Despite their ecological importance, mosses remain under-represented in ecological studies due to their challenging detection in field surveys and morphological identification, exacerbated by the lack of expert botanists. In this study, we optimise an environmental DNA method for the detection of bryophytes from river water samples, with the aim of facilitating their inclusion in biodiversity assessments. We compared three different methods in terms of species detection and community dissimilarity at seven sites along a river. The methods include (i) visual transect surveys conducted by bryologists based on macro- and micro-morphology, (ii) multi-marker PCR metabarcoding of the rbcL and the ITS2 markers with newly designed primers targeting bryophytes, and (iii) hybridisation capture (HC) for the same markers. We found that PCR metabarcoding recovered over 50% (n = 37) of the species observed in the field, whereas hybridization capture detected only 16% (N = 11). PCR metabarcoding identified the most species, 101 species compared to 68 observed in the field and 27 with HC. Both the PCR and HC metabarcoding approaches identified bryophyte species not recorded in field surveys but expected in the catchment. Molecular methods, particularly PCR metabarcoding, recovered elusive and rare species difficult to observed in the field and occurring outside our transect. The two markers used in the molecular approaches contributed uniquely to species detection, making a multi-marker approach necessary to study this group. Environmental DNA and field surveys are complementary methods and together allow the discoverability of inconspicuous species and provide the most exhaustive species inventory.