Molecular clock fork phylogenies: closed form analytic maximum likelihood solutions

Maximum likelihood (ML) is increasingly used as an optimality criterion for selecting evolutionary trees, but finding the global optimum is a hard computational task. Because no general analytic solution is known, numeric techniques such as hill climbing or expectation maximization (EM) are used in order to find optimal parameters for a given tree. So far, analytic solutions were derived only for the simplest model-three-taxa, two-state characters, under a molecular clock. Quoting Ziheng Yang, who initiated the analytic approach,"this seems to be the simplest case, but has many of the conceptual and statistical complexities involved in phylogenetic estimation."In this work, we give general analytic solutions for a family of trees with four-taxa, two-state characters, under a molecular clock. The change from three to four taxa incurs a major increase in the complexity of the underlying algebraic system, and requires novel techniques and approaches. We start by presenting the general maximum likelihood problem on phylogenetic trees as a constrained optimization problem, and the resulting system of polynomial equations. In full generality, it is infeasible to solve this system, therefore specialized tools for the molecular clock case are developed. Four-taxa rooted trees have two topologies-the fork (two subtrees with two leaves each) and the comb (one subtree with three leaves, the other with a single leaf). We combine the ultrametric properties of molecular clock fork trees with the Hadamard conjugation to derive a number of topology dependent identities. Employing these identities, we substantially simplify the system of polynomial equations for the fork. We finally employ symbolic algebra software to obtain closed formanalytic solutions (expressed parametrically in the input data). In general, four-taxa trees can have multiple ML points. In contrast, we can now prove that each fork topology has a unique(local and global) ML point.     

Analytic solutions of maximum likelihood on forks of four taxa

This work deals with symbolic mathematical solutions to maximum likelihood on small phylogenetic trees. Maximum likelihood (ML) is increasingly used as an optimality criterion for selecting evolutionary trees, but finding the global optimum is a hard computational task. In this work, we give general analytic solutions for a family of trees with four taxa, two state characters, under a molecular clock. Previously, analytical solutions were known only for three taxa trees. The change from three to four taxa incurs a major increase in the complexity of the underlying algebraic system, and requires novel techniques and approaches. Despite the simplicity of our model, solving ML analytically in it is close to the limit of today's tractability. Four taxa rooted trees have two topologies--the fork (two subtrees with two leaves each) and the comb (one subtree with three leaves, the other with a single leaf). Combining the properties of molecular clock fork trees with the Hadamard conjugation, and employing the symbolic algebra software Maple, we derive a number of topology dependent identities. Using these identities, we substantially simplify the system of polynomial equations for the fork. We finally employ the symbolic algebra software to obtain closed form analytic solutions (expressed parametrically in the input data).     

Maximum likelihood molecular clock comb: analytic solutions.

Maximum likelihood (ML) is increasingly used as an optimality criterion for selecting evolutionary trees, but finding the global optimum is a hard computational task. Because no general analytic solution is known, numeric techniques such as hill climbing or expectation maximization (EM), are used in order to find optimal parameters for a given tree. So far, analytic solutions were derived only for the simplest model--three taxa, two state characters, under a molecular clock. Four taxa rooted trees have two topologies--the fork (two subtrees with two leaves each) and the comb (one subtree with three leaves, the other with a single leaf). In a previous work, we devised a closed form analytic solution for the ML molecular clock fork. In this work, we extend the state of the art in the area of analytic solutions ML trees to the family of all four taxa trees under the molecular clock assumption. The change from the fork topology to the comb incurs a major increase in the complexity of the underlying algebraic system and requires novel techniques and approaches. We combine the ultrametric properties of molecular clock trees with the Hadamard conjugation to derive a number of topology dependent identities. Employing these identities, we substantially simplify the system of polynomial equations. We finally use tools from algebraic geometry (e.g., Gr?bner bases, ideal saturation, resultants) and employ symbolic algebra software to obtain analytic solutions for the comb. We show that in contrast to the fork, the comb has no closed form solutions (expressed by radicals in the input data). In general, four taxa trees can have multiple ML points. In contrast, we can now prove that under the molecular clock assumption, the comb has a unique (local and global) ML point. (Such uniqueness was previously shown for the fork.).     

Interordinal relationships of birds and other reptiles based on whole mitochondrial genomes

Several different groups of birds have been proposed as being the oldest or earliest diverging extant lineage within the avian phylogenetic tree, particularly ratites (Struthioniformes), waterfowl (Anseriformes), and shorebirds (Charadriiformes). Difficulty in resolving this issue stems from several factors, including the relatively rapid radiation of primary (ordinal) bird lineages and the lack of characters from an extant outgroup for birds that is closely related to them by measure of time. To help resolve this question, we have sequenced entire mitochondrial genomes for five birds (a rhea, a duck, a falcon, and two perching birds), one crocodilian, and one turtle. Maximum parsimony and maximum likelihood analyses of these new sequences together with published sequences (18 taxa total) yield the same optimal tree topology, in which a perching bird (Passeriformes) is sister to all the other bird taxa. A basal position for waterfowl among the bird study taxa is rejected by maximum likelihood analyses. However, neither the conventional view, in which ratites (including rhea) are basal to other birds, nor tree topologies with falcon or chicken basal among birds could be rejected in the same manner. In likelihood analyses of a subset of seven birds, alligator, and turtle (9 taxa total), we find that increasing the number of parameters in the model shifts the optimal topology from one with a perching bird basal among birds to the conventional view with ratites diverging basally; moreover, likelihood scores for the two trees are not significantly different. Thus, although our largest set of taxa and characters supports a tree with perching birds diverging basally among birds, the position of this earliest divergence among birds appears unstable. Our analyses indicate a sister relationship between a waterfowl/chicken clade and ratites, relative to perching birds and falcon. We find support for a sister relationship between turtles and a bird/crocodilian clade, and for rejecting both the Haemothermia hypothesis (birds and mammals as sister taxa) and the placement of turtles as basal within the phylogenetic tree for amniote animals.     

Reconstructing the evolutionary history of the Lorisidae using morphological, molecular, and geological data

Major aspects of lorisid phylogeny and systematics remain unresolved, despite several studies (involving morphology, histology, karyology, immunology, and DNA sequencing) aimed at elucidating them. Our study is the first to investigate the evolution of this enigmatic group using molecular and morphological data for all four well-established genera: Arctocebus, Loris, Nycticebus, and Perodicticus. Data sets consisting of 386 bp of 12S rRNA, 535 bp of 16S rRNA, and 36 craniodental characters were analyzed separately and in combination, using maximum parsimony and maximum likelihood. Outgroups, consisting of two galagid taxa (Otolemur and Galagoides) and a lemuroid (Microcebus), were also varied. The morphological data set yielded a paraphyletic lorisid clade with the robust Nycticebus and Perodicticus grouped as sister taxa, and the galagids allied with Arctocebus. All molecular analyses maximum parsimony (MP) or maximum likelihood (ML) which included Microcebus as an outgroup rendered a paraphyletic lorisid clade, with one exception: the 12S + 16S data set analyzed with ML. The position of the galagids in these paraphyletic topologies was inconsistent, however, and bootstrap values were low. Exclusion of Microcebus generated a monophyletic Lorisidae with Asian and African subclades; bootstrap values for all three clades in the total evidence tree were over 90%. We estimated mean genetic distances for lemuroids vs. lorisoids, lorisids vs. galagids, and Asian vs. African lorisids as a guide to relative divergence times. We present information regarding a temporary land bridge that linked the two now widely separated regions inhabited by lorisids that may explain their distribution. Finally, we make taxonomic recommendations based on our results.     

Treelength optimization for phylogeny estimation

The standard approach to phylogeny estimation uses two phases, in which the first phase produces an alignment on a set of homologous sequences, and the second phase estimates a tree on the multiple sequence alignment. POY, a method which seeks a tree/alignment pair minimizing the total treelength, is the most widely used alternative to this two-phase approach. The topological accuracy of trees computed under treelength optimization is, however, controversial. In particular, one study showed that treelength optimization using simple gap penalties produced poor trees and alignments, and suggested the possibility that if POY were used with an affine gap penalty, it might be able to be competitive with the best two-phase methods. In this paper we report on a study addressing this possibility. We present a new heuristic for treelength, called BeeTLe (Better Treelength), that is guaranteed to produce trees at least as short as POY. We then use this heuristic to analyze a large number of simulated and biological datasets, and compare the resultant trees and alignments to those produced using POY and also maximum likelihood (ML) and maximum parsimony (MP) trees computed on a number of alignments. In general, we find that trees produced by BeeTLe are shorter and more topologically accurate than POY trees, but that neither POY nor BeeTLe produces trees as topologically accurate as ML trees produced on standard alignments. These findings, taken as a whole, suggest that treelength optimization is not as good an approach to phylogenetic tree estimation as maximum likelihood based upon good alignment methods.     

A tree island approach to inferring phylogeny in the ant subfamily Formicinae,with especial reference to the evolution of weaving

The ant subfamily Formicinae is a large assemblage (2458 species (J. Nat. Hist. 29 (1995) 1037), including species that weave leaf nests together with larval silk and in which the metapleural gland-the ancestrally defining ant character-has been secondarily lost. We used sequences from two mitochondrial genes (cytochrome b and cytochrome oxidase 2) from 18 formicine and 4 outgroup taxa to derive a robust phylogeny, employing a search for tree islands using 10000 randomly constructed trees as starting points and deriving a maximum likelihood consensus tree from the ML tree and those not significantly different from it. Non-parametric bootstrapping showed that the ML consensus tree fit the data significantly better than three scenarios based on morphology, with that of Bolton (Identification Guide to the Ant Genera of the World, Harvard University Press, Cambridge, MA) being the best among these alternative trees. Trait mapping showed that weaving had arisen at least four times and possibly been lost once. A maximum likelihood analysis showed that loss of the metapleural gland is significantly associated with the weaver life-pattern. The graph of the frequencies with which trees were discovered versus their likelihood indicates that trees with high likelihoods have much larger basins of attraction than those with lower likelihoods. While this result indicates that single searches are more likely to find high- than low-likelihood tree islands, it also indicates that searching only for the single best tree may lose important information.     

Stochastic search strategy for estimation of maximum likelihood phylogenetic trees

The maximum likelihood (ML) method of phylogenetic tree construction is not as widely used as other tree construction methods (e.g., parsimony, neighbor-joining) because of the prohibitive amount of time required to find the ML tree when the number of sequences under consideration is large. To overcome this difficulty, we propose a stochastic search strategy for estimation of the ML tree that is based on a simulated annealing algorithm. The algorithm works by moving through tree space by way of a "local rearrangement" strategy so that topologies that improve the likelihood are always accepted, whereas those that decrease the likelihood are accepted with a probability that is related to the proportionate decrease in likelihood. Besides greatly reducing the time required to estimate the ML tree, the stochastic search strategy is less likely to become trapped in local optima than are existing algorithms for ML tree estimation. We demonstrate the success of the modified simulated annealing algorithm by comparing it with two existing algorithms (Swofford's PAUP* and Felsenstein's DNAMLK) for several theoretical and real data examples.     

Shrinkage Effect in Ancestral Maximum Likelihood

Ancestral maximum likelihood (AML) is a method that simultaneously reconstructs a phylogenetic tree and ancestral sequences from extant data (sequences at the leaves). The tree and ancestral sequences maximize the probability of observing the given data under a Markov model of sequence evolution, in which branch lengths are also optimized but constrained to take the same value on any edge across all sequence sites. AML differs from the more usual form of maximum likelihood (ML) in phylogenetics because ML averages over all possible ancestral sequences. ML has long been know to be statistically consistent - that is, it converges on the correct tree with probability approaching 1 as the sequence length grows. However, the statistical consistency of AML has not been formally determined, despite informal remarks in a literature that dates back 20 years. In this short note we prove a general result that implies that AML is statistically inconsistent. In particular we show that AML can 'shrink' short edges in a tree, resulting in a tree that has no internal resolution as the sequence length grows. Our results apply to any number of taxa.     

Ancestral maximum likelihood of evolutionary trees is hard

Maximum likelihood (ML) (Neyman, 1971) is an increasingly popular optimality criterion for selecting evolutionary trees. Finding optimal ML trees appears to be a very hard computational task--in particular, algorithms and heuristics for ML take longer to run than algorithms and heuristics for maximum parsimony (MP). However, while MP has been known to be NP-complete for over 20 years, no such hardness result has been obtained so far for ML. In this work we make a first step in this direction by proving that ancestral maximum likelihood (AML) is NP-complete. The input to this problem is a set of aligned sequences of equal length and the goal is to find a tree and an assignment of ancestral sequences for all of that tree's internal vertices such that the likelihood of generating both the ancestral and contemporary sequences is maximized. Our NP-hardness proof follows that for MP given in (Day, Johnson and Sankoff, 1986) in that we use the same reduction from Vertex Cover; however, the proof of correctness for this reduction relative to AML is different and substantially more involved.     

Accurate Phylogenetic Tree Reconstruction from Quartets: A Heuristic Approach

Supertree methods construct trees on a set of taxa (species) combining many smaller trees on the overlapping subsets of the entire set of taxa. A ‘quartet’ is an unrooted tree over taxa, hence the quartet-based supertree methods combine many -taxon unrooted trees into a single and coherent tree over the complete set of taxa. Quartet-based phylogeny reconstruction methods have been receiving considerable attentions in the recent years. An accurate and efficient quartet-based method might be competitive with the current best phylogenetic tree reconstruction methods (such as maximum likelihood or Bayesian MCMC analyses), without being as computationally intensive. In this paper, we present a novel and highly accurate quartet-based phylogenetic tree reconstruction method. We performed an extensive experimental study to evaluate the accuracy and scalability of our approach on both simulated and biological datasets.     

Reconstructing phylogeny by quadratically approximated maximum likelihood

Maximum likelihood (ML) for phylogenetic inference from sequence data remains a method of choice, but has computational limitations. In particular, it cannot be applied for a global search through all potential trees when the number of taxa is large, and hence a heuristic restriction in the search space is required. In this paper, we derive a quadratic approximation, QAML, to the likelihood function whose maximum is easily determined for a given tree. The derivation depends on Hadamard conjugation, and hence is limited to the simple symmetric models of Kimura and of Jukes and Cantor. Preliminary testing has demonstrated the accuracy of QAML is close to that of ML.     

Improving the efficiency of SPR moves in phylogenetic tree search methods based on maximum likelihood

MOTIVATION: Maximum likelihood (ML) methods have become very popular for constructing phylogenetic trees from sequence data. However, despite noticeable recent progress, with large and difficult datasets (e.g. multiple genes with conflicting signals) current ML programs still require huge computing time and can become trapped in bad local optima of the likelihood function. When this occurs, the resulting trees may still show some of the defects (e.g. long branch attraction) of starting trees obtained using fast distance or parsimony programs. METHODS: Subtree pruning and regrafting (SPR) topological rearrangements are usually sufficient to intensively search the tree space. Here, we propose two new methods to make SPR moves more efficient. The first method uses a fast distance-based approach to detect the least promising candidate SPR moves, which are then simply discarded. The second method locally estimates the change in likelihood for any remaining potential SPRs, as opposed to globally evaluating the entire tree for each possible move. These two methods are implemented in a new algorithm with a sophisticated filtering strategy, which efficiently selects potential SPRs and concentrates most of the likelihood computation on the promising moves. RESULTS: Experiments with real datasets comprising 35-250 taxa show that, while indeed greatly reducing the amount of computation, our approach provides likelihood values at least as good as those of the best-known ML methods so far and is very robust to poor starting trees. Furthermore, combining our new SPR algorithm with local moves such as PHYML's nearest neighbor interchanges, the time needed to find good solutions can sometimes be reduced even more.     

A method for inferring the rate of evolution of homologous characters that can potentially improve phylogenetic inference, resolve deep divergence and correct systematic biases

Current phylogenetic methods attempt to account for evolutionary rate variation across characters in a matrix. This is generally achieved by the use of sophisticated evolutionary models, combined with dense sampling of large numbers of characters. However, systematic biases and superimposed substitutions make this task very difficult. Model adequacy can sometimes be achieved at the cost of adding large numbers of free parameters, with each parameter being optimized according to some criterion, resulting in increased computation times and large variances in the model estimates. In this study, we develop a simple approach that estimates the relative evolutionary rate of each homologous character. The method that we describe uses the similarity between characters as a proxy for evolutionary rate. In this article, we work on the premise that if the character-state distribution of a homologous character is similar to many other characters, then this character is likely to be relatively slowly evolving. If the character-state distribution of a homologous character is not similar to many or any of the rest of the characters in a data set, then it is likely to be the result of rapid evolution. We show that in some test cases, at least, the premise can hold and the inferences are robust. Importantly, the method does not use a "starting tree" to make the inference and therefore is tree independent. We demonstrate that this approach can work as well as a maximum likelihood (ML) approach, though the ML method needs to have a known phylogeny, or at least a very good estimate of that phylogeny. We then demonstrate some uses for this method of analysis, including the improvement in phylogeny reconstruction for both deep-level and recent relationships and overcoming systematic biases such as base composition bias. Furthermore, we compare this approach to two well-established methods for reweighting or removing characters. These other methods are tree-based and we show that they can be systematically biased. We feel this method can be useful for phylogeny reconstruction, understanding evolutionary rate variation, and for understanding selection variation on different characters.     

A structural EM algorithm for phylogenetic inference.

A central task in the study of molecular evolution is the reconstruction of a phylogenetic tree from sequences of current-day taxa. The most established approach to tree reconstruction is maximum likelihood (ML) analysis. Unfortunately, searching for the maximum likelihood phylogenetic tree is computationally prohibitive for large data sets. In this paper, we describe a new algorithm that uses Structural Expectation Maximization (EM) for learning maximum likelihood phylogenetic trees. This algorithm is similar to the standard EM method for edge-length estimation, except that during iterations of the Structural EM algorithm the topology is improved as well as the edge length. Our algorithm performs iterations of two steps. In the E-step, we use the current tree topology and edge lengths to compute expected sufficient statistics, which summarize the data. In the M-Step, we search for a topology that maximizes the likelihood with respect to these expected sufficient statistics. We show that searching for better topologies inside the M-step can be done efficiently, as opposed to standard methods for topology search. We prove that each iteration of this procedure increases the likelihood of the topology, and thus the procedure must converge. This convergence point, however, can be a suboptimal one. To escape from such "local optima," we further enhance our basic EM procedure by incorporating moves in the flavor of simulated annealing. We evaluate these new algorithms on both synthetic and real sequence data and show that for protein sequences even our basic algorithm finds more plausible trees than existing methods for searching maximum likelihood phylogenies. Furthermore, our algorithms are dramatically faster than such methods, enabling, for the first time, phylogenetic analysis of large protein data sets in the maximum likelihood framework.     

Multiple maxima of likelihood in phylogenetic trees: an analytic approach

Maximum likelihood (ML) is a widely used criterion for selecting optimal evolutionary trees. However, the nature of the likelihood surface for trees is still not sufficiently understood, especially with regard to the frequency of multiple optima. Here, we initiate an analytic study for identifying sequences that generate multiple optima. We concentrate on the problem of optimizing edge weights for a given tree or trees (as opposed to searching through the space of all trees). We report a new approach to computing ML directly, which we have used to find large families of sequences that have multiple optima, including sequences with a continuum of optimal points. Such data sets are best supported by different (two or more) phylogenies that vary significantly in their timings of evolutionary events. Some standard biological processes can lead to data with multiple optima, and consequently the field needs further investigation. Our results imply that hill-climbing techniques as currently implemented in various software packages cannot guarantee that one will find the global ML point, even if it is unique.     

Increasing the efficiency of searches for the maximum likelihood tree in a phylogenetic analysis of up to 150 nucleotide sequences

Even when the maximum likelihood (ML) tree is a better estimate of the true phylogenetic tree than those produced by other methods, the result of a poor ML search may be no better than that of a more thorough search under some faster criterion. The ability to find the globally optimal ML tree is therefore important. Here, I compare a range of heuristic search strategies (and their associated computer programs) in terms of their success at locating the ML tree for 20 empirical data sets with 14 to 158 sequences and 411 to 120,762 aligned nucleotides. Three distinct topics are discussed: the success of the search strategies in relation to certain features of the data, the generation of starting trees for the search, and the exploration of multiple islands of trees. As a starting tree, there was little difference among the neighbor-joining tree based on absolute differences (including the BioNJ tree), the stepwise-addition parsimony tree (with or without nearest-neighbor-interchange (NNI) branch swapping), and the stepwise-addition ML tree. The latter produced the best ML score on average but was orders of magnitude slower than the alternatives. The BioNJ tree was second best on average. As search strategies, star decomposition and quartet puzzling were the slowest and produced the worst ML scores. The DPRml, IQPNNI, MultiPhyl, PhyML, PhyNav, and TreeFinder programs with default options produced qualitatively similar results, each locating a single tree that tended to be in an NNI suboptimum (rather than the global optimum) when the data set had low phylogenetic information. For such data sets, there were multiple tree islands with very similar ML scores. The likelihood surface only became relatively simple for data sets that contained approximately 500 aligned nucleotides for 50 sequences and 3,000 nucleotides for 100 sequences. The RAxML and GARLI programs allowed multiple islands to be explored easily, but both programs also tended to find NNI suboptima. A newly developed version of the likelihood ratchet using PAUP* successfully found the peaks of multiple islands, but its speed needs to be improved.     

Multiple genome alignments facilitate development of NPCL markers: a case study of tetrapod phylogeny focusing on the position of turtles

In recent years, the increasing availability of genomic resources has provided an opportunity to develop phylogenetic markers for phylogenomics. Efficient methods to search for candidate markers from the huge number of genes within genomic data are particularly needed in the era of phylogenomics. Here, rather than using the traditional approach of comparing genomes of two distantly related taxa to develop conserved primers, we take advantage of the multiple genome alignment resources from the the University of California-San Cruz Genome Browser and present a simple and straightforward bioinformatic approach to automatically screen for candidate nuclear protein-coding locus (NPCL) markers. We tested our protocol in tetrapods and successfully obtained 21 new NPCL markers with high success rates of polymerase chain reaction amplification (mostly over 80%) in 16 diverse tetrapod taxa. These 21 newly developed markers together with two reference genes (RAG1 and mitochondrial 12S-16S) are used to infer the higher level relationships of tetrapods, with emphasis on the debated position of turtles. Both maximum likelihood (ML) and Bayesian analyses on the concatenated data combining the 23 markers (21,137 bp) yield the same tree, with ML bootstrap values over 95% and Bayesian posterior probability equaling 1.0 for most nodes. Species tree estimation using the program BEST without data concatenation produces similar results. In all analyses, turtles are robustly recovered as the sister group of Archosauria (birds and crocodilians). The jackknife analysis on the concatenated data showed that the minimum sequence length needed to robustly resolve the position of turtles is 13-14 kb. Based on the large 23-gene data set and the well-resolved tree, we also estimated evolutionary timescales for tetrapods with the popular Bayesian method MultiDivTime. Most of the estimated ages among tetrapods are similar to the average estimates of the previous dating studies summarized by the book The Timetree of Life.     

Reticulate evolution on a global scale: a nuclear phylogeny for New World Dryopteris (Dryopteridaceae)

Reticulate, or non-bifurcating, evolution is now recognized as an important phenomenon shaping the histories of many organisms. It appears to be particularly common in plants, especially in ferns, which have relatively few barriers to intra- and interspecific hybridization. Reticulate evolutionary patterns have been recognized in many fern groups, though very few have been studied rigorously using modern molecular phylogenetic techniques in order to determine the causes of the reticulate patterns. In the current study, we examine patterns of branching and reticulate evolution in the genus Dryopteris, the woodferns. The North American members of this group have long been recognized as a classic example of reticulate evolution in plants, and we extend analysis of the genus to all 30 species in the New World, as well as numerous taxa from other regions. We employ sequence data from the plastid and nuclear genomes and use maximum parsimony (MP), maximum likelihood (ML), Bayesian inference (BI), and divergence time analyses to explore the relationships of New World Dryopteris to other regions and to reconstruct the timing and events which may have led to taxa displaying reticulate rather than strictly branching histories. We find evidence for reticulation among both the North and Central/South American groups of species, and our data support a classic hypothesis for reticulate evolution via allopolyploid speciation in the North America taxa, including an extinct diploid progenitor in this group. In the Central and South American species, we find evidence of extensive reticulation involving unknown ancestors from Asia, and we reject deep coalescent processes such as incomplete lineage sorting in favor of more recent intercontinental hybridization and chloroplast capture as an explanation for the origin of the Latin American reticulate taxa.     

No speed dating please! Patterns of social preference in male and female house mice

Glass sponges (Class Hexactinellida) are important components of deep-sea ecosystems and are of interest from geological and materials science perspectives. The reconstruction of their phylogeny with molecular data has only recently begun and shows a better agreement with morphology-based systematics than is typical for other sponge groups, likely because of a greater number of informative morphological characters. However, inconsistencies remain that have far-reaching implications for hypotheses about the evolution of their major skeletal construction types (body plans). Furthermore, less than half of all described extant genera have been sampled for molecular systematics, and several taxa important for understanding skeletal evolution are still missing. Increased taxon sampling for molecular phylogenetics of this group is therefore urgently needed. However, due to their remote habitat and often poorly preserved museum material, sequencing all 126 currently recognized extant genera will be difficult to achieve. Utilizing morphological data to incorporate unsequenced taxa into an integrative systematics framework therefore holds great promise, but it is unclear which methodological approach best suits this task. Here, we increase the taxon sampling of four previously established molecular markers (18S, 28S, and 16S ribosomal DNA, as well as cytochrome oxidase subunit I) by 12 genera, for the first time including representatives of the order Aulocalycoida and the type genus of Dactylocalycidae, taxa that are key to understanding hexactinellid body plan evolution. Phylogenetic analyses suggest that Aulocalycoida is diphyletic and provide further support for the paraphyly of order Hexactinosida; hence these orders are abolished from the Linnean classification. We further assembled morphological character matrices to integrate so far unsequenced genera into phylogenetic analyses in maximum parsimony (MP), maximum likelihood (ML), Bayesian, and morphology-based binning frameworks. We find that of these four approaches, total-evidence analysis using MP gave the most plausible results concerning congruence with existing phylogenetic and taxonomic hypotheses, whereas the other methods, especially ML and binning, performed more poorly. We use our total-evidence phylogeny of all extant glass sponge genera for ancestral state reconstruction of morphological characters in MP and ML frameworks, gaining new insights into the evolution of major hexactinellid body plans and other characters such as different spicule types. Our study demonstrates how a comprehensive, albeit in some parts provisional, phylogeny of a larger taxon can be achieved with an integrative approach utilizing molecular and morphological data, and how this can be used as a basis for understanding phenotypic evolution. The datasets and associated trees presented here are intended as a resource and starting point for future work on glass sponge evolution.