The IMPRINT study has given us a lot of information using shotgun metagenomic sequencing of longitudinal fecal samples: https://www.cell.com/cell/fulltext/S0092-8674(21)00660-7

There seem to be windows where bacterial populations are in flux until the microbiome stabilizes around 3 years of age. It is difficult to modify the microbiome after that. I think our models have to shift so that we prioritize early intervention, especially for those at risk for dysbiosis (like premature and c-section babies).