Data and code from: Onset of feeding behavior in intensively housed dairy calves
Data files
Feb 17, 2026 version files 2.91 MB
-
1_Descriptive_intake_and_figures_260122.pdf
236.73 KB
-
1_Descriptive_intake_and_figures_260122.Rmd
17.03 KB
-
1_Intake_and_rumination_complete_daily_data.xlsx
258.67 KB
-
2_Age_at_reaching_thresholds_calculations_260122.pdf
988.42 KB
-
2_Age_at_reaching_thresholds_calculations_260122.Rmd
22.17 KB
-
2_Age_at_reaching_thresholds.xlsx
26.56 KB
-
3_Onset_models_and_figures_260122.pdf
1.14 MB
-
3_Onset_models_and_figures_260122.Rmd
202.19 KB
-
README.md
12.84 KB
Abstract
Development of early feed and water intake is not well-documented in dairy calves, and little is known about how traditional, intensive feeding management influences the development of species-specific behaviors, like rumination. Our first objective was to describe grain, water, and hay intake in calves from birth. Our second objective was to evaluate whether early access to hay influenced the likelihood of reaching biologically relevant thresholds of grain and water intake (grain: 50, 100, 250, 500, 900 g; water: 1, 2, 3, 4, 5 L), and rumination (5, 10, 15, 20 % of a 24-h d) compared to grain alone. Our third objective was to evaluate whether hay provision affected the age at which calves reached these thresholds. We enrolled 49 Holstein heifer calves over 2 yr (2018: 22; 2019: 27) that were housed individually and fed milk replacer via a bottle (2018: 5.7-8.4 L/d; 2019: 3.8-5.6 L/d), as is common in the US. Calves received ad libitum water and grain (Control: n = 20) with additional access to chopped mountaingrass hay (Hay: n = 29) from d 0. Intake was measured daily until the start of weaning at 49-51 d. Calves were ear-tagged at 3 ± 2 d (mean ± SD) of age with an accelerometer (SenseHub, Merck Animal Health) that classified rumination by minute continuously through 38 ± 10 d (mean ± SD) of age. We analyzed age of onset as the first of 3 consecutive days that a calf met the threshold level with a zero-inflated hurdle model. The logistic regression component of this model evaluated whether access to hay affected the likelihood of reaching each onset threshold. The second, conditional component of the model evaluated, among the calves that met a given threshold, if having access to hay affected the age they reached it. We found that calves consumed grain, hay, and water within 1 d of birth. Calves consumed small quantities of hay throughout the milk-fed period, typically < 100 g/d, but this still led to differences in feeding development, with hay-fed calves reaching 5 % rumination at younger ages, and a greater proportion reaching higher rumination thresholds compared to controls: specifically, 89 % vs. 53 % reached the 15 % threshold, and 61% vs. 5 % reached the 20 % threshold (Hay vs. Control, respectively). Hay-fed calves also tended to be more likely to reach the 500 g grain threshold. There was no evidence that access to hay affected age at reaching higher rumination or grain thresholds, possibly because all calves were in similarly restrictive, milk-limited environments, where calories from solid food intake were particularly important. Calves with hay tended to reach the 3 L water threshold at a younger age than controls, but there was no evidence that hay affected likelihood of consuming increasing amounts of water. Overall, our results demonstrate that calves will consume grain, hay, and water as soon as they are provided, highlighting the importance of opportunities to perform species-specific feeding behaviors. Withholding hay from very young animals, which is common in the US and allowable under the EU 2008 Directive, limits the development of motivated behaviors like rumination.
B.C. Downey and C.B. Tucker
Data were collected from 49 Holstein heifer calves across 2 years (2018-2019) from birth until the start of step-down weaning (approximately d 50). Calves were fed ad libitum grain and water from birth (Control, n = 20) or given additional access to ad libitum hay (Hay, n = 29). All calves were individually housed and fed milk replacer via a bottle (2018: 5.7-8.4 L/d; 2019: 3.8-5.6 L/d). We measured feed and water intake daily from d 0. Calves were ear-tagged on d 3 ± 2 (mean ± SD; range: d 0-7) of life with an accelerometer (MSD Animal Health Technology) that classified rumination by minute continuously throughout the milk-fed period. Due to tag failure, 72 % of all possible calf-days for 49 calves from birth through the start of weaning (1770/2451 days) were available. We used these data to describe early life feed and water intake, and whether access to hay affected the likelihood of reaching biologically relevant thresholds of grain and water intake, and rumination, as well as whether hay affected the age at which calves reached these thresholds.
Description of the data and file structure
Data
- 1_Intake_and_rumination_complete_daily_data.xlsx;
- This file has daily intake and rumination values for each calf throughout the data collection period. Calves have data from day 0 (birth) until weaning began. Weaning began on day 49 to 51 for all calves, so the last day of data collection was day 48 to 50. Weaning age varied in one year due to another experiment.
- Each sheet includes columns for:
- "Year" (2018 or 2019; the year data collection occurred for each calf)
- "Calf" (calf ID), "Treatment" (Hay or Control)
- "Age_days" (the calf's age in days; day 0 = the day the calf was born)
- "Week" (the calf's age in weeks)
- and a final column for the data collected that day
- NAs represent missing or unknown data.
- For grain, hay, and water, these NAs typically occurred because of issues with drying feed samples or calves spilling or defecating in buckets, as these issues made it impossible to generate an accurate 24-h representation of what the calf consumed.
- NAs for rumination occurred due to issues with the accelerometers, either because they did not record any data, recorded < 1440 minutes (1 full day) of data, recorded "low activity" (1 of the 8 behavior categories possible) for > 80 % of the day (occurred within the first week of life; suggested to reflect an active tag that had not yet been placed on a calf for a full day) or failed entirely.
- This file is used in the RMarkdown files "1_Descriptive_intake_and_figures_260122.Rmd" and "2_Age_at_reaching_thresholds_calculations_260122.Rmd".
* "Grain" sheet; This sheet contains daily grain intake in grams. The "Grain_g" column is presented on a dry matter basis. All calves (20 Control, 29 Hay) are represented in this sheet.
* "Hay" sheet; This sheet contains daily hay intake in grams. The "Hay_g" column is presented on a dry matter basis. Only Hay treatment calves (n = 29) are represented in this sheet. There are a few 0s in this sheet; these are true 0s, i.e., calves did not consume any hay on that day.
* "Water" sheet; This sheet contains daily water intake in liters ("Water_L" column).
* "Rumination" sheet; This sheet contains rumination data ("Rumination_prop" column). This column represents the proportion of a 24-h day that the calf was recorded as ruminating. Tags collected data every minute continuously from the application until the end of the data collection period or the tag failed, whichever came first (see "Supplemental Table S1" for more details). NAs reflect missing data (either the tag did not record or recorded less than a full 24-h day), while 0s reflect true 0s, i.e., the tag recorded that calf did not ruminate at all on that day.
- 2_Age_at_reaching_thresholds.xlsx;
- This file is generated from "1_Intake_and_rumination_complete_daily_data.xlsx" in the RMarkdown file "2_Age_at_reaching_thresholds_calculations_260122.Rmd". It is used in the file "3_Onset_models_and_figures_260122.Rmd".
- This file has the age at which calves reached each threshold for grain, water, hay intake and rumination. Daily intake and rumination data were analyzed to identify when each calf met or exceeded each threhold for at least 3 consecutive days. The age reported in this file is the first day of that first 3 day period.
- Missing data (NAs in the complete daily data) were excluded; i.e., if a calf ate 50 g of grain on day 1, 60 g on day 2, had an NA on day 3, then ate 55 g on day 4, they would not be recorded as having met the threshold for 3 consecutive days due to the presence of the NA.
- "." in each sheet means that the calf did not reach that threshold, while "NA" means missing or excluded data.
* "Grain" sheet; This sheet contains age in days that calves reached 50, 100, 250, 500, and 900 g of grain intake.
* "Hay" sheet; This sheet contains age in days that calves reached 25, 50, 100, 150, and 200 g of hay intake. Only Hay Treatment calves are included on this sheet, as Control calves were never given access to hay.
* "Water" sheet; This sheet contains age in days that calves reached 1, 2, 3, 4, and 5 L of water intake.
* "Rumination" sheet; This sheet contains age in days that calves reached 5, 10, 15, and 20 % of a 24-h day ruminating. Two calves (2942, 3045) have NA values reflecting missing or incomplete daily tag data that prevented their analysis for age of meeting thresholds.
Code
All analyses were run in R using RStudio. Software versions, including those for all packages used, are included in each RMarkdown file. Each RMarkdown file is exported as a PDF as well, which has the same name. These PDF exports show both the code and outputs of the associated RMarkdown file, and are particularly useful for users that are not planning to run the code in real time, or are less familiar with R, but would like to view the processing steps.
- 1_Descriptive_intake_and_figures_260122.Rmd; This annotated RMarkdown script uses grain, water, and hay intake data collected daily in Holstein heifer calves from birth through the start of step-down weaning (~d 50) to calculate descriptive data on the development of hay, grain, and water intake, particularly in the first week of life, and create a figure of intake throughout the milk-fed period.
- 1_Descriptive_intake_and_figures_260122.pdf; This is the knit PDF output of the "1_Descriptive_intake_and_figures_260122.Rmd" coding file.
- 2_Age_at_reaching_thresholds_calculations_260122.Rmd; This file uses daily data ("1_Intake_and_rumination_complete_daily_data.xlsx") to determine calf age at reaching specific thresholds of intake and rumination. This code identifies the first of 3 consecutive days that each calf reached or surpassed each threshold. NA values do not count towards the 3 consecutive days. These ages are then exported ("2_Age_at_reaching_thresholds.xlsx").
- 2_Age_at_reaching_thresholds_calculations_260122.pdf; This is the knit PDF output of the "2_Age_at_reaching_thresholds_calculations_260122.Rmd" script.
- 3_Onset_models_and_figures_260122.Rmd; This file uses data on the ages at which calves reached each threshold ("2_Age_at_reaching_thresholds.xlsx"; calculated and generated in the second RMarkdown file, "2_Age_at_reaching_thresholds_calculations_260122.Rmd") to model the effect of hay on likelihood of reaching grain, water, and rumination thresholds, and whether, of the calves that met the thresholds, access to hay influenced the age at which calves reached these thresholds. This file also evaluates the development of hay intake in calves fed hay by describing the age at which hay intake thresholds were reached.
- 3_Onset_models_and_figures_260122.pdf; This is the knit PDF output of the "3_Onset_models_and_figures_260122.Rmd" script.
Supplemental
- Supplemental_Tables_and_Figures.pdf; This PDF contains 2 supplemental tables and 1 supplemental figure.
* Supplemental Table S1.; This table shows accelerometer ear tag (MSD Animal Health Technology) active days for all calves. Calves were tagged in the right ear on d 3±2 (mean±SD; range: d 0-7) of life. “First day tag was active” was considered the first full day the tag recorded after the tag was applied, or the first full day a tag reported a “low activity” percentage lower than 80 % after birth. “Low activity” values higher than 80 % were considered to reflect time when the tag was active but not yet applied to the calf, or had not yet registered the calf’s movement correctly. Tags collected behavioral data continuously from this starting point until the start of weaning, or until the tag failed, whichever came first. In some cases, tags began recording a few days after birth, and failed before weaning. This table presents columns for "Year" (2018 or 2019; the year data collection occurred for each calf), "Calf" (calf ID), "Treatment" (Hay or Control), "First day tag was active" (calf's age in days when the tag first began recording behavior for a complete 1440 min day), "Last day tag was active" (calf's age in days when the tag stopped recording behavior), "Total number of days tag recorded" (difference between first and last days tag recorded), "Ideal number of days tag would have recorded" (number of days from birth until the day before weaning began; due to variable weaning days, calves would ideally have 49-51 days where the tag was active), and "% of data collection period with active tag" (percentage of the entire data collection period that the calf had an active, recording tag, calculated by dividing "Total number of days tag recorded" by "Ideal number of days tag would have recorded"). NAs reflect missing data, as 1 calf (2942) had complete tag failure and had no data recorded.
* Supplemental Table S2.; This table shows model-predicted probabilities, means, and test statistics from hurdle models evaluating the effect of hay provision and year, in some cases, on the onset of grain and water intake and rumination in individually housed Holstein heifer calves fed milk replacer, water, and grain (Control, n = 20) and mountaingrass hay (Hay, n = 29). This table presents columns for "Outcome" (dependent variables, i.e. rumination, grain intake, and water intake), "Threshold" (predetermined levels of behavior or intake), "Likelihood of not reaching threshold" including model-predicted values (estimated marginal probabilities and standard error for Control and Hay calves at each threshold level for each outcome) and zero-inflated (ZI) model component test statistics (chi-square and P-value) for fixed effects (e.g. Treatment for all; Year and a Treatment:Year interaction for some models); "Age at reaching threshold (d)" including model predicted values (estimated marginal means and standard error for Control and Hay calves, or calves raised in 2018 and 2019, at each threshold level for each outcome) and conditional model component test statistics (chi-square and P-value), which was only evaluated when n (number of calves reaching the threshold) was greater than or equal to 2/treatment. "." reflects that model component not being run, i.e. in the "Likelihood of not reaching threshold" column it means that there were no zeros (all calves reached that threshold) and thus the ZI component was not evaluated; in the "Age at reaching threshold (d)" column it means that fewer than 2 calves per treatment reached that threshold and thus the conditional component was not evaluated. Test statistics for the ZI and conditional model components are also included in the associated manuscript, in a table alongside the observed (raw) means.
* Supplemental Figure S1.; This figure shows the results of a linear regression comparing proportion of a 24-h day engaged in rumination obtained through live observations using 1-0 sampling at 5-s intervals compared with behavior monitoring accelerometer ear tags (SenseHub, Merck Animal Health). The figure includes a red line representing a perfect relationship (coefficient of determination = 1, slope = 1, intercept = 0) and dotted line representing the estimate generated by the tag comparison. Comparisons were made on a subset of 16 calves from 2019 who were observed on week 4 of life and had tag data available on the same day; each point represents an individual calf. Detailed information about live observations is available in Downey and Tucker (2023). Coefficient of determination, slope estimate, and intercept estimate generated by the linear regression are reported in the figure, and p-values comparing the slope to 1 and intercept to 0 are presented in the caption.
