Different protein sources in concentrate feed for dairy cows affect cheese-making properties and yield

Soybean meal (SBM) is a commonly used protein source in feed. Yeast microbial protein could be used as a substitute for SBM, but its effect on cheese-making properties and yield is not known. Norwegian Red dairy cows (n = 48) in early or mid lactation were divided in 3 groups and fed a ration consisting of grass silage and concentrate, where the concentrates were barley based but with different additional protein sources. These were: completely barley based with no additional protein source (BAR), additional protein from SBM, or additional protein from yeast ( Cyberlindnera jadinii ; YEA). The SBM and YEA concentrates had a higher protein content than the barley concentrate. Four batches of cheese were made from pooled milk from each of the 3 groups of dairy cows. Milk samples were collected 5 times during the experiment. Milk from cows fed BAR concentrate showed inferior cheese-making properties (lower casein content, longer renneting time, lower content of phosphorus, and lower cheese yield) compared with SBM and YEA concentrates. Overall, SBM or YEA bulk milk had similar cheese-making properties, but when investigating individual milk samples, YEA milk showed better coagulation properties.


INTRODUCTION
Cheese-making efficiency is highly influenced by milk composition, which again is affected by the source and type of feed.The main indicators for cheese-making efficiency are renneting time, cheese yield, and loss of fat and protein in the whey.Curd structure, curd firmness, cheese yield, and renneting time are all directly related to the casein content of milk (Jenkins and McGuire, 2006;Jõudu et al., 2008).
Manipulating milk composition by adjusting the dairy cow diet has been of interest for many years, and already in the early 1980s it was clear that dietary control of milk composition had opportunities, but also restrictions (Jenkins and McGuire, 2006).Protein content is more responsive to diet than lactose, but less responsive than fat.A review by Jenkins and McGuire (2006) stated that the transfer efficiency of dietary protein to milk protein is only 25-30%, which explains the inability of the diet to markedly increase milk protein content.Because fat is the easiest milk component to manipulate, and considering human health issues connected to saturated fat, most research on the dairy cow diet has been concerned with fat content and fatty acid composition.
Due to sustainability issues and also to increase food security, the feed industry needs to develop novel nonfood protein sources.Countries located above ~55° north have limited areas of cultivated land and a challenging climate, and this has led to the need to import protein-rich feed ingredients.However, our research has shown that it is possible to use new bio-refining technology to make protein-rich yeast biomass from cellulose (Lapeña et al., 2020).If this technology can be upscaled to substitute or partially substitute, for example, soy in feed (Kidane et al., 2022) while maintaining the quality of cheese and other dairy products (Olsen et al., 2021), this can both reduce the climate footprint of animal feeds and increase food security.Soryal et al. (2004) tested the effect of pasture feeding in combination with different levels of concentrate feeds (0.66, 0.33, or 0 kg/d per 1.5 kg of milk) on Domiati cheese from goat milk.They found that milk from goats fed a high concentrate level (0.66 kg/d) during pasture feeding gave a higher yield of Domiati cheese.They attributed this to a higher fat, protein, and TS content in milk from goats given concentrate feed compared with milk from goats kept on pasture without concentrate feed or under a confined feeding system with hay and concentrate.Testroet et al. (2018) compared 2 diets given to midlactation Holstein cows.Diet 1 contained 13.5% of DM from soybean meal and diet 2 contained 19.5% of DM from reduced-fat dried distillers' grain.No differences were found between the diets regarding the suitability of milk for cheese-making (Baby Swiss cheese).Ferreira et al. (2017) studied the effect of replacing soybean meal and ground corn with licuri cake (a biodiesel by-product) at different concentrations (0, 200, 400, and 600 g/kg of DM).They found a linear increase in milk fat content when ground corn and soybean meal were replaced with licuri cake, which led to a higher fat content in Minas frescal cheese.No differences were found between the feeds regarding cheese yield, protein, lactose, TS, and SNF contents of either milk or cheese.
A lower milk casein content, but better coagulation properties were observed following pasture feeding during spring and summer in Poland compared with feeding silage or hay during autumn and winter (Teter et al., 2020).Kälber et al. (2013) found that feeding buckwheat silage to dairy cows resulted in shorter coagulation time and increased curd firmness compared with feeding chicory or ryegrass.However, the milk protein content did not differ between the treatments and casein was not analyzed.
The use of yeast as a protein source in rations to dairy cows and the subsequent effect on feed efficiency, milk yield, and metabolic status of the cow has been studied by several authors (Sabbia et al., 2012;Neal et al., 2014;Manthey et al., 2016;Kidane et al., 2022).Their results showed no clear differences in milk composition due to the different feed treatments.However, these studies have only to a minor extent focused on how the yeast influences milk quality more extensively than by only measuring the crude milk composition.None of these studies analyzed milk casein content, which is an important parameter for the dairy industry with regard to cheese-making.
The present work is based on results from the same feeding experiment as described by Kidane et al. (2022) and Olsen et al. (2021).These papers examined the effects of feeding different protein sources (barley, soybean meal, and yeast) in concentrate feed for dairy cows, hypothesizing that Cyberlindnera jadinii yeast protein can replace soybean meal or barley in early-to mid-lactation Norwegian Red (NR) dairy cow diets without adverse effects on milk yield, milk composition, and cheese quality.The results of Kidane et al. (2022) indicated that yeast could be used as a protein source for NR dairy cows without a negative effect on milk yield, and Olsen et al. (2021) found that all 3 protein sources resulted in cheeses of good quality.In addition to the quality of cheese, the cheese-making efficiency is highly important for the cheese maker.Therefore, in the present study, the performance of the milk during the cheese-making process was studied.
The main objective of this study was, therefore, to evaluate the effect of total substitution of soybean meal in concentrate feeds by C. jadinii yeast protein in grass silage-based rations of early-to mid-lactation NR cows on milk coagulation properties, cheese-making, and cheese yield.Furthermore, as barley can be produced in Norway and is the most used concentrate feed ingredient, a diet with barley replacing both yeast protein and soybean meal in the concentrate feed was compared with those 2 other protein sources.

Experimental Setup, Animal, and Feeding
The feeding experiment was performed at the Animal Production and Experimental Unit at the Norwegian University of Life Sciences (NMBU, Ås, Norway) with all animal procedures approved by the national animal research authority of the Norwegian Food Safety Authority (FOTS ID 18038).
The feeding experiment is described in detail by Kidane et al. (2022) and lasted for 10 wk comprising 2 wk of adaptation and 8 wk of experimental diet.In short, 48 early-to mid-lactation NR dairy cows were allocated into 3 treatment groups with 16 replicates per treatment based on parity, milk yield at start of the experiment (measured in the milking robot), DIM and milk protein genetic variants.An overview over the milk protein genetic variants is given in Olsen et al. (2021).The cows were fed a ration consisting of grass silage and concentrate.The concentrates were barley based, but with different additional protein sources.These were no additional protein source and completely barley based (BAR), additional protein from soybean meal (SBM), or additional protein from yeast (C.jadinii; YEA).The composition of concentrate feed and grass silage is shown in Table 1.
During the adaptation period (2 wk), cows in all 3 treatment groups were fed the concentrate feed with SBM.During the experimental period (the following 8 wk) the cows in each treatment group were given either the same SBM concentrate feed as in the adaptation period, or BAR or YEA concentrate feed.
The chemical composition of the basal diet (grass silage) and concentrate feed together with basic cow information is provided by Olsen et al. (2021).The experimental concentrate feeds were prepared in such a way that the SBM and YEA were iso-nitrogenous with a somewhat higher protein content compared with the BAR concentrate (161, 157, and 134 g of protein/kg of DM, respectively) and all 3 feeds were approximately iso-energetic.

Milk Sampling
Individual milk samples were collected in wk 2, 4, 6, 7, and 10.The samples (50 mL) were collected automatically at each milking in a Delaval Classic milking robot system (DeLaval International AB).The cows had access to the milking robot every sixth hour and on average, 5 samples were obtained from each cow during a 48-h period and kept cold until further preparation.On arrival at the analytic laboratory, all samples from the same cow were mixed and these pooled samples were used for further analysis.

Cheese-Making
Gouda-type cheeses were made during wk 8 and 9 of the feeding experiment, in the University dairy pilot plant.Milk from the specific cows of each group (BAR, SBM, and YEA) was collected in a separate milk tank over 2 d.
It was only possible to sample milk separately from one experimental group at a time; therefore, cheese was produced over 6 production days, 2 d for each type of milk.At each production day, 2 vats of cheese were made, and these were considered as replicates.This resulted in 4 cheese vats produced from the same type of milk (BAR, SBM, or YEA) and in total 12 vats of cheese were made.Cheeses were made as described by Olsen et al. (2021).

Analysis of Individual Milk Samples and Cheese Milk
Both the individual milk samples and the fatstandardized cheese milk before cheese-making were analyzed for gross composition.Samples for analysis of gross composition were preserved with bronopol (2-bromo-2-nitropane-1,3 diol, Broad-Spectrum Microtabs II, Advanced Instruments) and were analyzed by TINE S/A for fat, protein, lactose, and SCC using a DairySpec Combi (Bentley Instruments Inc.).The cheese milk was analyzed for fat, protein, casein, and lactose using MilkoScan FT1 (Foss Electric A/S) in the University dairy pilot plant.The pH was measured using a PHM 92 Lab pH meter (Radiometer).
The mineral content of individual milk samples and cheese milk was analyzed according to the method described by Jørgensen et al. (2015) using SRM 1549A (National Institute of Standards & Technology) as reference material.
Rennet coagulation properties (i.e., rennet clotting time (RCT), time until 20 mm width between the pendulums is achieved in the Lattodinamografo (K20), and firmness after 30 min (A30) of the individual milk samples were analyzed using Lattodinamografo (Foss-Italia SpA) according to the method described by Inglingstad et al. (2014).This analysis was made on the same day as the samples arrived at the laboratory.The K20 results were transformed to binary data (0 = samples that did not attain firmness of 20 mm and 1 = samples that did attain firmness of 20 mm).

Cheese Analysis and Calculations
Renneting time during production of cheese was defined as the time from adding the rennet until cutting the coagulum.Curd firmness and time to cut was evaluated by an experienced cheesemaker.The amount of cheese milk (L) and weight of cheese (kg) after brining was measured.
Twenty-four hours after the start of cheese-making, cheese was analyzed for DM (IDF standard 50C; IDF, 1995) and pH (using a PHM 92 Lab pH meter; Radiometer).
Predicted cheese yield (PY) was calculated by using the Van Slyke formula (Fox et al., 2017a) using the fat and casein content and a constant for loss of fines and other solids included in the cheese in Equation [1].Actual yield (Ya) in Equation [2] and moisture-adjusted cheese yield (MACY) in Equation [3] were calculated according to Banks (2007).Yield efficiency (YE) using Ya and PY was calculated in Equation [4] according to Fox et al. (2017a).
where F is the fat in milk (%); C is the casein in milk (%); W is the desired water content in cheese*; 0.1 is  where * is the mean moisture content of all 24-h cheeses (n = 12); and

Statistical Analysis
Data for milk composition and coagulation properties (RCT and A30) were analyzed using the mixed procedure of SAS (SAS Enterprise Guide 7.1, SAS).Somatic cell counts were log 10 transformed before analysis because of non-normal distribution.The model included the fixed effects of concentrate feed (BAR, SBM, or YEA), weeks (4, 6, 7, 10), parity (primiparous or multiparous), a covariate value (the respective variables from each cow from the end of the adaptation period (first 2 wk), and the interaction between concentrate feed and week, as well as the repeated effect of week and random effect of cow nested within concentrate feed and parity.Tukey-Kramer was used to test for pairwise differences between least squares means.Data are presented as least squares means, with statistical significance declared at P < 0.05.
The effect of concentrate feed on K20 was tested using the logistic procedure in SAS Enterprise Guide 7.1.As many of the samples did not attain K20, data were converted to a binary format (samples that did attain K20 and samples that did not attain K20), as previously described.The model used included the fixed effects of concentrate feed (BAR, SBM, or YEA), week (4, 6, 7, 10), parity (primiparous or multiparous), covariate (wk 2), and the interaction between concentrate feed and week.
Significant effects (P ≤ 0.05) of experimental factors on the cheese milk, cheeses, and production parameters were found using the mixed procedure of SAS Enterprise Guide 7.1.The experimental factors used were concentrate feed as the main factor (n = 3) and cheesemaking day (n = 6) as a random factor.Least Square Post Hoc (Tukey) was used to test differences between means (all pairwise differences).

RESULTS
All data used in this paper can be found under NMBU Open Research Data (Olsen, 2022).

Individual Milk Samples
Gross composition and coagulation properties (RCT and A30) of individual milk samples are shown in Table 2.The concentrate feed did not affect the gross composition of milk or its content of somatic cells, but milk protein content increased toward the end of the experiment.YEA milk had a significantly higher content of phosphorus than to the BAR milk, and BAR milk contained a significantly more selenium and iodine compared with YEA and SBM milk.
Most of the milk samples demonstrated poor coagulation properties, showing a long RCT and low A30.The RCT of the milk was not influenced by the types of feed, although SBM milk had a borderline significantly shorter RCT compared with BAR milk (P = 0.051; 17.9 vs. 19.8min respectively).The A30 was considered weak as most of the milk gels had a firmness well below 20 mm.BAR milk obtained the least firm gel with a mean A30 of 12.33 ± 1.04 mm, whereas YEA milk obtained the highest A30 with a mean of 15.29 ± 1.04 mm.Out of a total of 236 analyzed samples, only 77 samples (33%) attained K20.
There was a greater probability that the milk gel would attain a firmness of at least 20 mm if the cows were fed YEA concentrate feed compared with both SBM and BAR milk (Figure 1).If they were fed SBM concentrate feed, it was more likely that the milk gel would attain K20 compared with feeding BAR concentrate feed.Although no treatment*week interaction was found, it appears as the proportion of samples attaining K20 increased gradually (except for wk 10) for the YEA treatment (44, 50, 63, and 44% in wk 4, 6, 7, and 10 respectively).Milk from primiparous cows was less likely to attain K20 than milk from multiparous cows (results not shown).
In total, 13 milk samples from 9 cows (distribution: BAR = 2, SBM = 5, and YEA = 2) were noncoagulating, i.e., they did not form a curd within 30 min.The SBM group had a higher proportion of noncoagulating samples, but this group included 2 cows that gave milk that did not coagulate at 3 out of the 5 samplings.This

Cheese Milk and Cheese-Making
The composition of fat-standardized cheese milk and of cheese the day after production is shown in Table 3, while different yield parameters are shown in Table 4. Gross composition and pH of individual milk samples were not influenced by the different concentrate feeds used.However, YEA and SBM milk had a significantly (P = 0.0005) higher content of casein compared with BAR milk, and this resulted in >0.44 kg more casein in the SBM and YEA cheese vats compared with the BAR cheese vats.
BAR cheese milk differed from SBM and YEA milk with regard to the content of several minerals (Table 3).The BAR cheese milk had a significantly lower concentration of phosphorus than YEA and SBM milk.In addition, the BAR cheese milk had a higher concentration of sodium compared with SBM cheese milk and a higher concentration of iodine compared with YEA and SBM cheese milk.
Due to differences in the casein content, the rennet to casein ratio differed between the experimental groups, as the rennet was added according to volume of milk and not according to kg of casein.Significantly more rennet in relation to casein (mL/kg of casein) was added to the BAR milk vats compared with YEA and SBM milk vats.Despite this, the BAR milk had a significantly longer renneting time compared with the YEA milk.Due to the higher content of casein in YEA and SBM milk, the PY from cheese vats from these groups was significantly higher than cheese made from BAR milk.Both the Ya and MACY confirmed the results calculated for the PY.There were no significant differences in YE due to high standard deviations, but a tendency indicated that it could be more efficient to make cheeses from SBM cheese milk compared with BAR cheese milk.

DISCUSSION
This study showed that feeding YEA concentrate feed gave a higher probability that the individual milk samples attained good coagulation properties, and in the cheese vat, the YEA cheese milk was superior to the BAR cheese milk.This can probably be attributed to the higher casein content in the YEA milk compared with the BAR cheese milk.Higher casein content is cor- Different superscript letters represent significant differences between the different concentrate feeds at P < 0.05 for the diet variable. 1 Values are presented as LSM (n = 240; on some occasions the amount of milk was not sufficient for all analyses).related with better coagulation properties and has also been shown to be more important than total protein content (Auldist et al., 2002;Jõudu et al., 2008).
The coagulation properties of the individual milk samples were in general poor, and may be due to factors such as late lactation, high SCC, casein content, and polymorphism of the milk proteins, among others (Fox et al., 2017a,b).Those factors of relevance for this experiment are discussed further.None of the cows were in late lactation during this experiment and the SCC were low, therefore the whey protein: casein ratio in the individual milk samples was most probably fairly constant (not analyzed).The noncoagulating samples came from all the diet-groups, suggesting it is unlikely that the feed type caused the difference.When dealing with coagulation experiments the genetic variants of the milk proteins for the cows used in the experiment should be balanced, because these genetic variants affect cheese-making properties such as coagulation properties and cheese yield (Ng- Kwai-Hang, 2006;Gustavsson et al., 2014;Ketto et al., 2017).When grouping the cows, it was decided to use those cows with genotypes having the highest frequency at Animal Production and Experimental Unit (Olsen et al., 2021), and these cows unfortunately had a high prevalence of genetic protein variants related to inferior milk coagulation properties, similar to κ-CN AA and β-CN A2A2 (Ketto et al., 2017).The occurrence of these variants was high within the experimental herd, in total 35 out of 48 cows (73%) had the AA-variant of κ-CN in this experiment, which Proportion of milk samples (%) that attained a firmness of at least 20 mm (K20: time in minutes taken for the width of the curves to increase to 20 mm) during the 30-min run using a Lattodinamografo for each diet group (green bar = barley; gray bar = soybean meal; and yellow bar = yeast) in the adaptation and experimental periods.(Results in the experimental period are the calculated mean of wk 4, 6, 7, and 10).Different letters (a, b) indicate significant differences (P < 0.05) between the concentrate feeds in the experimental period.In the adaptation period, all groups were fed soybean meal; the color mixed with gray shows which feeding each group was further allocated to in the experimental period.
in NR is associated with poorer coagulation properties than the BB variant of κ-CN (Ketto et al., 2017).All of the noncoagulating milk samples had the AA-variant of κ-CN.This may explain the poor coagulation properties of the individual milk samples in this experiment.
During cheese-making, rennet was added at a concentration of 25 mL/100 L of milk without any adjustment for casein concentration, as is normal practice in Norway.However, because the milk in this experiment had greater differences in casein concentration than the normal variation, an adjustment of the rennet addition should preferably have been done.Previously, the average casein content in milk from cows at Animal Production and Experimental Unit has been 2.65% and this has been used to standardize the rennet: casein ratio.This gives 9.38 mL of rennet used (Chy-Max Plus, Chr.Hansen, Hørsholm, Denmark) per kg casein, which corresponds to using 25 mL of rennet used per 100 L of milk.When comparing the actual amount of rennet added in this experiment with the calculated amount of  rennet needed if 9.38 mL of rennet/kg of casein should be used, the correct amount of rennet was added to BAR cheese milk, but less rennet than optimal was added to YEA and SBM cheese milk.Probably, if the amount of rennet had been adjusted to the casein content, an even greater difference in renneting time would have been found between BAR milk and the 2 others (SBM and YEA).Moreover, as the casein content was not analyzed in the individual milk samples, and there were probably differences in casein concentration in those samples as well, then the same rennet: casein situation would also apply for the individual samples.
It is well known that casein content and composition of bovine milk influences the cheese-making efficiency and is therefore of great importance for profitability.An increase of the casein content in milk does not only normally result in better coagulation properties, but also in a higher cheese yield (Bobe et al., 1999;Banks, 2007;Fox et al., 2017a).This was also observed in this trial and may be attributed to the casein content and also mineral content as discussed further.The milk salts, especially calcium and phosphate, play a vital role in the structure of casein micelles and affect not only milk coagulation but also other aspects of cheesemaking such as buffer capacity and cheese texture (Lucey and Fox, 1993).Stocco et al. (2021) studied the effect of minerals on milk coagulation properties and yield of model cheeses.They found that phosphorus was associated with good cheese-making traits and an increased cheese yield (curd solids), and that a higher concentration of sodium in milk was associated with lower protein recovery in model cheese.Therefore, the higher content of phosphorus and casein in SBM and YEA cheese milks compared with BAR milk may have affected the coagulation and cheese-making properties and contributed to the higher cheese yield.In addition, the higher sodium content of BAR cheese milk compared with SBM cheese milk might have contributed to an undesirable longer renneting time and a lower protein recovery, thereby resulting in a lower Ya and MACY compared with the other groups.Although no differences in YE were found between the groups due to the high standard deviation, BAR milk showed a tendency to be less efficient for cheese-making.This may be due to the higher sodium content.Several other authors have also found a link between rennet coagulation properties and the mineral content of milk, both Malacarne et al. (2014) and Jensen et al. (2012) showed that milk with good coagulation properties had a higher content of calcium, phosphorus and magnesium, compared with poorly coagulating and noncoagulating milk.The mineral concentration in milk is affected by the mineral composition in the feed and by soil conditions where the feed is grown (Alothman et al., 2019).
This study shows that changing the protein content or protein source in feed to dairy cows can have several side-effects additional to changing the gross composition of milk, and that these changes can influence milk properties during the processing of different dairy products.
During feeding trials with dairy cows, milk protein, fat, lactose and milk yield are usually measured, but casein is not usually analyzed.Normally, feeding trials do not include a cheese-making experiment, and we have managed to identify only a few feeding studies where cheese has also been made.
Our results indicates that if casein content was measured in the studies of for example Sabbia et al. (2012), Manthey et al. (2016), andNeal et al. (2014), where alternative protein sources for dairy cows were investigated, differences could actually have been obtained.
In this study we used grass silage which is the commonly used silage type in northern countries.Further work is needed to see if similar results would be obtained by using other types of silage, such as in example maize silage, used in regions suitable for such crops.In addition, an interesting approach would be further testing of different protein sources and protein levels in concentrate feeds also in relation to the cheese-making efficiency and cheese quality.

CONCLUSIONS
With increasing global population and climate change, it is necessary to find alternative nonfood protein sources for farm animal feed and to allocate food-grade protein to human consumption.Yeast production, using cellulose as raw material, is a possible alternative in countries with limited cultivated land.By using such resources, more countries could have a self-sufficient supply of feed ingredients and therefore limit long distance transportation due to export/ import.This experiment shows that it is possible to substitute or partly substitute soybean meal with yeast as a protein source in concentrate feed to dairy cows, without negative effects on cheese-making properties.However, when comparing yeast or soy with barley, the cheese-making properties of the milk were clearly different, Therefore, the protein source and protein content of the feed are of importance when addressing cheese-making properties and cheese yield.This should be considered when planning dairy cow rations, as it influences the economy of the entire dairy chain.cheese manufacture; Helene Tynes Farstad at the Animal Production and Experimental Unit (NMBU) for cooperation with managing the milk collection logistics; TINE S/A (Oslo, Norway) for analyzing crude composition of milk and sensory analysis; and May Helene Aaberg, Kari Olsen, and Ahmed Abdelghani (NMBU) for technical assistance during sampling and analysis.This study was funded by Foods of Norway, Centre for Research-Based Innovation (Research Council of Norway, RCN, Oslo; grant no.237841/030).This research used the Food Pilot Plant facilities at Norwegian University of Life Sciences (Ås, Norway) that received a grant from RCN (grant 296083).The authors have not stated any conflicts of interest.
Olsen et al.: EFFECT OF CONCENTRATE PROTEIN ON CHEESE-MAKING indicates more of an individual cow problem rather than a feed problem.
Figure1.Proportion of milk samples (%) that attained a firmness of at least 20 mm (K20: time in minutes taken for the width of the curves to increase to 20 mm) during the 30-min run using a Lattodinamografo for each diet group (green bar = barley; gray bar = soybean meal; and yellow bar = yeast) in the adaptation and experimental periods.(Results in the experimental period are the calculated mean of wk 4, 6, 7, and 10).Different letters (a, b) indicate significant differences (P < 0.05) between the concentrate feeds in the experimental period.In the adaptation period, all groups were fed soybean meal; the color mixed with gray shows which feeding each group was further allocated to in the experimental period.
Olsen et al.: EFFECT OF CONCENTRATE PROTEIN ON CHEESE-MAKING

Table 1 .
Olsen et al.: EFFECT OF CONCENTRATE PROTEIN ON CHEESE-MAKING Composition of concentrate feeds [barley (BAR), soybean meal (SBM), and yeast (YEA)] and grass silage 1

Table 2 .
Olsen et al.: EFFECT OF CONCENTRATE PROTEIN ON CHEESE-MAKING Milk composition and coagulation properties of individual milk samples from dairy cows fed concentrate feed based on 3 different protein sources [barley (BAR), soybean meal (SBM), and yeast (YEA)] 1

Table 3 .
Olsen et al.:EFFECT OF CONCENTRATE PROTEIN ON CHEESE-MAKING Gross composition of cheese milk and renneting properties of the produced cheeses within each diet group [barley (BAR), soybean meal (SBM), and yeast (YEA)] 1

Table 4 .
Different cheese yield parameters from the production of cheeses within each diet group [barley (BAR), soybean meal (SBM), and yeast (YEA)] 1 1Values are presented as mean of 4 replicates ± SD.