Diversification of A

Home , Actenoides, Beach kingfisher, Ceyx (bird), Collared kingfisher, Common cicadabird, Common kingfisher

Ibis (2018) doi: 10.1111/ibi.12688

Diversiﬁcation of a ‘great speciator’ in the Wallacea region: differing responses of closely related resident and migratory kingﬁsher species (Aves: Alcedinidae: Todiramphus)

DARREN P. O’CONNELL,1,2* DAVID J. KELLY,1,2 NAOMI LAWLESS,1,2 ADI KARYA,3 KANGKUSO ANALUDDIN3 & NICOLA M. MARPLES1,2 1Department of Zoology, School of Natural Sciences, Trinity College Dublin, Dublin DO2 CX56, Ireland 2Trinity Centre for Biodiversity Research, Trinity College Dublin, Dublin DO2 CX56, Ireland 3Department of Biology and Biotechnology, Universitas Halu Oleo, Kendari, Southeast Sulawesi, Indonesia

The Collared Kingfisher species complex is the most widespread of the ‘great speciator’ lineages of the Indo-Pacific. They have shown a remarkable ability to spread and diversify. As a result of this rapid diversification, Todiramphus species are often found in secondary sympatry. In Southeast Sulawesi, Indonesia, two Todiramphus species are present, the breeding resident Collared Kingfisher Todiramphus chloris and the overwintering migratory Sacred Kingfisher Todiramphus sanctus. We investigated the effect of isolation on these closely related species by comparing their populations on mainland Sulawesi and its larger continental islands, with populations on the small, oceanic Wakatobi Islands. Within our wider analysis we provide further support for the distinctiveness of the Sulawesi Collared Kingfisher population, perhaps isolated by the deep water barrier of Wallace’s line. Within Sulawesi we found that populations of Collared Kingfisher on the Wakatobi Islands had diverged from those on mainland Sulawesi, differing both in morphology and in mitochondrial DNA. In contrast, there was no divergence between Sacred Kingfisher populations in either morphology or mitochondrial DNA. We propose that a difference in habitat occupied by Collared Kingfisher populations between the mainland and continental islands vs. oceanic islands has caused this divergence. Mainland Collared Kingfishers are predomi- nately found inland, whereas Wakatobi Collared Kingfishers are also found in coastal habitats. The larger body size of Wakatobi Collared Kingfisher populations may be a result of increased competition with predominantly coastal Sacred Kingfisher populations. The uniform nature of Sacred Kingfisher populations in this region probably reflects their consistent habitat choice (coastal mangrove) and their migratory nature. The demands of their breeding range are likely to have an even stronger selective influence than their Sulawesi wintering range, limiting their scope for divergence. These results provide insight into the adaptability of the widespread Todiramphus lineage and are evidence of the need for further taxonomic revision of Collared Kingfisher populations. Keywords: competition, evolution, great speciator, Indo-Pacific, islands, kingfisher, Todiramphus, Wallacea.

Island bird populations have historically been of biogeography (Darwin 1859, Wallace 1869). They great importance in the study of evolution and provide discrete units which provide insight into how species adapt and change in relative isolation (MacArthur & Wilson 1967, Whittaker & Fernan- *Corresponding author. dez-Palacios 2007). Modern molecular tools have Email: [email protected] provided new avenues for this research, allowing Twitter: @oconned5

for both greater insight into the evolutionary his- (Andersen et al. 2015b), making it one of the fast- tory of the taxa (Jetz et al. 2012) and the discov- est diversifying lineages of birds (Moyle et al. ery of previously unrecognized ‘cryptic’ species 2009, Jetz et al. 2012). The colonization of ocea- (Bickford et al. 2007). The Indo-Pacific is of par- nic islands is thought to have played a major part ticular importance to this area of research. This in the extraordinary diversification of the Collared region has great potential for cryptic diversity Kingfisher species complex. The rapid diversifica- (Lohman et al. 2010) and its many islands make it tion seen in this group occurred when colonizing perfect for studying the tempo and mode of speci- the oceanic islands of Wallacea, the Philippines ation in birds. The Indo-Pacific is home to a num- and the Pacific (Andersen et al. 2018). ber of groups of birds known as ‘great speciators’ This ability to colonize islands has led to multi- (Diamond et al. 1976), taxa renowned for their ple instances of secondary sympatry, where two or large geographical ranges and rapid diversifications more recently diverged Todiramphus species are (Mayr & Diamond 2001). found on the same island (Woodall 2001). Such Mayr and Diamond (2001) developed the ‘great closely related species have similar ecological speciator’ concept for their study system in North- requirements so they might be expected to com- ern Melanesia to describe a group of birds with pete strongly for resources (MacArthur & Levins high inter-island geographical variation, including 1967, Martin & Martin 2001, Lovette & diverse taxa found across many islands (e.g. Louisi- Hochachka 2006). However, multiple coloniza- ade White-eye Zosterops griseotinctus, Moluccan tions of island archipelagos have occurred in many Dwarf Kingfisher Ceyx lepidus, Australian Golden taxa, with very different outcomes, depending on Whistler Pachycephala pectoralis and Common the traits of those taxa. For example, although Cicadabird Edolisoma tenuirostre). Because of their multiple colonizations of Pacific reed-warblers wide ranges and multiple distinct populations, the (Acrocephalus) occurred in three archipelagos, ‘great speciators’ have provided ideal study systems species from different lineages do not co-occur on for developing many key concepts in evolutionary any island (Cibois et al. 2011). Reed-warblers live biology (Mayr 1942, Diamond 1974, Diamond in high-density populations of territorial pairs and et al. 1976). In recent years modern molecular trios, potentially saturating available habitat and methods have enabled researchers to begin to preventing the establishment of new immigrants uncover the intricate evolutionary histories of the (Craig 1992, Graves 1992, Thibault & Cibois ‘great speciators’, showing them to be complexes 2006). In contrast, two species of white-eye (Zos- of closely related species (Andersen et al. 2013, teropidae) coexist on several Mariana Islands (Sli- 2014, Irestedt et al. 2013, Pedersen et al. 2018). kas et al. 2000), probably aided by the social The Collared Kingfisher Todiramphus chloris flocking behaviour of white-eyes (van Balen 2008). species complex is one of the most widespread of The phenomenon of multiple colonizations of the ‘great speciator’ lineages of the Indo-Pacific, island archipelagos is perhaps best studied in the covering over 16 000 km from the Red Sea to Indian Ocean archipelago, a diverse collection of Polynesia (Woodall 2001, Andersen et al. 2015b). islands in terms of both island area and ecology This species complex shows great diversification (Whittaker & Fernandez-Palacios 2007). White- across its wide range, encompassing the Collared eyes and sunbirds (Nectarinia) show a complex Kingfisher (14 subspecies) and five species recently pattern of island occupancy in the Indian Ocean taxonomically split from the Collared Kingfisher in archipelago. Most islands are home to only one the IOC World Bird List (v. 8.1), based on work species from each genus and competition with by Andersen et al. (2015b): the Torresian King- congeneric species limits the further diversification fisher Todiramphus sordidus (three subspecies), of colonists (Warren et al. 2003, 2006). There are Islet Kingfisher Todiramphus colonus (monotypic), only a few exceptions. White-eye species live in Mariana Kingfisher Todiramphus albicilla (three sympatry on only three Indian Ocean islands, seg- subspecies), Melanesian Kingfisher Todiramphus regated according to altitude (Warren et al. 2006). tristrami (seven subspecies) and Pacific Kingfisher Sunbirds are also only found in sympatry on three (22 subspecies) (Gill & Donsker 2018). This level islands, partitioned by morphological niche (Bij- of diversification is particularly remarkable given nens et al. 1987). Sunbirds are even excluded from that the Collared Kingfisher species complex La Reunion and Mauritius by the endemic white- started diversifying within the last 0.57–0.85 Ma eyes (Reunion Olive White-eye Zosterops olivaceus

and Mauritius Olive White-eye Zosterops chloro- significantly revised the taxonomy of this remark- nothos respectively), which have abnormally long able diversification, the morphological and ecologi- bills for white-eyes and fill the sunbird niche (Gill cal adaptations that led to the isolation of the 1971, Cheke 1987). Clearly, interactions with tax- different populations remain to be studied. While onomically and ecologically similar species, as well morphology and phylogenetics have co-varied in as island characteristics, are important for the many taxa (Jablonski & Finarelli 2009, McKay spread of island-colonizing species (Franzen et al. et al. 2010, Dong et al. 2015, Liu et al. 2016) 2012). there are multiple examples of differing morpho- Todiramphus populations in secondary sympatry logical and phylogenetic patterns, particularly in have been shown to exhibit separation by habitat recently diverged island fauna (Cibois et al. 2007, preference, which allow these closely related spe- Phillimore et al. 2008, Saitoh et al. 2012). cies to partition niches and maintain reproductive The southeastern peninsula of Sulawesi provides isolation (Fry 1980, Woodall 2001). The Collared an excellent study system to test the effect of iso- Kingfisher complex shows great adaptability in its lation on species (Fig. 1). It includes continental habitat preferences in different parts of its range islands (Buton, Muna and Kabaena), which were (Woodall 2018b), facilitating this partitioning. connected to Sulawesi at the time of the last gla- Human modification of the environment can also cial maximum, around 10 000 years ago (Voris alter the dynamic of this habitat partitioning. 2000), and also oceanic islands (the Wakatobi Ward (1968) noted that one result of increasing Islands), which have never been connected to the urbanization in Singapore was that the Collared Sulawesi mainland (Carstensen et al. 2012). Kingfisher replaced the White-breasted Kingfisher Although the Wakatobi Islands are only separated Halcyon smyrnensis as the most common garden from Buton by 27 km, they are home to five kingfisher. The Collared Kingfisher species com- unique bird subspecies (Kelly & Marples 2010) plex is found in sympatry with multiple local and a proposed new species of flowerpecker (Kelly endemic congeneric taxa throughout its range, et al. 2014). However, the Wakatobi’s kingfisher including: the Beach Kingfisher Todiramphus populations have not been investigated since early saurophagus (north and east New Guinea and sur- taxonomic assessments of the region (Hartert rounding islands), Vanuatu Kingfisher Todiramphus 1903). Both the Collared Kingfisher and the farquhari (central Pacific), Talaud Kingfisher Todi- Sacred Kingfisher are present in Southeast Sula- ramphus enigma (Talaud Island), Rusty-capped wesi, allowing the effect of isolation to be tested Kingfisher Todiramphus pelewensis (Palau Island) on closely related Todiramphus kingfishers, one res- and Ultramarine Kingfisher Todiramphus leucopy- ident and one migratory. gius (Solomon Islands). The Collared Kingfisher This study aims: (1) to assess the genetic struc- species complex is also found in sympatry with ture of Todiramphus populations in Southeast the migratory Sacred Kingfisher Todiramphus sanc- Sulawesi using mitochondrial molecular markers, tus. Its migratory nature makes the Sacred King- with the prediction that the Wakatobi Collared fisher unique among Todiramphus kingfishers. This Kingfisher population may have diverged from the highly vagile lifestyle may be a vestige of the mainland population; (2) to assess whether the remarkable dispersal ability that allowed Todiram- morphology of the populations mirrors the genetic phus kingfishers to diversify across the Pacific structure, and to interpret the ecological relevance (Mayr & Diamond 2001). The Sacred Kingfisher of any morphological divergence seen; and (3) to has a wide range from Western Australia to New assess whether the resident Collared Kingfisher Caledonia (Woodall & Kirwan 2018b). shows greater evidence of local adaptation to the Although molecular work has begun uncovering Wakatobi Islands compared with the migratory the evolutionary history of the ‘great speciator’ Sacred Kingfisher. radiations (Moyle et al. 2009, Andersen et al. 2013), more focus on the morphological adapta- METHODS tions of these birds is needed for a greater under- standing of the ecological requirements of island Study site and sampling colonization (but see Irestedt et al. 2013). This is also the case for the Collared Kingfisher species Sampling was carried out throughout Southeast complex. Although Andersen et al. (2015b) Sulawesi (Fig. 1) on research expeditions

Wallace's Line Menui T. c. collaris Mainland Sulawesi Wawonii

Borneo Sulawesi

T. c. laubmannianus Papua Muna

Buton

Todiramphus chloris chloris Kabaena

Wakatobi Islands Java Bali Timor Flores T. c. palmeri 0 Lombok 0 500 km 100 km

Figure 1. Map showing the Sulawesi region of Indonesia (left panel) and the study region of Southeast Sulawesi (right panel). Cur- rent Collared Kingﬁsher subspecies divisions (Woodall 2018b) are shown by dotted lines, with Wallace’s line shown by a dashed line (left panel). Locations where Todiramphus kingﬁshers were sampled for this study are marked by .

undertaken between 1999 and 2017 in the months measurements of wing (maximum chord), bill, and of June–September by D.J.K., N.M.M., K.A., skull lengths and mass (Redfern & Clark 2001) D.O.C. and A.K. Todiramphus species were sam- were taken. All measurements were made by a sin- pled on 12 islands throughout the region. For gle recorder (N.M.M.). Only adult birds were additional details on sampling locations see included in this analysis. Collared Kingfishers and Table S1. Collared Kingfishers and Sacred King- Sacred Kingfishers are morphologically monomor- fishers were caught on both the Wakatobi Islands phic (Rogers et al. 1986, Higgins 1999), so sam- and ‘mainland’ islands (mainland Sulawesi and the pled individuals were not separated according to large continental islands of Buton, Muna and sex in the morphological analyses. Approximately Kabaena). Only Sacred Kingfishers were caught on 5–10 contour feathers were plucked from the flank the isolated island of Menui (Fig. 1), so no assess- of each bird and stored in sealed paper envelopes. ment of Collared Kingfishers could be made there. Contour feathers were sampled to minimize the Mist-nets were used to trap birds for sampling. risk of injury to the birds and avoid disruption of Care was taken with the identification and ageing flight ability and plumage-based visual signals of these similar species. Collared Kingfisher have a (McDonald & Griffith 2011). Mist-netting was car- ‘clean white collar and underparts’ (MacKinnon & ried out in a variety of habitats used by Todiram- Phillipps 1993) with ‘white (not buff) lores’ (Hig- phus species including plantation, forest edge, gins 1999). Juvenile Collared Kingfishers were dis- farmland and mangroves. tinguished by ‘forehead and secondary upperwing- coverts finely scaled buff’ (Higgins 1999). Sacred DNA sequencing Kingfishers are ‘smaller than Collared Kingfisher, duller greenish blue with buffy wash to underparts DNA was extracted from feathers using a Qiagen and lores’ (MacKinnon & Phillipps 1993). Juvenile DNeasy Blood and Tissue Kit (Qiagen, California, Sacred Kingfishers were distinguished by ‘feathers USA), following Kelly et al. (2014). We sequenced of forehead and secondary coverts of upperwing three mitochondrial genes: the entire second and fringed buff’ (Coates & Bishop 1997, Higgins third subunits of mitochondrial nicotinamide ade- 1999). Woodall (2001) and Eaton et al. (2016) nine dinucleotide dehydrogenase (ND2 and ND3, provided additional reference information for iden- respectively) and a 626-bp region of the cyto- tification and ageing. The morphological chrome c oxidase subunit 1 (COI) gene. We used

the established primers LTyr, COI908aH2 COI sequences of at least the same length as those (Elbourne 2011) and COIHT (Tavares & Baker produced by this study were used in our analysis. 2008) for COI, and L10755 for ND3 (Chesser Our analysis incorporated 29 Todiramphus COI 1999). Several novel primers were developed for sequences (21 produced by this study, with use with ND2 and ND3 (Table S2), primarily to another eight sourced from GenBank), including facilitate the sequencing of ND2 in two halves. the Collared Kingfisher (n = 12), Sacred King- The polymerase chain reaction (PCR) procedure fisher (n = 14), Forest Kingfisher Todiramphus was adapted from Kelly et al. (2014). All PCR macleayii (n = 2) and Mangareva Kingfisher Todi- amplifications were performed in 20-lL reactions, ramphus gambieri (n = 1). Halcyon whole mito- consisting of 8.1 lL double-distilled H2O, 0.4 lL genome sequences (accession nos NC_028177, 10 mM deoxynucleoside triphosphates (dNTPs), NC_024198 and KY940559) provided outgroups 2 lL109 PCR buffer, 2.4 lL25mM MgCl2, for both ND2/ND3 and COI analysis, with the 1 lL10lM forward primer, 1 lL10lM reverse addition of Syma (ND2/ND3), Actenoides and primer, 0.1 lL Taq polymerase (New England Dacelo (COI) samples to represent closer relatives Biolabs) and 5 lL template DNA. Annealing tem- to Todiramphus. perature was 55 °C for ND2, 50 °C for ND3 and 57 °C for COI. All reactions were amplified under Phylogenetic and genetic analyses the following thermal cycler conditions: 4 min at 94 °C, followed by 45 cycles of 1 min at 94 °C, Sequences were aligned using ClustalW multiple 1.5 min at the gene-specific annealing temperature alignment in BioEdit v7.2.5 (Hall 1999) and the and 1.5 min at 72 °C, finishing with 5 min at 72 ND2 and ND3 genes were concatenated using °C. Amplified PCR products were screened on 2% Mesquite v3.40 (Maddison & Maddison 2018). agarose gels stained with Gel Red. Sanger sequenc- Only one representative of each haplotype for ing was carried out in both directions by GATC ND2/ND3 and COI was included in each model; Biotech (Cologne, Germany) using an ABI 3730xl a full list of the samples and their haplotypes is DNA analyser system. All sequences were submit- provided in Table S1. Using the Bayesian informa- ted to GenBank (Benson et al. 2013); the acces- tion criterion (BIC) (Jhwueng et al. 2014), imple- sion numbers of all sequences are provided in mented in the ‘Find best DNA model’ tool in Table S1. MEGA v.7.0 (Kumar et al. 2016), the optimal nucleotide substitution models for concatenated ND2/3 and COI were selected. This tool tests Taxon sampling iterations of 24 different substitution models In addition to our focal study populations in (covering all model types possible in MEGA and Southeast Sulawesi, sequence information for Todi- MrBayes) and provides BIC, Akaike information ramphus species and other comparison groups criterion corrected (AICC) and maximum likeli- were sourced from GenBank (Benson et al. 2013) hood (lnL) scores of the model ‘goodness of fit’.A to facilitate robust phylogenetic analyses general time reversible (GTR) model was selected (Table S1). ND2 and ND3 sequences were con- for concatenated ND2/ND3 and a Hasegawa- catenated and analysed separately to COI Kishino-Yano (HKY) model for COI, both with sequences, due to a much larger sample of Todi- five gamma categories (5Γ). Maximum likelihood ramphus ND2 and ND3 genes available on Gen- analysis was carried out in MEGA v.7.0 using Bank (Andersen et al. 2015b). Our ND2/ND3 these model types and run for 1000 bootstrap analysis considered 83 Todiramphus samples (29 replicates. Genes were partitioned by codon posi- produced by this study, 54 by Andersen et al. tion, to allow for different substitution rates 2015b), including all available Sacred Kingfisher between codons (Shapiro et al. 2006). Concate- (n = 39) and Beach Kingfisher (n = 4) samples. nated ND2/3 was partitioned globally across the For the Collared Kingfisher species complex we two genes following Andersen et al. (2013, focused on its central Indo-Pacific range, clade H 2015b). in Andersen et al. (2015b) and adjacent popula- Bayesian phylogenetic inference of haplotypes tions, including all samples of the Collared King- was carried out with MrBayes v.3.2.6 (Ronquist & fisher (n = 30), Torresian Kingfisher (n = 5) and Huelsenbeck 2003) using the same models and Islet Kingfisher (n = 5). All Todiramphus GenBank partition strategy as above. We used two

independent Markov chain Monte Carlo (MCMC) means were 0 and their variances were 1 (Thomas runs, with four chains per run. Convergence was et al. 2015). To test whether the different popula- assessed using Tracer v.1.6 (Rambaut et al. 2014), tions of kingfishers in southeast Sulawesi (mainland, with convergence in runs accepted when the aver- Wakatobi or Menui) differed morphologically, anal- age standard deviation in split frequencies ysis of variance (ANOVA) was carried out on (ASDSF) reached 0.01 (Ronquist et al. 2012) and principal components with eigenvalues > 1. the effective sample size (ESS) of model parameters exceeded 200 (Drummond et al. 2006). The Ethics statement GTR + 5Γ model of concatenated ND2/3 reached ASDSF of 0.01 and an ESS of > 200 for all model The necessary permits and approvals for this study parameters after 5 million generations. The were obtained from Kementerian Riset Teknologi HKY + 5Γ model for COI reached ASDSF of 0.01 Dan Pendidikan Tinggi (RISTEKDIKTI), permit and an ESS of > 200 for all model parameters after numbers: 278/SlP/FRP/SM/Vll/2012, 279/SIP/FRP/ 3.5 million generations. Both models were sampled SM/VIII/2012, 174/SIP/FRP/E5/Dit.KI/V/2016, every 1000 generations, with a burn-in of 25%. 159/SIP/FRP/E5/Fit.KIVII/2017 and 160/SIP/FRP/ Phylogenetic tree topology was taken from the E5/Fit.KIVII/2017. We obtained prior permission Bayesian phylogenetic inference and produced in from all landowners and no protected species FigTree v.1.4.2, with annotations added in were sampled. We are committed to reproducibil- Inkscape v.0.48.5. ity, and aliquots of the extracted DNA for all A TCS haplotype network of Todiramphus con- sampled individuals are available upon request catenated ND2/3 was constructed using POPART (subject to the Material Transfer policies of Trin- (Leigh & Bryant 2015). A TCS network is con- ity College Dublin, Halu Oleo University and structed using an agglomerative approach where RISTEKDIKTI). clusters are progressively combined with one or more connecting edges (Clement et al. 2002). RESULTS Pairwise comparisons were carried out in MEGA v.7.0 to calculate mean uncorrected pro- Phylogenetic analyses portional genetic distances (p-distances) within and between sampled populations for both longer Results from our maximum likelihood and Baye- ‘barcoding’ genes: ND2 and COI (where avail- sian analyses produced highly concordant topolo- able). The distances between Collared Kingfisher gies for both concatenated ND2/ND3 and COI populations were then compared with interspecific haplotypes. The concatenated ND2/ND3 tree was distances for Todiramphus species. the most informative because more comparative material was available on GenBank (Fig. 2). Our focal population of T. c. chloris lay with the other Morphological analyses central Indo-Pacific Collared Kingfisher popula- All morphological statistical analyses were carried tions, with a deep split separating them from their out in R software v.3.4.2 (R Development Core closest relatives, the Torresian Kingfisher and Islet Team 2017). Two types of morphological analy- Kingfisher in Australia/New Guinea. Within the ses were carried out: discriminant function analy- central Indo-Pacific, Collared Kingfisher popula- sis (DFA) and principal component analysis tions were broadly split into a Philippines/Palau/ (PCA). Borneo (T. c. collaris, T. c. teraokai and T. c. laub- DFA was conducted with package ‘MASS’ mannianus, respectively) population, T. c. humii in (Ripley et al. 2016) to investigate how well the Singapore and the T. c. chloris population on Sula- morphological data supported the groupings pro- wesi. Individuals from central Sulawesi (sampled vided by our molecular phylogenies. All available by Andersen et al. (2015b)) and from our study morphological measurements were used. population on mainland Southeast Sulawesi, and PCA was also carried out to investigate which its large continental islands, grouped together traits showed the greatest morphological variabil- (Fig. 2, Table S1). However, the Wakatobi popu- ity. As the morphological variables were on differ- lation was entirely distinct, sharing none of the ent scales, all were re-scaled for inclusion in the haplotypes present on mainland Sulawesi (Fig. 3). PCA using the scale function in R, such that their Our COI tree (Fig. 4) provided further evidence

Halcyon coromanda 1.0 Halcyon pileata 99 Halcyon symrnensis 1.0 Syma megarhyncha [PNG] 99 Syma torotoro [PNG] T. sordidus sordidus hapY03 [NT]

0.84 T. sordidus sordidus hapY02 [NT]

94 T. sordidus sordidus hapY04 [QLD] Australia 0.95 T. sordidus colcloughi hapY05 [QLD] 85 T. sordidus sordidus hapY01 [NT]

0.99 T. colonus hapZ02 [PNG]

96 T. colonus hapZ01 [PNG]

] T. sanctus sanctus & ] canacorum hapS01 Throughout ssp. range 1.0 T. sanctus sanctus hapS20 [PNG] 99 T. sanctus sanctus hapS19 [SE Sulawesi: Wakatobi Is.] T. sanctus sanctus hapS18 [SE Sulawesi: Menui] T. sanctus sanctus hapS10 [Solomon Is.] T. sanctus sanctus hapS17 [Solomon Is.] T. sanctus sanctus hapS14 [SE Sulawesi: Kabaena] 0.77 T. sanctus sanctus hapS12 [SE Sulawesi: Menui] 66 0.78 54 T. sanctus sanctus hapS11 [Solomon Is.] T. sanctus sanctus hapS09 [Vanuata] T. sanctus sanctus hapS08 [Australia: NT] 1.0 T. sanctus sanctus hapS07 [Australia: NT] 0.95 99 55 T. sanctus sanctus hapS06 [SE Sulawesi: Wakatobi Is.] T. sanctus sanctus hapS04 [PNG] T. sanctus sanctus hapS03 [SE Sulawesi: Kabaena]

T. sanctus sanctus hapS02 [PNG]

1.0 T. sanctus sanctus hapS16 [SE Sulawesi: Kabaena] ] 87 T. sanctus sanctus, vagans ] SE Sulawesi: Menui, New Zealand & canacorum hapS21 & New Caledonia T. sanctus sanctus hapS15 [Western Australia] 1.0 T. sanctus sanctus hapS13 [SE Sulawesi: Kabaena] 81 T. sanctus sanctus hapS05 [Western Australia] T. chloris teraokai hapC09 [Palau] T. chloris collaris hapC05 [Philippines] T. chloris collaris hapC07 [Philippines]

1.0

99 T. chloris teraokai hapC08 [Palau] ] T. chloris collaris & Philippines & ] laubmannianus hapC04 Malaysia: Borneo 1.0 86 T. chloris collaris hapC06 [Philippines] T. chloris humii hapC01

1.0 T. chloris humii hapC03 Singapore 97 0.98 T. chloris humii hapC02 67 T. chloris chloris hapC14 0.85 SE Sulawesi: 64 T. chloris chloris hapC15 Wakatobi Is. 1.0 T. chloris chloris hapC13

0.58 99 T. chloris chloris hapC11 1.0 99 T. chloris chloris hapC12 Sulawesi: 99 mainland T. chloris chloris hapC10

T. saurophagus saurophagus hapX01 [PNG] 1.0 T. saurophagus saurophagus hapX02 [Solomon Is.] 99 T. saurophagus saurophagus hapX03 [Solomon Is.]

Figure 2. Bayesian consensus tree for concatenated ND2 and ND3 haplotypes, showing Bayesian posterior probabilities (above) and bootstrap values from our maximum likelihood analysis (below) for each node. Populations are listed as their currently described subspecies (Gill & Donsker 2018), the haplotype they represent and the geographical range for each haplotype. Square brackets are used for geographical range when referring to a single node. A curly bracket is used for geographical range when referring to more than one node. Where a haplotype is shared between subspecies this is noted. Branch lengths for Halcyon were reduced to save space. PNG, Papua New Guinea; SE Sulawesi, Southeast Sulawesi; Is., Islands; NT, Northern Territory; QLD, Queensland. of the separation of Wakatobi and Sulawesi main- No clear phylogeographical pattern was seen in land Collared Kingfisher populations, and the the Sacred Kingfisher (Figs 2 & 4). Some structure strong separation between Sulawesi and Philippine was evident, with haplotypes S05, S13, S15 and populations of Collared Kingfisher. S21 forming a discrete clade for concatenated

Figure 3. Haplotype network of Todiramphus population samples, based on concatenated ND2/3 sequences. One bar indicates one mutation, black nodes are hypothetical ancestral states, and the size of the circles corresponds to the number of sampled individuals sharing that haplotype. Individual haplotype details are given in Table S1.

ND2/ND3, and haplotype S06 for COI. However, between the T. c. chloris mainland Sulawesi popu- the separation was weak, with only one to three lation and the Sacred Kingfisher (ND2: 1.3%) and mutations separating even the most divergent hap- Beach Kingfisher (ND2: 1.4%) provide an indica- lotypes from other haplotypes (Fig. 3). No obvious tion of the levels of divergence found between geographical splits were apparent, except perhaps Todiramphus species which have been established distinctiveness in the Western Australian breeding as taxonomically distinct. Comparisons between population. the Collared Kingfisher populations sampled showed that T. c. chloris on mainland Sulawesi has diverged from both T. c. humii in Singapore Genetic distance (ND2: 1.3% divergence) and the subspecies cov- Collared Kingfisher ering the Philippines, Palau and Borneo (T. c. col- Calculations of pairwise genetic distance provided laris, T. c. teraokai and T. c. laubmannianus) an indication of the level of divergence between (ND2: 1.3%). The Philippines/Palau/Borneo pop- the populations described in our phylogenetic ulation shows minimal divergence within this trees (Tables S3 and S4). Divergence values clade (ND2: 0.0004%). The Philippines/Palau/

Halcyon coromanda

Halcyon pileata 1.0 92 Halcyon symrnensis

Actenoides lindsayi lindsayi hapAL1 1.0 Philippines 100 Actenoides lindsayi lindsayi hapAL2

Dacelo novaeguineae 0.93 [Australia] 59 T. macleayii incintus hapM01 0.99 Australia: 85 QLD T. macleayii incintus hapM02 0.82 51 T. chloris collaris hapC01

1.0 Philippines

T. chloris collaris hapC02

] ] Australia (NSW) T. sanctus sanctus hapS01 and throughout SE Sulawesi 0.68 T. sanctus sanctus hapS02 70 [SE Sulawesi: Wakatobi Is.] 0.95 T. sanctus sanctus hapS03 84 [SE Sulawesi: Wakatobi Is.] T. sanctus sanctus hapS04 0.95 [SE Sulawesi: Menui] 71 1.0 T. sanctus sanctus hapS05 94 [SE Sulawesi mainland] T. sanctus sanctus hapS06 [SE Sulawesi: Wakatobi Is.]

T. gambieri gertrudae [Tuomatu Is.]

0.69 T. chloris chloris hapC04 49 0.86 SE Sulawesi: 71 Wakatobi Is. 1.0 T. chloris chloris hapC05 93 T. chloris chloris hapC03 [SE Sulawesi mainland]

Figure 4. Bayesian consensus tree for COI haplotypes, showing Bayesian posterior probabilities (above) and bootstrap values from our maximum likelihood analysis (below) for each node. Populations are listed as their currently described subspecies (Gill & Dons- ker 2018), the haplotype they represent and the geographical range for each haplotype. Square brackets are used for geographical range when referring to a single node. A curly bracket is used for geographical range when referring to more than one node. Branch lengths for Halcyon were reduced to save space. SE Sulawesi, Southeast Sulawesi; Is., Islands; NSW, New South Wales; QLD, Queensland.

Borneo population and T. c. humii in Singapore COI was not available for all populations but differ less from each other (ND2: 0.9%), than followed the same pattern as ND2 where available either differs from T. c. chloris. Within the focal (Table S4). region of Southeast Sulawesi, divergence between the Wakatobi and Sulawesi mainland populations Morphological analyses of T. c. chloris is smaller (ND2: 0.4%), although this ‘across group’ value is much greater than A summary of the raw morphological measure- those of within-group divergence (Wakatobi ments collected for this analysis is given in ND2: 0.0007%; mainland Sulawesi ND2: Table 1. 0.0003%). DFA incorporating wing, bill, skull and mass measurements proved 100% successful at catego- Sacred Kingfisher rizing the Southeast Sulawesi Todiramphus king- Genetic divergence within the Sacred Kingfisher fishers into the taxonomic groupings suggested by was minimal. The within-group divergence for all our molecular phylogenies – ‘mainland’ Collared samples (and all three subspecies sampled) was Kingfishers, ‘Wakatobi’ Collared Kingfishers and only 0.0001% for ND2. Sacred Kingfishers (Table S5).

Table 1. Morphological measurements, showing mean values Æ standard error.

Collared Kingﬁsher Sacred Kingﬁsher

Variable Mainland (n = 15) Wakatobi (n = 10) Mainland (n = 7) Wakatobi (n = 8) Menui (n = 3)

Wing length (mm) 102.2 Æ 0.6 113.4 Æ 0.9 89.5 Æ 1.2 92.0 Æ 1.6 89.2 Æ 1.8 Bill length (mm) 47.2 Æ 0.6 54.0 Æ 0.8 42.2 Æ 0.9 44.6 Æ 0.5 43.2 Æ 1.2 Skull length (mm) 27.3 Æ 0.3 28.0 Æ 0.5 24.3 Æ 0.4 25.2 Æ 0.2 24.3 Æ 0.7 Mass (g) 60.6 Æ 1.3 71.2 Æ 2.1 39.1 Æ 1.3 40.9 Æ 1.4 40.3 Æ 2.2

The first two principal components explained Wakatobi Collared Kingfishers were significantly the majority of the variation seen in both species larger than those from the mainland (PC1, (Table 2) and were the components taken forward ANOVA: F1,23 = 82.91, P < 0.001) (Fig. 5, for further analysis: PC1 gave an overall indication Table 2). The bill to skull ratio (PC2) of Collared of body size and PC2 was most strongly influenced Kingfisher populations was not found to differ. by bill length and skull length, giving an indication There was no difference in morphology between of the bill to skull ratio. Sacred Kingfisher populations (mainland, Waka- tobi and Menui) in either PC1 or PC2.

Table 2. Summary of the loading of the different variables in DISCUSSION ﬁ the rst two PCs of the PCA and the proportion of the variance Our results demonstrated a clear split in Collared explained by these PCs. Kingﬁshers between the Southeast Sulawesi mainland population and the Wakatobi Islands popula- Variable PC1 PC2 tion in both genetics and morphology. Genetically, Wing length À0.534 0.071 the mainland and Wakatobi populations were Bill length À0.485 0.608 reciprocally monophyletic with respect to both Skull length À0.454 À0.790 Mass À0.523 0.049 concatenated ND2/ND3 and COI sequences. The Proportion of variance 84.6% 11.5% mainland Southeast Sulawesi population (this study) grouped with the central Sulawesi

Collared Kingfisher − Mainland 2.0 ● ● Collared Kingfisher − Wakatobi ● Sacred Kingfisher

1.5 ●

1.0 ●

● ● 0.5 ● ● ● ● ● ● ● ● ● ● ● ● ● 0.0 ● ● ● ● ● ●

Bill to skull ratio (PC2) Bill to skull ratio ● −0.5 ●

● ● −1.0

−1.5

−4 −2 0 2 4 Body size (PC1)

Figure 5. Scatterplot of the first two principal components of kingfisher morphology for the Sacred Kingfisher, the Collared Kingfisher mainland population and the Collared Kingfisher Wakatobi Islands population. PC1 reflects body size and PC2 reflects bill/skull ratio. Note: negative values denote larger size in morphological traits for PC1.

population (Andersen et al. 2015b), while all the T. s. sanctus subspecies (Western, Eastern and Wakatobi birds represented an independent evolu- Northern Australia), in addition to haplotypes tionary lineage. The Wakatobi birds are signifi- which match the supposedly sedentary T. s. cana- cantly larger than those from the Sulawesi corum and T. s. vagans. Additionally, Sacred King- mainland, suggesting ecological adaptation to the fishers from Southeast Sulawesi showed no local conditions on these small islands. Although a difference in morphology between Sulawesi, number of vouchered specimens of Collared King- Wakatobi and Menui birds, suggesting either that fisher (then named Halcyon chloris) were collected these birds were from the same breeding popula- by Kuhn€ in the early 20th century (Table S6) tion or that the Sacred Kingfisher shows remark- (Hartert 1903), morphological differences in the able morphological uniformity across its range. Wakatobi population have not been previously It is possible that the radiation of Sacred King- described. fisher populations is so recent that genetic differen- In addition, our results bring further resolution to tiation is still minimal, or that there is continuing the Collared Kingfisher populations of the central gene flow between these populations facilitated by Indo-Pacific, giving deeper insight into the Sulawesi the migratory nature of many Sacred Kingfisher populations and corroborating the findings of populations (Andersen et al. 2015b). The lack of Andersen et al. (2015b). Our analyses confirm that diversification in such a widespread bird seems the central Indo-Pacific Collared Kingfisher popula- remarkable, particularly considering the level of tions separate into three distinct groups: (1) diversification in the Collared Kingfisher species T. c. chloris in Sulawesi (with a more recent separa- complex (Andersen et al. 2015b) and several other tion between mainland Sulawesi and the Wakatobi vagile taxa (Cibois et al. 2014, Andersen et al. Islands), (2) T. c. humii in Singapore and (3) 2015a, Garcia-R et al. 2017) over the same area as T. c. collaris in the Philippines, T. c. teraokai in the Sacred Kingfisher’s range. However, these find- Palau and T. c. laubmannianus in Borneo. As ings for the Sacred Kingfisher tally with those of outlined in Andersen et al. (2015b) there was Pedersen et al. (2018), who found that populations almost no divergence between the T. c. collaris, of the Common Cicadabird species complex T. c. teraokai and T. c. laubmannianus subspecies in (E. t. pellingi and E. t. obiense) remained morpho- the Philippines, Palau and Borneo. By contrast, the logically and genetically similar, despite being sepa- Sulawesi T. c. chloris population was clearly distinct rated by 500 km of deep ocean. One population of (Figs 2–4). The distinctiveness of the T. c. chloris the Common Cicadabird (E. t. tenuirostre) also population on Sulawesi could be explained by the migrates between Australia and New Guinea (Tay- isolating effect of the deep-water trench that under- lor & Kirwan 2018). The migratory nature and lies Wallace’s line. This trench ensured Sulawesi population connectedness over large water barriers was isolated by a water barrier during the last glacial seen in the Sacred Kingfisher, and populations of maximum, when the islands to the west of Sulawesi the Common Cicadabird species complex (this were connected to mainland Southeast Asia (Essel- complex also includes distinct isolated popula- styn et al. 2010), and separates T. c. chloris from all tions), may indicate that these populations are still other Collared Kingfisher subspecies (Fig. 1). in an early ‘colonization phase’ (stage 1) of their Unlike the Collared Kingfisher, the migratory taxon cycle (Mayr & Diamond 2001). The taxon Sacred Kingfisher shows no consistent phylogenetic cycle concept (TCC) was developed by Wilson structure, mirroring the results of Andersen et al. (1961) for Melanesian ants but was applied to birds (2015b). Our phylogenetic work documented 21 by Ricklefs and Cox (1972) and Diamond et al. haplotypes (concatenated ND2/ND3) within the (1976). The TCC describes how taxa go through a Sacred Kingfisher, but there was little differentia- cycle of decreasing vagility as the taxon ages, start- tion between these haplotypes (Fig. 3). Both of ing with a colonization phase and ending with the haplotypes found in sedentary populations endemism. It helps to explain the paradox of the (hapS01 and hapS21) were also found in migra- great speciators, how species capable of such wide tory populations. Haplotype S01 was the most differentiation stop spreading and begin to diverge common haplotype sampled, representing 14/39 in isolation (Diamond et al. 1976). The Sacred individuals (Fig. 3, Table S1). Within Southeast Kingfisher’s preference for coastal edge habitats Sulawesi we found representative haplotypes from (Woodall & Kirwan 2018b) matches habitats the full breeding range of the migratory expected of a species early in its taxon cycle

(Mayr & Diamond 2001). These results further Kingfisher (n = 11) and the Sacred Kingfisher highlight the need for wider assessment of Sacred (n = 15) were exclusively caught in coastal habi- Kingfisher populations in both their breeding and tats, particularly mangrove. their wintering range, to understand the relation- Although the inland habitats typical for Col- ships between the different populations. lared Kingfishers are present on the Wakatobi A number of factors may explain the divergence Islands, the small land area of each island means seen in the Wakatobi Collared Kingfisher popula- that habitat diversity is low, with a predominance tion. We cannot rule out pure genetic drift (Rune- of coastal scrub. It is probable that the marine mark et al. 2010) or a founder effect (Spurgin edge influence dominates the whole of these et al. 2014) as origins for the changes seen in the islands. This is reflected in the depauperate fauna Wakatobi Collared Kingfishers. However, the dis- on the Wakatobi Islands in comparison with the tance between the Wakatobi Islands and mainland larger islands in Southeast Sulawesi (Kelly & Mar- is relatively short (27 km) (Kelly et al. 2014) and ples 2010, Martin et al. 2012, O’Connell et al. Collared Kingfishers are highly vagile (Woodall 2017), as species which depend on the greater 2018b). Therefore it appears unlikely that the habitat richness of larger islands cannot persist Wakatobi population is sufficiently geographically there (Pimm et al. 1988). To persist on the Waka- isolated for drift or a founder effect alone to be tobi Islands, the Collared Kingfisher population viable explanations (but see Mayr 1942, Diamond may have had to adapt to a more general habitat 1998, Leisler & Schulze-Hagen 2011 and Ander- niche. The decrease in land area the Wakatobi sen et al. 2015a for examples of species from other Islands would have experienced due to sea level bird groups with excellent dispersal abilities (Zos- rises in the late Quaternary may have accelerated teropidae, Acrocephalidae and Monarchidae, this process (Weigelt et al. 2016). Larger body size respectively), isolated by smaller water barriers). in island birds (particularly bill size) has been asso- Ecologically divergent habitats (such as mainland ciated with a more generalist niche in several Sulawesi and the Wakatobi Islands) can create bar- island bird groups (Grant 1965, Scott et al. 2003, riers to gene flow (Orsini et al. 2013). Under these Leisler & Schulze-Hagen 2011). Leisler and Win- conditions, neutral loci (such as the mitochondrial kler (2015) found longer bills in several Pacific genes used in this study) can diverge by genetic island warblers; these allow birds to handle a drift, even between populations which are not sep- greater range of prey sizes (Herrera 1978). The arated by strong geographical boundaries. This increase in body and bill size seen in the Wakatobi process is known as isolation by adaptation (IBA; Collared Kingfisher (Table 1) may reflect such a Nosil et al. 2009, Orsini et al. 2013). shift to a more generalist niche. The Wakatobi Islands present a very different More thorough surveys of Todiramphus species habitat to the ecologically complex mainland. The density at both mainland and Wakatobi sites would Wakatobi Islands are uplifted coral islands that sit be needed to confirm this proposed expansion of atop a platform of Australian origin and have habitat use. However, our observations tally with never been connected to mainland Sulawesi (Mil- the habitat descriptions given for these species by som et al. 1999). This seems to have resulted in a White and Bruce (1986), stating that the Collared change in habitat use by Collared Kingfishers Kingfisher is most common in gardens, plantations between the mainland and the Wakatobi Islands. and open wooded country in Sulawesi, Buton and Our observations indicated the Collared Kingfisher Muna, and is much less associated with mangroves partitioned habitat with the Sacred Kingfisher on in Wallacea than elsewhere in its range, whereas the mainland; the Collared Kingfisher was found the Sacred Kingfisher is mostly associated with inland in scrub, farmland and plantations, and the mangroves. The habitat on the Wakatobi Islands Sacred Kingfisher occupied a thin coastal strip, may more closely mirror that seen on small islands always adjacent to water and mostly in mangrove such as Palau, where the Collared Kingfisher is also habitats. Of the Todiramphus kingfishers mist- associated with mangroves (Woodall 2018b). netted at mainland sites during this research, Therefore the different selection pressures experi- 15/16 Collared Kingfishers were mist-netted inland enced by mainland Sulawesi and the Wakatobi (one in coastal mangrove) and 7/7 Sacred Kingfish- Islands may be promoting reproductive isolation ers were netted in coastal mangrove. In contrast, between Collared Kingfisher populations on the on the Wakatobi Islands, both the Collared mainland and on the Wakatobi Islands.

Although a difference in habitat may be enough Mayr and Diamond (2001) also note that the to account for the morphological divergence seen Melanesian Kingfisher and Sacred Kingfisher parti- in the Collared Kingfisher Wakatobi population, a tion resources ecologically, in Melanesia, by body lack of habitat partitioning between Collared and size and the Sacred Kingfisher’s preference for Sacred Kingfishers on the Wakatobi Islands may more open habitats. Diamond and Marshall bring these two species into closer competition. (1977) found that on Santo and Malakula Islands, Such ecologically similar species are likely to have the PacificKingfisher partitions habitats with the to partition resources to coexist (Schoener 1974, Vanuatu Kingfisher, being confined to coastal Jonsson et al. 2008, Robertson et al. 2014). areas, clearings and open spaces, whereas the Van- Although the Collared Kingfisher is larger than the uatu Kingfisher is found in closed forest; however, Sacred Kingfisher on mainland Sulawesi, a small whereas the Vanuatu Kingfisher is absent on Efate, difference in body size between competing species Erromanga and Tanna Islands, the Pacific King- does not guarantee complete competitive domi- fisher is found throughout those habitats. How- nance, perhaps necessitating the habitat partition- ever, this pattern does not always hold. Andersen ing between these species seen on the mainland et al. (2017) found these two species in sympatry, (Reif et al. 2018). As outlined by Grant (1968), inland on Malakula on the edge of primary forest, competing species on islands must segregate by: in disturbed habitats near human habitation. (1) habitat, (2) habitat use, (3) food size and (4) Potentially the Pacific Kingfisher’s tolerance for food type. Sympatric kingfisher species partition disturbed areas has allowed it to move further foraging niche by body size (Kasahara & Katoh inland. Clearly the presence/absence of competi- 2008, Borah et al. 2012), as larger kingfishers are tors is of great ecological importance within this able to take larger prey. The Todiramphus kingfish- species complex and differences in habitat struc- ers on the Wakatobi Islands are not experiencing ture can influence these interactions. competitive exclusion, so the accentuated differ- Similar patterns are seen in other widely dis- ence in body size between these species may allow tributed, island-colonizing taxa, as multiple colo- them to partition their niches by food size. This nization events force congeneric species to means the Wakatobi population of the Collared segregate niches or face competitive exclusion Kingfisher may have experienced ecological char- (Mayr & Diamond 2001). Andersen et al. (2013) acter displacement, with selection favouring indi- noted that Ceyx kingfishers segregate by habitat viduals which differed most from the Sacred when in sympatry in the Philippines. Lack (1971) Kingfisher (Brown & Wilson 1956, Dayan & Sim- noted that two species of white-eye are seldom berloff 2005, Stuart & Losos 2013). found in sympatry on the mainland, generally par- In Southeast Sulawesi the only kingfisher titioning by altitude or by habitat choice. How- species which is both larger than the Collared ever, sympatric pairs of white-eyes are found on Kingfisher and found in coastal habitats is the multiple Indian Ocean Islands, where they parti- Great-billed Kingfisher Pelargopsis melanorhyncha. tion morphologically, with one species larger than The Great-billed Kingfisher is absent from the the other (Gill 1971, Lack 1971, Warren et al. Wakatobi Islands (Woodall & Kirwan 2018a), so 2006). Wijesundara and Freed (2018) found that those kingfishers present experience no competi- the endemic Sri Lankan White-eye Zosterops ceylo- tion for larger prey. Members of the Collared nensis increased in bill and body size when in sym- Kingfisher species complex show remarkable niche patry with the widespread Oriental White-eye flexibility elsewhere in their range, depending on Zosterops palpebrosus. Our results suggest a similar which competitors are present. The Collared King- process may occur in sympatric pairs of Todiram- fisher is restricted to the coast when in sympatry phus kingfisher. with the Talaud Kingfisher or Rusty-capped King- Two main factors may explain why the Sacred fisher (Eaton et al. 2016, del Hoyo et al. 2018, Kingfisher has not experienced selection pressures Woodall 2018c) on the small islands of Talaud in a similar way to the Collared Kingfisher in and Palau, respectively. The Melanesian Kingfisher Southeast Sulawesi: its consistent habitat choice (a member of the Collared Kingfisher species com- and its migratory lifestyle. The Sacred Kingfisher is plex) segregates habitat with the Beach Kingfisher, found in mangrove habitat throughout the region, being restricted to inland areas where the Beach regardless of island size (but with the addition of Kingfisher is present (Mayr & Diamond 2001). the Collared Kingfisher as a potential competitor

on smaller islands). Although wintering grounds threatens Southeast Asia (Sodhi et al. 2004, Wil- have strong carry-over effects on breeding success cove et al. 2013). Much biodiversity, and the evo- (Bearhop et al. 2004, Latta et al. 2016, Rockwell lutionary lessons it can teach us, faces extinction et al. 2017), and Sacred Kingfisher individuals are before being formally recognized. known to be faithful to the same wintering site (Woodall & Kirwan 2018b), the ability of the We thank Kementerian Riset Teknologi Dan Pendidikan Sacred Kingfisher to adapt to its wintering grounds Tinggi (RISTEKDIKTI) for providing the necessary per- will be limited by the constraints placed upon the mits and approvals for this study. We also thank Opera- tion Wallacea for providing us with invaluable logistical population by the demands of its breeding grounds support while working in Indonesia. This research was and its migration route. In contrast, the resident supported by the Irish Research Council. We thank Collared Kingfisher needs only adapt to its local Rebecca Kimball, Gary Voelker, Michael Andersen and conditions. three anonymous reviewers for their thoughtful com- Our genetic and morphological findings suggest ments, which greatly improved the manuscript. that the Wakatobi Collared Kingfisher may represent an undescribed subspecies and we recom- AUTHOR CONTRIBUTIONS mend investigations of plumage (facilitated by the previously collected type specimens, Table S6) D.J.K., N.M.M., K.A., D.O.C. and A.K. conceived and song differences to confirm this (Uy et al. this study and carried out fieldwork. D.O.C. and 2009). This study relied on mitochondrial DNA to N.L. carried out the lab work. D.O.C. conducted separate populations of Todiramphus kingfishers. the analyses and led the writing. All authors con- Although mitochondrial DNA barcoding tributed to revising and improving the manuscript. approaches have proved successful in species delimitation (Hebert et al. 2004, 2016, Kerr et al. 2007), mitochondrial and nuclear DNA have been REFERENCES shown to infer different evolutionary histories in Andersen, M.J., Oliveros, C.H., Filardi, C.E. & Moyle, R.G. many cases (Rubinoff & Holland 2005, Phillimore 2013. Phylogeography of the variable dwarf-kingfisher Ceyx et al. 2008). Introgression of mitochondrial DNA lepidus (Aves: Alcedinidae) inferred from mitochondrial and nuclear DNA sequences. Auk 130: 118–131. and sex-biased asymmetries can affect phylogenies that depend solely on mitochondrial DNA, partic- Andersen, M.J., Nyari, A.S., Mason, I., Joseph, L., Dumbacher, J.P., Filardi, C.E. & Moyle, R.G. 2014. ularly when attempting to infer older evolutionary Molecular systematics of the world’s most polytypic bird: the relationships (Toews & Brelsford 2012). However, Pachycephala pectoralis/melanura (Aves: Pachycephalidae) Andersen et al. (2015b) found ND2 and ND3 to species complex. Zool. J. Linnean Soc. 170: 566–588. provide the best phylogenetic resolution for Todi- Andersen, M.J., Hosner, P.A., Filardi, C.E. & Moyle, R.G. ramphus kingfishers, with nuclear genes largely 2015a. Phylogeny of the Monarch Flycatchers reveals extensive paraphyly and novel relationships within a major uninformative due to the shallow scale of the Todi- Australo-Pacific radiation. Mol. Phylogenet. Evol. 83: 118– ramphus diversification. 136. This research illustrates a possible mode of Andersen, M.J., Shult, H.T., Cibois, A., Thibault, J.-C., divergence at the early stages of speciation in an Filardi, C.E. & Moyle, R.G. 2015b. Rapid diversification and fi fi isolated Todiramphus population. Our results also secondary sympatry in Australo-Paci c king shers (Aves: Alcedinidae: Todiramphus). R. Soc. Open Sci. 2: 140375. highlight how much is still to be resolved about Andersen, M.J., Fatdal, L., Mauck, W.M., III & Smith, B.T. the relationships between Sacred Kingfisher popu- 2017. An ornithological survey of Vanuatu on the islands of lations and their distribution on their wintering Efat e, Malakula, Gaua, and Vanua Lava. Check List 13: grounds. Assessing the populations of ‘great specia- 755–782. tor’ lineages such as the Collared Kingfisher pro- Andersen, M.J., McCullough, J.M., Mauck, W.M., Smith, B.T. & Moyle, R.G. 2018. A phylogeny of kingfishers vides the opportunity to contribute both reveals an Indomalayan origin and elevated rates of taxonomic revision and insight into the early stages diversification on oceanic islands. J. Biogeogr. 45: 269– of speciation. Their rapid speed of evolutionary 281. change, ability to colonize islands and the fre- van Balen, S. 2008. Family Zosteropidae (white-eyes). In Del Hoyo, J., Elliot, A. & Christie, D.A. (eds) Handbook of the quency with which they are found in secondary – ‘ ’ Birds of the World, Vol. 13: 402 485. Barcelona: Lynx sympatry makes great speciators ideal groups in Edicions. which to study speciation. This work is given Bearhop, S., Hilton, G.M., Votier, S.C. & Waldron, S. 2004. impetus by the looming biodiversity crisis that Stable isotope ratios indicate that body condition in

migrating passerines is influenced by winter habitat. Proc. R. Diamond, J.M., Gilpin, M.E. & Mayr, E. 1976. Species- Soc. Lond. B 271: S215–S218. distance relation for birds of the Solomon Archipelago, and Benson, D.A., Cavanaugh, M., Clark, K., Karsch-Mizrachi, the paradox of the great speciators. Proc. Natl Acad. Sci. I., Lipman, D.J., Ostell, J. & Sayers, E.W. 2013. GenBank. USA 73: 2160–2164. Nucleic Acids Res. 41: D36–D42. Dong, L., Wei, M., Alstrom,€ P., Huang, X., Olsson, U., Bickford, D., Lohman, D.J., Sodhi, N.S., Ng, P.K.L., Meier, Shigeta, Y., Zhang, Y. & Zheng, G. 2015. Taxonomy of the R., Winker, K., Ingram, K.K. & Das, I. 2007. Cryptic Narcissus Flycatcher Ficedula narcissina complex: an species as a window on diversity and conservation. Trends integrative approach using morphological, bioacoustic and Ecol. Evol. 22: 148–155. multilocus DNA data. Ibis 157: 312–325. Bijnens, L., Stevens, J., Janssens, L. & Louette, M. 1987. Drummond, A.J., Ho, S.Y., Phillips, M.J. & Rambaut, A. Community structure of Grand Comoro land birds with 2006. Relaxed phylogenetics and dating with confidence. special reference to the ecology of congeneric species. Rev. PLoS Biol. 4: e88. Zool. Afr. 101: 221–232. Eaton, J.A., van Balen, B., Brickle, N.W. & Rheindt, F.E. Borah, J., Ghosh, M., Harihar, A., Pandav, B. & Gopi, G. 2016. Birds of the Indonesian Archipelago: Greater Sundas 2012. Food-niche partitioning among sympatric kingfishers and Wallacea. Barcelona: Lynx Edicions. in Bhitarkanika mangroves, Odisha. J. Bombay Nat. Hist. Elbourne, R. 2011. COI barcoding of the shorebirds: Rates of Soc. 109:72–77. evolution and the identification of species. MSc. Thesis, Brown, W.L., Jr & Wilson, E.O. 1956. Character University of Toronto, Department of Ecology and displacement. Syst. Biol. 5:49–64. Evolutionary Biology. Carstensen, D.W., Dalsgaard, B., Svenning, J.C., Rahbek, Esselstyn, J.A., Oliveros, C.H., Moyle, R.G., Peterson, A.T., C., Fjeldsa, J., Sutherland, W.J. & Olesen, J.M. 2012. McGuire, J.A. & Brown, R.M. 2010. Integrating Biogeographical modules and island roles: a comparison of phylogenetic and taxonomic evidence illuminates complex Wallacea and the West Indies. J. Biogeogr. 39: 739–749. biogeographic patterns along Huxley’s modification of Cheke, A.S. 1987. The ecology of the smaller land-birds of Wallace’s line. J. Biogeogr. 37: 2054–2066. Mauritius. In Diamond, A.W. (ed.) Studies of Mascarene Franzen, M., Schweiger, O. & Betzholtz, P.E. 2012. Species- Island Birds: 151–207. Cambridge: Cambridge University area relationships are controlled by species traits. PLoS Press. ONE 7: e37359. Chesser, R.T. 1999. Molecular systematics of the rhinocryptid Fry, C.H. 1980. The evolutionary biology of kingfishers genus Pteroptochus. Condor 101: 439–446. (Alcedinidae). Living Bird 18: 1980. Cibois, A., Thibault, J.C. & Pasquet, E. 2007. Uniform Garcia-R, J.C., Joseph, L., Adcock, G., Reid, J. & phenotype conceals double colonization by reed-warblers of Trewick, S.A. 2017. Interisland gene flow among a remote Pacific archipelago. J. Biogeogr. 34: 1150–1166. populations of the buff-banded rail (Aves: Rallidae) and its Cibois, A., Beadell, J.S., Graves, G.R., Pasquet, E., Slikas, implications for insular endemism in Oceania. J. Avian Biol. B., Sonsthagen, S.A., Thibault, J.-C. & Fleischer, R.C. 48: 679–690. 2011. Charting the course of reed-warblers across the Gill, F.B. 1971. Ecology and evolution of the sympatric Pacific islands. J. Biogeogr. 38: 1963–1975. Mascarene white-eyes, Zosterops borbonica and Zosterops Cibois, A., Thibault, J.-C., Bonillo, C., Filardi, C.E., Watling, olivacea. Auk 88:35–60. D. & Pasquet, E. 2014. Phylogeny and biogeography of the Gill, F. & Donsker, D.E. 2018. IOC World Bird List (v8.1). fruit doves (Aves: Columbidae). Mol. Phylogenet. Evol. 70: Available at: https://www.worldbirdnames.org/ioc-lists/maste 442–453. r-list-2/ (accessed 6 April 2018). Clement, M., Snell, Q., Walke, P., Posada, D. & Crandall, K. Grant, P.R. 1965. The adaptive significance of some size 2002. TCS: estimating gene genealogies. Proc. 16th Int trends in island birds. Evolution 19: 355–367. Parallel Distrib. Process Symp. 2: 184. Grant, P.R. 1968. Bill size, body size, and the ecological Coates, B.J. & Bishop, K.D. 1997. A Guide to the Birds of adaptations of bird species to competitive situations on Wallacea, Sulawesi, The Moluccas and Lesser Sunda islands. Syst. Zool. 17: 319–333. Islands, Indonesia. Alderley: Dove Publications. Graves, G.R. 1992. The endemic land birds of Henderson Craig, R.J. 1992. Territoriality, habitat use and ecological Island, southeastern Polynesia. Notes on natural history and distinctness of an endangered Pacific island reed-warbler. J. conservation. Wilson J. Ornithol. 104:32–43. Field Ornithol. 63: 436–444. Hall, T.A. 1999. BioEdit: a user-friendly biological sequence Darwin, C. 1859. On the Origins of Species by Means of alignment editor and analysis program for Windows 95/98/ Natural Selection. London: Murray. NT. Nucl. Acids Symp. Ser. 41:95–98. Dayan, T. & Simberloff, D. 2005. Ecological and community- Hartert, E. 1903. On the birds collected on the Tukang-Besi wide character displacement: the next generation. Ecol. Lett. Islands and Buton, south-east of Celebes, by Mr Heinrich 8: 875–894. Kuhn.€ Novit. Zool. 10:18–38. Diamond, J.M. 1974. Colonization of exploded volcanic Hebert, P.D.N., Stoeckle, M.Y., Zemlak, T.S. & Francis, islands by birds: the supertramp strategy. Science 184: 803– C.M. 2004. Identification of birds through DNA barcodes. 806. PLoS Biol. 2: e312. Diamond, J.M. 1998. Geographic variation in vocalisations of Hebert, P.D.N., Ratnasingham, S., Zakharov, E.V., Telfer, the white-eye superspecies Zosterops [griseotinctus] in the A.C., Levesque-Beaudin, V., Milton, M.A., Pedersen, S., New Georgia Group. Emu 98:70–74. Jannetta, P. & deWaard, J.R. 2016. Counting animal Diamond, J.M. & Marshall, A.G. 1977. Niche shifts in New species with DNA barcodes: Canadian insects. Philos. Hebridean birds. Emu 77:61–72. Trans. R. Soc. B 371: 20150333.

Herrera, C.M. 1978. Individual dietary differences associated Leisler, B. & Winkler, H. 2015. Evolution of island warblers: with morphological variation in Robins Erithacus rubecula. beyond bills and masses. J. Avian Biol. 46: 236–244. Ibis 120: 542–545. Liu, Y., Chen, G., Huang, Q., Jia, C., Carey, G., Leader, P., Higgins, P.J. (ed.) 1999. Handbook of Australian, New Li, Y., Zou, F., Yang, Z., Olsson, U. & Alstrom,€ P. 2016. Zealand & Antarctic birds. Volume 4, Parrots to Dollarbird. Species delimitation of the White-tailed Rubythroat Calliope Melbourne, Vic.: Oxford University Press. pectoralis complex (Aves, Muscicapidae) using an del Hoyo, J., Collar, N. & Kirwan, G.M. 2018. Palau integrative taxonomic approach. J. Avian Biol. 47: 899–910. Kingfisher (Todiramphus pelewensis). In del Hoyo, J., Elliott, Lohman, D.J., Ingram, K.K., Prawiradilaga, D.M., Winker, A., Sargatal, J., Christie, D.A. & de Juana, E. (eds) K., Sheldon, F.H., Moyle, R.G., Ng, P.K.L., Ong, P.S., Handbook of the Birds of the World Alive. Barcelona: Lynx Wang, L.K. & Braile, T.M. 2010. Cryptic genetic diversity in Edicions. Available at: https://www.hbw.com/node/467404 ‘widespread’ Southeast Asian bird species suggests that (accessed 15 August 2018). Philippine avian endemism is gravely underestimated. Biol. Irestedt, M., Fabre, P.-H., Batalha-Filho, H., Jønsson, K.A., Conserv. 143: 1885–1890. Roselaar, C.S., Sangster, G. & Ericson, P.G. 2013. The Lovette, I.J. & Hochachka, W.M. 2006. Simultaneous effects spatio-temporal colonization and diversification across the of phylogenetic niche conservatism and competition on Indo-Pacificbya‘great speciator’(Aves, Erythropitta avian community structure. Ecology 87: S14–S28. erythrogaster). Proc. R. Soc. Lond. B 280: 20130309. MacArthur, R. & Levins, R. 1967. The limiting similarity, Jablonski, D. & Finarelli, J.A. 2009. Congruence of convergence, and divergence of coexisting species. Am. morphologically-defined genera with molecular phylogenies. Nat. 101: 377–385. Proc. Natl Acad. Sci. USA 106: 8262–8266. MacArthur, R.H. & Wilson, E.O. 1967. The Theory of Island Jetz, W., Thomas, G.H., Joy, J.B., Hartmann, K. & Mooers, Biogeography. Princeton, NJ: Princeton University Press. A.O. 2012. The global diversity of birds in space and time. MacKinnon, J. & Phillipps, K. 1993. A Field Guide to the Nature 491: 444–448. Birds of Borneo, Sumatra, Java, and Bali, the Greater Jhwueng, D.C., Huzurbazar, S., O’Meara, B.C. & Liu, L. 2014. Sunda Islands. Oxford: Oxford University Press. Investigating the performance of AIC in selecting phylogenetic Maddison, W.P. & Maddison, D.R. 2018. Mesquite: a models. Stat. Appl. Genet. Mol. Biol. 13: 459–475. modular system for evolutionary analysis. Version 3.40. Jonsson, B., Jonsson, N., Hindar, K., Northcote, T.G. & Available at: http://mesquiteproject.org Engen, S. 2008. Asymmetric competition drives lake use of Martin, P.R. & Martin, T.E. 2001. Ecological and fitness coexisting salmonids. Oecologia 157: 553–560. consequences of species coexistence: a removal Kasahara, S. & Katoh, K. 2008. Food-niche differentiation in experiment with wood warblers. Ecology 82: 189–206. sympatric species of kingfishers, the Common Kingfisher Martin, T.E., Kelly, D.J., Keogh, N.T., Heriyadi, D., Singer, Alcedo atthis and the Greater Pied Kingfisher Ceryle H.A. & Blackburn, G.A. 2012. The avifauna of Lambusango lugubris. Ornithol. Sci. 7: 123–134. Forest Reserve, Buton Island, south-east Sulawesi, with Kelly, D.J. & Marples, N.M. 2010. Bird populations of the additional sightings from southern Buton. Forktail 28:107–112. Wakatobi. In Clifton, J., Unsworth, R.K.F. & Smith, D.J. Mayr, E. 1942. Systematics and the Origin of Species. New (eds) Marine Research and Conservation in the Coral York, NY: Columbia University Press. Triangle: The Wakatobi National Park: 149–170. New York, Mayr, E. & Diamond, J.M. 2001. The birds of Northern NY: Nova. Melanesia: Speciation, Ecology & Biogeography. Oxford: Kelly, S.B.A., Kelly, D.J., Cooper, N., Bahrun, A., Oxford University Press. Analuddin, K. & Marples, N.M. 2014. Molecular and McDonald, P.G. & Griffith, S.C. 2011. To pluck or not to pluck: phenotypic data support the recognition of the Wakatobi the hidden ethical and scientific costs of relying on feathers as Flowerpecker (Dicaeum kuehni) from the unique and a primary source of DNA. J. Avian Biol. 42: 197–203. understudied Sulawesi Region. PLoS ONE 9: e98694. McKay, B.D., Reynolds, M.B.J., Hayes, W.K. & Lee, D.S. Kerr, K.C., Stoeckle, M.Y., Dove, C.J., Weigt, L.A., Francis, 2010. Evidence for the species status of the Bahama C.M. & Hebert, P.D. 2007. Comprehensive DNA barcode Yellow-throated Warbler (Dendroica ‘dominica’ flavescens). coverage of North American birds. Mol. Ecol. Notes 7: 535– Auk 127: 932–939. 543. Milsom, J., Ali, J. & Sudarwono. 1999. Structure and Kumar, S., Stecher, G. & Tamura, K. 2016. MEGA7: collision history of the Buton continental fragment, eastern molecular evolutionary genetics analysis version 7.0 for Indonesia. AAPG Bull. 83: 1666–1689. bigger datasets. Mol. Biol. Evol. 33: 1870–1874. Moyle, R.G., Filardi, C.E., Smith, C.E. & Diamond, J. 2009. Lack, D.L. 1971. Ecological Isolation in Birds. Cambridge: Explosive Pleistocene diversification and hemispheric Harvard University Press. expansion of a ‘great speciator’. Proc. Natl Acad. Sci. USA Latta, S.C., Cabezas, S., Mejia, D.A., Paulino, M.M., 106: 1863–1868. Almonte, H., Miller-Butterworth, C.M. & Bortolotti, G.R. Nosil, P., Funk, D.J. & Ortiz-Barrientos, D. 2009. Divergent 2016. Carry-over effects provide linkages across the annual selection and heterogeneous genomic divergence. Mol. cycle of a Neotropical migratory bird, the Louisiana Ecol. 18: 375–402. Waterthrush Parkesia motacilla. Ibis 158: 395–406. O’Connell, D.P., Sealy, S., O Marcaigh, F., Karya, A., Leigh, J.W. & Bryant, D. 2015. PopART: full-feature software Bahrun, A., Analuddin, K., Kelly, D.J. & Marples, N.M. for haplotype network construction. Methods Ecol. Evol. 6: 2017. The avifauna of Kabaena Island, south-east Sulawesi, 1110–1116. Indonesia. Forktail 33:14–19. Leisler, B. & Schulze-Hagen, K. 2011. The Reed Warblers: Orsini, L., Vanoverbeke, J., Swillen, I., Mergeay, J. & Diversity in a Uniform Bird Family. Zeist: KNNV Publishing. Meester, L. 2013. Drivers of population genetic

differentiation in the wild: isolation by dispersal limitation, Saitoh, T., Cibois, A., Kobayashi, S., Pasquet, E. & isolation by adaptation and isolation by colonization. Mol. Thibault, J.C. 2012. The complex systematics of the Ecol. 22: 5983–5999. Acrocephalus of the Mariana Islands, western Pacific. Emu Pedersen, M.P., Irestedt, M., Joseph, L., Rahbek, C. & 112: 343–349. Jønsson, K.A. 2018. Phylogeography of a ‘great speciator’ Schoener, T.W. 1974. Resource partitioning in ecological (Aves: Edolisoma tenuirostre) reveals complex dispersal and communities. Science 185:27–39. diversification dynamics across the Indo-Pacific. J. Biogeogr. Scott, S.N., Clegg, S.M., Blomberg, S.P., Kikkawa, J. & 45: 826–837. Owens, I.P. 2003. Morphological shifts in island-dwelling Phillimore, A.B., Owens, I.P.F., Black, R.A., Chittock, J., birds: the roles of generalist foraging and niche expansion. Burke, T. & Clegg, S.M. 2008. Complex patterns of genetic Evolution 57: 2147–2156. and phenotypic divergence in an island bird and the Shapiro, B., Rambaut, A. & Drummond, A.J. 2006. consequences for delimiting conservation units. Mol. Ecol. Choosing appropriate substitution models for the 17: 2839–2853. phylogenetic analysis of protein-coding sequences. Mol. Pimm, S.L., Jones, H.L. & Diamond, J. 1988. On the risk of Biol. Evol. 23:7–9. extinction. Am. Nat. 132: 757–785. Slikas, B., Jones, I.B., Derrickson, S.R. & Fleischer, R.C. R Development Core Team. 2017. R: A Language and 2000. Phylogenetic relationships of Micronesian white-eyes Environment for Statistical Computing. Vienna: R based on mitochondrial sequence data. Auk 117: 355–365. Foundation for Statistical Computing. Available at: http:// Sodhi, N.S., Koh, L.P., Brook, B.W. & Ng, P.K.L. 2004. www.R-project.org/ Southeast Asian biodiversity: an impending disaster. Trends Rambaut, A., Suchard, M., Xie, W. & Drummond, A. 2014. Ecol. Evol. 19: 654–660. Tracer v. 1.6. Available at http://tree.bio.ed.ac.uk/software/ Spurgin, L.G., Illera, J.C., Jorgensen, T.H., Dawson, D.A. & tracer/ (accessed 06 April 2018). Richardson, D.S. 2014. Genetic and phenotypic divergence Redfern, C.P.F. & Clark, J.A. 2001. Ringer’s Manual. in an island bird: isolation by distance, by colonization or by Thetford: British Trust for Ornithology. adaptation? Mol. Ecol. 23: 1028–1039. Reif, J., Reifova, R., Skoracka, A. & Kuczynski, L. 2018. Stuart, Y.E. & Losos, J.B. 2013. Ecological character Competition-driven niche segregation on a landscape scale: displacement: glass half full or half empty?. Trends Ecol. evidence for escaping from syntopy towards allotopy in two Evol. 28: 402–408. coexisting sibling passerine species. J. Anim. Ecol. 00:1–16. Tavares, E.S. & Baker, A.J. 2008. Single mitochondrial gene Ricklefs, R.E. & Cox, G.C. 1972. Taxon cycles in the West barcodes reliably identify sister-species in diverse clades of Indian avifauna. Am. Nat. 106: 195–219. birds. BMC Evol. Biol. 8: 81. Ripley, B., Venables, V., Bates, D.M., Hornik, K., Gebhardt, Taylor, B. & Kirwan, G.M. 2018. Slender-billed Cicadabird A. & Frith, D. 2016. MASS: Support functions and datasets (Edolisoma tenuirostre). In del Hoyo, J., Elliott, A., Sargatal, for Venables and Ripley’s MASS. R package version 3.1-120. J., Christie, D.A. & de Juana, E. (eds) Handbook of the Robertson, G.S., Bolton, M., Grecian, W.J., Wilson, L.J., Birds of the World Alive. Barcelona: Lynx Edicions. Davies, W. & Monaghan, P. 2014. Resource partitioning in Available at: https://www.hbw.com/node/57864 (accessed 15 three congeneric sympatrically breeding seabirds: foraging August 2018). areas and prey utilization. Auk 131: 434–446. Thibault, J.-C. & Cibois, A. 2006. The natural history and Rockwell, S.M., Wunderle, J.M., Sillett, T.S., Bocetti, C.I., conservation of Acrocephalus rimitarae, the endemic Ewert, D.N., Currie, D., White, J.D. & Marra, P.P. 2017. reedwarbler of Rimatara Island, Oceania. Bull. Br. Ornithol. Seasonal survival estimation for a long-distance migratory Club 126: 201–206. bird and the influence of winter precipitation. Oecologia 183: Thomas, R., Lello, J., Medeiros, R., Pollard, A., Seawrd, A., 715–726. Smith, J., Vafidis, J. & Vaughan, I. 2015. Data Analysis Rogers, K., Rogers, A., Rogers, D., Lane, B.A. & Male, B. with R Statistical Software. A Guidebook for Scientists. 1986. Bander’s Aid: a Guide to Ageing and Sexing Bush Newport: Eco-explore. Birds. St. Andrew’s: A. Rogers. Toews, D.P. & Brelsford, A. 2012. The biogeography of Ronquist, F. & Huelsenbeck, J.P. 2003. MrBayes 3: mitochondrial and nuclear discordance in animals. Mol. Ecol. Bayesian phylogenetic inference under mixed models. 21: 3907–3930. Bioinformatics 19: 1572–1574. Uy, J.A.C., Moyle, R.G. & Filardi, C.E. 2009. Plumage and Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D.L., song differences mediate species recognition between Darling, A., Hohna,€ S., Larget, B., Liu, L., Suchard, M.A. incipient flycatcher species of the Solomon Islands. & Huelsenbeck, J.P. 2012. MrBayes 3.2: efficient Bayesian Evolution 63: 153–164. phylogenetic inference and model choice across a large Voris, H.K. 2000. Maps of Pleistocene sea levels in Southeast model space. Syst. Biol. 61: 539–542. Asia: shorelines, river systems and time durations. J. Rubinoff, D. & Holland, B.S. 2005. Between two extremes: Biogeogr. 27: 1153–1167. mitochondrial DNA is neither the panacea nor the nemesis Wallace, A.R. 1869. The Malay Archipelago: The Land of the of phylogenetic and taxonomic inference. Syst. Biol. 54: Oranguatan, and the Bird of Paradise. A Narrative of Travel, 952–961. with Studies of Man and Nature. New York, NY: Harper. Runemark, A., Hansson, B., Pafilis, P., Valakos, E.D. & Ward, P. 1968. Origin of the avifauna of urban and suburban Svensson, E.I. 2010. Island biology and morphological Singapore. Ibis 110: 239–255. divergence of the Skyros wall lizard Podarcis gaigeae:a Warren, B.H., Bermingham, E., Bowie, R.C., Prys-Jones, combined role for local selection and genetic drift on color R.P. & Thebaud, C. 2003. Molecular phylogeography morph frequency divergence? BMC Evol. Biol. 10: 269. reveals island colonization history and diversification of

western Indian Ocean sunbirds (Nectarinia: Nectariniidae). Sargatal, J., Christie, D.A. & de Juana, E. (eds) Handbook Mol. Phylogenet. Evol. 29:67–85. of the Birds of the World Alive. Barcelona: Lynx Edicions. Warren, B.H., Bermingham, E., Prys-Jones, R.P. & Available at: https://www.hbw.com/node/55745 (accessed 15 Thebaud, C. 2006. Immigration, species radiation and August 2018). extinction in a highly diverse songbird lineage: white-eyes Woodall, P.F. & Kirwan, G.M. 2018b. Sacred Kingfisher on Indian Ocean islands. Mol. Ecol. 15: 3769–3786. (Todiramphus sanctus). In del Hoyo, J., Elliott, A., Sargatal, Weigelt, P., Steinbauer, M.J., Cabral, J.S. & Kreft, H. 2016. J., Christie, D.A. & de Juana, E. (eds) Handbook of the Late Quaternary climate change shapes island biodiversity. Birds of the World Alive. Barcelona: Lynx Edicions. Nature 532: 99. Available at: https://www.hbw.com/node/55771 (accessed 15 White, C.M.N. & Bruce, M.D. 1986. The Birds of Wallacea August 2018). (Sulawesi, the Moluccas and Lesser Sunda Islands, Indonesia): an annoTated Check-list. London: British Received 7 April 2018; Ornithologists’ Union (Check list no. 7). revision accepted 9 October 2018. Whittaker, R.J. & Fernandez-Palacios, J.M. 2007. Island Associate Editor: Gary Voelker Biogeography: Ecology, Evolution and Conservation. Oxford: Oxford University Press. Wijesundara, C.S. & Freed, L.A. 2018. Divergence of SUPPORTING INFORMATION morphological characters in two white-eye species (Passeriformes: Zosteropidae) in sympatry. Ecosphere 9: Additional supporting information may be found e02317. online in the Supporting Information section at Wilcove, D.S., Giam, X., Edwards, D.P., Fisher, B. & Koh, L.P. the end of the article. ’ 2013. Navjot s nightmare revisited: logging, agriculture, and Table S1. Full list of samples utilized in the biodiversity in Southeast Asia. Trends Ecol. Evol. 28:531–540. Wilson, E.O. 1961. The nature of the taxon cycle in the phylogenetic analysis, detailing the species, sam- Melanesian ant fauna. Am. Nat. 95: 169–193. pling location and museum ID. Woodall, P.F. 2001. Family Acedinidae (Kingfishers). In del Table S2. Novel primers developed for this Hoyo, J., Elliott, A. & Christie, D.A. (eds), Handbook of the study. DOC-ND3-R1 was developed as a match – Birds of the World. Volume 6: Mousebirds to Hornbills: 130 for L10755 (Chesser 1999) to sequence the whole 249. Barcelona: Lynx Edicions. Woodall, P.F. 2018b. Collared Kingfisher (Todiramphus ND3 region. All other listed primers are internal chloris). In del Hoyo, J., Elliott, A., Sargatal, J., Christie, D.A. primers for ND2. & de Juana, E. (eds), Handbook of the Birds of the World Tables S3. Pairwise distances between sampled Alive. Barcelona: Lynx Edicions. Available at: https://www.hb taxa for ND2. w.com/node/55767 (accessed 15 August 2018). Table S4. Pairwise distances between sampled Woodall, P.F. 2018c. Talaud Kingfisher (Todiramphus enigma). In del Hoyo, J., Elliott, A., Sargatal, J., Christie, taxa for COI. D.A. & de Juana, E. (eds) Handbook of the Birds of the Table S5. DFA categorization of Sulawesi Todi- World Alive. Barcelona: Lynx Edicions. Available at: https:// ramphus populations. www.hbw.com/node/55768 (accessed 15 August 2018). Table S6. Existing vouchered specimens of Col- fi Woodall, P.F. & Kirwan, G.M. 2018a. Black-billed King sher lared Kingfishers from Southeast Sulawesi. (Pelargopsis melanorhyncha). In del Hoyo, J., Elliott, A.,