Figures
Abstract
Hypotheses of hybrid origin are common. Here we use next generation sequencing to test a hybrid hypothesis for a non-model insect with a large genome. We compared a putative hybrid triploid stick insect species (Acanthoxyla geisovii) with its putative paternal diploid taxon (Clitarchus hookeri), a relationship that provides clear predictions for the relative genetic diversity within each genome. The parental taxon is expected to have comparatively low allelic diversity that is nested within the diversity of the hybrid daughter genome. The scale of genome sequencing required was conveniently achieved by extracting mRNA and sequencing cDNA to examine expressed allelic diversity. This allowed us to test hybrid-progenitor relationships among non-model organisms with large genomes and different ploidy levels. Examination of thousands of independent loci avoids potential problems produced by the silencing of parts of one or other of the parental genomes, a phenomenon sometimes associated with the process of stabilisation of a hybrid genome. Transcript assembles were assessed for evidence of paralogs and/or alternative splice variants before proceeding. Comparison of transcript assemblies was not an appropriate measure of genetic variability, but by mapping reads back to clusters derived from each species we determined levels of allelic diversity. We found greater cDNA sequence diversity among alleles in the putative hybrid species (Acanthoxyla geisovii) than the non-hybrid. The allelic diversity within the putative paternal species (Clitachus hookeri) nested within the hybrid-daughter genome, supports the current view of a hybrid-progenitor relationship for these stick insect species. Next generation sequencing technology provides opportunities for testing evolutionary hypotheses with non-model organisms, including, as here, genomes that are large due to polyploidy.
Citation: Morgan-Richards M, Hills SFK, Biggs PJ, Trewick SA (2016) Sticky Genomes: Using NGS Evidence to Test Hybrid Speciation Hypotheses. PLoS ONE 11(5): e0154911. https://doi.org/10.1371/journal.pone.0154911
Editor: Hikmet Budak, Montana State University Bozeman, UNITED STATES
Received: January 4, 2016; Accepted: April 21, 2016; Published: May 17, 2016
Copyright: © 2016 Morgan-Richards et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Data are available as fasta and fq files at Dryad. The data package has been assigned the unique identifier doi:10.5061/dryad.h5g60.
Funding: Funding came from Massey University (http://www.massey.ac.nz/massey/home.cfm; PBRF allocation [facilitated by Susan Adams] and Massey University Research Fund Women's Award for 2016: MMR) & Te Tipu Putaiao postdoctoral funding from the New Zealand Foundation Science and Technology (http://www.mbie.govt.nz, SFKH).
Competing interests: The authors have declared that no competing interests exist.
Introduction
Hybridisation between species can combine divergent genomes and produce new species when reproductive isolation from parentals accompanies novel genome fusion [1]. Polyploidy and selfing commonly co-occur with hybridisation in plants leading to a high frequency and multiple origins of hybrid plant taxa [2]. In fungi, hyphal fusion generates hybrids when normally geographically isolated species are brought into contact. In animals, the origin of new species via hybridisation might be relatively rare, but in those taxa where parthenogenetic reproduction has evolved many times, as in phasmids [3], geckos [4], and frogs [5], hybrid species are well documented. Hybrid species can be recognized by the presence in a single genome of alleles that are otherwise distinct to two separate evolutionary lineages or species (Fig 1). How hybrid genomes become stabilized and how fitness costs influence hybrid survival is poorly understood [6], but our estimate of hybrid species frequency is improving [7]; [2].
(A) Two lineages of stick insects have been sampled across their range in New Zealand and by contrasting maternal relationships from mitochondrial DNA sequences with bi-parental multicopy nuclear markers a role for hybridisation has been inferred. Diploid Clitarchus hookeri (orange squares) has both sexual and asexual populations. No males of any of the Acanthoxyla forms are known (purple squares diploid females, purple circles triploid females). (B) Hybrid species are the product of interspecific mating resulting in genomes that are a mix of the two parental species but are reproductively isolated from both these parent taxa. The resulting allelic diversity is illustrated and compared to the diversity expected within non-hybrids and autopolyploids. When short DNA sequence reads are mapped to parents, related and non-related species, allelic similarities can be used to infer origins.
Hypotheses of hybrid origin based on morphologically intermediate traits were, in the past, tested with genetic evidence that relied on universal markers [8] or lengthy development of species’ specific loci [9]. Multicopy genes present additional complexity, and as polyploidy is frequently associated with successful hybrid species [10] this is not a trivial problem. Next generation (high throughput) DNA sequencing provides an opportunity to generate information suitable for testing hybrid origin hypotheses in non-model organisms, an important objective if theory based on model organisms is to be applied to our understanding of Earth’s biodiversity. Here we set out a procedure for evaluating such data using assembled transcripts to compare ‘allelic’ diversity in a putative hybrid lineage and its putative paternal taxon. The process of stabilising a hybrid genome might involve cellular mechanisms such as gene silencing of alleles from one or other parental genome, and this might establish rapidly [11–13]. By sampling a large number of loci and comparing both within and between samples of cDNA, potential problems arising from differential silencing are minimized.
In New Zealand a genus of eight morphologically distinct species of stick insect (Phasmida) have been studied because the entire group lacks males. Each species of Acanthoxyla differs in how spiny it is, the presence/absence of abdominal flanges, and the sculpturing of its eggs [14]. Every individual is female and reproduces parthenogenetically producing viable offspring without males. A hybrid origin for the genus involving the ancestor of a related endemic bisexual species, Clitarchus hookeri, was inferred from a combination of mtDNA and nuclear markers [15]. A maternal bisexual species has not been identified and is likely to be extinct [16, 17]. In addition, many lineages of Acanthoxyla are mosaic triploids [18]. This pattern of polyploidy and hybrid origin has been inferred for many organisms including stick insect lineages in Europe and north Africa [19–21].
The whole Acanthoxyla genus has shallow mitochondrial divergence (<3%; COI-COII) and morphological “species” are not reciprocally monophyletic [15]. No partitioning by geography or diet has been suggested for the eight Acanthoxyla morphospecies. A hybrid origin for the genus and subsequent loss of heterozygosity might explain the current morphological diversity. Multi-copy nuclear markers (ITS, PGI, EF1a) and chromosome evidence identify Clitarchus hookeri as a likely parent taxon of the Acanthoxyla lineage (Fig 1a), although hybridisation between Acanthoxyla species and introgression from Clitarchus hookeri is possible if male Acanthoxyla existed in the past [8, 15, 18]. However, the absence of C. hookeri alleles in some Acanthoxyla individuals [8] could be the result of recombination we know occurs [18]. Clitarchus hookeri has all-female populations in the south of its range but crucially for hybridisation has males in sexual populations further north [22]. All Acanthoxyla lineages are sympatric with Clitarchus hookeri [15, 22].
Acanthoxyla sticks insects have large genomes, with approximately three times as much DNA per cell as humans (~9 pg; [18, 23]), and have not yet had their genome sequenced. To generate a manageable amount of data we reduced the genome sample by extracting mRNA. We used cDNA sequences generated from this to examine the expressed allelic diversity within one Acanthoxyla lineage and compared this to sequences from Clitarchus hookeri. The hybrid origin hypothesis predicts that Acanthoxyla geisovii will share alleles with the putative parental species Clitarchus hookeri, but will also contain alleles unique to the Acanthoxyla geisovii genome (inherited from the maternal parent species). The Acanthoxyla geisovii lineage investigated here is triploid and Clitarchus hookeri is diploid [15, 18]. If Acanthoxyla geisovii is not of hybrid origin then at each locus, all alleles within Acanthoxyla geisovii will be more similar to each other than they are to Clitarchus hookeri alleles. Even if the paternal species has not been correctly identified, the extent of allelic diversity in a parthenogenetic lineage ought to provide evidence of genome fusion. We used approaches to ensure our datasets were robust for assumptions of homology including manual curation of a subset of data to establish that sequence clusters did not contain mixed products of multigene families. We identified and separately analysed the higher GC content components of our data to ensure non-protein coding transcripts were not misleading us, and we searched public databases to confirm our samples were contaminant free. Preliminary functional analyses compared translated sequences with arthropod datasets to identify orthologous groups.
We mapped sequence reads from each stick insect against transcript assemblies of both to reveal asymmetry in allele diversity (Fig 1b). We compared the proportion of putative loci (transcript assemblies) where identical reads were found after mapping total sequence datasets back to putative loci, for each genome. We calculated sequence divergence between assembled transcripts and raw reads to determine relative genetic diversity within each genome. This provided data that supported the existing hybrid origin hypotheses involving these stick insect species.
Materials and Methods
Source material
Total RNA was simultaneously but independently extracted from femur muscle samples from two adult female parthenogenetic stick insects using the QIAGEN RNeasy Mini kit. Acanthoxyla was represented by a green spiny triploid form (A. geisovii) collected from the host plant Podocarpus totara (totara) in Manawatu, New Zealand (40°24'55.86"S, 175°39'50.05"E; Ax.PN-762). Clitarchus hookeri was represented by a brown female collected from the host plant Rubus fruticosus (bramble) in Kapiti Coast, New Zealand (40°51'35.37"S, 175° 3'4.62"E; Ch.W-765). Insects were collected from private property with the owner’s permission.
RNA was stored at -80°C until required, and analysed using a BioAnalyser, total RNA samples for mRNA: Quant-iT RNA: 90–100 ug/mL; Quant-iT dsDNA 18–31 ug/mL; Quant-iT ssDNA: 24–87 ug/mL; Quant-iT Protein 714–737 ug/mL. High protein content does not interfere with mRNA sequencing library preparation. The polyA mRNA was extracted using oligo dT magnetic-beads (protein and DNA remained in the supernatant). The BioAnalyzer measurements revealed a low level of RNA degradation in the Acanthoxyla sample only.
Sequence generation
The RNA samples were made into libraries using an Illumina mRNA-Seq Sample Preparation Kit (part no. RS-100-0801) and were indexed using the index sequence 5´CAAGCAGAAGACGGCATACGAGATAAGCTAGTGACTGGAGTTC for Acanthoxyla geisovii, and 5´CAAGCAGAAGACGGCATACGAGATGTAGCCGTGACTGGAGTTC for Clitarchus hookeri. The libraries were then pooled by equal molarity and loaded at 42.5 pM and run in a single lane on an Illumina GAIIX to generate single 100 base reads. The run was processed with RTA v.1.6 and Casava v.1.6 to demultiplex the data and generate the short read sequence files for each species. The short reads were analysed with SolexaQA [24] to assess the quality of the run, and to inform the trimming of reads to the desired quality for subsequent assemblies.
De novo assembly and clustering
De novo assemblers Velvet v.1.2.10; [25] and ABySS v.1.3.2; [26] were used to assemble the short reads from each species. A combination of k-mer assembly parameters (from 25 to 61 in steps of 4), pre-trimming of the data to remove any Illumina TruSeq adapters (resulting from small inserts for example) and quality-trimmed data (DynamicTrim at quality thresholds of 0.01, 0.003 and 0.001) were tried for each species, resulting in 120 combinations of assembly parameters (S1 Table). In addition, Velvet assemblies were performed with a minimum contig output length of 200bp. As this k-mer sweep approach generated many sequences that were almost identical (varying only at the ends due to the k-mers), a custom Perl script was used to generate a unique set of contigs from each k-mer sweep (S2 Table). The trimmed and unique contigs were then used as input for a clustering procedure using OrthoMCL v.2.0 [27, 28] with default parameters to generate clusters of sequence contigs for further analysis (Fig 2). Clustering of sequences was necessary so that each transcript could be treated as an independent locus in downstream analyses; sequences within each cluster are not independent units.
Manual curation of a subsample of clusters
A stratified random sample of 270 clusters (~10%) was taken from each single species set for manual curation and assessment (a total of 540 clusters). The sample was stratified to favor clusters containing a greater number of sequences as these clusters contained a greater proportion of the useful data and also avoided oversampling from the large number of clusters containing only two sequences. Clusters were assembled using the de novo assembly tool in Geneious v.6.1.6 (Biomatters; http://www.geneious.com). The quality of the resulting transcript assemblies was assessed on the number of transcript assemblies returned from the assembly of each cluster, and the quantity and distribution of sequence variability across the transcript assemblies. The number of transcript assemblies provides a measure of the effectiveness on the clustering algorithm; where more than one super-contig is generated for a given cluster this indicates that at least two different transcripts have been incorrectly assigned to the same cluster. Where more than one super-contig was generated for a cluster, the longest was taken for further analysis. Assessment of the degree and distribution of nucleotide variability across the transcript assembly of a given cluster provides an indication of the erroneous clustering of closely related paralogs or splice variants. As such artefacts can be especially misleading when analyzing data from polyploids, dubious sequences were removed to produce transcript assemblies of contig sequences representing single loci. In addition, observed nucleotide disagreements involving the first or last five bases of contributing contig sequences were resolved by deleting these ends. These errors appear to have resulted from miss-calls in the initial assembly of sequence reads into contigs in which the depth of coverage is reduced at the ends of contigs. Consensus sequences of each of the transcript assemblies were generated in Geneious. Open reading frames were predicted for each consensus sequence, and GC content was calculated. In order to confirm the nucleotide variability observed in the transcript assemblies, sequence reads were mapped back to these consensus sequences. The GC content was also calculated for primate protein coding genes with apparent homologs in our stick insect dataset. This was compared to the GC content of the stick insect sequence in order to determine a threshold for GC content of protein coding DNA (S1 Fig). The identity of putative protein coding ORFs was examined by a BLAST homology search using BLAST2GO [29]. The allelic diversity of a subset of genes identified in both species was further tested on a small scale by remapping reads to consensus sequences and calling SNPs (at 10% minimum frequency) in Geneious.
Contig clustering
For each species, the clustering process generated sets of sequences that overlapped to varying degrees, and it was generally the case that the longest contig in a cluster did not cover the full consensus sequence of that cluster. To generate such a consensus sequence Phrap v.0.990329 [30] was used with default parameters on each cluster from the most stringent dataset combination (data assembled with a quality cut-off of 0.001 on the short reads that had been processed to remove adapters). In the small proportion of cases where this introduced more than one sequence only the longest was analysed further. The identity of the consensus sequences was initially assessed by a BLAST homology search using default parameters in BLAST2GO [29]. Few sequences deposited in Genbank shared similarity with our assembled stick insect transcripts (S2 Fig) so we took a functional orthology approach using the eggNOG (evolutionary genealogy of groups: Non-supervised Orthologous Groups) classification system [31]. The Arthopoda (artNOG) HMM (Hidden Markov Model) files, members and annotations datasets (http://eggnogdb.embl.de/#/app/home) were downloaded to make a local HMMER (http://hmmer.org) database following the recommended procedure on the website. As reading frame was not known, transcript assemblies were translated into all six reading frames using BioPerl to generate amino acid sequences (including stops) with their transcript names appended with the frame. The resulting sequences were then searched with ‘hmmscan’ against the artNOG HMMER reference database to generate tabular output [31]. The highest bitscore was used to select one result per assembled transcript and data summaries were generated using a MySQL database.
SNP analysis
In order to assess allelic diversity at each locus we used a reciprocal mapping approach with short reads from each species being mapped back to transcript assemblies (loci) from itself, and the other species. It should be noted that the contig assembly process resulted in a consensus sequence from the transcript assemblies, and so some real allelic diversity would have been lost in that process. Short reads were mapped back to the longest contig from each transcript assembly cluster using the short read mapper Bowtie2 (Langmead and Salzberg 2012). We mapped short reads to all transcript assemblies simultaneously, preventing a read mapping to more than one transcript (c.f. sequential mapping). The resulting SAM files were then parsed so that only reads mapped over a certain length (12 nucleotides) were included in the subsequent SNP analysis, and a new SAM file was generated for each transcript assembly. Repeating this using 25 and 50 nucleotide length mappings had no significant effect on our findings (data not shown) so we present data for a single mapping here. The mappings from these individual SAM files were analysed with the variant detector VarScan v.2.3.2; [32], with a conservative threshold of 10% for variant frequency being taken as evidence for different alleles being present at a nucleotide position. The number of SNPs in a given transcript assembly was normalized by the transcript assembly length to get a SNP rate per nucleotide. The results of these analyses were visualised using R software [33].
Results
Data quantity and quality were similar for each stick insect species following Illumina high throughput DNA sequencing (Table 1). After adaptor and quality trimming, clustering of cDNA sequences resulted in more than 2,500 transcript assemblies per species (Table 1; Fig 2; S1 and S2 Tables). These transcript assemblies had similar length distributions (Fig 3a) meeting our assumptions of equal coverage in the two taxa. Data are deposited with the Dryad Digital Repository and publically available at http://datadryad.org (doi:10.5061/dryad.h5g60), or can be downloaded from http://evolves.massey.ac.nz/DNA_Toolkit.htm.
Length distributions of transcript assemblies produced from the cDNA sequence of two stick insects were similar. A log length frequency distribution plot used values rounded to 1 decimal place for the longest consensus sequence generated from each cluster. (B) Sequence divergence (measured by SNP density per nucleotide) observed when reads were mapped to ~2,600 loci (transcript assemblies). Loci without variation (SNP-free) were removed. The putative parental Clitarchus hookeri genome contains many loci with low allelic diversity. SNPs detected in less than 10% of the short reads were ignored but reads were included whether or not they passed the strand bias filter within VarScan. Only the longest assembled transcripts generated per cluster were included. (C) SNPs detectable on all transcript assemblies by BWA mapping to ~2,600 Acanthoxyla and Clitarchus transcript assemblies using VarScan with minimum variant frequency of 10% irrespective of strand filter results. The first violin of each color comprises all data, and the second excludes transcript assemblies with no sequence variation (SNP-free). Purple–Acanthoxyla reads mapped onto Acanthoxyla transcript assemblies; Pale green–Clitarchus reads mapped onto Acanthoxyla transcript assemblies; Pale purple–Acanthoxyla reads mapped onto Clitarchus transcript assemblies; Green–Clitarchus reads mapped onto Clitarchus transcript assemblies.
Manually curated subsample
A subsample comprising 10% of the >2,500 clusters from each species was manually curated to identify anomalies within our pipeline. We determined that the clustering algorithm was at least 97% accurate; only 7 Acanthoxyla geisovii and 4 Clitarchus hookeri clusters generated more than one transcript assembly. Alternate splice variants were included in many transcript assemblies, but these comprise <50 bp of the alternative exon. No instances of paralogous transcript assemblies within the mismatch parameters were identified. Mapping short sequence reads back to transcript assembly consensus sequences revealed more nucleotide variability than was seen in the short read assembled transcripts. Thus, comparison of the transcript assemblies was an inappropriate measure of the genetic variability between the two stick insect species. Open reading frame (ORF) prediction and GC content calculation indicated a positive relationship between ORF length and GC content (S1 Fig). By considering sequences with a GC content of 48% or more, it was possible to exclude almost all sequences that were not dominated by predicted ORF, and thus consider a dataset consisting predominantly of protein coding DNA (S3 Fig). Manually curating a subset of the data allowed us to identify genes from conserved families such as alpha-actinin (Table 2). Expression of this actin binding protein in muscle tissue revealed greater cDNA sequence variation in Acanthoxyla geisovii than in Clitarchus hookeri. Diploid Clitarchus hookeri had three nucleotide polymorphisms with equal frequency of reads, while Acanthoxyla geisovii had 26 substitutions where frequency of reads was close to 33%, as expected of a triploid (Table 2).
Within Acanthoxyla geisovii the SNP variant (allele) most similar to Clitarchus hookeri has approximately one third the expression level (inferred from read coverage) because this genome is triploid.
Full dataset
Only 13.3% of the assembled transcripts matched sequences in the National Center for Biotechnology Information (NCBI) non-redundant (nr) protein database, but 2/3 of these had their closest match to an arthropod sequence (S2 Fig, S3 and S4 Tables). A limited number of stick insect sequences readily matched published data using this approach [34]. The longest assembled transcript from both species was Twitchin as expected of mRNA derived from insect muscle tissue [35]. No human contamination was detected (S3 and S4 Tables). Our preliminary functional assessment using eggNOG found orthologous groups for 52.77% of the transcript assemblies (Table 3). For Acanthoxyla geisovii 1446 transcripts (55.19%) and for Clitarchus hookeri 1294 (50.31%) transcripts were assigned to functional groups. The individual eggNOG hits for the two species are available at http://evolves.massey.ac.nz/DNA_Toolkit.htm (Tables “domainSummary_Ac.txt” and “domainSummary_Cl.txt”). There was a unique set of 127 combinations of Cluster of Orthologous Groups (COG) codes in the artNOG dataset. Of these, 26 were the single category codes (18,077 of 18,837 (95.97%)) and the remaining 101 were non-single category codes (760 of 18,837 (4.23%)), indicating that nearly all the annotations were categorised into single category codes. An overview of the functional categories found from the assembled transcripts set was obtained using the COG classifications. In comparison to the full artNOG set, the stick insect transcript assemblies showed about 4-fold underrepresentation of categories L (Replication, recombination and repair), D (Cell cycle control, cell division, chromosome partitioning) and K (Transcription), and >4-fold overrepresentation of J (Translation, ribosomal structure and biogenesis), Z (Cytoskeleton) and C (Energy production and conversion) (Table 3). A higher frequency of genes involved in the cytoskeleton and energy production in our transcript assemblies than in the arthropod database is compatible with the source of mRNA being leg muscle.
COG = Cluster of Orthologous Groups.
A hybrid genome is expected to have higher heterozygosity than a parental genome. Thus we determined how many loci (transcript assemblies) contained no sequence variation (Single Nucleotide Polymorphism; SNPs) when raw reads were mapped back to the assembled transcripts compiled from the respective genome (Fig 2). The putative parental species Clitarchus hookeri had 1900/2572 (73.87%) loci lacking SNPs, whereas the putative hybrid Acanthoxyla geisovii had only 728/2620 (28.3%) of loci without SNPs. The low level of allelic variation within Clitarchus hookeri was evident in a frequency analysis (data not shown).
A parental genome is expected to nest within the diversity present in the hybrid genome. Thus when the putative hybrid (Acanthoxyla geisovii) reads were mapped back to the gene sequences from the putative parent (Clitarchus hookeri) we expected a similar proportion of loci to be homozygous as observed within the putative hybrid (28% SNP free loci). Our data did not meet these expectations as the proportion of homozygous loci was significantly lower (568/2572; 22.1%; chi-squared p < 0.0001), however, similarity will naturally decline over time since hybridisation. When sequence reads from the putative parent (Clitarchus hookeri) were mapped back to the assembled transcripts from the putative hybrid (Acanthoxyla geisovii) there were fewer SNP free loci than expected (456/2620; 17.4%; chi-squared p < 0.0001). This might be a product of the triploid genome of Acanthoxyla (Myers et al. 2013) and could be assessed with additional taxon sampling.
Repeating these mapping procedures using only putative protein coding sequences (transcript assemblies passing a 48% GC content threshold), produced in near identical results except for a reduction in the size of density peaks at 0 SNPs per nucleotide, and a general shift of the Clitarchus hookeri reads vs. Acanthoxyla geisovii transcript assembly curve on the x-axis towards slightly larger SNPs per nucleotide densities (S2 Fig).
The degree to which the sequences of alleles differed within a hybrid represents the genetic divergence of the two parental species. Therefore, allelic diversity between the maternal alleles and the paternal alleles can be observed within a single hybrid genome. This leads to the prediction that the difference between Acanthoxyla and Clitarchus alleles will be similar to the amount of allelic difference seen within Acanthoxyla, if Clitarchus is a parental taxon. By examining the number of SNPs as a proportion of the transcript assembly length we observed that alleles differed by a similar amount between the two stick insect genomes as they did within the putative hybrid (Acanthoxyla) genome (Fig 4), as expected. The symmetry of the heat map results from the similarity of sequence divergence within the hybrid and between the parent and hybrid alleles. As the divergence between the two subgenomes of the hybrid is higher than the diversity within a single parent (Fig 3b), we are able to distinguish the difference between alleles and paralogs.
Heat maps of sequence similarity between short-read cDNA sequences mapped to transcript assemblies. (A) Acanthoxyla transcript assemblies with Clitarchus reads (upper right) and Acanthoxyla reads (lower left). (B) Clitarchus transcript assemblies with Clitarchus reads (upper right) and Acanthoxyla reads (lower left).
Discussion
Are hybrids important?
Hybrid genomes are abundant and result from a range of processes; yeast fuse without sex [36], plant chimaeras are produced from grafting [37], and even some viral genomes appear to be the product of recombination between DNA and RNA viruses [38]. More than 70% of angiosperms are neo- or paleo-polyploids of which allopolyploids are most numerous [2], and the ancestor of vertebrates was involved in a whole genome duplication event [39]. So there is no doubt that many lineages are reticulate and that further tests of hybridisation are needed.
The phenotypic effects of hybridisation are many and varied and the role hybridisation plays in generating and reducing biodiversity is much debated [40, 41]. Furthermore, hybrid vigor has been associated with invasive species [42, 43], and conservation issues [44]. Thus the role and analysis of hybrid genomes are increasingly in the spotlight. The methods presented here provide the potential to explore any non-model organisms, even where ploidy level variation exists, as we have demonstrated by comparing diploid and triploid lineages. Distinguishing between paralogs and alleles is essential for downstream analyses, but where the divergence between two subgenomes of the hybrid is higher than the diversity within a single parent, then this is possible [45]. Artificial construction of hybrids, and recent natural hybrids reveal that alterations in gene expression can occur quickly due to many factors [46, 47], however, although this leads to important phenotypic variation it does not weaken our ability to test hybrid hypotheses using NGS. During manual curation of a subset of our data we found that expression levels as inferred from read coverage were strongly linked to ploidy level. A hybrid origin hypothesis sets up two testable predictions: 1. The putative hybrid will have greater allelic diversity than related non-hybrids; 2. Alleles of the parental taxa will co-occur in the genome of the hybrid. Both these predictions were met when we examine >2,500 loci expressed in Acanthoxyla geisovii (putative hybrid) and Clitarchus hookeri (putative paternal taxon).
Stick insect evolution
Stick insects produce successful hybrids because they readily reproduce via parthenogenesis, simultaneously providing both potential fitness advantage over parental taxa and reproductive isolation from those parents [19, 48–50]. Diverse reproductive mechanisms are employed by stick insects, including both hybridogenesis and androgenesis, where just one parental genome is transmitted to offspring without recombination, resulting in hemiclonal inheritance of maternal or paternal genotypes [21]. In addition, parthenogenesis with or without recombination (automixis, thelytoky) is common; [21, 49], and each stick insect taxon may include numerous independent origins of parthenogenetic lineages [3, 20, 49, 51, 52]. The genome of the stick insect Acanthoxyla geisovii fits our predictions of a hybrid, containing high allelic diversity and sharing alleles with its putative paternal taxon, Clitarchus hookeri. Although no males have been described from Acanthoxyla, eight morphologically distinct species are recognized. The genus may have multiple origins given the variation detected in nuclear markers from the full range of morphological diversity [8, 15]. However, the species do not form monophyletic clades, as expected of clonal taxa [18], and possibly much (or all) of the morphological variation arises from recombination within the hybrid genome (automixis). The distinctive character combinations could also result from loss of heterozygosity during the transition from triploid to diploid [18]. Alternatively, multiple hybridisation events between different Acanthoxyla lineages and different Clitarchus lineages might have been involved in generating diversity. Sampling across the range of Acanthoxyla phenotypes using multiple markers will help resolve this question. Variation is the indispensable basis of evolution but males may not be needed to generate variation in this system due to the allelic diversity we have observed within the hybrid genome.
Supporting Information
S1 Fig. Open reading frame (ORF) prediction and GC content from stick insect mRNA.
https://doi.org/10.1371/journal.pone.0154911.s001
(DOCX)
S2 Fig. The majority of transcript assemblies generated from cDNA from non-model organisms will find no match using BLAST searches against non-redundant (nr) protein databases, as in this example of two stick insects.
Those transcript assemblies that have matches are predominantly from well resourced insect species.
https://doi.org/10.1371/journal.pone.0154911.s002
(DOCX)
S3 Fig. Kernel density plot generated for the GC rich subset of data (48% or more).
Sequence divergence of stick insect protein coding DNA (measured by SNP density per nucleotide) observed when reads were mapped to loci (transcript assemblies). Putative parental genome (Clitarchus hookeri) contains many loci with no or low allelic diversity.
https://doi.org/10.1371/journal.pone.0154911.s003
(DOCX)
S1 Table. Short reads of cDNA from two New Zealand stick insects was assembled with a variety of different settings and software.
Assembly statistics for assembly combinations, the number of contigs, and overall contig length for each species and de novo assembler grouped by kmer and data trim type.
https://doi.org/10.1371/journal.pone.0154911.s004
(DOCX)
S2 Table. After running custom scripts to generate a unique subset of contigs for each stick insect species, the number of sequences was greatly reduced, as shown.
It should be noted that this process reduced the number of sequences contained within longer sequences; there was no reduction for any sequences that overlapped any other sequence.
https://doi.org/10.1371/journal.pone.0154911.s005
(DOCX)
S3 Table. The identity of transcript assemblies from Acanthoxyla geisovii was assessed by a BLAST homology search in which 13.3% matched sequences in the National Center for Biotechnology Information (NCBI) non-redundant (nr) protein database.
https://doi.org/10.1371/journal.pone.0154911.s006
(PDF)
S4 Table. The identity of transcript assemblies from Clitarchus hookeri was assessed by a BLAST homology search in which 13.3% matched sequences in the National Center for Biotechnology Information (NCBI) non-redundant (nr) protein database.
https://doi.org/10.1371/journal.pone.0154911.s007
(PDF)
Acknowledgments
Ted Trewick and Prasad Doddala collected the stick insects. PBRF allocation was facilitated by Susan Adams. Thanks to the Phoenix group (http://evolves.massey.ac.nz) and everyone from the Farside lab for lively discussions.
Author Contributions
Conceived and designed the experiments: MMR SAT. Performed the experiments: MMR SFKH PJB SAT. Analyzed the data: MMR SFKH PJB SAT. Contributed reagents/materials/analysis tools: MMR SFKH PJB SAT. Wrote the paper: MMR SAT SFKH PJB.
References
- 1. Bullini L. Origin and evolution of animal hybrid species. Trends in ecology & evolution. 1994;9(11):422–6. Epub 1994/11/01. pmid:21236911.
- 2. Koh J, Soltis PS, Soltis DE. Homeolog loss and expression changes in natural populations of the recently and repeatedly formed allotetraploid Tragopogon mirus (Asteraceae). BMC Genomics. 2010;11:97. Epub 2010/02/10. pmid:20141639; PubMed Central PMCID: PMCPMC2829515.
- 3. Scali V, Passamonti M, Marescalchi O, Mantovani B. Linkage between sexual and asexual lineages: genome evolution in Bacillus stick insects. Biological Journal of the Linnean Society. 2003;79(1):137–50.
- 4. Radtkey RR, Donnellan SC, Fisher RN, Moritz C, Hanley KA, Case TJ. When Species Collide: The Origin and Spread of an Asexual Species of Gecko. Proceedings: Biological Sciences. 1995;259(1355):145–52.
- 5. Vrijenhoek RC. Polyploid Hybrids: Multiple Origins of a Treefrog Species. Current Biology. 2006;16(7):R245–R7. pmid:16581499
- 6. Brasier C. Plant pathology: The rise of the hybrid fungi. Nature. 2000;405(6783):134–5.
- 7. Otto SP. The Evolutionary Consequences of Polyploidy. Cell. 2007;131(3):452–62. pmid:17981114
- 8. Buckley TR, Attanayake D, Park D, Ravindran S, Jewell TR, Normark BB. Investigating hybridization in the parthenogenetic New Zealand stick insect Acanthoxyla (Phasmatodea) using single-copy nuclear loci. Molecular phylogenetics and evolution. 2008;48(1):335–49. Epub 2008/03/28. pmid:18367411.
- 9. Joly S, Heenan PB, Lockhart PJ. A Pleistocene inter-tribal allopolyploidization event precedes the species radiation of Pachycladon (Brassicaceae) in New Zealand. Molecular phylogenetics and evolution. 2009;51(2):365–72. Epub 2009/03/04. pmid:19254769.
- 10. Kearney M. Hybridization, glaciation and geographical parthenogenesis. Trends in ecology & evolution. 2005;20(9):495–502. Epub 2006/05/17. pmid:16701426.
- 11. Adams KL. Evolution of Duplicate Gene Expression in Polyploid and Hybrid Plants. Journal of Heredity. 2007;98(2):136–41. pmid:17208934
- 12. Bulla GA, Luong Q, Shrestha S, Reeb S, Hickman S. Genome-wide analysis of hepatic gene silencing in mammalian cell hybrids. Genomics. 2010;96(6):323–32. Epub 2010/08/31. pmid:20801210.
- 13. Flowers JM, Burton RS. Ribosomal RNA Gene Silencing in Interpopulation Hybrids of Tigriopus californicus: Nucleolar Dominance in the Absence of Intergenic Spacer Subrepeats. Genetics. 2006;173(3):1479–86. pmid:16648582
- 14. Jewell TR, Brock P. A review of the New Zealand stick insects: new genera and synonymy, keys, and a catalogue. Journal of Orthoptera Research. 2002;11:189–97.
- 15. Morgan-Richards M, Trewick SA. Hybrid origin of a parthenogenetic genus? Molecular ecology. 2005;14(7):2133–42. Epub 2005/05/25. pmid:15910332.
- 16. Buckley TR, Attanayake D, Nylander JAA, Bradler S. The phylogenetic placement and biogeographical origins of the New Zealand stick insects (Phasmatodea). Systematic Entomology. 2010;35(2):207–25.
- 17. Trewick SA, Morgan-Richards M, Collins LJ. Are you my mother? Phylogenetic analysis reveals orphan hybrid stick insect genus is part of a monophyletic New Zealand clade. Molecular phylogenetics and evolution. 2008;48(3):799–808. Epub 2008/07/18. pmid:18632289.
- 18. Myers SS, Trewick SA, Morgan-Richards M. Multiple lines of evidence suggest mosaic polyploidy in the hybrid parthenogenetic stick insect lineage Acanthoxyla. Insect Conservation and Diversity. 2013;6(4):537–48.
- 19. Milani L, Ghiselli F, Pellecchia M, Scali V, Passamonti M. Reticulate evolution in stick insects: the case of Clonopsis (Insecta Phasmida). BMC Evolutionary Biology. 2010;10(1):1–15.
- 20. Ghiselli F, Milani L, Scali V, Passamonti M. The Leptynia hispanica species complex (Insecta Phasmida): polyploidy, parthenogenesis, hybridization and more. Molecular ecology. 2007;16(20):4256–68. pmid:17725570
- 21. Mantovani B, Scali V. Hybridogenesis and Androgenesis in the Stick-Insect Bacillus rossius- Grandii benazzii (Insecta, Phasmatodea). Evolution. 1992;46(3):783–96.
- 22. Morgan-Richards M, Trewick SA, Stringer IA. Geographic parthenogenesis and the common tea-tree stick insect of New Zealand. Molecular ecology. 2010;19(6):1227–38. Epub 2010/02/19. pmid:20163549.
- 23.
Parfitt RG. Cytology and Feulgen—DNA microdensitometry of New Zealand stick insetcs (Phasmatatodea: Phasmatidae). Wellington, New Zealand: Victoria University of Wellington; 1980.
- 24. Cox MP, Peterson DA, Biggs PJ. SolexaQA: At-a-glance quality assessment of Illumina second-generation sequencing data. BMC Bioinformatics. 2010;11:485. Epub 2010/09/30. pmid:20875133; PubMed Central PMCID: PMCPMC2956736.
- 25. Zerbino DR, Birney E. Velvet: algorithms for de novo short read assembly using de Bruijn graphs. Genome Res. 2008;18(5):821–9. Epub 2008/03/20. pmid:18349386; PubMed Central PMCID: PMCPMC2336801.
- 26. Simpson JT, Wong K, Jackman SD, Schein JE, Jones SJ, Birol I. ABySS: a parallel assembler for short read sequence data. Genome Res. 2009;19(6):1117–23. Epub 2009/03/03. pmid:19251739; PubMed Central PMCID: PMCPMC2694472.
- 27. Chen F, Mackey AJ, Stoeckert CJ Jr, Roos DS. OrthoMCL-DB: querying a comprehensive multi-species collection of ortholog groups. Nucleic Acids Res. 2006;34(Database issue):D363–8. Epub 2005/12/31. pmid:16381887; PubMed Central PMCID: PMCPMC1347485.
- 28. Li L, Stoeckert CJ Jr, Roos DS. OrthoMCL: identification of ortholog groups for eukaryotic genomes. Genome Res. 2003;13(9):2178–89. Epub 2003/09/04. pmid:12952885; PubMed Central PMCID: PMCPMC403725.
- 29. Conesa A, Gotz S, Garcia-Gomez JM, Terol J, Talon M, Robles M. Blast2GO: a universal tool for annotation, visualization and analysis in functional genomics research. Bioinformatics. 2005;21(18):3674–6. Epub 2005/08/06. pmid:16081474.
- 30. de la Bastide M, McCombie WR. Assembling genomic DNA sequences with PHRAP. Curr Protoc Bioinformatics. 2007;Chapter 11:Unit11 4. Epub 2008/04/23. pmid:18428783.
- 31. Huerta-Cepas J, Szklarczyk D, Forslund K, Cook H, Heller D, Walter MC, et al. eggNOG 4.5: a hierarchical orthology framework with improved functional annotations for eukaryotic, prokaryotic and viral sequences. Nucleic Acids Research. 2015.
- 32. Koboldt DC, Zhang Q, Larson DE, Shen D, McLellan MD, Lin L, et al. VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Res. 2012;22(3):568–76. Epub 2012/02/04. pmid:22300766; PubMed Central PMCID: PMCPMC3290792.
- 33.
Team RDC. A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2012. Available: http://www.R-project.org.
- 34. Gibson AK, Smith Z, Fuqua C, Clay K, Colbourne JK. Why so many unknown genes? Partitioning orphans from a representative transcriptome of the lone star tick Amblyomma americanum. BMC Genomics. 2013;14:135-. PMC3616916. pmid:23445305
- 35. Dunning LT, Dennis AB, Park D, Sinclair BJ, Newcomb RD, Buckley TR. Identification of cold-responsive genes in a New Zealand alpine stick insect using RNA-Seq. Comparative Biochemistry and Physiology Part D: Genomics and Proteomics. 2013;8(1):24–31.
- 36. Morales L, Dujon B. Evolutionary role of interspecies hybridization and genetic exchanges in yeasts. Microbiol Mol Biol Rev. 2012;76(4):721–39. Epub 2012/12/04. pmid:23204364; PubMed Central PMCID: PMCPMC3510521.
- 37. Stegemann S, Bock R. Exchange of genetic material between cells in plant tissue grafts. Science. 2009;324(5927):649–51. Epub 2009/05/02. pmid:19407205.
- 38. Diemer G, Stedman K. A novel virus genome discovered in an extreme environment suggests recombination between unrelated groups of RNA and DNA viruses. Biology Direct. 2012;7(1):13.
- 39. Dehal P, Boore JL. Two Rounds of Whole Genome Duplication in the Ancestral Vertebrate. PLoS Biol. 2005;3(10):e314. pmid:16128622
- 40. Mallet J. Hybridization, ecological races and the nature of species: empirical evidence for the ease of speciation. Philosophical Transactions of the Royal Society B: Biological Sciences. 2008;363(1506):2971–86.
- 41. Seehausen O, Takimoto G, Roy D, Jokela J. Speciation reversal and biodiversity dynamics with hybridization in changing environments. Molecular ecology. 2008;17(1):30–44. Epub 2007/11/24. pmid:18034800.
- 42. Lee CE. Evolutionary genetics of invasive species. Trends in ecology & evolution. 2002;17(8):386–91.
- 43. Morgan-Richards M, Trewick SA, Chapman HM, Krahulcova A. Interspecific hybridization among Hieracium species in New Zealand: evidence from flow cytometry. Heredity (Edinb). 2004;93(1):34–42. Epub 2004/05/13. pmid:15138450.
- 44. Allendorf FW, Leary RF, Spruell P, Wenburg JK. The problems with hybrids: setting conservation guidelines. Trends in ecology & evolution. 2001;16(11):613–22.
- 45. Tang H, Bowers JE, Wang X, Paterson AH. Angiosperm genome comparisons reveal early polyploidy in the monocot lineage. Proceedings of the National Academy of Sciences. 2010;107(1):472–7.
- 46. Chelaifa H, Monnier A, Ainouche M. Transcriptomic changes following recent natural hybridization and allopolyploidy in the salt marsh species Spartina x townsendii and Spartina anglica (Poaceae). New Phytol. 2010;186(1):161–74. Epub 2010/02/13. pmid:20149114.
- 47. Paun O, Bateman RM, Fay MF, Luna JA, Moat J, Hedren M, et al. Altered gene expression and ecological divergence in sibling allopolyploids of Dactylorhiza (Orchidaceae). BMC Evol Biol. 2011;11:113. Epub 2011/04/28. pmid:21521507; PubMed Central PMCID: PMCPMC3112086.
- 48. Law JH, Crespi BJ. The evolution of geographic parthenogenesis in Timema walking-sticks. Molecular ecology. 2002;11(8):1471–89. Epub 2002/07/30. pmid:12144667.
- 49. Schwander T, Crespi BJ. Multiple direct transitions from sexual reproduction to apomictic parthenogenesis in Timema stick insects. Evolution. 2009;63(1):84–103. Epub 2008/09/23. pmid:18803687.
- 50. Scali V, Tinti F, Mantovani B, Marescalchi O. Mate recognition and gamete cytology features allow hybrid species production and evolution in Bacillus Stick-insects. It J Zool. 1995;62.
- 51. Ghiselli F, Milani L, Scali V, Passamonti M. The Leptynia hispanica species complex (Insecta Phasmida): polyploidy, parthenogenesis, hybridization and more. Molecular ecology. 2007;16.
- 52. Schwander T, Henry L, Crespi Bernard J. Molecular Evidence for Ancient Asexuality in Timema Stick Insects. Current Biology. 2011;21(13):1129–34. pmid:21683598