Genes Genet. Syst. (2016) 91, p. 209–215Divergence times of the giant rhinoceros beetles 209 Cretaceous origin of giant rhinoceros beetles (Dynastini; Coleoptera) and correlation of their evolution with the Pangean breakup Haofei Jin1, Takahiro Yonezawa1,2*, Yang Zhong1, Hirohisa Kishino3 and Masami Hasegawa1,2 1School of Life Sciences, Fudan University, SongHu Rd. 2005, Shanghai 200438, China 2Institute of Statistical Mathematics, Midori-cho 10-3, Tachikawa, Tokyo 190-8562, Japan 3Graduate School of Agricultural and Life Sciences, The University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-8657, Japan (Received 8 January 2016, accepted 30 April 2016; J-STAGE Advance published date: 23 August 2016) The giant rhinoceros beetles (Dynastini, Scarabaeidae, Coleoptera) are distributed in tropical and temperate regions in Asia, America and Africa. Recent molecular phylogenetic studies have revealed that the giant rhinoceros beetles can be divided into three clades representing Asia, America and Africa. Although a correlation between their evolution and the continental drift during the Pangean breakup was suggested, there is no accurate divergence time estimation among the three clades based on molecular data. Moreover, there is a long chronological gap between the timing of the Pangean breakup (Cretaceous: 110–148 Ma) and the emergence of the oldest fossil record (Oligocene: 33 Ma). In this study, we estimated their divergence times based on molecular data, using several combinations of fossil calibration sets, and obtained robust estimates. The inter-continental divergence events among the clades were estimated to have occurred about 99 Ma (Asian clade and others) and 78 Ma (American clade and African clade), both of which are after the Pangean breakup. These estimates suggest their inter-continental divergences occurred by overseas sweepstakes dispersal, rather than by vicariances of the population caused by the Pangean breakup. Key words: Dynastinae, molecular phylogeny, time tree, fossil calibrations attempted to reconstruct their phylogenetic tree with INTRODUCTION extensive taxon sampling until recently (Rowland and The tribe Dynastini, known as the giant rhinoceros Miller, 2012). beetles, includes some of the largest species within the Rowland and Miller (2012) inferred the phylogenetic Coleoptera, which is the most species-rich order among all tree on the basis of morphological characters as well living organisms. These species show remarkable sexual as molecular data including mitochondrial and nuclear dimorphism, such as very long horns in the male’s head gene sequences. They indicated that this tribe can and pronotum that are usually absent in females (Endrödi, be divided into three subtribes, Dynastina (Dynastes, 1985). Naturalists have been attracted for centuries by Golofa, Megasoma, Augosoma), Xylotrupina (Allomyrina, such marked morphological variety and excess male orna- Xylotrupes, Trypoxylus, Xyloscaptes) and Chalcosomina mentation (Rowland and Miller, 2012). It is important to (Haploscapanes, Chalcosoma, Beckius, Eupatorus, clarify how this sexual ornamentation evolved, to gain a Pachyoryctes) (Supplementary Table S1). The first sub- better understanding of natural selection and especially tribe consists of the African clade (Augosoma) and the sexual selection. To elucidate this issue, reconstruction New World clade (Dynastes, Golofa, Megasoma). On the of a reliable phylogenetic tree is prerequisite. Never- other hand, the latter two subtribes are distributed in theless, no molecular studies of the Dynastini had been Asia and Oceania, and constitute a large clade (here- after called the Asian clade). From this categoriza- Edited by Hidenori Nishihara tion, Rowland and Miller (2012) suggested a correlation * Corresponding author. E-mail: [email protected] Abbreviations: ML (maximum likelihood), Ma (mega annum), between Dynastini evolution and the Pangean breakup in OKR (oldest known record) the Early Cretaceous. DOI: http://doi.org/10.1266/ggs.16-00003 Indeed, the correlations between continental drift and 210 H. JIN et al. phylogenetic branching patterns of terrestrial animals tide substitutions among genes and codon positions. The such as Eutherian mammals (Nishihara et al., 2009; dos best partition was found by the PartitionFinder program Reis et al., 2012), Palaeognathae (ratite and tinamou: (Lanfear et al., 2012). The best partition is shown in Cooper et al., 2001; Haddrath and Baker, 2012; Yonezawa et Supplementary Table S2, and it was applied for both the al., 2017), Pleurodira (side-necked turtle: Vargas-Ramírez phylogenetic inference and the divergence time estima- et al., 2008), Anura (frogs: Bossuyt et al., 2006), Cichlidae tions under the GTR+Γ model. The phylogenetic tree (cichlid fish: Friedman et al., 2013) and Plecoptera was inferred by the maximum likelihood (ML) method (stoneflies: McCulloch et al., 2016) have been vigorously using the RAxML program ver. 8.0.20 (Stamatakis et al., argued. Some of these studies (e.g., Haddrath and Baker, 2008). The rapid bootstrap method (Stamatakis et al., 2012; dos Reis et al., 2012; Yonezawa et al., 2017) sug- 2008) was applied to evaluate the confidence of the inter- gested that biogeographical distribution patterns were not nal branches with 1000 replications. Divergence times necessarily established merely by continental drift, but were estimated by the Bayesian relaxed clock method that the overseas ‘sweepstakes dispersal route’ (Simpson, (Thorne et al., 1998) with the MCMCTREE program 1940) should be considered. A sweepstakes dispersal implemented in PAML ver. 4.7 (Yang, 2007; Inoue et al., route is “a possible route of faunal interchange which is 2010). To reduce the computational burden, the normal unlikely to be used by most animals, but which will, by approximation method was applied. The autocorrelated chance, be used by some” (Allaby, 1999). Since we focus rates model (Thorne et al., 1998) and the independent on the establishment of intercontinental distribution pat- model (Rannala and Yang, 2007) were used for the drift terns, hereafter we call them ‘overseas sweepstakes dis- model of the evolutionary rates along the branches. The persal’ as an alternative to the hypothesis that vicariance prior distributions of the root rates were set to 4 and 2.26 was caused by the Pangean breakup. for the α and β parameters of Γ distribution, respectively, In addition, no paleontological evidence supports the and the parameters of σ2 were set to 1 and 0.2. Other hypothesis that the breaking up of Pangea triggered the parameters such as the Birth and Death rate were set to speciation of the subfamily Dynastinae because their default. The condition of the MCMC was as follows: the fossil record can be traced back to after the Oligocene first 200,000 generations were discarded as burnin; then, (Krell, 2000, 2007). Therefore, reliable divergence times 20,000 trees were sampled per 200 generations. within the Dynastini based on molecular data are impor- The history of morphological character traits was tant for a better understanding of the correlation between inferred using MESQUITE ver. 3.3 (http://mesquiteproject. their evolution and geological events such as continen- org) with the maximum parsimony method on the basis of tal drift. Here, we present the first clear picture of the the 17 characters reported by Rowland and Miller (2012). time-scale of giant rhinoceros beetle evolution and its cor- relation with geological events by detailed reanalyses of Fossil calibrations Reliable fossil records of the family Rowland and Miller (2012)’s molecular and morphological Scarabaeidae (Krell, 2000, 2007) were used as calibration data. points. The age of the oldest known record (OKR) of a fossil in a particular crown group does not necessarily correspond to the exact timing of its emergence because MATERIALS AND METHODS of the incompleteness of the fossil record. Therefore, the Phylogenetic inference and divergence time estima- emergence timing of the crown group should be assumed tions Detailed procedures for tree inference and diver- between the OKR of crown taxa and the OKR of the stem gence time estimation were as follows. The nucleotide taxa (Yonezawa et al., 2007). Here we suggest the cali- sequence data published by Rowland and Miller (2012) brations of divergence times (maximum and minimum (cytochrome c oxidase II, 16S rRNA, histone III and arginine boundaries of the nodes) in the Dynastini based on this kinase) were retrieved from NCBI. Rowland and Miller idea. The first calibration is the split between the sub- (2012)’s data cover all genera among the tribe Dynastini. families Dynastinae and Melolonthinae (node 2 in Fig. In addition, the corresponding nucleotide sequences of 1). The subfamily Dynastinae (Dynastini species, Orizabus Tribolium castaneum (family Tenebrionidae) were also and Cyclocephala) is closer to the subfamily Cetoniinae included as an outgroup. The accession numbers are pre- (Cotinus mutabilis) than to the subfamily Melolonthinae sented in Supplementary Table S1. Retrieved sequences (Polyphylla decemlineata). This relationship is sup- were automatically aligned by the MAFFT program (Katoh ported by morphological (Krell, 2000) and molecular data and Standley, 2013), and carefully checked by eye. The (Rowland and Miller, 2012, this study). The reliable impact of the partition model on both tree inference and OKR of Melolonthinae (Cretomelolontha transbaikalica: divergence time estimation was discussed in detail in our Fossil 1 in Fig. 1)