Mapping the Past to Model the Future: Exploring the Potential Effects of Climate Change on Coffee Flavor in Veracruz, Mexico | 25, Issue 22
For her master's degree, environmental science researcher LORENA PIEDRA CASTILLO set out to model how climate change could affect coffee flavor, in the hope of providing coffee growers with valuable information to develop strategies to ensure the continued availability of high-quality coffee in Veracruz state, Mexico in the future.
I would bet that most readers of 25 know that climate change is affecting the coffee industry. Readers likely also understand that the coffee farmers bear the brunt of the negative impacts of climate change, including the potential loss of their livelihoods due to extreme weather events. For more than a decade, coffee industry publications (including this one) have covered studies on the impacts of climate change on coffee production, and those stories are getting picked up by popular press outlets and turned into articles with increasing frequency. The titles for these articles often include words like “destroy,” “danger,” “extinction,” and “disappear,” framing the future of coffee in very bleak terms. That framing creates a sense of urgency—a strategy commonly used by media companies to attract reader attention—and while it is based on facts, lately I have found myself wondering about the consequences of telling an audience of coffee drinkers that coffee is disappearing. Does that story motivate people to act, and if so, what kind of action do they take?
This is also an article about climate change and coffee, but it is different for two reasons: first, because its primary focus is the flavor of coffee, and second, because it points to potential strategies for coffee farmers to adapt their practices in light of the predicted impacts of climate change. The focus on flavor is significant because flavor attributes are specific and measurable, unlike less tangible concepts such as quality. A particular flavor—dried fruit, let’s say, or floral—can be a key driver for a buyer’s preference for a coffee, and that flavor may result from the climate where the coffee grows, the variety of tree, the post-harvest process, or a combination of variables. Some flavors, like those associated with post-harvest processing, can be controlled by the producer (or processor) and changed from year to year or even within a single harvest. The opportunities to control desirable attributes offer protection against the potential damage of climate change.
It is important for all coffee stakeholders, including coffee drinkers, to recognize that climate change is impacting the whole sector, and particularly the producers. The choices and investments we make now will determine what the future of coffee looks (and tastes) like. That can feel overwhelming, but fear of the future and grief over what is being lost are unlikely to drive our best decision-making, whether individual or collective. Adaptation requires the willingness and resources to change, and everyone along the supply chain will need to adapt as producers continue to innovate.
KIM ELENA IONESCU
Chief Sustainability and Knowledge Development Officer
The coffee sector is no stranger to climate change: over the last several decades, many studies have predicted that the suitable area for coffee farming will decrease in most producing countries.
Put simply, as the climate warms up, optimal areas for arabica coffee production are shifting to cooler regions, usually at higher elevations. In many cases, simply moving coffee production further up the mountains is not recommended. It can mean converting remaining fragments of biodiverse tropical cloud forest and potentially impacting the clean water available to downstream communities by altering the hydrological functioning of steep mountainsides. Additionally, many coffee regions are simply running out of space at high elevations as coffee farming reaches the mountaintop. While conversations about the increased consumption of quality Coffea canephora (robusta) that tolerates higher temperatures are important to any discussion of coffee and climate change, predicting changes in Coffea arabica is essential in future planning for the coffee sector.
Most studies performed to date have used global or regional spatial suitability models to predict climate change–induced changes in coffee production. A modest number of studies have addressed the potential impacts on coffee quality and, to our knowledge, only a few have looked at the effects of spatial changes on coffee flavor. Determining the exact impacts of climate change on coffee quality is difficult.
A very thorough review by Ahmed et al. (2021)[1] concludes that “coffee quality is sensitive to shifts in environmental and management variables linked to climate change and climate adaptation. Synthesis of the totality of evidence across studies points to consistent trends regarding the effects of high altitude and low light exposure on enhanced coffee sensory attributes.” They noted critical research gaps in existing scholarship. Few studies address subjects such as “the effects of carbon dioxide, temperature, and water stress on the directionality of coffee quality (increase, decrease, or non-linear) as well as the interaction of multiple environmental and management conditions.”
Climatic factors affect the plant’s metabolism, flowering, and fruit maturation, which in turn impact coffee flavor upon roasting, and subsequently a taster’s impression of quality of that coffee. Higher temperatures normally accelerate the development and size of fruits and limit the generation of secondary metabolites that are important for coffee quality. Coffee cuppers have also noted that warm (but not extreme) temperatures are usually associated with sweet profiles (e.g., chocolate, which is considered optimal for some markets), while more complex flavor profiles are associated with cooler weather conditions.
Moving beyond these general observations (and generating practical recommendations for coffee growers) has been difficult due to the large number of variables involved, the complexity of their interactions, and the unique conditions of each coffee-growing region. Scientific teams attempting to answer this question— although often excelling in disciplines such as climate change, ecology, and plant metabolism—frequently lack the coffee knowledge (including sensory skills, supply-chain understanding, and market preferences) needed to establish the connections between causal factors affecting coffee quality and especially flavor. Advances in spatial modeling, climate change scenarios, and remote sensing data layers have made it easier to trace the causal chain from climatic conditions to coffee production and to provide generalized estimates of the effects of climate change on coffee quality. However, to the best of our knowledge, there have been virtually no prior studies attempting to link climate change to coffee flavor before this one.[2]
About the Study
This study was part of a master of science thesis at Mexico’s Institute of Ecology (INECOL). INECOL is a federal research institute based in central Veracruz, the second most important coffee-growing region in Mexico, with a long tradition of producing high-quality coffee. INECOL co-founded the Center for Coffee Agroecology (CAFECOL), a nonprofit promoting more sustainable production in coffee highlands through both agroecological knowledge and a focus on the specialty coffee market. Since 2012, CAFECOL has maintained a descriptive cupping database with the help of a panel of Q Graders.[3] Almost half of the 1,200 records from 2012 to 2017 also contained detailed information about farm location, coffee variety, processing method, and flavors. These descriptive cupping records were the raw material for this study, which could not have been accomplished without CAFECOL’s detailed record-keeping.
Data were analyzed using a widely accepted method in ecology studies, known as “ecological niche modeling.” This method maps the geographic distribution of a plant or animal species based on recorded areas of observation, collection, or capture. The map of a species’ location is then overlaid with climate data. Modelers can then construct a mathematical model that predicts the presence of the species based on different climate variables. The model’s predictive power is validated by comparing its results to additional data of the species’ presence not included in the model construction. In our case, instead of modeling the presence of a real plant or animal species, we modeled the occurrence of high-quality coffee, defined as a score of 83 points or above according to the 2004 SCA cupping system, and the occurrence of a specific flavor category.[4]
Once the model to predict coffee quality and flavor was working and validated for the current climate conditions,[5] two climate change scenarios for the year 2039 were applied: an intermediate-emission scenario, in which the climate change is expected to be milder; and a high-emission scenario, in which effects are expected to be more severe.[6] This analysis was done with a special type of bivariate regression analysis (logistic regression) with only two response levels (1 or 0, presence or absence) and combined with a geographic information system (GIS) tool to map where high coffee quality and different coffee flavors were most likely to be produced for coffee-growing areas in central Veracruz, considering both historical climate and future climate change scenarios.
The CAFECOL database was curated by first removing records without GPS coordinates, entries not located in the area of study in central Veracruz, or those involving varieties and processing methods that were not considered to be representative of this region.[7] The remaining data consisted of cupping records from 222 farms. Of these, 139 recorded the “presence” of high-quality coffee (for the purpose of this study in 2019, “high quality” was defined as coffees with a score of 83 and above), while 120 farms had lower cupping scores consistent with the “absence” of high-quality coffee. The exact locations of each farm were then used to map the values of 19 different climate variables for each, using historical climate data from Mexico’s Center for Atmospheric Sciences. These climate variables included detailed temperature and rainfall data, including seasonal extremes. Through iterations of the regression analysis, a mathematical model was produced to predict the “absence” or “presence” of high-quality coffee, using climate variables deemed to be significant predictive factors. A similar approach was used for subsets of the database corresponding to the presence of specific flavor categories: caramel, chocolate, spice, floral, fruity, and nutty.
Five climate variables were determined to be the most useful in predicting the presence of a quality score above 83 points: mean annual temperature, mean temperature of warmest quarter, mean temperature of the coolest quarter, rainfall of wettest month, and rainfall of warmest quarter. When the presence or absence of specific flavor categories was modeled, we found that different climate variables best predicted specific flavor categories. For example, in the case of the presence of a floral flavor, seven variables were found to be important. In decreasing order of importance, these were maximum temperature of the warmest month, mean temperature of the wettest quarter, total annual rainfall, rainfall of the coolest quarter, mean daily temperature range, temperature of the coolest quarter, and mean annual temperature.
Figure 1.
Aptitude for the production of high-quality coffee (83 points or above) in Central Veracruz, based on historical climatic conditions in the region, divided into areas with the potential to produce low, medium, and high quality.
Initial Findings
Using these mathematical models, we began to map locations in central Veracruz that would have the potential to produce coffees with scores of 83 or above (described as “aptitude,” figure 1). We also mapped locations that would have the potential to produce coffees with each of the flavor categories that we are interested in. For example, figure 2 shows those regions with the potential to produce coffees with a floral flavor. Although high-scoring coffee may be found in a large area of central Veracruz at a certain altitude, the maps we produced showed that the conditions for increasing the probability for certain flavor categories (e.g., floral) are much more localized than high-scoring coffee in general. This means that the altitude and climatic conditions required to produce certain flavors are more specific than the conditions required to produce quality. Similarly, we also found certain flavor categories (e.g., nutty) outside the area identified as suitable for this study’s high-scoring coffees. This indicates that the presence of flavors that some tasters value (depending on their preference) is not exclusive to high-quality coffee.
The maps of suitability for obtaining high scores and specific flavors under the historical climatic conditions are a great potential aid for producers in central Veracruz and buyers sourcing coffee from the region. Producers may be able to improve the climate conditions on their own farm through agricultural practices such as shade management, or they might plant new varieties that will be more resilient to the impacts of climate change. Alternatively, they could also adopt novel processing techniques, thus increasing the aptitude of their farm in generating higher scores in general or a particular flavor category. Buyers, on the other hand, may locate regions within central Veracruz that are likely to produce the sensory attributes and flavors their consumers seek.
Plotting overlapping multivariate estimates of suitability for each of the flavor categories on a graph together with mean annual temperature and total annual rainfall (figure 3) highlights how the floral flavor category requires the coolest temperatures, whereas the chocolate category requires slightly warmer conditions year-round. This offers another explanation as to why floral profiles are only found in certain higher-altitude regions.
Figure 2.
Aptitude (0–1 or proportionally low to high) for floral coffee flavor in central Veracruz based on historical climate conditions. Note how the probability of generating this flavor increases in the Coatepec growing region (the area with the highest concentration of red pixels in the center of the study region).
What Does It All Mean?
Having identified the climate variables and the models that best predict high scores and specific flavor categories under historical climate conditions, our study then proceeded to model what could happen to coffee scores and specific flavors in the area under different climate change scenarios. These scenarios describe intermediate and severe emissions of greenhouse gases and corresponding changes in temperature and precipitation patterns, including shifts in the mapped values of the 19 climate variables used in our analyses.
Figure 3.
This perceptual map displays multidimensional averages (centroids) of what climatic conditions are required to produce certain coffee flavor characteristics. The centroids (shown as red dots) represent the average of 19 different climate variables, including annual rainfall, mean temperature, and seasonal extremes. Two specific variables are represented as lines on the plot: mean annual temperature as green lines, and total annual rainfall as orange lines. From left to right, temperatures become hotter (from 18.8°C to 19.5°C) and annual rainfall becomes lower (1987.8 mL to 1967.46 mL). You can see that some flavors, such as floral, require cooler temperatures and higher rainfall, compared to chocolate, which can be produced in hotter temperatures with lower rainfall.
Under the intermediate climate change scenario, although one region in central Veracruz (Atzalan) gains suitable area for high-scoring coffee, another region (Tezonapa) loses all suitable area and, overall, the area suitable for high-scoring coffee in central Veracruz is reduced to 57,341 hectares (42% loss of current suitable area) by 2039. Under the severe scenario (and assuming no change in farm management) two regions (Atzalan and Tezonapa) completely lose their ability to generate high-scoring coffee. The growing areas in central Veracruz suitable for high-scoring coffees are reduced to 31,074 hectares overall (a 69% loss compared to the current area). As these numbers account for new areas that become suitable for coffee as the climate changes—usually higher-elevation areas that were previously too cold to grow coffee but that may be hard to convert for reasons already mentioned— the loss of present-day farms could be even greater than these statistics suggest.
Similar trends are expected for shifts in areas with potential to produce coffee with specific flavor categories. As climate change advances, some areas will become more suitable for a certain flavor, while others will become less suitable or will be lost completely. For example, figure 4 shows which areas will gain or lose their aptitude to produce coffee with a floral flavor. What is most worrying about these scenarios is that, although there will still be some suitable areas for high-scoring coffee in central Veracruz by 2039, certain cherished flavor profiles (e.g., floral) will become extremely scarce, at least in the study area.
Figure 4.
Changes in suitability (aptitude) for floral flavor in coffee-growing areas in central Veracruz by 2039, under an intermediate (left) or a severe (right) emissions scenario: light green represents gains, orange represents losses, and dark green represents no change.
But there is hope. Although climate models take a lot of variables into account, these particular models didn’t account for certain human factors such as skills in harvesting or processing or differences in agronomy practices such as shade management, which could have also impacted the tasters’ impression of quality. All models are imbued with a level of uncertainty: in a complex system like the global climate, changes in human behavior or unexpected feedback between variables could result in different outcomes from those shown here. Even the best model predictions are still just that—predictions—so scientists must keep forecasting and fine-tuning their models because they represent our best hope for timely and effective responses to the threat of climate change.
More studies correlating coffee flavor with climate factors and projecting climate change impacts, especially regional modeling with finer-scale data, are urgently needed to gain a better understanding of what will happen in other coffee-producing regions around the world and to provide growers with vital information for planning future farm management strategies. Strategies for mitigating and adapting to climate change in the coffee sector should explicitly consider the possible consequences on coffee quality and flavor.
LORENA PIEDRA-CASTILLO was awarded a master’s degree in biological sciences at the Institute of Ecology in Veracruz, Mexico (INECOL), for this research exploring the intersection of climatic conditions, micro-regions, and distinct cup profiles in the central region of the state of Veracruz.
This article was co-authored by her supervisor, ROBERT HUNTER MANSON, PhD, tenured researcher at INECOL, MARIO ROBERTO FERNÁNDEZ-ALDUENDA, PhD, Technical Officer at the SCA, and GERARDO HERNÁNDEZ-MARTÍNEZ, PhD, Executive Director of the Center for Coffee Agroecology (CAFECOL).
References
[1] Selena Ahmed, Sarah Brinkley, Erin Smith, Ariella Sela, Mitchell Theisen, Cyrena Thibodeau, Teresa Warne, Evan Anderson, Natalie Van Dusen, Peter Giuliano, Kim Elena Ionescu, and Sean B. Cash, “Climate Change and Coffee Quality: Systematic Review on the Effects of Environmental and Management Variation on Secondary Metabolites and Sensory Attributes of Coffea arabica and Coffea canephora,” Frontiers in Plant Science 12 (2021), https://doi.org/10.3389/fpls.2021.708013.
[2] “Delimitación de micro regiones con mayor potencial para producir café de alta calidad y perfiles de taza diferenciados, bajo el reto del cambio climático, en el centro del estado de Veracruz” (master’s thesis, INECOL, 2019), https://bit.ly/piedra-castillo-thesis.
[3] Database generated using the 2004 SCA cupping protocol supplemented with a descriptive check-all-that-apply list based on the early 2000s version of the Coffee Taster’s Flavor Wheel.
[4] The flavor categories used were floral, vegetative, fruity, caramelized, chocolate, nutty, spices, pyrolytic, resinous, fermented, phenolic, and earthy.
[5] Historical conditions (1902–2011) were modeled using data from the Autonomous National University of Mexico’s Unit for Information on Atmospheric and Environmental Sciences (UNIATMOS), https://uniatmos.atmosfera.unam.mx.
[6] Near-future (2039) climate change scenarios used Representative Concentration Pathways (RCPs) 4.5 and 8.5 to predict greenhouse gas emissions and expected impacts on global climate, under assumptions of moderate or minimal changes in human behavior to reduce these emissions, respectively. For more information, see https://coastadapt.com.au/sites/default/files/infographics/15-117-NCCARFINFOGRAPHICS-01-UPLOADED-WEB%2827Feb%29.pdf.
[7] To minimize excess variation in our dataset that could make interpreting our results more difficult, we eliminated recently introduced coffee varieties (e.g., Geisha or recently developed hybrids) and novel processing techniques such as “natural” and “honey” with a limited geographic distribution in our study region.
We hope you are as excited as we are about the release of 25, Issue 22. This issue of 25 is made possible with the contributions of specialty coffee businesses who support the activities of the Specialty Coffee Association through its underwriting and sponsorship programs. Learn more about our underwriters here.