3.4 What is the prevalence of NTS in samples of poultry from the Americas?
As summarized in Table 2 , the prevalence of NTS varied both according to the type of the sample assessed and the country. The pooled overall prevalence of NTS was 17.9% (95% CI: 10.8 to 26.3) in 43 studies that reported 24,113 samples of birds from 11 countries with a significant test of effect size (Z=7.5, p =0.00) and substantial heterogeneity as judged by the I 2 (99.5%,p =0.00). For these samples, the subgroup analysis revealed a lower prevalence of 9.8% of NTS in individual data (3.8 to 18.2, n=25) in comparison to the value of 33.8% (16.6 to 53.3, n=18) estimated in flock data (Supplementary Figure 1 ). At the national level, the prevalence of NTS ranged from 23.2 to 50% in samples of birds reported in studies from Mexico, the USA, Uruguay, and Venezuela. In contrast, the prevalence of NTS was lower in studies from Brazil, Chile, Paraguay, and Trinidad and Tobago where the estimations varied between 0.8 and 3.7% (Table 2 ).
In 90 studies that reported 173,354 samples of products and subproducts from 10 countries, the overall pooled prevalence of NTS was 21.8% (17.7 to 26.1) with a significant test of effect size (Z=16.8, p =0.00) and a significant variation attributable to heterogeneity (I 2=99.6%, p =0.00). According to subgroup analysis, the prevalence estimations were consistent between the individual (21.2%, 17.7 to 26.1, n=78) and collective data (28.1%, 12.6 to 46.4, n=12) of these samples (Supplementary Figure 2 ). The studies from Canada showed the highest pooled prevalence of NTS in samples from products and subproducts, while prevalence values ranging from 31.2 to 34.4% were estimated in studies from Guatemala, Mexico, and Puerto Rico. Finally, Argentina, Brazil, Costa Rica, and Trinidad and Tobago had prevalence values below-average (Table 2 ).
Regarding the environmental samples, the pooled prevalence of NTS was 29.5% (24.2 to 35.1) in 61 studies that comprised 63,941 samples from eight countries. There was a significant test of effect size (Z=17.1,p =0.00) and substantial heterogeneity across studies (I 2=94.4%, p =0.00), and the subgroup analysis showed a contrasting difference of 21.9 and 52.4% at the individual (17.6 to 26.6, n=43) and collective (41.7 to 62.9, n=18) level, respectively (Supplementary Figure 3 ). At the national level, the studies from Canada and the USA had prevalence values above the overall estimation, whereas with estimations varying 3.3 to 9.0%, the studies from Argentina and Costa Rica showed the lowest prevalence of NTS in environmental samples (Table 2 ).
Which are the main serovars of NTS identified in each type of sample?
In total, 131 NTS serovars were identified from 13,388 isolates from poultry samples reported in 94 publications from 13 countries (Fig. 3 ). The USA and Canada contributed both with the highest percentage of isolates (84.0%) and the greatest diversity of NTS serovars (68 and 57, respectively). Likewise, Brazil and Colombia provided a great diversity of serovars (53 and 40, respectively), whereas the remaining countries reported a lower diversity of NTS serovars (range 1-19). A detailed list of the distribution of the 131 NTS serovars per country is provided in Supplementary Table 5 .
Overall, 84.3% of the isolates were distributed among 10 NTS serovars, Heidelberg, Kentucky, and Enteritidis were the most prevalent (20.8, 20.6, and 17.6%, respectively), whereas Mbandaka, Schwarzengrund, and Montevideo were less prevalent with values ranging from 2.2 to 2.4%. Outside the top 10 serovars, the category “Others” comprised the remaining 121 serovars that were found in 2,103 isolates. As shown inFigure 3 , there was a heterogenous pattern in the top 10 ranking of the NTS serovars across countries, which additionally had several serovars not included in the overall top 10 ranking. Additionally, in Costa Rica, Ecuador, Mexico, and Trinidad and Tobago their 1st top serovar was not included in the overall top 10 ranking list. Even though the serovars Heidelberg and Kentucky were ranked as 1st and 2nd overall, these serovars were reported only in 5 and 6 countries, respectively, where their ranks varied across countries. In contrast, serovars Enteritidis and Typhimurium (3rd and 4th overall) were identified in most of the countries (10 and 8 respectively); thus, these two serovars had the broadest distribution in poultry samples across countries from the Americas. Furthermore, Enteritidis was the 1st top-ranked serovar in the USA, Brazil, Colombia, Paraguay, and Uruguay where its prevalence ranged from 23.4 to 100% and the 2ndtop-ranked serovar in Argentina, Chile, and Ecuador. Similarly, Typhimurium was the 1st top-ranked serovar in Chile and the 3rd ranked in Colombia. Despite the serovars Hadar, Thompson, and Montevideo were ranked in the top 10 overall, they were reported only in two countries each and thus showed a narrow geographical distribution.
We constructed Sankey diagrams to examine and compare the distribution of the top five NTS serovars according to the type of poultry sample (Fig. 4 ). Laying hens, broilers, and a mixture category that grouped chicks and hens were the main sources of NTS isolates in animal samples, whereas carcasses, eggs, and subproducts for consumption (such as meat and processed products) provided most of the isolates for the products and subproducts group. Finally, among environmental samples, most of the isolates were obtained from production wastes (such as litter and feces), breeding facilities, and production supplies (such as water, feed, and litter). Serovars Enteritidis and Heidelberg were identified in all the poultry samples, whereas the samples from products and subproducts and environmental shared 4 out of 5 serovars (Enteritidis, Kentucky, Heidelberg, and Typhimurium). Among the samples from birds and products and subproducts, Enteritidis was the most prevalent serovar with 32.4 and 22.8%, respectively, followed by Senftenberg and Kentucky (10.6% for birds and 21.5% for products and subproducts). Heidelberg and Kentucky were the most prevalent serovars in environmental samples (32.1 and 28.8%, respectively).
What is the prevalence and the profile of antimicrobial resistance in NTS isolates of poultry?
Of the 46 publications that assessed the AMR of the NTS isolates, 38/46 reported the prevalence in 3,078 isolates of which 2,223 were resistant to at least one antibiotic. As depicted in Figure 5a , the prevalence of AMR tended to be higher according to the increase in the number of antibiotics to which the NTS isolates were resistant: 1 antibiotic, 24.1% (14.7 to 34.9%, n=440); 2-3 antibiotics, 36.2% (24.5 to 48.6%, n=758); and ≥ 4 antibiotics, 49.6% (36.9 to 62.2%, n=1,025). Besides, there was a heterogenous pattern of prevalence to AMR among the three types of samples because in samples from birds and products and subproducts, resistance to ≥ 4 antibiotics was highly prevalent (61.2 and 48.7%), whereas resistance to 2-3 antibiotics was highly prevalent in environmental samples (47.1%) (Supplementary Figure 4 ).
A total of 26/46 publications reported 50 serovars from 1,688 NTS isolates that were resistant to at least one antibiotic. As shown inFigure 5b , 10 serovars contributed with 79.2% of the total, among which Kentucky (29.5%), Heidelberg (16.7%), Typhimurium (9.8%), and Enteritidis (7.1%) were the top four serovars most frequently reported as resistant to at least one antibiotic. Outside the top 10 ranking, the category “Others” comprised 40 serovars from 331 NTS isolates of which Typhimurium var. 5-, 4,5, 12:1:, Pullorum, Anatum, Lithfield, Montevideo, Hadar, Ohio, Newport, and Senftenberg were reported in 10 to 33 isolates each. In contrast, serovars such as Havana, Idikan, Ouakam, Bredeney, Alachua, Corvallis, Istambul, Livingstone, Worthington, and I 4,5, 12: - were reported only once across the publications.
In total, 40/46 publications reported 15 groups of antibiotics that included 57 antibiotics to which 8,911 NTS isolates were resistant. Each group included between 1 and 11 antibiotics that varied in the frequency of report (Supplementary Table 6 ). In Figure 5c are depicted the main six groups of drugs and their corresponding antibiotics to which 90.6% of the isolates were resistant. Cephalosporins, penicillins, tetracyclines, and aminoglycosides were the top four groups, which concentrated 71.9% of the resistant isolates of NTS. As shown in Figure 5d , with a prevalence that ranged from 5.8 to 16.2%, tetracycline, ampicillin, streptomycin, ceftiofur, and amoxicillin-clavulanic acid were among the top 10 antibiotics with the highest overall prevalence of resistance in Salmonella isolates. In the bottom rank, gentamicin and trimethoprim-sulfamethoxazole were the antibiotics with the lowest prevalence of resistance in NTS. Regarding the three poultry samples, the profile of antimicrobial resistant was heterogenous. Except for tetracycline, which was the 1st top-ranked antibiotic among resistant isolates, the remaining antibiotics varied in their position within the rank. For instance, streptomycin that was the second most prevalent in birds, ranked 4th and 5th in environmental samples and products and subproducts. Likewise, ampicillin ranked 5th in environmental samples, whereas in the two other samples this antibiotic was among the top three, nalidixic acid that was the third to last most prevalent overall, in environmental samples was ranked 3rd. Besides, in the three samples of poultry there were 1-2 antibiotics that were not included in the overall top 10 ranking.