| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ANIMAL GROWTH, PHYSIOLOGY, AND REPRODUCTION |
New South Wales Department of Primary Industries, Elizabeth Macarthur Agricultural Institute, Camden, NSW 2570, Australia
 |
Abstract |
|---|
 Top Abstract INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: feed efficiency • feed intake • growth • pig
|
INTRODUCTION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Improvement in efficiency of feed utilization is currently a major consideration in most pig breeding programs (Stewart, 1998). In some of the recent studies on feed intake and efficiency in pigs, the test duration has ranged from 28 to 110 d (Eissen et al., 1999; Hermesch et al., 2000; Nguyen et al., 2000). There are a number of reports on feed intake that show a curvilinear growth pattern from the young growing pig to maturity (Von Felde et al., 1996; Schnyder et al., 2001). Although this was confirmed in Eissen et al. (1999), the report also indicated that the first-order polynomial sufficiently described the feed intake pattern of pigs from 28 to 108 kg. This implies that, depending on the repeatability of the measurement, it might not be necessary to measure feed intake for the whole length of the growing period.
Management and feed costs increase as the duration of test increases; hence, it would be beneficial to identify the optimal test duration to reduce the cost of management without compromising data accuracy and reliability. A shortened test also means that more pigs can be tested using the same equipment. In recent years, the optimum length of test for feed efficiency in beef cattle has been examined in several studies (Archer et al., 1997; Wang et al., 2006), and the results indicated that the traditional 120- to 140-d test can be shortened considerably to as short as a 70-d test without significant loss of accuracy. There is a need for similar analysis for other species of livestock. The objective of this study was to determine the optimum duration of test for the measurement of ADG, feed intake, feed:gain ratio [F:G, which is the reciprocal of the efficiency of gain (G:F)], and residual feed intake in growing pigs.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Description of Animals and Data
Growth and feed intake data from a major pig energetics project at the EMAI in Camden, Australia, were used in this study. The experiment was replicated in time as 2 runs conducted during 2003 and 2005, respectively. Each run consisted of 2 parallel components. One component was a longitudinal study of live performance and body composition using computed tomography (CT) imaging. The second component, run concurrently with the longitudinal study, was a serial slaughter experiment to study the association between body composition assessed from CT imaging and whole-body chemical analysis.
For each of the 2 replicates, 90 hybrid (mainly Large White x Landrace) pigs were sourced from a Pig Improvement Company multiplier herd (Kings Partnership piggery, Bendigo, Victoria). One female and 2 male piglets were selected from each of 30 litters at 7 d of age. The selected pigs were born within a 7-d period. The 90 selected pigs remained at the Kings Partnership piggery until they were transported to EMAI at 42 ± 1 d of age (mean ± SD). During the period at the piggery, the pigs were with their sows and littermates and were managed like all the other pigs at the piggery. One of the 2 selected male piglets from each litter was surgically castrated at 10 ± 1 d of age (mean ± SD). The pigs were weaned at 28 ± 1 d of age (mean ± SD) and fed, ad libitum, a standard commercial weaner diet until they were transported to EMAI. When the 90 pigs arrived at EMAI at 42 ± 1 d of age (mean ± SD), they were housed in group pens with supplementary heating. A schematic representation of the time frame in relation to the different parts of the experiment is presented in Figure 1.
|
Three cereal-based pelleted diets were fed to the pigs, and the composition of each diet is presented in Table 1. The pigs were fed a grower diet (diet 1) until 133 ± 0.1 d of age (mean ± SD) then switched to a finisher diet (diet 2) until 189 ± 0.1 d of age (mean ± SD), after which they were switched to a heavy pig diet (diet 3). For each diet, the essential AA were 25% in excess of recommendations of the SCA (1987), to maximize protein accretion. Fresh feed was weighed and offered daily, but the weigh-back of unconsumed feed was conducted at 1-wk intervals. Water was provided using nipple drinkers. Each diet was offered ad libitum to each pig, and feed intake and BW were recorded at weekly intervals, on the same day of the week each time.
|
All pigs had access to feed and water in their pens before CT scanning. Before CT scanning, each pig was taken from its pen, weighed, restrained with a nose snare, and anaesthetized with a mixture of ketamine (10 mg/kg) and xylazine (1 mg/kg) by i.v. injection into the external jugular vein. The trade names for ketamine and xylazine are Ketamil (100 mg/mL) and Xylazil-100 (100 mg/mL), respectively, and are supplied by Ilium Veterinary Products, Troy Laboratories Pty. Ltd., Smithfield, New South Wales, Australia. After CT scanning, each pig was given an i.v. injection of yohimbine (5 to 10 mg) via the ear vein, to reduce the likelihood of paralytic ileus and restore intestinal peristalsis. The trade name for yohimbine is Reverzine (10 mg/mL) and is supplied by Parnell Laboratories (Australia) Pty. Ltd., Alexandria, New South Wales, Australia. Each pig was monitored until it could stand (usually within 30 min of yohimbine administration) and then walked back to its pen. The whole process of taking the pig from its pen for weighing, anesthetizing, CT scanning, and returning it to its pen took approximately 90 min. Feeding activity was usually observed within 30 min of the pig returning to its pen.
Data for this Study
The CT scan and body composition data are not the subject of this report and will be presented in subsequent papers. For this paper, the performance data for the first 10 wk were used, and they consisted of records on all the pigs (except 1 with incomplete records) on the longitudinal study component and those pigs from the serial slaughter component that were slaughtered after the first 10 wk and had complete weekly feed intake and BW records. From the 2 replicates, 26 boars, 25 gilts, and 26 barrows were used (n = 53 pigs from the longitudinal study and n = 24 pigs from the serial slaughter study). In addition, slaughter data on 107 pigs (35 boars, 36 gilts, and 36 barrows) comprised of 89 pigs (1 out of the 90 pigs was excluded due to incomplete records) from the serial slaughter study and 18 pigs (mean ± SD of 185 ± 5.8 kg of BW) from the longitudinal study slaughtered at the end of the experiment were used to examine the variability of gut fill and its effect on BW measurements. The pigs were slaughtered at EMAI, and they all had access to feed and water before slaughter. They were taken from their pens, in groups of 3, directly to slaughter.
Derivation of Traits
The first 2 wk of the study was considered as a pretest adjustment period, followed by a 56-d test period. The mean (±SD) BW of the pigs at the beginning of the test (after the adjustment period) was 46.5 ± 5.5 kg. The pretest adjustment period was necessary to allow the pigs to adjust to their individual pens and to being weighed on a weekly basis.
The traits studied were ADG (kg/d), ADFI (kg), F:G, and residual feed intake (kg/d). Each trait was calculated separately for each pig and for each of the test durations considered in the study. The growth of each pig was modeled by linear regression of BW against time (days on test). This approach was adopted because it enables the maximum amount of information to be utilized and minimizes the effect of measurement errors. The model fitted was
 |
where Y = the BW of animal i at wk t; βo = the intercept; β1 = the regression coefficient; Xit = the number of days on test; and eit = the residual error term. The regression coefficients were used to calculate ADG (β1), the BW at the beginning (βo) and end of the test, the mid-BW (i.e., the mean of the BW at the beginning and at the end), and the metabolic mid-BW (mid-BW raised to the power 0.75) from each test.
For each of the test durations, the total feed intake for each pig was divided by the number of days on test to obtain ADFI. Average daily feed intake was used in preference to total feed intake for the period so that feed intake during tests of different durations could be compared.
The F:G was calculated by dividing ADFI by ADG, to produce a value in killigrams of feed per kilogram of BW gain. This is the reciprocal of the efficiency of gain (G:F) and therefore increases as the efficiency of gain decreases and vice versa.
Residual feed intake was calculated by using ADG and metabolic mid-BW to model feed intake. A separate model was fitted for each sex/castrate status within replicate. The model fitted was
 |
where DFIi = the ADFI of animal i; βo = the intercept; β1 = the partial regression coefficient of DFI on ADG; β2 = the partial regression coefficient of DFI on metabolic mid-BW (MMWT); and ei = the residual error term. Residual feed intake was equated to the residual error term in the model.
Statistical Analysis
Each trait was first assessed by examining the repeatability of the trait over time and by evaluating different, shorter test durations for its accuracy relative to the full length 56-d test. To examine the repeatability of traits, each of the traits was computed for each of the 8 wk of the study. Pearson correlations were computed among the weekly measurements for each trait. The repeatability of ADG was examined further to see if gut fill contributed to inaccuracies in the measurement of BW, hence the low repeatability for ADG. If gut fill was proportional to BW, the regression of BW on EBW is expected to be exact, with no significant error other than measurement error (i.e., BW = β x EBW). To test the degree of variability of gut fill at slaughter, a simple linear regression through origin was used to analyze the data, first within sex/castrate status and then with the pooled data. The residual sum of squares was then tested against a 
2 value with the corresponding degrees of freedom.
To evaluate duration of test for each trait, the duration was progressively increased by 1 wk, from 7 to 56 d, with all tests commencing at d 0. Each test was assessed using 2 main criteria. One criterion used was the phenotypic residual variance for each test. This was used to determine whether adding extra data to the test was effective in reducing the amount of unexplained variation in the trait. Variance components were estimated using REML procedures in ASREML (Gilmour et al., 1999). The model fitted included the fixed effects of replicate (2 levels), sex/castrate status (3 levels), and their interaction. Random effects fitted included room, dam, and residual. To make the variance changes comparable among traits and across test duration, relative variance was computed for each trait and for each of the test durations. Relative variance was defined as the residual variance of a trait for a specified test duration, relative to the variance of the 56-d full test, expressed as a percentage, and was calculated as
 |
where RVi = the relative variance for test duration i (7-, 14-, 21-, 28-, 35-, 42-, 49-, or 56-d test); 
2i = the variance for test duration i (7-, 14-, 21-, 28-, 35-, 42-, 49-, or 56-d test); and 256 = the variance for the 56-d test. A 56-d test therefore had a relative variance of 100%.
The second criterion used to assess the accuracy of a shortened test was the phenotypic correlation of the trait measured using a shortened test with the same trait measured over the full 56-d test, which included the maximum amount of data. This criterion helps to assess the degree of reranking among pigs on a shortened relative to the full test duration. Pearson correlations were computed.
|
RESULTS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
. The means and SD for growth and feed efficiency traits for each week of the study period are presented in Table 2. The pigs had a relatively steady growth of 1.1 ± 0.17 kg/d using an average of 2.5 ± 0.35 kg of feed per day. There was a trend toward greater ADFI as the weeks progressed, which was likely a reflection of increasing BW, hence greater maintenance requirements and changes in the composition of gain of the pigs. The trend in ADFI was mirrored by a trend in F:G. At the beginning of the study when the pigs were lighter, they tended to be more efficient than later in the study when they were heavier. By definition, the mean for residual feed intake for any group of animals and for any specified period should be zero. The residual feed intake values presented in Table 2 were computed for each week separately, and although they were computed separately for each sex/castrate status within each replicate, the pooled (across sex/castrate status) means were still close to zero.
|
|
Phenotypic correlations among weekly measurements of ADG and ADFI are presented in Table 3. For ADFI, the correlations between any 2 consecutive weeks were moderate to high, indicating good repeatability for this trait. Although the magnitude of the correlations reduced progressively as weeks that were further apart were compared, all the correlations for ADFI were significantly different from zero.
|
). The residual sum of squares from the regression of BW on EBW was significant (P < 0.01) for all sex/castrate status groups (Table 4). A parallelogram analysis showed the slopes were not significantly (P = 0.07) different between sex/castrate status groups, and hence the results of the regression using the pooled data can be used to measure the variation of gut fill in the pigs. The sum of squares from the regression using the pooled data was significant (P < 0.01), indicating that gut fill did affect the accuracy in the measurement of BW. Mean gut fill was 4.2% of BW but was characterized by large variation among the pigs (SD = 1.8; CV = 42.2%).
|
|
. Correlations among weekly estimates of F:G were moderate to high, except for wk 1, in which the estimate was lowly correlated with those for the other weeks. For residual feed intake, the correlations among the weekly estimates were moderate to high, indicating a good repeatability for this trait.
|
Phenotypic residual variances for growth and feed efficiency traits for different test durations are presented in Table 6. Relative variance and phenotypic correlations of different short-duration tests compared with the full-length 56-d test for the growth and feed efficiency traits are presented in Figure 4. In general, the residual variances and relative variances decreased, whereas the phenotypic correlations increased, as the duration of test increased for each of the traits.
|
|
). With the 56-d test set at 100%, the relative variance of the 7-d test was 409% (Figure 4a). This is a reflection of the relatively low repeatability of the trait (Table 3). The relative variance was, however, characterized by a steady drop as the duration of test increased, such that the relative variance was 127% by the 35-d test, 109% by the 42-d test, and 100% by the 49-d test. The trend in relative variance indicates that there has been a massive drop up to the 35-d test and that data collection for ADG beyond 35 d resulted in relatively little improvement in accuracy. The phenotypic correlation between the 7-d test and the full-length 56-d test for ADG was only 0.13 (Figure 4b). This correlation, however, increased as the duration of test increased, such that the correlation was 0.86 by the 28-d test and 0.91 by the 35-d test. The high correlation (>0.80) indicates that, compared with a 56-d test, the degree of reranking among pigs for any test longer than that lasting 28 d was small. Based on the changes in variance and the correlations, a 35-d test for ADG in growing pigs could be recommended with a reasonable level of confidence.
For ADFI, the variance for the 7-d test was 0.097 compared with the variance of 0.063 for the full-length 56-d test (Table 6). With the 56-d test set at 100%, the relative variance of the 7-d test was 154% (Figure 4a), a value which is several magnitudes less than the 409% obtained for ADG for the same duration. This is a clear reflection of the relatively high repeatability of ADFI (Table 3). The relative variance remained virtually stable until after the 21-d test when it began to drop, to reach 100% by the 42-d test. The trend in relative variance was relatively flat, and the drop achieved with increasing the test duration was low. It is therefore not clear whether the small improvement in accuracy after the 21-d test to the 42-d test is significant enough to go beyond 7 d of data collection for feed intake. The phenotypic correlation (Figure 4b) between the 7-d test and the full-length 56-d test for ADFI was high (0.74) and continued to increase as the duration of test increased, such that the correlation was 0.83 by the 14-d test and 0.91 by the 28-d test. The high (>0.80) correlation indicates that, compared with a 56-d test, the degree of reranking among pigs for any test longer than 14 d was small. Based on the changes in variance and the correlations, a test for ADFI in growing pigs lasting 28 d could be recommended as a conservative but reasonable test duration.
The variance for F:G for the 7-d test was 0.165, whereas that for the full-length 56-d test was 0.039 (Table 6). The F:G has 2 components: ADG and ADFI. However, the changes in relative variance over the different test durations follow a pattern similar to that of ADG and not ADFI (Figure 4a). Starting off at 423% for the 7-d test, the relative variance for F:G increased initially (14-d test) before dropping sharply, such that by the 35-d test it was 147%, followed by relatively gradual decreases. It is acknowledged that the relative value of 123% for the 28-d test was less than expected, given the relatively greater values for the flanking test durations (21-d and 35-d tests). No obvious explanation was found for this anomaly. The trend in relative variance indicates that beyond 35 d there was relatively little improvement in accuracy. The phenotypic correlation between the 7-d test and the full-length 56-d test for F:G was only 0.25 (Figure 4b). This correlation, however, increased as the duration of test increased, such that the correlation was 0.85 by the 28-d test and 0.93 by the 42-d test. The high correlation (>0.80) indicates that, compared with a 56-d test, the degree of reranking among pigs for any test longer than 28 d was small. Based on the changes in variance and the correlations, a 35-d test for F:G in growing pigs could be recommended with a reasonable level of confidence.
For residual feed intake, the variance for the 7-d test was 0.054 compared with the variance of 0.028 for the full-length 56-d test (Table 6). With the 56-d test set at 100%, the relative variance of the 7-d test was 193% (Figure 4a), a value which is several magnitudes lower than those for ADG and F:G for the same test duration. Although residual feed intake is a component trait involving ADFI, metabolic mid-BW, and ADG, the changes in relative variance over the different test durations follow a pattern closer to that of ADFI and not ADG (Figure 4a). The relative variance from the shortest to the full-length test is characterized by a gradual steady decline reaching 132% by the 35-d test. The trend in relative variance indicates that beyond 35 d, there was relatively little improvement in accuracy. The phenotypic correlation (Figure 4b) between the 7-d test and the full-length 56-d test for residual feed intake was moderate (0.48) and continued to increase as the duration of test increased, such that the correlation was 0.84 by the 35-d test. The high (>0.80) correlation indicates that, compared with a 56-d test, the degree of reranking among pigs for any test longer than 35 d was small. Based on the changes in variance and the correlations, a 35-d test for residual feed intake in growing pigs could be recommended with a reasonable level of confidence.
|
DISCUSSION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Schnyder et al., 2001; Lorenzo Bermejo et al., 2003) in the scientific literature on feed intake patterns and models from the growing pig to mature BW. However, there seem to be very few reports describing optimum duration of test for feed intake and efficiency in growing pigs, with which the results can be compared. In a study by Von Felde et al. (1996), feed intake was measured in pigs in alternating weeks such that each pig had 5 weekly periods of feed intake measurements over a 10-wk period. The phenotypic correlations between wk 1 and wk 3, wk 3 and wk 5, and wk 5 and wk 7 of that study were 0.30, 0.44, and 0.90, respectively. The corresponding values from the current study are 0.69, 0.58, and 0.55. In the study by Von Felde et al. (1996), feed intake was recorded by automatic feeders, which were alternated between pens on a weekly basis. Hence, each time the feeders were reintroduced, the pigs had to adjust to feeding from the automatic feeders. This disturbance to the feeding behavior of the pigs could have been responsible for the lower correlations obtained in the earlier periods of that study relative to the current study. In a study on barrows by Rauw et al. (2006), correlations among different periods were examined. The correlations for feed intake between consecutive periods ranged from 0.36 to 0.89, whereas correlations for ADG between consecutive periods ranged from 0.20 to 0.38. Although the periods were longer than the weekly intervals used in the current study, the results in the Rauw et al. (2006) study confirm the finding that feed intake is relatively more accurate to measure in a short period than ADG. In beef cattle, Archer et al. (1997) and Wang et al. (2006) also found that the optimum duration of test for growth rate is longer than that for feed intake. Although variation in gut fill has been suspected (Kearney et al., 2004) as a major contributing factor to the lower accuracy of successive BW measurements, this study provides the first empirical evidence that variation in gut fill does influence the accuracy of BW measurements. Other factors, such as bladder fill, could influence the accuracy of measuring BW but could not be investigated in this study.
The results of this study indicate that the duration of test for ADG, and not feed intake, is the determinant of the optimum duration for testing of the feed efficiency traits. Similar results were obtained in beef cattle (Archer et al., 1997; Wang et al., 2006). Hence, any improvement in the measurement of ADG that results in a reduction of its duration of test will also reduce the duration of test for F:G and residual feed intake. Several strategies have been examined in other livestock species to reduce the duration of test for ADG. These include fasting of animals overnight or for a longer period before weighing, to reduce the variation in gut fill. This strategy has met with limited success. Working with data on cattle that were moved directly from the pens to the weighing scales, Archer et al. (1997) recommended that a 70-d test was adequate for the measurement of ADG and a 35-d test for feed intake. Using similar statistical analysis procedures for data on cattle fasted for 12 h before weighing, Archer and Bergh (2000) found that the optimum duration of test for ADG can be reduced to 42 to 56 d; however, the optimum duration of test for feed intake was 56 to 70 d. These studies in beef cattle indicate that fasting cattle before weighing improved the accuracy of measuring ADG but had a disrupting effect on the feeding patterns and feed intake of the cattle, hence negatively affecting the accuracy of measurement of feed intake.
A second strategy that has been used tries to minimize the effect of gut fill by more frequent weighing, but without disturbing the feeding behavior and feed intake of the animals. This strategy involves placement of automatic weighing platforms next to the feeders so that any time the animal goes to feed its BW is taken automatically. Thus, an individual animal may have over 5 BW measurements in a single day. Using this strategy, Kearney et al. (2004) reported that 56 d was adequate for testing for ADG in beef cattle, compared with the standard recommended 70 d.
If feed intake is the sole trait of interest, such as is required for incorporation directly into a selection index, then the 28-d test for feed intake can be used. If the interest is in feed efficiency traits, there is the opportunity to stop recording feed intake after 28 d but continue BW measurements up to 35 d, to obtain savings in management cost and greater throughput for the feed intake equipment. It should be noted that the animals should be on the same diet for the additional week when feed intake records are not being taken. They should also be managed in a similar manner as when the feed intake records were being taken, avoiding disturbance (such as mixing animals, moving them to another pen, or both) of the animals so that they are not unduly stressed and their feeding patterns are not disturbed. Another strategy that is commonly used to make maximum use of automatic feeding equipment is to measure feed intake in alternate weeks and then use a linear model to predict feed intake of the unrecorded weeks (Lorenzo Bermejo et al., 2003). This system was not evaluated in this study, but the same principle of managing the pigs in a similar manner, with or without the feeders, apply. In addition, it might be necessary to discard the feed intake data for 1 or 2 d after the automatic feeding equipment has been reintroduced to the pigs, to reduce the influence of adaptation to the feeder.
In most livestock species, including pigs, temporal changes in BW follow a sigmoid curve, which is characterized by a linear phase in the middle. This linear phase corresponds to the period of active growth in the pig (López et al., 2000). The onset and duration of the active growth period will differ among breeds and genetic lines and is influenced by the level of nutrition and whether the pigs were fed ad libitum or a restricted diet (Kusec et al., 2007). The pigs used in this study were in the active growing period and exhibited linear growth. The recommendations of this study apply to pigs in the linear phase of the growth curve, under good nutrition, and on ad libitum feeding.
In practice, whenever the efficiency of feed utilization is being considered, the traits of interest include those related to growth and feed efficiency, and a single test is usually conducted to assess these traits. Hence, any overarching recommendation on test duration should adequately cater for accurate measurement of all the growth, feed intake, and feed efficiency traits. Therefore, under good nutrition and ad libitum feeding, a 35-d test is recommended. This recommendation is based on a conservative interpretation of the results of this study. This is probably realistic, given that, for industry-wide application, the data collected from industry may not be as accurate as data used in this study, which were collected for research purposes. It is expected that these recommendations are robust enough to apply to situations in which feed is dispensed and recorded manually (as in this study) or by an automatic feeding equipment. In this study, the pigs were kept in individual pens, but in practice, pigs are penned in groups. There are reports that the performance of group-penned pigs is different from that of single-penned pigs (de Haer and de Vries, 1993). Hence, it is not known if these recommended test durations may vary in situations in which pigs are group-penned and should be the subject of further investigation.
|
Footnotes |
|---|
2 The contributions by former staff members P. J. Nicholls (experimental design) and L. J. Barker and D. Nicholson (technical assistance) are gratefully appreciated.
3 Corresponding author: paul.arthur{at}dpi.nsw.gov.au
Received for publication October 2, 2007. Accepted for publication January 9, 2008.
|
LITERATURE CITED |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. R. Giles, G. J. Eamens, P. F. Arthur, I. M. Barchia, K. J. James, and R. D. Taylor Differential growth and development of pigs as assessed by X-ray computed tomography J Anim Sci, May 1, 2009; 87(5): 1648 - 1658. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
A 
represents the 2-wk adjustment period for this study; EMAI* denotes the Elizabeth Macarthur Agricultural Institute.
), barrows (•), and gilts (
) during the test period (excluding the adjustment period).
