Physiological and Biochemical Analysis of Coffea arabica Cultivars in the Early Stage of Development Subjected to Water Stress for the Selection of Cultivars Adapted to Drought

Abstract

Drought events intensified by climate change severely compromise the physiological stability and productivity of Coffea arabica, particularly in rainfed systems, underscoring the need to identify cultivars with greater functional resilience. This study evaluated the physiological, nutritional and biochemical responses of seedlings from ten cultivars subjected to adequate irrigation (AW), severe water deficit (SWD) and rehydration (RI). Water potential, gas exchange, oxidative stress markers, stomatal traits and foliar macro- and micronutrients were quantified. Most cultivars exhibited pronounced reductions in the pre-dawn leaf water potential (Ψ_pd), photosynthesis (A), stomatal conductance (g_s) and transpiration (E), together with increases in oxidative stress indicators under SWD. In contrast, Obatá amarillo, Castillo, and Arará maintained greater hydraulic stability, more efficient stomatal regulation, higher water-use efficiency, and lower oxidative stress, accompanied by a more effective post-stress recovery after RI. Regarding nutrient dynamics, Geisha, Castillo, and Arará showed higher K⁺ accumulation, while Catimor bolo presented elevated Ca²⁺, P, and Fe²⁺ contents, elements associated with metabolic reactivation and structural recovery after stress. Geisha and Marsellesa displayed an adaptive, recovery-driven resilience strategy following drought stress. Overall, the findings identify Obatá amarillo, Castillo, and Arará as the most drought-tolerant cultivars, highlighting their potential relevance for breeding programs aimed at improving drought resilience in coffee.

1. Introduction

Coffee (Coffea arabica L.) is one of the most important crops worldwide, not only because of its economic relevance in generating foreign exchange and jobs, but also because of its social and cultural impact in producing regions, mainly in tropical and subtropical countries. Factors such as population growth and the expansion of urban areas suggest future production increases [1,2].

In the field of world coffee growing, Peru remains one of the main producers and exporters worldwide, surpassed only by Brazil, Colombia, Ethiopia and Honduras (Ministry of Agrarian Development and Irrigation [3]. However, during 2024, national production (3986 million 60 kg bags) was 9.65% lower compared to the previous year due to climate variability and the increase in infestations and pest distribution [4]. In addition to this concern, recent climate change projections warn that the increase in temperature, the variability of rainfall, and increases in CO₂ generally linked to abiotic stress will cause direct and/or indirect effects on agriculture [5]. In Brazil, droughts and abrupt thermal increases have been reported to have caused significant losses in production, compromising seed filling and cause dehydration of berries [6]. Thus, for Latin America, predictions in coffee cultivation foresee a significant decrease in coffee production, yield, and quality [7].

In this context of climate change, droughts and elevated temperatures are recognized as the most limiting stressors for crop growth and productivity [8]. Stress due to water scarcity stands out by triggering a series of morphological, physiological and biochemical effects in plants [9]. Water deficit induces alterations in physiological parameters such as stomatal closure, decreased foliar water potential, reduced photosynthetic rate, and solute transport [10]. At the same time, at the biochemical level, there is an increase in oxidative stress caused by the overproduction of reactive oxygen species (ROS) [11]. Therefore, the selection of genotypes that best respond to these effects are considered essential genetic material for the genetic improvement of coffee plants for the continuity and profitability of the crop [12,13].

In this scenario, the identification and selection of coffee cultivars adaptable to drought conditions becomes a primary strategy. It has been shown that plants, in order to cope with water stress, develop a series of morphological, physiological and biochemical responses aimed at preserving their water status and maintaining vital functions [10,14]. The most evaluated tolerance mechanisms include hydraulic responses, gas exchange, and morphoanatomical structural variations [15]; while, at the biochemical level, the accumulation of antioxidant metabolites (proline and glutamine) and polyamines, in addition to the synthesis of plant proteins and hormones (abscisic acid) play a fundamental role in the tolerance to water deficit [16].

Therefore, the present study aimed to analyze the physiological and biochemical responses of ten coffee cultivars (Coffea arabica L.) in the initial stage of development under water deficit conditions, in order to identify genotypes with greater capacity to adapt to drought and provide criteria for the selection of tolerant cultivars in the face of climate change scenarios.

2. Results

2.1. Foliar Water Potential

The pre-dawn leaf water potential (Ψ_pd) in C. arabica seedlings showed marked responses to water deficit in all the cultivars evaluated (Figure 1). At the end of the AW period, the values ranged between −0.57 and −1.23 MPa, while at the end of SWD significant reductions were observed in most of the cultivars (p < 0.05) Parainema (−3.12 MPa) and Bourbon rosado (−3.12 MPa) and Catimor (−3.10 MPa) the ones that presented the steepest drops, in contrast, Castillo (−1.03 MPa) and Obatá amarillo (−1.17 MPa) stood out for exhibiting the lowest drops in Ψ_pd. After the RI, most of the cultivars recovered their initial values, with Parainema (−0.48 MPa), Bourbon rosado (−0.5 MPa), Obatá amarillo (−0.55 MPa) and Castillo (−0.65 MPa) showing values higher than those recorded at the end of AW.

2.2. Gas Exchange

The stomatal conductance to water vapor (g_s) showed a pattern strongly associated with the applied water regime (Figure 2). The evaluations at the end of the AW and SDW periods maintained low and relatively homogeneous values between cultivars (p > 0.05), with means ranging from 0.004 to 0.041 mol H₂O m⁻² s⁻¹ and pronounced reductions in Castillo (0.009 mol H₂O m⁻² s⁻¹) and Obatá amarillo (0.010 mol H₂O m⁻² s⁻¹). followed by Bourbon rosado and Arará. However, after RI, significant increases were observed in several cultivars, highlighting Catimor bolo, Geisha and Marsellesa, whose values reached 0.20, 0.23 and 0.14 mol H₂O m⁻² s⁻¹, respectively, while Villalobos, Parainema, Catimor and Obatá amarillo maintained values similar to those observed in AW and SWD.

The net photosynthesis rate (A) showed significant variations between cultivars and water periods (AW, SWD and RI) (p < 0.05) (Figure 3). Highest values were registered in Obata amarillo, Bourbon rosado, Catimor bolo, and Castillo under AW, while Geisha, Villalobos and Parainema showed the lowest rates. The SWD period drastically reduced A in all cultivars; however, Arará (3.01 μmol CO₂ m⁻² s⁻¹), Castillo (2.34 μmol CO₂ m⁻² s⁻¹) and Obatá amarillo (2.17 μmol CO₂ m⁻² s⁻¹) maintained moderate values. At the end of RI, all cultivars increased the values of A, highlighting Marsellesa, Geisha and Bourbon rosado as the ones with the highest recovery, whose values reached 9.6, 9.1 and 8.7 μmol CO₂ m⁻² s⁻¹, respectively. While Catimor, Arará, Obatá amarillo and Castillo showed intermediate recoveries.

Transpiration rates (E) varied markedly according to the cultivar and the water regime applied, reaching their maximums at the end of the recovery phase (RI), in clear contrast to the consistently low values observed in the evaluation of AW and SWD (Figure 4). At the end of RI, Geisha had the highest level (3.30 mmol H₂O m⁻² s⁻¹), followed by Catimor bolo (3.02 mmol H₂O m⁻² s⁻¹), Marsellesa (2.20 mmol H₂O m⁻² s⁻¹) and Castillo (1.76 mmol H₂O m⁻² s⁻¹). Both at the end of AW and in SWD, most cultivars maintained relatively homogeneous values (0.17–0.89 mmol H₂O m⁻² s⁻¹), in contrast to Obatá amarillo (0.18 mmol H₂O m⁻² s⁻¹) and Castillo (0.17 mmol H₂O m⁻² s⁻¹) which presented the lowest values of E.

Water use efficiency (WUE) showed stark contrasts between cultivars and water regimes (Figure 5). Under AW conditions, the highest value was recorded in Geisha (24.6 μmol CO₂/mmol H₂O m⁻² s⁻¹); however, under SWD the cultivars Geisha, Bourbon rosado, Parainema, and Catimor drastically reduced their WUE (0.87–1.10 μmol CO₂/mmol H₂O m⁻² s⁻¹). In contrast, Obatá amarillo (11.87 μmol CO₂/mmol H₂O m⁻² s⁻¹), Castillo (15.76 μmol CO₂/mmol H₂O m⁻² s⁻¹) and Arará (6.49 μmol CO₂/mmol H₂O m⁻² s⁻¹) increased their WUE during SWD. At the end of the RI period, Villalobos, Obatá amarillo, Parainema and Catimor exhibited the highest WUE values, even exceeding their levels recorded in AW and SWD.

2.3. Stomatal Density and Stomatal Index

Stomatal density (SD) at the end of the IR period showed differences between the cultivars evaluated (Figure 6). The highest SD was recorded in Marsellesa, Arará, and Geisha, followed by Parainema and Villalobos. On the other hand, the lowest values were recorded in Bourbon rosado, Obatá amarillo, and Castillo. Similarly, the stomatal index (SI), which expresses the proportion of stomata in relation to the total number of epidermal cells, also showed variability among cultivars (Figure 6), with Arará standing out, followed by Catimor, while Parainema, Obatá amarillo, and Castillo recorded the lowest SI values. Representative micrographs confirm the morphological differences observed in the organization and abundance of stomata among cultivars.

2.4. Elemental Análisis

After the recovery period (RI), marked differences were evident between cultivars in the foliar accumulation of nutrients (

Table 1. Nutrient content (μg g⁻¹ DW) in seedlings of coffee cultivars after the RI period.

Cultivars	P	Ca2+	Na+	K+	Fe2+	Mg2+	Zn2+	Cu2+	Mn2+
Catimor bolo	242.1 ± 89.82 a	9553.31 ± 195.59 a	784.69 ± 29.57 h	3943.7 ± 124.39 b	123 ± 2.5 a	1538.15 ± 106.11 a	8.57 ± 0.37 g	18.64 ± 0.79 c	476.4 ± 55.91 a
Geisha	1911.67 ± 27.48 c	7807.62 ± 126.86 b	1107.19 ± 22.81 a	4470.02 ± 221.8 a	79.28 ± 3.81 c	1365.31 ± 65.15 b	6.99 ± 0.32 g	12.91 ± 0.53 e	301.68 ± 17.33 d
Marsellesa	1772.7 ± 14.72 c	6608.18 ± 169.25 b	954.04 ± 9.13 e	3811.13 ± 146.81 b	87.28 ± 1.18 b	1257.34 ± 79.05 b	6.92 ± 0.38 g	24.93 ± 1.24 a	192 ± 5.73 f
Bourbon rosado	1930.49 ± 39.37 c	6423.03 ± 177.08 c	1037.9 ± 11.51 b	3693.67 ± 37.69 b	77.85 ± 2.48 c	1253.96 ± 77.22 b	14.01 ± 0.45 c	19.81 ± 0.58 b	254.37 ± 29.37 e
Villalobos	1817.27 ± 41.96 c	9621.61 ± 74.87 a	974.67 ± 23.42 d	3973.75 ± 294.8 a	59.53 ± 2.71 e	1550.7 ± 120.34 a	16.71 ± 0.52 a	15.63 ± 0.69 d	323.87 ± 20.12 c
Parainema	1726.61 ± 29.47 d	7361.06 ± 146.5 b	870.66 ± 17.38 f	3635.24 ± 87.97 b	27.63 ± 2.76 g	1050.92 ± 36.17 c	8.99 ± 0.54 f	18.18 ± 1.06 c	140.15 ± 13.91 g
Catimor	1952.17 ± 68.49 c	7381.77 ± 180.2 b	789.53 ± 24.8 h	2764.86 ± 124.17 d	35.72 ± 2.19 g	1274.51 ± 74.86 b	14.27 ± 0.61 b	9.27 ± 0.08 f	186.65 ± 16.58 f
Arará	1892.45 ± 36.95 c	8323.66 ± 186.51 b	850.58 ± 48.95 g	4267.62 ± 250.63 a	67.68 ± 1.29 d	1624.73 ± 71.95 a	7.89 ± 0.49 g	12.42 ± 0.62 e	351.84 ± 10.18 b
Obatá amarillo	1873.69 ± 22.57 c	6855.02 ± 173.09 b	1052.81 ± 36.55 b	3104.92 ± 126.8 c	55.43 ± 2.03 f	1340.2 ± 72.82 b	11.86 ± 0.41 d	8.96 ± 0.53 f	234.55 ± 26.38 e
Castillo	2004.06 ± 56.16 b	8206.39 ± 197.46 b	985.69 ± 16.79 c	4272.86 ± 67.16 a	67.74 ± 1.79 d	1556.19 ± 75.17 a	10.86 ± 0.33 e	19.54 ± 1.12 c	274.05 ± 22.89 d
p-valor	<0.001	0.0012	<0.001	<0.001	<0.001	0.0012	<0.001	<0.001	<0.001
CV (%)	10.1%	14.2%	12.2%	14.9%	38.5%	15.2%	31.3%	31.6%	36.7%

The letters indicate significant differences between cultivars according to mean comparisons, according to Tukey HSD test.

). Potassium (K⁺) has its highest concentrations in Geisha (4470.02 ± 221.8 μg g⁻¹ DW), followed by Arará (1624.73 ± 71.95 μg g⁻¹ DW) and Castillo (1556.19 ± 75.17 μg g⁻¹ DW). For calcium (Ca²⁺) Villalobos (9621.61 ± 74.87 μg g⁻¹ DW) and Catimor bolo (9553.31 ± 195.59 μg g⁻¹ DW), showed the highest accumulations, while Bourbon rosé recorded the lowest accumulation.

Regarding phosphorus (P), Catimor had the highest concentration (2412.1 ± 89.82 μg g⁻¹ DW) followed by Castillo, while Parainema showed the lowest levels. In the case of magnesium (Mg²⁺), Castillo (1556.19 ± 75.17 μg g⁻¹ DW), Arará (1624.73 ± 71.95 μg g⁻¹ DW), Villalobos and Catimor stood out for their high contents. Finally, for iron (Fe²⁺), Catimor bolo exhibited the highest accumulation among the cultivars evaluated.

2.5. Proline Content

The standardization of proline by z-score values showed definite variations between the periods evaluated (Figure 7). In the SWD period, most of the cultivars evaluated presented an above-average proline accumulation (z > 0), standing out with increases in Castilla, Obatá amarillo, Catimor, Bourbon rosado, with the exception of Marsellesa, which remains stable compared to the AW period. In contrast, under RI, a generalized decrease in standardized values (z < 0) was observed in most cultivars, while Marsellesa presented a positive value in this period (z > 0), thus showing a late accumulation during the recovery period.

2.6. Lipid Peroxidation and Hydrogen Peroxide

The MDA, an indicator of the degree of lipid peroxidation, varied between cultivars and water regimes (

Table 2. MDA/H₂O₂ content in seedling leaves of ten cultivars of C. arabica evaluated at the end of three water periods (AW, SDW, and RI).

Cultivars	Period	MDA (nmol g−1 FW)	H2O2 (nmol g−1 FW)
Catimor bolo	AW	4.060 ± 0.344 d–j	313.432 ± 16.578 bcd
	SWD	5.215 ± 0.224 b–g	466.803 ± 39.970 ab
	RI	6.718 ± 0.357 b	416.231 ± 25.459 ab
Geisha	AW	3.504 ± 0.344 f–j	280.419 ± 16.578 bcd
	SWD	3.929 ± 0.224 e–j	252.493 ± 39.970 bcd
	RI	5.844 ± 0.357 b–d	346.224 ± 25.459 bcd
Marsellesa	AW	2.921 ± 0.344 j	233.988 ± 16.578 cd
	SWD	4.716 ± 0.224 c–i	411.526 ± 39.970 abc
	RI	5.437 ± 0.357 b–f	342.569 ± 25.459 bcd
Bourbon rosado	AW	2.346 ± 0.344 j	301.032 ± 16.578 bcd
	SWD	4.688 ± 0.224 c–i	415.834 ± 39.970 abc
	RI	5.567 ± 0.357 b–e	330.352 ± 25.459 bcd
Villalobos	AW	3.429 ± 0.344 f–j	294.902 ± 16.578 bcd
	SWD	4.903 ± 0.224 c–h	425.724 ± 39.970 ab
	RI	6.306 ± 0.357 bc	360.469 ± 25.459 bc
Parainema	AW	2.652 ± 0.344 j	255.278 ± 16.578 bcd
	SWD	5.576 ± 0.224 b–d	468.444 ± 39.970 ab
	RI	5.416 ± 0.357 b–g	376.148 ± 25.459 bc
Catimor	AW	3.386 ± 0.344 g–j	282.47 ± 16.5780 bcd
	SWD	9.492 ± 0.224 a	643.006 ± 39.970 a
	RI	4.197 ± 0.357 d–j	316.155 ± 25.459 bcd
Arará	AW	2.455 ± 0.344 j	350.954 ± 16.578 bcd
	SWD	3.354 ± 0.224 h–j	439.803 ± 39.970 ab
	RI	3.611 ± 0.357 e–j	287.908 ± 25.459 bcd
Obatá amarillo	AW	2.292 ± 0.344 j	204.316 ± 16.578 d
	SWD	4.034 ± 0.224 e–j	311.193 ± 39.970 bcd
	RI	3.044 ± 0.357 ij	228.989 ± 25.459 cd
Castillo	AW	2.504 ± 0.344 j	208.516 ± 16.578 d
	SWD	3.361 ± 0.224 h–j	360.013 ± 39.970 bcd
	RI	3.179 ± 0.357 h–j	193.829 ± 25.459 d
CV %		13.85%	8.18%

The letters indicate significant differences between the cultivar combinations × period according to comparisons of marginal means adjusted (emmeans) using Sidak (α = 0.05). The analysis was performed using a mixed heteroscedastic model (varIdent).

). Under AW, values remained low, between 2.3 and 4.1 nmol g⁻¹ FW. Under SWD conditions, most cultivars showed marked increases, with Catimor recording the highest concentration (≈9.5 nmol g⁻¹ FW), followed by Parainema, Catimor bolo, Villalobos, and Marsellesa. In contrast, Obatá amarillo, Arará, and Castillo had the lowest values, ranging between 3.4 and 4.0 nmol g⁻¹ FW. During RI, Catimor, Castillo, and Obatá amarillo showed lower concentrations compared to SWD, while Catimor bolo and Villalobos maintained relatively high levels.

H₂O₂ concentration also showed differences between cultivars and water conditions (Table 2). In AW, values ranged from ≈200 to 351 nmol g⁻¹ FW. Under SWD, increases were observed in most cultivars, with Catimor again reaching the highest concentration (≈643 nmol g⁻¹ FW), while Geisha, Obatá amarillo, and Arará had lower concentrations. During RI, some cultivars maintained relatively high levels, such as Catimor bolo, Parainema, and Villalobos (≈360–416 nmol g⁻¹ FW), while others, particularly Castillo and Obatá amarillo, had lower values.

2.7. Principal Component Analysis

Principal component analysis (PCA) reveals defined physiological patterns along the imposed water gradient (Figure 8). In AW, samples are grouped in the upper left quadrant (Figure 8A), oriented in the direction of A, E, g_s and WUE, indicating a functional state characterized by high photosynthetic efficiency, active transpiration and wide stomatal opening under adequate water availability. With the transition to SWD, seedlings show a marked shift towards the right and lower quadrants, as opposed to the vectors of A, E, and g_s (Figure 8A). This pattern denotes progressive stomatal closure and reduced gas exchange. At the same time, the orientation of the points towards the vectors of Proline, MDA and H₂O₂ reflects a greater accumulation of osmoprotectants and markers of oxidative damage.

At the end of RI, the samples migrate to the middle-right zone (Figure 8A), reflecting a partial recovery of photosynthetic and stomatal activity, especially in A and g_s, although their proximity to MDA and H₂O₂ suggests persistence of stress in some cultivars (Figure 8B). The biplot by cultivars (Figure 8B) reinforces these trends, showing that each cultivar occupies specific regions of the multivariate space, reflecting contrasting strategies against water stress.

2.8. Correlation Analysis

The correlation matrix reveals differential patterns between water periods (Figure 9). In AW, positive correlations between g_s, A, and E reflect a coordinated functioning of gas exchange under adequate hydration, while negative associations with MDA and H₂O₂ indicate minimal levels of oxidative stress.

During SWD, the correlations are remarkably reorganized: A, E and g_s maintain positive associations with each other, but show negative relationships with Ψ_pd, confirming the effect of the decrease in water potential on stomatal closure. In this period, Proline, MDA and H₂O₂ are positively correlated, evidencing the convergence between osmoregulation and oxidative stress as the main physiological responses under water deficit.

After rehydration (IR), the correlation pattern is extended to include nutritional variables assessed at the end of this period. Nutrients such as Fe²⁺, K⁺, Mg²⁺, and Zn²⁺ show positive associations with A, E, and WUE, suggesting that physiological recovery is accompanied by nutritional restructuring. Likewise, DS and SI present consistent correlations with post-stress physiological performance, indicating a potential role in resilience. However, the persistence of positive correlations between MDA, H₂O₂ and some micronutrients reflects that certain cultivars maintain residual signs of stress, despite water recovery.

3. Discussion

Water deficit is one of the most limiting abiotic factors affecting growth, development, and productivity of Coffea spp., as it directly disrupts plant water relations, carbon assimilation, and redox homeostasis [16,17]. In perennial crops such as coffee, drought tolerance cannot be attributed to a single trait but rather emerges from the coordinated regulation of hydraulic integrity, stomatal behavior, metabolic flexibility, and the capacity to recover physiological function after stress release, a feature increasingly recognized as central to plant resilience under climate variability [15,18,19]. Therefore, evaluating stress responses exclusively at the peak of drought may underestimate tolerance mechanisms that become evident during recovery phases, particularly in woody species [20].

In the present study, the imposition of severe water deficit (SWD) following the complete cessation of irrigation resulted in pre-dawn leaf water potentials (Ψ_pd) within ranges known to severely constrain hydraulic conductivity and leaf gas Exchange [16,21]. However, the comparatively stable Ψ_pd observed in Castillo and Obatá amarillo suggests a greater capacity to sustain leaf hydration through osmotic adjustment and/or enhanced root–soil water uptake, mechanisms previously reported as critical for drought tolerance in Coffea arabica and other woody crops [22,23,24].

Importantly, maintaining Ψ_pd under drought does not necessarily imply tolerance per se, as it may result from either effective physiological regulation or from growth inhibition that reduces water demand [25]. Thus, Ψ_pd must be interpreted in conjunction with gas exchange, water-use efficiency, and recovery performance, as done in the present study, to distinguish adaptive regulation from passive dehydration avoidance.

The coordinated reduction in net photosynthesis (A) and stomatal conductance (g_s) observed under SWD reflects a conservative stomatal regulation strategy, particularly evident in Obatá amarillo, Arará, and Castillo. Such coordinated downregulation is widely recognized as an adaptive response that limits transpirational water loss while preserving mesophyll integrity and photosynthetic machinery, rather than indicating irreversible metabolic damage [26,27,28]. Experimental evidence in coffee and other evergreen species shows that early stomatal closure under drought reduces oxidative pressure on chloroplasts by limiting excess excitation energy and ROS production [29,30,31].

The rapid recovery of A and g_s following rehydration further supports this interpretation. Recovery kinetics are increasingly regarded as a critical component of drought tolerance, as tolerant genotypes typically exhibit faster and more complete restoration of gas exchange after stress relief, reflecting preserved photosystem II function and metabolic readiness [11,20,27]. Thus, the observed recovery patterns indicate functional elasticity of the photosynthetic apparatus, rather than simple survival during drought.

The contrasting responses in transpiration (E) and instantaneous water-use efficiency (WUE) among cultivars under SWD reveal distinct drought-response strategies. Increases in WUE associated with reduced E, as observed in Castillo, Obatá amarillo, and Arará, reflect an optimization of carbon gain relative to water loss, a key determinant of drought adaptation in both annual and perennial species [27,32,33,34]. Importantly, WUE enhancement under drought is not driven solely by transient stomatal closure but is strongly influenced by intrinsic stomatal traits that define maximum transpiration capacity [35,36].

Consistent with this framework, cultivars exhibiting higher WUE also showed lower stomatal density and stomatal index. Reduced stomatal density has been shown to decrease leaf conductance to water vapor while maintaining sufficient CO₂ diffusion, particularly when accompanied by dynamic stomatal regulation, thereby improving whole-plant water-use efficiency under drought [14,35,37,38]. This coordination between stomatal anatomy and physiology is increasingly recognized as a structural basis of drought resilience, particularly in woody crops exposed to recurrent stress cycles [36,39].

Marked differences in foliar nutrient accumulation after rehydration indicate that nutrient dynamics are actively involved in post-drought recovery processes rather than representing residual stress effects. Elevated potassium (K⁺) levels in Geisha, Castillo, and Arará are particularly relevant, as K⁺ plays a central role in stomatal function, osmotic regulation, enzyme activation, and phloem transport [40,41]. Recent studies demonstrate that adequate K⁺ supply enhances photosynthetic recovery and accelerates the restoration of leaf turgor following drought, thereby improving resilience [36,42].

Calcium (Ca²⁺) accumulation in Villalobos and Catimor bolo likely reflects its dual structural and signaling functions during stress alleviation. Ca²⁺ stabilizes membrane phospholipids and cell walls while acting as a secondary messenger in stress-recovery signaling cascades, modulating antioxidant defenses and membrane repair processes [43,44]. Similarly, the concurrent accumulation of phosphorus (P) and iron (Fe²⁺) in Catimor bolo suggests enhanced capacity for metabolic reactivation, as P is essential for ATP-dependent processes and Fe is integral to electron transport chains and redox regulation within chloroplasts and mitochondria [26,37,45]. Elevated magnesium (Mg²⁺) levels further support the maintenance of chlorophyll integrity and Rubisco activation during recovery [46,47,48].

Proline accumulation under SWD reflects an important component of osmotic adjustment; however, its functional significance depends on the magnitude, timing, and reversibility of accumulation [49,50]. Beyond osmoprotection, proline acts as a stabilizer of proteins and membranes, a ROS scavenger, and a regulator of cellular redox balance, linking carbon and nitrogen metabolism under stress [49,51]. The differential accumulation observed among cultivars indicates distinct metabolic strategies rather than uniform stress severity, as excessive or persistent proline accumulation has been associated with stress injury rather than tolerance in some systems [52].

Controlled and reversible proline accumulation, followed by a decline during recovery, has been proposed as a hallmark of effective drought acclimation, allowing plants to buffer dehydration without compromising metabolic recovery [14,15,16,53].

Oxidative stress markers further elucidate cultivar-specific tolerance mechanisms. Lower malondialdehyde (MDA) levels under SWD and during recovery indicate reduced lipid peroxidation and enhanced membrane stability, reflecting effective control of oxidative damage [11,54]. MDA moderation is increasingly interpreted as an integrative outcome of coordinated antioxidant activity rather than the absence of oxidative processes [55].

Similarly, reduced hydrogen peroxide (H₂O₂) accumulation in tolerant cultivars suggests a finely regulated oxidative balance, where ROS act as signaling molecules rather than damaging agents [55,56]. Controlled ROS production supports stress signaling and acclimation, whereas excessive accumulation leads to irreversible cellular injury, particularly under repeated drought cycles [57]. The observed oxidative profiles therefore reinforce the notion that drought tolerance in coffee is closely linked to the capacity to regulate, rather than suppress, oxidative processes.

Overall, the coordinated regulation of hydraulic status, stomatal behavior, photosynthetic resilience, nutrient redistribution, osmotic adjustment, and oxidative balance reveals distinct drought-resilience strategies among C. arabica cultivars. Importantly, recovery capacity emerges as a central determinant of tolerance, highlighting that resilience under water deficit depends not only on survival during stress but also on the ability to rapidly restore physiological function once favorable conditions return [20,39]. These findings provide a physiological framework to support tolerance-oriented selection and management strategies in coffee under increasingly variable climatic conditions.

4. Materials and Methods

4.1. Plant Material and Experimental Conditions

Seedlings of ten cultivars of C. arabica (

Table 3. Genealogy of the cultivars of C. arabica evaluated in this study.

Namber	Trade Name	Crossing
1	Marsellesa	Hybrid Timor 832/2 × Villa Sarchi CIFC 971/10
2	Geisha	Ethiopian landrace
3	Catimor	Timor Hybrid 1343 × Caturra
4	Parainema	Selection of T5296
5	Arará	Natural cross between Obatá and Catuaí amarillo
6	Obatá amarillo	Hybrid Timor 832/2 × Villa Sarchi CIFC 971/10
7	Castillo	Timor Hybrid 832/2 × Caturra
8	Bourbon rosado	Bourbon genetic background
9	Catimor bolo	Local cultivate
10	Villalobos	Local cultivate

) were established in 3 L pots, using as substrate a mixture of agricultural soil, sand and peat in a proportion of 4:2:1 (v/v/v), with pH 5.57 and an organic matter content of 17.94% (see Supplementary Material S1). The evaluations began when the seedlings reached an average height of 26 cm and presented seven pairs of true leaves, selected for their vigor and uniformity between cultivars.

Information according to the World Coffee Research catalogue of coffee varieties: https://varieties.worldcoffeeresearch.org/arabica/varieties (accessed on 3 August 2025).The experiment was carried out under the completely randomized design in a polyethylene roof greenhouse with photosynthetically active radiation of 850 μmol m⁻² s⁻¹, the microclimatic conditions were recorded using an Elitech RC-4HC Data Logger (Elitech Technology Inc., Shenzhen, China), obtaining temperatures of 19.5 °C and relative humidity of 74.4%. The vapor pressure deficit (VPD) was calculated following the equation of Tetens [58]: (Figure 10).

$$\mathrm{V}\mathrm{P}\mathrm{D}=0.6108\mathrm{e}\mathrm{x}\mathrm{p}(\mathrm{T}+\mathrm{237.317.27}\mathrm{T})(1-100\mathrm{H}\mathrm{R})$$

T = Temperature (°C).

HR = Relative humidity (%).

Water regimes were established and monitored in three periods (Figure 10): AW, corresponding to irrigation at 100% of water availability; SWD, phase of severe water deficit, was imposed by the complete cessation of irrigation, during which pre-dawn leaf water potential (Ψ_pd) was periodically monitored in all cultivars, and which was terminated when at least one cultivar reached a Ψ_pd ≤ −3.0 MPa, used as an operational threshold following the criteria of Taiz and Zeiger and Rodrigues et al. [30,59]; and RI: rehydration period under adequate irrigation conditions for 15 days.

4.2. Monitoring of Foliar Water Potential

The pre-dawn leaf water potential (Ψ_pd) was determined immediately after leaf incision [27], using a Scholander-type pressure chamber (PMS-1000, Instrument Company, Albany, OR, USA). The evaluations were carried out on fully extended sheets at the end of the three periods: AW, SDW and RI.

4.3. Measurement of Gas Exchange

Gas exchange was measured at the end of each period on fully expanded leaves in the middle third of the seedlings between 10 a.m. and 1 p.m., using a LI-6800 portable photosynthesis meter (LI-COR Bioscienses, Lincoln, NE, USA). The parameters analyzed included net photosynthetic rate (A, μmol CO₂ m⁻² s⁻¹), stomatal conductance (g_s, mol H₂O m⁻² s⁻¹), transpiration rate (E, mmol H₂O m⁻² s⁻¹), and instantaneous water use efficiency (WUE, μmol CO₂/mmol H₂O m⁻² s⁻¹) (A/E). Chamber conditions were adjusted to: 600 μmol s⁻¹ airflow, 800 μmol m⁻² s⁻¹ photosynthetic photon flux density, 21 °C leaf temperature, 45% relative humidity, and 2.5 rpm fan speed.

4.4. Stomatal Features

Impressions of the leaf abaxial surface at the point of maximum leaf width (two leaves per plant and three different areas per leaf) were taken at the end of the RI period, which were observed under a Leica DMi8 M inverted optical microscope (Leica Microsystems GmbH, Wetzlar, Germany) [60]. Stomatal density (SD) was calculated as the number of stomata per unit of leaf area, in a visual field of 270.40 μm × 152.04 μm. The stomatal index (SI) was determined by the equation:

$$\mathrm{S}\mathrm{I}(\%)=\left[\right(\mathrm{s}\mathrm{t}\mathrm{o}\mathrm{m}\mathrm{a}\mathrm{t}\mathrm{a})/(\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{c}\mathrm{e}\mathrm{l}\mathrm{l}\mathrm{s}+\mathrm{s}\mathrm{t}\mathrm{o}\mathrm{m}\mathrm{a}\mathrm{t}\mathrm{a}\left)\right]\times 100$$

4.5. Elemental Analysis of Leaves

Leaf samples were freeze-dried for five days in a 4.5 L FreeZone (Labconco Corp., Kansas City, MO, USA) for elemental analysis. Subsequently, 0.2 g of dry and crushed tissue were taken, which underwent digestion in a nitric acid-chloric acid mixture [6:2, v/v) [61]. The resulting solution was filtered and measured to a final volume of 25 mL with ultrapure water. The quantification of macronutrients (K⁺, P, Ca²⁺ and Mg²⁺) and micronutrients (Fe²⁺, Cu²⁺, Mn²⁺ and Zn²⁺) was performed by atomic emission spectroscopy with microwave-coupled plasma (MP-AES 4100, Agilent Technologies, Santa Clara, CA, USA). The results were expressed in μg g⁻¹ dry mass after interpolating on calibration curves prepared with standard multi-elemental solutions in the ranges of 5, 10, 15 and 20 mg L⁻¹ for P, Ca²⁺, Na⁺, K and Fe²⁺, and 2, 5, 10 and 15 mg L⁻¹ for Zn²⁺, Mg²⁺, Cu²⁺ and Mn²⁺.

4.6. Analysis of Oxidative Stress Markers

Two fully extended leaves were collected (between 12 and 13 h) and crushed after being immersed in liquid nitrogen [11]. Subsamples were then taken for biochemical analysis.

4.6.1. Colorimetric Quantification of Proline Content

It was performed using the ninhydrin reaction method de Bates et al. [62]. To do this, 500 mg of leaf tissue was homogenized in 5 mL of 3% sulfosalicylic acid, followed by centrifugation at 10,000 rpm for 30 min. From the supernatant obtained, 1 mL was taken and mixed with 1 mL of acid ninhydrin (0.1 M) and 1 mL of glacial acetic acid (99.9%), in amber Falcon tubes. The reaction mixture was incubated at 90 °C for 60 min and rapidly cooled in an ice bath to stabilize the formed complex. For the final extraction of the chromophore, 3 mL of toluene was added and the upper phase was recovered, whose absorbance was determined at 520 nm in a Genesys 180 UV/Vis spectrophotometer (Thermo Scientific™, Madiminson, WI, USA). Proline content was calculated from a standard L-proline curve (Sigma-Aldrich, Darmstadt, Germany) and expressed as μg g⁻¹ fresh weight (FW).

4.6.2. Quantification of Lipid Peroxidation

Lipid peroxidation was assessed by quantifying malondialdehyde (MDA) using the thiobarbituric acid (TBA) assay, following the methodology proposed by Velikova et al. [63]. 200 mg of fresh tissue was homogenized with 2 mL of 0.1% trichloroacetic acid (TCA) (w/v). The homogenate was centrifuged at 14,000 rpm for 10 min, and 1 mL of the supernatant obtained was mixed with 2 mL of a 0.5% thiobarbituric acid (TBA) solution prepared in 20% TCA. The mixture was incubated at 90 °C for 30 min, followed by rapid cooling in an ice bath. It was then centrifuged again at 14,000 rpm for 5 min. Supernatant absorbance readings were taken at 450, 532, and 600 nm on a Genesys 180 UV/Vis spectrophotometer (Thermo Scientific™, Madison, WI, USA).

Malondialdehyde (MDA) content was calculated according to the equation [64]:

$$\mathrm{M}\mathrm{D}\mathrm{A} (\mathrm{\mu }\mathrm{m}\mathrm{o}\mathrm{l} {\mathrm{g}}^{-1} \mathrm{F}\mathrm{W})=(6.45\times (\mathrm{O}{\mathrm{D}}_{532}-\mathrm{O}{\mathrm{D}}_{600})-0.56\times \mathrm{O}{\mathrm{D}}_{450}\times \mathrm{V}/\mathrm{W}$$

V = final volume of the extract (mL)

W = fresh weight of the sample (g)

4.6.3. Spectrophotometric Quantification of Hydrogen Peroxide Content

The H₂O₂ content was quantified spectrophotometrically by its reaction with potassium iodide (KI), following the method described by Loreto and Velikova [65]. 200 mg of fresh tissue was homogenized in an ice bath with 2 mL of 0.1% trichloroacetic acid (TCA) (w/v). The homogenate was centrifuged at 12,000 rpm for 15 min and 1 mL was taken from the supernatant, which was mixed with 1 mL of potassium phosphate buffer 10 mM (pH 7.0) and 2 mL of potassium iodide (KI) 1 M. The reaction mixture was incubated at room temperature for 60 min, and absorbance was measured at 390 nm in a Genesys 180 UV/Vis spectrophotometer (Thermo Scientific™, Madison, WI, USA). The content was determined from a standard curve elaborated with known concentrations of H₂O₂ and was expressed in μmol g⁻¹ fresh weight (FW).

4.7. Experimental Design and Statistical Analysis

The experiment was carried out under a completely randomized design, with ten coffee cultivars and three water periods evaluated. Statistical analysis was performed using generalized mixed linear models (GLMM). The best model for each variable was selected based on the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). Adjusted marginal mean comparisons were performed using the emmeans function [66], and when appropriate, the Sidak correction was applied to control the error rate for farming using the cld function of the multcompView package [67]. Significant differences between cultivar combinations × period were represented by letters. Stomatal traits and elemental analysis were analyzed by single-way ANOVA, followed by Tukey’s HSD test for comparison of means in normal-distribution data, and by Kruskal–Wallis test for nonparametric data. All analyses were performed in R (version 4.5.2) using the RStudio environment.

5. Conclusions

The present study demonstrates that the adaptive response of Coffea arabica cultivars to drought at early developmental stages results from a multidimensional interaction emerging from the coordinated regulation of plant water status, gas exchange, stomatal architecture, biochemical adjustment, oxidative balance, and post-stress recovery capacity.

Based on the integrative evaluation of these responses, Castillo, Obatá amarillo, and Arará consistently exhibited the highest drought tolerance, particularly under severe water deficit conditions and a greater post-rehydration recovery capacity. These cultivars were characterized by a more stable water status, conservative stomatal regulation, higher water-use efficiency, and rapid recovery of gas exchange following rehydration, accompanied by structural traits favorable for long-term water conservation, greater osmoprotective capacity, and lower oxidative damage, which together support their superior drought tolerance.

In contrast, Geisha and Marsellesa displayed a distinct adaptive strategy, characterized by a pronounced recovery of photosynthetic activity after stress, despite experiencing marked reductions during the drought phase. This response suggests high physiological elasticity, indicating that recovery capacity, rather than stress avoidance, constitutes the main adaptive trait in these cultivars. These findings highlight the importance of considering recovery dynamics, in addition to performance during the stress phase, when evaluating drought tolerance.

Overall, the combined assessment of gas exchange regulation, water-use efficiency, stomatal architecture, and recovery capacity clearly identifies Castillo, Obatá amarillo, and Arará as the most drought-tolerant cultivars at early developmental stages, whereas Geisha and Marsellesa exhibit high resilience primarily associated with post-stress recovery, but with lower consistency across the water gradient. This integrative framework positions the tolerant cultivars as valuable candidates for breeding programs aimed at strengthening drought resilience in coffee under future climate change scenarios.