Impact of climate variability on the flowering and cherry filling in robusta coffee (Coffea canephora Pierre) in Yangambi, Democratic Republic of the Congo

Introduction

The coffee tree is one of the main cash crops in many tropical regions, where it plays a significant role in the income of small producers and international trade (Bunn et al., 2015) [6] . Among the cultivated species, Coffea canephora (robusta) holds an important place in hot and humid low and mid-altitude areas, due to its relative tolerance to high temperatures and its high productivity in environments where Arabica is less suitable (DaMatta & Ramalho, 2006; DaMatta et al., 2007; DaMatta, 2019) [13,14,15,16] . In Democratic Republic of the Congo, particularly in central Congolese basin, robusta represents a potential pillar for coffee revival, but this ambition faces recent significant climate change. Climate change is already evident through rising temperatures, altered rainfall patterns, and an increase in extreme events, with notable impacts on perennial crops (IPCC, 2021) [17] . Contrary to its image as a 'robust' plant, C. canephora exhibits thermal and water thresholds beyond which its physiology, flowering, and fruiting are disrupted (DaMatta, 2019; Jaramillo et al., 2011; Kath et al., 2020) [13,15,18,22] . In humid tropical areas, where there is more or less marked variation between dry and wet periods, this intra-annual variability in rainfall and temperatures can alter the synchronization of flowering, the load of flowers and fruits, as well as the final seeds quality (Bunn et al., 2015; Magrach et al., 2020) [6,23] . The flowering phenology of the coffee plant, particularly frequency, intensity, and flowering time, plays a crucial role in harvests regularity and coffee produced quality (DaMatta, 2019) [13,15] . Coffee flowering is often triggered by specific climatic sequences, such as a short period of water deficit followed by the return of rains, which synchronize flowering and influence the uniformity of cherry ripening (DaMatta et al., 2019; Magrach et al., 2020) [13,15,23] . When these sequences are disrupted (prolonged drought episodes, irregular or shifted rainfall, temperature anomalies), flowering can become spread out, the fruit set less effective, and maturation more heterogeneous, which results in potential yield losses and increased harvest and post-harvest constraints (Bunn et al., 2015; DaMatta et al., 2019) [6,13,15] . Beyond the triggering of flowering, ovary filling rate is also sensitive to the water and thermal conditions encountered during the reproductive cycle. Rainfall deficits at the beginning of cherry development or abnormally high temperatures can reduce the ovary filling rate, increase the proportion of malformed or aborted fruits, and degrade the physiological quality of the seeds (DaMatta, 2019; DaMatta et al., 2019) [13,15] . Despite the agronomic importance of these processes, the quantitative relationships between climate change, flowering dynamics, and ovary filling rate remain insufficiently documented for robusta in a humid tropical rainforest climate of the central Congo basin at Yangambi.

Material and Methods

Study environment

This study was conducted in Yangambi, a historic Robusta coffee-growing region, at the National Coffee Research Program site in northern Yangambi, about 100 km west from Kisangani (Tshopo Province), on the right of the Congo River in Democratic Republic of the Congo. The coordinates, recorded with a Garmin 65s GPS (global positioning system), were 00° 51'00.0'' N latitude, 024° 27'49.9'' E longitude, and an altitude of 458 meters. These details helped to create the map shown in figure 1.

Materials and method

Direct observation used as the primary method, with a probabilistic sampling approach based on continuous monitoring of flowering, its frequency and intensity, and mixed Robusta Lula coffee tree cherry filling rate over a period of three years (2023-2025). Lula mixed coffee tree, which included more than 263 genotypes from the National Coffee Research Program collection at Yangambi.

Collection of climatic parameters

Temperature, precipitation, and relative humidity are the main climatic parameters studied. Data collected at INERA climatological station in Yangambi, at PK 5, from daily observations. Precipitation was recorded each day and logged in a register, indicating the amount of rainfall (mm), as well as maximum and minimum temperatures and relative humidity. Cumulative totals and monthly and annual averages were calculated. These data cover the period from 2023 to 2025.

Selection of trees and sampling

Among 509 mixed Lula coffee trees recorded in the coffee program collection, 493 were selected after excluding those affected by various constraints (tight spacing, diseases, stunted plants, excessive shading) in order to better represent the potential of the material. In total, 105 trees randomly chosen as observation units, representing 21% of the population. These trees come from a coffee plantation established approximately 6 to 8 years ago.

Each selected tree received a unique code (tree number) to enable ongoing monitoring over three years. On each tree, four branches were chosen at three positions within the canopy (bottom, middle, top) and permanently marked with plastic tags to facilitate flowering events observations.

Followed by flowering and fruiting

Regular monitoring carried out on the selected trees throughout the production cycles. Observations conducted weekly, with increased monitoring during critical periods (the end of the dry season, the start of the rains, and phases of mass flowering). Observations concerned following variables: flowering, its intensity, its frequency, and cherry filling rate. For flowering, the aim was to record the presence or absence of flowers on the trees and to count them on each sampled node (to determine the female fertility category F) throughout the year (each month). Flowering frequency determined by counting number of flowering events per month throughout each year. [2] . This correspondence remains a qualitative approximation intended to link visual assessment to the phenological sequence coded by BBCH, without claiming strict metric equivalence. Cherry filling rate involved counting young fruits number after fertilization and before fruit ripening (before maturity) to classify them by female fertility. Fruits categorized based on the cherry filling rate, natural (free) fertilization was observed over the months. For this purpose, 100 cherries were randomly collected using cross-section at different canopy levels (top, middle, bottom) during months' observation.

Fig 2:

Illustrates the morphological type of cherries as defined by Reffye (1974) [27] .

Figure 2. Morphological types of coffee cherries (Reffye, 1974) [27] . Categories: A, cherry with two normal seeds; B, cherry with one empty locule and one normal seed (late endosperm abortion); C, cherry with one curved (caracol) seed and one empty scale (early ovule abortion); D, cherry with one empty locule and one scale; E, cherry with two empty locules (late abortion); F, ovary not fused or drying out within two months after fertilization.

Statistical analysis

The variables studied include flowering (the presence or absence of flowers on the trees), the frequency and intensity of flowering, and the filling rate of the cavities.

Results

Climatic parameters

Results related to climatic parameters are presented in Figures 3 and4 below: The flowering (fig. 5) occurs almost continuously throughout the year, regardless of the period or year analyzed, with a majority of trees in bloom, often up to 100%. In 2023, all the trees flowered from January to March (100%). This rate decreased slightly in April (96.2%) and May (86.7%), then rises again between June and July (from 93.3% to 95.2%).

A marked drop appears in August (15.2%), followed by a return to 100% from September to December. The same dynamic observed at the beginning of 2024, with 100% in January and February, then 99.0% in March. A significant decline observed in April (41.0%), followed by a gradual increase in May

Annual variations in flowering frequency

The results related to flowering frequency are presented in Figure 6. The regression model based on precipitation with a two-month lag (Precipitation_lag2) shows a significant positive effect (β = 0.0034; p = 0.046; R² = 0.118). An increase of 100 mm in precipitation corresponds to an average rise of about 0.34 units in flowering intensity.

Monthly and annual variations in the ovary cherry filling rates.

Figure 8 below presents the occupancy rates of the compartments based on monthly and annual flowering periods. . Figure 2. Morphological types of coffee cherries (Reffye, 1974) [27] . Categories: A, cherry with two normal seeds; B, cherry with one empty locule and one normal seed (late endosperm abortion); C, cherry with one curved (caracol) seed and one empty scale (early ovule abortion); D, cherry with one empty locule and one scale; E, cherry with two empty locules (late abortion); F, ovary not fused or drying out within two months after fertilization. A dominance of fruits with two normal seeds (category A) was observed throughout much of the study period, with monthly proportions ranging from 0% to 100%. These filling rates were often higher at the beginning and end of each year (ranging from 71% to 100%). Furthermore, the other categories (B to F), associated with varying levels of abortion or malformation, showed significant variability over the months. This indicates that the flowering periods in January and February consistently exhibit the best performance, with high proportions of normal fruits. Conversely, the months from July to October correspond to critical phases, characterized by a sharp decline in well-formed fruits and an increase in anomalies. Statistical models indicated a positive effect of temperature on the proportion of category A (β = 0.230; p = 0.012), as well as a marginal effect of precipitation. For category C, the maximum temperature has a significant negative effect (β = -0.084; p = 0.020). Similarly, for category E, the maximum temperature exerts a significant negative effect (β = -0.057; p = 0.003; R² = 0.246), with no notable effect from precipitation. These results show that moderate thermal conditions, combined with sufficient water availability, promote normal seed development, whereas more restrictive conditions are associated with increased anomalies. The observed variability, however, remains linked to multiple factors beyond climatic conditions.

Climate determinism on floral variables

Figure 9 illustrates the climatic determinism on the floral variables of the coffee plant. There is a marked negative linear relationship between periods of heavy rainfall and high humidity and the onset of coffee flowering. This indicates that the initiation of flowering and the differentiation of buds in the coffee plant require a prior phase of water stress (dry season). However, an excess of water during flowering causes mechanical abscission of the flowers, impairs pollen viability, and prevents the expression of a widespread and intense flowering, which explains the diametrically opposite positioning of these vectors on the graph (fig. 9).

Regarding temperature, an increase in minimum temperatures has led to a decrease in the frequency and floral load of coffee plants. Meanwhile, variations in daily maximum temperatures have exerted no direct linear influence (positive or negative) on the success gradient or the intensity of flowering.

Although the plant's metabolism depends on heat, the opening of the fully developed coffee flower is modulated by the plant's internal water potential and internal hormonal stimuli rather than by daily temperature peaks.

The climatic determinism on the cherries filling rate The studied categories are not randomly distributed with respect to climatic parameters. Dimension 1 can be interpreted as a heat index, while Dimension 2 relates to a water supply index.

Categories A and E constitute the two most discriminating climatic signatures of the model, whereas category D appears to be 'temporally or spatially independent' of the direct meteorological fluctuations captured by the sensors.

Discussion

The climatic conditions observed in Yangambi between 2023 and 2025 are part of a warming trend already documented in the region. High temperatures, peak in 2024, confirm the gradual increase in the average annual temperature and the increased frequency of hot days and nights (Kasongo et al., 2023; 2024) [20] . The progressive decrease in relative humidity aligns with long-term trends associated with rising temperatures (Kasongo, 2020) [19,21] . Precipitation remains abundant, but its distribution becomes more contrasting, with a strengthening of seasonality characterized by intense rainy episodes and more pronounced dry periods (Kasongo et al., 2023; Rainiaux et al., 2018) [26] . Under these conditions, coffee flowering appears continuous throughout the year, with 72.2% of months showing 100% of trees in bloom. This pattern differs from traditional coffee systems, in which flowering is seasonal and synchronized following a period of water deficit (Trottet, 2021). The absence of a marked floral rest period at Yangambi is linked to nearly constant rainfall, which limits the development of sufficient water stress to structure the reproductive cycle. This mechanism aligns with the work of Alvim (1960) [1] and Crisosto et al. (1985), who show that flowering depends on an alternation between water deficit and rehydration rather than cumulative rainfall. The frequency of flowering highlights a multimodal dynamic, with a main peak at the beginning of the year and a more diffuse second peak at the end of the year. This organization is consistent with observations by DaMatta et al. (2007) [16] , who relate flowering to variations in the plant's water status. The irregularity observed, especially in the latter part of the year, resembles the multiple flowering events sdescribed by Camargo (2010) [7] in unstable climatic contexts. The low explanatory power of the rainfall-based model (R² = 0.089) confirms that the frequency of flowering episodes depends more on the succession of dry and wet phases than on monthly rainfall amounts.

The intensity of flowering follows a similar pattern, with high values at the start of the year and lower values mid-year. The role of water in preparing floral buds is well established (Parada-Molina et al., 2025) [24] , while the combination of water deficit and rehydration remains crucial for abundant flowering (Alvim, 1960; Schuch et al., 1992) [1,28] . The variations observed in the endof the year may be linked to competition for assimilation, with a heavy fruit load limiting the expression of secondary flowering (Vaast et al., 2005) [30] . The filling of the coffee cherries shows high variability, with good performance at the beginning of the year and marked deterioration between July and October. This period corresponds to an increase in anomalies (caracoli grains, empty cherries, abortions), already associated with environmental stresses and physiological imbalances ( [3,9,25] . The limitation of carbon resources during periods of high reproductive load also contributes to these abortions (Vaast et al., 2005) [30] . Principal component analysis highlights a clear opposition between hydric variables (rainfall, humidity) and flowering parameters. Continuous humidity is associated with reduced floral expression, consistent with Borchert et al. (2004) [5] . The rise in minimum temperatures, particularly at night, is associated with decreased floral activity due to increased respiration at the expense of carbon reserves (Craufurd & Wheeler, 2009) [11] . Conversely, maximum temperatures do not directly affect the observed ranges unless critical thresholds are reached (Cleland et al., 2007) [10] . These results confirm that coffee flowering depends on a balance between hydric and thermal conditions, and that its desynchronization at Yangambi occurs within an increasingly variable climate context, in line with classical phenological models (Camargo, 2010; Carr, 2001) [7,8] . The correlation between humidity and failure of fruit set (Category F) can be interpreted through Cannell's (1985) "Source-Sink" theory, which states that very high relative humidity (saturation) reduces vapor pressure deficit (VPD), thereby blocking leaf transpiration and the overall transpiratory flux of the coffee plant. Without this mechanism, the transport of sap and nutrients to young fruits (which act as primary metabolic "sinks") is disrupted, causing their early physiological abortion.

Conclusion

This study highlights the high sensitivity of the Robusta coffee tree to microclimatic fluctuations in Yangambi. The recorded results indicate that the water regime is the primary structuring factor for flowering, while rising minimum temperatures and high humidity limit its expression. Climatic disturbances are accompanied by increased anomalies in fruit development.

In this context, the adaptation of production systems relies on strategies to mitigate climatic constraints, notably through agroforestry and the selection of genotypes tolerant to thermohydric stresses.

Acknowledgement

AcknowledgementsThe authors express their deep gratitude to the Management of the National Institute for Study and Agronomic Research, Yangambi Research Center (INERA-CRY), for granting access to the coffee concessions and to the station's climatic data. Special thanks the Botanical Garden of Meise and CIFOR (Center for International Forestry Research) for their crucial scientific, material, and financial support, which enabled to conduct field monitoring and complete this work in the central Congo Basin.

References

1. Alvim PT. Water stress as a necessary condition for the flowering of the coffee plant. Science. 1960; 132(3423):354. https://DOI.org/10.1126/science.132.3423.354
2. Arcila-Pulgarín J, Buhr L, Steffens JC, Martínez JA, Rubiano Y. Application of the extended BBCH scale to the description of coffee plant (Coffea spp.) growth stages. 2002. https://DOI.org/10.1080/14620316.2002.11511459
3. Barbosa J, de Castro EM, de Souza K, Thiebaut JTL. Anatomical and physiological plasticity of coffee leaves cultivated under different shaded environments. 2015. https://DOI.org/10.5539/jas.v7n10p184
4. Berthaud J, Charrier A. Genetic resources of coffee. In: Charrier A, Eskes S, Mossu M, Désamoré M, Jasmin J, editors. Coffee: Agronomy. CAB International; 1988. p. 142. https://DOI.org/10.1079/9780851985796.0001
5. Borchert R, Meyer SA, Felger RS, Porter-Bolland L. Environmental control of flowering periodicity in the dry tropical forests of Costa Rica and Mexico. Global Ecol Biogeogr. 2004; 13(5):409-425. https://DOI.org/10.1111/j.1466-822X.2004.00111.x
6. Bunn C, Läderach P, Rivera OO, Kirschke D. A bitter cut: climate change profiles on global Arabica and Robusta coffee production. Clim Change. 2015; 129:89-101. https://DOI.org/10.1007/s10584-014-1306-x
7. Camargo MBP. Impact of climate variability and climate change on Arabica coffee cultivation in Brazil. Bragantia. 2010; 69(1):239-247. https://DOI.org/10.1590/S000687052010000100030
8. Carr MKV. Water relations and irrigation needs of the coffee plant. Exp Agric. 2001;37(1):1-36. https://DOI.org/10.1017/S001447970100101X
9. Carvalho A, Monaco LC. Genetics of coffee. In: Ferwerda FP, Wit F, editors. Insights into the selection of perennial crops in the tropics. Wageningen: H. Veenman & Zonen; 1969. p. 199-221.
10. Cleland EE, Chuine I, Menzel A, Mooney HA, Schwartz MD. Phenological changes in plants in response to global changes. Trends Ecol Evol. 2007;22(7):357-365. https://DOI.org/10.1016/j.tree.2007.04.003
11. Craufurd P, Wheeler T. Climate change and flowering periods of annual crops. J Exp Bot. 2009. https://DOI.org/10.1093/jxb/erp196
12. Crisosto CH, Grantz DA, Meinzer FC. Effects of water deficit on flowering of coffee plants (Coffea arabica L.). Tree Physiol. 1992; 10(2):127-133. https://DOI.org/10.1093/treephys/10.2.127
13. DaMatta F. Ecophysiology of coffee growth and production in the context of climate change. 2019. https://DOI.org/10.1016/bs.agron.2019.10.001
14. DaMatta F, Ramalho JD. Impacts of water and heat stress on coffee physiology and production. 2006. https://DOI.org/10.1590/S1677-04202006000100006
15. DaMatta F, Rahn E, Läderach P, Ghini R, Ramalho JDC. Why might coffee cultivation be better able to withstand climate change and global warming than previously estimated? 2019. https://DOI.org/10.1007/s10584-0182346-4
16. DaMatta F, Ronchi CP, Maestri M, Barros RS. Ecophysiology of coffee growth and production. 2007. https://DOI.org/10.1590/S1677-04202007000400014
17. IPCC. Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the IPCC. Cambridge: Cambridge University Press; 2021. https://DOI.org/10.1017/9781009157896
18. Jaramillo J, Muchugu E, Vega FE, Davis A, Borgemeister C, Chabi-Olaye A. Some like it hot: influence and consequences of climate change on the coffee berry borer (Hypothenemus hampei) and coffee production in East Africa. PLoS ONE. 2011. https://DOI.org/10.1371/journal.pone.0024528
19. Kasongo YE, Hubau W, Ekila D, Kalundu EB, Mukendi J, Matabaro J, Syampungani S, et al. Ground climate data show evidence of warming and intensification of the seasonal rainfall cycle during the period 1960-2020 in Yangambi, in the central Congo basin. Clim Change. 2023; 176:144. https://DOI.org/10.1007/s10584-023-03606-0
20. Kasongo YE, Hubau W, Kalundu EB, Ekila D, Mukendi J, Matabaro J, Syampungani S, et al. Climate change is real in the Congo Basin, reveals a study on the Yangambi Biosphere Reserve. Greenafia -Environment and Climate in Central Africa. 2024. https://greenafia.com/le-changementclimatique-est-reel-dans-le-bassin-du-congo-revele-uneetude-sur-la-reserve-de-la-biosphere/
21. Kasongo YE. Recent climate trend in the Yangambi region, DRC. 2020. https://DOI.org/10.32474/AOMB.2020.03.000530
22. Kath J, Craparo A, Frojmovic D, Phan A, Brauman K, Das I, Garibaldi LA, Machado Dias, et al. Impacts of climate change on coffee. Clim Change. 2020. https://DOI.org/10.1007/s10584-020-02715-z
23. Magrach A, Ghazoul J, Koh L, Rice R, Bowles T, Perfecto I. Climate and farming practices jointly influence flowering, productivity, and yield variability of coffee plants. Agric For Meteorol. 2020. https://DOI.org/10.1016/j.agrformet.2020.108066
24. Parada-Molina J, Cerdán-Cabrera CR, Gálvez-López D, Chacón-Hernández SS, López-Cano A, Hernández-Ramos, et al. Impact of climate on water status, growth, yield, and phenology of coffee plants in the central region of Veracruz, Mexico. PLoS ONE. 2025; 0319670. https://DOI.org/10.1371/journal.pone.0319670
25. Pezzopane JRM, Pedro Júnior MJ, Thomaziello RA, Camargo MBP. Phenological scale for Arabica coffee cultivation. Bragantia. 2007. https://DOI.org/10.1590/S0006-87052007000300006
26. Rainiaux H, Moïse JF, Cogho M, Ndongala P, Mbala G, Mpunga B, Kalonda B. Climatic data and analyses of the Congo Basin (rainfall databases and trends). 2018. http://www.appsolutelydigital.com/CongoBasin/chapter12.html
27. Reffye P. Control of flowering and its anomalies in Coffea arabica, Robusta, and their Arabusta hybrids. Café Cacao Thé (Paris). 1974; 18(4):237-254.
28. Schuch UK, Fuchigami LH, Nagao MA. Flowering, ethylene production, and bud abscission induced by ethephon in coffee plants. J Am Soc Hortic Sci. 1992; 117(3):462-466. https://DOI.org/10.21273/JASHS.117.3.462
29. Trottet. La floraison des caféiers, l'indispensable. Cafés Trottet Blog. 2021. https://www.cafes-trottet.fr/blog/lafloraison-des-cafeiers-lindispensable/
30. Vaast P, Angrand J, Franck N, Dauzat J, Génard M. The fruit load and ring-barking of branches influence the distribution of carbon and the photosynthesis of fruits and leaves in Coffea arabica. Tree Physiol. 2005; 25(6):753760. https://DOI.org/10.1093/treephys/25.6.753