The frequency of extreme weather events—or climatic shocks—is increasing in all land areas. Figure 2 depicts the change to four extreme climate events that are critical for agriculture and human health. The number of extremely hot days is increasing globally, with temperatures above 40 °C projected to become increasingly common in much of the lower latitudes by 2100 in a high climate change scenario (Fig. 2a). On the other end, extreme cold events are decreasing in much of the world. The number of frost days, in which low temperatures drop below freezing, will decrease in most of the extratropics. This can allow for changes and adaptations in agriculture that could potentially increase production in these regions, although a lack of winter frosts could facilitate year-round pressure from microorganisms and insects that harm people, so the net effect is ambiguous [22].

Projected climate changes 2081–2100 under the RCP8.5 scenario, CMIP6 multimodel means. a Change in the annual number of days above 40 °C. b Change in the annual number of frost days (below 0 °C). c Percent change in precipitation on the wettest day of the year. d Change in the number of days in the longest string of consecutive dry days per year. Variables are defined in Table 1

Extreme precipitation events are increasing in almost all land areas. The wettest day of the year is projected to increase almost everywhere (Fig. 2c), with high percent changes in much of Asia and Africa. Hydrological projections for changes to river flows vary from region to region [23], and are further influenced due to the high level of human intervention in hydrological systems. The probability of drought is increasing in the Mediterranean, much of South America, and Southern Africa. As one indicator, the number of consecutive dry days is projected to increase dramatically in each of those regions (Fig. 2d). This coincides with an increased risk of extreme precipitation in several of those locations, indicating increased variability in a future climate with more droughts as well as more floods.

Crop model simulations for the long-term effects of climate change on agriculture present a varied picture (Fig. 3) of potential increases and decreases in calorie production globally. Assuming that people continue to grow the same crops in the same locations as was done historically (2010 cropland estimates), and assuming that the percent change in modeled achievable yield is a sufficient proxy for the change in crop production due to climate change, Fig. 3 presents the projected change in calorie production due to climate change. This can also be interpreted as the percent change in attainable yield, with each crop weighted according to historical caloric production.

Projected change between the present period (2011–2040) and the future period (2041–2070) in calorie production from agriculture, assuming actual agricultural production changes proportionally to attainable production and there is no change to the crops planted. Data are expressed as the natural log of kilocalories per gridbox, multimodel mean of 5 climate models. Individual model results are available in Supplementary Information Fig. 1

Total calorie production would increase under this no-adaptation scenario in many high-latitude regions, including the eastern United States, almost all of Canada, Northern Europe and Russia, Eastern China, much of Southeast Asia, and parts of Argentina, Uruguay, and Brazil. Other areas of the world have mixed signals or projected decreases in calorie production, including much of Africa, the remaining parts of Central and South America, South Asia, the Middle East, and Australia. For example, most of India and Myanmar would be projected to see decreased calorie production, as well as much of the Sahel and South Africa.

Since these are also local areas that would experience increases in shocks, they represent locations where the pressure from climatic factors will likely increase significantly. However, this does not mean that it will lead to famines since that would require holds in addition to the pressure.

In this scenario, we explore two ways in which climate change adaptation could increase agricultural production in a world with climate change. First, improvements to farming techniques, such as sustainable intensification, infrastructure investments, and technological improvements can increase crop yields in a given area. Known widely as the yield gap, one opportunity for adaptation is to increase local production to be closer to modeled attainable yields. Figure 4 illustrates the current yield gap, with much of Africa, Asia, and Latin America seeing yields that are less than half of what is believed to be the attainable yields in their region.

Opportunities for adaptation. The estimated yield gap of current agriculture: current yields as a percentage of attainable yield

A second major opportunity for adaptation is in the shifting of which crops and which cultivars are planted in which locations. Figure 5 shows that for almost all agricultural regions, there is at least one crop that is projected to increase yields under the pessimistic climate change scenario of RCP 8.5, even without including irrigation.

Opportunities for adaptation: the projected change in attainable yield between the current period (2011–2040) and future period (2041–2070) of the crop that will have the highest projected improvement to yield in each location

This analysis does not account for the range of outcomes given interannual variability; therefore, catastrophic yields from extreme weather events or compound events are not included in the agricultural projections. We also do not account for the impacts of climate change on irrigation and the viability of expanded irrigation as an adaptation strategy. The analysis also fails to include the effects of climate change on livestock production, pest and pathogen outbreaks, or the effect of heatwaves on agricultural labor [24,25,26].

Another factor not included in this analysis is the effect of climate change on disease rates, which affect the likelihood of famine in food scarce conditions [27]. Climate change could increase the incidence or geographic spread of tuberculosis in dry conditions or of waterborne diseases such as typhoid and cholera during floods [28, 29]. The uncertainties inherent in crop models are also a major limiting factor for this analysis; even when a suitable choice of a crop simulation model is selected, data limitations hamper some of the models’ effective role in projections [30].

We do not account for feedback loops in this analysis, in which extreme weather events or low agricultural production might influence conflict dynamics and the decision to maintain a “hold” on a region that seems otherwise vulnerable. Finally, during conflict situations, local agricultural production might not be possible, rendering this analysis irrelevant. For example, in the 2023 conflict in the Gaza Strip, 44% of pre-war consumption of agricultural commodities came from the Gaza Strip, but the conflict has bombarded agricultural areas and prevented people’s access to their land.

Our results suggest that in several locations around the world climate change will lead to an increase in famine-related pressure through more frequent and severe shocks and an increased vulnerability of populations engaged in agricultural livelihoods to these shocks. In much of Africa, much of Central and South America, South Asia, the Middle East, and Australia, agricultural models project a decrease in calorie production under conditions of no adaptation. Although not analyzed in this paper, a continued global prioritization of climate mitigation would be critical to limit the frequency and severity of these shocks. In already vulnerable regions due to conflicts and other existing holds, this can increase the risk of a humanitarian crisis.

Moreover, if given priority, the potential for adaptation to address this vulnerability under the current models is high in all of these regions. Current yield gaps can be closed with global financial investments in areas with high yield gaps, supporting technology, infrastructure, inputs, and sustainable intensification. In almost every region of the world, at least one crop is projected to see improved yields with climate change, and this does not account for the potential for improved varieties within crop types. Given the potential for adaptation, the extent to which climate change will impact local food production is under human control. For example, planting different crops and crop varieties can reduce the impact of climate change on agriculture, and yield gaps can be improved.

If there is a case in which a famine is declared, the role of climate change will be proportional to the societal decisions to invest or not invest in adaptation. In particular, policy options on climate finance are worth exploring especially at the national level, as countries craft transformative adaptation strategies as part of their national adaptation plans. Recent years have seen advancements in adaptation finance ambitions and mechanisms. Under Article 9 of the Paris Agreement [31], developed countries agreed to take the lead in mobilizing climate finance to support country-driven strategies, “taking into account the needs and priorities of developing country Parties.” The Green Climate Fund (GCF), the world’s largest climate funding facility, targets the most climate-vulnerable countries. The GCF plans to manage up to USD 50 billion by 2030 to catalyze climate actions for developing countries [32]. In addition, a new climate financing facility called the Loss and Damage Fund was created as a dedicated fund aimed at addressing and compensating for the losses and damage incurred by victims of climate impacts [33]. Major investments in climate-smart agriculture promote the adoption of climate-resilient crop varieties and water management strategies to reduce vulnerability to droughts, pests, diseases and other climate-related risks and shocks. World Bank investments in this area amount to almost USD 3 billion annually [34]. Other mechanisms include pre-arranged deployment of Climate and Disaster Risk Financing and Insurance (CDRFI), essentially crop insurance that offers timely payouts as close as possible to the occurrence of failed harvests, otherwise known as early action [35]. Despite these new mechanisms, mechanisms and finance are not well designed for adaptation in conflict-affected areas, which might be at risk of a conflict-related famine hold [36, 37].

In this respect, our study contributes to the evidence that disasters are not natural; disasters are political choices [38] and are directly related to the investments that governments and institutions allocate for climate adaptation and disaster risk reduction. In conflict-affected settings, there is a critical climate adaptation finance gap [39]; therefore, adaptation practices are limited [40]. This study argues that in these contexts, under a scenario of no adaptation in the agriculture sector, it is expected that disasters are more severe, exacerbating even further existing humanitarian challenges due to high levels of needs and unmet financial targets. The current political momentum, including the COP28 Declaration on Climate, Relief, Recovery and Peace [41], offers a timely opportunity to address the existing climate adaptation gap in conflict contexts. Yet, these high-level processes must be translated into contextual and flexible adaptation practices, adjusted to the high-risk volatility of conflict contexts. Our second scenario offers a critical perspective on the future impact of conflict in the context of famines; if agricultural systems in fragile and conflict-affected contexts are adapted and can cope with climate shocks, this can reduce pressures, even if there is a hold that limits humanitarian aid in a country or region. By prioritizing investment in adaptation of agricultural systems in the most complex humanitarian contexts, food crises can be reduced, which can be a relevant factor in reducing future famines.