The mitochondrial genome of Grapsus albolineatus (Decapoda: Brachyura: Grapsidae) and phylogenetic associations in Brachyura

Complete mitochondrial genomes (mitogenomes) can provide useful information for phylogenetic relationships, gene rearrangement, and evolutionary traits. In this study, we determined the complete mitochondrial DNA sequence of the herbivorous crab Grapsus albolineatus. It is a typical metazoan mitochondrial genome. The total size is 15,583 bp, contains the entire set of 37 genes, and has an AT-rich region. Then, 23 of the 37 genes were encoded by the heavy (+) strand while 14 are encoded by the light (−) strand. Compared with the pan-crustacean ground pattern, two tRNA genes (tRNA-His and tRNA-Gln) were rearranged and the tandem duplication/random loss model was used to explain the observed gene rearrangements. The phylogenetic results showed that all Grapsidae crabs clustered together as a group. Furthermore, the monophyly of each family was well supported, with the exception of Menippidae. In general, the results obtained in this study will contribute to the better understanding of gene rearrangements in Grapsidae crab mitogenomes and provide new insights into the phylogeny of Brachyura.


Results and discussion
Genome structure and composition. The complete mitogenome sequence of G. albolineatus is a typical closed-circular molecule of 15,583 bp in size (GenBank accession number MZ262276), which is similar in length to the published Grapsidae mitogenomes 4,[9][10][11][12] , a size range from 15,406 to 15,920 bp ( Table 1). The mitogenome contents of G. albolineatus is the same as most other published Brachyura which includes 37 genes, 13 PCGs, 22 tRNAs, and 2 rRNA (rrnl and rrns), as well as a brief non-coding region, all the genes were identified (Fig. 1, Table 2). Most of the 37 genes are located on the heavy (H-) strand, except 4 PCGs (ND5, ND4, ND4L, ND1), 8 tRNAs (tRNA-Cys, Tyr, Gln, Val, Leu, Pro, Phe, and His), and 2 rRNA which are located on the light (L-) strand ( Fig. 1, Table 2). There are 13 regions with overlap in the total G. albolineatus mitogenome, with 3 of them more than 10 bp (trnT (41 bp), trnL 1 (25 bp), and cox2/trnS 2 (20 bp)) and the other 10 shorter than 10 bp (nad4 (7 bp), atp8 (4 bp), cox3/atp6/rnK/nad6/trnW (1 bp), trnG (3 bp), and nad3/nad2 (2 bp)) ( Table 2). The G. albolineatus mitogenome also contains 328 bp of intergenic spacers located in 17 regions, ranging from 1 to 122 bp ( Table 2) and indicating the occurrence of tandem duplications and the deletions of redundant genes. GC-skew of the complete mitogenomes of 6 Grapsidae species were calculated and compared (Tables 3, 4). The nucleotide composition of the G. albolineatus mitogenome is A (33.4%), T (34.04%), G (12.02%), and C (20.54%), with a high A-T bias. The A + T (%) content of the mitogenomes was 66.74%. The AT-skew and GC-skew value are calculated for the chosen complete mitogenomes (Table 3). Both AT-skew and GC-skew of the G. albolineatus mitogenome are slightly negative, −0.009 and −0.262, informing T's and C's are more abundant than A's and G's. Similar results were observed for the other selected Grapsidae mitogenomes. In general, the AT-skew and GCskew of the overall mitogenomes, nucleotide composition, and gene lengths of the G. albolineatus were the same as those of the other Grapsidae species 4,9-12 . PCGs and codon usage. The initial and terminal codons of all PCGs of G. albolineatus are listed in Table 2.
The average A + T content is 65.26%, ranging from 39.63% (ND5) to 74.21% (ATP8) ( Table 3). The AT-skew and GC-skew are −0.159 and −0.034, respectively ( Table 3). All of the PCGs are initiated by the start codon ATN (ATT, ATG, and ATC), except ATP8 (GTG). The majority of the PCGs are terminated with TAA, whereas the other three PCGs (cox1, nad1, and nad2) use TAG as the stop condon ( Table 2). The most frequently used amion acid in G. albolineatus is Leu, and the least common anion acid is Trp (Fig. 2). The relative synonymous codon usage (RSCU) values for G. albolineatus of the 13 PCGs are shown in Table 5 and Fig. 2 24 . The three most frequently detected codons are GCU (Ala), UCU (Ser2), and GUA (Val), whereas GCU (Ala) is the least common codon. Based on CDspT and RSCU, comparative analyses showed that the codon usage pattern of G. albolineatus is conserved. The codon usage patterns of 13 PCGs are similar to those of other Grapsidae species.
Transfer RNAs and ribosomal RNAs. Like most Grapsidae species, G. albolineatus mitogenome contains 22 tRNA genes 20,25,26 . Fourteen of them are encoded by the heavy strain (H-) and the rest are encoded by the light strain (L-). In the whole mitogenome, the size of tRNAs range from 50 to 73 bp and have a total length of 1402 bp, with an obvious AT bias (71.54%) ( Table 2). The AT-skew and GC-skew are −0.009 and 0.158, respectively, showing a slight bias toward the use of Ts and an apparent bias toward Cs (Table 3).
The 12S and 16S rRNA genes are 1331 and 827 bp, respectively, which are typically separated by tRNA-Val (Table 2). These sizes are similar to those of other Grapsidae species [15][16][17][18][19] . The A-T content of rRNAs is 72.57%. The AT-skew and GC-skew are −0.001 and 0.284, respectively, suggesting a slight bias toward the use of Ts and an apparent bias toward Cs (Table 3). As most typical mitogenomes of other crabs, CR is located between 12S rRNA and tRNA-Ile. The 617 bp CR is obviously AT biased (77.63%). The AT-skew and GC-skew are 0.173 and −0.203, respectively (Table 3), indicating an obvious bias toward the use of A's and C's. The index of substitution saturation (Iss) was measured as an implemention in DAMBE 5 and the GTR substitution model 25 . Iss is for the combined dataset of all PCGs of the 59 Brachyura mitogenomes and was significantly lower (Iss = 0.674) than the critical values (Iss, cSym = 0.859). The genes are not saturated, so the reconstructed phylogeny was reliable.  www.nature.com/scientificreports/ Gene rearrangement. Mitochondrial gene rearrangement is an important molecular marker and is considered to be an effective tool for studying mitochondrial evolution 26 . A large number of studies and results have shown that gene rearrangements in metazoan mitochondrial genomes are conserved 20 and the occurrence of gene rearrangements is relatively random and rare 1,19,20,27 . However, it can be used as direct evidence of evolutionary relationships between species 28 . Mapping the gene layout based on the complete mitochondrial sequences of 70 species. Through comparison and analysis with the ancestor of Decapoda (Fig. 3A), we found that G. albolineatus and another 5 species from Grapsidae have a trnH translocation 4,[9][10][11][12][13] , which the trnH shifted into trnE and trnF instead of the usual location between nad5 and nad4 (Fig. 3C). It is widely believed that the tandem duplication/random loss model (TDRL) can explain the movement of trnH, occur from tandem duplication in the region between trnE and nad4, followed by deletions of redundant genes producing trnH-trnF-nad5. Additionally, 45 species from 14 families (Grapsidae, Mictyridae, Ocypodidae, Bythograeidae, Calappidae, Dotillidae, Matutidae, Menippidae, Oziidae, Xanthidae, Oregoniidae, Geryonidae, Portunidae, and Carpiliidae) had the same gene rearrangement, which are consistent with the ancestral of Brachyura (Fig. 3B). However, the gene order in 4 families (Sesarmidae, Varunidae, Macrophthalmidae, and Xenograpsidae) 30,32 displayed 4 patterns of gene rearrangements. The family Sesarmidae observed trnQ and trnI invertred, which has been described in previous studies ( Fig. 3D) 3,19,20,33 . The gene order of the Varunidae (Grapsoidea) and Macrophthalmidae (Ocypodoidea) have the same high level rearrangementa (Fig. 3E). It is worth noting that the two families come from two different superfamilies, but they form a sister clade in phylogenetic trees. The gene order of the Xenograpsidae have a more complex rearrangement and such within-genus rearrangements were infrequent 34 (Fig. 3F,G), which seems to be related to their particular habitat. Xenograpsidae have been found thus far only in shallow-water, volcanically active, and sulphur-rich hydrothermal vents 35 .
Phylogenetic relationships. In the present study, the phylogenetic relationships were analyzed based on the sequences of the 13 PCGs to clarify the relationships in Brachyura. G. albolineatus and other 68 known brachyuran specie were analyzed, with P. nigrofascia and P. gracilipes as outgroups. The two phylogenetic trees (Maximum Likelihood (ML) tree and Bayesian Inference (BI) tree) resulted in identical topological structuring with different supporting value. Then, only one topology (ML) with both support values was presented displayed (Fig. 4). Both trees showed that all the species of Grapsidae clustered together as a solid monophyletic group and consist of three sister clades ((Grapsus + Pachygrapsus) + Metapograpsus). It is obvious that G. albolineatus had the closest relationship with G. tenuicrustatus, and that these two species form a sister clade with high support values (BI posterior probabilities PP = 1, ML bootstrap BP = 100), constituting a Grapsus group. However, recent molecular studies, including our dataset, have not reached an agreement about closest relatives in Grapsidae.
The main phylogenetic structure of our tree is consistent with previous results, but some controversial findings were observed. Here, the families Macrophthalmidae and Varunidae were grouped into one clade, and Mictyridae as basal group which supports the previous findings revealed in Wang et al. and Zhang et al. 9,33 . However, previous researchers revealed that Macrophthalmidae and Varunidae were grouped into one clade, then into another clade with Varunidae ((Macrophthalmidae + Varunidae) + Mictyridae) 38,39 , which was conflict with our results. The classification of Grapsoidea and Ocypodoidea has long been controversial. Previous studies based on morphological characteristics considered them to be monophyletic branches. However, an increasing number of molecular studies, including ours, challenge the inconsistent views on the traditional classification system that are put forward. Although the polyphyly of Grapsoidea, Ocypodoidea, and Eriphioidea is well supported, the phylogenetic relationships of these superfamilies need to be further analyzed by integrating additional molecular data [32][33][34][35][36] . Previous studies on mitochondrial phylogeny have confirmed the importance of mitochondrial www.nature.com/scientificreports/ genomic data in elucidating the Grapsidae phylogeny 13,19 . On the contrary, many families contained only one representative, which may produce unstable phylogenetic relationships. Therefore, it is necessary to perform further mitogenome sequence studies to obtain a more comprehensive taxon sampling and understand the phylogeny and evolution of Grapsidae.

Materials and methods
Sampling and DNA extraction. A specimen of G. albolineatus was collected from Yangjiang, Guangdong Province, China (21°28′45″ N, 111°16′35″ E). The specimen was immediately preserved in absolute ethanol after collection and then stored at −20 °C. This specimen was identified by morphology and fresh tissues were dissected from the operculum and preserved in absolute ethanol before DNA extraction. The total genomic DNA was extracted using the salt-extraction procedure with a slight modification 40 and stored at −20 °C.  www.nature.com/scientificreports/ Genome sequencing, assembly, and annotation. The mitogenomes of G. albolineatus was sequenced by Origin gene Co. Ltd., Shanghai, China and was sequenced on the Illumina HiSeq X Ten platform. HiSeq X Ten libraries with an insert size of 300-500 bp were generated from the genomic DNA. About 10 Gb of raw data was generated for each library. Low-quality reads, adapters, and sequences with high "N" ratios and length less than 25 bp were removed. The clean reads were assembled using the software NOVOPlasty (https:// github. com/ ndier ckx/ NOVOP lasty) 42 , annotated, and manually corrected on the basis of the complete mitogenome sets assembled de novo by using MITOS tools (http:// mitos2. bioinf. uni-leipz ig. de/ index. py) 43 . To confirm the correct sequences, we compared the assembled mitochondrial genes with those of other Grapsus species and identified the mitogenomic sequences by checking the cox1 barcode sequence with NCBI BLAST 43 . The abnormal start and stop codons were determined by comparing them with the start and stop codons of other marine gastropods. Then, the reads were reconstructed using the de novo assembly program. The complete mtDNA was annotated using the software Sequin version 16.0 (https:// trace. ncbi. nlm. nih. gov/ Traces/ sra). The mitogenome map of the G. albolineatus was drawn using the online tool CGView Server (http:// cgview. ca/) 45 . The secondary structures predicted of the tRNA genes were plotted by using MITOS Web Server. The relative synonymous codon usage (RSCU) values and substitution saturation for the 13 PCGs, calculated by DAMBE 5 45 , were analyzed with MEGA 7 46 . The GC-skews and AT-skews were used to determine the base compositional difference and strand asymmetry among the samples. According to the following formulas 46 , composition skew values were calculated as AT-skew = A − T/A + T and GC skew = G − C/G + C. Substitution saturation for the 13 PCGs was calculated by DAMBE 5 45 .
Phylogenetic analysis. The phylogenetic relationships within Brachyura were reconstructed using the sequences of the 13 PCGs of a total of 57 complete mitogenome sequences downloaded from the GenBank database (https:// www. ncbi. nlm. nih. gov/ genba nk/) and adding two species of Paguridae to serve as the outgroup ( Table 1). The phylogenetic relationships were analyzed with Maximum Likelihood (ML) by using IQ-TREE 1.6.2 and Bayesian Inference (BI) methods in MrBayes 3.2 version program [47][48][49] . The ML analysis was inferred with 1000 ultrafast likelihood bootstrap replicates by using IQ-TREE 1.6.2. The best-fit model for each partition was GTR + F + R6, selected according to the Bayesian information criterion (BIC). BI was performed in MrBayes 3.2, and the best-fit evolutionary models were determined using MrMTgui 50 . MrMTgui was used to associate PAUP, ModelTest, and MrModelTest across platforms. MrBayes settings for the best-fit model (GTR + I + G) were selected by Akaike Information Criterion (AIC) in MrModelTest 2.3 51,52 . The Bayesian phylogenetic analyses were performed using the parameter values estimated with the commands in MrModelTest or ModelTest (nst = 6, rates = invgamma) 53 . With three hot chains and one cold chain, they were run simultaneously twice by Markov Chain Monte Carlo (MCMC) sampling, and the posterior distribution was estimated. The MCMC chains were set for 2,000,000 generations and sampled every 1000 steps, with a relative burn-in of 25%. The convergence of the independent runs was evaluated by mean standard deviation of the split frequencies (< 0.01). The phylogenetic trees were visualized and edited using Figure

Conclusions
In this study, the mitogenome of G. albolineatus was sequenced by next-generation sequencing, thereby generating new mitochondrial data for Grapsidae and confirming its ancestral gene order. The G. albolineatus mitogenome is a typical closed-circular molecule including 13 PCGs, 22 tRNA genes, two rRNA genes, and a CR. The AT-skew and GC-skew are both negative in the mitogenome of G. albolineatus, showing an obvious www.nature.com/scientificreports/ bias towards the use of T's and C's, consistent with published findings in most Brachyura crabs. G. albolineatus exhibits a novel gene rearrangement, which is similar to G. tenuicrustatus, P. crassipes, P. marmoratu, M. frontalis, and M. quadridentatus. Compared with the pan-crustacean ground pattern, the trnH of G. albolineatus shifted into trnE and trnF instead of the usual location between nad5 and nad4. By adding 62 Brachyura mitochondrial genomes, rearrangement and the phylogeny of Brachyura was reanalyzed. The phylogenetic analyses indicated