The Effects of Pleistocene Climatic Change on the Phylogeography of Euploea Mulciber (Lepidoptera; Nymphalidae; Danainae)

City University of New York (CUNY) CUNY Academic Works

Dissertations and Theses City College of New York

2012

The Effects of Pleistocene Climatic Change on the Phylogeography of Euploea mulciber (Lepidoptera; Nymphalidae; Danainae)

Elva Yang CUNY City College

How does access to this work benefit ou?y Let us know!

More information about this work at: https://academicworks.cuny.edu/cc_etds_theses/153 Discover additional works at: https://academicworks.cuny.edu

This work is made publicly available by the City University of New York (CUNY). Contact: [email protected]

The Effects of Pleistocene Climatic Change on the Phylogeography of Euploea mulciber (Lepidoptera: Nymphalidae: Danainae)

Elva J. H. Yang

Abstract

Southeast Asia is one of the most geologically and biogeographically complex regions in the world. Changing climate and fluctuating sea levels has had profound impacts on species diversification and faunal distribution in the region. During the Last Glacial Maximum (LGM) sea levels were lower than at present, and the Malay Peninsula, Sumatra, and Borneo formed one large landmass. However, there are great biogeographic disjunctions between eastern and western Sundaland, and the “savannah corridor” hypothesis has been proposed as an explanation of this phenomenon. We used a widespread, tropical butterfly species in Southeast Asia, Euploea mulciber, as a model organism to reassess the validity of the savannah hypothesis. We reconstructed the potential distribution of E. mulciber during the LGM and compared it to an intraspecific phylogeny of the species. The ecological niche models predicted that suitable habitat in eastern and western Sundaland were separated by a corridor of unsuitable habitat during the LGM, but lack of genetic differentiation of specimens from these two areas refutes the idea that this historical event structures contemporary genetic diversity, and there are many alternative explanations for this genetic pattern. In the future, we will reconstruct the phylogeny of Euploea with rapidly evolving molecular markers to determine the timing of island colonization and to assess whether dispersal events occurred during periods of lower sea level.

1 Introduction

Species diversification and the taxonomic composition of regional biota are influenced by three fundamental evolutionary processes ⎯ speciation, extinction and dispersal. However, the relative importance of dispersal has been a subject of great debate. An organism’s dispersal ability is evolutionarily labile. The dispersal hypothesis posits that long-distance dispersal usually requires organisms to survive for significant periods of time in environments that are unusual, or even unfavorable as compared to their typical habitats. In contrast, the isolation hypothesis suggests that dispersal is prevented by biological and physical barriers if the organisms are unable to survive in environments. Under this hypothesis, organisms remain in isolated areas despite land connectivity. This isolation would result in genetic divergence of organisms under different selective regimes in the absence of gene flow (Lomolino et al., 2010).

Barriers are often taxon-specific. The effectiveness of these barriers for preventing dispersal depends on both changes in environments and the characteristics of organisms. Many ancient, transient connections have served as corridors for exchange among terrestrial biota (Lomolino et al., 2010). During the Last Glacial Maximum (LGM) in Southeast Asia, temperatures were ~ 2-6 ˚C lower, and the climate was drier and more seasonal than today. At times, sea levels were lowered by at least 100 meters because of global water sequestration by Pleistocene glaciers, connecting many terrestrial regions that are now separated by oceanic barriers. The Malay Peninsula, Sumatra, and Borneo formed a single large landmass at times of lowered eustatic sea level. Heaney (1991) proposed that a continuous north-south corridor of open savannah from the Malay Peninsula to Java during the LGM divided perhumid rainforest in Sumatra/peninsular Malaysia and Borneo (Bird et al., 2005). Recent phylogeographic and biogeographic studies on vertebrates like Yellow-vented Bulbul (Lohman et al., 2010; Lohman et al., 2011) and terrestrial mammals, such as the short-tailed mongoose (Patou et al., 2009) showed that populations in Sumatra and Borneo are genetically distinct. This suggests that the populations were isolated between western and eastern Sundaland due to emergent land may in the form of a dry savanna. However, terrestrial vertebrates that favor more open habitats, such as the small Indian civet and Javan mongoose (Jennings &

2 Veron, 2011) may have been able to migrate across the land bridges of the Sunda Shelf during the LGM (Bird et al., 2005). However, this hypothesis for the existence of the savannah corridor has remained controversial. Pollen records, phylogeographic studies, distributional models, and other data conflict about its existence (Lohman et al., 2011).

Butterflies have been used as biological indicators of climate change because the life cycle of butterflies is relatively short, they can react rapidly to changes in climate (Thomas, 2005). Fluctuating climate would have a profound impact on the survival of butterflies. Therefore, adaptive mechanisms may have evolved to respond to unfavorable environmental stress. Butterflies are not only useful as biological indicators, but the species richness of butterflies is frequently used as a surrogate for the richness of other terrestrial invertebrates as well. They are more responsive than other insect groups or birds to external environmental factors (Thomas, 2005).

Euploea is one of the most widespread butterfly genera in Southeast Asia. This genus comprises more than 60 species. Many species are endemic to particular islands, but several species are widespread on several landmasses. Euploea mulciber, a tropical species, is widely distributed from India to the Philippines and southern China (Ackery & Vane-Wright, 1984). The exceptional richness and the wide distribution of this species in the tropics make it a model for investigating temporal and geographic patterns of diversification. The availability of data on its geographic distribution also makes E. mulciber as suitable for ecological niche modeling studies.

This study reconstructs the potential distribution of E. mulciber during the LGM and compares these predictions with an intraspecific phylogeny of the species to assess whether the predicted habitat disjunction was a dispersal barrier to E. mulciber between Borneo and Sumatra despite terrestrial connectivity during the LGM.

3 Method and Materials

Molecular procedures

Specimens of 16 E. mulciber were collected across the islands of Southeast Asia. Some of the specimens were preserved dry while others were preserved in 95% ethanol. Genomic DNA of butterflies was extracted from three legs of dried specimens and from the abdomen of fresh specimens with a Qiagen DNeasy Extraction kit. Since legs desiccate quickly after butterflies die, DNA of butterflies from legs is usually well preserved.

Primer names, sequences, lengths and annealing temperatures are provided in Table 1. Some markers were amplified in smaller segments because they were too long to sequence with a single read. All primers were concatenated with nondegenerate, nonhomologous 5’ tails, which improved PCR success (Wahlberg & Wheat, 2008). All PCR amplifications were performed in 20 µL reactions, including 1 µL of gDNA template. The PCR protocol was 95˚ C for 3 min, followed by 40 cycles of 94˚ C for 3 min, 55˚ C for 50 sec and 72˚ C for 90 sec, followed by a final extension period of 72˚ C for 5 min. The amplified double-stranded PCR products were visualized on 1% agarose gels; 4 µL of the PCR product was mixed with SybrGreen stain and loading dye to run the gel. Successfully amplified PCR products were cleaned with SureClean (Bioline sent to GENEWIZ (genewiz.com) for sequencing in both directions.

Phylogenetic Analysis

DNA sequences were aligned with Sequencher 4.10.1 (genecodes.com) and the multi-gene dataset was concatenated with Sequence Matrix (Vaidya et al., 2011). No insertions or deletions (indels) were detected in any protein encoding genes. Bayesian phylogenetic analyses were performed with MrBayes 3.1.2 (Ronquist & Huelsenbeck, 2003). The TPM + gamma and TrN + gamma models were chosen by jModeltest 0.1. (Posada 2008). However, because these models are not implemented in MrBayes 3.1.2, the GTR + gamma model was used for the concatenated data. Two runs searched simultaneously for 8 million generations, and trees were sampled every 1000 generations.

4 Changes in the posterior probabilities of up to twenty splits were plotted over the generations of the analysis with the computer program “Are We There Yet?” (Nylander et al., 2008) to assess whether the chains had converged. After completion of each analysis, the first 25% of the sampled trees were discarded as burn-in before a majority- rule consensus tree was calculated from the remaining trees.

Potential Distribution Reconstruction

Occurrence Records Collection

Occurrence data of E. mulciber were obtained from the Global Biodiversity Information Facility (GBIF) (www.gbif.org). The x and y coordinates corresponded to the latitude and longitude of the data points, respectively. These data and the global country boundaries shape file (www.diva-gis.org) were imported into the program DIVA- GIS to create a shape file comprising a map of Southeast Asia with collection localities marked.

Environmental layers

Current environmental layers were downloaded at 2.5 arc-minutes (about 5 km pixels) resolution from the Worldclim database (www.worldclim.org/current). Nineteen Bioclim environmental variables (Table 2) summarize the minimum, mean and maximum values of temperature and precipitation, which could affect species survival (Hijmans et al., 2005). The environmental layers of the LGM were also downloaded from Worldclim (www.worldclim.org/past).

Modeling

MaxEnt 3.33k (Phillips et al., 2006) used 126 present-day records of the species to predict its geographic distribution based on the theory of maximum entropy. The model was developed and used to compare the present day distribution of the species to its potential distribution during the LGM. In order to avoid bias, each model was repeated 5 times with randomly selected presence points. The Area Under the receiver

5 operating characteristic Curve (AUC) was used to examine model accuracy in each of 5 runs per model. AUC is used to compare the likelihood that a random presence site would have a higher predicted value in the model than a random absence site (Elith et al., 2006). Clamping grids were written while the data were projected onto LGM surfaces. Model output gave pixel values as percentages, where higher values indicate more suitable habitats. The maximum, minimum, median, and standard deviation were calculated for each of the 5 runs. A statistical analysis of the 5 runs were averaged over the model output by giving geographic range of suitable habitats for E. mulciber. The output from MaxEnt was examined with DIVA-GIS. MaxEnt results were summarized in a binary fashion (suitable or unsuitable habitat) by using the median of the Minimum Training Presence and the Equal Training Sensitivity and Specificity of the 5 runs to make a single threshold for the MaxEnt predicted current and projected models (Fig. 3).

Results

The mean AUC value for the current ecological niche model of E. mulciber was moderately high (mean AUC = 0. 882, SD = 0. 0014), indicating that our model has a high predictive power for habitat suitability. The logistic threshold for each run is summarized in Table 3. The logistic threshold of minimum training presence and equal training sensitivity plus specificity were consistent from all runs. P-values for each run were less than 0.05. Table 4 shows the four environmental variables that contributed the most information to the MaxEnt model, along with their percent contribution and permutation importance. Environmental layer Bio7 (temperature annual range) contributed 26.5%, and Bio 4 (temperature seasonality) contributed 23.6% to the model. Response curves (Fig. 1) demonstrate how each environmental variable affected the MaxEnt prediction. The red line represents the behavior of changing values and the blue area is the standard deviation for the data points. The environmental layers Bio7 and Bio 4 showed that the response was high for relatively small values, and quickly dropped towards 0 when it reached higher values. Visual comparison of outputs from MaxEnt showed differences in the

6 geographical distributions of E. mulciber in the LGM compared to the present-day (Fig. 2). Clamping grids were used to inspect whether the bioclim variables were outside the range of present conditions in the LGM projected model. While projecting the MaxEnt model onto the environmental variables of LGM, the values for absolute difference in predictions when using vs. not using clamping in all five runs were small, indicating that most of the bioclim variables in the projected LGM model are inside the range of present- day condition on our study region. Thus, they had little effect on predicted suitability (Elith & Leathwick, 2009). All colored areas were defined as suitable habitats by multiple runs. Regions in red represent the most suitable habitats for the species, and regions in yellow and green are fair or unsuitable habitats, respectively. The distributional models showed similar patterns of the species distribution at the LGM and present-day. The Malay Peninsula, Sumatra, Java, Western Borneo, Sulawesi and Eastern New Guinea were likely suitable places for the species. However, despite terrestrial connectivity in the past between what are currently islands, central Sundaland did not contain suitable habitats for E. mulciber butterflies. The trees derived from the analysis of six protein-encoding genes of 15 specimens are presented in Fig. 4. The color of taxon corresponds to the color of the localities shown on Fig. 3. Absence of values on the nodes of the phylogram from the Bayesian method indicates branches supported with less than 0.5. In separate analyses of mitochondrial (Fig. 4B), nuclear (Fig. 4C), and combined (Fig. 4A) datasets, no geographically definable contemporary populations were monophyletic in our molecular analysis.

Discussion

Despite a hypothesized “savannah corridor” dividing the eastern and western halves of Sundaland during the LGM and niche model predictions supporting this separation, there is no contemporary east-west differentiation in genetic diversity across the Sunda Shelf. Moreover, no geographically defined populations are recovered as monophyletic using any subset of the data.

7 The lack of monophyly of populations from Borneo and western Sundaland (Malay Peninsula/Sumatra) was shown in all three phylogenetic trees (mtDNA, nDNA, and mt + nDNA; Fig. 4). This result is surprising, and could result from several factors. Firstly, although the model indicates a yellow corridor of less-than-ideal conditions during the LGM (Fig. 2), this area still may have been traversed. Another possibility is that hallmarks of past isolation have been obliterated by more recent dispersal. Finally, the species is not as habitat-specific as previously assumed, but Euploea species usually have very strict environmental requirements, such as preferred host plant which only occurs during the wet season (Canzano et al., 2006). Their ecological needs suggest that they may not be able to survive in open dry grassland, and these data may thus argue against the hypothesis of a savannah corridor. The savannah-corridor hypothesis is based on the climatic conditions during the LGM, in which sea levels were low, and the climate was drier and more seasonal. However, these conditions existed for only “17% of time during the last 250,000 years”, making the savannah hypothesis less likely to explain observed biogeographic differences (Slik et al., 2011). It is argued that the biogeographic difference between western and eastern Sundaland might be related to the soil conditions of exposed sea floor in central Sundaland. Since the species has been recorded from lowland elevations up to 2000 m in montane areas, tropical forest restricted to mountainous regions of western Sumatra and central Borneo might have acted as refugia for the species during the LGM (Meijaard, 2003). Furthermore, some model projections predicted that there existed a belt of evergreen forest from east to west, bisecting the northern and southern savannah in central Sundaland at the LGM (Cannon et al., 2009). The phylogeographic study of Lithocarpus based on chloroplast DNA sequence also supported the continuous presence of tropical rain forest in Southeast Asia throughout its evolutionary history (Cannon & Manos, 2003). Under this hypothesis, a dispersal route would have existed for forest- dependent species to move between Sumatra to Borneo, and this would better explain our results. However, the historical events at the LGM might not have left a detectable genetic signal (genetic variation within the species) since the genetic markers selected may be too conserved over time. In the future, it is important to utilize markers with fast

8 evolving rate as a reliable molecular chronometer to reveal the specific biological and geological events during LGM. Fluctuations in temperature and sea level play an important role in shaping current distribution patterns and biodiversity in non-volant mammals, and even in birds (Woodruff, 2010; Lohman et al., 2011). Because of the extensive shallow exposed land area in the central Sundaland during the LGM, land connectivity allowed terrestrial animals to migrate across the land bridges of the Sunda Shelf during the LGM while blocked forest-dependent species to disperse based on strong genetic differentiation between population of eastern and western Sundaland. Dispersal barriers allowed divergence of the isolated populations to occur. However, if Euploea butterflies regularly fly among islands, the timing of colonization would not necessarily be associated with periods of lowered sea level (Outlaw & Voelker, 2008). Despite representing a minority of terrestrial diversity, vertebrates are strongly over-represented in biogeographic studies. The strong dispersal ability of butterflies—either by strong winds or their own powered flight—calls into question the generality of patterns documented only from vertebrates.

9 References

Ackery, P.R. & Vane-Wright, R.I. (1984) Milkweed Butterflies. British Museum (Natural History), London. Bird, M.I., Taylor, D. & Hunt, C. (2005) Palaeoenvironments of insular Southeast Asia during the Last Glacial Period: a savanna corridor in Sundaland? Quaternary Science Reviews, 24, 2228-2242. Cannon, C.H. & Manos, P.S. (2003) Phylogeography of the Southeast Asian stone oaks (Lithocarpus). Journal of Biogeography, 30, 211-226. Cannon, C.H., Morley, R.J. & Bush, A.B.G. (2009) The current refugial rainforests of Sundaland are unrepresentative of their biogeographic past and highly vulnerable to disturbance. Proceedings of the National Academy of Sciences of the United States of America, 106, 11188-11193. Canzano, A.A., Krockenberger, A.A., Jones, R.E. & Seymour, J.E. (2006) Rates of metabolism in diapausing and reproductively active tropical butterflies, Euploea core and Euploea sylvester (Lepidoptera: Nymphalidae). Physiological Entomology, 31, 184-189. Elith, J. & Leathwick, J.R. (2009) Species distribution models: ecological explanation and prediction across space and time. Annual Review of Ecology, Evolution, and Systematics, 40, 677-697. Elith, J., H. Graham, C., P. Anderson, R., Dudík, M., Ferrier, S., Guisan, A., J. Hijmans, R., Huettmann, F., R. Leathwick, J., Lehmann, A., Li, J., G. Lohmann, L., A. Loiselle, B., Manion, G., Moritz, C., Nakamura, M., Nakazawa, Y., McC. M. Overton, J., Townsend Peterson, A., J. Phillips, S., Richardson, K., Scachetti- Pereira, R., E. Schapire, R., Soberón, J., Williams, S., S. Wisz, M. & E. Zimmermann, N. (2006) Novel methods improve prediction of species’ distributions from occurrence data. Ecography, 29, 129-151. Heaney, L.R. (1991) A synopsis of climatic and vegetational change in Southeast Asia. Climatic Change, 19, 53-61. Jennings, A.P. & Veron, G. (2011) Predicted distributions and ecological niches of 8 civet and mongoose species in Southeast Asia. Journal of Mammalogy, 92, 316- 327. Lohman, D.J., de Bruyn, M., Page, T., von Rintelen, K., Hall, R., Ng, P.K.L., Shih, H.-T., Carvalho, G.C. & von Rintelen, T. (2011) Biogeography of the Indo-Australian Archipelago. Annual Review of Ecology, Evolution, and Systematics, 42, 205-226. Lohman, D.J., Ingram, K.K., Prawiradilaga, D.M., Winker, K., Sheldon, F.H., Moyle, R.G., Ng, P.K.L., Ong, P.S., Wang, L.K., Braile, T.M., Astuti, D. & Meier, R. (2010) Cryptic genetic diversity in ‘‘widespread” Southeast Asian bird species suggests that Philippine avian endemism is gravely underestimated. Biological Conservation, 143, 1885-1890. Lomolino, M.V., Riddle, B.R., Whittaker, R.J. & Brown, J.H. (2010) Biogeography. pp. 167-206. Sinauer Association., Sunderland, MA. Nylander, J.A.A., Wilgenbusch, J.C., Warren, D.L. & Swofford, D.L. (2008) AWTY (are we there yet?): a system for graphical exploration of MCMC convergence in Bayesian phylogenetics. Bioinformatics, 24, 581-583.

10 Outlaw, D.C. & Voelker, G. (2008) Pliocene climatic change in insular Southeast Asia as an engine of diversification in Ficedula flycatchers. Journal of Biogeography, 35, 739-752. Patou, M.-L., McLenachan, P.A., Morley, C.G., Couloux, A., Jennings, A.P. & Veron, G. (2009) Molecular phylogeny of the Herpestidae (Mammalia, Carnivora) with a special emphasis on the Asian Herpestes. Molecular Phylogenetics and Evolution, 53, 69-80. Phillips, S.J., Anderson, R.P. & Schapire, R.E. (2006) Maximum entropy modeling of species geographic distributions. Ecological Modelling, 190, 231-259. Posada, D. 2008. jModelTest: phylogenetic model averaging. Molecular Biology and Evolution, 25,1253-1256. Ronquist, F. & Huelsenbeck, J.P. (2003) MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics, 19, 1572-1574. Slik, J.W.F., Aiba, S.-I., Bastian, M., Brearley, F.Q., Cannon, C.H., Eichhorn, K.A.O., Fredriksson, G., Kartawinata, K., Laumonier, Y., Mansor, A., Marjokorpi, A., Meijaard, E., Morley, R.J., Nagamasu, H., Nilus, R., Nurtjahya, E., Payne, J., Permana, A., Poulsen, A.D., Raes, N., Riswan, S., van Schaik, C.P., Sheil, D., Sidiyasa, K., Suzuki, E., van Valkenburg, J.L.C.H., Webb, C.O., Wich, S., Yoneda, T., Zakaria, R. & Zweifel, N. (2011) Soils on exposed Sunda Shelf shaped biogeographic patterns in the equatorial forests of Southeast Asia. Proceedings of the National Academy of Sciences, 108, 12343-12347. Thomas, J.A. (2005) Monitoring change in the abundance and distribution of insects using butterflies and other indicator groups. Philosophical Transactions of the Royal Society B: Biological Sciences, 360, 339-357. Vaidya, G., Lohman, D.J. & Meier, R. (2011) SequenceMatrix: concatenation software for the fast assembly of multi-gene datasets with character set and codon information. Cladistics, 27, 171-180. Wahlberg, N. & Wheat, C.W. (2008) Genomic outposts serve the phylogenomic pioneers: designing novel nuclear markers for genomic DNA extractions of Lepidoptera. Systematic Biology, 57, 231 - 242. Woodruff, D. (2010) Biogeography and conservation in Southeast Asia: how 2.7 million years of repeated environmental fluctuations affect today’s patterns and the future of the remaining refugial-phase biodiversity. Biodiversity and Conservation, 19, 919-941.

11 Table 1. Primers and associated annealing temperatures used in this study.

Gene Annealing Fragments F/R Length Primer Name Primer Sequence 5' to 3' Temperature cyt b F 579 REVC-B2H-tF TAA TAC GAC TCA CTA TAG GGT GAG GAC AAA TAT CAT TTT GAG GW 53˚ C R REVC-BJ-tR ATT AAC CCT CAC TAA AGA CTG GTC GAG CTC CAA TTC ATG T COIa F 764 LCO1490-tF TAA TAC GAC TCA CTA TAG GGG GTC AAC AAA TCA TAA AGA TAT TGG 53˚C R Nancy-tR ATT AAC CCT CAC TAA AGC CCG GTA AAA TTA AAA TAT AAA CTT C COIb F 660 COIb-tF1 TAA TAC GAC TCA CTA TAG GGC AAC AYT TAT TTT GAT TYT TTG GYC 53˚C R COIb-tR2 ATT AAC CCT CAC TAA AGC GWC GAG GTA TTC CWG CTA AAC C EF1α 1&2 F 800 ef44-tF TAA TAC GAC TCA CTA TAG GGG CYG ARC GYG ARC GTG GTA TYA C 60˚C R ef52. 6-tR ATT AAC CCT CAC TAA AGG CYT CGT GGT GCA TYT CSA C EF1α 3 F 591 EF51. 9-tF TAA TAC GAC TCA CTA TAG GGC ARG ACG TAT ACA AAA TCG G 60˚C R EFrcM4-tR ATT AAC CCT CAC TAA AGA CAG CVA CKG TYT GYC TCA TRT C wg F 454 Wg1n-tF TAA TAC GAC TCA CTA TAG GGC GGA GAT GCG MCA GGA RTG C 50˚C R Wg2n-tR ATT AAC CCT CAC TAA AGC TTT TTC CGT SCG ACA CAG YTT GC RpS2 F 448 RpS2-tF TAA TAC GAC TCA CTA TAG GGA TCW CGY GGT GGY GAT AGA G 55˚C R RpS2-tR ATT AAC CCT CAC TAA AGA TGR GGC TTK CCR ATC TTG T RpS5 F 690 RpS5-tF TAA TAC GAC TCA CTA TAG GGA TGG CNG ARG ARA AYT GGA AYG A 55˚C R RpS5-tR ATT AAC CCT CAC TAA AGC GGT TRG AYT TRG CAA CAC G

12 Table 2. Environmental variables from worldclim.org used in the construction of the ecological niche models BIO1 Annual Mean Temperature BIO2 Mean Diurnal Range (Mean of monthly (max temp - min temp)) BIO3 Isothermality (BIO2/BIO7) (* 100) BIO4 Temperature Seasonality (standard deviation *100) BIO5 Max Temperature of Warmest Month BIO6 Min Temperature of Coldest Month BIO7 Temperature Annual Range (BIO5- BIO6) BIO8 Mean Temperature of Wettest Quarter BIO9 Mean Temperature of Driest Quarter BIO10 Mean Temperature of Warmest Quarter BIO11 Mean Temperature of Coldest Quarter BIO12 Annual Precipitation BIO13 Precipitation of Wettest Month BIO14 Precipitation of Driest Month BIO15 Precipitation Seasonality (Coefficient of Variation) BIO16 Precipitation of Wettest Quarter BIO17 Precipitation of Driest Quarter BIO18 Precipitation of Warmest Quarter BIO19 Precipitation of Coldest Quarter

13 Table 3. Values of each run for minimum training presence and equal training sensitivity plus specificity Logistic Description Run threshold Mean/Median P-value 0 0.025 2.62E-04 1 0.148 1.78E-07 Minimum training 2 0.03 0.0245 2.45E-04 presence 3 0.024 3.85E-04 4 0.024 3.69E-04 0 0.346 1.65E-05 1 0.327 6.78E-09 Equal training 2 0.36 0.3386 4.64E-07 sensitivity plus specificity 3 0.325 9.86E-07 4 0.335 8.72E-08

Table 4. The percentage contribution of each environmental variable to the MaxEnt models

Percentage Permutation Variable Contribution Importance

bio7 26.5 37.9 bio4 23.6 22.5 bio17 11.1 10.2 bio14 10.9 1.1

Figure 1. The response curves of environmental layers, BIO 7 and BIO 4. The curves show how the logistic prediction changes as each environmental parameter varies.

15 Euploea mulciber

0 - 0.0245 0.0245 - 0.3368 0.3368 - 0.9999 No Data

Figure 2. Potential geographical distributions of Euploea mulciber during the Last Glacial Maximum (top) and the present-day (bottom). The color gradient indicates the most suitable habitats for the species, ranging from 0 - 1: least suitable to most suitable.

E u p l o e a

m u l c i b e r

0 - 0 . 0 2 4 5

Figure 3. Map of Southeast Asia with major islands labeled. The colors of the taxon labels 0 in Fig. 4 correspond to the colors on this map. . 0 2 4 5

0 . 3 3 6

0 . 3 3 17 6 8 -

0 . 9 (A)

(B) (C)

Figure 4. Bayesian phylogenies of Euploea mulciber (A) based on combined nuclear and mitochondrial data, (B) based on mitochondrial DNA only, and (C) based on nuclear DNA alone. Numbers above braches denote posterior probabilities.