Multi-Locus Plastid Phylogenetics Resolves Evolutionary Relationships and Buffers Rate Heterogeneity Across 13 Plant Taxa

Abstract Single-locus phylogenetic reconstructions frequently suffer from limited topological resolution, low node support, and susceptibility to lineage-specific evolutionary rate heterogeneity. In plant molecular systematics, single plastid markers such as rbcL often fail to provide sufficient informative sites across disparate evolutionary timescales. This study presents a comparative phylogenetic evaluation across 13 diverse plant taxa—spanning green algae, mosses, gymnosperms, monocots, and eudicots—to assess whether concatenating two plastid markers (rbcL and atpB) improves tree topology, resolves node support, and mitigates distance-based and character-based topological artifacts relative to single-gene approaches. Nucleotide sequences were aligned using MUSCLE and concatenated into a combined matrix of 3,172 base pairs (bp). Phylogenetic trees were constructed in MEGA using both Neighbor-Joining (NJ) and Maximum Likelihood (ML) methods. The single-gene tree (Case I: rbcL, ∽1,400 bp / 1,665 positions) under ML exhibited topological instability, severe long-branch behavior, and failed to recover major evolutionary clades, displaying a distorted, polyphyletic placement of the Poaceae family (Oryza sativa, Zea mays, Triticum aestivum). Conversely, the multi-locus concatenated dataset (Case II: rbcL + atpB, 3,172 bp) successfully resolved major lineages under both NJ and ML inference frameworks, recovering Poaceae monophyly with high bootstrap confidence (85%–100%) and placing the basal outgroup Gonium multicoccum (green algae) cleanly at the root. Furthermore, topological artifacts observed in basal taxa highlight the impact of heterotachy and rate variation, wherein single-locus algorithms misinterpret rapid sequence divergence in early-diverging lineages as deep phylogenetic distance (Long Branch Attraction). Overall, multi-locus concatenation effectively buffers localized rate variation, demonstrating its necessity for robust plant evolutionary inference.