In this study, we first examined the temporal succession of specific bacterial groups in the infant stool microbiota sampled from two geographical locations (i.e., Singapore and Yogyakarta, Indonesia). These two populations in Asia are considered diverse, in terms of geographical distance, cultural ethnicity and overall social economic development. Despite these differences between the two populations, our findings showed that the dynamic of colonization was similar in both cohorts. For example, Enterobacteriaceae and Bifidobacterium constitute the predominant bacterial groups in stool microbiota before three months of age, and were present at a relative abundance of up to 98% of total bacteria. This observation is in agreement with past reports which found that healthy infants from Netherlands, breastfed Indian infants from Guatemala, preterm infants from Nigeria and 6-week old infants across Europe also had a similar predominance of Enterobacteriaceae and Bifidobacterium [10, 13–15]. As the infants age, our study also showed that Firmicutes represented by members of the Eubacterium rectale-Clostridium coccoides group increased in its abundance, and gradually resembled that of an adult stool microbiota i.e. mainly populated with members of the Firmicutes and Bacteroidetes phylum . The similarities in the pattern of colonization from early till late infancy despite geographical differences may be related to multiple factors. For example, the prevalence of facultative anaerobes Enterobacteriaceae during early life may be due to a relatively aerobic gastrointestinal tract, and the need for the facultative anaerobes to deplete the oxygen content so as to provide an anoxic environment suitable for other commensal microbes to establish .
There remains no clear explanation for the predominance of Bifidobacterium in most infants, including those who were exclusively formula-fed, but not in adults. A possible reason may be related to the diet consumed by the human host at different stages of life. To illustrate, dietary carbohydrates that are consumed by infants comprise mainly disaccharides (lactose) and oligosaccharides [18, 19], which are in turn rapidly hydrolyzed to form galactose and glucose monosaccharides [20, 21]. A portion of these monosaccharides becomes available for the commensal microbiota, and because Bifidobacterium spp. produce more ATP per mole of glucose through the bifidus pathway , there remains a selective advantage for Bifidobacterium to out compete the other commensal bacterial groups fermenting carbohydrates through the conventional glycolysis and 6-phosphogluconate pathways. Subsequently, as the host matures and undergoes weaning, the dietary carbohydrates become more complex and eventually favour the establishment of other bacterial members belonging to the Bacteroides and Clostridium for instance, which are known to contain a wide repertoire of polysaccharides-utilizing gene clusters that can effectively degrade complex dietary carbohydrates [23–25]. A recent study conducted in five European centres reported a decrease in the abundance of Bifidobacterium, Enterobacteriaceae and certain species of Clostridium, whereas the abundance of members belonging to the Clostridium coccoides and Clostridium leptum groups increased after weaning . Variations in dietary habits between Singapore and Indonesia may explain the differences in rates of colonization of these bacterial groups between Singapore and Indonesian subjects and therefore the slopes of the curves with age for Bifidobacterium, Clostridium leptum and Bacteroides (Figure 2).
A low relative abundance of the Bacteroides-Prevotella group was observed throughout all time points up till the age of 12 month (mean 7.31%). Our previous publication based on 16S rRNA pyrosequencing reported similar proportion of Bacteroides (8.90%) in healthy infants at 12 months  and substantiates the findings in this current study. On the contrary in adult the Bacteroidetes co-inhabits with the Firmicutes and both phyla dominate the bacterial community of the human gut microbiome [16, 27, 28]. The structure of the infant gut microbiome is dynamic and evolves over the first years of life toward an adult-like microbiota [29–31].
Besides monitoring for the temporal succession of stool microbiota, we further evaluate if demographic and lifestyle differences in the two studied geographical locations (Singapore, SG and Indonesia, IN) would influence the abundance of specific bacterial groups. A study conducted across Europe showed that the geographic origin had an impact on the composition of the gut microbiota , and it remains unknown if the structure of the microbiota is influenced to the same extent in Asia. In this study, both SG and IN differ in its extent of development and urbanization, and we observed a higher relative abundance of Bifidobacterium in the SG cohort compared to IN. This might be a common feature of urban populations, as it has also been reported previously for Northern European countries such as Stockholm to have a higher abundance of Bifidobacterium in infants stool microbiota as compared to those sampled in the Spanish province of Granada .
In addition, the two geographical locations in this study differ significantly in various aspects, for instance in mode of delivery, feeding history, occurrence of antibiotics consumption and sibling number. Interestingly, these factors studied have also been associated with the development of allergic diseases [32–35]. It has been postulated that the influence of these factors have on atopic disease may at least be in part through the effects on profile of gut microbiota. When we examined the effects of demographic and lifestyle factors, we found that the mode of delivery had the largest effect on stool microbiota of infants. These observations are supported by previous studies, where higher numbers of bacterial members belonging to the genus Bifidobacterium [36, 37], Bacteroides and Atopobium group were observed for vaginal delivered infants compared to caesarean delivered infants [8, 10]. Compared to caesarean delivered infants, the general notion is that infants who are vaginally delivered are exposed to their mothers' vaginal microbiota, in particular the Lactobacillus spp. , and therefore would have a higher incidence of bacterial transmission to their gut. However, in contrast to previous report which detected a higher abundance of Lactobacillus spp. in vaginally delivered infants , we detected a lower abundance of Lactobacilli-Enterococci group in our studied cohort. This discrepancy may be due to the specificities of different oligonucleotide primers/probes used to target the Lactobacillus-Enterococci group. Alternatively, the close adherence of Lactobacillus spp. to mucosal layers might hinder its transmission to the infants while the other vaginal microbiota gets transmitted to the infant . Future validations on a larger cohort of vaginally delivered infants residing in SG and IN will be needed to verify the associated low abundance of Lactobacillus.
Our study also showed that vaginal delivered infants had a significantly higher number of terminal restriction fragments (T-RFs) and microbial richness at 12 months of age. Previous studies had reported that the diversity of stool microbiota increased over time . We postulate that the higher abundance of beneficial bacteria such as Bifidobacterium associated with vaginal delivery may promote the diversity of overall gut microbiota as the infant ages.
Our findings also suggest that antibiotics consumption and sibling number are potential factors that influence the bacterial composition of the human fecal microbiota. For example, the consumption of postnatal antibiotic exposure resulted in a higher relative abundance of members of the Clostridium leptum group at one year of age. Previous studies have also found that postnatal antibiotic intake were associated with decreased numbers of Bifidobacterium and Bacteroides [11, 42], further suggesting that antibiotics consumption can perturb the structure of the commensal microbiota. A higher abundance of Bifidobacterium was observed to be associated with the presence of older siblings . Furthermore, we noted a corresponding decrease in the abundance of Enterobacteriaceae with the number of siblings. Interestingly, Lewis and colleagues have previously reported a decrease in the incidence of allergy with the number of siblings , while our past studies have found higher abundance of Bifidobacterium spp. and decreased abundance of Enterobacteriaceae in healthy infants compared to infants with eczema [5, 6]. It remains to be further established if these multitude of factors: the sibship size and abundance of Bifidobacterium spp. and Enterobacteriaceae are intricately linked with the development of allergy and its related disorders. Besides demographic and lifestyle characteristics, the genetic make-up of the host has been proposed to be an important contributing factor in shaping the composition of the gut microbiota. Ambiguous findings were reported for twin studies on the influence of host genetics on gut microbiota [27, 43], whereas absence, mutation or variation of certain single host genes were shown to affect the composition of gut microbiota .
FISH-FC approach showed a phylogenetic gap ranging from 22.89% to 37.40% of total bacteria for the four time points. A similar bacterial coverage was reported by Fallani et al using the same method, where the sum of bacterial cells detected were 72.7% ± 24.5%  and 74.3% ± 18.9%  with a panel of 10 non-overlapping probes. We acknowledge that the molecular techniques applied in this study do not permit a thorough description of the bacterial population inhabiting the human colon. Future studies would aim to utilize deep sequencing of the 16S rRNA genes so as to delve in depth the bacterial communities populating the human microbiome [46, 47]. Their greater depths of sampling offer the opportunity to explore within the phylogenetic gap and beyond, therefore allowing high-resolution association studies involving the bacterial populations of the human microbiome as "quantitative traits".