Site selection: climate, altitude and terrain

Temperature limits in detail#

Arabica. The 18–21 °C optimum dates back to a 1959 source cited in the 2006 review.¹ Above 23 °C, fruit development and ripening are accelerated, often with a loss of quality.¹ Continuous exposure to temperatures as high as 30 °C could depress growth.¹ Where the mean annual temperature is below 17–18 °C, growth is largely depressed.¹

Robusta. The optimum of 22–26 °C (or 24–30 °C) in the review rests on sources from 1998 and 1999 cited in the review.¹ The 2020 study instead fitted a hierarchical Bayesian model to ten years of yield records from 798 farms in South-East Asia.⁴ It put the optimum below 20.5 °C, or a mean minimum and maximum of ≤16.2 and ≤24.1 °C: 1.5–9 °C lower than earlier estimates.⁴ Above those values, yields fell by about 14% for each 1 °C of warming in the growing season.⁴ At the cold end, robusta is much less adaptable than arabica.¹

Altitude and quality. A 2007 review explains why altitude matters for the cup. At higher elevations ripening slows down, which allows more time for complete bean filling.² The beans are described as denser and more intense in flavour than those from lower altitudes.^2,3 A 2017 textbook chapter links the slower ripening to lower air temperatures and notes that shade can also slow it.³

Vapour pressure deficit and heat#

Temperature and rainfall do not capture how strongly the air dries the plant. Vapour pressure deficit (VPD) is calculated from temperature and humidity.^5,6 A 2022 global analysis compared it with temperature, precipitation and soil moisture as predictors of arabica productivity.⁵

VPD during fruit development proved a key indicator: yield declined rapidly above 0.82 kPa.^5,6 The risk of crossing this threshold rises sharply in most of the countries assessed if global warming exceeds 2 °C.⁵ At 2.9 °C of warming, countries making up 90% of global supply are more likely than not to exceed it.⁵ Only the abstract of the study could be opened for this entry; any further thresholds in its full text are left out.

Warm nights and ripening#

In one Tanzanian study, night temperature mattered more than day temperature for when arabica ripened. The study combined high-resolution microclimate data with generalised additive models.⁷ Warming nights, rather than day-time or maximum temperatures, had the strongest effect on the timing of ripening and harvest.⁷

The authors built a warm night index (WNI) from the mean night temperature. It predicted the start of the harvest better than growing degree days.⁷ A WNI of 15 °C went with the latest ripening. Past an inflection point of about 16 °C, ripening came earlier by about 17 ± 1.95 days for every 1 °C rise in the WNI.⁷ Faster ripening links back to the quality loss the 2006 review describes above 23 °C.¹

Altitude, latitude and the lapse rate#

The Colombian study used 34 weather stations in the coffee zone.⁸ Altitude was the main control of air temperature: −0.7 °C per 100 m for the annual mean maximum and −0.5 °C per 100 m for the annual mean minimum.⁸ The authors note that an earlier figure of −0.0061 °C per metre for mean temperature in the Colombian Andes is an average of the two.⁸

Latitude adds a second, smaller effect. Per degree of latitude, the maximum fell by 0.065 °C and the minimum by 0.216 °C (the paper’s Table 2 gives 0.064 and 0.214 °C).⁸ The latitude effect appeared only beyond about 7° of latitude.⁸ The authors call the analysis preliminary because it rests on 34 stations.⁸

As a worked example with the Colombian rates, a site 500 m higher would have a mean maximum about 3.5 °C lower and a mean minimum about 2.5 °C lower (our calculation).⁸ Rates elsewhere depend on local conditions and are not established by the sources opened here.

Frost and wind#

Frost easily damages and can kill the tree.¹ In a trial at Londrina, Paraná, Brazil, frosts in 1978 and 1979 hit young plants.¹³ Cold air collecting near the ground killed the phloem 10–20 cm above the soil, a lesion known in Portuguese as canela de geada.¹³ Plants whose stems had been banked with soil before the frosts showed no such lesions.¹³ On one frost night the weather-shelter minimum was 0.0 °C while the grass minimum was −2.8 °C.¹³

Wind stress reduces leaf area and internode length.¹ According to a 2017 textbook chapter, strong wind over several days also makes flowers and fruit drop. Hot wind in the dry season stops growth and leads to poor fruit set and empty fruits.³ The chapter lists windbreaks, adapted cultivars, pruning, irrigation and shade as responses.³ No source opened here gives a wind-speed limit.

Slope and mechanisation#

A 2016 Brazilian trial in Santo Antônio do Amparo, Minas Gerais, tested a self-propelled harvester on stands of 10%, 15%, 20%, 25% and 30% slope.¹² On 20% slope, harvesting took 21.6% longer than on flatter ground.¹² Downtime made up 10.66–29.18% of the total harvest because of more manoeuvres.¹² On steep land, some mechanised operations, harvesting among them, cannot be carried out.¹² Hand harvesting methods are covered in Harvesting and cherry sorting.

Agro-ecological zones#

A 2015 study grouped the places where arabica grows into five agro-ecological zones, based on climate data.⁹ Under the climate of 1950–2000, arabica grows on 7.2% of the land in the belt from 30°N to 33°S.⁹

Source: Bunn et al. 2015, scenario RCP 6.0 for the 2050s.⁹ According to a 2017 textbook chapter, specialty arabica comes mainly from the Constant zone.³

Climate change and suitable land#

Several global modelling studies agree on the direction: coffee land is projected to shrink and move.

• Arabica zones, 2015. With 19 climate models and the intermediate RCP 6.0 scenario, the share of suitable land in the 30°N–33°S belt halves to 3.6% by the 2050s.⁹ Zones suited to arabica move upwards by about 500 m.⁹ The Hot-dry zone is hit hardest: only 16% of its current land stays suitable, and 78% becomes unsuitable.⁹
• Both species, 2015. A second study projects that climate change will reduce the global area suitable for coffee by about 50% across emission scenarios.¹⁰ Higher temperatures may reduce arabica yields, while robusta could suffer from more variable temperatures within seasons.¹⁰
• Coffee, cashew and avocado, 2022. With 14 climate models and three emission scenarios for 2050, coffee proved the most vulnerable of the three crops.¹¹ Climate factors limited suitable land more than soil factors did. Long dry seasons, mean temperatures, low minimum temperatures and annual rainfall mattered most; on the soil side, low pH, poor texture and steep slopes.¹¹ High-latitude and high-altitude areas may gain from rising minimum temperatures.¹¹

These are projections from models, not forecasts by Coffearium. The VPD and warm-night findings above describe mechanisms through which such shifts would reach yields and harvest timing.^5,7

Reference values#

Open questions#

• A robusta optimum for other regions. The 2020 yield study covers South-East Asia. Whether its lower optimum holds in Africa or Brazil is not settled by the sources opened here.⁴
• Further thresholds. Only the abstract of the 2022 VPD study could be opened, so any further thresholds in its full text are left out, and no VPD threshold for robusta was found.
• Numeric climate envelopes per zone. A 2017 textbook chapter gives maximum and minimum temperatures for each arabica zone and for four robusta zones, but the primary paper opened here gives these only qualitatively. The numbers are left out until a second source confirms them.
• Frost and wind limits. No source opened here gives a lethal leaf temperature for frost or a damaging wind speed.

Key studies#