Agricultural and Forest Meteorology, 55 ( 199 1) 25...
Agricultural and Forest Meteorology, 55 ( 199 1) 25 l-263
Elsevier Science Publishers B.V., Amsterdam
çkle
14 i-L&
Determination of potential lengths of the trop
growing period in semi-arid regions of Senegal
J.-L.B. Khalfaoui
Institut Sénégalais de Recherches Agricoles (ISRA), CNRA, Bambey, Senegal
(Received 8 August 199’0; revision accepted 19 November 1990)
‘4BSTRACT
Khalfaoui, J.-L.B., 199 1. Determination of potential lengths of the trop growing period in semi-arid
regions of Senegal. Agric. For. Meteorol., 55: 251-263.
A study of rainfall data from the Senegal semi-arid region shows that one of the major components
of the deterioration in rainfall in the Sahel is the very marked reduction in the length of the rainy
season since 1970. This reduction means that lengths of the trop growing period have to be reviewed
for each region. A study method is described which takes into account the inherent climatic, economic
and technical constraints in Sahelian agriculture. It enables potential lengths of the growing period to
be determined, adapted to each region. Among the various means used to study drought, this method,
in particular, enables the setting of plant breeding targets. It has been applied to the case of the Senegal
semi-arid region.
INTRODUCTION
Rainfall totals since the start of the 1940s show that in India, the drought
is the result of signifïcant rainfall fluctuations from year to year, with no over-
a11 drop. However, for the Sudan-Sahel region of Africa and for Australia, a
tendency towards lower rainfall began during the 1950s with an apparent
stabilisation from 1972- 1973 onwards. This also applies to North and South
hmerica, particularly northeastern Brazil, since 1945, and, more recently, for
eastern and southern Africa (UNESCO, 1984).
In Africa, there is statistical evidence for a degree of periodicity as regards
series of rainfall figures. The previous two dry periods occurred during the
1910s and 1940s both of which lasted for a shorter time than the current
drought. Nonetheless,.this periodicity is of no value in forecasting, owing to
random phase modifications of these chronological series (UNESCO, 1984).
The dynamic causes of drought are little known, as climatologists still have
great diffculties distinguishing causes from correlated effects which may be
synergistic. Some partial explanations have been put forward, such as in-
016%1923/91/$03.50
0 199 1 - Elsevier Science Publishers B.V.
252
J.-L.B. KHALFAOUI
creased soi1 reflectivity and variations in the surface temperature of the oceans,
the latter being a result of modifications in ocean current dynamics.
It is now established that human activity, through overgrazing, deforesta-
tion and poor farming practices, plays a decisive role in the drought and de-
sertilïcation process. It seems that such adverse activity may lead to an irre-
versible deterioration in the Sahel.
Taking the case of the Senegal semi-arid region, which stretches northwards
to Gambia, this article sets out (i) to show that one of the major components
of the drought in the Sahel is the shortening of the rainy season, (ii) to ex-
amine the consequences of this on determining lengths of the trop growing
period, with the use of a simple study model.
DISCUSSION
The rainfall data used in this study have been provided by the Agroclima-
tology Service of ISRA. They represent 36 years, from 1953 to 1988, and have
been collected from ten different locations with very few missing values.
Reductions in annual rainfall
For Senegal, a comparison of the isohyet chart for the period 195 1- 1967
with that for the period 1968-1985 (Fig. 1) shows a north-south slide, cor-
responding to an overall reduction in rainfall (Diagne, 1988 ).
Figure 2 gives mean annual rainfall figures for ten sites representative of
the dry region in Senegal, for the years 1953-l 986. It is clear that the drop in
rainfall began around 1970. The inter-annual mean fell from 588 mm during
the ‘wet period’ to 4 10 mm during the ‘dry period’, i.e. a drop of 178 mm,
representing a 30% reduction. A statistical comparison of these two means
indicates that the difference is very highly signilïcant (PC 0.00 1). However,
a comparison of variantes for the two periods shows that they are not signif-
icantly different. Inter-annual variation is therefore the same for the two pe-
riods. This indicates that the drop in rainfall does not stem from the appear-
ance of extremely dry years, but rather from a general decrease in the level of
rainfall.
Reduction in the ‘useful length’ of the rainy season
In the semi-arid region in Senegal, rainfall data over 36 years ( 1953-1988)
for three locations (Louga in the north, Bambey in the central-nor-th region
and Nioro-du-Rip in the central-south region; Fig. 1) serve to illustrate the
phenomenon of the reduced ‘useful length’ of the rainy season. The so-called
‘useful length’ (Khalfaoui and Annerose, 1987) is delïned as the number of
days between ‘the first rain enabling sowing’ and ‘the last useful rainfall’.
POTENTIALLENGTHSOFCROPGROWINGPERIOD,SENEGAL
253
GUINEA-BISSAU
;
GUINEA
MAURITANIA
( 1968- 1985 1
I
656.
-Y----+
- +*+
- 700 =s
;/>,?j+
.---
- 1100 v'---
--C
.T
-++-,+
t
GUINEA-BISSAU
*
:a.
GUINEA
(Source: Bioclimatology Service (ISRA))
Fig. 1. Isohyet chart for Senegal for the periods 1951-1967 and 1968-1985 (mm).
254
J.-L.B. KHALFAOUI
M=499 mm
1
I
I I I I t
I
I I I t I
I
I 1 I I
1953 55
60
65
10
75
BO
85
Years
WET PERIOD
DRY PERIOD
Fig. 2. Mean annual rainfall for 10 locations representative of semi-arid regions in Senegal, for
years 1953-1986.
“The fïrst rain enabling sowing” (Forest and Dancette, 1982) is de-
fined, in accordance with the time that has elapsed since the possible
star-t of the rainy season, as the minimum amount of precipitation that
Will enable a trop to become well established if sowing is carried out
after this rain. This idea only concerns crops such as sorghum, ground-
nut and cowpea, which are sown just after rain, and not millet, which is
sown “dry” before the star-t of the rainy season. With this trop, the idea
of deciding when to sow depending on precipitation does not exist, and
that of the “first rain enabling sowing” corresponds to the ‘Yïrst success-
ful rainfall after dry sowing”. Since the chances of success are less the
earlier in the season sowing takes place, the minimum amount of rain in
the first fa11 enabling sowing Will be higher the earlier in the season the
grower wishes to sow. Table 1 was produced by Forest and Dancette
(1982).
“The last useful rain” is defïned as the last precipitation that partici-
pates in supplying water to a trop sown after the first rain enabling
sowing.
At Louga (Fig. 3 ), it is evident that the rainy season has shortened since
1970. A comparison of the mean useful length of rainy season for the years
prior to 1970 with those for the years after 1970, shows that this drop - 25
days on average - is very highly signifïcant (PC 0.00 1) . However, a compar-
ison of variante for season length during the two periods indicates that they
are not significantly different, and thus that the useful length of the rainy sea-
son is no more irregular after 1970 than before that date.
At Bambey (Fig. 4)) the reduction in the useful length of the rainy season
from 1970 onwards is signifïcant (Pc 0.05 ), and represents 12 days on aver-
POTENTIAL LENGTHS OF CROP GROWING PERIOD, SENEGAL
255
TABLE 1
Minimum amount of rain required to enable successful sowing, according to region and date in the
semi-arid region in Senegal
Site
Sowing period
Rain enabling sowing
(mm)
North region (Louga)
1-15 June
>Ci0
16-30 June
130
l-20 July
>20
20 July onwards
>15
Central-north region (Bambey )
1-15 June
>JO
16 June- 15 July
>20
15 July onwards
>15
Centre-south region (Nioro-du-Rip)
l-10 June
>40
11 June- 10 July
>20
10 July onwards
>15
After Forest and Dancette ( 1982).
Fig. 3. Useful length of the rainy season at Louga for the years 1953-1988.
age between the wet period and the dry period, without signifïcantly modify-
ing inter-annual regularity.
At Nioro-du-Rip (Fig. 5 ), a sharp and highly significant fa11 (P-C 0.1) in
256
J.-L.B. KHALFAOUI
I
99i3 ’ k ’ ’ ’ ’ 60 ’ ’ ’ ’ ’ 6’ 5 ’ ’ ’ 69 ‘i10 ’ ’ ’ ’ II ’ ’ ’ ’ 80 ’ ’ ’ ’ B5 ’ ’
YearS ’
WET PERIOD
DRY PERIOD
Fig. 4. Useful length of the rainy season at Bambey for the years 1953-1988.
k5~'r,111'~~*1t,1'1',,'t,',",,,",,1",'j
6 5
69 10
1 5
80
85
YSU-S
WET PERIOD
DRY PERIOD
(*):NO figure aveible
Fig. 5. Useful length of the rainy season at Nioro-du-Rip for the years 1953-1988.
the useful length of the rainy season is seen from 1970 onwards. The drop is
12 days on average, and does not signifkantly modify inter-annual regularity.
POTENTIAL LENGTHS OF CROP GROWING PERIOD. SENEGAL
251
Betermining the optimum length of the trop growing period
Principle
The reduction in the useful length of the rainy season means that the length
of the growing period for crops grown in the region has to be revised. A simple
method, based on bioclimatological data, enables an optimum growing pe-
riod length for each region to be determined. It is suffïciently general to be
applied to a11 crops, and cari be adapted to each species by oupling it with a
mode1 simulating trop water balance.
One basic principle should be retained: it is generally acknowledged that a
reduction in the cycle length of cultivars leads to a fa11 in production poten-
tial. In order to optimise the trop, the length of the cycle should coincide as
closely as possible with that of the rainy season.
The main constraint lies in the marked inter-annual fluctuation in the length
of the rainy season, which makes it diffïcult to determine the length of grow-
ing period to be adopted. In the case of useful lengths of the rainy season over
the period 1953- 1988 for the towns of Louga, Bambey and Nioro (see Figs.
3-5 ), the coefficients of variation are 24%, 18% and 13%, respectively. It cari
be seen that as we progress northwards, rainfall totals decrease and the inter-
annual variations increase. This constitutes one of the components of the in-
creased climatic risk in the northern regions. Optimisation is therefore nec-
essary, and consists of identifying the proportion of years in which the grow-
ing period would fit into the rainy season, which is equivalent to choosing the
probability of success. This choice should first and foremost take account of
the following two economic factors.
( 1) On an individual level, the economic situation of farmers in develop-
ing countries is characterised by the lack of inter-annual financial resources,
preventing them from speculating on ventures with a fairly high risk of fail-
ure. Some of these ventures would yield a higher overall productivity, but
would involve large fluctuations in annual income, particularly if national
agricultural development policy is based on agricultural credit, meaning that
they have to make repayments on loans.
(2) On a national level, market stability, hence that of production levels,
is an economic priority for a11 countries, especially developing countries, where
agricultural product processing, distribution and marketing networks, al-
ready in existence or yet to be set up, need to be safeguarded, owing to their
‘fragility’.
These two economic factors mean that a low risk of failure which may be
brought about by sufficiently short growing periods, should be ensured. This
Will also mean that the farmer is able to safeguard his trop by guarding against
‘false starts’ to the rainy season, i.e. sowing after the tïrst heavy fa11 of rain,
early in the season, which may be followed by a waiting period before the
258
J.-L.B. KHALFAOUI
rainy season really begins (Forest and Dancette, 1982). This implies taking
a ‘safety margin’ into account in the determination of sowing time.
The length of the growing period should also take account of the farming
calendar at the beginning and the end of the rainy season, which represents a
period of heavy workload for the farmer. Cultivars with a cycle length slightly
shorter than the rainy season would fit most years, and provide work flexibil-
ity by enabling the farmer to stagger sowing and harvesting. Furthermore,
these cultivars would enable wet ploughing which, if applicable, is an effective
trop technique that conserves and makes best possible use of water (Dan-
cette, 1985). This approach demands a ‘degree of freedom’ with regard to the
choice of technique to be used.
Mode1
The ‘potential length of the trop growing period’ is defïned in the following
way: the useful length of the rainy season, plus the time, after the last useful
fa11 of rain, during which the trop is able to draw water from the soil’s re-
serves, minus 5 days, corresponding empirically to the ‘safety margin’ plus
the ‘degree of freedom’ with regard to technique.
The length of time during which the soil’s usable water reserves cari be used
is evaluated using the total rainfall over the previous month. During the
month, the daily evapotranspiration is fïxed at approximately 2 mm. This
value corresponds to the water supply needed at the end of the growing pe-
riod, calculated by the method of Fréteaud et al. ( 1984)) and weighted by an
index of minimal satisfaction equal to 50%, compatible with a satisfying ter-
mination of the trop. Therefore, the consumption is 60 mm for the month as
a whole. Total rainfall over the month minus 60 mm gives an estimate of
available water reserves, which are used at a rate of 3.5 mm day-‘, with evap-
oration increasing after rainfall. This value is calculated according to the
method described above.
Given that a is the useful length of the rainy season (days), b is number of
days during which the trop cari draw on soi1 water reserves and c is total rain-
fa11 over the previous month of the rainy season, then: Potential length of the
trop growing period = a + b - 5 (in days )
where
b= c~~om~m(days)
Given the flexibility that some crops show, it is acknowledged that there
may be no adverse effects on production when, for a given year, the potential
length of the growing period is up to 5 days shorter than the theoretical dura-
tion of the trop.
POTENTIAL LENGTHS OF CROP GROWING PERIOD, SENEGAL
259
Application to the case of Senegal
The potential length of the growing period was determined for each year in
the period from 1953 to 1988, for each of the three locations in the semi-arid
region in Senegal (Figs. 6-8 ) .
From 1970 onwards, a reduction in the potential length of the growing pe-
riod was seen for a11 three localities: at Louga there was a reduction of 26 days
(P-C 0.00 1); at Bambey a reduction of 16 days (P-C 0.05); at Nioro-du-Rip a
reduction of 17 days (PC 0.05 ).
Table 2 shows the variability in the length of the growing period in the three
regions of the semi-arid zone in Senegal during the wet period ( 1953-l 969)
and the dry period ( 1970- 1988 ). It reveals the marked reduction in the lengths
of the growing period in each region, caused by the drought from the end of
the 1960s onwards.
The rainfall data for 14 locations over the whole of the semi-arid region in
Senegal enable a map to be drawn up of inter-annual average potential grow-
ing period length during the dry period of 1970- 1986 (Fig. 9). This does not,
however, make it possible to Select a value for the length of growing period
which is effective in the various regions, as it does not include the probability
of success, which takes account of inter-annual variations.
However, the choice of potential length cari be based on the map of the
lengths of growing period satisfïed in at least 80% of years, i.e. 14 out of the
17 years in the dry period (Fig. 10).
10 -1
4
,
I
, , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,
%k- 5 5
6 0
6 5
69 10
15
80
85 Year
W E T PKRIO”
DRY PEHIOD
Fig. 6. Potential length of the trop growing period at Louga for the years 1953-1988.
260
J.-L.B. KHALFAOUI
‘rection of the yeers il ” 1rhich cycle
is satisfied
1
I
--
O,,
--- -- -_
--
-- TII--
-.~
- -------ï17i
10 -
[ , , , , , , , , , , , ,
I
,
F
~
1911 II
60
61'
0
II
-or i
WET PERIOD
DRY PERIOD
Fig. 7. Potential length of the trop growing period at Bambey for the years 1953-1988.
Fraction of the years in which cyc
is natisfied
~Q-E-ly__-__-----__--_-___‘6~!
Ii- , , , , , , , , , , , , , ( , , ,
1951 II
OI*
61
69
WET PERIOD
DRY PERIOD
(*):No figure avaible
Fig. 8. Potential length of the trop growing period at Nioro-du-Rip for the years 1953-l 988.
POTENTIAL LENGTHS OF CROP GROWING PERIOD, SENEGAL
261
TABLE 2
Adaptability of various lengths of trop growing period for three regions in the semi-arid zone of
Senegal. Data are given for the periods 1953-1969 and 1970-1988
Region
Length of trop
Percentage ofyears in which length of
growing period
trop growing period fits into the
1
(Ws)
potential length
1953-1969
1970-1988
North (Louga)
120
29
0
110
35
5
90
82
26
75
100
68
70
100
84
60
100
84
Central-north
120
59
26
(Bambey )
110
71
37
95
88
84
90
94
89
Central-south
120
75
63
( Nioro-du-Rip )
110
100
74
105
100
79
90
100
95
MAUHI1
.++++
GUINEA-BISSAU ’
Fig. 9. Map of inter-annual means for potential lengths of the trop growing periods for the
period 1970-l 986 in the semi-arid region of Senegal (days).
u
262
J . - L . B . K H A L F A O U I
105
t+
???? ? ?
????
?
GAHBIA +++++++++
++. .
GUINEA-BISSA” :
Fig. 10. Map of lengths of the trop growing period satisfied in at least 80% of years for the period
1970- 1986 in the semi-arid region of Senegal (days).
Implications for the case ofplant breeding
The reduction in the length of the rainy season makes a certain number of
currently extended cultivars poorly adapted to these regions, as their cycle
lengths have become too long. This means that targets for new cultivars to be
obtained either by varietal creation or by the introduction of already existing
genotypes or new crops, need to be revised. This is illustrated by the case of
groundnut, cultivated in the Louga region. The cultivar released has a short
cycle (90 days). Examination of potential lengths of the trop growing period
for the years 1953- 1988 (Fig. 6 ) indicates that this cultivar was able to reach
the end of its growing period in 14 out of 17 years during the wet period but
only 5 out of 19 years during the dry period.
Determining potential cycles using rainfall data means that the best possi-
ble choice of cycle length cari be made, adapted to each region, whereas this
choice is usually empiric.
Taking a safety margin regarding sowing date and degree of freedom of
choice of technique into account, in this case 5 days, means that the following
cari be allowed for: inter-annual variations in the length of the rainy season
and the farmer’s calendar, and that effective farming techniques to control
drought cari be used, such as ploughing at the end of the cycle.
*
POTENTIAL LENGTHS OF CROP GROWING PERIOD, SENEGAL
263
ACKNOWLEDGEMENT
Support for this project was received from the EEC STD II program. The
author is a Research Engineer with CIRAD/IRHO, on secondment to the
Institut Sénégalais de Recherches Agricoles.
REFERENCES
Dancette, C., 1985. Contrariétés pédoclimatiques et adaptation de l’agriculture à la sécheresse
en zone intertropicale. In: Actes Colloque Résistance à la Sécheresse en Milieu Intertropical,
September 1984, Dakar, Conseil International de la Langue Français, Paris, pp. 27-42.
Diagne, M., 1988. Evolution Historique de la Pluviometrie au Sénégal: Synthése des Données
1950-87, ISRA, Centre National de Recherches Agricoles, Bambey, 108 pp.
Forest, F. and Dancette, C., 1982. Situation du bilan hydrique de l’arachide en vue d’une meil-
leure adaptation de cette culture aux conditions tropicales. In: Actes Symposium Interna-
tional sur la Production Arachidière, le Marché Mondial des Oléagineux, les Échanges lntra-
africains de YArachide et ses Dérivés, June 1982, Bandjul.
Fréteaud, J.-P., Lidon, B. and Marlet, S., 1984. La détermination des coefficients culturaux en
zone Soudano Sahélienne. Proposition d’une méthode générale et pratique. Institut de Re-
cherches Agronomiques Tropical (IRAT)-Division Economie et Valorisation de l’eau (DE),
Montpellier, 28 pp.
Khalfaoui, J.-L. and Annerose, D., 1986. Création variétale d’arachide adaptée aux contraintes
pluviométriques des zones semi-arides. In: Agrometeorology of groundnut. Proc. Int. Symp.,
2 l-26 August 1985, ICRISAT Sahelian Center, Niamey, Niger. ICRISAT, Patancheru.
UNESCO, 1984. Climate, drought and desertification. Nature and Resources 20( 1 ), pp. 2-8.
Elsevier Science Publishers B.V., Amsterdam
çkle
14 i-L&
Determination of potential lengths of the trop
growing period in semi-arid regions of Senegal
J.-L.B. Khalfaoui
Institut Sénégalais de Recherches Agricoles (ISRA), CNRA, Bambey, Senegal
(Received 8 August 199’0; revision accepted 19 November 1990)
‘4BSTRACT
Khalfaoui, J.-L.B., 199 1. Determination of potential lengths of the trop growing period in semi-arid
regions of Senegal. Agric. For. Meteorol., 55: 251-263.
A study of rainfall data from the Senegal semi-arid region shows that one of the major components
of the deterioration in rainfall in the Sahel is the very marked reduction in the length of the rainy
season since 1970. This reduction means that lengths of the trop growing period have to be reviewed
for each region. A study method is described which takes into account the inherent climatic, economic
and technical constraints in Sahelian agriculture. It enables potential lengths of the growing period to
be determined, adapted to each region. Among the various means used to study drought, this method,
in particular, enables the setting of plant breeding targets. It has been applied to the case of the Senegal
semi-arid region.
INTRODUCTION
Rainfall totals since the start of the 1940s show that in India, the drought
is the result of signifïcant rainfall fluctuations from year to year, with no over-
a11 drop. However, for the Sudan-Sahel region of Africa and for Australia, a
tendency towards lower rainfall began during the 1950s with an apparent
stabilisation from 1972- 1973 onwards. This also applies to North and South
hmerica, particularly northeastern Brazil, since 1945, and, more recently, for
eastern and southern Africa (UNESCO, 1984).
In Africa, there is statistical evidence for a degree of periodicity as regards
series of rainfall figures. The previous two dry periods occurred during the
1910s and 1940s both of which lasted for a shorter time than the current
drought. Nonetheless,.this periodicity is of no value in forecasting, owing to
random phase modifications of these chronological series (UNESCO, 1984).
The dynamic causes of drought are little known, as climatologists still have
great diffculties distinguishing causes from correlated effects which may be
synergistic. Some partial explanations have been put forward, such as in-
016%1923/91/$03.50
0 199 1 - Elsevier Science Publishers B.V.
252
J.-L.B. KHALFAOUI
creased soi1 reflectivity and variations in the surface temperature of the oceans,
the latter being a result of modifications in ocean current dynamics.
It is now established that human activity, through overgrazing, deforesta-
tion and poor farming practices, plays a decisive role in the drought and de-
sertilïcation process. It seems that such adverse activity may lead to an irre-
versible deterioration in the Sahel.
Taking the case of the Senegal semi-arid region, which stretches northwards
to Gambia, this article sets out (i) to show that one of the major components
of the drought in the Sahel is the shortening of the rainy season, (ii) to ex-
amine the consequences of this on determining lengths of the trop growing
period, with the use of a simple study model.
DISCUSSION
The rainfall data used in this study have been provided by the Agroclima-
tology Service of ISRA. They represent 36 years, from 1953 to 1988, and have
been collected from ten different locations with very few missing values.
Reductions in annual rainfall
For Senegal, a comparison of the isohyet chart for the period 195 1- 1967
with that for the period 1968-1985 (Fig. 1) shows a north-south slide, cor-
responding to an overall reduction in rainfall (Diagne, 1988 ).
Figure 2 gives mean annual rainfall figures for ten sites representative of
the dry region in Senegal, for the years 1953-l 986. It is clear that the drop in
rainfall began around 1970. The inter-annual mean fell from 588 mm during
the ‘wet period’ to 4 10 mm during the ‘dry period’, i.e. a drop of 178 mm,
representing a 30% reduction. A statistical comparison of these two means
indicates that the difference is very highly signilïcant (PC 0.00 1). However,
a comparison of variantes for the two periods shows that they are not signif-
icantly different. Inter-annual variation is therefore the same for the two pe-
riods. This indicates that the drop in rainfall does not stem from the appear-
ance of extremely dry years, but rather from a general decrease in the level of
rainfall.
Reduction in the ‘useful length’ of the rainy season
In the semi-arid region in Senegal, rainfall data over 36 years ( 1953-1988)
for three locations (Louga in the north, Bambey in the central-nor-th region
and Nioro-du-Rip in the central-south region; Fig. 1) serve to illustrate the
phenomenon of the reduced ‘useful length’ of the rainy season. The so-called
‘useful length’ (Khalfaoui and Annerose, 1987) is delïned as the number of
days between ‘the first rain enabling sowing’ and ‘the last useful rainfall’.
POTENTIALLENGTHSOFCROPGROWINGPERIOD,SENEGAL
253
GUINEA-BISSAU
;
GUINEA
MAURITANIA
( 1968- 1985 1
I
656.
-Y----+
- +*+
- 700 =s
;/>,?j+
.---
- 1100 v'---
--C
.T
-++-,+
t
GUINEA-BISSAU
*
:a.
GUINEA
(Source: Bioclimatology Service (ISRA))
Fig. 1. Isohyet chart for Senegal for the periods 1951-1967 and 1968-1985 (mm).
254
J.-L.B. KHALFAOUI
M=499 mm
1
I
I I I I t
I
I I I t I
I
I 1 I I
1953 55
60
65
10
75
BO
85
Years
WET PERIOD
DRY PERIOD
Fig. 2. Mean annual rainfall for 10 locations representative of semi-arid regions in Senegal, for
years 1953-1986.
“The fïrst rain enabling sowing” (Forest and Dancette, 1982) is de-
fined, in accordance with the time that has elapsed since the possible
star-t of the rainy season, as the minimum amount of precipitation that
Will enable a trop to become well established if sowing is carried out
after this rain. This idea only concerns crops such as sorghum, ground-
nut and cowpea, which are sown just after rain, and not millet, which is
sown “dry” before the star-t of the rainy season. With this trop, the idea
of deciding when to sow depending on precipitation does not exist, and
that of the “first rain enabling sowing” corresponds to the ‘Yïrst success-
ful rainfall after dry sowing”. Since the chances of success are less the
earlier in the season sowing takes place, the minimum amount of rain in
the first fa11 enabling sowing Will be higher the earlier in the season the
grower wishes to sow. Table 1 was produced by Forest and Dancette
(1982).
“The last useful rain” is defïned as the last precipitation that partici-
pates in supplying water to a trop sown after the first rain enabling
sowing.
At Louga (Fig. 3 ), it is evident that the rainy season has shortened since
1970. A comparison of the mean useful length of rainy season for the years
prior to 1970 with those for the years after 1970, shows that this drop - 25
days on average - is very highly signifïcant (PC 0.00 1) . However, a compar-
ison of variante for season length during the two periods indicates that they
are not significantly different, and thus that the useful length of the rainy sea-
son is no more irregular after 1970 than before that date.
At Bambey (Fig. 4)) the reduction in the useful length of the rainy season
from 1970 onwards is signifïcant (Pc 0.05 ), and represents 12 days on aver-
POTENTIAL LENGTHS OF CROP GROWING PERIOD, SENEGAL
255
TABLE 1
Minimum amount of rain required to enable successful sowing, according to region and date in the
semi-arid region in Senegal
Site
Sowing period
Rain enabling sowing
(mm)
North region (Louga)
1-15 June
>Ci0
16-30 June
130
l-20 July
>20
20 July onwards
>15
Central-north region (Bambey )
1-15 June
>JO
16 June- 15 July
>20
15 July onwards
>15
Centre-south region (Nioro-du-Rip)
l-10 June
>40
11 June- 10 July
>20
10 July onwards
>15
After Forest and Dancette ( 1982).
Fig. 3. Useful length of the rainy season at Louga for the years 1953-1988.
age between the wet period and the dry period, without signifïcantly modify-
ing inter-annual regularity.
At Nioro-du-Rip (Fig. 5 ), a sharp and highly significant fa11 (P-C 0.1) in
256
J.-L.B. KHALFAOUI
I
99i3 ’ k ’ ’ ’ ’ 60 ’ ’ ’ ’ ’ 6’ 5 ’ ’ ’ 69 ‘i10 ’ ’ ’ ’ II ’ ’ ’ ’ 80 ’ ’ ’ ’ B5 ’ ’
YearS ’
WET PERIOD
DRY PERIOD
Fig. 4. Useful length of the rainy season at Bambey for the years 1953-1988.
k5~'r,111'~~*1t,1'1',,'t,',",,,",,1",'j
6 5
69 10
1 5
80
85
YSU-S
WET PERIOD
DRY PERIOD
(*):NO figure aveible
Fig. 5. Useful length of the rainy season at Nioro-du-Rip for the years 1953-1988.
the useful length of the rainy season is seen from 1970 onwards. The drop is
12 days on average, and does not signifkantly modify inter-annual regularity.
POTENTIAL LENGTHS OF CROP GROWING PERIOD. SENEGAL
251
Betermining the optimum length of the trop growing period
Principle
The reduction in the useful length of the rainy season means that the length
of the growing period for crops grown in the region has to be revised. A simple
method, based on bioclimatological data, enables an optimum growing pe-
riod length for each region to be determined. It is suffïciently general to be
applied to a11 crops, and cari be adapted to each species by oupling it with a
mode1 simulating trop water balance.
One basic principle should be retained: it is generally acknowledged that a
reduction in the cycle length of cultivars leads to a fa11 in production poten-
tial. In order to optimise the trop, the length of the cycle should coincide as
closely as possible with that of the rainy season.
The main constraint lies in the marked inter-annual fluctuation in the length
of the rainy season, which makes it diffïcult to determine the length of grow-
ing period to be adopted. In the case of useful lengths of the rainy season over
the period 1953- 1988 for the towns of Louga, Bambey and Nioro (see Figs.
3-5 ), the coefficients of variation are 24%, 18% and 13%, respectively. It cari
be seen that as we progress northwards, rainfall totals decrease and the inter-
annual variations increase. This constitutes one of the components of the in-
creased climatic risk in the northern regions. Optimisation is therefore nec-
essary, and consists of identifying the proportion of years in which the grow-
ing period would fit into the rainy season, which is equivalent to choosing the
probability of success. This choice should first and foremost take account of
the following two economic factors.
( 1) On an individual level, the economic situation of farmers in develop-
ing countries is characterised by the lack of inter-annual financial resources,
preventing them from speculating on ventures with a fairly high risk of fail-
ure. Some of these ventures would yield a higher overall productivity, but
would involve large fluctuations in annual income, particularly if national
agricultural development policy is based on agricultural credit, meaning that
they have to make repayments on loans.
(2) On a national level, market stability, hence that of production levels,
is an economic priority for a11 countries, especially developing countries, where
agricultural product processing, distribution and marketing networks, al-
ready in existence or yet to be set up, need to be safeguarded, owing to their
‘fragility’.
These two economic factors mean that a low risk of failure which may be
brought about by sufficiently short growing periods, should be ensured. This
Will also mean that the farmer is able to safeguard his trop by guarding against
‘false starts’ to the rainy season, i.e. sowing after the tïrst heavy fa11 of rain,
early in the season, which may be followed by a waiting period before the
258
J.-L.B. KHALFAOUI
rainy season really begins (Forest and Dancette, 1982). This implies taking
a ‘safety margin’ into account in the determination of sowing time.
The length of the growing period should also take account of the farming
calendar at the beginning and the end of the rainy season, which represents a
period of heavy workload for the farmer. Cultivars with a cycle length slightly
shorter than the rainy season would fit most years, and provide work flexibil-
ity by enabling the farmer to stagger sowing and harvesting. Furthermore,
these cultivars would enable wet ploughing which, if applicable, is an effective
trop technique that conserves and makes best possible use of water (Dan-
cette, 1985). This approach demands a ‘degree of freedom’ with regard to the
choice of technique to be used.
Mode1
The ‘potential length of the trop growing period’ is defïned in the following
way: the useful length of the rainy season, plus the time, after the last useful
fa11 of rain, during which the trop is able to draw water from the soil’s re-
serves, minus 5 days, corresponding empirically to the ‘safety margin’ plus
the ‘degree of freedom’ with regard to technique.
The length of time during which the soil’s usable water reserves cari be used
is evaluated using the total rainfall over the previous month. During the
month, the daily evapotranspiration is fïxed at approximately 2 mm. This
value corresponds to the water supply needed at the end of the growing pe-
riod, calculated by the method of Fréteaud et al. ( 1984)) and weighted by an
index of minimal satisfaction equal to 50%, compatible with a satisfying ter-
mination of the trop. Therefore, the consumption is 60 mm for the month as
a whole. Total rainfall over the month minus 60 mm gives an estimate of
available water reserves, which are used at a rate of 3.5 mm day-‘, with evap-
oration increasing after rainfall. This value is calculated according to the
method described above.
Given that a is the useful length of the rainy season (days), b is number of
days during which the trop cari draw on soi1 water reserves and c is total rain-
fa11 over the previous month of the rainy season, then: Potential length of the
trop growing period = a + b - 5 (in days )
where
b= c~~om~m(days)
Given the flexibility that some crops show, it is acknowledged that there
may be no adverse effects on production when, for a given year, the potential
length of the growing period is up to 5 days shorter than the theoretical dura-
tion of the trop.
POTENTIAL LENGTHS OF CROP GROWING PERIOD, SENEGAL
259
Application to the case of Senegal
The potential length of the growing period was determined for each year in
the period from 1953 to 1988, for each of the three locations in the semi-arid
region in Senegal (Figs. 6-8 ) .
From 1970 onwards, a reduction in the potential length of the growing pe-
riod was seen for a11 three localities: at Louga there was a reduction of 26 days
(P-C 0.00 1); at Bambey a reduction of 16 days (P-C 0.05); at Nioro-du-Rip a
reduction of 17 days (PC 0.05 ).
Table 2 shows the variability in the length of the growing period in the three
regions of the semi-arid zone in Senegal during the wet period ( 1953-l 969)
and the dry period ( 1970- 1988 ). It reveals the marked reduction in the lengths
of the growing period in each region, caused by the drought from the end of
the 1960s onwards.
The rainfall data for 14 locations over the whole of the semi-arid region in
Senegal enable a map to be drawn up of inter-annual average potential grow-
ing period length during the dry period of 1970- 1986 (Fig. 9). This does not,
however, make it possible to Select a value for the length of growing period
which is effective in the various regions, as it does not include the probability
of success, which takes account of inter-annual variations.
However, the choice of potential length cari be based on the map of the
lengths of growing period satisfïed in at least 80% of years, i.e. 14 out of the
17 years in the dry period (Fig. 10).
10 -1
4
,
I
, , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,
%k- 5 5
6 0
6 5
69 10
15
80
85 Year
W E T PKRIO”
DRY PEHIOD
Fig. 6. Potential length of the trop growing period at Louga for the years 1953-1988.
260
J.-L.B. KHALFAOUI
‘rection of the yeers il ” 1rhich cycle
is satisfied
1
I
--
O,,
--- -- -_
--
-- TII--
-.~
- -------ï17i
10 -
[ , , , , , , , , , , , ,
I
,
F
~
1911 II
60
61'
0
II
-or i
WET PERIOD
DRY PERIOD
Fig. 7. Potential length of the trop growing period at Bambey for the years 1953-1988.
Fraction of the years in which cyc
is natisfied
~Q-E-ly__-__-----__--_-___‘6~!
Ii- , , , , , , , , , , , , , ( , , ,
1951 II
OI*
61
69
WET PERIOD
DRY PERIOD
(*):No figure avaible
Fig. 8. Potential length of the trop growing period at Nioro-du-Rip for the years 1953-l 988.
POTENTIAL LENGTHS OF CROP GROWING PERIOD, SENEGAL
261
TABLE 2
Adaptability of various lengths of trop growing period for three regions in the semi-arid zone of
Senegal. Data are given for the periods 1953-1969 and 1970-1988
Region
Length of trop
Percentage ofyears in which length of
growing period
trop growing period fits into the
1
(Ws)
potential length
1953-1969
1970-1988
North (Louga)
120
29
0
110
35
5
90
82
26
75
100
68
70
100
84
60
100
84
Central-north
120
59
26
(Bambey )
110
71
37
95
88
84
90
94
89
Central-south
120
75
63
( Nioro-du-Rip )
110
100
74
105
100
79
90
100
95
MAUHI1
.++++
GUINEA-BISSAU ’
Fig. 9. Map of inter-annual means for potential lengths of the trop growing periods for the
period 1970-l 986 in the semi-arid region of Senegal (days).
u
262
J . - L . B . K H A L F A O U I
105
t+
???? ? ?
????
?
GAHBIA +++++++++
++. .
GUINEA-BISSA” :
Fig. 10. Map of lengths of the trop growing period satisfied in at least 80% of years for the period
1970- 1986 in the semi-arid region of Senegal (days).
Implications for the case ofplant breeding
The reduction in the length of the rainy season makes a certain number of
currently extended cultivars poorly adapted to these regions, as their cycle
lengths have become too long. This means that targets for new cultivars to be
obtained either by varietal creation or by the introduction of already existing
genotypes or new crops, need to be revised. This is illustrated by the case of
groundnut, cultivated in the Louga region. The cultivar released has a short
cycle (90 days). Examination of potential lengths of the trop growing period
for the years 1953- 1988 (Fig. 6 ) indicates that this cultivar was able to reach
the end of its growing period in 14 out of 17 years during the wet period but
only 5 out of 19 years during the dry period.
Determining potential cycles using rainfall data means that the best possi-
ble choice of cycle length cari be made, adapted to each region, whereas this
choice is usually empiric.
Taking a safety margin regarding sowing date and degree of freedom of
choice of technique into account, in this case 5 days, means that the following
cari be allowed for: inter-annual variations in the length of the rainy season
and the farmer’s calendar, and that effective farming techniques to control
drought cari be used, such as ploughing at the end of the cycle.
*
POTENTIAL LENGTHS OF CROP GROWING PERIOD, SENEGAL
263
ACKNOWLEDGEMENT
Support for this project was received from the EEC STD II program. The
author is a Research Engineer with CIRAD/IRHO, on secondment to the
Institut Sénégalais de Recherches Agricoles.
REFERENCES
Dancette, C., 1985. Contrariétés pédoclimatiques et adaptation de l’agriculture à la sécheresse
en zone intertropicale. In: Actes Colloque Résistance à la Sécheresse en Milieu Intertropical,
September 1984, Dakar, Conseil International de la Langue Français, Paris, pp. 27-42.
Diagne, M., 1988. Evolution Historique de la Pluviometrie au Sénégal: Synthése des Données
1950-87, ISRA, Centre National de Recherches Agricoles, Bambey, 108 pp.
Forest, F. and Dancette, C., 1982. Situation du bilan hydrique de l’arachide en vue d’une meil-
leure adaptation de cette culture aux conditions tropicales. In: Actes Symposium Interna-
tional sur la Production Arachidière, le Marché Mondial des Oléagineux, les Échanges lntra-
africains de YArachide et ses Dérivés, June 1982, Bandjul.
Fréteaud, J.-P., Lidon, B. and Marlet, S., 1984. La détermination des coefficients culturaux en
zone Soudano Sahélienne. Proposition d’une méthode générale et pratique. Institut de Re-
cherches Agronomiques Tropical (IRAT)-Division Economie et Valorisation de l’eau (DE),
Montpellier, 28 pp.
Khalfaoui, J.-L. and Annerose, D., 1986. Création variétale d’arachide adaptée aux contraintes
pluviométriques des zones semi-arides. In: Agrometeorology of groundnut. Proc. Int. Symp.,
2 l-26 August 1985, ICRISAT Sahelian Center, Niamey, Niger. ICRISAT, Patancheru.
UNESCO, 1984. Climate, drought and desertification. Nature and Resources 20( 1 ), pp. 2-8.