A method for evaluating climate change adaptation strategies for small-scale

farmers using survey, experimental and modeled data

L. Claessens
a,b,
, J.M. Antle
c
, J.J. Stoorvogel
b
, R.O. Valdivia
c
, P.K. Thornton
d,e
, M. Herrero
d
a
International Potato Center (CIP), P.O. Box 25171, Nairobi, Kenya
b
Wageningen University, P.O. Box 47, 6700AA Wageningen, The Netherlands
c
Oregon State University, Corvallis, OR, USA
d
International Livestock Research Institute (ILRI), P.O. Box 30709, Nairobi, Kenya
e
CCAFS, ILRI, P.O. Box 30709, Nairobi, Kenya
a r t i c l e i n f o
Article history:
Received 12 January 2011
Received in revised form 23 March 2012
Accepted 17 May 2012
Available online 27 June 2012
Keywords:
Adaptation
Climate change
East Africa
Impact assessment
Socio-economic scenarios
TOA-MD model
a b s t r a c t
Sub-Saharan Africa (SSA) is predicted to experience considerable negative impacts of climate change. The
IPCC Fourth Assessment emphasizes that adaptation strategies are essential. Addressing adaptation in the
context of small-scale, semi-subsistence agriculture raises special challenges. High data demands includ-
ing site-specic bio-physical and economic data are an important constraint. This paper applies a new
approach to impact assessment, the Tradeoff Analysis model for Multi-Dimensional Impact Assessment
(TOA-MD), which simulates technology adoption and associated economic, environmental and social out-
comes in a heterogeneous farm population for a regional impact assessment. The methodology uses the
kinds of survey, experimental and modeled data that are typically available in countries where semi-sub-
sistence systems are important, combined with future socio-economic scenarios based on new scenario
pathway concepts being developed by the climate change and impact assessment modeling communi-
ties. Characteristics of current and future agricultural systems, including land use, output, output price,
cost of production, and farm and household size are analyzed and compared for both current and pro-
jected future climate (2030), with and without adaptation, and for different socio-economic scenarios.
The methodology is applied to two study areas in Kenya. These case studies show the potential of this
approach to provide a exible, generic framework that can use available and modeled data to evaluate
climate impact and adaptation strategies under a range of socio-economic scenarios.
2012 Elsevier Ltd. All rights reserved.
1. Introduction
The changing climate is exacerbating existing vulnerabilities of
the poorest people who depend on semi-subsistence agriculture
for their survival (Slingo et al., 2005; IPCC, 2007; Nelson et al.,
2009). Sub-Saharan Africa (SSA) in particular is predicted to expe-
rience considerable negative impacts of climate change (e.g.,
Thornton et al., 2006). The IPCC Fourth Assessment emphasizes
that adaptation strategies are essential and these must be devel-
oped within the broader economic development policy context
(IPCC, 2007). Addressing adaptation in the context of small-scale,
semi-subsistence agriculture in SSA raises special challenges that
cannot be addressed adequately by the approaches taken thus far
in most studies (Adger et al., 2003). Most of the existing research
has focused on impacts of climate change and adaptation to cli-
mate change in the agricultures of industrialized countries. In the
relatively few studies conducted in Africa, agricultural research
has either focused on individual crops (e.g., Hijmans, 2003; Jones
and Thornton, 2003), has used aggregated data and models (e.g.,
Winters et al., 1998; Mendelsohn et al., 2000), or used statistical
analysis that does not allow for site-specic adaptation strategies
(e.g., Kurukulasuriya and Mendelsohn, 2006). IPCC and some re-
cent studies at the sub-continental scale for Africa indicate the
importance of assessing the effects of climate change and possible
adaptation strategies at the agricultural system and/or household
level, rather than focusing on aggregated results that hide a large
amount of variability (Burke et al., 2009; Nelson et al., 2009;
Thornton et al., 2009a, 2010; Baethgen, 2010). High data demands
are one of the important constraints for this type of analysis be-
cause site-specic bio-physical and economic data are required,
typically obtained from costly multi-year farm-level surveys. At
the spatial resolution required, another drawback is that projec-
tions of climate change and simulations of the effects on crop
and livestock productivity come with a high degree of variability
and associated uncertainties depending on the climate models
and methodologies used. The development and application of
0308-521X/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.agsy.2012.05.003

Corresponding author. Present address: International Crops Research Institute

for the Semi-Arid Tropics (ICRISAT), P.O. Box 39063, 00623 Nairobi, Kenya. Tel.:
+254 20 722 4568; fax: +254 20 722 4001.
E-mail addresses: l.claessens@cgiar.org (L. Claessens), john.antle@oregonstate.
edu (J.M. Antle), jetse.stoorvogel@wur.nl (J.J. Stoorvogel), roberto.valdivia@
oregonstate.edu (R.O. Valdivia), p.thornton@cgiar.org (P.K. Thornton), m.herrero@
cgiar.org (M. Herrero).
Agricultural Systems 111 (2012) 8595
Contents lists available at SciVerse ScienceDirect
Agricultural Systems
j our nal homepage: www. el sevi er . com/ l ocat e/ agsy
relatively simple and reliable methods for ex ante evaluation of
adaptation strategies at the household and agricultural system lev-
els are needed to provide timely assessments of the potential im-
pacts in the context of climate change.
This paper describes and applies a new approach to regional
technology and environmental impact assessment using a novel
simulation approach to impact assessment, implemented with
the Tradeoff Analysis for Multi-Dimensional Impact Assessment
model (TOA-MD). The methodology is applied to two study areas
in Kenya. It makes use of the kinds of data survey, experimental,
modeled, and expert that are typically available to assess future
environmental changes and prospective technologies, especially in
countries where complex, semi-subsistence systems predominate.
The approach integrates socio-economic and bio-physical data on
farmers land allocation, outputs and cost of production and char-
acterizes the spatial heterogeneity in economic returns to baseline
and alternative systems under current and possible future climate.
Productivity characteristics of alternative systems that may be bet-
ter adapted to future climates are characterized using available
data which may include laboratory and eld experiments, simula-
tion model data, and expert data. Using these data, the model sim-
ulates the adoption of alternative systems and their economic and
social impacts among farms that may lose or gain from climate
change. A variety of possible management strategies is then as-
sessed for their capability to facilitate adaptation to climate change
under different socio-economic scenarios.
2. Study areas
2.1. Vihiga
Vihiga district in western Kenya lies between 1300 and 1500 m
above sea level and covers an area of 563 km
2
of which 419 km
2
is
arable land (CBS, 2003) (Fig. 1). Vihiga district is broadly represen-
tative of other areas of the East African highlands found in Uganda,
Ethiopia, and Madagascar in terms of soils, climate, technology,
and production potential (Soule and Shepherd, 2000). The districts
high potential agricultural area is characterized by well drained
nitisols that support the growing of various cash and food crops
(Waithaka et al., 2006). Soil fertility is low due to leaching and con-
tinuous cropping without sufcient replenishment (AFRENA, 1998;
Salasya et al., 1998). Nitrogen and phosphorus are the main limit-
ing nutrients for food crops (Soule and Shepherd, 2000). Currently
the area receives adequate bimodal rainfall that ranges between
18002000 mm per year with heavier long rains from March to
June and short rains between September and December. Tempera-
tures are moderate and range from 14 to 32 C with limited diurnal
variations.
In 1999 Vihiga district had a total population of 500,000 with a
population growth rate of 2.2% (CBS, 2001). According to the latest
census in 2009, the population has increased to 550,000 (KNBS,
2010). Poverty mapping in Kenya places Vihiga among the poorest
districts in the country (CBS, 2003). Most farm households in Vih-
iga show a maize decit of 200400 kg per year, which is equiva-
lent to shortage in six to ten months each year (Waithaka et al.,
2005). The shortage is aggravated by the increasing conicts
among food, cash and fodder crops as farm sizes continue to de-
cline due to growing population pressure and dividing farm land
among family members. This has greatly reduced available fodder
with hardly enough to feed livestock all year round. With high pov-
erty levels, farmers do not use high-return inputs such as certied
seeds, fertilizers, disease and pest control measures, and rotations,
but are limited to low-input, low-return enterprises (Waithaka
et al., 2006). The average farm household has 4.7 persons living
on a 0.5 ha farm creating a greater need for intensied agricultural
production.
About 60% of the Vihiga population falls below the poverty line
of 1 US$/person/day (CBS, 2003) with an average total income of
56 Kenyan Shillings (KSh) per farm household per day (1 US$ was
equivalent to KSh 77 in mid-2005). Farm households obtain 65%
Fig. 1. Location of the study areas in Kenya. The Machakos study area comprises both Machakos and Makueni districts.
86 L. Claessens et al. / Agricultural Systems 111 (2012) 8595
of their income from off-farm sources in the form of wages and
remittances (Waithaka et al., 2006). A high proportion of farm in-
come is obtained from milk sales. In an effort to enhance farm
household income and food security, farmers in the district appear
to pursue risk management strategies, such as producing much of
the household consumed food to avoid market risk, and diversica-
tion, and hence grow many crops on their small land holdings.
The main food crops are maize, beans, sorghum, groundnuts,
bananas and a variety of vegetables and the main cash crop is
tea. The predominant livestock is local Zebu, which is mainly used
for dairy production. Most farmers practice zero grazing and grow
Napier grass for animal feed, which competes with high value
crops in the small holdings. With investments in transportation
infrastructure, the area could have improved market opportunities
as most farms are within 50 km of the large urban centers of
Kakamega and Kisumu with more than 500,000 people each
(CBS, 2001).
2.2. Machakos
The 13,500 km
2
study area is located in the Eastern Province of
Kenya and contains both Machakos and Makueni districts. Eleva-
tion ranges from 400 to 2100 m above sea level (Fig. 1). Almost half
of the total surface of the study area is under agricultural use
(6615 km
2
). Most of the soils in the area are deep, friable, with tex-
tures ranging from sandy clay loam to sandy clay. Inherent fertility
is very poor and soils are generally decient in nitrogen, phospho-
rus and soil organic carbon (<1%) (Onduru et al., 2001; Mora-
Vallejo et al., 2008). The semi-arid climate in the area has low,
highly variable rainfall, distributed in two rainy seasons. Short
rains occur from November to January and long rains from March
to June. Average annual rainfall ranges from 500 to 1300 mm and
mean annual temperature varies from 15 C to 25 C, resulting in
a wide range of agro-ecological conditions (MoA, 1987). Drought
events do happen in cycles of four or ve years, normally in runs
of two or more seasons, and they have great impact on food secu-
rity (Tiffen et al., 1994).
In 1999 Machakos and Makueni had a total population of
around 900,000 and 770,000 respectively (CBS, 2001). According
to the latest census in 2009, the population has increased to
1,100,000 in Machakos and 880,000 in Makueni (KNBS, 2010).
60% of the population in Machakos and 62% in Makueni fall below
the poverty line of 1 US$/person/day (CBS, 2003).
Agriculture is dominated by subsistence-oriented mixed farm-
ing systems that include both crop and livestock production,
although some coffee and cotton are cultivated in the area as cash
crops. Farm households in general own between 1.5 and 6 ha of
land, of which 1.53.5 ha is cultivated (de Jager et al., 2001). Maize
is the most important staple crop, but a wide variety of other food
crops are grown (beans, millet and sorghum), vegetables (tomatoes
and kales), fruit trees (orange, banana, mango and pawpaw) and
tubers (cassava). For all crops, yields are generally low and crop
failure is a common problem. Soil nutrient management through
application of manure and chemical fertilizer is practiced by farm-
ers. However, due to the relatively high prices of chemical fertil-
izer, this is only applied on plots that are of good quality and
have less risk of crop failure, manure is more often applied on plots
with degraded soil fertility and health (de Jager et al., 2004; Mora-
Vallejo et al., 2008). On elds with continuous cultivation without
external inputs, a sharp yield decline has been observed (Lal,
2010). Soil conservation practices have been implemented in the
area since colonial times. While in the 1930s the building of ero-
sion control structures was enforced after severe land degradation
took place, nowadays the majority of the farmers (almost 75%)
voluntarily maintain these structures and the area is well known
for the widespread use of terrace cultivation (de Jager et al.,
2004; Tiffen et al., 1994). Irrigation is only available for a minority
of farms but some cases exist in locations neighboring Athi River.
Access to simple small-scale irrigation allows the cultivation of
vegetables such as chilli peppers, tomatoes, onions and eggplant
for commercial production. In such cases, where water and mar-
keting constraints are alleviated, farmers directly respond by
applying higher doses of mineral and organic fertilizer. This change
in farm management results in higher and more stable yields and
higher nancial returns (de Jager et al., 2004). In the majority of
the households, livestock (cattle, sheep, goats) represents an
important component of the farming system. The major functions
of livestock are provision of draught power, manure production
and capital assets (saving and insurance) (de Jager et al., 2001).
3. Materials and methods
3.1. Survey data
The data for Vihiga originate from the project System prototyp-
ing and impact assessment for sustainable alternatives in mixed
farming systems in high-potential areas of Eastern Africa (PRO-
SAM) for which farm survey data were collected in 2000 and
2002 (Waithaka et al., 2005; Salasya, 2005). For this analysis, a
selection of 119 farms was extracted from the database for which
complete data (quantities and prices) on inputs (such as seeds, la-
bor, fertilizer, and manure), outputs (crop yields, milk production
and land areas), and farm management were available (Table 1).
For Machakos we used similar farmsurvey data for 120 households
in six villages obtained from studies conducted in the Nutrient
Monitoring project (NUTMON) in 2000 (de Jager et al., 2001;
Gachimbi et al., 2005). The survey data are used to calculate statis-
tics needed to implement the TOA-MD model for the different
activities (crops and milk production) in each study area (Table
1). Annual crops (such as fruits and non-irrigated vegetables), that
are grown heterogeneously across farms and occupy very small
land units in both study areas are grouped for this analysis under
one activity called mixed crops. Crops such as tea, coffee, sugar
cane, and woodlots are treated as xed activities and thus are
not included in this analysis.
3.2. TOA-MD as a climate impact and adaptation assessment tool
For the analysis of climate change impact, adaptation strategies
and poverty we use the Tradeoff Analysis model for Multi-Dimen-
sional Impact Assessment (TOA-MD). The TOA-MD model is a par-
simonious, generic model for analysis of technology adoption and
impact assessment, and ecosystem services analysis. TOA-MD has
been used for the analysis of technology adoption (Claessens
et al., 2009, 2010) and payments for environmental services (Antle
and Valdivia, 2006; Nalukenge et al., 2006; Antle and Stoorvogel,
2006, 2008; Immerzeel et al., 2008; Antle et al., 2010). Antle
et al. (2010) present a validation of the TOA-MD approach against
more complex, spatially-explicit models of semi-subsistence agri-
cultural systems, including a model for the Machakos case study
presented here. Further details on the impact assessment aspects
of the model are provided in Antle (2011a). The model software
and the data used in this and other studies are available to
researchers with documentation and self-guided learning modules
at tradeoffs.oregonstate.edu.
The TOA-MD model simulates technology adoption and impact
in a population of heterogeneous farms. Farms are assumed to be
economically rational and to choose between systems based on ex-
pected economic returns. The simulation model uses data on the
spatial variability in economic returns to represent heterogeneity
in the farm population. When used to analyze technology adop-
L. Claessens et al. / Agricultural Systems 111 (2012) 8595 87
tion, an important implication of this model is that incomplete
adoption of a new technology can be due simply to heterogeneity
in the conditions determining the value of a system to farmers
such as heterogeneity in soils, climate, transportation costs, and
the farm households characteristics the conditions typical of
most technologies and most farm populations. This fact is impor-
tant to emphasize because much of the literature attributes incom-
plete adoption to attitudes such as risk aversion or constraints such
as access to the technology or nancing constraints that are typ-
ically difcult to observe and quantify (for a discussion of relevant
literature, see Suri, 2011). While such factors may indeed contrib-
ute to low adoption rates in some cases, recent research has shown
in important cases that observable heterogeneity in bio-physical
and economic characteristics of farms can be sufcient to explain
observed and often low adoption rates (Antle et al., 2005; Suri,
2011). In climate change assessment, the TOA-MD model implies
that not all farms are affected in the same way in most cases,
some farms lose and some farms gain from climate change.
Similarly, some farms may be willing to adopt technologies that
facilitate adaptation to climate change, while others will not. The
TOA-MD model allows researchers to simulate the impacts of the
full range of adoption rates from zero to 100%.
In the TOA-MD model, farmers are presented with a simple
binary choice: they can operate with a current or base production
system 1, or they can switch to an alternative system. Under the
climate change analysis, it is necessary to distinguish between
three factors affecting the expected value of a production system:
the production methods used, referred to here as the technology,
and the physical environment in which the system is operated,
i.e., the climate, and the economic and social environment in which
the system is operated, i.e., the socio-economic setting that we shall
refer to as a Representative Agricultural Pathway, or RAP. RAPs are
qualitative storylines that can be translated into model parameters
such as farm and household size, prices and costs of production,
and policy. They are being developed by the Agricultural Model
Intercomparison and Improvement Project (AgMIP, Rosenzweig
et al., 2012; Antle, 2011b) to be consistent with new scenario con-
cepts being developed by the climate modeling and impact assess-
ment communities (further details below, Section 3.5). Thus, in the
climate change analysis a production system is dened as a partic-
ular technology used in a particular climate regime and RAP. These
factors combine to determine the productivity and economic value
of the system.
A major challenge in scenario design for climate impact assess-
ment is the dimensionality of the analysis, given that there are
potentially a large number of technology, climate and RAP combi-
nations. To simplify the presentation, assume for the time being
that the analysis is for a given RAP. The logic of the analysis is then
summarized as follows: Farmers are initially operating a base tech-
nology with a base climate. This combination is dened as system 1.
System 2 is dened as the case where farmers continue using the
base technology under a perturbed climate. System 3 is dened
Table 1
Summary of data used in the TOA-MD sensitivity and scenario analysis. Characterization of base system from survey data (means with standard deviation between brackets).
Alternative systems are expressed as % of base system values for productivity (100%).
Vihiga Base system Alternative systems
a
System 1 System 2 System 3
Area Crop yield Net returns CC 2030 imz dpsplw dpsp dpsp1 dpsp12 RAP1 RAP2
Ha/season/farm kg/ha/season KSh/ha % of Base system
Maize-beans 0.24 (0.21) 1512 (1319) 15360 (16726) 70 95 70 70 70 70
Napier grass 0.16 (0.19) 34450 (23643) 22366 (23545) 80 80 80 80 80 80
Mixed 0.23 (0.21) 4031 (1701) 27551 (15971) 80 80 80 80 80 80
Sweet potato 0.05 (0.04) 4006 (2092) 11587 (8352) 100 100 100 100 100 100
Dpsp roots 8000 (3676) 27618 (13233) 38 100 100 100
Dpsp vines 14800 (8036) 21018 (12054) 70 100 100 100
Liters/season/farm
Milk 3211 (2473) 52317 (51723) 80 80 80 80 100 120
120 80
Maize price 130 100
Maize cost 90 110
Milk price 140 100
Farm size 120 80
Off farm income 150 100
Machakos Base system Alternative systems
a
System 1 System 2 System 3
Area Crop yield Net returns CC 2030 imz dpsplw dpsp dpsp1 dpsp12 RAP1 RAP2
Ha/season/farm kg/ha/season KSh/ha % of base system
Mixed 0.95 (1.39) 1187 (1631) 7085 (13313) 80 80 80 80 80 80
Maize 0.78 (0.79) 1597 (1624) 12704 (16996) 74 95 74 74 74 74
Beans 0.44 (0.59) 1390 (1374) 24658 (17942) 74 74 74 74 74 74
Vegetables 0.75 (1.00) 4121 (3369) 40718 (139490) 100 100 100 100 100 100
Napier Grass 1.49 (3.10) 12318 (14435) 11310 (18146) 80 80 80 80 80 80
Dpsp roots 7100 (4501) 24475 (16204) 42 100 100 100
Dpsp vines 12600 (9013) 18900 (13520) 83 100 100 100
Liters/season/farm
Milk 1784 (1992) 39238 (48208) 80 80 80 80 100 120 120 80
Maize price 130 100
Maize cost 90 110
Milk price 140 100
Farm size 120 80
Off farm income 150 100
a
CC = climate change, imz = improved maize, dpsp = dual purpose sweet potato, dpsplw = low yielding dpsp, dpsp1 = dpsp with 100% of base milk yield under CC,
dpsp12 = dpsp with 120% of base milk yield under CC, RAP = Representative Agricultural Pathway (as explained in the text).
88 L. Claessens et al. / Agricultural Systems 111 (2012) 8595
as an adapted technology used with the perturbed climate. We can
compare the outcomes associated with each system to quantify the
impacts of climate change, without adaptation and with adapta-
tion. Using the TOA-MD model, impacts that can be simulated in-
clude changes in farm income or poverty rates, as well as other
environmental or social outcomes of interest such as changes in
greenhouse gas emissions or human health.
It is important to note that in most analyses of climate change
adaptation, it is assumed all farms use the adapted technology.
However, that assumption is not appropriate in most cases, be-
cause the population of farms is heterogeneous (see for example
Salasya and Stoorvogel, 2010). Farms will be impacted differently
by climate change, and all farms are not necessarily adversely im-
pacted. In this study we use the TOA-MD model, which simulates
technology adoption together with impact to simulate the follow-
ing alternative scenarios:
The impact of climate change without adaptation, i.e., assuming
all farmers use the base technology (compare systems 1 and 2).
The impact of the adapted technology, when farmers choose
whether to adopt the adapted technology under the perturbed
climate (compare systems 2 and 3 with economically rational
adoption).
In the TOA-MD model, a farmer at a site s using a production
system h (dened as a combination of technology, climate and
RAP) earns per-hectare returns each period equal to v
t
= v
t
(s, h).
Over T time periods, system h provides a discounted net return of
Vs; h
X
T
t1
d
t
v
t
s; h; 1
where d
t
is the relevant discount factor. When the production
system changes, because of a change in technology or climate or
both, expected returns at each site also change. The effect on a
farms returns of changing from system j to system k is x(s, j, k) =
V(s, j) V(s, k). Thus, if x(s, j, k) is positive it represents the loss,
or opportunity cost, associated with switching from system j to sys-
tem k, and if negative it represents the gain from switching from
system j to k. Dene the density u(x|j, k) as the spatial distribution
of gains or losses in the population of farms indexed by s. The per-
centage of farms with x(s, j, k) < a (with a an amount in, e.g. dollars
per hectare) is
ra; j; k 100
Z
a
1
uxjj; k dx: 2
Two interpretations of (2) are possible in the context of the
TOA-MD model.
3.2.1. Technology adoption
In the standard technology adoption analysis, farms may be able
to choose to continue using system j which embodies one type of
technology (say, the base technology dened above), or to switch
to system k which embodies a different technology (say, the
adapted technology). In this case of voluntary technology adoption,
note that a farm will switch if x(s, j, k) = V(s, j) V(s, k) < a, which
implies V(s, j) < V(s, k) + a. Thus r(a, j, k) can be interpreted as the
proportion of adopters of system k when they experience the gain
or loss x(s, j, k) from switching, and are also given a payment (if po-
sitive) or made to pay a penalty or tax (if negative) of a dollars per
hectare to switch. If x(s, j, k) < 0, the farmer will switch from j to k
without any positive incentive payment, and r(0, j, k) is interpreted
as the adoption rate that would occur without any incentive pay-
ment or penalty. If a government or other entity wants to encour-
age additional adoption, a positive incentive can be offered to
adopters, in which case the adoption rate is r(a, j, k) > r(0, j, k) for
a > 0. Conversely, to discourage adoption, a negative incentive
(i.e., a penalty or tax) can be imposed on adopters (say, a tax on
the decrease in environmental services associated with the use of
system k). This type of adoption analysis can be used to assess
the rate of adoption of an adapted technology under climate
change by setting a = 0, j = 2 and k = 3, i.e., by comparing systems
2 and 3 as dened above.
3.2.2. Climate impact assessment and adaptation
In a climate impact assessment, farmers are initially using system
1 as dened above, and if no adaptation is possible, their only op-
tion is to use the same technology when the climate changes (sys-
tem 2 as dened above). In this type of analysis, Eq. (2) can be
interpreted as showing the proportion of farms with losses less
than a, i.e., with x(s, 1, 2) < a. Thus, r(0, 1, 2) is interpreted as the
proportion of farms that are positively impacted, and 1 r(0, 1, 2)
is interpreted as the proportion of farms that is negatively im-
pacted. For a > 0, r(a, 1, 2) r(0, 1, 2) can be interpreted as the pro-
portion of farms that have losses between zero and a, for a < 0,
r(0, 1, 2) r(a, p1, 2) can be interpreted as the proportion of farms
with gains between zero and a.
When farmers are confronted with an environmental change
such as climate change, they may choose a different technology
that performs better in the new environment, if one is available.
This adaptation to climate change can be evaluated by considering
an adapted technology (e.g., the system 3 described above) under
the perturbed climate. Following the system denitions presented
above, 1 r(0, 1, 2) is the proportion of farms whose net returns are
impacted negatively by climate change, and we expect that with
adaptation this proportion would decrease so that 1 r(0, 1, 2) >
1 r(0, 1, 3). Antle (2011a) shows that in an economic adaptation
analysis, accurate measurement of the economic, environmental
and social impacts of technology adoption must take into account
the statistical correlation between factors affecting adoption (e.g.,
economic returns) and the other outcomes of interest. The TOA-
MD model is designed to incorporate these correlations into the
simulation of impacts on farm income and income-based poverty.
In the analysis presented below, we use these relationships to as-
sess the distributional impacts of technology adoption.
3.3. Climate change projections
For climate change projections to 2030, we used data from the
IPCC Fourth Assessment report (2007, TYN SC 2.0 dataset (Mitchell
et al., 2004)) and spatial and temporal downscaling techniques as
described in Thornton et al. (2009a, 2010) and Jones et al. (2009).
Data from two Global Circulation Models (GCMs) and two scenar-
ios from the Special Report on Emissions Scenarios (SRES, Nakice-
novic et al., 2000) were used. There are considerable differences
between different GCMs and SRES scenarios in terms of projected
changes in temperatures and rainfall. In this study we used a com-
bination of the HadCM3 (Mitchell et al., 1998) and ECHam4 (Roe-
ckner et al., 1996) GCMs, and SRES A1FI and B1 (very high and
low emissions) on the basis of the arguments provided in Thornton
et al. (2009a): An analysis by McHugh (2005) on multi-model
trends in rainfall for East Africa suggests that certain GCMs are bet-
ter able to simulate observed rainfall patterns in this region than
others. ECHam4 is a wet model while HadCM3 is a drier model.
For emission scenarios, a low-emission scenario (B1) and a high-
emission scenario (A1F1) were chosen, which span the range of
best estimates of temperature change to 20902099 relative to
19801999 (IPCC, 2007). While the differences between the SRES
scenarios become much more marked by 2100 and beyond, in
terms of impacts on temperature, even by 2030 there are differ-
ences between the A1FI and B1 scenarios depending on the GCM
used. The choices of combination of GCM and emission scenario
L. Claessens et al. / Agricultural Systems 111 (2012) 8595 89
thus span a wide range of projected temperature and rainfall
changes.
3.4. Effects of climate change on productivity and simulated
adaptation strategies
A summary of model settings for the base, climate change and
adaptation scenarios is given in Table 1. To simulate the potential
effects of climate change on crop yields, crop growth simulation
models as currently implemented in version 4.0 of the Decision
Support System for Agrotechnology Transfer (DSSAT, Jones et al.,
2003; ICASA, 2007) were used for maize and beans (Thornton
et al., 2009a). The yields used are the mean simulated yields of
30 replications (weather years) of four combinations of the two
GCMs and SRES scenarios described above. Regarding the impacts
of carbon fertilization on crop yields, there is considerable on-
going debate as to the size of the effects on the physiology of crops
(Ainsworth et al., 2008; Boote et al., 2011), and there are more
uncertainties concerning yield benets in low-input, rainfed sub-
sistence production systems such as those that prevail in the study
region (Thornton et al., 2010). There are also substantial knowl-
edge gaps concerning the impacts of CO
2
concentrations and how
they may interact with changing ozone concentrations and with
other biotic and abiotic stresses (Challinor et al., 2009; Boote
et al., 2011). Carbon fertilization effects are incorporated to some
degree in the DSSAT models, but given the uncertainty surrounding
this issue, they were not used. Carbon dioxide concentrations were
held constant at 330 ppm (Thornton et al., 2009a).
For both crops and both study areas, a declining yield trend is
projected, caused by an increased temperature without adequate
rainfall (Thornton et al., 2010). A range of feasible adaptation strat-
egies was simulated based on stakeholder consultations (farmers,
extension agents, policy makers) in the region. The introduction
and adoption of an improved (heat/drought tolerant) maize vari-
ety, bringing yields back to 95% of the base level, is tested as an
adaptation strategy (imz in Table 1). For sweet potato, Napier
grass and the mixed crops, no crop growth simulation models
are currently available. For Napier grass and mixed crops, a 20%
yield decline was estimated. The vegetables in Machakos are irri-
gated and no yield reduction was assumed under climate change.
Drought and heat tolerant varieties of sweet potato as well as
dual-purpose varieties (roots for food and vines for livestock feed)
are currently being promoted in Kenya and in general sweet pota-
toes are known as a reliable food security crop giving good yields in
marginal climatic and soil conditions (Diop, 1998; Bovell-Benja-
min, 2007; Andrade et al., 2009). For Vihiga, no yield decline was
assumed under climate change for (non dual-purpose) sweet pota-
to. Adoption of dual-purpose sweet potato (DPSP) is tested as an
adaptation strategy and treated essentially as a new crop in both
study areas (as in Claessens et al., 2009). For Vihiga, yield data
for DPSP were taken from on farm eld trials in nearby locations
in western Kenya and eastern Uganda (Ndolo et al., 2007; Mwanga
et al., 2006). For Machakos, yield data for DPSP were obtained from
on farm eld trials in Kibwezi, which is centrally located in the
study area. We tested both the lowest quartile and average yields
from these datasets (dpsplw and dpsp respectively in Table 1). As
an example, the option of substituting half of the area under the
mixed crops with DPSP was tested for both study areas.
The physiological effects of increased temperatures on animals
and their productivity are well known (Blaxter, 1962; SCA, 1990).
However, little information is available at the local level to be able
to assess impacts on animal productivity in different systems,
especially in the tropics. The main effect of increased temperature
is a decreased feed intake due to the inability of animals to dissi-
pate the heat associated with digestion and metabolism of feeds
(Blaxter, 1962). Above the thermo-neutral zone, intake may
decrease to a point where signicant reductions in milk yield can
be observed (for systems with little control over the exposure of
animals to climate typical of Kenya). For the climate change sce-
nario, as a proxy for heat stress, we tested a 20% milk yield reduc-
tion for both study areas, caused by a declined intake of fodder and
their respective yield declines (less on farm produced fodder avail-
able for feed). Introducing DPSP as an adaptation strategy increases
both quantity and quality (mainly crude protein content) of on
farm produced animal feed and can substantially improve milk
yields and farm incomes, even under the current climate (Claes-
sens et al., 2009). In combination with an anticipated genetic
improvement of the animals through both natural and articial
selection as adaptation strategy (Ser et al., 2008; Thornton et al.,
2009b), we tested bringing the milk yield back to 100% and 120%
of the base production for both study areas (dpsp1 and dpsp12
respectively in Table 1).
In order to obtain a more in-depth analysis of different sub-
groups of farmers in the two study areas, the data from Table 1
were disaggregated in the TOA-MD model for farmers with and
without dairy production. A third category of farmers, those with
access to irrigation, was differentiated for Machakos.
3.5. Representative agricultural Pathways and socio-economic
scenarios
Most agricultural climate impact assessments in the published
literature have evaluated the impacts of climate change on current
or adapted technologies within historical or present socio-eco-
nomic conditions. It would be desirable to evaluate potential cli-
mate change impacts under plausible future socio-economic
scenarios that are consistent with the assumptions that were used
to generate climate change simulations. Until now, the most
widely used scenarios for climate impact assessment are the ones
presented in the Special Report on Emissions Scenarios (SRESs,
Nakicenovic et al., 2000). However, these scenarios had a number
of important methodological limitations; in an attempt to improve
the approach, the global climate modeling and impact assessment
communities are developing two new concepts, Representative
Concentration Pathways or RCPs and Shared Socio-economic Path-
ways or SSPs (Carter et al., 2011). The concept of Representative
Agricultural Pathways (RAPs) has been proposed as a way to ex-
tend these scenario concepts to be more relevant to agricultural
models (Antle, 2011b). RAPs include global economic conditions,
such as rates of growth in aggregate agricultural productivity, as
well as region-specic agricultural and economic development
conditions that can be used in disaggregate models such as the
TOA-MD model. However, as of this writing, a consensus has not
been reached on the details of SSPs, and as a result fully developed
RAPs are not yet available. In this study, in addition to current con-
ditions we propose two RAPs that are broadly consistent with the
types of SSPs that are currently under development. The proposed
SSPs correspond to future worlds characterized by different de-
grees of adaptation challenges and mitigation challenges. Here we fo-
cus on RAPs that correspond to different degrees of adaptation
challenges, because we are not going to directly incorporate anal-
ysis of greenhouse gas mitigation into the analysis. Thus we con-
sider two RAPs that correspond to low and high degrees of
adaptation challenges, based on the following qualitative
storylines:
3.5.1. RAP1: low adaptation challenges
Kenya follows a more positive economic development trajec-
tory than the past 30 years, with higher rates of economic growth,
movement of labor out of agriculture into other sectors, reductions
in rural household size, and increases in farm size. Investments in
transportation and communication infrastructure, and more open
90 L. Claessens et al. / Agricultural Systems 111 (2012) 8595
trade and liberalized domestic policies lead to higher real prices for
traded agricultural commodities such as maize in Kenya, in part
due to the projected higher real prices of maize due to global de-
mand growth and slowing global productivity growth due to cli-
mate change and other factors. Policy changes and infrastructure
improvements also lead to lower real prices of critical agricultural
inputs such as mineral fertilizer and improved seeds.
3.5.2. RAP2: high adaptation challenges
Even greater than the current situation. Kenya continues to
experience a low rate of economic growth. Population growth rates
remain high and rural populations increase, farm size declines, rur-
al household sizes increase. Transportation infrastructure deterio-
rates, trade policy discourages exports so that prices to farmers
remain at current levels, but policy imposes high taxes on imports
of critical inputs such as fertilizers. Soil fertility and agricultural
productivity continue to decline to an even lower level equilibrium
than was observed in the early part of the 21st Century.
Corresponding to these RAPs, we develop parametric scenarios
for 2030 as shown in Table 1. We translated the qualitative RAPs
into quantitative scenarios by making assumptions about com-
pound annual rates of change from the baseline values from
2012 to 2030 for prices, costs of production and farm size in the
range of 1% per year. For maize price changes, we followed the re-
sults of global modeling studies which indicate that the combined
effects of demand growth and climate change are likely to increase
real maize prices from the present until at least mid-century
(Foresight, 2011). For RAP1, in addition to the socio-economic
parameter settings in Table 1, more land was allocated to maize
compared to the base system, consistent with the increased ex-
pected returns from maize under RAP1 (half of the area under
mixed and Napier grass in the base system was converted to maize
for both study areas). In order to be consistent with the qualitative
storylines of the RAPs, we analyzed the impacts of climate change
under both RAP1 and RAP2 scenarios but only simulated adapta-
tion strategies (imz and dpsp) for RAP1 (low adaptation
challenges).
3.6. Effects of climate change, adaptation strategies and
socio-economic scenarios
We start our analysis by assessing the impacts of climate change
and adaptation, keeping socio-economic parameters of the base and
alternative systems constant. Results of the TOA-MDanalysis aggre-
gated across farmtypes for both study areas are shown in Fig. 2. The
interpretation of the curves representing the farm population is as
follows: The point where a curve crosses the x-axis shows the per-
centage of farms that gain from the scenario (i.e., the value r(0, j, k)
discussed in Section 3.2). Accordingly, the points on a curve to the
left of where it crosses the x-axis showthe percentage of farms with
gains (i.e., negative losses) greater than the amount shown on the
vertical axis. Conversely, points to the right of where a curve crosses
the x-axis show the percentage of farms with losses less than or
equal to the amount on the vertical axis. Fig. 2 shows that climate
change is projected to have a negative economic impact on 76% of
the farmers in Vihiga and on 62% in Machakos. By testing different
adaptation strategies with the TOA-MD model, we can simulate
aggregate economic impacts on the farm population in each of the
study areas. Fig. 2 shows that, in the aggregate, the different adap-
tation strategies simulated have a higher impact on the farm popu-
lation in Vihiga than in Machakos. The best adaptation strategy
(dpsp12) can bring back the percentage of farmers loosing from
Fig. 2. Economic impact of climate change and simulated adaptation strategies on farmers in Vihiga and Machakos, Kenya. Notation of legend as in Table 1.
L. Claessens et al. / Agricultural Systems 111 (2012) 8595 91
climate change from76% to 37% in Vihiga, but only from62% to 50%
in Machakos.
As explained in the methodology, the TOA-MD model can also
be used to analyze the effects of different adaptation strategies
on poverty rates, net losses (calculated here as the percentage loss
relative to the base system agricultural income), and can calculate
adoption rates (economic feasibility) of these strategies. Both
disaggregated and aggregated impacts of climate change and the
simulated adaptation strategies are shown in Table 2. The base
poverty rate (expressed as % of farm population living on <$1 a
day) in Machakos is higher than in Vihiga (73% vs. 62%), but in
terms of net losses in percent of mean agricultural income, Vihiga
and Machakos are equally negatively impacted (2732% loss). The
introduction of an improved maize variety as adaptation strategy
(bringing back yields to 95% of the base level), has a profound ef-
fect in Machakos, offsetting the negative effects of climate change
at the aggregate level (poverty rates are back to the base level of
73% and there is an aggregate net gain of 20%; dairy farmers are
still losing tough, as they have less land allocated to maize than
non-dairy farmers). In Vihiga this introduction of improved maize
brings back the poverty rates from 69% to 65% but farmers are still
losing compared to the base system. Substituting half of the mixed
system with low yielding DPSP (dpsplw) hardly reduces the per-
centage of farmers that are negatively affected by climate change
in Vihiga but reduces this from 62% to 57% in Machakos (Fig. 2).
This strategy has a similar effect than the improved maize option
in Machakos. Increasing the average yield of DPSP to the observed
levels but keeping the loss in milk yield at 20% (dpsp), has a posi-
tive effect in both study areas (negatively affected farmers to 63%
in Vihiga and to 54% in Machakos). On aggregate, Vihiga is still
negatively impacted by climate change but non-dairy farmers are
already gaining from this strategy (the poverty rate is back to the
base level and there is a net revenue gain of 7%). By increasing milk
yields to 100% and 120% of the base level, Machakos has limited
additional gains, whereas in Vihiga the percentage of negatively
affected farmers goes down to 50% and 37% respectively and adop-
tion rates (dened here as the percentage of farmers economically
beneting from an adaptation strategy) go up to 84% in the aggre-
gate and 90% for the dairy farmers. In general, this analysis indi-
cates that introduction of an improved maize variety or a low
yielding DPSP in the cropping system of Machakos would be suf-
cient to offset the negative impacts of climate change. For Vihiga,
average yielding DPSP together with improved feed and/or live-
stock breeds that can produce 100% of the base milk yield under
climate change, are needed to fully offset the impacts of climate
change. The disaggregated results in Table 2 show that farmers
with dairy in Vihiga benet relatively more from increases in milk
yield than dairy farmers in Machakos. Vihiga has a larger percent-
age of dairy farmers in the population (62% vs. 15% in Machakos)
and higher base milk yields and net returns (Table 1). Accordingly,
the increased milk yield does not affect the adoption rates in
Machakos as much as it does in Vihiga.
By introducing the socio-economic scenarios (RAPs) in the anal-
ysis, we can look at the effect of these scenarios on the impacts of
climate change and adaptation strategies (imz and dpsp for RAP1
only as explained above). For the optimistic low adaptation chal-
lenges scenario (RAP1), the base poverty rates are much lower com-
pared to the previous analysis (41% vs. 62% for Vihiga, 60% vs. 73%
Table 2
Impacts of climate change, simulated adaptation strategies and socio-economic scenarios on farmers in Vihiga and Machakos.
Vihiga Machakos
Scenario No dairy Dairy Total No dairy Dairy Irrigated Total
Poverty rate (% of farm population living on <$1 per day)
base 85 38 62 85 43 54 73
CC 89 49 69 89 51 57 78
imz 87 42 65 85 44 50 73
dpsplw 88 42 66 85 44 50 73
dpsp 85 41 63 83 43 50 71
dpsp1 85 36 60 83 41 49 71
dpsp12 85 30 58 83 38 48 70
RAP1 base 65 17 41 72 30 46 60
RAP1 CC 71 18 44 77 33 47 64
RAP1 imz 66 15 41 70 27 40 58
RAP1 dpsp 65 15 40 69 27 40 57
RAP2 base 89 48 68 91 50 57 79
RAP2 CC 91 50 71 93 53 57 81
Net loss (percentage of mean agricultural income in base system)
CC 26 27 27 32 31 33 32
imz 8 11 11 16 6 50 20
dpsplw 13 12 12 23 5 49 23
dpsp 7 9 6 31 3 51 27
dpsp1 7 5 6 31 7 65 34
dpsp12 7 23 21 31 19 80 43
RAP1 CC 30 5 8 35 11 12 19
RAP1 imz 4 5 3 23 8 44 27
RAP1 dpsp 2 6 5 27 8 42 28
RAP2 CC 26 7 10 25 14 8 16
Adoption rate (percentage of farm population)
imz 62 52 56 54 51 51 53
dpsplw 52 51 51 58 53 50 56
dpsp 74 57 64 61 55 51 59
dpsp1 74 77 77 61 65 55 61
dpsp12 74 90 84 61 74 59 63
RAP1 imz 71 56 62 57 54 52 56
RAP1 dpsp 73 58 64 60 55 51 58
92 L. Claessens et al. / Agricultural Systems 111 (2012) 8595
for Machakos). Climate change is bringing up the poverty rates to
44% and 64% and causes aggregate net losses of 8% and 19%. Both
adaptation strategies simulated are able to offset the negative im-
pacts of climate change in both study areas, which is consistent
with the low adaptation challenges concept under RAP1. The effect
of the adaptation strategies on poverty rates is less important be-
cause increased off-farm income in the scenario dominates the ef-
fect of increased production under climate change adaptation. For
the negative high adaptation challenges scenario (RAP2), base pov-
erty rates are higher than for the previous analysis (68% and
79%). Climate change is bringing the poverty rates up to 71% and
81% and causes aggregate net losses of 10% and 16%. No adaptation
strategies were simulated, consistent with this scenario.
4. Discussion
In this paper we proposed the TOA-MD model as a methodology
to test climate change adaptation strategies with limited data, that
is with data that are typically available in countries where small-
scale, semi-subsistence agricultural systems predominate. There
is however an inherent weakness in using this type of data, which
limits the condence in modeled outcomes. The real issue of
course is how much certainty one can gain with expensive and
time consuming gathering of additional data or parameterizing
more complex models. For both examples, existing datasets (farm
surveys) were used to characterize the base systems. Various
assumptions had to be made; hence substantial uncertainty is
associated with the parameterization of the climate change, adap-
tation and socio-economic scenarios.
The results in Table 2 for the irrigated system in Machakos pro-
vide some insight into the challenges researchers face in making
reliable estimates of impact and adaptation with limited data. In
the case of the irrigated farms, Table 2 shows a very large positive
impact of the adapted technologies. However, this result appears
to be due to the fact that the maize and other crop yields are
unusually high on these farms, perhaps because the available irri-
gation is benetting not only vegetables but other crops as well.
As a result, the estimates of climate change impacts on maize,
and the potential benets of an adapted maize variety or the
introduction of dual-purpose sweet potato based on data repre-
senting rainfed systems are not appropriate if these crops are
also benetting from the available irrigation. However, estimates
of climate change impacts on these crops in irrigated systems
are not available, thus illustrating the data challenges that
researchers can face when highly heterogeneous conditions are
encountered.
As mentioned in the methodology, there is already a high level
of variability in climate projections between different GCMs and
SRES scenarios and we chose to use a specic combination of GCMs
and SRES spanning a wide range of projected temperature and
rainfall changes. More uncertainty is added when downscaling
GCM data to the appropriate spatial and temporal resolution to
be used in the DSSAT crop growth simulation models. And there
is the parameterization of the crop growth models itself which is
generalizing genetic coefcients for maize and beans. Hence it is
worth stressing that the projected yield changes to 2030 should
merely be seen as a mean response in relation to all possible com-
binations of GCMs and SRES scenarios, making use of the best
information we can currently obtain from crop growth simulation
and downscaled climate projections.
It is also likely that the impacts of climate change on productiv-
ity are underestimated because the effects of increasing climate
and weather variability have not been included (and this is one
of the biggest constraints in rainfed agriculture). Methods are
becoming available to evaluate the effects of climate variability
at the agricultural system level (Cooper et al., 2008), but there
are still challenges to use the relatively coarse data of GCMs at
an appropriate spatial and temporal resolution (Baethgen, 2010;
Thornton et al., 2010).
CO
2
fertilization effects were not included in our analysis based
on the many uncertainties associated with it (see methodology and
Thornton et al., 2010).
Productivity changes for crops other than maize and beans and
for milk were estimated based on the trends for maize and beans
and information we obtained from the literature. As our analysis
of dual-purpose sweet potato and dairy illustrates, changes in pro-
ductivity can be subjected to a sensitivity analysis. If the results are
found to be sensitive to the inherent uncertainty in the available
data, then further analysis is called for. For example, in the case
of dairy productivity, it may be possible to improve the quality of
estimates using more detailed livestock models that account for
the effects of climate and feed quantity and quality on livestock
productivity (e.g. Herrero et al., 2008).
The TOA-MD methodology utilized here is based on economic
feasibility (expected protability). Accordingly, the simulated
adoption rates can be interpreted as providing an upper bound
on likely adoption rates because other factors that have been found
to be important in technology adoption, such as nancial con-
straints, risk aversion and cultural considerations are not taken
into account (see Sunding and Zilberman, 2001; Claessens et al.,
2009).
The main reasons why introduction of DPSP as an adaptation
strategy is economically viable for a relatively large percentage
of farms appear to be the relatively high yields, net returns and
the positive effect of increased feed quality on milk production
and income. The researchers who collected the DPSP yield data
from on farm eld trials (Mwanga et al., 2006; Ndolo et al., 2007)
suggest that the yields they observed are higher than most farmers
would achieve because crop management and soil conditions
would be less favorable than in the farm trial sites. A sensitivity
analysis was therefore conducted using low yielding DPSP as one
of the adaptation strategies.
Both Vihiga and Machakos are broadly representative of agri-
cultural systems in other parts of East Africa. By using an agricul-
tural systems classication like the one by Ser and Steinfeld
(1996) or Robinson et al. (2011), e.g., results of the analysis, when
interpreted with caution, could be scaled up to the agricultural sys-
tem and regional scale levels.
5. Conclusion
The development and application of relatively simple and reli-
able methods for assessing the impacts of climate change and
adaptation strategies at the agricultural system and/or household
level are needed to provide timely recommendations on the poten-
tial impacts of alternative technologies and policies. In this paper,
the TOA-MD model was presented as a method to evaluate the im-
pacts of climate change and the economic viability of adaptation
strategies using the kinds of data that are typically available in
countries where semi-subsistence systems are important. The
method was applied to the mixed crop-livestock systems of the
Vihiga and Machakos study areas in Kenya. With a combination
of simulated and estimated changes in crop and livestock produc-
tivity, the economic impacts of climate change to 2030 were ana-
lyzed. Climate change is projected to have a negative economic
impact on 76% of the farmers in Vihiga and on 62% in Machakos.
Different adaptation strategies were tested by changing crop and
livestock productivity under climate change and by introducing so-
cio-economic scenarios based on Representative Agricultural Path-
ways. The analysis suggests that introducing an improved maize
L. Claessens et al. / Agricultural Systems 111 (2012) 8595 93
variety or low yielding DPSP in the cropping system of Machakos
may be sufcient to offset the negative effects of climate change,
whereas improved feed quantity and quality in combination with
improved livestock breeds that can perform better under climate
change are an additional requirement for adaptation to climate
change and improvement of farmers livelihoods in Vihiga. As in
all scenario studies using models, and especially in the context of
climate change, various assumptions and uncertainties are associ-
ated with using the proposed approach and results should be inter-
preted with caution. Despite these limitations, the methodology
presented in this study shows the potential to yield new insights
into the way that realistic adaptation strategies could improve
the livelihoods of smallholder farmers operating in the mixed
crop-livestock systems in East Africa and other parts of the world.
Acknowledgments
This research was conducted within the framework of a project
funded by BMZ/GIZ on behalf of the Government of the Federal
Republic of Germany. This study used data from the PROSAM
and NUTMON projects, co-funded by the Ecoregional Fund to Sup-
port Methodological Initiatives of the Dutch government. Support
from the CGIAR Research Program on Climate Change, Agriculture
and Food Security (CCAFS), and also the Agricultural Model Inter-
comparison and Improvement Project (AgMIP), is acknowledged.
We thank the editor and two anonymous reviewers for their con-
structive comments on an earlier version of the manuscript.
References
Adger, W.N., Huq, S., Brown, K., Conway, D., Hulme, M., 2003. Adaptation to
climate change in the developing world. Prog. Develop. Stud. 3 (3), 179
195.
Ainsworth, E.A., Leakey, A.D.B., Ort, D.R., Long, S.P., 2008. FACE-ing the facts:
inconsistencies and interdependence among eld, chamber and modeling
studies of elevated CO
2
impacts on crop yield and food supply. New Phytol. 179,
59.
AFRENA, 1998. Agroforestry Potential for Land Use Systems in the Bimodal
Highlands of Eastern Africa. AFRENA Report No. 3, ICRAF, Nairobi, Kenya.
Andrade, M., Barker, I., Cole, D., Dapaah, H., Elliott, H., Fuentes, S., Grneberg, W.,
Kapinga, R., Kroschel, J., Labarta, R., Lemaga, B., Loechl, C., Low, J., Lynam, J.,
Mwanga, R., Ortiz, O., Oswald, A., Thiele, G., 2009. Unleashing the potential of
sweetpotato in sub-Saharan Africa: current challenges and way forward.
International Potato Center (CIP), Lima, Peru. Working Paper 20091, p. 197.
Antle, J.M., 2011a. Parsimonious multi-dimensional impact assessment. Am. J. Agric.
Econ. 93 (5), 12921311.
Antle, J.M., 2011b. Representative Agricultural Pathways for Model Inter-
comparison and Impact Assessment. <www.agmip.org/RAPs>.
Antle, J.M., Valdivia, R.O., Crissman, C.C., Stoorvogel, J.J., Yanggen, D., 2005. Spatial
heterogeneity and adoption of soil conservation investments: integrated
assessment of slow formation terraces in the Andes. J. Int. Agric. Trade
Develop. 1 (1), 2953.
Antle, J.M., Stoorvogel, J.J., 2006. Predicting the supply of ecosystem services from
agriculture. Am. J. Agric. Econ. 88 (5), 11741180.
Antle, J.M., Valdivia, R.O., 2006. Modelling the supply of ecosystem services from
agriculture: a minimum-data approach. Australian J. Agric. Res. Econ. 50, 1
15.
Antle, J.M., Stoorvogel, J.J., 2008. Agricultural carbon sequestration, poverty and
sustainability. Environ. Dev. Econ. 13, 327352.
Antle, J.M., Diagana, B., Stoorvogel, J.J., Valdivia, R.O., 2010. Minimum-data analysis
of ecosystem service supply in semi-subsistence agricultural systems: evidence
from Kenya and Senegal. Australian J. Agric. Res. Econ. 54, 601617.
Baethgen, W.E., 2010. Climate risk management for adaptation to climate variability
and change. Crop Sci. 50, 7076.
Blaxter, K.L., 1962. The Energy Metabolism of Ruminants. Hutchingson Press,
London, UK.
Boote, K.J., Ibrahim, A.M.H., Latte, R., McCulley, R., Messina, C., Murray, S.C., Specht,
J.E., Taylor, S., Westgate, M.E., Glasener, K., Bijl, C.G., Giese, J.H., 2011. Position
statement on crop adaptation to climate change. Crop Sci. 51, 23372343.
Bovell-Benjamin, A., 2007. Sweet potato: a review of its past, present, and future
role in human nutrition. Adv. Food Nutr. Res. 52, 159.
Burke, M.B., Lobell, D.B., Guarino, L., 2009. Shifts in African crop climates by 2050,
and the implications for crop improvement and genetic resources conservation.
Glob. Environ. Change 19 (3), 317325.
Carter, T., Ebi, K., Edmonds, J., Hallegatte, S., Kriegler, E., et al., 2011. A Framework
for a New Generation of Socioeconomic Scenarios for Climate Change Impact,
Adaptation, Vulnerability, and Mitigation Research. <www.isp.ucar.edu/socio-
economic-pathways>.
Central Bureau of Statistics (CBSs), 2001. Population and housing census. Counting
our people for development, vol. 1. Population Distribution by Administrative
Areas and Urban Centres. Ministry of Finance and Planning, Nairobi, Kenya.
Central Bureau of Statistics (CBSs), 2003. Geographic dimensions of well-being in
Kenya. Where are the poor? From districts to locations vol. 1. CBS, Ministry of
Planning and National Development, Nairobi, Kenya.
Challinor, A.J., Ewert, F., Arnold, S., Simelton, E., Fraser, E., 2009. Crops and climate
change: progress, trends, and challenges in simulating impacts and informing
adaptation. J. Exp. Bot. 60 (10), 27752789.
Claessens, L., Stoorvogel, J.J., Antle, J.M., 2009. Ex ante assessment of dual-purpose
sweet potato in the crop-livestock system of western Kenya: a minimum-data
approach. Agric. Syst. 99 (1), 1322.
Claessens, L., Stoorvogel, J.J., Antle, J.M., 2010. Exploring the impacts of eld
interactions on an integrated assessment of terraced crop systems in the
Peruvian Andes. J. Land Use Sci. 5 (4), 259275.
Cooper, P.J.M., Dimes, J., Rao, K.P.C., Shapiro, B., Shiferawa, B., Twomlow, S., 2008.
Coping better with current climatic variability in the rain-fed farming systems
of sub-Saharan Africa: an essential rst step in adapting to future climate
change? Agric. Ecosyst. Environ. 126, 2435.
de Jager, A., Onduru, D., van Wijk, M.S., Vlaming, J., Gachini, G.N., 2001. Assessing
sustainability of low-external-input farm management systems with the
nutrient monitoring approach, a case study in Kenya. Agric. Syst. 69 (12),
99118.
de Jager, A., van Keulen, H., Mainah, F., Gachimbi, L.N., Itabari, J.K., Thuranira,
E.G., Karuku, A.M., 2004. Attaining sustainable farm management systems in
semi-arid areas in Kenya: few technical options, many policy challenges.
Final Project Report (draft). Wageningen University and Research Center, The
Netherlands.
Diop, A., 1998. Storage and processing of roots and tubers in the tropics. In:
Calverley, D.J.B. (Ed.), Food and Agriculture Organization of the United Nations,
Agro-Industries and Post-Harvest Management Service. Agricultural Support
Systems Division, Food and Agriculture Organization, Rome, Italy, pp. 3850.
Foresight, 2011. The Future of Food and Farming. Final Project Report. The
Government Ofce for Science, London.
Gachimbi, L.N., van Keulen, H., Thuranira, E.G., Karuku, A.M., de Jager, A., Nguluu, S.,
Ikombo, B.M., Kinama, J.M., Itabari, J.K., Nandwa, S.M., 2005. Nutrient balances
at farm level in Machakos (Kenya), using a participatory nutrient monitoring
(NUTMON) approach. Land Use Policy 22 (1), 1322.
Herrero, M., Thornton, P.K., Kruska, R., Reid, R.S., 2008. Systems dynamics and the
spatial distribution of methane emissions from African domestic ruminants to
2030. Agric. Ecosyst. Environ. 126, 122137.
Hijmans, R.J., 2003. The effect of climate change on global potato production. Am. J.
Potato Res. 80 (4), 271280.
ICASA, 2007. The International Consortium for Agricultural Systems Applications
Website. <www.icasa.net>.
Immerzeel, W., Stoorvogel, J.J., Antle, J.M., 2008. Can payments for ecosystem
services secure the water tower of Tibet? Agric. Syst. 96 (13), 5263.
IPCC (Intergovernmental Panel on Climate Change), 2007. Climate Change 2007:
Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the
Fourth Assessment Report of the Intergovernmental Panel on Climate Change.
Cambridge University Press, Cambridge, UK.
Jones, J.W., Hoogenboom, G., Porter, C.H., Boote, K.J., Batchelor, W.D., Hunt, L.A.,
Wilkens, P.W., Singh, U., Gijsman, A.J., Ritchie, J.T., 2003. The DSSAT cropping
system model. Eur. J. Agron. 18, 235265.
Jones, P.G., Thornton, P.K., 2003. The potential impacts of climate change on maize
production in Africa and Latin America in 2055. Glob. Environ. Change 13 (1),
5159.
Jones, P.G., Thornton, P.K., Heinke, J., 2009. Software to generate characteristic daily
weather data at any resolution using downscaled climate model data from the
IPCCs fourth assessment. BMZ Project Discussion Paper, ILRI, Nairobi.
Kenya National Bureau of Statistics, 2010. 2009 Kenya Population and Housing
Census: Volume 1A: Population Distribution by Administrative Units.
Government of Kenya, Nairobi, Kenya.
Kurukulasuriya, P., Mendelsohn, R., 2006. Crop Selection: Adapting Climate Change
in Africa. CEEPA Publication 26. <http://www.ceepa.co.za/docs/CDPNo26.
pdf>.
Lal, R., 2010. Enhancing eco-efciency in agro-ecosystems through soil carbon
sequestration. Crop Sci. 50, 120131.
McHugh, M.J., 2005. Multi-model trends in East African rainfall associated with
increased CO
2
. Geophys. Res. Lett. 32, L01707.
Mendelsohn, R., Morrison, W., Schlesinger, M.E., Andronova, N.G., 2000. Country-
specic market impacts of climate change. Climatic Change 45 (34), 553
569.
Mitchell, J.F.B., Johns, T.C., Senior, C.A., 1998. Transient response to increasing
greenhouse gases using models with and without ux adjustment. Hadley
Centre Technical Note 2. UK Met Ofce, London Road, Bracknell, UK.
Mitchell, T.D., Carter, T.R., Jones, P.D., Hulme, M., New, M., 2004. A comprehensive
set of high-resolution grids of monthly climate for Europe and the globe: the
observed record (19012000) and 16 scenarios (20012100). Tyndall Centre for
Climate Change Research Working Paper 55.
MoA, 1987. Fertilizer Use Recommendation Project (Phase I). Ministry of
Agriculture, National Agricultural Research Laboratories, Nairobi, Kenya.
Mora-Vallejo, A., Claessens, L., Stoorvogel, J.J., Heuvelink, G.B.M., 2008. Small scale
digital soil mapping in Southeastern Kenya. Catena 76, 4453.
94 L. Claessens et al. / Agricultural Systems 111 (2012) 8595
Mwanga, R.O.M., Stevenson, R.C., Yencho, C.G., 2006. Development of High Yielding
Multiple Resistant Sweet Potato Germplasm. Harvest Plus Research Progress
Report April 1 2005 through March 31 2006.
Nakicenovic, N. et al., 2000. IPCC Special Report on Emissions Scenarios (SRESs).
Cambridge University Press. <http://www.grida.no/publications/other/ipcc_sr/
?src=/climate/ipcc/emission/>.
Nalukenge, I., Antle, J.M., Stoorvogel J.J., Claessens, L., 2006. Payments for
environmental services (PESs) as a new approach to tackle land degradation
in Pallisa District. Workshop Report USAID (Soil Management CRSP) and FAO,
Pallisa, Uganda.
Ndolo J., P., Gethi, J., Agili, S., 2007. Potential of improved orange eshed sweet
potato clones in different agro-ecologies in Kenya. In: Eighth African Crop
Science Society Conference, El-Minia, Egypt, 2007 October 2731.
Nelson, G.C., Rosegrant, M.W., Koo, J., Robertson, R., Sulser, T., Zhu, T., Ringler, C.,
Msangi, S., Palazzo, A., Batka, M., Magalhaes, M., Valmonte-Santos, R., Ewing, M.,
Lee, D., 2009. Climate change: impact on agriculture and costs of adaptation.
Food Policy Report 21, International Food Policy Research Institute (IFPRI),
Washington DC, USA.
Onduru, D.D., de Jager, A., Gachini, G., Diop, J.M., 2001. Exploring new pathways for
innovative soil fertility management in Kenya managing Africas soils. IIED-
Drylands Programme.
Robinson, T.P., Thornton, P.K., Franceschini, G., Kruska, R.L., Chiozza, F., Notenbaert,
A., Cecchi, G., Herrero, M., Epprecht, M., Fritz, S., You, L., Conchedda, G., See, L.,
2011. Global livestock production systems. Rome, Food and Agriculture
Organization of the United Nations (FAOs) and International Livestock
Research Institute (ILRI), p. 152.
Roeckner, E., Oberhuber, J.M., Bacher, A., Christoph, M., Kirchner, I., 1996. ENSO
variability and atmospheric response in a global coupled atmosphere-ocean
GCM. Clim. Dyn. 12, 737754.
Rosenzweig, C., Jones, J.W., Hateld, J.L., Ruane, A.C., Boote, K.J., Thorburn, P., Antle,
J.M., Nelson, G.C., Porter, C., Janssen, S., Asseng, S., Basso, B., Ewert, F., Wallach,
D., Bairgorria, G., Winter, J.M., 2012. The agricultural model intercomparison
and improvement project (AgMIP): protocols and pilot studies. Submitted to
Agricultural and Forest Metereology.
Salasya, B.D.S., 2005. Crop production and soil nutrient management: an economic
analysis of households in western and central Kenya. Ph.D. Thesis, Wageningen
University, Wageningen, The Netherlands.
Salasya, B.D.S., Mwangi, W., Hugo, V., 1998. Assessment of the adoption of seed and
fertilizer packages and the role of credit in smallholder maize production in
western Kenya. In: Earo, C.A. (Ed.), 6th East and Southern Africa Regional Maize
conference, Addis Ababa, Ethiopia, pp. 357360.
Salasya, B.D.S., Stoorvogel, J.J., 2010. Fuzzy classication for farm household
characterization. Outlook Agric. 39, 5763.
SCA, 1990. Feeding systems for Australian livestock: Ruminants. Standing
Subcommittee on Agriculture, Commonwealth Scientic and Industrial
Research Organisation (CSIRO). CSIRO Publications, Melbourne, Australia.
Ser, C., Steinfeld, H., 1996. World livestock production systems: current status,
issues and trends. FAO Animal Production and Health Paper 127. FAO (Food and
Agriculture Organization of the United Nations), Rome, Italy.
Ser, C., van der Zijpp, A., Persley, G., Rege, E., 2008. Dynamics of livestock
production systems, drivers of change and prospects for animal genetic
resources. Animal Genet. Res. Inform. 42, 327.
Slingo, J.M., Challinor, A.J., Hoskins, B.J., Wheeler, T.R., 2005. Food crops in a
changing climate. Philos. Trans. Roy. Soc. B 360 (1463), 19831989.
Soule, M.J., Shepherd, K.D., 2000. An ecological and economic analysis of
phosphorus replenishment for Vihiga division, western Kenya. Agric. Syst. 64
(2), 8398.
Sunding, D., Zilberman, D., 2001. The agricultural innovation process: research and
technology adoption in a changing agricultural sector. In: Gardner, B., Rausser,
G. (Eds.), Handbook of Agricultural Economics, vol. 1. Elsevier, Amsterdam, pp.
207261.
Suri, T., 2011. Selection and comparative advantage in technology adoption.
Econometrica 79 (1), 159209.
Thornton, P.K., Jones, P.G., Owiyo, T., Kruska, R.L., Herrero, M., Kristjanson, P.,
Notenbaert, A., Bekele, N., Omolo, A., with contributions from Orindi, V.,
Ochieng, A., Otiende, B., Bhadwal, S., Anantram, K., Nair, S., Kumar, V., Kelkar, U.,
2006. Mapping climate vulnerability and poverty in Africa. Report to the
Department for International Development, ILRI, Nairobi, Kenya, May 2006, p.
200. <http://www.dd.gov.uk/research/mapping-climate.pdf>.
Thornton, P.K., Jones, P.G., Alagarswamy, G., Andresen, J., 2009a. Spatial variation of
crop yield response to climate change in east Africa. Glob. Environ. Change 19
(1), 5465.
Thornton, P.K., van de Steeg, J., Notenbaert, A., Herrero, M., 2009b. The impacts of
climate change on livestock and livestock systems in developing countries: a
review of what we know and what we need to know. Agric. Syst. 101, 113
127.
Thornton, P.K., Jones, P.G., Alagarswamy, G., Andresen, J., Herrero, M., 2010.
Adapting to climate change: agricultural system and household impacts in
east Africa. Agric. Syst. 103, 7382.
Tiffen, M., Mortimore, M., Gichuki, F., 1994. More People, Less Erosion:
Environmental Recovery in Kenya. Wiley & Sons, Sussex.
Waithaka, M.M., Thornton, P.K., Herrero, M., Shepherd, K.D., Stoorvogel, J.J., Salasya,
B., Ndiwa, N., Bekele, N., Croucher, M., Karanja, S., Booltink, H., 2005. System
prototyping and impact assessment for sustainable alternatives in mixed
farming systems in high-potential areas of eastern Africa. Final Program Report
to the Ecoregional Fund to Support Methodological Initiatives.
Waithaka, M.M., Thornton, P.K., Herrero, M., Shepherd, K.D., 2006. Bio-economic
evaluation of farmers perceptions of viable farms in western Kenya. Agric. Syst.
90 (13), 243271.
Winters, P., Murgai, R., Sadoulet, E., de Janvry, A., Frisvold, G., 1998. Economic and
welfare impacts of climate change on developing countries. Environ. Res. Econ.
12 (1), 124.
L. Claessens et al. / Agricultural Systems 111 (2012) 8595 95