Effects of 00-rapeseed meal inclusion in Parmigiano Reggiano hay-based ration on dairy cows’ production, reticular pH and fibre digestibility

Abstract Hay-based diets are typically used in Parmigiano Reggiano cheese production. Parmigiano Reggiano feeding regulation prohibits 00-rapeseed dietary inclusion. The objectives of this study were to investigate the effects of substituting soybean-meal with different levels of 00-rapeseed-meal in dairy cows’ diets, to evaluate the possibility to include it in Parmigiano Reggiano regulation. The study had a Latin square design with 8 tie stall dairy cows. The isoenergetic and isoproteic dietary treatments differed in protein source (% of DM): S (0.0% 00-rapeseed-meal, 9.3% of soybean-meal), LR (3.8% 00-rapeseed-meal, 6.8% of soybean-meal), MR (8.5% 00-rapeseed-meal, 3.4% of soybean-meal), and HR (13.2% 00-rapeseed-meal, 0.0% of soybean-meal). DMI, milk production and composition, rumination and reticular-pH were recorded daily. Dietary fibre digestibility was evaluated by in vitro fermentation and milk gointrin content was quantified by liquid chromatography–tandem mass spectrometry. Statistical analysis was performed by mixed model. Milk yield, protein and casein content increased for incremental 00-rapeseed-meal dietary levels. MR treatment compared to the others resulted in higher daily reticular-pH (5.92), and fewer minutes with pH below 5.8 (467) and 5.5 (72). Goitrin was detectable in milk when 00-rapeseed-meal was included in the diet. Overall, the inclusion of 00-rapeseed-meal in the Parmigiano Reggiano type ratio did not compromise the performances of cows pointing to it as a reliable substitute for soybean-meal. Our results suggest that 00-rapeseed-meal should be tested in feeding studies to determine its effects on milk organoleptic characteristics and cheese production and quality to see if it can be included in the Italian PDO cheese regulation. Highlights 00-rapeseed inclusion on hay-based diet improve DMI, ECM and rumen health 00-rapeseed-meal introduction in Italian PDO cheese regulation is suggested


Introduction
Parmigiano Reggiano (PR) is one of the best-known and most exported Italian cheeses all over the world (Mordenti et al. 2017). It is an extra-hard, long-ripened cheese produced from partially skimmed raw cow milk and it is registered as a Protected Designation of Origin (PDO) cheese (Council Regulation (EC) No. 510, 2006). To obtain the PDO status, a product must be entirely manufactured within a delimited geographic area and follow specific processing techniques and feeding rules for cows (Ministero Politiche Agricole Ambientale e Forestali (MIPAAF) 2016; Mordenti et al. 2017; Disciplinare di Produzione del Formaggio Parmigiano Reggiano 2019).
In PR production, dietary inclusion of silages, nonprotein nitrogen sources (i.e. urea), added fat and many byproducts (like distillers) are not permitted. Dry hays must be included at least of 45-50% of dietary dry matter (DM) and alfalfa hay represent one of the main protein source in the rations. It is well known that the forage quality impacts cow performance improving dry matter intake and productivity, as well as rumen fermentations (Ferraretto et al. 2015;Fustini et al. 2017;Miller et al. 2021). Hays of good quality can be obtained harvesting plants at the proper stage of maturity and well preserved over the year . However, other protein sources are usually included in the diets to cover specific cow's requirements. Major protein sources used are represented by different by-products, especially soybean or sunflower meal, corn gluten feed, wheat bran and other protein sources like peas and lupins (Mordenti et al. 2007).
In Europe, rapeseed with low levels of erucic acid (<2%) and glucosinolates (<30 mmol/g) are called "double-zero" or "double-low" rapeseeds or 00-rapeseeds, while in North America, such varieties are identified as canola (Shahidi 1990;Newkirk 2009). 00-rapeseed is commonly cultivated in Europe, and represent the second most traded protein source in the global animal feed market (Carr e and Pouzet 2014). Actually, 00-rapeseed meal cannot be fed to the cows producing milk for PR and Grana Padano cheese (Disciplinare Grana Padano DOP 2019). The use of 00-rapeseed is not allowed because of the concern that some compounds like glucosinolates, among which goitrin is the most concentrated, could be transferred to the milk and cheese. These molecules have several effects on animal and human health, and can also compromise milk and cheese sensorial characteristics (EFSA: Scientific Panel on Contaminants in the Food Chain 2008). Bertuzzi et al. (2016) showed the possibility to detect those compounds in milk by liquid chromatography-tandem mass spectrometry.
Previous studies indicated that replacing soybean with canola meal leads to an increased dry matter intake (DMI) and milk composition, due to its better amino acid profile (Huhtanen et al. 2011;Martineau et al. 2013). Broderick et al. (2015) reported that feeding canola meal instead of soybean meal increased the feed intake, milk yield, energy-corrected milk (ECM), and true protein. However, PR cow diets, based on dry hays, have important differences compared to diets commonly fed to dairy cows in others part of the world commonly based on silages (Puhakka et al. 2016;S anchez-Duarte et al. 2019;Paula et al. 2020) or hay-silage mixture (Martineau et al. 2014(Martineau et al. , 2019. Thus, the replacement of soybean with 00-rapeseed meal in these conditions could have different effects on cow performances, rumen health and fibre digestibility.
Therefore, the objective of this study was to test the effect of 00-rapeseed meal inclusion in a typical Italian PR diet. In particular, we tested these effects in high producing dairy cows, fed hay-based diets with different protein sources, soybean or 00-rapeseed meal at different levels, and, at the same time, we verified the possibility to detect goitrin in milk produced. Our hypothesis is based on the possibility to include 00-rapeseed meal on PR type diets without any negative effects on production and rumen health.

Materials and methods
The study was conducted at the experimental dairy farm of the Department of Veterinary Medical Science of University of Bologna (Italy). All experimental procedures involving animals were approved by the University of Bologna Animal Care and Use Committee.

Experimental design and data collection
Eight multiparous Holstein cows (avg. BW 699 ± 55 kg, mean ± SD) were blocked by parity (3.13 ± 1.09 d, mean ± SD), milk production (38.53 ± 7.59 kg/d, mean ± SD), and days in milk (87 ± 28 d, mean ± SD), and used in a replicated 4 Â 4 Latin square design study with 21d periods, with 14 d of adaptation and 7 d of collection.
Animals were housed in a naturally ventilated tiestall barn and milked twice a day (0800 am, and 0730 pm) in a double-five herring-bone milking parlour. Meadow hay-based diets differed for major protein sources (soybean or 00-rapeseed meal), while other components were the same in all diets. Rations were formulated (DinaMilk5, Fabermatica, Ostiano, Italy) to mimic common diets used in the PR cheese production area of Italy, consisting of all dry and nonfermented components given as TMR, replacing soybean meal with increasing levels of 00-rapeseed meal. The objective was to formulate isonitrogenous and isoenergetic diets, and these results were achieved with different inclusion of steam flaked corn and grass hay to balance the relative content in protein and fibre of soybean and 00-rapeseed meal. Treatments were identified as control (S; 0% DM basis of 00-rapeseed meal and 9.3% of soybean-meal), low 00-rapeseed meal inclusion (LR; 3.8% 00-rapeseed meal, 6.8% of soybean-meal), medium inclusion (MR; 8.5% 00rapeseed meal, 3.4% of soybean-meal) and high 00rapeseed meal (HR; 13.2% 00-rapeseed meal, 0.0% of soybean-meal). Diets were offered in individual feed bunks as a TMR for ad libitum intake (approximately 1.10Â expected intake) once a day at 0800 am (Zago Mixer; Padova, IT).
Individual DMI was determined daily by recording feed offered and refused. Water intake was determined daily by individual water metre. Then, water intake per unit DMI was calculated. Individual feed ingredients were collected weekly and dried in a forced-air oven at 65 C for 48 h for DM. Samples of diets and orts were collected daily and dried in a forced-air oven at 65 C for 48 h for DM determination. Feedstuff and TMR were analysed for chemical composition according to the following methods: neutral detergent fibre (aNDFom) and acid detergent fibre (ADF), and crude protein (CP) according to Mertens et al. (2002), and AOAC 973.18, respectively. Starch was determined according to AOAC official method (AOAC 996.11) and ether extract according to AOAC 920.390020. Composite samples of each diet were used to evaluate particle size distribution on as-fed basis to determine the physical effectiveness factor (PEF) using a RoTap Separator (W.S. Tyler, Mentor, OH). Physically effective neutral detergent fibre (peNDF) of diets was calculated as: where PEF is the physical effectiveness factor, as described by Heinrichs (2013).
In vitro aNDFom digestibility (24, and 240 h) were determined using in vitro fermentation in buffered media containing ruminal fluid ). The four donor animals were fed respectively two the S and two the HR diet. Rumen fluid used for the inoculation was a mix of the four collected. Digestibility was performed on forage, TMR and faeces according to the procedure described by Palmonari et al. (2017aPalmonari et al. ( , 2017b. Briefly, in vitro aNDFom digestibility at 24 h and 240 h (IVNDFD24h and IVNDFD240h) was performed using the Tilley and Terry modified technique (Palmonari et al. 2017a(Palmonari et al. , 2017b. Faeces were analysed for aNDFom, ADF, acid detergent lignin (ADL), undigestible NDF (uNDF 240 ) and potentially digestible NDFom (pdNDFom), in order to calculate total tract digestibility of pdNDFom (TTpdNDFomD) according to the procedure described by Fustini et al. (2017).
In each collection period, multiple time points faecal samples were collected: on day 16 at 0800 am and 1100 pm, on day 17 at 0200 pm, on day 18 at 0500 am and 0800 pm, on day 19 at 1100 am and 0500 pm, and on day 20 at 0200 am.
TMR composition is reported in Table 1 Cows were milked twice a day and individual milk yield was recorded daily (Afimilk Information Management System; Afikim, Israel) during d 15-21 of each experimental period. Milk samples from two consecutive milkings for each cow were collected on d 17 and 18 of each period, preserved (Bronolab-W II Liquid Preservative; D & F Control Systems, Inc., Dublin, CA) and analysed by a qualified laboratory (Artest SpA, Modena, Italy) for fat, total protein, lactose, casein, somatic cell score (SCS), expressed as Log 10 base (Lacy-Hulbert et al. 1999), urea, titratable acidity, and pH, as described in Mammi et al. (2018a), and goitrin as described in Bertuzzi et al. (2016). Energy-corrected milk yield (ECM) and 3.5% fat-corrected milk (FCM) were calculated according to Davidson et al. (2008). Feed efficiency as milk/DMI (kg/kg), 3.5% fat-corrected milk/dry matter intake, and solids-corrected milk/dry matter intake were calculated during the experimental weeks.
Cows were monitored for rumination activity from day 15 to 21 of each period. Rumination time was measured using the Hi-Tag rumination monitoring system (SCR Engineers Ltd., Netanya, Israel). Data flow software analysed the rumination time with a resolution of 2 h (Schirmann et al. 2009) and calculated the rumination time during the last 24 h.
To continuously monitor pH, all cows received an indwelled wireless pH-transmitting unit (SmaXtec Animal Care Sales GmbH, Graz, Austria), which has been validated with rumen-cannulated dairy cows (Klevenhusen et al. 2014). These units (3.5 cm i.d. 12 cm long, and weighing 210 g) were calibrated following the manufacturer's instruction protocol and then manually inserted into the reticulorumen via the oesophagus, on day 14 of the first period. The units measured pH and temperature every 10 min and transmitted the data in real-time to a dock station using the ISM band (433 MHz). Both data of pH and temperature were collected using an analog-to-digital converter and stored in an external memory chip. Data of pH from d 15 to d 21of each period were analysed as daily mean pH and time (min/d) below specific cut-off points (5.5 and 5.8). Because wireless sensors were located in the ventral reticulorumen and little differences were observed between pH measured by wireless units and in the ventral rumen sac (Klevenhusen et al. 2014), the term "reticular pH" (r-pH) will be used.

Statistical analysis
The Latin square is designed to evaluate any possible statistically significant inferences even with a small number of animals involved in a study (Kononoff and Hanford 2006).
As reported in a similar study (Cavallini et al. 2018), data for DMI, milk yield and composition, feed efficiency, fibre digestibility and body weight were analysed as a replicated Latin square design with model effects for diet using the RELM procedure of JMP (version 12.0 pro, Statistical Analysis Systems Institute Inc., Cary, NC). Cow within replicate and period was considered as a random effect. Least square means were separated using the Tukey's procedure when a significant F test (p < .05) was detected. Moreover, sums of squares were further separated using orthogonal contrasts into single degree of freedom comparisons to test the significance of linear, quadratic and cubic effects of 00-rapeseed meal inclusion.
were observed in LR, HR, MR respectively, compared to control treatment (S diet). Moreover, the different 00-rapeseed levels influenced the water intake (p < .001) and the water intake/DMI (p < .001) in a cubic manner. Medium inclusion level (8.5% DM) resulted in the lowest water intake per unit DMI (p < .01). BW resulted quadratically influenced by the treatment (p ¼ .001). Daily rumination time was not influenced by the treatments, however, it linearly decreased (p ¼ .01) according to the incremental 00rapeseed and corn flakes levels, and the reduction of grass hay in the ration.
Data of daily reticular pH, time and AUC under values of 5.8 and 5.5 are reported in Table 3.
Inclusion of rapeseed led to a significant variation in daily average r-pH (p < .01), and the major difference was observed between S and HR (5.88 vs. 5.71, respectively). Diet LR showed intermediate values (5.84) between S and HR, while MR registered the highest pH value (5.92) and lower minutes and area (p ¼ .04) under the curve (AUC). In particular, AUC for pH < 5.8 resulted as: 72.1, 107.2, 165.0, and 255.0 in MR, S, LR and HR, respectively and 6.0, 15.0, 23.4, and 97.1 in MR, S, LR, and HR treatment, respectively, for pH < 5.5. Finally, inclusion of 00-rapeseed led to a cubic variation in r-pH parameters (p < .001).
Data of milk yield and components are reported in Table 4.
Milk protein concentration (g/kg) tended to be higher in rapeseed diets (p ¼ .08), as well as casein content (þ0.14 on avg, P ¼ .07), and when the daily  yield was calculated these differences resulted significant (p ¼ .01). Finally, lactose concentration (g/kg) slightly decreased (p ¼ .075) linearly. Urea values, on the contrary, tended to be lower (-4.21 mg/dl in HR compared with S diet, p ¼ .08), and negatively influenced by incremental levels of 00-rapeseed dietary inclusion (p ¼ .004, .008 and .001 for linear, quadratic and cubic contrast, respectively). Somatic Cell Score (SCS), showed no differences between diets. No differences were observed also in titrable milk acidity expressed in SH 50 degrees and pH. Analysis of goitrin content pointed out detectable values in milk produced by cows fed with 00-rapeseed meal and it increased according to incremental dietary levels (p < .001).
Data of faecal fibre fractions and fibre digestibility are reported in Table 5.
The level of 00-rapeseed meal inclusion had no effect on faecal aNDFom (g/kg DM), however, affected the other faecal parameters evaluated. Estimated faecal uNDF and TTpdNDFomD declined linearly (p < .001) from control diet S (0.0% inclusion) to the Table 4. Effect of rapeseed meal on milk production and quality in dairy cow fed PR type ration.  HR diet (13.2% inclusion). Accordingly, faecal pdNDFom content was higher in rapeseed diets (p ¼ .015 and 0.01 for linear and quadratic contrast respectively).

Discussion
TMRs were adequate in terms of nutrient composition (NRC, 2001). Results of crude protein content and other major nutrients show the isonitrogenous and isoenergetic diet formulation, similar to those observed in other commercial farms of PR area (Mammi et al. 2018a(Mammi et al. , 2018b. The dietary CP content resulted adequate (145 g/kg DM on average) to sustain the productions and pointed out the possibility to maintain relatively low CP levels in dairy rations when the diets are well managed (Formigoni and Piva 1996) and diet is well balanced in terms of energy (Formigoni and Biagi 2007). The experimental diets were balanced thanks to an adjustment of the corn flakes and grass hay proportions among the treatments, thus no substantial changes were observed in chemical composition and theoretical supply of ME and MP. TMRs were lower in particle size than usually recommended (Heinrichs 2013), but in accordance with previous trial (Fustini et al. 2017, Cavallini et al. 2018) and similar to those observed in several commercial farms of PR area ). This level of dietary particle size minimised the risk of cows sorting . Feed and water intake were similar to those observed in previous studies conducted under the same experimental conditions (Cavallini et al. 2018).
In all diets, rumination time was lower than the threshold value (390 min/d.) proposed by other authors (Zebeli et al. 2010). In particular, 00-rapeseed diets led to lower rumination times, this result could be explained by the lower aNDFom and peNDF content of these rations (Table 1). No differences were observed on DMI of rapeseed diets, confirming that the inclusion of this feed did not compromise the diet palatability. This result is in line with those obtained by Broderick et al. (2015), where intakes and yield were increased by feeding canola meal.
All the treatments showed an average r-pH lower than 6, which could be explained by the low forage inclusion in the diets, especially in treatment HR (33.1% DM), in which also the starch level was the highest (26.5% DM). Consistently with the higher daily r-pH, pH dynamics, minutes and AUC were lower in MR compared to other treatments. Since experimental diets were balanced for theoretical metabolisable energy and protein content, dietary starch and hay content differed among treatments, due to the higher fibre amount of 00-rapeseed meal. This condition could be responsible for the differences in rumination time and r-pH values, being the lowest in the HR diet (Table 1).
We observed an increase of milk production, especially in MR diet. Similar results are presented in other trials, in which authors related the higher milk production to a more balanced amino acid profile when canola was included in the diets in place of soy (Huhtanen et al. 2011;Martineau et al. 2013;Paula et al. 2020). For the same reason, according to these authors, it could be possible to correlate canola meal inclusion and increased content of milk protein and casein (Huhtanen et al. 2011;Martineau et al. 2019), as reported in the current study, where 00-rapeseed was used. It should be noted that in our experiment the lowest urea values were observed in the HR compared to S milk, suggesting a major nitrogen efficiency due to a better aminoacid profile and/or to a different level of starch in the diet. No differences were also observed in milk acidity expressed in SH 50 degrees and pH, suggesting that milk obtained from treated animals could be considered absolutely suitable for cheese making.
Analysis of goitrin content pointed out detectable values in milk produced by cows fed with 00-rapeseed meal. This fact is interesting since it could be considered as a marker to identify milk produced by cows fed with 00-rapeseed meal. The importance of this detection is even higher when the use of 00-rapeseed meal is prohibited, like in PDO production (Ministero Politiche Agricole Ambientale e Forestali (MIPAAF) 2016). Additionally, detected levels of goitrin in milk were lower than those considered potentially harmful for human health as set by EFSA: Scientific Panel on Contaminants in the Food Chain (2008). In this report, thresholds of 0.7 mmol/g of goitrin and 0.18 mmol/g of thiocyanate were considered safe in milk used for human consumption. Results of fibre digestibility appeared lower than values obtained in other experiments in similar conditions (Fustini et al. 2017) and in other trials (Cotanch et al. 2014). Faecal uNDFom content was lower in 00-rapeseed diets. The decrease of the observed digestibility in 00-rapeseed-based diets could be explained considering the higher passage rate of concentrates, such as rapeseed meal, instead of dry forages and the higher fibre content of rapeseed meal instead of soybean meal (Krizsan et al. 2010).

Conclusions
The results obtained in this study confirm that 00rapeseed meal could be used in PR type diets without negatively influence palatability and feed intake. Replacing soybean with medium levels of 00-rapeseed meal (8.5% DM) led to an improved milk production and milk protein content, with similar or higher results compared to PR type TMR with soybean meal. At this level of inclusion, we can also confirm the safe use of rapeseed meal on rumen health considering the more stable ruminal pH and the lower time spent below subacidosis threshold. We confirm the possibility to detect 00-rapeseed meal used in diets by milk goitrin analysis as a process control check in PDO cheese chain production. In conclusion, this study showed that 00-rapeseed meal could be used in diets without negative effects on performances and milk quality in PR type rations.