Herd-level infectious disease surveillance of livestock populations using aggregate samples

Abstract All sectors of livestock production are in the process of shifting from small populations on many farms to large populations on fewer farms. A concurrent shift has occurred in the number of livestock moved across political boundaries. The unintended consequence of these changes has been the appearance of multifactorial diseases that are resistant to traditional methods of prevention and control. The need to understand complex animal health conditions mandates a shift toward the collection of longitudinal animal health data. Historically, collection of such data has frustrated and challenged animal health specialists. A promising trend in the evolution toward more efficient and effective livestock disease surveillance is the increased use of aggregate samples, e.g. bulk tank milk and oral fluid specimens. These sample types provide the means to monitor disease, estimate herd prevalence, and evaluate spatiotemporal trends in disease distribution. Thus, this article provides an overview of the use of bulk tank milk and pen-based oral fluids in the surveillance of livestock populations for infectious diseases.


Introduction
Globally and locally, achieving the control of historically impactful infectious diseases of livestock continues to frustrate producers and challenge animal health specialists. A core requirement of a successful control program is the ongoing collection of disease data from populations. Schwabe (1982) describes this as the process of establishing baseline levels 'against which effects of intervention (control) efforts can be measured'.
The ongoing burden of disease in endemic areas and the expansion of infectious agents into previously free areas expose the frailty of current surveillance and response/control programs (Backer et al., 2009;Lee, 2015;Saeed et al., 2015;Neira et al., 2017). Foot-and-mouth disease virus (FMDV) was identified in 1897, but 116 years later, endemic FMDV losses were estimated at $6.5-$21 billion dollars annually and only 66 of the 181 (36.5%) OIE-member countries are 'FMD free where vaccination is not practiced' (Longjam et al., 2011;Knight-Jones and Rushton, 2013;OIE, 2017a). Classical swine fever virus (CSFV) was identified in 1903(de Schweinitz, Dorset, 1903, but in 2017, just 32 of the 181 (17.7%) OIE-member countries are considered free of CSFV (OIE, 2017b). This, despite the profound global economic burden of CSFV and the clear benefits of eradication, e.g. the benefit:cost ratio of CSFV eradication in the USA was estimated at ≥13.2 (USDA, 1981;Pinto et al., 2011). Initially identified on the basis of outbreaks of unknown origin in the 1980s, porcine reproductive and respiratory syndrome virus (PRRSV) was isolated in 1991 and has become endemic in most major pork-producing regions of the world (Wensvoort et al., 1991;Zimmerman et al., 2012). Holtkamp et al. (2013) estimated the US pork producers' losses to PRRSV at $664 million annually. Nathues et al. (2017) estimated losses to European producers at €126.79 per sow per year and €3.77 per pig marketed in herds with 'slight' PRRS.
A promising trend in the evolution toward more efficient and effective livestock disease surveillance is the increased use of aggregate samples (Thurmond and Perez, 2006;Strutzberg-Minder et al., 2015;Gibert et al., 2017;Rotolo, et al., 2017). By definition, an aggregate sample represents two or more animals at a specific location and time, e.g. bulk tank milk and pen-based oral fluid samples.
Depending on the governmental standards or ordinances, tanks are agitated for ≥10 min after which samples are collected aseptically from the top of the tank using a sterile pipette, syringe, or sanitized dipper (Bickett-Weddle et al., 2011;U.S. Food and Drug Administration, 2015). Although bulk tank milk samples do not represent dry cows or cows on milk withhold, they provide an economical, convenient, and timely approach for the detection of specific pathogens and/or estimation of herd prevalence (Olde Riekerink et al., 2006;Sekiya et al., 2013;Lanyon et al., 2014;Collins et al., 2017). Economically significant pathogens detectable in bulk tank milk samples and reported in the refereed literature are discussed below and listed in Table 1.

aureus
Olde Riekerink et al. (2006Riekerink et al. ( , 2010 S. agalactiae Keefe (1997) in milk, but antibodies to SBV can be detected in an individual cow and bulk tank milk samples using commercial indirect enzyme-linked immunosorbent assay (ELISAs) (Balmer et al., 2014;Johnson et al., 2014;Daly et al., 2015). Although test performance estimates are not available (diagnostic sensitivity, diagnostic specificity), results of bulk tank milk ELISA testing were predictive of within-herd seroprevalence and herd immunity (Collins et al., 2017). Analyses based on bulk tank milk testing results have been used to assess the spatial distribution, rate of spread, direction of the spread, and effect of farm altitude on the prevalence of SBV (Balmer et al., 2014;Johnson et al., 2014).
Persistently infected (PI) animals, the result of fetal infection during the first trimester of pregnancy (immunotolerance), serve as a continuous source of infection (Houe, 1999;Fray et al., 2000;Renshaw et al., 2000). PI cows produce little-to-no BVD antibody, but continuously shed real-time reverse transcription polymerase chain reaction (RT-rtPCR)-detectable levels of BVDV in milk (Radwan et al., 1995;Kramps et al., 1999;Houe, 1999;Renshaw et al., 2000). Drew et al. (1999) reported 100% diagnostic sensitivity and specificity for PCR-based detection of BVDV RNA in bulk tank milk samples from herds with PI cows.
Strategically, antibody detection is used to identify the herds with circulating BVDV, and nucleic acid detection is used to identify the herds with PI cattle (Lanyon et al., 2014). Monitoring changes in antibody prevalence has been used to determine whether a BVDV infection is ongoing or recent (Lanyon et al., 2014). ELISA testing has also been used to monitor declining antibody levels after removal of PI cattle (Houe, 1999).

Border disease virus
First reported in England and Wales in 1958 and closely related to BVDV, border disease virus (BDV) is a pestivirus of ovine and caprine species (Nettleton et al., 1998). BDV is transmitted through direct contact or transplacentally, with infection during early pregnancy resulting in PI offspring (Garcia-Perez et al., 2010). Goats are susceptible to BDV, but infection is rare and typically results in abortion (Nettleton et al., 1998). In sheep, clinical signs of BDV include abortion, stillbirths, and nonviable lambs.
As in the case of BVDV, PI animals shed BDV continuously and do not produce antibodies. Bulk tank milk samples can be tested for BDV by RT-rtPCR; however, estimates of diagnostic performance have not been reported (Berriatua et al., 2006). Immunocompetent animals produce antibodies detectable in bulk tank milk (Garcia-Perez et al., 2010). In one study, the diagnostic sensitivity and specificity of a blocking ELISA for BDV detection in bulk tank milk samples was reported as 100 and 85.2%, respectively (Corbiere et al., 2012). A high seroprevalence of BDV in lactating animals suggests the presence of PI animals (Berriatua et al., 2006). Thus, ELISA testing of bulk tank milk samples provides the means to estimate the prevalence of BDV in flocks and may indirectly reveal the presence of PI animals (Berriatua et al., 2006;Garcia-Perez et al., 2010).

Foot-and-mouth disease virus
FMDV is a highly impactful picornavirus of cloven-hoofed animals (Reid et al., 2006;Thurmond and Perez, 2006;Knight-Jones and Rushton, 2013). FMDV can be transmitted by direct or indirect contact (Bravo de Rueda et al., 2014). Clinical signs of FMDV infection include vesicular lesions, decrease in milk yield in lactating cattle, and pyrexia (Armstrong and Mathew, 2001).
FMDV was detected in milk samples from individual cows by RT-rtPCR for 23 days post-inoculation (Reid et al., 2006). Estimates of the diagnostic sensitivity and specificity of RT-rtPCR for the detection of FMDV in bulk tank milk samples has not been reported, but Thurmond and Perez (2006) predicted that RT-rtPCR testing of bulk tank milk samples would detect FMDV 4-7 days earlier than detection based on the recognition/reporting of clinical signs.
FMDV antibodies may be detected in ovine and bovine milk using blocking ELISAs (Armstrong, 1997a(Armstrong, , 1997b. Estimates for diagnostic sensitivity and specificity of these ELISAs are not available, but Armstrong and Mathew found a statistically significant correlation (r = 0.53) between serum and milk antibody titers (Armstrong and Mathew, 2001). On this basis, these researchers suggested that antibody testing of bulk tank milk samples would be an effective approach for monitoring herd immunity and/or evaluating population susceptibility to FMDV.

Mycobacterium avium subspecies paratuberculosis
Mycobacterium avium subspecies paratuberculosis (MAP) is the etiologic agent of Johne's disease in ruminants (Mortier et al., 2014). Most commonly acquired via fecal-oral transmission, Johne's disease is characterized by enteritis, decreased milk yield, weight loss, diarrhea, and death (Wilson et al., 2010;Mortier et al., 2014). A causal role for MAP in Crohn's disease has been postulated, but was neither confirmed nor rejected by an assessment of the available data (Feller et al., 2007).
MAP is detectable in milk via culture and PCR testing, but culture of bulk tank milk samples is not practical because the procedure is neither diagnostically sensitive nor timely, i.e. culture can take 18-52 weeks (Slana et al., 2008). The most common target of PCR assays is multiple copy insertion sequence IS900 in the MAP genome (Slana et al., 2008). The analytical sensitivity of the IS900 PCR is reported as 5-6 MAP cells ml −1 of bulk tank milk versus 83 MAP cells ml −1 for a PCR targeting F57. However, IS900 PCRs may have issues with analytical specificity because of the homology of this region across mycobacteria species (Cousins et al., 1999;Tasara et al., 2005;Slana et al., 2008). Jayaro et al. (2004) reported a diagnostic sensitivity of 21% and diagnostic specificity of 50% for bulk tank milk samples using an IS900 PCR. No estimates of diagnostic sensitivity and specificity are available for F57-based PCRs.
ELISA-detectable MAP antibodies are present in bulk tank milk samples, but interpretation of testing results has not been clearly established (Nielsen et al., 2000;van Weering et al., 2007;Wilson et al., 2010;Beaver et al., 2016). Regardless, some researchers believe that ELISA testing of bulk tank milk samples can be used effectively by monitoring changes over time (van Weering et al., 2007;Beaver et al., 2016). Alternatively, Beaver et al. (2016), suggested the concurrent use of both assays for bulk tank milk monitoring programs for MAP (Beaver et al., 2016). Thus, herds with positive PCR results and high ELISA titers reflected active infection; whereas, herds with positive PCR results but low ELISA titers reflected environmental contamination (Beaver et al., 2016).

Coxiella burnetii (Q fever)
Coxiella burnetii is an obligate, intracellular rickettsial organism and the cause of Q fever in animals and humans (Kim et al., 2005). Infection with C. burnetii results in reproductive disease, including metritis and infertility in cattle and abortion in goats and sheep (Kim et al., 2005;Rodolakis et al., 2007). Shedding patterns of C. burnetii in milk is species-dependent and varies among cattle, sheep, and goats (Rodolakis et al., 2007). Cattle shed C. burnetii in milk for several months, goats shed for a shorter time, and sheep do not reliably shed in milk (Rodolakis et al., 2007;Astobiza et al., 2012). Antibody to and nucleic acids of C. burnetii are detectable in bulk tank milk samples with ELISA and PCR, respectively (Rodolakis et al., 2007;van den Brom et al., 2012). Muskens et al. (2011) reported diagnostic sensitivity and specificity of 82 and 70%, respectively, when testing bulk tank milk samples by a commercial real-time PCR. The diagnostic sensitivity and specificity of a commercial C. burnetii antibody ELISA for bulk tank milk was reported as 88.2 and 94.6%, respectively, using manufacturer-recommended cutoffs (van den Brom et al., 2012). When used in combination, ELISA testing of bulk tank milk samples can be used to determine herd exposure and estimate prevalence of C. burnetii, while PCR testing can be used to determine shedding and prevalence (Muskens et al., 2011;Astobiza et al., 2012).

Detection of bacterial pathogens associated with mastitis
Streptococcus agalactiae is a highly contagious, obligate pathogen of the bovine mammary gland and a cause of subclinical and clinical mastitis (Keefe, 1997;Phuektes et al., 2003;Olde Riekerink et al., 2006;Mweu et al., 2012). Streptococcus agalactiae may be detected in bulk tank milk samples by culture or PCR (Keefe, 1997;Phuektes et al., 2003). As reviewed by Phuektes et al. (2003), estimates of the diagnostic sensitivity of culture range from 20 to 84%. Estimates of the diagnostic sensitivity and specificity are not available, but as would be expected, testing multiple bulk tank milk samples was shown to increase the likelihood of detecting S. agalactiae by PCR (Phuektes et al., 2003;Soltau et al., 2017). ELISA-detectable S. agalactiae antibodies have been reported in individual milk samples, but this approach has not been evaluated for bulk tank milk testing (Logan et al., 1982).
Staphylococcus aureus is an opportunistic pathogen and a cause of subclinical and clinical mastitis in cattle, sheep, and goats (Olde Riekerink et al., 2006;Haran et al., 2012;Zanardi et al., 2014;Merz et al., 2016). As reviewed by Olde Riekerink et al. (2010), culture of bulk tank milk for S. aureus had an estimated diagnostic sensitivity of 21-42% and a diagnostic specificity of 100%. Repeated sampling is recognized to improve the probability of detection by culture (Olde Riekerink et al., 2006Riekerink et al., , 2010. PCR testing of bulk tank milk samples can be used to detect S. aureus, estimate herd prevalence of the infection, and assess for the presence of methicillin-resistant strains (Haran et al., 2012). The diagnostic sensitivity and specificity of PCR testing for S. aureus in bulk tank milk samples is reported at 99 and 67%, respectively (Zanardi et al., 2014). Using individual milk, ELISA testing for antibodies against S. aureus may be used to as a screening tool to detect infected animals (Fox and Adams, 2000).
Mycoplasma bovis is a highly pathogenic mycoplasma causing both mastitis and respiratory disease in adult cattle (Parker et al., 2017a). Mycoplasma bovis is detectable in bulk tank milk samples by culture, but the assay can take 7-10 days and overgrowth of bacteria is problematic (Parker et al., 2017a(Parker et al., , 2017b. The diagnostic sensitivity of M. bovis culture is reported as 50%, with diagnostic specificity estimates as high as 100% (Justice-Allen et al., 2011;Maunsell et al., 2011). The diagnostic sensitivity and specificity of M. bovis PCR for individual milk samples is reportedly 100 and 99.3%, respectively, but estimates of PCR performance for bulk tank milk samples have not been reported (Cai et al., 2005). PCR testing allows for more rapid detection of M. bovis versus culture and herd prevalence estimates can be extrapolated from the results (Arcangioli et al., 2011). A commercial antibody ELISA is available for bulk tank milk testing, and estimates for diagnostic sensitivity and specificity are 60.4 and 97.3%, respectively (Nielsen et al., 2015). The combination of PCR and ELISA testing can reveal M. bovis infection in a herd and is an effective approach for surveillance (Nielsen et al., 2015).

Oral fluid samples
Oral fluids are collected from swine or cattle by providing access to a rope suspended in the pen, then recovering the sample for diagnostic testing (Smith et al., 2004;Prickett et al., 2008aPrickett et al., , 2008bStanford et al., 2009;. Oral fluid samples are an aggregate sample composed of saliva and transudate originating from capillaries within the buccal and gingival mucosa (Prickett et al., 2008a). Oral fluids contain both local and serumderived antibodies and pathogens (Prickett et al., 2008a(Prickett et al., , 2008b. In addition, viruses, bacteria, and other test analytes in feed, water, or the environment may be present in oral fluids as a result of normal exploratory behavior (Kittawornrat and Zimmerman, 2011;Johnson et al., 2012). This explains the detection of porcine epidemic diarrhea virus (PEDV) in swine oral fluid samples and Escherichia coli and salmonella in cattle (Smith et al., 2005a(Smith et al., , 2005bRenter et al., 2008;Bjustrom-Kraft et al., 2016). In cattle, oral fluids have been used in the observational studies in feedlot cattle (Renter et al., 2008;Smith et al., 2005aSmith et al., , 2005b), but have not been routinely utilized in surveillance. In contrast, oral fluids have been used extensively for disease surveillance in swine populations. Therefore, the remainder of this section will focus exclusively on this subject.
Oral fluids can be collected from groups or individual pigs (White et al., 2014;Pepin et al., 2015aPepin et al., , 2015b. In group-housed animals, oral fluids offer a higher probability of detection with fewer samples when compared with individual serum samples . Sampling guidelines for oral fluid collection at the barn or site level have been published (Rotolo et al., 2017).
Diagnostic assays optimized for swine oral fluid specimens have been available in North American veterinary diagnostic laboratories since 2010 Bjustrom-Kraft et al., 2018). In three North American swine-interest veterinary diagnostic laboratories, the number of oral fluid tests performed increased from 20,963 in 2010 to 369,439 in 2016 (Bjustrom-Kraft et al., 2018). Pathogens detectable in oral fluid samples and reported in the refereed literature are listed in Table 2. Selected pathogens are reviewed below.

Foot-and-mouth-disease virus
Rapid screening of swine herds is critical in the control of FMDV because pigs aerosolize a large amount of virus compared with cattle and promulgate virus transmission (Stenfeldt et al., 2016). Under experimental conditions, FMDV was isolated from swine oral fluids on day post-inoculation (DPI) 1-5 (Senthilkumaran et al., 2017). By RT-rtPCR, FMDV was detected from one DPI, i.e. prior to the appearance of clinical signs, and up to 21 DPI (Mouchantat et al., 2014;Senthilkumaran et al., 2017). RNA was detected in oral fluids one day earlier than oral or nasal swab samples and continued ∼7 days longer (Senthilkumaran et al., 2017). A field-deployable reverse transcription-insulated isothermal PCR has also been used to detect FMDV RNA in oral fluids (Ambagala et al., 2016). FMDV antigens were detected in oral fluids 1-6 DPI using lateral flow immunochromatographic strip tests and 2-3 DPI using a double-antibody sandwich ELISA (Senthilkumaran et al., 2017). FMDV IgA was detected in oral fluids using a solid-phase competitive ELISA beginning at 14 DPI (Senthilkumaran et al., 2017). Pacheco et al. (2010) were not successful in detecting FMDV IgM or IgG in oral fluid samples. Estimates of diagnostic sensitivity and specificity have not been reported for the assays reported in this paragraph. Although FMDV oral fluid assay development is in its early stages, preliminary results support the use of nucleic acid and antibody detection as a method to rapidly screen herds (Ambagala et al., 2016;Senthilkumaran et al., 2017).

Classical swine fever virus
CSFV is a pestivirus with significant economic consequences resulting from clinical disease, lost export markets, and costs related to control and eradication efforts (Fernández-Carrión et al., 2016). CSFV can be transmitted by direct or indirect contact and, depending on the virulence of the strain, causes pyrexia, anorexia, lethargy, conjunctivitis, enlarged and discolored lymph nodes, constipation, and diarrhea in affected pigs (Moennig et al., 2003;Petrini et al., 2017). Under experimental settings, CSFV was detected in oral fluids by RT-rtPCR from seven up to 30 DPI, with a higher detection rate in oral fluid than blood samples (40 vs 28%) (Dietze et al., 2017;Petrini et al., 2017). Estimates of diagnostic sensitivity and specificity have not been reported for these assays and, overall, research on CSFV oral fluid diagnostics is in its initial phases.

African swine fever virus
Infection with African swine fever virus (ASFV), the only member of family Asfarviridae, is a cause of fever, hemorrhage, and mortality in domestic and feral pigs (Sanchez-Vizcaino and Neira, 2012;Guinat et al., 2014;Gimenez-Lirola et al., 2016). Transmitted through direct and indirect contact, ASFV is of particular concern because, since its introduction into the Democratic Republic of Georgia in 2007, it has steadily advanced westwardly into Europe via feral swine and threatens to spread eastwardly into China (Guinat et al., 2014;Vergne et al., 2017).
Under experimental conditions, ASFV was detected in oral fluid 3-5 DPI by PCR (Guinat et al., 2014;Grau et al., 2015). ASFV antibodies were detected at 11 DPI in individual oral fluid samples by indirect ELISA under experimental conditions (Mur et al., 2013). The pattern of antibody response in oral fluids was similar to the pattern seen in serum (Mur et al., 2013). ASFV antibodies were also detected using a p30-based indirect ELISA in oral fluids (Gimenez-Lirola et al., 2016). Diagnostic sensitivities and specificities for these assays have not been reported. As in the cases of FMDV and CSFV, further studies are needed to optimize ASFV oral fluid assays and assess their use in the field (Grau et al., 2015).

Porcine reproductive and respiratory syndrome virus
PRRSV is an arterivirus transmitted through direct and indirect contact . Clinical signs of PRRSV vary based on the age of the pig and the virulence of the isolate. In sows, clinical signs include abortion, stillbirths, anorexia, and mortality . PRRSV is often an etiological component of the porcine respiratory disease complex in growing pigs .
IgG, IgA, and IgM antibody isotypes were detected in oral fluids collected from individual boars using a commercial PRRS serum antibody indirect ELISA modified for oral fluids . The pattern of PRRSV antibody ontogeny was similar in serum and oral fluid, with IgM detected in oral fluids at three DPI, IgA at seven DPI, and IgG at eight DPI . Commercial PRRSV oral fluid ELISAs have since become available. Antibodies were also detected in oral fluid using a fluorescent microsphere immunoassay with a reported diagnostic sensitivity of 92% and diagnostic specificity of 91% (Langenhorst et al., 2012).
Testing of oral fluids can be used to assess the effectiveness of PRRSV control and elimination programs (Biernacka et al., 2016;Rotolo et al., 2017). A distinct advantage of PRRSV oral fluid-based surveillance is that pen-based oral fluid sampling provides a higher probability of detection than individual animal sampling using either RT-rtPCR or ELISA .

Influenza A virus
Influenza A virus (IAV) is an orthomyxovirus of human beings, horses, sea mammals, birds, and pigs, transmitted via direct and indirect contact (Hughes et al., 2015;Neira et al., 2016). IAV in commercial swine herds results in chronic, endemic infection with respiratory or reproductive clinical signs, as well as clinically inapparent infections Panyasing et al., 2013). IAV is an important pathogen to surveil in pigs because of its zoonotic potential (Vincent et al., 2014;Hughes et al., 2015).
Under experimental conditions, IAV RNA was detected in swine oral fluids by one DPI and up to 69 DPI (Allerson et al., 2014;Decorte et al., 2015). Decorte et al. (2015) reported the duration of detection in oral fluids as 14 days longer than detection in nasal swabs by RT-rtPCR (Decorte et al., 2015). Compared with individual nasal swabs, the diagnostic sensitivity and specificity of pen-based oral fluid RT-rtPCR testing was estimated at 80 and 100%, respectively (Romagosa et al., 2012). Although further optimization is necessary, IAV has also been isolated from oral fluids . Sequencing of IAV from oral fluids has been reported (Panyasing et al., 2014a). RT-rtPCR testing of oral fluids can  Culture or isolation Influenza A virus Goodell et al. (2013) be used to track viral circulation and to monitor the effect of vaccination and control programs in commercial swine herds . Panyasing et al. (2013) reported the ontogeny of IAV IgM, IgA, and IgG in pigs housed under experimental conditions, using isotype-specific indirect ELISAs. Serum and oral fluid IgG responses were highly correlated (r = 0.80) . Detection of IAV antibody has also been reported using blocking or competitive ELISA formats (Panyasing et al., 2014b;Strutzberg-Minder et al., 2015). Diagnostic sensitivity and specificity estimates have not been established for these assays. Antibody detection in oral fluids allows for the detection of IAV infection in the absence of clinical signs .

Coronaviruses
PEDV is an enteric coronavirus transmitted via the fecal-oral route (Crawford et al., 2015;Bjustrom-Kraft et al., 2016). Clinical signs of PEDV infection in swine include watery diarrhea, vomiting, and mortality in neonates (Bjustrom-Kraft et al., 2016). In the field, Bjustrom-Kraft et al. (2016) reported the detection of PEDV nucleic acid in oral fluids from 6 days post-exposure (DPE) to 69 DPE. PEDV was detected 15 days longer in oral fluid samples compared with pen fecal samples, and, compared with individual rectal swabs, oral fluids demonstrated a higher concentration of detectable virus and higher rate of detection. In the same study, Bjustrom-Kraft et al. (2016) reported the detection of PEDV antibody (IgG and IgA) by 13 DPE in oral fluids. The diagnostic sensitivity and specificity of a PEDV IgG oral fluid ELISA was reported as 69 and 97%, respectively. In contrast, the diagnostic sensitivity and specificity of a PEDV IgA oral fluid ELISA were reported as 100 and 100%, respectively (Bjustrom-Kraft et al., 2016). Although estimates of diagnostic sensitivity and specificity have not been reported, the oral fluid RT-rtPCR is an effective tool to monitor for PEDV presence in herds, and IgA antibody testing offers an effective method to evaluate herd-level immunity (Bjustrom-Kraft et al., 2016).
Like PEDV, porcine deltacoronavirus (PDCoV) is an enteric coronavirus that causes diarrhea and vomiting in pigs (Homwong et al., 2016). PDCoV can be detected in oral fluids by RT-rtPCR, although estimates of diagnostic sensitivity and specificity are not available (Sinha et al., 2015;Homwong et al., 2016;Zhang, 2016). Homwong et al. (2016) reported that the detection of PDCoV nucleic acid in oral fluids was 1.89 times more likely than detection in feces. PDCoV antibody ontogeny in serum and oral fluids has not yet been reported.

Discussion
Globally, the production of livestockpoultry, cattle, and swine is in the process of shifting from small populations on many farms to large populations on fewer farms (Hoban et al., 1997;Marquer, 2010;Barkema et al., 2015;Gale, 2017). Readily accessible USDA data from the dairy and swine industries highlight this trend. In 1982, ∼275,000 US dairy farms housed ∼11,000,000 dairy cows. By 2012, the number of dairy farms dropped to ∼64,000, while animal numbers remained relatively stable at ∼9,250,000 (USDA, 2014). Pork production has followed the same trend. In 1982, ∼330,000 US farms housed ∼55,000,000 pigs in 1982. By 2012, the number of farms with pigs declined to ∼63,000, while the number of pigs increased to ∼66,000,000 (USDA, 2014). Increases in herd size are important to disease control because herd immunity becomes more difficult to achieve as population increases, which in turn leads to pathogen endemicity (LeBlanc et al., 2006;Pitzer et al., 2016).
Over the same time period, a shift occurred in the movement of livestock across political boundaries. In 1960, 13,500,000 live cattle crossed US state lines for feeding or breeding purposes (Hennessy et al., 2005). By 2015, this number had risen to 20,500,000 (USDA, 2017). Similarly, ∼2,500,000 pigs were moved across US stateliness in 1960, in contrast to ∼52,500,000 moved in 2016 (Shields and Mathews, 2003;USDA, 2017). Similar patterns have emerged in Europe. For example, Denmark, France, Germany, Italy, the Netherlands, Poland, and Spain cumulatively imported ∼910,000 live pigs and exported ∼937,000 live pigs in 1961 (FAO, 2017). In contrast, these countries imported ∼22,000,000 live pigs and exported ∼27,000,000 in 2013 (FAO, 2017). Trends in livestock movement are important because of the well-established role of animal transport in the spread of disease, e.g. the 2001 FMDV outbreak in the UK and, more recently, spread of PEDV throughout the Western Hemisphere (Davies, 2015;Guinat et al., 2016).
The unintended consequences of changes in the structure and management of livestock populations have manifested themselves in the appearance of multifactorial diseases resistant to traditional methods of prevention and control, e.g. bovine and porcine respiratory disease complexes (Schwabe, 1982;Gardner et al., 2002;Hagglund et al., 2006;LeBlanc et al., 2006;Bochev, 2007;Edwards, 2010;Pitzer et al., 2016). The need to understand complex animal health conditions mandates a shift toward the collection of longitudinal animal health data. New intervention strategies or unanticipated events, e.g. the introduction of an exotic pathogen, can then be evaluated in the context of their impact on baseline values.
Cumulatively, peer-reviewed research supports the conclusion that aggregate samples offer the opportunity to expand the scope of applied surveillance. Testing of bulk tank milk samples provides bovine and small ruminant practitioners and producers the means to monitor disease and estimate herd prevalence and provides animal health researchers the means to evaluate the spatial distribution and rate of disease transmission (Berriatua et al., 2006;Garcia-Perez et al., 2010;Balmer et al., 2014;Johnson et al., 2014;Collins et al., 2017). Swine oral fluids offer a more analytically sensitive detection system than individual pig samples, and at a lower cost Olsen et al., 2013). Continued progress toward the goal of effective surveillance using aggregate sampling requires research in two areas: (1) continued development and adaption of diagnostic technology for the most globally impactful diseases of animals and human beings (zoonoses); (2) continued development of statistically valid sampling guidelines including probability of detection estimates by sample size, sampling allocation, and frequency of sampling for farm and regional surveillance.