INTRODUCTION
Variation in weaning to estrus contributes to variation
in traits such as weaning to successful service, nonproductive
sow days, and farrowing intervals. These traits are components
of variation in litters per sow per year and number of live pigs
per sow per year, traits often used to measure the reproductive
efficiency of herds.
Many factors, including weaning age, nutrition, parity
of sow, and type of housing, affect the interval from weaning
to first observed estrus in sows. Variation is quite great for
sows of all parities, but is most obvious in sows after weaning
their first litter. Much of the variation is due to management
and environmental effects; but genetic variation also exists because
mean interval from weaning to first service differs among breeds
and breed crosses.
The purpose of this report is to summarize research
on genetics of rebreeding intervals. Distributions of weaning
to estrus and weaning to farrowing intervals, and their heritability
and relation to production traits are presented. Procedures
to include weaning to first service interval in genetic evaluation
programs are discussed.
DISTRIBUTION OF WEANING TO ESTRUS
INTERVAL
ten Napel et al. (1995) determined that variation
in the interval from weaning to first observed estrus is a mixture
of two types of distributions. It is illustrated in Figure 1.
Part of the distribution is normal and has the bellshaped curve
typical of most traits. But the distribution has a long right
hand tail caused by sows with prolonged intervals. Some sows
have a mildly prolonged interval and cannot be separated from
those sows in the right hand side of the normal distribution.
Other sows have a distinctly prolonged interval that extends
far to the right of the normal curve. This tail is best described
with an exponential distribution. The observed distribution is
the sum of the two components, each weighted by their frequency.
The distribution in Figure 1 is for the control line
in an experiment done in The Netherlands in which selection was
for decreased weaning to estrus interval in gilts after weaning
their first litter between 32 and 38 days of lactation. There
were 31.4 % of the sows in the normal part of the distribution
with a mean weaning to estrus interval of 5.6 days and a standard
deviation of .97 days. Approximately 95 % of sows with normal
intervals expressed estrus between 3.6 and 7.6 days after weaning.
However, 68.6 % of the sows were in the exponential distribution
which had a starting point of 5 days and a mean of 36.9 days.
A property of exponential distributions is that the mean and
variance are equal, thus the standard deviation of records in
the exponential part was 6.1 days.
SELECTION FOR REDUCED WEANING
TO ESTRUS INTERVAL
Eight generations of selection for decreased weaning to estrus interval were practiced in the experiment of ten Napel et al. (1995). Almost all selected dams in the select line had weaning to estrus intervals of 4 to 6 days. Control sows were selected randomly and mated when they expressed estrus. The response is summarized in Table 1. The decrease in weaning to estrus interval in the select line occurred entirely from a decrease in the proportion of sows with prolonged intervals. Selection did not change the parameters of the distribution. The average and variance for sows in the normal and exponential parts were nearly identical for both lines, as were the starting points of the exponential distribution. The parameters in Table 1 are averages over all 8 generations of selection, and thus reflect differences after 4 generations. Total response in the select line was predicted to be an increase to 47.2% of sows in the normal part of the distribution after 8 generations.
Table 1. Average distribution of weaning to estrus interval during 8 generation of selection (after ten Napel et al. (1995)  
Normal distribution  
Fraction  .393  .314 
Mean  5.57  5.63 
Standard deviation  1.08  .971 
Exponential distribution  
Fraction  .607  .686 
Starting day  36.8  36.9 
Mean (= variance)  5  5 
Variation explained  94.9  92.9 
DIFFERENCES AMONG POPULATIONS
IN DISTRIBUTIONS OF REBREEDING PERFORMANCE
Other studies confirm that the distributions of rebreeding
intervals have the general shape of those found by ten Napel.
ten Napel and Johnson (1997) studied the distribution of weaning
to farrowing interval in Large White and Landrace sows in two
farms. A total of 3,243 intervals were analyzed. Litters were
weaned between 17 and 23 days of age.
The weaning to farrowing interval includes the weaning
to first service interval, the interval between services for repeat
matings, and variation in gestation length. Some truncation of
data occurred as sows not expressing estrus within 25 days of
weaning were usually culled. All populations had very similar
distributions with nearly identical parameter estimates (Table
2). Populations differed because of the different proportion
in each part of the distribution.
The proportion of sows with normal intervals was
greater for Large White than Landrace, and was greater for second
parity sows than first parity sows. Subtracting 114 days, a normal
gestation length, from the mean of sows in the normal distribution,
and from the starting point of the exponential distribution, gives
values of approximately 8 days for the normal mean, and 6 days
for the starting point of the exponential distribution. These
values are very similar to those of the distribution of weaning
to estrus interval found by ten Napel et al. (1995).
Table 2. Parameter estimates for distribution of weaning to farrowing intervals after ten Napel and Johnson (1997)  
Farm 1  
Parity 1 




 
Parity 2 




 
LW 




 
LR 




 
Farm 2  
Parity 1 




 
Parity 2 




 
LW 




 
LR 





Greater variation in weaning to farrowing interval than weaning
to estrus interval is due to the addition of intervals between
services due to conception rates being less than one, and the
small amount of variation in gestation length. The distribution
of intervals after first, second and third parities in Figure
2 illustrates that sows of different parities differ only because
of different proportions in the normal and prolonged parts of
the distribution. Means and standard deviations for sows in the
normal part of the distribution were nearly equal for all parities.
Adamec and Johnson (1997) found distributions of weaning to service
and weaning to conception intervals in sows in the Czech National
herd were similar to those described above. Records of 2,896
Landrace and Large White sows in 7 farms were analyzed. Weaning
age averaged 27.7 days, and varied from 15 to 49 days for individual
sows. The mixture distribution explained a high percentage of
the variation in both weaning to service and weaning to conception
intervals. The proportion of sows in the normal part of the distribution
of weaning to conception interval varied from .35 to .69 among
farms. The mean of the normal distribution was 8.7 days with
a standard deviation of 2.76 days. The starting day of the exponential
distribution was 5 days.
HERITABILITIES AND CORRELATIONS
OF REBREEDING INTERVALS
Studies done in the 1970's and 80's found heritability of weaning to estrus interval of approximately .2, but heritabilities of farrowing intervals were lower, approximately .1. Results of more recent studies are very similar (Table 3). The estimate from ten Napel et al. (1995) is a realized heritability from selection; the other estimates are maximum likelihood estimates calculated with relationships among relatives.
Table 3. Heritability estimates of rebreeding performance  
Interval 
 
Weaning to estrus, parity 1 


 
Weaning to farrowing, parity 1, farm1 


 
Weaning to farrowing, parity 2, farm1 


 
Weaning to farrowing, parity 3, farm1 


 
Weaning to farrowing, parity 1, farm2 


 
Weaning to farrowing, parity 2, farm 2 


 
Weaning to farrowing, parity 3, farm2 


 
Weaning to service, across parities 


 
Weaning to conception, across parities 


 
Weaning to farrowing, across parities 



^{a}
1 = ten Napel et al. (1995);
2 = ten Napel & Johnson (1997); 3 = Adamec & Johnson (1997).
The realized heritability of weaning to estrus interval
was .17. Heritabilities of other intervals for first parity sows
were generally less. Heritability of intervals after second and
third parities were greater than intervals after first parity.
Greater environmental variation affecting first parity intervals
is likely the explanation.
Genetic correlations among weaning to farrowing intervals
after first, second, and third parities were .78 in Farm 1 data,
and .74 in Farm 2 data (ten Napel and Johnson , 1997). Genetic
correlations between weaning to estrus intervals in different
parities are expected to be at least this large, and probably
greater. Phenotypic correlations were much lower, ranging from
.01 to .15. Phenotypic correlations are estimates of repeatability
of traits, thus environmental effects common to intervals after
different parities appear to be unimportant.
CORRELATIONS WITH OTHER REPRODUCTION
AND PRODUCTION TRAITS
In a review, ten Napel et al. (1995) hypothesized
that genetic selection for increased daily gain, decreased backfat,
increased litter size, and increased litter weaning weight may
impair normal resumption of cycling activity after weaning. ten
Napel and Johnson (1997) studied these relationships in two farms.
Genetic correlations of weaning to farrowing intervals with average
daily gain ranged from .04 to .21, and those with backfat were
between .07 to .05. Significant differences in estimated breeding
values for daily gain and backfat were found only in parity 1
sows. In general results were consistent with the hypothesis
that selection for faster growth and leaner pigs may increase
the incidence of prolonged rebreeding intervals, although differences
were small.
Adamec and Johnson also found low genetic correlations
of production traits with rebreeding performancce. Genetic correlations
of average daily gain and different intervals were between .06
and .14, those with backfat were between .06 and .17. Rebreeding
performance seems not to be related to production traits or to
have very low correlations with them.
Genetic correlations with number born were also low
and inconsistent across studies, but correlations with litter
weaning weight were consistently positive (approximately .15).
Sows with genetic potential for above average milk production
had some tendency for expressing prolonged weaning to estrus intervals,
although the relationship is quite weak.
Genetic correlations of intervals from weaning to
estrus or weaning to farrowing with production and other reproduction
traits may not be useful in selection. It is likely that correlations
are close to zero for sows with normal intervals, but may be significant
and undesirable for sows with prolonged intervals. Calculated
correlations are some pooled value of these two relationships
and may differ across populations depending on the incidence of
prolonged intervals.
IMPLEMENTATION
The expected result of selection to improve rebreeding
performance is a reduction in the proportion of sows with prolonged
rebreeding intervals, those in the exponential part of the distribution.
Selection is not expected to decrease the mean of sows in the
normal part. Therefore, whether to implement selection for improved
rebreeding performance depends on the current distribution of
intervals of sows in the population. If a high percentage are
returning to estrus within normal intervals, selection will have
little effect. If prolonged intervals are prevalent, selection
for shorter intervals is expected to cause improvement.
Selection to improve rebreeding performance can be
implemented in several ways. A very simple procedure is to retain
all replacement boars and gilts from sows that rebreed within
normal intervals and to cull sows with prolonged intervals. When
sows were weaned at 17 days lactation or greater, intervals of
7 days or more to first observed estrus were nearly always in
the exponential part of the distribution. Therefore, retaining
replacements only from sows that expressed estrus within 78 days
following weaning is expected to be effective.
A somewhat more effective procedure is to select
on estimated breeding values for rebreeding performance. The
first step is to define the trait to be recorded. Either weaning
to first observed estrus interval, weaning to farrowing interval,
or farrowing interval are recommended as they can be systematically
recorded in databases. Repeat intervals can be treated as the
same trait with repeatability equal to or slightly greater than
heritability, as common environmental effects across intervals
seem not to be important. Heritabilities of .17 for weaning to
estrus interval, and .10 for other intervals are recommended.
Alternatively, intervals after weaning each litter can be considered
a different trait with genetic correlations among intervals of
approximately .75. Singletrait analyses are appropriate as
genetic correlations with other production traits seem to be very
low.
Once breeding values are estimated, they are weighted
along with those for other traits according to their relative
economic value and combined into an index. Value of reduction
in rebreeding intervals will vary across populations. Value may
be low in populations with a high proportion of sows with normal
intervals, but will be greater in populations with greater incidence
of sows with prolonged intervals. Determining economic value
of rebreeding intervals was not an objective of this paper. But
obtaining estimates of these values is as important as estimating
breeding values and should be done before selection is implemented.
A last word of caution. Most data in the literature
are in sows with litters weaned at 17 days or greater. Earlier
weaning is becoming more common. Rebreeding intervals when 10
to 14 day weaning is practiced have not been studied. Whether
parameters reported here apply to that management practice is
not known.
BIBLIOGRAPHY
ten Napel et al. 1995. A biological approach to
examine genetic variation in weaningtooestrus interval in firstlitter
sows. Livest. Prod. Sci. 41:81.
ten Napel et al. 1995. Genetics of the interval
from weaning to estrus in first parity sows: distribution of data,
direct response of selection, and heritability. J. Anim. Sci.
73:2193.
ten Napel and Johnson. 1997. Genetic relationships
among production traits and rebreeding performance. J. Anim. Sci.
75:51.
Adamec and Johnson. 1997. Genetic analysis of rebreeding
intervals, litter traits, and production traits in sows of the
National Czech nucleus. Livest. Prod. Sci. 48: 13.