Network-based Phylogenetic Analysis — Phylogenetic Network Inference
Phylogenetic Network Analysis · Also known as: phylogenetic network, reticulate phylogenetics, split network analysis, evolutionary network inference
Network-based phylogenetic analysis constructs graph-structured representations of evolutionary relationships that explicitly accommodate reticulate events — including hybridization, horizontal gene transfer, recombination, and incomplete lineage sorting — which strictly bifurcating phylogenetic trees cannot represent. Instead of forcing sequences into a single bifurcating tree, the method infers splits or reticulations in the data and visualises them as a network, revealing conflicting phylogenetic signals that are biologically informative.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use network-based phylogenetic analysis when organisms under study are known or suspected to undergo reticulate evolution — including plants with documented hybridisation, bacteria or archaea with widespread horizontal gene transfer, and RNA viruses prone to recombination. It is also the right choice when different genomic loci yield strongly conflicting trees, signalling genuine mosaic ancestry rather than estimation noise. Do not use this method as a drop-in replacement for standard tree inference when data are purely tree-like and the research goal is estimating divergence times or branch-specific rates: tree-based maximum likelihood or Bayesian methods (e.g., RAxML, BEAST) are more powerful and interpretable in that setting. Network methods also require good-quality multiple sequence alignment; poorly aligned data amplify artefactual reticulation signals.
Strengths & limitations
- Directly visualises conflicting phylogenetic signals without forcing a single tree topology onto the data.
- Biologically appropriate for organisms with documented hybridisation, recombination, or horizontal gene transfer.
- Split networks (Neighbor-Net, split decomposition) scale well to hundreds of taxa with modest computational cost.
- Distinguishes genuine reticulation from noise by quantifying bootstrap support on individual splits.
- Provides an honest exploratory tool — wide parallelogram boxes are a diagnostic cue, not a hidden artefact.
- Reticulate network methods that infer the full ancestry graph (ancestral recombination graphs, hybridisation networks) are computationally intensive and do not scale to thousands of taxa without approximations.
- Distinguishing true biological reticulation from alignment error, long-branch attraction, or model misspecification is difficult and requires domain expertise.
- Rooted reticulate networks require an outgroup or molecular clock assumption; unrooted split networks carry no directional information about time or ancestry.
- Software ecosystems (SplitsTree, PhyloNet) are less standardised than tree-inference pipelines, creating reproducibility challenges across studies.
- Interpretation of reticulation nodes as biological events demands independent corroborating evidence and should not rest solely on the network topology.
Frequently asked
What is the difference between a split network and a reticulate network?
A split network (e.g., Neighbor-Net output from SplitsTree) is an unrooted graph that represents conflicting signals as parallelogram boxes; it makes no explicit claim about which conflicts are due to hybridisation versus noise. A reticulate network places directional reticulation edges on a rooted topology and explicitly models each reticulation node as a biological event such as hybridisation with estimated inheritance proportions. Split networks are faster and exploratory; reticulate networks are statistically demanding but biologically interpretable.
Can I use this method on whole-genome SNP data instead of aligned sequences?
Yes. Distance-based methods such as Neighbor-Net accept pairwise SNP distance matrices computed from genotype data. Some tools also accept SNP alignment files directly. However, missing genotype data and SNP ascertainment bias can distort distance estimates, so filtering and imputation steps should precede network construction.
How do I know whether a wide box in my Neighbor-Net network is biologically meaningful?
Wide boxes indicate incompatible splits — positions where no single tree explains all patterns. Run a four-gamete test to confirm that the incompatibility is real and not an alignment artefact. Then check whether bootstrap support on the bounding splits is high (typically above 70–80%). Finally, look for independent biological evidence of hybridisation or recombination (e.g., intermediate morphology, discordant gene trees) before claiming the box represents a true reticulation event.
What software should I use?
SplitsTree5 is the standard for split network visualisation and analysis. PhyloNet handles explicit reticulate network inference under a coalescent framework. Dendroscope is useful for rendering large networks. For recombination detection in viruses, RDP4 and GARD (in HyPhy) are widely used. R packages such as ape and phangorn offer network functions for scripted workflows.
Is network-based analysis appropriate when I have only a few taxa?
Split networks can be constructed with as few as four taxa, but the diagnostic value of the network increases with sample size. With very few taxa (fewer than eight to ten), the network is unlikely to reveal complex reticulation patterns. For small taxon sets, explicit reticulate methods that test specific hybridisation hypotheses (e.g., D-statistics, ABBA-BABA tests) may provide clearer answers than a network visualisation.
Sources
- Bandelt, H.-J., & Dress, A. W. M. (1992). Split decomposition: A new and useful approach to phylogenetic analysis of distance data. Molecular Phylogenetics and Evolution, 1(3), 242–252. link ↗
- Bryant, D., & Moulton, V. (2004). Neighbor-Net: An agglomerative method for the construction of phylogenetic networks. Molecular Biology and Evolution, 21(2), 255–265. link ↗
How to cite this page
ScholarGate. (2026, June 3). Phylogenetic Network Analysis. ScholarGate. https://scholargate.app/en/bioinformatics/network-based-phylogenetic-analysis
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Bayesian Phylogenetic AnalysisBioinformatics↔ compare
- Genome-wide association studyBioinformatics↔ compare
- Multi-omics Phylogenetic AnalysisBioinformatics↔ compare
- Phylogenetic AnalysisBioinformatics↔ compare
- Sequence AlignmentBioinformatics↔ compare
- Variant CallingBioinformatics↔ compare