J. Anim Sci.
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Wellock, I. J.
Right arrow Articles by Kyriazakis, I.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Wellock, I. J.
Right arrow Articles by Kyriazakis, I.
J. Anim. Sci. 2004. 82:2442-2450
© 2004 American Society of Animal Science

ANIMAL PRODUCTION

Modeling the effects of stressors on the performance of populations of pigs1

I. J. Wellock2, G. C. Emmans and I. Kyriazakis

Animal Nutrition and Health Department, Scottish Agricultural College, West Mains Road, Edinburgh EH9 3JG, Scotland


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 Implications
 Literature Cited
 
A simulation model that predicts the effect of the social, physical, and nutritional environments on pig food intake and performance was extended to deal with individual variation. The aim was to investigate the effect of between-animal variation on the performance of a population of growing pigs. Variation was generated in initial state, growth potential, and ability to cope when exposed to social "stressors" (EX). Variation in initial state is described by initial body weight (BW0), from which the chemical composition of the pig is calculated. Variation in growth potential is described by creating variation in the genetic growth descriptors. Variation in EX exists between genotypes, where it has been suggested that leaner, more modern genotype pigs tend to be less able to cope. It is expected that within a population or group that the social environment (i.e., position within the social hierarchy) also affects an individual’s ability to cope. In the model, it is assumed that the larger, more dominant individuals are better able to cope when exposed to social stressors. Consequently, within a population, EX is correlated with body weight around the genotype mean. Model predictions showed that increasing the variation in BW0 and EX increased the variation in pig performance. This is an important practical consideration in commercial pig production, where the heterogeneity of the population at slaughter may affect the profitability of an enterprise. The way a stressor constrains performance determines whether the mean population response to a given stressor is the same as the average individual response. If all pigs in a group are affected at the same stressor intensity (e.g., all are either mixed or not), then the predicted average individual and mean population responses will be the same. If, however, the intensity of stressor at which performance becomes limiting differs between individuals (such as space allowance or temperature), differences between the individual and mean population responses will be predicted. Variation in the growth response of a population was determined to a greater extent by variation in EX and BW0 than by variation in growth potential, when pigs were housed in simulated conditions likely to be encountered in commercial environments. Consequently, decreasing the variation in initial body weight and improving ability of pigs to cope may be a better way of improving pig performance than selecting only for increased growth potential.

Key Words: Genetic Variation • Growth • Pig • Simulation Models • Stress


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 Implications
 Literature Cited
 
Models intended to simulate animal performance typically represent a single animal. Examples are for poultry (Emmans, 1981Go), cattle (Williams and Jenkins, 2003Go), sheep (Black, 1974Go; Blaxter et al., 1980Go), dairy cows (Baldwin et al., 1987Go), and growing pigs (Whittemore and Fawcett, 1976Go; Black et al., 1986Go). The assumption necessarily made is that the response of the population will be the same as that of the deterministically simulated response of the "average" individual. However, this will necessarily be the case only if all animals in the population have an equal growth potential, all are at the same stage of growth, and all react in the same way to encountered stressors. Therefore, to attempt to predict adequately the response of a population in a given environment, it is necessary to take account of between-animal variation (Fisher et al., 1973Go; Curnow, 1973Go; Emmans and Fisher, 1986Go).

The stochastic pig growth models of Ferguson et al. (1997)Go, Knap (2000a)Go, and Pomar et al. (2003)Go deal only with variation in growth potential. Any variation that may exist between individuals in initial state and ability to cope when exposed to social stressors, such as mixing, is ignored. Even under the best experimental conditions, there is likely to be variation in initial state between pigs at the start of the trial period and it is expected that variation in ability to cope exists within populations. It is thought that group composition and position within the social hierarchy will affect the ability of an individual to cope in a given social environment.

The starting point of this study was the pig growth model of Wellock et al. (2003c)Go, which predicts the effect of the social, physical, and nutritional environment on pig performance. The objective was to extend the model so that it could deal with between-animal variation in growth potential, initial state, and ability to cope when pigs were exposed to environmental stressors and to investigate the impact on performance.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 Implications
 Literature Cited
 
Model Description
The individual pig growth model of Wellock et al. (2003aGo,b)Go was later extended to include the effect of social stressors on individual pig performance (Wellock et al., 2003cGo). Here it is further developed to predict the effect of between-animal variation. The theory behind the model is outlined below.

The individual pig is described by four genetic characteristics. Three of these are used to predict its potential for growth: protein weight at maturity (Pm, kg), the ratio of lipid to protein at maturity (Lm/Pm, kg/kg), and a growth rate parameter (B, per day). The fourth parameter describes the ability to cope when exposed to social stressors (EX). The initial state of the pig is described by initial body weight (BW0, kg) from which the chemical composition of the pig is calculated assuming the pig has its ideal composition set by its genotype. The potential rate of protein retention (PRmax, kg/d) is determined by pig genotype and current protein weight only. Its value is used to determine the potential gains of the other chemical components (Emmans, 1988Go; Emmans and Kyriazakis, 1997Go; Wellock et al., 2003aGo). Potential average daily gain (ADGp, kg/d) is the sum of the potential gains of the four chemical components. Five percent of the gain is assumed to be gut fill.

It is assumed that all pigs will attempt to consume an amount of feed that will satisfy their energy and protein requirements for ADGp, maintenance, and any compensatory lipid gain as described by Wellock et al. (2003a)Go. The amount of feed that allows this to be achieved, termed the desired feed intake (FId, kg/d), is calculated from the composition of the given feed. Any costs of thermoregulation are calculated separately and any increase in requirements from exposure to pathogens is ignored. The only feed resources considered are energy and protein; any of the essential amino acids may be first-limiting. Actual feed intake and the consequent actual gains in chemical component weights are predicted taking into account the capacity of the animal to consume feed bulk, its ability to maintain thermoneutrality, and any consequences of the social environment. Gains of the chemical components are calculated by partitioning the energy and ideal protein supplies above maintenance between protein (PR, kg/d) and lipid (LR, kg/d) retention according to Kyriazakis and Emmans (1992aGo,b)Go.

The physical environment is described by the ambient temperature, air velocity, floor type, and relative humidity, and these set the maximum (HLmax, MJ/d) and minimum (HLmin, MJ/d) heat losses in the given environment. A comparison with the pigs calculated heat production (HP, MJ/d) determines whether the pig is hot (HP > HLmax), cold (HP < HLmin), or thermoneutral (HLmin < HP < HLmax). A constraint on intake will operate in hot environments owing to an inability to lose the heat produced by maintenance and growth to the surrounding environment. In cold environments, there is an extra thermal demand placed on the pig. If conditions are thermoneutral, no further action is taken.

The social environment is described by group size (N), pen area (A, square meters), feeder space allowance, (FSA, either as feeder spaces per pig or meters per pig), and the occurrence or not of mixing. The effective space allowance (SPA, m2/BW0.67) is calculated from N, A, and BW (kg). All of these factors may act as social stressors, and it is assumed in the model that they decrease performance by lowering the capacity of the animal to attain its potential for whole body growth following Emmans (1981)Go and Chapple (1993)Go. The exception is FSA, which directly constrains intake when limiting. The descriptor EX adjusts both the intensity of the stressor at which the pig becomes stressed and the extent to which each stressor reduces performance at a given stressor intensity (Wellock et al., 2003cGo). It is assumed in the model that these two factors are perfectly correlated; pigs that show signs of stress at a lower stressor intensity are also stressed to a greater degree at any given stressor intensity. Increasing the value of EX represents a decreasing ability to cope when stressed. The model was calibrated so that a unit change in EX produced deviations of approximately 1% from the mean performance.

The model can be run either to a final BW (BWf, kg) or for a given time period (t, days). For a complete description of the model including inputs, see Wellock et al. (2003aGo,c)Go.

Creating a Population
Genetic Characteristics.
The potential growth of individuals within a population can be described by generating variation around the population means of each of the genetic parameters, B, Pm, and Lm/Pm (Emmans, 1989Go; Ferguson et al., 1997Go). Between animals of the same population, there is likely to be a negative correlation between B and Pm (Emmans, 1988Go; Knap, 2000aGo; Lewis et al., 2002Go). The scaled rate parameter, B* = B. , described by Emmans and Fisher (1986)Go, is used as an alternative to B to avoid the problems caused by the correlation between B and Pm. The values of B*, Pm, and Lm/Pm are assumed to be uncorrelated and normally distributed (Ferguson et al., 1997Go; Knap 2000aGo; Pomar et al., 2003Go).

Initial State.
Individual variation in BW0 is generated from the assigned genotype mean, (µBW0, kg) and standard deviation ({sigma}BW0, kg) of BW0 using the simulated genetic parameters of the individual to correlate BW0 with potential growth. By this means, individuals in the group with the greatest potential will tend to have the highest BW0 as would be expected from nonlimiting growth. The initial weight of pig i, (BW0i, kg) was calculated as


[1]

The parameters, Pmi, and Lm/Pmi are the genetic parameter values for pig i. The parameters µB*, µPm, and µLm/Pm are the mean values, and {sigma}B*, {sigma}Pm, and µLm/Pm are the standard deviations of B*, Pm, and Lm/Pm, respectively. The parameter a is a general scaling factor; it is set at 0.6 to generate expected values. The parameters b1, b2, and b3 determine the degree of correlation between each of BW0i and , Lm/Pmi, and Pmi, respectively. All three parameters were set at unity. The value of residuali is drawn at random from a normal distribution with a mean value chosen to account for the expected variation in BW0. It adds a nongenetic component to BW0i. The initial chemical composition of each pig is calculated from BW0i assuming it has its ideal genetic composition at the start of the trial period. In this way, at the same BW0 genetically fatter pigs (higher values for Lm/Pm) will have a lower initial protein weight (P, kg) and a higher initial lipid weight (L, kg) than genetically thinner pigs.

Ability to Cope.
It is assumed in the model that there is a negative correlation between BW0 and EX. Individual values for EX (EXi) are generated around the assigned genotype mean (µEX) and standard deviation ({sigma}EX) of EX, whereas being negatively correlated with BW0. This results in EX being normally distributed with pigs with the highest BW0 tending to have the lowest values for EX.


[2]

The parameter b4 determines the degree of correlation between BW0 to EX and is set equal to 1. The residuali is drawn at random taking account of {sigma}EX. Within a population, EX is not directly correlated to leanness (Eq. [2]Go). However, leaner animals will tend to have higher EX values owing to the positive correlation between Lm/Pm and BW0 (Eq. [1]Go) and the negative correlation between BW0 and EX (Eq. [2]Go). Between populations, it is expected that modern "leaner" genotypes will have higher values of EX than traditional "fatter" genotypes (Grandin, 1994Go; Torrey et al., 2001Go; Schinckel et al., 2003Go). A value of 10 represents the mean response for the average pig type (changed from 5 in the model of Wellock et al. [2003c]Go). It is expected that 5 ≤ µEX ≤ 15 and {sigma}EX ≤ 2.5 are conditions that will hold for all populations. This is to avoid problems of generating nonpositive values for EXi.

Drawing Individual Pigs.
For each simulated pig within a population, values for

, Pmi, and Lm/Pmi are drawn at random from uncorrelated normal distributions (Ferguson et al., 1997Go; Knap 2000aGo; Pomar et al., 2003Go). Values for BW0i and EXi are then generated from their respective means and standard deviations (model inputs) while taking into account the generated genetic parameter values of the individual (Eq. [1]Go and [2]Go). The values that characterize each animal are drawn before each simulation run and are able to be maintained for multiple simulation runs.

Simulations
The model was used to simulate some relevant experimental conditions with environmental stressors as the experimental factors. Particular attention is given to the social stressors. The genetic line of van Lunen (1994)Go as characterized by Knap (2000b)Go was used in all model simulations. The estimated means and coefficients of variation (shown in brackets) for the genetic parameters, B*, Pm, and Lm/Pm were 0.0408 (0.03), 32.0 (0.07), and 1.2 (0.15), respectively. These values were kept constant throughout all model simulations.

Where model inputs are not stated below, the default values of the input variables used are those shown in Table 1Go. Other than where described, SPA, FSA, and feed bulk are expected to be nonlimiting and the temperature thermoneutral for the average animal. In all simulations, 500 animals were drawn at random. Group sizes of 20 were used in all simulations, except when the effect of N was investigated. Using 500 pigs is thus equivalent to simulating 25 replicates of 20 pigs. The simulated phenotypic variation that is predicted results from interactions between the generated variation of the individual pigs and the thermal, dietary, and social environments in which they are kept.


View this table:
[in this window]
[in a new window]
  		Table 1. Default values of the input variables used in model predictions
 
The Effect of Variation in Initial Body Weight and Ability to Cope with Social Stressors on Pig Performance
The effects of variation in BW0 and EX on the variation in t for all individuals to reach a given BWf (set BWf simulation) and the variation in BWf over a given number of days (set t simulation) were predicted. Variation ADG (kg/d), ADFI (kg/d), and body composition was also predicted. The weight range used is that of a typical finisher system from 60 to 100 kg. Pigs were mixed at 75 kg and kept with an area of 0.7 m2/pig. In the model, this space allowance becomes limiting at about 75 kg, coinciding with the occurrence of mixing. The number of days used in the simulations with a set period of time was the average number of days predicted to grow from 60 to 100 kg in the set weight simulations, t = 50 d. Variation was generated in BW0 by increasing {sigma}BW0 from 0 to 12 kg in 2-kg intervals, with {sigma}EX = 0, and in EX by increasing {sigma}EX from 0 to 2.5 in half-unit intervals, with {sigma}BW0 = 0. A value of 10 for µEX was maintained throughout. Variation was also generated in BW0 and EX simultaneously. In one case, values of 6 and 1.5 were used for {sigma}BW0 and {sigma}EX, respectively. In another, the effects of increasing {sigma}BW0 and {sigma}EX to 12 and 2.5, respectively, were simulated.

Comparison of the Average Pig Response with the Mean Population Response.
The model was used to predict the response of the average individual in the population and the mean population response. The same 500 pigs (µEX = 10 and {sigma}EX = 1) were used in each simulation, and the mean population response compared with that of the average pig (i.e., zero variation). The effects of four environmental stressors at differing intensities were simulated. First, group size was increased from 1 to 100 with a simulation interval from 20 ({sigma}BW0 = 1) kg to 60 kg. Second, space allowance was reduced by decreasing pen area in 0.02-m2 intervals from 0.80 to 0.40 m2/pig. The simulation period was for 1 d, and µBW0 and {sigma}BW0 were set at 60 and 4 kg, respectively. Third, mixing either occurred or not on d 1, and the simulation period was from 60 kg ({sigma}BW0 = 4 kg) to 100 kg. Fourth, temperature was increased from 20 to 30°C at intervals of 1°C. The simulation period was for 1 d, and µBW0 and {sigma}BW0 were set at 60 and 4 kg, respectively.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 Implications
 Literature Cited
 
The Effect of Variation in Initial Body Weight and Ability to Cope with Social Stressors on Pig Performance
Tables 2Go and 3Go show the predicted effect of within-population variation in EX and BW0 on the performance of growing pigs over a given weight range and time period, respectively. As variation in BW0 and EX increases, variation in both t (set BWf simulations) and BWf (set t simulations) increases, whereas the mean population responses remain unchanged. At zero variation in BW0 and EX, all pigs have the same initial BW and ability to cope and so all of the variation in performance is a result of differences between pigs in their potential only. This case is shown in the first line of Tables 2Go and 3Go. The standard deviation of t ({sigma}t, days) increased from 3.93 to 17.42 d as {sigma}BW0 was increased from 0 to 12 kg. For the same increase in {sigma}BW0, the value of {sigma}BWf increased from 2.85 to 12.77 kg. As {sigma}EX was increased from 0 to 2.5, {sigma}t increased from 3.93 to 5.52 d and {sigma}BWf from 2.85 to 4.13 kg.


View this table:
[in this window]
[in a new window]
  		Table 2. Effect of variation in initial BW (BW0, kg) and ability to cope (EX) on the variation in the time taken (t, days) to reach 100 kg from a mean BW0 of 60 kg. The variables P and L are the protein and lipid content, respectively. Mixing occurred at 75 kg, and pigs were given a space allowance of 0.7 m2/pig throughout
 

View this table:
[in this window]
[in a new window]
  		Table 3. Effect of variation in initial BW (BW0, kg) and ability to cope (EX) on the variation in the final BW (BWf, kg) achieved after a simulation period of 50 d from a mean BW0 of 60 kg. The variables P and L are the protein and lipid content, respectively. Mixing occurred at 75 kg, and pigs were given a space allowance of 0.7 m2/pig throughout
 
Mean values of 0.80 and 2.17 kg/d were predicted for ADG and ADFI, respectively. These were not affected by increasing {sigma}BW0 and {sigma}EX. Variation in ADFI increased with increasing variation in both BW0 and EX, whereas variation in ADG increased only with increasing variation in EX. The final body composition, as mean weights of L and P, was not affected by increasing {sigma}BW0 and {sigma}EX. Variation in P and L was predicted to increase with increases in {sigma}BW0 and {sigma}EX only in the set time simulations. Simultaneously generating variation in BW0 and EX affected variation in the performance parameters in a nonadditive manner (Tables 2Go and 3Go).

Comparison of the Average Pig Response with the Mean Population Response
Group Size.
There were no differences in the predicted response of the average pig and the mean population response for any of the performance parameters. Mean performance was predicted to decrease as N increased. Variation in performance of the mean population was predicted to increase with increasing N. An increase in {sigma}t from 2.9 to 4.6 d was predicted to occur as N was increased from 1 to 100 (results not shown).

Space Allowance.
As A decreased, the ADG and ADFI of both the average pig and the population decreased once the value of SPA fell below a critical value (SPAcrit, m2/BW0.67). The population response was predicted to differ from the response of the average pig, with a curvilinear, as opposed to a linear-plateau, response (Figure 1Go). A value of 0.66 m2/pig was predicted for the SPAcrit of the mean population compared with 0.62 m2/pig for the average pig. A decrease in the variation of the mean population response was found as SPA was decreased.



View larger version (15K):
[in this window]
[in a new window]
  		Figure 1. Predicted effect of space allowance on the ADG response of the average individual and the mean population of 60-kg pigs.

 
Mixing.
No differences were found in the responses to mixing of the average pig and the mean population. In both cases, mixing increased t from 53 to 58 d and decreased both ADG and ADFI from 0.76 to 0.69 kg/d and from 2.10 to 1.98 kg/d, respectively. Figure 2Go shows the time course of the effect of mixing on ADG for two particular individuals chosen from the population. Pig A was predicted to show a 6% decrease in ADG and take an extra 3 d to reach BWf owing to mixing. Pig B was predicted to show a 12% decrease in ADG and to need an extra 10 d to reach BWf.



View larger version (15K):
[in this window]
[in a new window]
  		Figure 2. Time course (t, days) effect of mixing on the ADG of two randomly chosen individuals (A and B) from the population. Mixing either occurred or not on d 1, and the simulation period was from an initial mean BW of 60 kg to a final BW of 100 kg.

 
Temperature.
As temperature increased, the ADG and ADFI of both the average pig and the population decreased once the value of temperature exceeded the upper critical temperature (UCT, °C). A linear-plateau response was predicted for the average pig and a curvilinear-plateau response for the population response. The performance of the mean population and average pig was predicted to decrease at 26 and 28°C, respectively. A decrease in the variation of the mean population response was predicted as temperature increased (results not shown).


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
Implications
 Literature Cited
 
Proper allowance for population variation is important when models are used to predict nutrient requirements (Fisher et al., 1973Go; Curnow, 1973Go, 1986Go), optimize pig production systems (Pomar et al., 2003Go), and devise animal breeding strategies (Knap, 1995Go). Knowledge of between-animal variation is important in commercial situations, especially for all-in-all-out systems as variation in carcass weight and composition (i.e., the heterogeneity of a group) will partly determine enterprise profitability. The aim here was to investigate the possible impact on pig performance of individual variation in potential growth and initial state and in the ability to cope when exposed to social stressors using a model.

Variation in Initial State
One of the main factors determining the degree of heterogeneity of a group at slaughter is the degree of heterogeneity at the start of the growth period. Variation in BW0 was predicted to increase both the time taken to reach a given BWf and the BWf achieved over a given time period. If there is no variation in BW0 (BW0 = 0), all pigs need to gain the same amount of weight to achieve a given BWf. As {sigma}BW0 is increased, the BW gain needed to achieve BWf varies owing to simulated differences in BW0. Similarly, over a specified number of days, individuals with greater BW0 are generally able to achieve a greater BWf than counterparts with lower BW0.

Predicted variation in body composition was affected by increasing {sigma}BW0 only in the set time simulations. This is because increasing variation in BW0 necessarily leads to increased variation in BWf over a set number of days, which in turn leads to greater variance in P and L weights. The point can be illustrated by an example. At the end of the simulation period, one individual was predicted to have a BWf of 105 kg with 18.5 and 15.9 kg of P and L, respectively. Another was predicted to have a BWf of 98 kg, with 17.3 and 15.4 kg of P and L, respectively. Variation in BW0 was predicted to affect the variation in the growth response to a substantial extent.

Variation in Ability to Cope
It has been shown in a number of studies that pigs classified as dominant tend to outperform their subordinates. This has been demonstrated when pigs are grouped (McBride et al., 1964Go; Hansen et al., 1982Go), mixed (Hessing et al., 1994Go; D’Eath, 2002Go), and when FSA is limiting (Giroux et al., 2000Go). There is also evidence that social dominance is positively correlated with BW in pigs (Brouns and Edwards, 1994Go; Erhard and Mendl, 1997Go; Drickamer et al., 1999Go; D’Eath, 2002Go). Taken together, these results suggest that the larger pigs within a group tend to be dominant and to cope better when conditions are suboptimal (i.e., when pigs are exposed to stressors). It was assumed in the model that there is a negative correlation between BW0 and EX.

Variation in EX was generated as a first step toward accounting for behavioral differences between individuals of a population and quantifying the resulting effects on population performance. This has not been achieved in previous stochastic modeling attempts, such as Knap (2000a)Go and Pomar et al. (2003)Go, where the effects of social stressors and any differences in ability to cope are absent. It was predicted that variation in the growth response of a population would be increased by the inclusion of variation in EX when pigs were housed in conditions likely to be encountered in commercial environments. Any improvement in the pigs’ ability to cope would allow a greater proportion of their potential to be attained under stressful conditions and may be a better way of improving pig performance and enterprise profitability than increasing potential alone. If increased growth rate and ability to cope are antagonistic, as suggested by Grandin and Deesing (1998)Go, then trying to increase pig performance achieved under excellent conditions (i.e., improving potential alone) may not prove to be the best selection strategy.

There is literature suggesting that the ability to cope is negatively correlated with rapid growth rate and lean content. Schinckel et al. (2003)Go noted that "pigs from populations with above-average percent carcass lean have a greater percentage reduction in live weight and carcass lean growth than pigs of average percent carcass lean" when exposed to stressors. Torrey et al. (2001)Go reported a genetic relationship between loin eye area and the ability to adjust to mixing with unfamiliar pigs. Grandin (1994)Go noted that "the appearance of highly excitable and difficult-to-handle animals appeared to coincide with the genetic selection for both rapid growth and high lean meat yield." If EX and lean growth rate are adversely correlated, there may be negative implications regarding the welfare of pigs selected for lean growth. Selection for improved lean growth rate would then indirectly lead to selection for poorer ability to cope in the population. Since EX depends in part on the structure of the group, then group selection may be necessary in order to improve the ability of animals to cope when exposed to social stressors. The experiments of Muir and Schinckel (2002)Go with quail and Muir (1996)Go and Muir and Craig (1998)Go with poultry demonstrate that selection for desirable associate effects within a group may be a means to select animals that are better adapted to their rearing environment. Any genetic correlation between EX and the growth parameters that can be evaluated could be included in the model by incorporating the covariation between the identified parameters and EX.

The effect of varying values of EX was discussed by Wellock et al. (2003c)Go. The apparent quantitative effects of different amounts of variation in EX (i.e., the value of {sigma}EX) follows from the scaling used. This was set so that a unit change in EX produced deviations of approximately 1% from the mean performance. However, it may be that the scaling made is inappropriate and that the model is too insensitive to between-animal variation in EX. More information is needed to quantify the variation of EX within a population and its effects on performance.

Individual and Mean Population Responses
Ferguson et al. (1997)Go stated that "there is a marked difference in the response of the average individual in the population and the mean population response." Pomar et al. (2003)Go demonstrated that there are clear differences between the predicted average individual and the mean population response for the rate of protein retention in response to increasing dietary protein intake. However, differences between the average pig and mean population responses should not always be expected, and it will depend partly on the stressors to which the pigs are exposed. When all individuals become adversely affected at the same stressor intensity (e.g., being housed individually as opposed to being in a group or being mixed or not), then no differences between the average individual and mean population response is to be expected. This is because all individuals are either affected or not, although this may be to varying extents. If, however, the intensity at which the stressor becomes limiting is able to differ between individuals, such as SPAcrit, critical FSA, and UCT, differences between the average individual and mean population response are expected.

The linear-plateau response of the average individual to decreasing SPA is a direct outcome of the assumption used in the model (see Wellock et al., 2003cGo for further details). The curvilinear-plateau response of the population, however, can be explained by individual differences in SPAcrit, generated from between-animal variation in BW and EX (Figure 1Go). The plateau is predicted to occur when SPA > SPAcrit for all pigs in the population and the curvilinear transition phase occurs when only a proportion of the population is constrained (i.e., SPA < SPAcrit for only some individuals). As the intensity of the stressor increases, the proportion of the population that is constrained increases until all individuals are affected. At a fixed SPA, the proportion of pigs limited will increase with increasing population variance, and this will result in a greater degree of curvature. This was demonstrated by Pomar et al. (2003)Go for average daily rate of protein retention in response to increasing protein intake. The mean population and individual responses to decreasing SPA are predicted to differ over only a small range of pen area. However, this quantitative finding may underestimate the position in commercial enterprises and could have important financial consequences when space is at a premium.

The type of stressor also influences the amount of variation around the mean population response to increasing stressor intensity. If the critical limit does not vary between individuals, then an increase in variation is expected as the intensity of the stressor increases. This is a direct result of individual differences in the ability to cope with the stressor. Conversely, a decrease in variation is expected if the critical limit is able to vary between individuals. This is because some individuals will be limited at lower stressor intensity than others, and, consequently, a narrower range of variation is expected as the intensity of the stressor increases. For example, the increased range of ADGp and FId of individuals with increased potential will not be expressed at high temperatures and low SPA because of thermal and space constraints, respectively. Individuals with a lower potential will be less affected at a given temperature and space allowance, as they will have a lower critical space allowance and higher UCT. Consequently, a narrower range of ADFI and ADG is predicted as SPA decreases and temperature increases.

Particular reasons for individual differences in ability to cope in differing thermal environments are not dealt with explicitly in the model. The way in which individual variation in BW, potential rates of gain, fatness, and levels of activity affect the between-animal variation in ability to cope with the thermal environment are simulated. These differences directly influence the individual’s upper and lower critical temperature (LCT, °C) and enable some individuals to cope better in a given thermal environment than others. For example, smaller, slower-growing, thinner, and less-active individuals will be able to cope better at high temperatures and less able to cope at low temperatures than larger, faster-growing, fatter, and more-active individuals. This is a consequence of lower UCT and higher LCT, respectively.

Future Developments
The introduction of variation in EX and BW0 into the model is likely to allow better estimates of the phenotypic variation in pig performance observed in experiments with real pigs to be derived. A comparison between model predictions and experimental data is a necessary next step to enable validation of the model. However, relevant information on populations of pigs exposed to various stressors with sufficient detail to allow validation is at present scarce in the literature.

The pig’s response to encountered stressors may be as important as the pig’s genetic potential for growth when pigs are reared in commercial conditions (Schinckel et al., 2003Go). Methods to characterize mean values of B*, Pm, and Lm/Pm have been suggested by Ferguson and Gous (1993)Go and Knap et al. (2002)Go and genetic variances estimated by Ferguson et al. (1997)Go and Knap (2000a)Go. A measure of µEX and {sigma}EX for specific populations is still required. Quantifying the variation in EX may improve the rate of breeding for improved ability to cope, as the amount of variation determines the degree of selection pressure able to be applied. Comparing the variation predicted by the model with experimental variation will allow an initial estimate of the variation in EX to be made. This was the method used by Ferguson et al. (1997)Go when predicting the variation in B*, Lm/Pm, and Pm from variability in ADG and ADFI and external estimates of the heritabilities of the traits.

Model building is an iterative process and, in this sense, is never complete (Pomar et al., 1991Go). The model described here provides a framework able to incorporate other factors when the relevant information becomes available. For example, if the simplistic assumption that individuals react in the same way to all types of social stressors is incorrect (i.e., an individual that copes well with one stressor may not necessarily cope equally well with all other stressors), then the introduction of further parameters, in addition to EX, may be required for a sufficient descriptor of ability to cope when exposed to social stressors. Furthermore, the immune response of an animal may be affected by exposure to social stressors (e.g., de Groot et al., 2001Go; Bolhuis et al., 2003Go), and, although this is ignored in the current model, it is thought that this is one area of research that should be actively pursued and included in future model developments. The main value of the model detailed here rests in its heuristic function and in the manner that it can help to direct effort in appropriate areas of research.


   Implications
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Implications
Literature Cited
 
Proper allowance for population variation is important when models are used to predict nutrient requirements, optimize pig production systems, and devise animal breeding strategies because there may be differences between the response of the individual and the mean population owing to between-animal variation. The variation in the growth response of a population was determined to a greater extent by variation in initial state and ability to cope than by variation in growth potential when pigs were simulated in conditions likely to be encountered in commercial environments. This is an important practical consideration in commercial pig production, where the heterogeneity of the population at slaughter may affect the profitability of an enterprise. Consequently, decreasing the variation in initial state and improving the ability of pigs to cope may be a better way of improving pig performance and enterprise profitability than selecting only for increased growth potential.


   Footnotes
 
1 This work was supported by the Biotechnology and Biological Sciences Research Council of the United Kingdom and the Scottish Executive, Environment, and Rural Affairs Department. Thanks are also due to D. Green for helpful programming advice and to R. D’Eath for constructive discussion. Back

2 Correspondence: Bush Estate, Penicuik, EH26 0PH (phone: +44 131 5353299; fax: +44 131 5353121; e-mail: i.wellock{at}ed.sac.ac.uk).

Received for publication October 7, 2003. Accepted for publication April 13, 2004.


   Literature Cited
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Implications
 Literature Cited
 


Baldwin, R. L., J. France, and M. Gill. 1987. Metabolism of the lactating cow. I. Animal elements of a mechanistic model. J. Dairy Res. 54:77–105.[Medline]

Beilharz, R. G., and D. F. Cox. 1967. Social dominance in swine. Anim. Behav. 15:117–122.[Medline]

Black, J. L. 1974. Manipulation of body composition through nutrition. Proc. Aust. Soc. Anim. Prod. 10:211–218.

Black, J. L., R. G. Campbell, I. H. Williams, K. J. James, and G. T. Davies. 1986. Simulation of energy and amino acid utilisation in the pig. Res. Dev. Agric. 3:121–145.

Blaxter, K. L., V. R. Fowler, and J. C. Gill. 1980. A study of the growth of sheep to maturity. J. Agric. Sci. 98:405–420.

Bolhuis, J. E., H. K. Parmentier, W. G. P. Schouten, J. W. Schrama, and V. A. Wiegant. 2003. Effects of housing and individual coping characteristics on immune responses of pigs. Physiol. Behav. 79:289–296.[Medline]

Brouns, F., and S. A. Edwards. 1994. Social rank and feeding behaviour of group housed sows fed competitively or ad libitum. Appl. Anim. Behav. Sci. 39:225–235.

Chapple, R. P. 1993. Effect of stocking arrangement on pig performance. Pages 87–97 in Manipulating Pig Production IV. E. S. Batterham, ed. Aust. Pig Sci. Assoc., Victoria.

Curnow, R. N. 1973. A smooth population response curve based on an abrupt threshold and plateau model for individuals. Biometrics 29:1–10.

Curnow, R. N. 1986. The statistical approach to nutrient requirements. Pages 79–89 in Nutrient Requirements of Poultry and Nutritional Research. C. Fisher and K. N. Boorman, ed. Butterworths, London.

de Groot, J., M. A. W. Ruis, J. W. Scholten, J. M. Koolhaas, and W. J. A. Boersma. 2001. Long-term effects of social stress on antiviral immunity in pigs. Physiol. Behav. 73:145–158.[Medline]

D’Eath, R. B. 2002. Individual aggressiveness measured in a resident-intruder test predicts the persistence of aggressive behaviour and weight gain of young pigs after mixing. Appl. Anim. Behav. Sci. 77:267–283.

Drickamer, L. C., R. D. Arthur, and T. L. Rosenthal. 1999. Predictors of social dominance and aggression in gilts. Appl. Anim. Behav. Sci. 63:121–129.

Emmans, G. C. 1981. A model of growth and feed intake of ad libitum fed animals, particularly poultry. Pages 103–110 in Computers in Animal Production. G. M. Hillyer, C. T. Whittemore, and R. G. Gunn, ed. Occasional Publ. No. 5, Br. Assoc. Anim. Sci., Penicuik, U.K.

Emmans, G. C. 1988. Genetic components of potential and actual growth. Pages 153–181 in Animal Breeding Opportunities. R. B. Land, G. Bulfield, and W. G. Hill, ed. Br. Soc. Anim. Prod. Occasional Publ. No. 12, Br. Assoc. Anim. Sci, Penicuik, U.K.

Emmans, G. C. 1989. The growth of turkeys. Pages 153–181 in Recent Advances in Turkey Science. Poultry Science Symposium 21. R. B. Land, G. Bulfield, and W.G. Hill, ed. Br. Soc. Anim. Prod., Edinburgh, UK.

Emmans, G. C., and C. Fisher. 1986. Problems in nutritional theory. Pages 9–39 in Nutrient Requirements of Poultry and Nutritional Research. C. Fisher and K. N. Boorman, ed. Butterworths, London.

Emmans, G. C., and I. Kyriazakis. 1997. Models of pig growth: problems and proposed solutions. Livest. Prod. Sci. 51:119–129.

Erhard, H. W., and M. Mendl. 1997. Measuring aggressiveness in growing pigs in a resident intruder situation. Appl. Anim. Behav. Sci. 54:123–136.

Ferguson, N. S., and R. M. Gous. 1993. Evaluation of pig genotypes. 1. Theoretical aspects of measuring genetic parameters. Anim. Prod. 56:233–244.

Ferguson, N. S., R. M. Gous, and G. C. Emmans. 1997. Predicting the effects of animal variation on growth and feed intake in growing pigs using simulation modelling. Anim. Sci. 64:513–522.

Fisher, C., T. R. Morris, and R. G. Jennings. 1973. A model for the description and prediction of the response of laying hens to amino acid intake. Br. J. Poult. Sci. 14:469–484.

Giroux, S., G. P. Martineau, and S. Robert. 2000. Relationships between individual behavioural traits and post-weaning growth in segregated early-weaned piglets. Appl. Anim. Behav. Sci. 70:41–48.[Medline]

Grandin, T. 1994. Solving livestock handling problems. Vet. Med. 89:989–998.

Grandin, T., and M. J. Deesing. 1998. Genetics and animal welfare. Pages 319–346 in Genetics and the Behavior of Domestic Animals. T. Grandin, ed. Academic Press/Elsevier, St. Louis, MO.

Hansen, L. L., A. M. Hagelso, and A. Madsen. 1982. Behavioural results and performance of bacon pigs fed ad libitum from one or several self-feeders. Appl. Anim. Ethol. 8:307–333.

Hessing, M. J. C., W. G. P. Schouten, P. R. Wiepkema, and M. J. M. Tielen. 1994. Implications of individual behavioural characteristics on performance in pigs. Livest. Prod. Sci. 40:187–196.

Knap, P. W. 1995. Aspects of stochasticity: variation between animals. Pages 165–172 in Modelling Growth in the Pig. EAAP Publ. No. 78. P. J. Moughan, M. W. A. Verstegen, and M. I. Visser-Reyneveld, ed. Wageningen Pers, the Netherlands.

Knap, P. W. 2000a. Stochastic simulation of growth in pigs: relations between body composition and maintenance requirements as mediated through protein turn-over and thermoregulation. Anim. Sci. 71:11–30.

Knap, P. W. 2000b. Time trends of Gompertz growth parameters in ‘meat-type’ pigs. Anim. Sci. 70:39–49.

Knap, P. W., R. Roehe, K. Kolstad, C. Pomar, and P. Luiting. 2002. Characterization of pig genotypes. J. Anim. Sci. 80(Suppl. 1):174. (Abstr.)

Kyriazakis, I., and G. C. Emmans. 1992a. The effects of varying protein and energy intakes on growth and body composition of pigs. 1. The effects of energy intake at constant, high protein intake. Br. J. Nutr. 68:603–613.[Medline]

Kyriazakis, I., and G. C. Emmans. 1992b. The effects of varying protein and energy intakes on growth and body composition of pigs. 2. The effects of varying both energy and protein intake. Br. J. Nutr. 68:615–625.[Medline]

Lewis, R. M., G. C. Emmans, W. S. Dingwall, and G. Simm. 2002. A description of the growth of sheep and its genetic analysis. Anim. Sci. 74:51–62.

McBride, G., J. W. James, and N. Hodgens. 1964. Social behaviour of domestic animals: IV. Growing pigs. Anim. Prod. 6:129–139.

Muir, W. M. 1996. Group selection for adaptation to multiple-hen cages: Selection program and direct responses. Poult. Sci. 75:447–458[Medline]

Muir, W. M., and J. V. Craig. 1998. Improving animal well-being through genetic selection. Poult. Sci. 77:1781–1788.[Abstract/Free Full Text]

Muir, W. M., and A. Schinckel. 2002. Incorporation of competitive effects in breeding programs to improve productivity and animal well being. Proc. 7th World Cong. Gen. Appl. Livest. Prod. 32:35–38. (Abstr.)

Pomar, C., D. L. Harris, and F. Minvielle. 1991. Computer simulation model of swine production systems 1. Modelling the growth of young swine. J. Anim. Sci. 69:1468–1488.[Abstract]

Pomar, C., I. Kyriazakis, G. C. Emmans, and P. W. Knap. 2003. Modeling stochasticity: dealing with populations rather than individual pigs. J. Anim. Sci. 81:E178–E186.[Abstract/Free Full Text]

Schinckel, A. P., M. E. Spurlock, R. T. Richert, and T. E. Weber. 2003. Modelling health status-environmental effects on swine growth. Book of Abstracts of the 54th Annu. Mtg. European Assoc. Anim. Prod.

Torrey, S., E. A. Pajor, S. Weaver, D. Kuhklers, and T. S. Stewart. 2001. Effect of genetic selection for loin-eye area on behaviour after mixing in Landrace pigs. Page 88 in Proc. 35th Intl. Conf. ISAE. J. P. Garner, J. A. Mench, and S. P. Heekin, ed. Center for Animal Welfare, Univ. of California, Davis.

Van Lunen, T. A. 1994. A study of the growth and nutrient requirements of highly selected pigs. Ph.D. Thesis Univ. Nottingham, U.K.

Wellock, I. J., G. C. Emmans, and I. Kyriazakis. 2003a. Modelling the effects of thermal environment and dietary composition on pig performance: model logic and concepts. Anim. Sci. 77:255–266.

Wellock, I. J., G. C. Emmans, and I. Kyriazakis. 2003b. Modelling the effects of thermal environment and dietary composition on pig performance: model testing and evaluation. Anim. Sci. 77:267–276.

Wellock, I. J., G. C. Emmans, and I. Kyriazakis. 2003c. Predicting the consequences of social stressors on pig food intake and performance. J. Anim. Sci. 81.

Whittemore, C. T., and R. H. Fawcett. 1976. Theoretical aspects of a flexible model to simulate protein and lipid growth in pigs. Anim. Prod. 45:437–445.

Williams, C. B., and T. G. Jenkins. 2003. A dynamic model of metabolizable energy utilization in growing and mature cattle. I. Metabolizable energy utilization for maintenance and support metabolism. J. Anim. Sci. 81:1371–1381.[Abstract/Free Full Text]


This article has been cited by other articles:


Home page
 J ANIM SCIHome page
A. P. Schinckel, D. C. Mahan, T. G. Wiseman, and M. E. Einstein
Growth of protein, moisture, lipid, and ash of two genetic lines of barrows and gilts from twenty to one hundred twenty-five kilograms of body weight
J Anim Sci, February 1, 2008; 86(2): 460 - 471.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Wellock, I. J.
Right arrow Articles by Kyriazakis, I.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Wellock, I. J.
Right arrow Articles by Kyriazakis, I.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS