A Decrease of Incidence Cases of Fumonisins in South Korean Feedstuff between 2011 and 2016

Several plant pathogen Fusarium species produce fumonisins (FUMs); which can end up in food and feed and; when ingested; can exhibit harmful effects on humans and livestock. Mycotoxin intoxication by fumonisin B1 (FB1) and fumonisin B2 (FB2) can cause porcine pulmonary edema; leukoencephalomalacia in equines; esophageal cancer and birth defects by natural contamination. Herein; the occurrence of FB1 and FB2 in feedstuff (compound feed and feed ingredients) was investigated between 2011 and 2016 in South Korea. A total of 535 animal feed samples (425 compound feed samples and 110 feed ingredients) produced domestically were sampled four times between 2011 and 2016 (2011; 2012; 2014 and 2016) from feed factories in South Korea. The limit of detection (LOD) for FB1 and FB2 was 20 μg/kg and 25 μg/kg; respectively; and the limit of quantitation (LOQ) was 30 μg/kg and 35 μg/kg; respectively. The recovery range (%) was between 86.4% and 108.8%; and the relative standard deviation (RSD) (%) was 4.7–12.1%. Seven (swine feed samples) out of the 425 feed samples exceeded the European Union (EU) and South Korea commission regulations over the six-year test period; and no feed ingredients exceeded the guidelines.


Introduction
Toxins produced by filamentous fungi are called mycotoxins and can contaminate food and feed products [1].Mycotoxin contamination can occur at various steps during food and feed production, including storage and distribution, and it decreases the quality of the product [2].Indeed, mycotoxin contamination reduces the efficiency of livestock and crop production in many countries [3,4].Furthermore, if a previously-contaminated ingredient is added to foodstuff or animal feed, human and animal health is threatened [5].
Fusarium species are toxigenic and pathogenic to plants or humans, and the trichothecenes and the fumonisins produced by them are the most well-known mycotoxins [6].Fumonisins (FUMs) were first isolated from F. moniliforme in 1988 [7], and they can be produced by other Fusarium species such as F. verticillioides, F. proliferatum and F. nygamai [8].FUMs are soluble in water, unlike other hydrophobic mycotoxins, and this property makes it difficult to study fumonisins [9].Fumonisin B 1 (FB1) is composed of two units of propane-1,2,3-tricarboxylic acid and a 2S-amino-12S,16R-dimethyl-3S,5R,10R,14S,15R-pentahydroxy-eicosane backbone connected by a diester bridge at the C-14 and C-15 hydroxy groups [10].The chemical structures of FB1 and FB2 are shown in Figure 1.
(A) (B) These toxins can disrupt sphingolipid biosynthesis leading to serious problems in membrane biosynthesis because sphingolipids regulate cell growth and differentiation through participation in communication and interactions related to the cell [11,12].They also cause leukoencephalomalacia, pulmonary edema, hydrothorax, apoptosis, hepatotoxic and carcinogenic effects in animals [7,[13][14][15][16][17][18], as well as esophageal cancer in humans [19].FB1 was assigned to group 2B, "possibly carcinogenic to humans", by the International Agency for Research on Cancer [20].
Because of the serious damage that mycotoxins can cause to human and animal health, mycotoxins are an important concern for the feed industry, feed supply chain and international trade [21].Binder et al. [22] reported that the amount and type of fungal contamination exhibits significant regional differences.For example, in their study, European samples were mostly contaminated by deoxynivalenol (DON), T-2 toxin and zearalenone (ZEN), but Asian and Pacific samples had aflatoxins (AFs), DON, FUMs and ZEN as major contaminants.In 2015, the levels of FB1 and glycated FB1 in a total of 30 compound feed samples were measured, and all measured concentrations of FB1 were below the regulation limits set by the European Union and the U.S. Food and Drug Administration [23].
Grains, grain byproducts and vegetable proteins found in animal feed are good nutrients for the fungi that produce mycotoxins.Therefore, mycotoxins in animal feed need to be closely monitored.In 2012, 18,640 tons of compound feed were distributed in Korea, and 15,350 tons of feed ingredients were purchased by Korea from many places including China, the USA, Europe, Canada, South Africa, South East Asia, Australia and India.Feed ingredients from different countries make it difficult to control mycotoxin levels in compound feed.In one study in Slovakia, a total of 50 poultry feed samples was collected during 2003 and 2004, and the level of FUMs was measured.FB1 and FB2 were detected in 49 samples (98.0%) and 42 samples (84.0%), respectively [24].In another study, brewers' grain, used as dairy cattle feed, was analyzed using high performance liquid chromatography (HPLC) in Brazil, and FUMs contaminated 72.5% (58) of the samples with rates between 50.3 and 908.5 μg/kg [25].In China, six mycotoxins were quantified in 83 feed samples including FUMs, and 82.7% were contaminated [26].These toxins can disrupt sphingolipid biosynthesis leading to serious problems in membrane biosynthesis because sphingolipids regulate cell growth and differentiation through participation in communication and interactions related to the cell [11,12].They also cause leukoencephalomalacia, pulmonary edema, hydrothorax, apoptosis, hepatotoxic and carcinogenic effects in animals [7,[13][14][15][16][17][18], as well as esophageal cancer in humans [19].FB 1 was assigned to group 2B, "possibly carcinogenic to humans", by the International Agency for Research on Cancer [20].
Because of the serious damage that mycotoxins can cause to human and animal health, mycotoxins are an important concern for the feed industry, feed supply chain and international trade [21].Binder et al. [22] reported that the amount and type of fungal contamination exhibits significant regional differences.For example, in their study, European samples were mostly contaminated by deoxynivalenol (DON), T-2 toxin and zearalenone (ZEN), but Asian and Pacific samples had aflatoxins (AFs), DON, FUMs and ZEN as major contaminants.In 2015, the levels of FB 1 and glycated FB 1 in a total of 30 compound feed samples were measured, and all measured concentrations of FB 1 were below the regulation limits set by the European Union and the U.S. Food and Drug Administration [23].
Grains, grain byproducts and vegetable proteins found in animal feed are good nutrients for the fungi that produce mycotoxins.Therefore, mycotoxins in animal feed need to be closely monitored.In 2012, 18,640 tons of compound feed were distributed in Korea, and 15,350 tons of feed ingredients were purchased by Korea from many places including China, the USA, Europe, Canada, South Africa, South East Asia, Australia and India.Feed ingredients from different countries make it difficult to control mycotoxin levels in compound feed.In one study in Slovakia, a total of 50 poultry feed samples was collected during 2003 and 2004, and the level of FUMs was measured.FB 1 and FB 2 were detected in 49 samples (98.0%) and 42 samples (84.0%), respectively [24].In another study, brewers' grain, used as dairy cattle feed, was analyzed using high performance liquid chromatography (HPLC) in Brazil, and FUMs contaminated 72.5% (58) of the samples with rates between 50.3 and 908.5 µg/kg [25].In China, six mycotoxins were quantified in 83 feed samples including FUMs, and 82.7% were contaminated [26].According to the European Commission (EC), maximum levels for Fusarium toxins were established in products intended for animal feeding in 2006 [27], and the Food and Drug Administration (FDA) in the U.S. set guidance levels for FUMs in human food and animal feed in 2001 [28].In South Korea, the guidance value for Fusarium mycotoxins in animal feed was established based on guidelines of the EC and the results of studies in Korea.The guidelines for the EC and South Korea are summarized in Table 1.
The object of this study was to confirm the contamination level of FUMs in feed and feed ingredients and to compare the level of FUMs with the guidelines for animal feed in South Korea.For this purpose, we collected feed samples over the course of six years, analyzed the levels of FUMs contamination and assessed the risk of FUMs contamination in animal feed in Korea.

Method Validation
Fumonisin B 1 and B 2 could be clearly separated from any potential interference in the HPLC chromatogram after pretreatment with an immunoaffinity column (IAC).The LOD and LOQ were estimated at 20 µg/kg and 30 µg/kg for FB 1 and 25 µg/kg and 35 µg/kg for FB 2 , respectively.The regression coefficient was over 0.999 in the solvent-based standard curve for FUMs (Figure 2A,B).The HPLC chromatogram of the standard solutions of FUMs spiked into an animal feed sample did not show any interfering peaks in the HPLC chromatogram (Figure 2C).Therefore, this analytical method was suitable for the analysis of FUMs.The accuracy for FUMs was expressed as the average recovery ratio in the recovery test.Precision was calculated from the same tests and expressed as percent relative standard deviation (%RSD).The range of recovery ratio was 86.4-108.8%,and %RSD was 4.7-12.1% (Table 2).These results satisfied the commission regulation for FUMs in the EU that defines an accuracy of 70-120% and %RSD for precision under 20% [30].The results of the method validation study are summarized in Table 2.After analysis of FUMs in the HPLC chromatogram, LC-MS analysis was further applied to identify FUMs with its extracted ion chromatogram (XIC) and spectrum of mass (Figure 3).Extracted ion chromatograms of FB 1 and FB 2 in a standard solution (Figure 3A,B) and in a representative sample (Figure 3C,D

Occurrence of FUMs in Compound Feed Samples between 2011 and 2016
The contamination levels of FUMs were estimated from 425 compound feed samples consisting of cattle feed (165), pig feed (140) and poultry feed (140), which were collected in 2011, 2012, 2014 and 2016.FB1 was detected in 90.6% of all of the compound feed samples; it was detected in 97.9% of cattle feed samples, followed by 90.7% of poultry feed samples and 82.9% of pig feed samples.FB2 also contaminated 63.3% of all compound feed samples: 78.6% of cattle feed samples, 45.7% of pig feed samples and 65.0% of poultry feed samples.Detailed data including mean contamination level of FUMs by feed type are shown in Table 3. Almost all compound feed samples were contaminated with FB1, and the mean contamination level in all of the compound feed samples was 1448.1 μg/kg, with the level of FB1 in the range of 36.4-14,876.8μg/kg.FB2 was detected in over half of compound  R 2 : coefficient of determination, LOD: limit of detection, LOQ: limit of quantitation, SD: standard deviation, RSD: relative standard deviation.

Occurrence of FUMs in Compound Feed Samples between 2011 and 2016
The contamination levels of FUMs were estimated from 425 compound feed samples consisting of cattle feed (165), pig feed (140) and poultry feed (140), which were collected in 2011, 2012, 2014 and 2016.FB 1 was detected in 90.6% of all of the compound feed samples; it was detected in 97.9% of cattle feed samples, followed by 90.7% of poultry feed samples and 82.9% of pig feed samples.FB 2 also contaminated 63.3% of all compound feed samples: 78.6% of cattle feed samples, 45.7% of pig feed samples and 65.0% of poultry feed samples.Detailed data including mean contamination level of FUMs by feed type are shown in Table 3. Almost all compound feed samples were contaminated with FB 1 , and the mean contamination level in all of the compound feed samples was 1448.1 µg/kg, with the level of FB 1 in the range of 36.4-14,876.8µg/kg.FB 2 was detected in over half of compound feed samples.Its mean contamination level was 210.8 µg/kg with the contamination level in the range of 14.9-9235.8µg/kg.
The mean concentration level of FB 1 in cattle feed, pig feed and poultry feed was 1779.0,1171.1 and 1382.5 µg/kg, respectively, so cattle feed had the highest mean contamination level among the three types of compound feed.In the case of FB 2 , the highest mean contamination level was shown in poultry compound feed.There was no statistically significant difference between the three groups of compound feed samples in a one-way analysis of variance (ANOVA, p < 0.05) (Table 4).The highest contamination value for FB 1 (14,876.8µg/kg) was detected in a weanling piglet feed, and this sample had an FB 2 level of 221.1 µg/kg.The sum of these value was over the EC and Korean regulation limits (5000 µg/kg).In addition, two growing pig feed samples, two gestating sow feed samples, a lactating sow feed sample, a sucking piglet feed sample and a weanling piglet feed sample exceeded the EC and Korean limits.Among all of the animal compound feed, only several pig feeds (1.41%) were above the limit.In the case of FB 2 , the highest level of FB 1 (8343.4µg/kg) was detected in an early laying hens' feed, which shows FB 2 co-contamination (1048.2µg/kg), and the highest level of FB 2 (9235.8µg/kg) was detected in an early broiler feed exhibiting 446.1 µg/kg of FB 1 content.The distribution of FUMs according to the type of compound feed is shown in Figure 4.The distributions of FUMs by year of sampling are shown in Figure 5.In 2011, 2012, 2014 and 2016, the mean contamination level of FB 1 was 2817.5, 805.1, 839.2 and 341.2 µg/kg, respectively, and for FB 2 , it was 192.5, 324.4,112.1 and 84.1 µg/kg, respectively.In 2011, a weanling piglet feed sample showed the highest level of FB 1 (14,876.8µg/kg), and a middle beef calf sample showed the highest level of FB 2 (2277.9µg/kg).In 2012, the highest level of FB 1 (8343.4µg/kg) was detected in an early laying hens' feed, and the highest level of FB 2 (9235.8µg/kg) was detected in an early broiler feed.In 2014, a middle beef calf sample had the highest level of FB 1 (4576.1 µg/kg), and an early broiler feed showed the highest level of FB 2 (821.2µg/kg).Lastly, in 2016, the highest contamination level was evaluated in a middle dairy calf feed with an FB 1 concentration of 3225.0 µg/kg and an FB 2 concentration of 670.0 µg/kg. of FB2 (9235.8μg/kg) was detected in an early broiler feed exhibiting 446.1 μg/kg of FB1 content.The distribution of FUMs according to the type of compound feed is shown in Figure 4.The distributions of FUMs by year of sampling are shown in Figure 5.In 2011, 2012, 2014 and 2016, the mean contamination level of FB1 was 2817.5, 805.1, 839.2 and 341.2 μg/kg, respectively, and for FB2, it was 192.5, 324.4,112.1 and 84.1 μg/kg, respectively.In 2011, a weanling piglet feed sample showed the highest level of FB1 (14,876.8μg/kg), and a middle beef calf sample showed the highest level of FB2 (2277.9μg/kg).In 2012, the highest level of FB1 (8343.4μg/kg) was detected in an early laying hens' feed, and the highest level of FB2 (9235.8μg/kg) was detected in an early broiler feed.In 2014, a middle beef calf sample had the highest level of FB1 (4576.1 μg/kg), and an early broiler feed showed the highest level of FB2 (821.2 μg/kg).Lastly, in 2016, the highest contamination level was evaluated in a middle dairy calf feed with an FB1 concentration of 3225.0 μg/kg and an FB2 concentration of 670.0 μg/kg.(A)

Occurrence of FUMs in Feed Ingredients between 2011 and 2016
One-hundred and ten feed ingredient samples were collected four times over six years: in 2011, 2012, 2014 and 2016.The contamination levels of these samples were evaluated using HPLC.The collected feed ingredient samples were composed of grains (seven samples), grain by-products (bran, 27 samples), meal (vegetable protein, 55 samples), fibrous feed (nine samples), food by-products (eight samples) and other feed ingredients (one sample of beans, two samples of seeds/nuts and a mixed formulation sample).The detailed data related to the collected feed ingredient samples are shown in Table 5.
The contamination rates of FB1 were 14.3% in grains, 70.4% in grain by-products (bran), 63.6% in meal, 33.3% in fibrous feed, 37.5% in food by-products and one bean sample.Overall, 56.4% of the tested feed ingredients were contaminated by FB1.In the case of FB2, the contamination rates were 48.1, 45.5, 44.4 and 12.5 in grain by-products (brans), meal, fibrous feed and food by-products, respectively.However, grains, seeds/nuts and mixed formulations were not contaminated by FB2, so 40.0% of the total samples exhibited contamination with FB2.The mean contamination levels of FUMs were 624.9 and 169.3 μg/kg for FB1 and FB2, respectively.The FUMs concentrations in total feed ingredients ranged from 15.5-6357.5 μg/kg for FB1 and from 21.5-1798.9μg/kg for FB2.
The calculated mean contamination values of FB1 were 12.5, 1224.8,580.8, 354.0 and 50.5 μg/kg in grains, grain by-products, meal, fibrous feed and food by-products, respectively; and the mean contamination values of FB2 were 309.6, 169.4,94.7 and 9.4 μg/kg in grain by-products, meal, fibrous feed and food by-products, respectively.The maximum contaminated concentrations of FB1 and FB2 were 6357.5 and 1798.9 μg/kg in corn bran, 4816.0 and 1662.7 μg/kg in corn gluten meal, 1815.0 and 580.0 μg/kg in fibrous feed and 169.5 and 75.5 μg/kg in food by-products, respectively; and only one grain product was contaminated by FB1 at 87.3 μg/kg.FUMs were not found in seeds/nuts or mixed formulations.Forty-four percent of all feed ingredient samples had lower FB1 concentrations than the LOD, and 60.0% of the total feed ingredients had lower FB2 values than the LOD.The distributions of FB1 and FB2 in accordance with feed ingredients and class of feed ingredients are shown in Figure 6.

Occurrence of FUMs in Feed Ingredients between 2011 and 2016
One-hundred and ten feed ingredient samples were collected four times over six years: in 2011, 2012, 2014 and 2016.The contamination levels of these samples were evaluated using HPLC.The collected feed ingredient samples were composed of grains (seven samples), grain by-products (bran, 27 samples), meal (vegetable protein, 55 samples), fibrous feed (nine samples), food by-products (eight samples) and other feed ingredients (one sample of beans, two samples of seeds/nuts and a mixed formulation sample).The detailed data related to the collected feed ingredient samples are shown in Table 5.
The contamination rates of FB 1 were 14.3% in grains, 70.4% in grain by-products (bran), 63.6% in meal, 33.3% in fibrous feed, 37.5% in food by-products and one bean sample.Overall, 56.4% of the tested feed ingredients were contaminated by FB 1 .In the case of FB 2 , the contamination rates were 48.1, 45.5, 44.4 and 12.5 in grain by-products (brans), meal, fibrous feed and food by-products, respectively.However, grains, seeds/nuts and mixed formulations were not contaminated by FB 2 , so 40.0% of the total samples exhibited contamination with FB 2 .The mean contamination levels of FUMs were 624.9 and 169.3 µg/kg for FB 1 and FB 2 , respectively.The FUMs concentrations in total feed ingredients ranged from 15.5-6357.5 µg/kg for FB 1 and from 21.5-1798.9µg/kg for FB 2 .
The calculated mean contamination values of FB 1 were 12.5, 1224.8,580.8, 354.0 and 50.5 µg/kg in grains, grain by-products, meal, fibrous feed and food by-products, respectively; and the mean contamination values of FB 2 were 309.6, 169.4,94.7 and 9.4 µg/kg in grain by-products, meal, fibrous feed and food by-products, respectively.The maximum contaminated concentrations of FB 1 and FB 2 were 6357.5 and 1798.9 µg/kg in corn bran, 4816.0 and 1662.7 µg/kg in corn gluten meal, 1815.0 and 580.0 µg/kg in fibrous feed and 169.5 and 75.5 µg/kg in food by-products, respectively; and only one grain product was contaminated by FB 1 at 87.3 µg/kg.FUMs were not found in seeds/nuts or mixed formulations.Forty-four percent of all feed ingredient samples had lower FB 1 concentrations than the LOD, and 60.0% of the total feed ingredients had lower FB 2 values than the LOD.The distributions of FB 1 and FB 2 in accordance with feed ingredients and class of feed ingredients are shown in Figure 6.According to guidelines set by the EC and Korea, the limit for FUMs was 60,000 μg/kg in maize by-products.The maximum concentration of FUMs was observed in a corn bran sample with 8156.4 μg/kg FB1 + FB2.This value is under the limit in the EU and Korea.In other feed ingredient types, the maximum contamination values for FUMs were 6478.6).Maximum levels of FUMs were estimated as follows: 4816.0 μg/kg of FB1 and 1662.7 μg/kg of FB2 in corn gluten meal in 2011, 6357.5 μg/kg of FB1 and 1798.9 μg/kg of FB2 in corn bran in 2012, 5354.9 μg/kg of FB1 and 1033.9 μg/kg of FB2 in corn bran in 2014 and 1815.0 μg/kg of FB1 and 580.0 μg/kg of FB2 in fibrous feed in 2016.The occurrence of FUMs in each year (2011, 2012, 2014 and 2016) is shown in Figure 7.There was no statistically-significant difference year-to-year for feed ingredients in a one-way ANOVA (p > 0.05, Table 6).According to guidelines set by the EC and Korea, the limit for FUMs was 60,000 µg/kg in maize by-products.The maximum concentration of FUMs was observed in a corn bran sample with 8156.4 µg/kg FB 1 + FB 2 .This value is under the limit in the EU and Korea.In other feed ingredient types, the maximum contamination values for FUMs were 6478.7 µg/kg in corn gluten meal and 1138.7 µg/kg in corn germ meal.Therefore, no feed ingredient samples exceeded the guidance level.
The number of collected feed ingredient samples was 110 samples including 40 samples in 2011, 30 samples in 2012, 17 samples in 2014 and 23 samples in 2016.Mean contamination values were 680.6, 947.6, 559.2 and 155.1 µg/kg for FB 1 and 161.0, 266.2, 177.4 and 51.7 µg/kg for FB 2 in 2011, 2012, 2014 and 2016, respectively (Table 6).Maximum levels of FUMs were estimated as follows: 4816.0 µg/kg of FB 1 and 1662.7 µg/kg of FB 2 in corn gluten meal in 2011, 6357.5 µg/kg of FB 1 and 1798.9 µg/kg of FB 2 in corn bran in 2012, 5354.9 µg/kg of FB 1 and 1033.9 µg/kg of FB 2 in corn bran in 2014 and 1815.0 µg/kg of FB 1 and 580.0 µg/kg of FB 2 in fibrous feed in 2016.The occurrence of FUMs in each year (2011, 2012, 2014 and 2016) is shown in Figure 7.There was no statistically-significant difference year-to-year for feed ingredients in a one-way ANOVA (p > 0.05, Table 6).

Discussion
Many studies related to FUMs have been performed worldwide in order to determine contamination levels in compound feed.For example, Marquardt and Madhyastha [32] surveyed levels of various toxins in feed material and feedstuff worldwide.The highest incidence, with more than 50% positive samples, was found in North, South and Southeast Asia, South America, Southern Europe and Africa.A study in Kuwait [33] reported that all samples of poultry feed that they tested (53 samples) were contaminated with FUMs in the range of 220-6000 μg/kg with the mean contamination level of 2733 μg/kg.Similar results were reported by Greco et al. [34]: FUMs were detected in all of the 46 poultry feed samples with the concentration between 222 and 6000 μg/kg (median 1750 μg/kg).Kocasari et al. (2013) [35] analyzed 180 compound feed samples in Turkey for the contamination of FUMs; approximately 10.6% of samples were contaminated by FUMs, and the mean contamination value was measured at 3190 μg/kg with range of 2690-4960 μg/kg.In the

Discussion
Many studies related to FUMs have been performed worldwide in order to determine contamination levels in compound feed.For example, Marquardt and Madhyastha [32] surveyed levels of various toxins in feed material and feedstuff worldwide.The highest incidence, with more than 50% positive samples, was found in North, South and Southeast Asia, South America, Southern Europe and Africa.A study in Kuwait [33] reported that all samples of poultry feed that they tested (53 samples) were contaminated with FUMs in the range of 220-6000 µg/kg with the mean contamination level of 2733 µg/kg.Similar results were reported by Greco et al. [34]: FUMs were detected in all of the 46 poultry feed samples with the concentration between 222 and 6000 µg/kg (median 1750 µg/kg).Kocasari et al. (2013) [35] analyzed 180 compound feed samples in Turkey for the contamination of FUMs; approximately 10.6% of samples were contaminated by FUMs, and the mean contamination value was measured at 3190 µg/kg with range of 2690-4960 µg/kg.In the Republic of South Africa, a total of 92 commercial compound feed samples was examined for contamination with various mycotoxins, and FUMs were detected in most of the compound feed samples [36].Specifically, cattle feed was severely contaminated by FB 1 and FB 2 .The mean contamination level of FUMs was 916 µg/kg, and the highest contamination level was 2999 µg/kg.In addition, the mean contamination value of FUMs in cattle, chicken and swine feed was 903, 975 and 313 µg/kg, respectively.The range of the contamination level of FB 2 was 103-673 µg/kg, which is lower than the contamination values for FB 1 .This research showed that no compound feed samples exceeded the guidance level of the EC.Another study [37] was conducted to determine the level of FUMs in a total of 30 pig feed samples.The mean contamination level of FUMs was 405 µg/kg, and 80% of the feed samples were contaminated by FUMs, ranging from 30-1043 µg/kg.The level of FUMs in all of samples was below 5000 µg/kg, the EC guidance value.Zachariasova et al. [38] conducted a study in order to quantify 56 mycotoxins in 343 samples of animal feed.Those samples were non-fermented and fermented feedstuff, feedstuff supplements and complex compound feed collected from the Czech Republic (256 samples) and the United Kingdom (87 samples) between 2008 and 2012.Ninety-five compound feed samples (for pigs, chickens and laying hens, birds, rodents and dairy cows) were analyzed among the collected 343 samples, and only 13 compound feed samples for dairy cows were contaminated with FB 1 ; three compound feed samples for dairy cows were contaminated by FB 2 .The median level of FB 1 was 10 µg/kg, and the maximum level was 58 µg/kg.In the case of FB 2 , the median was 9 µg/kg, and the maximum level was 14 µg/kg.Those results were much lower than the EC legislation.In that article, FUMs were detected below 100 µg/kg in all 278 samples.In Poland, a study [39] was performed to evaluate the level of several mycotoxins in 1384 samples including 480 complete feed samples during a four-year period.In that study, FUMs were present in 85.7% of the 14 complete feed samples analyzed.The maximum level of FUMs was measured at 1063 µg/kg.A total of 74 ready-to-consume feed samples for dairy cattle was collected in Konya, Turkey and the surrounding provinces [40].Seven mycotoxins were analyzed in collected feed samples, and FUMs were the most common mycotoxin; 93.2% of samples had detectable levels of FUMs.The maximum detected level was 1208 µg/kg, which was below the 50,000 µg/kg maximum residue limit in Turkey.A similar study [41] was also performed in Turkey in which 22 cattle feed samples were collected every quarter over one year (total 88 samples) from some provinces in Turkey; 94.3% of those samples were contaminated by FUMs.The contamination level was between 0 and 3900 µg/kg.Samples collected in the summer exhibited a higher mean contamination level than in any other season.
In this study, the contamination level of FUMs was estimated in 425 feed samples consisting of 145 samples of cattle feed, 140 samples of pig feed and 140 samples of poultry feed collected in 2011, 2012, 2014 and 2016 in the Republic of Korea.FB 1 and FB 2 were detected in 90.6% and 63.3% of compound feed samples.FUMs mainly contaminated cattle feed (97.9% for FB 1 and 78.6% for FB 2 ), followed by poultry feed (90.7% for FB 1 and 65.0% for FB 2 ) and pig feed (82.9% for FB 1 and 45.7% for FB 2 ).The incidence of FB 1 contamination in pig feed was similar to the incidence of FUMs in pig feed reported by Pleadin et al. (2012) [37] (80%).The mean concentration of FB 1 was 1779.0 µg/kg for cattle feed, 1171.1 µg/kg for swine feed and 1382.5 µg/kg for poultry feed, and the mean concentrations of FB 2 were 251.8, 90.5 and 287.9 µg/kg for cattle, swine and poultry feed, respectively.The mean concentrations of FB 1 and FB 2 were 1382.5 and 287.9 µg/kg in poultry feed, which were lower than the estimated value (2733 µg/kg) in poultry feed in Kuwait [33].The maximum contamination value of FUMs was detected at 15,098 µg/kg in a weanling piglet feed sample.This contamination level was over the EC and Korean regulation limit (5000 µg/kg).The concentration of FUMs in six other pig feed samples, as well as the weanling piglet feed mentioned above also exceeded the EC and Korean guidance level.In this study, the concentration ranges of FB 1 (36.4-14,876.8µg/kg) and FB 2 (14.9-9235.8µg/kg) contamination were broader than the ranges from Kuwait (220-6000 µg/kg) [33] and Turkey (2690-4960 µg/kg) [35].
In Kuwait [33], wheat bran, soybean meal and corn feed ingredients were contaminated by FUMs (89.8% of samples), and the mean contamination value was 2040 µg/kg with a range of 0-6000 µg/kg.Another study [42] was conducted to determine the occurrence of various mycotoxins including fumonisins.A total of 7049 samples consisting of corn, soybean/soybean meal, wheat, dried distillers' grains with solubles (DDGS) and finished feed samples was collected from the Americas, Europe and the Asia-Pacific region.FUMs contaminated most of these samples with mean contamination values of 1965 µg/kg.Corn from South America and Southern Europe was mainly contaminated with FUMs, and the contamination incidence was 92% and 90%, respectively.The mean contamination levels in both regions differed; the mean level of FUMs in South American samples (3226 µg/kg) was higher than that in Southern European ones (2271 µg/kg).In 2011-2014, a total of 1384 samples consisting of 295 maize samples, 143 maize silage samples, 466 small grain cereal samples and 480 complete feed samples was collected in Poland [39].Those samples were used to estimate the level of various mycotoxins including FUMs.Among them, 83 maize samples, 21 maize silage samples and three cereal grain samples were analyzed for levels of FUMs.FUMs were detected in 59% of the maize samples with 1885 µg/kg as the maximum level and in 53% of maize silage samples with 108 µg/kg as the maximum level.There were no positive cereal grain samples.The contamination concentration of FUMs in all of the positive samples was lower than the EC's guidance values.In Germany, 84 maize samples were collected in 2006 (44 samples) and 2007 (40 samples).In all, 34% of maize samples were contaminated with FUMs in 2006, and no samples were contaminated with FUMs in 2007 [43].Researchers estimated that the samples collected in 2006 were contaminated with a higher level of FUMs than the ones in 2007 because of the high temperature and low rainfall during anthesis and early grain filling.In 2006, FB 1 was detected in 34% of samples with 1910 µg/kg as the mean concentration, and FB 2 was detected in 23% of samples with 460 µg/kg as the mean concentration.Zachariasova et al. [38] reported that 11 maize silage samples were contaminated with a mean concentration of 35 µg/kg and 26 µg/kg for FB 1 and FB 2 , respectively.In that study, FUMs were mainly found in maize, maize silage and maize-based DDGS.In our previous study, separately performed as part of another research project with different samples in 2012 [44], the results were similar to the results of this study; most compound cattle feed was contaminated with FUMs (98%).Compound feed for early broilers was contaminated with a level of mean contamination of 3459 µg/kg and a range of 333-12,823 µg/kg.The range of contamination levels was between 0.059 and 8.239 for feed ingredient samples (83.3%).
In the present study, feed ingredients were collected in 2011, 2012, 2014 and 2016 in Korea.The incidence of FB 1 was 14.3, 70.4,63.6, 33.3 and 37.5 in grains, grain by-products, meal, fibrous feed and food by-products, respectively.FB 2 contaminated 48.1, 45.5, 44.4 and 12.5% of grain by-products, meal, fibrous feed and food by-products, respectively.FB 1 and FB 2 were found in 56.4 and 40.0% of feed ingredient samples.Kosichi et al. [39] reported that 59% of 83 maize samples were contaminated with a mean of 1885 µg/kg FUMs.In this study, the incidence of FUMs was 92.3% in 26 samples consisting of corn gluten feed, corn bran, corn gluten meal and corn germ meal with mean levels of 1440.5 and 517.6 µg/kg for FB 1 and FB 2 , respectively.The maximum detected levels of FB 1 and FB 2 were 6357.5 and 1798.9 µg/kg in corn bran and 4816.0 and 1662.7 µg/kg in corn gluten meal.As a result, all FUMs concentrations in feed ingredient samples were lower than the guidance limits established by the EC and Korea.

Chemicals and Reagents
Analytical standards of FB 1 and FB 2 were purchased from Sigma Chemical Co.(St.Louis, MO, USA).Acetonitrile and methanol, purchased from Honeywell Burdick & Jackson (Morris Plains, NJ, USA), were used for the extraction of FB 1 and FB 2 .Phosphate-buffered saline (PBS; Sigma-Aldrich, St. Louis, MO, USA) was used as a buffer solution for elution from the immune-affinity column.Ortho-phthaldialdehyde (OPA; Pickering Laboratories, Mountain View, CA, USA) was used for derivatization of FUMs for fluorescence analysis.Fumoniprep (R-Biopharm ® , Darmstadt, Germany) and FumoniTest (Vicam Co., Milford, MA, USA) were applied for solid-phase extraction and purification.A standard solution of FB 1 and FB 2 was diluted with 50% acetonitrile solution (acetonitrile:water = 50:50, v:v).

Sampling of Compound Feed and Feed Ingredients
A total of 535 animal feed samples (425 compound feed and 110 feed ingredients) produced domestically between 2011 and 2016 in Korea were collected in order to evaluate the contamination level of FUMs.These samples were provided by the National Agriculture Products Quality Management Service Experiment Research Institute.These feed samples were prepared by the general guidelines on sampling [45].One kilogram of sample was randomly collected from every ton of compound feed and feed ingredient.Four different samples collected from the same lot were mixed and divided into another four groups according to the guidelines mentioned above.All of the divided samples of 200 g were kept at 4 • C for analysis.Tables 7 and 8 describe detailed data related to collecting compound feed samples and feed ingredients.The classification of compound feed and feed ingredients is shown in the supplementary data (Table S1-S5).

Figure 3 .
Figure 3. Identification of FUMs by LC-MS/MS.An extracted ion chromatogram of standard FB 1 and FB 2 at 250 ppb (A,B) and FUMs in a feed sample (C,D).The ion spectrum (product ion) of standard FB 1 and FB 2 (E,F) and those in a feed sample (G,H).

Figure 4 .
Figure 4. Distribution of FB1 (A) and FB2 (B) in compound feed types (box-plot: whiskers at minimum and maximum, box at P25 and P75 with the line at P50; ° values above the 75th percentile plus 1.5-times the inter-quartile distance; * values above the 75th percentile plus 3.0-times the inter-quartile distance).

Figure 4 .
Figure 4. Distribution of FB 1 (A) and FB 2 (B) in compound feed types (box-plot: whiskers at minimum and maximum, box at P25 and P75 with the line at P50; • values above the 75th percentile plus 1.5-times the inter-quartile distance; * values above the 75th percentile plus 3.0-times the inter-quartile distance).

Figure 4 .Figure 5 .
Figure 4. Distribution of FB1 (A) and FB2 (B) in compound feed types (box-plot: whiskers at minimum and maximum, box at P25 and P75 with the line at P50; ° values above the 75th percentile plus 1.5-times the inter-quartile distance; * values above the 75th percentile plus 3.0-times the inter-quartile distance).

Figure 5 .
Figure 5. Distribution of FB 1 (A) and FB 2 (B) in compound feed between 2011 and 2016 (box-plot: whiskers at minimum and maximum, box at P25 and P75 with the line at P50; • values above the 75th percentile plus 1.5-times the inter-quartile distance; * values above the 75th percentile plus 3.0-times the inter-quartile distance).

Figure 6 .
Figure 6.The distribution of FB 1 (A) and FB 2 (B) in feed ingredients (box-plot: whiskers at minimum and maximum, box at P25 and P75 with the line at P50; • values above the 75th percentile plus 1.5-times the inter-quartile distance; * values above the 75th percentile plus 3.0-times the inter-quartile distance;values mean median value for each feed ingredient).

Figure 7 .
Figure 7. Distribution of FB1 (A) and FB2 (B) in feed ingredients across the years (box-plot: whiskers at minimum and maximum, box at P25 and P75 with the line at P50; ° values above the 75th percentile plus 1.5-times the inter-quartile distance; * values above the 75th percentile plus 3.0-times the inter-quartile distance).

Figure 7 .
Figure 7. Distribution of FB 1 (A) and FB 2 (B) in feed ingredients across the years (box-plot: whiskers at minimum and maximum, box at P25 and P75 with the line at P50; • values above the 75th percentile plus 1.5-times the inter-quartile distance; * values above the 75th percentile plus 3.0-times the inter-quartile distance).

Table 1 .
Guidance values for fumonisins (FUMs) in animal feed in the EU and South Korea.

Table 2 .
Summary of the method validation study.

Table 2 .
Summary of the method validation study.

Table 3 .
FUMs concentrations in various compound feed types.
[31]: number of samples,(b)LC: percentage of left censored results.This value in this experiment is the percentage of data below the LOD[31].

Table 4 .
Concentration mean and differences of mycotoxins in compound feed.

Table 4 .
Concentration mean and differences of mycotoxins in compound feed.

Table 6 .
Concentration mean and differences of FUMs in feed ingredients across the years.

Table 6 .
Concentration mean and differences of FUMs in feed ingredients across the years.