Adaptation in a Changing Climate: Practices of
Adaptation in a Changing Climate:
Practices of Smallholder Farmers in Uganda
Environmental Studies A.B. 2015
Advisor: Dr. Leah VanWey
Institute at Brown for Environment and Society
Deputy Director of Research;
Associate Professor of Sociology
May 3, 2015
Dr. Leah VanWey
Adaptation in a Changing Climate: Practices of Smallholder Farmers
Purpose of Research:
As climate change takes place and the effects continue to spread and become more intense, a
focus on adaptation is an important consideration. Farmers, and especially smallholder farmers,
are some of the most susceptible to the effects of climate change. When the rain and weather
are unreliable and changing from year to year it is much more difficult to count on consistent or
normal yields. Uganda is one developing country that depends on smallholder farmers and is
experiencing the effects of climate change. Though Uganda is not a large country, there are
diverse climates, types of soil, farming techniques, crops, and other difficulties that farmers
encounter. Generally, there are two periods of rain and one dry season. Farmers have worked
with these conditions for their entire lives and learned from the generations before them how
to best farm in these conditions. We wanted to find out how much climate variation farmers
were experiencing, and if they were making changes in their practices in reaction to these
In order to address these research questions we used a large social survey data set collected in
Uganda, and gridded climate data on precipitation and temperature. The survey data were
analyzed using Stata and combined with the climate data. Climate data was mapped and
analyzed using ArcGIS, Excel, and Stata. A regression model was used to analyze the variables
and determine their significance in changing the use of slash and burn practices, herbicide and
pesticide use, and use of inorganic fertilizer.
Results, Conclusions, and Recommendations:
The results indicate that farmers are responding to variability in recent weather, though
primarily to changes in precipitation. This could be because farmers already adjusted to
changes in temperature since this was already a factor in 2003 (across the country 14-18 of the
core six months of the agricultural season had temperatures at least one standard deviation
above the long term average). This research could be greatly influenced by qualitative
interviews and further investment. Smallholder farmers in Uganda to have the ability to adapt
to climate change, and many are willing and already making changes in order to adapt. These
data and results could be much further analyzed, and could be well put to use by the Ugandan
government in supporting farmers in adaptation as climate change continues.
As climate change takes place and the effects continue to spread and become more
intense, a focus on adaptation is an important consideration. In the developed countries, where
many of the causes of climate change originate, there are available resources and funds for
adaptation. However, in many developing countries resources are not as readily available to
devote to finding adaptations for climate change effects. These are often the countries most at
risk for weather related disasters or general climate change effects because of their specific
locations (Roberts and Parks, 2007). Farmers, and especially smallholder farmers, are some of
the most susceptible to the effects of climate change. They rely on consistent seasons to know
when and how to best plant and harvest their crops. When the rain and weather are unreliable and
changing from year to year it is much more difficult to count on consistent or normal yields.
One developing country that depends on smallholder farmers and is experiencing the
effects of climate change is Uganda. Though Uganda is not a large country, there are diverse
climates, types of soil, farming techniques, crops, and difficulties that farmers encounter.
Generally, there are two periods of rain and one dry season. Farmers have worked with these
conditions for their entire lives and learned from the generations before them how to best farm in
these conditions. While weather changes on a smaller scale over the course of history are a
constant, the seasons are now changing much more rapidly.
This thesis brings together data from several sources to study the effect of the climate on
farmers throughout Uganda. The data set on the farmers includes data from 2003 and 2013 from
eight different districts in Uganda, and over one hundred small communities. The data include
extensive survey information on the communities, the farmers, their individual plots, and the
soil. Comparing the data between the two years will allow us to determine what, if any, were the
changes in the practices of the farmers. Combining this with climate data from the time period,
as well as using a baseline from the previous 50 years, will show what changes have occurred in
precipitation and temperature and how they are related to the practices of farmers. Use of
ArcGIS software will allow for specific data and mapping options with the climate data,
analyzing each community’s climate changes and linking them with social survey data. Logistic
regression analyses in Stata software then allow us to relate the farmer behavior to these climate
Uganda is an appropriate choice through which to study these topics in order to get a
better idea of the effects East Africa as a whole is facing. Many countries in sub-Saharan Africa
are experiencing very low farm productivity already, and as climate change continues, this will
continue to worsen (Omamo, 2005). Uganda has similarly experienced climate changes, and has
importantly seen different patterns of climate and climate change in different regions. Combining
information on practices within households over ten years with climate information for Uganda
over those ten years and including the historical climate information will allow for a
comprehensive overview of the effects that are already occurring and the possibilities for
adaptation and preparing for the future. This is only a small part of so much more work that
could be done with this dataset, especially if qualitative interview data were incorporated.
Adapting to anthropogenic climate change presents new and important challenges.
Climate change impacts increase daily in new ways and we need to respond to that in order to
adapt most efficiently. While everyone will be affected, there are certain groups whose
adaptation to climate change will be critical to the survival of others. One such group is
smallholder farmers. Smallholder farms are small farms, often less than one acre, which are
typically owned and operated by a family. In developing countries, the farmers are often
subsistence farmers and rely on the farm for their food. Many are also able to produce enough to
provide food for others and make a small income. More than 90% of Africa’s agricultural
production depends on subsistence or local food production; 65% of the population are
themselves smallholder farmers that depend on food and income from their farms (IFPRI, 2004).
Smallholders, while they may be easily adaptable in some ways, also face larger threats
and constraints unique or especially intense due to their situation. Smallholders are often
diversified across a wider variety of income sources or crops, but they often have low capital
with which to adjust to risk and changes (World Bank, 2007). This can be good if there is a heat
wave or frost or flood or drought, because some of their crops will be better adapted than others
and they have a better chance of some crops surviving even while others fail. However it can
also mean that if they lose crops, they lose a lot of their income and food source because the
farms are so small and they generally lack insurance. Smallholder farmers are located
predominately in rural areas, which means they face additional disadvantages (World Bank,
2007). Farms are widespread and roads are difficult to traverse, so it can be difficult for farmers
to get their food to markets and for banks to be able to finance or insure them when they are not
easily accessible. These smallholder farmers are essential to feeding the world in a changing
climate, but need to be able to adapt to the changing climate conditions in order to survive and
This paper focuses on the experiences of smallholder farmers in Uganda, called the
“Pearl of Africa” by Winston Churchill due to its bountiful and vast vegetation and variety of
species. The soil is fertile and water is widely available. It is in a tropical area along the equator
with two annual rainy seasons, one having higher rains than the other, and two shorter dry
periods. Temperature stays fairly high throughout the year. Africa as a whole relies heavily on
agriculture in the economy and labor force, and in Uganda in particular, 60% of the work force is
in the agricultural sector (FAO, 2003). Recent climate change could threaten livelihoods of this
large population. Over the entire country, according to the climate data we studied, precipitation
has increased by 3% and temperature by 6% in the last decade relative to the second half of the
twentieth century. With climate change, the reliable wet and dry season system is fluctuating and
upsetting the normal decisions farmers make on planting times and which crops to grow (World
Smallholder Farmers & Rural Livelihoods
Agriculture has changed dramatically in the last fifty years. The changes post-World War
II and industrialization have revolutionized the way we farm, eat, and distribute food in much of
the world. Especially, in developing countries such as the United States, this has meant a
decrease in the number of farmers and the number of farms, and a large increase in the size of
farms (USDA, 2014). However, this is not a phenomenon that occurred everywhere or to every
farmer. As of 2013, the World Bank reports that 47% of the world’s population lives in a rural
area (World Bank, 2013). The World Development Report breaks countries into three groups
based on their agricultural systems – “one agriculture-based, one transforming, one urbanized”
(World Bank, 2007). It is the agriculture-based countries that continue to be dominated by
smallholder farmers and rely on them to feed the population.
Smallholders are a unique group in their countries as they live in rural areas, and often in
poverty, but own or can work on farming property. While most smallholders have little access to
capital and thus less ability to invest in risky new technologies or crops than do larger farmers,
they often use non-market methods for insuring against risks, including climate-related risks
(World Bank, 2007). They often grow multiple crops instead of just one. This means that if an
unexpected frost hits, or prolonged drought, or flooding, or a certain pest, then there is a higher
diversity of crops planted and a better chance that some of them will survive.
However, they also face a number of constraints. Many advantages that farmers in
developing countries take for granted are unavailable to smallholder farmers – food storage,
easily accessible markets and roads, access to loans, insurance, and credit, and use of modern
technologies and inputs such as nonorganic fertilizer (Morton, 2007). Smallholder farmers are
normally unable to store excess crops and sell them when prices are better or when food is
needed more because they do not have this infrastructure. Increasing populations and land
fragmentation can stress the land and soil, market expansion can create threats in competition,
and a lack of insurance makes them vulnerable to crop disease epidemics (Morton 2007). They
can pool resources with neighbors when possible, but this can be difficult if a shock affects the
whole community (World Bank, 2007). A great way for smallholders to increase income and
diversify their risks is to expand to off farm activities or even to on farm activities that involve
livestock or products such as honey (Peck and Pearce 2005). Research shows that a larger
percentage of income from off-farm activities is positively correlated with a higher total income
(Ellis, 1999). Yet this can be hard to implement without some outside help because most
smallholder families have low levels of education because the whole family works on the farm.
Increased education in the family could influence off farm activities and income strategies
There are many innovative ways for smallholder farmers to invest more in their land and
become less vulnerable to climate change; however many of these require upfront costs that most
farmers cannot make. Financial resources can be and are often an obstacle for smallholders in
their investments and decisions that they make in taking care of their land. Rural areas have a
much harder time accessing credit because banks encounter higher financial costs and constraints
in these areas due to rural locations and related risks (Kloeppinger-Todd and Sharma, 2010).
Farmers may be unable to take out loans because they do not qualify, or because they decide that
the collateral is not worth the risk – what Boucher et al. call “risk-rationing” (2008). Problems of
land tenure and security also make investments difficult. When smallholders do not have a clear
title for their land not only can they not use it as collateral, but also they are much less likely to
invest in land they do not know if they will continue to own it (Deininger and Ali, 2007). Even
when land is owned it is usually not worth enough to use as collateral.
Farming is also difficult to insure or grant credit for because so much is uncertain.
Farmers need to make decisions on what to grow and how much of it to grow without knowing
who else is growing it or what extreme weather could occur to hurt crops during that season
(Peck and Pearce, 2005). Because timing of payments is so variable in agriculture, it is best for
the borrower and the lender if they can operate closely, which is difficult for bigger banks to do
(Peck and Pearce, 2005). This means that many rural areas are left with nowhere to turn for
credit or loans, making the opportunity for them to make further investments in their land much
more difficult. Especially for less wealthy borrowers, this has led to non-traditional forms of
insurance and credit, such as microfinancing, and cooperatives (World Bank, 2007). Even with
these more local options farmers can be unable to get enough money to make a big difference
without putting up too much collateral.
Insurance is a whole other aspect of market access for smallholders, and can bring further
peace of mind for farmers, especially in light of weather shocks due to climate change.
Traditional forms of insurance often occur between neighbors, but this system can leave the
poorest groups out of luck (World Bank, 2007). Formal insurance is difficult to obtain because it
requires upfront sums of money before the harvest that most farmers do not have, it requires a
level of financial literacy that they often do not have, it is not beneficial every year, and it
requires that a formal insurance system functions in their rural area (World Bank, 2007). One of
the suggestions that Christen and Pearce make in relation to agricultural microfinance is
providing weather indexed insurance at a local level, so this can be more specific than it would
be at a country or district level (Peck and Pearce, 2005). Weather indexed insurance allows for
objective indicators such as rainfall and temperature to estimate risk and shock of different areas;
a difficult measure because it can affect individuals in so many different ways (World Bank,
2007). Insurance providers set certain thresholds for different regions based on risks and they
pay out if the index falls above or below these thresholds, no matter if damage was incurred by
every farmer insured or not. This could be very useful for smallholder farmers, especially those
that do not have a diverse income, because the payouts would be directly related to crop prices
and the weather in their region, and involvement in this type of insurance has even been related
to an increase in input use (Cole et al. 2012). It is better for farmers who do not have to assess
multiple risks or pay bribes to collectors, and for benefits providers so that they can reduce
overhead and pay only for certain risks, not cover all agricultural risk (Peck and Pearce 2005,
Kloeppinger-Todd, Sharma 2010). If insurance is run well it can benefit all parties and reduce
some of the risks that farmers take on. However, even though weather-indexed insurance has
good benefits for farmers and providers, and even work is done to increase the financial literacy
and trust in insurance, smallholders just do not have the financial liquidity that is required to
purchase insurance (Binswanger-Mkhize 2012, Cole et al 2012).
Even in the absence of these financial mechanisms, behavioral changes on the farm can
allow farmers to adapt better to climate change and even contribute to combatting climate
change. Farmers can coordinate with neighbors to track new weather patterns and share
successful strategies for changing traditional practices to work more effectively in the changing
environment. Some of these changes will require more financing and farmers on their own
already implement some, often times if one change is made on the farm it will lead to more
change in behavior. For example, implementing irrigation on the farm affects other land
management, such as increased use of mulch and crop rotation (Nkonya et al. 2004). For many
practices, the biggest barrier is knowledge, because activities such as composting, mulching,
rotating crops, or letting fields or strips lie fallow do not usually require any extra materials, they
just require the expertise to implement them, which can be learned from a neighbor or nearby
farmer (Nkonya et al. 2004). Another knowledge limitation exists wherein not every farmer
knows that he or she needs to adapt to climate variations, let alone how to adapt. There is a long
process that smallholder households need to go through in order to actually reach the point at
which they can implement adaptation mechanisms.
Previous studies have looked at the changing practices of food insecure households,
taking climate change into account, but not using it as a primary focus. Kristjanson et al’s 2008
study found that smallholder famers are making changes, but they are usually small and do not
involve innovative new practices (Kristjanson et al 2008). These practice changes include
management of land such as “introducing crop cover, micro-catchments, ridges, rotations,
improved pastures, planting trees, and new technologies such as improved seeds, shorter cycle
varieties, and drought tolerant varieties” (Kristjanson et al 2008). In another study, it was found
that something as simple as having a rain gauge on the farm allowed farmers to manage their
crops better and get better yields (Hellmuth 2007). Drawing on these literatures, we focus on the
relationships of climate and agricultural training to the use of the traditional practices as well as
change in the use of inorganic fertilizers, herbicides, and pesticides – all more serious inputs.
Agriculture, Climate Change, and East Africa
The Intergovernmental Panel on Climate Change (IPCC) has reported on the increase in
greenhouse gas emissions throughout the world, contributed largely by developed countries. This
increase has caused shifts in precipitation, weather, and seasons (Nelson et al. 2013). Africa is
undergoing a variety of effects from climate change. As a very large continent, there will be
different effects everywhere. East Africa will likely have an increase in rain but all of Africa will
suffer from more extreme events, droughts and floods (Collier et al. 2008). Many countries in
sub-Saharan Africa are experiencing very low farm productivity already, and as climate change
continues, this will continue to worsen (Omamo et al. 2005). Higher temperatures will also
accompany global warming and shorten growing seasons and prevent certain crops from growing
at all. Staple crops such as wheat and maize could be especially threatened (Collier et al. 2008).
The yields of these stable crops will be also be affected, research showing that they are already
decreasing due to increased variability and extreme events. As rain and temperature are likely to
continue increasing in variability, these yields could decrease even further (Collier et al. 2008).
These changes in rainfall and temperature also cause secondary effects, such as
expanding the range of pests, increased resource competition, and even increasing violent and
non-violent social conflict (Ziervogel 2008, Hendrix and Salehyan 2012). Population is still
growing rapidly in Africa, and agricultural productivity cannot keep up (Nelson et al. 2013),
putting additional pressure on smallholders to keep yields and production up. This is a problem
particularly in many East African countries, where there is a very low level of labor productivity
compared to the rest of Africa and the world. As temperature and precipitation variations
increase, food security could fall further (Nelson et al. 2013). Most of the countries in East
Africa, though their soil types and topography vary wildly, rely on agriculture as the main staple
of the economy (Collier et al. 2008). Thus, understanding smallholder adaptation strategies is
essential for planning for the near future in East Africa.
Agriculture, Climate Change, and Uganda
East Africa is a large area and has different topography throughout, but smallholder
farmers are affected similarly by climate change throughout the region. Though Uganda is not a
large country, there are diverse climates, types of soil, farming techniques, crops, and difficulties
that farmers encounter. There is a large amount of variation in the type and amount of fertilizers
or other modern inputs Ugandan farmers use, and to date they use much less than most African
countries (Nkonya 2004). Generally, there are two periods of rain and two shorter dry seasons
(Nkonya 2004). Farmers have worked with these conditions for their entire lives and learned
from the generations before them how to best farm in these conditions. It is unknown, however,
how and how well they will adapt to today’s rapidly changing climate. It is probable that some
farmers will be able to adapt their techniques moving forward, but many will encounter problems
with changing weather patterns such as flooding or droughts.
Uganda has better access to water than many other countries in Africa; however this
could quickly change as weather becomes more extreme and extended dry seasons or rainy
seasons change water storage and supply (Mukwaya et al., 2011). Although the agriculture sector
in Uganda now only generates twenty percent of the GDP, seventy percent of the country is still
employed by this sector and as of 2010, eighty five percent of the country still lived in rural areas
(Mukwaya et al., 2011). The tenure system in place is similar to elsewhere in East Africa,
causing insecurity about land ownership among farmers and sometimes increasing the likelihood
that they will work the land without considering the future because they do not know what the
future will hold (Nkonya 2004). Some smallholders choose to sell or abandon land and migrate
to urban areas but unfortunately, the urban economy is not able to incorporate this new influx
and people often end up unemployed, living in the slums, and need to farm anyway, finding
small space on the edges of the city (Mukwaya et al., 2011). While this may allow them to
continue growing some food, land pressure is even greater in these areas. Uganda has recognized
that they are encountering problems with decreasing food supplies and there are some
government efforts operating and in the works to support farmers throughout the country, though
they are still vulnerable (Omamo et al. 2005).
Uganda faces additional region-specific challenges throughout the country. In Northern
Uganda, they will focus on rebuilding agriculture after the war with the Lord’s Resistance Army.
Agricultural knowledge was lost when Ugandans were forced into displacement camps during
the war. As they have been repopulating villages post-conflict, the transition back into
agricultural production is crucial, both for increased food security, and for reintegration and
incorporation for the communities (Birner et al. 2011). This difficult task requires a lot of
support because it is most successful when implemented at small scales and tailored to specific
areas (Birner et al. 2011).
Uganda also experiences differences between rural and urban areas. There is worse
poverty in rural areas even as the urban areas and the slums that surround them grow (Mukwaya
et al. 2012). When rural poverty is worse than urban poverty, people abandon farms and try and
make it in the urban areas, even though the modern sectors have not grown enough to
incorporate all of these people. This move to the city is often unsuccessful for them, and
decreases food production throughout the country (Mukwaya et al. 2012). This problem is
persistent throughout the country, but particularly significant in the south near the capitol and
largest city, Kampala (Mukwaya et al. 2012).
Uganda confronts many unique systemic and agricultural problems that are specific to the
country. Uganda is also still dealing with effects of war in the Northern region and the recovery
of land, people, and farming knowledge in this area. However, many of the problems the farmers
here face are going to be similar if not the same as farmers elsewhere in East Africa. While
programs and adaptation suggestions will need to be tailored to specific communities, not just in
East Africa, but also in Uganda, the overall behavior patterns of how smallholders in Uganda are
adapting or not adapting to climate change will indicate how smallholders elsewhere might be
adapting their behaviors to climate change. Thus, the analysis I present below of farmers in
Uganda is broadly representative of the adaptation process underway in much of Africa and
indeed among smallholders throughout the tropics.
As we continue to explore this area, we will focus on some questions that have not been
answered in relation to this topic. In Uganda, precipitation has increased by 3% and temperature
by 6% in the last decade relative to the second half of the twentieth century. Which farmers and
communities have experienced the most changes? How have farmers adjusted their agriculture
practices to reflect these community level changes?
Data and Methods
In order to address these research questions we used a large social survey data set
collected in Uganda, and gridded climate data on precipitation and temperature. The survey data
were analyzed using Stata and combined with the climate data. Climate data was mapped and
analyzed using ArcGIS, Excel, and Stata.
Data and Measures
The social survey data in Uganda were collected as the “Project on Policies for Improved
Land Management in Uganda” in both 2003 and 2013. These data refer to the previous year, so
they cover the 2002 and 2012 agricultural practices. Many groups collaborated on the data
collection, including the International Food Policy Research Institute (IFPRI), the World Bank,
the Makerere University Faculty of Agriculture, the National Agricultural Research
Organization, and the Norwegian Trust Fund. Clark Gray, Assistant Professor of Geography at
the University of North Carolina (UNC), served as Principal Investigator for the 2013 wave of
data collection, funded by a grant from the National Science Foundation (NSF) with
collaboration from Brown University.
The data were collected through a series of four questionnaires in 2003, and five
questionnaires in 2013. While they were collected in 2003 and 2013, the questions asked were
referring to 2002 and 2012, so the years 2002 and 2003 refer to the first set of data and the years
2012 and 2013 refer to the second set. These questionnaires were administered in-person by
interviewers fluent in local languages. One collected information about the community as a
whole and was administered to community leaders. It included general information about
community programs, use of private and public lands, and environmental and resource
management. The other three surveys provide information about individual households;
collecting extensive information such as household makeup, how the agricultural land and plots
are maintained, and how many crops are harvested and sold. In 2013, an additional module on
reproductive history of women, and prenatal and child care, was added. These surveys were
given in eight districts throughout the country and over one hundred communities within these
districts. The houses were originally interviewed in 2003, and then in 2013 the same places were
interviewed again, it was indicated if the household was the same or if it had split. Some
households split after 2003 and these households were study in their multiple iterations in 2013,
indicating if it was a split household in the 2013 set.
The analysis presented here draws from the household survey data, the survey on the
individual plots, and the survey on all the plots. Data were cleaned and manipulated to generate
measures of agricultural practices undertaken by each household (regardless of on which plot the
practice was used), and measures of household characteristics. We wanted to highlight the
changes that farmers made in between 2003 and 2013 to see if they were adapting to climate
change. Thus our key dependent variable is the use of certain farming practices, and particularly
the initiation or discontinuation of certain practices between 2003 and 2013. The survey asked
about 27 different types of practices, which we initially grouped into eight different types of
practices – irrigation, traditional practices, use of fallow, mulch, machines, herbicides or
pesticides, integrated pest management, and inorganic inputs – as shown in Table 1.
Table 1 Agricultural Practices: New Groupings and Original Codes
The data included every crop that any of the farmers in the study grew, so we focused on
the crops with the biggest influence and prevalence. These five crops were bananas, beans,
maize, cassava, and sweet potato. As Table 2 shows, their numbers were much larger in the
districts than the other top nine crops.
Number of Farmers
Growing This Crop in
Original Data Set
Table 2 Crop Representation in the
Data (*Top five – chosen for study).
Our key independent variables are measures of climate and differences in climate
between 2002 and 2012. This information was downloaded from the University of Delaware’s
site as raster files and specified to Uganda at the 0.5 x 0.5 degree for the years 1951-2012 (see
figure 2 for examples of the raster maps in ArcGIS). Using ArcGIS, the locations of the survey
communities were overlaid on the climate data and then historical weather data were extracted
for every community. These data included monthly mean temperature (degrees Celsius) and
monthly total precipitation (mm) for each month in each year 1951-2012. We focused on the
months of March through August, during which the two rainy seasons fall. We averaged these
monthly temperature and precipitation measures for the years 1951-2000 to establish a halfcentury baseline of historical climate data for the middle of the agricultural season. There are two
rainy seasons in Uganda In Southern Uganda these are fairly distinct, with a dry season between,
but in the North, the dry season is more of a small dip between the two more wet seasons
(Uganda: Climate and Agriculture). This is also clearly demonstrated in the climate data that we
gathered, as can be seen in Figure 1 – the map of the community points and climate graphs of
three districts in each area.
Our analysis focuses on variation across communities experiencing different weather
during the years immediately preceding each survey. We expressed the recent weather in terms
of deviations from the long-term averages. Once the long-term averages and standard deviations
were established using Stata, we found the differences between the long-term mean and the
monthly average temperature or total rainfall for each of our six focal months in each of the
years 2000 – 2002 and 2010 – 2012. Then the number of times during the three years prior to
each survey that a month fell above or below one standard deviation from the mean was
calculated (maximum eighteen, minimum zero). These data show the amount of climate shock
that farmers experienced in their region in the two years before, and during the year that data was
Figure 1: Precipitation Trends: Map of the
community points and graphs of precipitation records in
Figure 2: ArcGIS Climate Maps: Raster maps in Arc GIS of
temperature (left, orange) and precipitation (right, blue) in
We also include other control variables in our models – the number of plots, the total area
of the plots, whether or not they grew any of the top five crops – beans, maize, cassava, sweet
potatoes, and bananas - the size of the household, the age of the head of the household, if any
member of the household had received any agricultural training, and if any member of the house
is employed in a non-agricultural sector.
We used a logistic regression model, regressing use of each of the three practices on
climate and control variables. We did not include the 2003 climate variables, instead including
the agricultural practices in 2003 (which reflect previous climate conditions) and focusing on the
effect that the climate variability in 2010-2012 had on practices in 2013. Logit coefficients
presented in Table 5 thus show the relationship between the independent variables and the logodds of the household using the practice in 2013 given the value of the other variables we were
testing. Results significant at the 0.05 and 0.10 level are shown in this table.
Results and Findings
Though the dataset started off large and messy, we were able to use it to find interesting
results about smallholder farmers and how climate influences the decisions they make in their
farming practices. Of the eight different groups of types of practices we created – irrigation,
traditional practices, use of fallow, mulch, machines, herbicides or pesticides, integrated pest
management, and inorganic inputs – three that changed between 2002 and 2012 and represented
a continuum from traditional to modern practices were traditional (slash and burn) practices,
herbicides or pesticides, and inorganic inputs. These three changes in practice were all affected
by a few different variables, though not those that we most expected would affect the
We ran regressions to estimate the relationships between key independent variables and
smallholders’ use of these practices in 2013. In each case, we controlled for their use of the
practice in 2003, meaning that the coefficients on other variables represent their relationship with
the change in use of the practice over the decade. The results of these regression models are
shown in Table 5. Interestingly, the use of a practice in 2003 is not consistently significantly
related to the use of the practice in 2013, suggesting there is some variability in behavior
reflecting experimentation and/or responsiveness to inter-annual variability.
The focus of our analysis is on the impacts of recent weather patterns on use of these
practices. We seek to examine how farmers are responding to recent weather anomalies (relative
to long-term averages) as an indication of likely adaptation to climate change in the absence of
additional interventions. As shown in Table 3 (Descriptive Statistics I – Climate Variability),
Uganda has already been experiencing hotter conditions than the long-term averages and drier
conditions in some places. In this recent period, farmers are responding by increasing use of
herbicides and pesticides during years when it has been drier and decreasing their use when it has
been wetter. Their probability of using traditional (slash and burn) methods or of using inorganic
fertilizer, however, is unrelated to the recent weather anomalies. In contrast to studies in other
areas in the world, all of the changes in practices seem to be unrelated to temperature anomalies.
These results, however, could reflect that practices have already changed in response to
temperature changes. That is, farmers have already effectively adapted to a hotter climate (both
2003 and 2013 showed a range of 11-18 in the number of above average months in the past three
agricultural seasons, suggesting no change in the weather patterns between the two survey dates,
see Table 3).
Beyond the focus on weather, we examined the relationship between other farmer
behaviors and characteristics and the change in the use of the three types of practice between
2003 and 2013. Looking first at whether changing practices reflected previous crop choices, we
examine the relationship between planting each of the five most common crops (separated into
the first and second rainy periods) and use of the slash/burn, herbicide and pesticides, and
inorganic fertilizer. Whether or not they grew a certain crop in the first or second season affected
all of the practices in some way. Their use of traditional slash and burn methods was impacted if
they grew maize in the first season, and significantly impacted if they grew it in the second
season as well. Growing bananas also significantly impacted their use of traditional slash and
burn methods. Growing cassava in the second season also significantly impacted the change in
the use of herbicides and pesticides. As demand for these crops increases, especially
international demand for bananas, farmers produce them as a monoculture and do not rotate
them, making it much easier for pests to target them (Lunder 2013). It is also possible that the
prevalence of pests is increasing with the large variation in rainfall and farmers are trying to
compensate for this. Use of inorganic fertilizer was additionally related to growing beans in the
first season. This is interesting because beans should not need that much fertilizer as they are
legumes and can fix nitrogen themselves, suggesting that they might be attempting to further
increase yields, or alleviate deficiencies in a different macronutrient. Growing maize in the
second season is also significantly related to increasing use of inorganic fertilizer, reflecting
either inherently low soil fertility or the high nitrogen demands of maize production.
Turning to the characteristics of the household and farmer, the area of the farm, the size
of the household, and the participation of anyone in the household in a non-agricultural activity
did not have any significant effect on any of the practices. This could have been because the farm
areas were similar because smallholder farmers all operate on similarly sized plots of land
(Figure 3). Whereas large variations in land could make a large difference in practices used
because certain practices could be much more difficult or even more expensive if they are
practiced on larger swaths of land, the small variations in land size are unlikely to affect what
practices they use. Similarly, household size was largely concentrated (Figure 4) and no
background literature indicated that this would be an important factor in how a household makes
a decision. The age of the head of the household was significantly related to the use of inorganic
fertilizer only. This could be because the older farmers are used to farming in a traditional way
and are uninterested in changing and the younger farmers are more excited and willing to adopt
new methods in their farming. Participation in a non-agricultural activity was something that we
had expected might have some effect because it could mean that the household has access to
more funds, as the literature does show that households with a non-agricultural income have
higher incomes that might allow them to bear the cost of the inorganic fertilizer or modern
herbicides and pesticides. These factors did not have significant coefficients in the regression
model. However, someone in the house having received agricultural training was significantly
positively related to herbicide and pesticide use. This is something we would have expected. If a
household receives some agricultural training than they will know more about herbicides and
pesticides, how effective they may be for their crop, how to use them, where to get them, and
maybe even help purchasing them.
Figure 3: Histogram of the area of the
land owned by a household.
Figure 4: Histogram of the amount of
people living in a household.
Table 3: Descriptive Statistics I – Climate variability
Table 4: Descriptive Statistics II – Household characteristics and agricultural practices.
Table 5: Regression model results
The analysis we undertook offered interesting conclusions, and offered many more
questions that could be addressed both through this quantitative data and with further qualitative
data research. We found some interaction between the changing climate and a change in farmer’s
practices, but not as much as expected.
The results indicate that farmers are responding to variability in recent weather in a few
ways. Use of pesticides and herbicides increased in drier years and decreased in wet years.
Temperature did not seem to affect change in practices between 2003 and 2013, but this could be
because that change took place previous to 2003. The temperature in the six primary months of
the agricultural season (March through August) was higher than one standard deviation above
the mean for fourteen to eighteen of the eighteen months in 2000-2002 throughout the region and
for eleven to fourteen of the eighteen months form 2010-2012. While we focused on the change
from 2003 to 2013, the changes they made in response to temperature could have taken place
before the 2003 study, something that would not reflect in these responses. In 1988, Matlon and
Kristjanson had already completed a study looking at how farmers coped with variability of
rainfall, demonstrating that farmers have been adapting for decades, which makes the possibility
that farmers had already made changes in practices by 2003 even more likely.
Studies show that East Africa will continue to have an increase in rain, temperature, and
extreme events such as drought and flooding (Collier et al. 2008). If the results from these data
hold, this will mean an increased in use of herbicides and pesticides in the future. This may be
hindered if financing for farmers does not improve, as herbicides and pesticides can be a
financial burden on the household. As the data is only quantitative, we cannot know for sure why
certain farmers decided to use more herbicides and pesticides in the drier seasons, and less in the
wetter seasons. This could be further explored through interviews, or even in studying the
interactions of certain results with other variables within this quantitative dataset.
Changes in practices could also be deterred because of lack of training. The use of
herbicides and pesticides was slightly increased if a member of the household had taken part in
any agricultural training, and was not significant in the use of inorganic fertilizers or traditional
practices. Perhaps if training were more widespread or available it could influence these practice
changes more drastically, especially because the literature indicated that training programs
should be influencing farmers to change practices in order to better adapt (World Bank 2007).
The literature also showed that engagement in non-agricultural activities would allow families to
increase resources and invest more in the farm, which meant I expected this to influence these
practices more than it did in these results (Natoya 2004). Non-agricultural activities should be
bringing in more income to the household and allowing for more access to training or resources
to advance the farm, but this did not seem to be the case with these three practices in the study
This research is an interesting step towards more that could be found in this data, and that
could be enhanced further with qualitative additions and study of interactions. As we saw that the
age of the head of the household significantly impacted the use of inorganic fertilizer, it seems
that younger households may be more willing to make changes. Was this same thing true with
any other practices? Are younger households overall more willing to make changes? Do they
have better access to resources? These questions could be looked at in the data further through
interactions, and even more could be discovered if the farmers were interviewed about this. As
previously mentioned, it would also be interesting to further explore the effect of the climate
change in the few years before the 2002 data was collected in 2003 in order to determine if
changes were made then due to climate change. This could also be enhanced through interviews
with farmers on why they make certain decisions and what considerations go into these decisionmaking processes.
We were able to answer the questions on whether or not farmers are adapting to climate
change, and which practices they were changing as a result of the climate changes, but we still
cannot answer the underlying questions of why exactly they are adapting. If we know more about
why they choose to adapt, and how, then it will be easier to support them in future adaptation
measures. It is important that programs for farmers critically affected by climate change are
established and administrative capacity is increased dramatically so that farmers can benefit from
local modeling systems and can receive specific recommendations on how to adapt to the climate
change that affects their area (Nelson et al. 2013). Even within this one country, we can see
variation in climate, so this is even more important at the regional East Africa area. While we
support farmers in adaptation, it is also crucial that mitigation increase immediately so that the
effects do not continue to be exacerbated.
Smallholder farmers in Uganda to have the ability to adapt to climate change, and many
are willing and already making changes in order to adapt. However, this needs to continue to be
a subject of research as climate continues to change more dramatically. Hopefully research can
lead to policy changes that support farmers in the most effective ways so that not only they
continue to survive, but so that they can support and feed their communities.
Bagamba, F., Burger, K., & Kuyvenhoven, A. (2009). Determinant of Smallholder Farmer Labor
Allocation Decisions in Uganda. IFPRI Discussion Paper, 00887.
Binswanger-Mkhize, H. (2012). Is There Too Much Hype about Index-based Agricultural
Insurance? Journal of Development Studies, 187-200.
Birner, R., Cohen, M., & Ilukor, J. (2011). Rebuilding Agricultural Livelihoods in Post-Conflict
Situations: What are the Governance Challenges? The Case of Northern Uganda. Uganda
Strategy Support Program (USSP), Working Paper no. USSP 07.
Bolwig, S. (2012). Poverty and Gender Effects of Smallholder Organic Contract Farming in
Uganda. Uganda Strategy Support Program (USSP), No. 8.
Boucher, S., Carter, M., & Guirkinger, C. (2008). Risk Rationing and Wealth Effects in Credit
Markets: Theory and Implications for Agricultural Development. American Journal of
Agricultural Economics, 90(2), 409-423.
Cole S, Bastian G, Vyas S, Wendel C, Stein D (2012) The effectiveness of index- based microinsurance in helping smallholders manage weather-related risks. London: EPPI-Centre,
Social Science Research Unit, Institute of Education, University of London.
Collier, P., Conway, G., & Venables, T. (2008). Climate change and Africa. Oxford Review of
Economic Policy, 24(2), 337–353-337–353. Retrieved June 1, 2014.
Deininger, K., & Ali, D. (2007). Do Overlapping Land Rights Reduce Agricultural Investment?
Evidence from Uganda. American Journal of Agricultural Economics, 869-882.
Ellis, F. (1999). RURAL LIVELIHOOD DIVERSITY IN DEVELOPING COUNTRIES:
EVIDENCE AND POLICY IMPLICATIONS. Overseas Development Institute, 40.
Gine, X., Townsend, R., & Vickery, J. (2008). Patterns of Rainfall Insurance Participation in
Rural India. The World Bank Economic Review, 539-566.
Hellmuth, M.E., Moorhead, A., Thomson, M.C., and Williams, J. (eds) 2007. Climate Risk
Management in Africa: Learning from Practice. International Research Institute for
Climate and Society (IRI), Columbia University, New York, USA.
Hendrix, C., & Salehyan, I. (2012). Climate change, rainfall, and social conflict in Africa.
Journal of Peace Research, 49(35), 35-50. Retrieved October 7, 2014, from
Kaspersen, L., & Føyn, T. (2010). Price transmission for agricultural commodities in Uganda:
An empirical vector autoregressive analysis. Uganda Strategy Support Program (USSP),
Working Paper No. 06.
Kristjanson, P., Neufeldt, H., Gassner, A., Mango, J., Kyazze, F., Desta, S., . . . Coe, R. (2012).
Are food insecure smallholder households making changes in their farming practices?
Evidence from East Africa. Food Security, 381-397.
Lunder, S. (2014, April 28). Banana Cultivation Is Pesticide-Intensive. Retrieved March 1, 2015,
Mukwaya, P., Bamutaze, Y., Mugarura, S. & Benson, T. 2012. Rural-urban transformation in
Uganda. IFPRI – USSP Working Paper no. 10. http://www.ifpri.org/publication/ruralurban-transformation-uganda.
Nabbumba, R., Bahiigwa, G., Okello, B., & Laker-Ojok, R. (2005). Part One - Agricultural
Productivity, Natural Resource Management and Food Security. In The Future of
Smallholder Farming in Eastern Africa.
Nelson, G., Waithaka, M., Thomas, T., & Kyotalimye, M. (Eds.). (2013). East African
agriculture and climate change. International Food Policy Research Institute (IFPRI).
Nkonya, E., & Kaizzi, C. (2003). Poverty-Natural Resource Management Linkages: Empirical
Evidence from Uganda.
Nkonya, E., Pender, J., Jagger, P., Sserunkuuma, D., Kaizzi, C. & Ssali, H. 2004. Strategies for
sustainable land management and poverty reduction in Uganda. IFPRI Research Report
no. 133. http://www.ifpri.org/publication/strategies-sustainable-land-management-andpoverty-reduction-uganda-0.
Omamo, S.W., Babu, S. & Temu, A. (editors). 2005. The future of smallholder farming in
eastern Africa. The roles of states, markets, and civil society. IFPRI Eastern Africa Food
Policy Network. Kampala: IFPRI.
Peck- Christen, R., & Pearce, D. (2005). MANAGING RISKS AND DESIGNING PRODUCTS
FOR AGRICULTURAL MICROFINANCE: FEATURES OF AN EMERGING
MODEL. Occasional Paper, No. 11.
Pender, J., Ssewanyana, S., Edward, K., & Nkonya, E. (2004). Linkages Between Poverty and
Land Management in Rural Uganda: Evidence from the Uganda National Household
Survey, 1999/00. Environment and Production Technology Division.
Preliminary Report Highlights U.S. Farms and Farmers. (2014). National Agricultural Statistics
Service. Retrieved from
Roberts, J. Timmons, and Bradley C. Parks. A Climate of Injustice Global Inequality, NorthSouth Politics, and Climate Policy. Cambridge, Mass.: MIT, 2007. Print.
Ruecker, G., Park, S., Ssali, H., & Pender, J. (2003). Strategic Targeting of Development
Policies to a Complex Region: A GIS-Based Stratification Applied to Uganda. ZEF –
Discussion Papers on Development Policy, 69.
Uganda: Climate & Agriculture. (n.d.). Retrieved 2015, from http://www.ourafrica.org/uganda/climate-agriculture.
Water management. (2013). Retrieved from http://www.fao.org/ag/save-andgrow/cassava/en/4/index.html.
World Bank. 2007. World Development Report 2008: Agriculture for Development.
Washington, DC. World Bank.
Zeller, M. (n.d.). Models of Rural Financial Institutions.
Ziervogel G, Zermoglio F (2009) Climate change scenarios and the development of adaptation
strategies in Africa: challenges and opportunities. Climate Research 30.