Introduction
Preservation of cereal grains by crimping is based on lactic acid fermentation by anaerobic lactic acid bacteria (McDonald et al., Reference McDonald, Henderson and Heron1991). The concept is the same as with grass and maize silage preservation, and the method is referred to generally as grain ensiling. If the ensiling is conducted carefully, the pH of the ensiled feed reduces to 4 or even below due to the presence of undissociated organic acids (lactic, formic, acetic, propionic and other volatile fatty acids [VFAs]), which are either added or produced by fermentation and prevent the growth of microbes. Crimping the grain breaks and flattens the whole grain, exposing the carbohydrate and protein, which can positively increase grain compaction and create a more anaerobic environment inside the silo, stimulating microbial activity due to greater surface area and substrate availability (Klemola et al., Reference Klemola, Järvenpää and Peltola1994; Bern, Reference Bern1998).
Crimping grains enables farmers to harvest, process, store and preserve moist cereals as animal feed without the use of expensive drying facilities and energy consumption for drying. Crimping grain offers an opportunity to harvest cereals earlier than conventional harvesting for dry grain (generally 15–20% moisture content [MC]) and harvesting can be carried out under more humid conditions, reducing the weather dependency of harvesting. Dry weather conditions may result in drier than optimal grains and this reduced moisture has technological advantages in harvesting and logistics but may compromise preservation due to aerobic spoilage. The unpredictable weather conditions due to climate change may alter the harvesting conditions of grain and thus flexibility of storage methods may be increasingly important in the future.
Ensiling of crimped grain has been successfully performed when the crop is harvested with 30–40% MC (Huhtanen et al., Reference Huhtanen, Jaakkola and Nousiainen2013) and when ensiling is based on a low pH generated by fermentation and/or acid-based additives (Vanhatalo et al., Reference Vanhatalo, Jaakkola, Rauramaa, Nousiainen and Tommila1999; Jaakkola et al., Reference Jaakkola, Saarisalo and Heikkilä2009). According to Huhtanen et al. (Reference Huhtanen, Jaakkola and Nousiainen2013), extensive research in Finland in the 1970s demonstrated that ensiling of high-moisture grain is an efficient storage method as an alternative to grain drying. Nowadays the ensiling of low-MC crimped grain is common, although there is a high risk of losses when additives are not used (Olstorpe et al., Reference Olstorpe, Schnürer and Passoth2010; Pieper et al., Reference Pieper, Hackl, Korn, Zeyner, Souffrant and Pieper2011). In crimped grain with MC between 20 and 30%, fermentation is restricted, so efficient protection is needed against aerobic deteriorating organisms. Harvesting should start when the grain has started ripening, since the MC required for effective lactic acid fermentation is high. However, if MC of grain is too low, water can be added to grain during the crimping to increase MC (Klemola et al., Reference Klemola, Järvenpää and Peltola1994; Palva et al., Reference Palva, Jaakkola, Siljander-Rasi, Valaja, Root, Peltonen, Palva, Kirkkari and Teräväinen2005), although under practical conditions this may not always be feasible.
In some cases, the conventional method of evaluating crimped grain aerobic stability (AS) by a rise in temperature may not be sensitive enough. In a previous experiment (Seppälä et al., Reference Seppälä, Mäki, Orkola, Rinne and Udén2015), rather dry (MC 19%) crimped grain did not show elevated temperatures although it was visually clearly mouldy during the measurement period. It is possible that the heat is lost to the environment from relatively dry grain so that it is not retained in the sample. Similarly, Seppälä et al. (Reference Seppälä, Heikkilä, Mäki and Rinne2016) did not observe heating of pre-wilted grass silage (MC 460 g/kg), which may have been due to the same reason, suggesting that a more sensitive methodology might be needed for other types of feeds as well.
An alternative method to study microbial activity in a sample is to follow the formation of carbon dioxide (CO2; Ashbell et al., Reference Ashbell, Weinberg, Azieli, Hen and Horev1990). Further, visual inspection of moulds growing on the samples can be conducted (Seppälä et al., Reference Seppälä, Mäki, Orkola, Rinne and Udén2015). This method gave useful results in the experiment of Rinne et al. (Reference Rinne, Franco, Jalava, Järvenpää, Kahala, Blasco, Siljander-Rasi and Kuoppala2019), when the stability of moist carrot by-products was evaluated.
The objective of the current study was to evaluate the fermentation quality, microbial composition and AS of low-MC crimped wheat grain when manipulated by the use of different organic acid-based additives. The feeds were used to compare different methods of evaluating the AS and comparing their ability in detecting differences between treatments. Three methods in evaluating the AS were compared: (1) increase in temperature, (2) production of CO2 and (3) visual appearance of moulds.
Materials and methods
Raw materials
Autumn wheat (Triticum aestivum, variety Urho, Boreal Plant Breeding Ltd., Jokioinen, Finland) sown on 7 September 2016 and harvested on 25 September 2017 at the Natural Resources Institute Finland (Luke) in Jokioinen, Finland (60°48′N, 23°29′E) was used as the raw material. The field was fertilized on 2 May 2017 (108 kg nitrogen (N)/ha, in the form of ammonium nitrate) and sprayed against weeds (Ariane S, 200 litres/ha) on 22 May 2017. After harvesting, the grain was crimped immediately without additives at the Häme University of Applied Sciences in Tammela, Finland using a farm scale crimper (MD 700 HD, Murska Ltd., Ylivieska, Finland). The material was stored in a refrigerator overnight before ensiling was conducted at the Luke Laboratory.
Treatments and experimental procedures
Eight additive treatments were used:
(1) control, without any additive;
(2) FA1 (pure formic acid [FA] diluted to 80% [81.63 ml FA + 18.37 ml water = 100 ml 80% FA], 5 litres/tonne);
(3) FA2 (58% FA, 20% propionic acid [PA], 2.5% potassium sorbate, 5.2% sodium formate [SF]; AIV Ässä Na, 5 litres/tonne);
(4) FA3 (AIV Ässä Na, 7 litres/tonne);
(5) FA4 (76% FA, 5.5% SF; AIV 2000 plus Na, 7 litres/tonne);
(6) FA5 (37% FA, 22% SF, 18% PA, 7.3% sodium, 1% sorbic acid; GrasAAT SX, 5 litres/tonne);
(7) PA1 (54% PA, 31.3% ammonium propionate, 5% sodium benzoate [SB], 0.1% glycerine, 0.1% propylene glycol; Eastman Stabilizer Crimp, 5 litres/tonne) and
(8) PA2 (37% PA, 14% SB, 10% FA, 11% sodium propionate; Kofa Feed EP, 4 litres/tonne).
Additive FA1 was sole laboratory grade formic acid from the Merck Group (Darmstadt, Germany, Emsure®, ACS, Reag. 1.00264, 98–100%). Additives FA2, FA3, FA4 and PA1 were products of the Eastman Chemical Company (Oulu, Finland) and FA5 and PA2 were produced by ADDCON GmbH (Bitterfeld-Wolfen, Germany). Details of the additives are presented in Table 1. Additives were dosed according to the commercial instructions, except for FA3 which was the same product as FA2 but used at a higher amount (5 v. 7 litres/tonne) to evaluate the dose response.
Description of additives used in the experiment
Glass jars of 1.5 litre volume were used to preserve the crimped grain. Three replicates were made for each additive treatment, which resulted in 24 different batches for which additives were applied separately. To have enough material, three jars were prepared per replicate. Four kilograms of crimped wheat grain were weighed into a plastic container and the additive was mixed carefully. Additives were diluted in water to have an even application so that total amount of liquid was 25 ml for one 4 kg batch. The volume of preservative solution (additive plus water) was the same for all treatments. The glass jars were filled as full as possible to minimize the volume of air in the headspace, resulting in a density of 577 kg dry matter (DM)/m3. The top of the jar was covered with a plastic film, closed airtight and stored at room temperature (20 °C) in the dark.
The ensiling period lasted for 57 days. When opening the jars, spoiled parts were discarded and the three jars from the same replicate were combined and mixed carefully before sampling.
Preservation losses
The jars were weighed before opening. The weight losses were calculated according to Knicky and Spörndly (Reference Knicky and Spörndly2015) by assuming the lost weight to be CO2 leaving the jar during the fermentation phase. It was assumed that for each mole of CO2, 1 mol of water (H2O) was produced: for each gram of weight decrease due to CO2 loss, 0.44 g of DM in the silo became water, which represents loss even if it is still inside the jar. Therefore, the DM loss was calculated as the decrease in weight of the silo multiplied by a factor of 1.44, expressed as grams per kg initial DM.
Visual inspection of mould was conducted before emptying the jars. A mould scale from 0 to 3 (0 = no mould, 1 = slight mouldiness, 2 = moderate mouldiness, 3 = severe mouldiness) was used for assessing the visual presence of moulds separately on the sides and top of the jars.
Aerobic stability measurements
The AS of the crimped grain samples after opening the jars was measured using three different methods.
Temperature measurement
As the standard method, samples of crimped grain (580 g DM) were inserted into 2.5 dm3 polystyrene boxes in plastic bags, which were left open. Thermocouple wires were inserted into the feed material within the polystyrene boxes and connected to a data logger, which measured the temperature changes of the samples automatically at 10-min intervals. AS was defined as the time taken to increase the temperature of samples by 2 °C above the ambient temperature, which was 22 °C. The data collection lasted for 174 h. A similar approach is commonly used in AS measurements internationally, with some modifications related to practicalities, and has been used routinely in the Luke Laboratory (e.g. Seppälä et al., Reference Seppälä, Heikkilä, Mäki and Rinne2016).
Increase of carbon dioxide concentration
For this method, 50 g of sample was placed in 0.5 litre glass bottles and closed, airtight, with a lid that allowed sampling of the gas from the headspace of the bottle. The large headspace volume was used intentionally so that the growth of aerobic bacteria should not be limited by availability of oxygen. The bottles were kept at room temperature (25 °C). Once daily, 200 µl of gas was withdrawn from the headspace of each bottle with a gas-tight syringe and analysed for CO2 using a gas chromatograph (Perkin Elmer, Arnel Clarus 500, Illinois, USA, equipped with a thermal conductivity detector and a Supelco Carboxen™ 1010 PLOT fused silica capillary column [30 m × 0.53 mm]). First, gas was pumped in and out to ensure mixing of the gas. The amount of gas produced was also recorded. Thresholds of 1 and 2 mg of CO2 per g DM were used to indicate the loss of AS.
Visual appearance of mould
A third method to evaluate AS was the visual appearance of mould. Visual inspection of the bottles used for CO2 measurement was conducted once a day. The spoilage of the samples was evaluated visually using the following scores: 0 = no mould, 1 = slight mouldiness, 2 = moderate mouldiness, 3 = severe mouldiness. Samples were discarded when they reached score 3. The results of this test are presented as the number of days the samples remained slightly moulded (score 1) or moderately moulded (score 2) and as the average daily mouldiness scores.
Laboratory analyses
Crimped wheat grain was sampled as a raw material before applying the additive treatments and analysed for DM, ash, crude protein (CP), neutral detergent fibre (NDF), starch and microbial quality (aerobic bacteria, yeasts and moulds). Ensiled wheat grain was analysed for DM, water soluble carbohydrates (WSCs), pH, ammonia-N, ethanol, lactic acid, VFAs and microbial quality (yeasts and moulds).
Chemical analyses
Chemical analyses were carried out at the Luke Laboratory in Jokioinen. The laboratory has a quality system which follows the SFS-EN ISO/IEC 17025:2005 standards and is accredited by FINAS (the Finnish Accreditation Service) with number T024. The %DM was determined by drying at 105 °C for 16 h and corrected for volatile losses. Ash content was determined by ignition of the samples at 600 °C for 2 h (AOAC, 1990, method 942.05). Nitrogen (N) content was determined by the Dumas method (AOAC, 1990, method 968.06) using a Leco FP 428 nitrogen analyser (Leco, Saint Joseph, USA). CP content was calculated as 6.25 × N content.
Ash-free NDF on organic matter basis (aNDFom) was determined according to Van Soest et al. (Reference Van Soest, Robertson and Lewis1991) using sodium-sulphite and α-amylase. Starch was determined according to Salo and Salmi (Reference Salo and Salmi1968). VFAs were determined according to Huhtanen et al. (Reference Huhtanen, Blauwiekel and Saastamoinen1998), lactic acid according to Haacker et al. (Reference Haacker, Block and Weissbach1983), WSC according to Somogyi (Reference Somogyi1945) and ammonia according to McCullough (Reference McCullough1967). The N content of the original material was used to express ammonia-N proportions in total N after fermentation. Ethanol was measured using an enzymatic kit (cat no. 981680, KONE Instruments Corporation, Espoo, Finland) and the selective clinical chemistry analyser Pro 981489 (KONE Instruments Corporation, Espoo, Finland) according to application instructions given by KONE. Ammonia and propionic acid were corrected for the amounts added in the additives. Further, a correction factor of 0.8 was used (corrected value = analysed value − 0.8 × amount added in the additive) to represent the losses of added substances based on unpublished observations to avoid overcorrection.
Microbiological analyses
The samples for microbiological analyses were mixed and 25 g was weighed in stomacher bags and mixed with 225 ml of ¼-strength Ringer solution (Merck 1.15525.0001). The samples were homogenized using the stomacher (Stomacher® 400 Circulator) for 2 min at 230 rpm. Serial decimal dilutions were prepared by mixing 1 ml of sample with 9 ml of Ringer solution.
Yeasts and moulds were determined in Dichloran Rose Bengal Chloramphenicol Agar medium (Lab M Ltd., Bury, UK, LAB217) which was supplemented with 50 µg/ml of oxytetracycline hydrochloride (AppliChem BioChemica, Billingham, UK, A5257). The Petri dishes were incubated at 25 °C. The colonies were counted after 3 and 5 days. The aerobic plate count was determined on Plate Count Agar (Lab M Ltd., Lab010) dishes incubated at 30 °C for 72 h.
Statistical analysis
The data were analysed using the MIXED procedure (SAS Inc., 2002–2012, Release 9.4; SAS Inst., Inc., Cary, NC, USA) of SAS at P < 0.05. Additives were considered fixed effects, while replicates were taken into account in the model as the random effect. Differences between least square means of response variables were estimated with Tukey's test. A linear effect for dose response of the additives FA2 and FA3 (same compound in different doses) was also evaluated. A linear correlation between AS measurement methods was evaluated using the CORR procedure of SAS.
Results
MC of wheat raw material was 19% and contained 20 g ash, 139 g CP, 670 g starch and 126 g aNDFom per kg DM. The MC of wheat grain after addition of water was 28%. Total bacteria, yeast and mould counts of the wheat raw material were 1.4 × 107, 7 × 104 and 2.3 × 105 CFU/g, respectively.
Table 2 presents the fermentation quality and microbial composition of crimped wheat grain according to additive treatments. Overall, MC was similar for all treatments (P > 0.05) ranging from 26.9 to 27.6%. This resulted in no production of lactic acid for the control, FA1, FA2, FA3, FA4 and FA5 and very low amount for PA1 and PA2. Significant ethanol formation occurred (P < 0.05) in the control. Additive treatments had lower pH than the control (P < 0.05) and FA3 resulted in the lowest pH value among treatments (P < 0.05). There was very little acetic and butyric acid in the samples, ranging from 0.55 to 1.35 g/kg and from 0.03 to 0.04 g/kg, respectively. For propionic acid, the results changed clearly after they were corrected for the amount of propionic acid delivered in the additives, and significant differences among treatments disappeared. Ammonia-N was similar for the control and additive treatments (P > 0.05), but higher for PA1 (P < 0.05) when the correction of ammonia originating from the additive was not taken into account. However, when the correction was applied, the amount of ammonia-N was similar (P > 0.05) among treatments.
Fermentation quality and microbial composition of crimped wheat grain according to additive treatments
FA1, formic acid; FA2, AIV Ässä Na; FA3, AIV Ässä Na; FA4, AIV 2000 plus Na; FA5, GrasAAT SX; PA1, Eastman Stabilizer Crimp; PA2, Kofa Feed EP; SEM, standard error of the mean; N, nitrogen; CFU, colony forming unit.
Values with the same letter in a row are not significantly different (P > 0.05) based on Tukey's test. For additive treatments, see Table 1.
a Standard error.
b Effect of treatment.
c Not corrected for the amount delivered in the additive.
d Corrected for the amount delivered in the additive.
e Colony forming unit.
Numbers of yeasts were lower in all treatments compared to the raw material (Table 2). Treatments with PA1, FA1 and FA4 resulted in the highest number of moulds, which were even higher than in the raw material, while the control and FA3 showed the lowest mould numbers.
The control resulted in the lowest (P < 0.05) amount of WSC (Table 2), followed by PA2, with consequently greater (P < 0.05) amounts of ethanol and acetic acid for those treatments compared to the others.
The average ensiling losses were rather low for all treatments as they remained below 15 g/kg of initial DM (Table 3). Among the treatments, the control had the greatest losses (P < 0.05) followed by PA2 while FA3 resulted in the lowest loss (P < 0.05) and the other treatments resulted in similar losses ranging from 2.9 to 4.1 g/kg of initial DM.
Ensiling losses and spoilage score on top and sides of jars prior opening and mould score according to days after air exposure of crimped wheat grain treated with additive
FA1, formic acid; FA2, AIV Ässä Na; FA3, AIV Ässä Na; FA4, AIV 2000 plus Na; FA5, GrasAAT SX; PA1, Eastman Stabilizer Crimp; PA2, Kofa Feed EP; SEM, standard error of the mean; DM, dry matter.
Values with the same letter in a row are not significantly different (P > 0.05) based on Tukey's test. For additive treatments, see Table 1.
a Standard error.
b Effect of treatment.
c Corrected values according to Knicky and Spörndly (Reference Knicky and Spörndly2015).
After 57 days of ensiling, the jars were reasonably mouldy on both the top and sides (Table 3). PA1 had the lowest mould score (P < 0.05), followed by the control and FA3. The other treatments (FA4, FA1 and FA5) presented high appearance of spoiled parts (P < 0.05) both on the top and sides of the jars, which were in general above a score of 2.
The AS based on temperature rise was the shortest in treatments FA1, FA5, PA2 and the control (P < 0.05; Table 4) followed by FA4. The longest AS was achieved in FA3 (P < 0.05); however, the temperature for this treatment did not even increase 2 °C above the ambient during the whole observation period of 174 h; and PA1 with AS of 168 h. Additives were effective at increasing (P < 0.01) AS when compared to the control or formic acid alone (FA1). There was a positive linear (P < 0.01) dose response to the same additive which was used in FA2 and FA3 treatments in incremental amounts.
AS (h) of crimped wheat grain according to additive treatments measured through standard and alternative methods, and correlation between increase in temperature and the alternative methods
FA1, formic acid; FA2, AIV Ässä Na; FA3, AIV Ässä Na; FA4, AIV 2000 plus Na; FA5, GrasAAT SX; PA1, Eastman Stabilizer Crimp; PA2, Kofa Feed EP; SEM, standard error of the mean.
Values with the same letter in a row are not significantly different (P > 0.05) based on Tukey's test. For additive treatments, see Table 1.
a Standard error.
b Effect of treatment.
c Correlation coefficients between increase in temperatured and the alternative methods to evaluate AS.
d Time taken to increase 2 °C above ambient temperature.
e Treatment did not reach the threshold during the evaluation period.
f Time taken to increase the carbon dioxide concentration up to 1 and 2 mg/g DM, respectively.
g Time taken to sample reach mould score 1 and mould score 2, respectively.
Cumulative CO2 production of crimped wheat grain preserved with FA1 was the highest and fastest among the treatments (Fig. 1). Its cumulative CO2 production decreased after 6 days of evaluation, but still remained higher than in other treatments. The control, FA4, FA5 and PA2 exhibited similar patterns for cumulative CO2, where production suddenly increased between days 3 and 5 and then reached a plateau with a flat stationary phase between days 5 and 7. Treatment FA2 kept a low CO2 production during the initial 4 days of evaluation and afterwards increased linearly up to 7 days. Treatment PA1 presented a low cumulative CO2 production throughout the evaluation period with a slight increase during the last 2 days. Cumulative CO2 production in FA3 remained rather close to zero throughout the whole evaluation period.
Cumulative carbon dioxide (CO2) production of ensiled crimped wheat grain according to days after air exposure and additive treatments. C: control; FA1: formic acid, 800 g/kg, 5 litres/tonne; FA2: formic acid, 580 g/kg, 5 litres/tonne; FA3: formic acid, 580 g/kg, 7 litres/tonne; FA4: formic acid, 760 g/kg, 7 litres/tonne; FA5: formic acid, 370 g/kg, 5 litres/tonne; PA1: propionic acid, 540 g/kg, 5 litres/tonne; PA2: propionic acid, 370 g/kg, 4 litres/tonne.
According to visual inspection of mould in crimped wheat grain after air exposure (Table 3), the mould score was greatest for FA1, which was severely mouldy after 5 days, followed by FA5, FA4 and FA2. The control and PA2 were moderately mouldy after 1 week, followed by PA1, which was slightly spoiled at days 6 and 7 of evaluation. FA3 treatment did not show any spoilage during the whole evaluation period.
There was a positive correlation between AS (measured by increase in temperature) and CO2 production (Table 4). Increase in temperature above the ambient and CO2 production methods ranked the treatments in the same order. However, the concentrations of CO2 at the time the treatment reached 2 °C above ambient temperature were not the same, indicating that the amount of microbial activity was not the same. The R 2 between increase in temperature and increase in CO2 concentration using 1 mg/g DM as a threshold was 0.936 and even higher when 2 mg/g DM was used as the threshold, when R 2 reached 0.979. Visual mould score also showed a positive correlation with an increase in temperature, but with lower R 2 values (0.567 and 0.528 for samples reaching mould scores 1 and 2, respectively).
Discussion
Manipulation of crimped grain quality by additives
The MC of wheat grain at harvest was relatively low but still higher than the default value of 14% for dry cereal grains, while ash, CP, starch and aNDFom amounts were typical, corresponding to average values in the Finnish Feed Tables (Luke, Reference Luke2019). Water addition was based on microwave DM determination, which slightly underestimated MC so the crimped wheat grain MC was 28% at ensiling, which was slightly higher than the target but still lower than recommended for crimping. According to Wilkinson and Davies (Reference Wilkinson and Davies2013), low counts of yeasts and moulds at harvest are primordial factors to increase the AS of silage. Low contamination of the raw material was achieved in the current experiment according to Kristensen et al. (Reference Kristensen, Sloth, Højberg, Spliid, Jensen and Thøgersen2010), which indicated that at the time of harvest, counts of yeasts and moulds above 6.36 and 5.64 log10/g, respectively, are correlated with low stability in maize.
The low amounts of lactic acid and VFA are due to the low MC (Kung, Reference Kung2010), which inhibits fermentation. As a result of low organic acid production, the grain is more prone to aerobic spoilage as undissociated organic acids are strong inhibitors for aerobic microbes. These results are also supported by Seppälä et al. (Reference Seppälä, Nysand, Mäki, Miettinen, Rinne, Kuoppala, Rinne and Vanhatalo2012), who studied the effect of propionic and formic acid-based additives on crimped barley grain. Under these conditions, low-MC grains need to be artificially preserved with the use of additives. The absence of lactic acid also increases the pH of crimped grain to a level that allows opportunistic bacteria and moulds to grow and further reduce silage quality (McDonald et al., Reference McDonald, Henderson and Heron1991).
The proportion of ammonia N in total N is generally considered as a good indicator of silage fermentation quality (McDonald et al., Reference McDonald, Henderson and Heron1991), but if additives that include ammonia have been used, it is necessary to correct for the amount delivered in the additives. Ammonia-N was similar for the control and other additive treatments except for PA1, which was treated with an additive containing ammonium propionate. Correction of ammonia-N provided by the additive resulted in a similar ammonia-N amount in PA1 as in the other treatments. According to Wilkinson (Reference Wilkinson1990), grass silage with ammonia N in a range of 50–100 g/kg total N is regarded as well fermented, while grass silage having ammonium-N below 50 g/kg total N is very well fermented. In the current work, all samples remained far below this value, but it is noteworthy that guidelines for crimped grain have not been established.
The readily available sugar, presented herein as WSC, was almost totally consumed in the control treatment, followed by PA2, which was responsible for greater amounts of ethanol and acetic acid in both treatments compared to the other treatments. In other treatments, WSC remained high due to restricted fermentation, which may represent a potential source of readily available substrate for the growth of aerobic microorganisms when the silos are opened and exposed to air (Wilkinson and Davies, Reference Wilkinson and Davies2013). However, this concept is not well established, since other researchers (Ohyama et al., Reference Ohyama, Hara, Masaki and Thomas1980; Pahlow et al., Reference Pahlow, Merry, O'Kiely, Pauly, Greet, Park and Stronge2005; Wróbel et al., Reference Wróbel, Zielinska and Suterska2008) have identified no direct relationship between AS and residual WSC amount, and in the current data those treatments did not show poor AS.
In general, the majority of DM losses are due to aerobic deterioration occurring during storage and even short-term exposure to air results in losses (Kung, Reference Kung2010). Furthermore, much of the lost DM is of high quality, e.g. in terms of digestibility. The DM losses of additive treated grain remained rather low when compared to the control, indicating the great efficiency of organic acid-based additives to prevent preservation losses.
The visual appearance of mould reflects an advanced stage of spoilage and extensive aerobic deterioration, but the beginning of spoilage is associated with the development of yeasts and aerobic acetic acid bacteria, which are not visually detectable (McDonald et al., Reference McDonald, Henderson and Heron1991; Wilkinson and Davies, Reference Wilkinson and Davies2013). In addition to gaseous losses and the need to discard feed, there is potential for mycotoxins to be produced in feed and they are known to pose serious health and production risks to animals (da Rocha et al., Reference da Rocha, Freire, Maia, Guedes and Rondina2014); however, mycotoxins were not analysed in the current study.
A relatively large amount of crimped grain was discarded from each jar to avoid sampling in the spoiled area. As moulds are aerobic microorganisms, their appearance shows that there was oxygen in the jars either from the beginning of ensiling or the lids were not properly airtight. Moulds are able to grow in lower moisture compared to yeasts, as indicated by the results of microbiological analysis. This reveals how challenging this moisture area is for ensiling. In the current experiment, yeast counts for all treatments remained below the established level of 105 CFU/g (McDonald et al., Reference McDonald, Henderson and Heron1991; Borreani and Tabacco, Reference Borreani and Tabacco2010) set as a threshold ensuring no reduction in grain AS. In practice, mouldy spots or mouldy layers are the problem detected when ensiling low moisture-crimped grains. Kung (Reference Kung2010) reported that management factors in controlling microorganism development are more important and efficient at the time of ensiling than afterwards, during feed out or total mixed ration preparation.
The control treatment, although without additive, resulted in rather low presence of mould spots. This was most likely due to high fermentation during the ensiling phase, converting WSC into ethanol and acetic acid, which inhibited microbial activity during the aerobic phase. The rapid reduction in pH of the grain silage is a direct effect of incremental concentrations of lactic acid; however, it has poor antifungal features. According to Woolford (Reference Woolford1975), acetic and propionic acids have a good effect on controlling microbial growth in grain silage, which could be clearly seen in the control and both propionic acid-based additive treatments in the current study.
The FA2, FA3 and PA1 additives tested in the current experiment were efficient at increasing the AS of wheat crimped grain. According to Wilkinson and Davies (Reference Wilkinson and Davies2013), evaluation of AS is a primordial factor in ensuring that the feed is well-preserved and safe in terms of minimal presence of moulds before providing it to ruminants. Fermentation acids start to oxidize as soon as the silo is opened and exposed to air and lactic acid is a substrate for growth of aerobic microbes (Pahlow et al., Reference Pahlow, Muck, Driehuis, Oude Elferink, Spoelstra, Buxton, Muck and Harrison2003). Formic acid-based additives, except for pure formic acid (80%), and the propionic acid and ammonium propionate-based additive (PA1), proved to be efficient at prolonging the AS of crimped wheat grain. According to commercial instructions, PA2 should not be mixed with water, which was done for all additives to promote even distribution in the raw material. This practice may have resulted in crystallization of benzoic acid and subsequently reduced the efficiency of PA2 in the current study.
Comparison of methods to evaluate aerobic stability
The microbial oxidation of organic acids and WSC leads to an increase in temperature of silages above ambient during the aerobic phase, converting those compounds into carbon dioxide and water (Ranjit and Kung, Reference Ranjit and Kung2000). This effect is easily observed in grass silages with higher %DM, for example, 300–500 g DM/kg fresh weight (Wilkinson and Davies, Reference Wilkinson and Davies2013). According to McDonald et al. (Reference McDonald, Henderson and Heron1991), a greater amount of heat is needed to increase the temperature of high-MC feed, but on the other hand, low-MC grains dissipate the heat produced to the environment rather than accumulate it in the feed, although signs of spoilage are spotted (Seppälä et al., Reference Seppälä, Mäki, Orkola, Rinne and Udén2015). More accurate methods are thus required to evaluate AS of low-MC materials.
As described previously, the additive treatments used in the current experiment resulted in different ASs that could be detected with different methods, which allows a comparison of the methods. The correlation between CO2 production (Ashbell et al., Reference Ashbell, Weinberg, Azieli, Hen and Horev1990) and increasing temperature was very close, indicating that CO2 produced by aerobic bacteria can be used as a method to evaluate AS. The labour and analytical costs of the CO2 production method were higher than that of the temperature method and only a single measurement point per day is realistic, at least if the CO2 measurement is carried out manually. However, a smaller sample size can be used and it may be useful also in the cases of drier grain samples that may not accumulate heat. An ideal situation would be where CO2 measurements can be automated and it could be used when a sensitive method is needed to follow-up aerobic spoilage.
Theoretically, the reason for the reduction of CO2 during the later phases of the follow-up could be carbon absorption by the grains through the homoacetogenic acetyl-CoA (Wood-Ljungdahl) pathway performed by bacteria (Schmidt et al., Reference Schmidt, Novinski, Zopollatto, Gerlach and Südekum2018). However, this pathway has not been described for silages or any other feed during the aerobic phase. According to Wood and Ljungdahl (Reference Wood, Ljungdahl, Shively and Barton1991), anaerobic bacteria are able to reduce CO2 into acetate, which is a well-established pathway. Although still not proven in the feed production system, this approach demonstrates a potential way to improve fixation of CO2 into silage, which might even be used to mitigate greenhouse gas emissions. In the current work, a longer follow-up of CO2 production would have been needed to confirm the trend.
The visual evaluation of mould did not rank the treatments in exactly the same order as the production of heat and CO2. However, this methodology is practical, easy, efficient, low cost and yet provides useful results to compare efficacy of treatments. The visual inspection method has been used successfully in evaluating the stability of e.g. moist vegetable by-products (Franco et al., Reference Franco, Jalava, Kahala, Järvenpää, Lehto, Rinne, Udén, Eriksson, Spörndly, Rustas and Liljeholm2018).
Conclusions
The current results show significant potential to modify drier than optimal crimped grain preservation and AS by using different organic acid-based additives. There were clear differences in the efficacy of additives in improving the AS of relatively low moisture-crimped wheat grains, with FA3 (high dose of formic acid-based additive including several ingredients) being the most efficient followed by PA1 (propionic acid-based additive), while sole formic acid had the lowest performance among the treatments.
It is possible to use different methods to measure AS, since they all provided results applicable to the identification of spoilage signs of the feed. The correlation between temperature rise and CO2 production was very high, indicating that CO2 produced by aerobic bacteria can be used as a method to evaluate AS and it may be more sensitive than the method based on temperature rise for relatively dry feed materials. Visual appearance of mould ranked the additives somewhat differently, but could still provide useful information on the efficacy of treatments.
Financial support
This work was carried out as part of Rehvi project funded by the Centre for Economic Development, Transport and the Environment for Northern Ostrobothnia, Finland (Grant number 42237/58904). Eastman Chemical Company and Berner Ltd. provided the additives used and partial funding for the project.
Conflict of interest
The authors have declared that there is no conflict of interest.
Ethical standards
Not applicable.
