1. Phylogenetic comparative methods often assume strictly bifurcating trees, yet many lineages evolve through reticulation (e.g., hybridization, horizontal gene transfer: HGT). The MPSEM package implements Phylogenetic Eigenvector Maps (PEMs) but is limited to trees, excluding systems with network-like histories. 2. We introduce RPEM (v1.0), an open-source R package that extends PEM computation to general phylogenetic graphs, including reticulate networks. RPEM provides modular tools for graph construction (graph(), crosslink()), trait-aware parameter estimation (evolution.model()), and prediction at arbitrary locations (locate, predict). It interoperates with ape and supports cross-validated modeling. 3. Using 13 odd-toed ungulates, we compare trait predictions across 11 graphs (tree + 10 networks). A model with 5 reticulations achieved the highest prediction coefficient (P²=0.943), substantially outperforming the tree-only model (0.643), demonstrating that moderate reticulation improves predictive accuracy. 4. Synthesis and applications: RPEM enables ecologists and evolutionary biologists to apply PEM-based modeling to reticulate systems—such as hybridizing plants, HGT-prone microbes, or viral recombinants—where tree-based methods fail. By generalizing PEM to graphs, RPEM bridges a critical gap between theoretical flexibility and practical implementation in comparative biology.