Modeling the response of growing broiler chickens

MISSING IMAGE

Material Information

Title:
Modeling the response of growing broiler chickens methodology for the evaluation of alternate feed ingredients under a variety of environmental and economic situations
Physical Description:
Book
Language:
English
Creator:
Fattori, Thomas Richard, 1950-
Publisher:
s.n.

Subjects

Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Statement of Responsibility:
by Thomas Richard Fattori.
Funding:
Florida Historical Agriculture and Rural Life

Record Information

Source Institution:
Marston Science Library, George A. Smathers Libraries, University of Florida
Holding Location:
Florida Agricultural Experiment Station, Florida Cooperative Extension Service, Florida Department of Agriculture and Consumer Services, and the Engineering and Industrial Experiment Station; Institute for Food and Agricultural Services (IFAS), University of Florida
Rights Management:
All rights reserved, Board of Trustees of the University of Florida
Resource Identifier:
aleph - 001020614
oclc - 17886157
notis - AFA2028
System ID:
UF00054862:00001

Table of Contents
    Front Cover
        Front Cover
    Title Page
        Page I
    Dedication
        Page II
    Acknowledgement
        Page III
    Table of Contents
        Page IV
        Page V
    List of Tables
        Page VI
        Page VII
    List of Figures
        Page VIII
        Page IX
        Page X
    Abstract
        Page XI
        Page XII
    Introduction
        Page 1
        Problematic situation
            Page 2
        Researchable problem
            Page 3
        Hypotheses
            Page 4
        Experimental objectives
            Page 5
        Research emphasis
            Page 6
            Page 7
    Literature review
        Page 8
        Experimental design
            Page 8
            Page 9
        Relevance to farming systems research and extension
            Page 10
            Page 11
        Environmental considerations
            Page 12
            Page 13
            Page 14
            Page 15
            Page 16
    Materials and methods
        Page 17
        Experimental methods and procedure
            Page 17
        Feed form, formulation and treatments
            Page 18
            Page 19
            Page 20
            Page 21
        Simulation of a farm environment
            Page 22
        Environmental factors
            Page 23
            Page 24
            Page 25
            Page 26
            Page 27
    Partitioning broiler growers into homogeneous production environments
        Page 28
        Introduction
            Page 28
        Materials and methods
            Page 29
            Page 30
            Page 31
            Page 32
        Results and discussion
            Page 33
            Page 34
            Page 35
            Page 36
            Page 37
            Page 38
            Page 39
            Page 40
            Page 41
            Page 42
            Page 43
            Page 44
            Page 45
            Page 46
            Page 47
            Page 48
            Page 49
            Page 50
            Page 51
            Page 52
            Page 53
            Page 54
            Page 55
            Page 56
            Page 57
            Page 58
            Page 59
            Page 60
            Page 61
            Page 62
            Page 63
            Page 64
            Page 65
    Estimation of a reliable broiler production funtion when using alternate feed ingredients
        Page 66
        Introduction
            Page 66
        Data and methods
            Page 67
            Page 68
            Page 69
        Results and discussion
            Page 70
            Page 71
            Page 72
            Page 73
            Page 74
            Page 75
            Page 76
            Page 77
            Page 78
            Page 79
            Page 80
            Page 81
            Page 82
            Page 83
            Page 84
            Page 85
            Page 86
    Economic analysis of broiler production when using alternate feed ingredients
        Page 87
        Introduction
            Page 87
        Materials and methods
            Page 88
            Page 89
            Page 90
            Page 91
            Page 92
            Page 93
            Page 94
            Page 95
            Page 96
            Page 97
            Page 98
            Page 99
            Page 100
            Page 101
        Results and discussion
            Page 102
            Page 103
            Page 104
            Page 105
            Page 106
            Page 107
            Page 108
            Page 109
            Page 110
            Page 111
            Page 112
            Page 113
            Page 114
            Page 115
            Page 116
            Page 117
            Page 118
            Page 119
    Summary and conclusions
        Page 120
        Problematic situation
            Page 120
        Partitioning environments
            Page 121
            Page 122
            Page 123
        Modeling the response
            Page 124
            Page 125
        Economic significance
            Page 126
            Page 127
            Page 128
    Experimental data
        Page 129
        Page 130
        Page 131
        Page 132
        Page 133
        Page 134
    Reference
        Page 135
        Page 136
        Page 137
        Page 138
        Page 139
    Biographical sketch
        Page 140
        Page 141
    Signature page
        Page 142
Full Text










MODELING THE RESPONSE OF GROWING BROILER CHICKENS:
METHODOLOGY FOR THE EVALUATION OF ALTERNATE FEED INGREDIENTS
UNDER A VARIETY OF ENVIRONMENTAL AND ECONOMIC SITUATIONS -






By








THOMAS RICHARD FATTORI


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF MASTER OF SCIENCE


UNIVERSITY OF FLORIDA


1987














MODELING THE RESPONSE OF GROWING BROILER CHICKENS:
METHODOLOGY FOR THE EVALUATION OF ALTERNATE FEED INGREDIENTS
UNDER A VARIETY OF ENVIRONMENTAL AND ECONOMIC SITUATIONS






By








THOMAS RICHARD FATTORI


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF MASTER OF SCIENCE


UNIVERSITY OF FLORIDA


1987
































To Maisha, Jonathan and Benjamin with the future in mind.















ACKNOWLEDGMENTS


The author wishes to express his most sincere appreciation to his

major advisor, Dr. F. Ben Mather, for his guidance throughout the

author's Master of Science program. Dr. Mather's steadfast interest

and support of the selected coursework and research project ensured

the completion of this learning experience.

Special appreciation is extended to Dr. Joseph H. Conrad and Dr.

Peter E. Hildebrand for their encouragement, technical advisement and

tolerance in guiding me throughout the research program. Without

their continued and open confidence in the author and the program, the

motivation to venture out of the traditional circuit would have been

lost.

Sincere gratitude is extended to Dr. D. Comer for her advice and

guidance in the analysis of the experimental data and to Dr. Nelson

Ruiz for his technical advice and personal encouragement.

Sincere appreciation is expressed to the author's parents, Mr.

and Mrs. L. A. Fattori, for their support and vote of confidence in

the decision to return to school.

The author wishes to extend his deepest appreciation to his wife

Maisha for her care and support of their home and family which freed

the many hours needed to complete this program. Without her patient

understanding this endeavor would have never been completed.















TABLE OF CONTENTS


Page

ACKNOWLEDGMENTS ................................................ iii

LIST OF TABLES ......................... .......... ............ vi

LIST OF FIGURES ................................................ viii

ABSTRACT .............................. ........ ............... xi

CHAPTERS

I INTRODUCTION ......................................... 1

Problematic Situation ................... ............ 2
Researchable Problem ................................ 3
Hypotheses .......................................... 4
Experimental Objectives .............................. 5
Research Emphasis ................. ................. 6

II LITERATURE REVIEW ................................... 8

Experimental Design ..... ..................... ........ 8
Relevance to Farming Systems Research and Extension .. 10
Environmental Considerations ......................... 12

III MATERIALS AND METHODS ................................ 17

Experimental Methods and Procedures .................. 17
Feed Form, Formulation and Treatments ................ 18
Simulation of a Farm Environment ..................... 22
Environmental Factors ................................ 23

IV PARTITIONING BROILER GROWERS INTO
HOMOGENEOUS PRODUCTION ENVIRONMENTS .................. 28

Introduction ........ .................. ............. 28
Materials and Methods ................................ 29
Results and Discussion .............................. 33












V ESTIMATION OF A RELIABLE BROILER PRODUCTION FUNCTION
WHEN USING ALTERNATE FEED INGREDIENTS ................ 66

Introduction ................................. ........ 66
Data and Methods ..................................... 67
Results and Discussion ................... ......... 70

VI ECONOMIC ANALYSIS OF BROILER PRODUCTION
WHEN USING ALTERNATE FEED INGREDIENTS ................ 87

Introduction ...................... .......... .. 87
Materials and Methods .......... .................... 88
Results and Discussion ............................... 102

VII SUMMARY AND CONCLUSIONS .............................. 120

Problematic Situation ................................ 120
Partitioning Environments ............................ 121
Modeling the Response ................................ 124
Economic Significance ................................ 126

APPENDIX EXPERIMENTAL DATA ................................... 130

REFERENCES .. .... .. ... ................ ........ ................ 135

BIOGRAPHICAL SKETCH ............................................ 140















LIST OF TABLES


Table Page

3-1. Composition of the experimental diets ................... 19

3-2. Composition of the vitamin-mineral premix ............... 20

3-3. Nutrient values used to calculate the diets ............. 21

3-4. Calculated amino acid profile of the diets .............. 22

3-5. Averaged weekly range of temperatures by replication .... 26

3-6. Mortality and culls by environment, sex and feed form ... 27

4-1. Broiler gain response to protein treatments for
pellet form at weeks 4, 6, and 8 in kg/pen (complete
data set) ............................................... 34

4-2. Broiler gain response to protein treatments for
mash form at weeks 4, 6, and 8 in kg/pen (complete
data set) ............................................... 35

4-3. Linear equations for modified stability analysis
calculated from broiler gain (complete data set) ........ 37

4-4. Broiler gain response to protein treatments for pellet
form at weeks 4, 6, and 8 in kg/pen (partitioned data
set) .................................................... 42

4-5. Broiler gain response to protein treatments for mash
form at weeks 4, 6, and 8 in kg/pen (partitioned data
set) .................................................... 43

4-6. Distribution of confidence intervals of broiler gain
to protein treatments for pellet form at weeks 4, 6,
and 8 in kg/pen (complete data set) ..................... 48

4-7. Distribution of confidence intervals of broiler gain to
protein treatments for mash form at weeks 4, 6, and 8
in kg/pen (complete data set) ........................... 49









Table Page

4-8. Distribution of confidence intervals of broiler gain to
protein treatments for pellet form at weeks 4, 6, and 8
in kg/pen (partitioned data set) ........................ 57

4-9. Distribution of confidence intervals of broiler gain to
protein treatments for mash feed at weeks 4, 6, and 8
in kg/pen (partitioned data set) ........................ 59

4-10. Ranking of the graphed distribution of confidence
intervals to protein treatments for feed form and
environments ...................... ..................... 63

5-1. Estimation of the coefficients of regression for
broiler gain to protein treatments and feed form from
3 to 10 weeks and under varying temperatures (estimates
+ standard error) .................................. .... 71

5-2. Regression analysis of variance for broiler gain to
protein treatments and feed from at 3 to 10 weeks and
under varying temperatures .............................. 71

5-3. Estimation of the coefficients of regression for broiler
feed intake to protein treatments and feed form from 3
to 10 weeks and under varying temperatures (estimates +
standard error) ............................... ......... 72

5-4. Regression analysis of variance for broiler feed intake
to protein treatments and feed form from at 3 to 10
weeks and under varying temperatures ................... 72

5-5. Estimates of the coefficients of regression for broiler
gain when fed milo and peanut meal in mash and pellet
form from 1 to 10 weeks and under varying temperatures
(estimates + standard error) ............................ 73

5-6. Regression analysis of variance for broiler gain when
fed milo and peanut meal in mash and pellet form from
1 to 10 weeks of age and under varying temperatures ..... 73

6-1. Optimum production for varying economic and
environmental stations in mash or pelleted form kg
of gain per square meter ................................ 110

6-2. An example of recommended feeding programs for
environment-specific broiler growers at a fixed
economic situation ...................................... 115


vii















LIST OF FIGURES


Figure Page

3-1. Broiler house pen arrangement showing good and poor
environments, completely randomized blocks of eight feed
treatments representing a growers farm and examples of
the location of feed treatments. M and P indicate mash
and pellet form, respectively ........................... 24

4-1. Broiler gain response to protein treatments for mash
and pelleted feed form at weeks 4, 6, and 8 (complete
data set) ................................................ 36

4-2. Broiler gain response to protein treatments for mash
form at 4 weeks of age as influenced by environment ..... 38

4-3. Broiler gain response to protein treatments for pellet
form at 4 weeks of age as influenced by environment ..... 38

4-4. Broiler gain response to protein treatments for mash
form at 6 weeks of age as influenced by environment ..... 39

4-5. Broiler gain response to protein treatments for pellet
form at 6 weeks of age as influenced by environment ..... 39

4-6. Broiler gain response to protein treatments for mash
form at 8 weeks of age as influenced by environment ..... 40

4-7. Broiler gain response to protein treatments for pellet
form at 8 weeks of age as influenced by environment ..... 40

4-8. Broiler gain response to protein treatments for pellet
form at 4, 6, and 8 weeks based on good and poor
environments ............................ ............... 46

4-9. Broiler gain response to protein treatments for mash
form at 4, 6, and 8 weeks based on good and poor
environments .......................................... 47

4-10. Distribution of confidence intervals for broiler gain
to protein treatments for mash feed at week 4.
Complete data set (a), poor environment (b), and good
environment (c) ......................................... 50


viii









Figure Page

4-11 Distribution of confidence intervals for broiler gain
to protein treatments for pellet feed at week 4.
Complete data set (a), poor environment (b), and good
environment (c) ......................................... 51

4-12. Distribution of confidence intervals for broiler gain
to protein treatments for mash feed at week 6.
Complete data set (a), poor environment (b), and good
environment (c) ............. .... ...... ....... .... ... 52

4-13. Distribution of confidence intervals for broiler gain
to protein treatments for pellet feed at week 6.
Complete data set (a), poor environment (b), and good
environment (c) ........................ .... .... ...* 53

4-14. Distribution of confidence intervals for broiler gain
to protein treatments for mash feed at week 8.
Complete data set (a), poor environment (b), and good
environment (c) ......................................... 54

4-15. Distribution of confidence intervals for broiler gain
to protein treatments for pellet feed at week 8.
Complete data set (a), poor environment (b), and good
environment (c) .............. ......... ........ .......... 55

5-1. Broiler gain response surfaces to protein treatments for
pellet feed in a good environment by weeks of age ....... 75

5-2. Broiler gain response surfaces to protein treatments for
pellet feed in a poor environment by weeks of age ....... 76

5-3. Broiler gain response surfaces to protein treatments for
mash feed in a good environment by weeks of age ......... 77

5-4. Broiler gain response surfaces to protein treatments for
mash feed in a poor environment by weeks of age ......... 78

5-5. Predicted broiler gain response surface with milo and
peanut meal inputs for pellet form in a good
environment ............................ .............. 81

5-6. Predicted broiler gain response surface with milo and
peanut meal inputs for pellet form in a poor
environment .... .... ... ................. ............. 82

5-7. Predicted broiler gain response surface with milo and
peanut meal inputs for mash form in a good environment .. 83










Figure rage

5-8. Predicted broiler gain response surface with milo and
peanut meal inputs for mash form in a poor environment .. 84

6-1. Broiler gain isoquants derived from milo and peanut meal
feed inputs for pellet form in a poor environment ....... 91

6-2. Protein treatment lines depicting a fixed proportion of
milo and peanut meal, and broiler gain isoquants for
pellet form in a poor environment ....................... 93

6-3. Isocost lines based on a fixed price ratio of 2.5:1
(price of PM to price of milo) and corresponding gain
isoquants for pellet form in a poor environment ......... 99

6-4. Least-cost expansion path based on a fixed price ratio
of 2.5:1 for pellet form and in a poor environment ...... 104

6-5a. Location of maximum production and maximum profit at a
fixed price ratio of inputs; least-cost feeding program
for pellet form in a poor environment ................... 107

6-5b. Location of maximum production and maximum profit at
a fixed price ratio of inputs; time isoquants (age)
derived for pellet form in a poor environment ........... 108

6-6a. Broiler feeding program at a fixed price ratio;
pellet form in a good environment ....................... 112

6-6b. Broiler feeding program at a fixed price ratio; time
isoquants (age) derived for pellet form in a good
environment .................. .......... ... ....... 113

6-6c. Broiler feeding programs at a fixed price ratio;
comparison of the programs developed for good and
poor environments of milo and peanut meal in pellet
form .................................. ................. 114

6-7a. Mash form in a good environment; broiler gain isoquants
derived from milo and peanut meal feed inputs ........... 116

6-7b. Mash form in a good environment; derived time
isoquants (age) ........................ ........... .... 117

6-8a. Mash form in a poor environment; broiler gain isoquants
derived for milo and peanut meal feed inputs ............ 118

6-8b. Mash form in a poor environment; derived time
isoquants (age) ........................................ 119















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science


MODELING THE RESPONSE OF GROWING BROILER CHICKENS:
METHODOLOGY FOR THE EVALUATION OF ALTERNATE FEED INGREDIENTS
UNDER A VARIETY OF ENVIRONMENTAL AND ECONOMIC SITUATIONS

By

Thomas Richard Fattori

August, 1987


Chairman: F. Ben Mather
Major Department: Poultry Science


If world broiler production is to meet its share of demand for

animal protein then more efficient ways to utilize feed resources and

more appropriate feeding technologies suited for specific environments

must be found. A methodology rooted in the farming systems approach

to technology generation was tested through a simulation study on the

broiler gain response to alternate feed ingredients under a variety of

environmental and economic situations.

The use of modified stability analysis combined with the graphic

frequency distribution of confidence intervals was demonstrated to be

an efficient tool for 1) quantifying the differences between broiler

growers due to environment (temperature), 2) partitioning growers into

homogeneous production environments (good and poor environments), 3)

detecting stability of technologies (protein levels) across
xi








environments, 4) evaluating trends in production due to protein

treatment levels and 5) identifying discrepancies in the hypothesized

versus actual response to treatments in a good or poor environment.

Procedures described in the use of response surface methodology

were followed so a reliable estimate of the broiler gain response

function to the alternate feed ingredients milo and peanut meal could

be defined. A technical model was developed to describe the

biological response of broilers (live-weight gain) to varying levels

of dietary protein and feed form at specific ages and in good and poor

environments. Then, an efficiency model was established describing

broiler gain as a function of cumulative milo and peanut meal intake,

feed form and environment. The models demonstrated that 96 and 98% of

the variation in live-weight gain was explained by the terms in the

technical and efficiency models, respectively.

The estimated model for efficiency was then used to determine the

economic significance of different feeding programs recommended for

particular environments. Broiler gain isoquants were derived from the

experimental data and were considered to be representative of the

underlying biological relationship of broiler gain as a function of

the cumulative intake of milo and peanut meal while exposed to

different environmental temperatures. Economic analysis was then

performed by superimposing a variety of economic considerations on

this estimated biological response. It was demonstrated how a

schedule of recommendations that optimize the use of feed resources in

the production of broiler gain could be established for varying

environmental and economic situations.















CHAPTER I

INTRODUCTION


Growth of the world's poultrymeat sector was a remarkable 62

percent for the ten year period covering 1974 to 1984. In volume

terms this expansion translated into an additional million tons of

poultry meat in the world marketplace each year (Anon., 1986). Over

the past 20 years the production of poultrymeat and eggs in developing

countries has grown faster than for any other major food. Between

1970 and 1980, beef, sheep, goat, and milk production increased less

than three percent per year. Pork production rose four percent while

poultrymeat and eggs increased by nearly six and four percent per

year, respectively (Krostitz, 1984). Reasons for the rise in poultry

meat production are attributed to declining prices for poultry

products relative to other animal products, population and income

growth, and urbanization. All those trends are expected to continue

into the future. Between now and the year 2000 poultry products are

the most likely source of animal protein to keep pace with demand in

the developing countries of the world.










Problematic Situation



To meet this agenda world broiler producers, especially

developing country producers, must find more efficient ways to utilize

available feed resources and develop feeding technologies suited for

specific achievable environments (Fetuga, 1983). The underproduction

of cereals and tubers and the resulting human food shortages have made

the use of agro-industrial by-products an attractive proposition in

many developing countries. Even if the use of these by-products

results in some depression in rate of live-weight gain and poorer feed

efficiency, the true test of their potential is when they are examined

in relation to their overall economic advantage of possible reductions

in feed cost and acceptable profits for the producer.

Numerous review and scientific articles describing the

constraints on poultry production in developing countries emphasize

that feed quality and quantity are the major constraints to increased

production (Ikpi and Akinwumi, 1981; Rao, 1982; Leong and Jalaludin,

1982; Ballard, 1985; Adeyemo, 1986; Randall, 1985; and Wilson, 1986).

Feed costs are the major component of total production costs for

broilers. In the U.S. feed costs represent over 60 percent of total

production costs. In developing countries the percentage is likely to

be higher.

The major objectives of broiler producers are to optimize

live-weight gain per unit of feed input (biological response) and to

maximize profit (economic returns). Together these objectives imply

that optimum nutrient levels in a broiler feeding program are an









3

economic as well as biological question. This optimization problem is

especially complex in developing countries where wide fluctuations in

availability and cost of feed ingredients as well as the price

received for broiler meat occur not only rapidly but in an

unpredictable fashion.

One of the greatest challenges to poultry researchers in the

developing countries of the world is to find new feeding technologies

that are more appropriate to a complex and highly variable local

production environment. They typically work in a research environment

with a great deal more socio-economic as well as agro-biological

variability than is found in the temperate, developed countries of the

world.

Although the basis for most feeding recommendations comes from a

general pool of knowledge, poultry consultants generally agree that

the best recommendations often are those that are custom-fit to the

immediate situation. This is especially true when the producer

operates in an environment of high variability.



Researchable Problem



To assist the broiler producer in making production decisions,

especially in matters concerning substitution rates of alternative

feed ingredients, a methodology is needed that is capable of

generating technical recommendations that are more efficient in their

application and more appropriate for the clients in need of their use.

Feeding recommendations are more efficient if they consider a range of








4

possible economic situations that broiler growers could experience and

more appropriate if they are made for environment-specific growers.



Hypotheses



Hypothesis 1

If the differences in environment on broiler growers' farms can

be characterized and quantified then these farms can be partitioned

into homogeneous production environments. The purpose of such a

partitioning is to increase the efficiency of feeding recommendations

by formulating feeds specific to particular production environments.



Hypothesis 2

If the functional form of the broiler gain response to alternate

feed ingredients (in this case milo and peanut meal) can be defined

mathematically (modeled) then predictions of broiler gain from these

ingredients can be made. A model that describes the efficiency of

alternate feeds in broiler production could be estimated for a range

of feed treatments (protein levels) and include temperature as a

variable in the production process. This model can then be used to

predict gain for a particular environment (temperature) and for a wide

range of milo and peanut meal combinations.



Hypothesis 3

If the estimated model proves to be an accurate description of

the broiler gain response to the alternate ingredients then this model










can be used in the derivation of a range of feeding programs for

varying economic and environmental situations. By developing

appropriate feeding programs for environment-specific growers the

efficient use of feed resources and profit from production are

maximized.



Experimental Objectives



The research project had the following experimental objectives.



Regarding Hypothesis 1

1. Demonstrate how levels of gain from all treatments can be

averaged into an environmental index providing a means for quantifying

the differences in broiler growers' environments.

2. Illustrate how broiler growers can be partitioned into

relatively homogeneous production environments.

3. Illustrate how the stability of different feed technologies

(protein levels) is affected by environment.

4. Identify production trends from different feed technologies

over time.

5. Identify a need to recharacterize the production environment

if an environment by treatment interaction is not consistent with the

hypothesized interaction.











Regarding Hypothesis 2

1. Estimate a reliable experimental production function

(efficiency model) of the broiler gain response to the alternate feed

ingredients milo and peanut meal, under varying environmental

situations.

2. Develop a biological model of the broiler gain response to

protein treatments with age and under varying environmental

situations.



Regarding Hypothesis 3

1. Derive a range of least-cost and least-time feeding programs

when feeding milo and peanut meal in mash or pellet form.

2. Demonstrate how the optimal protein concentration of a diet

can be determined for environment-specific growers in response to

changing feed ingredient costs and broiler meat prices.



Research Emphasis



The main emphasis of this research project concerned the testing

of the prescribed methodology for evaluating alternate feed

ingredients under varying economic and environmental situations rather

than the specific results obtained. When an accurate description of

the nutrient profile of a feed ingredient is not known due to

variability in storage, humidity, processing or in the product itself,

then potential errors in feed formulation may result, especially if

these feed ingredients are fed under varying environmental situations.








7

Therefore, nutritional considerations in this study were allowed to be

consistent with circumstances such as these in order that they reflect

a situation common to a developing country.

Clearly, the client for whom this research endeavor was conducted

is the broiler producer and/or researcher world-wide, with emphasis on

problems associated with developing countries where the factors that

impact on broiler production are more varied and difficult to manage.















CHAPTER II

LITERATURE REVIEW


Experimental Design



Recent studies conducted in Nigeria (Olomu and Offiong, 1980a and

1980b; and Onwudike, 1983) address the problem of a shortage of

research on the effects of protein and energy on the performance of

broiler chickens in the tropics. Also, they recognize the importance

of investigating the use of alternate feed ingredients (peanut meal,

wheat middlings and fish meal) as well as the cost/benefit

implications of using these ingredients for the local poultry

producer. These references, as well as many others from developing

countries, use the traditional approach of research by evaluating the

effect of one variable (protein or energy) while holding all other

variables constant. The shortcoming of such an experimental design is

in the failure to account for such variability as feed quality,

temperature, water quality and management practices all of which

researchers acknowledge have a significant influence on broiler

production. Also, the relative inefficiencies of these designs in the

application of economic analysis needs to be underscored.

Morton and Wyllie (1983) explained that often in the tropics no

theory exists as to how the birds should or do respond to treatments















CHAPTER II

LITERATURE REVIEW


Experimental Design



Recent studies conducted in Nigeria (Olomu and Offiong, 1980a and

1980b; and Onwudike, 1983) address the problem of a shortage of

research on the effects of protein and energy on the performance of

broiler chickens in the tropics. Also, they recognize the importance

of investigating the use of alternate feed ingredients (peanut meal,

wheat middlings and fish meal) as well as the cost/benefit

implications of using these ingredients for the local poultry

producer. These references, as well as many others from developing

countries, use the traditional approach of research by evaluating the

effect of one variable (protein or energy) while holding all other

variables constant. The shortcoming of such an experimental design is

in the failure to account for such variability as feed quality,

temperature, water quality and management practices all of which

researchers acknowledge have a significant influence on broiler

production. Also, the relative inefficiencies of these designs in the

application of economic analysis needs to be underscored.

Morton and Wyllie (1983) explained that often in the tropics no

theory exists as to how the birds should or do respond to treatments










such as protein and energy levels from alternate feeds. Therefore,

researchers should explore those empirical relationships which will

give production responses over a wide range treatments. Experimental

designs incorporating production functions or response surfaces are

the most useful in the exploration of likely agro-economic situations

(Dillon, 1966 and 1977; Heady and Dillon, 1961; Heady and Shashanka,

1983).

Response surface methodology (RSM) was described by Roush et al.

(1979) as the integration of experimental strategy, mathematical

methods and statistical inference. Together these concepts can be

used to efficiently make empirical explorations of the relationships

among variables that are of interest to the broiler producer. This

approach is different from the more traditional procedure of varying

one factor and holding the rest constant in that it results in the

simultaneous solution of several factors to find the quantitative

levels which will give the most desirable (economic) response.

Roush (1982) applied the RSM technique in the investigation of

protein levels for broiler starter and finisher diets and demonstrated

the power of this methodology in the determination of the optimal

biological response and economic returns of a range of potential

feeding programs for broilers. Furthermore, by using a three

dimensional central composite design (Cochran and Cox, 1957) along

with the RSM technique, far more scientific information can be

generated and evaluated with an equivalent amount of facilities,

animals, money and trained personnel than by using traditional

complete factorial designs. This point is particularly relevant to










developing countries where such resources are a normal constraint on

research programs. Since the objectives of Roush (1982) were to

quantitatively examine the combination of starter and finisher protein

levels and the time of ration change to produce optimal biological and

economic response for broiler males, environment as a variable that

affects production was not considered. The broilers were confined to

battery brooders on raised wire floors and did not experience the

environmental pressures (temperature, humidity, management, water

quality, etc.) that would be characteristic of a real broiler growout

farm.



Relevance to Farming Systems Research and Extension



The Farming Systems Research and Extension (FSR/E) approach to

technology generation and promotion is viewed by many as the state of

the art in applied agricultural research and development (Hildebrand,

1986). One of the tools that is often used in an FSR/E program is

modified stability analysis (MSA) (Hildebrand, 1984; Hildebrand and

Poey, 1985). MSA is a design and analysis technique that incorporates

the variability among broiler growers into the experimental

methodology. Often in developing countries there is a great deal of

variability among broiler growers due to differences in management

practices, housing design, altitude, ambient temperature and other
"environmental" factors. MSA is used because of its value in making

production oriented recommendations to specific producers that have

been characterized and then classified by their common or more











homogeneous environment. Environment can then be treated as a

continuous quantifiable variable whose range, as in the case of a

broiler study, is the range of cumulative live-weight gains for a

number of feed treatments at a particular location. This measure is

related to environment by simple linear regression.

The FSR/E approach suggests that a targeted region, for which and

perhaps in which research is to be conducted, be identified as a

research domain (Hildebrand, 1986). Within this research domain the

biophysical or agro-climatic attributes will span a range of

production environments that will impact on broiler production in

different ways. In the U.S. a research domain is likely to be

identified as the region around a vertically integrated broiler

producer and encompasses all the growers associated with that

integrator. Lacy and Benoff (1986) have demonstrated that by

examining the production records of 250 broiler growers for a year the

20 top (good) and 20 bottom (poor) performers are differentiated by

their ability to control the broiler environment.

These findings imply that the agro-socioeconomic conditions of an

individual grower has significant economic implications for both the

grower and the integrator. Lacy and Benoff (1986) reported that the

differences between broiler production on the good and poor farms, in

terms of feed conversion, final body weight, livability, uniformity

and condemnations, were all of significant economic importance.

Therefore, the FSR/E approach suggests that the integrators

research domain should be partitioned into recommendation domains

(Hildebrand, 1986) where the partitioning defines a group of growers








12

as having roughly homogeneous production environments. The purpose of

this partitioning of the research domain into one or more

recommendation domains is to enable the integrator to develop more

appropriate feeding programs for environment-specific growers. The

development of environment-specific feeding programs would add to the

efficient use of feed resources and the profits of both the grower and

integrator.



Environmental Considerations



Temperature and Production

House temperature is a principal factor in the matrix of

environmental variables that affects the economic optimization of

broiler production. Charles (1986), in his review of the literature

on temperature for broilers, found a bewildering variety of possible

conclusions concerning brooding and finishing temperatures, air speed,

humidity, nutrition, sex and temperature fluctuations. He reported

that sex by temperature interactions could be expected due to

differences in growth rate and fat deposition. Methods of production

such as stocking rate and housing design will also interact with

temperature and make prediction of an optimum temperature for growth

difficult. Charles et al. (1981) found that both weight gain and feed

intake were depressed by increasing temperature in the range of 15 to

270C (60 to 800F). Also, Harris et al. (1974) noted that commercial

broilers are not grown at constant temperatures but are subjected to

variations in house temperature that are directly related to diurnal









13

changes in outside air temperature. The general conclusion being that

depressions in broiler performance cannot entirely be prevented by

dietary formulation, although, as increasing temperatures restrict

intake, more concentrated diets are used in an attempt to meet

nutrient requirements.



Choice of Feed Ingredients

Grain sorghum (milo). Milo is used extensively as a direct

replacement for maize in poultry feeding in many developing countries.

It is an important food crop in arid and semi-arid tropics and is

believed to have originated in Africa about 5000 years ago (Olentine,

1986). Kramer and Matz (1969) reported that milo is the third most

utilized cereal grain in the world.

An early study on the replacement value of milo for corn (Harms

et al., 1958) found that the substitution of milo for corn in the diet

of broilers did not have a significant affect on feed efficiency at

the rates utilized in the experiment. However, a decrease in body

weight gain was obtained as yellow corn was replaced by milo.

New varieties of milo that are being developed in Africa were

examined by Okoh et al. (1982) for amino acid composition and tannin

content. Major differences among varieties were observed for lysine

and tannin. The primary dietary role of milo, like that of corn, is

to supply energy. The proximate analysis of the milo samples

investigated by these researchers indicate that milo has a similar

nutrient profile to that of corn. However, milo is slightly higher in

energy and protein content than corn. Although, the protein content








14

is slightly higher the efficiency of utilization of this protein will

depend on the amino acid balance and tannin content of each variety.

They concluded that the suitability of most of the varieties for

poultry feeds in Africa was likely to be affected by their high tannin

content. Antongiovanni et al. (1980) reported that two high protein

varieties of milo, although rich in protein, were limited by a severe

decrease in the concentration of the limiting amino acid lysine.

Broiler feeding trials conducted by Mohamedain et al. (1986) noted

that food intake of milo diets was reduced as compared to maize due in

part to the presence of tannins which, being orally astringent, reduce

palatability and thus intake. Tannins in milo are also known to

reduce protein digestibility and to inhibit the activity of various

enzyme systems. However, a high tannin level is desirable for wild

bird resistance in the fields.

When one considers the growing importance of milo in terms of

available quantities in areas of the world where other cereal grains

are relatively scarce, weighed against the inherent variability in

nutrient profile, tannin content and lysine limitations, a need is

indicated for research that explores the feeding potential of this

cereal in light of its availability and economic advantages over corn.

Peanut meal (PM). In general, attempts to utilize peanut meal as

a source of vegetable protein for growing chickens have met with

favorable results when the rations were fortified with methionine and

lysine (Douglas and Harms, 1959). Driggers and Tarver (1958) reported

that up to half the soybean meal can be replaced by peanut meal

depending upon the level of fish meal and supplemental lysine used.









15

Without the fishmeal the diets required methionine supplementation for

a growth response comparable to that of a corn-soybean meal diet.

Peanut meal is the major protein supplement in poultry diets in

India and many other subtropical countries. Singh and Prasad (1979)

found that the replacement of peanut meal in the diets of growing

chickens by sunflower meal improved growth rate and feed efficiency.

Diets based on sunflower meal needed only four percent fish meal

supplementation as compared to eight percent needed with peanut meal

for equal weights in broilers to 70 days of age. The amino acid

profile of the peanut meal was complemented by the sunflower meals

higher sulfur amino acid and lysine content in much the same way as

the previously reported supplementation of fish meal or synthetic

lysine and methionine.

It is interesting to note that Douglas and Harms (1959) reported

a significant sex by peanut meal treatment interaction and they

concluded that the slower growing females were more tolerant than

males to low protein levels when using peanut meal in the diets.

Feed Form. To attain the greatest level of efficiency in the

allocation of feed resources the possible economic advantages of

pelleting feed were examined in the simulation study. A summary of 14

research reports (Jones, 1979) comparing the value of pellets relative

to mash, showed that in all cases either growth rate and/or feed

conversion was improved by pelleting. The reason or reasons for the

improvement in production is not fully understood. Dymsza et al.

(1955), using rations containing 5, 10 and 15 percent fiber, showed

that pelleting and crumbling concentrated the nutrients relative to








16

mash and counteracted some of the adverse effects of low energy diets

when fed to turkey poults. Broiler studies conducted by Newcombe and

Summers (1985) reported that when cellulose was added to the basal

diet, birds fed on crumbles performed better than those fed mash.

These researchers attributed this difference to poor palatability

associated with the flour-like texture of the mash diets. Additional

research on the possible advantages of pelleting when feeding high

fiber diets with low palatability in the tropics was conducted by

Abdelsomie et al. (1983) where pelleting was shown to significantly

improve both growth rate and feed efficiency but not carcass yield.

Huile-Shen et al. (1985) concluded that steam pelleting enhanced the

utilization of lower energy diets and could be an important economic

consideration in parts of the world where dietary energy is in short

supply.
















CHAPTER III

MATERIALS AND METHODS


Experimental Methods and Procedures



Live-weight gain, feed intake and house temperature data from a

ten-week broiler response study were collected during the summer

months (Aug. 1 Oct. 10, 1986) at the University of Florida Poultry

Science research station. Individual birds of a commercial strain of

feather sexed broiler chicks (Cobb Cobb) were assigned at random to

64 pens subject to including 16 chicks in each pen and maintaining a

1:1 sex ratio. Protein treatments were specified in terms of four

isocaloric diets containing 16, 19, 22, and 25% crude protein. The

four protein treatments were fed in both mash and pellet form in a

four by two factorial arrangement for a total of eight feed treatments

and replicated eight times. Chicks were placed on experimental diets

at one day of age and fed ad libitum those same respective diets to

ten weeks of age.

Observations were recorded on the basis of a fixed time approach

where records were maintained on the cumulative live-weight and feed

consumed per pen on a weekly basis through ten weeks after the

beginning of the experiment. These records included 640 observations

on each of the production criteria. Data on cumulative live-weight
















CHAPTER III

MATERIALS AND METHODS


Experimental Methods and Procedures



Live-weight gain, feed intake and house temperature data from a

ten-week broiler response study were collected during the summer

months (Aug. 1 Oct. 10, 1986) at the University of Florida Poultry

Science research station. Individual birds of a commercial strain of

feather sexed broiler chicks (Cobb Cobb) were assigned at random to

64 pens subject to including 16 chicks in each pen and maintaining a

1:1 sex ratio. Protein treatments were specified in terms of four

isocaloric diets containing 16, 19, 22, and 25% crude protein. The

four protein treatments were fed in both mash and pellet form in a

four by two factorial arrangement for a total of eight feed treatments

and replicated eight times. Chicks were placed on experimental diets

at one day of age and fed ad libitum those same respective diets to

ten weeks of age.

Observations were recorded on the basis of a fixed time approach

where records were maintained on the cumulative live-weight and feed

consumed per pen on a weekly basis through ten weeks after the

beginning of the experiment. These records included 640 observations

on each of the production criteria. Data on cumulative live-weight










gain (G) in kilograms per pen, as well as cumulative milo and peanut

meal (PM) consumption in kilograms per pen were converted to kilograms

per square meter of pen space so that different size farms could use

the data in the evaluation of the economic implication. The

experimental unit was a pen (2.3225 square meters) with each weighing

of feed or birds an observation.



Feed Form, Formulation and Treatments



The Brill feed formulation program (Brill, Inc., 1985) was used

to balance four protein treatment diets given the alternate feed

ingredients milo, peanut meal and wheat middlings. Composition of the

diets is detailed in Table 3-1. Wheat middlings, salt, limestone,

dicalcium phosphate, vitamin-mineral premix, methionine, lysine and

coccidiostat were all held at constant levels for all treatments.

Composition of the vitamin-mineral premix is shown in Table 3-2.

Vegetable fat (corn oil) was varied between 0.5 and 5.5%, to formulate

the four isocaloric diets (2860 kcal ME/kg) for the 16, 19, 22 and 25%

crude protein treatments. Therefore, the principal changes in dietary

composition were due to changes in milo and peanut meal concentrations

enabling the evaluation of the efficiency of these ingredients in

producing live-weight gain in broilers.

Ingredient composition files in the Brill formulation system were

updated to values taken from standard industrial feed composition

tables (Table 3-3).






















Table 3-1. Composition of the diets


IngredientI 16% 19% 22% 25%

Milo 69.98 60.71 51.44 42.17
Peanut meal (solvent) 14.00 21.73 29.45 37.18

Wheat middlings 10.00 10.00 10.00 10.00
Vegetable fatz 0.77 2.32 3.86 5.41
Dicalcium phos.3 2.00 2.00 2.00 2.00
Limestone 1.40 1.40 1.40 1.40
Salt 0.30 0.30 0.30 0.30
Vit.-min.4 0.50 0.50 0.50 0.50
Meth.,dry D-L 0.40 0.40 0.40 0.40
Lysine-L 0.55 0.55 0.55 0.55
Coban 0.10 0.10 0.10 0.10

Calculated values
Protein (%) 16 19 22 25
ME (kcal/kg) 2860 2860 2860 2860


1 Ingredient levels expressed as
2 Corn oil (8,800 Kcal/kg).
3 Contains 22% Ca and 18.5% P.
4 Standard University of Florida


g/lOOg of diet.


chick microingredient premix.





















Composition of the vitamin-mineral premix


Vit.-min. Broiler Improved1 Units2
---------------- -- - -
Vitamin A 6,600 6,600 IU
Vitamin D3 2,200 2,200 ICU
Vitamin E 0.0 11.0 IU
Vitamin K3 2.2 2.2 mg
Riboflavin 4.4 4.4 mg
Niacin 39.6 59.6 mg
Pantothenic acid 13.2 13.2 mg
Choline chloride 499.0 998.8 mg
Vitamin B12 22.0 22.0 mcg
Biotin 0.0 0.11 mg
Folic acid 0.0 1.0 mg
Pyridoxine(82.2%) 0.0 3.0 mg
Ethoxyquin 125.0 125.0 mg
Manganese 60.0 60.0 mg
Iron 50.0 50.0 mg
Copper 6.0 6.0 mg
Cobalt 0.1980 0.1980 mg
Iodine 1.1 1.1 mg
Zinc 35.0 35.0 mg
I==-----=P~e ==E -e= =----=-- -- -- -

1 Added vitamin E, biotin, folic acid and pyridoxine.
2 Vitamin and mineral concentrations are expressed as their
activity per kilogram of finished feed.
3 Menadione Dimethylpyrimidinol Bisulfate (MPB).


Table 3-2.













Table 3-3. Nutrient values used to calculate the diets


Nutrient1

Protein
Fat
Calcium
Total phos.
Arginine
Lysine
Methionine
Meth. + cyst.
Tryptophan
Threonine
Crude fiber
ME(kcal/kg)


Milo

9.30
2.80
0.02
0.30
0.29
0.19
0.13
0.27
0.09
0.30
- 2.50
3,300


Peanut meal

45.00
1.20
0.15
0.63
4.80
1.60
0.45
1.15
0.46
1.44
12.00
2,750


Wheat middlings

16.00
4.00
0.10
0.85
1.10
0.75
0.26
0.63
0.23
0.59
7.50
1,606


1 Nutrient values expressed as g/lOOg of ingredient.












Amino Acid Profile

The amino acid profile of each protein treatment was examined for

deficiencies and presented in Table 3-4. Threonine was found to be

deficient in all treatments when expressed as a percent of total

weight and as a percent of protein for a particular treatment. This

amino acid is the first limiting amino acid after lysine and

methionine supplementation of the peanut meal and milo diets used in

this study.











Table 3-4. Calculated amino acid profile of the diets


Amino acid

Arginine
Lysine
Methionine
Meth.+cyst.
Tryptophan
Threonine


NRC
Requ.

1.44
1.05
0.50
0.85
0.23
0.80


% of
Prot.

5.0
4.0
2.0
3.6
1.0
3.5


al

1.08
0.88
0.57
0.82
0.18
0.47


16%

b2

6.75
5.50
3.56
5.13
1.13
2.94


a

1.48
1.00
0.60
0.88
0.22
0.55


19%

b

7.79
5.26
3.16
4.63
1.16
2.89


22%
I- -


a

1.88
1.12
0.62
0.95
0.26
0.64


b a

8.55 2.28
5.09 1.24
2.82 0.64
4.32 1.02
1.18 0.30
2.91 0.72


--ln-xei~:~= ====== = ==-1Sn,


1 Calculated level of amino acids
2 Calculated level of amino acids
in a particular treatment.


expressed as
expressed as


g/lOOg of diet.
a g/lOOg of protein


Simulation of a Farm Environment


A broiler house containing 64 pens (2.3225 sqm/pen) simulated a

broiler integrator's geographical production area and was considered a

research domain. Within this research domain a completely randomized

block of four protein treatments (16, 19, 22 and 25% crude protein)

each in mash and pellet form were used (eight feed treatments) as

shown in Figure 3-1. These eight pens simulated a growers farm on

which research was conducted.

To simulate the effects of different environments the lower wall

flaps and the roof vents associated with the eight pens (one

environment) in each of the four corners of the building were closed

so that these four environments would experience higher ambient


25%

b

9.12
4.96
2.56
4.08
1.20
2.88


-,u--= E---------u--==----~:-------I------- --------------5==


--


--


'"


--









23
temperatures (poor environments) as compared to the four environments

with opened wall flaps and roof vents (good environments) (Fig. 3-1).

The resulting differential in live-weight gain between the poor

(hot) and good (cool) environments enabled a partitioning of the

research domain into environment-specific recommendation domains.



Environmental Factors



Management

Broiler management was held constant over all pens so that the

effects of ambient temperature on broiler gain could be estimated.

Standard broiler management practices concerning the feeder, water,

litter, brooder lamps and house lighting practices were performed as

recommended in a standard broiler service manual (Cobb, Inc., 1984).



Vaccination Program

Day-old chicks were vaccinated subcutaneously against Marek's

disease. A commercial Newcastle-Bronchitis live virus vaccine

(B-type, B-strain) was administered at 14 days of age as a coarse

spray. At five weeks of age Newcastle-Bronchitis live virus vaccine

was administered in the drinking water.



Medication Program

Starting at seven days of age, when broilers on the low protein

treatments experienced signs of nutritional deficiencies (perosis) a

commercial vitamin stress pack with electrolytes was added to the










BROILER HOUSE


GOOD ENVIRONMENT
PROTEIN, FEED FORM
POOR ENVIRONMENT
PROTEIN, FEED FORM


Figure 3-1.


Broiler house pen arrangement showing good and poor
environments, completely randomized blocks of eight feed
treatments representing a growers farm and examples of
the location of feed treatments. M and P indicate mash
and pellet form, respectively.


F-7 -171717


Wv


S + I










water. This medication was administered to all pens in the broiler

house and was continued until 21 days of age at which time an improved

vitamin-mineral premix was used in the feed (Table 3-2).

At 18 days of age a generalized respiratory infection became

apparent throughout all pens. It was believed to be a typical

reaction to the vaccination program administered four days earlier,

but, as a precaution a commercial brand of the antibiotic erythromycin

was administered in the drinking water for five days.



Temperature

Two maximum-minimum thermometers were placed at bird level in

each completely randomized block of pens. Temperature data was

recorded daily and the range of temperatures (maximum minus minimum)

was computed and then averaged for the week. The average weekly

temperature range (AWTR) of ambient temperature for each replication

(grower's environment) is reported in Table 3-5.

The simulated "good" production environments (cooler

temperatures) ranged from a 12.21 to 18.46 AWTR for the ten-week

period. The "poor" environments (hot temperatures) ranged from a 9.36

to 14.61 AWTR. The overall AWTR for all replications and for the

entire study period was 14.12. A higher AWTR value indicated that

those pens (good environment) cooled down to a lower temperature than

those in the poor environments since the average maximum temperatures

were similar. The highest AWTR values recorded were in the good

environments on the south side of the broiler house.










Table 3-5. Average weekly temperature range (AWTR) in the house by
replication


Rep. Wk-1 Wk-2 Wk-3 Wk-4 Wk-5 Wk-6 Wk-7 Wk-8 Wk-9 Wk-10 avg.

R1 13.68 13.14 9.36 13.61 9.50 13.29 13.50 15.54 14.93 15.71 13.23
R2 15.89 16.64 12.68 15.93 12.21 14.86 15.36 17.93 17.57 18.46 15.75
R3 16.15 16.32 11.57 15.57 11.71 14.54 15.01 17.43 17.04 18.18 15.35
R4 14.29 13.93 10.21 13.32 9.86 11.93 12.93 14.61 14.50 16.11 13.17
R5 13.75 14.14 10.50 13.32 10.32 12.29 12.18 14.57 13.75 16.11 13.09
R6 15.95 15.93 12.07 15.25 11.57 13.79 13.71 16.25 14.61 16.50 14.56
R8 13.43 12.89 9.68 11.89 11.04 12.96 12.64 15.07 13.32 14.75 12.77

avg.14.79 14.77 11.02 13.91 10.86 13.40 14.15 16.27 15.45 16.61 14.12














Therefore, temperature values used in the modeling exercise were

14.0 AWTR representing an average environment, a 10.0 AWTR for a poor

(hot) environment and an 18.0 AWTR for a good (cool) environment.



Mortality

Mortality was recorded daily and feed intake data were adjusted

when deaths occurred. Birds that would be considered culls in a

typical broiler operation were culled from the experiment on those

days when body-weights and feed intake were taken. This was often due

to cases of severe perosis. The total number of culls and mortality

by feed treatment and sex are shown in Table 3-6.





















Table 3-6. Mortality and culls by environment (e), sex and
feed form


Feed treatment

16% 19% 22% 25%

(e) Sex M1 p2 M P M P M P

Mort. poor males 0 1 1 1 0 0 2 2
good males 0 2 0 2 3 1 0 3

poor females 2 1 1 1 1 0 0 1
good females 0 1 1 1 1 0 0 1

Culls poor males 5 7 5 5 2 5 0 1
good males 1 13 7 5 2 2 0 2

poor females 2 6 2 2 1 0 0 0
good females 4 1 2 3 2 0 0 1

Total males 6 23 13 13 7 8 2 8
Total females 8 9 6 7 3 1 0 2

Total treatment 46 39 19 12


1 Mash.
2 Pellet.















CHAPTER IV

PARTITIONING BROILER GROWERS
INTO HOMOGENEOUS PRODUCTION ENVIRONMENTS


Introduction



Broiler producers throughout the world realize that farms and

farm managers are not created equal and yet averaged data from

research stations are commonly used as the basis for making feeding

recommendations to individual broiler growers. If the decision makers

of a vertically integrated broiler production system have the

objective of improving the efficiency or quality of feeding

recommendations to individual contract growers, then a method for

partitioning the growers into groups with homogeneous production

environments is required. The purpose of this chapter is to

demonstrate the use of modified stability analysis combined with a

frequency distribution of confidence intervals (Hildebrand, 1984;

Hildebrand and Poey, 1985) as a tool for meeting that objective.

Variability among growers is due to differences in the overall

environment each grower provides for the broiler. Environment

encompasses those agro-biological and socio-economic factors that

impact on broiler production. For example, under the socio-economic

factors to be considered, the differences among growers may range from

management ability, education, experience, and age to the type of

28















CHAPTER IV

PARTITIONING BROILER GROWERS
INTO HOMOGENEOUS PRODUCTION ENVIRONMENTS


Introduction



Broiler producers throughout the world realize that farms and

farm managers are not created equal and yet averaged data from

research stations are commonly used as the basis for making feeding

recommendations to individual broiler growers. If the decision makers

of a vertically integrated broiler production system have the

objective of improving the efficiency or quality of feeding

recommendations to individual contract growers, then a method for

partitioning the growers into groups with homogeneous production

environments is required. The purpose of this chapter is to

demonstrate the use of modified stability analysis combined with a

frequency distribution of confidence intervals (Hildebrand, 1984;

Hildebrand and Poey, 1985) as a tool for meeting that objective.

Variability among growers is due to differences in the overall

environment each grower provides for the broiler. Environment

encompasses those agro-biological and socio-economic factors that

impact on broiler production. For example, under the socio-economic

factors to be considered, the differences among growers may range from

management ability, education, experience, and age to the type of

28










equipment used on the farm. Likewise, under the agro-biological

factors, the differences may range from feed form (mash or pellet),

bird density, and breed to house temperature.

It is important to note that the following methodology is as

relevant to large broiler integrators in the U.S. as it is to smaller

integrators in developing countries or even regional consultants for

independent growers in those countries where the industry is not

vertically integrated.



Materials and Methods



Modified Stability Analysis and Broiler Production

Modified stability analysis is used to examine the need to

partition the research domain into environment-specific recommendation

*domains (Hildebrand, 1986). It is an analysis technique that, as used

here, enables accounting for variability among broiler growers. By

averaging the results of all experimental feed treatments on a

grower's farm into an environmental index (e), a means for measuring

and quantifying the effect of all the factors that influence the gain

response of broilers to a feed treatment can be achieved. This

quantified variability found on different growers' farms then becomes

a practical means for partitioning the research domain.

To conceptualize the application of modified stability analysis

to broiler production, imagine an integrated broiler producer with

growers dispersed over an area with a radius of 50 miles. Consider

this area a research domain, where the integrator desires to










investigate the efficiency of the feeding recommendations currently

practiced by all growers within this study area. All growers are

feeding four protein treatment feeds in two forms (mash or pellet).

No other changes are made from the growers' usual practices. The only

constants on the farm are those factors that are controllable by the

integrator, feed treatments, strain of bird and medication programs.

The broilers on each grower's farm will be exposed naturally to

different environmental conditions, such as water quality, house

temperature, equipment type, housing design and management in general.

If the average production of broiler meat (gain) from the four protein

treatments is high for whatever reason then this grower's farm is

classified as a "good" environment for broilers. A grower's farm for

which production of broiler meat is low for whatever reason is

classified as a "poor" environment for broilers.

Broiler live-weight (gain) for each of the four protein

treatments can be related to environment by simple linear regression.



(Gi) = a + b (ej) (1)

where (G) = gain from feed treatment i, and

(e) = environmental index for farm j, equal to

the average gain of all feed treatments

at each grower's farm.



The environmental index (e) becomes a continuous variable whose

range is the range of cumulative average live-weight gains from the

experiment.










By fitting the linear equation independently for each treatment

level, then plotting the broiler gain response to environment for each

treatment on the same graph, it is possible to visually compare

treatments across a range of environments (e).



Frequency Distribution of Confidence Intervals and Broiler Production

A graphic distribution of confidence intervals can be used to

evaluate the variability in the results of each protein treatment

within each partition of the research domain. These partitions are

tentatively considered recommendation domains until the analysis is

complete.

The confidence intervals are calculated for a 50, 60, 80, 90, 95,

and 99% level of confidence from the equation:



(G) + ta S / n (2)



Where (G) = mean treatment gain for the partitioned

group of farmers, and

t = table value of t,

a = the level of confidence,

S/ n = standard error of the mean.



Simulation of an On-Farm Broiler Response Study

Live-weight gain, feed intake and house temperature data from a

ten week broiler response study were collected during the summer

months (Aug. 1 Oct. 10, 1986) at the University of Florida Poultry








32

Science research station. Individual birds of a commercial strain of

feather sexed broiler chicks (Cobb Cobb) were assigned at random to

64 pens subject to including 16 chicks in each pen and maintaining a

1:1 sex ratio. Protein treatments were specified in terms of four

isocaloric diets containing 16, 19, 22, and 25% crude protein. The

four protein treatments were fed in both mash and pellet form in a

four by two factorial arrangement for a total of eight feed treatments

and replicated eight times. A broiler grower's environment was then

represented by a randomized block of eight pens (four protein levels

by two forms of feed).

To simulate the effect of different environments the lower wall

flaps and the roof vents associated with the eight pens (one

environment) in each of the four corners of the building were closed

so that these four environments would experience a higher house

temperature as compared to the four environments with opened wall

flaps and roof vents (Fig. 3-1, pg. 24). The entire building

simulated a research domain, with four growers anticipated to be in a

"good" environment and four growers in a "poor" environment. This

simulation could be representative of farms such as those having

different housing designs, different elevations or proximity to large

bodies of water.

Of all the variables considered in the environmental matrix,

house temperature is of major importance due to its effect on feed

intake and thus live-weight gain.










Results and Discussion



The simulation study was designed and conducted to test a number

of hypotheses one of which was the hypothesis that broiler gain

response to different feed treatments and feed forms would vary

depending upon environmental conditions. The treatment averages for

all growers (complete data set) from the broiler response study are

presented in Tables 4-1 and 4-2 for weeks 4, 6, and 8.

Examination of the graphed broiler response surface to averaged

feed treatments from the combined data set (Fig. 4-1) gives no

indication of a need to partition the complete data set (research

domain) into more homogeneous production environments.



Partitioning the Growers by Production Environment

Application of modified stability analysis begins by averaging

all treatments tested on each grower's farm. This value is expressed

as the environmental index (e), as shown in Tables 4-1 and 4-2. The

broiler gain data for each treatment and for the 8 growers were fit to

equation (1) by simple linear regression and presented in Table 4-3.

By plotting individual broiler gain data points from the broiler

response study with its corresponding environmental index value (e),

along with the simple linear regression functions for each protein

level, a means for observing treatment by environmental interactions

can be made (Fig. 4-2 to 4-7).














Table 4-1.


Broiler gain response to protein treatments for pellet
form at weeks 4, 6, and 8 in kg/pen (complete data set)


Protein (%)
Broiler ------------------------ Environmental
grower index (e)
No. 25% 22% 19% 16% avg.

Week 4 1 15.400 12.227 9.540 5.919 10.772
2 14.841 13.178 10.753 6.546 11.330
3 14.917 13.263 10.014 6.039 11.058
4 13.614 12.988 10.118 9.169 11.472
5 13.735 13.596 9.252 6.550 10.783
6 15.843 13.595 11.693 7.332 12.116
7 14.629 15.144 11.028 8.521 12.331
8 15.301 14.230 9.732 5.386 11.162

avg. 14.785 13.528 10.266 6.933 11.378

Week 6 1 23.846 21.811 17.291 11.395 18.586
2 24.831 24.420 19.446 11.679 20.094
3 26.254 23.438 19.206 10.247 19.786
4 24.521 22.592 17.047 15.554 19.929
5 22.745 21.108 16.355 12.888 18.274
6 27.340 24.641 20.810 14.961 21.938
7 23.596 26.094 18.529 15.905 21.031
8 26.175 25.405 17.887 10.320 19.947

avg. 24.194 23.689 18.321 12.869 19.948

Week 8 1 33.296 27.886 24.996 15.495 25.418
2 33.429 34.505 27.136 16.789 27.965
3 36.904 33.828 27.738 15.337 28.452
4 34.196 31.287 23.832 20.929 27.561
5 28.120 31.978 20.130 19.038 24.817
6 39.841 34.325 29.216 22.086 31.367
7 32.921 38.769 25.654 23.180 30.131
8 35.745 36.490 24.912 15.845 28.248

avg. 34.307 33.634 25.452 18.587 27.995
- - - -- - - - - -- -















Table 4-2.


Broiler gain response to protein treatments for mash
form at weeks 4, 6, and 8 in kg/pen (complete data set)


-- -- -- --- -- P II -- -- -- --- -
Protein (%)
Broiler ------- ---------- Environmental
grower index (e)
No. 25% 22% 19% 16% avg.

Week 4 1 13.273 12.416 7.635 7.722 10.262
2 14.461 13.159 10.105 7.303 11.257
3 15.136 13.219 9.623 8.255 11.558
4 13.875 13.644 9.809 7.106 11.109
5 13.231 12.680 11.416 8.275 11.401
6 13.251 13.091 11.235 8.413 11.498
7 15.271 13.156 11.125 8.776 12.082
8 13.621 12.528 10.870 6.503 10.881

avg. 14.015 12.987 10.227 7.794 11.256
S.3 ---------- -
Week 6 1 24.012 23.545 13.902 14.902 18.895
2 25.212 24.551 17.829 13.904 20.374
3 26.231 23.568 18.643 15.519 20.990
4 23.838 23.629 18.478 13.095 19.760
5 23.504 22.095 20.197 16.086 20.471
6 24.755 23.854 20.806 16.121 21.384
7 26.704 23.991 19.529 16.809 21.758
8 25.753 22.585 20.289 12.827 20.364

avg. 25.001 23.477 18.709 14.810 20.499
Il~eIIXPPI-=- =-- 5==--------=--- -
Week 8 1 33.137 34.370 20.687 19.811 27.001
2 35.437 35.526 24.604 20.219 28.947
3 35.021 33.432 27.193 22.494 29.535
4 34.166 34.219 25.978 19.665 28.507
5 33.404 31.545 25.536 23.169 28.414
6 34.677 33.304 30.611 22.931 30.381
7 37.504 34.087 29.069 23.800 31.115
8 35.793 32.235 30.185 17.387 28.900

avg. 34.892 33.590 26.733 21.185 29.100
_--------- -----= -





























z



w
~25-
0





-J-



15





10.


Figure 4-1.


............. mash

pellet


.. ........... Wk-8


..** .... *" Wk-6
--*.























set).e
Wk-4
.. .* *.










19 22 25
CRUDE PROTEIN, (%)


Broiler gain response to protein treatments for mash and
pelleted feed form at weeks 4, 6, and 8 (complete data
set).















Table 4-3. Linear equations for modified stability analysis
calculated from broiler gain completet data set)

-= ------------- -------- -
Protein
Treatment a b R

Pellet Form

Week 4 16% -10.130 1.500 .66
19% 4.267 1.277 .89
22% 2.620 .959 .64
25% 11.777 .264 .20
S----------11- --e5=- -s~:--=5--= l------ =
Week 6 16% -6.653 .979 .45
19% -1.789 1.008 .82
22% .282 1.173 .79
25% 8.160 .840 .64

Week 8 16% -2.535 .755 .52
19% -2.236 .989 .60
22% 3.916 1.062 .70
25% .856 1.195 .76


Mash Form

Week 4 16% -2.463 .911 .63
19% -8.708 1.682 .72
22% 8.304 .416 .54
25% 2.867 .990 .40

Week 6 16% -9.177 1.170 .70
19% -18.955 1.837 .76
22% 20.525 .144 .17
25% 7.607 .849 .66
---I------------ ---- ---- -=~---
Week 8 16% -8.500 1.020 .58
19% -34.347 2.099 .81
22% 33.761 -.006 .01
25% 9.086 .887 .80


















,_ -*-- ~"- ~ ~ ,- -* -- -- -- ~
---------- -------

^---~ *



... .. ..... ... .



I.... ----' '
.,.o..- --- -
*..... I


ENVIRONMENTAL INDEX (e), KG


Figure 4-2.










15
Z
w
a.



^ 10

z
03


Figure 4-3.


Broiler gain response to protein treatments for mash
form at 4 weeks of age as influenced by environment.








S25% CP
.--'- 22% CP









S* 16% CP
S-4-----------*










ENVIRONMENTAL INDEX (e), KG







Broiler gain response to protein treatments for pellet
form at 4 weeks of age as influenced by environment.


25% CP


22% CP


19% CP






16% CP










































Figure 4-4.


25% CP


22% CP


19% CP





16% CP


19 20 21 22

ENVIRONMENTAL INDEX (e), KG




Broiler gain response to protein treatments for mash
form at 6 weeks of age as influenced by environment.


25% CP
22% CP





19% CP





16% CP


ENVIRONMENTAL INDEX (e). KG


Figure 4-5.


Broiler gain response to protein treatments for pellet
form at 6 weeks of age as influenced by environment.


*

Ic


U
^---*-c"
~ -.-.-




"*.,,' -. '
....A ..
..... *.,---'" "



*
....,-"
.... ...f~o'''' '" "











40


25% CP


22% CP

19% CP







16% CP


ENVIRONMENTAL INDEX (e). KG


Figure 4-6.








r


Broiler gain response to protein treatments for mash
form at 8 weeks of age as influenced by environment.


25% CP

22% CP





19% CP

16% CP


ENVIRONMENTAL INDEX (e), KG


Figure 4-7.


Broiler gain response to protein treatments for pellet
form at 8 weeks of age as influenced by environment.


- --- --^=- --- --C--- ----------- -------- -- -- -- -




----------
.' -' '




S........... -
...--' *....-
..-- --- ""
.....-- -*-- ""

I


_.. .... .... .... ." '.


.................""-
S....-.----
..............----.--.
I


S**


40








41

The advantages of partitioning the growers in the research domain

into homogeneous production environments (recommendation domains) is

clearly to develop more appropriate feeding recommendations for

environment-specific growers. Where, the greater the difference

between a good and poor broiler grower's environment the greater the

need to partition the environments. The development of environment-

specific feeding programs will contribute to the optimization of the

use of feed resources in the production of broiler live-weight gain.

For example, tentative partitioning of the research domain can be

made for week 4, pellet form (Fig. 4-3), by declaring those growers

above the environmental index e = 11.5 as "good" growers and those

lower than e = 11.5 as "poor" growers. If the spread of (e) is wide

the broiler integrator may partition the research domain into 3

recommendation domains. For growers above e = 11.5, below e = 11 and

between e = 11 and e = 11.5

If the integrator averaged data from all of his growers by not

partitioning the research domain, those growers within the e = 11 to

11.5 range would benefit from the recommended feeding practices.

Those growers above 11.5 would be feeding a protein level higher than

they would require and those growers below e = 11 would not be feeding

adequate protein levels to meet the maximum growth requirement.

In this case, the complete data set was partitioned into "good"

and "poor" broiler production environments as differentiated by the

open or closed wall flaps and roof vents, as shown in Tables 4-4 and

4-5. This partitioning was based on the assigned environmental groups

as designed in the simulation study.











Table 4-4.


Broiler gain response to protein treatments for pellet
form at weeks 4, 6, and 8 in kg/pen (partitioned data set)


Protein (%)
Broiler ----- ----Environmental
grower index (e)
No. 25% 22% 19% 16% avg.

Week 4

poor 1 15.400 12.227 9.540 5.919 10.772
4 13.614 12.988 10.118 9.169 11.472
5 13.735 13.596 9.252 6.550 10.783
8 15.301 14.230 9.732 5.386 11.162
poor avg. 14.513 13.260 9.661 6.756 11.047

good 2 14.841 13.178 10.753 6.546 11.330
3 14.917 13.263 10.014 6.039 11.058
6 15.843 13.595 11.693 7.332 12.116
7 14.629 15.144 11.028 8.521 12.331
good avg. 15.058 13.795 10.872 7.110 11.709
----- ----- --- -----=-cr=-==
Week 6

poor 1 23.846 21.811 17.291 11.395 18.586
4 24.521 22.592 17.047 15.554 19.929
5 22.745 21.108 16.355 12.888 18.274
8 26.175 25.405 17.887 10.320 19.947
poor avg. 24.322 22.729 17.145 12.539 19.184

good 2 24.831 24.420 19.446 11.679 20.094
3 26.254 23.438 19.206 10.247 19.786
6 27.340 24.641 20.810 14.961 21.938
7 23.596 26.094 18.529 15.905 21.031
good avg. 25.505 24.648 19.498 13.198 20.712

Week 8

poor 1 33.296 27.886 24.996 15.495 25.418
4 34.196 31.287 23.832 20.929 27.561
5 28.120 31.978 20.130 19.038 24.817
8 35.745 36.490 24.912 15.845 28.248
poor avg. 32.839 31.910 23.468 17.827 26.511

good 2 33.429 34.505 27.136 16.789 27.965
3 36.904 33.828 27.738 15.337 28.452
6 39.841 34.325 29.216 22.086 31.367
7 32.921 38.769 25.654 23.180 30.131
good avg. 35.774 35.357 27.436 19.348 29.479
's-;"--^ -",--- -- -T - --^: r" W ~V K~ 'T '~'~ ~ -'~ ~ ~ "^ '^-* ~ *~w'''' "~ *-- i_-_ _*- -- y 1;










Table 4-5.


Broiler gain response to protein treatments for mash form
at weeks 4, 6, and 8 in kg/pen (partitioned data set)


==------------P-r=e tE,-=- =
Protein (%)
Broiler -------------------- Environmental
grower index (e)
No. 25% 22% 19% 16% avg.

Week 4

poor 1 13.273 12.416 7.635 7.722 10.262
4 13.875 13.644 9.809 7.106 11.109
5 13.231 12.680 11.416 8.275 11.401
8 13.621 12.528 10.870 6.503 10.881
poor avg. 13.500 12.817 9.933 7.402 10.913

good 2 14.461 13.159 10.105 7.303 11.257
3 15.136 13.219 9.623 8.255 11.558
6 13.251 13.091 11.235 8.413 11.498
7 15.271 13.156 11.125 8.776 12.082
good avg. 14.530 13.156 10.522 8.187 11.599

Week 6

poor 1 24.012 23.545 13.902 14.120 18.895
4 23.838 23.629 18.478 13.095 19.760
5 23.504 22.095 20.197 16.086 20.471
8 25.753 22.585 20.289 12.827 20.364
poor avg. 24.277 22.964 18.217 14.032 19.873

good 2 25.212 24.551 17.829 13.904 20.374
3 26.231 23.568 18.643 15.519 20.990
6 24.755 23.854 20.806 16.121 21.384
7 26.704 23.991 19.529 16.809 21.758
good avg. 25.726 23.991 19.202 15.588 21.127

Week 8

poor 1 33.137 34.370 20.687 19.811 27.001
4 34.166 34.219 25.978 19.665 28.507
5 33.404 31.545 25.536 23.169 28.414
8 35.793 32.235 30.185 17.387 28.900
poor avg. 34.125 33.092 25.597 20.008 28.206

good 2 35.437 35.526 24.604 20.219 28.947
3 35.021 33.432 27.193 22.494 29.535
6 34.677 33.304 30.611 22.931 30.381
7 37.504 34.087 29.069 23.800 31.115
good avg. 35.660 34.087 27.869 22.361 29.995
-- -------- -- ------ ====,,=====-~='tl-~










Stability of Treatments

An indication of the relative stability of protein treatments

across a range of environments can be seen in Figures 4-2 to 4-7. For

example, in Figure 4-3 the regression line describing the 25% protein

treatment as a function of the environmental index (e) is almost

horizontal (slope = .264). This indicates that at four weeks of age

the 25% protein treatment, pellet form, produced a stable gain

response across all environments. Whereas, the regression line

describing the 16% protein treatment produced a relatively unstable

response (slope = 1.5) across all environments. The level of gain

from the 16% treatment in the good environment (e > 11.5) was nearly

7.11 kg per pen and the gain from the poor environment (e < 11.5) was

only 6.76 kg per pen, which. suggests a relatively strong 16% protein

treatment by environment interaction.

The relative stability of a protein treatment in varying

environments has practical implications in the making of feeding

recommendations. In this study the implications are two fold; first,

high protein (22 and 25%) treatments can help "overcome" the effects

of poor environments (temperature) and secondly, caution should be

advised when feeding low protein levels (16 and 19%) in poor

environments.



Broiler Gain Response to Varying Environments

Differences in the predicted broiler gain response for good and

poor broiler production environments can be visualized by graphing the

average gain response to protein treatments for the good and poor










environments (Fig. 4-8 and 4-9). The impact of environment (e) on

gain becomes evident in this visual representation and further

examination indicates that a) the differences between good and poor

environments were greater for pellet than for mash form, b) with age

and in poor environments the gain response curve becomes more cubic in

form and less quadratic and c) with age the differences between poor

and good environment becomes greater.

The distribution of confidence intervals for gain were calculated

for the complete data set (all growers), each feed form, ages 4, 6,

and 8 weeks, and presented in Tables 4-6 and 4-7. Graphing the

distribution of confidence intervals for the four treatments by mash

and then pellet form, taking care to set the axis to equal intervals,

enables an overlay of the graphs which gives a preliminary indication

of a greater gain response to pellet form at 22 and 25% protein and

mash being greater than pellet at 16% protein (Fig. 4-10a to 4-15a).

Overlaying graphs for similar feed forms at different weeks

illustrates the general trends in broiler response to protein levels

and feed form. But, as the graphed gain response to the protein

treatments from the complete data set (Fig. 4-1) failed to discern

variability due to environment, neither do the graphs of the

corresponding confidence intervals (Fig. 4-10a to 4-15a).

In contrast, the variability due to environment can be discerned

with the following approach. Treatment averages for the good and poor

growers representing the partitioned research domain (Tables 4-4 and

4-5) are graphed (Fig. 4-5 and 4-6) and a visual inspection of the

magnitude and form of the curves from these partitioned environments















GOOD


o. '--. ,. ,., ,'* ,* ,..',. W k- 8




30- POOR ENVIRONMENT




25- .,,--,*, ,.-, W k-6



20- "
>_




15- k-4










16 19 22 25
CRUDE PROTEIN, (%)


Figure 4-8. Broiler gain response to protein treatments for pellet
form at 4, 6, and 8 weeks based on good and poor
environments.
















GOOD ENVIRONMENT


00'0'ooo Wk-8





POOR ENVIRONMENT.




.....0- 0` -'i?00*00 's O
00000 a00 Wk-6











oooo Wk-4


CRUDE PROTEIN, (%)


Figure 4-9.


Broiler gain response to protein treatments for mash
form at 4, 6, and 8 weeks based on good and poor
environments.


Z
w
0.
25-


-J

S20

Z]














Table 4-6.


Distribution of confidence intervals of broiler gain to
protein treatments for pellet form at weeks 4, 6, and 8
in kg/pen (complete data set)


Probability of a larger value (a)
0.50 0.40 0.20 0.10 0.05 0.01

Protein (%) .711 .896 1.415 1.895 2.365 3.499
-=-ei------------- --------------------
Week 4

16% 7.265 7.351 7.594 7.818 8.037 8.567
6.601 6.515 6.272 6.048 5.829 5.299
19% 10.474 10.528 10.679 10.819 10.957 11.288
10.058 10.004 9.853 9.713 9.575 9.244
22% 13.746 13.803 13.962 14.110 14.254 14.602
13.310 13.253 13.094 12.946 12.802 12.454
25% 14.982 15.033 15.177 15.310 15.440 15.754
14.588 14.537 14.393 14.260 14.130 13.816

Week 6

16% 15.190 15.289 15.567 15.824 16.075 16.682
14.430 14.331 14.053 13.796 13.545 12.938
19% 19.261 19.404 19.807 20.180 20.544 21.424
18.157 18.014 17.611 17.238 16.874 15.994
22% 23.675 23.726 23.870 24.003 24.133 24.447
23.281 23.230 23.086 22.953 22.823 22.509
25% 25.296 25.373 25.588 25.787 25.982 26.453
24.706 24.629 24.414 24.215 24.020 23.549
------ =------- -------
Week 8

16% 21.748 21.895 22.306 22.686 23.058 23.956
20.622 20.475 20.064 19.684 19.312 18.414
19% 27.560 27.775 28.379 .28.937 29.483 30.802
25.906 25.691 25.087 24.529 23.983 22.664
22% 33.907 33.990 34.221 34.435 34.645 35.151
33.273 33.190 32.959 32.745 32.535 32.029
25% 35.245 35.337 35.595 35.834 36.067 36.631
34.539 34.447 34.189 33.950 33.717 33.153















Table 4-7.


Distribution of confidence intervals of broiler gain to
protein treatments for mash form at weeks 4, 6, and 8 in
kg/pen (complete data set)


Probability of a larger value (a)
0.50 0.40 0.20 0.10 0.05 0.01

Protein (%) .711 .896 1.415 1.895 2.365 3.499

Week 4

16% 7.988 8.039 8.180 8.311 8.440 8.749
7.600 7.549 7.408 7.277 7.148 6.839
19% 10.541 10.623 10.852 11.065 11.272 11.774
9.913 9.831 9.602 9.389 9.182 8.680
22% 13.090 13.117 13.192 13.262 13.330 13.494
12.884 12.857 12.782 12.712 12.644 12.480
25% 14.227 14.282 14.437 14.580 14.720 15.058
13.803 13.748 13.593 13.450 13.310 12.972

Week 6

16% 13.453 13.605 14.031 14.425 14.811 15.742
12.285 12.133 11.707 11.313 10.927 9.996
19% 18.689 18.785 19.054 19.303 19.546 20.133
17.953 17.857 17.588 17.339 17.096 16.509
22% 24.131 24.245 24.568 24.866 25.158 25.862
23.247 23.133 22.810 22.512 22.220 21.516
25% 25.306 25.408 25.694 25.958 26.217 26.842
24.522 24.420 24.134 23.870 23.611 22.986
- - - -- ---- -
Week 8

16% 19.381 19.588 20.168 20.704 21.229 22.495
17.793 17.586 17.006 16.470 15.945 14.679
19% 26.148 26.329 26.837 27.307 27.767 28.878
24.756 24.575 24.067 23.597 23.137 22.026
22% 34.468 34.685 35.294 35.857 36.408 37.738
32.800 32.583 31.974 31.411 30.860 29.530
25% 35.164 35.388 36.013 36.592 37.159 38.527
33.450 33.226 32.601 32.022 31.455 30.087
---- ---- -=---------










50

60

70

80

90

S100

z 50
w
60
LL
W 70
0

w 80
0
z
U
w 90

z 100
0
0 50


60-

70-

80-

90

100
1

16% CP


5 10 15


GAIN (G), KG/PEN
* c19% CP \I 22% CP


Figure 4-10. Distribution of confidence intervals for broiler gain to
protein treatments for mash feed at week 4. Complete
data set (a), poor environment (b), and good environment
(c).


7 25% CP










50-

60-

70-

80-

90-


100-

50-

60-

70-

80-

90-

100
50-


60-

70-

80-

90-

100-


4


1


|E16% CP


Figure 4-11.


I I 1
5 10 15


GAIN (G), KG/PEN
~19% CP E 22% CP


25% CP


Distribution of confidence intervals for broiler gain to
protein treatments for pellet feed at week 4. Complete
data set (a), poor environment (b), and good environment
(c).


I I I


I


+--m----










50

60

70

80

90,

100
50'

60-

70-

80-


100
50


60

70

80

90

100 ---
5

16% CP


Figure 4-12.


1


0 15 20 2
GAIN (G), KG/PEN
E 19% CP I 22% CP


25% CP


Distribution of confidence intervals for broiler gain to
protein treatments for mash feed at week 6. Complete
data set (a), poor environment (b), and good environment
(c).










50

60

70


z
0
LU
U.

0
Ui

0
z
z

0
0
W,


60-

70-

80-

90 -

100


Eli


5 10 15 20 25 30
GAIN (G), KG/PEN
16% CP L 19% CP \\ 22% CP 1 25%


Figure 4-13.


Distribution of confidence intervals for broiler gain to
protein treatments for pellet feed at week 6. Complete
data set (a), poor environment (b), and good environment
(c).


CP










50

60

70

80


90- -

100 ---

50- B

60-

70--

80--

90

100
50-


O-
60-

70-

80-

90-


GAIN
L 19% CP


I I
30
(G), KG/PEN
\\ 22% CP


Figure 4-14. Distribution of confidence intervals for broiler gain to
protein treatments for mash feed at week 8. Complete
data set (a), poor environment (b), and good environment
(c).


100


1s6% CP


25% CP


I


l I I


.










50

60

70

80

90

100
F:
z 50
LU
60
i- 60
LL.
W 70
0

w 80
O
z
w 90

z 100
0
0 50


60-

70-

80-

90

100 -
10

<16% CP


20 30
GAIN (G), KG/PEN
E]19% CP \ 22% CP


Figure 4-15. Distribution of confidence intervals for broiler gain to
protein treatments for pellet feed at week 8. Complete
data set (a), poor environment (b), and good environment
(c).


25% CP










may be compared to the curves generated from the complete data set

(Fig. 4-1). Obvious visual differences in these graphs reveal the

masked effects of environment on broiler gain and indicate the degree

of error if recommendations were made from averaged data. Comparison

of the graphs for the poor and good environments revealed a potential

means for generating more appropriate feeding recommendations for each

environment. By comparing the differences between the poor and good

environments for any age, a means for quantifying the technical

potential for the poor environment can be established. Further

investigation, through farm characterization can then be made to

isolate those factors that are contributing to the grower's good or

poor performance.

Examination of the graphs in Figures 4-8 and 4-9 will enable

comparison of feed form in different environments, providing a means

for refining feeding recommendations for specific environments.

Support for the above mentioned findings was then found in the

analysis of confidence intervals by partitioned environment.

Visualization of differences in the broiler gain response surface to

treatments and feed form under different environments can be made by

calculating the distribution of confidence intervals for good and poor

environment (Tables 4-8 and 4-9) and then graphing these values for

the different feed forms for week 4, 6 and 8 (Fig. 4-10b,c to

4-15b,c).











Table 4-8.


Distribution of confidence intervals of broiler gain to
protein treatments, for pellet form at weeks 4, 6, and 8
in kg/pen (partitioned data set)


Probability of a larger value (a)
0.50 0.40 0.20 0.10 0.05 0.01
----- -----------------
Protein (%) .765 .978 1.638 2.353 3.182 5.841
e-EEE=-----==--=:~===,_--,------- --
Week 4

poor 16% 7.398 7.577 8.130 8.730 9.426 11.657
6.114 5.935 5.382 4.782 4.086 1.855
19% 9.800 9.839 9.959 10.089 10.240 10.724
9.522 9.483 9.363 9.233 9.082 8.598
22% 13.587 13.679 13.961 14.267 14.622 15.760
12.933 12.841 12.559 12.253 11.898 10.760
25% 14.884 14.987 15.307 15.654 16.056 17.346
14.142 14.039 13.719 13.372 12.970 11.680

good 16% 7.523 7.638 7.995 8.381 8.828 10.264
6.697 6.582 6.225 5.839 5.392 3.956
19% 11.137 11.211 11.440 11.688 11.976 12.899
10.607 10.533 10.304 10.056 9.768 8.845
22% 14.146 14.244 14.547 14.875 15.256 16.476
13.444 13.346 13.043 12.715 12.334 11.114
25% 15.262 15.319 15.497 15.689 15.912 16.627
14.850 14.793 14.615 14.423 14.200 13.485

Week 6

poor 16% 13.407 13.648 14.396 15.207 16.147 19.163
11.671 11.430 10.682 9.871 8.931 5.915
19% 17.388 17.455 17.664 17.891 18.154 18.997
16.902 16.835 16.626 16.399 16.136 15.293
22% 23.450 23.650 24.272 24.946 25.726 28.231
22.008 21.808 21.186 20.512 19.732 17.227
25% 24.871 25.024 25.498 26.011 26.607 28.516
23.773 23.620 23.146 22.633 22.037 20.128

good 16% 14.221 14.506 15.388 16.344 17.452 21.007
12.175 11.890 11.008 10.052 8.944 5.389
19% 19.864 19.966 20.283 20.625 21.022 22.296
19.132 19.030 18.713 18.371 17.974 16.700
22% 25.067 25.184 25.546 25.937 26.392 27.849
24.229 24.112 23.750 23.359 22.904 21.447
25% 26.131 26.305 26.845 27.430 28.108 30.283
24.879 24.705 24.165 23.580 22.902 20.727
-= = ----===-----------------E----- ---






















Table 4-8.


Week 8

Probability of a larger value (a)
0.50 0.40 0.20 0.10 0.05 0.01
-------------------
Protein (%) .765 .978 1.638 2.353 3.182 5.841
==r~eer= ---------r=,=-- r n--23 ;
poor 16% 18.826 19.104 19.966 20.900 21.983 25.455
16.828 16.550 15.688 14.754 13.671 10.199
19% 24.343 24.587 25.342 26.160 27.108 30.150
22.593 22.349 21.594 20.776 19.828 16.786
22% 33.263 33.640 34.808 36.072 37.539 42.243
30.557 30.180 29.012 27.748 26.281 21.577
25% 34.103 34.455 35.545 36.726 38.096 42.488
31.575 31.223 30.133 28.952 27.582 23.190

good 16% 20.827 21.238 22.514 23.896 25.499 30.639
17.869 17.458 16.182 14.800 13.197 8.057
19% 28.000 28.157 28.643 29.170 29.781 31.741
26.872 26.715 26.229 25.702 25.091 23.131
22% 36.234 36.478 37.234 38.054 39.004 42.051
34.480 34.236 33.480 32.660 31.710 28.663
25% 37.013 37.357 38.426 39.584 40.926 45.231
34.535 34.191 33.122 31.964 30.622 26.317


Cont.












Table 4-9.


Distribution of confidence intervals of broiler gain to
protein treatments for mash form at weeks 4, 6, and 8 in
kg/pen (partitioned data set)


Probability of a larger value (a)
0.50 0.40 0.20 0.10 0.05 0.01
------ --- ------- --------
Protein (%) .765 .978 1.638 2.353 3.182 5.841

Week 4
Wee4------------ ------------
poor 16% 7.695 7.777 8.029 8.303 8.621 9.639
7.109 7.027 6.775 6.501 6.183 5.165
19% 10.572 10.750 11.301 11.898 12.590 14.810
9.294 9.116 8.565 7.968 7.276 5.056
22% 13.032 13.092 13.277 13.478 13.711 14.458
12.602 12.542 12.357 12.156 11.923 11.176
25% 13.617 13.650 13.751 13.860 13.987 14.394
13.383 13.350 13.249 13.140 13.013 12.606
--------" ------------- ---------
good 16% 8.427 8.494 8.701 8.926 9.186 10.021
7.947 7.880 7.673 7.448 7.188 6.353
19% 10.823 10.906 11.166 11.447 11.773 12.818
10.221 10.138 9.878 9.597 9.271 8.226
22% 13.176 13.181 13.199 13.217 13.239 13.308
13.136 13.131 13.113 13.095 13.073 13.004
25% 14.883 14.982 15.287 15.617 16.000 17.229
14.177 14.078 13.773 13.443 13.060 11.831

Week 6
------ ----- -------------
poor 16% 14.597 14.755 15.242 15.771 16.383 18.348
13.467 13.309 12.822 12.293 11.681 9.716
19% 19.362 19.681 20.669 21.739 22.980 26.961
17.072 16.753 15.765 14.695 13.454 9.473
22% 23.250 23.330 23.577 23.844 24.154 25.149
22.678 22.598 22.351 22.084 21.774 20.779
25% 24.662 24.769 25.101 25.461 25.878 27.215
23.892 23.785 23.453 23.093 22.676 21.339
-------------------- -------------
good 16% 16.062 16.194 16.604 17.047 17.561 19.209
15.114 14.982 14.572 14.129 13.615 11.967
19% 19.690 19.826 20.247 20.703 21.232 22.929
18.714 18.578 18.157 17.701 17.172 15.475
22% 24.149 24.192 24.328 24.476 24.646 25.194
23.833 23.790 23.654 23.506 23.336 22.788
25% 26.069 26.165 26.461 26.782 27.155 28.349
25.383 25.287 24.991 24.670 24.297 23.103






















Table 4-9.


Week 8

Probability of a larger value (a)
0.50 0.40 0.20 0.10 0.05 0.01

Protein (%) .765 .978 1.638 2.353 3.182 5.841
---r:r=~=====----------Le=S-=-----==----- --~~r~~=-L
poor 16% 20.919 21.173 21.959 22.810 23.798 26.965
19.097 18.843 18.057 17.206 16.218 13.051
19% 27.083 27.497 28.780 30.169 31.780 36.946
24.111 23.697 22.414 21.025 19.414 14.248
22% 33.634 33.785 34.253 34.760 35.348 37.233
32.550 32.399 31.931 31.424 30.836 28.951
25% 34.582 34.709 35.103 35.530 36.025 37.612
33.668 33.541 33.147 32.720 32.225 30.638

good 16% 22.945 23.108 23.612 24.159 24.792 26.824
21.777 21.614 21.110 20.563 19.930 17.898
19% 28.858 29.134 29.987 30.911 31.983 35.421
26.880 26.604 25.751 24.827 23.755 20.317
22% 34.476 34.585 34.921 35.285 35.707 37.060
33.698 33.589 33.253 32.889 32.467 31.114
25% 36.145 36.280 36.698 37.152 37.677 39.363
35.175 35.040 34.622 34.168 33.643 31.957
1-,-~--; E1,--------E--- -- =----iS-


Cont.









61

Table 4-10 compares the ranking of the visual differences in the

graphed distribution of confidence intervals in three ways:

1. Comparison of feed form and protein treatment within the

combined data set (overlay Fig. 4-10a with 4-11a; 4-12a with 4-13a;

4-14a with 4-15a).

Examination of the ranking from this comparison would lead to the

conclusion that with the 16 and 19% protein level the mash was

superior to pellets whereas at the 22 and 25% protein levels pellet

was superior to mash. Also, mash was more stable (narrower confidence

intervals) than pellets for all treatments except 19% protein. This

discrepancy may be explained by a review of the mortality records,

which indicated the deads and culls from the 19% treatment were

contributing to the instability in earlier weeks. With their removal

there was increased stability as compared to the mash treatment.

2. Comparison of feed form and treatment within a specific

environment (overlay Fig. 4-10b,c with 4-11b,c; 4-12b,c with 4-13b,c;

4-14b,c with 4-15b,c).

The scores noted in this comparison trial (Table 4-10), reveal an

obvious difference in the response of broiler gain to environment at

the 19% protein level, where pelleted feed in the good environment is

superior to mash. Stability scores demonstrate a similar response to

treatments in the partitioned environment as it did in the complete

data set.

3. Comparison of specific environments and treatments by feed

form (overlay all b Fig. with all c Fig., e.g.4-10b with 4-10c).











Table 4-10 demonstrates that in regard to mean broiler gain,

there is a consistent beneficial effect of a good environment over a

poor one for both feed forms. Stability of mash was clearly superior

in the good environment where the pellets produced a discrepancy in

the 16 and 19% pellet treatments as explained earlier. The 16%

pelleted treatment experienced heavy mortality at later ages.



Application of the Methodology

The broiler integrator should ask the following questions. Can

the differences between the good and poor growers be quantified? Can

these different production environments (good and poor) be

characterized in such a way as to be able to distinguish those factors

that impact on broiler production? What are the costs and benefits

for recommending more accurate feeding programs to good and poor

growers? What are the costs and benefits to the integrator for aiding

the poor growers to improve production in their environments?

Although the simulation study clearly demonstrated the ability

and advantages of interpolating results to partition a research domain

by environment, care should be taken when extrapolating results for

environments beyond the range of data.

In order to enhance the predictive capability of the procedure

the broiler integrator needs to incorporate and complete a detailed

characterization of a full range of grower environments within the

research domain. Data from existing production records along with the

detailed characterization would allow for the simulation and testing

on-station of a preliminary model. The researcher/integrator's













Table 4-10. Ranking of the graphed distribution of confidence
intervals to protein treatments for feed form and
environments


1) Comparison of feed form and protein treatment within the combined
data set.

MEANS STABILITY

Week 16% 19% 22% 25% 16% 19% 22% 25%

M P M P M P M P M P M P M P M P

4 ** -*
6 __. .* .* *
8 *

2) Comparison of feed form and protein treatment within an
environment.

4 poor *
6 poor *
8 poor *

4 good *
6 good *
8 good *

3) Comparison of particular environments and treatments by feed form.

Week 16% 19% 22% 25% 16% 19% 22% 25%

pr gd pr gd pr gd pr gd pr gd pr gd pr gd pr gd

4 mash *-.
6 mash *-
8 mash *-.4

4 pellet *_ *
6 pellet 44 *
8 pellet *


* denotes higher treatment mean or greater stability.
*--* denotes similar treatment means or stability.
M = mash, P = pellet, gd = good and pr = poor.










confidence in the accuracy of the characterization process and the

results of the simulation study should enable research station results

to be tested on-farm. A methodology for the scientific evaluation of

treatments on-farm would have to be determined so that production

costs to the integrator would be minimized.

Movement of research on-farm could be step-wise, where the

initial testing would be on the extremes in environment. If the

methodology proves to be sensitive in its ability to distinguish

between the differences in treatments due to environment and the

benefits of partitioning the research domain outweigh the costs, then

the on-farm research program could be expanded. This expansion should

include an adequate number of growers that span a wide range of

environments so that confidence in the data set in terms of stability

in the estimate of the treatment means is sound.

This simulation study emphasized the differences between growers

based on house temperature. Modified stability analysis enables the

partitioning of the research domain by the sum of all uncontrollable

factors (e.g. management, water quality, temperature, etc.) expressed

as an environmental index (e). Therefore, research conducted on

growers farms and managed by the farmers provides a more accurate

means for testing a proposed technology and a realistic way to

evaluate the effect of farm differences which arise from

socio-economic factors as well as agro-biological influences on the

broiler production process.

Results from the characterization, simulation and on-farm

research will enable the integrator to make important production









65

decisions based on knowledge generated from the application of

scientific methodologies.















CHAPTER V

ESTIMATION OF A RELIABLE BROILER PRODUCTION FUNCTION
WHEN USING ALTERNATE FEED INGREDIENTS


Introduction



Before the optimum biological response and economic returns from

a particular broiler feeding program can be evaluated, a reliable

estimate of the broiler gain production function must be established.

This response function is a mathematical expression or approximation

that defines the way in which live-weight gain (dependent variable)

changes as a result of changes in one or more independent variables,

such as feed ingredients, temperature, feed form, feed protein level

and broiler age. Numerous investigations have been conducted where

the functional form of the broiler gain response to corn-soybean meal

diets have been established (Dillon, 1977; Heady and Dillon,1961;

Heady and Shashanka, 1983; Hoepner and Freund, 1964; Naiyana and

Anderson, 1982; Pesti, 1982; Pesti and Fletcher, 1984; Roush, 1982).

Often, the empirical model chosen for approximating the broiler

response is a second order polynomial and is fitted to the

experimental data by least square regression methods. Typically,

corn-soybean meal studies were analyzed under the assumption that

satisfactory levels of all nutrients and environmental conditions

required for growth were given as adequate and constant.

66















CHAPTER V

ESTIMATION OF A RELIABLE BROILER PRODUCTION FUNCTION
WHEN USING ALTERNATE FEED INGREDIENTS


Introduction



Before the optimum biological response and economic returns from

a particular broiler feeding program can be evaluated, a reliable

estimate of the broiler gain production function must be established.

This response function is a mathematical expression or approximation

that defines the way in which live-weight gain (dependent variable)

changes as a result of changes in one or more independent variables,

such as feed ingredients, temperature, feed form, feed protein level

and broiler age. Numerous investigations have been conducted where

the functional form of the broiler gain response to corn-soybean meal

diets have been established (Dillon, 1977; Heady and Dillon,1961;

Heady and Shashanka, 1983; Hoepner and Freund, 1964; Naiyana and

Anderson, 1982; Pesti, 1982; Pesti and Fletcher, 1984; Roush, 1982).

Often, the empirical model chosen for approximating the broiler

response is a second order polynomial and is fitted to the

experimental data by least square regression methods. Typically,

corn-soybean meal studies were analyzed under the assumption that

satisfactory levels of all nutrients and environmental conditions

required for growth were given as adequate and constant.

66










If an accurate description of the nutrient profile of a feed

ingredient is not known due to variability in storage, humidity,

processing or in the product itself, then potential errors in feed

formulation may result, especially if these feed ingredients are fed

under varying environmental situations.

Therefore, under conditions typical to a developing country,

where the use of agro-industrial by-products as alternate feeds for

broiler production, and the environment (temperature) in which the

broilers are grown are both highly variable in their composition, then

it is advisable to estimate the functional form of the broiler gain

response under these specific conditions so that, accurate feeding

recommendations can be made.



Data and Methods



Experimental Data

The source of data for this analysis was from a broiler response

study conducted during the summer months (Aug.l-Oct.10,1986), at the

University of Florida Poultry Science research station. A commercial

strain of feather sexed broiler chicks (Cobb Cobb) were assigned at

random to 64 pens subject to including 16 chicks in each pen and

maintaining the 1:1 sex ratio. Feed treatments were specified in

terms of four different isocaloric diets containing 16, 19, 22, and

25% crude protein. The four protein treatments were fed in both mash

and pellet form in a four by two factorial arrangement for a total of

eight feed treatments and replicated eight times. To simulate the









68
effect of different environments the lower window flaps and roof vents

that cover the eight pens in each of the four corners of the building

were closed so these four environments (poor) would experience higher

ambient temperature as compared to the four environments (good) with

opened window flaps and roof vents. The experimental unit was the

total live-weight of the survivors in one pen (2.3225 square meters)

and expressed as live-weight gain (G) per pen.

Observations were made on live-weight gain and feed intake each

week through 10 weeks of age, as summarized in Appendix A. There

were 640 observations on gain, measured in kg/pen of live-weight; 640

observations on feed intake, measured in kg/pen of feed treatment; and

1120 observations on temperature, where the maximum and minimum

temperatures were recorded daily, then combined into 80 observations

on average weekly temperature ranges (AWTR), for each of the eight

environments over the ten-week period.



Fitting the Broiler Production Function

Three forms of the broiler gain response were estimated:

(1) Gain (G) = f(protein, temperature, feed form, age)

(2) Intake (I) = f(protein, temperature, feed form, age)

(3) Gain (G) = f(milo, peanut meal, temperature, feed form)


Evaluation of the Biological (Technical) Response

Model (1) was estimated so that an evaluation of the impact of

feed protein level, fed in different forms and under different









69

environmental temperatures, on broiler live-weight gain could be made

at different ages in the broiler production process.

Model (2) was estimated so that a comparison of the terms that

are significant in the broiler gain model can be examined in relation

to the significant terms in the intake model.



Evaluation of the Efficiency of Feed Inputs

Model (3) was estimated so that an evaluation of the efficiency

of milo and peanut meal, fed in different forms and under different

environmental temperatures, on broiler live-weight gain could be made

at different ages in the broiler production process.

The functional form for both the broiler gain response and intake

as the dependent variables was estimated in terms of the independent

variables protein, temperature, feed form, and age in model (1) and

(2); and milo, peanut meal, temperature and feed form in model (3).

This was accomplished by using the interactive multiple linear

regression program in the statistical analysis package Statpak (North-

west Analytical, 1986).



Partitioning the Broiler Gain Response by Environment

Analysis of the graphic representation of the broiler gain

response to alternate feeds after partitioning the complete data set

into more homogeneous environments, indicates that the functional form

of the broiler gain response will differ between environments, feed

form and age, as explained in Chapter 4, pg #.










The procedure used here was first, to estimate both broiler

production function models with the full set of independent variables,

secondly, to partition the data set by feed form and broiler age, and

thirdly, to fix the temperature variable by incorporating a constant

temperature value into the appropriate coefficients (parameters).

This procedure was useful for the testing of the significance of

feed form in the complete model as well as to reduce the number of

independent variables to two so that a three dimensional graphic

representation of the broiler response could be produced. All figures

presented here (Fig. 5-1 to 5-8) were developed using the G3D option

of the SAS/GRAPH program of the SAS statistical package (Council and

Helwig, 1981) and then plotted on a Versetec plotter.



Results and Discussion



Data on live-weight gain, cumulative feed intake, cumulative milo

and peanut meal intake, and feed efficiency results for 3 to 10 weeks

of age appear in Appendix A.



Technical Response

The coefficients of determination (R-squared) for 3 to 10 week

live-weight gain (Table 5-1) and 3 to 10 weeks of broiler feed intake

(Table 5-3) indicate that these equations (models) account for

approximately 96 percent of the variation observed in the simulation

study. The estimates of the coefficients (parameters) for all

variables and interactions tested for both gain and intake (combined










Table 5-1.


Estimation of the coefficients of regression for broiler
gain from 3 to 10 weeks production when fed various
protein levels, in mash or pellet form and under varying
temperatures (estimates + standard error)


Performance Gain Gain Gain
parameters (combined) (pellet form) (mash form)

Intercept -11.313 +1.993** -10.162 +3.0377** -11.152 +2.502**
Age 39.967 +6.956** 47.056 +10.641** 32.367 +8.769**
Temperature 0.513 +0.152** 0.480 +0.232* 0.548 +0.191**
Age x temp. -0.080 +0.022** -0.095 +0.232** -0.066 +0.028**
Age x protein -6.028 +1.045** -7.224 +1.600** -4.827 +1.317**
Age x protein2 0.327 +0.052** 0.391 +0.079** 0.263 +0.065*
Age x protein3 -0.006 +0.001* -0.007 +0.001* -0.005 +0.001**

Age x form 0.296 +0.057**
Feed form x
protein3 0.001 +0.001**

Coefficient of
determination (R2) 0.959 0.942 0.961

* Significant (P<.025).
**Significant (P<.01).




Table 5-2. Regression analysis of variance for broiler gain from 3 to
10 weeks production when fed various protein levels, in
mash or pellet form and under varying temperatures

Source Sum of Degrees of Mean
squares freedom squares F

1. Gain (combined mash and pellet)
Due to regression 53,739 8 6,717 1,195**
Residual 2,826 503 5.619
Total 56,565 511 110.69

2. Gain (pellet form)
Due to regression 26,579 6 4,430 673**
Residual 1,637 249 6.576
Total 28,217 255 110.65

3. Gain (mash form)
Due to regression 27,189 6 4,531 1,015**
Residual 1,112 249 4.465
Total 28,300 255 110.98

** Significant (P<.01).










Table 5-3.


Estimates of the coefficients of regression for broiler
feed intake from 3 to 10 weeks when fed various protein
levels, in mash or pellet form and under varying
temperatures (estimates + standard error)


Performance Intake Intake Intake
parameter (combined) (pellet form) (mash form)

Intercept -5.236 +4.340** -6.016 +5.931** -1.872 +5.730**
Age 43.878 +15.143** 63.212 +20.778** -6.315 +2.225**
Temperature -1.478 +0.330** -1.342 +0.453** -1.608 +0.438**
Age x temp. 0.283 +0.049** 0,223 +0.067** 0.341 +0.064**
Age x protein -6.813 +2.275** -10.076 +3.122** 0.950 +0.200**
Age x protein2 0.384 +0.002** 0.561 +0.154** -0.016 +0.005**
Age x protein3 -0.007 +0.002** -0.010 +0.003**

Age x form -0.669 +0.125**

Feed form x
protein3 0.001 +0.001**

Coefficient of
determination (R2) 0.966 0.969 0.971

** Significant (P<.01).




Table 5-4. Regression analysis of variance for broiler intake from 3
to 10 weeks production when fed various protein levels, in
mash or pellet form and under varying temperatures

Source Sum of Degrees of Mean
squares freedom squares F

1. Intake (combined mash and pellet)
Due to regression 385,175 8 48,147 1,808*
Residual 13,395 503 26.631
Total 398,570 511 779.98

2. Intake (pellet form)
Due to regression 192,435 6 32,073 1,279*
Residual 6,244 249 25.074
Total 198,679 255 779.13

3. Intake (mash form)
Due to regression 193,617 6 38,723 1,653**
Residual 5,854 249 23.415
Total 199,471 255 782.24
--* Significant (P<.01).
", Significant (P<.01).











Table 5-5.


Estimates of the coefficients of regression for broiler
gain from 1 to 10 weeks of age when fed milo and peanut
meal, in mash or pellet form and under varying
temperatures (estimates + standard error)


Performance Gain Gain Gain
parameters (combined) (pellet form) (mash form)

Intercept 0.727 + 0.113** 0.849 + 0.163** 0.658 + 0.121**
Milo 0.465 + 0.070** 0.487 + 0.026** 0.447 + 0.017**
Peanut meal 1.428 + 0.045** 1.593 + 0.064** 1.280 + 0.047**
Milo2 -0.005 + 0.001** -0.006 + 0.001** -0.004 + 0.001**
Peanut meal2 -0.017 + 0.001** -0.019 + 0.001** -0.014 + 0.001**
Temp.x PM -0.021 + 0.001** -0.038 + 0.005** -0.008 + 0.004**
Temp.xPMxMilo 0.001 + 0.001** 0.001 + 0.001** 0.001 + 0.001*

Coefficient of
determination (R2) 0.989 0.989 0.994

** Significant (P<.01).









Table 5-6. Regression analysis of variance for broiler gain from 1 to
10 weeks of age when fed milo and peanut meal, in mash and
pellet form and under varying temperatures

Source Sum of Degrees of Mean
squares freedom squares F
--------------------- ------------
1. Gain (combined mash and pellet)
Due to regression 92,879 6 15,480 9,456**
Residual 1,036 633 1.637
Total 93,915 639 146.97

2. Gain (pellet)
Due to regression 45,744 6 7,624 4,508**
Residual 529 313 1.69
Total 46,273 319 145.06
--------------------------
3. Gain (mash)
Due to regression 47,307 6 7,884 8,324**
Residual 296 313 0.947
Total 47,604 319 149.23
-- Significant (P<.).---------------------------
" Significant (P<.01).










data set) were significant (P<.01). The regression analysis of

variance for broiler live-weight gain (Table 5-2) and broiler feed

intake (Table 5-4) show a significant F value (P<.01) indicating that

the models are relatively precise descriptions of the relations

defined by the production function. This result is in agreement with

other studies (Pesti and Fletcher, 1984; Arraes, 1983 as cited in

Pesti and Fletcher, 1984) where the inclusion of age as a continuous

variable improved the fit for protein intake.

The significant effect of the interaction of feed form with age

or protein on both gain and intake was used as the basis for dividing

the combined data set into separate mash and pellet data sets. All

estimated coefficients remained significant (P<.01) after separation

of the data, with the exception of the temperature coefficient in

pellet form and on gain (P<.025). Also, the interaction of age with

the cubic effect of protein on intake (mash form) was not significant

(P>.05).

The models predicting gain for mash and pellet form, (Gm) and

(Gp), were then used to generate three dimensional graphs of the

broiler live-weight gain response to protein treatment at 3 to 10

weeks of age and for good and poor environments (Fig. 5-1 to 5-4). It

can be seen from these figures that the response curves became

increasingly cubic in form as the flock aged. The characteristics of

a cubic form are a slight depression of the response on the 19 and 25%

protein treatments. This result was in contrast to the reports by

Pesti (1982) and Heady (1983) when feeding corn-soybean meal diets.

These three dimensional plots suggest that the nutrient profile of the














GRIN = KG/SQM
RGE = WEEKS
PROTEIN = PERCENT


GRIN


17.98





12.77





7.56


Figure 5-1.


Broiler gain response surfaces to protein treatments for
pellet feed in a good environment by weeks of age.















GRIN = KG/SQM
RGE = WEEKS
PROTEIN = PERCENT


GRIN


19.60





13.63





7.65





1.68


9





















Figure 5-2.


25




22

7 PROTEIN

AGES
19


4

3 16









Broiler gain response surfaces to protein treatments
for pellet feed in a poor environment by weeks of age.















GAIN = KG/SOQ
AGE = WEEKS
PROTEIN = PERCENT


Figure 5-3.


Broiler gain response surfaces to protein treatments
for mash feed in a good environment by weeks of age.


GAIN


18.06





13.03





7.99


3 16
3














GRIN = KG/SOU
RGE = WEEKS
PROTEIN = PERCENT


GAIN


18.45





12.88





7.31





1.75
1


Figure 5-4.


25






7 PROTEIN

RGE65
19
7

4

3 16









Broiler gain response surfaces to protein treatments
for mash feed in a poor environment by weeks of age.










alternate feeds used in this study decreased at a more rapid rate

between 22 and 19% protein than was found between similar treatments

using corn-soybean meal diets. This was especially true for the

pelleted treatments in poor environments. It was also possible to see

how the gain response to 25% protein decreased relative to 22% protein

as the flock aged. This indicated that protein requirements decreased

with increasing age, as previously reported by Roush (1982). Also, it

appears that compensatory growth on the 22% protein treatment was

superior to all other treatments probably due to the well balanced

nutrient profile of this feed treatment.



Efficiency Model

The coefficient of determination (R-squared) for the 3 to 10 week

live-weight gain data, mash, pellet and combined data sets (Table 5-5)

indicates that these models account for approximately 99 percent of

the variability observed in the broiler study. The estimates of the

coefficients for the linear, quadratic and interaction terms, were all

significant (P<.01) in the mash, pellet and combined data set for the

equations describing live-weight gain.

The regression analysis of variance for the gain models all had

significant F values (P<.01) indicating that broiler live-weight gain

can accurately be described by the models estimated in this study.

The significant effect of temperature with milo and peanut meal

was to be expected since feed intake was shown to be significantly

affected by temperature in the technical model. The significance of

the interaction of temperature with peanut meal suggests that the gain










response to protein derived from peanut meal was affected by

temperature.

The (Gm) and (Gp) models (mash and pellet form of model 3) as a

function of milo and peanut meal intake under varying environments

(temperature) were used to generate three dimensional response

surfaces (Fig. 5-5 to 5-8). As with the technical model temperature

was held constant at 10 and 18 F (AWTR) representing good and poor

environments, respectively. Inspection of the plotted broiler gain

response surfaces gives a clear view of the diminishing returns of

live-weight from both peanut meal and milo.



Characteristics of the Response

The quadratic broiler gain response to corn soybean meal diets

reported by Dillon (1977), Heady and Dillon (1961), Heady and

Shashanka (1983), Pesti and Fletcher (1984), was the basis of a new

least-cost feed formulation computer program for the broiler industry

(Pesti et al., 1986). The diets used in those studies were well

balanced for all nutrients which enabled a relatively good response to

low protein treatments (16 and 19%). In this study, the imbalance of

nutrients at the low protein levels provoked a rapid decrease in the

gain response between the 22 and 19% protein treatments resulting in a

cubic rather than a quadratic gain response curve. Still, it was

possible to model the gain response and to accurately predict broiler

gain for a range of protein treatments and environments. These

results show that under circumstances often found in developing

countries, where the nutritional profiles of feed ingredients have not














ALL UNITS (KG/SQM)


GAIN
22.51

20.25

17.99

15.73

13.48

11.22

8.96

6.70

4.44

2.18
34.4
31.6
28.






















Figure 5-5.


21.53


18.58


Predicted broiler gain response surface with milo and
peanut meal inputs for pellet form in a good
environment.














ALL UNITS (KG/SQl)































21.53


18.58


Figure 5-6.


Predicted broiler gain response surface with milo and
peanut meal inputs for pellet form in a poor
environment.


GRIN
19.32

17.40

15.48

13.56

11.64

9.72

7.80

5.88

3.96

2.04
34















ALL UNITS (KG/SQM)
































21.53


18.58


Figure 5-7.


Predicted broiler gain response surface with milo and
peanut meal inputs for mash form in a good environment.


GRIN
19.90

17.99

16.08

14.17

12.25

10.34

8.43

6.52

4.61

2.70
34.4
31