Are there consistent effects of gut microbiota composition on performance, productivity and condition in poultry?

Abstract

Microbiome of the gastrointestinal tract (GIT) has been identified as one of the crucial factors influencing the health and condition of domestic animals. The global poultry industry faces the challenge of understanding the complex relationship between gut microbiota composition and performance-related traits in birds. Considerable variation exists in the results of correlational studies using either 16S rRNA profiling or metagenomics to identify bacterial taxa associated with performance, productivity, or condition in poultry (e.g., body weight, growth rate, feeding efficiency, or egg yield). In this review, we survey the existing reports, discuss variation in research approaches, and identify bacterial taxa consistently linked to improved or deteriorated performance across individual poultry-focused studies. Our survey revealed high methodological heterogeneity, which was in contrast with vastly uniform focus of the research mainly on the domestic chicken (Gallus gallus) as a model. We also show that the bacterial taxa most frequently used in manipulative experiments and commercial probiotics intended for use in poultry (e.g., species of Lactobacillus, Bacillus, Enterococcus, or Bifidobacterium) do not overlap with the bacteria consistently correlated with their improved performance (Candidatus Arthromitus, Methanobrevibacter). Our conclusions urge for increased methodological standardization of the veterinary research in this field. We highlight the need to bridge the gap between correlational results and experimental applications in animal science. To better understand causality in the observed relationships, future research should involve a broader range of host species that includes both agricultural and wild models, as well as a broader range of age groups.

INTRODUCTION

Poultry ranks amongst the most widespread livestock: in 2021, worldwide production included 74 billion chickens, 4 billion ducks, and 0.6 billion turkeys (http://www.fao.org/faostat). Although still novel and developing, microbiota research in the past decade has fundamentally changed our views on factors conditioning poultry health and performance in production-related traits (Dehau et al., 2022). On top of traditional condition determinants, microbiota composition, namely in the gastrointestinal tract (GIT), emerged as a key modulator of animal physiology. Whereas poultry microbiota comprises a rich community of microorganisms including bacteria, viruses, fungi, and protists, only bacterial microbiota is presently extensively studied (Berg et al., 2020). Recent research reveals that various symbiotic bacteria affect the fitness of their hosts in different ways (Abdel-Kafy et al., 2022). Although most GIT bacteria are represented by commensals that are neutral to their hosts under normal circumstances, positively acting mutualists as well as detrimental pathogens and opportunistic pathogens are present (Proctor et al., 2019; Berg et al., 2020). The major challenge of the present research in applied veterinary and agricultural microbiology is to determine the direction of interactions of various bacterial species and strains with their hosts. Presumably, due to co-evolutionary arms races between the hosts and pathogens diversifying these interactions (Zilber-Rosenberg and Rosenberg, 2008), the effects of individual bacteria on host condition will vary among animal taxa. As a consequence, different livestock species and breeds can be predicted to interact differently with various microbes, creating heterogeneity of responses to manipulative treatments. This can affect the outcomes of therapeutic interventions and of veterinary probiotics designed to improve condition.

Probiotic treatments in poultry are aimed at improving productivity, but the estimated effects of microbiota on condition or performance depend on their definition adopted for the purposes of research. Although often related, several distinct biological concepts have been proposed to refer to condition (Gosler and Harper, 2000). Condition in a broad sense includes nutritional state, health, and physiological vigor (Schluter and Gustafsson, 1993). In a narrow sense, condition was defined by Gosler and Harper (2000) as a component of body mass independent of body size, which is easy to measure. However, this narrow definition covers only one component of the broad-sense condition, leaving out several relatively important health determinants, such as micronutrient intake or defective metabolite synthesis or trafficking. In agriculture, condition is typically defined as a measure of production-related traits (Roche et al., 2009). The term can be thus synonymized with animal performance, which refers to efficiency in providing the desired benefit, such as growth, reproduction, or egg production. It should be noted that improved productivity-linked performance does not always correlate with improved health. For example, higher body mass can be associated with impaired immunity (Van der Most et al., 2011). Hence, the broad-sense condition and animal performance can be linked negatively. Despite heterogeneous methodological approaches to condition estimates used in poultry, the most commonly measured parameters of performance are body weight, body weight gain over time, feeding efficiency, and egg-laying rate (see Supplementary Table S2 in electronic supplementary material).

Here we review the results of correlational studies that used 16S rRNA profiling or a metagenomic approach in birds to identify bacterial taxa associated with improved or deteriorated performance. Although we adopted the narrow concept of condition, the methodological variation among the reviewed studies is still high and the microbial content can vary dramatically, making true meta-analysis of the data impossible. However, a detailed survey of the available correlational results is essential, as these data should serve as the primary source of information forming the basis for subsequent experimental research. Currently, the selection of bacterial taxa used in experiments and commercial probiotics is highly uniform across a wide range of hosts including poultry, livestock, pet birds and mammals, including even humans (Abd El-Hack et al., 2020; Barreto et al., 2021). Thus, little host specificity is anticipated in the present research, so we expect positive associations of the bacterial species used in probiotics with animal performance to be common to all species. In this review, we question this assumption for non-mammalian species. First, based on a comprehensive survey, we identify GIT bacterial taxa that are consistently linked with good or poor condition across a variety of poultry-oriented studies. Second, we examine whether the bacteria that consistently correlate with body condition in birds overlap with the bacterial taxa most commonly used in commercial poultry probiotics and manipulative research (recently reviewed elsewhere – e.g., Royan, 2018; Yousaf et al., 2022). Our survey aims to bridge the gap between basic correlational veterinary research and experimental animal science applications.

METHODS

A literature survey on correlations of GIT microbiota composition and condition or productivity in poultry was performed through the Web of Science (WOS), using relevant keywords (for query settings, see Table S1). Out of the 941 research articles retrieved in our search for the past 15 yr (2008–2023), when NGS 16S rRNA profiling has been available, only 34 articles represented correlational studies published on the specific topic and were chosen for this comparative review (selection based on a manual check of the information provided in the title and abstract). One study (Davidson et al., 2021) was later excluded because it was the only correlational study focused on non-poultry birds (the great tit, Parus major).

We emphasize that this review has not gained the originally intended form of a meta-analysis, since during our survey we revealed too high variation among the studies in their methods and result representation. Our goal is to highlight this heterogeneity while indicating whether any certain bacteria show consistent effects on condition across multiple studies. Although our literature survey was standardized, using strictly defined criteria, we admit that the final list of studies included may lack full comprehensiveness.

For each of the reviewed 33 studies, we listed the basic description (host species, sex, and age; sampling site; sequencing method; 16S rRNA gene region; bacterial phylogenetic level; condition markers) in Table S2 (summarized in Figure 1). From each study, all bacterial taxa that were reported to significantly correlate with performance were listed (Table S3). The taxonomical classification was added using the ITIS database (https://www.itis.gov/) or Silva database (https://www.arb-silva.de/) for taxa that were not present in ITIS. We excluded mitochondria, chloroplasts, and all bacterial taxa that were absent from both databases. It is important to note that our only source of information about bacterial taxa significantly correlating with performance were the tables or figures provided by the authors in their published research articles. As we have already mentioned, the methodology of these studies was highly heterogeneous and so were the ways of presenting the results. When creating Table S3, we have always listed all taxa reported by the authors to significantly correlate with condition, including those that were sex-, sample site- or condition parameter-specific. Furthermore, authors of the research papers covered in our survey performed their analyses on various taxonomic levels (see Table 1, Table S2 and Table S3). Lacking access to the original data, here we strictly review the results as reported in the publications. Therefore, wherever the higher taxonomic groups were reported as significant (e.g., correlation with condition revealed for a genus), we show the same taxonomic level here (i.e., the result obtained for a merged category of all OTUs belonging to the particular genus). In cases where multiple taxonomic levels were tested in one study, we treated each of them here separately.

Figure 1.

Summary of methodological approaches taken by the target 33 correlational studies focusing on associations between microbiota composition and performance in poultry. (A) host organism and age; (B) tissue sampled with lines connecting cases of multiple-site sampling within studies; (C) DNA fragments selected for sequencing, V = 16S rRNA hypervariable region; (D) sequencing methods.

Table 1.

Bacterial taxa consistently associated with good or poor condition based on results from 3 or more studies, and without any contradictory evidence. G = genus, F = family.

Bacterial taxon	Tax. level	Condition	Phylum	Gram-positive/negative	References
Methanobrevibacter	G	Good	Euryarchaeota		(Han et al., 2016; Yan et al., 2017; Wen et al., 2019)
Candidatus Arthromitus	G	Good	Firmicutes	G+	(Danzeisen et al., 2013; Vollmar et al., 2020; Bindari et al., 2021; Fu et al., 2022)
Ruminococcaceae	F	Good	Firmicutes	G+/G-	(Singh et al., 2012; Stanley et al., 2016; Huang et al., 2021)
Prevotella	G	Poor	Bacteroidetes	G-	(Singh et al., 2014; Han et al., 2016; Hou et al., 2016)
Streptococcus	G	Poor	Firmicutes	G+	(Singh et al., 2014; Han et al., 2016; Vollmar et al., 2020; Fu et al., 2022; Zhang et al., 2022a)
Subdoligranulum	G	Poor	Firmicutes	G+	(Singh et al., 2014; Hou et al., 2016; Vollmar et al., 2020)

Next, we wanted to link the correlational studies with manipulative experiments and commercial poultry probiotic products. We searched for poultry probiotics registered in the Food and Feed Information Portal of European Commission and the Czech Institute for State Control of Veterinary Biologicals and Medicines. All relevant products were listed in a table (Table S4), and their bacterial content was noted from the information reported on the manufacturer's webpage. Additional products were added through a 3-h Google search (term “poultry probiotics”) and from reviews on poultry probiotics (Abd El-Hack et al., 2020; Krysiak et al., 2021). From these reviews, we excluded any products that 1) lacked the manufacturer's official web-shared information on the bacterial content, 2) were designated for primary use in humans or animals other than poultry, and 3) did not contain living bacterial cells. The frequency of use of individual probiotic bacterial species in manipulative studies has been estimated based on a search performed through WOS (for query settings, see Table S1).

RESULTS AND DISCUSSION

Methodological Variation in Poultry Microbiota Studies

Comparison of the approaches taken by the 33 correlational studies reviewed (Figure 1, Table S2) shows that most research has been conducted using domestic chicken (Gallus gallus f. domestica). This limits our understanding of the host-microbiota interactions in birds to a single dominant model species. Despite this relative uniformity, studies significantly differ in the age groups of the experimental animals. Both host species and age can contribute to high variability in the reported results (Liu et al., 2021b). Also, the number of individuals examined per study varied notably (from only 20 to 3,480, from which, however, only tens to hundreds of individuals were chosen for microbiota sampling). Frequently, the research adopted the strategy of comparing 2 distinct groups of animals, one with the lowest and the other one with the highest performance selected for the microbial analysis out of the total number of the experimental individuals forming a continuum (as described in Han et al. 2022). Typically, researchers performed a multivariate analysis aimed at identifying bacterial taxa associated with performance. Yet, other studies often individually tested for partial correlations between the condition-related parameters and abundances of specific bacterial taxa. Nonetheless, this approach requires careful statistical correction for multiple testing within the same dataset, which is sometimes neglected.

Several different traits to measure animal performance were used by the correlational studies surveyed: body weight, body weight gain, feeding efficiency (counted as feed conversion ratio, or residual feed intake), apparent metabolizable energy, egg yield and different measures of fat deposition (Table S2). It should be pointed out that the use of different performance traits can lead to distinct results. For example, bacteria that are linked with fast weight gain, but also contribute to increased feed intake, will be considered beneficial for body weight gain, but likely detrimental to feeding efficiency (e.g., family Burkholderiaceae; Díaz-Sánchez et al. 2019).

Variation between tissues in their microbial composition can be another source of differences among the studies. The correlational studies analysed in this review were based on samples from 7 different compartments of GIT and feces. As known from poultry (Su et al., 2021), but also from other birds (Schmiedová et al., 2023), these sites differ in their microbial composition, and the abundances of specific bacterial taxa can even show opposing correlations with values of condition markers in different GIT compartments (e.g., genus Bifidobacterium in the crop vs. ileum; Han et al. 2016). Variation also exists in the sequencing strategies adopted by the researchers, where the majority of the studies used 16S rRNA profiling, but the precise gene region varied (most frequently the sequences covered variable region V4 or a combination of V3 and V4). Only 6 studies used metagenomic data (Singh et al., 2014; Hou et al., 2016; Du et al., 2020; Jing et al., 2022; Liu et al., 2022; Bai et al., 2023). The most common platforms used for sequencing were Illumina and 454 pyrosequencing, but other approaches were also applied in a few cases. Additional heterogeneity also exists in the filtering steps of microbial composition analysis.

Bacterial Taxa Consistently Associated With Condition-Related Performance

In total, across the 33 reviewed studies, significant correlations with the condition were reported for 300 bacterial taxa (Table S3). However, our survey revealed that in poultry, there are presently only 3 bacterial taxa reported as associated with good condition (family Ruminococcaceae, and genera Methanobrevibacter, and Candidatus Arthromitus) and 3 taxa reported as associated with poor condition (genera Prevotella, Streptococcus, and Subdoligranulum) consistently in multiple (≥3) studies and without any contradictory evidence (Table 1; for a summary of the results from all reviewed publications, see Table S3).

In 4 studies, a significant positive association between condition and the abundance of Candidatus Arthromitus was shown (Table 1). Candidatus Arthromitus is a collective name for segmented filamentous bacteria, but in the old concept, it includes 2 unrelated groups – one belonging to Lachnospiraceae (inhabiting GIT of Arthropods) and the second belonging to Clostridiaceae (inhabiting GIT of vertebrates) (Thompson et al., 2012). Therefore, a new name, Candidatus Savagella, has been proposed for the vertebrate SFBs (Thompson et al., 2012). Yet, the former name still remains in the Silva database and subsequently appears in the results of studies describing poultry microbiome composition. Although no association with poor condition has been found for Candidatus Arthromitus in our survey, there is some evidence of pathogenicity caused by these bacteria in older livestock microbiological literature (Angel et al., 1990; Goodwin et al., 1991). Despite this, recent literature considers SFBs to be commensals with positive effects on condition and a potential for application in probiotics (Danzeisen et al., 2013; Schnupf et al., 2015; Hedblom et al., 2018). Yet, manipulative experiments using SFBs in poultry are rare. Due to their reduced genomes, SFBs are highly dependent on their hosts for many metabolic functions and consequently hard to culture (Danzeisen et al., 2013). They can be efficiently administrated as spores (Redweik et al., 2020). Meinen-Jochum et al. (2023) showed that SFB treatment in chicken can suppress Salmonella infection and cause a significant decrease in total Enterobacteriaceae.

Besides Candidatus Arthromitus, Methanobrevibacter was also repeatedly correlated with good condition in our survey. Species of this archaean genus are the most dominant methanogens in the GIT of various vertebrates and invertebrates (Thomas et al., 2022). Despite the generally limited information on the relationship between host condition and representation of Archea, there is some evidence for the association of Methanobrevibacter with increased body weight in humans. Abundance of M. smithii is increased in obese children (Mbakwa et al., 2015) and decreased in patients with IBD (Ghavami et al., 2018) or severe malnutrition (Camara et al., 2021). That is in agreement with our finding that Methanobrevibacter abundance correlates with better performance in poultry (e.g., associated with increased body weight in Han et al. 2016).

With regard to the family Ruminococcaceae, the total abundance of all operational taxonomic units (OTUs) belonging to this family correlated with good condition in 3 studies and with poor condition in none. However, several species/genera from this family were linked to poor condition in multiple studies that performed the analysis on lower taxonomical levels (Stanley et al., 2012; Hou et al., 2016; Yan et al., 2017). Consistent association with poor condition was found for Subdoligranulum (Singh et al., 2014; Hou et al., 2016; Vollmar et al., 2020) and very variable results were obtained for Faecalibacterium (discussed below). It is evident from this example that the analyses performed on family or higher levels (e.g., phyla in Elokil et al. 2020 or Lyu et al. 2021) should always be interpreted with caution because their results are highly dependent on the relative abundances of individual species within the higher taxa represented in each sample (Liu et al., 2021a).

Apart from Subdoligranulum, two other genera were identified in this review as consistently associated with poor condition – Prevotella, and Streptococcus. Prevotella is considered beneficial for human health in the context of decreasing obesity (Tett et al., 2021), which is in agreement with our results that this bacterium is linked to decreased performance in poultry. Whereas some Streptococcus species are pathogenic for birds (Roy et al., 2013), other species, such as S. salivarius, are present in the healthy gut microbiota of ducks and chickens, and some strains are even used in probiotic products for poultry because they have shown positive effects on host health and production-related traits (Naseem et al., 2012; Zhang et al., 2021).

Bacterial Taxa With Inconsistent Associations to Body Condition Across Correlational Studies

In general, for many bacterial taxa, the correlational research on the relationship to animal performance in poultry provides highly inconsistent and sometimes even contradictory results (Figure 2). Across the 33 correlational studies reviewed, as many as 72% of the significant identifications of bacterial taxa associated with good or poor condition were never independently confirmed by any other study. Furthermore, contradictory effects were reported in different studies for another 20% of the taxa. This can be partially attributed to the biologically relevant factors affecting gut microbiota modulation of the host physiology, but also partially to the distinct methodology of the studies (differences in host species, sex and age, sampling site, or the performance marker tested). Contradictory effects of some bacterial taxa were shown even within a single study. For example, Stanley et al. 2016 showed that the effects of Lactobacillus reuteri were dependent on the choice of performance marker. Lee et al. (2017) reported sex-dependent results. However, it should also be noted that more OTUs with different effects can be assigned to the same higher taxon (genus/species), contributing to the apparent inconsistency of the overall results (e.g., Siegerstetter et al. 2017).

Figure 2.

Numbers of bacterial taxa associated with good or poor condition identified in one or more studies.

In our survey, we found the most contradictory results for effects of the phylum Bacteroidetes (positive in 4 studies, negative in another 3), which was also true for the specific genus Bacteroides (7 positive, 3 negative). Similar ambiguity was detected for some representatives of other phyla, including Enterococcus (4 positive, 4 negative), Lactobacillus (3 positive, 5 negative), Oscillospira (2 positive, 2 negative), Faecalibacterium (5 positive, 6 negative) and Ruminococcus (4 positive, 2 negative); and the species Lactobacillus crispatus (2 positive, 3 negative). In the case of the phylum Bacteroidetes, this apparent inconsistency could be an effect of age because those studies describing positive association with condition were performed in young individuals, but those studies showing negative relationships were done in adult fowl. The effect of Ruminococcus could be sampling site-dependent, as 4 studies reported the association with good condition for several parts of the intestine, and 2 studies reported the link to poor condition where only the feces were sampled.

In our survey, 23 of the 33 reviewed correlational studies also considered the effect of alpha diversity in poultry. However, the positive association with performance was shown only in a single study (Lv et al., 2021), and the negative relationship in 3 studies (Konsak et al., 2013; Duquenoy et al., 2020; Zhang et al., 2022a). Others reported sampling site-dependent or not significant associations of alpha diversity with performance.

Experimental Research and Probiotics

Over the past 20 yr, there has been an exponential increase in the use of probiotics in animal agriculture, linked to the gradual decrease in the preventive use of antibiotics (Abd El-Hack et al., 2020; Shini and Bryden, 2022). In farm animals, probiotics are applied to improve production, but also behavioral and physical well-being (Kraimi et al., 2019). However, many questions remain unanswered, mainly concerning the optimal strain choice, dosing and the mechanisms by which particular probiotics act (Applegate et al., 2010; Jha et al., 2020).

Our search showed that the bacteria most commonly used in commercial poultry probiotics are (order based on the commonality of their use): Bacillus subtilis, Enterococcus faecium, Lactobacillus acidophilus, Bacillus amyloliquefaciens, Bacillus licheniformis, Lactobacillus plantarum, Lactobacillus casei, Lactobacillus rhamnosus, Lactococcus lactis, Pediococcus acidilactici, Bacillus coagulans, Bifidobacterium bifidum, Carnobacterium divergens, Lactobacillus delbrueckii, Lactobacillus farciminis, Lactobacillus fermentum, Lactobacillus paracasei, Streptococcus salivarius (for frequencies, see Figure 3). Surprisingly, from these bacterial species, only Lactobacillus delbrueckii had been identified based on metabarcoding as positively associated with FE in one correlational study (Yan et al., 2017). Other bacteria were never shown as significantly associated with condition in any of the correlational studies. However, this could result from the fact that most of the studies were using higher taxonomical categories than the species level, which can result in masking the effects of specific probiotic bacterial species.

Figure 3.

Frequency of the use of bacterial species in commercial probiotics recommended for poultry and in manipulative experimental studies. Right-hand axis: number of commercial probiotic products containing specific bacterial species (orange bars), left-hand axis: number of research studies reporting manipulative procedures including the specific bacterial species used in the probiotics (blue bars; the parameters of the survey through WOS are specified in Table S1).

To show causality, correlative studies investigating microbial effects on poultry performance need to be followed by manipulative experiments. The experiments performed used treatments consisting of dietary supplementation with either one or more strains of the potentially beneficial probiotic bacteria. Unfortunately, most experimental studies published to date lack specific explanation for their selection of the probiotic species/strain used. Often the rationale is limited to a statement that the beneficial effects of the selected treatment bacterium on avian performance have already been shown by another study, but rarely does the experimental research refer to the results of correlational studies. This approach can potentially introduce substantial experimental bias, which could then impact upon livestock management and veterinary practice. This notion could be supported by the fact that the frequency of the actual use of bacterial species in commercial poultry probiotics correlates well with their frequency of application in experimental research (Figure 3).

Most bacterial species used in probiotics belong to genera where correlational studies provided contradictory results (i.e., Lactobacillus, Bacillus, Enterococcus). This may be because these genera include species with both positive and negative effects, and the results of correlational studies depend on the particular species of these genera dominating the dataset. This urges for more selective screening to be adopted in the correlational research as well as for stronger rationale for the bacterial species and strains used in the experimental research, preferably across multiple different host species and age categories. Surprisingly, one probiotic product contains Streptococcus salivarius where the abundance of bacteria belonging to the genus Streptococcus universally negatively correlated with condition across 5 different studies, and not a single positive association was reported. Even in manipulative experiments, S. salivarius has been used in poultry only twice (Kahraman et al., 2000; Naseem et al., 2012) and in both cases as part of a multistrain probiotic treatment, where the effects of individual bacterial species are indistinguishable. These examples suggest that in some bacterial species used in poultry probiotics, there is at present limited evidence for their positive effect on condition.

Comparison of Poultry With Other Birds

Currently, a very limited number of research articles focus on gut microbiota-condition relationships in nonpoultry, namely wild bird species. Wild birds host a different spectrum of gut bacteria than domestic fowl, with a higher relative abundance of Actinobacteria and a lower relative abundance of Firmicutes (Grond et al., 2018). Also, the condition parameters studied in wild birds often differ from the poultry performance markers (except for body weight and reproductive success). Therefore, it is not surprising that condition appears to be linked with different bacterial taxa in wild birds than in poultry.

Benskin et al. (2015) found a correlation between reduced survival and the presence of an OTU related to Campylobacter lari in blue tits (Cyanistes caeruleus). Enhanced survival, on the other hand, correlated with the presence of an OTU close to Arthrobacter spp. In the great tit, Davidson et al. (2021) identified 3 OTUs from the Lactobacillaceae family that were associated with increased survival of nestlings, and the OTUs belonging to Enterobacteriaceae, Methylopilaceae and Acidothermaceae families indicated lower survival probabilities. Higher abundances of the Lactobacillaceae OTUs most strongly associated with survival at d 8 post-hatching also predicted higher body weight at d 16. Our literature survey indicated that both the bacterial families Lactobacillaceae and Enterobacteriaceae contain many species linked to improved, but also to deteriorated performance in poultry. Therefore, for interspecific comparison of the effects of bacteria from these families between the passerine birds and poultry, identification of the individual species involved in the associations would be necessary. As yet, the Methylopilaceae and Acidothermaceae families have not been found relevant to the condition in poultry.

In the house sparrow (Passer domesticus), the abundance of the family Streptococcaceae (which contains Lactococcus) was significantly more abundant in the mucosal communities of antibiotic-treated birds, which grew faster than birds from the control group (Kohl et al. 2018). This is consistent with 2 studies in poultry, where the positive effects of Lactococcus on body weight were revealed (Danzeisen et al., 2013; Han et al., 2016). Furthermore, Worsley et al. (2021) compared the gut microbiota composition of Seychelles warblers (Acrocephalus sechellensis) that either survived or died in the next season. Warblers that survived showed higher abundances of OTUs belonging to Desulfovibrio, Bartonella, Pragia, Pararhodospirullum or Anaerovorax, and individuals that died by the following year had higher abundances of OTUs identified as Thermomicrobiales, Microlunatus, Rubrobacter, Kocuria, Microbacterium, Mycobacterium, Methylobacterium, Aureimonas and Rubellimicrobium. None of these bacterial taxa was associated with condition in any of the reviewed poultry studies, except for Desulfovibrio (Singh et al., 2014) and Microbacterium (Zhang et al., 2022b), where the association with improved condition was found in both cases.

Conclusions and Future Directions

Our survey provides a comprehensive summary of results reporting the microbiota composition associations with condition and performance in poultry. Taken altogether, this survey shows limited consistency in the results of correlational studies associating individual bacterial taxa to condition and an even weaker overlap of these results with the target bacteria tested in manipulative experiments. On the other hand, we found that frequencies of bacterial taxa tested in experimental studies correspond well with the composition of poultry probiotics available in commercial products designed for birds. In our opinion, these main findings would hold regardless of some additional existing studies that could have escaped our standardized literature survey, because the results reflect the generally high heterogeneity of the study designs and analytical approaches observed across the current studies. There are 2 important applications of correlational studies focused on the microbiota-performance relationship in poultry. First, the microbiota profiles can be used for predictions of weight gain or feeding efficiency to improve selective breeding (Díaz-Sánchez et al., 2019). Second, novel bacteria can be identified for use as probiotics in the poultry industry. Both of these applications are currently limited by the fact that, even under controlled conditions, different GIT microbial communities develop across individuals. Furthermore, different bacterial taxa seem to provide benefits in different environments and microbial backgrounds (Torok et al., 2011; Stanley et al., 2016). Our survey also shows that the methodology used in poultry microbiota-performance studies is highly heterogeneous.

All the important variation highlighted by our review urges further and more standardized research in this field. Large-scale correlational studies comparing the links to multiple trials in various environments and age groups might be an effective way to overcome these issues and identify bacteria linked to improved conditions across wider ranges of environmental conditions. To understand the causalities, this should serve as a basis for subsequent manipulative experiments. Most importantly, such research needs to include a larger pool of host species and age classes, ideally also including wild models, to illuminate the consistency in the microbial effects on the physiology and productivity of avian organisms.