History of Litter Size Selection
University of Nebraska, Lincoln
Litter size in the national pig herd has increased at the rate of approximately .052 pigs per year from 1980 to 2000 (figure 1). The total change is approximately 13%. However, improvements have also occurred in other economic traits. Therefore, economic value of litter size is probably as great today as it was in the past.
Economic value of a trait for within-line selection depends on how the line is used in crossbreeding systems. Dr. C. Smith, in a paper published in 1983, wrote “animal geneticists have been rather cavalier in their use of economic values in index selection.” He explained that much more emphasis has been placed on parameter estimation and prediction of breeding values than on determining relative economic values of traits based on life-cycle costs of production and value of product produced. In that paper, Smith and co-workers (Smith et al., 1983) used a deterministic bio-economic model of life cycle pig production to examine relative economic values of traits. Using production parameters and economic values of the early 1980s, they determined that the economic value of traits varies with the value of lean and fat in the carcass. When the criterion of evaluation was cost ($) of producing lean tissue only, in which case fat has no value, then the economic value of litter size for general purpose lines was $11.60 per phenotypic standard deviation (approximately $4.64/pig). However, the value of litter size was $31.40 and $17.60 per phenotypic standard deviation ($12.56/pig and $7.04/pig) in maternal granddam and maternal grandsire lines, respectively. When the criterion of evaluation was cost to produce live weight, in which case carcass fat and lean have equal value, relative economic values were approximately 50% less.
Economic values of litter size published in past NSIF Guidelines have been around the value of $12 per pig. The value in the current edition is $13.50. These values are calculated for general use of a line as it would be used in a rotation crossing system and are somewhat greater than those obtained from the bio-economic models of Smith et al., 1983.
The economic value of traits for life cycle production efficiency should be based on a criterion in which both lean and fat have value, but heavily weighted toward producing lean tissue. Because of genetic selection against fatness, today’s pigs are considerably leaner than those of 1980. Therefore, the value of further decreased selection against fat has decreased relative to that of other traits. In fact, several lines are considered to be lean enough for today’s market. Certain buying grids (e.g., Hormel foods, Fremont NE; Carcass Lean Value Report) actually penalize very lean carcasses (below .51 in backfat). In lines for which no premium is realized for further increases in carcass leanness, there has been a shift in relative economic value of traits toward those that decrease costs of production (growth rate, feed efficiency and reproduction) and meat quality traits that increase value.
Litter size has economic value only because with each additional pig costs associated with reproduction are spread over more output, thus reducing per unit production costs. Thus, we might conclude that the economic value of increasing litter size in maternal lines is greater today than in the past. However, to investigate this question, we must consider how those costs of production, which can be affected by increased reproduction rate, have changed relative to total costs of production.
Figure 2 illustrates changes in whole-herd feed efficiency of the nation's pig crop during the last 20 years. It has decreased at the rate of approximately -.024 lbs. of feed per lb. of live weight marketed. Most of this change presumably has occurred from improved feed efficiency of growing pigs, although it includes advantages realized from increased reproductive rate.
Figure 3 is an attempt to determine whether feed costs relative to total costs have changed. Total costs of production were available from 1980, but the proportion of these costs due to feed were available beginning in 1988. Values above the feed cost data points are the percentages of total costs due to non-feed costs. We might expect that proportion to increase because of improvements in feed efficiency of growing pigs. This of course assumes that the cost of feed relative to other variable and fixed inputs has remained relatively stable. However, from 1988 to 1995, non-feed costs were 39% of total costs. It was only in 1999 during a period of low grain prices that the proportion of total costs due to non-feed increased sharply.
My rather off hand approach to comparing relative economic values is exactly of the nature that Dr. Smith made reference to in his earlier paper. However, it does give us some comfort in concluding that the value of increasing litter size relative to other traits that reduce costs of production has not changed greatly in the last 15-20 years. In general, economic return from selection for traits that decrease costs of production relative to those that increase carcass value has probably increased in most lines of pigs.
Early experiments on selection for litter size at birth were only marginally successful. One of the first was conducted in France in a Large White population. Selection was based on mean litter size in the first two parities. Olivier and Bolet (1981) reported results after 10 generations. Ovulation rate and embryonic survival at 30 d of gestation had increased, but litter size at birth was not consistent with the response observed early in gestation. A later analysis (Bolet et al., 1989) found a response of .024 ± .077 pigs per generation (year) after 11 generations. The small response was attributed to lower than expected heritability and to failure to achieve desired selection intensity. Other early experiments of short duration and in lines with small population sizes also reported no significant genetic or phenotypic trends from direct litter size selection (Rutledge et al., 1980; Vangen, 1981).
Results like these generally discouraged swine producers from including litter size in selection objectives. It was known to have low heritability, responses to direct selection in experimental herds were discouraging, and real economic improvements could be realized from selection for growth and carcass traits. Thus, there appeared to be little real selection emphasis on litter size in breeder herds until the mid to late 1980s. During this time, a number of papers were published showing that litter size could be increased at the rate of as much as .27 to .51 pigs per year if selection was on EBV from adequate pedigree and family data (Haley et al., 1988; Avalos and Smith, 1987). It was also during this time that we began to consider models of litter size based on ovulation rate, embryonic survival and uterine capacity (Bennett and Leymaster, 1989; Johnson et al., 1999). Theoretical work indicated that response depended on mean values of these component traits and that response could be increased by selection for optimum weights on the components.
Reproductive rate in the nations pig herd has increased rather dramatically in the last 20 years (Figure 4). Increases of .0085 in litters per sow per year and .052 pigs per litter caused an increase of .165 pigs per sow per year, a total change over the last 20 years of approximately 25%. This change is due to several factors that cannot be separated. Earlier weaning, improved sow management, and continuous breeding/farrowing systems have most likely explained the increased number of litters per sow per year.
Improved diets, management and production environments have undoubtedly increased litter size, although the exact effects of these improvements are difficult to document. Genetic changes have also occurred. Specific crossing systems are now standard practice. More efficient use of heterosis and breed resources, especially use of lines superior in maternal performance as F1 females, might explain a large part of the genetic improvement in litter size. Responses from within-line/breed selection have also contributed to increased litter size in the nation's pig herd, but this source of change is probably less than that due to other genetic and environmental changes.
Real genetic change in litter size in nucleus populations is occurring. Improvements began to occur with introduction of BLUP procedures to estimate genetic values. For example responses in the national Yorkshire and Landrace populations are estimated to be approximately .25 pigs per litter from 1991 to 1999 (Figure 5). Similar genetic changes have probably occurred in most maternal populations.
Figure 5. Estimated litter size genetic trends in Yorkshire and Landrace (Source: http://www.ansc.purdue.edu/users/dlofgren/stages/)
Selection experiments to increase litter size were initiated in Nebraska in 1981. Today, five genetic lines exist, three selection lines and two control lines, all originating from the same base population. Responses in litter size are illustrated in Figure 6. All data are for first parity gilts.
The base population was a composite of 50% Large White and 50% Landrace. Breeds were crossed in 1979, and random mating was used until 1981, when selection was initiated. Generation interval was one year. Generation numbers in figure 6 correspond with year of birth of dams with year 0 being those born in 1980.
Line I was selected on an index of ovulation rate and embryonic survival for 11 generations, and then it was selected for increased litter size at birth. Line C is its control line. Response in line I was .19 pigs per generation. Total response after 19 generations is approximately 3.6 pigs per litter. In latter generations, litter size has averaged between 13 and 14 pigs per litter. However, this response was only 40-50% of the increase in ovulation rate. Measurement of embryonic survival, which was done at 50 days of gestation, was an inadequate measure of uterine capacity. Although litter size in line I increased greatly, so did fetal losses late in gestation.
To address the problem of fetal losses in late gestation, selection for ovulation rate and uterine capacity was implemented. Three lines were formed from lines I and C after 8 generations of selection. They were established by re-mating selected generation 8 parents to produce second parity litters. Two of these lines (IOL, derived from line I, and COL, derived from line C) were selected in two stages for ovulation rate and litter size at birth. The third line (C2) also derived from line C was a randomly selected control.
Gilts in lines IOL and COL were first selected for ovulation rate using laporatomy to count number of corpora lutea (CL). At farrowing, gilts from one-half of the litters with greatest number of pigs were selected, ovulation rate was recorded and the 50% with the greatest number of CL were mated. The next cycle of selection was initiated among the litters when ovulation rate was again recorded in all gilts from the largest litters. The hypothesis was that ovulation rate of gilts selected on CL number was greater than their uterine capacity so litter size at birth measured uterine capacity.
Litter size in lines IOL and COL increased at the rate of .29 pigs per generation, approximately 50% greater than the response in line I. Also, rate of response was the same in lines IOL and COL, even though they had greatly different ovulation rates, uterine capacities, and litter size in generation 8 (generation 0 of 2-stage selection).
The ovulation rate/uterine capacity model of litter size appears to be very effective. Currently, it is a difficult model to implement in breeder herds because ovulation rate can be measured accurately only by laparotomy to count CL. It would be an easy model to implement if an accurate, non-invasive procedure to estimate CL were found. An interesting result is that heritability of litter size in all selection lines is approximately .17, higher than the value of .10 reported for most populations.
In summary, litter size continues to be an important economic trait. In 20 years, significant improvements have occurred in the nation's pig herd. Implementation of BLUP selection in nucleus populations has been effective. Selection response can be enhanced from selection on models of ovulation rate and uterine capacity.
Avalos, E., and C. Smith. 1987. Genetic improvement of litter size in pigs. Anim. Prod. 44:153.
Bennett, G., and K. A. Leymaster. 1989. Integration of ovulation rate, potential embryonic viability and uterine capacity into a model of litter size in swine. J. Anim. Sci. 67:1230.
Bolet, G., L. Ollivier, and P. Dando. 1989. Selection sur la prolificite chez le porc. I. Resultats d’une experience de selection sur onze generations. Genet. Sel. Evol. 21:93.
Haley, C. S., E. Avalos, and C. Smith. 1988. Selection for litter size in the pig. Anim. Breed Abstr. 56:317.
Johnson, R. K., M. K. Nielsen, and D. S. Casey. 1999. Responses in ovulation rate, embryonic survival and litter traits in swine from 14 generations of selection to increase litter size. J. Anim. Sci. 77:541.
Ollivier, L., and G. Bolet. 1981. La selection sus la prolificite chez le porc: Resultats d’une experience de selection sur dix generations. J. Rech. Porcine en France. 13:261.
Rutledge, J. J., 1981. Fraternity size and swine production. I. Effect on fecundity of gilts. J. Anim. Sci. 51:868.
Smith, C., G. E. Dickerson, M. W. Tess, and G. L.Bennett. 1983. Expected relative responses to selection for alternative measures of life cycle economic efficiency of pork production. J. Anim. Sci. 56:1306.
Vangen, O. 1981. Problems and possibilities for selection for fecundity in multiparous species. Pig News and Information 2:257.