Novel Insights from the Summit Rat (Rattus Baluensis), a Borneo Mountain Endemic

DOI: 10.1111/ddi.12761

BIODIVERSITY RESEARCH

Interglacial refugia on tropical mountains: Novel insights from the summit rat (Rattus baluensis), a Borneo mountain endemic

1Conservation and Evolutionary Genetics Group, Estación Biológica de Doñana (EBD- Abstract CSIC), Seville, Spain Aim: The genetics of organisms currently isolated in refugia has received little atten- 2 Smithsonian Conservation Biology tion compared to post-glacial expansions. We study the population history and con- Institute, Center for Conservation Genomics, National Zoological Park, nectivity of a rat endemic to montane habitat in Borneo to better understand the Washington, DC, USA history and potential of populations in interglacial mountain refugia. 3Division of Mammals, National Museum Location: Sabah, Borneo, Malaysia. of Natural History, Smithsonian Institution, Washington, DC, USA Methods: We performed a field survey of the summit rat (Rattus baluensis) on two 4Sabah Parks, Sabah, Malaysia mountains, Mt. Kinabalu and Mt. Tambuyukon, its entire known distribution. We se- 5 Environmental Futures Research quenced mitogenomes and 27 introns (19 of which were polymorphic) in 49 individu- Institute, School of Environment, Griffith University, Brisbane, QLD, Australia als from both populations. We analysed their current genetic structure and diversity, and inferred their demographic history with approximate Bayesian computation. Correspondence Jennifer Leonard, Conservation and Results: Summit rats were tightly associated with mountain mossy forest and scrub- Evolutionary Genetics Group, Estación land above 2,000 m, facilitating the prediction of their past and future distributions. Biológica de Doñana (EBD-CSIC), Seville, Spain. The genetic analysis supports a Holocene fragmentation of a larger population into Email: [email protected] smaller ones that are now isolated in interglacial refugia on mountaintops. These

Present address findings are consistent with climatic reconstructions and the retreat of upland forest Melissa T. R. Hawkins, Department of to higher elevations after the Last Glacial Maximum (LGM), ~21 kya. Biological Sciences, Humboldt State University, Arcata, CA, USA. Main conclusions: The two isolated populations of summit rats formed through the upland shift of their habitat after the LGM. The current trend of global warming will Funding information likely lead to diminishing suitable upland habitat and result in the extinction of the Ministerio de Economía y Competitividad, Grant/Award Number: BES-2011-049186, population on Mt. Tambuyukon. The population on Mt. Kinabalu, the higher peak, CGL2010-21524, CGL2014-58793-P and could persist at higher elevations, highlighting the singular value of high tropical EEBB-I-12-05317 mountains as reservoirs of biodiversity during past and ongoing climate change. Editor: Jeremy Austin KEYWORDS approximate Bayesian computation, climate change, demographic history, introns, mitochondrial genomes, multiplex, Sunda Shelf, Sundaland

1 | INTRODUCTION processes in species that experience range contractions during cold periods to glacial refugia in lower latitudes and range expan- The Quaternary is characterized by pronounced environmental sions during interglacials (Hewitt, 1996, 1999, 2000; Petit et al., fluctuations, which have driven changes in the ranges of many 2003; Stewart, Lister, Barnes, & Dalén, 2010). The opposite pat- species. These dynamics have been closely studied in temperate tern, refugia during interglacials, is also possible in other habitats habitats to explain common phylogeographic patterns and genetic (Stewart et al., 2010), including on tropical mountains (Hewitt,

2000). This lesser studied scenario, referred to as “interglacial re- 2 | METHODS fugia” (Bennett & Provan, 2008; Stewart et al., 2010), has been recognized in a handful of studies on tropical mountains, including 2.1 | Study system beetles from Central America (MacVean & Schuster, 1981), plants The summit rat is apparently present continuously across the high in Africa (Kebede, Ehrich, Taberlet, Nemomissa, & Brochmann, montane habitat (Figure 1b). Its lower distribution limit is situ- 2007) and mammals from southern China (He, Hu, Chen, Li, & ated in the mossy forest just above 2,000 m (Figure 1a). On Mt. Jiang, 2016; He & Jiang, 2014), and is associated with sky islands Tambuyukon, we recorded it as low as 2,040 m (Table 1) and the (McCormack, Huang, & Knowles, 2009). However, there is still a lowest records from Mt. Kinabalu are around 2,100 (Musser, 1986; significant lack of knowledge on the phylogeographic patterns and Phillipps & Phillipps, 2016), although it has been observed with genetic processes in these interglacial refugia. Tropical mountains camera traps just over 2,000 m in association with pitcher plants offer the opportunity to study these processes as organisms may (Greenwood, Clarke, Ch’ien, Gunsalam, & Clarke, 2011; Wells, Lakim, shift their ranges to higher elevations instead of higher latitudes & Beaucournu, 2011; Wells, Lakim, Schulz, & Ayasse, 2011). At these with increasing temperatures (Colwell, Brehm, Cardelus, Gilman, lower elevations, it is scarce, but becomes much more abundant in & Longino, 2008). upper montane dwarf forest and scrubland, up to the summit on Mt. At 4,095 m, Mt. Kinabalu, Borneo, is the highest peak in Tambuyukon (2,579 m), and up to at least 3,200–3,426 m on Mt. Southeast Asia between the Himalayas and New Guinea, and the Kinabalu (Figure 1c; Musser, 1986; Nor, 2001; Phillipps & Phillipps, most studied mountain in the region. It is protected within Kinabalu 2016). Despite the lack of abundance data above this elevation, there National Park, which was declared a UNESCO World Heritage Site are museum records up to 3,810 m (Musser, 1986), and it likely even in 2000 for its great diversity and endemism. Much of this diversity reaches the summit (Park staff, personal communication). The sum- is associated with its montane habitats (Merckx et al., 2015), where mit rat has a mutualistic relationship with Nepenthes rajah, a pitcher high levels of endemism are found in plants (Beaman, 2005; van der plant found on montane ultramafic outcrops: the rat climbs on to Ent, Repin, Sugau, & Meng Wong, 2015; Raes, Roos, Slik, Van Loon, the pitchers to lick sugar exudates on their lids and the pitcher col- & Ter Steege, 2009), birds (Smythies, 1964) and mammals (Phillipps lects its droppings, which are rich in nitrogen (Figure 1b; Greenwood & Phillipps, 2016). During the Last Glacial Maximum (LGM), ~ 21 et al., 2011; Wells, Lakim, et al., 2011). The IUCN Red List classifies thousand years ago (Kya), the top of Mt. Kinabalu was covered by the status of the summit rat as “Least Concern” due to a presumably glaciers (Stauffer, 1968; 3,658 m snow line; 3,665 m in Porter, 2001). large population size and a distribution within a well-protected area At that time, the montane forest reached its maximum extension in (Aplin, 2016). Sundaland, and ever since has been in regression to higher elevations (Cannon, Morley, & Bush, 2009). As montane habitat became reduced and fragmented, it is likely that the species associated with 2.2 | Fieldwork this habitat became isolated in mountaintop refugia. We trapped small mammals including the summit rat during two field Such seems to be the case for the summit rat (Rattus baluensis expeditions to Kinabalu National Park (Sabah, Malaysia) in 2012 and Thomas, 1894), a high-elevation endemic which seems to now be 2013, across two different peaks, Mt. Kinabalu and Mt. Tambuyukon in a refugial state. It was previously only known from the upper (Figure 1a). Box-style live traps were set on transects from 331 to slopes of Mt. Kinabalu (6.07°N 116.56°E) above 2,100 m (Phillipps & 2,509 m following the mountain trail on Mt. Tambuyukon for a Phillipps, 2016), but here, we report our discovery of a second popu- total effort of 5,957 trap-nights. On Mt. Kinabalu, we trapped for lation on Mt. Tambuyukon (2,579 m; 6.20°N 116.66°E; Figure 1a), a 2,022 trap-nights, from 503 to 1,007 m in and around Poring Hot peak 18 km away in the same range but with a disjunct upper mon- Springs and from 1,512 to 3,466 m following the Kinabalu summit tane forest isolated by lower montane forest. We use this system trail (Figure 1a; Hawkins, Camacho-Sanchez, Yuh, Maldonado, & to characterize the genetic processes and degree of connectivity Leonard, 2018). All elevations were extracted using field GPS co- between populations of a species in fragmented and isolated inter- ordinates on a 1 arc-second SRTM digital elevation model (http:// glacial refugia. We trapped summit rats along an altitudinal gradient earthexplorer.usgs.gov/). Field samples were collected accord- on Mt. Kinabalu and Mt. Tambuyukon and sequenced complete mi- ing to the guidelines of the American Society of Mammalogists tochondrial genomes and a novel panel of nuclear markers. Using an (Sikes, et al. 2016), as approved by Institutional Animal Care and approximate Bayesian framework, we show an ancestral population Use Committees (Estación Biológica de Doñana Proposal Number was reduced and fragmented in the Holocene, leading to complete CGL2010-21524 and Smithsonian Institution, National Museum of genetic isolation. With ongoing climate change, we predict an ups- Natural History, Proposal Number 2012-04), with permission from lope shift of its suitable habitat of around 500 m by the end of this Sabah Parks (TS/PTD/5/4 Jld. 45 (33) and TS/PTD/5/4 Jld. 47 (25)) century. Mt. Kinabalu is the highest of the two peaks and its summit and the Economic Planning Unit (100-24/1/299), and exported with rat population could persist on its upper slopes, but the recently dis- permissions from the Sabah Wildlife Department (JHL.600-3/7 covered population on Mt. Tambuyukon will likely go extinct as the Jld.7/19 and JHL.600-3/7 Jld.8/) and Sabah Biodiversity Council montane habitat on this lower peak contracts. 1254 | CAMACHO-SANCHEZ et al.

(Ref: TK/PP:8/8Jld.2). Voucher specimens were deposited at the ethanol precipitation protocol and then quantified with a Nanodrop Doñana Biological Station (EBD), Spain, and Sabah Parks, Malaysia. ND-1000 Spectrophotometer (Nano-Drop Technologies, Inc., Wilmington, DE, USA).

2.3 | Materials 2.4 | Sequencing and genotyping We included liver and ear samples of 47 summit rats from our field surveys and two additional DNA samples from previous fieldwork We amplified a panel of 30 introns, based on the list proposed by for a total of 49 samples: 27 from Mt. Kinabalu and 22 from Mt. Igea, Juste, and Castresana (2010) to study phylogenies of closely re- Tambuyukon (Table 1). The tissues we collected were preserved lated non-rodent mammals, although their variation at the intraspe- in the field at ambient temperature using NAP buffer (Camacho- cies level was unknown. After a test run, we selected a panel of 27 Sanchez, Burraco, Gomez-Mestre, & Leonard, 2013). We ex- primer pairs that amplified target introns in a single reaction opti- tracted the DNA following a standard phenol–chloroform and mized to reduce artefacts (Brandariz-Fontes et al., 2015; Table 2).

(a) (b)

FIGURE 1 (a) Study area with trapping locations (open circles) and approximate distribution of the summit rat (hashed area) in Kinabalu Park (marked with (c) dashed line). (b) Top: Summit rat licking sugar exudates from the lid of a Nepenthes rajah pitcher (photograph: Ch’ien C. Lee). Bottom: mountain scrubland habitat. (c) Elevation profile across Mt. Kinabalu and Mt. Tambuyukon, with trapping effort across elevation (bottom scale) and number of summit rats trapped (shaded area of pie charts), relative to captures of other small mammals (light area pie charts) at each trapping location. The trapping locations below 2,000 m, where summit rats were not caught, are not depicted. (d) Ancestry of the samples from Mt. Kinabalu, left side, and Mt. Tambuyukon, right side, estimated for (d) the most likely number of populations, K = 2, with structure. The horizontal dotted line indicates a threshold of 0.10 ancestry [Colour figure can be viewed at wileyonlinelibrary.com] CAMACHO-SANCHEZ et al. | 1255

TABLE 1 Samples studied. Collector field number, museum code, tissue type sampled, elevation where animal was caught, mountain of origin (K, Mt. Kinabalu; T, Mt. Tambuyukon), and average sequence coverage per intron and for complete mitogenomes

Coverage

#Field Museum number Tissue Elevation (m) Location Intron mitDNA

BOR326 EBD 30372M Liver 2,701 K 84.7 26.9 BOR343 EBD 30374M Liver 2,757 K 71.1 44.2 BOR344 - Liver 2,757 K 26.1 83.0 BOR348 - Liver 2,730 K 73.7 74.2 BOR354 - Liver 2,730 K 104.6 43.1 BOR362 EBD 30377M Liver 2,757 K 108.9 28.7 BOR364 - Ear 2,683 K 87.5 - BOR372 - Ear 2,752 K 64.0 - BOR373 - Ear 2,757 K 74.8 42.0 BOR374 - Ear 2,745 K 81.0 - BOR376 - Ear 2,670 K 94.4 - BOR377 - Ear 2,683 K 80.1 - BOR378 - Ear 2,730 K 91.1 - BOR379 - Ear 2,737 K 76.3 - BOR383 - Liver 3,275 K 88.3 81.3 BOR384 EBD 30380M Liver 3,294 K 78.9 74.9 BOR391 - Liver 3,317 K 106.6 46.5 BOR392 - Liver 3,367 K 70.0 19.9 BOR393 EBD 30382M Liver 3,336 K 85.7 17.5 BOR398 EBD 30383M Liver 3,426 K 166.7 206.1 BOR399 - Ear 3,317 K 120.1 115.2 BOR400 - Ear 3,235 K 88.6 - BOR401 - Ear 3,275 K 87.2 - BOR403 - Ear 3,382 K 98.8 - BOR405 - Ear 3,222 K 78.3 - BOR161 - Liver 2,050 T 95.7 37.7 BOR201 - Liver 2,449 T 101.0 37.4 BOR207 EBD 30360M Liver 2,377 T 118.4 140.9 BOR209 - Liver 2,363 T 116.2 96.6 BOR210 EBD 30361M Liver 2,381 T 112.0 26.9 BOR212 - Liver 2,477 T 89.3 36.1 BOR216 - Liver 2,363 T 96.1 28.9 BOR223 - Ear 2,449 T 108.3 93.6 BOR226 - Ear 2,363 T 116.6 - BOR230 - Liver 2,363 T 119.9 21.0 BOR519 - Liver 2,040 T 134.0 190.5 BOR528 EBD 30395M Liver 2,274 T 101.4 189.2 BOR529 EBD 30396M Liver 2,283 T 81.8 96.9 BOR531 - Ear 2,291 T 85.5 - BOR532 - Ear 2,305 T 81.2 120.7 BOR533 - Ear 2,363 T 108.7 131.9 BOR534 - Ear 2,344 T 92.7 - BOR540 EBD 30398M Liver 2,194 T 101.2 197.0

(Continues) 1256 | CAMACHO-SANCHEZ et al.

TABLE 1 (Continued)

Coverage

#Field Museum number Tissue Elevation (m) Location Intron mitDNA

BOR543 - Ear 2,291 T 64.2 - BOR548 EBD 30400M Liver 2,240 T 175.1 167.0 BOR556 - Ear 2,274 T 61.6 - BOR557 - Ear 2,349 T 156.7 - B0993 - - - K 57.5 S0903 - - - K 67.2

Sequencing primers and individual indices were added in a second To visualize the haplotype diversity in the introns, we aligned the PCR, and the amplicon libraries were sequenced on Roche 454 fol- alleles in all the samples for each of the 19 polymorphic loci. We lowing manufacturer’s instructions. Three individuals were repli- considered indels as a fifth character and used these alignments to cated. Genotype calling was performed manually, considering only build tcs haplotype networks (Clement, Posada, & Crandall, 2000) in those loci with a minimum of 10× coverage. To ensure reliability in PopART (http://popart.otago.ac.nz). calling heterozygotes, only alleles present in both forward and reverse reads and that had a minimum of 4× coverage were considered 2.6 | Mitochondrial structure and diversity (details in Appendix 1). We sequenced complete mitochondrial genomes for all 49 sam- The complete mitogenomes of the 32 samples (16 from Mt. Kinabalu ples (Table 1). Dual-indexed shotgun libraries were sequenced on and 16 from Mt. Tambuyukon) were aligned with the mafft v7.017 an Illumina MiSeq using 250-bp PE or an Illumina HiSeq 2000 with (Katoh & Standley, 2013) plug-in in geneious 8.1.5. The mitochondrial 100-bp PE reads. Three samples were replicated using the differ- matrix only had 0.01% missing data. We used the alignment to build ent library preparation protocols to ensure repeatability (details in a haplotype network in PopART (http://popart.otago.ac.nz). Most Appendix 1). We assembled mitogenomes with a custom pipeline population genetics and phylogeography studies on rodents use ei-

(Appendix 1) and generated consensus sequences in geneious using ther cytochrome b or the control region alone. We also built haplo- a 75% threshold and a minimum of 2× coverage. Only mitogenomes type networks for these regions extracted from the mitogenomes to with more than 99.5% of their sequence reconstructed were consid- compare the level of information provided by each of them. ered for downstream analysis, yielding a total of 32 mitogenomes We used DnaSP v5.10.01 to calculate the number of polymorphic from the 49 samples (Table 1). sites, parsimony-informative sites, haplotype diversity, number of haplotypes, nucleotide diversity (π) and Tajima’s D for the complete mi- togenome alignment, and for the cytochrome b and control region 2.5 | Nuclear genetic structure and diversity separately. We assessed the partition of the molecular variance be-

We assessed population structure with a Bayesian clustering tween and within populations with an AMOVA in GenAlEx 6.5, in method implemented in structure v 2.3.4 (Pritchard, Stephens, & which each polymorphic position was considered a different locus. Donnelly, 2000). We used an admixture model using locations as The complete mitochondrial alignment had a total of 16,315 posi- prior and assuming correlation of allele frequencies among populations. Indels and missing data are not considered in PopART haplo- tions. For each K from 1 to 6, we performed four independent runs type networks or DnaSP, and so wer e r emoved in GenAlEx, yielding a with 50,000 MCMC repetitions and a burn-in period of 10,000. We matrix of 16,264 positions of which 157 were polymorphic. generated consensus solutions for each K (1–6) by clustering the four runs with clumpak (Kopelman, Mayzel, Jakobsson, Rosenberg, 2.7 | Demographic history & Mayrose, 2015). Then, structure harvester (http://taylor0.biology.ucla.edu/structureHarvester, Earl & vonHoldt, 2012) was used We included 18 polymorphic introns to estimate a set of demo- to infer the most probable number of populations through the ΔK graphic parameters with approximate Bayesian computation method (Evanno, Regnaut, & Goudet, 2005). (ABC) in PopABC 1.0 (Lopes, Balding, & Beaumont, 2009; https:// The number of alleles, the observed and expected heterozy- sites.google.com/site/jsollarilopes/). These included the time of gosity, and the fixation index for each locality were calculated in the split between the Mt. Kinabalu and Mt. Tambuyukon popu-

GenAlEx 6.5 (Peakall & Smouse, 2012). We also evaluated the parti- lations (tev), the effective population size on Mt. Kinabalu (Ne1) tioning of nuclear genetic variation with an AMOVA and the genetic and Mt. Tambuyukon (Ne2), and the effective size of the ancestral differentiation, Fst, in GenAlEx. We calculated the nucleotide and population (NeA; Table 3). We explored a range of priors sampled haplotype diversity of introns with DnaSP version 5.10.1 (Librado & from uniform distributions, using upper limits of 20 ky for tev Rozas, 2009) with indels coded as a fifth state. and overestimations of effective population sizes from area and CAMACHO-SANCHEZ et al. | 1257 1 1 1 1 1 1 1 1 1 1 43,677,246:- 166,524,733:1 227,433,026:- 72,323,056:1 82,072,661 51,728,622:1 59,320,794:1 224,766,884:- 118,721,810:- 31,090,582:1 84756917:- 56,583,432:1 62,981,572:- 9,533,108:1 116225863:- 15,214,732:1 73,190,192:1 9,048,157:- 7,916,593:1 Location in Rnor_5.0 6:7,897,005- 9:9,046,025- 8:116216748- 4:224,761,678- 14:43,670,533- 5:166,490,225- 5:62,973,516- 11:84745603- 10:110199929:110209824:1 17:15469301:15471060:1 19:51,714,803- 4:56,572,293- 1:227,423,230- 3:167597817:167605882:1 1:235019784:235027733:1 10:82,051,977- 6:85803391:85807823:- 63,672,094:1 5:63,653,695- 10:59,305,745- 66,673,403:- 10:66,660,854- 13:72,316,370- 1:83095962:83099071:1 18:31,076,952- 10:9,528,490- 3:73,152,918- 3:118,677,063- 3:15,181,259- , size in base pairs Size in summit rat 390 389 415 379 408 396 447 393–394 403 402 397 385 390 409–419 370 416 482–486 424–428 420–421 390 422 411 374–376 394–398 383 400–403 403 m 0.026 0.06 0.034 0.03 0.04 0.05 0.04 0.03 0.06 0.04 0.032 0.04 0.03 0.07 0.07 0.054 0.05 0.07 0.06 0.022 0.04 0.05 0.06 0.04 0.05 0.07 0.034 Primer concentration Primer Reverse primer GGCAAAGAAATAAGGACCAGCA CTGGCCTTTCCCTGTTGTCT GGGGTGGCCTTGGTTGTATA TCCGGATCTTTGTGACCTTCT TTGGGAAARGATGAACCGGC GCACRGAATTGAGGCACACA GAGCATTACCTGGACCTGCT GCCCTGCAGAACATCAACAG AGTGKCCATAGAAGGATGCT GGATACAGTTCCGGAAAGCAC CCAGCAATGTCTGAACCTCC CTGGACCATGGGCTGCAA TCAGGATAAGCGCCGGGA GTCTCGCGGCAAGTCTTC CTYCGCTTATATGAACWACTGCA GGTCAATAACGGCACKGTGG TGAGGAGAGCTATGACACGG TGGTTGAACTGRACATCTCTCA TCTGCTTTTCCTGGAGTGCA CCCACCACATCTAYGGACTG TGAGTTTGAAGTGGTAACCGC CAGAAGGACCAGCAGAAGGA CCTTTGCCTGGGATGYGAAG GAGGCACAGCTTGTTGAGG ATTCGGACAGAGTTCCGCA TGTCAGAGGAAGATGARGAGGA AGGAAGGTGTGGTTGGTGAT TTTCCAATGACTTCCGGGAC Forward primer Forward GCTGGAGGTGGCTCTTCTG GAAAGGATGCTGTCTTGGCC CAGGCTCRGGTTTCCGGT TTATCTGGGAGAGCGTGACC CTCCTCAAGTCCCAGCATGT CCTTTAGTTGTGACCAGGCC ACAGAGAKCCCATGTCTTCC AGATGGACATGGTGGAGAAGA GACCTCCAAGTCCCAGCTG GGTACTCAGAGTGGCCAACA AAACCCCGAGAGAAGAAGCT AAAGCCCTACCTGTTTGCG GATGATGGGAGAGAAGCAGTC ACACCAGTTCCTGTCAAACA GGCAGAATCACACCTGGGA GCCTCCAAACACACAGTCAT TGAACTCAAACACGTACGTACT CATTGTGCAGAGGTGAGGAC TCTGAGAAGGTTTTGGGGCA GTGTATCCYCCCACAGTCAAA AATACTCCACGTTGCGWACC CTTCGGAGGCATGTTCTTCC AGAARCCGGCTCACTACCC ACCCTCATATGACAGAGGAGG CTGCTTGGTACTTCTCATTTTCA AACGTGATCAATGGGCAGTG (Rnor_5.0) ENSRNOG00000005420 Ensembl ID for Rattus norvegicus ENSRNOG00000047089 ENSRNOG00000029572 ENSRNOG00000027466 ENSRNOG00000002484 ENSRNOG00000015736 ENSRNOG00000010424 ENSRNOG00000001806 ENSRNOG00000036660 ENSRNOG00000013090 ENSRNOG00000017602 ENSRNOG00000007437 ENSRNOG00000020299 ENSRNOG00000017539 ENSRNOG00000020993 ENSRNOG00000042912 ENSRNOG00000007390 ENSRNOG00000015991 ENSRNOG00000017606 ENSRNOG00000008798 ENSRNOG00000002525 ENSRNOG00000020233 ENSRNOG00000019276 ENSRNOG00000003125 ENSRNOG00000005865 ENSRNOG00000008455 ENSRNOG00000007710 - 5 16 1 11 - - - 1 g 1 - 9 17 intron # intron - - 5 - 5 - 3 9 - 2 - 5 - - 1 13 - 3 - 3 17 - - 5 10 - - - 7 - - 5 10 - - - 3 - 7 Gene name- abcg8 alkbh7 apeh cd27 chrna9 dhrs3 fancg fetub fn3krp gadd45 il34 irf5 klc2 mmp9 ms4a2 mycbpap nfkbia npr2 p2rx1 pipox ptgs2- 7 rabac1 rgd735029 rogdi ssfa2 tmem87a usp20 PrimersTABLE 2 used to amplify the introns 27 and their location in the rat genome. All primer sequences Primer to concentration 3′. 5′ in μ 1258 | CAMACHO-SANCHEZ et al.

TABLE 3 Prior and posterior Posterior distribution distributions of the parameters simulated Parameter Description Prior Mode 95% HPD with PopABC

tev Years from split Uniform [1, 20,000] 3,005 777–6,167 Ne1 Population size on Mt. Uniform [10, 100,000] 25,741 6,819–63,806 Kinabalu Ne2 Population size on Mt. Uniform [10, 25,000] 3,611 1,103–8,141 Tambuyukon NeA Ancestral effective Uniform [1, 200,000] 36,872 20,317–63,494 population size −6 μ Average mutation rate 4.46 × 10 - - −6 μsd SD of μ across loci 1.74 × 10 - -

density estimates from the literature (Table 3; details in Appendix rats were trapped in mossy forest (2,022 m–~2,290 m), but it was −6 2). We provided a fixed mutation rate (μ) of 4.46 × 10 substitu- relatively common (~1/3 of the captures) in the montane dwarf tions per generation per locus (details in Appendix 2). PopABC was forest and scrubland (~2,290 m–2,509 m), particularly in areas run for 500,000 simulations in which the algorithm sampled priors with the pitcher plant Nepenthes rajah (Figure S3.1). On Mt. from the given distributions (Table 3). A set of nine population ge- Kinabalu, we did not trap any summit rat in the mossy forest at netic statistics for sequence data were computed by PopABC. A the 2,250–2,268 m transect, but it was again relatively abundant rejection threshold of 0.01 was used to keep only 1% of the simula- in upper dwarf forest and scrubland in the two upper transects: tions whose statistics were closest to those of the real data (Figure 2,615–2,759 m, ~1/4 of the captures, and at 3,222–3,466 m (~1/3 S2.1). Then, we ran a regression step (Beaumont, Zhang, & Balding, of the captures). Assuming a 2,000 lower distribution limit for

2002), using the r scripts make_pd2.r, loc2plot_d.r and reg_step.r, Rattus baluensis (see Discussion), its current potential distribu- 2 2 distributed with PopABC 1.0 in r 3.1.3 (R Core Team, 2015). Here, tion area spans 10.1 km on Mt. Tambuyukon and 114.7 km in each accepted set of parameters is given a weight according to Mt. Kinabalu, as computed from a SRTM DEM with 1 arc-second its distance to the real data. We plotted the fitted data following resolution. reg_step.r and calculated the mode and the 95% highest posterior density (HPD) intervals of the posterior distribution of the param- 3.2 | Sequencing eters with loc1statsx in loc2plot_d.r. Additionally, we explored the potential effect of a longer generation time of 2.5 years and mu- Between the two Roche 454 GS Junior runs, a total of 126,825 tation rates 5 times faster (Figures S2.2–S2.3 and Table S2.1 in reads more than 250 bp long were assigned to one of the 47 ani- Appendix S2). mals sequenced (not considering replicates). Most of those reads, 121,636 in total, mapped to one of the 27 introns in Rattus norvegicus, yielding an average coverage of 96 reads per locus per individual 3 | RESULTS (Table 1). The genotypes for all three replicates were consistent, although some allelic dropout was observed in cases of low coverage. 3.1 | Trapping and ecology The alleles are in GenBank (MG423625–MG425911). Although the summit rat was considered endemic to Mt. Kinabalu Whole mitochondrial genomes (≥ 99.5% completeness) were (Phillipps & Phillipps, 2016), we discovered a second popula- reconstructed from 32 of the 49 samples attempted (GenBank tion on Mt. Tambuyukon, another peak in the same mountain KY611359–KY611390) with an average coverage of 82.5× (Table 1). range, 18 km north (Figure 1a). We recorded summit rats on We found no mismatches in any of the replicated samples. transects that ranged in elevation from 2,040 to 2,477 m on Mt. Tambuyukon and from 2,670 to 3,426 m on Mt. Kinabalu (Table 1) 3.3 | Nuclear genetic diversity and structure in a total of 2,276 trap-nights. At these locations, we trapped a total of 211 different small mammals of which 48 were summit rats We amplified and sequenced the 27 targeted introns. One of them, (Figure 1c), with an overall trapping success ranging from 5.6% to mycbpap-11, was discarded because of low coverage. Of the remain- 15.4% depending on the transect (Hawkins et al., 2018). Despite ing 26 introns, six were monomorphic (alkbh7-3, apeh-17, fancg-9, a larger trapping effort of 5,703 trap-nights, the summit rat was irf5-7, pipox-5 and ptgs2-7) and were thus discarded. One of the never trapped in transects at lower elevations: 2,250–2,268 m polymorphic introns (fetub-1) had a very high observed/expected and below on Mt. Kinabalu, and 1,504–1,881 m and below on Mt. heterozygosity (Kinabalu: 0.84/0.49; Tambuyukon: 0.91/0.51), Tambuyukon (Figure 1c). On Mt. Tambuyukon, only two summit with two alleles of similar size (393 and 394 bp) but a high number CAMACHO-SANCHEZ et al. | 1259 ) ), −3 θ 9.28 4.72 1.04 1.09 1.03 4.10 1.94 1.88 0.30 0.96 0.05 0.93 0.89 0.11 0.90 0.32 0.26 3.47 2.25 2.35 2.2 π (×10 0.14 0.61 0.40 0.02 0.36 0.52 0.44 0.16 0.36 0.40 0.25 0.40 0.05 0.57 0.46 0.37 0.04 0.47 0.10 0.32 0.2 θ 3 4 2 2 2 4 2 4 4 2 3 h 5 2 3 3 2 2 2 3 2.8 1.0 2 5 1 1 1 5 1 5 4 1 5 S 4 1 8 Overall 4 1 3 3 2 3.0 2.0 - - - 0.64 0.26 0.64 0.29 0.11 0.00 0.06 −0.27 −0.03 −0.05 −0.20 −0.11 F −0.10 −0.29 −0.02 −0.13 −0.02 −0.05 0.50 0.13 0.52 0.05 0.13 0.10 0.49 0.17 0.00 0.17 0.58 0.00 0.35 0.00 0.04 0.20 0.04 0.46 0.17 0.43 0.04 He 0.64 0.38 0.05 0.05 0.05 0.10 0.59 0.19 0.00 0.18 0.41 0.00 0.45 0.00 0.05 0.23 0.05 0.41 0.17 0.43 0.04 Ho ) −3 1.31 1.31 1.71 4.95 1.69 1.19 1.09 1.5 0.12 0.27 0.32 0.25 0.00 0.89 0.00 0.00 0.38 0.11 3.71 3.57 2.93 π (×10 0.53 0.51 0.05 0.13 0.13 0.10 0.50 0.17 0.36 0.00 0.17 0.60 0.00 0.47 0.00 0.05 0.05 0.21 0.23 0.44 0.2 θ 3 2 2 2 2 2 2 2 2 1 2 3 1 3 h 1 2 2 2 2.0 2 0.6 4 1 1 1 1 1 1 4 4 0 2 4 0 8 Tambuyukon S 0 3 1 3 2.1 1 2.0 - - 0.12 0.03 0.12 0.07 0.01 0.05 0.02 −0.24 −0.09 −0.20 −0.02 −0.05 −0.15 −0.26 F −0.04 −0.02 −0.06 −0.17 −0.05 0.63 0.49 0.15 0.49 0.50 0.30 0.53 0.04 0.00 0.53 0.45 0.28 0.33 0.00 0.08 0.04 0.15 0.48 0.23 0.30 0.05 He 0.56 0.61 0.16 0.48 0.44 0.28 0.64 0.04 0.00 0.56 0.52 0.28 0.42 0.00 0.08 0.04 0.16 0.56 0.24 0.28 0.05 Ho ) −3 5.33 5.86 1.19 1.22 1.61 4.52 1.76 1.64 1.63 1.8 0.31 0.10 0.00 0.00 0.20 0.31 0.40 0.75 3.60 2.89 2.57 π (×10 0.65 0.50 0.15 0.50 0.51 0.31 0.54 0.04 0.00 0.55 0.46 0.29 0.33 0.00 0.08 0.04 0.49 0.15 0.31 0.30 0.21 θ 4 2 4 2 2 3 3 2 1 5 4 4 2 1 h 2 2 2 3 2.6 2 1.1 ), observed heterozygosity (Ho), expected heterozygosity (He) and fixation index (F) for the polymorphic 19 nuclear loci π 5 1 4 1 1 5 4 1 0 4 5 5 7 0 S Kinabalu 1 3 3 2 2.8 1 2.1 - 5 16 1 - - 1 g 1 - 9 17 - - 5 - 5 - 2 - 5 - - 13 - 3 - 3 10 - - 7 - 5 10 - - - 3 npr2 p2rx1 rgd735029 cd27 mmp9 ms4a2 nfkbia rabac1 rogdi ssfa2 tmem87a usp20 Locus abcg8 klc2 chrna9 dhrs3 MEAN gadd45 il34 fn3krp SD PopulationTABLE 4 genetic statistics on summit rats from each mountain separately and all together. Number of polymorphic sites (S), number of haplotypes ( diversity nucleotide (h), haplotype diversity ( 1260 | CAMACHO-SANCHEZ et al.

abcg8-9 dhrs3-3

cd27-5 chrna9-1

klc2-10 il34-3 gadd4g-1 fn3krp-5

nfkbia-5 mmp9-2 ms4a2-5 npr2-10

rogdi-7 rgd735029-5

p2rx1-3 rabac1-1

tmem87a-16 50 alleles 10 alleles usp20-17 1 allele

ssfa2-13 Mt Tambuyukon Mt Kinabalu

FIGURE 2 Minimum spanning haplotype networks for each of the 19 polymorphic introns studied. The size of the pie charts corresponds to the frequency of the haplotype; transversal lines represent single base pair differences or indels; small black circles represent missing haplotypes [Colour figure can be viewed at wileyonlinelibrary.com]

of polymorphic sites (S = 22). This pointed to different loci instead 50 alleles were private to Mt. Kinabalu, and five of 38 were private to of different alleles, and it was consequently discarded from down- Mt. Tambuyukon. The average nucleotide and haplotype diversities stream analysis. The remaining 19 loci were used to assess the nu- were also higher on Mt. Kinabalu (π = 0.00163; Hd = 0.31) than on Mt. clear genetic diversity and the structure of the populations. Tambuyukon (π = 0.00119; Hd = 0.23; Table 4). The mean observed The polymorphic introns had from two to five alleles per locus: heterozygosity was higher in Mt. Kinabalu (mean ± SD: 0.24 ± 0.05) nine introns had two alleles, five introns had three alleles, four introns than in Mt. Tambuyukon (mean ± SD: 0.17 ± 0.04; Table 4). The fix- had four alleles, and one had five alleles (Table 4; Figure 2). Fifteen of ation index (F), a measure of the inbreeding coefficient, was not CAMACHO-SANCHEZ et al. | 1261 different from zero in either population (mean ± SD; Mt. Kinabalu: An AMOVA assigned 22% of the genetic variance (p = .001) to −0.05 ± 0.02; Mt. Tambuyukon: −0.05 ± 0.06). the differences between the two mountains.

The structure analysis revealed that the most likely number of genetic groups was two (K = 2), one for each mountain (Figures 1d 3.5 | Demographic history and S3.2). All the samples had nearly 100% ancestry associated with their geographical population. Increasing the number of clus- The posterior distributions of the parameters in the PopABC anal y- ters up to K = 6 did not reveal any further genetic structure (Figure ses show that the summit rat populations from Mt. Kinabalu and Mt. S3.3). An AMOVA indicated that 22% of the genetic variation was Tambuyukon became isolated approximately 2,000 years ago (tev partitioned between the two mountains, whereas the variation mode, 3,005; 95% HPD, 777–6,167 years; Table 3; Figure 4). Before within individuals accounted for 68%, and only 10% among in- the split, the ancestral population size (NeA mode, 36,872; 95% HPD, dividuals within the clusters. The Fst was 0.22 (p = .01) indicat- 20,317–63,494) averaged about twice the size the current popula- ing a strong genetic differentiation between the rats on the two tion on Mt. Kinabalu (Ne1 mode, 25,741; 95% HPD, 6,819–63,806) mountains. and was around 10 times larger than the one on Mt. Tambuyukon (Ne2 mode, 3,611; 95% HPD, 1,103–8,141; Table 3; Figure 4). Assuming the potential distribution of this species as the area above 3.4 | Mitochondrial diversity and structure 2,000 m and a factor of 10 between effective population size and There were 12 different haplotypes in the 32 complete mitog- census size (Frankham, 1995), the densities of the summit rat on Mt. enomes. All haplotypes were unique to one of the two moun- Kinabalu and Mt. Tambuyukon are 2,243 rats per km2 and 3,584 rats tains (Figure 3). Haplotype diversity was higher on Mt. Kinabalu per km2, respectively. These densities are in the high range of densi- (Hd = 0.800), with eight haplotypes, than on Mt. Tambuyukon ties reported for other species of Rattus in PanTHERIA (Jones et al., (Hd = 0.642), with four haplotypes (Table 5). All the haplotypes 2009). differed by few base pairs (1–18) except one haplotype found in only one animal from Mt. Tambuyukon, which was 131 base 4 | DISCUSSION pairs different from the most similar haplotype in the network (Figure 2). This divergent haplotype led to increased nucleotide 4.1 | Distribution and ecology of the summit rat diversity and number of polymorphic sites in Mt. Tambuyukon with respect to Mt. Kinabalu (Table 5). When only cytochrome The summit rat was previously described as endemic to Mt. Kinabalu b or the control region was considered, the number of haplo- (Phillipps & Phillipps, 2016). We discovered a new population on types decreased to five and seven, respectively. The amount Mt. Tambuyukon, a peak in the same mountain range 18 km north, of information as measured by parsimony-informative sites where we trapped individuals between 2,040 and 2,477 m. These decreased from 22 when considering complete mitogenomes, data, along with previous records (Nor, 2001; Wells, Lakim, & to two with only cytochrome b or five with the control region Beaucournu, 2011), suggest a lower distribution limit for the sum- (Table 5; Figure S3.4). mit rat corresponding to the lower delineation of the cloud forest

FIGURE 3 Minimum spanning haplotype network for the complete mitochondrial genomes; 16 from Mt. Kinabalu and 16 from Mt. Tambuyukon. The pie chart size corresponds to the frequency of each haplotype; transversal lines represent a single base pair differences; small black circles represent missing haplotypes [Colour figure can be viewed at wileyonlinelibrary.com] 1262 | CAMACHO-SANCHEZ et al.

(mossy forest or upper montane forest) at ~2,000 m (Kitayama, 1992), although it is at lower densities in this habitat. Both nuclear and mitochondrial DNA supported genetic isolation of the popula- ns−1.40 0.00323 7 0.831 5 19 Overall tions on Mt. Kinabalu and Mt. Tambuyukon despite their proximity, <18 km. Thus, the lower montane forest (1,200–2,000 m; Kitayama, 1992) that connects the two mountains seems a barrier to dispersal between these two populations. We found summit rats to be most abundant in the higher altitude

informative sites, PI; dwarf forest and montane scrubland above ~2,300 m and up to at ns −1.43 0.00366 3 0.575 3 17 Mt. Tambuyukon Mt. least 3,426 m (Figure 1c) as previously reported (Musser, 1986; Nor, 2001; Phillipps & Phillipps, 2016). There is no abundance data above that elevation, just occasional records up to 3,810 m on Kinabalu (Musser, 1986). Densities are probably lower at these elevations given the low productivity in these subalpine and alpine zones on Mt. ns Kinabalu (Kitayama, 1992). This hypothesis is supported by the esti- 0.25 0.00114 4 0.725 2 3 Control region Kinabalu Mt. mated population densities for Mt. Kinabalu, 2,243 rats/km2, much lower than that estimated for Mt. Tambuyukon, 3,5848 rats/km2. This lower density for Mt. Kinabalu likely reflects averaging across the more productive mountain scrubland where densities could be −1.64 ns 0.00153 5 0.762 2 14 Overall similar to Mt. Tambuyukon, and the much less productive alpine and subalpine habitats where densities are likely much lower (Kitayama, 1992). These alpine habitats are absent on Mt. Tambuyukon because of its lower elevation. The summit rat seemed particularly abundant in areas with the

* presence of Nepenthes rajah (Figure S3.1). They have previously been reported to have a mutualistic relationship with this pitcher plant, 0.00177 3 0.575 1 13 Mt. Tambuyukon Mt. 1.86 that is also a Kinabalu park endemic (Greenwood et al., 2011; Wells, Lakim, et al., 2011). The summit rat feeds on its sugar exudates mainly at night, while the mountain treeshrew, Tupaia montana, is a daytime visitor to this plant. Given the high rate of visits to the plants by the two species, these pitcher plants may represent an import- 0.00082 4 0.717 2 2 Mt. Kinabalu Mt. Cytochrome b 1.40 ns ant food source in the areas where they are syntopic (Greenwood et al., 2011; Wells, Lakim, et al., 2011). In turn, the faeces of the rats and treeshrews may be an important source of nitrogen for N. rajah ** in the impoverished ultramafic outcrops it is restricted to (Clarke −2.22 0.00101 12 0.865 22 157 Overall et al., 2009; van der Ent et al., 2015). These ultramafic outcrops have a patchy distribution within Kinabalu Park, including the high elevation of Mt. Tambuyukon and several areas on Mt. Kinabalu, such as Mesilau, where the lowest elevation records of summit rats have been made, ~2,000 m, visiting N. rajah. However, R. baluensis

* and T. montana are also very abundant in areas without N. rajah. Therefore, N. rajah may improve the habitat of R. baluensis, but the −1.97 0.00141 4 0.642 15 140 Mt. Tambuyukon Mt. mutualism is not obligatory.

4.2 | Effects of late Quaternary changes

; number of haplotypes, nucleotide h; diversity, π We found strong genetic isolation between the populations of θ −0.91 ns 0.00035 8 0.800 8 24 Mt. Kinabalu Mt. Complete mitogenomes Complete summit rats on Mt. Tambuyukon and Mt. Kinabalu that probably derives from a Holocene fragmentation of a more widespread ancestral one. This is similar to the phylogeographic pattern of another

significant. Bornean mountain endemic, the mountain black-eye (Manthey et al.,

.05. 2017). These findings are consistent with models that predict Late p < .01. Tajima’s D π h θ PI S p < GeneticTABLE 5 diversity for complete mitochondrial genomes, cytochrome b and control region of 32 samples. Number of polymorphic sites, S; parsimony-haplotype diversity, ns, non- * ** Quaternary retreat of the upland vegetation to higher elevations in CAMACHO-SANCHEZ et al. | 1263

(a) (b)

(c) (d) FIGURE 4 Prior (dashed line) and posterior (solid line) distributions for the time of split (tev), ancestral effective population size (NeA) and effective population size of summit rats on Mt. Kinabalu (Ne1) and Mt. Tambuyukon (Ne2) used in the PopABC analyses [Colour figure can be viewed at wileyonlinelibrary. com]

difficult (Still, Foster, & Schneider, 1999). At the LGM, lower temperatures and lower humidity dropped the level of cloud formation on Mt. Kinabalu by around 439 m with temperatures 2.1°C cooler (Still et al., 1999). Considering only temperature, geological evidence and palynological evidence reveal an even more severe change in the mountains of tropical Southeast Asia since the LGM. On Mt. Kinabalu, a glaciated cap reached around 3,660 m, which implies depression of the snowline by around 990 m, equivalent to around ∆ 5.4°C from the LGM to recent (Porter, 2001). Evidence from glaciers in New Guinea also suggests a similar pattern, with temperature increase of 5–6°C on their mountaintops from the LGM (Hope, 2007). These values are also similar to the 6–9°C increase that pollen data suggest for mountains in tropical Southeast Asia and Australia FIGURE 5 Hypsographic curve of Mt. Kinabalu, binning raster since the LGM (Pickett et al., 2004). At the LGM, summit rats could values from a STRM DEM with 1 arc-second resolution (approx. have expanded their distribution to much lower elevations of around 31 m) every 20 m in elevation starting from the 1,200 m contour 720 m assuming a ∆ 6°C on Mt. Kinabalu since the LGM, a lapse rate line. The shape of the mountain is pyramidal (Elsen & Tingley, of 5°C/km for tropical mountains (Sekercioglu, Schneider, Fay, & 2015); thus, area always decreases with elevation. A dashed blue Loarie, 2008), and a correction of −120 m due to sea level change line marks the current lower distribution of Rattus baluensis (blue + red area) at 2,000 m. Dotted red line marks a mild prediction on (Cannon et al., 2009). If this estimation is accurate, it is possible that upslope shift of the isothermal by the end of the 21st century by there could be other remnant populations of the summit rat on other +490 m, with a consequent reduction in the potential distribution high mountains in northern Borneo (Appendix 4, Figure S4.1). Any of for Rattus baluensis (red area). The black area corresponds to the these scenarios is consistent with the summit rat populations on Mt. overplotting of the raster from Mt. Tambuyukon [Colour figure can Tambuyukon and Mt. Kinabalu sharing a large, continuous ancestral be viewed at wileyonlinelibrary.com] population. Genetic diversity was lower in the summit rats from Mt. parallel with temperature increase after the LGM. Recent models Tambuyukon than those from Mt. Kinabalu, likely a consequence of predict an increase of 3°C in temperature and a vegetation lapse more intense drift in the smaller Mt. Tambuyukon population after rate of 166 m/∆°C for the upland forest since the LGM in Sundaland, their isolation in higher elevations during the present interglacial. suggesting that habitats have shifted up mountains by approxi- We identified a single very divergent mitochondrial haplotype (sam- mately 500 m (Cannon et al., 2009). Cloud forests, which we found ple BOR 161): 131 substitutions apart from the closest haplotype in correlated with the lower distribution limit of R. baluensis, are very the network, despite all other haplotypes being only 1 to 18 bases sensitive ecosystems, dependent not only on temperatures but also apart. This animal was not divergent at any nuclear locus (Figure 1d), on humidity, so reconstruction of their historical distributions is and it could be either the retention of ancestral polymorphism or a 1264 | CAMACHO-SANCHEZ et al. case of mitochondrial introgression from another distant, unknown rat tightly follows the elevation shift in the thermal isoline derived population, or from its lowland sister species, Rattus tiomanicus from climate warming, the population isolated on Mt. Tambuyukon (Aplin et al., 2011). and its unique genetic variation will likely disappear by the end of the current century. However, the deleterious effects of global warming will likely be milder on Mt. Kinabalu, where the height of the mountain 4.3 | Predicted responses to ongoing climate change provides room for an elevational shift to a higher elevation (Figure 5). Elevational shifts induced by climate change have been previously This prediction should be applicable to other species with limited dis- observed in animal communities (Parmesan, 2006). For instance, persal capabilities that are tightly associated with this habitat. half of the small mammal community in Yosemite, USA, have shifted their ranges an average of 500 m uphill in one century (Moritz et al., ACKNOWLEDGEMENTS 2008). On a global scale, the effects of climate change do not mani- fest equally and are difficult to predict for a specific location because We thank all the staff that participated in the fieldwork. Ch’ien C. Lee they depend not only on the concentration of gases with green- kindly provided summit rat pictures. Anna Cornellas helped with lab- house effect but also on complex atmosphere–ocean global circula- oratory work. Vicente García Navas, Carles Vilà, Santiago Montero, tion models (IPCC, 2013). On a local scale, for most high-elevation Giovanni Forcina, Juanma Peralta, Inés Sánchez Donoso and other species studied, there is an upslope shift correlated with local rise members of CONSEVOL (www.consevol.org), Lillian Parker, Nancy of temperatures. This shift can be heterogeneous and often lags be- McInerney, Rob Fleischer and other members of the CCG at SCBI hind temperature change, probably due to local effects and species- (Smithsonian National Zoo), and Kristofer Helgen kindly provided in- specific responses (time delays, biotic interactions or physiological sight. We further thank the Malaysian institutions that allowed us to constraints) (Chen, Hill, Ohlemuller, Roy, & Thomas, 2011; Rowe et al., do fieldwork and export the samples: the Sabah Biodiversity Centre 2015). A re-survey of the moth community on Mt. Kinabalu showed for issuing a research and export permit, Sabah Parks for research that climate warming caused an uphill shift of 67 m in just 42 years and collection permits, as well as support and cooperation during (Chen et al., 2009), and a review of recent and historical distribution of our time in Malaysia, and the Sabah Wildlife Department and the the birds on Mt. Kinabalu points to a similar effect in their community Economic Planning Unit for research and export permits. Logistical (Harris et al., 2012). Upslope shifts of around 100 m have also been support was provided by Laboratorio de Ecología Molecular, Estación observed already in birds on mountains from New Guinea (Freeman Biológica de Doñana, CSIC (LEM-EBD). Digital elevation models for & Class Freeman, 2014). As climate change progresses, populations maps was obtained from the U.S. Geological Survey. The Spanish on some mountains may find themselves trapped on mountaintops Ministry of Science and Innovation grants CGL2010-21524 and (Colwell et al., 2008; Sauer, Domisch, Nowak, & Haase, 2011,Struebig CGL2014-58793-P supported this work. MCS is supported by the et al. 2015). Scenarios with CO2 concentrations of 690 ppm (twice the Spanish Ministry of Science and Innovation Predoctoral Fellowship concentration at the end of the 20th century) predict temperatures BES-2011-049186, and part of his fieldwork was also funded by +2.2°C warmer, increased absolute humidity and an upslope shift of EEBB-I-12-05317. the cloud forest by ~492 m on Mt. Kinabalu (Still et al., 1999). These CO concentrations best match mild IPCC scenarios RCP6.0 for 2100 2 ORCID (IPCC, 2013). Other models predict slightly higher temperature increases in Southeast Asian mountains, 2.3–3.1°C, for year 2085 Miguel Camacho-Sanchez http://orcid. (Nogués-Bravo, Araújo, Errea, & Martínez-Rica, 2007). Assuming org/0000-0002-6385-7963 an increase of 2.3 to 3.1°C on Mt. Kinabalu and a lapse rate of 5°C/ Konstans Wells http://orcid.org/0000-0003-0377-2463 km (Sekercioglu et al., 2008), by the end of this century the montane Jesus E. Maldonado http://orcid.org/0000-0002-4282-1072 communities in northern Borneo will have shifted up the mountains 460–620 m in elevation. The pyramidal shape of Mt. Kinabalu and Mt. Jennifer A. Leonard http://orcid.org/0000-0003-0291-7819 Tambuyukon (Figure 5) makes their highland community particularly vulnerable to climate change (Elsen & Tingley, 2015). A mild scenario REFERENCES of +490 m upslope shift of the lower distribution limit of R. baluensis from 2,000 m to 2,490 m will lead to a great loss in its potential area Aplin, K. (2016) Rattus baluensis. The IUCN Red List of Threatened Species 2016: e.T19323A22443731. Downloaded on 21 November available from 10.1 to 0.4 km2 (−96%) on Mt. Tambuyukon and from 2016. 2 114.7 to 60.7 km (−47%) on Mt. Kinabalu (Figure 5). Aplin, K. P., Suzuki, H., Chinen, A. A., Chesser, R. T., Ten Have, J., The consequent reduction in the distribution area for summit rats Donnellan, S. C., … Catzeflis, F. (2011). Multiple geographic origins of is likely to reduce population sizes and increase drift, eroding levels commensalism and complex dispersal history of black rats. PLoS ONE, 6, e26357. https://doi.org/10.1371/journal.pone.0026357 of genetic diversity both on Mt. Kinabalu and Mt. Tambuyukon. This Bálint, M., Domisch, S., Engelhardt, C. H., Haase, P., Lehrian, S., Sauer, J., genetic erosion can be particularly severe for genetically isolated Theissinger, K., … Nowak, C. (2011). Cryptic biodiversity loss linked populations such as species with sky-island distribution (Bálint et al., to global climate change. Nature Climate Change, 1, 313–318. https:// 2011), such as R. baluensis. Assuming the distribution of the summit doi.org/10.1038/nclimate1191 CAMACHO-SANCHEZ et al. | 1265

Beaman, J. H. (2005). Mount Kinabalu: Hotspot of plant diversity in Nepenthes rajah, and members of a small mammal community. PLoS Borneo. Biologiske Skrifter, 55, 103–127. ONE, 6, e21114. https://doi.org/10.1371/journal.pone.0021114 Beaumont, M. A., Zhang, W., & Balding, D. J. (2002). Approximate Harris, J. B., Yong, D. L., Sheldon, F. H., Boyce, A. J., Eaton, J. A., Bernard, Bayesian computation in population genetics. Genetics, 162, H., & Wei, D. (2012). Using diverse data sources to detect elevational 2025–2035. range changes of birds on Mount Kinabalu, Malaysian Borneo. The Bennett, K. D., & Provan, J. (2008). What do we mean by “refu- Raffles Bulletin of Zoology, 25, 197–247. gia”? Quaternary Science Reviews, 27, 2449–2455. https://doi. Hawkins, M. T., Camacho-Sanchez, M., Yuh, F. T., Maldonado, J. E., & org/10.1016/j.quascirev.2008.08.019 Leonard, J. A. (2018). Small mammal diversity along two neighbor- Brandariz-Fontes, C., Camacho-Sanchez, M., Vila, C., Vega- Pla, J. L., ing Bornean mountains. PeerJ Preprints, 6, e26523v1. https://doi. Rico, C., & Leonard, J. A. (2015). Effect of the enzyme and PCR con- org/10.7287/peerj.preprints.26523v1 ditions on the quality of high-throughput DNA sequencing results. He, K., Hu, N.-Q., Chen, X., Li, J.-T., & Jiang, X.-L. (2016). Interglacial re- Scientific Reports, 5, 8056. https://doi.org/10.1038/srep08056 fugia preserved high genetic diversity of the Chinese mole shrew in Camacho-Sanchez, M., Burraco, P., Gomez-Mestre, I., & Leonard, J. A. the mountains of southwest China. Heredity, 116, 23–32. https://doi. (2013). Preservation of RNA and DNA from mammal samples under org/10.1038/hdy.2015.62 field conditions. Molecular Ecology Resources, 13, 663–673. https:// He, K., & Jiang, X. (2014). Sky islands of southwest China. I: An overview doi.org/10.1111/1755-0998.12108 of phylogeographic patterns. Chinese Science Bulletin, 59, 585–597. Cannon, C. H., Morley, R. J., & Bush, A. B. G. (2009). The current refugial https://doi.org/10.1007/s11434-013-0089-1 rainforests of Sundaland are unrepresentative of their biogeographic Hewitt, G. (1996). Some genetic consequences of ice ages, and their role past and highly vulnerable to disturbance. Proceedings of the National in divergence and speciation. Biological Journal of the Linnean Society, Academy of Sciences, 106, 11188–11193. https://doi.org/10.1073/ 58, 247–276. https://doi.org/10.1111/j.1095-8312.1996.tb01434.x pnas.0809865106 Hewitt, G. (1999). Post-glacial re-colonization of European biota. Chen, I.-C., Hill, J. K., Ohlemuller, R., Roy, D. B., & Thomas, C. D. (2011). Biological Journal of the Linnean Society, 68, 87–112. https://doi. Rapid range shifts of species associated with high levels of cli- org/10.1111/j.1095-8312.1999.tb01160.x mate warming. Science, 333, 1024–1026. https://doi.org/10.1126/ Hewitt, G. (2000). The genetic legacy of the Quaternary ice ages. Nature, science.1206432 405, 907–913. https://doi.org/10.1038/35016000 Chen, I. C., Shiu, H. J., Benedick, S., Holloway, J. D., Chey, V. K., Barlow, H. Hope, G. S. (2007). Paleoecology and paleoenvironments of Papua. In A. S., … Thomas, C. D. (2009). Elevation increases in moth assemblages J. Marshall & B. M. Beehler (Eds.), The ecology of Papua: Part one (pp. over 42 years on a tropical mountain. Proceedings of the National 255–266). Singapore: Tuttle Publishing. Academy of Sciences of the United States of America, 106, 1479–1483. Igea, J., Juste, J., & Castresana, J. (2010). Novel intron markers to study the https://doi.org/10.1073/pnas.0809320106 phylogeny of closely related mammalian species. BMC Evolutionary Clarke, C. M., Bauer, U., Ch’ien, C. L., Tuen, A. A., Rembold, K., & Moran, Biology, 10, 369. https://doi.org/10.1186/1471-2148-10-369 J. A. (2009). Tree shrew lavatories: A novel nitrogen sequestration IPCC (2013) IPCC, 2013: Climate change 2013: The physical science basis. strategy in a tropical pitcher plant. Biology Letters, 5, 632–635. Contribution of working group I to the fifth assessment report of the https://doi.org/10.1098/rsbl.2009.0311 intergovernmental panel on climate change. Cambridge, UK and New Clement, M., Posada, D., & Crandall, K. A. (2000). TCS: A computer pro- York, NY, USA: Cambridge University Press. gram to estimate gene genealogies. Molecular Ecology, 9, 1657–1659. Jones, K. E., Bielby, J., Cardillo, M., Fritz, S. A., O’Dell, J., Orme, C. D., https://doi.org/10.1046/j.1365-294x.2000.01020.x … Connolly, C. (2009). PanTHERIA: A species-level database of life Colwell, R. K., Brehm, G., Cardelus, C. L., Gilman, A. C., & Longino, J. history, ecology, and geography of extant and recently extinct mam- T. (2008). Global warming, elevational range shifts, and lowland bi- mals. Ecology, 90, 2648. https://doi.org/10.1890/08-1494.1 otic attrition in the wet tropics. Science, 322, 258–261. https://doi. Katoh, K., & Standley, D. M. (2013). MAFFT multiple sequence align- org/10.1126/science.1162547 ment software version 7: Improvements in performance and us- Earl, D. A., & vonHoldt, B. M. (2012). STRUCTURE HARVESTER: A web- ability. Molecular Biology and Evolution, 30, 772–780. https://doi. site and program for visualizing STRUCTURE output and implement- org/10.1093/molbev/mst010 ing the Evanno method. Conservation Genetics Resources, 4, 359–361. Kebede, M., Ehrich, D., Taberlet, P., Nemomissa, S., & Brochmann, https://doi.org/10.1007/s12686-011-9548-7 C. (2007). Phylogeography and conservation genetics of a giant Elsen, P. R., & Tingley, M. W. (2015). Global mountain topography and lobelia (Lobelia giberroa) in Ethiopian and Tropical East African the fate of montane species under climate change. Nature Climate mountains. Molecular Ecology, 16, 1233–1243. https://doi. Change, 5, 772–776. https://doi.org/10.1038/nclimate2656 org/10.1111/j.1365-294X.2007.03232.x van der Ent, A., Repin, R., Sugau, J., & Meng Wong, K. (2015). Plant Kitayama, K. (1992). An altitudinal transect study of the vegetation diversity and ecology of ultramafic outcrops in Sabah (Malaysia). on Mount Kinabalu, Borneo. Vegetatio, 102, 149–171. https://doi. Australian Journal of Botany, 63, 204–215. org/10.1007/BF00044731 Evanno, G., Regnaut, S., & Goudet, J. (2005). Detecting the num- Kopelman, N. M., Mayzel, J., Jakobsson, M., Rosenberg, N. A., & Mayrose, ber of clusters of individuals using the software STRUCTURE: A I. (2015). Clumpak: A program for identifying clustering modes and simulation study. Molecular Ecology, 14, 2611–2620. https://doi. packaging population structure inferences across K. Molecular Ecology org/10.1111/j.1365-294X.2005.02553.x Resources, 15, 1179–1191. https://doi.org/10.1111/1755-0998.12387 Frankham, R. (1995). Effective population size/adult population size ra- Librado, P., & Rozas, J. (2009). DnaSP v5: A software for comprehensive tios in wildlife: A review. Genetical Research, 66, 95–107. https://doi. analysis of DNA polymorphism data. Bioinformatics, 25, 1451–1452. org/10.1017/S0016672300034455 https://doi.org/10.1093/bioinformatics/btp187 Freeman, B. G., & Class Freeman, A. M. (2014). Rapid upslope shifts in Lopes, J. S., Balding, D., & Beaumont, M. A. (2009). PopABC: A program New Guinean birds illustrate strong distributional responses of trop- to infer historical demographic parameters. Bioinformatics, 25, 2747– ical montane species to global warming. Proceedings of the National 2749. https://doi.org/10.1093/bioinformatics/btp487 Academy of Sciences, 111, 4490–4494. https://doi.org/10.1073/ MacVean, C., & Schuster, J. C. (1981). Altitudinal distribution of passa- pnas.1318190111 lid beetles (Coleoptera, Passalidae) and Pleistocene dispersal on the Greenwood, M., Clarke, C., Ch’ien, C. L., Gunsalam, A., & Clarke, R. H. (2011). volcanic chain of northern Central America. Biotropica, 13, 29–38. A unique resource mutualism between the giant Bornean pitcher plant, https://doi.org/10.2307/2387868 1266 | CAMACHO-SANCHEZ et al.

Manthey, J. D., Moyle, R. G., Gawin, D. F., Rahman, M. A., Ramji, M. F., Sekercioglu, C. H., Schneider, S. H., Fay, J. P., & Loarie, S. R. & Sheldon, F. H. (2017). Genomic phylogeography of the endemic (2008). Climate change, elevational range shifts, and bird ex- Mountain Black-eye of Borneo (Chlorocharis emiliae): Montane and tinctions. Conservation Biology, 22, 140–150. https://doi. lowland populations differ in patterns of Pleistocene diversification. org/10.1111/j.1523-1739.2007.00852.x Journal of Biogeography, 44, 1–12. Sikes, R. S., Gannon, W. L., & Animal Care and Use Committee of the McCormack, J. E., Huang, H., & Knowles, L. (2009) Sky islands. In R. G. American Society of Mammalogists (2016). 2016 Guidelines of the Gillespie & D. Clague (Eds.), Encyclopedia of islands (pp. 839–843). American Society of Mammalogists for the use of wild mammals in Berkeley, CA: University of California Press. research and education. Journal of Mammalogy, 97, 663–688. https:// Merckx, V. S., Hendriks, K. P., Beentjes, K. K., Mennes, C. B., Becking, doi.org/10.1093/jmammal/gyw078 L. E., Peijnenburg, K. T., … Buang, M. M. (2015). Evolution of ende- Smythies, B. E. (1964). The birds of Mt Kinabalu and their zoogeographical mism on a young tropical mountain. Nature, 524, 347–350. https:// relationships. Proceedings of the Royal Society of London B: Biological doi.org/10.1038/nature14949 Sciences, 161, 75–80. https://doi.org/10.1098/rspb.1964.0079 Moritz, C., Patton, J. L., Conroy, C. J., Parra, J. L., White, G. C., & Stauffer, P. H. (1968). Glaciation of Mount Kinabalu. Geological Survey of Beissinger, S. R. (2008). Impact of a century of climate change on Malaysia Bulletin, 1, 63. small-mammal communities in Yosemite National Park, USA. Science, Stewart, J. R., Lister, A. M., Barnes, I., & Dalén, L. (2010). Refugia re- 322, 261–264. https://doi.org/10.1126/science.1163428 visited: Individualistic responses of species in space and time. Musser, G. G. (1986). Sundaic Rattus: Definitions of Rattus baluensis and Proceedings of the Royal Society of London B: Biological Sciences, 277, Rattus korinchi. American Museum Novitates, 2862, 1–24. 661–671. https://doi.org/10.1098/rspb.2009.1272 Nogués-Bravo, D., Araújo, M. B., Errea, M. P., & Martínez-Rica, J. P. Still, C. J., Foster, P. N., & Schneider, S. H. (1999). Simulating the effects of (2007). Exposure of global mountain systems to climate warming climate change on tropical montane cloud forests. Nature, 398, 15–17. during the 21st Century. Global Environmental Change, 17, 420–428. Struebig, M. J., Wilting, A., Gaveau, D. L., Meijaard, E., Smith, R. J., https://doi.org/10.1016/j.gloenvcha.2006.11.007 Abdullah, T., … Belant, J. L. (2015). Targeted conservation to safeguard Nor, S. M. (2001). Elevational diversity patterns of small mammals on a biodiversity hotspot from climate and land-cover change. Current Mount Kinabalu, Sabah, Malaysia. Global Ecology and Biogeography, Biology, 25, 372–378. https://doi.org/10.1016/j.cub.2014.11.067 10, 41–62. https://doi.org/10.1046/j.1466-822x.2001.00231.x Wells, K., Lakim, M. B., & Beaucournu, J. C. (2011). Host specificity and Parmesan, C. (2006). Ecological and evolutionary responses to recent cli- niche partitioning in flea–small mammal networks in Bornean rain- mate change. Annual Review of Ecology, Evolution, and Systematics, 37, forests. Medical and Veterinary Entomology, 25, 311–319. https://doi. 637–669. https://doi.org/10.1146/annurev.ecolsys.37.091305.110100 org/10.1111/j.1365-2915.2010.00940.x Peakall, R., & Smouse, P. E. (2012). GenALEx 6.5: Genetic analy- Wells, K., Lakim, M. B., Schulz, S., & Ayasse, M. (2011). Pitchers of sis in Excel. Population genetic software for teaching and re- Nepenthes rajah collect faecal droppings from both diurnal and search. Bioinformatics, 28, 2537–2539. https://doi.org/10.1093/ nocturnal small mammals and emit fruity odour. Journal of Tropical bioinformatics/bts460 Ecology, 27, 347–353. https://doi.org/10.1017/S0266467411000162 Petit, R. J., Aguinagalde, I., de Beaulieu, J. L., Bittkau, C., Brewer, S., Cheddadi, R., … Mohanty, A. (2003). Glacial refugia: Hotspots but not melting pots of genetic diversity. Science, 300, 1563–1565. https:// BIOSKETCH doi.org/10.1126/science.1083264 Phillipps, Q., & Phillipps, K. (2016). Phillips’ field guide to the mammals of This research group is studying the ecology and evolution of the Borneo and their ecology. Kota Kinabalu: Natural History Publications biodiversity of the Sunda Shelf, with a focus on forest-dependent (Borneo). vertebrates. This system is particularly well suited to study alter- Pickett, E. J., Harrison, S. P., Hope, G., Harle, K., Dodson, J. R., Peter Kershaw, A., … Flenley, J. (2004). Pollen-based reconstructions of biome distri- native hypotheses regarding differentiation and speciation in the butions for Australia, Southeast Asia and the Pacific (SEAPAC region) tropics due to the geography and history of isolation and con- at 0, 6000 and 18,000 14C yr BP. Journal of Biogeography, 31, 1381– nection associated with Quaternary climatic fluctuations. More 1444. https://doi.org/10.1111/j.1365-2699.2004.01001.x information can be found at http://www.consevol.org/ Porter, S. C. (2001). Snowline depression in the tropics during the last glaciation. Quaternary Science Reviews, 20, 1067–1091. Author contributions: M.T.R.H., M.C.S. and K.W. collected the Pritchard, J. K., Stephens, M., & Donnelly, P. (2000). Inference of pop- samples. I.Q.R. and M.C.S did the laboratory work and data anal- ulation structure using multilocus genotype data. Genetics, 155, 945–959. ysis. M.C.S., I.Q.R. and J.A.L. wrote the manuscript. All authors R Core Team (2015) R: A Language and Environment for Statistical reviewed the final version of the manuscript. Computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, Retrieved from http://www.R-project.org/., 55, 275–286. SUPPORTING INFORMATION Raes, N., Roos, M. C., Slik, J. W. F., Van Loon, E. E., & Ter Steege, H. (2009). Botanical richness and endemicity patterns of Borneo de- Additional Supporting Information may be found online in the sup- rived from species distribution models. Ecography, 32, 180–192. porting information section at the end of the article. https://doi.org/10.1111/j.1600-0587.2009.05800.x Rowe, K. C., Rowe, K. M., Tingley, M. W., Koo, M. S., Patton, J. L., Conroy, C. J., … Moritz, C. (2015). Spatially heterogeneous impact of climate How to cite this article: Camacho-Sanchez M, Quintanilla I, change on small mammals of montane California. Proceedings of the Royal Society of London B: Biological Sciences, 282, 1–10. Hawkins MTR, et al. Interglacial refugia on tropical Sauer, J., Domisch, S., Nowak, C., & Haase, P. (2011). Low mountain mountains: Novel insights from the summit rat (Rattus ranges: Summit traps for montane freshwater species under climate baluensis), a Borneo mountain endemic. Divers Distrib. change. Biodiversity and Conservation, 20, 3133–3146. https://doi. 2018;24:1252–1266. https://doi.org/10.1111/ddi.12761 org/10.1007/s10531-011-0140-y