Reliability is the top concern to runtimes. This paper studies how to test Wasm runtime, by presenting Wapplique, the first Wasm bytecode mutation-based fuzzing tool. Wapplique solves the diversity/efficiency dilemma in generating test cases with a specifically-tailored code-fragment substitution approach for Wasm. In particular, Wapplique appliqués code fragments from real-world programs to seed programs to enhance the diversity of the seeds. Via sophisticated code analysis algorithms we design, Wapplique also guarantees the validity of the resulting programs. This allows Wapplique to generate tremendous valid and diverse Wasm programs as test cases to well exercise target runtimes. Our experiences on applying Wapplique in testing four prevalent real-world runtimes indicate that it can generate test cases efficiently, achieve high coverage, and find 20 previously unknown bugs.