Is that documented somewhere by chance?
The best I can offer is the changelog (“Major software changes” → “Statistical inference” → “Multiple hypotheses”), which admittedly failed to explicitly state that the -negative command-line option was correspondingly removed (now added).