The anisotropic thermal transport within Selective Laser Melting (SLM) melt pools, primarily induced by Marangoni flow, significantly influences molten pool morphology [1]. To computationally capture this phenomenon, this work develops a nonlocal, bond-based peridynamic (PD) framework [2] for simulating melt pool geometry in SLM. First, a novel anisotropic heat conduction PD model is established through a micro-conductivity formulation that ensures a positive-definite conductivity tensor for arbitrary anisotropy ratios. Building on this, an enthalpy-based phase-change PD model for anisotropic materials is developed to incorporate latent heat effects and track solid-liquid interface evolution. By integrating these advances with a Gaussian-exponential volumetric laser heat source, a unified PD-based computational framework for SLM is constructed. The proposed models are validated against finite-element solutions and experimental data using several benchmarks. Results demonstrate that the anisotropic heat conduction PD models accurately reproduce temperature distributions and interface evolution during directional phase change. Single-track SLM simulations further confirm that the PD-based framework can predict melt pool geometry and characteristic features across varying laser energy densities without explicit fluid dynamics coupling. This work thus provides a fundamental step toward fully coupled thermomechanical PD modelling for additive manufacturing process optimization.