Unraveling Virus Evolution: How Phylogenetic Trees Reveal Hidden Relationships

Biological species, including viruses, change over generations through evolution, with mutations becoming fixed in successful genomes and forming genetic lineages. Limited replication fidelity or environmental factors can alter, insert, or delete nucleotides, while recombination mechanisms introduce additional genomic innovation. Accepted changes—mutations—may be neutral, beneficial, or harmful, and their fate depends on population size and environment.

Phylogeny describes the relationships among lineages sharing common ancestry and the methods used to reconstruct those relationships. Although viruses were once thought unsuitable for phylogenetic reconstruction due to scarce fossils and high mutation rates, molecular data (nucleotide and amino‑acid sequences, occasionally three‑dimensional polymer structures) now enable virus phylogeny, typically visualized as trees such as the Tree of Life (ToL).

Virus phylogeny follows the same theoretical framework developed for cellular life. To infer phylogeny, one compares sequences of presumed homologous taxa. If all lineages evolve at a constant rate (a molecular clock), the model may be limited for viruses, which often evolve at variable, fluctuating rates, with site‑specific reversions. Consequently, reconstructing the full record of changes incurs increasing uncertainty with each new mutation, and accumulated inter‑species differences may develop non‑linearly over time.

The ultimate goal is to reconstruct relationships among all viral isolates and species. Because viruses lack a universal molecular denominator, a comprehensive viral phylogeny would require comparing viral and cellular genomes—a task still largely ongoing. Most viral phylogenetic work focuses on well‑sampled lineages of medical relevance. High‑throughput next‑generation sequencing (NGS) and metagenomics have recently enabled large‑scale phylogenetic studies of viral groups and entire viromes, shedding light on virus evolution, life cycles, and host interactions.

Our knowledge of contemporary viral diversity is steadily improving, yet only a fraction of the total diversity has been described, suggesting many undiscovered lineages may exist.

Definition of Trees

Similarity among species depends on evolutionary speed and divergence time. The process from a common ancestor to present‑day diversity is modeled as a chain of intermediate ancestors. In tree visualizations, the root represents the common ancestor, internal nodes correspond to intermediate ancestors, and terminal nodes (leaves) represent extant taxa. Leaves are often called operational taxonomic units (OTUs), while unobservable internal nodes and the root are termed hypothetical taxonomic units (HTUs). Nodes are connected by branches (edges).

Trees can be characterized by topology, branch length, shape, and root position. Topology reflects the relative arrangement of internal and terminal nodes and defines the branching events that generated current diversity. Congruent trees share identical topologies. Branch length may represent a fixed amount of change or elapsed time, leading to “additive” or “ultrametric” trees. Shape can relate to evolutionary details such as population size changes and selection pressures. The root determines evolutionary direction; descendants of a rooted internal node form a clade whose most recent common ancestor (MRCA) indicates monophyly. When branch lengths or root positions are undefined, the tree is called a cladogram or an unrooted tree, respectively.

Phylogenetic Tree Analysis

Multiple nucleotide or amino‑acid sequence alignments that maximize similarity among the taxa are the traditional input for phylogenetic analysis. Alignment quality is a critical determinant of inference quality. Because of the redundancy of the genetic code, nucleotide sequences accumulate changes faster than protein sequences. In viruses (including RNA viruses), this disparity is not offset by constraints such as dinucleotide frequencies or secondary structure, so nucleotide data are usually limited to closely related taxa, while protein sequences retain stronger phylogenetic signal for more distant relationships.

Differences derived from alignments can be quantified as pairwise distances (forming a distance matrix) or as discrete character states at specific sites. Distance methods, praised for speed, are preferred for very large datasets, though character‑state methods have narrowed the performance gap. UPGMA was the first clustering method, iteratively merging the most similar pairs while recomputing distances. Neighbor‑Joining (NJ) employs a more sophisticated algorithm that minimizes total branch length and is the most popular distance method.

Character‑based approaches evaluate many alternative trees to find the best fit under computationally intensive criteria. Maximum Parsimony (MP) seeks the tree requiring the fewest substitutions, while Maximum Likelihood (ML) provides a statistical framework that incorporates models of population size change, mutation rates, and other parameters. ML is mathematically robust and can be combined with other tree‑building techniques. Bayesian variants of ML incorporate prior knowledge (e.g., known substitution rates or fossil calibrations) and generate a forest of trees that reflect uncertainty, from which a consensus tree and branch support values are derived. In viral phylogenetics, Bayesian dating often supplies MRCA dates, while fossil calibrations are used for cellular trees.

Different reconstruction methods can yield trees with varying topologies and branch lengths, though ML and Bayesian trees often show better agreement, especially regarding branch lengths. No single method outperforms all others in every aspect; consequently, applying multiple methods and trusting only the congruent results is a common practice.

Support values for internal nodes are often assessed by bootstrap analysis, which generates many pseudo‑replicate datasets by random resampling of the original alignment. The proportion of replicates in which a node appears constitutes its bootstrap value; high values indicate reliability. In Bayesian analyses, posterior probabilities serve a similar purpose.

If taxa evolve according to a molecular clock, the root can be placed directly from observed inter‑species differences, or external knowledge (e.g., known out‑group taxa) can be used to root the tree. Out‑groups are taxa assumed to have diverged before the ingroup’s radiation. Unrooted trees are common in viral studies because the applicability of a strict molecular clock is often untested, and reliable out‑groups may be unavailable. Relaxed‑clock models aim to infer rooted trees without imposing a constant rate.

Viral phylogenies can be inferred from whole genomes or individual genes, both standard approaches in phylogenomics. Whole‑genome alignments work best for relatively close viruses; recombination can complicate analysis. Genes lacking recombination evidence can be concatenated to improve signal. For viruses with small genomes, single‑gene trees are typical, though their representativeness is debated. Network methods have recently been employed to depict multi‑gene evolutionary relationships, accounting for gene‑specific affinities.

When a gene tree is used to represent the whole genome’s phylogeny, it is often assumed that topology, rather than branch length, reflects shared evolutionary history despite differing substitution rates across genome regions. This assumption can be violated by homologous gene exchange, gene duplication, horizontal gene transfer, and other recombination events, leading to incongruent trees for different regions. Technical issues related to dataset size and diversity can also produce inconsistencies. Such conflicts are examined using consistency tests that identify recombination. Additional challenges include unresolved deep splits, long‑branch attraction (LBA), and the inability to resolve highly divergent lineages.

Applications of Viral Phylogenetic Trees

Phylogenetic analysis is widely applied to viral research, addressing both applied and fundamental questions such as epidemiology, diagnostics, forensic investigations, phylogeography, and viral origin, evolution, and taxonomy. During outbreaks, phylogeny helps determine a virus’s characteristics and origins, informing immediate control measures, vaccine design, and antiviral development.

Well‑sampled lineages—e.g., influenza, HIV, hepatitis C, poliovirus—benefit from extensive databases that capture known natural diversity. When a new virus clusters with known taxa, its evolutionary origin becomes apparent. Combined gene‑specific and whole‑genome phylogenies can reveal whether recombination contributed to a lineage’s emergence, as seen in rare recombination events in hepatitis C versus frequent recombination in poliovirus.

For truly novel zoonotic infections (e.g., Nipah, SARS‑CoV, MERS‑CoV, Ebola, Zika), phylogeny aids classification, identifies potential animal reservoirs, and traces transmission dynamics. Early uncertainties surrounding SARS‑CoV’s placement highlighted challenges posed by limited sampling and divergent lineages, which were later resolved.

Large‑scale sampling and extensive phylogenetic work are required to catalogue emerging zoonoses. Phylogenies have clarified the origins of HIV‑SIV inter‑species transmission, traced “HIV‑dentist” outbreaks, and illustrated how geographic isolation shapes viral spread, as demonstrated by JC polyomavirus and West Nile virus distributions.

Phylogenetic studies also reveal the strength and timing of virus‑host associations. Frequent host‑jump events in coronaviruses have produced multiple human pathogens (SARS‑CoV, MERS‑CoV, HCoV‑OC43). Conversely, herpesvirus phylogenies show remarkable co‑speciation with hosts over ~400 million years.

In taxonomy, phylogenetic insights drive reclassification: human hepatitis E virus was moved out of the Caliciviridae based on genomic evidence, while new families such as Marnaviridae and Dicistroviridae were established. For large DNA bacteriophages, phylogeny plays a smaller role, but tree‑based analyses still inform family‑level groupings.

Reference:

Gorbalenya AE, Lauber C. Phylogeny of Viruses. Reference Module in Biomedical Sciences. 2017:B978-0-12-801238-3.95723-4. doi: 10.1016/B978-0-12-801238-3.95723-4. Epub 2017 Jun 26. PMCID: PMC7157450.