Two cropped genotypes of C. canephora Pierre ex A. Froehner cv. Conilon Clone 153 (CL153) and C. arabica L. cv. Icatu Vermelho (an introgressed variety resulting from a cross of C. canephora and C. arabica cv. Bourbon Vermelho, then further crossed with C. arabica cv. Mundo Novo) were evaluated following an experimental design previously set (Dubberstein et al., 2020). Briefly, 32 plants in total were grown for 7 years in 80 L pots in two walk-in growth chambers (EHHF 10000, ARALAB, Portugal), in a substrate consisting of a mixture of soil, peat, and sand (3:1:3, v/v/v), with pH 6.5, and under controlled temperature (25°C/20°C, day/night, ± 1°C), PAR irradiance (ca. 700–800 μmol m⁻² s⁻¹, at the upper canopy level), relative humidity (70% ± 2%), photoperiod (12 h), and air [CO₂] (380 ± 5 μL CO₂ L⁻¹).

Irradiance was provided by a combination of fluorescent (Lumilux L58W/840, Osram, Germany) and halogen (100 W, Halolux Ceram, Osram) lamps. Plants were fertilized (see Ramalho et al., 2013) and well-watered (WW) by adequate irrigation every 2 to 3 days. Water deficit and heat conditions were sequentially imposed in a gradual manner in order to allow plant acclimation (see below), in eight plants per treatment and genotype.

Water conditions were first imposed under adequate temperature (25°C/20°C), considering the exposure to approx. 80% (WW) or 10% (severe water deficit, SWD) of maximal pot water availability, exactly as described earlier (Dubberstein et al., 2020; Rodrigues et al., 2024). For that, WW plants were kept fully irrigated (predawn water potential, Ψ _pd ≥ −0.35 MPa; relative water content, RWC ≥ 91%), while SWD plants were subjected to a partial withholding irrigation (through a partial reposition of water that was lost in each pot, every 2 days) for 2 weeks to promote SWD conditions (reaching values of Ψ _pd below −3.7 MPa and of RWC close to 60%). After this, the SWD plants were maintained under these conditions for another 5 days before evaluation at the control 25°C/20°C temperature, as well as during the entire exposure to increased temperature (see 2.3).

The mentioned WW and SWD conditions were maintained along a gradual temperature increase of 0.5°C day⁻¹ (of diurnal temperature) from 25°C/20°C up to 42°C/30°C, with stabilization of 5 days at 31°C/25°C, 37°C/28°C, and 42°C/30°C to enable the programmed evaluations, exactly as described in Rodrigues et al. (2024). Finally, temperature was readjusted to 25°C/20°C, and then all the plants were irrigated to the initial watering conditions (80% of maximal pot water availability), and their potential recovery was monitored for 14 days (Rec14).

Leaf Ψ _pd was measured using a pressure chamber (Model 1000, PMS Instrument Co., Albany, OR, USA), according to Schölander et al. (1965), in individual leaves of five or six plants per treatment and genotype. Samplings every 2 or 3 days allowed a close monitoring, though only data from key points during temperature increase and recovery periods were shown.

Leaf RWC was estimated as described for Coffea spp (Ramalho et al., 2018), using 10 foliar discs of 0.5 cm² each, punched from the same leaves used for Ψ _pd determinations. RWC values (%) were calculated as = [(FW − DW)/(TW – DW)] × 100, where FW represents the fresh weight determined immediately after cutting the discs, TW is the turgid weight obtained after overnight rehydration of the discs in a humid chamber at ca. 20°C, and DW is the dry weight obtained after drying the discs at 80°C for 48 h.

Both RWC_pd and Ψ _pd measurements were performed at predawn on five to six replicates per treatment, every 2 to 3 days, but only the data at the main temperature points for data samplings (considering temperature rise and both heat and drought recoveries) were presented.

Samplings were made using newly matured leaves from the upper (well-illuminated) part of six to eight plants per treatment and genotype, under photosynthetic steady-state conditions (after ca. 2 h of light) at 25°C/20°C (control), 37°C/28°C, 42°C/30°C, and Rec14. The same leaf pool per plant was used for all evaluations. The freshly collected leaf material was immediately used for lipid analysis or flash frozen in liquid N₂ and stored at –80°C, which was finely powdered in liquid N₂ prior to protein and gene expression analysis. Leaf tissue extractions were performed using an ice-cold mortar and pestle and cold homogenizing solutions.

Total RNA was extracted following Marques et al. (2024) using the Analytik-Jena InnuSPEED Plant RNA Kit (Analytik Jena Innuscreen GmbH, Jena, Germany). cDNA was synthesized from 1 μg of total RNA using the SensiFAST™ cDNA Synthesis kit (Meridian Bioscience, Cincinnati, OH, USA), according to the manufacturer’s recommendations. PCR reactions were prepared using the SensiFAST™ SYBR No-ROX kit (Meridian Bioscience) following the protocol and the parameters described in Marques et al. (2023). Reactions were carried out in 96-well plates using a qTOWER 2.2 Thermal Cycler (Analytik, Jena, Germany) using the following parameters: hot start activation of the Taq DNA polymerase at 95°C for 10 min, followed by 40 cycles of denaturation at 95°C for 15 s, annealing at 60°C for 30 s, and elongation at 72°C for 30 s. A melting curve analysis was performed at the end of the PCR run by a continuous fluorescence measurement from 55°C to 95°C with sequential steps of 0.5°C for 15 s. A single peak was obtained, and no signal was detected in the negative controls. Three technical replicates were performed. Expression studies included 14 selected genes ( Supplementary Tables S1 ; S2 ), namely, aquaporins (CaPIP2, CaTIP2), dehydrin (DH1a), dehydration-responsive element-binding protein 1D (DREB1D-F1), chloroplast 70 kDa heat shock-related protein (HSP70), chloroplast early light-induced protein (ELIP), chloroplast 20 kDa chaperonin (Chape 20), mitochondria chaperonin CPN60 (Chape 60), antioxidative enzymes [e.g., catalase (CAT), cytosolic (APXCyt), and chloroplast ascorbate (APXChl) peroxidases], and violaxanthin de-epoxidase (VDE2). Malate dehydrogenase (MDH) and ubiquitin (UBQ10) were used as reference genes (Martins et al., 2017). This selection of genes (and most corresponding proteins, as shown below) was based on prior evidence of involvement in abiotic stress responses (drought and/or heat), with emphasis on those known to play roles in oxidative stress protection, membrane stabilization, water transport, and stress signaling pathways. Specifically, aquaporins (PIP and TIP families), chaperonins, dehydrins, heat shock proteins, antioxidative enzymes (APXs, CAT, SOD), and regulatory transcription factors (DREB1D-F1) were targeted. Candidate genes were identified from Coffea spp. transcriptomic databases and previous studies (e.g., Fernandes et al., 2021; Marques et al., 2022a, b) and validated by expression profiles under similar stress conditions.

Protein extraction from ca. 200 mg FW of powdered frozen leaves, liquid chromatography and high-resolution mass spectrometry (NanoLC-MS/MS) peptide analysis, and protein identification and quantification were carried out as described in Dubberstein et al. (2020). The reference database from C. canephora (Denoeud et al., 2014) downloaded on 1 July 2019 was used for peptide and protein inference by the MASCOT Daemon 2.6.1 search algorithm (Matrix Science). A targeted approach was used, selecting a set of 17 proteins ( Supplementary Table S3 ) usually involved in plant stress responses and closely associated with the selected genes (see 2.5.2). Protein annotation was obtained from the UniProt Knowledgebase (UniProtKB) (https://www.uniprot.org/uniprot/?query=&sort=score). Mass spectrometry proteomics data were deposited in the ProteomeXchange Consortium via the PRIDE partner repository, with the dataset identifier PXD019474 and DOI: 10.6019/PXD019474 for the C. arabica proteome and the dataset identifier PXD019541 and DOI: 10.6019/PXD019541 for the C. canephora proteome.

Plants of CL153 and Icatu genotypes were independently subjected to eight treatment combinations, forming a 2 × 4 factorial consisting of two water availability levels (WW or SWD) and four levels of temperature (25°C/20°C, 37°C/28°C, 42°C/30°C, and Rec14) under a completely randomized design, with eight plants per treatment and genotype. Altogether, the entire experiment lasted 82 days: SWD plants reached the desired Ψ _pd within 14 days upon gradual drought imposition and were kept in these conditions another 5 days before the temperature was increased to 42°C/30°C (49 days). Soil was later fully watered and the temperature was set to 25°C/20°C, and plants were analyzed for stress relief for 14 days (Rec14).

Datasets were analyzed using a two-way ANOVA to assess differences between water availability levels, temperature treatments, and their interaction. Mean comparisons (independently for each genotype) were conducted using Tukey’s HSD test. Statistical analyses were performed in STATISTICA v7.0 (StatSoft, Hamburg, Germany) with a 95% confidence level applied to all the tests.

Water restriction resulted in SWD, assessed by Ψ _pd mean values of ca. −3.9 (CL153) and −3.7 MPa (Icatu) at 25°C/20°C, with concomitant values of RWC close to 60% in both genotypes ( Table 1 ). In contrast, though single heat (at the two highest temperatures) doubled Ψ _pd values, no significant changes occurred in water status in WW plants at 42°C/30°C, when they reached −0.8 MPa (CL153) and −0.7 MPa (Icatu). The good hydration status of WW plants at the highest temperature was further confirmed by the RWC value, which was maintained at 95% in both genotypes.

The combined stress exposure led to further Ψ _pd decline, to minimal values close to −4.5 MPa at 37°C/28°C in both cultivars (not significantly different from the values at 42°C/30°C), together with a greater dehydration only in Icatu-SWD plants that reached minimal RWC values below 50% at 42°C/30°C.

Notably, after the simultaneous restoration of water and temperature control conditions, SWD plants of both genotypes showed almost complete recoveries of Ψ _pd and RWC from 4 days onward up to 2 weeks (Rec14), especially in WW plants, as compared to their respective controls.

A strong impact of the imposed water and/or temperature conditions was found in selected genes ( Figure 1 ; Supplementary Table S2 ) coding for proteins associated with plant stress response. In both genotypes, genes coding for intrinsic proteins of the plasma membrane (CcPIP2) and tonoplast (CcTIP2) subfamilies of aquaporins, as well as the dehydration-responsive element-binding transcription factor gene (DREB1D-F1), were unresponsive to heat. In contrast, these genes were highly sensitive to drought (regardless of temperature), distinctly downregulated (aquaporins) or upregulated (DREB1D-F1), a pattern sustained by Rec14. Notably, in CL153, the combined stress exposure (SWD at 37°C/28°C and 42°C/30°C) further amplified the expression of DREB1D-F1 found in SWD plants at 25°C/20°C.

The dehydrin DH1a gene exhibited minimal responsiveness, if any, to all stress conditions in CL153, whereas in Icatu, a somewhat higher expression was observed under single heat stress (37°C/28°C and 42°C/30°C) or combined with SWD.

In general, genes coding for three chloroplast proteins (stroma heat shock, HSP70; early light-induced protein, ELIP; 20 kDa chaperonin, Chape 20), along with the mitochondrial chaperonin CPN60 (Chape 60) were slightly but systematically upregulated by single drought in both genotypes, particularly in Icatu (except for Chape 20). Heat stress alone (37°C/28°C and, especially, 42°C/30°C) promoted an even greater upregulation of these genes. That was somewhat further amplified under the combined stress exposure (SWD, 37°C/28°C) concerning Chape 20 in CL153 and HSP70, ELIP, and Chape 60 in Icatu. The Chape 20 gene was one of the few genes with a stronger response in CL153 than Icatu, especially under SWD and combined stresses. Although with a decline by Rec14, both WW and SWD plants often had higher expression than in their initial controls, especially for ELIP and Chape 60 in Icatu, as well as HSP70 and Chape 20 in both genotypes.

Among the genes coding for enzymes directly involved in ROS control, the transcript abundance of catalase isozyme 1 (CAT), Cu,Zn-superoxide dismutase (CuSOD₂ ), peroxisomal ascorbate peroxidase (APXt+s), and violaxanthin de-epoxidase (VDE2) barely responded to a single drought in either genotype. Still, a strong gene upregulation was found for cytosol ascorbate APX (APXCyt) in CL153 and chloroplast APX (APXChl) in Icatu. In contrast, heat alone consistently upregulated these genes, strongly in CL153, with APXCyt standing out with the greatest value among all the studied genes in this genotype. Notably, stress superimposition (SWD, 42°C/30°C) prompted an even greater expression of CAT in CL153 and of all the genes in Icatu (except VDE2). Such Icatu overresponse was particularly strong in APX genes, especially in APXChl, which is associated with chloroplast antioxidative protection. Finally, it was noteworthy that, as noted for several of the abovementioned genes, the upregulation was usually maintained by Rec14, but to a lower extent than at 42°C/30°C.

As referred above, most selected proteins were coded by the presented genes so that transcriptomic and proteomic patterns of changes could be integrated ( Figure 2 ; Supplementary Table S3 ). The intrinsic protein subfamilies of AQPs from plasma (PIP) and tonoplast (TIP) membranes showed different changes, both between subfamilies and among individual members within each subfamily.

Overall, AQPs were more responsive to drought in Icatu, although CL153 denoted a higher constitutive level (25°C/20°C; WW plants). In detail, PiP2 protein abundance significantly increased in Icatu under both single drought and heat (42°C/30°C). In CL153, no changes were observed in either condition, but it should be noted that there was a higher constitutive value along the entire experiment (including Rec14), similar to the maximal values found in Icatu. Among the other protoplasm AQPs (PIP1.2, PIP2.1, PIP2.7), protein abundance usually tended to rise under the single stresses in both genotypes, although significant only for PIP1.2 in Icatu. The opposite was observed under stress superimposition (SWD, 42°C/30°C) as compared with the single stresses (except for PIP2 in Icatu).

The tonoplast aquaporin TIP1.1 showed a significant rise in CL153 only under single drought, but although without statistical significance, it seems noteworthy to mention that Icatu showed a double abundance under both single SWD and 42°C/30°C and even greater with stress combination. TIP4.1 was undetected in most treatments, regardless of genotype.

Dehydrin DH1a protein abundance was not impacted by single heat stress, but greatly increased under SWD conditions in both genotypes, at all temperatures and with a positive interaction at 37°C/28°C in CL153. In Rec14, this protein returned to control values.

Among the group of protective proteins in chloroplasts (HSP70, ELIP, and Chape 20) and mitochondria (Chape 60), Chape 20 showed a particularly high constitutive abundance (WW, 25°C/20°C) in both genotypes. This protein was also one of the few that showed a higher value or response pattern to single drought and heat conditions in CL153 than in Icatu plants. All the abovementioned chloroplast proteins tended to have higher values in response to single drought or 42°C/30°C (except ELIP), although it was non-significant. In contrast, the mitochondria Chape 60 greatly accumulated under either single stress in both genotypes, more prominently under heat in Icatu (with ca. 30-fold increase). Notably, ELIP markedly increased in Icatu under the stress combination, in comparison to the respective WW plants at 42°C/30°C. By Rec14, only Chape 20 and 60 maintained increased levels in WW and SWD plants.

In contrast with the small gene expression changes, the abundance of enzymes associated with oxidative stress control (CAT, APX Cyt, APX Chl, APXt+s, and VDE) was among the most responsive ones to drought or heat in both genotypes, particularly in Icatu. The exception was Cu/ZnSOD₂, which showed moderate increases in response to drought and/or heat in both genotypes, whereas VDE abundance increased only under SWD (greatly in Icatu). Also, only Icatu showed an additional abundance increase of all APX proteins under the stress combination and maximal values for CAT and APXs in all stress conditions. Except Cu/ZnSOD₂ and VDE, increased levels of all of these enzymes were maintained by Rec14, in both genotypes, especially in SWD plants.

A quantitative FA analysis showed that the TFA content of chloroplast membranes was mostly unaffected in CL153 plants, regardless of the imposed stress conditions ( Table 2 ). This contrasted with Icatu showing significant TFA increments due to drought (68% at 25°C/20°C), heat (47% at 37°C/28°C), and stress combination (142% under SWD by 37°C/28°C), although by 42°C/30°C, these values declined, approaching those found at 25°C/20°C. Also, contrasting patterns were observed among genotypes by Rec14, with higher (Icatu) and lower (CL153) TFA values than at the initial WW-25°C/20°C conditions.

The qualitative changes were assessed through FAs’ relative weight and their unsaturation degree. DBI rise reflected significant increases in FA unsaturation under single drought, in both genotypes, whereas single heat increased DBI at 37°C/28°C only in Icatu. However, a DBI decline was found from 37°C/28°C to 42°C/30°C, together with higher DBI values in SWD plants at the maximal imposed temperature (as compared with WW counterparts), in both genotypes. By Rec14, the values tended to those of the initial control in both genotypes. Such DBI modifications mainly resulted from the opposite changes in the relative weight of the major saturated palmitic (C16:0) and the highly unsaturated linolenic (C18:3) acids, which together accounted for ca. 75% (CL153) and 72% (Icatu) of the chloroplast membranes’ TFAs.

The most represented FA, C18:3, significantly increased upon both single drought and heat (37°C/28°C) imposition in Icatu, while in CL153, it rose under SWD and was unresponsive to heat. Still, under 42°C/30°C, the C18:3 values in WW plants were similar to those of plants at control temperature in both genotypes. The stress combination at maximal temperature reduced C18:3 as compared to the SWD value at 25°C/20°C, but these were still higher than in WW plants at 42°C/30°C, especially in Icatu. By Rec14, the C18:3 values approached those of the initial control, but SWD plants of both genotypes kept greater C18:3 values than their WW counterparts, similar to what happened due to SWD impact alone at 25°C/20°C.

The second most represented FA, C16:0, declined under single drought and showed a stress interaction (SWD, 42°C/30°C) in both genotypes, being significant only in Icatu. Single heat significantly reduced C16:0 in Icatu plants up to 37°C/28°C (but not at 42°C/30°C), whereas in CL153, an increase was found only at the highest temperature. By Rec14, these FA values were close to the initial control conditions in WW plants of both genotypes, but the SWD ones maintained lower values than the WW counterparts (significant in CL153).

The third and fourth most represented FAs, the unsaturated linoleic acid (C18:2) and the saturated stearic acid (C18:0), respectively, were responsive only to single drought in both genotypes, showing consistent declines (that were maintained at 37°C/28°C and 42°C/30°C for C18:2). By Rec14, the values from SWD plants approached those of the initial control conditions.

Finally, the moderately unsaturated cis+trans palmitoleic (C16:1c+t) and oleic (C18:1) acids differed from the rest of the FAs, with a negligible response to single SWD in both genotypes. With single heat, C16:1c+t declined in CL153 and C18:1 rose in both genotypes, but both FAs increased under stress combination (SWD, 42°C/30°C) when compared with their respective WW plants at 42°C/30°C, especially in Icatu. By Rec14, the values of C16:1c+t (Icatu) and C18:1 (both genotypes) were higher than under initial control conditions.

Values of Ψ _pd ≤ −3.5 MPa reflect extreme water deficits (Pinheiro et al., 2004; Semedo et al., 2021), clearly below −2.15 MPa, causing leaf wilting in coffee plants (Santos and Mazzafera, 2012). Also, temperatures above 37°C (or even 39°C) exceeded thermal tolerance in Coffea spp (Rodrigues et al., 2016), impairing physiological, biochemical, and molecular functions (Rodrigues et al., 2016; Dubberstein et al., 2020; Marques et al., 2021). Therefore, a severe drought degree was imposed in SWD plants at 25°C/20°C, with Ψ _pd values of ca. −3.7 to −3.8 MPa, further confirmed by the strong declines of RWC_pd to values close to 58% in CL153 and even lower (47%) in Icatu ( Table 1 ). With stress co-occurrence, the Ψ _pd values of SWD plants were further reduced to ca. −4.5 MPa at 42°C/30°C, regardless of genotype, with RWC decreasing below 50% in Icatu, thus allowing to pinpoint genotype- and stress-dependent responses of selected genes, proteins, and chloroplast membrane FAs.

Notably, by Rec14 it was observed an almost full recovery of g _s (Dubberstein et al., 2020), in agreement with the resumption of RWC_pd and Ψ _pd ( Table 1 ), although with a tendency to lower values of these parameters in the SWD plants of both genotypes by Rec14. This suggests a considerable tolerance of hydraulic traits under harsh conditions of combined water deficit and heat and is in line with the full recovery of the same parameters after 1-week recovery to drought and cold stress in coffee genotypes, including Icatu (Ramalho et al., 2018). In fact, dehydration sensitivity has been ascribed to a lack of full recovery in water potential and stomatal conductance, accompanied by intense leaf shedding (Martins et al., 2019), none of which occurred in our study, since leaf senescence was negligible (if any) and none of the plants died during the stress period or in the subsequent recovery period. Indeed, under harsh drought under field conditions, coffee plants can tolerate extreme drought (Ψ _pd values as low as −4.4 MPa) with no plant mortality (Martins et al., 2019). This high tolerance to dehydration was associated with irrelevant (or absent) hydraulic conductivity failure and xylem vulnerability to embolism (Martins et al., 2019) and in line with the findings of Pereira et al. (2016), who predicted that coffee plant death will occur at Ψ _w as low as −7 to −8 MPa.

Coffee plants can activate a range of defense mechanisms in a species-/genotype-dependent manner (Ramalho et al., 2018; Dubberstein et al., 2020; Semedo et al., 2021). These responses support plant resilience to drought and heat and involve the synthesis and regulation of proteins that protect cellular structures, stabilize membranes, scavenge ROS (e.g., HSPs, DHNs, antioxidant enzymes), regulate water transport across cell membranes, and help maintain cellular water balance (AQPs) (Araújo et al., 2019).

Unlike CL153, which showed minimal quantitative changes in TFAs in response to drought and/or heat, Icatu exhibited marked lipid de-novo synthesis under drought, moderate heat (37°C/28°C) and, especially, under stress combination (SWD-37°C/28°C) ( Table 2 ). This greater responsiveness of the C. arabica genotype reflects greater lipid metabolic plasticity, a key to stress acclimation to cold (Partelli et al., 2011; Scotti-Campos et al., 2014) and heat (Scotti-Campos et al., 2019). Such lipid metabolism flexibility was further reflected in important qualitative changes. The reprogramming of the FA profile occurred under drought (both genotypes) or heat (only in Icatu, up to 37°C/28°C), through an increase in the unsaturation level (reflected in DBI rise). This shifted the balance between the two most represented FAs, C16:0 and C18:3, which were reduced and increased, respectively. Additionally, single heat (42°C/30°C) increased C16:0 in CL153. Such drought- and/or heat-prompted (37°C/28°C) unsaturation (greater in Icatu) would support membrane fluidity and functionality (Gombos and Murata, 1998; Routaboul et al., 2000). This is highly relevant for maintaining C-assimilation, since photosynthetic performance closely depends on higher FA unsaturation to preserve lipid acyl motion in thylakoid membranes (Harwood, 1998; Siegenthaler and Tremolieres, 1998), concerning PSI and PSII thylakoid electron transport rates, which are membrane-based events. These traits are also crucial to integrate newly synthesized D1 (Kern and Zouni, 2009) related to the repair processes needed to sustain PSII function under stress. This agrees with the preservation of the photosynthetic machinery potential in Coffea spp. under SWD (Dubberstein et al., 2020; Semedo et al., 2021) and up to 37°C/28°C, but with impacts at 42°C/30°C (Dubberstein et al., 2020; Rodrigues et al., 2024), when DBI falls as compared with 37°C/28°C. Still, a higher abundance of polyunsaturated FAs also increases lipoperoxidation risk, as double bonds are preferential targets of hydrolytic enzymes, peroxidases, ROS, and free radicals (Girotti, 1990; Öquist, 1982), thus requiring a complementary strengthening of antioxidative defenses, as reported under cold (Fortunato et al., 2010) and heat (Martins et al., 2016). That was the case here under SWD and heat (37°C/28°C), since although with different extent, both genotypes, particularly Icatu, showed an enhanced coordination of lipid matrix remodeling together with greater abundance of antioxidant enzymes (APXs, CAT) and their activity (Rodrigues et al., 2024), stress proteins (HSP70, Chaperonins), and membrane protectors, such as AQPs (PIP1.2, PIP2.1, PIP2.7, TIP1.1) and DH1a, all contributing to membrane protection. For instance, the presence of several PIPs (PIP1, PIP1,1, PIP2,1, PIP2,5, PIP2,6, PIP2,7) and TIPs (TIP1,2, TIP4,1-1) was associated with enhanced antioxidant capability and lowered ROS presence and lipid peroxidation level (Patel and Mishra, 2021), whereas DHNs, which bind to membrane lipids, have antioxidative activity, and their accumulation protects membrane integrity, acting against lipoperoxidation (Theocharis et al., 2012; Tiwari and Chakrabarty, 2021; Szlachtowska and Rurek, 2023).

Notably, only Icatu increased C16:1c+t values (due to greater TFA abundance) under single drought, stress interaction (SWD, 42°C/30°C), and Rec14 (WW plants), likely contributing to chloroplast membrane stability (Scotti-Campos et al., 2014). C16:1t is an exclusive component of thylakoid phosphatidylglycerol (PG) (Öquist, 1982; Siegenthaler, 1998), and both PG and C16:1t are involved in the supramolecular thylakoid membrane organization of LCHII proteins and pigments, stabilizing the PS complexes and allowing an efficient non-cyclic electron flow (Siegenthaler and Tremolieres, 1998; Vijayan et al., 1998; Yang et al., 2005). This FA is also involved in the process of replacement of damaged D1 protein (Siegenthaler and Tremolieres, 1998; Harwood, 1998; Gray et al., 2005), reducing PSII photoinhibition and promoting a faster recovery from the photoinhibited state (Gombos and Murata, 1998; Siegenthaler and Tremolieres, 1998), in accordance with Icatu resilience and recovery under drought (Semedo et al., 2021) and heat (Dubberstein et al., 2020).

Overall, the ability to remodel the lipid matrix of chloroplast membranes is a crucial feature to drought/heat resilience in Coffea spp., working in tandem with complementary protection mechanisms (e.g., antioxidative), ultimately sustaining productivity under adverse conditions.

Accelerating the identification of drought- and heat-resilient genotypes is a key objective in breeding programs, especially under the pressing challenges of climate change. Traditional phenotyping approaches rely on the evaluation of adult plants in field conditions, which are time-consuming, highly variable, and resource-intensive. Our findings suggested that several stress-responsive traits (at the molecular, biochemical, and physiological levels) are detectable in the leaves, providing a potential shortcut for selecting resilient genotypes. Also, they reinforced the possibility of using environmental controlled stress experiments to assess genotypic plasticity during the vegetative phenophase. Integrating gene expression (e.g., DREB1D-F1, APXChl), enzyme protein abundance and activity (e.g., CAT, APX), and lipidomic profiling and unsaturation of FAs (e.g., DBI, C16:0, C18:3, C16:1t) could form the basis of high-throughput screening tools for the pre-field selection of elite genotypes.

Future studies should focus on validating the here identified biomarkers across developmental stages and diverse genetic backgrounds, defining threshold expression or activity levels that reliably predict long-term stress performance, developing cost-effective protocols (e.g., qPCR panels, ELISA kits, targeted omics) applicable in breeding nurseries and controlled environments. Still, it is noteworthy that gene transcripts and corresponding protein abundance and activity often showed a distinct pattern of response, pointing to the need for an integrated proteomic, transcriptomic, and ecophysiological analysis to get a truly accurate perspective. Overall, the identification of these reliable stress-responsive traits opens new avenues for accelerating the selection of drought- and heat-resilient Coffea spp. cultivars, reducing breeding cycles and enhancing genetic gain under climate stress scenarios.