I've implemented SHA-3 and Keccak in hardware (FPGA) and software (CPU, GPU) countless times: there's zero scenario where this single byte change occurring before 24 rounds of massive permutation has any measurable effect on the security of this hash function.