Datalog is used in essential areas that require soundness, however testing Datalog engines is difficult because the correct output is often infeasible to verify manually. Building on the insight that e-graphs can drive metamorphic testing, we apply this approach to Datalog engines, a setting that introduces new technical challenges absent in prior work. This paper presents EDlogMT, an e-graph-based metamorphic testing framework for Datalog engines. Rather than transforming programs directly, EDlogMT lowers them into a custom intermediate representation, saturates an e-graph with semantics-preserving rewrites, and extracts semantically equivalent variants for equality-based metamorphic checking. A custom Rust implementation reveals a previously unreported bug in FlowLog, reproduces the relevant previously reported query bugs in its targeted setting, and improves observed code coverage across all evaluated source test-case generation methods. These results suggest that e-graphs are useful not only for optimization, but also as a practical foundation for systematic testing of Datalog engines.