Association between intestinal microbiota composition and tail-biting in pigs CURRENT STATUS: POSTED

Background: Tail-biting (TB) is a serious behavioral disorder in pigs. It is defined as a pig chewing the tail of another pig. It is an important challenge in swine production as it impacts the animal welfare, its health and the economics and safety of the pork meat supply chain. Some treatments have been proposed but have not proven optimal. Nutrition, feed type and composition, appears to be an important factor in TB behavior, perhaps by modulating the intestinal microbiota (IM). In humans, IM is increasingly recognized as a modulator of behavior. Our aim was to assess an association between TB behavior and IM in pigs through comparisons of IM in groups of biter or bitten pigs to the IM in a non-biter and non-bitten negative control group, and through comparisons of IM between the biter and bitten groups. Each group, composed of 12 pigs, was formed at the beginning of the growing/finishing phase based on a target behavior analysis centered on TB behavior for the biter group and a score of damages caused to the tail for the bitten group. Fecal and blood samples were collected from each pig during a TB episode, at time 0, t0, and after a TB episode, four weeks later, at time 1, t1. Serum cortisol level was determined by ELISA and used as an indicator of stress. The pig’s fecal microbiota was analyzed from DNA extracted from freshly collected fecal matter using amplicon sequencing of the V4 hypervariable region of the 16S rRNA gene. Results: Serum cortisol levels were significantly higher in either the biter or bitten pig groups compared to the negative control group ( p =0.02 and p =0.01, respectively). Interestingly, no significant difference was revealed between the biter and bitten groups. The microbiota alpha-diversity was not significantly different between all groups, biter, bitten and negative control. Analyses of beta-diversity, however, revealed a significant difference between either the biter or the bitten group in comparison to the negative control group in terms of structure and composition of the microbiota. Lactobacillus were significantly more abundant in the negative control group compared to the two other groups ( p =0.001). No significant difference was revealed between the biter and bitten groups. Quantitative real-time PCR


Background
Tail-biting (TB) is a behavior often observed on pig farms and is expressed by a pig nibbling the tail of a pen mate. Tail-biting is considered a severe behavioral disorder related to cannibalism (1,2), and causes pain, infection, stress (3)(4)(5) and reduced growth performance (6) in bitten pigs.
The average incidence of TB behavior reported by several studies varies between 1.3 and 7.2% (1,6).
These results are, however, based on the assessment of ante-and post-mortem lesions and not based on observation of TB behavior at the farm. At slaughter, the degree of lesions on the carcass varies from detectable (70%) to severe (1 to 3%), resulting in partial or total loss of the tail (7)(8)(9). Tail-biting does not only have consequences on animal welfare but also on the economics of the pork meat supply chain due to veterinary treatment (9), reduction of growth performances and total or partial carcass condemnation at the abattoir (8,9).
Several risk factors have been associated with TB, most linked to a stressful farm environment such as housing on slatted floor and indoors (10), lower space allowance in the pen (11) and poor air quality (7,(12)(13)(14). A strong correlation between the prevalence of diseases, such as respiratory disorders and TB has also been reported (11,15). Individual and genetic characteristics appear to play a significant role in the expression of TB behavior in pigs, with for example, the Yorkshire pigs being more likely to be victims of TB compared to the Landrace pigs (16). Age and sex have been reported as risk factors. Tail-biting is more frequently reported at the fattening stage, and females bite more than males (17). However, insufficient information is available about the possible correlation between these different risk factors, and it is difficult to clearly identify the triggering cause of TB (18). Nutrition is also considered a risk factor for TB (12). Attraction to blood from a bleeding tail may increase TB when diet is low in proteins, minerals, or fibres (18)(19)(20). In addition, studies have found that the dietary supplementation with tryptophan decreased the general activity of pigs and their attraction to blood in farms where TB is present (19,21,22). A link has been shown between serotonin, a precursor of tryptophan, and TB in biter pigs (6,23). Pigs are also more active during daylight, and it has been shown that TB is more frequent before midday (12,24).
Biter, bitten and control pigs exhibit different jejunal morphology, blood metabolites, activity of the immune system and stress levels (4,25,26). It is tempting to speculate that biter, bitten and control pigs may harbor different intestinal microbiota (IM) composition (18) which can be regulated by nutrition, feed composition and stress conditions (18,27). Studies on mice have shown that a modification of IM composition may influence the hypothalamic-pituitary-adrenal (HPA) response to stress (28,29), resulting in anxiety and depression (30,31). Furthermore, some bacteria in the IM may regulate the pathophysiology of stress-related disorders through the synthesis of serotonin (e.g. Candida spp., Streptococcus spp., Escherichia spp. and Enterococcus spp.) (32,33) and gammaaminobutyric acid (some Lactobacillus spp. strains and Bifidobacterium spp.) (34). Cortisol secretion is regulated by the brain and increases with an animal's coping strategy to stress (35). It affects some permeability parameters of the intestinal epithelial barrier which leads to a modification of the IM (35,36).
The aim of this study was to assess the association between TB behavior and IM in pigs by characterizing the animal stress response as measured by its serum cortisol level and by comparing the IM structure and composition between biters, bitten and control pigs, during and after a TB episode at the farm.

Animal selection
Tail-biting behavior was observed in ten pens (10/32; 31%) after the first 5 weeks of the growing/finishing phase at a commercial farm. About 40 biter pigs were pre-selected within these pens using the "behavior sampling method" (two observations of 2 h/24 h for 7 days). Initial pig selection was determined based on total TB behavior frequency ranging from 4 to 39 bites over the one-week pre-selection period (Fig. 1). To obtain the targeted sample size of 12 pigs per group, the biters with the highest biting frequency were selected (> 20 bites) in the different observation periods within the same week (Fig. 1). All biter pigs came from seven pens.
Within the ten pens where TB behavior was observed, 70% of pigs presented no tail damage (0 score), 13% and 17% showed some scratches (score 1) and a small bleeding lesion (score 2) on the tail. Twelve pigs with a score 2 tail damage and from the same seven pens that housed the biters were selected to form the bitten group. In addition, 12 pigs randomly selected from three pens with no TB (0 score) formed the control group.

Behavior And Tails Condition
In biter pigs, TB behavior was at its highest peak at the time of selection (time 0, week 1; t0, w1) and decreased over the following weeks until it became almost non-observed, at the onset of the fourth week (t1, w4) (Fig. 2).
Tails of all bitten pigs healed between 2 and 4 weeks after selection and remained intact until the end of the finishing period. Biters were never bitten; they were true biters, and the negative control pigs never became biters or bitten pigs.

Variation Of Serum Cortisol Concentration
Serum cortisol levels were higher in biter and bitten pigs compared to negative control pigs at t0 (p = 0.02 and p < 0.01, respectively; Fig. 3). No significant difference in cortisol level was observed between the biters and bitten groups at t0 and t1 or between all three groups, biter, bitten, negative control, at t1 (p > 0.05). Based on the composition of the microbial community positive sequencing control, an acceptable error rate of 0.094% was calculated.

Microbiota Description And Analysis
No significant difference in the alpha-diversity indices, OTUs, Shannon-even and inverse Simpson, was revealed between all three groups, biter, bitten and negative control, at t0 and t1 ( Table 2). Touches the tail with a reaction from the receiver, who is startled or moves away. Touches the tail without a reaction from the receiver, who doesn't move.
Moves the tail with a reaction from the receiver, who is startled or moves away. Moves the tail without a reaction from the receiver, who doesn't move. Harmful social behavior (biting of the tail) PQG − PQG Takes the tail in its mouth (with a reaction -) Takes the tail in its mouth (without reaction) Takes the tail in its mouth with a reaction from the receiver, who is startled or moves away. Takes the tail in its mouth without a reaction from the receiver, who doesn't move. Table 2 Comparison of alpha-diversity indices of the intestinal microbiota of three groups of pigs, biter, bitten and the negative control pigs. 1 t0: immediately following selection of the pig groups; and t1: four weeks later.
Beta-diversity was compared between the three pig groups at t0 and t1 (Table 3). A significant difference in microbiota structure was observed between the biter and the negative control, and between the bitten and the negative control group, both at t0 with the Yue & Clayton index (Fig. 4). Table 3 Comparison of the intestinal microbiota structure of pig groups 1  Tables S1, S2).
Likewise, when the bitten group was compared to the negative control group at t0, two genera, The composition and diversity of the intestinal microbiota was not significantly different between the biter and the bitten groups (p > 0.05).
As expected, when the positive control group (non-biter and non-bitten-antibiotic-treated) was compared with the negative control group (non-biter and non-bitten), significant differences were found in alpha-diversity indices at t0 (Shannon-even p = 0.002 and inverse Simpson p = 0.002), and beta-diversities at t0 (Yue & Clayton p = 0.001; Jaccard p < 0.001) and t1 (Yue & Clayton p = 0.028; Figure S2). This validates the experimental model and subsequent bioinformatics analysis abilities to measure and identify microbiota modification in the studied animals.

Real-time Quantitative Polymerase Chain Reaction
A qPCR assay was performed to quantify and validate the results obtained from LEfSe for the genus Lactobacillus. A significant difference was observed when comparing either the biter or the bitten pig groups to the negative control pig group at t0 (during TB), with respective averages of 1.15 and 1.11 log of gene copies per ng of DNA. However, no difference was observed at t1 (four weeks after TB) ( Fig. 5).

Discussion
Our aim was to study a possible association between IM composition and tail-biting. First, animal stress was assessed by serum cortisol level, second, IM structure and composition were compared between either biters or bitten pigs and control pigs, during and after a TB episode at the farm. Four pig groups, biter, bitten, a non-biter, non-bitten negative control, and an antibiotic-treated positive control for the IM composition, were formed in a commercial fattening farm during a TB behavior episode that was video-recorded. Tail-biting appeared two weeks after the beginning of the growing period and initially only in two non-contiguous pens. This is in agreement with similar studies on pig TB behavior (7,(37)(38)(39). Tail-biting within the farm appeared variable over time, as previously described (40). Once present, TB behavior frequency decreased gradually in the 12 selected biter pigs four weeks after pig recruitment (40).
Cortisol levels were compared among groups. The levels were higher in both biter and bitten pigs at t0, when TB behavior was at its peak in selected animals, than in the negative control group, presumably a result of increased response to acute stress in pigs (41,42). Our results for the biter and bitten pigs are in agreement with those of Smulders et al. (2006) (43,44). Four weeks after the animal selection, at t1, cortisol levels were high in all three groups, biter, bitten and control. The cortisol levels between biter and bitten pig groups, showed no difference either at t0 or t1, indicative that both groups contained stressed animals.
Additionally, we also studied a possible association between TB and IM and we generated novel information on the structure and composition of the IM in these stressed animals. Numerous studies have used high throughput sequencing methods to explore the composition of animal's IM (45)(46)(47)(48).
Here, the structure and the composition of the IM in feces samples of biter and bitten pigs were OTUs associated respectively to the biter, the genus Coprococcus, and bitten pig groups, the genus Sphaerochaeta and the Phascolarctobacterium OTU, both stressed groups, had not been previously associated with behavior disorders.
We also showed that the relative abundance determined by qPCR of Lactobacillus, could distinguish biter and bitten groups from the negative control group. It has been reported that some Lactobacillus species are associated with behavior disorders such as anxiety and depression in both humans (31,50,51) and mice (52,53). Conversely, different Lactobacillus species are used in probiotics to reduce behavior disorders, such as Lactobacillus rhamnosus JB-1 for anxiety and stress (54) or Lactobacillus helveticus R0052 for depressive behaviors (55). It would be interesting to test whether a manipulation of the IM of pigs exhibiting TB behavior through the introduction of either or both Lactobacillus species could shorten the disorder.
Because this study was carried out on a commercial farm, it is important to emphasize that several risk factors, most of which are also stress factors associated with TB behavior, were present, such as intact tails, slatted floors pens without enrichment and a relatively high number of pigs per pen (11 pigs per pen) (7,10). However, it is noteworthy that biter and bitten pigs were in the same pens and all subjected to the same environment. In this study, no specific event could be recorded as a trigger for TB behavior.
With respect to the relationship between IM and TB, at least two hypotheses can be proposed.

Conclusion
To the best of our knowledge, our results provide the first evidence of a relationship between the occurrence of a behavior problem, TB, in biters and bitten pigs and IM. We showed that biter and bitten pigs were stressed in comparison to the negative control group, as revealed by higher cortisol levels. Pigs in the negative control group had more Lactobacillus in their IM than those expressing TB, which is consistent with human and mice studies on the relationship between behavioral disorders and microbiota composition. However, the mechanisms underlying the association between TB and IM are still unknown. Further studies are needed to gain a better understanding of the cause-effect relationship between both.  In order to further define the groups of animals to be studied, the "target behavior sampling" method (58), centered on observations of the behavior of interest, here TB, was used on recordings of the 10 pens where TB behavior was present, based on the ethogram. Sampling observations were distributed among two periods: from 10:00 am to noon and from 8:00 pm to 10:00 pm, which corresponded to the peaks of TB behavior. Sampling survey lasted for eight consecutive days until the study groups A positive control group consisting of 12 non-biter/non-bitten antibiotic-treated pigs was also formed.

Animals and housing
Blood samples for serum cortisol analysis were not taken. Chlortetracycline at 1210 ppm was added to their diet seven days prior to each sampling date to induce conformational changes in their IM. This group served as a positive control in our later 16S rRNA gene amplification, sequencing and analysis.
The average bite number of biter pigs was plotted on a weekly basis, w1, w2, w3 and w4.

Blood And Feces Sampling
Blood and feces were sampled in the biter, bitten and negative control groups, during two different periods, t0 and t1. No sampling was done on the AT group. At t0, pigs were 12 weeks of age, and TB behavior episode was beginning. Four weeks later, at t1, the TB behavior episode was considered finished.
Blood (approx. 5 ml) was collected from the jugular vein of snared pigs by a single experienced animal technician in the pen. Blood samples were kept at room temperature for 2 h to allow blood to clot prior to centrifugation (15 min at 1,000 x g). The serum was transferred to 1.5-mL Eppendorf tubes and stored at -80ºC pending cortisol concentration analysis. Serum cortisol concentration was determined using the Cortisol ELISA Kit (Pig) (Abnova, Taiwan) according to the manufacturer's recommendation. The minimum detectable concentration of cortisol was 0.2 ng/ml.

Consent for publication
All co-authors have read and approved the final version of the manuscript and consent to its publication.

Availability of data and material
All data and material will be made freely available upon acceptance of the manuscript for publication.

Competing interests
The authors declare that they have no competing interests.     indicate statistically significant differences between groups (* p < 0.003; ** p < 0.003). NS: No difference in average log of gene copy number/ng of DNA within the t1 period (p > 0.05).

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.