Projected expansion of hottest climate zones over Africa during the mid and late 21st century

Projected shifts in thermal climate zones over Africa during the mid and late 21st century are assessed by employing the Thornthwaite thermal classification applied to 40 CMIP6 global climate models under the SSP1-2.6, SSP2-4.5 and SSP5-8.5 forcing scenarios. The CMIP6 multimodel ensemble mean reproduces the observed pattern of thermal zones during the reference period, albeit with some discrepancies. The projections reveal a gradual expansion of the hottest thermal type consisting of a northward and southward displacement of torrid climate zones, with this effect intensifying as greenhouse gas (GHG) forcing increases and the time horizon moves from the mid to the end of the century. In particular, the Mediterranean region, almost all southern African countries, part of East Africa and most Madagascar predominantly warm in present-day conditions, are projected to face mostly hot climates in the mid—21st century and torrid by the end of the 21st century in the high-end forcing scenario. Generally, in the mid—21st century, torrid climates expand by up to ∼15%, 20% and 27% of total Africa’s land areas for the SSP1-2.6, SSP2-4.5 and SSP5-8.5, respectively, with these fractions increasing to ∼16%, 28% and 42% in the late 21st century. Therefore, at the end of the 21st century for the high-end GHG concentration scenario, the African continent will be covered by 81%–87% of torrid climate type, which will have enormous impacts on the sustainable development of African countries.


Introduction
Human induced climate change by increased anthropogenic greenhouse gas (GHG) forcing affects the globe ubiquitously (IPCC 2021), in particular causing changes in local and regional climates (Sylla et al 2016a, Dosio et al 2021, Ranasinghe et al 2021, Seneviratne et al 2021. These changes have lead to disastrous impacts including heatwaves, droughts, floods, landslides and bush fires (e.g. Russo et al 2015, Cloutier et al 2016, Mukherjee et al 2018, Tabari 2020. Of great concern within the global change debate is the occurrence of possible shifts in thermal climate zones. These are areas predominantly identified by temperature patterns and impact for example biodiversity, agricultural systems, public health and the energy sector (Mahlstein et al 2013, Porter et al 2014. In fact, thermal climate zones distribution determines the ecosystem of living species and the energy consumption from residential buildings (Bai et al 2019, Kim and Bae 2021). As temperatures will continue to rise due to anthropogenic warming (IPCC 2021), changes in thermal climate zones will accelerate ecosystem 2. Data and methods

Data description
The ensemble mean of 40 GCMs from CMIP6 for the reference period  and two future periods, i.e. mid  and late (2071-2100) 21st century, are used to examine changes in thermal climate zones over Africa (see figure 1). To consider a range of possible future climates, we analyze different future SSPs (O'Neill et al 2016): SSP1-2.6 (i.e. low forcing: sustainability pathway), SSP2-4.5 (i.e. medium forcing: middle-of-the-road pathway) and SSP5-8.5 (high-end forcing: fossil fueled development pathway). The list of GCMs in the ensemble is presented in table 1. The analysis also includes regionalization over nine (9) sub-regions over Africa (see figure 1) identified by the Intergovernmental Panel on Climate Change (IPCC) Assessment Report 6 (AR6) (IPCC AR6 WGI 2021). Monthly near-surface air temperature from the GCMs is used to compute monthly PET to be used into the thermal climate classification methods as previously done by (Feddema 2005, Elguindi and Grundstein 2013, Sylla et al 2016b, Lang et al 2017. For model validation purpose we use the near-surface monthly mean temperature data from the Climatic Research Unit (i.e. CRU TS4.05; 0.5 • × 0.5 • of resolution; www.cru.uea.acuk/data) (Harris et al 2020) and from the fifth generation of European Re-analysis (ERA5; 31 km of resolution) (Hersbach andDee 2016, Hersbach et al 2020). Use of different observation datasets allows us to account for uncertainties in historical temperature data over Africa (Diallo et al 2012).

Methods
The revised Thornthwaite thermal climate classification method (Feddema 2005) is applied to monthly near-surface mean temperature of CRU, ERA5 and each of the CMIP6 GCMs (see table 1). This thermal climate classification uses PET and we compute annual PET from the Thornthwaite formulation. This method is widely used to estimate PET for climate classification (Elguindi and Grundstein 2013, Elguindi et al 2014, Sylla et al 2016b, Rahimi et al 2019. In order to enhance robustness of our results we also use the Hamon PET method (Lu et al 2005, McCabe et al 2015, Zhao et al 2021, Abeysiriwardana et al 2022, which also employs near-surface mean temperature. All datasets (i.e. CRU, ERA5, CMIP6 GCMs) are reggrided onto a 1 • × 1 • common grid. The following section describes the different formulations of the Thornthwaite and Hamon PET formulations.

Thornthwaite method
The Thornthwaite PET formulation was developed by (Thornthwaite 1948) and calculated monthly PET as follows: where, PET: monthly potential evapotranspiration [mm.month −1 ] T mean,i : monthly mean near-surface air temperature [ • C] for month i N: mean length of daylight I: annual heat index I i : monthly heat index for month i a: function of the annual heat index.

Hamon method
The Hamon method was developed by Hamon (1963) to estimate PET from mean monthly near-surface air temperature and day length. The formulation is as follows: Each of the estimated monthly PET is computed annually before performing the climatological mean for the reference period  and the two future periods (2041-2070; 2071-2100) under each of the SSP forcing scenario. The thermal climate classification (i.e. thermal types; Feddema 2005) derived from total annual PET is given in table 2. Six categories of thermal index are identified: frigid (PET between 0 and 300 mm yr −1 ), cold (PET between 300 and 600 mm yr −1 ), cool (PET between 600 and 900 mm yr −1 ), warm (PET between 900 and 1200 mm yr −1 ), hot (PET between 1200 and 1500 mm yr −1 ) and torrid (PET above 1500 mm yr −1 ), depending on the range of values of the total annual PET. These thermal types are expressed in mm.yr −1 and the range of values of each type is assigned in table 2. Future shift of thermal zones with respect to the reference period is assessed under SSP1-2.6, SSP2-4.5 and SSP5-8.5 forcing scenarios for the CMIP6 MME. A shift is defined here as an extension or recession of a thermal climate type or a change in the thermal category occupying a region. We consider the shift robust if the information is similar when using both methodologies to calculate PET (i.e. Thornthwaite and Hamon).
For a quantitative assessment of the shift in each thermal type, historical and projected changes (i.e. SSP1-2.6/SSP2-4.5/SSP5-8.5 minus reference period) of land-only area extent are computed over the Africa continent and for each of the 9 IPCC AR6 sub-regions (as defined in figure 1). The area cover of a climate type is expressed as percentage of the total Africa area.

Validation
The thermal classification applied to CRU, ERA5 and CMIP6 MME using the Thornthwaite and Hamon PET methods during the reference period  are intercompared in figure 2. For all methods, robust features of torrid climate type are observed (in both CRU and ERA5 and for both methodologies) over the WAF region including the Sahel (i.e. Senegal, Mali, Niger, Chad, Sudan, Burkina Faso) and the northern part of Gulf of Guinea (i.e. Côte d'Ivoire, Ghana, Benin, Togo, Nigeria), the western and central Saharan Desert (SAH), parts of northern and central East Africa (i.e. NEAF and CEAF) including South Sudan, northeast Kenya and Horn of Africa. The hot climate type mainly prevails over part of the northeastern SAH (i.e. Libya and Egypt), Central Africa (i.e. Congo Basin, Congo, Central Africa Republic), in some coastal countries of ESAF (i.e. Mozambique) and in western Madagascar. The warm climate type is mainly found in the WSAF region including Angola, Namibia, western Zambia, parts of ESAF (i.e. eastern Zambia, Zimbabwe, southern Botswana), and eastern Madagascar. The cool climate type occurs in South Africa and the coastal Mediterranean regions of Africa (MED-AF).
A few discrepancies, however, exist across the results with different datasets and methodologies. The most notable ones are in the coastal areas of MED-AF and in South Africa, along the coastlines of the Gulf of Guinea in WAF and over the CAF. In fact, the cool thermal type is less extended in the coastal region of MED-AF (i.e. in Morocco and northern Algeria) and in South Africa when using the Hamon PET compared to the Thornthwaite method. In addition, along the coastlines of the Gulf of Guinea (i.e. southern Côte d'Ivoire and southern Ghana), the torrid thermal type prevails based on CRU data while the hot thermal type is found based on ERA5, indicating that the latter has lower temperatures than CRU. Finally, over Cameroon and Gabon, using CRU with both methods and ERA5 with Hamon PET yields a hot dominant climate type, whereas ERA5 along with the Thornthwaite PET produces a prevailing warm thermal type. Note, however, that these discrepancies are very localized and do not alter the estimated global pattern of thermal zones in Africa.
The CMIP6 MME yields an overall agreement with CRU and ERA5 in terms of patterns of thermal types over most of the regions. In particular, it captures the torrid climate type over WAF, western and central SAH and part of northern and central East Africa (i.e. NEAF and CEAF), including the Horn of Africa. It also captures the dominant hot thermal type over the CAF, northeastern SAH, the coastal countries of CEAF and ESAF and the western region of Madagascar. In addition, it reproduces areas of warm climate conditions, and specifically along the coastal regions of MED-AF, some localized areas in CEAF, most of WSAF and eastern Madagascar. The cool thermal type observed over South Africa country is also well captured.
The CMIP6 MME also exhibits some biases, such as the extension of the hot type in the northern portions of the Southern Africa region (i.e. northern WSAF and ESAF), and the lack of cool thermal type in the coastal MED-AF region (i.e. Morocco and northern Algeria).
These biases in thermal types are evidently related to corresponding biases in surface air temperature. Figures SM1 and SM2 show the distribution of mean monthly temperature for the CRU observations (figure SM1(a)), ERA5 reanalysis (figure SM1(b)) and CMIP6 MME (figure SM1(c)) along with the mean monthly temperature bias in the CMIP6 MME with respect to the CRU observations (figure SM2(a)) and the ERA5 reanalysis (figure SM2(b)) during the reference period . The CMIP6 MME shows a warm bias in the northern region of WSAF of up to 1.2 • C compared to both the CRU dataset and ERA5 reanalysis. Over northern ESAF, CMIP6 shows a warm bias compared to both CRU and ERA5 but more pronounced with respect to the latter. In addition, a warm bias is also present in MED-AF of about 0.5 • C with respect to both the CRU and ERA5 datasets.
For a more quantitative assessment of the CMIP6 MME thermal type simulation during present climate conditions, the percentage of land-only areas occupied by each thermal type for the reference period in the 9 IPCC AR6 sub-regions and in the entire African continent are intercompared in figure 3 for the CRU observations, the ERA5 reanalysis and the CMIP6 MME using both the Thornthwaite and Hamon PET formulations. Figures SM3 and SM4 provide similar information but separately for the Thornthwaite and Hamon methods.
As a result, for the whole Africa (land-only areas) based on the CRU observation dataset (ERA5 reanalysis), torrid, hot and warm climate conditions dominate with a range of 37.77%-47.36% (35.95%-43.98%), 24.3%-41.35% (22.29%-40.48%) and 18.13%-22.18% (20.03%-25.52%) area extent, respectively. The proportion of observed cool climate varies between 1.7%-6.13% (3.09%-7.76%). It should be emphasized that for the whole Africa the climate classification using the Thornthwaite PET shows larger areas with torrid and warm thermal types compared to the Hamon PET, while this exhibits greater areal extents occupied by hot climate type.
Compared to the observed thermal-based estimates, the CMIP6 MME captures very well the fraction of land occupied by each thermal type, with only a few exceptions. For example, over the CAF, the CMIP6 MME slightly overestimates the areas covered by torrid climate types, whereas it somewhat underestimates the areas under hot climate conditions in both PET formulations. Furthermore, in the SAH, the models slightly underestimate the torrid thermal type area extent while overestimating the warm thermal type fraction. Overall, for the whole Africa, the CMIP6 MME captures very well the observed fractional area extent of each thermal category considering both PET formulation.
In summary, torrid, hot and warm climate conditions are predominant in Africa. These categories are primarily found in the SAH, WAF, CAF and NEAF regions irrespective of the PET formulation considered. SAH experiences the largest proportion of areas under torrid climate, followed by WAF, CAF and NEAF, with greater fractions of hot conditions in CAF. In CEAF, ESAF and WSAF hot and warm conditions prevail with similar proportions. The CMIP6 MME is able to capture this regional distribution, with small differences compared to observations over a few regions (i.e. SAH and CAF).

Future changes of thermal climate zones 3.2.1. Mid-21st century (2041-2070)
The reference period and projections of thermal climate types under the SSP1-2.6, SSP2-4.5 and SSP5-8.5 forcing scenarios using the Thornthwaite and Hamon PET methods for the mid-century period (i.e. 2041-2070) are shown in figure 4 while figure 5 displays their future shifts with respect to their reference period. Robust increases in torrid and hot climates spatial extent are observed in all SSP scenarios. Under the SSP1-2.6, the torrid climate type, initially confined within the Sahara Desert (i.e. SAH), WAF and part of East Africa (i.e. NEAF and CEAF), extends north into the Mediterranean region and south to cover part of Central Africa (i.e. northern Congo basin). The Southern Africa region (i.e. WSAF and ESAF), dominated in present day climate by the warm type, shifts to a hot thermal type in the mid-century, mainly in countries such as Namibia, part of Angola, southern Botswana and western Zambia.
Under the SSP2-4.5 scenario, the areas covered by torrid thermal type increase towards the south reaching the central Congo basin and the coastlines of CEAF and ESAF (i.e. Kenya, Tanzania and Mozambique) at the expenses of hot and warm climate conditions.
Under the most extreme SSP5-8.5 scenario, torrid climate is projected to extend north and south covering most areas of the continent, with only a few exceptions in localized areas. For example, a large part of WSAF and ESAF is still dominated by the hot thermal type, the southern part of ESAF still experiences cool thermal conditions, and the coastal areas of Morocco and Algeria, central Angola, Ethiopian highlands and East African highlands are still subject to the warm thermal type.  SSP1-2.6 (2041SSP1-2.6 ( -2070, SSP2-4.5 (2041SSP2-4.5 ( -2070 and SSP5-8.5 (2041SSP5-8.5 ( -2070 relative to reference period  using Thornthwaite (a), (c), (e) and Hamon (b), (d), (f) potential evapotranspiration (PET) method. no chg stands for no change, Ht -> Td: change from hot to torrid, Wm -> Td: change from warm to torrid, Wm -> Ht: change from warm to hot, Cl -> Wm: change from cool to warm.
There is thus a robust recession of hot and warm climates and an extension of the torrid climate type with increasing GHG concentrations already at mid-century. In fact, the Mediterranean region, part of CEAF and most countries in WSAF, ESAF and Madagascar progressively shift from warm to hot while the remaining areas of the Sahara Desert and most part of central Africa (SAH and CAF) gradually shift from hot to torrid (i.e. figure 5). Note that there are some differences between results with the Thornthwaite and Hamon methods. For instance, for the low forcing scenario (i.e. SSP1-2.6), the northward and southward displacement of torrid climate boundaries is more extended when using Thornthwaite compared to Hamon, while for the higher forcing scenario (i.e. SSP5-8.5) the spatial extension of climate type change is similar with both methods. Figure 6 presents the reference period and projected climate types under the SSP1-2.6, SSP2-4.5 and SSP5-8.5 scenarios using the Thornthwaite and Hamon PET methods for the late 21st century period (i.e. 2071-2100). Figure 7 shows the corresponding shifts from one climate type to another. Similar patterns of changes are projected as for the mid-century period, with larger spatial expansion of areas under torrid conditions. For example, under the SSP1-2.6 scenario in the Congo basin (in CAF) the torrid climate has wider spatial extent compared to mid-century. For the SSP2-4.5 scenario, more areas in southern Africa (i.e. WSAF and ESAF) experience torrid climate, including northern Botswana, compared to the mid-century.

Late 21st century (2071-2100)
In the SSP5-8.5 scenario, torrid climate is projected to cover almost the entire Africa continent, stretching to the coastal Mediterranean region in the north and to WSAF and ESAF in the south. The coastal areas of South Africa previously dominated by cool thermal type during the reference period and the mid-century, shift to warm climate type during the late 21st century.
These changes in thermal types are closely related to corresponding changes in near-surface temperatures pattern. Figure SM5 presents the distribution of the late 21st century change in annual mean, maximum and minimum temperature climatologies for SSP1-2.6, SSP2-4.5 and SSP5-8.5 scenarios. The annual mean, maximum and minimum temperatures display a spatial pattern of a uniform increase of about 0.5 • C under SSP1-2.6. Under SSP2-4.5, maximum increase of 3 • C-3.5 • C is found over the Sahara, the Mediterranean and Sahelian regions of West Africa and a minimum increase of 1.5 • C-2 • C is found over southern part of West Africa, Central Africa and part of East Africa (WAF, CAF, CEAF and NEAF). We also note a considerable warming of 3 • C prevailing for maximum temperature over the southern part of the continent (i.e. WSAF and ESAF). Under SSP5-8.5, the larger increase of up to 5 • C is projected over most of the regions except in West and Central Africa regions that will face an increase in minimum temperature of about 4 • C. As expected, areas of pronounced temperature changes coincident with areas of shift toward hottest thermal type.
Overall, there is a clear shift of thermal climate zones toward the torrid thermal type across Africa throughout the 21st century, with this expansion covering more regions during the late century compared to Figure 7. Spatial shifts of thermal climate zones for future SSP1-2.6 (2071-2100), SSP2-4.5 (2071-2100) and SSP5-8.5 (2071-2100) relative to reference period (1985-2014) using Thornthwaite (a), (c), (e) and Hamon (b), (d), (f) potential evapotranspiration (PET) method. no chg stands for no change, Ht -> Td: change from hot to torrid, Wm -> Td: change from warm to torrid, Wm -> Ht: change from warm to hot, Cl -> Wm: change from cool to warm. the mid-century. At the end of the century and under the high-end fossil fueled development pathway (i.e. SSP5-8.5), torrid climate occupies almost the entire Africa continent, except for localized areas, e.g. in the southern part of South Africa. In fact, the Mediterranean region, a large part of CEAF and most countries in southern Africa (i.e. WSAF and ESAF) and Madagascar steadily shift from warm to torrid. In addition, the shift from hot to torrid in many areas of the central, eastern and northern Sahara Desert and most part of central Africa (SAH and CAF) during the mid-century increasingly expands spatially in the late 21st century.
These findings are partly consistent with previous findings by Elguindi et al (2014) and Sylla et al (2016b), who projected an extension of torrid climate into central and parts of southern Africa using CMIP5 and in West Africa using CORDEX, respectively.

Projected areal extent
To quantify the shift in thermal zones, the change in area extent of each thermal type is assessed over the 9 Africa sub-regions and the whole African continent (land only) in figures 8 and 9, which present the changes in the fractional cover of different thermal types using the Thornthwaite method under the SSP1-2.6, SSP2-4.5 and SSP5-8.5 scenarios in the mid (2041-2070; figure 8) and late century (2071-2100; figure 9).  The unit for the changes is percentage of total land area of the whole Africa continent. Results using the Hamon method are not shown as they show similar changes compared to Thornthwaite (see supplementary materials figures SM6 and SM7). Our results first indicate that the sign of change is the same across all SSP scenarios and the changes are greater under the SSP5-8.5 scenario than the SSP2-4.5 and SSP1-2.6. Figure 8 shows a general increase in area extent of torrid thermal type across all regions and scenarios with the largest increases projected in SAH (8.25%) and CAF (5.45%) under SSP5-8.5. We also find a decrease in fractional area cover of hot and warm types of about −5.15%% and −3.08% in SAH and −3.25% and −2.2% in CAF, respectively, for the same scenario and period. Another notable feature is the slight increase in area extent of both torrid and hot thermal types and a decrease of warm climate type area coverage over the ESAF and WSAF regions. In the MDG and MED-AF regions, the changes are small and not significant. Considering the whole Africa continent, there is an increase of torrid thermal type area extent in the range of 15.2%-27%, whilst hot and warm climate area coverages decrease by 6.7%-1.35% and 16.56%-11.12%, respectively. As expected, the increase and decrease in area extent are larger under the high end SSP5-8.5 scenario compared to the other two. Figure 9 shows similar changes as in figure 8, but with larger magnitudes. For example, the change in the proportion of torrid climate type during the late 21st century reaches 9.45%% in SAH and 7.85% in CAF. The corresponding decreases in hot and warm conditions area coverage are 6.2% and 3.2% in SAH and 5.31% and 2.55% in CAF under the SSP5-8.5. In addition, over WSAF and ESAF a moderate spatial expansion of torrid and hot climates and a recession of warm areas prevail as in the mid-century but again with larger values. Consequently, considering the whole Africa continent, the torrid climate type increases in spatial extent by up to ∼16% for SSP1-2.6, 28% for SSP2-4.5 and 42% under SSP5-8.5 scenario at the expenses of warm and hot zones.
Overall over Africa, warm and hot thermal types are projected to be progressively converted into the torrid thermal type across almost all land areas as GHG concentrations increase into the 21st century. By mid-century, the torrid climate type using both Thornthwaite and Hamon PET estimates covers a total fractional area of the African continent in the range of 49%-60%, 53%-65% and 61%-72% under SSP1-2.6, SSP2-4.5 and SSP5-8.5. At the end of the century, Africa will be covered by the torrid climate type at 50%-61%, 62%-73% and 81%-87% under SSP1-2.6, SSP2-4.5 and SSP5-8.5, respectively.

Conclusion and discussion
In this paper, we investigated the shift in thermal climate zones over Africa during the mid (2041-2070) and late 21st century (2071-2100) with respect to the reference period  under three SSP forcing scenarios: a low forcing scenario (SSP1-2.6), a medium forcing scenario (SSP2-4.5) and a high-end forcing scenario (SSP5-8.5). This was achieved using multiple PET temperature-based estimation methods, the Thornthwaite and Hamon methods, applied to CRU observations, the ERA5 reanalysis and the CMIP6 MME mean (MME).
The MME is able to reproduce the general thermal conditions of the African continent, although it overestimates the spatial extent of hot climate zones in the northern part of Southern Africa region compared to the observation and reanalysis. In particular, among the well simulated thermal types by the MME we identified the torrid climate type of West Africa (i.e. Sahel and Gulf of Guinea), western and central Sahara Desert (SAH), and part of north and central Eastern Africa (i.e. NEAF and CEAF), the hot climate type in central Africa, northeastern SAH and western Madagascar, the warm and cool types in some portions of the coastal Mediterranean of Africa (MED-AF) and CEAF, most part of WSAF and eastern Madagascar, and the cool type characterizing the South Africa country.
Overall in Africa, torrid, hot and warm climates are the dominant thermal conditions, with the torrid type exhibiting the largest fractional area cover and in general, the CMIP6 MME is able to capture this area distribution.
The most important climate shifts occur in the Mediterranean region, CEAF, southern Africa and Madagascar. These areas initially warm will become mostly hot by mid-21st century and overall torrid by late 21st century. In addition, areas in central Africa, particularly the Congo basin previously hot will become torrid by the end of the 21st century.
It is important to note that although the multimodel ensemble of the 40 CMIP6 GCMs shows a good performance in simulating temperature and PET, these results are still subject to some uncertainties. They can be originated from future emissions, internal climate variability and inter-model difference (RÄISÄNEN 2007, Knutti and Sedláček 2013, Northrop and Chandler 2014, Zhang and Chen 2021. Nevertheless, the use of a large ensemble (i.e. 40 CMIP6 GCMs here) provides more robust results (Semenov and Stratonovitch 2010, Gao et al 2019, Doi & Kim 2022.
From our study, it is clearly evident that anthropogenic climate change will cause a shift of thermal zones and further displace the boundaries of the hottest thermal types over all African regions during the mid and late 21st century. In particular, most of the continent will be subject to torrid climate conditions by the end of the 21st century in the SSP5-8.5 scenario.
The projected shifts (i.e. extention of torrid thermal zones and recession of hot, warm and cool thermal types) will result in a disruption of existing ecosystem structure and a loss of biodiversity (Mahlstein et al 2013, Román-Palacios andWiens 2020). In fact, climate controls the distribution of species ranges and rate of primary productivity (Grimm et al 2013, Elguindi et al 2014.
Another implication of our results is an increase in the level of heat stress for human, crops and animals. Shifting towards the hottest thermal climate type can substantially affect the most vulnerable population to heat stress, i.e. outdoor intensive workers in sectors such as agriculture, forestry, fishing, construction, and utility suppliers (Xiang et al 2014, Acharya et al 2018, Moda et al 2019. In addition, changes towards hottest thermal climate conditions in regions previously favorable to farming, can cause heat stress on crops resulting in loss of crop production (Teixeira et al 2013). Finally, as livestock cattles substantially contributing to food production, grow in specific climate conditions, the expansion of the hottest thermal type can trigger heat stress on livestock in previously favorable regions, resulting in an alteration of the livestock production systems (Salem et al 2011, Renaudeau et al 2012.
A final concern is the energy demand and consumption and more generally the design for energy-efficient buildings. Because of the displacement of boundaries of the torrid thermal type, some regions will experience novel climate types and unprecedented warming, leading to an increase in energy demand for African countries that already struggle to meed to the present-day demand. Our results thus call for an update in the design of buildings for more efficient energy savings.
An assessment of the impacts of these thermal zones shifts on heat stress for human, crops and animals as well as energy demand and availability will be presented in subsequent papers.

Data availability statement
The data that support the findings of this study are available upon request from the authors.