• TABLE OF CONTENTS
HIDE
 Title Page
 Dedication
 Acknowledgement
 Table of Contents
 List of Tables
 List of Figures
 Abstract
 Chapter I: Introduction
 Chapter II: Modified stability...
 Chapter III: Effects of fertilization...
 Chapter IV: Summary and conclu...
 Appendix
 Literature cited
 Biographical sketch






Title: Yield and yield stability of pure and mixed stands of sorghum (Sorghum bicolor (L.) Moench) varieties in north Cameroon
CITATION PAGE IMAGE ZOOMABLE PAGE TEXT
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00056220/00001
 Material Information
Title: Yield and yield stability of pure and mixed stands of sorghum (Sorghum bicolor (L.) Moench) varieties in north Cameroon
Physical Description: xiii, 184 leaves : ill. ; 28 cm.
Language: English
Creator: Russell, John T., 1951-
Publication Date: 1991
 Subjects
Subject: Sorghum -- Yield -- Cameroon   ( lcsh )
Agronomy thesis Ph. D
Dissertations, Academic -- Agronomy -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
 Notes
Thesis: Thesis (Ph. D.)--University of Florida, 1991.
Bibliography: Includes bibliographical references (leaves 177-182).
Statement of Responsibility: by John T. Russell.
General Note: Typescript.
General Note: Vita.
Funding: Electronic resources created as part of a prototype UF Institutional Repository and Faculty Papers project by the University of Florida.
 Record Information
Bibliographic ID: UF00056220
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: aleph - 001690069
oclc - 25151733
notis - AJA2111

Table of Contents
    Title Page
        Page i
    Dedication
        Page ii
    Acknowledgement
        Page iii
        Page iv
    Table of Contents
        Page v
    List of Tables
        Page vi
        Page vii
    List of Figures
        Page viii
        Page ix
        Page x
        Page xi
    Abstract
        Page xii
        Page xiii
    Chapter I: Introduction
        Page 1
        Sorghums of Northern Cameroon
            Page 1
            Page 2
            Page 3
            Page 4
            Page 5
        Sorghum research in the center north zone
            Page 6
            Page 7
            Page 8
            Page 9
        Sorghum yield stability
            Page 10
            Page 11
    Chapter II: Modified stability analysis of Sorghum variety tests in north Cameroon, 1984-87
        Page 12
        Introduction
            Page 12
            Page 13
            Page 14
            Page 15
            Page 16
            Page 17
            Page 18
            Page 19
        Materials and methods
            Page 20
            Page 21
            Page 22
            Page 23
        Results and discussion
            Page 24
            Page 25
            Page 26
            Page 27
            Page 28
            Page 29
            Page 30
            Page 31
            Page 32
            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
            Page 66
            Page 67
            Page 68
            Page 69
            Page 70
    Chapter III: Effects of fertilization and planting pattern on grain yield and yield stability of mixed stands of long and short cycle Sorghum varieties
        Page 71
        Introduction
            Page 71
            Page 72
            Page 73
            Page 74
            Page 75
            Page 76
            Page 77
            Page 78
            Page 79
            Page 80
        Materials and methods
            Page 81
            Page 82
            Page 83
            Page 84
            Page 85
            Page 86
            Page 87
            Page 88
            Page 89
            Page 90
            Page 91
            Page 92
        Results and discussion
            Page 93
            Page 94
            Page 95
            Page 96
            Page 97
            Page 98
            Page 99
            Page 100
            Page 101
            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
            Page 120
            Page 121
            Page 122
            Page 123
            Page 124
            Page 125
            Page 126
            Page 127
            Page 128
            Page 129
            Page 130
            Page 131
            Page 132
            Page 133
            Page 134
            Page 135
            Page 136
            Page 137
            Page 138
            Page 139
            Page 140
            Page 141
            Page 142
            Page 143
            Page 144
            Page 145
            Page 146
            Page 147
            Page 148
            Page 149
            Page 150
            Page 151
            Page 152
            Page 153
    Chapter IV: Summary and conclusions
        Page 154
        Page 155
        Page 156
        Page 157
        Page 158
        Page 159
        Page 160
        Page 161
        Page 162
        Page 163
    Appendix
        Page 164
        Page 165
        Page 166
        Page 167
        Page 168
        Page 169
        Page 170
        Page 171
        Page 172
        Page 173
        Page 174
        Page 175
        Page 176
    Literature cited
        Page 177
        Page 178
        Page 179
        Page 180
        Page 181
        Page 182
    Biographical sketch
        Page 183
        Page 184
        Page 185
        Page 186
Full Text












YIELD AND YIELD.. STABILITY OF PURE AND MIXED
STANDS..OF SORGHUM (Sorghum bicolor (L.) Moench]
VARIETIES IN NORTH CAMEROON












By

JOHN T. RUSSELL


A DISSERTATION PRESENTED TO THE GRADUATED SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY






UNIVERSITY OF FLORIDA


1991


























This work is dedicated to the memory of my father

John T. Russell Sr.















ACKNOWLEDGEMENTS


I extend my sincere appreciation to Dr. Clifton K.

Hiebsch, committee cochair, for his invaluable guidance and

encouragement throughout this research. I also thank Drs.

D. A. Knauft (chair), K. L. Buhr, R. A. Littell, and P. E.

Hildebrand for serving on my committee. Their interest and

enthusiasm were always a great help and comfort. I thank

the IFAS Office of International Programs, particularly Dr.

Hugh Popenoe, for financing my farming systems assistantship

and for consenting to my doing my research in Cameroon.

The research presented here was conducted while the

author was employed by the International Institute of

Tropical Agriculture, Ibadan, Nigeria. I thank I.I.T.A.,

especially Drs. L. D. Stifel, J-P. Eckebil, and D. S. C.

Spencer, for allowing and, in fact, encouraging this work.

The research in North Cameroon could not have been completed

without the support and assistance of the following people:

Dr. J. Ayuk-Takem, Director, Institute of Agronomic Research

(IRA), Cameroon; Mr B. Z. Boli, Chief of Center, CRA, IRA

Maroua; Drs. E. A. Atayi, Chief of Party, and T. Stilwell,

Deputy Chief of Party, National Cereals Research and

Extension Project (IRA/IITA/USAID). Particular thanks are


iii









due Drs. O. P. Dangi and L. Singh for sharing with me their

extensive knowledge of sorghum and sorghum production, and

to Mr. R. Ndikawa for his collaboration in the on-station

sorghum intercrop trial discussed in this work.

I will long be indebted to Dr. K. McDermott for first

seeing the possibility of sending me to Cameroon, and to Mr.

John Dorman and Mr. Gary Cohen for helping to make it

happen.

Finally, my deepest gratitude and respect is offered to

the staff, former and present, of the Testing and Liaison

Unit, IRA Maroua, including Jerry J. Johnson, Dr. M.

Kamuanga, M. T. Fobasso, the late Jean Nzoning, Njomaha

Charles, Samaki Joseph, Diak Albert, Koda Oumarou, Tsabgou

Tonfack Mathias, Hamidou Mal Toukour, Boubakary Hamadou,

Mamoudou Hamadou, Haman Guipili, Mrs. Yvette Yepmo, and

Yaouba. Their hard work and dedication were a constant

source of reassurance, motivation, and inspiration.

















TABLE OF CONTENTS

oacte
ACKNOWLEDGEMENTS...................................... iii

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

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

ABSTRACT................. ................................. xii

CHAPTERS

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

Sorghums of Northern Cameroon........................ 1
Sorghum Research in the Center North Zone............6
Sorghum Yield Stability..............................10


II MODIFIED STABILITY ANALYSIS OF SORGHUM
[Sorghum bicolor (L.) Moench] VARIETY
TESTS IN NORTH CAMEROON, 1984-87.................... 12

Introduction................................ ....... 12
Materials and Methods..............................20
Results and Discussion.............................. 24


III EFFECTS OF FERTILIZATION AND PLANTING
PATTERN ON GRAIN YIELD AND YIELD
STABILITY OF MIXED STANDS OF LONG AND
SHORT CYCLE SORGHUM VARIETIES........................71

Introduction. ........... ............... .............71
Materials and Methods..............................81
Results and Discussion............................. 93


IV SUMMARY AND CONCLUSIONS............................154

APPENDIX... ............................................... 164

LITERATURE CITED................. ......................... 177

BIOGRAPHICAL SKETCH............. .........................183

v
















LIST OF TABLES


TABLE Mace

2.1 Summary of 1984 on-farm variety test
results (from SAFGRAD, 1984; and
Johnson, 1988) ................................... 22

2.2 Summary of 1985 on-farm variety test
results (from SAFGRAD, 1985; and
Johnson, 1988)........................ .... ....... 22

2.3 Summary of 1986 and 1987 on-farm variety
test results (from Testing and Liaison
Unit, Maroua, 1986, 1987; and Johnson, 1988).........23

2.4 Regressions of yields of S35 and locals on
three segments of the range of E, 1984-87...........48

2.5 Regressions of E (S35 and locals) on
date of seeding (DOS) and on rainfall,
1984-87, 1984, 1985, and 1984-85....................65

2.6 Correspondence of date of seeding and
sectors where S35 was adopted, 1984-87 data........66

2.7 Correspondence of date of seeding and
sectors where S35 was adopted, 1984-85 data.........68

3.1 Soil physical and chemical properties,
on-station mixed stands trial, 1989-90..............83

3.2. Subplot treatments for on-station trial
of mixed stands of sorghum varieties............... 84

3.3 Dates of cultural operation, on-station
sorghum mixed stands trial, 1989-90................86

3.4 Number of workers, and soil physical and
chemical properties, on-farm sorghum
mixture tests......... ...........................91

3.5 Years each field farmed, previous crops,
and field operation dates, on-farm sorghum
mixture test................................. ..... ....92










3.6 Mean rainfall by decade of days for
on-station sorghum mixed stand trial................94

3.7 ANOVA by year for grain yield; on-station
trial of sorghum mixed stands, 1989-90..............95

3.8 Main effects of (a) location and (b)
fertilizer level on grain yield, by year;
on-station trial of sorghum mixed stands,
1989-90............................................. ..97

3.9 Significance of sources of variation,
contrasts, and interactions, for grain
yield, by year and location; on-station
trial of sorghum mixed stands, 1989-90..............98

3.10 Means of intercrop treatments compared
to highest-yielding pure crop component,
by year; on-station trial of sorghum
mixed stands, Guetale............................. 100

3.11 Means of intercrop treatments compared
to highest-yielding pure crop component,
by year; on-station trial of sorghum
mixed stands, Mouda ............................... 101

3.12 Means of intercrop treatments compared
to highest-yielding pure crop component,
by year; on-station trial of sorghum
mixed stands, Tchatibali........................... 102

3.13 Environments (36 reps) in on-station
trial of sorghum mixed stands, sorted
into low (E<2000) and high (E>2000)
recommendation domains. ............................114

3.14 Characteristics of on-farm tests sites,
1989-90 sorghum mixtures, ranked by
environmental index......................... ....... 140

3.15 Simple linear regressions of
environmental index from 1989-90
sorghum mixture tests on quantitative
environmental characteristics.....................141

A.1 Data from 1984-87 on-farm variety tests............ 165

A.2 Data from on-station mixed stands trial.............170

A.3 Data from 1989 and 1990 on-farm mixtures test.......176


vii
















LIST OF FIGURES


FIGURE page

2.1 Modified Stability Analysis, all
varieties, 42 sites, north regions,
1984..................................................28

2.2 Modified Stability Analysis, S35
and locals, 42 sites, north regions,
1984................................................29

2.3 Modified Stability Analysis, all
varieties, 46 sites, central regions,
1984...................... ........................... 30

2.4 Modified Stability Analysis, S35
and locals, 46 sites, central regions,
1984 ..............................................31

2.5 Modified Stability Analysis, S35
and locals, 87 sites, 1984........................32

2.6 Modified Stability Analysis, S35 and
locals, 42 early-seeded sites, 1985...............35

2.7 Modified Stability Analysis, S35 and
locals, 16 late-seeded sites, 1985..................36

2.8 Modified Stability Analysis, S35 and
locals, 38 sites, 1986..............................37

2.9 Modified Stability Analysis, S35 and
locals, 35 sites, 1987.............................38

2.10 Modified Stability Analysis, S35 and
locals, 239 sites, 1984-87.........................45

2.11 Modified Stability Analysis, hypothetical
curved response of S35 and locals, 1984-87..........46

2.12 S35 and locals plotted on environmental
index, 239 sites, 1984-87................ ...... .....47

2.13 S35 and locals plotted on environmental
index, 87 sites, 1984................... ............ 49

viii










2.14 S35 and locals plotted on environmental
index, 46 sites, 1984-85.........................50

2.15 Modified Stability Analysis, S35 and
locals, 146 sites, 1984-85......................... 51

2.16 Modified Stability Analysis, S35 and
locals, 58 sites, 1985 ......................... 67

2.17 Lower confidence limits for S35 and locals,
by early- and late-seeded sites....................69

2.18 Lower confidence limits for S35 and locals,
by early- and late-seeded sectors..................70

3.1 'Modified stability analysis regression
of pure stand yields of CS-54 (C),
walaganari (W), and damougari (D).................. 108

3.2 Modified stability analysis regression
of mixed and pure stand yields of
damougari (D) and walaganari (W).................109

3.3 Modified stability analysis regression
of mixed and pure stand yields of
CS-54 (C) and walaganari (W).......................110

3.4 Modified stability analysis regression
of 3-variety mixed stand yields and
pure stand yields of CS-54 (C),
walaganari (W), and damougari (D).................11.

3.5 Component grain yields of damougari (D),
CS-54 (C), and walaganari (W), low fertility
domain, as pure stands (P), mixtures (M), or
alternate-row stands (A).......................... 117

3.6 Component grain yields of damougari (D),
CS-54 (C), and walaganari (W), high fertility
domain, as pure stands (P), mixtures (M), or
alternate-row stands (A)..........................118

3.7 Lower confidence limits, damougari (D) and
walaganari (W) mixed stands, high fertility........124

3.8 Lower confidence limits, damougari (D) and
walaganari (W) mixed stands, low fertility.........125

3.9 Lower confidence limits, CS-54 (C) and
walaganari (W) mixed stands, high fertility........126










3.10 Lower confidence limits, CS-54 (C) and
walaganari (W) mixed stands, low fertility......... 127

3.11 Lower confidence limits,'damougari (D),
walaganari (W), and CS-54 (C) mixed stands,
high fertility..................................... 129

3.12 Lower confidence limits, damougari (D),
walaganari (W), and CS-54 (C) mixed stands,
low fertility..................................... 13C

3.13 MSA regression of S35 and djigari pure stands,
on-farm sorghum mixture tests.....................132

3.14 Scatter plot of S35 and djigari, on-farm
tests of sorghum mixtures..........................133

3.15 MSA regression of S35 and djigari mixtures,
on-farm test of sorghum mixtures..................134

3.16 MSA regression of mixture dominated by S35,
compared to pure stands of its components..........135

3.17 MSA regression of mixture dominated by
djigari, compared to pure stands of its
components .............................13

3.18 MSA regression of 2:2 mixture of S35 and
djigari, compared to pure stands of its
components ......................................... 13

3.19 Replacement series, early-seeded on-farm tests
of sorghum mixtures.............................. 14

3.20 Replacement series, late-seeded on-farm tests
of sorghum mixtures ............................... 14!

3.21 Lower confidence limits, mixture dominated by
S35, early-seeded environments (S=S35,
D=djigari)........................ .................. 141

3.22 Lower confidence limits, mixture dominated by
djigari, early-seeded environments (S=S35,
D=djigari) ......................................... 14*

3.23 Lower confidence limits, 2:2 mixture,
early-seeded environments (S=S35,
D=djigari) ....... ..... .......... ............

3.24 Lower confidence limits, mixture dominated
by S35, late-seeded environments (S=S35,
D=djigari) .... ........ ........ ... ... ........... ..... 15










3.25 Lower confidence limits, mixture dominated
by djigari, late-seeded environments (S=S35,
D=djigari)........................... ............. 15

3.26 Lower confidence limits, 2:2 mixture,
late-seeded environments (S=S35,
D=djigari)............. .............................. 15:















Abstract of Dissertation Presented to the Graduate
School of the University of Florida in Partial
Fulfillment of the Requirements for the Degree
of Doctor of Philosophy


YIELD AND YIELD STABILITY OF PURE AND MIXED
STANDS OF SORGHUM [Sorghum bicolor (L.) Moench]
VARIETIES IN NORTH CAMEROON

By

JOHN T. RUSSELL

August, 1991

Chair: Dr. D. A. Knauft
Cochair: Dr. C. K. Hiebsch
Major Department: Agronomy



The stability of crop yields under diverse and

unpredictable environmental conditions is increasingly

recognized as an important goal of limited-resource farmers

in developing nations, perhaps even more important than

overall average yield as a criterion for evaluating crop

varieties or other production technologies.

Modified Stability Analysis (MSA) was done on the

results of four years of on-farm sorghum varietal tests in

northern Cameroon. The improved variety S35 was more stable

in different environments than farmers' local varieties.

The major identifiable determinant of productivity was date

of seeding. Differences in the known adoption pattern of


xii










S35 in northern Cameroon were concluded to be due to a

greater superiority of S35 over local varieties, both in

terms of yield and of reduced risk, when seeding is late

rather than early.

A research station trial of mixed stands of sorghum

varieties, done in three locations over two years, supported

the conclusion that benefits from intraspecific mixed stands

of cereals usually result from positive mixing effects due

to complementarity in high-yielding environments.

Compensation-induced mixing effects, expected in poor

environments, were not evident. Mixed-row stands were

usually superior to seed mixtures. Mixed stands

occasionally yielded more than the mean pure stand yield of

their components, but did not perform better than the

highest-yielding component. Mixed stands were no more

stable, according to MSA, than their most stable component.

On-farm tests of S35 and djigari also showed mixtures

of an improved and a local sorghum to be no more stable

across environments than the more stable component, and

usually less stable than the mean pure stand yields of the

two components. Mixtures appeared to have no major benefit

in terms of grain yield or of reduced risks in either early

seeded or late-seeded recommendation domains.


xiii
















CHAPTER I


INTRODUCTION



Sorghums of Northern Cameroon

Northern Cameroon, which consisted previously of a

single large province, presently comprises three--Adamaoua,

North, and Extreme North. It is a remarkably rich and

diverse region, ranging from the sparsely-populated, wooded,

Guinea savanna of the southern Adamaoua plateau, at a

latitude of about 60N, through the Sudanian and Sudan-

Sahelian savannas of North and Extreme North Provinces, to

the true Sahel of the Lake Chad region, at almost 130N. The

striking ethnic, cultural, geographical, and ecological

diversity of this region has inspired a large and detailed

literature, largely descriptive in nature.

The cotton-growing portion of Extreme North Province

and the Mayo Louti Department of North Province was the

target zone of a large-scale development project, the Center

North project, and this area is still often called the

Center North Zone. In this region, sorghum [Sorghum bicolor

(L.) Moench] is the staple food crop. In many of the more

northern parts of this zone, pearl millet [Pennisetum
















CHAPTER I


INTRODUCTION



Sorghums of Northern Cameroon

Northern Cameroon, which consisted previously of a

single large province, presently comprises three--Adamaoua,

North, and Extreme North. It is a remarkably rich and

diverse region, ranging from the sparsely-populated, wooded,

Guinea savanna of the southern Adamaoua plateau, at a

latitude of about 60N, through the Sudanian and Sudan-

Sahelian savannas of North and Extreme North Provinces, to

the true Sahel of the Lake Chad region, at almost 130N. The

striking ethnic, cultural, geographical, and ecological

diversity of this region has inspired a large and detailed

literature, largely descriptive in nature.

The cotton-growing portion of Extreme North Province

and the Mayo Louti Department of North Province was the

target zone of a large-scale development project, the Center

North project, and this area is still often called the

Center North Zone. In this region, sorghum [Sorghum bicolor

(L.) Moench] is the staple food crop. In many of the more

northern parts of this zone, pearl millet [Pennisetum











americanum (L.) Leeke] is increasingly grown as annual

rainfall progressively decreases. In the Mandara Mountains

to the west of the Center North Zone, millet is important in

alternate-year rotation with sorghum, as a pest control

strategy, and to the east it is frequently grown because of

its tolerance of the very sandy soils there (Testing and

Liaison Unit, Maroua, 1990a,b). Nevertheless, sorghum

remains by far the most important and widely grown cereal.

Most of the sorghums of northern Cameroon are

rainfed, depending entirely on rains that in Extreme North

Province average around 750 mm per year and generally begin

in late May to early June and end in early October, although

erratic and late onset of rains is a perennial problem.

Rainfall distribution is monomodal, with a peak in August,

but annual distribution at any given site is notoriously

unpredictable. As in many parts of semi-arid West and

Central Africa, northern Cameroon has suffered from an

extended drought episode, that is, a period in which

drought years are more frequent than usual. Farmers tend

to think that the current drought episode, which has lasted

more than a decade and a half, is due primarily to

decreasing rainfall. Nicholson (1986) cites several factors

that may have contributed to drought, including overgrazing,

overcultivation, and removal of vegetation, but concludes

that the fundamental cause of the current drought is

meteorological.











Estimates of average sorghum yields vary from

approximately 750 kg ha-1 (Dangi, 1987) to 1100 kg ha-1 (NCRE,

1989), but most estimates are in the 900 kg ha-1 range. In

addition to rainy season sorghums, some 30% of sorghum

production comes from types (called mouskwari or babouri)

seeded into nurseries in August and transplanted into

vertisols known as karal toward the end of the rainy

season. These transplanted types mature on residual soil

moisture and are harvested from mid-January to early

February.

Sorghum is usually consumed as a stiff porridge known

locally in French as boule, and also as a watery gruel

(bouille). Local brewing of sorghum beer is also very

common. In addition to human consumption of sorghum grain,

sorghum stalks are important as a source of fodder and in

some regions as a building material.

Traditional rainy season sorghums in the Center North

zone can generally be grouped into three broad classes. The

first are medium to long cycle (100-130 day), tall (3-4 m)

varieties with compact panicles and a brown subcoat which

gives the boule made from them a dark red color. Within

this class (all from the Snowden sub-series Caudatum), there

are varieties with red to brown, floury grain (generally

known as djigari) and others with white to grey, floury

grain (boulbassiri). These varieties are adapted to a wide

range of soil types but produce best on sandy loams.












Varieties of djigari in particular are very widely grown.

Medium to long cycle varieties without a brown subcoat

exist, notably walaganari, and are highly appreciated by the

dominant ethnic group, the Fulbe, for the white boule they

produce.

A second broad class of short to very short cycle (85-

100 day) varieties, such as damougari and makalari, are

found, if not very widely grown. These are sometimes

considered to be types of diigari by farmers, who tend to

use that name for any red, rainy-season sorghum, regardless

of cycle.

A third class of very long cycle (130-180 day)

varieties, types such as yolobri or mbairi, once common as

far north as Maroua (10060'N), are now rarely found much

farther north than Garoua (9020'N). The northernmost

boundary of these types is steadily moving southward in

response to the long-term trend of decreasing rainfall in

the Sudan-Sahelian zone (Hallaire, 1984). All of these

traditional varieties are photoperiod sensitive; farmers

tend to select at any given planting date--often beyond the

control of farmers in the case of late rains--varieties

whose cycle will allow them to mature with or soon after the

end of the rains.

It is of interest to note that this sort of

classification of sorghum types by maturity group in the

Center North Zone is slowly changing in response to the











changing rainfall pattern. Thus the 120-day varieties of

djigari, considered short cycle varieties in the 1960s and

early 1970s, are increasingly considered long cycle ones now

that the 150-180 day varieties are moving southward and the

very short cycle varieties--both locals and those bred by

the research service--are increasing in importance.

The diversity of sorghums grown in northern Cameroon is

as great as that of the region itself. A systematic census

of all types of local sorghums in 1968 resulted in notation

of approximately 2700 varieties and in final tabulation of

more than 1000 (Eckebil, 1970a). In addition to differences

in cycle, farmers recognize and value for different uses a

range of grain types, forage qualities, and adaptations to

local climatic, edaphic, ecological, and cultural

conditions. Differences in taste and culinary preferences,

even within a single ethnic group, are also reasons for a

large number of local varieties.

A given field of traditional sorghum in the Center

North is unlikely to be very homogeneous. Part of the

heterogeneity may be due to microclimatic and edaphic

variability within the field, but a considerable part is

often due to the genetic variability of the crop itself.

Farmers in fact seldom grow a single variety, even within a

single field. A field of "djigari," for example, is

frequently a mixture of two or several kinds of djigari.

Farmers will mix varieties of equal maturity cycle as well











as those of different cycles. When asked why they use such

mixtures, farmers will often first respond only that they

have always done so. Further questioning almost always

shows that farmers recognize advantages from this practice,

primarily that it may reduce their risk of low yields or

total crop loss. That their traditional varieties differ in

resistance to environmental stresses such as drought, Stricra

hermonthica, or weed, insect and disease pressure, is well

known by farmers, as is the ability of individual components

of a mixture to compensate for poor performance by other

components. Use of mixtures, though rarely an openly

expressed strategy, appears to be an attempt (either

conscious or unconscious) to minimize risk in the face of

variable and unpredictable environmental conditions.


Sorghum Research in the Center North Zone

Agronomic research in northern Cameroon is the province

of the Institute of Agronomic Research (IRA), which has its

center in Maroua, capital of Extreme North Province.

Although much money and effort is devoted to research on the

major cash crop, cotton, primary emphasis for some time has

been placed--as a matter of official policy--on development

and improvement of indigenous food crop production

(N'Sangou, 1978).

Since the late 1970s, research on sorghum, as well as

on other food crops in the sorghum-based cropping system,












has been conducted by two USAID-funded projects. Until

1986, the Semi-Arid Food Grains Research and Development

Project (SAFGRAD) conducted both on-station and (after 1981)

on-farm trials on sorghum, millet, maize, peanuts, and

cowpeas. When SAFGRAD ended in northern Cameroon in 1986,

the on-farm testing component of food crops research in

Maroua was taken over by the National Cereals Research and

Extension (NCRE) Project. The NCRE project has attempted to

institutionalize Farming Systems Research and Extension

(FSR/E) in Cameroon by developing a number of Testing and

Liaison Units (TLUs) within IRA centers such as Maroua.

These TLUs are responsible for farming systems diagnosis,

on-farm research, and research-extension linkage. A major

focus of both SAFGRAD and TLU/NCRE on-farm research has been

regional tests of improved sorghum varieties and agronomic

practices.

In addition to its TLU component, NCRE has helped IRA

develop its capacity for cereal crop improvement. The NCRE

sorghum and millet breeding program in Maroua dates from

1981. These efforts have built upon more than two decades

of breeding work by IRAT, which included much important

classification and collection of local germplasm, both of

rainy season and dry season transplanted types (Eckebil,

1970a,b; Monthe, 1977).

The sorghum improvement program has for a number of

years based its program, with on-going SAFGRAD and TLU










8

collaboration, on the actual constraints to improved sorghum

production faced by farmers of the Center North Zone. These

constraints include poor and erratic rainfall, often

disastrously distributed during the growing season; striga,

which is increasing in importance as both soil fertility and

the length of fallow period decrease; labor constraints at

the time of sowing and weeding, which impede improvement in

land preparation and weed control; and lack of credit for

yield-enhancing inputs such as animal traction, fertilizer,

and pesticides. Further constraints include a variety of

insects, primarily several species of head bugs, stalk

borers (especially Busseola fusca), nutgrass armyworm

(Spodoptera exempta), and shoot fly (Atherigona soccata),

all of which can be severe at times. There exists also an

array of endemic leaf diseases, but their economic

importance is undocumented.

In response to these constraints, the sorghum breeding

program in Maroua has concentrated on developing for the

Center North Zone short cycle (85-95 day) drought tolerant

varieties, medium in height (2.5 m), and resistant to

striga, diseases, and insects. These varieties generally

have white grain of medium vitreousness, and are of a "tan"

plant type (Dangi and Djonnewa, 1988). While much

appreciated by farmers for high yield potential, stalk

forage quality, and a cycle which allows late planting in

response to retarded onset of rains, these varieties have











met with some farmer resistance. The short cycle of these

varieties is linked to a lack of photoperiod sensitivity.

While contributing to their resistance to drought (through

avoidance), this same characteristic leads to an increased

susceptibility to grain mold (Fusarium moniliforme, Fusarium

semitectum, Curvularia lunata, etc.) when the varieties are

planted too early. Of equal concern is that these

varieties, whose grain lacks tannins and other polyphenolic

compounds, are as highly appreciated by birds as by humans.

Losses to birds can occasionally be quite severe and are

aggravated by the relatively small areas devoted to improved

varieties at present. Also, their short cycle leads to

maturity earlier than the longer cycle white-seeded

varieties that might otherwise equally attract birds and

thus dilute the damage caused by them.

Some of these varieties have produced high yields both

in on-station and on-farm trials; a few, due to this

performance, have been extended to farmers. The most widely

extended and successful of these varieties, all of which are

open-pollinated, is S35, a selection from varieties sent to

Cameroon by Dr. N. G. P. Rao, working then in Zaria,

Nigeria. Adoption rates, however, have generally been

disappointing. In fact, after five years of serious

extension efforts, adoption of these varieties as part of an

improved production package, has been limited to one or two

specific geographical areas.












Sorghum Yield Stability

The Center North Zone, for which these 85- to 95-day

varieties have been selected, is a region with a strong

rainfall gradient from north to south, averaging 850 mm or

more per annum in the Mayo Louti Department of North

Province, but well less than 600 mm in the northern Mora

Plain in Extreme North Province. There also exists an east-

west rainfall gradient, with lower rainfall in the eastern

Mayo Danay Department accentuated by the very sandy soils

found there. Of even more concern than average total annual

rainfall, however, is the extreme variability of rainfall,

both from year to year and from one site to another (even

sites only hundreds of meters apart) within a year.

Despite the many other factors that can limit sorghum

yields, farmers almost always cite poor and erratic rains as

the major constraint to production. Variability in yields

can therefore be extreme from one farm to the next in any

given year. More crucially, total regional yields can vary

as much as 300% from year to year, and given the sensitivity

of local markets to supply fluctuations, grain prices can

vary 1000% or more between a good and a very bad year

(Johnson, 1988). Risk to farmers from erratic rainfall,

aggravated by additional risks from insects, diseases, and

striga--all affected to some extent by rainfall--is great.

For this reason, an ever increasing emphasis within IRA

Maroua has been placed, both by the sorghum breeding program












and by the TLU, not just on increasing average sorghum

yields, but also on increasing yield stability across

environments.

The problem to be addressed in this dissertation can

best be expressed by posing two questions. First, if

stability of production and minimization of risk are

important criteria in a sorghum improvement program in the

Sudan-Sahelian zone, how can these factors be effectively

evaluated and used in establishing recommendation domains

for new varieties? In, particular, how can Modified

Stability Analysis contribute to such a program? Second, to

what extent can mixtures of long cycle and short cycle

cultivars of sorghum, particularly mixtures of local and

improved types, increase yields and yield stability across

environments and decrease risk to farmers?

The first objective of the research presented here,

then, is to examine the usefulness of Modified Stability

Analysis (Hildebrand, 1984; 1990) in the development and

dissemination of improved sorghum varieties in the semi-arid

zone of West Africa, using the example of the variety S35 in

northern Cameroon. The second objective is to investigate

the possible advantages in this region of mixtures of long

and short cycle sorghum varieties, particularly in terms of

increasing yield stability and thus decreasing risk to

small-scale farmers.















CHAPTER II

MODIFIED STABILITY ANALYSIS OF SORGHUM
[Sorghum bicolor (L.) Moench] VARIETY
TESTS IN NORTH CAMEROON, 1984-87


Introduction


Various methods of analyzing genotype-by-environment

(GE) interaction have been proposed and developed over the

years. These had as their impetus a recognition that the

interaction of a variety with the environments in which it

is grown is often as important or more important than its

average yield over all environments.

Some, usually earlier, methods of estimating GE

interaction followed the example of Sprague and Federer

(1951), who separated the effects of genotype, environment,

and GE interaction by comparing the observed mean squares

from analysis of variance to the expected mean squares of

the random model. This approach gave rise to a number of

stability parameters. Wricke (1962), for example, proposed

ecovalence, W2j, and Shukla (1972) identified a stability

variance, a2i. These, as well as the mean variance component

for pairwise GE interaction, 8, (Plaisted and Peterson,

1959) and the variance component for GE interaction, 9(

(Plaisted, 1960), all involved partitioning of the overall















CHAPTER II

MODIFIED STABILITY ANALYSIS OF SORGHUM
[Sorghum bicolor (L.) Moench] VARIETY
TESTS IN NORTH CAMEROON, 1984-87


Introduction


Various methods of analyzing genotype-by-environment

(GE) interaction have been proposed and developed over the

years. These had as their impetus a recognition that the

interaction of a variety with the environments in which it

is grown is often as important or more important than its

average yield over all environments.

Some, usually earlier, methods of estimating GE

interaction followed the example of Sprague and Federer

(1951), who separated the effects of genotype, environment,

and GE interaction by comparing the observed mean squares

from analysis of variance to the expected mean squares of

the random model. This approach gave rise to a number of

stability parameters. Wricke (1962), for example, proposed

ecovalence, W2j, and Shukla (1972) identified a stability

variance, a2i. These, as well as the mean variance component

for pairwise GE interaction, 8, (Plaisted and Peterson,

1959) and the variance component for GE interaction, 9(

(Plaisted, 1960), all involved partitioning of the overall











GE variability. Francis and Kannenberg (1978) considered

the standard coefficient of variability, CVi, as a measure

of stability. Use of these methods varied; agronomists and

breeders tried to minimize GE effects, and geneticists tried

to determine the underlying causes of such interactions.

Regression of mean varietal yields on overall site

means, as a method of analyzing GE interaction, has a long

history. First proposed by Mooers (1921), it was developed

by Yates and Cochran (1938), but largely ignored thereafter,

until Finlay and Wilkinson (1963) rediscovered it and used

it to define "general adaptability" of a variety i as high

yield and a linear regression coefficient of unity (bi = 1).

Eberhart and Russell (1966), using a similar technique,

added the deviation from regression mean square (a2di) as a

parameter of interest, and defined a "stable variety" as one

with high yield, bi = 1, and a2i = 0.

Criticisms of the technique of regressing individual

variety yields on overall site mean yields have not been

lacking. Such criticism centers on the obvious mathematical

interdependence between the theoretically independent

variable (site mean), and the several dependent variables

(individual variety yields) regressed on it (Freeman and

Perkins, 1971). In response to this and other perceived

shortcomings, a number of alternative stability parameters

have been advanced. These include methods for grouping

genotypes by clustering techniques (Hanson, 1970; Lin and











Thompson, 1974; Lin and Binns, 1985), and by pattern

analysis (Mungomery et al., 1974; Shorter et al., 1977).

Tai (1971) proposed a parameter ai, an unbiased estimate of

(b-1), where b is Eberhart and Russell's regression

coefficient. Hardwick and Wood (1972) proposed regression

on a set of environmental variables external to and

independent of yields of the test varieties, a technique

expanded upon by Wood (1976).

General reviews of the many methods of stability and GE

interaction analysis have defended and justified the

approach of linear regression on an environment index (site

mean) by comparing its advantages to those of competing

measures. These advantages include simplicity of

calculation and ability to provide information on variety

response over a range of test environments, information

essential for cultivar recommendations (Freeman, 1973; Lin

et al., 1986). Until prediction of yield response as a

function of external environmental variables, as attempted

by Nor and Cady (1979), becomes feasible, this technique is

likely to remain popular. It has been employed in stability

analysis of sweet potato (Martin et al., 1988), peanut

(Anderson et al., 1989; Schilling et al., 1983), maize (Yue

et al., 1990), and wheat (Musaviun and Ehdaie, 1987).

One of the major advantages of regression on an

environmental index is that it permits assessment of GE (or

any treatment-by-environment) interaction without











replication within a test site. This advantage is of

particular interest to Farming Systems Research and

Extension (FSR/E) researchers conducting on-farm trials

which are often limited to a single replication per farm.

Hildebrand (1984) proposed a Modified Stability Analysis

(MSA), based on Eberhart and Russell's technique, as a tool

for the analysis of farmer managed on-farm trials.

As use of MSA by FSR/E practitioners increases, the

question is often posed: Just what is "modified" about

Modified Stability Analysis? First, while Eberhart and

Russell consider a "stable" variety to be one with a bi of

unity, MSA defines stability across environments as bi = 0.

A difference of this sort makes sense in the FSR/E context.

Whereas breeders in the U.S. may want varieties that respond

in step to improving environmental conditions, farmers in

most developing nations, while desiring good production in

high-yielding environments, consider maintenance of adequate

production in low- or very low-yielding environments to be

more important. This is especially true since conscious

improvement of growing conditions--through irrigation,

fertilization, etc.--is often beyond the means of most

resource-limited farmers. Many crop breeders now use b=0

rather than b=l as a criterion of stability in different

environments. A significant difference still exists between

MSA and the stability analysis of most breeders in that

breeders tend to consider other criteria such as mean square











deviations from regression in addition to the regression

coefficient, while MSA limits itself to b.

Second, MSA extends the use of regression on

environmental index beyond varietal trials to trials of any

production technology. Third, by adding to linear

regression a graphic distribution of confidence intervals

for treatment means, MSA includes the concept of low risk to

farmers in the definition of stability. Finally, MSA is an

effort to identify specific adaptability of varieties or

other technologies to particular sets of environments, i.e.,

recommendation domains, rather than the general adaptation

to all environments proposed by Finlay and Wilkinson

(Hildebrand, 1990).

Two further characteristics of MSA highlight its

usefulness as a FSR/E analytical tool. Rather than being

limited to yield (i.e., kg ha'1) response only, regression of

other variables of interest such as monetary return per unit

land area, per unit seed planted, per unit labor expended,

etc., can also be done. In this event, however, the

environmental index (E) is still calculated on a kg ha'1

basis (P. E. Hildebrand, 1991, personal communication).

Also, the environment, as expressed by E, is implicitly

thought of as defined by all those parameters determined by

"good" or "poor" farmer management--weeding, degree of pest

control, etc.--as well as the more commonly considered

"biophysical parameters" such as rainfall, soil fertility,











toposequence, etc. This complexity of factors determining

the value of E for any given on-farm test site complicates

the predictive classification of environments into low and

high domains, a classification essential for recommendation

of specifically adapted technologies. While the regression

component of MSA is often encountered in agronomic and FSR/E

literature, graphing of confidence intervals is less common;

one exeception is a study of peanut varieties in stressed

and unstressed environments (Knauft and Gorbet, 1989).

Classification of environments is almost universally

neglected, but Singh (1990) used it for fertilizer

technologies in the Amazon basin of Brazil.
Several studies have investigated differences in

stability of sorghum varieties in the semi-arid savanna

zones of Africa. In Uganda, Jowett (1972) found eight

varieties tested to have generally smaller coefficients of

regression (mean b = 0.81) than either eight single crosses

(mean b = 1.11) or five three-way crosses (mean b = 1.09).

Deviations from regression were similarly lower for the

varieties. Unfortunately, insistence on the concept that

b = 1 rather than b = 0 denotes stability led to the

conclusion that the hybrids were more stable than the

varieties. Kambal and Mahmoud (1978) also based selection

of the most stable of 16 varieties tested in Sudan on

Eberhart and Russell's three criteria of high yield,

regression coefficient of unity, and low deviations from









18

regression. In Nigeria, Obilana and El-Rouby (1980) tested

varieties in four savanna zones over three years. They

considered the regression coefficient to be a "parameter of

adaptation," i.e., adaptation to high- or low-yielding

environments. Deviations from regression and coefficient of

determination (R2) they considered "parameters of

stability," applicable chiefly to high-yielding

environments. While implicitly recognizing the goal of

specific adaptability, Obilana and El-Rouby, like Kambal and

Mahmoud, stressed partitioning the GE interaction into

components for deciding on the relative importance of years,

locations, and replications in a sorghum breeding program.

In northern Cameroon, on-farm variety tests on a large

number of farms in 1984 indicated tremendous yield

superiority of one improved variety, S35, over local types.

Since 1984 was a catastrophically poor rainfall year

throughout the Sudan-Sahelian zone, the economic value of

this superiority was even greater than the absolute yield

differences might suggest. Because of these results, S35

was multiplied in 1985, and its extension, as part of an

improved sorghum production package, began in 1986.

Despite further on-farm confirmation from 1985 to 1987

that S35 yields were superior or equal to those of local

varieties, adoption has never quite "taken off." Rather,

for the entire Center North zone, it has varied from year to

year between 300 and 650 ha, as estimated by the local











extension agency. Reasons variously offered for this

limited adoptions have included the susceptibility of S35 to

grain mold and birds, as noted above. Some farmers also

maintain that S35 requires fertilizer and/or plowing to

produce well, or that it does not do well on their

particular soil types, or that it does not store well, or

that its culinary characteristics or taste are not locally

acceptable. In some individual SODECOTON (Societe de

developpement du coton du Cameroun) sectors, however,

adoption of S35 is widespread and increasing. This adoption

of S35 is not always accurately measured by SODECOTON since

some farmers plant it without adopting the entire production

package, and SODECOTON only keeps records on the package.

The purpose of the research reported in this chapter is

to evaluate, using the on-farm test data from 1984 to 1987,

the usefulness of Modified Stability Analysis in assessing

the stability--and concomitant risk to farmers--of improved

short cycle sorghum varieties and in delimiting

recommendation domains for their adoption in the semi-arid

West African savanna zone. In light of more than five

years' experience of extension of S35 throughout the Center

North Zone of Cameroon, this analysis hopes to answer two

questions. First, why is S35 being adopted in some places

and not in others? Second, could MSA in the first year or

two of on-farm tests of S35 have predicted where this

variety is now well accepted by farmers and where it is not?












Materials and Methods



On-farm sorghum variety tests were conducted by SAFGRAD

Maroua in 1984, 1985 and 1986, and by TLU/NCRE Maroua in

1987, in collaboration with SODECOTON. In each test, a

farmer-selected local variety was used as a check. In 1984,

tests were divided into two groups, with 42 on-farm sites in

the two "northern" SODECOTON regions of Diamare and Mora-

Mokolo, and 46 sites in the three "central" regions of

Kaele, Mayo Danay and Mayo Louti. In response to good

performance in 1984 of S35, a variety thought to do best

when seeded late, on-farm tests in 1985 were again divided

into groups, with 42 sites seeded before 15 June, and 16

seeded after 15 June. Varietal tests, not divided into

groups, were done at 38 sites in 1986 and at 35 in 1987.

Each on-farm test was conducted on a 0.25 ha field

(i.e., 50 m X 50 m), with a collaborating farmer chosen by

the local SODECOTON field extension agent and monitored

throughout the season by that agent. Between-row spacing

was 0.80 m; within-row spacing was 0.40 m between hills.

Plants were thinned, beginning seven days after planting, to

two per hill, resulting in densities of 62,500 plants ha1.

In 1984, recommended seeding dates were 20 June to 10 July,

choice of preceding crop was left to the farmer, and all

sites were fertilized with 61-13-25 kg ha-1 N-P-K. In 1985,

fertilizer was reduced to 46 kg N ha'1 as urea, and all sites












had cotton as the preceding crop. Test conditions in 1986

and 1987 were as in 1985 except that recommended date of

seeding was 10 to 20 June, in order to avoid favoring too

much the early maturing, improved varieties (Johnson, 1988).

Means, CVs, and standard errors from analysis of variance of

these trials are presented in Tables 2.1 to 2.3.

For the present study the linear regression component

of Modified Stability Analysis was carried out, by year, for

each of the tests described above. Since only S35 and the

farmers' local checks were common across years, MSA was also

done, by year, using only S35 and locals regressed on an

environmental index based on these two. The same sort of

regression was done on the set of all test sites across all

four years. This set of 239 sites included 21 striga-

infested sites in 1985 not included in the year-by-year

regression analyses.

Based on the regression of S35 and locals across years,

an attempt was made to hypothesize a higher order (quadratic

or other) response of each of these varieties to

environment. In an effort to partition the test sites into

recommendation domains based on environmental index (E), a

cursory search for patterns of coincidence of nonnumerical

site characteristics such as soil type, geographical

location, etc. with E was done but proved unsuccessful.

Simple and multiple regression of E on quantitative

environmental measurements, using both all possible













Table 2.1 Summary of 1984 on-farm variety test results
(from SAFGRAD, 1984; and Johnson, 1988).


North Regions (42 Sites)


Variety


S35
Local
38-3


S.E.
LSD(o.o0


Rainfall(mm)
Mean Seeding Date


Central Regions (46 Sites)


Variety


Yield
flr ha-1


1070 at
598 b
596 b


S35
Local
E35-1


98.8
271


393
7/3


Yield
(ka ha-)


1573 a
829 -b
975 b

109.0
305


359
7/6


t Means in a row followed by the same letter are not
significantly different (p=0.05) by the LSD procedure.
$ Rainfall is total seasonal rainfall.





Table 2.2 Summary of 1985 on-farm variety test results
(from SAFGRAD, 1985; and Johnson, 1988).


Early-Seeded (42 Sites) Late-Seeded (16 Sites)


Variety Yield Variety Yield
(ka ha-l) (kg ha-1)


S35 1866 S35 1416 at
Local 1721 Local 1156 ab
S34 1666 NS S36 806 bc
S20 601 c

S.E. 117.7 160.1
LSD(o.o 330 453

Rainfall(mm)$ 504 515
Mean Seeding Date 6/14 6/28

t Means in a row followed by the same letter are not
significantly different (p=0.05) by the LSD procedure.
$ Rainfall is total seasonal rainfall.


ri ru
rri~ ru r --- ---













Table 2.3 Summary of 1986 and 1987 on-farm variety test
results (from Testing and Liaison Unit, Maroua, 1986, 1987;
and Johnson, 1988).



1986 (38 Sites) 1987 (35 Sites)

Variety Yield Variety Yield
(kg ha-1) (kg ha-1)


CS54
CS61
S35
Local

S.E.
LSD(O. 0

Rainfall(mm)f
Mean Seeding Date


CS54
CS61
S35
Local


2201
2168
2164
2128 NS

145.4
406

621
6/22


2013
1934
1889
1825 NS


123.4
345


604
6/19


t Rainfall is total seasonal rainfall.












regressions and stepwise techniques, was carried out.

SODECOTON sectors were ordered by those variables important

in determining E to see whether this order corresponded to

degree of adoption of S35. Within each of the

"recommendation domains" identified in the preceding steps,

confidence intervals were graphed for each variety in order

to estimate the role that reduced risk of low yields may

have played in adoption of S35 in each domain. Finally, the

site characterization, recommendation domain identification,

and risk estimation steps were repeated for the 1984 and

1985 test subsets to see whether the same conclusions might

have been reached using MSA earlier in the testing and

dissemination process.


Results and Discussion



1984

In the two northern regions of the Center North zone,

in 1984, the regression component of MSA indicated that the

improved, short-cycle variety S35 was superior in yield to

38-3 and locals across all test environments; this

superiority increased as E increased (Figure 2.1). While

deviations from regression mean squares for S35 and locals

were greater than those for 38-3, their similarity to each

other suggested that the regression responses of the two

varieties could be safely compared. Eliminating 38-3, which









25

is common to none of the later tests, from the analysis, and

regressing the yields of S35 and locals on an E based on

only these two varieties, gave much the same results as MSA

of all three varieties (Figure 2.2). S35 was again superior

across the entire range of environments. The relationship

of the responses of S35 and local yields can be considered

one of "low-end convergence." There was little or no

difference in very low-yielding environments (E = 0)--which

is to be expected, given the nature of the calculation of

E--and an ever-increasing difference in the yields of the

two varieties as E increases. Examination of the range of

E, from very near zero to 2300, showed a concentration of

sites with E between 750 and 1200, and another concentration

at E<300.

MSA regression for the three central regions of the

Center North zone in 1984 also indicated superiority of S35

across all environments, and deviations from regression mean

squares for S35 and locals similar to each other and greater

than those of the third variety, E35-1 (Figure 2.3). Again,

relative responses of S35 and locals analyzed alone (Figure

2.4) were little different from those of the analysis

including three varieties.

For all of the separate tests to be discussed here,

this similarity between S35 and local responses in a two-

variety analysis and their responses when all the varieties

in the tests were included held true. For this reason, as











well as because the comparison of S35 to locals is the one

of chief interest, for the sake of clarity only the two-

variety regressions will be presented. It could be argued

that the statistical anomaly of interdependence of E and

individual component yields for which MSA has been

criticized can only be aggravated by restricting the

analysis to two varieties. It should be recalled, however,

that E is a measure not just of "the environment" but of the

productivity of the environment, and more specifically of

the environment's capacity to produce a particular thing, in

this case sorghum grain. Further, one must remember that

inferences from this kind of regression are applicable only

to the varieties included in the analysis, and only for the

environments represented by the test environments. If we

are concerned, then, with the yield and stability of S35

relative not to all other sorghum varieties but to varieties

specifically adapted to local conditions, the estimate of

"environment" we use should logically be the capacity of the

environment to produce S35 and/or locals; there is little to

be gained by muddying E by including its capacity to produce

other varieties not of interest for the comparison at hand.

This reductionist approach to MSA, when justified by

the goals of the analysis, can provide other beneficial

insights into the mechanics and implications of the

technique. As can be gleaned from study of the figures

illustrating the two-variety regressions, the mean of the









27

slopes of the two regression lines is always unity. Perhaps

not as obvious is that since E for any given site will be

the mean of only two varieties or varietal means, the

deviations from regression mean squares will be the same for

each of the two varieties. It follows that R2 for the

variety with b>l is greater than R2 for the other with b
and that the difference in R2 corresponds to the difference

in b; such was the case for each of the analyses presented

here. These measures of "goodness of fit" are of no

interest, then, in regressions of two varieties on an

environmental index calculated as the mean of two yields.

Comparing the central regions in 1984 (Figure 2.4) to

the northern ones (Figure 2.2), the relative responses of

the two varieties, while still converging somewhat at the

low end of the continuum of environments, converged so much

less than for the northern regions that they could be

considered relatively "parallel" across the range of E. The

distribution of E showed generally higher values than in the

northern regions, with a concentration between 1000 and

1500, and no concentration at values less than 500.

Combining the northern and central regions, the

responses of S35 and locals across the whole of the Center

North zone were, not surprisingly, midway between those of

each of the two groups of regions. They were perhaps more

like those of the north regions, with a distinct low-end

convergence (Figure 2.5). The reason for this was not just












1984, North Regions


Grain Yield (t/ha)

835: b-1.34, MS Dev. Reg.-129,000
Locals: b-0.82, MS Dev. Reg.-106,000
38-3: b-0.84, MS Dev. Reg.-62,000


500 1000 1500 2000
Environmental Index (E)


Figure 2.1


835
-4- Locals
- 38-3


2500 3000


Modified Stability Analysis, all varieties, 42
sites, north regions, 1984.












1984, North Regions, S35 vs. Locals


Grain Yield (t/ha)


835
-4- Locals
0 E


0 500 1000 1500 2000 2500 3000
Environmental Index (E)


Figure 2.2


Modified Stability Analysis, S35 and locals, 42
sites, north regions, 1984.













1984, Central Regions


Grain Yield (t/ha)

S35: b-1.04, MS Dev. Reg.-144,000
-Locals: b-0.89, MS Dev. Reg.-136,000
E35-1: b-1.06, MS Dev. Reg.-92,000


S35
--Locals
--- E35-1


0 500 1000 1500 2000 2500 3000 3500 4000
Environmental Index (E)









Figure 2.3 Modified Stability Analysis, all varieties, 46
sites, central regions, 1984.


I I I


I 1 i 1












1984, Central Regions, S35 vs. Locals


Grain Yield (t/ha)


S35
--- Locals
0 E


0 500 1000 1500 2000 2500 3000 3500 4000
Environmental Index (E)


Figure 2.4


Modified Stability Analysis, S35 and locals, 46
sites, central regions, 1984.











1984, S35 vs. Locals


Grain Yield (t/ha)


S35
-4- Locals
0 E


0 500 1000 1500 2000 2500 3000 3500
Environmental Index (E)








Figure 2.5 Modified Stability Analysis, S35 and locals,
87 sites, 1984.











chance, but rather can be explained by examination of the

distribution of environmental indices. Over all 88 sites,

there was a general concentration of Es between 500 and

1500. There is still, nevertheless, an important

concentration of E values very near zero which exert great

influence on the overall shape of the relative responses.



1985

The responses of S35 and locals in the early-seeded

tests of 1985 (Figure 2.6) were different from those in the

1984 tests. The relative linear responses of the two

varieties were, if not parallel, then slightly convergent

toward the high end of the E continuum. Yield of S35 was

not at all different from local yields in high-yielding

environments. While ANOVA of early-seeded sites showed no

overall difference in yields between the two varieties

(Table 2.2, p. 21), MSA regression indicated that S35 may in

fact be superior in the lower-yielding environments. The

distribution of environments is characterized by higher

overall yields than in 1984, with no sites yielding less

than 500 kg ha-', several with E>2500, and a quite broad

concentration of E values between 1000 and 2500.

Late-seeded tests in 1985, with distribution of E

values more similar to 1984 than to early-seeded 1985 tests

also had yield responses across environments similar to 1984

(Figure 2.7). Differences from one group of tests to











another in 1984 and 1985 seemed very much tied to,

if not entirely determined by, the distribution of E.



1986 and 1987

Regressions of the two varieties in 1986 and 1987 were

more like each other than like any of the 1984 or 1985

tests. It might be argued that the 1986 tests (Figure 2.8)

displayed a low-end convergence, as in 1984, and that in

1987 there was a cross-over of the regression lines, with

S35 producing better than locals in low-yielding

environments and worse than locals in high-yielding

environments (Figure 2.9). More to the point, perhaps, the

yields of S35 and locals, with no overall mean difference

(Table 2.3), also appeared to be very close to each other or

identical across the entire range of environments.

Modified Stability Analysis is intended to allow

identification of specifically adapted varieties or

technologies in a limited time by substituting locations for

years. One would hope, therefore, that any trials with a

certain minimum number of sites would produce pretty much

similar results. If the first year of on-farm tests with

S35 had been conducted in 1985, particularly if there had

been no second group of sites where farmers had been

instructed purposefully to seed late, the prospects for

extension of the variety might have been quite different

than they were after the 1984 tests. Even had MSA been












1985, Early-Seeded, S35 vs. Locals

Grain Yield (t/ha)
4A.,


- S35
-+- Locals
0 E


0 500 1000 1500 2000 2500 3000
Environmental Index (E)


Figure 2.6


3500 4000


Modified Stability Analysis, S35 and locals, 42
early-seeded sites, 1985.


835: b-0.96
3 Locals: b-1.04


0 I0M8=wa)OD W90aw moo 0













1985, Late-Seeded, S35 vs. Locals


Grain Yield (t/ha)

SS35: b-1.05
Locals: b-0.95









0 (z 000 0 0 0 0 0 0 00


0 500 1000 1500 2000 2500 3000
Environmental Index (E)


Figure 2.7


S35
--- Locals
0 E


Modified Stability Analysis, S35 and locals, 16
late-seeded sites, 1985.












1986, S35 vs. Locals


Grain Yield (t/ha)

S35: b-1.01
Locals: b-0.99


S35
--- Locals
O E


0 500 1000 1500 2000 2500 3000 3500 4000
Environmental Index (E)


Figure 2.8


Modified Stability Analysis, S35 and locals, 38
sites, 1986.


0 0 0 0 0 0 000 no 0 00 OMM8D 000I 0












1987, S35 vs. Locals


Grain Yield (t/ha)


S35: b-0.94
3 Locals: b-1.06


0 500 1000 1500 2000 2500 3000
Environmental Index (E)


Figure 2.9


S35
-4- Locals
0 E


3500 4000


Modified Stability Analysis, S35 and locals, 35
sites, 1987.


0 <3 000OWO 08 00 0 0 I O 0 0









39

used, one might have concluded in 1984 that S35 was superior

in high-yielding environments (Figure 2.5)--what one

traditionally expects from an "improved" variety. Had 1985

been the first year of testing, and had no specifically

late-seeded tests been done that year, MSA might have been

interpreted to show S35 superior in low-yielding

environments (Figure 2.6). On the other hand, however, had

either 1986 or 1987 been the first year of testing, one

might have concluded either that S35 was superior in low-

yielding environments, or that there was, in fact, no real

difference between S35 and locals in either group of

environments.

It could also be argued that the year-to-year

differences between relative response across environments of

S35 and of locals are unimportant. This view would hold

simply that S35 in all groups of tests is either superior to

locals or at least no worse than locals. From this

argument, S35 should be recommended everywhere; there is, in

FSR/E terms, a singe recommendation domain. We know,

however, that farmers are not adopting S35 everywhere. The

problem is twofold: how do we reconcile the differences

between the various groups of on-farm tests; and how do we

reconcile the performance of S35 in high- and low-yielding

environments with the pattern of S35 adoption since

extension of this variety began.











One step in resolving at least the first of these

dilemmas is to recognize that the various groups of tests

represent essentially different research domains. Tests in

1984 had extremely late dates of seeding, both in the

northern and the central regions (Table 2.1). This lateness

was not just due to the lateness of rains, but due to the

mechanics of the testing program that year; test inputs and

protocols were distributed much later in 1984 than in

succeeding years because of a change in the leadership of

the SAFGRAD program which necessitated retarding somewhat

the planning and design phases of testing. Thus the 1984

sample of sites represented not just a target population of

farms ("research domain") in a very low rainfall year, but a

population of farms seeded late in a very dry year.

It is more immediately obvious that there were two de

facto research domains in 1985, determined by the test

structure: farms seeded late in a moderately good rainfall

year, and farms seeded early in the same year. Similarly,

the research domain represented in 1986 and 1987 was a

population of farms seeded between 10 and 20 June in good

rainfall years.

It should also be noted that the sample of test farms

in 1984 differed from those of later years in terms of rate

of fertilizer applied. Since fertilization is a factor

expected to be in interaction with other factors that

determine E, such as rainfall, any definition of the











research domain, i.e., the population of farms for which

inferences can be drawn from a set of tests, must also

include fertilization. In the present study, differences in

environment caused by native fertility are confounded by

different fertilizer rates applied in the tests.

To look at a larger research domain, more nearly

approximating the population of all small farms in the

Center North for all years, all of the test sites from 1984

to 1987 were combined. This was, of course, only an

approximation; it did not include, for example, early-seeded

or low-fertility sites in a poor rainfall year nor very

early- or very late-seeded sites in a good rainfall year

(Tables 2.1 to 2.3). Whether exclusion of these kinds of

environments from the sample of tests is important depends

on the validity of an implied assumption of MSA regression:

that the treatment-by-environment interaction is independent

of the reasons for an environment being "high" or "low."

MSA regression for 239 on-farm test sites over four

years showed linear responses of S35 and locals that were

slightly convergent at the high end of the E continuum

(Figure 2.10). While S35 appeared superior across all

environments, this superiority was greater in low-yielding

environments and practically non-existent in high-yielding

ones. This was similar only to the response for the early-

seeded test in 1985 (Figure 2.6). Note that the

distribution of E for this set of all sites, like that of












the early-seeded 1985 tests (and unlike any of the other

subsets), had a wide range and a broad concentration of

values from under 1000 to well over 2500, weighted perhaps

on the low end. It was also not dominated by concentrations

of either very high or very low values.

The purely linear regression predicted yields of locals

to be less than zero at very low values of E. Since this is

clearly impossible, it is certain that no matter how good

the linear estimate of local yield response is at Es greater

than 300 or 400, at some point that linear relationship

between E and yield of locals must change. A hypothetical

illustration of this necessary modification is presented in

Figure 2.11. A change in the locals' response will of

course, by the nature of the calculation of E, entail a

symmetrical change in the shape of the S35 response. These

changes in the varietal responses to E may be either two

different linear responses, as illustrated, or some sort of

curved (quadratic or other) response. A "best" fit of the

predicted response functions) can be determined by

appropriate statistical techniques, but since modelling of

the response was unlikely to improve delineation of specific

adaptability to environments, it was not attempted.

The nonlinearity of the response of S35 and locals

across environments could be seen by plotting the 1984-87

yield data (Figure 2.12). It was clear that the data points

for S35 yield and for yield of locals, converging at zero E,











diverged as E increased to about 1000 or 1500. Yields of

locals close to zero at a number of sites up to E = 1000

contributed strongly to this impression of divergence. The

data for Es greater than 1500 seemed to show that yields of

the two varieties, separate at E = 1000 or so, converged as

Es became greater than 2000-2500.

Further verification of this idea of a "curved"

response of varieties to E was furnished by performing MSA

regression analysis on three different segments of the E

continuum from the 1984-87 data (Table 2.4). Regression of

S35 and local yields on Es less than 1000 produced responses

which converged sharply at the low end (intercepts nearly

equal, slope of S35 above unity and slope of locals well

below unity). Note the similarity of this result to the

regressions from the 1984 northern regions, data (Figure

2.2). This sort of response could be expected not only from

data with no E values above 1000, but also from data

strongly dominated by very low values. Examination of yield

data for all of the 1984 tests plotted on E showed that

yields of S35 and locals at E greater than 1500 were not so

sharply divergent (Figure 2.13). Being few in number,

however, they could not counteract the influence of many

low-end data points in determining the shape of the

regression response (Figure 2.5).

Regression on the range of Es between 500 and 1500

(Table 2.4) produced estimated'responses which, although











still convergent toward the low end, are much less so than

for data dominated by low values of E. These responses

resemble those for the 1984 central region data (Figure 2.4)

and the 1985 late-seeded data (Figure 2.7), dominated

neither by high nor by low values of E. Grouping 1984 and

1985 sites together produced a data set with neither high-

nor low-end dominance (Figure 2.14); estimated responses of

S35 and locals were almost parallel (Figure 2.15).

Returning to Table 2.4, regression on the range of Es

greater than 1000 produced response lines for S35 and locals

that were slightly convergent at the high end (slope of S35

less than unity, slope of locals greater than unity), with

less overall mean yield difference than for the subsets with

lower E values. This type of response corresponded well to

those from the 1985 early-seeded data, which were somewhat

more dominated by environments with E>1000 (Figure 2.6).

Similarly, and in line with this trend, regression on

E>1700, not shown in Table 2.4, produced responses with

practically equal slopes (S35 = 1.02, locals = 0.98) and

very little difference in overall mean yield (=175 kg ha').

This was quite similar to the results of MSA for the 1986

and 1987 data (Figures 2.8 and 2.9).

It will be assumed, then, that the linear response of

S35 and locals across all 239 sites from the combined 1984-

87 data set was an acceptably accurate picture of the

variety-by-environment interaction for these two types of











1984-87, S35 vs. Locals

Grain Yield (t/ha)


835
-- Locals
0 E


0 500 1000 1500 2000 2500 3000 3500 4000
Environmental Index (E)


Figure 2.10


Modified Stability Analysis, S35 and locals,
239 sites, 1984-87.












1984-87, S35 vs. Locals (Hypothesis)

Grain Yield (t/ha)


S35
- LOCALS


0 500 1000 1500 2000 2500 3000 3500 4000
Environmental Index (E)


Figure 2.11


Modified Stability Analysis, hypothetical
curved response of S35 and locals, 1984-87.











1984-87, S35 vs. Locals, Scatter


Grain Yield (t/ha)


* 835
x Locals


0 500 1000 1500 2000 2500 3000 3500 4000
Environmental Index (E)







Figure 2.12 S35 and locals plotted on environmental index,
239 sites, 1984-87.




















Table 2.4 Regressions of yields of S35 and locals on three
segments of the range of E, 1984-87.


E<1000 500
S35 Locals S35 Locals S35 Locals

Intercept 10 -10 144 -144 253 -253

b 1.31 0.68 1.05 0.95 0.93 1.06

Mean 769 389 1104 722 1992 1733












1984, S35 vs. Locals, Scatter


Grain Yield (t/ha)


* S35
x Locals


0 500 1000 1500 2000
Environmental


Figure 2.13


2500
Index


3000 3500 4000
(E)


S35 and locals plotted on environmental index,
87 sites, 1984.


x
x
*. "X x

"X" y X
.* X^^ (X XA












1984-85, S35 vs. Locals, Scatter


Grain Yield (t/ha)


* S35
x Locals


0 500 1000 1500 2000 2500 3000 3500 4000
Environmental Index (E)








Figure 2.14 S35 and locals plotted on environmental index,
146 sites, 1984-85.












1984-85, S35 vs. Locals


Grain Yield (t/ha)


S35
-I- Locals


0 500 1000 1500 2000 2500 3000 3500
Environmental Index (E)


Figure 2.15


Modified Stability Analysis, S35 and locals,
146 sites, 1984-85.











sorghum in the Center North zone. Differences in the

individual year test data could be explained by considering

these data merely subsets of the larger research domain,

each with its own characteristic distribution of E. Those

with distributions of E most similar to that of the overall

dataset were more representative samples of the whole

research domain; the others were representative of

incomplete segments of the research domain.

How could this information be used to predict specific

adaptability of S35 and identify recommendation domains for

it? Further, how could it be used to help judge the

usefulness MSA might have had in that identification process

early in the on-farm testing of S35? The key seemed to be

in appropriate classification of environments. Ranking

sites by E, then looking for patterns of correspondence of

non-numeric site characteristics such as soil texture,

geographical location, previous crop, etc. with E proved

unsuccessful. The two numerical variables for which data

existed over all four test years were rainfall and date of

seeding. Simple and multiple linear regression of E on

these variables was performed. The simple regressions, for

all four years, for 1984 only, for 1985 only, and for 1984

and 1985 together, are presented in Table 2.5.

For the 1984-87 data taken together, there was a

significant negative linear relationship between E and date

of seeding; i.e., the later the date of seeding, the lower











the site mean. There was also a significant positive

relationship between rainfall in the first 90 days after

seeding and E, although this parameter accounted for much

less of the variability in E than did date of seeding--7%

versus 17% (Table 2.5).

It was hypothesized from the shape of the responses of

S35 and locals across all years (Figure 2.10) that S35 might

be more adapted, in terms of grain yield at least, to low-

yielding environments, since the difference between its

yields and those of locals is greatest in that domain. Even

if the two response lines were parallel, rather than

slightly divergent at the low end, as indicated by MSA, a

given difference in low-yielding environments would be much

more important to farmers (as a large percentage of total

yield) than would be an equal difference in high-yielding

environments.

Given the relationship between E and date of seeding,

it was further hypothesized that S35 might be more

acceptable to farmers as a substitute for their locals in

those areas where late seeding is common. Based on the same

1984-87 varietal test data, SODECOTON sectors were ranked by

date of seeding of those varietal tests (Table 2.6). The

sectors in which S35 is known ex post facto to be more

readily adopted were, in general, among the later-seeding

sectors in this ranking. Further, the five latest-seeding












sectors are also low-rainfall sectors on the northernmost

edge of the Center North zone, as is the sector Moulvoudaye.

The fact that there was less than perfect

correspondence (i.e., that the very latest-seeded sectors

were not those with the highest adoption of S35) was due no

doubt both to the many factors other than date of seeding

that determine E and to the many factors other than grain

yield that influence adoption of a sorghum variety.

Hina sector, for example, is not a low rainfall sector; in

addition, the Hina possess an unusually small number of

local varieties to which they are highly attached and resist

S35 strongly for its taste and cooking qualities. Koza and

Kourgui sectors, on the other hand, while not ranking among

the latest-seeding sectors from these test data have human

populations which are more accepting of the taste of S35.

In addition, proximity to both an IRA research site and a

seed multiplication project, may have increased farmer

contact and familiarity with the new variety, and a relative

lack of bird pressure may also have served to increase

acceptance.

Could the same sort of recommendation domain for S35,

i.e., sectors--particularly low-rainfall ones--where farmers

tend to seed late, have been identified by Modified

Stability Analysis after the 1984 test season? Although the

significant negative linear relationship between date of

seeding and E holds true for the 1984 data alone (Table











2.5b), it accounts for only 3% of variability in E, due no

doubt to the fact that all tests sites were seeded unusually

late that year. More importantly, the relative responses of

S35 and locals would probably have led one to think that S35

should be recommended either everywhere or in high-yielding

sites. This impression would have been confounded with the

low-yielding nature of all sites in 1984, making 1984 test

sites an unrepresentative sample of all years, as discussed

above. Similar confounding would result from the high

fertility level of 1984 tests. It is possible, for example,

that the higher-yielding, early-seeded (and presumably

higher rainfall) sites in 1984 favored S35 because S35 was

under these conditions better able to respond to high

fertilizer levels than locals--levels that did not exist in

later tests. In any event, it is unlikely that a

relationship between late-seeded sites (or sectors) and

increased superiority of S35 over locals would have been

recognized.

Had the on-farm testing of S35 begun in 1985, MSA would

have indicated the increased yield superiority of S35 over

locals in low-yielding environments (Figure 2.16), and

regression of E on date of seeding would have shown the

negative relationship of date of seeding and E (Table 2.5c).

If MSA had been performed after the 1985 season on the

1984 and 1985 data combined, the yield superiority of S35 in

low-yielding environments--in terms of relative, if not











absolute yield differences--might very well have been

recognized (Figure 2.15). Regression of date of seeding on

E for the combined 1984-85 dataset would also have

identified a negative relationship as significant and

accounting for as much of E as for the entire 4-year dataset

(Table 2.5d). Ranking SODECOTON sectors by mean date of

seeding of 1984-85 tests resulted in a recognizable

correspondence of late seeding with S35 adoption, but this

correspondence was much less clear than for the 1984-87 data

taken together (Table 2.7). This is probably because so

many of the 1984 test sites were seeded late for test-

related reasons rather than because of farmers' decisions

based on environmental conditions; the 1984 tests

comprise more than half of the 1984-85 data, but only a

third of the data for all four years.

From the above, it was evident that the 1985 tests

alone allowed better identification of the recommendation

domain of late-seeded sectors than did the 1984 tests alone

because the set of 1985 tests more closely approximated the

larger research domain in terms of range and distribution of

environmental indices and in terms of range of seeding

dates. The combined 1984-85 data were even more

representative and would have been more likely to produce

recommendation domains similar to those from the 1984-87

data than would have been either the 1984 or 1985 data

alone. It should be noted here that the 1985 data were









57

uncharacteristic in terms of the relationship between E and

rainfall, which was actually negative in 1985. The reasons

for this anomaly were unclear, but the hypothesized

recommendation domain for S35 drawn from 1984-87 data--i.e.,

late-seeded, low-rainfall sectors--could not have been

deduced from either the 1985 data alone nor the combined

1984-85 data (Table 2.5c,d).

It is worth noting that had the on-farm testing of S35

begun in 1986 rather than in 1984, both the data from 1986

alone and those from 1986 and 1987 combined would most

probably have led to the conclusion that S35 was not

superior to local varieties. This would certainly have been

the case had analysis relied on ANOVA alone (Table 2.3).

Even MSA would not have indicated a superiority after the

1986 season (Figure 2.8). Modified Stability Analysis might

have led to recognition of yield advantages of S35 in poor

environments in 1987 (Figure 2.9), but these advantages are

considerably less striking than those exhibited in 1984.

There are two ways of going about establishing

recommendation domains when interpreting the superiority of

S35 to locals in late-seeded sites. The first is to

consider the recommendation domain as "farmers who seed

late" or even "late-seeded fields of sorghum". The problem

with this approach is that it is impossible in any given

year to identify those farmers or those fields ex ante. The

most one can do is sell S35 seed to a farmer and say "this









58

is for planting late." Farmers in the semi-arid regions of

Africa, however, are used to planting sorghum with the first

good rains, knowing that it is possible that the next rains

may not come until too late to plant, or at least so late

that sorghum planting will interfere with planting other,

higher value crops. Some no doubt will wait to plant their

S35 in the recommended range of planting dates; many, moved

by habit and adherence to a proven traditional survival

strategy, will not.

Farmers who do not wait to seed will predictably have

yields of S35 not much better, perhaps worse than their

local yields. If farmers who tend to seed early comprise

the vast majority of those planting S35 in a sector, it is

unlikely that adoption will take off in that sector. If

farmers generally seed late, adoption is more likely. This

is especially true if S35 is part of a package that requires

behavioral modifications other than retarding seeding date.

The question then is one of efficiency of extension

efforts. To achieve efficiency, current farmer practices

must be taken into account. Should one extend S35 and

concurrently try to change farmer practices in order to

maximize S35's inherent advantages, or should one extend S35

where farmers' current practices are most likely to permit

S35's advantages to manifest themselves? If extension has

the resources to educate farmers over the long term and a











long-term perspective not easily given to discouragement,

perhaps the first option is a viable one.

The first option of furthering S35 adoption by

educating farmers to take advantage of its ability to be

planted late is an ideal one. If S35 could be extended in a

recommendation domain comprised of farmers who seed late,

its advantages in terms of reduced risk would be maximized.

This was illustrated by confidence limits of S35 and locals

in the domain of 1984-87 test sites seeded before 25 June

versus S35 and locals in the domain of sites seeded after 25

June (Figure 2.17). Only the lower confidence limits are

illustrated, calculated as the mean of the variety within a

given domain minus the product of the standard deviation of

the mean and the t value corresponding to a given a level

and the number of degrees of freedom for the mean. The risk

coefficient is the chance, expressed as a percentage, that

the true value of the mean is less than the lower confidence

limit. The reduced risk of very low yields from seeding S35

rather than locals in the late-seeded sites was clearly

greater than the reduction in risk in the early-seeded

sites. The problem from a practical extension point of view

is that in some sectors there will be many farmers

benefitting from this greater reduction in risk and in other

sectors there will be very few farmers; both sets of

sectors, however, will be receiving the same level of

extension efforts.









60

The second extension option, that of targeting for S35

extension only those sectors where farmers generally seed

late is far less ideal, in terms of maximizing S35's

potential for reducing low-end risk, but is still a viable

one. Confidence limits for S35 and locals were graphed for

the domain of generally early-seeding sectors (where average

date of seeding was before 25 June) and for the domain of

generally late-seeding sectors (Figure 2.18). Since each

domain comprised both early-seeded and late-seeded sites,

the greater reduction in risk of late-seeded S35 versus

late-seeded locals was less evident, being in effect

somewhat "diluted." The reduction in risk was nevertheless

illustrated. Using the accepted, if not statistically pure,

interpretation of this graph, locals in the late-seeded

sectors had almost a 25% chance of yielding less than 1200

kg ha'1, while S35 in the same sectors had less than a 1%

chance. Early-seeded sectors also had a great reduction in

risk from S35 (primarily because they too comprised some

late-seeded sites), but the importance of this reduction in

risk is less than in late-seeded sites since overall mean

yields are higher. More to the point, in terms of

efficiency of extension efforts, the number of farmers who

benefit from reduction in risk is greater in the late-seeded

sectors than in the early-seeded ones.

The results of this study, while having particular

reference to northern Cameroon, could be applicable to









61

efforts to develop and extend improved sorghum varieties in

other parts of semi-arid West Africa, especially since so

many of those efforts are concentrating on short-cycle,

relatively photoperiod-insensitive varieties. In a broader

sense, some aspects might be applicable to all on-farm

research in an FSR/E context. Modified Stability Analysis

is not a technique that will magically and simply identify

superior technologies and define appropriate recommendation

domains by a series of rote operations: regression on site

means; characterization of environments; delineation of

domains according to high- and low-yielding environments,

and graphing of confidence limits. MSA, in short, is not a

substitute for intimate knowledge of the farming system (the

research domain) for which one wants to draw inferences from

the on-farm tests, but can be useful in complementing that

knowledge.

In the present study, knowledge of the sector-to-sector

differences in average date of seeding and average rainfall

were as important as knowledge of how well S35 and locals

tolerate late seeding and poor rainfall. Even so, those two

parameters accounted for only 25% of the environmental

index, and would not have been precise in predicting which

sectors would adopt S35. Knowledge of sector-to-sector

differences in soil type, soil fertility, and bird and

striga pressure might also have increased precision in

predicting where S35 would be adopted, as would knowledge of












sector-to-sector differences in taste or storage

preferences, degree of animal traction, etc.

Knowledge of the research domain is also necessary in

deciding if the set of test environments is an adequately

representative sample of all sites in the domain over all

years. FSR/E researchers often criticize on-station

researchers for drawing conclusions aimed at small-scale,

limited-resource farmers from trials done on research

stations that are very different from farmers' fields. In

northern Cameroon, for example, on-station varietal

performance trials commonly have grand mean yields of 4000-

6000 kg ha1. These trials are very good for identifying the

yield potential of varieties, but have little immediate

relevance to farmers; thus, the need for on-farm tests. In

the present study, the representativeness of the 1986 data

alone or the 1987 data alone was doubtful; it seemed obvious

on the face of it, that data with mean local yields of

approximately 2000 kg ha'' were probably of little

inferential value for a research domain of farmers with mean

local yields of 900 kg ha-. On the other hand, the data for

1984 alone was unrepresentative for different reasons.

It is not that the 1984, 1986, and 1987 data were not

valuable, just that in themselves they gave an inaccurate,

i.e., incomplete, view of the broader situation. Perhaps it

is true that a large number of on-farm test sites is most

often a better sample than a small number, but only because









63

it is more likely to be a more representative sample of the

larger research domain. Such does not have to be true,

however, if one knows the range, distribution, and make-up

of on-farm environments that characterize the research

domain. The 239 environments from all four years was

undoubtedly a better sample than the 88 environments in

1984; but these 88 alone, as well as the 73 environments in

1986-87, were probably a less representative sample than the

58 environments in 1985.

The simple linear regression component of MSA, as

usually employed, would have certainly indicated to

researchers a superiority of 535 corresponding to the later-

known pattern of adoption in only one year taken alone,

i.e., 1985. It might possibly have done so in 1987, but

this is far less certain. Only by using some multiple

regression of yields on E would the 1984 data have shown

such a pattern, and even this possibility is arguable. MSA

as a tool allowing substitutions of many test locations in a

single year to substitute for two or more years of testing

depends, it would seem from this study, on the single year's

data (environments) being quite representative of the entire

population of environments, over both space and time, in the

research domain.

It is true that in the present study, the advantage of

hindsight was ever-present. Nevertheless, the potential

advantages of MSA shown herein, seemed plausible, although









64

highly dependent upon judicial selection of an appropriate

test sample which might not always be easy to identify ex

ante, and upon a highly refined familiarity with the target

research domain and with the goals and capabilities of the

existing extension services.












Table 2.5 Regressions of E (S35 and locals) on date of
seeding (DOS) and on rainfall*, 1984-87, 1984, 1985, and
1984-85.

a) 1984-871

DOS Rainfall

Intercept 5979 864
b -25.4 1.3
R2 0.17 0.07
p 0.0001 0.0003

b) 1984

DOS Rainfall

Intercept 3140 645
b -11.4 1.04
R2 0.03 0.03
p 0.022 0.088

c) 1985

DOS Rainfall

Intercept 5470 2483
b -22.6 -1.6
R2 0.13 0.09
p 0.006 0.02

d) 1984-85

DOS Rainfall

Intercept 5209 1034
b -22.0 0.63
R2 0.17 0.01
p 0.0001 0.126



t Date of seeding = days after 1 January.

$ Rainfall = mm, 1-90 days after seeding.

1987 excluded from regression of E on rainfall since
rainfall data aggregated over whole year.















Table 2.6 Correspondence of date of seeding and sectors
where S35 was adopted, 1984-87 data.

N Sector Mean DOS
(Days after 1 Jan.).


10 Bidzar 168.2
10 Lara 169.7
7 Ardaf 169.3
10 Dziguilao 169.8
10 Guidiguis 170.3
11 Karhay I 170.4
13 Gobo 170.4
11 Moutouroua 171.9
14 Mokong 172.0
11 Sorawel 172.9
10 Dana 173.4
9 Mayo Oulo 174.8
14 Zongoya 176.9
10 Karhay II 177.1
8 Kourgui 178.4 **
7 Koza 178.7 **
9 Moulvoudaye 179.1 *
12 Hina 179.7
10 Yoldeo 180.0 *
10 Meme 181.8 **
10 Mindif 183.1 *
8 Dogba 183.1 *
13 Bogo 184.0 *


** Sectors with very good S35 adoption
* Sectors with good S35 adoption












1985, S35 vs. Locals


Grain Yield (t/ha)


S35: b-0.98
Locals: b-1.02








0 @QOO030

0 500 1000 1500 2000 2500 3000 3500
Environmental Index (E)


S35
-+- Locals
0 E


Figure 2.16 Modified Stability Analysis, S35 and locals,
58 sites, 1985.















Table 2.7 Correspondence of date of seeding and sectors
where S35 was adopted, 1984-85 data.

N Sector Mean DOS
(Days after 1 Jan.)


5 Ardaf 168.6
6 Lara 170.6
7 Bidzar 171.6
7 Karhay I 171.6
6 Dziguilao 171.8
10 Mokong 171.9
9 Gobo 172.3
7 Dana 173.7
6 Guidiguis 174.0
8 Moutouroua 175.1
6 Kourgui 175.2 **
5 Mayo Oulo 176.4
6 Koza 176.5 **
7 Sorawel 177.3
10 Hina 177.3
7 Meme 180.4 **
10 Zongoya 180.7
7 Karhay II 180.7
6 Moulvoudaye 181.2 *
8 Mindif 182.1 *
7 Yoldeo 182.3 *
9 Bogo 186.8 *
5 Dogba 191.4 *


** Sectors with very good S35 adoption
* Sectors with good S35 adoption










Early vs. Late Dates of Seeding,
S35 and Locals, 1984-87


Risk Coefficient (%)


0-
400


800 1200 1600


Grain Yield (kg/ha)

-L-- Loc. late --'-- S35 late Loc. early S35 early


Figure 2.17


Lower confidence limits for S35 and locals, by
early- and late-seeded sites.


2000










Early vs. Late Seeded Sectors,
S35 and Locals, 1984-87

Risk Coefficient (%)


0 "--
750 1000 1250 1500 1750
Grain Yield (kg/ha)


-4-- Loc. late


SLoc. early


---- S35 late


- S35 early


Figure 2.18 Lower confidence limits for S35 and locals, by
early- and late-seeded sectors.


2000















CHAPTER III

EFFECTS OF FERTILIZATION AND PLANTING PATTERN
ON GRAIN YIELD AND YIELD STABILITY OF MIXED STANDS
OF LONG AND SHORT CYCLE SORGHUM VARIETIES



Introduction



The use of mixed stands, either of cereal species or of

cultivars of a single species, in the improvement of cereal

crop yields has been the subject of much conjecture and

research. It is hypothesized that improvements--increased

productivity (i.e., higher yields) and increased stability

of production across diverse environments--would result from

the genetic diversity of the different components of the

mixed stands. From a theoretical viewpoint, yield

advantages might be induced by complementarity, where the

component varieties or species exploit environmental

resources in different ways, or by compensation, where

growth and yield formation of individual
components are adversely affected by factors
other than competition, with the unaffected
components being able to benefit from the
increased availability of resources. (Panse
et al., 1989, p. 347)



Research on advantages of mixed stands of cereals, while

often promising, has generally been less conclusive than
71















CHAPTER III

EFFECTS OF FERTILIZATION AND PLANTING PATTERN
ON GRAIN YIELD AND YIELD STABILITY OF MIXED STANDS
OF LONG AND SHORT CYCLE SORGHUM VARIETIES



Introduction



The use of mixed stands, either of cereal species or of

cultivars of a single species, in the improvement of cereal

crop yields has been the subject of much conjecture and

research. It is hypothesized that improvements--increased

productivity (i.e., higher yields) and increased stability

of production across diverse environments--would result from

the genetic diversity of the different components of the

mixed stands. From a theoretical viewpoint, yield

advantages might be induced by complementarity, where the

component varieties or species exploit environmental

resources in different ways, or by compensation, where

growth and yield formation of individual
components are adversely affected by factors
other than competition, with the unaffected
components being able to benefit from the
increased availability of resources. (Panse
et al., 1989, p. 347)



Research on advantages of mixed stands of cereals, while

often promising, has generally been less conclusive than
71












that on intercropping of cereals with legumes or with

longer-cycle root or tuber crops.

The term "multiple cropping" has usually been reserved

to mean growing two or more different crop species in the

same field during a year, and "intercropping" has tended to

be used to indicate that the crops are grown simultaneously,

rather than successively, as is the case in "sequential

cropping" (Francis, 1986).

Intercrops are generally thought of as interspecific

multiple cropping. While there is no real reason for thus

limiting the term, rather than add to the confusion of the

already overburdened and contradictory terminology found in

the multiple cropping literature (in which, for example,

"mixed intercrop" has come to mean an intercrop with no

distinct row arrangement), the present work will simply

posit the existence of a number of forms of intraspecific

"mixed stands". Whether or not these are "intercrops", each

reader may decide.

The most common of these mixed stands are various types

of mixtures, either multiline cultivars, bulk populations of

hybrids or varieties, or mechanical mixtures or blends of

distinctly different cultivars. In addition to these types

of mixtures, intraspecific mixed stands have been obtained

by alternating rows or pairs of rows of different cultivars;

these have been referred to simply as "mixed stands" (Prasad

and Sharma, 1980), or as "mixed row cropping" (Reddy et al.,












1991). For the purposes of this work--and in order, again,

to avoid confusion with multiple cropping terms already

accepted by researchers--two forms of intraspecific mixed

stands will be distinguished: 1) seed mixtures, and

2) mixed-row stands.

StUtzel and Vanderlip (1988) found no yield

advantages, as measured by the Land Equivalency Ratio (LER),

from intercropping three sorghum hybrids with 90- to 110-day

cycles and a pearl millet hybrid with a growth cycle similar

to that of the early sorghum. Only under good moisture

conditions, and using the latest-maturing sorghum, was there

even a tendency toward yield increases. The importance of a

large temporal difference in the component species or

cultivars recurs elsewhere in the literature. Baker (1979)

determined that for cereal mixtures in northern Nigeria,

yield advantages occurred only when components had growth

cycle differences greater than 43 days; for mixtures of

sorghum varieties, maturity differences of more than 51

days, plus height differences of 0.6 m, were required for

yield increases. Similar yield advantages in northern

Nigeria from intercropping very long-cycle (180-day) sorghum

with shorter-cycle millet were shown by Norman (1984).

Studies in India by Willey and Rao (1983), however, could

not account for more than 22% of the variability in LERs

from sorghum/millet intercrops by differences in height or

growth cycle, although growth cycle appeared to have more










74

influence than height. Yield advantages from maize/sorghum

intercropping have been identified in central America by

Hawkins (1984), and in northern Cameroon, where it is a

theme actively extended by the local cotton and rural

development agency.

In mixed stands of wheat (Triticum aestivum L.) and

barley (Hordeum vulgare L.), Prasad et al. (1988) showed

that 1:1 and 2:1 wheat:barley stands produced higher grain

yields than either the mean of the component yields or the

yield of the higher-yielding component (wheat). A 1:1 seed

mixture of the two components yielded less than the 1:1

alternate-row stand. Noworolnik et al. (1984), however,

found little yield advantage from mixtures of oats (Avena

sativa L.) and barley.

An appreciable amount of research has been done on

intraspecific cultivar mixtures. In the case of legumes,

Panse et al. (1989) identified compensation-induced yield

stability advantages in mixtures of common bean (Phaseolus

vulgaris L.). Allard (1961) studied the relationship

between level of genetic diversity of lima beans (Phaseolus

lunatus L.) and yield stability by growing three pure-line

varieties, four annually reconstituted mechanical mixtures

of these varieties, and three populations from successive

bulk propagation, from F2 to F7 or F9, of hybrids made from

the three pure-line varieties, in four distinctly different

locations over four successive years. In terms of









75

productivity, bulks yielded more than pure lines, which were

superior to mixtures. Using consistency of rank order and

relative magnitude of variances as criteria for stability,

mixtures and bulks were more stable than the pure lines, but

bulks were not appreciably more stable than mixtures.

In a study of stability of two multiline peanut

(Arachis hypogaea L.) varieties compared to their component

lines, Schilling et al. (1983) identified component pure

lines of peanut which were as stable across environments as

the multilines derived from them. Although a regression

coefficient of unity (in regression on E) was one of the

stability parameters used, the same conclusion would be

true, in this particular case, using a regression

coefficient of zero. It was noted in this study that

component pure lines of peanut multilines are more

genetically similar to one another than are those of the

multilines of other crops.

Knauft and Gorbet (1991) found that mixtures of five

peanut lines experienced genetic shifts over time, with the

highest-yielding of the five lines increasing in proportion

after only two years. Two other high-yielding lines

maintained or decreased their original proportion, while the

two lowest-yielding lines also decreased in proportion over

time.

In the case of cereal cultivar mixtures, Frey and

Maldonado (1967) tested six oat cultivars and 57 seed












mixtures of two, three, four, or five cultivars, at both

early and late planting dates, over three years. Averaged

over both planting dates, nine of the mixtures yielded

higher than the best cultivar, but the advantage never

exceeded 2.5%. Compared to the mean of the component

yields, the advantage of mixtures was greater in highly

stressed, i.e., late-seeded environments. Production of

mixtures was also more stable across planting dates than was

production of cultivars.

Funk and Anderson (1963) found that two-hybrid mixtures

of double-cross maize (Zea mays L.) hybrids did not yield

more than the mean of the component hybrids. Stability of

the mixtures, however, was greater than that of the

components, as measured by magnitude of the hybrid-by-

location interaction mean square from a combined analysis of

variance. Stitzel and Aufhammer (1990) identified yield

increases in barley cultivar mixtures, which they attributed

to compensatory mixing effects; the compensation could not

be attributed to reduced disease incidence.

Several studies have been done on the relation of

genetic diversity of sorghum, represented by cultivar

mixtures, and improved stability and productivity. Ross

(1964) compared five single-cross hybrids with the ten

possible 1:1 two-hybrid blends. In five years at one

location, blends outyielded the mean of their components in

only one (high-yield) year. None of the blends outyielded










77

its component mean (its "mid-component") over all years, and

only two yielded better than the mid-component within a

year, in both cases during the highest-yielding year.

When eight parental lines of sorghum and 16 F, hybrids

were compared to 16 two-component blends of parental lines

and 16 two-component hybrid blends, in 9 environments over

two years, average yield of hybrid blends was greater than

that of hybrids, which in turn outyielded the parental lines

and blends of parental lines (Reich and Atkins, 1970).

There was little difference between the average coefficients

for regression on environmental index of these four groups;

average deviations from regression for blends were smaller

than the average deviations of components.

Marshall and Allard (1974) determined that the

intergenotypic interactions in mixtures of sorghum genotypes

tended to be additive in nature. While this would

theoretically mean that expected yields of mixtures would

increase with the number of components, the higher-order

components of the intergenotypic interaction effects were

often greater than lower-order components, and opposite in

sign, thus having a canceling effect in mixtures with three

or more components. This conclusion and the results of

their study of two-, four-, five-, six-, twelve-, 20-, 50-,

and 80-component mixtures of randomly chosen inbred lines

supported the often-reported conclusion that "with respect

to yield, ...mixtures often, but by no means invariably,











outperform their average component. However, they seldom

outperform their better or best component" (p. 151).

Mercer-Quarshie (1979) compared five sorghum cultivars

to two-, three-, four-, and five-component mixtures of these

cultivars in eleven environments in northern Ghana. The

results were difficult to relate to the present work, since

the data themselves are not presented, and the author

followed the lead of Eberhart and Russell (1966) in defining

a stable variety as one with regression coefficient of

unity. This choice of criteria was remarkable in that this

study concludes that mixtures were more "stable," i.e., had

regression coefficients closer to unity than the component

cultivars, and that this stability explained why farmers in

this region of highly variable rainfall choose to grow

mixtures. Interestingly, average mean square deviations

from regression in this study were, in general, inversely

related to complexity of the cultivar mixture.

Much less research has been reported on mixed-row

stands of cereal cultivars than on cultivar mixtures.

Prasad and Sharma (1980) found that with a tall, a semi-

dwarf, and a dwarf cultivar of spring wheat, both a row

arrangement resulting in a "pyramidal" canopy (2:1:2:1:2

rows of tall:medium:short:medium:tall) and one resulting in

a "columnar" canopy (2:2:2 rows of tall:short:medium)

produced higher overall straw yields than the "flat"

canopies of pure stands. Grain yields of the columnar









79

mixed-row stand were significantly greater than those of the

pure stand of the best component under high N rates.

Mixed-row stands of either of two semi-tall, early-

maturing rice (Oryza sativa L.) varieties plus a semi-dwarf,

long-cycle modern variety produced grain yield increases of

19% over the best component variety under intermediate

deepwater conditions in India (Reddy et al., 1991). Yields

of the mixed-row stands were also more stable across years.

Increased stability and productivity were attributed in part

to decreased lodging of the early varieties. Neither row

arrangements of 2:1 or 1:2, early:late, nor decreased inter-

row spacing were any more productive or stable than a 1:1

row arrangement with 0.20 m between rows.

Mixtures of local sorghum varieties are common in

northern Cameroon. Unfortunately, the rationale and

strategy behind this practice have not been the objects of

much study and are consequently poorly understood. A

mixture common to the North Province is mixing long-cycle

(150+-day) varieties such as yolobri with shorter-cycle

varieties such as djigari. This practice is less common in

the Extreme-North Province; at an earlier time, when longer-

cycle varieties were more widely grown in this province,

such mixtures were perhaps more often encountered.

After two years of extension of the improved, short-

cycle variety S35, it was discovered through field visits

and farmer interviews that a number of farmers were











incorporating S35 into their traditional sorghum-based

cropping systems by mixing it with their local varieties.

Two types of mixtures were identified: a "catch-up" mixture

(association de ratrappage), in which the local variety was

seeded with the first rains, and in the event of poor stand

establishment, missing hills ("pockets") were later seeded

to S35; and a "hungry period" mixture (association de

soudure), in which seed of S35 and the usually longer-cycle

local variety or varieties were mixed and planted together.

In light of this interest in mixing S35 with local

varieties, initiated by farmers themselves, a series of

trials was initiated. At least some evidence existed that

such mixtures might be, if not necessarily more productive,

at least more stable across highly variable environmental

conditions. The objective of these studies was to verify

possible yield advantages from mixed stands of long- and

short-cycle sorghum varieties, and to identify the most

productive and/or most stable of these forms of multiple

cropping under conditions representative of those facing

limited-resource farmers. Since stability of production

under farmer conditions was perhaps the most likely benefit

of mixed stands, on-farm tests were begun concurrently with

more complex on-station trials. It was thought that this

short-cutting of the usual process of preceding on-farm

tests with one or more on-station trials posed little

potential risk to farmers since variety mixtures were










81

already known to and practiced by farmers. In addition, S35

was a variety already being extended to them.

Although mixed-row stands had not been observed in

farmers' fields, the relative ease of some cultural

operations (particularly harvesting) to be offered by this

planting arrangement, as opposed to seed mixtures, led to

the inclusion of this type of mixed stand in the on-station

trial reported here. The possibility, indicated in the

literature, that the relative advantage of mixed over pure

stands might manifest itself differently under high and low

soil fertility conditions led to the inclusion of level of

applied fertilizer as a factor in this same trial.


Materials and Methods



On-Station Trial

A field experiment was conducted at three sites in the

Center North zone of northern Cameroon in 1989 and 1990.

The first site was Mouda research station, in the southern

portion of the Diamare plain, at approximately 10020'N,

14010E; the second was Guetale research station, in the

southern Koza plain, at approximately 10050'N, 13055'E; and

the third was Tchatibali research station, in the southern

Kalfou plain, at approximately 10010'N, 15005'E. The soils

at Mouda are classified in the French system as shallow,

sedimentary soils of volcanic origin; texturally, they are












loams or sandy loams. The soils of Guetale are alluvial

soils, slightly weathered (peu 6voluB), and are loams or

sandy loams. The soils of Tchatibali are sands or loamy

sands. The chemical and physical soil characteristics of

these sites are presented in Table 3.1.

At each location, the experiment was a split plot with

the main plots arranged in a randomized complete block

design, with three replications. The main plot factor was

fertilization, at two levels: no fertilizer (FO), and 40-9-

17 kg ha'1 N-P-K (Fl), applied at time of planting.

The subplot factor was crop combination and planting

pattern, with nine treatments completely randomized within

each whole plot. The three varieties of sorghum included in

the experiment were damougari (D), an 85- to 90-day, red-

seeded, short (2 m), local type; walaganari (W), a 120-day,

tall (3 to 3.5 m), white-seeded, local type; and CS-54 (C),

an 85-day, 2.5 m, white-seeded, relatively photoperiod

insensitive, improved variety. CS-54 was chosen as the

improved variety rather than S35, since the two are very

similar, CS-54 differing only by maturing 5 days sooner than

S35. The variety combinations and planting patterns of the

nine subplot treatments are presented in Table 3.2.

Each location was plowed before planting and seeded by

the resident research station chief in response to the

arrival of reliable rains. Plot size was 10 x 4.8 m, with

six rows per plot and 0.80 m between rows. The two-variety,















Table 3.1 Soil physical and chemical properties, on-station mixed stands trial, 1989-90


Site O.C. T.N. Avail. P Exchangeable cations (me/100q) ECEC pH Mechanical Comp.
Year (%) (%) (Bray-1) Ca Mg Mn K Na Exch. (me 1Og-') water (%)
(ppm) Acid. Sand Silt Clay


Guetale

1989 0.34 0.033 32.4 3.56 1.00 0.04 0.32 0.32 0.31 5.55 5.1 57 30 13

1990 0.56 0.041 39.2 6.61 2.47 0.09 0.34 0.23 0.01 9.75 6.4 -


Mouda

1989 0.76 0.071 2.1 3.83 1.31 0.08 0.35 0.39 0.01 5.97 5.2 55 32 13

1990 1.20 0.082 1.8 7.03 2.56 0.13 0.35 0.15 0.13 10.35 6.2 -


Tchatibali

1989 0.35 0.028 7.6 0.76 0.35 0.00 0.25 0.19 0.00 1.55 5.7 79 14 7

1990 0.40 0.025 4.6 3.44 0.45 0.07 0.21 0.07 0.00 4.24 7.0 -


t O.C. = organic carbon
T.N. = total nitrogen
Avail. P = available phosphorus, using the
Exch. Acid. = exchangeable acidity


Bray-1 procedure





















Table 3.2. Subplot treatments for
mixed stands of sorghum varieties.


on-station trial of


-Pure Crops


Alternate-Row
Mixed Stands


Seed Mixtures


- Damougari (D)

- Walaganari (W)

- CS-54 (C)


- D+W

- C+W

- D+C+W



- D+W

- C+W

- D+W+C




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs