Culturing bacteria from plant material is well known to be conducive to strong bias compared to the actual diversity in the original samples. This bias is related to the bacterial cultivability, chemical composition of the media and culture conditions. Recovery bias is often observed but has never been quantified on different media using an amplicon barcoding approach whereby plant microbiota DNA extractions are compared to DNA extracted from serial dilutions of the same plant tissues grown on bacterial culture media. In this study, we: i) quantified the bacterial culturing diversity bias using 16S amplicon barcode sequencing by comparing a culture-dependent approach (CDA) focused on rice roots on four commonly used bacterial media (10% and 50% TSA, plant-based medium with rice flour, nitrogen free medium NGN and NFb) versus a culture-independent approach (CIA) assessed with DNA extracted directly from root and rhizosphere samples; ii) assessed enriched and missing taxa detected on the different media; iii) used biostatistics functional predictions to highlight metabolic profiles that could potentially be enriched in the CDA and CIA. A comparative analysis of the two approaches revealed that among the 22 phyla present in microbiota of the studied rice root samples, only five were present in the CDA (Proteobacteria, Firmicutes, Bacteroidetes, Actinobacteria, Verrucomicrobia). The Proteobacteria phylum was the most abundant in all CDA samples, showing high gamma-Proteobacteria enrichment. The diversity of the combined culture media represented about a third of the total microbiota diversity, and its genus diversity and frequency …