Theriogenology 85 (2016) 495–501

Contents lists available at ScienceDirect

Theriogenology
journal homepage: www.theriojournal.com

Ultrasonographic reproductive tract measures and pelvis

measures as predictors of pregnancy failure and anestrus in
restricted bred beef heifers
Dietmar E. Holm a, *, Mirjam Nielen b, Ruurd Jorritsma b, Peter C. Irons a,
Peter N. Thompson a
a
Department of Production Animal Studies, Faculty of Veterinary Science, University of Pretoria, Onderstepoort, South Africa
b
Department of Farm Animal Health, Faculty of Veterinary Medicine, Utrecht University, Utrecht, Netherlands

a r t i c l e i n f o a b s t r a c t

Article history: Previous reports have shown that reproductive tract score (RTS) can predict reproduction
Received 23 June 2015 outcomes in seasonally bred beef heifers, although the accuracy can vary. Some ultrasono-
Received in revised form 25 August 2015 graphic measures of the female reproductive tract and pelvis area have also been associated
Accepted 16 September 2015
with reproductive outcome in young heifers. The objectives of this study were to determine
which transrectal ultrasound or pelvis measures taken at a single examination are inde-
Keywords:
pendent predictors of reproductive failure and whether the RTS system can be optimized
Beef cattle
with this information. In this observational study, year-old beef heifers (n ¼ 488) in 2 birth
Fertility
Heifer selection cohorts were followed from just before the first breeding until confirmation of pregnancy. A
Pelvis area single pre-breeding examination included body condition score, RTS, ultrasound measures
Reproductive tract score of the reproductive tract (length and diameter of the left and right ovaries, presence and
Ultrasonography diameter of a CL, largest follicle diameter and left uterus horn diameter) and transverse and
vertical diameters of the pelvis. Additional farm records including dam parity, sire, birth
weight and birth date, weaning weight, weaning date, prebreeding body weight, AI dates,
and semen used were available. Breeding consisted of 50 days of AI, followed 5 to 7 days later
by a 42-day bull breeding period. Pregnancy failure was defined as the failure to become
pregnant after the AI and bull breeding periods, while anestrus was defined as the failure to
be detected in estrus during the 50-day AI period. From the prebreeding data and farm
records, independent predictors of pregnancy failure and anestrus were identified using
stepwise reduction in multiple logistic regression models. Age at the onset of breeding was
the only consistent independent predictor of pregnancy failure and anestrus in both cohorts
of this study (P < 0.05). Body condition score, uterus horn diameter, absence of a CL, largest
follicle of less than 13 mm, and pelvis area (PA) were the prebreeding examination variables
that remained in prognostic models (P < 0.1). Combining either the model based on the 3
remaining ultrasound measures or RTS with PA provided more accurate prognostic models
for pregnancy failure and anestrus than using RTS alone (P < 0.05). It is concluded that ul-
trasound measures have prognostic value for pregnancy failure in restricted bred yearling
heifers as a result of their association with anestrus, and that smaller PA has additional
prognostic value for poorly performing heifers.
Ó 2016 Elsevier Inc. All rights reserved.

1. Introduction

The ability to select young heifers that will reproduce

* Corresponding author. Tel.: þ27 125298000; fax: þ27 125298315. effectively in a seasonal breeding system has advantages
E-mail address: dietmar.holm@up.ac.za (D.E. Holm). over the alternative approach of waiting until reproductive

0093-691X/$ – see front matter Ó 2016 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.theriogenology.2015.09.031
496 D.E. Holm et al. / Theriogenology 85 (2016) 495–501

failure occurs [1,2]. Reproductive tract score (RTS) predicts pelvimeter; Lane Manufacturing, Denver, Colorado)
anestrus and pregnancy failure in heifers independently of [20,28,29]. Farm management and staff were blinded to all
age, body weight (BW), and body condition score (BCS) and the measured prebreeding data throughout the trial, except
is a valid selection tool to enhance reproductive performance for the prebreeding BW.
of herds [3–7]. However, estrus cycle stage and proportion of The MSD was October 15 of each year, and breeding
heifers in anestrus affect the accuracy of RTS, the complexity consisted of 50 days of continuous estrus observation by
of the RTS system affects its repeatability, and other tests visual inspection, and once daily AI of all heifers identified
with potential to improve RTS are available [5,7–9]. in estrus during the preceding 24 hours by the same
The ultrasonographic presence of a CL has been used to inseminator. Five to 7 days after each 50-day AI season, all
mark the onset of puberty, has substantial repeatability, is heifers were joined with bulls in a single multisire group at
more accurate than blood progesterone determination, and a heifer:bull ratio of 30-35:1 for 42 days. Pregnancy di-
is a predictor of reproductive outcome in seasonally bred agnoses were performed by transrectal palpation 138 or
cows and heifers [9–12]. Ovary size is associated with 165 days after MSD (2007 and 2008 cohorts, respectively).
antral follicle count (AFC), which in turn is associated with For the purpose of regression models, BCS was catego-
follicular reserve and fertility, whereas AFC is not affected rized into 2 categories (<6 and 6) and RTS into 3 cate-
by estrus cycle stage [13,14]. Maximum follicle diameter is gories (1–2, 3, and 4–5) [30]. Diameter of the largest follicle
correlated with uterus, cervix, and vaginal diameter and was used either as a continuous variable or was dichoto-
increases in the 10 weeks before first ovulation in heifers mized using various cutoffs (7, 8, 9, 12, 13, and 14 mm).
because of increased LH pulse frequency [13,15–19]. Larger Pelvis area (PA) was calculated as the product of the TD and
PA has been associated with early onset of puberty in VD, and rescaled values of PA (RPA) and uterus diameter
heifers and improved libido in bulls [20–22]. (RUD) were calculated within birth cohort using the
The objective of this study was to determine which in- following formula:
dividual transrectal ultrasound or pelvis measures taken at
one point in time before breeding are independent pre- X  ¼ ðX  X minimum ÞOðX maximum  X minimum Þ
dictors of reproductive failure in seasonally bred beef
heifers and whether this knowledge can be used to opti- where X* ¼ RPA or RUD and X ¼ PA or UD.
mize RTS. If a heifer was not detected in estrus, it was assumed
that she remained prepubertal until the end of the 50-day
2. Materials and methods AI season and was defined as anestrus, whereas pregnancy
failure was defined as a negative pregnancy test at the end
This was an observational study of 488 uniquely iden- of the AI and bull breeding periods.
tified Bovelder beef heifers born in either 2007 (n ¼ 259) or Correlations were estimated using Spearman correla-
2008 (n ¼ 229) (2007 and 2008 cohorts) that were fol- tion for non-normally distributed data (only age in this
lowed from just before their first breeding season to study) and Pearson correlation for normally distributed
confirmation of pregnancy. The farming system, breed, and data. Independent proportions were compared using the
location have been described previously [5,7,20,23–25]. Fisher exact test, and means and medians were compared
Farm data collected included the following: birth weight using analysis of variance with the Tukey–Kramer multiple
and birth date, parity of dam, sire, bull allocated, and first to comparison test and the Kruskal–Wallis one-way analysis
fourth AI day numbered from the mating start date (MSD). of variance, respectively.
Heifers were weighed either 22 days (2007 cohort) or 27 Multiple linear regression models (for length of the
or 24 days (2008 cohort) before the MSD (prebreeding BW), longest ovary, diameter of the largest follicle, UD, and PA)
and a single prebreeding examination was performed and logistic regression models (for absence of a CL, absence
7 days (2007 cohort) or 27 or 24 days (2008 cohort) before of a follicle 13 mm, anestrus, and pregnancy failure) were
the MSD. During the prebreeding examination, heifers constructed using a backward elimination process [31]
were restrained individually in a chute, and the following with P < 0.20 for initial inclusion and PWald < 0.10 for
data were collected in the same order by one experienced retention in models. Predictors that were considered
veterinarian: First, BCS was determined using a 9-point included year of birth, dam parity (1, 2, or 3), prebreeding
scale [26]. This was followed by RTS by transrectal palpa- BW (kg), growth rate (kg/day), age at onset of breeding
tion using a 5-point scale [3], then followed by transrectal (days) and BCS category at examination, presence or
ultrasonographic measurements of the reproductive tract diameter (mm) of the CL, diameter (mm) of the largest
[27], using a real-time digital ultrasound imaging system follicle or presence of a follicle of at least 7, 8, 9, 12, 13, or
set in B-mode with a variable frequency linear probe set at 14 mm, RUD, RPA, length of the longest or shortest ovary
5 MHz (SIUI CTS-900V; Shantou Institute of Ultrasonic In- (mm) or combined length of the 2 ovaries (mm), or ovary
struments, Shantou, China). The interpolar length of the left length difference (mm). Once only independent variables
and right ovaries, the diameter of the left and right ovaries remained in each model (PWald < 0.10), each of the elimi-
at the deepest point (2008 cohort only), the presence and nated variables was included individually again to test for
diameter of a CL, the diameter of the largest follicle, and confounding. Confounding was considered if inclusion of a
diameter of the left uterus horn near the base (UD) were variable changed the coefficient of one of the independent
recorded. Finally, internal vertical diameter (VD) and predictors by more than 15%.
transverse diameter (TD) of the pelvis were measured by Finally, independent prebreeding examination pre-
transrectal placement of a caliper type pelvimeter (Rice dictors of anestrus and pregnancy failure were combined
 D.E. Holm et al. / Theriogenology 85 (2016) 495–501 497

Table 1
Prebreeding measures and reproductive outcomes per year of birth and per sampling day

Variable Sampling day

Day 7 (n ¼ 259) Day 27 (n ¼ 134) Day 24 (n ¼ 95)

Born 2007 Born 2008 (n ¼ 229)

Age at examination (d)d 401  31a (311–449) 383  27b (308–453)
Age at the onset of breeding (d)d 407  31a (317–455) 408  27a (331–479)
Prebreeding body weight (kg)d 292  36a (195–392) 272  34b (184–349)
Vertical pelvis diameter (cm)d 12.9  0.9a (11–16) 12.6  1.1b (8–15)
Transverse pelvis diameter (cm)d 11.0  0.9a (8.5–13) 10.6  1.0b (8–13)
Largest follicle diameter (mm)d 10.7  2.8a (4–18) 11.1  2.4a (4–17)
Proportion with CL 101/259 (39%)a 56/229 (24%)b
CL diameter (mm)d 21.9  4.5a (11–30) 20.7  4.4a (11–30)
Left ovary interpolar length (mm)d 24.3  5.7a (13–43) 23.1  4.7b (12–36)
Left ovary diameter (mm)d N/De 14.0  4.2 (5–40)
Right ovary interpolar length (mm)d 25.8  5.6a (13–42) 24.1  5.1b (14–43) 26.5  6.0a (11–42)
Right ovary diameter (mm)d N/D 14.0  4.2a (8–30) 15.1  4.1b (8–29)
Left uterus horn diameter (mm)d 15.3  2.6a (10–24) 12.2  1.9b (7–17) 11.7  1.5c (8–15)
Proportion with pregnancy failure 56/258 (22%)a 38/219 (17%)a
Proportion with anestrus 51/259 (20%)a 50/229 (22%)a
a,b,c
Means or proportions in rows with differing superscripts differ significantly (P < 0.05).
d
Mean  standard deviation (minimum and maximum).
e
Not done.

into different prognostic models to estimate which models proportions were similar between birth cohorts (P ¼ 0.87).
provided the best predictions of the outcomes. Areas under Pelvis and ultrasound measures of the reproductive tract
the receiver operating characteristic (ROC) curves are reported in Table 1. The UD differed between the 2 birth
(ROC-AUC) of prognostic models for anestrus and preg- cohorts (Table 1). Furthermore, in the 2008 cohort, sam-
nancy failure were compared using the algorithm of pling day was associated with right ovary length and
DeLong et al. [32]. diameter and with UD (P < 0.05, Table 1). The left ovaries
Data analysis was done using NCSS 2007 (NCSS, Kays- had shorter mean interpolar length than that of the right
ville, UT, USA) and STATA 11.1 (StataCorp, TX, USA). (23.7 and 25.5 mm, respectively, P < 0.01), but the mean
diameter of the left and right ovaries for heifers born in
3. Results 2008 did not differ (14.0 and 14.4 mm, respectively,
P ¼ 0.26).
The age (mean  standard deviation) of heifers at pre- Age, BCS, length of the longest ovary, diameter of the CL
breeding weighing (384  28.8 days) and at the onset of and UD at the time of examination, and prebreeding BW
breeding (407  28.7 days) were similar for the 2 birth were all positively correlated with each other (P < 0.05).
cohorts (P ¼ 0.74 and 0.27, respectively), but heifers born in The diameter of the largest follicle was positively correlated
2007 were examined at an older age than those born in with the length of the longest ovary and the length of the
2008 (P < 0.01, Table 1). Heifers born in 2007 were shortest ovary (P < 0.05). Reproductive tract score was
significantly heavier prebreeding than those born in 2008 most markedly associated with the length of the longest
(P < 0.01, Table 1) and BCS (median, interquartile range ovary, the length of the shortest ovary, and the absence of a
[IQR]) was also higher in the 2007 cohort (6 [5–6] and 5 [5– CL (Table 2). It was also associated with the diameter of the
6], respectively, P < 0.01). More heifers were in RTS cate- CL and the diameter of the largest follicle, less so with UD
gories 4 or 5 (247 of 488) compared to categories 1 or 2 and the absence of a follicle 8 mm but not associated with
(102 of 488) and RTS 3 (139 of 488; P < 0.01), and the the absence of a follicle 13 mm (Table 2).

Table 2
Different ultrasonographic measures of the reproductive tract per reproductive tract score (RTS) category.

Ultrasound variable RTS 1 (n ¼ 15) RTS 2 (n ¼ 87) RTS 3 (n ¼ 139) RTS 4 (n ¼ 120) RTS 5 (n ¼ 127)
Longest ovary length (mm)f 20.7  4.3a 23.0  3.0b 25.6  3.2c 28.4  3.5d 33.0  4.5e
Shortest ovary length (mm)f 16.8  2.9a 19.1  2.6b 21.3  3.2c 22.8  3.6d 23.0  4.0d
Absence of a CLg 15/15 (100%)a,b 87/87 (100%)a 127/138 (92%)b 82/120 (68%)c 19/127 (15%)d
CL diameter (mm)f d d 18.0  4.7a 18.9  3.9a 22.7  4.1b
Largest follicle diameter <8 mmg 5/15 (33%)a 12/87 (14%)a,b 16/138 (12%)b 8/120 (7%)b 12/127 (9%)b
Largest follicle diameter <13 mmg 14/15 (93%)a 69/87 (79%)a 105/138 (76%)a 82/120 (68%)a 93/127 (73%)a
Largest follicle diameter (mm)f 8.7  3.3a 10.4  2.5b 10.7  2.5b 11.4  2.5c 11.1  2.7b,c
Uterus horn diameter (mm)f 12.7  3.0a 13.4  2.6a 13.6  2.9a 14.0  2.9a,b 14.2  2.6b
a,b,c,d,e
Values in rows with different superscripts differ significantly (P < 0.05).
f
Mean  standard deviation.
g
Proportion of the total number of heifers in each RTS category (%).
498 D.E. Holm et al. / Theriogenology 85 (2016) 495–501

Table 3
Summary of the multiple logistic or linear regression models of selected prebreeding measures.

Predictor Multivariable model outcome

Presence of a CLa Largest follicle Uterus diameter Longest ovary Pelvis area (cm2)b
13 mma (mm)b length (mm)b
Dam parity >1 (vs. 1) # # # # 4.15 (7.63, 0.66)
Wean weight (10 kg units) 1.16 (1.08, 1.25) # # # 0.89 (0.05, 1.74)
Prebreeding body weight (10 kg units) # 1.08 (1.02, 1.14) 0.06 (0.01, 0.13) # 2.35 (1.61, 3.09)
Age at examination (wk) 1.13 (1.06, 1.20) # 0.07 (0.01, 0.12) 0.09 (0.00, 0.18) 0.58 (0.16, 1.00)
BCS at examination 6 # # # # 3.25 (5.92, 0.59)
Presence of a CL N/a N/a 0.58 (0.14, 1.01) 6.38 (5.56, 7.21) 7.64 (5.00, 10.30)
Largest follicle diameter (mm) N/a N/a # 0.31 (0.16, 0.45) #
Year of birth 2007 1.55 (1.00, 2.40) # 3.13 (2.72, 3.55) # 3.68 (0.78, 6.59)

Abbreviations: BCS, body condition score; N/a, not analyzed; #, not an independent predictor (P > 0.10).
a
Odds ratios (95% confidence interval [CI]) of independent predictors (P < 0.10) in logistic regression models.
b
Regression coefficients (95% CI) of independent predictors (P < 0.10) in multiple regression models.

In the multivariate models, weaning weight and age at 4. Discussion

examination were independently associated with presence
of a CL, whereas only prebreeding BW was independently In this study, ultrasonographic measures of the repro-
associated with presence of a follicle 13 mm (Table 3). ductive tract and pelvis measures were compared as pre-
Prebreeding BW, age at examination, and presence of a CL breeding predictors of pregnancy failure and anestrus in
were independently associated with UD, and age at ex- seasonally bred beef heifers, and this information was used
amination, presence of a CL, and largest follicle diameter to determine if the current RTS system can be optimized.
were independently associated with the length of the Despite good heritability of age at puberty [11,33]
longest ovary (Table 3). Dam parity 1 (vs. > 1), weaning (h2 ¼ 0.52 and 0.43, respectively), environmental factors
weight, prebreeding BW, age at examination, BCS, and from fetal development to puberty can influence the
presence of a CL were all independently associated with PA phenotypic expression of reproductive potential [5,11,19],
(Table 3). which may have led to different levels of pubertal develop-
Body condition score, absence of a CL, largest follicle less ment achieved by the MSD in the 2 birth cohorts in this study.
than 13 mm, RUD, and RPA were the prebreeding exami- However, the anestrus and pregnancy failure proportions
nation variables that remained in multivariable models of were not different, and we assumed that considering year
pregnancy failure and anestrus (P < 0.1, Table 4). The of birth as a potential covariate in models would adequately
prognostic model for anestrus using the 3 remaining pre- adjust for any differences between the years.
breeding ultrasonographic measures of the reproductive The difference in UD between the 2 birth cohorts in this
tract in combination with RPA (US þ RPA model) yielded an study may have occurred as a result of a true biological
ROC-AUC of 0.81 (Table 5, Fig. 1). The US þ RPA model and difference in the younger group of heifers in 2008 [19] or
the model combining RTS with RPA (RTS þ RPA model) may have been a systematic measuring error that occurred
provided more accurate predictions of pregnancy failure during sampling [34]. The relatively small UD reported in
and anestrus than using RTS alone (P < 0.05, Table 6). The the 2008 cohort of the present study compared to the 2007
RPA model and the model using ultrasonographic absence cohort and previous reports [3,19] indicates that a sys-
of a CL, absence of a follicle 13 mm, and RUD (US model) tematic error was more likely to have occurred in the 2008
tended to predict anestrus better than RTS (P ¼ 0.09 and cohort than in the 2007 cohort. The possibility of such an
0.06, respectively). The RPA model tended to predict error to occur justifies the use of a rescaled value of the raw
pregnancy failure better than RTS (P ¼ 0.09). data in analyses, and we assumed that the relative uterus
diameter within birth cohort in the present study, rather
than the absolute diameter, provided a better indication of
Table 4
Summary of logistic regression models for pregnancy failure and anestrus
in the 2 birth cohorts combined.
Table 5
Independent predictor Outcome: Outcome: Logistic regression model for prediction of anestrus using independent
pregnancy failurea anestrusa transrectal ultrasound measures of the reproductive tract in combination
CL absent 1.69 (0.94–3.05) 6.13 (2.32–16.21) with pelvis area (n ¼ 488).a
Largest follicle <13 mm 2.07 (1.10–3.90) 2.13 (1.09–4.16)
Rescaled uterus diameter 0.43 (0.18–1.02) 0.39 (0.15–1.01) Predictor Coefficient OR 95% CI P
Rescaled pelvis area # 0.15 (0.05–0.48) Constant 2.13 0.12 0.03 0.41 <0.01
Age at onset of breeding (wk) 0.88 (0.83–0.93) 0.90 (0.84–0.96) CL absent 1.97 7.15 2.75 18.57 <0.01
BCS <6 # 2.90 (1.66–5.08) Largest follicle < 13 mm 0.81 2.25 1.19 4.27 0.01
Rescaled uterus diameter 0.96 0.38 0.16 0.94 0.04
Abbreviations: BCS, body condition score; #, not an independent predictor
Rescaled pelvis area 2.77 0.06 0.02 0.18 <0.01
(P > 0.10).
a
Odds ratios (95% confidence interval) of independent predictors Abbreviations: CI, confidence interval; OR, odds ratio.
a
(P < 0.10). Area under the receiver operating characteristic curve ¼ 0.81.
 D.E. Holm et al. / Theriogenology 85 (2016) 495–501 499

1.00 Age at examination was associated with UD and PA in-

dependent of BW, confirming that the development of the
reproductive system is a function of both age and BW and
that age and BW when puberty is reached vary between
0.75

animals, even in a uniform group such as the study popu-

lation [5]. The age range of the study population fell in or
Sensitivity
0.50

just after the second phase of rapid development of the

reproductive tract [15,19], and as such, a lot of variance
could be expected because of the proximity to puberty. The
0.25

results of this study may therefore not necessarily be

extrapolated to heifers in other age ranges.
None of the ovary length variables were independently
0.00

associated with reproductive outcomes in this study. How-

0.00 0.25 0.50 0.75 1.00
1 - Specificity ever, 2 significantly independent predictors of reproductive
Area under ROC curve = 0.8152 outcomes, largest follicle 13 mm and the presence of a CL
at the time of examination, were both also independent
Fig. 1. Receiver operating characteristic (ROC) curve for anestrus using the predictors of ovary length, and we conclude that the effect
model based on the ultrasonographic absence of a CL, absence of a follicle of ovary length on reproductive outcome is confounded by
13 mm, rescaled uterus diameter, and rescaled pelvis area (Table 5).
the presence of ovarian structures. Cushman et al. [13] and
Ireland et al. [14] suggested that the size of the ovaries may
give a reflection of antral follicle count, which is associated
the relative pubertal development stage of an animal
with fertility in young adult cattle. Antral follicle count was
within a group. We assumed the same for PA data.
not considered as an input variable in the present study;
The difference in length between the left and right
however, we assumed that either the longest, the shortest,
ovaries could either have been a true biological difference
or the combined ovary length would provide a reflection of
[35] or it could also have been a systematic measuring error
antral follicle count after adjustment for the size of the
due to the operator using the same hand and the alignment
largest follicle and the size or presence of a CL.
of the ultrasound probe being different on the left and right
The diameter of the largest follicle was not correlated
ovaries [27]. Honaramooz et al. [19] could not demonstrate
with UD as was the case in the study of Honaramooz et al.
a difference in ultrasonographic size between the left and
[19] and was only independently associated with the length
right ovaries; however, the largest follicle on the right side
of the longest ovary, both which appear in this study to be
was 1 mm larger than on the left side in their data. The fact
confounded in their prognostic value for reproductive out-
that there was a numerical difference in ovary diameter
comes. After testing several cutoff points to dichotomize the
between the 2 sides in the 2008 cohort, that was not sig-
diameter of the largest follicle, less than 13 mm was the only
nificant, is not useful to support either of the 2 hypotheses.
predictor of anestrus and pregnancy failure in this study,
It may be that an adjustment for the side of the largest
which is in agreement with the observation by Honaramooz
ovary may improve the ability of the length of the longest
et al. [19] that the maximum follicle size increases before
ovary to predict reproductive outcomes, but because of
puberty from 10 to 12 mm. We conclude that heifers in the
obvious confounding by the size of the largest follicle and
current trial that did not have a CL and also had a largest
the presence of a CL on the predictive ability of ovary
follicle diameter less than 13 mm were at risk of being too
length, this was not investigated any further.
far from puberty at the time of examination to show estrus
during the 50-day AI season or to become pregnant during
the breeding season. None of the other follicle size cutoffs
Table 6 tested had any significant associations in this study, indi-
Areas under the receiver operating characteristic curves (ROC-AUC) of
cating that whether dominance of a follicle has occurred
different predictive models for pregnancy failure and anestrus in the two
birth cohorts separately and combined. (using a cutoff between 7 and 9 mm [36]) did not have
prognostic value for reproductive outcome in our study.
Model Pregnancy failure Anestrus
Previous findings indicating the superior ability of ul-
2007 2008 Combined 2007 2008 Combined trasonography to detect the presence of a CL [10] are sup-
Cohort Cohort data Cohort Cohort data
ported by this study because of the fact that significant
RTSd 0.59a 0.63a 0.60a 0.72a 0.69a 0.71a proportions of heifers with ultrasonographically visible CLs
RPAe 0.62a,b 0.63a 0.62a 0.74a,b 0.78b 0.76a
Ultrasoundf 0.70b,c 0.64a 0.65a,b 0.77a,b,c 0.76a,b 0.76a,b
were assigned RTS 3 or 4. These CLs were smaller than
RTS þ RPA 0.63a,b 0.65a 0.64a,b 0.79b,c 0.80b,c 0.79b,c those of heifers with RTS 5 and were most likely not easily
Ultrasound 0.71c 0.67a 0.68b 0.81c 0.83c 0.81c palpable; however, the tendency of the ultrasound model
þ RPA to have a better predictive value for anestrus and preg-
a,b,c
ROC-AUC values in columns with differing superscripts differ signifi- nancy failure compared to RTS is likely partly as a result of
cantly (P < 0.05). the better sensitivity of ultrasound to detect a CL. In the
d
Reproductive tract score (RTS) categorized as 1 to 2, 3, and 4 to 5, with present study, the absence of a CL predicted not only
4 to 5 as reference value.
e
Rescaled pelvis area.
anestrus but also pregnancy failure. Keeping in mind that
f
Ultrasonographic absence of a CL, absence of a follicle 13 mm, and the total breeding season length was 90 days, this can be
rescaled uterus diameter. partly explained by the fact that the first few ovulations
500 D.E. Holm et al. / Theriogenology 85 (2016) 495–501

after puberty have lower fertility than later ovulations anestrus and pregnancy failure. We suggest that emphasis
[37,38], which will further decrease the ability of heifers should be placed on different predictors depending on the
that reach puberty after the MSD to become pregnant age of heifers at the time of examination or depending on
during a restricted breeding season. the proportion of heifers with CLs at the time of examina-
Although not validated against estrus or pregnancy tion. When heifers are examined long before the MSD or
outcomes during the breeding period after examination, when only a small proportion of heifers have CLs, more
Archbold et al. [6] estimated the sensitivity of ultrasonog- emphasis should be placed on the relative diameter of the
raphy to determine pubertal status to be reduced during uterus horn and the relative PA, whereas when examina-
proestrus and metestrus, because of the relatively poor tion is done shortly before the MSD or when a larger pro-
ability to visualize the regressing corpus albicans and the portion of heifers have CLs, more emphasis should be
corpus hemoragicum, respectively. We therefore assume placed on the absence of a CL and the absence of a follicle
that the reason why the absence of a follicle 13 mm 13 mm diameter. Further research is needed, possibly
remained an independent predictor in our models was using Bayesian modeling, to establish if different prog-
either that some heifers that were pubertal at the time of nostic models should be applied based on herd status.
examination had their first estrus in the few days after the
examination or that in postpubertal heifers, a CL was not 4.1. Conclusions
detected because of stage of the estrus cycle. Because of the
fact that some heifers may have been at stages of the Transrectal ultrasonography of the reproductive tracts
follicular wave before divergence of the dominant follicle at of beef heifers can provide prognostic models of pregnancy
the time of examination [36], the absence of a follicle failure because of its association with anestrus during a
13 mm cannot completely rule out cyclicity but improves restricted breeding season. The ultrasonographic measures
the predictive ability when a CL is not present. that remained independent predictors of pregnancy failure
Because of the inaccuracy of transrectal palpation rela- and anestrus were the absence of a CL, absence of a follicle
tive to ultrasonography to detect a CL, to distinguish be- 13 mm, and relatively smaller uterus horn diameter.
tween follicles less than 13 mm and 13 mm or greater, and Relatively smaller PA can either replace or add value to
to estimate the uterus horn diameter, transrectal ultraso- reproductive tract scoring by transrectal palpation or ul-
nography tended to provide better prognostic models for trasonography as predictor of poor reproductive perfor-
reproductive failure than the current palpation model of mance in restricted bred beef heifers.
Andersen et al. [3]. However, the accuracy of RTS by
palpation may be improved by putting more emphasis on Acknowledgments
the presence of a CL, the size of the largest follicle, and the
diameter of the uterus horn and less emphasis on the ab- The authors thank Roger and Lynette Wood of Johan-
solute size of the 2 ovaries. Our data confirm that the nesburg Water’s Northern Farm for providing the animals,
operator assigning the RTS scores weighed the size of the facilities, and farm data, and Reinette van Reenen and Daleen
ovaries relatively heavily in the scoring system, but this Anderson for assistance with data capture. This research was
study further indicated that the size of the ovaries after performed under protocol no V089/13, as approved by the
adjusting for ultrasonographically visible structures on the Animal Ethics Committee of the University of Pretoria.
ovaries was not an independent predictor of reproductive
outcome and should preferably not be emphasized.
References
Pelvis area, a measure previously used only to predict
dystocia in heifers [28,29], had a strong and independent [1] Chenoweth PJ. Cow/calf production principles. In: Chenoweth PJ,
association with reproductive outcome in this study and Sanderson MW, editors. Beef practice; cow-calf production medi-
added significant prognostic value to models based on cine. Ames, Iowa: Blackwell Publishing; 2005. p. 9–28.
[2] Cushman RA, Kill LK, Funston RN, Mousel EM, Perry GA. Heifers
palpation or ultrasonography of the reproductive tract. This calving date positively influences calf weaning weights through six
is in agreement with previous reports of associations be- parturitions. J Anim Sci 2013;91:4486–91.
tween PA and reproductive outcomes [21,22]. Similar to [3] Andersen KJ, LeFever DG, Brinks JS, Odde KG. The use of reproduc-
tive tract scoring in beef heifers. Agri-Practice 1991;12:19–26.
uterus diameter, PA has the potential to overcome the in- [4] Pence M, Ensley D, Berghaus R, Rossi J, Wilson T, Cannon PT.
accuracy to predict pubertal stage at a single point in time Improving reproductive efficiency through the use of reproductive
caused by different stages of the estrus cycle. We assume tract scoring in a group of beef replacement heifers. Bov Prac 2007;
41:35–40.
that the reason why PA predicted reproductive outcome [5] Holm DE, Thompson PN, Irons PC. The value of reproductive tract
independent of other measures and why it added significant scoring as a predictor of fertility and production outcomes in beef
prognostic value to models predicting reproductive failure heifers. J Anim Sci 2009;87:1934–40.
[6] Archbold H, Shalloo L, Kennedy E, Pierce KM, Buckley F. Influence of
is because it develops gradually over time and is probably age, body weight and body condition score before mating start date
not significantly associated with the daily estrus cycle stage. on the pubertal rate of maiden Holstein-Friesian heifers and im-
The absence of a CL, being the best predictor of repro- plications for subsequent cow performance and profitability. Animal
2012;6:1143–51.
ductive failure, was a particularly good predictor in the
[7] Holm DE, Nielen M, Jorritsma R, Irons PC, Thompson PN. Evaluation
2007 birth cohort, in which case, heifers were examined of pre-breeding reproductive tract scoring as a predictor of long
closer to the MSD and the proportion of heifers with a CL term reproductive performance of beef heifers. Prev Vet Med 2015;
was also higher. In the 2008 cohort, when heifers were 118:56–63.
[8] Rosenkrans KS, Hardin DK. Repeatability and accuracy of repro-
examined more than 3 weeks before the MSD, BCS, uterus ductive tract scoring to determine pubertal status in beef heifers.
diameter, and PA were more important predictors of Theriogenology 2003;59:1087–92.
 D.E. Holm et al. / Theriogenology 85 (2016) 495–501 501

[9] Archbold H, Buckley F, Pierce KM, Begley N, Mee JF. Ultrasonography [22] Singh J, Kumar A, Dhaliwal GS. Pelvic area and scrotal circumference
and progesterone concentration at a single time point for the in relation to libido in breeding bulls. Indian Vet J 2010;87:396–7.
detection of puberty in dairy heifers. Reprod Domest Anim 2012;47: [23] Paterson AG, Venter HAW, Harwin GO. Preweaning growth of
407–11. British, Bos Indicus, Charolais and dual purpose type cattle under
[10] Lean IJ, Abe N, Duggan S, Kingsford N. Within and between observer intensive pasture conditions. S Afr J Anim Sci 1980;10:125–34.
agreement on ultrasonic evaluation of bovine ovarian structures. [24] Schoeman SJ, Jordaan GG. Animal x testing environment interaction
Aust Vet J 1992;69:279–82. on postweaning liveweight gains of young bulls. Aust J Agric Res
[11] Johnston DJ, Barwick SA, Corbet NJ, Fordyce G, Holroyd RG, 1998;49:607–12.
Williams PJ, et al. Genetics of heifer puberty in two tropical beef [25] Holm DE, Thompson PN, Irons PC. The economic effects of an estrus
genotypes in northern Australia and associations with heifer- and synchronization protocol using prostaglandin in beef heifers.
steer-production traits. Anim Prod Sci 2009;49:399–412. Theriogenology 2008;70:1507–15.
[12] Mee JF, Buckley F, Ryan D, Dillon P. Pre-breeding ovaro-uterine ul- [26] Marston TT. Beef cowherd nutrition and management. In:
trasonography and its relationship with first service pregnancy rate in Chenoweth PJ, Sanderson MW, editors. Beef practice; cow-calf pro-
seasonal calving dairy herds. Reprod Domest Anim 2009;44:331–7. duction medicine. Ames, Iowa: Blackwell Publishing; 2005. p. 89–108.
[13] Cushman RA, Allan MF, Kuehn LA, Snelling WM, Cupp AS, [27] DesCôteaux L, Gnemmi G, Colloton J. Ultrasonography of the bovine
Freetly HC. Evaluation of antral follicle count and ovarian female genital tract. Vet Clin North Am Food Anim Pract 2009;25:
morphology in crossbred beef cows: Investigation of influence of 733–52.
stage of the estrous cycle, age, and birth weight. J Anim Sci 2009;87: [28] Deutscher GH. Pelvis measurements: key to reducing incidence of
1971–80. bovine dystocia. Norden News 1988;63:18–25.
[14] Ireland JJ, Smith GW, Scheetz D, Jimenez-Krassel F, Folger JK, [29] Van Donkersgoed J. Pelvimetry. In: Youngquist RS, editor. Current
Ireland JLH, et al. Does size matter in females? an overview of the therapy in large animal theriogenology. Philadelphia: W.B.Saun-
impact of the high variation in the ovarian reserve on ovarian ders; 1997. p. 306–9.
function and fertility, utility of anti-Müllerian hormone as a diag- [30] Stevenson JL, Rodrigues JA, Braga FA, Bitente S, Dalton JC, Santos JEP,
nostic marker for fertility and causes of variation in the ovarian et al. Effect of breeding protocols and reproductive tract score on
reserve in cattle. Reprod Fertil Dev 2011;23:1–14. reproductive performance of dairy heifers and economic outcome of
[15] Desjardin C, Hafs HD. Maturation of bovine female genitalia from breeding programs. J Dairy Sci 2008;91:3424–38.
birth through puberty. J Anim Sci 1969;28:502–7. [31] Dohoo I, Martin W, Stryhn H. Veterinary epidemiologic research.
[16] Day ML, Imakawa K, Garcia-Winder M, Zalesky DD, Schanbacher BD, Charlottetown, Canada: AVC Inc; 2003.
Kittok RJ, et al. Endocrine mechanisms of puberty in heifers: [32] DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas
estradiol negative feedback regulation of luteinizing hormone under two or more correlated receiver operating curves: a
secretion. Biol Reprod 1984;31:332–41. nonparametric approach. Biometrics 1988;44:837–45.
[17] Day ML, Imakawa K, Wolfe PL, Kittok RJ, Kinder JE. Endocrine [33] Brinks JS. Genetic influences on reproductive performance of two-
mechanisms of puberty in heifers. Role of hypothalamo- pituitary year-old beef females. In: Fields MJ, Sand RJ, editors. Factors
estradiol receptors in the negative feedback of estradiol on lutei- affecting calf crop. Boca Raton: CRC Press; 1994. p. 45–53.
nizing hormone secretion. Biol Reprod 1987;37:1054–65. [34] Melendez P, McHale J, Bartolome J, Archbald LF, Donovan GA.
[18] Bergfelt EGM, Kojima FN, Cupp AS, Wehrman ME, Peters KE, Garcia- Uterine involution and fertility of Holstein cows subsequent to early
Winder M, et al. Ovarian follicular development in prepubertal postpartum PGF2a treatment for acute puerperal metritis. J Dairy Sci
heifers is influenced by level of dietary energy intake. Biol Reprod 2004;87:3238–46.
1994;51:1046–50. [35] Petskol PG. Age and functional changes in mammalian ovaries. The
[19] Honaramooz A, Aravindakshan J, Chandolia RK, Beard AP, age morphology of cattle ovaries (abstract). Trud Kirovsk Zooteh –vet
Bartlewski PM, Pierson RA, et al. Ultrasonographic evaluation of the Inst 1941;4:49–74.
pre-pubertal development of the reproductive tract in beef heifers. [36] Wiltbank MC, Gümen A, Sartori R. Physiological classification of
Anim Reprod Sci 2004;80:15–29. anovulatory conditions in cattle. Theriogenology 2002;57:21–52.
[20] Holm DE, Webb EC, Thompson PN. A new application of pelvis area [37] Byerley DJ, Staigmiller RB, Berardinelli JG, Short RE. Pregnancy rates
data as culling tool to aid in the management of dystocia in heifers. J of beef heifers bred either on puberal or third estrus. J Anim Sci
Anim Sci 2014;92:2296–303. 1987;65:645–50.
[21] Ramin AG, Daniel RCW, Fenwick DC, Verrall RG. Pelvic parameters, [38] Rawlings NC, Evans ACO, Honaramooz A, Bartlewski PM. Antral
growth rate, puberty, and their interrelationships in young dairy follicle growth and endocrine changes in prepubertal cattle, sheep
heifers. Reprod Domest Anim 1995;30:117–23. and goats. Anim Reprod Sci 2003;78:259–70.