Opposite Direction for Seasonal Variation of A atoxin M1 in Bulk Milk and A atoxin B1 in Rations: Results From a Prospective Study in Selected Dairy Farms of Qazvin Province, Iran


 The presence of aflatoxin M1 (AFM1) in 24h bulk milk and aflatoxin B1 (AFB1) in concurrent total mixed rations (TMR) and feed ingredients were assessed in 12 large dairy operations. The bulk milk was sampled on days 1, 15 and 30 during winter and summer (n=72). Total mixed rations (n=48) and feed ingredients (n=230) were sampled two times with a 30-day interval. Aflatoxin was measured using direct competitive ELISA kits with detection limits of 1-81 ngkg−1 for milk and 1.25-101.25 ngkg−1 for feeds. Aflatoxin M1 was identified in all milk samples (100%), ranging from 2.03 to >81 ngkg−1, with a median of 70 ngkg−1 and averaging 61.25±28.91 ngkg−1 in winter and 54.20±25.51 ngkg−1 in summer (P=0.279). Contaminations <81 ngkg−1 (below the Iranian standard of 100 ngkg-1) were detected in 76% (n=55/72) of samples. Contaminations >81 ngkg−1 were detected in 24% (n=17/72) of samples and were more frequent in winter than in summer (42% vs. 6%). Sixty-nine percent of the winter milk samples (n=25/36) had contaminations above the median (70 ngkg−1). A reverse result was detected in summer. The chance of contaminations above median was higher in winter than in summer (OR=5.33, P=0.007). All TMR and ingredient samples had higher AFB1 contaminations in summer (P<0.05). Six TMR samples had non-detectable (<1.25 ngkg-1) values (5 in winter) and 7 samples had levels >101.25 ngkg-1 (all in summer). The chance of TMR contamination above the median (716 ngkg-1) was 5.57 times higher in summer than in winter (P=0.002). Seventy percent of the TMR samples had contaminations above the median in summer. Elevated levels of AFB1 of rations in summer (1375.50±905.02 vs. 537.05±558.79; P<0.002) did not result in elevation of AFM1 in milk, probably due to reduced AFB1 metabolism in the liver and lower dry matter intakes caused by heat stress. The AFB1 content of grain mix succeeded by corn silage, wet beet pulp, dry beet pulp and alfalfa hay were correlated with TMR contamination. Ration AFB1 and milk AFM1 were not correlated. Based on the results, a great majority of milk produced in the studied farms could have AFM1 contaminations below the Iranian standard limit (100 ngkg-1). Contaminations below 50 ngkg-1 appear to be achievable and affordable. Intensifying the controlling measures in summer, when the feed contaminations are elevated, may reduce the overall milk contamination.

. Widespread contamination of feedstuffs by a atoxins necessitates some regular actions in dairy farms for hazard analysis and de ning the critical control points. A atoxin regulatory limits vary between countries based on their speci c considerations. The most proposed limits for AFM 1 in raw milk are 50 ngkg −1 (as low as reasonably achievable) and 500 ngkg −1 (regards to the carcinogenic effects of AFM 1 ) (FAO/ WHO, 2002;2001). It has been stated in some FAO/WHO reports (2001; 2002) that 500 ngkg −1 AFM 1 may be high because any level of genotoxic carcinogens might pose a health risk to consumers. However, it has also been noted that the limit of 50 ngkg −1 may not be achievable in some regions of the world and the limit of 500 ngkg −1 is adequate for the protection of consumer health. To achieve the strict limit of 50 ngkg −1 , the maximum allowable concentrations of AFB 1  that the level of AFM 1 in raw and pasteurized milk may be well below the ISIRI limit (100 ngkg −1 ) but may exceed the limit of 50 ngkg −1 , and few samples may exceed the limit of 500 ngkg −1 . Also, some studies have assessed AFB 1 contamination of feeds including corn silage, alfalfa hay, straw, barley, wheat bran, beet pulp, canola meal and grain mix and have reported various prevalence of contamination (19. The AFM 1 contamination of milk and its seasonal variations would be best controlled if the level of AFB 1 in the concurrent rations and feed ingredients are de ned. In the present study, performed in a number of large dairy operations during cold and warm seasons, the level and the seasonal changes of AFM 1 were screened in bulk milk to assess the probable hazard. The AFB 1 contaminations of the concurrent rations and separate feed ingredients were also assessed to explain the most critical dietary contaminants in the studied farms.

Farms and samplings
The study was done in twelve selected dairy farms in Qazvin province, north-west, Iran, during winter and summer, 2019. The criteria for selecting the farms were having distinct role in the provision of the total milk produced in the province, being scattered in various parts of the province, willing for cooperation and provision of the requested data. The farms owned about 17,700 milking cows during the study period and produced about 40% of the total daily raw milk in the province (Cooperative Organization of Dairy Farms, Qazvin province; personal communication). The average milk production of the farms within the year of study was 30.5 kg/cow/day. In each season, in a 30-day period, bulk milk and rations were sampled. The bulk milk produced within the last 24 hours was sampled on days 1, 15 and 30 (n=36/season, 100ml in screw tap sterile bottles). Samples were taken according to the procedure No 87-44-05 communicated by Iranian Veterinary Organization (IVO, 2005) from a depth of about 20cm of the top surface in the milk tank. The stirrers were working for about 10 minutes before sampling. The samples were transferred cool to the laboratory and were kept frozen (-19°C) for about 30 days before the measurement of AFM 1 .
Sampling of feeds was preformed 2 days before the rst and the last bulk milk sampling. The total mixed rations (TMR) of the milking cows were sampled two times (30 days apart) from the mangers immediately after the distribution of the morning meal according the procedure described by Robinson and Meyer (2010) (n=24/season; about 0.5kg each). The ingredients of the rations (Table 1) were also sampled (n=230, both seasons; about 0.5kg each) in zipped nylon bags (Procedure No 84-44-06; IVO, 2008). The wet samples (TMRs, corn silage and wet beet pulp) were transferred cool to the laboratory and were oven dried at 50 to 70°C (24 hours; Memmert, UM200, Germany). The samples were powdered in a mill carefully cleaned for each sample, were heated at 105°C for 24 hours and AFB 1 was measured in dry matter. Completion of sampling, transferring and collecting the required data (e.g. the formulations of the rations) all were performed by one trained person and took about two months in each season. Storing and preparation of the samples and measurement of AFM 1 and AFB 1 on milk and feed samples were done at the central diagnostic laboratory of the Veterinary Organization, Qazvin, Iran.
Measurement of AFM 1 in milk and AFB 1 in feeds The concentration of AFM 1 in milk and AFB 1 in feeds were determined using direct competitive ELISA kits (Bioassay Technology Laboratory, China). The kits were stored at 2-8ºC, and before using they were put at room temperature (25±2ºC) for two hours. All steps of the experiments were done avoiding direct sun light.

AFM 1
The frozen milk samples were thawed at room temperature (25±2ºC) and after reaching to thermal equilibrium, they were centrifuged at 3000g for 10 minutes.
The skimmed milk was used for AFM 1 determination. Ten ml of each sample was mixed with 20 ml of 70% ethanol in a screw-tap vial and was shaken vigorously for 3 minutes. Then, the mixtures were centrifuged at 3000g and 100 µL of the supernatant liquid was mixed with 400 µL of the diluent solution.
Using the provided microplate, 50 µL of the standard solutions (0.1, 0.3. 0.9, 2.7 and 8.1 ppb of AFM 1 ) were put in the corresponding wells for preparing the standard curve. The same amounts of the prepared samples were put in the other wells. To all wells, was added 50 µL of anti-AFM 1 conjugate antibody. The plate was shaken gently for a few seconds and then was warmed in 37ºC bath for 30 minutes. The liquid in the holes was emptied and the microplate was washed 5 times (Bio-Tek ELx50, Bio-Tek Instruments, USA) using the washing solution. To all holes, were added 50µL of color solutions A and B and the plate was put in 37ºC bath for 10 minutes (formation of blue color). Finally, 50µL of the stop solution was added to each hole to terminate the reaction (formation of yellow color). The absorbance of the samples was determined at 450nm within 5 minutes (Bio-Tek ELx800, Bio-Tek Instruments, USA). The absorbance value was inversely related to the concentration of AFM 1 . The analytical range of the AFM 1 kit was 1-81 ngkg −1 . The values were reported as ngkg −1 (ppb).

AFB 1
Five grams of each powdered feed sample was mixed with 25 ml of 70% ethanol in a screw-tap vial, was left at room temperature overnight and then was shaken vigorously for 10 minutes. Then, the samples were centrifuged at 3000g and the supernatant liquids were ltered on Whatman 1 papers. Exactly 100 µL of the ltered liquid was mixed with 400 µL of the diluent solution. The other steps for measuring AFB 1 were the same as described for AFM 1 (Table 1) were reported as measures calculated from the weights and AFB 1 contaminations of separate ingredients (calculated TMR) and also as the contaminations of the samples taken from the bunks (bunk TMR).

Statistical analysis
The results were statistically studied using SPSS software (version 24, Illinois, USA). For a atoxin values out of the analytical range of the corresponding kit (1-81 ngkg −1 for AFM 1 ; 1.25-10.25 ngkg −1 for AFB 1 ) the lower and the upper limits were substituted as the lowest and the highest detected values, respectively.
Descriptive statistics were presented as means and standard deviations. The a atoxin levels in milk and bunk TMR samples were classi ed into two relatively equal-sized groups based on their medians. Association of the a atoxin with season was evaluated using generalized estimating equations (GEE) with binomial distribution and logit function. Farm was introduced into the model as subject effect and season as well as number of samplings in each season were considered as repeated effects to account for the dependence between measurements. Separate models were constructed for AFM 1 and AFB 1 in milk and bunk TMR, respectively. The a atoxin concentration in the calculated TMR and ration ingredients as well as ration DM (bunk and calculated, see below) of winter and summer were compared using linear mixed models. Farm was considered as random effect, and season and number of samples as xed effects in the model. For ration ingredients, an additional xed effect for type of feeds was introduced into the model followed by multiple comparisons with Bonferroni adjustment. The relationships between milk AFM 1 with AFB 1 in the TMR (bunk and calculated) and the separate ration ingredients were assessed by Spearman's rho correlation test. The P-value less than 0.05 was considered signi cant. Table 1 Weight, AFB 1 concentration and the contribution of ration ingredients in total contamination of the ration   Table 1 Weight, AFB 1 concentration and the contribution of ration ingredients in total contamination of the ration   Regarding the seasonal variations, contaminations >81 ngkg −1 were more frequent in winter (42% in winter vs. 6% in summer; Figure 2). Examining the median AFM 1 contamination, 69% of winter samples (n=25/36) had contaminations above 70 ngkg −1 and a reverse result was detected in summer (Table 2). In fact, the results of GEE for association of milk a atoxin with season showed the chance of contaminations above 70 ngkg −1 was 5.33 times higher in winter than in summer (OR=5.33, P=0.007).
A atoxin B 1 in Feed ingredients and TMRs: The weights (DM), AFB 1 contamination and the share of each ingredient in the whole ration contamination, and total AFB 1 content of the TMR are shown in Table 1. The data on TMR were calculated from the weight and AFB 1 contamination of separate ingredients (calculated TMR). The contamination of the TMR samples taken from the bunks are shown in separate rows (bunk TMR). Two farms did not provide the ration details and so, only the contamination of their feed samples is shown.
The average AFB 1 contaminations of separate ration ingredients (n=230) and the frequencies of non-detectable (<1.25 ngkg -1 ) and high (>101.25 ngkg -1 ) contaminations are depicted in Table 3  The average AFB 1 content of the TMRs (Table 4) was higher in summer than in winter for both calculated (P<0.001) and bunk (P=0.002) TMRs, so that each cow could potentially consume at least 2.4-2.6 times more a atoxin in summer. The bunk samples were contaminated higher than the calculated TMRs.
Among the bunk TMRs (n=48), 6 samples had non-detectable values (5 in winter) and 7 samples had levels above 101.25 ngkg -1 (all in summer; Table 4). The median contamination of bunk TMRs was 716 ngkg -1 and the GEE results showed that the chance of contaminations above median was 5.57 times higher in summer than in winter (OR=5.57, P=0.002). Seventy percent of the bunk TMR samples had contaminations above the median in summer (Figure 3). Table 4 The concentration (ngkg -1 DM) and the total daily content (ng) of AFB 1 (mean±SD)* in TMRs of the studied farms * Assuming that the values of 1.25 and 101.25ngkg-1 (the lower and the upper limits of the experimental kit) were the lowest and the highest contamination rates; TMR: total mixed ration ** Excluding the two farms with incomplete data (totally 40 bunk TMR samples), 5 samples had non-detectable values (4 in winter) and 5 samples had levels above 101.25ng/kg (all in summer)

Correlations
No correlation was detected in neither of seasons between the ration AFB 1 and milk AFM 1 . There were some correlations between the AFB 1 content of TMRs and the amount of AFB 1 added to the ration by each ingredient ( Table 5). The most frequent and the strongest correlations were seen for grain mix succeeded by corn silage, wet beet pulp and dry beet pulp, respectively. The AFB1 in alfalfa hay was correlated only with that of calculated TMR in the sum of both seasons, irrespective of its higher contaminations compared with the other ingredients. The correlations were less prominent for bunk TMRs.
The AFB 1 levels in calculated and bunk TMRs were related in summer (r=0.48, P=0.033) and in the sum of both seasons (r=0.43, P=0.006), but not in winter (r=-0.05, P=0.83). Table 5 Correlations (r values) between the a atoxin level in calculated or bunk TMR and the a atoxin added to the ration by each ingredient (P values in parentheses) In the present study, however, the AFB 1 contaminations of all rations and feed ingredients were signi cantly higher in summer than in winter (see below).

A atoxin B 1 in Feed ingredients and TMRs
Ninety percent of the examined ingredients (n=207/230) were positive for AFB 1 , 10% (n=23/230) had non-detectable values and 13.9% (n=32/230) had values above 101.25 ngkg −1 . All ingredients exhibited higher contaminations in summer (P<0.05). The non-detectable AFB 1 values were mostly seen in winter (n=17/23) and the values above 101.25 ngkg −1 were detected mainly in summer (n=27/32). These differences could be due to the synergistic effects of ambient temperature and feed moisture on behavior of mycotoxigenic fungi in summer, which is affected by climatic changes at any stage of production Thus, the true contaminations of the ingredients could be higher than the detected levels. As the AFB 1 content of TMR samples taken from bunks were higher than those calculated from the contaminations of separate ingredients, the contamination could be increased during processing. Common devices such as mills and mixing wagons (Pinotti et al., 2016) and remainders of contaminated feeds in wagons, preparation areas and bunks may increase the level of a atoxin in the ration. The difference could also be due to the small quantities of samples taken from huge volumes of feeds.
The lack of correlation between feed AFB 1 and milk AFM 1 could be due to variations in daily intake of AFB 1 . contamination of milk would potentially be elevated. Thus, the most logic way to control the a atoxin contamination in milk is to control the critical contaminants of the rations.

Critical feed contaminants
The AFB 1 content of all ration ingredients showed some correlations with AFB 1 content of TMRs. The strongest correlations were seen for grain mix and corn silage, which had the highest proportions in the rations. However, the impact of all other ingredients should also be taken into account. Beet pulp also could be an important source of contamination (strong correlations) regardless of its low incorporation or absence in rations. Alfalfa hay also could be an important contaminant, despite that its AFB 1 contents were not correlated with those of the total rations. It is incorporated in nearly all rations in considerable amounts in Iranian farms and may be substituted for corn silage in many conditions. The co-occurrence and the synergistic effects of various mycotoxins even at relatively low levels (Kovalskey et al., 2016) should be kept in mind. In addition to controlling the AFB 1 contaminations of feed ingredients, tainting of feedstuffs during processing should also be controlled. Intensifying the controlling measures in summer, with probable higher contaminations of feedstuffs, may reduce the overall milk contamination. Responses to these efforts may be rapid as AFM 1 enters the milk 12-24 hours after consumption of AFB 1 and drops to non-detectable levels about 72 hours of removal of AFB 1 from ration (Pettersson, 1998). Based on the results of this study, a great majority of milk produced in the studied farms could have AFM 1 contaminations below the ISIRI limit (100 ngkg −1 ). Contaminations below 50 ngkg −1 appear to be achievable and affordable in the studied farms. Figure 1 Frequencies (%) of bulk milk samples (n=72) with AFB1 contaminations below 50, 50-81 and above 81 ngkg-1.

Figure 2
Frequencies (%) of bulk milk samples with AFM1 contaminations below and above 81 ng/kg during winter and summer (n=36 each season).