Evidence of viral survival in representative volumes of feed and feed ingredients during long‐distance commercial transport across the continental United States

Abstract The hypothesis that feed ingredients could serve as vehicles for the transport and transmission of viral pathogens was first validated under laboratory conditions. To bridge the gap from the laboratory to the field, this current project tested whether three significant viruses of swine could survive in feed ingredients during long‐distance commercial transport across the continental US. One‐metric tonne totes of soybean meal (organic and conventional) and complete feed were spiked with a 10 ml mixture of PRRSV 174, PEDV and SVA and transported for 23 days in a commercial semi‐trailer truck, crossing 29 states, and 10,183 km. Samples were tested for the presence of viral RNA by PCR, and for viable virus in soy‐based samples by swine bioassay and in complete feed samples by natural feeding. Viable PRRSV, PEDV and SVA were detected in both soy products and viable PEDV and SVA in complete feed. These results provide the first evidence that viral pathogens of pigs can survive in representative volumes of feed and feed ingredients during long‐distance commercial transport across the continental United States.


| INTRODUC TI ON
In 2014, it was first reported that pigs could become infected with porcine epidemic diarrhoea virus (PEDV) following consumption of contaminated feed via natural feeding behaviour (Dee et al., 2014).
Since that time, similar observations have been reported for Seneca virus A (SVA), porcine reproductive and respiratory syndrome virus (PRRSV) and African swine fever virus (ASFV) (Dee et al., 2020a;Niederwerder et al., 2019,). These and other studies have also confirmed that certain feed ingredients, that is, soy-based products, are protective to viruses and enhance their survival for extended periods under simulated conditions of transoceanic shipping. (Dee et al., 2015(Dee et al., , 2016(Dee et al., , 2018Stoian et al., 2020).
In support of these laboratory-based findings, a demonstration project was conducted to evaluate survival of viruses in feed ingredients under real-world shipping conditions (Dee et al., 2020b).
Thirty-gram samples of several feed ingredients feed were spiked with 2-mL mixture of PRRSV 174, PEDV and SVA and transported for 21 days in the trailer of a commercial transport vehicle, crossing 14 states and over 9,741 km. While the study successfully demonstrated that infectious PRRSV, PEDV and SVA were present in both soy products, it possessed inherent limitations as the experimental design did not accurately represent the commercial trucking industry

| INTRODUC TI ON
African swine fever virus (ASFV), the causative agent of an acute haemorrhagic fever of domestic and wild suids, was first identified in East Africa in the early 1900s (Dixon et al., 2019). In the last decade, African swine fever (ASF) has seen a remarkable expansion across Eurasia after an initial re-introduction in Georgia, followed by a spread through Russia and towards Europe (Sánchez-Cordón et al., 2018). Health, 2020). The Eurasian wild boar (Sus scrofa) is endemic to forested areas throughout Southeast Asia. Deforestation and over-hunting have reduced the numbers and diversity of the wild ungulate community, but the resilience of wild boar has allowed them to persist (Gray et al., 2012;Rasphone et al., 2019;Son et al., 2014). Despite their known presence, limited information is available on the distribution, density and ecology of wild boar populations in Southeast Asia (for some existing information see Gray et al., 2012;Guo et al., 2017;Rasphone et al., 2019), and their interface with domestic pigs. While we know that wild boar are susceptible to ASFV (Blome et al., 2013), the virus had yet to be documented in free-ranging populations in this region. Consequently, it is unknown if they are able to maintain the virus in a wild boar-habitat cycle in Southeast Asia as they do in eastern Europe, and if they will become a reservoir for domestic pigs.
In the wake of the ASF epizootic in Viet Nam, Cambodia and Laos, we sought to detect the disease in the wild boar population using cost-effective, risk-based surveillance methods based on participatory engagement of local actors in each country. We conducted a preliminary assessment of the risk of ASFV transmission at the interface between domestic and wild pigs in several pre-selected villages located within or near wild pig habitat. The objectives were to characterize the interface between domestic pigs and wild boar to ascertain the risk of viral spillover from domestic pigs to wild boar, and to determine whether ASF transmission to the wild boar population could be detected in each country. Follow-up site visits and sampling were conducted to identify and document suspected viral spillover.

| Site selection
Investigative teams first identified villages in Laos, Cambodia and Viet Nam with a confirmed ASF outbreak in domestic pigs or with unconfirmed reports of high pig mortality numbers, and in close proximity to forested areas with known or suspected presence of wild boar. These sites were then visited for preliminary assessments of potential interfaces between wild and domestic pigs with risk of ASFV transmission. Based on relevant information obtained from the community members, forest rangers and protected area staff at these sites, additional villages were added to the investigation and visited. Three villages in Houaphanh province in Laos, two in Dong Nai province in Viet Nam and eleven in Ratanakiri province in Cambodia were selected for the assessments which were conducted in July, October/December and April/June 2019, respectively ( Figure 1) and mapped using ArcGIS 10.7.1 (ESRI, Redlands, CA, USA). Sites were not selected following a formal risk assessment, but rather were a convenience sample based on local reports received and field team time availability and resources.

| General approach
Investigative field teams performing site visits consisted of central, provincial, and district-level animal health and environmental sector representatives from local government, as well as field staff and veterinarians from the Wildlife Conservation Society (WCS). In each or the commercial feed industry, since it utilized very small volumes of feed (30g) which did not accurately portray the challenges associated with testing bulk ingredients. In addition, the 30-g samples were inoculated with relatively large volumes of liquid (2ml per sample), and viral viability was only assessed via swine bioassay and an evaluation of viral transmission via natural feeding behaviour was not included.
To address these acknowledged limitations, we conducted a new study to better bridge the gap between the laboratory and the field. The experimental design incorporated several characteristics of the commercial trucking and feed industries, that is, the use of semi-trailer truck and a commercial route of transit, along with the use of larger volumes of feed, which were sampled using a standardized method for the testing of bulk feed. In addition, a viral challenge designed to simulate a 'hot spot' of contamination, as seen with aflatoxin contamination of grain, was used, and both viability and transmission were assessed via bioassay and natural feeding behaviour. The study was based on the hypothesis that certain viruses can survive in select feed and feed ingredients during long-distance commercial transport under real-world conditions.

| Animal care and use
Pigs used in the study were housed in the Pipestone Applied Research biosafety level 2 facility in accordance with the institutional animal care and use guidelines approved by the investigators ethical review board (Pipestone Applied Research IACUC trial number 2021-01).

| Feed preparation
Types of feed used in the study included conventional soybean meal, (1 to 2% fat and 46 to 47% protein), organic soybean meal, (6 to 7% fat and 44 to 45% protein) (Dee et al., 2016(Dee et al., , 2018 and complete grow-finish swine feed. These ingredients were added in bulk to new polypropylene bags, each with a capacity of 1.74 m 3 (National Bulk Bag, Champlin, MN, US), resulting in totes with a final volume of 1-metric tonne per tote. For this study, two 1-metric tonne totes of conventional soybean meal, two 1-metric tonne totes of organic soybean meal and three metric tonnes of complete feed, seven totes in total, were prepared. Totes were then delivered to a dispatching point in Fridley, MN, to prepare for embarkation.

| Tote inoculation
To simulate a 'hot spot' model of feed contamination, 10-mL ice cubes containing a mixture of PRRSV 174, PEDV and SVA at a total dose of 1 x 10 5 TCID 50 per virus was prepared. Each virus was diluted in 30 ml of minimum essential medium (MEM, Sigma-Aldrich, St. Louis, MO, USA) to a concentration of 1 x 10 5 TCID 50 /mL per virus and mixed (three viruses for a total of 90 ml) followed by an addition of 210-mL MEM, to bring the total volume to 300 ml. Ice cubes were prepared by freezing 10-mL aliquots of the mixture in 50-mL conical centrifuge tubes (Corning Inc. Corning, NY, USA) at −80 0 C. Six totes, two containing conventional soybean meal, two containing organic soybean meal and two containing complete feed, were inoculated. The final tote of complete feed was used as a negative control.
To inoculate the six designated totes, a previously filled tote was elevated using a forklift and placed 15 cm directly above an empty tote, with its duffle top held open in a fixed position. The spout bottom of the upper tote was then opened, allowing feed to flow via gravity into the opening of the empty tote. When the lower tote was approximately half full, an ice cube containing the described viral mixture was blindly dropped into the lower tote. The remainder from the upper tote was then added to the lower tote, burying the cube from sight, and the duffle top was tied shut after the lower tote was filled to completion.

| Controls
For controls, twelve 30-g allotments of feed (four conventional soybean meal samples, four conventional organic soybean meal samples and four complete feed samples) were weighed into individual 50-mL mini-bioreactor tubes with vented caps. Six of the twelve samples were individually spiked with a 2-mL aliquot from the viral mixture described previously, to serve as positive controls. The aliquot was injected directly into the centre of each 30-g ingredient sample using a 3-mL syringe with an 18-gauge, 3.81-cm needle. The remaining six samples (30g-feed, no virus) served as negative controls.

| Transport vehicle
To transport the seven totes and control tubes, a commercial semi-trailer truck with a 15.8-m trailer was used (Csp Delivery,  transport vehicle was used to track location, time in transit and distance travelled.

| Details of travel plan
The study utilized a route of delivery representative of the commercial trucking industry which involved travel through 29 US states. The

| Quality control and project oversight
Prior to the study, details of the project were relayed to the Food and Drug Administration Center for Veterinary Medicine (FDA CVM), the United States Department of Agriculture, and to the directors of the respective boards of animal health in states where the truck and trailer were planning to stay overnight. Totes were labelled per FDA CVM instructions stating that feed was 'Not for human or animal consumption/for research use only', and a letter of approval from the agency, co-signed by the PI with respective contact information was carried by the driver during the entire trip. Prior to departure, a barrier was inserted into the trailer to secure the totes within the proximal half of the trailer to minimize the risk of tote movement, spillage, etc., during transport. Other than the seven totes and the controls, no other products were included on the trailer. The truck did not plan to stop anywhere during the transit period, other than for refuelling, meals and hotel stays.

| Bulk sampling and processing
On day 0 and day 23 post-inoculation, the seven totes were sam- the virus had yet to be documented in free-ranging populations in this region. Consequently, it is unknown if they are able to maintain the virus in a wild boar-habitat cycle in Southeast Asia as they do in eastern Europe, and if they will become a reservoir for domestic pigs.
In the wake of the ASF epizootic in Viet Nam, Cambodia and Laos, we sought to detect the disease in the wild boar population using cost-effective, risk-based surveillance methods based on participatory engagement of local actors in each country. We conducted a preliminary assessment of the risk of ASFV transmission at the interface between domestic and wild pigs in several pre-selected villages located within or near wild pig habitat. The objectives were to characterize the interface between domestic pigs and wild boar to ascertain the risk of viral spillover from domestic pigs to wild boar, and to determine whether ASF transmission to the wild boar population could be detected in each country. Follow-up site visits and sampling were conducted to identify and document suspected viral spillover.

| Site selection
Investigative teams first identified villages in Laos, Cambodia and Viet Nam with a confirmed ASF outbreak in domestic pigs or with unconfirmed reports of high pig mortality numbers, and in close proximity to forested areas with known or suspected presence of wild boar. These sites were then visited for preliminary assessments of potential interfaces between wild and domestic pigs with risk of ASFV transmission. Based on relevant information obtained from the community members, forest rangers and protected area staff at these sites, additional villages were added to the investigation and visited. Three villages in Houaphanh province in Laos, two in Dong Nai province in Viet Nam and eleven in Ratanakiri province in Cambodia were selected for the assessments which were conducted in July, October/December and April/June 2019, respectively ( Figure 1) and mapped using ArcGIS 10.7.1 (ESRI, Redlands, CA, USA). Sites were not selected following a formal risk assessment, but rather were a convenience sample based on local reports received and field team time availability and resources.

| General approach
Investigative field teams performing site visits consisted of central, provincial, and district-level animal health and environmental sector representatives from local government, as well as field staff and veterinarians from the Wildlife Conservation Society (WCS). In each handle and six openings (Seedburo Equipment Co., Des Plaines, IL, US). According to protocol, 10 samples were collected from each tote using two 'X' patterns (AAFCO, 2014) and then mixed in a 1-litre plastic bag (Ziploc, S.C. Johnson & Son, Racine, WI, US) to create a single composite sample ( Figure 2). After each tote was sampled, feed dust was expelled from the probe using forced air, sprayed with 70% ethanol, wiped with a clean cloth and the ethanol allowed to evaporate prior to the next sampling. Gloves were also changed between every tote. Following collection, samples from the four soy-based products were processed to prepare inoculums for PCR and bioassay testing. Specifically, each soy-based bulk feed sample collected from its respective tote was mixed with 1,000 ml of sterile phosphate-buffered saline in a 4-litre metal can, the can sealed, inverted, shaken vigorously by hand and then placed on a pneumatic paint shaker (Astro pneumatic tool 4,550, Astro pneumatic tool company, South el Monte, CA, US). Each mixture was shaken for two minutes, the liquid decanted into 250-mL sterile plastic tubes, centrifuged at 4000g for 10 min, supernatant decanted into 50-mL sterile plastic tubes and recentrifuged at 4000g for 10 min. The four samples were frozen at −80 0 C in preparation for testing and inoculation. For testing of the positive control samples, each of the six samples was added into a 250-mL conical tube, followed by the addition of 60 ml of sterile saline. The sample was then homogenized and centrifuged 4000g for 10 min, with supernatant decanted into a clean 50-mL tube and recentrifuged at 4000g for 10 min. Supernatant was then decanted into 10-mL tubes and frozen at −80 0 C, in preparation for testing and inoculation.

| Diagnostic testing
Following processing, samples were evaluated for the presence of viral RNA by PCR and for virus viability by swine bioassay for soybased products and by natural feeding behaviour for complete feed.
For PCR, samples were tested at the South Dakota State University Animal Disease Research and Diagnostic Laboratory (SDSU ADRDL) using published methods (Dee et al., 2014(Dee et al., , 2015(Dee et al., , 2016. For viability testing, pigs were housed in the Pipestone Research biosafety level 2 facility for a 21-day period. For testing of soy-based ingredients by swine bioassay, 24 five-week-old pigs, originating from a farm known to be naïve for PRRSV, PEDV and SVA were housed in a single room, four pigs per pen, across six pens. As the experimental unit for the study was the pen, pens were organized according to ingredient, specifically: pen 1 represented conventional soybean meal tote one, pen 2 represented conventional soybean meal tote two, pen 3 represented organic soybean meal tote one, pen 4 represented organic soybean meal tote two, pen 5 represented positive control samples, while pen 6 represented negative control samples.
Solid pen dividers were placed between pens to prevent nose-tonose contact between groups and minimize cross-contamination.
For the assessment of viable virus in conventional soybean meal tote one, the four pigs in pen one were each inoculated with 2 ml via the intramuscular route, 2 ml via the oral route and 2 ml via the intranasal route. For the assessment of viable virus in conventional soybean meal tote two, four pigs in pen two were each inoculated with 2 ml via the intramuscular route, 2 ml via the oral route and 2 ml via the intranasal route. The same inoculation procedures were followed for samples of from organic soybean meal totes (pen 3 and pen 4). All six positive control samples (two samples from conventional soy totes, two samples from organic soy totes and two samples from complete feed totes) were pooled, and the four pigs in pen 5 were inoculated as described. Finally, all six negative control samples were pooled, and pigs in pen 6 were inoculated as described. For the testing of complete feed for the presence of viable virus, the three totes of complete feed from the transport vehicle were loaded into a single feed bin and 18-100 kg pigs, originating from the same source as the bioassay pigs, were housed in a single room (six pens, three pigs per pen) and allowed to consume the complete feed via natural feeding behaviour. During the 21-day period, a pen-based sampling protocol was employed to determine the status of each pen using oral fluid samples that were collected from each of the 12 pens on days 0, 7 and 14 post-inoculation. To support results from the oral fluid samples, clinically affected pigs were humanely euthanized, and tonsil tissue, rectal swabs and blood samples collected. All samples were tested by PCR at the SDSU ADRDL.

| Data analysis
Temperature and % RH data from the two soybean meal totes collected during the transport period were summarized using descriptive statistics. Differences in mean temperature and mean % RH F I G U R E 2 Schematic of AAFCO protocol for sampling bulk ingredients used in the study. Ten samples were collected from each tote using two 'X' patterns resulting in a composite sample per tote  between the conventional soybean meal and the organic soybean meal were analysed for significance using a two-sample t-test.

| Summary of the transport period
The transport period took place over 23 days, from November 30, 2020, to December 22, 2020. The route covered 29 states, for a total of 100.2 hr in transit over 10,183 km ( Figure 1)

| Feed samples
The mean temperature of the conventional soybean meal and the mean temperature of the organic soybean meal were significantly different (p <.0001) from one another, as were the mean % RH of the conventional soybean meal and the mean % RH of the organic soybean meal (p <.0001) (

| Presence of viral nucleic acid in feed
The results of the PCR testing of samples from the totes are summarized in Table 2a (Table 2b).

| Presence of viable virus in feed
Prior to inoculation, all pigs were confirmed to be naïve to all three  (Table 2c).

| D ISCUSS I ON
The ability of feed and feed ingredients to serve as vehicles for the transport and transmission of viral pathogens is a relatively new area ern Europe, and if they will become a reservoir for domestic pigs.
In the wake of the ASF epizootic in Viet Nam, Cambodia and Laos, we sought to detect the disease in the wild boar population using cost-effective, risk-based surveillance methods based on participatory engagement of local actors in each country. We conducted a preliminary assessment of the risk of ASFV transmission at the interface between domestic and wild pigs in several pre-selected villages located within or near wild pig habitat. The objectives were to characterize the interface between domestic pigs and wild boar to ascertain the risk of viral spillover from domestic pigs to wild boar, and to determine whether ASF transmission to the wild boar population could be detected in each country. Follow-up site visits and sampling were conducted to identify and document suspected viral spillover.

| Site selection
Investigative teams first identified villages in Laos, Cambodia and Viet Nam with a confirmed ASF outbreak in domestic pigs or with unconfirmed reports of high pig mortality numbers, and in close proximity to forested areas with known or suspected presence of wild boar. These sites were then visited for preliminary assessments of potential interfaces between wild and domestic pigs with risk of ASFV transmission. Based on relevant information obtained from the community members, forest rangers and protected area staff at these sites, additional villages were added to the investigation and visited. Three villages in Houaphanh province in Laos, two in Dong Nai province in Viet Nam and eleven in Ratanakiri province in Cambodia were selected for the assessments which were conducted in July, October/December and April/June 2019, respectively ( Figure 1) and mapped using ArcGIS 10.7.1 (ESRI, Redlands, CA, USA). Sites were not selected following a formal risk assessment, but rather were a convenience sample based on local reports received and field team time availability and resources.

| General approach
Investigative field teams performing site visits consisted of central, provincial, and district-level animal health and environmental sector representatives from local government, as well as field staff and veterinarians from the Wildlife Conservation Society (WCS). In each of science. Since the initial description of PEDV transmission in feed, an extensive amount of experimental evidence has been compiled to the point where comprehensive literature reviews can now be written on the topic and risk assessments can be conducted (Dee et al., 2020a, Jones et al., 2020b). Yet, to continue to challenge the hypoth- States. Despite the small sample size, it was interesting to note the significant difference in both the mean temperature and the mean % RH in the organic soybean meal tote versus the conventional soybean meal tote. While this outcome is based on only two totes, a large number (2,132) of datapoints were collected from each tote, resulting in information that could be used to generate new hypotheses on why certain ingredients are protective to certain viruses. In the case of point number 2, it is the opinion of the authors that the bulk sampling protocol worked well overall, particularly, since we were blinded to the location of the 'hot spot'.
It was interesting to note that once again, soy-based ingredients were protective to all three viruses, stabilizing both viral genome and viable virus. In contrast, complete feed was not as forgiving.
For example, all complete feed samples were PEDV PCR negative at day 23 post-inoculation; however, infection still occurred following natural feeding behaviour, suggesting that virus went undetected during sampling. Regarding SVA, RNA was detected in complete feed totes and infection was documented post-feeding, once again demonstrating the ability of this virus to survive for extended periods in feed. In contrast, PRRSV RNA was not detected in samples from complete feed totes and infection did not take place following feeding, suggesting that PRRSV did not survive in this feed matrix during long-distance transport, similar to what had been reported under laboratory conditions (Dee et al., 2018).
Finally, as it pertains to point number 3, this study demonstrated the ability of all three viruses to survive and to be infectious to   in combination with the current body of experimental evidence, will help to unify opinions across the swine industry, the veterinary profession and governmental agencies regarding the significance of the risk of feed for viral movement. Until we are united, we cannot make progress, and until that time, we all are at risk.

ACK N OWLED G EM ENTS
Dr. Dee would like to recognize and thank Cory Powell and Steve Hawkenson of SAM Nutrition for sharing their extensive knowledge of the transport process of feed and feed ingredients along with their high level of technical expertise that helped make the project a success. In addition, the support and assistance of the FDA CVM and the USDA are very much appreciated.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

E TH I C A L A PPROVA L
Animals in this study were managed in accordance with the institutional animal care and use guidelines observed by the investigators' ethical review board, Pipestone Applied Research IACUC, trial number 2021-01.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data from this study have been disclosed.  ern Europe, and if they will become a reservoir for domestic pigs.
In the wake of the ASF epizootic in Viet Nam, Cambodia and Laos, we sought to detect the disease in the wild boar population using cost-effective, risk-based surveillance methods based on participatory engagement of local actors in each country. We conducted a preliminary assessment of the risk of ASFV transmission at the interface between domestic and wild pigs in several pre-selected villages located within or near wild pig habitat. The objectives were to characterize the interface between domestic pigs and wild boar to ascertain the risk of viral spillover from domestic pigs to wild boar, and to determine whether ASF transmission to the wild boar population could be detected in each country. Follow-up site visits and sampling were conducted to identify and document suspected viral spillover.

| Site selection
Investigative teams first identified villages in Laos, Cambodia and Viet Nam with a confirmed ASF outbreak in domestic pigs or with unconfirmed reports of high pig mortality numbers, and in close proximity to forested areas with known or suspected presence of wild boar. These sites were then visited for preliminary assessments of potential interfaces between wild and domestic pigs with risk of ASFV transmission. Based on relevant information obtained from the community members, forest rangers and protected area staff at these sites, additional villages were added to the investigation and visited. Three villages in Houaphanh province in Laos, two in Dong Nai province in Viet Nam and eleven in Ratanakiri province in Cambodia were selected for the assessments which were conducted in July, October/December and April/June 2019, respectively ( Figure 1) and mapped using ArcGIS 10.7.1 (ESRI, Redlands, CA, USA). Sites were not selected following a formal risk assessment, but rather were a convenience sample based on local reports received and field team time availability and resources.

| General approach
Investigative field teams performing site visits consisted of central, provincial, and district-level animal health and environmental sector representatives from local government, as well as field staff and veterinarians from the Wildlife Conservation Society (WCS). In each