The quality and quantity of boar semen may be the forgotten denominator in the march to increase litter size and maximize genetic potential of terminal market hogs.

Artificial insemination has had an undeniable impact on the swine industry. It is commonly held that the efficiency of the breeding herd is highly dependent on high pregnancy rates and the survivability and performance of the offspring.

Why then do we pay so little attention to the quality of boar semen used for those inseminations, asked Michael Dyck during the Optimizing Sow Breeding Management and Litter Outcomes workshop held in conjunction with the Leman Swine Conference last fall in St. Paul, MN.

The near-universal acceptance of artificial insemination (AI) has reduced labor in the breeding herd, limited the presence of boars to those needed for heat-checking duties, reduced the risk of disease transmission, and created a safer work environment.

In a broader sense, using fewer boars accelerates genetic improvement and produces a more consistent product; plus new and better genetics can be integrated into the national herd more rapidly.

These are all very positive attributes, Dyck admits, but he wonders if we can do better.

“We instinctively know that not all boars and the semen ejaculates from those boars are of equal quality,” states the University of Alberta swine reproductive physiologist. “Given the polygamous structure of swine production, poor-quality boars will affect the reproductive outcome of numerous females, and the use of pooled semen from poorly defined males breaks the link between known genetic value of individual boars and the paternity of progeny produced.”

If greater value is to come from AI, the next logical steps are to reduce the number of sperm per insemination, and to capitalize on the true genetic merit of the superior sires.

Fewer Sperm, Same Outcome

At the outset, Dyck reinforces the genetic impact that a single boar can have. In natural service, a boar can reasonably produce 50 tons of pork a year, he says. With on-farm semen collection and AI, a boar's influence increases ten-fold, so he's capable of producing 500 tons of pork annually. Move the boar to an off-site AI stud and he can produce half again as much — about 765 tons of pork. If the number of sperm needed to produce a litter could be reduced by one-third, a sire could produce 2,295 tons of pork, he stresses.

“Instead of producing 765 tons/boar, can we produce over 2,000 tons/boar? And, assuming we use the best boars, they will produce a better product and the pigs will be easier to raise,” he says.

Dyck says there are three things keeping the industry from making these gains:

Dyck reviewed two sperm traits that affect fertility:

Identifying Sub-Fertile Boars

Conventional semen evaluation generally includes seminal volume, sperm concentration and the percentage of motile and morphologically normal sperm. While some of these parameters are correlated with fertility, some are not good predictors.

In commercial boar studs, it is generally agreed that a boar with good motility (more than 80%) and good morphology (less than 15% abnormal sperm), has a pretty good chance of being relatively fertile, Dyck notes.

“The almost universal use of pooled semen doses in commercial boar studs severely limits the collection of data on relative boar fertility at the production level,” he continues.

Dyck cites recent field data that tracked litter size in sows bred to commercial Landrace boars using single-sire matings (three billion sperm/dose). The study included 31 boars that sired at least 20 litters. Each dot in the Figure 1 scatter graph indicates an average of 20 or more litters. Over 800 single-sire matings are represented.

Figure 1 shows the productivity of the top two-thirds of the boars is very high, averaging over 13 pigs born/litter, topping out at 16 and 18 pigs born/litter. However, when the lower one-third of boars is added to the mix, overall average productivity is reduced (12.26 vs.13.29 pigs/litter).

“If we identified and eliminated the 10 worst boars, we would gain a pig/litter,” Dyck emphasizes. “We talk about managing sows better and how we can make gains in productivity. By managing the sire genetics, we could make the advances we are trying to target in the sow herd.

“There is heritability to capture, too,” he continues. “If we could define and select our boars a little better, we could make the (targeted) gains just on that level alone.

“Moreover, if the genetic merit of the three boars in Figure 1 (averaging over 14 total pigs born) was high, the application of more efficient AI technologies would allow the merits of these ‘elite’ boars to spread across a larger portion of the sows bred. However, with current AI practices, these substantial differences in boar productivity and the link to known progeny produced by individual boars are confounded by the use of pooled semen and the high sperm numbers per AI dose,” he emphasizes.

Lowering Semen Doses

Next, Dyck turned his attention to a study using nine boars known to have acceptable semen characteristics (greater than 80% sperm motility; less than 15% abnormal sperm). Instead of using the typical three billion sperm/dose, the sperm count was cut in half and each boar was mated to at least 45 gilts.

“We expected some of the boars to not do very well and drop out,” he explains. “Surprisingly, the majority of the boars did fine. We really didn't see much difference in pregnancy rate and farrowing rate. But a couple of boars didn't perform quite as well (see Boars R-1 and G-1 in Table 1). When we looked at litter size, those two boars didn't perform as well either. Still, those two boars would have been acceptable in a boar stud. Do you really want them to be part of a (semen) pool?” he asks.

“We know that not all boars are created equal, but when we go below about two billion sperm (per dose), you're no longer compensating for a minority of boars.”

By consistently using three billion sperm per insemination, and by pooling the semen from several boars, sperm numbers of the more fertile boars in the pooled semen dose are effectively being diluted. “Some boars will actually produce very few offspring, which is not a very effective strategy to maximize the impact of certain animals,” he reminds.

Why Pool Boar Semen?

“We pool semen because it seems to make sense,” Dyck continues. “If the semen from one boar is not very good, we want to compensate for it. By the same token, we will not expose that animal. He is kept in the population. Shouldn't we be exposing these animals and getting rid of them?

“The other issue with pooling semen is the averaging effect. We're diluting out the reproductive potential — positive or negative. Therefore, if you mix semen from a less-fertile boar with a highly fertile boar — one boar is going to produce more piglets than the other — but will it be the genetically superior boar?” he challenges.

Dyck sees other problems with pooling semen when optimizing production efficiency is the goal. In a recent preliminary study, two boars that met the normal criteria for acceptable semen quality were compared in single-sire and pooled-semen protocols (Table 2). The boars were designated “red” and “blue,” and two billion sperm/dose were used in all matings. Sows' ovulation rates averaged above 20.

Results in Table 2 show both boars performed well, whether used individually or pooled. Interestingly, the number of live embryos at 30 days post-insemination was essentially the same for pooled semen and the single-sire inseminations using the red boar. However, the blue boar averaged 2.5 more embryos at Day 30 — a 15% difference in fertilization rate and/or embryo survival.

“Notably, the outstanding performance of the most fertile (blue) boar was masked by using pooled semen,” Dyck notes. “A single-sire AI strategy would improve total pigs born by allowing the blue boar to express his true reproductive potential. It seems reasonable to assume that a similar ‘averaging effect’ results from the pooling of semen from the best boars shown in Figure 1. We've got potential for more live embryos, but when we pooled semen, it tended to average or reduce them. It's a small study and a bit anecdotal, but we have done larger studies that demonstrate this tendency as well.”

Proving Paternity

In another study, boars with dramatically different fertility levels helped illustrate the problems of linking progeny with sires when semen is pooled in a commercial herd setting. Again, all boars appeared to have similar, acceptable semen characteristics.

Initially, Boar A and Boar B were used in single-sire inseminations. Major differences are evident in farrowing rate and litter size born (Table 3).

Clearly, Boar B's 100% farrowing rate and 12.04 pigs/litter average were superior to Boar A's 45% farrowing rate and 9.22 pigs/litter average. “You can see the difference in these two boars, even though the sperm of both boars was normal,” Dyck observes.

Next, the semen from those two boars and three additional boars was pooled, with equal amounts of semen from each. Ten sows were inseminated; nine became pregnant and farrowed 104 piglets.

“Due to the averaging effect, results were not great, but not bad — averaging 10.4 vs. 12.04 pigs born/litter for the pooled semen vs. Boar B (single-sire semen),” he notes.

Tissue samples were collected on the pigs from the pooled-semen litters to establish which sires actually produced the piglets in the litters (paternity).

“The really interesting thing from the paternity testing was even though sperm only represented one-fifth (20%) of the pooled semen, it produced 30% of the piglets. Boar A only produced one piglet out of 104 (1%).

“From this you would assume that all Boar A is doing is diluting; he has no real genetic purpose; he's contributing nothing. Do we really want those boars in the system?” Dyck asks again.

The message seems clear — there's a good chance the pork industry is using some sub-fertile boars. They are not removed from the system because they cannot be easily identified.

“Using pooled semen doesn't allow us to link paternity to the progeny and, therefore, we can't really benefit from the genetic value of the individuals,” Dyck notes. “Plus, we are using excessive numbers of sperm — three billion per dose — to compensate for what may be a minority of animals that don't perform. We're giving 2-3 inseminations, using 6-9 billion sperm to produce a single litter, which obviously increases the number of boars needed for production and it reduces the genetic impact of the better animals.”

From the various studies cited, Dyck reinforced two key points:

The options really boil down to selecting for superior fertility characteristics, but in the absence of good indicative genetic markers, it is difficult to identify those animals without using single-sire matings. “If we did that, we possibly could eliminate some of the boars with limited reproductive capacity and select for boars with strong reproductive performance — even when we reduce AI doses to two billion sperm,” he says.

The Bottom Line

University of Alberta swine reproduction specialist George Foxcroft, coordinator of the annual workshop, reminded: “In the early days of AI, the results had to be as good as natural mating or the industry would not have bought into this technology. That's why we pool boar semen. It's the only reason why we pool semen.”

Foxcroft acknowledges the common argument that pooling improves the efficiency of processing and packaging semen. Still, he challenges, if there are boars in the pool that are only capable of producing 7-8 pigs per litter, “we don't even know they are there; we are blind to it.”

“If we can gain a half pig/litter by simply screening our boars, that's more than we can do with sow management — probably in five years. Technically, it is easier to do something about these sub-fertile boars, and try to get most of them producing 13-14 pigs/litter, than the amount of time we're currently spending trying to manage our sows better. It seems to me, this is something we haven't even addressed,” he adds.

^x,y,zMeans with different superscripts within each column were different by ×2 analysis (P <0.05).

^abLSM with different superscripts within each column were different (P <0.05). Values in the table are least means.

Source: Ruiz-Sanchez 2006.