identi ed on four nuclear introns ( Supplementary Table S8 ), two found on the X chromosome ( BGN and PHKA2 ), the Y chromosome ( SMCY7 ) and one on chromosome 16 for Homo sapiens ( ZP2 ). For these ve SNPs, we designed custom TaqMan SNP assays ( Applied Biosystems ; Supplementary Table S9 ) and screened the 659 individuals ( Supplementary Methods ). Genetic diversity indices . To test for di erences in genetic diversity between the ai and Myanmar populations, we grouped colonies into two groups corresponding to ailand and Myanmar and compared their genetic diversity at the nuclear and mitochondrial level in Fstat version 2.9.3.2. Signi cance was tested using 10,000 permutations. Samples were randomly allocated to the two groups, and the genetic diversity was calculated for each permutation. e P -value of the test was the proportion of randomized data sets giving a larger value for the genetic index than the observed one (one-sided test). Population history and structure . e ai population  s demographic history was examined using the mismatch distribution of 462 mitochondrial DNA sequences 55 . Population structure was analysed using principal component analysis on microsatellite data with R version 2.12.0 (ref. 46) using the adegenet package version 1.3-0 (ref. 56). Simulations of a spatially expanding population mimicking the expansion of the ai population were carried out in SPLATCHE 57 , and outputs analysed in Arlequin ver 3.5 (ref. 58; via arlecore) and displayed in R 2.12.0 (ref. 46). A 100 × 10 lattice was used, representing the distribution of Craseonycteris along the Kwae River valley 16 . Each deme was considered to be sending migrants to its neighbours at a rate m . Once a deme was colonized, its population size grew

PAGE 8

ARTICLE 8 NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1582 NATURE COMMUNICATIONS | 2:573 | DOI: 10.1038/ncomms1582 | www.nature.com/naturecommunications© 2011 Macmillan Publishers Limited. All rights reserved. following a logistic model with rate r and carrying capacity C . e initial population size was set to 2,000 and the  AllowSourcePopulationOver ow  parameter was set to  0  . A homogene friction map was used. A set of 12 locations spaced evenly along the  lat  axis of the lattice (starting at deme 4 and every 8 demes) were sampled for 20 individuals. Sampling was carried out just a er the entire lattice was completely lled in. A set of 14 unlinked microsatellite loci with a mutation rate of 5 × 10 Š 4 were simulated. On the basis of available information on Craseonycteris colony size 16 , the simulations were run for two values of the carrying capacity C , 100 and 250. For each value of C , 12 combinations of the m (0.1, 0.3, 0.5, 0.7) and r (0.1, 0.3, 0.5) parameters were simulated. One hundred replicates were analysed for each combination of parameters. A simulation was considered to give an isolation by distance pattern similar to the observed, if it contained at least one replicate with a maximal F st comprised between 0.2 and 0.3 and a Mantel r greater than 0.9 ( Supplementary Table S3 ). Causal modelling . Causal modelling on resemblance matrices 23,24 was used to investigate the combination of factors driving genetic di erentiation between colonies and the causal relationships between these factors. Mantel and partial Mantel tests were used within the causal modelling framework to assess the support for all possible organizational models 23,24 . All tests were conducted with R version 2.12.0 (ref. 46) using the package ecodist 59 version 1.1.4, and signi cance was assessed with 9,999 permutations ( Supplementary Methods ). Selection tests . e 15 microsatellites from 463 individuals were tested for selection based on the F st outlier approach 27 using two models, and island model and a hierarchical island model in LOSITAN 60 and Arlequin 58 version 3.5, respectively ( Supplementary Methods ). Microsatellite CTC1 genomic position . Before a Blast search, the microsatellite sequence including the anking regions (total of 599 nt) was repeatmasked (RepeatMasker version open-3.2.8). Only simple repeats corresponding to the microsatellite motif were identi ed (from nt 347 to nt 408; TCCA motif). A Blastn search with the default settings and  word size  set to 7 was used to identify genomic regions similar to the microsatellite locus (CTC1) anking regions. For the 5 anking region (346 nt), the best match ( E -value: 9e Š 39; coverage = 84 % , identities = 221 / 302 (73 % )) was much better than the second ( E -value: 0.07; coverage = 14 % , identities = 41 / 51 (80 % )) or third match ( E -value: 0.25; coverage = 10 % , identities = 32 / 37 (86 % )). Similarly, for the 3 anking region (191 nt), the best match ( E -value: 2e Š 06; coverage = 92 % , identities = 136 / 181 (75 % )) was much better than the second ( E -value: 0.003; coverage = 16 % , identities = 29 / 31 (93 % )) or third match ( E -value: 0.01; coverage = 15 % , identities = 28 / 30 (93 % )). Both anking region  s best matches were situated on Human chromosome number 4 (GRCh37 reference primary), around 50 kb (53,653 for 3 anking region and 52,745 for 5 anking region) at 5 side of the  recombining binding protein suppressor of hairless isoform 3 , also known as RBP-J ( H. sapiens geneID: 3516; chromosome: 4, location: 4p15.2). As we do not know if there is synteny around RBP-J between C. thonglongyai and H. sapiens , we then used Ensembl to identify potential synteny around RBP-J between H. sapiens and 12 other genomes sequenced at high coverage ( Supplementary Table S6 ). We showed that large syntenic blocks (average = 30.2 M) around the RBP-J gene in H. sapiens were also present in all species investigated ( Supplementary Table S6 ). Given the conserved synteny observed in other species investigated, it is therefore likely that microsatellite CTC1 in C. thonglongyai is also in the 5 -side of the RBP-J gene. References 1 . Doebeli , M . & Dieckmann , U . Speciation along environmental gradients . Nature 421 , 259 … 264 ( 2003 ). 2 . Kawata , M . , Shoji , A . , Kawamura , S . & Seehausen , O . A geneticall y explicit model of speciation by sensory drive within a continuous population in aquatic environments . BMC Evol. Biol. 7 , 99 ( 2007 ). 3 . Seehausen , O . et al. Speciation through sensory drive in cichlid sh . Nature 455 , 620 … 626 ( 2008 ). 4 . Boughman , J . W . Divergent sexual selection enhances reproductive isolation in sticklebacks . Nature 411 , 944 … 948 ( 2001 ). 5 . Kingston , T . & Rossiter , S . J . Harmonic-hopping in Wallaceas bats . Nature 429 , 654 … 657 ( 2004 ). 6 . Nosil , P . , Harmon , L . J . & Seehausen , O . Ecological explanations for (incomplete) speciation . Trends Ecol. Evol. 24 , 145 … 156 ( 2009 ). 7 . Via , S . Natural selection in action during speciation . Proc. Natl Acad. Sci. USA 106 , 9939 … 9946 ( 2009 ). 8 . Harrison , R . Molecular changes at speciation . Annu. Rev. Ecol. Syst. 22 , 281 … 308 ( 1991 ). 9 . Templeton , A . (ed). e Role of Molecular Genetics in Speciation Studies ( Birkhauser-Verlag , 1998 ) . 10 . Losos , J . B . & Glor , R . E . Phylogenetic comparative methods and the geography of speciation . Trends Ecol. Evol. 18 , 220 … 227 ( 2003 ). 11 . Endler , J . A . Signals, signal conditions, and the direction of evolution . Am. Nat. 139 , S125 … S153 ( 1992 ). 12 . Fenton , M . B . Eavesdropping on the echolocation and social calls of bats . Mammal Rev. 33 , 193 … 204 ( 2003 ). 13 . omas , J . A . , Moss , C . F . & Vater , M . (eds) Echolocation in Bats and Dolphins ( e University of Chicago Press , 2004 ) . 14 . Jones , G . & Siemers , B . e communicative potential of bat echolocation pulses . J. Comp. Physiol. A 197 , 447 … 457 ( 2011 ). 15 . Boughman , J . W . How sensory drive can promote speciation . Trends Ecol. Evol. 17 , 571 … 577 ( 2002 ). 16 . Puechmaille , S . J . et al. Population size, distribution, threats and conservation status of two endangered bat species: Craseonycteris thonglongyai and Hipposideros turpis . Endang. Species Res. 8 , 15 … 23 ( 2009 ). 17 . Pereira , M . J . R . et al. Conservation of the worlds smallest mammal, the bumble-bee bat Craseonycteris thonglongyai , in Myanmar . Oryx 40 , 456 … 463 ( 2006 ). 18 . Surlykke , A . et al. Echolocation in two very small bats from ailand: Craseonycteris thonglongyai and Myotis siligorensis . Behav. Ecol. Sociobiol. 33 , 1 … 12 ( 1993 ). 19 . Khin Mie Mie PhD dissertation, University of Yangon, 2004 . 20 . Templeton , A . Mechanisms of speciation--a population genetic approach . Annu. Rev. Ecol. Syst. 12 , 23 … 48 ( 1981 ). 21 . Ray , N . , Currat , M . & Exco er , L . Intra-deme molecular diversity in spatially expanding populations . Mol. Biol. Evol. 20 , 76 … 86 ( 2003 ). 22 . Neuweiler , G . e Biology of Bats ( Oxford University Press , 2000 ) . 23 . Legendre , P . & Trousselier , M . Heterotrophic bacteria: modeling in the presence of spatial autocorrelation . Limnol. Oceanogr. 33 , 1055 … 1067 ( 1988 ). 24 . Legendre , P . Autocorrelation: trouble or paradigm . Ecology 74 , 1659 … 1673 ( 1993 ). 25 . Hartl , D . L . & Clark , A . G . Principles of Population Genetics ( Sinauer Associates, Inc. , 1997 ) . 26 . Mullen , L . M . & Hoekstra , H . E . Natural selection along an environmental gradient: a classic cline in mouse pigmentation . Evolution 62 , 1555 … 1570 ( 2008 ). 27 . Beaumont , M . A . & Nichols , R . A . Evaluating loci for use in the genetic analysis of population structure . Proc. R. Soc. Lond. B 263 , 1619 … 1626 ( 1996 ). 28 . Yamamoto , N . et al. Inhibition of Notch/RBP-J signaling induces hair cell formation in neonate mouse cochleas . J. Mol. Med. 84 , 37 … 45 ( 2006 ). 29 . Hori , R . et al. Pharmacological inhibitio n of Notch signal in the mature guinea pig cochlea . Neuroreport 18 , 1911 … 1914 ( 2007 ). 30 . Lawrence , B . D . & Simmons , J . A . Measurements of atmospheric attenuation at ultrasonic frequencies and the signi cance for echolocation by bats . J. Acoust. Soc. Am. 71 , 585 … 590 ( 1982 ). 31 . Heller , K.G . & Von Helversen , O . Resource partitioning of sonar frequency bands in rhinolophoid bats . Oecologia 80 , 176 … 186 ( 1989 ). 32 . Orr , H . e population genetics of speciation: the evolution of hybrid incompatibilities . Genetics 139 , 1805 … 1813 ( 1995 ). 33 . Matute , D . , Butler , I . , Turissini , D . & Coyne , J . A test of th e snowball theory for the rate of evolution of hybrid incompatibilities . Science 329 , 1518 … 1521 ( 2010 ). 34 . Duangkhae , S . Ecology and behaviour of the Kittis Hog-nosed bat ( Craseonycteris thonglongyai ) in Western ailand . Nat. Hist. Bull. Siam Soc. 38 , 135 … 161 ( 1990 ). 35 . Simpson , G . G . Mammals and land bridges . J. Wash. Acad. Sci. 30 , 137 … 163 ( 1940 ). 36 . Liu , Z . et al. Clay mineral assemblages in the northern South China Sea: implications for East Asian monsoon evolution over the past 2 million years . Mar. Geol. 201 , 133 … 146 ( 2003 ). 37 . Beaufort , L . , de Garideloron , T . , Linsley , B . , Oppo , D . & Buchet , N . Biomass burning and oceanic primary production estimates in the Sulu Sea area over the last 380 kyr and the East Asian monsoon dynamics . Mar. Geol. 201 , 53 … 65 ( 2003 ). 38 . Hoelzer , G . A . , Drewes , R . , Meier , J . & Doursat , R . Isolati on-by-distance and outbreeding depression are su cient to drive parapatric speciation in the absence of environmental in uences . PLoS Comput. Biol. 4 , e1000126 ( 2008 ). 39 . Nosil , P . Adaptive population divergence in cryptic color-pattern following a reduction in gene ow . Evolution 63 , 1902 … 1912 ( 2009 ). 40 . Pye , J . D . Is delity futile? e true signal is illusory, especially with ultrasound . Bioacoustics 4 , 271 … 286 ( 1993 ). 41 . Sa , K . & Siemers , B . M . Implications of sensory ecology for species coexistence: biased perception links predator diversity to prey size distribution . Evol. Ecol. 24 , 703 … 713 ( 2010 ). 42 . Houston , R . D . , Boonman , A . M . & Jones , G . Do echolocation signal parameters restrict bats choice of prey? In Echolocation in Bats and Dolphins (eds Jeanette A. omas, Cynthia F. Moss, & Marianne Vater) 339 … 345 ( University of Chicago Press , 2004 ) . 43 . Siemers , B . M . & Schniltzer , H.U . Echolocation signals re ect niche di erentiation in ve sympatric congeneric bat species . Nature 429 , 657 … 661 ( 2004 ). 44 . Roverud , R . C . & Grinnell , A . D . Echolocation sound features processed to provide distance information in the CF/FM bat, Noctilio albiventris : evidence for a gated time window utilizing both CF and FM components . J. Comp. Physiol. A 156 , 457 … 469 ( 1985 ). 45 . Neuweiler , G . Auditory adaptations for prey capture in echolocating bats . Physiol. Rev. 70 , 615 … 641 ( 1990 ).

PAGE 9

ARTICLE 9 NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1582 NATURE COMMUNICATIONS | 2:573 | DOI: 10.1038/ncomms1582 | www.nature.com/naturecommunications© 2011 Macmillan Publishers Limited. All rights reserved. 46 . R Development Core Team . R: a language and environment for statistical computing, http://cran.r-project.org/, Vienna (Austria) ( R Foundation for Statistical Computing , 2010 ) . 47 . Miller , S . A . , Dykes , D . D . & Polesky , H . F . A simple salting out pr ocedure for extracting DNA from human nucleated cells . Nucleic Acids Res. 16 , 1215 ( 1988 ). 48 . Puechmaille , S . J . et al. Characterization and multiplex genotyping of 16 polymorphic microsatellite loci in the endangered bumblebee-bat, Craseonycteris thonglongyai (Chiroptera: Craseonycteridae) . Conserv. Genet. 10 , 1073 … 1076 ( 2009 ). 49 . Wilberg , M . J . & Dreher , B . P . GENECAP: a program for analysis of multilocus genotype data for non-invasive sampling and capture-recapture population estimation . Mol. Ecol. Notes 4 , 783 … 785 ( 2004 ). 50 . Puechmaille , S . J . & Petit , E . J . Empirical evaluation of non-invasive capturemark-recapture estimates of population size based on a single sampling session . J. Appl. Ecol. 44 , 843 … 852 ( 2007 ). 51 . Pritchard , J . K . , Stephens , M . & Donnelly , P . Inference of population st ructure using multilocus genotype data . Genetics 155 , 945 … 959 ( 2000 ). 52 . Evanno , G . , Regnaut , S . & Goudet , J . Detecting the number of clusters of individuals using the so ware STRUCTURE: a simulation study . Mol. Ecol. 14 , 2611 … 2620 ( 2005 ). 53 . Rousset , F . genepop  007: a complete re-implementation of the genepop so ware for Windows and Linux . Mol. Ecol. Notes 8 , 103 … 106 ( 2008 ). 54 . Drummond , A . J . & Rambaut , A . BEAST: Bayesian evolutionary analysis by sampling trees . BMC Evol. Biol. 7 , 214 ( 2007 ). 55 . Slatkin , M . & Hudson , R . R . Pairwise comparisons of mitochondrial DNA sequences in stable and exponentially growing populations . Genetics 129 , 555 … 562 ( 1991 ). 56 . Jombart , T . adegenet: a R package for the multivariate analysis of genetic markers . Bioinformatics 24 , 1403 … 1405 ( 2008 ). 57 . Ray , N . , Currat , M . , Foll , M . & Exco er , L . SPLATCHE2: a spatially explicit simulation framework for complex demography, genetic admixture and recombination . Bioinformatics 26 , 2993 … 2994 ( 2010 ). 58 . Exco er , L . & Lischer , H . E . L . Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows . Mol Ecol Resour 10 , 564 … 567 ( 2010 ). 59 . Goslee , S . C . & Urban , D . L . e ecodist package for dissimilarity-based analysis of ecological data . J. Stat. So w. 22 , 1 … 19 ( 2007 ). 60 . Antao , T . , Lopes , A . , Lopes , R . J . , Beja-Pereira , A . & Luika rt , G . LOSITAN: a workbench to detect molecular adaptation based on a Fst -outlier method . BMC Bioinformatics 9 , 323 ( 2008 ). Acknowledgements is paper is dedicated to the memory of Professor Daw Tin Nwe of Yangon University for her encouragement and assistance with the research and for her kindness and humour, which she shared with all who knew her. We thank Sai Yok and Erawan National Park ( ailand) for assisting us during the survey; the Institut pour la Recherche et le D é veloppement, Unit é de Recherche 178, and Mahidol University for their logistic support; Nu Nu Aye, Wai Wai Myint, Aye ida and ida Tin for their help in the eld. is project was funded by Science Foundation Ireland (RFP GEN0056) and supported by the Royal Irish Academy, the Darwin Initiative (DEFRA) and the Royal Society (International project). Author contributions E.C.T. and P.J.B. conceived the study. E.C.T, P.J.B., C.S., T.N. and S.S.H.B. supervised the project either during eldwork in ailand or Myanmar and / or in the laboratory. S.J.P., M.A.G., P.P., M.Y., I.J.M. and K.M.M. surveyed caves and collected genetic samples. S.J.P. collected the acoustic data, generated the genetic data and performed all the analysis. E.J.P. and E.C.T. helped with the genetic analysis. S.J.P and E.C.T. wrote the paper. All authors discussed the results, implications and edited the manuscript. Additional information Accession codes: e data have been deposited in the GenBank database under accession codes GU247601 to GU247751. Supplementary Information accompanies this paper at http://www.nature.com/ naturecommunications Competing nancial interests: e authors declare no competing nancial interests. Reprints and permission information is available online at http://npg.nature.com/ reprintsandpermissions/ How to cite this article: Puechmaille, S.J. et al. e evolution of sensory divergence in the context of limited gene ow in the bumblebee bat. Nat. Commun. 2:573 doi: 10.1038 / ncomms1582 (2011). License: is work is licensed under a Creative Commons Attribution-NonCommercialNoDerivative Works 3.0 Unported License. To view a copy of this license, visit http:// creativecommons.org/licenses/by-nc-nd/3.0/


printinsert_linkshareget_appmore_horiz

Download Options

close


  • info Info

    There are only PDFs associated with this resource.

  • link PDF(s)



Cite this item close

APA

Cras ut cursus ante, a fringilla nunc. Mauris lorem nunc, cursus sit amet enim ac, vehicula vestibulum mi. Mauris viverra nisl vel enim faucibus porta. Praesent sit amet ornare diam, non finibus nulla.

MLA

Cras efficitur magna et sapien varius, luctus ullamcorper dolor convallis. Orci varius natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Fusce sit amet justo ut erat laoreet congue sed a ante.

CHICAGO

Phasellus ornare in augue eu imperdiet. Donec malesuada sapien ante, at vehicula orci tempor molestie. Proin vitae urna elit. Pellentesque vitae nisi et diam euismod malesuada aliquet non erat.

WIKIPEDIA

Nunc fringilla dolor ut dictum placerat. Proin ac neque rutrum, consectetur ligula id, laoreet ligula. Nulla lorem massa, consectetur vitae consequat in, lobortis at dolor. Nunc sed leo odio.