Occurrence of Mycotoxins in Grass and Whole-Crop Cereal Silages—A Farm Survey

Abstract

Mycotoxin incidence in forage may heavily affect the amount of toxins consumed by cattle. However, many studies have focused on mycotoxin occurrence in cereals and there are less studies of forages, particularly of grass silages. For determining the occurrence of mycotoxins in farm silages under Northern European conditions in Finland, 37 grass silage and 6 whole-crop cereal silage batches were analysed separately for surface, core and, if present, visibly mouldy spots. Mycotoxins were found in 92% of the samples. All mouldy samples contained mycotoxins. Beauvericin was the most common mycotoxin in grass silages and roquefortine C in whole-crop cereal silages. In mouldy samples, beauvericin, mycophenolic acid and roquefortine C were the most common mycotoxins in the grass silage and mycophenolic acid in the whole-crop cereal silage. Aflatoxins were not found in any of the samples. On average, all samples contained more than one type of mycotoxin. Concentrations of mycotoxins varied considerably from very low to very high values. The results of this survey indicate that silage-fed ruminants can be exposed to a broad range of mycotoxins. The absence of visible moulds does not always indicate mycotoxin-free feed. All moulded samples contained mycotoxins and some at very high concentrations, and they contained more different types of mycotoxins than samples without visible mould. Thus, feeding mouldy feeds to animals should be avoided.

1. Introduction

Mycotoxins are a group of secondary metabolites naturally produced by fungal organisms [1]. More than 300 mycotoxins are known, of which around 30 have been considered to be a risk for human or animal health [2]. However, only five mycotoxins, aflatoxin B1, deoxynivalenol, zearalenone, ochratoxin A and fumonisins, are regulated by the EU legislation for animal feed [3,4]. The obligatory regulation applies for aflatoxin B1 [3], and recommended guidance values are available for the other above-mentioned mycotoxins [4].

Various factors are known to influence the incidence of mycotoxins and they may occur in a wide range of animal feeds, including concentrates and forages. Mycotoxins can be produced in the field during plant growth, but also during harvesting and when stored either dried or ensiled [5,6,7,8]. The verification of mycotoxins in feeds is difficult. Not all moulds produce mycotoxins and visible mould in silage does not indicate the presence of mycotoxins but, on the other hand, mycotoxins may occur without visible mould [8,9]. In addition, mycotoxins are invisible, odourless and they cannot be detected by smell or taste [10]. Therefore, assessing the occurrence of mycotoxins subjectively is not possible. Moulds and mycotoxins are not homogenously distributed in silages, which makes representative sampling for analyses difficult [7,11].

Mycotoxins have many adverse effects. They can, e.g., decrease crop yields and quality and impair the productivity, health and welfare of the animals and, in consequence, cause economic losses [1,12,13,14]. Exposure to mycotoxins may cause various symptoms and health problems such as a reduced feed intake, digestive disorders, reproductive problems, immune suppression and carcinogenic effects [10]. The severity of the symptoms caused by mycotoxins depends on many factors, including the type(s) of mycotoxin(s), ingested dose, duration of exposure and physiological status and age of the animal [1,15,16,17].

Ruminants are less sensitive to the harmful effects of mycotoxins than monogastric animals because rumen microbiota can degrade or inactivate many mycotoxins, but not all [15]. However, the rumen capacity to detoxify mycotoxins may be exceeded or it can vary depending on, e.g., the type and concentration of mycotoxins, diet, metabolic diseases and ruminal conditions [15,18,19]. Additionally, although rumen microbes may degrade toxins, in some cases, their metabolites may be more toxic than the parent toxin, which is the case when zearalenone is converted into α-zearalenol [16,18]. Furthermore, aflatoxin B1 can be a risk for the safety of milk and milk products because of the carry-over of aflatoxin B1 from feed to aflatoxin M1 in milk [20]. However, feed aflatoxin contaminations are usually associated with regions of high temperature and high humidity [11] and, thus, the risk of aflatoxin occurrence in feeds is not high under the climate conditions of Northern Europe.

Ensiled forages are globally an important component in cattle diets [21]. Therefore, mycotoxin incidence in silage may heavily affect the amount of toxins consumed by cattle. However, most studies have focused on mycotoxin occurrence in cereals and there are less studies of forages [17]. If the occurrence of mycotoxins in forages has been studied, the focus has mostly been on maize silage so that there is a lack of information of mycotoxin prevalence in grass silage. Although the use of maize silage has increased and it is commonly used worldwide, in some countries, such as in Ireland and Nordic countries, grass silage is a more common forage than maize silage. Additionally, small-grain cereals and grain legumes harvested as a whole crop silage can be used as a forage.

In practice, silages are not analysed routinely for mycotoxins due to a lack of rapid low-cost methods available for farmers. Additionally, the results of mycotoxin analyses are still not in a consensus across laboratories [22]. In consequence, there is lack of wide-ranging information about the prevalence of mycotoxins in silages at the farm level. Additionally, increased humidity towards the end of the growing season can increase the risk of mycotoxin occurrence in the regrowth harvest compared to primary growth harvest, which requires more consideration. The objective of this survey is to identify and quantify the occurrence of mycotoxins in farm silages under Northern European environmental conditions. This kind of survey has not been previously conducted in Finland. The main target is to produce information of the occurrence of mycotoxins in grass silage, which is limited in the scientific literature.

It is hypothesized that several mycotoxins are present in silages produced under farm conditions. Additionally, it is hypothesized that the occurrence of mycotoxins is higher in the second and third cuts of grass silage than in the first cut and that none of the silage samples contain aflatoxin B1.

2. Materials and Methods

2.1. Sample Collection for Mycotoxin Analysis

Silage samples were collected from 20 Finnish beef cattle farms in the regions of North Savo, North Karelia, South Ostrobothnia, Central Ostrobothnia and North Ostrobothnia, which represented Northern European climatic conditions (Figure 1). These are the main areas for beef and milk production in Finland and cattle feeding is based on grass silage. Thus, these are significant grass production areas. Four beef-producing farms from each region were selected arbitrarily related to anticipated mycotoxin occurrence in silages. The emphasis was to collect samples from the second and third cuts of grass silage due to the higher risk of mycotoxin contamination; however, samples from the first cut were also included. Further, if whole-crop cereal silage was present at the farm in addition to grass silage, it was also sampled.

Samples were collected between February and March 2020 from 43 different silages. Silages were stored in 13 clamp silos, 5 stacks and 24 round bales, while information of storage method from one silage was missing. In the current dataset, 13 silages were preserved with formic acid-based additives, 8 with lactic acid bacteria inoculants, 4 without any additives and for 18 the information was not available. Most of the grass silages were mixtures of different grass and legume species, timothy (Phleum pratense) and meadow fescue (Festuca pratensis) being the most common species included. Whole-crop cereal silages were different mixtures of wheat (Triticum aestivum), barley (Hordeum vulgare), pea (Pisum sativum) and previously mentioned grass species in variable proportions.

All silages were sampled from the surface and core and in case of visible mould present, a separate sample of the moulded feed was taken for comparing mycotoxin occurrence in visually non-mouldy and mouldy feeds. Sampling surface and core separately was based on the tendency of the surface to be more prone to spoilage due to greater aerobic exposure. The sampling was conducted according to the procedure used by Driehuis et al. [23]. The surface layer was defined as a maximum 20 cm depth from the surface. The core was defined to be at least 30 cm deep from the surface. If mould existed, at least 15 cm distance was kept from moulded spots when core and surface samples were taken. Samples from both the surface and core comprised 10 subsamples, that were manually taken from different positions, and from both approximately 500 g sample was taken for the mycotoxin analysis. When visibly mouldy feed was sampled, a maximum of 10 moulded spots was collected, mixed and then a sample of approximately 500 g was taken for analysis. All samples were stored frozen at −20 °C prior to analysis.

A total of 102 samples was collected, which contained surface, core and mouldy samples. Both surface and core samples were collected from 37 grass silages and from 6 whole-crop cereal silages. In addition, 13 mouldy samples were collected from grass silages and 3 from whole-crop cereal silages. Most of the grass silage surface and core samples were from the second cut (56.8%), then from the third cut (29.7%) and least from the first cut (13.5%). Distribution of mouldy samples over cuts was 7.7%, 53.8% and 38.5% in first, second and third cuts, respectively.

2.2. Dry Matter and Mycotoxin Analyses

All collected 102 samples were thawed and analysed for dry matter (DM) and mycotoxins at the laboratory of the Natural Resources Institute Finland (Luke) in Jokioinen, Finland. The DM concentration was determined by drying the samples in a forced-air oven at 105 °C overnight (+2 h at 50 °C at first) until no more weight loss occurred. For mycotoxin extraction, silage samples were dried in a forced-air oven at 60 °C for 18 h. After drying, the samples were ground using a sample mill (Sakomylly KT-120, Koneteollisuus Oy, Helsinki, Finland) using a 1 mm sieve. Secondary DM content of the same samples that were analysed for mycotoxins (originally dried at 60 °C) was determined by drying the air-dried samples at 105 °C for 16 h to be able to present the results on DM basis (g/kg DM).

In total, 32 different mycotoxins were analysed (Table S1). All mycotoxins were extracted from 1 g of samples according to Rämö et al. [24] with one modification; BondElut 6982–6650 (Agilent Technologies, Santa Clara, CA, USA) was used instead of AHO-9041 (Phenomenex, Torrance, CA, USA) as QuEChERS extraction salt. Mycotoxins were quantified both from raw extract and from cleaned-up and concentrated extract of each sample according to Rämö et al. [24] by filtering the sample extracts through 0.2 µm GHP Acrodisc 13 Teflon filter (Pall Corporation, Ann Arbor, MI, USA) in UHPLC vials.

Both solvent and matrix-matched calibration samples were prepared as real grass silage samples for Waters Acquity UPLC system equipped with Xevo TQ MS triple–quadrupole mass spectrometer (Milford, MA, USA), which was used for identification and quantification of mycotoxins. A mycotoxin-free grass silage sample was used for matrix-matched calibration. Both calibration curves contained seven different concentration levels, including level zero without mycotoxins. Caffeine was used as an internal standard [24], and nandrolone was used to follow how well extraction of a single sample was succeeded. Mycotoxins were separated in gradient run on Acquity BEH C18 1.7 µm -UPLC reversed phase columns; VanQuard pre-column (2.1 × 5 mm) was connected to an analytical column (2.1 × 100 mm) (Waters, Milford, MA, USA) according to Huuskonen et al. [25].

Duplicate analyses were conducted for all mouldy and whole-crop cereal samples, and later additional duplicate analyses were conducted when mycotoxins were detected in a sample. The relative difference between duplicate silage samples was used for the calculation of repeatability of the method. If the mycotoxin concentrations in the samples exceeded the highest calibration level, they were diluted with the same sample extract which was used for the matrix-matched calibration (1:9 v:v), which enabled quantitation of 10 times higher concentrations. The lowest calibration concentration of a mycotoxin was used as the limit of quantification (LOQ) if (a) both multiple reaction monitoring values (MRM) of identification and qualification existed with a correct ratio and (b) the signal to noise ratio was satisfactory. However, matrix interferences were detected with lower calibration levels, which caused higher LOQ for several mycotoxins. For the quantitative range of calibration curve (µg/kg), the highest quantitative concentration (µg/kg), LOQ and repeatability (%) for 10 detected mycotoxins are given in

Table 1. Quantitative range of calibration curve (μg/kg), the highest quantitative concentration (μg/kg), limit of quantification (LOQ, μg/kg) and repeatability (%) of detected mycotoxins.

	Quantitative Range	The Highest Quantitative Concentration 1	LOQ	Repeatability 2
Alternariol	50–2500	25,000	50	NC 3
Beauvericin	5–250	2500	5	13.3
Deoxynivalenol	130–6800	68,000	130	NC 3
Enniatin A	10–600	6000	10	9.7
Enniatin A1	10–650	6500	10	8.6
Enniatin B	10–720	7200	10	11.6
Enniatin B1	10–720	7200	10	7.9
Mycophenolic acid	30–2570	25,700	30	2.9
Roquefortine C	10–3200	32,000	10	3.0
Zearalenone	30–2560	25,600	30	8.1

¹ Accessible with 10 times dilution with blank silage extract. ² Calculated with relative difference between duplicate silage samples. ³ NC—not calculated. Too few results for repeatability calculation.

. Molecular weights, MRM values, cone voltages and collision energies of all 32 mycotoxins and LOQ of all analysed 29 mycotoxins are given in Table S1. Moniliformin, citrinin and nivalenol could not be quantified by this method: Moniliformin was eluted at the beginning of UHPLC gradient, its MS response was weak and it was interfered by silage matrix. Citrinin was able to calibrate, but it was not stable in silage extract during the analysis. Nivalenol could not be quantified with UHPLC-MS/MS, because of its poor repeatability and producibility.

The matrix effect of grass silage was controlled with an internal standard matrix-matched calibration. The sample matrix was found to suppress the areas of caffeine and mycotoxins. However, the slopes of both solvent and matrix-matched calibration curves with the internal standard were almost identical. The correlations (R²) were >0.99, which indicated linearity of calibration curves. The multi-point matrix-matched calibration was used for separation and identification of mycotoxins in grass silage samples, because of suppressing matrix effect detected for all mycotoxins and detected interference signals in retention area of several mycotoxins, which also caused higher LOQ.

Two earlier analysed red clover silage samples were used as in-house reference samples in Rämö et al. [24]. Mycophenolic acid and roquefortine C were quantified in the first reference sample with average concentrations 225 µg/kg and 3500 µg/kg, respectively (earlier values 220 µg/kg and 3800 µg/kg). Zearalenone was quantified in the second reference sample with average concentration 1050 µg/kg (earlier value 1800 µg/kg). One whole-crop cereal silage sample was also used as an in-house reference: deoxynivalenol was quantified in concentrated silage extract with average concentration 370 µg/kg (earlier value 470 µg/kg). When deoxynivalenol was detected, the samples were re-extracted with accredited trichothecene method determined as their trimethylsilyl imidazole derivatives by GC-MS (Agilent 6890 N-5973, Agilent Technologies, Santa Clara, CA, USA), and the method was described in more detail by Huuskonen et al. [25]. Deoxynivalenol concentrations in those samples were higher when using the GC-MS method than when using the UHPLC-MS/MS method. Reported concentrations of deoxynivalenol were analysed using the GC-MS-method.

2.3. Statistical Analyses

When average, minimum and maximum mycotoxin concentrations and standard deviations were calculated, half of the LOQ value was used for detected values over zero, but below LOQ to include all samples where mycotoxins were detected in the results. When duplicate analyses were conducted, mean value of the analytical replicates was used. The LOQ values for all determined individual mycotoxins are presented in Table S1. When mycotoxin concentrations were presented, values represented only samples where specific mycotoxin was detected and samples with zero mycotoxin concentration were excluded. However, DM concentrations represent results of all samples. Pearson correlations in grass silage between DM and total mycotoxin concentrations in core and surface samples and mycotoxin concentrations between core and surface samples were computed using SAS CORR procedure (version 9.4, SAS Institute Inc., Cary, NC, USA).

3. Results

3.1. Samples

Variation in the sample DM concentration was high (

Table 2. Number of samples and dry matter concentrations of grass and whole-crop cereal silage samples. Results of different cuts include both surface and core samples, but not mouldy samples.

	n	Dry Matter, g/kg
		Mean	Sd	Min	Max
Grass silage
Core	37	325	125.8	167	720
Surface	37	325	122.7	179	709
Mouldy	13	308	145.9	138	606
First cut	10	407	186.9	204	720
Second cut	42	360	101.3	196	623
Third cut	22	221	31.1	167	286
Whole-crop cereal silage
Core	6	295	104.3	168	458
Surface	6	320	67.0	263	435
Mouldy	3	234	34.4	200	269

). Numerically, the average DM concentration in the core, surface and mouldy samples was slightly higher in grass silages compared to the corresponding samples in whole-crop cereal silages. The average DM concentration of grass silage core and surface samples was equal. In whole-crop cereal silages, the average DM concentration was numerically higher in surface samples than in core samples. In both grass and whole-crop cereal silages, the mouldy samples had a numerically lower average DM concentration than the core and surface samples. They had a 13% and 84% higher DM concentration than those in the second and third cut, respectively.

3.2. Incidence of Mycotoxins

A total of 10 different mycotoxins of 32 analysed mycotoxins was detected from the samples, 9 different mycotoxins in grass silage and 8 different in whole-crop cereal silage. The incidence, mean, minimum and maximum concentrations and standard deviation of each detected mycotoxin in the grass silage and whole-crop cereal silage are presented in

Table 3. Incidence and concentrations of mycotoxins (μg/kg dry matter) in grass silage. Values represent only samples where the specific mycotoxin was detected.

	Core, n = 37						Surface, n = 37						Mouldy, n = 13
	Positive Samples		Concentration				Positive Samples		Concentration				Positive Samples		Concentration
	n	%	Mean	Sd	Min	Max	n	%	Mean	Sd	Min	Max	n	%	Mean	Sd	Min	Max
Alternariol	1	3	736		736	736	1	3	493		493	493	0	0
Beauvericin	26	70	10	9.4	3	32	30	81	14	19.3	3	101	11	85	1025	2669.0	3	8854
Enniatin A	2	5	27	0.4	27	27	2	5	1257	1690.4	61	2452	5	38	48	66.9	6	167
Enniatin A1	2	5	6	0.0	6	6	2	5	780	1072.6	22	1538	5	38	51	56.9	6	138
Enniatin B	14	38	70	64.6	6	215	16	43	45	58.1	6	210	8	62	499	748.4	6	2155
Enniatin B1	15	41	13	13.3	6	55	15	41	31	94.2	3	371	7	54	186	218.4	6	509
Mycophenolic acid	8	22	515	1305.0	17	3737	11	30	940	1588.4	17	4843	11	85	22,291	29,989.7	27	84,830
Roquefortine C	18	49	1220	4962.9	6	21,101	19	51	501	950.4	6	3266	11	85	18,847	22,393.9	6	60,962
Zearalenone	3	8	175	81.9	100	262	2	5	77	84.8	17	137	5	38	7908	10,479.9	106	19,992

and

Table 4. Incidence and concentrations of mycotoxins (μg/kg dry matter) in whole-crop cereal silage. Values represent only samples where the specific mycotoxin was detected.

	Core, n = 6						Surface, n = 6						Mouldy, n = 3
	Positive Samples		Concentration				Positive Samples		Concentration				Positive Samples		Concentration
	n	%	Mean	Sd	Min	Max	n	%	Mean	Sd	Min	Max	n	%	Mean	Sd	Min	Max
Beauvericin	2	33	9	8.4	3	15	3	50	15	16.3	3	34	2	67	1106	1559.6	3	2208
Deoxynivalenol	1	17	126		126	126	1	17	425		425	425	0	0
Enniatin A	0	0					0	0					1	33	97		97	97
Enniatin A1	0	0					0	0					1	33	55		55	55
Enniatin B	2	33	235	214.0	84	387	1	17	299		299	299	0	0
Enniatin B1	2	33	21	7.3	16	26	1	17	16		16	16	1	33	13		13	13
Mycophenolic acid	2	33	522	673.0	46	998	2	33	754	1004.0	44	1464	3	100	2210	3472.6	154	6220
Roquefortine C	3	50	68	62.5	30	140	4	67	876	1689.2	6	3409	2	67	1177	1427.4	168	2186

, respectively.

Mycotoxins were found in 92% of all samples. For the grass silage, 94%, and for the whole-crop cereal silage, 80% of the samples contained mycotoxins. All mouldy samples contained mycotoxins. Mycotoxins were found in 91% of non-mouldy samples. Alternariol and zearalenone were detected only in the grass silage and deoxynivalenol only in the whole-crop cereal silage, while other detected mycotoxins were found both in grass and whole-crop cereal silages. In the grass silage, alternariol was found only in non-mouldy samples, but other detected mycotoxins were found both in non-mouldy and mouldy samples. In the whole-crop cereal silage, deoxynivalenol and enniatin B were found only in non-mouldy samples and enniatin A and enniatin A1 were found only in mouldy samples, while other detected mycotoxins were found in both non-mouldy and mouldy samples.

Beauvericin was the most common mycotoxin in the core and surface samples of the grass silage. The incidence of enniatin B, enniatin B1, mycophenolic acid and roquefortine C was also considerable. In addition, alternariol, enniatin A, enniatin A1 and zearalenone were detected, but their proportions of the detected mycotoxins were rather small. In mouldy samples of grass silage beauvericin, mycophenolic acid and roquefortine C were the most common mycotoxins, but also enniatin A, enniatin A1, enniatin B, enniatin B1 and zearalenone were detected. In core and surface samples of the whole-crop cereal silage, roquefortine C was the most common, but also beauvericin, deoxynivalenol, enniatin B, enniatin B1 and mycophenolic acid were detected. In the mouldy samples of the whole-crop cereal silage, the most common mycotoxin was mycophenolic acid, and it was detected in all samples. Additionally, the incidence of beauvericin, enniatin A, enniatin A1, enniatin B1 and roquefortine C was considerable. Aflatoxins were not found in any of the samples.

Beauvericin was the most common mycotoxin in all cuts of grass silage (

Table 5. Incidence and concentrations (μg/kg dry matter) of mycotoxins in different cuts of grass silage. Mouldy samples were not included in the results. Values represent only samples where the specific mycotoxin was detected.

	First Cut, Core and Surface, n = 10						Second Cut, Core and Surface, n = 42						Third Cut, Core and Surface, n = 22
	Positive Samples		Concentration				Positive Samples		Concentration				Positive Samples		Concentration
	n	%	Mean	Sd	Min	Max	n	%	Mean	Sd	Min	Max	n	%	Mean	Sd	Min	Max
Alternariol	2	20	615	171.9	493	736	0	0					0	0
Beauvericin	8	80	22	33.2	3	101	28	67	10	9.8	3	38	20	91	11	10.0	3	42
Enniatin A	2	20	44	23.9	27	61	2	5	1239	1714.7	27	2452	0	0
Enniatin A1	2	20	14	11.2	6	22	1	2	1538		1538	1538	1	5	6		6	6
Enniatin B	2	20	84	35.1	59	109	13	31	64	76.8	6	215	15	68	46	49.3	6	187
Enniatin B1	2	20	6	0.03	6	6	12	29	41	105.1	3	371	16	73	9	6.09	6	22
Mycophenolic acid	1	10	17		17	17	11	26	244	520.9	17	1787	7	32	1681	2076.0	17	4842
Roquefortine C	4	40	113	156.6	6	344	20	48	297	668.1	6	2583	13	59	1929	5829.2	6	21,101
Zearalenone	0	0					0	0					5	23	136	89.6	17	262

). Additionally, the incidence of enniatin B, enniatin B1 and roquefortine C was considerable in all cuts. The incidence of mycophenolic acid and roquefortine C was more common in the second and third cuts than in the first cut. Enniatin B and enniatin B1 were clearly more common in the third cut than in the first and second cuts. Zearalenone was detected only in the samples of the third cut and alternariol only in the samples of the first cut. Enniatin A and enniatin A1 were more common in the first cut than in the second and third cuts.

The average number of different mycotoxins in the core, surface and mouldy samples of the grass silage was 2.4, 2.7 and 4.8, respectively. The maximum numbers were seven, six and eight in the core, surface and mouldy samples, respectively. In the whole-crop cereal silage, the corresponding numbers were 2.0, 2.0 and 3.3, and the maximum numbers were five, four and six, respectively. In the different cuts of grass silage, the average number of mycotoxins in the core, surface and mouldy samples were 2.2, 2.4 and 5.0 in the first cut, 1.8, 2.4 and 4.9 in the second cut and 3.7, 3.3 and 4.8 in the third cut, respectively.

3.3. Concentrations of Mycotoxins

Concentrations of mycotoxins varied widely from very low to very high (Table 3 and Table 4). In the grass silage, the highest average mycotoxin concentrations were found for roquefortine C in the core and mycophenolic acid in the surface and mouldy samples. Particularly high average concentrations of mycophenolic acid, roquefortine C and zearalenone were detected in mouldy samples of the grass silage. In the whole-crop cereal silage, mycophenolic acid in the core, roquefortine C in the surface and mycophenolic acid in the mouldy samples showed the highest concentrations of the detected mycotoxins. When comparing different cuts of grass silage, the highest detected mycotoxin concentration was found for alternariol in the first cut and roquefortine C in the second and third cuts (Table 5).

Correlations between DM and total mycotoxin concentration in the core and surface samples and of total mycotoxin concentrations between the core and surface samples were not found.

4. Discussion

This survey showed that the incidence of several mycotoxins was common in farm silages produced under Norther European conditions as hypothesized. However, concentrations of mycotoxins varied considerably, which was demonstrated by the high standard deviations and the large differences between the minimum and maximum concentrations also in line with earlier reports [8,23,26]. When assessing the results, it should be noted that each sample represented only those 10 spots of silage mass where it was taken. The occurrence and distribution of mycotoxins in silos may vary depending on the physical and environmental conditions within a silo [11]. In consequence, it is difficult to obtain a representative sample. This causes it to be difficult to assess the real amounts of mycotoxins in feeds and how much animals ingest them when fed with silage.

Mycotoxin contamination may occur both pre- and postharvesting [5,6,8,27]. Thus, the prevention of fungal growth and mycotoxin production in silage should be started in the field and continue during harvesting, ensiling and feeding [7]. The ensiling process of forages may affect mycotoxin levels by degrading the mycotoxins that are formed in the field and present in the ensiled crop by modifying the formation of new mycotoxins during the ensiling period or during the aerobic phase after opening the silos. All these processes can potentially be affected by the use of silage additives [28,29]. In the current survey data, effects of different types of silage additives could not be correlated with silage mycotoxin levels, particularly since the information of additives used was missing in 18 of the 43 silages included in the survey.

Commonly occurring mycotoxin producers in silage are Aspergillus, Fusarium and Penicillium [11,14], but others such as Alternaria have also regularly been isolated [11]. Alternaria and Fusarium are often categorized as field fungi, but Aspergillus species have also been found in the preharvest stage [11]. Fusarium species are detected also after ensiling [6,7]. Penicillium is commonly detected in ensiled material [7]. Field fungi proliferate during plant growth and maturation [6,7,14]. Good agronomic practises to minimize environmental stress on the plants are needed to prevent fungal growth and mycotoxin production in the field [6,7,14]. During harvesting, the cutting height should be set so high that soil contamination does not occur to reduce the contamination risk of Fusarium spores from the soil [27]. Furthermore, harvested feed should be preserved rapidly to minimize the risk of continued growth of fungi under favourable aerobic conditions [30]. Fungal development and mycotoxin production in stored feed may occur at any time during storage if the circumstances, particularly aerobic exposure, are favourable [7,14].

Mycotoxins that were detected in the present survey are typically produced by Alternaria, Fusarium and Penicillium fungi, so they may have been produced both pre- and postharvesting. The incidence of some mycotoxins that are typically produced by Fusarium and Penicillium species were high both in grass and in whole-crop cereal silages, in line with McElhinney et al.’s [31] findings. Similarly, a high incidence of mycotoxins typically produced by Fusarium fungi was found in Driehuis et al. [23] and Schenck et al.’s [8] studies. In northern temperate regions such as Finland and Sweden, Fusarium are typical fungi-producing mycotoxins.

Penicillium species are able to tolerate typical silage environments, i.e., low pH and anaerobic conditions [32]. This could explain the incidence of toxins produced by Penicillium fungi. In the present survey, mycotoxins typically produced by Alternaria fungi were not common, as alternariol was detected in only two samples of grass silage. However, the incidence of alternariol was higher in Schenck et al.’s [8] work, where it was detected in 8% of Swedish grass silage samples. Typically, Alternaria mycotoxins are detected in grains, fruits and vegetables [33].

This survey showed that the incidence of the Fusarium mycotoxins beauvericin, but also enniatin B and enniatin B1, were high in grass silage. The high prevalence of Fusarium toxins beauvericin and enniatin B in grass silage was consistent with the findings of Panasiuk et al. [34], where silage samples were collected from farms in Poland. In Schenck et al. [8], in addition to deoxynivalenol, beauvericin and enniatin B were the most common mycotoxins detected in grass silage. Additionally, in the present survey, the incidence of enniatin B1 was in agreement with results found by McElhinney et al. [31] and Panasiuk et al. [34], where mycotoxins were analysed in grass silages collected from farms in Ireland and Poland, respectively. The available data on the toxicity of beauvericin and enniatins are limited and there is no EU Commission recommendation for the maximum values. According to EFSA [35], acute adverse health effects of feed containing beauvericin and enniatins are unlikely under current feeding practices. However, in the present survey and in previous studies [8,31,34], grass silage was a notable source of beauvericin and enniatins and, in some cases, high values were detected. Thus, grass silages may require more attention when assessing animal exposure to beauvericin and enniatins and the potential health risks related to it.

Average concentrations of beauvericin in core and surface samples were lower compared to previous studies. In the work by Schenk et al. [8], the average concentration was 248 μg/kg in the air-dried sample, and in McElhinney et al.’s [31] work it varied between 22 and 55 μg/kg DM in bales and clamps in two consecutive years. However, in the present survey, the average concentration was very high in mouldy samples, both in grass and in whole-crop cereal silages.

In the present survey, the average concentrations of enniatin A and enniatin A1 in the surface samples of grass silage were considerably higher compared to the results of McElhinney et al. [31] where concentrations varied between 11 and 29 and 32 and 50 μg/kg DM, respectively. The average concentrations of enniatin B and enniatin B1 in the grass silage, and enniatin B1 in the whole-crop cereal silage, were lower in both core and surface samples compared to McElhinney et al.’s [31] work, where concentrations of enniatin B varied between 240 and 364 and for enniatin B1 between 63 and 142 μg/kg DM. However, concentrations of enniatin B and B1 in mouldy samples in the grass silage were high in the present survey. In Schenck et al.’s [8] study, enniatin B was the most common mycotoxin in grass silage and the average concentration of 56 μg/kg air-dried sample was at the same level as in the present survey in the grass silages.

The recommendations of EU guidance values in products intended for animal feed are provided for deoxynivalenol, zearalenone, ochratoxin A and fumonisin B1 and B2 [4]. These recommendations are for cereals, cereal products, maize, maize by-products and complementary and complete feeding stuffs, and they are toxin and animal-dependent. Because there are no specified recommendations for grass-based forages, the values for cereals and cereal products were considered to apply for grass-based forages.

In the present survey, deoxynivalenol in the whole-crop cereal silage and zearalenone in the grass silage were the only EU-regulated mycotoxins that were detected. Deoxynivalenol is one of the most common Fusarium mycotoxins in cereals [35]. It was detected in two whole-crop cereal silage samples in the present survey. However, the average and maximum concentrations were clearly below the EU guidance values for cereals and cereal products. Grass silage was not a source of deoxynivalenol, consistent with results provided by Driehuis et al. [25] and McElhinney et al. [31], although Schenck et al. [8] detected deoxynivalenol in 12% of Swedish grass silage samples and, on average, 352 μg/kg DM was found in Finnish experimental pilot-scale grass silages [36]. Similarly, with the results of Driehuis et al. [37] and McElhinney et al. [31], grass silage was a minor source of zearalenone, although in Franco et al.’s [36] study, a relatively high concentration of 523 μg/kg DM was found in grass silage. The average and maximum concentrations were clearly below the EU guidance values with the exception of the average and maximum concentrations in the mouldy samples that exceeded the recommended value.

Mycophenolic acid was detected in both grass and whole-crop cereal silages, similar with the results from Schneweis et al. [38], where the proportion of the positive German samples was 37% in grass silage and 28% in maize silage. Driehuis et al. [23] found mycophenolic acid in surface samples of grass and maize silage in 13% and 50% of Dutch samples, respectively. However, in the core samples, it was not detected contrary to the present survey. Consistent with Driehuis et al. [23], concentrations in mouldy samples were clearly higher compared with non-mouldy samples. In the study by McElhinney et al. [31], the incidence of mycophenolic acid was low, and it varied between 0 and 8.9% of the samples in the two-year survey, where mycotoxin incidence in baled and clamp silages was studied. Mycophenolic acid is considered an immunosuppressing mycotoxin that may increase the risk of infectious diseases in livestock, but the level that exposes to the risk and the effects is unclear [14,38].

In the present survey, roquefortine C was detected in half or more of the core and surface samples both in grass silage and in whole-crop cereal silage. The greatest incidence and concentrations were detected in mouldy samples of grass silage and in surface and mouldy samples of whole-crop cereal silage. These results were similar with earlier studies, where a rather high incidence of roquefortine C in German silages was detected [23], although in other studies in Ireland and the Netherlands, respectively, clearly lower incidences were reported [31,37].

A notable, but expected, result of this survey was that none of the silage samples contained aflatoxin B1 in line with our hypothesis. The result was consistent with Driehuis et al. [37], where in the Netherlands both grass and whole-crop cereal silages were analysed and no positive samples of aflatoxin B1 were found. Environmental conditions expose to aflatoxins, particularly a high ambient temperature (26–35 °C) and humidity [11]. Thus, aflatoxin contaminations are usually associated with geographical regions with a tropical or subtropical climate [11], clearly differing from the climate conditions of Northern Europe. Aflatoxin B1 is transformed into hydroxylated metabolites, which can be found in milk and dairy products obtained from livestock that have ingested contaminated feed [39]. According to Zain [9] and EFSA [20], some of these compounds ingested by humans may be carcinogenic and, thus, milk and dairy products contaminated with hydroxylated metabolites from aflatoxin are a safety risk for human health. The European Union has set a maximum content for aflatoxin B1 in products intended for animal feed [3]. These values are presented in a moisture content of 12%. If the mycotoxin results are presented based on a 100% DM concentration as in the present survey, they can be converted to represent EU mycotoxin regulatory values by multiplying the results by a factor of 0.88.

There was no clear evidence for the hypothesis of a higher occurrence of mycotoxins in the second and third cuts of grass silage compared to the first cut. Mycotoxins were detected from all silage cuts in contrast to the findings of Huuskonen et al. [25], where zearalenone, roquefortine C, mycophenolic acid and HT-2 were found in the second cut, but none of the first cut grass silage samples contained any of the analysed mycotoxins. In the present survey, there was a tendency for enniatin B, enniatin B1, mycophenolic acid and roquefortine C to be more common in the second and third cuts than in the first cut. Additionally, zearalenone was detected only from the third cut samples. The more humid and hot weather during the growth period of regrowth grass and the more vegetative rather than generative type of swards could contribute to greater fungal activity and subsequent mycotoxin production in the regrowth rather than in the primary growth grass. Schenck et al. [8] found that the risk to find fungi in the regrowth harvest was higher compared to the primary growth harvest, and when fungal counts were higher, the risk to find mycotoxins increased, but the correlation was low. However, Skladanka et al. [5] detected mycotoxins in the first cut grass silage, similar to the current survey.

In the present survey, mycotoxins were found in both mouldy and non-mouldy samples. Thus, the absence of visible moulds does not always indicate mycotoxin-free feed in agreement with Raymond et al. [40]. Schenk et al. [8] found that in baled silage, the presence of visible fungi on bale surfaces did not correlate with mycotoxin presence in the core samples. However, they also found that the risk of mycotoxin prevalence increased when fungal counts in core samples increased. Schenk et al. [8] concluded that visible fungi were not a sufficient factor to assess the risk of mycotoxin presence in the forage.

In the present survey, all mouldy samples contained mycotoxins, and some at very high concentrations. In addition, mouldy samples contained more different types of mycotoxins compared to samples without visible mould. Based on these results, feeding mouldy feed to animals clearly increases the risk of exposure to mycotoxins and is a significant indication that mouldy feed should not be fed to animals.

This survey indicated that feeding silage to ruminants can expose the animals to a broad range of mycotoxins. Additionally, mycotoxin-contaminated feeds may contain several types of mycotoxins, as was found in the present survey and by others [6,8,31,34]. This increases the risk of adverse effects and may be a significant threat to human and animal health as concluded by Šegvić Klarić [41].

Although regulatory limits for some mycotoxins in feed can be found, the interpretation of mycotoxin concentrations in feed is difficult because there is a lack of guidance values regarding most mycotoxins. In addition to the need for guidance values for individual mycotoxins in feed, the total intake and the interaction of several mycotoxins should also be considered. However, experimentally testing animals with various cocktails of mycotoxins may be considered unethical, making it difficult to establish clear safety limits for mycotoxins. Further, the lack of an affordable and reliable rapid test of mycotoxins of farm silages prevents the routine evaluation of feeds. The effect of mycotoxins on rumen microbes could be detected in vitro, but effects on feed intake, immunosuppression, fertility, etc., would require large-scale in vivo experiments.

Reporting the concentrations of mycotoxins in literature is confusing, making it difficult to compare the results between individual studies. In some studies, concentrations are reported in fresh matter or on an air-dried basis without reporting the DM concentration of the samples, while in some studies results are reported on a DM basis. Further, EU maximum and guidance values of given in a moisture content of 12%. In addition, when calculating animal exposure to mycotoxins based on feed intake, it should be calculated on a DM basis rather than in fresh matter. Particularly in ensiled forages, the DM concentration may vary considerably and should be taken into account.

5. Conclusions

The incidence of mycotoxins was common in farm silages and feeding silage to ruminants can expose the animals to a broad range of mycotoxins. The high presence of mycotoxins in samples without visible mould was noteworthy. However, the concentrations of mycotoxins varied widely, and both very high and low concentrations were detected. All mouldy samples contained mycotoxins, some at very high concentrations. In addition, mouldy samples contained more of different types of mycotoxins compared to non-mouldy samples. These findings confirmed our hypothesis and emphasized that feeding mouldy feed to animals clearly increases the risk of exposure to mycotoxins and should not be conducted. The main health risks of mycotoxins carried to humans via animal products are related to aflatoxin, and it is noteworthy that it was not detected in any of the samples collected under Northern European conditions. The interpretation of mycotoxin concentration in silages is difficult because there is lack of guidance values. In addition, the total intake and the interaction of several mycotoxins should also be considered when assessing the risk of mycotoxins in feed. Grass silage requires more attention and research when assessing the exposure of ruminants to mycotoxins.