Bulk de novo mitogenome assembly from pooled total DNA elucidates the phylogeny of weevils (Coleoptera: Curculionoidea)

Abstract

Complete mitochondrial genomes have been shown to be reliable markers for phylogeny reconstruction among diverse animal groups. However, the relative difficulty and high cost associated with obtaining de novo full mitogenomes have frequently led to conspicuously low taxon sampling in ensuing studies. Here, we report the successful use of an economical and accessible method for assembling complete or near-complete mitogenomes through shot-gun next-generation sequencing of a single library made from pooled total DNA extracts of numerous target species. To avoid the use of separate indexed libraries for each specimen, and an associated increase in cost, we incorporate standard polymerase chain reaction-based "bait" sequences to identify the assembled mitogenomes. The method was applied to study the higher level phylogenetic relationships in the weevils (Coleoptera: Curculionoidea), producing 92 newly assembled mitogenomes obtained in a single Illumina MiSeq run. The analysis supported a separate origin of wood-boring behavior by the subfamilies Scolytinae, Platypodinae, and Cossoninae. This finding contradicts morphological hypotheses proposing a close relationship between the first two of these but is congruent with previous molecular studies, reinforcing the utility of mitogenomes in phylogeny reconstruction. Our methodology provides a technically simple procedure for generating densely sampled trees from whole mitogenomes and is widely applicable to groups of animals for which bait sequences are the only required prior genome knowledge.

Keywords: MiSeq; genomics; mitochondria; next-generation sequencing; phylogenetics; wood-boring.

Fig. 1.

Cumulative distribution of assembly lengths from the Celera, IDBA-UD, and the combined Minimus2-generated assemblies.

Fig. 2.

Frequency distribution of assembly lengths from the Celera, IDBA-UD, and the combined Minimus2-generated assemblies.

Fig. 3.

Relative proportions, by gene, of total “bait” sequences available, “bait” sequences with matching “hits” to the assembled genes and matching hits that contributed to a successful mitogenome identification following a BLAST search.

Fig. 4.

Mean sequencing coverage versus (A) assembly (contig) length (bp) and (B) approximate mass of genomic DNA in the sample pool, for identified mitogenomic assemblies. Thirty-one samples that were not assayed for DNA concentration are shown at bottom of graph B.

Fig. 5.

(Parts 1 and 2) ML tree resulting from the analysis of the “all genes” data set partitioned according to the six PartitionFinder partitions (see table 1). Within Curculionidae s. str. (sensu Bouchard et al. 2011) branches are colored according to subfamily. Other curculionoid families have their name labels colored by family. Numbers adjacent to nodes are RAxML rapid bootstrap scores, with values more than 80% highlighted in red. The three principal wood-boring subfamilies are represented by dashed branches and the nodes labeled A and B indicate the two large divisions within Curculionidae referred to in the text. Nodes indicated in green correspond to nodes present in the strict consensus tree and nodes indicated in blue are consistent with it. The positions of the three tRNA rearrangements are indicated. Scale bar represents substitution rate. Family and subfamily codes precede taxa names as follows: Anthribidae (ANTH), Attelabidae (ATTE), Brachyceridae (BRAC), Brentidae (BREN), Dryophthoridae (DRYO), Nemonychidae (NEMO), Bagoinae (BAGO), Baridinae (BARI), Ceutorhynchinae (CEUT), Conoderinae (CONO), Cossoninae (COSS), Cryptorhynchinae (CRYP), Curculioninae (CURC), Lixinae (LIXI), Mesoptillinae (MESO), Molytinae (MOLY), Platypodinae (PLAT), and Scolytinae (SCOL).

Fig. 5.

(Parts 1 and 2) ML tree resulting from the analysis of the “all genes” data set partitioned according to the six PartitionFinder partitions (see table 1). Within Curculionidae s. str. (sensu Bouchard et al. 2011) branches are colored according to subfamily. Other curculionoid families have their name labels colored by family. Numbers adjacent to nodes are RAxML rapid bootstrap scores, with values more than 80% highlighted in red. The three principal wood-boring subfamilies are represented by dashed branches and the nodes labeled A and B indicate the two large divisions within Curculionidae referred to in the text. Nodes indicated in green correspond to nodes present in the strict consensus tree and nodes indicated in blue are consistent with it. The positions of the three tRNA rearrangements are indicated. Scale bar represents substitution rate. Family and subfamily codes precede taxa names as follows: Anthribidae (ANTH), Attelabidae (ATTE), Brachyceridae (BRAC), Brentidae (BREN), Dryophthoridae (DRYO), Nemonychidae (NEMO), Bagoinae (BAGO), Baridinae (BARI), Ceutorhynchinae (CEUT), Conoderinae (CONO), Cossoninae (COSS), Cryptorhynchinae (CRYP), Curculioninae (CURC), Lixinae (LIXI), Mesoptillinae (MESO), Molytinae (MOLY), Platypodinae (PLAT), and Scolytinae (SCOL).

Fig. 6.

Schematic flowchart of the principal steps for the bulk de novo assembly of mitogenomes and identification with PCR-amplified “bait” sequences.