New Mitogenomes of Two Chinese Stag Beetles (Coleoptera, Lucanidae) and Their Implications for Systematics.

Although conspicuous and well-studied, stag beetles have been slow to join the genomic era. In this study, mitochondrial genomes of two stag beetles, Sinodendron yunnanense and Prosopocoilus confucius, are sequenced for the first time. Both of their genomes consisted of 13 protein-coding genes (PCGs), 22 transfer RNA genes (tRNAs), 2 ribosomal RNAs (rRNAs), and a control region. The mitogenome of S. yunnanense was 16,921 bp in length, and P. confucius was 16,951 bp. The location of the gene trnL(UUR), between the A + T-rich and control region in S. yunnanense, is the first observed in Lucanidae. In P. confucius, an unexpected noncoding region of 580 bp was discovered. Maximum likelihood and Bayesian inference on the 13 mitochondrial PCGs were used to infer the phylogenetic relationships among 12 representative stag beetles and three scarab beetles. The topology of the two phylogenetic trees was almost identical: S. yunnanense was recovered as the most basal Lucanid, and the genus Prosopocoilus was polyphyletic due to P. gracilis being recovered sister to the genera Dorcus and Hemisodorcus. The phylogenetic results, genetic distances and mitogenomic characteristics call into question the cohesion of the genus Prosopocoilus. The genetic resources and findings herein attempts to redress understudied systematics and mitogenomics of the stag beetles.

Lucanidae (stag beetles) is a small family in Scarabaeoidea, containing about 1,800 described species and subspecies worldwide (Krajcik 2001, Bartolozzi and Sprecher-Uebersax 2006, Fujita 2010). Most studies support Lucanidae as a monophyletic group and an early diverging lineage in scarab beetles (Howden 1982, Scholtz et al. 1994, Ahrens et al. 2014) although a recent study has questioned Lucanid monophyly (Mckenna et al. 2015). There is a robust body of literature on taxonomy of this family (Arrow 1950, Didier and Séguy 1953, Benesh 1960, Lawrence and Newton 1995, Krajcik 2001, Holloway 2007), with particular interest in sexual selection behavior and male polymorphism. As typical, molecular markers have more recently been used to enrich stag beetle systematics. Hosoya and Araya (2005) reconstructed the first molecular phylogeny of stag beetles (based on the molecular markers cox1 and rrnL) although using a small data set of only Japanese species. There has been particular attention to species delimitation and conservation of lucanids from Japan and Taiwan Island, partly because these beetles are popular as pets in these places (Huang and Lin 2010, Kubota et al. 2011, Lin et al. 2011, Tsai et al. 2014, Tsai and Yeh 2016). Kim and Farrell (2015) performed the first comprehensive phylogenetic study, and revealed a Gondwanan origin of Darwin’s stag beetle.

Undoubtedly these studies have shed light on the evolution of Lucanidae although using short gene fragments. Further, although widely used in phylogenetics, phylogeography and population genetics in insects (Timmermans et al. 2010, Cameron 2014a, Yuan et al. 2016), only three complete mitogenomes have been reported for lucanids; Lucanus mazuma (Sheffield et al. 2009), Prosopocoilus blanchardi (Kim et al. 2015), and Prosopocoilus gracilis (Wu et al. 2016, our recent work).

To address these insufficiencies, we herein report complete mitogenomes of two representatives from the Chinese fauna, Sinodendron yunnanense and Prosopocoilus confucius. In Sinodendron, four species are recognized worldwide; three are distributed in the Palearctic and Nearctic, whereas S. yunnanense is oriental, with a restricted habitat in Southwest China and quite rare in the field. The species P. confucius is a large but otherwise typical member of the genus, also with limited distribution in South China. We aim to create a more robust understanding of the evolution of lucanid beetles by developing mitochondrial phylogenomics, with a particular focus on Prosopocoilus, including a comparison among P. confucius, P. blanchardi and P. gracilis.

Materials and Methods

Sample Collection and DNA Extraction

The specimen of S. yunnanense was collected by Dr. Hong-liang Shi from Lanping, Yunnan, China, in July 2012, and the specimen of P. confucius was collected from Wuming, Guangxi, China, by Ying-Hua Wu in July 2011. Total genomic DNA was extracted from the muscle of a single S. yunnanense using the Qiagen DNAeasy Kit. For P. confucius, a small portion of the muscle from a specimen preserved in 100% ethanol was used for total DNA extraction with the Blood & Tissue Kit (Qiagen, Germany). The voucher specimens (S. yunnanense: MAHUSi012; P. confucius: MAHU00102) were deposited at the Museum of Anhui University, China, and the sequences were submitted to GenBank with the accession numbers KP735804 (S. yunnanense) and KP552119 (P. confucius).

Primer Design, Polymerase Chain Reaction Amplification, and Sequencing

Complete mitogenomes were assembled from amplified fragments, with all primers used for amplification listed in Table 1. PCR amplification reactions were carried out in 25 μl volumes containing 10 μM of each primer (forward and reverse) 1 μl, 2 μl template DNA, 12.5 μl 2 × EasyTaq SuperMix (+dye), and 8.5 μl sterile double-distilled water to make up a final volume of 25 μl. The polymerase chain reaction amplifications were performed under the following conditions: an initial denaturation at 94 °C for 2 min, followed by 35–37 cycles of denaturation at 94 °C for 40 seconds, annealing at 52–58 °C for 50 seconds, and elongation at 70 °C for 1 min, and then a final extension step at 72 °C for 7 min. The temperature of annealing was determined by the length of fragments. Sequencing was conducted with the Illumina HiSeq 2000 platform. Cluster strands created by bridge amplification were primed and all four fluorescently labeled, and 3-OH blocked nucleotides were added to the flow cell with DNA polymerase. The cluster strands were extended in single nucleotides. Following the incorporation step, the unused nucleotides and DNA polymerase molecules were washed away, a scan buffer added to the flow cell, then the optics system scanned each lane of the flow cell in imaging units (tiles). Once imaging was completed, chemicals that effect cleavage of the fluorescent labels and the 3-OH blocking groups were added to the flow cell, which prepares the cluster strands for another round of fluorescent nucleotide incorporation.

Table 1

Details on primers used in this study

Gene	Primer name	Sequence (5′–3′)	Reference
cox1	COI–F1	CAACATTTATTTTGATTTTTTGG	Simon et al. 1994
	COI–R1	TCCAATGCACTAATCTGCCATATTA	Simon et al. 1994
cytb	Cytb–F2	GAGGAGCAACTGTAATTACTAA	Balke et al. 2004
	Cytb–R2	AAAAGAAARTATCATTCAGGTTGAAT	Balke et al. 2004
rrnL	16S–F1	CCGGTTTGAACTCAGATCATG	Hosoya et al. 2001
	16S–R1	TAATTTATTGTACCTTGTGTATCAG	Hosoya et al. 2001
nad4	PcND4F1	AATCATTGACCCTGAAACAG	This study
	PcND4R1	AGGGAGATTTCTTTGAGGTG	This study

Mitogenome Assembly, Annotation, and Analysis

The mitogenomes were assembled using SOAP denovo (BGI company, Shenzhen, China) and preliminary annotations made with the MITOS WebServer (http://mitos.bioinf.uni-leipzig.de/index.py). tRNA genes and their secondary structures were inferred using tRNAscan-SE 1.21 (http://lowelab.ucsc.edu/tRNAscan-SE/). Those not identified by tRNAscanoSE, in addition to 16S ribosomal RNA (rrnL, lrRNA), and 12S ribosomal RNA (rrnS, srRNA), were determined according to sequence similarity with related species. The protein-coding genes (PCGs) were determined by ORF Finder (http://www.ncbi.nlm.nih.gov/gorf/gorf.html) under the invertebrate mitochondrial genetic code. Nucleotide compositions, codon usage, and relative synonymous codon usage (RSCU) values of PCGs were calculated with MEGA version 6.05 (Tamura et al. 2013). PCGs were translated with DNAMAN v7.0.2.176 (Lynnon Biosoft, Vaudreuil-Dorion, Canada). Mitogenomes were mapped with CGView (Grant and Stothard 2008, available at http://stothard.afns.ualberta.ca/cgviewserver/). Composition skew analysis was conducted according to formulas AT skew = [A − T]/[A + T] and GC skew = [G − C]/[G + C] (Perna and Kocher 1995).

Phylogenetic Analyses

Phylogenetic analyses of stag beetles, including the newly sequenced S. yunnanense and P. confucius and 13 species retrieved from GenBank, were performed using a concatenated dataset of the 13 PCGs. Models of nucleotide substitution were selected according to the Akaike Information Criterion (AIC) with jModelTest v2.1.4 (Posada 2008). Phylogenetic trees were generated from ML analysis using RAxML (Stamatakis 2014) and Bayesian inference (BI) with MrBayes v3.2.5 (Huelsenbeck and Ronquist 2001), both under the GTR + I + G model. Node support in the ML tree was estimated through bootstrap analysis with 1,000 replicates. The BI was conducted with two simultaneous Markov chain Monte Carlo runs of 2 million generations, sampled every 1,000 steps, with the first 25% discarded as burn-in. Phylogenetic trees were viewed and edited in Figtree v1.4.3 Rambaut 2016.

Results and Discussion

Genome Composition and Base Content

The complete mitogenomes of S. yunnanense and P. confucius were 16,921 bp (Fig. 1), and 16,951 bp (Fig. 2), respectively. Details of the 13 PCGs, 22 tRNAs, 2 rRNAs, and control region are given in Tables 2 and 3. Their compositions were typical of the Coleoptera (Sheffield et al. 2009, Kim et al. 2015, Wu et al. 2016, Du et al. 2016, Yang et al. 2016). Of the 37 genes, 23 (9 PCGs and 14 tRNAs) genes were located on the J-strand, with the remaining 4 PCGs, 2 rRNAs, and 8 tRNAs on the N-strand. Typical of metazoan mitogenomes (Cameron 2014a), the single tRNA gene trnS (AGN) lacked the cloverleaf secondary structure, due to the structure of the dihydrouridine arm. The lengths of the tRNAs ranged from 61 to 71 nucleotides in both species. In both species lrRNA was located between trnL (CUN) and trnV, and was 1,278 bp in S. yunnanense, and 1,275 bp in P. confucius. srRNA was 767 bp in length and located upstream of the A + T–rich region in S. yunnanense, whereas 747 bp and upstream of the control region in P. confucius.

Fig. 1

Mitochondrial genome maps of Sinodendron yunnanense. The abbreviations for the genes are as follows: cox1–3 refer to the cytochrome C oxidase subunits; cytb refers to cytochrome B; and nad1–6 refers to NADH dehydrogenase subunits; atp6 and atp8 refer to subunits 6 and 8 of ATPase; rrnL and rrnS refer to ribosomal RNA of 12S and 16S. From outer to innermost, the first circle shows the gene map, the second circle shows GC content, and the third shows GC skew calculated as (G − C)/(G + C). There is a unique gene rearrangement: trnL2 (UUR) located between the A + T‒rich and control region.

Fig. 2

The mitochondrial genome map of Prosopocoilus confucius. There is a noncoding region of 580 bp between trnI and trnQ.

Table 2

Mitogenome organization of Sinodendron yunnanense

Gene	Strand	Region	Length (bp)	Start codon	Stop codon	Anticodon	Intergenic nucleotides (bp)
trnI	J	1–65	65	–	–	GAT	–3
trnQ	N	63–131	69	–	–	TTG	–1
trnM	J	131–199	69	–	–	CAT	0
nad2	J	200–1213	1014	–	TAA	–	2
trnW	J	1,216–1,281	66	–	–	TCA	–8
trnC	N	1,274–1,335	62	–	–	GCA	0
trnY	N	1,336–1,401	66	–	–	GTA	1
cox1	J	1,403–2,938	1536	AAT	TAA	–	46
cox2	J	2,985–3,672	688	ATG	T	–	0
trnK	J	3,673–3,743	71	–	–	CTT	–1
trnD	J	3,743–3,806	64	–	–	GTC	0
atp8	J	3,807–3,962	156	ATT	TAA	–	–7
atp6	J	3,956–4,636	681	ATG	TAA	–	–1
cox3	J	4,636–5,424	789	ATG	TAA	–	2
trnG	J	5,427–5,489	63	–	–	TCC	0
nad3	J	5,490–5,843	354	ATT	TAG	–	–2
trnA	J	5,842–5,906	65	–	–	TGC	–1
trnR	J	5,906–5,971	66	–	–	TCG	–1
trnN	J	5,971–6,035	65	–	–	GTT	–1
trnS(AGN)	J	6,035–6,103	69	–	–	ACT	–1
trnE	J	6,103–6,165	63	–	–	TTC	–2
trnF	N	6,164–6,228	65	–	–	GAA	0
nad5	N	6,229–7,951	1723	ATA	T	–	–3
trnH	N	7,949–8,011	63	–	–	GTG	0
nad4	N	8,012–9,347	1336	ATG	T	–	–7
nad4L	N	9,341–9,631	291	ATT	TAA	–	2
trnT	J	9,634–9,697	64	–	–	TGT	0
trnP	N	9,698–9,760	63	–	–	TGG	1
nad6	J	9,762–10,259	498	ATC	TAA	–	–1
cytb	J	10,259–11,401	1143	ATG	TAG	–	–2
trnS(UCN)	J	11,400–11,464	65	–	–	TGA	19
nad1	N	11,484–12,434	951	ATA	TAG	–	1
trnL(CUN)	N	12,436–12,499	64	–	–	TAG	0
rrnL	N	12,500–13,777	1278	–	–	–	0
trnV	N	13,778–13,847	70	–	–	TAC	0
rrnS	N	13,848–14,614	767	–	–	–	0
A+T–rich region	–	14,615–16,301	1687	–	–	–	0
trnL(UUR)	J	16,302–16,367	66	–	–	TAA	0
Control region	–	16,368–16,921	554	–	–	–	0

Table 3

Mitogenome organization of Prosopocoilus confucius

Gene	Strand	Region	Length (bp)	Start codon	Stop codon	Anticodon	Intergenic nucleotides (bp)
trnI	J	1–62	62	–	–	GAT	0
noncoding region	–	63–642	580	–	–	–	0
trnQ	N	643–711	69	–	–	TTG	–1
trnM	J	711–778	68	–	–	CAT	0
nad2	J	779–1,792	1014	ATA	TAG	–	2
trnW	J	1,795–1,860	66	–	–	TCA	–8
trnC	N	1,853–1,913	61	–	–	GCA	–1
trnY	N	1,913–1,977	65	–	–	GTA	1
cox1	J	1,979–3,509	1531	AAT	T	–	0
trnL(UUR)	J	3,510–3,573	64	–	–	TAA	0
cox2	J	3,574–4,261	688	ATA	T	–	0
trnK	J	4,262–4,332	71	–	–	CTT	–1
trnD	J	4,332–4,393	62	–	–	GTC	0
atp8	J	4,394–4,549	156	ATT	TAG	–	–7
atp6	J	4,543–5,211	669	ATG	TAA	–	–1
cox3	J	5,211–5,995	785	ATG	TA	–	–1
trnG	J	5,995–6,056	62	–	–	TCC	0
nad3	J	6,057–6,410	354	ATA	TAG	–	–2
trnA	J	6,409–6,473	65	–	–	TGC	–1
trnR	J	6,473–6,536	64	–	–	TCG	–1
trnN	J	6,536–6,602	67	–	–	GTT	–1
trnS(AGN)	J	6,602–6,670	69	–	–	TCT	–1
trnE	J	6,670–6,732	63	–	–	TTC	–2
trnF	N	6,731–6,794	64	–	–	GAA	0
nad5	N	6,795–8,511	1717	ATT	T	–	0
trnH	N	8,512–8,578	67	–	–	GTG	0
nad4	N	8,579–9,914	1336	ATG	T	–	–7
nad4L	N	9,908–10,195	288	ATG	TAA	–	2
trnT	J	10,198–10,261	64	–	–	TGT	0
trnP	N	10,262–10,327	66	–	–	TGG	5
nad6	J	10,333–10,830	498	ATG	TAA	–	–1
cytb	J	10,830–11,972	1143	ATG	TAA	–	–2
trnS(UCN)	J	11,971–12,035	65	–	–	TGA	18
nad1	N	12,054–13,004	951	ATA	TAG	–	0
trnL(CUN)	J	13,005–13,067	63	–	–	TAG	0
rrnL	N	13,068–14,342	1275	–	–	–	–8
trnV	N	14,335–14,403	69	–	–	TAC	0
rrnS	N	14,404–15,150	747	–	–	–	–14
Control region	–	15,137–16,951	1815	––	–	–	0

The mitogenome nucleotide composition was A (38.8%), C (15.8%), T (36.2%), and G (9.1%) in S. yunnanense, and A (36.5%), C (21.9%), T (31.3%), and G (10.3%) in P. confucius, with an AT content of 75% and 67.8%, respectively. The higher AT content of S. yunnanense is relative to the A + T–rich region. In five lucanid species, the heavy strand of the mitogenome has a positive AT-skew (0.035 to 0.107) and negative GC-skew (‒0.362 to ‒0.271). In contrast, S. yunnanense rRNA genes have negative AT-skew and positive GC-skew. Interestingly, the AT-skew and GC-skew of PCGs were negative, and positive in the tRNA genes (Table 4). It has been proposed that nucleotide composition strand bias might be related to replication and transcription mechanisms (Hassanin et al. 2005, Song et al. 2016a).

Table 4

AT-content, AT-skew, and GC-skew of five Lucanidae mitochondrial genomes

		Genome			PCGs			tRNAs			rRNAs
Genus	Species	A+T%	AT-skew	GC-skew	A+T%	AT-skew	GC-skew	A+T%	AT-skew	GC-skew	A+T%	AT-skew	GC-skew
Lucanus	mazama	67.125	0.074	−0.272	65.646	−0.154	−0.068	72.426	0.020	0.141	72.192	−0.093	0.350
Prosopocoilus	confucius	67.851	0.077	−0.362	66.092	−0.154	−0.072	71.309	0.070	0.146	70.277	−0.065	0.431
Prosopocoilus	gracilis	66.091	0.106	−0.332	64.469	−0.137	−0.077	69.729	0.042	0.113	69.797	−0.102	0.417
Prosopocoilus	blanchardi	67.015	0.107	−0.265	65.220	−0.149	−0.074	70.146	0.050	0.096	70.620	−0.127	0.368
Sinodendron	yunnanense	75.061	0.035	−0.271	73.109	−0.140	−0.057	76.326	0.023	0.115	78.582	−0.035	0.370

PCGs and Codon Usage

All PCGs of S. yunnanense and P. confucius had the conventional start codon for invertebrate mitochondrial PCGs (ATN) with the exception of cox1 (AAT) as found previously (Sheffield et al. 2008). In S. yunnanense, 10 (eight in P. confucius) of the 13 PCGs shared the typical termination codons TAA and TAG, whereas in the remaining genes an incomplete stop codon T or TA was inferred. It is not uncommon in insects that incomplete codon structures signal a halt of protein translation (Wu et al. 2014, Li 2014, Cheng et al. 2016). The RSCU analysis indicated that codons including A or T at the third position were overrepresented in comparison to other synonymous codons (Fig. 3), which could reflect nucleotide bias.

Fig. 3

RSCU in the S. yunnanense, P. confucius, P. gracilis and P. blanchardi. Lower bars give color codes. Y axis gives RSCU.

Intergenic Spacers and Noncoding Region

There was frequent occurrence of intergenic spacers in the two new mitogenomes, interspersed throughout the PCGs and RNA genes, and ranging from 1 to 580 bp in length (Tables 2 and 3). Among them, a 7-bp conserved motif (TACTAAA) upstream of nad1 was proposed as a binding site for the mitochondrial transcription termination factor, and noted in various beetles and other insects (Cameron and Whiting 2008, Sheffield et al. 2009, Cameron 2014b, Kim et al. 2015, Wu et al. 2016, Yang et al. 2016).

A notable result in P. confucius was a 580 bp noncoding region located between trnI and trnQ, with three structural sections: (C)14 in the initial third; (A)14 in the center, and a (C)9 in the latter third. This noncoding region is also present and at the same position in P. blanchardi, albeit longer (4,051 bp) and more complex (with 17 tandem repetitive sequence, each of which is composed of two subunits; Kim et al. 2015). These features appear to reflect both congeneric and phylogenetic distances (Fig. 6). Moreover, this noncoding region did not occur in P. gracilis, despite being a longstanding member of Prosopocoilus. Indeed, the mitogenome of this species was structured more typical of insects. Absence of this noncoding region in P. gracilis and paraphyly of Prosopocoilus in relation to this species (Fig. 6) raises a question on its taxonomic status.

Fig. 6

Phylogenetic reconstruction of 15 beetles using a concatenated dataset of mitochondrial PCGs. (A) ML and (B) BI.

New Gene Rearrangement

Gene arrangements of most insect mitogenomes are of the type exemplified by Drosophila yakuba, considered to be the ancestral state (Cameron 2014a). However, gene rearrangement events have been reported in many lineages (Timmermans and Vogler 2012, Song et al. 2016b), with a moderate frequency in Coleoptera (Wei and Chen 2011, Li et al. 2013). Although as the most diverse animal order (with >390,000 described species; Zhang 2013), there are merely ∼80 reported mitogenomes of Coleopterans, belonging to 32 families and representing all four suborders (as of February 2016). Gene rearrangement events have been reported in 10 of these, in the form of reversal or absence of tRNAs; in Anthicidae (1), Byrrhoidea (3), Chrysomelidae (3), Curculionidae (1), Lycidae (1), and Mordellidae (1) (Timmermans and Vogler 2012, Nie and Yang 2014).

In this study, we report a novel gene rearrangement of S. yunnanense, a first for Lucanidae. trnL (UUR), ancestrally located between cox1 and cox2, was translocated to a position between A + T‒rich region and control region (Fig. 4). Moreover, intergenic spacers were observed between trnL (UUR) and trnI, and cox1 and cox2, which is thought to be evidence of random gene losses (San et al. 2006). The TDRL model (tandem duplication followed by random loss; Moritz et al. 1987) was introduced to interpret the transposition of trnL (UUR) (Fig. 5). Additionally, recombination is another possible mechanism since illicit priming is unlikely due to the lack of a replication origin in this region (Wei et al. 2009). Data for more representatives are needed to determine the universality of the rearrangement in Lucanidae and to better understand the causal mechanisms.

Fig. 4

Mitochondrial gene order of putative ancestor, S. yunnanense, P. confucius, P. gracilis and P. blanchardi. There is a unique gene rearrangement: trnL(UUR) located between the A + T‒rich and control region in S. yunnanense. A non–coding region is in the same position of P. confucius and P. blanchardi. The mitochondrial gene order of P. gracilis is as per the ancestor.

Fig. 5

The TDRL model was introduced to interpret the gene rearrangement in S. yunnanense. (A) The putative ancestral arrangement. (B) Genes in gray were duplicated to form the intermediate. (C) Some regions were lost to form the current arrangement in S. yunnanense. The X in B indicates the randomly lost regions.

Phylogenetic Analysis

ML and BI phylogenies were reconstructed using a supermatrix of the 13 PCGs. The phylogeny showed that Lucanidae was a monophyletic group and S. yunnanense was sister to other Lucanidae (Fig. 6). Similarly, the genetic distance between this species and others was high (0.285‒0.359) (Supp Table 1 [online only]). These results are consistent with previous studies based on morphological characters (Holloway 1960, 2007) and molecular data (Kim and Farrell 2015).

The species P. confucius and P. blanchardi were recovered as sister taxa (1.00 BPP, 83% MLB), and had low genetic distance (0.241; Supp Table 1 [online only]). This result is consistent with their traditional taxonomy. These two species have long been considered archetypal Prosopocoilus, sharing characters such as large, stout and shiny body, with straight and well developed mandibles in the males. However, the genus Prosopocoilus formed a polyphyly due to the distinctive positioning of P. gracilis (1.00 BPP, 100% MLB), sister to the pair Dorcus paralleipipedus and Hemisodorcus rubrofemoratus. The genetic distance among the three taxa of Prosopocoilus is 0.241, 0.287, and 0.296 (Supp Table 1 [online only]), whereas the value between P. gracilis and D. paralleipipedus is 0.267, and 0.260 with H. rubrofemoratus (Supp Table 1 [online only]). The status of P. gracilis has been revised often in morphological taxonomy. P. gracilis was first erected by Benesh (1950). Séguy (1954) established the genus Epidorcus, including P. gracilis and a congener, despite a lack of convincing generic characters (see Zhong et al. 2014). Actually, its genital characters are quite distinct, and much more similar to those of species of Serrognathus (Wan 2007). Huang and Chen (2013) resurrected the genus Epidorcus for P. gracilis and its congeners after morphological analysis although they acknowledged characters shared by Epidorcus and Serrognathus species.

In our study, the phylogenomic analysis supported the morphological conclusion, i.e., P. gracilis should be removed from Prosopocoilus. Instead, P. gracilis probably belongs in Dorcus (s.l.). This is just one of many issues warranting more comprehensive work in Lucanidae phylogenetics, including higher-level classification, doubtful monophylies of some key genera, and species/subspecies boundaries and nomenclature, all of which would benefit from more comprehensive genetic data.

Click here for additional data file.