The survival and cecal microbiota profile of broilers subjected to acute heat stress are contingent upon their body weight

Abstract

Heat stress (HS) is a growing challenge in poultry farming, driven by rising temperatures, which alters the gut and its microbiota. It is also assumed that body weight might be strongly correlated with a bird’s ability to tolerate HS. However, there is no clear evidence of the potential interaction between HS exposure, body weight, and their effects on the host’s microbiota in the short term. Hence, this study evaluated the effects of acute HS on the cecal bacterial community in chickens, relative to host body weight, and explored potential links between the microbiota and HS resistance. A total of ninety (n = 6 replicates of 5 birds) 36-day-old birds were grouped by body weight class (BWC): Low (2373.9 ± 32.1 g), Med (2681.1 ± 11.2 g), and High (3009.5 ± 36.0 g), and were subsequently either exposed to acute HS (35 ± 1 °C, 8 hours) or kept under thermoneutral conditions (CON). At the end of the trial, their cecal microbiota were subjected to metataxonomic analysis. Results revealed that, compared with their High-BWC counterparts (median survival time: 300 min), birds in the Low-BWC group exhibited greater resistance to HS (median survival time: 359 min). Metataxonomic analyses showed that acute HS increased evenness indices in the HS_Low group compared with the CON_Low group. The HS_Low group also demonstrated significantly higher evenness than the HS_High group. Conversely, beta-diversity analyses revealed no significant effects of HS or BWC. Bacterial communities were dominated by Bacillota and Bacteroidota phyla, with Bacteroides fragilis, Blautia hominis, and Enterococcus cecorum as the most abundant species. More microbial taxa were enriched in the HS group than in the CON groups. Identified biomarkers included twelve for the HS_Low, five for the HS_Med, and three for the HS_High, with Mediterraneibacter glycyrrhizinilyticus, Pseudoflavonifractor capillosus, and Clostridium phoceensis being the most significant for each group, respectively. Relative abundances of Erysipelotrichia, Erysipelotrichales, and Erysipelotrichaceae exhibited a linear decrease with the ratio between initial and final BW. Microbial co-occurrence networks were resilient to acute HS, with no structural differences between the CON and HS groups. Overall, these findings suggest that the BWC may serve as a reliable predictor of acute HS survival in broiler chickens. However, the greater HS resistance observed in lighter birds may be associated with rapid changes in cecal microbiota evenness during exposure. Besides, Erysipelotrichaceae may serve as a microbial indicator of HS resistance in broilers, warranting further investigation.

Introduction

The adverse effects of heat stress (HS) on the physiology, production, and metabolism of chickens cannot be overstated. HS occurs when an organism's cooling mechanisms are insufficient to cope with elevated ambient temperatures (Ncho et al., 2023). In chickens, HS is among the most prevalent abiotic stressors, as temperature plays a crucial role in their development from the embryonic stage to maturity (Ncho et al., 2025a). Although chickens exhibit behavioral responses such as panting, increased water intake, and reduced locomotor activity to mitigate HS, sudden changes and extremely high temperatures still pose a threat to their homeostasis (Lara and Rostagno, 2013). Furthermore, as their production performance increases, it correlates with an increase in their metabolic heat, which significantly elevates their thermoneutral heat production (Syafwan et al., 2011). This is particularly true for broilers, which have been selected for high meat yields and can often increase their hatching weight sixty-fold within 40 days. This impressive growth curve necessitates a similar amount of feed intake, adding to the total basal heat production, as digestion and nutrient utilization in the gut are exothermic processes (Syafwan et al., 2011). Earlier studies have reported positive correlations between body weight (BW), feed intake, feed conversion ratio, and rectal temperature of broilers exposed to HS (Cooper and Washburn, 1998). It should therefore be noted that the genetic predisposition of modern broiler lines, primarily due to their rapid growth, accounts for their lower heat tolerance.

In poultry, three HS models have been extensively studied based on their temporal characteristics: acute, cyclic, and chronic HS. A single, short-term high-temperature challenge lasting from a few hours to a maximum of 24 hours per day is defined as acute HS (Gupta et al., 2024; Quinteiro-Filho et al., 2012). Cyclic HS, on the other hand, is characterised by repeated daily temperature variations, with a period of thermal stress usually followed by a recovery period. This challenge usually spans multiple days or even weeks, mimicking the natural seasonal variation (Gupta et al., 2026; Yuan et al., 2023). Finally, chronic HS involves continuous exposure to high temperatures over an extended period without any recovery period (Azad et al., 2010). Different exposure periods and variations elicit different physiological consequences. While cyclic HS might instil acclimatisation, chronic HS may lead to sustained metabolic suppression and compensatory physiological adaptation (Lara and Rostagno, 2013). Acute HS causes urgent hypothalamic-piyuitary-adrenal (HPA) axis activation, leading to rapid stress hormone (such as catecholamines and corticosterone) surges. This might cause immediate gut barrier disruption and high mortality (Pandupuspitasari et al., 2025).

In chickens, the gastrointestinal tract is one of the primary organs affected by HS. Exposure to acute HS, characterized by either a sudden or gradual increase in ambient temperature beyond the birds' thermoneutral zone for up to 24 hours, is sufficient to induce intestinal injury. Previous research has indicated that exposure to HS for 10 h in 35-day-old broilers resulted in multifocal acute enteritis, characterised by increased counts of immune cells in the gut lamina propria (Quinteiro-Filho et al., 2012). Further histological analyses demonstrated that 10 hours of HS exposure in similarly aged broilers resulted in significant alterations in the villus height-to-crypt depth ratio of jejunal mucosa (Hao et al., 2012). At the molecular level, acute HS was also found to modulate intestinal heat shock protein expression as early as 3 hour post-exposure in 21-day-old broilers (Hasan Siddiqui et al., 2020). The rapid physiological changes in the chicken gut during HS are often associated with leaky gut syndrome, which impairs nutrient absorption and exacerbates inflammation, thereby reducing performance (Goel et al., 2021b).

Recent advancements in "omics" technologies have revealed that the intestinal microbial population in chickens can be estimated at approximately 10¹¹ bacteria per gram of gut content (Apajalahti and Kettunen, 2006). Several factors, including genetic makeup, age, diet, and environmental conditions, are also thought to influence the composition and diversity of the microbiota (Yeoman et al., 2012). Regarding HS, researchers have reported varying findings, ranging from marginal compositional differences to entirely distinct microbial community structures. For instance, Yuan et al. (2023) did not observe differences in cecal microbial alpha and beta diversity between cyclic HS and thermoneutral birds. The authors also noted that cyclic HS was associated with higher abundances of Enterococcus, Weissella, and Reyranella and a reduced presence of Romboutsia and Ruminiclostridium. Conversely, Liu et al. (2020) reported that cyclic HS affected both alpha and beta diversity, with approximately eight genera identified as microbial biomarkers of HS. Therefore, not only is there no conclusive evidence on how HS may shape the intestinal microbiota of chickens, but there is also a scarcity of studies that apply acute HS. As previously explained, a few hours of HS exposure can lead to drastic alterations in gut structure (Hao et al., 2012; Quinteiro-Filho et al., 2012). It can be conjectured that the gut microbial population might be similarly affected. Hence, it is essential to provide an overview of how intestinal microbiota dynamics evolve during the early stages of HS.

Overall, research on the dynamics of chicken gut microbiota in response to elevated ambient temperatures remains an active area of investigation. It is crucial to identify potential microbial biomarkers associated with the acquisition of heat tolerance. In this study, we aimed to elucidate the effects of acute HS on the cecal microbiota of broiler chickens. Additionally, we examined the relationship between different body weight class (BWC) and heat tolerance.

Methods

This study was conducted in the animal research facility of Gyeongsang National University, Korea. The experimental procedures were approved by the Institutional Animal Care and Use Committee of Gyeongsang National University (GNU-250515-C0100).

Experimental design and birds’ management

Day-old mixed-sex Arbor Acres broiler chicks (n = 90) were procured from a Harim Hatchery, Iksan, South Korea. The chicks were housed in battery cages at a brooding temperature of 36 ± 1 °C and a relative humidity (RH) of 50 ± 5%. The temperature was systematically decreased by 1 °C every other day until it reached 21 °C ± 1 °C, 50 ± 5% RH by day 28 of rearing. The birds were maintained at this temperature until day 35, with a lighting schedule of 23 hours of light and 1 hour of darkness. On day 36, each bird was individually tagged, weighed, and ranked in ascending order based on body weight (BW), from the lightest to the heaviest. The birds were then categorized into three body BWC using Microsoft Excel software (Microsoft Corp, Washington, USA): Low (2373.9 ± 32.1 g), Medium (2681.1 ± 11.2 g), and High (3009.5 ± 36.0 g), with each class comprising 30 birds. These birds were further allocated to two rooms, where they were either maintained at thermoneutral conditions (CON: 21 ± 1 °C) or exposed to a high ambient temperature (HS: 35 ± 1 °C). Humidity in both rooms was controlled at 50 ± 5%. Each room contained 15 birds from each BWC, divided into 3 replicates of 5 birds each. A single acute HS challenge was followed on day 36 of age in the HS room for 8 hours. The HS protocol was implemented from 9 am to 4 pm (Supplementary Fig. 1). The temperature was increased at 3.5 °C per hour to reach 35 °C within 4 hours, maintained for 3 hours, and then reduced back to 21 °C over the next hour. Birds in the CON room were maintained at 21 ± 1 °C throughout this period. Feed and water were provided ad libitum throughout the experimental period. Commercial broiler finisher feed (Nonghyup Feed, Gyeongju, South Korea) was administered during the experiment (Supplementary Table 1). During the HS period, birds were closely monitored for observable behavioral changes such as rapid open-beak breathing, wing spreading, and panting. Mortality was recorded throughout the trial.

Experimental variables measurement and sample collection

The initial body weight (BWi) of each bird was documented before the HS exposure, while the final body weight (BWf) was recorded either after the HS period or upon the bird's death, whichever occurred first. The BW ratio was determined as the ratio between BWi and BWf (Supplementary Table 2). Throughout the experimental period, the birds were closely monitored during the HS period. Mortality was confirmed with the complete cessation of movement and respiration. The time of death for each bird was recorded immediately upon observation of mortality. The survival time, measured in minutes, was calculated as the time the bird remained alive from the onset of the HS challenge. Rectal temperature (RT) was recorded from six birds per the body weight class (BWC), with two birds per cage, before euthanasia. For HS-exposed birds, the first 6 birds to die in each BWC were selected from their respective cages at the time of death for RT measurement. A digital thermometer (Hanna Instruments Inc., Padova, Italy) was inserted 3 cm into the cloaca, and the temperature was recorded once the readings stabilized, typically within 30 to 45 s. Before sample collection, the birds were euthanized using a carbon dioxide chamber, and the cecum was excised. However, for the birds under HS, after recording the RT, the carcass was immediately taken to the postmortem room to collect cecal samples. The total time taken from confirmation of death to collection of ceca was approximately 10 min per carcass. The excised ceca were tied with sterile thread and transferred to labeled 25 mL Eppendorf tubes (#0030118405, SnapTec Cap, 25 mL, Eppendorf, Hamburg, Germany). These tubes were immediately placed in liquid nitrogen to preserve the cecal samples for subsequent analyses.

16S rRNA gene amplicon sequencing

A total of thirty-six cecal samples (6 per group) were utilized for 16S rRNA gene amplicon sequencing. The laboratory procedures associated with the metagenomic sequencing analysis of the cecal microbiota samples were outsourced to Macrogen, Inc. (Seoul, Korea). In brief, genomic DNA from microbial communities in broiler cecal contents was extracted using the DNeasy PowerSoil Kit (Qiagen, Hilden, Germany), adhering to the manufacturer's standardized protocol. The extracted DNA was amplified with specific bacterial primers, 341F (5′-CCTACGGGNBGCASCAG-3′) and 805R (5′-GACTACNVGGGTATCTAATCC-3′), which target the V3-V4 variable region of the 16S rRNA gene. Subsequent paired-end sequencing analyses were performed using the Illumina Hiseq PE250 sequencing platform (Illumina, Inc., California, USA). After sequencing, Cutadapt (v3.2) was utilized to remove primer sequences from the raw data. To generate Amplicon Sequence Variants (ASV) sequences, the reads were error-corrected and denoised with DADA2 (v1.18.0). Sequences with an expected error of 2 or more were excluded. Erroneous reads were denoised based on an established error model. Chimeric sequences were eliminated using the consensus method with the removeBimeraDenovo function in DADA2. ASVs with lengths shorter than 1000 bp or longer than 1600 bp were filtered out using R (v4.0.3). Each obtained ASV was aligned to the organism with the highest similarity in the corresponding Reference Database (NCBI_16S), using algorithms such as BLAST+ (v2.9.0, Query coverage > 85% and identity > 85%) and the Bayesian classifier (DADA2_v1.18.0, confidence value: 50).

Ecological and statistical analyses

All subsequent analyses and illustrations were conducted utilizing the R programming language (version 4.4.3) through the R Studio integrated development environment (version 2024.12.1). Data on BW ratio and rectal temperature were initially evaluated for normality of distribution and homoscedasticity before analysis via a two-way ANOVA, with temperature and BWC as main factors. Kaplan-Meier survival curves for BWC exposed to HS were analysed using a log-rank test. Subsequently, a Cox proportional hazards model was employed to estimate and compare the hazard ratios (HRs) of the same groups (Ncho et al., 2022).

Microbiota analyses were conducted following established procedures. Alpha diversity metrics, including Observed ASVs, Chao1, Shannon, and Inverse Simpson indices, were calculated and subsequently analyzed using the Scheirer-Ray-Hare test, with temperature and BWC as the main factors (Goel et al., 2023). When significance was detected, multiple post hoc comparisons were performed using pairwise Wilcoxon rank-sum tests. Beta diversity metrics, such as Bray-Curtis, unweighted, and weighted UniFrac distances, were computed and visualized through principal coordinate analysis (PCoA). A two-way permutational multivariate analysis of variance (PERMANOVA) was used to evaluate potential differences in beta diversity, with temperature and the BWC as main factors. Additionally, permutational multivariate analysis of dispersion (PERMDISP) was utilized to assess the homogeneity of dispersion across treatment groups. The taxonomic composition and distribution of the cecal samples were visualized using stacked bar plots. Unique and shared ASVs were depicted using Venn diagrams, with either temperature or BWC as the main factors. To identify potential microbial biomarkers associated with each treatment group, linear discriminant analysis (LDA) effect size (LEfSe) was applied (| LDA effect size | > 2.0). Furthermore, variables such as BWi, RT, and BW ratio were used as regressors to detect linear associations with the relative abundance of microbial taxa at the phylum-to-species level, employing total sum scaling (TSS) log2 linear regression (Ncho et al., 2025b). Finally, microbial co-occurrence network analyses (|r| > 0.6, minimum prevalence = 10%) were conducted to visualize the effects of temperature on the interactions between microbial taxa in birds maintained under thermoneutral conditions or exposed to HS.

All statistical tests were considered significant at a threshold of p < 0.05 unless specified otherwise. All software packages utilized for the analyses were updated to their latest versions as of April 2025. The two-way ANOVA was conducted using the base R functions. Survival analysis was performed employing the "survival" and "survminer" packages (Kassambara, 2018). Diversity and compositional analyses were executed using the "phyloseq" (McMurdie and Holmes, 2013), "vegan" (Dixon, 2003), "ape" (Paradis and Schliep, 2018), "pairwiseAdonis" (Martinez Arbizu, 2020), and "VenDiagram" (Chen and Boutros, 2011) packages. The LEfSe, TSS log2 linear regression, and stacked bar analyses were performed using the "Microviz" package (Barnett et al., 2021). Microbial co-occurrence networks were constructed using the "igraph" package (Csardi, 2013).

Results

Heat stress resilience: the ability of broiler chickens to withstand heat stress is determined by their body weight class

Fig. 1 illustrates the effects of HS and BWC on physiological parameters associated with heat tolerance. Survival analysis revealed a significant difference in HS resistance among broilers with different BWC. Specifically, birds with low BWCs had a lower mortality rate during the HS trial. The Cox proportional hazards model estimated HRs and their confidence intervals (CIs) to be significantly reduced in the low BWC (HR: 0.32, 95% CI: 0.14–0.72, p < 0.01) compared to the high BWC (Fig. 1A). No significant differences were observed when comparing the medium BWC to the high BWC. Similarly, median survival times, as estimated by Kaplan-Meier curves, were significantly higher (p = 0.021) in the low BWC (median: 359 min, 95% CI: 325–384) compared to the high BWC (median: 300 min, 95% CI: 286–327), while the medium BWC (median: 338 min, 95% CI: 295–375) did not differ significantly from either (Fig. 1B). Regarding the BW ratio (Fig. 1C), there was a significant effect of temperature (p < 0.01) and an interaction between BWC and temperature (p = 0.036). Overall, birds in the CON group exhibited an increase in body weight post-trial, whereas the opposite trend was observed in the HS group. Notably, the CON_High group experienced the greatest increase (1.05 ± 0.02) in body weight, while the HS_High group showed the most pronounced decrease (0.97 ± 0.01). Rectal temperature values (Fig. 1D) followed a similar pattern, with a significant effect of temperature (p < 0.01) observed. Birds in the HS groups (45.7 ± 0.46 °C) had higher rectal temperatures than their CON counterparts (41.0 ± 0.44 °C).

Fig. 1.

Physiological responses of birds across various body weight class (BWC) under conditions of heat stress (HS) or thermoneutral temperature: (A) Forest plot illustrating the probability of mortality in birds subjected to HS stratified by BWC; (B) Kaplan-Meier survival curves for birds exposed to HS, categorized by BWC; (C) The ratio of initial to final body weight in birds following HS exposure; (D) Rectal temperature measurements of birds in both the thermoneutral and HS groups. The experimental treatments were categorized as: CON: thermoneutral temperature; HS: acute heat stress (35 °C for 8 hours at 36 days of age); Low/Med/High: Corresponds to specific body weight class as low, medium, and high, respectively.

Microbiota diversity: short-term acute heat stress primarily influences the evenness of the cecal microbiota in broiler chickens

The impact of temperature and BWC on alpha diversity metrics is depicted in Fig. 2A. Temperature did not significantly alter either richness or evenness indices (Supplementary Fig. 2). The same was observed for BWC (Supplementary Fig. 2). Significant interactions (p < 0.05) between temperature and BWC were observed in the Shannon, Pielou’s evenness, and Inverse Simpson indices, necessitating further multiple comparisons. Specifically, the lowest values were observed in the HS_High group, resulting in a significant reduction (p < 0.05) in these indices compared with the HS_Low group. Additionally, these indices showed significant differences between the CON_Low and HS_Low groups, with the latter consistently showing higher values (p < 0.01). Beta diversity analyses revealed a similar pattern, with no differences between clusters of samples concerning either temperature or body weight class as the main factor (Supplementary Fig. 3), regardless of the distance matrix employed (Bray-Curtis, weighted and unweighted UniFrac). However, only when considering the interaction between temperature and body weight class, and using the Bray-Curtis distances, were there significant differences among groups (PERMANOVA: p = 0.022, R = 0.19, Fig. 2E). Furthermore, regardless of the main factors or interaction, there was no evidence (PERMDISP: p > 0.05) of heterogeneity of dispersion among treatment groups.

Fig. 2.

Diversity analyses of the cecal microbiota of the birds used in the experimental trial: (A) alpha diversity indexes of cecal bacterial communities; (B) Beta diversity analyses were visualized using principal component analysis (PCoA) based on Bray-Curtis distances; (C) Beta diversity analyses were visualized using PCoA based on unweighted UniFrac distances; (D) Beta diversity analyses were visualized using PCoA based on weighted UniFrac distances. The experimental treatments were categorized as: CON: thermoneutral temperature; HS: acute heat stress (35 °C for 8 hours at 36 days of age); Low/Med/High: Corresponds to specific body weight class as low, medium, and high, respectively.

Microbiota composition: the cecal microbiota composition in broiler chickens is more susceptible to alterations due to heat stress than to variations in the host's body weight class

The microbial composition of the cecal samples from the birds was depicted using stacked bar charts (Fig. 3). At the phylum level across treatment groups, the cecal composition exhibited minimal variation, with most samples being predominantly composed (above 95%) of Bacillota and Bacteroidota (Fig. 3A). Although these two phyla were the most prevalent, a total of five phyla were identified, with the less abundant being Pseudomonadota, Actinomycetota, and Candidatus Melanaibacteria. At the genus level, the top ten genera constituted approximately 60% of the total bacterial population, namely Bacteroides, Blautia, Faecalibacterium, Lactobacillus, Enterococcus, Mediterraneibacter, Ruminococcus, Romboutsia, Anaerostipes, and Guopingia (Fig. 3B). Similarly, at the species level, the top ten taxa accounted for approximately 50% of the total composition of the cecal microbiota. Ranked by relative abundance, species such as Bacteroides fragilis, Blautia hominis, Enterococcus cecorum, and Faecalibacterium gallinarum were particularly prevalent across all samples (Fig. 3C and D).

Fig. 3.

Cecal bacterial community composition of the birds used in the experimental trial: (A) Bacterial composition across treatment groups at the phylum level; (B) Bacterial composition across treatment groups at the genus level; (C) Bacterial composition of each sample at the species level; (D) Bacterial composition across treatment groups at the species level. The experimental treatments were categorized as: CON: thermoneutral temperature; HS: acute heat stress (35 °C for 8 hours at 36 days of age); Low/Med/High: Corresponds to specific body weight class as low, medium, and high, respectively.

Fig. 4 illustrates the microbial taxa enriched and unique across treatments. Lefse analysis (Fig. 4A) indicates that more microbial taxa were enriched in the HS groups than in the CON groups. Specifically, twelve, five, and three microbial taxa were identified as biomarkers for the HS_Low, HS_Med, and HS_High groups, respectively. While species and genera were predominantly detected as biomarkers, higher taxonomic ranks, such as the class Clostridia and the family Erysipelotrichaceae, were significantly enriched in the HS_Low group. At the genus level, Mediterraneibacter, Lachnoclostridium, Lacrimispora, and Merdibacter were associated with the HS_Low group. Conversely, Intestimona and Pseudoflavonifractor were enriched in the HS_Med group, and only Evtepia was associated with the HS_High group. The trend at the species level was similar, with the number of species decreasing progressively from the HS_Low to the HS_High group. Specifically, five species (Mediterraneibacter glycyrrhizinilyticus, Lachnoclostridum urinimassiliense, Massilimicrobiota timonensis, Merdibacter massiliensis, and Intestimonas butyriciproducens) were associated with the HS_Low group, two species (Pseudoflavonifractor capillosus and Oscillibacter massiliensis) with the HS_Med group, and two species (Clostridium phoceensis and Evtpia gabavorous) with the HS_High group. In the CON groups, four taxa (Neglectibacter, Neglectibacter timonensis, Blautia shinkii, and Lacrimispora celerecrescens) were identified as biomarkers of the CON_med group, while Limosilactobacillus urinaemulieris and Faecalibacterium hattorii were identified as biomarkers of the CON_Low and CON_High groups, respectively.

Fig. 4.

Differential and shared features of the cecal microbiota of the birds used in the experimental trial: (A) Treatment-specific microbial biomarkers were identified using linear discriminant analysis effect size (LEfSe); (B) Venn diagrams show the unique and overlapping species between each temperature group; (C) Venn diagrams show the unique and overlapping species across body weight class (BWC). The experimental treatments were categorized as: CON: thermoneutral temperature; HS: acute heat stress (35 °C for 8 hours at 36 days of age); Low/Med/High: Corresponds to specific body weight class as low, medium, and high, respectively.

Fig. 4B and C present a Venn diagram depicting the number of bacterial species identified across all groups. Notably, more species were uniquely identified in the CON samples (16) than in the HS samples (7). Furthermore, among the BWC, the Medium and High groups exhibited only three and four unique species, respectively, whereas the Low group exhibited eight unique species. Additionally, linear associations between microbial taxa, BWi, BW ratio, and RT were evaluated. No significant relationships between microbial populations were detected, except with the BW ratio. Consequently, Fig. 5A illustrates these specific linear relationships. A significant decrease in the populations of Erysipelotrichia (Fig. 5B), Erysipelotrichaciales (Fig. 5C), and Erysipelotrichaceae (Fig. 5D) was observed as the BW ratio increased.

Fig. 5.

Relationship between bacterial relative abundance and physiological parameters measured during the trial: (A) Tree displaying associations between taxa at all taxonomic levels and the ratio between the ratio of initial to final body weight (BW ratio) in birds following the HS challenge; (B) Scatter plot of the linear relationship between Erysipelotrichia and the BW ratio; (C) Scatter plot of the linear relationship between Erysipelotrichaciales and the BW ratio; (D) Scatter plot of the linear relationship between Erysipelotrichaceae and the BW ratio. The birds were subjected to a temperature of 35 °C for 8 hours at 36 days of age.

Microbial co-occurrence networks: interactions among microbial taxa in the cecum of broiler chickens demonstrate resilience to short-term acute heat stress

Microbial co-occurrence network analyses were conducted to evaluate the interaction dynamics among microbial species in the cecum of birds subjected to either thermoneutral or HS conditions (Fig. 6). Consequently, two distinct networks were constructed using input data sets from the HS (Fig. 6A) and control (CON) groups (Fig. 6B) . A common prevalence threshold of 10% and a minimum correlation threshold of |r| > 0.6, with p < 0.05, were established to facilitate comparison between the networks. Overall, both networks showed similar numbers of vertices (115 for HS versus 112 for CON) and edges (157 for HS and 153 for CON). Additionally, the number of negative correlations between species was comparable, with the HS network displaying 36 negative edges and the CON network 40. Other key network metrics, such as edge density (0.024 for both HS and CON), average path length (4.0 for HS and 4.4 for CON), and transitivity (0.21 for HS and 0.19 for CON), followed a similar trend. The primary difference was observed in the identification of keystone species: Bacillus pseudomycoides in the HS network and Acetivibrio saccicola in the CON network. Furthermore, only one unclassified species present in the HS group was absent in the CON group, with other phyla, except Bacillota, exhibiting the same number of microbes in both networks.

Fig. 6.

Microbial co-occurrence network of broilers’ cecal microbiota at the phylum level: (A) Network structure of samples belonging to birds exposed to HS; (B) Network structure of samples belonging to birds kept under thermoneutral conditions. The birds were subjected to a temperature of 35 °C for 8 hours at 36 days of age.

Discussion

A growing body of evidence is evaluating the potential relationship between HS exposure and the gut microbiota of chickens (Lyte et al., 2025; Sayed et al., 2025). Most studies have employed a cyclic HS approach during their trials (Xing et al., 2019; Zhou et al., 2022). While cyclic HS aims to replicate daily temperature variations over an extended period, heat waves can often occur suddenly. However, there remains a paucity of studies focusing on the association between acute HS and gut microbiota. The present study investigates how rapid changes in ambient temperature during acute HS may influence microbiota dynamics in the avian gut. Additionally, it seeks to elucidate the relationships between these changes and heat tolerance.

In this study, it was hypothesized that BW may affect the responses of broilers exposed to HS. Consequently, a survival analysis was conducted. As anticipated, birds with lower BWC exhibited a reduced likelihood of succumbing to HS during the trial. It is important to emphasize that, at the same age, a weight difference of approximately 400 g between birds resulted in more than a 50% reduction in the risk of death during HS exposure. Specifically, there was a 68% reduction in the risk of death between the HS_Low and HS_High groups. To our knowledge, this is one of the few studies to quantify the potential risk of death during HS in relation to body weight variation, an important indicator of flock heterogeneity. Modern broiler lines are characterized by their high metabolic rate, which is correlated with increased feed intake (Syafwan et al., 2011). Quentin et al. (2004) demonstrated that, regardless of feed form, fast-growing birds consistently exhibited greater feed intake compared to their slow-growing counterparts, which were not heavily subjected to genetic selection. Additionally, within flocks of similar age, heavier birds tend to be those with the highest feed intake. Recent advancements in precision feeding have confirmed this observation. For instance, Zuidhof et al. (2017) demonstrated that controlling broiler breeders' intake with a precision feeding station led to greater flock uniformity, with body weight variation of less than 2% among individuals. High feed intake correlates with increased basal heat production due to not only elevated maintenance metabolism but also increased energy use associated with digestion and nutrient utilization (MacLeod et al., 1993). Consequently, during HS, an already elevated basal heat production may challenge heat dissipation, primarily through posture adjustments and increased respiratory rate. In fact, panting, which can account for up to 80% of total evaporative heat loss (Ahmad and Sarwar, 2006), may be hindered in heavier birds as they are more susceptible to respiratory and cardiovascular anomalies (Nawab et al., 2018). While the current trial confirmed the assumption that BWC can be a reliable predictor of HS tolerance, the observations made on microbiota dynamics were unexpected.

Microbiota diversity is a primary outcome assessed in microbiome studies, serving as a robust indicator of the number of microbial species inhabiting the host, their distribution, and the diversity of their genetic composition (Lozupone et al., 2012). In the present study, the alpha diversity of the cecal microbial population in each group was evaluated. Notably, HS did not significantly affect richness; however, evenness indices varied significantly among groups, with distinct patterns. Specifically, only the birds in the CON_Low group showed significantly lower evenness indices than those in the HS_Low group. Furthermore, HS_Low maintained higher evenness during HS than the HS_High group. This observation is directionally consistent with the superior survival outcomes of low BWC. Previous studies have reported an association between increased evenness-related indices and HS. Emami et al. (2022) provided evidence of a higher Shannon index in the ileal luminal content of chickens exposed to cyclic HS, irrespective of their lines. Similarly, Liu et al. (2023b) found a significantly higher Shannon index in heat-stressed birds following cyclic HS. Supporting this observation, a trend toward a higher Shannon index in acutely heat-stressed versus thermoneutral broilers (p = 0.071) was reported previously . Microbial diversity evenness quantifies the uniformity of microbial species distribution within a sample (Johnson and and Burnet, 2016). It is hypothesized that microbial communities with higher evenness are more resilient to invasions and exhibit greater stability (Philippot et al., 2021). For instance, increased evenness under environmental stress, such as drought, was characteristic of more resilient fungal communities, facilitating their recovery post-rewetting (de Vries et al., 2018). Given that the CON_Low group did not exhibit higher evenness compared to the CON_High group, it can be inferred that the higher evenness observed in the HS_Low group is an adaptive response to HS. This may imply that the HS_High group either did not survive long enough to exhibit these cecal microbial characteristics or that the lower evenness observed may, in turn, contribute to their reduced survival. Higher evenness is often associated with community resilience and resistance to external stressors. This concept in the microbiome literature has been referred to as the insurance hypothesis of biodiversity, which posits that a more evenly distributed set of taxa buffers the community against environmental perturbations (Wongkuna et al., 2024). However, another interesting and straightforward explanation for the superior survivability of the low BWC groups could be a more favorable surface area-to-body mass ratio for heat dissipation (Liang et al., 2026). Moreover, HS is known to compromise gut integrity and structure in chickens, promoting pathogen proliferation (Cao et al., 2021). As microbiota richness was not significantly affected, a more uniform distribution of commensal bacterial populations may serve as an initial defense against potential heat-induced pathogen colonization (Lan et al., 2005). Nevertheless, these interpretations are speculative hypotheses because the current study’s dataset does not include pathogen abundance data, gut barrier function, and host immune parameters.

Although alpha-diversity analyses may support the hypothesis that BWC correlates with HS resistance, beta-diversity results did not align. Specifically, there was no evidence of group-related clustering when HS and BWC were treated as primary factors or their interaction. Furthermore, the utilization of the three distinct distance metrics, namely Bray-Curtis, unweighted UniFrac, and weighted UniFrac, did not appear to influence the results. This observation suggests, first, that microbial community structures across groups did not exhibit phylogenetic variation, as indicated by UniFrac distances. Secondly, it implies that changes in relative abundances, as referenced by the Bray-Curtis distance, were not substantial enough to impact beta-diversity. This discordance between alpha and beta diversity has been reported previously, with the Shannon index significantly higher in the ileal contents of HS broilers, whereas beta diversity did not clearly separate CON and HS birds (Emami et al., 2022). However, most studies employed cyclic and/or chronic HS models. The results might need further validation and more studies for better interpretation.

An examination of the cecal microbiota composition in the birds utilized in the trial revealed a pronounced predominance of Bacillota and Bacteroidota. These findings align with previous studies that have provided a cross-sectional description of the chicken gut microbiota (Bajagai et al., 2024; Han et al., 2016). It is anticipated that Bacillota (formerly Firmicutes) would comprise 50-90% of the cecal microbiota, particularly at this developmental stage (Yeoman et al., 2012). Furthermore, the Lactobacillus genus is expected to constitute a significant proportion of the Bacillota population, ranking among the top 10 genera in the cecal microbiota. This trend was observed in the current trial and in prior studies on the same subject. Goel et al. (2021a) and Goel et al. (2022) consistently reported Lactobacillus as one of the top 10 genera and a major component of the Firmicute population. Although the overall cecal composition of the birds was typical of broilers at this age, a comparative analysis revealed treatment-related differences. Indeed, environmental conditions exerted a greater influence on microbial composition than BWC. Specifically, HS resulted in the suppression of sixteen microbial species and the emergence of seven new ones. Notably, Clostridium perfringens was identified as one of the species newly present in the cecal microbiota of birds following HS exposure (Supplementary Fig 4). C. perfringens is a Gram-positive bacterium known to produce toxins and enzymes associated with necrotic enteritis (Shojadoost et al., 2012). One of the primary factors known to facilitate the colonization and proliferation of C. perfringens is intestinal damage (Immerseel et al., 2004). As previously mentioned, early consequences of HS in the chicken gut include structural alterations leading to leaky gut syndrome. Disruption of the gut barrier not only compromises immunity but also results in the release of plasma proteins into the intestinal lumen (Brugaletta et al., 2022). This nutrient-rich leakage may serve as a substrate for the growth of various pathogens, including C. perfringens (Immerseel et al., 2004). In addition to treatment-specific taxa, several microbial biomarkers were identified, most of which belonged to the HS groups. A particularly notable example was the Erysipelotrichaceae family. It was not only identified as a statistical biomarker of HS but also exhibited a linear decrease in relative abundance with the BW ratio. It is crucial to note that the BW ratio indirectly reflects the level of stress during the trial. Indeed, birds maintained under thermoneutral conditions all displayed a BW ratio above 1 (indicating an increase in BW), while those exposed to HS had a BW ratio below 1 (indicating a decrease in BW). It can thus be inferred that as the birds experienced increasing levels of stress, their BW ratio correspondingly declined. A higher relative abundance of Erysipelotrichaceae has been associated with inflammation and metabolic disorders (Kaakoush, 2015). There was a significantly higher increase in their population in the tumor group of an animal model of 1,2-dimethylhydrazine-induced colon cancer (Zhu et al., 2014). Another study has also demonstrated a positive correlation between this bacterial family and host cholesterol metabolite levels (Martinez et al., 2009). In poultry, HS is known to induce both inflammation and excessive lipid deposition (Ncho et al., 2025a), which may account for the observations reported in this study. However, it is important to note that most evidence regarding the association between Erysipelotrichaceae and host physiology has been derived primarily from human and rodent models. The potential for inter-host variation cannot be disregarded, and therefore, further research in poultry species is warranted. However, statistical biomarker confirmation of Erysipelotrichaceae was not followed with functional validation, and hence, warrants further work.

The analysis of parameters such as microbial diversity and composition has been extensively studied in trials assessing the potential impact of HS on chicken microbiota (Campos et al., 2023; Goel et al., 2023). However, given that microbial assemblages comprise numerous microorganisms that interact not only with their host but also among themselves, it is crucial to provide a three-way visualization of these relationships (Liu et al., 2023a). To achieve this, we conducted a microbial co-occurrence network analysis, treating HS as a primary factor to assess whether microbial interactions were disrupted in hosts subjected to acute HS. The analysis revealed that both networks were similar, with comparable numbers of microbial species and interactions. In fact, aside from keystone species, other network characteristics were highly similar. The relatively short exposure of birds to HS in this trial may be the primary factor explaining the similarities between networks. Indeed, as previously discussed, compositional changes between the cecum of birds exposed to two environmental conditions were minimal. A longer exposure may have been necessary to result in changes in microbial interactions. For instance, an 18-day HS exposure led to distinct microbial networks in the rumen of dairy cows (Park et al., 2022). These findings were also associated with changes in beta diversity, which were not present in our trial. Hence, these results may suggest that short-term acute HS did not constitute a strong enough disturbance to modulate the cecal microbial networks of the broilers.

While the present study facilitated a comparison of how HS may influence chickens of varying BWC and shape their microbiota, several limitations must be considered when interpreting the findings. Firstly, there was no direct evaluation of the relationships between HS resistance and microbial taxa dynamics. Only half of the subjects in the trial were exposed to HS, and a limited number of microbiota samples were analyzed. Consequently, BWC and other parameters, such as the BW ratio, were used as indicators of potential HS resistance because they correlated with survival times. Survival time is arguably the most effective measure of HS resistance, as it reflects the chickens' ability to withstand elevated ambient temperatures before mortality. Ideally, larger trials that focus on survival time under HS conditions and incorporate more microbiota samples should be conducted. This would enable the identification of plausible microbial biomarkers of HS resistance. Secondly, the BWC was arbitrarily defined and related to the age and body-weight distribution of the birds included in the trial. Therefore, the outcomes may have differed if the trial had been conducted at a different age or involved birds with different growth trajectories. It is also acknowledged that the cecal samples for the broilers under HS were collected postmortem. An attempt was made to collect the samples quickly (within 10 min of confirmation of mortality), a brief postmortem period was inevitable in this study. This might have arguably altered oxygen-sensitive anaerobic taxa in the cecum. The dearth of studies on acute HS (compared with cyclic and/or chronic HS) was another factor limiting interpretation. Finally, only the cecal microbiota was analyzed, and changes may have been more pronounced in other parts of the gut, particularly the small intestine, which is known to be especially sensitive to HS. Future studies could also investigate other physiological stress markers, such as corticosterone, oxidative stress-related enzymes, and heat shock proteins, to provide a complete mechanistic picture. Further line of experimental approach, such as transplanting microbiota from low BWC donors to high BWC recipients, could clearly identify if the different microbiota play a causal role in mediating heat resistance.

Conclusions

In the present study, BWC showed a significant association with the survivability of broiler under acute HS, suggesting that it may serve as a potential non-invase indicator of HS resistance. Furthermore, acute HS was observed to influence the cecal microbiota of chickens, primarily by impacting evenness diversity indices. Several microbial biomarkers associated with HS were identified, with Erysipelotrichaceae being the most significant. Further validation across diverse commercial production environments, broiler strains, ages and HS conditions are necessary to comprehensively elucidate the modulatory effects of HS on the avian gut microbiota.

Ethics declaration

This study was conducted in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. This study was approved by the Institutional Animal Care and Use Committee of Gyeongsang National University. (Approval No. GNU-250515-C01000)

Authors contributions

VG: Made substantial contribution in conception, designing and executing the study. Involved in sample data collection, analysis, original draft preparation and reviewing the manuscript. CMN: Contributed to original draft preparation, data analysis and visualisation, handling the software and in reviewing the manuscript. MAD: Contributed to data and cecal sample acquisition. Observing and recording mortality during heat stress and reviewing the manuscript draft. YJH: Contributed to data and cecal sample acquisition. Observing and recording mortality time along with MAD during heat stress, and reviewing the manuscript draft. WKK: Critically contributed to study design and conception. Gave important feedback for the original manuscript, reviewed and revised the final draft for submission. YHC: Acquired funding for the project, conceptualised, and supervised the entire study, along with a critical review of the original and revised drafts for submission. All the author(s) read and approved the final version of the manuscript.

Funding

This work was supported by the Ministry of Science and ICT through the National Research Foundation of the government of Korea grant (No. RS-2024-00350745).

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Disclosures

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.

Data availability

The datasets supporting the conclusions of this article are included within the article (and its additional files).