Multiple historical processes obscure phylogenetic relationships in a taxonomically difficult group (Lobariaceae, Ascomycota)

Article metrics

Abstract

In the age of next-generation sequencing, the number of loci available for phylogenetic analyses has increased by orders of magnitude. But despite this dramatic increase in the amount of data, some phylogenomic studies have revealed rampant gene-tree discordance that can be caused by many historical processes, such as rapid diversification, gene duplication, or reticulate evolution. We used a target enrichment approach to sample 400 single-copy nuclear genes and estimate the phylogenetic relationships of 13 genera in the lichen-forming family Lobariaceae to address the effect of data type (nucleotides and amino acids) and phylogenetic reconstruction method (concatenation and species tree approaches). Furthermore, we examined datasets for evidence of historical processes, such as rapid diversification and reticulate evolution. We found incongruence associated with sequence data types (nucleotide vs. amino acid sequences) and with different methods of phylogenetic reconstruction (species tree vs. concatenation). The resulting phylogenetic trees provided evidence for rapid and reticulate evolution based on extremely short branches in the backbone of the phylogenies. The observed rapid and reticulate diversifications may explain conflicts among gene trees and the challenges to resolving evolutionary relationships. Based on divergence times, the diversification at the backbone occurred near the Cretaceous-Paleogene (K-Pg) boundary (65 Mya) which is consistent with other rapid diversifications in the tree of life. Although some phylogenetic relationships within the Lobariaceae family remain with low support, even with our powerful phylogenomic dataset of up to 376 genes, our use of target-capturing data allowed for the novel exploration of the mechanisms underlying phylogenetic and systematic incongruence.

Introduction

With the advent of next-generation sequencing (NGS) technology, the evolutionary relationships of many groups on the tree of life are increasingly resolved and our understanding of the diversification of these groups has been significantly improved1,2,3. However, in many groups, despite the use of NGS data, certain nodes have resisted unambiguous resolution. Conflicting topologies have been inferred from independent NGS data throughout the tree of life. For example, the placement of ctenophores and sponges have proven difficult as some studies place either sponges or ctenophores as sister to all other animals4,5. Phylogenomic reconstructions of birds also yielded conflicting relationships for the earliest divergence within Neoaves6, perhaps due to inferences from unequal data and taxon sampling: 42 Mbp from 48 bird genomes7 versus, 0.4 Mbp from 259 loci sampled from 198 species8. In the plant kingdom, inferences from NGS datasets resolve Amborella either sister to all other angiosperms9,10 or sister to water lilies11,12. Similarly, the Gnetales may be sister to pines, all conifers, or all seed plants13.

Several reasons have been invoked to explain gene-tree discordance14. Gene duplication can cause problems in phylogenetic reconstruction if paralogous loci with different histories are not distinguished within taxa and erroneously analyzed as homologs between taxa15,16. Rapid diversifications may lead to the fixation of fewer substitutions and hence to difficulties in resolving phylogenetic relationships. When too many speciation events occur in a relatively short period of time, gene genealogies are not expected to be fully sorted among evolutionary lineages leading to incomplete lineage sorting (ILS). Species tree methods can be used to mitigate the effects of ILS, but computational constraints prohibit fully parameterized methods such as *BEAST17 to estimate a species tree directly from hundreds of loci. Instead, species trees for large datasets are estimated from reconstructed gene trees, which can underestimate the support of relationships since these methods use summary statistics or pseudolikelihood18. Moreover, reticulate evolution, in the form of hybridization or horizontal gene transfer (HGT), can lead to incongruence among gene trees and obscure phylogenetic relationships19. Reticulations are not modeled in the commonly used species tree and concatenation approaches and hence specific, computationally intensive programs are needed to examine whether relationships of organisms are more complex than bifurcation20.

Reconstructing phylogenies can be done with nucleotide and amino acid data and these sources of data can yield incongruent phylogenies21. Nucleotide data can suffer from substitutional saturation at particular sites in a genome22. Such homoplasy is difficult to model in tree reconstruction methods and hence result in phylogenetic signatures being erased23. In case of ancient divergences, amino acid sequences may be preferable, since they are less prone to saturation21.

Fungi of Lobariaceae (recently also treated as a subfamily within Peltigeraceae24) develop conspicuous foliose macrolichens. Nearly 400 species are currently accepted25, but the diversity is predicted to reach 800 species26. Inferences from variation in three loci failed to resolved either of the three traditional genera (i.e., Lobaria, Pseudocyphellaria and Sticta) as monophyletic, and these were consequently broken up in several genera (e.g. Crocodia, Parmostictina, Podostictina and Yarrumia)27. Despite the discovery of highly-supported, genus-level clades, the relationships among these new genera remain partially unresolved. None of the phenotypic traits, essential to traditional generic concepts in Lobariaceae, i.e., the presence and type of pores in the lower cortex, defined a clade wherein all descendants exhibit the particular traits: pseudocyphellae and cyphellae no longer define a monophyletic Pseudocyphellaria and Sticta respectively. These pores have been shown to facilitate gas diffusion into the thallus28,29,30 and may provide an adaptive advantage in temperate environments31. Cyphellae likely arose independently in Dendriscosticta, which is consistently resolved within the Lobaria s.lat clade composed of genera lacking pores27. Phenotypic characters of the lichen association have been repeatedly shown to be poor phylogenetic predictors for the mycobiont32,33,34, and the newly described genera, Crocodia, Parmostictina, Podostictina, and Yarrumia, may represent morphological/ecological chimeric forms between Pseudocyphellaria and Sticta.

No study has yet critically estimated divergence times in Lobariaceae, but one study35 included three specimens from Sticta, Lobaria and Pseudocyphellaria in a fossil calibrated tree of eukaryotes. These samples composed a monophyletic sister group to Peltigera diverging roughly 150 million years ago (Mya) and having a stem age of around 70 Mya. However, another study estimated more recent divergences, with the split from Peltigera around 90 Mya and a stem age of nearly 50 Mya36. Furthermore, two additional studies have estimated divergence times in Lobaria and Sticta, both reporting a stem age of nearly 30 Mya for each of the genera37,38.

We have reassessed phylogenetic relationships in Lobariaceae using 400 target captured nuclear protein coding loci, and a variety of tree inference methods to reconstruct relationships among genera. We sought to assess how historical processes may confound phylogenetic reconstructions and obscure relationships of major lineages within the family. Specifically, we addressed the effect of (1) data type (nucleotides and amino acids), (2) phylogenetic reconstruction method (concatenation and species tree approaches), and (3) missing data. Furthermore, we examined our dataset for evidence of historical processes, such as (1) rapid diversification, and (2) reticulate evolution. Finally, we produced a fossil-calibrated phylogeny to estimate the timing of historical events during the diversification of Lobariaceae.

Results

Efficiency of sequencing data recovery and assembly of datasets

Following the read assembly via Hybpiper39 we recovered 337.67 of the 400 target genes with 75% coverage. Most coverage (all 400 genes mapped and 398 having 75% coverage) was achieved for Lobaria pulmonaria which was used as the reference in HybPiper and for the bait design. The lowest coverage was in Sticta cinereoglauca and Pseudocyphellaria crocata for which only 386 and 387 targets were recovered and only with partial coverage to the extent that at 75% coverage only 32 and 0 genes were recovered (Supplementary Table S1). This is probably due to the DNA extract quality reducing the hybridization of these sample to the baits. Of the 400 loci sequenced in this study, 138 generated a paralog warnings (an average of 14 paralogs per sample) involving between one and 95 samples (average 11 taxa with multiple sequences at a given locus). The trees produced after running the HybPiper script paralog_retriever.py to obtain all paralogous sequences for all flagged loci exhibited two patterns: (1) the two sequences retrieved would cluster and form a clade or (2) the two sequences would be resolved in two very different clades, separated by a long branch, which is indicative of an ancestral gene duplication event but could also be explained by allopolyploid hybridization (data not shown). In all 138 cases, HybPiper selected only one sequence that had the most sequencing depth and higher percent identity to the target sequence.

Topological patterns among data types and phylogeny reconstruction methods

Initially we generated a nucleotide dataset (376 × 96 nuc; Fig. 1A and Table 1) where all included loci had at least 50% specimen representation (at least 48 sequences regardless of recovered length in HybPiper), which was the case for 376 of the 400 loci. In both RAxML and ASTRAL analyses, Lobariaceae formed a well-supported (100% bootstrap support) clade but some of the backbone nodes were poorly supported (Fig. 1A). In the concatenated RAxML analysis, the first split gave rise to samples of Podostictina, forming a sister-group relationship to the remaining Lobariaceae. The next split reflects the divergence of Lobaria and its closely related genera (Anomolobaria, Dendriscosticta, Lobariella, Lobarina, Ricasolia, and Yoshimuriella). The following split gives rise to the clade composed of two sister genera, Sticta and Yarrumia. The most derived split among genera yields Crocodia sister to Pseudocyphellaria (RAxML tree; Fig. 1A).

Figure 1
figure1

Tree topologies estimated with nucleotide (A) and amino acid (B) data and different tree reconstruction methods (RAxML concatenated vs. ASTRAL species tree). Nodal support, either as bootstrap (RAxML) or local posterior probability (ASTRAL) is depicted under the branches. Abbreviations are as follows: N = Nephroma (outgroup); L = Lobaria s.lat. clade (also including Anomolobaria, Dendriscosticta, Lobariella, Lobarina, Ricasolia, and Yoshimuriella); Po = Podostictina; S = Sticta; Y = Yarrumia; C = Crocodia; P = Pseudocyphellaria.

Table 1 Summary of the datasets used in this study.

Subsequently, we analyzed amino acid sequences from the same dataset (376 × 96 aa; Figs 1B and 2). Again, the calculated tree supported Lobariaceae as a well-supported monophyletic clade, but the RAxML tree had even lower backbone support (37%) for the unique ancestry to the Podostictina clade, and the clade including Sticta, Yarrumia, Crocodia and Pseudocyphellaria. (Figs 1B and 2).

Figure 2
figure2

Congruence of topologies inferred with concatenation (A: RAxML) and species tree approaches (B: ASTRAL) from dataset composed of 376 amino acid sequences. The colors of branches are assigned to clades as follows: Orange: Nephroma (outgroup); Red: Anomolobaria, Dendriscosticta, Lobaria, Lobariella, Lobarina, Ricasolia, Yoshimuriella; Yellow: Crocodia, Parmostictina, Podostictina, and Yarrumia; Green: Sticta; Purple: Pseudocyphellaria.

The species tree inferred in ASTRAL with nucleotide sequences conflicted with the same dataset reconstructed from concatenated data in RAxML (376 × 96 nuc; Fig. 1A), while the amino acid species tree produced in ASTRAL had the same backbone topology as the concatenated amino acid tree (376 × 96 aa; Figs 1B and 2B). However, in the amino acid tree reconstruction from ASTRAL, two nodes were poorly supported; one associated with the splitting of Podostictina (0.71 local posterior probability) and another associated with the branching of Sticta and Yarrumia, both still forming a monophyletic clade (0.41 local posterior probability). The nucleotide species tree differed from the amino acid species tree, with Yarrumia being sister to Crocodia and Pseudocyphellaria with poor support (0.41 local posterior probability) and with Podostictina being sister to Sticta albeit also with poor support (0.44 local posterior probability; Fig. 1A). Crocodia was paraphyletic because Parmostictina obvoluta (DNA# 15843) clustered in this clade. Parmostictina is polyphyletic, with samples clustering in clades containing samples of Crocodia and Podostictina (Fig. 2).

Evidence for rapid diversification and gene-tree discordance

The presence of short backbone branches leading to Crocodia, Parmostictina, Podostictina, Pseudocyphellaria, Sticta, and Yarrumia is suggestive of a rapid diversification (Fig. 2). To investigate the cause for inconsistent phylogenetic tree results based on potential rapid diversification and gene-tree discordance, we created a pruned dataset (Fig. 3) (297 × 17). This created a complete matrix without missing data which was analyzed with RAxML and ASTRAL based on nuclear data and their amino acid translations. All four trees differed in their backbone branching patterns (Fig. 3A), with incongruent clades receiving strong support. Removing Podostictina and Yarrumia from the datasets, especially Podostictina, increased congruence and node support among all trimmed datasets (Supplementary Fig. S1). To find the most common topology we used SumTrees, which recovered 297 unique topologies of each tree reconstruction from the pruned nucleotide and amino acid datasets. We visualized this gene-tree discordance by using DensiTree, which creates a figure by overlaying all 297 gene trees. Although both datasets recovered clear groupings representing well supported clades, the backbone is depicted as a diffuse cloud of branches (Fig. 3B). Using the pairwise Robinson-Foulds distances (topological distances) of gene-trees and multidimensional scaling (MDS), we found that clustering of topologies was not significant and that both the amino acid and nucleotide datasets exhibited a diffuse pattern in all dimensions (Fig. 3C). This suggests that the gene-trees in both amino acid and nucleotide datasets have no clear pattern of a particular topology.

Figure 3
figure3

Poor support, incongruence of relationships of genera and gene tree discordance in Lobariaceae with (A) different methods (RAxML vs. ASTRAL) and different datatypes (amino acid vs. nucleotide) of the reduced dataset of 297 loci and 17 tips with no missing data. Nodal support is depicted by the size of the gray circles in the nodes, with larger circles being higher supported. (B) DensiTree plots of the gene trees used to produce the 297 × 17 datasets. (C) Dimension one of multidimensional scaling (MDS) plots of one-to-one Robinson Foulds distances of 297 trees produced from amino acid and nucleotide sequences.

We also used the program PhyParts to visualize the amount of gene trees that were in support or conflict with each node in the 297 × 17 aa and 297 × 17 nuc datasets (Fig. 4). ASTRAL species trees generated from the gene trees served as reference trees. The proportion of gene trees that supported and conflicted with reference bipartitions were mapped to the reference trees. The PhyParts output for both datasets showed that the deep backbone nodes leading to the genera Crocodia, Podostictina, Pseudocyphellaria, Sticta, and Yarrumia were supported by a small plurality of gene trees (Fig. 4, pink portions of the pie charts). The node reprsenting the common ancestor of Crocodia, Podostictina, Pseudocyphellaria, Sticta, and Yarrumia was only supported by 13% and 20% of the gene trees in the amino acid (Fig. 4A) and nucleotide (Fig. 4B) datasets respectively. However, some nodes within this clade were congruent among a vast majority of amino acid and nucleotide gene trees, such as the stem node leading to the four Sticta species and the node leading to P. chloroleuca and P. glabra. The deeper reference tree node leading to Dendriscosticta, Lobaria, Lobariella, Lobarina, Ricasolia, and Yoshimuriella was supported by 44% and 63% of the amino acid and nucleotide gene trees respectively. The more recent nodes representing the relationships of Lobariella, Ricasolia, and Yoshmuriella also conflicted with amino acid and nucleotide gene trees, but the remaining nodes in this clade were generally more supported.

Figure 4
figure4

ASTRAL species trees. (A) Estimated with the 297 amino acid trees. (B) Estimated with 297 nucleotide trees. Pie charts show the gene tree conflict evaluation at each node with light blue proportions representing concordant topologies, pink portions representing conflicting topologies, and yellow proportions representing one dominant alternative topology. Numbers above the branches report the number of gene trees supporting that node, while the number below reports the number that are in conflict.

Evidence for reticulate evolution in Lobariaceae

We used the same pruned 297 × 17 nuc dataset to infer evidence of reticulate evolution in Lobariaceae by using PhyloNet. Although, the maximum pseudo-likelihood (MPL) option (InferNetwork_MPL) in PhyloNet indicated that a three-reticulation model had the best fit to our data, none of the networks produced by PhyloNet had three reticulations (Table 2). At most, two reticulations were inferred (Fig. 5). The reticulations reconstructed in the networks generally originated in the ancestral nodes or branches of samples from Crocodia, Podostictina, Pseudocyphellaria, Sticta and Yarrumia. Using a full likelihood approach (InferNetwork_ML) on 17 taxa is too computationally heavy for most servers hence we produced a dataset that had only the five taxa that were most involved in putatively reticulate patterns in the 17-tip analyses (Crocodia, Podostictina, Pseudocyphellaria, Sticta, and Yarrumia). As with the MPL approach, the three-reticulation scenario had the best fit to the data in the ML analysis and the most likely network had three reticulations (Table 3 and Supplementary Figs S2 and S3).

Table 2 Total log probabilities and AIC scores for different reticulation scenarios using the MPL approach.
Figure 5
figure5

PhyloNet network showing one of the most likely reticulation scenarios.

Table 3 Likelihood and AIC scores for different reticulation scenarios using the ML approach.

To further investigate the evidence for reticulations, we used the MedianNetwork analysis implemented in the program SplitsTree. Instead of gene trees for input, which is what is used in Phylonet, we used the concatenated 297 × 17 nuc alignment. We removed the outgroup Nephroma antarcticum for the network analysis. The SplitsTree MedianNetwork analysis recovered the well-supported and differentiated ML and ASTRAL lineages such as the Sticta, Pseudocyphellaria and the clade with Dendriscosticta, Lobariella, Ricasolia, and Yoshmuriella, but also showed numerous box-like relationships indicating conflicting phylogenetic signals (Fig. 6).

Figure 6
figure6

A SplitsTree network estimated with the MedianNetwork method. The network shows relationships of genera based on the 297 × 17 nuc dataset with no outgroup.

Timing of divergence in Lobariaceae

Both runs of MCMCTree converged on similar estimations (Table 4). The average stem age of Lobariaceae was estimated to be 64.4 Mya (91.9–43.4 Mya). Crown divergences of the major clades in Lobariaceae are as follows: Lobaria s.lat clade (which includes Anomolobaria, Dendriscosticta, Lobariella, Lobarina, Ricasolia, and Yoshimuriella), 57.6 Mya (84.0–38.1 Mya); Podostictina, 29.5 Mya (53.0–14.2 Mya); Yarrumia, 12.4 Mya (21.2-5.9 Mya); Sticta, 25.2 Mya (39.9–15.2 Mya); Crocodia, 33.4 Mya (56.0–16.7 Mya); and Pseudocyphellaria, 54.1 Mya (78.7–35.2 Mya). The estimates of the first run are depicted in Fig. 7.

Table 4 Node ages estimated in MCMCTree with error ranges in millions of years.
Figure 7
figure7

A chronogram estimated by Bayesian relaxed molecular clock implemented in MCMCTree. The ML tree inferred with concatenated amino acid sequences (376 × 96 aa - congruent with the ASTRAL species tree) was used a scaffold for the dating analysis. The grey vertical bar depicts the K-Pg boundary. The timescale is in units of millions of years. Colors of clades: Orange: Nephroma (outgroup); Red: Anomolobaria, Dendriscosticta, Lobaria, Lobariella, Lobarina, Ricasolia, Yoshimuriella; Yellow: Crocodia, Parmostictina, Podostictina, and Yarrumia (not monophyletic); Green: Sticta; Purple: Pseudocyphellaria.

Discussion

We produced the first target enrichment dataset of a lineage of lichenized fungi with a goal of resolving the higher-level relationships in Lobariaceae, and found that the evolutionary history exhibited a rapid and reticulate diversification and that we could not confidently reconstruct the relationships assuming a strictly bifurcating tree despite the wide sampling of nuclear loci. The phylogenetic positions of certain lineages differ when using (1) different data types (nucleotides versus amino acids) or (2) different tree reconstruction methods (Fig. 1). However, more slowly-evolving amino acid sequence data yielded more consistent results, especially with the 376-locus dataset, where the concatenated RAxML and ASTRAL species tree backbone branching patterns were identical (Fig. 2). This topology was more supported in ASTRAL because the node leading to the Podostictina clade in the RAxML tree was poorly supported (Fig. 1B). An increasing number of studies are producing datasets of hundreds, even thousands of loci, and are finding that certain nodes in the tree of life are still very challenging to resolve with our current inference methods4,5,6,7,8,11,12,13. We found no exception in Lobariaceae. However, even though phylogenetic relationships among genera remained uncertain (c.f., Moncada, Lücking, and Betancourt-Macuase 2013), we were able to further study the factors that are preventing resolution. As a result, our study is the first on lichenized fungi to investigate with target capture, the underlying mechanisms that may lead to poorly-resolved and conflicting relationships.

Regardless of the dataset or analysis, backbone branches marking shared ancestry of Pseudocyphellaria (including Crocodia, Podostictina, and Yarrumia) and Sticta were very short, suggesting a rapid diversification. Short branches can be difficult to resolve with confidence, even with large subgenomic datasets40. Rapid diversifications can also lead to conflicting topologies among individual gene trees15. This is consistent with our study, even in our reduced dataset (297 × 17) with 297 unique rooted topologies (Figs 3 and 4). The multi-species coalescent-based species tree approach can account for ILS41. However, species trees with short, deep branches, such as those recovered in this study, can result in uninformative gene-trees, and the gene trees may result in erroneous species tree reconstruction42. The inconsistency in phylogenetic reconstruction and low support for certain branching patterns revealed in our study suggest that the short branches in our trees confound species tree approaches in this case. The only time we found a consensus between the two methods was when we used the largest amino acid data set (376 × 96 aa), but node support was low for some of these relationships (Figs 1B and 2).

Concatenation of genetic loci has a long history in phylogenetics, but it has also been known to produce highly supported yet conflicting results, especially in the area of tree space known as the “anomaly zone”43 (when most gene trees in the genome do not reflect the true speciation history). Furthermore, concatenated analyses cannot account for the different evolutionary histories at different loci, which can also produce misleading topologies44. They can also be extremely sensitive to outlier loci45. The probability of sampling an outlier locus increases as the amount of data increases, and it is therefore expected that the effect of such loci will be stronger in larger subgenomic data set45.

Models that incorporated reticulations always had better fit to the data than models assuming a strictly bifurcating tree (Fig. 5, Tables 2 and 3). Strictly bifurcating trees are optimized and reduced visualizations of a more complicated biological reality and when using hundreds of genes, a network will generally be a better explanation of the data46. In most of the phylogenetic networks generated, the taxa with labile topological positions (Podostictina, Sticta and Yarrumia) were usually associated with the reticulations in the network. The Phylonet reticulation estimations were supported by the PhyParts analysis (Fig. 4), which showed that most of the gene trees were in conflict with the ASTRAL species trees at the deeper nodes connecting to Crocodia, Podostictina, Pseudocyphellaria, Sticta and Yarrumia, the genera associated with reticulations. Furthermore, the SplitsTree network (Fig. 6) showed that most of the nodes associated with the clade containing genera with pored lower cortices contain relatively more reticulations than the Lobaria s.lat. clade and that the phylogenetic placement of Podostictina and Yarrumia is unclear, as they originate from the base of the network. Reticulation events can confound both concatenated ML and coalescent-based species tree approaches, causing incongruence and poorly supported relationships14,20. This could also contribute to the inconsistency among phylogenetic reconstructions and poor nodal supports found in our study. Additionally, the effects of reticulation and rapid diversification may not be independent. Reticulations have been hypothesized to promote subsequent rapid diversification events in cichlids47, yeasts48, and fungal pathogens49. Reticulate evolution is most likely to induce rapid diversification when parental lineages occupy highly dissimilar ecological niches but are only moderately genetically differentiated50. Recombination among distinct yet compatible evolutionary lineages may generate novel phenotypes to exploit new niches. Although we have shown that reticulate models have a better fit, it is unclear whether they led to a rapid diversification (suggested by the short and poorly supported nodes) in Lobariaceae as the two are not mutually exclusive. Rapid radiations could also increase the chances of reticulate evolution, because so many closely related taxa may not have had the chance to develop effective reproductive barriers before secondary contact occurred in nature. Reticulate evolution may characterize the evolution of a number of lichenized fungi. Hybridization has been invoked to account for patterns in secondary chemistry in Hypotrachyna51 and evidence for introgressive hybridization was proposed in the Peltigera didactyla complex52. Also gene-tree incongruence in Cladonia53 and Lobaria54 were seen as compatible with reticulation. In Letharia a reticulate evolutionary history was proposed based on patterns of recombination in nuclear DNA markers55. The other form of reticulate evolution, horizontal gene transfer (HGT), has been discovered with fungal genes being found in the Trebouxia photobionts56, and with polyketide synthases of actinobacterial origin being found in lichenized fungi57, but HGT has not yet been reported among lichenized fungi and is not hypothesized in this study.

Our topology (376 × 96 aa data sets in Figs 1B and 2) was similar to that of Moncada et al.27 with the main exceptions being the relationships of Sticta and genera previously segregated from Pseudocyphellaria (e.g. Crocodia, Parmostictina, Podostictina, and Yarrumia). In the previous study27, Parmostictina, Podostictina and Crocodia formed a well-supported clade (Yarrumia was not included in their analysis) sister to Pseudocyphellaria. In our analyses, the Crocodia and Pseudocyphellaria still formed a sister-group relationship in all datasets, but the relationships of Parmostictina and Podostictina were poorly supported and differed depending on data type and method of inference: it could be placed sister to either all Lobariaceae, to (Crocodia + Pseudocyphellaria + Sticta + Yarrumia), to Yarrumia, or only to Sticta (Figs 1, 2, and 3). The erratic placement of Parmostictina and Podostictina, when using different data types and inference methods, exhibits patterns found in other unrelated groups of organisms with unresolved phylogenetic relationships, such as the deep, poorly supported nodes leading to ctenophores and sponges4,5, Amborella9,10,11,12, and Gnetales13,58. Another relationship that was often but not always recovered was the Sticta and Yarrumia sister relationship.

Our time-calibrated phylogeny (Fig. 7) recovered a nearly 70 Mya stem age of Lobariaceae which is in agreement with the estimate of Gaya et al.35 but not with Simon et al.36 where a 50 Mya stem age was reported. Furthermore, our crown age estimates agree with other dated phylogenies generated for the genera Sticta38 and Lobaria37, both of which have crown ages around 30 Mya. The timing of the rapid and reticulate Lobariaceae divergence is near the end of the Cretaceous around 70–60 Mya. The Cretaceous Terrestrial Revolution59,60 (CTR) occurred between 100 and 70 Mya and is associated with the massive diversification events that gave rise to the angiosperms61,62. The increase in angiosperm diversity in the late Cretaceous created new ecological opportunities63,64 and these may have triggered bursts of diversification in spore-dispersing plants63,64,65,66 mammals67, and ants68. Secondary diversification pulses following the CTR and rise of angiosperms are especially linked to the evolution of epiphytism and has been demonstrated in leafy liverworts65 and ferns63,64. Most Lobariaceae species are epiphytic and the rapid diversification event seen in our study could be another example of a subsequent pulse in divergence linked to the CTR. The rapid diversification of Lobariaceae was also associated with the Cretaceous–Paleogene (K-Pg) boundary around 66 Mya. Mass extinctions associated with K-Pg boundary may have reduced competition and provided an opportunity for surviving lineages to radiate69. The macrolichen growth form is associated with multiple diversification rate increases following the K-Pg boundary (Huang et al. 2019 unpublished data70). Lobariaceae, a lineage composed solely of macrolichens, may have also experienced a similar diversification rate increase as in other clades of lichenized fungi.

Incongruence in phylogeny reconstruction can be caused by stochastic and systematic errors. Stochastic error is exacerbated by gene length. The shorter the sequence, the more stochastic error can cause misleading results. With hundreds of genes, phylogenomic approaches, such as target-capturing applied in this study alleviate the issue of stochastic error, but systematic errors may be amplified when phylogenetic reconstruction methods do not account for the properties in the data14,71. Overestimation of divergence times has been demonstrated when molecular substitution rates vary among clades and when only a subset of taxa are used in the phylogeny reconstruction72. While it is unclear whether our dataset is characterized by clade-specific rate heterogeneity, our taxon sampling is not complete. This incomplete sampling could lead to an over estimation of divergence times in our dataset and shift the occurrence of the rapid diversification event after the K-Pg boundary. Furthermore the presence of short branches in our phylogenetic reconstructions and a high level of gene tree discordance could be caused by ILS and reticulate evolution. It is possible that these historical processes also influenced divergence time estimations as well. If ILS is present at certain loci, divergence times will be overestimated with these genes because coalescence of the loci precedes divergence of species73. Hybridization could have multiple effects on divergence time estimates depending on whether the hybrid locus coalesced before or after species divergence. Hybrid loci that coalesce before species divergence would lead to overestimated divergence times similar to ILS. However, if hybrid loci coalesce after lineage splitting, divergence times will be underestimated73. Because we did not identify which loci were influenced by ILS and hybridization, it is important to state that the divergence times estimated in this study should be interpreted with caution.

Since the application of genetic data to reconstructing the evolutionary relationships of lichenized fungi, our understanding of the diversification of most lineages has significantly improved. Single and multiple locus phylogenies have discovered many relationships that were obscured by homoplasy in phenotypic characters. With the advent of next-generation sequencing (NGS) technology, only a handful of studies have been conducted on groups of lichenized fungi to resolve phylogenetic relationships74,75. In traditional phylogenetic studies, fungal-specific primers got around the issue of mixed genomes in DNA isolations, but with genomic studies, the only way to get purely fungal DNA is to culture the fungal symbiont. The ability of lichenized fungi to be cultured varies among groups and it has not been successful in many. Regardless, over ten genomes have been sequenced and are available for study. Recently, studies using purely fungal reference genomes from cultured lichenized fungi have gotten around the issue of using metagenomic DNA isolations, by applying a mapping step that pulls out the fungal reads for use in phylogenomic analyses75.

This study is the first to apply the target capturing approach in lichenized fungi to investigate historical processes influencing the evolution of major clades in Lobariaceae. Although we were not able to confidently resolve certain relationships, we were able to investigate why certain nodes remained unresolved, and we have a more thorough understanding of the evolutionary history of the group. We provide evidence for rapid diversification and reticulate evolution in our data set, which would not be possible to infer with only a few loci. However, it is still not clear what the relative effects of rapid diversification and reticulations are on gene tree discordance and difficulties in species tree reconstruction. We encourage the phylogenetic study of taxonomic groups with uncertain phylogenetic relationships so, as a scientific community, we can understand how historical processes are most likely to obscure relationships in the tree of life. Resolution of some nodes in the tree of life may not always be possible with increasing datasets, but a detailed understanding of the evolutionary events that confound the history of a group are the next best steps to take to make informed decisions about their taxonomic classification.

Methods

Taxon sampling

We sampled representatives of all the current and tentative genera in Lobariaceae. Our dataset includes representatives of the three major classic genera Lobaria, Pseudocyphellaria, Sticta, along with representative samples of the later segregated genera (Anomolobaria, Crocodia, Dendriscosticta, Lobariella, Lobarina, Parmostictina, Podostictina, Ricasolia, Yarrumia, and Yoshimuriella (Supplementary Table S2).

Bait design

We designed baits for target capture using Markerminer76, a bioinformatics pipeline that finds single copy loci in the genome using genome and transcriptome data as inputs. Markerminer is designed for use with angiosperms, and has databases for 15 angiosperm genomes, but other genome databases can be added for customized use of the program. We used the Lobaria pulmonaria genome and the gene annotation file (Lobpul1_AssemblyScaffolds_Repeatmasked.fasta and Lobpul1_GeneCatalog_20170213.gff3) from JGI (https://jgi.doe.gov/) to develop a custom database. The result is a simplified reference genome that only contains gene regions and the introns are hard-masked as “N”s in the sequence77.

We assembled transcriptome data (published in Meiser et al.77) from Evernia prunastri, Pseudevernia furfuracea and Lasallia pustulata using Trinity78. The resulting transcriptomes and the custom L. pulmonaria database were used to identify clusters of single-copy gene transcripts present in the transcriptome assemblies. Next, these were aligned and filtered against the L. pulmonaria reference proteome from JGI (Lobpul1_GeneCatalog_proteins_20170213.aa.fasta) using BLAST. These single-copy genes were re-aligned to the intron hard-masked genome and intron-exon boundaries were identified.

Markerminer identified and aligned 1,714 single-copy genes to the hard-masked L. pulmonaria reference genome. We selected loci that had at least one exon of at least 500 bp and indicated clear intron-exon boundaries on the hard-masked alignment. The designed baits were collected as 800 separate fasta DNA sequence files from the L. pulmonaria and E. prunastri (400 each). These sequences were provided to Arbor Biosciences (Ann Arbor, MI, USA) for MYbaits bait design. These filtered baits covered 92% of the desired target positions with at least one bait, therefore all 800 target sequences are represented with at least one bait. One hundred nucleotide-long baits were designed with two times tiling density resulting in 18,139 raw unfiltered baits. Following Arbor Biosciences recommended filtering process (baits passing “Moderate” BLAST filtering), 17,941 baits were retained for target-capturing.

Library preparation

DNA was obtained using the ZR Fungal/Bacterial DNA MiniPrep™ (Zymo Research, Irvine, CA, USA) or by a CTAB extraction protocol. The concentration of all DNA isolates was quantified with the Qubit (Thermo Fisher Scientific, Waltham, MA, USA). Two hundred ng of meta-genomic DNA was normalized to a final volume of 52.5 μL in resuspension buffer. Fragments of DNA around 500 bp were generated with M220 Focused-ultrasonicator™ (Woburn, MA, USA) and then 50 μL was cleaned up with 80 μL SeraPure beads which are an inexpensive alternative to commercially purchased magnetic beads79. The Adapterama dual-indexing system79 was used to uniquely barcode all samples using the KAPA Hyper Prep Kit (KAPABiosystems, Wilmington, MA, USA). Twenty-five μL of the sheared, cleaned bead elution was used in a 30 μL end-repair and A-tailing reaction followed by a 55 μL ligation to attach the Adapterama stubby y-yolk adapter. The ligation products were subjected to bead-based size selection with SeraPure bead to enrich for fragments of around 550 bp which was eluted in 22 μL of RSB. Twenty μL of the elution was used in a limited-cycle (9–11 cycles) polymerase chain reaction (PCR) to attach the barcoded iTru5 and iTru7 Adapterama primers. Subsequently, the PCR products were cleaned with 1X SeraPure beads and eluted in 43 μL of nuclease-free, PCR-grade water. DNA concentrations of all samples were quantified with a Qubit fluorometer (Thermo Fisher Scientific, Waltham, MA, USA) and a subset were checked for proper size distribution of fragments on the Bioanalyzer (Agilent, Santa Clara, CA, USA).

Samples were pooled by phylogenetic relatedness, corresponding to major clades (corresponding to the colors in Figs 1, 2, and 3) of the previous Lobariaceae three-locus phylogeny27, for hybridization with RNA baits. For each sample, 100–200 ng of DNA was mixed with all samples of each of the five pools and then concentrated in a heated vacuum centrifuge. Pools were hybridized with reagents provided with the baits from Arbor Biosciences for ~20 h at 65 °C. After incubation, the baits were attached to Dynabeads® M-280 Streptavidin beads (Carlsbad, CA) and then washed according to the Arbor Biosciences protocol followed by a post-wash enrichment PCR cycle with KAPA Hifi Hotstart ReadyMix. Each pool went through 11 cycles of PCR except for the outgroup (Nephroma samples), which was subjected to 14 cycles. These were cleaned with 1X SeraPure beads and the DNA concentration was quantified on the Qubit and size of distribution of the DNA fragments in the pools were observed on the Bioanalyzer. These pools were mixed together to have 3 ng of DNA per sample in the final 96-sample pool which was used for sequencing at the Field Museum’s Pritzker Laboratory with a single 300-cycle v2 MiSeq reagent kit (Illumina, San Diego, CA, USA).

Data processing

The MiSeq run (300 cycles using V2 chemistry) produced 15,835,491 150-bp paired-end reads, which were demultiplexed and adapter-trimmed by Illumina BaseSpace. The raw reads were downloaded to the Field Museum server and quality trimmed with Trimmomatic80 using a quality cut off of 15 in a 4-bp sliding window, discarding any reads under 35 bp. Only paired, trimmed reads were used in downstream analyses, which is an average of 158,443 reads per sample remaining after trimming with a range from 7929 to 317,874 reads (Supplementary Table S1). These sequences were used for the read files using the program HybPiper39 which assembles gene regions and extracts exon sequences for each sample. We generated a target file from the L. pulmonaria transcriptome on JGI (Lobpul1_GeneCatalog_proteins_20170213.aa.fasta) that has the complete amino acid sequences of each of the 400 target genes that were used for bait design. The sorted, trimmed reads of each sample were mapped to the targets using the default BLASTX81 option, which we found recovered much more data (a more complete dataset), especially for the outgroup (Nephroma taxa) than the BWA option that uses a nucleotide target file. Upon completion of the HybPiper assembly, the 399 gene sequences (one gene did not produce any sequences) were extracted using retrieve_sequences.py and then batch aligned using MAFFT82.

HybPiper will flag a gene if it identifies multiple sequences spanning at least 85% of the gene length39. When this occurs at a specific gene, HybPiper will choose only one of the sequences, first by selecting the sequence with the highest depth of sequencing and then, if all sequences are similar in coverage, it uses the closest match to the specified target sequence file. For the paralog warning genes, all of the copies were extracted (paralog_retriever.py), aligned and phylogenetically analyzed to check the patterns of paralogy for each of these flagged genes. However, paralogous sequences could be present in some datasets that did not have a warning. For example, if one or more taxa yield only one sequence that is paralogous to the others. Furthermore, multiple copies could be alleles or a duplication that postdates a bifurcation which HybPiper would flag, but this case would not cause problems with phylogenetic inference. Future studies will need to use more thorough analyses to have a detailed understanding of how paralogs are influencing phylogenetic reconstructions.

Concatenation of datasets

Multiple nucleotide alignments of single genes were concatenated with FASconCAT-G83. One alignment contained 376 loci and 96 tips (376 × 96 nuc) and had 438,036 nucleotide positions with 13.75% gaps and undetermined characters. Only genes that had at least 48 sequences (50% sample representation) were used. Another reduced dataset (297 × 17 nuc) with only 17 taxa, representing all Lobariaceae genera and the well supported subclades in Sticta and Pseudocyphellaria was generated in HybPiper. With this dataset 297 of the genes had all 17 sequences and these alignments were used to generated gene trees and a concatenated alignment with 338,019 nucleotide positions and 4.04% gaps and undetermined characters. We also generated the same concatenated datasets with amino acid sequences using FASconCAT-G. The dataset with 376 loci and 96 tips (376 × 96 aa) and had 145,467 amino acid positions with 6.34% missing data and 7.26% indels. Finally, the reduced dataset (297 × 17 aa) of 297 loci and 17 tips had 112,265 amino acid positions with no missing data, but it did contain 3.89% indels. A summary of all datasets can be found in Table 1.

Maximum likelihood and species tree analyses

For all full and reduced datasets, concatenated phylogenies and all single gene trees were estimated using the RAxML84 rapid hill climbing algorithm and performing 100 bootstrap replicates. All single gene alignments and all partitions of each concatenated dataset had substitution model selection using either the GTRGAMMA or PROTGAMMA -AUTO models for nucleotide and amino acid models of sequences respectively. Since concatenated datasets can be prone to false positive topologies43,85, we also analyzed our datasets with the multi-species coalescent model the program Accurate Species Tree Algorithm (ASTRAL 5.6.1)86. For the ASTRAL analyses, we inferred species trees from single, fully resolved, and unrooted gene trees mentioned above. Branch support was reported as local posterior probabilities (i.e., the support for a quadripartition).

Analysis of gene-tree discordance and tree space

Regardless of the data type or tree building approach, all of the datasets produced backbones with short internal branches (Fig. 3). During rapid diversification, speciation and gene flow occur simultaneously in a relatively short period of time, which can lead to high levels of gene-tree discordance. Discordance among gene-trees can be the result of incomplete lineage sorting (ILS), reticulate evolution, and homoplasy. The coalescent model implemented in ASTRAL, accounts for ILS, but this approach was still unable to completely resolve the deep divergences in the backbone of the amino acid and nucleotide datasets. Although this does not completely rule out ILS as a source of gene-tree discordance, it does suggest that other historical processes, such as homoplasy or reticulate evolution, which are not accounted for in ASTRAL, were potentially contributing to the poorly resolved backbones in our reconstructions. To conduct analyses on and visualize the gene-tree discordance the best trees from the RAxML analyses for each gene were converted to ultrametric trees using the chronos() function in the R package APE (Paradis, Claude, and Strimmer 2004) with a lambda = 1. This was done using the program SumTrees87 implemented in DendroPy 4.0.088 to find if there was a common tree topology among the reduced (297 × 17) amino acid and nucleotide datasets. DensiTree 2.2.589 was used to visualize the discordance of gene-tree topologies.

Tree space of the reduced datasets (297 × 17 nuc and 297 × 17 aa) were described using multidimensional scaling (MDS) in R using the customized topclustMDS function described elsewhere. These datasets were used because complete trees (all having 17 tips) were required. Gene trees were loaded into R using the APE package90, then MDS scaling and functions in the cluster package91 were used to identify the most appropriate number of clusters and the identity of loci comprising each cluster.

We used the program PhyParts92 (https://bitbucket.org/blackrim/phyparts) to assess the level of concordance and conflict among gene trees in the reduced datasets (297 × 17 aa and 297 × 17 nuc). The ASTRAL species trees generated from each dataset were used as reference trees. Individual amino acid and nucleotide gene trees were rooted with the root.multiphylo() function in R package APE with N. antarcticum as outgroup. PhyParts conducts a bipartition analysis across all rooted gene trees and maps the amount of trees that support and conflict with each bipartition in the reference tree. The output of PhyParts was visualized by plotting pie charts that depict the amount of gene trees that supported and conflicted with each bipartition on the ASTRAL species tree with the script phypartspiecharts.py (https://github.com/mossmatters/MJPythonNotebooks).

Tests for reticulate evolution

Another known source of poor backbone node support in phylogenies is reticulate evolution. This process is not modeled in the most used phylogenomic inference programs such as RAxML or ASTRAL as the output can only be bifurcating. A current program, PhyloNet93, does model this process, and we used it to investigate whether parts of the relationships in the Lobariaceae phylogeny are better supported with a model of reticulate evolution rather than a strictly bifurcating tree. PhyloNet identifies reticulation events using a multi-species network coalescent model that accounts for ILS and reticulate evolution. We first used the maximum pseudo-likelihood approach (InferNetwork_MPL option) on the 297 × 17 nuc dataset, specifying the -po option which optimizes the branch lengths and inheritance probabilities under full likelihood for the returned species networks. We conducted six analyses with reticulation scenarios ranging from zero to five and then ten with each analysis conducting ten runs and inferring five networks. The zero-reticulation scenario served as a null model and we then increased the reticulation events step-wise to five reticulations to see how each increase fit to the data by calculating AIC for each network. Finally, we conducted a ten-reticulation analysis to see if more than five reticulations could be recovered from our dataset.

Although the MPL approach is computationally fast, under certain conditions, the true reticulation history may not be identifiable94. Thus, using the drop.tip() function in the R package APE90, we produce a five-tip dataset that contain only Crocodia, Podostictina, Pseudocyphellaria, Sticta, and Yarrumia (the genera that were usually involved in the reticulations of the MPL analyses), so that we could apply a fully parameterized ML approach (InferNetwork_ML). Due to the taxing computational demands of the fully parameterized ML option of PhyloNet, we only tested zero to three reticulations (using the -po option described above), because the best fitting scenarios in the MPL analyses included three reticulations (see results). Each analysis conducted five runs and produced one network and the total log probabilities were used to calculate AIC to assess model fit.

To visualize uncertainty and phylogenetic conflict within our reduced dataset, we used SplitsTree v4.14.895. The input was the concatenated reduced nucleotide dataset (297 × 17 nuc) and excluded the outgroup N. antarcticum since evidence for reticulate evolution was only detected among the ingroup taxa in Phylonet. The data were analyzed using the MedianNetwork method which uses all sites in the nucleotide alignment that contain exactly two different states while excluding gaps or missing states to generate an unreduced median network.

Dated phylogeny of Lobariaceae

The program MCMCTree 1.296,97 was used to infer divergence times in Lobariaceae. The ML tree (376 × 96 aa), inferred with amino acids and congruent with the one inferred in ASTRAL (Fig. 2A) was used for the dating analyses and the concatenated amino acid matrix. We used CODEML to generate the Hessian matrix using an empirical rate matrix with gamma rates among sites (WAG + Gamma). Using this matrix, we ran MCMCTree with appropriate rates under the approximate method. Settings were set as follows: independent-rates model, time unit at 100 Ma, sampling fraction at ρ = 0.1, and the birth and death rates at λ = µ = 1. We ran the analyses with gaps and ambiguities removed (cleandata = 0). The gamma prior for the substitution model was set at six transitions to two transversions rate ratio. The gamma shape parameter for variable rates among sites was set as α = 1 The Dirichlet-gamma prior was set to α = 2 and β = 20 for a diffuse prior. The MCMC was run for 40,000 iterations, with samples taken every other iteration, after a 5% burnin. Two runs were conducted to check for consistency.

We calibrated the phylogeny with the only known fossil from the family, obtained from a 12–24 Myr-old Miocene deposit from northern California. The fossil is an impression that resembles the genus Lobaria. A recent study by Cornejo and Scheidegger37, used a molecular clock-calibrated phylogeny of Lobaria, to estimate the age of the fossil impression and found that it fits into the time frame of Lobaria diversification and placed it near the crown of the genus. Based on that, we calibrated the 376 × 96 aa concatenated phylogeny with the impression fossil, by placing it on the branch between L. linita and the remaining Lobaria species, with a range of 12 to 24 million years.

Accession codes

Genbank BioSample submission: SUB4924261.

References

  1. 1.

    Rokas, A., Williams, B. L., King, N. & Carroll, S. B. Genome-scale approaches to resolving incongruence in molecular phylogenies. Nature 425, 798 (2003).

  2. 2.

    Gee, H. Evolution: ending incongruence. Nature 425, 782 (2003).

  3. 3.

    Hipp, A. L. et al. A framework phylogeny of the American oak clade based on sequenced RAD data. PLoS One 9, e93975 (2014).

  4. 4.

    Dunn, C. W. et al. Broad phylogenomic sampling improves resolution of the animal tree of life. Nature 452, 745 (2008).

  5. 5.

    Simion, P. et al. A large and consistent phylogenomic dataset supports sponges as the sister group to all other animals. Curr. Biol. 27, 958–967 (2017).

  6. 6.

    Reddy, S. et al. Why do phylogenomic data sets yield conflicting trees? Data type influences the avian tree of life more than taxon sampling. Syst. Biol. 66, 857–879 (2017).

  7. 7.

    Jarvis, E. D. et al. Whole-genome analyses resolve early branches in the tree of life of modern birds. Science (80-.). 346, 1320–1331 (2014).

  8. 8.

    Prum, R. O. et al. A comprehensive phylogeny of birds (Aves) using targeted next-generation DNA sequencing. Nature 526, 569 (2015).

  9. 9.

    Drew, B. T. et al. Another look at the root of the angiosperms reveals a familiar tale. Syst. Biol. 63, 368–382 (2014).

  10. 10.

    Zhang, N., Zeng, L., Shan, H. & Ma, H. Highly conserved low‐copy nuclear genes as effective markers for phylogenetic analyses in angiosperms. New Phytol. 195, 923–937 (2012).

  11. 11.

    Goremykin, V. V., Nikiforova, S. V., Cavalieri, D., Pindo, M. & Lockhart, P. The root of flowering plants and total evidence. Syst. Biol. 64, 879–891 (2015).

  12. 12.

    Xi, Z., Liu, L., Rest, J. S. & Davis, C. C. Coalescent versus concatenation methods and the placement of Amborella as sister to water lilies. Syst. Biol. 63, 919–932 (2014).

  13. 13.

    Burleigh, J. G. & Mathews, S. Phylogenetic signal in nucleotide data from seed plants: implications for resolving the seed plant tree of life. Am. J. Bot. 91, 1599–1613 (2004).

  14. 14.

    Som, A. Causes, consequences and solutions of phylogenetic incongruence. Brief. Bioinform. 16, 536–548 (2015).

  15. 15.

    Maddison, W. P. Gene trees in species trees. Syst. Biol. 46, 523–536 (1997).

  16. 16.

    Martin, A. P. & Burg, T. M. Perils of paralogy: using HSP70 genes for inferring organismal phylogenies. Syst. Biol. 51, 570–587 (2002).

  17. 17.

    Bouckaert, R. et al. BEAST 2: a software platform for Bayesian evolutionary analysis. PLoS Comput Biol 10, e1003537 (2014).

  18. 18.

    Liu, L., Xi, Z., Wu, S., Davis, C. C. & Edwards, S. V. Estimating phylogenetic trees from genome‐scale data. Ann. N. Y. Acad. Sci. 1360, 36–53 (2015).

  19. 19.

    Xu, S. Phylogenetic analysis under reticulate evolution. Mol. Biol. Evol. 17, 897–907 (2000).

  20. 20.

    Nakhleh, L., Warnow, T., Linder, C. R. & John, K. S. Reconstructing reticulate evolution in species—theory and practice. J. Comput. Biol. 12, 796–811 (2005).

  21. 21.

    Liu, Y., Cox, C. J., Wang, W. & Goffinet, B. Mitochondrial phylogenomics of early land plants: mitigating the effects of saturation, compositional heterogeneity, and codon-usage bias. Syst. Biol. 63, 862–878 (2014).

  22. 22.

    Delsuc, F., Brinkmann, H. & Philippe, H. Phylogenomics and the reconstruction of the tree of life. Nat. Rev. Genet. 6, 361 (2005).

  23. 23.

    Philippe, H. et al. Resolving difficult phylogenetic questions: why more sequences are not enough. PLOS Biol. 9, e1000602 (2011).

  24. 24.

    Kraichak, E., Huang, J.-P., Nelsen, M. P., Leavitt, S. D. & Lumbsch, H. T. A revised classification of orders and families in the two major subclasses of Lecanoromycetes (Ascomycota) based on a temporal approach. Bot. J. Linn. Soc. in press (2018).

  25. 25.

    Kirk, P. M., Cannon, P. F., Minter, D. W. & Stalpers, J. A. Dictionary of the Fungi. (CAB International, 2008).

  26. 26.

    Moncada, B. & Lücking, R. Ten new species of Sticta and counting: Colombia as a hot spot for unrecognized diversification in a conspicuous macrolichen genus. Phytotaxa 74, 1–29 (2012).

  27. 27.

    Moncada, B., Lücking, R. & Betancourt-Macuase, L. Phylogeny of the Lobariaceae (lichenized Ascomycota: Peltigerales), with a reappraisal of the genus Lobariella. Lichenol. 45, 203–263 (2013).

  28. 28.

    Nash, T. H. Lichen Biology. (Cambridge University Press, 1996).

  29. 29.

    Lange, O. L., Green, T. G. A. & Ziegler, H. Water status related photosynthesis and carbon isotope discrimination in species of the lichen genus Pseudocyphellaria with green or blue-green photobionts and in photosymbiodemes. Oecologia 75, 494–501 (1988).

  30. 30.

    Green, T. G. A. & Lange, O. L. Ecophysiological adaptations of the lichen genera Pseudocyphellaria and Sticta to south temperate rainforests. Lichenol. 23, 267–282 (2007).

  31. 31.

    Hale, M. E. Pseudocyphellae and pored epicortex in the Parmeliaceae: their delimitation and evolutionary significance. Lichenol. 13, 1–10 (1981).

  32. 32.

    Lumbsch, H. T. & Leavitt, S. D. Goodbye morphology? A paradigm shift in the delimitation of species in lichenized fungi. Fungal Divers. 50, 59–72 (2011).

  33. 33.

    Printzen, C. Lichen systematics: the role of morphological and molecular data to reconstruct phylogenetic relationships BT - Progress In Botany 71. in (eds Lüttge, U., Beyschlag, W., Büdel, B. & Francis, D.) 233–275, https://doi.org/10.1007/978-3-642-02167-1_10 (Springer Berlin Heidelberg, 2010).

  34. 34.

    Jaklitsch, W., Baral, H.-O., Lücking, R., Lumbsch, H. T. & Frey, W. Syllabus of Plant Families-A. Engler’s Syllabus der Pflanzenfamilien Part 1/2. (Schweizerbart and Borntraeger Science Publishers 2016).

  35. 35.

    Gaya, E. et al. The adaptive radiation of lichen-forming Teloschistaceae is associated with sunscreening pigments and a bark-to-rock substrate shift. Proc. Natl. Acad. Sci. 112, 11600–11605 (2015).

  36. 36.

    Simon, A., Magain, N., Goffinet, B. & Sérusiaux, E. High diversity, high insular endemism and recent origin in the lichen genus Sticta (lichenized Ascomycota, Peltigerales) in Madagascar and the Mascarenes. Mol. Phylogenet. Evol. 122, 15–28 (2018).

  37. 37.

    Cornejo, C. & Scheidegger, C. Estimating the timescale of Lobaria diversification. Lichenol. 50, 113–121 (2018).

  38. 38.

    Widhelm, T. J. et al. Oligocene origin and drivers of diversification in the genus Sticta (Lobariaceae, Ascomycota). Mol. Phylogenet. Evol. 126, 58–73 (2018).

  39. 39.

    Johnson, M. G. et al. HybPiper: Extracting coding sequence and introns for phylogenetics from high-throughput sequencing reads using target enrichment. Appl. Plant Sci. 4, 1600016 (2016).

  40. 40.

    Wiens, J. J. et al. Branch lengths, support, and congruence: testing the phylogenomic approach with 20 nuclear loci in snakes. Syst. Biol. 57, 420–431 (2008).

  41. 41.

    Liu, L. & Edwards, S. V. Phylogenetic analysis in the anomaly zone. Syst. Biol. 58, 452–460 (2009).

  42. 42.

    Huang, H., He, Q., Kubatko, L. S. & Knowles, L. L. Sources of error inherent in species-tree estimation: impact of mutational and coalescent effects on accuracy and implications for choosing among different methods. Syst. Biol. 59, 573–583 (2010).

  43. 43.

    Kubatko, L. S. & Degnan, J. H. Inconsistency of phylogenetic estimates from concatenated data under coalescence. Syst. Biol. 56, 17–24 (2007).

  44. 44.

    Degnan, J. H. & Rosenberg, N. A. Discordance of species trees with their most likely gene trees. PLOS Genet. 2, e68 (2006).

  45. 45.

    Lacey, K. L., Huateng, H., Jeet, S. & Smith, S. A. A matter of phylogenetic scale: Distinguishing incomplete lineage sorting from lateral gene transfer as the cause of gene tree discord in recent versus deep diversification histories. Am. J. Bot. 105, 376–384 (2018).

  46. 46.

    Bapteste, E. et al. Networks: expanding evolutionary thinking. Trends Genet. 29, 439–441 (2013).

  47. 47.

    Meier, J. I. et al. Ancient hybridization fuels rapid cichlid fish adaptive radiations. Nat. Commun. 8, 14363 (2017).

  48. 48.

    Leducq, J.-B. et al. Speciation driven by hybridization and chromosomal plasticity in a wild yeast. Nat. Microbiol. 1, 15003 (2016).

  49. 49.

    Stukenbrock, E. H. The role of hybridization in the evolution and emergence of new fungal plant pathogens. Phytopathology 106, 104–112 (2016).

  50. 50.

    Kagawa, K. & Takimoto, G. Hybridization can promote adaptive radiation by means of transgressive segregation. Ecol. Lett. 21, 264–274 (2018).

  51. 51.

    Culberson, C. F. & Hale, M. E. Chemical and morphological evolution in Parmelia sect. Hypotrachyna: Product of ancient hybridization? Brittonia 25, 162–173 (1973).

  52. 52.

    Goffinet, B. & Hastings, R. I. Two new sorediate taxa of Peltigera. Lichenol. 27, 43–58 (1995).

  53. 53.

    Jana, S., Stenroos, S., Grube, M. & Škaloud, P. Genetic diversity and species delimitation of the zeorin-containing red-fruited Cladonia species (lichenized Ascomycota) assessed with ITS rDNA and β-tubulin data. Lichenol. 45, 665–684 (2013).

  54. 54.

    Cornejo, C., Chabanenko, S. & Scheidegger, C. Are species-pairs diverging lineages? A nine-locus analysis uncovers speciation among species-pairs of the Lobaria meridionalis-group (Ascomycota). Mol. Phylogenet. Evol (2018).

  55. 55.

    Kroken, S. & Taylor, J. W. Outcrossing and recombination in the lichenized fungus. Letharia. Fungal Genet. Biol. 34, 83–92 (2001).

  56. 56.

    Beck, A., Divakar, P. K., Zhang, N., Molina, M. C. & Struwe, L. Evidence of ancient horizontal gene transfer between fungi and the terrestrial alga Trebouxia. Org. Divers. Evol. 15, 235–248 (2015).

  57. 57.

    Schmitt, I. & Lumbsch, H. T. Ancient horizontal gene transfer from bacteria enhances biosynthetic capabilities of fungi. PLoS One 4, e4437 (2009).

  58. 58.

    Mathews, S. Phylogenetic relationships among seed plants: persistent questions and the limits of molecular data. Am. J. Bot. 96, 228–236 (2009).

  59. 59.

    Vermeij, G. J. The energetics of modernization: the last one hundred million years of biotic evolution. Paleontol. Res. 15, 54–61 (2011).

  60. 60.

    Benton, M. J. The origins of modern biodiversity on land. Philos. Trans. R. Soc. B Biol. Sci. 365, 3667–3679 (2010).

  61. 61.

    Lidgard, S. & Crane, P. R. Quantitative analyses of the early angiosperm radiation. Nature 331, 344 (1988).

  62. 62.

    Lidgard, S. & Crane, P. R. Angiosperm diversification and Cretaceous floristic trends: a comparison of palynofloras and leaf macrofloras. Paleobiology 16, 77–93 (1990).

  63. 63.

    Schuettpelz, E. & Pryer, K. M. Evidence for a Cenozoic radiation of ferns in an angiosperm-dominated canopy. Proc. Natl. Acad. Sci. 106, 11200–11205 (2009).

  64. 64.

    Schneider, H. et al. Ferns diversified in the shadow of angiosperms. Nature 428, 553 (2004).

  65. 65.

    Feldberg, K. et al. Epiphytic leafy liverworts diversified in angiosperm-dominated forests. Sci. Rep. 4, 5974 (2014).

  66. 66.

    Laenen, B. et al. Extant diversity of bryophytes emerged from successive post-Mesozoic diversification bursts. Nat. Commun. 5, 5134 (2014).

  67. 67.

    Meredith, R. W. et al. Impacts of the Cretaceous Terrestrial Revolution and KPg extinction on mammal diversification. Science. 334, 521–524 (2011).

  68. 68.

    Moreau, C. S., Bell, C. D., Vila, R., Archibald, S. B. & Pierce, N. E. Phylogeny of the ants: Diversification in the age of angiosperms. Science. 312, 101–104 (2006).

  69. 69.

    Renne, P. R. et al. Time scales of critical events around the Cretaceous-Paleogene boundary. Science. 339, 684–687 (2013).

  70. 70.

    Huang, J.-P., Kraichak, E., Leavitt, S. D., Nelsen, M. P. & Lumbsch, H. T. Accelerated diversification in three diverse clades of morphologically complex lichen-forming fungi after the K-Pg boundary. Sci. Rep. under review (2019).

  71. 71.

    Jeffroy, O., Brinkmann, H., Delsuc, F. & Philippe, H. Phylogenomics: the beginning of incongruence? TRENDS Genet. 22, 225–231 (2006).

  72. 72.

    Beaulieu, J. M., O’meara, B. C., Crane, P. & Donoghue, M. J. Heterogeneous rates of molecular evolution and diversification could explain the Triassic age estimate for angiosperms. Syst. Biol. 64, 869–878 (2015).

  73. 73.

    Joly, S., McLenachan, P. A. & Lockhart, P. J. A statistical approach for distinguishing hybridization and incomplete lineage sorting. Am. Nat. 174, E54–E70 (2009).

  74. 74.

    Leavitt, S. D. et al. Resolving evolutionary relationships in lichen-forming fungi using diverse phylogenomic datasets and analytical approaches. Sci. Rep. 6, 22262 (2016).

  75. 75.

    Grewe, F., Huang, J.-P., Leavitt, S. D. & Lumbsch, H. T. Reference-based RADseq resolves robust relationships among closely related species of lichen-forming fungi using metagenomic DNA. Sci. Rep. 7, 9884 (2017).

  76. 76.

    Chamala, S. et al. MarkerMiner 1.0: A new application for phylogenetic marker development using angiosperm transcriptomes. Appl. Plant Sci. 3, 1400115 (2015).

  77. 77.

    Meiser, A., Otte, J., Schmitt, I. & Grande, F. D. Sequencing genomes from mixed DNA samples - evaluating the metagenome skimming approach in lichenized fungi. Sci. Rep. 7, 14881 (2017).

  78. 78.

    Grabherr, M. G. et al. Trinity: reconstructing a full-length transcriptome without a genome from RNA-Seq data. Nat. Biotechnol. 29, (644 (2011).

  79. 79.

    Glenn, T. C. et al. Adapterama I: universal stubs and primers for thousands of dual-indexed Illumina libraries (iTru & iNext). BioRxiv 49114 (2016).

  80. 80.

    Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114–2120 (2014).

  81. 81.

    Camacho, C. et al. BLAST+: architecture and applications. BMC Bioinformatics 10, 421 (2009).

  82. 82.

    Katoh, K. & Standley, D. M. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol. Biol. Evol. 30, 772–780 (2013).

  83. 83.

    Kück, P. & Longo, G. C. FASconCAT-G: extensive functions for multiple sequence alignment preparations concerning phylogenetic studies. Front. Zool. 11, 81 (2014).

  84. 84.

    Stamatakis, A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30, 1312–1313 (2014).

  85. 85.

    Heled, J. & Drummond, A. J. Bayesian inference of species trees from multilocus data. Mol. Biol. Evol. 27, 570–580 (2010).

  86. 86.

    Mirarab, S. & Warnow, T. ASTRAL-II: coalescent-based species tree estimation with many hundreds of taxa and thousands of genes. Bioinformatics 31, i44–i52 (2015).

  87. 87.

    Sukumaran, J. & Holder, M. T. SumTrees: Phylogenetic tree summarization. 4.0. 0. Progr. Doc. available from authors, https//github.com/jeetsukumaran/Dendrophy (2015).

  88. 88.

    Sukumaran, J. & Holder, M. T. DendroPy: a Python library for phylogenetic computing. Bioinformatics 26, 1569–1571 (2010).

  89. 89.

    Bouckaert, R. R. DensiTree: making sense of sets of phylogenetic trees. Bioinformatics 26, 1372–1373 (2010).

  90. 90.

    Paradis, E., Claude, J. & Strimmer, K. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20, (289–290 (2004).

  91. 91.

    Maechler, M. Finding groups in data”: Cluster analysis extended Rousseeuw et al. Doc. Softw. Packag. Compr. R Arch. Netw. Wien (2018).

  92. 92.

    Smith, S. A., Moore, M. J., Brown, J. W. & Yang, Y. Analysis of phylogenomic datasets reveals conflict, concordance, and gene duplications with examples from animals and plants. BMC Evol. Biol. 15, 150 (2015).

  93. 93.

    Than, C., Ruths, D. & Nakhleh, L. PhyloNet: a software package for analyzing and reconstructing reticulate evolutionary relationships. BMC Bioinformatics 9, 322 (2008).

  94. 94.

    Yu, Y. & Nakhleh, L. A maximum pseudo-likelihood approach for phylogenetic networks. BMC Genomics 16, S10 (2015).

  95. 95.

    Huson, D. H. SplitsTree: analyzing and visualizing evolutionary data. Bioinformatics 14, 68–73 (1998).

  96. 96.

    Yang, Z. PAML 4: phylogenetic analysis by maximum likelihood. Mol. Biol. Evol. 24, 1586–1591 (2007).

  97. 97.

    Yang, Z. & Rannala, B. Bayesian estimation of species divergence times under a molecular clock using multiple fossil calibrations with soft bounds. Mol. Biol. Evol. 23, 212–226 (2005).

Download references

Acknowledgements

This project is part of the first author’s dissertation project at the University of Illinois at Chicago and the Grainger Bioinformatics Center at the Field Museum and was financially supported by NSF grant DEB-1354884 to Thorsten Lumbsch and DEB-1354631 to Bernard Goffinet. The study would not have been as complete without the samples provided by Luis Coca, Antoine Simon, Bruce McCune, Taylor Quedensley, and Tor Tønsberg and to Imke Schmitt, who provided the transcriptome data for bait design. We appreciated the guidance of Matt Johnson who advised in the bait design and data assembly. We would like to acknowledge our colleagues in New Zealand: Peter deLange, Dan Blanchon, Allison Knight, and Jeremy Rolfe for their expertise and guidance while collecting specimens there. Similarly, we also are grateful to Juan Larraín and Matt von Konrat for their assistance while collecting specimens in Chile. We like to thank Kevin Feldheim and Isabel Distefano for assistance with NGS sequencing and Huini Wu, Erin Conely, and Yejun Kim for help with specimen processing and DNA extractions. We appreciate the comments from the anonymous reviews.

Author information

T.W. wrote the manuscript, prepared and sequenced the target capture samples, and conducted phylogenetic analyses as part of his dissertation research. F.G., J.H., R.L., B.G., R.M. and H.T.L. provided assistance with research design and writing the paper. F.G. provided bioinformatic guidance for bait design and dataset assembly. B.G. provided funding for baits. J.M. and B.M. provided specimens for the study. All authors reviewed the manuscript.

Correspondence to Todd J. Widhelm.

Ethics declarations

Competing Interests

The authors declare no competing interests.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.