Insect Phylogenomics

Abstract

With the advent of next-generation sequencing methods, phylogenetics has taken a new turn in the recent years. Phylogenomics, the integration of phylogenetics with genome data, has emerged as a powerful approach to study systematics and evolution of species. Recently, breakthrough researches employing phylogenomic tools have provided better insights into the timing and pattern of insect evolution. The next-generation sequencing methods are now increasingly used by entomologists to generate genomic and transcript sequences of various insect species and strains. These data provide opportunities for comparative genomics and large-scale multigene phylogenies of diverse lineages of insects. Phylogenomic investigations help us better understand systematic and evolutionary relationships of insect species that play important roles as herbivores, predators, detritivores, pollinators, or disease vectors. It is important that we critically assess the prospects and limitations of phylogenomic methods. In this review, I describe the current status, outline the major challenges, and remark on potential future applications of phylogenomic tools in studying insect systematics and evolution.

Introduction

As species share a common history through their ancestry, reconstructing evolutionary history provides information about systematic relationships among species. Traditionally, morphological or ultrastructural characters were exploited in phylogenetic reconstruction. The major groups of animals and plants on earth have been delineated by using comparative anatomy of fossils and extant. However, the number of reliable morphological features as homolog characters among species remains a major disadvantage of this approach. With the introduction of DNA sequencing in the early 1970s, the application of molecular data in reconstructing phylogenies has gained huge popularity. The small subunit ribosomal RNA gene is used as reference for constructing phylogenetic trees due to considerable sequence conservation among species (Olsen and Woese 1993). Subsequently, extensive studies have been conducted to combine multiple genes and/or morphology data to infer phylogenetic relationships among species (Legg 2013). Since the last decade or so, genomic revolution has ushered in application of large scale multigene data in phylogenetics. The use of genome sequences to infer species phylogenies constitutes the central theme of phylogenomics. It is the intersection of the fields of evolution and genomics for reconstructing evolutionary histories among species. This is an emerging area that integrates large scale sequence data, computational tools and evolutionary principles to study phylogenetics (Blair and Murphy 2011). Phylogenomics separates from phylogenetics primarily in the input data. Unlike phylogenetics, phylogenomics investigations make use of large scale genomic or transcriptomic data to generate the input data based on orthology prediction of genes or whole-genome alignments.

For many lineages of insects, the resolution of the true phylogeny has been a challenge. One example is the Polyneopteran lineage. The branching patterns of the Polyneopteran insects such as cockroaches, mantids, earwigs, grasshoppers, and phasmids are highly ambiguous (Wipfler et al. 2014). The Polyneoptera are composed of 11 orders: Blattodea, Dermaptera, Embiodea, Grylloblattodea, Isoptera, Mantodea, Mantophasmatodea, Orthoptera, Phasmatodea, Plecoptera and Zoraptera. Establishing relationships among these orders has been difficult (Kristensen 1991). Using phylogenomic approaches, new insights are beginning to emerge on the internodal relationships of these insects (Wipfler et al. 2014). Moreover, application of phylogenomics in insects is appropriate because of the reason that insects are highly diverse. With about one million identified species, insects represent the highest proportion (∼75%) of all animals (May 1988, Foottit and Adler 2009). Hence, phylogenomics analyses are important in entomology because they can help us better understand evolutionary relationships of insects that are relevant to health, environment and agriculture. Moreover, during the last several years, genome sequencing of numerous insect species has been initiated, in addition to completion of almost hundreds of insect genomes. These are opening up new avenues for identification of orthologs for large-scale multigene phylogenetic studies. With the rising trends in insect genomics and transcriptomic studies, phylogenomic studies are gaining popularity that has potential to provide new insights into true phylogenies of many insect lineages.

Phylogenomic methods

Phylogenomic analyses are broadly carried out using three methods: 1) sequence alignment, 2) comparison of occurrences of ‘DNA strings’, and 3) comparison of gene content or order in the genome. The sequence alignment method is the most widely used approach in phylogenomic investigations. This approach relies on an accurate prediction of single copy orthologs representing genes in different species that have evolved from a common ancestral gene via speciation. Multiple alignment of orthologous genes are carried out either using individual orthologs or their concatenated sequences. The phylogenomic inferences are made from constructing a supertree of the individual genes (Bininda-Emonds and Sanderson 2001) or from supermatrix data that is generated by the alignment of concatenated sequences (Delsuc et al. 2005, Savard et al. 2006). The supertree based method relies on genes from overlapping taxa, whereas in the supermatrix method genes from non-overlapping taxa are treated as missing data. The independent evolutionary rates of multigene sequences are accommodated by methods such as partitioned-likelihoods to infer the phylogenetic relationships among the species. In addition to the sequence alignment based method, DNA strings (distribution patterns of short sequences among genomes) and random anchor (a longer stretch of nucleotides) are also used in phylogenomics studies, particularly among the closely related species (Vishnoi et al. 2010).

The DNA string method does not require sequences to be aligned, and hence has an advantage over sequence alignment method. It is based on the frequency of short-oligonucleotide sequence combinations in the genomes that is used to directly compare higher order features of non-homologous DNA sequences (Edwards et al. 2002). The DNA string is also an appropriate method to delineate phylogenetic inference across deeper taxonomic levels where genomic homology is difficult to establish (Qi et al. 2004). The gene content and gene order information are also applicable in performing phylogenomic inferences (Murphy et al. 2004, Bourque and Tesler 2008). Gene order information are determined from scoring the presence or absence of pairs of orthologous genes that is then used as the measure of genetic distance between species (Wolf et al. 2001, Zhang et al. 2009). Gene rearrangement and duplications are critical factors that influence the accuracy of phylogenies based on gene orders. Recently, Hu et al. (2014) developed a maximum likelihood approach that takes into account of gene rearrangements, insertions, deletions and duplications for reconstruction of phylogeny from gene-order data. Phylogenomic studies using DNA strings or gene order methods are infrequent compared to those using sequence alignment methods. Although the majority of insect phylogenomics studies have been carried out using the sequence alignment methods, inference of phylogenies based on rearrangement of genomic regions, also known as ‘chromosomal phylogeny’, have provided useful information about evolution of specific insect lineages such as the species complex of Anopheles gambiae (Kamali et al. 2012, Sharakhov et al. 2013).

Current status of insect phylogenomics

In the recent years, phylogenomics studies have provided valuable insights into evolutionary relationships of insects. Numerous phylogenomic studies have been performed in diverse lineages of insects that have addressed several fundamental issues of classification and evolution of insect species. A non-comprehensive list of recent phylogenomic studies of insects is provided in Table 1 along with the study objectives. The expressed sequence tags (ESTs) (Behura 2006) and whole transcriptome data (Trautwein et al. 2012) are predominantly used to develop data for phylogenomic investigations of many insects. Although the application of EST datasets have been useful for phylogenetic analysis (Theodorides et al. 2002, Hughes et al. 2006, Parkinson and Blaxter 2009, Meusemann et al. 2010), they may often bias the results. This is because the representation of these data to the coding sequences may vary between species. It has been shown that EST-based phylogenies may also lead to inconsistent results when compared with the known speciation events and life history of the organisms (Andrew 2011).

Table 1.

Recent developments in insect phylogenomics. A list of studies reported within the last 2 years only are provided^*.

Study objective	No. of species/taxa	No. of genes	Reference
Pattern and time of insect evolution	144	1478	Misof et al. 2014
Phylogeny of aculeate hymenoptera	18	308	Johnson et al. 2013
Phylogeny of malaria mosquitoes	6	49	Kamali et al. 2014
Phylogeny of malaria mosquitoes	43	1085	Neafsey et al. 2015
Phylogeny of malaria mosquitoes	8	whole-genome alignment	Fontaine et al. 2015
Phylogeny of lower neopteran orders	48	229	Letsch and Simon 2013
Phylogeny of holometabolan insects	13	1,343	Peters et al. 2014
Phylogeny of Lepidoptera	46	2,696	Kawahara and Breinholt 2014
Phylogeny of wingless insects	73	1,866	Dell'Ampio et al. 2014
Phylogeny of Neuropteroidea	36	668	Boussau et al. 2014

^*The list may not be complete.

Currently, multigene datasets and whole-transcriptome data are extensively used in phylogenomic analyses (Chan and Ragan 2013). In a recent publication, Misof et al (2014) conducted a large-scale phylogenomic analysis of 144 insect species that provided a holistic view about origin and evolution of insects. Using 1478 orthologous genes, this study suggested that insects originated approximately 479 million years ago. By partitioning transcriptome sequence data into protein domains, Misof et al. (2014) used maximum likelihood models to show diversification of insects into four groups: Palaeoptera. Polyneoptera, Condylognatha and Holometabola. The study further showed that diversification within modern winged insects started in the Paleozoic era, and suggested that insect flight occurred approximately 406 million years ago. This study has clarified our understanding of insect origin and evolution and arguably has laid foundation towards building the insect tree-of-life in finer detail (Jones 2015).

Phylogenomics has also been helpful in resolving species relationship of specific lineages of insects. Species phylogenies of mosquitoes is one of the ongoing goals of vector biologists. Over 3,500 species, belonging to at least 43 genera, of the Anophelinae and Culicinae mosquitoes are known. Several species of these mosquitoes act as global vectors of malaria, dengue, West Nile virus and others. In addition to a rich literature on classical and molecular systematics of mosquitoes (Munstermann and Conn 1997), evolutionary investigations of Anopheles gambiae (malaria vector), Aedes aegypti (dengue vector) and Culex quinquefasciatus (lymphatic fialriasis and West Nile virus vector) have gained a lot of momentum in the recent years after the availability of genome sequences of these species (Severson and Behura 2012, Neafsey et al. 2013, Kamali et al. 2014, Fontaine et al. 2015). Using a combined dataset of six nuclear protein-coding genes and an array of morphological characters (n =80), Reidenbach et al (2009) performed maximum parsimony and maximum likelihood analyses to understand phylogenetic radiation among mosquitoes representing 25 genera. Their analysis provided a renewed and stronger evidence for the basal position of the Anophelinae subfamily. It was further suggested from this study that divergence times for major culicid lineages might date back to the early Cretaceous. The recent study by Kamali et al (2014) further shows that Anopheles nili occupies the basal clade that diversified from other studied malaria mosquito species some 47.6 million years ago. Recently, 16 Anopheles species have been sequenced (Neafsey et al 2015), and this has set the stage for phylogenomic investigation of malaria vector mosquitoes. Fontaine et al (2015) performed a phylogenomic analysis among eight Anopheles species and showed that lineages leading to the principal vectors of human malaria were among the first to split within the species complex. Their data further revealed extensive introgression in autosomes that may have important implications in vectorial capacity of Anopheles to transmission of malaria. Recent studies also demonstrate the confounding effects of introgression and shared mutations on applying genomic data in inferring true speciation histories of Anopheles gambiae species complex (O'Loughlin et al. 2014, Crawford et al. 2014). Introgression, an important source of genetic variation in natural populations, is caused by stable integration of genetic material from one species into another through repeated back-crossing. Introgression is particularly a problem in phylogeny reconstruction of among closely related lineages partly due to the fact that commonly sequenced genetic markers often lack sufficient phylogenetic signal at the lowest taxonomic levels. When introgression occurs, a substantial fraction of their genomes can be permeable to alleles from related species. Discordant genealogies of closely related species can also occur by incomplete lineage sorting where two lineages fail to coalesce within a population. Incomplete lineage sorting generally arises from stochastic coalescence that leads to mask the signatures of true phylogeny. Distinguishing introgression from incomplete lineage sorting is important in evolutionary studies of closely related species. Topology based phylogeny discordance, test of isolation model (speciation with no gene flow) or detection of gene flow are commonly used methods to distinguish incomplete lineage sorting from genetic introgression.

Phylogenomics of Polyneoptera is gaining a lot of interest in recent times. Employing large-scale EST datasets, Letsch et al (2012) performed phylogenomics analysis among Polyneopteran and Paraneopteran insects and showed that Polyneoptera and Eumetabola (Paraneoptera + Holometabola) have a monophyletic origin. In parallel to this study, Simon et al (2012) performed further phylogenomic analyses among three polyneopteran orders; Dermaptera, Plecoptera, and Zoraptera, the results of which provided conclusive support for monophyletic Polyneoptera. However, later results of Letsch and Simon (2013) rejected the previous classification of Parametabola (= Zoraptera + Paraneoptera), Mystroptera (= Embioptera + Zoraptera) and Orthopterida (= Orthoptera + Phasmatodea) and indicated that several polyneopteran orders still show unstable positions within a monophyletic Polyneoptera.

Phylogenomics investigations have aided resolution of holometabolous insects. Savard et al. (2006) conducted a study using 185 orthologous genes among six species (Drosophila melanogaster, Anopheles gambiae, Bombyx mori, Tribolium castaneum, Apis mellifera, Nasonia vitripennis and Nasonia giraulti) representing four insect orders (diptera, lepidoptera, coleopteran and hymenoptera). The results of this phylogenomics study suggested that bees and wasps retain the basal position of holometabolous insects' phylogeny. Later, maximum-likelihood tree generated from concatenated sequences of 1,150 single-copy orthologs among 10 metazoan species (Richards et al. 2008) also confirmed the results of Savard et al. (2006). A recent phylogenomics study (Peters et al. 2014) exploited de novo transcriptome data to study holometabolous insects. By analyzing 1,343 single-copy orthologous genes among 13 species, this study suggested that hymenopteran insects belong to a sister group of other holometabolan insects that comprises Mecopterida and Neuropteroidea. The results of this study strongly supported the relations of Raphidioptera + (Neuroptera  +  monophyletic Megaloptera), and Diptera + (Siphonaptera + Mecoptera) within Neuropterida and Antliophora.

The order Lepidoptera has shown several ambiguities in the phylogenetic placement of specific species such as moths and butterflies. Kawahara and Breinholt (2014) performed a phylogenomics study based on 2,696 single copy ortholog genes among 46 taxa that provided strong evidence that butterflies and moths have shared evolutionary relationships. The study revealed monophyly of butterflies with Hesperiidae (skippers) and Hedylidae (moth-butterflies), and provided support for placing butterflies sister to the obtectomeran Lepidoptera. Similarly, the aculeate Hymenoptera that are extensively investigated for eusocial behavior of many species such as ants, bees and stinging insects has unclear phylogenies. Johnson et al. (2013) performed phylogenomic analysis among 18 species by exploiting gene portioning from transcriptome data. They found that the eusocial insects are contained within two major groups which are interpolated among three other clades of wasps. It also showed that ants are the sister group of spheciform wasps and bees (Apoidea).

Major challenges

Although significant progress has been made in the field of phylogenomics, several factors have been identified that impose challenges in these studies. Here, I discuss three broad areas that are often problematic in phylogenomic studies: A) accurate prediction of gene orthology, B) gene tree heterogeneity, and C) assumption and statistics. A prerequisite step in preparing phylogenomic data generally involves identification of 1:1 orthologous genes among the species. This has been a challenging step particularly when genes are present in duplicated copies within genome (Jensen 2001, Dalquen et al. 2013), and also for the reason that building orthology models relies upon all-against-all gene comparisons (Sonnhammer et al. 2014). Similarly, the presence of horizontally transferred genes further confound the problem of identification of orthologs (Dalquen et al. 2013). Furthermore, the number of single copy ortholog groups sharply decreases as we consider more diverse species (Figure 1). This is based on known orthologies predicted among different insect species where genome sequences are available (Waterhouse et al. 2013). Thus, the number of single copy orthologous genes that can be used in the phylogenomic analysis becomes a limiting factor to carry out phylogenomic investigations among distantly related species. Besides single copy orthologs, phylogenomics studies are also carried out with partitioning the data into predicted protein domains. This approach may also have a limitation in accurate partitioning of sequence data into orthologous domains when the number of species compared is large and the size of protein domains is small (Storm and Sonnhammer 2003). Furthermore, differential evolution and architecture of protein domains (Buljan and Bateman 2009) may impact identification of orthologous protein domains.

Figure 1.

Number of single copy orthologs among different insect species. The screenshot image shows phylogenetic grouping of different insect species where genome sequences are available (OrthoDB7, http://cegg.unige.ch/orthodb7). The vertical lines on the right represent different groups with number of species on the top and number of single copy orthologous genes on the bottom.

Accounting for heterogeneity among gene trees is a major challenge in phylogenomics investigations. Discrepancies between gene trees and species trees is well known that imposes uncertainty in predicting species phylogenies based on multigene datasets (Maddison 1997). Several factors such as recombination, hybridization and introgression (Siepel 2009), gene duplication (Page and Charleston 1997), horizontal gene transfer (Doolittle 1999) and incomplete lineage sorting (Pamilo and Nei 1988) may contribute to heterogeneity among gene trees. Gene tree heterogeneity is a problem in phylogenomics because genealogical histories often vary among different genes throughout the genome (Degnan and Rosenberg 2009). The discrepancy in the inferred gene trees and the actual species tree stems from different coalescent events among genes due to differential lineage sorting, selection and drift of ancestral polymorphisms in the orthologous genes (Rannala and Yang 2008). Incongruence, the topological conflict between different gene trees, is often a critical issue of phylogenomics studies. Several recent studies have addressed the issue of incongruence and suggested approaches to overcome them (Salichos and Rokas 2013, Salichos et al. 2014, Dell'Ampio et al. 2014).

Long-branch attraction (where species with high evolutionary rates tend to group together) and heterotachy (site specific variation of evolutionary rate) can also be potential causes of inaccurate phylogenies (Zhang et al. 2009). In a recent study, Boussau et al. (2014) showed that correcting long-branch attraction by usage of appropriate models accurately places Strepsiptera as a sister group to Coleoptera. Although the long-branch attraction problem was initially thought to be associated with parsimony methods (Felsenstein 1978), study (Huelsenbeck and Lander 2003) has suggest that this problem is frequent, and may arise with other methods as well (Swofford et al. 2001). Heterotachy is a potential problem in phylogenomics when an appropriate model is not implemented to account for site-specific changes in evolutionary rates across the tree. The homotachous techniques may produce inaccurate phylogenies as these models assume that relatively fast-evolving sites are fast across the entire tree, whereas slower sites always evolve at relatively slower rates. Kolaczkowski and Thornton (2008) used simulation to show that the mixed branch length model is more accurate than homotachous techniques. In addition to selection of appropriate evolutionary models, insufficient sequence data (Philippe et al. 2004) and sampling bias (Driskell et al. 2004) may also lead to conflicting phylogenetic inferences. By analyzing ∼300,000 protein sequences across eukaryotes and prokaryotes, Driskell et al. (2004) showed that the ‘supermatrix’ approach may provide useful insights into broad sections of the tree of life even with incomplete datasets.

The third but serious concern in phylogenomics study stems from statistical analyses. The inaccurate results primarily originate from unrealistic assumptions and statistical methods of multigene phylogenies. For example, when genes are acted upon by episodic positive selection in specific lineages, the statistical inference of evolutionary relationships can be misleading as the amount of sequence data required to accommodate the evolutionary model is invariably underestimated (Kumar et al. 2012). Similarly, bias in data matrix composition is known to mislead phylogenomic inferences. The phylogenomic investigation by Letsch et al (2012), using expressed sequence tags of different Polyneoptera and Paraneoptera species, clearly demonstrated the effect of matrix composition on phylogenetic placement of Pediculus humanus, the human body louse. It was observed that biased gene overlap resulting from orthology predictions from transcript data was the reason for inaccurate placement of the human louse along with Holometabola. This study suggested that when transcript sequences used in phylogenomic analyses are not comprehensive for each genome, the results may have potential pitfalls.

Future prospects and concluding remarks

Despite the inherent challenges, phylogenomic approaches have provided valuable information about systematics and the evolutionary history of insects. Within the last two years, several phylogenomic investigations have provided new insights into insect evolution of different lineages. The evolution of eusocialiaty (Johnson et al. 2013) and vectorial ability of malaria mosquitoes (Fontaine et al. 2015) are recent examples of this progress.

In the recent years, phylogenomics tools have been applied in diverse domains of molecular evolutionary studies ranging from predicting gene function, studying evolutionary patterns of macromolecules and molecular adaptation, and resolving relationships and divergence times of genes and species (Kumar et al. 2012). The emerging concept of ‘integrative phylogenomics’ is now gaining popularity in diverse domains of evolutionary biology (Bapteste and Burian 2010). Integration of anatomical data generated by the non-invasive three dimensional reconstruction techniques along with fossil and genomic data (Giribet and Edgecombe 2012) is a clear example of that concept. Combining genomic data from protein coding and non-coding genes is another way of integrating data for phylogenomics analyses (Rota-Stabelli et al. 2010). One example would be the application of microRNA gene sequences in phylogenomics studies (Rota-Stabelli et al. 2010, Campbell et al. 2011). MicroRNA genes are appropriate for phylogenomics investigations because 1) their stem-loop precursors are associated with differential nucleotide diversity among different species groups (Behura 2007), 2) they mostly lack convergent evolution (Sperling and Peterson 2009), 3) their birth rate is higher than death rate in most metazoan taxa (microRNA families are continually added to metazoan genomes), and 4) generating genome-wide microRNA sequences is becoming very routine these days via next-generation sequencing.

Application of phylogenomic tools to predict gene functions (Eisen 1998, Sjölander 2004, Brown and Sjölander 2006) is an important area of research that may find utilities in future phylogenomic studies of insects. The phylogenomic approaches to study functional adaption of a large number of genes may be beneficial compared to the phylogenetic approaches that rely on the correlation between single gene tree and the associated trait (gene function) (Pagel 1999). Besides, phylogenomic tools are also applicable in studying evolutionary histories of endosymbionts (Comas et al. 2007, Kembel et al. 2011). They may also find utility in cell biology to define evolutionary origins of cell lineages. This is a particular area that holds exceptional promise in cell fate mapping projects of various organisms. While phylogenetics approaches have been successfully applied in mapping cell fates (Salipante and Horwitz 2007), recent advances in single-cell sequencing (Shapiro et al. 2013, Liang et al. 2014) is expected to aid phylogenomic approaches to reconstruct cell lineages. These specific examples not only indicate the broad utilities of phylogenomic tools but also attest that they can be harnessed in deeper understanding of biology and evolution in a holistic manner. In particular, the explosion of genome sequencing efforts of 5,000 insects and other arthropods (Robinson et al. 2011, i5K consortium 2013) is expected to aid integration of phylogenomics tools to provide new insights into systematics and evolutionary relationships of several unresolved or poorly resolved insect lineages. With no doubt, insect phylogenomics holds huge promise in understanding insect biology in evolutionary terms; and sooner or later, it is possible that they may unravel some of the many secrets of why insects are so diverse and adaptive in nature.