Skip to main content

Data from: Providing long hay in a novel pipe feeder or a bucket reduces some, but not all abnormal oral behaviors in milk-fed dairy calves

Cite this dataset

Downey, Blair Caitlin; Tucker, Cassandra Blaine (2023). Data from: Providing long hay in a novel pipe feeder or a bucket reduces some, but not all abnormal oral behaviors in milk-fed dairy calves [Dataset]. Dryad.


Many milk-fed dairy calves are not provided forage. In these settings, calves often perform abnormal repetitive behaviors (ARBs), including tongue rolling and nonnutritive oral manipulation (NNOM), that, based on their form, seem similar to movements used when processing feed. Feeding hay, typically presented as a short chop (≤5 cm) in a bucket, reduces ARBs. Our objective was to evaluate whether altering the presentation method of long hay (approx. 19 cm), by providing it in a bucket or a novel polyvinyl chloride (PVC) pipe feeder, could reduce ARBs. Holstein heifer calves were housed individually on sand and fed ad-libitum starter grain and limited milk replacer (5.7-8.4 L/d step-up) via a bottle (Control, n=9) or given access to mountaingrass hay in a bucket (Bucket, n=9) or in a PVC pipe feeder (Pipe, n=9). The 56 x 10.2 cm (length x diameter) PVC pipe feeder had four 6.4 cm-wide openings that required the calf to insert her tongue into the pipe and curl her tongue to extract hay. Treatments were applied from birth through 50 d of age, when step-down weaning began and TMR was provided to all calves. Calves were fully weaned at d 60. At wk 4 and 6, oral behaviors (eating, ruminating, drinking water, sucking milk, self-grooming, NNOM, tongue rolling, tongue flicking, and panting) were recorded by direct observation for 24 h using 1-0 sampling during 5-s intervals. Feeding long hay, regardless of presentation method, increased overall DMI, grain intake, and ADG compared to Control calves. Hay provision also increased rumination (25% vs. 15% of 24-h observations in Control) and eating time (5.5% vs. 2% in Control). Abnormal behaviors were seen in all calves. Hay provision reduced some of these, including NNOM (5% vs. 9% in Control). There was no difference in NNOM between calves fed hay in a pipe or bucket, even though Bucket calves ate more hay (178 vs. 129 g/d in wk 6) and tended to spend more observations eating hay than Pipe calves (4.5% vs. 3%). Hay provision did not affect other behaviors: drinking water (0.5%), grooming (3%), or tongue flicking (3%). We also found evidence of other abnormal oral behaviors that have received less attention. Calves showed signs of polydipsia, and displayed excessive grooming, the latter indicated by overall duration, number of bouts per day, and duration of individual bouts (up to 25 min). Tongue rolling was expressed at low levels (up to 0.4% of intervals) but by 85% of calves. Feeding hay, both in a bucket and using novel methods, was not enough to counteract the welfare challenges associated with individual housing and limited ability to suck milk (<1% of time). Provision of long hay, regardless of presentation method, promotes rumination, improves performance (higher grain intake and ADG) and reduces at least some, but not all, of the considerable abnormal oral behaviors these calves performed.


Please see the associated manuscript for details.

Usage notes

Supplemental material:

The supplemental materials in this section are referenced in the corresponding paper, and provide additional information and context for the experiment’s design and results. Raw data, RMarkdown files, and tables are presented as proportions, which is how the data were analyzed. RMarkdown files that produce figures convert these proportions to percentages, and the corresponding manuscript similarly converts all proportions to percentages to improve readability.

Supplemental tables and figures:

Supplemental Tables and Figures.pdf

  • Supplemental Table S1. Feed and ADG model outputs: Model outputs for feed (grain, hay, and TMR) intake and water intake in the preweaning and weaning periods, and average daily gain (ADG) during the preweaning period.
  • Supplemental Table S2. Behavior model outputs: Back-transformed model-predicted means (using type = “response” in emmeans in R), SE, and 95% CI for all oral behaviors scored, along with model outputs.
  • Supplemental Figure S1: Bucket and pipe set up for Control (left), Bucket (middle), and Pipe (right) treatments. Control calves had grain, water, and empty buckets (left to right and an empty pipe feeder; Bucket calves had grain, water, and hay buckets (left to right) and an empty pipe feeder; Pipe calves had grain, water, and empty buckets (left to right) and a hay pipe feeder. Perforated rubber mats placed on top of sand bedding for newborn calves (d 0-5) are seen in the middle and right photos. All treatments had access to buckets and pipes beginning on d 0. At the start of step-down weaning (50-60 d), TMR was provided for Control and Pipe calves in the empty bucket (right) and for Hay calves in place of hay (right).
  • Supplemental Figure S2: Starter grain fed to all treatments (left; Starter Calf Feed 901033, Associated Feed and Supply Co.), and long chop (~19 cm) of mountaingrass hay fed to Bucket and Pipe treatment calves (right).
  • Supplemental Figure S3: Instructions for creating PVC pipe feeders.
  • Supplemental Figure S4: Overall percentage of 24-h observations spent drinking water in weeks 4 and 6 compared to amount of water consumed on the corresponding day. Each point represents an individual calf.
  • Supplemental Figure S5: Overall percentage of 24-h observations engaged in tongue rolling in weeks 4 and 6. Each colored line represents an individual calf.
  • Supplemental Figure S6: Percentage of time engaged in tongue rolling in weeks 4 and 6 by individual calf averaged over 15-min bins in a 24-h day. Each colored line represents a different calf. Dashed lines represent milk feedings, which occurred at 0930 and 1630 h.


Short videos illustrating ruminating, eating grain, eating hay, sucking milk, drinking water, grooming, NNOM: Other, tongue flicks, tongue rolling, and panting are available via Downey, Blair Caitlin, Jensen, Margit Bak, & Tucker, Cassandra Blaine. (2022). Data from: Hay provision affects 24-h performance of normal and abnormal oral behaviors in dairy calves. Zenodo.

Additional videos are provided here. These videos show what eating hay from the pipe feeder looked like: a calf feeding in real time (Video 1), a calf feeding in slow motion (Video 2), a calf feeding by biting and pulling hay out (Video 3), and a video of a non-trial heifer using the pipe from the side to show what tongue movements inside the pipe looked like (Video 4). Another video demonstrates an example of what the behavior “NNOM: Pipe” looked like (Video 5).

Raw data and code:

This section includes the data and code used for this paper.

All data in this section were collected from 27 calves fed a standard diet of grain, milk, and water (Control treatment, n=9) with additional access to ~19-cm hay in a bucket (Bucket treatment, n=9) or a PVC pipe feeder (Pipe treatment, n=9) from birth until weaning at 60 d. The column “Age_days” reflects the calf’s age when the intake was recorded, while “Date” reflects the exact date of intake or body weight measurement. “Week” corresponds to the calf’s age in days (i.e. 1-6 days = week 0, 7-13 days = week 1, etc.).


  • FEiC_Behavior proportion data.csv: Proportion of time engaged in oral behaviors (drinking water, eating grain, eating hay, eating, grooming, panting, NNOM: Pipe, NNOM: Other, NNOM: Total, ruminating, sucking milk, tongue flicks, and tongue rolling) generated from 1-0 sampling at 5-s intervals for a continuous 24-h day at weeks 4 and 6 of life. The column labeled “Eating” is “eating.grain" + “eating.hay” combined while “NNOM: Total” is “NNOM.pipe” + “NNOM.other.”
  • FEiC_Grooming bout data.csv: Consecutive grooming bouts generated from 1-0 sampling at 5-s intervals for a continuous 24-h day at weeks 4 and 6 of life. The column “Duration_5s_intervals” indicates the number of 5-s intervals that grooming occurred in. For example, a duration of 3 = three 5-s intervals, or approximately a 15-s bout.
  • FEiC_Feed intake data.xlsx: The “Grain intake” sheet includes daily intake of starter grain in grams (g) from birth through the end of weaning at 60 d. The “Hay intake” sheet includes daily intake of hay in grams (g) from birth through the start of weaning at 50 d. The “TMR intake” sheet includes daily intake of TMR in grams (g) through the weaning period (50-60 d). The “Water intake” sheet includes daily intake of water in liters (L) from birth through the end of weaning at 60 d. For grain, hay, and TMR, the column labeled “[feed type]_consumed_dried_g” indicates that these values have been converted to a dry matter basis (see paper methods for more information about this weekly feed sampling). For all solid feed and water, NAs reflect instances where 24h intake could not be reliably captured, e.g. because of fecal contamination in the bucket, spillage, negative intake values, or lack of DM conversion. The sheet “Water intake vs. drinking time” includes water intake on the day of each 24h observation for each calf.
  • FEiC_Weekly calf weights and ADG.xlsx: “Weekly calf weights” sheet presents weights for calves obtained via a scale at d 0, then weekly, that were used to determine preweaning average daily gain (ADG), ADG across the full trial, and final weight differences. The second tab of this sheet shows only ADG by calf (produced by the corresponding .Rmd script, but also included here). In this sheet, “ADG_kg/d” reflects kg/d difference from the last recorded weight for each calf. NAs are present on d 0 for all calves for “ADG_kg/d” as this was their first recorded weight (birthweight) and thus gain over time could not be calculated.

RMarkdown code:

All codes in this section use the data files listed above to build and validate models to describe effects of treatment (Control, Bucket, Pipe) and time (Week or Day) on behavior, feed intake, and ADG, and/or produce the figures used in the corresponding paper. Rmd files can be downloaded and run in R, or downloaded and viewed via the pdf output, which includes the code and output previews. Both versions are annotated.

  • FEiC_Average daily gain models (pdf and .Rmd)
  • FEiC_Feed intake data models and figures (pdf and .Rmd)
  • FEiC_Behavior models and boxplots (pdf and .Rmd)
  • FEiC_Repetitive grooming bout analysis and figure (pdf and .Rmd)