Header Side Menu Content Footer
Government of Manitoba
Agriculture
Get Started
Stay Connected
to Manitoba Agriculture

Stay Connected
to the Manitoba Government

To view PDF files, you must have a copy of the Adobe Acrobat Reader which is available as a free download:

Get Adobe Reader

Managing and Interpreting the Records of the Breeding Herd

Gary D. Dial College of Veterinary Medicine, University of Minnesota, St. Paul, MN

Producers and their consultants often become frustrated by their inability to accurately assess the efficiencies of swine operations. In an effort to determine which performance measures need improving, most of us nowadays can only subjectively compare the reproductive performance for a herd with empirically derived industry standards. If a producer uses a recording system in which data from multiple farms is collected, a farm's performance can be compared with an existing data base. However, frustration arises from the lack of objectivity in ascertaining the relative importance that different measures of suboptimal female performance have on overall herd productivity. None of us want to waste time assessing measures that have only modest affect in overall profitability, preferring instead to analyze those things that will best reflect the potential profitability of the farm while ensuring its solvency. With historical approaches, outside analysts of production records are forced to blindly address, in a nonprioritized fashion, any factor deviating from industry norms regardless of its real effect on productivity. The intent of this paper is to show the relative importance of the various measures of breeding herd productivity. This article focuses on breeding herd productive efficiency, recognizing it is as the most difficult to assess even though the majority of costs are incurred during the finishing phase.

Assessing Capacity Utilization

One of the fundamental objectives of all farms is to maximize the capacity utilization of a farm. Capacity utilization is a term commonly used in the manufacturing businesses to define how well current production matches the maximum output of an operation. As production is increased toward a farm's capacity, the cost of production for all pigs produced declines. This is a classical reflection of marginal profit. That is, changes in production (i.e. pigs weaned per unit time) allow fixed and overhead costs to be spread over more pigs, resulting in cheaper pigs. Relative to the breeding herd, some of the costs that are classically described as variable costs, such as sow feed and labor, behave as fixed costs, not as variable costs; thus, increased production results in some of the variable costs being reduced on a per weaned pig basis. Thus, assessments of a farm's production relative to its capacity is a critical first step to understanding the link between biological performance and profitability.

The capacity of a commercial farm can be looked at in two ways. The first method determines capacity as the number of pigs weaned from a farm annually. With this approach, capacity is a factor of the number of breeding females in the inventory and the efficiency by which they produce pigs (Figure 1). Capacity is assessed using this method on a farm-level basis and, thus, it is most useful when evaluating the factors affecting the capacities of multiple farms in a large database. The second method determines capacity on a number of pigs weaned/group basis. With this approach, capacity is a factor of the number of sows farrowing/group and the number of pigs weaned/sow (Figure 2). Capacity is determined with this approach on a group-level basis and, thus, it is useful in evaluating the factors that affect the capacity of individual farms. The producer will most commonly use the group-level approach for understanding the factors that contribute the most to herd capacity. In contrast, the external examiner of a herd's records will likely use the herd-level model to contrast farms, thereby determining relative weaknesses and strengths.

Our studies at the University of Minnesota have given us several takehome lessons regarding capacity, specifically regarding the capacity of the breeding herd.

  • Of the two major factors driving a farm's capacity, output efficiency dominates throughput efficiency. That is, breeding female inventory and the number of females weaned/group, called "output" measures in Figures 1 and 2, typically have substantially more influence on capacity than their counterpart "throughput" measures, the number of pigs weaned/female/year and number of pigs weaned/litter.
  • Of the output measures, the number of females served/group is more important than farrowing rate in its influence on capacity. Therefore, the old adage "keep your crates full" appears to hold true.
  • It is difficult to influence the number of weaned sows, repeat breeders, and postweaning culls; thus, the number of gilts served/group has a more important influence on capacity than the number of sows served/group. Gilt pool size and management is critical to a farm's capacity. It also is typically the most ignored phase of production.
  • Farrowing schedule (i.e. number of days between service group) affects several factors driving capacity and, thus, has a compounded influence on capacity. However, since farrowing schedule cannot be changed on most farms, it typically has no real effect on capacity.
  • Nonproductive days affect both throughput and output efficiency. That is, they not only affect the efficiency of female production (i.e. pigs weaned/female/year) but also herd female inventory.

Assessing Efficiency

While measures of output may be better predictors of farm capacity than measures of throughput, assessing the efficiency of reproductive performance remains critical to external audits of a farm's records. Our group has used the PigCHAMP data set, which is updated annually, to study the interrelationship between the common measures of breeding herd performance and a herd's overall efficiency, as measured in terms of pigs weaned/inventoried female/year (PWIFY). This data set contains only those farms remaining after removal of those that had missing data, suspect data or dynamic herd inventories. Table 1 gives the descriptive statistics for the various measures of breeding herd productivity that compose the PWIFY productivity tree shown in Figure 3. As shown, the number of PWIFY is driven by the number of pigs each sow weans and the number of litters she farrows annually.

Such reference tables are useful as yardsticks for comparing the individual measures of performance of farms using the same database. Because of differences in how measures are calculated by various information systems, comparisons of farms using different databases can be misleading for many variables. That is, PigCHAMP calculates such things as pigs/sow/year and nonproductive days entirely different than do systems such as PigTales and Easicare . For the same reason, comparing production summaries for various databases is extremely misleading.

An understanding of the interrelationship (Figure 3) among various measures of the components of a productivity tree enables you to better understand the relative importance that various measures have on overall productivity. In general, as long as you remain in one branch of the tree, parameters higher in the tree are more important than lower measures. For example, total-born litter size predicts PWIFY less well than born-alive litter size.

The interrelationship among the various measures shown in Figure 3 allows a general understanding of the relative importance of production measures. Their absolute importance in influencing PWIFY across the swine industry has not been known until recent studies in our laboratory. By using the PigCHAMP data base as a reference, we have developed approaches that allow the relative contributions of various measures to PWIFY to be dynamically determined for individual farms. Using a statistical procedure, called standardized regression analysis, we have determined the strength of association between the various diagnostic indicators (branches of the productivity tree) and PWIFY. This statistical procedure determines the mathematical relationship between two or more variables (e.g. preweaning mortality, stillbirth rates, nonproductive days) affecting a common endpoint (e.g. PWIFY) and uses that relationship to predict how changes in one variable affect the endpoint variable. For example, Table 2 lists the relative change in key diagnostic indicators needed to produce an increase of 1 pig weaned/inventoried female/year on an average PigCHAMP farm. For example, a herd that reduces the number of NPD/female by 16 days can expect an increase of 1 PWIFY.

For herds using PigCHAMP , the relationship between the components of the PWIFY productivity tree can be defined by the simple equation:

productive efficiency, recognizing it is as the most difficult to assess even though the majority of costs are incurred during the finishing phase.

PWIFY = 18.63 - 1.55 NPD + 1.17 TB - 0.85 PWM - 0.51 SBM - 0.48 LL

Where, 18.63=PigCHAMP average for PWIFY, NPD=nonproductive/female/yr, TB=total pigs born/litter, PWM=% bornalive piglets dying before weaning, SBM=no. stillbirths plus mummies/litter, and LL= lactation length.

The numbers in front of the reproductive measures included in the equation reflect the effect that 1 standard deviation unit (a measure of variation) change in that parameter has on PWIFY. For example, if NPD for the average herd is improved by one SD, then PWIFY would improve by one pig. Further, one SD unit change in NPD has a larger effect on PWIFY than one unit change in TB. The signs (+ or -) in front of the numbers indicate the direction of change that must be accomplished to improve PWIFY. For example, NPD, which has a negative number, must be reduced to improve PWIFY, whereas TB, which has a positive number, must be increased. Gestation length, one of the components of the PWIFY productivity tree, does not have a significant influence on PWIFY.

There are some obvious limitations to our approach. It is highly likely that there would be a difference among the variables in the effort and, perhaps, cost required to make a 1-pig change. Changing a herd's average NPD from 95 days (10th percentile PigCHAMP herd) to 65 days (50th percentile) may be much easier to make on some farms than improving the herd's total-born litter size from 10.2 pigs (10th percentile) to 11.0 pigs (50th percentile). Therefore, subjective consideration of the ease of implementing change is an important prelude to initiating corrective m

TABLE 2 - Influence of Various Measures of Breeding Herd Productivity on Pigs Weaned/Inventoried Female/Year
PARAMETER

PigCHAMP Mean

CHANGE TO PRODUCE 1 PWIFY INCREASE

NPD/female/yr

29

-16.1 days

Lactation Length

25

-7.5 days

Stillbirths+mummies

.8

-0.59 pigs

Totalborn

11.0

+0.56 pigs

Preweaning Mortality

1.38

-4.7 %

Farms differ considerably in productivity, with substantial variation among farms in the levels of performance for the commonly used measures of productivity. Thus, it is not entirely unexpected that we have found a considerable variation among PigCHAMP farms in the relative impact of various measures of breeding productivity on PWIFY. Across the database, NPD ranked as the most important variable effecting PWIFY, followed by total-born litter size and preweaning mortality. However, when productivity measures were ranked for individual farms, total-born litter size was the most important problem on 41% of 641 farms, NPD was most important on 32% of farms, and preweaning mortality had the greatest effect on 25% of farms. There were less than 4% of farms in which stillbirth plus mummy rates were the highest ranking parameter.

There is considerable farm-to-farm variation in the ranking of the relative importance of the variables affecting pigs weaned/sow/year. The variation that we observed among three farms is illustrated in Table 3.

TABLE 3 - Estimated Potential Gains in Pigs Weaned/Inventoried Female/Year on Three Farms by Improving Reproductive Parameters to the 80th Percentile PigCHAMP Ranking

The priority of action on an individual farm is dictated by the current level of performance for each of the key diagnostic variables, how each varies from achievable targets, and the relative amount of effort or cost required to make a change in each measure. The initial step to getting top performance from a herd is to keep accurate records that can be used to compare to a defined data base. To minimize the chance of an Òapple-versus-orangeÓ type comparison, the farm being assessed should be compared to similar farms using the same record system. Secondly, select targets for each measure of performance. For example, should the farm be compared to the 80th percentile, the 90th percentile, or the 98th percentile? Thirdly, calculate the added number of PWIFY that results from improving one or more measures of productivity to the target value. Easy-to- use algorithms for calculating the potential impact of improvements on PWIFY are being made available to PigCHAMP users.

As production improves over time, the ranking of the parameters changes. For example, Table 4 shows quarterly production measures for the same farm at two times approximately 2 years apart. Note that there were improvements in NPD, litter size, preweaning mortality, and stillbirth rates over the two-year period. NPD and preweaning deaths became less important factors over time; litter size remained the most limiting parameter; and the influence of stillbirths/litter became greater.

TABLE 4 - Change in Ranking of Importance of

Cheating with Records

Computerized records promise to give producers more accurate, comprehensive, useable, and timely access to information. While farm-to-farm comparisons and comparisons to a central database are highly useful methods for assessing an individual farm's performance, these comparisons must be done with a great deal of caution. Here are some guidelines for minimizing the chance that erroneous conclusions will be made from a farm's records.

  • Avoid comparing the data collected and reported with one information system with that collected with another. There are often substantive differences in the algorithms used and data may be collected differently.
  • Understand the impact of missing data. Missing events, such as service, weaning, and farrowing events, can make data look artificially good or bad.
  • Understand the impact of inventory on performance parameters. Several key parameters are substantially effected by the time at which gilts are recorded after actual entry into the information system and culled sows are electronically removed from the database relative to their physical removal from the herd.
  • Understand if data is being entered erroneously. Pigs can be recorded as being born or weaned; matings can be recorded when they have never occurred; feed deliveries may not be recorded as they occur, etc.

Progressive computer programs have built in data checks and data entry reminders, that tell the user when they have done something that does not make biological sense. PigCHAMP has a data integrity report that details potential sources of data integrity problems. An essential prelude to an analysis of a farm's database is the evaluation of its data integrity and an understanding of the farm's management practices.

Conclusions

A farm's production data offers valuable and, perhaps, essential information, if it is recorded accurately and related objectively to similar farms using the same record system. In order to be competitive and continue making clear progress, farms should be analyzed regularly, perhaps at quarterly intervals. Broadly used information systems, such as PigCHAMP , not only provide valuable production references and are useful on-farm decision aids, but they also give producers and their lenders vital tools for objectively interpreting a farm's data as well as a clear recognition of the problem areas most likely to give the greatest return on improvement efforts.


Bookmark and Share