Regular outdoor access decreases stress with positive effects on the health of growing pigs raised on a conventional farm

Abstract

Conventional pig production systems are characterized by fully slatted floors and space restrictions, hindering pigs from foraging and roaming. In contrast, providing pigs with outdoor access allows them to express a wide range of behaviors, but might challenge their health. In this context, we hypothesized that giving pigs regular outdoor access might lower stress exposure with positive impacts for health and welfare. A total of 145 male and 148 female pigs, housed in a conventional barn, were provided (OUT) or not (IN) with access to a grassy paddock twice a week from the age of 77 days to slaughter. The OUT pigs had lower salivary cortisol levels than the IN pigs. Regular access to a pasture also lowered the severity of body lesions. While it had no consequences on animal growth, it was associated with a reduced frequency of diarrhea in pigs. Despite lower leukocytes counts in the OUT pigs compared to the IN ones, higher proportion of the OUT pigs developed humoral response against L. intracellularis. Thus, providing regular access to a paddock for growing pigs had positive impacts on their health and welfare, with no consequence on their performance.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-38461-3.

Introduction

In modern societies, the place devoted to animals is extending, with citizens showing a growing interest in animal welfare. As a consequence, European as well as American people call for a change in the way production animals are raised, and giving outdoor access to animals appears as a clear societal expectation^1,2. Pasture rearing systems provide a high-welfare environment to pigs, associated with reduced distress and increased positive emotions³. This may result from increased space allowance and/or diversity of the environment, which favors specific behaviors such as rooting, pawing, and chewing. Indeed, in semi-natural conditions, pigs spend significant time rooting (> 20%) and exploring (> 10%)^4,5.

To date, only a few systems provide outdoor access to pigs, either on an outdoor paddock or, more frequently, on organic farms, through a concrete outdoor area connected to an indoor pen^6,7. Indeed, in the European Union and worldwide, most pigs are still raised indoors, on fully slatted floors, with no or limited bedding⁸. These kinds of poorly enriched housing, developed for both economic and sanitary reasons, do not allow the pigs to express such a wide range of behaviors⁹. Because they cannot explore and root properly, pigs sometimes redirect their motivation towards objects, or even pen mates, leading to tail- or ear-biting episodes^10,11. In this context, improving the housing conditions of conventionally-raised pigs is mandatory.

When conventional farmers are questioned about the barriers to giving pigs access to the outdoors, most of them fear for the health of their animals^12,13. One concern associated with outdoor housing is the higher risk of exposure to parasites that may have deleterious effects on pigs and require the use of anti-parasitic drugs^14,15. Also, because foodborne pathogens can originate from farms, the question arises whether raising pigs outdoors increases the risk of transmission of zoonotic agents, such as Trichinella, Toxoplasma, and Salmonella¹⁶. Access to the outdoors is also associated with extended locomotion, which might lead to increased leg disorders and associated lameness. Etterlin et al. already showed in one herd that free-range pigs displayed more numerous and severe osteochondrosis lesions than did confined pigs, but without lameness increase¹⁷. In a large study in the UK, Pandolfi et al. reported that outdoor pigs seemed to be more exposed to lameness¹⁸. In contrast, permanent outdoor access was associated with a lower prevalence of respiratory symptoms in pigs¹⁹. Thus, although well-being was shown to favor disease resilience in humans and animals²⁰, it is still unclear whether giving outdoor access to pigs could be beneficial to health.

Outdoor-raised pigs exhibit increased exploratory behavior, which might increase energy demands and impair feed conversion, thereby negatively affecting their growth²¹. Furthermore, they might be more exposed to extreme temperatures and spend more energy for thermoregulation, either to produce or to lose heat, which can also affect their performance²². In this context, temporary access to a pasture appears as an alternative solution to improve pigs’ quality of life through positive experiences, while not being detrimental to individual performance. This kind of outdoor access may particularly benefit finishing pigs, since they are young adults willing to explore a richer environment. Also, at this stage, their immune system is mature and able to cope with infectious challenges.

To our knowledge, no study has yet compared the health and performance of finishing pigs raised on a conventional farm, with or without temporary access to a paddock. For this purpose, during the entire finishing period, we provided growing pigs access to an outdoor pasture twice a week. We compared their health and performance to those of siblings, housed in the same conventional barn, without outdoor access. Stress exposure of the animals was evaluated through salivary and hair cortisol levels, while performance was measured using weight gain. Animal health was assessed through the recording of clinical signs, the scoring of body lesions, and the analysis of circulating immune and inflammation markers.

Methods

Ethics statement

Animal experiments were carried out in strict compliance with the EU directive 2010/063, under the authority of the French Ministry for Education and Research (APAFIS#41824–202306051736732 v2), after being reviewed and approved by the French ethics committee n°84. All experiments involving animals were performed in accordance with the ARRIVE guidelines. All efforts were made to limit the number of pigs involved in the study and to minimize animal suffering. Animals were inspected daily by technicians, and individuals presenting clinical signs or injuries that could cause pain or compromise their welfare were either removed from the experiment or euthanized, depending on the severity. The euthanasia protocol consisted of electrical stunning followed by exsanguination, in compliance with the European regulation (EU directive 2010/63, annex 4).

Animals and housing

The data presented here were collected from a total number of 293 [Large White x Pietrain] finishing pigs, housed at the INRAE EU GenESI (Rouillé, France), an experimental farm characterized by a high health status, as previously defined²³. Three batches of pigs were included between May 2023 and May 2024. For each batch, about a hundred piglets born from 9 to 12 Large White sows were transported to Rouillé (75 km away) at weaning at the age of 27 days (27.1+/−1.6).

At the beginning of the finishing phase, when pigs were approximately 70-day old (67.4+/−1.8), they were all weighed and vaccinated against Erysipelothrix rhusiopathiae (Eryseng^®, Hipra, France). Animals were blocked by sex, litter, and weight, and assigned to the two treatments, with (OUT) or without (IN) regular access to a 5000 m² pasture. At this step, ten pigs already displayed lameness (five OUT versus two IN), or hernia (two OUT vs. one IN) (Supplementary Table 1) and those were kept in the study. Pigs were housed in two different rooms according to sex. Each room (10 × 6 m) was separated into two pens (5 × 6 m), the back one for the IN pigs, and the front one for the OUT ones (23 to 25 pigs/pen). Each pen contained one feeder, four water troughs, and four chains hung on the walls, along with 2 pieces of wood. Feed in pellet form was distributed from 8.30 am ad libitum (Alicoop, France). The light was automatically switched on every day from 6.30 am to 7.30 pm.

For the OUT pigs, the first outdoor access occurred the week following entrance in the finishing room at the theoretical age of 77 days. Pigs acceded to the pasture twice a week (on Mondays and Wednesdays for males and on Tuesdays and Thursdays for females) for four hours in the mornings, from 8.00 am to 12.00 pm. The same animal technician took the pigs in and out during the whole study. The paddock was mainly composed of grass and bordered by three fenced-in oak trees and natural hedges (Supplementary Fig. 1). A double fence surrounded the pasture to comply with biosafety requirements. An automatic drinking trough was placed at the entry of the pasture.

Sampling procedures

Samples from males and females were systematically collected one day apart. On days 104/105 and 148/149, individual saliva samples were taken from all pigs, using a saliva swab (from Salivette^® cortisol, Sarstedt, France) connected to a surgical thread (Filamid, Génia, France) to allow 30s- to 1min-long spontaneous chewing. Saliva samples were collected first from the IN pigs, while the OUT pigs were outside, then from the OUT pigs after they were back in the pen. After centrifugation, saliva samples were stored at −80°C. On days 135/136, blood samples were collected from the jugular vein of snare-restrained pigs, using 20G × 1.5’’ needles (BD Vacutainer PrecisionGlide, Becton Dickinson, UK), by two experienced staff, in less than one minute per pig. Three tubes per pig were collected (dry, EDTA-and heparin-coated), and the whole procedure lasted less than one hour from 8.00 am to 9.00 am per sex. To assess chronic stress of pigs during the finishing phase, we used a “shave-reshave” method. Pigs were shaved on the neck at the beginning of the study on day 70, then hair samples were collected after reshaving the same area on day 150. Before cortisol extraction, hair samples were stored in the dark, at room temperature.

Analysis of salivary and hair cortisol levels

Hair cortisol levels were quantified following cortisol extraction as previously described²⁴. Briefly, after washing in isopropanol, the dried hair samples were cut with scissors. Per sample, about 50 mg were ground to powder using 3 mm beads in a Retsch Mixer Mill (Verder Scientific, France), and cortisol was extracted using methanol. Both saliva and hair cortisol levels were determined using the cortisol-free saliva test kit (Demeditec Diagnostics GmbH, Germany). The intra-assay and inter-assay coefficients of variation were respectively of 8.1+/−4.0 and 11.3+/−5.6%. Hair cortisol concentrations are expressed in pg/mg of hair.

Immune competence assessment

Complete blood cell counts and lymphocyte subpopulations analyses

Complete blood cell counts were obtained from EDTA-treated blood samples using the XN-1000 analyzer (Sysmex, Villepinte, France). Lymphocyte subpopulations’ analyses were determined by flow cytometry on EDTA-treated blood samples as previously described²⁴. The following monoclonal antibodies were used: (i) Alexa Fluor^® 405-conjugated anti-pig CD3 (clone PPT3, Bio-Techne, Rennes, France), fluorescein isothiocyanate (FITC)-conjugated anti-pig CD4 (clone 74-12-4, Interchim, Montlucon, France), and phycoerythrin (PE)-conjugated anti-pig CD8 (clone 76-2-11, BD Biosciences), or (ii) PE-conjugated anti-pig CD21 (clone BB6-11C9.6, Bio-Techne). After red blood cell lysis (FACS lysis™, BD Biosciences), samples were stained with 1 µM DRAQ5™ fluorescent probe (Thermo Scientific, Courtaboeuf, France) before analysis using a Macsquant flow cytometer (Miltenyi Biotech, Germany). Data were analyzed using the FlowJo™ v10 software (FlowJo, Ashland, USA). Subpopulation cell counts were finally determined using white blood cell counts.

Whole blood assay

Blood was diluted, according to complete blood counts, in RPMI 1640 medium supplemented with 2 mM L-glutamine, 100 IU/mL penicillin, and 100 mg/mL streptomycin to ensure a final concentration of 3 × 10⁶ leukocytes/mL per well of 24-well plates. Cells were stimulated in duplicate wells using 100 ng/mL O111:B4 E. coli lipopolysaccharide (LPS) (Sigma-Aldrich) according to Gavaud et al.²⁴. Supernatants were collected after an 18 h-incubation at 37 °C and 5% CO2 and stored at − 80 °C until assays. Porcine TNFα and IL-8 were quantified using ELISA (references DY690B and DY535 from Bio-Techne), with respective detection ranges of 31.3–2000 pg/mL and 125–8000 pg/mL.

Phagocytosis activity

The phagocytosis activity of white blood cells was quantified by flow cytometry, according to the protocol given by the manufacturer of the Phagotest™ Kit (BD Biosciences) and as previously described²⁴. Results were computed using the FlowJo™ software and expressed as a percentage of phagocytes among mononuclear cells and granulocytes.

Immunoglobulin and haptoglobin quantification

Total immunoglobulin G (IgG) and haptoglobin levels in serum were quantified on D135, only in females, on diluted serum using specific porcine ELISA kits (references NBP3-12524 and NBP2-60642 from Bio-Techne), with respective following detection ranges [1.37–1000 ng/mL] and [0.94–60.94 ng/mL].

Detection of anti-Lawsonia intracellularis antibodies

D135 serum of diarrheic pigs were assessed in duplicate for anti-L. intracellularis antibodies using the Svanovir^® L. intracellularis antibody competitive ELISA (Biosellal, France). Results are expressed as a percentage of inhibition. The cut-off value for seropositivity was set to 35% according to Jacobson et al.²⁵. The value of the positive control was 95%.

Recording of clinical signs and assessment of body and tail injuries

Throughout the study, pigs were individually examined daily by an animal technician and weekly by researchers. Morphological abnormalities such as inguinal or umbilical hernia, othematoma, bursitis, and rectal prolapse were recorded. The posture of animals, and their ability to move, was checked, whilst walking into the pen and quietly encouraging pigs to move, to detect lameness or paralysis at the origin of locomotor problems. We paid special attention to the gait of pigs when they were moved for weighing and sampling. Skin was visually inspected to detect abscesses. Signs associated with respiratory diseases (coughing, sneezing, dyspnea…) were recorded. All pigs were specially screened for diarrhea on D133 by systematic examination of the feces we collected. For each clinical sign, animals were scored zero if absent or one if present to evaluate prevalence.

Tail lesions were scored weekly according to Caroll et al.²⁶: 0 = intact tail, no sign of bite; 1 = slight or healed lesions, tail tip slightly red; 2 = signs of chewing or puncture without swelling; 3 = signs of chewing or perforation with swelling and possible signs of infection; 4 = partial or total loss of the tail, necrosis. On D150, scoring of body lesions was performed by a single operator as described²⁷. Four regions (head/neck, shoulder/foreleg, back/ventral and hind/rump) were scored from 0 to 5 [with 0 = no lesion; 1 = a single small and superficial lesion (< 2 cm); 2 = more than one score 1 lesion or a single red lesion deeper than score 1; 3 = one or more extensive and deep lesion(s) (over 2–5 cm); 4 = a single deep, red and extensive (> 5 cm) lesion, or many score 3 lesions; 5 = many score 4 lesions] and the mean body lesion score per individual was reported. Finally, mortality was reported, as well as the reason when identified (hemorrhagic diarrhea).

Growth performance

Pigs were individually weighed on D70 (at the beginning of the finishing period) and D150, and the average daily gain (ADG) was calculated. Per batch, two departures for slaughter were organized around D160 and D175 according to pig weight.

Statistical analyses

Statistical analyses were performed with R (version 4.5.1²⁸). For the salivary cortisol, which was measured on two timepoints, we used a linear mixed effect model (lmer function in R) on natural log-transformed salivary cortisol with the treatment (IN vs. OUT), the time (D104 vs. D150), the sex (M vs. F), and the batch (1, 2, or 3) as fixed factors, and the individual as random factor and all possible interactions. For the other parameters, we used linear models (lm function in R) on raw (weight, ADG, and lesion scores) or natural log-transformed data, with the treatment (IN vs. OUT), the sex (M vs. F), and the batch (1, 2, or 3) as factors and all possible interactions. For IL-8 data, when above the maximal detection threshold of 8000 pg/mL, values were arbitrarily assigned a concentration of 8100 pg/mL. Similarly, for TNFα values exceeding the maximum threshold of 2000 pg/mL, the value assigned was 2100 pg/mL, while for values below the detection limit of 31.3 pg/mL, we assigned a value of 30.3 pg/mL. The goodness of fit of all models (normality of residuals, homogeneity of variance, and colinearity) was checked by visual inspection. We tested significance for the fixed effects and interaction effects on each parameter using the anova function of the car package. Significant interactions (p < 0.05) involving the treatment were further investigated with post hoc pairwise comparisons using the mvt correction. A p-value < 0.05 was considered significant. Estimated marginal means and confidence intervals were calculated using the emmeans package (emmeans). The emmeans and the upper and lower limits of the 95% confidence interval [IC95] were represented graphically using GraphPad Prism software (version 9.3.1) or written in text. Significant pairwise differences were indicated on graphs with stars: * for p < 0.05; ** for p < 0.01, and *** for p < 0.001. Finally, we performed Chi-square tests for analyses of contingency tables (body lesions, clinical signs, mortality, antibodies against L. intracellularis).

Results

Data overview

In this study, we compared 22 physiological parameters in male and female pigs, housed in the same rooms, without (IN) or with (OUT) regular access to a pasture. Figure 1 depicts the timeline of the experiment. Individual raw data are available in the supplementary Table 1 and are summarized in Table 1. We observed a batch effect on most parameters (Fig. 2 for salivary cortisol and Supplementary Table 2). These batch effects were generally not associated with a significant interaction effect between the batch and the treatment, highlighting similar treatment-related effects in all batches. In contrast, when ANOVA indicated a significant interaction between the batch and the treatment (either alone or in combination with the sex), the batch effect was further described and discussed. As shown in Table 2, and Fig. 2 for salivary cortisol, ten parameters among 22 were affected by the treatment. All results are detailed in the following sections.

Fig. 1.

Timeline of the experiment. Only the first theoretical day is indicated per time point, although samplings were performed on two consecutive days according to sex on D104, D135, and D148. Weighing, blood sampling, and hair collection were performed in a room dedicated to experiments. Saliva was collected in home pens on a day the OUT pigs went outside. First, saliva from the IN pigs was collected from 11.00 am to 12.00 pm. Then, the saliva of the OUT pigs was collected from 12.30 pm, after they were back in their home pen.

Table 1.

Raw data of the parameters assessed in pigs raised without any outdoor access (IN pigs) and with regular access to a paddock (OUT pigs), from the beginning of the finishing phase to slaughter. N indicates the number of samples that were analyzed for each parameter. IL-8, Interleukin 8; TNFα, tumor necrosis factor; ND, not done.

	Treatment
	IN pigs						OUT pigs
	Males			Females			Males			Females
Parameters	n	Median [min-max]		n	Median [min-max]		n	Median [min-max]		n	Median [min-max]
Stress mediators
D104 salivary cortisol (ng/ml)	61	4.5 [1.6–32.2]		59	4.9 [1.0–20.3]		69	1.9 [0.5–8.7]		66	2.3 [0.6–10.5]
D150 salivary cortisol (ng/mL)	61	4.0 [0.7–26.7]		67	4.4 [1.0–27.5]		70	1.9 [0.5–11.8]		63	2.1 [0.4–17.6]
Hair cortisol (pg/mg)	59	74.8 [28–174]		69	96.4 [37.5–227]		58	68.8 [30.3–173.7]		67	74.7 [24.9–241]
Body lesions
Severity of body injuries	71	2 [0–4]		73	2 [0–3]		71	2 [0–3]		72	1 [0–3]
Leukocyte counts (103/µL)
Total count	72	24.3 [15.1–40.2]		74	23.8 [16.7–32.6]		71	21.1 [14.6–39.8]		72	21.1 [15.3–36.4]
Monocytes	72	1.4 [0.9–2.3]		74	1.4 [0.8–2.6]		71	1.3 [0.8–2.6]		72	1.3 [0.9–2.2]
Eosinophils	72	0.5 [0.1–1.6]		74	0.6 [0.2–1.9]		71	0.6 [0.1–2.1]		72	0.6 [0.3–2.3]
Basophils	72	0.3 [0.1–0.7]		74	0.3 [0.1–1.1]		71	0.3 [0.1–0.5]		72	0.3 [0.1–1.5]
Neutrophils	72	6.1 [2.7–14.6]		74	6.5 [3.9–13.5]		71	6.0 [3.3–15.7]		72	5.1 [3.0–12.6]
Lymphocytes	72	14.8 [10–25.6]		74	14.5 [9.5–22.9]		71	13.1 [6.8–22.2]		72	13.7 [9.7–28.6]
Cytotoxic T cells	68	3.4 [1.7–6.1]		72	3.5 [1.7–5.5]		70	3.0 [1.6–5.4]		69	3.2 [1.6–5.8]
Naïve T helper lymphocytes	68	1.5 [0.9–2.5]		72	1.4 [0.4–2.5]		70	1.4 [0.7–2.7]		69	1.4 [0.7–2.9]
Activated T helper lymphocytes	68	1.5 [0.7–3.7]		72	1.6 [0.6–2.8]		70	1.5 [0.8–3.2]		69	1.6 [0.8–3.2]
γδ T cells	68	4.9 [2.3–10.7]		72	4.7 [2.3–10.3]		70	4.0 [1.6–7.8]		69	4.2 [2.1–9.6]
B lymphocytes	70	1.5 [0.6–2.9]		73	1.4 [0.3–3.3]		68	1.4 [0.4–2.6]		68	1.3 [0.2–3.3]
Immune competence
IL-8 secretion in WBA (pg/mL)	71	4139 [261–8100]		60	3229 [475–8100]		68	4726 [343–8100]		59	3636 [326–8100]
TNFα secretion in WBA (pg/mL)	62	200 [61–1671]		61	202 [60–2100]		59	237 [30–741]		59	204 [96–1738]
% phagocytic granulocytes	49	76.5 [48–90]		74	78.7 [46.5–91.5]		47	78.1 [38.7–88.2]		72	76.5 [55.1–88.8]
% phagocytic mononuclear cells	49	12.7 [7.7–23.7]		74	15.8 [6.3–25.1]		47	12.9 [6.8–22.6]		72	14.4 [6.1–26.5]
IgG (mg/mL)	ND	/		59	55.4 [18–172.2]		ND	/		61	50.8 [20.2–144]
Haptoglobin (mg/mL)	ND	/		56	0.6 [0.1–3.7]		ND	/		60	0.5 [0.2–2.9]
Growth
D70 weight (kg)	72	30.9 [19.3–41.2]		74	29.9 [20.7–38.7]		73	31.1 [19.4–40.6]		74	30.0 [19.2–40.5]
D150 weight (kg)	71	108 [69–137]		73	105 [81–112]		70	106 [69–126]		72	105 [80–129]
Average daily gain (g/day)	71	967 [888–1040]		73	949 [682–1004]		70	952 [861–1016]		72	929 [631–1001]

Fig. 2.

Salivary cortisol levels in pigs that had access (OUT) or not (IN) to the pasture according to the batch. Salivary cortisol levels were measured on D104 and D148 in males and females, and the effect of the treatment was analyzed using a linear mixed effect model A Results of the statistical analysis. Chisq and p-values of the linear mixed effect model are indicated for the main factors and for the interactions involving treatment. Of note, the effects of the other interactions were not statistically significant. B Emmeans ± IC95 calculated from the linear mixed effect model are graphed. Significant pairwise comparisons are indicated with stars.

Table 2.

Results of the statistical analyses describing the effect of the treatment (outdoor access), the sex, and their interaction on various parameters measured in the pigs with (OUT) or without (IN) access to the pasture. For each parameter in the linear model, F and p-values of the treatment, the sex, and the treatment x sex interaction, are indicated.

Parameters	Treatment	Sex	Treatment x Sex
Hair cortisol	F(1,241) = 5.68, p < 0.05	F(1,241) = 18.4, p < 0.001	F(1,241) = 0.00, p = 0.95
Severity of body injuries	F(1,274) = 41.4, p < 0.001	F(1,274) = 0.44, p = 0.51	F(1,274) = 0.24, p = 0.62
White blood cells	F(1,277) = 15.7, p < 0.001	F(1,277) = 0.05, p = 0.82	F(1,277) = 0.35, p = 0.55
Neutrophils	F(1,277) = 16.4, p < 0.001	F(1,277) = 4.2, p < 0.05	F(1,277) = 7.0, p < 0.01
Eosinophils	F(1,277) = 0.4, p = 0.54	F(1,277) = 3.6, p = 0.06	F(1,277) = 0.01, p = 0.91
Monocytes	F(1,277) = 3.88, p < 0.05	F(1,277) = 0.06, p = 0.80	F(1,277) = 0.26, p = 0.61
Lymphocytes	F(1,277) = 9.02, p < 0.01	F(1,277) = 0.37, p = 0.54	F(1,277) = 0.90, p = 0.34
Cytotoxic T cells	F(1,267) = 10.5, p < 0.01	F(1,267) = 1.61, p = 0.21	F(1,267) = 0.0, p = 0.95
γδ T lymphocytes	F(1,267) = 9.2, p < 0.01	F(1,267) = 0.0, p = 0.94	F(1,267) = 0.61, p = 0.45
B lymphocytes	F(1,267) = 3.54, p = 0.06	F(1,267) = 4.30, p < 0.05	F(1,267) = 0.00, p = 0.97
Naive helper T lymphocytes	F(1,267) = 0.20, p = 0.65	F(1,267) = 1.04, p = 0.31	F(1,267) = 2.97, p = 0.09
Activated helper T cells	F(1,267) = 0.31, p = 0.57	F(1,267) = 2.24, p = 0.13	F(1,267) = 0.31, p = 0.58
% phagocytic mononuclear cells	F(1,230) = 0.43, p = 0.51	F(1,230) = 14.9, p < 0.001	F(1,230) = 0.64, p = 0.42
% phagocytic polynuclear cells	F(1,230) = 0.26, p = 0.61	F(1,230) = 1.57, p = 0.21	F(1,230) = 1.68, p = 0.20
IL-8 secretion	F(1,246) = 3.8, p = 0.05	F(1,246) = 3.5, p = 0.06	F(1,246) = 0.17, p = 0.68
TNFα secretion	F(1,229) = 0.24, p = 0.62	F(1,229) = 35.9, p < 0.001	F(1,229) = 0.00, p = 0.99
Serum IgG	F(1,114) = 2.43, p = 0.12	/	/
Haptoglobin	F(1,110) = 0.90, p = 0.34	/	/
D150 weight	F(1,274) = 1.71, p = 0.19	F(1,274) = 1.64, p = 0.20	F(1,274) = 0.00, p = 0.99
Average daily gain	F(1,274) = 2.23, p = 0.14	F(1,274) = 0.17, p = 0.68	F(1,274) = 0.00, p = 0.97

Stress exposure analysis

We measured cortisol levels in saliva, collected on D104, about one month after the OUT pigs had started accessing the pasture regularly, and again on D148. As shown in Fig. 2, salivary cortisol was significantly affected by the treatment (p < 0.001), the sex (p < 0.05), the batch (p < 0.001), and the treatment x batch interaction (p < 0.01), but not by the time (p = 0.09). Pairwise comparisons per batch demonstrate significantly lower salivary cortisol in the OUT pigs compared to the IN ones in the three batches (Fig. 2). Day 150 hair cortisol was affected by the treatment (p < 0.05) with significant interaction effects with sex and batch (Table 2 and Supplementary Table 2). Pairwise comparisons in males and females further show that the level of hair cortisol was significantly lower in females from batch 2 only (Fig. 3).

Fig. 3.

D150 hair cortisol levels in females that had access (OUT) or not (IN) to the pasture according to the batch. Emmeans ± IC95 calculated from the linear model are graphed, and significant pairwise comparison is indicated with stars.

Immune competence assessment

The total number of white blood cells was affected by the treatment (Table 2). Indeed, the IN pigs had significantly higher leukocyte counts than the OUT ones (p < 0.001). Lymphocytes represent the most abundant cell population in pig blood. Accordingly, the total lymphocyte count was also affected by the treatment (p < 0.01), the IN pigs displaying greater lymphocyte numbers than the OUT ones (Fig. 4A).

Fig. 4.

Analysis of blood leukocytes on D135 in pigs that had access (OUT) or not (IN) to the pasture. Total lymphocyte (A), cytotoxic lymphocyte (B), γδ T lymphocyte (C), and monocyte (D) counts in males and females were aggregated. Stars denote the significant treatment effect from the ANOVA analyses. Neutrophil counts (E) were graphed according to the treatment and the sex. Emmeans ± IC95 are graphed, and the significant pairwise comparison is shown.

Cytotoxic CD8+ (Fig. 4B) and γδ (Fig. 4C) T lymphocyte numbers were significantly modulated by the treatment (p < 0.01 for both). Again, both cytotoxic and γδ T lymphocyte numbers were greater in the IN pigs compared to the OUT ones (Fig. 4), while B lymphocyte counts also tended to be higher in the IN pigs compared to the OUT ones (Table 2). In contrast, naïve and activated helper T cell counts were not affected by the treatment.

Monocyte count differed significantly depending on the treatment (p < 0.05), with the IN pigs exhibiting larger monocyte numbers than the OUT pigs (Fig. 4D). Neutrophil count was significantly affected by the treatment (p < 0.001), the sex (p < 0.05), and the treatment x sex interaction (p < 0.01). In females only, the neutrophil count was affected by the treatment (p < 0.001), with the IN females displaying higher neutrophil counts than the OUT females (Fig. 4E). Eosinophil numbers were not different between the IN and OUT groups of pigs (Table 2).

For TNFα secretion, the ANOVA demonstrated that the effect of the treatment was not significant (p = 0.62, Table 2), despite significant effects of the treatment x batch (p < 0.05) and the treatment x sex x batch (p < 0.05) interactions. Pairwise comparisons between IN and OUT pigs demonstrated that outdoor access had no significant effect on TNFα secretion either in males or in females, in any batch (Supplementary Fig. 2). Interleukin-8 secretion after LPS stimulation was also not affected by the treatment (Supplementary Fig. 2). The percentages of phagocytes among mononuclear and polynuclear cells were also similar in both groups of pigs (respective p-values for treatment effect: 0.51 and 0.61, Table 2). We did not observe any effect of the treatment on the total serum IgG levels (Table 2, p = 0.12). In contrast, serum haptoglobin concentration was affected by the treatment x batch interaction (Supplementary Table 2, p < 0.05). Post hoc comparisons indicate significantly lower haptoglobin levels in the OUT pigs compared to the IN ones in batch 2 only (Supplementary Fig. 3).

Finally, immune response against L. intracellularis was evaluated through the detection of antibodies in the serum of diarrheic pigs on D135. Antibodies were found in 12 out of 18 IN pigs, while all the OUT pigs displayed antibodies against L. intracellularis (Fig. 5, p < 0.01).

Fig. 5.

D135 serum specific immunoglobulins against L. intracellularis in diarrheic pigs that had access (OUT) or not (IN) to the pasture. Results are expressed as a percentage of inhibition (PI). Each dot represents one animal. The dashed line indicates the cut-off value for seropositivity.

Health monitoring

On day 150, 10 of the 146 (6.9%) IN pigs did not display body lesions (score 0) compared to 22 of the 147 (14.9%) OUT ones (p < 0.05, see Fig. 6A for distribution of the scores in the two groups of pigs, according to batch). The overall severity of body lesions was significantly affected by the treatment (Table 2, p < 0.001), the OUT pigs exhibiting body lesions of lower severity than the IN pigs (Fig. 6B). Of note, we only detected two cases of transient tail lesions of low severity (score of 2) on D135 in the OUT pigs versus none in the IN ones.

Fig. 6.

Severity of the body lesions of 150-day-old pigs that had access (OUT) or not (IN) to the pasture. A Violin plot presenting raw data of the body lesion scores according to the treatment (IN vs. OUT) and the batch. B Individual data for males and females were aggregated. Emmeans ± IC95 are graphed, and stars denote the significant treatment effect from the ANOVA analysis.

Table 3 details clinical symptoms and deaths that occurred after the beginning of the treatment. There was no significant difference in clinical symptoms’ occurrence between the two groups (19.8% occurrence in the IN pigs vs. 15.0% in the OUT animals, p = 0.28). Lameness incidence was not significantly different between the groups, with four cases for the IN pigs and six for the OUT ones (2.7% vs. 4.1%, p = 0.75). We observed two new cases of hernia in the IN pigs as well as in the OUT ones. Among the different clinical signs observed, non-hemorrhagic diarrhea, detected on D133, was the most frequent, with a total of 26 cases among 293 individuals. Interestingly, its frequency was significantly higher in the IN pigs compared to the OUT ones (12.3% vs. 5.4%, p < 0.05).

Table 3.

Recording of clinical signs and deaths observed after the first exits. The number of asterisks indicates the number of animals that were removed from the protocol (three of them were euthanized, while the other two were isolated in the infirmary).

	IN males n = 72	IN females n = 74	OUT males n = 73		OUT females n = 74
Bursitis (n = 1)	0	0	0		1
Paralysis (n = 1)	0	0	1 *		0
Rectal prolapse (n = 1)	0	1	0		0
Othematoma (n = 3)	2	0	0		1
Abscess (n = 4)	2	0	1		1
Hernia (n = 5)	2	0	0		3 **
Lameness (n = 10)	2	2	2 **		4
Diarrhea (n = 26)	12	6	4		4
TOTAL	20	9	8		14
Death (n = 5)	3	2	0		0
Hemorrhagic diarrhea	2	1	0		0
Unknown reason	1	1	0		0

Noteworthy, we had to remove five pigs from the OUT group only between D104 and D144: one because of a pre-existing hernia that became too large, the four others due to the severity of lameness (two males), hernia (one female), and paralysis (one male, repeatedly ridden by the others). Furthermore, in the IN group, we reported five deaths after D145. While two pigs died from an unknown reason, the other three died from hemorrhagic diarrhea. We also detected one rectal prolapse in one IN female, while none in the OUT group. Finally, we only observed sporadic signs evocative of respiratory diseases (coughing or sneezing) in pigs from both groups.

Performance

As shown in Table 2, the ADG was not affected by the treatment, but by a significant treatment x sex x batch interaction effect (Suppl. Table 2). Post hoc comparisons per sex and batch show no effect of the outdoor access on the ADG. Accordingly, D150 pig weight was neither affected by the treatment nor by the treatment x batch interaction (Table 2). The weight of the IN pigs was 107 [105–108] kg compared to 105 [103–107] kg for the OUT ones. The number of pigs sent to the slaughterhouse at the first departure was also not affected by the treatment (76 out of 141 for the IN pigs vs. 74 out of 142 for the OUT ones). Finally, there was no carcass condemnation at the slaughterhouse.

Discussion

We describe herein the physiological consequences of offering finishing pigs regular outdoor access to a 5000 m² paddock. The outdoor breaks provided pigs with a larger and richer environment, which allowed increased exercise as well as human-animal interactions for getting in and out. During the time of the experiment, we never had to face extreme climatic conditions and never had to maintain pigs inside for meteorological conditions (Supplementary Table 3 for meteorological data monitored during the study). Thus, pigs were offered 24 outdoor breaks during the finishing phase.

We first asked whether this type of regular outdoor access could reduce pig stress, using salivary cortisol level as a non-invasive acute stress indicator²⁹. We observed lower cortisol levels in the saliva of the OUT pigs after they went back from the pasture, as compared to the control ones on D104 and D148. To our knowledge, this is the first report of a long-term decrease in acute stress exposure induced by a non-permanent enrichment of housing conditions in conventional pigs. This suggests an efficient enhancement of pigs’ quality of life, contrasting with the usual “housing supplementations” done through the sole addition of resources to a relatively barren environment³⁰.

Despite two cases of moderate tail lesions in OUT pigs, we observed a significant reduction in the severity of body lesions in D150-old pigs raised with regular access to the outside compared to siblings kept indoors. This is an indirect proof of the decrease of agonistic behaviors between the OUT pigs, which may result from the increased richness of their environment. Interestingly, at slaughter on day 150, Foury et al. previously reported a similar reduction of skin lesions on whole carcasses, but no difference in urinary cortisol levels, on pigs raised with permanent outdoor access (150 m²/pig), as compared to conventional ones (0,65 m²/pig)³¹.

While regular access to an outdoor pasture alleviated acute stress response and fights in finishing pigs, indicative of improved welfare, we did not detect a difference in hair cortisol levels at the end of the study, except in females from batch 2. This could be partly due to the strong batch effect observed for this parameter, with batch 2 displaying the lowest level (Supplementary Fig. 4). Interestingly, hair cortisol level was the highest for batch 1, which was run during the summer, confirming that finishing pigs are more sensitive to heat compared to cold stress. Indeed, the upper limit of the thermoneutral zone of pigs exceeding 60 kg is below 25 °C, and the average indoor temperature measured for batch 1 was 26.7 °C (24.1–31.4 °C, Supplementary Table 3). Furthermore, hair cortisol levels were particularly high in this farm, with a mean of 92 +/- 39 pg/mg compared to an average value of 25.9 +/- 16.2 pg/mg obtained from 950 finishing pigs housed in 20 different farms³². Thus, we can presume that four hours of outdoor access twice a week was not sufficient to dampen the high chronic stress exposure in this farm. In this context, it would be interesting to test this treatment in several commercial conventional farms and/or to study the effects of giving daily outdoor access to pigs.

While accessing the pasture induced increased locomotor activity and allowed pigs to express a wider range of behaviors, we did not observe any detrimental consequences on their growth. It was previously demonstrated that giving permanent outdoor access to pigs (350m²/pig, from the age of 2 months) had no significant effect on their growth rate³³. In the current study, average daily gain was also similar between the IN and OUT pigs. Interestingly, the OUT pigs were observed chewing grass during recreations, and we can hypothesize that the uptake of nutrients from pasture might have contributed to sustaining pig growth, as already reported³⁴.

Outdoor access did not significantly increase clinical lameness in pigs. The prevalence we observed (< 5%) was similar to the one recently reported by Buijs et al. on fattening pigs reared in grassland and agroforestry systems³⁵. There was no joint condemnation at the slaughterhouse, either for the IN pigs or for the OUT ones. An increased level of osteochondrosis leading to joint condemnations had previously been observed in free-range as compared to confined fattening pigs¹⁷. In our study, we might have prevented lameness by vaccinating pigs against Erysipelothrix rhusiopathiae, a major cause of arthritis. Exercise might also have contributed to reinforcing the musculoskeletal system of the OUT pigs³⁶.

When exposed to the outdoors, pigs are more prone to encounter gastrointestinal parasites¹⁴, including zoonotic ones such as Toxoplasma gondii³⁷. In our study, we did not observe signs of parasitic infestations in the OUT pigs, on the farm, and at the slaughterhouse, although pigs did not receive any anti-parasitic treatment. However, we did not perform coprological analysis, and we cannot exclude that the OUT pigs were more infested by parasites than the IN ones. Indeed, these types of infestations are generally subclinical^37,38.

Environmental and social enrichment was previously demonstrated to reduce disease susceptibility and improve clinical outcome in pigs co-infected with Porcine Reproductive and Respiratory Syndrome Virus and Actinobacillus pleuropneumoniae⁴⁰. In the current study, regular outdoor access has prevented digestive disorders in pigs. Indeed, the frequency of diarrhea was significantly higher in the IN pigs, and three IN pigs finally died from hemorrhagic diarrhea. L. intracellularis, a well-known bacteria associated with enteric disease in pigs⁴¹, was present on the farm. Indeed, we detected antibodies against this pathogen in the serum of diarrheic pigs (14/26), while pigs had not been vaccinated. Interestingly, the four diarrheic IN pigs that died after D147 did not exhibit antibodies against L. intracellularis, although the pathogen was detected on D133 in the feces of two of them (not shown). In contrast, all the diarrheic OUT pigs displayed antibodies against L. intracellularis, suggesting that the OUT pigs were either more exposed to the bacteria and/or more efficient in developing specific immune responses. Since stress, known to induce immune dysfunctions (reviewed by⁴²), was previously described as a risk factor for this disease⁴³, we can hypothesize that the reduced stress exposure helped the OUT pigs to better fight against L. intracellularis infection. A similar protocol conducted in commercial farms characterized by contrasting health status, associated with concomitant follow-up of pathogen exposure, would help to decipher whether outdoor access significantly improves disease resilience or not.

In the current study, the OUT pigs not only went outdoors but also exercised more than the IN ones, and this might have contributed to the positive effects we detected. For example, play behavior and associated exercise were previously shown to allow reduction in the clinical severity of a PRRSV infection in pigs⁴⁴. In this study, Steinerová et al. demonstrated lower leukocyte count (affecting especially monocytes) in PLAY pigs compared to controls, which they attributed to increased exercise. We also found lower white blood cell counts in the OUT pigs compared to the IN ones. Lower leukocyte count was also previously associated with enriched indoor-housing conditions in 6-week-old pigs⁴⁵. This study reported lower haptoglobin levels and higher antibody titers following KLH (Keyhole Limpet Hemocyanin) immunization, suggesting that enriched housing conditions lowered immune system stimulation but improved specific antibody response in finishing pigs. In our study, haptoglobin levels were similar, except for one group of females, and we only explored innate immunity traits. Outdoor access did affect neither the ability of leukocytes to secrete IL-8 and TNFα in response to LPS ex vivo, nor their phagocytic activity. This is not surprising since these parameters were greatly influenced by a batch effect. We previously described a significant batch effect in pigs fully raised indoors⁴⁶. In this case, we cannot exclude that season might have modulated some immune parameters, while affecting the abiotic and biotic pig environment, as previously described for humans⁴⁷. We also observed an effect of sex on several immune parameters. Data were collected from adult uncastrated males and females, and sex is known to influence immune traits (reviewed in⁴⁸).

Although they were raised in a conventional barn and sold at the end of the experiment, the point of profitability was not as important as in commercial farms. Indeed, in an experimental farm, human resources are well-adapted. Taking pigs in and out required time, even for short distances (70 m, including 10 m indoors), which increased the workload of the farmer. At the beginning of each repetition, pigs had to understand what was going on and needed to adapt their behavior. With time, the duration required to take the pigs out decreased, while the time needed to get them back to their home pen increased, although they still went in spontaneously at the opening of the barriers. We read this as an indirect proof that the pigs appreciated recreations, which was also a source of satisfaction for the farmer. Thus, despite an increased workload, providing pigs with temporary outdoor access might also contribute to the well-being of farmers. This could help to increase the attractiveness of the sector, but it needs to be confirmed in other commercial farms.

Giving outdoor access to conventional pigs raises the question of its environmental impact (recently reviewed by Ruckli et al.⁴⁹. Indeed, the development of concrete outdoor runs in organic farms was identified as the main environmental risk factor due to ammonia emission from soiled surfaces⁵⁰. In contrast, while farms with outdoor access are classically associated with higher environmental costs, the trade-offs between land use, greenhouse gas emission, antimicrobial use, and animal welfare are not inevitable⁵¹. In the current study, providing pigs with regular access to a pasture appears as an opportunity that might limit the use of arable land while strongly supporting pig welfare.

Thus, providing regular access to an outdoor pasture appears as a valuable option to improve the quality of life and the health of conventionally-housed finishing pigs. Nevertheless, this study was conducted in a single farm, and the data now need to be confirmed in different commercial farms, characterized by contrasted health status.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

BL, CT, and JH designed the experiment. CT and JH supervised the project. AJ, FG, JL, SG, KH, TT, SF, BL, and JH conducted the experiments. AJ, JL, KH, BL, and JH analyzed the results. AJ, BL, and JH interpreted the results and wrote the main manuscript text. All authors have read and approved the submitted version of the manuscript.

Funding

AJ was supported by the INRAE through her PhD fellowship. The study was funded by the INRAE Metaprogram SANBA through the PANORAMA project, supervised by CT and JH.

Data availability

The data generated and analyzed in the current study are available in Supplementary Table 1.

Declarations

Competing interests

The authors declare no competing interests.