African Crop Science Journal, Vol. 8. No. 3, pp. ...
African Crop Science Journal, Vol. 8. No. 3, pp. 243~249,200O
ISSN 1021-9730/2000 $4.00
Printed in Uganda. Al1 rights reserved
02000, African Crop Science Society
AN APPRAISAL OF IRRIGATED TElhPERATE AND TROPICAL MILLET
VARIETIES IN THE SEM1AkI.D REGION OF SENEGAL
SALIOU DIANGAR, TANGU BA and C. F. YAMOAH’
Institut Senegalais de Recherches Agricoles, ISRA CNRA, Bambey, BP 53 Senegal
‘Soi1 Institute, kumasi, Ghana
(Received 6 November, 19p8; accepted 2 May, 2000)
ABSTitACT
Pearl millet (Pennisetum glaucum (L). R. Br.) constitutes a major food trop in the semiarid region of West Africa
but yields are extremely low in subsistence cropping sysfems because of inappropriate management and scarcity
of water. This study was designed to see if Pearl millet could become a component of trop rotation in an irrigated
scheme, and additionally, at a particular season (hot dry season - February to May when evapotranspiration levels
are too high for other crops) when water supplies are low. The effects of fertilisation, plant density and land
preparation on yields of improved varieties were tested under irrigation. Land preparation did not affect mean
yields of the two very early dwarfmillet genotypes, hybrid 68 A x MLS from Nebraska (4025 kg ha-‘) and a local
synthetic, variety GAM 8201 (4018 kg ha’). However, $ield of taller, but early, local synthetic variety increased
by 60% when planted on flat as opposed to ridges. Effect of fertiliser on yield of hybrid GB 87-35 was signifïcant.
In addition, method of land preparation and plant den&y increased yields of 68 A x MLS but not GB87-35.
Application of 10 N, 9 P and 17 kg K ha-’ and 45 kg N ha-’ as urea gave the highest grain yield of variety 68 A
x MLS. Neither frequency nor amount of irrigated wa@ significantly (PcO.05) affected yield of pearl millet
varlety 68 A x MLS, but we noted a trend (PcO.12) where plots irrigated once per week with 75% maximum,
seasonal trop water requirement of millet (526 mm) produced higher yields. Millet yields could be sustained in
semiarid environments by ensuring a minimum and reliable water supply (526 mm during the dry season),
optimum plant population (60 cm x 20 or 30 cm) as well as moderate fertilisation (10 N, 9 P and 17 K kg ha1 +
45 kg N ha-r as urea), and suitable land preparation method, depending on variety.
Key Words: Agro-ecological zone, fertiliser use, irrigation, Pearl millet, semi-arid
RE@MI?
Le mil est la principale culture vivrière dans les régions semi-arides de l’Afrique Occidentale. Les faibles
rendements enregistrés dans les systèmes de culture de subsistance sont le fait d’une gestion inadéquate et de la
rarett5 de l’eau. Cette étude a pour but de déterminer l’aptitude du mil à s*intégrer dans une rotation avec d’autres
cultures en culture irriguée, et particulièrement en contre saison (saison sèche chaude - de février a mai - quand
l’indice d’évaporation est très élevé pour les autres cultures) durant laquelle les ressources en eau sont limitées.
Les effets de la fertilisation, de la densité des plantes et &e la préparation du sol sur les rendements des variétés
ameliorées ont etté étudiés sous irrigation. La préparati 4n du sol n’a pas affecté le rendement moyen des deux
variétés du mil, l’hybride nain, 68 AX MLS de Nebrask’ et la variété synthétique locale de grande taille, GAM
8201 produisant respectivement 4025 kg ha’ et 4018 kg i:a-‘. Cependant, le rendement de la variété synthétique
locale a augmenté de 60% sur la culture à plat par rapport à la culture sur billon. L’apport d’engrais a un effet
signifïcatif sur le rendement de l’hybride GB 87-35. Par ailleurs, le mode de préparation du sol et la densité des
plantes ont améliorélesrendementsdelavariété68AXMLS
, mais pas CeuxdelavariétéGB
87-35. L’application
de ION 9P et 17 kg K ha-l et 45 kg N ha-’ comme urée a~ produit le rendement le plus élevé chez la variété 68A
244
SALICW DIANGAR ç!t al.
X MLS. Si la fréquence et ies doses d’irrigation n’ont pas affecté d’une façon significative (fiO.05) le: rendement
de la variété de mil 68A x MLS, les parcelles irriguées une fois par semaine et recevant un maximum de 75% dc
leurs besoins en eau (526 mm) ont produit les rendements les plus élevés (P<O. I2). II ressort de I’en~semble des
résultats obtenus que les rendement? du mil peuvent être sécurisés dans la zone semi-aride du Sénégal grâce 2 la
satisfaction des besoins minimaux e:n eau, le semis à une densité optimale de 83 333 plants parhectanz, l’efficacité
d’itinéraires techniques appropriés tels que la préparation du sol, et une fertilisation raisonnée, adaptés aux
variétés.
M~ors CMs: Utilisation d’engrais, irrigation, mil, zone agroécologique semi aride
INTRODUCTION
irrigation water and provide significantly greater
food production than flooded rice and utilise
Pearl millet (Pennisetum
gluucr~~z (L). R. BIT.)~ is
irrigation schcme resources at periods
the main staple in the diet of people in northern
unfavourable for other crops. A millet hybrid
Senegal. Yields are generally low because it is
from Nebraska (USA) and an early local improved
entirely rainfed and rainfall in this region is
variety were tested undervarious cu1tura.l practices.
presently low (~300 mm/year) and fluctuates
The objective of the study was to determine the
from month to month and even within mon&
response of improved millet varieties to trop
(Dancette, 1974; INTSORMIL, 1993). Limited
management (fertilisation, plant population and
rainfall in northern Senegal is patily attributed 1.0
land preparation) under irrigation.
advancing desertification of the Sahelo-Sudaniam
region. Furthermore, soils of the semiarid region
MATERIALS AND METHODS
are sandy, infertile and contain low levels of
organic matter. Itack of improved varieties as
The study was conducted in 1991 and 1992 with
well as poor trop and soi1 management also
double cropping each year (i.e., four test periods)
contribute to the low productivity under pea.sant
at the Thiago .and Fanaye villages located in
conditions in ibis environment.
Senegal River Valley in north Senegal.. The two
Researchers
in Senegal,Niger and other semiarid
villages belong to the Sahelo-Sudanian
countries have found that increa.sed productivity
agroecological zone of Senegal (Bèye, 1977).
of millet is possible with judicious management
The soils are Alfisols (USDA, 1975) and generally
under favourable climatic condi,tions (Bationo (~1
deficient in major nutrients (Nicou, 1976; Ndiaye,
al., 1993). Management decisions under farmers
1989; Piéri, 1989). Rainfall is low (250 to 300
control include moderate application of manure
mm) and uncertain; followed by a long dry season
and compost, a mixture of low doses of inorganic
usually lasting about 8 months. Mean annual
fertilisers with manure or compost, rotation with
temperature i,s 20°C to 35°C. Additional
legumes, varietal choices, appropriate land
information on the Sahelo-Sudani.an agro-
preparation techniques and proper sowing dates.
ecological zone of Senegal is provide:d by Sagna
Farmers do not have any direct control on
(1976).
rainfall but indirectly they manipulate it through
Six field trials under irrigation were conducted,
moistureretention strategies suchaslocal irrigation
four in 1991 with two planted on 13 o:TFebruary
canais, mulching and ditches. Added to that, in a
at Thiago (experiments 1 and 2) and two on 2 1 of
move to intensify land use and boost cere,al
August at Fanaye (experiments 3 and 4.), and two
production in the light of growing human and
in 1992,both planted on 11 of February
livestock populations, the government l-nas
(experiments 5 and 6). The design of the first tria1
provided irrigation facilities on 240,000 ha in
in 199 1 was a split plot with four replications. The
northern Senegal. Since millet has much lower
main plots were millet genotypes, early dwarf
waterrequirement than floodedrice, amuch la.rger
hybrid 68 A x MLS from University of Nebraska,
area of land could be cultivated with millet. ‘This
Lincol,n, Nebraska, USA and a local dwarf
could be a more efficient way to use scarce
synthetic variety, GAM 820 1. Subplots were either
An appraisal of irrigated te@erate and tropical millet
245
three land preparation techniques or fertiliser at ’replications. The weekly or twice weekly irrigation
three levels. The genotype differences were ~ treatments consisted of a total of 693 mm water
important and were deliberately chosen, 68A x ~ (maximum seasonal amount of water used by
MLS and GAM 8201 were both dwarf and likely ~millet) or 526 mm.water (75% of the maximum
to respond to higher plant densities; GB87-35 is Iconsumption). The design was a 2 by 2 factorial
medlum tall. Al1 are shorter season, reaching ~ in a randomised complete block with four
physiological maturity in about 80 days, 10-15 ~ replications. Maximum water consumption
(Mc)
days earlier than indigenous Senegalese varieties. ~ was calculated as: Mc = Kc * EV, where Kc = trop
Land preparation techniques were: 1) flat; 2)
coefficient and Ev=evapotranspiration.
Statistical
single ridges (width = 30 cm and height = 40 cm); ~ analyses were performed using SPSS (Norusis,
and 3) double ridges (width = 100 cm and
1997) and MSTAT-C (MSU, 1988).
height=30 cm). Al1 land preparation treatments ~
received 22 N, 21 P and 39 K kg ha-’ at planting
RESULTS AND DISCUSSIONS
and 67 N kg ha-’ urea as a top dressing 2 weeks ~
after planting. The fertiliser subplot treatments ~ Land preparation did not signifïcantly (-0.05)
applied in 199 1 were: 1) 15 N, 14 P and 26 K kg
~affect average yields of the Nebraska (68 A x
ha-‘; 2) 22 N, 21 P and 39 K kg ha-’ + 67 kg N , MLS) and the local synthetic (GAM 8201)
ha-’ as urea; and 3) 30 N, 28 P and 52 K kg ha-’ +
~varieties (Table 1). However, mean yields of the
90 kg N ha-’ as urea a11 applied at planting.
local synthetic varietyplantedontheflatincreased
The third and fourth experiments in 199 1 minor ~ by about 58% and 26%, relative to planting on
season were modified to include plant density as
single anddouble ridges, respectively. This
~
finding
an additional treatment and the synthetic variety is noteworthy, because flat planting is cheaper
GAM 820 I was replaced with variety GB 87-35. ~than ridge planting. Conversely, the Nebraska
Research protocols of the third and fourth ~ variety performed better in single ridges, showing
experiments were split-split-plot with four ~ a 14% increase over the flat. Average yields of
replications. Main plots werethreefertiliserlevels: ~ both millet genotypes, the Nebraska (4025 kg
1)15N,14Pand26Kkgha-‘+45Nkgha-‘asurea
~
ha-‘) and the local synthetic (4018 kg ha-‘) were
(divided into equal halves for top dressing at first ; not significantly different (P>O.O5) under
weeding and 30 days later), 2) 22 N, 21 P and 39 irrigation, but yields of the two varieties were
K kg ha-’ at planting and 67 N kg ha-’ as urea ~ much higher than yields on farmers’ fields under
(divided into equal halves for top dressing at first
rainfed conditions (430 - 1230 kg ha-‘) in the
weeding and 30 days later) and 3) 30 N, 28 P and
~
semiarid environment (Bationo et al., 1993).
52 K kg ha.’ + 90 kg N ha-’ as urea (divided into
~ The main effect of fertiliser was not significant
equal halves for top dressing at first weeding and ~ (fiO.05) nor was average varietal difference
30 days later). Subplots were land preparation on ~ (Table 1), even though the yield of the local
flat, simple ridges and double ridges. Sub-sub : synthetic outperformed the Nebraska hybrid by
plots for variety GB 87-35 were plants spaced 80 ~ 15 %. However, doubling the fertiliser rate of 15
x 80 cm, 80 x 60 cm, and 80 x 40 cm and sub- ~ N, 14 P and 26 K kg ha-’ + 45 N kg ha*’ urea,
subplot size was 11.52 m*. Spacings for variety
resulted in a significant (PcO.05) 20% yield
68A x MLS were 70 x 30 cm; 70 x 20 cm; 60 x 30
~
increase of 3295 kg ha-’ to 3911 kg ha’ for the
cm, and 60 x 20 cm.
~
Nebraska hybrid whereas the local synthetic
The rest of the studies conducted in 1992 ~ variety had a 10% yield reduction. Apart from
examined the use of reduced levels of fertilisers ~ genetic differences of the two genotypes, reasons
on hybrid 68 A x MLS and frequency and I for the varietal differences are unclear; we suggest
quantity of water applied through irrigation. i further research on this subject.
Fertiliser treatments were controlled with no
Results of yields for the new improved variety
fertiliser, 5 N, 4 P and 8 K kg ha’, 5 N, 4 P and 8
~
(GB 87-35) and the Nebraska hybrid (68 A x
K kg ha-’ + 22 N kg ha-’ as urea, and 10 N, 9 P and ) MLS) in separate experiments are given in Tables
19 K kg ha-’ + 45 N kg ha-’ urea. Experimental ~ 2 and 3. Low yields in Tables 2 and 3 were
design was a randomised block design with four ~ principally due to infestation of stalk borers
---
SALIOU DIANGAR et al.
(Coniestaignefusalis) andflower-suckinginsects
kg ha-’ with 30 N, 28 P and 52 K kg ha* + 90 N kg
(Psalydollitafuscaand Cylindrothoraxdussaulti).
ha-’ as urea but both of these were significantly
These insect pests caused about 30 % yield
higher than yields from 15 N, 14 P and 26 K kg
reduction. Fertiliser affected yield of GB 87-135
ha-’ + 415 N kg ha’ as urea. T~US, fiom this the
(PcO.05) but methods of land preparation and
recommended fertiliser rate is 22 N, 21. P and 39
plant density did not (-0.05) (Table2). The yield
K kg h;x* and 67 N kg ha-’ urea in combination
of 1525 kg ha’ with 22 N, 21 P and 39 K kg ha-’
with any of the planting arrangements and land
and 67 N kg ha’ urea was not different from 1601
preparation methods for variety GB 87-35.
TABLE 1. Effects of land preparation and fertiliser on grain yields of irrigated millet varieties in the semiarid area of
Senegal (Experiments 1 and 2)
Variety
Land preparation
Yield (kg ha-l)
Fertiliser
Yield (kg ha”)
68A x MLS
Flat
3,700
60 hl, 14 P and 26 K kg ha-l
3,300
Simple ridge
4,230
90 hl, 21 P and 39 K kg ha“
3,680
Double ridge
4,145
120 N, 28 P and 52 K kg ha-l
3.920
GAM 8201
Flat
4,974
60 hl, 14 P and 26 K kg ha-l
4,390
Simple ridge
3,146
90 hl, 21 P and 39 K kg ha-l
4,060
Double ridge
3,930
120 N, 28 P and 52 K kg ha-l
3903
Level of
Level of
significance
significance
Vanety (V)
n s
Variety
0.079
Land preparation (LP)
Fertlliser
n s
VXLP
z43
Varisty x fertiliser
n s
TABLE 2. Effect of fertiliser, land preparation and plant density on grain yields of irrigated millet (variety GB 87-35)
in the semiarid area of Senegal (Experiment 3)
Fertiliser
Density
Flat
Simple ridge
L.arge tidge
-
Yield (kg ha -l)
60N,14Pand26Kkghaw1
80cmxBOcm
1,120
1,170
1,330
80cmx60cm
il ,290
1,040
1,590
80cmx40cm
11,170
1,250
1,490
QON,21 Pand39Kkgha-1
80cmx80cm
il ,950
1,410
1,700
80cmx60cm
II ,490
1,270
1,610
80cmx40cm
II ,300
1,660
1.350
120 N, 28 P and 52 K kg ha -l
80cmx80cm
II ,770
1,510
1,800
80cmx60cm
11,640
1,460
1,610
80cmx40cm
‘1,630
1,610
1,370
Level of slgnificance
Density (D)
Fertiliser (F)
:so13
Land preparation (LP)
n s
LP x F
LP x D
:y047
LPxDxF
ns
An appraisal of irrigated tt perate and tropical millet
2 4 7
The resulls shown in Table 3 indicate that
(60 cm x 30 cm or 60 cm x 20 cm) were superior
fertiliser did not (tiO.05) affect yields of the
to wider (70 cm x 20 or 30 cm) spacings. It appears
Nebraska variety but land preparation techniques
that the best cultural practices to attain optimum
and cropping patterns did (PcO.05). Also, ridges,
yields of the Nebraska hybrid are ridges (large or
both single and double, improved yields of the
$mall)andahighplantpopulationdensity.Average
Nebraska millet variety relative to planting on the
yields were higher with the Nebraska hybrid.
flat, consistent with the other study mentioned
Results of the effect of reduced fertiliser rates
above. YieIds of closer spacings between rows
m the Nebraska variety are presented in Table 4.
TABLE 3. Effe~t of fertiliser, land preparation and plant dc
;ityon grain yields of irrigated millet (hybrid 68 A x MLS)
in the semiarid area of Senegal (Experiment 4)
Fertiliser
Density
Flat
Simple ridge
Large ridge
kg ha -l
60 N, 14 P and 26 K kg ha-l
70cmx30cm
2,030
2,340
2,030
70cmx20cm
2,170
1,790
2,380
60cmx30cm
2,060
2,660
2,420
60cmx20cm
2,730
2,630
2,330
90 N, 21 P and 39 K kg ha-’
70cmx30cm
1,850
2,250
2,200
70cmx20cm
1,680
2,170
2,420
60cmx30cm
2,370
2,730
2,830
60cmx20cm
1,890
2,530
2,280
120 N, 28 P and 52 K kg ha-’
70cmx30cm
1,280
2,650
2,820
70cmx20cm
1,790
2,420
2,680
60cmx30cm
2,370
2,680
2,980
60cmx20cm
1,740
2,630
3,130
Level of significance
Plant density (D)
0.0001
Fertiliser (F)
Land preparation (LP)
:5004
LPxF
0.094
LPxD
n s
LPxDxF
n s
TABLE 4. Effects of reduced rates of minera1 fertiliser on g
n and stoveryields of irrigated millet (hybrid 68 Ax MLS)
in the semiarid zone of Senegal in 1992 (Experiment 5)
Fertiliser
Grain yield (kg ha’
Stover yield (kg ha-l)
Control (without fertiliser)
3,300
4,090
5 N, 4 P and 8 K kg ha-l
2,440
3,020
27N,4Pand8Kkgham1
3,810
4,740
55N,9Pand25Kkgham1
4,880
6,330
Level of significance
Fertiliser
0.062
0.004
248
SALIOU DIANGAR et al,
TABLE5. ,Millet(hybrid68AxMLS)glrain yieldandyieldcomponentsasaffected byirriga6onfmquencyandamount
of water in 1992 (Experiment 6)
Frsquency
Amount
Yield
Stover Non productive
Productive - Head
tlllers
tîllers
S/d
kg ha“
(no. of tillers ha-l)
l/week
Maximum
75%t maximum
3,620
4,660
5,423 6,915
274,210 172,590
637,135
3,155
727,465
3,641
2lweek
Maximum
75%tmaximurn
3,560
4,390
5,202 6,472
220,980 196,766
562,290
2,974
640,361
3,644
Level of signifkance
Fr-(0
0.17
0.13
-nt (4
Ti2
;:2
ns
0.12
zl3
FxA
ns
0.05
ns
ns
ns
tAbout 75 % of 693 mm water per ysar
Both grain and stover yields of millet responded water needed by millet tended to yield more than
to fertiliser (PcO.05). Highest yields for grain
plots with full water dose. Millet is acrop adapted
(4880 kg ha-l) and stover (6330 kg hx’) were to low moisture conditions, particularly
obtained with the application of 110 N, 9 P and 117 unsaturated soils, thus having lower water
K kg ha-* + 45 N kg ha-’ urea. The lowest yields
requirements (Vachaud et al., 1978; Sarr et al.,
came from 5 N, 4 P and 8 K kg ha-‘. The 36%
1998). Also, irrigation once per week yielded
difference in yield between the control(3300 k8
slightly higher than twice per wek (Table 5).
ha-‘) and the 5 N, 4 P and 8 K kg ha-’ (2440 k8
Govemments and private organisations of the sub
ha-‘) plots is not significantly different (p>O.OS)l.
region should consider developing low level
Yields of the Nebraska hybrib in Table 1
irrigation facilities to promote higher and more
compared favourably with the reduced yields in
stable yields of millet and other cereal crops.
Table 4, thus, it may be safe to recommend less
These crops may offer a more efficient way to use
fertiliser (10 N, 9 P and 17 K kg ha-’ + 45 N k8
scarce irrigation water to produce more food per
ha-’ as urea) to make this technology affordable to
unit area than high water-requiring crops such as
many farmers. Recent and past ex.periences in the
lowland rice. However, environmental impact
semiarid region of Senegal have demonstrated
assessment studies should be done before
that acombination ofmineral fertilisers and animal
implementing such programmes.
manure or compost improved rai.nfed millet and
peanut yields betterthan either manure or fertiliser
A C K N O W L E D G M E N T S
alone(Chan-eauandNicou, 1971;Badiane, 1988).
Additional research is needed on irrigated millet This project is a contribution of Natural IResource
to determine whether adding organic amendments
Based Agricultural Research project and was
could lùrther improve yield in the presence of
supported by a grant from USAID/CID/Senegal
inorganic fertiliser.
project #685-0285-C-00-2329-00. Appreciation
Data on yield and yield components of the
is expressed to Dr. David Andrews of University
Nebraska hybrid as affected by frequency and
of Nebraska-Lincoln, for supplying 68 A X MLS
amount of irrigation water are given in Table 5.
and his advice on the design of the studies and Dr.
Neither the frequency nor the amount of water
Richard P. Dick, Oregon State University, for
significantly (fiO.05) affected millet yield.
editing tbe nianuscript.
However, plots irrigated witb ‘75% maximum
An appraisal of irrigated tel pperate and tropical millet
249
l
REFERENCES
~Ndiaye, J.P. et Sagna, A. 1989. La fertilisation des
cultures au Sénégal. Bilan diagnostic et
Badiane, A.N., 1988. Courbe de réponse à des
,
perspectives. Ministère du Développement
doses croissantes de fumier (Thilmakha). Essai ~ R u r a l . 99pp.
travail du sol (Sole III Nord du CNRA de ~Norusis, M. J. 1997. SPSS 7.5. Guide to data
Bambey). Essai régénération des sols
analysis. Prentice Hall, Upper Saddle River,
Ndi&mane. Résultats de 1987. ISRA-CNRA ~ NJ.
de Bambey.
~Piéri, C. 1989. Fertilité des terres des savanes.
Bationo, A., Christianson, C.B. and Klaij, M.C.
i
Bilans de 30 ans de recherche et de
1993. The effect of trop residues and fertilizer ~
développement agricole au Sud du Sahara.
on use of Pearl millet yields in Niger. ~
Min. Coopération et CIRAD-IRAT. 444 pp.
Fertilizer Research 34251-258.
~
Sagna, A. 1976. Le bilan de pluies au Senegal de
Bèye, G. 1977. Dégradation des sols au Sénégal.
~
1944 a 1973. Dakar, Faculte des Lettres et des
Situation actuelle et perspectives. ISRA-
Sciences Humaines, Dept. de Geographié,
CNRA. 23pp.
i
267 pp.
Charreau, C. et Nicou, C. 1971. L’amélioration
~Sarr, B., Boggio, D., Annerose, D. and Macauley,
du profil cultural dans les sols sableux et sablo
R.H. 1998. In: Approach in modeling
argileux de la zone tropicale sèche Ouest-
environmental factors and genotypes
Africaine et ses incidences agronomiques.
interaction: a case study of millet (Pennisetum
Agronomy Tropicale 26: 1183- 1247.
~
Glaucum). European Society of Agronomy,
Dancette, C. 1974. Les besions en eau des plantes
~
5th ESA Congress, Nitra, Rep. Slovaque, 28
des grandes culture au Senegal. In: Isotopes ~
Juin - 3 Juillet 1998.
and radiation techniques in soi1 physics and ~USDA. 1975. Soi1 taxonomy. A basic of soi1
irrigation studies. Vienne, AIEA. pp. 351- ~
classification for making and interpreting soi1
371.
~
surveys. U.S. Soi1 Conservation Service.
INTSORMIL. 1993.Annualreportl992.
Senegal ~Vachaud, G., Dancette, C., Sonko, S. et Thony,
Country Report. University of Nebraska- J.L. 1978. Méthodes de caractérisation
Lincoln, NE. pp. 277-230.
~
hydrodynamique in situ d’un sol non saturé
MSU. 1988. Microcomputer Statistical Program.
du Sénégal en vue de la détermination des
Michigan State University, East Lansing, MI. ~
termes du bilan hydrique. Ann. Agron. 29: l-
~
36.
ISSN 1021-9730/2000 $4.00
Printed in Uganda. Al1 rights reserved
02000, African Crop Science Society
AN APPRAISAL OF IRRIGATED TElhPERATE AND TROPICAL MILLET
VARIETIES IN THE SEM1AkI.D REGION OF SENEGAL
SALIOU DIANGAR, TANGU BA and C. F. YAMOAH’
Institut Senegalais de Recherches Agricoles, ISRA CNRA, Bambey, BP 53 Senegal
‘Soi1 Institute, kumasi, Ghana
(Received 6 November, 19p8; accepted 2 May, 2000)
ABSTitACT
Pearl millet (Pennisetum glaucum (L). R. Br.) constitutes a major food trop in the semiarid region of West Africa
but yields are extremely low in subsistence cropping sysfems because of inappropriate management and scarcity
of water. This study was designed to see if Pearl millet could become a component of trop rotation in an irrigated
scheme, and additionally, at a particular season (hot dry season - February to May when evapotranspiration levels
are too high for other crops) when water supplies are low. The effects of fertilisation, plant density and land
preparation on yields of improved varieties were tested under irrigation. Land preparation did not affect mean
yields of the two very early dwarfmillet genotypes, hybrid 68 A x MLS from Nebraska (4025 kg ha-‘) and a local
synthetic, variety GAM 8201 (4018 kg ha’). However, $ield of taller, but early, local synthetic variety increased
by 60% when planted on flat as opposed to ridges. Effect of fertiliser on yield of hybrid GB 87-35 was signifïcant.
In addition, method of land preparation and plant den&y increased yields of 68 A x MLS but not GB87-35.
Application of 10 N, 9 P and 17 kg K ha-’ and 45 kg N ha-’ as urea gave the highest grain yield of variety 68 A
x MLS. Neither frequency nor amount of irrigated wa@ significantly (PcO.05) affected yield of pearl millet
varlety 68 A x MLS, but we noted a trend (PcO.12) where plots irrigated once per week with 75% maximum,
seasonal trop water requirement of millet (526 mm) produced higher yields. Millet yields could be sustained in
semiarid environments by ensuring a minimum and reliable water supply (526 mm during the dry season),
optimum plant population (60 cm x 20 or 30 cm) as well as moderate fertilisation (10 N, 9 P and 17 K kg ha1 +
45 kg N ha-r as urea), and suitable land preparation method, depending on variety.
Key Words: Agro-ecological zone, fertiliser use, irrigation, Pearl millet, semi-arid
RE@MI?
Le mil est la principale culture vivrière dans les régions semi-arides de l’Afrique Occidentale. Les faibles
rendements enregistrés dans les systèmes de culture de subsistance sont le fait d’une gestion inadéquate et de la
rarett5 de l’eau. Cette étude a pour but de déterminer l’aptitude du mil à s*intégrer dans une rotation avec d’autres
cultures en culture irriguée, et particulièrement en contre saison (saison sèche chaude - de février a mai - quand
l’indice d’évaporation est très élevé pour les autres cultures) durant laquelle les ressources en eau sont limitées.
Les effets de la fertilisation, de la densité des plantes et &e la préparation du sol sur les rendements des variétés
ameliorées ont etté étudiés sous irrigation. La préparati 4n du sol n’a pas affecté le rendement moyen des deux
variétés du mil, l’hybride nain, 68 AX MLS de Nebrask’ et la variété synthétique locale de grande taille, GAM
8201 produisant respectivement 4025 kg ha’ et 4018 kg i:a-‘. Cependant, le rendement de la variété synthétique
locale a augmenté de 60% sur la culture à plat par rapport à la culture sur billon. L’apport d’engrais a un effet
signifïcatif sur le rendement de l’hybride GB 87-35. Par ailleurs, le mode de préparation du sol et la densité des
plantes ont améliorélesrendementsdelavariété68AXMLS
, mais pas CeuxdelavariétéGB
87-35. L’application
de ION 9P et 17 kg K ha-l et 45 kg N ha-’ comme urée a~ produit le rendement le plus élevé chez la variété 68A
244
SALICW DIANGAR ç!t al.
X MLS. Si la fréquence et ies doses d’irrigation n’ont pas affecté d’une façon significative (fiO.05) le: rendement
de la variété de mil 68A x MLS, les parcelles irriguées une fois par semaine et recevant un maximum de 75% dc
leurs besoins en eau (526 mm) ont produit les rendements les plus élevés (P<O. I2). II ressort de I’en~semble des
résultats obtenus que les rendement? du mil peuvent être sécurisés dans la zone semi-aride du Sénégal grâce 2 la
satisfaction des besoins minimaux e:n eau, le semis à une densité optimale de 83 333 plants parhectanz, l’efficacité
d’itinéraires techniques appropriés tels que la préparation du sol, et une fertilisation raisonnée, adaptés aux
variétés.
M~ors CMs: Utilisation d’engrais, irrigation, mil, zone agroécologique semi aride
INTRODUCTION
irrigation water and provide significantly greater
food production than flooded rice and utilise
Pearl millet (Pennisetum
gluucr~~z (L). R. BIT.)~ is
irrigation schcme resources at periods
the main staple in the diet of people in northern
unfavourable for other crops. A millet hybrid
Senegal. Yields are generally low because it is
from Nebraska (USA) and an early local improved
entirely rainfed and rainfall in this region is
variety were tested undervarious cu1tura.l practices.
presently low (~300 mm/year) and fluctuates
The objective of the study was to determine the
from month to month and even within mon&
response of improved millet varieties to trop
(Dancette, 1974; INTSORMIL, 1993). Limited
management (fertilisation, plant population and
rainfall in northern Senegal is patily attributed 1.0
land preparation) under irrigation.
advancing desertification of the Sahelo-Sudaniam
region. Furthermore, soils of the semiarid region
MATERIALS AND METHODS
are sandy, infertile and contain low levels of
organic matter. Itack of improved varieties as
The study was conducted in 1991 and 1992 with
well as poor trop and soi1 management also
double cropping each year (i.e., four test periods)
contribute to the low productivity under pea.sant
at the Thiago .and Fanaye villages located in
conditions in ibis environment.
Senegal River Valley in north Senegal.. The two
Researchers
in Senegal,Niger and other semiarid
villages belong to the Sahelo-Sudanian
countries have found that increa.sed productivity
agroecological zone of Senegal (Bèye, 1977).
of millet is possible with judicious management
The soils are Alfisols (USDA, 1975) and generally
under favourable climatic condi,tions (Bationo (~1
deficient in major nutrients (Nicou, 1976; Ndiaye,
al., 1993). Management decisions under farmers
1989; Piéri, 1989). Rainfall is low (250 to 300
control include moderate application of manure
mm) and uncertain; followed by a long dry season
and compost, a mixture of low doses of inorganic
usually lasting about 8 months. Mean annual
fertilisers with manure or compost, rotation with
temperature i,s 20°C to 35°C. Additional
legumes, varietal choices, appropriate land
information on the Sahelo-Sudani.an agro-
preparation techniques and proper sowing dates.
ecological zone of Senegal is provide:d by Sagna
Farmers do not have any direct control on
(1976).
rainfall but indirectly they manipulate it through
Six field trials under irrigation were conducted,
moistureretention strategies suchaslocal irrigation
four in 1991 with two planted on 13 o:TFebruary
canais, mulching and ditches. Added to that, in a
at Thiago (experiments 1 and 2) and two on 2 1 of
move to intensify land use and boost cere,al
August at Fanaye (experiments 3 and 4.), and two
production in the light of growing human and
in 1992,both planted on 11 of February
livestock populations, the government l-nas
(experiments 5 and 6). The design of the first tria1
provided irrigation facilities on 240,000 ha in
in 199 1 was a split plot with four replications. The
northern Senegal. Since millet has much lower
main plots were millet genotypes, early dwarf
waterrequirement than floodedrice, amuch la.rger
hybrid 68 A x MLS from University of Nebraska,
area of land could be cultivated with millet. ‘This
Lincol,n, Nebraska, USA and a local dwarf
could be a more efficient way to use scarce
synthetic variety, GAM 820 1. Subplots were either
An appraisal of irrigated te@erate and tropical millet
245
three land preparation techniques or fertiliser at ’replications. The weekly or twice weekly irrigation
three levels. The genotype differences were ~ treatments consisted of a total of 693 mm water
important and were deliberately chosen, 68A x ~ (maximum seasonal amount of water used by
MLS and GAM 8201 were both dwarf and likely ~millet) or 526 mm.water (75% of the maximum
to respond to higher plant densities; GB87-35 is Iconsumption). The design was a 2 by 2 factorial
medlum tall. Al1 are shorter season, reaching ~ in a randomised complete block with four
physiological maturity in about 80 days, 10-15 ~ replications. Maximum water consumption
(Mc)
days earlier than indigenous Senegalese varieties. ~ was calculated as: Mc = Kc * EV, where Kc = trop
Land preparation techniques were: 1) flat; 2)
coefficient and Ev=evapotranspiration.
Statistical
single ridges (width = 30 cm and height = 40 cm); ~ analyses were performed using SPSS (Norusis,
and 3) double ridges (width = 100 cm and
1997) and MSTAT-C (MSU, 1988).
height=30 cm). Al1 land preparation treatments ~
received 22 N, 21 P and 39 K kg ha-’ at planting
RESULTS AND DISCUSSIONS
and 67 N kg ha-’ urea as a top dressing 2 weeks ~
after planting. The fertiliser subplot treatments ~ Land preparation did not signifïcantly (-0.05)
applied in 199 1 were: 1) 15 N, 14 P and 26 K kg
~affect average yields of the Nebraska (68 A x
ha-‘; 2) 22 N, 21 P and 39 K kg ha-’ + 67 kg N , MLS) and the local synthetic (GAM 8201)
ha-’ as urea; and 3) 30 N, 28 P and 52 K kg ha-’ +
~varieties (Table 1). However, mean yields of the
90 kg N ha-’ as urea a11 applied at planting.
local synthetic varietyplantedontheflatincreased
The third and fourth experiments in 199 1 minor ~ by about 58% and 26%, relative to planting on
season were modified to include plant density as
single anddouble ridges, respectively. This
~
finding
an additional treatment and the synthetic variety is noteworthy, because flat planting is cheaper
GAM 820 I was replaced with variety GB 87-35. ~than ridge planting. Conversely, the Nebraska
Research protocols of the third and fourth ~ variety performed better in single ridges, showing
experiments were split-split-plot with four ~ a 14% increase over the flat. Average yields of
replications. Main plots werethreefertiliserlevels: ~ both millet genotypes, the Nebraska (4025 kg
1)15N,14Pand26Kkgha-‘+45Nkgha-‘asurea
~
ha-‘) and the local synthetic (4018 kg ha-‘) were
(divided into equal halves for top dressing at first ; not significantly different (P>O.O5) under
weeding and 30 days later), 2) 22 N, 21 P and 39 irrigation, but yields of the two varieties were
K kg ha-’ at planting and 67 N kg ha-’ as urea ~ much higher than yields on farmers’ fields under
(divided into equal halves for top dressing at first
rainfed conditions (430 - 1230 kg ha-‘) in the
weeding and 30 days later) and 3) 30 N, 28 P and
~
semiarid environment (Bationo et al., 1993).
52 K kg ha.’ + 90 kg N ha-’ as urea (divided into
~ The main effect of fertiliser was not significant
equal halves for top dressing at first weeding and ~ (fiO.05) nor was average varietal difference
30 days later). Subplots were land preparation on ~ (Table 1), even though the yield of the local
flat, simple ridges and double ridges. Sub-sub : synthetic outperformed the Nebraska hybrid by
plots for variety GB 87-35 were plants spaced 80 ~ 15 %. However, doubling the fertiliser rate of 15
x 80 cm, 80 x 60 cm, and 80 x 40 cm and sub- ~ N, 14 P and 26 K kg ha-’ + 45 N kg ha*’ urea,
subplot size was 11.52 m*. Spacings for variety
resulted in a significant (PcO.05) 20% yield
68A x MLS were 70 x 30 cm; 70 x 20 cm; 60 x 30
~
increase of 3295 kg ha-’ to 3911 kg ha’ for the
cm, and 60 x 20 cm.
~
Nebraska hybrid whereas the local synthetic
The rest of the studies conducted in 1992 ~ variety had a 10% yield reduction. Apart from
examined the use of reduced levels of fertilisers ~ genetic differences of the two genotypes, reasons
on hybrid 68 A x MLS and frequency and I for the varietal differences are unclear; we suggest
quantity of water applied through irrigation. i further research on this subject.
Fertiliser treatments were controlled with no
Results of yields for the new improved variety
fertiliser, 5 N, 4 P and 8 K kg ha’, 5 N, 4 P and 8
~
(GB 87-35) and the Nebraska hybrid (68 A x
K kg ha-’ + 22 N kg ha-’ as urea, and 10 N, 9 P and ) MLS) in separate experiments are given in Tables
19 K kg ha-’ + 45 N kg ha-’ urea. Experimental ~ 2 and 3. Low yields in Tables 2 and 3 were
design was a randomised block design with four ~ principally due to infestation of stalk borers
---
SALIOU DIANGAR et al.
(Coniestaignefusalis) andflower-suckinginsects
kg ha-’ with 30 N, 28 P and 52 K kg ha* + 90 N kg
(Psalydollitafuscaand Cylindrothoraxdussaulti).
ha-’ as urea but both of these were significantly
These insect pests caused about 30 % yield
higher than yields from 15 N, 14 P and 26 K kg
reduction. Fertiliser affected yield of GB 87-135
ha-’ + 415 N kg ha’ as urea. T~US, fiom this the
(PcO.05) but methods of land preparation and
recommended fertiliser rate is 22 N, 21. P and 39
plant density did not (-0.05) (Table2). The yield
K kg h;x* and 67 N kg ha-’ urea in combination
of 1525 kg ha’ with 22 N, 21 P and 39 K kg ha-’
with any of the planting arrangements and land
and 67 N kg ha’ urea was not different from 1601
preparation methods for variety GB 87-35.
TABLE 1. Effects of land preparation and fertiliser on grain yields of irrigated millet varieties in the semiarid area of
Senegal (Experiments 1 and 2)
Variety
Land preparation
Yield (kg ha-l)
Fertiliser
Yield (kg ha”)
68A x MLS
Flat
3,700
60 hl, 14 P and 26 K kg ha-l
3,300
Simple ridge
4,230
90 hl, 21 P and 39 K kg ha“
3,680
Double ridge
4,145
120 N, 28 P and 52 K kg ha-l
3.920
GAM 8201
Flat
4,974
60 hl, 14 P and 26 K kg ha-l
4,390
Simple ridge
3,146
90 hl, 21 P and 39 K kg ha-l
4,060
Double ridge
3,930
120 N, 28 P and 52 K kg ha-l
3903
Level of
Level of
significance
significance
Vanety (V)
n s
Variety
0.079
Land preparation (LP)
Fertlliser
n s
VXLP
z43
Varisty x fertiliser
n s
TABLE 2. Effect of fertiliser, land preparation and plant density on grain yields of irrigated millet (variety GB 87-35)
in the semiarid area of Senegal (Experiment 3)
Fertiliser
Density
Flat
Simple ridge
L.arge tidge
-
Yield (kg ha -l)
60N,14Pand26Kkghaw1
80cmxBOcm
1,120
1,170
1,330
80cmx60cm
il ,290
1,040
1,590
80cmx40cm
11,170
1,250
1,490
QON,21 Pand39Kkgha-1
80cmx80cm
il ,950
1,410
1,700
80cmx60cm
II ,490
1,270
1,610
80cmx40cm
II ,300
1,660
1.350
120 N, 28 P and 52 K kg ha -l
80cmx80cm
II ,770
1,510
1,800
80cmx60cm
11,640
1,460
1,610
80cmx40cm
‘1,630
1,610
1,370
Level of slgnificance
Density (D)
Fertiliser (F)
:so13
Land preparation (LP)
n s
LP x F
LP x D
:y047
LPxDxF
ns
An appraisal of irrigated tt perate and tropical millet
2 4 7
The resulls shown in Table 3 indicate that
(60 cm x 30 cm or 60 cm x 20 cm) were superior
fertiliser did not (tiO.05) affect yields of the
to wider (70 cm x 20 or 30 cm) spacings. It appears
Nebraska variety but land preparation techniques
that the best cultural practices to attain optimum
and cropping patterns did (PcO.05). Also, ridges,
yields of the Nebraska hybrid are ridges (large or
both single and double, improved yields of the
$mall)andahighplantpopulationdensity.Average
Nebraska millet variety relative to planting on the
yields were higher with the Nebraska hybrid.
flat, consistent with the other study mentioned
Results of the effect of reduced fertiliser rates
above. YieIds of closer spacings between rows
m the Nebraska variety are presented in Table 4.
TABLE 3. Effe~t of fertiliser, land preparation and plant dc
;ityon grain yields of irrigated millet (hybrid 68 A x MLS)
in the semiarid area of Senegal (Experiment 4)
Fertiliser
Density
Flat
Simple ridge
Large ridge
kg ha -l
60 N, 14 P and 26 K kg ha-l
70cmx30cm
2,030
2,340
2,030
70cmx20cm
2,170
1,790
2,380
60cmx30cm
2,060
2,660
2,420
60cmx20cm
2,730
2,630
2,330
90 N, 21 P and 39 K kg ha-’
70cmx30cm
1,850
2,250
2,200
70cmx20cm
1,680
2,170
2,420
60cmx30cm
2,370
2,730
2,830
60cmx20cm
1,890
2,530
2,280
120 N, 28 P and 52 K kg ha-’
70cmx30cm
1,280
2,650
2,820
70cmx20cm
1,790
2,420
2,680
60cmx30cm
2,370
2,680
2,980
60cmx20cm
1,740
2,630
3,130
Level of significance
Plant density (D)
0.0001
Fertiliser (F)
Land preparation (LP)
:5004
LPxF
0.094
LPxD
n s
LPxDxF
n s
TABLE 4. Effects of reduced rates of minera1 fertiliser on g
n and stoveryields of irrigated millet (hybrid 68 Ax MLS)
in the semiarid zone of Senegal in 1992 (Experiment 5)
Fertiliser
Grain yield (kg ha’
Stover yield (kg ha-l)
Control (without fertiliser)
3,300
4,090
5 N, 4 P and 8 K kg ha-l
2,440
3,020
27N,4Pand8Kkgham1
3,810
4,740
55N,9Pand25Kkgham1
4,880
6,330
Level of significance
Fertiliser
0.062
0.004
248
SALIOU DIANGAR et al,
TABLE5. ,Millet(hybrid68AxMLS)glrain yieldandyieldcomponentsasaffected byirriga6onfmquencyandamount
of water in 1992 (Experiment 6)
Frsquency
Amount
Yield
Stover Non productive
Productive - Head
tlllers
tîllers
S/d
kg ha“
(no. of tillers ha-l)
l/week
Maximum
75%t maximum
3,620
4,660
5,423 6,915
274,210 172,590
637,135
3,155
727,465
3,641
2lweek
Maximum
75%tmaximurn
3,560
4,390
5,202 6,472
220,980 196,766
562,290
2,974
640,361
3,644
Level of signifkance
Fr-(0
0.17
0.13
-nt (4
Ti2
;:2
ns
0.12
zl3
FxA
ns
0.05
ns
ns
ns
tAbout 75 % of 693 mm water per ysar
Both grain and stover yields of millet responded water needed by millet tended to yield more than
to fertiliser (PcO.05). Highest yields for grain
plots with full water dose. Millet is acrop adapted
(4880 kg ha-l) and stover (6330 kg hx’) were to low moisture conditions, particularly
obtained with the application of 110 N, 9 P and 117 unsaturated soils, thus having lower water
K kg ha-* + 45 N kg ha-’ urea. The lowest yields
requirements (Vachaud et al., 1978; Sarr et al.,
came from 5 N, 4 P and 8 K kg ha-‘. The 36%
1998). Also, irrigation once per week yielded
difference in yield between the control(3300 k8
slightly higher than twice per wek (Table 5).
ha-‘) and the 5 N, 4 P and 8 K kg ha-’ (2440 k8
Govemments and private organisations of the sub
ha-‘) plots is not significantly different (p>O.OS)l.
region should consider developing low level
Yields of the Nebraska hybrib in Table 1
irrigation facilities to promote higher and more
compared favourably with the reduced yields in
stable yields of millet and other cereal crops.
Table 4, thus, it may be safe to recommend less
These crops may offer a more efficient way to use
fertiliser (10 N, 9 P and 17 K kg ha-’ + 45 N k8
scarce irrigation water to produce more food per
ha-’ as urea) to make this technology affordable to
unit area than high water-requiring crops such as
many farmers. Recent and past ex.periences in the
lowland rice. However, environmental impact
semiarid region of Senegal have demonstrated
assessment studies should be done before
that acombination ofmineral fertilisers and animal
implementing such programmes.
manure or compost improved rai.nfed millet and
peanut yields betterthan either manure or fertiliser
A C K N O W L E D G M E N T S
alone(Chan-eauandNicou, 1971;Badiane, 1988).
Additional research is needed on irrigated millet This project is a contribution of Natural IResource
to determine whether adding organic amendments
Based Agricultural Research project and was
could lùrther improve yield in the presence of
supported by a grant from USAID/CID/Senegal
inorganic fertiliser.
project #685-0285-C-00-2329-00. Appreciation
Data on yield and yield components of the
is expressed to Dr. David Andrews of University
Nebraska hybrid as affected by frequency and
of Nebraska-Lincoln, for supplying 68 A X MLS
amount of irrigation water are given in Table 5.
and his advice on the design of the studies and Dr.
Neither the frequency nor the amount of water
Richard P. Dick, Oregon State University, for
significantly (fiO.05) affected millet yield.
editing tbe nianuscript.
However, plots irrigated witb ‘75% maximum
An appraisal of irrigated tel pperate and tropical millet
249
l
REFERENCES
~Ndiaye, J.P. et Sagna, A. 1989. La fertilisation des
cultures au Sénégal. Bilan diagnostic et
Badiane, A.N., 1988. Courbe de réponse à des
,
perspectives. Ministère du Développement
doses croissantes de fumier (Thilmakha). Essai ~ R u r a l . 99pp.
travail du sol (Sole III Nord du CNRA de ~Norusis, M. J. 1997. SPSS 7.5. Guide to data
Bambey). Essai régénération des sols
analysis. Prentice Hall, Upper Saddle River,
Ndi&mane. Résultats de 1987. ISRA-CNRA ~ NJ.
de Bambey.
~Piéri, C. 1989. Fertilité des terres des savanes.
Bationo, A., Christianson, C.B. and Klaij, M.C.
i
Bilans de 30 ans de recherche et de
1993. The effect of trop residues and fertilizer ~
développement agricole au Sud du Sahara.
on use of Pearl millet yields in Niger. ~
Min. Coopération et CIRAD-IRAT. 444 pp.
Fertilizer Research 34251-258.
~
Sagna, A. 1976. Le bilan de pluies au Senegal de
Bèye, G. 1977. Dégradation des sols au Sénégal.
~
1944 a 1973. Dakar, Faculte des Lettres et des
Situation actuelle et perspectives. ISRA-
Sciences Humaines, Dept. de Geographié,
CNRA. 23pp.
i
267 pp.
Charreau, C. et Nicou, C. 1971. L’amélioration
~Sarr, B., Boggio, D., Annerose, D. and Macauley,
du profil cultural dans les sols sableux et sablo
R.H. 1998. In: Approach in modeling
argileux de la zone tropicale sèche Ouest-
environmental factors and genotypes
Africaine et ses incidences agronomiques.
interaction: a case study of millet (Pennisetum
Agronomy Tropicale 26: 1183- 1247.
~
Glaucum). European Society of Agronomy,
Dancette, C. 1974. Les besions en eau des plantes
~
5th ESA Congress, Nitra, Rep. Slovaque, 28
des grandes culture au Senegal. In: Isotopes ~
Juin - 3 Juillet 1998.
and radiation techniques in soi1 physics and ~USDA. 1975. Soi1 taxonomy. A basic of soi1
irrigation studies. Vienne, AIEA. pp. 351- ~
classification for making and interpreting soi1
371.
~
surveys. U.S. Soi1 Conservation Service.
INTSORMIL. 1993.Annualreportl992.
Senegal ~Vachaud, G., Dancette, C., Sonko, S. et Thony,
Country Report. University of Nebraska- J.L. 1978. Méthodes de caractérisation
Lincoln, NE. pp. 277-230.
~
hydrodynamique in situ d’un sol non saturé
MSU. 1988. Microcomputer Statistical Program.
du Sénégal en vue de la détermination des
Michigan State University, East Lansing, MI. ~
termes du bilan hydrique. Ann. Agron. 29: l-
~
36.