Trends and impacts of climate change on crop production in Burkina Faso

Global climate change (CC) constitutes an important challenge worldwide with severe impacts on sustainable development. Over the 20th century, climate conditions have experienced significant changes in many regions worldwide driven by a continuous increase in global greenhouse gas emissions. This has caused a long-term warming trend since pre-industrial times, estimated at 0.2 °C per decade due to past and ongoing emissions (IPCC 2022).

This century revealed an increasing trend of the global mean surface temperature (Wedajo et al. 2021; Worku et al. 2022) and precipitations (Nouaceur & Murarescu 2020; Wedajo et al. 2021). At the regional or local scale, both increases and decreases in the spatiotemporal trends of precipitations are observed (Berg & Sheffield 2018; Atiah et al. 2021). Several studies reported increased and decreased trends of climate extremes, such as hot days/heat index and cold/frost days, were experienced, respectively, for nearly all land areas during this century (Sillmann et al. 2013).

Mean temperature was projected to rise from 3 to 4 °C leading to a 15–35% loss in crop yields in Africa and West Asia this century (Chadalavada et al. 2021). Similarly, about 25–35% of yield loss is expected in the Middle East with a temperature rise of 3–4 °C (Chadalavada et al. 2021).

West Africa is a CC hotspot region characterised by high climate variability, extreme climatic events, severe exposure and low adaptive capacity (Heubes et al. 2013). CC in this region negatively impacts agricultural production (Heubes et al. 2013; Nelson et al. 2018) and strongly affects the well-being of poor households relying on natural resources (Jones & Thornton 2009) and cropping for their survival. Altered climate conditions in West Africa are reported to cause yield declines of 10–20% for millet and 5–15% for sorghum (El Bilali 2021). The decline in crop yield implies income reduction for farmers and the country's economy (El Bilali 2021) and less resilience to global changes. Inappropriate or poor adoption of mixed cropping in the region contributes to low resilience (Ifejika Speranza 2010; Devkota et al. 2022).

Like most West Sahelian countries, Burkina Faso strongly depends on agriculture and animal breeding. Farmers in Burkina Faso are more vulnerable to CC because usually, they have very few alternatives facing climate risk. Cropping systems are predominantly subsistence-oriented, with mainly small holdings characterised by small farm size and highly variable herd size and composition. Cereal crops such as sorghum (Sorghum bicolor), millet (Panicum sp.) and maize (Zea mays L.) constitute the main pillars of Burkina Faso's food security and across West Africa (Waongo et al. 2015). Besides these cereal crops are groundnut (Arachis hypogaea) and cowpea (Vigna unguiculata), which are double staple and cash crops for vulnerable smallholder farmers in Burkina Faso.

In recent decades, there has been a growing literature on CC and variability in the West African Sahel (Pirret & Daron 2019). Yet, these studies were more based on farmers' perceptions and impact simulations from models. Unavailability of long-term data is a common challenge for most West African countries. In Burkina Faso, the longest available time series data on the targeted study zones was about 37 years. Therefore, most studies relied on satellite data or resorted to surveys in assessing CC impact on crop production. Accordingly, updated data on crop's sensitivity to CC, namely to climate extremes, are lacking in Burkina Faso. Furthermore, most of the studies focused only on the impact of raw climate variables (mean temperature and rainfall) on crop production. This may hide some effects posed by climate extremes. From this perspective, the current study used climate cold, hot and wet indices as predictors in the analysis of climate implications on crop production across climatic zones of Burkina Faso. Specifically, this study sought to:

• (i)

  analyse extreme climate indices across different climatic zones of Burkina Faso;

• (ii)

  analyse the influence of extreme climate indices on the yield of five major crops in Burkina Faso. We hypothesise an increasing CC impacts on crop yield with the climate gradient.

Findings from this study may contribute to a better understanding of the influence of CC and variability on crop yield in the West African Sahel, which is of great relevance for farmers and policymakers to address food security.

Long time series of climate records spanning the last 60 years (1961–2020) were obtained from the National Agency of Meteorology of Burkina Faso. The daily climate datasets (rainfall, maximum and minimum temperature) were from three weather stations, each located in one of the three climatic zones of Burkina Faso, Dori in the Sahelian zone, Ouagadougou in the Sudano-Sahelian zone and Gaoua in the Sudanian zone.

This study considered five major crops: maize, millet, sorghum, cowpea and groundnut. Maize, millet and sorghum represent staple crops and greatly contribute to food security in the West African region (Waongo et al. 2015). Cowpea and groundnut are cash crops and sources of income for smallholder farmers. In each climatic zone, we collected data on annual yields (kg/ha) of the five crops from 1984 to 2020 from the Ministry of Agriculture. The crops data were obtained from three provinces: Seno in the Sahelian zone, Kourweogo in the Sudano-Sahelian zone and Ioba in the Sahelian zone.

Data were quality controlled before subsequent analysis. Missing data were not filled but substituted with the value −99.9 in the input text files as required by ClimPACT2 (Alexander & Herold 2016). Homogeneity of data was tested with RhtestsV4 package and the Quality Control (QC) option in ClimPACT2 to remove outliers like minimum temperature greater than maximum temperatures and negative temperature records (Alexander & Herold 2016). The homogenised data was subsequently used in the multiple regression analysis between climate extremes and crop production.

To assess climate trends in the three climatic zones, 23 climate extreme indices known as best climate descriptors were selected among the 64 indices commonly used for climate trends analysis (Alexander & Herold 2016). The first 10 out of the 23 selected indices (Table 2) have important implications in the sectors of agriculture and food security, water resources and hydrology (Alexander & Herold 2016). These indices were computed using the R package ClimPACT2 GUI (Alexander & Herold 2016). We also considered other indices such as warm spell, cold spell, dry spell, wet spell, rainfall anomaly and intensity which better describe climate conditions in Burkina Faso where high temperature and irregular rainfall strongly determine water flow and availability for cropping. Moreover, water availability for plant use was highly influenced by the evapotranspiration, which is also influenced by temperature. Climate indices trends were based on Sen's slope (using the zyp package in R) (Alexander & Herold 2016). Sen's slope reflects the median slope of all ordered pairs of points in a dataset and is more appropriate for calculating trends in extreme values. Sen's slope analysis is a non-parametric method (Sen 1999).

To determine the importance of each selected index and the most affected decades in terms of climate deterioration within each climatic zone, we first computed the means values of climate indices over six decades (1961–1970, 1971–1980, 1981–1990, 1991–2000, 2001–2010 and 2011–2020). Afterwards, climate indices were subjected to principal component analysis (PCA) using the R package ‘FactomineR’ (Lê et al. 2008). For this purpose, time series of chosen indices were split and analysed according to the six defined decades within each climatic zone.

Pearson correlation test and multiple linear regression analyses were performed to assess the influence of extreme climatic conditions on the yields of the five crops. Mean temperature, total precipitation, precipitation intensity and climate extreme indicators were used as predictors. Climate data from 1984 to 2020 fit the time series length of crop yield data.

We assumed non-climatic factors (physiographic, edaphic, biotic and socio-economic) affecting crop production to be constant to investigate how extreme climatic indices alone could influence yield patterns in the study area.

Generally, the climate extremes indices revealed a warming trend throughout the three climatic zones from 1961 to 2020 (Supplementary Appendix 3). Cold extremes indices (Cold spell duration indicator (CSDI), Amount of cool days (TX10p), Amount of cold night (TN10p)) (Supplementary Appendix 4) showed a decreasing trend. In contrast, hot extremes (Hottest day (TXx), Hottest night (TNx), Warm spell duration indicator (WSDI), Amount of hot days (TX90p), Amount of warm nights (TN90p)) revealed an increasing trend across all the studied zones (p < 0.05). The findings are within the range reported by Panda et al. (2014) for India for the period 1971–2005. It could imply that the patterns of extreme temperature indices after the pre-industrial age in the tropics are similar. Furthermore, the coldest night (TNn) and coldest day (TXn) generally showed an increasing trend in all zones, indicating a reduction in the intensity of cold nights and days in the country. Our findings suggest an overall warming trend in Burkina Faso, as reported by several studies conducted in West Africa (New et al. 2006; Barry et al. 2018). According to Barry et al. (2018), warm days and nights have become more frequent in Burkina Faso during 1960–2010. A warming trend of both days and nights between 1961 and 2000 was also observed within West Africa (New et al. 2006).

These warming trends have several implications on the agricultural sector. Temperature increase within an optimal range induces an increase in yield whereas below or beyond the optimal range results in reduced yield (Parthasarathi et al. 2022). This is because crops speed through their development with a temperature beyond the optimal temperature range, producing less grain (Cline 2008). Excessive temperature stresses lead to various physiological and biochemical changes in crops during their phenological development (Parthasarathi et al. 2022; Bita & Gerats 2013). High temperatures decrease seed germination percentage, plant emergence, results in poor seedlings vigour, abnormal seedlings, and further decrease radicle and plumule growth (Parthasarathi et al. 2022). Furthermore, excessive temperatures disturb the water cycle and interfere with crops ability to access and use moisture (Cline 2008). Both evaporation from soil and plants and transpiration are accelerated by a temperature rise (Cline 2008).

The number of normal nights is continuously converted into both extremes of hot or cold nights, implying a lack of a clear pattern in the trend of night extremes in the Sudanian zone (night cooling or warming) as observed (night warming) in the Sudano-Sahelian and the Sahelian zone (p < 0.05). This trend was consistent with the diurnal temperature range (DTR), which showed no significant trend in the Sudanian climatic zone but significantly increased in the Sahelian and Sudano-Sahelian zones. Our findings disagree with previous studies on climate extremes within West African regions (New et al. 2006; Barry et al. 2018) and highlight the accuracy of climate trend-related studies at the local scales. These results suggest more changes in climate conditions across the Sahelian and Sudano-Sahelian zones of Burkina Faso, contrary to the Sudanian zone.

The Sahelian and Sudano-Sahelian zones experienced longer cold spells (CSDI) between 1961 and 2020, while shorter cold spells characterised the Sudanian zone. Results in the Sudanian zone are consistent with New et al. (2006), who reported a decrease of a cold spell in West Africa. The Sudano-Sahelian zone, which constitutes the transition between the Sahelian zone and the Sudanian zone experienced a persistence in warm period as revealed by the WSDI within the study period (1961–2020). This result corroborates previous studies (New et al. 2006) supporting an average increase of warm spells by 2.4 days per decade in the West African region.

The comparison between cold and hot climate extremes revealed that hot extremes evolve faster than cold extremes. This finding suggests that the warm tails of the daily temperature distributions are changing faster than the cold tails.

The PCA revealed different patterns of decadal climate conditions within the study area over the last six decades for both rainfall and temperature, with the Sahelian and the Sudano-Sahelian being the most vulnerable to CC compared to the Sudanian zone.

The 2011–2020 decade was the warmest, with a mean temperature of 28.0, 29.38 and 30.2 °C in Sudanian, Sudano-Sahelian and Sahelian zones, respectively (Supplementary Appendix 7). Also, this decade generally experienced the hottest day and night of 1961–2020. This aligned with the warming trends revealed by several climate studies in West Africa (New et al. 2006; Barry et al. 2018). These findings contrast those of WMO (2013) that indicated the decade 2001–2010 to be the warmest in Africa and even worldwide in all the domains (land, ocean and land–ocean). This contrasting result could be due to the periods of studies: 1901–2010 for the WMO versus 1961–2020 for the current study. Africa experienced warmer than normal conditions every year of the decade 2001–2010 for nearly 94% of reporting countries from WMO's survey (WMO 2013). The coldest decade was not identic across the three climatic zones studied. Indeed, the decade 1961–1970 was the coldest (28.8 °C) in the Sahelian zone while the decade 1971–1980 was instead the coldest in Sudano-Sahelian (28.0 °C) and Sudanian (27.4 °C) zones. These results also differ from that of WMO (2013) that rather indicated the decade 1981–1990 to be the coldest decade on land. The contrasting findings could be due to the spatiotemporal difference in studies scales (global versus local).

Furthermore, the highest frequencies of both warm nights and hot days were not experienced in the same ways across the zones. The 2001–2010 decade was characterised by the highest frequencies of warm nights and hot days in Sudanian and Sudano-Sahelian zones. For WMO (2013), 2001–2010 was a very exceptional hot decade of 1901–2010. The Sahelian zone experienced the highest frequencies of warm nights and hot days during the decade 2011–2020. Similarly, cold extremes did not show the same patterns across zones. The highest frequencies of cool days and cold nights were observed in the Sahelian zone from 1961 to 1970. These frequencies generally were observed from 1971 to 1980 in the Sudano-Sahelian and Sudanian zones.

Generally, the last three decades distinguished themselves by hot climate extremes from the first three decades, mainly characterised by cold climate extremes in all three climatic zones. This aligned with previous research findings (Panda et al. 2014).

The multiple linear regression between yields and climate extremes suggested that climate variability and change have different directions of influence (negatively or positively) on crop production (Table 3). Similar findings of this study were reported on cowpea yields in Nigeria during 1961–2006 (Ajetomobi & Abiodun 2010) and in Mali (Butt et al. 2005). Furthermore, Waha et al. (2013) also reported that CC adversely affected maize production (10–33% yield decrease) in Sub-Saharan Africa, with positive impacts found in mountainous and cooler regions of South and East Africa (6% yield increase). About 67% (14 over 23 indices) of the studied extreme climate indices significantly influenced crop yields across the three climatic zones. Some of these indices adversely impacted yields. This result is congruent with several African authors that similarly revealed the negative implication of climate extremes events on crop germination, growth and production (Abubakar et al. 2020). The negative impacts consisted of a severe disruption in plant development through several morphological, physiological, biochemical and molecular changes leading definitely to yield decline (Chadalavada et al. 2021). These impacts were found to be more severe within the Sudano-Sahelian and Sahelian zones than the Sudanian zone (Table 3). Crop production in the Sudanian and Sahelian zones was influenced by six and seven climate indices, respectively, while five climate indices affected crop yields in the Sudano-Sahelian zone.

The negative impact of extreme warm indices from this study may not vary much across sub-Saharan Africa due to the continuous rise in temperature on the continent and could alter future food security in Africa and even at a global scale (Chadalavada et al. 2021).

Compared with the other zones, maize production was more influenced in the Sudano-Sahelian zone where both positive and negative influences of extreme climate conditions were observed. A one-unit increase in the coldest night (night warming) and 1 day increase of cold spell duration (day cooling) could have caused a decrease and an increase in maize yield by 85.5 and 50.7 kg/ha, respectively. It seems that night and day temperature trends could affect maize production differently in the Sudano-Sahelian zone of Burkina Faso. Day cooling tends to be favourable to maize production, while the warming trend of the coldest night could have stemmed it. A global-scale study indicated that unusually cold and warm days negatively affect maize yields (Vogel et al. 2019).

A decline in millet yield has been generally observed across all zones due to longer dry spells and climate warming, except in the Sudanian zone where both increased and decreased millet yield trends were observed. A unit warming of the coldest day could have increased millet yield by 97.7 kg/ha, while a one-degree increase in the coldest night (warming of the coldest night) may have caused 45.6 kg/ha reduction in millet yield. Millet yield could have also decreased by 1.6 kg/ha for a 1-day increase in consecutive dry days (dry spell). Such crop failure due to dry climate conditions is highlighted from previous findings (Abubakar et al. 2020). In the Sudano-Sahelian zone, millet yield was also declining by 77.6 kg/ha due to a one-degree increase in the coldest night. Finally, in the Sahelian zone, a 1% increase in cold night frequency (nights cooling) causes a decline in millet yield by about 23.8 kg/ha. It seems that millet production has been more sensitive to changes in the night temperature. The negative impact of the changes on millet yield is more pronounced in the Sudano-Sahelian and Sudanian zones than in the Sahelian zone, where such changes were more observed. Furthermore, our findings related to wet extreme climate indices are in line with research that indicated crop failure is favoured by a decline in rainfall (Abubakar et al. 2020) while rain resumption is potentially favourable to crop yields. Indeed, heavy rain days favourably increase millet yield by 11.8, 17.8 and 26.7 kg/ha in Sudanian, Sudano-Sahelian and Sahelian zones, respectively.

The warming trend in the zone might have offset to some extent the observed favourable effects of rain on millet yields (Wheeler et al. 2000; Panda et al. 2014; Abubakar et al. 2020). Indeed, hot temperatures at the time of flowering can reduce the potential number of seeds or grains formed or developed (Wheeler et al. 2000).

Suppose sorghum production could have been drastically affected by the average temperature (yield decline of 391.2 kg/ha due to one-degree increase in average temperature) in the Sudanian zone. In that case, this crop production seems rather to have been favoured by changes in hot climate extremes in the Sudanian and Sahelian zones of the country. Indeed, a unit increase in warm nights (nights warming) in the Sudanian zone could have increased sorghum yield by 24.6 kg/ha. Furthermore, a 1% increase in hot days (day warming) in the Sahelian zone could have induced an increase (by 18.3 kg/ha) of sorghum yield. Thus, the present stade of climate warming seems to have been favourable to sorghum production in the abovementioned zones.

Moreover, a millimetre increase in the total precipitation could have induced sorghum yield increase by 1.2 kg/ha in the Sahelian zone. The Sudano-Sahelian zone did not show significant impacts of climate indices on Sorghum yield. Groundnut yield was also observed to be negatively associated with the coldest day and consecutive dry days. A yield decline by 57.7 and 3.4 kg/kg could have been observed in the zone due to a degree and a day increase in the coldest days and dry spell, respectively.

Groundnut production was seriously threatened only in the Sudano-Sahelian, while Sorghum was the less disturbed crop by extreme hot days and nights temperature. The results could be due to the warm-weather crop nature of the Sorghum (Du Plessis 2008).

Cowpea yield seems to have been more affected in the Sudanian and Sahelian zones, particularly, by cold climate extremes (cool days and cold nights). Indeed, a one-unit increase in cool days (day cooling) within the Sudanian zone could have reduced cowpea yield by 58.4 kg/ha. Inversely, a degree increase in the coldest night (warming of the coldest night) and a 1% increase in cold nights (night cooling) frequency could have caused a reduction of 28.6 and 56.2 kg/ha, respectively, in cowpea yield. The findings were consistent with Panda et al. (2014) that attested that the extreme state of day and night temperatures adversely impact major crops through changes in phenological development and physiological processes. Night and day cooling is thus unfavourable to cowpea production in the Sudanian and Sahelian zone of Burkina Faso. Despite the increased trend of hot climate extremes experienced within each climatic zone, these extremes did not negatively influence cowpea production. This result did not support the claim that hot temperatures during flowering can reduce the potential number of seeds or grains yield and stem growth (Wheeler et al. 2000). The drought-tolerant nature of cowpea varieties (Agossou et al. 2020) may be an explanation.

Generally, the study found that the extreme climate indices explained almost one-quarter to half of the variability in crop yields (maize, groundnut, cowpea, sorghum and millet) across the study zones. In all the zones, the adverse impacts of climate extremes were more observed than their positive influences on crop yields. In Sudanian and Sudano-Sahelian zones, the average temperature and rainfall over the study period (1984–2020) were in the range required for crop's optimal growth (27–30 °C and 400–1,200 mm/y) (Du Plessis 2008; Hatfield et al. 2011; Agossou et al. 2020). Nevertheless, yields were significantly affected by changes in extreme climate conditions. This means that climate extremes (cold days and nights, warm nights, coldest days and tropical nights) more than changes in the average climate conditions are responsible for observed variation of crop yield in the study zones (Vogel et al. 2019).

Moreover, in the Sahelian zone, despite the persistence in some years (after 2005) of average daily temperature (Supplementary Appendix 2) beyond the required maximum temperature (30 °C) for the majority of crop growth (Du Plessis 2008; Agossou et al. 2020), no significant interrelation was observed with some of the studied crops. This further supports that climate extremes (hottest days and nights, frequency of cold nights) are more responsible for crop yield variations. Furthermore, as indicated by Vogel et al. (2019), our findings revealed that temperature-related extremes are more responsible for yield anomalies than precipitation-related extremes.

Changes in extreme climate conditions have explained between 21–47 and 29–50% of the variation in crop yields within Sudanian and Sahelian zones, respectively, and about 30–45% of the variation in crop yields within the Sudano-Sahelian zone. Consequently, this underlines the possible important role played by non-climatic factors in crop yield determination. Therefore, about 53–79 and 50–71% of crop yield variation in Sudanian and Sahelian zones, respectively, then about 55–70% of yields variation in the Sudano-Sahelian zone may be attributed to non-climatic factors such as improved seed, fertility of the soils and farming methods, among others (Atiah et al. 2021). These authors suggested in their research a need to study the impacts of non-climatic factors on maize yield to maximise its production in Ghana. This suggestion is paramount for Burkina Faso and SSA agriculture sector as well. Therefore, more efforts at policies and research levels must be done to discriminate between climatic and non-climatic impacts, thus allowing a better understanding of non-climatic factor's impact on yields, including soil fertility, seeds and farmers' practices.

The observed variations in crop yields attributable to extreme climate indices are lower than the value reported at a global scale by Vogel et al. (2019). According to these authors, climate extremes indices account for more than half of the explained variances of yield anomalies (maize, rice, soybeans) and nearly half of spring wheat at the global scale. The findings may be attributed to differences in crop types and climate conditions of the respective studied zones.

The current study found evidence of change and variability in rainfall and temperature patterns in the study areas. However, the change is much more pronounced in temperature than in rainfall, at all scales of assessment. Moreover, the study revealed that hot extremes indices are evolving much faster than their counterpart cold extremes within the study zones, witnessing a warming climate across the three climatic zones studied. Decadal climate indices analysis showed that the last three decades in each climatic zone distinguished themselves by hot climate extremes from the first three decades that were generally characterised by cold climate extremes. Likewise, recent decades of the assessed period (1961–2020) were wetter. The warming climate and rainfall variability experienced within the study zones pose more stress for crops productions, thereby threatening the livelihoods of farming households in such zones. The induced threats led to crop yield reduction. The Sudano-Sahelian and Sahelian zones, where CC was more pronounced, are likewise more exposed to such threats. Across all three zones, crop yields were significantly associated with some climate indices. It can be deduced that climate variability and CC highlighted through indices trend in the current study is affecting the climate-dependent agricultural sector in Burkina Faso and could affect farmers' livelihoods. The major crops (maize, groundnut, cowpea, sorghum, millet) produced within the study zones responded differently (positively and negatively) to climate extremes indices. They were found to be more sensitive to these extremes than the average climate conditions.

A climate-smart policy option consisting of monitoring and/or addressing climate impacts could harness certain indices' positive influence (yield increase) on crops or mitigate adverse effects (yield decline) of other indices. Mitigation could consist of breeding more resilient crops and adopting a crop-livestock integration practices to enhance the livelihood of smallholder farmers in Burkina Faso.

We are grateful to the German Federal Ministry of Education and Research (BMBF), which funded this work through the West Africa Science Centre of Climate Change and Adapted Land Use (WASCAL). We also appreciate the Meteorological Agency and the Ministry of Agriculture of Burkina Faso for providing the data.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.