The main reason is speed. The first priority in USEARCH is always to provide the fasted possible algorithms. Gzipped FASTQs are slow because (1) Linux pipes are slow, and (2) decompression is slow. A gzipped FASTQ is typically around 60% of the decompressed file, so the space saving is only 40%. In my opinion, the small space saving is not worth the slower speeds and implementation effort of supporting this file format. If you really do care about the space savings, then you can decompress the file before running usearch and compress again when usearch completes.