Local level adaptation strategies: case study of the 2001/2002 drought
in Buhera district, Zimbabwe1
1Patrick Mushove, 2Elliot Thomas and 3Dominick Kwesha
(Pvt) Ltd, Office 802 Eighth Floor, Robinson House, P.O. Box UA 526, Union Avenue,
Harare, Zimbabwe. patrickmushove@teledata.mz
2 Department of Meteorological Services, P.O. Box BE 150, Belvedere, Harare, Zimbabwe.
ethomas@weather.utande.co.zw
3Forest Research Centre, P.O. Box HG 595, Highlands, Harare, Zimbabwe. dkwesha@frchigh.co.zw
1ENVIRONMASY
INTRODUCTION
AIACC Project AF38 has included in its plan of work analyses of case studies in the four countries
(Malawi, Mozambique, Zambia and Zimbabwe) covered by that project. The idea of case studies
was endorsed by participants from Project AF38 during the Nairobi Kick-Off Meeting in early
February 2002. This presentation is the result of that endorsement.
Buhera District is located in south central Zimbabwe, in the eastern province of Manicaland.
Manicaland province has the highest population density in rural Zimbabwe, estimated at 49
persons km-2 (Central Statistical Office, 1997). Buhera is 100% communal land, which means that
there are no commercial farms or conservation areas in the whole district. The district is divided
into 32 administrative units called wards. Each ward is represented by a popularly elected
councillor who sits in the District Council Chambers. Buhera District’s eastern boundary is the
river Save (see Figure 1), which is an important water course not only in Zimbabwe but in
neighbouring Mozambique as well where its flooded middle and lower courses have caused havoc
in recent times. The district is divided into three agro-ecological zones, with region V being the
most arid category in Zimbabwe and region III somewhat suitable for rain-fed agriculture. Like
the rest of Zimbabwe, the 2001/2002 agricultural season has been the worst since 1947.
PROJECT DESIGN AND METHODS
We conducted literature search on droughts and drought-related disasters during the past 100
years in Zimbabwe. We also collated rainfall data from Buhera rainfall station for the period
1960/1961 to 2000/2001. Available data for the 2001/2002 season only go up to March 2002.
During the period 11 to 15 March, 2002, a team of five enumerators served a 20-question semistructured questionnaire to 214 households selected randomly from 11 of the 32 wards that
make up Buhera District. The interview venues were opportunistically selected to coincide with
business centres where drought relief maize was being sold on the particular day. The data were
coded and captured onto spreadsheet. This is one of the first case studies of its kind under
Project AF38 and we hope that it will stimulate improvements in formulating and designing the
many other case studies still to be conducted under this project.
OUTCOMES
Rainfall Anomalies
By the end of March 2002 Buhera had received 426.2 mm of rain, or 55% of normal (based on a
41-year average from the 60/61 season to the 2000/2001 season. However, the final percentage
is likely to exceed 80 due to the abnormally high rainfall received in April. The normal decadal
rainfall graph for Buhera is slightly bi-modal with a dip in early to mid January and two peaks,
1
Working Paper, data still under analysis. Comments are most welcome.
1
one in mid-December and the other towards the end of January (Figure 2). The absolute
differences between the normal January dip and the two peaks range from 32 to 37 mm.
This year (for Masvingo and Zaka, two rainfall stations nearest Buhera and for which all data for
the 2001/2002 season are available), both peaks have shifted to the left (mid November and
mid December) and the absolute differences between the January dip and the two new peaks
was between 150 and 300 mm for Zaka and between 100 and 180 mm for Masvingo (see Figure
3). As if that was not enough, Masvingo also received a third anomalous peak in mid April 2002
(a total of 83 mm of rain compared to the 10 mm normal expected for that period).
Local Knowledge Systems regarding rainfall and Droughts
The years seasons 82/83, 84/85, 92/93, 95/96, and 98/99 were classified by both the
respondents in the Buhera survey as well as by the OFDA/CRED International Disaster Database
as disaster years in terms of drought and famine in Zimbabwe. Besides these years, however, the
Buhera farmers also mentioned seven other seasons which were drought years. These were:
67/68, 72/73, 83/84, 85/86, 93/94, 94/95, and 97/98.
A closer look at Table 1 reveals that this second lot of drought seasons were genuine drought
seasons in terms of the amount of rainfall received expressed as a percentage of normal. This
should provide some homework as to who knows best which season was a drought year and
which season was not.
2
Dekadal Normal Rainfall for Buhera
80
70
Millimetres
60
50
40
30
20
Jun
May
Apr
Mar
Feb
Jan
Dec
Nov
Oct
Sep
Aug
0
Jul
10
Month
Figure 2. Ten-day normal rainfall distribution pattern for Buhera
Table 1. Local knowledge versus international drought and famine databases
Season
67/68
72/73
82/83
83/84
84/85
85/86
92/93
93/94
94/95
95/96
97/98
98/99
Mentioned by
Buhera Farmers?
Mentioned by
OFDA/CRED?
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
N
N
Y
N
Y
N
Y
N
N
Y
N
Y
Seasonal
Rainfall as %
of normal
51
46
51
76
137
114
97
71
56
112
81
113
One of the questions asked during the Buhera survey was “ What are the normal signs of a
drought year?” Twenty percent of the respondents said that winds blowing from the south and
bringing the early rains were a sure sign of an imminent drought year. We are still not quite clear
as to the significance of this response but we realize that it may be an interesting issue to study.
Another question sought to find out why farmers were convinced that the current drought was
worse than the 2000/2001 drought. A surprisingly high percentage (20%) of the respondents
said that the unstable political and economic situation in the country was worsening the current
drought. Our analysis of this response is that unfavourable political and economic circumstances
3
especially among the rural poor can become significant predisposing factors if a drought were to
strike.
Masvingo Dekadal Rainfall
Millimetres
200
150
100
50
Jun
May
Apr
Mar
Feb
Jan
Dec
Nov
Oct
Sep
Aug
Jul
0
Month
2001/2002 Season
Normal
Figure 3. Anomalous rainfall distribution pattern in Masvingo, 2001/2002 season.
Household Welfare and Mitigation Strategies
The average household in Buhera is made up of seven people. Fifty-seven percent of the
respondents classified their households into the 6-10 category. Land-holding is fairly low (1.6 ha
per household, with a range of 0.7 – 2.4 ha), besides most of the land (agro-ecological regions
IV and V) being marginal for rain-fed agriculture. Twenty-eight percent of the households
sampled had no cattle, 36.9% had between one and three cattle while 21% had between four
and six. When asked about the survival strategies they use in times of drought, 21% of the
respondents said that they purchased grain from the Grain Marketing Board (GMB), another
15.4% said they purchased from lucky neighbours who might have managed to harvest
something and 19.6% said that they resorted to selling vegetables and/or wild fruits, especially
Berchemia fruit, which fetches high prices in all urban centres. Still another 12.6% said they
resort to trading (buying and selling clothes and other commodities) while 9% said they may end
up selling their livestock and farm implements, thus disposing of some of their vital capital assets
for future seasons. Responding to a question on proposals to minimise the effects of droughts,
17.7% said that the government should give out soft loans in cash and in kind, 17.3% said that
the government should distribute food for free during drought years, 10.2% said that the
government should fund small-scale irrigation schemes and 9.8% said that the government
should seek help from other countries.
INTERIM CONCLUSIONS AND FOLLOW-UP WORK
In a seasonal rain-fed agricultural economy, it is not the cumulative annual rainfall that matters,
but rather the timing of the seasonal peaks, which should coincide with the critical period
(January and February) when crops are in their most vulnerable physiological state. In the face
4
of such rainfall anomalies as experienced in Zimbabwe this season, the choice of crop varieties
that mature early becomes most urgent. This anomaly is also responsible for upsetting even the
wild plants’ biological clocks: most Brachystegia trees in Zimbabwe’s miombo vegetation had
started producing new leaves in April instead of waiting until September.
In the medium term, the implication is that small-scale schemes for harvesting water should take
priority Such small water reservoirs would become an invaluable safety measure for small-scale
intensive production units (e.g. family or village gardens) to compensate for traditional crop
failures. In normal seasons, the same production units would boost the yields, and hence income
per household.
Future work on this case study will include analysis of decadal monthly rainfall trends over a
number of drought years in an effort to test further the hypothesis that decadal trends are the
most reliable variable for use in early warning systems. Another aspect still to be pursued is the
vulnerability ranking or sensitivity analysis of different categories of households (in function of
size of household, wealth or poverty index, access to major production factors and range of offfarm income sources).
********************************************************************
5