Variation: Real hidden cost in pork production
Since we rarely measure it, we can’t really manage it
By John Patience
Variation is obviously a part of raising any species of livestock, and pigs are no different. In some respects, variation within a herd of pigs is an essential part of pork production because it is the basis for genetic selection of animals with superior traits. If all animals were the same, then classical genetic selection would not be possible.
However, variation is also a big challenge for pork producers because variation costs money and complicates barn and animal management. Because we rarely measure variation in the barn, it is very difficult to manage and a big challenge to even understand.
Variation reduces the efficiency with which we utilize barn floor space. It increases the work required to market hogs within desired carcass weights. Variation in lactation feed intake leads to all kinds of issues with sow nursing performance and rebreeding success. Variation in colostrum production by the sow and intake by the newborn piglet causes significant problems with piglet livability, growth performance and disease resistance. Variation also leads to less uniform consumer products, in such things as color, tenderness and the size of various cuts. In other words, variation affects all stages of production, from the breeding herd to the butcher.
Let’s look at one example of variation in detail. It is well accepted that some pigs get onto feed earlier than others at the time of weaning. There are not a lot of data on this topic, but one study showed that even in pigs weaned at four weeks of age, about 10% had not eaten any feed by the end of the first day. About 5% had not eaten by the end of the second day; in a typical 2,400 head barn, this would represent about 120 pigs.
We assume that most of these pigs figure it out and do well once they start eating, but one thing is certain. If a newly weaned pig is not eating to at least its maintenance energy requirement, no matter how warm the barn may be, that pig will lose body heat to its surroundings; in other words, it will be chilled and that will certainly lead to weight loss, and greater susceptibility to disease and other stressors in the environment. The pig generates body heat by digesting and metabolizing feed; if the pig is not eating, chilling is the natural consequence with attendant health and growth problems.
Even in relatively uniform groups of nursery pigs, the lightest pigs in the batch will not reach the weaning weight of the average pigs until about 3 weeks postweaning. And the average pigs in the group will not reach the weight of the heaviest pigs until at least 2 weeks post weaning. Following these same pigs to market, at first cut, body weights will vary by 100 pounds or more. Variation is not a trivial matter! Imagine the savings in time and the improved utilization of barn capacity if all pigs could be dumped in a single day, like a broiler barn!
Let’s consider an example of variation with direct consequences on the cost of production. Looking at pigs from 165 to 265 lb. with an average protein deposition rate of 160 g/d (a decent pig but not a rock star), the lysine requirement will vary by as much as 0.12% just based in differences in feed intake. If we consider the combination of differences in both protein deposition rate and feed intake, the lysine requirement might vary by another 50% or more – eg. 0.18%.
Remember that when we determine amino acid requirements, we do so on the basis of the average for the group of pigs, and as a result, we do not get a very good understanding of how much the requirement varies within the group of pigs being studied. I have conducted dozens of amino acid titration studies in my career, and I was always left wondering how much the requirement I had just determined varied within the group of pigs I studied. Only once in 35 years was I able to run a study on individual pigs that allowed us to generate the numbers presented above. These data clearly show how much variation in the basic biology of the pig can impact nutrient requirements.
So, the fact remains that we are feeding a single diet even though it means overfeeding a significant portion of the animals; under most conditions, we are not prepared to sacrifice the performance of the faster growing pigs in order to avoid overfeeding slower growing pigs. Prevailing economics dictates that we feed pigs to their genetic potential, or close to their genetic potential. Help may be on the way, however. We hear more and more about precision feeding in the pig industry; as this technology evolves and becomes more affordable, targeted feeding programs based on actual performance is going to get much more attention.
By the numbersWhat does variation look like? Table 1 reports the results of collecting individual body weights (BW) on almost 2,600 pigs at three ages: 19, 68 and 140 days. At weaning, while BW averaged 11.9 lb., it ranged from a low of 5.3 lb. to a high of 20.3 lb. I think we can all agree that feeding and managing a 5 lb. pig at weaning is very different than a 20 lb. pig, or even a 12 lb. pig. As the pigs get older, variation increases; at about 10 weeks of age, the BW of the pigs ranged from 52.5 lb. to 90.2 lb. and at 20 weeks of age, BW ranged from 164.1 to 275.4 lb., a range of 111.4 lb. No matter how you look at it, this is a lot of variation!
If this is what variation looks like, how is variation normally measured? For performance measures such as body weight, standard deviation is perhaps the best option. Because it requires weighing a large number of individual pigs, most barns don’t have the available labor to do this. Consequently, while standard deviation can be determined in commercial barns, it is most likely measured in research barns. Even then, it is time consuming and laborious, and as a result, is rarely measured.
Standard deviation is a statistical term that was developed 130 years ago to measure the dispersion of data. Standard deviation is illustrated in Figure 1. This curve represents all of the data in a given population; the area under the curve represents the number of observations at each weight along the X-axis.
It is easy to see that most of the pigs have weights that are close to the mean; the further one gets from the mean, the number of pigs gets smaller and smaller, so at the extreme weights, there are very few pigs. One standard deviation, depicted in red in Figure 1, represents 68.3% of all pigs, while two standard deviations, depicted in green, represent 95.5% of all pigs. In the previously mentioned Table 1, the standard deviation of BW increased from 2.6 lb. at weaning to 18.3 lb. at first marketing cut. It is very normal for standard deviation to increase as pigs grow.
Another term that is often used to describe variation is called the coefficient of variation or CV. Very simply, it is calculated by dividing the standard deviation by the average. In Table 1, the CV of BW at weaning can be calculated by dividing 2.6 by 11.9 which equals 22%; CV is always expressed as a percentage.
I like to use CV when studying variation, but I have to remind myself that I can only do so with caution. Because both the standard deviation and the mean are used to calculate CV, if the mean changes and the standard deviation stays the same, the CV will change. Since the standard deviation is the number that best represents variation in the population, it is easy to be confused if the mean changes.
Let us take the example of birth weight and a comparison of litter sizes. As litter size increases, the average birth weight will obviously decline. If the standard deviation stays the same, the CV will increase due to the decrease in birth weight; on the basis of CV, one would conclude that variation in birth weight increases with increased litter size but in reality, variation in this instance did not increase because the standard deviation of birth weight stayed the same.
What confounds the issue is the fact that as litter size increases, average birth weight declines, which means that more light weight pigs will be born and those light weight pigs create more headaches for the barn staff. In other words, because piglet birth weight declines with litter size, by definition, there will be more lightweight pigs.
Because there are more light weight piglets born, it is assumed that birth weight variation increases. However, variation may not have changed or changed very little; simply, the mean birth weight declines which means that even with equal distribution of weights, more piglets will slide into the underweight category.
I raise this point because, like many aspects of variation, understanding the root cause of production problems helps us to identify the best strategies for improvement. Erring in understanding the cause of production problems associated with variation can cause us to make mistakes in developing solutions.
One example of this is the former practice of sorting pigs by body weight when barns are loaded with weaned piglets. Indeed, sorting pigs when loading barns became a fairly common practice – that is until actual research was conducted on the subject and revealed that such a practice does not reduce overall variation in the barn, and may in fact reduce pig performance.
It has thus been concluded that sorting pigs may be desirable from a management perspective but not to reduce variation. For example, sorting the smallest pigs into six or eight pens in a 2,400 head barn will allow adoption of a feed budget different from the rest of the barn that better suits smaller pigs at weaning.
Another management practice that makes sense practically and scientifically involves the creation of special care pens for pigs that require medical treatment, have been injured or are generally falling behind the rest of the group. These two practices help us to manage variation but they do not reduce the total variation in the barn. More on this later.
Coefficient of variation is very useful in benchmarking. For example, a CV for BW of 8% at first cut would be considered very, very good; it would be highly unlikely that less variation could be achieved in commercial practice. In contrast, Dr. Kate Dewey at the University of Guelph reported an average CV of BW at first cut of 20% across six different farms, about 2.5 times bigger than we observed in the barn used in the study reported in Table 1.
An important question to ask is how many pigs do we need to weigh to obtain a reasonable estimate of CV? Unfortunately, it is a very big number (Table 2). Assuming pigs are selected randomly from the barn, at weaning, something in the range of 7.5% of the pigs, or 90 animals, would provide a reasonable estimate of the average weight, but all of the pigs would have to be weighed to estimate the standard deviation.
At 10 weeks of age, even 18 pigs, selected randomly from the barn would provide an estimate of the average that was within 5% of the actual average. However, 25% of the pigs, or 175 animals, would have to be weighed to obtain a reasonable estimate of the standard deviation, and even then, the result would not be as accurate as the estimate of the mean. At 20 weeks of age, weighing 47 pigs or 7.5% of the group would provide a reasonable estimate of the mean, but 311 pigs, or half the animals would have to be randomly selected and weighed to obtain a reasonable estimate of the standard deviation.
Thus, it is pretty clear why standard deviation is so rarely measured, even in most research facilities. Yet, we cannot manage what we do not measure, and if we do not measure variation in the barn, it is very hard to manage it effectively.
Managing versus reducingPork producers have two choices when it comes to dealing with variation. The first is to work to reduce variation, and the second is to manage it. By reducing variation, I am referring to taking action that actually lowers the standard deviation of body weight, to create a more uniform barn fill. For example, Improving the health of pigs will reduce variability; illness in a barn greatly increases variation in growth performance.
The second option available to pork producers is to manage variability. This includes actions taken to minimize the impact of variation on barn operation without changing the standard deviation. Perhaps the best example is marketing a barn using three or four cuts, with each cut removing the heaviest pigs available for market; the final cut or dump cut ships all remaining pigs to market. While this approach does not reduce the overall variation of the barn, it does result in more uniform pigs going to market.
How does one decide whether to reduce variation or manage it? One effective tool is to use body weight CV as a benchmark; if the CV at weaning is above about 20 to 22%, then options to reduce variability might be considered. If the CV is already around 18-20%, it is unlikely that it can be further reduced so managing variability makes more sense than reducing it. The same benchmark approach can be used at nursery exit, if a three-site system is being used. A baseline body weight coefficient of variation at nursery exit might be 12 to 15%; in the same way, a baseline of 8 to 10% can be applied at barn first cut.
A CV for BW of 8% at first cut would be considered very, very good; it would be highly unlikely that less variation could be achieved in commercial practice. In contrast, Dr. Kate Dewey at the University of Guelph reported an average CV of BW at first cut of 20% across six different farms, about 2.5 times bigger than we observed in the barn used in the study reported in Table 1.
Please note that collecting weights as the pigs are marketed in each cut is not a suitable way to evaluate barn variation because cuts are selected according to BW and thus will not reflect the overall variation in the total barn population.
Reducing variationA good way to evaluate strategies that might reduce variation is to consider the social behavior model which says that any factor in the barn that gives an advantage to dominant pigs in a pen is a candidate for reducing variation.
For example, if there is inadequate feeder space or waterers in the pen, or if feeders are adjusted too tightly, dominant pigs can push subordinate pigs away; the dominant pigs can get all the feed and water they want, while subordinates will have to settle for less feed or water, resulting in dominant pigs growing faster and subordinate pigs growing slower. By definition, this leads to greater variation.
Therefore, the following are options for reducing variability:
Ensuring that there is adequate feeder space available, and that feed flow is not restricted in any way.
Ensuring that there is adequate drinker capacity in the pen, and also ensuring that drinker flow rates meet standards required by the pigs in the pen.
There are other strategies unrelated to the social behavior model which have been shown to reduce variability:
Reduce disease load. Anything that reduces pathogen load in the barn is likely to reduce variability. Farm surveys have shown that variation is increased in barns with a high disease load.
Improve the quality of the phase 1 and 2 nursery diets. This strategy may not make economic sense, because the improvement in uniformity is rarely sustained through to market.
Practice strategic euthanasia. A recent study of commercial data (Knap et al., 2023) showed that lactation mortality increases sharply when birth weights are less than 2.4 lb. Mortality appears to increase linearly as birth weight declines. Individual producers select different birthweights for euthanasia. Generally, pigs born weighing less than 1.5 lb. have a very low likelihood of survival. Even pigs weighing less than 1.8 lb. at birth face a very uncertain future and euthanasia may be the best and most humane option. Euthanasia can also be practiced in the nursery and finisher barn when compromised pigs fail to respond to treatment and special care.
Managing variation
Depending on sow herd size, and other management variables, weaning more frequently will reduce variation in weaning weights.
To the greatest extent possible, load wean-to-finish barns and finishing barns in the shortest possible time.
Segregated parity production reduces variability because gilt offspring are different in numerous ways from sow litters.
Take whatever steps are necessary to increase growth rate. Increasing overall barn growth rate will increase the portion of full value pigs. However, if the increased growth rate is used to turn the barn over faster, then this advantage will be lost.
Increasing weaning age and thus weight will result in heavier pigs delivered to the finishing barn with the same advantages listed in the previous item.
Rarely practiced in the US, separating barrows and gilts will improve uniformity of growth. It will also free up the barn loaded with barrows for re-fill before the gilt barn; the difference between males and females can be as much as a week. However, few systems are able to load barns according to gender, explaining why it is not commonly practiced here. It is more common when smaller finishing barns are used, because they are easier to fill in a short time.
ConclusionVariation is clearly an important issue to the pig industry. While variation is a fact of life, it is also a factor in our barns that impairs optimum barn utilization, complicates the marketing process, and increases feed and other costs.
Variation that begins at birth and is exacerbated at weaning contributes to some of the variation that is observed at the time of harvest. Therefore, anything which reduces variation early in life is likely to be beneficial.
Before major efforts to reduce variation can take place, we must first measure variation in our barns, to better learn the magnitude of the issue, and to help identify practice that will be beneficial to the pigs, and their mothers.
Unfortunately, measuring variation is time consuming and requires large numbers of animals to be weighed. The refinement of technology which weighs live pigs in the barn using infrared cameras or other automated systems will provide a major boost in our fight to reduce variation and thus simplify barn operation and improve net income.
Patience is with Oxford Nutrition Consulting LLC and is a professor emeritus at Iowa State University.
