Effects of Ruminal Crabtree-negative Yeast Ensiled Rice Straw on Feed Intake, Rumen Fermentation, and Performance in Tropical Crossbred Lactating Holstein Cows

This research aimed to determine the effects of ruminal Crabtree-negative yeast ensiled rice straw (RS) on feed intake, ruminal fermentation, milk production, and milk composition in tropical crossbred lactating Holstein cows. This study used 6 multiparous crossbreds between Holstein Frisian × Zebu dairy cows in their mid-lactation period (165.5 ± 44.0 of day-in-milk) with an initial body weight of 363.9 ± 55.80 kg (average milk yield 8.58 kg/d). Dairy cows were randomly allocated to three ensiled RS with various yeast species including S. cerevisiae, P. kudriavzevii KKU20, and C. tropicalis KKU20 according to a 3 × 3 replicated Latin square design. The ruminal yeasts were obtained by isolating, screening, and identifying the rumen of crossbred Thai-Holstein Friesian dairy cattle. The yeast species did not change the RS intake, concentrate diet, and total intake (P > 0.05). Crabtree-negative yeast (P. kudriavzevii and C. tropicalis) increased the apparent digestibility of dry matter by about 6.9% when compare with Crabtree-positive yeast (S. cerevisiae). Rumen pH and ammonia-nitrogen concentration were not changed among yeast species (P > 0.05). The bacterial populations at both 0 hours and 4 hours after feeding and the mean value were highest (P < 0.05) with ensiled RS with C. tropicalis KKU20. Ensiled RS with P. kudriavzevii KKU20 and C. tropicalis KKU20 were signicantly increased with a total volatile fatty acids (VFAs) at 0 and 4 hours after feeding (P < 0.05) when compared with S. cerevisiae, whereas yeasts ensiled RS had no effect on the VFAs’ prole (P > 0.05). The yeast strains’ effects were not observed (P > 0.05) on actual milk yields. The treatments did not alter the milk composition (P > 0.05); except for when the protein in the milk was highest in the C. tropicalis KKU20 acid was produced when ensiled RS with C. tropicalis KKU20 and P. kudriavzevii KKU20 at 14 day. NH 3 -N concentration in ensiled RS within the range of 1.80 to 2.00 g/kg DM indicated the normal standards for estimating silage. These results are similar to those of Li et al. 41 , who collected information on various types of RS parameters and concluded that RS silage has a NH 3 -N concentration of approximately 1.61 to 2.36 g/kg DM. Other parameter such as C 2 show great value for preserved silage within range 20 to 25 g/kg DM 39 . Moreover, after the fermentation process, the moisture content should range from 650 to 750 g/kg to be optimum 16 , which in our study showed an average of 722.1 g/kg. Therefore, our study proposes that the nutrients in ensiled RS are still well preserved.


Introduction
For several years, yeast has been the model organism used for enhancing animal e ciency and is the traditional practice for ruminant feed additives 1 . In many studies, using S. cerevisiae fermented with an agricultural by-product or rice straw (RS) has been shown to enhance their nutritional value, silage quality, Table 1 Dietary ingredients and chemical composition of different yeast species in ensiled rice straw and concentrate diet. Premix = Vitamins and minerals; A: 10,000,000 IU; Vitamin E: 70,000 IU; Vitamin D: 1,600,000 IU; Fe: 50 g; Zn: 40 g; Mn: 40 g; Co: 0.1 g; Cu: 10 g; Se: 0.1 g; I: 0.5 g. Hemicellulose = NDF-ADF Cellulose = ADF-lignin Metabolizable energy calculated according to the equation described by Robinson et al. (2004).

Item
Ensiled rice straw (g/kg fresh matter) Concentrate diet (g/kg DM) S. cerevisiae P. kudriavzevii  Feed intake, nutrient intake, and nutrient apparent digestibility The impacts of different yeast species ensiled RS on the effectiveness of feed utilization in dairy cattle is illustrated in Table 2.. The yeast species did not change the RS intake, concentrate diet, and total intake (P > 0.05). Total intake ranged from 111.7 to 121.1 g/kg BW 0.75 . OM and CP intake were 8.9 to 9.6 kg/day and 1.3 to 1.4 kg/day, respectively, which was not altered among treatments (P > 0.05). Crabtree-negative yeast (P. kudriavzevii KKU20 and C. tropicalis KKU20) increased the apparent digestibility of DM by about 6.9% when compare with Crabtree-positive yeast (S. cerevisiae). However, the data achieved in this study showed that apparent digestibility of OM (OMD), CP (CPD), NDF (NDFD), and ADF (ADFD) were not altered among yeast species and ranged from 762.1 to 791.5, 752.0 to 791.5, 601.5 to 641.3, and 492.8 to 525.4 g/kg, respectively. Furthermore, the total digestible nutrients were the same among yeast species and ranged from 734.8 to 767.7 g/kg (P > 0.05). Effect on rumen pH, NH 3 -N, blood metabolites and microbial communities The bacterial populations at both 0 hours and 4 hours after feeding and the mean value were highest (P < 0.05) with ensiled RS with C. tropicalis KKU20 by 9.9, 12.5 and 11.2 Log10 cell/ml, respectively. However, the fungal zoospore and protozoa populations were not affected by any treatments (P > 0.05). Table 3 Effect of different yeast species in rice straw ensiled on ruminal pH, NH 3 -N concentration, blood urea-nitrogen concentration, and microbial communities in crossbred lactating dairy cows. a, b Means in the same row with different superscript letters differ (P < 0.01, P < 0.05). S. cerevisiae = Saccharomyce cerevisiae, P. kudriavzevii = Pichia kudriavzevii, C. tropicalis = Candida tropicalis. Effect on ruminal volatile fatty acid The total VFA, acetic acid (C 2 ), propionic acid (C 3 ), butyric acid (C 4 ) proportions, and acetic acid to propionic acid ratio are illustrated in Table 4. Ensiled RS with P. kudriavzevii KKU20 and C. tropicalis KKU20 were signi cantly increased with a total VFAs at 0 hours (5.15 and 5.06%, respectively), and 4 hours (5.07 and 8.83%, respectively) after feeding (P < 0.05) when compared with S. cerevisiae, whereas yeasts ensiled RS had no effect on the VFAs' pro le (P > 0.05). The mean value of C 2 , C 3, and C 4 were 67.4, 22.3, and 10.3 mol/100 mol, respectively. Effect on milk production, milk composition, and feed e ciency The effects of ensiled RS with various yeast species on milk production, composition of milk, and feed e ciency in dairy cows are shown in Table 5. The yeast strains' effects were not observed (P > 0.05) on actual milk yields (8.5 to 8.8 kg/h/d), 4.0% FCM (7.6 to 8.3 kg/h/d), and ECM (7.7 to 8.3 kg/h/d). The treatments did not alter the milk composition (P > 0.05); except for when the protein in the milk was highest in the C. tropicalis KKU20 fed group at 35.6 g/kg (P < 0.01). Feed e ciency did not changed for any diets (P > 0.05). Table 5 Effect of different yeast species in rice straw ensiled on milk yield, milk composition, feed e ciency and economic e ciency in crossbred lactating dairy cows. a, b Means in the same row with different superscript letters differ (P < 0.01, P < 0.05). S. cerevisiae = Saccharomyce cerevisiae, P. kudriavzevii = Pichia kudriavzevii, C. supporting dairy cows' performance. Furthermore, ensiled RS with P. kudriavzevii KKU20 and C. tropicalis KKU20 (Crabtree-negative yeast) were established as having a low ber content when compared with adding S. cerevisiae (Crabtree-positive yeast). The low ber content can be clari ed by the yeast's ability to release cellulase enzymes and digest ber during the fermentation process. Suntara and Cherdthong 8 con rmed that C. tropicalis KKU20 and P. kudriavzevii KKU20 were more capable to releasing cellulase enzymes than S. cerevisiae by about 0.7 to 6.8 times, respectively. Moreover, the experiment on in vitro gas production of ensiled RS at 14 days with the P. kudriavzevii KKU20 could decrease the NDF content by about 6.7% when compared with S. cerevisiae 15 . Ilmén et al. 37 discovered yeast isolated from a plant named C. konsanensis species could excrete cellulase enzymes and digests ber, and it is a new yeast strain that had not been reported previously. Similar with our study, C. tropicalis KKU20 and P. kudriavzevii KKU20 are great potential yeasts to improve feedstuffs and this study is the rst report in ruminant nutrition feed research.
The fermentation quality of ensiled RS with different yeast species indicated that the silage was well preserved. The ensiled RS still maintained appropriate pH, high lactic acid content, and a low NH 3 -N level.
Acceptable silage was de ned by the pH value and the composition of their fermentation products 38 .
The pH is the main indicator for evaluating silage quality and our study showed ensiled RS still has a satisfactory score of about 4.1 to 4.3 39 . In addition, pH is highly related with lactic acid content, which in this study showed a consistent range of about 19.8-22.1 g/kg DM. Lactic acid content in silage should range between 21 to 25 g/kg DM to be considered of high quality, according to Flieg's score 40 ; therefore, it is close to the high quality of silage. In addition, our result showed lactic acid content similar to an earlier study by Suntara et al. 15 who revealed that about 20.53 to 26.14 g/kg DM of lactic acid was produced when ensiled RS with C. tropicalis KKU20 and P. kudriavzevii KKU20 at 14 day. NH 3 -N concentration in ensiled RS within the range of 1.80 to 2.00 g/kg DM indicated the normal standards for estimating silage. These results are similar to those of Li et al. 41 , who collected information on various types of RS parameters and concluded that RS silage has a NH 3 -N concentration of approximately 1.61 to 2.36 g/kg DM. Other parameter such as C 2 show great value for preserved silage within range 20 to 25 g/kg DM 39 . Moreover, after the fermentation process, the moisture content should range from 650 to 750 g/kg to be optimum 16 , which in our study showed an average of 722.1 g/kg. Therefore, our study proposes that the nutrients in ensiled RS are still well preserved.
Crabtree-negative or -positive yeast has no effect on the dry matter intake (DMI However, Aquino et al. 46 reported that the amount of RS that ruminants can consume can be as high as 1.2% BW, which is similar with our result of 0.8-1.0%BW. The intake of OM, EE, NDF, and ADF was similar with previous studies of lactating crossbred dairy cows 47,48 . The CP intake (CPI) in this study was also similar with Wanapat, et al. 43 , which used lactating crossbred dairy cows (50% Holstein Frisian × 50% Thai native cows) and BW around 365.5 kg, and the CPI was about 1.0 to 1.2 kg/d. Typically, the CP found in tropical forage plants is often relatively low 49 . Especially in RS (3%CP) when using a roughage source it can have an effect on the animal's yield adequacy 50 . However, our study showed that ensiled RS with yeast could support protein from yeast to low quality roughage as RS, and the enhanced intake of protein were high enough to meet the requirement of tropical lactation dairy cows.
The dry matter digestibility (DMD) was increased when ensiled RS with Crabtree negative yeast was offered to animals. This strain is outstanding in terms of high proliferation ability and its high yield of cellulase enzymes 15 . The improved digestion may be due to the potential of how rumen micro ora are promoted for better digestibility. Yeast is an important biological responder in the rumen fermentation, live yeast cells improve microorganisms in rumen 51 and stabilizes pH in the rumen 52 . Habeeb 53 stated that yeast could provide rumen with biological stimulants, which is necessary for microorganisms' growth in the rumen. Therefore, yeast contributes to establishing microbiota 54 and is why the digestibility was apparently enhanced. This is consistence with Wang et al. 6 , who found that Crabtree-negative yeast as C. tropicalis could increase digestion in the in vitro technique and that it generated 3.03% more gas production than did S. cerevisiae.
However, Crabtree-negative yeast did not change the apparent digestibility of OM, CP, NDF, and ADF. The digestibility of NDF and ADF are similar among Cabtree-negative and positive yeast (601.50 vs 650.05 g/kg DM and 492.8 vs 518.15 g/kg DM, respectively). Noticeable changes occurred after the silage process was complete, but when the animal intakes the feed, its digestion was not altered. The reason for this is still not clear, but it is possible that yeast does not react directly on RS. Rather, digestion in the rumen occurred by the cooperation of microbes' synergy until the resulting values were not statistically different. This is similar with an experiment by Suntara et al. 15 , who compared the effect of Crabtree-negative and -positive yeast on ensiled RS on the in vitro gas and con rmed that in the rumen, there was no difference among yeast species in the digestibility of NDF and ADF (705.2 vs 703.6 and 464.8 vs 464.4 g/kg DM).
Ensiled RS with the P. kudriavzevii KKU20 and C. tropicalis KKU20 (Crabtree-negative yeast) could increase bacterial populations when compared to S. cerevisiae (Crabtree-positive yeast) by about 4.76%.
The ruminal bacterial populations depend on su cient nutrients or stimulants supply 53 . Yeast is a great supply to stimulate bacteria because it is enriched in essential substances 55 . Previous studies have con rmed that yeast could supply essential amino acids, vitamins, and minerals to increase the ruminal bacteria more than without yeast 2,56 . The key explanation is that under aerobic conditions, Crabtreenegative yeast may proliferate more than Crabtree-positive yeast since the enzyme mechanism functions differently 9,57 . Suntara and Cherdthong 8 found that at 72 h of incubation time, P. kudriavzevii KKU20, C. tropicalis KKU20, and S. cerevisiae had growth by about 10.02, 9.6, and 8.87 Log cells/ml, respectively. The high amount of Crabtree-negative yeast creates a greater supply of essential nutrients to the rumen bacteria 15 , thus the amount of rumen bacteria is increased in response to the Crabtree-negative yeast.
The ensiled RS with Crabtree-negative yeast has more effect on the total VFAs than with Crabtree-positive yeast by about 6.1% at the mean value. The high production of total VFAs in rumen uids is related to the amount of ruminal bacteria 58 . The great bacterial population could enhance carbohydrate digestion and then the animal obtains the greater VFAs 59 . This is similar to Castillo-González et al. 60 , who stated that the expansion of rumen microorganisms could increase the quantity of rumen VFAs. Certainly, a high bacterial population in our experiment was related with the Crabtree-negative yeast's effect. Nonetheless, the direct in uence of the Crabtree-negative yeast on rumen bacterial populations was unclear and this hypothesis required further research to be conducted. However, expanding the Crabtree-negative yeast population (during fermentation process) may be more effective than expanding that of the Crabtreepositive yeast (S. cerevisiae). This suggests that animals have a greater chance of obtaining stimulants for activate rumen bacteria. In agreement with our results, Wang et al. 6 compared the effect between Crabtree-negative yeast (C. tropicalis) and Crabtree-positive yeast (S. cerevisiae) for in vitro gas technique and found that the inclusion of 0.25 × 10 7 of Crabtree-negative yeast could enhanced the total VFAs by 7.7%. Moreover, Suntara et al. 15 reported that Crabtree-negative yeast (P. kudriavzevii KKU20) increased the total VFAs by 2.3% for in vitro gas study more than Crabtree-positive yeast.
The milk yield and milk composition of ensiled RS with Crabtree-negative yeast did not have any impact.
Our study showed that the actual milk yields are about 8.5 to 8.8 kg/h/d, which are slightly lower than previous trials using early to mid-lactation cows (12.6 kg/h/d according to Supapong and Cherdthong 61 ; 11.1 kg/h/d according to Wanapat et al. 43 ). To produce milk, cows must calve and split its lactation cycle into four phases (early, mid, late lactation and dry period) 62 . The milk yield response was greater in the early lactation, and in the mid-lactation period, the milk yield begins to decline from its peak 63 .
Therefore, the lower actual milk yields in this study may be because dairy cows were in mid to late lactation (DIM 165.5 to 186.5). Our study indicated that daily protein yields in milk of the C. tropicalis KKU20 group was highest at 35.6 g/kg and lowest when applied with S. cerevisiae and P. kudriavzevii KKU20 in ensiled RS at 34.5 and 34.1 g/kg, respectively. Milk protein is associated with the feed degradation energy supply as VFAs and microbial protein (MCP) synthesis 64 . High amounts of microorganisms in rumen could affect the MCP synthesis. This will be the supply protein and amino acids (AA) in the small intestine and could enhance the milk protein yields 65 . Our result clearly demonstrated that C. tropicalis KKU20 was unique in the highest bacterial population (11.2 Log10 cell / ml), which is why the increase in milk protein yields occurred. Furthermore, there were no differences in milk proteins between S. cerevisiae and P. kudriavzevii KKU20. This suggests that the in uence of Crabtree-negative yeast may play different roles in terms of milk quality. This thought is support by Intanoo et al. 66 , who compared different yeast strains that were in the same group of Crabtree-negative, and found that P. kudriavzevii KKU20 decreased daily protein yields in milk by 14.9% when compared with Kluyveromyces marxianus in crossbred lactating cows. This yeast species could provide high biomass, which possibly supplies more amino acid sources for milk protein synthesis. This is similar to Wardrop et al. 11 who stated that K. marxianus has an outstanding ability to provide high biomass when compared with other strain. The explanation is limited in regard to P. kudriavzevii's impact on daily protein yields in milk. A few studies have focused on applying non-S. cerevisiae to dairy cows and further research about the in uence of each strain is required.
Based on this study, we conclude that Crabtree-negative yeast-treated RS, especially C. tropicalis KKU20, could enhance the RS's nutrition value through increasing DMD, the ruminal bacterial population, and total VFAs. In addition, C. tropicalis KKU20 could increase the milk protein when compared with other groups. However, there are certain drawbacks associated with the high-producing lactating cows in uenced by C. tropicalis KKU20 treated RS, which requires further investigation.

Methods
The animals participating in this study have been certi ed by the Khon Kaen University Animal Ethics Committee (Record No. IACUC-KKU 38/62), based on the Ethics of Animal Experimentation of the National Research Council of Thailand. In addition, we con rmed that all methods were performed in accordance with the relevant guidelines and regulations.

Animals and experimental design
This study used 6 multiparous crossbreds between Holstein Frisian × Zebu dairy cows in their midlactation period (165.5 ± 44.0 of day-in-milk) with an initial body weight of 363.9 ± 55.80 kg (average milk yield 8.58 kg/d) and a mean age of 5 years. The milk yield reported was slightly higher than the previous studies, which Holstein Frisian × Zebu cow's milk yields were 2,897 kg/year or 8.05 kg/d 14 . Dairy cows were randomly allocated to three ensiled RS with various yeast species including S. cerevisiae, P. kudriavzevii KKU20, and C. tropicalis KKU20 according to a 3 × 3 replicated Latin square design.
Ensiling rice straw with yeast from rumen uid The ruminal yeasts were obtained by isolating, screening, and identifying the rumen of crossbred Thai-Holstein Friesian dairy cattle 8 . The P. kudriavzevii KKU20 and C. tropicalis KKU20 were tested for their high-potential on in vitro study, which has an outstanding bene t for feed digestion and in vitro gas production 15 . The S. cerevisae was obtained from the commercial baker's yeast (Perfect yeast Co., Ltd, Ubon Ratchathani, Thailand). Isolated homogenous yeast suspension from rumen (about 10 6 cells per ml) were multiplied in media solution including 250 g molasses (Khon Kaen Dairy cooperative Co., Ltd., Khon Kaen, Thailand) plus 10 g urea per 1000 ml of water. After that, the solution's pH was then modi ed using formic acid (L.C. industrial Co., Ltd, Nakhon Pathom, Thailand) to reach a nal pH of 3.5 2 . Media solution directly into electromagnetic air compressor (Hailea aco-318 oxygen pump, Sagar aquarium ®, Gujarat, India) ushed with oxygen to complete respiration for maximum cell growth at 72 hours ( nal estimated yeast as 1 × 10 9 cfu/ml, 15 . The media solution was mix on the RS ( 2:1 ratio) and adjusted with a moisture content of 650-750 g/kg to provide su cient ensilage conditions 16 . Fifteen kilograms of ensiled RS were put into plastic bags (size 24 × 42 inch, P.P Plastic Pagchong Co., Ltd, Nakhonrachasrima, Thailand), and sealed with a vacuum machine (Ima ex 1400W VC-921, Imar ex Industrial Co., Ltd., Bangkok, Thailand). To ensure anaerobic environment, the bags were securely sealed and fermented at room temperature for 14 days 13 .

Feeding and samples collection
The feeding trial lasted for 63 days (21 days/period with 3 periods); dairy cows were held in independent pens and individually fed roughage and concentrate diets at 07:00 and 16:00. Ensiled RS offered ad libitum for all cows. The experimental diet was formulated by using the KCF 2006 Program 17 . The ingredients and nutrient composition of ensiled RS and concentrate diet were provided in Table 1. During the experiment, mineral blocks and fresh water were accessible. The experiment was performed over 3 periods with double squares. The period lasted for 21 days, the rst 14 days for treatment adjustments and the last 7 days for sample intake and collection assessment.
In the time of the feeding trial, orts were obtained and weights were collected every day, and the feeding rate was adjusted daily to yield orts between 50 to 100 g/kg of intake. Individual voluntary feed determined consumption difference between the feed offered and orts.  23 . Silage uid subsamples were centrifuged for 15 minutes at 16,000 rpm and the liquid above the solid residue was ltered using a 0.45 micron syringe lter. High-performance liquid chromatography (HPLC) devices (Shimadzu LC-20A, Shimadzu Industrial Systems Co., Ltd, Kyoto, Japan) were used to conduct lactic acid (LA), acetic acid (C 2 ), propionic acid (C 3 ), and butyric acid (C 4 ) analyses 24 . The ammonia-nitrogen (NH 3 -N) concentration was calculated according to the Kjeldahl process 18 .
Jugular blood and rumen uid samples were obtained at 0 and 4 hours after feeding on the last day of each period. A blood sample (approximately 10 mL) was obtained in tubes containing 12 mg of ethylene diamine tetra-acetic acid (EDTA) from the jugular vein. The plasma was isolated by centrifugation for 10 min at 500 × g and preserved at − 20 C until blood urea-nitrogen (BUN) analysis, according to Crocker 25 . Approximately 200 mL of rumen uid was collected from the rumen by a stomach tube connected to a vacuum pump. Rumen uid was assessed immediately by the pH meter (Hanna Instruments HI 8424 microcomputer, Hanna Instruments (Thailand) Ltd, Bangkok, Thailand) for determining the pH and temperature. Rumen uid samples were then ltered through 4 cheesecloth layers. A uid sample containing 5 mL of 1 mol/L of H 2 SO 4 applied to 45 mL of rumen uid was put into the bottle. The rumen uid mixture was centrifuged for 15 min at 16,000 × g and used for analyzing the NH 3 -N (AOAC, 1998) and volatile fatty acid (VFA) (gas chromatography, Model HP6890-Hewlett, NY, USA; 26 . Methane (CH 4 ) production was calculated using VFA pro les following the equation CH 4 (g/d) = 0.45 × C 2 (mmol/L) -0.275 × C 3 (mmol/L) + 0.40 × C 4 (mmol/L) according to Moss et al. 27 .. Ruminal bacteria, protozoa, and fungal zoospores were numbered under a hemocytometer using the direct counting method 28 .
During the last 7 days of each experimental period, milk samples were taken according to the yield for morning and afternoon milking, preserved with 2-bromo-2 nitropropane-1, 3-dial, and stored at 4 C until analysis by using Milko-Scan (Foss Electric, Hillerod, Demark) for fat, true protein, lactose, total solids (TS), and solids-not-fat (SNF) content. Milk urea nitrogen (MUN) was estimated by the diacetyl monoxime method using UV/Vis-spectrophotometer (PG Instruments Ltd., London, UK) according to Ochei

Statistical analysis
All data from the experiment were analyzed according to a 3 × 3 replicated Latin square design using the GLM procedure 34 according to the model: where Yijk, observation from cow j, receiving ensiled RS i, in period k; µ, the overall of mean, Sl, the effect of square (l = 1, 2); Mi, effect of yeast species in RS silage (i = 1, 2, 3); Aj, the effect of cows (j = 1, 2, 3, 4, 5, 6); Pk, the effect of period (k = 1, 2, 3); and εijk, the residual effect. Signi cant differences between individual means were evaluated using the Duncan's multiple comparison tests when a signi cant (P < 0.05) effect was detected 35 . Standard errors of means were calculated from the residual mean squares in the analysis of variance.