1 Introduction

During the last decades climate has been changing and this change has been affecting global crop production (Lobell et al. 2011). Global temperatures will likely continue to increase throughout the twenty-first century (IPCC 2007), primarily due to increases in the concentration of greenhouse gases such as carbon dioxide, which are the direct effects of human activities, mainly land-use change and burning fossil fuels (Darwin 2001; Githeko et al. 2000). This has already resulted in sea-level rise, an increase of severe weather events and radical changes in weather patterns, all of which are expected to get noticeably worse over the coming decades (Schmidhuber and Tubiello 2007). Climate change has also caused reduced yields of numerous agricultural crops around the world due to the expanding range of crop diseases (Epstein et al. 1998) and temperature and water stresses. These severely jeopardize food security on global and local scales (Jarvis et al. 2010; Jones and Thornton 2003; Lobell et al. 2008; Ludi 2009; Parry 2007; Schmidhuber and Tubiello 2007).

Agriculture is the basis for food security and economic sustainability of people’s livelihoods (FAO 2009). Food security in this context means that all people, at all times, have physical, social, and economic access to sufficient, safe, and nutritious food that meets their dietary needs and food preferences for an active and healthy life (FAO 2003). According to the Food and Agricultural Organization of the United Nations (FAO) and the International Fund for Agricultural Development (IFAD) in 2009 over one billion people were undernourished or unable to meet their basic food needs (Godfrey 2009; Ludi 2009). This global food problem is expected to worsen progressively in the coming years as numerous factors continue to hamper the production and access to food, including environmental degradation, and socio-political and economic instability throughout the world (Ludi 2009). The changing global climate is yet another obstacle, impeding the production and access to food in many of the most impoverished regions of the world (Magrin and Gay Garcia 2007).

The majority of the Colombian population lives in the Andean region, one of the most food insecure regions in the world (Lobell et al. 2008). This has made the country vulnerable to the effects of climate change in terms of food security. In general, small farmers in Colombia have long been plagued by numerous factors that have compromised their livelihoods. For example, according to the Instituto Colombiano de la Reforma Agraria (INCORA), in the 1970’s one-third of the country’s population was living in poverty in rural areas (Rochin 1977). Additionally, at least three million Colombians have been displaced, mainly in rural areas, as a result of internal armed conflicts (Amnesty International 2009). A recent evaluation of the impacts of climate change on the most vulnerable sectors of Colombian agriculture shows that if no adaptation measures are taken, 80 % of crops would be impacted in more than 60 % of their current areas of cultivation (Ramirez-Villegas et al. 2012).

Bogotá, the capital city of Colombia is located in the Andes at 2,600 mean average sea level (masl). It is home to more than 8 million people and hosts the second largest market place for food items (Corabastos) in South America (Lee 2002). Approximately 47 % of food items offered in the market place come from Cundinamarca, a department adjoining Bogotá (Montoya and Mondragón 2010; Rodriguez 2004). Another 33 % of the food passing through the Corabastos comes from the departments of Boyacá, Tolima and Meta (Fig. 1). Approximately 90 % of the farmers in these departments are smallholders, producing roughly 70 % of Bogota’s food, owning or cultivating less than 20 ha of land or less (Montoya and Mondragón 2010).

Fig. 1
figure 1

Bogotá and its surrounding departments where food for the capital is grown

Full size image

In light of the numerous challenges faced by the farmers as well as of the implications these challenges have for Bogotá's food security, a new supply chain dubbed mercado de campesinos (farmers’ market) was initiated. Supported by several local and international NGOs, this new initiative aims at removing the middle men from the supply chain and at providing farmers with the necessary transport, organization, marketing facilities and support that allows them to sell directly to their consumers. As a result, farmers are able to command a higher price for their products, while consumers profit by obtaining their food items at a lower price (Montoya and Mondragón 2010). Regardless of the numerous advantages of the mercado de campesino, it is unable to address completely the diverse challenges that farmers have been facing and are likely to face in the future.

It will be crucial to explore options for adaptation to climate change to improve food security in developing countries. There is often a lack of information, financial resources, and governance to guide smallholder farmer to improve their adaptive capacity.

The objective of this paper is to address the vulnerability of smallholder farmers to climate change in the area surrounding Bogotá. We quantified farmers’ exposure by analyzing the direct impact of climate change through crop modeling, and then we conducted participatory workshops and questionnaires to determine their sensitivity and adaptive capacity. We also examined farm-specific vulnerability and distinguished the differences in vulnerability between the farmers who use the mercado de campesinos and those who deal with intermediaries.

First we describe the context of food security and climate change in Bogotá and explain the methodologies and the definition of vulnerability to climate change. We then present the vulnerability analysis using crop and socio-economic modeling. We conclude with an assessment of implications for Bogotá’s food security and for the ecological landscape of its surrounding area.

2 Methods

A block diagram of the methodology is shown in Fig. 2. We used a mechanistic crop model to determine the direct impact of climate change on the supply chain in terms of its impact on the climatic suitability for crop production. We complemented this analysis by assessing the indirect sensitivity and adaptive capacity to climate anomalies using the sustainable livelihoods framework (Farrington et al. 1999; Scoones 1998). Based on this analysis, we developed indicators to compare the vulnerability of the distinct supply chains. This approach is appropriate when compare vulnerable people, communities, and regions within narrowly-defined local systems (Hinkel 2011).

Fig. 2
figure 2

The methodologies used in the assessment

Full size image

We followed the United Nations Intergovernmental Panel on Climate Change (IPCC) definition of vulnerability (IPCC 2001) to assess the exposure, the sensitivity, and the adaptive capacity of Bogotá’s food supply chains. Site-specific vulnerability at the farm level is given by:

$$ Vulnerability=Exposure+Sensitivity-Adaptive\,Capacity $$

We calculated exposure for each farm as the direct impacts of climate change on the main crop and used the defined set of indicators for sensitivity and adaptive capacity derived from the survey.

2.1 Case study area

We selected areas (Fig. 1) in the departments of Boyacá and Cundinamarca, which surround Bogotá and supply it with most of its food. The selected areas are characterized by diverse landscapes and climatic conditions. The predominant highlands (páramos), for example, are 2,000–3,000 (masl), have cool climates. Here farmers grow a wide array of staple crops such as potato (Solanum tuberosum), onion (Allium fistulosum), and corn (Zea mays). The páramos are also an important ecosystem for water supply to Bogotá. In contrast, in areas below 1,000 masl farmers grow hot climate crops such as banana (Musa paradisiaca), sugarcane (Saccharum officinarum), oranges (Citrus sinensis), and other tree crops such as mango (Mangifera indica) and papaya (Carica papaya). The warm-temperate transition zones between these two areas provide the environmental niches for crops like common bean (Phaseolus vulgaris), green pea (Pisum sativum), blackberry (Rubus glaucus), tomato (Lycopersicon esculentum), and tree tomato (Cyphomandra betacea).

2.2 Climate data

As current (baseline) climate we used historical climate data from the WorldClim database (Hijmans et al. 2005). We generated regional future climates using the global climate model (GCM) outputs from the IPCC, downloaded from the Earth System Grid (ESG) data portal. We applied a statistical downscaling to 19 GCMs from the IPCC Fourth Assessment Report (IPCC 2007) for the Special Report on Emissions Scenario (SRES) A2 (regionally-oriented economic development) and for two different 30-year periods (i.e., 2020–2049 representing the 2030 time-slice and 2040–2069 representing the 2050 time-slice). The downscaling that we used is based on the sum of interpolated anomalies to a 30 arc-seconds resolution of the monthly climate surfaces from WorldClim (Hijmans et al. 2005). This method produced a smoothed, interpolated surface of changes in climates forecast by the particular GCM and applies it to the baseline climate (Ramirez-Villegas and Jarvis 2010).

2.3 Impact on crops as “Exposure”

We input both the current and generated, down-scaled climates to ECOCROP (Hijmans et al. 2001), a mechanistic model based on the FAO crop environmental requirements databases (FAO 2000) that can be used to predict crop climate-suitability where no prior knowledge or data is available. ECOCROP uses the mean monthly temperatures, and accumulated growing season rainfall during an iteratively-defined growth period. It estimates a suitability index (0–100) based on temperature and rainfall data separately, and an overall index by calculating the product of the temperature and rainfall indices.

We calibrated the crop parameters from the FAO database by adjusting the agro-climatic values following the method of Ramirez-Villegas et al. (2011), so that they accurately represent crop evidence data collected from the study area. For each crop we collected 60–100 geo-referenced, random evidence samples to recalculate the environmental factors. We did this by dividing them into discrete constant-value ranges, and predicted current crop-climate suitability, based on the current crop distribution. The resulting suitability maps for current conditions were confirmed using the expert opinion of a local agricultural expert from the Boyacá department.

We applied ECOCROP to data of each of the 19 GCMs and estimated the uncertainty of future crop climate-suitability predictions by computing two measures of confidence: the coefficient of variation (CV) of precipitation and temperature on the raw GCM data, and the percentage of ECOCROP estimates for the separate GCM data that predicted changes in the same direction as the mean of all 19 GCMs for each location. Suitability in this paper refers to climatic suitability; in other words the percentage values shown in the graphs and tables (Table 1 and Fig. 3) refer to the probability that the crop grows well. In this case, it points to the presence of the combination of suitable climatic variables. ECOCROP spatially extrapolates climatically suitable areas, which are different to areas where the crop is actually grown, since some of the suitable areas may be occupied by settlements, protected areas, or by different cropping systems.

Table 1 Change in climate-suitability change for 19 crops by the years 2030 and 2050
Full size table
Fig. 3
figure 3

Suitability change for selected crops by 2050. Areas without color are unsuitable for the crop in question. ‘Areas not available’ are zones that contain water bodies, constructed zones or bare land. Forests and protected areas are labeled as Areas unfavorable for cultivation

Full size image

Therefore in the final step, (Table 2) we intersected the results of crop climate-suitability to take account of land suitability considering restrictions to land use (water surface, populated areas, forest cover, and cultivated land), landscape protection (including ecological buffer strips), and proximity to access roads (to take account of increased transport costs from remote areas). We assumed that any land not subject to one of these restrictions is available for crop cultivation.

Table 2 Changing climate-suitability versus land availability. Numbers in grey are changes in area and numbers in red are crops losing suitability in highly favourable areas
Full size table

2.4 Vulnerability analysis

To assess the impacts of climate change on cropping systems within the context of rural livelihoods, we estimated producers’ available resources in the form of livelihood capital stocks (Farrington et al. 1999). Based on expert knowledge, literature review, and our previous research experience, we differentiated between physical, natural, human, social, and financial capital and designed climate-related indicators for both sensitivity and adaptive capacity that could be measured on an ordinal scale. We held participatory workshops with farmers’ groups to learn about their first-hand experience with historical climatic events. We used the participatory rural appraisal (PRA) method, taking account of the respondents’ gender (Bhandari 2003). Facilitators guided the workshops to draw out the required information using diagramming and visual sharing, where charts are used to communicate ideas with the farmers (Chambers 1994).

Adaptive capacity (in relation to climate change impacts), is the ability of a system to adjust to climate change (including climate variability and extremes), to moderate potential damage, to take advantage of opportunities, or to cope with the consequences. Sensitivity, on the other hand, is the degree to which a system is affected, either adversely or beneficially, by climate variability or change. The effect may be direct (e.g.: a change in crop yield in response to a change in the mean, range, or variability of temperature), or indirect (e.g.: damage caused by an increase in the frequency of coastal flooding due to sea-level rise).

We collected socio-economic information throughout 120 semi-structured interviews conducted in six different regions in the departments of Cundinamarca and Boyacá. We randomly selected farmers from the areas where workshops were conducted and in adjoining communities and asked them questions that shed light on their level of resilience to climate change. The core purpose of the interviews was to identify key vulnerabilities of supply chains at the farm level, which is where risks to livelihoods caused by climate change will need to be addressed.

The research team analyzed the data using the statistics software for social sciences SPSS V.17.0. In particular, using visual analyses, correlation analysis, multiple regression analysis, and t-Tests to identify relationships and differences between households and forms of capital. We then conducted a cluster analysis to identify principal clusters of individuals sharing common characteristics and the principal classifiers. We prepared final output diagrams (Figs. 5 and 6) for each type of capital that represents high sensitivity and low adaptive capacity. In these diagrams 3 represents high adaptive capacity and 1 means low adaptive capacity. In contrast, 3 means low sensitivity and 1 high sensitivity. Two 3s for adaptive capacity and sensitivity therefore represents low vulnerability, while two 1s represents high vulnerability to climate-related, socio-economic indicators. We chose modal values, which gives clear indications of the state of each type of capital, even though it masks detailed information. Thus, for each supply chain and municipality the modal values indicated vulnerability. This was used in stakeholder workshops to identify adaptation strategies.

We calculated exposure for each farm as the direct impact on the main crop from Ecocrop (0–100 scale divided by 10) and used the defined set of indicators for sensitivity and adaptive capacity as the average of all capitals determined in the survey. To illustrate site-specific vulnerability and its change between 2030 and 2050 we mapped exposure, sensitivity, and adaptive capacity of each surveyed farm together (Figs. 7 and 8).

3 Results

3.1 Exposure to climate change

According to the average output of the 19 GCMs for the future climate scenarios (2030, 2050), annual rainfall will increase in all selected areas by an average of 100 mm by 2050. The biggest increase will be 122 mm at San Antonio de Tequendama. Mean annual temperature will increase by 2.1 °C over the study area. At Duitama, in Boyacá, mean annual maximum temperature is predicted to increase by 2.9 °C by 2050, while by 2030, mean annual maximum temperature will increase by 2.5 °C in La Vega and Duitama. The maximum number of cumulative dry months will stay constant in most of the region, except for 1 more dry month in La Vega and Duitama and 1 less dry month in Cáqueza. In general, the wettest quarter of the year will get wetter and the warmest quarter will get drier in the whole region. Half the rainfall increase will be in the wettest quarter, followed by the coldest quarter.

Table 1 shows the climate suitability for 19 crops. Ecocrop indicates that currently all crops have suitability values 60–100, which means that growing conditions are very good or excellent. By 2030, however, suitability will be reduced for most crops, although growing conditions will remain very good or excellent for most of them, except for mango (Mangifera indica), orange (Citrus sinensis), papaya (Carica papaya), and corn (Zea mays), all of which will lose considerable suitability. By 2050, most crops lose climatic suitability. Guava (Psidium guajava), mango, orange, and papaya will be the worst affected, falling to 34–47. Suitability declines for all crops and sites by 13 %. Only papaya makes a marginal gain in suitability by 2050.

In Table 2, climate suitability and land favorability in units of 1,000 ha are combined. Rice (Oryza sativa) and tomato (Solanum lycopersicum) are currently highly suited climatically on 1,127,000 and 1,271,000 ha of land and are the crops for which the area of land with highly favourable climate increases by 2050. In contrast, blackberry (Rubus glaucus) and tree tomato (Cyphomandra betacea) are currently much more restricted with only 268,000 and 324,000 ha respectively of land with highly favourable climate. Most other crops will lose area of highly favourable land, with orange and mango losing the most (682,000 and 231,000 ha respectively by 2030). By 2050, the area of highly favourable land will be less for most crops. Rice, banana (Musa paradisiaca), onion (Allium fistulosum), beans (Phaseolus vulgaris), tomato, and tree tomato are the exceptions.

Combining Tables 1 and 2, the maps in Fig. 3 show the changes in climatic suitability for some selected crops by 2050 accounting for restrictions of land availability. Cassava (Manihot esculenta) and plantain (Musa balbisiana) lose most of their suitable areas in the lowlands, while gaining suitability in some narrow zones in areas of medium altitude, in places competing with other crops such as beans or corn. Beans shift to higher altitudes and corn loses lots of production areas in the lower areas of the Meta department. Suitable areas for potato (Solanum tuberosum) shift to higher altitudes, but much of the highly suitable areas are páramos, which are classified as unfavorable for cultivation because they are important sources of water for Bogotá. Many crops shift to higher altitude, which will cause conflicts in land use and water, and place more stress on protected and the remaining forest areas.

3.2 Measures of uncertainty

The coefficient of variation for the predicted values of the climate variables ranges between 0 % and 20 % in most of the study area, which suggests that the predictions are acceptable. There are exceptions, however, at higher elevations where there is a low density of rainfall stations (Fig. 4). The measure of agreement of the GCMs predicting changes in the same direction as the average of all models is generally high (Fig. 4).

Fig. 4
figure 4

Coefficient of climate variation between models for 2050 and the uncertainty of predicted climate-suitability for all crops

Full size image

3.3 Farmers sensitivity and adaptive capacity

Regarding sensitivity to climate change, we separated farmers into two groups according to their marketing channels. Farmers who sell their produce to intermediaries (Fig. 5) showed a low sensitivity (score 3) to financial, physical and natural capital, while human capital was highly sensitive (score 1). Social capital had medium sensitivity (score 2). On the other hand, adaptive capacity was high (score 3) for natural and human capital, but low (score 1) for social and financial capitals.

Fig. 5
figure 5

Spider diagram of sensitivity and adaptive capacity for producer groups selling to Intermediaries

Full size image

Farmers who sell their produce directly to consumers (Fig. 6) differed substantially, with low sensitivity and high adaptive capacity in the financial, natural and human capitals coupled with a low sensitivity and a medium adaptive capacity for physical capital. These farmers had medium sensitivity and high adaptive capacity in social capital. These trends indicate that farmers who sell directly are less vulnerable than farmers who sell to intermediaries.

Fig. 6
figure 6

Spider diagram of sensitivity and adaptive capacity for producer groups who sells direct to the consumer

Full size image

In the case of farmers selling directly to the consumer, high adaptive capacity in social capital (Fig. 6) can be explained by their active participation in the organized farmers’ market in Bogotá. They also receive technical assistance and training, which explains their low sensitivity in human capital. The farmers selling to intermediaries (Fig. 5) typically receive no training, which accounts for their high sensitivity in human capital. Furthermore, their low adaptive capacity to financial capital is because their most frequent answer to credit related questions was “does not apply”, which implies a lack of access to credit.

We mapped site-specific vulnerability for 2030 in 2050 (Figs. 7 and 8) to show which groups of farmers are most vulnerable to climate change, with exposure plotted on the abscissa as the climate suitability change of a major crop on a particular farm (−4 is high negative and 7 is high positive change); while the ordinate shows sensitivity (0 is low, 3 is high). The size of the circles indicates adaptive capacity (big size is low adaptive capacity, small size high). The background color shows the combination of exposure and sensitivity in traffic light colors. The red section represents high exposure and high sensitivity while the green section represents low exposure and low sensitivity. The only changing variable between Figs. 7 and 8 is exposure, which represents the crop climate-suitability as modeled by EcoCrop. Sensitivity and adaptive capacity result from the analysis of the collected socio-economic information and are not projected into the future. The big sample points in the upper left, red-colored corner are those that are highly vulnerable to predicted climate change. In contrast, a small bubble in the green section represents farmers that are less vulnerable to the impacts of climate change.

Fig. 7
figure 7

Site-specific vulnerability by the year 2030 derived from Exposure, Sensitivity and Adaptive Capacity

Full size image
Fig. 8
figure 8

Site-specific vulnerability by the year 2050 derived from Exposure, Sensitivity and Adaptive Capacity

Full size image

The graphs show clearly that for 2030, Bogotá’s farmers are still only moderately vulnerable (Fig. 7), but by 2050 range of vulnerability is more widespread and many more of them move towards the left corner into higher vulnerability (Fig. 8).

3.4 Farmers’ perceptions of climate change

During participatory workshops, farmers reported that deforestation is a noticeable problem within their communities but it is not something that they can control. They added that nearly all forest was transformed to agriculture last century, mainly through slash-and-burn. When asked whether the general climate (rain, temperature, and wind) have gotten better or worse over the past 20 years, the response was very different between sites. For example, farmers in Choconta and Caqueza perceive that present precipitation is quite more favourable for agricultural production than 20 years ago. In contrast, farmers in Duitama thought it was less.

3.5 Implication for Bogota’s food security and farmers’ livelihood

Some major cash crops will face substantially reduced climatic suitability, while others will need to move their geographic location to maintain suitability, which has serious implications for Bogotá’s food security and farmers’ livelihoods. Moreover, increasing temperature and changing precipitation patterns may alter flowering dates and affect other biotic factors such as the spatial and temporal range of pests, diseases, and weeds. This could increase farmers’ production costs as they are forced to purchase products such as chemicals to deal with these challenges (Jarvis et al. 2011).

Changing precipitation patterns will change soil water availability. Heavy rains will lead to more flooding, soil erosion, and possibly massive crop loss. These stresses may reduce yields and quality, and thereby reduce farmers’ incomes. If the principle of supply and demand holds, a reduced food supply in a future where demand for food is expected to increase exponentially will cause food prices to rise to levels that would not allow Bogotá’s citizens to meet their basic food needs.

The geographical shift in climate suitability may also force farmers to shift their crops to higher elevations, including remaining forests and the páramos. The effects of this will be an exacerbation of the current level of deforestation, which in turn will exacerbate the frequency and intensity of flooding, soil erosion, and crop losses due to increased precipitation. Disturbance of the natural vegetation in the páramos, which act as natural basins and filtration system for Bogotá’s water supply, would compromise its quality and quantity. This situation looks even bleaker when we consider the increased variation of annual precipitation that is forecasted.

Farmers who only sell to the traditional intermediaries are more vulnerable than those who trade in the mercado de campesino system, mainly because the latter are given technical assistance to overcome a number of their challenges. Unfortunately, the mercado is only able to support a limited number of farmers and is currently dependent on external funding, so that it is not sustainable in its current form. Furthermore, it is difficult to overcome the bureaucratic hurdles to license a site to host the mercado campesino, so there may be a role for the state. Nevertheless, if better organizational and marketing practices are not implemented and sustainably internalized at the farm level, the farmers who supply the mercado may become more vulnerable.

4 Conclusions

Small holder farmers represent the weakest part of the food market chain. They need to be informed comprehensively what the expected impacts of climate change are, where and when adaptation measures will be required and what the consequence for their livelihoods may be. On the other hand, a strong commitment of the national government is needed to implement effective adaptation measures at a regional scale. This paper shows a detailed and quantitative vulnerability analysis for a few departments in Colombia. Our findings show that increasing temperature and precipitation will affect the distribution of crops such as mango, papaya, corn, and plantain, which will become less suitable by 2050 in current production areas. The groups of farmers that are most sensitive to climate change, and whose adaptive capacity to cope with its consequences is low, are those who sell their produce directly to intermediaries. On the other hand, farmers who sell their produce at the mercado de campesino are less vulnerable. Our socio-economic analysis shows that changes in social capital made farmers more or less vulnerable. Further work is required to focus on how these farmers can cope with potential threats of ongoing climate change and climate variability.

We conclude that the surrounding area of Bogotá will likely face severe changes in climate suitability for selected number of crops, farmers will need options for crop diversification, and the food for Bogota is likely to come from different, and probably more remote, rural areas than today. Conservation areas, such as the páramos will likely suffer increased invasion with the shifting of the agriculture frontier. Finally as a consequence of farming systems migrating to higher altitudes, the ecosystems (páramos) that provide essential environmental services to downstream populations will become threatened.

There will be profound implications for Bogotá’s food security, for the ecological landscape around the capital, and for farmers’ livelihoods unless proper adaptation measures are implemented. Here, we recommend a more integrated research on the study area, including interacting components of agriculture, ecosystem services, and conservation of biodiversity which could provide sustainable adaptation measures for the region.