The study of Måge et al. [7] reported a contamination prevalence with DON in fish feed components of 33%, whereas others reported 50% DON contamination in samples from cereals, straw, silage and finished feed from all over the world [32]. In contrast, we were able to show that DON could be found in more than 80% of the samples from commercial fish feed showing average values of 289 μg kg⁻¹ feed (Table 1). A maximum value for DON of 825 μg kg⁻¹ was noted for feed #6. However, its metabolite DOM-1 was not found in the samples. Thus, the DON values in fish feed do not exceed the guidance values for complete feedingstuff [31].

Again Måge et al. [7] noted that 33% of the samples taken from fish feed components were contaminated with ZEN, whereas another study reported that in 56% of the samples from finished animal feed ZEN was found with average levels of 99 μg kg⁻¹ feed [6]. In Asian countries ZEN was found in 9 fish feed samples showing average concentrations of 76.2 μg kg⁻¹ feed [8]. Similarly, in our study ZEN was found in all samples (Table 1), although the concentrations varied considerably (average 67.9 μg kg⁻¹ feed, ranging from 3 to 511 μg kg⁻¹ feed). Consequently, the ZEN values found in our study do not exceed the values currently recommended by the European Commission [31].

According to the recommendation of the European Commission [31] ZEN should not exeed levels of 2 mg kg⁻¹ feed material (with the exception of maize by-products) and DON should be lower than 5 mg kg⁻¹ complete feedstuff. According to our findings for the samples from fish feed these values are not exeeded. However, recent studies showed that DON can be harmful to fishes at levels below these recommendations [23,24,25]. For ZEN the consequences of food-borne exposure remain mostly unknown [23]. Still, the frequent exposure of fish to single mycotoxins and combinations of different mycotoxins in aquaculture represents a continuous health risk [29,33,34]. Unfortunately, the actual risks remain more or less unknown due to a lack of data which provide evidence for effects in aquatic animals and which clarify the mechanism(s) of action of these substances. In particular, the effects of DON on certain aspects of fish health and cell function in vivo have rarely been investigated in fish [24]. However, this is an important issue for future research since economic consequences for the animal industry can be minimized.

Chemical substances and natural toxins in the food chain received increasing attention recently which led to risk assessments and the development of different prevention strategies [35,36]. One strategy would include the selection of proper ingredients for feed production. Not surprisingly, even within one production firm the batches of ingredients seem to influence the mycotoxin content of the final feed enormously. This can be seen from feed #3 to #5 which were produced by the same company by using similar ingredients and the same production processes but differed in final pellet size. Nevertheless, the contents of DON and ZEN differ considerably between these three samples (Table 1). Moreover, the contamination of the tested fish feed samples in our study was more prevalent than previously expected from other studies [6,7]. This is certainly due to the combination of ingredients that were chosen for feed production for cyprinids. A quite compatible protein source in fish feed would be reached by inclusion of fish meal at high percentages. However, the availability of fish meal for production of aquaculture feeds is decreasing globally over the last decades and the fish meal prices are increasing accordingly [5]. Thus, alternative protein sources need to be used for fish feed production. A higher percentage of cereals is commonly used for feed production for cyprinids compared to carnivorous fishes [11,12]. But natural contamination of commercial fish feed with DON and ZEN seems to be higher when corn and cereal components are used at higher percentages whereas fish meal-based feed mostly contained less of these mycotoxins (Table 1 and Table 2). This assumption is supported by the finding that the most prevalent mycotoxin in central Europe was DON with maximum values of 14.1 mg kg⁻¹ barley and 49.0 mg kg⁻¹ wheat (found in Austria) and average levels of ZEN and DON of 1.8 mg kg⁻¹ and 1.7 mg kg⁻¹ in corn gluten meal world-wide, respectively [6]. Soybean and soybean meal are considered to be less contaminated with DON and ZEN [6,37] and the usage of these plant components instead of cereals probably leads to less contamination of animal feedstuffs with Fusarium toxins. However, soybean and soybean by-products can be used for fish feed production only at relatively low percentages (depending on the fish species) because they exert estrogenic effects and negative impacts on fish growth at higher levels [38,39]. Thus, the contribution of certain ingredients to the mycotoxin concentrations in fish feed and their possible consequences on fish health should be the subject of future research.

However, a general problem during our collection of samples was that although the producers of fish feed have to give an ingredient declaration of compounds in feedingstuffs by (percentage) weight of inclusion, only the producer of feed #10 displayed percentages on the packaging. All others refused to declare the percentages even on repeated enquiry. Therefore, an improved labeling policy would help to identify and prevent sources of mycotoxin inclusion in animal feed.

It is also known that the storage of ingredients and the production processes influence mycotoxin contents in fish feed. Pelleted feed are increasingly replaced by extruded diets. Extrusion of feeds increases the digestibility of nutrients. The process of extrusion comprises an increase of temperatures to 110–160 °C. In our study feed #7 and feed #9 are extruded but, both of them still contain ZEN and feed #7 still contains DON. Studies on effects of high temperatures and high pressure on DON concentrations during production of food and feed have yielded contradicting results [40,41,42]. However, extrusion cooking clearly led to the reduction of DON concentrations in wheat grits depending on the adjustment of physicochemical parameters [43]. Moreover, extrusion can influence the content of ZEN [44]. We have no proof whether the ZEN content in samples from feed #7 and feed #9 was influenced by the production process. Nevertheless, according to the literature the selection of the appropriate course of production processes in addition to optimized storage conditions of ingredients and finished feed should lead to less mycotoxins in feedstuffs to some extent.

In addition, some research group focus on strategies of biological detoxification of mycotoxins, the use of different sorbents or the use of genetically-modified crops for feed production in order to achieve less mycotoxin contamination in feed [45,46,47].

For analyses of DON and ZEN 11 samples from fish feed from central Europe were collected. The samples that were chosen were mainly designed for use of carp feeding, since cyprinids display a major group of species in freshwater aquaculture [5]. All samples were taken from complete feedstuffs with the exception of feed #1 which is designated as a complementary feed for carp. Samples were stored at −20 °C before extraction of mycotoxins.

DON and DOM-1 in fish feed were analyzed by HPLC-DAD (high performance liquid chromatography (consisting of a pump (LC-10ADVP), an autoinjector (SIL-10ADVP), and a column oven (CTO-10ACVP)) from Shimadzu (Duisburg, Germany)) with diode array detection using the detector SPD-M10AVP (Shimadzu, Duisburg, Germany) after a clean-up with IAC (immuno-affinity columns, DONprep^TM, R-Biopharm, Darmstadt, Germany) according to manufacturer’s procedure with slight modifications as described previously [48]. The detection limit was 30 μg kg⁻¹, the mean recovery was approximately 90%.

ZEN in feed was determined by HPLC with fluorescence detection after a clean-up with IAC (ZearalaTest, Vicam, Klaus Ruttmann, Hamburg, Germany) according to a slightly modified method of the “Verband Deutscher Landwirtschaftlicher Untersuchungs- und Forschungsanstalten” [49]. The detection limit was 2 μg kg⁻¹ and mean recovery was approximately 79%.