Effects of rumen-protected lysine and dry matter intake on metabolism of dairy cows

Fehlberg, Laura K.

Permalink

https://hdl.handle.net/2142/115502 Copy

Description

Title

Effects of rumen-protected lysine and dry matter intake on metabolism of dairy cows

Author(s)

Fehlberg, Laura K.

Issue Date

2021-12-22

Director of Research (if dissertation) or Advisor (if thesis)

Cardoso, Felipe C

Doctoral Committee Chair(s)

Cardoso, Felipe C

Committee Member(s)

• Drackley, James K
• Loor, Juan J
• Condotta, Isabella
• Pan, Yuan-Xiang

Department of Study

Animal Sciences

Discipline

Animal Sciences

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• amino acid
• dry matter intake
• lysine
• liver composition
• machine learning
• transition period

Abstract

Throughout the lifespan of a dairy cow, she will encounter multiple types of stress, which have the ability to decrease production. During the transition period, dairy cows undergo metabolic and reproductive changes that pose challenges to completing a smooth transition into early lactation and may result in metabolic disorders that have large economic repercussions. This includes a decrease in dry matter intake (DMI) that does not support the requirements of lactation, resulting in a negative energy and protein balance. To combat this, it is vital to supply the most limiting amino acids, notably Lys and Met for dairy cows consuming corn, soy-based diets, in an intestinally available form. Losses in DMI can have additional economic ramifications; therefore, limiting fluctuations in DMI while allowing for adequate DMI would be beneficial to the dairy industry. However, classical modeling techniques may not be adequate to predict fluctuations in DMI; therefore, machine learning methods should be explored. Stress in the dairy industry is one of the most limiting aspects of management that affects production and economic success; therefore, this research was conducted to determine potential dietary modification to ameliorate stress during the transition period and limit large fluctuations in DMI in non-lactating or lactating dairy cows. Experiments 1 and 2 were companion studies and were conducted to determine the effect of feeding rumen-protected Lys (RPL) prepartum [-26 d relative to calving (DRC)], postpartum (28 DRC), or both on performance, health, blood metabolites, immunometabolic status, and liver composition in dairy cows. Seventy-five multiparous Holstein cows were assigned to 1 of 4 dietary treatments in a randomized, complete block design with a 2 × 2 factorial arrangement. Treatments prepartum consisted of TMR top-dressed with RPL (PRE-L) or without (PRE-C) and postpartum treatments consisted of TMR top-dress with RPL prepartum and postpartum (PRE-L POST-L), with RPL prepartum and without RPL postpartum (PRE-L POST-C), without RPL prepartum and with RPL postpartum (PRE-C POST-L), and without RPL prepartum and postpartum (PRE-C POST-C) in 300 g of molasses. Cows in PRE-L had greater body weight (BW) at -2 and -1 wk prior to calving compared to those in PRE-C. Body weight was greater and DMI tended to be greater for cows in PRE-L POST-L and PRE-L POST-C compared to those that were in PRE-C POST-L and PRE-C POST-C. Energy-corrected milk, milk fat, milk true protein, milk casein, and milk lactose yields were greater for cows in PRE-L POST-L and PRE-L POST-C compared to those that were in PRE-C POST-L and PRE-C POST-C. Plasma concentrations of Lys prepartum increased for cows in PRE-L compared to those in PRE-C. Concentrations of haptoglobin in plasma and glutathione peroxidase activity in serum were lower at d 0 for cows in PRE-L compared to cows in PRE-C. Glutathione peroxidase activity in serum was also less postpartum for cows in PRE-L compared to cows in PRE-C. Oxidative burst capacity in monocytes tended to be greater on d 7 postpartum for cows in PRE-L than cows in PRE-C. Additionally, mRNA expression of SAA and SOD1 were downregulated postpartum when RPL was consumed pre- and/or postpartum. There was a PRE × DRC interaction for plasma concentrations of glucose, cholesterol, total protein, and globulin and a tendency for triglyceride and albumin in which cows in PRE-L had greater concentrations in plasma prepartum than those in PRE-C. Postpartum plasma concentrations of albumin were greatest for cows in PRE-L POST-C compared to all others. Plasma concentrations of insulin were greater for cows in POST-L compared to cows in POST-C. Protein abundance of SLC7A7 pre- and postpartum tended to be greater when RPL was consumed prepartum while abundance of BBOX1 tended to be less pre- and postpartum compared to cows that did not consume RPL. Experiment 3 was conducted to predict DMI if individual cow DMI is not available, identify peaks (maximum) and valleys (minimum) in DMI, and forecast future DMI to simulate if decreasing DMI at peak results in a less dramatic decrease in DMI following the peak using machine learning modeling techniques. A dataset was compiled to include variables of interest from experiments conducted at the University of Illinois, including a total of 510 dairy cows from 12 different experiments. Cows ranged from -50 days relative to calving (DRC) to 350 DRC and included primiparous and multiparous. Cows were identified as being in stage 1 (-30 to -1 DRC), stage 2 (1 to 30 DRC), or stage 3 (31 to 350 DRC) and DMI curves were smoothed to decrease noise. Peaks and valleys were identified for individual cows on smoothed DMI. Regression models were utilized to predict DMI and long-short term memory deep learning time series was utilized to forecast 7 d of DMI based on a 7-d observation period from either actual or predicted DMI. Two regression models, decision tree and Gaussian process regression (GPR) were compared to determine accuracy. The GPR rational quadratic model was most accurate compared to the fine tree with a root mean square error (RMSE) of 5.98 and 6.98 and an r2 of 0.82 and 0.76, respectively when the model included days in milk, body condition score, body weight, and diet components. The average length from peak to peak was least for stage 1 and greatest for stage 3. The average length from peak to valley was least for stage 2 but greatest for stage 3. The RMSE of the forecasting model for actual, smoothed DMI was greater than the predicted, smoothed DMI model. Additionally, the forecasted days at peak, DMI at peak, difference between forecasted day at peak and observed day at peak, and difference between forecasted DMI at peak and observed DMI at peak were not different when the actual or predicted DMI was utilized as the input for the forecasting model. However, stage did have an effect on all variables, notably a difference between day at peak and DMI at peak, in which stage 2 forecasted a peak sooner and at a lesser peak DMI compared to stage 1 and 3. Forecast simulations to decrease peak DMI by 5 or 10 % of peak during the 7-d observation period and again forecast 7 d only altered the difference in DMI at peak in which the forecast with a 10 % reduction in DMI at observed peak resulted in a forecasted peak that was less compared to the forecast in which a peak was not reduced. In conclusion, prepartum supply of RPL is pertinent to postpartum performance. However, postpartum RPL had no effect on cows’ performance. Additionally, even though Lys is recognized as an amino acid that is primarily absorbed and utilized by the mammary gland, consuming RPL during the transition period resulted in molecular changes in liver composition and enhanced liver function. This may also be due to adipose tissue metabolism related to carnitine and β-oxidation in the liver and through improved immune status. Lastly, predicting DMI was successful using the GPR rational quadratic and peaks and valleys of DMI were determined. Predicted DMI can be utilized to forecast future DMI and to determine when the next peak in DMI will occur. Also, reducing observed peak DMI by 10 % resulted in a lower DMI for the subsequent forecasted peak. Overall, RPL had a positive impact on transition dairy cow production and immunometabolic status. Also, DMI was successfully predicted and reduction of peak DMI resulted in a lower subsequent peak.

Graduation Semester

2022-05

Type of Resource

Thesis

Copyright and License Information

Copyright 2022 Laura Fehlberg

Owning Collections