Cocktails of Mycotoxins, Phytoestrogens, and Other Secondary Metabolites in Diets of Dairy Cows in Austria: Inferences from Diet Composition and Geo-Climatic Factors

Dairy production is a pivotal economic sector of Austrian and European agriculture. Dietary toxins and endocrine disruptors of natural origin such as mycotoxins and phytoestrogens can affect animal health, reproduction, and productivity. This study characterized the profile of a wide spectrum of fungal, plant, and unspecific secondary metabolites, including regulated, emerging, and modified mycotoxins, phytoestrogens, and cyanogenic glucosides, in complete diets of lactating cows from 100 Austrian dairy farms. To achieve this, a validated multi-metabolite liquid chromatography/electrospray ionization–tandem mass spectrometric (LC/ESI–MS/MS) method was employed, detecting 155 of >800 tested metabolites. Additionally, the most influential dietary and geo-climatic factors related to the dietary mycotoxin contamination of Austrian dairy cattle were recognized. We evidenced that the diets of Austrian dairy cows presented ubiquitous contamination with mixtures of mycotoxins and phytoestrogens. Metabolites derived from Fusarium spp. presented the highest concentrations, were the most recurrent, and had the highest diversity among the detected fungal compounds. Zearalenone, deoxynivalenol, and fumonisin B1 were the most frequently occurring mycotoxins considered in the EU legislation, with detection frequencies >70%. Among the investigated dietary factors, inclusion of maize silage (MS) and straw in the diets was the most influential factor in contamination with Fusarium-derived and other fungal toxins and metabolites, and temperature was the most influential among the geo-climatic factors.


Introduction
Dairy production is the most important agricultural sector in the Republic of Austria, representing 18% of the national agricultural production [1]. Animal feeding is a fundamental element of milk production, affecting the rest of the productive chain, including aspects such as animal health and performance as well the quality and safety of the derived foods [2]. The composition of dairy cattle diets varies widely among farms and production systems worldwide, incorporating a broad range of ingredients including roughage, cereal grains, and agroindustrial by-products [2]. The physiological nature of ruminants makes forages (including pastures and conserved forages: silages, hay, and straw) the most adequate and important feed sources for dairy cattle [3]. Additionally, the incorporation of high-density energy dietary sources (concentrate feeds) is essential to achieve the high milk yields demanded and expected in modern dairy farming [2,4,5]. Such diversity of ingredients contributes to the dietary exposure to a broad spectrum of toxic, potentially toxic, and endocrine-disrupting fungal and plant secondary metabolites [6][7][8][9][10].
Crops and feedstuffs are susceptible to mould infection and colonization with subsequent contamination with mycotoxins and other fungal secondary metabolites during the feed-production chain, both pre-and post-harvest, influenced by several biotic and abiotic factors [11]. Probably based on the paradigm that ruminants are less susceptible to the negative effects of fungal toxins [12], most studies concerning mycotoxins and animal feeds have focused on monogastric animals and their main dietary sources (cereal grains). However, a wide spectrum of fungal metabolites (several of them toxic and potentially toxic), primarily produced by Fusarium, Alternaria, Aspergillus, Penicillium, and other fungal species, has been found in cattle feed sources beyond cereal grains [6,13]. Some of the mycotoxins are included in the European legislation, which currently establishes a maximum limit for aflatoxin B1 (AFB1) and guidance values (GV) for zearalenone (ZEN), deoxynivalenol (DON), T-2 and HT-2 toxins, fumonisins B1 and B2 (FB1 and FB2), and ochratoxin A (OTA), and thus, their occurrences and levels in feeds and diets have received strong attention [14][15][16]. More recently, monitoring studies on contamination frequency and levels of ergot alkaloids (EAs) and emerging mycotoxins in animal feeds have been highly advocated [17][18][19][20][21][22][23]. Consequently, the characterization of the implicated mycotoxin mixtures needs to be performed with an innovative and holistic approach based on multimetabolite analyses to achieve an optimal risk assessment [24]. Such multi-metabolite analytic approaches are relevant because, additionally to single negative effects, there are multiple toxicological interactions (such as addition, synergism, potentiation, and antagonism) among mixtures of mycotoxins and other metabolites, which could have implications for health and reproduction. These interactions require more investigation [25,26]. Beyond toxic fungal metabolites, the dairy cattle diet contains substantial levels of plant secondary metabolites, some of which may induce unfavourable impacts on the health and/or reproduction of livestock, such as pyrrolizidine alkaloids, cyanogenic glucosides (CGs), and phytoestrogens (PEs) [27]. Phytoestrogens can act as endocrine disruptors, impairing reproductive functions, generating temporal infertility, and potentially reducing the productive efficiency of dairy herds [9,[28][29][30][31]. Interestingly, mycoestrogens (such as ZEN, alternariol (AOH), and their modified forms) and PEs (such as isoflavones) have synergistic effects [32][33][34], which must be considered in the context of a complete risk assessment on livestock reproductive performance [35][36][37][38]. Fungal and plant growth as well as concentrations of secondary metabolites in the dietary components and finally in the complete rations are influenced by multiple factors such as plant species/varieties, infecting/colonizing fungal species/varieties, climatic conditions, geography, parasitic/symbiotic interactions, use of pesticides, and other agricultural practices utilized [39][40][41][42][43][44][45][46]. The most influential factors favouring mycotoxin contamination and PE production of feedstuffs and diets of dairy cows should be studied. More data in this field would contribute to developing preand post-harvest preventive and management strategies to reduce exposure and optimize the health and productive performance of livestock farming [6,39,46]. Several studies have analysed the occurrence of some mycotoxins in different types of feed ingredients, including in pastures, cereals, and silages [6,7,22,42,[47][48][49]. Research on the incidence of mycotoxins and other fungal secondary metabolites in complete diets (i.e., TMR) of cattle has been carried out during the last decade; however, it is still scarce [50][51][52][53][54][55][56]. Targeting the dietary levels of toxins and endocrine-disrupting metabolites is vital to assessing the risks for impacts on health, reproduction, and production [10,24]. Moreover, the whole-diet approach applied across many farms with different farm characteristics and feeding management could reveal true high-risk ingredients in dairy rations. Thus, the current study determined the frequency, levels, and co-occurrences of a wide spectrum of mycotoxins, PEs, and other secondary metabolites in representative samples of lactating cows' diets in 100 Austrian dairy farms, using a validated multi-metabolite liquid chromatography/electrospray ionization-tandem mass spectrometric (LC/ESI-MS/MS) method. Inclusion levels of the basal feed ingredients and their characteristics (chemical composition, particle size, hygienic status), dietary forage proportion, and geo-climatic factors (such as altitude, temperature, relative air humidity, and rainfall) were evaluated for their contribution to the dietary concentrations of mycotoxins, PEs, and other secondary metabolites.

Type of Rations and Main Dietary Components
The participating farms fed three kinds of dietary rations to cows: (i) partial mixed ration (87%) and (ii) exclusively forage-based mixed rations (11%), both with separately fed concentrate, as well as (iii) total mixed rations (2%). The frequency and rate of the inclusion levels of the main dietary ingredients in the rations of Austrian dairy cows are shown in Table 1. Grass silage (GS) and MS were the most common forages incorporated in the rations of the visited Austrian dairy farms, presenting frequencies of inclusion over 80% and representing maximums of around 87% and 59% of the rations, respectively. About 60% of the farms used straw in the rations, with maximal inclusion of 10% on a dry-matter basis. Hay was included in around 18% of the evaluated diets, representing from 0.6% to 30% of the ration. Wet brewery's spent grains (BSG) were included in 27% of the diets, with the maximal inclusion level of 13.5% of the total diet. Other silages (e.g., wheat, oats, barley, sunflower, and beep pulp) were included in 10% of the diets, with a maximal inclusion of 23.6% of the rations. The average forage-to-concentrate ratio was 66:34 (Table 1).

Chemical Composition and Particle Size Distribution of Basal Rations
Farms showed variation in the chemical (proximate) composition of the basal ration ( Table 1). The dry matter of the basal rations ranged from 25.7% to 54.6%. The basal rations contained an average of around 50% neutral detergent fibre (NDF), ranging from 36.8% to 75.2%. Non-fibre carbohydrate (NFC) ranged from 0.4% to 41.3% (average: 23.3%), and crude protein ranged from 10% to 21.2% (average: 15.4%). Values of ash and crude fat also showed a wide range. Farms used rations with considerable variation in terms of the distribution of the particle sizes (Table 1). Large particles (>19 mm) represented the main particle size in the ration, accounting for 46.8 ± 18% (mean ± SD) of the ration (as-fed basis). Particles of 8-19 mm and 1.18-8 mm represented similar proportions in the ration, with averages of 22.7% and 25.6%, respectively. Finally, the proportion of fine particles (<1.18 mm) represented on average 4.6% of the ration. The value reached a maximum of 13.7% (Table 1).

Hygienic Status of the Main Dietary Ingredients
The hygienic status of the main components of basal rations (GS, MS, straw, hay, BSG, and concentrate) was determined by sensory evaluation and scored as "proper", "minor deficiencies", "significant deficiencies", and "vast deficiencies" according to Kamphues et al., 2014 [57]. Most samples (>80%) of dried feedstuffs including straw, hay, and concentrates showed a proper hygiene score (Table 1). Wet conserved feeds presented major hygienic status concerns. MS was the feedstuff most often (over 50%) detected for hygiene deficiencies (minor to vast deficiencies). Ensiled grass presented minor deficiencies in hygienic status in 30% of the samples, significant deficiencies in 8%, and vast deficiencies in 3%. Around 44% of the BSG was not in proper hygienic conditions.  (Table 2). Their occurrences and respective average (with SD), median, and range of concentrations (expressed on a dry-matter basis in µg/kg) are indicated in Table 2. The detected metabolites were categorized in groups based on their main producers, consisting of Alternaria, Aspergillus, Fusarium, Penicillium, lichen-associated fungi, other fungal species, other plant metabolites, and unspecific (i.e., derived from fungi, bacterial and/or plants) metabolites, or corresponding to the kind of metabolites, such as EAs and PEs, according to previous reports [42,58,59]. Fusarial metabolites were detected in all samples and with the highest grade of diversity, with 35 different compounds identified ( Table 2). Lower numbers of detected metabolites were derived from Penicillium (23), other fungal species (21), Aspergillus (16), Alternaria (11), and EAs (13). High occurrences (>90%) were detected for the groups of fungal metabolites (Fusarium, Alternaria, Aspergillus, and Penicillium), except for the total EAs (32.3% of total samples) and compounds produced by lichen-associated fungi (16.2%) ( Table 2). Regarding the dietary contamination levels, the group of fungal metabolites with the highest average concentration was Fusarium (1380 µg/kg), followed by Alternaria (445 µg/kg), Penicillium (205 µg/kg), Aspergillus (177 µg/kg), other fungi (115 µg/kg), EAs (19.5 µg/kg), and minor grade lichen-associated fungi (4.57 µg/kg) ( Table 2). As displayed in Figure 1, the distribution of the concentrations among groups of metabolites varied widely.
As presented in Table 2, ten different PEs and six additional plant metabolites were identified across all samples. Most of these plant metabolites occurred in high frequencies and high concentrations, with average concentrations of total PEs and other plant metabolites above 70,000 µg/kg and 3000 µg/kg, respectively. A high degree of variation among the samples was marked (Figure 1), with ranges from 1080 µg/kg to 411,000 µg/kg for total PEs and from 5.37 µg/kg to 24,500 µg/kg for the total of other secondary plant metabolites ( Table 2). All diets were detected for unspecific metabolites ( Table 2). The total concentrations of this category presented an average of 20,000 µg/kg and ranged from 3740 µg/kg to 52,400 µg/kg. The concentration heterogeneity was evident for this group of metabolites ( Figure 1).     As presented in Table 2, ten different PEs and six additional plant metabolites w identified across all samples. Most of these plant metabolites occurred in high frequenc and high concentrations, with average concentrations of total PEs and other plant met olites above 70,000 μg/kg and 3000 μg/kg, respectively. A high degree of variation amo the samples was marked (Figure 1), with ranges from 1080 μg/kg to 411,000 μg/kg for to

Mycotoxins Included in the EU Legislation and Related Compounds
The mycotoxins with GV in the European legislation but not the strongly regulated AFB1 were found in the dietary rations tested in the present study ( Table 2). The level of occurrences and heterogeneity in concentrations across samples differed among these mycotoxins ( Figure 2A). Accordingly, DON, ZEN, and FB1 were the most abundant and frequently found regulated mycotoxins ( Table 2). Type A trichothecenes, T-2 toxin, and HT-2 toxin were detected in frequencies <30%. Metabolites structurally and toxicologically related to the regulated fusarial metabolites, including DON-3-glucoside, nivalenol (NIV), monoacetoxyscirpenol, HT-2 glucoside, FA1, FB3, and FB4, occurred in the studied diets but at lower frequency compared to their parental form (Table 2, Figure 2A). Of these, NIV showed the highest concentration (range: 34.6-804; mean 311 µg/kg). The mycotoxin OTA (produced mainly by Penicillium spp. but also by Aspergillus spp.) was detected only in 1% of the samples and in low concentrations (<8 µg/kg). In total, 13 different EAs were identified. The individual levels of EAs detected in the evaluated samples of diets' averages were below 12 µg/kg and presented maximum concentrations less than 65 µg/kg, and their occurrences were lower than 20% ( Table 2). The concentration distribution across samples was similar among the EAs ( Figure 2B).

Other Mycotoxins and Metabolites from Fusarium, Alternaria, Aspergillus, and Penicillium
Additionally, many other less-known mycotoxins and metabolites associated with Fusarium, Alternaria, Aspergillus, and Penicillium were found in the diets of Austrian dairy cows (Table 2, Figure 2). Metabolites produced by Fusarium, including 15-hydroxyculmorin, antibiotic metabolites, and W493, were found in more than 40% of the samples, whereas acuminatum B, apicidin D2, chrysogine, fusaproliferin, and fusapyrone had lower occurrences at below 10%. Concerning other compounds derived from the genus Alternaria, infectopyrone (78.3%) and altersetin (47%) were the most frequently found metabolites, after the previously mentioned TeA (Table 2). In terms of concentrations, infectopyrone was the major contaminant produced by Alternaria ( Figure 2D). Multiple compounds produced by members of the genus Aspergillus and Penicillium were detected in diverse frequencies of occurrence and contamination levels. Most of the Aspergillus and Penicillium secondary metabolites were detected in rates lower than 10% of the samples and presented average and median concentrations below 100 µg/kg. For Aspergillus-derived metabolites, while kojic acid showed the highest mean concentration (165 µg/kg), flavogluacin was the most frequently found metabolite and presented high concentration heterogeneity across samples (Table 2, Figure 2E). Fellutanine A and phenopyrrozin were the most frequently found Penicillium metabolites and had relatively high mean concentrations as compared to other Penicillium metabolites (Table 2, Figure 2F).
The co-occurrence rates of PEs, other plant-derived metabolites, and mycoestrogens (AOH, AME, TeA, and ZEN) are illustrated in Figure 6. All tested samples presented cocontamination between biochanin and genistein. Samples often presented with mixtures of PEs with high occurrences (>70%), including the metabolites coumestrol, daidzein, daidzin, genistein, and genistin. Many of the PEs co-occurred with the mycoestrogens in more than 30% of the samples. Particularly, ZEN and TeA showed relatively higher co-occurrences of PEs compared with the mycoestrogens from Alternaria. The co-occurrence rates of PEs, other plant-derived metabolites, and mycoestrogens (AOH, AME, TeA, and ZEN) are illustrated in Figure 6. All tested samples presented cocontamination between biochanin and genistein. Samples often presented with mixtures of PEs with high occurrences (>70%), including the metabolites coumestrol, daidzein, daidzin, genistein, and genistin. Many of the PEs co-occurred with the mycoestrogens in more than 30% of the samples. Particularly, ZEN and TeA showed relatively higher cooccurrences of PEs compared with the mycoestrogens from Alternaria.

Dietary Composition and Geo-Climatic Factors in Relation to the Concentration of Mycotoxins, Phytoestrogens, and Other Secondary Metabolites
Correlations between recorded dietary and geo-climatic factors (see Table 1) with the contamination with fungal (toxic) metabolites of interest were screened using Spearman
Multiple regression models of the log-transformed concentration values of compounds derived from species of Alternaria, Fusarium, and Penicillium, total fungal and some individual mycotoxins such as ZEN, DON, ENNs, BEA, CUL, as well as the sum of FB1 and FB2 are presented in Table 3. Influences of some dietary factors based on the simple correlation method were confirmed by a multiple regression approach. Importantly, the multiple regression approach revealed a joint effect of multiple factors attributed to the dietary concentration of mycotoxins. Inclusion levels of MS and straw, the proportion of particles >19 mm, and dietary NFC content affected total concentrations of Fusarium metabolites. Together, these factors explained 52% of the variance, which is the highest value observed in this present study. Specifically, the proportion of MS and its combination with straw positively influenced the contamination levels of Fusarium-derived metabolites (slope = 0.004, p = 0.042, Table 3 and Figure S1). As shown in Figure 7, at the same level of MS, farms using more straw showed higher Fusarium contamination and vice versa. Interestingly, a quadratic effect of the proportion of MS was observed, and the total Fusarium metabolite peaked at an MS level around 30-35% of the basal diet DM before dropping to a higher MS level (Figure 7). A similar outcome was observed via logistic regression analysis that estimated an odds ratio of 1.05 (95% confidence limits: 1.01-1.08) and predicted a close to 75% chance for high loads of Fusarium metabolites at MS inclusion level of 30% of the diet DM (Supplementary Data Table S2 and Figure S1). For individual fusarial mycotoxins, the inclusion of MS positively influenced the contamination level of DON and emerging mycotoxins BEA, CUL, and SIC, while the proportion of particles >19 mm negatively influenced the contamination of ZEN, CUL, SIC, and total Fusarium metabolites. However, this depended on the inclusion level of MS. As shown in Figure 7, when no MS was used, the contamination of Fusarium metabolites increased with an increment in the proportion of particles >19 mm. With the inclusion of MS, the effect of MS dominated the effect of the proportion of particles >19 mm. Only ZEN was related to the level of ether extract (i.e., crude fat) of basal diet and hygiene score of GS. In agreement with simple correlation analysis, the inclusion level of straw affected the contamination level from Alternaria but with a significant quadratic effect. The influence of the temperature during the maize's growing season was only confirmed for the BEA contamination. The influence of the dietary proportion of BSG on the total ENN commination was confirmed by the multiple regression approach; however, it explained only 19% of the variance. In addition to the factors listed above, the ash content in basal diets (with a positive quadratic response) contributed to the concentration of total fungal metabolites. None of the factors studied substantially explained the concentration of phytoestrogens and plant metabolites (data not shown).  temperature during the maize's growing season was only confirmed for the BEA contamination. The influence of the dietary proportion of BSG on the total ENN commination was confirmed by the multiple regression approach; however, it explained only 19% of the variance. In addition to the factors listed above, the ash content in basal diets (with a positive quadratic response) contributed to the concentration of total fungal metabolites. None of the factors studied substantially explained the concentration of phytoestrogens and plant metabolites (data not shown).

Discussion
Fusarium metabolites were the most relevant fungal contaminants in the rations of dairy cattle surveyed in the present study, corroborating again the importance of Fusarium as one of the most widespread mycotoxigenic species in crops and the main contributor to mycotoxin contamination in animal feeds [7,64,65]. Among the EU-regulated mycotoxins, the type B trichothecene DON (occurrence: 92%) was predominant, followed by the mycoestrogen ZEN (77%) and FB1 (71%). The type A trichothecenes, T-2 and HT-2 toxin, which are more cytotoxic than the type B trichothecenes [66], were detected in low frequencies (12% and 18%, respectively) and concentrations (on average < 30 µg/kg). The contamination levels of these regulated mycotoxins were not over the guidance values of the European Union for feeds of dairy cattle. However, it has been proven that even dietary contamination under the EU values can negatively affect the performance, digestion, and immunity of dairy as well as beef cattle [67]. We showed that numerous non-regulated emerging and modified toxins produced by Fusarium spp. were even more recurrent and presented higher contamination levels. Although the tangible implications resulting from exposure to modified and emerging mycotoxins are not properly characterized, it is known that these compounds interact with other well-recognised fungal toxins, increasing their toxicological activity [24,25]. The high occurrences, concentrations, and diversity of metabolites derived from Fusarium spp. confirm the omnipresence and relevance of this genus in the mycotoxin contamination of crops and animal feeds [64,65]. Emerging Fusarium mycotoxins ENNs and BEA have antibacterial and cytotoxic properties; however, their implications for health and performance in ruminants are underexplored [21,68]. Research on the impact of such kinds of compounds on rumen ecology and functionality is crucial [12,61,69]. Maize silage and straw were the main forage components that drove the increasing concentration of Fusarium metabolites, which lined up with previous studies in the Netherlands [70,71] and Spain [50]. Viable Fusarium spp. is rarely isolated in ensiled maize, suggesting that Fusarium species do not grow properly during the ensiling process [72]. However, it has been widely proposed that mycotoxins of Fusarium spp. are mainly produced during crop growing [73,74] and, therefore, field conditions such as temperature influence mould proliferation and mycotoxin synthesis, supporting the notion that global warming promotes mycotoxin contamination in crops and feeds [46,[75][76][77]. Various studies mark the key effects of temperature and humidity on mycotoxin contamination [39,42,46]. Based on our correlation analysis, environmental temperature increments during the crop's late growing season (June to September) and the sampling month were associated with a higher accumulated concentration of some Fusarium mycotoxins (type B trichothecens, AUR, and BEA) and Penicillium metabolites, respectively. However, its significance was not confirmed by the multiple regression approach, except for BEA. We also did not observe significance for humidity. This might be explained by the accuracy of available climatic data when studying dietary contaminations coming from multiple sources (self-produced and purchased feed as well as different time of storage). This means that spot or average climatic data do not match the concentration at sampling as precisely as studies of single feed sources such as pasture [42].
In general, we did not observe dietary concentrations of regulated mycotoxins exceeding the EU maximum limit and GVs. Compared to the earlier study in Spain by Rodriguez-Blanco et al. (2020), we observed higher occurrences of regulated Fusarium mycotoxins. The researchers studied a similar number of total mixed rations (n = 193) from different areas of Spain during the period from February 2016 to January 2018 and found that DON (16.6%), ZEN (16.0%), and the sum of FB1 and FB2 (34.2%) presented lower occurrences and slightly higher average concentrations than those found in our study. However, all the samples showed values under the EU recommendations [78]. Other mycotoxin surveys performed in several European countries have also evidenced high occurrences and contamination levels of Fusarium mycotoxin in MS [61,79]. Dreihuis et al.
(2008) estimated the dietary intake of four mycotoxins (DON, ZEN, ROQC, and MPA) of high-producing dairy cows in different regions of the Netherlands. The detected mean concentrations of DON, ZEN, ROQC, and MPA in complete diets were 273 µg/kg, 28 µg/kg, 114 µg/kg, and 54 µg/kg, respectively. Consistent with our findings, they reported that MS was the major feed source of these mycotoxins in the diet [70]. Similarly, other studies underline MS as the potential feed source of Fusarium mycotoxins [61,78]. Matching our results, Fusarium-derived mycotoxins were the most recurrent fungal contaminants with the highest concentrations detected in total mixed rations of Brazilian feedlots [54]. Europe-based studies, including the present research, rarely report the detection of AFB1. Nevertheless, this was the case in a recent study on Lithuanian dairy farms [56]. In that study, the analysis of total mixed rations (n = 51) collected in 2019-2020 showed that 60.8% of the rations were positive for AFB1, 54.9% for DON, 49% for ZEN, and 29.4% for T-2 toxin, and AFB1 exceeded the maximum concentration limits in haylage samples [56]. Moreover, the maximum average concentrations of AFB1 and T-2 toxin were found in the GS samples, while some samples of ensiled maize had ZEN and DON concentrations exceeding the EU GVs. Relating to toxic compounds produced by Aspergilli, the absence of strongly regulated AFB1 and other AFs was expected, because the occurrence of these mycotoxins in central Europe has been considered rare [46]. However, we detected precursors of AFs, such as averufin, STC, and versicolorin C [79,80], albeit at low frequencies (<20%) and concentrations (<11 µg/kg). Regarding STC, it has been suggested that this mycotoxin can be produced pre-and post-harvest [81]. Like AFs, STC is a known carcinogenic with immunotoxic and immunomodulatory activity. In general, the information available on exposure data of dairy cows to these precursors of AF is still very limited [18,60]. Fungi of the genus Bipolaris, Chaetomium, and Emiricella are able to synthesize STC [82]. OTA, considered in the European regulation, is produced by Penicillium and Aspergillus spp. and presented very low occurrence and contamination levels in the present survey, which suggests that this mycotoxin presents a minor risk for Austrian dairy herds. Additionally, kojic acid, produced primarily by Aspergillus spp. but also by Penicillium and Acetobacter fungi [83], has been shown to have low toxicity for human macrophages, along with antibacterial and immunomodulatory properties [84][85][86]. In the present study, due to low frequencies as well as high heterogeneity of the metabolite composition among farms, we did not identify factors associated with the contamination of Aspergillus metabolites.
Other potentially harmful contaminants occurring in dairy cows' diets were compounds derived from the genus Alternaria, some of which are considered emerging mycotoxins, such as AOH, AME, and TeA. Our study indicates that they are commonly presented in the diets of Austrian dairy cows. Alternaria spp. can grow and produce toxins in various crops in the field and post-harvest stage, causing considerable losses due to decomposition [87,88]. Our analysis further indicates that straw contributes to contamination from Alternaria. Data and information regarding occurrence in the feeds and toxicological implications of Alternaria toxins for livestock systems are still missing [88][89][90]. Our survey suggests that the occurrence of metabolites of Alternaria should not be ignored. For instance, TeA was the most frequently detected Alternaria metabolite in the diets of Austrian dairy cows. This mycoestrogen targets protein synthesis inhibition at the ribosomal level and is considered, concerning toxicity, the most important metabolite produced by Alternaria spp. [91]. The benzopyrene derivatives AOH and AME are not related to acute toxicity but are known for their genotoxic effects [92][93][94]. Moreover, AME, AOH, and TeA are also classified as mycoestrogens, showing strong synergistic estrogenic effects in combination with mycoestrogen ZEN even at very low concentrations [32,33,95]. Our co-occurrence analysis showed that 30% to 60% of the samples displayed co-contamination between ZEN and Alternaria-derived AOH, AME, and TeA.
The analysed diets presented several Penicillium-derived toxins, which are considered the most relevant post-harvest mycotoxins contained in silages [6,[96][97][98][99][100]. However, the production of such toxins is also possible in the field [72,101]. MPA and ROQs are considered the most investigated Penicillium metabolites occurring in silage [6]. A common feature of many Penicillium-derived exometabolites such as MPA, ROQs, CIT, and OTA is their immunotoxic properties [102,103], which could interfere with the activity of innate and adaptative immune responses, predisposing the animals to secondary infectious diseases [104]. Penicillium toxins have been linked with appetite reduction, affecting nutrient efficiency, and increasing the incidence of abomasal ulcers, laminitis, gastroenteritis, abortion, and paralysis [105]. Additionally, toxins produced by Penicillium spp. Such as ROQ C have neurotoxic activity [106]. Despite their abundance in feeds and their potential harmful properties, the economic relevance of Penicillium mycotoxins in livestock farming is considered underestimated, because even though mycotoxins are believed to be rapidly metabolized by gut microbiota and hepatic enzymes [104,[107][108][109], the detoxification process of mycotoxins can still be disrupted by their antimicrobial and hepatotoxic properties [104,107,[110][111][112][113][114]. Penicillium-derived mycotoxins are mostly associated with storage, being detected frequently in mouldy spots of silages [100,115,116]. Although the temperature of the samplings' month presented a negligible correlation (ρ = 0.20, p = 0.004) in our study, several studies performed under controlled conditions have proven that Penicillium growth and toxin production were strongly increased by higher temperatures [117][118][119][120]. Penicillium roqueforti has been described as the most predominant fungi in mouldy sections of silages in Austrian dairy farms [100]. Contamination with storage mycotoxins (mainly associated with Penicillium) can occur even in good-quality silages, since aerobic spoilage is practically unavoidable during feed-out [121]. Our findings did not reveal relationships between the hygienic status of the main feedstuffs (GS, MS, straw, hay, BSG, and concentrate) and the contamination levels, which has been reported previously in forages [122]. This can be explained by the fact that toxin production by a fungus does not correlate directly with its growth [123]. Over 30% of the evaluated diets contained EAs, toxic compounds associated with diverse endocrine, vascular, and neurological effects [124]. These can be commonly detected in cereal grains as well as in pastures [42,125,126]. Dietary exposure to EAs in dairy cattle can produce unspecific effects such as reduced productive and reproductive performance and acute clinical signs of ergotism including hyperthermia, convulsions, gangrene in distal portions of the body, and fatalities [127][128][129]. It was stated that feeds exceeding 250 µg/kg of EAs should not be fed to pregnant or lactating animals, because it could increase the risk of abortion and agalactia syndrome [126]. Additionally, further less-known metabolites are produced by other fungi detected in the diets of dairy cows. Some of them have antibacterial activity, for example, the anthraquinone rubellin D [130,131], illicicolins [132], monocerin [133,134], and cytochalasins [135,136].
Interestingly, the recent analysis indicates that as compared with contamination from other fungal groups, contamination of Fusarium metabolites can be explained to a greater extent by dietary factors that are mainly related to forage components. We demonstrated a complex relationship between MS, straw, and proportions of NFC and large particles (>19 mm) that drives the contamination of Fusarium metabolites in dairy cow diets. With our multiple regression approach, the independent factors can explain 50% of the variance, substantially higher than a previous study that used a simple correlation analysis [137]. As explained before, other studies have underlined MS as a potential feed source of Fusarium mycotoxins. This could be explained by the fact that starch induces mycotoxin production (e.g., trichothecenes) in F. graminearum [137]. Thus, the superior content of non-fibre carbohydrates (such as starch) in maize and cereal plants compared to other forages such as GS and hay could explain the elevated levels of mycotoxins and other secondary metabolites. Furthermore, we found that, in addition to MS, straw was likewise an influential forage component. Straw is often added to dairy cow diets containing high grains and high MS to compensate for physical characteristics (long fibre) of the diet. As minor dietary components, the hygienic as well as chemical characteristics of straw likely receive less attention as compared to main forage sources such as MS, GS, and hay. Mould infection could be present in straw but might not be screened out before feeding. We found that, in addition to Fusarium metabolites, straw was also a determinant for contamination with Alternaria metabolites. The black mould genus Alternaria includes various saprophytic, endophytic, and pathogenic species, which occur worldwide in different habitats such as soil, as well as on dead or dying plant tissues such as straw [138]. A recent Swiss survey targeting a broad spectrum of mycotoxins in barley products found higher concentrations of total fungal metabolites in straw than in grains [139]. Interaction of dietary large particle size with MS and with straw partly represented shifts in the physical characteristics of the diet based on the combination of forage choices. Dietary ash content did not influence concentrations of metabolites from Fusarium, Alternaria, or Penicillium, but it did influence total fungal metabolites. Its positive quadratic effect indicates that high fungal metabolite loads are associated with high dietary ash content. High dietary ash contents are an indicator of contamination with soil, which affects the hygienic quality of the feedstuffs. All in all, although the current data could prove partial roles of the main dietary factors, the outcome underlines that there is no single factor that dominantly influences the dietary contamination. Rather, the dominant influence comes through the combination of forage choice, management (particle length), and the hygienic status of feed sources.
Another novel outcome of the present study was related to PEs, which constitute the extensively recurrent class of metabolites contained in dairy rations. PEs are of concern in veterinary medicine and public health due to their endocrine-disrupting activity. These substances especially affect the reproductive organs and process, inducing infertility in livestock [140,141]. These metabolites are found primarily in Leguminosae plants, such as soy, but also in clovers (Trifolium spp.) and alfalfa/lucerne (Medicago sativa) [9,28,142]. Coumetrans such as coumestrol seem to be more potent in estrogenic activity [9,31]. The levels of coumestrol detected in diets of Austrian dairy herds in the present study were below the reported critical range (18-180 mg/kg) [141]. Their interaction with other estrogenic substances (such as mycoestrogens) is currently the focus of interest [38]. Other plantderived compounds such as the cyanogenic glucosides linamarin and lotaustralin observed in the present study did not exceed the maximum limit (50 mg/kg) of total cyanogenic compounds established by the European Union [143]. Both compounds (linamarin and lotaustralin) have a relatively broad distribution in the plant kingdom, being found in high concentrations in cassava, soy, cereal, clovers, and other plant species [53,144]. In general, levels of these compounds in clover are not high enough to cause acute toxicity. Some clinical manifestations include dyspnoea, muscular contractions, and oedemas in mucous membranes [145]. Nevertheless, reports of cyanide poisoning of livestock are rare, suggesting that levels of cyanide-or HCN-producing compounds in the feed are generally low [143], as is also the case for the present study. The inclusion of hay showed a major correlation with both linamarin and lotastratin in this study. Among the unspecific metabolites detected were molecules of some biologically active toxins, which increase the toxicological complexity of the cocktails of secondary metabolites evidenced. These include, for instance, emodin (antibacterial and immunosuppressive) [146,147], 3-nitropropionic acid (neurotoxic) [148,149], skyrin [147], brevianamide F (cyclo-L-Trp-L-Pro) (antifungal and antibacterial) [150], cyclo (L-Pro-L-Tyr), and cyclo (L-Pro-L-Val) (antibacterial) [151,152]. The complex profiles of co-contamination with different mycotoxins, PEs, CGs, and other metabolites occurring in the diets of high-yielding dairy cows suggest unexplored and unpredictable synergistic as well as antagonistic toxic effects. Most of the detected metabolites represent unregulated compounds with a high diversity of biological and toxic activity, indicating that the characterization of the regulated contaminant in dairy feeds is only the tip of the iceberg of fungal and other environmental toxins.

Conclusions
This study underlined the omnipresence of a broad number of mycotoxins (most of them unregulated), Pes, and other metabolites occurring in diets of dairy cows in Austria. Overall, the Austrian dairy rations are safe when considering that the detected contamination levels were below the guidance values of the EU commission. Nevertheless, a vast majority of mycotoxins and metabolites are emerging ones, as well as less-known and lessstudied fungal metabolites. Overall, Fusarium-produced metabolites and mycotoxins were the dominant fungal contaminants. Additionally, we found that dietary factors related to the use of forages, rather than concentrating sources, contribute to increased contamination of mycotoxins in Austrian dairy rations. Among typical forage sources, the content of MS and straw were the most influential factors linked to the concentration of Fusarium metabolites in the complete rations. The analysis further addressed the influences of characteristics of diets and hygienic substandard of forages. Individually, the detected mycotoxins represented a relatively low or safe level based on EU regulation and literature. However, the co-exposure to mycotoxins and other (fungal and plant) secondary metabolites has unpredictable effects. Our findings make clear that the evaluation of contamination with only regulated mycotoxins offers a limited picture of the possible toxicological risks to animal health, reproduction, and productivity. Therefore, it is crucial to elaborate surveillance and monitoring programs for a broad spectrum of metabolites in the dairy feed chain and to understand their toxicological effects. Furthermore, there is a need to increase awareness of the importance of feed management and nutrition as reduction and prevention measures for mycotoxin contamination in dairy production. Monitoring and further research based on multi-metabolite approaches in the dairy industry in other geographic regions are still necessary.

Sampling and Sample Preparation
Under the agreement of written informed consent with the farmers, 100 dairy farms located in Lower Austria (n = 33), Upper Austria (n = 51), and Styria (n = 16), representing the 3 provinces leading the country's dairy production, were involved in the survey, lasting from 2019-2020 ( Figure 8A). The herd sizes (number of lactating cows) during both visits were on average 59 ± 15 SD lactating cows per farm, varying from 32 to 140 lactating cows per farm. Each representative sample of complete diets (n = 198) consisted of at least 30 incremental samples of mixed rations from the feeding table (feed bunk), and at least 30 subsamples of concentrate feed on the automatic feeders were collected. The final sample amount was 1-1.5 kg of each kind of sample (basal feed ration and additional concentrate) ( Figure 8B). An additional sample of basal ration (approx. 1 kg) was collected for particle size determination. The samples were immediately vacuum-packed (−0.7 psi) and stored in the dark at −20 • C to avoid subsequent microbial spoilation until sampling preparation ( Figure 8C). Sampling was performed during the period April 2019 to September 2020, at two time points with a divergence of at least six months between the first (n = 100) and the second sampling (n = 98; two farms did not continue in the study). Since the formulations, feed components, and batches of the different feedstuffs varied between the two visits, both visits within each farm were treated independently (n = 198). The frozen basal feed samples were thawed at room temperature for 24 h and air-dried at 65 • C for 48 h. The average dry-matter content of basal feed samples was 37.06% ± 4.72% (mean ± SD, range: 25.73-54.72%). The dried samples were sequentially milled to a final particle size of ≤0.5 mm. Firstly, they were milled in the cutting mill (SM 300, Retsch GmbH, Haan, Germany) at 1500 rpm for approximately 1 min. The non-milled residues (mostly hard fragments of seeds) were subsequently milled using an ultra-centrifugal mill (ZM 200, Retsch GmbH, Haan, Germany) at 10,000 rpm for approximately 30 s. All milled fractions of each kind of sample were combined, homogeneously mixed, and packed in plastic bags ( Figure 8D). Twenty grams (±0.01 g) of the whole diet representative samples was obtained by mixing proportionally milled basal and the additional concentrated feeds (supplemented based on the daily milk production) according to the average intake of each farm provided by the farmers (see Section 5.2). Then, five grams (±0.01 g) of each homogenized representative sample of the diets intended for multi-analysis was weighed in 50-mL polypropylene conical tubes (Sarstedt, Nümbrecht, Germany), and 100 g of basal feed was utilized for the chemical (proximate) analysis and stored at −20 • C until analysis.  Figure 8D). Twenty grams (±0.01 g) of the whole diet representative samples was obtained by mixing proportionally milled basal and the additional concentrated feeds (supplemented based on the daily milk production) according to the average intake of each farm provided by the farmers (see Section 5.2). Then, five grams (±0.01 g) of each homogenized representative sample of the diets intended for multi-analysis was weighed in 50-mL polypropylene conical tubes (Sarstedt, Nümbrecht, Germany), and 100 g of basal feed was utilized for the chemical (proximate) analysis and stored at −20 °C until analysis. The representative sampling consisted of basal feed (total, partial, or forage mixed ration) collected from the feeding table (*) as well as samples of concentrated feeds (°). (C) Vacuum packing and preservation at −20 °C until sample preparation and subsequent analysis. Sampling preparation consisted of drying, (D) milling (to a particle size of ≤ 0.5 mm), and subsequent (E) pooling and homogenization according to the reported average intakes of basal feed and additional concentrate.

Data Collection
Information regarding the kind of farming system (organic or conventional), the composition of the basal feed (major ingredients and their proportions), and total intakes of basal feeds (forage, partial, or total mixed rations), as well as the amount of additional concentrate and feed supplemented (based on the daily milk production) were obtained from those responsible for feeding management via personal interview guided by questionnaire. Per farm, the hygienic status of conserved forages and concentrates included in the rations of the lactating cows were evaluated. For the hygienic status assessment, representative samples (of at least 10 subsamples) were composited and immediately assessed. The sensory evaluation was performed considering characteristics of the appearance (colour was considered along with the presence of impurities), odour, and texture

Data Collection
Information regarding the kind of farming system (organic or conventional), the composition of the basal feed (major ingredients and their proportions), and total intakes of basal feeds (forage, partial, or total mixed rations), as well as the amount of additional concentrate and feed supplemented (based on the daily milk production) were obtained from those responsible for feeding management via personal interview guided by questionnaire. Per farm, the hygienic status of conserved forages and concentrates included in the rations of the lactating cows were evaluated. For the hygienic status assessment, representative samples (of at least 10 subsamples) were composited and immediately assessed. The sensory evaluation was performed considering characteristics of the appearance (colour was considered along with the presence of impurities), odour, and texture based on the methodological approaches described by Kamphues et al., 2014 [57]. The geo-climatic data, including altitude, average air temperature of the month of sampling, the average air temperature, relative humidity, and the accumulated rainfall during the growing season of maize (June-September of the previous year); average relative humidity of summer (June-September, maize's growing season); rainfall during the summer (June-September, maize's growing season); and averages of the air temperature of the municipalities/districts of the farms, were retrieved from the website of the Central Institution for Meteorology and Geodynamics (in German: Zentralanstalt für Meteorologie und Geodynamik-ZAMG) (available at https://www.zamg.ac.at/cms/de/klima/klimauebersichten/jahrbuch) (accessed on 1 June 2021). Summarized data are illustrated in Table 1.

Chemical Proximate Analysis and Particle Size Distribution of the Rations
The chemical proximate (nutrient) analysis of the samples of basal feed rations was conducted according to the protocols of the Association of German Agricultural Analytic and Research Institutes (VDLUFA, Darmstadt, Germany, 2012) [153]. The dry-matter content was determined by oven-drying the samples at 103 • C for at least 4 h (method 3.1). Ash was analysed by combustion in a muffle furnace at 550 • C overnight (method 8.1). Crude protein was determined using the Kjeldahl method (method 4.1.1) and ether extract using the Soxhlet extraction system (method 5.1.2). Analyses of NDF and the estimation of NFC were performed following the methods described by Van Soest et al. (1991) [154]. Particle size distribution of the basal rations was determined using a manually operated Penn State Particle Separator (PSPS) (model C24682N, Nasco, Fort Atkinson, WI, USA) with three sieves with aperture diameters of 19 mm, 8 mm, and 1.18 mm in diameter, according to Lammers et al. (1996) [155] and Kononoff et al. (2003) [156]. For each visit, the test was performed in duplicate, and the sieve fraction values (%) were averaged.

Sample Extraction and Multi-Metabolite Analysis (LC-ESI-MS/MS)
For simultaneous multi-metabolite quantification, five grams (±0.01 g) of each homogenized sample was extracted in 20 mL of the extraction solvent (acetonitrile/water/acetic acid 79:20:1, v/v/v) and following the procedures reported by Sulyok et al. [157]. Glacial acetic acid (p.a.) and methanol (LC gradient grade) were acquired from Merck, Darmstadt, Germany), and the water was reverse-osmosis-purified using an Elga Purelab ultra-analytic system (Veolia Water, High Wycombe, UK). Then, for sedimentation, the samples were put in a vertical position for 10-15 min. A supernatant of 500 µL of the raw extract was diluted 1:1 with a dilution solvent (acetonitrile/water/acetic acid 20:79:1, v/v/v) in vials. The injection volume of both raw extracts of the samples and standard solutions of the analytes was 5 µL. These volumes were put into the QTrap 5500 LC-MS/MS system (Applied Biosystems, Foster City, CA, USA) equipped with a TurboV electrospray ionization (ESI) source, which was coupled to a 1290 series UHPLC system (Agilent Technologies, Waldbronn, Germany) as described by Sulyok et al., 2020 [157]. A subsequent quantification from external calibration by serial dilutions of a stock solution of analysed compounds was completed. Finally, the results were adjusted for apparent recoveries defined through spiking experiments according to Steiner [158]. Standards of fungal, plant, and unspecific secondary metabolites were purchased from several commercial suppliers or obtained via a donation from different research institutions [157,158]. This analytical methodology has been validated [157,158] and has been employed to study multi-mycotoxin occurrence in complex feedstuff matrices such as silage, pastures, concentrate feed, and total mix rations [53,61,159,160]. The accuracy of the method is verified on a routine basis by participation in proficiency testing organized by BIPEA (Genneviliers, France). Satisfactory z-scores between −2 and 2 have been obtained for >95% of the >1700 results submitted so far. In particular, 17 out of 18 results submitted for a sample of MS were in this range, the exception being zearalenone exhibiting z = −2.05.

Statistical Analysis
Frequencies of contamination (occurrences) and the descriptive statistics of the concentrations of metabolites (average, SD, median, and range values) were calculated considering values over the limit of detection (LOD). Values lower than the limit of quantification (LOQ) were processed as LOQ/2. Concentrations of metabolites are expressed in µg/kg parts per billion (ppb) on a dry-matter basis and plotted on a logarithmic scale (Log 10 ) where applicable. The co-occurrence analyses of mycotoxins and plant metabolites were performed separately using Microsoft Excel, constructing matrices that included metabolites with detection frequencies over 20%. Spearman's correlation coefficients were computed, and heatmaps were plotted using GraphPad Prism (Prism version 9.1, GraphPad Software, San Diego, CA, USA). The correlation analysis was interpreted considering only significative correlations with ρ ≥ 0.3, based on Hinkle et al. (2003) [161]. Multiple regression analysis was performed using SAS (version 9.4; SAS Institute Inc., Cary, NC, USA) to investigate the influences of dietary and geoclimatic factors on dependent variables of interest: concentrations of total metabolites produced by fungi, plants, Alternaria, Aspergillus, Fusarium, Penicillium, EAs, DON, ZEN, FUM (the sum of FB1 and FB2), BEA, ENNs, CUL, siccanol, and phytoestrogens. Data were log-transformed to normalize the data. For some variables (DON, FUM, siccanol, total Alternaria metabolites, total Penicillium metabolites, and total fungal metabolites), extreme data that still led to screwed data were manually excluded. The normality of data based on the Shapiro-Wilk test (p > 0.05) and Q-Q-plot were ensured before further data analysis. For each dependent variable, a set of independent variables including dietary proportions of MS, straw, hay, BSG, feed particle size > 19 mm and between 8-1.18 mm, the content of crude protein, ash, ether extract, ash, and non-fibre carbohydrate, the hygienic status of MS and GS, altitude, temperature, relative humidity, and rainfall were tested, and the candidate independent variables were selected based on a step-wise selection using the procedure SELECT of SAS. All candidate independent variables passed the collinearity test, having a variance inflation factor less than 10. Next, the effects of candidate variables, including their squared terms and interactions, were investigated using the mixed procedure of SAS. The model also included the random effect of two rounds of visits. Backward elimination was performed to obtain the final model using the protocol described previously [162]. Additionally, R 2 and RMSE of the final model were calculated. In addition, the odds ratio and predicted probabilities for high contamination of Fusarium metabolites due to the inclusion levels of forage sources were determined using PROC LOGISTIC (SAS version 9.4; SAS Institute Inc., Cary, NC, USA). For this analysis, data classified as low (25 percentile, n = 49) and high Fusarium metabolite concentrations (75 percentile, n = 60) were used. The model included dietary levels of MS, GS, straw, hay, BGS, other silages, and quadratic terms of MS, because they were found to show significance in multiple regression analysis.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/toxins14070493/s1. Table S1. List of 863 targeted metabolites via a validated multi-metabolite liquid chromatography/electrospray ionization-tandem mass spectrometric (LC/ESI-MS/MS) method. Table S2. Odds ratio estimates and profile-likelihood confidence intervals of forage inclusion levels as dietary risk factors for high Fusarium mycotoxin loads (above 75th percentile concentrations). Figure S1. Predicted probabilities for Fusarium mycotoxin loads (above 75th percentile concentrations) related to the proportion of (a) maize silage and (b) straw in the dietary rations of Austrian dairy cows.