Mitochondrial genomes for the Afrobatrachian Hyperolius substriatus (Anura: Hyperoliidae) obtained from museum specimens

Abstract. We report the first mitochondrial genomes of the Afrobatrachian frog species Hyperolius substriatus, providing insights into the gene arrangements, duplications, and evolutionary dynamics within this taxonomically complex genus. Assemblies were gained from two museum specimens and included all 37 typical vertebrate mitogenomic genes. In both cases, the mitogenome of H. substriatus shows duplications of tRNAs (trnL, trnT and trnP). The WANCY region exhibits the characteristic Afrobatrachian pattern but with unique intergenic spacers. Both copies of duplicated non-coding and control regions show high sequence similarity within and between specimens. The distinct gene order and homology of duplications observed can be explained under the tandem duplication and random loss model of mitochondrial evolution. Furthermore, a comparison with Hyperolius marmoratus highlights differences in gene copy number and synteny, and the presence of a duplication of trnM in the latter appears to be a derived condition. Together, these findings provide valuable insights into the taxonomy and evolution of Hyperolius, contributing to understanding interspecific gene reorganisation in this diverse group and suggesting that gene arrangements could help resolve some of the taxonomic problems of this complex genus.

Vertebrate mitogenomes generally have conserved gene organisation.However, the WANCY region, which includes the origin of replication of the light strand (O L ) and tRNAs in the order: trnW, trnA, trnN, O L , trnC, and trnY (Seutin et al. 1994), is considered a hotspot of gene rearrangement.Different arrangements observed in this region have been explained by the tandem duplication-random loss (TDRL) model, which presupposes that novel gene orders arise from tandem repeats, followed by random deletions of each pair of the duplicated genes (Boore 1999, Moritz et al. 2017).Kurabayashi & Sumida (2013) propose that the WANCY region, denoted as WNO L ACY, is a synapomorphy of Afrobatrachia, which comprises 426 species of frogs distributed in the families Arthroleptidae, Brevicipitidae, Hemisotidae, and Hyperoliidae (sensu Frost et al. 2006).The family Hyperoliidae Laurent, 1943 comprises over 200 species distributed in 17 genera in sub-Saharan Africa.Hyperolius Rapp, 1842 is the most speciose genus with 144 species.Six complete (Kurabayashi & Sumida 2013, Hemmi et al. 2020) and five nearcomplete (Zhang et al. 2013) mitogenomes are available for Afrobatrachians.To date, there is only one complete and one partial mitogenome available for the genus Hyperolius (Hyperolius marmoratus (GenBank: AB777218) and H. ocellatus, (GenBank: JX564872)).Here, we present two near-complete but uncircularised mitogenomes of the East African reed frog Hyperolius substriatus Ahl, 1931.

Material and Methods
The two specimens of H. substriatus (BMNH 2002.638 andBMNH 2002.654)from the Nilo Nature Forest Reserve (-4.93333, 38.65000), northwest of the East Usambara Mountains in Tanzania, are part of the collection of the Natural History Museum in London, UK.These specimens, collected in 2002, were fixed and preserved in 70% ethanol.A small fragment of liver was dissected from each specimen using sterile forceps.The samples were lysed overnight in 180 µl Qiagen® ATL buffer and 10 µl Proteinase K at 56 °C.Total DNA was extracted from each lysate using the Qiagen® DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany) per the manufacturer's protocol.Genomic libraries were prepared using NEBNext® Ultra™ II DNA Library Prep Kit for Illumina® enzymatic fragmentation (Illumina, San Diego) and sequenced on a NovaSeq PE150 (Novogene Europe, Cambridge).Low-quality ends and adapters were trimmed using Trim Galore v. 0.6.10 (https:// github.com/FelixKrueger/TrimGalore),applying a Phred score threshold of 20 and stringency of 4. NOVOPlasty v.4.2 (Dierckxsens et al. 2017) was used to independently de novo assemble the mitogenomes using k-mer size 33, and a cytochrome c oxidase subunit I sequence from H. substriatus used as a seed (GenBank: KY177140).Contigs generated by NOVOPlasty were then visualised in Geneious 9.0.5 (https://www.geneious.com)for final assembly.

Results
Two near-complete assemblies for specimens BMNH 2002.638 andBMNH 2002.654were produced (GenBank accessions: OR987482 and OR987483) using NOVOPlasty.Of the two libraries used, only the mitogenome for specimen BMNH 2002.654resulted in a circularised mitogenome; however, upon manual inspection, one of the four final contigs comprised a 21,853 bp repeated motif.As such, this contig was excluded.While neither final assembly was circularised by NOVOPlasty, both resulted in a single contiguous sequence with the same gene order and contained all expected vertebrate mitochondrial genes.This outcome allowed for the first characterisation of the H. substriatus mitogenome and further comparative analyses.Table 1 summarises the size and gene types of mitogenomes of H. substriatus and Hyperolius marmoratus (GenBank: AB777218).
The mitogenome of H. substriatus contains a total of 40 genes because there is a duplication of tRNAs trnL (CUN), trnT, and trnP (Fig. 1A).Thirty genes are encoded on the H-strand and ten on the L-strand, in accord with the typical vertebrate arrangement.Nucleotide composition (average between both specimens) is 32% A, 33% T, 22% C, and 13% G.
As expected, the WANCY region of H. substriatus follows the Afrobatrachia pattern (Fig. 1A).Additionally, a short intergenic spacer (IGS) of 5 bp is found between trnW and trnN, followed by two longer ones of 38 and 14 nucleotides that are respectively found before and after the O L .Presumed homologues of these IGSs (same position and similar sizes) are also present but previously unreported in other Afrobatrachians (B.adspersus, Hemisus marmoratus, Hyperolius marmoratus and T. robustus).
Finally, a fourth IGS (4 bp) is observed between trnA and trnC, a pattern which is only similarly observed in Hemisus marmoratus (Fig. 2).
The cytochrome b (cytb) gene of H. substriatus lacks a complete stop codon, presumably generated upon post-transcriptional polyadenylation (see comments in Discussion).The MITOS annotation of this gene  was shorter than expected (1,134 bp; 378 amino acids), so it was manually annotated extending a further nine base pairs, resulting in 1,143 bp and 381 amino acids (Table 2).
In H. substriatus, the CR + LTPF tRNA cluster (trnL (CUN)-trnT-trnP-trnF) differs from the typical Neobatrachian arrangement (Kurabayashi & Sumida 2013) in having two copies of trnL (CUN)-trnT-trnP, with a non-coding region (NC) inserted between trnL (CUN) and trnT (Fig. 1A,C).The two copies of trnT and trnP have identical nucleotide sequences, suggesting that both are functional (i.e.neither has become a pseudogene).The two copies of the trnL (CUN) have the same length and similar structure but differ by six nucleotides (Fig. 3).A further distinguishing feature is that the two copies of the CR are arranged after the trnP.The first copy of the CR has a similar length in both specimens of H. substriatus, whereas the second copy varies substantially in length (BMNH 2002.638= 2,023 bp; BMNH 2002.654= 2,695 bp) (Fig. 1A).These copies of the CR are incomplete and probably are connected to the trnF (Fig. 1A), which is adjacent to the SSU (12S).
Comparisons of the two copies of the NC region and the two CRs were conducted to explore homologies further.Pairwise sequence similarity was calculated for the NC regions using 654 bp (89% of the original 728 bp).The first NC region was identical in both  specimens, and the second was revealed to vary from 72 to 73% (Fig. 1B).Comparisons of the CRs were based on 1,993 bp (73% of the original 2,703 bp).The first CR is identical in both specimens, whereas the second is 98% similar.A comparison of each copy within the specimen revealed 94-99% similarity (Fig. 1B).
The phylogeny confirms that H. substriatus groups with Hyperolius marmoratus (Fig. 4).Low branch support for other clades is likely due to the small selection of genes used to infer the phylogeny.However, the close relationship between Afrobatrachia and Microhylidae and their sister relationship with Natatanura is well supported elsewhere (see Kurabayashi & Sumida 2013).

Discussion
Here, we provide the first mitochondrial genome records for the Afrobatrachian H. substriatus.Both assemblies are similar in length to the complete mitogenome available for Hyperolius marmoratus (Table 2), suggesting non-circularisation is due to incomplete cover of the CR.Because the available mitogenome of H. ocellatus is only partial (9,457 bp) and does not include the complete WANCY region or CR + LTPF cluster, no further comparisons can be made now.
A single copy of trnM was found in H. substriatus, while two were found in Hyperolius marmoratus.
Because typical Neobatrachia and vertebrate mitogenomes only have one copy of trnM, the condition in Hyperolius marmoratus is a derived feature.The organisation of the WANCY region and IGSs is consistent in the other five species of Afrobatrachians compared here, except for one extra IGS in H. substriatus and Hemisus marmoratus (see Fig. 2).These small fragments of non-coding DNA are possibly remnants of gene duplication and are expected under the TDRL model.However, at least three TDRL events would be necessary to explain the pattern observed in Afrobatrachians (Fig. 2).
It is not uncommon for the cytb of vertebrates to lack or have incomplete stop codons (e.g.Görtz & Feldmann 1982, Murray et al. 1994, McKnight & Shaffer 1997, Glenn et al. 2002).The cytb gene usually encodes 380 amino acids in vertebrates, but deviations from this number are not unusual.For instance, within the Afrobatrachians analysed here, the number of amino acids varies between 379 and 384 (Table 2).
Duplicated NC regions and CRs of similar sizes to those found in H. substriatus also occur in Hyperolius marmoratus but with different synteny (see Fig. 1A).
Given the high similarity between the two copies of the NC regions and CRs observed in H. substriatus, we speculate that the two copies corresponding to the CR + LTPF tRNA cluster are homologous and could be explained by three TDRL events (Fig. 1C).The arrangement of this gene cluster in H. substriatus is a derived condition from Neobatrachia.
This study provides the first mitogenome-wide comparison of gene orders and arrangements for the genus Hyperolius, showing significant interspecific reorganisation and duplication-loss events.The unique variation within Hyperolius mitogenomes revealed here contributes to growing research on interspecific gene reorganisation and improves taxonomic resolution within this speciose lineage.

Fig. 2 .
Fig. 2. Afrobatrachian gene order and arrangement of the WANCY region.Arrows indicate intergenic spacers common to all species and numbers denote the base pair length of each spacer.

Fig. 3 .
Fig. 3. Secondary structure and DNA sequences of two copies of trnL (CUN).A) Arrows and boxes indicate mismatches between copies.B) DNA sequence differences in bases (highlighted in grey and underscored).

Fig. 4 .
Fig. 4. Maximum likelihood phylogeny based on the two ribosomal RNAs (12S and 16S) and trnV.Values above the branch indicate bootstrap support.GenBank accession numbers are shown in parentheses.

Table 2 .
Size and number of encoded amino acids of the cytb gene of some Afrobatrachians.