See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/328592867

Dimensions of the Modern Pig

Article · January 2018

DOI: 10.13031/trans.12826

CITATIONS READS
8 8,161

6 authors, including:

Isabella Condotta Tami Brown-Brandl

University of Illinois, Urbana-Champaign University of Nebraska at Lincoln
13 PUBLICATIONS 64 CITATIONS 182 PUBLICATIONS 3,626 CITATIONS

SEE PROFILE SEE PROFILE

John Stinn J.D. Davis

11 PUBLICATIONS 47 CITATIONS
Auburn University
61 PUBLICATIONS 342 CITATIONS
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Livestock systems View project

Identificação eletrônica para avaliação do comportamento dos suínos na fase de gestação View project

All content following this page was uploaded by J.D. Davis on 14 February 2019.

The user has requested enhancement of the downloaded file.

 DIMENSIONS OF THE MODERN PIG
I. C. F. S. Condotta, T. M. Brown-Brandl, J. P. Stinn,
G. A. Rohrer, J. D. Davis, K. O. Silva-Miranda

ABSTRACT. It is important to know the physical dimensions of livestock to properly design confined animal housing facilities
as well as feeding and drinking equipment. An engineering standard for the dimensions of livestock and poultry published
by ASABE reports swine dimensions that were originally published in 1968. Changes in animal husbandry practices for
swine, such as improved and new genetic lines, nutrition and feed form, and improved facility and equipment design, make
it necessary to validate or update these dimensions for modern animals. The objective of this study was to evaluate dimension
data for the grow-finish stages of modern pigs. A total of 150 growing-finishing pigs were sampled at five approximate
ages: 4, 8, 12, 16, and 20 weeks old (30 animals at each age). The animals equally represented three commercial sire lines
(Landrace, Duroc, and Yorkshire), and equal numbers of barrows and gilts were sampled. Dorsal and lateral color digital
and depth images were collected using a Kinect sensor as the pigs were held individually in a stanchion or scale, and the
images were analyzed by manual and automated methods. Measured physical dimensions included height from top of back
to the floor, length from nose to base of the tail, width at shoulders, jowl length, front leg height, body depth from top of
back to lowest point of the belly, and others. It was determined that the conformation of modern pigs has changed from the
dimensions reported in current engineering standards such that modern pigs tend to be wider (15.1%) and shorter in height
(-10.2%) and length (-4.9% on average) between 4 and 20 weeks of age. These updated pig dimensions will enable engineers
to better design modern swine equipment and facilities.
Keywords. Depth sensor, Dimensions, Image analysis, Precision livestock farming, Swine.

D
esign of facilities and equipment for farm ani- standard (ASABE, 2011) presents numerous physical meas-
mals needs to consider the mass as well as the urements taken from pigs throughout the grow-finish cycle.
size and conformation of modern animals. Ani- These dimension measurements are necessary for many dif-
mal facilities are typically segregated by growth ferent aspects of swine facilities, including feeder and
stage to meet the animals’ needs at each age. Animal space drinker design, physical space allowance, and different pen-
needs and equipment type and size are determined by the ning aspects. The measurements that are reported for the
typical animal size for each growth stage. Thus, the ability head, neck, jowl, and overall width are particularly im-
to specify the space required for animals is necessary portant for the design of pig feeders and the spacing of water
(Petherick, 1983); therefore, it is important to know how an- drinkers. The sizing of the feeders, including the width and
imals’ dimensions vary over time. depth, is critical to ensure that pigs have access to feed
In 1968, ASABE published dimension curves for farm throughout the grow-out period and to ensure that the space
animals (cattle, dairy, pigs, sheep, horses, and poultry), and is not large enough to promote feed wastage. These animal
these dimension curves are still in the standard today. This dimensions are also used to predict the mass of animals
(Schofield, 1990, 1999; Frost et al., 1997; Kashiha et al.,
2014; Wu et al., 2004; Kongsro, 2014).
Submitted for review in February 2018 as manuscript number PAFS Changes in production practices over the years, such as
12826; approved for publication by the Plant, Animal, & Facility Systems nutrition, genetic lines, and facility design, make it necessary
Community of ASABE in August 2018.
Mention of company or trade names is for description only and does not to validate or update these dimension curves for modern
imply endorsement by the USDA. The USDA is an equal opportunity swine. In addition, the current standard only has a graphical
provider and employer. representation of dimensions over time and not equations,
The authors are Isabella C. F. S. Condotta, Graduate Student, Luiz de
Queiroz College of Agriculture (ESALQ), University of São Paulo,
making it difficult to use accurately. The objective of this
Piracicaba, Brazil; Tami M. Brown-Brandl, Agricultural Engineer, study was to evaluate and update the physical dimension
USDA-ARS Meat Animal Research Center, Clay Center, Nebraska; John data for the grow-finish stages of modern pigs.
P. Stinn, Environmental Projects Engineer, Iowa Select Farms, Iowa Falls,
Iowa; Gary A. Rohrer, Research Geneticist, USDA-ARS Meat Animal
Research Center, Clay Center, Nebraska; Jeremiah D. Davis, Associate
Professor and Associate Director, Department of Biosystems Engineering MATERIAL AND METHODS
and National Poultry Technology Center, Auburn University, Auburn,
Alabama; Késia O. Silva-Miranda, Associate Professor, Luiz de Queiroz
The experiment was conducted in a nursery and a grow-
College of Agriculture (ESALQ), University of São Paulo, Piracicaba, finish building at the USDA-ARS Meat Animal Research
Brazil. Corresponding author: Tami M. Brown-Brandl, USDA-ARS- Center (USMARC) in Clay Center, Nebraska (-98.13° W,
MARC, P.O. Box 166, Clay Center, NE 68933; phone: 402-762-4279; 42.52° N). Digital RGB color images, depth images, and
e-mail: [email protected].

Transactions of the ASABE

Vol. 61(5): 1729-1739 2018 American Society of Agricultural and Biological Engineers ISSN 2151-0032 https://doi.org/10.13031/trans.12826 1729
masses were collected on a population of grow-finish pigs at in a single five-hole feeder, with one feeder space for ap-
five different time points. The first data acquisition time was proximately eight pigs (Korthals, 2000). Water was provided
weaning (4 weeks of age) when the animals were moved to with four nipple drinkers per pen, with one drinker for ap-
the nursery. The remaining four data acquisition times were proximately ten pigs. All animals (150 pigs) were weighed
during the grow-finish period (8, 12, 16, and 20 weeks of at the same time as images were acquired. Ages were calcu-
age) in the finishing building. All animal procedures were lated from the animal’s exact birth date to the day on which
approved by the USMARC IACUC and followed recognized images were collected.
guidelines for animal use and care (FASS, 2010).
IMAGE ACQUISITION
ANIMAL SPECIFICS An image acquisition program was developed in numeri-
A total of 150 grow-finish pigs were randomly sampled cal computing software (MATLAB, ver. R2015b) to acquire
at five nominal ages: 4, 8, 12, 16, and 20 weeks old (30 ani- both digital RGB color and depth images with a commer-
mals at each age from a total building population of cially available depth sensor (Kinect, ver. 1). The program
234 pigs). The animals equally represented three commer- was deployed on a Windows-based laptop for data collec-
cial sire lines (Landrace, Duroc, and Yorkshire) with 50 pigs tion. At 4 weeks of age, each animal was placed inside a con-
per sire line. The maternal line was a mix of Landrace and structed stanchion (fig. 1) to prevent it from turning around.
Yorkshire. A balance of barrows and gilts and sire lines was The stanchion had an acrylic side panel that did not interfere
represented during each measurement period such that five with digital image acquisition of the animal profile. For the
animals of each sex and sire line were measured at each time last four ages (8, 12, 16, and 20 weeks), image acquisitions
point. were made during weighing of the animals. A Kinect sensor
The pigs were housed in separate nursery and finishing was positioned above the animal to acquire dorsal color and
buildings. During the nursery phase, the pigs were housed at depth images (figs. 2a and 2c). A second sensor was posi-
20 pigs per pen (0.24 m2 pig-1). The nursery building was tioned to the side of the animal to acquire lateral color and
equipped with evaporative cooling pads. During the finish- depth images (figs. 2b and 2d). The lateral depth images
ing phase, the pigs were housed at 39 pigs per pen (0.93 m2 were used to measure the distance between the sensor and
pig-1), exceeding the recommendations from MidWest Plan the animal. The scale used for mass acquisition of all pigs
Service (MWPS, 1983), thus ensuring maximum growth was a Rice Lake Weighing Systems digital scale that was
performance (McGlone and Newby, 1994). The finishing calibrated by the manufacturer and regularly checked with a
building was equipped with a sprinkle cooling system. The 22.6 kg mass. Digital RGB color and depth images were
temperatures in the building were within the range 17°C to taken at approximately 1 s intervals. One or more images per
35°C, with an average temperature of 25°C. The thermo- animal were selected for analysis based on the clarity of the
neutral zone was shown to be between 17.4°C and 23.2°C image, ensuring that the entire pig was captured within the
(Brown-Brandl et al., 2013). Throughout the grow-finish image in a selected posture (standing, head down, and
phase, pigs had ad libitum access to feed and water. Diets straight legs) and standing naturally with specific measure-
were a mix of corn and soybean meal formulated to meet or ment regions of the animal visible. If this posture was not
exceed National Research Council recommendations (NRC, perfectly met, different images were used for acquiring the
2012). Specific diets are shown in table 1. Feed was provided different dimensions to ensure accuracy.

Table 1. Compositions of diets. Ration 309B302 was used for pigs entering the growing-finishing phase up to approximately 50 kg, ration 309B402
was used for pigs ranging from 50 to 70 kg, and ration 309B502 was used for pigs ranging from 70 to 130 kg.
Ration (and Weight Range)
309B102 309B202 309B302 309B402 309B502
Ingredient (0 to 10 kg) (10 to 25 kg) (25 to 50 kg) (50 to 70 kg) (70 to 130 kg)
BMD 60 - - - - 0.03
Carbodox 1.00 1.00 - - -
Chlortetracycline 50 - - 0.10 0.10 -
Choline 0.10 - - - -
Copper sulfate - 0.05 - - -
Corn 53.88 65.32 71.19 75.47 80.21
Dicalcium phosphate 0.60 1.35 1.00 0.60 0.45
Fishmeal 5.00 - - - -
L-lysine 0.30 0.35 0.25 0.23 0.15
Limestone 0.55 0.75 0.95 0.85 0.80
Methionine 0.13 0.13 0.05 0.01 -
Oat, steam rolled 5.00 - - - -
Phytase 600 0.09 0.09 0.05 0.05 0.05
Salt 0.30 0.30 0.30 0.30 0.30
Soybean meal 21.30 29.15 24.65 20.95 16.60
Soybean oil 1.00 1.00 1.00 1.00 1.00
Swine vitamin premix #10 0.20 0.20 0.20 0.20 0.20
Threonine 0.11 0.13 0.06 0.05 0.01
Trace minerals H - swine 0.20 0.20 0.20 0.20 0.20
Whey 10.00 - - - -
Zinc oxide 0.25 - - - -

1730 TRANSACTIONS OF THE ASABE

 portable network graphics (.png) format. The depth images
were used for acquiring the width and average height of the
animals and the distance from the sensor to the animal. The
depth images (640 × 480 pixels) were saved in a space-de-
limited text file (.txt). The mass of each animal was recorded
simultaneously. A similar procedure was used for the pigs at
4 weeks of age, and the mass was recorded on the same day.

IMAGE PROCESSING
Figure 3 illustrates the dimensions of each animal that
were obtained from the collected images. Two image analy-
sis methods were used: automated analysis of the dorsal
depth images, and manual analysis of the color lateral im-
ages. Automated analysis was performed with the depth data
captured by the sensor positioned above the animal. An al-
gorithm was written to select the region of interest within the
image and to acquire selected dimensions. Depth data pro-
Figure 1. Constructed stanchion with acrylic side panel used to restrain vided by the sensor located laterally to the animal were used
weaning pigs at 4 weeks.
to collect the distance between the animal and the sensor.
The presence of the bars in front of the animal made auto-
The digital RGB color images were used for animal iden- mated acquisition of dimensions difficult to replicate. There-
tification and for acquiring animal dimensions from the lat- fore, the dimensions were acquired manually to ensure pre-
eral views of the animals. As the pigs walked on to the scale, cise measurements.
their number was written on a small white board and held in An algorithm was written in numerical computing soft-
front of the camera to ensure that each image could be iden- ware (MATLAB, ver. R2015b) for automated acquisition of
tified. The digital images (640 × 480 pixels) were saved in dimensions from the dorsal view. This automation of the

Figure 2. Positioning of Kinect sensors at (a) dorsal and (b) lateral positions of the constructed stanchion used for imaging pigs at 4 weeks of age
and at (c) dorsal and (d) lateral positions of the scale used for weighing pigs between 8 and 20 weeks of age.

61(5): 1729-1739 1731

 were used for this study.
The lateral digital color images obtained with the sensor
located on the side of the scale (figs. 5a and 6a) was used to
manually acquire animal height (H), animal depth (D),
length (L), taper length (TL), jowl length (J), head height
(HH), and face length (F). This was done using the imtool
function in MATLAB to measure these distances on the im-
age in pixels. Length of the front legs (FL) was calculated by
subtracting D from H.
The lateral depth images (fig. 5b) were used to obtain the
distance between the sensor and the animal for unit conver-
sion (eq. 1). This distance was measured using a single point
on the depth image (marked with an “x” in fig. 6b). Figure 6
illustrates the manually and automatically measured regions.
Equation 1 (Condotta, 2016) was used to transform the di-
Figure 3. Important dimensions of a pig (L = total length, TL = taper mensions from pixels to cm:
length (from shoulders to snout), W = width of body at shoulders, F =
face length (from nose to base of ears), HH = head height (from base of lcm = l px × 2.49 ×10−3 × Z 0.95 (1)
jowl to top of head between ears), J = jowl length (from nose to front
legs), FL = length of front legs, D = depth (from lowest point on belly to where
top of back), and H = height.
lcm = length (cm)
lpx = length (pixels)
dorsal images (fig. 4) was developed as part of another study
Z = distance from sensor to animal (cm).
on estimating mass from projected volume (Condotta et al.,
The average residual error using equation 1 was 2.78%.
2018). This process yielded the average height, width at
Dimension curves were generated using the regression func-
shoulders, length of body from neck to rump, area, and pro-
tion in Microsoft Excel.
jected volume. The width at shoulders and average height

Figure 4. (a) Dorsal RGB color image and (b) dorsal depth image obtained with the Kinect sensor located above the pig, (c) elimination of areas
in the depth image that were outside of the pre-established range, and (d) dorsal surface of the pig after eliminating the head and tail.

1732 TRANSACTIONS OF THE ASABE

Figure 5. (a) RGB image and (b) depth image obtained with the Kinect sensor located on the side of the pig. The white “x” in (b) represents the
point at which the distance from the Kinect sensor was measured.

Figure 6. White lines in (a) represent manually measured physical dimensions: total length, taper length (from shoulders to snout), face length
(from nose to base of ears), head height (from base of jowl to top of head between ears), jowl length (from nose to front legs), depth (from lowest
point on belly to top of back), and height. Black line in (b) represents width at shoulders. The black dot is the centroid of the white area.

DATA ANALYSIS
The growth profiles (both mass and dimensions) were de-
F ( n,d ) =
( SSr − SS g ) / ( DFr − DFg ) (2)
scribed using power equations calculated in SigmaPlot (ver. SS g / DFg
12.0). Log transformation was applied to the physical dimen-
sions (total length, taper length, width of shoulders, face where
length, head height, jowl length, length of front legs, depth, SSr = sum of squares of the residual of the reduced model
and height) and weight data to obtain prediction and confi- SSg = sum of squares of the residual of the global model
dence intervals. Global multiple linear regressions were gen- DFr = degrees of freedom of the residual of the global
erated for each transformed dimension against transformed model
weight, sex, and sire line. A reduced simple linear regression DFg = degrees of freedom of the residual of the reduced
model was generated for each transformed dimension model.
against transformed weight. Efroymson’s algorithm (step-
wise regression; Efroymson, 1960) was used to test the level
of significance of sex and sire line in the multiple linear re-
gression equations. The null hypothesis considered the re-
RESULTS AND DISCUSSION
A total of 150 digital and depth images, each from a
duced model equivalent to the global model, and the alterna-
unique animal, were captured and analyzed for this study.
tive hypothesis considered the global model different from
the reduced model. The global model was developed in Mi- The pigs weighed 7.6 ±1.5 kg, 17.5 ±3.0 kg, 45.4 ±5.6 kg,
crosoft Excel and considered the effects of sex and sire line 72.6 ±7.7 kg, and 119.75 ±11.5 kg (average ±SD) at each of
using dummy variables (Draper and Smith, 1998). The test the five data acquisition times. The growth of the pigs over
statistic is given in equation 2: time is shown in figure 7. On average, modern pigs increase

61(5): 1729-1739 1733

Figure 7. Change in mass with age of grow-finish pigs raised in typical
nursery and finishing barns. Pigs were a representative sample from
three different sire lines (Duroc, Landrace, and Yorkshire) and a single
maternal line (Yorkshire × Landrace).

in mass 22% faster during the grow-finish period than re-

ported for 1963 pigs in the ASABE standard (ASABE,
2011). The pigs measured in this study (0.85 kg d-1) matched
the growth rate of 2016 animals (0.84 kg d-1), as reported in
the 2016 Pork Industry Productivity Analysis (Stalder,
2017). However, the curves in figure 7 show two different
growth profiles. It appears that the modern pig growth rate
is slower during the nursery period compared to the older
(1963) genetic lines. This may be due to the earlier weaning
Figure 8. Dimensions (cm) versus mass (kg) of pigs (L = total length,
age of modern piglets (3 to 4 weeks). In the 1960s, pigs were H = height, TL = taper length, D = depth (from lowest point on belly to
weaned at an older age (6 to 8 weeks; Valsickle, 2004), and top of back), W = width of shoulders, J = jowl length, FL = length of
the later weaning could have resulted in less of a depression front legs, HH = head height (from a point in front of front legs to top
in growth in the nursery. However, despite the slower of head between ears), and F = face length (from nose to ears).
growth in the nursery, it is apparent that the growth of the
Table 2. Power equation coefficients for dimensions determined from
modern genetic lines exceeds the growth of the older genetic body mass (L = total length, H = height, TL= taper length, D = depth,
lines starting at approximately 12 weeks. W = width of shoulders, J = jowl length, FL = front leg length, HH =
There was a relative increase in taper length (35%) and head height, and F = face length).
head height (8%) as well as a relative reductions in jowl Dimension[a] a b R2
length (24%) and face length (6%) when comparing the data L 26.8 ±0.7 0.33 ±0.01 0.95
H 14.4 ±0.4 0.35 ±0.01 0.96
acquired in this study with the ASABE standard (ASABE, TL 10.7 ±0.6 0.30 ±0.01 0.81
2011). Figure 8 illustrates the increase in physical dimen- D 7.7 ±0.2 0.37 ±0.01 0.97
sions (cm) with the increase in animal mass (kg). The largest W 8.7 ±0.2 0.30 ±0.01 0.96
changes in dimensions were for taper length (+34%), jowl J 7.6 ±0.5 0.31 ±0.02 0.74
length (-24%), and width (+16%). The total length (-1%) and FL 6.7 ±0.4 0.31 ±0.01 0.80
front leg height (-3%) changed the least. Table 2 contains HH 6.3 ±0.5 0.33 ±0.02 0.75
F 7.7 ±0.4 0.21 ±0.01 0.66
coefficients for a power equation (eq. 3) of each dimension [a]
Dimension = aMb, where M is body mass.
curve over the range of body mass as well as each respective
coefficient of determination (R2):
performing measurements of these dimensions. For the other
Di = aM b
(3) dimensions, the coefficients of determination were all
greater than 0.9, which indicates that the model fits well for
where those variables.
Di = physical dimension (cm) When comparing the modern pig dimension data with the
a = coefficient 1963 pig dimensions, pigs today are approximately 1% shorter
M = body mass (kg) in total length (L), 5% shorter in height (H), 6% smaller in
b = coefficient. depth (D), but 17% wider in width (W) at equivalent mass
(table 3). While it is understood that current genetic lines grow
The physical dimension data related to the head of the an- at a faster rate than the 1963 genetic lines measured as part of
imal (jowl length, head height, face length, and taper length) the ASABE standard (ASABE, 2011), this study illustrates
and the length of the front legs as related to the mass of the that the conformation of the pigs has changed. Commercial
animal had smaller coefficients of determination. This could genetic selection emphasizes a saleable product, which could
be due to the relatively small change in these physical di- explain the shorter and wider skeletal frame and increased
mensions over the growth of the pigs and/or the difficulty in muscle as compared to past genetic lines.

1734 TRANSACTIONS OF THE ASABE

Table 3. Percent difference between newly obtained dimension data depth are smaller than the prediction errors for other dimen-
and ASABE (2011) dimension data for different weight ranges (L =
sions (front leg length, head height, jowl length, face length,
total length, H = height, D = depth, and W = width of shoulders).
Mass Dimension and taper length). This may be because repeatability is
(kg) L H D W harder to maintain (for both automated and manual image
1 to 10 3.36% -7.18% -11.15% 26.83% acquisition) when acquiring dimensions related to the ani-
10 to 30 0.36% -5.81% -7.17% 19.56% mal’s head because the position of the head affects the di-
30 to 50 -1.06% -5.13% -5.16% 16.19%
50 to 70 -1.86% -4.75% -4.00% 14.31%
mensions (e.g., due to loose skin on the neck or change in
70 to 90 -2.43% -4.47% -3.17% 13.01% the position of the ears). The confidence intervals for all di-
90 to 110 -2.86% -4.26% -2.52% 12.00% mensions are given in table 4. These ranges are important for
Average -0.75% -5.27% -5.53% 16.98% the design of swine equipment to ensure functionality for the
smallest and largest pigs in each weight interval.
There was no effect of sex or commercial sire line on the Understanding how pig physical dimensions have
predicted width (p = 0.2423), length (p = 0.1143), height (p = changed over time is important for the design of modern
0.1173), depth (p = 0.0942), front leg height (p = 0.1176), housing structures and equipment. Feeder spacing is one
head height (p = 0.2777), jowl length (p = 0.6963), face such design parameter that depends on shoulder width (W).
length (p = 0.4911), and taper length (0.7748). The predic- In 1983, Petherick (1983) estimated shoulder width (cm) us-
tion and confidence intervals for these dimensions versus ing equation 2 with coefficients of a = 6.1 and b = 0.33. For
mass regressions after log transformation are shown in fig- a 20 kg pig, the estimate for W in this study was 30.5% wider
ure 9. The prediction errors for width, length, height, and than the shoulder width reported by Petherick (1983), i.e.,

Figure 9. Confidence intervals for physical dimensions (width, length, height, depth, leg length, height, head height, taper length, jowl length, and
face length) versus mass (kg) after log transformation.

61(5): 1729-1739 1735

Table 4. Predicted ranges (95% confidence intervals) for dimensions (cm) at different weights (kg): W = width of shoulders, L = total length, H =
height, D = depth (from lowest point on belly to top of back), FL = length of front legs, HH = head height (measured from a point in front of the
front legs to top of head between the ears), J = jowl length, F = face length (from nose to ears), and TL = taper length.
Mass Dimension
(kg) W L H D FL HH J F TL
5 13.8 to 14.3 44.4 to 46.4 24.5 to 25.7 13.7 to 14.2 10.6 to 11.7 10.0 to 11.2 11.9 to 13.4 10.4 to 11.3 16.6 to 18.2
10 16.8 to 17.8 55.1 to 58.7 30.9 to 32.9 17.5 to 18.5 12.9 to 14.8 12.2 to 14.4 14.5 to 17.0 11.8 to 13.2 20.2 to 22.9
20 20.5 to 22.1 68.5 to 74.4 38.8 to 42.2 22.4 to 24.1 15.7 to 18.8 15.0 to 18.5 17.5 to 21.7 13.3 to 15.6 24.4 to 28.8
30 23.1 to 25.0 77.7 to 85.4 44.4 to 48.8 25.9 to 28.1 17.7 to 21.6 16.9 to 21.5 19.6 to 25.0 14.4 to 17.1 27.3 to 33.0
40 25.1 to 27.4 85.1 to 94.2 48.9 to 54.1 28.7 to 31.3 19.2 to 23.9 18.3 to 23.9 21.3 to 27.7 15.1 to 18.3 29.6 to 36.3
50 26.7 to 29.3 91.2 to 101.7 52.6 to 58.6 31.0 to 34.1 20.4 to 25.7 19.6 to 25.9 22.6 to 29.9 15.8 to 19.3 31.4 to 39.1
60 28.2 to 31.0 96.6 to 108.2 55.9 to 62.6 33.1 to 36.5 21.5 to 27.4 20.6 to 27.6 23.8 to 31.9 16.3 to 20.1 33.1 to 41.5
70 29.4 to 32.6 101.4 to 114.0 58.8 to 66.2 35.0 to 38.7 22.4 to 28.9 21.6 to 29.2 24.9 to 33.7 16.8 to 20.9 34.5 to 43.7
80 30.6 to 33.9 105.7 to 119.3 61.5 to 69.4 36.7 to 40.7 23.3 to 30.2 22.5 to 30.7 25.8 to 35.3 17.2 to 21.5 35.8 to 45.6
90 31.6 to 35.2 109.7 to 124.2 63.9 to 72.4 38.3 to 42.6 24.1 to 31.5 23.2 to 32.0 26.7 to 36.8 17.5 to 22.1 37.0 to 47.5
100 32.6 to 36.4 113.3 to 128.8 66.2 to 75.2 39.7 to 44.3 24.8 to 32.6 24.0 to 33.3 27.5 to 38.1 17.9 to 22.7 38.1 to 49.1
110 33.5 to 37.5 116.8 to 133.0 68.3 to 77.8 41.1 to 46.0 25.5 to 33.7 24.7 to 34.5 28.2 to 39.4 18.2 to 23.2 39.1 to 50.7
136 35.6 to 40.0 124.8 to 143.0 73.3 to 84.0 44.3 to 49.8 27.1 to 36.3 26.2 to 37.2 29.9 to 42.5 18.9 to 24.4 41.4 to 54.4
181 38.7 to 43.7 136.5 to 157.6 80.5 to 93.0 49.1 to 55.5 29.4 to 40.0 28.5 to 41.3 32.4 to 46.9 19.9 to 26.0 44.9 to 59.8
227 41.3 to 46.9 146.5 to 170.3 86.8 to 100.9 53.2 to 60.5 31.3 to 43.2 30.5 to 44.9 34.5 to 50.8 20.8 to 27.5 47.7 to 64.5

21.4 cm versus 16.4 cm. For a 120 kg pig, W in this study article estimate the dimensions of pigs at 136, 181, and
was 23.6% wider than in the previous study, i.e., 36.6 cm 227 kg to be 84 × 143 × 40 cm, 93 × 158 × 44 cm, and 101
versus 29.6 cm. Baxter (1986) used the shoulder width di- × 170 × 47 cm (height × length × width), respectively, with
mension plus 10% length to size feeder spacing, which al- 95% confidence (table 4), indicating that an increase in scale
lowed for animal variability, but also discussed behavioral dimensions is needed. For example, using the dimension
considerations. More recently, DEFRA (2003) provided equations presented in this article, the scale stanchion for a
feeder spacing recommendations that varied by animal size, 136 kg pig should be at least 5 cm higher, 19 cm longer, and
ranging from 10 cm pig-1 at 5 kg to 30 cm pig-1 at 120 kg. At 2 cm wider to accommodate the largest finishing pigs.
120 kg, W was 22.0% wider in this study than the required Changes in animal conformation can also have an impact
feeder spacing of 30 cm. The recommendations by DEFRA on thermal regulation. A pig that is shorter in length and height
(2003) may be inadequate for modern swine genetics in the and wider at a similar mass would equate to a pig with a
U.S. A feeder spacing of 38 cm was reported by PIC (2014) smaller ratio of surface area to volume. With a smaller surface
with a suggested number of pigs per feeder by body weight. area to volume ratio, modern pigs would have more difficulty
PIC (2014) suggested that this spacing should be the mini- dissipating heat to the environment. Estimates of surface area
mum W allowance for large market animals. The 38 cm to volume ratios were calculated and are presented in the Ap-
spacing would allow an average pig of up to 136 kg; how- pendix. Modern pigs have been shown to have increased heat
ever, a 40 cm spacing would allow for 95% of the pigs at this production (Brown-Brandl et al., 2004, 2014). This has been
weight. theorized to be due to faster growth, higher lean percentage,
Another example of a design parameter is nipple drinker and more muscle mass. Greater heat production coupled with
height, which depends on pig height. Harmon and Meyer a decrease in surface area to volume ratio increases the heat
(2008) suggested drinker height ranges of approximately 10- sensitivity of modern pigs. This highlights the need to re-eval-
15 cm, 15-30 cm, 30-45 cm, 45-60 cm, and 60-75 cm for pig uate the thermal comfort zone of modern pigs.
weight ranges of approximately 0-5 kg, 5-15 kg, 15-35 kg,
35-70 kg, and 70-110 kg. Harmon and Meyer (2008) stated
that nipple drinkers should be at mid-shoulder height or CONCLUSIONS
higher. The height ranges of the pigs in this study, at the
A total of 150 growing-finishing pigs from three modern
same weight ranges, were 14-25 cm, 25-33 cm, 35-52 cm,
sire lines were used to measure various physical dimensions
47-66 cm, and 59-78 cm with 95% confidence (table 4).
reported in 1963 and currently used in an ASABE standard.
Therefore, the nipple drinker height recommendations for
It was determined that modern pigs increase in mass at a
pigs larger than 15 kg matches the height ranges presented
faster rate and have a different conformation than previously
in this article. However, based on this study, nipple drinker
reported. Modern pigs are shorter in height and length; how-
heights for pigs up to 15 kg need to be increased.
ever, they are up to 20% wider than pigs with older (1963)
For the design of weighing scales, most of the commer-
genetics. These changes may affect the physical arrangement
cial options available today underestimate pig sizes. Finish-
of equipment and the space requirements for modern hous-
ing weights are increasing, and scale manufacturers, alt-
ing systems. Changes in conformation have resulted in mod-
hough adapting for weight capacity, are not adapting for an-
ern pigs with a smaller surface area to volume ratio. This
imal dimensions. It is possible to find swine scales that can
conformation, coupled with the increased heat production re-
weigh animals up to 136, 181, and 227 kg (300, 400, and
ported previously in the literature, highlights the need to re-
500 lb) with dimensions of approximately 79 × 124 × 38 cm, evaluate the thermal comfort zone for modern grow-finish
77 × 122 × 44 cm, and 98 × 140 × 44 cm (height × length × pigs. The information provided by this study will improve
width), respectively. The dimension curves presented in this the data in the ASABE standard.

1736 TRANSACTIONS OF THE ASABE

ACKNOWLEDGEMENTS Pamphlet PM 1493. Ames, IA: Iowa State University Extension.
The authors would like to thank John Holman, Dale Kashiha, M., Bahr, C., Ott, S., Moons, C. P., Niewold, T. A.,
Janssen, and Hannah Speer for their help in collecting data, Ödberg, F. O., & Berckmans, D. (2014). Automatic weight
and Donna Griess for her help in preparing the manuscript for estimation of individual pigs using image analysis. Comput.
Electron. Agric., 107, 38-44.
publication. This research was funded in part by the USDA https://doi.org/10.1016/j.compag.2014.06.003
Agricultural Research Service, the São Paulo Research Foun- Kongsro, J. (2014). Estimation of pig weight using a Microsoft
dation (FAPESP), and the National Council for Scientific and Kinect prototype imaging system. Comput. Electron. Agric.,
Technological Development (CNPq) of Brazil. 109, 32-35. https://doi.org/10.1016/j.compag.2014.08.008
Korthals, R. L. (2000). Evaluation of space requirements for swine
finishing feeders. Trans. ASAE, 43(2), 395-398.
https://doi.org/10.13031/2013.2717
REFERENCES McGlone, J. J., & Newby, B. E. (1994). Space requirements for
ASABE. (2011). D321.2 (R2011): Dimensions of livestock and finishing pigs in confinement: Behavior and performance while
poultry. St. Joseph, MI: ASABE.
group size and space vary. Appl. Animal Behav. Sci., 39(3-4),
Baxter, M. R. (1986). The design of the feeding environment for the 331-338. https://doi.org/10.1016/0168-1591(94)90166-X
pig. PhD diss. Aberdeen, UK: University of Aberdeen.
MWPS. (1983). Structures and environment handbook. MWPS-1.
Brown-Brandl, T. M., Eigenberg, R. A., & Purswell, J. L. (2013). Ames: Iowa State University, MidWest Plan Service.
Using thermal imaging as a method of investigating thermal
Nienaber, J. A., Hahn, G. L., Eigenberg, R. A., Korthals, R. L., Yen,
thresholds in finishing pigs. Biosyst. Eng., 114(3), 327-333. J. T., & Harris, D. L. (1997). Genetic and heat stress interaction
https://doi.org/10.1016/j.biosystemseng.2012.11.015
effects on finishing swine. Proc. 5th Intl. Livestock Environment
Brown-Brandl, T. M., Hayes, M. D., Xin, H., Nienaber, J. A., Li, Symp. (vol. 2, pp. 1017-1023). St. Joseph, MI: ASAE.
H., Eigenberg, R. A., ... Shepherd, T. (2014). Heat and moisture
NRC. (2012). Nutrient requirements of swine. Washington, DC:
production of modern swine. ASHRAE Trans., 120(1), 469-489. National Research Council.
Brown-Brandl, T. M., Nienaber, J. A., Xin, H., & Gates, R. S.
Petherick, J. C. (1983). A biological basis for the design of space in
(2004). A literature review of swine heat production. Trans. livestock housing. In S. H. Baxter, M. R. Baxter, & J. A. D.
ASAE, 47(1), 259-270. https://doi.org/10.13031/2013.15867
MacCormack (Eds.), Farm animal housing and welfare (pp.
Condotta, I. C. F. S. (2016). Automatic assess of growing-finishing 103-120). Boston, MA: Martinous Nijhoff.
pigs weight through depth image analysis. MS thesis. Piracicaba, PIC. (2014). Wean to finish manual. Hendersonville, TN: PIC North
Brazil: Universidade de São Paulo. America. Retrieved from
Condotta, I. C. F. S., Brown-Brandl, T. M., Silva-Miranda, K. O., & http://na.picgenus.com/sites/genuspic_com/Uploads/Wean%20T
Stinn, J. P. (2018). Evaluation of a depth sensor for mass o%20Finish%20Manual.pdf
estimation of growing and finishing pigs. Biosyst. Eng. (in Schofield, C. P. (1990). Evaluation of image analysis as a means of
press). https://doi.org/10.1016/j.biosystemseng.2018.03.002 estimating the weight of pigs. J. Agric. Eng. Res., 47, 287-296.
DEFRA. (2003). Code of recommendations for the welfare of https://doi.org/10.1016/0021-8634(90)80048-Y
livestock: Pigs. London, UK: Department for Environment, Schofield, C. P., Marchant, J. A., White, R. P., Brandl, N., &
Food, and Rural Affairs. Wilson, M. (1999). Monitoring pig growth using a prototype
Draper, N. R., & Smith, H. (1998) Chapter 14: “Dummy” variables. imaging system. J. Agric. Eng. Res., 72(3), 205-210.
In Applied regression analysis (pp. 299-326). New York, NY: https://doi.org/10.1006/jaer.1998.0365
John Wiley & Sons. Stalder, K. J. (2017). 2016 Pork industry productivity analysis. Des
Efroymson, M. A. (1960). Multiple regression analysis. In A. Moines, IA: National Pork Board. Retrieved from
Ralston & H. S. Wilf (Eds.), Mathematical methods for digital https://www.pork.org/wp-content/uploads/2017/11/2017-pork-
computers (pp. 191-203). New York, NY: John Wiley & Sons. industry-productivity-analysis.pdf
FASS. (2010). Guide for the care and use of agricultural animals in Valsickle, J. (2004). Later weaning is a no brainer. National Hog
research and teaching (3rd Ed.). Champaign, IL: Federation of Farmer (15. Feb. 2004). Retrieved from
Animal Science Societies. https://www.nationalhogfarmer.com/mag/farming_later_weanin
Frost, A. R., Schofield, C. P., Beaulah, S. A., Mottram, T. T., Lines, g_no
J. A., & Wathes, C. M. (1997). A review of livestock monitoring Wu, J., Tillett, R., McFarlane, N., Ju, X., Siebert, J. P., & Schofield,
and the need for integrated systems. Comput. Electron. Agric., P. (2004). Extracting the three-dimensional shape of live pigs
17(2), 139-159. https://doi.org/10.1016/S0168-1699(96)01301-4 using stereo photogrammetry. Comput. Electron. Agric., 44(3),
Harmon, J., & Meyer, V. (2008). Nipple waterers for swine. 203-222. https://doi.org/10.1016/j.compag.2004.05.003

APPENDIX
The surface area to volume ratio is an important parameter
for thermal regulation. The larger the surface area for a given
volume, the easier it is for an animal to dissipate heat by both
sensible and latent means. Changes in animal conformation
can lead to changes in the surface area to volume ratio. Calcu-
lating the surface area and volume of an ellipsoid cylinder
(fig. A1) using the depth, width, and length dimensions of pigs
provides a simple estimation of pig volume and surface area.
The surface area and volume equations are as follows: Figure A1. Ellipsoid cylinder where a is the radius of the minor axis, b
is the radius of the major axis, and L is the length.

61(5): 1729-1739 1737

 (
 a 2 + b2
A = 2π 
)  L (A1)
a = minor axis radius (cm)
b = major axis radius (cm)
 2  L = cylinder length (cm).
 
  For estimation of the surface area and volume of a pig,
the major axis was the pig width, the minor axis was the pig
where depth, and the length of the cylinder was the length of the
A = surface area (cm2) pig (fig. A2). For this estimation, the legs were not consid-
a = minor axis radius (cm) ered. As stated in the text of the article, the conformation of
b = major axis radius (cm) pigs has changed over time, resulting in pigs that are approx-
L = cylinder length (cm). imately 1% shorter in total length (L), 5% shorter in height
(H), 6% smaller in depth (D), but 17% wider in width (W) at
V = abπL (A2) an equivalent mass (table 3). The surface areas and volumes
where of 1963 and 2017 pigs at various weights were calculated
V = volume (cm3) using equations A1 and A2, respectively (table A1). The sur-

(a) (b)
Figure A2. How an ellipsoid cylinder estimates the surface area to volume ratio of (a) 1963 and (b) 2017 genetic lines of pigs. Note the shorter
stature and length, smaller depth, and great width of the 2017 genetics compared to the 1963 genetics.
Table A1. Dimensions and calculations of the surface area, volume, surface area to mass ratio (SA:M), surface area to volume ratio (SA:V), and
percent change from 2017 to 1963 genetics.
Mass Surface Area Volume Percent Change
(M, kg) L D W (SA, cm2) (V, cm3) SA:M SA:V SA:M SA:V
2017 Genetics
5 45.58 13.97 14.10 2010 7050 402 0.29 18.62% -10.61
10 57.30 18.05 17.36 3188 14101 319 0.23 -2.04% -8.33
20 72.02 23.33 21.37 5062 28201 253 0.18 1.32% -7.30
30 82.34 27.10 24.14 6638 42302 221 0.16 -2.86% -4.55
40 90.53 30.15 26.31 8048 56402 201 0.14 -0.42% -5.10
50 97.45 32.74 28.13 9345 70503 187 0.13 1.49% -4.91
60 103.50 35.03 29.71 10561 84603 176 0.12 2.99% -6.54
70 108.90 37.08 31.12 11711 98704 167 0.12 1.96% -6.88
80 113.80 38.96 32.39 12809 112804 160 0.11 3.28% -5.74
90 118.31 40.70 33.56 13864 126905 154 0.11 3.23% -6.49
100 122.50 42.31 34.64 14880 141005 149 0.11 4.07% -7.92
110 126.41 43.83 35.64 15865 155106 144 0.10 3.63% -9.06
1963 Genetics
5 41.00 15.00 11.00 1694 5313 339 0.32 - -
10 60.00 20.00 14.00 3254 13195 325 0.25 - -
20 73.00 25.00 18.00 4996 25800 250 0.19 - -
30 84.00 30.00 21.00 6833 41563 228 0.16 - -
40 91.00 32.70 23.00 8082 53754 202 0.15 - -
50 96.50 35.00 24.90 9208 66052 184 0.14 - -
60 103.00 36.50 26.00 10254 76770 171 0.13 - -
70 109.00 39.00 27.00 11486 90146 164 0.13 - -
80 113.00 40.00 29.00 12402 102950 155 0.12 - -
90 119.00 41.00 30.00 13430 114959 149 0.12 - -
100 124.00 42.00 30.50 14298 124756 143 0.11 - -
110 130.00 43.00 31.00 15308 136102 139 0.11 - -

1738 TRANSACTIONS OF THE ASABE

 face area to volume ratios were also calculated. The esti- face area to volume ratio even further. While important for
mated surface area to volume ratio of the 2017 genetic lines thermal regulation, ear size and skin folds are difficult to
decreased by 4.5% to 10.6% compared to the 1963 pigs, with measure and are not an economically important trait; thus,
the largest difference occurring at the extreme weights there are no historical measurements of these parameters.
(lightest and heaviest). While this is only an estimate of the However, using these simple estimations, it is clear that to-
body volume, i.e., neglecting the legs, skin folds, and ears, day’s pigs have a decreased ability to dissipate heat (Niena-
the legs of modern pigs are shorter, thus decreasing the sur- ber et al., 1997).

61(5): 1729-1739 1739

V i e w p u b l i c a t i o n s t a t s