Manitoba Agriculture, Food and Rural Initiatives

Growth Curve Analysis: Practical Tools for Making Farm-Specific Decisions

Bob Goodband, Steve Dritz, Mike Tokach, and Jim Nelssen
Department of Animal Sciences and Industry
Kansas State University

We use growth curve information for two main reasons. The first is to study and determine nutrient requirements in actual commercial production situations. The second is to examine the economics of various feeding regimens or determine marginal economics to estimate optimal slaughter weight. Thus, we are able to fine tune dietary recommendations on a farm-specific basis in order to enhance the profitability of feeding strategies. We estimate that overfeeding lysine by .1% results in an increased feed cost by approximately $1.20 per pig. In contrast, under feeding lysine that lowers lean percentage by .5% leads to a reduction in net income by approximately $.60 per pig and increases feed cost at least that much as a result of poorer feed efficiency.

Some researchers and feed companies have tried to develop models that categorize various factors impacting growth. Growth and nutrient needs are predicted from the sum of these factors. We have taken a different approach, we observe growth in commercial production systems and then calculate the amount of nutrients to drive the observed growth, instead of trying to predict the growth based on categories of factors.

We also believe the responses to dietary energy and protein in actual commercial production situations are not the same as measured in controlled research trials. Energy intake or feed intake is usually much higher in research settings than encountered in most commercial production situations. Thus, we do not observe the same responses to energy density in the research station compared to commercial production. This concept is illustrated by the average daily gain (ADG) response of added dietary fat for high-lean genetic pigs. We have performed two experiments at Kansas State University examining the response to added dietary fat from 50 to 115 kg.^1,2 These pigs were high-health, high lean growth potential pigs housed two pigs per pen and had excellent feed intake. Increasing dietary fat had no effect on ADG, but as expected, it reduced average daily feed intake (ADFI) and improved feed efficiency. This is contrasted with recent research with pigs housed in a commercial production system. In that study we observed a linear increase in ADG with increasing added fat to the diet.³ It is important to note that the ADFI in the commercial research setting was approximately 15% lower than we normally observe in our university research environment. This indicates that energy is a more limiting nutrient at feed intakes observed in commercial production compared to those observed at the K-State research farm.

Until adequate models are developed to categorize and quantify the factors affecting growth, experimental data will have to be evaluated in actual commercial production situations. Our first case study will illustrate how we are using growth curve analysis in commercial production situations to fine-tune dietary recommendations.

The second reason we use growth curve analysis is to determine the economics of feeding programs to determine optimal nutrition programs or market weights. Sometimes retrospective records such as close outs or PIGCHAMP data are used to make economic decisions. Unfortunately, many times this retrospective data is confounded with several unknown factors and may be biased. In our second case study number, we will examine how growth curves are needed to make marginal economic evaluations while retrospective data may be confounded and lead to misleading decisions.

The use of growth curve analysis has allowed us to characterize the biology of growth in actual production situations and translate that biology into economic decisions. The objectives of this paper are to present some practical guidelines that we use for data collection and present some examples of how we have used the data for making decisions.

Approaches for Growth Curve Analysis

We have taken two approaches when developing growth curves. The two approaches depend on whether the objective is to determine nutrient requirements or develop feed budgets.

Objective 1 - Determining Nutrient Requirements

The basic concepts we use for the collection and translation of growth curve data into nutrient requirements are based on the concepts presented by Schinckel and de Lange, 1996.⁴ Briefly, this involves obtaining weights and ultrasound measurements of backfat and loin eye at approximately 5 to 6 points during the growth period. The ultrasound and weight measurements are then used to determine the amount of body protein and lipid at each weight. Daily protein (P) and lipid (LP) accretion curves are then calculated. The daily lysine requirement (grams per day) can then be calculated from daily protein accretion by using constants for the lysine content of protein, the efficiency of lysine utilization, and the maintenance requirement (Table 1). We then calculate the requirements for the remaining amino acids based on a ratio relative to lysine according to the concepts presented by Baker, 1997.⁵

Table 1. Equation to convert daily protein accretion into daily lysine requirement.

Parameter	Constant or Equation
Lysine content of body protein, L	6.6%
Post-absorptive efficiency of lysine utilization, E	65%
Digestibility of lysine in the diet, D	80%
Lysine needed for maintenance, M	.036 x Body Weight, kg .75
Daily total lysine requirement, g = ((M) + (P * L) / E) / D.

The daily energy intake requirement to produce the observed growth is then calculated from the daily protein and lipid accretion plus an allowance for the maintenance energy requirement. The grams of lysine intake can then be divided by the daily energy intake to derive a lysine to calorie ratio that can then be converted to a dietary percentage based on the dietary energy concentration. We then convert the dietary percentage into a curve based on body weight as depicted in case study number one. The curve can then be used to determine dietary lysine requirements for each phase.

A standard protocol for the collection of growth curve data that can be used to determine nutrient requirements is listed in Appendix A. Key concepts to remember when collecting the data are:

1. Randomly select a group or groups of pigs that are representative of the production system.
2. Follow the same pigs over time.
3. Obtain individual pig age.
4. Start ultrasound measurements at approximately 20 kg.
5. Pig weights at earlier intervals (< 20 kg) will increase accuracy.
6. Maintain the group as long as possible or as heavy of weight as possible.
7. Remove pigs from test all at the same time.

Producers provide us with copies of the data set and we translate the curves into nutrient requirements in cooperation with Alan Schinckel at Purdue University. The data is relatively inexpensive to obtain and is even less expensive if groups of producers that have common facilities, genetics, and feeding programs collaborate together for data collection.

Recently we have developed standardized growth curves based on differences observed in fat free lean index. This eliminates the need for obtaining ultrasound measurements on every farm. Furthermore, we have made a conversion from the U.S. Fat Free Lean Index to the Canadian marketing grid using the Hennesy Probe (Figure 1). Calorie:lysine ratio requirements and actual lysine requirement estimates have been calculated using the above described procedures. This may allow producers the option of either developing farm specific growth curves and corresponding nutrient requirements, or using generalize growth curve equations (Figures 2 through 5).
 

Figure 1	Figure 2	Figure 3
Figure 4	Figure 5

Objective 2 - Feed Budgeting

Although a daily feed requirement to drive growth can be determined from the body composition data, this does not take into account feed wastage or variation between groups. Feed budget development has evolved from our frustration in determining accurate feed intake data. The major concept that moved us forward in this area was the observation that feed intake varies widely in commercial production situations; however, a large part of this variation can be correlated with growth performance.⁶ A good example is depicted in the figure adapted from data presented by Bahnson and Dial, 1995 (Figure 6).⁷
The graph depicts the seasonal variation on close-out feed intake. However, there is almost a one to one correlation between feed intake and ADG for pigs housed in modern environmentally controlled buildings. Although we acknowledge that the environmental effects on feed intake in the summer have an impact on the biology of energetic feed conversion efficiency, it does not appear that the amount is large enough to make practical adjustments in feed budgets (approximately 3% across season). We believe that this is probably true for other factors such as herd health.

Therefore, our approach has been to develop cumulative feed budgets based on cumulative weight gain. We initially developed a feed budget curve based on variety of data which included close outs, on-farm experiments, and research data. The cumulative feed budgets have then been converted into a table based on a standard close out feed efficiency. We then placed the curve in an Excel spreadsheet so that the budgets for each phase can be easily adjusted based on overall close out feed efficiency and customized weight breaks (KSU Feed Budget Program).⁸

With the advent of many large production systems that have similar feeding programs, genetics, and buildings, we also have developed specific feed budgets for individual production systems. The overlying concept is to develop a statistical inference to the production system. The basic approach is to randomly select 6 groups for each gender and then track feed deliveries to the group. In addition, a random sampling of pens (3 or 4) in the group are weighed to determine the average pig weight of the group. Feed is inventoried on each weigh day and cumulative feed intake determined. The groups are followed as long as possible with the removal of as few of pigs as possible. A minimum of 5 data points is needed to develop the curve. A curve can then be fit to the data and an equation derived to determine the cumulative feed intake. This can be easily accomplished by making an X-Y scatterplot in Excel and using the trendline function to obtain the equation for the curve. The customized budget for each phase can then be calculated by subtracting the cumulative intake at two different points. An example is depicted in Figure 7.
Because feed is delivered to an individual group, the average curve is developed from a sub sampling of groups within the production system. The development of feed budgets take into account both the feed required for growth and feed disappearance due to wastage and can be customized for application to specific production systems.

Case Studies

Two case studies are presented that illustrate how we have used on-farm growth curve data to aid in our decision making process to optimize farm profitability.

Case Study 1 Fine-tuning farm-specific dietary recommendations from growth curve data.

Growth and ultrasound measurements were collected and translated to nutrient requirements according to the procedures presented in this paper in two different production systems with similar genetics. The major difference between systems is that farm 1 uses multiple-site finishing with all-in/all-out by barn production and farm 2 uses single-site finishing with all-in/all-out by room. The protein accretion, lipid accretion, and lysine requirement as a percentage of the diet are listed in Figures 8, 9, and 10, respectively.
 

Figure 8	Figure 9	Figure 10

It is striking to note that the difference between farms was much greater than the differences between barrows and gilts. The other striking note is the dramatic decrease in protein deposition and increase in lipid deposition in the late finishing phase (i.e., greater than 70 to 90 kg). This indicates that growth is dramatically slowing and feed efficiency rapidly deteriorating at heavier weights. Similar results have been observed from several other production systems. This data is in contrast to the traditional thinking of many producers, veterinarians, and nutritionists that growth rate keeps increasing as pigs get closer to market weight. However, the data is consistent with the principles used by many researchers who study and model growth.

Evaluation of the lysine requirement curves compared to the feeding programs in place at the time of the experiment indicated that farm 1 was overfeeding lysine by approximately .1% from 90 to 115 kg. Reducing dietary lysine for this phase decreased feed cost by $.45 per pig. The data also indicated that farm 2 was overfeeding lysine by approximately .15% from 60 to 115 kg. Reducing dietary lysine for this phase decreased feed cost by $1.03 per pig. In addition, there were indications that farm 2 was under feeding lysine in the early period since the calculated lysine requirement was higher than what was actually fed.

Case 2 – Correlation of Retrospective Data with Biologic Principles.

Figure 11 depicts the overall ADG from 117 close out-groups from one production system. Within this production system there is a fairly strong correlation (r²=. 42) between market weight and ADG (ADG increased as market weight increases).

Figure 12 depicts the growth curve from two groups within the same production system measured prospectively. The data in Figures 11 and 12 seem contradictory. However, further examination will explain that the data in Figure 12 is confounded with time. This is because the barns are emptied after a set amount of time (i.e., 16 weeks). Thus, faster growing groups of pigs will be heavier at market because of their faster growth throughout the finishing period. This analysis can be used to determine the impact of increasing ADG for the entire system on the pounds of pork generated or determine the economics on a system wide basis. However, does this data accurately reflect the biologic growth of pigs over the 105 to 125 kg weight range or can it be used to determine the marginal economics of an additional pound of gain? The answer is probably not. The biology of growth is more accurately reflected in Figure 12. This figure indicates that for pigs greater than 90 kg the marginal ADG is decreasing due to the physics of decreased amount of protein gain and increased amount of lipid gain which is energetically less efficient.

What is the disadvantage of using the prospective growth curve in to make economic decisions? The major disadvantage is whether the growth curve adequately makes an inference to the production system. Measurement on a single group of pigs is potentially prone to bias. For example, the group may be an exceptionally fast growing group of pigs and thus the growth curve is much better than encountered on the average. However, even with limited data, the prospective curve is probably a better tool than the retrospective data to determine the marginal value of gain.

This case study illustrates that when interpreting retrospective data consider biologic principles to help determine potential confounding factors in the data. The second implication is that application of new technology or dietary strategies that alter growth performance will have to be inferred to the production system when determining economics.

In conclusion, growth curve analysis has allowed us to better understand nutrient requirements under commercial conditions, determine farm-specific recommendations, and evaluate economic scenarios. Furthermore, these techniques have lead to a better understanding of the marginal economics of feeding heavy weight pigs.

References

1. Smith JW et al. The effects of poultry fat and choice white grease on finishing pig growth performance and carcass characteristics. J Anim Sci. 1997; 75(Suppl 1):196.

2. Smith JW et al. The effects of increasing dietary energy density on growing-finishing pig growth performance and carcass characteristics. KSU Swine Day 1995;108-111.

3. Van Heugten E et al. Effects of pelleting and fat supplementation on growth performance of growing-finishing pigs. J Anim Sci. 1997; 75(Suppl 1):194.

4.Schinckel, AP and de Lange CFM, Characterization of growth parameters needed as inputs for pig growth models. J Anim Sci. 1996; 74:2021-2036.

5. Baker DH. Ideal amino acid profiles for swine and poultry and their applications in feed formulation. Nutr-Quest, Inc. Chesterfield, MO. 1997. BioKyowa Technical Review #9.

6. Rademacher C, et al. Benchmarking: Building a regional database and establishing performance targets for the growing pig. Proc of AASP. 1997; 271-281.

7. Bahnson P and Dial G. Factors associated with output and efficiency in growing and finishing swine. Proc of AASP. 1995; 305-310.

8. Dhyuvetter K and Tokach MD. KSU Feed Budget Program. Kansas Cooperative Extension Service, Northeast Area Extension Office.

Appendix A. Standard Protocol for Collection of Growth Curve Data

1. Randomly, select and double ear tag 39 or 40 pigs per group per gender. Select 13 pigs from each of 3 pens if the group contains 10 or less pens per group. Select 10 pigs from each of 4 pens per group if the group contains more than 10 pens. Also, exclude any sort or sick pens.
2. Randomly select the pens across the length of the building by thirds or quarters and excluding any sick pens. For example, a barn contains 10 pens on each side with barrows (Pens 1 to 10) housed on one side and gilts (Pens 11 to 20) on the other and Pen 1 is a sort pen. Therefore, select 13 barrows from either pen 2, 3, or 4, the next 13 barrows from either pen 5, 6, or, 7, the next 13 barrows from pens 8, 9, or 10, select 13 gilts from pen 11, 12, or 13, the next 13 gilts from pen 14, 15, or 16, and the last 13 gilts from pen 17, 18, 19, or 20. Double ear tag each selected pig.
3. At the initial weighing, record the pig identification, date, gender, weight, backfat, loin eye area, and age for each pig. Ultrasound and weight measurements will be obtained once every three weeks with the final measurement as close to market and at as heavy of weight as possible. When collecting growth curve data in a wean to finish barn the ultrasound measurements will not be obtained until the 4^th weight period (9 weeks after placement).
4. Record the individual pig identification, date, weight, backfat depth, and loin eye area for each pig on test on each ultrasound and weigh day. Place the data in an Excel spreadsheet with pig identification, date, weight, backfat, and loin eye area each in a separate column and each pig on an individual row (See Table 1).
5. If pigs are sold from the barn at multiple times, retain the tagged pigs on test until the last marketing time. It is important that the average weight of the pigs is heavier than the normal market weight for accurate interpretation of the curves. A certified ultrasound technician will be required to perform the ultrasound scans (a list of certified technicians can be obtained from the National Swine Improvement Federation web site).
6. Maintain a record of the date and nature of any clinical disease outbreaks and if any diagnosis is made. Also, maintain a listing of the diets fed with date and amount each was fed.
7. Collect information to provide ADG, ADFI, and F/G for the close out-group.
8. At the end of the experiment, Kansas State University personnel will then analyze the data and provide an estimate of the lysine requirements for the pigs in that group.

Table 1. Example spreadsheet layout.