RRBS is reduced representation bisulfite sequencing. So, it's BS-seq, just not of the whole genome (WGBS, or "whole-genome bisulfite sequencing"). This is typically done by predigesting the genome with a restriction enzyme (e.g. MspI) and then selecting a size-range of the resulting fragments for sequencing. RRBS gives higher coverage of target* areas at lower sequencing cost.
*"target" is perhaps the wrong word, you aren't really targeting a particular region of the genome, but rather enriching for a subset of the genome.
Just want to add some to the already excellent answer by Devon.
Because the recognition sequence of MspI is CCGG, RRBS reads are extremely enriched in regions with high GC content. These regions are typically CG islands or promoter regions. Therefore, RRBS is a great way to get high coverage methylation data on promoters but it generally misses the intergenic regions (where around half of the enhancers are located). Although people used RRBS to study distal regions such as enhancers and distal transcription factor binding sites, only a tiny fraction of them can be covered and there is no guarantee that the covered ones are representative. Just be cautious about conclusions about distal regions drawn simply based on RRBS.