An integrative phylogenomic approach to elucidate the evolutionary history and divergence times of Neuropterida (Insecta: Holometabola)

Abstract

Background: The latest advancements in DNA sequencing technologies have facilitated the resolution of the phylogeny of insects, yet parts of the tree of Holometabola remain unresolved. The phylogeny of Neuropterida has been extensively studied, but no strong consensus exists concerning the phylogenetic relationships within the order Neuroptera. Here, we assembled a novel transcriptomic dataset to address previously unresolved issues in the phylogeny of Neuropterida and to infer divergence times within the group. We tested the robustness of our phylogenetic estimates by comparing summary coalescent and concatenation-based phylogenetic approaches and by employing different quartet-based measures of phylogenomic incongruence, combined with data permutations.

Results: Our results suggest that the order Raphidioptera is sister to Neuroptera + Megaloptera. Coniopterygidae is inferred as sister to all remaining neuropteran families suggesting that larval cryptonephry could be a ground plan feature of Neuroptera. A clade that includes Nevrorthidae, Osmylidae, and Sisyridae (i.e. Osmyloidea) is inferred as sister to all other Neuroptera except Coniopterygidae, and Dilaridae is placed as sister to all remaining neuropteran families. Ithonidae is inferred as the sister group of monophyletic Myrmeleontiformia. The phylogenetic affinities of Chrysopidae and Hemerobiidae were dependent on the data type analyzed, and quartet-based analyses showed only weak support for the placement of Hemerobiidae as sister to Ithonidae + Myrmeleontiformia. Our molecular dating analyses suggest that most families of Neuropterida started to diversify in the Jurassic and our ancestral character state reconstructions suggest a primarily terrestrial environment of the larvae of Neuropterida and Neuroptera.

Conclusion: Our extensive phylogenomic analyses consolidate several key aspects in the backbone phylogeny of Neuropterida, such as the basal placement of Coniopterygidae within Neuroptera and the monophyly of Osmyloidea. Furthermore, they provide new insights into the timing of diversification of Neuropterida. Despite the vast amount of analyzed molecular data, we found that certain nodes in the tree of Neuroptera are not robustly resolved. Therefore, we emphasize the importance of integrating the results of morphological analyses with those of sequence-based phylogenomics. We also suggest that comparative analyses of genomic meta-characters should be incorporated into future phylogenomic studies of Neuropterida.

Keywords: Endopterygota; Four-cluster likelihood mapping; Megaloptera; Multi-species coalescent; Neuroptera; RNA-seq; Raphidioptera; Supermatrices; Transcriptomics.

Fig. 1

Phylogenetic relationships of Neuropterida based on the analyses of the concatenated amino-acid sequence data of supermatrix E. Colored circles depict phylogenetic branch support values based on 100 non-parametric bootstrap replicates. Bars on the individual nodes show the 95% confidence intervals (equal-tail CI) of the posterior divergence time estimates. Blue squares indicate the time-calibrated nodes. Divergence time estimates were calculated from a single summarized MCMC chain (first independent analysis, run 1) that included all parameter values from each individual meta-partition analysis when including all fossil calibrations. Insect photos from top to bottom: Dichrostigma flavipes, Sialis lutaria, Chrysopa perla (all photos by O. Niehuis)

Fig. 2

Gene tree-based and concatenation-based quartet analyses of the phylogenetic relationships of Neuropterida. a Phylogenetic relationships of Neuropterida, as they resulted from the summary coalescent phylogenetic analysis with ASTRAL, when analyzing the full set of gene trees (3983 gene trees inferred at the amino-acid sequence level). Pie charts on branches show ASTRAL quartet support (quartet-based frequencies of alternative quadripartition topologies around a given internode). Arrows indicate the numbers of the corresponding tree nodes in Fig. 1, and the corresponding hypotheses in the FcLM analyses. b Results of FcLM analyses for a selection of phylogenetic hypotheses applied at the amino-acid sequence level (supermatrix E). The first column shows the results of FcLM when the original data of supermatrix E were analyzed. The second column shows the results of FcLM after phylogenetic signal had been eliminated from supermatrix E (i.e. permutation no. I, see Additional file 2)

Fig. 3

Distribution of the median posterior node ages among the different meta-partitions. Arrows indicate the corresponding crown groups of Neuropterida and outgroups. Numbers on x-axis correspond to the node number ids of the tree in Fig. 1. The distribution of the median posterior age estimates of the individual meta-partitions from the first independent dating analysis (α = 2, run 1) is shown in blue. The distribution of the median age estimates when running the analyses without data (i.e. marginal prior) is shown in red

Fig. 4

Summarized results of stochastic character mapping analyses (SCM) for the evolution of larval ecologies based on 10,000 sampled character histories. Stochastic character maps were generated under the ER model and by using the topology and branch lengths of the chronogram of Fig. 1. Colored circles at the tips show the coded state for each species. Pie charts on internal tree nodes show posterior probabilities of states at each node under the model used. Internal nodes with a posterior probability lower than 1.00 are depicted in larger size (note: for the SCM analyses we assumed that larval ecologies remain constant within the same family)