Fig. 4

Effect of different maternal predictors moderated by Age at sampling on the differential relative abundance of bacterial genera during the rich season. Panel A shows the effect of Early-preGC moderated by Age (Early-preGC*Age), panel B shows the effect of Late-preGC moderated by Age (Late-preGC*Age), and panel C shows the effect of PostGC moderated by Age (PostGC*Age). “UN” refers to undetermined/unclassified genus, and family or order is therefore reported. Colored bars indicate the effect direction (blue = positive, yellow = negative) of an increase in 1 SD of specific predictors (Early-preGC = 59.3 ng/g; Late-preGC = 67.0 ng/g; PostGC = 52.7 ng/g) estimated at the mean value of age (mean = 4.7 years), and at the mean value of all the other predictors. Red and black lines indicate the effect estimated at the mean plus 1 SD (red = 7.1 years) and at the mean minus 1 SD (black = 2.3) of age. Case example: Panel A 1st taxon from the right: when subjects are 4.7 years old (mean age), the abundance of Bifidobacterium experiences a 2.3 log-fold decrease with an increase in maternal GCs during the early gestation of 59 ng/g (1SD). When the subjects are 2.3 years old (mean age-1SD), the same increase in maternal GCs is associated with an 0.12 log fold increase in Bifidobacterium abundance (black whisker). Finally, when the subjects are 7.1 years old (mean age + 1 SD), the same increase in maternal GCs is associated with a 4.6 log-fold decrease in the abundance (red whisker). When the bar is not plotted, the effect of the same increase in maternal GCs at the mean age is 0. Taxa showing the black head (estimates at 2.3 years) further from 0 than the red head (estimates at 7.1 years) show an effect that is reduced with age