Article Text

Pig-shed air polluted by α-haemolytic cocci and ammonia causes subclinical disease and production losses
  1. T. Murphy, BAgSc1,
  2. C. Cargill, BVSc, PhD2,
  3. D. Rutley, BAgSc, PhD1 and
  4. P. Stott, BVSc, MSc, PhD, Dip Ed1
  1. School of Animal and Veterinary Sciences, The University of Adelaide, Roseworthy Campus, Roseworthy SA 5371, Australia
  2. South Australian Research and Development Institute, Roseworthy Campus, Roseworthy SA 5371, Australia
  1. E-mail for correspondence Dr P. Stott, e-mail: philip.stott{at}

There is mounting evidence that bacteria originating from pigs degrade the environment of the pig shed and adversely affect the health of the animals and the pig-shed workers. α-haemolytic cocci (AHC) occur in pig-shed environments, but are regarded as commensals. Ammonia is also a component of the pig-shed environment, and is known to damage upper respiratory tract epithelia. The aim of this study was to determine whether polluted air in pig sheds adversely affected performance indicators in pigs. Modelling revealed a direct effect of AHC on voluntary feed intake and hence AHC are not commensal. No direct effect of ammonia on the pigs was detected, but the combination of AHC and ammonia stimulated the immune system in a progressive manner, and there were direct effects of immune stimulation on food intake and growth resulting in poorer feed-conversion efficiency, even though the effects remained subclinical. The authors conclude that exposure of the respiratory epithelia of pigs to viable AHC in the presence of ammonia redirects nutrients away from production and towards the immune system, explaining the impact of poor pig-shed hygiene on production parameters.

Statistics from

GLOBAL economic and agricultural policies have driven agricultural enterprises in advanced economies to become larger, more intensive and more specialised (Donham 2000), with greater capital investment and less labour input. The modern pig industry is a high cost, high technology, intensive industry with narrow margins, which must achieve high levels of efficiency if adequate profitability is to be maintained (Hope 1990). Consequently, pig sheds are designed to house pigs as densely as possible, but the confined airspace retains airborne pollutants, and there are links between surface hygiene, suboptimal air quality and the prevalence and severity of disease in pigs, which lead to reduced growth rates (Donham 1991, Renaudeau 2009), affecting profit margins. Air-borne pollutants include microorganisms, their endotoxic cell-wall fragments and ammonia (Banhazi and others 2008).

The link between poor air quality and poor growth rates may be mediated through an effect of pollutants on the animals' immune function. Poor sanitary conditions in pig sheds are associated with the induction of inflammatory responses (Le Floc'h and others 2009), and the inflammatory activation leads to slower growth, partly because of reduced voluntary food intake (VFI; Escobar and others 2004, Renaudeau 2009), but suppression of food conversion efficiency may also occur (Le Floc'h and others 2009), and the immune response itself has a nutrient demand (Le Floc'h and others 2004, 2007).

Ammonia is the most important gaseous pollutant in pig sheds (Subramanian and others 1996). It is highly water soluble and reacts with the moisture on mucosal surfaces to produce an alkaline, corrosive and irritating solution of ammonium hydroxide (Brautbar 1998). Its solubility means that most of the gas is absorbed in the nasopharynx (Urbain and others 1996a) and damages respiratory epithelia at levels commonly occurring in pig sheds (Urbain and others 1996b). It induces inflammatory responses in the respiratory system (von Borell and others 2007), and suppresses the cough reflex (Moreaux and others 2000). Even short-term exposure will depress the defences of the respiratory tract against inhaled microorganisms (Gustin and others 1991).

The purpose of this study was to examine the impact of the exposure of the upper respiratory system to ammonia gas and to the airborne microbes found in pig sheds on growth rate, feed utilisation and immune-system parameters. The authors chose α-haemolytic cocci (AHC) including viridians-group streptococci (VGS) because they occur in the gut or faeces of some species (VGS, Thanantong and others 2006, Aerococcus viridians, Guo and others 2007, Budzinska and others 2009, Byrne-Bailey and others 2009), they are prevalent in piggery airspaces (VGS, Cargill and Skirrow 1997, Done and others 2005), they are considered to be generally non pathogenic (VGS, Van der Hoeven and Camp 1991, Aerococcus viridians, Park and others 2002), and they have been found, in some host species, to be upper respiratory-tract commensals (VGS, Van der Hoeven and Camp 1991, Aerococcus viridians, Silvanose and others 2001). The authors hypothesised that AHC have subclinical pathogenic influences on growth rate and feed production and that the effect would be exacerbated by exposure to ammonia. Hence, the authors examined impacts on production parameters and immune function of pigs exposed to ammonia and/or to a mixed culture of AHC.

Materials and methods

Study site

The study was undertaken at the Roseworthy Campus, The University of Adelaide, Australia. Australia is free from transmissible gastroenteritis and porcine reproductive and respiratory syndrome, and the piggery is free of helminth parasites, swine dysentery, and atrophic rhinitis, and maintains control of mycoplasmal pneumonia, erysipelas, Glasser's Disease, leptospirosis and clostridial diseases through vaccination.

The experiments were conducted in a fully enclosed room containing 20 1.6 m2 individual pens with partially slatted floors. Filtered air was supplied under positive pressure, and temperature and relative humidity were set at 24°C and 55 per cent, respectively. The pens were flushed with potable water three times daily into a slurry pit, which in turn was flushed every second day while the pigs were out of the room, and allowed to dry before the pigs were returned. Dirty pigs were washed and dried before being returned to their pens.

The pigs were weighed daily. They were offered a daily ration of 3.0 kg of a commercial diet. Uneaten food was collected and weighed. Voluntary food intake (VFI, kg/d), average daily weight gain (ADG, kg), and food conversion ratio (FCR) were calculated. The daily feed loss due to immune challenge was calculated as the product of the mean VFI of the challenged animals minus the product of the mean FCR of the unchallenged animals and the mean ADG of the challenged animals.

Experimental design

The study was conducted as a 2 × 4 factorial with the main effects of bacterial challenge (AHC or AHC+) and ammonia (NH3 0, 10, 25 or 50 ppm) in eight blocks (AHC-NH30; AHC-NH310; … AHC+NH350). In each block, 20 pneumonia-free 16-week-old Large White × Landrace female pigs were used. Ammonia at 0, 10, 25 or 50 ppm was supplied in nitrogen (BOC gases Australia) into each feed bin during a 15 minutes feeding period at a rate of 12 l/minute, based on the United States Occupational Safety and Health Administration permissible short-term exposure limit for ammonia of 35 ppm for 15 minutes (OHSA 1998). The short-term protocol was chosen because it is regarded as a mild to moderate exposure for human beings and presumably therefore also for pigs, because the exposure is readily standardised, and because there would be a clear association between the exposure to ammonia and any subsequent inoculation with AHC.

AHC were obtained from a grower room, a finisher room, and a weaner room in adjacent pig sheds using a six-stage Andersen Sampler loaded with Columbia Horse Blood Agar plates. Each airspace was sampled three times at its central point 0.5 m above the floor for five minutes at a flow rate of 1.9 l/minute. α-Haemolytic colonies with differing colony morphologies were selected from each group of plates for phenotypic identification using API 20 Strep strips in accordance with the manufacturer's instructions (bioMérieux, La Balme les Grottes). Organisms yielding unique API profiles were then sequenced by 16S rRNA gene sequencing. Sequence matches were sought in the GenBank+EMBL+DDBJ+PDB sequences using the program BLASTN ver. 2.2.24+ (Altschul and others 1997). The matching was performed after the GenBank accession date for Aerococcus suis partial 16S rRNA gene (type strain 1821/02T; June 29, 2009).

Half of the experimental pigs were exposed to AHC by intranasal inoculation of 3 ml of a mixed suspension of 6 × 105 cfu/ml in buffered saline with a composition matching the proportions of each organism as identified above. The inoculation was undertaken 30 minutes after the start of the first exposure to ammonia.

Immunological parameters

Leucocytes were extracted from heparinised jugular blood samples obtained at both the start and the end of the trial. Stimulation of the immune systems of the pigs was assessed using a lymphocyte proliferation index (LPI), an assessment of the phagocytic potential of heterophilic polymorphonuclear leucocytes (HPP), measures of the proportions of lymphocytes expressing cluster of differentiation (CD) group 4, CD8 or CD21 markers, and the proportion of double-positive (DP) CD4 and CD8 lymphocytes. The ratio of the CD4 to CD8 markers was calculated. For each pig, the changes in the markers over the duration of the trial were recorded, and the differences in the markers between the control pigs and the inoculated pigs at the end of the trial were calculated.

HPP was measured by the uptake of fluorescent 1 µm diameter microspheres (Polysciences) by polymorphonuclear leucocytes (Kato and others 2000). The flow cytometric profiles were acquired using a FACSCalibur Flow Cytometer in conjunction with CELLQuest software (BD Biosciences). The analysis gate was set around heterophils on forward- and side-scatter profiles. The results were expressed as the percentage of heterophils that contained fluorescent microspheres.

LPI was based on the incorporation of tritiated-thymidine (Amersham Biosciences) into the replicating DNA (Maluish and Strong 1986) of peripheral blood mononuclear cells (PBMC). Incorporated tritiated-thymidine activity was determined by a Microbeta Trilux 1450 beta counter (EG&G Wallac). Tests were performed in triplicate. LPI was expressed as the mean counts of stimulated cultures/mean counts of medium control.

Lymphocyte subsets were identified by measuring cell-surface immunofluorescent labelling (Chamorro and others 2000) with fluorescein isothiocyanate (FITC), phycoerythrin (PE), streptavidin peridinin chlorophyll protein (PerCP)/streptavidin-Cy-Chromme (CyC) or allophycocyanin (APC; BD Biosciences Pharmingen). Three- or four-colour flow cytometric analysis was performed using an FACSCalibur Flow Cytometer in conjunction with CELLQuest software (BD Biosciences). The analysis gate was set around lymphocytes on forward- and side-scatter profiles. Dual expression of CD4 and CD8 was determined by gating on CD3 CD8 bright lymphocytes. The results are expressed as the mean percentage of cells expressing the particular phenotypic marker. The CD4:CD8 ratio was calculated from the available data.

Pathological parameters

The pigs were slaughtered at the end of the trial, and the lungs were retrieved and examined for gross pathology. From each pig, a small portion of the dorsal diaphragmatic lobe of the right lung, together with a section of the trachea just anterior of the bifurcation, was preserved in 10 per cent buffered formalin, and sections stained with haematoxylin and eosin were prepared. For each animal, 10 sections, chosen at random, were examined to assess the state of the epithelial layer, and the number, percentage and type of inflammatory and immune cells present.

Statistical analyses

All statistical analyses were undertaken using PASW Statistical Software, Version 18.0 (IBM 2009). A general linear model was developed (Nelder 1994) with the response variable being either voluntary feed intake (VFI) or average daily gain (ADG). The explanatory variables were as follows: (1) factors, -AHC (Yes or No) NH3 (0, 10, 25 or 50 ppm); (2) covariates, LPI, HPP (per cent), CD21 (per cent), CD4 (per cent) and CD8 (per cent), all recorded at day 0 and at day 14; and 3) quadratic terms of covariates.

To avoid cross-contamination of the effects of either bacterial or ammonia treatment, block was necessarily confounded with treatment and so the block effect was not included in these analyses. The effect of block was tested and found to be systematic rather than random, supporting the decision to remove the block effect from these analyses. This systematic effect of block was expected as it was actually the treatment effect.

The final statistical model was developed using stepwise backwards elimination from the maximum model. Interactions and effects were eliminated based on tests of significance (P<0.01) using Type III sums of squares while ensuring the marginality requirements of Nelder (1994) were maintained. The maximum model included factors, covariates and quadratic terms as main effects, with all two-way interactions and three-way interactions between factors and linear terms of covariates. Results are expressed as least squares means ± se.


Airborne α-haemolytic cocci

Twenty-seven isolates of α-haemolytic cocci were characterised. Genotyping identified 21 of the isolates as Aerococcus viridians, three as Streptococcus alactolyticus, one as S pluranimalium and one as Vagococcus lutrae. One phenotypic identification as Aerococcus viridans had an unacceptable number of mismatches among the 456 base pairs to all GenBank+EMBL+DDBJ+PDB sequences. Aerococcus suis was not isolated.


At slaughter, no lesions indicative of either pneumonia or pleurisy were observed, though a mild to severe alveolitis dominated by mononuclear leucocytes was evident proportional to the concentration of ammonia. Similar changes were observed in pigs exposed to both ammonia and AHC, with the addition of monocytic inflammatory cells at the higher concentrations of ammonia.

Immune responses

Technically but necessarily, the treatment was confounded by block, but we were not aware of any differences between blocks other than the expected treatment effects.

The authors detected immune-system activation in the pigs in response to the challenges with AHC and NH3. The interaction between AHC and NH3 accounted for 66 per cent of the increase in LPI at day 14 (Table 1). LPI responded in a progressive manner with increasing NH3 concentration (Fig 1e). The interaction between AHC and LPI at day 0 accounted for 19 per cent of the variation in LPI at day 14, while complex interactions between NH3, LPI and CD4 and NH3, LPI and CD21, including a quadratic effect of LPI, were highly significant, though only explaining a relatively small amount (14 per cent) of the variation in LPI at day 14. These complex interactions have not been presented and the final model was corrected for the effects at day 0.

Fig 1

(a) Mean (se) VFI (voluntary feed intake), adjusted for immune effects at day 0, for AHC+ pigs (inoculated with α-haemolytic cocci) and AHC– pigs exposed to ammonia at different concentrations. (b) Mean (se) ADG (average daily gain), adjusted for immune effects at day 0, for AHC+ and AHC-pigs exposed to ammonia at different concentrations. (c) Predicted mean (se) VFI (voluntary feed intake) for AHC+ and AHC- pigs exposed to ammonia at different concentrations. (d) Predicted mean (se) ADG (average daily gain) for AHC+ and AHC- pigs exposed to ammonia at different concentrations. (e) Predicted mean (se) LPI (lymphocyte proliferation index) at day 14 for AHC+ and AHC-pigs exposed to ammonia at different concentrations. (f) Predicted mean (se) HPP (heterophil phagocytic potential) at day 14 for AHC+ and AHC- pigs exposed to ammonia at different concentrations.

Table 1

Between subjects effects on LPI (lymphocyte proliferation index)

As for LPI, the interaction between AHC and NH3 accounted for >60 per cent of the variation in HPP; again, the effect was progressive with increasing NH3 (Table 2, Fig 1f). At day 14, this immune parameter was affected by its level at day 0 interacting with AHC treatment, explaining 19 per cent of the variation in HPP. Similarly, complex interactions between NH3 and HPP at day 0 explained a small but significant amount of variation in HPP at day 14. These complex interactions have not been presented, though the final model was corrected for variation at day 0.

Table 2

Between subjects effects on HPP (heterophil phagocytic potential)

Growth rate and feed utilisation

The results, adjusted for immune parameters at day 0 (Fig 1a, b), demonstrate that the combination of ammonia and AHC adversely impacted VFI and ADG in a dose-dependent manner, resulting in a progressive increase in FCR. Adjusting the model for immune effects at day 14 showed that there was an effect of AHC explaining part of the response in VFI (Table 3) but no significant interactions between ammonia and AHC (Fig 1c, d); rather, interactions between each of the factors and the covariate measures of immune function affected the response variables (Tables 3 and 4). Hence, the effects on VFI and ADG in pigs exposed to AHC and ammonia were primarily due to the immune responses between days 0 and 14.

Table 3

Between subjects effects on voluntary food intake (VFI)

Table 4

Between subjects effects on ADG (average daily gain)

The predicted decrease in VFI due to inoculation with AHC alone was 51 per cent of the observed decrease in the pigs inoculated with AHC while exposed to 50 ppm of ammonia, and the actual FCR in these pigs was 4.58, compared with unexposed controls with FCR of 3.29. The model predicted that the direct effect of AHC in the presence of 50 ppm ammonia contributed 44.7 per cent, and the immune responses 55.3 per cent, of the difference in FCR.


The airborne bacteria in the Roseworthy pig-sheds were not harmless commensals in these pigs. Although there was no clinical expression of disease, both feed intake and growth rates were reduced as a function of the immune response elicited by the inoculation of AHC onto respiratory epithelia exposed at the time to ammonia.

There is an increasing body of evidence that some species within the AHC group may have pathogenic impacts. In human beings, VGS usually act as commensals utilising mucin as an energy substrate (Van der Hoeven and Camp 1991), but may act as periodontal pathogens (Robertson and Smith 2009) and may cause rhinosinusitis (Hwang and Tan 2007). They are also common secondary colonisers in the distal airways of people with chronic lung diseases (Cabello and others 1997). The factors that make the organisms pathogenic are not known (Hwang and Tan 2007), but it is known that VGS represent a particular risk to human beings with neutropenia (Tunkel and Sepkowitz 2002), and that in pigs they readily colonise the aortic valve following mechanical damage to the valve, resulting in endocarditis (Ramirezronda 1978). A viridans may also colonise heart valves, although it is very rare as a cause of clinical disease in human beings (Popescu and others 2005). A viridans was first described as a common airborne organism in human-occupied places (Williams and others 1953) and it has also been found in human faecal slurry (Budzinska and others 2009). It is generally considered a saprophyte (Park and others 2002). However, A viridans has been isolated in pure culture from 11.5 per cent of a sample of pigs with arthritis, 2.2 per cent of pigs with meningitis, and 1 per cent of pigs with pneumonia, apparently as an opportunistic pathogen (Martin and others 2007). Vagococcus species have been isolated from carcases of pigs condemned because of pathological changes initially attributed to swine erysipelas and also from field cases with a presumptive diagnosis of swine erysipelas (Bender and others 2009) and V fluvialis has been isolated from a number of pigs with clinical disease (Pot and others 1994, Teixeira and others 1997), but the clinical significance of the Vagococcus species have yet to be determined in pigs. Vagococcus elongatus has been isolated from a pig effluent pit (Lawson and others 2007).

Subclinical impacts are also recorded. VGS has been previously described in human lungs (Hanage and Cohen 2002), evidenced by upregulation of adhesion molecules and neutrophil aggregation, and A viridans has been isolated from the milk of cows with subclinical mastitis (Devriese and others 1999). Subclinical disease has been shown to have an impact on the growth rate of calves (Forbes and others 2002), and was evident in the pigs in the present study, in which there was no clinical disease but feed intake and growth (VFI and ADG) were adversely affected. In those pigs exposed to AHC but not ammonia, the daily feed loss was 268 g of feed per inoculated pig per day, indicating that 11.0 per cent of the feed was lost to subclinical disease.

AHC may have affected VFI through central appetite suppression. VGS are known to stimulate monocytes to rapidly produce very large amounts of interleukin (IL)-1β (Hanage and Cohen 2002, Hahn and others 2007), which could be expected to impact VFI because IL-1β acts directly on the brain to suppress appetite (DeBoer and others 2009). However, no significant direct impact on ADG was detected.

The responses of lymphocytes and heterophils to exposure to the combination of AHC and NH3 demonstrated that the immune system was activated by supposedly commensal bacteria, and that immune activation was associated with reduced ADG, indicating that the impact of AHC on ADG arose because of diversion of nutrients to the immune system. The present findings are consistent with those of studies that have compared production parameters in unhygienic sheds with the same measures of immune stimulation (LPI and CD4, Galina-Pantoja and others 2006, CD21, Clapperton and others 2005b, 2008); the difference with the present study was that the sheds were hygienic and the air pollution in the present study was artificial, controlled and applied individually.

Measures of immune markers at day 0 were quite variable, demonstrating different levels of immune-system activation in the pigs at the start of the trial. The data revealed a complex suite of interactions between the marker covariates at day 0 that accounting for about a third of the variation in both LPI and HPP at day 14. In particular, the interactions between AHC on the one hand and LPI at day 0 and HPP at day 0 on the other each accounted for 19 per cent of the respective responses on day 14. However, it was not the purpose of this study to dissect these complex relationships, and so the model was adjusted to remove the effects of immune status at day 0.

In the pigs, NH3 exacerbated the impact of AHC on immune parameters in a substantial and progressive manner. A similar interaction was observed by Hamilton and others (1999) who examined the impact on the turbinates of pigs exposed to a continuous supply of ammonia before inoculation with Pasteurella multocida. There are several mechanisms that might explain such synergistic interactions. Ammonia could facilitate the impact of a microorganism residing on the nasal mucosa through its ability to breach protective mucous and epithelial barriers (Brautbar 1998), allowing penetration by the microorganism into subepithelial tissues, which is consistent with the history of VGS as an invader secondary to other agents of epithelial damage (Johnson and Bowie 1992, Cabello and others 1997, Hanage and Cohen 2002), leading to immune stimulation. Another mechanism could be the affinity of VGS to fibronectin, which selectively promotes the attachment of VGS to oral epithelial cells (Sinner and Tunkel 2010); but fibronectin is also secreted by endothelial cells, platelets and fibroblasts in response to vascular injury (Sinner and Tunkel 2010), and so it may be that ammonia is capable of eliciting a fibronectic response by the vascular tissues that in some way enhances the presentation of VGS to the immune system. The pathogenesis of Aerococcus species is much less well understood, but an aggregation of platelets and fibrin in response to Aerococcus urinae has been noted (Shannon and others 2010). A third explanation for the synergistic effect might be that the relative infective dose of the organisms is increased if they are able to exploit ammonia as a source of nitrogen, and many Streptococcus species have this capability in vitro, for example, S bovis (Atasoglu and Wallace 2002); S thermophilus (Monnet and others 2005) and S mutans, (St Martin and Wittenberger 1980). An increased effective dose would likely result in greater immune stimulation.

The daily feed loss due to the disease process for each pig inoculated with AHC and exposed to 50 ppm ammonia was 546 g or 28.1 per cent of the feed intake. This is higher than the 15.4 per cent penalty that may be calculated from data presented by Williams and others (1997) for 102 kg pigs from an Iowa herd in which the recognised pathogens Actinobacillus pleuropneumoniae, Mycoplasma hyopneumoniae, swine influenza virus, and transmissible gastroenteritis virus are endemic, but whether the losses in the pigs would have been sustained over an equally long time scale, is not known. It has been opined (Clapperton and others 2005a) that associations between performance and immune traits can be attributed to the nutrient demands of subclinical diseases, such as subclinical enteric Salmonella infection (Galina-Pantoja and others 2006), that divert energy from growth. Maintaining an immune response is known to have a high energy demand, and is partly attributable to thermogenesis, establishing and maintaining fever, that requires a 10 per cent increase in basal metabolic rate for each degree rise in body temperature in human beings (Kluger 1991), particularly in species with poor insulation (Hart 1988). Energy demand may also be attributed to the energetic costs of cell-based immunity, although Klasing (2004) maintained that these energetic costs are relatively low. The present study findings disagree with that hypothesis; day 14 measures of LPI, HPP and CD21 accounted for much of the variation in ADG.

Viridans-group streptococci are present in the faeces of pigs in Australia (Skirrow and others 1995), and Aerococcus viridians has been found in porcine faecal slurries elsewhere (Guo and others 2007, Budzinska and others 2009, Byrne-Bailey and others 2009). Vagococcus elongatus has also been isolated from a piggery manure storage pit (Lawson and others 2007). With poor hygiene, higher concentrations of airborne AHC would be expected, and the authors suggest that it is an explanation for the well-documented impact of poor shed hygiene on production parameters and economic performance (Le Floc'h and others 2009). Ammonia is released from slurries of urine and faeces (Groot Koerkamp and others 1998) and so atmospheric ammonia is another consequence of poor shed hygiene (Banhazi and others 2008).

The authors conclude that air polluted with AHC causes subclinical disease in pig-sheds with poor hygiene, and that the effect is exacerbated by the atmospheric ammonia prevalent in these sheds, which together activate the pigs' immune systems. Further, the authors conclude that there is a substantial nutrient demand inherent in the immune response to organisms such as AHC that are generally thought to be commensal.


The study was supported by the Pig Research and Development Corporation (PRDC), now Australia Pork Limited (APL). For assistance with analyses, the authors thank Andrew Bean, Matthew Bruce and Vijaya Janardhana at the Australian Animal Health Laboratory; Rachel Pratt and Mary Barton at the University of South Australia and Lance Mickan and Alan Goodwin of the South Australian Department of Health. The project was approved the Animal Ethics Committee at The University of Adelaide (Approval Number W-53–1997).


View Abstract


  • Provenance not commissioned; externally peer reviewed

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.