Maternal stress is associated with oral microbiota diversity patterns in pregnant women in Quito, Ecuador

Abstract

Background

Psychological stress is present in almost all pregnancies and contributes to maternal and fetal physiology. Human microbiota is dynamic and highly responsive to maternal stress.

Aims

To characterize maternal stress levels and their association with oral microbiota diversity patterns in pregnant women attending a public hospital in Quito, Ecuador.

Methods

Pregnant women were recruited at a public hospital upon presentation for labor. Stress levels were assessed using the 13-item Perceived Stress Scale. Oral mucosa samples were collected before the mothers entered the delivery room. The V3–V4 region of the 16S rRNA gene, from the extracted DNA of the samples, was sequenced using the Illumina MiSeq® platform. Alpha and beta diversity were evaluated using Shannon, Faith's PD, and weighted UniFrac metrics. Group differences were tested with PERMANOVA and Kruskal–Wallis tests (p < 0.05).

Results

Fifty-six women were included, with low (35.7%), moderate (50.0%), and very high (14.3%) perceived stress levels. Alpha diversity was greater in the very high-perceived stress group compared with low and moderate groups (Shannon 4.58 vs. 3.85 and 3.99; p = 0.01–0.04; Faith 5.15 vs. 4.17 and 4.43; p = 0.01–0.02). Patescibacteria and Bacillota (Firmicutes) were depleted in highly stressed women. Beta diversity showed no differences.

Conclusion

Maternal stress was associated with increased oral alpha diversity, primarily driven by changes in community evenness, without significant shifts in beta diversity.

Highlights

• •Maternal stress increases oral microbial alpha diversity.
• •High stress depletes Bacillota and Patescibacteria.
• •Changes driven by microbial evenness, not composition.
• •No significant beta diversity differences across groups.
• •First microbiome evidence in an Ecuadorian pregnant cohort​.

1. Introduction

The trends in the global demographics of women's reproductive health are concerning. In 2024, an estimated 132 million babies were born, continuing a trend following global total fertility rates of 2.3 births per woman recorded in 2023. The United Nations projects that although global population growth will persist over the next few decades, reaching a peak in the mid-2080s, the annual number of births will steadily decrease while death rates will rise, ultimately leading to global population contraction by the end of the century [[1], [2], [3]].

The individual experience of pregnancy emerges as a critical determinant of maternal and child health. Pregnancy imposes multidimensional burdens such as health, educational, financial, social and inequity-related burdens [4], with health encompassing physical, mental and emotional domains. These burdens are shaped by modifiable factors, such as social conditions, discrimination and racism, and non-modifiable factors, including physiological and psychological states [4,5]. Understanding these determinants highlights the need for evidence-based interventions in maternal healthcare.

Stress is present in the majority of pregnancies with varying intensity [[6], [7], [8]]; although it emerges in early gestation [9], it typically intensifies during late pregnancy and often persists into the postpartum period [10]. Physiologically, prenatal stress has been linked to heightened activation of the endocrine and immune systems, disrupting maternal and possibly even fetal physiology [11]. Consequently, this systemic dysregulation has been associated with adverse fetal outcomes, including preterm labor, preterm delivery, low birth weight, and shortened gestation, as well as maternal complications such as preeclampsia and gestational diabetes [12]. These findings suggest that maternal stress may both contribute to and be a consequence of the overall pregnancy burden.

The relationship between stress and microbiota stability remains an emerging field [13,14]. Microbial communities across distinct body sites may respond differently to host physiological changes and inflammatory signaling, which modulate local ecological conditions and resource availability. Under these conditions, microorganisms that successfully colonize a given niche and efficiently exploit available resources undergo positive selection, thereby proliferating at the expense of competing taxa to achieve higher relative abundance within the community) [15].

While higher intestinal microbial diversity is typically associated with health, there is no consensus on the pattern of the oral microbiota, where higher and lower diversities often correlate with the presence of diseases [16,17]. Differences in microbial patterns between individuals with disease and healthy controls may represent reactive changes, contributors to pathology, or even causal factors [17]. Current evidence largely rests on correlations, with a need for causation studies to clarify these changes.

We aim to characterize maternal stress levels and their association with oral microbiota diversity patterns in pregnant women attending a public hospital in Quito, Ecuador.

2. Methods

2.1. Study design and clinical setting

We performed a cross-sectional observational study of pregnant women attending a public hospital in Quito, Ecuador, between 2022 and 2023. The study was approved by the Research Ethics Committee on Humans from the Universidad Central del Ecuador on August 30, 2022 (Approval No. 329-CEISH-UCE-2022). This study was not pre-registered due to its exploratory nature, aiming to characterize the oral microbiota in this specific population and identify potential patterns of association.

We excluded women with a diagnosis of diabetes, hypertension, HIV infection, preeclampsia, those who smoked, or those who had received any antibiotics in the previous three months. Both primiparous and multiparous women were eligible, and participants were enrolled in the obstetric unit at the onset of labor.

Sample size was determined by the available funding. Most participants self-identified as Mestiza, the most representative ethnic group in Ecuador. All participants were recruited consecutively and met the predefined inclusion and exclusion criteria, reducing selection and procedural bias.

2.2. Perceived stress assessment

Perceived stress was measured by a psychologist via the application of the 13-item Perceived Stress Scale (PSS-13) [[18], [19], [20]], which was administered at least 24 h before delivery. The items were rated on a five-point Likert scale (1 = “never” to 5 = “very often”). Scores were summed and transformed into percentile scales (0–100) and T-scores (20–80). The T-score model was selected and categorized into five levels: very high, high, moderate, low, and very low perceived stress.

2.3. Sample collection and DNA extraction

Oral mucosa samples were collected before the mothers entered the delivery room in the obstetric unit. The participants were instructed not to wear lip makeup or brush their teeth prior to sampling to avoid gingival bleeding. Four swabs (two per cheek) were collected and transported in 1X PBS under a cold chain (∼8 °C). Genomic DNA was extracted via the AccuPrep® Genomic DNA Extraction Kit (Bioneer, USA) following the manufacturer's protocol for buccal samples (Becton Dickinson, USA).

2.4. 16S rRNA gene sequencing

The V3–V4 region of the 16S rRNA gene was amplified from the DNA of the oral samples via PCR using the forward 5′-TCGTCGGCAGCGTCAGATGTGTATAAGACACCCTACGGGGGCWGCAG-3′ and the reverse 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGCAGGACTACHVGGGTATCTAATCC-3′ primers, with an amplicon size of approximately 480 bp. Two negative controls for extraction and PCR were included. Amplicons were purified with AMPure Beads (Beckman Coulter, USA), and libraries were prepared using the Nextera XT DNA Kit (Illumina, USA). We determined nucleic acid library concentrations with a Quantus™ fluorometer (Promega, Corp., USA) before sequencing on the Illumina MiSeq® platform, version 4.1.0.656 (Illumina, USA). We used QIIME2 v2024.2 to generate amplicon sequence variants (ASVs). Quality filtering was performed with the DADA2 plugin in QIIME2 v2024.2, using a minimum of 1000 sequences per sample, and rarefaction curves confirmed sufficient sequencing depth (Supplementary Fig. S1). Taxonomic assignment was performed against the SILVA 16S database [21].

2.5. Statistical analysis

Alpha diversity was assessed using the Shannon diversity, Faith's phylogenetic diversity, Observed ASVs, and Chao1 richness indices. Alpha rarefaction curves were generated to evaluate sequencing-depth adequacy, and samples with fewer than 1000 sequences were excluded from downstream diversity analyses (Supplementary Fig. S1). Differences in alpha diversity among perceived stress groups were evaluated using the nonparametric Kruskal–Wallis test due to non-normal distributions. Pairwise Kruskal–Wallis comparisons were subsequently performed between stress groups, and multiple-comparison-adjusted q-values were used to assess pairwise significance.

Associations between continuous PSS-13 scores and alpha-diversity metrics were evaluated using Spearman's rank correlation. The resulting p-values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate procedure.

Beta diversity was assessed using both weighted and unweighted UniFrac distances and visualized using principal coordinate analysis (PCoA). Differences in microbial community composition among perceived stress groups were evaluated using permutational multivariate analysis of variance (PERMANOVA) with 999 permutations. Because PERMANOVA may be influenced by differences in within-group dispersion, homogeneity of multivariate dispersion was additionally evaluated using PERMDISP with 999 permutations.

Differential abundance according to perceived stress categories was assessed using Analysis of Compositions of Microbiomes with Bias Correction (ANCOM-BC), implemented in QIIME 2 (v2024.2), at the phylum, class, order, family, and genus taxonomic levels. In addition, the association between microbial abundance and PSS-13 score as a continuous variable was evaluated using ANCOM-BC at the genus level. Taxa with a prevalence below 10% of samples were excluded from this analysis. Multiple-testing-adjusted q-values were used to determine statistical significance.

Finally, a post hoc power analysis was performed for the comparison between the low- and very high-perceived stress groups. Effect sizes were estimated using Cohen's d, and statistical power was calculated separately for Shannon diversity and Faith's phylogenetic diversity. Statistical significance was established at p < 0.05 for global tests and at q < 0.05 for analyses involving multiple comparisons.

3. Results

3.1. Participant characteristics and prevalence of stress

A total of 56 pregnant women were included in this study. The patients’ demographic characteristics are summarized in Table 1. Briefly, the median age was 25.1 years (range 22.1 - 29.4), 70% were single, all had at-term pregnancies of 39.35 weeks (range 38.3-40.1), 83% had cephalovaginal deliveries, and 17% had cesarean deliveries. The sex of the newborns was 63% male and 37% female. Only 48% of mothers had completed secondary education.

Table 1.

Maternal and neonatal characteristics by stress diagnosis.

Variable	N	Total (N = 56)	Low (N = 20)	Moderate (N = 28)	Very high (N = 8)	p value1
Age (years)	54	25.1 (22.1–29.4)	24.2 (21.9–28.2)	26.0 (23.0–30.7)	22.3 (20.2–30.0)	0.3
Gestational age (weeks)	54	39.35 (38.30–40.10)	39.10 (38.20–40.10)	39.40 (38.25–40.40)	39.35 (38.98–39.70)	0.6
Maternal height (cm)	52	155.0 (152.0–158.1)	155.0 (150.0–159.6)	153.0 (152.0–159.0)	158.5 (155.5–160.0)	0.2
Maternal weight (kg)	52	67 (60–73)	65 (57–72)	70 (61–74)	64 (58–71)	0.6
Delivery type	54					0.12
Vaginal		45 (83%)	14 (70%)	24 (92%)	7 (88%)
Cesarean		9 (17%)	6 (30%)	2 (7.7%)	1 (13%)
Neonate sex	54					0.4
Female		20 (37%)	5 (25%)	12 (46%)	3 (38%)
Male		34 (63%)	15 (75%)	14 (54%)	5 (63%)
Marital status	50					0.13
Married		6 (12%)	1 (6.6%)	2 (8.0%)	3 (43%)
Divorced		1 (2.0%)	1 (6.6%)	0 (0%)	0 (0%)
Single		25 (50%)	12 (67%)	20 (80%)	3 (43%)
Cohabiting		8 (16%)	4 (22%)	3 (12%)	1 (14%)
Maternal education	54					0.3
High school		3 (5.6%)	1 (5.0%)	2 (7.7%)	0 (0%)
None		8 (15%)	6 (30%)	2 (7.7%)	0 (0%)
Primary		12 (22%)	4 (20%)	7 (27%)	1 (13%)
Secondary		26 (48%)	8 (40%)	11 (42%)	7 (88%)
Tertiary		5 (9.3%)	1 (5.0%)	4 (15%)	0 (0%)

¹Kruskal‒Wallis rank sum test; Fisher's exact test.

The assessment of perceived stress with the PSS-13 instrument and categorization via the T-score model revealed that 35.7% of the participants experienced low perceived stress, 50.0% experienced moderate perceived stress, and 14.3% experienced very high perceived stress. No participants were classified as having very low or high perceived stress levels.

When demographic variables were analyzed by perceived stress category, no statistically significant differences were observed among mothers with low, moderate, or very high levels (Table 1).

3.2. Alpha diversity

After quality filtering, 56 samples were retained for downstream analyses. The sequencing quality was 36.25 according to the Phred score, and across the retained samples, 981 amplicon sequence variants (ASVs) were identified. The retained samples had a mean sequencing depth of 19,862.91 ± 10,265.46 reads per sample, with a median of 18,505.5 reads and a range of 1146–46,191 reads. Five samples with sequencing depths below the rarefaction threshold of 1000 reads were excluded. Among the retained samples, only one sample had a sequencing depth close to the rarefaction threshold, with 1146 reads, whereas the next lowest sequencing depths were 2,707, 6,472, and 6686 reads. Rarefaction curves showed that Shannon diversity and Faith's phylogenetic diversity approached a plateau by approximately 1000 reads in most samples, supporting the adequacy of the selected rarefaction depth (Supplementary Fig. S1).

The main phyla found in the oral microbiota were Bacillota (Firmicutes), Bacteroidota, and Pseudomonadota (Proteobacteria) across all stress groups (Fig. 1). Additionally, at the ASV level, the most abundant correspondant genera were Streptococcus, Prevotella, Veillonella, Neisseria, Haemophilus, Gemella, Porphyromonas, and Alloprevotella (Fig. 2).

Fig. 1.

Relative abundance profiles of major bacterial phyla in the oral microbiota of pregnant women with low (n = 20), moderate (n = 28), and very high (n = 8) perceived stress. The main phyla are shared among stress levels, although variation in evenness (uniformity) is most evident in the very high-stress group.

Fig. 2.

Relative abundance of amplicon sequence variants (ASVs) in the oral microbiota of pregnant women with low (n = 20), moderate (n = 28), and very high (n = 8) perceived stress. The main genera were shared among stress levels.

Comparison of alpha diversity revealed differences between perceived stress groups (Fig. 3). Shannon diversity was higher in the very high-perceived stress group (H = 4.58) than in the low-perceived stress group (H = 3.85; q = 0.025). No significant differences were observed between the low- and moderate-perceived stress groups (q = 0.452) or between the moderate- and very high-perceived stress groups (q = 0.078). Although Faith's phylogenetic diversity tended to be higher in the very high-perceived stress group, pairwise comparisons did not remain statistically significant after multiple-testing correction (low vs. moderate, q = 0.691; moderate vs. very high, q = 0.078; low vs. very high, q = 0.102).

Fig. 3.

Alpha diversity of the oral microbiota according to perceived stress level. Alpha diversity was evaluated using (A) Shannon diversity, (B) Faith's phylogenetic diversity (Faith's PD), (C) Observed ASVs, and (D) Chao1 richness. Participants were classified into low (n = 20), moderate (n = 28), and very high (n = 8) perceived stress groups. Boxplots show the distribution of each alpha-diversity metric within stress groups. Pairwise comparisons were performed using the Kruskal–Wallis test, and q-values represent multiple-testing-adjusted p-values. A significant difference was observed for Shannon diversity between the low- and very high-stress groups (q = 0.025), whereas the remaining pairwise comparisons for Shannon, Faith's PD, Observed ASVs, and Chao1 were not statistically significant after multiple-testing correction. Statistical significance was defined as q < 0.05.

To determine whether differences in alpha diversity were accompanied by differences in richness, Observed ASVs and Chao1 were additionally evaluated. Neither metric showed statistically significant differences among perceived stress groups after multiple-testing correction. For Observed ASVs, pairwise comparisons yielded q = 0.523 for low versus moderate, q = 0.089 for moderate versus very high, and q = 0.089 for low versus very high perceived stress. Similarly, Chao1 comparisons yielded q = 0.834, q = 0.120, and q = 0.120, respectively (Fig. 3). These results indicate that the significant difference observed in Shannon diversity was not accompanied by statistically significant differences in richness.

A post hoc power analysis based on the comparison between the low- and very high-perceived stress groups revealed large effect sizes for Shannon (Cohen's d = 1.00) and Faith's phylogenetic diversity (Cohen's d = 0.98), with statistical power values of 0.63 and 0.61, respectively.

When perceived stress was analyzed as a continuous variable, PSS-13 scores showed positive correlations with Shannon diversity (Spearman's ρ = 0.275, p = 0.040, q = 0.060), Faith's phylogenetic diversity (ρ = 0.278, p = 0.038, q = 0.060), and Observed ASVs (ρ = 0.269, p = 0.045, q = 0.060). Chao1 richness showed a weaker positive correlation (ρ = 0.219, p = 0.105, q = 0.105). However, none of these associations remained statistically significant after multiple-testing correction.

3.3. Beta diversity

Beta diversity analyses based on both weighted and unweighted UniFrac distances showed no significant differences in oral microbial community composition among perceived stress groups (Fig. 4). For weighted UniFrac, PERMANOVA and PERMDISP analyses revealed no significant differences among groups (pseudo-F = 1.081, p = 0.371) or in multivariate dispersion (F = 0.796, p = 0.480). Unweighted UniFrac analysis yielded similarly non-significant results for both community composition (pseudo-F = 0.793, p = 0.805) and dispersion (F = 0.639, p = 0.566). These findings indicate that low, moderate, and very high perceived stress levels are not significantly associated with differences in overall oral microbial community structure, whether based on abundance-weighted or presence/absence-based phylogenetic distances.

Fig. 4.

Oral microbiota beta diversity according to perceived stress level. Principal coordinate analysis (PCoA) based on weighted (A) and unweighted (B) UniFrac distances showing the beta diversity of the oral microbiota in women with low (n = 20), moderate (n = 28), and very high (n = 8) perceived stress. Each point represents the microbiota of one participant and is colored according to stress group. No significant differences in microbial community composition were detected among stress groups by PERMANOVA for weighted UniFrac (pseudo-F = 1.081, p = 0.371) or unweighted UniFrac (pseudo-F = 0.793, p = 0.805).

3.4. Differential abundance of bacterial phyla

Differential abundance analysis was performed using ANCOM-BC at the phylum, class, order, family, and genus levels. Significant differences were detected at the phylum and genus levels. At the phylum level, Bacillota (Firmicutes), one of the predominant phyla, and Patescibacteria, one of the least represented phyla, were significantly less abundant in women with very high perceived stress than in those with low perceived stress (Fig. 5). In the genus-level analysis, two taxonomic features, assigned as Saccharimonadales within Patescibacteria and Clostridia_UCG-014 within Firmicutes by the taxonomic classification pipeline, were significantly depleted in women with very high perceived stress compared with those with low perceived stress (LFC = −1.44, q = 0.0097 and LFC = −1.73, q = 0.0102, respectively; Fig. 5). No significant differences were detected at the class, order, or family levels. Additionally, no significant differences were identified between the moderate and very high perceived stress groups or between the moderate and low perceived stress groups.

Fig. 5.

Differential abundance analysis at the genus level according to perceived stress. ANCOM-BC identified two taxonomic features significantly depleted in women with very high perceived stress (n = 8) compared with those with low perceived stress (n = 20): a feature assigned to Saccharimonadales within Patescibacteria (LFC = −1.44, q = 0.0097) and a feature assigned to Clostridia_UCG-014 within Firmicutes (LFC = −1.73, q = 0.0102). Error bars represent the standard error of the estimated log fold change.

When perceived stress was analyzed as a continuous variable using the PSS-13 score, ANCOM-BC identified a significant positive association between Rothia abundance and increasing stress scores (LFC = 0.186, W = 3.97, p = 7.15 × 10⁻⁵, q = 0.0027). Rothia was the only genus that remained significantly associated with PSS-13 scores after multiple-testing correction (Fig. 6).

Fig. 6.

Association between perceived stress and the relative abundance of Rothia. Scatter plot showing the relationship between PSS-13 score and the relative abundance of Rothia in the oral microbiota. Each point represents one participant (n = 56). The line represents the fitted linear trend with its 95% confidence interval. ANCOM-BC analysis using PSS-13 score as a continuous variable identified a significant positive association between perceived stress and Rothia abundance (LFC = 0.186, q = 0.0027).

4. Discussion

Our findings are consistent with the hypothesis that pregnancy-related physiological drivers may influence both psychological stress responses and oral microbiota patterns.

Higher stress levels were associated with increased alpha diversity of the oral microbiota, primarily driven by changes in community evenness rather than by a compositional restructuring of taxa. The increase in alpha diversity without significant differences in beta diversity suggests that maternal stress may influence the internal ecological balance of the oral microbiota rather than driving large-scale community restructuring. Such patterns may reflect microbial resilience, where host physiological changes modify relative abundances within microbial communities without altering the overall community composition across individuals.

The observed association between stress levels and microbial diversity is consistent with previous studies linking psychological stress with alterations in the oral microbiome. For instance, Alex et al. [9], who studied pregnant women under 20 weeks of gestation in Michigan, USA, reported a correlation between oral microbiota patterns and life stress, anxiety, posttraumatic stress disorder (PTSD) and depression levels. Their results revealed that higher levels of such conditions were correlated with greater alpha diversity. Only PTSD levels were associated with differences in beta diversity. Notably, PTSD physiology differs from anxiety and stress responses [22]. This suggests that a common driver linking psychological stress and microbiota changes may be present throughout pregnancy.

A plausible biological mechanism involves activation of the hypothalamic–pituitary–adrenal (HPA) axis, leading to systemic endocrine shifts and immune modulation [11]. These host physiological responses may influence microbial ecology at mucosal surfaces, including the oral cavity [17]. However, while these mechanisms are extensively characterized in the gastrointestinal tract—specifically through gut–brain axis and HPA interactions—their role in the oral cavity remains under-researched [23]. Future studies integrating hormonal profiling, immune markers, and multi-omic microbiome analysis are essential to clarify the mechanistic pathways underlying these associations.

In this study, changes in diversity at the phylum level were attributable to depletion of the Patescibacteria and Bacillota (Firmicutes) phyla in the oral microbiota of women with very high perceived stress, which may demonstrate their susceptibility to reshaping. Bacillota (Firmicutes), the predominant phylum in the oral microbiota, shows a marked tendency to reshape under many other local or systemic conditions. For example, Bacillota (Firmicutes) may also increase in abundance in patients under orthodontic treatment at the 6-month follow-up [24], and was also reported to correlate with maternal anxiety in early pregnancy stages [9]. In the same context, Patescibacteria, which is not a predominant group in the oral microbiota, has also shown susceptibility to reshape in some conditions. For example, Erdem et al. reported increased Patescibacteria abundance in the oral microbiota of women with Hashimoto's thyroiditis compared with healthy controls, while differences were not observed in patients under levothyroxine treatment, and no functional changes were demonstrated [25]. And, among glioma patients, Patescibacteria abundance in the oral microbiota was inversely correlated with the degree of malignancy of the tumor [26].

Other dominant groups of oral bacteria such as Leptotrichia are also susceptible to reshaping. For example, gestational maternal diabetes and the oral microbiota were found strongly correlated with a significant reduction in their abundance [27,28], which remains depleted even after 9 months of the delivery [27]. This evidence raises the possibility that these phyla are particularly sensitive to modulation. This finding is supported in our analysis by the weighted UniFrac which takes account on evolutionary divergence and abundance. Our data did not show significant divergence between stress levels in this regard, representing steady richness of indigenous taxa between groups. Moreover, while our methodology successfully resolved community composition down to the genus level, it was unable to reach the strain and genotype levels, where functional consequences of microbiota patterns can also occur [29,30], as this would require whole-genome shotgun metagenomic sequencing.

On the other hand, the onset of microbiota shifts has not been ruled out, highlighting the need to follow up women before pregnancy to determine if such changes react to pregnancy-related physiological changes, if previous microbiota patterns are already present, or even if the pregnancy burden exacerbates previous patterns. Importantly, early microbiome studies demonstrated that taxonomic composition varies substantially across individuals, whereas functional pathways remain conserved, a phenomenon known as functional redundancy [31]. Accordingly, shifts in taxonomic profiles alone should not be regarded as sufficient evidence of dysbiosis; instead, they must be accompanied by alterations in metabolic pathways, metabolite production, or bacterial antagonism. Oral microbiota dysbiosis has only directly been part of the etiology in specific cases, such as periodontal diseases [17]. Thus, taxonomic variation may reflect either a detrimental disruption of functionality or an adaptive reconfiguration that preserves functional capacity under altered conditions and alone it cannot be proposed as a potential predictor of any autoimmune, inflammatory or non-inflammation-mediated disease risk as some suggested [13,25].

The taxonomic diversity of human microbiota has long been shown to shift in association with various syndromes and diseases. These alterations, often referred to as dysbiosis, have been hypothesized to contribute to disease progression or pathogenesis [14]. Based on this framework, some have suggested that restoring the microbiome to a core composition could reverse pathological processes [16].

However, several key considerations challenge this assumption. First, the definition of a core healthy microbiome or a general index of healthy/disease microbiota is still under discussion [32,33]. Moreover, a core microbiome change has not been consistently observed in individuals following inflammatory stress [27,34], and feasible strategies to find microbiota modifiers are still not available [35]. In contrast to symbiotic systems in other organisms, such as Wolbachia in insects, fermentative consortia in ruminants, or the Euprymna scolopes–Vibrio fischeri association, the human microbiota is largely commensal, with only a minor symbiotic component [29].

Despite its primarily commensal role, the human microbiota exerts a critical protective function through bacterial antagonism, whereby resident taxa inhibit colonization by external microbes. This antagonism is mediated by various mechanisms, including type VI secretion systems, antimicrobial peptide production, bacteriocin release, and quorum-sensing regulation [17]. Such processes not only preserve niche integrity but also may stabilize community composition against ecological perturbations [36]. Thus, the return to homeostasis after a microbiota-altering trigger may be accompanied by a reshaping of the microbiota at the expense of abundance or evenness; however, such changes may not be significant if functional redundancy remained stabilized during the trigger period. This highlights the importance of integrating multiomics approaches, such as transcriptomics, metabolomics, and inflammation profiling, into microbiome research, with emphasis on when core compositional changes are observed and to address its triggers [17,32,33].

Advances in metagenomic technologies, which enable comprehensive sequencing of all genes within a sample, offer powerful tools to unravel the complex genomic interactions underlying these processes. Such approaches can move beyond taxonomic shifts to clarify the functional dynamics of the microbiota. Future investigations should also integrate hormonal and immunological parameters to elucidate the mechanisms linking host physiology to microbiota modulation. Furthermore, future studies incorporating paired oral and fecal sampling would permit the simultaneous evaluation of oral and gut microbiomes in response to stress, offering a broader perspective on stress-associated microbial shifts.

5. Limitations

The measurement of stress in this study presents certain limitations. First, the stress scale employed was not originally developed for pregnant populations, which may constrain its sensitivity to prenatal stress-specific domains. However, the PSS-13 was selected since a Spanish-language version has been previously validated in a Latin American adult population. This validation provides methodological consistency and comparability for the current study, supporting its use over other widely used scales such as the PSS-10 or PSS-14.

Additionally, administering the PSS-13 upon admission for delivery introduces a temporal confound, as it can be difficult to distinguish chronic perceived stress from the acute anxiety inherent to the peripartum window, potentially overestimating the proportion of women categorized with moderate and very high stress. Furthermore, this self-reported questionnaire was not complemented by physiological biomarkers, such as salivary or serum cortisol. Consequently, the findings of this study should be interpreted as associations with perceived stress rather than with physiological stress responses.

Beyond the limitations of the evaluation approach, the cross-sectional design and the absence of longitudinal antepartum assessments further limit the interpretation of the temporal relationship between perceived stress and oral microbiota composition. Because both stress and microbiota were assessed at a single peripartum time point, it is not possible to determine whether the observed microbial patterns represent stable differences established before or during pregnancy or transient changes associated with the acute peripartum period. Longitudinal studies incorporating assessments from preconception, throughout early and late pregnancy, and into the postpartum period are therefore needed to clarify the temporal dynamics of these associations.

Moreover, although dietary information, including fermented food consumption, was collected, diet was not comprehensively controlled as a potential confounding factor in the microbiome analyses. In addition, specific exclusion criteria or predefined washout periods for probiotic and prebiotic supplementation were not applied; therefore, their potential influence on oral microbiota composition cannot be excluded. A detailed assessment of the relationship between dietary patterns and oral microbiota was beyond the scope of the present study.

Oral health status was not systematically assessed or recorded as part of the study protocol. Although dental care is included as part of routine prenatal care at the primary healthcare level, individual information regarding periodontal disease, dental caries, recent dental procedures, and oral hygiene practices was not available for all participants and therefore could not be evaluated or controlled for in the present analyses. Given their potential influence on oral microbiota composition, these factors should be considered when interpreting the observed associations between perceived stress and the oral microbiota.

Finally, an additional limitation relates to statistical power. A post hoc power analysis based on the comparison between the low- and very high-perceived stress groups revealed large effect sizes for both Shannon (Cohen's d = 1.00) and Faith's phylogenetic diversity (Cohen's d = 0.98), but moderate statistical power (0.63 and 0.61, respectively). This suggests a limited ability to detect differences with high confidence, particularly due to the small sample size of the very high-perceived stress group. Nevertheless, the consistency of the observed patterns across diversity metrics supports the potential biological relevance of these findings, which should be confirmed in studies with larger sample sizes.

6. Conclusion

Together, these findings emphasize the need to move beyond taxonomic microbiota descriptions toward integrative models that consider the onset of triggers, functional redundancy, immune and hormonal modulation, and host–microbiota interactions in the context of stress and pregnancy. Our data suggest that physiological drivers associated with pregnancy may contribute to shifts in microbial evenness while also being linked to maternal stress responses. We also highlight the need to address what occurs after the driver returns to physiological balance. Therefore, omics approaches and longitudinal follow-ups are needed to make concluding remarks about how microbial patterns and stress are dynamically reshaped.

CRediT authorship contribution statement

Gabriela Vasco: Writing – review & editing, Writing – original draft, Validation, Supervision. Juan Jácome-Navarrete: Writing – original draft, Software, Methodology, Investigation, Formal analysis, Data curation, Writing – review & editing. Vinicio Ponce: Writing – original draft, Methodology, Data curation. Andrés Mateus: Writing – original draft, Investigation, Data curation, Writing – review & editing. Carmen Salvador Pinos: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization.

Declaration of generative AI and AI-assisted technologies

During the preparation of this work, the authors used ChatGPT (OpenAI) and Curie's AI-assisted editing features to improve the readability and clarity of selected sections of the manuscript. The content and scientific ideas were originally developed by the authors. After using these tools, the manuscript was carefully reviewed and edited, and full responsibility for the content of the published article is assumed by the authors.

Funding

This work was supported by the Dirección de Investigación at the Universidad Central del Ecuador [Senior 2022 project: “Metagenómica de la microbiota materna y del neonato durante el año 2022, en el Hospital Docente de Calderón y establecimiento de banco de cepas”].

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Glossary

Abundance: represents the number of individuals from one species in a community.

Alpha diversity: refers to the diversity of the species in each community. It takes account richness, abundance and evenness.

ASVs: Amplicon sequence variants are designed for unique nucleotide variations of given DNA sequences obtained via polymerase chain reaction. They are used to characterize microbial diversity.

Beta diversity: Beta diversity compares the diversity of species between different communities.

Communities: represent an ecological niche. In the context of this manuscript, each patient represents a community.

Evenness: or the uniformity or how the individuals of each species are represented in comparison to those of other species. It measures whether a community has a dominant species or if different species are uniformly represented.

Faith index: Faith index is an alpha diversity index. It sums the length of the branches from a phylogenetic tree of species represented in a community. A higher index represents a more diverse community.

Richness: the total number of species in a community.

Shannon index H: represents the diversity of species in a community, taking account the richness of species and the uniformity of how the individuals of each species are represented. It is an alpha diversity index. A higher index represents a more diverse community.

Weighed UniFrac: is a beta diversity index that uses phylogenetic distances and abundance between different species from different communities.

Appendix A. Supplementary data

The following is the Supplementary data to this article: